| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
a Departamento de Ciencias y Recursos Agrícolas y Forestales, Universidad de Córdoba, Edificio C4, Campus de Rabanales, 14071 Córdoba, Spain
b School of Ocean and Earth Science, National Oceanography Centre, Univ. of Southampton, European Way, Southampton SO14 3ZH, UK
c Dep. of Geography, Roxby Building, Univ. of Liverpool, Liverpool L69 7ZT, UK
* Corresponding author (torrent{at}uco.es).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODS4. RESULTS AND DISCUSSION11. CONCLUSIONSREFERENCES |
|---|
Hm) and the frequency-dependent magnetic susceptibility (
FD), which is used as a proxy for the concentration of fine-grained pedogenic maghemite, were found to be linearly correlated (R2 = 0.825, P < 0.001). This supports the idea that these two minerals were formed concomitantly during pedogenesis, which is consistent with the results of previous in vitro experiments showing that the ferrihydrite 
maghemite hematite transformation takes place under aerobic conditions. By contrast, the concentration of pedogenic goethite (Gt) was only weakly correlated with either FD or Hm, which suggests that goethite formed through an alternative pathway. The paleosols above 40 m (S4, S5, corresponding to marine isotope stages 9 and 11, respectively) exhibit a higher degree of weathering and higher Hm/(Hm + Gt) ratio than those below such a depth (S6–S8). This was ascribed to differences in paleoclimatic conditions, which are moister and warmer in the former paleosols than in the latter, rather than to differences in pedogenesis duration.
Abbreviations: CLP, Chinese Loess Plateau • Fed, citrate–bicarbonate–dithionite-extractable iron • Fet, total iron • GSD, grain size distribution • Gt and Hm, goethite and hematite concentrations, respectively • PSD, pseudo-single domain • SD, single domain • SP, superparamagnetic • Gt and Hm, pedogenic goethite and hematite concentration, respectively • ARM, anhysteretic remanent magnetization susceptibility • FD, frequency-dependent mass magnetic susceptibility • HF, high-frequency magnetic susceptibility • LF, low-frequency magnetic susceptibility
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODS4. RESULTS AND DISCUSSION11. CONCLUSIONSREFERENCES |
|---|
, and the degree of pedogenesis; see Maher et al., 1994; Vidic et al., 2004; and references therein).
The main phase accounting for magnetic enhancement is maghemite in the superparamagnetic (SP, <20–25 nm) and single-domain (SD, between 20—25 and 
100 nm) grain size regions (Fine et al., 1993; Verosub et al., 1993; Heller and Evans, 1995; Liu et al., 2005). The role of both SP and SD particles has been discussed in a number of studies (e.g., Geiss and Zanner, 2006, and references therein). Although SP maghemite exhibits inherently higher magnetic susceptibility values than does SD maghemite, the latter seems to contribute substantially to the total bulk magnetic susceptibility by virtue of its increased volume percentage (Liu et al., 2004b). It can therefore be concluded that the magnetic enhancement of paleosols is caused by the neoformation of pedogenic, fine-grained (SP + SD) maghemite, which, in contrast to the coarse ferrimagnetic particles present in the parent loess, is soluble in a citrate–bicarbonate–dithionite mixture (Verosub et al., 1993).
Because the CLP local climate (which is reflected, for example, in ) and the global climate (as represented, for example, by the marine O-isotope records) are related, the corresponding proxies have been extensively correlated (Heller and Liu, 1984, 1986; Balsam et al., 2005; and references therein). Moreover, has been further used to quantify variations in the amount of paleorainfall (Maher et al., 2003, and references therein). The contention that paleorainfall is the main factor affecting has been qualified, however, by studies that suggest the influence of soil temperature and the duration of pedogenesis (Maher, 1998; Fine et al., 1989; Vidic and Verosub, 1999; Vidic et al., 2004). Because the latter factors obviously affect other soil properties (e.g., clay, carbonate, and Fe oxide contents and color), a multiproxy approach is preferable with a view to reconstructing the nature of paleoclimates (Balsam et al., 2004; Vidic et al., 2004).
Interpreting the magnetic enhancement data for paleosols is complicated by the uncertainty in the relative significance of the different pathways from Fe-bearing minerals to maghemite in soils. As discussed by Cornell and Schwertmann (2003), maghemite can be formed by oxidation of detrital magnetite and or by thermal (fire) transformation of other Fe oxides [a term that is used here to designate all Fe(III) oxides, oxyhydroxides, and hydroxides]. It has also been hypothesized that, under reducing conditions—which determine the presence of Fe2+ in solution—magnetite is formed abiotically (Maher and Taylor, 1988) or by bacterial mediation (Fassbinder et al., 1993; Lovley et al., 1987; Chen et al., 2005), and later oxidized to maghemite with the onset of oxidizing conditions (van Velzen and Dekkers, 1999; Liu et al., 2003, 2004a). Recently, the hypothesis has been put forward that maghemite is an intermediate in the transformation of ferrihydrite—which is considered to be the first product of the weathering of Fe-bearing minerals—to hematite in aerobic soils (Barrón and Torrent, 2002; Barrón et al., 2003; Torrent et al., 2006). This is supported by the covariance of and hematite concentration in several CLP sections and in soil profiles on loess and other parent materials in various world regions (Torrent et al., 2006).
One other problem is the relationship between the concentration of maghemite and those of hematite (Hm) and goethite (Gt), which are the main Fe oxides produced in the course of pedogenesis. It is well established that the Hm/(Hm + Gt) ratio is highly dependent on the particular pedoenvironment, with colder, moister, more acidic, organic-matter-rich environments favoring goethite over hematite (Schwertmann, 1985). Balsam et al. (2004) related this ratio to and other paleoclimatic indicators in the Luochuan and Lingtai sections of the CLP. These researchers hypothesized that, above certain rainfall and temperature thresholds, hematite, goethite, and magnetite (which is subsequently oxidized to maghemite) are formed; with increasing rainfall (and temperature), the ratio of magnetite to hematite increases because the duration of the relatively dry period needed for the dehydration of ferrihydrite to hematite decreases and the duration of the anaerobic period [needed for reduction of Fe(III) to Fe(II) and subsequent formation of magnetite] also increases. With a further increase in rainfall, hematite is no longer produced. Eventually, if the soil is saturated with water (i.e., under anaerobic conditions) for long periods, one can expect Fe oxides to be reductively dissolved and the soil to become depleted in Fe (Bigham et al., 2002). Balsam et al. (2004) also found that the Hm/(Hm + Gt) ratio had steadily decreased in the last 2.5 x 106 yr, which suggests a change in the timing of precipitation, in temperature, or in both.
The first aim of the present work was to characterize in detail the relationships between the concentrations of pedogenic hematite, goethite, and fine-grained maghemite in the upper part of the Luochuan loess–paleosol sequence. For this purpose, pedogenic maghemite was represented by the frequency-dependent susceptibility, FD, defined as LF – HF, where LF and HF are the values of measured at low and high frequency, respectively. Such relationships were then used (i) to validate the various models proposed to explain the formation of the different Fe oxides, and (ii) to identify the paleoenvironmental differences responsible for the mineralogical differences among paleosols. Finally, several unresolved questions concerning the proposed pedogenic models were addressed.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODS4. RESULTS AND DISCUSSION11. CONCLUSIONSREFERENCES |
|---|
570 mm and mean annual temperature of 9.8°C. The surface 50-cm sediments were removed before sampling. The depth interval studied was about 23 to 55 m, which, according to the conventional notation used in the CLP (Bronger, 2003), comprises the following sequence of loess (L) and paleosol (S) units: L4–S4–L5–S5–L6–S6–S7–S8–L9 (Fig. 2
). The soil units consist mainly of rubified Bw and Bt horizons underlain by Ck horizons (Bronger, 2003). Samples were collected at 5-cm intervals and stored in polyethylene bags. All were used for magnetic measurements and 53 of them, spaced in accordance with the changes observed in magnetic properties, were used for comprehensive chemical and mineralogical analysis.
|
|
, expressed here in 10–6 m3 kg–1, was measured at two frequencies (LF = 0.47 kHz and HF = 4.7 kHz) with a Bartington MS 2B dual-frequency sensor (Bartington Instruments Ltd., Oxford, UK) and the frequency-dependent susceptibility, FD, was then calculated as LF – HF. This parameter is sensitive only to a very narrow grain size region crossing the SP/SD threshold (20–25 nm for maghemite) (Worm and Jackson, 1999). For natural samples, however, which generally exhibit a continuous and nearly constant grain size distribution (GSD), FD can be used as a proxy for relative changes in concentration in pedogenic fine-grained magnetic particles (Liu et al., 2005; see below). The relative FD (FD% defined as 100 x FD/LF) was then calculated to determine the relative variations in GSD of pedogenic maghemite particles.
Anhysteretic remanent magnetization (ARM) was imparted by using a peak alternating field of 100 mT with a biasing field of 40 µT. The ARM is particularly sensitive to the content of stable single-domain (typically larger than 20–25 nm, but less than 100 nm for maghemite) ferrimagnets (Dunlop and Özdemir, 1997). In what follows, ARM is expressed as ARM susceptibility (ARM = ARM/40 µT).
The combined frequency (1 and 10 Hz in fields of 240 A m–1, or 0.3 mT) and low-temperature dependence (10–300 K) of susceptibility curves (hereafter referred to as FD – T, where FD = 1Hz – 10Hz and T represents temperature) for samples at depths of 32.0, 33.5, and 37.2 m were measured using a Quantum Design Magnetic Properties Measurement System (Quantum Design, San Diego, CA). These samples were selected on the basis of their distinct bulk magnetic susceptibility across the S5–L6 boundary.
3.3 Diffuse Reflectance Spectroscopy Measurements and Calculations
Diffuse reflectance spectra were recorded from 380 to 900 nm in 0.5-nm steps at a scan rate of 30 nm min–1, using a Varian Cary 1E spectrophotometer equipped with a BaSO4–coated integrating sphere 73 mm in diameter (Varian Inc., Palo Alto, CA). Samples were previously powdered (<10 µm) and pressed by hand into the 8- by 17-mm rectangular holes of white plastic holders (thickness 2.5 mm) at a pressure >500 kPa. The resulting mounts were self supporting, thus allowing the holders to be vertically placed without the powder falling into the sphere. The Kubelka–Munk (K–M) remission function [F(R
)] at each wavelength was calculated from
 | [1] |
is the reflectance of a thick layer of sample (1–2 mm for samples containing Fe oxides). The first and second derivatives of the K–M function were calculated by using a cubic spline procedure. This method involves end-to-end joining of a number of cubic polynomial segments based on a certain number of adjacent data points with continuity in the first and second derivative at the joints. Segments with 30 data points were adopted because they provided second-derivative spectra with both well-resolved absorption bands and relatively low background noise.
The difference in ordinate between the minimum and the next maximum at a longer wavelength, which is hereafter referred to as band intensity, was used as a proxy for the true band amplitude (Scheinost et al., 1998). The intensities of the bands at 425 nm (I425) and 535 nm (I535) are proportional to the concentration of goethite and hematite, respectively (Scheinost et al., 1998), and can thus be used as proxies for relative changes in the mass concentration of goethite and hematite.
The concentrations of goethite and hematite were then quantified with two methods. One was based on the following calibration curve:
 | [2] |
Such a curve, where Y is the Hm/(Hm + Gt) ratio and X is the I535/(I425 + I535) ratio, was constructed from 22 samples of soils from the Mediterranean region (A and B horizons of Alfisols and Inceptisols of variable age and contents in goethite and hematite). The "true" concentrations of hematite and goethite in the samples were determined by differential x-ray diffraction. Equation [2] accounted for 90% of the variance in Y and held at 0.07 < X < 0.4, a condition which was fulfilled by the studied loess and paleosol samples. The mass concentrations of hematite and goethite were estimated from Y by assigning all citrate–bicarbonate–dithionite (CBD)-extractable Fe (Fed) to these two minerals. The contribution of ferrihydrite to Fed was neglected because oxalate-extractable Fe, which provides an estimate of the ferrihydrite concentration, was typically <8% of Fed. The mass contribution of maghemite, which is also dissolved by CBD, was also neglected on the basis of the results of Liu et al. (2004b). These researchers showed the saturation magnetization (Ms) of a Yuanbao paleosol sample with a bulk susceptibility of 1.3 x 10–6 m3 kg–1 to be only 0.05 A m2 kg–1. After correcting for the contribution from the parent loess (0.02 A m2 kg–1), the mass concentration of pedogenic maghemite was 0.4 g kg–1, which is much smaller than Fed in comparable samples in the present study (>10 g kg–1). Therefore, for simplicity, we assigned Fed to the combination of Fe in stoichiometric hematite and goethite:
 | [3] |
The second method was based on the following regression equation:
 | [4] |
Based on the good correlations between the estimates provided by the two methods [Hm(Method 2) = 0.81 + 0.846Hm(Method 1); R2 = 0.994; n = 53, P < 0.001; Gt(Method 2) = 1.16 + 0.883Gt(Method 1); R2 = 0.873; n = 53, P < 0.001], the average value was finally adopted for both hematite and goethite.
3.4 Chemical Analyses
Samples (ground to <2 mm) were analyzed for Fed by the method of Mehra and Jackson (1960), except that extraction temperature was 25°C and the extraction time 16 h. Oxalate-extractable Fe was determined by using the method of Schwertmann (1964), which involves one 2-h extraction with 0.2 M ammonium oxalate at pH 3. The total concentrations of Fe (Fet) and other elements were determined by isotope-source x-ray fluorescence, using a Metorex XMET920 system (Metorex International Oy, Espoo, Finland) with a 3 to 20 keV probe (109Cd source). Spectra were deconvoluted using the in-house PASCAL software DECONV (Boyle, 2000) and calibrated using standard samples of known elemental concentration.
Expressing the concentrations of hematite and goethite on a carbonate-free basis required estimating the mass concentration of CaCO3 equivalent in the sample, which was calculated by multiplying by a stoichiometric factor of 2.5 the total concentration of Ca less the mean Ca concentration in carbonate-free samples (6 g kg–1).
As noted above, we used Fed as a measure of Fe present in the Fe oxides, and the Fed/Fet ratio as an index of the degree of weathering of Fe-bearing minerals in the loess into secondary Fe oxides.
|
4. RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODS4. RESULTS AND DISCUSSION11. CONCLUSIONSREFERENCES |
|---|
LF, FD, and ARM (Fig. 2a–2c) show that the magnetic signals for the paleosol units are enhanced relative to the underlying loess, which can be ascribed to the pedogenic neoformation of fine-grained (SP plus SD) maghemite (Zhou et al., 1990; Forster et al., 1994; Maher et al., 2003; Liu et al., 2005). The values of these three magnetic parameters are systematically lower, however, in the peak paleosols below L6 (40 m) than those above this unit.
The FD–LF and the ARM–FD correlations are illustrated in Fig. 3a and 3b
, respectively. Previous studies have shown that FD% is determined mainly by the GSD of fine-grained magnetic particles in the SP and SD regions (Worm, 1998; Worm and Jackson, 1999). When the bulk susceptibility is relatively low, however, FD% can also be seriously distorted by contributions from coarse-grained pseudo-single-domain (PSD) and multidomain (MD) magnetite, as well as by Fe silicates in the loess (Liu et al., 2004c). Such contributions can be estimated from the x axis intercept, o, which represents contributions to the bulk from the parent material (Forster et al., 1994); for our profile, o was 0.2 x 10–6 m3 kg–1 (Fig. 3a). The same rationale can also be applied to ARM. Although ARM is sensitive to the concentration of SD magnetic particles, contributions from PSD and MD particles are not negligible, especially in the unaltered loess and intermediate paleosol samples. The contribution of aeolian inputs to ARM can be estimated from the y axis intercept in Fig. 3b.
|
FD% relates inversely to the width of the GSD of fine-grained particles. For many moderately weathered soils and paleosols, FD% is 8 to 12% (Maher, 1988; Dearing et al., 1996), therefore pedogenic particles exhibit a relatively broad GSD. Our results (11.3%; slope of the regression line in Fig. 3a) are consistent with those previous studies. The narrow range spanned by the FD% values thus indicates that the GSD of pedogenic SP + SD maghemite is very similar across the studied sequence (Liu et al., 2005).
The use of biplots of ARM and for magnetic granulometry is traditionally named the "King method" (Banerjee et al., 1981). The /ARM ratio has later been used more often as a grain size proxy (dimensionless, or equivalently ARM/). Stable SD particles exhibit the highest ARM, but the lowest , yielding a minimum /ARM. By contrast, both SP and PSD/MD particles exhibit a higher , but a lower ARM SP particles are supposed to carry no remanence. Thus, higher /ARM ratios are observed in these two cases.
The /ARM ratio (0.33) for the parent loess can be estimated as (o – para)/ARMo, where para (0.05 x 10–6 m3 kg–1; Liu et al., 2004b) is the susceptibility contributed by the paramagnetic background. The /ARM ratio (0.145) for pedogenic maghemite particles is estimated as 1/(60.7 x 0.113), where 60.7 is the slope of the regression line (Fig. 3b) with a physical meaning of (ARM – ARMo)/FD, and 0.113 the slope of the regression line in Fig. 3a. Clearly, aeolian inputs have a higher /ARM ratio than pedogenic particles, consistent with the fact that the GSD of aeolian magnetite is coarser than that of pedogenic maghemite. For pedogenic particles, the /ARM ratio (0.145) is higher than the typical 0.09 to 0.1 value found for stable SD particles (Maher, 1988). The difference (0.045) is contributed by SP maghemites; therefore, the contribution of SD particles to the susceptibility enhancement is about twice (0.09/0.04 = 2) that of SP particles, which is consistent with data for samples from the last glacial–interglacial intervals (Liu et al., 2004b).
10.2 Low-Temperature Dependence of FD
The room-temperature FD% has limitations in representing the total concentration of SP particles because this parameter is sensitive only to viscous SP particles with size near the SP/SD threshold (Worm, 1998; Worm and Jackson, 1999). To determine the GSD of pedogenic maghemite particles, Liu et al. (2005) used the Néel theory and the low-temperature variations in FD. At room temperature, SD particles (< 100 nm) change their magnetic properties sharply around 20 to 25 nm. For SD particles larger than this size threshold, susceptibility is independent of frequency because they are blocked. For SD particles much smaller than this threshold, susceptibility is also independent of frequency because the time constants for the susceptibility measurements are much larger than the relaxation time. Therefore, FD peaks at the SD/SP threshold. As the temperature is lowered, a decrease in grain size occurs that corresponds to the maximum FD value. Thus, temperature can be used as a window with a view to determining the grain size distribution of finer SP particles. In other words, the FD–T curve can be taken as an analog of the GSD curve, even though there is a nonlinear transformation between grain size and temperature (Liu et al., 2005).
As can be seen in Fig. 4a
, the FD–T curves for three representative samples across the L6/S5 boundary are very similar. This suggests that the GSD of the pedogenically produced SP particles differs little among samples. Moreover, the FD–T curves for these samples from the Luochuan section of the CLP (Fig. 4a) resemble those from sections in both the western and eastern parts of the CLP (Fig. 4b). Therefore, we can conclude that the GSD of pedogenic SP particles is independent of the degree of pedogenesis. Consequently, the room-temperature FD is proportional to the concentration of these pedogenic SP particles, and thus can be confidently used as a proxy to quantify their relative concentration changes.
|
20 nm and <3% dissolution for crystals of 40 nm; unpublished data; see also Reyes and Torrent,1997). In the 10 paleosol samples studied, the oxalate treatment caused a 6 to 18% decrease in LF (mean = 10%), but no significant change in FD%. This result is consistent with the idea that the pedogenic ferrimagnets consist of maghemite rather than magnetite.
In summary, the linear trends between FD, LF, and ARM and the consistent low-T FD–T behavior strongly suggest that the GSD of the pedogenic ferrimagnets is similar throughout the Luochuan section and ranges from the SP to SD grain region. The FD% and the contribution of SP and SD maghemites to the bulk susceptibility are both consistent with previous results for other loess profiles, which suggests that the genesis of these minerals has been uniform throughout the CLP. A recent study on modern loessic soil profiles from the midwestern USA (Geiss and Zanner, 2006) also exposed the similarity of the pedogenic ferrimagnetic component (which consisted of a mixture of SP, SD, and possible PSD particles). Thus, in terms of GSD, one consistent magnetic enhancement process functions across rather a wide range of precipitation conditions (Geiss and Zanner, 2006).
10.3 Depth Profiles of Hematite and Goethite
Figures 2d, 2e, and 2f show the depth distribution of hematite, goethite, and the Fed/Fet ratio, respectively. The concentration of Fet on a carbonate-free basis (not shown) was relatively constant (34–40 g kg–1), which suggests that the parent loess was essentially homogeneous. By contrast, the concentration of Fed spanned a wider range (9–15 g kg–1) as a result of the loess being weathered in the paleosol layers. Based on the Fed/Fet ratio, the less weathered layers are the L4, L5, L6, and L9 loess units, with Fed/Fet = 0.24; on a carbonate-free basis, this corresponds to a background Fed concentration in the region of 7.5 g kg–1, and background hematite and goethite concentrations of approximately 4 and 7 g kg–1, respectively. A moderately high correlation exists between Fed/Fet and FD (Fed/Fet = 0.239 + 0.367FD; R2 = 0.531; n = 53, P < 0.001). This indicates that weathering of the loess Fe-bearing minerals (mainly Fe chlorite; Ji et al., 2006) has resulted in the formation of significant amounts of fine-grained ferrimagnets.
The hematite concentration ranges from 2.9 to 11.2 g kg–1 (3.6–11.2 g kg–1 on a carbonate-free basis) and its depth distribution is roughly parallel to that of LF and FD; therefore, Hm and FD are highly correlated (Hm = 3.53 + 28.8FD; R2 = 0.825; n = 53, P < 0.001; Fig. 5a
). If we now consider that the concentration of pedogenic hematite, Hm, is the Hm – Hm0 difference, where Hm0 is the y axis intercept—which provides an estimate for the hematite concentration at FD = 0, i.e., in the background loess, then the previous correlation simply indicates that Hm/FD is essentially constant throughout the profile. The concentration of goethite, Gt, ranges from 6.3 to 11.8 g kg–1 (7.2–11.8 g kg–1 on a carbonate-free basis). In contrast to hematite, Gt, or Gt (i.e., the concentration of pedogenic goethite as calculated similarly to Hm) is only weakly correlated with FD (R2 = 0.09; P < 0.05; Fig. 5b). Accordingly, the correlation between Fed and FD adopts an intermediate R2 value (Fed = –0.219 + 0.0319FD; R2 = 0.754; n = 53, P < 0.001).
|
Balsam et al. (2004) reported similar goethite concentrations, but hematite concentrations two to three times lower than those of Fig. 2 (1.6 g kg–1 for the loess units and 3.4 g kg–1 for the S5 paleosol peak). The difference can be tentatively attributed to the standard used by these researchers (a synthetic hematite) rather than to the diffuse reflectance method they used for quantification (described in Ji et al., 2002). The idea that the nature of the standard resulted in underestimated hematite concentrations is supported by Fig. 2 in Ji et al. (2002), where the combined Fe contents in hematite and goethite was about 20 to 30% lower than Fed in the calibration samples.
10.4 Relationships between Maghemite, Hematite, and Goethite
The high correlation between FD—a proxy for the concentration of pedogenic maghemite—and the hematite concentration supports the hypothesis of Torrent et al. (2006) that maghemite is an intermediate product in the ferrihydrite to hematite transformation in aerobic soils. This model implies that various ligands are sorbed on the surface of ferrihydrite particles, which are thus "blocked" and can undergo internal rearrangement and slow dehydroxylation to hematite via maghemite (Barrón and Torrent, 2002; Barrón et al., 2003). The experiments of Barrón and Torrent (2002) revealed complete conversion of maghemite to hematite at temperatures above 100°C. Should this model be applicable to soil environments, one could expect a gradual increase in the hematite/maghemite ratio with weathering and progress of the ferrihydrite maghemite hematite transformation, which is rarely the case (Torrent et al., 2006). It is therefore likely that part of the maghemite formed does not evolve to hematite, the partitioning between maghemite and hematite depending on the particular soil environment. For instance, maghemite would tend to be stabilized in the presence of large concentrations of blocking ligands. The soil environment is also likely to determine the fraction of the precursor ferrihydrite that is directly transformed to hematite, as suggested by synthesis experiments in ligand-poor media (Barrón and Torrent, 2002); in fact, irrespective of the synthesis conditions, some hematite particles are always formed at the earlier stages of the ferrihydrite transformation. In any case, the high correlation between the pedogenic maghemite and hematite concentrations supports the hypothesis that these two minerals form concomitantly.
A soil environment favoring dissolution of ferrihydrite and further polymerization of the Fe3+ ions in solution in turn favors the formation of goethite, which predominates in soils that are cool, moist, and rich in organic matter (Schwertmann, 1985). The previous considerations led to the well-established idea that the Hm/(Hm + Gt) ratio largely reflects the nature of the environment in which pedogenic Fe oxides have formed, and can thus be used as a paleoenvironmental indicator.
Based on the ferrihydrite maghemite hematite model, FD is influenced, among other factors, by (i) conditions favoring the blocking of ferrihydrite by some ligands and its further transformation into maghemite and hematite—less pedogenic maghemite is generally found in aerobic, hematite-free soils relative to similar hematite-containing soils (Torrent et al., 2006), and (ii) the rate and duration of weathering of the primary Fe-bearing minerals, which determine the amount of precursor ferrihydrite. This obviously limits the usefulness of FD as the sole paleoclimatic indicator. A multiproxy approach based on magnetic properties, color, Fed/Fet ratio, and other pedogenetic indices is thus required (Vidic et al., 2004). Specifically, as discussed below, the parallel variation of FD and the hematite concentration in the Luochuan profile makes it reasonable to support any paleoclimatic reconstruction on the Hm/(Hm + Gt) ratio as well.
Based on the Fed/Fet ratio, the paleosols above 40 m (S4 and S5) are more weathered on average than are those below this depth, which may be the result of differences in paleoclimate or pedogenesis duration. The differences in Fed/Fet are not reflected in significant differences in the Hm/FD ratio between the two groups of paleosols, however. This suggests that the nature of the processes leading to the concomitant formation of maghemite and hematite was essentially similar for both groups.
To further compare the two paleosol groups, we used the Hmcf/(Hmcf + Gtcf) ratio, where Hmcf and Gtcf represent the concentrations of pedogenic hematite and goethite, respectively. They were calculated as (Hmcf – Hmcf0) and (Gtcf – Gtcf0), where Hmcf and Gtcf are the respective carbonate-free concentrations of hematite and goethite in the sample, and Hmcf0 and Gtcf0 are the lowest values of these concentrations in the parent loess. The depth profile of the ratio is shown in Fig. 6
, where only the samples with Fed/Fet >0.26 were plotted to exclude weakly weathered soil samples. The ratio is significantly larger in the paleosols above 40 m (S4 and S5) than in those below such a depth (S6–S8). One cannot ascribe the difference to the higher degree of weathering (due, for example, to a longer pedogenesis) in S4 and S5 because no correlation exists between the Hmcf/(Hmcf + Gtcf) ratio and Fed/Fet—which is a good proxy for the degree of weathering. It must then be concluded that the paleosols above 40 m developed in an environment that was more favorable to the formation of hematite than those below it.
|
Hmcf/(Hmcf + Gtcf) ratio, as it is well established that an increase in temperature favors hematite over goethite (Schwertmann, 1985). Since present and past climates in the CLP have been governed by alternations in the warm, moist southwest summer monsoon and the dry, cool northwest winter monsoon, an increase in paleorainfall is likely to be associated with an increase in temperature. This hypothesis is consistent with the observed differences in marine benthic 
18O and 13C records, which suggest strengthened summer monsoons in S4 and S5 relative to S6 to S8 (Guo et al., 2000).
Our results point to the usefulness of the hematite–goethite relationship as a paleoclimatic indicator for the loess–paleosol sections, consistent with previous studies (Vidic et al., 2004; Balsam et al., 2004). Unlike other useful pedological indicators (e.g., the absolute hematite and goethite concentrations, FD, Fed/Fet, and Rb/Sr), the ratio of pedogenic hematite to goethite is likely to be relatively independent of the duration of pedogenesis, as suggested by data for various soil chronosequences (Torrent, 1976; Torrent et al., 1980).
10.5 Unresolved Questions
A long-sustained hypothesis is that biotic or abiotic formation of fine-grained magnetite (Maher, 1998; Maher et al., 2003, and references therein; Chen et al., 2005), which is later oxidized to maghemite, contributes substantially to the soil magnetic enhancement. Significant reduction of Fe(III) to Fe(II), however one strict requirement for magnetite formation is unlikely to occur in well-drained soils, where hematite can form and substantial magnetic enhancement can develop. In somewhat imperfectly drained soils, where the Fe2+ needed for the formation of magnetite is likely to appear in the solution during the wet season, little magnetic enhancement occurs and goethite rather than hematite is formed. Consequently, the coexistence of pedogenic magnetite (whether oxidized to maghemite or not) and hematite in the Chinese loess paleosols is paradoxical in the light of that hypothesis. Such a paradox does not exist in the ferrihydrite maghemite hematite model because maghemite not magnetite and hematite are products at different stages of a consistent transformation process. Moreover, a biotic origin is inconsistent with the weak correlation found between total N, a proxy for organic matter and hence biotic activity in soil, and FD (R2 = 0.204; P < 0.01). Indeed, we cannot exclude the possibility of magnetite forming under microanaerobic environments in otherwise well-drained soils, as suggested by some electron microscopy studies (e.g., Chen et al., 2005). The relative significance of this pathway, however, remains to be established, particularly in magnetically enhanced soil horizons where organic compounds—the main electron donors for the reduction of Fe(III)—are present at low concentrations.
The goethite concentration in the least altered loess samples is about 2.5 times higher than that of hematite (i.e., the parent loess contains more goethite than hematite). Because aeolian inputs were transported from the source regions by Asian winter monsoons, the higher initial goethite contents might reflect the fact that more goethite was formed in the source region or during the transportation process during cold periods. Confirming this hypothesis would require further research into the spatial distribution of goethite and, especially, the goethite content of the parent material in the source region.
One other question is the relationship between hematite and maghemite. Our conceptual model predicts that both minerals formed through the same pedogenic pathway. If that were the case, we could expect them to exhibit some crystallochemical similarities (e.g., substitution of foreign ions for Fe). More elaborate experiments that were not performed in this study are needed to check this assumption. Finally, the coexistence of hematite and maghemite, and SD maghemite and SP maghemite, requires testable hypotheses on which factors affect the rates of the transformation steps in the pathway from ferrihydrite to hematite.
|
11. CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODS4. RESULTS AND DISCUSSION11. CONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODS4. RESULTS AND DISCUSSION11. CONCLUSIONSREFERENCES |
|---|
Received for publication September 15, 2006.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODS4. RESULTS AND DISCUSSION11. CONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. Cabello, M. P. Morales, C. J. Serna, V. Barron, and J. Torrent MAGNETIC ENHANCEMENT DURING THE CRYSTALLIZATION OF FERRIHYDRITE AT 25 AND 50{degrees}C Clays and Clay Minerals, February 1, 2009; 57(1): 46 - 53. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
