Nanoparticle metrology with atomic force microscope (AFM) aims to determine an average particle size from measurements of individual nanoparticles derived by image analysis. This constrains the statistical relevance of the measurement due to the limited number of particles which can be practically imaged and analyzed. Consequently, the number density of particles on samples prepared for particle measurement is an important parameter of sample preparation. A number density that is too low makes it difficult to obtain sufficient measurement statistics, whereas a number density that is too high can result in particle agglomeration on the substrate and limit the area of uncovered substrate that is required to obtain a reliable reference for measuring the particle height. We present imaging and measurement results of a particle number density gradient of 16 nm gold nanoparticles deposited using a gradual immersion process. Results demonstrate how samples with particle number density gradients can facilitate identification of an area on a sample with optimal particle number density for AFM particle metrology and thereby improve measurement efficiency and reliability.