Artificial Intelligence-Aided Massively Parallel Spectroscopy of Freely Diffusing Nanoscale Entities

Massively parallelspectroscopy (MPS) of many single nanoparticlesin an aqueous dispersion is reported. As a model system, bioconjugatedphoton-upconversion nanoparticles (UCNPs) with a near-infrared excitationare prepared. The UCNPs are doped either with Tm3+ (emission450 and 802 nm) or Er3+ (emission 554 and 660 nm). TheseUCNPs are conjugated to biotinylated bovine serum albumin (Tm3+-doped) or streptavidin (Er3+-doped). MPS is correlatedwith an ensemble spectra measurement, and the limit of detection (1.6fmol L-1) and the linearity range (4.8 fmol L-1 to 40 pmol L-1) for bioconjugatedUCNPs are estimated. MPS is used for observing the bioaffinity clusteringof bioconjugated UCNPs. This observation is correlated with a nativeelectrophoresis and bioaffinity assay on a microtiter plate. A competitiveMPS bioaffinity assay for biotin is developed and characterized witha limit of detection of 6.6 nmol L-1. MPS from complexbiological matrices (cell cultivation medium) is performed withoutincreasing background. The compatibility with polydimethylsiloxanemicrofluidics is proven by recording MPS from a 30 & mu;m deep microfluidicchannel.