terrarosa 2011_banerjee_merino_jgeol_tr_iii_52011.pdf

8/9/2019 terrarosa 2011_Banerjee_Merino_JGeol_TR_III_52011.pdf

1/17

The University of Chicago Press

http://www.jstor.org/stable/10.1086/659146 .

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .http://www.jstor.org/action/showPublisher?publisherCode=ucpress . .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Geology.

http://www.jstor.org


2/17

[The Journal of Geology, 2011, volume 119, p. 259–274] 2011 by The University of Chicago.All rights reserved. 0022-1376/2011/11902-0002$15.00. DOI: 10.1086/659146

259

Terra Rossa Genesis by Replacement of Limestone by Kaolinite.III. Dynamic Quantitative Model

Amlan Banerjee 1,* and Enrique Merino 1,†

1. Department of Geological Sciences, Indiana University, Bloomington, Indiana 47405, U.S.A.

A B S T R A C T

We model quantitatively the chemical dynamics of terra rossa formation that we deduced earlier from new eld andpetrographic evidence. The key phenomenon is that authigenic kaolinite (a proxy for kaolinite solid solutions andother clays) replaces limestone at a downward-moving reaction front at the base of the existing terra rossa. Thechemical elements needed to make the authigenic clay are assumed to come from dissolution of eolian clay dustdeposited at the surface. The model consists of equations of mass conservation (in one spatial dimension) that

incorporate inltration, diffusion, and replacement and dissolution reactions. Distinctive elements of the model arethat (1) the kinetics of the kaolinite-for-calcite replacement is dominated by the crystallization stress generated bythe kaolinite growth, in accordance with the view that replacement happens not by dissolution-precipitation but byguest-mineral-growth-driven pressure solution of the host; (2) the eld observation that the terra rossa–forming frontis narrow indicates that the growth of kaolinite at the front proceeds roughly at the pace at which the appropriateaqueous reactants reach the front, which in turn warrants scaling time, length, and concentrations such that transportand reaction terms are of the same order of magnitude; and (3) the huge contrast between aqueous concentrationsand mineral densities, in effect, makes the scaled equation of mass conservation steady state. Numerical simulationscorrectly reproduce the formation of two adjacent zones in the moving reaction front—replacement and leaching—and of considerable secondary porosity in the leaching zone. This leaching porosity maximum constitutes a solitarywave, or soliton, that moves with the front. The simulations produce a range of predicted front widths that bracketthe observed front width. Similarly, model-predicted front velocities of a few meters per million years agree well,within an order of magnitude, with both (1) a paleomagnetically derived rate of terra rossa formation at Bloomingtonand (2) the rate of authigenic clay formation that could be expected to form—by clay dissolution at the surface,

downward inltration of aqueous solutes, and precipitation of new clay at the front—from average clay dust depositionrates as measured today.

Online enhancements: appendixes.

Introduction

We proposed, in an earlier work, a new chemical-dynamical theory to account for the formation ofterra rossa by replacement of the underlying lime-stone by authigenic red kaolinite and other clays(Merino and Banerjee 2008). The theory, summa-rized in the next section, was based on excellent

Manuscript receivedNovember18, 2009; acceptedDecember18, 2010.

* Author for correspondence; present address: Departmentof Earth and Environmental Science, New Mexico Institute ofMining and Technology, Socorro, New Mexico 87801, U.S.A.;e-mail: [email protected].

† E-mail: [email protected].

petrographic evidence of clay-for-limestone re-placement obtained on samples collected at thebottom of the terra rossa at Bloomington, Indiana.The microscopic evidence of volume-for-volumereplacement agreed with macroscopic, paleomag-netic evidence of constant rock volume reported inMeert et al. (2009).

The purpose of this article, the last of the series,is to model the new terra rossa–forming processquantitatively and, in particular, to comparemodel-calculated rates of formation of terra rossato rates obtained paleomagnetically (Meert et al.2009).


3/17

260 A . B A N E R J E E A N D E . M E R I N O

Figure 1. Schematic prole of a terra rossa claystone,with its associated underlying karst limestone. The terrarossa is bound by a thin reaction front at its bottom andthe current erosion surface at its top. It contains ironoxide pisolites and in situ limestone “oaters,” both ex-cellent indicators of the authigenic origin of the terrarossa. The aqueous Al, Si, and Fe resulting from disso-lution of eolian clay dust at the surface percolate downto the replacement front, driving growth of the authi-genic clay in it. The approximate position of the reactionfront shown in gure 2 a is marked with a rectangle.

Summary of the Qualitative Model

The qualitative model is described in detail in Me-rino and Banerjee (2008) and Meert et al. (2009).The new model of terra rossa genesis is shown ingure 1. The terra rossa clay grows authigenicallyat a narrow reaction front at the base of the terrarossa, where it replaces the underlying limestone.The front travels downward by virtue of the min-eral reactions that take place in it. The reactionfront (g. 2 a ) is about 9 cm thick and consists of aporous, leached-limestone zone, A, and a replace-ment zone, B. Zone B is the region where the au-thigenic replacement of limestone by red clay waslast in progress. The aqueous reactants needed tomake the new clays reach the reaction front byinltration from above; they are the aqueous Al,Si, and Fe produced at the current land surface by

dissolution of eolian-dust clay particles by rain andsoil water. (From Amram and Ganor’s [2005] ex-periments, one can derive dissolution half-lives oftens of years for smectite dust particles of size 1–100 mm dissolved in pH 5 water, and of tens ofmonths for 0.1- mm smectite particles.) As aqueousAl, Si, and Fe reach zone B, new clay (which, forsimplicity, we represent as pure kaolinite) growsvia

30.33Al 0.33SiO 0.83H O p2 2

[Al Si O (OH) ] H (1)2 2 5 4 1/6

and simultaneously pressure-dissolves an equalvolume (see “Kinetics of the Kaolinite-for-CalciteReplacement”) of host calcite, according to

2 CaCO H p Ca HCO , (2)3 3

leading to the local mass balance,30.45CaCO 0.33Al 0.33SiO 0.83H O p3(calc) 2 2

2 [Al Si O (OH) ] 0.45Ca 0.45HCO 0.55H , (3)2 2 5 4 1/6 3

where the coefcient 0.45 before the calcite for-mula accounts for mineral volume conservation.For simplicity, the predominant aluminum and car-

bonate species were assumed to be Al3

and, respectively (Merino and Banerjee 2008), anHCO 3assumption that results in a moderate overestimateof the amount of H released by reaction (3). Thereason for choosing a formula for kaolinite one-sixth the size of the standard nine-oxygen formulawill become apparent below. The H released inzone B by the mass balance (eq. [3]) travels down-ow and quickly dissolves out of pores in an ad-jacent slice of fresh limestone according to reaction(2), giving rise to the observed secondary pores thatdene zone A of the front (g. 2 c). Note that calcite

is dissolved simultaneously by two mechanisms:(1) in zone B, calcite is dissolved by pressure so-lution driven by kaolinite growth (as discussed indetail in “Kinetics of the Kaolinite-for-Calcite Re-placement”), and (2) in zone A, calcite is dissolvedchemically by the H released by the mass balanceequation (3) in zone B, at the same rate as these Hions are released by equation (3). When zone B hasbeen completely replaced, the replacement willstart in what is now zone A, and the acid releasedwill start dissolving pores in the calcite of a newzone A; this is how the front advances (downwardin nature, g. 1, and to the right in the sketch ofg. 4).

The actual kaolinite in terra rossa is red (and or-ange in thin section; see g. 2) because it is Fe 3

bearing, like that in laterites (Muller at al. 1995).Although the actual kaolinite is represented in thequantitative model below by pure kaolinite to keepthe number of components and mass conservation

equations to a minimum, it is simulated here bysimply taking for it an equilibrium constant greaterthan that for pure kaolinite (see “ParametersUsed”).

In short, the chemical dissolution of calcite inzone A is coupled via the H ions to the calcitepressure solution in zone B. This, in turn,is coupledvia the crystallization stress to the growth of ka-olinite, which is driven by the arrival at the frontof aqueous Al, Si, and Fe, which are derived fromdissolution of clay dust at the surface.

Meert et al. (2009) showed that the directions of


4/17

Journal of Geology D Y N A M I C S O F T E R R A R O S S A G E N E S I S B Y R E P L A C E M E N T 261

Figure 2. Summary of eld and petrographic evidence, reproduced from Merino and Banerjee (2008). a, Reactionfront between terra rossa and Salem Limestone on the Bloomington campus of Indiana University. The front consistsof a bleached zone ( A) ∼3 cm wide and a replacement zone ( B) ∼6 cm wide. Approximate positions of the photomi-crographs in b –e are marked. b , Crinoid columnal partly replaced by orange clay (at arrows) in the replacement zone(B). The circular shape of the replaced portions is preserved, indicating that the replacement was volume-for-volume,a fact used in adjusting the mass balance (eq. [3]). Plane-polarized light. c, Large secondary pores (areas in extinction)in the bleached zone ( A). Crossed polars. d , Incipient replacement of a crinoid columnal by the orange clay (whitearrows) in the replacement zone ( B). Plane-polarized light. e, Crinoid columnal completely replaced by clay. Theoccurrence in the replacement zone of crinoid columnals in various stages of replacement by kaolinite, shown in b ,d , and e, is additional evidence (beyond that described in Merino and Banerjee 2008) that the growth of kaolinite andthe dissolution of calcite proceeded simultaneously and at the same rate.

remanent magnetism in a 1.8-m prole of terrarossa at Bloomington, Indiana, cluster closelyaround the earth’s current magnetic-eld direction,a fact that both points to in situ growth of the terra

rossa clay (in agreement with the petrographic evi-dence of authigenesis) and indicates that its growthoccurred within the past 0.78 m.yr., that is, sincethe earth’s latest magnetic-eld reversal, indicatinga minimum terra rossa formation rate of

m/m.yr. Also, paleomagnetic direc-1.8/0.78 p 2tions obtained from unreplaced limestone “oat-ers” within the terra rossa are indistinguishablefrom directions in the bedrock limestone, indicat-ing that the oaters preserve their original strati-graphic position. This, in turn, implies that theterra rossa under each oater occupies a volume

equal to that of the limestone that has been re-placed, which conrms macroscopically the mi-croscopic evidence of volume conservation re-ported in Merino and Banerjee (2008).

We turn now to the kinetics of mineral replace-ment, especially applied to the kaolinite-for-calcitereplacement of interest in our model.

Kinetics of the Kaolinite-for-Calcite Replacement

In the preceding section, the kaolinite-for-calcitereplacement was held to proceed not by dissolu-tion-precipitation but by a kaolinite growth–drivenpressure solution of the host limestone. This mech-anism was explained in detail in “ReplacementPhysics” in Merino and Banerjee (2008, p. 68–69)


5/17


Figure 3. A, How guest-mineral growth automatically equalizes the volumetric rates of guest growth and hostdissolution, explaining why replacement preserves mineral volume. See text, appendix A, available in the onlineedition or from the Journal of Geology ofce, and Nahon and Merino (1997). In the general case shown here, theguest and host minerals have rate constants not very different from each other, which causes the slopes of the tworate-versus-time curves to be steep. B, For the particular case of the kaolinite-for-calcite replacement, the kaolinite-growth branch is practically horizontal, whereas that for calcite dissolution is very steep, because ; see k K kkaol calcappendix A. Consequently, the replacement rate is practically equal to the unconstrained initial growth rate of thekaolinite, a crucial factor for the quantitative model of “Quantitative Reaction-Transport Model.”

and references therein. Its great advantage over theidea of dissolution-precipitation is that it providesa strong, instantaneous coupling—the crystalliza-tion stress—between growthand dissolution, a cou-pling that automatically forces the growth of the

guest and the dissolution of the host to proceed atthe same rate, explaining why replacement pre-serves mineral volume. The demonstration that thecrystallization stress automatically equalizes guestmineral growth and host mineral dissolution ratesis reproduced from Nahon and Merino (1997) ingure 3 A in a general case. The analogous plot ingure 3 B shows the stress-coupled kinetics for theparticular case of interest here, characterized by thefact that , which makes the kaolinite k K kA(k aol) B(c alc)grow at practically its unconstrained rate and thecalcite pressure-dissolve at that same very low rate,regardless of its own superhigh dissolution kinet-ics. This will be essential in “Quantitative Reac-tion-Transport Model” below.

Essentially, the replacement proceeds as follows.The guest mineral A, driven by supersaturation,starts to grow within the host rock B at its uncon-strained rate and immediately starts to ex-growthRA,initialert an induced stress, or crystallization stress, onthe adjacent host as well as on itself. This stress

increases the Gibbs free energies (Kamb 1961) andthus the equilibrium constants for both minerals,K A and K B, simultaneously causing the A (guest)volumetric growth rate ( ≈ ) to de-A k S [(Q /K ) 1]A 0 A Acrease and the B (host) volumetric dissolution rate

(≈ ) to increase until the two curvesB

k S [1 (Q /K )]B 0 B Bintersect. These trends are shown qualitatively ingure 3 A. The point of intersection is at time . ∗tFrom then on, the two volumetric rates remainequal to each other, explaining with generality whyreplacement automatically preserves mineral vol-ume. However, for the kaolinite-for-calcite replace-ment of interest here, and because the rate con-stants of kaolinite and calcite differ by many ordersof magnitude, with , the rate of re- k K kA (k aol) B(c alc)placement is practically equal to the unconstrainedgrowth rate of kaolinite (see app. A, available in theonline edition or from the Journal of Geology of-ce),

R p k ln Q p R , (4)replacement A A unconstrained

as shown in gure 3 B by the practically horizontalkaolinite growth rate. This replacement kineticswill be used to advantage in our quantitative modelof terra rossa formation below, “QuantitativeReaction-Transport Model,” eliminating the need


6/17


Table 1. Notation

Symbol Denition Units

c i Concentration of i th species mol/L of pore volumeC Average concentration used as scale for concentrations mol/L of pore volume

,ic bcc c i i Initial and boundary concentrations of species i mol/L of pore volumeg i Scaled concentration of species Dimensionlesst Time sT Timescale st Scaled time, dened as t p T t Dimensionless x Length cm X Length scale cml Scaled length, dened as x p X l DimensionlessR j Rate of reaction for reaction j mol/(bulk cm 3 s) rj Scaled rate of reaction for reaction j Dimensionless k j Rate constant for reaction j mol/cm 2 shkaol Volumetric rate constant for kaolinite; eq. (14a) mol/cm 3 sb calc Scaled rate constant for calcite; eq. (16a)D Average effective diffusion coefcient of aqueous species cm 2/sU Water velocity through pores cm/sf Porosity (cm 3 of pore volume)/(cm 3

of bulk rock)f Leaching porosity (cm 3 of pore volume)/(cm 3

of bulk rock)G A factor of 10 -3 L/cm 3; see eq. (6)y i, j Stoichiometric coefcient of species i in j th reactionS0 Specic surface area of a mineral with volume fraction of 1 cm 2/(bulk cm 3) f Factor equal to 0 or 1, dened in eq. (7)

jK eq Equilibrium constant for chemical reaction j r i Molar density of mineral j mol/cm 3of jv j Volume fraction of mineral j cm 3 of mineral j /(bulk cm 3)p, d Superscripts denoting precipitation and dissolution, respectivelyv Quantity dened in eq. (30)Q A, Q B Ion activity products of guest mineral A and host mineral B,

respectively

to compute the crystallization stress. Nonetheless,the crystallization stress generated by the kaolinitegrowth and responsible for pressure-dissolving thecalcite,

k ln QA Astress p , (5)xln B k V B 0

turns out to be minute, because extremely littlework is required for kaolinite to pressure-dissolvecalcite (app. A; also gs. 3, 4 in Fletcher and Merino2001).

Quantitative Reaction-Transport Model

Reaction-Transport Equations. The process of clayformation, its replacement of calcite, and the re-sulting leaching, as outlined above, can be de-scribed by a set of continuity reaction-transportequations for mass, together with mineral rate lawsand initial and boundary conditions. For each rel-evant aqueous species i in the system (H , Al 3,SiO 2, , and Ca 2), the continuity equation isHCO 3

22c c c i i iGf p Gf D Gf U y R . (6) i, j j2t x x j p 1

Symbols are listed in table 1. The rst two terms

on the right represent the gradient of diffusive andadvective uxes, respectively. The last term rep-resents the rates of production or consumption ofaqueous species i by each of the j reactions in thefront: , kaolinite growth in zone B (eq. [1]) andj p 1

, chemical dissolution of calcite in zone A (eq.j p 2[2]). The factor L/cm 3 appears because we3G p 10later use equilibrium constants consistent withmol/L but mineral reaction rates in mol/cm 3 s(Wang et al. 1995).

Mineral Reaction Rates. For each reaction in-volved in the reaction front, we adopt a basic linearrate law whereby rate is proportional to the ther-

modynamic afnity (Aagaard and Helgeson 1982).For our reactions (1) and (2) the afnity is propor-tional to RT ln Q, where the supersaturation is

and the ion activity product Q is writtenQ p Q /K eqin terms of all relevant aqueous concentrations


7/17


(taken as proxies for the activities), ensuring thatall of them interact in the calculations.

Kaolinite Growth in Zone B . For kaolinitegrowth (reaction [1]) in zone B, the rate law adoptedis

1/3 1/3(c ) (c )3Al SiOp 2R p k S v f 1 , (7)k aol ka ol 0, kaol k aol kaol[ ]c K H eqwhere S0 is specic surface area of kaolinite, and

is the kaolinite volume fraction (see app. B, avail-v able in the online edition or from the Journal of Geology ofce). The factor f in the rate law ensuresthat kaolinite grows only where there is calcite tobe replaced, that is, only at the reaction front. With

denoting the volume fraction of calcite, ifv calc, then , that is, no kaolinite grows;v ≤ 0.01 f p 0.0calc

if , then , that is, kaolinite mayv 1 0.01 f p 1.0calcgrow.

We assume that all activity coefcients are equalto 1, and we use molarity for molality units in allequations. Our choice of kaolinite formula([Al2Si2O 5(OH) 4]1/6 ) in reaction (1) makes the expo-nent on in equation (7) unity. If we had chosenc Has the kaolinite formula the usual Al 2Si2O 5(OH) 4,then the exponent on aluminum and silica con-centrations would have been 2 in equation (7), andthat on would have been 6. This high exponentc Hwould exaggerate the effect of , and even veryc Hsmall changes in would make the solution nu-c Hmerically unstable.

Pressure Solution of Host Calcite in Zone B. Asexplained in “Kinetics of the Kaolinite-for-CalciteReplacement” and in gure 3 B, the pressure solu-tion of calcite driven by the crystallization stressgenerated by kaolinite growth takes place not ac-cording to the conventional calcite rate law, butaccording to the volumetric, unconstrained growthrate of kaolinite, .pR / rkaol kaol

Chemical Dissolution of Calcite in Zone A. The rate law adopted for chemical dissolutionof calcite (by the released H during replacementin zone B; reaction [3]) in zone A is

c c2 Ca HCOd 3R p k S v a 1 . (8)ca lc ca lc 0, calc calc H calc( )c K H eqThis rate law includes calcite surface area (app. B),and it coincides with the rst and last terms of thefour-term rate law determined by Busenberg andPlummer (1986).

Initial and Boundary Conditions. Initial Condi-tions . The concentrations of the ve relevantaqueous species Al 3, SiO 2, Ca 2, , and HHCO 3

throughout the limestone are initially such that thepore uid has a pH of 7.0 and is in equilibrium withboth calcite and kaolinite.

Boundary Condition at x p 0. Through the leftboundary, gure 4 a , enters a weakly acid (pH 5)

aqueous solution slightly supersaturated with re-spect to kaolinite ( ) and undersaturatedQ p 3.0kaolwith respect to calcite.

Boundary Condition at x p . On the rightside we impose the condition that the gradient ofall aqueous concentrations be 0, .c / x p 0 i

Volume Change. As a result of chemical disso-lution and pressure dissolution of calcite and pre-cipitation of kaolinite, the volume fractions ofthese minerals change in time and space. The vol-umetric growth rate of kaolinite in zone B at anymoment is given by

p

v Rkaol kaolp . (9)t r kaol

The rate of volume change of calcite consists oftwo terms, calcite pressure-dissolved in zone B andcalcite chemically dissolved in zone A:

p dv R Rcalc kaol calcp (1 f ) . (10)[ ]t r rkaol calc

In addition, all volume fractions must satisfy therelation

v

v

f p

1 f , (11)calc kaol

where f is the leaching porosity produced earlierby chemical dissolution of calcite and f is the (as-sumed constant) rock porosity of both terra rossaand limestone outside the reaction front.

Scaling of the Dynamic Equations. We now scalethe system of equations (6)–(10), taking advantageof the fact that the reaction front is narrow (see g.2a ), a feature of terra rossa formation that indicatesthat kaolinite grows and replaces calcite in the re-action zone B as fast as its aqueous ingredients aredelivered to the front, which in turn implies that

mineral growth rates and transport terms in equa-tion (4) should have the same order of magnitude.This kind of self-consistent scaling was also carriedout by Wang et al. (1995) and Ortoleva et al. (1982,p. 625–627) in models of lateritic weathering andmetamorphic layering, respectively. Here, it con-sists of the following steps.

1. Dimensionless, Order-One Variables . We de-ne new scaled parameters for time ( t ), length ( l ),and concentration ( g) that are dimensionless and oforder unity, according to


8/17


Figure 4. Model system and expected qualitative proles of mineral modes and aqueous species across the replace-ment zone ( B) and the leaching zone ( A) of gure 2 b . a , With supply of aqueous Al 3 and SiO 2 from the left, kaolinitegrows and partly replaces calcite in zone B. This replacement releases H (eq. [1]), which quickly dissolves morelimestone by reaction (3), generating pores (g. 2 c) in the bleached zone A. b , Expected prole of the reaction ratesof replacement and dissolution in the front. The two rates are coupled via H . c, Expected prole of aqueous-speciesconcentrations across the system. Al 3 and SiO 2 are consumed in the replacement zone, while H is produced thereand is immediately consumed by dissolving limestone in the leaching zone. d , Schematic proles of clay, calcite,and secondary porosity at an intermediate position of the front.

tt p ,

T

x l p , (12)

X

c ig p ; i C

the scales T , X , and C are determined below.

2. Scaling Mineral Rates . For kaolinite precip-itation in zone B, by inserting into equa-Cg p c i ition (7) we obtain

p pR p h r , (13)kaol kaol kaol

where

h p k S (14a)kaol kaol 0, kaol

and


9/17


Table 2. Typical Ranges of Values Adopted in Numerical Simulations

Variable Symbol Values adopted

Concentration scale C 10 5 mol/LDiffusion coefcient D 10 5 cm 2/sAnnual rainfall RF 2–0.5 m/yrPorosity f 0.01–0.1Velocity of water a U 1.6 # 10 5 to 4 # 10 6 cm/sGrain size L 0.1–1 mmSpecic surface area S0 600–60 cm 1Factor (see eq. [4]) G 10 3 L/cm 3Equilibrium constant, kaolinite b K eq, kaol 17.1, 54.1, 100Equilibrium constant, calcite K eq, calc 51.3Rate constant, kaolinite b kkaol 5 # 10 15 to 6 # 10 17 mol/cm 2 sRate constant, calcite kcalc 10 5 mol/cm 2 sScaled rate constant, equation (16c) c b calc 1.0Kaolinite molar density b r kaol 0.06 mol/cm 3Calcite molar density r calc 0.027 mol/cm 3Supersaturation with kaolinite Qkaol 3 (boundary condition)Initial kaolinite modal abundances 0.01–0.1Initial calcite modal abundances 0.98–0.8a We set rainfall in the Bloomington area between 2 and 0.5 m/yr. We assume that only 25% of it inltrates anddivide it by the porosity (0.01–0.1) to obtain water velocity U within terra rossa and limestone.b Values refer to a kaolinite formula one-sixth the size of its usual formula, Al 2Si2O 5(OH)4.c Where . This scaled rate constant is 10 7 for uncoupled dissolution of calcite. But dissolution ofb p Ck / kcalc calc kaolcalcite in leaching zone A is coupled to release of H ions by the replacement reaction (1) in zone B; thus, b calc mustbe set to 1 to keep the chemical leaching of calcite of the same order of magnitude as the kaolinite-for-calcitereplacement.

1/3 1/3g g3Al SiOp 2 r p v f 1 . (14b)kaol kaol 1/3 kaol( )C g K H eqHere, hkaol is the new rate constant (one that in-corporates the specic grain surface and is in unitsof mol/cm 3 s), and is the scaled kaolinitep rkaolgrowth rate of order 1.

Similarly, for the calcite chemical dissolution inzone A, the rate law, equation (8), becomes

d dR p h b r , (15)calc kaol ca lc calc

where

Ck calcb p , (16a)calc kkaol

h p k S , (16b)kaol kaol 0

and

Cg g2 Ca HCOd 3 r p g v 1 . (16c)calc H calc calc( )g K H eqDividing equation (15) by equation (13), we obtain

d dR rcalc calcp b p b , (17)calc calcp pR rkaol kaol

because both and are of order 1. With valuesp d r rkaol calcof C , k kaol , and k calc from table 2, we obtain

7b p 10 . (18)calc

This expresses the fact that the chemical dissolu-tion of calcite—if uncoupled—is seven orders ofmagnitude faster than the growth of kaolinite.However, in our case, the chemical dissolution ofcalcite effected in zone A is coupled to, and there-

fore cannot be faster than, the rate of release of Hions by replacement in zone B. This coupling re-quires that in our model, regardless of the hugevalue of bcalc in equation (18), we must set

b p 1. (19)calc

3. Scaling Time on Kaolinite Growth . The rateof volume fraction change of kaolinite, from equa-tions (9), (12), and (13), is

v T hkaol kaol pp r . (20)kaol( )t r kaol

We tie the timescale on kaolinite growth, the driv-ing reaction, by simply setting the right side ofequation (18) to be of order 1. Because the left sideof equation (18) and are already of order 1, itp rkaolfollows that

T h kaolp 1, (21)

r kaol

from which

r kaolT p . (22)h kaol


10/17


This denes the timescale T . Depending on valueschosen in this article for the kaolinite rate constantand the specic surface area, T is of the order of10 3–10 5 yr. Equation (20) becomes

v kaol pp r , (23)kaolt

and equation (10) becomes

v rcalc kaolp dp (1 f ) r r . (24) kaol calc[ ]t r calc

4. Scaling the Continuity Equation . We intro-duce the scaled variables in the continuity equation(eq. [6]),

2Gf C g Gf DC g i ip

2 2T t X l2Gf UC g i h y r , (25)kaol i, j j X l j p 1

replace T with in equation (25), and divide r /hkaol kao lby hkaol to leave the reaction term of order 1:

22Gf C g Gf DC g Gf UC g i i ip y r . (26) i, j j2 2 r t h X l h X l j p 1kaol kaol kaol

5. Making Transport and Reaction of the SameOrder of Magnitude . Since the reaction term andall the derivatives in equation (26) are now of order1, we can impose the condition that kaolinitegrowth must be as fast as transport of aqueous spe-cies by making the coefcient of the diffusion termalso of order 1,

Gf DCp 1. (27)2h X kaol

This relationship connects the concentration ( C )and length scales ( X ) as follows.

6. Concentration and Length Scales . Thechoice in equation (18) tied the timescale T to ka-olinite growth and led to equation (19). We nowtake mol/L, a probably typical order of5C p 10magnitude of aqueous concentrations in our model,

where the relevant species Ca2

, , Al3

, SiO 2,HCO 3and H are expected to be between 10 4 and 10 6

mol/L. The third choice was equation (27), whichyields the self-consistent value of the length scale X ,

0.5 0.5

Gf DC Gf DC X p p . (28)p( ) ( )h k Skaol kaol 0

The scales T , C , and X are now dened. See tables2, D1, and D2 (the latter two available in the onlineedition or from the Journal of Geology ofce). Put-

ting the scaled variables and scaling choices intoplace, equation (26) reduces to

22Gf C g g g i i ip v y r , (29) i, j j2 r t l l j p 1kaol

where

Gf UCp v. (30)

h X kaol

7. Natural Scaling . Finally, as pointed out byOrtoleva et al. (1987) and Wang et al. (1995), thecoefcient of the g i/ t term on the left side ofequation (29) is always very small because of theinherently huge difference between typical aqueousconcentrations and mineral densities, especially inlow-temperature environments. With values fromtable 2 and the choice of 10 5 mol/L as the con-

centration scale C, the coefcient is in(Gf C/ r )kaolour case between 10 8 and 10 9. This indeed makesthe left side of equation (29) vanish, leaving

22 g g i iv y r p 0, (31) i, j j2l l j p 1

where i is H , Al 3, SiO 2, , or Ca 2. Note thatHCO 3the self-consistent scaling results in a differentialequation of mass conservation with only one param-eter, v, given by equation (30), whichcombines manyof the relevant individual parameters of the terrarossa–formingprocess (inltrationvelocity,porosity,concentration length scales, and timescales).

Numerical Simulations

The scaled system of equations (eqq. [14b], [16c],[23], [24], and [31]) has been solved numerically for24 sets of values of the relevant parameters, listedin tables 3, D1, and D2. The calculation scheme isgiven in appendix C, available in the online editionor from the Journal of Geology ofce.

Parameters Used. Many parameters enter intoeach simulation: inltration velocity, limestoneand terra rossa porosity, kaolinite specic surface

area, kaolinite rate constant, equilibrium constant,supersaturation of the aqueous solution reachingthe front with respect to kaolinite, initial andboundary conditions imposed, etc. Each parameterhas been given values within a certain range, asdiscussed next. All parameter values for each nu-merical run are listed in tables D1 and D2.

The kaolinite equilibrium constant, K eq , wasgiven values ranging from that for pure kaoliniteupward. This assumes that the equilibrium con-stant for Fe 3 -bearing kaolinite solid solutions—which are the clays actually making up terra


11/17


Table 3. Initial and Boundary Conditions Adopted forthe Base Run

Chosen (mol/L) Scaled

Initial concentrations:Al 3 10 9 10 4

H (pH) 1.66 # 107

(6.78) 1.66 # 102

SiO 2 1.6 # 10 8 1.6 # 10 3Ca 2 5.41 # 10 3 5.41 # 10 2HCO 3 1.64 # 10 3 1.64 # 10 2

Boundary concentrations, left side: aAl 3 10 5 1H (pH) 10 5 (5)SiO 2 10 5 1Ca 2 5.41 # 10 5 5.41HCO 3 3 # 10 5 3.0

Note. Scaled values refer to a concentration scale ofmol/L.5C p 10

a Concentration of the aqueous species in the water en-tering the limestone model system.

Table 4. Reported Rates of Kaolinite Growth or Dissolution

Rate (mol/m 2 s) pH References

10 13 to 4 # 10 13 3 Metz and Ganor 20011.5 # 10 13 to 2 # 10 13 3 Carroll-Webb and Walther 19887.5 # 10 13 3 Wieland and Stumm 19927 # 10 13 to 8 # 10 13 3 Ganor et al. 19951.24 # 10 12 3 Nagy et al. 1991

Note. From these measured rates, we deduce a rate constant of ∼10 17 mol (Al 2Si2O 5(OH) 4/cm 2 s for kaolinite,equivalent to mol [(Al 2Si2O 5(OH) 4]1/6 /cm 2 s and consistent with the rate law in equation (7). For Fe 3-bearing176 # 10kaolinite solid solution, we increase the rate constant to 10 16 and 10 15 in the numerical simulations (see text).

rossa—is greater than that for pure kaolinite, anassumption consistent with previous work on thestability of phyllosilicate solid solutions (Merino1975; Merino and Ransom 1982; Vieillard 2000).The K eq value for pure kaolinite, Al 2Si2O 5(OH)4,namely, 10 7.4 , was taken from Bowers et al. (1984).The equilibrium constant for the pure “one-sixthkaolinite,” [Al 2Si2O 5(OH) 4]1/6 , adopted in this articlethus becomes 10 (7.4/6) (p 17). For the more realistickaolinite solid solutions in some of the runs in ta-bles D1 and D2, and 100.K p 54eq

The volumetric rate constant for kaolinite, hkaol(see g. 6 and tables D1 and D2), was assigned val-ues ranging from 10 15 mol one-sixth kaolinite/cm 3

s for pure kaolinite to an assumed 10 12 mol one-sixth kaolinite (solid solution)/cm 3 s for the morerealistic red clay minerals, in particular kaolinitesolid solutions, that make up terra rossa. The val-ues of kkaol used cover the range from mol/176 # 10cm 2 s for pure one-sixth kaolinite (estimated frompublished data; see table 4 and its footnote) upward.The values of S0 used cover one order of magnitude,from 600 to 60 cm 1, according to grain size; seeappendix B and tables D1 and D2. The rock poros-ity, f , of both limestone and terra rossa was set in

the simulations as either 0.01, which is very low,or 0.1, which is moderate.

Adopted values of inltration velocity, diffusioncoefcient, clay grain size, clay molar density, su-persaturation with kaolinite, and initial kaolinite

and calcite modal abundances are given in table 2.The inltration velocity was assumed to resultfrom one-tenth of rainfall divided by rock porosity,with rainfall ranging from 2 to 0.5m/yr. The bound-ary-condition supersaturation with clay was only3 for all 24 runs in tables D1 and D2, and valuesof up to 10 were used for the runs carried out toassess sensitivity (see “Sensitivity of ModelPredictions”).

Numerical Simulations. Twenty-four numericalsimulations were carried out with the scaled sys-tem of equations (eqq. [14b], [16c], [23], [24], and[31]), each for a set of relevant parameter valueslisted in tables D1 and D2. For each simulation,the time and length scales T and X , calculated fromequations (22) and (28), respectively, and the cal-culated front thickness and front velocity are givenin tables D1 and D2. The latter two quantities arealso plotted in gure 6 versus K eq and hkaol for twoporosities, 0.1 and 0.01, and are compared to eldobservations.

The evolution of several calculated quantitiesversus scaled space and time is shown graphicallyfor run 6 in gure 5. The quantities shown are vol-ume fraction of clay and calcite (g. 5 A), scaled

replacement rate in zone B and dissolution rate inzone A (g. 5 B), saturation states for clay and cal-cite (g. 5 C ), and dissolution porosity and pH (g.5D ). Each frame contains two sets of curves: thethin curves correspond to a duration of 2 scaledtime units, and the thick curves correspond to aduration of 4 scaled time units.

Each run has its own set of time and length scalesresulting from the self-consistent scaling. For run6, the time and length scales are yr andT p 6600

cm, respectively. In equation (22), we see X p 0.42that the timescale depends simply on the clay’svolumetric rate constant, hkaol . Since, as mentioned


12/17


Figure 5. Numerical results for run 6, table D1, available in the online edition or from the Journal of Geology ofce,shown as functions of scaled distance and for two scaled times, (13,200 yr; thin lines ) and (26,400 yr;t p 2.0 t p 4.0thick lines ). Simulation parameters include a high kaolinite rate constant and high rock porosity. Time and lengthscales for this run are 6600 yr and 0.42 cm. The front velocity calculated from the proles is about 4 scaled lengthunits per 2 scaled time units, or 1.7 cm/13,200 yr (1.2 m/m.yr.), in good agreement with the minimum rate of 2 m/m.yr. derived by Meert et al. (2009) from paleomagnetic work summarized in “Summary of the Qualitative Model.”The front width predicted from these proles is 6 scaled length units, or ∼3 cm, in good order-of-magnitude agreementwith the observed 9 cm of gure 2 a . See “Discussion and Conclusions.”

above, hkaol is assigned values that span three ordersof magnitude, the model-predicted timescales ofthe terra rossa–forming process for the 24 runs (ta-bles D1, D2) also span three orders of magnitude,from 700 to 528,000 yr.

The timescale of each run is critical in comput-ing the clay formation rate, or front velocity,predicted by the run. For run 6, we see from theproles of gure 5 that the clay-forming front ad-vances about 4 scaled length units (equivalent to

cm, or 1.7 cm) in 2 scaled time units4 # 0.42(equivalent to yr). Thus, the2 # 6600 p 13,200front velocity calculated for run 6 is 1.7 cm/13,200yr, or 1.2 m/m.yr., the value given in table D1. Allthe front velocities calculated in this way, one perrun, are shown in the top frames of gure 6, andthey are compared to independent estimates below.

The front thickness predicted in run 6 can beestimated visually from any of the panels of gure5, especially gure 5 B, where we see that the totalwidth of zones B and A is about 10 scaled lengthunits, which yields a front of cm,10 # 0.42 p 4.2

the value given in the last row of table D1.Comparing Predictions to Observations. As ex-pected, all the simulations carried out in this studyreproduced correctly several qualitative spatial at-tributes observed in eld (g. 2 a ), namely, a reactionfront consisting of two adjacent zones—replacement(zone B) and leaching (zone A)—in the correct se-quence (g. 5 B), with clay-for-calcite replacement inthe replacement zone and high dissolution porosityin the leaching zone. A more stringent test involvescomparing the numerical simulations for frontthickness and front velocity to independent values.


13/17

Figure 6. Predicted front velocity, or rate of terra rossa formation, in meters per million years ( top ) and front widthin centimeters ( bottom ) of a clay-making front for a series of volumetric rate constants ( kS0) of kaolinite ranging from10 15 mol/cm 3 s (equivalent to mol/cm 2 s, referred to “one-sixth kaolinite”) to 10 12 mol/cm 3 s (equiv-17 k p 6 # 10alent to mol/cm 2/s), kaolinite equilibrium constants ranging from 17 (referred to pure “one-sixth kaolinite”)156 # 10to 100, and porosity of both terra rossa and limestone (10% and 1%). Predictions for aqueous concentrations, clayand limestone volume fractions, and leaching porosity are not shown; they have the general trends shown for run 6in gure 5. The front width of 9 cm observed in Bloomington (g. 2 a ) is shown ( plane in c and d ) for comparison.The paleomagnetically derived minimum front velocity of 2 m/m.yr. is also shown for comparison ( plane in a andb ). Simulation parameters for each run are given in tables D1 and D2, available in the online edition or from theJournal of Geology ofce. In general, for a given rate constant, front velocity decreases but front width increases withincreasing equilibrium constant. On the other hand, for a given equilibrium constant, an increase in the rate constantincreases both front width and front velocity. The simulations show that for greater porosity (10%), the predictedfront width ( c) and front velocity ( a ) bracket the observed values fairly well.


14/17


For the front thickness, we see in gure 6 c and 6 dthat the predicted values that best approximate theobserved front thickness of 9 cm, which is markedin gure 6 by a horizontal rectangle, are those cal-culated for a fairly high volumetric rate constant of

about 1013

mol one-sixth kaolinite/cm2

s at a mod-erate porosity of 0.1 (g. 6 c) and for a high rate con-stant of about 10 12 mol one-sixth kaolinite/cm 2 sat a low porosity of 0.01 (g. 6 d ). It is clear in gure6 that for the volumetric rate constant of pure ka-olinite, about 10 15 mol one-sixth kaolinite/cm 2 s,the predicted front thickness is more than an orderof magnitude lower than the observed 9 cm. This isconsistent with the fact that the actual clay thatalways forms in terra rossa is never pure kaolinitebut consists of red kaolinite (and other red clays) thatcontains considerable amounts of Fe and other cat-ions, which, we estimated above, are responsible forits higher rate constant.

Finally, we compare the calculated front velocity,plotted vertically for the 24 runs in gures 6 a and6b , against two independent estimates: that ob-tained from paleomagnetic measurements in Meertet al. (2009) and that obtained from reasonable claydust deposition rates.

1. As explained in “Summary of the QualitativeModel,” paleomagnetic directions of terra rossasamples covering 1.8 m coincided with the currentnorth, implying that they crystallized in less than0.78 m.yr. (i.e., since the date of latest reversal) andthus that the minimum rate of authigenic terrarossa formation is 2 m/m.yr. This value is markedwith a horizontal rectangle in gures 6 a and 6 b ,which shows that only the front velocity simula-tions for moderate porosity (0.1) and highest vol-umetric rate constant used ( ∼10 12 mol one-sixthkaolinite/cm 2 s) can be commensurate with, orgreater than, the “paleomagnetic” minimum of 2m/m.yr. (see g. 6 a ). This is consistent with theestimate we made earlier that the actual red kao-linite solid solution that mostly makes up terrarossa is bound to have a much higher rate constantthan does pure kaolinite.

2. If, as proposed by Merino and Banerjee (2008),terra rossa is the authigenic equivalent of theeolianclay dust settled at a site, according to gure 1, thenits formation rate should be roughly equivalent tothe deposition rate of clay dust. An average supplyof 10 g eolian clay dust/m 2 yr—about one-half ofthe total dust deposition rates measured by Smithet al. (1970), Rabenhorst et al. (1984), and Reheisand Rademaekers (1997)—would yield (by disso-lution at the land surface, downward inltration,and precipitation as new clay at the reaction front)an average of 5 mm of terra rossa (of assumed den-

sity 2 g/cm 3) per 1000 yr, or 5 m/m.yr., in roughagreement with the minimum value obtained pa-leomagnetically for Bloomington.

In summary, rates of terra rossa formation, orfront velocities, predicted by the reaction-transport

calculations for high clay volumetric rate constantsand moderate porosity agree well within one orderof magnitude with the two independent estimatesdiscussed above, which agree with each other. Sim-ilarly, the front thicknesses predicted by the modelfor runs using high rate constants and moderateporosity come out equal to 10–20 cm, in good agree-ment with the value observed in the eld (g. 2 a ).These agreements between predicted and observedvalues appear to support the qualitative new theoryof terra rossa formation proposed in Merino andBanerjee (2008).

Sensitivity of Model Predictions. The sensitivity ofthe calculated front velocity and front thickness toporosity, clay equilibrium constant K eq , and volu-metric clay rate constant hkaol can be assessed fromthe top and bottom panels of gure 6, respectively.It is apparent in these gures that sensitivity to theclay equilibrium constant is low and that sensitiv-ity to the porosity is low or moderate but that (be-cause of eq. [19]) sensitivity is high to the volu-metric rate constant hkaol (p ), neither of k Skaol 0,kaolwhose factors, conventional rate constant and spe-cic surface area, is well known. For the specicsurface area, we have used the grain size depen-dence described in appendix C.

An additional 20 numerical simulations were car-ried out specically to assess how sensitive the cal-culated front velocity and front thickness are toaqueous-ion diffusivity, boundary supersaturationwith respect to kaolinite, clay grain size, and bound-ary pH of the clay-forming aqueous solution. Theresults are summarized in gure 7. It is apparent thatthe inuence of boundary pH is negligible betweenpH 5 and 6 but considerable between pH 5 and 4and that the inuence of both diffusivity and su-persaturation with kaolinite is minor.

Discussion and ConclusionsGeochemical Modeling and Petrography. The

reaction-transport model of “Quantitative Reac-tion-Transport Model” and “Numerical Simula-tions” tests quantitatively the chemical dynamicsof terra rossa formation proposed in Merino andBanerjee (2008), a dynamics driven by kaolinitegrowth. The model incorporates some eld and pet-rographic observations and produces predictionsthat are compared to other observations not intro-duced in the model. Both the microscopic (petro-


15/17


16/17


zone B, despite the fact that the rate constant forcalcite dissolution is orders of magnitude greaterthan the replacement rate.

The strategy of incorporating evidence of replace-ment and its kinetic and volumetric implications

into the geochemical modeling of terra rossa gen-esis should be viewed not as circular but as nec-essary. If one looks again at gures 2 b , 2d , and 2 e,showing crinoid columnals in various stages of re-placement—intermediate, incipient, and complete,respectively—one realizes that at all stages of re-placement the kaolinite growth and the calcite dis-solution must have been simultaneous and musthave proceeded at the same rate, preserving mineralvolume all along, an astonishing fact that the stan-dard kinetic laws available would be unable to pre-dict if we viewed replacement as resulting from justcalcite dissolution and kaolinite precipitation, be-cause the kinetic rate constant of calcite dissolu-tion is many orders of magnitude greater than thatfor kaolinite growth. The equality of rates char-acteristic of any replacement, however, stops beinga physicochemical conundrum if replacement isviewed as produced by guest-growth-driven pres-sure-solution of the host, a mechanism that has notrouble making the two rates mutually equal andpreserving the solid volume, as demonstrated ingure 3, “Kinetics of the Kaolinite-for-Calcite Re-placement,” and references therein.

Some Simplifying Assumptions. 1. To keep thenumber of components low, the several clay min-erals that usually form terra rossa claystones arerepresented in the model by pure kaolinite. Morerealistic clay-mineral solid solutions are simulatedin the numerical runs simply by adopting values ofthe equilibrium and rate constants higher thanthose for pure kaolinite.

2. The number of relevant aqueous species in-cluded in the model also was kept to a minimum(ve) to minimize the number of mass conservationdifferential equations to be solved simultaneously.This, in effect, amounted to neglecting the speciesAlOH 2 and H 2CO 3; thus, the model probably over-

estimates both the H released by reaction (1) inthe reaction zone and the secondary porosity, f ,produced by that reaction (g. 5 D ).

Conclusions. 1. Quantitative reaction-transport

simulations of the downward-advancing replace-ment of limestone by authigenic clays (representedby kaolinite) reproduce the formation of twoadjacent subzones in the reaction front and the for-mation of considerable secondary porosity in sub-

zone A, as observed. The moving secondary-poros-ity porosity wave, or soliton, predicted in gure 5 Dand conrmed by the occurrence of pores in theleaching zone (g. 2 a , 2c), may have a crucial rolein triggering a reactive-inltration instability thatis probably responsible for producing the typicalkarst-sink morphology in the limestone that un-derlies terra rossa (Merino and Banerjee 2008).

2. Numerical simulations for moderate terrarossa porosity and for high values of the clay growthrate constant yield front widths of several centi-meters, in good agreement with the observedthickness.

3. Similarly, model-predicted front velocities ofseveral meters per million years agree well withtwo independent estimates of the rate of terra rossaformation, one derived from paleomagnetic mea-surements and the other derived from reasonablerates of eolian clay dust deposited at the landsurface.

A C K N O W L E D G M E N T S

We thank M. Person of the New Mexico Instituteof Mining and Technology for generous help andnancial support to A. Banerjee; the Departmentof Geological Sciences at Indiana for support to A.Banerjee as an associate instructor; Y. Wang of San-dia Labs, Albuquerque, and R. T. Glassey of theMathematics Department at Indiana University foradvice at several stages of the modeling; J. Waltherof Southern Methodist University for advice on theeffects of solid solution on the kinetics of mineralreactions; S. Dworkin of Baylor University and X.Querol of the Spanish Research Council at Barce-lona for references and data on dust deposition,chemistry, and mineralogy; J. Ganor of Ben GurionUniversity for help with deriving the half-life of

eolian smectite particles at the earth’s surface; andB. Wilkinson of Syracuse University and an anon-ymous reviewer for very helpful critical reviews ofthe manuscript.

R E F E R E N C E S C I T E D

Aagaard, P., and Helgeson, H. C. 1982. Thermodynamicand kinetic constraints on reaction rates among min-erals and aqueous solutions. I. Theoretical consider-ations. Am. J. Sci. 282:237–285.

Amram, K., and Ganor, J. 2005. The combined effect ofpH and temperature on smectite dissolution rate un-der acidic conditions. Geochim. Cosmochim. Acta 69:2535–2546.


17/17


Bowers, T. S.; Jackson, K. J.; and Helgeson, H. C. 1984.Equilibrium activity diagrams. New York, Springer,397 p.

Busenberg, E., and Plummer, N. L. 1986. A comparativestudy of the dissolution and crystal growth kineticsof calcite andaragonite. In Mumpton, F. A.,ed. Studiesin diagenesis. U.S. Geol. Surv. Bull. 1578:139–168.

Carroll-Webb, S. A., and Walther, J. V. 1988. A surfacecomplex reaction model for the pH-dependence of co-rundum and kaolinite dissolution rates. Geochim.Cosmochim. Acta 52:2609–2623.

Fletcher, R., and Merino, E. 2001. Mineral growth inrocks: kinetic-rheological models of replacement, veinformation, and syntectonic crystallization. Geochim.Cosmochim. Acta 65:3733–3748.

Ganor, J.; Mogollón, J. L.; and Lasaga, A. C. 1995. Theeffect of pH on kaolinite dissolution rates and on ac-tivation energy.Geochim. Cosmochim. Acta59:1037–1052.

Kamb, W. B. 1961. The thermodynamic theory of non-hydrostatically stressed solids. J. Geophys. Res. 66:259–271.

Meert, J. G.; Pruett, F.; and Merino, E. 2009. An “inverseconglomerate” paleomagnetic test and timing of insitu terra rossa formation at Bloomington, Indiana. J.Geol. 117:126–138.

Merino, E. 1975. Diagenesis in Tertiary sandstones fromKettleman North Dome, California. II. Interstitial so-lutions: distribution of aqueous species at l00 C andchemical relation to the diagenetic mineralogy. Geo-chim. Cosmochim. Acta 39:l629–l645.

Merino, E., and Banerjee, A. 2008. Terra rossa genesis,implications for karst, and eolian dust: a geodynamicthread. J. Geol. 116:62–75.

Merino, E., and Ransom, B. 1982. Free energies of for-mation of illite solid solutions and their composi-tional dependence. Clays Clay Miner. 30:29–39.

Metz, V., and Ganor, J. 2001. Stirring effect on kaolinitedissolution rate. Geochim. Cosmochim. Acta 65:3475–3490.

Muller, J.-P.; Manceau, A.; Calas, G.; Allard, T.; Ildefonse,P.; and Hazemann, J.-L. 1995. Crystal chemistry of

kaolinite and Fe-Mn oxides: relation with formationconditions of low temperature systems. Am. J. Sci.295:1115–1155.

Nagy, K. L.; Blum, A. E.; and Lasaga, A. C. 1991. Dis-solution andprecipitation kinetics of kaoliniteat 80 Cand pH 3: the dependence on solution saturation state.Am. J. Sci. 291:649–686.

Nahon, D., and Merino, E. 1997. Pseudomorphic replace-ment in tropical weathering: evidence, geochemicalconsequences, and kinetic-rheological origin. Am. J.Sci. 297:393–417.

Ortoleva, P.; Merino, E.; Moore, C.; and Chadam, J. 1987.Geochemical self-organization I: reaction-transportfeedbacksand modeling approach.Am. J. Sci. 287:979–1007.

Ortoleva P.; Merino, E.; and Strickholm, P. 1982. Kineticsof metamorphic layering in anisotropically stressedrocks. Am. J. Sci. 282:617–643.

Rabenhorst, M. C.;Wilding,L. P.; and Girdner, C. L. 1984.

Airborne dusts in the Edwards Plateau regionof Texas.Soil Sci. Soc. Am. J. 48:621–627.

Reheis, M., and Rademaekers, J. 1997. Predicted dustemission vs. measured dust deposition in the south-western United States. In Impact of Climate Changeand Land Use in the Southwestern United States.USGS Web workshop, http://geochange.er.usgs.gov/sw/.

Smith, R. M.; Twiss, P. C.; Kraus, R. K.; and Brown, M.J. 1970. Dust deposition in relation to site, season, andclimatic variables. Soil Sci. Soc. Am. J. 34:112–117.

Vieillard, P. 2000. A new method for the prediction ofGibbs free energies of formation of hydrated clay min-erals based on the electronegativity scale. Clays ClayMiner. 48:459–473.

Wang, Y.; Wang, Y.; and Merino, E. 1995. Dynamicweathering model: constraints required by coupleddissolution and replacement. Geochim. Cosmochim.Acta 59:1559–1570.

Wieland, E., and Stumm, W. 1992. Dissolution kineticsof kaolinite in acidic aqueous solutions at 25 C. Geo-chim. Cosmochim. Acta 56:3339–3355.

terrarosa 2011_banerjee_merino_jgeol_tr_iii_52011.pdf

Documents