Fig. 1. (a) Schematic of the metagrating with controlled energy distribution among the supported diffraction orders. The meta-atoms of the metagrating are used to control the scattering directivity to a specific diffraction order. (b) Relationship between the diffraction period and bending angle of a metagrating with a normally incident terahertz wave. (c)–(e) The simulated electric field distributions of a metagrating with the energy concentrated at different diffraction angles (α), which is achieved by controlling the diffraction period of the metagrating arrays. (f) Transmission spectra of different diffraction orders T0, T−1, and T+1.

Fig. 2. (a) The inverse design model and (b) its training and testing processes.

Fig. 3. (a) Comparison of the testing dataset prediction results for different models. (b) Prediction performance of different models for different structure parameters. (c)–(f) The prediction errors and spectra obtained via inverse design models with typical design parameters. The insets in (c)–(f) are the FEM-simulated spectra and corresponding predicted spectra of the inverse design.

Fig. 4. (a) Implementation procedure of the inverse design for metadevices. (b) Process of fabrication and characterization for 3D printing of the predicted metagrating.

Fig. 5. (a) Fabricated metagrating. (b) Optical microscope images of the fabricated metagrating. (c) Size of the fabricated metagrating. (d) Experimental system for measurement of the transmission spectra. (e)–(h) FEM-designed transmission spectra and corresponding inversely designed spectra used in the experiments.

Fig. 6. (a) Schematic description of terahertz metagrating for microorganism sensing. (b) Measured transmission spectra of the samples filled with air (blank), different fungi, and bacteria. (c) Analysis of resonance peaks for different microorganisms.