Surface Plasmon–Phonon Hyperspectral Imaging: Exploring a New Pathway for Viral Protein Detection

Science Advances has published an important research achievement on mid-infrared hyperspectral imaging (MIR-HSI) technology. By integrating surface plasmons with surface phonons, the research team constructed a novel dual-mode nanoantenna system, successfully achieving high-sensitivity and high-spatial-resolution imaging of complex biomolecules, demonstrating strong potential particularly in viral protein recognition.

Technical Core: Dual-Mode Nanoantennas and Polarization Control

The research team designed an asymmetric cross-shaped bilayer nanoantenna structure, with a gold plasmonic antenna on the top layer and a silicon dioxide phononic antenna on the bottom layer. By controlling the polarization direction of the incident light, surface plasmon modes and surface phonon modes can be selectively excited. The plasmonic mode is mainly used to enhance the detection of molecular vibrational signals, while the phononic mode responds to changes in the refractive index of molecules. Together, they form complementary dimensions of information.

Key Breakthrough: Addressing Spectral Overlap to Improve Identification Accuracy

In the detection of pathogens such as SARS-CoV-

2, the infrared absorption spectra of different proteins or variants often exhibit severe overlap, posing challenges for accurate identification. In this study, by simultaneously acquiring plasmonic signals and phononic signals and processing them with a multimodal deep neural network, the researchers effectively distinguished mixed samples of two highly overlapping SARS-CoV-2 spike proteins, significantly improving classification accuracy.

Application Demonstrations: From Laboratory Samples to Clinical Saliva Testing

The study validated the practicality of the system at multiple levels:

• Protein Type Classification: Achieved a classification accuracy of 93.4% for mixed spike proteins.

• Concentration Prediction and Spectral Reconstruction: Using deep learning models, the concentrations of individual components in mixed samples were predicted, and their individual spectra were reconstructed.

• Single-Monolayer Hyperspectral Imaging: Successfully performed spatial distribution imaging of protein single-molecule layers immobilized on the device.

• Clinical Saliva Sample Analysis: Saliva samples from SARS-CoV-2–infected individuals were tested, revealing trends in viral protein signals over the course of disease progression.

Technical Characteristics and Significance

The main advantages of this method are reflected in the following aspects:

• Label-free detection: Directly detects the intrinsic infrared characteristics of proteins, avoiding complex labeling procedures.

• High sensitivity: The nanoantenna structure enhances light–matter interaction, facilitating low-concentration detection.

• Spatial information acquisition: Combined with hyperspectral imaging, it enables visualization of target molecule distributions.

• Rich information dimensions: Dual plasmonic and phononic modes provide complementary spectral features, aiding in the analysis of complex samples.

Outlook

This work successfully introduces surface phonon polaritons into the field of mid-infrared hyperspectral bioimaging. By combining nanophotonics with artificial intelligence algorithms, it provides a new imaging-guided molecular screening tool for pathogen detection, proteomics analysis, and drug screening. In the future, with further optimization of device fabrication processes, such technologies are expected to play an active role in biomedical research and clinical diagnostics.