Illinois
Fertilizer Conference Proceedings
January 24-26, 1994
| Home | 1994 Index | Search |
| Home | 1994 Index | Search |
Siyuan Shen and Joseph W. Stucki1
Potassium is one of the most important plant nutrients in soils, and has thus been studied extensively (Potash and Phosphate Institute, 1980; Munson, 1985). Despite much study and diligent efforts, the fundamental chemical and physical phenomena that govern its fate, movement, and availability to plants in soils have yet to be characterized fully (Bertsch and Thomas, 1985). Soil tests for K often fail to reveal the true fertilizer demand in the field, resulting in frequently unreliable and inefficient fertilizer recommendations (Munson, 1980; Ted R. Peck, University of Illinois Soil Testing Laboratory, Personal Communication). Many factors contribute to this problem, but perhaps the one factor making the solution to this problem so elusive is that K, which in the soil phase is usually distributed among soluble, exchangeable, fixed, and insoluble forms, becomes redistributed among these forms in an unpredictable manner. In other words, the fate and movement of K in the soil have yet to be well understood. Because soil tests measure primarily only the soluble and exchangeable forms of K, any uncontrolled or unrecognized transitions of K among its forms between the time of testing and the time of attempted uptake by the plant roots will produce an erroneous prediction of available K.
While previous studies provided empirical evidence that K behavior is correlated to a number of different soil and environmental factors (Munson, 1980), such as the types of soil minerals present (Rich, 1964), moisture regime (Barber, 1960; Bates and Scott, 1964; Mengel and von Branschweig, 1972), cropping and fertilizer history (Doll et al., 1965; T.R. Peck, personal communication; Kaspar et al., 1989), temperature fluctuations (Sparks and Liebhardt, 1982), and weathering (Mortland et al., 1956), no unified explanation linking all of these variables in a consistent manner has been established.
Soil testing methods have evolved over the years. As science recognized little relationship between crop-available and total K in the soil (Reitemeier et al., 1951), the exchangeable form of K was used to estimate plant-available forms. But, again, correlations were wrong about as often as they were right. Hence, an improved soil test method was proposed by McLean et al. (1982) utilizing the "Quantity-Intensity" (Q/I) concept advanced by Beckett (1964). This test is based on the ratio of total K (Q) to an intensity (1) term, which is the quotient of thermodynamic activities of K, Ca, and Mg. While this method is somewhat more rigorous from a chemical and mathematical viewpoint than other methods, it is still largely empirical insofar as the mechanism for K availability is concerned. The concept recognizes the fact that the soil matrix consists of a heterogeneous array of energy sites for K retention, and those sites that bind with lower energies will more readily release K to the plant. Utilizing this principle of varying binding energies, electroultrafiltration (EUF) has been applied to differentiate the chemical dynamics of soil K in terms of nutrient availability (Grimme and Nemeth, 1979). Sparks and Jardine (1982) investigated the kinetics of K adsorption and desorption in a soil, and provided further rigorous chemical methods (Sparks and Huang, 1985) for treating K data. But, again, the mechanisms are incomplete.
One puzzling observation is that the amount of exchangeable K in a routine soil test changes with moisture content -- sometimes increasing (Bohannon, 1957) and sometimes decreasing (Bates and Scott, 1964). One explanation that is sometimes suggested for increased exchangeability upon drying is that K+ is expelled from the inter-laminar spaces of the clay minerals along with the evaporating water; but this appears to contradict fundamental electrostatic neutrality requirements. Bates and Scott (1964) found that low-volatile organics and sucrose both inhibit release of K+, which they explained as the result of diffusion inhibition.
These and other results indicate that other processes may be involved which have yet to receive adequate consideration. One such process is oxidation and reduction of expanding soil minerals. Oxidation and reduction (redox) is the terminology used to describe the chemical process which changes the electrical charge of a chemical element. Some of the elements in soils which are susceptible to redox changes are iron (Fe), manganese (Mn), nitrogen (N), and sulfur (S), and such processes are commonly found in agricultural soils as a result of biological activity and alternate wetting and drying events. The diffusion inhibition mechanism proposed by Bates and Scott (1964) may actually be the result of microbial growth and reducing conditions in the soil, producing a change in redox activity.
Redox reactions involving the above-mentioned elements, and particularly Fe,
are responsible for many changes in the physical and chemical properties of
soils (Yu, 1985). While the transformation of N to its various forms, ranging
from nitrate (NO3-) to ammonium (NH4+),
is probably the most-studied redox process in soils, significant changes in
oxidation state also occur in many soil minerals because of the presence of
Fe in their crystal structures. The weathering of primary minerals is an example
of an oxidation process in which Fe 2+ is converted to more expansive
types which release various nutrients, including K+, to the soil
solution. The in situ reduction of Fe 3+ to Fe 2+ in secondary
minerals (vermiculite, montmorillonite, illite) also occurs, and creates a climate
in which some of the beneficial effects of weathering may be reversed, such
as the fixation of K+. This is due to numerous changes that are invoked
in the physical-chemical properties of the mineral phase of soils, including
swellability in water (Stucki et al., 1984), electrical charge (Stucki and Roth,
1977; Lear and Stucki, 1985), and surface area (Lear and Stucki, 1989). These
redox phenomena are of great importance to soil fertility because the availability
of plant nutrients depends in large degree on the surface chemistry of the minerals.
Since the oxidation state will vary with different environmental conditions,
the associated properties also may change significantly throughout the year.
The purpose of this chapter is to review the present state of knowledge of the
role of Fe oxidation state in determining K availability in soils.
The levels of Fe(II) attained by the chemical reduction of six clay samples are reported in Table 1 (Shen and Stucki, 1994). Sample SWa-1 contained the most total Fe of all samples, and the Drummer soil from Urbana was highest among the soil clays. The Fe(II) content of these samples in their initial, oxidized (unaltered) state varied from less than 0.16% of the total Fe in smectite to more than 25% in Fithian illite. The initial Fe(II) content of the soil clays was similar to smectite API 25, ranging from 3% to 10% of their total Fe.
When Shen and Stucki (1994) reduced samples using sodium dithionite, the Fe(II)
content exceeded 60% of total Fe in all samples except the Drummer soil from
Dekalb was less than 40%, which they attributed to some reoxidation during the
post-reduction washing procedure. The levels of Fe(II) achieved were not necessarily
the limiting case because longer treatments are known to reduce all of the Fe
in sample SWa-1 (Komadel et al., 1990), which was only reduced 75% by the treatment
reported in Table 1.
Smectites
Recent studies showed that the change in surface charge that occurs during
Fe reduction increases the ability of the mineral to fix interlayer cations
(Figure 1), including Na+ (Lear
and Stucki, 1989), Ca2+, Cu2+, and Zn2+ (Khaled
and Stucki, 1991), and K+ (Chen et al., 1987). Chen et al. (1987)
observed a rapid increase in the non-exchangeable form of K+ with
Fe2+ content of freezedried soil clays and standard reference clays,
wherein the total fixation capacity reached as high as 30% of the total exchange
capacity. Khaled and Stucki (1991) followed the distribution of K+
between exchangeable and non-exchangeable fractions with increasing Fe2+
content of Upton montmorillonite gels that were never dried, and found that
fixation increased steadily with increasing Fe" but that the exchangeable
fraction remained
almost constant (Figure 2).
Khaled and Stucki (1991) also offered the following sample calculation showing the potentially great effect of oxidation state on the availability of K in soils. By interpolating the data in Figure 2 to a reduced level of only 20% (% Fe(II) to Total Fe = 20), the estimated amount of fixation is about 10 meq/100 g of clay. In a soil having a clay content of 15% by weight, of which 2/3 is smectite, the K+ fixation capacity would be about 900 lbs K2O/acre 6-inches. Reduction levels may greatly exceed this value in the soil under various environmental conditions, so the potential for reduced clay to serve as a sink for K is tremendous. This point is even more poignant in view of the fact that a high application rate for K fertilizer is in the range of 200 lbs K2O/acre. A modest change in the oxidation state of Fe in the mineral could, therefore, remove virtually all of such an application from plant availability -- at least temporarily.
Recent studies (Shen and Stucki, 1994) of two reference smectite clays (SWa-1 and API 25), one reference illite (Fithian), and the clay fractions from the A horizon of three Illinois soils (Drummer from Dekalb, Drummer from Urbana, and Cisne from Brownstown), confirmed that the total and fixed K contents of smectites increased when the clays were reduced, while the exchangeable K contents remained constant (Table 2). This result is consistent with observations by Chen et al. (1987) and Khaled and Stucki (1991). So we can generalize by stating that structural Fe reduction increases K fixation in smectites; and thus, likely decreases K availability in smectite soils.
Illite
The Fithian illite sample behaved very differently from the smectites. The amount of fixed K decreased upon treatment with the reducing agent, while Ke, the exchangeable K, increased (Table 2). The total K content also decreased. Soils containing a mixture of illite and smectite would, therefore, either fix or release K during reducing conditions depending on which clay dominates; or, if these minerals are precisely balanced, no change in K availability would occur because the amount that one releases would be fixed by the other. One possible explanation for the behavior of illite is that as more Fe(II) is added to the octahedral sheet during further reduction, the dipole moment of structural OHgroups become tilted more along the c-axis (the axis perpendicular to the clay plates) and thereby create a weaker attractive (or stronger repulsive) force between the clay layer and the interlayer K ions (Juo and White, 1969). Release of K to soil solution subsequently would be increased.
Soil Clays
Shen and Stucki (1994) also found (Table 3)
that the total K content of both Drummer soils was constant during one redox
cycle, but it varied slightly in the Cisne soil. The distribution of K between
exchangeable and fixed fractions, however, fluctuated measurably, but less than
in the pure clay samples. The behavior of soil clays appeared to be closer to
that of illite than of smectite, suggesting that the soil clays may be dominated
by illite..X-ray powder diffraction (Figure
3) revealed that the clay fraction of these three soils was about two-thirds
illite and one-third smectite.
Iron oxides in the soil may significantly influence K availability. Shen and Stucki (1994) studied soils with and without Fe oxides present. The Fe oxides in the clay fraction of three soil samples were removed by CBD treatment before the reduction experiment. Results from this experiment (Table 4) revealed that after oxides were removed from the soil clays, both exchangeable and fixed K increased when the clays were reduced, resulting in an increase in total K as well. Hence; removal of Fe oxides caused the soil clays to behave somewhat more like smectite than like illite. Although the difference in fixed K was not very large between the reduced and unreduced samples, it was significant for the levels of Fe reduction achieved in the samples. Chen et al. (1987) achieved similar amounts of fixed K when they reduced and freeze-dried the soil samples in their experiment.
Linear regression analysis reported in Table
5 reveals the quantitative relationship between fixed or exchangeable K
and the Fe(II) content of the various soil clays. Notice that the correlation
between fixed K and Fe oxidation state was excellent in the two Drummer soils,
but was rather poor in the Cisne soil. We attribute the latter to variability
in the Cisne sample.
All samples, including K-saturated reference and soil clays with oxides, were subjected to cyclic reduction and reoxidation treatments. The reoxidation treatment consisted of bubbling O2 gas through the reduced suspension for 30 min, exposing it to air for three days, or adding sodium hypochlorite solution to the reduced suspension. The Fe(II) and fixed K contents in the reduced and reoxidized samples were measured with each cycle, and varied depending on the mineralogy.
Smectite
The level of Fe(II) in ferruginous smectite, where Fe ions are clustered within the clay crystal, is reflected in the intervalence electron transfer (IT) band observed at about 730 nm (Lear and Stucki, 1987). The continuous monitoring of the IT band in Fe-rich smectite (SWa-1) suspensions revealed that this band increases in intensity with- the number of Fe(II)-0-Fe(III) linkages in the clay structure, and reaches a maximum at about 45% of total Fe reduced to Fe(II) (Lear and Stucki, 1987; Komadel et al., 1990).
Spectrum A in Figure 4 shows that in the
first 10-20 min of reduction, about 45% of structural Fe was reduced in the
K-saturated SWa-1 clay. But more than 3 hr was required before an additional
20% of structural Fe was reduced. When O2 was introduced into the
suspension to reoxidize the clay, this latter 20% of Fe(II) was oxidized quickly
(in about 0.5 hr). But the 45% reduced initially was more difficult to reoxidize.
Even after a 6-hr exposure to O2 about 2001o of structural Fe was
still in the Fe(II) state. Chemical analysis (method of Komadel et al., 1988)
of thesamples treated similarly revealed that 17% of the Fe
remained as Fe(II) after 6 hr of reoxidation. By contrast, the Na-saturated
SWa-1 was completely reoxidized in half an hour (Komadel et al., 1990). The
difference between Na- and K-saturated SWa-1 reduced in suspension indicates
that K fixed in the interlayer of the clay during reduction can protect the
structural Fe(II) from re-oxidation to a large extent. This probably is due
to the collapse of superimposed clay layers to the exclusion of interlayer water.
Six redox cycles were completed with Fe-rich smectite SWa-1 (Figure 4). In spectra B and C are reported the results from the third and fifth redox cycles of the same sample described by spectrum A. The rate of initial reduction and the time required for completion of the reaction both decreased with increasing number of cycles, presumably because the Fe(II) contents were a little higher at the end of each cycle, and thus less Fe(III) remained in the sample. The first peak maximum decreased with each cycle, indicating that the available number of Fe(II)-O-Fe(III) pairs in the sample was less after each cycle. The Fe(II) remaining in the clay at the end of each cycle, therefore, must be clustered and uninvolved in subsequent redox cycles.
In Figure 5 the residual Fe(II) that is present after the reoxidation step of the respective redox cycles, as determined by wet chemical analysis (Komadel et al., 1988), is plotted vs. redox cycle number. Notice that teh residual Fe(II) level increases steadily to about 24% of the total Fe. The rate of Fe reduction appears to slow after 5 or 6 cycles, but conceivably the trend of increasing residual Fe(II) with each cycle could continue indefinitely to produce Fe(II) contents similar to that observed in Fithian illite (60%, Table 2).
Also reported in Figure 5 is the amount of K that is nonexchangeable or fixed in the clay after several redox cycles. Reduction of the Fe increases fixed K in the smectite from about 2.5 to about 35 meq/100 g. Reoxidation decreases back to only about 25 meq/100 g, thus failing to release the K that was fixed by the Fe reduction reaction. Subsequent redox cycles produced similar values upon reduction and reoxidation, indicating that a stable level of K fixation apparently is reached immediately with the first reduction treatment. Similar results were found for smectite API 25. Perhaps the combination of K fixation and the gradually increasing residual Fe(II) in the reoxidized clays could constitute a mechanism for illite formation in soils. In any case, these processes certainly will alter the kinetics of K availability in smectitic soils.
Illite and Soil Clays
Fithian illite changed color from dark gray to white and the clay fraction of Cisne soil changed color from a yellowish brown to light gray when reduced. No intervalence transfer (IT) band was observed in these clays because the Fe ions in the clay structure were not located in adjacent crystallographic sites, but charge transfer (CT) did occur between O and Fe(III), giving a change in visible absorbance intensity at about 400 nm. The approximate position of the band for this electronic transition is 245 nm, and only the tail is seen at 400 nm. The peak itself cannot be studied by this system because the reducing agent absorbs uv light at approximately the same wavelength, but the reducing agent does not interfere with the tail of the O-Fe(III) charge transfer at 400 nm. The intensity of the absorbance at 400 nm is directly proportional to the Fe(III) concentration in the sample, as given by the Beer-Lambert Law, viz.,
| Io | ||||
| A = | log | ------- | = | elc |
| I |
where A is the absorbance; to and 1, the incident and transimtted photon intensities, respectively; epsilon, the absorptivity; 1, the sample thickness; and c, the concentration of O-Fe(III) moieties in the sample.
The changes in intensity at 400 nm with time of redox treatment are reported in Figure 6 for soil clays. The absorbance decreased with Fe reduction due to the loss of intensity from the CT band at 245 nm as explained above. The intensity was initially high, but dropped rapidly with time after the commencement of reduction. When O2 gas was introduced, the intensity increased steadily, reflecting the increasing level of Fe(III) in the sample. The final state of re-oxidation recorded was below that of the unaltered sample, indicating a resistance to complete reoxidation similar to that observed in ferruginous smectite (sample SWa-1).
The CT band was also monitored continuously at 400 nm for K-saturated Fithian illite during reduction and reoxidation with the hope that some understanding of the behavior of illite in the soil clay fraction could be derived. Recall that in this clay 35% of the Fe initially is Fe(II). When subjected to further reduction by citratebicarbonate-dithionite, the color changed noticeably. As illustrated by the patterns in Figure 7, during the first half hour of reduction of the K saturated illite the absorbance decreased rapidly. After 3.5 hr of reduction, and thereafter, the absorbance remained essentially constant. When O2 was used to reoxidize the clay suspension, the absorbance increased quickly in about half an hour, then after 2 hr it returned to the level that existed before reduction. No further change in absorbance was observed despite continuous treatment with O2, and even the introduction of sodium hypochlorite (a strong oxidizing agent) could not increase the absorbance further. These results, combined with chemical analyses (Table 2), signify that the original Fe(II) in K-saturated Fithian illite is not easily oxidized, but that which was reduced in the laboratory was rather readily reoxidized.
When the illite was reoxidized, the amount of fixed K increased and the exchangeable decreased, while the total changed only slightly. So, during one redox cycle the illite released K into its environment, in contrast to the smectite which removed K from its surroundings.
For the Drummer soils (from Dekalb and Urbana), both fixed and exchangeable K were little changed after one redox cycle. But in the Cisne soil, by comparison, the amount of fixed K decreased and exchangeable K increased after one reduction-reoxidation step, and total K was virtually unchanged.
The trend for K behavior is, therefore, in the direction of lower K availability for smectites and greater K availability for illites after one redox cycle. The mineralogy of soils is, therefore, very important for determining the effect of Fe oxidation state on K availability in the soil. A soil containing a mixture of illite and smectite would either fix or release K during reducing conditions depending on which clay dominates; or, if these minerals are precisely balanced, no change in K availability would occur because the amount that one releases would be fixed by the other. Further study to quantify the comprehensive effects of Fe reduction on K availability for different mixtures of clay mineral types in soil clays is needed.
The reduction of Fe in the crystal structures of reference and soil clay minerals
alters the distribution of K between exchangeable and fixed states. In smectite
clay minerals, Fe reduction increased K fixation without changing the
amount available in exchangeable sites. The rate of K release was slowed and
some fixed K was retained in the clay even when the sample was re-oxidized.
Several cyclic reduction and reoxidation steps produced increased levels of
Fe(II) and fixed K that apparently are difficult to reverse, and these levels
may steadily increase over many such cycles. These results can be explained
by the clay layers becoming collapsed as a result of Fe reduction, and thereby
forming larger particles due to K fixation.
Figure 1. Effect of Fe oxidation state on cation fixation (from Khaled and Stucki, 1991)
Figure 5. Effect of redox cycles on Fe(II) and fixed K contents of ferruginous smectite (SWa-1)
Table 1: Iron contents of oxidized, reduced, and reoxidized reference clays and Illinois soil clays
Table 2: Total, fixed, and exchangeable K contents of reference clays (n = 4)
Table 3: Total, fixed, and exchangeable K contents of three Illinois soil clays (with oxides)
1 Graduate Research Assistant and Professor of Soil Physical Chemistry, respectively, University of Illinois, W-317 Turner Hall, 1102 South Goodwin Avenue, Urbana, IL 61801.
Barber, S. A. 1960. The influence of moisture and temperature on phosphorus and potassium availability. Trans. Int. Congr. Soil Sci. Proc., 7th 3:435-440. Bates, T. E., and A. D. Scott. 1964. Changes in exchangeable potassium observed on drying soils after treatment with organic compounds: I. Release. Soil Sci. Soc. Am. Proc. 28:769-772.
Beckett, P. H. T. 1964. Studies on soil potassium: I. Confirmation of the ratio law: Measurement of potassium potential. J. Soil Sci. 15:1-8.
Bertsch, P. M., and G. W. Thomas. 1985. Potassium status of temperate region soils. p. 131-162. In R. D. Munson (ed.) Potassium in Agriculture. Soil Science Society of America, Madison.
Bohannon, R. A. 1957. The effect of drying on exchangeable K in soils from Illinois and Kansas. Ph.D. Thesis, University of Illinois.
Chen, S. Z., P. F. Low, and C. B. Roth. 1987. Relation between potassium fixation and the oxidation state of octahedral iron. Soil Sci. Soc. Am. J. 41:82-86.
Doll, E. C., M. M. Mortland, K. Lawton, and B. G. Ellis. 1965. Release of potassium from soil fractions during cropping. Soil Sci. Soc. Am. Proc. 29:699-702.
Grimme, H., and K. Nemeth. 1979. The evaluation of soil K status by means of soil testing. Proc. Congr. Int. Potash Inst. 11:99-108.
Juo, A. S. R., and J. L. White. 1969. Orientation of the dipole moments of hydroxyl groups in oxidized and unoxidized biotite. Science 165:804-805.
Kaspar, T. C., J. B. Zahler, and D. R. Timmons. 1989. Soybean response to phosphorous and potassium fertilizers as affected by soil drying. Soil Sci. Soc. Am. J. 53:1448-1454.
Khaled, E. M., and J. W. Stucki. 1991. Effects of iron oxidation state on cation fixation in smectites. Soil Sci. Soc. Am. J. 55:(In Press). Komadel, P., P. R. Lear, and J. W. Stucki. 1990. Reduction and reoxidation of iron in nontronites: Rate of reaction and extent of reduction. Clays Clay Miner. 38:203-208.
Komadel, P., and J. W. Stucki. 1988. The quantitative assay of minerals for Fe2+ and Fe3+ using, 1,10-phenanthroline. III. A rapid photochemical method. Clay Miner. 36:379-381.
Lear, P. R., and J. W. Stucki. 1985. Role of structural hydrogen in the reduction and reoxidation of iron in nonronite. Clays Clay Miner. 33:539-545.
Lear, P. R., and J. W. Stucki. 1987. Intervalence electron transfer and magnetic exchange interactions in reduced nontronite. Clays Clay Miner. 35:373-378.
Lear, P. R. and J. W. Stucki. 1989. Effects of iron oxidation state on the specific surface area of nontronite. Clays Clay Miner. 37:547-552. McLean, E. O., J. L. Adams, and R. C. Hartwig. 1982. Improved corrective fertilizer recommendations based on a two-step alternative usage of soil tests: II. Recovery of soil-equilibrated K. Soil Sci. Soc. Am. J. 46:1198-1201.
McLean, E.O., J.L. Adams, and R.C. Hartwig. 1982. Improved corrective fertilizer recommendations based on a two-step alternative usage of soil tests: II. Recovery of soil-equilibriated K. Soil Sci. Soc. Am. J. 46:1198-1201
Mengal, K., and L. C. von Branschweig. 1972. The effect of soil moisture upon the availability of potassium and its influence on the growth of many maize plants (Zea mays L.). Soi Sci. 134:142-148.
Mortland, M. M., K. Lawton, and G. Uehara. 1956. Alteration of biotite to vermiculite by plant growth. Soil Sci. Soc. Am. Proc. 31:286-287. Munson, R. D. 1980. Potassium availability and uptake. p. 28-66. In Potassium for Agriculture -- A Situation Analysis. Potash and Phosphate Institute, Atlanta.
Munson, R.D. (ed.) 1985. Potassium in Agriculture. American Society of Agronomy, Madison. 1223 p.
Potash and Phosphate Institute. 1980. Potassium for Agriculture - A Situation Analysis. Potash and Phosphate Institute, Atlanta. 216 p. Reitemeier, R. F., I. C. Brown, and R. S. Holmes. 1951. Release of native and fixed nonexchangeable potassium of soils containing hydrous mica. USDA Tech Bull. 1049.
Reitemeier, R.F., I.C. Brown, and R.S. Holmes. 1951. Release of native and fixed nonexchangeable potassium of soils containing hydrous mica. USDA Tech Bull 1049.
Rich, C.I. 1964. Effect of cation size and pH on potassium exchange in Nason soil. Soil Sci. 98:100-106.
Shen, S., and J. W. Stucki. 1994. Effects of cyclic redox processes on K fixation in smectite. (In Preparation).
Sparks, D. L., and P. M. Huang. 1985. Physical chemistry of soil potassium. page 201-276. In R.D. Munson (ed.) Potassium in Agriculture. Soil Science Society of America, Madison.
Sparks, D. L., and P. M. Jardine. 1981. Thermodynamics of potassium exchange in soil using a kinetics approach. Soil Sci. Soc. Am. J. 45:1094-1099.
Sparks, D. L., and W. C. Liebhardt.1982. Temperature effects on potassium exchange and selectivity in Delaware soils. Soil Sci. 133:10-17. Stucki, J. W., and C. B. Roth. 1977. Oxidation-reduction mechanism for structural iron in nontronite. Soil Sci. So. Am. J. 41:808-814. Stucki, J. W., P. F. Low, C. B. Roth, and D. C. Golden. 1984. Effects of oxidation state of octahedral iron on clay swelling. Clays Clay Miner. 32:357-362.
Stucki, J.W., and C.B. Roth. 1977. Oxidation-reduction mechanism for structural iron in nontronite. Soil Sci. So. Am. J. 41:808-814.
Stucki, J.W., P.F. Low, C.B. Roth, and D.C. Golden. 1984. Effects of oxidation state of octahedral iron on clay swelling. Clays Clay Miner. 32:357-362.
Yu, T. R. 1985. Physical Chemistry of Paddy Soils. Springer-Verlag, New York.
217 p.
