University of Southern California Dissertations and Theses

Exploring temporal changes in surface species on weathered feldspar mineral surfaces using solid-state NMR spectroscopy

(USC Thesis Other)

Exploring temporal changes in surface species on weathered feldspar mineral surfaces using solid-state NMR spectroscopy

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
EXPLORING TEMPORAL CHANGES IN SURFACE SPECIES ON WEATHERED FELDSPAR MINERAL
SURFACES USING SOLID-STATE NMR SPECTROSCOPY

by

Eric Kleinsasser
______________________________________________________________________________
A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOLOGICAL SCIENCES)

December 2014

ii
Acknowledgements
This research was funded by the University of Southern California Dornsife College of Letters,
Arts and Sciences, and by the Department of Energy’s William R. Wiley Environmental
Molecular Sciences Laboratory (EMSL). Much of the analytical work would have been
impossible without collaboration with and guidance from Nancy M. Washton, Karl T. Mueller,
and Jamie Weaver at EMSL. Special thanks to Mark Torres for considerable assistance in one of
the field areas of this project, and for vital guidance with mineral dissolution protocols and our
research group’s MP-OES. Thanks also goes to Ansgar Siemer, who generously offered the use
of his solid-state NMR spectrometer at USC’s Zilkha Neurogenetic Institute. Additional field
assistance was given by John Guzman, who helped collect samples used to calibrate our lab
group’s ASAP 2010 BET physisorption instrument. Particular thanks go to my thesis
committee, Dr. Doug Hammond and Dr. Doug LaRowe. And most certainly many thanks and
recognition must go to Josh West, my advisor, for insights on the primary aspects of this project,
field assistance, laboratory assistance, and general guidance on the unique challenge of doing
graduate research.

iii
Table of Contents
Abstract 1
1_ Introduction 2
1.1_ Motivation and Background 2
1.2_ Theoretical Basis for Proposed Silicate Dissolution Mechanisms 6
1.3_ Experimental Exploration of Silicate Dissolution Mechansims 7
1.4_ Extent of weathering and “Reactive Surface Area” 10
1.5_ Probing the Silicate Surface 13
1.6_ NMR Spectroscopy 14
1.7_ Study Areas 16
2_ Methods 21
2.1_ Sample Collection 21
2.2_ Sample Preparation 22
2.3_ Nuclear Magnetic Resonance Spectroscopy 23
2.4_ Microwave-Plasma Optical Emission Spectroscopy 24
2.5_ Brunauer Emmett Teller (BET) and geometric SSA measurements 24
2.6_ X-Ray Diffraction 25
2.7_ Staining 25
3_ Results 26
3.1_
19
F NMR Experiments 26
3.2_ Temporal Variations Determined by MP-OES Experiments 30
3.3_ X-Ray Diffraction 30
3.4_
27
Al NMR Experiments 33
4_ Discussion 34
4.1_ Implications of the Chronosequence Results 34
4.2_ Elemental Analysis and Staining 42
4.3_
27
Al NMR and Future Directions 43
5_ Closing Remarks 44
References 45
Appendix A: NMR Acquisition Parameters A1
Appendix B: Solid-state NMR spectroscopy and Magic Angle Spinning B1
Appendix C: Determination of BET sample input precision threshold C1
iv
List of Tables
Table 1. Samples from both field areas, site coordinates, soil exposure ages,
specific surface areas, OH counts, and elemental data. 31
Table 2. Mole fractions of cations relative to one another in samples. 36
Table C1. Input mass, measured SSA, and calculated SA for the two different
reference materials. C3
v
List of Figures
Figure 1. Dissolution rates of silicate minerals measured by cation depletion
through a soil profile, or effluent concentrations out of a column reactor
experiment, as a function of soil age or experiment duration. 3
Figure 2. Rate constants for silicate mineral dissolution, as a function of either
fluid residence time (left) or reaction age (right). 5
Figure 3. Mineral surface charge (a) and dissolution rate (b) as a function of pH,
for potentiometric titrations of albite done by Blum and Lasaga (1991). 9
Figure 4. Representation of the silicate dissolution mechanism proposed by
Oelkers (2001), as it would theoretically apply to anorthite (CaAl
2
Si
2
O
8
). 12
Figure 5. Proposed mechanism for the silanization of a silicate mineral surface. 15
Figure 6. Aerial map of Bishop Creek and vicinity, with shaded regions showing
moraines and their approximate ages. 18
Figure 7. Maps of the two field areas for this study, Bishop Creek in the Sierra
Nevada and the Front Range in the Colorado Rocky Mountains. 20
Figure 8. Examples of processed spectra from (a)
19
F and (b)
27
Al MAS NMR
experiments. 27
Figure 9. Q
3
OH site concentrations versus age of deposit for samples from the
Bishop Creek chronosequence. 28
Figure 10. Net Q
3
OH site accumulation per m
2
mineral surface area, as a
function of sample exposure age, for samples from Bishop Creek. 32
Figure 11. XRD Pattern of one sample from Bishop Creek. 35
Figure 12. Three-dimensional ternary diagrams for samples from the Bishop
Creek chronosequence. 37
Figure 13. Semi-log plot showing proportions for different samples of one split in
the
27
Al peak associated with tetrahedral Al, relative to the whole peak. 38
Figure 14. Images of feldspar grains from a Bishop Creek chronosequence site,
stained with sodium cobaltinitrite and amaranth. 41
Figure B1. Graphic representation of anisotropic effects in a crystalline sample. B3
Figure B2. Graphic representation of spinning sidebands. B4
vi
List of Figures (continued)
Figure C1. BET specific surface area (SSA) of samples vs. input sample mass,
for the Micromeritics silica alumina reference material (upper plot—blue
squares), and anorthosite from the San Gabriel Mountains (lower plot—purple
squares). C2
Figure C2. BET specific surface area (SSA) of aluminosilicate samples vs. actual
total surface area (SA) of each sample. C4
Figure C3. Slopes of least squares linear regressions run on subsets of the 55
aluminosilicate reference samples. C5
Figure C4. Measured total surface area plotted as a function of sample mass, for
two different types of reference material, silica alumina (left) and anorthosite
(right). C6
1

Abstract
A method developed to quantify hydroxyl (OH) sites on silicate minerals is applied for the first
time to feldspar grains from soils in two glacial chronosequences and a saprolite profile.
Illuminating the interplay between accumulation of surface OH sites and the total time primary
silicate minerals have spent in the weathering environment could be a step towards
mechanistically understanding the reasons mineral dissolution rates are observed to decrease
with time. Soil was sampled from discrete deposits—with ages ranging from 10 ka to 200 ka—
along glacial chronosequences in the Sierra Nevada and the Colorado Front Range. Feldspar
grains were picked from these soils and treated with (3,3,3trifluoropropyl) dimethylchlorosilane
(TFS). This molecule selectively binds to Q
3
OH sites on silica and alumina tetrahedra—sites
wherein one tetrahedral anion forms a hydroxyl at the surface during dissolution and the other
three remain coordinated with the crystal. NMR spectroscopy is a sensitive, non-destructive
analytical method, and here it is used to quantify
19
F in TFS on TFS-treated feldspar samples. Q
3

OH concentrations were inferred from these data, and normalized to specific surface area (the
physical surface area per unit mass). Treated samples were also dissolved and run through an
MP-OES for elemental analysis, to determine whether or not time-dependent mineralogical
evolution was partly responsible for the changing hydroxyl concentrations. NMR experimental
results reveal a power-law decrease in Q
3
OH site accumulation rate with time. Elemental
analysis by MP-OES suggests that mineralogical differences among samples may be a notable
complicating factor in Front Range samples, but far less so in the Bishop Creek samples. This
study confirms that the NMR procedure is sensitive enough to quantify very low concentrations
of TFS condensed to OH sites, as found on feldspar surfaces. Continued investigation using the
2

TFS treatment and NMR method will hopefully shed light on the molecular-scale processes that
drive interface-limited systems.

1_ Introduction
1.1_ Motivation and Background
Silicate dissolution has been a subject of major interest in environmental geochemistry
for much of the past half century. Chemical weathering of silicate minerals has critical influence
on global-scale biogeochemical cycles (Berner et al. 1983, White and Brantley 1995).
Aluminosilicates buffer acidic soils (Ruess 1983, Van Breemen et al. 1983, Chadwick and
Chorover 2001, Mol et al. 2003), affect nutrient fluxes and advection of material through the
Critical Zone (Marschner 1995, Porder et al. 2007, Jin et al. 2010), and regulate long term
climate change (Berner 1983, Volk 1987, Goudie and Viles 2012). Numerous investigations,
both based on laboratory experimentation and field study, have explored silicate weathering to
build a better understanding of the main constraints that affect dissolution rate (see especially
reviews by Brantley 2003, White and Brantley 1996, and White and Brantley 2003).
Both field study and laboratory experimentation have shown that weathering rates of soil-
contained silicates appear to decrease with time, according to a power-law relationship (figure 1)
(White and Brantley 2003). It is noteworthy, however, that rates of dissolution in nature and
experimentally determined rates are often disparate by orders of magnitude. In addition, natural
weathering rates span a significant range contingent on environment, lithology, and other
parameters.
Recently, Reeves and Rothman (2013) have partitioned weathering scenarios by their
dominant governing processes, and used various models to describe different weathering
3

Figure 1. Dissolution rates of silicate minerals measured by cation depletion through a soil
profile, or effluent concentrations out of a column reactor experiment, as a function of soil age or
experiment duration. Time is profile age or experiment duration. Adapted from White and
Brantley (2003).

4

mechanisms. This was partly done by deconstructing the data presented in White and Brantley
(2003) (see more discussion in “Theoretical Basis” below). They have established that, in
different cases, extrinsic (solution-controlled) or intrinsic (surface/solution interface-controlled)
mechanisms may have varying degrees of influence on weathering rate. This was demonstrated
by defining time either as fluid residence time, or total duration of mineral dissolution—a value
the authors call the “age of reaction” (Reeves and Rothman 2013) (figure 2). Reeves and
Rothman approximate the overall rate of dissolution via what they call an “apparent rate constant
k
ap
” which they describe as accounting for “catalytic and inhibitory species, the specific surface
area of the sample, surface geometry, and the effects of averaging over heterogeneous samples.”
The general form of the rate equation from which values of k are derived (figure 2) is
R = k∙(1 - Ω) (1)
where Ω is the relative saturation (unit-less), i.e. the ratio of bulk concentration of weathering
material to equilibrium concentration of weathering material (Reeves and Rothman 2013, Maher
et al. 2010).
As is apparent from figure 2, the age of reaction, i.e. total duration a soil is subject to
intrinsic weathering processes, is important in affecting the overall rate of weathering through
time. A number of datasets (Clow and Drever 1996, Kim 2002, Maher et al. 2004, Fantle and
DePaolo 2007, Jin et al. 2008, White et al. 2008, Maher 2010) used in the compilation by Reeves
and Rothman (2013) have noted this temporal rate change, and the question remains: why is this
change observed in myriad cases? Looking for more ways to understand the evolution mineral
surfaces at a molecular level during dissolution promises potential next steps towards answering
this question, at least in the cases where intrinsic processes are in control.

5

Figure 2. Rate constants for silicate mineral dissolution, as a function of either fluid
residence time (left) or reaction age (right). The lines plotted are for rate constants k = 4*10
-5

(yr
-1
) (left), and k = 0.1 (yr
-1
) (right), where k is in units of inverse time (derived from methods of
Maher et al. 2004). White-colored points do not follow the general power-law relationship with
the time axis defined as fluid residence time (left panel), but seem to follow a power law when
time is defined as the duration of reaction (left panel). Adapted from Reeves and Rothman
(2013).

6

1.2_ Theoretical Basis for Proposed Silicate Dissolution Mechanisms
Mechanistic explanations for how silicate minerals dissolve completely have long utilized
Transition State Theory (TST), which attempts to describe the kinetic energy barrier created by
activated complexes as the surface-solution interface approaches equilibrium (Eyring 1935,
Glasstone et al. 1941, Lin et al. 1975, Boudart 1976, Aagaard and Helgeson 1982, Lasaga 1998).
Aagaard and Helgeson (1982) were the first authors to rigorously express silicate hydrolysis in
terms of complexation at the mineral surface, such that far-from-equilibrium and near-
equilibrium conditions were differentiated. They proposed:
(2)
where ξ denotes the reaction progress, k is the rate constant (in units of inverse time), s is the
effective surface area (i.e. the ratio of reactive-to-total surface area), a
i
is the activity of species i,
n is the stoichiometric coefficient of species i undergoing reaction j with respect to the formation
of a complex, A is the chemical affinity for the overall reaction (-∂G/∂ξ, where G is the Gibbs
energy), and σ (a unit-less constant) is the rate at which the activated complex decomposes
relative to the overall reaction rate (Aagaard and Helgeson 1982).
This framework has been the basis for theoretically describing silicate dissolution in
numerous subsequent studies, notably the above-mentioned White and Brantley (1995), Brantley
(2003), White and Brantley (2003), and Reeves and Rothman (2013), as well as others (see
detailed exploration in Lasaga 1998).
As an example, White and Brantley (2003) modified the equation above to delinate near-
equilibrium and far-from equilibrium behavior using TST. They divide weathering rate into two
separate terms, both containing a Gibbs Energy component with the form (1 - e
ΔG/RT
), for Gibbs
Energy ΔG and temperature T, where R is the universal gas constant. These terms represent two
7

parallel reactions—rapid dissolution far from equilibrium, and gradual dissolution that dominates
as the system approaches equilibrium (White and Brantley 2003). Depending on the rate of fluid
transport through the system, the dissolution rate will be either limited by the interface (far-from-
equilibrium term dominates), or limited by transport (near-equilibrium term dominates).
1.3_ Experimental Exploration of Silicate Dissolution Mechanisms
Theoretical work on silicate dissolution has been complemented by a variety of
laboratory and modelling techniques. Experimentation in the lab often employs a reactor setup
designed to simulate weathering over faster-than-geologic timescales.
There are a few different common reactor types for this application. Chou and Wollast
(1984 and 1985) utilized the first fluidized bed reactor system for experiments involving the
dissolution of albite. The fluidized bed reactor overcame a problem in earlier experimental
setups: fluctuations in solute concentrations could not be controlled during an experiment. Bed
and continuous flow-through reactors allow the system to reach steady-state (Chou and Wollast
1985). Using bed reactors, these authors were able to infer near-surface cation-H
+
exchange that
leads to the development of a leached layer 10-20 angstroms thick.
Subsequent experimental designs have attempted to better simulate natural weathering.
Two classes of such systems are commonly known as batch and column-type reactors. Van
Grinsven and van Riemsdijk (1991) provided a thorough review of these, comparing several
different designs with respect to their effectiveness in measuring weathering rates. In the batch-
type systems, solution chemistry can be allowed to develop naturally, or controlled via titration.
The column design, also known as plug-flow (Brantley 2003) is a setup through which fluid
passes vertically, intended to simulate a natural soil profile. Fluid flow can be made constant or
varied through time (van Grinsven and van Riemsdijk 1991, Brantley 2003).
8

On a smaller scale, much insight has come from work using potentiometric titration
techniques on 1-10g aliquots of silicate material. Early on, Davis et al. (1977) modelled
complexation parameters of oxide-solution interfaces using data from potentiometric titrations.
Their work gave rise to a series of more specialized investigations on silicate minerals (Helgeson
et al. 1984, Chou and Wollast 1984, Chou and Wollast 1985, Blum and Lasaga 1988).
Building on this framework, Blum and Lasaga (1991) explored speciation on albite
(NaAlSi
3
O
8
) by titrating small (1-2g) mineral suspensions. Monitoring solute concentrations and
calculating surface charge below and above neutral pH (figure 3a), they inferred protonation of
the surface in the acidic region and deprotonation in the basic region. When inverse surface
charge was plotted versus inverse H+ or OH- concentration, the data have reasonably linear
trends, indicating that they can be modelled with Langmuir adsorption isotherms. Such pH-
dependent protonation/deprotonation of the mineral surface supports theoretical mechanisms of
silicate dissolution involving the breakdown of the surface solution interface by the formation of
protonated complexes that are eventually released into solution. Using rate data from Chou and
Wollast (1985), they also inferred dissolution rates for their experiments, and plotted these rates
versus pH (figure 3b). The authors observed a very strong pH-dependence for dissolution rates,
with a steady decrease below, and steady increase above, neutral. However, the adsorption
isotherm and dissolution rate plot do not show the same pattern at pH < 4 (figure 3), as surface
charge appears to reach saturation but dissolution rate does not. Blum and Lasaga also noted that
the surface adsorption reactions were extremely fast, and “do not in themselves present a kinetic
barrier to albite dissolution” (Blum and Lasaga, 1991). Their experiments inferentially offer
9

Figure 3. Mineral surface charge (a) and dissolution rate (b) as a function of pH, for
potentiometric titrations of albite done by Blum and Lasaga (1991). In (a), the units are moles of
M—OH
2
+
sites per m
2
in the acidic region, and moles of M—O
-
sites per m
2
in the basic region
(where M denotes a tetrahedrally coordinated Si or Al). Lines are fit to the steep parts of the
trends. Adapted from Blum and Lasaga (1991).
a

b

10

broad evidence of theoretical elementary reactions. This is an important step forward, since
confirmation of elementary reactions in dissolution would permit the application of Transition
State Theory with much greater confidence than if only non-elementary reactions were known.
However, Blum and Lasaga concluded by noting that because adsorption takes place at cation-
exchange sites, terminal isolated and geminal oxygens, and bridging oxygens, a limited amount
can be inferred from adsorption isotherm modelling alone.
1.4_ Extent of weathering and “Reactive Surface Area”
Surface sites that contribute to overall silicate dissolution comprise what has been called
the “reactive surface area” (Lasaga 1995, Luttge et al. 1999, Metz et al. 2005), inspiring work to
identify specific reactive sites at the surface (Teng et al. 2001, Fenter et al. 2014). However, no
consistent definition for reactive surface area has been agreed upon.
It is clear that the reactive portions of a mineral’s surface are a subset of the overall
physical surface area (Brantley 2003). The term s used by Aagaard and Helgeson (1982) (see
discussion above) was defined by those authors as “surface area exposed to solution,” but this
should not imply that, in reality, all physical surfaces will react at comparable rates. Terms like s
are necessarily vague because the domain of “surface area” is difficult to assess and remains a
source of inconsistency in the literature (Lee and Parsons 1995, Brantley 2003) (see discussion).
A rigorous surface area term must be predicated on the changing chemistry of the
surface-solution interface. Many studies have pointed out the necessity of understanding what
regions of the mineral surface are actually reactive, and the role of etch pits and other surface
features in facilitating dissolution has been clearly identified (Parsons 1994, Lasaga 1995,
Hochella and Banfield 1995, Lee and Parsons 1995, Parsons et al. 2005).
11

Potentiometric titrations (such as those done by Blum and Lasaga 1991) have been used
to make inferences about how dissolving silicates behave in nature, i.e. what will happen as they
weather over geologic time. However, these titrations are not capable of determining the specific
surface species that are reacting, where reactive sites develop, or the exact number of elementary
reactions involved in dissolution. Blum and Stillings (1995) also point out that, although
titrations offer significant insight into dissolution processes, these analyses are subject to non-
negligible experimental error.
Physical surface area measurements using BET adsorption methodology (Brunauer et al.
1938) are a common way to assess how reactive silicate material is. However, using BET surface
area measurements to represent reactive surface area assumes unrealistic uniformity, i.e. that all
physical surface sites react equally, which is unrealistic for the above-mentioned reasons.
None of the approaches discussed as yet make distinctions among the different sites that
form hydroxyl groups at the surface (Washton et al. 2008). Such sites can be generally classified
by the coordination number of the silica or alumina tetrahedron being protonated, i.e. Q
4
, Q
3
, Q
2
,
or Q
1
, where Q
n
denotes n oxygens linked to the bulk crystal (Brantley 2003, see Q
3
site in figure
4, right panel). Protonation is presumed to initiate at Q
3
sites, and it is thought that the
coordination number is successively reduced until the Al or Si—or possibly in some cases a
silica polymer—is released into solution (Brantley 2003).
This concept of Q
n
site reduction fits into an overall mechanism of silicate dissolution
proposed and outlined by Oelkers (2001). In this model, dissolution involves ion-H
+
exchange at
the surface-solution interface, followed by protonation of silica and alumina tetrahedra that
terminate at the interface, and eventually stoichiometric dissolution of the mineral (figure 4).
The relevant question is then how much of a mineral surface actually undergoes protonation.
12

Figure 4. Representation of the silicate dissolution mechanism proposed by Oelkers (2001), as it
would theoretically apply to anorthite (CaAl
2
Si
2
O
8
). The left panel depicts near-surface cation
exchange. In the right panel, tetrahedral sites are protonated. O
term
denotes a terminal tetrahedral
oxygen at the surface-solution interface. A Q
3
site is labelled. Lattice schematic adapted from
Weber (2006).

13

Rates of surface protonation cannot be verified without a non-destructive quantitative analytical
procedure that measures concentrations of surface hydroxyl sites.
In this study, Q
3
hydroxyl site concentrations are measured on naturally weathered
feldspar surfaces, as a function of exposure age of their parent soils, using an instrumental
method applied to feldspars for the first time. It is important to stress that reacted surface areas
reported here may be a useful analog to reactivity, but the reactive surface area of these samples
at present is not what is being measured. The objective is to assess the interplay between
hydroxyl site development and progression of weathering, thus developing a better
understanding of what drives changes in weathering rates through time.
1.5_ Probing the Silicate Surface
Relative to secondary minerals, primary silicates like feldspars have very low surface
areas and porosities (Sanders et al. 2010). The surface-selective method chosen must therefore
be sufficiently sensitive to measure low areas. A recently-developed method of chemical
treatment and Nuclear Magnetic Resonance (NMR) analysis is applied to feldspar grains in this
study. The method was previously developed and used on highly protonated silicate minerals
(i.e. clays) and glasses (Washton et al. 2008), but this study is its first application to studying
primary silicate minerals.
The operative experimental step is treatment of mineral grains with (3,3,3trifluoropropyl)
dimethylchlorosilane (TFS). In a mixture with silicate minerals, TFS serves as a probe molecule
which binds selectively to Q
3
sites at a mineral surface (Q
3
sites are represented in figure 3b).
This binding of an organosilane to the silicate surface (silanization) takes place via a two-step
reaction, involving hydrolysis of TFS and subsequent condensation of the hydrolyzed molecule
to a Q
3
site (Washton et al. 2008) (figure 5).
14

Silanization reactions such as the one using TFS have been shown to take place
selectively at Q
3
sites. A review of the terminology and experimental history on silanization of
silica materials can be found in Van Der Voort and Vansant (1996). Working mainly with
trichlorosilane (TCS) on silica, they identify five reaction mechanisms for TCS condensation.
During condensation, a “complete disappearance of the free [i.e. isolated Q
3
] hydroxyl groups”
takes place (Van Der Voort and Vansant 1996). Their review follows a fair amount of analytical
work done on silica surfaces using infrared (IR) spectroscopy (Chuiko et al. 1966, fluorescence
spectroscopy (Lochmuller and Kersey 1988),
1
H and
29
Si NMR spectroscopy (Sindorf and
Maciel 1983, Bronnimann et al. 1988, Morrow and Gay 1988, Legrand et al. 1990, Leonardelli et
al. 1992), and computer modeling (Ferrari et al. 1993), that have collectively assessed the
adsorption kinetics of silanes on silica. Van Der Voort and Vansant conclude that these silanes
react completely with isolated (Q
3
) sites.
1.6_ NMR Spectroscopy
In the approach used here, mineral samples treated with TFS are run through one-pulse
19
F MAS NMR experiments. NMR subjects the sample to a strong magnetic field to induce
nuclear precession. Sensitivity is maximized in this setup, because
19
F is a highly receptive to
NMR experiments (Washton and Mueller 2008). With the repeated pulsing of an orthogonal
radiofrequency field, one can infer nuclear resonance from the relaxation of nuclei between the
pulses. Because resonance is dependent on electron shielding, the same nuclide will have
different resonant frequencies contingent on its chemical environment (Lambert and Mazzola
15

Figure 5. Proposed mechanism for the silanization of a silicate mineral surface. (a) TFS is first
hydrolyzed in solution, and subsequently (b) condenses with the mineral surface, selectively
binding to Q
3
hydroxyl sites (see text for discussion). Adapted from Washton et al. (2008).
a

b

16

2003). This makes it possible to quantify
19
F in TFS that has adsorbed to Q
3
hydroxyl sites. The
sensitivity of
19
F NMR makes it an ideal method for this study.
Crystalline samples, however, present an obstacle to analysis: because nuclei in these
solids are part of a lattice, the magnetic dipoles of adjacent nuclei are anisotropic (MacKenzie
and Smith 2002). In nuclei with a quantum spin of ½, such as
19
F, this anisotropy causes a
problematic broadening of the signal. With fine-grained solids, the broadening is reduced by
spinning the sample in a rotor, oriented at an angle of 54.74° to the main field. At that angle, the
anisotropy caused by interactions of adjacent dipoles is largely removed (MacKenzie and Smith
2002). This is called Magic Angle Spinning, or MAS NMR (See Appendix B for a more
detailed discussion of theory).
The approach in this study is to use NMR experiments to quantify
19
F in TFS on treated
feldspar samples. Q
3
hydroxyl site concentrations are calculated from this analysis (see
Methods).
1.7_ Study Areas
Samples were taken from moraines in two glacial chronosequences. These are comprised
of discrete soil deposits with well-constrained ages, such that the weathering duration can be
assumed as approximately equal to deposit age (Bockheim 1990, Taylor and Blum 1995, Hodson
et al. 1998, White et al. 2008, Schaller et al. 2009, Phillips et al. 2009, Dahms et al 2012). This
sampling approach was taken in order to draw a comparison with the literature on temporal
changes in dissolution rates.
Glacial chronosequences form in mountainous regions and record successive glacial-
interglacial events (Madole 1980, Birkeland et al. 2003, Phillips et al. 2009, Ward et al. 2009).
In the case of most Pleistocene-age chronosequences, each major glaciation tended to have lower
17

intensity than its predecessor. Consequently, some regions with sufficient relief feature a
succession of discrete, temporally diverse post-glacial deposits (figure 6). The exposure age of
each individual member can be determined by various methods, including stratigraphic
relationships and radiocarbon dating (Hodson et al. 1998), and
36
Cl and
10
Be cosmogenic nuclide
dating (Phillips et al. 2009, Ward et al. 2009).
Another advantage of glacial chronosequences and similar field areas is that such sites
have been the subject of numerous investigations on silicate weathering rates in nature (Taylor
and Blum 1995, Clow et al. 1996, Jin et al. 2008). As such, the results from this study—on
feldspar surface hydroxyl site concentration changes—are compared to reported weathering rates
from this literature.
Field areas in this study are chronosequences in Bishop Creek, CA, and the Front Range
of the Rocky Mountains, CO. In addition to the chronosequence in the Front Range, a
feldspathic saprolite in the Betasso Gulch, which is located on the north side of Boulder Canyon
two miles west of Boulder, CO (Dethier et al. 2012), was also sampled at several depths. This
soil has likely developed for 35 to 60 kyr (Dethier 2009).
Bishop Creek
Early investigations proposed relative ages for the most recent Pleistocene glaciations in
the Sierra Nevada Mountains—the Tioga and Tahoe events—by correlating them with the
Wisconsin and Iowan events, respectively (Blackwelder 1931, Burke and Birkeland 1979).
More recent paleoclimatological studies of lacustrine sediments have confirmed that each major
glaciation was partitioned by numerous sub-events (Benson et al. 1996, Bischoff et al. 1997,
Menking et al. 1997).
18

Figure 6. Aerial map of Bishop Creek and vicinity, with shaded regions showing moraines and
their approximate ages. The chronosequence effect is a consequence of having a sequence of
successively less intense glaciations—as such, each subsequent termination is higher than the
last. Red scale bar is 2 km. Imagery and elevation model courtesy of Google Earth. Moraine
ages adapted from Phillips (2009).

19

Phillips et al. (2009) used
36
Cl cosmogenic exposure-age dating on moraine boulders in
the vicinity of Bishop Creek to constrain the ages of these events. Glacial advance-retreat cycles
of Tahoe and Tioga ages, as well as younger deposits, were identified. Sites selected for this
field area are based on the work of Phillips et al. (2009) (figure 7).
Front Range
In the Rocky Mountains, the Pleistocene glaciations contemporaneous with the Tahoe
and Tioga events were identified as, respectively, the Bull Lake and Pinedale glaciations
(Blackwelder 1915, Richmond 1965, Madole 1980, Birkeland 2003).
Numerical models combined with
10
Be cosmogenic dating in Ward et al. (2009) provide a
detailed assessment of Pinedale-age Front Range glacial chronology. Schildgen (2000) used
10
Be,
26
Al, and
36
Cl isotopic dating techniques to identify Pinedale and Bull Lake-age deposits.
Front Range sample site selection for the current study is based on the work of these authors
(figure 7).

20

Figure 7. Maps of the two field areas for this study, Bishop Creek in the Sierra Nevada and the
Front Range in the Colorado Rocky Mountains. Front Range field areas are regionally located
within the dotted box in each inset map. Each point represents a sample site. Numbers adjacent
to sample sites, and the color of each point, denote the age of soil exposure after glacial retreat,
in ka. Orange points are creek or riverbed samples. The question mark above the oldest Front
Range sample signifies that its age is somewhat ambiguous (see text). Relief base-maps made
using data from the USGS National Map interface.

?
21

2_ Methods
2.1_ Sample Collection
Moraines
Sediment samples were collected from Bishop Creek in the Sierra Nevada Mountains,
CA, and the southern Front Range of the Rocky Mountains, CO (figure 7). Samples from
thirteen sites along Bishop Creek and six sites along the Front Range were taken from below the
upper soil horizons of glacial moraines dated in separate studies. Chronosequence deposits from
both field areas span approximately 140 ky of the late Pleistocene glaciations (10 to 150 ka)
(Madole 1980, Birkeland et al. 2003, Phillips et al. 2009, Ward et al. 2009).
Profiles were dug on lateral moraines to a depth of up to 1.5 meters. Sediment samples
were taken below the subsoil horizon at each site, to minimize variability in the manner of
weathering from among samples. At sites for the youngest deposits, this material was at near the
surface at approximately 0.1 m depth.
River beds
In addition to the moraine profiles, samples were taken from the beds of major rivers in
each field area: Bishop Creek in the Sierra Nevada, and Boulder Creek in the Rocky Mountains.
Bed sediment from both Bishop and Boulder Creeks was removed directly from each bank, air-
dried briefly in the field and dried in the lab for several hours at 100 °C.
Soils
To examine the changes in hydroxyl concentration through a classic weathering profile,
samples were also collected at thirteen depths of a feldspathic saprolite adjacent to the Betasso
Treatment Plant west of Boulder, CO.
22

Samples were collected from an exposed cut of the profile in 0.2 meter increments from
the surface to 2 meters depth. In addition, two samples were taken at approximately 0.7 meters
depth from a mafic enclave within the profile.
2.2_ Sample Preparation
Subsets of each sample were sonicated in acetone at 5 minute intervals. Samples were
decanted to remove clays, and this procedure was repeated until the sonicated mixture was
transparent. Sonicated samples were dried for 30 to 45 minutes at approximately 100 °C to
remove residual acetone (boiling point 56 °C). Separation of feldspar mineral grains was
completed by hand with an optical microscope and forceps. Quartz, mafic mineral grains,
organic matter, and aggregated grains were discarded. One to 3 grams of feldspars were
separated from each sample.
Each sample was ground with an agate mortar and pestle and sieved to a size fraction of
53-150 microns to ensure that material was fine enough to spin evenly in the spectrometer. The
unground 1-2mm grains were too large to facilitate even spinning. The ground samples were
subjected to a second round of sonication to remove any remaining fine particles. Cleaned
samples were dried once more for 12 to 14 hours at 125 °C.
Samples were allowed to cool and placed into 100 mL Schlenk flasks. One mL of
(3,3,3trifluoropropyl) dimethylchlorosilane (TFS) was pipetted into each flask and dissolved in
25 mL of anhydrous toluene. Flasks were evacuated and purged with argon gas. Flasks were
then mechanically shaken for 72 hours at a moderate speed to allow complete TFS reaction with
all hydroxyl sites. After this, samples were either dried in a vacuum oven for 1 hour at
approximately 120 °C, or run in a centrifuge-evaporator until completely dry. Treatment
methodology was adapted from the approach of Sanders et al. (2010).
23

2.3_ Nuclear Magnetic Resonance (NMR) Spectroscopy
19
F
Experiments were done on one Bruker 750 MHz and two Varian 600 MHz solid state
MAS NMR spectrometers, all with two-channel probes. Two of these spectrometers are located
at the Pacific Northwest National Lab, and one is located at the Zilkha Neurogenetic Institute at
USC. Dried samples were packed into 4mm-diameter ceramic rotors. Bruker rotors accepted
between 135 and 145 mg of material, and Varian rotors accepted between 60 and 70 mg. A
quantitative standard of sodium trifluoroacetate (Na TFA) diluted in silica gel was also prepared
and analyzed. The concentration of the prepared dilution was 1.68 * 10
18

19
F spins per gram
(2.79 * 10
-5
moles of fluorine per gram of dilution). Samples were run through one-pulse MAS
19
F experiments at a spin frequency of 15 kHz. Aside from the recycle delay and number of
scans, the standard was run with the same acquisition parameters as the samples.
The data collected from both spectrometers were processed with NUTS NMR software
(Acorn, Inc.). There was a background
19
F signal due to fluorine-bearing Teflon in the lining of
the Bruker probe. Initially,
19
F peaks from TFS were indiscernible because this background
convoluted the baseline of the Fourier-transformed raw data. However, it was possible to
digitally shift background points off each free induction decay signal. By doing a linear back
prediction in NUTS, the TFS data were resolved. Thirty two points on each Bruker experiment
were shifted off and back predicted with a NUTS linear prediction routine. There was also a
minor fluorine background due to the Varian probe, but this was much less significant. Only
twelve points per experiment were shifted off and back predicted for experiments done on the
Varian instrument.
24

All resultant TFS peaks were phased, and peak areas were integrated with a NUTS line
fitting routine. Peak areas for the samples were compared to the peak area for the quantitative
standard to determine concentrations of fluorine in each treated sample. Replicates of each
signal were processed multiple times to ensure precision. A Q
3
hydroxyl site count for each
sample was calculated from this result by dividing by three since
19
F exists as an F
3
group in
each TFS molecule.
27
Al
One-pulse
27
Al MAS NMR experiments were also run for some of the samples. This was
done to assess whether there was a significant quantity of clay minerals with octahedral (six-
coordinate) Al on the surfaces of the feldspar grains. Crystalline feldspar is characterized by
tetrahedral, or four-coordinate, Al. The
27
Al MAS NMR was also done to assess potential trends
in the
27
Al spectral characteristics across samples, i.e. as a function of age.
2.4_ Microwave-Plasma Optical Emission Spectrometry (MP-OES)
Elemental analyses were completed to test for compositional variability among samples.
Samples that had been TFS-treated and analyzed by NMR were subjected to a multi-step
dissolution procedure to obtain Na, K, and Ca salts. Aliquots of approximately 100 mg each
were taken from ground, TFS-treated samples. These were loaded with 70% distilled HNO
3
and
58% HF, dissolved for 24 hours, and brought to 85 °C to evaporate the solvent. Subsequently,
samples were loaded with aqua regia and dissolved for another 24 hours. After completely
drying, samples were diluted in 5% distilled HNO
3
. Dilutions were run on an Agilent MP-OES
instrument alongside Na, K, and Ca standards with cation concentrations ranging from 1-50 ppm.
2.5_ Brunauer Emmett Teller (BET) and geometric specific surface area measurements
To normalize hydroxyl site concentrations to specific surface area (SSA) of the mineral
25

grains, unground aliquots of sample were run on a Micromeritics ASAP 2010 physisoption
instrument employing the method of Brunauer, Emmett, and Teller (1938). The minimum
sample input threshold of reasonable precision was determined for this instrument (Appendix C).
Samples were loaded into sample tubes that had been dried. Each sample was degassed
for three hours, or longer if necessary, until all residual moisture had been removed from mineral
surfaces. Analyses were 10-point BET isotherms with N
2
as the analysis gas. Only a subset of
the samples was analyzed for SSA. For the other samples, a geometric surface area calculation
was done, by applying a surface roughness coefficient calculated using the method of White and
Brantley (2003). All surface area data are reported in Table 1.
2.6_ X-ray diffraction (XRD)
A subset of the ground samples were run in a PANalytical MPD X-ray diffractometer
with a Cu anode. Approximately 50 mg of sample was loaded onto a single-crystal quartz slide
and subjected to X-rays through a 2θ range of 10° to 75°. Data were processed and referenced
using JADE software.
2.7_ Staining
A separate subset of unground samples were stained with sodium cobaltinitrite and
amaranth, using adapted methodology from Baily and Stevens (1960) and Hynek et al. (2011).
This was done in order to distinguish plagioclase, K-feldspar, and perthites. Individual grains
were mounted in plastic petri dishes coated with contact adhesive. Care was taken to avoid
immersing grains in the adhesive. Each set of grains was etched with hydrofluoric acid vapor for
5 minutes, after which the grains were soaked separately in supersaturated sodium cobaltinitrite
and amaranth solutions, with a barium chloride rinse in between. Grains were allowed to dry for
12 hours at room temperature.
26

3_ Results
3.1_
19
F NMR experiments
In the
19
F spectra from these experiments in this study, TFS peaks (at approximately -73
ppm) were of sufficient amplitude to integrate their areas. Still, signal-to-noise ratios were high
in some of the spectra. As shown in figure 8a, noise can affect the flatness of the spectral
baseline (compared to the relatively flat
27
Al baseline in figure 8b). Integrating these peaks using
a combination of Gaussian and Lorentzian approximations therefore introduces a small source of
error.
Q
3
Hydroxyl site concentrations, normalized to specific surface area (SSA) of each
sample, are reported in Table 1. In the Front Range chronosequence, these data show
considerable scatter as a function of sample age. This may be a due to analytical error associated
with running samples on multiple NMR spectrometers, and referencing them to separately
prepared Na TFA dilutions. There is also some question as to the exact age of the oldest sample,
because it was collected along the edge of a much younger moraine which was difficult to
distinguish geomorphologically.
Bishop Creek chronosequence Q
3
OH site concentrations were more reproducible
between spectrometers. They are plotted in figure 9 as a function of sample age. Concentrations
are normalized to specific surface area. There appears to be a notable increase in concentration
from younger to older samples.
Net accumulation rates of Q
3
OH sites for each sample were calculated. It is assumed
that fresh, unweathered feldspar is not appreciably protonated. As such, the net accumulation
d[OH]/dt, was calculated as:
d[OH]
x
/dt = ΔC/Δt (3)
27

Figure 8. Examples of processed spectra from (a)
19
F and (b)
27
Al MAS NMR experiments. (a)
The central peak at approximately -72 ppm is associated with
19
F in TFS. (b) Each of the
27
Al
experiments run from this sample set had a split peak similar to the one shown here at
approximately 60 ppm, associated with tetrahedral Al (four-coordinate, or Al
[4]
). There were
also very minimal peaks near 5 ppm associated with octahedral Al (six-coordinate, or Al
[6]
). In
both spectra shown, asterisks denote spinning sidebands, not separate peaks. Sidebands are a
consequence of spinning the sample (see Appendix B page B4). They are part of the signal
associated with TFS, and must be integrated as such when quantifying abundance.
a

b

28

Figure 9. Q
3
OH site concentrations versus age of deposit for samples from the Bishop Creek
chronosequence. Concentrations for all samples are normalized to BET specific surface area.

29

where ΔC is the concentration of Q
3
sites (Δ indicates that this is the net change from an
unprotonated feldspar surface), and Δt is the time between deposition and sampling, i.e. the
sample exposure age. It should be noted that this is a linear rate calculation that is used to
maintain consistency with weathering rate calculations done in the literature (Taylor and Blum
1995, White and Brantley 2003, Maher et al. 2004, Reeves and Rothman 2013). This approach
ignores the potential nonlinearity in rate trends and consequently is an incomplete representation
of weathering rate evolution.
As shown in figure 10, samples from the Bishop Creek chronosequence show a power-
law decrease in Q
3
OH site accumulation rate as a function of deposit age. Net accumulations
(d[OH]/dt) are reported in table 1, along with sample identification information.
In figure 10, Q
3
OH site accumulation rates are reported. This was done in order to draw
a comparison with silicate dissolution rates compiled by White and Brantley (2003). The best-fit
trend line shown in figure 9 can be represented by the equation
d[OH]/dt = 9*10
-19
t
-0.58
(4)
This compares to an overall silicate weathering rate equation reported by White and
Brantley (2003) of
d[W]/dt = 3.1*10
-13
t
-0.61
(5)
where [W] is the concentration of weathered material. Rates in equations (4) and (5) are
in units of mol/m
2
/s.
OH site concentrations at different depths in the Betasso saprolite profile are reported in
table 1. There was notable scatter in SSA-normalized Q
3
OH site concentrations as a function of
depth, similar to the scatter noted in the Front Range chronosequence. As mentioned above, this
may be a consequence of analytical error related to running on separate spectrometers with
30

separate references. With the Betasso profile, an additional factor contributing to scatter may be
the normalization to specific surface area, which itself was quite variable throughout the profile
(SSA values reported in table 1).
3.2_ Elemental distributions and temporal variations determined by MP-OES experiments
In the youngest samples from both field areas, distributions of Na and K are very similar,
with much lower relative amounts of Ca. This suggests the mineral distribution of the samples is
mainly of alkali feldspar ([Na,K]AlSi
3
O
8
) with small amounts of anorthite (CaAl
2
Si
2
O
8
) or other
low-albite plagioclase feldspars.
Relative to pristine feldspar, all samples show low absolute cation concentrations (table
1). If the feldspar in these samples is comprised mainly of albite and microcline (see XRD
results below), it can be assumed that the bulk molar mass should be somewhere in the range of
270 g/mol. Also note that the molar ratio of Ca, Na, or K to their respective feldspar phases is
1:1. As such, cation concentrations should be in the range of 3.7 mmol/g. However, the actual
cation-to-mineral ratio ranges from 20% to 70% of the expected value (see total cation-to-
expected cation ratios in table 2).
In samples across the Bishop Creek chronoseuqence, some compositional variability with
age was observed. Depletion of Ca and Na relative to K with age was notable, as well as some
depletion of Na relative to Ca (figure 12 and table 2). This suggests that plagioclase was
preferentially weathered relative to potassium feldspar, as might be expected (Siefert 1967,
Wilson and McHardy 1980, White et al. 2001, White and Brantley 2003).
3.3_ X-ray Diffraction
XRD patterns show albite, microcline, and quartz (example in figure 11). Several peaks
do not match neatly with reference patterns, and thus remain unaccounted for. Overlapping
31

Table 1. Samples from both field areas, site coordinates, soil exposure ages, specific surface areas, OH counts, and elemental data.
[Ca]

mmol/g
0.12
-
-
-
0.30
0.20
0.18
-
0.22
-
-
0.24
-
0.18
-
0.16
0.08
0.09
-
0.14
-
0.54
-
-
-
-
0.27
[K]
mmol/g
0.38
-
-
-
0.94
0.65
1.19
-
0.65
-
-
0.79
-
1.04
-
1.26
0.77
0.93
-
1.39
-
1.34
-
-
-
-
0.76
[Na]

mmol/g
0.22
-
-
-
1.02
0.68
1.08
-
0.67
-
-
0.93
-
1.34
-
0.93
1.57
1.81
-
1.18
-
0.87
-
-
-
-
1.63
d[OH]
/
dt
*10
-19

mol
/
m^ 2∙s

6.58
7.09
-
-
12.12
-
-
-
16.09
15.93
-
29.96
-
2.59
-
2.42
2.36
-
-
-
-
-
-
-
-
-
-
[OH] *10
-7

mol/m
2

2.07
1.57
-
-
0.65
-
-
-
1.01
0.65
-
0.94
-
0.25
3.00
0.39
0.31
-
-
-
1.60
0.26
-
6.76
-
1.97
0.20
SSA
m
2
/g
0.9
1.3
1.1
1.1
0.9
-
-
-
1.0
1.0
-
0.4
-
0.9
1.0
1.1
0.9
-
0.2
-
2.2
2.0
-
-
0.4
0.8
0.6
Age (ka)
100
70
14
14
17
14
-
20
20
20
-
10
-
17
14
14
121*
-
5
5
35-60
35-60
35-60
35-60
35-60
35-60
35-60
Lon. (W)
118° 31' 24.49"
118° 30' 48.13"
118° 31' 37.20"
118° 32' 06.00"
118° 32' 16.58"
118° 31' 53.20"
118° 32' 03.26"
118° 32' 05.10"
118° 32' 03.12"
118°32' 07.44"
118° 30' 55.44"
118° 35' 5.568"
105° 31' 35.52"
105° 31' 25.56"
105° 38' 05.94"
105° 37' 55.08"
105° 31' 19.38"
105° 21' 06.22"
105° 38' 51.93"
105° 38' 51.77"
105° 20' 19.90"
105° 20' 19.90"
105° 20' 19.90"
105° 20' 19.90"
105° 20' 19.90"
105° 20' 19.90"
105° 20' 19.90"
Lat. (N)
37° 19' 37.38"
37° 19' 27.88"
37° 18' 32.00"
37° 18' 2.04"
37° 17' 55.03"
37° 18' 28.44"
37° 17' 58.92"
37° 18' 8.496"
37° 18' 15.91"
37° 18' 16.16"
37° 19' 28.02"
37° 14' 50.64"
39° 56' 53.52"
39° 56' 56.10"
39° 59' 51.72"
39° 59' 35.88"
39° 56' 49.32"
40° 00' 17.06"
40° 09' 12.56"
40° 09' 09.61"
40° 56' 16.44"
40° 56' 16.44"
40° 56' 16.44"
40° 56' 16.44"
40° 56' 16.44"
40° 56' 16.44"
40° 56' 16.44"
Sample
Type
moraine
moraine
moraine
moraine
moraine
moraine
creek
moraine
moraine
moraine
dry creek
moraine
moraine
moraine
moraine
moraine
moraine
creek
moraine
moraine
saprolite
saprolite
saprolite
saprolite
saprolite
saprolite
saprolite
ID
E.5.11.01
E.5.11.02
E.5.11.03
E.5.11.04
E.5.11.05
E.5.11.06
E.5.11.07
E.5.11.08
E.5.11.09
E.5.11.10
E.5.11.11
E.5.11.12
E.6.11.01
E.6.11.02
E.6.11.03
E.6.11.04
E.6.11.05
E.6.11.06
St. V.l
St. V.u
B.B.0.4
B.B.0.7
B.B.1.0
B.B.1.2
B.B.1.4
B.B.1.6
B.B.2.0
Field
Area
Bishop
Creek
Front
Range

*The age of sample 05 from the Front Range chonosequence is somewhat in question.
32

Figure 10. Net Q
3
OH site accumulation per m
2
mineral surface area, as a function of sample
exposure age, for samples from Bishop Creek. This represents the net change in OH
concentration on the mineral surface from a fresh unweathered state to its current state. It is
assumed here that unweathered feldspar is not protonated and represents the initial state for
chronosequence development. Rate of net accumulation is reported in mol/m
2
/s in order to be
consistent with dissolution rate units used by White and Brantley (2003).

33

peaks of both feldspar and quartz could not be resolved, so these results are not quantitative. The
patterns do show that some quartz is contained in the samples, possibly as small crystals attached
to larger feldspar crystals, or as improperly picked grains.
3.4_
27
Al NMR experiments
Extremely low quantities of octahedral (six-coordinate) aluminum, or Al
[6]
, were
observed in several samples from Bishop Creek. However, relative to the quantity of tetrahedral
(four-coordinate, or Al
[4]
) aluminum, the octahedral peaks were almost indistinguishable from
noise (figure 8b).
In
27
Al spectra for all samples run, as in figure 8, the tetrahedral peak at approximately 60
ppm is noticeably split into two components. This is indicative of two classes of tetrahedral Al
in slightly different chemical environments in the crystal (background in Appendix B). It is
possible that these constituent peaks delineate Al with different numbers of oxygens bridging to
Si, as noted by Pedone et al. (2012), although the chemical shifts in this study are slightly
different than those reported by those authors.
Across samples of varying age, the intensity of the component closer to 58 ppm is not
constant relative to the intensity of the component near 61 ppm (figure 13). It appears that in
older samples the ~58 ppm component is proportionally more intense than it is in younger
samples. An age dependence may imply that the split peak is actually indicative of changing
numbers of Al tetrahedra at the surface-solution interface forming OH sites. The contrast in
figure 13 is as much as 11% —considerably less than the proportional difference in Q
3
OH site
concentrations between the youngest and oldest samples. The difference in OH concentrations,
however, includes sites at both silica and alumina tetrahedra. Therefore, it is not unreasonable to
34

suggest that split
27
Al peaks like the one in figure 8b delineate different types of Q
n
alumina sites
at the feldspar surface, e.g. Q
2
and Q
3
.

4_ Discussion
4.1_ Implications of the Chronosequence Results
The net Q
3
OH site accumulation rate for the Bishop Creek chronosequence can be
compared to the silicate dissolution rate data compiled by White and Brantley (2003). The
exponent for time in White and Brantley’s trend for bulk mineral dissolution rate (-0.61, equation
5) is similar to that exponent for time in the derivative of net accumulation rate for the Bishop
Creek chronosequence (-0.58, equation 4). Thus, the Bishop Creek data suggest that the net rate
of hydroxyl site accumulation in feldspar may scale with silicate weathering rates as minerals
reside in the weathering environment through time.
This supports the dissolution mechanism proposed by Oelkers (2001) wherein the rate of
silicate surface protonation is correlated with total mineral dissolution. The theoretical
mechanism suggests that over time, a larger proportion of the exposed mineral grain surface
becomes protonated as the dissolution front propagates into the grain. As new surface is created
by dissolution, the derivative of this hydroxyl site increase gradually decreases. Concurrently,
the dissolution rate decreases, which has been observed in the lab and field (Taylor and Blum
1995, White and Brantley 2001, Maher et al. 2004, White et al. 2008, Reeves and Rothman
2013). Therefore, this derivative of hydroxyl site development can be assumed to run parallel to
the rate of overall dissolution of the mineral.
None of this demonstrates mechanistic causation, however. Whether protonation controls
stoichiometric dissolution, or they merely scale together, remains in question. Still, these
35

Figure 11. XRD Pattern of one sample from Bishop Creek. Note that, in addition to the
identified albite, microcline, and quartz peaks, there are several other peaks not consistent with
reference patterns in JADE software.

36

Table 2. Mole fractions of cations relative to one another in samples.
Field Area
Sample ID Age
(ka) Na fraction K fraction Ca fraction
Cation
total
/
Cation
expected

Bishop Creek
EAK.5.11.12
10 0.47 0.40 0.12
0.5
EAK.5.11.06
14 0.45 0.42 0.13
0.4
EAK.5.11.05
17 0.45 0.42 0.13
0.6
EAK.5.11.09
20 0.44 0.42 0.14
0.4
EAK.5.11.01
100 0.31 0.53 0.16
0.2

Front Range
St.V.l
5 0.51 0.43 0.05
0.7
EAK.6.11.04
14 0.54 0.40 0.07
0.6
EAK.6.11.02
17 0.32 0.65 0.03
0.7
EAK.6.11.05
121* 0.41 0.52 0.07
0.7

*The age of sample 05 from the Front Range chonosequence is somewhat in question.

37

Figure 12. Three-dimensional ternary diagrams for samples from the Bishop Creek
chronosequence. Each peak represents a sample. The position of a peak corresponds to the
distributions of Na, K, and Ca cations relative to one another in that given sample. Age of the
sample is represented by peak height, and identified by a label (in ka).

38

Figure 13. Semi-log plot showing proportions for different samples of one split in the
27
Al peak
associated with tetrahedral Al (Al
[4]
), relative to the whole peak (as shown in figure 9b, this peak
is split and has two parts). There are two main components to the Al
[4]
peak (see figure 9b), with
the proportions plotted here reflecting the peak closer to 58 ppm. Samples are plotted as a
function of age.

39

findings demonstrate for the first time that scaling of OH site development with overall
dissolution at near-neutral pH takes place over millennial timescales. This encourages additional
exploration of the TFS-treatment and NMR analysis of primary silicate minerals, because it
offers insights into the alteration layer that forms during weathering (Brantley 2003). The
leached alteration layer is thought to consist of tetrahedral Si and Al sites with varying degrees of
connectedness to the crystal, as well as octahedral Al—see Tsomaia et al. (2003) and Criscenti et
al. (2005) for discussion. It is also argued that in many cases the alteration layer might be
spatially heterogeneous (Brantley 2003). Like Oelkers (2001), Brantley (2003) offers a
complexation model controlled mainly by protonation of the interface. Continued exploration of
naturally weathered silicates—and quantitative analysis of hydroxyl groups at their surfaces—
may offer potential for determining the most important controls in modeling surface
complexation.
It should be noted once more that the sites being measured in this study are specifically
Q
3
hydroxyl sites. Brantley (2003) asserts that for feldspars, Q
3
sites develop as one step in a
continuous process that ends with release of tetrahedral Al or Si into solution.
Assuming this is the case, the trend observed here should represent the overall hydroxyl
site concentration. All of this cannot be verified in the scope of the current study, however. In
this respect, the TFS condensation reaction—and in general the overall mechanism of sylilation
of silicates by organosilanes as outlined by Van Der Voort and Vansant (1996)—provides only
one piece of information about the silicate surface. It is valuable in some ways to have a probe
molecule that selectively condenses to specific hydroxyl sites, but these results would be
enhanced with complementary measurements of hydroxyl sites with different coordination, i.e.
Q
2
, Q
1
. A fluorine bearing silane probe that can be shown to condense with germinal (Q
2
) or
40

vicinal (bridged) hydroxyl sites would be useful for the approach used in this study, since
19
F is a
very NMR-sensitive nucleus (Washton et al. 2008). If such a silane exists, however, it’s efficacy
for this application remains undetermined at present.
It is also interesting to consider the absolute concentration of Q
3
OH sites measured on
these samples relative to the total number of oxygens that theoretically terminate Si and Al
tetrahedra at the surface. Assuming a specific surface area of 1 m
2
/g for simplicity—this is a
reasonable approximation of the silicates in this study—and assuming a molar mass (n
silicate
) of
approximately 270 g/mol and specific gravity (ρ) of 2.5 (relative to H
2
O), the bulk moles per unit
surface area is given by
ρ * 1/n
silicate
* 1/[
s
/
v
] = 3.7∙10
-3
(6)
where [
s
/
v
] is the surface-area-to-volume ratio. Given that the molar ratio of oxygen to silicate in
feldspar is approximately 0.57, and assuming that 2 out of 3 oxygens in a unit crystal will
terminate at the surface (given feldspar crystallography), 1.4∙10
-3
moles, or 8.4∙10
23
atoms, of
oxygen should terminate per m
2
of silicate. This compares with measured values on the order of
1∙10
17
Q
3
hydroxyl sites per m
2
of silicate. The much lower measured value suggests that either
the feldspar surfaces are minimally protonated, or surfaces are occupied by Q
2
and Q
4
sites in
concentrations several orders of magnitude in excess of Q
3
.
In any case, quantifying Q
3
surface sites offers useful information about surface structure,
but it does not elucidate the relationship between protonated sites and morphologic surface
features, nor does it show whether the development of such features is controlled by the degree
of protonation. To provide insight into the role of surface morphology in this context, it might
be useful to accompany the current analyses with surface-sensitive methods such as Reflective
High-Energy Electron Diffraction.
41

Figure 14. Images of feldspar grains from a Bishop Creek chronosequence site, stained with
sodium cobaltinitrite and amaranth. (a) Potassium-feldspar stains yellow. (b) Plagioclase stains
red. (c,e) Perthite texures. (d) A grain with clear feldspar cleavage but very little stain. All scale
bars in these images denote 1 mm.

42

4.2_ Elemental Analysis and Staining
There are considerably lower cation concentrations in the samples than would be
expected from pristine feldspar (table 2). This implies non-stoichiometric leaching of cations, or
that aggregates of quartz were taken up during picking. XRD results suggest that some quartz
was taken up during picking, but not enough to account for cation concentrations as low as those
reported. Still, the presence of any quartz is important to note, because it implies that some of
the measured Q
3
hydroxyl sites may have been on quartz rather than feldspar. This potentially
skews the observed temporal trend to some degree, since quartz and feldspar weather at different
rates. Because the stability of quartz is so high, though, OH concentrations on quartz surfaces
are likely to be extremely low, and thus contribute minimally to the measured Q
3
OH site
concentrations.
Overall, the XRD provides some insight, but staining of separate sample aliquots was
also done to better understand the mineralogical attributes of samples.
Results of the sodium cobaltinitrite and amaranth stains are shown in figure 14. Using
the procedure described in Methods, potassium feldspar should stain yellow (Baily and Stevens
1960) (figure 14a) and plagioclase should stain red (figure 14b) (Baily and Stevens 1960, Hynek
et al. 2011), while quartz or leached material will remain unstained (Hynek et al. 2011). As
shown in the figure, some grains stained completely (figure 14a-c) while others have very little
to no stain (figure 14d). It is possible, as mentioned above, that grains which are highly altered
by weathering can partly account for the low measured cation concentrations. In figure 14d in
particular, a very clear feldspar cleavage plane can be seen, but the grain sustained very little
stain, indicating extensive leaching (Hynek et al. 2011).
43

In addition to evidence of leaching, the stains are also indicative of perthites, since grains
with albite exsolution lamellae intergrown in orthoclase would be exhibit both types of stain on a
single surface (figure 14c and e) (see Parsons et al. 2005 for perthite definition).
As shown in the results of the MP-OES experiments, there is some compositional
variability in the Bishop Creek chronosequence, particularly the depletion of K relative to Na
and Ca.
This implies preferential weathering of albite (NaAlSi
3
O
8
) and anorthite (CaAl
2
Si
2
O
8
),
relative to K-feldspar (KAlSi
3
O
8
), or the preferential weathering of albite lamellae in perthitic
feldspar (for discussion see Wilson and McHardy 1980). Such variable weathering is of
potential concern for this study, because changes in OH site accumulation rate might be the
result of changing mineralogy, and not the overall weathering of feldspar.
For Bishop Creek samples, OH site accumulation rates span an order of magnitude (table
1), a much greater change than the compositional changes for that chronosequence (table 2,
figure 12). Still, with future studies it will be valuable to address the challenging issue of
developing a non-destructive method for mineralogical partitioning of feldspar. This may
require careful selection of field areas, since perthites present an obstacle in this respect (figure
14c and e). Once that is done, each separate mineral phase can be treated and analyzed in turn.
4.3_
27
Al NMR and future directions
As mentioned in the results section, the
27
Al NMR experiments may indicate changing
alumina species at the feldspar surface. The literature on split tetrahedral
27
Al peaks in silicates
is very limited, and it would be of use to explore this issue in more depth.
This could be done using Cross Polarized Magic Angle Spinning NMR (CP-MAS NMR)
experiments which employ a pulse sequence that allows spin transfer between two nuclei
44

(MacKenzie and Smith 2002). In
1
H—
27
Al CP-MAS NMR, the experiment is selective to Al-
bearing species near the protonated surfaces of mineral grains (Criscenti et al. 2005).
Such experimental methods aimed at tetrahedral Al species could be an informative next
step in studying aluminosilicate surfaces. Criscenti et al. (2005) did extensive modeling of
aluminosilicate by calculating parameters for different tetrahedral and octahedral Al species. In
their experimental work, though, these authors were mainly concerned with the speciation of
octahedral Al to determine whether clay formation was a process that took place directly on the
surfaces of primary silicate minerals, or in solution. Using
1
H—
27
Al CP-MAS NMR to study
tetrahedral Al on feldspars, and in particular to coax out subtleties of the tetrahedral Al peak, has
the potential to elucidate the interdependence of surface tetrahedral species and overall
dissolution. The split
27
Al peak found in Bishop Creek samples provides a possible motivation
to continue
27
Al NMR on silicates with an emphasis on surface-sensitive techniques, i.e. cross
polarization. If different types of hydroxyl sites on feldspar surfaces can be measured directly
with NMR rather than inferred through treatment with an organosilane (e.g. TFS), the process of
exploring silicate surfaces could be greatly streamlined and enhanced.

5_ Closing remarks
The question of why weathering rates decrease over long timescales certainly remains.
However, each step in the quantitative exploration of the silicate surface is one closer to
understanding the intrinsic controls on these rates. In this study, the viability of the TFS-
treatment procedure and NMR analysis on naturally weathered feldspars has been demonstrated.
It has been shown that a power-law decrease in OH-site accumulation rates may exist for
weathering feldspars over long timescales (figure 10). Comparing these rate changes to overall
45

silicate weathering rate changes compiled by White and Brantley (2003), it appears at this early
stage that OH site accumulation may scale with overall dissolution as silicates undergo natural
weathering.

References

Aagaard P, Helgeson HC. “Thermodynamic and kinetic constraints on reaction rates among
minerals and aqueous solutions.I. Theoretical considerations.”American Journal of Science, 282,
237-285. 1982.

Andrew ER, Bradbury A, Eades RG. “Nuclear Magnetic Resonance spectra from a crystal
rotated at high speed.” Nature 182, 1659. 1958.

Andrew ER, Newing RA. “The Narrowing of Nuclear Magnetic Resonance Spectra by
Molecular Rotation in Solids.” Proceedings of the Physical Society, 72, 6, 959-972. 1958.

Baily EH, Stevens RE. “Selective Staining of K-Feldspar and Plagioclase on rock slabs and thin
sections.” American Mineralogist, 45, 1020-1025. 1960.

Benson LV, Burdett JW, Kashgarian M, Lund SP, Phillips FM, Rye RO. “Climatic and
hydrologic oscillations in the Owens Lake Basin and adjacent Sierra Nevada, California”
Science, 274, 746-749. 1996.

Berner RA, Lasaga AC, Garrels RM. “The carbonate-silicate geochemical cycle and its effect on
atmospheric carbon dioxide over the past 100 million years.”American Journal of Science, 283,
641-683. 1983.

Birkeland PW, Shroba RR, Burns SF, Price AB, Tonkin PJ.“Integrating soils and
geomorphology in mountains—an example from the Front Range of Colorado.”Geomorphology,
55, 329-344. 2003.

Bischoff JL, Menking KM, Fitts JP, Fitzpatrick JA. ”Climatic oscillations 10,000–155,000 yr
B.P. at Owens Lake, California, reflected in glacial rock flour abundance and lake salinity in
Core OL-92” Quaternary Research, 48, 313-325. 1997.

Blackwelder E.“Cenozoic history of the mountains of central Wyoming” Journal of Geology, 23,
97-117, 193-217, 307-340. 1915.

46

Blackwelder E. “Pleistocene glaciation in the Sierra Nevada and Basin Ranges” Geological
Society of America Bulletin, 42, 865-922. 1931.

Blum AE, Lasaga AC. “Role of surface speciation in the low temperature dissolution of
minerals.” Nature, 4, 431-433, 1988.

Blum AE, Lasaga AC. “The role of surface speciation in the dissolution of albite.” Geochimica
et CosmochimicaActa, 55, 2193-2201. 1991.

Blum AE, Stillings LL. “Feldspar dissolution kinetics.”Chemical Weathering Rates of Silicate
Minerals, ed. White AF, Brantley SL, 291-351, 1995.

Bockheim JG. “Soil development rates in the Transantarctic Mountains.” Geoderma, 47, 59-77.
1990.

Boudart M. “Consistency between kinetics and thermodynamics.” Journal of Physical
Chemistry, 80, 2869-2870. 1976.

Brantley SL. “Reaction kinetics of primary rock forming minerals under ambient conditions.”
Treatise on Geochemistry, 5, 73-117, 2003.

Breene RG. The Shift and Shape of Spectral Lines. In Geophysical Research Papers, 41. 1955.

Brunauer S, Emmett PH, Teller E. “Adsorption of gases in multimolecular layers.”Journal of the
American Chemical Society, 60, 309-319. 1938.

Burke RM, Birkeland PW. “Re-evaluation of multiparameter relative dating techniques and their
application to the glacial sequence along the eastern escarpment of the Sierra Nevada,
California” Quaternary Research, 11, 21-51. 1979.

Chadwick OA, Chorover J. “The chemistry of pedogenicthresolds.” Geoderma, 100, 321-353.
2001.

Choh SH, Stager CV. “Nuclear Spin Hamiltonian for NMR in Antiferromagnets.” Physics
Letters, 30A, 3, 184-185. 1969.

Chou L, Wollast R. “Study of the weathering of albite at room temperature and pressure with a
fluidized bed reactor.” Geochimica et Cosmochimica Acta, 48, 2205-2217. 1984.

47

Chou L, Wollast R. “The steady-state kinetics and dissolution mechanisms of albite.” American
Journal of Science, 285, 963-993. 1985.

Chuiko AA, Turtykh VA, Khranovskii VA, Egorov YP, Roev LM. “Spectroscopic study of the
interaction of chlorosilanes with the surface of Aerosil.” Teoreticheskaya I Eksperimental ‘naya,
2, 2, 247—English Translation 2, 2, 189-193. 1966.

Clow DW, Drever JI. “Weathering rates as a function of flow through an alpine soil.” Chemical
Geology, 132, 131-141. 1996.

Corriu RJP, Guerin C. “Nucleophilic displacement at silicon: stereochemistry and mechanistic
implications.” Journal of Organometallic Chemistry, 198, 231-320. 1980.

Criscenti LJ, Brantley SL, Mueller KT, Tsomaia N, Kubicki JD. “Theoretical and 27Al CPMAS
NMR investigation of aluminum coordination changes during aluminosilicate dissolution.”
Geochimica et CosmochimicaActa, 69, 9, 2205-2220. 2005.

Dahms D, Favilli F, Krebs R, Egli M. “Soil weathering and accumulation rates of oxalate-
extractable phases derived from alpine chronosequences of up to 1 Ma in age.” Geomorphology,
151, 15, 99-113. 2012.

Davis JA, James RO, Leckie JO. “Surface ionization and complexation at the oxide/water
interface.” Journal of Colloid and Interface Science, 63, 3, 480-499. 1977.

Dethier DP. Talk: “Evolution of soils derived from weathering of saprolitic bedrock and
colluvium, Betasso Catchment, Front Range, CO.” Geological Society of America Annual
Meeting (Abstracts with Programs), 41, 7, 336. 2009.

Dethier DP, Birkeland PW, McCarthy JA. “Using the accumulation of CBD-extractable iron and
clay content to estimate soil age on stable surfaces and nearby slopes, Front Range, Colorado.”
Geomorphology, 173, 1, 17-29. 2012.

Doskocilova D, Schneider B. “Narrowing of Proton NMR lines by Magic Angle Rotation.”
Chemical Physics Letters, 6, 4, 381-384. 1970.

Eyring H. “The activated complex in chemical reactions.” Journal of Chemical Physics, 3, 107-
115. 1935.

48

Fantle MS, DePaolo DJ.“Ca isotopes in carbonate sediment and pore fluid from ODP Site 807A:
The Ca
2+
(aq)–calcite equilibrium fractionation factor and calcite recrystallization rates in
Pleistocene sediments.” Geochimica et Cosmochimica Acta, 71, 2524-2546. 2007.

Fenter P, Zapol P, He H, Sturchio NC. “On the variation of dissolution rates at the orthoclase
(001) surface with pH and temperature.” Geochimica et Cosmochimica Acta, 141, 598-611.
2014.

Ferrari AM, Ugliengo P, Garrone E. “Geminal Silica Hydroxyls as Adsorbing Sites: an ab initio
study.” Journal of Physical Chemistry, 97, 11, 2671-2676. 1993.

Goudie AS, Viles HA. “Weathering and the global carbon cycle: geomorphological perspetives.”
Earth Science, 113, 59-71, 2012.

Glasstone S, Laidler KJ, Eyring H. Theory of rate processes. 611 p. 1941.

van Grinsven JJM, van Reimsdijk WH. “Evaluation of batch and column techniques to measure
weathering rates in soils.” Geoderma, 52, 41-57. 1992.

Helgeson HC, Murphy WM, Aagaard P. “Thermodynamic and kinetic contraints on reaction
rates among minerals and aqueous solutions: Rate constants, effective surface area, and the
hydrolysis of feldspar.” Geochimica et Cosmochimica Acta, 48, 2405-2432. 1984.

Hochella MF, Banfield JF. “Chemical weathering of silicates in nature: a microscopic
perspective with theoretical considerations.” Chemical Weathering Rates of Silicate Minerals,
ed. White AF, Brantley SL, 353-406, 1995.

Hodson ME, Langan SJ, Kennedy FM, Bain DC. “Variation in soild surface area in a
chronosequence of soild from Glen Feshie, Scotland and its implications for mineral weathering
rate calculations.” Geoderma, 85, 1-18. 1998.

Hynek SA, Brown FH, Fernandez DP. “A rapid method for hand picking potassium-rich feldspar
from silicic tephra.” Quaternary Geochronology, 6, 285-288. 2011.

Jin L, Hamilton SK, Walter LM.“Mineral weathering rates in glacial drift soils (SW Michigan,
USA), New constraints from seasonal sampling of waters and gases at soil monoliths.” Chemical
Geology, 249, 129-154. 2008.

49

Jin L, Ravella R, Ketchum B, Bierman PR, Heaney P, White T, Brantley S. “Mineral weathering
and elemental transport during hillslope evolution at the Susquenhanna/Shale Hills Critical Zone
Observatory.” Geochimica et Cosmochimica Acta, 74, 3669-3691. 2010.

Kim K. “Plagioclase weathering in the groundwater system of a sandy, silicate aquifer.”
Hydrological Processes, 16, 1793-1806. 2002.

Lambert HB, Mazzola EP.Nuclear Magnetic Resonance Spectroscopy: An Introduction to
Principles, Applications, and Experiemental Methods. Pearson Education, Prentice Hall, Inc.,
Upper Saddle River. 2004.

Lasaga AC. Kinetic Theory in the Earth Sciences.Princeton University Press, Princeton, NJ, 781
p. 1998.

Lee MR, Parsons I. “Microtextural controls of weathering of perthitic alkali feldspars.”
Geochimica et Cosmochimica Acta, 59, 4465-4488. 1995.

Lin SH, Li KP, Eyring H. “Theory of reaction rates in condensed phases.”Physical chemistry: An
advanced treatise v. 7, ed. Eyring H, 1-56. 1975.

Lochmuller CH, Kersey MT. “Effect of Thermal Pretreatment on the Surface Reactivity of
Amorphous Silica.” Langmuir, 4, 3, 572-578. 1988.

Luttge A. Bolton EW, Lasaga AC. “And interferometric study of the dissolution kinetics of
anorthite: the role of reactive surface area.” American Journal of Science 299, 652-678. 1999.

MacKenzieKJD, Smith ME.Multinuclear Solid-State Nuclear Magnetic Resonance of Inorganic
Materials.Pergamon. 2002.

Madole R. “Time of Pinedale deglaciation in North Central Colorado: further considerations.”
Geology, 8, 118-122. 1980.

Maher K, DePaolo DJ, Lin JC. “Rates of silicate dissolution in deep-sea sediment: In situ
measurement using
234
U/
238
U of pore fluids.” GeochimicaetCosmochimicaActa, 68, 22, 4629-
4648. 2004.

Maher K. “The dependence of chemical weathering on fluid residence time.” Earth and
Planetary Science Letters, 294, 101-110. 2010.

Marschner H. Mineral Nutrition of Higher Plants. London, Academic Press, 529 p. 1995.
50

Menking KM, Bischoff JL, Fitzpatrick JA, Burdette JW, Rye RO. “Climatic/hydrologic
oscillations since 155,000 yrB.P. at Owens Lake, California, reflected in abundance and stable
isotope composition of sediment carbonate” Quaternary Research, 48, 58-68. 1997.

Metz V, Raanan H, Pieper H, Bosbach D, Ganor J. “Towards the establishment of a reliable
proxy for the reactive surface area of smectite.” Geochimica et Cosmochimica Acta, 69, 10,
2581-2591. 2005.

Mol G, Vriend SP, van Gaans PFM. “Feldspar weathering as the key to understanding soil
acidification monitoring data; a study of acid sandy soils in the Netherlands.”Chemical Geology,
202, 417- 441, 2003.

Oelkers, EH.“General kinetic description of multioxide silicate mineral and glass
dissolution.”GeochimicaetCosmochimicaActa, 65, 21, 3703-3719. 2001.

Parsons I, editor.Feldspars and their Reactions. Springer Verlag. 1994.

Parsons I. “Role of microtexture in Isotopic Exchange and Weathering in Alkali Feldspars.”
Mineralogical Magazine, 58A, 690. 1994.

Parsons I, Thompson P, Lee MR, Cayzer N. “Alkali feldspar microtextures as provenance
indicators in siliciclastic rocks and their role in feldspar dissolution during transport and
diagenesis.” Journal of Sedimentary Research, 75, 921-942. 2005.

Pedone A, Gambuzzi E, Malavasi G, Menziani MC. “First-principles simulations of the
27
Al and
17
O solid-state NMR spectra of the CaAl2Si3O10 glass.”Theoretical Chemistry accounts, 131, 3,
1-11. 2012.

Phillips FM, Zreda M, Plummer MA, Elmore D, Clark DH.“Glacial geology and chronology of
Bishop Creek and vicinity, Eastern Sierra Nevada, California.”GSA Bulletin, 121.7-121.8, 1013-
1033. 2009.

Primas H. “A new method for analyzing spectra in high resolution NMR spectroscopy.”
Spectrochimica Acta, 14, 17-23. 1959.

Reuss JO. “Implications of the calcium-aluminum exchange system for the effect of acid
precipitation on soils.” Journal of Environmental Quality, 12, 4, 591-595. 1983.

51

Reeves D, Rothman DH. “Age dependence of mineral dissolution and precipitation rates.”Global
Biogeochemical Cycles 27, 1-14. 2013.

Richmond GM.“Glaciation of the Rocky Mountains”,The Quaternaryof the United States,ed.
Wright HE Jr., Frey, DG, PrincetonUniversity Press, 217-230. 1965.

Sanders RL, Wasthon NM, Mueller KT. “Measurement of the Reactive Surface Area of Clay
Minerals Using Solid-State NMR Studies of a Probe Molecule.”Journal of Physical Chemistry,
114, 5491-5498. 2010.

Schaller M, Blum JD, Ehlers TA. “Combining cosmogenic nuclides and major elements from
moraine soil profiles to improve weathering rate [models].”Geomorphology, 106, 198-205. 2009.

Schildgen TF. Thesis: “Fire and Ice: The Geomorphic History of Middle Boulder Creek as
Determined by Isotopic Dating Techniques, Colorado Front Range.” Williams College, 114 pgs.
2000.

Siefert KE. “Electron microscopy of etched plagioclase feldspar.” Journal of the American
Ceramic Society, 50, 12, 660-661. 1967.

Sindorf DW, Maciel GE. “Silicon-29 Nuclear Magnetic Resonance Study of Hydroxyl Sites on
Dehydrated Silica Gel Surfaces, using Silylation as a probe.” Journal of Physical Chemistry, 87,
5516-5521. 1983.

Spokas JJ. “Nuclear Magnetic Relaxation in Aluminum.” Thesis, Doctor of Philosophy,
University of Illinois Dept. of Physics. 1958.

Taylor A, Blum JD. “Relation between soil age and silicate weathering rates determined from
chemical evolution of a glacial chronosequence.” Geology, 23, 979-982. 1995.

Teng HH, Fenter P, Cheng L, Sturchio NC. “Resolving orthoclase dissolution processes with
atomic force microscopy and X-ray reflectivity.” Geochimica et Cosmochimica Acta, 65, 20,
3459-3474. 2001.

Tsomaia N, Brantley SL, Hamilton JP, Pantano CG, Mueller KT. “NMR evidence for formation
of octahedral and tetrahedral Al and repolymerization of the Si network during dissolution of
aluminosilicate glass and crystal,” American Mineralogist, 88, 54–67. 2003.

Van Breemen N, Mulder J, Driscoll CT. “Acidification and alkalization of soils.”Plant Soil, 75,
283-308. 1983.
52

Van Der Voort P, Vansant EF. “Silylation of the silica surface.”Journal of Liquid
Chromatography and Related Technologies, 19, 17, 2723-2752. 1996.

Van Grinsven JJM, Van Riemsdijk WH. “Evaluation of batch and column techniques to measure
weathering rates in soils.” Geoderma, 52, 41-57. 1992.

Volk T. “Feedbacks between weathering and atmospheric CO2 over the last 100 million years.”
American Journal of Science, 287, 763-779. 1987.

Ward DJ, Anderson RS, Guido ZS, Briner JP. “Numeric modeling of cosmogenic deglaciation
records, Front Range and San Juan Mountains, Colorado.” Journal of Geophysical Research,
114. 2009.

Washton NM, Brantley SL, Mueller KT. “Probing the molecular level control of aluminosilicate
dissolution: a sensitive solid NMR proxy for reactive surface area.”
GeochimicaetCosmochimicaActa, 72, 5949-5961, 2008.

Weber S. Anorthite.http://www.webmineral.com/anorthite.shtml. 2006.

White AF, Brantley SL., ed. Chemical Weathering Rates of Silicate Minerals, Reviews in
Mineralogy, 1995.

White AF, Bullen TD, Schulz MS, Blum AE, Huntington TG, Peters NE. “Differential rates of
feldspar weathering in granitic regoliths.” Geochimica et Cosmochimica Acta, 65, 847-869.
2001.

White AF, Brantley SL. “The effect of time on weathering rates of silicate minerals: why do
weathering rates differ in the laboratory and field?” Chemical Geology, 202, 479-506. 2003.

White AF, Schulz MS, Vivit DV, Blum AE, Stonestrom DA, Anderson SP. “Chemical
weathering of a marine terrace chronosequence, Santa Cruz, California I: Interpreting rates and
controls based on soil concentration-depth profiles.” GeochimicaetCosmochimicaActa, 72, 36-
68. 2008.

Wilson MJ, McHardy WJ. “Experimental etching of a microcline perthite and implications
regarding natural weathering.” Journal of Microscopy, 120, 291-302. 1980.
A1

Appendix A: NMR acquisition parameters for Bruker 750 MHz spectrometer (set, not
calculated)

Experiment: 1-pulse MAS NMR
MAS spin rate: 15 kHz
Number of scans per experiment: 11,520
Recycle Delay: 2 s
Acquisition time: 0.08197 s
Excitation Pulse: 4.5 μs
Power level for excitation pulse: 100 W
Receiver Gain: 16
Dwell time: 2.5 μs
Pre-scan delay: 6.5 s
X-channel frequency: -73.51 MHz

B1

Appendix B: Solid-state NMR spectroscopy and Magic Angle Spinning

Nuclear Interactions and solid vs. solution
Internal energy of a nucleus is expressed by its total Hamiltonian, the sum of several
component functions, describing various interactions between the nucleus and its local
environment. In an NMR experiment, these include the interaction with a strong applied
magnetic field (Zeeman interaction), perturbation with a pulsing radiofrequency field (RF
interaction), the shielding effect of local electron clouds (chemical shielding), and interactions
with the dipoles created by adjacent nuclei (dipole interaction) (Breene 1955, Spokas 1958,
Primas 1959, Choh and Stager 1969). Chemical shielding is the interaction that gave NMR its
utility when the method was first being developed (MacKenzie and Smith 2002). Because of
local fields in electrons induced by the applied magnetic field, the resonance frequency—and
thereby free induction decay—of a nucleus is modified by this interaction. As such, the position
of a nuclear peak on a frequency spectrum shifts as a consequence of this shielding. This is
termed the “chemical shift” (Lambert and Mazzola 2004). Because of this interaction, NMR
spectroscopy observes not merely the nuclei being measured, but the distributions of different
chemical environments which those nuclei occupy.
The Hamiltonian for the chemical shift can be expressed as
H
CS
= γ
I
∙h∙I∙σ∙B
o
where γ
I
is the gyromagnetic ratio of the nucleus, h is Planck’s constant, I is the nuclear spin
angular momentum, and B
o
is the applied magnetic field (MacKenzie and Smith 2002). σ is an
orientation-dependent tensor that can be broken into principal components σ
XX
, σ
YY
, and σ
ZZ
. By
convention, the z-axis in this reference frame is chosen such that σ will have no effect on H
CS

when the z-axis is aligned with B
o
. Because this will virtually always not be the case for most
B2

grains in a powder, σ accounts for a large component of the peak-broadening that is observed in
solids. In solution, molecules—and their constituent nuclei—undergo rapid translation and
tumbling, which averages out the σ
ZZ
component of the tensor for each nucleus. This isotropic
averaging of orientation results in high resolution spectra. By contrast, in a crystalline solid, the
nuclear orientations are static and distinct for each grain. Shift tensor anisotropies in the
varyingly oriented grains spread out the nuclear resonances in the frequency domain, resulting in
broad peaks (figure B1).
In addition to the chemical shielding interaction, other nuclear interactions also have
similar anisotropic tensors. The Knight Shift—caused by electrons in a conducting solid—and
the quadrupole shift—due to the spin properties of quadrupolar nuclei—are two prominent ones
(MacKenzie and Smith 2002). However, in most cases the overwhelming effect of peak-
broadening in solids is caused by the H
CS
.

Overcoming chemical shift anisotropy via Magic Angle Spinning
Spinning a sample makes H
CS
time-dependent, and a modified form of it can be broken
into static and time-dependent constituents. The part that is apparently static to an outside frame
of reference has an anisotropic component like the initial form of H
CS
.
However, the new anisotropic part contains the term ½(3cos
2
θ – 1), where θ is the angle
at which the sample is rotated relative to B
o
. This is the second-order Legendre polynomial. At
θ = 54.74°, this term is zero, and therefore the anisotropic part of the Hamiltonian goes to zero.
This significantly reduces the broadening effects mentioned above, so spinning solid samples at
this angle has become a common practice in solid-state NMR spectroscopy. The method has
B3

been termed “Magic Angle Spinning” or “Magic Angle Rotation” (Andrew and Newing 1958,
Andrew et al. 1958, Doskockilova and Schneider 1970).

Figure B1. Graphic representation of anisotropic effects in a crystalline sample. The left portion
of the figure depicts a solution-state NMR spectrum, with a sampling of the individual nuclear
spins contributing to it. Due to random tumbling and translation of molecules in a fluid, the
anisotropic part of the chemical shielding Hamiltonian is averaged out for each nucleus, and
therefore the chemical shift tensor does not significantly alter nuclear resonance. By contrast,
the nuclear spins comprising a solid-state spectrum, to the right, are coordinated in the lattices of
grains in many different—static—orientations. Thus, the anisotropic tensor in the chemical shift
Hamiltonian effects each nuclear resonance slightly differently. The result is spectral
broadening, as shown.

B4

Spinning Sidebands
As mentioned, the Hamiltonian for a spinning sample is split into static and time-
dependent parts. The time-dependent part of the Hamiltonian is not a concern if the spinning
speed is sufficiently high, but at lower spin rates there is a modulation effect that is manifested
on the spectrum as a series of bands. They are separated from the main peak by a length on the
spectrum associated with the rotor period, e.g., if the sample rotor was spinning at 10 kHz,
sidebands would appear as smaller peaks ± 10kHz from the main peak. At very low speeds,
three or four sets of sidebands may develop, each an additional rotor period away (figure B2).

Figure B2. Graphic representation of spinning sidebands. At the center of the figure is the main
peak, flanked with two sets of sidebands.

C1

Appendix C: Determination of BET sample input precision threshold
NMR experiments on the instruments used for this study only require between 50 and
100 mg of sample, limited by the capacity of the sample rotor. However, surface area
measurements done on the ASAP 2010 physisorption instrument require significantly more
sample. In order to attain accurate BET specific surface area (SSA) results, sufficient material is
necessary. Measurements on samples with too little total surface area are unreliable. To
determine the minimum necessary sample input amount for the ASAP 2010, 55 sets of 2 kinds of
aluminosilicate, each with known SSA, were run through the instrument, varying input mass
over two orders of magnitude. SSA is plotted as a function of measured input mass in figure C1.
Values are reported in table C1.
Measured SSA was expressed as a function of known total surface area (SA) for each
sample. 54 least squares linear regressions were calculated for subsets of these data, first with all
55 samples, then excluding the lowest total SA sample, and so on in this fashion until only two
samples remained. It is presumed that when a sufficient number of high-total-SA samples are
used, the least squares line for that subset of samples has a slope of zero.
After running aluminosilicate reference materials it was found that a minimum of
approximately 0.9 m
2
total surface area is required for 5% precision (dotted line in figure C2),
determined by the data analysis shown in figure C3. Running a linear regression on only the 19
samples with the highest total SA values yields a slope of zero, to 95% confidence (figure C3).
Including any lower-SA samples yielded a nonzero slope and more unconstrained confidence
intervals. This implies that the lowest-SA sample in the above-mentioned subset of 19 (with a
total SA of ~0.9 m
2
) represents the minimum input total surface area allowable when running
aluminosilicate materials on the ASAP 2010. For the feldspars in this study, that value of 0.9 m
2

C2

ended up corresponding to between 1 and 2 g of sample. There appears to be only a minor
relationship between age and SSAs of the samples (table 1, main text).

Figure C1. BET specific surface area (SSA) of samples vs. input sample mass, for the
Micromeritics silica alumina reference material (upper plot—blue squares), and anorthosite from
the San Gabriel Mountains (lower plot—purple squares).

C3

Table C1. Input mass, measured SSA, and calculated SA for the two different reference materials.
Anorthosite

Silica Alumina
Mass (g) SSA (m
2
/g) Total SA (m
2
)

Mass (g) SSA (m
2
/g) Total SA (m
2
)
0.01 28.42 0.31

0.03 6.39 0.16
0.02 16.48 0.38

0.05 3.24 0.16
0.02 17.00 0.39

0.06 4.17 0.25
0.04 8.17 0.31

0.19 0.65 0.12
0.04 14.54 0.58

0.21 1.27 0.26
0.13 9.00 1.14

0.21 1.61 0.33
0.31 7.80 2.40

0.25 0.58 0.15
0.70 7.09 4.96

0.38 0.99 0.38
0.72 6.61 4.76

0.38 1.01 0.38
1.54 6.77 10.42

0.39 0.51 0.20
1.86 6.99 12.99

0.43 0.95 0.41
1.86 6.99 12.99

0.44 0.92 0.40
1.97 6.62 13.06

0.55 0.82 0.45
1.97 6.60 13.03

0.90 0.69 0.62
2.26 7.26 16.45

1.06 0.70 0.74
3.40 6.96 23.63

1.23 0.65 0.79
4.14 7.59 31.46

1.33 0.59 0.79
7.60 6.84 51.99

1.34 0.63 0.84

1.34 0.64 0.85

1.35 0.67 0.91

1.35 0.60 0.80

1.41 0.64 0.90

1.49 0.59 0.87

1.68 0.61 1.02

2.24 0.56 1.24

3.18 0.57 1.81

C4

Figure C2. BET specific surface area (SSA) of aluminosilicate samples vs. actual total surface
area (SA) of each sample. The solid gray line represents the known SSA of these materials. The
vertical dashed line at 0.9 m
2
total SA denotes the approximate minimum amount of sample
required for a reliable result. Measured SSA is normalized to known SSA, consequently the
vertical axis is unitless.

C5

Figure C3. Slopes of least squares linear regressions run on subsets of the 55 aluminosilicate
reference samples. The samples were ordered from lowest to highest total SA. Every sample
was included in the first regression. For the next regression, the sample with the lowest total SA
was omitted. This sequence was continued until there were not enough samples to do a valid
regression, i.e. the residuals of the regression were not normally distributed. This happened with
approximately 15 samples remaining, i.e. 40 samples omitted. Slopes for each regression are
plotted as a solid black line, and confidence intervals within standard error ε are shown as dotted
blue lines.

C6

In addition to the regressions, an assessment of the instrument blank was done by plotting
measured total surface area as a function of input sample mass, separately for the two reference
materials, silica alumina and anorthosite (figure C4). These plots suggest that there is a blank of
approximately 0.15 m
2
, but it should be noted that the data do not have linear behavior at very
low surface areas (figure C2). Also, experimental runs were done with empty sample tubes,
which yielded SSA values of less than 0.1 m
2
/g, suggesting a lower blank than indicated by the
plots in figure C4.

Figure C4. Measured total surface area plotted as a function of sample mass, for two different
types of reference material, silica alumina (left) and anorthosite (right). Vertical axis intercepts
for silica alumina and anorthosite are, respectively, 0.14 and 0.18 m
2
.

Abstract (if available)

Abstract A method developed to quantify hydroxyl (OH) sites on silicate minerals is applied for the first time to feldspar grains from soils in two glacial chronosequences and a saprolite profile. Illuminating the interplay between accumulation of surface OH sites and the total time primary silicate minerals have spent in the weathering environment could be a step towards mechanistically understanding the reasons mineral dissolution rates are observed to decrease with time. Soil was sampled from discrete deposits—with ages ranging from 10 ka to 200 ka—along glacial chronosequences in the Sierra Nevada and the Colorado Front Range. Feldspar grains were picked from these soils and treated with (3,3,3trifluoropropyl) dimethylchlorosilane (TFS). This molecule selectively binds to Q³ OH sites on silica and alumina tetrahedra—sites wherein one tetrahedral anion forms a hydroxyl at the surface during dissolution and the other three remain coordinated with the crystal. NMR spectroscopy is a sensitive, non-destructive analytical method, and here it is used to quantify ¹⁹F in TFS on TFS-treated feldspar samples. Q³ OH concentrations were inferred from these data, and normalized to specific surface area (the physical surface area per unit mass). Treated samples were also dissolved and run through an MP-OES for elemental analysis, to determine whether or not time-dependent mineralogical evolution was partly responsible for the changing hydroxyl concentrations. NMR experimental results reveal a power-law decrease in Q³ OH site accumulation rate with time. Elemental analysis by MP-OES suggests that mineralogical differences among samples may be a notable complicating factor in Front Range samples, but far less so in the Bishop Creek samples. This study confirms that the NMR procedure is sensitive enough to quantify very low concentrations of TFS condensed to OH sites, as found on feldspar surfaces. Continued investigation using the TFS treatment and NMR method will hopefully shed light on the molecular-scale processes that drive interface-limited systems.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Environmental controls on alkalinity generation from mineral dissolution: from the mineral surface to the global ocean

PDF

Germanium and silicon isotope geochemistry in terrestrial and marine low-temperature environments

Asset Metadata

Creator Kleinsasser, Eric A. (author)

Core Title Exploring temporal changes in surface species on weathered feldspar mineral surfaces using solid-state NMR spectroscopy

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geological Sciences

Publication Date 10/29/2014

Defense Date 10/29/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag ¹⁹F,¹⁹F hydroxyl,3,3 trifluoropropyl dimethylchlorosilane,Bishop Creek,dissolution rate,feldspar,glacial chronosequence,kinetics,mineral surface area,NMR,nuclear magnetic resonance,OAI-PMH Harvest,Rocky Mountains,silicate,spectroscopy,transition state theory,weathering

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor West, A. Joshua (committee chair), Hammond, Douglas E. (committee member), LaRowe, Douglas (committee member)

Creator Email eakleinsasser@gmail.com,kleinsas@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-510823

Unique identifier UC11297704

Identifier etd-Kleinsasse-3041.pdf (filename),usctheses-c3-510823 (legacy record id)

Legacy Identifier etd-Kleinsasse-3041.pdf

Dmrecord 510823

Document Type Thesis

Format application/pdf (imt)

Rights Kleinsasser, Eric A.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA