Multilayer Soft Lithography for Microfluidic Devices | Stanford Foundry

PDMS (polydimethylsiloxane) dominates microfluidics due to transparency, non-toxicity, gas permeability, and ease of fabrication. Oxygen plasma treatment converts hydrophobic PDMS to hydrophilic surfaces for bonding to glass, silicon, or other PDMS layers—bonds form within ~10 minutes before surface reversion. Anti-stick coatings prevent adhesion during casting, crucial for high-aspect-ratio features. Limitations include softness causing deformation, limited aspect ratios, and ~1% shrinkage during curing. Solutions involve adjusting cross-linker ratios for stiffness, accounting for shrinkage in design, and minimizing fabrication time between steps.

PDMS naturally exhibits hydrophobic properties, causing liquids to bead up on its surface. However, for many biological applications, hydrophilic surfaces are needed to ensure proper protein attachment and cell adhesion. Plasma treatment converts PDMS surfaces to hydrophilic by breaking surface bonds and reacting them with oxygen molecules. The degree of hydrophilicity depends on treatment parameters including oxygen flow rate and exposure duration. This surface modification is essential for successful microcontact printing of biological molecules.

This section covers wafer-level bonding techniques essential for MEMS device fabrication. Field-assisted (anodic) bonding uses electric fields to drive sodium ions in glass and oxygen ions on silicon toward each other, creating a depletion layer. Temperature then causes irreversible bonding between glass and silicon surfaces. Intermediate layer bonding uses adhesives like BCB between substrates for electrical interconnects, though limited for MEMS with microstructures. Direct bonding of polymers to glass/silicon uses plasma treatment to lower surface energy, making surfaces hydrophilic for hydroxy group formation and polycondensation reactions. Oxygen plasma treatment of PDMS replaces methyl groups with hydroxy groups (Si-OH), reducing contact angle from ~109° to ~20°. Bond strength correlates with contact angle—lower angles indicate higher bond strengths from Si-OH linkages formed through polycondensation. Bond strength is measured by pressurizing microchambers until layers separate at critical pressure values. After plasma treatment, PDMS surfaces undergo hydrophobicity recovery as oligomers migrate back to the surface, crowding it with methyl groups. Therefore, bonding between plasma-treated surfaces should be performed immediately after treatment to prevent reduced bond strength.

Creating functional microfluidic devices requires proper surface preparation before assembly. Both PDMS and glass substrates must be cleaned with isopropanol and dried thoroughly using compressed air or nitrogen. Glass slides are baked at 100°C for five minutes to remove moisture. Plasma treatment creates reactive oxygen species on surfaces that enable strong covalent bonding between PDMS and glass. The process requires a clean workstation free of electronics and metal objects that could interfere with plasma generation. Understanding these preparation fundamentals ensures reliable bonding and prevents experimental failures due to poor adhesion or contamination.

Microfluidic cell culture overcomes limitations of conventional approaches by enabling automated, high-throughput experimentation with precise control over cellular microenvironments. Integration of gas exchange, temperature control, and flow management eliminates dependence on external incubators while enabling combinatorial testing of drugs, growth factors, and signaling molecules. Microfluidic tissue engineering recreates physiological conditions through controlled biochemical gradients, fluid shear stress, and spatial organization of cellular components. Organ-on-a-chip platforms such as lung-on-a-chip devices replicate complex tissue interactions including vascularization and mechanical stimulation. Stem cell engineering utilizes BioMEMS to control differentiation and self-renewal through manipulation of biochemical factors, fluid shear stress, cell-ECM interactions, and cell-cell interactions. Microfluidic platforms enable systematic investigation of how transcription factors, epigenetic modifications, and combinatorial ECM composition direct cellular fate decisions. These capabilities enable fundamental research into developmental biology and provide tools for regenerative medicine applications, personalized drug screening, and disease modeling at unprecedented resolution and throughput.

Flow stability critically determines success in biological microfluidic applications. In cell culture experiments, unstable flow patterns fail to mimic in vivo conditions and cause cell damage through uncontrolled shear stresses. Peristaltic pumps produced approximately 40% flow variation, resulting in 30% cell detachment and loss of viability, while pressure-controlled systems maintained stable flow rates preserving cell confluence and morphology. Organ-on-a-chip systems represent the pinnacle of biological microfluidics, defined as microfluidic devices containing living engineered organ substructures in controlled microenvironments that recapitulate in vivo organ dynamics, functionality, and responses with real-time monitoring capabilities. Three core components define these systems: microfluidic interfaces for fluid handling, cell culture chambers for tissue growth, and biosensor/microscopy interfaces for observation. These systems can interconnect multiple organs to recreate physiological interactions impossible with conventional models, offering advantages over small animal models by providing human physiology while retaining in vitro experimental flexibility.

Bio-MEMS integrates micro/nanofabrication technology with life sciences to create devices for biomedical applications, requiring biocompatible materials like polymers (PDMS, PMMA) and silicon for sensors and actuators that can detect biological entities (cells, proteins, DNA), monitor physiological parameters (blood pressure, muscle movement), and enable targeted drug delivery and neural interfacing, with key challenges including biocompatibility, miniaturization, and reliable detection mechanisms.

Developmental processes are controlled by mechanical forces as important as chemical signals and genes. Cells generate mechanical forces through actin cytoskeleton contraction that pull on each other and accumulate extracellular matrix, physically sculpting organs within hours. Cells exist in a state of isometric tension or pre-stress, like a bow and bowstring. Cells can be controlled to adopt different fates by manipulating their degree of stretching: island size determines cell shape, and when growth factors are saturated, cells can be switched between proliferation and apoptosis by controlling stretch. Cells function like tensegrities, where a flexible membrane is stabilized by internal poles (cytoskeleton) connected to external pegs (integrins). Integrins are the primary mechanoreceptors that sense mechanical forces and transmit them across the cell surface. The fastest known mechanotransduction event involves pulling on integrins on endothelial cells with magnetic beads, inducing calcium influx within 5 milliseconds through beta-1 integrin to TRPV4 ion channels. Organs-on-a-chip are microfluidic devices containing living human cells that reconstitute organ-level functions to accelerate drug development and replace animal testing. The lung-on-a-chip consists of three hollow channels separated by a porous membrane with alveolar cells on top and capillary cells on bottom. Cyclic suction creates breathing-like stretching. This platform can culture whole human blood without anticoagulants and mimic complete inflammatory responses where epithelial cells produce cytokines that activate endothelial cells to recruit immune cells. The lung-on-a-chip enabled successful translation of a drug candidate for pulmonary edema, a dose-limiting toxicity of the FDA-approved cancer drug interleukin-2. Bronchial epithelium on chip shows functional cilia and mucociliary clearance at rates matching human lungs. Organ chips reveal synergistic epithelial-endothelial crosstalk during viral infection that is missed in static cultures. COPD patient cells differentiated on chip retain disease phenotypes and show exaggerated inflammatory responses to viral and bacterial stimuli. A cigarette smoking robot delivers controlled smoke directly to air-liquid interfaces, revealing that COPD chips show doubled inflammatory responses to smoke compared to normal chips. Organ chips reveal that biological responses require multi-component interactions, not single factors. When immune cells are added to gut-on-a-chip and exposed to LPS endotoxin, villi blunting and barrier function loss occur. Analysis shows only a specific combination of four cytokines (IL-1 beta, IL-6, IL-8, and TNF-alpha) at specific concentrations can replicate the injury. Lung cancer cells grown on airway chips versus alveolar chips show dramatically different growth patterns, reflecting in vivo behavior where cancer cells grow at the bronchiole-alveolar interface. The microenvironment on the chip, not just the cells themselves, determines tumor growth behavior. Breathing motions suppress tumor growth by 50% and inhibit invasion by 50%, creating a positive feedback loop where tumor growth rigidifies the alveolus, promoting further growth.

Droplet microfluidics creates discrete, uniform droplets (40 pl to 75 fl) at T-junctions, enabling 1,000-20,000 droplets/second with precise chemical payload control. Key applications include: (1) Digital PCR achieving single-copy amplification through encapsulation in inert carriers, (2) Single-cell analysis where picoliter confinement prevents dilution upon lysis, (3) High-throughput screening where half a million experiments can be performed in one hour. These droplet-based approaches solve surface interaction problems while enabling massive parallel experimentation impossible with conventional methods.

Microfluidic droplet generation creates uniform, monodisperse droplets in the femto- to nanoliter scale by using a microfluidic chip with a defined nozzle where two immiscible phases (continuous and dispersed) meet; droplet size is controlled by nozzle dimensions and phase ratio, while throughput depends on flow rates, enabling applications like digital PCR, single-cell analysis, and high-throughput screening without cross-contamination between droplets.

To further increase throughput and reduce batch effects while minimizing contamination and protein losses, researchers developed droplet-based microfluidic approaches. Tiny non-liter-sized droplets are deposited on fluorocarbon-coated glass surfaces, arranged flexibly with groups of 13 droplets accommodating TMT labeling. Perimeter water droplets control local humidity to prevent evaporation losses. The complete workflow—including DMSO lysis, trypsin digestion, TMT labeling, hydroxylamine quenching, and peptide pooling—is performed automatically within individual droplets without moving the glass slide. This enables preparation of hundreds or thousands of single cells in parallel, dramatically reducing sample volumes to nanoliter scales while maintaining analytical precision and throughput.

Microfluidic droplet generators create monodisperse water-in-oil or oil-in-water droplets for single-cell isolation, Drop-seq, and digital PCR applications by precisely controlling flow rates through specialized chips; the system requires priming PTFE tubing with DI water and oil, connecting reservoirs with appropriate adapters, and optimizing pressure settings while ensuring larger particles are directed to central inlets for uniform encapsulation.

Droplet microfluidics generates discrete aqueous droplets suspended in immiscible oils, creating isolated reaction compartments at ultra-high throughput (potentially 8,000 droplets/second). Each droplet acts as an independent experimental unit, enabling single-cell encapsulation and massive parallelization of biochemical assays. Digital PCR (ddPCR) exploits this technology for absolute quantification of target molecules, offering higher sensitivity than conventional qPCR. The confined droplet environment preserves metabolite concentrations, enabling study of metabolic interactions between cells that cannot be maintained in dilute liquid cultures. Applications include gut microbiome cultivation with greater diversity, pathogen-host interaction studies, and combinatorial screening experiments where cell-containing droplets are merged with stimulus-containing droplets in controlled manners.

Hot embossing presses molds into heated polymer substrates to create 2.5D parts from plastics, silicon, nickel, brass, and copper, serving as a bridge between prototyping and mass production. Injection molding injects molten polymer into closed cavities, achieving temperatures up to 140°C and shot weights up to 20g, producing channels as small as 50×50 μm. Cold runner tools with diamond-milled brass inserts achieve optical-quality surfaces. Adhesive bonding applies UV-curable adhesives using capillary forces for uniform coating without air inclusion, achieving pressures exceeding 3 bar.

Micro molding techniques (injection molding, hot embossing, jet molding, replica molding, micro transfer molding) enable mass production of polymer microstructures. The basic process involves dissolving the polymer in solvent, degassing to remove bubbles, depositing onto molds, curing, and demolding. Master molds can be made from polymers (using photolithography or stereolithography), metals (electroplating or LIGA), or silicon (etching). PDMS is particularly popular as a mold material for large-scale production of microreactors, microchips, and fluidic systems due to its transparency, optical properties, and ability to replicate structures down to micron scales.

Injection molding is the primary technology for mass-producing custom plastic parts, fundamentally different from 3D printing which is suited only for prototyping. The process uses a mold with two metal pieces forming a cavity where molten plastic is injected under high pressure. Plastic pellets enter through a hopper, are melted by a heated screw mechanism, and are forced through gates into the mold cavity. This pressure-driven process, developed over 100 years ago, enables filling complex geometries but requires strict adherence to design rules. Understanding this distinction between prototyping and mass production technologies is essential for successful product development.

Microfluidic manufacturing requires understanding multiple technologies and their trade-offs. Mechanical machining offers large dynamic range in structural dimensions and complex 3D geometries but has minimum radius limitations (~25 microns). Lithographic methods provide precise features but limited dimensional range. Injection molding dominates production due to scalability, handling features from macroscopic blisters to sub-micron channels. Critical manufacturing considerations include: PDMS has significant drawbacks for industrial production (high cost, long curing times, molecule absorption, batch inconsistency); back-end processing (assembly, metallization, surface modification, quality control) accounts for approximately 80% of costs; design for manufacturability requires simplifying assembly steps and considering material costs which become proportionally more significant at higher production volumes. Academic development often overlooks these manufacturing realities, leading to costly redesigns when scaling from prototyping to production.

The transition from prototyping to mass production involves moving from 3D nanoprinted components to replicated versions using micro-injection molding in thermoplastic polymers. This approach enables scaling production to thousands, tens of thousands, or even millions of parts at significantly reduced costs. The replication process maintains the optical performance characteristics developed during the prototyping phase.