Association of trace elements and dissolution rates of soil iron oxides

Additional keywords: acid dissolution, batch method, Fe oxide concentrate, non-stirred flow through reactor method, total element concentration, tropical red soils.

Iron (Fe) oxides are the most abundant metallic oxides in soils; they are present in various mineral forms depending upon present or past pedogenic conditions (Schwertmann and Taylor 1989). The distribution of Fe oxides in soils is also varied, ranging from a uniform distribution throughout the soil matrix to concentrated forms in segregations such as concretions and mottles (Schwertmann 1991; Singh and Gilkes 1992; Wiriyakitnateekul et al. 2007). Because of the variability of the chemical environment in soil during the precipitation of Fe oxides, and their ability to incorporate other trace elements into their structures, Fe oxides are an important source and sink of environmentally significant trace elements. Incorporation of several elements such as cadmium (Cd), cobalt (Co), chromium (Cr), copper (Cu), manganese (Mn), nickel (Ni), titanium (Ti), and zinc (Zn) occurs in the structure of synthetic goethites (Fitzpatrick et al. 1978; Lim-Nunez and Gilkes 1987; Singh et al. 2002; Cornell and Schwertmann 2003; Huynh et al. 2003; Kaur et al. 2009a, 2009b), which lends further support to the association of various trace elements with Fe oxides. Trace elements are commonly associated with Fe oxides in soils; however, it is not known whether these elements occur within a separate phase or substitute for Fe³⁺ in the structure of Fe oxides (Singh and Gilkes 1992; Trolard et al. 1995). The substitution of aluminium (Al) in soil goethite and hematite has been well established; however, the association of trace elements has received only limited attention (Fontes and Weed 1991; da Motta and Kämpf 1992; Singh and Gilkes 1992; Trakoonyingcharoen et al. 2006; Wiriyakitnateekul et al. 2007). Singh and Gilkes (1992) suggested that some portions of the Co, Cr, Cu, Mn, Ni and Zn, associated with Fe oxides in soils from south-western Australia, may be present within the structure of the two commonly occurring Fe oxides—goethite and hematite. A close association of trace elements, including Mn, Ni, Cr and vanadium (V), with Fe oxides exists in Thai soils (Wiriyakitnateekul et al. 2007). The nature of the association between trace elements and Fe oxides in soils has not been established for several other environmentally important trace elements.

Characterisation of the dissolution behaviour of natural Fe oxides containing several trace elements is important for predicting the availability of trace elements in soil environments. Dissolution studies of synthetic goethites have shown that factors such as the nature of the substituent metal, the extent of metal substitution, and the structural and morphological properties of minerals have a major control on their dissolution rate (Schwertmann 1991). Individual substitution of Al, Cr and Co into goethite structure decreases the dissolution rate of goethite in HCl (Schwertmann 1984; Lim-Nunez and Gilkes 1987; Ruan and Gilkes 1995). By contrast, the incorporation of Mn into goethite accelerated the dissolution rate compared with pure goethite (Lim-Nunez and Gilkes 1987; Schwertmann 1991; Alvarez et al. 2007). More recently, Kaur et al. (2010) found that the dissolution rate of single-metal-substituted goethite was in the order: Zn- > Pb(II)- ≥ Pb(IV)- > unsubstituted > Cd- > Cr-goethite. They observed complex dissolution behaviour for multi-metal-substituted goethite that was linearly related to steric strains determined from the lattice parameters of the mineral.

The objectives of this study were (i) to evaluate the association of several major and trace elements with Fe oxides in soils from Thailand using HCl dissolution kinetics and total element analysis, and (ii) to study the dissolution rate of soil Fe-oxide concentrates using a batch dissolution method and at the steady-state using flow-through reactors in highly acid solutions.

Soils from the B horizon of nine Oxisols and five Ultisols from Thailand were used for the study. The soils have developed on basalt, limestone, sandstone or alluvium. Soil descriptions and characteristics, site locations and land use are summarised in Table 1. The clay fraction of the soils was separated by repeated dispersion and sedimentation. Iron-oxide concentrates were obtained by boiling the clay fractions with 5 m NaOH to remove kaolinite and gibbsite (Singh and Gilkes 1991a).

Table 1.  Soil classification, parent material, horizon, depth, available P and land use of the soil samples used for extracting the Fe-oxide concentrates in this study

Random powder X-ray diffraction (XRD) patterns of Fe-oxide concentrates were obtained by scanning from 3 to 80° 2θ with a step size of 0.02° at 0.2° min^–1 using Cu Kα radiation, λ = 1.5418 Å on a diffractometer (MMA, 35 kV and 28.5 mA; GBC Scientific Equipment Pty Ltd, Hampshire, IL, USA). The proportion of goethite to hematite (Goe/(Goe + Hem)) of Fe-oxide concentrate was estimated by a comparison of peak area of goethite (110) and hematite (104) in XRD patterns. Aluminium substitution was determined from the c-axis dimension of the goethite unit cell obtained from d(110) and d(111) and calculated using the relationship: mol% Al = 1730 – 572.0(c₀) (Schulze 1984). For hematite, Al substitution was determined from the a-axis dimension of the hematite unit cell obtained from d(110) and calculated using the relationship: mol% Al = 3109 – 617.1a₀ (Schwertmann et al. 1979). Mean crystallite dimension (MCD) of goethite and hematite were calculated from the width of reflections at half-maximum using the Scherrer formula (Klug and Alexander 1974).

Synthetic goethite used in this study was synthesised by rapidly adding 5 m KOH to 1 m Fe(NO₃)₃ solution while stirring, and ageing the precipitate for 2 weeks at 40°C (Schwertmann and Cornell 2000). This synthetic goethite was found by XRD to be pure goethite.

The Fe-oxide concentrates and synthetic goethite were digested in concentrated HCl at 80°C until all Fe oxides dissolved; anatase and rutile were still present in the residues of the digests. The concentration of Fe, Al, Ca, Cu, Cr, Mg, Mn, Ni, lead (Pb), phosphorus (P), silicon (Si), V, Ti and Zn in the acid digests was determined using an inductively coupled plasma-atomic emission spectrometer (ICP-AES, Varian Vista AX CCD; Agilent Technologies, Santa Clara, CA, USA).

To determine the nature of the association of trace elements with Fe oxides, dissolution kinetics measurements were obtained by a batch dissolution method in acid solution. Briefly, 250 mg of Fe-oxide concentrate was added with 200 mL of 1 m HCl in a 250-mL plastic bottle, and the samples were shaken on a horizontal shaker (300 rpm) at 20 ± 1°C. After 1, 2, 3, 5, 7, 10, 20, 34, 48, 72, 96, 120, 144, 168 and 192 h, 5 mL of the suspension was withdrawn using an automatic pipette. The suspensions were immediately filtered through a 0.22-µm Millipore filter. The filtered solutions were analysed for Fe, Al, Ca, Co, Cu, Cr, Mg, Mn, Ni, Pb, P, Si, V, Ti and Zn using an ICP-AES.

Equations to describe the dissolution kinetics of Fe-oxide concentrates were cube-root law (Hixson and Crowell 1931), Avrami-Erofejev equation (Cornell and Giovanoli 1993) and Kabai equation (Kabai 1973). The cube root law is:

where W₀ and W are the weight of sample at t = 0 and after time t, respectively, and k is a rate constant. The Avrami-Erofejev equation is:

where α is the fraction Fe dissolved at time t. The Kabai equation is:

where C₀ and C are the amounts of Fe dissolved at t = 0 and at time t, respectively, and a is a phase-specific constant. A plot of ln[ln C₀/(C₀ – C)] against lnt results in a straight line from which k and a can be derived.

Non-stirred flow-through reactors were used to evaluate the dissolution rate at the steady-state of a soil Fe oxide (sample Kk) and the synthetic goethite. The specific surface area (SSA) was measured by the BET N₂ method (Aylmore et al. 1970), with values of 129 and 32 m² g^–1 for sample Kk and synthetic goethite, respectively. The flow-through reactor (46 mL in volume) used in this study has been described previously by Bibi et al. (2011). We used 100 mg Fe oxide for the dissolution experiment; the input solution was 0.01 m NaCl, which was prepared by using AR-grade NaCl, and solution pH was adjusted to the required value (1.0, 2.0 and 3.0) by adding AR-grade HCl. The flow rate of the input solution was 1.6 ± 0.2 mL h^–1, which was maintained by a Gilson peristaltic pump. The reactors were immersed in a constant-temperature water bath at 25 ± 1°C. Output solution was collected after every 12 h, the solution pH was measured immediately using a calibrated pH meter (PHM 210 pH meter; Radiometer Analytical SAS, Villeurbanne, France), and the solution was stored at 5°C for Fe determination using ICP-AES (detection limit 0.01 mg L^–1). The experiment was continued until the steady-state was reached, as indicated when the Fe in the output solution remained approximately constant (the difference of at least seven consecutive samples was <6%) (Bibi et al. 2011).

The dissolution rate (R) of Fe-oxide concentrates and synthetic goethite was calculated following equation below (Cervini-Silva and Sposito 2002; Cheah et al. 2003):

where [Fe] is the concentration of Fe in the output solution (µm) under steady-state conditions, q is the solution flow rate (L h^–1), and m is the sample mass in the reactor (kg). The resulting units of R are then µmol kg^–1 h^–1.

The general reaction (protonation) between protons and goethite can be written as:

The speciation calculations were used to examine the extent of equilibration of the reaction in terms of the Gibbs free energy (ΔG) by using Eqn 6, which was modified from Bibi et al. (2011), but the resulting units of ΔG are kJ mol^–1:

where SI is the saturation index of goethite and hematite, calculated using PHREEQC Version 2.18 (Parkhurst and Appelo 1999). The concentration of Fe (see Table 6) in the output solution at the steady-state was used in modelling with PHREEQC to calculate the SI value for the synthetic goethite and Kk samples.

The crystal morphology of both the synthetic goethite and the Kk sample was examined using transmission electron microscopy (TEM) before and after termination of the dissolution at the steady-state. A very dilute suspension of the samples was dispersed by an ultrasonic probe. A drop of suspension was deposited onto a carbon-coated grid and dried at room temperature. Micrographs were obtained using a JEOL 1400 electron microscope (JEOL Ltd, Tokyo).

The mineralogical properties of Fe oxide concentrates are shown in Table 2. Hematite and goethite were the main constituents of Fe oxide concentrates from these Thai red soils (Trakoonyingcharoen et al. 2006; Wiriyakitnateekul et al. 2007). Maghemite, anatase and rutile were present in small amounts in most of the Fe oxide concentrates in the study. The Goe/(Goe + Hem) ratio ranged from 0.12 to 0.74. Most samples had a higher content of hematite than goethite, except for Ti1, Ti2 and Ak1, which had similar amounts of the two minerals. Sample Kk was the only sample where the goethite content exceeded the hematite content. Aluminium substitution in goethite ranged from 10 to 20 mol%, compared with 3–11 mol% Al in hematite in the samples. The substitution of Al in goethite was always greater than the corresponding value for hematite in the same sample. Similar observations were made in earlier studies on highly weathered Brazilian, Australian and Indonesian soils (Fontes and Weed 1991; da Motta and Kämpf 1992; Singh and Gilkes 1992; Prasetyo and Gilkes 1994). The MCD from XRD line-broadening indicated that goethite crystals had a smaller size than associated hematite, as is generally observed for tropical soils (Fontes and Weed 1991; Singh and Gilkes 1992; Prasetyo and Gilkes 1994; Muggler et al. 2001; Trakoonyingcharoen et al. 2006; Wiriyakitnateekul et al. 2007). Aluminium substitution in Fe oxides increases structural strain, slows the rate of crystal growth and reduces crystal size (Schwertmann 1988). Goethite from Ultisols showed similar crystal development along the three crystallographic axes (MCD₁₁₀ : MCD₁₁₁ close to 1 : 1). Higher values of MCD along [110] than [104] for hematite indicate better crystal development along X–Y axes compared with the Z-axis; this is indicative of the platy nature of hematite crystals (Fontes and Weed 1991).

Table 2.  Mineral composition and some properties of goethite and hematite in Fe-oxide concentrates used in this study
Hem, Hematite; Goe, goethite; Mhm, maghemite; Ant, anatase; Rut, rutile; MCD, mean crystallite dimension

Total element concentrations in the Fe-oxide concentrates and synthetic goethite are presented in Table 3; however, some Ti did not dissolve and persisted as Ti oxides in residues. The concentration of Fe in the concentrates from soils ranged from 389 to 672 g kg^–1, compared with 632 g kg^–1 in the synthetic goethite. Samples Ci, Kk and Sd had lower Fe concentrations than the other samples, possible due to relatively large amounts of other minerals (Table 2) and the presence of residual alumino-silicate amorphous phase. Synthetic goethite had negligible to very low concentrations of other elements except Si and Ca, which resulted from the use of glassware in the synthesis. Aluminium and Ti were the major impurities in all Fe-oxide concentrates. Aluminium can be present as a structural substituent for Fe³⁺ in goethite and hematite, the two major Fe oxides in the Fe-oxide concentrates (Trakoonyingcharoen et al. 2006; Wiriyakitnateekul et al. 2007). With the exception of two samples (Ci and Ak3), Fe-oxide concentrates with high content of goethite also contained more Al (Tables 2 and 3), which is consistent with the greater Al substitution in goethite than hematite when they coexist. Aluminium in Ci and Ak3 samples originated from alumino-silcates and was not associated with Fe oxides. Titanium was mostly present in rutile and anatase, which did not fully dissolve in acid; however, some of the Ti can be in Fe oxides (Table 2). Other elements (Mn, Si, Ca, Mg, P, Cu, Zn, Ni, Cr, V, Pb and As) were commonly present in relatively smaller concentrations, with values ranging from below detection to 11 278 mg kg^–1. Manganese was the most abundant among the other elements, with concentrations of 2396–11 278 mg kg^–1, whereas the concentrations of Cu, Zn, Pb and As were generally low with values <250 mg kg^–1 in the Fe-oxide concentrates. The high concentrations of Al, Si, Ca and Mg in some of the Fe oxide concentrates (e.g. sample Kk) are possibly due to an amorphous phase, composed mostly of Al and Si, precipitated during the dissolution of clay minerals while obtaining Fe-oxide concentrates; this phase could not be detected by XRD (Singh and Gilkes 1992). Samples Ti1 and Ti2 contained higher amounts of P than the other samples, which is reflected by their relatively high available-P values (Table 1). The Fe-oxide concentrates from soils derived from limestone and shale/limestone had relatively higher concentrations of V and As than the concentrates of soils derived from other parent materials. Darunsontaya et al. (2010) also reported high concentrations of V and As in soil clays derived from limestone; As was particularly high in the clay fraction of limestone-derived soils, with values ranging from 21 to 221 mg kg^–1, values much larger than for soil clays derived from basalt, sedimentary rock and granite.

Table 3.  Total element analysis of the soil Fe oxide concentrates and a synthetic goethite used in the study
n.d., Not detectable

Principal component analysis of the total element concentration of Fe-oxide concentrates was used to identify elements and samples that have similar geochemical behaviour. Two factors could explain only ~50% of the variation in the data (Fig. 1a, b). These results are similar to those of Wiriyakitnateekul et al. (2007), in which first two factors explained only ~51% of the compositional variations for total element concentrations of Fe-oxide concentrates from Thai soils. They reported that most of the elements measured had a similar chemical behaviour and that Fe oxides in soils derived from basalt had quite different chemical composition from Fe oxides from other soils. However, for this study, the elements could be allocated into four groups, with Zn not belonging to any group (Fig. 1a). The first group included Fe, Cr, V, Pb, As and Ca. These elements had similar geochemical behaviour without systematic differences among soil parent materials. Nevertheless, sample Pc2 differed from other samples due to the high content of Fe-group elements (Fig. 1b). A second group included Si, Mg and Cu, and a third group included Al and Mn caused Kk to differ from other samples. The last group consisted of Ti, P and Ni; these elements were most abundant in samples Ti1 and Ti2, which formed a discrete group, with elevated P being due to fertiliser use.

Fig. 1.  Factor analysis for total element concentration of the Fe oxide concentrates from Thai red soils derived from different parent materials: (a) elements in Fe oxide concentrates, (b) soil samples.

Singh and Gilkes (1992) investigated the dissolution kinetics of Fe and trace elements in Fe-oxide concentrates, and based on the shape of the relationship between the proportions of Fe and metal dissolved, they suggested different mechanisms for the associated trace elements with Fe oxides in soils. They proposed that if an element was uniformly distributed throughout Fe-oxide crystals, Fe and the element will dissolve congruently, with a linear relationship and slope equal to 1 in plots of percentage metal v. percentage Fe dissolved, as indicated by line (iv) in Fig. 2a. If the element is present as a discrete phase that is very soluble, then 100% of the metal will dissolve before the dissolution of Fe commenced (line (i)). An initial rapid dissolution of a proportion (50%) of the element followed by congruent dissolution with Fe is indicated by line (ii). The converse scenario, where no dissolution of the element commences until 50% of Fe has dissolved, followed by congruent dissolution, is indicated by line (vi). The dissolution behaviour with trend lines similar to lines (ii) and (vi) are likely to indicate that at least 50% of the metal was not associated with Fe oxides. Intermediate behaviours for the association of elements and Fe are shown by the convex curve (iii) and concave curve (v).

Fig. 2.  Dissolution of trace elements plotted against Fe dissolved in 1 m HCl at 20 ± 1°C at various time intervals for the soil Fe-oxide concentrates: (a) theoretical dissolution behaviours; (b–o) observed dissolution for individual trace elements.

The dissolution curves of Al, Cr and V (Fig. 2b, c, d, respectively) showed that for Fe-oxide concentrates, dissolution of these elements and Fe followed a pattern quite similar to the schematic dissolution line (iv) (congruent dissolution) with a few exceptions. For samples Ci and Ak3, Al dissolved faster than Fe (60–80% of Al released), similar to the schematic dissolution lines (i) and (ii), whereas for sample Cpg, Cr dissolved faster than Fe, similar to the schematic dissolution curve (iii). These results suggested that Al (in samples Ci and Ak3) and Cr (in sample Cpg) were present in a discrete phase or were adsorbed onto the surface of Fe-oxide particles, which was consistent with the total Al concentration that was mentioned before. The congruent dissolution of Al, Cr and V with Fe for the majority of samples suggests that these elements are in the structure of goethite and hematite. These metals are known to substitute for Fe³⁺ in the structures of goethite and hematite (Trolard et al. 1995; Manceau et al. 2000; Wells et al. 2001; Singh et al. 2002; Perrier et al. 2006). Aluminium substitution in hematite, the main Fe oxide, in sample Ci (Table 2) was very low but total Al was much higher than in the other samples (Table 3), which suggest that most of the Al in sample Ci was not present in the structure of Fe oxides.

Dissolution behaviour of Mn, Ni, Co, Pb and Si (Fig. 2e, f, g, h, i, respectively) were similar to the schematic dissolution curve (iii) and similar to the results reported for Mn, Ni and Co by Singh and Gilkes (1992). These results indicate that some proportion of Mn, Ni, Co, Pb and Si was present within the structure of Fe oxides. The initial greater release of these elements relative to Fe also suggests that some fraction of the element was present in a discrete solid or soluble phase or adsorbed on to Fe oxides. There are several reports indicating that these elements can be adsorbed by Fe oxides and/or incorporated into Fe-oxide structures (Quin et al. 1988; Angove et al. 1999; Christophi and Axe 2000; Manceau et al. 2000; Wells et al. 2006; Kaur et al. 2009a; Marcussen et al. 2009). The dissolution pattern for Ti was varied (Fig. 2j), showing similarity to schematic dissolution line (ii) and curve (iii). With only a small fraction of the total Ti in Fe-oxide concentrates dissolved during the experiment, we suspect that only a small proportion of Ti was associated with Fe oxides. Most of the Ti that was not dissolved existed in anatase and rutile. The initial faster release of Ti than Fe suggests that a small fraction of Ti was possibly present in nano-crystalline forms of Ti minerals or Ti adsorbed onto Fe oxides.

The dissolution behaviour of Mg, Cu, Zn and Ca (Fig. 2k, l, m, n, respectively) showed an intermediate pattern between lines (i) and (ii) and curve (iii). These results imply that at least 20% of Mg, Zn and Ca and 10% of Cu was not directly associated with Fe oxides. These elements were released during the Fe-oxide concentration process, by oxidation of organic carbon and dissolution of phyllosilicates, and then re-adsorbed onto Fe oxides. Phosphorus did not show any systematic relationship with Fe in the dissolution data (Fig. 2o). However, the greater release of P than Fe during the initial stage of dissolution indicates that some of the P was adsorbed on Fe-oxide surfaces. There are several reports indicating that Fe and Al oxides play a major role in controlling the availability of P in soils through the adsorption–desorption processes (Singh and Gilkes 1991b; Börling et al. 2001; Agbenin 2003; Trakoonyingcharoen et al. 2005; Wiriyakitnateekul et al. 2005; Wisawapipat et al. 2009).

The extent of Fe dissolution plotted against time is shown in Fig. 3. None of the samples was completely dissolved in 1 m HCl at 20°C within the 192-h dissolution period used in the study. The proportion of Fe dissolved at 192 h ranged from 0.07 to 0.42, except for Ti1 and Ti2, with values of 0.59 and 0.62, respectively. The dissolution data for the 14 Fe-oxide concentrates fitted very well to the equations for three models employed (Table 4). The dissolution rate constant (10³k) ranged from 0.44 to 6.11 h^–1 for the cube root law, from 1.01 to 4.40 h^–1 for the Avrami-Erofejev equation and from 0.03 to 4.12 h^–1, for the Kabai equation. These values are lower than values of 10³k obtained at 40°C (1 m HCl) for 10 Fe-oxide concentrates from south-western Australia, with values ranging from 10.52 to 35.48 h^–1 for the cube root law and from 7.56 to 32.16 h^–1 for the Kabai equation (Singh and Gilkes 1992). The dissolution rate of goethite and hematite has been known to increase with increasing dissolution temperature, SSA and acid concentration (Chiarizia and Horwitz 1991; Cornell and Giovanoli 1993; Ruan and Gilkes 1995). The values of rate constants are also lower than values (64.8 to 3386.4 h^–1) reported for the dissolution of six synthetic hematite in 2 m HCl at 65°C obtained using the Avrami-Erofejev equation (Cornell and Giovanoli 1993). Furthermore, values of 10³k in this study are similar to values for synthetic goethite obtained by the first-order rate equation (ln(1 – α) = –kt) at 25°C with 2 m HCl (7.2 h^–1), but lower than at 80°C with 1 m HCl (720 h^–1) and 2 m HCl (4680 h^–1) (Chiarizia and Horwitz 1991). Similarly, the dissolution rate constants observed in our study are comparable to the values obtained for goethite dissolution (3.96 × 10³k h^–1) and Al-goethite dissolution (3.42−3.96 × 10³k h^–1) in 1 m HCl at 30°C for the initial linear part of the graph described by the Kabai equation (Ruan and Gilkes 1995).

Fig. 3.  Cumulative fraction of Fe dissolved in 1 m HCl at 20 ± 1°C from the soil Fe-oxide concentrates during the 192-h batch dissolution experiment.

Table 4.  Dissolution parameters of Fe-oxide concentrates derived from the cube-root law, Avrami–Erofejev equation and Kabai equation
***P < 0.001

The MCD of hematite had significantly negative relationships with dissolution rate constants derived from the three equations. As hematite was the dominant mineral in most of the Fe-oxide concentrates, the crystal size of this mineral has a major influence on the dissolution rate of the Fe-oxide concentrates (Table 5). Several previous studies have shown an inverse relationship between crystal size (and SSA) and the dissolution rate of Fe oxides (Cornell and Giovanoli 1993; Wells et al. 2001). Furthermore, Ruan and Gilkes (1995) reported that the dissolution rate of synthetic goethite decreased with increasing Al substitution and increased with increasing dissolution temperature, SSA and heating temperature. Other properties of Fe-oxide concentrates including Goe/(Goe + Hem) ratio, Al substitution in both Fe oxides and MCD of goethite did not show any relationship with the dissolution rate constants (Table 5).

Table 5.  Correlation coefficients for the relationships between some properties of Fe-oxide concentrates and dissolution rate constants (10³k)
*P < 0.05 and **P < 0.01

The steady-state dissolution measurement takes several days and requires special equipment, and hence for this measurement, we used only two samples, a synthetic goethite and a goethite-rich soil Fe-oxide concentrate (sample Kk). The dissolution rate for the two samples was measured using non-stirred flow-through reactors and using 0.01 m NaCl as the electrolyte. The dissolution curve and dissolution rate of synthetic goethite and Kk samples are shown in Fig. 4 and Table 6, respectively. Dissolution of synthetic goethite at pH 1.0 plotted against time (Fig. 4a) showed that the rate of Fe release was rapid during the initial period (up to 83 h) and then decreased before reaching the steady-state (~336 h). Iron release from synthetic goethite was generally low at pH 2.0 and 3.0 (Fig. 4b, c, respectively). The Fe concentration declining with time at pH 1.0 and 2.0 showed no systematic trend at pH 3.0. Little Fe was released rapidly at pH 1.0 from sample Kk (Fig. 4d), but by 80 h a constant rate of dissolution occurred. A similar trend occurred for pH 2.0 (Fig. 4e); however, the steady-state at this pH was reached later (after 288 h) than at pH 1.0. At pH 3.0 (Fig. 4f), Fe release from sample Kk was very low (close to detection limit 0.01 mg L^–1 of ICP-AES) and it increased slowly to reach a steady-state at ~372 h.

Fig. 4.  Dissolution behaviour of Fe in 0.01 m NaCl at pH 1.0, 2.0 and 3.0 from (a–c) synthetic goethite and (d–f) sample Kk (Fe-oxide concentrate from the Kk soil) using flow-through reactor method at 25 ± 1°C.

Table 6.  Solution pH, flow rate, Fe concentration and dissolution rate (R) of synthetic goethite and sample Kk (a soil Fe-oxide concentrate) at the steady-state (with ± s.e.), and Δ_rG of reaction

The dissolution rate of both the synthetic goethite and Kk samples followed a decreasing trend with solution pH, with the maximum dissolution rate observed at pH 1.0 and the lowest rate at pH 3.0 (Table 6). The dissolution rate of both Fe oxides decreased linearly with increasing solution pH from 1.0 to 3.0 (Fig. 5); however, the effect was more pronounced for sample Kk than the synthetic goethite. The dissolution rate of sample Kk was much greater than the synthetic goethite at pH 1.0 and 2.0, but at pH 3.0 the dissolution rate was similar for the two samples. The much higher dissolution rate of sample Kk than the synthetic goethite at pH 1.0 and 2.0 was possibly due in part to the higher SSA of sample Kk (129 m² g^–1) than the synthetic goethite (32 m² g^–1) (Cornell and Giovanoli 1993; Wells et al. 2001). Structural disorder in sample Kk, possibly resulting from the substitution of multiple elements (other than Al), may also have contributed to the higher dissolution rate of this sample compared with the synthetic goethite (Fig. 2). Individual substitutions of foreign elements in the goethite structure are known to produce different effects on the dissolution of the mineral. For example, the substitution of Mn, Co, Pb and Zn in goethite is known to increase its dissolution rate (Alvarez et al. 2005, 2008; Kaur et al. 2010), whereas the substitution of Al, Cr and Cd decreased the dissolution rate of goethite (Alvarez et al. 2007; Kaur et al. 2010). However, the presence of Zn, Pb and Cd in di- and tri-metal (Zn-Pb, Zn-Cd and Zn-Pb-Cd) incorporated goethite increased the dissolution rate of goethite (Kaur et al. 2010). The dissolution rates of the synthetic goethite and Kk samples at pH 3.0 in this study were close to the dissolution rates of goethite at pH 5.0 in the presence of 200 µm oxalate (dissolution rate was 3.3 µmol kg^–1 h^–1) reported by Cheah et al. (2003).

Fig. 5.  Dissolution rate of a synthetic goethite and an Fe-oxide concentrate from an Ultisol (sample Kk) plotted against solution pH using flow-through reactor method; the experiments were conducted using 0.01 m NaCl background solution at 25 ± 1°C.

The approximate steady-state dissolution of the synthetic goethite and Kk samples in this study showed under-saturation with respect to goethite and hematite, with Δ_rG <0 (Table 6). The Δ_rG for both the synthetic goethite and Kk samples increased with increasing pH. The Δ_rG value of synthetic goethite at pH 1.0 and 2.0 was slightly lower than that for Kk, and was very similar at pH 3.0. The Δ_rG values of the sample in this study were lower than the Δ_rG for goethite dissolution reported by Cheah et al. (2003) for various pH values.

As shown by transmission electron micrographs, the crystals of original synthetic goethite were acicular, with particle sizes ranging from around 0.5 to 2 µm (Fig. 6a). For sample Kk, the crystals had a flat, rounded morphology and small size (<20 nm) (Fig. 6c), which is consistent with MCD₁₁₀ and MCD₁₁₁ values of goethite in Table 2. The small crystal size of sample Kk corresponds to higher SSA than for synthetic goethite. There was no observable difference in the crystal size and shape before and after steady-state dissolution was reached at pH 1.0 for either the synthetic goethite (Fig. 6a, b) or Kk (Fig. 6c, d) samples. Although only a small proportion of the samples was dissolved, nevertheless these observations suggest that there was no selective dissolution along a particular crystal axis of the synthetic and natural samples. Schwertmann (1984) reported that both sides of the lath are attacked by protons, and rounded corrosion features are formed for Al-goethite, but unsubstituted Al-goethite showed preferential dissolution between domains and showed formation of a rhombic holes feature. Furthermore, Cornell and Giovanoli (1993) reported that all crystal faces of hematite dissolved equally well except where susceptible regions were present, and some preferential attack took place at the (001) faces of the platy crystal, leading to hole formation. Similarly, rounded hematite in soil from south-eastern Ohio (Washington county; 39°28′00″N, 81°33′30″W) exhibited dissolution cavities attributable to instability within the modern pedo-environment (Bigham et al. 1991).

Fig. 6.  Transmission electron micrographs of a (a, b) synthetic goethite and (c, d) an Fe-oxide concentrate from an Ultisol (sample Kk) before and after dissolution (steady-state conditions established), respectively, at pH 1.0 in 0.01 m NaCl.

Hematite and goethite are abundant Fe oxides in the Oxisols and Ultisols from Thailand examined in this study. Hematite was the dominant Fe oxide in most of the Fe-oxide concentrates of these soils. Iron, Al and Ti were present in significant quantities in the Fe-oxide concentrates. Manganese was the most abundant trace element in the Fe-oxide concentrates. Near-congruency of dissolution indicates that Al, Cr and V are substituted in goethite and hematite structures of most samples. Much of the Mn, Ni, Co, Pb, Si and Ti also appeared to present within the structure of these minerals, whereas Mg, Cu, Zn and Ca were closely associated with Fe oxides but perhaps not substituted in the minerals.

Dissolution behaviour of Fe-oxide concentrates in 1 m HCl at 20°C was well described by the three models investigated. The dissolution rate constant (10³k) values obtained using the three models are lower value than many reported values in the literature, due to the lower concentration of the acid and lower temperature used in our study. The dissolution constants were inversely related to the MCD along [110] and [104] of hematite, the dominant mineral in the Fe-oxide concentrates. The near steady-state dissolution rate of the soil Fe-oxide concentrate (Kk) was much higher than that of the synthetic goethite at pH 1.0 and 2.0; however, at pH 3.0 their dissolution rates were similar. The difference could be partly attributed to the higher SSA of sample Kk than the synthetic goethite. These results may have consequences on the availability of several elements including essential and toxic elements.

The results of this study substantiate the early data obtained for Fe-oxide concentrates for Australian and Thai soils (Singh and Gilkes 1992; Wiriyakitnateekul et al. 2007). The results of this study identified that in addition to Al, Cr, Ni, Mn and V, some other elements including Co, Cu and Pb may be associated with Fe oxides in highly weathered soils. Structural incorporation of small (except for Al) and variable amounts of these elements produces structural disorder resulting in much smaller crystal size of goethite and hematite. Our results suggest that Fe oxides isolated from highly weathered soils dissolved at a much faster rate than pure synthetic goethite under highly acidic conditions.