Aqueous polydisperse suspensions of angular (milled) or spherical (atomised) ferrosilicon alloy powders are widely used in mineral, aggregate and scrap separation processes as a dense medium. Table I shows the typical range of Dense Media Separation water analyses (analysing the supernatant after allowing the ferrosilicon solids to settle) taken from three diamondiferous-kimberlite circuits, from which it is evident that ferrosilicon powders are immersed in a near-neutral to alkaline pH electrolyte.

Surface Chemistry of Atomised Particles

At a given temperature and pressure, the predominant surface phase(s) and solution species in an electrochemical system may be uniquely defined by the solution pH and electrochemical potential (EH). For example, in the form of a potential-pH diagram derived from free energy of formacion (ΔGf°) values.

The diagram is substantially the same for other Fex Si compositions and predicts, for example, that at low potentials preferential oxidation of the silicon component in the alloy will occur by reactions such as,

Considering the alloy electrode, the first two current peaks (I, II) of the anodic-going potential sweep on the first (n=1) potential cycle correspond to iron oxidation to ferrous species and surface oxidation Fe(II)- to Fe(III) -oxide, respectively. On subsequent potential cycles (n>1) peak I disappeared, but peak II, and its corresponding reduction peak III, remained. Independent measurements on peaks II and III indicated that a reversible one electron/one proton Fe(II)/Fe(III) surface transformation was occuring.

Particle Microelectrophoresis

The nature of the particle/electrolyte interphase must now be considered. Such information can be accessed only via electrokinetic techniques, such as electrophoresis.

 

As the pH was increased beyond pzr1, the mobility became increasingly negative until around pH 7.0, where it changed abruptly. This charge reversal was observed over a narrow range 7.0 < pH < 7.6, before going through pzr3, and becoming increasingly negative. Above pH 9.0 the mobilities decreased less rapidly with pH and in some cases became slightly less negative (due to ionic strength effects), thus prohibiting quantitative interpretation in these zones at the extremities of the pH range.

In the microelectrophoresis experiments, the ferrosilicon particles were almost certainly in a non-equilibrium state, i.e. they were corroding albeit at a low specific rate. Indeed, some of the particles were seen to change direction during the measurements, presumably due to dissolut ion-precipitation/adsorption of iron species, though in the unbuffered solutions, local pH changes could also have contributed to the effect.

For the larger monovalent FeOH ions, this energy is significantly less than that for Fe²+ ions (Hunter, 1981, probably more than compensating for the FeOH+ ion activity being two orders or magnitude lower in the bulk solution at pH 7.

In addition, as the solutions used in the electrophoresis experiments were unbuffered (because many buffer anions have significant complexing power and hence a propensity to specifically adsorb), the local pH at the surface of the corroding particles is likely to have been signficantly higher than in the bulk solution, as protons were being consumed in the hydrogen evolution (and oxygen reduction) reaction driving the ferrosilicon anodic oxidation process(es).

Surface Analysis

A VG Scientific (ESCALAB II) X-ray photoelectron spectrometer, with Al K-∝ radiation (1486.6 eV), was used to investigate the surface composition of “Cyclone 60” powder (Samancor Ltd) and compared with a metalIographically- polished 15.8 % Si alloy (reference) sample (Samancor Ltd.). Sequential XPS and Ar+ bombardment was used to obtain a compositional depth profile through the surface of the powder and alloy specimen.

Interpretation of the Isoelectric Point

The results of XPS and microelectrophoretic experiments confirm the presence of a pre-existing (thermally) oxidised particle surface, due to the atomisation procedures, whose properties depend upon the potential and pH of the aqueous media. Unfortunately, details regarding the spatial distribution of these oxides on dry powders are now known – although preferential silicon oxidation produces surface accumulation of silica, it is not clear whether the accompanying iron oxides exist as discrete island sites, or are admixed with silica.

In general terms, the total charge density (σ) comprises contribution from each of the i types of hydroxylated site present in the surface at fractional atomic concentration Xi, hence

where Ki is a constant.

These examples also serve to emphasise the necessity of obtaining surface rather than bulk compositional analyses when dealing with surface phenomena, and the sensitivity of electrophoretic mobility measurements to examine such effects.

Rheological Measurements at ‘Constant’ Zeta-Potential

Examination of dimensionless groups shows that the zeta potential will have no effect on the rheological properties of a clean suspension of atomised particles in an aqueous electrolyte if substantial shear (say, r > 30 s-¹) is experienced. Under such circumstances the applied shear forces far exceed any electrostatic forces or related interactions of the coarse (1-100µm) polydisperse suspension of dense (7000 kgm-³) solid particles.

 

Sedimentation Measurements at ‘Constant’ Zeta Potential

Even for DMS units operating under relatively quiescent conditions, such as bath or cone separators, the gravitational force still exceeds any diffusional forces associated with the translational kinetic energy of all except the very small (<2µm) particles. However, under certain conditions electrical double layer forces may influence the behaviour of a clean sedimenting suspension, particularly for powders containing a large proportion of fine particles.

For many colloidal-size systems also increases with increasing zeta-potential, since mutual electrostatic repulsion allows a sedimenting particle more time to find and occupy an optimum-sized space in the sediment bed, thus minimising the bed porosity. An examination of the morphology of the setted bed suggested that these arguments apply for ferrosilicon, however the most marked effect was the extent of differential settling within the sediment, even at high solids fraction (∅ > 0.3). This is contrary to the commonly-held belief that all concentrated suspensions settle in a truly hindered fashion (i.e. all particles sediment at the same rate independent of their size) .

Use of Indicator Electrodes

Various attempts have been made to assess the corrosion rate of freshly manufactured ferrosilicon particles (usually the milled variety). A better indication of the actual electrochemical state of a DM suspension is to be gained by measuring the electrochemical potential in-situ.

The rate of ferrosilicon corrosion is largely determined by the hydrodynamics of the DMS circuit. For example, normally the oxidation of iron (eqn.3) will be driven via oxygen reduction (O2 + 2H2O + 4e = 4OH-), or in the absence of oxygen, by reduction of water (2H2O + 2e = H2(g) + 2OH-). This is congruent with commonly observed phenomena of hydrogen gas evolution, particularly from stagnant DM sumps used to store the medium during plant shut-down, under which conditions differential aeration will further enhance the corrosion rate.

There were several significant features associated with the ageing process, viz:

1. The supernatant potential decreased with time – this may reflect a deletion in the oxygen content (initially 2×10 -6 mol dm-³) and changes in the iron(II)/iron(III) concentration, due to corrosion processes.
2. The potential of the indicator electrode in contact with the sediment was sensitive to the condition of the powder. After 3 days the potential decreaess by at least 500 mV, into the region of active corrosion (Figure 1). Hydrogen gas bubbles were visible on gently tapping the glass vessel containing the sediment/supernatant.
3. The potential at the bottom of the sediment was consistently lower than that near the surface, thus reflecting the additional resistance to the mass transport-controlled processes occurring at the solid-solution interface. Hence the depletion in oxygen content with depth creates a differential aeration cell i.e. the current due to oxygen reduction in the (passivated) upper part of the suspension is fed into anodic dissolution/oxidation in the lower region.
4. After resuspending the actively corroding powder with fresh electrolyte and allowing it to sediment, the measured potential increased only a little, but sufficiently to passivate (Figure 1) at pH 9.2.
5. After 13 days the sediment bed was greatly expanded due to profuse evolution of hydrogen gas.
6. A temperature differential of +0.5°C existed between the sediment and the supernatant (the cooler component).

Observations on Operating Plants: The in- circuit behaviour of DM is complicated by the unavoidable presence of mineral/gangue slimes.

Measurements were made on samples taken from the circuit, which were allowed to sediment, except for Plant C where the in-situ slurry potential, pH and tempeature were measured. The medium taken from Plant A was relatively free of slimes contamination, and after storage exhibited vigorous hydrogen evolution and a temperature difference between the sediment and overlying supernatant. Similar results were observed for material taken from Plant B.

Erosion/Corrosion and Suspension Stability in a Contaminated Medium

EDAX analysis confirmed that most of the features appearing as “small particles adhering to the surface” were composed of alloy constituents and were not of kimberlitic origin, thereby implying that they were either regions of oxide growth, uneroded/corroded zones resilient to degradation, or small fragments of ferrosilicon cemented to the surface of larger atomised particles. The latter two options seem the most likely explanation.

The prediction of suspension stability and viscosity in a diamond DMS plant presents a formidable task, and must take into account many of the features described above in addition to operational considerations (flowrates, medium losses, demagnetisation etc.).