Surface-enhanced Raman spectroscopy(SERS)has been developed as a versatile tool for trace-molecule detection and biomolecular analysis by coupled gold or silver nanostructures in the past two decades [1,2].However,SERS suffers from a long-term limitation of application for surface analysis of general materials.That is because the SERS hotspots in inter-particle nanogap generated from coupled nanostructures are spatially tiny(~25 nm2),and the typical materials such as silicon wafer and blades in motors could not be squeezed into hotspots in the inter-particle nanogap.And the SERS hotspots were not specially designed on the surface of probe materials.Furthermore,several interfered Raman signals typically could not be ruled out if we employ the contact-mode SERS based on bare Au or Ag nanoparticles [3].Here,we try to address the fundamental issues associated with plasmon-enhanced Raman spectroscopy(PERS)including SERS,tip-enhanced Raman spectroscopy,and shell-isolated nanoparticle-enhanced Raman spectroscopy for surface analysis.Especially,we discuss new,third-generation PERS hotspots generated by electromagnetic coupling in hybrid structures consisting both PERS-active nanostructures and probe materials [3].Furthermore,we discuss our very recent discovery of third-generation hotspots generated by nanosphere,nanocube or nanobar clusters on a flat metal surface with an area as large as ten thousand nm2(which could be termed as hot domains).The hot domains could be effectively excited by modulating the reflection phase and relative value of intrinsic and radiative decay rate of a metal-insulator-metal metasurfaces [4].The featured hot domains could in principle support a uniformly distributed and giant Raman signals,which could significantly improve the spot-to-spot reproducibility.