Enhancing the Hemocompatibility of 3D-Printable Silicone Elastomers for Artificial Lung Applications

Abstract

Clinical hollow fiber artificial lungs are prone to clotting, necessitating the use of systemic anticoagulation and thus increasing the risk of bleeding events. This study seeks to address these limitations by creating hemocompatible and biomimetic 3D printed artificial lungs. This study investigates the nonthrombogenic effects of imbuingpolydimethylsiloxane (PDMS or silicone elastomer) based 3D-printable resin with hydrophilic molecules with the goal of reducing the body’s natural coagulation response to foreign materials, increasing device lifetime, and reducing systemic anticoagulation, thereby coming closer to mimicking the native in vivo blood interface. First, contact angle (hydrophilicity) tests were done to narrow down the number of candidate modifications for the development of a high-resolution PDMS resin for vat photopolymerization (VPP). Then, dynamic blood flow through testing was performed using a high-resolution PDMS base resin modified by 1) adding 1% 2-Methacryloyloxyethyl phosphorylcholine (MPC) to the base resin; 2) adding 1.8% Poly(ethylene glycol) methacrylate (PEGMA) to the base resin; or 3) infusing the neat PDMS devices with 2% dimethylsiloxane-[60-70% ethylene oxide] (PEO-PDMS) in ethanol post-printing. Biomimetic microfluidic capillary devices designed in SOLIDWORKS were 3D printed via SLA, cleared of uncured resin, and tested for coagulation with freshly-drawn ovine whole blood. Devices (n ≥ 12 per group) were exposed to blood for 10 minutes at 0.8 mL/min and evaluated for clotting via fluorescent confocal microscopy, percent clotting area analyses, pressure logging data, and flow cytometry. The 1% MPC, 1.8% PEGMA, and 2% PEO-PDMS infusion resin groups demonstrated a significant decrease in clotting area and fluorescence intensity when compared to the unmodified base resin and a commercially-available resin (FTD Nano Clear). The top performing modification (PEO-PDMS infusion) decreased clotting area by 57.5 and 65.2% and fluorescence intensity by 84.6 and 88.3% relative to the unmodified base resin and FTD Nano Clear resin, respectively.

Graphical Abstract

INTRODUCTION

Extracorporeal membrane oxygenation (ECMO) is a form of life support used in infants, children, and adults with life-threatening heart and/or lung conditions. Currently, clinical artificial lungs, known as hollow-fiber oxygenators, are utilized in thousands of extracorporeal membrane oxygenation (ECMO) and millions of cardiopulmonary bypass (CPB) cases per year.^1,2 These hollow-fiber devices have a short shelf life, are unconducive to an active or even normal lifestyle, and are susceptible to clotting. The burgeoning field of microfluidics, particularly biomicrofluidics, offers a transformative avenue for advancing ECMO technology. Microfluidics has become a growing industry with the number of publications on microfluidics doubling every 15 months.³ However, it has become imperative to choose the correct materials with which to develop blood contacting microfluidic devices to avoid device sensitivity and rejection as a result of protein buildup and thrombosis.

In the clinical setting, patients on ECMO require around the clock care and systemic anticoagulation, thus increasing the incidence of bleeding events and complications.⁴ Coagulation and clotting are the major causes of device failure.⁵ Additionally, the non-hemocompatible nature of these hollow-fiber devices puts further strain on blood. During oxygenation, the blood traverses a dense bundle of gas-permeable hollow fibers and experiences non-physiological flow paths and shear stresses, triggering the blood coagulation cascade and inflammatory responses in the patient.^6–8

Microfluidic artificial lungs (μALs) have emerged as a pioneering class of membrane oxygenators, offering potentially significant advantages over conventional hollow-fiber counterparts. By closely emulating the alveolar microenvironment of the natural lung in terms of size, scale, and features,¹ μALs hold promise for enhanced gas exchange efficiency, superior hemocompatibility, the creation of biomimetic blood flow networks, and physiologically relevant blood vessel pressures and shear stresses.⁹ These innovations address critical challenges such as thrombosis, clotting, and inflammatory responses associated with traditional oxygenation methods. However, a significant challenge hindering their clinical translation lies in the fouling and protein buildup on hydrophobic surfaces inherent to materials like polydimethylsiloxane (PDMS).¹⁰

Protein fouling refers to the accumulation of proteins on surfaces, particularly in biomedical devices, leading to impaired function and potentially adverse biological responses. In the context of microfluidic devices, protein fouling can occur when blood comes into contact with non-native device surfaces, leading to the adsorption of proteins from the blood onto the foreign material surface. This accumulation triggers a cascade of biological responses, including thrombosis, inflammation, and immune reactions, ultimately compromising the device’s performance and biocompatibility.¹¹

Hydrophilic modifications offer a promising strategy to mitigate protein fouling in microfluidic devices. By introducing hydrophilic molecules into the structure of the device that can provide a polar blood-interacting surface, the affinity for water is increased. This forms a tightly bound water layer close to the blood-interacting surface that repels proteins, leading to reduced protein adsorption and improved fouling resistance. Furthermore, hydrophilic modifications can enhance biocompatibility of microfluidic devices by mimicking the properties of biological interfaces, such as those found in native human capillaries and blood surface membranes. This approach reduces protein fouling and promotes interactions with blood components that are more akin to those observed in vivo, thereby minimizing adverse immunological responses.

Prior research in our laboratory has developed a high-resolution 3D-printable photopolymerizable polydimethylsiloxane (PDMS) resin capable, for the first time, of producing truly micron scale features, flow channels, and membranes via vat photopolymerization 3D printing.^12–17 PDMS is one of the most popular materials for microfluidic devices, lab-on-chip devices, and other biomimetic applications due to its optical transparency, robust mechanical properties, and high gas permeability.¹⁸ However, owing to its hydrophobicity, PDMS quickly initiates the coagulation cascade when brought into contact with blood. There have been many previous demonstrations of hydrophilic modifications of PDMS to reduce fouling.^19,20 PDMS surface modification strategies span physical methods (e.g., oxygen plasma, UV/ozone, thermal treatment) that break Si–CH₃ bonds to form silanol (Si–OH) groups, and chemical approaches including polymer adsorption, surfactants, and grafting.²¹ Plasma treatments rapidly enhance hydrophilicity (contact angles <20°) and biocompatibility but suffer hydrophobic recovery within hours-days and cannot access enclosed microchannels post-fabrication.²² UV/thermal methods offer similar transient polarity via oxidation without plasma equipment.²³ Chemical modifications provide greater versatility: passive adsorption of hydrophilic polymers (e.g., PVA, PEG) relies on hydrogen bonding with surface siloxanes or water layers, while surfactants (e.g., PEO-PDMS, Pluronic^®, Triton X-100) create amphiphilic surfaces via hydrophobic anchoring and hydrophilic tails oriented outward.²⁰ Grafting yields covalent stability (e.g., PEG-silanes forming Si–O–Si bonds via silanol condensation).²⁴

While adsorption and surfactants enable simpler, plasma-free modification, they risk leaching during solvent extraction common in 3D printing workflows. Additionally, nearly all previous techniques require a surface modification step, such as plasma activation, and are not effective for enclosed microchannels lacking surface access. PDMS-zwitterion microfluidic fabrication remains nascent, with only isolated reports of Sylgard 184-SB-diallyl hybrids (static blood contact, comparable gas transfer) and surface-segregating copolymer blends (0.025 wt%, 6-month stability, soft lithography).^25,26 Critically absent are high-resolution vat photopolymerization-compatible formulations with dynamic whole blood validation. Further, there is a dearth of literature on developing blood compatible 3D printing resins. In fact, all current 3D-printable materials including our custom PDMS resin lack the requisite hemocompatibility for blood contacting applications, creating a bottleneck for translation to clinical applications.^20,21

To overcome these challenges, our custom PDMS resin was modulated through the direct addition of various hydrophilic molecules with the goal of improving hemocompatibility, eliminating the need for a surface activation step via plasma or chemical modification. Methacrylated phosphorylcholine (MPC), Poly(ethylene glycol) methacrylate (PEGMA), and dimethylsiloxane-ethylene oxide (PEO-PDMS) were either directly mixed into the base resin pre-printing (MPC and PEGMA) or infused into the base resin post-printing (PEO-PDMS). MPC,a zwitterionic methacrylate, and PEGMA, a non-ionic methacrylate polymer, are formed by the removal of a proton from the carboxylic acid group of methacrylic acid.²⁷ These methacrylated molecules are known to bind to polymer matrices, thereby permeating the resulting polymer with hydrophilic properties. PEO-PDMS does not bind into the resin polymer matrix. Instead, PEO-PDMS is known to migrate to the surface of interfaces when exposed to a polar compound (e.g. water), where the polar portion of the molecule is presented at the surface and the nonpolar portion remains within the PDMS matrix.²⁸ Initially, PEO-PDMS was mixed directly into the base resin, but was leached out of the polymer during the mandatory 72 h ethanol extraction required to remove cytotoxic uncured oligomers.¹⁷ Thus, a post-extraction infusion strategy was implemented that leveraged ethanol swelling for surface incorporation while preserving device biocompatibility. These methods for integrating hydrophilic molecules into the PDMS resin streamline the manufacturing process without necessitating extensive post-printing modifications.

EXPERIMENTAL

Resin Formulation

A 3D-printable UV resin for vat photopolymerization previously developed in our laboratory²⁹ was modulated by the addition of 2-Methacryloyloxyethyl phosphorylcholine (MPC), Poly(ethylene glycol) methacrylate (PEGMA), and dimethylsiloxane-[60-70% ethylene oxide] (PEO-PDMS) in order to create the experimental groups presented in this paper. The base resin is composed of 1) the PDMS telechelic polymer DMS-R22 (diluent) (Gelest, Inc. Morrisville, PA); 2) 7–9% methacryloxypropyl)methylsiloxane] dimethylsiloxane side-chain co-polymer RMS-083 (Gelest, Inc. Morrisville, PA, USA); 3) Sudan 1 photoabsorber (Sigma-Aldrich, St. Louis, MO, USA); 4) 2,4,6-trimethyl benzoyl diphenylphosphine oxide (TPO-L; photoinitiator) (PL Industries of Esstech, Inc. Essington, PA, USA); and 5) 2-isopropylthioxanthone (ITX; photosensitizer) (VWR International (Radnor, PA, USA). The weight fraction (w/w%) of each listed base molecule was maintained at 80% DMS-R22, 0.8% TPO-L, 0.09% Sudan I, and 0.4% 2-Isopropylthiocanthone (ITX).²⁶ The weight fraction of RMS-083 only was adjusted according to the amount of hydrophilic molecule added to the base resin.

Resin Dose Curves and Curing Dynamics

The resin curing dynamics for all initial resins (Base Resin and six experimental groups) were determined by placing uncured resin on a glass slide and exposing it to a small circle of light (5.2 mm²) from the printer (MAX X27 UV) at 15 mW/m² at various time points (1, 2, 3, 5, 7, 9, 11, 13, 15, 20, 25, 30, 40, and 60 seconds). The excess uncured resin was rinsed from the glass slide with ACS-grade ≥ 99.5% isopropyl alcohol (IPA) (LabChem, Zelienople, PA, USA) purchased from Fisher Scientific Company (Hampton, New Hampshire, USA). The thickness of the cured resin was measured by taking side-view images of the cured resin spot with an AM413T Dino-Lite Digital Microscope using DinoCapture 2.0 software (Dunwell Tech, Inc., Torrance, CA, USA; camera resolution was ± 3 μm).

Resins were cured and measured at least three separate times with triplicate measurements taken for each thickness (n ≥ 9). Simple linear regressions were run in GraphPad Prism (Figure S1) to determine the slope, x-intercepts, and standard error of these curves.³⁰ Cure energy (D_c) and resulting cure height (h_a) were input into a material file for each resin for printing with an Asiga MAX X27 UV printer.

Preliminary Resin Modifications and Surface Testing

In order to narrow down the number of experimental groups to test in the in vitro blood flow studies (to ensure feasibility and completion), a larger number of preliminary resin modifications were implemented and tested using contact angle analysis (to determine impact on surface hydrophilicity). For these preliminary studies, the PDMS base resin was modified in seven different ways by: 1) adding 1% methacrylated phosphorylcholine (MPC) to the base resin; 2) adding 1.8% Poly(ethylene glycol) methacrylate (PEGMA) to the base resin; 3) adding 3% PEGMA to the base resin; 4) adding 2% dimethylsiloxane- ethylene oxide (PEO-PDMS) to the base resin; 5) adding a combination of 3% PEGMA and 2% PEO-PDMS to the base resin; 6) adding a combination of 1.8% PEGMA and 1.2% PEO-PDMS to the base resin; and 7) infusing the neat PDMS devices with 2% dimethylsiloxane-[60-70% ethylene oxide] (PEO-PDMS) in ethanol post-printing. A chart denoting which formulations were tested and at what stage is included below (Figure 1). More rationale for the choice of these experimental groups is provided in Supplementary Information.

Figure 1.

Experimental workflow showing resin formulations tested at each experimental stage.

All resin ingredients were measured in 50 g batches on a Quintix 125D-1S Semi-Micro Balance (Sartorius Lab Instruments GmbH & Co. KG, Göttingen, Germany), mixed on a VMS-C7 S1 hot plate (VWR International, Radnor, PA, USA) at 70 degrees Celsius for 2 hours, sonicated with a Q700 sonicator (Qsonica LLC, Newtown, CT, USA) to ensure uniform mixing and particle size reduction, and analyzed under a Dino Microscope at 150x magnification to assess particle size <20 μm to ensure resin homogeneity. All preliminary and final prints were printed on an Asiga MAX X27 UV printer and post-processed to remove any uncured resin, unreacted compounds, and excess photoabsorbers (Sudan 1 and ITX). Post-processing included sonicating printed parts and devices in isopropyl alcohol and then soaking for 72 hours in Ethyl alcohol. A prior study demonstrated that 72 hours of soaking was sufficient in ensuring that all uncured and unreacted groups were leached out of the resin, proving to significantly increase cell viability and proliferation post extraction.²⁹

The 2% PEO-PDMS infusion experimental group was prepared after post-processing the printed material by soaking a printed Base Resin device in 2% PEO-PDMS w/w% dissolved in ethanol for 48 hours. Prior to testing, the 2% PEO-PDMS infused surfaces, along with all other devices for the sake of consistency, were submerged in DI water for 1 hour to permit the PEO-PDMS molecules to migrate to the surface. Figure 2 depicts chemical structures for the Base Resin and additional molecules utilized in this study.

Figure 2:

Schematic demonstrating chemical composition of Base Resin and additional molecules (MPC, PEGMA, and PEO-PDMS).

Contact Angle Testing

Films for contact angle testing were created in a similar fashion to spot testing. Resin was dispensed onto a glass slide and exposed to UV light for 60 seconds to create 200 μm thick, 6 mm diameter films. All films were soaked in ethanol for 48 hours and submerged in DI water for 1 hour prior to testing. The contact angle of the following preliminary resin formulations was determined: Base Resin, 1% MPC, 3% PEGMA, 2% PEO-PDMS, 3% PEGMA+2% PEO-PDMS, and 1.8% PEGMA+1.2% PEO-PDMS. Static contact angles were measured using the sessile drop method (deionized water, 8 μL drop size). Three drops of deionized water were allowed to sit on the resin film for up to 90 minutes. Images with the droplets were taken using a custom-built goniometer with an AmScope UCMOS series microscope camera (Figure S2).³¹ Images were taken every 10 seconds for 2 minutes, and then at 3, 5, 7, 10, 15, 20, 30, 40, 50, 60, 75, and 90 minutes. Finally, the images were analyzed using the contact angle plugin for ImageJ.^32,33

Microfluidic Device Design

Computer aided drawing (CAD) designs for the biomimetic, microfluidic branching devices for flow-through testing were creating using SOLIDWORKS (Dassault Systems, Waltham, MA, USA). Designs were implemented according to Murray’s Law, which describes the size, scaling, and branching of the natural vasculature.³⁴ The bifurcation angle of each vessel branching was 37.5. The 3D printed microfluidic devices contained 16 channels each and were printed 4 at a time per glass slide. The smallest blood flow channels (i.e. “capillaries”) were designed nominally 300 μm tall, 324 μm wide, 10 mm long, and separated laterally 324 μm apart. The flow network bifurcated three times at both the inlet and the outlet sides of the device. See Supplementary Information for relevant formulas, equations, and CAD deign schematics for printed microfluidic device (Figure S4).

3D Printing

All devices, spots, and films were printed using the Asiga MAX X27 UV printer (Asiga, Alexandria, Australia 2015). The printer has a 385 nm light source (wavelength range 370-400 nm), with an X and Y pixel resolution of 27 μm, and a Z (vertical) resolution of 1 μm and uses digital light processing (DLP) technology. 3D models were generated in SOLIDWORKS (Dassault Systems, Waltham, MA, USA) and exported to STL file format. Asiga Composer Software version 1.3.8 (Asiga, Alexandria, Australia 2015) was the interface used for all STL files, for establishing print parameters, and for initiating prints. Devices were printed with the following parameters: layer thickness of 20 μm, light intensity of 5.00 mJ/cm², burn-in layer exposure time of 48.056 s, non-burn-in layer exposure time of 27.851 s (10% over exposure), and wait time after each exposure of 30 s (to permit time for the resin to polymerize).¹⁷

Glass slides, first silanized with 3-(trimethoxysilyl) propyl methacrylate (Sigma Aldrich, St. Louis, MO, USA), were adhered to the build platform using a UV epoxy (Proto Glass, Proto Products, Ashland City, TN, USA) before the start of each device print in order to provide a pristine, smooth printing surface.³⁵ The build platform was calibrated and levelled with the glass slide attached prior to printing.¹

Microfluidic networks were created with the Base PDMS Resin, FunToDo Nano Clear 3D printing resin (positive control), and three experimental resins: 1% MPC (mixed into base resin), 1.8% PEGMA (mixed into base resin), and 2% PEO-PDMS Infusion (applied post printing), onto silanized glass slides with an Asiga MAX 27X UV printer.

Microfluidic Device Surface Roughness Measurement

Surface roughness values of each 3D printed resin formulation were obtained using an Olympus LEXT OLS4000 laser confocal microscope. A 20× objective lens was employed to acquire 640 μm × 640 μm images of each surface. To optimize laser imaging, the top and bottom ranges for the topographical scans were adjusted to minimize background noise during 3D imaging. Brightness levels were calibrated to prevent oversaturation, as excessive saturation reduces the accuracy of topographical imaging and obscures variations in the signal.

Triplicate measurements were obtained from three distinct regions of the top surface of each 3D printed surface. The acquired images were analyzed using the surface roughness measurement tool within the LEXT application. Height was selected as the primary analysis parameter. A Gaussian filter with a cutoff value set at 80 μm was applied post-measurement according to ISO 4288. The Gaussian filter was used to eliminate form and waviness components, ensuring that any surface curvature on the devices did not influence the roughness measurement. The data were subsequently compiled into a report displaying the following analysis parameters: S_q (root mean square height), S_ku (kurtosis), S_v (maximum pit height), S_a (arithmetical mean height), S_sk (skewness), S_p (maximum peak height), and S_z (maximum pit height).

Microfluidic Device Post-Processing, Assembly, and Clearing

3D printed flow-through devices were removed from the build platform and sonicated for 10 minutes in IPA using a Cole Parmer Ultrasonic cleaner (model 08849-00; purchased from Cole Parmer, Vernon Hills, IL, USA) to wash the exterior of the device of uncured resin. Before assembly and clearing, devices were extracted in ethanol for 72 hours. Ethanol was changed every 24 hours to ensure efficient extraction and removal of uncured resin. These microfluidic devices were assembled by attaching 0.062 in ID x 0.125 in OD silastic tubing (LIVEO^™ Silicone Laboratory Tubing, Avantor, Radnor, PA, USA) to the inlet and outlet ports using DOWSIL 3140 MIL-A-46146 RTV Silicone Conformal Coating adhesive (Ellsworth Adhesives, Germantown, WI, USA). After the adhesive dried, parts were flushed with ethanol to remove uncured resin from the internal channels. For flushing, parts were first placed on a continuous flow circuit containing ethanol and driven with a peristaltic roller pump (Masterflex 7523-80 Peristaltic Pump, Cole Parmer, Vernon Hills, IL, USA) for at least 48 hours. Ethanol was pushed through the device starting at 0.2 mL/min and slowly ramped up by 0.1 mL/min every 10-20 minutes until fluid flow of at least 1.5 mL/min was achieved. After pump clearing, channels were hand cleared using a 1/16” 30 mL BD Luer-Lok^™ Syringe (Grainger, Lake Forest, IL, USA) filled with ethanol. Pressure was applied to the syringe to push the ethanol through the devices slowly with the goal of flushing out any excess resin. Each device was cleared in this way under constant pressure until all channels were cleared.

Dynamic Blood Flow Testing of Microfluidic Networks

After assessing the results from the contact angle testing, poor-performing resin experimental groups were eliminated from further dynamic testing. The following experimental groups alongside the Base Resin and a commercially available resin, FunToDo Nano Clear (Ergometa, Irvine, CA, USA; positive control), were used to create biomimetic flow networks for blood testing: 1) 1% Methacrylated phosphorylcholine (MPC) added to the base resin; 2) 1.8% Poly(ethylene glycol) methacrylate (PEGMA) added to the base resin; or 3) base PDMS resin devices infused with 2% dimethylsiloxane-[60-70% ethylene oxide] (PEO-PDMS) in ethanol post-printing.

Biomimetic microfluidic flow networks were tested for coagulation with freshly drawn ovine whole blood treated with fluorescent cell stain. Fresh ovine blood was obtained via acute venous puncture (new venipuncture per testing day) into syringes containing 3.8% w/v Sodium Citrate anticoagulant (1 μL 0.129 M citrate/mL blood) from a single live sheep, rather than chronic catheterization or indwelling ports. This approach ensured high-quality, minimally manipulated whole blood with native coagulation competence for dynamic testing, avoiding catheter-induced platelet activation or port-related contamination. All blood was drawn from the same sheep, ensuring consistency in the biological source. However, due to the practical constraints of testing over 60 devices, different devices were tested on different days across 4 months. This was necessary as it was not feasible to test all devices in a single session or draw that much blood from the animal at once.

First, the heparin dose required to achieve an ACT of 200-250 s was determined using small 1 mL aliquots of blood. Next, DiI (1,1’-dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine perchlorate; ThermoFisher Scientific, Waltham, MA, USA) fluorescent dye (Excitation: 549 nm; Emission: 665 nm) and Heparin were added directly to blood and incubated at room temperature in the dark for 20 minutes before reversing citrate with calcium gluconate. Before testing, all devices were immersed in deionized (DI) water for 1 hour. Devices (n ≥ 12 per group) were then exposed to blood flow for 10 minutes at 0.8 mL/min (shear of 0.771 Pa) on a peristaltic roller pump (Masterflex 7523-80 Peristaltic Pump Cole Parmer, Vernon Hills, IL, USA), flushed with 50 mL phosphate-buffered saline, fixed with 2% Paraformaldehyde, then evaluated for clotting via fluorescent confocal microscopy (Leica Stellaris 5), clotting area analyses, and flow cytometry. 18 μm Hemo-Nate Blood Filters (Utah Medical Products Inc 4020009, West Midvale, UT, USA) were used upstream of the inlet of each device to filter any microemboli already in the blood.

During dynamic blood testing, pressure data were recorded at various time intervals. Pressure drop across the device during flow testing was monitored using a clinical pressure sensor (Hewlett Packard, M1176A, Model66). The average percent change in pressure was computed by comparing readings at time 0 with those at 5 minutes. These measurements directly correlate with the extent of clotting accumulation within the channels, where higher pressure readings indicate more severe thrombosis.

Flow Cytometry of Microfluidic Blood Flow Samples to Monitor Platelet Sequestering

Black 1.5 mL Eppendorf tubes were prepared by adding in 55 μL of sodium citrate as an anticoagulant. Inlet and outlet samples of 0.5 mL were taken at 5 minutes timepoints during dynamic flow-through blood testing for each device. After testing was concluded, 20 μL of the anticoagulated blood was transferred to new Eppendorf tubes containing 980 μL Hanks’ Balanced Salt Solution (HBSS) (Gibco, ThermoFisher Scientific, Waltham, MA, USA) each and vortexed for a few seconds. 100 μL from the whole blood HBSS dilution tube were then transferred into six flow sample tubes. Tubes (1)-(4) were inlet samples (use 400 μL total) and tubes (5)-(6) were outlet samples (use 200 μL total). 4 μL of phosphate-buffered saline was pipetted into tubes (1) (3) (5) followed by mixing on the vortex. 4 μL of collagen, meant to activate the platelets and amplify the signal, were pipetted into tubes (2) (4) (6) followed by mixing on the vortex. Samples were then incubated for 2 minutes at room temperature. After incubating, 20 μL Mouse Anti-Immunoglobulin G (IgG) (negative control) Antibody was transferred into tubes (1) (2) followed by vortexing. Next, 20 μL Anti-CD62P PE Alexa Fluor 647 was pipetted into sample tubes (3) (4) (5) (6) followed by vortexing. Samples were incubated for 15 minutes in the dark at room temperature. Finally, 700 μL of 1% formalin was pipetted into each tube and vortexed in order to fix the platelet stain.

200 μL of each flow sample was then transferred into a labelled 96-well plate and placed in a −20°C fridge. All samples were analyzed and tested within 5 days on the Cytek Aurora Spectral Analyzer at the University of Michigan North Campus Research Complex Flow Core. The Aurora uses information from every single detector to assess cells. The raw .fcs files report information from all ~60 detectors (UV1-16, V1-16, B1-14, Yg1-10, R1-8). The unmixed .fcs files had been run through the software’s algorithm to disentangle the different emission spectra of stain colors so that only the data for the intended scatter parameters and fluorochromes in the experiment were visualized. Gating strategies and specific gating methods used are included in the Supplementary Information (Figure S6).

Fluorescent Imaging of Microfluidic Devices

All devices were imaged for fluorescence intensity via confocal microscopy on a Leica Stellaris 5 Microscope. Imaging was carried out with a 2.5X objective, an excitation wavelength of 549 nm, and a detection range of 410 to 850 nm. Each entire device was imaged automatic stepping; focus was set at 9 points across the device to ensure accurate intensity readout.

Analysis of Fluorescent Imaging

Captured images were analyzed for fluorescence intensity and percent area clotting using ImageJ. To determine percent area clotting, images were converted to 16-bit grayscale before being converted to binary, with black representing blood cells and white representing the background.³⁶ Clotting area % was normalized to the amount of actual blood exposure by dividing the ImageJ clotting area value by the actual channel area exposed to blood flow during testing (a small fraction of the channels was blocked and did not permit blood flow during testing).

In order to determine fluorescence intensity for each device, ImageJ was used to take triplicate measurements of the total fluorescence intensity over the device area minus the background fluorescence intensity. Background fluorescence intensity was calculated as the average intensity over 3 separate background regions across the device. The corrected total cell fluorescence (CTCF) was calculated using the following formula: CTCF = Integrated Density – (Area of Selected Cell x Mean Fluorescence of Background readings. This process was repeated for each device.

Statistical Analysis

Statistical significance between groups for flow cytometry, fluorescence intensity, percent pressure increase, and percent area clotting was determined. Group differences were assessed using one-way ANOVA (Excel, α = 0.05) for multiple-group comparisons (≥3 resin formulations) and one-sided unequal variance t-tests (Excel) for pairwise comparisons between two specific groups (e.g., experimental vs. Base PDMS). All comparisons yielding p < 0.05 were deemed significant.

RESULTS AND DISCUSSION

Resin Dose Curves and Curing Dynamics

Curing depth versus cure energy plots (i.e. resin dose curves) are provided in Supplementary Information (Figure S1). Penetration depth and the critical dose required for polymerization to form a non-flowable material (Dc) were calculated for each resin via linear regression of the resin dose curves and are provided in Table 1.

Table 1.

Penetration depth (ha) and cure energy (Dc) for each resin modification, calculated via linear regression of each respective resin dose curve.

Formula	ha (um)	Dc (mJ/cm2)
Base Formula	49.4	5.3
1.8% PEGMA	52.7	2
3% PEGMA	46.4	1.8
1% MPC	61.4	5.7
2% PEO-PDMS	51.5	5.1
1.8%PEGMA, 1.2%PEO-PDMS	44.4	1.8
3%PEGMA, 2%PEO-PDMS	58.4	2.8

Contact Angle Testing

Measured water contact angle over time for five tested resin formulations are shown in Figure 3. At 90 minutes, the average triplicate contact angle was 63.1° for Base Resin; 39.2° for 1% MPC; 52.3 for 1.8% PEGMA; 53.3° for 1.8% PEGMA+1.2% PEO-PDMS; 51.3° for 3% PEGMA; 53.2° for 3% PEGMA+2% PEO-PDMS; 43.5° for 2% PEO-PDMS No Soak; 43.3° for 2% PEO-PDMS Infusion; and 58.0° for 2% PEO-PDMS 72 Hour Soak. Of note, PEO-PDMS resins had a significantly higher contact angle after soaking in ethanol indicating that ethanol soaking was leeching PEO-PDMS out of the resin films. Further, the unsoaked PEO-PDMS film displayed a rapid decrease in contact angle and maintained the smallest contact angle throughout most of the testing period. Based on this data, it was decided that PEO-PDMS would be infused into the material post printing for the microfluidic network devices in order to avoid loss of the PEO-PDMS during ethanol soaking, a crucial manufacturing step used to reduce cytotoxicity.³⁷ This 2% PEO-PDMS Infusion group started out at a higher contact angle than the 2% PEO-PDMS No Soak group, but had a similar contact angle at the 90 min time point. 1.8% PEGMA+1.2% PEO-PDMS and 3% PEGMA+2% PEO-PDMS performed similarly at the 90 minute time point, informing the decision to eliminate the 3% PEGMA+2% PEO-PDMS group in further studies in favor of less added in molecules to the base resin. 1% MPC was the top performing modification at the 90 minutes timepoint with 2% PEO-PDMS Infusion as second best, solidifying both as experimental groups for dynamic blood flow testing.

Figure 3.

(A) Contact angle analysis over time for the following resin formulas (n=3 per resin group): 3% PEGMA, 2% PEO-PDMS, 1% MPC, 3% PEGMA+2% PEO-PDMS, 1.8% PEGMA+1.2% PEO-PDMS, Base Resin, and Sylgard 184. (B) Closeup of 5 minutes of contact angle testing.

Microfluidic Device Design & 3D Printing

True (as-printed) channel dimensions of PDMS resin devices were determined via microscopic examination to be approximately 200 μm tall and 240 μm wide due to UV light scattering effects. However, realized channel dimensions for FTD Nano Clear devices remained congruous with drawn dimensions of 300 μm tall and 340 μm wide. All modified and unmodified PDMS resin devices took ~6 hours to print. FTD Nano Clear devices printed in 50 minutes. Extended (>10 hr). EtOH soaking of FTD Nano Clear devices resulted in deterioration such that the solvent appeared to be dissolving the device. Thus, these devices were not soaked prior to dynamic blood flow testing, but simply cleared by flowing EtOH through the channels, then dried with compressed air. This was not the case for PDMS devices, which remained structurally unaffected.

Microfluidic Device Surface Roughness Measurement

Surface Roughness measurements and images were collected for all devices using an Olympus OLS 4000 LEXT laser confocal microscope. Measurements indicate no statistically significant difference in surface roughness between devices (Figure 4), suggesting that the observed variations between experimental groups are primarily due to differences in resin formulation rather than surface topology. LEXT surface images included in Supplementary Information (Figure S7).

Figure 4.

Surface roughness data comparing control (FTD NanoClear), Base PDMS Resin, and three experimental groups. S_q refers to root mean square height in μm, representing the average deviation of surface heights from the mean plane

Dynamic Blood Flow Testing of Microfluidic Networks

PDMS resin (modified and unmodified) and control FTD Nano Clear devices (n ≥ 12 for each testing group) were successfully tested with anticoagulated whole ovine blood at physiologic (venous) blood flow rates for 10 min. Figure 5 shows photographs of representative devices during and after blood testing and flushing. Cell deposition quantified by fluorescence intensity (Figure 6A) showed a 54.2%, 69.2%, and 88.3% decrease compared to FTD Nano Clear and a 40.1%, 59.7%, and 84.6% decrease compared to the unmodified PDMS base resin for the 1% MPC resin, 1.8% PEGMA resin, and 2% PEO-PDMS infusion, respectively. One-way ANOVA revealed a significant effect of resin formulation on fluorescent intensity, F(4, 148) = 14.57, p < 0.0001. Representative fluorescent images used for quantifying fluorescence intensity can be found in Supplementary Information (Figure S8). Clotting area analysis (Figure 6B) determined a 42.5%, 58.3%, and 65.2% decrease between positive control resin (FTD Nano Clear) and 1% MPC, 1.8% PEGMA, and 2% PEO-PDMS infusion, respectively, and an 11.4%, 49.1%, and 57.5% decrease relative to the unmodified base PDMS resin. Similarly to fluorescence intensity results, one-way ANOVA revealed a significant effect of resin formulation on percent area clotting, F(4, 80) = 10.41, p < 0.0001. Percent clotting and fluorescence intensity analyses demonstrated that PDMS resin devices with 1% MPC, 1.8% PEGMA, and 2% PEO-PDMS Infusion all significantly reduced clotting and thrombosis, with 2% PEO-PDMS performing the best.

Figure 5.

3D-printed capillary devices during (top) and after blood exposure (bottom) (A) Positive control FTD Nano Clear, (B) Base PDMS Resin, (C) 1% MPC, (D) 1.8% PEGMA, and (E) 2% PEG-PDMS infusion.

Figure 6.

(A) Average fluorescence intensity and (B) Percent area clotting data comparing control (FTD Nano Clear), Base PDMS Resin, and three experimental groups. (* = p < 0.05, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001)

Figure 7 illustrates the percent increase in pressure drop over time (0 vs 5 min time points) within the microfluidic capillary channels of each resin variant (n ≥ 12 per resin group). Across all devices, the average percent pressure increase was noted as follows: 178.5%, 135.5%, 68.0%, 41.8%, and 80.7% for FTD Nano Clear (positive control), Base Resin, 1% MPC, 2% PEO-PDMS Infusion, and 1.8% PEGMA, respectively. Notably, the 2% PEO-PDMS Infusion group exhibited a statistically significant lower pressure change compared to both FTD Nano Clear and Base Resin, with p-values of 0.0428 and 0.0341, respectively. In conjunction with metrics such as percent area clotting and fluorescence intensity, the pressure variation further underscores the superiority of the 2% PEO-PDMS Infusion resin as the top-performing resin modification.

Figure 7.

Average percentage pressure increase comparing control (FTD Nano Clear), Base PDMS Resin, and three experimental groups with raw plotted data.

Platelet Sequestering

Flow cytometry data revealed that platelet activation (measured via Anti-CD62P PE Alexa Fluor 647) was the same in untreated samples and samples treated with collagen (a platelet activator), indicated that all platelets were activated prior to entering the device under all experimental conditions (Figure S5). As platelets were activated prior to entering each device, inlet compared to outlet platelet numbers served as a measure of the retention of activated platelets by each experimental group. These results are shown in Figure 8 as percentage decrease in inlet to outlet platelets. The number of platelets decreased from inlet to outlet in all devices, demonstrating that some platelets adhered to the interior of the device, forming residual clots. Average percent decrease in platelet counts measured 55.0%, 41.2%, 42.0%, 22.9%, and 40.3% for FTD Nano Clear (positive control), Base Resin, 1% MPC, 2% PEO-PDMS Infusion, and 1.8% PEGMA, respectively. It is noteworthy that there was a statistically significant decrease in platelet events between 2% PEO-PDMS Infusion and FTD NanoClear (p < 0.001), suggesting reduced clotting and platelet retention within the device channels, consistent with previously established patterns in the pressure drop, percent clotting, and fluorescent intensity data.

Figure 8.

Percent decrease in platelet count at inlet and outlet comparing control (FTD NanoClear), Base PDMS Resin, and three experimental groups.

Discussion

Current clinical hollow-fiber oxygenators pose inherent challenges and limitations as they are particularly susceptible to clotting, resulting in the need for systemic anticoagulation, and thus heightening the risk of bleeding events. These ECMO devices necessitate around the clock monitoring and limit patient mobility and quality of life. The use of microfluidic artificial lungs (μALs) and 3D printing technology has grown tremendously and shows strong potential to combat some of the challenges that come with hollow-fiber devices. However, there are still limitations that persist as there is a lack of biocompatible materials for device fabrication.

In order to combat this limitation and overcome the shortcomings of existing oxygenation technology and nonnative 3D printing materials, this paper aimed to formulate a biocompatible, hydrophilic, 3D-printable, high-resolution, and gas permeable resin. By modifying an established custom-made polydimethylsiloxane (PDMS) based 3D-printable resin via the incorporation of hydrophilic molecules such as MPC, PEGMA, and PEO-PDMS, hemocompatibility was significantly enhanced and clotting significantly decreased during in vitro tests with whole blood flow. These modulations combat the natural coagulation response to foreign materials by increasing surface hydrophilicity, thereby decreasing blood protein deposition and slowing the progression of clotting.

Preliminary Resin Modifications and Surface Testing

Preliminary investigations focused on assessing the hydrophilic properties of various resin formulations. In the initial formulation phase of the study, it was found that there were challenges with resin formula stability when more than 3-5% of the base formula consisted of additives. There was precipitate formation when the content of additives was too high, leading to the decision to limit hydrophilic molecule addition to 5%. This informed the creation of the 3% PEGMA+2% PEO-PDMS group (5% total additions to base resin) and the 1.8% PEGMA+1.2% PEO-PDMS group (3% total addition) to test which formula was most successful. It would later be determined that PEO-PDMS was leached out of the resin during extraction phases and that 1.8% PEGMA was superior to 3% PEGMA in terms of contact angle, suggesting that lower w/w% additions provide a more stable resin and resulting polymer. Additionally, it was found that concentrations of MPC above 1% began to form micelles in liquid resin which resulted in the chosen addition.

Contact angle measurements provided initial indications of surface characteristics, with notable differences observed between formulations containing PEO-PDMS and those without. Note that testing conditions for each film was controlled, so any evaporation would be similar for all resins and testing groups. Additionally, volume estimates of water droplets from contact angle images determined that there was no significant water loss (Figure S3), indicating that decrease in contact angle was owed to the hydrophilicity of the resin surface rather than evaporation. The leaching of unbound PEO-PDMS from post-processed devices that underwent a solvent extraction phase highlighted the importance of optimizing resin compositions and post-processing to ensure long-term stability and functionality through all potential processing steps and use cases.

The increase in contact angle of resins containing PEO-PDMS after an EtOH soak indicated that solvent soaking was leeching away PEO-PDMS from the resin. PEO-PDMS is a non-methacrylated amphiphilic compound that does not covalently bind into the polymer backbone but will migrate to the surface when exposed to a polar solute such as water.¹⁷ It became evident that when devices that were printed with PEO-PDMS added into the resin formula were post-processed in ethanol, the PEO-PDMS that was unbound was leached out, reducing or eliminating the intended hydrophilic properties (Figure 3). Due to this discovery, a 2% PEO-PDMS infusion group in which PEO-PDMS was infused into the device post printing was implemented and investigated during dynamic blood flow-through testing. Given that PEO-PDMS is not methacrylated and does not bind into the polymer matrix, the base PDMS resin devices were able to take on the hydrophilic properties of the PEO-PDMS simply via infusion. By setting up a concentration gradient and solubilizing PEO-PDMS in the ethanol soak after printing, the base resin was imbued and infused with the PEO-PDMS, effectively increasing hydrophilic characteristics.

The described PEO-PDMS infusion technique exploits ethanol-induced PDMS swelling, facilitating diffusion of amphiphilic macromolecules from the solvent into the polymer matrix, a deswelling/incorporation mechanism analogous to established strategies.³⁷ Optimizing this process has the potential to yield higher PEO concentrations at blood-contacting interfaces (~10x bulk concentration), enhancing localized antifouling efficacy.³⁷ This swelling-deswelling process is thus a promising oxygen-plasma-free route to hemocompatible 3D-printed microchannels, where traditional surface activation strategies remains inaccessible. Future optimization can leverage these findings to propel imbued hydrophilicity in PDMS-based devices and materials. Future studies should validate the long-term durability of this approach under flow.

Previous PDMS modification strategies by our and other research groups deserve discussion in regard to their potential application to 3D printed microfluidic devices. Zhang et al. used a flow through approach to coat pre-assembled 350 x 600 μm PDMS channels with a 1-2 μm layer of polybetaine.³⁸ This approach demonstrated significant reductions in water contact angle, fibrinogen deposition, and clotting in in vitro flow tests with whole bovine blood compared to unmodified PDMS controls, but the coating’s impact on gas permeability was not measured and smaller channel dimensions were not tested. DeMeulemeester et al. demonstrated a PDMS modification strategy that grafted polyethylene glycol and corn trypsin inhibitor (a Factor XII inhibitor) to the surface of PDMS flow chambers, exhibiting long term stability and a reduction in contact angle and clotting compared to uncoated controls, but required a plasma surface activaction step that is not compatible with enclosed channels formed via 3D printing. Jukarainen et al. investigated the surface adsorption from solution of PDMS-PEO copolymers on polystyrene microspheres, demonstrating reduced fibrinogen deposition. The work was not applied to enclosed channels, but potentially would work in that application.⁴⁰

The plasma-free PEO-PDMS infusion method described here enables the post-fabrication modification of monolithic enclosed channels (200×240 μm capillaries) after a solvent extraction, a step necessary to remove unreacted and potentially cytotoxic molecules remaining in the polymer as demonstrated previously.²⁹ Under physiological venous shear (0.77 Pa) testing with ovine whole blood, it delivered significant clotting reduction versus the base resin.

Despite the promising performance of the 2% PEO-PDMS infusion method, physiosorbed amphiphiles risk delamination under prolonged flow, a recognized limitation of non-covalent surface modification strategies. Here, the PEO-PDMS infusion was exposed to 1 h of DI water (to permit the PEO-PDMS to migrate to the surface) and 10 min of testing with whole blood, demonstrating short-term stability. Previous studies demonstrated that the PEO-PDMS copolymer mixed into Sylgard 184 is shelf stable for up to 20 months of ambient storage.^19,41 However, long-term stability under prolonged blood exposure remains untested. Further, durability comparisons versus covalently bound MPC/PEGMA were not performed in this study. Such studies will need to be completed in the future as clinical oxygenators require hours to weeks of operation under continuous blood flow.

Contact Angle surface testing aided in the elimination of resin formulations 3% PEGMA, 3% PEGMA+2% PEO-PDMS, and 1.8% PEGMA+1.2% PEO-PDMS from blood testing. The combined resin groups 1.8% PEGMA+1.2% PEO-PDMS and 3% PEGMA+2% PEO-PDMS performed similarly to each other and similar or worse than the 1.8% PEGMA and 3% PEGMA resin groups, and were eliminated in favor of formulations with fewer added in molecules. Additionally, 1.8% PEGMA was insignificantly different from 3% PEGMA and was thus favored for the same reason. Thus, 3% PEGMA, 3% PEGMA+2% PEO-PDMS, and 1.8% PEGMA+1.2% PEO-PDMS resin formulations were eliminated prior to dynamic blood flow testing. Contact angles for several formulations, particularly Sylgard 184, appeared to exhibit a continued decrease beyond the 90 min testing time (Figure 3). This likely reflects dynamic molecular reorganization within PDMS networks, where unbound low-molecular-weight species migrate to the air-water interface surface methyl groups rearrange, and/or water slowly absorbs into the hydrophobic matrix. This behavior aligns with PDMS's well-documented elastic surface restructuring, where contact angles can evolve over hours-days as equilibrium interfacial composition establishes.⁴² This phenomenon is complicated, however, by that fact that each film underwent an ethanol soak prior to testing, which should have removed unbound molecules from the polymer. It is possible that a small number of unbound molecules remain and continue to migrate to the surface of the PDMS over time. It is more likely that, despite its hydrophobicity, a small amount of water continues to absorb into the material, decreasing its hydrophobicity over time. These time-dependent effects highlight the importance of extended measurement protocols for accurate steady-state hydrophilic assessment in future resin optimization.

Microfluidic Device Post-Processing, Assembly, and Clearing

The extraction of unreacted monomers and photo absorbing compounds is a critical postprocessing step to ensure visual clarity, transmission of the cured resin, and mitigate concerns regarding toxicity and biocompatibility in biological applications.⁴³ The extraction of unreacted molecules and toxic photoabsorbing compounds from 3D-printed resins is known to improve the biocompatibility of 3D-printed parts.⁴³ Previous studies have demonstrated that a 72-hour immersion in EtOH, is sufficient to remove all extractable compounds from the printed PDMS parts. This is the reason that all PDMS resin microfluidic devices went through a 72-hr extraction process before clearing and testing. Extraction also increased optical clarity and visualization under confocal microscopy, with the excitation and emission wavelengths chosen to be in the high transparency range (>695 nm) for the extracted material.

Microfluidic Device Design and Printing, & Testing

The fabrication of small-scale biomimetic, microfluidic capillary devices using custom PDMS 3D printing resin, augmented with biocompatible modifications, has the potential to create a new generation of microfluidic membrane oxygenators. The small penetration depth of the custom PDMS resin alongside the demonstrated hemocompatible modifications enabled 3D printing of high-resolution channels.

Anatomy has already determined the principles upon which to design efficient and blood compatible blood distribution networks. It was determined that natural in vivo vasculature follows certain laws aimed at minimizing energy consumption.^44,45 There are also established rules which simultaneously describe both the natural vasculature, and a biomimetic blood distribution network that minimizes energy consumption and coagulation risk.^46–49 Bifurcations, splitting angles, and channel radii were determined by adhering to these laws, rules, and applied equations in order to minimize energy consumption and coagulation risk.

Surface roughness measurements demonstrated that there was no significant difference in surface quality between resin formulations, allowing greater confidence in the ability to compare device formulation and resulting level of thrombogenesis.

Despite identical CAD dimensions, realized channel dimensions differed between PDMS (200×240 μm) and FTD Nano Clear (300×340 μm) devices, yielding modestly higher shear stress in PDMS channels at fixed 0.8 mL/min flow—a condition ordinarily expected to exacerbate, not mitigate, surface-induced thrombosis.⁵⁰ This discrepancy was identified post-blood testing and could not be experimentally corrected. Despite this unfavorable comparison, hydrophilic PDMS modifications substantially outperformed FTD Nano Clear across clotting, fluorescence, pressure, and platelet metrics, suggesting their hemocompatibility advantages are robust and conservative. Future studies will employ rigorously matched post-printing geometries to enable precise shear- and flow-normalized comparisons.

Dynamic Blood Flow Testing of Microfluidic Networks

Dynamic blood flow-through testing provided quantitative and visual data (Figure 5) on clotting area and cell deposition, demonstrating the superior performance of modified resin formulations compared to unmodified and commercially available controls. Minor distal capillary clotting observed in some 2% PEO-PDMS infusion devices (Fig. 5E, bottom-right) likely reflects heterogeneous amphiphile distribution across the branched 16-capillary network due to passive 48 h diffusion limitations from inlet/outlet ports. Despite this, 1% MPC, 1.8% PEGMA, and 2% PEO-PDMS Infusion testing groups performed significantly better than the Base Resin and FTD Nano Clear. All experimental groups demonstrated significantly less clotting and cell deposition with 2% PEO-PDMS Infusion devices performing the best. It is likely that the 2% PEO-PDMS Infusion group performed best due to its mobile nature. PEO-PDMS is a molecule that does not bind to the polymer matrix, enabling a large portion of the infused molecules to migrate from the bulk PDMS to the surface of the material, likely forming a higher density surface coverage than the molecules that are integrated into the polymer backbone and spread throughout the polymer volume.

The shear stress experienced by blood flowing through the capillary channels fell within the lower range of physiological values. In humans, normal physiological blood flow generates a range of shear stress of 0.1–9.5 Pa.^49,51 The maximum shear rate at the wall of the capillary channel devices was determined to be 0.771 Pa for as-drawn channel dimensions. This is within the low end of the physiological range. This was a deliberate choice as 0.8 mL/min permits facile testing (minimal clotting in the tubing, enough flow to permit blood sample collection for flow cytometry) and allows for a larger blood residence time in the device, thereby facilitating clotting for better comparative visualization between resin formulations.

The addition of hydrophilic molecules to the PDMS resin had another unforeseen benefit, mainly that all added hydrophilic molecules greatly increased the ease of removing uncured resin from the capillary channels after printing, with 1.8% PEGMA devices being the most efficient to clear (i.e. required the least pressure to remove uncured resin from the channels).

Channel pressure measurements (Figure 7) closely mirrored the trends observed in percent area clotting and fluorescent intensity, revealing that all three resin groups exhibited lower pressure changes over time compared to FTD Nano Clear and Base Resin, with the 2% PEG-PDMS showing the most favorable performance. However, FTD Nano Clear saw the lowest absolute pressures during testing while exhibiting the highest amount of clotting. These discrepancies signify that observed absolute pressures were a larger indication of channel clarity, size, and clearing rather than hemocompatibility. As previously discussed, FTD Nano Clear devices saw realized channel dimensions closer to the drawn dimensions.

A limitation of this work is that all dynamic blood flow experiments were conducted using blood drawn from a single ovine donor over the four-month testing period. Coagulation profiles can vary substantially between individuals, which may limit the generalizability of these findings. However, a single-donor design was intentionally chosen to avoid inter-donor variability and confounding, which would have necessitated a substantially larger and impractical sample size to detect statistically significant differences between resin formulations. While pooled blood from multiple donors would represent an attractive strategy to better capture biological heterogeneity, there was inconsistent access to a sufficient number of animals to implement this approach over the course of the study. Future work will incorporate pooled or multi-donor blood sources to validate and generalize the hemocompatibility improvements observed here.

Flow Cytometry of Microfluidic Blood Flow Samples to Monitor Platelet Sequestering

Flow cytometry samples were meticulously prepared post-printing and subjected to analysis to assess platelet counts across different resin devices and testing groups. Platelets were fluorescently activated prior to flowing through devices so all collected data quantified the number of platelets as opposed to activation itself. These data suggested trends in the percent of platelets trapped in each device and yielded statistically significant conclusions between FTD NanoClear compared to 2% PEO-PDMS Infusion. There were lower absolute platelet counts witnessed in both inlet and outlet samples for FTD NanoClear compared to all other testing groups. Although blood was taken from the same sheep for all tested microfluidic devices, they were tested with different batches of blood on different days. This potentially led to variation in the blood sample conditions (i.e. sheep stress, platelet count, etc.) across the different weeks testing was performed. Despite lower platelet counts for the blood used to test, FTD NanoClear devices still demonstrated the greatest level of thrombosis compared to other resin testing groups.

Lastly, the protocol used for flow cytometry in this study called for the addition of collagen as a platelet activator. Though effective for this experiment, future experiments will employ convulxin, a highly specific GPVI/GPIb agonist derived from snake venom, to achieve more accurate static platelet activation.⁵² Thrombin could have also been used as it is known to activate Par1 and Par4 receptors under static conditions and is very sensitive in the nanomolar range. Additionally, a platelet marker such as CD41 or CD42 could be added in future experiments as it was difficult to discern between platelet and singlet populations in FlowJo during data analysis.

CONCLUSIONS

This study is a step toward addressing the limitations of clinical hollow-fiber oxygenators through the development of hemocompatible 3D-printable PDMS resins for artificial lung applications. High-resolution biomimetic microfluidic capillary devices were fabricated using resins modified by: (1) pre-printing incorporation of hydrophilic MPC (1 wt%) or PEGMA (1.8 wt%) into the polymer backbone, or (2) post-extraction infusion of PEO-PDMS (2 wt%).

Dynamic testing with freshly drawn ovine whole blood demonstrated significant reductions in clotting area, fluorescence intensity, pressure increase, and platelet sequestration across all modifications, with PEO-PDMS infusion performing best across the various metrics. PEO-PDMS infusion overcame solvent extraction challenges by leveraging ethanol-induced swelling-deswelling for enclosed microchannel modification—a geometry incompatible with plasma activation surface modification strategies. MPC/PEGMA covalent integration facilitated uncured resin clearing while ensuring permanent hydrophilicity. These plasma-free strategies enhance manufacturability and biocompatibility for complex 3D-printed microfluidic geometries, while maintaining channel integrity and dimensions. Overall, the results of this study underscore the importance of meticulous material selection and fabrication techniques in the development of biocompatible microfluidic devices. The significant reductions in clotting and thrombosis observed in experimental groups highlight the potential future clinical relevance of these findings, paving the way for further advancements in hemocompatible materials and microfluidics. The outcomes presented herein provide a step toward a new generation of biomimetic membrane oxygenators, with the potential to improve respiratory support, patient care, and health outcomes in critical care settings.

Supplementary Material

Additional figure S1–S7 as mentioned in the text. Attached as separate document.