Bazilevskaya PhD Dissertation 2 December2009

8/18/2019 Bazilevskaya PhD Dissertation 2 December2009

1/187

The Pennsylvania State University

The Graduate School

College of Agricultural Sciences

IRON AND ALUMINUM HYDROXIDE NANOPARTICLES IN THE

ENVIRONMENT: FROM NANO-SCALE TO THE FIELD PROCESSES

A Dissertation in

Soil Science and Biogeochemistry

by

Ekaterina Bazilevskaya

2009 Ekaterina Bazilevskaya

Submitted in Partial Fulfillment

of the Requirements

for the Degree of

Doctor of Philosophy

December 2009


2/187

ii

The dissertation of Ekaterina Bazilevskaya was reviewed and approved* by the

following:

Carmen Enid MartnezAssociate Professor of Environmental and Soil ChemistryDissertation AdvisorChair of Committee

Douglas Archibald

Research Associate in Agricultural Analytical Chemistry

Edward CiolkoszProfessor Emeritus of Soil Genesis and Morphology

James KubickiProfessor of Geosciences

Kwadwo Osseo-AsareDistinguished Professor of Metallurgy and Energy and Geo-environmentalEngineering

David SylviaProfessor of Soil MicrobiologyHead of the Department of Crop and Soil Sciences

*Signatures are on file in the Graduate School


3/187

iii

ABSTRACT

The objective of this doctoral research was to increase scientific understanding of

the behavior of Fe and Al hydroxide nanoparticles in soils. These particles are of great

environmental importance due to their ability to retain and transport nutrients and

contaminants. Three studies were undertaken at different scales, which are documented in

three manuscripts included in this dissertation

The first study examined the rate constants for goethite (α-FeOOH) crystallization

from nano-particulate Fe hydroxide suspensions in the absence (0% Al) and presence

(2% Al) of aluminum. One of the merits of this study was the application of a

multivariate curve resolution analysis (MCR) of infrared spectra to environmentally

important mixed Fe-Al hydroxide colloids in order to quantify goethite content in poorly-

crystalline mixtures. Obtained rate constants were found to be equal to (7.64±0.67)×10-7

s-1 for 0% Al and (4.5±0.21) ×10-7 s-1 for 2% Al hydroxides. Dissolution-precipitation

mechanism was dominant in the process of goethite transformation to ferrihydrite.

Further growth of goethite crystals took place either by aggregation mechanism to form

polycrystalline agglomerates or alternatively by Oswald ripening to form large single

crystals. The presence of aqueous Al species „poisoned“ goethite’s surface by disrupting

the formation of hydrogen bonds thus increasing the number of non-stoichiometric

hydroxyls.

The second study addressed changes of mineral composition in mixed Fe-Al

hydroxide nanoparticles as a function of Al-substitution and reaction time. It was found

that low Al concentrations (2-8 mol. %) lead to formation of moderately crystalline Al-


4/187

iv

goethite upon ageing, while at medium Al concentrations (10-20%) colloidal suspensions

remained stable for the duration of the whole experiment (54 days), goethite formation

was completely retarded, and less crystalline intermediate structure were formed. At 25%

Al substitution, gibbsite Al(OH)3 microcrystalline structures appeared within the first

days of experiment. In addition, to understand the mechanism of Al substitution in

goethite, we explored the most energetically favorable arrangement of Al atoms within

goethite by ab initio periodic density functional theory (DFT) calculations. These

calculations showed that Al may form Al-O-Al clusters as opposed to evenly distributed

isolated Al atoms (Al-O-Fe) in goethite structure.

The third study was conducted to investigate the relative importance of Fe and Al

hydroxide nanoparticles in migration and accumulation of these elements in soils. I

approached this goal by studying soil water and coatings on mineral grains from

Spodosol soil, which is characterized by intensive leaching of Fe, Al, and organic matter

(OM) and their accumulation in Spodosol profile. While Fe, Al and Si were mostly

transported as inorganic colloids, Al also showed close association with organic matter.

Fe, Al, and Si have the highest mobility in organic-rich A and Bh horizons of the

spodosol profile, which suggests that the presence of organic matter facilitates transport

of these elements by stabilizing the inorganic colloids. The two major mechanisms of

immobilization of Fe, Al, Si and OM are (1) polymerization of metal-OM complexes and

(2) surface charge neutralization of OM-inorganic colloidal aggregates. Both of these

processes presumably occur when the (Fe+Al) to C ratios in the colloidal fraction

increase in Bh horizon.


5/187

v

The presented research findings contribute to our understanding of the

fundamental properties of mixed Fe- and Al-hydroxide nanoparticles and also help to

predict the interactions and environmental fate of natural nanocolloids in soil profile.


6/187

vi

TABLE OF CONTENTS

LIST OF FIGURES ................................................................................................................. ix

LIST OF TABLES...................................................................................................................xiii

ACKNOWLEDGMENTS ....................................................................................................... xiv

Chapter 1 INTRODUCTION...................................................................................................1

1.1. Background...............................................................................................................1 1.2. Dissertation structure ................................................................................................2 References........................................................................................................................3

Chapter 2 NANO-GOETHITE CRYSTALLIZATION IN THE PRESENCE OF LOWALUMINUM CONCENTRATIONS UNDER ENVIRONMENTALLY

RELEVANT CONDITIONS...........................................................................................5

Abstract............................................................................................................................5 2.1. Introduction...............................................................................................................6 2.2. Experimental Section................................................................................................8

2.2.1. Synthesis of Fe-hydroxide and Al-doped Fe-hydroxide nanoparticles ..........8 2.2.2. Synchrotron-based X-ray diffraction..............................................................10 2.2.3. ATR-FTIR spectroscopy .................................................................................10

2.2.3.1. ATR-FTIR data collection and Gaussian band analyses.....................10 2.2.3.2. Multivariate curve resolution (MCR) analysis of ATR-FTIR

spectra.......................................................................................................12 2.2.4. Dynamic light scattering (DLS) .....................................................................13

2.2.5. Transmission electron microscopy (TEM) .....................................................13 2.3. Results.......................................................................................................................14 2.3.1. XRD data ........................................................................................................14 2.3.2. Infrared spectra..............................................................................................15

2.3.2.1. OH-bending vibrations........................................................................15 2.3.2.2. OH-stretching vibrations: Gaussian deconvolution.............................15 2.3.2.3. MCR analysis of IR O-H stretching region.........................................18

2.4. Discussion.................................................................................................................20 2.4.1. Mechanisms for nano-goethite crystallization ...............................................20 2.4.2. Rate constants for ferrihydrite-goethite transformation ................................23

2.5. Conclusions...............................................................................................................25 Acknowledgments............................................................................................................26

References........................................................................................................................26 Figures..............................................................................................................................32 Chapter 3 ATR-FTIR, XRD AND PERIODIC-DENSITY-FUNCTIONAL

THEORY STUDIES OF MINERAL PHASES IN CO-PRECIPITATED FEAND AL HYDROXIDE..........................................................................................46

Abstract............................................................................................................................46 3.1. Introduction...............................................................................................................47 3.2. Methods.....................................................................................................................50

3.2.1. Experimental ..................................................................................................50


7/187

vii

3.2.1.1. Synthesis and aging of Fe-hydroxide and Al-doped Fe-hydroxidenanoparticles.............................................................................................50

3.2.1.2. Characterization of Fe-hydroxide and Al-doped Fe-hydroxidenanoparticles.............................................................................................51

3.2.2. Periodic density functional theory calculations of bulk Al-goethite modelstructures..........................................................................................................52

3.3 Results and discussion ...............................................................................................54 3.3.1. Phase transformation in Fe-Al hydroxide nano-particle suspensions ...........54 3.3.2. Arrangement and concentration of Al in Al-goethites: energies and unit

cell parameters from DFT calculations ...........................................................57 3.3.3. Goethite versus Al-goethites: stability considerations...................................60 3.3.4. Relative abundance of gibbsite and diaspore in soils ....................................62

3.4. Conclusions...............................................................................................................64 Acknowledgements..........................................................................................................65 References........................................................................................................................65 Tables...............................................................................................................................69

Figures..............................................................................................................................72 Chapter 4 SPECTROSCOPIC AND MICROSCOPIC INVESTIGATIONS OF THE

COMPOSITION OF SOIL WATER AND ACCUMULATIONS FORMED INSITU IN A SPODOSOL..................................................................................................81

Abstract............................................................................................................................81 4.1. Introduction...............................................................................................................82 4.2. Methods.....................................................................................................................84

4.2.1. Field site description and samples .................................................................84 4.2. 2. Analyses of soil solids ...................................................................................85

4.2.2.1. Soil pH and selective extractions ........................................................85

4.2.2.2. Total Organic Carbon (TOC) and Total Nitrogen...............................86 4.2.2.3. X-ray diffraction (XRD) of clay size fraction (< 2 µm) fromSpodosol horizons.....................................................................................87

4.2.3. Analyses of soil waters ...................................................................................87 4.2.3.1. Soil water extraction............................................................................87 4.2.3.2. Dissolved organic carbon (DOC) and Colloidal organic carbon

(COC) ....................................................................................................... 88 4.2.3.3. Attenuated Total Reflectance Fourier Transform Infrared ATR-

FTIR .........................................................................................................89 4.2.3.4. Scanning electron microscopy - Energy dispersive spectroscopy

(SEM-EDS) ..............................................................................................89 4.2.3.5. Transmission electron microscopy (TEM)..........................................90

4.2.4. In situ coating formation within the Spodosol profile ....................................90 4.2.4.1. Experimental setup..............................................................................90 4.2.4.2. Scanning Electron Microscopy - Energy Dispersive Spectroscopy

(SEM-EDS) ..............................................................................................90 4.2.4.3. Fe-GIXAS (Gracing Incidence X-ray Absorption Spectroscopy).......91

4.3. Results.......................................................................................................................92 4.3.1. Characteristics of Soil Solids .........................................................................92 4.3.2. Soil water: dissolved and colloidal fraction...................................................93

4.3.2.1. Dissolved concentration of Fe, Al, Si and C .......................................94


8/187

viii

4.3.2.2. Concentrations of Fe, Al, Si and C in colloidal fraction of soilwaters........................................................................................................94

4.3.2.3. ATR-FTIR analysis of soil water ........................................................95 4.3.2.4. SEM-EDS analyses of the colloidal fraction (0.45-1.2 µm particles

size) of soil waters....................................................................................98 4.3.2.5. TEM-EDS of the colloidal fraction of soil waters...............................99 4.3.3. Characteristics of in situ coatings formed in the Spodosol profile ................100

4.3.3.1. SEM-EDS of coatings of wafers .........................................................100 4.3.3.2. Fe-EXAFS analysis of quartz wafers .................................................. 102

4.4. Discussion.................................................................................................................103 4.4.1. Transport and accumulation forms of Fe, Al, and Si in Spodosols................103

4.4.1.1. Aluminum............................................................................................103 4.4.1.2. Iron ......................................................................................................104 4.4.1.3. Silica....................................................................................................105

4.4.2. Role of organic matter in podzolization process............................................106 4.4.2.1. Transport forms ...................................................................................106

4.4.2.2. Immobilization ....................................................................................107 4.4.5. Environmental significance of this study........................................................108 4.5. Conclusions...............................................................................................................108 Acknowledgements..........................................................................................................109 References........................................................................................................................109 Tables...............................................................................................................................115 Figures..............................................................................................................................117

Chapter 5 CONCLUSIONS AND FUTURE WORK..............................................................129

APPENDIX A. Composition and morphology of soil colloids (0.45µm to 1.2 µm) bySEM-EDS.........................................................................................................................131

APPENDIX B. SEM-EDS analyses of quartz wafers implanted in the Spodosol soil............143

APPENDIX C. X-ray diffraction (XRD) of Spodosol clay fraction (< 2 µm) ........................164


9/187

ix

LIST OF FIGURES

Figure 2-1. X-ray diffractograms for FeIII-based nanocolloids synthesized with 0 or 2 mol%

incorporated-Al at initial conditions (0 days) and after aging in dilute suspensions for 2, 9,23, and 54 days. Go stands for goethite, numbers in parenthesis indicate crystal face; Fhdenotes ferrihydrite................................................................................................................32

Figure 2-2. Crystallite sizes for goethite as a function of time. Crystallite sizes were estimatedfrom the XRD (110) peak using the Scherrer equation (0%Al, diamonds; 2% Al, squares). 33

Figure 2-3. (A) Crystal structure of goethite. Fe is in octahedral coordination surrounded by O2- and OH- ligands (O, red; H, white). (B) Closer look showing hydrogen bonds, O-H ···O(dashed lines).........................................................................................................................34

Figure 2-4. ATR-FTIR spectra as a function of time for Fe-hydroxide suspensions with 0% Al(upper panel) and 2% Al (lower panel). Multiple lines at a given time reflect both batch andinstrument replicate variability. ............................................................................................. 35

Figure 2-5. (A) Example of Gaussian band analysis for the OH-stretching region (0%Alsuspensions: upper plot, suspension aged for 54 days; bottom plot, initial suspension at 0days). Solid lines are experimental spectra, dotted lines are the Gaussian components, andbroken lines are the sum of the Gaussian components. Stoichiometric OH peak position (B)and full width at half maximum (FWHM) (C) as a function of time for 0%Al and 2% Alsuspensions. ........................................................................................................................... 36

Figure 2-6. MCR analysis of OH-stretching region. (A) Components extracted from the 0% Alexperimental dataset; these two components describe all spectral variability. (B) Scores(relative ratios) for Component 1 (“ferrihydrite-like”) and Component 2 (“goethite-like”) asa function of time. (C) Gaussian band analysis indicates each component (for both 0% and2% Al suspensions) is consistent with combination of stoichiometric hydroxyls, non-

stoichiomeric hydroxyls, and adsorbed water........................................................................37Figure 2-7. (A) Particle sizes as determined by dynamic light scattering for 0%Al (upper panel)

and 2%Al (bottom panel) suspensions. Dark grey symbols show the results for a suspensionmeasured after aging for 0, 2, 9, 23 and 54 days. Light symbols show data obtained by in-situ heating of an initial suspension (0 days) at 50 °C every 2 hours for about 2 days inautomatic mode. (B) Volume-based particle size distributions for 0%Al (upper panel) and2%Al (bottom panel) suspensions aged for 0, 2, 9, 23, and 54 days (indicated by numberswith arrows)...........................................................................................................................38

Figure 2-8. Representative TEM images of iron hydroxide nanoparticles. Images A and B shownon-aged particles (0 days) from 0%Al and 2%Al suspensions, respectively. TEM revealedsmall (


10/187

x

Figure 2-9. TEM images of diluted samples. Images A and B show non-aged particles (0 days)from 0%Al and 2%Al suspensions, respectively; C and D - Al-free and 2%Al suspensionsaged for 54 days, respectively................................................................................................40

Figure 2-10. High-resolution TEM images of 0% Al suspensions aged for 0 days (A) and 54 days(B and C). (A) A needle formed from aggregated particles (0 days). (B) Large goethitecrystal shows single-crystal lattice fringes in agreement with a spot diffraction pattern(insert). (C) Goethite needle consisted of smaller grains with different lattice orientation.Diffraction pattern produced diffraction rings characteristic of polycrystalline materials....41

Figure 2-11. Schematic diagram illustrating the transformation of ferrihydrite to goethite. ........ 42

Figure 2-12. Plots of ATR-FTIR- and XRD- derived parameters for 0%Al and 2%Al suspensionsas a function of time. These parameters reflect the presence of the crystallization product(goethite). Solid circles for 0% Al and open circles for 2% Al suspensions. Solid (0% Al)and dotted (2% Al) lines are data fits using the first order equation [Goethite] = [Goethite]0

*(1– exp-kt)...............................................................................................................................43

Figure 2-13. Rate constants for goethite appearance obtained from ATR-FTIR- and XRD-derived parameters for 0%Al and 2%Al suspensions............................................................44

Figure 2-14. Comparison of first-order rate constants for goethite crystallization obtained in thiswork to available literature data. Authors are indicated by numbers: 1. Shaw et al. (2005); 2.Nagano et al. (1994); 3. Fischer (1971); 4. Yee et al.(2006); 5. Ford et al. (1999, 2005).Methods used in calculating the rate constants: ◊ - synchrotron XRD; ○ - oxalate extraction;x - colorimetry. The aging temperature (oC) at which experiments were conducted is alsoindicated.................................................................................................................................45

Figure 3-1. Titration curves for Fe-Al mixed solutions adjusted to pH 5 by addition of 0.1 MKOH. Numbers on the plots show initial mol% Al of the 10 mmol [Al + Fe] in solution.Titration curves of solutions with 30%, 50% and 75% Al have 2 inflection points whichindicate the formation of two separate phases (Fe-hydroxides and Al-hydroxides).............. 72

Figure. 3-2. Millimoles added hydroxide at half-conversion pH points (see Fig. 3-1) versus mol%Al in initial solutions..............................................................................................................73

Figure 3-3. Aluminum in initial preparation solutions versus that observed in dialyzed nano-particle suspensions. The solid line indicates a 1:1 ratio.......................................................73

Figure 3-4. ATR-FTIR spectra of particles from Fe-Al-hydroxide suspensions (A) at initialconditions (0 days) and (B) after aging at 50 °C for 54 days. Numbers in the center of eachspectrum indicate the initial mol% Al as a fraction of [Fe + Al] in suspension.Abbreviations: Go, goethite; Gib, gibbsite . .......................................................................... 74

Figure 3-5. XRD patterns for particles from Fe-Al-hydroxide suspensions: (A) at initialconditions (0 days), and (B) after aging at 50 °C for 54 days. Numbers between panelsindicate mol% Al in suspensions as a fraction of [Fe + Al]. Abbreviations: Fh, ferrihydrite;Go, goethite; Hm, hematite; Gib, gibbsite.............................................................................75

Figure 3-6. Theoretical solubility curves for (A) iron hydroxides and (B) aluminum hydroxides inequilibrium with total aqueous Fe(III) and Al(III) species, respectively. Solubility curveswere constructed using Ksp values for Fe-hydroxides from Stumm and Morgan (1981), andfor gibbsite and diaspore from Peryea and Kittrick (1988). Shaded areas in panel B show Alconcentrations in our synthetic mixed oxide nanoparticles, with numerical ranges indicatingthe corresponding mol%Al as a fraction of [Fe + Al]. .......................................................... 76


11/187

xi

Figure 3-7. Optimized structural models for goethite with (A) 2 clustered Al atoms, and (B) 2isolated Al atoms. The distance between two Al atoms is 3.15 Å and 6.73 Å in clustered andisolated structures, respectively ............................................................................................. 77

Figure 3-8. Optimized structural models for goethite with (A) 6 clustered Al atoms, and (B) 6isolated Al atoms. The distances among Al atoms are 3-3.5 Å and 4.5-5.5 Å in clustered andisolated structures, respectively. ............................................................................................ 78

Figure 3-9. Experimental and calculated unit cell volume (A) and cell parameters a (B), b (C) andc (D) for goethite and Al-substituted goethites as a function of mol% Al substitution andtype of substitution (clustered vs. isolated). Model calculations for clustered substitution(upward pointing triangles) yield parameters that are closer to measured values.Experimental unit cell parameters were obtained from Szytula et al. (1968); Schulze (1984);Kosmas et al. (1986); Fazey and O'Connor (1991); Piszora and Wolska (1998); Gualtieri andVenturelli (1999); Scheinost et al. (2001); Ruan et al. (2002); Majzlan and Navrotsky(2003); Wells et al. (2006); Alvarez et al. (2007); Blanch et al. (2008). Thin solid linesindicate theoretical dependence of goethite unit cell parameters on mol% of Al substitution

in goethite (Vegard’s law) ..................................................................................................... 79Figure 3-10. VASP calculated free energies for iso-structural goethite and diaspore and their Al-

containing solid solutions ...................................................................................................... 80

Figure 4-1. Placement of quartz wafers into soil profile.............................................................117

Figure 4-2. Fe, Al and Si forms in soil obtained from selective extractions...............................118

Figure 4-3. Dissolved (


12/187

xii

Figure 4-10. TEM-EDS data of colloids (0.45 to 1.2 µm) from Bhs horizon containingamorphous organo-mineral colloids containing C, Al (major elements) and traces of Si, Ca,Fe, P, and S..........................................................................................................................126

Figure 4-11. Fourier transforms of Fe-EXAFS spectra of Fe-organic standards (Fe-citrate, Fe-EDTA and Fe-catechol), Fe-mineral standards (goethite and hematite) and field samples.These plots show inter atomic distances from Fe atom to its nearest neighbors (not correctedfor phase shifts)....................................................................................................................127

Figure 4-12. Schematic diagram depicting suggested composition of metal-organo-complexes insoil water. Two-end members are shown: mineral particles with organic groups attached totheir surface (left) and organic particles bridged by Fe and Al ions (right). The relativeproportion of these two end-members and composition of the intergrades between themdepends on the supply of metals and organics to the soil water. ......................................... 128

Figure A-1. SEM images of colloids (0.45-1.2 µm) from A horizon .........................................133

Figure A-2. SEM images of colloids (0.45-1.2 µm) from Bh horizon........................................135

Figure A-3. SEM images of colloids (0.45-1.2 µm) from Bhs horizon ......................................138Figure B-1-a. SEM image and EDS spectra of A-horizon, area1...............................................145

Figure B-1-b. SEM image, EDS spectra and elemental map of A-horizon, area 2. ................... 146

Figure B-1-c. SEM image and elemental map of A-horizon, area 3...........................................147

Figure B-1-d. SEM image, EDS spectra and elemental map of A-horizon, area 4. ................... 148

Figure B-1-e. SEM image, EDS spectra of A-horizon, area 5....................................................149

Figure B-2-a. SEM image and EDS spectra of Bh-horizon, area 1. ........................................... 150

Figure B-2-b. SEM image, EDS spectra of Bh-horizon, area 2..................................................150

Figure B-2-c. SEM image and EDS spectra of Bh-horizon, area 3 ............................................ 151Figure B-2-d. SEM image and EDS spectra of Bh-horizon, area 4............................................151

Figure B-2-e. SEM image and EDS spectra of Bh-horizon, area 5 ............................................ 152

Figure B-2-f. SEM image and EDS spectra of Bh-horizon, area 6.............................................153

Figure B-2-f . SEM image and EDS spectra of Bh-horizon, area 6 (magnified).........................154

Figure B-2-g. SEM image and EDS spectra of Bh-horizon, area 7 ............................................ 155

Figure B-2-h. SEM image and EDS spectra of Bh-horizon, area 8............................................156

Figure B-3-a SEM image and EDS spectra of Bhs-horizon, area 1............................................157

Figure B-3-b SEM image and EDS spectra of Bhs-horizon, area 2 ........................................... 158Figure B-3-c SEM image and EDS spectra of Bhs-horizon, area 3............................................159

Figure B-3-d SEM image and EDS spectra of Bhs-horizon, area 4 ........................................... 160

Figure B-3-e SEM image and EDS spectra of Bhs-horizon, area 5............................................161

Figure B-3-f SEM image and elemental maps of Bhs-horizon, area 6....................................... 162

Figure B-3-g SEM image and EDS spectra of Bhs-horizon, area 7............................................163


13/187

xiii

Figure B-3-h. SEM image and EDS spectra of Bhs-horizon, area 8 .......................................... 163

Figure C-1. XRD spectra of different treatments for the clay fraction (< 2 µm) from A-horizon.............................................................................................................................................168

Figure C-2. XRD spectra of different treatments for the clay fraction (< 2 µm) from E-horizon.............................................................................................................................................169

Figure C-3. XRD spectra of different treatments for the clay fraction (< 2 µm) from Bh-horizon.............................................................................................................................................170

Figure C-4. XRD spectra of different treatments for the clay fraction (< 2 µm) from Bhs-horizon.............................................................................................................................................171

Figure C-5. XRD spectra of different treatments for the clay fraction (< 2 µm) from C-horizon.............................................................................................................................................172

LIST OF TABLES

Table 3-1. ATR-FTIR- and XRD-determined mineral phases in mixed Fe and Alhydroxide nano-particle suspensions as a function of aging time and initial Al molesas a fraction of total Al and Fe. Abbreviations: Fh, ferrihydrite, Go, goethite, Hm,hematite, Gib, gibbsite. Numbers represent estimates of relative peak intensity(higher numbers for higher intensities). .................................................................... 69

Table 3-2. Calculated and experimental crystallographic parameters and calculated freeenergy for goethite, Al-goethites, diaspore and gibbsite........................................... 70

Table 3-3. Relative stability of Al-substituted goethites and diaspore with respect togoethite (∆Hreaction) .................................................................................................... 71

Table 3-4. Relative stability of gibbsite and diaspore with respect to corundum (∆Hreaction)................................................................................................................................... 71

Table 4-1. Description of soil horizons of Black Moshannon Spodosol site................. 115

Table 4-2. Infrared band assignments in the 4000-600 cm-1 region............................... 116

Table A-1. Elemental composition of colloids (0.45µm


14/187

xiv

ACKNOWLEDGMENTS

First of all I would like to express my greatest appreciation to my academic

adviser, Dr. Carmen Enid Martinez for her guidance and constant support. It was a

pleasure to work with her. I am also deeply indebted to my PhD Committee members

Douglas Archibald, Edward Ciolkosz, James Kubicki, and Kwadwo Osseo-Asare for the

invaluable advice throughout my research and the comments to my manuscripts. They

helped me immensely to believe in myself as a scientist. I also want to thank Ephraim

Govere for his advice and discussions on different analytical approaches. I was really

lucky to be a Fellow of the Center for Environmental Kinetics Analysis (CEKA) at Penn

State, which gave me an opportunity to communicate with many bright scientists at and

outside Penn State and to present my work in the interdisciplinary environment.

On a more personal note, I would like to thank my children Nikita, Natasha, and

Kristina for their unselfish understanding and for keeping me sane during the years of my

graduate studies. Finally, I thank my husband Mark Fedkin for his constant

encouragement, support, and sacrifice.

Funding for this work was provided by the National Science Foundation, Grant #

CHE-0431328, and The Penn State Biogeochemical Research Initiative for Education

(BRIE) under National Science Foundation Grant # DGE-9972759.


15/187

1

Chapter 1

INTRODUCTION

1.1. Background

Migration of mineral nanoparticles through the soil profile and their interaction with

dissolved elements, organic matter and soil minerals are important processes in soil development

(Hochella et al., 2008, Banfield and Zhang, 2001). Due to a high specific surface area,

nanocolloids are known to have a greatly increased reactivity towards many geochemically

important solute species and hence are largely responsible for the retention and transport of metal

ions, contaminants, and nutrients in the environment (Zhang, 2003; Madden et al., 2006;Novikov et al., 2006; Wigginton et al., 2007). Some of their distinct properties are believed to

result from variations (or distortions) in their atomic structure at the nano-scale (Waychunas et

al., 2005). It is also known that thermodynamic properties such as surface enthalpies, enthalpies

of mineral phase transition, and aqueous solubility may be functions of particle size, favoring the

stability of particular minerals at different size ranges (Navrotsky, 2008).

Among the mineral nanoparticles that exist in soils, iron and aluminum hydroxides

(Fe(OH)3, Al(OH)3,) are the most abundant and environmentally relevant. These low-crystalline

materials of colloidal size (


16/187

2

Some of the questions that need to be addressed are: (1) what is the mechanism of

goethite crystallization from amorphous Fe-hydroxide (ferrihydrite) precipitated in the presence

of Al, and how Al influences the rate of ferrihydrite-goethite transformation? (2) how does Al

incorporate in the structure of goethite during crystallization? (3) are co-precipitated hydroxides

more stable compared to pure Fe- and Al-hydroxide nanoparticles? (4) how do Fe- and Al-

hydroxide nanoparticles form, migrate and interact in complex soil environments?

To address these questions, in my dissertation I use a combination of laboratory

experiments, field work, and molecular modeling to explore the issues of Fe and Al hydroxide

nano-particle crystallization, transformation and interaction in soil environments.

1.2. Dissertation structure

This dissertation consists of three main chapters that will be submitted for publication in

peer reviewed journals. The chapter titled “Nano-goethite (α-FeOOH) crystallization in the

presence of low aluminum concentrations under environmentally relevant conditions” (Chapter

2) focuses on quantification of the rate of ferrihydrite-goethite transformation in nano-colloidal

suspensions in the presence of low concentrations of Al, and on understanding the mechanisms

of goethite crystallization. My hypothesis was the following: even low Al concentrations in

nano-colloidal suspensions may significantly decrease the rate of goethite crystallization from

ferrihydrite suspensions. To quantify the rate of this transformation, we applied a novel

analytical approach to estimate goethite contents in nano-colloidal suspensions using infrared

data.

Chapter 3, “Co-precipitation of Fe and Al hydroxide nanoparticles: ATR-FTIR, XRD and

Periodic Density Functional Theory Calculations” tests the following hypotheses: (1) the kinetics

of crystallization of minerals (goethite and/or gibbsite) from mixed Fe-Al hydroxide


17/187

3

nanoparticles is much slower than that of pure phases due to the formation of intermediate

phases that are indefinitely metastable in low-temperature soil environments; (2) Al incorporates

into goethite by forming Al-clusters, which is energetically more favorable compared to isolated

substitution of Al for Fe; however, even in clustered arrangement, Al creates strain on the

goethite structure which delays its further crystallization. To test these hypotheses, we used a

combination of experimental data and ab initio molecular modeling to follow phase

transformations in co-precipitated Fe-Al hydroxides with a wide range of Fe/Al ratios.

Finally, Chapter 4 (“Mobility and accumulation of Fe, Al and Si organic and inorganic

forms in soil and soil solutions”) is an attempt to go across scales and extend knowledge on thebehavior of Fe- and Al-hydroxides nanoparticles to soil environments. For this purpose, we

studied a spodosol characterized by intensive transport and accumulation of Fe and Al within the

soil profile. We hypothesized that: in this soil, Fe and Al migrate from surface horizons to the

subsoil in the form of inorganic colloids (Fe and Al hydroxides, short-range aluminosilicates)

that accumulate in lower profiles due to interaction with soil organic matter. To test the

hypothesis, we used various spectroscopic and microscopic techniques to explore the

composition of soil water and soil coatings.

References

1. Banfield, J.F. and Zhang, H. (2001) Nanoparticles in the environment. In: Nanoparticlesand the environment. Eds: Banfield, J.F. and Navrotsky, A. Reviews in mineralogy and

geochemistry, V. 44, pp. 1-582. Blanch, A.J, Quinton, J.S., Lenehan, C.E. , and Pring, A.(2008) The crystal chemistry of

Al-bearing goethites: an infrared spectroscopic study. Miner. Magazine, 72, 1043-1056

3. Colombo, C. and Violante, A. (1996) Effect of time and temperature on the chemicalcomposition and crystallization of mixed iron and aluminum species. Clays and Clay Minerals, 44, 113-120.


18/187

4

4. Fey, M.V., and Dixon J.B. (1981) Synthesis and properties of poorly crystalline hydratedaluminous goethites. Clay and Clay Miner . 29, 91-100

5. Hochella, M.F., Lower, S.K., Maurice, P.A., Penn, R.L., Sahai, N., Sparks. D.L. andTwining, B.S. (2008) Nanominerals, mineral nanoparticles and earth systems. Science,319, 1631-1635

6.

Madden, A.S., Hochella, M.F, and Luxton, T.D. (2006) Insights for size dependentreactivity of hematite nanomineral surfaces through Cu2+ sorption, Geochim etCosmochim Acta, 70, 4095-4014

7. Navrotsky, A., Mazeina, L., and Majzlan, J. (2008) Size-driven structural andthermodynamic complexity in iron oxides. Science, 319, 1635-1638

8. Novikov, A.P., Kalmykov, S.N., Utsunomiua, S., Ewing, R.C., Horreard, F., Merkulov,A, Clark, S.B., Tkachev, V.V., and Myasoedov, B.F. (2006) Colloidal transport ofplutonium in the far-field of the Mayak Production association, Russia. Science, 314,638-641

9. Waychunas, G. A., Kim, C.S, and Banfield, J.F. (2005)

Nanoparticulate iron oxideminerals in soils and sediments: unique properties and contaminant scavengingmechanisms. J. of Nanoparticle Res. 7, 325-571

10. Wigginton, N.S., Haus, K.L., and Hochella, M.F. (2007) Aquatic environmentalnanoparticles. J. Env. Monitoring, 9, 1306-1316

11. Zhang, W.X. The nanoscale iron particles for environmental remediation: An overview. Journal of nanoparticle research, 5, 323-332


19/187

5

Chapter 2NANO-GOETHITE CRYSTALLIZATION IN THE PRESENCE OF LOW ALUMINUM

CONCENTRATIONS UNDER ENVIRONMENTALLY RELEVANT CONDITIONS

Abstract

We quantified rate constants for goethite (α-FeOOH) crystallization from dilute nano-

particulate Fe hydroxide suspensions (Fe(III) = 10 mM) aged for up to 54 days at 50 oC in the

absence (0% Al) and presence (2% Al) of aluminum using attenuated total reflectance Fourier

transform infrared (ATR-FTIR) spectroscopy and synchrotron-based X-ray diffraction (s-XRD).

We used multivariate curve resolution (MCR) analyses of the OH-stretching region of the FTIR

spectra to derive “ferrihydrite-like” and “goethite-like” components. Each of these components

consisted of (1) non-stoichiometric OH stretching vibrations (reflect the combination of surface

hydroxyls and disordering of the Fe-hydroxide structure) and (2) stoichiometric OH-stretching

vibrations (reflect characteristic hydrogen bond (O-H...H) stretching vibrations in the goethite

structure) as assumed from band component analysis (Gaussian functions). Rate constants were

obtained by fitting the contribution of goethite-like MCR component to the overall spectra, and

were equal to (7.4 ± 1.1)*10-7 s-1 for 0% Al and (4.2 ± 0.4)*10-7 s-1 for 2% Al-doped iron

hydroxides. Rate constants derived from intensities of the 790 cm-1 and 890 cm-1 OH-bending

infrared vibration showed similar values (within error) for both 0% Al and 2% Al. The presence

of 2% Al decreased the rate constants determined from analyses of infrared OH-stretching and

OH-bending vibrations by 43-57%, but did not change the rate constant determined from

synchrotron-XRD. The presence of hydrogen bond types (O-H…H) characteristic of goethite in

all nano-particle suspensions suggests nano-goethite- and amorphous Fe-hydroxide-like

structures co-existed and formed during the synthesis of Fe-nanoparticles. A mechanism for


20/187

6

nano-goethite crystallization under experimental conditions close to environmental is also

discussed.

2.1. Introduction

In terrestrial and aquatic environments, nano-sized Fe (oxy)hydroxides exhibit a variety

of amorphous and crystalline structures, among which ferrihydrite (Fe(OH)3·H2O) and goethite

(α-FeOOH) are the most common phases (Waychunas et al., 2005; Wigginton et al., 2007).

Goethite co-exist with ferrihydrite in spodosol, volcanic ash, bog iron, and hydromorphic soils

formed in moist and cool temperate climate (Schwertmann and Taylor, 1989; Bigham et al.,2002), and in acid mine drainage waters at pH


21/187

7

have been shown Al retards ferrihydrite crystallization (Colombo and Violante, 1996; Cornell

and Schwertmann, 2003). To our knowledge, no attempt has been made to quantify the rate of

nano-goethite crystallization in the presence of Al, although numerous experimental studies have

shown that Al retards the formation and crystallization of Fe-hydroxides by forming Fe-Al

coprecipitates and/or by sorption to the surface of already formed Fe-hydroxides (Anand and

Gilkes, 1987; Fontes et al., 1992).

When studying structural transformations of nanoparticulate and poorly-crystalline

substances, it is always a challenge to properly identify and quantify mineral phases. Rate

constants for the transformation of ferrihydrite to goethite have been determined fromoperationally defined chemical extractions using ammonium oxalate (Schwertmann and Murad,

1983; Cornell et al., 1989) or 0.4 M HCl (Ford et al., 1999), which are presumed to selectively

dissolve ferrihydrite. However, McCarty et al. (1998) showed that ammonium oxalate dissolved

considerable amounts of nano-crystalline goethite in Fe-rich precipitates from an acid mine zone,

which indicates that selective extraction methodologies might not be appropriate in the study of

nano-particle transformations. This concern is also supported by the earlier observation by

Schwertmann (1973) that oxalate extraction could dissolve nano-sized goethite crystallites.

Colorimetric techniques (Nagano et al., 1994; Hamzaoui et al., 2002) and synchrotron-based X-

ray diffraction (XRD) (Shaw et al., 2005; Yee et al., 2006) were also used to determine the rate

constants for the ferrihydrite to goethite transformation by detection of the fraction of goethite in

samples aged for up to 15 hours. However, most of these laboratory studies were conducted

under alkaline pH conditions (pH > 8), which are rare in natural environments. Rate constants of

goethite crystallization in alkaline solutions were found to be at least one order of magnitude

higher than rate constants obtained for more typical environmental pH (4 < pH < 8) conditions

(Fischer and Schwertmann, 1975). In addition, Hockridge et al. (2009) found that diffraction


22/187

8

patterns of goethite are not detectable in the mixtures of nanocyrstalline goethite and ferrihydrite

until the amount of goethite present exceeds about 10%. On the other hand, infrared

spectroscopy provides significant advantages over XRD for studying the transformation of

ferrihydrite to nano-goethite in nano-colloidal suspensions. It is more sensitive for the

determination of small amounts of goethite in poorly-crystalline ferrihydrite suspensions because

it probes short-order molecular vibrations.

In this work we synthesize iron hydroxide nanoparticulate suspension under

environmentally relevant conditions (pH=5, T=50 °C, slow titration rate, ionic strength of

solution < 5x10

-3

). We use synchrotron-XRD and Attenuated total reflectance Fourier transforminfrared (ATR-FTIR) spectroscopy to (1) follow structural transformations of ferrihydrite to

goethite in nanoparticle suspensions, (2) quantify the relative amounts of ferrihydrite and

goethite, (3) calculate the rate constants of nano-goethite crystallization in the absence and in the

presence of low concentrations of aluminum, and (4) to obtain some insight on the mechanism of

ferrihydrite-to-goethite transformation.

2.2. Experimental Section

2.2.1. Synthesis of Fe-hydroxide and Al-doped Fe-hydroxide nanoparticles

To mimic realistic environmental conditions, pure Fe- and Al-doped-Fe-hydroxide

nanoparticles were synthesized at 25 oC using a relatively low Fe (III) concentration (0.01 M)

and a slow rate (1 mL/min-1 to 1 L) of base addition to a pH of 5. In preliminary experiments,

series of Fe-Al mixed hydroxide were prepared from homogeneous solutions with Al

concentrations 2, 4, 8, 12, 16, and 20. It was found that for 4 and 8% Al hematite appeared along

with goethite as a second crystallize phase, and at higher Al concentrations (12, 16 and 20)

goethite was not detected for the duration of experiment. Therefore, we used suspensions with 2


23/187

9

mol% Al to evaluate the effect of low Al concentrations on goethite crystallization. For each 0%

and 2%Al composition, two replicate suspensions were synthesized. The following procedure,

modified from Bakoyannakis et al. (2003), was specifically designed to produce nano-sized Fe

and Fe-Al hydroxide particles. Specific amounts of Fe(III)-nitrate and Al-nitrate salts were

dissolved in 100 mL of de-ionized (DI) water to obtain the desired 0% and 2% Al in solutions.

Solutions were slowly titrated to pH 5 with 0.1 N KOH (1 mL min-1) using a pH-controlled

Masterflex C/L TM Pump system for ~3-6 hours. The final volume of all samples was adjusted to

1 L, and a final total metal ([Fe+Al]) concentration of 10-2 M was obtained. The suspensions

were then dialyzed at 25 °C against Milli-Q water using 1 nm pore size Spectropor/7 dialysistubes (molecular weight cutoff 1000) to remove excess salts and excess Al, so all the Fe and Al

contained in the dialysis tubes were in nano-particulate form. The water was replaced several

times over the course of several days, until the conductivity was reduced to 0.3 S/cm (ionic

strength value < 0.005 M). Aliquots of each suspension (~100 ml) were taken immediately after

dialysis (time = 0 days) and after the suspensions were aged for 2, 9, 23, and 54 days at 50°C. No

precipitates were formed during the course of the experiment, and all suspensions remained

visually clear without any precipitates. A portion of the aliquots (~40 ml) collected at each time

interval was freeze-dried for s-XRD analysis, while another portion was stored at 4 °C in the

dark in polystyrene bottles until used for ATR-FTIR, DLS, and TEM analyses as described

below. All nano-colloidal suspensions were yellow-brown in color and remained optically

transparent for the duration of the experiment and measurements.

The elemental composition of dialyzed suspensions was determined by dissolving 1 ml of

suspension in 2 ml of concentrated HNO3 and heating at ~80 °C for 1 hour. The digest was

diluted 100 times with DI water, and total dissolved Fe and Al concentrations determined by

atomic absorption spectroscopy. As expected, for the samples with 2% Al in initial solution,


24/187

10

chemical analysis revealed 1.95% Al (n=2) , which indicate that almost all of the Al introduced

to the system either formed a coprecipitate with the iron hydroxides or retained at its surface

(Lovgren et al., 1990; Fey and Dixon, 1981).

2.2.2. Synchrotron-based X-ray diffraction

High-resolution synchrotron-based XRD patterns were collected at beamline X16C,

National Synchrotron Light Source, Brookhaven National Laboratory. An X-ray wavelength of

0.70051(7) Å was selected with a Si(111) double crystal monochromator. Freeze-dried powder

samples were loaded into a glass capillary to consistent density (0.7 mm internal diameter) and

rotated about the longitudinal axis during data collection. XRD patterns were collected in the 2θ-range from 2° to 40° using a 0.05º step size and a counting time of 5 seconds per step. Jade+

software (version 8.2, Materials Data, Inc., Livermore, CA) was used for the identification of

crystalline phases and measurement of peak areas. Goethite crystallite sizes were estimated using

the Scherrer equation for the most intense and non-overlapping (110) peak. The full-width at

half-maximum (FWHM) values were corrected for instrument broadening using silicon powder

standard.

2.2.3. ATR-FTIR spectroscopy

2.2.3.1. ATR-FTIR data collection and Gaussian band analyses

ATR-FTIR measurements were performed using a Bruker Tensor 27 FTIR instrument

with a La DTGS detector and equipped with single-reflection diamond/ZnSe ATR accessory. A

0.5 µL aliquot of the hydroxide suspension was placed on the sensor and air-dried for 5 min to

form a thin coating before data collection. The sensor background spectrum was measured before

each sample. Spectra of 8 cm−1 resolution were acquired by co-addition of 256 scans within the

4000-600 cm-1 region. Three or more replicates were measured for each sample. Spectral

manipulation, including subtraction of background water and CO2 vapor, baseline correction, and


25/187

11

spectra normalization were performed using in-house developed codes under the MATLAB 7.4®

computing environment. Baseline correction was performed by subtracting a 2-point linear

baseline in the background region and applying standard normal variate correction across the

selected range. The 4000-2500 cm-1 spectral regions were normalized to a unit area (i.e. rescaled

in order to set the area under the peak equal to 1). Due to intense sloping background in the

1000-650 cm-1 region, for each spectrum the absorption at each wavelength was divided by the

area of corresponding peak in the 4000-2500 cm-1 region.

Gaussian band analysis (deconvolution) of the iron (oxy)hydroxide OH-stretching (4000-

2500 cm

-1

) region of each ATR-FTIR spectrum was performed using the nonlinear least-squaresfitting routine SOLVER in MS Excel (de Levie, 2001). The choice of number of bands and their

positions were made based on the available literature data (Farmer, 1974; Morterra et al., 1984;

Weckler and Lutz,, 1998; Ruan et al., 2002). Spectral de-convolution of measured FTIR peaks of

FeOOH is typically performed to extract information about the contributions of different OH-

groups (Hug, 1997; Ruan et al., 2001; Boily et al., 2006). The choice of number of bands used in

fitting a band envelope and the band positions largely depends on a researcher’s model, and

sometimes mathematically the best fit may not be the best fit to explain the chemistry of the

system (Meier, 2005). This univariate approach, when each spectral group is assigned to a

specific component, can be challenging in the case of highly overlapped varying intervals bands

in the OH-stretching region. In addition to analyses of overlapping OH-stretching vibrations,

peak heights for non-overlapping OH bending vibrations were determined from Gaussian curves

fitted to experimental spectra.


26/187

12

2.2.3.2. Multivariate curve resolution (MCR) analysis of ATR-FTIR spectra

To complement our Gaussian band deconvolution approach we brought a new technique,

multivariate curve resolution-alternating least squares (MCR-ALS). MCR permits extraction of

chemically reasonable spectral components to describe the whole system (Garrido et al., 2008)

and assumes that each spectrum in a dataset in composed of “pure” component spectra. These

pure components contribute to the spectral intensity in various proportions (“scores”) and can

completely explain the observed spectral variance in the dataset, which in our case spans an

aging (crystallization) process. MCR analysis is able to reduce and analyze matrices of ATR-

FTIR spectra by extracting major spectral features (MCR components) and their correspondentweights (concentrations) to describe the whole original experimental dataset (Tauler, 2003). This

approach is especially useful in its application to nano-particle crystallization studies where

infrared spectra are collected as a function of a perturbation to the chemical system (i.e., aging)

and when no end-member standards are available for quantification (Schoonover et al., 2003).

MCR analysis consists of 3 major steps: (1) estimation of the number of pure components; (2)

eigenvalue decomposition of the full dataset matrix to produce abstract factors and scores,

arranged with contributions of each factor to the measured spectra in decreasing order; and (3)

refining abstract factors into real spectra of pure components by fitting the sum of factors, each

with a particular score, to the experimental spectra by an iterative ALS process so that the

percent of the spectral variance not explained by MCR model was minimal. The non-negativity

constraint was applied where only positive values were allowed for pure components and scores.

The choice of the number of components was based on the physical meaning of extracted

components, on the percent of the variance explained by the fit, and on a plot of eigenvalues

versus number of factors that allows the differentiation of real signal from noise (Schoonover et

al., 2004). MCR analyses were conducted using MATLAB (The Mathworks) with the PLS


27/187

13

toolbox Version (Eigenvector Inc.) and in-house developed functions. Normalized spectra were

used in all analyses.

2.2.4. Dynamic light scattering (DLS)

Particle size distribution for the Fe- and Fe-Al- hydroxide nanoparticle suspensions was

determined by dynamic light scattering using a ZetaSizer Nano S (Malvern Instruments, UK).

Particle size distributions were obtained by fitting the correlation functions to multiple

exponentials using non-negatively constrained least squares (NNLS) analysis. The diffusion

coefficient (D) was calculated by fitting the correlation curve to an exponential function, with D

being proportional to the lifetime of the exponential decay. The hydrodynamic radius (RH) was

then calculated from the diffusion coefficient using the Stokes-Einstein equation, H R

kT D

πη 6=

where k is the Boltzmann constant, T is the temperature, and η is the medium viscosity. Since

particle size distributions were polydispersed for most of the samples, the intensity distributions

were converted to volume distributions according to the Mie theory (Dahneke, 1983).

Suspensions of iron (oxy)hydroxide with 0% and with 2% aluminum-doping aged for 0, 2, 9, 23

and 54 days were placed in disposable square polystyrene cuvettes and light scattering measured

with a laser beam at an angle of 173° between the incident and diffracted beams. Additionally, in

situ heating experiments at 50 °C were performed for both Fe- and Fe-Al suspensions by

measuring the light scattered from an aliquot of the initial suspensions (time = 0 days) every 2

hours for about 2 days in automatic mode.

2.2.5. Transmission electron microscopy (TEM)

Samples for TEM were prepared by placing a single drop of the Fe and Fe-Al

suspensions onto a 200 mesh holey-carbon-coated copper grid and allowed to air-dry. TEM

images and selected area electron diffraction (SAED) patterns were collected using a JEOL EM-


28/187

14

2010 with a LaB6 filament operated at 200 kV. TEM images were obtained for non-diluted and

diluted suspensions. High-resolution TEM (HRTEM) images were obtained for diluted

suspensions using a JEOL EM-2010F (Field Emission TEM/STEM) operated at 300 kV.

2.3. Results

2.3.1. XRD data

The characteristic broad bands at d-spacings of 0.15 and 0.26 nm indicate that poorly

ordered 2-line ferrihydrite (Cornell and Schwertmann, 2003) was the dominant phase in undoped

and 2%-Al-doped ferric (oxy)hydroxide nano-colloid suspension at initial conditions and in 2-day aged 2%Al suspension (Fig. 2-1). The intensity of the main XRD peak (110) (0.42 nm in d-

space), which is characteristic of goethite, increased with time for the 0% and 2% Al

suspensions. X-ray diffraction patterns of Al-free suspensions, however, revealed fairly

crystalline goethite was present after 54 days of aging at 50°C, while in suspensions with 2% Al

the intensities of the goethite peaks were lower in intensity with little goethite formed until 9

days (Fig. 2-1). No other crystalline phases except goethite were determined in suspensions.

Since the (110) peak reflaction is distinct from ferrihydrite and does not overlap with other

peaks, its area was used to determine the rate constant for the crystallization of goethite from

ferrihydrite (section 4.2). Goethite (110) reflection was also used to determine crystallite sizes

using the Scherrer equation (Patterson, 1939). Analyses (Fig. 2-2) showed crystallite size

increased with time from 2.5 to 5.5 nm in 0% Al suspensions. In the presence of 2% Al,

crystallite sizes were in the range from 3.5 to 4.8 nm with no clear trend with time (Fig.2-2).

These results suggest the presence of Al decreased the degree of structural order in nanoparticle

suspensions, as is also evidenced from XRD patterns (Fig.2-1).


29/187

15

2.3.2. Infrared spectra

2.3.2.1. OH-bending vibrations

Goethite’s structure consists of double sheets of edge-sharing FeO3(OH)3 octahedra

connected to each other by corner-sharing (O bridges) that form 2x1 octahedra tunnels

(Schwertmann and Cornell, 1991). Hydrogen bonds (O-H···O) that cross the tunnels and link the

chains of octahedra are structurally equivalent, but point in different directions (Fig. 2-3). The

presence of these hydrogen bonds in a Fe hydroxide mixture is an indication of the presence of

goethite and can be used to follow the ferrihydrite to goethite crystallization process. The

presence of these hydrogen bonds is verified by characteristic infrared stretching (3120-3200 cm

-

1) and bending (895 and 795 cm-1) vibrations in bulk goethite (Morterra et al., 1984, Cornell and

Schwertman, 2003). Hydroxyl bending vibrations at 895 cm-1 and 795 cm-1 are diagnostic for

goethite and reflect OH bending deformations in the a-b plane and out of the a-b plane

deformations, respectively (Schwertmann et al., 2003). Their intensities increased upon aging

due to the increase in the number of goethite hydrogen bonds in suspension (Fig. 2-4). It is

important to note that in Al-free suspensions the OH-bending vibrations, although broad and of

low intensity, started to appear even at 0 days. In contrast, OH-bending vibrations were not

observed at 0 days in suspensions with 2% Al, and lower intensity peaks were developed at

corresponding times compared to Al-free suspensions (Fig.2-4). The intensities (peak height) of

the OH-bending vibrations were used to determine rate constants of ferrihydrite to goethite

transformation (section 2.4.2).

2.3.2.2. OH-stretching vibrations: Gaussian deconvolution

Unlike OH-bending vibrations, IR bands in the OH-stretching region overlap strongly

making it difficult to directly estimate intensity changes due to system perturbation (Fig. 2-3).

The types of spectral envelopes of the OH-stretching region were similar for Al-free and 2% Al-


30/187

16

containing Fe-hydroxides suspensions. Replicate variability was higher in 2% Al suspensions

(Fig.2-4) possibly due to heterogeneity in particle shapes and sizes. As shown by our results, the

infrared OH-stretching region of iron hydroxides is characterized by relatively broad bands

which contain multiple OH contributions under broad peak areas (Kubicki et al., 2008). For our

analysis, the OH-stretching envelopes were deconvoluted using Gaussian functions to model

three OH contributions: stoichiometric OH vibrations, non-stoichiometric OH vibrations, and

adsorbed water (Fig. 2-5A). Stoichiometric, or bulk, OH groups are those not reactive in

adsorption/desorption processes and are difficult to exchange with deuterium (D2O) vapor

(Morterra et al., 1984). They usually yield infrared bands in the 3050 to 3240 cm

-1

rangedepending on crystallinity and crystal size of goethite and on the infrared technique used. These

OH groups form hydrogen bonds (O-H···O) between double chains of octahedra (Weckler and

Lutz, 1998), and are characteristic of the goethite structure. Bands in the 3400-3480 cm -1 range

have been attributed to weakly H-bonded surface hydroxyls (Rochester and Topham, 1978), bulk

OH groups in the ferrihydrite structure (Russel, 1979), and to non-stoichiometric hydroxyl units

incorporated into the goethite structure during the process of synthesis, when, for example, one

iron ion is replaced by 3 protons (Ruan et al., 2001; Boily et al., 2006). Most of these hydroxyls

can be readily exchanged with deuterium, which suggest high reactivity and mobility, and by

definition, indicate greater degree of disorder in the goethite structure. Furthermore, Boily et al.

(2006), who studied goethite suspensions evaporated on an ATR cell under dry N2 gas

atmosphere, assigned OH-stretching vibrations in the 2600-3800 cm-1 region to contributions

from (1) structural, or stoichiometric, hydroxyls in bulk goethite at ~3027 cm-1, (2) non-

stoichiometric hydroxyls in bulk goethite at ~3384 cm-1, and (3) OH stretching of isolated

surface hydroxyls at ~3640-3660 cm-1. In agreement with their data, our ATR-FTIR spectra also

had two broad bands that can be attributed to stoichiometric and non-stoichiometric hydroxyls,


31/187

17

but lack surface hydroxyl bands at high frequencies (Fig.2-5A). A possible explanation for the

absence of these bands is that our measurements were done under ambient atmosphere, and the

sorption of water vapor onto the Fe hydroxide film could have suppressed surface hydroxyl

features. Adsorbed water may also cause a significant asymmetric broadening at lower

frequencies (~2900-3000 cm-1; Figs. 2-3 and 2-4A) (Morterra et al., 1984). As depicted in

Figure 2-4A, band component analysis was used to differentiate among the contribution of

specific OH vibrations to the OH-stretching envelope, thus isolating stoichiometric OH

vibrations characteristic of the goethite structure. The best fit to the experimental spectra was

obtained by using three Gaussian functions, assigned as non-stoichiometric (~3460 cm

-1

),stoichiometric (~3240 cm-1), and adsorbed water (2960 cm-1) (Fig. 2-5A). These band

component assignments are in good agreement with the data of Ruan et al. (2002) who obtained

bands for stoichiometric and non-stoichiometric hydroxyls at 3233–3206 and 3476–3450 cm−1,

respectively, and two other bands (~2804, 2930 cm-1) at lower frequencies. One of the

approaches we undertook to follow goethite’s crystallization process was to use changes in

frequency and in full width at half maximum (FWHM) values of stoichiometric hydroxyls peaks

with time, which were obtained by Gaussian deconvolution of the OH-stretching region. In

general, the width of the Gaussians that describe the OH stoichiometric components of aged (i.e.,

54 days) suspensions decreased and their intensity increased compared to initial (i.e., 0 days)

suspensions, thus suggesting a continuous increase in structural order of nano-goethite with

aging time. This is evidenced by a decrease in the peak position (wavenumbers at center of

stoichiometric OH vibrations) at which the 0%Al and 2% Al suspensions appeared with time

(Fig. 2-5B). In addition, FWHM values for the stoichiometric OH vibrations decreased steeply

within the first 10 days of aging (Fig. 2-4C). Peak positions and FWHM values are slightly

higher for 2% Al suspensions. The shape and position of adsorbed water remained constant for


32/187

18

all suspensions. It was suggested by Waychunas et al (2005) that during crystallization double

chains of edge-sharing FeO3(OH)3 octahedra, the building block of the goethite structure, begin

to stack together by Fe-O-Fe corner sharing and by the formation of hydrogen bonds between

these double chains (Fig. 2-3). We hypothesize that as the number of free hydroxyls decrease

(Fe-O-H bonds present in the octahedra) and the number of H-bonded hydroxyls increase (Fe-O-

H···O-Fe that connect double chains, the energy of the hydroxyl bond decreases and their

frequencies shift to lower wavenumbers, as observed in our results (Fig. 2-5B). On the other

hand, the relatively large initial FWHM value indicates there is variability in the energy of H-

bonded hydroxyls that contribute to stoichiometric OH vibrations (Fig. 2-5C). A substantialdecrease (Fig. 2-4C) in the FWHM value for stoichiometric OH vibrations with time

(approximately 40%) indicates a decrease in the variability of H-bonded hydroxyls and therefore

an increase in structural order. In the presence of 2% Al, the formation of hydrogen bonds

between double chains of octahedra is hindered compared to Al-free suspensions as evidenced

from slightly higher peak positions and FWHM values with time. A lower degree of structural

order in Al-doped suspensions may result from disturbance to the network of hydrogen bonds

that form across double chains of Fe octahedra by substitution with smaller Al octahedra. The

presence of Al might then increase the number of non-stoichiometric hydroxyls and/or increase

variability within stoichiometric hydroxyls in Al-doped nano-goethite.

2.3.2.3. MCR analysis of IR O-H stretching region

Multivariate curve resolution (MCR) analysis was used to extract pure components and to

describe their contribution to spectral variations that occurred during the crystallization process

(Tauler, 2003). MCR analysis is appropriate for systems in which there is a significant change in

spectral features as the system evolves with time, i.e. amorphous to crystalline transformations as

in the case of this study. A 2-component system, in which the components were inversely


33/187

19

related, resulted from MCR analysis of the OH-stretching region (Fig. 2-6). A 2-component

system is in agreement with our Gaussian deconvolution results and with principal component

analysis (PCA), both of which revealed the contribution of two primary “components”

(stoichiometric and non-stoichiometric OH vibrations in Gaussian deconvolution) in the dataset.

MCR is more advantageous than PCA because it is possible to add non-negativity constraints in

the MCR algorithm.

MCR-extracted components (spectra) and scores, along with experimental spectra for the

0%Al dataset are shown in Fig. 2-6. Component 1 characterized the system initially, while

Component 2 resembled the most aged samples. It is clear that none of these components can beattributed to a single type of OH-stretching vibration, but rather consists of overlapping bands.

To bring chemical relevance to this statistical method and aid in its interpretation, each of the

main components was deconvolved into single bands in the same manner we did for the

experimental spectra (Gaussian deconvolution). Similar to the experimental spectra discussed

above (section 2.3.2.2), each component exhibited non-stoichiometric hydroxyls, stoichiometric

hydroxyls and adsorbed water (Fig. 2-6B). Conceptually, Component 1 can be presented as

disordered Fe-hydroxide or “ferrihydrite-like” phase, in which non-stoichiometric hydroxyls are

dominant (>30% of the total volume; Zhao et al., 1994) due to structural disorder and large

surface area characteristic for ferrihydrite, but minor concentrations of goethite nanoparticles are

already present. In contrast, Component 2 can be considered a “goethite-like” phase since it

exhibits well-expressed stoichiometric OH-stretching vibrations at ~3100 cm-1 with some

structural disorder/surface water at ~3300 cm-1 (Fig. 2-6B). These two components described

99.2% and 99.6% of the variance in the 0%Al and 2%Al datasets, respectively, and may be

considered the two end-members of the goethite crystallization reaction. Their contributions to

spectral changes as a function of time are reflected in the MCR-derived scores (Fig. 2-6B),


34/187

20

which clearly indicate a gradual decrease of the “ferrihydrite-like” component and a concomitant

increase of the “goethite-like” component (Fig. 2-6B). Scores for the “goethite-like” component

can be directly related to the number of goethite stoichiometric OH-groups in the suspension and

can therefore describe the kinetics of goethite crystallization.

2.4. Discussion

2.4.1. Mechanisms for nano-goethite crystallization

The rate and mechanism of ferrihydrite-goethite transformation in laboratory experiments

depend on many factors among which pH, Fe concentration, particle size, and the presence oftrace metal impurities are the most important ones. For example, the effect of particle size on

phase transformation kinetics becomes clear when several authors suggested that in addition to

conventional Ostwald ripening (defined by a first-order equation) the kinetics of phase

transformation in nanoparticulate systems can be described by oriented aggregation, a

mechanism specific for nanoparticles (Banfield et al., 2000; Waychunas, 2001; Zhong et al.,

2006; Jia et al., 2006; Penn et al., 2007). The kinetics of oriented aggregation was quantified

from particle sizes obtained by transmission electron microscopy (TEM) and dynamic light

scattering (DLS), and the rate law was found to be second order in the concentration of primary

nanoparticles (Penn, 2004). These two mechanisms are not exclusive: aggregation can

predominate at first, followed by Ostwald ripening (Waychunas et al., 2005), or they may occur

simultaneously. Thus, it is important to understand the mechanism of ferrihydrite-goethite

transformation in particular systems to correctly model kinetic data and derive rate constants.

Goethite crystallization kinetics is usually described by a dissolution-precipitation mechanism

(i.e. Ostwald-ripening) wherein that small particles with high surface potential dissolve due to

their higher solubility compared to larger particles which grow by incorporating Fe mono- and


35/187

21

poly-nuclear species from solution into the crystal structure by diffusion (Jolivet, 2004). This

mechanism obeys a first-order type reaction and considers only the volume of goethite

transformed with time; Ostwald ripening assumes intermediate Fe-hydroxide species are not

involved. In our suspensions, dynamic light scattering (DLS) revealed no measurable increase in

average particle size at ≤ 2 days (average particle size = 5.04 ± 3.7). An increase in particle size

(to 40-55 nm) was observed for 0% Al suspensions after 9 days (Fig. 2-7). However, the 2% Al

containing suspensions showed more variability in particle sizes during the first 2 days of aging,

with values ranging from 2 to 30 nm and no trend with time (Fig. 2-7). A possible explanation

of such differences in particle size between pure and Al-doped suspensions is that the presenceof Al could increase the rapidity of Fe hydrolysis (Singh and Kodama, 1994) thus favoring the

formation of ferrihydrite as a more metastable phase. Alternation explanation is that Al

suppressed ferrihydrite dissolution and transformation, however it is not in agreement with

Jentzsch and Penn (2006) who found that doping of small amounts of Al (


36/187

22

the corresponding nano-goethite crystallite size (Fig. 2-7A), which suggests aggregation of nano-

goethite crystallites into larger polycrystalline goethite particles. This interpretation is in

agreement with particle size distribution (PSD) analysis of the DLS data (Fig. 2-7B). For time 0

and 2 days, PSD exhibited narrow size distributions within 10 nm; however, as the suspensions

aged, PSD broadened and presented a tail (right-skewed) of large particles with sizes up to 100

nm. If there was a constant particle growth rate (dissolution-precipitation mechanism), then PSD

would only shift laterally on the size axis (Li et al., 2005), however, PSD broadening signified

particle aggregation.

TEM and DLS results are in good agreement since TEM images showed small roundparticles (< 5 nm) for both 0% and 2% Al suspensions at time 0 days (Figs. 2-8A, B and 2-9A,

B). Three populations of nanoparticles of various shapes and sizes (5-100 nm range) were

present in Al-free suspensions aged for 54 days at 50° C (Figs. 2-8C and 2-9C): (1) round

particles less than 10 nm in diameter; (2) thin goethite needles about 50 nm in length and several

nm wide; (3) large goethite crystals (prisms) 50-100 nm in length and 10-20 nm wide. In the

presence of 2% Al (Figures 2-8D and 2-9D), ferrihydrite (round) particles were the most

abundant, yet goethite needles and crystals were also present. It is important to note that TEM

images showed small needle-like (


37/187

23

crystals (Fig. 2-10B) contained single-crystal lattice fringes and showed a pattern of spots in the

SAED image; thus suggesting it was formed by monomeric addition (Ostwald ripening). In

contrast, goethite needles consisted of smaller grains with different lattice orientation (Fig. 2-

10C) and a diffraction pattern showing characteristic polycrystalline rings, which suggests

goethite needles were formed by aggregation of round nano-goethite particles. Suspensions with

2% Al showed similar features.

Based on our (HR)TEM and DLS results, we hypothesize that at first, some of the

primary (round) particles of ferrihydrite were transformed to nano-goethite round particles by a

dissolution-precipitation mechanism that probably occurred at the dialysis stage (for non-dopedFe-hydroxide suspensions) or during the first 2 days of aging. Then, nano-goethite particles

started to grow either by aggregation to form goethite needles or by monomeric addition to form

single-crystal goethite (Fig. 2-11). Since nano-goethite aggregation took place after ferrihydrite

to nano-goethite transformation of small round particles, we concluded Ostwald ripening was the

dominant mechanism controlling the rate of this reaction. Gradual increase with time of MCR-

derived “goethite-like” component supports Ostwald ripening mechanism. The presence of Al

does not change the mechanism of goethite crystallization, but retards formation of goethite

needles and prisms as evidenced from TEM data

2.4.2. Rate constants for ferrihydrite-goethite transformation

Ostwald ripening (dissolution-precipitation) is usually described by a pseudo-first order

equation when only the formation of a crystalline end-product (i.e., goethite) or the decrease in

the amorphous end-member (i.e., ferrihydrite) is considered; the concentration of additional

chemical species taking part in the reaction are not considered (Lasaga, 1998). This reaction can

be written as Ferrihydrite → Goethite, or in exponential form:

)exp1(*][][ 0kt GoethiteGoethite −−= (eq.1)


38/187

24

where the concentration of goethite, [Goethite], is expressed in terms of XRD (110) peak

areas, intensities of OH-bending (790 cm-1 and 890 cm-1) vibrations, and MCR scores as

indicators of the presence of goethite in Fe-hydroxide and Al-doped Fe-hydroxide mixtures at a

given time t (sec); [Goethite]0, is the initial concentration of goethite in the mineral mixture, and

k (sec-1) is the first-order rate constant. Both [Goethite]0 and rate constant were determined by

fitting experimental data plotted versus reaction time to equation 1.

XRD peak areas, the intensity of OH-bending vibrations, and MCR scores increased with

time for Al-free and Al-doped Fe-hydroxide suspensions (Fig. 2-12). Corresponding values for

2% Al were lower compared to 0% Al, with the largest difference resulting from the intensitiesof OH bending vibrations. The rate constants obtained from different methods were consistent

within experimental error for both 0% and 2% Al suspensions (Fig. 2-13). The presence of 2%

Al decreased the rate constants determined from OH-stretching and OH-bending vibrations by

~47-50%. For example, MCR derived rate constant decreased from (7.4 ± 1.1)*10-7 (0%Al) to

(4.2 ± 0.4)*10-7 s-1 (2%Al). However, the values of the rate constants determined from XRD

peak areas were the same for both 0% and 2% Al suspensions. This discrepancy can be

explained by a large experimental error due to lower sensitivity of XRD to structural changes in

poorly-crystalline mixtures (Burleson and Penn, 2006).

The fact that both the rate of goethite crystallization and the stability of colloidal

suspension are influenced by factors such as pH, temperature, [Fe], ionic strength of the solution

and the presence of foreign ions, makes it difficult to obtain a consistent picture from already

available experimental work (Fig. 2-14). Many studies on ferrihydrite to goethite transformation

were conducted under very high alkaline (pH 10-13) conditions (Nagano et al., 1992; Nagano et

al., 1994; Shaw et al., 2005), where the rate of goethite transformation is relatively fast compared

to environmental pH values of 4-9 (Fig. 2-14). Even studies conducted at circumneutral pH (Fig.


39/187

25

2-14) show that the rate constants highly depend on experimental parameters. For example, the

presence of Fe2+ in solution greatly accelerates the rate of goethite crystallization (Yee et al.,

2006). Our values of rate constants are similar to the values of Ford et al. (2002) and Fischer and

Schwertmann (1975), both of which derived their rate constants from ammonium oxalate

extractions. What we can deduce from published rate constants is that both increases in pH and

temperature result in a faster rate of transformation (Fig. 2-14). Differences in the numerical

value of the rate constants obtained by various methods (XRD, infrared, colorimerty, oxalate

extraction) may not be as significant as differences attributed to synthesis procedures or

environmental conditions (i.e., pH, temperature, [Fe], etc.).

2.5. Conclusions

We succesfully applied MCR analysis of overlapped IR OH-stretching bands to study

ferrihydrite-goethite transformations and to quantify rate constants in poorly-crystalline nano-

colloidal suspensions. To our knowledge, this is the first time the rate constant for ferrihydrite

crystallization in the presence of low concentrations of Al has been quantified. The presence of

2% Al decreased the rate constants determined from analyses of infrared OH-stretching and OH-

bending vibrations by ~45%, while TEM data showed that in the presence of Al the growth of

larger goethite particles was hindered. Our results also indicated dissolution-precipitation was the

dominant mechanism in the reaction of ferrihydrite transformation to nano-goethite. Further

growth of nano-goethite crystals took

Documents

Bazilevskaya PhD Dissertation 2 December2009