Since the first report on the spectral detection of single metal nanoparticles by single-particle dark-field spectroscopy in 2000 [1], the single-particle plasmon sensing method has evolved into an important tool to investigate physical and chemical processes on nanoscale interfaces [2–6]. In a typical single-particle plasmon sensing experiment, changes in spectral characteristics (resonance frequency Δω_res, linewidth ΔΓ, and scattering intensity ΔI_sca) of the single metal nanoparticle are monitored. Then the relative changes of the spectral characteristics are interpreted as the changes in the particle’s properties or its local environment. Foerster et al. [7] developed an elegant method to disentangle the changes of the electron density and refractive index on the surface of the nanoparticles. In a recent Letter, Du et al. [8] proposed a deep learning model to determine the layer refractive index and thickness from the whole scattering spectrum of a nanoparticle. However, this method ignores the local curvature of the layer and the surface charging effect in its training data. The polymers poly(allylamine hydrochloride) (PAH) and poly(sodium 4-styrenesulfonate) (PSS) used in their experiments have surface charging effects on nanoparticles, which usually leads to a relatively large change in spectral characteristics (e.g., blueshift or redshift of ω_res, depending on the surface charge of the layer). Du et al. [8] do not describe how to account for such effects in the model. Therefore, their method might not be an appropriate way to evaluate changes in spectral characteristics in single-particle plasmon sensing experiments. Moreover, Du et al. [8] could use atomic layer deposition (ALD) to further verify whether their model is effective for dielectric layers without any local curvature and the surface charging effect. For such an experiment, it should be noted that the interface damping caused by the dielectric layer can produce additional changes in spectral characteristics.

As a general guide, in single-particle plasmon sensing experiments, the spectrum for the first layer of coated nanoparticles or surface functionalized nanoparticles should be used as the reference spectrum [6]. Moreover, in some experiments, an atomic force microscope (AFM) should also be used to image the morphology of nanoparticles before and after layer deposition; this allows the surface morphology of the nanoparticles to be checked. It is also very important to obtain a stable reference spectrum for the experiments using “naked” gold nanoparticles as sensors. According to the authors, single crystal (SC) gold nanoparticles stabilized by cetyltrimethylammonium bromide (CTAB) prepared using the silver-assisted method showed slow changes in the spectral characteristics during the first hour in the liquid environment—even for gold nanorods prepared many years previously, because of silver ions (Ag⁺) leaving the nanoparticle’s surface [9]. Before any sensing experiments, it is feasible to thoroughly clean the particles in a liquid environment. Taking into account these effects is of high importance for all single-particle plasmon sensing applications that rely on an accurate evaluation of the spectral characteristics of single metal nanoparticles.