Mechanical actuation on surface (MAOS) microfluidics: compression for preparation in next-generation sequencing

Abstract

We present mechanical actuation on surface (MAOS), a programmable microfluidic platform that manipulates droplets via localized mechanical compression—eliminating the need for embedded electronics or fixed microchannel geometries. MAOS integrates essential fluidic operations—including droplet transport, magnetic bead-based purification, and thermal cycling—within a benchtop instrument and single-use cartridge. The system accommodates droplet volumes from nL to μL, enabling precise control over sequential biochemical processes. By studying the dynamic behavior of diverse fluids under compression, we identified the key physical variables—surface tension, contact angle, and viscosity—that dictate the onset of droplet motion. We observed sharp transitions in mobility around specific thresholds and validated interfacial encapsulation as a general strategy to overcome resistive pinning. We validated MAOS by first implementing and testing miniaturized next-generation sequencing (NGS) library preparation sub-processes. Magnetic bead-based cleanup showed DNA recovery and fragment size selection comparable to manual methods, and PCR amplification was carried out reliably in low-volume (5 μL) reactions with minimal evaporation. Subsequently, the full NGS library preparation workflow was executed in a plexed format, processing eight libraries in parallel on a single disposable cartridge using as little as 10% of standard reagent volumes. Short- and long-read sequencing outputs from MAOS libraries aligned with manual protocols across key quality metrics. These results establish MAOS as a scalable and user-friendly alternative to conventional microfluidics, suitable for diverse applications in molecular biology, chemistry, and high-throughput workflows.

---

Introduction

Sample preparation remains a key bottleneck in biological and chemical workflows, particularly across multi-omics fields such as genomics, proteomics, and metabolomics.¹ These workflows often involve multistep protocols—including reagent mixing, extraction, amplification, and purification—that are labor-intensive, variable, and rate-limiting. While downstream technologies like next-generation sequencing (NGS) and mass spectrometry have advanced rapidly, upstream processing continues to constrain throughput, reproducibility, and scalability.

Microfluidics has emerged as a potential solution to this challenge. By miniaturizing fluid handling, microfluidic platforms offer precise control over biochemical reactions while reducing reagent consumption and processing time.^2,3 Existing platforms span a wide range of microfluidic actuation modalities, including pressure-driven microchannels, electrowetting-based digital microfluidics (DMF), and centrifugal and acoustic systems.^3–7 Of these, channel-based systems and DMF are the most widely adopted, but both have limitations for sample preparation. Channel-based systems leverage pressure gradients or electrokinetic forces to drive fluid flow through microfabricated networks.⁸ These platforms are well-suited for continuous flow applications but require external pumps and valves, and are prone to clogging, fabrication complexity, and high cost-factors that hinder scalability.^4,9 DMF, in contrast, manipulates discrete droplets (e.g., dispensing, splitting, and merging) via electrowetting actuation, enabling flexible, programmable workflows.^10–12 However, DMF devices rely on patterned electrodes and dielectric layers, which are susceptible to breakdown and require complex, costly fabrication—limiting their utility in disposable or field-deployable formats.¹³

In contrast, conventional liquid handling robots—widely used in high-throughput laboratories—are often bulky, expensive systems that rely heavily on tip-based fluid transfer. These systems consume large numbers of disposable plastic pipette tips to shuttle reagents between functional stations such as heaters, magnetic racks, and shakers. This approach generates significant plastic waste and introduces fragility into the workflow. Tips frequently disengage or misalign, compromising automation and increasing the risk of error. More critically, the time lag involved in physical transfer between modules can perturb time- or temperature-sensitive reactions, introducing variability and inefficiency into complex workflows.

To address these constraints, we developed mechanical actuation on surface (MAOS), a new droplet microfluidic platform based on localized mechanical compression. In MAOS, droplets are confined between two hydrophobic surfaces (with the top surface being elastic), and actuation is achieved by applying localized compression to the top surface to induce motion. As shown in Fig. 1, movement occurs when the driving force exceeds resistive forces, which can be tuned by adding surfactants or by encapsulating droplets in a low-surface-tension fluid such as oil. Importantly, MAOS requires no electrical components or external pumps, offering a mechanically driven alternative for programmable droplet manipulation with minimal instrumentation. Although prior approaches relied on basic droplet manipulation using a single flexible substrate through dynamic surface deformation,^14,15 MAOS leverages vertical confinement and dual-surface interactions to achieve reconfigurable, precise droplet control. It also supports robust execution of key molecular biology operations, including thermal cycling for PCR and magnetic bead-based cleanup. In contrast to conventional liquid handlers, MAOS performs all fluidic operations within a single, integrated plane—eliminating tip-based transfer, reducing plastic waste, and minimizing disruptions to sensitive reactions.

Fig. 1 MAOS-actuated droplet transport under compression. (A) A pure water droplet remains immobile due to high surface tension and contact angle, resulting in F_drive ≪ F_resist. (B) Surfactant addition lowers interfacial tension and contact angle, reducing F_resist; droplet motion is observed when F_drive > F_resist. (C) Oil-encapsulation minimizes interfacial resistance, resulting in smooth, reproducible droplet motion (F_drive ≫ F_resist).

We applied MAOS to NGS library preparation, a typical example of a multistep, resource-intensive workflow. Although sequencing costs have plummeted under $200 in 2024,¹⁶ sample preparation remains expensive and hardware-dependent, often requiring high-end liquid handlers and specialized consumables. MAOS addresses these limitations by enabling scalable, low-volume, and automated reagent handling using a compact benchtop format and single-use cartridges. In this work, we demonstrate that MAOS reliably performs key subprocesses of next-generation sequencing (NGS) workflows—including magnetic bead cleanup and thermal cycling—with performance comparable to manual methods. Building on this, we show that MAOS can prepare libraries for both short- and long-read sequencing platforms. The system processes eight samples in parallel on a single disposable cartridge, using significantly reduced reagent volumes while maintaining data quality. MAOS represents a significant advancement in microfluidic technology and a key step toward overcoming the long-standing bottleneck of sample preparation. Finally, we believe that its versatility extends beyond sequencing to other domains, including proteomics, synthetic biology, and combinatorial chemical synthesis.

Experimental

Reagents and materials

Tween 20, Triton X-100, Pluronic F-127, acetone, acetonitrile, ethanol (EtOH), methanol, 2-propanol, toluene, mineral oil, silicone oil, and a red dye colorant were purchased from Sigma-Aldrich (St. Louis, MO). Filtered and steam-sterilized water was obtained from Growcells (Irvine, CA) and used in the preparation of all experimental solutions. Surfactant solutions were prepared in water at concentrations of 0.05%, 0.5%, and 1.0% (v/v) to assess the effect of surfactant concentration on droplet mobility. Dropgloss (INTEGRA Biosciences, San Francisco, CA; PN: M-03-0001-001-02) was used as the proprietary encapsulation fluid unless otherwise noted. A proprietary elution buffer (INTEGRA Biosciences, PN: M-03-0001-001-03) was used for library retrieval and is optimized for DNA recovery and downstream compatibility. A Twist Library Preparation EF Kit 2.0 (Twist Bioscience, San Francisco, CA; PN: 104207) was used to validate the system's compatibility with automated library preparation workflows and to prepare DNA libraries for short-read sequencing using 50 ng of genomic DNA (Coriell Institute for Medical Research, Camden; PN: NA12878). A Ligation Sequencing Kit V14 (Oxford Nanopore Technologies, Oxford, UK; PN: SQK-LSK114) was used to prepare DNA libraries for long-read sequencing with 3 μg of high molecular weight (>100 kb), high-quality (DIN >8) genomic DNA extracted from HEK-293 cells.

System, cartridge fabrication and operation

The system depicted in Fig. 2A was developed in collaboration with an external engineering partner, guided by design specifications established by our laboratory. The MAOS (mechanical actuation on surface) system incorporates an automated linear compression actuator comprising eight rollers that apply localized vertical compression to the top elastic film of the cartridge, facilitating droplet translation along the X-axis. Compression is applied along the Z-axis, perpendicular to the cartridge surface, while droplet motion is directed horizontally along predefined hydrophobic tracks aligned with the X-axis. Actuation is driven by a stepper motor and coordinated through integrated electronics, including a motor driver, microcontroller, and power regulation circuitry, to ensure precise timing and synchronization of roller movement.

Fig. 2 MAOS system and cartridge architecture. (A) MAOS system equipped with eight programmable compression rollers that actuate droplets along defined paths within a cartridge. The system integrates alternating isothermal-magnetic regions (IMRs) and thermal cycling regions (TCRs) to enable temperature-controlled and magnetically controlled bioanalytical workflows across eight parallel lanes. (B) MAOS cartridge has an 8-lane layout with a 3.5 mm thick body that sets the gap between the top and bottom films. It includes lane dividers, inlet ports, and an absorbent pad to enable structured and contamination-free droplet handling.

The system's base supports six processing regions—three isothermal magnetic regions (IMRs) and three thermal cycling regions (TCRs)—arranged in an alternating configuration. Each IMR comprises eight circular reaction zones (8 mm in diameter), each incorporating a bottom-mounted magnet for magnetic bead manipulation and a resistive heater for temperature control up to 75 °C. A ferrite button is embedded to concentrate the magnetic field within the droplet volume. Each TCR consists of eight wells (8 mm in diameter, 3 mm deep), thermally actuated via individual Peltier modules capable of cycling temperatures between 4 °C and 100 °C. Real-time temperature regulation is achieved through embedded thermistors interfaced with a PID-controlled feedback loop. The final system was delivered fully assembled, calibrated, and integrated with custom control software developed by the engineering partner.

The cartridge (INTEGRA Biosciences, San Francisco, CA; PN: M-02-0001-002-04), illustrated in Fig. 2B, is composed of three key elements: top and bottom thermoplastic polyurethane (TPU) elastic films (Covestro LLC, Pittsburgh, PA) with hydrophobic coatings, a rigid structural body, and an internal absorbent pad for waste capture. The fabrication process begins with coating the bottom plate with a hydrophobic layer to promote droplet mobility and reduce surface fouling; this layer is then cured under controlled thermal conditions. Once cured, the bottom plate is precisely aligned and bonded to the cartridge body to establish a sealed, leak-resistant interface. The top film is then processed by punching reagent inlet holes using a precision cutting method, followed by the same hydrophobic coating and curing steps. Absorbent pads are placed at the terminal regions of each fluidic lane to manage waste. Finally, the top film is aligned and sealed to the base assembly, completing the enclosure of the cartridge. The cartridge spacer is fabricated from polycarbonate. The top and bottom elastic films are heat-sealed to the outer periphery of the spacer body to ensure complete reagent containment and prevent external leakage. To prevent the risk of oil leaking between adjacent lanes during experiments, the cartridge clamp compresses the top and bottom films firmly against the lane dividers, while the heat-sealed periphery preserves lane integrity.

To begin operation, the cartridge is placed onto the system's base. Once the cartridge's locating features align with those on the base, it is locked into position using a quarter-turn clamp. A vacuum system is then activated via a switch to apply uniform suction through machined air channels and grooves within the base, beneath the cartridge. This vacuum ensures firm contact between the cartridge's bottom film and the base, minimizing vertical displacement and enabling reliable droplet actuation, temperature regulation, and magnetic control. Once the cartridge is secured, the system is initialized through a custom user software interface, which controls actuation sequences, heating profiles, and magnetic bead manipulation across the IMR and TCR zones. The software also allows users to configure roller movement speeds along both the Z-axis and X-axis, set timing for compression steps, and monitor real-time temperature feedback from embedded thermistors.

MAOS droplet actuation

To evaluate droplet actuation in the MAOS platform, a series of test liquids were tested under standardized compression and motion conditions. Unless otherwise specified, general droplet actuation was performed using standardized parameters based on droplet volume. For small droplets (0.5–10 μL), the distance between the engaged roller and the cartridge inlet hole was set to approximately 2 mm, with the top film fully compressed along the Z-axis against the bottom film to achieve a 0 mm gap between films, and an X-axis actuation speed of 1 mm s⁻¹. For medium-volume droplets (11–80 μL), the roller-to-inlet distance was increased to approximately 3.7 mm, with a gap of ∼0.5 mm between films and an X-axis speed of 2 mm s⁻¹. For larger droplets (81–200 μL), the roller-to-inlet distance was further increased to approximately 4.7 mm, with a gap of ∼1 mm between films and an X-axis speed of 2 mm s⁻¹. Across all conditions, the maximum force required to achieve compression was 3 N. These parameters were optimized to ensure reliable droplet transport across all experimental workflows. Waste generated during processing was actuated toward the integrated absorbent pad using the same actuation parameters, depending on droplet volume.

To evaluate the mobility of various liquids in the MAOS system, as summarized in Table 1, each test liquid was pipetted into one of the eight designated inlet holes of a cartridge lane at a fixed volume of 10 μL. The droplet was then actuated from the inlet hole to TCR 1 and back to the inlet position. Liquids that exhibited this bidirectional movement reproducibly were classified as “movable” (Y), while those that failed to complete the round-trip motion were classified as “not movable” (N). Surface tension and static contact angle measurements were performed using an Ossila Contact Angle Goniometer (Ossila Ltd, Sheffield, UK). Contact angles were measured on a hydrophobic-coated film using polynomial curve fitting to the droplet edge, and surface tension was determined using the pendant drop technique, as recommended by the instrument.

Table 1 Assessment of droplet mobility of various liquids in MAOS

Liquid	Surface tension (mN m^−1)	Contact angle^a (°)	Movable
a Data were measured as described in the Experimental section; values without annotations were obtained from ref. 20. Liquids marked Y exhibited movement under the conditions tested, while those marked N showed no movement. b Dropgloss is provided by INTEGRA Biosciences.
Pure water	73	108	N
Pluronic F-127 (0.05%)	50^a	95	N
Pluronic F-127 (0.5%)	40^a	88	Y
Pluronic F-127 (1.0%)	36^a	85	Y
Triton X-100 (0.05%)	42^a	63	Y
Triton X-100 (0.5%)	40^a	61	Y
Triton X-100 (1.0%)	36^a	60	Y
Tween 20 (0.05%)	49^a	93	N
Tween 20 (0.5%)	46^a	88	Y
Tween 20 (1.0%)	43^a	85	Y
Mineral oil (67 cSt)	30	52	N
Silicone oil (5 cSt)	20	28	Y
Dropgloss^b (2.3 cSt)	27^a	50	Y
2-Propanol	22	38	Y
Acetone	24	48	Y
Acetonitrile	29	63	Y
Ethanol	22	50	Y
Methanol	22	52	Y
Toluene	27	48	Y

---

Magnetic bead purification and PCR amplification in MAOS

Magnetic bead cleanup was performed inside the MAOS cartridge on each of the isothermal-magnetic regions (IMR 1–3). To start, 7.5 μL of ligated DNA, 6 μL of magnetic beads, and 20 μL of Dropgloss were added to the cartridge. The mixture was actuated to the IMR and incubated at room temperature for 5 min to allow DNA binding. A magnet was then engaged for 2 min to pellet the beads. The supernatant was removed and transported to the waste zone, followed by two washes with 50 μL of 80% EtOH. After washing, the magnet was disengaged, and the beads were resuspended in 7 μL of water and 5 μL of Dropgloss. The droplet was incubated again for 5 min, after which the magnet was re-engaged to pellet the beads once more. Finally, the cleaned DNA droplet was transferred to the inlet port for collection and downstream analysis.

For PCR validation, reactions were made by mixing 1.5 μL of purified DNA, 1 μL of index primers, 2.5 μL of Equinox PCR master mix, and 45 μL of Dropgloss. Each droplet was moved to one of the three thermal cycling regions (TCR 1–3) in the MAOS cartridge. PCR was performed in reaction wells using this thermal program: 98 °C for 45 s; followed by 8 cycles of 98 °C for 15 s, 60 °C for 30 s, and 72 °C for 30 s; ending with a final step at 72 °C for 1 minute. After PCR, the amplified droplet was sent to a second IMR and cleaned up again using the same bead-based method described earlier.

Final DNA concentrations were measured using a Qubit 4 Fluorometer (Thermo Fisher Scientific, USA) with 1 μL of sample. Fragment size and quality were analyzed using an Agilent TapeStation with a D1000 ScreenTape (Agilent Technologies, USA) and 1 μL of input per sample.

NGS library preparation using MAOS

The following workflow describes the execution of the Twist Library Preparation protocol for Illumina sequencing on the MAOS platform using 10% of the manufacturer's recommended reagent volumes. The 20% volume condition, also tested, followed the exact same workflow with all reagent volumes doubled. Importantly, the volume of encapsulation Dropgloss remained consistent between both protocols.

The protocol begins with enzymatic fragmentation, where a 4 μL droplet of genomic DNA (12.5 ng μL⁻¹, 50 ng total) is mixed with a 1 μL droplet of fragmentation master mix and 30 μL of Dropgloss. This mixture is transported to TCR 3 and incubated at 30 °C for 4.5 min and 65 °C for 30 min. The fragmented droplet is then transferred TCR 2, where a 0.5 μL droplet of adapter and a 2 μL droplet of ligation master mix are added, followed by incubation at 20 °C for 15 min.

To purify the ligated DNA, it is moved to IMR 2, and a 6 μL droplet of magnetic beads is introduced and incubated for 5 min. The magnet is engaged to pellet the beads, and the supernatant is removed to the waste zone. The pellet is washed twice using 50 μL droplets of 80% EtOH. After washing, the beads are resuspended in 1.5 μL of elution buffer, incubated for 15 min, and magnetically separated. The purified supernatant is then transported to TCR 1 for PCR. For PCR amplification, a 1 μL droplet of PCR index primer, a 2.5 μL droplet of PCR master mix, and 35 μL of Dropgloss are combined with the purified DNA. The reaction mixture is thermally cycled using the following program: initial denaturation at 98 °C for 45 s, followed by 8 cycles of 98 °C for 15 s, 60 °C for 30 s, and 72 °C for 30 s, with a final extension at 72 °C for 1 min.

After amplification, a second magnetic bead-based cleanup is performed at IMR 1 using a 5 μL droplet of magnetic beads, followed by two 50 μL EtOH washes. The final DNA library is resuspended in a 20 μL elution droplet and transported to the inlet zone. The eight libraries obtained from a single 8-plex run are pooled volumetrically by taking 188 ng from each library and proceeded with whole-exome enrichment using a Twist Fast Hybridization and Wash Kit (Twist Bioscience, San Francisco, CA; PN: 104181). Enriched libraries were sequenced on an Illumina NovaSeq X platform using a 10B flow cell with paired-end 150 bp (PE150) reads, and data were processed using an in-house bioinformatics pipeline.

The following workflow describes the execution of the Ligation Sequencing Kit V14 protocol from Oxford Nanopore Technologies (ONT) on the MAOS platform using 25% of the manufacturer's recommended reagent volumes. The DNA library preparation protocol began with end-repair and dA-tailing. A 12 μL droplet DNA extracted from HEK293 cells (250 ng μL⁻¹, 3 μg total) was combined with three smaller droplets: 1.8 μL of DNA repair buffer, 0.5 μL of DNA repair mix, and 0.8 μL of enzyme mix. The combined reaction droplet was encapsulated in 20 μL of Dropgloss, actuated to TCR 2, and thermally processed at 20 °C for 9 min followed by 65 °C for 10 min.

Following enzymatic treatment, the droplet was transferred to IMR 2 for bead-based purification. A 15.1 μL droplet of magnetic beads (at a 1.0× ratio) was added, and the droplet was incubated at room temperature for 10 min. The magnet was then engaged to pellet the beads, and the supernatant was transported to the waste zone. Two 80 μL EtOH droplets were applied sequentially for washing. Instead of elution, 10 μL of nuclease-free water encapsulated in 5 μL Dropgloss was directly added to the bead pellet. This resulting droplet was then transferred to TCR 1 and merged with 1.5 μL of ligation adapter, 7.5 μL of ligation buffer, and 3.3 μL of DNA ligase to initiate adapter ligation. The reaction was encapsulated in Dropgloss and incubated at room temperature for 60 min. A second bead-based cleanup was then initiated by adding 8.92 μL of AMPure XP beads (at a 0.4× ratio), followed by incubation and magnetic separation. Two washes were performed using 62.5 μL droplets of long fragment buffer, favoring retention of long DNA fragments. The final library was resuspended in a 45 μL elution droplet, incubated sequentially at 48 °C for 10 min and 37 °C for 30 min in IMR 1, and collected from the inlet port for downstream ONT sequencing.

The Ligation Sequencing Kit V14 (Oxford Nanopore Technologies, Oxford, UK; PN: SQK-LSK114) was used to prepare DNA libraries for long-read sequencing. The prepared DNA library was then quantified and loaded onto a MinION Mk1D device (Oxford Nanopore Technologies; PN: MIN-101D) equipped with a MinION Flow Cell R10.4.1 (PN: FLO-MIN114). Sequencing was initiated using a MinKNOW software, and basecalling was performed with Dorado.

Results and discussion

Mechanical actuation on surface (MAOS)

We explored a new method for droplet actuation using mechanical actuation on surface (MAOS). As shown in Fig. 1, this approach employs two hydrophobic-coated surfaces—a fixed bottom layer and an elastic top film—between which droplets are confined and optionally encapsulated in an immiscible phase. Localized compression applied to one side of the top film generates a driving force (F_drive) that pushes the droplet in the direction of the applied pressure. This force arises primarily from compression-induced asymmetries in contact angle and Laplace pressure gradients within the droplet.^17,18 In parallel, the droplet resists motion through a set of interfacial forces (F_resist), including contact angle hysteresis, capillary adhesion, and viscous drag.¹⁹ Actuation occurs only when F_drive exceeds F_resist.

To experimentally evaluate this model, we applied compression to various liquid droplets and recorded their mobility. As summarized in Table 1, movement was strongly correlated with interfacial properties such as surface tension, contact angle, and viscosity. Pure water—with a surface tension of 73 mN m⁻¹ and contact angle of 108°—remained pinned, consistent with a high F_resist and negligible F_drive (Fig. 1A). Adding surfactants (Tween 20 and Pluronic F-127) at 0.05% modestly reduced the surface tension and contact angle, yet no movement was observed. Reliable droplet motion emerged at higher concentrations (0.5% and 1.0%) of these surfactants, where surface tension dropped to ≤43 mN m⁻¹ and contact angles fell to ≤88°. Notably, Triton X-100 enabled consistent actuation at all tested concentrations, with surface tensions ≤42 mN m⁻¹ and contact angles as low as 60°. These results suggest that surface tension ≤50 mN m⁻¹ and contact angle ≤90° are necessary—but not always sufficient—conditions for motion (Fig. 1B).

We next examined organic solvents to further explore the generality of MAOS actuation. Solvents such as 2-propanol, acetone, acetonitrile, ethanol, methanol, and toluene all exhibited smooth motion under compression, consistent with their relatively low surface tensions (22–29 mN m⁻¹) and moderate contact angles (38–63°)—values well below the empirically defined thresholds for droplet mobility. However, mineral oil, which has a surface tension of 30 mN m⁻¹ and a contact angle of 52°, failed to actuate under identical conditions. This discrepancy highlights the role of viscosity in addition to interfacial tension and contact angle. To investigate this further, we tested low-viscosity fluids such as silicone oil and Dropgloss. Despite differing viscosities, both supported smooth droplet transport under compression.

Finally, we evaluated a hybrid case: water droplets that are otherwise immobile on MAOS but were encapsulated in a low-viscosity interfacial medium. When encapsulated by silicone oil or Dropgloss, these droplets moved reliably. The lubricating encapsulating layer is hypothesized to reduce contact line pinning and capillary adhesion,²¹ allowing the compressed droplet to overcome resistive forces and achieve motion. These experiments further support the model that successful actuation requires a balance of low surface tension, low contact angle, and low viscosity, and they demonstrate the utility of encapsulation as a practical strategy to expand the range of actuated fluids (Fig. 1C). Encapsulation was achieved during loading by first introducing oil into the designated cartridge inlets, followed by dispensing aqueous reagent droplets into the oil, where they became fully surrounded by the immiscible phase before actuation began.

Collectively, these results indicate that successful actuation depends on a combination of interfacial tension, contact angle, and viscosity. While more detailed force quantification is warranted, these thresholds guide practical use. With these physical parameters defined, we next translated the MAOS actuation principle into a functional platform by engineering a system and cartridge tailored for automated biochemical workflows.

MAOS system design and cartridge

Building on the MAOS-driven droplet actuation framework, we developed an integrated system and cartridge to enable automated execution of core bioanalytical operations, including droplet transport, thermal cycling, and magnetic bead-based purification.

The MAOS instrument (Fig. 2A) is a benchtop device engineered for parallel processing of microliter-scale reactions. It incorporates a compression actuator with eight independently controlled rollers that apply localized force to the flexible top film of the cartridge, enabling precise droplet movement. Embedded in the system base are three isothermal-magnetic regions (IMRs) and three thermal cycling regions (TCRs), arranged in an alternating sequence. Each IMR contains eight addressable reaction zones capable of active heating (up to 85 °C) and magnetic control, providing a total of 24 IMR zones. Each TCR comprises eight thermal wells capable of cycling between 4 °C and 100 °C, totaling 24 TCR wells. This spatial arrangement supports multi-step workflows with regulated temperature and magnetic field profiles across all processing lanes.

The MAOS cartridge (Fig. 2B) is a single-use consumable composed of four primary components: a top and bottom elastic film, a 3.5 mm-thick spacer frame, and an absorbent waste pad. The use of inexpensive materials and simplified fabrication contributes to a cost-effective solution compared to standard microfluidic devices. The elastic film material was selected for its flexibility, thermal stability, and resilience, allowing the films to deform reversibly under compression and return to their original shape upon release. The same elastic film is used for both top and bottom substrates to ensure uniform mechanical response and thermal compatibility. Flexibility in the bottom film is essential during thermal cycling, as it enables conformal contact with the reaction wells under vacuum, ensuring efficient heat transfer. This bottom-side compliance is also critical during magnetic bead-based purification, where proximity to the underlying magnet enhances field strength and promotes effective bead capture. Both internal surfaces are hydrophobically treated to minimize droplet pinning and enable smooth, consistent droplet motion under mechanical actuation.

Each cartridge contains eight independent processing lanes, physically isolated from one another to prevent cross-contamination between samples during droplet manipulation and processing. Droplets are introduced through inlet ports along the bottom edge of each lane and collected in the absorbent pad at the top. During operation, the cartridge is held in place by a system-applied vacuum and reinforced with a mechanical clamp to ensure precise alignment and positional stability.

MAOS droplet actuation

The MAOS platform accommodates a wide range of droplet volumes—from nL to μL—enabling diverse liquid-handling operations within a single system. This versatility is made possible by the adjustable inter-film gap, which is facilitated by the elasticity of the top film. As shown in Fig. 3A(1–6), MAOS reliably actuated droplets spanning four orders of magnitude in volume. To illustrate this volumetric range and droplet control, we conducted a demonstration using dye and water droplets. All dye and water droplets used in this demonstration contained 0.5% (v/v) Pluronic F-127 surfactant to ensure consistent droplet mobility throughout the workflow.

Fig. 3 Volume scaling and droplet morphology in the MAOS platform. (A) Stepwise expansion (1–6) from 500 nL to 100 μL by sequential merging of dye and water droplets (all containing 0.5% v/v Pluronic F-127 for mobility). Droplets are loaded, merged, and transported across processing zones (IMR 1, TCR 1, IMR 3), followed by final transfer to the waste zone. (B) Morphological response to compression release. (1) Small droplets (500 nL) fragment due to surface tension, splitting volume between top and bottom films. (2) Large droplets (100 μL) retain a stable bridge-like geometry, with most fluid residing on the lower surface.

The demonstration began with eight 500 nL dye droplets dispensed into the cartridge and transported to IMR 1. Next, 5 μL water droplets were introduced and merged with the dye droplets, producing a total volume of 5.5 μL. These were mixed by oscillatory motion (Video S1) and moved to TCR 1. Subsequently, 20 μL water droplets were added to form 25.5 μL droplets, which were then transferred to IMR 2 for further processing. Following that, 74.5 μL droplets were merged, yielding a final volume of 100 μL, which was actuated to IMR 3. The process concluded with transport to the waste zone. This experiment showcases MAOS's ability to execute sequential operations—merging, mixing, dilution, and transport—in a single cartridge. The final 100 μL volume represents a 200-fold dilution, highlighting MAOS's potential for serial dilutions.

To maintain consistent actuation, we tuned the inter-film gap based on droplet size. Droplets ≤10 μL were handled with a fully compressed gap (0 mm), while those >10 μL required a slightly expanded gap (≥0.5 mm). Roller spacing was also adjusted to prevent interference with inlet ports during high-volume dispensing. Details on setup parameters—gap spacing, roller offsets, and actuation speeds—are in the Experimental section. These adjustments ensured reproducible and clean droplet movement across the full volume range.

We further investigated post-compression droplet stability. As shown in Fig. 3B, smaller droplets (e.g., 500 nL) were frequently fragmented upon release, with partial volume adhering to both films—a result of surface tension effects within the 3.5 mm cartridge gap. However, most of the volume settled on the bottom film and re-merged upon recompression. Larger droplets (e.g., 100 μL) remained intact, forming stable liquid bridges between the films, with the base broadened by gravitational effects. This volume-dependent behavior highlights the dominant role of cohesive forces in larger droplets, while surface tension governs fragmentation in smaller ones.

MAOS's flexible architecture distinguishes it from conventional microfluidics. Fixed-geometry channel systems and DMF platforms with rigid gap heights restrict the volume range. In contrast, MAOS's deformable top film adapts the cartridge gap in real-time, enabling continuous manipulation of droplets from 500 nL to 100 μL. This adaptability allows multi-step workflows—such as serial dilutions and reagent additions—within a single cartridge, eliminating the need for hardware modifications.

MAOS for automated NGS library preparation

We evaluated the MAOS platform for automated sample preparation in next-generation sequencing (NGS), targeting core unit operations including magnetic bead-based purification, thermal cycling, and end-to-end library construction. The system enables precise droplet control in a compact format and is ideally suited for multiplexed workflows where reproducibility, integration, and volume reduction are essential.

Magnetic bead-based cleanup was implemented using the isothermal-magnetic regions (IMRs). In these regions, a magnet positioned beneath the substrate enables spatially localized bead immobilization. As illustrated in Fig. 4A, droplets containing magnetic beads are actuated to the IMR where the magnet is engaged. The supernatant is removed by actuating the droplet to the waste zone. This is followed by sequential washing and elution steps, each delivered and actuated independently. After re-engaging the magnet, the eluate is directed to the inlet for retrieval. To validate bead-handling performance, we performed post-ligation cleanup on MAOS, while upstream steps (fragmentation, end-repair, A-tailing, ligation) were carried out manually. DNA recovery on MAOS reached 91.1%, comparable to the 90.7% recovery achieved with manual pipetting. Removal of low-molecular-weight species (80–150 bp) was also efficient, with 98.5% clearance on MAOS versus 98.1% for the control (Fig. 4C), indicating that solid-phase extraction is fully compatible with digital actuation and spatially confined bead handling.

Fig. 4 Magnetic bead-based purification and thermocycling on MAOS. (A) DNA cleanup workflow in IMR 3. (1) Magnetic bead-containing droplets are actuated into position. (2) A magnet immobilizes the beads; supernatant is discarded. Inset: Bead pellet localized at the zone center. (3) Beads are washed with EtOH and resuspended upon magnet release. (4) Elution droplet is added, and purified DNA is either directed downstream or retrieved at the inlet. (B) PCR in the thermal cycling region (TCR). Reaction droplets are encapsulated in Dropgloss and compressed to form a stable interfacial bridge between bottom and top surfaces. The encapsulation layer suppresses evaporation and maintains thermal isolation, enabling fixed-position amplification. (C) Post-ligation cleanup results comparing MAOS to manual processing in terms of DNA recovery and removal of short fragments. (D) PCR yields from MAOS thermal cycling regions versus standard thermocyclers. Each data point represents N = 72 replicate measurements, and error bars represent ±1 S.D.

To evaluate the MAOS platform's ability to support full workflows, we implemented fully automated, end-to-end NGS sample prep for both short- and long-read sequencing. All operations—including magnetic bead cleanup, PCR amplification, and droplet control—were performed on MAOS without user intervention. Eight libraries were prepared in parallel under each condition using significantly reduced reagent volumes (10–25% of recommended standard kits). Workflows on the MAOS platform started at the zone nearest to the waste reservoir and progressed backward toward the inlet—a directional layout that minimizes cross-contamination as droplets advance through each step. Fig. 5 illustrates this concept as a matrix, linking the cartridge architecture to key steps in the Twist library preparation protocol. The same directional workflow principle was used for Oxford Nanopore Technologies (ONT) library preparation. Sequencing was carried out on both Illumina and Oxford Nanopore Technologies (ONT) platforms to assess compatibility across read formats.

Fig. 5 MAOS NGS library preparation workflow. Single cartridge lane with labeled processing zones (top) — isothermal-magnetic region (IMR) for bead cleanup, thermal cycling region (TCR) for PCR, and other assay-specific zones — mapped to the key steps of the Twist library preparation protocol (left). Arrows indicate the flow direction: reagent movement from the inlet to processing zones (solid blue), reaction outputs moving backward toward the inlet (dashed orange), and waste from bead cleanup to the absorbent pad (dashed black). The inlet port serves as both the reagent entry and final sample retrieval point. The same directional principle was applied for ONT library preparation.

For Illumina-based short-read libraries, we tested 10% and 20% volume conditions and compared them against matched manual workflows. As shown in Fig. 6, key target enrichment metrics were maintained across all groups. 30× coverage remained consistent, and fold 80 base penalty values were only marginally elevated under automated conditions, indicating balanced coverage even at lower input volumes. Both AT and GC dropout levels were low and uniform, suggesting no compositional bias introduced by MAOS. Median insert size and percent duplicate reads showed minimal variation, and percent on-target alignment remained nearly identical across all runs. Collectively, these metrics confirm that MAOS reliably supports high-quality library construction for short-read sequencing, with performance comparable to manual prep even at a 10-fold volume reduction.

Fig. 6 Sequencing quality metrics across reduced-volume library prep. Comparison of MAOS and manual workflows using the Twist Library Preparation protocol at 10% and 20% reagent volumes relative to the standard protocol volume. Each data point represents at least N = 17 replicate measurements.

Thermal cycling was conducted in temperature-controlled regions (TCRs) equipped with embedded heaters and thermal sensors. Reaction droplets (5 μL) containing ligated DNA, primers, and PCR master mix were encapsulated with Dropgloss and compressed into shallow wells, creating a sealed interface that suppresses evaporation and ensures uniform thermal contact (Fig. 4B). Unlike droplet-shuttling thermocyclers, MAOS maintains the droplet in a stationary configuration throughout cycling, reducing inter-lane variability and minimizing heat loss. Across all eight TCRs, PCR yields exceeded 500 ng per reaction (Fig. 4D), equivalent to bench-scale results. Evaporation losses remained below 3% up to 20 PCR cycles, underscoring the effectiveness of the Dropgloss encapsulation geometry (Fig. 4B) in suppressing evaporation during high-temperature processing.

For ONT-based long-read sequencing, libraries were constructed using 25% reagent volumes. After executing the steps of DNA repair and end-prep, and adapter ligation with bead-based cleanup on MAOS, the sequencing results are summarized in Table 2 and Fig. 7. Despite a 75% reduction in reagent use, libraries yielded the same total base output as manual prep (1.1 Gb), with N50 values closely matched (44.0 kb vs. 45.2 kb). The mean read length was lower in MAOS (7.8 kb vs. 11.0 kb) likely reflecting a bias toward recovery of shorter fragments, as also indicated by the lower median read length. This suggests that optimizing fragment size selection conditions could further improve length distribution. Quality scores were comparable between workflows, confirming that MAOS supports library construction for long-read sequencing at miniaturized scale.

Table 2 Summary of ONT sequencing metrics for DNA libraries prepared with MAOS and manually at 25% volume

Sequencing metrics	MAOS	Manual
Mean read length (kb)	7.8	11.0
Median read length (kb)	1.0	2.6
Mean quality score	11.0	11.6
Median quality score	12.1	13.0
Read length N50 (kb)	44.0	45.2
Number of reads	139 000	97 000
Total bases (Gb)	1.1	1.1

---

Fig. 7 Read length distribution from Oxford Nanopore sequencing. Base yield versus read length is shown for DNA libraries prepared with MAOS (A) and manually (B). Both conditions generated long-read libraries with acceptable profiles (N = 3).

The above workflows revealed a critical learning regarding the final elution step in library preparation: encapsulation must be excluded to ensure compatibility with downstream sequencing platforms, as residual encapsulating media can interfere with performance. To address this, we developed a proprietary elution buffer formulated to be sequencing-compatible while preserving recovery efficiency. However, for ONT workflows, the required heating during elution—48 °C for 10 min followed by 37 °C for 30 min—introduced a significant challenge. Without encapsulation to suppress evaporation, we observed consistent volume loss. Based on an empirically measured ∼55% evaporation rate, we increased the elution volume to 45 μL to reliably recover the 20 μL needed for sequencing. The ability to implement this adjustment was made possible by the MAOS platform's unique design: the co-location of thermal and magnetic control within the same isothermal magnetic region. This feature allowed temperature-assisted elution to be performed in a stationary droplet, without user intervention or sample transfer.

A key advantage of the MAOS platform is its ability to operate at significantly reduced reagent volumes. Across all workflows, we demonstrated robust NGS library construction using just 10–25% of standard kit volumes—enabling reagent savings of up to 90% without compromising performance. For labs running high-throughput sequencing, this reduction translates to substantial cost efficiency, particularly in reagent-heavy steps like PCR and bead cleanup. Beyond cost, MAOS's small footprint and modular design allow it to be integrated into existing laboratory automation setups. It can interface with upstream or downstream liquid handlers, enabling fluid transitions between instruments and supporting automation of complex workflows. This flexibility makes MAOS valuable not only as a standalone unit, but also as a modular system for executing precision operations—such as thermal cycling and magnetic bead purification—within broader automated processes.

Conclusions

We developed MAOS, a mechanically actuated platform for programmable droplet manipulation that sidesteps the constraints of embedded electronics and fixed-channel designs. The system supports core biochemical operations—including magnetic bead cleanup, thermocycling, and droplet transport—within a compact, cartridge-based format, handling volumes from nL to μL. By probing the interplay of surface tension, contact angle, and viscosity, we defined the physical thresholds for reliable actuation and demonstrated oil encapsulation as a robust means to overcome pinning. In the context of next-generation sequencing, MAOS enabled efficient DNA purification and low-volume PCR with high reproducibility and scaled to multiplexed library preparation using just a fraction of conventional reagent volumes. Sequencing results were consistent with manual benchmarks. Together, these findings establish MAOS as a versatile, accessible platform for automated workflows in molecular biology and beyond.

Author contributions

P. N. conducted the MAOS experiments and performed data analysis to assess performance and reproducibility. P. N. also executed the short-read sequencing workflows and associated data analysis. R. L. was responsible for the complete development (design, microfabrication, and assembly) of the MAOS cartridges. C.-C. Lee contributed to understanding the magnetic-bead process involved in the MAOS system and performed the long-read sequencing experiments. E. C. contributed to early droplet mobility studies with MAOS and provided key insights into its theoretical foundations. M. J. J. initially discovered the MAOS phenomenon, and together with F. C., defined the conceptual framework and scientific objectives of the study. F. C. and M. J. J. provided ongoing supervision and guidance throughout the project. M. J. J. prepared the first draft of the manuscript. All authors contributed to the design of experiments, interpretation of results, and revision of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

Supplementary Information available: Video S1 demonstrating oscillatory droplet mixing during sequential expansion workflows. See DOI: https://doi.org/10.1039/D5LC00625B.

The authors confirm that all data supporting the findings and conclusions of this study are included in the main text of the article.

Acknowledgements

This work was funded by INTEGRA Biosciences. No external funding was received.

References

1. R. P. Adams and G. E. Homan, Nat. Rev. Genet., 2023, 24, 45–59.
2. A. M. Streets and Y. Huang, Curr. Opin. Biotechnol., 2014, 25, 69–77.
3. G. M. Whitesides, Nature, 2006, 442, 368–373.
4. D. Mark, S. Haeberle, G. Roth, F. von Stetten and R. Zengerle, Chem. Soc. Rev., 2010, 39, 1153–1182.
5. M. G. Pollack, R. B. Fair and A. D. Shenderov, Appl. Phys. Lett., 2000, 77, 1725–1726.
6. R. Gorkin, J. Park, J. Siegrist, M. Amasia, B. S. Lee, J. Park, J. Kim, H. Kim, M. Madou and Y.-K. Cho, Lab Chip, 2010, 10, 1758–1773.
7. J. Friend and L. Y. Yeo, Rev. Mod. Phys., 2011, 83, 647–704.
8. M. A. Unger, H.-P. Chou, T. Thorsen, A. Scherer and S. R. Quake, Science, 2000, 288, 113–116.
9. B. M. Paegel and R. A. Mathies, Anal. Chem., 2004, 76, 525–533.
10. S. K. Cho, H. Moon and C.-J. Kim, J. Microelectromech. Syst., 2003, 12, 70–80.
11. R. B. Fair, Microfluid. Nanofluid., 2007, 3, 245–281.
12. A. R. Wheeler, Science, 2008, 322, 539–540.
13. M. Abdelgawad and A. R. Wheeler, Microfluid. Nanofluid., 2008, 4, 349–355.
14. J. Seo, S.-K. Lee, J. Lee, J. S. Lee, H. Kwon, S.-W. Cho, J.-H. Ahn and T. Lee, Sci. Rep., 2015, 5, 12326.
15. G. Chen, Y. Gao, M. Li, B. Ji, R. Tong, M.-K. Law, W. Wen and B. Zhou, J. Mater. Sci., 2018, 53, 13253–13263.
16. Illumina, Inc., Illumina's revolutionary NovaSeq X exceeds 200th order milestone in first quarter 2023, Press Release, 4 April 2023, https://emea.illumina.com/company/news-center/press-releases/2023/4b48580b-42d4-419b-8512-5adcbb069836.html.
17. F. Mugele and J.-C. Baret, J. Phys.: Condens. Matter, 2005, 17, R705–R774.
18. P.-G. de Gennes, F. Brochard-Wyart and D. Quéré, Capillarity and Wetting Phenomena: Drops, Bubbles, Pearls, Waves, Springer, New York, 2004.
19. I. Swyer, R. Fobel and A. R. Wheeler, Langmuir, 2019, 35, 5342–5352.
20. PubChem, National Library of Medicine, https://pubchem.ncbi.nlm.nih.gov, accessed April 2025.
21. H. Xu, Y. Zhou, D. Daniel, J. Herzog, X. Wang, V. Sick and S. Adera, Nat. Commun., 2023, 14, 4901.

---