High durability and stability of 2D nanofluidic devices for long-term single-molecule sensing

The study addresses a central challenge in 2D solid-state nanopore sensors: instability of nanopores in atomically thin membranes, particularly monolayer MoS2. While 2D nanopores offer high spatial resolution for single-molecule sensing and have been applied to DNA, RNA, protein analysis, desalination, separations, and energy harvesting, their practical deployment is limited by device yield, variability, stability, and reliability. Prior work has suggested coating strategies and explored silicon-based nanopore stability, but systematic analysis of 2D nanopore stability is lacking. The authors hypothesize that two mechanisms—voltage-mediated delamination of the 2D layer from SiNx substrates and chemical oxidation-induced defect formation and pore growth in aqueous media—are primary causes of instability. The purpose is to identify, quantify, and mitigate these mechanisms to enable durable, long-term single-molecule sensing with 2D MoS2 nanopores.

The authors review the landscape of 2D nanopores across materials including graphene, MoS2, WS2, hBN, and MXenes, highlighting MoS2’s advantages for biosensing due to its tri-atomic thickness (~0.65 nm) and weaker DNA base adhesion compared to graphene. Prior studies achieved single-nucleotide discrimination and topological DNA detection, and demonstrated MoS2 nanopore FETs. Practical bottlenecks in solid-state nanopores include noise, robustness, and especially stability; attempts to improve robustness (e.g., TiO2 coating on graphene) increased membrane thickness. Stability research on silicon-based nanopores and reports of voltage-mediated delamination in 2D materials indicate adhesion issues on hydrophilic SiNx. Air exposure leads to spontaneous oxygen incorporation and oxidation in 2D TMDs; CVD MoS2 and WS2 can crack and oxidize along grain boundaries. Theoretical and experimental works indicate high barriers to basal-plane oxidation in pristine MoS2 that are lowered at defects, edges, and grain boundaries, suggesting defect-mediated oxidative degradation under ambient or aqueous conditions.

Device architecture and fabrication: SiNx membranes (~20 nm thick, ~30×30 µm) with 80–120 nm apertures were patterned by e-beam lithography and dry etching on wafer-scale substrates (100 mm). Monolayer MoS2 crystals grown by MOCVD (Mo(CO)6 and H2S precursors at 850 °C, sapphire substrates, NaCl and oxygen-assisted growth, post-growth H2S to suppress S vacancies) were transferred via PMMA-assisted transfer onto the SiNx membranes to suspend MoS2 over apertures. Substrate modification: To improve MoS2–substrate adhesion, SiNx surfaces were oxygen-plasma treated to form hydroxylated surfaces and then primed with HMDS vapor (dehydration bake, HMDS exposure, purge cycles), depositing a monolayer terminated with methyl groups to render the substrate hydrophobic. Wettability and surface free energy were characterized via contact angle (CA) and extended Fowkes method using water, diiodomethane, and ethylene glycol. Nanopore fabrication and electrical measurements: Nanopores were drilled in monolayer MoS2 using STEM/TEM at 80 kV. Devices were mounted in a PMMA flow cell. Ionic measurements were performed in 1 M KCl (unless stated), with typical filtering at 10 kHz and sampling at 100 kHz. I–V curves and open-pore conductance (Gopen) were measured, and nanopore diameters were estimated using the general conductance model with assumed membrane thickness L≈1 nm for monolayer MoS2. Delamination assessment: Abrupt, stepwise increases in open-pore current during voltage sweeps (up to ±500 mV) were monitored as signatures of delamination. Post-measurement bright-field TEM imaging verified MoS2 detachment around the SiNx aperture. Defect quantification: Aberration-corrected ADF-STEM (FEI Titan Themis, specified probe and collection conditions) was used to image pristine monolayer MoS2 and after incubation in aqueous ionic solution. Sulfur vacancy concentrations (Vs, Vs2) were counted across ~3500 nm2 areas and extrapolated considering e-beam dose to estimate intrinsic defects and changes after aqueous exposure. Oxidation and PL assays: Photoluminescence (PL) spectra of monolayer MoS2 in water were collected on a confocal setup (561 nm excitation, water immersion objective) under two dissolved oxygen (DO) conditions: air-saturated (~8 mg L−1 O2) and Ar-purged (<1 mg L−1 O2) in a sealed fluidic chamber. Time-dependent PL intensity and peak energy shifts were monitored to infer charge transfer and potential photo-induced oxidation/dissolution. Pore growth in aqueous buffers: TEM-fabricated single and double nanopores on HMDS-modified substrates were incubated 12 h in 1 M KCl TE buffer (pH ~7.5) at room temperature under either air-saturated DO (~8 mg L−1) or low-oxygen (~1 mg L−1) conditions without applied bias; pore sizes were imaged by TEM before/after. Long-term DNA sensing: A sealed flow cell with degassed, filtered 1 M KCl TE buffer (<1 mg L−1 O2) was used to translocate 1 kbp dsDNA at 500 mV through a ~6.5 nm MoS2 nanopore for >3 h. Events were detected and fitted using OpenNanopore and CUSUM algorithms. Open-pore conductance drift and conductance drops (ΔGdrop) distributions were analyzed at early (0–30 min) and late (150–180 min) time windows to infer pore size stability. Controls and ancillary procedures: Contact angle and surface free energy measurements validated HMDS coating (stability tracked up to 28 days). Leakage conductance of intact SiNx was assessed (<~300 pS). Dissolved oxygen was monitored with an optical O2 probe. Occasional pore clogging was addressed by reverse biasing or buffer exchange.

• Two primary instability mechanisms were identified: (1) voltage-mediated delamination of monolayer MoS2 from hydrophilic SiNx, and (2) oxidation-driven defect formation and pore enlargement in aqueous solutions containing dissolved oxygen. • Device failure statistics: Among unsuccessful devices (n=36), ~70% failed due to unstable MoS2 nanopores (instability/ membrane damage), consistent with prior observations in graphene. • Delamination signatures: For a device initially at Gopen≈13 nS (calculated pore ~4.2 nm), the conductance abruptly increased to ~225 nS within minutes—comparable to the bare SiNx aperture—indicating MoS2 detachment confirmed by TEM. In another device (dTEM≈2.5 nm), stepwise conductance increases began at ~200–300 mV, reaching Gopen ~150 nS at 200 mV and ~400 nS at 300 mV, far exceeding the expected ~25 nS for an intact pore, highlighting voltage-dependent delamination. • HMDS substrate modification improved adhesion and stability: Oxygen-plasma-treated SiNx showed CA ~10°, increasing to ~60–62° after HMDS, with surface free energy reduced from ~60 to ~40 mN/m, indicating successful hydrophobization. Over ~5 h, ΔG in HMDS-modified devices remained <50 nS, whereas unmodified hydrophilic devices often exceeded ΔG >400 nS; the rate of conductance change Ė was up to ~4 nS/min (unmodified) vs <1 nS/min (HMDS-modified). TEM confirmed intact MoS2 layers on HMDS/SiNx post-measurement. • Aqueous oxidation increases defect density: Sulfur vacancy concentration increased from 1.2 ± 0.3 × 10^13 to 1.9 ± 0.4 × 10^13 defects cm−2 after 12 h incubation in non-degassed 1 M KCl (DO ~8 mg L−1), indicating oxygen-assisted defect formation. • PL evidence for oxygen-mediated reactions: In water with DO ~8 mg L−1, MoS2 PL intensity initially increased >2× and blue-shifted by ~35 meV (transition from trion to exciton emission), followed by decay/red-shift, consistent with oxygen-driven charge transfer and potential material dissolution. Under low O2 (<1 mg L−1), PL intensity and energy were stable, indicating suppressed photo-induced reactions. • Pore growth depends on dissolved oxygen: TEM-tracked nanopores enlarged over 12 h in air-saturated buffer (~8 mg L−1 O2), including coalescence of double pores, while pores in low-O2 buffer (~1 mg L−1) showed only slight size increases. • Long-term single-molecule sensing achieved: In a sealed, low-O2 buffer, a ~6.5 nm MoS2 nanopore enabled continuous 1 kbp dsDNA measurements for >3 h at 500 mV. Gopen increased modestly from ~58 nS to ~62 nS (~7%; ~0.03 nS/min). ΔGdrop distributions for unfolded DNA remained close to model expectations: ~4.41 nS at 30 min (n=1832) and ~4.2 nS at 180 min (n=1195), indicating minimal pore enlargement and no formation of additional pores.

The results directly address the posed instability mechanisms in monolayer MoS2 nanopores. Abrupt, voltage-dependent increases in conductance and TEM confirmation establish that delamination from hydrophilic SiNx is a dominant failure mode under bias. Introducing a hydrophobic HMDS monolayer lowers surface free energy and increases contact angle, strengthening van der Waals interactions between MoS2 and the substrate, thereby suppressing voltage-triggered delamination and stabilizing membrane conductance over hours. The oxidation study demonstrates that dissolved oxygen in aqueous buffers elevates sulfur vacancy formation and accelerates pore edge dissolution and growth. PL measurements corroborate oxygen-driven charge withdrawal and indicate that reducing dissolved O2 below ~1 mg L−1 suppresses photo-induced chemistry and material dissolution. By combining HMDS-modified substrates with deoxygenated buffers, the authors achieve long-term, low-drift DNA sensing (>3 h at 500 mV), with ΔGdrop statistics and Gopen drift consistent with a single, stable pore. These findings are significant for robust 2D nanopore sensors, showing that addressing interfacial adhesion and buffer chemistry can unlock durable, high signal-to-noise single-molecule measurements and extend device shelf-life and reusability.

This work identifies and mitigates the two principal causes of instability in monolayer MoS2 nanopore devices—substrate delamination and aqueous oxidation. Hydrophobizing SiNx substrates with an HMDS monolayer markedly improves MoS2 adhesion and suppresses voltage-induced delamination. Lowering dissolved oxygen in measurement buffers diminishes oxidation-driven defect formation and pore enlargement. Together, these strategies enable hours-long, stable DNA translocation measurements through a single monolayer MoS2 pore with minimal conductance drift. The study provides practical guidelines for engineering durable 2D nanopore devices suitable for long-term biosensing. Future research could explore: optimizing alternative surface functionalizations or interlayers to further enhance adhesion; investigating other 2D materials and heterostructures; controlling pore edge chemistry to resist oxidation; integrating in situ oxygen scavenging or inert microfluidics; quantifying long-term shelf-life across storage conditions; and scaling to high-throughput, arrayed devices while maintaining stability and low noise.

Stability still depends on material quality and experimental conditions; oxidation is mitigated but not eliminated, with modest pore growth observed even under low-O2 conditions. Occasional clogging due to nanobubbles or contaminants persists, requiring operational interventions (reverse bias, buffer exchange). The study focuses on monolayer MoS2 on SiNx with HMDS; generalization to other 2D materials, substrates, and chemistries requires validation. Measurements were performed at specific ionic strengths, pH, and room temperature; different environments and higher biases may alter failure thresholds. Sample sizes for device-level statistics are limited, and long-term stability beyond several hours, as well as extended storage and reuse cycles, warrant further investigation.