Flocs Properties and Flotation Performance of Fine Diaspore with Energy Input Pretreatment Induced Using Sodium Oleate

Abstract

Energy input, an important factor affecting flocs properties and flotation performance, has rarely been studied in the field of diaspore flotation, which has severely limited our understanding of the flocculation flotation of fine diaspore. Therefore, in this study, the flocs properties and flotation performance of fine diaspore with energy input pretreatment were studied through flotation kinetics, flocs size measurements, and fractal dimension analysis. The results showed that the flocs size increased and the flocs structure became looser with the increasing energy input, while the flocs size decreased and the structure became compact when the energy input exceeded 10.93 kJ/m³. Meanwhile, there were significant differences in the flotation performance under different energy input pretreatment conditions, suggesting that the flotation performance of the fine diaspore was closely related with the flocs properties generated during the agitation process. In particular, the flotation performance was positively correlated with the flocculation degree of flocs, to a certain extent. The flocculation flotation of the fine diaspore benefited from a suitable energy input, and an excessive energy input was not conducive to flotation performance.

1. Introduction

The low aluminum oxide-to-silica mass ratio (A/S) of diaspore, typically ranging from 4 to 6, makes it uneconomical to feed the Bayer process directly because of the high energy and reagent consumption [1,2]. Therefore, it is essential to remove the gangue minerals, such as illite and kaolinite, in order to increase the A/S of the diaspore ores. Flocculation flotation, mainly composed in the flocculation stage and flotation stage, has been widely confirmed to be an effective solution to solve the problem [3].

In flotation, the flocculation of fine mineral particles generally occurs when the hydrocarbon chains of collectors contact the particle surface [4]. Energy input pretreatment is essential in order to allow the hydrophobic association to take place because of the electrostatic repulsion that acts as a barrier to the process, which means that the energy input is an important parameter in evaluating the flocculation process and flocs properties (i.e., flocs size, structure, density, shape, and porosity). Actually, numerous papers about flocs properties and the flocculation process influenced by the energy input have been published, most of which were achieved by controlling the shear rate or agitation conditions; for example, the relation between the flocs size distribution and hydrodynamics [5], the transfer of mineral particles between flocs under various conditions of agitation [6], the temporal evolution of flocs size and structure in low-shear flow [7], and the fragmentation and erosion of aggregates in shear flow [8].

In addition, modern characterization techniques of flocs properties have been developed. Flocs are easily destroyed during the transfer or measurement process, so the characterization techniques of flocs size or flocs size distribution (FSD) are mainly divided into two types: sampling techniques and in situ techniques. Sampling techniques include the laser particle size analyzer, microscope, dynamic image analysis (DIA), and so on. Meanwhile, in situ techniques include a focused beam reflectance measurement (FBRM), particle vision measurement (PVM), and so on. On the other hand, the characterization of the flocs structure using the mass fractal dimension (D_f) has been accepted and investigated by more and more researchers [9,10,11]. Any object we can find must have a mass fractal dimension of 1 ≤ D_f ≤ 3 [12]. However, the D_f ranges of different flocculation models are usually different. The cluster–cluster aggregation model has been the most widely investigated. When the average size of the cluster grows with time evolution in the power law, and the D_f of the clusters ranges from 1.75 to 1.85, the aggregation model is usually diffusion-limited cluster aggregation (DLCA). In the case that the average size of the cluster, accompanied by a compact structure, grows exponentially with time, and the D_f value of clusters is in the range between 2.0 and 2.1, and the aggregation model is known as reaction-limited cluster aggregation (RLCA) [13].

There is a considerable body of experimental and theoretical evidence to indicate that energy input pretreatment (or agitation) also plays an important role in the flotation process, especially in flotation kinetics [14,15]. The effects of particle size [16], bubble size [17], hydrodynamics [18], and multi-stage agitation [19] on flotation performance under energy input conditions have also been studied. However, to the authors’ knowledge, the papers about diaspore flotation mainly focus on flotation reagents [20,21]. The relationship between the flocs properties and the associated flotation performance as a function of the energy input has rarely been studied, which is not conducive to deepening our understanding of the effect of the energy input on flocs properties and flotation performance.

Thus, the effects of energy input pretreatment on the flocs properties and flotation performance of fine diaspore were investigated in this study. Sodium oleate (NaOl), acting as the flocculant and collector, was used to induce the surface hydrophobicity of diaspore. This study aimed at revealing the influence of the energy input on the flocs properties and the associated flotation performance of fine diaspore particles.

2. Materials and Methods

2.1. Materials and Reagents

The pure mineral samples of diaspore used in this study were obtained from Xiaoguan, located in Henan Province, China. The samples were first crushed using a laboratory jaw crusher, followed by handpicking to remove the impurity minerals, and then were ground in a vibration mill to obtain −38 μm size fraction. Figure 1 shows the primary particle size distribution of the samples measured using Malvern Mastersizer2000 (Malvern Panalytical Ltd., Malvern, UK). The D₁₀, D₅₀, and D₉₀ (the maximum particle size below which 10%, 50% and 90% of the sample volume existed) of the samples were 2.706, 11.147, and 35.475 μm, respectively. The XRD spectra are shown in Figure 2, and the chemical analysis results are presented in

Table 1. Chemical analysis results of the diaspore samples.

Components	Al2O3	SiO2	TiO2	Fe2O3	CaO	MgO	K2O	Na2O	LOI
Content (%)	80.77	0.89	2.76	0.756	0.01	0.053	0.008	0.025	14.728

LOI = Loss on ignition.

. The diaspore samples had a high purity of 94.441%, based on XRD spectra and chemical analysis. The sodium oleate (NaOl) used in this study was analytical reagent (AR), and was purchased from the Tianjin Kermil Chemical Reagents Development Centre (Tianjin, China). The hydrochloric acid (HCl) and sodium hydroxide (NaOH) were chemically pure and acted as pH regulators. Distilled water was used in all of the tests.

2.2. Hydrophobic Flocculation and Energy Input Measurement

The schematic of the experimental devices used to determine the flocs properties and flotation performance is shown in Figure 3. The hydrophobic flocculation of the diaspore in an aqueous solution was conducted in a 160 mm × 70 mm cylinder container with four baffles of 100 mm × 5 mm. The impeller was driven by an IKA Eurostar power control-visc6000 blender (Werke GmbH & CO. KG, Staufen, Germany), and a torque meter was installed to measure the torque. The semidiameter of the impeller was 20 mm, and its blade (10 mm × 10 mm) was vertical to the axis. The impeller was 20 mm away from the bottom of the cylinder container.

Mineral pulp was prepared by adding 15 g of diaspore to 285 mL of distilled water. A given dosage of NaOl, and a suitable HCl or NaOH solution were added too. After that, the suspension was pretreated in the cylinder container at a set speed for a certain time. During the agitation, the mineral pulp of the fine diaspore was carefully sampled using injection syringes for all the measurements. The energy input (ω) in the pulp system was calculated as follows [4]:

$$\omega =\frac{T·N·t}{9549V}$$

(1)

where ω is the energy input on the mineral pulp (J/m³), T is the torque to overcome the fluid resistance (N·m), N is agitation speed of the blender (rpm), t is the agitation time (s), and V is the pulp volume (m³). The energy input was a term that described the agitation intensity on the pulp. In this study, the agitation time and pulp volume were set at 6 min and 300 mL, respectively. The torque and energy input values under different agitation speed in this study were presented in

Table 2. Torque and energy input values under different agitation speed.

Agitation Speed(rpm)	900	1200	1500	2100	2700	3000	3300	3900
Torque (N∙m)	0.004	0.010	0.015	0.025	0.030	0.029	0.030	0.030
Energy input (kJ/m3)	0.45	1.51	2.83	6.60	10.18	10.93	12.44	14.70

.

2.3. Flocs Size and Fractal Dimension Measurements

The flocs size and fractal dimension of the diaspore flocs were measured by using a Malvern Mastersizer2000 (Malvern Panalytical Ltd., Malvern, UK). After the measurements, we could obtain the size of the flocs directly, and extract the data (scattered light intensity I(q) and incident angle θ) from the Mastersizer2000, through the supporting application program. Each pulp sample was measured three times, and the average of the three measurements was used as the result.

The flocculation behaviors of the fine diaspore were described using a self-defined parameter, flocculation degree (R_f), which was calculated as follows:

$${\mathrm{R}}_f=\frac{\left(D_b-D_a\right)}{D_a}\times 100\%$$

(2)

where D_a and D_b describe the volume weighted mean (directly obtained from the Mastersizer) of the diaspore flocs before and after the energy input pretreatment, respectively.

The fractal dimension (D_f) was calculated using the relationship between I(q) and the scattering vector q, as follows [22]:

$$I\left(q\right)\propto q^{-D_f}$$

(3)

where θ was converted to q using the following expression:

$$q=\left|q\right|=\frac{4\pi n}{\lambda }\sin (\frac{\theta }{2})$$

(4)

where n = 1.33 was the refractive index of the aqueous solution, and λ = 466 nm was the laser wavelength of the Mastersizer2000. We plotted the log(I(q)) and log(q) curves, and estimated D_f from the inverse of the slope by fitting a straight line in the range of −3.5 < q < −2. An example of determining the fractal dimension of the flocs by the relationship between I(q) and q is shown in Figure 4. Generally, the D_f of the flocs is positively correlated with the compactness of the flocs structure. The larger the fractal dimension of the flocs, the more compact the flocs structure, and vice versa [23].

2.4. Flotation Tests

The flotation tests were carried out in an XFG_ΙΙ flotation machine (Jilin Exploration Machinery Plant, Changchun, China) with a 300-mL cell at 1800 rpm after the hydrophobic flocculation pretreatment. The pulp was mixed for 2 min and then the flotation began. For each flotation test, the total flotation time was 4 min, and the concentrate was collected in batches using flotation times of 30, 30, 30, 30, 30, 30, and 60 s. The hydrophobic flocs could be mostly recovered and a stable flotation recovery could be obtained by appropriately prolonging the flotation time in the final stage of the flocculation flotation. The concentrate and tails were subsequently collected, filtered, and dried. The flotation recovery was calculated based on the solid mass distributions between the two products. Each flotation test was carried out three times, and the average value was taken as the final value.

A first-order model was used to describe the flotation kinetics of the pretreated pulp, which could be described as follows [24]:

$$\mathrm{R}\left(\mathrm{t}\right)=R_{max}\left(1-e^{-kt}\right)$$

(5)

where k is the flotation rate constant and R_max is the flotation recovery at an infinite time. A nonlinear least square regression was used to calculate k and R_max from the best fit of the curve of the experimental flotation recoveries versus flotation time.

3. Results

3.1. Effect of NaOl Concentration and pH on the Flotation Performance of Fine Diaspore

The effects of the NaOl concentration and pulp pH on the flotation performances of the fine diaspore are shown in Figure 5. It can be seen from Figure 5a that the flotation recovery increased significantly with the increase of NaOl concentration, until it reached the maximum at 5 × 10⁻⁴ mol/L, and then the flotation recovery declined. When the pulp pH ranged from 4 to 12 in the presence of NaOl, the diaspore flotation recovery significantly increased by increasing the pulp pH from 4 to 10, reaching its maximum of approximately 80% at a pH of 10.

The aforementioned flotation results were consistent with the findings reported by other researchers [25,26]. NaOl had a good collecting ability for diaspore when the pH was around 10. The collecting ability was derived from the chemisorption of NaOl on the surface Al active sites of the negatively charged diaspore, which has been demonstrated by Xu et al. using the zeta potential and Fourier transform infrared spectroscopy (FTIR) [27]. Here, the NaOl concentration of 5 × 10⁻⁴ mol/L and a pulp pH of 10 were used as the standard condition to investigate the flocs properties and flotation performance of the fine diaspore with different energy inputs.

3.2. Effect of Energy Input Pretreatment on Flocs Properties of Fine Diaspore

The FSD and fractal dimension (D_f) of the fine diaspore with different energy input pretreatments were investigated. The results are shown in Figure 6. As energy was input into the system, the FSD of the fine diaspore began to change from unimodal to bimodal, and the newly generated peaks were located at a size of around 40 μm (Figure 6a). With the increase of energy input from 0.45 kJ/m³ to 10.93 kJ/m³, the newly generated peaks grew until reaching their highest value. Nevertheless, the size of the diaspore flocs decreased when the energy input reached 14.70 kJ/m³. In order to evaluate the change of the FSD more clearly, the R_f of the diaspore flocs was calculated using the data in Figure 6a, and the results are shown in Figure 6b. It can be seen from Figure 6b that the R_f increased significantly from 4.01% to 103.59%, with the increasing energy input from 0.45 kJ/m³ to 10.93 kJ/m³, and then decreased to 86.84% when the energy input reached 14.70 kJ/m³. Interestingly, the D_f of the flocs showed a completely opposite trend to R_f. D_f first decreased remarkably from 1.724 to 1.450, which indicated that the flocs structure became looser and more open as the flocs grew. However, when the energy input reached 14.70 kJ/m³, D_f increased to 1.522.

According to previous research, the flocculation and destruction of flocs always occurred simultaneously in the agitating process [28,29]. When the rate at which the flocs grew was greater than that at which the flocs were destroyed, the size of the flocs increased, and conversely, the size decreased. The final FSD was an indicator of the balance of the flocs destruction and flocculation [6]. In this paper, the particles collided with each other with the energy input into the pulp system and flocculated into flocs. The rate of flocs formation under the condition of a low energy input (<10.93 kJ/m³) dominated, resulting in an increase in flocs size, accompanied by loose and porous flocs structures. In fact, it was difficult to form large and compact flocs simultaneously in an aqueous solution [23]. In the dilute slurries, the flocs caused by weak attraction were usually small and compact, while the flocs induced by strong attraction were usually large with a loose structure [30]. The excessive agitation and circulation of a high energy input (>10.93 kJ/m³) would lead to an increased likelihood of flocs destruction, and the flocs would break up at their weakest points, giving the flocs a stable structure [7]. Moreover, the internal voids of the flocs would be exposed because of damage to the flocs surface or from expelling inter-floc liquor, and fine particles could integrate into the flocs and make the structure more compact than before [31]. Thus, as the energy input into the pulp system increased, the size and D_f of the flocs decreased and increased, respectively. As for why the fractal dimension of the flocs increased slightly, this might be because the coarse particles in the slurry increased the shear yield stress of the pulp, which made the flocs not too sensitive to shear stress under high-energy input conditions [32].

Therefore, a low energy input could increase the size of the flocs and make the structure more porous. A high energy input would cause the size to decrease, but the structure would become more compact. Furthermore, it should be pointed out that the D_f in this paper was smaller than 1.7, although some papers also reported similar data [33,34], which could be because the particle size of the diaspore in this paper was larger than in other papers [35,36].

3.3. Effect of Energy Input Pretreatment on Flotation Performance of Fine Diaspore

Figure 7 shows the flotation performance of the diaspore flocs under different energy input conditions, and

Table 3. Fitting results of the flotation performance with different energy input conditions.

Energy Input (kJ/m3)	Flotation Recovery ± RSD (%)	Fitting Results
		Rmax (%)	k (s−1)	R2
0.45	66.32 ± 1.11	63.80	0.0244	0.9961
1.51	72.56 ± 0.57	70.69	0.0251	0.9963
2.83	73.73 ± 0.47	72.44	0.0253	0.9974
6.60	79.56 ± 0.38	77.24	0.0259	0.9967
10.18	81.95 ± 0.61	79.90	0.0262	0.9973
10.93	83.98 ± 0.28	82.30	0.0265	0.9972
12.44	81.39 ± 0.80	78.30	0.0260	0.9947
14.70	75.69 ± 0.36	74.61	0.0256	0.9986

presents the fitting results of the flotation kinetics using a first-order flotation model. It could be found from Figure 7 that the flotation performance became better with increasing energy input of the pulp system. When the energy input was 10.93 kJ/m³, the flotation recovery of the diaspore was 83.98%, and the flotation rate constant reached the maximum value of 0.0265 s⁻¹, which indicated that the flocs could float quickly. However, when the energy input exceeded 10.93 kJ/m³, both the flotation recovery and flotation rate constant began to decrease.

The variation of flotation performance with the energy input showed the same trend as the change of R_f and the opposite trend as that of the change of D_f for the diaspore flocs in Section 3.2. The increase in the size of the diaspore flocs could promote associated flotation performance, and the looser and more open structure of the flocs could also be another reason for promoting flotation performance because of the greater buoyancy in the pulp system [22]. Therefore, a suitable energy input promoted diaspore flotation, while excessive energy input was not conducive to diaspore flotation. Safari et al. reported similar findings by investigating the effect of the energy input on the flotation of three sulfide minerals and three oxide minerals [16].

4. Conclusions

In this study, the flocs properties and associated flotation performance of the fine diaspore with energy input pretreatment induced by NaOl were studied. The energy input significantly affected the flocs properties of the diaspore. The FSD of the fine diaspore changed from unimodal to bimodal as the energy input increased, and the newly generated peaks were located around 40-μm in size. Low energy input (<10.93 kJ/m³) was conducive to flocs growth, resulting in larger and less compact flocs forming as the energy input increased, while the size decreased and the structure became compact with a high energy input (>10.93 kJ/m³). Furthermore, the flotation performance under different energy input conditions was closely related to the flocs properties. With the increase of energy input, the flotation recovery and flotation rate constant increased first and then decreased, which showed the same trend as the change of R_f and the opposite trend as the change of D_f of the flocs. The flocculation flotation of fine diaspore benefited from a suitable energy input, and the excessive energy input was not conducive to the flotation performance.