The Proposal of Tungsten Ores Processing in Rwanda

Tungsten is one of the rare elements occurring in Earth. Its applicability and request for it causes that the production of this metal is very beneficial. One of the biggest deposits of this metal ores in the world is located in African country of Rwanda. Due to the lack of appropriate technology and lack of investments the current way of producing this valuable metal in this country causes that much of tungsten is wasted as well the production is slow and the results are not appropriate. That is why an attempt to propose an adequate processing way for this deposit was done and is presented in this paper. Authors performed several tests on the ore originating from Rwanda, including its mineralogical composition and tests performed by means of laboratory shaking table. The results are promising and the further tests, including other methods of beneficiation are planned, like second shaking table stage and flotation.


Introduction
Tungsten is a hard, rare and essential metal in many applications whether commercial or industrial. The alloys of tungsten are extensively used to make so many different things, such as incandescent light bulb filaments, electrodes in welding, superalloys and radiation shielding, to name a few. It is much used in several military applications due to its hardness and density. This made tungsten one of the strategic metals in the world [ Considering geography of tungsten occurring, Rwanda is one of the top producers of tungsten globally, its concentrate counted 830 000 Mg in 2018 [USGS U.S geological survey, 2019]. Still, it is processed mostly in artisanal and small-scale mining ways which means that most mineral processing in Rwanda involves manual techniques, particularly without mechanical ones. It is known that depending on the style of mineralization, these traditional beneficiating techniques are likely to be inefficient and may lead to loss of high amount of ore particles of economic interest to the tailings. Tungsten concentrates from Rwanda are not upgraded at a mine site, they have to be transported miles away for proper and profitable processes. So, it is due to inefficient processing methods, insufficient mount of processing plants as well lack of controlled comminution processes that some useful minerals are simply lost. Moreover, the distance transportation of materials that still need to be upgraded from their place of origin to another one for additional further processes causes also additional economic costs for the mining industry in Rwanda. Consequently, tungsten processing plants in Rwanda produce concentrates of grades lower than the ones achieved in 2013 and it is very important for Rwanda to introduce more modern and more efficient methods in the process.

Properties and applications of tungsten
Tungsten hardness and wear resistance make it valued. It has its highest melting point at the temperature of 3422°C and its density of 19.25 g/cm3. Moreover, it is among the substantial metals. More detailed tungsten properties are listed in Table 1.
Tungsten is used in many technological applications, due to its exceptional physical and chemical properties. Its main usage is found in the manufacture of cemented carbides which is the main consumer of tungsten today.
The other usage of tungsten is that it is used as the alloying element in the iron and steel industry. Also, it is found in many metal products, such as lighting filaments, electrodes, rods, electrical and electronic contacts, wires, sheets etc. The prime economic minerals of tungsten are scheelite, ferberite, hübnerite, and wolframite, the content of tungsten in the earth's crust is 0.007%. Tungsten grade in feed should be at least 0.3 to 1% concentration so the mining is beneficial [https://www.itia.info/about-tungsten.html]. There are more than 20 tungsten bearing mineral but only wolframite and scheelite are essential for industrial use. Their main features are listed in Table 2.
World tungsten resources are geographically widespread. Even though China has been topping other countries, in terms of tungsten resources and reserves, there are other countries with significant concentration including, Russia, Canada, US, Bolivia, Vietnam, Portugal, Spain, Austria, Rwanda, UK.
The production of tungsten concentrates in Rwanda and other countries in period 2017-2018 is shown in Table as  Today, the number of applications of tungsten is increasing very fast and the industrial demand for it increases. Hence, the request for low-grade complex ores is increasing and this leads to the complexity of its beneficiation.

Tungsten processing in Rwanda
Artisanal processing is a way of beneficiation that is made in a traditional or non-mechanized way. In Rwanda, it includes panning, handpicking, ground sluicing, air classification, manual magnetic separation.
How exactly these processes are being done can be found in [Heizmann and Liebetrau, 2017] with step by step guide, demonstration figures and a detailed comparison of both artisanal and mechanical processing in Rwanda, not only for wolframite but for cassiterite as well. However, the main techniques of processing used there are listed below.
• Panning: an artisanal method of separation which sorts particles by their specific gravity. It is made of ponds filled halfway with water, with 2 m * 2 m and 6 m of size and depth respectively. • Hand-picking: it is used to pick coarse particles; it is done by a miner who is familiar with that type of mineral and knows very well the physical properties of the minerals. Every grain is sorted manually. • Ground sluicing: density sorting method, but its negative side is that it requires high amount of water. • Air classification/tap and blow/winnowing: after the drying of the concentrate, artisanal air classification. Applicability of this technique depends on the grain size of the concentrate. It is only suitable for relatively low grain-sized pre-concentrates. light particles accumulate at the edge of the material cone and can be carefully blown away (blow) by the worker. • Artisanal magnetic separation: a manual magnet separation for the final processing step.
In recent years, a few of Rwanda mining companies introduced mechanical equipment to add more value and increase their production, including crushers, spirals and shaking tables, but still a lot is to be improved

Description
The sample was collected from Muyira Cell of Manihira Sector, Rutsiro District. Minerals in this block are handpicked from underground hydrothermal quartz veins which were hosted in black shales. The sample consists of brown material (clayish) with blocks of quartz and black crystals of wolframite. Often, the quartz coating of iron oxide could be visible in quartz. The brownish clay material silicates, iron hydroxide and sulfides (weathered or partly altered) could be visible too. Scheelite is also present in small particles form enclosed in quartz.

Sample characteristics -XRF analysis
The small representative amount of sample of granulation below 0.2 mm was prepared by means of mixing and XRF analysis. These analyses were conducted and repeated 3 times for ensuring sufficient accuracy. The average result obtained showed an overall head grade of 23.675% WO 3 , 11.687% Fe, 0.571% Mn and very low grade of arsenic of about 0.037%. Table 4 demonstrates all three repeated test results performed on the feed.
Ferberite (FeWO 4 ) and hübnerite (MnWO 4 ) are commonly the main wolframite minerals in such deposit types. The amount of manganese can be observed -average grade of 0.571%, but comparing to iron it is low amount -Fe content reaches in average to 11.687%.
Basing on the analysis, it is obvious to consider ferberite as the main tungsten mineral contained in the sample.

Sample preparation
The sample preparation aim was to get all particles in the feed of the preferred size. First, the sample material of a maximum of 31 mm size was put into screen, the material of size lower than 1 mm was removed and the remaining part was put into jaw crusher. Next, the material was transferred to a rod mill and then to screen. These stages were repeated continuously to get the expected particle size.
The sieving time was equal to 5 minutes. Each time the -0.2 mm mesh material was screened out and the remaining material was passed again through the laboratory rod mill or ball mill until the whole material was characterized with the size below 0.2 mm. The obtained products were well mixed and then were divided into 5 parts. Such representative samples were divided as well into parts, where 25% was dedicated to XRF analysis and the rest for sieve analysis.
The samples taken from screen analysis were examined under a binocular microscope. The mineralogical components were quartz with wolframite, mainly FeWO4 as well a significant amount of scheelite. The whole scheme of the sample preparation is shown on Figure 5.

Results and discussion
The results of sieve analysis are positioned in Table 5 and the particle size distribution is presented on Figure 6. Table 6 describes the amount of W, Fe, As, Mn in each particle size fraction, being: -0.063; 0.063-0.1; 0.1-0. 16  that the smallest amount of W occurred in a fraction 0.1-0. 16 [mm] and the highest amount of W was found in fraction 0. 16-0.2 [mm]. Considering the percentage difference between each analyzed element in raw feed and te balance of particle size fractions is visible that the biggest difference occurred for As (around 10%). In case of W this difference was equal to 8%. The results are presented in Table 7.
Next step were tests performed on a laboratory shaking table. The study was completed using 5 test products, with different masses and all of them showed similar characteristics with slight differences. For example, the difference of tungsten recovery in concentrate was equal to only 0.013% and the biggest observed difference was equal only to about 2%, which is acceptable. The same observations can be noticed for other investigated elements.
The dried weight of each test product was 464 g, 408 g, 440 g, 432.5 g, 450.5 g corresponding with 1st product, 2nd, 3rd, 4th, and 5th respectively. Table 8. shows the mass of the concentrate and tailings in each product.
The average amount of tungsten in the concentrate was equal to 29.13%, while the amount of it in tailings was equal to 11.12%. This shows that there is necessity to perform additional processes to recover W from the tailings. Table 9 il-Tab. 7. The percentage difference between elements in raw feed and elements from the balance of particle size fractions lustrate not only the results for W, but also for As, Fe and Mn. Considering that the yield of concentrate was equal to 62.36% (in average) and yield of tailings to 37.63% (in average) it can be said that the results are promising but the process requires additional stage or introduction of other processes, like flotation [Mohammadnejad et al., 2018] or magnetic separation [Lu et al., 2016]. In further steps, hydrometallurgy or pyrometallurgy can be applied to the process of producing tungsten products [Singh Gaur, 2006]. The most important results are shown in Tables 9-12. Basing on the obtained knowledge, the Authors propose to apply the tungsten processing scheme, like is presented on Figure 7. Although, it looks correctly this proposal needs to be verified empirically, which will be the next target to perform. The obtained results can be applied in Rwanda to process wolframite ores. Furthermore, the other products contained in the ore should be also the point of interest, including Fe, in particular.

Conclusions
The results of the work show that the grain class 0.16-0.2 mm contained the highest amount of useful minerals. So, it is necessary to control the grinding stage to not excess the production of the finest material, because it usually occurs in tailings. Gravity concentration (by means of shaking table) is undoubtedly the best method of tungsten beneficiation since it seemed efficient and showed a good recovery level. However, it is important to add some supportive beneficiation stages as another step of gravity separation or floatation to recover, particularly very fine particles. Also, the magnetic separation can be useful to fully separate useful metals, like iron.
Shaking table recovery results were equal to 80.15~83.56% of W; 69.42~73.00% of Fe and 78.93~82.81 of Mn. The highest recovery rate achieved in the test series was equal to about 83.56% which can be assumed as good. As it is not possible to achieve the recovery equal to 100% in practice, it can be increased by introduction of additional beneficiation stages as is mentioned above. Also, the control of grinding stages can add some percentage to the recovery rate.
The amount of the metals in tailings is significant in average, equal to about 20% of each considered element. Certain-ly, it is advised to do more research before adopting the idea to industrial conditions. The flowsheet used in the research work consisted only of one separation stage. This was caused by a limited time being a result of difficulties in obtaining the ore samples from Rwanda. However, the remaining proposed tests such as flotation and magnetic separation will be conducted in the near future. Also another step of gravity separation process related to wastes will be considered (cleaning shaking table process).