Recently, much attention has focused on an alternative approach involving the use of bacteria to modify the floatability of mineral particles. Many reports have suggested that certain types of bacteria, including T. ferrooxidans, may suppress floatability (20). The postulated mechanism of this suppression is an increase in surface hydrophilicity due to adhesion of bacterial cells (12). If true, the ability of specific bacteria to selectively adhere to and alter the surface properties of specific minerals could be highly useful. In that regard, T. ferrooxidans appears able to adhere well to iron-bearing minerals. If the bacterium adheres selectively to (iron-bearing) pyrite, its addition to a mineral mixture in a flotation column should suppress pyrite floatability without suppressing flotation of the other minerals contained within the mixture.

In this report, the ability of T. ferrooxidans to adhere to selected minerals was investigated with the aims of clarifying whether the bacterium selectively adheres to pyrite and determining whether it will effectively remove pyrite from mixtures of sulfide minerals.

Microorganism, medium, and conditions of cultivation. The iron-oxidizing bacterium T. ferrooxidans ATCC 23270 was used in this study. T. ferrooxidans was cultured in 9K medium containing (per liter of distilled water) 3.0 g of (NH₄)₂SO₄, 0.5 g of MgSO₄ · 7H₂O, 0.1 g of KCl, 0.5 g of K₂HPO₄, 0.01 g of Ca(NO₃)₂, and 44.2 g of FeSO₄ · 7H₂O; the pH was adjusted to 2.5 with 6 N H₂SO₄. The bacteria were cultivated for 3 days in a 10-liter carboy containing 7 liters of 9K medium aerated at 30°C.

Sulfide minerals. Pyrite (Cerro mine, Peru), chalcocite (Broken Hill mine, Australia), millerite (Nepean nickel mine, Australia), molybdenite (Wolfram Camp mine, Australia), and galena (Sweetwater mine, Missouri) were used in this study. The pyrite and chalcocite were purchased from Nihon Chikagaku Co. Ltd., Kyoto, Japan, and the other minerals were purchased from Iwamoto Mineral Co. Ltd., Tokyo, Japan. All of the minerals were museum grade and >90% pure. With the exception of molybdenite, the minerals were ground with an agate mortar; molybdenite was ground with an impeller mill. The crushed particles were fractionated with 200- and 270-mesh sieves. The fractions between minus 200 and plus 270 mesh were sonicated for several minutes in acetone to dissociate the very fine particles from the larger ones. The detached fine particles were removed by decanting; the large particles were dried at room temperature and used in subsequent experiments.

Specific surface areas of sulfide minerals. The surface areas of sulfide mineral particles were determined by the BET method, by surface area gas adsorption (Quantasorb 9S-13; Quantachrome, New York, N.Y.), or by direct microscopic visualization. By the BET method, 1.5 g of each mineral was subjected to three nitrogen-helium gas mixtures containing 9.92, 19.7, and 29.7% nitrogen; the surface area was estimated as a function of nitrogen adsorption. The microscopy entailed photographing each sulfide mineral under a stereoscopic microscope (×750), digitizing the images with an image scanner (GT-9000; Epson Co., Tokyo, Japan), and storing the digital images on a computer for later analysis. The major and minor lengths of >100 randomly selected individual mineral particles were measured with an image analyzer (IP Lab Spectrum; Signal Analytics Co., Vienna, Va.). The average particle size was calculated by assuming the shapes were rectangular parallelepipeds. The number of particles per mineral weight was estimated from both the volume of the average particle and the specific gravity. Surface area per mineral weight was then calculated by multiplying the total particle number per mineral weight by the surface area of a single particle.

Adhesion experiments. Culture medium containing T. ferrooxidans was passed through filter paper (no. 2; Advantec Toyo Co., Tokyo, Japan) to remove precipitated ferric compounds. The filtrate was then centrifuged at 15,000 × g for 15 min to collect the cells. The harvested cells were washed three times with sulfuric acid solution (pH 2.0) and then resuspended in sulfuric acid solution at various cell densities. Cell density was estimated from a previously constructed calibration curve in which optical density at 610 nm was plotted as a function of the density of T. ferrooxidans cells.

Adhesion experiments were carried out using both individual minerals and mineral mixtures. In the case of the former, 0.5 g of each sulfide mineral was added to 2 ml of cell suspension (3.2 × 10⁸ cells/ml). The suspension was then shaken for 1 min with a vortex mixer and then allowed to settle for 5 min. At that point, the optical density of the supernatant was measured to determine the cell density. In the case of mineral mixtures, 0.2 g each of four minerals (chalcocite, millerite, molybdenite, and galena) plus various amounts of pyrite (0.25, 0.5, and 0.8 g) was added to 2 ml of cell suspension (1.7 × 10⁸ cells/ml). The suspensions were shaken and allowed to settle, and the optical density of the supernatant was measured. The number of adherent cells was determined by subtracting the number of cells in the supernatant from the number initially added. As a control a mineral mixture lacking pyrite and pyrite alone were also tested separately.

Flotation experiments. The flotation experiments were conducted with a microflotation column (working volume, 270 ml; height and diameter, 38 by 3 cm). Sulfuric acid solution (pH 2.0) containing methyl isobutyl carbinol (25 μl/liter) was used as the flotation liquor. Air bubbles were generated with a porous glass filter situated at the bottom of the column.

The floatabilities of the minerals were measured by adding 0.5 g of each sulfide mineral to 2 ml of the cell suspension (3.2 × 10⁸ cells/ml). The suspension was then shaken for about 1 min and allowed to settle for 5 min. The settled mineral particles were applied to the flotation column, which was then aerated for 10 min at a rate of 100 ml/min. The particles that floated to the top of the column were collected as froth, while the particles that sank to the bottom were recovered as tailing. Froths and tailings were dried at 80°C and weighed. Floatability was calculated as the proportion of the froth weight in the total mineral weight (i.e., froth plus tailing).

The flotation experiments aimed at separating pyrite from a mineral mixture were carried out with two groups of mineral-pyrite mixtures. The simpler mixtures contained 0.2 g each of pyrite and a single sulfide mineral. Prior to the flotation test, the 0.4-g amount of mineral-pyrite mixture was added to 2 ml of cell suspension (3.7 × 10⁸ cells/ml) to induce bacterial adhesion. The more complex mineral-pyrite mixtures were composed of all five minerals. A low-pyrite mixture contained 0.2 g of pyrite and 0.2 g each of the other four minerals, yielding a 20% final pyrite content; a high-pyrite mixture contained 0.8 g of pyrite and 0.05 g each of the other four minerals, yielding an 80% final pyrite content. Totals of 0.4 g of low-pyrite mixture and 1.0 g of high-pyrite mixture were exposed to 7.3 × 10⁸ and 17.7 × 10⁸ cells, respectively, for 2 min. After exposure to the bacteria, the mineral-pyrite mixtures were subjected to flotation. To remove pyrite by flotation, mineral-pyrite mixtures were fed directly into the middle of the column containing 230 ml of flotation liquor and aerated at a rate of 500 ml/min at 0.1 MPa; 10 min of flotation was allowed to complete the separation.

Froths and tailings were recovered as described above and analyzed to determine the distribution of minerals: they were dissolved in nitric and hydrochloric acids, and the elements in the solution were identified and quantitated with an inductively coupled plasma atomic emission spectrometer (model JY48P; Seiko Industry Co., Tokyo, Japan). The absolute amounts of identified elements were converted to mineral weights on the basis of the chemical formula of each mineral. Each mineral’s floatability was then calculated from mineral weights in the froths and the tailings as described above.

Scanning electron microscopy. To observe the roughness of mineral surfaces, scanning electron micrographs were obtained (model JFM-T300; Joel Co. Ltd., Tokyo, Japan) following standard critical point fixation.

Contact angle. Contact angles were measured with a contact angle meter (type CA-A; Kyowa Surface Science Co. Ltd., Tokyo, Japan). The minerals were first polished with a fine abrasive (C800, A1500, and A3000 [Maruto Co., Ltd., Tokyo, Japan] and 6- and 0.25-μm-diameter diamond pastes [Buehler Co. Ltd., Lake Bluff, Ill.]) to produce mirror-like surfaces. Measurements were made after carefully dropping sulfuric acid solution (pH 2.0) onto the mineral surfaces. At least five droplets (10 μl/droplet) were measured for each mineral.

Zeta potentials. Zeta potentials of sulfide minerals and T. ferrooxidans were determined in a sulfuric acid solution (pH 2.76) with a Plus Zeta potential analyzer (Brookhaven Instruments, Inc., New York, N.Y.). The mineral particles used were further reduced by dry grinding with an agate mortar. The fine particles were suspended in the sulfuric acid solution and dispersed by sonication for 1 min. The suspension was then allowed to settle for 2 min, and the supernatant was analyzed.

Adhesion of T. ferrooxidans to sulfide minerals. For all five minerals tested, T. ferrooxidans adhesion increased with the number of cells added, although there were significant differences in the affinity of the bacteria for the various minerals (Fig. 1A). By far the largest number of cells adhered to pyrite, followed in descending order by molybdenite, chalcocite, millerite, and galena. The relative adhesiveness of the cells for each mineral was assessed by comparing the numbers of adherent cells while taking into consideration the mineral surface areas (Fig. 1B and C). In each case, the mineral surface area estimated by the BET method was substantially larger than that determined by direct microscopy (Table 1). However, for each method, the number of cells adhering per unit of surface area of pyrite was substantially greater than the number of those adhering to other minerals over the entire range of added cells.

FIG. 1 Adhesion of T. ferrooxidans to selected minerals. (A) Numbers of cells adhering to 0.5 g of each mineral expressed without regard to mineral surface area. (B and C) Numbers of adherent cells normalized to mineral surface areas estimated by BET gas adsorption (more ...)

Sulfide mineral	Surface area (cm2/g)
	BET methoda	Microscopic methodb
Pyrite	889	140(±4.6)
Millerite	4,064	159(±13.8)
Galena	326	102(±2.9)
Molybdenite	5,900	212(±42.8)
Chalcocite	1,548	121(±4.2)

The differences in estimated surface area between the BET and microscopy methods were likely due to surface roughness that could not be detected by light microscopy. The surface roughness of each mineral was, therefore, assessed by scanning electron microscopy (data not shown). Pyrite, molybdenite, and galena particles had smooth surfaces. The surfaces of chalcocite and millerite particles were not smooth. However, not every mineral showed the surface roughness on a micrometer scale. The largest irregularities on the surfaces were gaps between molybdenite particle lamina due to their laminar structures, but gap sizes were still smaller than a micrometer. The size of a single T. ferrooxidans cell is, comparatively, very large (0.5 μm wide and 1.0 μm long). Consequently, the contribution made by surface roughness to the available area for bacterial adhesion should be negligible.

The apparent specificity of T. ferrooxidans adhesion suggests that the bacterium may selectively adhere to pyrite, even when pyrite is mixed with other minerals. Therefore, the adhesiveness of T. ferrooxidans was assessed with mineral mixtures composed of 0.2 g each of molybdenite, chalcocite, millerite, and galena and selected quantities of pyrite; these findings were compared to pyrite adhesion in the absence of other minerals. When T. ferrooxidans cells were added to pyrite alone, adhesion increased linearly with increases in added pyrite (Fig. 2); the number of adherent cells reached 2.75 × 10⁸ in the presence of 0.8 g of pyrite. In contrast, in the absence of pyrite, adherence of T. ferrooxidans to 0.8 g of mineral mixture was 3.7 times lower. The affinity of T. ferrooxidans for pyrite was particularly evident when pyrite was mixed with other minerals. As with pyrite alone, the number of adherent cells increased linearly with the weight of the added pyrite when 0.3, 0.5, or 0.8 g of pyrite was added in combination with 0.8 g of the mineral mixture (Fig. 2). It was evident from the results that T. ferrooxidans selectively adhered to pyrite within mineral mixtures.

FIG. 2 Selective adhesion of T. ferrooxidans to pyrite in mixtures of sulfide minerals. Symbols: ●, the number of cells adhering to a mineral mixture; , the number of cells adhering to pyrite alone. The mineral mixtures were prepared by blending (more ...)

The effects of T. ferrooxidans on the floatability of sulfide minerals. The effect of the addition of T. ferrooxidans on mineral floatabilities was investigated by exposing mineral mixtures to cell suspensions containing 6.5 × 10⁸ cells (Table 2). In the absence of bacteria, all of the minerals exhibited high floatability (90.8 to 99.0%). Upon exposure to the bacteria, the floatability of pyrite declined dramatically from 95.9 to 19.3%. Exposure to bacteria also affected the floatabilities of millerite and galena, which decreased from 95.5 to 83.8% and from 90.8 to 81.8%, respectively. However, those decreases were much smaller than the decrease in pyrite floatability. Thus, pyrite floatability was specifically suppressed by exposure to T. ferrooxidans.

Sulfide mineral	Floatability (%)
	Without bacteria	With bacteria
Pyrite	95.9(±1.4)	19.3(±4.9)
Chalcocite	99.0(±1.4)	98.8(±1.1)
Molybdenite	97.9(±0.3)	96.5(±3.4)
Millerite	95.5(±1.6)	83.8(±8.8)
Galena	90.8(±1.4)	81.8(±4.4)

Pyrite removal from mixtures of sulfide minerals. The specific ability of T. ferrooxidans to suppress pyrite floatability is potentially useful for selectively removing pyrite from mineral mixtures. We initially tested this possibility by blending pyrite (0.5 g) with equal quantities of molybdenite, chalcocite, millerite, or galena, exposing the mixtures to cell suspensions, and then subjecting them to flotation (Table 3). In the absence of T. ferrooxidans, >90.2% of the pyrite was recovered as froth; i.e., flotation by itself was very effective for separating pyrite from other minerals. Exposing of mineral mixture to T. ferrooxidans dramatically reduced the pyrite content of the froth: 83.7 to 95.1% of the pyrite in each mixture was found in the tailings. In contrast, 72.3 to 100% of the other minerals remained in the froth. Consequently, the purity of the other minerals increased from 50% to 84.3 (±3.0)% (millerite), 86.0 (±8.0)% (galena), 84.4 (±4.0)% (chalcocite), and 79.2 (±7.5)% (molybdenite).

Presence of cells	Location	Distribution of each sulfide mineral to froth and tailing (%)
		Pyrite + molybdenite		Pyrite + chalcocite		Pyrite + millerite		Pyrite + galena
		FeS2	MoS2	FeS2	Cu2S	FeS2	NiS	FeS2	PbS
Without cells	Froth	98.6(±2.5)	100.0(±0.0)	90.2(±4.1)	99.9(±0.3)	95.3(±1.4)	99.2(±0.4)	100.0(±0.0)	99.0(±1.1)
	Tailing	1.4(±2.5)	0.0(±0.0)	9.8(±4.1)	0.1(±0.3)	4.7(±1.4)	0.8(±0.4)	0.0(±0.0)	1.0(±1.1)
With cells	Froth	7.4(±6.3)	100.0(±0.0)	4.9(±3.9)	99.2(±0.6)	16.3(±1.8)	72.3(±8.9)	8.3(±7.3)	91.1(±8.5)
	Tailing	92.6(±6.3)	0.0(±0.0)	95.1(±3.9)	0.8(±0.6)	83.7(±1.8)	27.7(±8.9)	91.7(±7.3)	8.9(±8.5)

We next examined the capacity of T. ferrooxidans to selectively remove pyrite from mixtures containing the five minerals, including either large or small quantities of pyrite (Table 4). In the latter case, 84.5 to 98.8% of the minerals, including pyrite, was recovered from the froth in the absence of cells. Upon addition of T. ferrooxidans to the mixture, pyrite was selectively rejected from the mixture and was recovered from the tailings; 76.9% of the pyrite was removed from the mixture even though the initial pyrite content of the mineral mixture was low. Conversely, 75.1 to 100% of the other minerals were recovered as froth. With high-pyrite mixtures, in the presence of T. ferrooxidans, 94.3% of the pyrite was rejected from the mixture and recovered as tailing whereas 82.2 to 97.2% of the other minerals remained as froth. Separation by flotation with T. ferrooxidans increased the purity of the other minerals from 20.0% (added) to 89.6 (±11.0)% (froth). Thus, by addition of the bacterium, the pyrite content of flotation froth can be substantially reduced without diminishing mineral recovery, even when the pyrite content is high.

Presence of cells	Location	Distribution of each sulfide mineral to froth and tailing (%)
		Low pyrite content					High pyrite content
		FeS2	MoS2	Cu2S	NiS	PbS	FeS2	MoS2	Cu2S	NiS	PbS
Without cells	Froth	87.3(±3.6)	98.8(±1.1)	84.5(±6.1)	94.0(±4.9)	95.4(±1.8)	95.7(±1.3)	99.3(±0.2)	87.1(±2.4)	97.3(±1.5)	97.3(±1.0)
	Tailing	12.7(±3.6)	1.2(±1.1)	15.5(±6.1)	6.0(±4.9)	4.6(±1.8)	4.3(±1.3)	0.7(±0.2)	12.9(±2.4)	2.7(±1.5)	2.7(±1.0)
With cells	Froth	23.1(±6.6)	100.0(±0.0)	99.3(±0.3)	75.1(±0.7)	98.4(±1.7)	5.7(±1.7)	97.2(±2.2)	88.5(±10.1)	82.2(±9.6)	89.8(±7.5)
	Tailing	76.9(±6.6)	0.0(±0.0)	0.7(±0.3)	24.9(±0.7)	1.6(±1.7)	94.3(±1.7)	2.8(±2.2)	11.5(±10.1)	17.8(±9.6)	10.2(±7.5)

Mineral surfaces available for adhesion of T. ferrooxidans. We employed two methods to estimate the surface areas of the minerals used in this study. With the BET method, areas were determined as a function of the amount of nitrogen adsorbed on the mineral surfaces. This means that even if mineral surfaces contain convexities and concavities (roughness) as small as a few nanometers, the small size of nitrogen molecules allows for easy adsorbance to the entire surface. T. ferrooxidans cells are, by contrast, approximately 0.5 μm wide and 1.0 μm long and are much too large to adhere within concavities in particle surfaces. Consequently, our measurements of bacterial adhesion should have been unaffected by surface roughness. Indeed, the surfaces of the minerals used in this study were fairly smooth (data not shown), and it seemed reasonable to consider the surfaces available for adhesion of T. ferrooxidans as flat and without roughness. As a result, the surface areas estimated microscopically were probably closer to the actual areas available for bacterial adhesion than the estimates obtained by the BET method.

Selective adhesion of T. ferrooxidans to pyrite and its interactions. The selectivity of T. ferrooxidans adhesion to pyrite was first suggested when it was observed that the cells adhered to pyrite within coal particles (3, 8). The same phenomenon was observed in iron-rich areas of sulfide ores (9). Adhesion of T. ferrooxidans to pyrite has been compared with its adhesion to other minerals (13), but the selectivity was not definitively shown. In contrast, the present study clearly demonstrates that T. ferrooxidans will selectively adhere to pyrite, despite the presence of large quantities of one or more other minerals.

Of particular interest to us were the interactions mediating T. ferrooxidans’ selective adhesion. It is generally understood that bacterial adhesion is governed by physical (e.g., electrostatic and/or hydrophobic) interactions (5, 7, 18). Electrostatic interactions depend on the charges present on the surfaces of the cell and the particle. Therefore, the zeta potentials of the minerals used in this study were estimated (Table 5). All of the minerals had negatively charged surfaces, and the surfaces of T. ferrooxidans cells are also negatively charged (4). Thus, electrostatic interaction would not produce the essential force necessary for T. ferrooxidans adhesion; in fact, the negative charges would repel one another (6).

Sulfide mineral	Zeta potentiala (mV)	Contact angle (degrees)	ΔGadhb (erg/cm2)
Pyrite	−28.12(±9.96)	83.8(±1.3)	−5.58
Millerite	−6.33(±2.31)	81.6(±4.0)	−5.87
Galena	−7.85(±3.44)	80.6(±2.9)	−6.56
Molybdenite	−35.12(±8.20)	96.4(±1.9)	−8.11
Chalcocite	−7.45(±7.03)	88.5(±2.6)	−8.33

Adhesion driven by hydrophobic interactions can be modeled thermodynamically, taking into consideration the surface tensions of adherent cells, the solid surfaces, and the suspending liquid medium (1). The surface tension of each mineral used in this study could be calculated from contact angles determined experimentally as shown in Table 5. The angles were measured toward the sulfuric acid solution that served as the suspending liquid medium in the cell adhesion experiments and which has a surface tension of 71.4 dynes/cm. The angle of T. ferrooxidans cells to the same sulfuric acid solution was reported as 24.0° (13). The above-mentioned values for angles and liquid surface tension were incorporated into a thermodynamic model to determine the change of surface free energy (ΔG_adh) that occurs when T. ferrooxidans adheres to the minerals shown in Table 5 (1, 10, 11). In general, bacterial adhesion via hydrophobic interaction with the mineral can occur spontaneously when ΔG_adh is negative, and the adhesion is more extensive the more negative the value of ΔG_adh (5, 18). According to this model, cells are most likely to adhere to molybdenite, followed by chalcocite, pyrite, millerite, and galena in descending order. In contrast, cells adhered to pyrite, followed by molybdenite, chalcocite, millerite, and galena (Fig. 1). With the exception of pyrite, this theoretical determination approximates the actual tendency in the adhesion observed experimentally. The large number of cells which adhered to pyrite did not merely match the theoretical determination made on the basis of adhesion via hydrophobic interactions. The adhesion of T. ferrooxidans to pyrite involves specific interactions other than hydrophobic interactions.

Research conducted over the past several years has frequently focused on whether there is a specific mechanism involved when T. ferrooxidans adheres to pyrite. For instance, Rojas et al. suggested that an organic capsule covering the cell surface is relevant to pyrite adhesion (16, 17). Devacia et al. and Arredondo et al. both suggested that cell surface proteins play an important role in the adhesion of T. ferrooxidans to solid surfaces (2, 6). Very recently, an apo form of rusticyanin was isolated as a surface protein responsible for T. ferrooxidans adhesion to pyrite, which may provide the key to its selectivity (14).

Application of microbial flotation to mineral processing. When pyrite is present as a contaminant in mixtures of sulfide minerals, commercial flotation processors commonly use cyanide to suppress its floatability (21). Because the suppressive effect of cyanide on pyrite flotation is enhanced in solutions made alkaline by the presence of Ca(OH)₂, the separation efficiency of flotation in the presence of cyanide plus Ca(OH)₂ was tested with a pyrite-galena mixture (results not shown). The pH of the flotation liquor was adjusted to 10.0 with Ca(OH)₂, and then cyanide (KCN) was added to a concentration of 1 mM. Under these conditions, pyrite rejection and galena recovery were 76 and 68%, respectively, which are less than the separation efficiency of T. ferrooxidans, which produced corresponding values of 91 and 91% (results not shown).

There is a growing consensus that bacteria can significantly affect mineral floatability; the present study showed that T. ferrooxidans suppresses pyrite floatability. Conversely, bacterial cells may also be able to function as collectors. For instance, Smith et al. reported that addition of the hydrophobic bacterium Mycobacterium phlei increased the hydrophobicity of mineral surfaces and thereby improved floatability (19). Thus, it may soon be possible to replace conventional collectors, such as fatty acids or cationic amines, with bacteria. In the past, oxidative metal leaching by bacteria has contributed significantly to mineral processing. It now appears that it may be useful to expand the role played by bacteria in mineral processing to include adhesive control of mineral floatability.