Association of Macrocrystalline Goethite and Humic Acidin Some Oxisols from Brazil

M. R. Fontes, S. B. Weed,* and L. H. Bowen

ABSTRACTOxisols from the cocoa-growing region of the state of Bahia, Brazil,

may contain large amounts of Fe oxides and significant amounts oforganic C to depths exceeding 2 m. The nature of the association oforganic C and crystalline oxides was the objective of this study. Alkaliextracts (0.1 M NaOH) of soil material taken from different depthsin three Typic Hapludox and one Humic Hapludox were characterizedby chemical analysis, x-ray diffraction, Mossbauer spcctroscopy, elec-tron microscopy, surface area determination (N2 adsorption), Fourier-transform infrared spectroscopy, and phosphate adsorption. Micro-crystalline goetbite was the only inorganic phase accompanying thehumic acid in the alkaline solution and humic acid was essential forgoethite to be extracted. Mossbauer spectra confirmed the absence ofFe oxides other than goethite in the humic-acid extracts although hem-atite and ilmenite were identified in the total soil material. The goeth-ite particles were 10 to 15 nm in diameter and nearly equidimensional.Infrared spectra of the humic acid-goethite complex indicated thatthe COOH group was in the salt form, suggesting bonding throughthis group to the goethite surface. Phosphate adsorption was markedlyhigher on the humic-acid-free goethite than on the complex, indicatingcompetition for adsorption on the goethite by the humic acid. Appar-ently, the interaction between the humic acid and the goethite involvesligand exchange in addition to coulombic attraction, and may signif-icantly modify the properties of the Fe oxide.

THE SURFACES of Al and Fe oxyhydroxides (re-ferred to for convenience as oxides) are widely

believed to interact with carboxyl groups in naturalhumic materials. The adsorption of the HS at oxidesurfaces would influence both dissolution and crys-tallization of the oxides (Schwertmann, 1966). Ac-cording to Schwertmann et al. (1986), if theenvironment into which Fe is released when a mineralweathers is highly humiferous, complexing of Fe byHS will compete with Fe3+ oxide formation. Susserand Schwertmann (1983) reported an increase of > 100

M.R. Fontes, Reforestadora Simpson, Lida., DIAGONAL 6, IT-97 ZONA 10, Guatemala; S.B. Weed, Dep. of Soil Science, andL.H. Bowen, Dep. of Chemistry, North Carolina State Univ.,Raleigh, NC 27695-8204. This work was supported in part by theNational Science Foundation, Grant EAR-850656. Received 23May 1991. "Corresponding author.

Published in Soil Sci. Soc. Am. J. 56:982-990 (1992).

m2 g-1 in external surface areas (N2 adsorption) whennatural ferrihydrites were treated with H2O2 to destroyorganic matter, suggesting substantial association ofthis disordered oxide with organic colloids. Studies ofadsorption of isolated HS on synthetic oxides haveestablished some model systems for these interactions(Watson et al., 1973; Parfitt et al., 1977a,b,c;Schwertmann, 1988). Parfitt et al. (1977a,b,c) showedthat oxalate, benzoate, and fulvic acids will adsorb ongoethite and gibbsite. The adsorption mechanism in-volves ligand exchange between COOH groups of or-ganic substances and metal atoms in the oxide surface,with displacement of surface OH groups. Infraredanalysis of the complexes identified singly coordi-nated OH groups on the goethite surface that may bedisplaced by anions such as phosphate, oxalate, andcarboxylate (Parfitt et al., 1977a,c). Adsorption stud-ies of humic acids on Fe and Al oxides have not beenso fully explored, partly because of the insolubility ofhumic acid in acid solutions (Parfitt, 1977b).

Humic acid (HA) is defined as the fraction of soilorganic matter that is soluble in alkali, insoluble inacid solutions, and dark brown to black in color(Schnitzer and Khan, 1972). Preparations of HA com-monly obtained by extraction of soil material withalkali may contain quite large amounts (e.g., 30%) ofinorganic components (Huffman and Stuber, 1985);the extraction is usually followed by purification pro-cedures to obtain an organic fraction free from ash,and the organo-inorganic complexes are destroyed ordiscarded. Thus, if an organic fraction is associatedwith an inorganic fraction, the latter will not be seen.

Preliminary analyses have shown that some Oxisolsof the humid Brazilian cocoa-growing region contain largeamounts of crystalline oxides of Fe and Al, predomi-nantly goethite, hematite, and gibbsite. Noncrystallineoxides are virtually absent. Some of these Oxisols con-tain significant amounts of organic C throughout the pro-

Abbreviations: HS, humic substances; HA, humic acid; CBD,citrate-bicarbonate-dithionite; ECEC, effective cation-exchangecapacity at the soil pH; FTIR, Fourier-transform infrared; XRD,x-ray diffraction; RT, room temperatures; QS, quadrupole split-ting; IS, ispmer shift; MS, Mossbauer spectra; MCDhk,, meancrystallite diameter.

FONTES ET AL.: GOETHITE AND HUMIC ACID ASSOCIATIONS IN OXISOLS 983

file to depths exceeding two meters. The nature of thisorganic material and its association with the crystallineoxides were investigated in this study.

MATERIALS AND METHODSSoil Samples

Soils selected for sampling were under natural vegetation(humid tropical forest) in the Brazilian cocoa-growing re-gion (southern region of Bahia state). The soils were formedfrom predominantly acid rocks and have udic soil moistureregimes and isohyperthermic soil temperature regimes.

The soils were sampled on shoulder slope positions (12—15% slope) near sites previously described in a ComissaoExecutiva Piano Lavoura Cacaueira (CEPLAC, 1975) soilsurvey. Sampling depths were 0 to 10, 10 to 20, 20 to 40,40 to 60, 100, and 200 cm. Selected properties of the soilsare given in Table 1. The analyses were performed at CE-PLAC Laboratories using the Brazilian Soil Survey meth-odologies (Empressa Brasileira de Pesquisa Agropecuaria,1982). These methods included organic C by wet oxidationwith K2Cr2O7 (organic C determined by total combustionis also included in Table 1); reductant-soluble Fe oxides onwhole soil samples using a CBD extract; clay content byhydrometer after dispersion of the sample with 0.1 M NaOH;soil reaction (pH) in a 1:2.5 soil/water ratio; and base sat-uration based on the ECEC.Soil Organic-Matter Fractionation

Humic acids were extracted with 0.1 M NaOH (Schnitzer,1982). The soil plus extractant were maintained under N2

Table 1. General properties of the four Oxisols used in thestudy.

OrganicfC

ProfileDepth

cm

0-1010-2020-4040-60

100200

0-1010-2020-4040-60

100200

0-1010-2020-4040-60

100200

0-1010-2020-4040-60

100200

1

22.113.09.67.65.73.3

34.322.915.211.57.28.0

22.613.111.16.24.42.9

28.916.312.18.65.62.2

Rasp2 Fedt Clay saturation pH

—— gkg-1 ————— %PI — Typic Hapludox (Valenca PontaDH

23.9 73 260 19 5.113.2 89 280 20 5.011.1 95 300 17 5.08.5 117 360 25 5.37.4 133 380 27 5.44.0 207 350 22 5.3

P2 — Humic Hapludox (Agua Sumida)45.8 77 530 17 4.626.9 90 570 9 4.618.5 78 590 7 4.714.9 74 610 7 4.710.6 78 620 10 4.610.3 80 650 15 4.6

P3 — Typic Hapludox (Una)28.1 83 380 27 5.115.6 111 420 18 4.914.1 115 480 19 4.97.7 126 550 25 5.05.0 119 520 27 5.03.2 106 470 22 5.0

P4 — Typic Hapludox (Una Mocambo)32.2 138 560 18 5.218.8 139 590 11 4.513.7 155 690 12 4.98.9 142 720 12 4.86.1 149 750 10 4.72.2 144 580 10 4.7

Munsellcolor§

10YR 4/410YR 4/4

7.5YR 4/47.5YR 4/47.5YR 5/6

4YR 5/6

10YR 3/210YR 3/310YR 4/410YR 5/610YR 5/610YR 5/6

10YR 3/310YR 4/410YR 5/510YR 5/69YR 5/6

7.5YR 5/6

SYR 3/4SYR 4/4SYR 4/6SYR 4/6SYR 4/64YR 4/5

11. Organic C by K2Cr207 oxidation; 2. organic C by total combustion.t Reductant (dithionite) soluble Fe.§ Moist color.1! Names in parentheses are soil series names.

gas while being shaken for 24 h at room temperature. Thedark-colored supernatant liquid was separated by centrifu-gation at 10 000 rpm (relative centrifugal force = 16 000)for 20 min in a Sorvall1 RC5C refrigerated centrifuge witha Model GSA fixed-angle rotor (Sorvall Products, Wil-mington, DE). The supernatant was transferred to anothercentrifuge tube and the centrifugation process was repeatedunder the same conditions. The soil residues were thensuspended in 50 mL of deionized water and centrifuged asbefore. These washings were added to the original alkalineextract. The soil residue, which included the "humin" frac-tion, was freeze-dried and stored for later analysis.

The alkaline extract was acidified to pH 3 with 1 M HC1and allowed to stand for 24 h at room temperature. Theprecipitate (HA) was then separated from the soluble ma-terial (fulvic acid) by centrifugation, washed with deionizedwater to remove soluble salts, and freeze-dried. The HumicHapludox, P2, yielded more HA per gram of soil at eachdepth than did the three Typic Hapludoxs (PI, P3, and P4).The results reported for material sampled at depths > 10 cmrefer mainly to material from soil P2.

A portion of the HA complex was treated with NaCIO, pH9.5, for organic-C removal (the product is labeled HAgt). An-other portion of the HA complex was treated with hot 6 MHC1 for Fe-oxide removal (labeled HAh). After these treat-ments, the residues of both samples were centrifuged, resus-pended, and washed twice with deionized water. The suspensionswere then transferred to dialysis tubing (molecular-weight cu-toff 12 000-14 000) and dialyzed against deionized water, afterwhich the salt-free samples were freeze-dried, passed througha sieve with 0.25-mm openings and stored.

Clay FractionationCoarse and fine clay were separated from the natural soil

material by centrifugation of aqueous suspensions aftertreatment with NaCIO to destroy organic matter (Anderson,1963; Jackson, 1979).

Physical AnalysisExternal (nonexpanded) surface areas of HA and HAgt

were determined by single-point N2-gas adsorption using aQuantachrome Quantasorb Surface Area Analyzer (Quan-tachrome Corp., Syosett, NY).

Transmission electron micrographs of flocculated anddispersed samples of the HA, HAgt, and HAh were ob-tained. Flocculated samples were suspended in distilled waterand dispersion of HA and HAh was achieved by ultrasonicagitation for 60 s in pH 10 water (0.111 g Na2CO3/L) usinga Branson Sonifier Cell Disruptor (Model W-350, BransonSonic Power Co., Danbury, CT) equipped with a microtip.Dispersion of HAg, samples required extensive dialysis againstdeionized water. Small droplets of the suspensions weretransferred to carbon-coated copper grids, dried, and viewedwith a JEOL Model 100S transmission electron microscope(Japan Electronics Optics Lab, Peabody, MA).

Fourier-transform infrared adsorption spectra were ob-tained using a Nicolet 5DX FTIR spectrometer (NicoletAnalytical Instruments, Madison, WI) equipped with a dataprocessing terminal. Potassium bromide pellets (300 mg)containing 10 mg of HA, HAh, or HAgt in 1000 mg of KBrwere scanned from 4600 to 400 cm-1. Background correc-tion was obtained using a blank KBr pellet. Difference spectrawere generated by subtracting spectra of HAgt from thoseof HA using software provided with' the instrument.

Magnesium-saturated coarse- and fine-clay samples andhumin-clay (clay separated from the soil residue after NaOHtreatment to extract HA) were prepared as smears on glass

1 The use of trade names in this publication does not implyendorsement by the North Carolina Agricultural Research Serviceof the product named, or criticism of similar ones not mentioned.

984 SOIL SCI. SOC. AM. J., VOL. 56, MAY-JUNE 1992

microscope slides (Theisen and Harward, 1962). A DianoXRD-700 diffractometer (Diano Corp., Woburn, MA) em-ploying Cu Ka radiation, a 1 ° divergent beam slit, and adiffracted-beam graphite monochromator was utilized forthe analysis. Scans were made at rates of 2 ° 20 min-1 and0.2° 26 min-1.

Randomly oriented HA and humin-clay specimens wereprepared by placing the dry HA into an Al powder specimenholder and pressing it gently against porous bibulous paper.When the amount of HA was not sufficient to fill the Alholder, the samples were mixed with ethyl alcohol and driedon a glass microscope slide.

Mossbauer spectra of the total soil sample, the humin-clay fraction, and the HA were obtained at 16 ± 0.1, 80± 0.1, and 295 ± 2 K as described by De Grave et al.(1982).

Aluminum substitution in the goethites was estimatedbased on the parameters taken at 80 K using equationsdeveloped by Amarasiriwardena et al. (1988). These werebased on synthetic samples, which were of somewhat largerparticle size than those found in the soils. Thus, Al substi-tution is overestimated by this method.

The size of goethite particles was estimated from XRDpeak broadening and from surface-area measurements. Meancrystallite diameter of goethite was determined for dno,dI30, djn, and d140 reflections in humin-clay and HA sam-ples using the Scherrer formula (Klug and Alexander, 1954).X-ray patterns for this purpose were obtained at a scanningrate of 0.2 ° 26 min-1 for both humin-clay and HA sam-ples. A diffraction pattern of 2- 5- |xm quartz was used toobtain the instrumental broadening parameter. Estimates ofparticle size from surface-area measurements of HA , as-sumed a particle density of 4.3 g cm-3 (Deer et al., 1975).

Phosphate AdsorptionPhosphate adsorption by HA and HA.,, was compared

using a single adsorption point from a solution with initialP concentration of 3.1 mol m-3. This concentration wasselected based on results with natural and artificial goethites(Fontes, 1988; Golden, 1978). Precise amounts of =50 mgof the HA were used for the analysis. Amounts of HAgtused corresponded with the total amount of oxides (Fe andAl) in the respective HA sample. The amounts used variedfrom 16 to 20 mg, based on the total chemical analysis ofthe HA from PI, P2, P3, and P4. Results are expressed asmillimoles of P adsorbed per gram of goethite (correctedfor Al substitution). Triplicate samples were weighed intoacid-washed 50-mL polyethylene centrifuge tubes. Potas-sium dihydrogen phosphate and KC1 were added to thetubes to give a solution that was 3.1 mol P m-3 in 0.1 MKC1 while maintaining a solid/solution ratio of 1:100. Thesamples were dispersed ultrasonically for 30 s with a Bran-son Sonifier Cell Disrupter equipped with a microtip andthen shaken continuously for 24 h. Suspension pH of theHAgt at equilibrium was approximately 6.5; that of the HAwas slightly lower. The soil suspensions were then centri-fuged at 10 000 rpm (RCF = 16 000) for 30 min in aSORVALL RC5C refrigerated centrifuge with a Model GSAfixed-angle rotor. Aliquots of the supernatant liquid wereanalyzed for P according to the method of Murphy andRiley (1962) as modified by Watanabe and Olsen (1965).The amount of P adsorbed was calculated from the differ-ence between the initial and final P concentrations in so-lution.

Chemical AnalysisTotal C and N in HA and whole-soil samples were de-

termined using a Perkin-Elmer Elemental Analyzer (Model2400 CHN, Perkin-Elmer, Norwalk, CT). Total Fe, Si, andAl were determined in HA samples using a double-step

digestion: 10 to 50 mg of the dark-colored HA were firsttreated with 15 mL of NaCIO (pH 9.5) to destroy organicmatter; the yellow-colored residue was then dissolved in 25mL of hot 6 M HC1. The Fe, Si, and Al were determinedusing a Perkin-Elmer Plasma II Emission Spectrophoto-meter with a 7500 Professional Computer.

RESULTS AND DISCUSSIONMineralogy of Alkaline Extract

X-ray DiffractionGoethite was the only crystalline phase identified in

the HA fraction (Fig. 1 and 2). Very little fulvic acidwas present and it was not characterized. Kaolinite andgibbsite dominate the fine- and coarse-clay fractions ofthese soils; however, they are largely absent from theHA fractions, indicating that centrifugation was effectivein removing most of these materials from the dispersedorganic fraction. Figure 1 illustrates these findings forsoil PI. A trace of gibbsite may have remained in theHA fraction.

The possibility that the NaOH treatment dispersed andextracted discrete particles of goethite, i.e., goethite notassociated with humic acid, was tested by treating thetotal soil material with NaCIO for destruction of organicmatter prior to NaOH extraction. The samples were thentreated with NaOH as for HA extraction. No goethite orany solid phase was detected in the supernatant liquid,

P3HA

P4HA

40 30 20 10Degrees 29

Fig. 1. X-ray diffraction patterns of coarse clay (PIC) and fineclay (P1F) from soil PI and humic acid (HA) from the 0- to10-cm depth of the respective soil profiles. K = kaolinite;Gb = gibbsite; Gt = goethite.

FONTES ET AL.: GOETHITE AND HUMIC ACID ASSOCIATIONS IN OXISOLS 985

10-20 CD

20-40 en

Fe/C2

40-60 en

100 ci

45 40 35

Degrees 26Fig. 1. X-ray diffraction patterns of humic acid (HA) from

different depths of profile P2.

indicating that HA was essential to goethite dispersionin the extraction procedures used.

There is some evidence for increased crystallinity ofthe goethites with depth. Although the goethite peaksfrom 0- to 10-cm material are well defined, there is aslight sharpening of some peaks near 0.254 nm (35 ° 29)in deeper horizons, as illustrated for soil P2 (Fig. 2).TheFe/C ratio also increases with depth in P2 (Fig. 3).

Mossbauer SpectroscopyMossbauer spectroscopic data (Tables 2 and 3) con-

firm the absence of Fe-oxide forms other than goethitein the HA fractions, although hematite and ilmenite areseen as components of the whole soil material at the 200-cm depth in P2 and P4 (Table 3). Goethite exhibits adoublet at RT and a sextet at reduced temperature. Thepresence of a doublet at 80 K plus the broad distributionof fields (Fig. 4; asymmetric hyperfine fields) suggestsvery small particle size or high Al substitution of thegoethite.

Spectra of HA from PI and P3 are very similar to thatof P4 shown in Fig. 4. However, the spectrum of P2obtained at 80 K suggests that the goethite in this ma-terial differs from that in the other soils studied in thatthere is a larger doublet contribution, i.e., conversion tothe magnetic state is less complete (Table 2). This maybe due to smaller particles or higher Al substitutions inP2 than in the other soils. This is supported by the datain Table 4, indicating generally greater Al substitutionfor the organic-matter-rich 0- to 10-cm horizon of P2than for the other profiles, although interpretation is con-founded by the effect of small particle size, to be dis-cussed below.

Results for whole-soil materials (Table 3) indicate thepresence of other ferruginous forms and corroborate thepreferential association of the humic acid and goethite,i.e., only goethite was extracted with the HA. The Fe2+

doublet due to ilmenite is distinguished by its low QS(0.6-0.7 mm s"1). Its absence in the humin-clay sug-gests that ilmenite occurs in coarser fractions. The pres-

eo

0)Q

100

200 JFig. 3. Relationship between Fe/C in the alkaline extract and

depth in the four soil profiles.

ence of hematite is shown by a sextet in the patternobtained at RT. Hematite is present in lesser amounts inthe humin-clay fraction than in total soil samples at anydepth, indicating its preferential occurrence as aggre-gates or perhaps concretions in fractions coarser thanclay. Mossbauer spectroscopy showed no significant dif-ference between the goethite in the humin-clay fractionand the goethite in the HA fraction for P4 (Tables 2 and3). The field values are essentially equal. Relative ab-sorption area of ilmenite is distinctly higher for all RTspectra than for 80 K, possibly due to a higher recoillessfraction at RT than the other components, or, less likely,due to the overlap of sextet distributions at 80 K. Rel-ative proportions of goethite, hematite, and Fe3+ doublet(probably Fe3+ in layer silicate) are about the same at80 K and 16 K. However, a doublet fit for Fe2+ at lowtemperature gave unreasonable QS and IS, because il-menite may order antiferromagnetically below 55 K(Syono et al., 1981). Thus, the 16 K spectra cannot beused for proper estimates of relative absorption area,although goethite/hematite ratios should be correct anddo agree with the 80 K spectra within an error of + 2.5%.Relative-area values at 80 K give the best relative pro-portions of the various Fe components. Hematite is clearlyidentified in P4 and at 200 cm in PI, the redder originalsoil samples (Table 1). The redder soil samples wereseen to yield smaller amounts of extracted HA, sug-gesting that HA may inhibit hematite formation in thesesystems (Schwertmann, 1966).

Chemical Composition of Alkaline ExtractChemical composition of the HA fraction (Table 4)

shows a very high inorganic content, predominantlyFe and Al. The low Si content indicates that little, ifany, layer silicate was included in the alkaline extract.

The rationale for calculating Al substitution (Table5) from the total contents of Fe and Al is based onthe premise that Fe occurs solely as crystalline oxides

986 SOIL SCI. SOC. AM. J., VOL. 56, MAY-JUNE 1992

Table 2. Mossbauer parameters! for Fe oxide associated with humic acids (HA).Goethite (sextet)

Profile

PI

P2

P3

P4

PI

Depth

0-10

0-10

0-10

0-10

200

Temperature^

RT8016RT8016RT8016RT8048RT

IS

_0.450.46_

0.490.47_

0.470.48

__

0.480.48-

QS

-0.26-0.18

__

-0.11-0.22

__

-0.17-0.12

__

-0.23-0.24-

B.v

36.544.9_ .

32.045.0_

36.145.0

_

37.546.3-

B™,- T ————

46.248.6

__

45.448.0

__

46.848.6

__

46.348.3-

RA<7

92100__

76100__

89100

_

93100-

IS

0.360.45

—0.350.50

—0.370.50

—0.360.49

—0.36

Fe3+ doublet

QSav

0.560.87

—0.580.84—

0.571.01

—0.550.84

—0.57

RA_

1008

—10024—

10011—100

7—100

t IS = isomer shift referenced to Fe metal at RT; QS = quadrupole splitting; Bav = average magnetic field; Bmax = magnetic field of maximum probability;QSav = average quadrupole splitting; RA = relative absorption peak area.

t RT = room temperature.

in these extracts and that the Al measured is containedwithin the oxide structure. The premise is reinforcedby estimates of Al substitution using XRD and MSparameters. Mossbauer spectroscopy may overesti-mate Al substitution due to the extremely small sizeof the goethite particles (Murad and Schwertmann,1983). However, the data indicate highly substitutedgoethites.

Schulze (1984) has shown that, in synthetic goeth-ites, the c dimension of the unit cell decreases linearlywith increasing Al substitution. Aluminum-substitu-tion data by chemical analysis, XRD, and MS for PI,P3, and P4 are in the range of high substitution fornatural goethites reported by Schwertmann (1988) and,in P2, the values calculated by XRD or MS equal orexceed the maximum reported values, suggesting higherreal Al substitution in this profile.

Chemical data indicate constant values of Al sub-stitution in goethite throughout profile P2 (Table 5).

This soil is a Humic Hapludox and the amounts ofHA extracted were much higher than for the othersoils. In the other profiles, the calculated mole frac-tion of Al in deeper horizons was extremely high,exceeding 0.9 (data not shown), suggesting the ex-traction of some Al compound with the HA. Theamounts of HA extracted from the deeper horizons ofthese three profiles were also very low (1-2 mg g-1

soil, compared with 20 mg g"1 soil for P2) so that asmall amount of Al could give an unreasonable valuefor estimated Al substitution.

Goethite Particle SizeParticle size of goethites in the HA and humin-clay

fractions, estimated by XRD line broadening, arecompared in Table 6. Broad diffraction lines and ashift of XRD line positions are effects associated withvery small crystallites, attributable to the physics of

Table 3. Mossbauer parameters! for humin clay and for whole soil samples at different soil depths.Hematite

Profile

PI

P4

P4

P4

P4

Depth Fractioncm200 whole

soil0-10 humin

clay

0-10 wholesoil

40-60 wholesoil

200 wholesoil

Temperature^ Bav B^K

RT

RT8016RT8016RT8016RT8016

_____ Tp _ ____

49.4 -

50.952.4 52.6§52.8 53.050.852.452.950.952.552.850.452.252.2

52.8§53.1

52.8§53.0

52.6§52.6

RA

19

121415261921221718221719

Goethite

«av B™,—— T ——

38.0 47.145.8 48.9_

37.344.0_

37.144.4_

37.645.5

__

47.149.0

__

47.048.9_

46.848.9

RA

7274__

5762__

6668_6869

Fe3+ doublet

QS.,mm s"1

0.58

0.560.610.540.530.680.710.540.610.610.580.650.59

RA

74

88141154111065108

7097

Fe2+ doublet

QS.,mm s"1

0.64

0.671.25

110.601.11

110.721.11

11

RA

7

20137

1476865

t IS = isomer shift referenced to Fe metal at RT; QS = quadrupole splitting; Bav = average magnetic field; Bmax = magnetic field of maximum probability;QSav = average quadrupole splitting; RA = relative absorption peak area.

t RT = room temperature.§ RT spectra fit to one sextet B = Bav = Bmax.H Fe1* doublet: ilmenite. At 16 K, magnetic broadening occurs and QSav is not a valid parameter. The RA was, however, estimated from a broadened doublet

fit.

FONTES ET AL.: GOETHITE AND HUMIC ACID ASSOCIATIONS IN OXISOLS 987

r- ,2 Table 6. Estimated particle size of goethite present in the humic-acid and humic-clay fractions, 0- to 10-cm depth.

-16 -{ -2 2 6 10

Velocity (mm/a)Fig. 4. MSssbauer spectra (80 K) of humic acid (HA) from

the 0- to 10-cm depth of soil profiles P2 and P4.

Table 4. Major elemental composition of the humic acid (HA)fraction.

Profile

PIP2

P3P4

Depth

0-100-10

10-2020-4040-601002000-100-10

C

224.0137.098.378.476.175.071.9

180.0220.0

Fe

204209239213246250264202171

AlIT -1

384040434445463022

Si

7.17.13.63.76.74.23.93.57.0

N

202396655

1523

Table 5. Aluminum substitution of goethite in the humic-acidfraction as calculated from chemical analysis (CA), x-raydiffraction (XRD), and Mossbauer spectroscopy at 80 K(MS).

Profile

PIP2

P3P4

Depth

0-100-10

10-2020-4040-60

1002000-100-10

CA

282828272729272321

Al SubstitutionXRDt

—— mol % ——2532

2424

MS*

3238

2832

t Mole % Al = 173.0 - 57.2c (Schulze, 1984), where c is the goethite unitcell c dimension in nm: c = l/(l/d2

nl - l/d2110)"2.

t Magnetic field B^, = 50.4 - 0.13 x mole % Al (Amarasiriwardena etal., 1988), where Bma, is the magnetic field of maximum portability intesla.

diffraction. In each instance, and for each line con-sidered, the goethite in the HA fraction is of smallerparticle size than the goethite in the humin-clay. Av-eraged values indicate a 25 to 40% difference in size.

Profile

Diametert

areafrom

duo

Particle sizex-ray diffraction peak broadening

d,,0 dm d,« it

HAPIP2P3P4

14111413

16131515

12101312

11111111

11101214

12111313

Humin-clayPIP2P3P4

24202826

14141827

14141816

12121818

16152022

t Assuming spherical particles and density of 4.3 g cm-'.t Mean of particle sizes for d,10, d,,0, d,,,, and d140 reflections.

Table 7. Surface area (N2 adsorption) and P adsorption of theHAt and HA^t fractions (0-10-cm depth).

Surface areaSample

PIP2P3P4

HA

<1<1

11

HA*- m2 g-1

109138107119

P adsorbedHA

lunol g-1

985792

103

HA*goethite

483492488484

t HA, humic acid-goethite complex; HAg,, goethite from the HA.

Particle size of the goethite in the HA fraction (cal-culated from external surface area of the HAg, frac-tion) agrees remarkably well with values calculatedfrom XRD. Thus, while the particles may not bespherical, they are nearly equidimensional, in agree-ment with the very similar values of MCDhkl for agiven sample. The goethite in P2 is the smallest ofthe four sources.

Surface AreaSurface areas (N2 adsorption) of the HA and HA^,

fractions are shown in Table 7. The HAgt fraction isbasically crystalline goethite of a very small particlesize, 10- to 15-nm diameter and highly Al substituted(Table 3). These small particles are characteristicallyyellow (10YR 6/8) and correspond to surface areas of>100m2g-1 (Table?).

Surface area is the major factor explaining the ad-sorptive characteristics of the oxyhydroxide compo-nent. Parfitt et al. 91977b) showed that adsorption offulvic acid by imogolite was almost tenfold more thanthe adsorption by gibbsite due to a surface area ap-proaching 100 m2 g-1. Results of Fontes (1988) sug-gest that surface area rather than the kind of Fe oxideis a major factor controlling P adsorption in someBrazilian Oxisols. The external surface area (N2 ad-sorption) of the HA fraction is very small (Table 7),indicating that the HA molecules interact with the soilFe oxides in such a way that most of the externalsurface area is blocked.

Electron MicrographsElectron micrographs support the surface-area find-

ings. Micrographs of the HAgt (Fig. 5a and c) indicate

988 SOIL SCI. SOC. AM. J., VOL. 56, MAY-JUNE 1992

Fig. 5. Electron micrographs of humic acids from the 0- to 10-cm depth (bar represents 50 nm): a. dispersed goethite from thehumic acid-goethite complex of soil P2 (P2HA,.,), b. Flocculated P2HA,,,, c. dispersed goethite from the humic acid-goethitecomplex of soil P3 (P3HAgt), d. flocculated humic acid from soil P2 (P2HAJ, e. dispersed humic acid-goethite complex fromsoil P2 (P2HA), and f. flocculated P2HA.

very small (8-12 nm) nearly equidimensional particlesof quite different morphology than that reported forwell-crystalline synthetic goethites (Schulze andSchwertmann, 1984). These features suggest that dif-ferent dimensions of the crystallite grew at about equalrates from a central nucleus. Micrographs of HAgt dis-persed at pH 7 allow better differentiation of particlemorphology and size than the micrographs of floc-culated samples, as illustrated for P2 HAgt (Fig. 5aand b).

Micrographs of the flocculated HAh fraction (Fig.

5d) suggested a platey appearance quite different thanthe HA fraction (Fig. 5f). The dispersed HA and HAhhave a gelatinous appearance, perhaps due to the higherpH of the suspension (pH = 7); well-defined micro-graphs of the dispersed HAh were not obtained. Themicrographs of the dispersed HA fraction show thevery small, almost equidimensional, oxide crystallitesembedded in the gelatinous organic medium and sug-gest that the crystalline Fe-oxide particles are bondedto a matrix of HA (Fig. 5e). The micrograph of theflocculated HA (Fig. 5f) suggests a large oxide-or-

FONTES ET AL.: GOETHITE AND HUMIC ACID ASSOCIATIONS IN OXISOLS 989

ganic clump surrounded by small, round Fe-oxide par-ticles.

Infrared SpectraThe FTIR spectra of the HAh are shown in Fig. 6.

The overall simplicity of the spectra is more apparentthan real because the broadness of the bands resultsfrom the fact that one is dealing with a complex mix-ture (MacCarthy and Rice, 1985).

Main absorption bands in the region of 1720 cnr1

in the spectra of the HAh (Fig. 6) are due to the C = Ostretching of the COOH groups and, in the 1620 cm"1

region, to aromatic C = C and H-bonded C = C. Otherbands are evident near 3100 cm"1 (H-bonded OHgroups). Bands near 2900 cm"1, attributable to ali-phatic C-H stretching, are weak and vary consider-ably among HA preparations (Stevenson, 1982).

Broad OH-bending bands are observed at 941 to918 and 820 cm"1 in the spectra of the HAgt fraction(Fig. 7). Amarasiriwardena et al. (1988) reported that,in unsubstituted synthetic goethite, these bands occurat 896 and 798 cm"1, although the exact position ap-parently depends on the degree of crystallinity of theoxide, increasing slightly in frequency and decreasingin band width as the crystallinity increases (Schwert-mann et al., 1985). As Al substitutes for Fe in thegoethite structure, both bands move to higher fre-quencies (Jonas and Solymar, 1970; Fey and Dixon,1981; Fysh and Fredericks, 1983; Amarasiriwardenaet al., 1988) and band width increases (Fysh and Fred-ericks, 1983). Values of 918 and 812 cm"1 for syn-thetic 25% Al-substituted goethite were reported byAmarasiriwardena et al. (1988); however, Fysh andFredericks (1983) noted that the relationship betweenband position and Al content depends on the mode ofsynthesis of goethite, presumably related to degree ofcrystallinity or to the inclusion of water moleculeswithin the oxide structure. Schulze and Schwertmann(1984) observed that the frequency difference betweenthe two bands also increases when Al substitution in-creases. They concluded that the shortening of the unitcell along the b and c dimensions due to Al substi-tution causes an increase in the H-bond strength. Thepatterns shown in Fig. 7 exhibit very broad and rel-

atively weak absorption bands corresponding to OH-bending vibrations. The band at 918 to 941 cm"1 ap-pears to be a doublet. On the basis of the frequencydifferences between the two OH bands reported bySchulze and Schwertmann (1984) the band at 918 cm"1

appears to be lower and the band at 941 cm"1 higherthan expected to correspond with the 820 cm"1 bandshown in Fig. 7. Very small particle size and high Alsubstitution of the goethites extracted make interpre-tation of the patterns uncertain. The other major banddue to goethite was a broad absorption at 3100 cm"1

(stretching band for H-bonded OH).A difference spectrum obtained by subtracting the

spectrum of HAgt from the spectrum of HA is com-pared with the spectrum of the HAh in Fig. 8. Themain absorption bands near 1720 and 1620 cm"1 areshifted to lower frequencies in the difference spec-trum, indicating that there is a decrease in the bondstrength and suggesting that the COOH group is inthe salt form. The spectra in Fig. 8 are for soil P2,but the other soils behaved similarly. The band at1400 cm"1 may also be caused by the COO" ion(Stevenson, 1982). These results suggest bonding of

Fig. 7. Expanded Fourier-transform infrared transmittancespectra from the goethite from the humic acid-goethitecomplex of (1) soil P3 (P3HAg,) and (2) soil P4 (P4HAB,).

46 38 18 14 8

Wavenumbers (x 10)Fig. 6. Fourier-transform infrared transmittance spectra of

humic acid (HAJ from the 0- to 10-cm depth of the fourOxisols.

Fig. 8. Expanded Fourier-transform infrared spectra from soilP2: (1) humic acid from the humic acid-goethite complex(HA,,); (2) spectrum of the humic acid-goethite complex (HA)minus the spectrum of the goethite from the humic acid-goethite complex

990 SOIL SCI. SOC. AM. J., VOL. 56, MAY-JUNE 1992

the humic substances to the goethite surface throughthe carboxyl group.

Phosphate AdsorptionPhosphate-adsorption data (Table 7) show clearly

the differential adsorptive characteristics of the HAfraction vs. the HA t fraction. Parfitt et al. (1975) haveshown that P is adsorbed by goethite via ligand ex-change of singly coordinated OH groups. Accordingto Torrent et al. (1990), maximum adsorption of P asa binuclear complex corresponds to 2.77 |xmol P m~2

on the d100 face, 5.96 jjunol P m~2 on d010, and 2.50u,mol P m-2 on dno. The "free" goethite (HAgt) isvery efficient in adsorbing the P added, correspondingto 3.57 to 4.56 |xmol P m-2 (Table 7). This is muchhigher than the value of 2.51 (jimol P m~2 found byTorrent et al. (1990) to represent goethite whose ex-posed surface is mainly dno. The micrographs (Fig.5) do not permit determination of the exposed facesof the goethite particles, but it is evident that morereactive hydroxyls are present than can be accountedfor in better crystallized goethites.

In the HA fraction, most of the adsorption sites areapparently blocked and much less P is retained, i.e.,it does not displace the humic compounds. Steric ef-fects may also prevent some replacement in the com-plex. The results indicate strong bonding of the humicacid to the goethite surface, whereby its COOH groupsdisplace surface OH groups. Further analysis of thedata in Table 7 suggests a somewhat different behav-iour of P2, i.e., a smaller value for P adsorption withinHA and a slightly higher value within the HAgt. Thegoethite from P2 has the higher external surface area(Table 7) and smaller particle diameter (Table 6),compared with the goethites from PI, P3, and P4.The net result is that P adsorption on P2 is less thanfor the other three materials.

CONCLUSIONSMicrocrystalline goethite was the only Fe oxide found

to be associated with the HA in the soil materialsstudied. The bonding between the organic and inor-ganic phases was sufficiently strong to modify theproperties of each and may explain the relatively highorganic-C contents extending to depths exceeding 1 min these soils.