Rock flour

Bridging the gap between engineering and geology, the aim of minerals engineering is to provide a better recovery in terms of the exploitation of deposits. Mineralogists from Liege and Madrid have specialized in the quantitative analysis of digital microscopic images. The objective is to better understand the intergrowth textures of the different minerals in the same rock. These techniques make it possible not only to understand how to better liberate the minerals captured in the rock for a better recovery, but also to rethink the entire recycling process for used metals. 

The journey from the mines to the copper pipe that connects the central heating circuit in our homes is a long and arduous one. This is because copper, like most metals, is not found conveniently separated from rock particles simply waiting to be harvested. On a scale of the order of tens of microns, the different minerals present in the same rock are interwoven. Between the extraction of the rock and the metallurgy phase, there is a rigorous and critical stage which involves mineral liberation and the separation of the different particles to be found in the same lump of rock.

Since 1986, Eric Pirard, Professor of Mineral Resources and Geo-imaging at the University of Liege has brought his precious scientific knowledge to bear on this phase of production for metals. Assisted by Laura Pérez-Barnuevo, a doctoral student at the Polytechnic University of Madrid, he has studied the liberation characteristics of chalcopyrite (CuFeS₂, a mineral composed of copper, iron and sulphur) from a mineral deposit located in Zambia (1), with a view to proposing and marketing an automatic digital imaging technique. This mineral presents three advantages. It exists in great quantities on our planet (copper exists in many sulphide forms, and is interesting to extract and study). Chalcopyrite is yellow in sharp contrast with other minerals and is therefore easy to identify and observe. Finally, this sulphide presents complex intergrowth textures with the other minerals that make up the rock in which it is found. In other words, copper is difficult to extract but displays a large mineralogical and textural diversity. These characteristics have led the researcher to concentrate on this mineral in order to develop a quantitative tool that might eventually be more widely used.  

Worse than a needle in a haystack

The theory of liberation involves the separation of the different molecules in a rock. It literally consists of separating the minerals from each other. Gold serves as a good example. “In the collective imagination we think of a prospector panning for gold fragments in a river, but that was an artisanal activity practiced during the gold rushes of the 19th century”, Eric Pirard reminds us. Most gold-extraction, as is the case with other minerals, takes place in large factories that process hundreds of thousands of tons of material. With a ratio of barely 5 grams of gold to every tonne of rock “you may just as well search for a needle in a haystack. It would be easier to find a needle in fact, as it does not stick to the hay. All you need is a magnet”, the researcher adds, with irony.

Evidently, a magnet cannot be used to extract gold from a tonne of rock. Firstly, the five grams of gold are not bunched together but are spread and trapped throughout the rock. And these particles of gold rarely exceed one-tenth of a millimetre.

From a block of stone to rock flour 

In order to separate gold from other minerals it is necessary to break up the rock. It is then ground into a fine flour which requires the use of heavy duty equipments. First, the blocks of stone are broken up in an industrial crusher that looks like a gigantic nut crusher. The pieces of rock which are a few centimetres in size are then placed in a ball mill (a large rotating drum filled with steel balls which are a speciality of Magotteaux, a Belgian company based in Vaux-sous-Chèvremont). A fine flour, the grains of which are only a few tenths of a millimetre thick, is then produced by the ball mill. 

The challenge facing the production process is to produce flour that is sufficiently fine so that only particles of gold and rock are left. “All mineralogists want to reach a liberation level of 100% but that would be the ideal level. This is the aim of the research we are conducting. We want to be able to maximize this liberation which, at the present time, is not always efficient”.

The extent of this liberation is crucial because, at this stage of production, the flour is still only a mishmash of the different minerals which made up the crushed rock. “Before the metallurgy phase, the minerals that will be used during the other stages still need to be separated”.

There are several separation techniques. For gold, differences in density are used. A 1 mm3  particle of gold has a mass of 19 milligrams and that of the rock is 2.5 milligrams. For other minerals containing sulphide such as chalcopyrite, the technique used is flotation. “We use the hydrophobic property of most sulphide minerals whereas the others are hydrophilic. We add a foaming agent to a vat containing a cubic meter of water and the foaming agent rises to the surface creating air bubbles. We then add the mineral particles from the crushing process,” explains the researcher. The hydrophobic particles tend to adhere to the air bubbles. In this way they are separated from the rock particles and form a froth that has a high concentration of chalcopyrite. “Two concentrates are formed when the rock flour is added. One is rich in sulphide and the other is not. It is mainly at this stage of the production that we intervene. Our study consists of observing what goes in and what comes out, and verifying if the separation has been sufficient or not which is of vital importance because, on an industrial scale, the crushing phase of the rock is the most taxing in terms of metal production. It is a very costly and energy-consuming phase involving thousands of tonnes of rock”

By repeating the process, this froth becomes more and more concentrated until a high percentage of chalcopyrite is reached. It can then be sent to metallurgy factories where the copper will be extracted and refined. Given that one tonne of rock only results in around twenty kilograms of copper, any tool or method of expertise that makes it possible to avoid wasting any of these very valuable particles will be welcome. 

The separation techniques for minerals are more efficient when the liberation phase functions well also and this is the stumbling block for engineers today.

Digital imaging forbetter performance 

After crushing the rock, it will never be possible to have particles containing 100% chalcopyrite or 100% gangue. Most of the particles are mixed. In order to better understand and therefore react to the level of liberation, mineralogists observe samples of this rock flour with an optical microscope using reflected light. “The particles are placed on a resin and then sectioned before being polished. What interests us is the internal structure of the particles and not their surface. Tomography or 3D analysis could be used but the material is too compact and dense”. Traditionally, these observations are made with the naked eye and, from a qualitative point of view, make it possible to obtain more or less precise information on the degree of liberation of minerals.

Since the middle of the 1980s, Eric Pirard has aimed to go further and develop a quantitative approach. Therefore he no longer uses his eyes but rather a camera fitted to the microscope which digitally transcribes the observed image. “Today, digital imaging is a widely-used technology, even by private individuals. Twenty-five years ago we were almost pioneers, the researcher says with some pride”.

The images which are then sent to the computer are made up of millions of pixels which assume different colours according to the type of material they represent. These pixels make possible a series of very precise calculations. For example, chalcopyrite as we have said is a yellow mineral. In an image of a cross-section of a particle, all the yellow pixels represent chalcopyrite. It is possible to calculate to the nearest pixel, the percentage of chalcopyrite still present in each particle and to establish a liberation curve. If the particle contains 100%, 90% or 80% this means that the liberation of minerals was a success during the crushing phase. Yet, very often, there is an important percentage of other minerals present in these particles. The thorny question is to know whether it would useful to recrush tens of tons of flour in the crusher in an attempt to have a greater liberation or whether this would be a waste of time, money and energy.

Better classification of the texture of particles

Precisely establishing the percentage of different minerals in a particle was certainly very useful but did not take into account two important characteristics. Firstly, the intergrowths can be very different and more or less complex in nature. Professor Pirard’s team identified four main families of particles: The simple particles (a), the stockwork particles (b), the coated particles (c), and the emulsions (d). It should be noted that in nature, there is an intermediate series of textures that make it difficult to achieve an exhaustive and efficient classification.



Assuming there are four groups of particles, each one belonging to one of these families and each having a 50% proportion of chalcopyrite, then in the case of group C, if the flour is re-exposed to a longer time in the crusher, there is a good chance that the particles will be ground and broken up during the separation of the minerals. There is a greater possibility that the chalcopyrite will be better liberated. Another spell in the crusher could therefore be useful. If the particles now present a type-A texture, we can imagine a better liberation but this already appears to be more complicated. With regard to the B-texture intergrowth, if these particles are ground even more, they will certainly be smaller but will still present the same texture. The chalcopyrite will not have been liberated. Sending particles like this to be recrushed would quite simply be wasteful.

The second important characteristic is that the section does not take account the texture of the intergrowths in the third dimension of the particle which could be very different. “This is a basic principle of stereology. If we take as an example a cake with layers of vanilla and chocolate. When the cake is cut transversely and I observe a layer of vanilla, I cannot make the deduction that it is made up of 100% vanilla. One section does not tell us anything. We would need several cross-sections of the same particle, but this is impossible. Today, as we study several particles we only make random cuts”

The creation of several indices has made it possible to better classify the particles according to the two problems mentioned above, and to better identify the nature of a deposit to enable more economical exploration. “The first index that emerged from the calculation by means of pixels made it possible to take account of the volume proportion, and therefore to establish the percentage of chalcopyrite in the particle. The second index is that of surface exposition. When the separation technique used is froth flotation, for example, the hydrophobic minerals must be at the surface of the particles so that the latter adhere to the bubbles and rise to the surface”. For example, a rim-type particle could contain 80% chalcopyrite and not float to the surface. 

These two first indices already make it possible to take account of the way the particles could react to crushing and the separation phase. But this is not yet sufficient to classify them into families. For example, the stockwork and emulsion particles will have indices that are similar, that is to say, the same percentage of chalcopyrite, and will be poorly represented on the surface. “To obtain an even more precise vision, we draw random lines. Then we measure the length covered and we look at the distances covered by the yellow phases, therefore chalcopyrite, and at what frequency”

 In the case where the yellow phases are short and numerous, we are dealing with an emulsion-type particle, in the case where the yellow phases are long and not very numerous, we are dealing with stockwork particles, etc. “All these indices make it possible to calculate and identify particle textures in different ways. This procedure may seem off-putting, but the addition of these results gives a more precise statistical analysis of the comparison and classification of these particles into identifiable families. This allows for a better exploitation of the raw material.”

Beyond the texture of rocks

The article is part of much wider research. Eric Pirard aims to develop an automatic optical microscope dedicated to the needs of the mineral industry and to perfect a marketable quantification technique for minerals. 

The researcher also mentions his knowledge with regard to improving recycling conditions for metals. “The crushing techniques that work for rocks also work for GSMs or for solar panels, for example”, the scientist explains. The constraints and problems encountered are the same. A way has to be found to liberate the different metals at the lowest possible cost in terms of economics and energy. Also, with regard to the exponential increase in the world population and the tools it uses, particularly in developing countries, it is vitally important that the researcher devotes his time to these procedures involving engineering, geology, recycling and exploitation of rock. “There has been a lot of talk about closing mines in Europe”, here the researcher takes a jump forward in time, “I predict that in the future mines will be soon be reopened”.

(1) Pérez-Barnuevo, L., et al. Automated characterization of intergrowth textures in mineral particles. A case study. Miner. Eng. (2013), http://dx.doi.org/10.1016/j.mineng.2013.05.001