Digital plasmonic nanobubble detection for rapid and ultrasensitive virus diagnostics

The study addresses the need for rapid, ultrasensitive, and simple diagnostics for infectious diseases highlighted by the COVID-19 pandemic. Conventional digital immunoassays achieve single-molecule detection and absolute quantification by partitioning samples into microwells or droplets for independent amplification and counting but require complex workflows (washing, partitioning, amplification). The research question is whether plasmonic nanobubble-based detection can enable a simplified, compartment-free, calibration-free digital immunoassay that maintains single-particle sensitivity and allows direct detection of intact viruses at room temperature without washing or amplification. The authors introduce DIAMOND, which uses transient plasmonic nanobubbles generated around gold nanoparticles by short laser pulses, detected optically in flow to create virtual detection zones for digital on/off counting. The aim is to validate single-NP detection, quantify analytes via Poisson statistics, implement a homogeneous immunoassay with high sensitivity, and demonstrate specific, rapid detection of respiratory syncytial virus (RSV), including in spiked human nasal swab samples.

Digital immunoassays have advanced sensitivity up to 10^3-fold over conventional ELISA by counting events in micron-scale compartments but are hindered by multistep operations. Alternative single-particle detection modalities (bright/dark-field, interferometric or fluorescence imaging, SERS, plasmon resonance microscopy) remove amplification but still need complex labeling, chip preparation, and advanced imaging. Homogeneous digital assays based on particle motion can avoid washings but remain laboratory-bound and multistep. Plasmonic nanobubbles (PNBs), transient vapor events around plasmonic nanoparticles induced by pulsed lasers, are sensitive to NP size, shape, and clustering, and provide amplified optical signatures suitable for digital counting without physical partitions. This prior body of work motivates DIAMOND as a simplified, compartment-free digital immunoassay platform leveraging PNB physics for single-particle sensitivity and absolute quantification.

• Principle and setup: DIAMOND employs an optofluidic configuration where NP-labeled analytes flow through a 200 µm ID glass micro-capillary. A pulsed 532 nm pump laser (28 ps, PL2230) generates plasmonic nanobubbles around gold nanoparticles; a continuous 633 nm HeNe probe laser detects transient absorption changes captured by a photodetector and oscilloscope. Aligned beams synchronously irradiate and detect, producing discrete, non-overlapping PNB events per pulse, defining ~16 pL virtual detection zones.
• Flow and scanning: Typical parameters included syringe pump flow 6 µL/min, laser scan speed 1000 µm/s with flow 2500 µm/s to avoid double counting, and acquisitions of 3000 pulses at 50 Hz (~60 s per sample). Event frequencies are reported as f_on (fraction of positive pulses).
• Poisson framework: Analyte concentration c is converted to expected occupancy λ=c·V (V≈16 pL). Event probabilities follow Poisson statistics P(k;λ)=λ^k e^−λ/k!. Calibration-free quantification is based on the linearity of experimental f_on versus theoretical Poisson probability.
• Signal features and gating: For each pulse, PNB signal amplitude and area-under-the-curve (AUC) are extracted (custom MATLAB). Thresholds T=μ+5σ for amplitude and AUC are determined from negative controls to define positive events. Bivariate plots (amplitude vs AUC) support gating for heterogeneous populations and specificity.
• Single-NP and size characterization: AuNPs (15, 35, 50, 75 nm; characterized by UV–vis, TEM, DLS) were tested at controlled λ and laser fluences (2,000–10,000 mJ/cm^2). Size dependence of PNB amplitude and AUC was quantified.
• Heterogeneous mixtures: Detection of rare larger NPs (75 nm, λ=0.5) in a high-background of small NPs (15 nm, λ=240) using thresholding from 15 nm controls to identify large-particle events and comparison to Poisson predictions.
• Homogeneous immunoassay model: SiO2 beads (aminated) incubated with 15 nm citrate-AuNPs to form core–satellite conjugates (30 min, RT). DIAMOND used to count bead-induced clustering events. Calibration via f_on vs λ(SiO2) with LOD computed as 3σ_control/slope.
• Virus immunoassay: 15 nm AuNPs were conjugated to Palivizumab (Synagis) via DTSSP linker; overnight conjugation and washing yielded AuNP-Synagis probes targeting RSV F protein at RT. For improved specificity in complex matrices, probes were backfilled with DTSSP-modified BSA (0.1%) to reduce nonspecific aggregation.
• Sample types: Purified RSV A2 and other respiratory viruses (hMPV, PIV, IVA) tested in buffer; for matrix testing, viruses were spiked (10^2–10^5 PFU/mL) into healthy human nasal swab viral transport medium (VTM). Incubation with probes for 30 min at RT; then DIAMOND measurement at 3000 mJ/cm^2 fluence.
• Comparative method: Digital LAMP (dLAMP) on microwell chips using commercial WarmStart kit and RSV primers. Synthetic RNA and extracted RSV RNA (after buffer extraction and purification) quantified by fraction of positive wells; LOD and calibration curves established.
• Data processing: Custom MATLAB scripts for PNB signal recognition (SNR-based), feature extraction, thresholding, and frequency counting; and for dLAMP image analysis (well recognition, thresholding, counting). Statistical analyses included triplicates, means±SD, two-sample t-tests where applicable.

• Single-NP detection and calibration-free quantification: At low occupancy (e.g., λ=0.04), discrete PNB events correspond to single AuNPs. f_on scales with λ with R^2=0.998; experimental f_on vs Poisson probability yields slope=0.985, R^2=0.998, indicating absolute quantification without calibration.
• Size dependence: PNB amplitude and AUC increase with AuNP diameter, fitting ~d^1.98 and ~d^3.51, respectively, consistent with size-dependent absorption and heat generation.
• Heterogeneous sample analysis: In mixtures (75 nm AuNPs at λ=0.5 within 15 nm AuNPs at λ=240), threshold-based gating identifies large-particle events. Experimental f_on for large NPs matches Poisson predictions with no significant difference, enabling detection of rare large particles (≈1 in 240) without separation.
• Homogeneous assay sensitivity (SiO2 model): Linear region R^2=0.99 for λ(SiO2)=0.0016–0.16. LOD=0.0028 in λ units, equivalent to 1.75×10^5 beads/mL or 290 aM—about 570-fold lower (better) than colorimetric readout. Background-subtracted frequency f_on' vs Poisson probability shows slope=1.007, R^2=0.999, supporting absolute quantification.
• RSV detection in buffer: AuNP-Synagis probes specifically bind RSV; colorimetric LOD=3.6×10^4 PFU/mL; commercial LFIA LOD≈1.6×10^4 PFU/mL. DIAMOND yields a linear range 10^2–10^4 PFU/mL (R^2=0.995) and LOD≈10^3 PFU/mL, representing ~333-fold and ~150-fold sensitivity gains over colorimetry and LFIA, respectively.
• Specificity in complex matrices: Without backfilling, nonspecific aggregation produced false positives for control viruses in buffer. BSA-backfilled probes greatly reduced nonspecific signals in nasal swab VTM while retaining strong RSV signals (f_on≈100% at high titers), improving specificity.
• RSV detection in nasal swabs: Using BSA-backfilled probes, DIAMOND detects RSV spiked into nasal swab samples with LOD≈10^2 PFU/mL. Linear calibration with R^2=0.95 over tested range.
• Comparison to dLAMP: dLAMP calibration with synthetic RNA showed LOD≈2 copies/µL. For RSV extracts from spiked samples, dLAMP LOD≈200 PFU/mL. Based on calibration, 100 PFU/mL corresponds to ~1 RNA copy/µL, indicating DIAMOND achieves single-molecule equivalent sensitivity (LOD ~100 PFU/mL ≈ 1 copy/µL) without nucleic acid amplification.
• Throughput and timing: Typical DIAMOND measurements acquire 3000 pulses at 50 Hz (~60 s per sample), enabling rapid readouts at room temperature.

DIAMOND addresses key limitations of traditional digital immunoassays by eliminating physical compartmentalization, washing, and signal amplification, while providing single-particle sensitivity and absolute quantification through Poisson-based digital counting. The method detects intact viruses directly at room temperature, reducing workflow complexity relative to molecular assays that require extraction, heating, and microdevice loading. Demonstrations include single-NP detection, size-resolved PNB signatures, identification of rare large particles within heterogeneous backgrounds, and a one-step homogeneous immunoassay with sub-femtomolar equivalent sensitivity in a model system. In virus diagnostics, DIAMOND specifically detects RSV over closely related respiratory viruses and achieves LODs down to ~10^2 PFU/mL in spiked nasal swab matrices with blocking, rivaling digital isothermal amplification while requiring minimal processing. The findings validate the hypothesis that PNB-based virtual compartment counting can deliver calibration-free, digital quantification in a simple flow-through format. The approach is broadly applicable to NP-labeled targets and may be extended to other pathogens and biomarkers. Future enhancements discussed by the authors—benchtop integration with compact nanosecond lasers (up to 2 kHz), microfluidic flow focusing to increase sampling efficiency, and operation near PNB thresholds to preferentially detect clusters—could improve sensitivity, dynamic range, and throughput. Additionally, strategies for recognizing events corresponding to defined NP multiplicities and for multiplexed detection are poised to enable analyte panels and protein biomarker assays.

The study introduces DIAMOND, a plasmonic nanobubble-based, compartment-free digital immunoassay platform that enables single-nanoparticle detection, absolute quantification via Poisson statistics, and sensitive homogeneous assays without washing or amplification. Applied to RSV, DIAMOND achieves LODs of ~10^3 PFU/mL in buffer and ~10^2 PFU/mL in spiked nasal swabs, with specificity improved by BSA backfilling, and performance comparable to digital LAMP (single-molecule-equivalent sensitivity). The platform offers rapid (∼1 min readouts), room-temperature detection of intact viruses and has potential as a general framework for ultrasensitive diagnostics. Future work should focus on benchtop instrument integration with higher repetition-rate lasers, microfluidic flow focusing to enhance sampling efficiency and dynamic range, threshold-tuned operation modes to target clusters, and development of multiplexed assays and strategies tailored for smaller protein analytes.

• Dynamic range limited by event counts: With 3000 pulses per run, the digital range spans roughly 2 orders of magnitude; additional pulses or higher repetition-rate lasers are needed to expand range and improve LOD.
• Sampling efficiency: In the current capillary geometry, only a fraction of the cross-section is interrogated (∼20–40% along and 5–10% orthogonal), reducing effective event counts and sensitivity; flow focusing is proposed to mitigate this.
• Matrix effects and specificity: In complex samples, unblocked probes can cause nonspecific aggregation and false positives; BSA backfilling is required to maintain specificity.
• Instrumentation: Current setup uses research-grade picosecond lasers; translation to compact, lower-cost nanosecond systems is proposed but requires engineering validation.
• Analyte scope: Detecting small protein biomarkers is challenging due to limited NP binding; requires diluted operation (low λ), robust gating to distinguish single vs multiple NP events, and potentially refined probe design.
• Operational mode dependence: Sensitivity and selectivity can depend on laser fluence relative to PNB thresholds; optimization is needed for different targets and assay formats.