Size Distributions of Gold Nanoparticles in Solution Measured by Single-Particle Mass Photometry

Abstract

Specialized applications of nanoparticles often call for particular, well-characterized particle size distributions in solution, but this property can prove difficult to measure. Highthroughput methods, such as dynamic light scattering, detect nanoparticles in solution with an efficiency that scales with diameter to the sixth power. This diminishes the accuracy of any determination that must span a range of particle sizes. The accurate classification of broadly distributed systems thus requires very large numbers of measurements. Mass-filtered particle-sensing techniques offer a better dynamic range but are labor-intensive and so have low throughput. Progress in many areas of nanotechnology requires a faster, lower-cost, and more accurate measure of particle size distributions, particularly for diameters smaller than 20 nm. Here, we present a tailored interferometric microscope system, combined with a high-speed image-processing strategy, optimized for real-time particle tracking that determines accurate size distributions in nominal 5, 10, and 15 nm colloidal gold nanoparticle systems by automatically sensing and classifying thousands of single particles sampled from solution at rates as high as 4000 particles per minute. We demonstrate this method by sensing the irreversible binding of gold nanoparticles to poly-Dlysine functionalized coverslips. Variations in the single-particle signal as a function of time and mass, calibrated by TEM, show clear evidence for the presence of diffusion-limited transport that most affects larger particles in solution.

Enhancing Nanoparticle Detection in Interferometric Scattering (iSCAT) Microscopy Using a Mask R-CNN

“…The intensity of the signal detected is determined by the superposition of the scattered and reflected lights I normali normalS normalC normalA normalT ∝ false| E normalr + E normals false| 2 = E normali 2 [ r 2 + s 2 + 2 r false| s false| cos false( ϕ false) ] where I iSCAT is the intensity at the detector; E r , E S , and E i are the reflected, scattered, and incident electric fields, respectively; r is the fraction of E i reflected; s is the fraction of E i scattered; and ϕ is the phase difference between the reflected and scattered lights due to a difference in optical path length. According to Mie scattering theory (applicable when the wavelength of light is of the order of the nanoparticle size), the magnitude of s scales linearly with the volume of the particle ( d NP 3 ), , and in the limit of small scattering objects, reflected light dominates the detected signal making the pure scattering contribution negligible ( r 2 ≫ s 2 ) eq then reduces to I normali normalS normalC normalA normalT = E normali 2 [ r 2 + 2 r false| s false| cos false( ϕ false) ] …”

“…In the past two decades, interferometric scattering (iSCAT) microscopy has enabled imaging and tracking of nano-objects such as gold nanoparticles, − viruses, and individual proteins − for dynamic studies on cell membranes, molecular motors, and quantitative mass/size measurements. − As a scattering-based technique, imaging of nano-objects in iSCAT experiments does not require fluorescent labeling . This advantage mitigates challenges inherent to fluorescence-based techniques such as photobleaching and photoblinking .…”

“…According to Mie scattering theory (applicable when the wavelength of light is of the order of the nanoparticle size), the magnitude of s scales linearly with the volume of the particle (d NP 3 ), 13,28 and in the limit of small scattering objects, reflected light dominates the detected signal making the pure scattering contribution negligible (r 2 ≫ s 2 ). 28 eq 1 then reduces to…”

“…By balancing images that follow a nanoparticle adsorption event with those that precede the event, ratiometric processing capitalizes on time-dependent variations in scattering signals during the collection of iSCAT videos to isolate adsorption events in background-suppressed images. In doing so, it enables the measurement of the scattering contrast and the binding times of analytes and has been previously used to detect nanoparticles 13 and proteins. 11 To apply ratiometric processing to detect nanoparticle adsorption events, we began by defining a time-binning window of N ratio = 5 frames.…”

“…If this probability was greater than 0.3, the pixel was classified as belonging to a particle, resulting in a segmented binary image that was then labeled in object detection. Although the Haarlike feature algorithm was used as a control in this study, it would be valuable in future studies to test other state-of-the-art object detection methods, such as the circular Hough transform used by Melo 13 et al, to see how performance compares to that of the mask R-CNN when the background scattering is high.…”