Residence time investigation of artificial silver ores in heap leaching using cyanide lixiviant

• Main Author: Gibson, Borbor Auzzel Kwaku Kudar
• Other Authors: Petersen, Joachim
• Format: Dissertation
• Language: English
• Published: Faculty of Engineering and the Built Environment 2021
• Subjects: Engineering
• Online Access: http://hdl.handle.net/11427/32642

Heap Leaching has gained much relevance in the processing of low-grade mineral resources - ores considered uneconomical for beneficiation through conventional concentration and tank leaching. However, it is at the same time characterized by extended leaching periods due to slow mineral conversion and low rates of recovery as a major challenge. Interactions within the heap bed are not fully understood as chemical leaching and hydrodynamics interact in a complex manner. To study these interactions, a number of investigations have focused on the hydrodynamic interaction using conventional residence time distribution (RTD) studies in laboratory columns. In these RTD studies, the flows of tracer exiting through the effluent stream provide information of its paths, where some flows might relate to fast movement, slow convoluted channels, or micro/macro stagnated regions. This information is usually interpreted through simplified reactor models representing the bed as a combination of plug flow reactors (PFRs), continuous stirred tank reactors (CSTRs) and dead zones. Using columns as reactors to approximate heap leaching on a laboratory scale, it is anticipated that the RTD flow distribution response should be similar to the distribution of a PFR with associated dead zones. While some literature sources have alluded to the response in columns being similar to a plug flow response, recent sources using a similar hydrodynamic RTD approach reported column reactor distribution resembling flow more typical of a continuous stirred tank, CSTR, system instead of plug flow. Given that packed ore beds are not agitated, this appears paradoxical. It is hypothesised that the CSTR-like response is a result of the distribution of convoluted flow channels through the ore bed, which perform overall like a bundle of PFRs of different lengths. To discern the two patterns the use of a reactive leaching on a well characterized ore material is proposed. Therefore, this work aims to study simultaneously the flow and leach performance of a laboratory column reactor, utilizing a novel reactive leaching approach with a lixiviant amendable to a well-characterized homogeneous solid material. The objective of this work is to establish flow distribution performance in packed bed columns and correlate such performance to the ultimate extraction from the packed bed. The study was performed using a nonreactive solution tracer (potassium nitrate) to characterise flow through a column packed with an artificial silver ‘ore' (silver grains embedded in concrete), followed by a reactive leaching study using sodium cyanide which would leach the silver. The artificial silver ore was developed with the aim to exhibit four ideal properties, namely homogeneous porosity, uniform grain size, homogenous dispersion of the grains throughout the ore, and even grade distribution of the different size fractions after crushing. Leaching and micro-XCT characterization studies were performed in order to determine the leaching properties of the artificial silver ore and validate the ore corresponds to these ideal properties. This validation was indeed achieved; however, the inner-particle pores were not found to be continuous at the scale of resolution of the instrument (4.8 microns). Poor extraction from the coarse particles in both the leaching characterization and reactive column leaching investigations suggested that this network was not well established and may exist only at the sub-micron scale. Leach tests were performed on individual particle sizes in both shake flask and circulating bed reactor tests. Extraction from the leaching of the coarse synthetic silver ore particles was observed to be very low relative to dissolution from pure silver metal grains. Diagnostic shrinking core and the extended mixed topology models were used to determine the controlling dissolution mechanism. Both models demonstrated that a diffusion-reaction mechanism governed the dissolution extraction from the large particles. RTD column leaching studies were performed utilizing flow rate and PSD as investigated parameters. The nonreactive tracer produced a step-change flow response that was more similar to a characteristic plug flow type distribution but showed distinct deviations towards CSTR behavior, especially for the beds containing a higher degree of fines. Reactive column leaching experiments were performed under similar conditions as the nonreactive RTD, introducing a step-change of the cyanide reagent. Rapid silver leaching occurred initially, but equally rapidly declined to very low rates. The leach curves were interpreted by translating the information obtained from the nonreactive RTDs into a distribution of parallel plug flow channels. The extent of reaction for each of these channels is derived from the surface reaction model for the individual size classes, put together for the corresponding PSD in each experiment. RTD specific PSDs tested using this approach assumed that longer residence times correspond to the prevalence of finer material. The validity of the approach was tested by comparing the extraction determined from the particle leaching kinetics studies to the reactive column data through modelling. The model is based on the weighted average leaching from a population of particles, calibrated against kinetic models formulated for individual size classes. This model is further linked to a distribution of flow channels determined from the RTD studies. The prediction of the model did not compare too well against the raw silver dissolution data of the columns. This was attributed to the model having been calibrated against kinetic data that did not fully consider the smaller size classes below -4/+2.8 mm – considered to be the key source of rapid surface reaction in the packed bed. Although the concept proposed in this project was not fully proven, further test work is recommended to expand on the approach presented here.