# Optical Microscopy and Deep Learning for Absolute Quantification of Nanoparticles on a Macroscopic Scale and Estimating Their Number Concentration

## TL;DR

This paper introduces a simple method using optical microscopy and deep learning to count nanoparticles in large volumes and estimate their concentration with high accuracy.

## Contribution

The novel method, called EVA, enables absolute quantification of nanoparticles on a macroscopic scale using evaporation and AI-based counting.

## Key findings

- EVA estimated the concentration of Tm3+-doped nanoparticles with a 2.7% relative standard uncertainty.
- The method was validated with polystyrene and silver nanoparticles, showing comparable results to reference methods.
- Theoretical limits like detection and quantification thresholds were analyzed for EVA.

## Abstract

We present a simplistic and absolute method for estimating
the
number concentration of nanoparticles. Macroscopic volumes of a nanoparticle
dispersion (several μL) are dropped on a glass surface and the
solvent is evaporated. The optical microscope scans the entire surface
of the dried droplet (several mm2), micrographs are stitched
together (several tens), and all nanoparticles are counted (several
thousand per droplet) by using an artificial neural network. We call
this method evaporated volume analysis (EVA) because nanoparticles
are counted after droplet volume evaporation. As a model, the concentration
of ∼60 nm Tm3+-doped photon-upconversion nanoparticles
coated in carboxylated silica shells is estimated with a combined
relative standard uncertainty of 2.7%. Two reference methods provided
comparable concentration values. A wider applicability is tested by
imaging ∼80 nm Nile red-doped polystyrene and ∼90 nm
silver nanoparticles. Theoretical limits of EVA such as the limit
of detection, limit of quantification, and optimal working range are
discussed.