DIVISION S-9-SOIL MINERALOGY

Phyllosilicate Distribution and Origin in Aridisols on a Granitic Pediment,Western Mojave Desert

J. L. Boettinger* and R. J. Southard

ABSTRACTThere is considerable uncertainty about the extent of mineral weath-

ering and neosynthesis in arid soils. Although researchers have specu-lated that smectite neosynthesis can occur in soils containing opalinesilica cement, no evidence has been presented to support this hypothe-sis. We investigated the mineralogy of two Aridisols in the westernMojave Desert, California, to study phyllosilicate distribution andorigin in soils with opaline silica. The clay fraction of both soilsis dominated by Al-rich, dioctahedral smectite, characterized by aMg-saturated d(001) spacing of 1.52 nm and a d(060) spacing of 0.149to 0.150 nm. This smectite is also present in silt and sand fractionsof deeper horizons, where it exists mainly as microagglomerates ofclay-sized crystals. Biotite is most abundant in silt and sand fractionsof near-surface horizons of both pedons, where physical weatheringis greatest. Deep, alkaline, silica-rich horizons of both pedons containmore silt- and sand-sized vermiculite than biotite, probably due torapid chemical weathering of biotite to trioctahedral vermiculite. Clay-and silt-sized hydroxy-interlayered 2:1 minerals are present in theupper horizons of these soils. Given the lack of gibbsite and substantialamounts of kaolinite and Al hydroxy-interlayered 2:1 phyllosilicates,we propose that neosynthetic dioctahedral smectite was the dominantsink for Al released by feldspar weathering in these Aridisols. Silicanot consumed in smectite neosynthesis can cement microagglomerates.We further speculate that minimal physical weathering in the alkaline,high-silica environment of deep horizons favors aggregation of smec-tites to silt- and sand-sized microagglomerates.

SMECTITES are the dominant phyllosilicate clays in soilsthat contain opaline silica cement in arid and semiarid

climates (Flach et al., 1974; Torrent et al., 1980; Blank,1987; Eghbal and Southard, 1993a). Apparently, thesilica cement in these duric soils contributes to the stabil-ity of smectite. Smectite may be inherited from the soilparent materials (Blank, 1987; Borchardt, 1989), formedby weathering of a 2:1 lattice precursor (Borchardt,1989; Fanning etal., 1989), or pedogenically precipitatedfrom solution (Kittrick, 1969). In pedogenic environ-ments where weathering products are not fully leachedfrom the solum, silica cement can be derived from theweathering of volcanic glass (Chadwick et al., 1989)and primary crystalline silicates (Blank and Fosberg,1991; Boettinger and Southard, 1991). Although it hasbeen proposed that neosynthesis of smectite can occur induric soils (Flach et al., 1974; Boettinger and Southard,

J.L. Boettinger, Dep. of Plants, Soils, and Biometeorology, Utah StateUniv., Logan, UT 84322-4820; and R.J. Southard, Soils and Bio-geochemistry, Dep. of Land, Air and Water Resources, Univ. of Califor-nia, Davis, CA 95616. This research was supported in part by the UtahAgric. Exp. Stn., Utah State Univ., Logan, UT 84322-4810. Approvedas journal paper no. 4713. Received 11 Mar. 1994. *Corresponding author(jlboett@cc.usu.edu).

Published in Soil Sci. Soc. Am. J. 59:1189-1198 (1995).

1991) and that silica cementation arises from excesssilica not consumed during the neogenesis of smectite(Southard et al., 1990), no evidence has been presentedto support or refute the formation of neosynthetic smec-tites in duric soils.

Reports of silt- and sand-sized smectites are rare;smectites are usually concentrated in the fine clay fraction(Borchardt, 1989). There is Mg-rich, tetrahedrally sub-stituted smectite, apparently weathered from saponite,in the sand and silt fractions of soils derived from hydro-thermally altered basalt in Scotland (Curtin and Smillie,1981). Inherited saponite occurs throughout sand andsilt fractions of Entisols formed from hydrothermallyaltered gabbro and anorthosite in southern California(Graham et al., 1988). High-charge beidellite was foundin the 2- to 8-M.m fraction of alluvial soils in Spain(Aragoneses and Garcia-Gonzales, 1991). This smectite,which occurs as silt-sized aggregates and as grains thatwere not dispersed when treated with Na2CO3, apparentlyweathered from illite inherited from shale. Eghbal andSouthard (1993a) concluded that smectite in the silt frac-tion of soils on alluvial fans in the western Mojave Desertconsists of clay-sized smectite cemented into aggregatesby noncrystalline and short-range-ordered silica and alu-minosilicates.

Aridisols containing opaline silica cement occur ongranitic pediments of the western Mojave Desert ofsouthern California (Valverde and Hill, 1981). The Sifor silica cement and Ca for carbonates are deriveddominantly from weathered primary silicates in the gra-nitic saprolite parent material (Boettinger and Southard,1991). We documented the distribution and major char-acteristics of phyllosilicates in a Haplodurid and a Haplargidforming on the granitic pediment, including the occurrenceof silt-sized smectite. We propose that the dioctahedralsmectite in the clay fraction and in the microagglomeratesof these duric soils is neosynthetic, serving as the sinkfor Al released in the dissolution of feldspars, biotite,and amphiboles.

MATERIALS AND METHODSStudy Site

The study site is on a low granitic pediment of the westernMojave Desert, about 100 km north of Los Angeles and 17km south of Mojave, CA (34°53'N, 118°10'W; Fig. 1). Theclimate in this area is arid and warm. Precipitation is «130mm, concentrated during the winter months (Dibblee, 1963),and mean annual air temperature is 18°C (Valverde and Hill,1981). Dominant vegetation on the pediment is creosote bush[Larrea tridentata (Sesse & Mocino ex DC.) Cov.], spiny hopsage [Grayia spinosa (Hook) Moq.], and annual grasses and

1189

1190 SOIL SCI. SOC. AM. J. , VOL. 59, JULY-AUGUST 1995

Fig. 1. Location of the study area in southern California.

forbs (desert needlegrass [Achnatherum speciosum (Trin. andRupr.) Barkworth = Stipa speciosa], red brome [Bromusmadritensis L. ssp. rubens (L.) Husnot. = Bromus rubens],and redstem filaree [Erodium cicutarium (L.) L'Her.]). Thepediment is composed of medium- to coarse-grain holocrystal-line Mesozoic granitic rock of the Rosamond Hills, whichvaries locally from quartz monzonite to granite and granodiorite(Dibblee, 1963). The granitic rock of this pediment is weath-ered in most places to saprolite. The pediment landform isrepresentative of the pediment-fan landscapes of the westernMojave Desert (Valverde and Hill, 1981). The age of thepediment surface is unknown.

Field MethodsTwo pedons, 70 m apart, were exposed, described, and

sampled by genetic horizon in hand-excavated pits on a south-

east-facing pediment (2% slope). Horizons below 100 cm weresampled by auger. Soils were classified according to SoilSurvey Staff (1994). One pedon, a loamy, mixed, thermic,shallow Typic Haplodurid (formerly classified as a TypicDurorthid) is calcareous throughout and has a calcareous duri-pan with an indurated laminar cap at 28 cm (Table 1). TheHaplodurid pedon is considered modal for the Muroc series;the study site is the type location for the Muroc series (Valverdeand Hill, 1981). The other pedon, a fine-loamy, mixed, thermicTypic Haplargid, is deep, has an argillic horizon beginningat 13 cm, and is calcareous below 80 cm (Table 1). TheHaplargid pedon is a taxadjunct to the HiVista series, whichis noncalcareous throughout and has a lithic contact between50 and 100 cm.

The dominant parent material for the pedons is deeplyweathered and coherent granitic saprolite, which is calcareousin parts, depending on local variations in initial composition(Boettinger and Southard, 1991). The original rock fabric wasevident below 47 cm in the Haplodurid and below 35 cm in theHaplargid. Granitic sediment derived from upslope probablycontributed to the soil horizons above the duripan of the Haplo-durid and above the Bt2 horizon of the Haplargid.

Laboratory MethodsAir-dried soil samples were crushed gently with a mortar

and pestle and sieved to remove rock fragments >2 mm. The<2-mm soil from all nonindurated horizons were treated with1 M NaOAc buffered to pH 5.0 to remove carbonates andsoluble salts, if present (Jackson, 1979). Organic matter wasdestroyed with 50 g L~' NaOCl buffered to pH 9.5 (Lavkulichand Weins, 1970). Free Fe oxides were removed with sodiumcitrate-dithionite-bicarbonate (Jackson, 1979). Sand (0.05-2 mm) was separated by wet sieving. With dilute

Table 1. Selected morphological, physical, and chemical properties of the Haplodurid and Haplargid pedons.

Horizon Depth

cm

Munsellcolor

moist

Clay Siltwt %V» I. i\>

Sand Gravelt

vol.%PH CaCO3 SiOrf

g kg-' soil-

A1203* m/AM4gkg-'/g kg-'

HaploduridAlA2A3Bqkml§Bqkm2§Bqkm3§BqkmttCrkttBkqBCkCBklCBk2CBk3

AlA2BtlBt2Bt3BklBk2Bk3Bk4BCklBCk2

0-44-14

14-2828-2929-3131-4747-7547-7575-101

101-127127-145145-155155-160

0-77-13

13-2424-3535-8080-100

100-115115-125125-135135-145145-180

10YR 3/310YR 4/310YR 4/410YR 6/310YR 8/310YR 8/310YR 8/2variegated7.5YR 7/27.5YR 7/27.5YR 7/27.5YR 7/27.5YR 7/2

10YR 3.5/410YR 4/47.5YR 4/47.5YR 4/47.5YR 4/47.5YR 6/67.5YR 5/67.5YR 5/410YR 6/410YR 6/410YR 6/4

12.912.715.1-1——

10.40.97.14.95.35.14.8

8.611.418.719.816.914.514.518.29.0

10.64.9

21.221.424.6-——

12.23.39.58.78.09.87.4

21.819.822.521.88.1

21.115.021.720.224.828.9

65.965.960.3-——

77.495.883.486.386.785.087.9

Haplar69.768.958.458.475.064.570.560.170.864.666.2

151515-——15501522303030

gid1510122030151515151515

7.58.08.1-——

8.18.38.28.68.48.28.3

7.57.57.57.57.58.07.97.98.08.08.0

221041

468713513204

1525164161811

0000054

12169

15

10.813.011.951.390.873.334.37.0

29.710.86.06.25.5

12.816.116.318.616.526.429.142.332.028.324.0

3.03.93.93.80.3#0.6.0.3.6.9.1.2.0

4.66.97.17.96.93.03.74.43.62.92.9

3.63.43.2

13.7338.3#119.333.95.2

46.912.15.75.25.3

2.82.32.32.32.48.77.99.79.09.98.4

t Field estimated volume.t Soluble in boiling 0.5 M NaOH.§ Duripan separated into sublayers based on morphology.1 Analysis not performed.tt Al measured at lowest limit of detection by atomic absorption spectroscopy.tt Portions of Crk/Bqkm horizon analyzed separately.

BOETTINGER AND SOUTHARD: PHYLLOSILICATES IN ARIDISOLS ON GRANITIC PEDIMENT 1191

coarse silt (20-50 urn) was separated by sedimentation, andfine plus medium silt (2-20 urn) was separated from clay (<2urn) by centrifugation. Soil separates >2 u,m were ground in anagate mortar and pestle before preparation for x-ray diffraction(XRD) analysis. Preferred oriented aggregates with Mg andK saturation and glycerol solvation treatments were preparedfor XRD analysis using porous ceramic tiles (Whittig andAllardice, 1986). Random powder mounts were back-loadedinto the 6-mm window of an Al slide using a method similarto that described by Moore and Reynolds (1989). Sampleswere continuously scanned with Cu Ka radiation generated at50 kV and 15 inA in a Diano 8000 x-ray diffractometer (DianoCorp., Woburn, MA). Time constants during scans were 2.5 sfor oriented specimens and 5 s for random powder mounts.The <2-um fraction was further analyzed using a modifiedGreene-Kelley test of Li saturation and overnight heating at250°C, solvation with glycerol at 90°C for 16 h, and XRDanalysis (Lim and Jackson, 1986). Sand (0.05-2 mm) andcoarse-silt (20-50 u,m) grains from selected horizons wereimmersed in oil (n = 1.544) and examined with a polarizinglight petrographic microscope. Selected soil fractions wereboiled in 0.5 M NaOH for 2.5 min (Jackson et al., 1986) toremove poorly crystalline silica and aluminosilicate cementsand reexamined with XRD analysis and optical microscopy.

Particle-size distribution was determined on <2-mm samplesfrom noncemented horizons according to Day (1965) afterpretreatment with 1 M NaOAc (pH 5.0) and 5% NaOCl (pH9.5) as described above. The pH was determined on saturatedpastes of <2-mm soil (Soil Conservation Service, 1972). Cal-cium carbonate equivalent (Williams, 1948) was determinedon samples crushed to <180-nm. Amorphous and poorly crys-talline silicates and aluminosilicates were extracted from < 180-(im, Fe-oxide-free (Jackson, 1979) samples by boiling in0.5 M NaOH for 2.5 min (Jackson et al., 1986); Si and Alwere measured with atomic absorption spectroscopy.

RESULTSClay Fraction

The dominant phyllosilicate in the <2-|J.m fractionthroughout both pedons was smectite (Table 2, Fig. 2aand 2b). Vermiculite, mica, hydroxy-interlayered 2:1phyllosilicates, and kaolinite were present in minor totrace amounts in the upper horizons of these pedons.Trace amounts of vermiculite occurred below the Bt2horizons of the Haplargid pedon, whereas smectite wasthe only phyllosilicate detected by XRD below the duri-pan of the Haplodurid. The Mg-saturated basal d-spacingof the smectite throughout the Haplodurid and in thehorizons underlying the Bt2 in the Haplargid pedon was1.52 nm (Fig. 2a). In contrast, the smectite dominatingthe upper horizons of the Haplargid pedons had a 1.42-nmbasal d-spacing with Mg saturation and some hydroxypolymers in the interlayer, indicated by complete collapseto 1.0 nm with K saturation only when heated to 550°C(Fig. 2b). In both pedons, the 1.84-nm Mg-saturatedand glycerol-solvated diffraction maximum of smectiteincreased greatly in sharpness and intensity with increas-ing depth and was especially intense in horizons whereNaOH-extractable SiOz/AhOs is >5 (Table 1).

Aluminous dioctahedral phyllosilicates have d(060)values of < 0.150 nm, whereas trioctahedral species haved(060) spacings of 0.152 to 0.1541 nm and dioctahedralnontronite has d(060) = 0.152 (Bailey, 1980; Brindley,1980). Throughout both pedons, the <2-nm fractionexhibited an intense d(060) peak at < 0.150 nm. Theintensity of this peak increased with depth as the spacingdecreased from 0.1496 to 0.1491 nm (Fig. 3a and b).

Table 2. Relative abundance of phyllosilicates in the clay, silt, and sand fractions of the Haplodurid and Haplargid pedons.t

Horizon

AlA2A3Bqkm§BqkmlCrk1BkqBCkCBklCBk2CBk3

AlA2BtlBt2Bt3BklBk2Bk3Bk4BCklBCk2

Major

SmSmSm-

SmSmSmSmSmSmSm

SmSmSmSmSmSmSmSmSmSmSm

<2 um

Minor

V,M,HI,KV,M,HI,KV,M,HI,K

-—-———--

V,HI,M,KV,HI,M,KV,HI,M,KV,HI,M,K

VVVVVVV

Major

M,VVV-

SmSm,VSm,VSmSmSmSm

M,VM,VV,MV,M

VV,SmSm,V

SmSmSmSm

2-20 um

Minor

HaploduridSm,HI

Sm,M,HISm,M,HI,IS

-V,IS,MHI.IS

HI,IS,MV,HI,IS,MV,HI,IS,M

V.IS.MV.IS.MHaplargid

Sm,HI,LHl.SmHI,Sm

HI,Sm,LM,Sm,HI,L

IS,MIS,M,L

V,IS,M,LV,M,LV,L,MV,L,M

20-50 |im

Major

M,VV,MV,M-

V,SmV,MV,SmV,SmSm,VSm,VSm,V

M,VM,VV,MV,M

VV

V,SmV,SmSm,V

Sm,V,LSm,V,L

Minor

SmSmSm-

M,ISIS.SmM,ISM,ISM,ISM,ISM,IS

SmSmSmSm

Sm,MSm,MIS,L,MIS,L,M

L,MMM

Major

M——-—-

V,M—V--

M——M-—

V,M—VV—

Sand}:

Minor

V——-—-

Sm,IS—

M,Sm,IS--

V——

V,IS-—

Sm,IS—

Sm,M,IS,LSm,M,IS,L

—t Sm = smectite, V = vermiculite, M = mica, HI = hydroxy-interlayered 2:1 phyllosilicate, IS = interstratifled mica-vermiculite, K = kaolinite, L =

laumontite (zeolite).t Only selected horizons were analyzed.§ Not analyzed because not fractionated into particle-size separates.1 Portions of Crk/Bqkm horizon analyzed separately.

1192 SOIL SCI. SOC. AM. J., VOL. 59, JULY-AUGUST 1995

a) Haplodurid CBkl b) Haplargid A1

1.521.0 nm j

,1.24; ;

K(25°C):

Mg-glycerol

0.71

10

°29Fig. 2. X-ray diffractograms (Cu Ka radiation) of the <2-\im fraction

of (a) the CBkl horizon of the Haplodurid (127-145 cm), and (b)the Al horizon of the Haplargid (0-7 cm). The diffractogram ofthe Haplargid Al is shown at a smaller intensity scale than thediffractogram of the Haplodurid CBkl. Treatments from top tobottom are K saturated and heated to 550°C; K saturated; Mgsaturated and glycerol solvated; and Mg saturated.

The < 0.150-nm peak intensity in these pedons coincidedwith the relative abundance of smectite, thus suggestingthat the smectite is dioctahedral, with mostly Al in theoctahedral sheet. A diffraction maximum centered at0.1531 nm occurred in the upper horizons of both pedons,where vermiculite and mica were detected. Trioctahedralvermiculite and biotite probably gave rise to the 0.153-nmpeak.

The modified Greene-Kelley test (Li saturation, heat-ing, and glycerol solvation) produced a dominant peakat = 1.0 nm, with only a minor peak at = 1.4 nm (diffrac-togram not shown), suggesting that the Al-rich smectiteis octahedrally substituted and is mostly montmorillonite,rather than beidellite or nontronite (Lim and Jackson,1986).

Fine Plus Medium Silt FractionGlycerol solvation of Mg-saturated samples revealed

smectite (1.84 nm), vermiculite (1.42 nm), and mica(1.0 nm) in the 2- to 20-nm fraction of both pedons(Fig. 4, Table 2). Mica and vermiculite dominated theupper horizons, whereas smectite dominated the lowerhorizons. Mica and vermiculite were more abundant inthe noncalcareous upper horizons of the Haplargid thanin the calcareous upper horizons of the Haplodurid. Theabundance of vermiculite relative to mica increased with

a) Haplodurid b) Haplargid0.149nm 0.149nm

A1

CBkl

62

Fig. 3. X-ray diffractogram (Cu Ka radiation) showing the d(060)lines of <2-um fraction of representative horizons of the (a) Haplo-durid and (b) Haplargid.

depth; mica was present only in trace amounts in horizonsbelow the duripan of the Haplodurid and below the Bt3horizon of the Haplargid. As in the clay fraction ofthese pedons, the intensity and sharpness of the smectitediffraction maxima in the 2- to 20-nm fraction generallyincreased with depth (Fig. 4). In the deepest horizons,the Mg-saturated 2- to 20-|im samples gave rise to adouble XRD peak at 1.42 and 1.52 nm. With glycerolsolvation, the doublet separated cleanly into the 1.42-nmpeak, probably due to vermiculite and hydroxy-interlayered 2:1 phyllosilicates, and the 1.84-nm peakof smectite (diffractogram not shown).

In addition to smectite, vermiculite, and mica, minoramounts of 2:1 phyllosilicates interlayered with hydroxypolymers are present in horizons above and in a fewhorizons below the duripan of the Haplodurid and inhorizons above the Bt3 horizon of the Haplargid (Table2). The hydroxy-interlayered phyllosilicates retainedXRD peaks at 1.42 nm with K saturation and eithercollapsed to a broad shelf centered « 1.24 nm or retaineda 1.4-nm peak with heating to 550°C (Fig. 5). Thedegree of collapse with heating reflected the extent towhich 2:1 phyllosilicate interlayers are filled with hy-droxy polymers (Barnhisel and Bertsch, 1989). Thereare interstratified minerals in the lower parts of thesepedons (Table 2). In the horizons below the A2 of theHaplodurid and in the Bkl, Bk2, and Bk3 horizons of thecalcareous Haplargid, there was evidence of a regularlyinterstratified mica-vermiculite (hydrobiotite): 1.24- and

BOETTINGER AND SOUTHARD: PHYLLOSILICATES IN ARIDISOLS ON GRANITIC PEDIMENT 1193

a) Haplodurid b) Haplargid1.84

1.421.0

0.71

A1

CBk1

Fig. 4. X-ray diffractograms (Cu Ka radiation) showing depth trendsof Mg-saturated and glycerol-solvated 2- to 20-um fractions fromrepresentative horizons in the (a) Haplodurid and (b) Haplargid.

2.45-nm peaks occurred with Mg saturation and glycerolsolvation (e.g., Fig. 5a) but apparently collapsed to 1.0nm with heating to 550°C. In the horizons below theduripan of the Haplodurid (diffractograms not shown),a mineral phase that collapsed to ~ 1.24 nm with Ksaturation and to 1.18 nm with heating to 550°C wasinterpreted as a vermiculite interstratified with a chlorite-like layer that underwent dehydroxylation with heating.The 2- to 20-nm fraction of horizons above the duripanof the Haplodurid and above the Bkl horizon of theHaplargid showed trace amounts of amphibole (0.84 nm;e.g., Fig. 5a and b).

In addition to the phyllosilicates, laumontite, a Ca-richzeolite (CaAkSuO^ • 4H2O), was identified in the Haplar-gid pedon (e.g., Fig. 4b and 5b). Laumontite occurredin the 2- to 20-nm fraction of the A, Bt2, Bt3, Bk2,Bk3, Bk4, BCkl, and BCk2 horizons of the Haplargid(Table 2). Diffraction maxima at 0.95 and 0.686 nmcorrespond to the d(110) and d(200) spacings of thepartially and reversibly dehydrated form of laumontite,leonhardite (Coombs, 1952). With heating to 550°C,the 0.95 and 0.686 XRD peaks disappeared and newpeaks at 0.86 and 0.55 nm appeared. The new peakscorrespond to peaks that Taylor et al. (1990) attributed tothe high-temperature dehydration products of laumontite.Laumontite, a product of weak hydrothermal alterationof plagioclase, also occurs in soils derived from plutonicrocks in the San Gabriel Mountains, south of the studyarea (Graham et al., 1988; Taylor et al., 1990).

In both pedons, the d(060) lines for the 2- to 20-(xmfraction followed the same pattern with depth as the<2-|j,m fraction (Fig. 6). As smectite became more domi-

a) Haplodurid A3 b) Haplargid Bt21-0nm 1.0 nm

0.71

K(550°C):

K(25°C);

Mg-glycerol

Mg(25°C)

10

Fig. 5. X-ray diffractograms (Cu Ka radiation) of the 2- to 20-umfraction of (a) the A3 horizon of the Haplodurid (14-28 cm), and(b) the Bt2 horizon of the Haplargid (24-35 cm). Treatments fromtop to bottom are K saturated and heated to 550°C; K saturated;Mg saturated and glycerol solvated; and Mg saturated.

nant with depth, the 0.150-nm peak increased in intensityand shifted slightly toward 0.149 nm. Based on thecoincidence of the smectite with the <0.150-nm peakin the deeper horizons of both pedons, the smectite hithe 2- to 20- nm fraction is probably dominantly dioctahe-dral and the octahedral sheet is Al-rich. The 0.154-nmd(060) spacing decreased in intensity with depth in theHaplodurid and Haplargid pedons, except in some deeperhorizons of the Haplodurid where quartz [d(211) = 0.154nm] was present.

Coarse Silt FractionCoarse silt-sized smectite occurred in the lower hori-

zons of the Haplodurid and Haplargid pedons (Table 2,Fig. 7). In both the Haplodurid and Haplargid, onlytrace amounts of smectite were detected by XRD analysisin the surface horizons, whereas only trace amounts ofmica were detected in the lower horizons. As in the finerfractions, the intensity and sharpness of the 20- to 50-nmsmectite diffraction maxima increased with depth, as therelative abundance of mica decreased. Smectite is mostabundant in horizons with SiO2/Al2O3 >5 (Table 1).Minor interstratified mica-vermiculite was also presentin horizons at mid-depth in both pedons (see broad 1.2-nmpeaks in Haplodurid Bkq and CBkl and Haplargid Bk2in Fig. 7). There appeared to be less hydroxy interlay-ering of 2:1 minerals in this fraction relative to finerfractions, with more complete collapse to 1.0 nm

1194 SOIL SCI. SOC. AM. J., VOL. 59, JULY-AUGUST 1995

a) Haplodurid b) Haplargid0.154nm

CBkt

A1

Bt2

Bk2

BCk1

62 58

°26Fig. 6. X-ray diffractograms (Cu Ka radiation) showing the d(060)

lines of 2- to 20-um fractions of representative horizons of the (a)Haplodurid and (b) Haplargid.

achieved with heating to 550°C (diffractogram notshown). Notable amounts of laumontite were present inthe coarse silt fraction of the Haplargid BCkl and BCk2horizons, and there were trace amounts in the Bk2, Bk3,and Bk4 horizons (Table 2, Fig. 7b). The coarse siltfractions of the upper horizons of both pedons containedamphibole, and all horizons contained coarse silt-sizedplagioclase (0.64 nm). The occurrence of smectite inthe coarse silt fraction corresponded with 0.150-nm peaksof the (060) phyllosilicate XRD lines (Fig. 8). The 0.150-nm spacing indicates that a portion of the smectite isdioctahedral and Al-rich. The 0.154-nm peaks were prob-ably due to trioctahedral mica and vermiculite and quartz.

The presence of smectite in fractions >2 urn maybe due to aggregation by short-range-order silica oraluminosilicate cements. Subsamples of selected 2- to20- and 20- to 50-um fractions from the Haplodurid andHaplargid that were boiled in 0.5 M NaOH for 2.5 minand refractionated by sedimentation showed only minorreductions in XRD peak intensities (Fig. 9). Opticalpetrographic analysis of subsamples of both NaOH-treated and untreated 20- to 50-um fractions of the BCklhorizon revealed clear, colorless grains with layer-silicate-like basal cleavage and low positive relief. Thesegrains must either be vermiculite or smectite, given theXRD data for this horizon (Fig. 7 and 9). Under crossedpolarized light, the grains exhibited near-uniform extinc-tion and first-order gray to white interference colors.The biaxial negative mineral showed a 2V of about 20°.The 2V of this mineral is larger than that reported fora trioctahedral vermiculite inherited from biotite orphlogopite (0-8°) but is in the range (0-30°) for an

a) Haplodurid b) Haplargid0.14

1.0 nm0.636

CBk1

Fig. 7. X-ray diffractograms (Cu Ka radiation) showing depth trendsof Mg-saturated and glycerol-solvated 20- to 50-um fractions fromrepresentative horizons in the (a) Haplodurid and (b) Haplargid.

aluminous dioctahedral smectite, montmorillonite orbeidellite (Phillips and Griffen, 1981). Although thesegrains appeared to be silt-sized crystals of aluminous,dioctahedral smectite, the average crystallite thicknesscalculated from d(001) and d(060) XRD peak widthsusing the Scherrer equation was only 13 nm. Silt-sizedsmectite in these Aridisols probably occurs mostly asaggregates that resist dispersion after boiling hi 0.5 MNaOH. The less likely existence of silt-sized, single-crystal smectite would have to be confirmed with electrondiffraction analysis, which was beyond the scope of thisstudy.

Sand FractionsThe sand fractions of the Al, Bkq, and CBkl horizons

of the Haplodurid and of the Al, Bt2, Bk2, Bk4, andBCkl horizons of the Haplargid were qualitatively ana-lyzed with optical microscopy and XRD analysis. Allhorizons were rich hi quartz, K-feldspar, plagioclase,and biotite and/or vermiculite and smectite, dependingon the soil depth. At greater depths, the amounts ofvermiculite and smectite increased relative to the amountof biotite (Table 2). Biotite was the only mica detectedin these Aridisols. Primary chlorite was not identified. Theplagioclase was dominandy oligoclase. The K-feldspar wasmostly microcline, with some orthoclase. Laumontitewas common in fractions up to medium sand in thedeepest horizons of the Haplargid. Hornblende is presentin trace amounts in all horizons examined. All horizonsexamined in the Haplargid contained trace epidote. TheAl and CBkl horizons of the Haplodurid contained trace

BOETTINGER AND SOUTHARD: PHYLLOSILICATES IN ARIDISOLS ON GRANITIC PEDIMENT

1.84

1195

a) Haplodurid b) Haplargid0.154 0.154

0.150 nm I

0.95 nm 1.42

CBk1

°20Fig. 8. X-ray diffractograms (Cu Ka radiation) showing the d(060)

lines of 20- to 50-um fractions of representative horizons of the (a)Haplodurid and (b) Haplargid.

amounts of a sodic amphibole. Smectite and vermiculitewere detected with XRD analysis in the sand fractionof the deepest subsoil horizons of both pedons. Weinitially speculated that smectite occurred in the sandfraction as discrete crystals, as direct transformationproducts on the edges of biotite/vermiculite grains, oras cemented aggregates, like the microagglomerates re-ported by Chadwick et al. (1989) or the turbid grainsreported by Eghbal and Southard (1993a,b).

Whitish, turbid, sand-sized microagglomerate grainswere hand picked from the silica-containing horizons ofthe Haplodurid and Haplargid. These microagglomeratesare either silica-rich nodules composed of clay- andsilt-sized grains cemented by opaline silica, or weatheredprimary silicate grains replaced by opaline silica andcalcite (Boettinger and Southard, 1990, 1991). X-raydiffraction analysis of these microagglomerates revealedthat the only phyllosilicate present was smectite with adiffraction pattern identical to the smectite in the clayfraction (Fig. 2a). The XRD analysis of sand fractionscontaining these grains had d(060) spacings of < 0.150nm, whereas near-surface horizons that lack these mi-croagglomerates did not have distinct peaks near 0.150nm. These microagglomerates, like those reported byEghbal and Southard (1993a), were completely removedfrom the sand fraction after boiling in 0.5 M NaOH for2.5 min. These turbid grains were composed of smectitemineral grains cemented by opaline silica.

Brown grains with the morphology of biotite werehand selected from the medium sand fraction of the BCklhorizon of the Haplargid and analyzed with XRD analysisto assess the direct weathering products of the biotite in

Fig. 9. X-ray diffractogram (Cu Ka radiation) of the Mg-saturatedand glycerol-solvated 20- to 50-urn fraction of the BCkl of theHaplargid (a) before and (b) after boiling for 2.5 min in 0.5 MNaOH and refractionation.

the parent material. The grains were composed domi-nantly of vermiculite with small amounts of biotite,regularly interstratified biotite-vermiculite and smectitewith a Mg-saturated d-spacing of 1.42 nm (diffractogramnot shown). Apparently biotite, derived from the graniticparent material, weathered almost completely to vermic-ulite via regularly interstratified biotite-vermiculite. Thesmectite is probably a direct simple transformation prod-uct of biotite weathering and may only occur at theweathered edges of, or as sand-sized sheets within, thesemacroscopic biotite-derived grains.

DISCUSSIONMica, Vermiculite, and Hydroxy-Interlayered

PhyllosilicatesBiotite in sand fractions was apparently inherited from

the granitic rock-derived parent material of these Aridisolpedons. Mica appeared in the clay and silt fractions ofthe surface and near-surface horizons of both pedons,but its relative abundance in these fractions decreasedwith increasing depth, which is opposite to the trendreported for most mica-containing soils with homoge-neous parent materials due to near-surface weatheringand lessivage (Fanning et al., 1989).

Several processes may account for the decrease inmica with depth in these duric soils. Mica in the clayand silt fractions of near-surface horizons may be ofeolian origin. The potential contribution of phyllosilicatesto arid soils from eolian deposition has been well docu-mented in the southwestern USA (Gile and Grossman,1979; McFaddenetal., 1986). However, eolian contribu-tions of carbonates and silica are apparently minimalcompared with in situ sources hi these Aridisols (Boet-tinger and Southard, 1991) and in other Aridisols in the

1196 SOIL SCI. SOC. AM. J., VOL. 59, JULY-AUGUST 1995

western Mojave Desert (Eghbal and Southard, 1993c);therefore, eolian mica contributions are probably mini-mal. We propose that the most likely explanation forthe surface accumulation of clay- and silt-sized mica isthe enhanced physical weathering of biotite grains innear-surface horizons due to large daily and seasonalfluctuations in temperature and moisture and to limitedlessivage in the aridic and thermic climate. Temperatureand moisture fluctuations are increasingly dampened withdepth, thus reducing the potential for physical weatheringof biotite to silt- and clay-sized particles. Limited physicalweathering and pedoturbation in deeper horizons wasclearly evident in the preservation of much of the originalrock structure below 47 cm in the Haplodurid and below35 cm in the Haplargid.

Mica in these Aridisols apparently weathered to ver-miculite via direct simple transformation, which is thegenerally accepted mode of trioctahedral vermiculite for-mation in soils (Douglas, 1989). Evidence for directtransformation of biotite to vermiculite was the domi-nance of vermiculite relative to mica in the sand fractionof deeper horizons and the association of vermiculite withinterstratified mica-vermiculite in sand and silt fractions.The occurrence of vermiculite in the coarser silt andsand fractions of deeper horizons indicated that biotitechemically weathered to vermiculite, but the absence ofbiotite and vermiculite in the clay fraction suggestedthat neither biotite nor vermiculite had been extensivelyphysically weathered into clay-sized particles. A similaroccurrence of macroscopic vermiculite was reported andexplained by Alexiades et al. (1973). Trioctahedral ver-miculite is considered to be an unstable intermediatephase, particularly in acid soils (Kittrick, 1973; Douglas,1989), but its occurrence and stability in calcareous andsilica-rich soil has not been documented. The persistenceof sand- and silt-sized vermiculite in these duric soilsmay reflect a relatively rapid removal of K from theinterlayer and hydration of biotite to vermiculite vs. theslower chemical alteration of vermiculite to smectite.Protonation and layer charge reduction of vermiculitemay be slow hi this limited-leaching, alkaline, silica-richenvironment.

Hydroxy-interlayered 2:1 minerals and a 1.4-nm min-eral that did not collapse with heating to 550°C werepresent in silt fractions of near-surface horizons of bothpedons, and an interstratified vermiculite-chlorite oc-curred in silt fractions of horizons below the duripan ofthe Haplodurid. The lack of primary chlorite in sandfractions of these soils means that the chlorite-like min-eral was probably not chlorite inherited from the parentmaterial and that the interstratified vermiculite-chloritepresent in finer fractions was not a chlorite weatheringproduct. The 1.4-nm peaks that remained after heatingto 550 °C were probably due to extensively hydroxy-interlayered 2:1 layer silicates that are pedogenic. Hy-droxy-interlayered minerals may result from the incorpo-ration of Fe, Mg, or Al hydroxy polymers in theinterlayer of vermiculite and smectite (Rich, 1968). TheAl hydroxy polymers are most common, especially insoils where wetting and drying cycles are frequent, or-ganic matter is low, and pH ranges from 4.6 to 5.8

(Rich, 1968). Rich (1968) proposed that Mg(OH)2 shouldbe the dominant interlayer group in alkaline soils, butthe interlayer composition of several chloritic intergradesof alkaline soils have not been identified (Whittig, 1959;Jackson, 1962; Klages and Southard, 1968). Consideringthat Mg(OH)2 was precipitated in the interlayer spacesof Wyoming bentonite at alkaline pH in laboratory experi-ments (Levy et al., 1983) and that dissolution of primarybiotite and amphibole can provide a source of Mg, thehydroxy-interlayer polymers in these soils may be Mg-rich.

SmectiteSmectite is the dominant phyllosilicate in the clay

and silt fractions of these Aridisols, and the relativeabundance and crystallinity of smectite increases withsoil depth. The lack of smectite in the silt fraction ofnear-surface horizons probably means that there wasminimal eolian input of silt-sized smectite to these soils.Although smectite-containing dust may have made somecontributions to these soils, the carbonate and opalinesilica distributions across the pediment indicate that eo-lian deposition, which would include smectite deposition,was probably not as important as in situ mineral weather-ing (Boettinger and Southard, 1991).

The majority of clay- and silt-sized smectite and thesmectite in sand-sized microagglomerates is apparentlyoctahedrally substituted with an Al-rich dioctahedralsheet, as was evident by the correspondence of the 0.149-to 0.150-nm d(060) line to the relative abundance ofsmectite detected by XRD analysis and by the results ofthe modified Greene-Kelley test. The presence of biotitein the parent material made it difficult to rule out com-pletely a 2:1 lattice precursor for the dioctahedral smec-tite (Borchardt, 1989). However, direct transformationof trioctahedral biotite to dioctahedral smectite was notsupported by the high relative abundance of biotite innear-surface horizons and by the possibly slow rate ofprotonation and layer charge reduction of trioctahedralvermiculite to tetrahedrally substituted smectite in theseAridisols.

Significant chemical weathering of primary silicates,such as feldspars, has occurred in these soils. Silicatedissolution, evidenced by etched and pitted sand grains(Boettinger and Southard, 1991), indicate that Si, Al,and cations have been released into solution and may beavailable for dioctahedral smectite neogenesis. Manyof the microagglomerates found in the Haplodurid andcalcareous Haplargid are probably feldspars that havebeen replaced by opaline silica, calcite (Boettinger andSouthard, 1990, 1991), and dioctahedral smectite, asdetermined by XRD analysis. In the limited-leachingenvironment, dissolution products are not removed fromthe soil. We suggest that neosynthetic dioctahedral smec-tite served as the major Al sink for the Al released fromthe dissolution of primary aluminosilicates hi these soils;other potential Al sinks (i.e., gibbsite, kaolinite, and2:1 phyllosilicates with Al hydroxy-interlayers) were notpresent hi substantial amounts, particularly in deeper,calcareous, and Si-rich horizons.

BOETTINGER AND SOUTHARD: PHYLLOSILICATES IN ARIDISOLS ON GRANITIC PEDIMENT 1197

To illustrate the conservation of feldspar dissolutionproducts in these soils, consider the dissolution of themost abundant Al-rich feldspar, oligoclase (80% albite,20% anorthite), in the granitic parent material. SufficientSi and Al from oligoclase hydrolysis combine with Mgfrom biotite or amphibole weathering in the presence ofCC>2 to form montmorillonite:15 Nao.gCao.2Ali.2Si2.gO8 + 2 Mg2+ + 17 H2O + 9 CO2

oligoclase= 5 Nao.4Si8(Al3.6Mgo.4)O2o(OH)4 + 10Na+

montmorillonite+ 3 CaCO3 + 2 H4SiO4 + 6 HCO3~

monosilicic acidMonosilicic acid not consumed in smectite neosynthesisforms opaline silica, and Ca2+ and HCO3~ precipitateas calcite. With dissolution of K-feldspars and limitedamounts of quartz providing additional silica, most of theAl released in dissolution of aluminosilicates is probablyincorporated into the octahedral sheet of the smectite.Excess silica not consumed in smectite neogenesis norlost by limited leaching is available for cementation ofmicroagglomerates, durinodes, and, ultimately, duripans.

The above silicate dissolution-smectite precipitationreaction was probably responsible for the formation ofAl-rich dioctahedral smectite in the clay fraction andin microagglomerates in the sand fraction. Silt-sizeddioctahedral smectite probably exists mainly as aggre-gates that are not dispersed in boiling 0.5 M NaOH.The presence of laumontite from hydrothermally alteredplagioclase in several deep horizons of the Haplargidsuggested that silt-sized smectite was also of hydrother-mal origin, although the lack of laumontite or othermineralogical indicators of hydrothermal origin in theHaplodurid pedon and the existence of dioctahedral smec-tite as microagglomerates suggested a pedogenic origin.Optimal chemical conditions for smectite neosynthesis(i.e., high pH, high silica), coupled with protection fromphysical weathering, could have created an environmentin which smectite was aggregated into silt-sized particles.

In summary, dioctahedral, Al-rich smectite is the dom-inant phyllosilicate in the clay fraction of both Aridisolpedons and is also present as silt- and sand-sized opalinesilica-cemented aggregates in deeper horizons. Dioctahe-dral smectite is probably neogenetic, forming from feld-spar dissolution products, with excess silica precipitatingas opaline silica. Silt- and sand-sized biotite is mostabundant in near-surface horizons, where it was mosthighly susceptible to physical weathering. In contrast,vermiculite is relatively more abundant in deeper hori-zons, probably due to rapid chemical weathering ofbiotite. The formation and stability of smectite and ver-miculite are favored in deeper horizons of Aridisols dueto longer periods of moisture availability for chemicalweathering and dampened temperature fluctuations rela-tive to near-surface horizons.

1198 SOIL SCI. SOC. AM. J., VOL. 59, JULY-AUGUST 1995