American Mineralogist, Volume 85, pages 17671783, 2000

Surface area and porosity of primary silicate minerals

SUSAN L. BRANTLEY AND NATHAN P. MELLOTT*
Department of Geosciences, Pennsylvania State University, University Park, Pennsylvania 16802, U.S.A.

ABSTRACT
Surface area is important in quantifying mineral-water reaction rates. Specific surface area (SSA) was measured to investigate controls on this parameter for several primary silicate minerals (PSM) used to estimate rates of weathering. The SSA measured by gas adsorption for a given particle size of relatively impurity-free, laboratory-ground samples generally increases in the order: quartz olivine albite < oligoclase bytownite < hornblende diopside. Reproducibility of BET SSA values range from 70% (SSA < 1000 cm2/g) to 5% (SSA > 4000 cm2/g) and values measured with N2 were observed to be up to 50% larger than values measured with Kr. For laboratory-ground Amelia albite and San Carlos olivine, SSA can be calculated using log (SSA, cm2/g) = b + m log (d), where d = grain diameter (m), b = 5.2 0.2 and m = 1.0 0.1. A similar equation was previously published for laboratory-ground quartz. Some other samples showed SSA higher than predicted by these equations. In some cases, high SSA is attributed to significant second phase particulate content, but for other laboratory-ground samples, high SSA increased with observed hysteresis in the adsorption-desorption isotherms. Such hysteresis is consistent with the presence of pores with diameters in the range 2 to 50 nm (mesopores). In particular, porosity that contributes to BET-measured SSA is inferred for examples of laboratory-ground diopside, hornblende, and all compositions of plagioclase except albite, plus naturally weathered quartz, plagioclase, and potassium feldspar. Previous workers documented similar porosity in laboratory-ground potassium feldspar. Surface area measured by gas adsorption may not be appropriate for extrapolation of interfacelimited rates of dissolution of many silicates if internal surface is present and if it does not dissolve equivalently to external surface. In addition, the large errors associated in measuring SSA of coarse and/or impurity-containing silicates suggest that surface area-normalized kinetics in both field and laboratory systems will be difficult to estimate precisely. Quantification of the porosity in laboratory-ground and naturally weathered samples may help to alleviate some of the discrepancy between laboratory- and field-based estimates of weathering rate.

INTRODUCTION
Laboratory-ground and naturally weathered primary silicates show generally similar trends in specific surface area (SSA) as a function of mineralogy. White (1995) and White et al. (1996) concluded that SSA of naturally weathered PSMs generally increases for a given grain size from quartz to potassium feldspar (denoted K-spar) to plagioclase, and a similar trend was observed for SSA of laboratory-ground samples based upon a comparison of literature data by Brantley et al. (1999). This increasing SSA parallels an increase in rate of dissolution across the same compositional trend (Brantley et al. 1999), but little is known regarding the effects of mineralogy on SSA in laboratory-ground or naturally weathered silicate grains. Although trends in SSA of laboratory-ground and naturally weathered silicates may be similar, many studies of weathered soil silicate grains (e.g., White and Peterson 1990; Anbeek 1992a, 1992b, 1993; Anbeek et al. 1994) have shown that specific surface area of silicates is larger for naturally weathered as compared to laboratory-ground samples. Some of this in* E-mail: npm113@psu.edu
0003-004X/00/11121767$05.00

creased surface area was attributed to increased roughness of the external surface. However, some of the increased surface area may be due to internal surface related to porosity developed during crystallization or during solution etching (e.g., Montgomery and Brace 1975; Worden et al. 1990; Anbeek 1992a, 1992b, 1993; Walker et al. 1995; Lee and Parsons 1995; White 1995). The contribution of surface area from porosity may be measured using standard techniques of nitrogen adsorption. Where investigators have used nitrogen adsorption isotherms to infer pores of diameter 250 nm (mesopores), the evidence suggests that such mesoporosity is a significant contributor to specific surface area (SSA) in many naturally weathered samples. For example, Titley et al. (1987), Mayer (1994), and Werth and Reinhard (1997) concluded that pores contribute significantly to surface area of several estuarine and marine sediment samples and several soils. Porosity in weathered quartz and feldspar was similarly documented by Ball et al. (1990), Wood et al. (1990), and Yau (1999) using gas adsorption, mercury porosimetry, and diffusion analysis on glacial outwash sediments. Although mesoporosity was not specifically measured, White and coworkers also inferred the presence of porosity in

1767

1768

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

weathered samples of plagioclase, K-spar, and hornblende based upon microscopic observations and variations in surface area measured by gas adsorption as a function of grain size for soils from Puerto Rico (Schulz et al. 1999) and Merced, California, (White et al. 1996). In addition, porosity in alkali feldspars has been documented by many workers (e.g., Worden et al. 1990; Walker et al. 1995; Lee and Parsons 1995). Pores have also been documented in laboratory-ground silicates: for example, Hodson and coworkers (Hodson et al. 1997; Hodson 1998) documented porosity in potassium feldspar and Brantley et al. (1999) documented mesoporosity in hornblende. A better understanding of the controls on mineral surface area and porosity in surficial environments is important for several reasons. For example, if internal surface area (e.g., Gregg and Sing 1982; Hochella and Banfield 1995) is present in primary silicate minerals, and if internal surface area is more or less reactive than external surface area, then quantification of the ratio of internal vs. external surface area could be important in the extrapolation of mineral-water kinetics from one system to another (see, for example, Lee and Parsons 1995; Hodson 1998). Whereas several authors have suggested that weathered and laboratory-ground samples differ with respect to the relative importance of internal and external area (e.g., Anbeek 1992a, 1992b, 1993; Anbeek et al. 1994), few have quantified the contribution from these two types of surface for minerals which are commonly used in comparisons of field and laboratory weathering rates (e.g., plagioclase, hornblende, olivine, diopside). Specifically, if the ratio of these two types of surface area differ between laboratory-ground and naturally weathered plagioclase, hornblende, olivine, and diopside, then perhaps these differences contribute to the large (up to five orders of magnitude) discrepancies between laboratory- and field-derived dissolution rates for these minerals (White and Brantley 1995). In addition, if porosity is present in silicate grains then the observed aging or sequestration of contaminants in soils and aquifers may be due to sequestration inside these pores (e.g., Farrell and Reinhard 1994; Werth and Reinhard 1997; Luthy et al. 1997). In a series of papers, Mayer (1994, 1999) also proposed that sequestration of organic matter into small pores may allow preservation of the organic matter on continental shelf sediments since hydrolytic enzymes should be excluded from such small pores. Again, however, little is known about mineral controls on the presence of porosity in the silicates making up the sediments. This paper addresses the following questions concerning SSA of primary silicates: What is the range of SSA measured using BET for laboratory-ground primary silicate powders? Can we infer the presence of porosity from adsorption data for laboratory-ground and weathered primary silicates? What do these observations imply for reaction kinetics in the laboratory and in the field? Our approach was to investigate relatively pristine mineral samples that have traditionally been used by geochemists to investigate mineral dissolution (samples largely derived from pegmatite deposits) using standard BET techniques for analysis of gas adsorption-desorption. We also used the same BET technique on seven naturally weathered samples to infer the presence or absence of porosity.

BACKGROUND
The specific geometric surface area, SSAgeo, can be expressed as a function of grain diameter d: SSAgeo = adD3 (1)

where is the density of the solid and a is a geometric parameter. For Euclidean solids (i.e., solids with no fractal properties), D = 2. Adsorption of gas on to a powder surface is extensively used to measure powder surface area (Gregg and Singh 1982; Lowell and Shields 1991). The ratio of the measured surface area using gas adsorption, SSAads, to the geometric surface area, SSAgeo, has typically been called the surface roughness, , by geochemists (Helgeson et al. 1984): = SSAads/SSAgeo (2)

The surface roughness differs from 1 because of surface topography and porosity. White (1995) and White et al. (1996) produced a general expression for the physical surface of a mineral grain in terms of external surface roughness, ext, and internal surface area, SSAint: SSAads = ext a/(d) + SSAint (3)

This equation can be applied to a suite of samples of varying grain size if surface roughness and internal surface area are both assumed to be independent of grain size.

METHODS
Samples The minerals were used previously to measure dissolution kinetics (e.g., White and Brantley 1995). Primary silicate minerals include examples of anorthite (Miyake Jima, Japan), anorthite (Grass Valley, California, provided by R. Holdren, Pacific Northwest National Laboratory, Washington), albite (Amelia Courthouse, Virginia), diopside (Herschel, Ontario, Canada), oligoclase (Madawaska, Ontario, Canada), labradorite (Labrador, Canada), bytownite (Duluth, Minnesota), olivine (San Carlos, Arizona), olivine (Twin Sisters Range, Washington), and hornblende (Gore Mountain, New York). Where specific collections are not indicated above, samples were obtained from Wards Scientific Inc. Glass of albite composition was also prepared by melting Amelia albite (Hamilton 1999). The Grass Valley anorthite was investigated by Holdren and Speyer (1987); sample preparation is described in that paper. Although we review some literature data for microcline, we make no attempt to summarize or investigate the potassium feldspars and we refer the reader to other publications (e.g., Worden et al. 1990; Walker et al. 1995; Lee and Parsons 1995; Hodson et al. 1997; Hodson 1998). Sample fragments were ground with an agate mortar and pestle and dry sieved until 1 to 5 g of each of the following size fractions were collected: 2035 mesh (840500 m), 3560 mesh (500250 m), 140200 mesh (10574 m), 230270 mesh (6253 m), 270325 mesh (5344 m), and 325400

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

1769

mesh (4437 m). For some samples, additional size fractions (e.g., mesh size 100200 or 75150 m) were also collected. Data for a labradorite, summarized by Brantley et al. (1999), are here included for completeness. Each size fraction of this sample and the Miyake Jima anorthite were also run through a magnetic separator up to three times to attempt to remove impurities. All powders were ultrasonically cleaned in deionized water several times for 69 minutes until supernatant fluid contained little if any fine material, and rinsed with Optima Acetone (Fisher Scientific) before surface analysis. Bulk chemical analysis or electron microprobe analysis was completed for samples lacking published analysis (Table 1). Grain mounts were prepared, polished, and carbon-coated for analysis with a Cameca SX-50 electron microprobe. In some cases, grain mounts were imaged on a Philips XL-20 scanning electron microscope (SEM). For several minerals in Table 1, small amounts of disseminated second phase inclusions are present. In other cases, second phase particulates were found in samples, despite careful cleaning (e.g., Grass Valley and Miyake Jima anorthite, Twin Sisters olivine). Adsorption-desorption isotherms were also measured on several weathered samples from the Merced soil chronosequence (White 1995; White et al. 1996). Samples described by White (1995) of the 500 to 1000 m fraction of quartz, potassium feldspar, and plagioclase from two soils (Modesto sampled at a depth of 413 cm, Turlock Lake sampled at a depth of 375 cm) were analyzed. Samples of quartz were also analyzed from the China Hat soil from the same chronosequence. These samples were the identical samples, cleaned using a citrate-dithionite extraction, analyzed by White et al. (1996). The ages of the surfaces of these soils were 40 ka (Modesto), 600 ka (Turlock Lake), and 3000 ka (China Hat). Surface area measurement SSA was determined with a Micromeritics ASAP 2010 surface area analyzer. Unless noted otherwise, ~1 to 5 g of sample

were degassed at 300 C for 412 h (until the pressure of the system reached 3040 mm Hg) before Kr measurement. In most cases, Kr and He were used as the analysis and backfill gases, respectively. Kr is recommended for determination of SSAs as low as 10 cm2/g, and is thus preferred over argon or nitrogen for PSM, which characteristically have low specific surface areas. Where noted, N2 was used as the adsorbate. Multi-point surface area (8 points recorded) was calculated for relative pressures of 0.010 to 0.25, and adsorption isotherms were drawn using the Brunauer, Emmet, and Teller (BET) method (Brunauer et al. 1938; Lowell and Shields 1991). A commercially available alumina-silica nonporous reference powder (P/N no. 004/ 16816/00, 46F-BA-106-6, available from Micromeritics with quoted surface area of 4600 cm2/g) was run periodically as a standard. The BET equation (Gregg and Sing 1982) models the measured number of moles, n, adsorbed on 1 g of adsorbent:

p / p0 c 1 1 = + ( p / p0 ) n(1 p / p0 ) nm c nm c

(4)

where nm is the calculated number of moles adsorbed as a monolayer on 1 g of adsorbent, p is the gas pressure, p0 is the saturation vapor pressure of the gas, and c equals: c = exp(H/RT). (5)

Here H is related to the net heat of adsorption of the gas on the surface, R is the gas constant, and T is the absolute temperature. In practice, (p/p0)/n(1p/p0) is plotted against p/p0, yielding values for (c1)/nmc (the slope, s) and 1/nm c (the intercept, i) for values between p/p0 = 0.05 and 0.3 (Webb and Orr 1997). The values of nm of SSA can then be calculated: nm = (s + i)1 (6)

TABLE 1. Composition of samples

Weight % oxides Estimated* Analysis SiO2 Na2O CaO FeO Fe2O3 Al2O3 MgO K2 O H2O (LOI) % method Albite (Amelia) 68.12 11.50 0.39 N/A N/A 19.59 N/A 0.20 N/A rare EM Oligoclase (Madawaska) 62.90 9.01 4.02 N/A 0.07 22.40 0.05 0.70 N/A N/A Bytownite (Duluth) 48.81 2.57 15.49 0.38 N/A 32.82 N/A N/A N/A rare EM Anorthite (Miyake Jima) 43.45 0.41 18.91 0.44 N/A 37.45 N/A 0.02 N/A 12 EM Anorthite (Grass Valley) 44.78 0.89 18.52 0.59 N/A 34.53 0.02 0.02 N/A >5 EM Forsteritic olivine (San Carlos)|| 40.72 N/A N/A 9.21 N/A N/A 49.24 N/A N/A rare XRF Olivine (Twin Sisters) 40.27 N/A N/A 8.26 N/A N/A 47.46 N/A N/A 25 EM Enstatite# 57.21 N/A 0.37 5.58 N/A N/A 36.65 N/A N/A EM Diopside (Herschel) 54.41 0.25 24.85 2.18 N/A 0.43 17.14 N/A N/A rare EM Hydrated Phase** 55.72 0.59 13.00 3.25 N/A 2.15 22.17 N/A N/A EM Hornblende (Gore Mt.) 46.00 3.16 8.34 6.9 7.66 17.60 10.20 0.40 1.65 5 XRF Hydrated Phase** 35.08 0.04 0.31 18.71 N/A 10.72 21.53 0.01 N/A EM Notes: N/A = not analyzed; rare = only a few or no second phase inclusions were observed; LOI = loss on ignition. * % of impurity, estimated visually from back-scattered electron photomicrograph or based upon literature descriptions. Analysis from Stillings and Brantley (1995). Sample contained dark red to gray secondary particulates that could not be probed (probably a Mg-Ca-Fe oxide). Black glassy inclusions were also present along grain boundaries. Analysis from Casey et al. (1991). Sample was provided by R. Holdren, PNNL. Sample contained significant iron-rich mineral (Holdren, personal communication). || Analysis from Wogelius and Walther (1992). # Second phase particulates observed with the previous phase. ** Impurity inclusions observed with the previous phase. Mineral and location

1770

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

SSA = Lnmam

(7)

where am is the area occupied by the adsorbate molecule (a constant for a given gas, calculated theoretically from the liquid density) in the monolayer and L is Avogadros number. Porosity measurement Adsorption-desorption hysteresis was measured for several minerals and for one blank (no sample) run with the Micromeritics ASAP 2010 surface analyzer. Adsorption-desorption isotherms and BET SSAs were determined using 1 to 5 g of 100200 mesh sample, with N2 and He as the analysis and backfill gases. Kr is not used for hysteresis measurements. We use the IUPAC (1972) convention for nomenclature of porosity even though other conventions have been used in the mineralogical literature (e.g., Montgomery and Brace 1975; Worden et al. 1990; Walker et al. 1995). By IUPAC convention, pores are characterized by their diameters, where micro- (<2 nm), meso- (250 nm), and macropores (>50 nm) are identified. The shape of the isotherm (see Lowell and Shields 1991; Gregg and Sing 1982; Hodson 1998) for PSMs in our experience can be classified as Type I (interpreted to contain significant microporosity), Type II (typically interpreted as nonporous), or Type IV (interpreted to contain significant mesoporosity). In brief, microporous samples (Type I) show enhanced adsorption below p/p0 = 0.10.2 and an adsorption plateau above this value (Lowell and Shields 1991). In addition, a value of c which is high (>100) or negative indicates micropores (Webb and Orr 1997). Conceptually, a Type I isotherm documents adsorption of gas into micropores at very low gas pressures. Mesoporous samples generally show hysteresis in the adsorption and desorption branches above p/p0 = 0.4, attributed to multilayer formation and, especially, capillary condensation in mesopores (Type IV). Nonporous samples show neither type of enhanced adsorption, and instead resemble curves for nonporous standards (Type II, no hysteresis, no enhanced adsorption). Combinations of pore types can make interpretation of isotherm shape difficult: for example, Gregg and Sing (1982) point out that a Type II isotherm can still have microporosity. For a Type II isotherm measured on a microporous sample, however, adsorption will be exceptionally high in the low pressure region and the value of c will be anomalous (several hundreds or negative). For such a case, the SSA calculated from BET analysis will be erroneously high (Eq. 4 cannot be used, Webb and Orr 1997). The presence of porosity is also documented using t-plots comparing the volume of nitrogen adsorbed on a sample to that adsorbed on a nonporous standard. The t-value is the statistical thickness of the adsorbed gas layer as calculated at a given value of p/p0 for a nonporous standard adsorbent (Webb and Orr 1997). The volume of microporosity can be determined from the intercept of the first straight line segment (Brunauer 1969; Webb and Orr 1997; Hodson 1998). If the sample isotherm is identical with that of the standard, the t-plot will be linear and pass through the origin. Gregg and Sing (1982) point out that a t-curve should be based upon a standard sample that is chemically similar to the study sample. For example, a standard sample could be chosen such that the BET c value of the

standard and sample are similar (Brunauer 1969). Because the t-plot relies upon a model parameter, nm, Sing (1969) advocates using a different method, the -plot, as presented here to compare standard isotherms. In essence, t- and -plots are identical, representing comparisons between the sample and a nonporous standard. Sing (1969) defines = n/ n0.4, where n is the number of moles of gas adsorbed at any relative pressure, and n0.4 is the number of moles adsorbed at p/ p0 = 0.4. In effect, equals the moles gas adsorbed normalized by the moles adsorbed at a relative pressure where mesoporosity should start causing enhanced adsorption. These curves show volume of gas adsorbed on a sample for a given p/p0 plotted vs. the value for a standard for the same p/p0. Both and t-plots have the same shape, but axes on each plot are scaled differently (Webb and Orr 1997). If the standard and the sample are both nonporous, then the (or t) plot should be linear and should pass through the origin. Microporous samples will show an (or t) plot with a large slope for << 1 (p/p0 < 0.2), and then a decrease in slope above these relative pressures: the entire plot is concave downward. Mesoporous samples show a linear portion through the origin and an increase in slope at relative pressures near = 1 where capillary condensation into mesopores occurs. Pore size distributions for samples with porosity can be calculated from adsorption data by inferring a pore geometry and using the BJH method (Barrett et al. 1951). This model assumes a geometry and apportions all surface area to pores of varying radii (Lowell and Shields 1991; Micromeritics 1995).

RESULTS
Values of surface area and observed reproducibility Values of SSA for measured samples varied from 90 (coarse albite glass powder) to 18 000 (fine anorthite powder) cm2/g (Table 2, Figs. 1 and 2). SSA generally increased as grain size decreased (Figs. 1 and 2). Two compositions generally showed lower SSA for given grain sizes than others: Amelia albite and San Carlos olivine. In contrast, hornblende, Miyake-Jima anorthite, and diopside exhibited generally high SSA. Reproducibility (number of measurements = N = 4) of the standard powder (SSA 5000 cm2/g) over five months of analysis (Table 2) yielded <2% RSD (with Kr) and 7% RSD (with N2). For PSM powders with SSA > 4000 cm2/g, reproducibility was similar. On the other hand, measurements of PSM showed considerable discrepancy for lower values of SSA (grain sizes > 100 m, Table 2). The average of the values of the surface area of the Si- and Al-containing standard measured using N2 was 40% higher than the average measured using Kr. Similar comparisons for PSMs showed a range wherein surface areas measured with N2 were 2050% larger than those measured by Kr. The value measured for the standard using Kr was within 4% of the accepted value. Porosity in primary silicate grains The volumes of gas both adsorbed and desorbed per grams of powder were measured for albite, bytownite, anorthite (Miyake-Jima and Grass Valley), olivine (Twin Sisters and San Carlos), diopside, and hornblende (e.g., Figs. 3 and 4), as well

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES TABLE 2. BET surface area results
Name Albite (Amelia) Gas Kr Kr Kr Kr Kr Kr Kr Kr N2 Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr N2 Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr N2 N2 Ar Grain size (m) 840500 500250 10575 6253 5344 4437 4437 4437 15075 840500 500250 10575 6253 5344 4437 840500 500250 10575 6253 5344 4437 505297 297250 250210 210149 149105 10574 7453 5344 840500 500250 15075 10575 6253 5344 4437 15075 840500 840500 840500 500250 500250 500250 10574 6253 5344 4437 15075 7537 7537 7537 840500 500250 15075 10575 6253 5344 4437 15075 Mean diameter* (m) 648 354 88 57 48 40 40 40 106 648 354 88 57 48 40 648 354 88 57 48 40 387 272 229 177 125 88 63 48 648 354 106 88 57 48 40 106 648 648 648 354 354 354 88 57 48 40 106 53 53 53 648 354 106 88 57 48 40 106 Mass for BET g 1.97 1.45 1.67 1.67 1.18 1.18 1.57 3.16 1.55 1.75 1.61 1.41 1.22 1.06 1.48 1.35 1.28 1.16 0.94 0.98 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 1.18 1.88 1.73 1.47 1.11 1.02 1.14 4.85 1.52 1.47 1.61 1.92 1.39 1.48 1.23 0.98 2.50 1.77 ? SSA (cm2/g) 225 497 1000 1670 2520 5450 5570 1190 86 197 838 1800 2680 6000 927 1080 1500 2760 2680 7260 1460 1230 1690 1620 1720 2190 3150 4310 1020 1180 1530 2380 2790 5440 1280 1850 414 750 41% 1200 420 68% 2830 7290 12000 17900 1750 6550 10,200 31% 1.25 1.70 2.35 1.20 1.12 1.24 1.03 3.85 223 329 1040 3550 3050 4620 520 1500 ~0.2 ~0.2 ~0.2 ~0.2 ~0.2 ~0.2 ~0.5 ~0.5 1.6 1.6 1.6 1.6 2.0 >12 y ? 1.6 2.0 RSD (%) Time|| ~4.5 ~4.5 ~4.5 ~4.5 ~4.5 ~4.5 ~4.5 2% ~4.5 ~4.2 ~4.2 ~4.2 ~4.2 ~4.2 ~4.2 ~0.2 ~0.2 ~0.2 ~0.5 ~0.5 ~0.5 ~2.0 ~2.0 ~2.0 ~2.0 ~2.0 ~2.0 ~2.0 ~2.0 ~0.2 ~0.2 ~0.2 ~0.2 ~0.2 0.03 ~0.5 ~0.5 1.5 1.7

1771

Albite Glass

Oligoclase (Madawaska)

Labradorite (Labrador)

Bytownite (Duluth)

Anorthite (Miyake-Jima)

Anorthite (Grass Valley) Anorthite (Grass Valley)

Olivine (San Carlos)

Kr Kr Kr Kr Kr Kr Kr N2

Olivine (Twin Sisters)

Kr 840500 648 1.67 3190 ~4.8 Kr 500250 354 1.85 2340 ~4.5 Kr 150125 136 1.69 2290 ~4.8 Kr 10574 88 1.99 1850 ~4.5 * Diameter calculated as the antilog of the mean of the log (maximum) and log (minimum) diameters. This calculation assumes that the particle distribution is log-normal. Where multiple measurements were made, the relative standard deviation is noted (see text). Original sample as reported by Holdren and Speyer (1987) measured in our laboratory in 1998. Same sample as above, measured and reported by Holdren and Speyer (1987). || Approximate time since grinding (months unless otherwise noted).

1772 TABLE 2Continued

Name

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

Gas Kr Kr

Grain size (m) 7444 4437 840500 500250 150125 125105 15075 10575 6253 5344 4437 15075 840500 840500 840500 500250 250150 15075 15075 15075 15075 15076 10575 6253 7537 53-44 4437 4437 4437 15075 15075 15075

Mean diameter* (m) 57 40 648 354 136 115 106 88 57 48 40 106 648 648 648 354 194 106 106 106 106 106 88 57 53 48 40 40 40 106 106 106

Mass for BET g 1.57 1.40 1.86 1.78 1.61 1.35 2.11 1.65 1.16 1.22 1.11 4.50 1.65 1.64 1.16 5.00 5.00 1.25 4.50 10.78 1.53 1.29 5.00 1.23 1.01 1.02 5.00 4.48

SSA (cm2/g) 4940 8100 1770 2020 2010 2360 1850 2620 4810 4910 13700 2750 1850 1580

RSD (%)

Time|| ~4.8 ~4.5 ~0.2 ~0.2 ~0.2 ~0.2 ~0.2 ~0.2 ~0.2 ~0.2 ~0.8 ~0.5 ~17 ~20

Diopside(Herschel)

Kr Kr Kr Kr Kr Kr Kr Kr Kr N2 Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr Kr N2 N2 N2

Hornblende (Gore Mtn.)

11% 1730 2110 3270 1930 2160 2500 24% 2820 5890 9240 10300 9120 9210 1% 3150 3430 6% 4600 Value by Micromeritics ~3.5 ~17 0.5 ~1 0.6 0.6 0.6 ~4.5 ~17 ~1 ~1 ~17 ~17 ~17

Al/Si Standard

Kr 4790 Kr 1.77 4800 Kr 4750 Kr 1.60 4720 Kr 1.60 1% 2.52 6430 N2 2.36 7310 N2 N2 2.65 6360 2.98 6610 N2 N2 2.98 7% * Diameter calculated as the antilog of the mean of the log (maximum) and log (minimum) diameters. This calculation assumes that the particle distribution is log-normal. Where multiple measurements were made, the relative standard deviation is noted (see text). Original sample as reported by Holdren and Speyer (1987) measured in our laboratory in 1998. Same sample as above, measured and reported by Holdren and Speyer (1987). || Approximate time since grinding (months unless otherwise noted).

as for the soil grains. Hysteresis (where volume of gas adsorbed at a given relative pressure is not equal during adsorption and desorption) was observed for every laboratory-ground sample except olivine (San Carlos and Twin Sisters) and Amelia albite (Figs. 3 and 4). Observed curves with hysteresis can be classified as Type B based upon their geometries (Gregg and Sing 1982; Lowell and Shields 1991). Such curves have been classically interpreted to indicate the condensation of gas in slit-like mesoporosity (Gregg and Sing 1982). For all laboratory-ground minerals, adsorption for gas at p/p0 < 0.1 or 0.2 was small and the BET c value (Eqs. 4 and 5) was << 100 (Table 3). Adsorption-desorption measurements for the soil samples (Fig. 5) revealed hysteresis in all samples except for the quartz

from the youngest and oldest soil studied (all data not shown). For these naturally weathered samples, the BET c value was <100 for all soil grains except the Turlock Lake plagioclase grains. Some isotherms (Fig. 35) reveal less gas sorbed during desorption than during adsorption for relative pressures <0.4. Such a discrepancy is probably related to errors in pressure transducer linearity, zero drift, degassing from rubber in the system, and/or small leaks. One blank measurement (adsorption-desorption run without a powder sample) similarly showed such a discrepancy for all values of relative pressure < 0.95. This error is shown as characteristic error bars on Figure 3, 4, and 6, normalized by appropriate sample mass. Based upon the blank run, hysteresis is always underestimated because of the error in measurement.

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

1773

FIGURE 1. SSA plotted vs. grain size (micrometers): (a) microcline, (b) albite, (c) oligoclase, (d) labradorite, (e) bytownite, and (f) anorthite. Stars and crosses are used for BET (Ar), solid symbols for BET (N2), and open symbols for BET (Kr).

1774

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

FIGURE 2. SSA plotted vs. mean grain size for (a) olivine, (b) diopside, (c) hornblende. Solid symbols are used for BET (N2), and open symbols or open symbols with a cross/slash embedded for BET (Kr).

TABLE 3. Porosity results

% Porosity % Pore % Pore as area as area as mesoporosity mesoporosity microporosity Albite (Amelia) 2.8 37% 93% bd Bytownite (Duluth) 6.9 37% 80% 11% Anorthite (Miyake-Jima) 9.3 56% 80% 17% Anorthite (Grass Valley) 24.9 40% 76% 17% Olivine (San Carlos) 2.4 57% 81% 17% Diopside (Hershel) 9.5 41% 80% 12% Hornblende (Gore Mtn.) 8.3 51% 88% 8% Hornblende (Gore Mtn.) 22.1 26% 86% 8% Olivine (Twin Sisters) 5.7 56% 89% 8% Potassium Feldspar (Modesto) 14.8 22% 84% bd Potassium Feldspar (Turlock Lake) 41.6 36% 92% bd Plagioclase (Modesto) 19.2 47% 70% 25% Plagioclase (Turlock Lake) 60.7 54% 90% 6% Quartz (Modesto) 6.0 40% 87% 13% Quartz (Turlock Lake) 6.6 42% 77% 13% Quartz (China Hat) 4.2 51% 76% 20% * Cumulative pore volume as calculated using BJH model for adsorption isotherm (Micromeritics, 1995). Note that <3% of cumulative pore volume was present as microporosity for all samples. Mesoporosity from 2 to 40 nm as calculated using BJH model. Microporosity from 1.7 to 2.0 nm as calculated using BJH model. Mineral Phase Pore volume* 104 (cm3/g) c Time since grinding (months) 4.5 0.5 2 >12 y 0.5 0.5 17 3.5 0.25 field field field field field field field

34 27 49 43 11 28 37 59 43 65 90 27 143 14 27 17

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

1775

FIGURE 3. Volume of gas sorbed (at STP) during N2 adsorption and desorption plotted vs. relative pressure for: (a) albite and bytownite, (b) anorthite (Miyake-Jima) (c) anorthite (Grass Valley). Characteristic error bars are shown for a blank run on the same instrument at various values of relative pressure, normalized for the mass of sample used. This error was always positive, as shown. Volume of gas adsorbed plotted for albite and anorthite vs. the value for the same relative pressure as measured for hydroxylated silica (see text) (d). = 1 implies relative pressure of 0.4. All grains except Grass Valley anorthite were 75150 micrometers in diameter; GV anorthite grains were 3775 micrometers in diameter.

DISCUSSION
Reproducibility of SSA measurements Reproducibility of measurement of SSA for coarse grain sizes of PSMs was disappointing, considering that the SSA of the alumina-silica standard can be reproduced within <2% using Kr. The extremely low specific surface area of many PSMs yields measurements that are highly irreproducible. Blind studies summarized by Gregg and Sing (1982) suggest that factors contributing to discrepancies include lack of reproducible conditions of outgassing, temperature changes during adsorption measurements, inaccurate measurements of p0, contamination of adsorbate gases, and discrepancies in assumptions made with

respect to the BET model. Several other variables contribute to discrepancies, as discussed here. Eggleston et al. (1989) previously reported possible decreases in SSA for <75 m diopside powder over 250 d and suggested that the decrease could be due to healing of microcracks. Knauss et al. (1993) reported for the same mineral (75125 m) that heating 4 h at 200 C decreased the BET SSA from 610 12 cm2/g to 550 15 cm2/g, and that this same powder measured one year later after similar heating decreased to 490 14 cm2/g. Stillings and Brantley (1995) reported decreases in SSA for feldspar powder over year long periods, attributing the decrease to healing of microcracks induced by grinding. For the coarser hornblende samples reported here,

1776

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

FIGURE 4. Volume of gas sorbed (at STP) during N2 adsorption and desorption plotted vs. relative pressure for (a) olivine (b) diopside (c) hornblende. Characteristic error bars are shown for a blank run on the same instrument at various values of relative pressure, normalized for the mass of sample used. This error was always positive as shown. plot showing volume adsorbed for a sample vs. the value for the same relative pressure as measured for hydroxylated silica (see text) for (d) olivine, diopside, and hornblende. = 1 implies a relative pressure of 0.4. All grains were 75150 micrometers in diameter.

measured SSA decreased with increasing time since grinding (Table 2). For these samples, the calculated pore volume also decreased (Table 3). In contrast, Miyake Jima anorthite showed both an increase and a decrease in SSA with time since grinding for individual samples. Especially for samples with impurities, lack of reproducibility may be related to entrainment of fine second phase particles. For example, as indicated in Table 2, SSA of Miyake Jima anorthite was measured for a sample (75150 m) using N2 as 1750 cm2/g; another sample, ground separately to the same grain size and cleaned similarly to the first, revealed a SSA of 3760 cm2/g. This second sample, although cleaned similarly to the first sample (and not reported in the table), was visibly more contaminated with second phase

particles, consistent with the suggestion that high SSA of the Miyake Jima anorthite is at least partially explained by the presence of a second phase, and furthermore suggesting that sample preparation can drastically change measured SSA. The BET method measures surface area with a ruler equivalent to the area of the adsorbate molecule: for example, 16.2 vs. 15.2 A2 (N2 vs. Kr at 78 K, Gregg and Sing 1982). For this reason, and because of nonidealities in adsorption, different molecules yield different values for the SSABET, with Kr < N2 for most materials (Gregg and Sing 1982), and with Ar < N2 (Brunauer 1969). In agreement with this, we observed that BET (N2) surface areas were larger than Kr areas by up to 50% (Table 2). For the low SSAs observed for these samples, Kr is more reproducible than N2 (Table 2).

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

1777

FIGURE 5. Volume gas sorbed (at STP) during N2 adsorption and desorption plotted vs. relative pressure for plagioclase from the (a) Modesto and (b) Turlock Lake soil (see text). Soil grains were collected and prepared by White et al. (1996).

FIGURE 6. (a) Log (SSA) vs. log (diameter) plotted for selected minerals as measured in our laboratory. Estimated regression line for quartz derived from Parks (1990). (b) Volume of gas sorbed during desorption, Vdes, minus volume of gas sorbed during adsorption,Vads at p/p0 = 0.6. Error bars estimated for a blank run assuming 3 g of sample.

High SSA values The SSA generally increases in the order albite olivine < oligoclase bytownite < hornblende diopside (Fig. 6A). If all mineral samples had been ground identically for identical periods of time (probably not strictly valid), then the SSA of a given grain size should vary inversely with fracture energy of the sample. For a given expenditure of energy during grinding, the surface area created should be inversely proportional to the fracture energy. The data available contradict this hypothesis: if anything, SSA increases rather than decreases with increasing fracture energy (Atkinson and Meredith 1987). However, as shown in Figure 6, SSA of both laboratory-

ground and naturally weathered samples generally increases with increasing hysteresis in adsorption-desorption. For several of the samples, hysteresis in adsorption-desorption qualifies as Type B hysteresis which is commonly interpreted as evidence for slit-shaped meso- or macropores (Lowell and Shields 1991). Assuming the presence of pores for samples with significant hysteresis, we can use the BJH method to calculate model pore diameters (Table 3). The BJH method calculates the presence of meso- and microporosity, and for the samples investigated here, mesoporosity contributes dominantly to the surface area. The presence of meso- and/or macropores in the laboratory-ground plagioclase (excluding albite), diopside, and horn-

1778

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

blende samples is inferred to at least partially explain the relatively high measured SSA in those minerals. One laboratory-ground sample showed high SSA but insignificant hysteresis: Twin Sisters olivine (Fig. 6). It is possible that the Twin Sisters olivine contains mesoporosity, since some mesoporous samples show no hysteresis (Lowell and Shields 1991). However, this olivine contained a significant volume of second phase particulates (Table 1). It may therefore be simpler to assume that the second phase in the Twin Sisters sample is either finer or more platy (e.g., with higher SSA) than the olivine, and that the impurity explains the high SSA observed (Table 2). In fact, every sample with a high content of second phase (Twin Sisters olivine, Grass Valley and Miyake Jima anorthite) also showed a high SSA. Two minerals from this study with significant impurity content also showed hysteresis in sorption behavior consistent with the presence of mesopores (both anorthite samples). Pores with diameters on the order of nanometers can form at triple junctions between some mineral grains in rocks if reactive fluids are present long enough for equilibration (e.g., Lee et al. 1991). For the impurity-containing laboratory-ground minerals that showed hysteresis in this study, SSA may therefore document porosity at triple junctions between the impurity and primary minerals. Such triple junctions are not, however, limited to multimineralic samples: interconnected triple junctions could also contribute measurable porosity in the impurity-free specimens that show hysteresis. Standard isotherms One way to analyze adsorption isotherms to determine the presence of porosity is to plot Vads for one sample against for a nonporous reference sample for the same value of p/p0. Such an plot (Gregg and Sing 1982; see also description in Methods section) should be linear if both samples are nonporous, but should show a deviation for porous samples compared to nonporous standards. The data for the standard phase used here to make the plots, based on work by Sing and coworkers (see original references in Gregg and Sing 1982, Table 2.14), is collated from nitrogen isotherms for phases of SiO2 including quartz with surface areas ranging from 1200 m2/g. For plots for all laboratory-ground silicates investigated (Figs. 3 and 4), a regression line was calculated through the first three data points. In every case, the intercept of this line equaled zero within error or was negative, consistent with absence of microporosity. From the plots, the Vads at = 1.0 (relative pressure = 0.4) increased from albite olivine diopside hornblende anorthite, consistent with increasing mesoporosity in these samples. Regression of the first linear region of any plot will yield a positive intercept if micropores are present or zero if no micropores are present (Webb and Orr 1997). The following observations therefore imply that the contribution of micropores to the total surface in our laboratory-ground samples is small (<25%): (1) generally low overall gas adsorption and small SSA (Figs. 3 and 4, Table 2), (2) little gas adsorption at low pressures (Figs. 3 and 4), (3) the first linear region of the plots extrapolate to zero or a negative value, and 4) the values of c are low (<65, Table 3).

Although plots are not presented for the naturally weathered samples, the hysteresis observed for all but two of these samples is consistent with the presence of porosity (Figs. 5 and 6). As calculated by the BJH model, most of the pore area in these samples is contributed by mesoporosity (Table 3). Consistent with mesoporosity rather than microporosity dominating the pore area, all but one of the c values for the naturally weathered samples is less than 100. The anomalously high value of c for the Turlock Lake plagioclase may suggest the presence of some microporosity in that sample. Comparison to literature data Porosity has been documented microscopically by many workers in alkali feldspars, and, in particular, numerous possible examples of mesopores consisting of subgrain walls and triple junctions have been observed (e.g., Worden et al. 1990; Walker et al. 1995; Lee and Parsons 1995). Some pores with diameters as small as 8 nm have been imaged by Lee et al. (1998, see their Fig. 7). To compare microscopic observations with BET measurements for alkali feldspar, Hodson et al. (1997) compared the SSA values measured by BET with calculated internal SSA based upon image analysis of pores of cross-sectional area from 0.01 to 20 m2. Hodson et al. (1997) concluded that the SSA of samples of the unweathered alkali feldspar samples studied were dominated by external surface topography and that the area of pores was insignificant. However, in that work, they did not consider micropores or mesopores (<50 nm) because no such features were imaged under SEM. Hodson (1998) cited BET adsorption isotherms to argue that 1284% of BET surface area in four of the previously investigated unweathered alkali feldspars was due to microporosity and that most of the rest of the SSA was explained by external surface. He assumed that mesopores did not significantly contribute to SSA in his samples largely on the basis of his previous SEM observations (Hodson et al. 1997). Arguing that no gas-permeable channels with diameter <2 nm exist in the alkali feldspar structure, he concluded that the inferred microporosity was related to cracks emplaced by grinding. The argument for the presence of microporosity in four alkali feldspars investigated by Hodson (1998) was also based upon the observation of a t-plot with an inflection point (c.f. his Fig. 2). He used an empirically fitted t calculation that did not use data from a standard nonporous reference sample (e.g., Harkins and Jura 1943). Hodson (personal communication) notes that recalculation of the t-plots he previously reported using standard reference nonporous data made no difference to his results or conclusions. For our samples, t-plots calculated using a standard equation based upon nonporous materials (Harkins and Jura 1943; see Eqs. 312 of Webb and Orr 1997) yielded negative intercepts, consistent with lack of microporosity. Lack of microporosity is also consistent with low gas adsorption, zero intercepts for alpha plots, and low values of c. Although we cannot rule out the presence of micropores in the porous PSMs investigated here, BET analysis is consistent with a significant contribution of SSA from mesoporosity. Based upon the hysteresis, the presence of interconnected meso-

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

1779

porosity is inferred in every primary silicate sample investigated except olivine and albite: in a pyroxene (diopside), in an amphibole (hornblende), and in three plagioclase feldspars (oligoclase, bytownite, anorthite). Zhang et al. (1993) also used hysteresis in gas sorption behavior to argue that slit-like pores were present in hornblende. His observations of porosity were similar to those reported here for hornblende. Mesoporosity also contributes SSA to natural samples of soils and sediments (e.g., Titley et al. 1987; Mayer 1994, 1999). Consistent with our observations, Mayer (1994) argued that the contribution of micropores (pores with diameters <2 nm) to the total BET surface area of marine sediments and several soils is small (<15%) compared to the contribution from mesopores. Porosity in weathered quartz and feldspar has also been documented by Ball et al. (1990), Wood et al. (1990), and Yau (1999) using gas adsorption, mercury porosimetry, and diffusion analysis in grains from glacial outwash comprising two groundwater aquifers in North America. For these samples, mesoporosity contributed significant SSA. In one of these aquifers, clays inside the pores of some grains were also thought to contribute to surface area (Yau 1999). Genesis of porosity Porosity that contributes to BET SSA may be present in the original hand samples, or be due to grinding. If the porosity was present before grinding, the pores may have formed by solution reactions with hydrothermal, diagenetic, or weathering fluids. Even relatively pristine unweathered samples may have experienced some episode of fluid infiltration. In the weathered soil grains analyzed here, the increase in hysteresis observed in grains of the Turlock Lake soil as compared to the Modesto soil probably attests to growth of porosity due to solution etching during weathering of those samples. Solution etching could also explain the correlation noted by Brantley et al. (1999) that the measured BET SSA of minerals generally increased with increasing dissolution rate. However, for extremely fast-weathering minerals such as olivine, either insignificant interaction with solution has occurred, or pores may have quickly grown into macropores that do not contribute to SSA, explaining the low SSA of this mineral. Coalescence of pores may also explain the decrease in hysteresis (Fig. 6) and surface area (White 1995) in the quartz grains from the China Hat soil as compared to the Turlock Lake. If solution etching occurs at defect sites in the crystal, then some defect-abundant minerals, such as perthitic feldspars, may be more likely to develop pores, as observed previously (e.g., Lee et al. 1998), whereas well-annealed minerals such as San Carlos olivine may not. As mentioned earlier, pores may also form at triple junctions between mineral grains in polycrystalline samples. Very few workers have searched for porosity in most silicate minerals other than potassium feldspar. It is also possible that much of the SSA and porosity measured in laboratory-ground PSMs is due to grinding, as suggested for several potassium feldspars by Hodson (1998). For the one case reported here where SSA decreased with time since grinding (e.g., hornblende, Table 2), some porosity may have been induced and then healed. However, in another case, SSA increased after grinding (e.g., anorthite). The relationship be-

tween duration of time since grinding and SSA and porosity remains to be investigated. Regardless of the cause of the porosity, the inferred presence of mesoporosity in naturally weathered quartz and feldspar and the similarity between isotherms for laboratory-ground and naturally weathered samples (Figs. 35), suggests that mesoporosity is important for both field and laboratory systems. SSA vs. grain size Equation 1 predicts that log (SSA) of non-porous non-Euclidean solids should vary linearly with log grain size (slope = 1). Such a model will not hold, however, if porosity contributes significant surface area. For two minerals where little porosity was inferred from adsorption-desorption (Amelia albite crystal, San Carlos olivine), slopes = 1 within one standard error (Figs. 7 and 8, Table 4). A slope of 1 was also observed for quartz by Parks (1990, plotted on Fig. 6A as a line). The equation estimated for the quartz data (for particle size determined by sieving), is log (SSA, cm2/g) = b + m log (d) (8)

where d = grain diameter (m), intercept b = 4.67 0.06 and slope m = 1. Comparable values are observed for Amelia albite crystal and San Carlos olivine: b = 5.2 0.3 and 5.3 0.4, respectively, and m = 1.0 0.2 and 1.1 0.2, respectively. The equation for quartz or the equation for Amelia albite and San Carlos olivine data regressed together (b = 5.2 0.2, m = 1.0 0.1) may be useful to estimate SSA as a function of grain size for impurity-free, laboratory-ground primary silicates without significant porosity, or may be useful to estimate the external surface area of silicates. Presumably, this equation would not be useful for platy or acicular primary silicates. Equation 3 also suggests that, for samples with porosity, a plot of SSA vs. 1/grain size should fit a line with intercept = internal surface area and slope equal to a term related to external roughness. One such inverse plot is presented here (Fig. 8). Curvature of such plots is universally convex upward, indi-

TABLE 4. Calculated slopes

Mineral phase Slope Quartz* 1 Microcline (Keystone) 0.8 0.2 Microcline (Keystone) 0.8 0.1 Albite glass 1.4 0.1 Albite (Amelia) 1.0 0.1 Albite (Evje) 0.7 0.1 Oligoclase (Madawaska) 0.6 0.2 Labradorite (Labrador) 0.5 0.1 Bytownite (Duluth) 0.5 0.1 Bytownite (Crystal Bay) 0.6 0.2 Anorthite (Miyake Jima) 1.2 0.2 Anorthite (Grass Valley) 0.2 0.05 Olivine (San Carlos) 1.1 0.2 Olivine (Twin Sisters) 0.3 0.2 Diopside (Herschel) 0.6 0.2 Hornblende (Gore Mtn.) 0.6 0.1 * Parks (1990). From Holdren and Speyer (1985). From Hodson (1999). Slope previously reported by Brantley et al. (1999).

1780

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

FIGURE 7. SSA plotted as log (SSA) vs. log (grain diameter) for datasets for the noted minerals where measurements were made on the same mineral in the same laboratory: (a) microcline (b) albite (c) bytownite (d) anorthite. Arrows between data points on some plots show changes in SSA measured after differences in duration since grinding (Table 2).

cating that the model does not fit the data. In particular, SSA of finest fractions are high compared to the prediction. The nonlinearity of inverse plots might be explained if Aint were a function of grain size. For example, if pore spacing averaged l, and if grains were ground until diameter d < l, then Aint would be a function of grain size for d< l because such fine grains might contain no pores. Such a model would predict, however, that SSA of fine grains should be unusually small, contradicting Figure 8. However, if internal surface area were grinding-induced, and if finer grains contain more grindinginduced cracks than coarser grains as suggested for some alkali feldspars by Hodson (1998), then inverse plots such as those shown might be produced. Another explanation for high SSA for fine fractions is that ext increases with decreasing grain size. Hodson (1998) reported changing ext as a function of grain size for four samples

of alkali feldspar; however, the question of roughness as a function of grain size has not been investigated for the minerals reported here. To document such roughness, one needs a probe capable of measuring roughness at the same scale as the BET adsorbate. SEM work will only, under optimal conditions, reveal features down to tens of nanometers and thus cannot yield insight into roughness measured at the scale of the BET measurement (e.g., Hodson 1998). Perhaps the only probe that can measure such atomic roughness is the atomic force microscope (AFM); however, the AFM is only useful in measuring roughness on surfaces that are relatively flat. Extrapolation of kinetics Variations in SSA for grain sizes above 75 m measured in different laboratories and for different compositions range from <2 (microcline), <1.5 (albite), <1.5 (oligoclase), <3 (labra-

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

1781

FIGURE 8. SSA plotted as log (SSA) vs. log (grain diameter) for datasets for the noted minerals where measurements were made on the same mineral in the same laboratory: (a) olivine (b) diopside (c) hornblende. Arrow on data point for diopside sample indicates that the sample may have contained significant fraction of fines. An inverse plot (SSA vs. 1/grain diameter) is also plotted for (d) hornblende.

dorite), <2 (bytownite), <15 (anorthite), <5 (diopside), <7 (hornblende), to <15 (olivine) (Figs. 1 and 2). For most of these cases, the discrepancies in SSA among measurements is larger than the variability observed for our samples of one composition, suggesting that variations in SSA summarized in Figures 1 and 2 are due to factors such as variations in composition, structure, defect density, impurity content, mineral treatment or history, adsorbate identity, and/or measurement protocol. Most of the SSA measurements summarized in the literature (and in Figs. 1 and 2) were used by investigators to normalize dissolution rates for determination of rate constants. Such investigations present rate data that typically vary by factors of at least 210 among laboratories for a given mineral composition (e.g., White and Brantley 1995). Given the large variations in SSA for minerals, it is possible that random and systematic errors in SSA contribute significantly to the

interlaboratory error in rate constants. For example, in the past, many compilations of geochemical rate data for silicate minerals have not distinguished between rates normalized by SSA measured by N2 BET vs. Kr BET (see examples within White and Brantley 1995). Differences in SSA of natural and laboratory-ground silicates have been associated with the large (orders of magnitude) discrepancies between field and laboratory dissolution rates. Anbeek (1992a, 1992b, 1993; Anbeek et al. 1994) argued that most of the new BET surface area created during natural weathering is relatively nonreactive, while laboratoryground surface is reactive (see also, Eggleston et al. 1989). Hodson (1998) argued that much of the BET surface area for the laboratory-ground alkali feldspars which he investigated could be due to grinding-induced microporosity. He implied that, if such induced surface area reacts differently than that of

1782

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

weathered samples (see for example, Lee et al. 1998), then the BET surface area may be an inappropriate parameter to use for extrapolating interface-limited kinetics from laboratory to field. If transport in and out of a pore is slow compared to interface-limited reaction within the pore, then solution chemistry in the pore will differ from bulk chemistry. How small must a pore be to create slow transport in and out of the pore? Derjaguin and Churaev (1986) argued on the basis of a capillary model that solute transport from pores smaller than 1000 nm in diameter must be controlled by diffusion. If this model is applicable here, then all pores identified in this study by gas adsorption would be accessible only by diffusional transport. Solution chemistry and reaction kinetics inside the pore probably differ from chemistry and kinetics outside the pore. In such a system, the BET surface area includes the pore area and is not appropriate for extrapolation of reaction kinetics based upon measurements of dissolution of external surface area. Although other authors have discussed this problem, data reported in this work document that for several silicates for which field and laboratory dissolution rates have been compared, porosity may contribute significantly to BET SSA. If mineral-water reaction within micro- and mesoporosity is transport-limited and if such reaction does not significantly contribute to the flux of dissolved constituents in a dissolving system, then laboratory and field dissolution rates should be compared only after normalization by SSAexternal rather than SSABET. For some minerals, such as the laboratory-ground quartz, albite, and San Carlos olivine samples reported here, external surface area may equal BET SSA. However, for other silicates such as the diopside, hornblende, and other plagioclase compositions analyzed here, the BET SSA differs from the external surface area. For such minerals, external SSA rather than BET SSA might be more useful to compare rates, and it may be possible to estimate the external SSA by using the SSA of Amelia albite, quartz, or San Carlos olivine. Normalization by external SSA will increase the apparent rates of dissolution compared to BET-normalized rates if SSABET > SSAexternal. To quantify these factors, we can assume that external surface area of the primary silicate equals the SSA of albite, quartz, or San Carlos olivine: with this assumption, the ratio of SSABET/SSAexternal for laboratory-ground diopside, hornblende, and plagioclase at a given grain size lies between ~28 (Table 2). The so-called field-lab discrepancy in which field rates are slower than laboratory rates will be alleviated only if the ratio SSABET/SSAexternal is larger for the field than for the laboratory, and if external grain surfaces dissolve more rapidly than internal pore surfaces. As an upper limit estimate of this ratio for naturally weathered samples from the Merced soils, we assume that the external surface area of the soil samples equals SSAalbite as measured for laboratory-ground samples of the same grain size. Using this assumption, the SSABET/SSAexternal for the Turlock Lake samples varied from 10 (quartz) to 49 (potassium feldspar) to 109 (plagioclase). These values are much larger than the ratios (SSABET/SSAexternal) estimated from data in Table 2 for laboratory-ground plagioclase samples (5). If dissolution rates were estimated for Merced plagioclase using BET surface area, and if the internal surface area was insignificantly reactive, then

the field rate would be underestimated by about a factor of 20 with respect to the laboratory rate because field samples have about a factor of 20 more internal surface area than laboratoryground samples. Furthermore, if some of the porosity in PSMs is grinding-induced, or if the internal surface area of laboratory-ground samples is more reactive than naturally weathered surfaces, such a discrepancy could further contribute to the observed laboratory-field problem (e.g., Anbeek 1992a, 1992b, 1993; Anbeek et al. 1994; Lee et al. 1998; Hodson 1998).

CONCLUSIONS
If internal and external surface is not equally reactive in primary silicates, then porosity in PSM samples as documented here implies that BET surface area measurements may not be appropriate for extrapolation of geochemical kinetics. Reactive surface area may be dominated by external surface area because reaction in pores may be rate-limited by transport rather than interfacial reaction. Research should focus on attempting to estimate the reactive surface area and porosity in field systems, and learning how to accurately estimate these parameters in the laboratory for use in extrapolation to the field.

ACKNOWLEDGMENTS
Don Voigt, Drew Stolar, Michelle Smith, Peter Heaney, Carlo Pantano, Rich Holdren, Art White, Jorie Schulz, and Mark Hodson are acknowledged for experimental work, access to instrumentation, samples, and/or advice. Funding from the Dept of Energy Office of Basic Energy Sciences grant DE-FG0295ER14547.A000 is also acknowledged. Ian Parsons and Mark Hodson provided extremely valuable reviews.

REFERENCES CITED
Anbeek, C. (1992a) Surface roughness for minerals and implications for dissolution studies. Geochimica et Cosmochimica Acta, 56, 14611469. (1992b) The dependence of dissolution rates on grain size for some fresh and weathered feldspars. Geochimica et Cosmochimica Acta, 56, 39573970. (1993) The effect of natural weathering on dissolution rates. Geochimica et Cosmochimica Acta, 57, 49634975. Anbeek, C., van Breemen, N., Meijer, E.L., and van der Plas, L. (1994) The dissolution of naturally weathered feldspar and quartz. Geochimica et Cosmochimica Acta, 58, 46014614. Atkinson, B.K. and Meredith, P.G. (1987) Experimental fracture mechanics data for rocks and minerals. In B.K., Atkinson, Ed., Fracture Mechanics of Rock, 111 166 Academic Press, Orlando. Ball, W.P., Buehler, C.H., Harmon, T.C., MacKay, D.M., and Roberts, P.V. (1990) Characterization of a sandy aquifer material at the grain scale. Journal of Contaminant Hydrology, 5, 253295. Barret, E.P., Joyner, L.G., and Halenda, P.P. (1951) The determination of pore volume and area distributions in porous substances. I. Computations from nitrogen isotherms, Journal of American Chemical Society, 73, 373380. Brantley, S.L., White, A.F., and Hodson, M.E. (1999) Surface area of primary silicate minerals. Growth and Dissolution, In B. Jamtveit and P. Meakin, Eds., Geosystems, p. 291326. Kluwer Academic Publishers, Dordrecht. Brunauer, S., Emmett, P.H., and Teller, E. (1938) Adsorption of gases in multimolecular layers. Journal of the American Chemical Society, 60, 309319. (1969) Surface Areas of Porous Solids. In Proceedings of the International Symposium on Surface Area Determination, p. 6397. Buttersworth, London. Casey, W.H., Westrich, H.R., and Holdren, G.R. (1991) Dissolution rates of plagioclase at pH = 2 and 3. American Mineralogist 76, 211217. Casey, W.H., Westrich, H.R., Banfield, J.F., Ferruzzi, G., and Arnold, G.W. (1993) Leaching and reconstruction at the surface of dissolving chain-silicate minerals, Nature, 366, 253256. Chen, Y. and Brantley, S.L. (1997) Temperature- and pH-dependence of albite dissolution rate at acid pH, Chemical Geology, 135, 275292. (1998) Diopside and anthophyllite dissolution at 25 C and 90 C and acid pH, Chemical Geology, 147, 233248. (2000) Dissolution of forsteritic olivine at 65 C and 2 < pH < 5, Chemical Geology, 165, 267281. Chou, L. and Wollast, R. (1984) Study of the weathering of albite at room temperature and pressure with a fluidized bed reactor: Geochimica et Cosmochimica

BRANTLEY AND MELLOTT: SURFACE AREA AND POROSITY OF SILICATES

Acta, 48, 22052217. (1985) Steady state kinetics and dissolution mechanism of albite: American Journal of Science, 285, 963993. Derjaguin, B.V. and Churaev, N.V. (1986) Properties of water layers adjacent to interfaces. In C.A. Croxton, Ed., Fluid Interfacial Phenomena, p. 748. Wiley, New York. Eggleston, C.M., Hochella, M.F., and Parks, G.A. (1989) Sample preparation and aging effects on the dissolution rate and surface composition of diopside. Geochimica et Cosmochimica Acta, 53, 797804. Farrell, J. and Reinhard, M. (1994) Desorption of halogenated organics from model solids, sediments, and soil under unsaturated conditions. 2. Kinetics. Environ. Science and Technology, 28, 6372. Gregg, S.J. and Sing, K.S.W. (1982) Adsorption, Surface Area and Porosity, p. 303 Academic Press, London. Hamilton, J.P., Pantano, C.G., and Brantley, S.L. (2000) A comparison of the dissolution and leaching behavior of albite crystal and glass in acidic, neutral, and basic conditions. Geochimica et Cosmochimica Acta, 64, 26032616. Harkins, W.D. and Jura, G. (1943) An adsorption method for the determination of the area of a solid without the assumption of a molecular area, and the area occupied by nitrogen molecules on the surfaces of solids. Journal of Chemistry and Physics 11, 431432. Helgeson, H.C., Murphy, W.M., and Aagard, P. (1984) Thermodynamic and kinetic constraints on reaction rates among minerals and aqueous solutions II. Rate constants, effective surface area, and the hydrolysis of feldspar. Geochimica et Cosmochimica Acta, 48, 24052432. Hochella, M.F. and Banfield, J. (1995) Chemical weathering of silicates in nature: A microscopic perspective with theoretical considerations. In A.F. White and S.L. Brantley, Eds., Chemical weathering rates of silicate minerals, 31, 353406. Mineralogical Society of America Reviews in Mineralogy, Washington, D.C. Hodson, M.E. (1998) Micropore surface area variation with grain size in unweathered alkali feldspars; implications for surface roughness and dissolution studies, Geochimica et Cosmochimica Acta, 62, 34293435. Hodson, M.E., Lee, M.R., and Parsons, I. (1997) Origins of the surface roughness of unweathered alkali feldspar grains. Geochimica et Cosmochimica Acta, 61, 38853896. Holdren G.R. Jr., and Speyer, P.M. (1985) Reaction rate-surface area relationships during the early stages of weathering: I. Initial observations. Geochimica et Cosmochimica Acta, 49, 675681. (1987) Reaction-rate-surface area relationships during the early stages of weathering: II. Data on eight additional feldspars. Geochimica et Cosmochimica Acta, 51, 23112318. IUPAC (1972) Manual of symbols and terminology for physicochemical quantities and units, Appendix 2, Definitions, Terminology, and Symbols in Colloid and Surface Chemistry. Part 1. Pure and Applied Chemistry, 31, 578638. Knauss, K.G., and Wolery, T.J. (1986) Dependence of albite kinetics on pH and time at 25 C and 70 C. Geochimica et Cosmochimica Acta, 50, 24812497. Knauss, K.G., Nguyen, S.N., Weed, H.C. (1993) Diopside dissolution kinetics as a function of pH, CO2, temperature, and time. Geochimica et Cosmochimica Acta, 57, 285294. Lee, M.R. and Parsons, I. (1995) Microtextural controls of weathering of perthitic alkali feldspars. Geochimica et Cosmochimica Acta 59, 44654488. Lee, V.W., Mackwell, S.J., and Brantley, S.L. (1991) The effect of fluid chemistry on wetting textures in novaculite. Journal Geophysical Research, 96, 10,023 10,037. Lee, M.R., Hodson, M.E., and Parsons, I. (1998) The role of intragranular microtextures and microstructures in chemical and mechanical weathering: direct comparisons of experimentally and naturally weathered alkali feldspars. Geochimica et Cosmochimica Acta, 62, 27712788. Lowell, S. and Shields, J.E. (1991) Powder surface area and porosity (3rd edition), p. 245. Chapman and Hall, U.K. Luthy, R.G., Aiken, G.R., Brusseau, M.L., Cunningham, S.D., Gschwend, P.M., Pignatello, J.J., Reinhard, M., Traina, S.J., Weber, W.J., and Westall, J.C. (1997) Sequestration of hydrophobic contaminatns by geosorbents, Environmental Science and Technology, 31, 33413347. Mayer, L. (1994) Relationships between mineral surfaces and organic carbon concentrations in soils and sediments. Chemical Geology, 114, 347363. Mayer, L.M. (1999) Extent of coverage of mineral surfaces by organic matter in marine sediments, Geochimica et Cosmochimica Acta, 63, 2, 207216. Micromeritics Instrument Corporation (1995) Micromeritics ASAP 2010 acceler-

1783

ated surface area and porosimetry system: Operators manual, Norcross, Georgia. Montgomery, C.W. and Brace, W.F. (1975) Micropores in plagioclase. Contributions to Mineralogy and Petrology, 52 1728. Nickel, E. (1973) Experimental dissolution of light and heavy minerals in comparison with weathering and instrastratal solution, Contributions to Sedimentology, 1, Stability of heavy minerals, 368. Parks, G.A. (1990) Surface energy and adsorption at mineral-water interfaces: An introduction. In Mineralogical Society of America Reviews in Mineralogy. Vol. 23, 133175. Rose, N.M. (1991) Dissolution rates of prehnite, epidote, and albite, Geochimica Cosmochimica Acta 55, 32733286. Schulz, M.S., White, A.F., and Art, F. (1999) Chemical weathering in a tropical watershed, Luquillo Mountains, Puerto Rico; III, Quartz dissolution rates, Geochimica Cosmochimica Acta, 63, 337350. Sing, K.S.W. (1969) Utilisation of adsorption data in the BET region. In Proceedings of the International Symposium on Surface Area Determination, p. 2543. Buttersworths, London. Stillings, L.L. and Brantley, S.L. (1995) Feldspar dissolution at 25 C and pH 3: reaction stoichiometry and the effect of cations. Geochimica et Cosmochimica Acta, 59, 14831496. Titley, J.G., Glegg, G.A., Glasson, D.R., and Millward, G.E. (1987) Surface areas and porosities of particulate matter in turbid estuaries. Continental Shelf Research, 7, 13631366. Walker, F.D.L., Lee, M.R., and Parsons, I. (1995) Micropores and micropermeable texture in alkali feldspars: geochemical and geophysical implications. Mineralogical Magazine, 59, 505534. Webb, P.A. and Orr, C. (1997) Analytical methods in Fine Particle Technology. Micromeritics Instrument Corp. Atlanta, Georgia. Werth, C.J. and Reinhard, M. (1997) Effects of temperature on trichlorethylene desorption from silica gel and natural sediments. 1. Isotherms. Environmental Science and Technology, 31, 689696. White, A.F. (1995a) Chemical weathering rates of silcate minerals in soils. In A.F. White and S.L. Brantley, Eds., Chemical weathering rates of silicate minerals, 31, 407461. Mineralogical Society of America Reviews in Mineralogy, Washington, D.C. White, A.F. and Brantley, S.L., Eds. (1995) Chemical weathering rates of silicate minerals, vol. 31. Mineralogical Society of America Reviews in Mineralogy, Washington, D.C. White, A.F. and Peterson, M.L. (1990) Role of reactive-surface-area characterization in geochemical kinetic models. In Chemical Modeling of Aqueous Systems II, American Chemical Society Symposium Series 416, 461475. White, A.F. and Yee, A (1985) Aqueous oxidation-reduction kinetics associated with coupled electron-cation transfer from iron-containing silicates at 25 C. Geochimica Cosmochimica Acta, 49, 12631275. White, A.F., Blum, A.E., Schulz, M.S., Bullen, T.D., Harden, J.W., and Peterson, M.L. (1996) Chemical Weathering of a Soil Chronosequnece on Granitic Alluvium 1. Reaction Rates Based on Changes in Soil Mineralogy. Geochimica et Cosmochimica Acta, 60, 25332550. Wogelius, R.A. and Walther, J.V. (1991) Olivine dissolution at 25 C: Effects of pH, CO2, and organic acids. Geochimica et Cosmochimica Acta 55, 943954. Wood, W.W., Kraemer, T.F., and Hearn, P.P. Jr. (1990) Intragranular diffusion: An important mechanism influencing solute transport in elastic aquifers? Science, 247, 15691572. Worden, R.H., Walker, F.D.L., Parsons, I., and Brown, W.L. (1990) Development of microporosity, diffusion channels and deuteric coarsening in perthitic alkali feldspars. Contributions to Mineralogy and Petrology, 104, 5, 507515. Yau, S. (1999) Dissolution Kinetics of Feldspar in the Cape Cod Aquifer, Massachusetts: Calculation of Ground Water Residence Time. Masters Thesis, The Pennsylvania State University. University Park, Pennsylvania. Zhang, H., Bloom, P.R., and Nater, E.A. (1993) Change in surface area and dissolution rates during hornblende dissolution at pH 4.0. Geochimica et Cosmochimica Acta 57, 16811689.

MANUSCRIPT RECEIVED MAY 10, 1999 MANUSCRIPT ACCEPTED JULY 7, 2000 PAPER HANDLED BY SUSAN L.S. STIPP