Using Machine Learning and Silver Nanoparticle-Based Surface-Enhanced Raman Spectroscopy for Classification of Cardiovascular Disease Biomarkers

Characterizing complex biofluids using surface-enhanced Raman spectroscopy (SERS) coupled with machine learning (ML) has been proposed as a powerful tool for point-of-care detection of clinical disease. ML is well-suited to categorizing otherwise uninterpretable, patient-derived SERS spectra that contain a multitude of low concentration, disease-specific molecular biomarkers among a dense spectral background of biological molecules. However, ML can generate false, non-generalizable models when data sets used for model training are inadequate. It is thus critical to determine how different SERS experimental methodologies and workflow parameters can potentially impact ML disease classification of clinical samples. In this study, a label-free, broadband, Ag nanoparticle-based SERS platform was coupled with ML to assess simulated clinical samples for cardiovascular disease (CVD), containing randomized combinations of five key CVD biomarkers at clinically relevant concentrations in serum. Raman spectra obtained at 532, 633, and 785 nm from up to 300 unique samples were classified into physiological and pathological categories using two standard ML models. Label-free SERS and ML could correctly classify randomized CVD samples with high accuracies of up to 90.0% at 532 nm using as few as 200 training samples. Spectra obtained at 532 nm produced the highest accuracies with no significant increase achieved using multiwavelength SERS. Sample preparation and measurement methodologies (e.g., different SERS substrate lots, sample volumes, sample sizes, and known variations in randomization and experimental handling) were shown to strongly influence the ML classification and could artificially increase classification accuracies by as much as 27%. This detailed investigation into the proper application of ML techniques for CVD classification can lead to improved data set acquisition required for the SERS community, such that ML on labeled and robust SERS data sets can be practically applied for future point-of-care testing in patients.

A one-step, tunable method of selective reactive sputter deposition as a wrinkling approach for silver/polydimethylsiloxane for electrically conductive pliable surfaces

The wrinkle period and morphology of a metal thin film on an elastic substrate is typically controlled by modifying the substrate before carrying out additional metal deposition steps. Herein, we show that a simultaneously selective and reactive sputtering plasma that modifies the surface of a polydimethylsiloxane (PDMS) substrate while not reacting with the metal during the deposition process decreases the wrinkle wavelength and induces additional wrinkling components and features such as ripples or folds. The selective reaction of the nitrogen plasma with PDMS functionalizes the siloxane surface into silicon oxynitride. This hardens the immediate surface of PDMS, with a quadratic increase in the Young's modulus as a function of the sputtering flow ratio. The increase in the critical strain mismatch and the corresponding presence of folds in the nitrogen-modified wrinkled silver film form a suitable plasmonic platform for surface-enhanced Raman spectroscopy (SERS), yielding an enhancement factor of 4.8 × 10⁵ for detecting lipids. This enhancement is linked to the emergence of electromagnetic hotspots from surface plasmon polariton coupling between the folds/wrinkles, which in turn enables the detection of low concentrations of organics using SERS. Furthermore, when strained, the nitrogen-modified wrinkles enhance electrical conductivity by a factor of 12 compared with unmodified films. Finally, the optical properties of the substrate can be tuned by altering the N₂ content. The simple addition of nonreactive nitrogen to silver sputtering enables simultaneous PDMS hardening and growth of the silver film and together provide a new avenue for tuning wrinkling parameters and enhancing the electrical conductivity of pliable surfaces.

Protein capture and SERS detection on multiwavelength rainbow-trapping width-graded nano-gratings

Surface-enhanced Raman scattering (SERS) substrates with multiwavelength rainbow-trapping properties hold the potential for a one-size-fits-all platform for rapid and multiplexed disease detection. We present the first report on the utilization of rainbow-trapping width-graded nano-gratings, a new class of chirped metamaterials, to detect protein biomarkers. Using cytochrome c (Cc), a charged analyte with inherent difficulty in adsorbing onto sputtered silver films, we investigated methods of binding Cc on the silver nano-grating in order to improve the SERS signal strength at both 532 and 638 nm excitation. Cc was not detectable on the Ag nano-gratings without surface functionalization at 1μM concentration. Upon charge reversal functionalization of the Ag nano-gratings, 1μM Cc was detectable albeit not reliably. By further crosslinking 1μM Cc to the functionalized Ag nano-gratings, the analyte-capture detection scheme greatly improved the SERS signal strength and reliability at both excitation wavelengths and allowed for quantification of their coefficients of variation with values down to 27%.

Integrated Assembly and Photopreservation of Topographical Micropatterns

Micromanipulation techniques that are capable of assembling nano/micromaterials into usable structures such as topographical micropatterns (TMPs) have proliferated rapidly in recent years, holding great promise in building artificial electronic and photonic microstructures. Here, a method is reported for forming TMPs based on optoelectronic tweezers in either "bottom-up" or "top-down" modes, combined with in situ photopolymerization to form permanent structures. This work demonstrates that the assembled/cured TMPs can be harvested and transferred to alternate substrates, and illustrates that how permanent conductive traces and capacitive circuits can be formed, paving the way toward applications in microelectronics. The integrated, optical assembly/preservation method described here is accessible, versatile, and applicable for a wide range of materials and structures, suggesting utility for myriad microassembly and microfabrication applications in the future.

Adiabatic mode transformation in width-graded nano-gratings enabling multiwavelength light localization

We delineate the four principal surface plasmon polariton coupling and interaction mechanisms in subwavelength gratings, and demonstrate their significant roles in shaping the optical response of plasmonic gratings. Within the framework of width-graded metal–insulator-metal nano-gratings, electromagnetic field confinement and wave guiding result in multiwavelength light localization provided conditions of adiabatic mode transformation are satisfied. The field is enhanced further through fine tuning of the groove-width (w), groove-depth (L) and groove-to-groove-separation (d). By juxtaposing the resonance modes of width-graded and non-graded gratings and defining the adiabaticity condition, we demonstrate the criticality of w and d in achieving adiabatic mode transformation among the grooves. We observe that the resonant wavelength of a graded grating corresponds to the properties of a single groove when the grooves are adiabatically coupled. We show that L plays an important function in defining the span of localized wavelengths. Specifically, we show that multiwavelength resonant modes with intensity enhancement exceeding three orders of magnitude are possible with w < 30 nm and 300 nm < d < 900 nm for a range of fixed values of L. This study presents a novel paradigm of deep-subwavelength adiabatically-coupled width-graded gratings—illustrating its versatility in design, hence its viability for applications ranging from surface enhanced Raman spectroscopy to multispectral imaging.

Tunable rainbow light trapping in ultrathin resonator arrays

Rainbow light trapping in plasmonic devices allows for field enhancement of multiple wavelengths within a single device. However, many of these devices lack precise control over spatial and spectral enhancement profiles and cannot provide extremely high localised field strengths. Here we present a versatile, analytical design paradigm for rainbow trapping in nanogroove arrays by utilising both the groove-width and groove-length as tuning parameters. We couple this design technique with fabrication through multilayer thin-film deposition and focused ion beam milling, which enables the realisation of unprecedented feature sizes down to 5 nm and corresponding extreme normalised local field enhancements up to 10³. We demonstrate rainbow trapping within the devices through hyperspectral microscopy and show agreement between the experimental results and simulation. The combination of expeditious design and precise fabrication underpins the implementation of these nanogroove arrays for manifold applications in sensing and nanoscale optics.