A TIME-RESOLVED SYNCHROTRON X-RAY DIFFRACTION STUDY OF THE IN SITU, HYDROTHERMAL SYNTHESIS OF GOETHITE FROM 2-‐LINE FERRIHYDRITE
Abstract
The hydrothermal transformation of 2-‐line ferrihydrite to goethite at pH 13.6
was studied at 80, 90, and 100 °C using time-‐resolved, angle-‐dispersive X-‐ray diffraction at the Advanced Photon Source, Argonne National Lab. We found that these in situ experiments could be successfully performed by injecting freshly made ferrihydrite gels
into polyimide capillaries that were then sealed with fast curing epoxy. The integrity of these reaction cells held steady when heated to 100 °C for up to 6 hr, and the low
background scattering from the polyimide allowed for an extremely high sensitivity for the detection of weak X-‐ray diffraction peaks from the first goethite nanocrystals to
nucleate.
Rietveld analysis of the time-‐resolved series of X-‐ray diffraction patterns as
goethite crystallized enabled a high-‐resolution extraction of crystallographic and
kinetic data. Particle sizes for goethite increased with time at similar rates for all three temperatures, from particle diameters of ≈135 Å for the first crystals detected to ≈290 Å after 6 hr. With increasing particle size, unit-‐cell volume decreased, primarily as a result of a contraction along the c-‐axis, the direction of closest-‐packing (S.G. Pnma). At 100 °C c decreased from 4.631 ± 0.001 to 4.613 ± 0.001 Å. We found no evidence for
high vacancy concentrations in incipient goethite nanocrystals, in contrast to our prior
study of hematite crystallization (Peterson, et al. 2015).
Two different kinetic models were used to calculate the activation energy of the
reaction: a pseudo first-‐order model based on the Johnson-‐Mehl-‐Avrami-‐Kolmogorov (JMAK) equation, and a linear Arrhenius analysis based on only the initial reaction rates. Both approaches yielded excellent fits to the data (R2 > 0.95), but the calculated activation energies using the JMAK and linear models were 72.74 and 100.1 kJ/mol, respectively. This disparity reveals that the transformation of ferrihydrite to goethite was most temperature-‐dependent during the initial stages of crystallization.
TABLE OF CONTENTS
List of tables……………………………………………………………………………………………..vi
List of figures…………………………………………………………………………………………..vii
Acknowledgements……………………………………………………………………………………..x
Chapter 1. INTRODUCTION………………………………………………………………………..1
Hydrothermal Experiments and X-‐ray Diffraction……………………………..1
Rationale for Present Study…………………………………………………………………………..6
Chapter 2. EXPERIMENTAL METHODS…………………………………………………………………13
Sample Preparation………………………………………………………………………………………13
Construction of Environmental Cell………………………………………………………….13
Synchrotron X-‐ray Diffraction…………………………………………………………………….16 Structure Refinement……………………………………………………………………………………18
Particle Size Determination………………………………………………………………………..19
Kinetic Modeling……………………………………………………………………………………………19
Chapter 3. RESULTS…………………………………………………………………………………………………..28 Changes in Particle Size……………………………………………………………………………….28
Crystal Structure Changes During Goethite Crystallization…………………28
Kinetics of Goethite Crystallization…………………………………………………………..29
Chapter 4. DISCUSSION……………………………………………………………………………………………..40
Kinetic Modeling……………………………………………………………………………………………40
Crystallographic Parameters………………………………………………………………………41
Chapter 5. CONCLUSION…………………………………………………………………………………………..42
Introduction
Hydrothermal Experiments and X-‐ray Diffraction
Crystallographers attempted to study the effect of temperature and pressure on
crystal structure almost immediately after X-‐ray diffraction (XRD) was first applied to the determination of atomic positions in minerals. For example, within a year of solving the structure of sodium chloride, W.H. Bragg conducted experiments on the thermal expansion of halite (Bragg, 1914). However, Bragg’s method allowed for only a single crystal to be studied in a few crystallographic orientations, severely limiting the statistical rigor of the data collected. In contrast, Albert Hull (1917) recognized that X-ray diffraction of powders enabled the sampling of a large number of crystals in many orientations. Powder XRD was advantageous because it required only a small amount of material (as little as 0.005 g) and extreme purity of monophasic powders was not
necessary, allowing whole rocks or soils to be studied in bulk (Hull, 1917).
Thermal diffraction of dry powders became conventional by the 1920s, but X-‐ray
diffraction of solid-‐fluid mixtures proved a continuing challenge. Hydrothermal experiments required the sample to be completely contained during heating in order to prevent water vapor loss. However, in early approaches to XRD, “angle-‐dispersive” powder diffraction (ADPD) was the only viable methodology. As with the original experiments by W.L. and W.H. Bragg, in ADPD, the diffraction angle θ is varied as a monochromatic beam of X-‐rays (of wavelength λ) samples a range of lattice planes of interplanar spacing dhkl, and constructive interference of the diffracted beams is
expressed by Bragg’s law: λ = 2dhkl sin θ (Clark, 1989). In the traditional Bragg-Brentano geometry, the X-‐ray source is spatially fixed but the sample and the detector are rotated, greatly limiting the varieties of sample cell that are compatible with
hydrothermal experiments.
Energy Dispersive Approaches. Giessen and Gordon (1968) innovated a new
technique that did not require either sample or detector rotation: Energy Dispersive Powder Diffraction (EDPD). Not only did EDPD allow for the sample to remain still, it also cut down on the capture time. Because the incident beam is polychromatic, λ rather than θ is the variable in Bragg’s law, and intensities from Bragg peaks for all d-spacings are collected simultaneously. Using a Siemens X-‐ray source with an Fe anode at 8 ma and 45 kV and a lithium-‐drifted Si semiconductor detector, Giessen and Gordon
(1968) obtained indexable diffraction patterns within 15 seconds. Even today a standard diffractometer with Bragg-‐Brentano geometry requires at least 10 minutes to
collect an indexable pattern.
The advent of EDPD opened up a wealth of directions to pursue with respect to
in situ XRD experiments for hydrothermal systems. In the 1980s, Paul Barnes and colleagues at the Daresbury Synchrotron Radiation Source (SRS) took advantage of the high intensity of synchrotron X-‐ray radiation, which allowed for short capture times and data collection from encapsulated samples. The high penetration of the synchrotron X-‐rays enabled samples to be encased by materials, such as steel, that would normally absorb most of the incident X-‐ray beam from standard diffractometers (Clark, 1988). Barnes pioneered his method at Station 9.7 of the 5 Tesla Wiggler
magnet beamline of SRS throughout the late 1980s and 1990s.
By 1991, Barnes et al. had developed three different environmental cell designs, 1) a flat plate in reflection; 2) a thin plate in transmission; and 3) a reaction tube in transmission. Each cell was developed for certain types of experiments and exhibited
both advantages and disadvantages.
- The flat plate in reflection (Figure 1a) is best used for experiments where the sample does not permit sufficient X-‐ray penetration for the transmission cells, or if the reactions studied are in a controlled gas The flat plate in reflection can be used with or without
heating.
- The thin plate in transmission (Figure 1b) is used with a heating plate that has a central aperture that allows an exit path for the diffracted
radiation from the sample.
- The reaction tube in transmission (Figure 1c) is the most versatile cell and can be used for studying bulk reactions under a variety of both solid state and hydrothermal The original reaction tube was a sealable sample container, similar in size to a laboratory test tube, surrounded by a block containing coils for circulating fluid allowing the
heating or cooling of the system. Barnes was able to collect EDPD patterns for a variety of processes using these cells, including the hydration of cements, the synthesis of ceramics, and the crystallization of
ferromagnetic metallic glasses (reviewed in Barnes, 1991).
In 1995, Evans et al. developed a more advanced version of Barnes’ reaction tube
in transmission (Figure 2). The sample holder in this cell was a stainless-‐steel tube closed at one end. The wall of the tube was machined down to a thickness of 0.7 mm over a 29 mm height range to allow the incident and diffracted beams to penetrate the sample. In order to prevent contact between reagents and the metal cell walls, a Teflon liner and lid were created to fit within the steel tube. This cell was attached to a Parr Instruments bomb cover fitted with a flat Teflon gasket. Connected to the bomb cover was a gauge block assembly, with both a pressure relief valve (set to 450 psi) and an Inconel rupture disk rated to 1002 psi. A blast containment chamber connected to the burst disk and surrounded the bomb itself. A digital pressure gauge was also fitted to the gauge block via a pressure transducer. This design allowed for a working pressure and temperature of 400 psi and 230°C respectively (Evans et al., 1995). To this day, many studies continue to use this experimental cell design (e.g. Shaw et al. 2005; Sun
and Ren, 2013).
EDPD, however, offers serious drawbacks. The main disadvantage of EDPD is its
low peak resolution compared to that of ADPD. EDPD’s momentum resolution is an order of magnitude worse than that of ADPD. This makes EDPD unsuitable for studying low-‐symmetry materials, which typically are characterized by overlapping XRD peaks. More significantly, it is not possible to obtain full crystal structure refinements from EDPD patterns because of the complex relationship between X-‐ray energy and sample absorption. Thus, EDPD is most appropriate for crystals that have an established crystal structure. The detector and counting chain also limit EDPD’s count rate to about
5 x 104 counts/s (Clark, 1988). Comparisons of EDPD and ADPD spectra can be seen in Figures 3a and 3b.
Constant Wavelength Techniques. For the reason outlined above, ADPD
remains the best option for accurate refinement of crystal structures. Throughout the mid-‐1990s, multiple cells were developed for compatibility with in situ ADPD. In 1994, Notten et al. innovated a high-‐pressure cell to investigate hydride formation reactions (Figure 4). The cell consisted of a stainless-‐steel cylindrical housing such that 180° of the cylinder wall was left open and, in places, a curved X-‐ray transparent Be window with a thickness of 0.3 mm was placed. This curved window allowed the cell to rotate while still enabling the X-‐rays to reach the sample. This experimental cell was used to determine the absorption and desorption kinetics of hydrogen to LaNi5 (Notten et al.,
1994).
All of the EDPD and the Notten ADPD experimental cells are complicated to
assemble and require expensive parts. In 1996, Poul Norby, who was a postdoctoral researcher at Brookhaven National Laboratory, developed an environmental cell that
was easy to put together, versatile over a range of experiments and compatible with ADPD for powder-‐fluid-‐gas mixtures. He contained the sample assemblages in 0.5-‐0.7 mm quartz capillaries, mounted using a ferrule in a T-‐connector, which, in turn was mounted on a goniometer head (Figure 5). The capillaries were sealed on one end and gas was pumped into the other end to apply a pressure sufficient to contain the sample. A hot air gun heated the sample externally. Norby used two different air guns depending on the desired temperatures; for temperatures up to 250°C, he employed a heating coil in an insulated glass tube, but for higher temperature experiments (up to 900°C) he exploited a commercial Enraf Nonius heater. The full schematic with gas flow and heater can be seen in Figure 6. Norby performed all experiments at the
National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory (BNL) on
the Chemistry Beamline, X7B.
Rationale for the Present Study
For this study, I sought to accomplish two goals. First, I attempted to modify the
Norby-‐type environmental cell to achieve the following targets:
- A decrease in the X-‐ray background contribution from the capsule relative to the quartz glass used in standard capillary containers;
- Minimization of the time between sample loading and X-‐ray data collection; and
- Compatibility with fluid-‐powder mixtures at temperatures at least as high
as 100 °C
Secondly, I aimed to test this modified cell through the hydrothermal synthesis of
goethite from 2-‐line ferrihydrite.
Goethite is extremely common in wet and oxidizing surface environments. It can
form as a weathering product of Fe-‐rich minerals, or as a primary precipitate from Fe-rich solutions at low or high pH (Cornell and Schwertmann, 2003). The high sorption capacity of goethite can control the mobility of trace metals, ions, and organic compounds (Gimenez et al., 2007; Tipping, 1981; Torrent et al., 1992). The sorption properties of goethite can vary depending on multiple factors, such as particle size and crystallinity. Therefore, a detailed understanding of the relationship among the kinetics of goethite formation, crystallite size, and atomic structure is needed in order to fully understand how goethite affects trace element migration on the Earth’s surface (Shaw et al., 2005). Goethite is used industrially as a yellow pigment (Cornell and
Schwertmann, 2003). Additionally goethite is produced as an intermediate in many iron oxide preparations, and a quantification of the kinetics and activation energies associated with goethite formation is industrially useful (Sharrock and Bodnar, 1985)
and (Sesigur et al., 1996).
Goethite crystallization has been the subject of numerous studies (Schwertmann
and Murad, 1983; Nagano et al., 1994; and Shaw et al., 2005). However, only Shaw et al. (2005) investigated hydrothermal goethite crystallization in real time. They employed EDPD in the hydrothermal autoclave developed by Evans et al. (1994). Although they succeeded in obtaining rate constants and activation energies for goethite crystallization under different solution conditions (pH 10.7 and 13.7), their data did not offer sufficient resolution to determine full structure refinements as a function of time and temperature. In the present study, I used my modified Norby cell to perform a time-‐resolved study of goethite crystallization at 80, 90, and 100 °C in aqueous solutions at pH at 13.6. These results then allowed a calculation of the rate constants and activation energies of goethite crystallization with high accuracy. Because hematite and goethite are both found naturally on the Earth’s surface, even though hematite is considered the stable phase in most environments, it was important to calculate activation energy to better understand why both minerals are present. This study will show that the activation energy for goethite is significantly lower than hematite for certain environments, leading us to believe the goethite-‐hematite relationship is kinetically driven instead of thermodynamically driven.
Figure 1. Sample stages developed by Barnes et al. (1991) at Station 9.7 of the
Daresbury Synchrotron Radiation Source (SRS). A) Flat plate in reflection; B) Thin plate in transmission; C) Reaction Tube in Transmission. Source: Barnes et al. (1991)
Figure 2. A schematic representation of the hydrothermal autoclave developed by Evans et al. (1994). Source: Evans et al. (1994)
dimensional representation of XRD patterns collected by ADPD of the crystallization of goethite from 2-‐line ferrihydrite. Note the difference in the number of peaks visible between the two figures.
Figure 4. A schematic of the high pressure X-‐ray diffraction cell developed by Notten et al. (1994). Be window allowed X-‐rays to penetrate into the cell. Source: Notten et al. (1994).
Figure 5. Close up view of the cell developed by Norby et al. (1996). A) T-‐piece; B)
Capillary; C) Tubing applied pressure or gas-‐flow; D) Goniometer head. Source: Norby et al. (1996)