Introduction

In kinetic ARD testwork, as in field behaviour, particle size, particle size distribution and individual mineral grain size are parameters that affect both acid generation and acid neutralization. For pyrite and pyrrhotite grains to oxidize, there must be exposure of these grains to the atmosphere. For neutralizing mineral grains to react with produced acid, such grains must be exposed to the acid. The degree to which individual grains are exposed depends to a great extent upon the size of the individual grains relative to the size of the rock sample in which they are contained.

It has been standard practice for many years, in most of the world, to use ISO Standards for screening analysis. These standards use square aperture size (in mm or Ám) for standard screening analysis, rather than the "mesh sizes" formerly used by various national organizations, and currently used in the United States under ASTM Standards. Current practice prefers or requires (depending upon regulatory authority) the use of SIU (SystŔme Internationale d'UnitÚs) units. "Mesh size" is dependent upon the weave of screen cloth - i.e wire diameter and spacing of the screen wires. Most mesh series were originally based on either the fourth root of the integer 2, or the square root of the integer 2, and on available wire size and weaving techniques. The ISO Standards maintain this basis, but use absolute dimensions to normalize individual national standards.

Unlike Acid Base Accounting procedures, where samples are finely ground (100% -75 Ám, -45 Ám or -38 Ám), samples for kinetic testwork may have significant size ranges. For example, for the ASTM humidity cell procedure the sample should be 100% passing 6.3 mm (1/4 inch), while column tests are commonly run on material less than about 25 mm and lysimeters may be operated with run-of-mine material that may be as coarse as 500 mm or greater.Although a specified mass may be used for a kinetic test (e.g. 1 kg for the ASTM humidity cell procedure), the sample specific surface area (usually expressed in square metres/kilogramme) depends upon the sample particle size distribution and the mineralogical characteristics of the sample.Because of these dependencies, two concepts from the mineral processing field must be introduced into ARD prediction:

  1. Particle size distribution

  2. Mineral liberation size (Liberation Size Determination and Surface Sulphur Determination Methods)

Particle Size Distribution

When rocks are broken (by blasting, crushing or grinding), particles are produced with a continuous size distribution from the coarsest particle in the product to the finest (which is theoretically a particle of zero size). The continuity of this particle size distribution may be broken when the fragmented material is subjected an imposed process such as screening or hydraulic classification - in which case any products of the process will have their own unique size distribution that will differ from the that of the original material.

Many methods, of varying complexity, have been developed to quantify the size distribution of particulates (Lynch, 1977, Bond, 1985, Lynch and Lees, 1985). One relatively uncomplicated method that has found favour in the North American metalliferous mining industry since 1940 is the Gates-Gaudin-Schuhmann equation, defined by:

y = [x/k]m

where y is the fraction of the sample finer than size x, and k (size modulus) and m (distribution modulus) are constants for the particular distribution. If the logarithm of the cumulative % weight passing size x is plotted against the logarithm of particle size x, a relatively straight line is often obtained, with a slope equal to m, the distribution modulus, and an intercept at 100% cumulative weight passing of k, the size modulus. The size distribution is therefore characterized by two simple parameters, k and m. For any given group of size distributions, the higher the value of k, the coarser the size distribution, and the higher the value of m, the narrower the size distribution (and vice-versa). For a fixed mass of sample and value of k, specific surface area increases as the width of the size distribution increases - i.e. as m decreases.

The following Figure 1 shows Gaudin-Schuhmann size distributions for 8 samples of crushed drill-core from the same deposit prepared for humidity cell testwork by the same method. All of the curves appear to converge at a 100% cumulative % weight passing size of about 16 mm or 5/8 inch, thus the value of k for all samples is 16 mm or 16,000 Ám. The slopes, or gradients, of the curves range from approximately 0.5 for Sample 5 to approximately 1.1 for Sample 4, although the curves for Samples 4, 6, 7 and cannot be described as a straight line. The 8 samples thus have a range of values for m, from 0.5 to 1.1 and Sample 5 has the widest, and Sample 4 the narrowest, size distributions. Sample 4 would be expected to have the least specific surface area, and Sample 5, the greatest.

Figure 1: Gaudin-Scuhmann Size Distributions for 8 Humidity Cell Samples (Data source confidential)

A characteristic that is of considerably more use than the simple particle size distributions in the interpretation of ARD testwork is an estimate of sample total or specific surface area. For the above 8 samples, these estimates have been made, assuming all particles to be spherical with a maximum particle size (k) of 16 mm and a minimum particle size of 2 Ám (chosen as being the likely finest sized particle retained in an operational humidity cell, and introducing a less than 1% error into calculations). Since testing screens are manufactured with apertures in series based on the fourth root of two, the geometric mean of the upper and lower screen size for each fraction should be used as a mean fraction particle size, rather than an arithmetic mean.The following Table 1 shows the calculated specific surface area (square metre per kilogramme) and the percentage of the total sample surface area contributed by the finest (-150+2 Ám) fraction of the sample for the 8 samples.

Table 1: Calculated Specific Surface and Percentage of Total Surface Area Contributed by -150+2Ám Fractions for Data in Figure 1

Sample Number

Calculated Specific Surface, square meter/kilogramme

Percentage of Total Surface Area Contributed by Minus 150 Plus 2 micrometre Fraction

1

12.0

93.2

2

9.77

90.8

3

4.51

86.4

4

1.28

79.1

5

12.5

92.0

6

2.75

86.9

7

4.42

89.6

8

3.86

82.7

The calculated specific surface area of the samples ranges from 1.28 (Sample 4) to 12.5 (Sample 5) square metre/kilogramme, which represents a range of almost an order of magnitude. In general the specific surface area values correlate well with the size distribution curves, in that the wider the size distribution, the greater the specific surface area.

A difference of an order of magnitude between the specific surface area of humidity cell samples, from the same location, and prepared in a similar manner is significant. To be of any validity, comparisons between the results of humidity cell testwork on these eight samples would need to be on a normalized surface area basis, and not on a mass basis.

The variation between the percentage of surface area contributed by the -150+2 Ám fractions is from 79.1% (Sample 4) to 93.2 (Sample 1), and this is also significant, representing an absolute difference of 14.1%.

The specific surface area of particles is proportional to the reciprocal of their size as shown in Figure 2 below for spherical particles. The points on the graph correspond to the geometric mean particle size for each size fraction in the above example - i.e. 14.1 mm, 7.7 mm, 3.1 mm, 1.3 mm, 652 Ám, 274 Ám and 17 Ám.

Figure 2: Specific Surface Area versus Size for Spheres

The range of specific surface area values for these mean particle sizes is from 0.16 to 129 square metre/kilogramme.

If fine size particle size distribution is critical to the evaluation of kinetic ARD tests, then a more rigorous size distribution determination procedure than the one used above should be employed.This should include dry screening at 105 Ám, 75 Ám and 44 or 37 Ám, wet screening at 44 or 37 Ám and Cyclosizing. A Warman Cyclosizer (a commonly used item in mineral processing laboratories, and shown below in Photograph 1) typically gives size splits for material with a Relative Density of 2.7 of 44 Ám, 29 Ám, 19 Ám, 14 Ám and 11 Ám.

Photograph 1: Warman Cyclosizer«

Further refinements to specific surface area determination can be made by determining the particle shape factor for each size fraction and applying this to the spherical particle approximation.

Mineral Liberation Size

The textural relationships between minerals within an ore, and their relation to process selection requires the introduction of the concept of liberation size. This is the size to which an ore must be crushed or ground to produce separate particles of either value mineral or gangue that can be removed from the ore (as concentrate or tailings) with an acceptable efficiency by a commercial unit process. Liberation size does not imply pure mineral species, but rather an economic trade-off of grade and recovery. The concept of liberation and liberation size may be extended to ARD prediction, where atmospheric leaching becomes the relevant unit process. Atmospheric pyrite and pyrrhotite oxidation requires that part of the mineral grain be exposed or liberated with respect to leaching. For any given pyritic or pyrrhotitic rock, the degree of pyrite or pyrrhotite mineral grain exposure increases with decreasing particle size, attaining 100% exposure or liberation with respect to leaching when all of the rock has been reduced in size to that of the smallest grain of pyrite or pyrrhotite. It is usually only at 100% exposure or liberation that acid generation potential determined by ABA (static) methods equals the acid generation potential of kinetic test or field samples.

Thus grains pyrite or pyrrhotite in whole rock will have a lesser degree of exposure (less liberation) than those in blasted waste rock, and those in blasted waste rock will have a lesser degree of exposure (less liberation) than those in mill tailings. Size reduction corresponds to an increase in total particle surface area. Particle specific surface area is proportional to the reciprocal of particle size.

Figure 3: Four Possible Scenarios for Pyrite or Pyrrhotite Liberation

Figure 3, above, shows four possible scenarios for pyrite, pyrrhotite or metallic sulphide mineral liberation from non-sulphide minerals such as silicates, oxides and carbonates. The particles are shown in section view as they would be seen when sawed through for macroscopic examination or microscopic viewing in transmitted or reflected light.

Both the first and fourth particles contain a sulphide mineral grain that is liberated with respect to atmospheric leaching. The second and third particles contain a grain or grains of a sulphide mineral that are not liberated with respect to atmospheric leaching.

It is quite possible for all four grains to have the same sulphide sulphur analysis, and thus the same AP, and to have the same NP. In this case, ABA (static) analysis would rate the four grains equally with respect to potential ARD generation. However, in a kinetic test only particles one and four would have the potential to generate acid, as they would in the field.

Liberation size can be estimated from petrographic studies, especially the microscopic examination of thin and polished sections.

Liberation size can also be determined by quantitative mineralogical testwork on size fractions of ground ore, although this is considerably less common because it is costly, time consuming, and often technically difficult. However for the determination of potential ARD generation problems with tailings, the methodology produces information that simple chemical analysis does not.

Liberation size is a function of the relevant physical or chemical process and may differ greatly between processes (Mills, 1995). Since ARD generation is a sulphide (pyrite) leaching process and base metal concentration is usually by flotation from gangue sulphide (pyrite) these differences may have significance in ARD prediction. The liberation size of pyrite for both flotation and leaching (and gravity concentration if it is included in the mill flowsheet) is therefore a most important parameter. This information is extremely difficult to determine without quantitative mineralogical studies.

Size reduction for liberation naturally produces a size range of particles. This range is dependent upon the type of size reduction method used, and the hardness, texture, friability and degree of weathering of the rock and its constituent minerals.

For ARD prediction the important characteristics of ore or waste are:

1. The amount of pyrite and/or pyrrhotite in the ore or waste.

2. The degree of pyrite and/or pyrrhotite liberation produced by size reduction with respect to potential oxidation and leaching.

3.The presence of, and association with pyrite of, trace metal sulphides.

4. The amounts of leachable metallic sulphide minerals in the ore or waste

5. The degree of liberation of leachable metallic sulphide mineral produced by size reduction with respect to leaching.

Most of the mineralogical testwork conducted in the field of ARD has focussed on the products of pyrite oxidation and the products of ARD neutralization within tailings impoundments. It is considerably easier to determine particle size distribution and liberation characteristics for tailings than it is for waste rock dumps. For existing tailings, size distribution may be representatively determined from testing screen results, while liberation characteristics may usually be obtained using visible light microscopic techniques. For existing waste rock dumps the acquisition of representative size distributions is very difficult since particle size ranges from the coarsest material produced by blasting to the finest material retained after rain and snow infiltration and flushing. Liberation characteristics for waste dump material may be (and usually are) inferred from examination of the finer dump material. Where only drill core is available, liberation characteristics can be determined by microscopy techniques, but estimated size distributions require information on the proposed mining method and milling flowsheet. Metallurgical testwork is usually conducted concurrently with ARD testwork, so that tailings from testwork programs are normally available for ARD study.

It is particularly important that metallurgical and ARD testwork be integrated because a well-designed mill will not only optimize metallurgical performance, but also minimize potential ARD problems. A good recent example of this integration is demonstrated by the design of the proposed Tulsequah Chief mine in British Columbia as reported by Rescan Environmental Services Ltd. (1997). In the proposed Tulsequah Chief mill, pyrite will be concentrated by froth flotation following base metal sulphide concentration and recovery. The pyrite concentrate will be deposited sub-aqueously in the underground mine. The mill tailings, containing some residual pyrite and potentially acid-generating, will be blended with ground limestone for sub-aqueous deposition in the tailings impoundment for the operational life of the mine. At mine closure the tailings impoundment will be allowed to drain and the tailings will progress to sub-aerial storage. Mill design with respect to mineralogy and potential ARD generation was examined for the Huckleberry mine in British Columbia by Mills (1995).

The effect of particle size upon some ARD static procedure measured values for waste rock dumps at 4 sites in British Columbia has been reported by Price and Kwong (1997) . In all cases only minus 100 mm material was examined and samples were taken only from the top 1 m of the dump. Sized fractions, -100+19 mm, -19+11 mm, -11+2 mm, -2 mm+50 Ám and -50 Ám, were subjected to paste pH, carbonate NP, Sobek NP and sulphide sulphur determinations. Although the results show variations in the values of these measured quantities with particle size, the authors concluded only that:

"The data provided illustrates: differences in the composition of different particle size fractions, the inadequacy of a whole sample assay as a means of characterizing weathering, and the importance of separately analysing a fine particle size fraction (< 2 mm) when evaluating weathering effects"

The static procedures used in this testwork all require that a sample be pulverized, so that liberation characteristics of acid-generating minerals and neutralizing minerals could not be observed. In a waste rock pile a size fraction of -2 mm would be expected to have a relatively high degree of liberation of both acid-generating and neutralizing minerals when compared to the coarser fractions. As a result, it would be expected that in-situ acid generation and neutralization had occurred with concurrent reduction in carbonate NP, Sobek NP and sulphide sulphur (AP) when compared to the coarser fractions. Other causes of differences between size fractions may be mineralogical in nature - for example the friability of pyrite relative to calcite can produce more fine pyrite particles of a given size than calcite particles of the same size after size reduction by impact.

Price and Kwong (1997) note that the -2 mm fraction of their samples represented 10 to 30% of the sample mass. Price (1997) notes that most laboratory (i.e. static) analyses do not distinguish between liberated and unliberated mineral species, and recommends that, in the absence of site-specific data (liberation size studies) that indicate otherwise, the -2 mm fraction of waste rock should be considered the reactive fraction in ARD prediction studies. He gives the following potentially negative effects of such an approach:

  • Sieving is required, and additional handling could alter the quality of the sample.
  • A large enough sub-sample of the -2 mm fraction may be unavailable, and too small a sample size may not yield representative data.
  • The analysis will not measure the contribution from the coarse fragments, which may be significant if coarse fragments are incompetent or porous, or the -2 mm fraction is unreactive.
  • The analysis assumes that most contaminant release comes from this size fraction. This assumption may not be correct for historic mine wastes and for naturally weathered materials in which weathering has progressively removed reactive minerals from the finer particles.

In western Canada, a widely used approximation for the reactive portion of waste rock dumps is 10 to 20%, based on observations of the magnitude of the -2 mm weight fraction of waste rock dumps. Price and Kwong (1997) reported a range of 10 to 30% -2 mm for their - 100 mm waste dump samples, and Murray (1977) has given an average of 15 to 25% with a range of 0 to 35% for waste rock at Canadian precious and base metal mines.

The above approximations and practices notwithstanding, ARD programs should include liberation studies in all but the most straightforward cases, and particularly in cases where predictions from static and kinetic testwork are inconsistent or contradictory.

Liberation Size Determination Methods

Low-power stereoscopic microscopy and high power petrographic microscopy (particularly using dark-field illumination) are usually the most productive tools. It is often preferable to size (screen) and concentrate samples prior to examination. The concentration process will depend upon the type and complexity of sample, but the aim is to identify the distribution and degree of liberation of pyrite, pyrrhotite, metallic sulphide minerals, and their mode of occurrence. Heavy liquids such as 1,1,2,2-tetrabromoethane, tri-iodomethane and (less commonly) Clerici solution (Mills, 1978, 1985) have been used to concentrate the more dense fractions of mineral bearing samples, giving a range of Relative Densities up to about 5 (hot Clerici solution). Because they can be used in centrifuges, heavy liquids may be used with particles as fine as 1-5 mm, and therefore offer great flexibility. This flexibility is offset by cost and toxicity considerations. For particles coarser than about 200 mm, the Magstream 50 (Domenico et al., 1994) offers some advantages for density separation. This unit is simple to operate, has high sample throughput, uses non-toxic chemicals, and can operate at a continuous range of Relative Densities comparable to heavy liquids.

Generally, any concentration of samples would be to aid microscopy, and not to isolate or separate pure mineral samples.

A most useful guide to the physical and chemical identification of mineral in granular form has been written by Jones and Fleming (1965).

Surface Sulphur Determination

Another method for the determination of exposed or liberated sulphide grains in particulate samples was developed by Acme Analytical Laboratories Ltd. of Vancouver in collaboration with Bruce Downing in 1996. This method, which has not seen widespread use and is currently under further development, utilizes a weak acid leach and leachate analysis for sulphur to determine the soluble sulphide of size fractions of crushed or ground rock. This "Surface Sulphur" procedure is intended to measure the proportion of contained sulphide minerals that are exposed at granular surfaces, and are thus liberated with respect to atmospheric leaching. Any sulphidic grains occluded in the rock particles, while contributing to sample total sulphur, will not be reported by the Surface Sulphur procedure and a value for liberated sulphides as a percentage of total sulphides may therefore be determined. In relation to ARD prediction, the procedure generates a percentage liberation value for the oxidizable sulphides in each size fraction of a fragmented sample.

This procedure appears to have considerable potential use in ARD prediction. However, for proprietary reasons we are unable to publish it on these pages.

Summary

It is not unusual for kinetic ARD testwork to produce results that are inconsistent with ABA results. This can usually be attributed to the fact that ABA tests measure total sulphides, whereas kinetic tests are dependent upon exposed or liberated sulphides. As a consequence, kinetic tests may produce results that indicate that a sample is not net acid generating when ABA analysis has indicated that the sample is likely to be net acid-generating. This situation arises most often when sulphides such as pyrite and pyrrhotite occur as inclusions finer than 5 Ám in size as well as in the form of much coarser grains. In such conditions, the kinetic test has more validity than the ABA test, since the coarser material of a waste rock pile will also contain much unliberated sulphide. It is also possible for the situation to be reversed, in that some neutralization potential measured in ABA tests may not be realized in kinetic tests or in the field because the neutralizing minerals are not adequately liberated. In such cases a sulphide and/or neutralizing mineral liberation study would generally clarify inconsistencies.

The generation and neutralization of acid in waste rock piles, tailings and exposed rock and ore in underground mines is surface area, not a mass, dependent. The example given above, shows clearly that 8 samples, from the same deposit and prepared in the same way for humidity cell tests, can vary by an order of magnitude in their specific surface area. In this particular case, metal leaching rates based on mass, rather than surface area, obtained from the humidity cell tests should not be directly compared with one another. Only when the metal leaching rates are normalized with respect to surface area can comparisons be accepted as valid.

 

ęThe contents of this web page are protected by copyright law. Please contact the authors for permission to re-use the contained information.


References

American Society for Testing and Materials (ASTM) (1996), Standard Test Method for Accelerated Weathering of Solid Materials Using a Modified Humidity Cell, Designation D5744-96, ASTM, Conshohocken, PA, 13p.

Bond, F.C. (1985), Testing and Calculations, in Weiss, N.L., Ed., SME Mineral Processing Handbook, American Institute of Mining, Metallurgical and Petroleum Engineers, New York, 3A: 16-27.

Domenico, J.A., Stouffer, N.J. and Faye, C. (1994), Magstream as a Heavy Liquid Separation Alternative for Mineral Sands Exploration, Society of Mining Engineers, Albuquerque meeting, March, 14p.

Jones, M.P. and Fleming, M.G. (1965), Identification of Mineral Grains, Elsevier, London, 102p.

Lynch, A.J. (1977), Mineral Crushing and Grinding Circuits - Their Simulation, Optimization, Design and Control, Developments in Mineral Processing Volume 1, Elsevier, Amsterdam, 18-25.

Lynch, A.J. and Lees, M.J. (1985), Simulation and Modeling, in Weiss, N.L., Ed., SME Mineral Processing Handbook, American Institute of Mining, Metallurgical and Petroleum Engineers, New York, 3A:28-53.

Mills, C. (1995), Technical Review of the Acid Rock Drainage (ARD) and Metal Leaching Aspects of the Metallurgical Testwork, Milling Practices and Tailings Monitoring for the Huckleberry Project, report to B.C. Ministry of Energy, Mines and Petroleum Resources, October, 34p.

Mills, C. (1985), Specific Gravity Fractionation and Testing with Heavy Liquids, in SME Mineral Processing Handbook, American Institute of Mining, Metallurgical and Petroleum Engineers, New York, 30:44-52.

Mills, C. (1978), Mineralogy and Heavy Liquid Analysis in Gravity Separation, Short Course on Gravity Concentration Technology, University of Nevada, Reno, October, 25p.

Murray, D.R. (1977), Pit Slope Manual Supplement 10-1, CANMET Report 77-31, Department of Energy, Mines and Resources Canada, Ottawa, Ontario.

Price, W.A. (1997), DRAFT Guidelines and Recommended Methods for the Prediction of Metal Leaching and Acid Rock Drainage at Minesites in British Columbia, British Columbia Ministry of Employment and Investment, Energy and Minerals Division, Smithers, BC, (April), 143p.

Price, W.A. and Kwong, Y.T.J. (1997), Waste Rock Weathering, Sampling and Analysis: Observations from the British Columbia Ministry of Employment and Investment Database, Proceedings 4th International Conference on Acid Rock Drainage, Vancouver, p31-45.

Rescan Environmental Services Ltd. (1997), Tulsequah Chief Project Report, Volumes II and IV, Vancouver, BC.


Return to ARD at Enviromine