January AMP_Digital

A D V A N C E D M A T E R I A L S & P R O C E S S E S | J A N U A R Y 2 0 1 9 1 9 PHYSICAL METHODS Several techniques are available to characterize particle properties. For granular aggregate materials, the sim- plest technique to determine particle size distribution is sieve analysis [4,5] . After sifting through meshes of gradu- ated sizes, the percent-by-mass of each size range and a fineness modulus of the material trapped by each sieve is calculated. The technique requires moving samples offline for analysis and it is unable to provide more detailed shape information. More sophisticated techniques such as acoustic emission avoid some of the limitations of sieve analysis. For example, acoustic emission is rapid, generates a continuous size distribu- tion, and does not require removing samples from the test apparatus [6] . It can measure very small (micron range) particle sizes, but is limited to a maxi- mum particle size. However, like sieve analysis, acoustic emission is limited to size measurements. Dynamic light scattering, also known as photon correlation spectros- copy [7] , is used to make size measure- ments (down to the nanometer range [8] ) of particles in suspension. As particles diffuse through the suspension, they are illuminated by a laser beam and scatter light. Particle size is determined mathematically, which requires assum- ing particular shape(s), so true shape information is lost. MICROGRAPH-BASED PARTICLE ANALYSIS The main benefit of using micro- graphs to analyze particles over phys- ical techniques is that particle shape information is not lost. Once particles are properly identified in a micrograph, many different shape descriptors such as aspect ratio and roughness can be reported. Using morphological image processing, more complicated analysis can be designed to report metrics such as number of satellite particles per par- ent, amount of cracking per particle, and many others (Fig. 3). Another bene- fit of micrograph analysis is that there is no minimum or maximum particle size that can be analyzed. This flexibility re- quires the use of more powerful and ex- pensive microscopes to achieve smaller resolutions. The basic image processing techniques are not scale dependent. Micrograph particle analysis, un- like physical techniques, does not re- quire particles to be free to move relative to each other. Particles can be measured as long as they are distin- guishable in the image, even if they are fixed in a concretion or composite ma- terial, such as after the molding step of a metal injection molding process [9] . This enables further analysis of particle size and shape after production. In ad- dition, measuring particle anisotropy across the image enables investigating nonuniform mechanical properties of the part. Particle shape information is cru- cial in powder-based additive manu- facturing techniques, as shape charac- teristics directly influence flow rate and packing density, which in turn influence mechanical and thermal properties of the final product [10] . REDUCING ERRORS IN GRAPHITE CLASSIFICATION Micrograph analysis is performed to quantify inclusions and porosity in a material using standard reference micrograph charts. However, simply comparing micrographs to a chart is subjective and introduces human error, leading to wrongly approved and reject- ed parts. Implementing point-count- ing techniques can reduce some bias, but this increases the time needed to quantify a sample. Also, these tech- niques do not eliminate the subjectiv- ity inherent in deciding which feature class is under each point. Using amicro- scope with a digital camera, the micro- graph can be digitized as an image and Fig. 3 — Examples of particles and their features identified using Mipar’s morphological image processing. (a) Original image of particles with satellites; (b) segmented images (green = parent particles, purple = satellite particles); (c) original image of particles with cracks; and (d) segment- ed images (green = particles, purple = cracks within those particles). (a) (b) (c) (d)