Porosity and surface area evolution during weathering of two igneous rocks

Available online at www.sciencedirect.com

Geochimica et Cosmochimica Acta 109 (2013) 400–413 www.elsevier.com/locate/gca

Porosity and surface area evolution during weathering of two igneous rocks Alexis K. Navarre-Sitchler a,b,⇑, David R. Cole c, Gernot Rother d, Lixin Jin b,e, Heather L. Buss f, Susan L. Brantley b b

a Geology and Geological Engineering, Colorado School of Mines, CO, United States Center for Environmental Kinetics Analysis, Earth and Environmental Systems Institute Pennsylvania State University, PA, United States c Department of Earth Sciences, The Ohio State University, Columbus, OH, United States d Geochemistry and Interfacial Sciences Group, Chemical Sciences Division, Oak Ridge National Laboratory, TN, United States e Department of Geological Sciences, University of Texas at El Paso, El Paso, TX, United States f School of Earth Sciences, University of Bristol, UK

Received 24 June 2012; accepted in revised form 7 February 2013; available online 16 February 2013

Abstract During weathering, rocks release nutrients and store water vital for growth of microbial and plant life. Thus, the growth of porosity as weathering advances into bedrock is a life-sustaining process for terrestrial ecosystems. Here, we use small-angle and ultra small-angle neutron scattering to show how porosity develops during initial weathering under tropical conditions of two igneous rock compositions, basaltic andesite and quartz diorite. The quartz diorite weathers spheroidally while the basaltic andesite does not. The weathering advance rates of the two systems also differ, perhaps due to this difference in mechanism, from 0.24 to 100 mm kyr1, respectively. The scattering data document how surfaces inside the feldspar-dominated rocks change as weathering advances into the protolith. In the unaltered rocks, neutrons scatter from two types of features whose dimensions vary from 6 nm to 40 lm: pores and bumps on pore–grain surfaces. These features result in scattering data for both unaltered rocks that document multi-fractal behavior: scattering is best described by a mass fractal dimension (Dm) and a surface fractal dimension (Ds) for features of length scales greater than and less than 1 lm, respectively. In the basaltic andesite, Dm is approximately 2.9 and Ds is approximately 2.7. The mechanism of solute transport during weathering of this rock is diffusion. Porosity and surface area increase from 1.5% to 8.5% and 3 to 23 m2 g1 respectively in a relatively consistent trend across the mm-thick plagioclase reaction front. Across this front, both fractal dimensions decrease, consistent with development of a more monodisperse pore network with smoother pore surfaces. Both changes are consistent largely with increasing connectivity of pores without significant surface roughening, as expected for transport-limited weathering. In contrast, porosity and surface area increase from 1.3% to 9.5% and 1.5 to 13 m2 g1 respectively across a many cm-thick reaction front in the spheroidally weathering quartz diorite. In that rock, Dm is approximately 2.8 and Ds is approximately 2.5 prior to weathering. These two fractals transform during weathering to multiple surface fractals as micro-cracking reduces the size of diffusion-limited subzones of the matrix. Across the reaction front of plagioclase in the quartz diorite, the specific surface area and porosity change very little until the point where the rock disaggregates into saprolite. The different patterns in porosity development of the two rocks are attributed to advective infiltration plus diffusion in the rock that spheroidally fractures versus diffusion-only in the rock that does not. Fracturing apparently diminishes the size of the diffusion-limited parts of the spheroidally weathering rock system to promote infiltration of meteoric fluids, therefore explaining the faster weathering advance rate into that rock. Ó 2013 Elsevier Ltd. All rights reserved. ⇑ Corresponding author at: Geology and Geological Engineering, Colorado School of Mines, CO, United States. Tel.: +1 303 384 2219. E-mail addresses: [email protected], [email protected] (A.K. Navarre-Sitchler).

0016-7037/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.gca.2013.02.012

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

1. INTRODUCTION Pristine igneous bedrock does not usually contain enough water to support significant biomass. Once meteoric water infiltrates and reacts with the rock however, it begins to disaggregate into soil grains (Hochella and Banfield, 1995; Oguchi and Matsukura, 1999; White and Brantley, 2003; Meunier et al., 2007; Jin et al., 2011). This weathering process releases nutrients to solution and increases the porosity and surface area – all processes that enhance water retention to sustain ecosystem growth (Torn et al., 1997; Meunier et al., 2007; Brantley, 2008; Graham et al., 2010). The initial growth of nanometer-scale porosity (here termed nanoporosity) is poorly understood (Montgomery and Brace, 1975; White et al., 1996; Putnis, 2002; Navarre-Sitchler et al., 2009). For example, although the health of terrestrial ecosystems relies on soil production, the rates of surface area change during water infiltration into rock cannot yet be predicted because our fundamental understanding of the water–rock interface inside rocks is limited (Hochella et al., 1988). This limitation is one of the reasons we cannot accurately extrapolate laboratory reaction rates to the field in predictive numerical models (NavarreSitchler and Brantley, 2007; Navarre-Sitchler et al., 2011). Traditionally, surface area is measured in Earth materials with sorption, Hg porosimetry, or microscopy (Ball et al., 1990; Rufe and Hochella, 1999; Brantley and Mellott, 2000; White and Brantley, 2003; Navarre-Sitchler and Brantley, 2007). However, sorption only characterizes accessible, connected pores, while microscopy only images extremely small subsamples (Fischer and Gaupp, 2004). Neutron scattering holds the promise of providing unique insights into how pores ranging in size from nm’s to 10s of lm’s in diameter develop during weathering of macroscopic rock volumes (Radlinski, 2006; Jin et al., 2011; Bazilevskaya et al., in press). Here we probe nm- to micron-scale changes in the pore structure and mineral surface area that occurs during weathering of two common igneous rock compositions under tropical climate conditions. The first sample, a basaltic andesite, is a hand sample-sized clast that weathers without spheroidal fracturing (Sak et al., 2004). The second sample was collected from a 2 m-sized corestone of the Rio Blanco quartz diorite from the Luquillo Critical Zone Observatory in Puerto Rico that weathers with spheroidal fracturing (Turner et al., 2003; Buss et al., 2008). In both cases, oxidation of iron-bearing minerals is the earliest weathering reaction but dissolution of plagioclase feldspar is the reaction that disaggregates the rock (Buss et al., 2008; NavarreSitchler et al., 2009, 2011). The plagioclase reaction front – i.e., the thickness of the zone across which plagioclase dissolves completely – is 1–2 mm in the basaltic andesite but 40 cm in the quartz diorite. Furthermore, the weathering advance rate of the quartz diorite is nearly three orders of magnitude faster than the basaltic andesite despite the generally observed faster dissolution in the laboratory of calcium-rich plagioclase in the basaltic andesite relative to the more sodium-rich plagioclase in the quartz diorite (Bandstra et al., 2008). We use neutron scattering to under-

401

stand how nanoporosity and surface area evolve during the onset of weathering with and without spheroidal fracturing in these rocks. 2. THE WEATHERING EXAMPLES The basaltic andesite clast (Fig. 1a) was collected from a fluvial terrace in Costa Rica deposited 125 ka (Sak et al., 2004; Navarre-Sitchler et al., 2011). Clasts from these terraces range from basalt to andesite in composition (Navarre-Sitchler et al., 2011). The basaltic andesitic clast analyzed for this study was 61 wt.% SiO2, 5 wt.% (Na2 O + K2O), and 4–5% total Fe oxide (reported as Fe2O3). The unaltered core of the clast contains 67 vol.% feldspar and 27 vol.% pyroxene. The core contains 1–3% porosity and abundant feldspar crystals (0.1–1 mm) in a matrix of small feldspar/pyroxene grains (1–10 lm) with trace ilmenite and magnetite (Sak et al., 2004; Navarre-Sitchler et al., 2009). The basaltic andesitic core is enveloped in a porous weathering rind (50% porosity) of iron and aluminum oxides with small amounts of kaolinite in some clasts (Sak et al., 2004). This rind is thought to have grown at a rate of 0.2–0.5 mm kyr1 (Sak et al., 2004; Pelt et al., 2008; Navarre-Sitchler et al., 2011). Feldspar and pyroxene dissolution occurred together over a reaction front thickness of 1 mm at the rind-core interface and was likely rate-limited by diffusion (Table 1, Fig. 1B) (Navarre-Sitchler et al., 2011). The rind is friable and disaggregates easily once plagioclase is completely weathered (0.4 cm from the core). The bulk density decreases from 2.8 in the unweathered core to 1.4 in the rind (Sak et al., 2004). The unweathered quartz diorite has <3% porosity and is dominated by crystals of plagioclase (56%) in an interstitial matrix of submillimeter-sized quartz (25%). The samples were collected from an outcrop in the Luquillo Critical Zone Observatory in the Luquillo Experimental Forest in Puerto Rico (Fig. 1C) (Murphy et al., 1998; White et al., 1998; Turner et al., 2003; Buss et al., 2008; Chabaux et al., 2013). Oxygenated waters infiltrate into meter-long vertical fractures into the granitic rock and iron in the biotite oxidizes as oxygen diffuses in from the fracture walls (Buss et al., 2008). Fletcher et al. (2006) demonstrate that this reaction causes volume expansion that can lead to spheroidal fractures spaced 2.5 cm apart. Fractures wrap the central corestone like onionskin (Fig. 1B). The intact but weathered rock layers between fractures are referred to as rindlets (Buss et al., 2008). Although relatively large fractures define the rindlets, very small micro-fractures are common within the individual rindlets, especially those that are more weathered. The set of fracture-delineated rindlets around the central corestone is 40 cm thick, i.e., starting at sample PR28 and grading through multiple rindlets into saprolite in the outer part above sample PR13 (Table 1, Fig. 1D). The spheroidal fractures anastomose and intersect to define rindlets that range in number from 13 to 16 around the corestone. Here, 10 rindlets along a vertical transect perpendicular to the corestone were sampled (Table 1, Fig. 1D). Only the outermost rindlets disaggregate easily.

402

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

A

B

~ 3 cm ~ 1 cm

C

D

PR13 PR14

PR15 PR16 PR17 PR20 PR21

PR25 PR27 PR28 PR35 PR03

~ 40 cm

~ 10 cm

Fig. 1. (A) Weathered basaltic andesite clast from Costa Rica. Orange/brown rind is approximately 2.9 cm thick in this image. (B) A photograph showing the thick section which was cut across the weathering interface of the basaltic andesite for neutron scattering analysis. A mask with a 1 mm rectangle cut-out was used to define the area of impingement of neutrons with the sample. This mask was placed over the thick section so that the cutout was parallel with the reaction front (imaged here as the transition from grey to orange) and was moved in 1 mm increments between data collection. (C) Roadcut showing spheroidally weathered quartz diorite corestone in Puerto Rico. Corestone is center grey part and darker grey rindlets envelop the corestone. Thickness of the rindlet set was approximately 40 cm in the upper part where sampled. (D) Diagram showing sample locations: parent samples (hachured, samples PR35, PR03) and rindlet samples (collected from individual rindlets as shown by labels). The rindlet zone begins to transition to saprolite at sample PR15 and PR13. All samples were cut into thick sections for neutron scattering. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3. METHODS 3.1. Neutron scattering experiments Neutrons penetrate through rock samples, scattering from interfaces between materials of different chemistry and density that define the scattering contrast. Analysis of small angle (SANS) and ultra-small angle (USANS) neutron scattering scattering data yields statistical information about the internal structures on length scales of 1 nm to 30 lm. Using appropriate models, scattering provides a detailed characterization of topology and architecture of pore networks (Mildner et al., 1986; Radlinski, 2006; Anovitz et al., 2009; Jin et al., 2011; Bazilevskaya et al., in press). Highly disordered and scale invariant pore systems can be

mathematically described by fractal and polydisperse hard-sphere models, which give access to the pore size distribution, pore volume, surface roughness and internal interfacial area (Hinde, 2004; Radlinski, 2006). Samples from each weathering system were prepared for SANS and USANS neutron scattering experiments using standard thick section preparation. Each section was glued to a quartz slide and polished to a thickness of 150 lm to prevent multiple neutron scattering (Anovitz et al., 2009). For the basaltic andesite clast, one thick section was cut across the entire mm-thick plagioclase reaction front (Fig. 1C). For the granitic system, individual sections were cut from rindlets at different distances from the core (Fig. 1D). For the basaltic andesite, a cadmium (SANS) or gadolinium (USANS) mask with a 1.8  0.1 cm

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

403

Table 1 Results from neutron scattering analysis. Density (g/cm3)

Total porosity (%)b (connected)

Dc@low Q

Dsd@mid Q

Ds@high Q

%e of parent feldspar remaining

Total surface area (m2/g)b (connected)

BET surf. area (m2/g)

Basaltic andesite 0.4 0.3 0.2 0.1 0 0 0 0

1.0 1.0 1.4 2.8 2.8 2.8 2.8 2.8

8.6 6.6 3.3 1.9 1.6 1.5 1.5 1.3

2.95 ± 0.02 2.61 ± 0.03 2.72 ± 0.07 2.83 ± 0.07 2.83 ± 0.04 2.97 ± 0.07 2.88 ± 0.03 2.91 ± 0.06

– – – – – – – –

Not linear 2.52 ± 0.05 2.66 ± 0.02 2.74 ± 0.03 2.70 ± 0.01 2.72 ± 0.01 2.68 ± 0.01 2.79 ± 0.09

0 12 18 79 100 100 100 100

23.5 (nm) 13.2 (13.0) 8.3 (8.2) 4.0 (3.9) 3.1 (2.7) 3.5 (3.1) 2.9 (nm) 2.4 (nm)

30 – 13 – – – – –

Quartz diorite PR13 40 PR15 38 PR16 37 PR17 36 PR20 30 PR21 27 PR25 18 PR27 10 PR28 5 PR35 0 PR03 0

1.5 2.0 2.0 2.0 2.0 2.0 2.0 2.7 2.7 2.7 2.7

9.4 6.3 3.3 1.2 2.2 2.2 2.3 2.2 2.4 1.4 1.3

Not linear 2.90  ± 0.02 2.85  ± 0.02 2.81  ± 0.04 2.85  ± 0.04 2.69  ± 0.05 2.75  ± 0.03 2.77  ± 0.04 2.90 ± 0.02 2.80 ± 0.03 2.75 ± 0.02

– – 2.22 ± 0.01 2.18 ± 0.01 2.18 ± 0.01 2.25 ± 0.01 2.33 ± 0.01 2.30 ± 0.01 2.22 ± 0.09 – –

2.30 ± 0.02 Not linear 2.59 ± 0.07 2.74 ± 0.02 2.63 ± 0.08 2.75 ± 0.06 2.65 ± 0.05 2.53 ± 0.05 2.35 ± 0.06 2.46 ± 0.04 2.52 ± 0.03

68 97 98 84 91 96 83 82 100 100 100

13.1 6.1 3.2 1.6 2.6 1.7 2.5 2.0 2.0 1.9 1.3

– – – – – – – – – – –

I.D.

Distance (cm)a

(nm) (5.5) (2.8) (1.8) (1.5) (1.48) (nm) (nm)

a

Distance from the pristine unweathered rock (=zero). Porosity and specific surface area determined from scattering data as described in text. Values in parentheses are connected porosity or surface area as measured for basaltic andesite samples; quartz diorite samples were not analyzed for unconnected porosity. nm = not measured. c Surface or mass fractal dimension: surface fractals are marked by  , all others are mass fractals. d The surface fractal dimension for the separate mid-Q region with distinct slope observed in some of the quartz diorite rindlet samples. e Calculated from the mass transfer coefficient s (Anderson et al., 2002) using measured Na and Ti concentrations. b

rectangular opening was used to define the area of analysis. The thick section was moved behind the stationary mask to measure scattering every 1 mm across the reaction front. For the rindlets, a cadmium mask with a 1.6 cm-diameter circular opening was used to define the area of analysis for both SANS and USANS. SANS and USANS data were measured at the National Institute of Standards and Technology Center for Neutron Research (NCNR). SANS was measured using the NG7 ˚ with a wavebeamline (Glinka et al., 1998) (k = 8.09 A length spread, Dk/k = 13%) at sample-to-detector distances of 1.0, 4.0, and 15.3 m (the latter using biconcave MgF2 lenses) for a Q (scattering vector) range of 8.9  104 to ˚ 1, equivalent to spherical scatterers ranging 4.3  101 A ˚ . USANS was measured on in diameter from 15 to 7000 A the BT5 double crystal diffractometer (Barker et al., 2005) ˚ with Dk/k = 5.9%, which using a wavelength of k = 2.38 A 5 ˚ 1. The overcovered a Q range of 4  10 to 3  103 A lap in Q-range of the SANS and USANS data allowed the two data sets to be merged. Contrast-matching experiments were performed for the basaltic andesite to assess connected versus unconnected porosity. In this experiment, scattering from connected pores was eliminated by saturating the sample with a 75% D2O–25%H2O mixture of the same scattering length density (calculated as described in Section 3.2 to be ˚ 2) as the bulk unweathered rock. Samples 4.3  106 A

were equilibrated with contrast-matched fluid for P72 h. Scattering was measured from the contrast-matched samples in the same manner as described above. 3.2. Neutron scattering data analysis Data were processed according to standard methods including empty cell correction (a blocked beam and a quartz slide with Krazy Gluee), background scattering correction, transmission correction, radial averaging and normalization to absolute intensities. All SANS data reported here were observed to be azimuthally symmetric. Finally, the different Q-segments were combined into one scattering curve. The normalized USANS data were desmeared following protocols of NCNR (Kline, 2006). After de-smearing, data from USANS and SANS overlapped continuously across the Q range (Fig. 2). The combined SANS/USANS dataset was then fit to Eq. (1) at high Q values to determine the contribution of incoherent scattering, mostly resulting from hydrogen in the sample: IðQÞ ¼ bQm þ c;

ð1Þ

where m is the apparent slope of the data and b is a fitting parameter. The incoherent scattering (c in Eq. (1)) was subtracted from the data.

404

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413 11

10

A 8

10

7

6

10

-1

Intensity, I(Q) (cm )

-1

Unweathered Quartz Diorite

9

10 Intensity, I(Q) (cm )

B

Unweathered Basaltic Andesite

4

10

2

10

0

10

10

5

10

3

10

1

10

-1

10

-2

-3

10

10 -4

-3

10

10

-2

10

-1

10 -1

Scattering Vector, Q (angstrom )

-5

10

-4

10

-3

10

-2

10

-1

10

0

10

-1

Scattering Vector, Q (angstrom )

Fig. 2. Plots of scattering intensity (I(Q)) versus scattering vector (Q). SANS (closed circles) and de-smeared USANS (open circles) data overlap and are well connected across Q values for the unweathered basaltic andesite (A) and granitic rock (B). Incoherent scattering from hydrogen in the samples produces the constant intensity values at high Q in both samples.

Table 2 Calculated scattering length densities (qj ). Density (g cm3)

˚ 2) qj (A

Basaltic andesite minerals Plagioclase Na0.387Ca0.613Al1.613Si2.387O8 Augite Ca0.35Mg0.42Fe(II)0.23SiO3 Ilmenite FeTiO3

2.69 3.3 4.79

3.94  106 4.85  106 4.40  106

Quartz diorite minerals Plagioclase Quartz Hornblende Biotite

2.69 2.65 4.3 3.1

4.22  106 4.18  106 4.52  106 3.50  106

2.7 2.6

4.3  106 4.45  106

Formula

Na0.50Ca0.48K0.01Al1.43Si2.56O8 SiO2 Ca1.73Na0.29K0.06Mg2.64Fe1.95Mn0.09Ti0.12Al0.37Si7.24Al0.76O22(OH)2 K0.89Fe1.28Mg1.19Mn0.02Ti0.18Al0.16Si2.88Al1.12O10(OH)2

Normalized rock formulas Basaltic andesite Na0.14K0.06Ca0.08Mg0.02Fe0.04Al0.42SiO2.88 Quartz diorite Na0.1K0.02Mg0.06Ca0.13Fe0.09Ti0.01Al0.36SiO2.95

The resulting curve contains information about scattering objects (pores) and interfaces between minerals and voids. The intensity of scattered neutrons is proportional to the volume of scatterers and the scattering contrast (Dqj ) i.e., the difference in scattering length densities (qj ) between rock phase and voids. The coherent scattering length densityqj for a mineral phase is given by Eq. (2)1: Pn bci qj ¼ i¼1 : ð2Þ vm Here, bci is the bound coherent scattering length of atom i of n atoms of a molecule and Vm is the molecular volume (g mol1). The values of the scattering length densities ˚ 2) calculated for minerals ranged from 3.94 to (qj , A ˚ 2 for the basaltic andesite and 3.5 to 4.85  106 A ˚ 2 for the quartz diorite (Table 2). The values 4.52  106 A  of qj were individually calculated from the bulk chemical ˚ 2 for the formula of each rock composition: 4.3  106 A 6 ˚ 2 unweathered basaltic andesite and 4.45  10 A for the unweathered quartz diorite (Table 2). While minerals within the samples have slightly different scattering length

1

http://www.ncnr.nist.gov/resources/sldcalc.html.

densities, scattering from mineral–mineral interfaces is minimal compared to the large scattering contrast of mineral-unfilled void interfaces where qpore ¼ 0. Therefore, pore-mineral interfaces scatter significantly stronger than mineral–mineral boundaries, allowing application of the two-phase approximation (Table 2) (e.g., Kahle et al., 2004; Radlinski, 2006; Anovitz et al., 2009).Rock neutron scattering data that display a power-law relationship between I(Q) and Q with an exponent of 3 to 2 over at least an order of magnitude of Q indicate a fractal distribution of pore sizes. In those cases, fractal scaling laws can be used to interpret pore network geometry. If those pores are modeled as an assemblage of polydisperse hard spheres, the size and surface area distributions can be calculated. Here we apply both fractal scaling laws and a poly-disperse sphere model to interpret the neutron scattering data comprehensively. I(Q) versus Q data were fit to a polydisperse sphere model and a histogram of pore sizes were calculated using the two-phase approximation and fitting routines implemented in the program PRINSAS (Hinde, 2004). Porosity and surface area were then calculated from this pore size distribution using macros implemented in PRINSAS (for details on fitting macros see Hinde, 2004). Scattering data were also plotted on Porod plots (log I(Q)  Q4 versus

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

log Q) to accentuate regions in Q-space described by different values of m in Eq. (1) (Anovitz et al., 2009). The data were assessed to determine where the Porod plots show linearity over ranges in Q of at least one order of magnitude to define regions of fractal behavior. Within each region of linearity, log I(Q) values were fit to Q following Eq. (3) to determine the slope m: IðQÞ ¼ bQm :

ð3Þ

3.3. BET surface area measurements Surface areas were determined using Brunauer Emmett Taylor (BET) analysis with N2 gas for two samples of basaltic andesite rind taken at 0.2–0.3 and >0.4 cm from the core–rind interface. Sample size limitations prevented BET measurements on material taken from closer to the core or on samples spaced closer together across the reaction front. Isotherms were measured over a relative pressure range from 3  105 to 1 on approximately 1.5 g of sample using a Quantachrome Autosorb1. To enable BET measurement, the small samples had to be ground to particles of diameter ranging from 0.125 to 0.25 mm. The BET surface area is thus an upper limit to the actual BET surface area of the intact rind. 3.4. Transmission electron microscopy (TEM) TEM images were collected from unweathered and weathered basaltic andesite and quartz diorite using a Hitachi HF-3300 cold-field-emission TEM, operated at 200 kV accelerating voltage at the High-Temperature Materials Laboratory at Oak Ridge National Laboratory. Brightfield images (1024  1024 pixels) were recorded on a Gatan 794 multi-scan CCD camera system. The images of unweathered basaltic andesite were taken from the core while images of weathered basaltic andesite were collected from two points located approximately 2 or 5 mm from the core–rind boundary. TEM images of the unweathered and weathered quartz diorite were taken from samples PR03 and PR16, respectively (Fig. 1D). Sample PR03 is from the core while sample PR16 was 37 cm from the corestone–rindlet interface. Energy dispersive spectra (EDS) were collected for some spots using a Bruker EDS system. 4. RESULTS AND DISCUSSION 4.1. Pore imaging Pores were observed in the transmission electron microscope images at the same length scales probed by SANS. Many of these pores are located at grain boundaries and triple junctions in the unweathered samples (Fig. 3, left-hand column of images). Dimensions of basaltic andesite pores imaged under TEM in the unweathered sample were generally <<1 lm but quartz diorite pores ranged from submicronto micron-sized. In both compositions of unweathered protolith, some grain–grain interfacial angles at triple junctions are <60°, consistent with an interconnected pore network along the 3-grain boundaries (Lee et al., 1991). In both samples, pores are more abundant and variable in size and shape after weathering (Fig. 3, right-hand col-

405

umn of images). In the weathered quartz diorite, microfractures were commonly observed, especially near feldspar (Fig. 3f and h). Microfractures often connected small pores (Fig. 3f). Precipitates of kaolinite, sometimes covered by a layer of Mn–Fe oxyhydroxide, were also observed in micro-fractures (Fig. 3h). Only a couple of Fe-oxide filled micro-fractures were observed in the basaltic andesite. Fe-oxide was mostly observed around augite grains in the basaltic andesite core. 4.2. Porosity and surface area with weathering SANS data for all samples were azimuthally symmetric and non-monotonous with respect to scattering angle, indicating randomly shaped and oriented features. The combination of SANS and USANS data provides information on features in the rocks of dimension or spacing, d, from 7 nm to 30 lm (d = 2p/Q). Scattering intensities generally increased along the weathering trajectory, i.e., with increasing distance from the unweathered rock towards the soil, espe˚ 1, Fig. 4A and B). cially for d >75 nm (Q < 8  103 A The scattering increase is consistent with increasing porosity and surface area with increasing weathering. Total porosity calculated from the SANS and USANS data (nanoporosity) increases in both weathering systems from <3% in the unweathered cores to 10% over the width of the completely developed reaction front for plagioclase dissolution (Table 1). In the basaltic andesite, where scattering was also measured on H2O/D2O soaked samples, total porosity increases more with distance from the corestone than connected porosity: 98% versus 83% of pores were connected before and after weathering, respectively (Fig. 4A and C, Table 1). Surface areas calculated from SANS and USANS increase in both cases by approximately a factor of 6 across the plagioclase reaction front: from 4 to 24 m2 g1 in the basaltic andesite and 2 to 13 m2 g1 in the quartz diorite. For the basaltic andesite, the average SANS-calculated surface area of samples from 0.2–0.3 cm from the core equaled 11 m2 g1. For comparison, surface area of the ground basaltic andesite measured by BET for roughly that location was 13 m2 g1. The BET surface area increases to 30 m2 g1 further out in the rind on another ground sample. While the increases in porosity and surface area in the two weathering systems are similar, the evolution of the surface area and porosity with distance across the plagioclase reaction front is not. Porosity and surface area increase steadily over a distance of 4 mm across the plagioclase reaction front in the basaltic andesite (Fig. 5). In contrast, despite plagioclase weathering in the rindlets, the specific surface area and porosity values derived from SANS and USANS for the intact quartz diorite rindlets remain constant over 35 cm of reaction front and increase significantly only in the last 5 cm (Fig. 5 35–40 cm from the corestone). In the center part of the quartz diorite rindlet zone, scatterers <20 nm were the only features in the network within the plagioclase reaction front where specific surface area increased consistently outward with distance from the core (Fig. 6B). In the outermost rindlet samples (PR15 and

406

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

Unweathered

Weathered

Basaltic Andesite

A

B

0.2 µm

700 nm

C

D

0.2 µm

Quartz Diorite

E

1 µm

PR03

1 µm

G

1 µm

F

PR16

1 µm

PR03

H

PR16

1 µm

Fig. 3. TEM micrographs of unweathered (left) and weathered (right) basaltic andesite (top) and quartz diorite (bottom). In both unweathered rocks (a and g), pores at triple junctions show contact angles <60o, consistent with interconnected tubules of cross-section <1 lm (Lee et al., 1991). Unconnected pores <0.5 lm diameter also decorate grain boundaries (c and e). Micro- and nano-porosity increase in the weathered samples (b and d: light grey areas are pores). Micro-cracking was also observed in the granitic example: fractures connect pores especially in feldspar-rich zones (f). Fracturing is attributed to oxidation of ferrous minerals (Buss et al., 2008). Only kaolinite precipitates, identified by X-ray dispersive spectra and texture, were observed in micro-fractures in early rindlets close to the corestone. In outer rindlets, Mn–Fe oxide coatings were also observed. Equant crystallites of kaolinite growing in a micro-fracture in plagioclase from an outer rindlet are shown in (h) with a coating of a dark Mn–Fe phase.

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

407

Fig. 4. Scattering plots for total porosity (open symbols) in unweathered (squares) and weathered (circles) basaltic andesite (A) and quartz diorite (B). The weathered samples were measured 0.4 mm from rind–core interface (basaltic andesite) or 40 cm above the corestone (quartz diorite). For unweathered (A) and weathered (C) zones of the basaltic andesite sample, contrast matching delineated scattering due to unconnected (filled symbols) and total porosity (open symbols). More than 80% of total porosity, calculated from this scattering data, is connected. The connected porosity is 1.5% and 5.5% of the total rock volume in the unweathered and weathered basaltic andesite samples, respectively. Samples in the granitic rindlet zone (PR25, stars, multiplied by 0.1 for plotting purposes) show three regions of fractal scattering, indicated by line fits (D). In contrast, unweathered (PRO3, data multiplied by 0.001 for plotting purposes) and weathered (PR13) quartz diorite samples show two regions of fractal scattering.

PR13), the surface area increased for features <1000 nm. In contrast, the surface area in the basaltic andesite increased consistently for features <3000 nm once weathering initiated 0.2 cm from the core–rind interface (Fig. 6A and C). Initial connectivity in the unweathered basaltic andesite is only indicated for scatterers <400 nm diameter, as shown in Fig. 7 where scattering from H2O–D2O contrast matched samples (filled symbols) are significantly less than from all pores, documenting that unconnected porosity is a small fraction of the total porosity. Pores observed at triple junctions in TEM images are in the size range of 100–200 nm (Fig. 3c) and therefore may constitute the connected pore network in the unweathered basaltic andesite. The scattering that occurs from unconnected features >400 nm in the unweathered basaltic andesite are presumed to be pores that are not preserved during preparation of the very tiny and thin TEM sections, explaining why large pores were not imaged by the TEM. During weathering of the basaltic

andesite clast, the pore connectivity increases especially in pores >400 nm. After weathering, the specific surface area of all features in the basaltic andesite has increased, and little unconnected surface area remains (Fig. 7). Despite the lack of evidence of an increase in porosity and surface area from the SANS data across much of the quartz diorite rindlet zone (Fig. 5B), weathering features were observed (Buss et al., 2008). Specifically, increasingly lower density and iron oxidation were manifestations of chemical weathering across the rindlet zone (Buss et al., 2008). Buss et al. (2008) and Fletcher et al. (2006) have suggested that oxygenated water infiltrates the rocks along fractures. As oxygen diffuses through the fracture walls into the unfractured corestone material and oxidizes ferrous minerals, elastic strain energy builds up in the corestone until eventually spheroidal fracturing occurs (Fletcher et al., 2006). In general, each new fracture forms within the unaltered corestone; therefore, reactive fluids are inferred to have reacted with rindlets that are further from the

408

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

Fig. 5. Total porosity (closed squares) or specific surface area (open circles) calculated from neutron scattering data for basaltic andesite (A) and quartz diorite (B) plotted versus distance from protolith (Table 1). The rindlet zone in (B) lies between 6 and 37 cm. Total porosity and specific surface area are calculated using PRINSAS. This surface area does not include the contributions from pores with radius larger than 30 microns.

A

B

C

D

Fig. 6. Specific surface area (SSA) determined from neutron scattering plotted versus scatterer dimension d for basaltic andesite (A and C) and quartz diorite (B and D) (note logarithm scale in C and D). Panels (A and B) show scattering data from all samples while panel (C) shows scattering data from 0, 0.1 and 0.4 mm from the unweathered core and panel (D) shows scattering data from 0, 18, and 40 mm from the core (PR03, PR25, and PR13, respectively). Weathered samples are shown as circles or stars (stars indicate samples from the plagioclase reaction front). Unweathered samples are shown as squares. In both basaltic andesite and quartz diorite, the total calculated SSA increases with weathering especially for smaller rather than larger scatterers (A and B).

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

409

4.3. Fractal dimensions

Fig. 7. Total and unconnected specific surface area as a function of scatterer diameter calculated from neutron scattering data for the basaltic andesite sample. In unweathered basaltic andesite only pores of diameter <400 nm are connected (arrow A delineates the difference between unconnected and connected in unweathered basaltic andesite). For example, pores along triple junctions with contact angles <60o (as imaged in Fig. 3A) are <400 nm and are likely to be connected. Basaltic andesite weathering connects the larger pores. Arrow B delineates connectivity of porosity in the rind after weathering, compared to very little connectivity in the core.

corestone for longer durations. For example, the outer rindlets (PR13, PR15 and PR16) are thought to have been exposed to reactive fluids for longer periods of time than those nearest the corestone. According to this model, the first reaction in each rindlet is oxidation of ferrous minerals such as biotite that can cause formation of a new rindletforming fracture. These large fractures are also accompanied by many small micro-fractures. This overall model may explain why, unlike the basaltic andesite clast, the initial generation of porosity and surface area in the quartz diorite does not correlate with the extent of feldspar depletion (Table 1). For example, in rindlets PR15 and PR16, feldspar content is 97% of the value in the original quartz diorite but porosity and specific surface area have increased 5- and 6-fold, respectively. The % feldspar values in samples summarized in Table 1 were calculated as the normalized concentration, i.e., the mass transfer coefficient s (Anderson et al., 2002), using measured Na and Ti concentrations. The value (1  s) reveals the fraction of the original Na that remains in the sample compared to the parent sample (corestone) under the assumption that Ti is immobile during weathering. These calculations and the assumption of immobility of Ti have been described previously by Buss et al. (2008). In contrast to the quartz diorite samples PR15 and PR16, when porosity and surface area have increased 5or 6-fold in the basaltic andesite, 100% of the primary feldspar has been dissolved away. This early development of porosity in the quartz diorite rindlets without total feldspar loss is attributed to the development of microcracks related to oxidation of biotite (Buss et al., 2008) and not to wholesale dissolution. These ideas are further discussed in the context of the fractal nature of the rocks in the next section.

Classic examples of fractal systems such as the Sierpinski gasket or Mandlebrot set are self-similar nonrandom geometric fractals. Pores in rocks are themselves not fractal in that they are not self-similar. The pores have different geometries and shapes at all length scales. However, when the statistical average of rock properties such as density or scattering intensity decreases as a power function of the length scale of measurement, the rocks can be considered as random fractals. Random fractals can be treated with the same equations as nonrandom geometric fractals (Stanley, 1991). In rocks, scattering over one order of magnitude in Q that fits Eq. (3) yields a slope m that defines either a surface (4 < m < 3) or mass fractal (3 < m < 2) (Radlinski, 2006). The fractal dimensions, calculated from m, vary from 2 to 3 for both surface and mass fractals and are defined as Ds and Dm, respectively (Eq. (4a,b)a and b): Ds ¼ 6  m; Dm ¼ m:

ð4a; bÞ

For a porous medium like a rock, the surface fractal dimension documents the roughness of the pore wall interface where Ds = 2 for smooth surfaces and 3 for extremely rough surfaces. Likewise, the mass fractal dimension provides insight into the self-affine character of the overall geometry of the pore volume, where Dm = 2 for monodisperse and 3 for polydisperse pore-size distributions. For example, to explain these concepts further, we can consider a Euclidean object such as a cube of size L. For such an object, the area or mass varies with L2 or L3, respectively. In contrast, a surface fractal is a material whose surface area varies with LDs where Ds is not an integer. Likewise, for a mass fractal, the mass varies as LDm where Dm is not an integer (Radlinski, 2006). We discuss these ideas with respect to the two rock examples in the next sections. 4.3.1. Basaltic andesite Scattering data across the plagioclase reaction front in the basaltic andesite fits Eq. (3) over two different portions of the data. These two fractal regions are defined by a break in slope at d  1 lm for all points within the reaction front except in the outermost sample 0.4 cm from the core–rind interface where the plagioclase has weathered completely and the material is friable and is easy to disaggregate. For ˚ 1, i.e., d = 1–24 lm), the slope m low Q (<6.5  104 A indicates a mass fractal, which is attributed here to scattering from the network of pores (Table 1). The mass fractal dimension at low Q decreases across the reaction front (0–0.3 cm from core) from 2.9 to 2.6 reflecting a shift in the pore size distribution with weathering from more polydisperse to more monodisperse (Fig. 6). Given our observation from TEM that some pores along triple-grain junctions have contact angles less than 60o, we infer that the unweathered rock has an interconnected pore network along triple grain junctions as well as unconnected pores (Fig. 3). This polydisperse pore network apparently becomes more monodispersed with increasing dissolution.

410

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

˚ 1), slopes In contrast, at high Q (6.5  104 to 0.1 A indicate a surface fractal that we attribute to scattering from bumps on pore–mineral interfaces at length scales from d = 6 nm to 1 lm (Table 1). The surface fractal dimension also decreases with weathering from 2.7 or 2.8 in the core to 2.5 at 0.3 cm from the core. This decrease in surface fractal dimension is consistent with smoothing of mineral surfaces from a more space-filling to less space-filling surface during weathering of the basaltic andesite. Like the increase in porosity and specific surface area (Fig. 5), the evolution of fractal dimensions across the reaction front are generally consistent with gradual changes in the pore network until 0.4 cm from the core. A calculated increase in specific surface area inside the rock samples can result either from increased roughness of surfaces or increased porosity (or both). The decrease in surface fractal dimension with distance from the unweathered core is most consistent with weathering resulting in smoother surfaces. Therefore, the increase in specific surface area during weathering of the basaltic andesite is attributed largely to increased porosity. In the outermost basaltic andesite rind sample, a nonlinear region was identified at high Q and no slope was determined. In contrast, the mass fractal dimension for that sample determined from the low Q data is 2.95. The nonlinear behavior in the rind is indicative of a very irregular pore network that does not exhibit self-similarity. Overall, the changes in mass and surface fractals across the rind are inferred to document the increasing disaggregation of the sample as plagioclase weathers: once disaggregation is significant, scattering no longer fits Eq. (3) at high Q (small scatterers), but at low Q (large scatterers) the material still retains a pore geometry best described with a mass fractal close to 3. 4.3.2. Granitic rock Similar to the basaltic andesite, the unweathered quartz diorite was fit with a mass fractal at low Q (Dm  2.8) and a surface fractal at high Q (Ds  2.5). The break in slope occurred at d  650 nm (Fig. 4D). Across the rindlet zone, however, three fractal regions were identified and, for all but the inner-most rindlet, all three of these regions are indicative of surface fractal behavior (Table 1). The surface fractal dimension of the high Q data shows a tendency to increase across the rindlet zone from 2.4 to 2.7, consistent with increased surface roughness at small length scales with weathering. A region at mid-Q was fit with a surface fractal dimension of 2.2–2.3. These relatively smooth surfaces at length scales 60–600 nm are most consistent with microcracks observed in TEM images. Finally, the low Q data for the rindlets are described by a surface fractal that varies in dimension from 2.7 to 2.8. At the point where the rindlet zone transitions to saprolite (PR15 (38 cm) and PR13 (40 cm)), the fractal behavior changes abruptly. PR15 is described by a surface fractal dimension at low Q with Ds = 2.9 but this sample has a non-linear (non-fractal) region at high Q. In contast, PR13 (40 cm) exhibits surface fractal behavior with Ds = 2.3 at high Q and non-fractal behavior at low Q. Thus, scattering from these more disaggregated samples is

inconsistent and difficult to interpret and is likely related to the process of disaggregation. Unlike the basaltic andesite, abrupt changes in fractal behavior in the quartz diorite were identified across the plagioclase reaction zone (rindlet zone). These changes in the fractal behavior across the rindlet zone demonstrate changes to the pore network that do not correlate to feldspar depletion, porosity, or surface area (Table 1 and Fig. 5B). 5. MECHANISMS OF WEATHERING Navarre-Sitchler et al. (2009, 2011) have shown ample evidence that the dominant mechanism of transport of reactants and products during weathering of the basaltic andesite clast is diffusion. Lack of roughening during dissolution of the crystalline surfaces as evidenced by the lowering of the surface fractal dimension across the rind is consistent with dissolution close to equilibrium, a feature consistent with transport limitation. Apparently, in the absence of fracturing, diffusion-controlled weathering dominates to produce relatively consistent changes in porosity, specific surface area and fractal behavior as a function of plagioclase dissolution extent across the mm-wide reaction front (Table 1, Fig. 5A). Throughout almost the entirety of weathering of this basaltic andesitic clast, the material was best characterized as both a mass and a surface fractal. In contrast, the quartz diorite weathering system shows complex transformation from a mass + surface fractal to a surface fractal across the plagioclase reaction front. Apparently, the fracturing of this sample causes relatively abrupt changes in the fractal distribution of scatterers across the 10’s of cm wide reaction front. Based on petrographic observation (Buss et al., 2008), each of the rindlets is akin to a separate weathering system where a reaction front develops between the fracture and the unweathered diffusion-limited core of the rindlet, experiencing inward-diffusion of reactants, not unlike the basaltic andesitic clast. However, the rindlets also experience further micro-cracking, reducing the size of the diffusion-limited part of the weathering rindlet rock further. In the zones between micro-fractures, the overall mechanism of chemical weathering is inferred to be the same in both systems (diffusionlimited mineral dissolution) – however, the micro-fracturing in the quartz diorite breaks up the original quartz diorite corestone into even smaller diffusion-limited weathering subregions as weathering progresses across the rindlet zone. Fracturing accelerates the weathering by forming rindlets and allowing infiltration of advecting fluids, and further (micro)fracturing accelerates the weathering even beyond that. The fracturing and microfracturing are the likely reasons why the weathering advance rate in the quartz diorite is nearly three orders of magnitude faster than that of the basaltic andesite (Fletcher et al., 2006; Pelt et al., 2008). This acceleration of weathering is particularly noteworthy given that dissolution of Ca-rich plagioclase measured in the laboratory for basaltic andesite is generally faster than that measured for the Na-rich plagioclase in the quartz diorite (Bandstra et al. (2008), and references therein). A

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413

similar argument has been made recently to explain why diabase weathers to very thin regolith and granitic rock weathers to thick regolith in the Piedmont of Virginia in the USA (Bazilevskaya et al., in press). In that case, the rates of weathering advance are thought to be similar but, like the examples of tropical weathering shown here, fracturing in the granitic rock allows solute transport by advection whereas the lack of fracturing in the diabase limits transport to diffusion only. In both cases, fracturing driven by oxidation of biotite is hypothesized to explain why advective transport occurs in the granitic lithology. Bazilevskaya et al. (in press) argue that weathering-driven fracturing of granitic lithologies may be much more common than on less silicic and more Fe-rich lithologies, possibly explaining the common observation of thick regolith on granite and thin regolith on diabase worldwide. The accelerated rate of weathering of the quartz diorite may also be affected by a bacterial ecosystem that has been observed at the bedrock–saprolite interface in that system (Buss et al., 2005; Minyard et al., 2011, 2012). Putatively, Fe-oxidizing primary producers support this ecosystem. Beyond identification of bacterial cells and species, mobilization of Fe(II)aq and changes in Fe isotopic signatures have also been identified at this interface, consistent with bacterial inputs of protons and organic ligands (Buss et al., 2010). Fe + Mn precipitates have also been identified in the outer rindlets (see for example, the Fe–Mn precipitate shown in Fig. 3h). Such precipitates, not observed in the innermost rinds, document downward infiltration of organic-rich, acidic, Fe, Mn-charged fluids into the spheroidally weathering rock. All of these lines of evidence are consistent with infiltration of advecting meteoric fluids into the rindlet zone. 6. CONCLUSIONS Neutron scattering was used to measure changes in nano-porosity and specific surface area associated with initial weathering of two low-porosity, igneous rocks. Before weathering, the pore networks of both rocks are described as a combination of mass and surface fractals. In both rocks, the transition between mass and surface fractal behavior occurs for scattering features of dimension d  1 lm. In both cases, the mass fractal is an object that can be conceptualized as the network of pores whereas the surface fractal is an object that can be conceptualized as the nanometer-scale bumpiness of the pores. One of the rocks (the basaltic andesite) dissolves without spheroidal weathering: in this rock, the dominant mechanism of transport is diffusion. During diffusive weathering of basaltic andesite, porosity and specific surface area increase relatively consistently across the mm-thick reaction front of plagioclase dissolution. The mass and surface fractal behavior is mostly retained until rind disaggregation, but the fractal dimensions decrease with weathering. These changes are consistent with increasing connectivity of pores without significant roughening of the pore surfaces. Lack of roughening during dissolution of crystalline surfaces is consistent with dissolution close to equilibrium, a feature consistent with transport limited behavior.

411

The second rock is a spheroidally weathered quartz diorite which shows evidence for both chemical reactions and fracturing across the 40 cm-thick reaction zone for plagioclase. The reaction front consists of 2.5-cm thick rindlets subtended by spheroidal fractures that wrap the granitic corestone like onionskin. In addition to the rindlet-defining fractures, the rindlets themselves are criss-crossed by microfractures imaged in optical and electron microscopy. Across this zone of micro-fractured rindlets, the rock transforms from a mass + surface fractal into three surface fractals. In the rindlets, the specific surface area and porosity determined from SANS and USANS does not increase significantly until actual grain-scale disaggregation begins. The transformation of the corestone from a mass + surface fractal to several surface fractals, the increase in the surface fractal dimension measured at high Q during weathering, and the observation of precipitated Fe and Mn in some rindlets are all taken together as evidence that the transport mechanism for reactants and products in this weathering rock is advective infiltration + diffusion rather than just diffusion. Nonetheless, it is inferred that the diffusion-limited weathering subzones within the rindlets are somewhat similar to the plagioclase reaction zone in the basaltic andesite. Thus, the spheroidal weathering in the quartz diorite leads to advective infiltration across many fracture interfaces where diffusion-limited weathering can occur. In contrast, weathering in the non-fractured basaltic andesite is diffusion-limited across a single interface, the core–rind boundary. The fracturing translates to a faster weathering advance rate in the quartz diorite compared to the basaltic andesite. All of these observations yield a more mechanistic picture of how porosity opens up inside igneous rocks during weathering, leading to solubilization of nutrients and storage of water that can nurture and grow terrestrial ecosystems. We have demonstrated here that neutron scattering provides unprecedented, quantitative information about the nano- to micro-scale structure of pore networks in rocks and the evolution of porosity and pore/mineral surface area with chemical reaction. The linkage between physical pore space and geochemical reaction has long been recognized in weathering, leaching, pseudomorphic replacement, diagenetic alteration, and metamorphic processes (e.g., Putnis and Mauthe, 2001; Putnis, 2002; Labokta et al., 2004; Jamtveit and Hammer, 2012 – and references therein). Pore space provides both pathways for reactive fluid to reach mineral surfaces and space for minerals to grow and we now have evidence that fluid transport, mineral reaction, and accommodation happens at even the smallest length scales (nanometers). In a short amount of time we have learned much about the collection and interpretation of neutron scattering data from geochemically altered rocks (Anovitz et al., 2009; Jin et al., 2011; Bazilevskaya et al., in press). With continued application of neutron scattering in weathering systems (along with other advanced characterization techniques) we can learn much more about the chemical and physical processes that control opening and occlusion of the pore space so vital to geochemical reactions and more quantitatively connect the physics with the chemistry attendant with water–rock interactions.

412

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413 ACKNOWLEDGEMENTS

S.L.B. acknowledges NSF CHE-0431328 and DOE DE-FG0205ER15675; D.R.C. acknowledges support from the Basic Energy Sciences Energy Frontier Research Center “Nanoscale Control of Geologic CO2”. G.R. acknowledges support from the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences, U.S. Department of Energy. Oak Ridge National Laboratory is managed by UT-Battelle, LLC for the U.S. Department of Energy under Contract DE-AC05-00OR22725. D. Mildner, A. Jackson and the National Institute of Standards and Technology, U.S. Department of Commerce, are acknowledged for neutron beam support and facilities. Anonymous reviewers are acknowledged for their thoughtful and very helpful comments on an earlier version of this manuscript. H. Buss acknowledges support from the Luquillo Critical Zone Observatory and funding from NSF EAR-0722476 to F. Scatena (Univ. of Pennsylvania). We thank Jorgen Rosenqvist for performing BET analysis on the basaltic andesite samples at Oak Ridge National Laboratory.

REFERENCES Anderson S. P., Dietrich W. E. and Brimhall G. H. (2002) Weathering profiles, mass-balance analysis, and rates of solute loss: linkages between weathering and erosion in a small, steep catchment. Geol. Soc. Am. Bull. 114, 1143–1158. Anovitz L. M., Lynn G. W., Cole D. R., Rother G., Allard L. F., Hamilton W. A., Porcar L. and Kim M. H. (2009) A new approach to quantification of metamorphism using ultra-small and small angle neutron scattering. Geochim. Cosmochim. Acta 73, 7303–7324. Ball W. P., Beuhler C. H., Harmon T. C., MacKay D. M. and Robers P. V. (1990) Characterization of a sandy aquifer material at the grain scale. J. Contam. Hydrol. 5, 253–295. Bandstra J., Buss H. L., Campen R. K., Liermann L. J., Moore J., Hausrath E., Navarre-Sitchler A. K., Jang J.-H. and Brantley S. L. (2008) Appendix: compilation of mineral dissolution rates. In Kinetics of Water–Rock Interaction (eds. S. L. Brantley, J. D. Kubicki and A. F. White). Springer, New York. Barker J. G., Glinka C. J., Kim M. H., Moyer J. J., Drews A. R. and Agamalian M. (2005) Design and performance of a thermal-neutron double-crystal diffractometer for USANS at NIST. J. Appl. Crystallogr. 38, 1004–1011. Bazilevskaya E., Lebdeva M., Pavich M., Rother G., Parkinson D., Cole D. and Brantley S. L. (in press). Where fast weathering creates thin regolith and slow weathering creates thick regolith. Earth Surf. Processes Landforms. http://dx.doi.org/10.1002/ esp.3369. Brantley S. L. and Mellott N. P. (2000) Surface area and porosity of primary silicate minerals. Am. Mineral. 85, 1767–1783. Brantley S. L. (2008) Understanding soil time. Science 321, 1454– 1455. Buss H. L., Sak P., Webb R. M. and Brantley S. (2008) Weathering of the Rio Blanco quartz diorite, Luquillo Mountains, Puerto Rico: coupling oxidation, dissolution and fracturing. Geochim. Cosmochim. Acta 72, 4488–4507. Chabaux F. E., Blaes P., di Stille R., Chiara Roupert E., Pelt A., Dosseto L. Ma., Buss H. L. and Brantley S. L. (2013) Regolith formation rate from U-series nuclides: implications from the study of a spheroidal weathering profile in the Rio Icacos watershed (Puerto Rico). Geochim. Cosmochim. Acta 100, 73– 95. Fischer C. and Gaupp R. (2004) Multi-scale rock surface area quantification – a systematic method to evaluate the reactive surface area of rocks. Chem. Erde 64, 241–256.

FletcherR.C.,BussH.L.andBrantleyS.L.(2006)Aspheroidalweathering model coupling porewater chemistry to soil thicknesses during steadystate denudation. Earth Planet. Sci. Lett. 244, 444–457. Glinka C. J., Barker J. G., Hammouda B., Krueger S., Moyer J. J. and Orts W. J. (1998) The 30 m small-angle neutron scattering instruments at the National Institute of Standards and Technology. J. Appl. Crystallogr. 31, 430–445. Graham R., Rossi A. and Hubbert K. (2010) Rock to regolith conversion: producing hospitable substrates for terrestrial ecosystems. GSA Today 20, 4–9. Hinde A. L. (2004) PRINSAS – a Windows-based computer program for the processing and interpretation of small-angle scattering data tailored to the analysis of sedimentary rocks. J. Appl. Crystallogr. 37, 1020–1024. Hochella M. F. and Banfield J. F. (1995) Chemical weatheirng of silicates in nature: a microscopic perspective with theoretical considerations. In Chemical Weathering Rates of Silicate Minerals (eds. A. F. White and S. Brantley). Mineralogical Society of America, Washington, DC. Hochella M. F., Ponader H. B., Turner A. M. and Harris D. W. (1988) The complexity of mineral dissolution as viewed by high resolution scanning Auger microscopy. Geochim. Cosmochim. Acta 52, 385–394. Labokta T., Cole D. R., Fayek M., Riciputi L. and Stadermann F. (2004) Coupled cation and oxygen isotope exchange between alkali feldspar and aqueous chloride solution. Am. Mineral. 89, 1822. Jamtveit B. and Hammer O. (2012) Sculpting of rocks by reactive fluids. Geochem. Perspect. 1, 341–481. Jin L., Rother G., Cole D. R., Mildner D., Duffy F. R. and Brantley S. L. (2011) Characterization of deep weathering and nanoporosity development in shale – a neutron study. Am. Mineral. 96, 498–512. Kahle A., Winkler B., Radulescu A. and Schreuer J. (2004) Smallangle neutron scattering study of volcanic rocks. Eur. J. Mineral. 16, 407–417. Kline S. R. (2006) Reduction and analysis of SANS and USANS data using IGOR Pro. J. Appl. Crystallogr. 39, 895–900. Lee V. W., Mackwell S. J. and Brantley S. L. (1991) The effect of fluid chemistry on wetting textures in novaculite. J. Geophys. Res. 96(023–010), 037. Meunier A., Sardini P., Robinet J. C. and Pret D. (2007) The petrography of weathering processes: facts and outlooks. Clay Mineral. 42, 415–435. Mildner D. F. R., Rezvani R., Hall P. L. and Borst R. L. (1986) Small-angle scattering of shaly rocks with fractal pore interfaces. Appl. Phys. Lett. 48, 1314–1316. Minyard M. L., Bruns M. A., Martinez C. E., Liermann L. J., Buss H. L. and Brantley S. L. (2011) Halloysite nanotubes and bacteria at the saprolite, bedrock interface, Rio Icacos watershed, Puerto Rico. Soil Sci. Soc. Am. J. 75, 348–356. Minyard M. L., Bruns M. A., Liermann L. J., Buss H. L. and Brantley S. L. (2012) Bacterial associations with weathering minerals at the regolith–bedrock interface, Luquillo Experimental Forest, Puerto Rico. Geomicrobiol. J. 29, 792–803. http://dx.doi.org/10.1080/01490451.2011.619640. Montgomery C. W. and Brace W. F. (1975) Micropores in plagioclase. Contrib. Mineral. Petrol. 52. Murphy S. F., Brantley S. L., Blum A. E., White A. F. and Dong H. (1998) Chemical weathering in a tropical watershed, Luquillo Mountain, Puerto Rico: I. Rate and mechanism of biotite weathering. Geochim. Cosmochim. Acta 62, 2404. Navarre-Sitchler A. and Brantley S. (2007) Basalt weathering across scales. Earth Planet. Sci. Lett. 261, 321–334. Navarre-Sitchler A., Steefel C. I., Sak P. B. and Brantley S. L. (2011) A reactive-transport model for weathering rind formation on basalt. Geochim. Cosmochim. Acta 75, 7644–7667.

A.K. Navarre-Sitchler et al. / Geochimica et Cosmochimica Acta 109 (2013) 400–413 Navarre-Sitchler A., Steefel C. I., Yang L. and Tomutsa L. (2009) Evolution of porosity and diffusivity associated with chemical weathering of a basalt clast. J. Geophys. Res. Earth Surf. 114, F02016. Oguchi C. and Matsukura Y. (1999) Effect of porosity on the increase in weathering-rind thicknesses of andesite gravel. Eng. Geol. 55, 77–89. Pelt E., Chabaux F., Innocent C., Navarre-Sitchler A., Sak P. and Brantley S. (2008) Uranium–throium chronometry of weathering rinds: rock alteration rate and paleo-isotopic record of weathering fluids. Earth Planet. Sci. Lett. 276, 98–105. Putnis A. (2002) Mineral replacement reactions: from macroscopic observations to microscopic mechanisms. Mineral. Mag. 66, 689–708. Radlinski A. P. (2006) Small-angle neutron scattering and the microstructure of rocks. In Neutron Scattering in Earth Sciences (ed. H. Wenk). Mineralogical Society of America, pp. 363–397. Rufe E. and Hochella M. F. (1999) Quantitative assessment of reactive surfacea rea of phlogopite during acid dissolution. Science 285, 874–876. Sak P. B., Fisher D. M., Gardner T. W., Murphy K. and Brantley S. L. (2004) Rates of weathering rind formation on Costa Rican basalt. Geochim. Cosmochim. Acta 68, 1453–1472. Stanley H. E. (1991) Fractals and multifractals: the interplay of physics and geometry. In Fractals and Disordered Systems (eds.

413

A. Bunde and S. Havlin). Springer-VerlagBerlin Heidelberg, pp. 1–45. Torn M. S., Trumbore S. E., Chadwick O. A., Vitousek P. M. and Hendricks D. M. (1997) Mineral control of soil organic carbon storage and turnover. Nature 389, 170–173. Turner B. F., Stallard R. F. and Brantley S. L. (2003) Investigation of in situ weatheirng of the quartz diorite bedrock in the Rio Icacos basin, Luquillo Experimental Forest, Puerto Rico. Chem. Geol. 202, 313–341. White A. F., Blum A. E., Schulz M. S., Bullen T. D., Harden J. W. and Peterson M. L. (1996) Chemical weathering rates of a soil chronosequence on granitic alluvium: I. Quantification of mineralogical and surface area changes and calculation of primary silicate reaction rates. Geochim. Cosmochim. Acta 60, 2533–2550. White A. F., Blum A. E., Schulz M. S., Vivit D. V., Stonestrom D. A., Larsen M. C., Murphy S. F. and Eberl D. (1998) Chemical weathering in a tropical watershed, Luquillo Mountains, Puerto Rico: I. Long-term versus short-term weathering fluxes. Geochim. Cosmochim. Acta 62, 209–226. White A. F. and Brantley S. L. (2003) The effect of time on the weathering of silicate minerals: why do weathering rates differ in the laboratory and field? Chem. Geol. 202, 479–506. Associate editor: Jiwchar Ganor