The liming effect of bauxite processing residue (red mud) on sandy soils.
Refining of bauxite ore for aluminium via the Bayer process produces red mud as a residue. Approximately 0.5-2.5 t of red mud is produced per t of alumina produced. The amount of red mud produced depends on the quality of the bauxite and the processing conditions. In Western Australia, 4 alumina refineries produce more than 14 Mt of bauxite residue each year, much of which consists of red mud (Summers 1994), the remainder being sand. Red mud has a fine particle size (mostly <150 [micro]m in diam.), contains large amounts of iron oxides, and thus has a very high sorption capacity for phosphorus. Due to residual caustic soda (used in the Bayer extraction process), the red mud has a very high pH (>11) and thus disposal is problematic. One possibility for disposal of this waste is as a liming agent onto acidic agricultural land.
Whittaker et al. (1955) were the first to investigate the liming potential of 'brown mud' in the USA. This mud had a CaC[O.sub.3] equivalent of 75%, which is much higher than for most red muds. Simons (1984) investigated red mud for liming acid soils in Surinam and found that pH and yields increased by the addition of red mud but that plant yields were less than for CaC[O.sub.3] limed soils. Application of red mud to soils in Western Australia has been investigated, and the increase in pH has been shown to increase pasture yield on acid sandy soils (Summers et al. 1996). Other Australian work has evaluated seawater neutralised red mud for ameliorating acid sulfate soils (Ward et al. 2002).
This paper reports on an investigation of chemical changes due to the application of several red muds of different CaC[O.sub.3] equivalence to 2 acid sandy soils from Western Australia.
Materials and methods
Two soils were sampled from 20-80 mm depth (i.e. plant debris and litter were removed) from long-term fertilised pastures on the Swan Coastal Plain, south of Perth, Western Australia. The soils were from the Bassendean Association and consisted of pasture litter and thin humus horizon on a sandy topsoil overlying deep grey sand (McArthur 1991). The soil is classified as a Podosol (Isbell 1996). These soils are commonly saturated to a shallow depth for much of the year and flooded in winter. The soils were air-dried and sieved to pass a 2-mm mesh.
Red mud from the 6 Australian alumina refineries and 5 refineries in other countries was air-dried, crushed, and sieved to <2 mm. The material had been air-dried after most of the free liquor had been drained out. A carbonated sample and bitterns concentrate-treated sample were also supplied by Alcoa (Alcoa World Alumina, Australia). An Alcoa red mud/gypsum (10%) sample was collected from a stockpile that had been exposed to the weather for ~4 years (identified as aged red mud/gypsum). A subsample of the Alcoa Kwinana refinery red mud was mixed with 10% w/w gypsum (red mud/gypsum). Subsamples of Alcoa Pinjarra refinery red mud were ameliorated to reduce alkalinity by (1) the addition of gypsum at 5% and thoroughly leached to remove water-soluble sodium sulfate (leached red mud/gypsum); and (2) thorough leaching with dilute acid to remove most alkaline salts (leached red mud). These materials were air-dried and crushed to pass a 2-mm mesh. Table 1 lists the sample identifications and origins of the red muds. A sample of desilication product (DSP) was provided by Alcoa.
The soils were analysed as random powders on a Philips PW3020 X-ray diffractometer (XRD) fitted with a graphite monochromator using Cu K[alpha] radiation at 1[degrees]/min scan speed. The following properties were determined using methods from Rayment and Higginson (1992) except where indicated: bicarbonate-extractable P (Colwell 1963); phosphorus retention index (Allen and Jeffery 1990); pH and electrical conductivity (EC) in a 1:5 soil:water ratio (soil pH was also in 1 M NaF); Fe and Al extracted by dithionite-citrate-bicarbonate (Mehra and Jackson 1960), ammonium oxalate, and sodium pyrophosphate; total carbon and nitrogen (LECO); Walkley-Black organic carbon (Nelson and Sommers 1982); particle size distribution (Gee and Bauder 1986); P fractionation (Olsen and Sommers 1982); cation exchange capacity using 0.01 M [(AgTU).sup.+]; titratable acidity; and elemental analysis by X-ray fluorescence (XRF) spectrometry.
The red muds were analysed for oxalate- and DCB-extractable Fe and Al, pH, EC, elemental analysis, and by random powder XRD by the same methods used for the soils. Deionised water and 0.5 M HCI in several 5-mL increments were added to 20-g samples of red muds, DSP, and CaC[O.sub.3] to bring solution additions to a total of 100 mL. Suspension pH was measured after 4 h, 1 day, 3 days, 7 days, 14 days, and 28 days of shaking. After the measurement at 28 days, the suspensions were centrifuged to separate solution from solid and the solid was air-dried. Samples at a range of final pH values were then analysed by XRD. Acid neutralisation capacity (mol [H.sup.+]/kg solid) was calculated from these titration curves to pH 7 after 28 days.
Samples (20 g) of the soils were thoroughly mixed with all the red muds, and separately with laboratory-grade CaC[O.sub.3], Ca[(OH).sub.2], and CaO at rates equivalent to 0, 2.5, 5, 10, 20, 40, 80, and 120 g/kg soil (~t/ha) and wetted to about 35% water content. NaOH was added to subsamples of soils in several amounts up to 375 mmol NaOH/kg soil. Replicates of the samples treated with Pinjarra red mud, the red mud/gypsum, and CaC[O.sub.3] had toluene added to depress microbial activity. The tops were firmly screwed onto the vials, which were placed in the dark at 20[degrees]C for 16 weeks. Subsamples were removed for analysis after 1 week, 4 weeks, and 16 weeks and were analysed for EC and pH (1:5 [H.sub.2]O).
Results and discussion
Soils and red muds
The 2 soils used were chosen on the basis of their contrasting exchange acidity and organic C (Table 2). The red mud samples varied in many respects and ranged in texture from clay to sandy loam, indicating that full separation of the sand content had not occurred for some samples. The term 'red mud' will be employed although some materials were more probably the entire Bayer residue. Red mud pH values ranged from 8.4 to 12.6 and acid neutralisation capacity to pH 7 ranged from 0.45 to 1.6 mol [H.sup.+]/kg solid. The only systematic relationship between red mud characteristics was a positive relationship between EC and [Na.sub.2]O (XRF) content ([R.sup.2] = 0.62). XRD showed that muscovite, hematite/ maghemite, goethite, quartz, calcite, anatase, DSP (as identified here has a zeolite type structure and resembles sodalite and cancrinite; Thornber and Hughes 1992), and gibbsite were the main crystalline minerals present in the red muds.
Every red mud produced a different buffering curve when added to soil. A representative selection of the buffering curves is shown in Fig. 1. Data for CaC[O.sub.3] are also plotted on this graph, which shows a much steeper initial slope for CaC[O.sub.3] than for the red muds and a CaC[O.sub.3] buffering curve of much simpler shape. This difference in buffering behaviour in the soil reflects the diverse acid buffering (titration) curves for red muds and CaC[O.sub.3], examples of which are shown in Fig. 2. Red mud has a more complex buffering curve than CaC[O.sub.3].
[FIGURE 1-2 OMITTED]
When buffering curves of the 2 soils for the same red mud are compared, the shapes of the curves are similar. Data for Pinjarra refinery red mud are shown in Figs 3, 4, 5, and 6, as this red mud is representative of the red mud applied to these soils in the field. Soil 1 has a smaller buffering capacity than soil 2 and thus reached a higher pH than soil 2 for each rate of red mud addition (Fig. 3). Buffering curves for some red muds converged at high rates of applied red mud, which reflects the similar intrinsic properties of the red muds rather than an effect of soil buffering.
[FIGURE 3-6 OMITTED]
A near-equilibrium pH value was generally reached after 4 weeks of incubation, although in some instances a difference in pH of up to 1 unit existed for different incubation times (e.g. soil 2, Pinjarra refinery red mud Fig. 4b). Addition of the microbial inhibitor toluene to the incubation had no effect on the buffering curves, so microbial action possibly does not play a role in the liming effect of red mud for these soils.
There was a significant difference in initial EC between the 2 soils and a linear relationship between rate of addition of red mud and EC (Fig. 5). When the different red muds were compared, there were different slopes, mostly due to the different EC of the red muds (Table 3). Thus the relationships shown in Fig. 5 are parallel because both soils received the same red mud.
As red mud is proposed as an alternative to lime it is important to determine the lime equivalence of red mud, and because of the complex shape of the soil-red mud buffer curves the lime equivalence will vary with red mud/lime application rate and target pH. Lime equivalence was calculated for each red mud for several target pH values at week 16 and for both soils. Figure 7 shows that the lime equivalence differed greatly between red muds, was always <100%, and decreased to very small values for higher target pH values. At target pH 6, which is a common target pH in practical agriculture, the lime equivalence of the red muds ranged from 11 to 42% for soil 1 and 13 to 50% for soil 2. The lime equivalence is linearly related to [Na.sub.2]O percentage in red mud ([R.sup.2] = 0.83 for soil 1 and 0.76 for soil 2) but is not closely related to any other red mud characteristic. The [Na.sub.2]O content may largely reflect the DSP (sodalite) content of the red mud and DSP is likely to provide a contribution to the liming effect of red mud. XRD analysis of red mud samples at different pH values showed that DSP and calcite buffer simultaneously and buffering at pH 6-8 is due to dissolution of both DSP and calcite. Samples with only small amounts of calcite and larger amounts of DSP still had long plateaus in this region of the buffering curve.
[FIGURE 7 OMITTED]
Most of the chemical reactions between red mud and soil apparently occurred in the first week of incubation, which is at least partly due to the soil and red mud being well mixed and maintained in a wet condition. The red mud was ground to <2 mm, which would also enable the red mud to react quickly with the soil. This mixing is difficult to attain in the field, especially on fragile sandy soils that have a tendency to erode if disturbed by tillage.
The complex nature of the soil/red mud buffering curves indicates that complex buffering reactions occur, in marked contrast with the simple shape of the soil/lime buffering curves (Fig. 1). Differences in the mineralogy of the red muds are a major cause of differences in buffering curves between red muds. Difference in the buffering curves of the 2 soils for the same red mud are probably due to the large difference in organic matter content of the 2 soils and associated differences in alkali buffering, as organic matter provides most of the surface area that retains [Al.sup.3+] and [H.sup.+] ions in these soils (McArthur 1991). As mentioned above, complete mixing on these soils is problematic and a liming agent that is fine and dissolves rapidly would be desirable. Unfortunately, these laboratory measurements cannot take account of the effect of minimal mixing in the field.
The buffering curves for CaC[O.sub.3], CaO, and Ca[(OH).sub.2] addition are much simpler curves than for red muds. The differences between these curves may partly reflect the conversion of CaO and Ca[(OH).sub.2] to CaC[O.sub.3] as part of the dissolution process in the soil/lime mixtures as data for all 3 types of lime gradually approach the same curve (Fig. 1b). Two minerals that are common constituents of red mud and appear to be the main liming agents are DSP and calcite (CaC[O.sub.3]). DSP appears to buffer at a slightly lower pH than CaC[O.sub.3]. From the titration curves shown in Fig. 2, CaC[O.sub.3] has a lower starting pH but strongly buffers at pH ~7.5. The DSP sample used to generate these data came from a Bayer reaction vessel and was not pure DSP--it contained minor amounts of calcite, hematite, goethite, and gibbsite and possibly a little free caustic. Reactive alkaline forms including free [OH.sup.-] and adsorbed caustic will react with added acid in the early (high pH) part of the curve before pH is buffered by a relatively small amount of CaC[O.sub.3] and subsequently by DSP at about pH 6.5-7. The buffering curve for Spanish red mud appears to reach a plateau at about pH 7.5, possibly due to calcite buffering.
The abundance of these 2 mineral buffering agents (CaC[O.sub.3], DSP) was reduced considerably when red mud was pre-treated with acid, resulting in much less complex soil/red mud buffering curves (Fig. 1, LRM line). Most of the free and adsorbed alkali had been neutralised and washed out of the LRM, so the initial slope of this buffering curve is not as steep as for the unleached red muds. The initial slopes of the buffering curves (Fig. 1a) are thus probably due to reactions of free and adsorbed alkali with buffering at lower pH values being due to calcite, DSP, and small amounts of other alkaline compounds.
The CRM sample possibly had a higher proportion of [Na.sub.2]C[O.sub.3] than the other red muds as it was treated with gaseous C[O.sub.2] to reduce the NaOH content. NaC[O.sub.3] would be expected to buffer at a higher pH than CaC[O.sub.3] and DSP, which explains why the CRM buffering curve is much steeper and does not reach a plateau below pH 8 as occurs for other red muds.
The lime equivalence of the red muds varied considerably due at least in part to differences in contents of calcite and DSP. Other forms of alkalinity that may be present include free [OH.sup.-], adsorbed/entrained alkalis, and calcium aluminate. The DSP group minerals have cage-like structures that can contain inter alia [OH.sup.-], Al[(OH).sup.-.sub.4] , and C[[O.sub.3].sup.2-] (Thornber and Hughes 1992), which would participate in buffering reactions. The lime equivalence of the red muds reacted with soil to reach pH 6; a target pH that would normally be applied to these soils is 11-42% (soil 1) and 13-50% (soil 2). From this it is seen that the amount of red mud required to increase soil pH is generally greater than the amount of lime required, so that the economic benefits of using red mud may be limited. However, the neutralising value of lime sand and limestone can vary between 50 and 95% of calcium carbonate (McLay and Porter 1996). Red mud has very few costs associated with its preparation and is provided free from the producer. The alumina refineries are on rail lines that extend into the Western Australian wheatbelt and rail transport can considerably reduce the cost of transport to a fifth of that of road transport (R. Summers, pers. comm.). Combining all of these factors into a simple economic analysis (Porter and Sandison 1996), red mud may be an economical alternative to lime, as has been found in field experiments on sandy soils in high rainfall areas by Summers et al. (1996) and Summers (2001). This economic analysis does not take into account other benefits of using red mud to lime soils, such as not depleting the natural lime resource, less effect on roads where road transport distances for red mud are less than for lime, and increased income due to reduced P leaching and water repellence for sandy soils.
The effect of CaC[O.sub.3] additions on soluble salt content as assessed by EC was much smaller than for most of the red muds. Some red muds caused a large increase in soil EC at 20 t/ha of red mud, to values that might affect growth for sensitive plant species (Hunt and Gilkes 1992). This increase in EC appears to be mostly associated with the high Na content that is variously present as dissolved, exchangeable, and soluble Na forms. In the field this increase in EC would possibly be only a short-term effect, as excess Na salts are rapidly leached from these well-drained sandy soils during the high rainfall winter (Ward 1983; Vlahos et al. 1989).
It is not possible to predict the final pH of soil treated with red mud on the basis of intrinsic properties of soil and red mud. Instead, the titration (buffering) curves of the 2 materials must be determined and compared (Wong et al. 1998). Soil buffering curves are determined using additions of NaOH solution. By plotting the soil and red mud buffer curves on the same graph on the basis of proportions that would be present in the particular soil/red mud mixture, the final pH can be predicted. An example is shown in Fig. 6 for 40 t/ha Pinjarra red mud applied to both soils. The intercept of the soil/alkali and red mud/acid curves indicates the final pH obtained for a 40 t/ha application of Pinjarra red mud. Figure 8 shows the relationship between predicted and measured pH for several rates of application for 17 red muds applied to both soils. The agreement between measured and predicted pH is satisfactory around pH 6-7, which approximates the usual target pH range for practical liming. For higher pH values (higher red mud rates) the predicted values are less than the measured values, and conversely, predicted pH values are higher than measured values at lower pH (lower red mud rates).
Addition of red mud to acid sandy soils will increase the pH of the soils. However, as the lime equivalence of red mud is generally <30%, the amount of red mud required to increase soil pH is higher than the amount of lime required, so that the economic benefits of using red mud may be limited. Further field work is needed to assess the impact of red mud on acid soils in low rainfall areas, where the rapid action of some of the liming material in red mud and the small particle size would be an advantage especially for ameliorating subsoil acidity. Another important consideration is that use of red mud as a liming agent removes the ongoing cost of managing red mud in impoundments at refineries, so that there is a benefit to the community in promoting the use of red mud to lime soils.
Table 1. Origins of red mud samples Sample Sample origin Kwinana Alcoa World Alumina Australia Kwinana refinery, W. Aust. Pinjarra Alcoa World Alumina Australia Pinjarra refinery, W. Aust. Wagerup Alcoa World Alumina Australia Wagerup refinery, W. Aust. Worsley Worsley Australia Pty Ltd, W. Aust. Nabalco Nabalco, Northern Territory (now Alcan Gove) QAL Queensland Alumina Ltd, Qld Brazil Alunorte, Brazil Germany Aluminimn Oxid Stade GmbH, Germany Italy Eurallumina, Italy Spain Alcoa Spain USA Reynolds Metal Co., USA CRM Carbonated red mud Kwinana, W. Aust. SWC Bitterns concentrate treated red mud Kwinana, W. Aust. RMG Kwinana red mud + 10% w/w gypsum ARMG Kwinana red mud + 10% w/w gypsum exposed to weather ~4 years LRMG Pinjarra red mud + 5% w/w gypsum with leaching to remove salts LRM Pinjarra red mud leached with dilute acid Table 2. Characteristics of the soils used in the liming experiment PH EC Exch. CEC Particle size (%) (1:5 [H. (1:5 [H. acidity ([cmol. sub.2]O) sub.2]O) (cmol [H. sup.c]/ Sand Silt (mS/cm) sup.+]/kg) kg) Soil 1 4.55 0.26 0.45 8.9 94 4 Soil 2 4.29 0.11 1.16 12.1 94 3 Particle size (%) Walkley-Black Clay org. C (%) Soil 1 2 4.9 Soil 2 3 8.6 Table 3. Characteristics of red muds used in the liming experiment Sample pH EC BET SSA ANC (A) CaO [Na.sub. (1:5 [H. (1:5 [H. ([m.sup (mol [H. (%) 2]O sub.2]O) sub.2]O) .2]/g) sup.+]/kg (XRF) (%) (mS/cm) solid) (XRF) Kwinana 11.5 3.0 27 0.75 5.3 3.4 Pinjarra 11.6 6.1 25 1.08 4.4 4.2 Wagerup 12.0 2.7 24 0.68 3.6 3.2 Worsley 12.6 6.3 25 0.88 2.4 2.2 Nabalco 12.4 10.8 29 1.64 2.3 7.1 QAL 10.2 8.2 29 0.92 2.5 8.6 Brazil 12.2 3.3 15 1.25 1.2 7.5 Germany 12.1 2.6 23 0.61 5.2 4.0 Italy 9.8 18.2 25 0.94 4.2 11.7 Spain 12.6 2.3 21 0.86 5.5 3.6 USA 11.0 3.6 27 0.77 7.7 6.1 CRM 10.9 6.3 19 1.05 5.8 4.4 SWC 9.9 6.7 27 1.30 4.9 4.3 RMG 11.4 8.5 23 1.04 6.9 4.3 ARMG 10.7 0.72 24 0.52 4.9 2.1 LRMG 9.2 2.4 26 0.88 6.2 2.6 LRM 8.4 0.67 30 0.45 2.3 0.9 (A) Acid neutralising capacity, calculated from buffering curves to pH 7.0 (mol [H.sup.+]/kg solid).
We thank staff at Alcoa World Alumina Australia, Worsley, QAL, Nabalco, Reynolds Metal Co., Alunorte. Aluminium Oxid Stade GmbH, Alcoa Spain, and Eurallumina for information and supplying the red muds, John Kargotich for access to his farm to collect the soils, and Rob Summers for stimulating debate.
Allen DG, Jeffery RC (1990) Methods for analysis of phosphorus in Western Australian soils. Report of Investigation No. 37, Chemistry Centre of Western Australia, East Perth.
Colwell JD (1963) The estimation of the phosphorus fertilizer requirements of wheat in southern New South Wales by soil analysis. Australian Journal of Experimental Agriculture and Animal Husbandry 3, 190-197.
Gee GW, Bauder JW (1986) Particle size analysis. In 'Methods of soil analysis. Part 1. Physical and mineralogical methods'. 2nd edn (Ed. A Klute) pp. 383-411. (ASA/SSSA: Madison. WI)
Hunt N, Gilkes R (1992) 'Farm monitoring handbook.' (University of Western Australia/Land Management Society: Perth, W. Aust.)
Isbell RF (1996) 'The Australian Soil Classification.' (CSIRO Publishing: Melbourne)
McArthur WM (1991) Reference soils of south-western Australia. Australian Soil Science Society Inc., WA Branch/Department of Agriculture, Western Australia.
McLay C, Porter W (1996) Lime. In 'Soil acidity: a reference manual'. Ch. 7. (Eds L Leonard, M Bolland) Miscellaneous Publication 1/96. (Agriculture Western Australia: Perth, W. Aust.)
Mehra OP, Jackson ML (1960) Iron oxide removal from soils and clay by a dithionite-citrate system buffered with sodium bicarbonate. Clays and Clay Minerals 7, 317-327.
Nelson DW, Sommers LE (1982) Total carbon, organic carbon and organic matter. In 'Methods of soil analysis. Part 2. Chemical and microbiological properties'. 2nd edn (Eds AL Page, RH Miller, DR Keeney) pp. 539-580. (ASA/SSSA: Madison, WI)
Olsen SR, Sommers LE (1982) Phosphorus. In 'Methods of soil analysis. Part 2. Chemical and microbiological properties'. 2nd edn (Eds AL Page, RH Miller, DR Keeney) pp. 403-430. (ASA/SSSA: Madison, WI)
Porter W, Sandison A (1996) Economics. In 'Soil acidity: a reference manual.' Ch 9. (Eds L Leonard, M Bolland) Miscellaneous Publication 1/96. (Agriculture Western Australia: Perth, W. Aust.)
Rayment GE, Higginson FR (1992) 'Australian laboratory handbook of soil and water chemical methods.' (Inkata Press: Melbourne)
Simons AP (1984) Red mud (bauxite residue) as a potential liming alternative for acid soils in Surinam: Crop yield improvement. De Surinaamse Landbouw 32, 100-113.
Summers RN (1994) Red mud--cutting pollution and boosting yields. Western Australian Journal of Agriculture 35, 55-59.
Summers RN, Guise NR, Smirk DD, Summers KJ (1996) Bauxite residue (red mud) improves pasture growth on sandy soils in W.A. Australian Journal of Soil Research 34, 569-581.
Summers RN (2001) The use of red mud residue from alumina refining to reduce phosphorus leaching and increase yield potential on sandy soils. PhD thesis, University of Western Australia.
Thornber MR, Hughes CA (1992) The alkalinity of residues from Alcoa of Australia Limited's refineries of south-west Australia. In 'International Bauxite Tailings Workshop'. Perth, W. Aust. pp. 136-147. (Executive Management Services: Perth, W. Aust.)
Vlahos S, Summers KJ, Bell DT, Gilkes RJ (1989) Reducing phosphorus leaching from sandy soils with red mud bauxite processing residues. Australian Journal of Soil Research 27, 651-662.
Ward NJ, Sullivan LA, Bush RT (2002) Sulfide oxidation and acidification of acid sulfate soil materials treated with CaC[O.sub.3] and seawater-neutralised bauxite refinery residue. Australian Journal of Soil Research 40, 1057-1067.
Ward SC (1983) Growth and fertiliser requirements of annual legumes on a sandy soil amended with fine residue from bauxite refining. Reclamation and Revegetation Research 2, 177-190.
Whittaker CW, Armiger WH, Chichilo PP, Hoffman WM (1955) Brown Mud" from the aluminum industry as a soil liming material. Soil Science Society of America Proceedings 19, 288-292.
Wang MTF, Nortcliff S, Swift RS (1998) Method for determining the acid ameliorating capacity of plant residue compost, urban waste compost, farmyard manure, and peat applied to tropical soils. Communications in Soil Science and Plant Analysis 29, 2927-2937.
K. E. Snars (A), R. J. Gilkes (A,C), and M. T. F. Wong (A,B)
(A) School of Earth and Geographical Sciences (Soil Science Discipline), Faculty of Natural and Agricultural Sciences, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia.
(B) CSIRO Land and Water, Underwood Avenue, Floreat, WA 6913, Australia.
(C) Corresponding author; email: firstname.lastname@example.org
Manuscript received 5 February 2003, accepted 22 December 2003
|Printer friendly Cite/link Email Feedback|
|Author:||Snars, K.E.; Gilkes, R.J.; Wong, M.T.F.|
|Publication:||Australian Journal of Soil Research|
|Date:||May 1, 2004|
|Previous Article:||A field study of phosphorus mobilisation from commercial fertilisers.|
|Next Article:||The nature and distribution of copper, lead, and zinc in soils of a highly urbanised sub-catchment (Iron Cove) of Port Jackson, Sydney.|