A statistical analysis of the factors affecting thermal debinding in powder injection molded parts

Abstract

Powder injection molding (PIM) technology allows the creation of complex ceramic geometries by combining ceramic powder with a binder, typically a thermoplastic, to form a homogeneous moldable feedstock, which is then removed prior to densification. The main stages in PIM processing include powder/binder selection, compounding, molding, binder removal, and densification. Although all processing stages in PIM are important in achieving a successful final part, the removal of the binder plays a more significant role as it is the slowest stage, and so requires more careful control [1]. Traditionally, thermal debinding or pyrolysis has been the method most widely practiced. However, as pyrolysis introduces porosity, this stage may present defects such as bloating, cracking and/or void formation that cannot be fixed in later stages [2, 3]. The purpose of this work is to illustrate the importance of the early stages of binder removal with respect to defect formation as well as to determine the critical process variables governing the early stages of debinding in a quantitative manner. Previous studies have revealed the importance of binder composition [3, 4], ceramic powder selection [1], debinding atmosphere [5, 6], heating rate or temperature [7, 8] and specimen geometry [1, 8] on thermal debinding of plastically formed ceramic bodies. However, these investigations have been generally qualitative and/or case-specific, and have not provided an experimentally verified theoretical basis for process control and materials selection [9]. Conversely, this study uses a factorial design methodology to determine the statistical significance of powder type, binder wax addition, debinding atmosphere and specimen thickness on the initial stages of debinding by pyrolysis of typical PIM parts. The samples used in the present study were prepared following an established procedure based on the initial stages of powder injection molding technology, a detailed description of which can be found in [10]. The specimens consisted of 60-vol% alumina and 50-vol% nickel powder in either a 3 : 1 mass ratio of polypropylene: polyethylene wax or a 3 : 1 ratio of polypropylene: polypropylene wax binder. These solids loadings were found to yield a homogenous feedstock at slightly below the critical solids loading for these systems [10]. The alumina (Calcined Alumina C-90LSB: Alcan Chemicals) and nickel (Nickel: Inco Nickel Powder Type 123) powders were chosen primarily because they are representative of the materials and particle sizes used in typical PIM applications, but also because of the large particle size difference between the two (1.04 μm and 8.51 μm, respectively), allowing the effect of powder type to be investigated. The major binder constituent, polypropylene (PM6100: Himont) was selected since it was a thermoplastic, and therefore suitable for PIM processing [4]. Two minor binder constituents, polypropylene wax (Epolene N-15: Eastman-Kodak) and polyethylene wax (Epolene N-14: Eastman-Kodak), were chosen because of their low molecular weights and their ability to reduce binder viscosity [4]. In practice pyrolysis is performed in air whenever possible for convenience; however, some specimens cannot be debound in air because of the possibility of oxidation, so that the effect on debinding time using a nitrogen atmosphere was of practical importance. Debinding was performed in a long tube furnace equipped with temperature control feedback. Weight loss measurements were taken at regular intervals so that a plot of weight loss vs time could be established for each sample studied [10]. Preliminary debinding experiments on the samples revealed that defects such as bloating occurred at very early stages of binder removal by pyrolysis in either air or nitrogen environments [10]. Specifically, in the case of 60-vol% alumina in the polypropylene/polyethylene wax binder, defects were found to occur in samples when as little as 5-mass% binder was removed in air. For 50-vol% nickel samples, large pores were observed in samples when as little as 4-mass% binder had been removed in nitrogen. For the polypropylene/polypropylene wax binder loaded with 60-vol% alumina, defects occurred when as little as 3.5-mass% binder was removed. These defects occurred when enough binder was still present in the samples or before a significant amount of interconnected open had developed porosity in the samples. Scanning electron micrographs (SEM) of partly burned out alumina and nickel samples provided some insight into when the transition between saturated flow and open pore flow occurred. This transition was important in defining an upper bound for binder loss at which enough binder was still present in the sample to allow unwanted bloating. Samples were pyrolyzed in air at 190 ◦C and their fractured cross-sections examined under SEM. In all cases, the samples were fractured in liquid nitrogen to avoid plastic deformation of the binder during bending. Fig. 1a and b show micrographs of 50-vol% nickel samples after 10 and 45 mass% binder loss, respectively. A comparison of the figures reveals