Improved Flotation of Fine Flake Graphite Using a Modified Thickening Process

Abstract

Natural graphite ores are usually upgraded by froth flotation. However, complex processes with multistage grinding and flotation are required to achieve decent liberation and separation of graphite and gangue minerals. This study reports a short and improved flotation process for fine flake graphite ore by employing a thickening stage. The results indicated that increasing the regrinding concentration via thickening can improve the grinding efficiency and, thus, shorten the separation process. With thickening, a high-grade intermediate concentrate of 96.01% was obtained after three steps of cleaner flotation, which is close to the final concentrate after five steps. Particle size distribution analysis and FIB-SEM-EDS studies suggested that the main contribution of thickening–regrinding was to achieve better abrasion rather than attrition of the graphite flakes, thus liberating the interlayer impurities without reducing the size of the graphite flakes. This study offers a more cost-effective pathway for the simplified flotation of natural graphite ores.

1. Introduction

Graphite is an indispensable and crucial raw material for emerging scientific and industrial fields, such as metallurgy, electronics, defense, and aerospace [1]. Natural graphite is mainly found in three types: crystalline flake, amorphous, and crystalline vein [2]. Natural graphite ore can be easily purified by froth flotation due to its natural hydrophobicity [3,4,5,6]. Flake graphite exhibits a stronger hydrophobicity and is more floatable than amorphous graphite [7,8,9,10,11,12]. As illustrated in Figure 1, graphite beneficiation generally consists of a multistage grinding and flotation process [13,14]. The main reasons for this are described as follows: (1) to preserve the high-value large graphite flakes from being damaged by overgrinding, and (2) to gradually liberate the gangue minerals embedded in between the graphite flakes or the graphite intercalated in the gangue minerals [15].

In particular, for those fine flake or amorphous graphite resources, up to 10 or more stages of grinding-flotation modules are usually required to allow sufficient dissociation and separation of the graphite and ultrafine scale gangue particles. However, multistage fine grinding drastically deteriorates flotation performance due to the intensified entrainment of fine gangue minerals [16,17,18] and severe slime coating [19]. In addition, an overlong process flowchart yields multiple middling products that require proper treatment [20], placing higher demands on plant operations and imposing a heavy economic burden. Many new attempts, such as scrubbing [21,22], ultrasonic treatment [23,24,25], and equipment retrofitting [26,27,28], have been employed to improve the flotation of graphite ore. However, none of these approaches could effectively simplify the flowchart. Instead, most of them ended up introducing additional processes or developing new techniques, which are usually complex and costly.

Moreover, the solids concentration of cleaner flotation is usually kept relatively low, around 3–10%, to ensure decent separation efficiency [21,29,30]. However, feed streams with low solids concentrations can lead to low efficiencies in regrinding. Since the aim of multistage grinding and flotation is to gradually dissociate the intercalated impurities from the graphite interlayers, increasing the solids concentration during grinding might be a good option to enhance the grinding efficiency and improve the liberation between graphite and gangue minerals. As a result, the overall flotation process might be shortened. Thickening is a typical dewatering process in the mineral process [31], which is usually used to separate liquids from the final concentrate or tailings via gravity sedimentation [32] or filtration [33]. However, few studies have reported the dewatering of the streams for the intermediate grinding stages in graphite ore beneficiation.

This work aimed to improve the traditional graphite flotation process by adding a thickening stage to increase the solids concentration of the feed subject to regrinding. The operational parameters of this process were optimized based on roughing flotation and open-circuit flotation tests. The improvement of thickening in the regrinding stage was confirmed by a particle size distribution analysis and visualized using focused ion beam scanning electron microscopy (FIB-SEM) techniques.

2. Experimental

2.1. Materials and Reagents

The fine flake graphite sample was obtained from Heilongjiang province, China, with an average fixed carbon (FC) content of 17.87%. The chemical composition and mineralogical properties of this sample are given in

Table 1. Chemical composition of the raw graphite ore (wt%).

Composition	SiO2	Al2O3	CaO	Fe2O3	SO3	K2O	MgO	TiO2	Na2O	P2O5
Content/%	46.662	9.645	8.397	5.292	4.186	1.417	1.409	0.502	0.286	0.280
Composition	V2O5	MnO	Cr2O3	SrO	ZrO2	Rb2O	Y2O3	MoO3	LOI	FC
Content/%	0.185	0.128	0.043	0.019	0.017	0.008	0.005	0.005	21.515	17.870

and Figure 2, respectively. The main chemical component in the raw ore was SiO₂ (46.66%), followed by Al₂O₃ (9.65%), CaO (8.40%), Fe₂O₃ (5.29%), SO₃ (4.19%), K₂O (1.42%), and MgO (1.41%), respectively, with other minor components accounting for less than 1%. In addition, the main associated gangue minerals were quartz, kaolinite, muscovite, and diopside. The particle size distribution results in Figure 3 show that nearly 90% of the graphite particles were less than 0.15 mm in size, of which around 60% were less than 0.074 mm, indicating the fine-flake nature of this graphite ore.

Kerosene and terpenic oil, which were used as the collector and frother for flotation, were purchased from Kermel Chemical Reagent Co., Ltd., Tianjin, China. All the chemicals were directly used without further purification.

2.2. Experimental Tests

For the roughing flotation test, 250 g samples were first ground in an RK/BM Φ250 × 100 ball mill with an agitation speed of 210 rpm, followed by flotation using a 1.0 L RK/FD mechanically agitated flotation cell at 1800 rpm (Wuhan Rock Crush & Grand Equipment Manufacture Co., Ltd., Wuhan, China).

For the open-circuit tests, the roughing concentrate was subjected to several steps of regrinding-cleaner flotation modules using an XMQ Φ150 × 50 conical ball mill and a 1.0 L RK/FD flotation cell, respectively. The experimental parameters were optimized by evaluating the flotation recovery and the grade of the concentrate product (FC content).

2.3. Characterization

The FC content of graphite products was analyzed using the National Standard GB3251-2008 Graphite Chemical Analysis Method. X-ray fluorescence (XRF) spectrum analysis (Zetium, Panalytical. B.V, Almelo, The Netherlands) and X-ray diffraction (XRD) analysis (Empyrean, Panalytical. B.V, The Netherlands) were used to characterize the chemical composition and mineralogy of the graphite sample, respectively. The morphology and surface species of the concentrate were examined by focused ion beam scanning electron microscopy (FIB-SEM, ZEISS Crossbeam 540), equipped with energy-dispersive X-ray spectroscopy (EDS, X-Max 50). The desiccated graphite flakes were coated with gold using a LUXOR AU sputter coater. For the tomography characterization of the graphite flakes, the stage was first tilted to around 52°, and a focused ion beam (FIB, 30 kV, 2.5 nA, and 1 μs dwell time) was then emitted to cut through and expose the selected area of interest for image acquisition. The average size distribution of the graphite concentrate was determined using a laser particle size analyzer (BT-9300S, Bettersize Instruments Ltd., Dandong, China).

3. Results and Discussion

3.1. Optimizing the Roughing Flotation Conditions

Graphite flotation usually follows multiple grinding-flotation steps, each of which improves the quality of the final concentrate to a greater or lesser extent. Notably, the highest separation efficiency is always accomplished by the very first roughing flotation. Therefore, the parameters of the roughing flotation, including the grinding fineness, the dosage of the collector or frother, and the flotation time, are of great importance and were evaluated first. The corresponding flotation results are given in Figure 4.

Proper coarse grinding is crucial to maximizing the liberation rate between graphite and gangue minerals while minimizing the damage to graphite flakes. Figure 4a shows the flotation recovery as a function of the coarse grinding fineness. Generally speaking, a −0.074 mm content, which denotes the percentage of particles below the 0.074 mm size fraction in all the grinding outcomes, is an important factor that has been widely used to represent the grinding fineness in mineral beneficiation and to evaluate the grinding efficiency. There were no exceptions for graphite beneficiation. It can be seen that the grade of roughing concentrates increased gradually from around 33% to 40% as the −0.074 mm content increased from 60% to 90%. Intensive grinding produced better liberation of gangues from the graphite interlayers; therefore, those graphite flakes with a higher purity could be preferentially floated. However, the flotation recovery in Figure 4a suffered an undesirable decrease under finer grinding conditions, mainly because those gangue minerals with underliberated interlocking graphite flakes were not easily captured by bubbles and remained in the flotation slurry.

Figure 4b,c show the flotation results in relation to kerosene and terpenic oil concentrations, respectively. Kerosene and terpenic oil, as surface-active reagents, are essential for enhancing the hydrophobicity of the graphite surfaces and improving the flotation efficiency. As expected, the flotation recovery showed a continuous upward trend from below 90% to nearly 100% with an increase in kerosene or terpenic oil dosage. However, the concentrate grade first rose to approximately 40% and then fell by 4% at higher reagent dosages due to the undesirable collection of graphite–gangue interlocked particles. Figure 4d shows the effect of flotation time (scraping time) on the flotation recovery and grade of the roughing concentrate. Increasing the scraping time allowed the collection of more graphite particles from the slurry; thus, a rising trend for flotation recovery was witnessed. Again, the continuous recycling of gangue-associated graphite particles resulted in the collection of more impurities. As a result, the concentrate grade showed a decreasing trend.

The optimal conditions chosen for the rougher grinding and flotation stages largely relied on a balance between maximizing the flotation recovery and ensuring a decent grade for the intermediate products. In other words, higher recovery means more graphite products being collected and subjected to sequent processes rather than discharged into the waste tailings, which is of great economic importance. However, a higher recovery usually results in a lower overall purity (grade) of the product. Although the grade of intermediate products could easily be upgraded by the sequent grinding-flotation steps, a decent grade is also important at this stage as the feed with a lower grade would pose a detrimental impact on the following grinding process. Specifically, the presence of more impurities (usually some harsh gangue particles, such as quartz) would easily break down the large graphite flakes during grinding, resulting in a decrease in the fraction size of the graphite product. Therefore, it is always recommended to seek a balance between recovery and the grade. As a result, considering both the flotation recovery and concentrate grade, an optimized roughing flotation condition was determined: a coarse grinding fineness of 80.13% of the −0.074 mm content, a kerosene dosage of 470 g/t, a terpenic oil dosage of 50 g/t, and a scrapping time of 4 min. In addition, the recovery and grade of the roughing concentrate were 97.91% and 41.23%, respectively.

3.2. Optimizing the First Regrinding Condition

The grade of the roughing concentrate increased from 17% to over 40%, which demonstrates that a considerable proportion of impurities were still present in the concentrate. These impurities were either associated with the graphite or sandwiched within the interlayers. Therefore, further grinding was needed, particularly the first regrinding, which was very important in liberating most of the gangues from the graphite.

Figure 5 shows the flotation results as a function of different first regrinding finenesses. Unlike roughing flotation, the recovery showed a monotonic increase as the flotation feed became progressively finer. In contrast, although the concentrate grade slightly increased from 66% to 69% with the increase of the −0.074 mm content from 71% to 76%, it dropped sharply again to 63% when further increasing the −0.074 mm content to 87%. It is hypothesized that intensive grinding resulted in finer gangue minerals, which were prone to enter the froth layer through entrainment, thereby deteriorating the grade of the final concentrate.

3.3. Comparing the Flotation Performance with and without the Thickening Stage

An open-circuit flotation process with four regrinding and five cleaner flotation steps was developed for this graphite ore, as illustrated in Figure 6, and thereby employed to further evaluate the feasibility of using a thickening stage. In detail, this open-circuit flowchart comprises one roughing grinding-flotation module, four stepwise regrinding-cleaner flotation modules, and one subsequent cleaner flotation. The only difference between the three flotation tests is whether or not a thickening process is used prior to the second regrinding, as annotated in Figure 6. Specifically, the three approaches are described as follows: (1) no thickening, (2) thickening and dewatering by gravity sedimentation, and (3) thickening by vacuum defoamation and filtration. In addition, the solids concentrations of the corresponding streams for second regrinding were 11.96%, 20.64%, and 58.30%, respectively. Detailed solids concentrations for each step of the three processes are compared in

Table 2. The solid concentrations of each step for the open-circuit flotation flowchart (wt%).

Processes	Process 1	Process 2	Process 3
Coarse grinding	65.00	65.00	65.00
Rougher flotation	25.14	25.02	25.21
Regrinding 1	30.31	31.47	31.82
Cleaner flotation 1	10.25	10.29	10.71
Regrinding 2	11.96	20.64	58.30
Cleaner flotation 2	5.61	5.46	5.62
Regrinding 3	10.28	9.56	10.32
Cleaner flotation 3	4.48	4.49	4.17
Regrinding 4	8.87	8.64	8.69
Cleaner flotation 4	3.94	4.07	3.94
Cleaner flotation 5	3.67	3.91	3.84

. It is noticeable that the solids concentrations for other grinding and flotation steps were almost identical among the three processes. The flotation results are compared in

Table 3. Comparison of the open-circuit flotation results with and without thickening.

Products	Process 1			Process 2			Process 3
	Yield %	FC %	Recovery %	Yield %	FC %	Recovery %	Yield %	FC %	Recovery %
Conc	14.33	95.86	83.88	15.48	96.78	88.18	15.13	97.12	87.53
M1	18.56	2.24	2.54	19.35	2.73	3.11	20.37	2.21	2.68
M2	4.51	10.92	3.00	3.88	7.96	1.82	5.79	7.23	2.50
M3	2.15	21.29	2.79	1.67	11.72	1.15	0.93	26.01	1.44
M4	1.09	42.04	2.79	0.63	31.93	1.18	0.39	54.39	1.28
M5	0.36	66.90	1.47	0.17	64.02	0.65	0.22	85.35	1.13
Tailing	59.00	0.98	3.53	58.83	1.13	3.91	57.16	1.01	3.44
Raw Ore	100.00	16.38	100.00	100.00	16.98	100.00	100.00	16.78	100.00

and Figure 7.

As shown in Table 3 and Figure 7, the increment in concentrate grade became less pronounced as the grinding-flotation module progressed. For example, the grade increased by nearly 30% after the first cleaner flotation, while less than a 5% grade increment could be obtained after the fourth or fifth cleaner flotation. Notably, the concentrate grade from Process 3 showed a more significant increase, followed by that from Process 2, while Process 1 without thickening showed the slowest increment. More straightforwardly, the concentrate grade from Process 3 reached 96.01% after three regrinding-flotation modules, which is equal to the final concentrate after five steps of cleaner flotation without thickening (Process 1). This is indicative that increasing the solids concentration enhances the second regrinding efficiency and contributes to better dissociation between graphite and gangue minerals. As a result, a qualified concentrate could be obtained with a shorter flotation process. Most significantly, grinding is one of the most energy-consuming processes in the beneficiation of minerals. With our design, the original process with four regrinding and five flotation stages could be potentially reduced to only two regrinding and three flotation stages. In other words, the overall process was almost cut in half, which is of great economic significance. For example, capital investment, operating cost, and most importantly, energy consumption were all reduced to a great extent. Moreover, the production capacity would be largely improved due to the faster processing circuit with the shortened process. Moreover, the recovery loss of the M2–M5 was also reduced after thickening, resulting in an increased flotation recovery from 83.88% to 87.53%. Therefore, in addition to simplifying the flotation process, employing a thickening stage is also of great economic importance as it could increase both the flotation recovery and the grade of the final product.

Note that the selection of thickening for the first regrinding-flotation concentrate was based on the following two considerations: a rational solids concentration and a suitable proportion of gangue particles. Specifically, as indicated by the solids concentrations in Table 2 and the yield % in Table 3 for the process without thickening, 59% of the raw feed was discharged as tailings after coarse flotation, with 41% of them being collected and subjected to the first regrinding and cleaner flotation. The solids concentration for the first regrinding was approximately 30%, which is in a reasonable range for good grinding. Besides, the grade (around 40%) of the coarse concentrate indicated the presence of a large proportion of gangues, and increasing the solids concentration for the first regrinding feed would enhance the abrasion of these harsh and sharp gangue particles to the graphite flakes. However, after the first cleaner flotation, the majority of liberated gangue particles were separated and discharged into the middlings (Middlings 1) (a yield of 18.56%), which indicates a significantly decreased yield of the first cleaner concentrate (around 21%). As a result, the solids concentration for the second regrinding was reduced dramatically to only 12% when compared to the first regrinding (30%). In addition, after the first cleaner flotation, most of the liberated gangue was removed from the circuit, with only a small proportion of impurities either associated or intercalated inside the graphite flakes (the grade of the first cleaner concentrate was close to 70%; Figure 7). Unlike the strong damage to graphite flakes, the presence of this small proportion of gangue could serve as supporting grinding media to improve the grinding efficiency. As a result, increasing the solids concentration is beneficial to improving the grinding efficiency for the second regrinding.

As for the third regrinding, it should be mentioned that the concentrate grade at this stage was nearly 95%, and a slight regrinding, even under low solids concentrations (8–10%), is sufficient to liberate the remaining gangue and to obtain qualified concentrate by flotation. In addition, with the increased purity of the graphite concentrate, the foam products become more stable and harder to deform, and therefore, employing the thickening process will reduce efficiency. In addition, there is no need to introduce another thickening process again for the subsequent process.

Moreover, as demonstrated in the flowchart in Figure 6, the original process without thickening consisted of four regrinding and five cleaner flotation steps, which is a very typical, and relatively simple process when compared with those with 10 or more regrinding-flotation modules. Nevertheless, introducing additional thickening is beneficial to enhancing the subsequent regrinding and contributing to a graphite concentrate of higher purity. However, the main reasons why we only introduced one step of filtration are as follows: (1) After introducing the thickening process, we could confidently obtain a qualified graphite concentrate after only two regrinding and three cleaner flotation steps, which dramatically simplified the process; and (2) the filtration efficiency is not very high for purified and finer graphite flakes due to the deforming difficulty. Instead, it would adversely increase the complexity of the process; thus, it was not recommended to introduce additional thickening from the perspective of economy and productivity.

3.4. Mechanism of the Thickening and Regrinding Process

The particle size distributions of the products after the second regrinding were measured to understand the impact of thickening on regrinding. The results are shown in Figure 8. It is evident that both the interval and cumulative size distribution curves for all three products overlap with each other, indicating that regrinding has an imperceptible effect on changing the size fraction, regardless of the solids concentration. However, the ensuing flotation concentrate grades differ: a higher FC content was obtained when using the thickening process. Therefore, it is assumed that the main contribution of regrinding is stripping thick graphite along the flake direction to liberate interlayer impurities rather than crushing the graphite flakes. In other words, regrinding mainly thins the graphite flakes while having little impact on their size, as schematically illustrated in Figure 9. Increasing the solids concentration of grinding improves the abrasion and, thus, facilitates the dissociation of gangue minerals from the graphite interlayers. As a result, a higher concentrate grade was observed after thickening.

FIB-SEM was used to characterize the microstructure of the graphite flakes after the second regrinding process. Figure 10a shows the morphology of one bulk graphite flake with a thickness of about 10 μm. The employment of the focused ion beam allowed us to cut the graphite flake open and expose a cross section, as shown in Figure 10b. Interestingly, some brighter streaks were detected in the magnified image of the cross section (Figure 10c), which are thought to be gangue minerals trapped inside the graphite interlayers. The elemental mapping of the same cross section in Figure 10d,e shows that the dominant constituent element was C, with only small amounts of other elements, such as Si, Fe, and Ga. In addition, the presence of these impurities is further confirmed by the selected spot analysis in Figure 10f. Detailed elemental information is given in

Table 4. The atomic information of the selected area from the graphite cross section.

Map Sum Spectrum			Spectrum 1			Spectrum 2
Element	Wt%	Atomic %	Element	Wt%	Atomic %	Element	Wt%	Atomic %
C	91.81	97.82	C	26.10	47.74	C	100.00	100.00
O	-	-	O	14.66	20.13	O	-	-
Si	1.86	0.85	Si	22.76	17.80	Si	-	-
S	-	-	S	0.46	0.32	S	-	-
Fe	3.64	0.84	Fe	33.90	13.34	Fe	-	-
Ga	2.69	0.49	Ga	2.13	0.67	Ga	-	-
Total:	100.00	100.00	Total:	100.00	100.00	Total:	100.00	100.00

. Specifically, the sum spectrum shows that C was the dominant element, accounting for 91.81% of the mass fraction. The spectrum of selected point 1 indicates the presence of C, O, Si, Fe, Ga, and S, with mass fractions of 26.10%, 14.66%, 22.76%, 33.90%, 2.13%, and 0.46%, respectively, probably assigned to the pyrite and quartz. Further grinding was needed to dissociate these impurities from the graphite interlayers to improve the quality of the final concentrate.

4. Conclusions

An improved flotation process was developed for a fine flake graphite ore. The parameters of the roughing flotation and the first regrinding-cleaner flotation step were optimized based on the flotation recovery and concentrate grade. A thickening stage was employed prior to the second regrinding process, which resulted in a significant 5% increase in flotation recovery and a 2% increase in concentrate grade compared to the process without thickening. More importantly, the particle size distribution analysis of the intermediate product after the second regrinding step indicated that the main mechanism of regrinding was to abrade the graphite flakes to dissociate the impurities within the interlayers, which was further confirmed by the FIB-SEM-EDS analysis. It is suggested that increasing the solids concentration by thickening can significantly improve the grinding efficiency and facilitate the liberation of gangues from the graphite interlayers.

Our study offers a feasible approach to shorten and improve the beneficiation process of graphite ore. Nevertheless, further investigations are desired to better reveal the topographic features of graphite flakes in our future study. For example, we intend to employ the FIB-SEM tomography technique, also known as “slice and view”, to gradually remove and expose new cross sections. By recording the multiple newly exposed faces and the 3D imaging process of the graphite flakes, we would be able to obtain more detailed and meaningful topographic features at high resolutions. Moreover, we are scheduled to further check the feasibility of our thickening process by scaling up the experiments (on the pilot scale or even industrial plant scale), starting with the same graphite sample provided by our cooperative partner.