Journal Information
Vol. 8. Issue 1.
Pages 1-7 (January - March 2019)
Download PDF
More article options
Vol. 8. Issue 1.
Pages 1-7 (January - March 2019)
Original Article
DOI: 10.1016/j.jmrt.2018.02.011
Open Access
Feldspar production from dimension stone tailings for application in the ceramic industry
André Carlos Silvaa,
Corresponding author

Corresponding author.
, Sílvio Divino Carolinab, Débora Nascimento Sousac, Elenice Maria Schons Silvad
a Modelling and Mineral Processing Research Lab, Engineering Faculty, Federal University of Goiás, Capes Scholarship POS-DOC – 88881.119296/2016-01, Catalão, Brazil
b METAGO, Mineral Processing Department, Goiânia, Brazil
c Mining Department, Goiano Federal Institute, Catalão, Brazil
d Modelling and Mineral Processing Research Lab, Engineering Faculty, Federal University of Goiás, Catalão, Brazil
This item has received

Under a Creative Commons license
Article information
Full Text
Download PDF
Figures (8)
Show moreShow less
Tables (2)
Table 1. Mass and metallurgical recoveries for Fe2O3, K2O, and feldspar at magnetic separation.
Table 2. Froth flotation bench tests metallurgical recovery.
Show moreShow less

Brazil has one of the largest reserves of granite in the world and is in part of the top ten producers of dimension stone. The dimension granite extensive mining in this country generates expressive volumes of tailings, among then rock fragments and fine particles, accounting for more than 50% of the material extracted from the quarries. The dimension stone, commercially known as Marrom-Guaíba granite, is an alkali-feldspar quartz syenite rich in potassium and sodium exploited by Serra Geral Mining in the southern part of Brazil. Due to its chemical composition, Marrom-Guaíba tailings have potential application in the ceramic industry. However, the grade of iron bearing minerals (considered a contaminant) is too high. Tailings samples were grinded, milled, chemical analyzed and tested using magnetic separation and froth flotation in order to reduce the iron and quartz content. The results show that the magnetic separation reduced the iron content from 3.20 to 0.48% and increased the feldspar content from 72.20 to 81.23% with 65.31% of mass recovery and 73.48% of metallurgical recovery. Flotation provided small decreased in quartz content, from 4.40 to 3.90%. A suitable feldspar concentrate for ceramic industry was obtained.

Magnetic separation
Froth flotation
Ceramic industry
Full Text

In 2014, the world production of dimension stone reached 136,500Mt, leaded by China with approximately 31.1%. Brazil is the fourth in the world ranking, contributing with 7.4% of the world's production. In the same year the Brazilian's granite exportation reached 2547Mt [1]. Granite extensive mining generates an expressive volume of tailings, accounting for more than 50% of the extracted material, causing many environmental impacts. Due to its mineralogical and physical characteristics, some granites present significant feldspar and quartz concentrations [2].

Ceramic and glass industries are the main consumers of feldspar in Brazil. In glass manufacture, feldspar is used as fluxing agent and as a source of alumina (Al2O3), alkali (Na2O and K2O), and silica [3]. Fluxing agents are substances, usually oxides, used in glasses, glazes and ceramic bodies to lower the high melting point of the main glass forming constituents, usually silica and alumina. These materials play a key role in the vitrification of clay bodies by reducing the overall melting point and the most common fluxes used in clay bodies are potassium oxide and sodium oxide, which are found in feldspars. In the manufacture of ceramics, feldspar is the second most important ingredient after clay. They improve the strength, toughness, and durability of the ceramic body and cement the crystalline phase of other ingredients, softening, melting and wetting other batch constituents. In the flooring sector, feldspar is the main constituent in the body composition [4].

Brazilian's feldspar gross production in 2014 was 497kt, which represented an increase of 41% over the previous year. The State of Paraná was responsible for 40.5% of the gross production, followed by Minas Gerais (35.9%), Santa Catarina (10.6%), Bahia (5.2%), Paraíba (4.4%), Rio Grande do Norte (3.1%), and São Paulo (0.3%). Processed production totalled around 418kt [3].

One of the main characteristics to be considered of feldspar for ceramic industry is the grade of iron bearing minerals (oxides or silicates) present on it. Although the ceramic industry requirements for raw material are quite diversified, since they depend on the type of material to be produced, if the grade of this contaminant is higher than 0.5% it promotes a dark colouring and other irregularities in the final product, detrimental to the processes and to the quality of the ceramic. In the production of white dishes, the maximum iron content (evaluated as Fe2O3) is 0.1%. However, if the product brightness is not important, which happens in cases where the ceramic receive a colourful pigmentation in the last stages of the process, the Fe2O3 content can reach 2 or 3%. Regarding the granulometry, feldspar used in the ceramic industry must be smaller than 74μm because its fluxing power is inversely proportional to its size [5].

This paper describes the tests in bench scale performed with samples from Marrom-Guaíba granite tailings in order to obtain a feldspar concentrate with chemical and granulometric properties suitable for use in the ceramic industry. A dry high-intensity magnetic separator was used in order to reduce the iron bearing minerals content on the sample. Cationic flotation of feldspar using amine and HF was used to reduce the quartz content.


A sample of 20kg of Marrom-Guaíba granite tailings was received in pieces with size between 25 and 50mm, donated by Serra Geral Mining Company, located in the Piquiri District, municipality of Cachoeira do Sul, State of Rio Grande do Sul, Brazil. The rock was characterized in the SED-FUNMINERAL mineral laboratory using an Atomic Absorption Spectrophotometry (AAS) Analyst 200 (Perkin Elmer). Samples were comminuted bellow 74μm and sent to XDR analysis using Ultima IV (Rigaku). The quartz content was determined semi-quantitatively by the solubilization of the silicates with hydrofluoric acid to a concentration of 30% using 10g of sample during 10min of reaction. Two samples of pure quartz were tested under the same conditions to generate a calibration curve for the rock samples silicates solubilization. A petrographic analysis was done for determining the composition and texture of the rock.

2.1Sample preparation and milling tests

The rock samples were crushed in a jaw crusher, to obtain particles below 10mm, and then milled in a HPGR mill, resulting in 80% of the particles bellow 3mm. The material was mixed and quartered in a longitudinal Chevron pile, assembled with three passes and aliquots withdrawals of 1kg.

To determine the optimum time for the rock comminution (100% below 600μm), milling tests were carried using a bar mill and three different residence times (5, 10, and 15min). The mill had 400mm length and 200mm diameter and operated with 6 bars of 25mm, 7 bars of 20mm, and 8 bars of 10mm diameter. The solids percentage in the mill was 55% and the rock mass was 1kg. The mill product was deslimed through wet sieving for 10min using a 38μm sieve, without dispersant. The resulting material (retained in 38μm) was dried in oven at 80°C for 4h.

2.2Magnetic separation tests

A high-intensity (22kG) dry magnetic separator was used in an attempt to separate mica from feldspar present in a feldspathic sand. The authors were able to reduce the mica content of the initial sample by 95% [6]. In order to test the presence of high magnetic minerals in the rock a magnetic separation tests was performed in the fraction above 38μm using a manual ferrite magnets with magnetic field approximately 1kG. Since some material was separated with a low magnetic field, a Frantz Isodynamic Separator was tested operating with magnetic field of 12kG. The chute inclination was set to 25° longitudinal and 20° transversal. A vibration intensity level 6 was adopted in all tests.

2.3Cationic flotation of feldspar (amine+HF)

After the magnetic separation, the non-magnetic particles @ 12kG field were sent to froth flotation bench tests using a Denver laboratory machine with a 2.5l cell. The direct flotation was opted in order to separate feldspar from quartz. Triameen Y12D, an amine with CAS name N,N-Bis(3-aminopropyl) dodecylamine, supplied by AkzoNobel was used as collector with dosage of 200g/t. The collector's alkyl chain distribution, according to the supplier, was maximum 1% for alkyl chain length C10 or C14 and minimum 98% for C12. Kerosene was used as feldspar activator and frother in a dosage of 50g/t. Hydrofluoric acid was used as silica depressant and feldspar activator [7] and to adjust the pulp pH to 2.5. The pulp density was 35wt% solids and the impeller speed 1400rpm. The conditioning time was 5min and the flotation time 5min. Tap water was used throughout the experiments. The flotation tests were performed in two sequential stages at the same conditions: rougher and scavenger.

3Results and discussion

Feldspar contents were calculated by stoichiometry based on sodium and potassium contents, as can be observed in Fig. 1. According to the petrographic analysis, the sample of Marrom-Guaíba granite, as shown in Fig. 2, corresponds to an igneous rock, classified as alkali-feldspar quartz syenite. The rock cannot be characterized as a granitic rock due to quartz low grade (5.7%). Granitic rocks normally have quartz modal composition ranging from 20 to 60% [8]. The macroscopic description of the rock is a reddish-brown rock, phaneritic textures, equigranular grains with average granulation, predominantly constituted by K-feldspar and amphibole (black colour), subordinately by titanite and rarely pyrite [9]. The sample presented moderate magnetism, which could be related with the presence of magnetite on it. In structural terms, it is anisotropic, featured by the preferred orientation of the crystals of feldspar and amphibole.

Fig. 1.

K and Na-feldspar chemical adopted composition [10,11].

Fig. 2.

(a) Macroscopic aspect of the sample with preferred minerals orientation on the horizontal axis. (b) Photomicrograph in binocular magnifying glass showing titanite crystals. (c) Photomicrography in natural and (d) polarized light of amphibole grains containing inclusions of clinopyroxene in association with opaque minerals and apatite. (e) Photomicrography in natural and (f) polarized light showing a Carlsbad germination and perthitic intergrowths in K-feldspar. (g) Photomicrography in natural and (h) polarized light of a quartz grain among a K-feldspar.


The microscopic description showed that the Marrom-Guaíba granite was constituted predominantly by K-feldspar and amphibole, which agrees with the macroscopic description, having secondarily titanite, apatite, and opaque minerals, and traces of clinopyroxene. The rock texture was a stacked arrangement of K-feldspar crystals, which showed preferred alignment. The modal composition estimated was 66% K-feldspar, 24% amphibole, 4% quartz, 3% titanite, 2% opaque minerals, and 1% apatite. The rock was classified as a syenite alkali-feldspar [12].

3.1Milling residence time tests

Fig. 3 presents the results of the bar mill tests. After 5min, 100% of the rock was below 600μm, with P80 approximately 280μm. The rock showed P80 around 160μm only after 15min of milling. Since Frantz Isodynamic Magnetic Separator usually shows better results for particles above 74μm no additional tests were performed for higher residence time than 15min. It is also possible to notice that the amount of fine particles (bellow 38μm) produced in all testes were similar, ranging from 5 to 10%. No finer milling was necessary once there was no presence of sulphides in the samples.

Fig. 3.

Particle size distribution of the grinding and feed products.

3.2Magnetic separation

Fig. 4 shows the Frantz Isodynamic Separator results for the three milling residence times tested. It is possible to notice that 5 and 15min milling showed the high non-magnetic minerals content (feldspar, quartz and silicates) in weigh, 62.13 and 51.14% respectively. This result could be explained because some magnetic particles could be remained lock or associated at 5min milling. For 15min milling the higher amount on fine particles present in the sample (above 74μm) could be lead to a poor separation at the Frantz Isodynamic Separator.

Fig. 4.

Mass fraction results for the magnetic separation after (a) 5, (b) 10 and (c) 15min milling.


Fig. 5 shows the chemical composition of the magnetic and non-magnetic materials produced at 12kG magnetic field for the three tested milling times. It was possible to enhance the grade of Fe2O3 from 3.20 to 11.2% with 5min of milling and to 15.90% with 15min. This result proves, as expected, that a higher milling time lead to a high liberation degree of the iron bearing particles. The Fe2O3 grade in the non-magnetic was 0.48% after 15min of milling, achieving the goal of reduce the iron content to levels below 0.5%. The Fe2O3 mass recovery was 17.64% and the metallurgical 87.85% for the some milling time (see Table 1).

Fig. 5.

Chemical analyses of the magnetic separation feed and products.

Table 1.

Mass and metallurgical recoveries for Fe2O3, K2O, and feldspar at magnetic separation.

Milling (min)  Fe2O3K2OFeldspar
  Mass (%)  Metal. (%)  Mass (%)  Metal. (%)  Mass (%)  Metal. (%) 
25.72  90.02  49.07  61.71  63.11  70.15 
10  23.99  91.45  48.79  61.65  58.96  64.93 
15  17.64  87.65  60.22  76.85  65.31  73.48 

Regarding the K2O, it was possible to enhance its grade from 6.41 to 8.18%, with mass recovery of 60.22% and metallurgical recovery of 76.85%. The total alkali mass recovery for 15min milling was 63.57% and the metallurgical recovery was 73.56%. The feldspar mass and metallurgical recovery reached the maximum after 15min milling, 65.31 and 73.48% respectively. On the other hand, the grade of feldspar on the paramagnetic flow was 55.20% for the same milling condition. The presence of this material in the magnetic stream could be explained due the action of mechanical drag forces present in the equipment.

Fig. 6 shows the X-Ray diffraction results for the magnetic and non-magnetic products produced at a 12kG magnetic field after 15min of milling. The magnetic product was identified as hornblende, anorthoclase, microcline, albite, and quartz and the non-magnetic product was composed by anorthoclase, microcline, albite, quartz, and cummingtonite.

Fig. 6.

X-ray diffraction results of the (a) magnetic and (b) non-magnetic products @ 12kG.

3.3Flotation results

Fig. 7 shows the mass fraction (in g) and recovery (in %) for the froth flotation bench tests regarding the three tested milling times. The flotation was not efficient for coarser particles, since for 5min milling the mass recovery was around 50%. However, after 15min milling the mass recovery reached 93%, an increase of 86%. On one hand the feldspar metallurgical recovery was around 91% (see Table 2) for the same milling time, but in the other hand the recovery of quartz in the concentrate was also high (82%). Those results confirm that the feldspar and quartz floated together. This result could be related with particle association between feldspar and quartz or a low concentration of the hydrofluoric acid.

Fig. 7.

Mass fraction (g) and recovery (%) after (a) 5, (b) 10 and (c) 15min milling.

Table 2.

Froth flotation bench tests metallurgical recovery.

Milling (min)  Metallurgical recovery (%)
  K2Quartz  Feldspar 
50.55  57.58  53.53 
10  76.83  90.75  81.55 
15  91.93  82.39  91.22 

Chemical analysis of the flotation concentrate were inconsistent for milling times of 5 and 10min, (see Fig. 8), since the grade of K2O and feldspar were similar in both concentrate and tailings. There was a slight increase in the feldspar grade in the concentrate for 5 and 10min milling, but a reduction for 15min (from 67.3 to 66.2%), which was unexpected. It is possible to notice the enrichment of quartz in the tailings, indicating an interaction between feldspar and amine and the depression of quartz due the hydrofluoric acid. The tailings chemical analysis results for 15min milling resulting in a low grade of alkalis (5.2%) and high grade of quartz (47.8%).

Fig. 8.

Chemical analyses of the froth flotation feed and products.


Marrom-Guaíba granite corresponds to an igneous rock, classified as alkali-feldspar quartz syenite due to its low amount of quartz (5.7% on petrographic and 8.4% on the semi-quantitative analysis). Therefore, the rock cannot be considered as a granitic, or a granitoid, geologically speaking.

The feldspar concentrate production with chemical and granulometric specifications suitable for the manufacture of ceramics, porcelain tiles, sanitary wares, glasses, and insulators was investigated in this work. After 15min of milling and a magnetic separation with a 12kG field the iron content (expressed as Fe2O3) was reduced from 3.20 to 0.48% and the feldspar content increased from 72.20 to 81.23% with 65.31% of mass recovery and 73.48% of metallurgical recovery, being this concentrate suitable to be used as ceramic raw material.

Flotation tests were not efficient for coarser particles (lower milling time). However, there was a slight increase in the alkalis content and consequently in the feldspar grade and a decrease in quartz content in those conditions, which could not be observed for 15min milling. A deficit in hydrofluoric acid or an association between quartz and feldspar could explain the found results.

Conflicts of interest

The authors declare no conflicts of interest.


The authors gratefully acknowledge the financial support of the Brazilian agencies CNPq, CAPES, FAPEG and FUNAPE. In addition, we would like to thank the companies Metago and Serra Geral Mining and the support of the Federal University of Goiás and Goiano Federal Institute.

M. Heider, C.M. Maia.
Rochas ornamentais e de revestimento.
Sumário Mineral 2015, pp. 100-101
R. Dutra.
Rochas ornamentais – feldspato e quartzo.
Proceedings of the II Encontro de Engenharia e Tecnologia dos Campos Gerais, August 14–18; Ponta Grossa, Brazil, pp. 1-8
L.G. Joaquim, R.F. Pereira Júnior.
Sumário Mineral 2015, pp. 64-65
A. Valchev, T. Ignatova, I. Grigorova, I. Nishkov.
Recovery of feldspar and silica sand from arkosic sandstones.
Proceedings of the XXVII IMPC, October 20–24, Santiago, Chile, pp. 1-13
M.M. Almada, T.F. Vlack.
Homogenization stacks: a new vision for ceramic feldspar.
Ceram Ind, 5 (2000), pp. 31-34
A.A. Díaz, J. Taboada, F.G. Bastante, M.A. Fernández.
Effects of flotation variables on feldspathic sand concentration.
Dyna, 81 (2014), pp. 132-139
T.R. Boulos, S.S. Ibrahim, A. Yehia.
Differential flotation of some Egyptian feldspars for separation of both silica and iron oxides contaminants.
J Miner Mater Charact Eng, 3 (2015), pp. 435-443
M.G. Best.
Igneous and metamorphic petrology.
2nd ed., Blackwell Science Ltd., (2003),
J.P. Bard.
Microtextures of igneous and metamorphic rocks.
Springer Netherlands, (1986),
Homepage Webmineral – Orthoclase Mineral Data [homepage on the internet].
Available from: [cited 1.09.17]
Homepage Webmineral – Albite Mineral Data [homepage on the internet].
Available from: [cited 1.09.17]
H.G. Dill.
Kaolin: soil, rock and ore: from the mineral to the magmatic, sedimentary and metamorphic environments.
Earth-Sci Rev, 161 (2016), pp. 16-129
Copyright © 2018. Brazilian Metallurgical, Materials and Mining Association
Journal of Materials Research and Technology

Subscribe to our newsletter

Article options
Cookies policy
To improve our services and products, we use cookies (own or third parties authorized) to show advertising related to client preferences through the analyses of navigation customer behavior. Continuing navigation will be considered as acceptance of this use. You can change the settings or obtain more information by clicking here.