**Oleg Popov 1, Irina Talovina 2, Holger Lieberwirth 1,\* and Asiia Duriagina <sup>2</sup>**


Received: 9 January 2020; Accepted: 24 January 2020; Published: 31 January 2020

**Abstract:** Profound knowledge of the structure and texture of rocks and ores as well as the behavior of the materials under external loads is essential to further improvements in size reduction processes, particularly in terms of liberation size. New analytical methods such as computer tomography (CT) were adopted to improve the understanding of material characteristics in rocks and ores relevant to mineral processing, particular the crushing and grinding and the modelling/simulation thereof. Results obtained on the texture and structure of identical samples of rather different rocks and ores (copper ore, granodiorite, kimberlite) are compared by CT with quantitative results from traditional optical microscopy obtained by quantitative microstructural analysis (QMA). While the two approaches show a good agreement of the results in many areas, the measurements with the two different methods also exhibit remarkable differences in other areas, which are discussed further. In conclusion, both methods have their specific advantages starting from sample preparation to the accuracy of information obtained concerning certain parameters of mode and fabric. While sample preparation is faster with CT and information on special distribution of metal minerals is more reliable, the information on mode, grain size and clustering seem to be more precise with QMA. Based on the results, it can be concluded that both methods are comparable in many areas, but in in the field of spatial distribution, they are merely complementary.

**Keywords:** quantitative microstructural analysis; X-ray computed tomography; selective comminution; texture; structure; mineral processing; crushing; grinding

### **1. Introduction**

An increasing demand for raw materials worldwide meets a general trend towards the exploitation of lower grade new ore deposits [1–3]. Thus, ever larger quantities of ore have to be processed to produce the required amounts of concentrates. Not only rising energy costs, per processed ton of ore as well as in absolute terms, but also the accompanying CO2 emissions and the related social license to proceed are reasons for the resource industries to look for smarter solutions, particularly for the energy-intensive comminution processes.

In a study comprising numerous Australian copper mines, it was found that on average, 36 percent of the energy utilized is consumed solely by comminution processes [4]. Comminution processes for copper production alone consume about 1.223 MWh/t Cu [4]. Milling, in particular fine grinding in the area of liberation size, consumes thereby the largest mass specific amounts of energy. Investigation of the distribution of the valuable components in the ore, their grain size as well as the behavior of the rock and ore under certain load is therefore essential, to select the right type, magnitude and frequency of load and thus, respectively, the right milling machine working at appropriate operating parameters [5,6].

The influence of rock and ore properties on the processing characteristics is a well-established fact [7–12]. A profound knowledge of the mineral behavior is required on all comminution steps, be it for the efficient production of preconcentrates by selective comminution on coarse particle size level [13] or the high voltage impulse comminution [14,15], or in the fine grinding. The importance of such an understanding on the microscopic level is emphasized by a number of publications and even global initiatives such as the Coalition for Energy Efficient Comminution (CEEC) [16,17]. It is also essential for the modelling and simulation of comminution processes by numerical methods, such as with the Discrete Element Method (DEM). Modelling of comminution processes with those methods provides fascinating new insights into comminution processes [18–21]. For those methods the calibration of the models for prognostic purposes poses a tremendous challenge if the differences of the mode and fabric of the various ores and rocks cannot be quantified [22–24].

The increased focus on the ore microstructure led to a number of efforts introducing new technologies into the analysis of ore microstructures. In recent years, new technologies such as the scanning electron microscope (SEM)-based Mineral Liberation Analysis (MLA) or Quantitative Evaluation of Minerals by Scanning Electron Microscopy (QEMSCAN) made their way into the mineralogist's laboratories [25,26]. More recently, also Computer Tomography (CT) was applied for mineralogical analyses [27–30]. CT is a nondestructive technique that allows for the visualization of internal structures of objects based on their different X-ray density. Originally, the method was used mainly for medical purposes, but since the 1990s it has also been used in other applications, first in petroleum geology [31–33] and later in process mineralogy [34] and process technology [27,29,30,35].

The investigation of structure and texture of ores is a pretty new but fast growing application area of CT [36,37]. While a number of publications support the usefulness of applying CT for analyzing mineral structures is still missing, a comparison with results acquired by traditional methods, e.g., optical microscopy. This may be caused by the missing quantitative criteria for such a comparison. The Quantitative Microstructural Analysis (QMA), a method based on optical microscopy and proven in numerous industrial applications, shall be used in this paper to compare the results obtained by the new method CT with the proven QMA. Optical microscopy has been used for decades to investigate mineral raw materials using thin or polished sections. Under polarized light, an experienced mineralogist directly identifies the various mineral varieties of the grains. Yet, optical microscopy has its limitations with regard to the image and grain size resolution, and the mineral identification, finally limited by the wave length of the light. QMA will be briefly introduced in the following chapter. A more detailed description can be found for instance in [9,10].

The research presented will therefore present a comparison of structural and textural data, obtained for identical samples of three rather different materials using both methods. This allows to evaluate the capacity of CT for obtaining relevant material data of rocks and ores on microstructural level for modelling, simulation, machine and process design in crushing and milling.

#### **2. Methods and Materials**

#### *2.1. QMA—Microstructural Analysis Based on Optical Microscopy*

Descriptions of texture and structure of rocks are of great value as they can be used in the interpretation of rock formations. Microstructural descriptions are traditionally obtained by geologists using thin or polished sections for petrographic analysis. The traditional method has limitations in describing the full 3D rock structure with quantitative measures and in drawing conclusions and deriving interpretations of the microstructural information.

Quantitative Microstructural Analysis (QMA) is a collective term for a number of methods for analysing the geometry and mechanical properties of microstructural constituents. The objective is usually to investigate the relationship between the microstructure on the one hand and the properties of the rock or its history of origin on the other. Microstructural characteristics are often used for the classification of rocks. Methods for microstructural assessment have long been part of rock research, rock testing and production monitoring. The widening range of possible applications (petrography, mineralogy, metallography, investigation of the structure of building materials, etc.) and the development of theoretical principles (stochastic geometry, digital image processing) have led to QMA becoming a separate field of science over recent years [9,10].

The basis of the QMA of a rock sample is normally a microscope analysis of thin or polished sections. As rock-forming minerals and rock microstructures mostly have a complex three-dimensional structure, the information that can be derived from one-dimensional cut surfaces is often insufficient for the spatial quantitative characterization of the minerals and microstructures of geological materials. A three-dimensional information or the reconstruction of minerals and rock microstructures of geological materials is required. The results of mathematical-petrographic rock characterization are a precondition for predicting the relationships between rock characteristics and relevant product properties or system characteristics (e.g., wear, energy consumption) with the help of mathematical statistical modelling (e.g., multiple regression and correlation analysis) [8,38].

All the strength properties of rocks are directly dependent on the type and degree of mutual bonding of the mineral grains in the microstructure. The greater the specific surface on which the mineral grains are in contact with one another, the more mechanically competent is the rock and the more resistant to the effects of comminution processes.

The degree of grain bonding depends on grain shape and grain size. The cohesiveness of the microstructure rises as grain size decreases and specific surface correspondingly increases. With otherwise identical properties, fine-grained rocks therefore always manifest greater strength than medium and coarse-grained types. The more irregular the grain shape, the greater the grain surface area and the more intense the mutual intergrowth of adjacent individual grains [11].

The QMA usually starts with a proper extraction of a defined number of rock samples from the deposit. According to the QMA approach developed at the Institute of Mineral Processing Machines (Freiberg, Germany), model three polished sections perpendicular to each other are prepared from each rock specimen. These thin sections are evaluated with the help of a polarization microscope, assessed by means of various stereological methods (Figure 1) and classified in a library of thin sections. A mathematical model is used to characterize the rocks quantitatively, based on the calculation of relevant structural and textural characteristic data [13,23].

**Figure 1.** Process flow for QMA rock analysis.

Rock characteristics are mathematically derived in the form of characteristic values (see Table 1). These are first determined completely independently from a specific application, but can be interpreted for use with specific applications in comminution or other processes. In the following, the most important characteristic values gained by QMA (according to [8]) are explained in greater detail:


#### Mode (Volume Percentage of the Mineral Phases)

The volume percentage of a mineral phase is defined as the quotient of the sum of the volume of the individual mineral grains in the sample and the sample volume. It can be most easily determined with the help of the point counting method [39].
