Preconcentrating Ultrafine Ilmenite Tailings Using a Laboratory-Scale Reflux Classifier

Abstract

China is rich in reserves of titanium, but a large amount of titanium resources is lost in the ultrafine tailings, and it is challenging to treat the ilmenite contained in ultrafine ore. The reflux classifier (RC), a novel gravity concentration technology, has been applied in the preconcentration of ultrafine ilmenite in this study. During this process, the feasibility of using RC for preconcentration of ultrafine ilmenite was explored through theory and conditional experiments. After one-stage preconcentration using RC, the ultrafine ilmenite ore with a TiO₂ grade of 8.77% can be concentrated into a product with a TiO₂ grade of 20.3% and a recovery rate of 82.8%. The tailings grade is as low as 2.44%, and the yield reaches 62.6%. The separation efficiency achieves 50.0%. Experimental results demonstrate that utilizing RC for the preconcentration of ultrafine ilmenite can avoid the influence of weakly magnetic gangue and achieve better results compared to a magnetic separator. Therefore, RC offers a more effective and affordable method for preconcentrating ultrafine ilmenite ore.

1. Introduction

China possesses vast reserves of titanium resources, comprising over 35% of the world’s total reserves. The majority of these resources, over 90% [1], are primarily concentrated in the Panzhihua area of Sichuan province, predominantly in the form of ilmenite. Common gangue minerals include feldspar and augite, with a minor quantity of olivine present [2,3]. Due to the valuable minerals being embedded in the gangue in the form of extremely fine particles, raw materials must be finely ground to ensure product quality, resulting in the generation of a significant volume of fine particle materials [3,4,5]. Consequently, a substantial quantity of fine-grained ilmenite is lost in the tailings, resulting in the wastage of valuable titanium resources [6,7,8]. The commonly used techniques for ilmenite concentration are magnetic separation and flotation [5,9]. In the case of magnetic separation, to attain optimal beneficiation products when separating ultrafine ilmenite, it is imperative to utilize a superconducting magnetic separator capable of generating a maximum magnetic field intensity of up to 5 T [7]. Indeed, this poses a substantial challenge given that conventional magnetic separators encounter difficulties in effectively separating ultrafine ilmenite particles [10,11,12]. Additionally, if the content of ilmenite in the ultrafine raw material is low, the presence of fine-grained gangue minerals, especially augite and chlorite, can exert a considerable adverse impact on both magnetic separation [13] and flotation processes [14,15,16,17].

Investigating the viability of utilizing gravity concentration techniques for the beneficiation of ultrafine ilmenite represents a valuable pursuit. However, it is crucial to acknowledge that fine-grained minerals in a liquid medium tend to increase slurry viscosity and are highly susceptible to variations in settling velocity caused by particle size. The gravity separation of ultrafine particles poses a significant challenge for equipment.

Professor Galvin [18] from Newcastle University has introduced an innovative gravity separation device known as the reflux classifier (RC). This separation apparatus has demonstrated exceptional performance in the field of coal beneficiation [19,20,21,22] and is also gaining recognition in the domain of iron ore beneficiation. Amariei [23] employed iron tailings, which had been processed through a spiral chute, as the feed material for a pilot-scale RC. The sample exhibited a particle size distribution of less than 150 μm, and the spaces of the inclined channels are 6 mm. Ultimately, the recovery rate of total iron in the resulting concentrate product exceeded 80%. However, the recovery rate for fine particles (−40 μm) within the concentrate was observed to be below 40%. D. M. Hunter [24] conducted experiments using a laboratory-scale RC (RC^TM100 with 6 mm spacing inclined channels) for hematite and siliceous gangue separation at ultrafine sizes. The initial feed contained 31.9% of particles with a size of −20 µm and exhibited an Fe_T grade of 32.0%. The concentrate product achieved an 80.2% recovery rate with an Fe_T grade of 66.1%. However, it is worth noting that the recovery rate for the −20 µm fraction was only 57.0%, significantly lower than that of other sizes. Rodrigues [25] conducted tests using the laboratory-scale RC for the beneficiation of two itabirite iron ore mines in Brazil. The ore samples had a particle size distribution of less than 150 μm, with approximately 20% of particles in the −20 μm range. However, the recovery rates for both materials in the −20 μm particle size range were below 35%.

Based on the aforementioned reports, it is apparent that the mentioned RC did not deliver satisfactory results for the ultrafine grain in iron ore. To enhance the separation efficiency for ultrafine particles, an RC with an inclined channel spacing of 1.5 mm was tested for the preconcentration of ultrafine ilmenite. Additionally, modifications were made to the structural proportions of the RC. As previously mentioned, the ratio of the length of inclined channels to the height of the fluidized bed column in the RC^TM100 was 1:1. In this study, the height of the fluidized bed was reduced, resulting in an increased ratio of the length of inclined channels to the height of the column, reaching 4.5:1. This adjustment emphasizes the pivotal role played by the inclined channels in the beneficiation process.

2. Theory

2.1. Hydrodynamic Model in the Fluidized Bed

Different particles with varying sizes and densities have distinct settling velocities in the slurry due to the medium resistance effect. In fluidized beds, the settling terminal velocity equation can be used to establish a relationship between free settling velocity. The relationship between hindered settling terminal velocity and free settling terminal velocity is as follows [26,27]:

$$u_0={\displaystyle \frac{d^2\left({\rho }_s-\rho \right)g}{18\mu +0.61d{\left[\left({\rho }_s-\rho \right)\rho gd\right]}^{1/2}}},$$

(1)

$$u_t=u_0{\left(1-\lambda \right)}^n,$$

(2)

where u₀ is the terminal velocity of free settling, u_t is the terminal velocity of hindered settling, d is the diameter of the particle, ρ_s is the density of the particle, ρ is the density of the fluid, μ is the viscosity of the fluid, g is the gravitational acceleration, and λ is the volume fraction of solids.

According to Equation (1), it is evident that particles with differing densities exhibit varying settling velocities, a crucial factor in fluidized beds for achieving particle separation based on density. In addition, a higher density of the fluidized bed, serving as the separation medium, proves advantageous for separating particles with different densities.

The solid volume fraction is proportional to the overall density of the liquid–solid fluidized bed; the higher the overall density of the liquid–solid fluidized bed, the higher the solid volume fraction. However, a higher overall density of the liquid–solid fluidized bed is more conducive to the separation of particles with different densities. According to Equation (2), the higher the solid volume fraction, the lower the terminal settling velocity of the particles, which will result in an extended sorting time. Hence, conducting experiments to strike a balance between the factors influencing bed concentration and density is imperative for exploring optimal separation conditions.

2.2. Hydrodynamic Model in the Inclined Channel

A planar diagram of the inclined channel is presented in Figure 1 [18], where the fluid passing through it exhibits laminar flow. The velocity distribution of the flow in the direction of thickness is parabolic. This assumption is made while considering the thickness of the channel (z).

Assume a spherical particle of diameter d and density ρ_s is present in a non-uniform flow field, at a distance x from the upward-facing surface. In this scenario, the flow velocities on both sides of the particle differ in the normal direction, resulting in the generation of a lifting force (L_f) acting perpendicular to the direction of u. Therefore, the particle experiences force due to gravity, buoyancy, drag force (F_d), and lifting force (L_f).

Galvin and Liu [28] revised and integrated the studies conducted by King and Leighton [29], and Vance and Moulton [30]. Based on their work, they proposed the equation for lifting force (L_f) and subsequently verified it.

$$L_f=0.0567{\displaystyle \frac{{\rho }^{1.8}{\gamma }^{2.8}{d}^{5.6}}{{\mu }^{0.8}}}=0.0567R{e}_s^{0.8}\rho {\gamma }^2{d}^4,$$

(3)

where Re_s is the shear Reynolds number, γ is the shear rate, and U′ is the water velocity in the inclined channel, m/s.

$$R{e}_s=\rho \gamma d^2/\mu ,$$

(4)

$$\gamma ={\displaystyle \frac{6U^{\prime }}{z}}\left(1-{\displaystyle \frac{2x}{z}}\right),$$

(5)

where z is the spacing of the inclined channel, and x is the distance from the particle center to the channel surface.

Hence, in the inclined channel, particles with densities exceeding the critical density ρ_s are prone to settle downwards, while particles with densities lower than ρ_s tend to be transported upward. If two particles of different densities are moving upwards within the inclined channel, lighter particles are more likely to be carried all the way to the overflow, while heavier particles are inclined to settle on the lower side of the inclined channel and return to the fluidized bed.

As illustrated in Figure 2, particles with different densities are ejected to varying distances. Yanfeng Li et al. [31] conducted computations and experimental verification of the particle motion distance along the channel (L_P) for densities at different apparent velocities.

$$L_P={\displaystyle \frac{z}{\cos \theta }}\left({\displaystyle \frac{U^{\prime }}{u_0}}-1\right),$$

(6)

Figure 3 illustrates the L_P values of particles of varying sizes and densities flowing through the inclined channel, and the curve in Figure 3 is calculated using Equation (6). In the inclined channel, as the particle diameter decreases, the difference in L_P values caused by the density becomes larger, resulting in a wider range of separating intervals for particles of different densities. Moreover, if the particles are extremely fine, high-density particles are more likely to settle near the entrance of the inclined channel, achieving separation from particles of different densities.

It can be seen from Figure 3 that as the spacing of inclined channels increases, the value of L_p increases. With the channels’ limited length, effectively capturing and recovering ultrafine heavy minerals becomes a challenging task. Separating ultrafine particles of varying densities becomes difficult, and the majority of fine particles end up being lost in the overflow. This factor contributes to the lower recovery rate of ultrafine particles in iron ore processing when wide-spaced inclined channels are utilized in the RC.

In conclusion, closely spaced inclined channels hold promise for separating ultrafine ilmenite effectively. Additionally, the apparent velocity within these channels will impact particle separation efficiency. Similarly, the concentration and density of the fluidized bed, as well as the velocity of the fluidization water, will significantly affect separation efficiency. Consequently, this study will focus on factors such as underflow rate, fluidization water velocity, and equipment processing capacity, all of which influence the parameters mentioned above.

3. Experiment Design

3.1. Equipment

The apparatus employed in this study is a laboratory-scale reflux classifier, as illustrated in Figure 4. The middle part is a prism and the inclined channel has an inclination angle of 70 degrees with respect to the horizontal plane. The inclined chamber is divided into 23 evenly distributed channels by 22 stainless steel plates, each with the same spacing and dimensions.

The equipment can be partitioned into three primary sections: the top inclined channels; the central vertical quadrangular prism, which serves as the fluidized bed separation zone; and the lower section, which contains the water distribution and underflow discharge port. The underflow is utilized to evacuate the heavy particles settled at the bottom of the fluidized bed, and its discharge capability is regulated by a peristaltic pump. Water from the tank is pressurized and sent via a pump into the distribution chamber, subsequently providing fluidized water uniformly into the fluidized bed via vertical minute holes present on the distribution plate. The particles present on the upper layer of the fluidized bed then progress towards the inclined channel for further separation, ultimately resulting in low-density product overflowing from the uppermost part of the inclined channel.

3.2. Sample

The samples used in this study were obtained from the Panzhihua area of Sichuan, China. The raw material was tailings from ilmenite flotation, which, after impurity removal, grinding, classification, and desliming, became the experimental samples.

The size distribution of the feed material for test work was measured by a Malvern Mastersizer 2000 laser diffractometer (Comprehensive Laboratory, Resource and Biological Experiment Center, School of Resources Processing and Bioengineering, Central South University, Changsha, China). Particle size is reported as equivalent spherical diameter. Figure 5 illustrates that the sample’s particle size distribution ranges within −80 μm, with roughly 50% of the particles measuring finer than 13.8 μm in size. The samples were screened using 38 and 18 μm sieves, and the mass percentages and TiO₂ grade are shown in

Table 1. Sieve analysis result of feed.

Size (μm)	Mass Fraction (%)	TiO2 Grade (%)	TiO2 Distribution (%)
+38	11.58	2.43	3.21
−38 + 18	31.46	5.94	21.32
−18	56.96	11.62	75.46
Total	100	8.77	100

. It can be seen from the table that the majority of particles were less than 18 μm. The material in the sample with a particle size less than 18 μm has the highest proportion, accounting for 56.96%, with a TiO₂ grade of 11.62% and a TiO₂ distribution rate of 75.46%. The majority of the Ti elements are present in the −18 μm fraction, which poses a significant challenge for gravity separation technology.

The result of X-ray diffraction (XRD) analysis is shown in Figure 6, XRD analysis of the samples revealed a complex mineral composition, consisting of ilmenite as the target mineral, and augite, chlorite, feldspar, and amphibole as the primary gangue minerals.

3.3. Method

Before conducting the experiment, it is essential to prepare a precise mass concentration of the slurry in a tank. A stirrer is employed to prevent any potential issues related to settling and layering. Once the fluidization water is activated, and the throughput and underflow rate are adjusted, the feed pump can be launched to supply the slurry to the equipment. The operating procedure of the RC is depicted in Figure 7.

One hour after the feed pump starts, the equipment reaches a completely stable state. At this juncture, the relationship between the mass and grade of the product and the feed is in equilibrium. During stable separation conditions of the RC, samples are collected from the fluidized bed, tailings, and concentrate. The particles of samples are weighed and subjected to chemical analysis to calculate the separation efficiency.

The control variates method is utilized to assess the performance of the RC in the processing of ultrafine ilmenite. The variables under investigation include the underflow rate, fluidization water velocity, and throughput. These three variables are regulated by the underflow pump, fluidization water pump, and feed pump, respectively.

Three consecutive experiments are conducted under the same experimental conditions, with each sampling interval lasting 60 min. The results are then summarized, and the standard deviation and mean values are calculated. The equation for calculating separation efficiency and recovery is given below:

$$R={\displaystyle \frac{C·\beta }{\alpha }},$$

(7)

$$\eta ={\displaystyle \frac{\left(\alpha -\epsilon \right)\left(\beta -\alpha \right)\times 100}{\alpha \left(\beta -\epsilon \right)\left(100-\alpha \right)}}\times 100\% ,$$

(8)

where R is the recovery of the concentrated product, η is the separation efficiency [27], C is the yield of concentrate, and α, β, and ε are (in %) the TiO₂ grade of feed, concentrate, and tailings.

4. Results

4.1. Effects of Underflow Rate

The specific information of the experimental results is listed in Appendix A 

Table A1. Summary of conditions used in the experimental program.

Run	Underflow Rate (mL/min)	Fluidization Water Velocity (m/s)	Throughput (t/(m2·h))	Concentration of Feed (%)
1	17.00	2.67 × 10−3	0.71	30.00
2	34.00	2.67 × 10−3	0.71	30.00
3	51.00	2.67 × 10−3	0.71	30.00
4	68.00	2.67 × 10−3	0.71	30.00
5	17.00	4.44 × 10−4	0.71	30.00
6	17.00	8.89 × 10−4	0.71	30.00
7	17.00	1.78 × 10−3	0.71	30.00
8	17.00	1.78 × 10−3	1.42	30.00
9	17.00	1.78 × 10−3	2.14	30.00
10	17.00	1.78 × 10−3	2.50	30.00

,

Table A2. Summary of product results of the experimental program.

Run	Grade of Tailings (%)	Grade of Concentrate (%)	Yield of Concentrate (%)	Recovery (%)	Separation Efficiency (%)
1	3.85	18.51	31.47	68.78	40.76
2	3.30	14.83	49.50	81.47	35.12
3	2.54	10.78	77.98	93.71	17.28
4	2.13	9.61	90.34	97.59	7.96
5	2.10	15.03	50.23	87.80	41.10
6	2.23	15.96	47.61	86.61	42.75
7	2.44	17.94	43.02	84.75	45.90
8	2.51	20.29	37.39	82.80	49.98
9	3.11	22.11	30.06	75.26	49.57
10	3.77	24.84	22.72	65.89	47.21

and

Table A3. Summary of fluidized-bed results of the experimental program.

Run	Density (t/m3)	Solid Concentration (wt %)	TiO2 Grade of Bed (%)
1	1.42	44.43	14.60
2	1.39	43.14	13.53
3	1.37	42.23	11.96
4	1.35	41.62	10.87
5	1.47	47.47	13.84
6	1.45	45.65	14.06
7	1.42	45.08	14.21
8	1.42	44.89	15.55
9	1.44	45.83	16.70
10	1.45	47.00	17.57

, which detail the mean results of multiple experiments. The experimental results for the underflow rate are displayed in Figure 8. As shown in Figure 8, the increase in the underflow rate will result in a synchronous decrease in the concentrate grade and tailings grade. More material enters the concentrate product, including ilmenite and gangue, which leads to a decrease in the grade and an increase in the recovery. As shown in Figure 8c, an increase in the underflow will lead to a reduction in bed density. According to Equations (1) and (2), the fact that the fluidization water has not changed but the density of the fluidized bed has decreased results in a smaller difference in particle settling velocities, which is the reason for the decrease in separation efficiency. Furthermore, rapid entrainment of heavy minerals into the underflow prevents effective enrichment within the bed. Consequently, this results in shorter separation times and a lower bed grade, which is the primary cause of the decrease in concentrate grade. An excessively high underflow yield could lead to insufficient separation based on density and result in incomplete separation of particles before being directed to the concentrate product, thereby decreasing the separation efficiency. The lowest underflow rate for this device is 17.00 mL/min, and a lower rate will lead to blockages and abnormal operation.

4.2. Effects of Fluidization Water Velocity

The fluidization water velocity mainly affects the drag force exerted on the particles and the solid concentration in the fluidized bed. An increase in fluidization water velocity will cause fine mineral particles to enter the overflow product more easily, while coarser gangue minerals are more likely to be discarded. Therefore, the fluidization water velocity needs to be experimentally determined.

As can be seen in Figure 9a,b, consistent with the pattern observed in the calculations, an increase in fluidization water velocity results in a marked reduction in the recovery and an increase in the quantity of solid entering the overflow. Additionally, concentrate and tailings grades were both improved with the escalation of fluidization water velocity.

As shown in Figure 9c, an increase in the fluidized water velocity results in the dilution of the fluidized bed, causing a simultaneous decrease in fluidized bed density and concentration. According to Equation (1), as the fluidization water velocity increases, lighter particles encounter more difficulty in settling into the fluidized bed, leading to a higher TiO₂ grade within the fluidized bed. At the same time, an increase in the fluidization water velocity also leads to an increase in the slurry velocity within the inclined channels. According to Equation (6), the Lp values of various particles increase, especially the fine, heavy particles, which are more likely to be lost in the overflow. This results in an increase in tailings yield and a decrease in recovery rate. However, this also results in the loss of finer heavy particles. Therefore, an increase in fluidization water velocity leads to an increase in concentrate grade but a decrease in recovery.

By computing the separation efficiency, it can be deduced that the highest efficiency emerges at a fluidization water velocity of 1.78 × 10⁻³ m/s.

4.3. Effects of Throughput

The throughput refers to the amount of solid material, measured in metric tonnes, that is processed per unit of horizontal vessel area per hour. This unit is usually expressed as square meters of horizontal cross-sectional area. As shown in Figure 10, an increase in throughput results in more particles being processed in the vessel, while also causing more heavy particles to enter the fluidized bed and the inclined channel over the same period. The overflow increases as the processing capacity increases. In addition, a greater mass of heavy minerals per unit area is deposited in the inclined channel. This results in a greater loss of heavy particles in the overflow, leading to a decrease in recovery. However, due to the enrichment of heavy particles in the fluidized bed, the concentration and grade of the fluidized bed are improved, thus resulting in an increase in the concentrate grade.

In Figure 10, an increase in throughput leads to a simultaneous increase in concentrate grade and tailings grade, while the recovery rate decreases significantly with an increase in tailings yield. The highest separation efficiency can be achieved when the throughput is 1.42 t/(m²·h). At this point, the fluidized bed density is 1.425 t/m³. Hence, in the processing of ultrafine ilmenite ore, the attainment of a notably high separation efficiency is observed at a fluidized bed density of approximately 1.425 t/m³. This results in a higher separation efficiency of 50.0%, accompanied by an impressive concentrate recovery of 82.8% with a TiO₂ grade in the concentrate of 20.3%, while the tailings yield is 62.6%, with a TiO₂ grade in the tailings of 2.51%.

4.4. Comparison Between Magnetic Separator and RC

Magnetic separation is one of the primary methods for recovering and concentrating ilmenite. In order to draw comparisons, the feed sample of the RC was tested using an XCRQ—50 × 70 wet high-intensity Jones magnetic separator produced by Wuhan Hengle Mineral Engineering Equipment Co., Ltd. (Wuhan, China). The solid concentration of feed for the magnetic separation experiment was 5%, with unloading taking place every 5 s. The clamp was demagnetized during unloading, after which the particles captured by the magnetic medium were washed into the concentrate collection bucket.

Figure 11 shows the effect of the magnetic intensity on the separation performance of the Jones magnetic separator. The magnetic intensity described in the figure is the highest magnetic intensity on the magnetic medium plate teeth in the Jones magnetic separator, the spacing between the magnetic medium teeth is 1.5 mm, and the tooth angle is 45°. Figure 11 indicates that as the magnetic intensity increases, the recovery gradually improves while the tailings yield steadily decreases. At a magnetic intensity of 1.0 T, the concentrate can attain a maximum TiO₂ grade of 15.4%, yet the recovery rate is relatively low, at only 61.2%. Despite achieving a higher recovery rate by increasing the magnetic intensity to its maximum of 2.2 T, the concentrate grade is lower, resulting in ineffective enrichment.

Figure 11 also includes the gravity separation index, which was derived from the optimal performance in Section 4.3. As illustrated, at a magnetic intensity of 1.4 T, recovery similar to those obtained through gravity separation, approximately 80%, can be achieved. However, the concentrate TiO₂ grade obtained through gravity separation is higher, about 1.5 times that of magnetic separation. Additionally, the tailings grade of RC is relatively lower compared to the condition of a magnetic intensity of 1.4 T, with a higher yield. This indicates that the effect of RC is superior when processing the ilmenite samples.

XRD analysis was performed on the concentrate products obtained through gravity separation and magnetic separation, and the results are shown in Figure 12. As depicted, the peak heights of augite and chlorite increased significantly as the magnetic intensity increased. This indicates that the augite and chlorite in the feed sample possess certain magnetic properties. With an increase in magnetic intensity, gangue minerals will be captured onto the magnetic medium together with the ilmenite, making it difficult to achieve the separation of gangue minerals from ilmenite.

In the XRD curve of the concentrate products of RC, the peak heights of gangue minerals, particularly chlorite and augite, are relatively low. This is due to the significant difference in density between ilmenite (4450 kg/m³) and gangue minerals such as augite (3230 kg/m³), chlorite (3210 kg/m³), and feldspar (2600 kg/m³) in the feed sample. Therefore, the density difference was utilized as the basis for separation in gravity separation. Consequently, the density difference between gangue minerals and ilmenite resulted in a more favorable separation effect during the reprocessing.

When the highest separation efficiency is achieved through gravity separation, the recovery rate is similar to that achieved by magnetic separation with a magnetic field intensity of 1.4 T. During magnetic separation, a higher recovery rate can be achieved with a magnetic field intensity of 2.2 T. Figure 13 presents information on various particle sizes of products at magnetic intensities of 1.4 T and 2.2 T and the RC highest separation efficiency.

As seen in Figure 13b,c, increasing the magnetic field intensity significantly improved the concentrate yield and recovery of all size particles. The stronger magnetic force enhanced the recovery of coarse particles that are more susceptible to loss. However, due to the interference of magnetic gangue minerals, as seen in Figure 13a, a large number of particles were adsorbed onto the medium and thus entered the concentrate products, resulting in poor enrichment of ilmenite.

By raising the magnetic intensity from 1.4 T to 2.2 T, the yield of the +38 μm coarse particles in the concentrate increased from 39.9% to 89.2%, while the grade decreased from 5.68% to 3.38%. The grade of the +38 μm particles in the feed was 2.43%. Hence, excessively high magnetic field intensity will result in a reduction in the selectivity of +38 μm particles.

As seen in Figure 13, increasing the magnetic intensity will similarly result in an elevation of the recovery rate of ultrafine ilmenite of −18 μm and can effectively reduce the content of TiO₂ in tailings products, as was reported in the introduction. However, it also leads to a selective reduction in which the grade of the concentrate is not significantly enhanced relative to the feed. Therefore, when employing magnetic separation to process this sample, lower magnetic field strength will drastically increase the tailings yield, especially for fine particles of −18 μm, which will cause a significant reduction in recovery. On the other hand, increasing the magnetic field strength will result in a reduction in separation efficiency, making it difficult to achieve effective concentration of concentrate products.

In contrast, as shown in Figure 13a, RC achieved a relatively high concentrate grade for all particle sizes and demonstrated a high enrichment ratio for the ultrafine −18 μm particles. As for the −18 μm material, the tailings grade (Figure 13d) generated by gravity separation was the same as the results obtained by magnetic separation at the highest intensity, but with a higher yield. Therefore, it can be seen that gravity separation has a good effect on the separation of ultrafine ilmenite minerals.

5. Conclusions

The efficient and cost-effective preconcentration of fine-grained ilmenite presents a challenge, as traditional beneficiation techniques are often prohibitively expensive or difficult to implement effectively. However, experimentation has demonstrated that the reflux classifier (RC) is highly effective for the preconcentration of low-grade ultrafine ilmenite.

• The results indicate that using the reflux classifier for the preconcentration of ultrafine ilmenite ore, with an underflow rate of 17 mL/min, a fluidization water velocity of 1.78 × 10⁻³ m/s, and a throughput of 1.42 t/(m²·h), yields a maximum achievable separation efficiency of 50.0%. Under these conditions, the concentrate grade of TiO₂ is 20.3%, with a recovery rate of 82.8%. The tailings yield is 62.6%, with a tailings TiO₂ grade of 2.44%.
• A comparative analysis of the beneficiation results between magnetic separation and the reflux classifier reveals that certain gangue minerals in the ore, such as augite and chlorite, exhibit some degree of magnetism and can be captured by the magnetic medium due to increased magnetic intensity. This negatively impacts the quality of the concentrate product. However, the magnetic properties of the gangue do not affect the efficiency of the gravity process. For this specific material, magnetic separation was ineffective, while the reflux classifier achieved favorable results.
• It is noteworthy that the ore sample used in the experiment was exceptionally fine-grained, with 50% of the material having particle sizes finer than 13.8 μm. Despite this, the reflux classifier demonstrated commendable performance, highlighting the significant potential of this technology for the cost-effective and efficient preconcentration of ultrafine ilmenite ore.