Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Investigating mineral dissolution kinetics by Flow-Through Time-Resolved Analysis (FT-TRA) De Baere, Bart 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_february_debaere_bart.pdf [ 62.71MB ]
Metadata
JSON: 24-1.0220867.json
JSON-LD: 24-1.0220867-ld.json
RDF/XML (Pretty): 24-1.0220867-rdf.xml
RDF/JSON: 24-1.0220867-rdf.json
Turtle: 24-1.0220867-turtle.txt
N-Triples: 24-1.0220867-rdf-ntriples.txt
Original Record: 24-1.0220867-source.json
Full Text
24-1.0220867-fulltext.txt
Citation
24-1.0220867.ris

Full Text

INVESTIGATING MINERAL DISSOLUTION KINETICS BY FLOW-THROUGH TIME-RESOLVED ANALYSIS (FT-TRA) by  Bart De Baere  MSc., The University of Southampton, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Oceanography)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2015  © Bart De Baere, 2015   ii Abstract  This thesis evaluates the applicability of flow-through time-resolved analysis (FT-TRA) to address problems ranging from determining mineral dissolution kinetics and dissolution regimes, to unraveling the elemental composition of multiple mineral phases in microfossils, to predicting drainage chemistry from mine waste. FT-TRA consists of a gradient pump, which continuously passes eluent of fixed or varying composition through a small flow-through reactor containing a small amount of solid sample. The effluent composition is then analyzed online using an inductively coupled plasma mass spectrometer in time-resolved mode, or is collected in a fraction collector for subsequent offline analysis, depending on the goal of the experiment.  It is found that FT-TRA is well suited to study mineral dissolution kinetics. Using forsterite as a case study, it is shown that FT-TRA can be used to rapidly determine mineral dissolution rate parameters. The high temporal resolution data generated by FT-TRA documents in detail, and in real-time the gradual formation of surface leached layers as well as sporadic and abrupt exfoliation events occurring during dissolution. A range in eluent residence times in the reactor can be applied by controlling the eluent flow with the gradient pump, allowing for the empirical determination of the dissolution regime (surface- or transport-controlled), which must be established prior to interpreting mineral dissolution rates measured during the experiment. When combining FT-TRA data with pore scale modeling, dissolution rate constants can still be determined, even when the dissolution experiment is conducted under transport-controlled conditions.  The added value of this continuous eluent flow system for assessing the leaching behavior of mine waste is also evaluated. The ability to carry out experiments in a relatively short time   iii period provides a new means to elucidate the mechanism and conditions resulting in the release of toxic metals from mine waste during weathering.  Finally, using insight gained from studying mineral dissolution kinetics, the premise on which FT-TRA was used to distinguish the elemental ratios of different biogenic mineral phases in microfossils for paleoceanographic reconstruction is re-evaluated.   iv Preface Contributions of Authors 1) I carried out the majority of the hardware, software development presented in chapter 2. Ferenc Tarnok at Neptune Research & Development Inc. provided feedback regarding hardware development. Paul De Baere designed and built the pH sensor buffer amplifier.  2) A version of chapter 3 has been published as: De Baere, B., François, R., Mayer, K.U.: Measuring mineral dissolution kinetics using on-line flow-through time resolved analysis (FT-TRA): an exploratory study with forsterite, Chemical Geology 413, 107 – 118, 2015, doi: 10.1016/j.chemgeo.2015.08.024. I carried out all experiments, and analyzed all the data. I wrote most of the manuscript in close collaboration with R.F. K.U.M. provided feedback regarding data analysis and interpretation.  3) A version of chapter 4 has been submitted as: De Baere, B., Molins, S., Mayer, K.U.: François, R., Empirical determination of mineral dissolution regimes using flow-through time-resolved analysis (FT-TRA). I carried out all experimental work and analyzed the experimental data. S.M. carried out the pore scale modeling, and provided me with the results. I interpreted the modeling results in close collaboration with K.U.M. and S.M. I wrote the manuscript in close collaboration with all co-authors.  4) A version of chapter 5 will be submitted as: De Baere, B., François, R and Kohfeld, K.: Evaluating flow-through time-resolved analysis (FT-TRA) as a tool to quantify ontogenetic molar ratios in foraminiferal tests. R.F. partly contributed to the research design. I carried out all experiments and analyzed all data. I wrote most of the manuscript in close collaboration with all co-authors.   v 5) The exploratory study presented in chapter 6 represents the result of a newly established collaboration with Lorax Environmental Services, Ltd. The experimental design was developed in close collaboration with Bruce Mattson at Lorax Environmental Services, Ltd. I carried out all experiments, and analyzed all data. Data interpretation was carried out in close collaboration with K.U.M and R.F.    vi Table of Contents  Abstract ............................................................................................................................................. ii	Preface .............................................................................................................................................. iv	Table of Contents ............................................................................................................................ vi	List of Tables ................................................................................................................................ xiii	List of Figures ................................................................................................................................ xvi	Acknowledgements .................................................................................................................... xxxii	Dedication .................................................................................................................................. xxxiv	Chapter 1: Introduction .................................................................................................................. 1	1.1	 Mineral dissolution ............................................................................................................... 1	1.2	 Mineral dissolution kinetics .................................................................................................. 1	1.3	 Experimental approaches used to study mineral dissolution ................................................ 4	1.3.1	 Batch reactor experiments .............................................................................................. 4	1.3.2	 Continually stirred flow-through reactor experiments ................................................... 5	1.3.3	 Surface topographic measurements ............................................................................... 7	1.3.3.1	 Atomic force microscopy (AFM) ........................................................................... 7	1.3.3.2	 Vertical Scanning Interferometry (VSI) ................................................................. 9	1.4	 Flow-through time-resolved analysis (FT-TRA) ................................................................ 10	1.5	 Thesis objectives ................................................................................................................. 11	Chapter 2: Flow-through time-resolved analysis (FT-TRA) module design ............................ 14	2.1	 Introduction ......................................................................................................................... 14	2.2	 FT-TRA experiments .......................................................................................................... 14	  vii 2.3	 FT-TRA leaching module development ............................................................................. 15	2.3.1	 Gradient pump ............................................................................................................. 17	2.3.2	 Flow-through reactors .................................................................................................. 18	2.3.3	 Solenoid valves ............................................................................................................ 20	2.3.4	 In-line pH measurement ............................................................................................... 23	2.3.5	 Fraction collection ....................................................................................................... 28	2.4	 Standards ............................................................................................................................. 29	2.5	 Eluent flow rate precision ................................................................................................... 32	2.6	 Time-resolved analysis ICP-MS ......................................................................................... 33	2.7	 General operating procedure ............................................................................................... 34	2.7.1	 Online FT-TRA ............................................................................................................ 36	2.7.2	 Offline FT-TRA ........................................................................................................... 38	Chapter 3: Measuring mineral dissolution kinetics using online flow-through time-resolved analysis (FT-TRA): an exploratory study with forsterite ........................................... 39	3.1	 Introduction ......................................................................................................................... 39	3.2	 Background ......................................................................................................................... 40	3.3	 Materials and methods ........................................................................................................ 43	3.3.1	 Materials ...................................................................................................................... 43	3.3.2	 Flow-through dissolution module ................................................................................ 44	3.3.3	 Measurements of magnesium and silicon concentration in the eluent by on-line quadrupole ICP-MS ................................................................................................................. 47	3.3.4	 Flow-through dissolution experiment design ............................................................... 48	3.3.4.1	 Dissolution rate parameters ................................................................................... 48	  viii 3.3.4.2	 Dissolution under transient conditions and replication ......................................... 49	3.3.5	 Calculation of surface normalized dissolution rates and dissolution rate parameters (k!! ,n!!) .............................................................................................................. 50	3.4	 Results and discussion ........................................................................................................ 51	3.4.1	 Determination of forsterite dissolution parameters ...................................................... 51	3.4.1.1	 Forsterite dissolution regime ................................................................................. 51	3.4.1.2	 Forsterite dissolution under acidic conditions ...................................................... 54	3.4.1.3	 Forsterite dissolution in DIW (pH 5.6) ................................................................. 60	3.4.1.4	 General applicability of FT-TRA to estimate mineral dissolution parameters ..... 62	3.4.2	 Replicate analyses and dissolution under transient conditions .................................... 64	3.4.2.1	 Reproducibility of forsterite dissolution under identical conditions ..................... 64	3.4.2.2	 Forsterite dissolution stoichiometry under transient eluent conditions ................ 68	3.4.2.2.1	 Monitoring the surface leached layer thickness during transient eluent conditions… ..................................................................................................................... 68	3.4.2.2.2	 Sporadic exfoliation of the Si-rich layer ........................................................ 72	3.5	 Summary and future prospects ............................................................................................ 74	Chapter 4: Empirical determination of mineral dissolution regimes using flow-through time-resolved analysis (FT-TRA) ................................................................................................. 77	4.1	 Introduction ......................................................................................................................... 77	4.1.1	 Dissolution regimes ..................................................................................................... 77	4.1.2	 Previous approaches to determine dissolution regimes ............................................... 80	4.2	 Materials and methods ........................................................................................................ 81	4.2.1	 Sample origin, preparation and surface area measurement ......................................... 81	  ix 4.2.2	 FT-TRA module ........................................................................................................... 83	4.2.3	 Experimental design to establish dissolution regimes ................................................. 86	4.2.4	 Rate equations .............................................................................................................. 89	4.2.5	 Pore scale modeling ..................................................................................................... 89	4.3	 Results and discussion ........................................................................................................ 92	4.3.1	 Forsterite ...................................................................................................................... 92	4.3.2	 Calcite .......................................................................................................................... 94	4.3.3	 Added value to FT-TRA ............................................................................................ 102	4.4	 Conclusions and future research direction ........................................................................ 103	Chapter 5: Evaluating FT-TRA dissolution as a tool to quantify ontogenetic Mg/Ca molar ratios in foraminiferal tests .............................................................................................. 105	5.1	 Introduction ....................................................................................................................... 105	5.1.1	 Brief introduction of the use of foraminiferal elemental ratios in paleoceanography .................................................................................................................. 105	5.1.2	 Foraminifera calcification, partial dissolution ........................................................... 106	5.1.3	 FT-TRA technique validation .................................................................................... 110	5.1.4	 Addressing FT-TRA calcite dissolution kinetics ....................................................... 112	5.2	 Methodology ..................................................................................................................... 114	5.2.1	 FT-TRA technique development ............................................................................... 114	5.2.2	 Sample origin, preparation and surface area characterization ................................... 115	5.2.3	 FT-TRA dissolution experiments .............................................................................. 117	5.2.4	 Measurement and calculation of [Ca, Mg, Sr] concentrations, dissolution rates ...... 119	5.2.5	 Modeling approach .................................................................................................... 120	  x 5.3	 Results ............................................................................................................................... 122	5.3.1	 Individual grain dissolution experiments (samples 1 – 5) ......................................... 122	5.3.2	 Powder experiments (samples 6 – 8) ......................................................................... 126	5.4	 Discussion ......................................................................................................................... 134	5.4.1	 Mineral grain dissolution experiments (samples 1 – 5) ............................................. 134	5.4.2	 Powder experiments (samples 6 – 8) ......................................................................... 135	5.4.3	 Previously reported evidence challenging the FT-TRA premise ............................... 137	5.4.4	 Alternative data interpretation and a possible path forward ...................................... 139	5.5	 Conclusions and future research directions ...................................................................... 142	Chapter 6: Measuring metal release rates from mine waste rock: a preliminary case study .............................................................................................................................................. 144	6.1	 Introduction ....................................................................................................................... 144	6.1.1	 Challenges associated with kinetic testing ................................................................. 145	6.1.2	 Study setting .............................................................................................................. 147	6.1.3	 Study rationale ........................................................................................................... 147	6.1.4	 Study objectives ......................................................................................................... 148	6.2	 Materials and methods ...................................................................................................... 149	6.2.1	 Sample characterization ............................................................................................. 149	6.2.2	 Trickle leach experiments .......................................................................................... 153	6.2.3	 Mixed Flow Reactor (MFR) experimental design ..................................................... 154	6.2.4	 MFR experiments ....................................................................................................... 156	6.2.5	 MFR effluent elemental and sulphate measurements ................................................ 157	6.3	 Results ............................................................................................................................... 158	  xi 6.3.1	 Trickle leach experiments .......................................................................................... 158	6.3.1.1	 pH ........................................................................................................................ 158	6.3.1.2	 Elemental and sulphate concentrations ............................................................... 158	6.3.2	 MFR experiments ....................................................................................................... 163	6.3.2.1	 pH ........................................................................................................................ 163	6.3.2.2	 Elemental and sulphate concentrations ............................................................... 165	6.3.2.2.1	 DIW experiment ........................................................................................... 165	6.3.2.2.2	 Acidic experiment ........................................................................................ 170	6.4	 Discussion ......................................................................................................................... 174	6.4.1	 pH ............................................................................................................................... 174	6.4.1.1	 Trickle leach experiment ..................................................................................... 174	6.4.1.2	 MFR experiments ................................................................................................ 176	6.4.2	 Metals (Al, Mg, Fe, K, Na, Ca, Ba), silicon and sulphate release ............................. 176	6.4.2.1	 Trickle leach experiment ..................................................................................... 176	6.4.2.2	 Comparison between DIW trickle leach and MFR experiments ........................ 183	6.4.2.3	 Comparison between DIW, acidic MFR experiments ........................................ 186	6.4.2.4	 MFR sulphate release, comparison to trickle leach experiments ........................ 190	6.5	 Environmental implications .............................................................................................. 193	6.6	 Key benefits associated with MFR and future work ......................................................... 193	Chapter 7: Conclusions and future outlook .............................................................................. 195	7.1	 Synthesis of thesis accomplishments ................................................................................ 195	7.2	 Future research directions ................................................................................................. 198	7.2.1	 Mineral dissolution kinetics studies ........................................................................... 198	  xii 7.2.2	 Paleoceanographic studies ......................................................................................... 200	7.2.3	 Environmental impact studies .................................................................................... 201	Bibliography ................................................................................................................................. 203	Appendices .................................................................................................................................... 220	Appendix A: chapter 3 ............................................................................................................... 220	Appendix B: chapter 4 ............................................................................................................... 233	Appendix C: chapter 5 ............................................................................................................... 236	Appendix D: chapter 6 ............................................................................................................... 258	   xiii List of Tables  Table 2.1 Overview of key components required for assembling a flow-through dissolution sample introduction module. ....................................................................................................... 16	Table 2.2 Overview of pump flow rate Q measurements. Measured flow rates (n experiments) provided in third column, where 1error represents the 95% confidence interval.  The accuracy of the measured flow rate relative to the nominal flow rate is provided as a percentage in the fourth column, and lies within 5.50 % across all flow-rates. Best flow-rate precision is achieved in the mid rang and decreases at high and low flow rates. ................................................................... 33	Table 3.1 Weight (g) and surface area (m2) of forsterite used in the “dissolution regime” experiment (sample 1), the “rate parameters” experiment (sample 2, 3) and the “replication” and “transient” experiments (sample 4 – 8). Surface areas are calculated from the sample weight and BET surface measurements (sample 1, 4 – 8: 2.57 × 10-! m2 g-1; sample 2, 3: 3.39 × 10-! m2 g-1). Sample loss is estimated by integrating the total moles of Mg released during each experiment. .................................................................................................................................. 49	Table 3.2 Dissolution rate constant k!! and reaction order n!! quantified in this study (based on [Mg] and [Si]), compared to other studies compiled by Rimstidt et al., 2012 (pH ≤ 6, log r BET surface area normalized Mg, Si data) ......................................................................................... 58	Table 4.1 Overview of constant incoming eluent pH experiments (indicated in bold) conducted as part of this study. Effluent [Mg, Si] for forsterite and [Ca] concentrations were monitored using time-resolved analysis until steady state dissolution was achieved. ........................................... 88	  xiv Table 4.2 List of aqueous complexation reactions and equilibrium constants sourced from the EQ3/6 database (Wolery and Daveler, 1992). ............................................................................ 92	Table 5.1 Eluents used in flow-through dissolution experiments (nr = not reported) as reported in FT-TRA studies aiming to isolate ontogenetic carbonate elemental composition. Nitric acid pH represents eluent pH used to dissolve the foraminiferal test. Hydroxylamine has been used to dissolve “high–Mg calcite and oxide coatings” (Haley and Klinkhammer, 2002); DTPA (diethylene triamine pentaacetic acid) has been used to dissolve sedimentary barite (Lea and Boyle, 1993), relevant for studying the Rare Earth Elements (Haley and Klinkhammer, 2002). Shaded areas represent eluent absence from study. .................................................................. 110	Table 5.2 Overview of samples (Arag. = aragonite; Cal. = calcite). ........................................ 117	Table 5.3 Measured Mg/Ca, Sr/Ca molar ratios in samples 6 – 8 (aragonite, calcite powder and powder mixture sample respectively) as well as calculated mixture molar ratios based on end-member composition (Arag. = aragonite; Cal. = calcite). ........................................................ 129	Table 6.1 Elemental content (expressed in % and ppm) of waste rock sample, obtained using aqua regia digestion (1st column) and a more extensive four-acid procedure (2nd column). Four-acid digestion range represents 2 separate analyses (n = 2). ............................................................ 150	Table 6.2 Sample composition expressed in percentage based on x-ray diffraction. .............. 152	Table 6.3 Absolute error, expressed in ppb (relative error on concentrations, expressed in %).157	Table 6.4 pH probe calibration data DIW MFR experiment, where pH = (slope × mV)+intercept. The coefficient of determination (R2) was always greater than 0.99. Note maximum shift in calibration intercept occurred between Oct. 22nd and Oct. 29th 2013 ........................... 164	Table 6.5 Integrated moles in four-acid digested sample (first column), trickle leach experiment (second column) and % leached during trickle leach experiment (third column). ................... 181	  xv Table 6.6 Total moles, calculated using four-acid digestion data (first column), DIW MFR experiment (second column), % leached during DIW MFR experiment (third column), acidic MFR experiment (fourth column) and % leached during acidic MFR experiment (fifth column). .................................................................................................................................... 186	   xvi List of Figures  Figure 1.1 Concentration C versus distance r from a crystal surface for three rate-controlling regimes: (A) transport control; (B) surface control; (C) mixed control. C!" = saturation concentration; C!"#$ = bulk aqueous phase concentration. Adapted from Berner, 1978. .................. 3	Figure 1.2 Conceptual representation of a mixed flow reactor experiment (Panel A) and a fluidized bed reactor experiment (Panel B, adapted from Chou and Wollast, 1984). In a MFR experiment, a single pump (P1) generates an identical fluid flow for both the inflow of the eluent and the outflow of the effluent. In a fluidized bed reactor, P1 generates the fluid flow required to keep the particles in suspension, P2 generates the fluid flow responsible for the eluent input and effluent output. ................................................................................................................................... 7	Figure 1.3 Flow-chart indicating thesis outline. Introductory and concluding chapters (chapters 1 and 7) not shown. .......................................................................................................................... 13	Figure 2.1 Diagram of the flow-through module developed (small internal volume flow-cell version shown). A Dionex gradient pump draws upon up to four eluent bottles after which the eluent flows to a specific flow-through reactor (depicted as reaction vessel – a total of 6 individual sample positions were present). Here, the sample gradually dissolves, and the dissolved sample (effluent) flows to a quadrupole ICP-MS (for online analysis, chapters 3, 4 and 5) or a fraction collector (for offline fraction collection at user-defined time-intervals, chapter 6). When using the online-set-up, the effluent merges with a constant flow of internal standard supplied by a separate, isocratic pump (used for temporal instrumental drift correction). Optional in-line pH sensors can be installed (as indicated here) if a measurable (pH sensor precision ± 0.1 pH units) pH difference is to be monitored. ............................................................................................................................ 15	  xvii Figure 2.2 Time-dependent eluent composition (FT-TRA leaching protocol) as programmed using Dionex Chromeleon™ software. Top graph shows time-dependent eluent bottle composition (%) versus time (min). Lower panel shows table used to program leaching dissolution sequence (% of eluent A is automatically calculated in software, based on user input of % eluent B, % eluent C and % eluent D). In this sequence, the sample is initially exposed to 40 min of 100% eluent A (yellow), after which eluent C is gradually introduced (40 min < t < 70 min) before it reaches a maximum of 10 % at t = 70 min. This eluent proportion is held constant between 70 min < t < 100 min. At 100 min < t < 120, the proportion of eluent C increases to 100%, at which level it remains constant for 30 min. At 150 min < t < 160, the eluent composition gradually returns to 100 % eluent A. Throughout this experiment, flow-rates are held constant at 0.70 ±0.0367  mL min-1 (error indicates the 95% confidence interval). Further details on flow rate precision are provided in section 2.5. © 2011 Thermo Fisher Scientific Inc., by permission. ............................. 18	Figure 2.3 Flow-through cells used during this thesis. Panel A: cross-section of an off-the-shelf 13 mm syringe filter. Blue arrows indicate eluent flow through filter. Blue rectangles represent open slits through which the eluent moves upwards. Syringe filters are held in place using luer fittings connected to solenoid valves, from where the effluent flows via tubing to the ICP-MS for online time-resolved analysis. Panel B: 50 mL Teflon® flow-though cell as used in chapter 6. Eluent enters the bottom of the flow-cell, exists at the top of the flow-cell. Aluminum support bracket ensures a tight seal. ............................................................................................................. 19	Figure 2.4 Flow-through cell volumes relevant to this thesis. Panel A: 25 µL volume (shown as y) relevant to chapters 3, 4, 5. Panel B: 50 µL volume (shown as x) relevant to chapters 4 and 5.20	  xviii Figure 2.5 Two parallel manifolds, each holding six separate solenoid valves (numbered on image). Flow-through reactors (syringe filters, as indicated) are located in between two parallel solenoid valves (upper and lower solenoid valves, respectively). ................................................... 21	Figure 2.6 CoolDrive® circuit board used to operate the solenoid valves. As each board controls 5 valves, a total of 3 boards were required (15 output connections) to control 2 parallel solenoid valves (12 valves in total). ............................................................................................................... 22	Figure 2.7 LabView control panel allowing the user to open and close two parallel solenoid valves (2 parallel manifolds operate 6 flow-cell positions), each holding a flow-through reaction cell. Here, two parallel valves named ‘Valve 1’ have been opened, routing the eluent to flow via the flow-through reaction cell located in between both valves. ....................................................... 23	Figure 2.8 Hardware setup used to record in-line effluent pH. The pH electrode is exposed to the effluent via a flow-cell (shown in gray) holding the pH electrode. The analog pH electrode signal is amplified before being sent to an analog input associated with the universal chromatography interface (UCI). There, the analog signal is digitized, and sent to a computer via USB output. ..... 24	Figure 2.9 Schematics showing pH sensor buffer amplifier. A: power supply (accomplished using 9 V block battery), B: offset compensator. ...................................................................................... 26	Figure 2.10 Electrode data obtained during pH calibration. Plateaus represent time intervals during which electrode was exposed to pH buffer solutions (pH 4, 7 and 10 respectively). .......... 27	Figure 2.11 Linear pH calibration curve (coefficient of determination = 1) where the electrode signal (average mV, x-axis) is plotted versus pH buffer solution (y-axis). ..................................... 28	Figure 2.12 Time-dependent eluent composition during a typical standard run. Here, eluent bottle D represents the standard matrix (1 % HNO3) and eluent bottle B represents the high standard solution in a 1 % HNO3 matrix. Note (1) standard concentrations increase over time and (2) eluent   xix flow rate is held constant at 0.70±  0.0367  mL min-1 (error indicated the 95% confidence interval). Further details on flow rate precision are provided in section 2.5. © 2011 Thermo Fisher Scientific Inc., by permission. .......................................................................................................... 30	Figure 2.13 Typical standard run (here, internal standard normalized 24Mg) showing time-resolved plateaus of known high standard : matrix proportions. Percentages shown above plateaus denote proportion of high standard. ................................................................................................. 31	Figure 2.14 Standard calibration curve (here: 24Mg). Internal standard normalized values shown represent averages obtained from plateaus shown in Fig. 2.13. ...................................................... 31	Figure 2.15 Flow-through module developed for online FT-TRA. ................................................ 35	Figure 2.16 Recommended protocol for online FT-TRA. .............................................................. 36	Figure 3.1 SEM image at low magnification (× 180) of the forsterite sample used in this study, after acetone cleaning. ..................................................................................................................... 44	Figure 3.2 Schematic of the flow-through dissolution module. E1 – E5 represent eluent bottles, where E1 – E4 are used to generate eluents or standards and E5 is used to continuously supply an internal standard. Eluent flows top to bottom of figure. Computer controlled solenoid valves with 6 sample positions allow for the efficient analysis of blanks, samples over the course of a day. When running standards, the flow-through cell is bypassed (as indicated by the red line). This flow-through module allows for time-resolved control of incoming eluent composition, and a constant, reproducible flow rate through the flow-through cell, for time-resolved analysis of the effluent stream by quadrupole ICP-MS. .......................................................................................... 46	Figure 3.3 Average steady state Mg (A) and Si (B) concentrations (sample 1) versus eluent residence time (calculated by dividing flow cell volume (25 µL) by varying eluent flow rates). ... 53	  xx Figure 3.4 (A) [Mg] (5-point running average in black), [Si] (5-point running average in gray) and Mg/Si molar ratio (red) in the eluent resulting from the dissolution of sample 2. Forsterite was subjected to acidic solutions (DIW and HNO3) of decreasing pH (5.6, 4.0, 3.5, 3.0, 2.3) maintained until steady state dissolution was reached or approached. Time step for Mg and Si measurements by ICP-MS is 1.8 seconds, hence the 5-point average is shown to represent the eluent reactor renewal time. The dashed lines indicate the stoichiometric Mg / Si ratio of forsterite (1.81 ± 0.07). (B) Forsterite dissolution rates estimated from average [Mg] (filled circles), and average [Si] (open diamonds) concentration plateaus at each pH. [Mg] were divided by 1.81 to take into account forsterite stoichiometry. Average [Mg, Si] were multiplied by the effluent flow rate (0.70 ± 0.01 ml min-1) and divided by BET surface area (Table 3.1). Weighted (according to the number of data points collected at each plateau) linear regression provides the reaction order (slope) and rate constant (intercept at pH = 0). 95 % confidence intervals shown are based on the weighted linear regression. .............................................................................................................. 56	Figure 3.5 Logarithm of the surface normalized forsterite dissolution rate (moles m-2 s-1) based on the Mg (A) and Si (B) release rates versus pH (sample 2). The dashed lines represent the 95 % confidence interval based on the uncertainties of the rate parameters (slope and intercept) obtained from the weighted linear regression (Table 3.2) and extrapolated to pH 1 – 6. Results from previous studies compiled in Rimstidt et al. (2012) are shown for comparison. Color-coded lines are obtained by linear regressions on Mg, Si data obtained at pH ≤ 6. Gray circles indicate studies yielding reaction orders very different from 0.5 (Luce et al., 1972; Grandstaff, 1986; Golubev et al., 2005; Table 3.2). ........................................................................................................................ 59	Figure 3.6 Concentrations of [Mg] (5-point running average in black), [Si] (5-point running average in gray) and Mg/Si molar ratio (red) in the eluent resulting from the dissolution of sample   xxi 3 in DIW. The time step for Mg and Si measurements by ICP-MS is 1.8 seconds, hence the 5-point average is shown to represent the eluent reactor renewal time. The dashed lines indicate the stoichiometric Mg / Si ratio of forsterite (1.81 ± 0.07). ................................................................... 61	Figure 3.7 (A) Concentrations of [Mg] (5-point running average in black), [Si] (5-point running average in gray) and Mg/Si molar ratio (red) measured in the effluent resulting from the dissolution of sample 6. The dashed lines indicate the stoichiometric Mg / Si ratio of forsterite (1.81 ± 0.07). Figures for samples 4, 5, 7 and 8 are supplied in Appendix A: Fig. A.1. (B) Changes in incoming eluent pH (≈ pH in flow-through reactor, as changes in pH resulting from forsterite dissolution in the flow cell are below detection limits). .................................................................. 66	Figure 3.8 Forsterite dissolution rates under varying pH estimated from time-resolved Mg (upper panel A, divided by 1.81 to take into account forsterite stoichiometry) and Si (lower panel B) concentration measured by on-line ICP-MS, multiplied by the eluent flow rate (0.70 ± 0.01 mL min-1) and divided by the BET surface area for sample 4 – 8 (Table 3.1). Time-resolved eluent pH provided in Fig. 3.7B. ...................................................................................................................... 67	Figure 3.9 Changes in the thickness of the Mg leached surface layer (solid red line) in sample 6 (thicknesses for samples 4, 5, 7, 8 are reported in Appendix A: Fig. A.2). The dashed red lines represent range in Mg – Si leached layer thickness calculated using the 95% confidence interval of the forsterite stoichiometry (Mg/Si = 1.81 ± 0.07). Mg/Si molar ratio (calculated using the 5-point running [Mg, Si] average) is shown in black. Time resolved eluent pH is reported in Fig. 3.7B.71	Figure 3.10 (A) Exfoliation event observed in sample 6 as an abrupt increase in [Si]; (B) Same event on an expanded time scale showing the temporal sequence of changes in [Si] (y-axis on right) preceding a smaller [Mg] peak (y-axis on left). ..................................................................... 74	  xxii Figure 4.1 SEM imagery associated with mineral samples used in this study. (A): forsterite powder used in this study (63 – 150 µm size fraction); (B): individual calcite rhomb as typically used in this study. ............................................................................................................................. 83	Figure 4.2 Schematic of the flow-through dissolution module. E1 – E5 represent eluent bottles, where E1 – E4 are used to generate eluents or standards (using an advanced gradient pump, indicted as AGP), and E5 is used to continuously supply an internal standard (here 115In, using an isocratic pump). This flow-through module allows for time-resolved control on incoming eluent composition (here, generating an eluent pH range) by mixing different proportion of DIW with concentrated acid or standard from bottle E1-E4, providing a constant, reproducible flow rate through the flow-through cell, and enabling time-resolved analysis of the effluent stream. A total of 6 flow-through cell positions are available by using 2 parallel computer-controlled solenoid valves, allowing for efficiently measuring blanks and multiple samples. When running standards, the solenoid valve holding the flow-through cells is bypassed, as indicated in red. ....................... 85	Figure 4.3 Conceptual representation of proposed FT-TRA approach to determine dissolution regime. If mineral dissolution is surface-controlled – and conditions are kept far-from-equilibrium – dissolution rates remain constant under varying flow rates (Panel A, slight curvature due to a minor saturation (IAP/K!") effect) and steady state dissolved species concentration (here: [Ca]) increases proportionally with eluent residence time (Panel B, similar slight curvature due to a minor saturation (IAP/K!") effect). In contrast, if dissolution rates vary according to flow rate (Panel C), transport processes (DBL formation) affect c!"#$%&', thereby lowering the dissolution rate. Here, steady state dissolved species concentration (here: [Ca]) will deviate from linearity with eluent residence time (Panel D). .............................................................................................. 88	  xxiii Figure 4.4 Pore-scale simulation domain showing the calcite grain placed at the inlet end of the reactor. Contours of the magnitude of the fluid velocity in the reactor are shown on a slice that divides the domain in half. Fluid flows from left to right at the slowest experimental rate used (1.67 × 10-! mL s-1), which is prescribed using a uniform fluid velocity distribution at the inlet face. .................................................................................................................................................. 91	Figure 4.5 Panels A and B show [Mg]steady state and [Si]steady state concentrations respectively (calculated using average [Mg, Si]selected obtained from applicable data as illustrated using data collected as shown in Panel C); error falls within symbol size; R2 = 0.999 for both [Mg] as well as [Si] versus eluent residence time. Panel C: Time-resolved [Mg, Si] concentration results collected at an eluent flow rate of 5 × 10-! mL s-! at eluent pH 2.3 illustrates data that is typically collected measured during forsterite powder dissolution experiments. Selected [Mg, Si] concentrations representing steady state conditions are shown as black filled circles for Mg and red filled circles for Si. Data recorded prior to [Mg, Si]selected plateaus are shown as gray circles. Panel D: BET surface area normalized Mg and Si based forsterite dissolution rates versus eluent flow rate (note logarithmic x- and y-axis). [Mg]selected concentrations were divided by 1.81 to take into account forsterite stoichiometry. Average dissolution rate shown as solid black line, 95 % CI based on 2σ shown as dotted black line. .............................................................................................................. 93	Figure 4.6 Panels A, B and C show [Ca]steady state concentrations collected under eluent pH of 2.3, 3.3 and 4.0 respectively plotted versus eluent residence time. Error falls within symbol size. [Ca]steady state concentrations were calculated using average [Ca]selected obtained from applicable data as illustrated using time-resolved data collected at flow rate Q = 5 × 10-! mL s-! at eluent   xxiv pH 2.3 in Panel D. Selected [Ca] concentrations assumed to represent steady state conditions are shown as black filled circles. ........................................................................................................... 95	Figure 4.7 Panels A, B, C show calcite dissolution rates on logarithmic scale from experimental data (black filled circles, error falls within symbol size), and the pore scale modeling (black diamonds) under incoming eluent pH of 2.3, 3.3 and 4.0 respectively. Solid lines in A and B display linear fit through experimental data. Dashed line in C displays linear fit through experimental data. Solid black line in C solely shown to represent approximate hypothetical linear fit with a slope of 0.5. Pore scale modeling data were obtained using k1, k2 from Chou et al. (1989), and k3 from Busenberg and Plummer (1986).  Modeled effluent pH data is also shown (black filled triangles, axis shown on right). .................................................................................... 97	Figure 4.8 Model domain cross-section (simulations are 3-dimensional), showing transport-controlled calcite rhomb (represented as white square) dissolution during highest left) and lowest (right) flow rate conditions at different incoming eluent pH conditions (pH = 2.3 at top, pH = 3.3 in middle, pH = 4.0 at bottom). Color gradient within flow-cell indicates pH. Eluent velocity direction and magnitude indicated by arrow direction and length (scaled in each plot), respectively. ..................................................................................................................................... 98	Figure 4.9 Simulated pH values plotted along the diameter of the cylinder used in the simulation domain, intersecting perpendicularly the mineral grain surface at 0.1 cm from the inlet. Departure from bulk pH at either side of the grain (center blank zone) shows the variation of the DBL thicknesses at 0.1 cm from the inlet face as a function of flow rates (as indicated) at (A) eluent pH 2.3, (B) eluent pH 3.3 and (C) eluent pH 4.0. .................................................................................. 99	Figure 4.10 Thickness of the DBL as a function of eluent flow rate. In symbols are DBL thicknesses extracted from Fig. 4.9 (distance from the point where pH departs from bulk solution   xxv value to the point at the mineral surface where pH is highest). Dashed lines indicate the steps corresponding to the grid cells in discretized numerical domain. Solid lines are a fit to simulated DBL thicknesses using the slope m obtained in the fit to experimental data (Fig. 4.7). ............... 101	Figure 5.1 Schematic of the flow-through dissolution module. E1 – E5 represent eluent bottles, where E1 – E4 are used to generate eluents or standards and E5 is used to continuously supply an internal standard. AGP = advanced gradient pumps. Eluent flows top to bottom of figure. When running standards, the flow-through cell is bypassed. This flow-through module allows for time-resolved control on incoming eluent composition, and a constant, reproducible flow rate through the sample holder, for time-resolved analysis of the effluent stream by quadrupole ICP-MS. ..... 115	Figure 5.2 Time-variable eluent composition associated with experiments carried out on individual calcite, aragonite grains. Eluent pH associated with calcite rhombs (samples 1 – 3) shown in black; eluent pH associated with aragonite grains (samples 4, 5) shown as red, dashed line. Eluent consists initially of DIW in equilibrium with the atmosphere after which acidity was gradually increased to either 0.5 or 2.5 mM HNO3 (pH = 3.3 or 2.6), then 5 mM HNO3 (pH = 2.3). Finally, eluent composition was gradually switched back to DIW. ...................................... 119	Figure 5.3 Measured and modeled Ca concentrations (moles L-1) associated with individual mineral grain experiments. Left hand axis refers to experimental [Ca] concentrations whereas right hand axis refers to modeled [Ca] concentrations. A: calcite rhomb dissolution experiments (samples 1 – 3). B: aragonite grain dissolution experiments (samples 4 - 5). ............................... 123	Figure 5.4 Experimental calcite, aragonite grain dissolution rates (shown on logarithmic scale) versus eluent pH (samples 1 – 4). Since no effluent pH was measured, experimental pH was calculated based on incoming eluent composition. Results associated with this study (samples 1 –   xxvi 4) are shown as filled color-coded circles. Some previously reported studies are also shown as indicated. ........................................................................................................................................ 125	Figure 5.5 Scanning Electron Microscope (SEM) images obtained before and after flow-through dissolution experiments, obtained during preliminary experiments. Shown here is an intact, as well as partially dissolved typical calcite rhomb exposed to identical eluent pH conditions as used in this study (pH 5.7 – 2.3). Figures on the left show intact calcite rhomb prior to dissolution experiments; figures on the right show calcite rhomb after dissolution experiments. Note smooth, generally rounded surfaces of partially dissolved calcite rhomb. .................................................. 126	Figure 5.6 Aragonite (sample 6), calcite (sample 7) end-member powder experiments. A: measured calcite, aragonite surface normalized dissolution rates (aragonite in red solid line, calcite in black solid; aragonite model results in dashed red line, calcite model results in dashed black line) B: 5-point running average end-member molar ratios, shown to represent the eluent reactor renewal time (Mg/Ca shown on left y-axis, Sr/Ca shown on right y-axis);. Two vertical lines indicate time-interval used to quantify end-member Mg/Ca, Sr/Ca molar ratios. ................ 128	Figure 5.7 Sample 8 (mineral mixture experiment) data. A: experimental (solid lines), modeled (dotted lines) [Ca] concentrations indicated on left y-axis, [Mg, Sr] concentrations (shown in black, red and green respectively. Dissolution eluent consisted of 10 mM HNO3 (pH = 5.0) until approximately 4000 seconds, after which acidity was increased to 100 mM HNO3 (pH = 4.0); B: measured dissolution rate (mol m-2 s-1) shown as a solid line, modeled dissolution rate shown as a dotted line. ...................................................................................................................................... 132	Figure 5.8 Calcite, aragonite mixture (sample 8) experimental, model results. A: Mg/Ca molar ratios; B: Sr/Ca molar ratios. Experimental 5-point running average experimental molar ratios (green) are shown to represent the eluent reactor renewal time. Purple dotted line denotes   xxvii calculated El/Ca molar ratio expected if the two minerals dissolve at the same rate. Red line indicates model results (which assumes surface-controlled dissolution). Black dashed lines represent end-member elemental ratios, as indicated. ................................................................... 133	Figure 6.1 Trickle leach effluent [Cu] concentrations (connected filled black dots), as well as effluent pH (connected red crosses). Area between dashed lines represents pH range of the incoming eluent (pH 5.3 – 5.5). ..................................................................................................... 148	Figure 6.2 Cumulate volume (%) of sample versus particle size (note log-scale). Samples used in trickle leach experiment (leached cap rock, filled black connected circles) as well as sample used in MFR experiment (milled leached cap rock, open red connected circles) are shown. Trickle leach experiment data obtained using particle sieving; flow-through experiment data obtained using a combination of particle sieving (> 0.15 mm), laser diffraction (< 0.15 mm) methods. .... 153	Figure 6.3 Experimental trickle leach design. 300 mL of DIW eluent is introduced (on a weekly basis) to the top of a plexiglass column holding 10kg of crushed waste rock. The trickle leach effluent is collected on a weekly basis. .......................................................................................... 154	Figure 6.4 Experimental MFR design. Eluent (blue arrow) is continuously pumped through a 50 mL mixed flow reactor in which the sample is placed (convex bottom, minimizing sample grinding due to rotating stirrer bar), after which the effluent (red arrows) passes through a 0.2 µm syringe filter (preventing sample loss), and continues via an in-line pH sensor prior to periodic fraction collection. When not collecting fractions, effluent is sent to waste. ................................ 156	Figure 6.5 Trickle leach experiment effluent concentrations, expressed in moles L-1. A: Ca (initial, highest value not shown but indicated), K and Mg concentrations; B: Cu, Mn, Sr, Al (initial, highest value not shown but indicated). ............................................................................ 160	  xxviii Figure 6.6 Trickle leach experiment effluent Na (connected filled black circles, left y-axis) and Si (connected filled light blue circles, right y-axis) concentrations, expressed in moles L-1. ............ 161	Figure 6.7 Trickle leach experiment effluent sulphate (connected filled black circles, left y-axis) and Fe (connected filled red circles, right y-axis) concentrations, expressed in moles L-1. .......... 162	Figure 6.8 Trickle leach experiment effluent Ba concentrations, expressed in moles L-1. ........... 163	Figure 6.9 Original DIW MFR experiment effluent pH (connected black filled circles) and follow-up DIW MFR experiment (connected red filled circles). Incoming eluent pH of 5.7± 0.1 is shown in between dashed lines. Error shown associated with original DIW MFR experiment represents error associated with pH calibration drift, which equals the difference in pH calculated by applying two consecutive pH calibration curves. ..................................................................... 165	Figure 6.10 DIW MFR experiment effluent Ca, Mg (connected filled blue and red circles, respectively; scale on left y-axis) and Sr, Cu and Mn concentrations (connected filled green, orange and black circles, respectively; scale on right y-axis), expressed in moles L-1. Gap in data from day 65 – 82 due to accidental pump shutdown. .................................................................... 167	Figure 6.11 DIW MFR experiment effluent K, Si (connected filled purple and light blue circles, respectively; scale on left y-axis) and Ba concentrations (connected filled black circles; scale on right y-axis), expressed in moles L-1. Gap in data from day 65 – 82 due to accidental pump shutdown. ....................................................................................................................................... 168	Figure 6.12 DIW MFR experiment effluent Na (connected filled black circles; scale on left y-axis) and Al concentrations (connected filled gray circles, respectively; scale on right y-axis), expressed in moles L-1. Gap in data from day 65 – 82 due to accidental pump shutdown. ........... 169	  xxix Figure 6.13 DIW MFR experiment effluent sulphate (connected filled black circles; scale on left y-axis) and Fe concentrations (connected filled red circles; scale on right y-axis), expressed in moles L-1. Gap in data from day 65 – 82 due to accidental pump shutdown. ................................ 170	Figure 6.14 Acidic MFR experiment effluent Ca, Si and K concentrations (connected filled blue, light blue, and prurple circles, respectively), expressed in moles L-1. ........................................... 171	Figure 6.15 Acidic MFR experiment effluent Mg, Na (connected filled red and black circles, respectively; scale on left y-axis) and Ba, Mn, Cu and Sr concentrations (connected filled green, blue, orange and light blue circles, respectively; scale on right y-axis), expressed in moles L-1. . 172	Figure 6.16 Acidic MFR experiment effluent Al (connected filled gray circles; scale on left y-axis) and Fe concentrations (connected filled red circles; scale on right y-axis), expressed in moles L-1. .................................................................................................................................................. 173	Figure 6.17 Acidic MFR experiment effluent sulphate concentrations (moles L-1). Error is based on relative standard deviation between replicate standard analyses within one day of analyses. Missing data can be attributed to sample loss during processing. ................................................. 174	Figure 6.18 Measured trickle leach effluent pH (connected black filled circles) and calculated pH based on sulphate effluent data (connected filled red circles). In this calculation, jarosite was assumed to be the sole source of hydrogen ions (as shown in equation 6.1). ................................ 175	Figure 6.19 Trickle leach experiment mineral saturation index (log Q/K) associated with jarosite minerals, ferrihydrite (Fe(OH)3). Except for the initial weeks, jarosite minerals remain undersaturated (log Q/K < 0) throughout the course of the trickle leach experiment. Amorphous ferrihydrite is relatively close to saturation throughout the trickle leach experiment. In comparison, goethite and hematite are supersaturated (log Q/K > 0) throughout the experiment.   xxx Calculated using Geochemist’s WorkBench 10.0.4, using WATEQ4F thermodynamic database (Ball and Nordstrom, 1991). .......................................................................................................... 178	Figure 6.20 A: K: SO!!-molar ratio associated with the trickle leach experiment. Stoichiometric jarosite dissolution (K: SO!!- = 0.5) is indicated as a dashed black line; B: trickle leach experiment effluent K:Cu molar ratio. .............................................................................................................. 182	Figure 6.21 Cu release rate (mg week-1 kg waste rock-1) in A: the DIW MFR experiment (initial, highest value not shown but indicated) plotted versus time; B: the trickle leach experiment (initial, highest value not shown but indicated) plotted versus time number and C: both experiments plotted versus number of pore volume replacements, shown on a logarithmic x, y scales. Note (1) close to order of magnitude higher Cu release rates associated with DIW MFR experiment, compared to the trickle leach experiment, both carried out using deionized water eluent; (2) order(s) of magnitude higher pore volume replacements in the MFR experimental design as compared to the trickle leach experimental design. Pore volume replacement calculated using (flow rate Q (volume per unit time) × t cumulative time)sample volume × porosity. ......... 185	Figure 6.22 DIW MFR experiment mineral saturation index (log Q/K) associated with hematite, goethite, jarosite minerals as well as ferrihydrite (Fe(OH)3). Saturation indices during the initial 125 days could not be calculated due to absence of pH data. Jarosite minerals remain undersaturated (log Q/K < 0) throughout experiment. Similarly, ferrihydrite is quite undersaturated during the final 81 days of the experiment. In contrast, hematite and goethite are supersaturated throughout the experiment. Calculated using Geochemist’s WorkBench 10.0.4, using WATEQ4F thermodynamic database (Ball and Nordstrom, 1991) ..................................... 185	  xxxi Figure 6.23 Acidic MFR experiment mineral saturation index (log Q/K) associated with jarosite minerals, ferrihydrite (Fe(OH)3). Note gaps in jarosite data at 60, 115 and 140 – 155 days due to absence of sulphate data. Jarosite minerals remain undersaturated (log Q/K < 0) throughout the course of the experiment. Ferrihydrite is undersaturated to a greater degree as compared to the DIW MFR experiment (Fig. 6.19). Although hematite and goethite remain supersaturated throughout the experiment, they are to a smaller degree compared to the DIW MFR experiment. Calculated using Geochemist’s WorkBench 10.0.4, using WATEQ4F thermodynamic database (Ball and Nordstrom, 1991). .......................................................................................................... 188	Figure 6.24 K: SO!!-molar ratio associated with the acidic MFR (connected red filled circles) and DIW MFR (connected blue filled circles) effluent. Stoichiometric jarosite dissolution (K: SO!!- =0.5) is indicated as a dashed black line. ......................................................................................... 189	Figure 6.25 Acidic MFR experiment effluent Cu: Fe molar ratio. ............................................... 190	Figure 6.26 Sulphate release rate (mg week-1 kg waste rock-1) plotted versus A: time (days) and B: number of pore volume replacements, shown on a logarithmic scale. Note orders of magnitude higher sulphate release rates associated with MFR experiments, compared to the trickle leach experiment. ..................................................................................................................................... 192	   xxxii Acknowledgements  First and foremost, I would like to acknowledge my graduate advisor, Roger François, for his valuable guidance and support throughout my Ph.D. program at the University of British Columbia. As I transferred my research project from Oregon State University (where I started my PhD studies 2 years earlier, for which I would like to acknowledge the support of Gary Klinkhammer) my start was somewhat unconventional, requiring funds to purchase equipment necessary to continue my research. For the resulting (successful) NSERC RTI proposal, I would like to thank Karen Kohfeld, Greg Dipple, as well as Tom Pedersen for their immediate backing during proposal writing.  As part of the hardware development which followed, I would like to acknowledge Ferenc Tarnok at NResearch Inc. for helping me connect some of the hardware components, as well as my father Paul De Baere for not only teaching me soldering skills from an early age which proved invaluable to this thesis research, but also for assisting me with some of the hardware development.  As research evolved, it quickly became clear ‘calibration’ work was required. As part of this work, I would like to thank Katy Ramsden (undergraduate summer research assistant) for her assistance in numerous, painstaking experiments over the course of 2 summers. Research was re-directed towards promising applications after a series of meetings with Ulrich Mayer, for which I would like to acknowledge him immensely. I also would like to acknowledge Ulrich Mayer for introducing me to his former Ph.D. student Sergi Molins, which resulted in a fruitful collaboration. In terms of software development, I’d like to thank Pejman Rasouli for his helpful   xxxiii insights when troubleshooting inverse modeling software, the results of which are excluded from the chapter for simplicity (however included in appendix A). In parallel, an application in paleolimnology continued to be explored in collaboration with Antje Schwalb and Nicole Börner at the University of Braunschweig, who should be acknowledged for their continued involvement as research findings evolved. I also would like to acknowledge Lorax Environmental Services Ltd. not only for participating in a successful NSERC industry postgraduate scholarship, but also for providing me with the opportunity to explore a more applied research project. In particular, I would like to thank Bruce Mattson and Timo Kirchner for numerous insightful discussions.  Research that ended up beyond the scope of this thesis involved oxygen isotope research carried out at the Jülich Research Centre, where I was kindly hosted by Andreas Lüttge. There, I would like to also acknowledge the technical assistance of Holger Wissel and Robert Moschen. As part of that project, local fieldwork would not have been possible without the help of Maureen Soon, Yiming Luo as well as departmental seagoing technicians.  I would like to thank Greg Dipple for his continued support during the final stretch of my studies, as well as his talented student Kate Carroll for being a quick learner. Also, I would like to thank Marg Amini and Vivian Lai at the Pacific Centre for Isotopic and Geochemical Research (PCIGR), led by Dominique Weis for accommodating new equipment at a logistically challenging time when the laboratory went through a major overhaul.  Finally, I’d like to express my deep and sincere gratitude to all my family members (be it newly established ones in Vancouver or blood relatives in Europe) as well as close friends for their emotional support allowing for the successful completion of this Ph.D. thesis.   xxxiv Dedication  To my parents, who constantly encourage me to pursue my interests.  1 Chapter 1: Introduction  1.1 Mineral dissolution Mineral dissolution is a ubiquitous process, which occurs both on land and in the ocean. The scope of studies related to mineral dissolution is vast. Mineral dissolution is the dominant process in continental weathering and sediment-water interactions, controlling soil formation, freshwater composition, oceanic elemental budgets (e.g. Jeandel and Oelkers, 2015), and the long-term evolution of global climate (e.g. Kump et al., 2000).  Hence, the measurement of mineral dissolution rates and understanding dissolution mechanisms are important for a wide range of fields of study. A number of experimental approaches have been developed to study and measure mineral dissolution, ranging from widely used batch reactors to more elaborate continually stirred flow-through reactors and more recently, surface topographic measurements such as atomic force microscopy (AFM) and vertical scanning interferometry (VSI). To add to the palette of methods used in mineral dissolution studies, this thesis explores the potential of the flow-through time resolved analysis (FT-TRA) – to date solely applied in paleoceanography to estimate the elemental composition of different mineral phases in carbonate microfossils– as an additional tool to measure the dissolution rates and elucidate the dissolution mechanisms of minerals.   1.2 Mineral dissolution kinetics On an atomic scale, mineral dissolution involves the detachment of atoms from the mineral surface. Dissolution rates are dictated by the structure of the mineral surface (e.g. Eggleton, 1986) and the aqueous boundary layer immediately adjacent to the mineral surface. Deciphering   2 dissolution processes at this scale requires monitoring at the Ångstrom scale (e.g. White and Brantley, 1995).  A basic concept in mineral dissolution kinetics is that mineral dissolution reactions consist of a series of different physical and chemical processes that can be broken down into different “steps”. For dissolution, these steps include (e.g. Morse and Arvidson, 2002): (1) transport of reactants from the aqueous phase to the mineral surface, (2) adsorption of the reactants on the mineral surface, (3) migration of the reactants on the mineral surface to an “active” site (such as a dislocation), (4) the chemical reaction between the absorbed reactant and the mineral, (5) migration of products away from the reaction site, (6) desorption of the reaction products to the aqueous phase, and finally (7) transport of the reaction products away from the surface to the aqueous phase.  An important concept in mineral dissolution kinetics is that one of these steps will be the slowest and that the reaction can therefore not proceed faster than this ‘rate-limiting’ step. Steps (1, 7) involve the transport of reactants and reaction products through the aqueous phase to and from the mineral surface. When this process can be considered rate-limiting, the dissolution regime is referred to as being ‘transport-controlled’. Steps (2 – 6) occur at the mineral surface. When any of these processes are rate-limiting, the dissolution regime is considered to be surface-controlled.  In a pure transport-controlled mineral dissolution scenario (Figure 1.1A), reaction products are detached so rapidly from the mineral surface that they build up to form a saturated solution adjacent to the mineral surface. Mineral dissolution is then controlled by transport of these reaction products via advection and diffusion into the ambient undersaturated aqueous phase. The rate of dissolution depends on the thickness of the boundary layer, which itself   3 depends on the flow velocity and/or the degree of stirring, with increased flow / stirring resulting in accelerated transport, and subsequently, faster dissolution (Berner, 1978). In a purely surface-controlled dissolution regime, reaction product detachment is sufficiently slow that buildup at the mineral surface cannot keep up with advection and diffusion, and the resulting concentration level adjacent to the mineral surface is essentially the same as that in the surrounding aqueous phase (Fig. 1.1B). In this case increased flow and / or stirring have no effect on the dissolution rate. In intermediate conditions between pure transport-controlled and pure surface-controlled dissolution regimes, surface detachment is sufficiently fast that the surface concentration builds up to levels greater than the surrounding solution but does not reach saturation with respect to the dissolving mineral phase (Fig. 1.1C). A rule-of-thumb proposed by Berner (1978) states that in aqueous solutions, very soluble minerals tend to undergo dissolution under a transport-controlled dissolution regime, whereas relatively insoluble minerals undergo dissolution under a surface-controlled dissolution regime.    Figure 1.1 Concentration C versus distance 𝑟 from a crystal surface for three rate-controlling regimes: (A) transport control; (B) surface control; (C) mixed control. 𝐶!" = saturation concentration; 𝐶!"#$ = bulk aqueous phase concentration. Adapted from Berner, 1978.   4 A central concept of mineral dissolution kinetics is that once a dissolution mechanism is implied and the appropriate, intrinsic dissolution rate parameters (e.g. dissolution rate constants, reaction orders) are measured (see section 1.3), the kinetic mechanism and parameters can be transferred to natural systems (Lasaga, 1998). However, it is important to recognize that mineral dissolution rates obtained in the laboratory represent the upper limit relative to the overall rates observed in the field. Lower field-based dissolution rates have been partly attributed to the presence of transport-controlled dissolution (e.g. Swoboda-Colberg and Drever, 1993).   1.3 Experimental approaches used to study mineral dissolution 1.3.1 Batch reactor experiments Historically, one of the most commonly used approaches involves a batch reactor containing the mineral being studied in a known volume 𝑉 of aqueous solution. Over the course of the experiment, the concentration of the reaction product 𝑖, 𝐶! is monitored as a function of time. The dissolution rate 𝑟 (moles m-2 s-1) is then calculated based on the d𝐶!/d𝑡 ratio using equation (1.1): 𝑟 = !!!!!  ×  !!! × !! × !!          (1.1) where 𝐴! and 𝑀! are the specific surface area (m2 g-1) and the mass (g) of the dissolving mineral and 𝑠! the stoichiometric coefficient of reaction product 𝑖, expressing the moles of reaction product 𝑖 released by dissolving 1 mole of mineral. Both the mass and surface area may change over the course of a batch dissolution experiment and need to be accounted for.  Over the course of a batch reactor experiment, reaction products are continuously added to the bulk solution, which may become oversaturated with respect to one or multiple secondary mineral phases. This represents one of the key limitations associated with the batch reactor   5 approach. In addition, in some cases, large variations in bulk solution pH may be of concern, which can be minimized by using suitable buffers. However, the addition of buffer solution to the bulk solution may interfere with the dissolution mechanism and bias the measured dissolution rate. Similarly, the variation in the bulk concentration of solutes such as Al affects mineral dissolution rate measurements (as demonstrated for aluminosilicates by Oelkers et al., 1994).   1.3.2 Continually stirred flow-through reactor experiments Also referred to as “mixed-flow reactor” experiments, the continually stirred flow-through reactor approach involves a reactor, which is kept under agitation by a stirring bar. The aqueous phase is continually pumped through the flow-through reactor at a constant flow rate, 𝑄. The concentration of the reaction products, generated by dissolution of the mineral undergoing dissolution is typically monitored before (𝐶!") and after (𝐶!"#) the flow-through reactor. The experiment is run until 𝐶!"# reaches a constant / steady-state value, which is used to calculate the steady-state dissolution rate using equation (1.2): 𝑟 =  ! × (!!"#!!!") !! × !! × !!            (1.2) In relation to the batch reactor experiments, a number of key advantages associated with flow-through reactors include: (1) the continuous renewal of the aqueous phase helps maintain far-from-equilibrium conditions with respect to the dissolving mineral, thereby minimizing the possibility of secondary mineral phase precipitation (e.g. Holdren and Speyer, 1986); (2) bulk aqueous phase composition can be held constant over the course of a dissolution experiment. As a result, measured dissolution rates will not be affected by a temporally changing bulk aqueous   6 phase composition (e.g. Holdren and Speyer, 1986); (3) each measurement provides a direct measure of the mineral dissolution rate. In comparison, in a batch reactor experiment, bulk aqueous phase composition is monitored as a function of time. Dissolution rates are deduced by using a regression through the acquired data, where the slope is used to quantify the dissolution rate. As this approach relies on statistics, it has been proven difficult to distinguish between dissolution rate laws (e.g. White and Claasen, 1979); (4) experiments are able to monitor nonstoichiometric dissolution behavior in multicomponent minerals (e.g. Putnis and Ruiz-Agudo, 2013). Non-stoichiometric dissolution is revealed when the ratio of elements in 𝐶!"# differs from the stoichiometric composition of the mineral.  The period of time necessary to reach steady-state values depends on two factors: the time required for the mineral surface to reach a steady dissolution rate (𝑇!"##), and the time required to fully replenish the aqueous phase within the flow-through reactor (𝑇!"#$!, typically taken as 4 – 5 times the residence time (e.g. Pokrovsky and Schott, 2000), 𝑉/𝑄). If 𝑇!"## ≫ 𝑇!"#$!, the amount of time required for the experiment to reach steady-state 𝐶!"# is mainly dictated by the properties of the mineral surface. This situation arises when 𝑇!"## is very long or 𝑇!"#$! is very short. On the contrary, if 𝑇!"#$! ≫  𝑇!"##, the time needed for the experiment to reach steady-state 𝐶!"# is mainly dictated by the aqueous phase residence time in the flow-through reactor.  The most commonly used type of mixed flow reactor experimental design involves a continuous supply of new eluent to the reactor (e.g. Pokrovsky and Schott, 2000, Fig. 1.2A). In comparison, when using a fluidized bed reactor (a type of flow-through reactor) there are two   7 fluid flows: one is very fast to keep the dissolving mineral particles under agitated suspension, and the other is monitored for 𝐶!" and 𝐶!"# (e.g. Chou and Wollast, 1984, Fig. 1.2B).   Figure 1.2 Conceptual representation of a mixed flow reactor experiment (Panel A) and a fluidized bed reactor experiment (Panel B, adapted from Chou and Wollast, 1984). In a MFR experiment, a single pump (P1) generates an identical fluid flow for both the inflow of the eluent and the outflow of the effluent. In a fluidized bed reactor, P1 generates the fluid flow required to keep the particles in suspension, P2 generates the fluid flow responsible for the eluent input and effluent output.   1.3.3 Surface topographic measurements 1.3.3.1 Atomic force microscopy (AFM) Atomic Force Microscopy has allowed for the observation of mineral surface topography at high (1 × 10!! 𝑚) spatial resolution (Binnig et al., 1986). Topography height resolution is at the Ångstrom level as AFM imagery is based on the measurement of interatomic forces between a peristaltic pumpeluenteffluentstirring barMIXED FLOW REACTOR(MFR)FLUIDIZED BED REACTOReluenteffluentfluidizedbedP2P2P1P1 + P2P1P1overlyingsolutionA B  8 probe tip and the electrons of the atoms in the sample surface (Ruiz-Agudo and Putnis, 2012). AFM was developed to be used in liquids by Marti et al. (1987) and has since been used to study mineral dissolution kinetics (e.g. De Giudici, 2002). In short, AFM continuously scans the mineral surface with a probe tip, which is attached to a cantilever. The deflection of the cantilever is continuously monitored using laser light. The distance the cantilever moves in the 𝑧 direction represents the topography at known 𝑥,𝑦 coordinates. The end-result is a topographic image of the mineral surface.   In order to illustrate the use of AFM in quantifying mineral dissolution rates, the dissolution of calcite is considered. The dissolution of calcite cleavage surfaces occurs by the retreat of surface steps in preferred crystallographic directions (e.g. Ruiz-Agudo and Putnis, 2012). Step motion occurs by preferentially removing ionic species in calcite by lateral advancement of kink sites (Ruiz-Agudo et al., 2012). Step motion – either originally present prior to exposure to an aqueous phase, or created upon contact to the aqueous phase by the generation of etch pit – remains straight and advances at constant velocity in pure solutions (Liang et al., 1996). The overall mineral dissolution rate (moles cm-2 s-1) can be deduced from AFM topography data using equation (1.3): 𝑅!"# =  ∆! !!"#!!(!!!!!)           (1.3)  where ∆𝑉 (cm3 etch pit-1) denotes the measured volume increase of etch pits; 𝑁!"#(etch pits cm-2) denotes the number of etch pits per unit surface area; 𝑉! represents the molar volume (cm3 mol-1) and (𝑡! − 𝑡!) (s) represents the time passed between two sequential images.     9 1.3.3.2 Vertical Scanning Interferometry (VSI) Vertical scanning interferometry can be used to measure the retreat of a mineral surface relative to a fixed, reference surface. The mineral is placed in a temperature controlled, aqueous flow-through system, and is exposed to a beam of light. Depending on the backscatter characteristics (for details refer to Lüttge et al., 2003), mineral height can be monitored at high precision (ängstrom to nanometer precision depending on experimental design). Mineral height measurements are made before, during, and after mineral dissolution during fixed time intervals. Time resolved height observations of individual topographic points (co-ordinates 𝑖, 𝑗) allow for the calculation of the velocity of dissolution at that topographic point on crystal surface (ℎ𝑘𝑙) (equation 1.4): !!!,!!! = (!!,!)!! (!!,!)!!!! !! = 𝑣!,! (!!")        (1.4) where 𝑣!,! is the velocity of mineral dissolution at coordinates 𝑖, 𝑗 at surface (ℎ𝑘𝑙) in units of cm s-1. The rate constant associated with that co-ordinate (𝑘!,!) is then calculated by dividing by the molar volume (cm3 mol-1) of the mineral undergoing dissolution (𝑉!, equation 1.5), resulting in rate constants expressed in mol cm-2 s-1.  𝑘!,! =  !(!!")!!            (1.5) An average rate constant associated with surface ℎ𝑘𝑙  can be obtained by summing all local rate constants, 𝑘!,! and take the average (equation 1.6) to obtain a rate constant 𝑘(!!") with the entire surface. 𝑁!,! represents the number of pixels measured in a VSI scan.  𝑘(!!") =  Σ !!,! (!!")!!,!             (1.6)   10 This normalization relative to number of pixels makes VSI rates independent of bulk surface determinations (e.g. Lüttge, 2004).   1.4 Flow-through time-resolved analysis (FT-TRA) Although flow-through time-resolved analysis (FT-TRA) was introduced over a decade ago (Haley and Klinkhammer, 2002), it has not yet been widely adopted by the scientific community. FT-TRA adheres to the principles associated with the mixed flow reactor (section 1.3.2), however typically incorporates a flow-through cell of much smaller volume (here 25 – 50 µL). This small flow-through cell volume results in very short eluent residence times in the reactor, allowing for high-resolution time-resolved effluent compositional analysis. This analytical capability is achieved by combining advanced eluent gradient pumps (commonly used in ion chromatography, capable of generating precise flow rates) with a small volume flow-through reactor, the effluent of which is sent to an Inductively Coupled Mass Spectrometer (ICP-MS), capable of time-resolved analysis.  To date, the FT-TRA approach has been applied exclusively in paleoceanography. In particular, FT-TRA has been used as a tool to extract the elemental composition of different mineral phases in calcareous marine microfossils (foraminifera). Both elemental ratios (Haley and Klinkhammer, 2002; Klinkhammer et al., 2004; Benway et al., 2003; Klinkhammer et al., 2009) and rare earth patterns (Haley et al., 2005) have been studied. Although some research carried out as part of this thesis addresses the use of FT-TRA for paleoceanographic studies (chapter 5), the main goal of this thesis is to explore alternative applications of the flow-through technique, in particular, for measuring pure mineral dissolution kinetics (chapters 3, 4) as well as metal leaching from mine waste (chapter 6).    11 1.5 Thesis objectives The objectives of this thesis are outlined below, and summarized as a flowchart in Fig. 1.3. Although FT-TRA instrumentation has been developed at Oregon State University (Haley and Klinkhammer, 2002; Haley and Klinkhammer, 2003; Benway et al., 2003; Klinkhammer et al., 2004; Haley et al., 2005) as well as the University of Bremen (Haarmann, 2012), accessible information outlining the steps required to assemble a flow-through module capable of FT-TRA is not readily available. Hence, the purpose of chapter 2 is to provide a detailed, yet accessible overview of the module development necessary to carry out FT-TRA.  The first data chapter (chapter 3) outlines how FT-TRA can be applied as a tool to measure mineral dissolution rates based on the example of forsterite (De Baere et al., 2015). Exploratory in nature, this chapter addresses whether data recorded during an online FT-TRA experiment can provide additional mechanistic information pertaining to dissolution processes occurring at the mineral surface. The second data chapter (chapter 4) illustrates the novel use of FT-TRA as a tool to empirically deduce the rate-limiting dissolution regime (surface-, mixed- or transport-controlled dissolution regimes) (De Baere et al., submitted). Although Lasaga (1998) stated that “the important data are the surface reaction rates”, this chapter explores through a combined approach including FT-TRA and pore scale modeling the effect of transport-limitations on mineral dissolution, thereby allowing for the quantification of meaningful dissolution rate parameters under transport controlled conditions.  The third data chapter (chapter 5) revisits the premise on which FT-TRA data analysis has been based to date in paleoceanography (e.g. Klinkhammer et al., 2004). Using carefully designed FT-TRA experiments using pure mineral mixtures, the assumption that mineral phases   12 are dissolved sequentially according to solubility is re-evaluated. This chapter draws upon expertise gained as part of chapters 3 and 4.  The fourth data chapter (chapter 6) is more applied in scope and explores the added value of a flow-through dissolution approach to better predict and understand the leaching of potentially toxic element from mine wastes.  The overarching aim of this thesis is to assess the potential of FT-TRA technique for this range of applications, with an emphasis on assessing its ability to rapidly determine mineral dissolution parameters and to investigate the transient dissolution of mineral phases subjected to variations in external forcings.   13   Figure 1.3 Flow-chart indicating thesis outline. Introductory and concluding chapters (chapters 1 and 7) not shown.     14 Chapter 2: Flow-through time-resolved analysis (FT-TRA) module design  2.1 Introduction The establishment of the flow-through time-resolved analysis (FT-TRA) technique requires the development of purpose-built hardware, as well as software. Although similar instrumentation has been developed at Oregon State University (Haley and Klinkhammer, 2002; Haley and Klinkhammer, 2003; Benway et al., 2003; Klinkhammer et al., 2004; Haley et al., 2005) as well as the University of Bremen (Haarmann, 2012), accessible information outlining the steps required to assemble a flow-through module is not readily available. Hence, the purpose of this chapter is to provide a detailed, yet accessible overview of the method development carried out as part of this thesis.    2.2 FT-TRA experiments During a typical FT-TRA experiment, a solid sample is placed in a flow-through sample holder and continuously exposed to an incoming stream of fresh eluent. The composition of the incoming eluent can be programmed to be temporally variable, if required. For example, this capability has been used to clean microfossils before dissolution under acidic conditions (e.g. Haley and Klinkhammer, 2002). The effluent containing elemental information associated from the dissolving solid is either sent to a fraction collector for periodic sample collection, or sent directly to a quadrupole ICP-MS for online, time-resolved elemental analysis. The quadrupole ICP-MS used as part of this research is an Agilent 7700×, housed at the Pacific Centre for Isotopic and Geochemical Research (PCIGR) at the University of British Columbia.    15 2.3 FT-TRA leaching module development Two flow-through modules were developed as part of the research involved in this thesis (Fig. 2.1). The initial module was developed for online, time-resolved analysis and was used for applications explored in data chapters 3, 4 and 5. A second module was developed incorporating a single, larger internal volume (50 mL) flow-through cell capable of accommodating larger samples, and is used in data chapter 6. An overview of key components required is provided in Table 2.1. Major components are described individually below.    Figure 2.1 Diagram of the flow-through module developed (small internal volume flow-cell version shown). A Dionex gradient pump draws upon up to four eluent bottles after which the eluent flows to a specific flow-through reactor (depicted as reaction vessel – a total of 6 individual sample positions were present). Here, the sample   16 gradually dissolves, and the dissolved sample (effluent) flows to a quadrupole ICP-MS (for online analysis, chapters 3, 4 and 5) or a fraction collector (for offline fraction collection at user-defined time-intervals, chapter 6). When using the online-set-up, the effluent merges with a constant flow of internal standard supplied by a separate, isocratic pump (used for temporal instrumental drift correction). Optional in-line pH sensors can be installed (as indicated here) if a measurable (pH sensor precision ± 0.1 pH units) pH difference is to be monitored.  Supplier Components (# part number, followed by description) Thermo Fisher Scientific Inc. #072034, Dionex Dual analytical gradient pump with degasser #5911.0010, Universal Chromatography Interface (UCI, 8 data channels) National Instruments Inc. #779975-01 USB-6509, 96-channel digital I/O and NI-DAQMX driver software #185095-02, Cable, 2 x 100-POS .050 series D-type, shielded, flex motion #776990-01 SCB-100, shielded connector block Neptune Research & Development Inc. #225T092C, Teflon coated solenoid valve #225D5X24, Cooldrive circuit board Bio-Rad Laboratories (Canada) Ltd. #7602042, pH electrode #7602044, flow cell FIAlab Instruments, Inc. #82001, custom flow cell: 50 mL internal volume Merck Millipore Ltd. #SLLGC13NL, IC certified filter units (pore size 0.2 µm) Teledyne Isco, Inc. Foxy® R1 fraction collector Table 2.1 Overview of key components required for assembling a flow-through dissolution sample introduction module.   17 2.3.1 Gradient pump A gradient pump capable of highly reproducible, precise flow-rates is the central component of the flow-through system. Therefore, pumps routinely used in ion chromatography (Dionex instrumentation supplied by Thermo Fisher Scientific Inc.) were used. These pumps have flow-paths made of chemically inert material, thereby eliminating the potential of chemical leaching. In addition, gradient pumps contain a proportioning valve, which allows for the generation of (pre-programmed) time dependent eluent composition (achieved using mixing known proportions of up to 4 eluent bottles). An example of a typical programmed dissolution sequence is shown in Fig. 2.2.     18 Figure 2.2 Time-dependent eluent composition (FT-TRA leaching protocol) as programmed using Dionex Chromeleon™ software. Top graph shows time-dependent eluent bottle composition (%) versus time (min). Lower panel shows table used to program leaching dissolution sequence (% of eluent A is automatically calculated in software, based on user input of % eluent B, % eluent C and % eluent D). In this sequence, the sample is initially exposed to 40 min of 100% eluent A (yellow), after which eluent C is gradually introduced (40 min < t < 70 min) before it reaches a maximum of 10 % at t = 70 min. This eluent proportion is held constant between 70 min < t < 100 min. At 100 min < t < 120, the proportion of eluent C increases to 100%, at which level it remains constant for 30 min. At 150 min < t < 160, the eluent composition gradually returns to 100 % eluent A. Throughout this experiment, flow-rates are held constant at 0.70 ±0.0367  mL min-1 (error indicates the 95% confidence interval). Further details on flow rate precision are provided in section 2.5. © 2011 Thermo Fisher Scientific Inc., by permission.   2.3.2 Flow-through reactors The relatively small internal volume flow-through reactor is an off-the-shelf, 13 mm diameter, PTFE syringe filter (0.2 µm mesh, Millipore Ltd.). These syringe filters were held in place by using two parallel solenoid valves equipped with standard luer fittings (Fig. 2.3A). The internal volume of the syringe filter is 25 µL (applicable to powder samples, Fig. 2.4A), with an overlying dead volume of 50 µL (applicable to calcite rhomb samples, Fig. 2.4B) The larger, 50 mL internal volume flow-through cell (manufactured by FIAlab Instruments Inc.) was entirely made out of chemically inert Teflon® material (Fig. 2.3B). Standard luer fittings were used for the inlet and outlet connections. The bottom of this flow-through reactor was engineered to be concave to limit any gradual grinding of the solid sample by a magnetic stirrer bar (limiting any gradual change in sample surface area). An aluminum bracket was designed to provide a press mechanism between the flow cell lid and the flow cell body, ensuring a tight seal.    19  Figure 2.3 Flow-through cells used during this thesis. Panel A: cross-section of an off-the-shelf 13 mm syringe filter. Blue arrows indicate eluent flow through filter. Blue rectangles represent open slits through which the eluent moves upwards. Syringe filters are held in place using luer fittings connected to solenoid valves, from where the effluent flows via tubing to the ICP-MS for online time-resolved analysis. Panel B: 50 mL Teflon® flow-though cell as used in chapter 6. Eluent enters the bottom of the flow-cell, exists at the top of the flow-cell. Aluminum support bracket ensures a tight seal. eluent ineffluent outO-ringAluminumsupport bracket Teflon flow cellChapters 3, 4, 5 Chapter 6A Beluent ineffluent out0.2 µm meshsolenoid valveluer fittingsolenoid valveluer fitting  20  Figure 2.4 Flow-through cell volumes relevant to this thesis. Panel A: 25 µL volume (shown as y) relevant to chapters 3, 4, 5. Panel B: 50 µL volume (shown as x) relevant to chapters 4 and 5.   2.3.3 Solenoid valves Two parallel manifolds which incorporate six separate normally - closed solenoid valves integrated in a single block of Teflon® each were used to hold six small volume flow-through reactors (Fig. 2.5). As a precaution, the outside of the solenoid valves were coated with Teflon®, in order to prevent long-term corrosion.  eluent ineffluent out0.2 µm meshsolenoid valveluer fittingsolenoid valveluer fittingeluent ineffluent out0.2 µm meshsolenoid valveluer fittingsolenoid valveluer fittingyyyyyyyyyyyyyyyyyyyyyxxxxxxxx xx     xx xx     xxpowdersamplesV = y (25 µL)individualcalcite rhombsV = x (50 µL)A B  21  Figure 2.5 Two parallel manifolds, each holding six separate solenoid valves (numbered on image). Flow-through reactors (syringe filters, as indicated) are located in between two parallel solenoid valves (upper and lower solenoid valves, respectively).  Since the solenoid valves are supplied with bare wires only, additional hardware and software was required to individually open and close two parallel solenoid valves. To this end, CoolDrive® circuit boards (supplied by Neptune Research & Development Inc.) were used (Fig. 2.6). These circuit boards apply a holding-voltage of 24VDC to open the valves (70mA current per solenoid valve) according to the logic level (= digital) signals on their input. In our configuration, two parallel six-position valves required a total of 12 signal outputs (hence, 3 CoolDrive® circuit boards were required). In addition to the 24VDC power supply required to power the valves, a 5VDC power supply was required to power each CoolDrive® board (1.5A per board).   22 Digital valve control was established using a 100-position connector block in conjunction with a 96-channel converter box (National Instruments Inc.). DAQMX driver software (National Instruments Inc.) allowed for signal input control, and was incorporated in a LabVIEW based control panel. The LabVIEW control panel (Fig. 2.7) allows the user to manually open and close two parallel solenoid valves.    Figure 2.6 CoolDrive® circuit board used to operate the solenoid valves. As each board controls 5 valves, a total of 3 boards were required (15 output connections) to control 2 parallel solenoid valves (12 valves in total).     23  Figure 2.7 LabView control panel allowing the user to open and close two parallel solenoid valves (2 parallel manifolds operate 6 flow-cell positions), each holding a flow-through reaction cell. Here, two parallel valves named ‘Valve 1’ have been opened, routing the eluent to flow via the flow-through reaction cell located in between both valves.  2.3.4 In-line pH measurement In-line pH measurement was carried out using off-the-shelf pH electrodes (Bio-Rad Laboratories Ltd.). The pH electrode used is a sealed Calomel combination electrode consisting of a pH and a reference electrode built into a single body. To allow for in-line pH measurement, a flow cell made out of chemically resistant polyether ether ketone (PEEK) was used to hold the pH electrode. This flow cell has an internal volume of approximately 80 µL when the pH electrode is inserted. In order to record the pH electrode conductivity data, a number of steps were required: (1) correcting for wire impedance; (2) converting the analogue output to a digital signal   24 and (3) monitoring, as well as exporting raw pH data using Dionex Chromeleon™ software. Each step in our set-up (summarized in Fig. 2.8) is described below.   Figure 2.8 Hardware setup used to record in-line effluent pH. The pH electrode is exposed to the effluent via a flow-cell (shown in gray) holding the pH electrode. The analog pH electrode signal is amplified before being sent to an analog input associated with the universal chromatography interface (UCI). There, the analog signal is digitized, and sent to a computer via USB output.  The challenge from an electronic standpoint is that the output impedance of a pH electrode is extremely high. The pH electrode acts as a high impedance voltage source, producing an output voltage relative to the pH value of the solution (approximately 140 mV at Universal Chromatography Interface (UCI)Channel 1ANALOG INPUTSpHelectrodeamplifierAnalog HighAnalog LowpHelectrodeUSB9VDC80 µLvolumeeffluent ineffluent out  25 pH 4.0 to -210 mV at pH 10, see Fig. 2.9). The series connection of the pH electrode and digital voltmeter acts as a 2:1 resistive voltage divider, effectively halving the signal. Hence, a buffer amplifier with extreme input resistance is required (Fig. 2.9). The input bias current of the amplifier must be extremely low, as 1 nA through a 100 MOhm resistance equals 0.1 volt (equivalent to a signal distortion close to 2 pH units). The CA3160 amplifier has a very low (2 × 10!!" A) input bias current. Using this amplifier, the maximum signal distortion lies below 0.1%. A simple circuit was built, with a single CA3160 amplifier and a trim potentiometer for offset compensation. The trim potentiometer is required to bring the buffer amplifier output voltage to zero volts when the electrode is short-circuited. This avoids a biased pH electrode signal.  The following steps are required for offset compensation: (1) disconnect pH electrode from input; (2) connect BNC short circuit on input; (3) adjust trimpot until zero mV is achieved at output; (4) remove BNC short from circuit and (5) reconnect pH electrode. Finally – in this simple circuit – the inherent temperature dependence of the pH probe is not compensated for, as it was used in a controlled laboratory environment. In order to convert the analogue pH signal to a digital signal, an off-the-shelf analogue – digital signal converter box was used (Universal Chromatography Interface, Table 2.1). Here, the pH signal is converted into a digital signal, which can be recorded (and monitored live) using Dionex Chromeleon™ software.  The following steps allow for exporting pH electrode data (mV, which can be converted into pH after calibration) within the Dionex Chromeleon™ software. Right click the sequence folder, select <Batch Report…>, select channel (associated with pH electrode), unselect <Printout>, select <Export> and click on the <Export Settings…> button. Here, define output folder, and select <ASCII text format >. Click <next>, deselect <Export results using sheet> and   26 select <Selected channel>. After clicking on <Finish>, click <OK> after which the raw data will be saved as a text file in the user-defined directory.    Figure 2.9 Schematics showing pH sensor buffer amplifier. A: power supply (accomplished using 9 V block battery), B: offset compensator.  For the pH electrode calibration, commercially available buffer solutions were used (pH 4, 7 and 10 respectively, Thermo Fisher Scientific Inc.). Electrode data (mV) was continuously recorded as the electrode was exposed to each buffer solution (Fig. 2.10). Averaged mV values   27 associated with each pH buffer plateau were plotted versus buffer pH, resulting in a linear calibration curve (Fig. 2.11).    Figure 2.10 Electrode data obtained during pH calibration. Plateaus represent time intervals during which electrode was exposed to pH buffer solutions (pH 4, 7 and 10 respectively).   28  Figure 2.11 Linear pH calibration curve (coefficient of determination = 1) where the electrode signal (average mV, x-axis) is plotted versus pH buffer solution (y-axis).   2.3.5 Fraction collection For fraction collection, a Foxy® fraction collector was used (Teledyne Isco Inc.). This fraction collector provides the ability to periodically sample the continuous stream of effluent (here, the effluent) in multiple collection vessels (here: 15 mL test tubes), switching vessels based on time interval. The fraction collector contains a diverter valve, which automatically directs effluent to waste during time intervals when no fraction collection is required. The fraction collector is connected to the Universal Chromatography Interface (UCI) via RS-232 serial connection, and controlled within the Chromeleon™ software (supplied with the gradient pumps), since Foxy® fraction collector support is built-in.   29  At low flow rates (such as 0.05 mL min-1), 300 minutes are required to fill up a single 15 mL test tube. As the Chromeleon™ software limits collection time to 9999 seconds (< 300 minutes), the next test tube is used to finalize fraction collection. Keeping this limitation in mind, collection times and/or flow rates can be adjusted to optimize the use of the fraction collector. Alternatively, the fraction collector can be programmed offline, bypassing this software limitation.  2.4 Standards Given the ability of the gradient pump to mix eluent bottles in pre-programmed proportions, time-dependent proportions of a high standard with the standard matrix can be automatically generated. As a result, a series of standard concentrations can be generated, allowing for the establishment of a calibration curve. As part of this work, a high standard was introduced at 1 – 5 – 10 – 25 – 50 – 75 – 100 % of eluent composition, with the remainder representing the standard matrix (here: 1% HNO3, Fig. 2.12). Resulting elemental signal plateaus associated with each standard : matrix proportion (Fig. 2.13) can be used to establish a calibration curve (Fig. 2.14).     30  Figure 2.12 Time-dependent eluent composition during a typical standard run. Here, eluent bottle D represents the standard matrix (1 % HNO3) and eluent bottle B represents the high standard solution in a 1 % HNO3 matrix. Note (1) standard concentrations increase over time and (2) eluent flow rate is held constant at 0.70 ±  0.0367  mL min-1 (error indicated the 95% confidence interval). Further details on flow rate precision are provided in section 2.5. © 2011 Thermo Fisher Scientific Inc., by permission.      31  Figure 2.13 Typical standard run (here, internal standard normalized 24Mg) showing time-resolved plateaus of known high standard : matrix proportions. Percentages shown above plateaus denote proportion of high standard.   Figure 2.14 Standard calibration curve (here: 24Mg). Internal standard normalized values shown represent averages obtained from plateaus shown in Fig. 2.13.    32 2.5 Eluent flow rate precision Using a fraction collector, DIW water was collected for a set period of time and weighed. Based on the difference between expected and measured sample weight measurements, flow rate precision was assessed. An overview of experiments conducted under a range of flow rates is provided in Table 2.2. It was found that the gradient pump was able to produce highly precise, reproducible (within a few percent) flow rates over a large range of flow rates (0.015 – 7.5 mL min-1).   Q (nominal) (mL min-1) n Q (measured) (mL min-1) 𝐀𝐜𝐜𝐮𝐫𝐚𝐜𝐲 % 0.015 7 0.014± 0.003! 5.50 0.1 10 0.101± 0.001! 1.27 0.2 10 0.203± 0.001!  1.37 0.35 10 0.356± 0.003! 1.84 0.7 10 0.705± 0.037! 0.76 1 10 1.015± 0.004! 1.51 2.5 10 2.511± 0.019! 0.43 3.5 10 3.483± 0.019! 0.49 4.5 10 4.468± 0.028! 0.70 5.5 10 5.424± 0.034! 1.37 6.5 10 6.363± 0.068! 2.11 7.5 10 7.308± 0.264! 2.57   33 Table 2.2 Overview of pump flow rate 𝑄 measurements. Measured flow rates (n experiments) provided in third column, where 1error represents the 95% confidence interval.  The accuracy of the measured flow rate relative to the nominal flow rate is provided as a percentage in the fourth column, and lies within 5.50 % across all flow-rates. Best flow-rate precision is achieved in the mid rang and decreases at high and low flow rates.   2.6 Time-resolved analysis ICP-MS Time-resolved data acquisition (TRA) using ICP-MS has to date been primarily applied to laser-ablation based (LA-ICP-MS), or ion chromatography ICP-MS analyses. Similar to FT-TRA, sample introduction during IC-ICP-MS analyses occurs through a gradient pump (e.g. Kohlmeyer et al., 2003), which allows for the separation of chemical species on a chromatography column over time.  The work presented here as part of this thesis uses the TRA capability of a quadrupole ICP-MS, which has been noted for its fast scanning capabilities over a wide mass range due to efficient peak hopping mass selector capabilities (e.g. Longerich et al., 1996). The actual analytes to be measured must be specified, and the number of channels to be used for each analyte needs to be selected accordingly. Typically, the number of masses to be scanned and the number of channels per peak determine duration needed to acquire one data point during the TRA experiment, which dictates the temporal resolution of the FT-TRA experiments (typically a few seconds). The collected data are acquired as sweeps. One sweep represents data collected during each time slice during which all masses are measured once. During TRA, each sweep is recorded individually allowing for the collection of data as a function of time. This enables careful monitoring of eluent blanks, tubing washout, etc. during FT-TRA.   34 All the work presented here was carried out on an Agilent 7700x quadrupole ICP-MS (located at the Pacific Centre for Isotopic and Geochemical Research, PCIGR) in He mode to minimize isobaric interferences (McCurdy and Woods, 2004).  2.7 General operating procedure Ideally, the flow-through module is installed on a cart allowing for easy transport to and from an ICP-MS (Fig. 2.15). As part of this work, we identified InterMetro (Metro) carts as most suitable, given their durable, corrosion proof polymer mats.  The recommended general operating procedure for FT-TRA explained in more detail is provided in section 2.7.1 (online FT-TRA) and 2.7.2 (offline FT-TRA) below. A flow-chart outlining the recommended protocol for online analysis is also provided in Fig. 2.16.     35  Figure 2.15 Flow-through module developed for online FT-TRA.    36  Figure 2.16 Recommended protocol for online FT-TRA.   2.7.1 Online FT-TRA The standard operating procedure for on-line FT-TRA involves the following steps: (1) Sample preparation: (a) load flow-through reactor(s) with sample(s); (b) program leaching protocol in Chromeleon™ software; (c) prepare eluents / high standard solution / standard matrix / internal standard as required.  (2) Calibrate pH sensor with commercial pH solution buffers.   37 (3) After quadrupole ICP-MS tuning has been completed, connect flow-through module to sample nebulizer. Note a higher volume nebulizer may be required depending on what flow rates are used (here, a Micromist nebulizer from Glass Expansion with 0.8 mL min-1 uptake was used, part number #AR35-1-UM08EX). Ensure rotation speed of ICP-MS peripump associated with waste line is increased (0.3 rpm when using a combined inflow of 0.8 mL min-1) after connecting flow-through module. Failure to do so results in the build-up of effluent waste and (if ignored) eventual ICP-MS shutdown.  (4) Manually adjust tubing to bypass solenoid valves holding samples. Bypassing solenoid valves lowers amount of time dedicated to running standards, and limits the amount of flushing time required to achieve baseline after running the highest standard. Run standards leaching program (e.g. Fig. 2.12).  (5) Reconnect eluent flow path via solenoid valves. Open solenoid valves holding an empty flow-through reactor for blank analysis. Run programmed sample leaching protocol (e.g. Fig. 2.2). (6) At the end of blank analysis, manually select eluent, which does not interfere with sample (e.g. DIW), preventing remaining eluent in flow-path to interfere with sample when initially opening sample solenoid valve. Flush tubing with this eluent prior to opening sample valve. After opening sample solenoid valves, close valves holding the blank flow-through reactor (e.g. open valve 2 then close valve 1 as shown in Figs. 2.5, 2.7).  (7) Run sample leaching protocol (e.g. Fig. 2.2). (8) After final sample, switch back to blank filter (position 1) and run 1 – 2 % nitric acid through flow-through system for 10 – 15 min, allowing for flow-path and ICP-MS glassware cleaning.    38 (9) Export time-resolved ICP-MS, pH data.   2.7.2 Offline FT-TRA The standard operating procedure for offline FT-TRA involves the following steps: (1) Experiment preparation: (a) load flow-through reactor with sample; (b) program leaching protocol in Chromeleon™ software; (c) prepare eluents as required.  (2) Calibrate pH sensor. For experiments lasting over a week, calibrate pH sensor as frequently as possible to monitor calibration curve. (3) Periodically close test tubes in fraction collector. (4) Prepare standards as required, add internal standard to fractions as required, analyze effluent fractions on ICP-MS using standard auto-sampler protocol.    39 Chapter 3: Measuring mineral dissolution kinetics using online flow-through time-resolved analysis (FT-TRA): an exploratory study with forsterite  3.1 Introduction Accurate determination of mineral dissolution rates over a wide range of conditions (pH, temperature, pCO2, etc.) is essential to investigate a wide variety of environmental processes. However, mineral dissolution rates measured in different laboratories can differ by several orders of magnitude (e.g. Rimstidt et al., 2012), and this problem is compounded by the fact that methods used to measure mineral dissolution rates tend to be labor intensive and time consuming. Methods capable of rapidly generating data on mineral dissolution rates would be beneficial to study the factors underpinning the variability of this process and to better constrain random errors. Likewise, methods providing information on dissolution rate and stoichiometry under non-steady-state or transient conditions would help elucidate the factors controlling mineral dissolution.  This study uses forsterite to explore the applicability of “flow-through time resolved analysis” (FT-TRA) with on-line measurement of dissolution products by inductively coupled plasma mass spectrometry (ICP-MS) to measure mineral dissolution rate and stoichiometry under constant and transient eluent conditions. Forsterite was chosen because its dissolution kinetics has been well characterized in a number of laboratory studies (Rimstidt et al., 2012), providing a robust benchmark to assess the result obtained with the new approach. In addition, its dissolution rate below pH 6 is relatively rapid and controlled by the activity of protons, resulting in a simple linear relationship between the logarithm of the surface normalized dissolution rate and pH (e.g. Chen and Brantley, 2000).    40 In the proposed flow-through system, the amount of mineral subjected to dissolution (10 – 20 mg) and the volume of the reactor (25 µL) are much smaller than for the mixed flow reactors commonly used to estimate mineral dissolution rate parameters, and the resulting ratio of mineral surface area to reactor volume (A/V) much higher. Because of its very small volume, much shorter eluent residence times can be achieved in the reactor (seconds versus hours), which is key to reducing the time needed to reach steady state dissolution. On-line analysis of the dissolution products by ICP-MS provides real-time information on the evolution of the mineral dissolution conditions (steady-state or not) and stoichiometry (congruent vs. incongruent dissolution). In addition, ICP-MS low detection limits and wide dynamic range allow for the simultaneous measurement of major and minor elements released during dissolution. Low detection limits could also facilitate the study of minerals dissolving very slowly. The goals of this study are to establish whether this approach can provide a means to measure mineral dissolution rates under a range of conditions more rapidly than with mixed flow reactors, whilst also providing insights into mineral dissolution under transient eluent and non-steady-state mineral surface conditions.  3.2 Background Most dissolution rate studies (including the present study) are conducted under “far-from-equilibrium” conditions. Under these conditions, mineral dissolution is affected by a number of intrinsic and extrinsic parameters (e.g. Fischer et al., 2014). Intrinsic parameters include mineral composition, crystal structure and surface nanoscale features (terraces, step edges, kinks, screw dislocations). Extrinsic parameters consist of temperature, the composition of the fluid in which the mineral is dissolving (pH, degree of undersaturation, ionic strength, pCO2, etc.) and the   41 dissolution regime (surface vs. transport-controlled). Rate laws most commonly used to describe the dissolution rate of minerals under far-from-equilibrium, surface controlled conditions can be generalized by equation (1) (e.g. Lasaga, 1984):  𝑅! = !!! !   𝐴  𝑘!  𝑎!!!  !!!!!          (3.1)  where   𝑅! = overall dissolution rate (moles s-1)   𝑚 = moles of mineral at a given time   𝑚! = initial moles of mineral present 𝑛 = factor that accounts for changes in surface area during dissolution (n = 2/3 for single uniformly dissolving spheres and cubes, Appelo and Postma, 2005). 𝐴 = surface area (m2)   𝑁! = number of parallel reaction pathways leading to dissolution    𝑘! = rate constant associated with aqueous species i (moles m-2 s-1)   𝑎! = activity of aqueous species i (dimensionless term)   𝑛! = reaction order  For a given mineral, its surface-normalized rate of dissolution (𝑅! / A; moles m-2 s-1) depends on its intrinsic properties, which dictate the rate constants (ki’s) and reaction orders (ni’s), and the activity of aqueous species that control dissolution (ai’s).  Mineral dissolution rates are most commonly measured in batch reactors or mixed flow reactors. Batch reactors are closed systems in which mineral dissolution is allowed to proceed   42 towards equilibrium. The concentrations of elements of interest are measured as a function of time and fitted to a presumed rate law. In contrast, mixed-flow reactors are open systems in which the solvent and dissolution products continuously flow in and out of the reactor. Measurements are typically made after the reactor has reached steady-state, i.e. when the eluent composition is constant, which is achieved when the rate of element release from mineral dissolution is equal to its rate of removal from the system with the effluent. A key benefit of the mixed flow reactors is that precipitation of secondary minerals can be avoided by removing the dissolution products before they reach saturation in the reactor. Hence, mixed flow reactors have generally been the preferred technique for measuring mineral dissolution rates.  Both batch and mixed flow reactors require a considerable amount of time to conduct the necessary measurements and calculate dissolution rate constants, from a few to 1000s of hours per experiment covering a single experimental condition (e.g. Brantley and Chen, 1995). Dissolution is typically monitored ex-situ, and a series of experiments need to be carried out to cover a range in experimental conditions before dissolution rate constants can be calculated. Furthermore, after compiling all available data on forsterite dissolution rate as a function of pH and temperature, Rimstidt et al. (2012) concluded that the standard deviation of dissolution rates measured in different studies is 300-fold, i.e. more than ten times larger than expected from analytical uncertainties alone (Rosso and Rimstidt, 2000). Discrepancies between studies have been mostly attributed to surface normalization, which is generally done using “geometric” surfaces, calculated from mineral grain diameter, or “BET” surface, measured by gas adsorption. These operationally defined surface areas are not necessarily proportional to the “reactive” surface area (White and Peterson, 1990) or the concentration of reactive surface sites that controls the rate of dissolution. The concentration of reactive surface sites may not be uniformly   43 distributed on mineral surfaces and may vary as mineral dissolution proceeds (e.g. Lüttge et al., 2013). Sample preparation (e.g. grinding methods) is another uncontrolled factor that could affect reactive site density and distribution and contribute to the variability in dissolution rate constants estimated in different laboratories.   3.3 Materials and methods 3.3.1 Materials Forsterite grains were obtained from Ward’s Natural Science (item #491557). Transparent mineral grains were handpicked, and ground down to a 63 – 150 µm size fraction. Sample treatment was limited to ultrasonic cleaning with acetone until a clear supernatant was obtained. The sample was dried overnight in an oven at 60 ºC. The mineral grains were examined by scanning electron microscopy (SEM), which revealed that they were generally free of adhering fine particles (Fig. 3.1). The sample had a purity of 98.8 % forsterite (quartz = 0.9 % and pyrite = 0.3%), determined by quantitative XRD. Forsterite stoichiometry was established by total digestion and elemental analysis (Mg1.81Fe0.18Ni0.01SiO4; Fo91). Surface area was measured using multi-point BET (Brunauer et al., 1938) with N2 adsorption using a Quantachrome Autosorb-1 surface area analyzer.       44  Figure 3.1 SEM image at low magnification (× 180) of the forsterite sample used in this study, after acetone cleaning.   3.3.2 Flow-through dissolution module A flow-through dissolution module was built by simplifying the original design developed by Haley and Klinkhammer (2002). A schematic of the FT-TRA module is shown in Fig. 3.2. Eluents are pumped from individual bottles (E1 – E4) by a Dionex ICS-3000 dual pump (designed using an entirely metal-free (PEEK) flow-path) equipped with time-programmable proportioning valves. The gradient pump was programmed to control the time-dependent composition of the eluent (for the experiment) or standard (for calibration). The isocratic pump continuously supplies an internal standard (indium, in bottle E5) to correct for instrumental drift associated with the on-line measurement of effluent concentration by Inductively Coupled Plasma Mass Spectrometry (Agilent 7700× quadrupole ICP-MS). Programming of the gradient   45 pump is done using the Chromeleon™ chromatography software. The generated eluent either directly joins the internal standard stream (when running standards) or passes through a flow-through cell (when running samples) before merging with the internal standard stream. Here, the flow-through cell containing the sample consists of a 13 mm diameter Millex® syringe filter with a PTFE membrane (0.2 µm, Millipore catalogue #SLLGC13NL) mounted between two computer controlled parallel solenoid valves (NResearch, Inc., #225T092C). Effluent pH measurement with a Bio-Rad DuoFlow in-line pH electrode (Catalogue number 760-2040) is optional for dissolution experiments expected to result in measureable pH changes (> 0.1 pH unit). When dissolving forsterite in our experiments, pH changes were << 0.1 pH unit and thus impossible to record with the present system. The pH reported in our experiments is the pH of the incoming solutions.  Forsterite subsamples were weighed on a Mettler Toledo XP6 microbalance, inserted into the upstream side of the filter and continuously exposed to a pre-programmed eluent sequence at room temperature. The eluent was in contact with ambient atmospheric pCO2.  The flow rate used in all experiments was 0.7 mL min-1 (± 0.0367 𝑚𝐿 𝑚𝑖𝑛!!; 95% CI), resulting in an eluent residence time of ~ 2.1 seconds in the reactor, based on a syringe filter internal volume of 25 µL. The effluent combined with the internal standard was then introduced directly into the plasma of an ICP-MS for time-resolved measurement of 24Mg, 28Si and 115In.    The main difference between this experimental set-up and mixed flow reactors is the very small size of the reactor and mineral samples, the short residence time of the eluent in the reactor and the short time of contact between eluent and mineral. Mixed flow reactors have to be well mixed because concentration measurements are conducted on a small subsample of the entire volume of the reactor. FT-TRA is not a mixed flow reactor in the sense that the reactor does not   46 need to be well mixed, because the volume of the cell is entirely renewed every 10 seconds (5 x 2.1 seconds). On this time scale, mass balance dictates that the elemental concentration of the eluent times flow rate must be equal to the rate of at which the element is released by dissolution of the mineral.   Figure 3.2 Schematic of the flow-through dissolution module. E1 – E5 represent eluent bottles, where E1 – E4 are used to generate eluents or standards and E5 is used to continuously supply an internal standard. Eluent flows top to bottom of figure. Computer controlled solenoid valves with 6 sample positions allow for the efficient analysis of   47 blanks, samples over the course of a day. When running standards, the flow-through cell is bypassed (as indicated by the red line). This flow-through module allows for time-resolved control of incoming eluent composition, and a constant, reproducible flow rate through the flow-through cell, for time-resolved analysis of the effluent stream by quadrupole ICP-MS.  3.3.3 Measurements of magnesium and silicon concentration in the eluent by on-line quadrupole ICP-MS 24Mg, 28Si, and 115In counts per second (cps) were recorded using the time resolved analysis setting on an Agilent 7700x quadrupole ICP-MS in He-mode. Counts per second obtained were corrected for dilution caused by merging eluent and internal standard streams (eluents and standards had a flow rate of 0.7 mL min-1 while the internal standard flow rate was 0.1 mL min-1). Dilution corrected counts per second were 115In normalized to correct for instrumental drift over the course of an experiment. A high concentration standard in 1% HNO3 was diluted in known time-resolved proportions with the standard matrix (1% HNO3) for calibration. The range of concentrations measured in this study was 0.6 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! to 3.5 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! for Mg and 0.3 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! to 1.8 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! for Si. Relative error on concentrations (95% confidence interval) ranged from 2.6% to 4.6% for Mg and from 4.9% to 14.4% for Si, from high to low concentration. The lower precision for Si measurements is attributed to higher blanks resulting from the use of a glass spray-chamber with the ICP-MS. Blank filters were run using identical eluents as used in sample runs. Blank corrections were 0.014 ± 0.002 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! for Mg and 0.19 ± 0.02 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! for Si. Although high, the Si blank was very stable, allowing measurements of blank-corrected Si concentrations as low as 0.2 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!!. Although the glass spray chamber is not prohibitive, in future   48 studies it will be preferable to replace it with a Teflon spray chamber. Blank-corrected concentrations were measured every 1.8 seconds and smoothed using a 5-point running average to reflect the renewal time of the eluent in the reactor.  3.3.4 Flow-through dissolution experiment design Several experiments were conducted to explore the potential and possible applications of the FT-TRA experimental set up.   3.3.4.1 Dissolution rate parameters To provide a comparison with previous studies, a dissolution experiment was specifically designed to calculate the dissolution rate parameters of forsterite (𝑘!! and 𝑛). For that purpose, dissolution rates must be measured at steady-state and under surface-controlled conditions. To establish whether the latter condition is met in our experimental design, a series of dissolution experiments was carried out using a forsterite subsample (sample 1, Table 3.1), which was exposed to a nitric acid solution at pH 2.3 under different flow rates ranging from 0.01 to 0.7 mL min-1, and maintained until steady-state [Mg, Si] concentrations were reached.    Once surface-controlled conditions were confirmed, another forsterite subsample (sample 2: BET surface area of 339.9 × 10!! m2 g-1; Table 3.1) was subjected to eluents with pH decreasing stepwise (5.6, 4.0, 3.5, 3.0, 2.3), and the durations of the pH plateaus were maintained until the on-line ICP-MS measurements indicated steady-state Mg and Si concentrations. The dissolution experiment on sample 2 lasted 234 minutes to cover all 5 pH steps. Because steady-state dissolution could not be achieved at pH 5.6, another experiment was conducted by exposing   49 a third forsterite subsample (sample 3; Table 3.1) to DIW only for a longer period of time (224 minutes).   Sample Experimental  pH Sample weight 𝑔 A × 10!! 𝑚! Experimental sample loss % 1 2.3  (constant) 8.17 × 10!! 2.10 0.30 2 5.6 – 4.0 – 3.5 – 3.0 – 2.3  (step-wise decrease) 20.24 × 10!! 6.88 0.28 3 5.6  (constant) 22.01 × 10!! 7.48 0.20 4  2.3 – 5.6 (transient/constant) 17.78 × 10!!  4.58 0.32 5 2.3 – 5.6 (transient/constant) 13.58 × 10!!  3.49 0.33 6 2.3 – 5.6 (transient/constant) 12.85 × 10!! 3.31 0.21 7 2.3 – 5.6 (transient/constant) 20.27 × 10!! 5.21 0.22 8 2.3 – 5.6 (transient/constant) 17.78 × 10!! 4.58 0.26 Table 3.1 Weight (g) and surface area (m2) of forsterite used in the “dissolution regime” experiment (sample 1), the “rate parameters” experiment (sample 2, 3) and the “replication” and “transient” experiments (sample 4 – 8). Surface areas are calculated from the sample weight and BET surface measurements (sample 1, 4 – 8: 2.57 × 10!! m2 g-1; sample 2, 3: 3.39 × 10!! m2 g-1). Sample loss is estimated by integrating the total moles of Mg released during each experiment.  3.3.4.2 Dissolution under transient conditions and replication Replicate analyses were conducted with five subsamples from the same forsterite sample. In each of the five experiments, a sample was weighed (samples 4 – 8, Table 3.1) and subjected first to a   50 continuous flow of distilled, de-ionized water (DIW; pH = 5.6) for ~1950 s (32.5 min). Eluent acidity was then gradually ramped up by mixing HNO3 (1%) to decrease eluent pH to 3.3, which was maintained for 1200 seconds (20 min). Afterwards, eluent acidity was ramped up further to decrease eluent pH to 2.3, which was maintained for ~1400 s (23 min) before raising the eluent pH back to pH = 5.6 (DIW).  Effluent Mg and Si concentrations were continuously measured, providing dissolution rate data over a range of pH (5.6 to 2.3) in less than 9000 seconds (150 min). The goals of this experiment are to assess reproducibility between subsamples and to document transient dissolution patterns.   3.3.5 Calculation of surface normalized dissolution rates and dissolution rate parameters (𝒌𝑯! ,𝒏𝑯!) Surface area normalized forsterite dissolution rates (moles m-2 s-1) as a function of pH are calculated by multiplying the time resolved concentrations (moles L-1) of Mg or Si by the eluent flow rate (𝑄; L s-1) and dividing by the sample surface area (𝐴; m2), taking into account the stoichiometry of the forsterite used in the experiment (Mg1.81Fe0.18Ni0.01SiO4; Fo91): Surface normalized dissolution rates (moles forsterite m-2 s-1) =  ([𝑀𝑔] 1.81  (𝑚𝑜𝑙𝑒𝑠 𝐿!!) × 𝑄 (𝐿 𝑠!!)) / 𝐴 (𝑚!)       (3.2a) ([𝑆𝑖] (𝑚𝑜𝑙𝑒𝑠 𝐿!!) × 𝑄 (𝐿 𝑠!!)) / 𝐴 (𝑚!)        (3.2b)  To describe forsterite dissolution in our experimental set up, equation (3.1) can be simplified. It has been previously determined that forsterite dissolution between pH 1 and 6 only depends on the activity of H+  (Np = 1). In addition, since only a very small fraction of the sample is dissolved during the experiments (Table 3.1), 𝑚 𝑚! ≈ 1. Thus, equation (3.1) can be simplified to:   51 𝑅!  (moles s-1) = 𝐴 × 𝑘!! × 𝑎!!!          (3.3) which results in a simple linear relationship between the logarithm of the surface normalized forsterite dissolution rate (𝑟 = 𝑅!/ 𝐴) and pH:  log (𝑟) = log  𝑘!! −  𝑛 pH         (3.4)  In parallel, an automated inverse modeling approach was developed to quantify dissolution rate parameters. Although not necessary for the simple rate law controlling forsterite dissolution, this approach would be useful for minerals with more complex dissolution rate equations and is described in Appendix A.  3.4 Results and discussion 3.4.1 Determination of forsterite dissolution parameters 3.4.1.1 Forsterite dissolution regime Dissolution under surface-controlled conditions, as opposed to transport-controlled conditions, is a pre-requisite to obtain dissolution rate constants that are intrinsic to the mineral studied and independent of experimental conditions during measurement (e.g. Colombani, 2008 and references therein). Sample 1 (Table 3.1) was dedicated to evaluating the dissolution regime of forsterite in our experimental set up at pH 2.3. Steady state [Mg, Si] concentrations reached at constant eluent pH 2.3 but at different flow rates increase linearly with the residence time of eluent in the reactor, indicating that forsterite dissolution rate is constant and independent of flow rate (Mg: Fig 3.3A; Si: Fig 3.3B). This is taken as an indication that dissolution in our experimental set up at pH 2.3 is surface-controlled, i.e. the rate of detachment of dissolving species from the mineral surface is slower than the rate at which dissolved products diffuse to the bulk solution, and the rate of diffusion of H+ from the bulk solution to the mineral surface is   52 faster than the rate of consumption of protons by the surface reaction. A telltale sign of transport control would be a decrease in dissolution rates with decreasing flow rate, indicating that transport through the thickening diffusive boundary layer (DBL) at the interface between solution and grain surface limits, or at least contributes to limiting dissolution rate (transport or mixed controlled regimes). We thus conclude that forsterite dissolution is surface-controlled at pH 2.3.  A more complete investigation of dissolution regime using FT-TRA is presented in chapter 4 (De Baere et al., submitted).     53  Figure 3.3 Average steady state Mg (A) and Si (B) concentrations (sample 1) versus eluent residence time (calculated by dividing flow cell volume (25 µL) by varying eluent flow rates). R2 = 0.999Steady state Mg (x 10-5  moles L-1)0123456R2 = 0.999Steady state Si (x 10-6 moles L-1)05101520253035Residence time (s)0 50 100 150AB  54 3.4.1.2 Forsterite dissolution under acidic conditions To validate our proposed methodology for the determination of dissolution rate parameters, the time-resolved [Mg, Si] concentrations and Mg/Si molar ratios measured during the dissolution of sample 2 (Fig. 3.4A) were used for comparison with previous studies. Mg and Si concentrations are initially high and decreased sharply during the first dissolution step in DIW at pH 5.6 without reaching steady-state after nearly 90 minutes of dissolution. Such dissolution pattern (parabolic kinetics) is attributed to the faster dissolution of high surface free energy sites that have been produced during sample preparation (Chou and Wollast, 1984). Subsequently, Mg and Si concentrations increase sharply at each step when pH decreases, and continue to increase more gradually thereafter, suggesting a gradual increase in active surface area (consistent with the increase in surface area when forsterite is immersed in acidic solutions for 24 hours; Pokrovsky and Schott, 2000). Concentrations eventually plateau at each pH step. To calculate dissolution rate parameters, we used the on-line ICP-MS data to monitor the time needed to reach or approach steady-state dissolution before changing the pH of the eluent. Sections with constant Mg and Si concentrations were then selected on each plateau (Fig. 3.4A) and averaged to quantify 𝑘!! and 𝑛!! by weighted linear regression (Fig. 3.4B), following equation (3.4). Weights were allocated according to the number of data points collected at constant Mg and Si concentrations. Since steady state dissolution could not be reached in DIW, only data obtained at pH 4.0, 3.5, 3.0 and 2.3 were considered.  Dissolution rate parameters obtained in this experiment are in reasonable agreement with previous studies, and fit best with the results from Pokrovsky and Schott (2000) (Table 3.2; Fig. 3.5). The dissolution rate measured at pH 4.0 deviate slightly from the linear trend, particularly when calculating forsterite dissolution rates from Mg release rates (Fig. 3.4B). This may indicate   55 that at this pH, although Mg concentrations appeared to have reached a plateau (Fig. 3.4A), these concentrations may still have been decreasing very slowly. This is further supported by the relatively high Mg/Si also obtained at pH 4.0 (2.13 ± 0.16) compared to forsterite stoichiometry (1.81 ± 0.07), indicating that the Si-rich surface leached layer was still deepening. Si-based dissolution rates deviate less (Fig. 3.4B), suggesting that under acidic conditions and when pH is decreasing Si release rates reach or approach steady state faster than Mg release rates. Nonetheless, these results confirm the suitability of FT-TRA and on-line ICP-MS to generate meaningful dissolution rate data relatively quickly (approximately 4 hours (240 min) for 4 pH steps and including the initial DIW step), which would allow, at least for minerals reaching steady state dissolution rapidly, replicate analyses, dissolution process studies, and the generation of a database suitable for statistical analysis.     56  Figure 3.4 (A) [Mg] (5-point running average in black), [Si] (5-point running average in gray) and Mg/Si molar ratio (red) in the eluent resulting from the dissolution of sample 2. Forsterite was subjected to acidic solutions (DIW [Mg, Si] (x 10-7  moles L-1)ABMg Si Mg / Si5.6 4.03.53.02.3051015202530Mg / Si (molar)00.51.01.52.02.53.03.54.0Time (s)0 2000 4000 6000 8000 10000 12000 14000Mg release rate 95% CI (Mg release rate)Si release rate 95% CI (Si release rate)log r forsterite (moles m-2 s-1)9.08.58.07.57.0pH2.0 2.5 3.0 3.5 4.0 4.5 5.0  57 and HNO3) of decreasing pH (5.6, 4.0, 3.5, 3.0, 2.3) maintained until steady state dissolution was reached or approached. Time step for Mg and Si measurements by ICP-MS is 1.8 seconds, hence the 5-point average is shown to represent the eluent reactor renewal time. The dashed lines indicate the stoichiometric Mg / Si ratio of forsterite (1.81 ± 0.07). (B) Forsterite dissolution rates estimated from average [Mg] (filled circles), and average [Si] (open diamonds) concentration plateaus at each pH. [Mg] were divided by 1.81 to take into account forsterite stoichiometry. Average [Mg, Si] were multiplied by the effluent flow rate (0.70 ± 0.01 ml min-1) and divided by BET surface area (Table 3.1). Weighted (according to the number of data points collected at each plateau) linear regression provides the reaction order (slope) and rate constant (intercept at pH = 0). 95 % confidence intervals shown are based on the weighted linear regression.                     58 Study rate constant 𝑘!! × 10!! moles m-2 s-1 reaction order  𝑛!! rate constant 𝑘!! × 10!! moles m-2 s-1 reaction order  𝑛!!  based on [Mg] based on [Si] this study (sample 2) 𝟐.𝟏𝟓 − 𝟐.𝟒𝟖(𝐚) 𝟎.𝟒𝟔 ± 𝟎.𝟎𝟏(𝒃) 𝟐.𝟒𝟎 − 𝟐.𝟕𝟑(𝐚) 𝟎.𝟒𝟗 ± 𝟎.𝟎𝟏(𝒃) Blum and Lasaga, 1988   0.52 − 1.78(!) 0.50 ± 0.08(!) Wogelius and Walther, 1991 0.19 − 5.09(!)  0.53 ± 0.16(!) 0.26 − 2.43(!) 0.53 ± 0.12(!) Pokrovsky and Schott, 2000 1.28 − 2.90(!)  0.51 ± 0.04(!) 1.97 − 3.91(!) 0.53 ± 0.04(!) Rosso and Rimstidt, 2000 0.93 − 1.22(!)  0.50 ± 0.02(!) 0.78 − 1.03(!) 0.44 ± 0.02(!) Olsen, 2007   0.70 − 3.11(!) 0.53 ± 0.10(!) Olsen and Rimstidt, 2008 0.46 − 1.39(!)  0.44 ± 0.09(!) 0.86 − 1.39(!) 0.48 ± 0.04(!) Luce et al., 1972 8.29 − 42.92(!)  0.32 ± 0.10(!) 8.94 − 102.05(!) 0.64 ± 0.45(!) Grandstaff, 1986 0.13 − 60.97(!)  1.09 ± 0.35(!) 0.12 − 64.36(!) 1.10 ± 0.35(!) Golubev et al., 2005 0.12 − 0.34(!)   0.16 ± 0.06(!) 0.16 − 0.41(!)  0.19 ± 0.05(!) Table 3.2 Dissolution rate constant 𝑘!! and reaction order 𝑛!!quantified in this study (based on [Mg] and [Si]), compared to other studies compiled by Rimstidt et al., 2012 (pH ≤ 6, log r BET surface area normalized Mg, Si data)  (a) range associated with 95% confidence interval (2σ) of weighted linear regression intercept. Weighted linear regression is based on number of data points associated with each [Mg, Si] steady state plateau (𝑛!"!!.! =241;  𝑛!"!!.! = 451;  𝑛!"!!.! = 644; 𝑛!"!!.! = 491) (b) error = 95% confidence interval (2σ) associated with weighted linear regression slope (c) range associated with 95% confidence interval (2σ) of linear regression intercept (d) error = 95% confidence interval (2σ) associated with linear regression slope    59  Figure 3.5 Logarithm of the surface normalized forsterite dissolution rate (moles m-2 s-1) based on the Mg (A) and Si (B) release rates versus pH (sample 2). The dashed lines represent the 95 % confidence interval based on the uncertainties of the rate parameters (slope and intercept) obtained from the weighted linear regression (Table 3.2) Luce et al., 1972Grandstaff, 1986  Golubev et al., 2005 ABsample 2 Wogelius and Walther, 1991 Pokrovsky and Schott, 2000Rosso and Rimstidt, 2000 Olsen and Rimstidt, 2008log r forsterite (moles m-2 s-1)11.010.510.09.59.08.58.07.57.06.56.0sample 2 Wogelius and Walther, 1991 Pokrovsky and Schott, 2000Rosso and Rimstidt, 2000 Olsen and Rimstidt, 2008Olsen, 2007Blum and Lasaga, 1988log r forsterite (moles m-2 s-1)11.010.510.09.59.08.58.07.57.06.5pH1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0Luce et al., 1972Grandstaff, 1986  Golubev et al., 2005   60 and extrapolated to pH 1 – 6. Results from previous studies compiled in Rimstidt et al. (2012) are shown for comparison. Color-coded lines are obtained by linear regressions on Mg, Si data obtained at pH ≤ 6. Gray circles indicate studies yielding reaction orders very different from 0.5 (Luce et al., 1972; Grandstaff, 1986; Golubev et al., 2005; Table 3.2).  3.4.1.3 Forsterite dissolution in DIW (pH 5.6) Since steady state concentrations could not be reached at pH = 5.6 after ~ 90 minutes, an additional dissolution experiment with DIW only was conducted for 224 minutes (sample 3, Fig. 3.6). Even with this extended period of exposure, Mg and Si release rates at the end of this experiment were still decreasing, although Mg/Si reached forsterite stoichiometry, indicating congruent dissolution, after approximately 6000 s (100 min). Unlike the gradual increase in dissolution rates occurring when pH decreases (Fig. 3.4A), which could be attributed to a gradual increase in surface–active area with the gradual deepening of the Si-rich surface layer, the gradual decrease in the dissolution rate observed in DIW may be, instead, an indication of gradual surface passivation once stoichiometric dissolution is achieved and the depth of Si-rich surface layer has stabilized. This could be a result of the gradual polymerization of silica tetrahedra in the leached layer (Pokrovsky and Schott, 2000). Once the silica-rich layer depth has reached steady-state, the rate limiting step for forsterite dissolution is the release of silica from the surface altered layer (Liu et al., 2006), accounting for the gradual decrease in Mg concentration. Notwithstanding stoichiometric dissolution, the rates of Mg and Si release at the end of this experiment are still significantly higher (log 𝑟 = -8.5) than predicted with the dissolution rate parameters derived at lower pH (Fig. 3.4B) and in previous studies (Fig. 3.5). This observation indicates that it takes more than 224 minutes to reach steady state dissolution of   61 forsterite at pH 5.6. It is well established that reaching steady state dissolution of forsterite at higher pH takes much longer than under more acidic conditions (Pokrovsky and Schott, 2000). This may be because of slow surface passivation, or because forsterite dissolution is not surface-controlled at this pH (Rimstidt, 2015). In fact, Rimstidt et al. (2012) document a change in dissolution reaction mechanism going from low to high pH at 25 ºC and split their dataset at pH 5.6. Regardless of the mechanism responsible, for forsterite dissolution, the use of FT-TRA and on-line ICP-MS measurements, as applied in this experiment, may be limited to lower pH conditions (section 3.4.1.2). Indeed, if the time needed to reach steady state dissolution exceeds 12 hours (720 minutes), it may be too costly to study mineral dissolution with our set-up by direct on-line ICP-MS analysis.  Figure 3.6 Concentrations of [Mg] (5-point running average in black), [Si] (5-point running average in gray) and Mg/Si molar ratio (red) in the eluent resulting from the dissolution of sample 3 in DIW. The time step for Mg and Si Mg Si Mg / Si (molar ratio) [Mg, Si] (x 10-7  moles L-1)02468101214161820Mg / Si (molar)00.51.01.52.02.53.03.54.0Time (s)0 5000 10000  62 measurements by ICP-MS is 1.8 seconds, hence the 5-point average is shown to represent the eluent reactor renewal time. The dashed lines indicate the stoichiometric Mg / Si ratio of forsterite (1.81 ± 0.07).  3.4.1.4 General applicability of FT-TRA to estimate mineral dissolution parameters The last conclusion in the previous section raises questions about the broader applicability of the proposed approach. One of the main advantages of FT-TRA over conventional flow-through reactors for measuring mineral dissolution parameters is the shorter time required to reach steady-state concentrations of dissolved products. In our experiment, quasi steady-state concentrations are reached in less than one hour at pH ≤ 4 (Fig. 3.4A), while it took 8 to 75 hours (480 to 4500 minutes) during the extensive forsterite dissolution experiments conducted by Pokrovsky and Schott (2000) under similar pH conditions, but using a conventional mixed flow reactor. The main differences between the two experimental designs are the larger surface area to volume ratio (A/V) and the much smaller volume of the reactor in our set up (25 µL vs. 250 mL). While there are no a priori reasons to expect that higher A/V would decrease the time needed to reach steady state concentrations, the much smaller volume of the reactor is an important characteristic, which can largely explain this observation. The time necessary to reach steady-state concentrations in a flow reactor depends on both the times it takes for the mineral surface to reach steady dissolution (Tdiss), and the time it takes to renew the solution within the reactor (Tflush). With their comparatively large volumes, when mixed flow reactors are used to measure dissolution rates of forsterite or other minerals reaching steady state dissolution relatively quickly, Tflush >> Tdiss and the time needed for the dissolution experiment to reach steady-state concentrations is largely dictated by the residence time of the eluent in the reactor (e.g. 6 – 50 hours (360 – 3000 min) in the pH < 4 experiments conducted by Pokrovsky and Schott (2000).   63 The eluent residence time in our flow-through cell is only 2.1 seconds (i.e. Tflush << Tdiss); therefore the time needed for dissolution experiments to reach steady-state concentrations is entirely dictated by the properties of the mineral surface.  The time advantage of FT-TRA is clearly apparent in our study because forsterite reaches steady-state dissolution rapidly, at least at pH ≤ 4. If other minerals or conditions (e.g. forsterite dissolution in DIW) require longer time periods (more than 720 min) to reach steady-state dissolution, on-line ICP-MS analysis may become prohibitively expensive. In this case, we suggest two possible alternatives: (1) pre-conditioning the mineral off line prior to on-line ICP-MS effluent measurement or (2) collection of time-series samples over a much longer period using an automated fraction collector, followed by off-line ICP-MS analysis. The choice between these two alternative approaches would depend on the goal of the experiment, and both approaches could be combined. If the goal were to determine dissolution rate constants, fraction collection would be particularly suited, since the experiment could be continued over much longer periods of time with minimal attendance. The advantage of high analytical time resolution would be lost but this is of no consequence for this particular application. Of course, for minerals with Tdiss >> Tflush, mixed flow reactors may be equally suited to measure dissolution rate parameters, but the automation of our set-up would still facilitate this task. On the other hand, if the goal of the experiment is to study rapidly evolving transient dissolution events, the high time resolution afforded by the present approach is needed, and neither FT-TRA with fraction collector nor conventional mixed flow reactor could be used effectively. In this case, FT-TRA and on-line ICP-MS would be the tool of choice to document transient dissolution of minerals or minerals having been subjected to various pre-treatments. For instance, off-line preconditioning of minerals under varying conditions prior to on-line ICP-MS measurements could prove   64 particularly useful to study the factors controlling surface passivation or exfoliation, two important processes which regulate weathering and carbonation of minerals (e.g. Olsson et al., 2012).  3.4.2 Replicate analyses and dissolution under transient conditions 3.4.2.1 Reproducibility of forsterite dissolution under identical conditions The shorter experimental durations afforded by our small volume flow-through cell facilitate replicate experiments to assess the inherent variability of forsterite dissolution between closely related subsamples. Results obtained with one of the five replicate subsamples (sample 6) subjected to the same dissolution sequence are illustrated in Fig. 3.7A (results from the other replicates are reported in Appendix A: Fig. A.1). Mg and Si concentrations follow a similar pattern as in the previous experiment. They start high and sharply decrease in DIW, increase as incoming effluent pH decreases and drop when pH increases. pH changes associated with forsterite dissolution calculated from the concentration of Mg2+ in the effluent are < 0.004 pH units and therefore negligible. The pH values reported in our experiments (Fig. 3.7B) are thus the pH of the incoming solutions.  Multiplying the time-resolved concentrations obtained with the five replicates by the effluent flow rate, which was identical for the five experiments, and dividing by the surface area of each subsample (Table 3.1) reveal a near 2-fold variability in the surface normalized rate of dissolution of the replicates (Fig. 3.8). In addition, not only are the surface-normalized dissolution rates generated during the 5 experiments different, but the rate at which steady-state dissolution is approached also varies between subsamples. Differences in surface-normalized dissolution rates between subsamples may reflect variations in reactive surface area, while the   65 gradual increase in dissolution rate at constant pH 2.3 and/or pH 3.3 may reflect a gradual increase in reactive surface area with time in 4 of the 5 subsamples. The variability observed in this experiment is surprising considering that the subsamples were treated and dissolved under the same conditions, and further points to the importance of replicating dissolution experiments (Rosso and Rimstidt, 2000) and the need to fully understand the underlying reasons for this variability.   pH23456Time (s)0 2000 4000 6000 80003.32.3BMg Si Mg / Si (molar ratio) [Mg, Si] (x 10-7  moles L-1)02468101214161820Mg/Si (molar)00.51.01.52.02.53.03.54.0A  66 Figure 3.7 (A) Concentrations of [Mg] (5-point running average in black), [Si] (5-point running average in gray) and Mg/Si molar ratio (red) measured in the effluent resulting from the dissolution of sample 6. The dashed lines indicate the stoichiometric Mg / Si ratio of forsterite (1.81 ± 0.07). Figures for samples 4, 5, 7 and 8 are supplied in Appendix A: Fig. A.1. (B) Changes in incoming eluent pH (≈ pH in flow-through reactor, as changes in pH resulting from forsterite dissolution in the flow cell are below detection limits).    67  Figure 3.8 Forsterite dissolution rates under varying pH estimated from time-resolved Mg (upper panel A, divided by 1.81 to take into account forsterite stoichiometry) and Si (lower panel B) concentration measured by on-line ICP-05101520253035404550Time (s)0 2000 4000 6000 8000r forsterite (x 10-9  moles m-2 s-1)BSample 4 Sample 5Sample 6Sample 7 Sample 805101520253035404550r forsterite (x 10-9  moles m-2 s-1)ApH = 2.3pH = 3.3pH = 3.3pH = 2.3  68 MS, multiplied by the eluent flow rate (0.70 ± 0.01 mL min-1) and divided by the BET surface area for sample 4 – 8 (Table 3.1). Time-resolved eluent pH provided in Fig. 3.7B.   3.4.2.2 Forsterite dissolution stoichiometry under transient eluent conditions FT-TRA analysis reveals that the stoichiometry of forsterite dissolution display a striking variability associated with eluent pH changes (Fig. 3.4A; 3.7A). Mg/Si molar ratio of the eluent is initially much higher and gradually decreases towards the stoichiometric ratio of the forsterite sample (1.81 ± 0.07). When the pH of the eluent gradually decreases, Mg/Si often increases sharply and relaxes back towards the forsterite stoichiometric ratio. In contrast, when the pH of the eluent increases (at t = 7500 s (125 min); Fig. 3.7A; Appendix A: Fig. A.1), invariably, Mg/Si drops below the stoichiometric ratio and gradually rises thereafter. In addition, the time resolved concentrations generated at pH 3.3 with sample 6 (Fig. 3.7A) documents a large and sudden change in Si concentrations, which is clearly reflected in the Mg/Si time series data. Likewise, smaller but significant and rapid changes in Mg concentration are observed during the first dissolution step in DIW. These peaks indicate that mineral dissolution is not necessarily a continuous process, but may be punctuated by abrupt events, which require the temporal resolution afforded by our approach to be documented.  3.4.2.2.1 Monitoring the surface leached layer thickness during transient eluent conditions The preferential release of Mg during the early stage of forsterite dissolution at low pH has been widely reported (e.g. Luce et al., 1972, Wogelius and Walther, 1991, Pokrovsky and Schott, 2000a, Rosso and Rimstidt, 2000), and attributed to the formation of a Si-rich leached layer due   69 to exchange reaction between 𝑀𝑔!! and 𝐻! (e.g. Pokrovsky and Schott, 2000a). Higher Mg/Si ratios of the dissolved products, compared to the Mg/Si ratio of the dissolving forsterite, indicate preferential release of Mg and deepening of the Si-rich surface layer. Quantum mechanical simulations by Liu et al. (2006) suggest that under acidic conditions, oxygen atoms at a specific lattice position (µ3-O) become protonated, which lengthens and weakens the bonds holding Mg. The rate of release of Mg by this reaction is faster at kinks or edges. This mechanism could account for the observed high Mg/Si released when forsterite is first exposed to more acidic solutions. On the other hand, eluent ratios lower than forsterite stoichiometry indicate preferential dissolution of Si and partial erosion of the Si-rich layer. Changes in the mean thickness of the Si-rich leached surface layer as dissolution proceeds are calculated following Chou and Wollast (1984): Σ𝑀𝑔(!) =  Σ 𝑀𝑔 !  × 𝑄 × Δ𝑡        (3.5a) Σ𝑆𝑖(!) =  Σ 𝑆𝑖 !  × 𝑄 × Δ𝑡        (3.5b) where Σ𝑀𝑔(!) and Σ𝑆𝑖(!) are the moles of Mg and Si dissolved since the beginning of the experiment to time 𝑡,  𝑀𝑔 ! and  𝑆𝑖 ! are the concentrations (moles L-1) measured at time 𝑖, Δ𝑡 is the time step of ICP-MS analysis (1.8 seconds) and 𝑄 is the effluent flow rate (L s-1) 𝑉!" (!) =  1.81 × Σ𝑀𝑔 ! × 𝑀𝑉!"#$%&#'%&      (3.6a) 𝑉!" (!) =  Σ𝑆𝑖 ! × 𝑀𝑉!"#$%&#'%&       (3.6b) where 𝑀𝑉!"#$%&#'%& is the molar volume of forsterite (43.79 cm3 mol-1; Cemič, 2005), 𝑉!" (!) and 𝑉!" (!) are the volumes of forsterite dissolved as a result of Mg and Si release, respectively (taking into account forsterite stoichiometry). 𝑍!"(!) = 𝑉!"(!) 𝐴         (3.7a)   70 𝑍!"(!) = 𝑉!"(!) 𝐴         (3.7b) where 𝐴 is the surface area of the sample, 𝑍!"(!) and 𝑍!"(!) are the thickness of the layer of forsterite dissolved by releasing Mg and Si respectively.  If the Mg/Si ratio of the eluent is higher than the stoichiometry of forsterite, 𝑍!"(!) > 𝑍!"(!), and 𝑍!"(!) −  𝑍!"(!) is the thickness of the Mg-leached surface layer at time 𝑡 (Fig. 3.9), if we assume that the leached layer has lost 100% of its Mg and the molar volume of the silica layer has retained the molar volume of forsterite. If we assume that there is a linear Mg/Si gradient in the leached layer, from zero at the top to forsterite stoichiometry at the bottom, estimated leached layer thicknesses would have to be doubled (Chou and Wollast, 1984). Although the calculated thickness is only an average over the entire surface area of the mineral and the assumption that the silica leached layer retain the molar volume of forsterite may not be correct (e.g. Kim et al., 2005), this simple calculation provides an useful visualization of the formation process of the leached layer (Fig. 3.9) and should estimate the true average thickness within of factor of ~ 2. The evolution of leached layer thickness depends both on Mg/Si and the rate of dissolution. The sample reported on Fig. 3.9 never reached stoichiometric dissolution (Mg/Si = 1.81) during the first 7000 s (117 min) of the experiment, therefore the thickness of the leached layer continuously increased over that time period. If we assume that it lost all its Mg and remained isovolumetric, the Si-rich layer reached an apparent mean thickness of 2 ångströms within the first few minutes of dissolution in DIW and then deepened much more gradually to 3 ångströms during the following 30 minutes (Fig. 3.9; similar figures for the four other replicates are provided in Appendix A: Fig. A.2). The rapid initial increase in thickness was due to the   71 initially high Mg/Si and dissolution rate, and the more gradual deepening that followed reflects near congruent dissolution and slower dissolution rates (Fig. 3.7A). At pH 3.3, the leached layer deepened faster because of higher dissolution rates and higher (but gradually decreasing) Mg/Si. At pH 2.3, Mg/Si was closer to congruency but the much higher dissolution rates (Fig. 3.8) still resulted in a rapid increase in thickness, reaching a maximum of 10 ångströms at the end of the pH 2.3 plateau. When reverting to DIW at the end of the dissolution sequence, the thickness of the leached layer decreased a little and stabilized, indicating a very slight erosion of Si-rich layer, corresponding to eluent Mg/Si below forsterite stoichiometry. This erosion was very small because dissolution stoichiometry remained below that of forsterite for a short period only, at a time when the rate of dissolution was slow.                                                                                                                                           Figure 3.9 Changes in the thickness of the Mg leached surface layer (solid red line) in sample 6 (thicknesses for samples 4, 5, 7, 8 are reported in Appendix A: Fig. A.2). The dashed red lines represent range in Mg – Si leached Mg / Si (molar ratio) Leached layer thickness (Å)Mg / Si (molar ratio)00.51.01.52.02.53.03.54.0Leached layer thickness (Å)024681012Time (s)0 2000 4000 6000 8000  72 layer thickness calculated using the 95% confidence interval of the forsterite stoichiometry (Mg/Si = 1.81 ± 0.07). Mg/Si molar ratio (calculated using the 5-point running [Mg, Si] average) is shown in black. Time resolved eluent pH is reported in Fig. 3.7B.  3.4.2.2.2 Sporadic exfoliation of the Si-rich layer Brief but pronounced episodes of rapid Si release are observed during the dissolution of forsterite (Fig. 3.8B). The most prominent of these events was observed during the dissolution of sample 6, resulting in a sharp decrease in Mg/Si (Fig. 3.9: t = 3420 seconds (57 min)). These features are tentatively interpreted as sporadic exfoliations of a section of the Si-rich layer from which all the Mg had been previously stripped. The high temporal resolution afforded by the short eluent residence time in the reactor (2.1 s) and time-resolved ICP-MS analysis (1.8 s) provides interesting details on the most prominent exfoliation episode illustrated in Fig. 3.9, which lasted ~ 20 s, and is thus well documented by about 12 consecutive ICP-MS data points (Fig. 3.10). In particular, while Mg was not released above background during the peak of the exfoliation, a small Mg peak followed (Fig. 3.10B), possibly reflecting exposure of the new unaltered forsterite surface, which released Mg at a faster rate for a brief period.  FT-TRA data also provide a means of estimating the extent of exfoliation, within the constraints of a few simplifying assumptions. Fig. 3.10 indicates that the section exfoliated during this episode consisted exclusively of silica. Integrating the amount of silicon released above background during this event (2.8 × 10!!" 𝑚𝑜𝑙𝑒𝑠) and assuming that the leached layer kept the molar volume of forsterite, we can calculate the volume of silica exfoliated (1.24 × 10!! 𝑐𝑚!). Since we know the average thickness of the leached layer at the time of the   73 exfoliation (4.5 × 10!! cm, Fig. 3.9), the surface area of exfoliation can be calculated (0.275 cm2), which represents 8% of the total surface area of the sample.  Exfoliation is triggered when the Si-enriched layer reaches a critical thickness (e.g. Jarvis et al., 2009) and has been attributed to molar volume differences between the forsterite crystal and the Si-enriched layer (e.g. Kim et al., 2005). Interestingly, the eluent generated by sample 6 also had the highest Mg/Si molar ratio over the entire run, indicating the formation of a more extensive silica rich layer, which also resulted in a more pronounced decrease in Mg/Si when raising the eluent pH back to 5.6. In contrast, sample 7 had the lowest Mg/Si and released relatively little Si in the last transition to DIW (see Appendix: Fig. A.1). This seems to indicate that the development of the surface Si-enriched layer differed among samples 4 – 8, even though they were treated under identical conditions. FT-TRA could thus also prove very useful to study the factors controlling exfoliation, which is a major process exposing fresh mineral surfaces necessary to optimize mineral carbonation and carbon sequestration (e.g. Béarat et al., 2006).    74  Figure 3.10 (A) Exfoliation event observed in sample 6 as an abrupt increase in [Si]; (B) Same event on an expanded time scale showing the temporal sequence of changes in [Si] (y-axis on right) preceding a smaller [Mg] peak (y-axis on left).  3.5 Summary and future prospects The results from this exploratory study validate FT-TRA and on-line ICP-MS as a meaningful and versatile new experimental method to study mineral dissolution. We have identified a Mg (stoichiometry normalized) Si [Mg, Si] (x 10-7  moles L-1)05101520Time (s)3000 3200 3400 3600 3800 4000[Mg] (x 10-7  moles L-1)AB2.02.22.42.62.83.03.23.43.63.84.0Time (s)3410 3420 3430 3440 3450[Si] (x 10 -7 moles L -1)05101520Mg (stoichiometry normalized) Si   75 number of advantages, and several applications, which could benefit from using this approach. The advantages are (1) faster attainment of steady state eluent concentration, which stems from the small reactor volume and extremely short residence time of the eluent in the reactor; (2) high temporal resolution resulting from the small volume of the reactor and on-line time-resolved ICP-MS analysis; and (3) simultaneous analysis of multiple elements occurring over a very wide range of concentrations (ppm to ppb). Several aspects relevant to mineral dissolution studies can benefit from these characteristics: (1) faster achievement of steady-state concentrations significantly accelerates the determination of dissolution rate parameters, allowing for replications and statistical analysis, and facilitating investigations of the influence of multiple variables; (2) high-temporal resolution documents abrupt (e.g. exfoliation) and gradual (e.g. leached surface formation, surface passivation) changes in dissolution rates and dissolution stoichiometry, which provide insight into dissolution processes and evolution; (3) simultaneous multi-element analysis over a wide range of concentrations facilitates the measurement of dissolution stoichiometry, including minor elements, and facilitate the study of slowly-dissolving minerals. Regarding future prospects, one of the major limitations in quantifying intrinsic mineral dissolution rates is surface normalization, and it has been recently suggested that a paradigm shift may be required to effectively address this question (Lüttge et al., 2013). Nanoscale imaging of mineral surface topography (Atomic Force Microscopy (AFM) or Vertical Scanning Interferometry (VSI)) has revealed that mineral dissolution rates depend on the density of mineral surface features, such as steps, kinks and screw dislocations, which act as active surface sites (e.g. Ruiz-Agido and Putnis, 2012). This insight lead to rationalizing the large range of dissolution rate constants, which have been carefully measured from bulk solution   76 measurements, as reflecting stochastic changes in the surface density of these active sites. For more accurate descriptions, dissolution rates must therefore be adjusted to the number and reactivity of active surface sites. The solution proposed by Lüttge et al. (2013) is to replace dissolution rate constants by spectra of rate constants, which would capture the variability in the energy level, and density of active surface sites. If the density and types of active surface sites were amenable to predictive modeling and anchored to a bulk mineral property that can be measured to determine ab-initio conditions, this information could then be applied to any conditions in the laboratory or in the field. Progress will require combining different modeling and measurement techniques, in which FT-TRA could contribute significantly. As already demonstrated by Duckworth and Martin (2003), AFM can be combined with chemical analysis to simultaneously document changes in surface topography and bulk solution composition during mineral dissolution. The FT-TRA module presented here could be a perfect complement to surface imaging. The flow cell used in this study could be replaced by the flow-through cell used with AFM (Duckworth and Martin, 2003). The multiple data sets produced by this approach could help assessing whether the evolution of reactive surface site during mineral dissolution can be predicted and detected with the high temporal resolution and sensitivity afforded by FT-TRA/ICP-MS to yield intrinsic dissolution rate constants normalized to active site distribution, possibly resolving the present surface normalization conundrum.   77 Chapter 4: Empirical determination of mineral dissolution regimes using flow-through time-resolved analysis (FT-TRA)  4.1 Introduction 4.1.1 Dissolution regimes Accurately predicting mineral dissolution rates is essential for the investigation of many processes in environmental geochemistry, material science, and a wide range of related fields. One approach to reach this goal has been to measure mineral dissolution rates in laboratory experiments, from which dissolution rate parameters can be deduced and then used to predict rates of mineral dissolution under a wide range of conditions (e.g. White and Brantley, 1995). To obtain meaningful dissolution rate parameters that can be applied to a range of environmental conditions, experimentally measured rates must reflect the intrinsic properties of the dissolution process, i.e. the rate of detachment of dissolving species from the surface of the mineral (e.g. Berner, 1978; Compton and Unwin, 1990; Morse and Arvidson, 2002; Morse et al., 2007). However, rates measured in laboratory experiments may be affected by the hydrodynamic conditions under which the dissolution measurements were conducted (e.g. Sjöberg and Rickard, 1984), and thus may not be readily transposed to other conditions or compared between minerals and even mineral crystallographic surfaces. When a mineral comes into contact with a fluid, a diffusive boundary layer (DBL) forms at the mineral-fluid interface, and dissolution occurs in two distinct steps: first the dissolving species must detach from the surface of the mineral, and second, they must diffuse through the DBL to the bulk solution (e.g. Morse and Arvidson, 2002). The slower of these steps controls the   78 rate of mineral dissolution (e.g. Berner, 1978). When the dissolution rate of a mineral is dictated by the rate of diffusion of dissolution products or reactants across the DBL, dissolution is “transport-controlled”. When dissolution rate is limited by the rate of detachment of dissolving species from the surface of the mineral, dissolution is “surface-controlled”. Under a surface-controlled dissolution regime, rates of mineral dissolution can be described with rate laws of the form (e.g. Lasaga, 1998): 𝑅!"#$%&' =   !!! !/! 𝑘!  𝑎!!!  !!!!! (1− (IAP /𝐾!"))!    (4.1) where 𝑅!"#$%&' (moles m-2 s-1) is the surface area normalized, surface-controlled dissolution rate, the !!! !/! term accounts for surface area loss over the course of a dissolution experiment, where the exponent 2/3 is applicable for uniformly dissolving spheres or cubes (e.g. Appelo and Postma, 2005), 𝑁! is the number of parallel reaction pathways, 𝑘! (moles m-2 s-1) is the dissolution rate constant for the 𝑖!! reaction pathway, 𝑎! is the activity of the 𝑖!!  species, 𝑛! is the reaction order of the 𝑖!! pathway, and IAP /𝐾!" defines the saturation state, which is the ratio of the ion activity product at the mineral surface (IAP ) to the solubility product of the mineral (𝐾!"). 𝑛 is an empirically determined exponent applied to the thermodynamic affinity term. The thermodynamic affinity terms modulates the reaction rate if equilibrium conditions are approached.  Under a transport-controlled dissolution regime, the dissolving species build up in a diffusive boundary layer (DBL), and Fick’s rate law dictates the rate of diffusion of products or reactants across the DBL (e.g. Lasaga, 1998): 𝑅!"#$%&'!" =  𝑘! 𝑐!"#$%&' − 𝑐!"#$        (4.2)   79 where 𝑅!"#$%&'"! (moles m-2 s-1) is the surface-normalized diffusion rate across the DBL, 𝑘! (m s-1) is the transport rate constant (with 𝑘! =  𝐷 𝛿, where 𝐷 (m2 s-1) is the diffusion coefficient of the dissolved species, and 𝛿 (m) is the DBL thickness), 𝑐!"#$%&' and 𝑐!"#$ (moles m-3) are the concentration of the dissolved species in contact with the surface of the mineral and in the bulk fluid, respectively. Between these two ‘end-member’ dissolution regimes, a mixed dissolution regime exists in which diffusion through the DBL is slow enough to allow for the build-up of dissolved species in the DBL, but the rate of detachment of dissolving species from the mineral surface is too slow to achieve saturation at the surface. For a given mineral and bulk solution composition, the dissolution regime shifts from transport- to mixed- to surface-controlled dissolution regime as the DBL decreases in thickness, or as the mineral approaches saturation.  Since intrinsic mineral dissolution rate constants must be measured under surface-controlled conditions, it is important to establish the dissolution regime under which dissolution rates of minerals are measured before interpreting or using the results. Many of the studies addressing mineral dissolution kinetics rely on measuring bulk solution concentrations to estimate mineral dissolution rates. With the development of high-resolution imaging techniques such as Atomic Force Microscopy (AFM) and Vertical Scanning Interferometry (VSI), some of the focus has now shifted from bulk solution measurements to variations in nanoscale surface topography to estimate mineral dissolution rates of the more soluble minerals (e.g. Liang and Baer, 1997; Arvidson et al, 2003; Ruiz-Agudo and Putnis, 2012). Using this approach, dissolution rates are estimated from volume of mineral loss to dissolution per unit time, converted into moles/s using the molar volume of the mineral.  Although this approach is fundamentally different from the methods using bulk solution composition, it also requires   80 knowledge of the dissolution regime to interpret the data. Thus, there exists a need to develop methodology to assess this critical parameter when measuring dissolution rates by any techniques.  4.1.2 Previous approaches to determine dissolution regimes The need to determine the rate-limiting mineral dissolution step was quickly identified when studying the rate of dissolution of more soluble minerals such as calcite (e.g. Plummer et al., 1978, 1979; Sjöberg and Rickard, 1984; Murphy et al., 1989). In principle, transport-controlled and mixed dissolution regimes can be identified when dissolution rates vary in response to changes in the thickness of the DBL. A commonly used approach involves measuring dissolution rates by dissolving fine-grained minerals and/or individual mineral crystals suspended in a stirred vessel. In this set up, controlling the thickness of the DBL is achieved by varying the stirring rate of the suspensions. Constant dissolution rates with varying stirring rates have been interpreted to indicate a surface-controlled dissolution regime (e.g. Sjöberg, 1978; Plummer et al., 1978; 1979; Busenberg and Plummer, 1986). However, Sjöberg and Rickard (1983) pointed out that such a system has poorly controlled hydrodynamics, and a lack of stirring dependence does not necessarily mean surface-controlled dissolution, but instead could indicate that, at some point, the effective thickness of the DBL does not change with stirring rate. This led to alternative approaches including the fluidized bed reactor, which produces a more even spatial distribution of suspended particles (e.g. Chou et al., 1989). The higher turbulence generated using this experimental design accelerates diffusional transport through the DBL, facilitating the establishment of surface-controlled dissolution conditions. However, it is difficult to determine the actual dissolution regime with this experimental set up. Another widely used approach relies   81 on a rotating disc made of the mineral of interest, which is immersed in the dissolving media (e.g. Sjöberg and Rickard, 1983; 1984; Pokrovsky et al., 2005). Using this approach, in laminar flow regime, the transport rate constant can be calculated with the solution of Levich (1962) based on a semi-infinite domain (Alkattan et al., 1998, Pokrovsky et al. 2005): 𝑘! = 0.62 𝐷!/! 𝜈!!/! 𝜔!/!          (4.3) where 𝜈 is the kinematic solution viscosity of the solution and 𝜔 the disc rotation speed. For mineral grains of arbitrary shape and size in finite reactor cells, however, solution of the flow, transport and reaction processes can only be achieved through numerical techniques. Recent advances in pore-scale modeling, involving the numerical solution of Navier – Stokes flow, advective-diffusive transport and dissolution equations, make it possible to capture transport limitations to reactive surfaces in complex domain geometries without having to make assumption about the thickness of the DBL (Molins et al., 2014).    This study aims to assess the potential of FT-TRA to distinguish between surface-, mixed- and transport-controlled dissolution regimes. Further, pore-scale modeling is used to support the interpretation of FT-TRA data under various flow conditions and to explore the development of a combined experimental-modeling approach to determine intrinsic reaction rate parameters.   4.2 Materials and methods 4.2.1 Sample origin, preparation and surface area measurement Calcite and forsterite samples were obtained from Ward’s Natural Science (items #49-5860 and #491557 respectively). Transparent forsterite grains (Mg1.81Fe0.18Ni0.01SiO4; Fo91) were handpicked, and ground down to a 63 – 150 µm size fraction. Pre-treatment was limited to   82 ultrasonic cleaning with acetone until a clear supernatant was obtained. A subsample (8.17 × 10!! 𝑔) was dried-overnight in an oven at 60 ºC. The mineral grains were examined by scanning electron microscopy (SEM), which revealed that they were generally free of adhering fine particles (Fig. 4.1A). Specific surface area (257.3 𝑐𝑚! 𝑔!!) was measured using multi-point BET (Brunauer et al., 1938) with N2 adsorption using a Quantachrome Autosorb-1 surface area analyzer. Calcite mineral rhombs were individually picked, ultrasonically cleaned with acetone to remove fine particles adhering to the mineral surface, and oven-dried at 60 ºC (Fig. 4.1B). Geometric surface areas were estimated from images obtained with a Hitachi S-4700 field emission scanning electron microscope housed at the University of British Columbia BioImaging Facility. Individual calcite rhombs were positioned on a platform, and were manually repositioned to expose each rhomb face for a top-down field-of-view to allow for geometric surface area estimation.     83  Figure 4.1 SEM imagery associated with mineral samples used in this study. (A): forsterite powder used in this study (63 – 150 µm size fraction); (B): individual calcite rhomb as typically used in this study.  4.2.2 FT-TRA module Dissolution experiments were conducted using a custom-built flow-through module connected to an Agilent 7700x quadrupole ICP-MS housed at the Pacific Center for Isotopic and Geochemical Research (Fig. 4.2; Chapter 2) to measure the release rate of Mg and Si to monitor forsterite   84 dissolution and Ca to monitor calcite dissolution. The flow-through cell consisted of a 13 mm diameter syringe filter (Millipore Inc. item #SLLGC13NL) mounted between two computer-controlled parallel solenoid valves (NResearch Inc. item#225T092C). This design provides 6 filter positions, allowing for efficient switching between blanks and samples, analyzed successively. Eluents were pumped from individual bottles (E1 – E4) with a Dionex ICS-3000 dual gradient pump that generates precise flow rates covering 4 orders of magnitude (0.16 × 10!! − 0.16 𝑚𝐿 𝑠!!). With this experimental design, mineral dissolution rates can be measured over a wide range of eluent residence times within the flow-through cell (𝑉 (0.05 𝑚𝐿) 𝑄). The pump was programmed with the Chromeleon™ chromatography software to control eluent composition for the experiment or standards for calibration. Another gradient pump was used isocratically in parallel, to continuously supply an internal standard (115In) allowing for instrumental drift correction during time-resolved ICP-MS measurement. The eluent produced by the first pump either directly joined the internal standard stream (when running standards) or passed through the flow-through cell (when running samples) before merging with the internal standard stream. The flow rate of the internal standard was held constant (1.66 × 10!! 𝑚𝐿 𝑠!!), while the flow generated by the other gradient pump was adjusted according to the experiment. 24Mg, 28Si, 44Ca and 115In were measured in time-resolved mode with helium to minimize isobaric interferences (McCurdy and Woods, 2004), and the ICP-MS output (cps) was corrected for dilution by the internal standard. Effluent [Mg, Si and Ca] concentrations were monitored in real time and experiments were terminated when steady release rates were observed. Forsterite and calcite dissolution rates were calculated by multiplying [Mg, Si; Ca] concentrations (𝑚𝑜𝑙𝑒𝑠 𝐿!!) by the flow rate of the effluent (𝐿 𝑠!!). For forsterite, the Mg value (𝑚𝑜𝑙𝑒𝑠 𝑠!!) was corrected for stoichiometry by dividing by 1.81, before dividing by the BET surface area   85 associated with the weighed subsample to obtain a surface area normalized dissolution rate (𝑚𝑜𝑙𝑒𝑠 𝑚!! 𝑠!!). For calcite, the Ca value was divided by the geometric surface area exposed to the incoming eluent (5 faces of the calcite rhomb), obtained from SEM imagery, to obtain surface area normalized dissolution rates (𝑚𝑜𝑙𝑒𝑠 𝑚!!𝑠!!).    Figure 4.2 Schematic of the flow-through dissolution module. E1 – E5 represent eluent bottles, where E1 – E4 are used to generate eluents or standards (using an advanced gradient pump, indicted as AGP), and E5 is used to continuously supply an internal standard (here 115In, using an isocratic pump). This flow-through module allows for time-resolved control on incoming eluent composition (here, generating an eluent pH range) by mixing different   86 proportion of DIW with concentrated acid or standard from bottle E1-E4, providing a constant, reproducible flow rate through the flow-through cell, and enabling time-resolved analysis of the effluent stream. A total of 6 flow-through cell positions are available by using 2 parallel computer-controlled solenoid valves, allowing for efficiently measuring blanks and multiple samples. When running standards, the solenoid valve holding the flow-through cells is bypassed, as indicated in red.   4.2.3 Experimental design to establish dissolution regimes The purpose-built leaching module described above was used to determine the dissolution regime from observing the evolution of mineral dissolution rates with eluent flow rate (Q). When mineral dissolution is surface-controlled, the rate of mineral dissolution is described by equation (4.1) and dissolution rates should be independent of DBL thickness. We hypothesize that if experimental conditions are maintained far-from-equilibrium (IAP ≪ 𝐾!") in the FT-TRA module, dissolution rates (𝑄  × c!""#$!%&) should remain constant under varying flow rates (Fig. 4.3A). Under such mineral dissolution conditions, the steady-state concentration of dissolved species ([effluent]!"#$%& !"#"$) must increase proportionally with eluent residence time (𝑐𝑒𝑙𝑙 𝑣𝑜𝑙𝑢𝑚𝑒 (𝑉) 𝑄, Fig. 4.3B). In contrast, if [effluent]!"#$%& !"#"$ deviates from linearity with residence time, dissolution must be partially or fully controlled by diffusive transport through the DBL (Fig. 4.3D). This stems from the increase in thickness of the DBL as flow rate decreases and residence time increases (equation 4.2). As a result, 𝑐!"#$%&' gradually increases from 𝑐!"#$ towards saturation during a mixed dissolution regime, and eventually reaches saturation when dissolution is fully transport-controlled (Fig. 4.3C).  To test this hypothesis, individual calcite rhombs or weighed forsterite powders were placed in the flow-through cell and exposed to incoming effluents of constant pH (2.3, 3.3 and 4.0) over a   87 range of flow rates spanning three orders of magnitude (1.67 × 10!! − 0.125 𝑚𝐿 𝑠!!). Dissolution was carried out until a steady-state eluent concentration was reached (monitored by online ICP-MS time-resolved analysis). An overview of experiments conducted is provided in Table 4.1. A range in analytical grade (Seastar™) HNO3 concentrations was used to generate acidic eluents at pH 2.3, 3.3 and 4.0. Experiments were carried out at room temperature. An additional set of calcite dissolution experiments was also carried out at pH 5.6, however results obtained are deemed beyond the scope of this thesis chapter (reader is referred to Appendix B).    88 Figure 4.3 Conceptual representation of proposed FT-TRA approach to determine dissolution regime. If mineral dissolution is surface-controlled – and conditions are kept far-from-equilibrium – dissolution rates remain constant under varying flow rates (Panel A, slight curvature due to a minor saturation (𝐼𝐴𝑃/𝐾!") effect) and steady state dissolved species concentration (here: [Ca]) increases proportionally with eluent residence time (Panel B, similar slight curvature due to a minor saturation (𝐼𝐴𝑃/𝐾!") effect). In contrast, if dissolution rates vary according to flow rate (Panel C), transport processes (DBL formation) affect 𝑐!"#$%&', thereby lowering the dissolution rate. Here, steady state dissolved species concentration (here: [Ca]) will deviate from linearity with eluent residence time (Panel D).  Flow rate (mL s-1) Forsterite Calcite  pH = 2.3 pH = 2.3 pH = 3.3 pH = 4.0 1.25 × 10!!  x x x 8.33 × 10!!  x x x 6.67 × 10!!   x x 5.00 × 10!!   x x 4.17 × 10!!  x   3.33 × 10!!  x x  1.67 × 10!!  x x  1.17 × 10!! x x (n = 2) x x 5.83 × 10!! x x (n = 2) x x 4.17 × 10!!  x x x 3.33 × 10!! x  x  2.50 × 10!!   x x 1.67 × 10!! x x x x 1.33 × 10!!  x (n = 2) x x 1.00 × 10!!  x   8.33 × 10!!   x x 5.00 × 10!! x x   3.33 × 10!! x x   2.50 × 10!! x    1.67 × 10!! x x   Table 4.1 Overview of constant incoming eluent pH experiments (indicated in bold) conducted as part of this study. Effluent [Mg, Si] for forsterite and [Ca] concentrations were monitored using time-resolved analysis until steady state dissolution was achieved.   89 4.2.4 Rate equations The calcite dissolution rate equation used in the pore-scale model (section 2.5) was derived from equation (4.1), excluding !!! !/! as sample loss remained limited to < 1% for all experiments (calculated using integrated [Ca]effluent concentrations). The rate expression shown in equation (4.4) (Plummer et al., 1978) was used for the far-from-equilibrium term of Eq. (4.1)  𝑅!"#$%&' = (𝑘!𝑎!! +  𝑘!𝑎!"! +  𝑘!𝑎!!!)(1− 𝐼𝐴𝑃/𝐾!")      (4.4)  The thermodynamic affinity term was included in the pore scale model, but found to play almost no role in the simulations as all experiments were run under far-from-equilibrium conditions.  For forsterite dissolution rate data interpretation, solely the H+ reaction pathway can be considered under acidic, far-from-equilibrium conditions (e.g. Rimstidt et al., 2012):  𝑅!"#$%&' = 𝑘!!(𝑎!!)!.! (1− 𝐼𝐴𝑃/𝐾!")        (4.5)  Similar to the calcite dissolution experiments, a constant surface area was assumed in the forsterite dissolution experiments, given the limited amount of sample loss (< 1%, calculated using integrated [Mg, Si]effluent concentrations).    4.2.5 Pore scale modeling To delineate transport-controlled processes occurring during calcite dissolution under varying incoming eluent velocities, high-resolution pore scale simulations were performed using the approach of Molins et al. (2012, 2014). A 3-dimensional domain was constructed consisting of a cylinder (0.4 cm in diameter and 0.4 cm in length) at the bottom of which a cubic calcite grain rested, reproducing the FT-TRA calcite dissolution experimental design (3-dimensional model domain shown in Fig. 4.4). Simulations were run at a 62.5 µm resolution. The dimensions of the   90 grains were varied in each simulation such that their surface areas were consistent with measured geometric surface area obtained from SEM imagery for each experiment. This resulted in slightly different grain sizes in each simulation. Fluid flow was obtained by solving the incompressible Navier–Stokes equations in the cell space (Molins et al., 2012), with the eluent solution entering the reactor in the space between the grain and the walls of the cylinder. Transport of 4 aqueous components (𝐻!,𝐶𝑂!(!"),𝐶𝑎!!,𝑁𝑂!!) was simulated with the advection-diffusion equation. The reactive components involved 8 aqueous equilibrium complexation reactions (Table 4.2). The rate of calcite dissolution was calculated at the mineral surface. 𝑘! 0.89 𝑚𝑜𝑙𝑒𝑠 𝑚!! 𝑠!! , 𝑘!(5.01 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝑚!! 𝑠!!) were sourced from Chou et al. (1989) and the 𝑘! 2.34 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝑚!! 𝑠!!  was sourced from Busenberg and Plummer (1986). The use of these rate constants was not intended to be part of a fitting exercise, but to provide a reasonable approximation to the rates measured in the experiments such that transport limitations could be investigated. Because the dissolution rate is calculated at the mineral surface in the pore scale model, rate constants used in the model are assumed free of transport limitations. Diffusive boundary layer thickness and resulting transport-controlled dissolution rates are quantified from the simulation.   91  Figure 4.4 Pore-scale simulation domain showing the calcite grain placed at the inlet end of the reactor. Contours of the magnitude of the fluid velocity in the reactor are shown on a slice that divides the domain in half. Fluid flows from left to right at the slowest experimental rate used (1.67 × 10!! mL s-1), which is prescribed using a uniform fluid velocity distribution at the inlet face.        92 𝐴𝑞𝑢𝑒𝑜𝑢𝑠 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 (𝑠𝑒𝑐𝑜𝑛𝑑𝑎𝑟𝑦 𝑠𝑝𝑒𝑐𝑖𝑒𝑠 ⇌ 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠)  log 𝐾!" 𝑂𝐻!  ⇌  −𝐻! +  𝐻!𝑂 13.9 𝐻𝐶𝑂!! ⇌  −𝐻! +  𝐻!𝐶𝑂! (!") 6.3 𝐶𝑂!!!  ⇌  −2𝐻! +  𝐻!𝐶𝑂! (!") 15.9 𝐶𝑎𝐶𝑂!  ⇌  −2𝐻! +  𝐻!𝐶𝑂! !" +  𝐶𝑎!! 13.3 𝐶𝑎𝐻𝐶𝑂!!  ⇌  −𝐻! +  𝐻!𝐶𝑂! !" +  𝐶𝑎!! 5.3 𝐶𝑎𝑂𝐻!  ⇌  −𝐻! +  𝐻!𝑂 +  𝐶𝑎!! 12.8 𝐻𝑁𝑂! (!")  ⇌  𝐻! +  𝑁𝑂!! 1.3 𝐶𝑎𝑁𝑂!!  ⇌  𝑁𝑂!! +  𝐶𝑎!! -1.3 Table 4.2 List of aqueous complexation reactions and equilibrium constants sourced from the EQ3/6 database (Wolery and Daveler, 1992).  4.3 Results and discussion 4.3.1 Forsterite Steady-state effluent [Mg, Si] concentrations were achieved within an hour (Fig. 4.5C) at eluent pH 2.3 and concentrations were found to increase proportionally with eluent residence time (Fig. 4.5A, 4.5B), following expectations if dissolution was surface-controlled for all eluent flow rates used. Forsterite dissolution rates vary within ±0.21 log units across all flow rates (error = 2𝜎, as shown in Fig. 4.5D), which is a typical range in dissolution rates found within one laboratory (e.g. Brantley, 2008: p.175). This variability is greater than expected from analytical uncertainties and may reflect variability in the number and distribution of reactive sites, as dissolution proceeds (Lüttge et al., 2013). The dissolution rates measured at all flow rates (2.53 ± 0.59 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝑚!! 𝑠!!) at pH 2.3 yield a dissolution rate constant (2.53 ± 0.59 × 10!! 10!!.! !.! = 0.36 ± 0.084 × 10!!𝑚𝑜𝑙𝑒𝑠 𝑚!! 𝑠!!) consistent with published values (Rimstidt et al., 2012). As the forsterite dissolution regime is widely recognized to be surface-controlled under acidic conditions (e.g. Pokrovsky and Schott, 2000), our findings agree with existing studies, and confirm the ability of FT-TRA to empirically confirm the presence of a surface-controlled dissolution regime.    93   Figure 4.5 Panels A and B show [Mg]steady state and [Si]steady state concentrations respectively (calculated using average [Mg, Si]selected obtained from applicable data as illustrated using data collected as shown in Panel C); error falls within symbol size; R2 = 0.999 for both [Mg] as well as [Si] versus eluent residence time. Panel C: Time-resolved [Mg, Si] concentration results collected at an eluent flow rate of 5 × 10!! 𝑚𝐿 𝑠!! at eluent pH 2.3 illustrates data that is typically collected measured during forsterite powder dissolution experiments. Selected [Mg, Si] concentrations representing steady state conditions are shown as black filled circles for Mg and red filled circles for Si. Data recorded prior to [Mg, Si]selected plateaus are shown as gray circles. Panel D: BET surface area normalized Mg and Si based forsterite dissolution rates versus eluent flow rate (note logarithmic x- and y-axis). [Mg]selected concentrations were divided by 1.81 to take into account forsterite stoichiometry. Average dissolution rate shown as solid black line, 95 % CI based on 2𝜎 shown as dotted black line.   94 4.3.2 Calcite Similar to the forsterite experiment, steady-state Ca concentrations ([Ca]steady state) were achieved within an hour at all flow rate conditions (Fig. 4.6D). However, during calcite dissolution, [Ca]effluent is not a linear function of residence time (Figs. 4.6A, 4.6B, 4.6C), which are interpreted as indicating transport-controlled conditions. Calcite dissolution rates calculated by multiplying averaged [Ca]steady state by flow rates at pH 2.3, 3.3 and 4.0 increase with decreasing pH and with increasing flow (shown in Figs. 4.7A, B and C respectively). The experimental data show a quasi linear dependence of the logarithm of the dissolution rate on the logarithm of the flow rate (Fig. 4.7A, B and C). A linear fit to these data (log 𝑟 = 𝑛 +  𝑚 log𝑄) yields 𝑚 = 0.513 (R = 0.992) for pH 2.3, 𝑚 = 0.502 (R = 0.986) for pH 3.3, and 𝑚 = 0.308 (R = 0.986) for pH 4.0. Under transport-controlled conditions (pH 2.3 and 3.3), the approximately square-root dependence (𝑚 ≈ 0.5) on the flow rate found here is comparable to the square-root dependence on the rotation speed of the rotating disc solution of Levich (1962, Eq. 4.3). Deviations from linearity at pH 4.0 for low flow rates (<0.05 mL s-1, hypothetical linear fit using 𝑚 = 0.5 illustrated in Fig. 4.7C) likely indicate the presence of a mixed dissolution regime, and heavily weigh the fit towards a value of 𝑚 smaller (0.308, dashed line in Fig. 4.7C) than at pH 2.3 and pH 3.3.    95  Figure 4.6 Panels A, B and C show [Ca]steady state concentrations collected under eluent pH of 2.3, 3.3 and 4.0 respectively plotted versus eluent residence time. Error falls within symbol size. [Ca]steady state concentrations were calculated using average [Ca]selected obtained from applicable data as illustrated using time-resolved data collected at flow rate 𝑄 = 5 × 10!! 𝑚𝐿 𝑠!! at eluent pH 2.3 in Panel D. Selected [Ca] concentrations assumed to represent steady state conditions are shown as black filled circles.  These experimental data could be reasonably well reproduced with the pore scale model, particularly at pH 2.3 and 3.3, where the departure of linearity at flow rates was not observed   96 (Fig. 4.7A, 4.7B and 4.7C). The pore scale model revealed the presence of a nearly stagnant fluid pocket on the lee side of the calcite rhomb, where reaction products build up and pH increases (Fig. 4.8). As a result, dissolution rates at this surface are much slower and contributed little to the overall calcite dissolution rate. In contrast, dissolution rates from the four calcite surfaces parallel to the mean flow direction are much faster due to the decreased thickness of the DBL both under high and low flow rate conditions (Fig. 4.8). The simulated thickness of the DBL increases with decreasing flow rates (Fig. 4.9). The existence of the DBL indicates partial or full transport control on rates in all simulations. Based on the agreement with effluent experimental data, the modeling results support the observation of transport-controlled rates made in the previous paragraph and based solely on FT-TRA experimental data. At pH 2.3 and 3.3, the concentrations in the solution immediately adjacent to the mineral surface are such that the 𝑘!𝑎!! term of Eq. (4.4) dominates the intrinsic dissolution rate. However, at pH 4.0 and slow flow rates, the activity of H+ at the surface decreases sufficiently from the bulk solution (due to transport limitations) that the zero-order term 𝑘! and the affinity term in Eq. (4.4) dominate the rate. This results in slower intrinsic dissolution rates, hence the effective rates become partially surface-controlled rather than exclusively transport-controlled. This observation was made also based solely on FT-TRA data that showed a departure from linearity at slow flow rates (Fig. 4.7C).     97  Figure 4.7 Panels A, B, C show calcite dissolution rates on logarithmic scale from experimental data (black filled circles, error falls within symbol size), and the pore scale modeling (black diamonds) under incoming eluent pH of 2.3, 3.3 and 4.0 respectively. Solid lines in A and B display linear fit through experimental data. Dashed line in C displays linear fit through experimental data. Solid black line in C solely shown to represent approximate hypothetical linear fit with a slope of 0.5. Pore scale modeling data were obtained using k1, k2 from Chou et al. (1989), and k3 from Busenberg and Plummer (1986).  Modeled effluent pH data is also shown (black filled triangles, axis shown on right).    98 2.3   3.3   4.0   Figure 4.8 Model domain cross-section (simulations are 3-dimensional), showing transport-controlled calcite rhomb (represented as white square) dissolution during highest left) and lowest (right) flow rate conditions at different   99 incoming eluent pH conditions (pH = 2.3 at top, pH = 3.3 in middle, pH = 4.0 at bottom). Color gradient within flow-cell indicates pH. Eluent velocity direction and magnitude indicated by arrow direction and length (scaled in each plot), respectively.   Figure 4.9 Simulated pH values plotted along the diameter of the cylinder used in the simulation domain, intersecting perpendicularly the mineral grain surface at 0.1 cm from the inlet. Departure from bulk pH at either side   100 of the grain (center blank zone) shows the variation of the DBL thicknesses at 0.1 cm from the inlet face as a function of flow rates (as indicated) at (A) eluent pH 2.3, (B) eluent pH 3.3 and (C) eluent pH 4.0.   In Fig. 4.10, the thicknesses of the DBL extracted from Fig. 4.9 are plotted against the flow rate, showing the decrease in thickness with increasing flow rate. By analogy between 𝑄 and 𝜔 from the Levich (1962) rotating disc solution, we assumed a relationship of the form 𝛿 = 𝑘𝑄!! to capture the dependence of the DBL thickness on flow rates observed in simulation results (Fig. 4.9). For this purpose, we used the value of 𝑚 obtained in the earlier linear fits to experimental data (Fig. 4.4). The fit to the simulated DBL thicknesses is remarkably good except at fast flow rates. At fast flow rates, the DBL shrinks significantly and the discretization used for the simulations is somehow coarse; i.e. the DBL is captured within only one or two grid cells (Fig. 4.10). This likely results in an underestimation of transport limitations and thus in an overestimation of the intrinsic dissolution rates.    101  Figure 4.10 Thickness of the DBL as a function of eluent flow rate. In symbols are DBL thicknesses extracted from Fig. 4.9 (distance from the point where pH departs from bulk solution value to the point at the mineral surface where pH is highest). Dashed lines indicate the steps corresponding to the grid cells in discretized numerical domain. Solid lines are a fit to simulated DBL thicknesses using the slope 𝑚 obtained in the fit to experimental data (Fig. 4.7).  Even when taking into account the model resolution issue, the good agreement between measured and modeled dissolution rates under transport-controlled conditions when using 𝑘! reported by Chou et al. (1989) (Fig. 4.7) could indicate that their fluidized bed reactor yielded dissolution conditions very close to surface-control. It could be argued that if their measurements were conducted in transport or mixed control regimes, their 𝑘! would have been underestimated and the pore scale model would have underestimated our experimental data. However, this conclusion is still very tenuous, because it would also imply that the surface normalization in   102 both studies were equivalent, which cannot be verified. In fact, our experimental surface normalized dissolution rates are based on the assumption that five of the six faces of the calcite rhomb are subjected to dissolution. However, the pore scale model indicates that the surface positioned in the lee of the flow does not contribute significantly to the overall dissolution and it may be more accurate to normalize to the 4 faces parallel to the flow. In doing so, our experimental surface-normalized dissolution rates would increase by 25%. It is clear that the problem of surface normalization is still a major hurdle in the study of mineral dissolution kinetics. Geometric or BET surfaces are inadequate for normalization because they do not take into account the distribution of reactive surface sites (e.g. Lüttge et al., 2013). Nonetheless, setting this problem aside, our results illustrate the ability to identify dissolution regimes with the FT-TRA module and how dissolution rate parameters may be constrained by combining experimental data obtained with FT-TRA and pore scale modeling.  4.3.3 Added value to FT-TRA FT-TRA is a new approach when applied to study the dissolution kinetics of minerals. While the present study demonstrates that FT-TRA is suitable for establishing the dissolution regime, it has also other advantages. One of these advantages, stemming from the very small size of the flow cell, is in generating dissolution rate data more rapidly than with mixed flow reactors, facilitating duplication, statistical analysis of the results, and process studies to unravel dissolution mechanisms (De Baere et al., 2015). It also provides detailed information on the evolution of dissolution rates and stoichiometries when a mineral is subjected to abrupt or transient dissolution episodes (op. cit.).    103 As demonstrated here, the FT-TRA approach could prove particularly powerful when combined with pore scale reactive-transport modeling to resolve transport and reactive processes within the experimental flow-through cell. The experimental set up described here is still exploratory and alternative flow cell geometry could be designed to facilitate modeling. The measurement of additional effluent parameters (e.g. pH) may also aid in further constraining modeling results and provide additional means to calibrate reaction rate parameters. Furthermore, a flow cell that would expose a single mineral cleavage face to a laminar flow of eluent, as used in the study by Compton and Unwin (1990) may be particularly suitable for a wide range of experiments. In particular, such cell design should be compatible with consecutive or on-line atomic force microscopy (Duckworth and Martin, 2003) to correlate bulk solution concentrations and changes in nano-topography, through which the fundamental problem of surface normalization could be addressed.  4.4 Conclusions and future research direction This study demonstrates that FT-TRA can be used to empirically establish mineral dissolution regimes. We illustrated how it can be used to identify surface-controlled dissolution regimes using the dissolution of forsterite. In contrast, applying FT-TRA to calcite dissolution illustrated how transport-controlled dissolution can be revealed. The need to know the dissolution regime during dissolution experiments stems from the fact that intrinsic mineral dissolution parameters, such as dissolution rate constants and order of dissolution reactions, can only be directly measured under surface-controlled conditions. Combining FT-TRA measurements with 3-dimensional pore scale reactive transport models provides a means to estimate intrinsic mineral dissolution parameters, even when the experimental conditions are inadequate to produce a   104 surface-controlled regime. The model establishes the hydrodynamic conditions of dissolution but is driven by the kinetics of surface reactions. As a result, intrinsic dissolution rate parameters may be obtained by fitting the model to experimental data. Future work will involve: developing different flow-through cell designs to facilitate modeling; measuring additional effluent parameters (e.g. pH) providing additional empirical data to constrain modeling results and exploring the possibility of associating FT-TRA with atomic force microscopy. The latter development may be particularly important in view of the long-standing problem of surface normalization (e.g. Lüttge et al., 2005 and references therein) and the recent debate concerning the validity of using single rate constants to describe mineral dissolution (e.g. Lüttge et al., 2013; Fischer et al., 2014). The multi-faceted applications and encouraging results obtained with this novel approach (Chapter 3) warrants further development.             105 Chapter 5: Evaluating FT-TRA dissolution as a tool to quantify ontogenetic Mg/Ca molar ratios in foraminiferal tests  5.1 Introduction The trace element/Ca molar ratios of calcite shells produced by foraminifera are widely used in paleoceanography to reconstruct past environmental conditions. However, clay contamination and partial dissolution have widely been documented to affect the reliability of some elemental proxies (e.g. paleotemperature from Mg/Ca: Barker et al., 2003; Rosenthal et al., 2004; Mekik and François, 2006). Moreover, the presence of multiple biogenic calcite layers added over the course of a foraminiferal lifecycle (e.g. Bé and Lott, 1964; Bé et al., 1983; Hemleben and Spindler, 1983; Hemleben et al., 1985; Hamilton et al., 2008) significantly complicates the geochemical analysis required to isolate the correct signal.  This introduction will (a) briefly introduce the importance of foraminiferal elemental ratios in paleoceanography; (b) briefly review calcification and partial dissolution in foraminifera (section 5.1.1); and (c) introduce how flow-through time-resolved analysis (FT-TRA) has been validated as a tool to dissolve and isolate the carbonate phase expected to contain the Mg/Ca molar ratio that represents calcification temperature (section 5.1.2; section 5.1.3).   5.1.1 Brief introduction of the use of foraminiferal elemental ratios in paleoceanography The trace element composition of foraminiferal shells records information pertaining to the environmental conditions under which the foraminifera grew. Hence, fossilized foraminiferal tests represent a powerful archive, allowing for – assuming accurate calibration and   106 measurement – reconstruction of past environmental conditions.  Some environmental conditions that have been studied (by no means an exhaustive overview, which is beyond the scope of this thesis chapter) include seawater temperature using Mg/Ca molar ratios (e.g. Nürnberg et al., 1996; Lea et al., 1999; Rosenthal et al., 2000), carbonate chemistry using U/Ca, B/Ca molar ratios (e.g. Russell et al., 2004; Yu et al., 2007), as well as nutrient utilization using Cd/Ca molar ratios (Elderfield and Rickaby, 2000). Calibrating the relationship between foraminiferal trace element composition and environmental variable remains challenging, as foraminiferal calcite is precipitated under biologically controlled processes (e.g. Bentov and Erez, 2005). In particular, the offset between the chemical composition of biogenic and inorganic calcite (e.g. Elderfield et al., 1996) varies not only with foraminiferal species (e.g. Hathorne and James, 2006), but also the size of the shell (e.g. Ni et al., 2007). Therefore, the effect of biomineralization processes on measured shell trace element chemistry need to be fully understood, thereby allowing for the extraction of the relevant signal.  5.1.2 Foraminifera calcification, partial dissolution Foraminifera calcify in discrete layers throughout their lifecycle, and although the shell is entirely composed of calcite, the layers consist of distinct crystal forms. Using Scanning Electron Microscopy (SEM) imagery, Bé et al. (1975) examined the cross-section of a fully developed Globorotalia truncatulinoides foraminiferal test (a symbiont-free species) from a surface sediment sample from the Sargasso Sea (35º06’N., 45º56’W., 3400 m). Several layers, each with distinct crystalline units could be distinguished: (1) an inner layer consisting of closely packed anhedral microgranules (about 0.2 µm in diameter), producing a smooth texture when seen from top; (2) an intermediate layer consisting of subhedral crystallite units (about 0.6 – 1.0 µm long),   107 which is interpreted as the initial stage in the calcite crust development and (3) an outer layer consisting of large euhedral calcite crystallites (10 – 15 µm long), which are the product of the gradual enlargement of the subhedral crystalline units. Hence, no distinct boundary was found between the subhedral crystallite units and euhedral calcite crystallites (Bé et al., 1975). In order to differentiate foraminiferal shells having gone though varying stages in their lifecycle or ontogeny (from juvenile to mature stage), ontogeny nomenclature needs to be defined. Here, we follow the distinction between calcite layers by Bé et al. (1975). The earliest, juvenile calcite, consisting of anhedral microgranules is referred to as ontogenetic calcite. The older, more mature calcite layer, consisting of the subhedral crystallite units and euhedral calcite crystallites is referred to as gametogenic calcite, after the notion this layer is secreted immediately prior to gamete release (e.g. Hamilton et al., 2008). New microanalytical techniques such as nanoSIMS (e.g. Kunioka et al., 2006) and laser ablation coupled to ICP-MS (LA-ICP-MS, e.g. Eggins et al., 2003; Hathorne et al., 2003) allow for accurate and precise measurement of low levels of a number of elements at high (µm) spatial resolution. Studies using such techniques have, for example, identified a layer of high Mg/Ca calcite intimately associated with the primary organic membrane in Pelleniatina obliquiloculata (Kunioka et al., 2006). Another recent study by Sadekov et al. (2005) using electron microprobe mapping combined with SEM analysis revealed large variations of Mg/Ca within individual tests, the pattern of which was found to vary between symbiont-bearing and symbiont-free species. Symbiotic species reveal cyclic Mg/Ca compositional banding, the variation of which exceeds potential calcification temperature changes. Hence, vital effects have been suggested to modulate foraminiferal test Mg/Ca composition. Symbiont-free species reveal wider bands of distinct Mg/Ca composition that may reflect changes in calcification temperature as the   108 foraminifera migrate through the water column during its lifecycle (e.g. the development of thick, low-Mg outer layers in foraminiferal tests, as shown by Eggins et al., 2003). Regardless of the processes involved, the need exists to distinguish between the geochemical composition of both ontogenetic and gametogenic calcite layers prior to (paleo)oceanographic data interpretation.  Assessing which (or whether) layers of calcite preferentially dissolve as the foraminiferal test settles on the seafloor is difficult to establish, in part because the calcite layering occurs during the foraminiferal lifecycle. A generally accepted explanation for post-depositional Mg/Ca alteration is the preferential dissolution of the more soluble, Mg-rich parts of the foraminifera test (e.g. Brown and Elderfield, 1996). A number of approaches have been developed to study the effect that partial dissolution of foraminifera on the seafloor has on test elemental composition, including: (1) modeling the variation in foraminiferal test chemistry that results from adding and removing test calcite of different composition (Lohmann, 1995). The model exploits the fact that as foraminiferal test size increases as new chambers are added, the mass of ontogenetic calcite can be deemed proportional to its size. In contrast, the addition of gametogenic calcite only changes foraminiferal test thickness.  Hence, the difference between mass implied by foraminiferal test size and the mass actually measured can be assumed to be the added mass of the gametogenic calcite crust. Although promising results have been published using Globorotalia truncatulinoides, model calibration data requires the determination of the relationship between foraminiferal test size and mass prior to gametogenic calcite addition (Lohmann, 1995);   109 (2) observing compositional changes in sample splits of the same species exposed to laboratory induced partial dissolution using dilute acids (e.g. Globorotalia tumida in Brown and Elderfield, 1996; Globorotalia menardii and Globigerinoides trilobus in Hecht et al., 1975). Both studies indicated a decrease in bulk Mg/Ca, which was attributed by Brown and Elderfield (1996) to the preferential dissolution of the ontogenetic calcite high in Mg/Ca.   (3) monitoring effluent composition during the gradual dissolution of foraminifera tests using a purpose-built flow-through dissolution module (e.g. Haley and Klinkhammer, 2002; Benway et al., 2003; Klinkhammer et al., 2004). The latter approach (flow-through time-resolved analysis, FT-TRA) was developed to mitigate batch cleaning limitations such as partial sample loss (preferentially dissolving high Mg/Ca), the absence of data allowing for statistical evaluation of carbonate homogeneity, and the lack of information about contaminant phases (e.g. Haley and Klinkhammer, 2002). Klinkhammer et al. (2004) stated FT-TRA is able to “sort biogenic calcite by susceptibility to dissolution and allows elemental ratios to be computed for the primary biogenic calcite fraction while excluding other ratio populations in the bulk shell and contaminant phases” (validations summarized in section 5.1.2). However, recent work by Sadekov et al. (2010) compared high spatial resolution laser-ablation data to high temporal resolution flow-through data, and did not find any evidence of preferential removal of Mg-rich calcite layers by progressive dissolution of Orbulina universa tests using the FT-TRA approach. Here, we test the assumptions upon which the flow-through dissolution is based, by designing simple FT-TRA based experiments to evaluate whether more soluble mineral phases preferentially dissolve in the flow-through module.    110 5.1.3 FT-TRA technique validation The flow-through time-resolved analysis (FT-TRA) technique (Haley and Klinkhammer, 2002) continuously exposes foraminiferal tests to a series of distinct cleaning eluents. In turn, these eluents dissolve distinct contaminant phases associated with foraminiferal tests (Table 5.1), whilst effluent elemental composition is continuously recorded using time-resolved ICP-MS analyses. Later work gradually simplified the eluent sequence (Benway et al., 2003; Klinkhammer et al., 2004), eventually restricting eluents to deionized water and dilute nitric acid (e.g. Klinkhammer et al., 2009, as indicated in Table 5.1).   Study Flow rate mL min-1 Hydroxylamine pH > 9 DTPA pH > 9 DIW Nitric acid, pH Haley and Klinkhammer, 2002 4.0 × × × 1.3 Benway et al., 2003 nr ×  × 2 Klinkhammer et al., 2004  ×  × nr: “pump mixes 1.0 N HNO3 with deionized water to produce the desired concentration” Klinkhammer et al., 2009    × 2 Sadekov et al., 2010 nr nr: “method described in Haley and Klinkhammer (2002) and  Klinkhammer et al. (2004)” Table 5.1 Eluents used in flow-through dissolution experiments (nr = not reported) as reported in FT-TRA studies aiming to isolate ontogenetic carbonate elemental composition. Nitric acid pH represents eluent pH used to dissolve the foraminiferal test. Hydroxylamine has been used to dissolve “high–Mg calcite and oxide coatings” (Haley and Klinkhammer, 2002); DTPA (diethylene triamine pentaacetic acid) has been used to dissolve sedimentary barite (Lea and Boyle, 1993), relevant for studying the Rare Earth Elements (Haley and Klinkhammer, 2002). Shaded areas represent eluent absence from study.    111 FT-TRA technique validation has been based on a combination of qualitative and quantitative arguments. Qualitatively, Benway et al. (2003) carried out a series of quenched dissolution experiments, during which varying degrees of partial dissolution of foraminiferal tests were simulated. Scanning Electron Microscopy (SEM) imagery of remaining shell material obtained revealed similarities between naturally dissolving foraminiferal tests and dissolution in the flow-through module. Ontogenic (microgranular inner layer) calcite appeared to be dissolving earlier, followed by gametogenic (outer layers) calcite (Fig. 1 in Benway et al., 2003). This visual agreement between laboratory-based and natural dissolution has been interpreted to validate the FT-TRA approach to “chemically sort genetically distinct layers of calcite” (Benway et al., 2003). The earliest calcite to dissolve appeared to be the inner calcite layer (ontogenetic calcite), followed by the outer layers (gametogenic calcite).  Quantitatively, the accuracy of the flow-through approach has been tested by comparing core-top Mg/Ca based temperature estimates from G. ruber species from a late Holocene Cocos Ridge core top (4º36.82’N, 86º42.24’W, 904 m) with World Ocean Atlas (WOA 98) temperature profiles. Good agreement (within 0.2 ºC) between observed and reconstructed temperatures was found, with reconstructed temperatures corresponding to the calcification depth range associated with the particular species (Benway et al., 2003 and references therein).  In a later study by Klinkhammer et al. (2004), Mg/Ca and Sr/Ca ratios obtained in core-top samples (G. tumida and G. sacculifer) collected across a depth transect on the Ontong Java Plateau did not seem to be affected by prohibitive preferential high Mg/Ca dissolution associated with batch processing sample preparation. Batch-based dissolution Mg/Ca, Sr/Ca data shows a pronounced decrease in Mg/Ca, Sr/Ca with increasing water depth, which is not as readily observed in the FT-TRA based Mg/Ca, Sr/Ca data (Fig. 6 in Klinkhammer et al., 2004). Hence,   112 the FT-TRA data was interpreted to “minimize the effects of partial dissolution on the seafloor” (Klinkhammer et al., 2004) as one can still distinguish the (partially dissolved) ontogenetic phase from the gametogenic phase, irrespective of changes to the ontogenetic : gametogenic mass ratio. In addition, FT-TRA-based Mg/Ca ratios across the depth transect were somewhat higher compared to batch analyses (the latter analyses carried out by Brown and Elderfield, 1996). This finding was interpreted to be expected “if a significant amount of the more susceptible (higher Mg/Ca) calcite had been lost on the sea floor or during pretreatment in the laboratory during batch processing” (Klinkhammer et al., 2004). Finally, the agreement between calculated calcification temperatures based on these FT-TRA based Mg/Ca measurements and World Ocean Atlas temperatures was used to further validate the ability of flow-through technique to isolate the ontogenetic calcite signal (Klinkhammer et al., 2004).   5.1.4 Addressing FT-TRA calcite dissolution kinetics The FT-TRA application re-evaluated in this chapter addresses the simultaneous dissolution of multiple biogenic carbonate phases within a single sample. In order for a FT-TRA experiment to exploit intrinsic physicochemical dissolution variability between several mineral phases, it is important to consider the rate-limiting step determining the dissolution regime (e.g. Berner, 1978). Mineral dissolution involves the following steps: (1) transport of the reactants to the mineral phase; (2) adsorption of the reactants to the mineral phase surface; (3) migration of the reactants on the surface to an “active” site; (4) the chemical reaction between the absorbed reactant and mineral phase; (5) migration of the reaction products away from the reaction site; (6) desorption of the reaction products to the ambient solution and (7) transport of the reaction products to the “bulk” solution (e.g. Morse and Arvidson, 2002).  The slowest step is the rate-  113 controlling step. Steps (1) and (7) involve the transport of reactants and products to and from the mineral phase surface, whereas steps (2 – 6) occur at the mineral phase surface. Hence, if one of the latter steps is the slowest step, the dissolution rate is surface controlled, i.e. dissolution is dictated by the intrinsic properties associated with the mineral undergoing dissolution (e.g. Colombani, 2008). Alternatively, if step (1) or (7) is deemed rate controlling, the dissolution rate is transport-controlled, which involves the establishment of a concentration gradient across the diffusive boundary layer (DBL) through which the dissolving species must diffuse. The thickness of the DBL and/or the reaction product concentration gradient across the DBL and its effect on mineral dissolution rates depends on external conditions such as flow-cell hydrodynamics (e.g. Levich, 1962, see Chapter 4). Under the latter conditions, transport through the DBL will dictate the mineral dissolution rate and mask the intrinsic (surface-controlled) dissolution rate of the mineral.  The goal of this study is to re-evaluate the existing premise stating that “FT-TRA sorts biogenic calcite by susceptibility to dissolution” (Klinkhammer et al., 2004). Chapter 4 has – in detail – demonstrated the presence of transport-limited calcite dissolution at and below pH 4.0 using the current FT-TRA configuration. Here, it is illustrated how this insight impacts the application of FT-TRA in paleoceanographic studies. This is achieved by (1) dissolving pure calcite mineral samples and comparing observed dissolution rates to previously published rates (Chou et al., 1989; Busenberg and Plummer, 1986; Plummer, 1978, Sjöberg and Rickard, 1984), and (2) dissolving a mixture of aragonite (orthorhombic crystal structure) and calcite (trigonal crystal structure) powder to assess whether FT-TRA can distinguish between these distinct calcium carbonate polymorphs with different solubility (aragonite is about 1.5 times more soluble than calcite, e.g. Mackenzie and Lerman, 2006).   114 5.2 Methodology 5.2.1 FT-TRA technique development In order to carry out FT-TRA experiments, a flow-through dissolution module was developed at the University of British Columbia (Fig. 5.1). Eluents are pumped from individual bottles (E1 – E4) by a Dionex ICS-3000 dual gradient pump, which features a time-programmable proportioning valve. Pumps are able to generate precise flow rates covering 4 orders of magnitude: 0.001 – 10 mL min-1. In the experiments presented here, eluent flow rates were held constant at 0.7 mL min-1 (± 0.0367 𝑚𝐿 𝑚𝑖𝑛!!; 95% CI). One gradient pump was programmed to control the time-dependent composition of the eluent (for the experiment) or standard (for calibration). The other gradient pump was used isocratically to continuously supply an internal standard (0.1 mL min-1) to correct for instrumental drift associated with the continuous ICP-MS measurement. Programming of the gradient pump is done using the Chromeleon™ chromatography software. The generated eluent either directly joins the internal standard stream (when running standards) or passes through the flow-through cell (when running samples) before merging with the internal standard stream. The flow-through cell housing the sample consists of a 13 mm diameter syringe filter (Millipore Inc. item #SLLGC13NL) mounted between two computer controlled parallel solenoid valves (NResearch, Inc., item #225T092C). The effluent merges with an internal standard (115In) solution, before being introduced online to an ICP-MS (here, an Agilent 7700x quadrupole ICP-MS) for time-resolved measurement of 44Ca and 115In.     115  Figure 5.1 Schematic of the flow-through dissolution module. E1 – E5 represent eluent bottles, where E1 – E4 are used to generate eluents or standards and E5 is used to continuously supply an internal standard. AGP = advanced gradient pumps. Eluent flows top to bottom of figure. When running standards, the flow-through cell is bypassed. This flow-through module allows for time-resolved control on incoming eluent composition, and a constant, reproducible flow rate through the sample holder, for time-resolved analysis of the effluent stream by quadrupole ICP-MS.  5.2.2 Sample origin, preparation and surface area characterization End member calcite and aragonite mineral samples (CaCO3 polymorphs) were obtained from Ward’s Natural Science (item #49-5860 and item #49-5920 respectively). For experiments   116 requiring individual calcite and aragonite grains, grains were individually picked from the sample, and sample treatment was limited to ultrasonic cleaning with acetone to remove fine particles adhering to the mineral surface, followed by oven drying at 60 ºC. For powder experiments, aragonite and calcite samples were ground to a fine powder. No further sample preparation was carried out on the powder samples.  Individual mineral grain surface area was estimated geometrically, based on Scanning Electron Microscopy (SEM) imaging carried out using a Hitachi S-4700 field emission scanning electron microscope. Individual calcite and aragonite grains were positioned on a standard sample platform and were manually repositioned to expose each face for a top-down field-of-view (images provided in Appendix C). As part of a preliminary set of experiments, an individual calcite grain underwent SEM imaging before and after a series of dissolution experiments (exposing the mineral to pH 5.7 – 2.3 eluent), to visually assess the impact of flow-through dissolution on the mineral surface topography.  Mineral powder surface area was measured using multi point BET (Brunauer et al., 1938) with N2 adsorption using a Quantachrome Autosorb-1 surface area analyzer. Aragonite sample surface area was 2.018 m2 g-1. Calcite sample surface area was 0.599 m2 g-1. Sample characteristics (weight, surface area) associated with all samples are summarized in Table 5.2.          117 Sample Description Sample weight g Surface Area cm2 Sample loss(1) % 1 Cal. grain 4.82 × 10!! 0.070(2) 2.3 2 Cal. grain 6.59 × 10!! 0.110(2) 2.8 3 Cal. grain 4.09 × 10!! 0.080(2) 2.9 4 Arag. grain 2.95 × 10!! 0.070(2) 2.6 5 Arag. grain 10.52 × 10!! 0.160(2) 1.5 6 Arag. powder 389 × 10!! 7.85(3) 23.9 7 Cal. powder 285 × 10!! 1.71(3) 81.8 8 Arag. – Cal. powder mixture 374 × 10!!: Arag. = 96 × 10!! Cal. = 278 × 10!! 3.6(4): Arag. = 1.94(3) Cal. = 1.66(3) 42.6(4) Table 5.2 Overview of samples (Arag. = aragonite; Cal. = calcite).  (1) Sample loss calculated based on integrated effluent [Ca].   (2) Geometric surface area based on SEM imagery (3) Total surface area based on BET measurement: surface area calcite = 0.599 m2 g-1; surface area aragonite = 2.018 m2 g-1.  (4) Bulk sample calculation.   5.2.3 FT-TRA dissolution experiments Individual calcite and aragonite grains (samples 1 – 5) were placed in the flow-through cell, and exposed to a continuous stream of time-variable eluent composition covering pH 5.7 – 2.3 (Fig. 5.2), similar to the eluent pH range used by both Benway et al. (2003) and Klinkhammer et al. (2009, see Table 5.1). Specifically, individual grains were exposed to DIW in equilibrium with atmospheric CO2, after which HNO3 was gradually introduced until 0.5 mM HNO3 (aragonite grains) or 2.5 mM HNO3 (calcite grains) concentrations were reached. Here, eluent acidity was held constant for 1000 seconds (pH = 3.3 or 2.6, respectively). Then, acidity was gradually increased to 5 mM HNO3 (pH = 2.3), where it was held constant for 1000 seconds, after which   118 eluent composition was gradually shifted back to DIW. This final step was originally included in an attempt to quantify the hydrolysis dissolution rate constant under eluent conditions no longer affected by the dissolution of fines during the initial DIW time interval. However, quantification of calcite rate constants is beyond the scope of this chapter.   For the powder experiments, weighed pure aragonite (sample 6 = 389 µg) and calcite (sample 7 = 285 µg) powders were dissolved separately followed by a calcite – aragonite mixture of known proportion (sample 8 = 96 µg of aragonite + 278 µg of calcite). Samples were placed in a flow-through filter. Samples 6 and 7 were dissolved for 2000 seconds each using 10 µM HNO3 (pH = 5.0). Sample 8 was dissolved for a longer period of time (approximately 5500 seconds) using 10 µM HNO3 (pH = 5.0) until t = 4000 seconds, after which acidity was increased to 100 µM HNO3 (pH = 4.0) until the end of the experiment. The rationale behind using these mildly acidic eluents lies in the fact that aragonite and calcite are very soluble minerals. Dissolution under acidic eluent pH conditions as has been done to date (pH 1.3 – 2, Table 5.1) would promote the establishment of a concentration gradient across the DBL (Chapter 4), preventing the exploitation of inherent mineral solubility differences between aragonite and calcite. Hence, relatively mild pH conditions (pH 4 – 5) were used, increasing the possibility of achieving some degree of surface-controlled dissolution conditions (e.g. Busenberg and Plummer (1986) quantified a difference in dissolution rate at eluent pH 5.0 established using HCl). All experiments were carried out using a constant eluent flow rate of 0.7± 0.0367 mL min-1 (where error represents 95% confidence interval, see Table 2.2).     119  Figure 5.2 Time-variable eluent composition associated with experiments carried out on individual calcite, aragonite grains. Eluent pH associated with calcite rhombs (samples 1 – 3) shown in black; eluent pH associated with aragonite grains (samples 4, 5) shown as red, dashed line. Eluent consists initially of DIW in equilibrium with the atmosphere after which acidity was gradually increased to either 0.5 or 2.5 mM HNO3 (pH = 3.3 or 2.6), then 5 mM HNO3 (pH = 2.3). Finally, eluent composition was gradually switched back to DIW.   5.2.4 Measurement and calculation of [Ca, Mg, Sr] concentrations, dissolution rates 24Mg, 44Ca, 88Sr and 115In were recorded in time-resolved analysis mode using an Agilent 7700x quadrupole ICP-MS. He-mode was used to minimize isobaric interferences, as noted in McCurdy and Woods (2004). Counts per second obtained were corrected for dilution caused by the merging eluent and internal standard streams according to equation (5.1). The eluents and standards had a flow rate of 0.7 mL min-1 while the internal standard flow rate was 0.1 mL min-1. 𝑐𝑝𝑠!"#$%"&' !"##$!%$& =  𝑓𝑙𝑜𝑤 𝑟𝑎𝑡𝑒!"!#$ 𝑓𝑙𝑜𝑤 𝑟𝑎𝑡𝑒!"#!$% !" !"#$%#&% × 𝑐𝑝𝑠   (5.1)  DIW DIW5 mM2.5 mMSample 1 - 3 (calcite) Sample 4, 5 (aragonite) 0.5 mMpH23456Time (s)0 2000 4000 6000 8000  120 A high concentration standard ([Mg] = 2.09 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! ; [Ca] = 4.55 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!!; [Sr] = 2.84 × 10!!𝑚𝑜𝑙𝑒𝑠 𝐿!! in 1% HNO3) was diluted in known time-resolved proportions with the standard matrix (1% HNO3). Dilution corrected counts per second were 115In normalized to correct for instrumental drift over the course of an experiment. Prior to a sample run, a blank filter was run for blank correction using identical eluents as used in a sample run.  Based on the time-resolved [Ca] concentration data, dissolution rates were calculated by multiplying the measured elemental concentrations by flow rate (moles L-1 x L s-1), resulting in moles s-1. This value was then divided by the sample surface area (mineral grains = geometric surface area, powder samples = BET surface area), resulting in a dissolution rate expressed in moles m-2 s-1.  5.2.5 Modeling approach A simple, 1-dimensional model was developed using the publicly available geochemical software PHREEQC (version 3.1.4.8929, Parkhurst and Appelo, 1999). A flow-through cell domain was defined using the volume of the applicable flow-through cell (25 or 50 µL). Time-dependent infilling solution (varying concentrations of nitric acid) continuously replaces the content of the flow-through cell at a rate defined by the eluent residence time, and the effluent concentration is provided in the model output. The WATEQ4F thermodynamic database was sourced to carry out all chemical speciation calculations. The rate equation used in the geochemical model is:          (5.2) where   𝑅! = overall dissolution rate (moles s-1)   121   𝑚 = moles of mineral at a given time   𝑚! = initial moles of mineral present 𝑛 = factor that accounts for changes in reactive surface area during dissolution (n = 2/3 for single uniformly dissolving spheres and cubes, Appelo and Postma, 2005). 𝑆𝐴 = surface area (m2)   𝑁! = number of parallel reaction pathways leading to dissolution    𝑘! = rate constant associated with aqueous species i (moles m-2 solid s-1)   𝑎! = activity of aqueous species i (dimensionless term)   𝑛! = reaction order  For calcite and aragonite, the reaction order 𝑛! equals 1 for calcite and aragonite (e.g. Chou et al., 1989). Plummer et al. (1978) suggests that the following reactions occur parallel during carbonate dissolution under ambient conditions: 𝐶𝑎𝐶𝑂!(!) +  𝐻(!")!   =  𝐶𝑎 !"!!   +  𝐻𝐶𝑂!(!")!          (5.3) 𝐶𝑎𝐶𝑂!(!) + 𝐻!𝐶𝑂! (!")∗  =   𝐶𝑎 !"!! + 2𝐻𝐶𝑂!(!")!        (5.4) 𝐶𝑎𝐶𝑂!(!) +  𝐻!𝑂(!") =  𝐶𝑎 !"!! +  𝐻𝐶𝑂!(!")! + 𝑂𝐻(!")!      (5.5) where 𝐻!𝐶𝑂!∗ represents the sum of 𝐻!𝐶𝑂! (!") and 𝐶𝑂! (!"). Rates associated with reactions 5.3 – 5.5 are provided in equations 5.6 – 5.8, where 𝑘!, 𝑘! and 𝑘! are rate constants and 𝑎! refers to the activity of the relevant species i. Calcite, aragonite rate constants were sourced from Chou et al., 1989. Model output includes time-resolved effluent [Ca, Sr, Mg] concentrations, dissolution   122 rates associated with each forward reaction, as well as flow-cell pH assuming the flow-through cell is well mixed.  𝑟 =  𝑘!𝑎!!           (5.6) 𝑟 =  𝑘!𝑎!!!!!∗             (5.7) 𝑟 =  𝑘!𝑎!!!           (5.8) Although chapter 4 has demonstrated the latter assumption does not hold during calcite dissolution at and low pH 4.0, the use of this PHREEQC model will nevertheless assist in quantifying the expected effluent composition assuming surface-controlled dissolution conditions. As will become clear in the discussion, future work must address limitations associated with the current FT-TRA experimental design.  5.3 Results 5.3.1 Individual grain dissolution experiments (samples 1 – 5) High [Ca] concentrations were measured during the initial 500 seconds of both calcite and aragonite grain experiments, which can be attributed to incomplete removal of fines during acetone cleaning or high energy surface area sites produced during sample preparation (Fig. 5.3A, 5.3B). Measured [Ca] concentrations show clear plateaus under constant pH conditions (pH 2.6, 2.3 for calcite and pH 3.3, 2.3 for aragonite). Similarly, Ca increases/decreases linearly as eluent acidity is linearly ramped up/down before or after steady eluent composition. However, compared to observed values, model-derived [Ca] concentrations – assuming surface-controlled mineral dissolution – are an order of magnitude higher for calcite and aragonite (Fig. 5.3A, 5.3B). In fact, due to rapid dissolution, the modeled decrease in sample mass (therefore, surface area) translates in the absence of steady [Ca] plateaus under constant pH conditions. This pattern   123 is not observed experimentally, with sample loss representing less than 3 % of the original sample mass (Table 5.2).    Figure 5.3 Measured and modeled Ca concentrations (moles L-1) associated with individual mineral grain experiments. Left hand axis refers to experimental [Ca] concentrations whereas right hand axis refers to modeled   124 [Ca] concentrations. A: calcite rhomb dissolution experiments (samples 1 – 3). B: aragonite grain dissolution experiments (samples 4 - 5).  No distinct difference in surface area normalized dissolution rates between aragonite and calcite dissolution rates (moles m-2 s-1) could be observed (Fig. 5.4). Calcite, aragonite dissolution rates obtained in this study are significantly slower than previously reported rates (Fig. 5.4). Calcite dissolution rates measured as part of this study (samples 1 – 3) start to merge with rates measured by Busenberg and Plummer (1986) at approximately pH = 4.5, when the dissolution rate is dominated by hydrolysis.  Similar to results reported by Busenberg and Plummer (1986), relatively steady dissolution rates were found above pH = 4.5. Aragonite dissolution rates found in samples 4 – 5 are not only lower than previously reported rates (Chou et al., 1989), they only start to merge with the previously published rates at pH > 5.5. In addition – and most importantly for the purpose of this chapter – experimental data do not show a clear distinction between aragonite and calcite dissolution rates.     125  Figure 5.4 Experimental calcite, aragonite grain dissolution rates (shown on logarithmic scale) versus eluent pH (samples 1 – 4). Since no effluent pH was measured, experimental pH was calculated based on incoming eluent composition. Results associated with this study (samples 1 – 4) are shown as filled color-coded circles. Some previously reported studies are also shown as indicated.  SEM images taken prior to exposing individual calcite rhombs to flow-through dissolution show defined crystallographic features such as sharp rhomb corners (Fig. 5.5). However, images obtained after flow-through dissolution exposure reveal much smoother surfaces with a general absence of previously observed surface features. A general rounding of calcite rhomb corners is also observed.    126  Figure 5.5 Scanning Electron Microscope (SEM) images obtained before and after flow-through dissolution experiments, obtained during preliminary experiments. Shown here is an intact, as well as partially dissolved typical calcite rhomb exposed to identical eluent pH conditions as used in this study (pH 5.7 – 2.3). Figures on the left show intact calcite rhomb prior to dissolution experiments; figures on the right show calcite rhomb after dissolution experiments. Note smooth, generally rounded surfaces of partially dissolved calcite rhomb.  5.3.2 Powder experiments (samples 6 – 8) The measured calcite, aragonite dissolution rates (affected by transport-limitations, as shown in chapter 4) reveal lower dissolution rates compared to model results (Fig. 5.6A). The aragonite powder (sample 6) dissolves approximately an order of magnitude lower as compared to modeled rates, whilst the calcite powder (sample 7) dissolves at multiple orders of magnitude slower rates as compared to modeled rates. Similar to the individual calcite rhomb experiments (section 5.3.1 above), the disagreement in dissolution rates between experimental and modeled   127 dissolution rates can be attributed to the PHREEQC model assumption that calcite dissolution occurs under a surface-controlled dissolution regime. However, the distinctly lower calcite powder rates remain difficult to explain. Heterogeneous sample distribution within the flow-through cell may play a role, resulting in the development of a thick diffusive boundary layer (DBL, see chapter 4). Mg/Ca molar ratios of similar order of magnitude were measured for calcite and aragonite, whereas an order of magnitude difference in Sr/Ca molar ratios were found between the calcite and aragonite mineral end members (Table 5.3, Fig. 5.6B). End-member molar ratios to be used in the geochemical model were calculated based on the 400 – 1000 second time interval (Fig. 5.6B, Table 5.3).    128  Figure 5.6 Aragonite (sample 6), calcite (sample 7) end-member powder experiments. A: measured calcite, aragonite surface normalized dissolution rates (aragonite in red solid line, calcite in black solid; aragonite model results in dashed red line, calcite model results in dashed black line) B: 5-point running average end-member molar   129 ratios, shown to represent the eluent reactor renewal time (Mg/Ca shown on left y-axis, Sr/Ca shown on right y-axis);. Two vertical lines indicate time-interval used to quantify end-member Mg/Ca, Sr/Ca molar ratios.  Sample Description Mg/Ca Sr/Ca  molar ratio 6 Aragonite powder 3.16 ± 0.05(!) × 10!! 7.96 ± 0.03(!) × 10!! 7 Calcite powder 4.61 ± 0.02(!) × 10!! 0.22 ± 0.001(!) × 10!! 8 Arag. – Cal. powder mixture 4.18 ± 0.44(!) × 10!! 1.91 ± 0.20(!) × 10!! 2.14 ± 0.09(!) × 10!! 8 Calculated(4) Arag. – Cal. powder mixture 4.24 × 10!! 2.21 × 10!! Table 5.3 Measured Mg/Ca, Sr/Ca molar ratios in samples 6 – 8 (aragonite, calcite powder and powder mixture sample respectively) as well as calculated mixture molar ratios based on end-member composition (Arag. = aragonite; Cal. = calcite). (1) Error indicates 95% confidence interval (2σ) associated with data recorded between 400 – 1000 seconds.  (2) Calculated throughout sample run. Error indicates standard deviation. (3) Calculated based on Sr/Ca plateau present from 4000 seconds onwards. Error indicates standard deviation.  (4) Calculated mixture molar ratios (El/Camixture) taking into account calcite, aragonite proportions present in sample (massArag. or massCal.), as well as end-member molar ratios: !"!"!"#$%&' = !"!"!"#$ × 𝑚𝑎𝑠𝑠!"#$ + !"!"!"# × 𝑚𝑎𝑠𝑠!"# 𝑚𝑎𝑠𝑠!"!#$   In the first 500 seconds of the aragonite – calcite mixture experiment (sample 8), the measured [Ca] concentrations decreases exponentially (Fig. 5.7A). Since no acetone cleaning step was applied to the powder samples, fine particles produced during sample preparation, or rapid dissolution at high surface energy sites can be held responsible for this pattern. Excluding the initial, rapid drop in elemental concentrations during the first 500 seconds, modeled [Ca, Mg, Sr] concentrations consistently fall above experimental values (Fig. 5.7A). At t = 4000 seconds,   130 the eluent acidity was increased from pH 5.0 to 4.0. The resulting increase in modeled [Ca, Mg, Sr] concentrations – assuming surface-controlled mineral dissolution – is considerably larger as compared to the experimentally observed increase. Moreover, modeled [Sr] concentrations decrease shortly after this acidity increase, since aragonite fully dissolves, resulting in Sr solely being supplied by the calcite end-member. In parallel, initially observed mixture dissolution rates exponentially decrease (t < 500 seconds), after which they gradually decrease (Fig. 5.7B) before increasing again as eluent acidity increases at t = 4000 seconds. Observed dissolution rates remain below modeled dissolution rates throughout the experiment (excluding the initial 500 seconds).  Measured Mg/Ca ratios for the aragonite and calcite mineral powder mixture experiment (sample 8, shown in Fig. 5.8A) show a general scatter between both mineral end-members. The calculated Mg/Ca molar ratio expected if calcite, aragonite dissolve at the same rate (purple dotted line in Fig. 5.8A; Table 5.3) lies within the Mg/Ca standard deviation associated with the measured mineral powder mixture Mg/Ca molar ratio (Table 5.3).  Measured Sr/Ca molar ratios gradually increase over the course of dissolution at pH 5.0, reaching quasi steady state values at approximately after 4000 seconds (Fig. 5.8B), when eluent acidity was increased to pH 4.0. Here, Sr/Ca molar ratios merge with calculated mixture Sr/Ca molar ratios (similar calculation as done for Mg/Ca, purple dotted line in Fig. 5.8B). In fact, the calculated Sr/Ca molar ratio lies within the standard deviation range associated with Sr/Ca measurements from 4500 seconds onwards (Table 5.3).  The PHREEQC model, which does not account for transport-controlled mineral dissolution, predicts an increase in Mg/Ca over time (red solid lines in Fig. 5.8), as the modeled aragonite (low Mg/Ca end-member) dissolution increases and the Mg/Ca signal associated with   131 calcite (high Mg/Ca end-member) eventually becomes predominant. As dissolution rates increase under more acidic eluent conditions (t = 4000 seconds onwards), Mg/Ca molar ratios increase more rapidly towards the Mg/Ca molar ratio associated with the calcite end-member. At t ≈	5250 seconds, aragonite fully dissolves, resulting in solely calcite end-member Mg/Ca molar ratios (Fig. 5.7A). Similarly, Sr/Ca molar ratios are predicted to decrease as aragonite (high Sr/Ca end-member) dissolves, and the signal associated with calcite (low Sr/Ca end-member) becomes predominant (Fig. 5.8B). As acidity increases at t = 4000 seconds, Sr/Ca rapidly decreases as aragonite fully dissolves. Similar to modeled Mg/Ca, as aragonite becomes fully dissolved, calcite end-member Sr/Ca molar ratios are achieved.   132  Figure 5.7 Sample 8 (mineral mixture experiment) data. A: experimental (solid lines), modeled (dotted lines) [Ca] concentrations indicated on left y-axis, [Mg, Sr] concentrations (shown in black, red and green respectively. Dissolution eluent consisted of 10 mM HNO3 (pH = 5.0) until approximately 4000 seconds, after which acidity was increased to 100 mM HNO3 (pH = 4.0); B: measured dissolution rate (mol m-2 s-1) shown as a solid line, modeled dissolution rate shown as a dotted line.    133  Figure 5.8 Calcite, aragonite mixture (sample 8) experimental, model results. A: Mg/Ca molar ratios; B: Sr/Ca molar ratios. Experimental 5-point running average experimental molar ratios (green) are shown to represent the eluent reactor renewal time. Purple dotted line denotes calculated El/Ca molar ratio expected if the two minerals dissolve at the same rate. Red line indicates model results (which assumes surface-controlled dissolution). Black dashed lines represent end-member elemental ratios, as indicated.   134 5.4 Discussion 5.4.1 Mineral grain dissolution experiments (samples 1 – 5) The [Ca] concentrations obtained during dissolution of samples 1 – 5 are significantly lower than modeled effluent concentrations (Fig. 5.3). This is not surprising, since the model assumes a surface-controlled dissolution regime, which is not representative of dissolution kinetics occurring in the flow-through cell. A study assessing the dissolution regimes controlling calcite dissolution across a range of pH and flow rate conditions has demonstrated the presence of a concentration gradients across the diffusive boundary layer (DBL) under the flow-through experimental conditions (incoming eluent pH, flow rate) used in this study (Chapter 4). The presence of a concentration gradient across the DBL at the mineral surface will result in decreased calcite, aragonite dissolution rates (as can be seen by the relatively low aragonite, calcite dissolution rates in Fig. 5.4). Therefore, the offset between modeled and measured [Ca] concentrations (Fig. 5.3) can be attributed to the presence of a transport–limited dissolution rates. Observed dissolution rates are lower than previously reported rates by at least an order of magnitude below pH 4.0 (Fig. 5.4). Moreover, the higher degree of surface roughness associated with aragonite grains when compared to rhombic calcite grains (SEM imagery supplied in Appendix C: Figs. C.1, C.2) has likely resulted in an underestimated surface area. As a result, aragonite dissolution rates (moles m-2 s-1) have likely been overestimated, which would further deviate (lower) our experimental results from previous studies (Fig. 5.4). Most importantly, however, the measured aragonite dissolution rates (samples 4, 5) are within the measured calcite dissolution rates range (samples 1 – 3, Fig. 5.4). This finding agrees with previous studies, which report very similar calcite, aragonite dissolution rates under far-from-equilibrium conditions (e.g. Chou et al., 1989, Busenberg and Plummer, 1986). Hence, although the higher solubility   135 associated with aragonite compared to calcite is well established (e.g. Mackenzie and Lerman, 2006), this difference is not reflected in far-from-equilibrium dissolution experiments. Finally, the presence of a transport-controlled dissolution regime is further supported by the microscopic examination of the surface morphology associated with partially dissolved calcite grains (Fig. 5.5). Smooth, generally rounded surfaces, as observed as part of this study, have previously been associated with transport-controlled dissolution conditions (e.g. Berner, 1978). No similar SEM work was carried out for the aragonite grains.   5.4.2 Powder experiments (samples 6 – 8) Given the similar dissolution rates measured with aragonite and calcite grains under transport limited conditions, we did not anticipate that the flow-through technique will be able to sort minerals “according to their susceptibility to dissolution” (Klinkhammer et al., 2004). Nevertheless, calcite and aragonite powder dissolution experiments were carried out to test the hypothesis. Measured powder end-member dissolution rates (BET surface area normalized, Brunauer et al., 1938) reveal aragonite dissolution rates that are as much as two orders of magnitude faster than the observed dissolution rates for calcite (Fig. 5.6A). The transport-control of dissolution in a powder sample is likely enhanced in comparison with individual grain samples because of the physical nature of the sample in the FT-TRA set-up. Unlike conventional batch stirred reactors, the small flow-through cell volume associated with FT-TRA prevents any control of sample stirring. This limited control may heterogeneously distribute powder particles within the flow-through cell, a factor that may affect how much surface area is exposed to the incoming eluent. This process may have contributed to the relatively low calcite dissolution rates observed in sample 7 (Fig. 5.6A).   136 Measured [Ca] concentrations and dissolution rates associated with a known mixture of aragonite and calcite powder (sample 8) fall below modeled (surface-controlled dissolution) rates (Fig. 5.7A), as can be expected given the presence of a transport-controlled dissolution regime. In fact, the modeled results predict a complete dissolution of aragonite shortly after eluent pH is decreased to pH 4.0 (t = 4000 seconds) at approximately t ≈	5250 seconds ([Sr] concentration drop in Fig. 5.7A). No such drop in [Sr] concentrations is observed experimentally.  For sample 8, we are able to quantify Mg/Ca ratio of the mineral mixture simply by considering end-member molar ratios of Mg/Ca along with the proportions of each end member in the mineral mixture (Table 5.3, Fig. 5.8A). In addition, the 10-second running average of the average Mg/Ca molar ratio remains close to the calcite end-member, which is expected if the two minerals dissolve at similar rates. In contrast, the model predicts Mg/Ca to lie closer (for the majority of the experiment) to the more soluble aragonite end-member, which is expected if the more soluble mineral dissolves faster. Similarly, modeled Sr/Ca molar ratios initially lie closer to the – more soluble – aragonite end-member (Fig. 5.8B). However, experimental results reveal that the calcite signal dominates the dissolution reaction, with Sr/Ca molar ratios gradually moving towards the Sr/Ca molar ratio calculated from the proportion of the two minerals in the sample, assuming identical dissolution rates. As the acidity is increased to pH 4.0 (t = 4000), both minerals dissolve at roughly equal rates, thereby agreeing with the Sr/Ca molar ratios calculated from the proportion of the two minerals in the sample, assuming identical dissolution rates.  (Table 5.3, Fig. 5.8B).  In addition, an unexpected observation in sample 8 is that the calcite powder dissolves faster than the aragonite powder during the initial 4000 seconds, resulting in lower Sr/Ca molar ratios compared to the Sr/Ca molar ratios calculated from the proportion of the two minerals in the   137 sample assuming identical dissolution rates (Fig. 5.8B). This behavior is opposite from what we would expect based on the mineral grain dissolution experiments (section 5.4.1). Although we are unable to provide a clear explanation for this observed behavior, this may the result of (a) transport-limited dissolution associated with the physical nature of a powder sample; (b) a problem associated with the BET surface area quantification; or (c) a combination of (a) and (b). At pH 4.0, the system is fully transport controlled (Chapter 4).  Two key conclusions can be made based on the mineral powder mixture experiment: (1) the end-member mineral solubility does not determine the bulk molar ratio signal, as the dissolution reaction is controlled by transport processes; (2) even if surface-controlled dissolution conditions were to be achieved using the flow-through dissolution technique flow-through dissolution would not sort minerals according to susceptibility to dissolution. This can be illustrated using the model result obtained for sample 8: no single Mg/Ca or Sr/Ca time-envelope can be selected which would be representative of the more soluble aragonite mineral end-member (Fig. 5.8). Fundamentally, all minerals physically exposed to dissolution will dissolve simultaneously, albeit at different rates, determined by (1) their intrinsic dissolution rate parameters and (2) their relative surface area.   5.4.3 Previously reported evidence challenging the FT-TRA premise (1) The premise on which FT-TRA data analysis has been based states that “material is sorted during dissolution based on its susceptibility to protons” (Klinkhammer et al., 2004). Haley and Klinkhammer (2002: Fig. 5.7) showed 2 plateaus of distinct Mg/Ca, Sr/Ca ratios recorded during nitric acid sample dissolution. However, results shown by Benway et al. (2003: Fig. 2) and Klinkhammer et al. (2004: Fig. 5.3) do not show steady Mg/Ca plateaus. Instead, a   138 gradual decline in Mg/Ca was observed, and interpreted to represent “ordered dissolution of distinct types of calcite in the foraminiferal test based on Mg content” (Benway et al., 2003). Assuming the above hypothesis is correct, a distinct plateau of relatively high Mg/Ca representative of ontogenetic calcite, followed by a distinctly lower Mg/Ca plateau representative of gametogenic calcite should have been found, as previously reported by Haley and Klinkhammer (2002).  (2) Assuming “FT-TRA sorts material based on susceptibility to dissolution” (Klinkhammer et al., 2004), sample pre-treatment should be of minimal importance. In reality, sample pre-treatment involving cracking the foraminifera shells and cleaning with ethanol led to inconsistent FT-TRA derived Mg/Ca molar ratios (Fig. 5.5 in Klinkhammer et al., 2004), illustrating that the absence of physical sample preparation is best for obtaining reproducible FT-TRA-based Mg/Ca. In particular, the use of the cleaning agent ethanol was found to play a considerable role: cleaning without cracking increases FT-TRA Mg/Ca as clays are preserved in the presumably intact foraminiferal test at the expense of ontogenetic calcite, whereas cleaning with cracking decreases FT-TRA Mg/Ca as the clays are more easily removed from the cracked pieces and the ontogenetic calcite is partially dissolved. This finding implies the importance of the physical nature of the sample undergoing FT-TRA. Cracking but not cleaning the foraminiferal test with ethanol would have been able to provide additional insight, as this would have exposed clays (much less soluble compared to biogenic carbonates) whilst preventing the preferential dissolution of ontogenetic calcite during cleaning. The greater the exposed surface area of the inner, ontogenetic layer (when cracked), the higher the likelihood of capturing the signal of interest.    139 (3) Initially, the FT-TRA technique used distinct cleaning reagents (Table 5.1) such as hydroxylamine, which was used for the removal of oxide coatings (Haley and Klinkhammer, 2002). This eluent was brought up to pH 9.0 with ammonium hydroxide, to avoid calcite dissolution. Later, this dissolution step was also deemed responsible for removing “high-Mg calcite”, even though very little Ca dissolved under such conditions (Fig. 2 in Klinkhammer et al., 2004). In fact, when calculating calcification temperatures based on the reported Mg, Ca data in Fig. 2 in Klinkhammer et al. (2004) using the sediment trap calibration from Anand et al. (2003), erroneously high calcification temperatures are obtained. Hence, the initial mineral phase removed using hydroxylamine remains to be determined. Furthermore, later work increasingly simplified the range in eluents used and was eventually limited to a combination of DIW and nitric acid eluents (e.g. Klinkhammer et al., 2009). In other words, the importance of the hydroxylamine-cleaning step remains unclear as it was eventually excluded.  (4) Sadekov et al. (2010) analyzed foraminiferal tests from the same core-top sample using both laser-ablation coupled to ICP-MS, as well as FT-TRA. These authors observed that (1) laser-ablation was able to locate and resolve much greater intra-test Mg/Ca molar ratio variation as compared to the FT-TRA approach; (2) no preferential dissolution of high Mg/Ca carbonate phases occurs during FT-TRA (Fig. 6 in Sadekov et al., 2010) and (3) merely exposed surfaces dictate which mineral phase will dissolve during FT-TRA.   5.4.4 Alternative data interpretation and a possible path forward In both the present, as well as previous studies far-from equilibrium dissolution experiments carried out on calcite, aragonite show similar dissolution rates (Fig. 5.4). This is most pronounced below pH 4.0, where Busenberg and Plummer (1986) measure close to identical   140 dissolution rates for both aragonite and calcite. Since existing FT-TRA methods use eluent acidity of pH < 4.0 (Table 5.1), ontogenetic and gametogenic calcite will dissolve at similar rates, irrespective of their difference in solubility. The net elemental signal recorded during FT-TRA is the sum of reaction products from each exposed mineral phase undergoing dissolution. The decreasing Mg/Ca signal over time (e.g. Fig. 2 in Benway et al., 2002), may be explained by the dissolution of an initially higher proportion of exposed ontogenetic calcite, high in Mg/Ca. Over time, the relative contribution of the dissolving gametogenic calcite will increase, increasingly dissolving a higher proportion of lower Mg/Ca calcite. However, this hypothesis requires the physical access of the incoming eluent to the inner, ontogenetic layer, which remains unclear. Therefore, we are at this stage unable to explain the decreasing Mg/Ca curve reported by Klinkhammer et al. (2004: Fig. 3, species not reported) and Benway et al. (2002: Fig. 2: G. ruber and G. sacculifer). In order to geochemically differentiate between ontogenetic and gametogenic calcite, a twofold approach is required:  (1) in terms of the experimental conditions, parameters (pH, flow rate, etc.) must be adjusted to allow for a mixed- or surface-controlled dissolution regime. Once a mixed or full surface-controlled dissolution regime is established, although ontogenetic and gametogenic calcite will dissolve concurrently, they will do so at slightly different rates. This small difference in dissolution rate – in combination with either equal exposed surface areas, or a higher exposure of the more soluble mineral phase – may be exploited as part of a redesigned data analysis approach to extract end-member composition such as ontogenetic !"!" ! and gametogenic !"!" ! molar ratios. Alternatively, even if it remains a challenge to achieve a mixed- or surface-  141 controlled dissolution regime, by exposing a different proportion of ontogenetic !"!" ! and gametogenic !"!" ! calcite through time, which is automatically achieved unless both minerals are present in equal amounts and dissolve at the same rate.  (2) the data analysis associated with FT-TRA data used to isolate the ontogenetic signal must be revisited. We propose a new approach, exploiting the vast Mg/Ca, Sr/Ca dataset typically generated over the course of a FT-TRA experiment. Consider the following scenario where solely two mineral phases are present: ontogenetic (𝐶𝑎!) and gametogenic calcite (𝐶𝑎!). These two minerals – to some degree – dissolve at different rates, and the measured, total signal is denoted subscript 𝑇: !"!" !!"#$!! =  !"!" ! ×  !"!!!"!!   +  !"!" ! ×  (!"!! )!(!"!! ) !"!!         (5.9) !"!" !!"#$!! =  !"!" ! ×  !"!!!"!!   +  !"!" ! ×  (!"!! )!(!"!! ) !"!!       (5.10) !"!" !!"#$!!!! =  !"!" ! ×  !"!!!!!"!!!!   +  !"!" ! ×  (!"!!!!)!(!"!!!!) !"!!        (5.11) !"!" !!"#$!!!! =  !"!" ! ×  !"!!!!!"!!!!   +  !"!" ! ×  (!"!!!!)!(!"!!!!) !"!!      (5.12) !"!" !!"#$!!!! =  !"!" ! ×  !"!!!!!"!!!!   +  !"!" ! ×  (!"!!!!)!(!"!!!!) !"!!        (5.13) !"!" !!"#$!!!! =  !"!" ! ×  !"!!!!!"!!!!   +  !"!" ! ×  (!"!!!!)!(!"!!!!) !"!!      (5.14) !"!" !!"#$!!!! =  !"!" ! ×  !"!!!!!"!!!!   +  !"!" ! ×  (!"!!!!)!(!"!!!!) !"!!        (5.15) !"!" !!"#$!!!! =  !"!" ! ×  !"!!!!!"!!!!   +  !"!" ! ×  (!"!!!!)!(!"!!!!) !"!!      (5.16)   142 The above eight equations (5.9 – 5.16) contain eight unknowns ( !"!" ! , !"!" ! , (𝐶𝑎!! ), !"!" ! , !"!" ! , 𝐶𝑎!!!!, 𝐶𝑎!!!!, 𝐶𝑎!!!!), which can therefore theoretically be solved. A minimum of 23 (8) equations is required to solve for two mineral phases, 33 (27) for three mineral phases, etc. Since a typical FT-TRA experiment generates a virtually unlimited amount of time-resolved data, the system becomes over-determined. Hence, this data analysis approach can take full advantage of the time-resolved analysis dataset.  Under the current FT-TRA experimental configuration, conditions are unable to exploit difference in mineral solubilities as demonstrated here using aragonite, calcite. Hence, the data analysis approach suggested above is not – yet – applicable.   5.5 Conclusions and future research directions The original idea behind FT-TRA was that it would provide a means of distinguishing between different mineral phases (such as ontogenetic versus gametogenic calcite). This study demonstrates that the sorting of material according to its susceptibility to dissolution is not possible using FT-TRA, as currently configured. By doing so, this study builds on similar observations made by Sadekov et al. (2010). Instead, mineral dissolution within the flow-through cell is dictated by: (1) the intrinsic rate parameters associated with each mineral and exposed surface area, (Chapter 3) and (2) the degree of transport-controlled dissolution (as illustrated in Chapter 4).  In order to potentially develop the use of FT-TRA in paleoceanography, the flow-through configuration needs to be adjusted to achieve at least a partially surface-controlled dissolution regime, allowing for a degree of differential dissolution between mineral phases, which could   143 come from different timing of exposure to the eluent. This data can then be interpreted using the approach outlined in section 5.4.4. However, given the physical nature of a foraminiferal shell sample, it remains technically challenging to achieve such conditions (surface-controlled calcite dissolution has previously only been achieved at low pH using a single exposed calcite face using a custom-made flow-cell developed by Compton and Unwin, 1990). Alternatively, if both biogenic minerals are present in different proportions, the difference in exposed surface area may be sufficient to apply the data analysis as suggested in section 5.4.4. However, this hinges on the assumption, that both biogenic calcite phases are homogeneous with respect to El/Ca molar ratios, which has been demonstrated to not be the case using microanalytical techniques (e.g. Sadekov et al., 2010). Therefore, we suggest the use of FT-TRA to remain limited in paleoceanography as an automated sequential leaching tool, using distinct reagents as originally done by Haley and Klinkhammer (2002). Alternative microanalytical techniques such as laser-ablation ICP-MS (LA-ICP-MS) are able to resolve much greater spatial Mg/Ca molar ratio variation (e.g. Sadekov et al., 2010), and should be explored further as a valuable tool to calibrate the Mg/Ca paleothermometry proxy.  Lastly, although beyond the scope of this thesis chapter, the possibility was explored to apply the FT-TRA technique to extract biogenic Mg/Ca, Sr/Ca molar ratios from calcareous ostracod shells (Appendix C).    144 Chapter 6: Measuring metal release rates from mine waste rock: a preliminary case study  6.1 Introduction During the weathering of mine waste, a number of processes facilitate the production of acidity and associated (heavy) metal release: iron sulfide oxidation, dissolution of soluble iron sulfate minerals, and the dissolution of less soluble sulfate minerals of the jarosite-alunite group (e.g. Bigham and Nordstrom, 2000). This contaminated effluent is often referred to as Acid Rock Drainage (ARD, e.g. Egiebor and Oni, 2007) and can result in significant environmental pollution (e.g. Simate and Ndlovu, 2014). Once initiated, ARD may persist for hundreds of years (e.g. Arnesen and Iversen, 1997). Therefore, the prediction of drainage chemistry remains an important component of environmental risk assessment studies (e.g. Price, 2009).  Numerous testing procedures have been devised and are presently available for the prediction of acid rock drainage generation. For example, kinetic tests were designed to monitor mine waste weathering, and aim to predict the characteristics of mine waste effluent over time. Kinetic tests include laboratory-based trickle leach columns, humidity cells as well as field-based test cells (e.g. Price, 2009).  Humidity cell experiments involve exposing mine waste to dry and humid air (at room temperature) in an alternating fashion. During the dry air cycle, pore water evaporates, resulting in the concentration of dissolved ions in the pores and likely precipitation of salts on particle surfaces. In addition, hydrated precipitates may partially dehydrate and stabilize. During the humid air cycle, water and air is introduced to the sulfides, allowing for the oxidation process to   145 occur. As part of the humidity cell procedure, weekly flushing with water is carried out, allowing for oxidation products formed during the dry and humid air cycle to dissolve and come in contact with neutralizing minerals. The effluent is used to monitor leachate water quality. Humidity cell experiments are typically restricted to more reactive, small particles (< 10 mm in diameter, Price, 2009). As the precipitation and dissolution of secondary weathering products often determines mine drainage chemistry, humidity cell experiments are most useful for constraining dissolution rates of primary minerals, but are more limited for predicting leachate composition under natural conditions, which is affected by precipitation of secondary minerals. In comparison, trickle leach column experiments study both primary and secondary reactions (secondary mineral precipitation) at the same time, thereby simulating natural conditions more accurately (Price, 2009). This is achieved by adjusting the design of trickle leach cells to simulate key aspects of the weathering and leaching conditions such as temperature, particle size distribution, eluent pH and flow rate (Price, 2009).  In the end, data collected during kinetic tests provides an indication of (1) time to ARD onset; (2) drainage chemistry, as well as (3) the degree of acid generation and neutralization (e.g. Sapsford et al., 2009). Combined with bulk digestion, elemental depletion times can be calculated if steady-state drainage chemistry conditions are reached.  6.1.1 Challenges associated with kinetic testing A number of challenges are associated with kinetic testing, including (a) scaling up of kinetic test results, and (b) the time-consuming, costly nature of these tests: (a) Scaling up of the kinetic test results to make predictions of the long-term weathering reactions of mine waste requires consideration of the differences between the laboratory-based   146 kinetic test and the actual mine waste dump (e.g. Frostad, 1999). The scaling up issue has been illustrated by Brodie et al. (1991) by comparing results from a standard humidity cell (0.5 kg waste rock, < 200 µm size fraction, flushed once a week, allowing for partial and temporary water saturation) to a larger scale modified humidity cell (50 – 65 kg waste rock, < 10,000 µm size fraction, flushed gradually over 5 days, never allowing for saturation) carried out on identical rock types. Over a 12 week leaching period, the effluent pH of the large scale modified humidity cell dropped from 7.4 to 3.6 whereas in the standard humidity cell effluent pH merely dropped from 7.3 to 6.0. Other scaling up discrepancies have been documented by Price (1997) when comparing humidity cell and field test pad sulphate release rates. Here, under both acidic and neutral conditions, sulphate release rates per kg of mine waste from the field test pads were 0.3% and 1.5% of sulphate release measured in humidity cells. Hence, the issue of scale remains difficult to address due to parameters such as spatial and temporal differences in flushing rates, as well as differences in exposed surface areas related to grain size distribution. (b) Kinetic tests to date are generally operated until analyte release rates become relatively stable. For humidity cells, this requires at least 40 weeks of testing, but may require over a year (e.g. Price, 2009). In some cases, even after years of testing, no relatively stable release rates are obtained (section 6.1.3). Hence, kinetic testing frequently becomes a relatively time-consuming, therefore costly procedure. One of the initial motivations behind this study was to explore the added value of a mixed flow reactor (MFR) as a useful tool in the kinetic testing toolbox. In particular, this study aims to assess whether a mixed flow reactor (section 1.3.2) can provide useful information in a more manageable experimental time frame (here: 6 months).    147 6.1.2 Study setting The proposed Casino Project consists of a copper, gold, molybdenum and silver deposit located in west-central Yukon, 300 km northwest of the territorial capital of Whitehorse (62º 44’ 10” N; 138º 49’ 45” W). The ore deposit is topped by a well-defined leached oxide cap (CAP), which contains jarosite. The jarosite-alunite group of sulphate minerals is a byproduct of the oxidative dissolution of sulfide minerals (explained in further detail in Appendix D).  The oxide CAP zone was produced by weathering that extends down to an average depth of 70 m. Primary sulphide minerals that were originally present in the CAP rock have weathered in-situ, which has resulted in the formation of boxwork textures partly filled by iron oxide minerals and jarosite (archetypal jarosite = AB! (OH)!(SO!)!, where A is K! and B is Fe!!, giving the ideal end-member composition 𝐾𝐹𝑒!(𝑆𝑂!) !(𝑂𝐻)!). Although jarosite has Fe3+ as the dominant trivalent cation in the B site, many other isostructural minerals occur via simple or coupled cation substitutions (including Cu2+, e.g. Table 1 in Welch et al., 2007).  6.1.3 Study rationale In order to establish an environmentally sound future mine waste management strategy, a trickle leach column kinetic test experiment was conducted on a sample collected from the leached oxide cap to predict drainage quality. In particular, a 3-year trickle leach experiment was conducted on 10 kg of rock sample to assess the potential ARD impact that could be produced from piles of excavated CAP material (experiment involved weekly flushing using 300 mL of DIW). The trickle leach experiment revealed elevated, slowly decreasing effluent [Cu] concentrations, as well as slightly depressed pH conditions (Fig. 6.1). As the dissolution reaction progresses slowly, it remains difficult to extrapolate the Cu release rate into the future. This case   148 study provides an opportunity to explore whether more useful information can be gained in a more timely and economical manner using a mixed flow reactor.   Figure 6.1 Trickle leach effluent [Cu] concentrations (connected filled black dots), as well as effluent pH (connected red crosses). Area between dashed lines represents pH range of the incoming eluent (pH 5.3 – 5.5).   6.1.4 Study objectives In this study, we aim to address the following questions: a) Explore and assess the added value of a mixed flow reactor (referred to as MFR, section 1.3.2) experimental design. After having developed FT-TRA, the development of a MFR merely requires replacing the small volume flow-cell with a larger volume flow-through reactor. In addition, instead of analyzing the effluent online as typically done in FT-TRA, Cu (moles L-1)pH Cu (x 10-6  moles L-1)246810121416pH4.44.64.85.05.25.45.65.86.06.2Time (days)0 100 200 300 400 500 600 700 800 900 1000 1100  149 effluent fractions are periodically collected using a fraction collector for offline analysis. As part of this study goal, MFR results will be compared to conventional trickle leach results; b) Compare and determine the implications of the leaching behavior using a MFR experimental design for different eluent pH conditions; c) Assess environmental implications based on knowledge gained from all kinetic tests.  6.2 Materials and methods 6.2.1 Sample characterization A leached oxide cap sample was collected in the field in January 2010. The sample was crushed in a 5x7 jaw crusher to approximately 80 weight % passing through < 3/8 inch. Representative splits of the crushed samples were made using a 12.7 mm opening splitter box. Material from these sample splits was used to carry out both the trickle leach, as well as the mixed flow reactor (MFR) leaching experiments. In order to produce a finely ground sample for sample characterization, 200 g of crushed sample was placed in a ring pulverizer to produce 80 weight % passing through < 200 µm mesh.  Total metal content was quantified on the finely ground sample using both aqua regia (Acme Analytical Laboratories Ltd.) as well as a four-acid digestion (ALS Canada Ltd.). In the aqua regia based sample digestion protocol, 0.5 g of pulverized sample is digested at 95ºC for 1 hour. The extract is then diluted to 10.0 mL and analyzed for metals by ICP-MS. In the four-acid digestion protocol, 0.25 g of pulverized sample is digested using HClO4, HNO3, HF and HCl acids. The residue is topped up with dilute hydrochloric acid and analyzed by ICP-AES followed by – after appropriate dilution – ICP-MS. Digestion results are provided in Table 6.1. As   150 expected, elements associated with alumino-silicates and barites (Na, Al, K, Sr and Ba) were extracted to a greater degree using the more aggressive four-acid digestion.    Aqua regia digestion Four-acid digestion Na (%) 0.044 0.51 – 0.56 Mg (%) 0.15 0.34 – 0.34 Al (%) 0.82 7.45 – 7.52 K (%) 0.31 3.46 – 3.58 Ca (%) 0.11 0.53 – 0.61 Mn (ppm) 33 36 – 37 Fe (%) 1.5 1.68 – 1.78 Cu (ppm) 123.8 101.5 – 115 Sr (ppm) 21 133 – 142.5 Ba (ppm) 328 1680 – 1700 S (%) 0.32 0.26 – 0.28 Table 6.1 Elemental content (expressed in % and ppm) of waste rock sample, obtained using aqua regia digestion (1st column) and a more extensive four-acid procedure (2nd column). Four-acid digestion range represents 2 separate analyses (n = 2).  In addition, the finely ground sample was analyzed using quantitative x-ray diffraction analysis. For this, the sample was reduced to the optimum grain-size range for quantitative X-ray analysis (<10 µm) by grinding under ethanol in a vibratory McCrone Micronising Mill for 7 minutes.  Continuous-scan X-ray powder-diffraction data were collected over a range 3-80°2θ with CoKα radiation on a Bruker D8 Advance Bragg-Brentano diffractometer equipped with an Fe monochromator foil, 0.6 mm (0.3°) divergence slit, incident- and diffracted-beam Soller slits   151 and a LynxEye XE detector. The long fine-focus Co X-ray tube was operated at 35 kV and 40 mA, using a take-off angle of 6°. The X-ray diffractogram was analyzed using the International Centre for Diffraction Database PDF-4+ and Search-Match software by Siemens (Bruker). X-ray powder-diffraction data were refined with Rietveld program Topas 4.2 (Bruker AXS). These amounts represent the relative amounts of crystalline phases normalized to 100%. Results are provided in Table 6.2. Errors (accuracy and precision) are difficult to assess for Rietveld refinement results. Although the computational precision of Rietveld refinements is high, often less than 1% relative, the actual precision, because of sample characteristics (crystallite size and shape, microabsorption, solid solutions, etc.) is of the order of several percent relative.  Accuracy can only be estimated by preparing synthetic equivalents of a particular sample and doing the analysis with repeats, and also by using independent methods such as bulk chemical analysis where applicable. The precision for major phases is high, a few percent relative, but precision is low for minor phases, especially below a few percent. Although the relative error may be high, the absolute error is very small. It must be noted that x-ray diffraction results are normalized to 100% so phases not detected (amorphous, extreme nanoscale) are not included.          152 Mineral Ideal Formula Sample (%) Quartz  SiO2 45.2 Clinochlore (Mg,Fe)5Al(Si3Al)O10(OH)8 1.8 Kaolinite Al2Si2O5(OH)4 1.4 Muscovite KAl2AlSi3O10(OH)2 29.8 Plagioclase NaAlSi3O8 – CaAl2Si2O8 10.7 K-feldspar KAlSi3O8 8.5 Jarosite KFe3(SO4)2(OH)6 2.6 Total  100.0 Table 6.2 Sample composition expressed in percentage based on x-ray diffraction.  The mineral grain size distribution associated with the trickle leach experiment was obtained by sieving. For the MFR experiments, mineral grain size distribution was obtained using a combination of dry sieving and laser diffraction. For laser diffraction, a suspension of 1 g of sample in 3 mL of DIW was prepared of the < 0.15 mm size fraction. This suspension was homogenized using a glass rod before an aliquot was drawn with a sample dropper for particle analysis. Ultrasonic treatment was applied for 1 min to disperse any aggregates in the sample prior to analysis using a Malvern Mastersizer 2000. The cumulative MFR sample size fraction breakdown obtained using sieving and laser diffraction is provided in Fig. 6.2.     153  Figure 6.2 Cumulate volume (%) of sample versus particle size (note log-scale). Samples used in trickle leach experiment (leached cap rock, filled black connected circles) as well as sample used in MFR experiment (milled leached cap rock, open red connected circles) are shown. Trickle leach experiment data obtained using particle sieving; flow-through experiment data obtained using a combination of particle sieving (> 0.15 mm), laser diffraction (< 0.15 mm) methods.  6.2.2 Trickle leach experiments A trickle leach experiment was carried out by SGS Canada Inc. between January 20th, 2010 and December 5th, 2012 under temperature-controlled conditions (20 – 22 °C).  A plexiglass column, (inner diameter = 10 cm, height = 116 cm) was filled with 10 kg of dried crushed rock sample (height of sample = 78 cm). 300 mL of DIW eluent was passed through the column per week, with effluent collection occurring once a week. The trickle leach experimental design is shown in Fig. 6.3. Trickle leach effluent pH and elemental composition (Na, Ca, K, Si, Mg, Cu, Mn, Sr, Fe, Ba, Fe) was determined by ICP-MS by SGS Canada Inc. flow-through experiment trickle leach experimentCumulative volume (%)0102030405060708090100size (mm)10−4 10−3 10−2 10−1 1 101  154  Figure 6.3 Experimental trickle leach design. 300 mL of DIW eluent is introduced (on a weekly basis) to the top of a plexiglass column holding 10kg of crushed waste rock. The trickle leach effluent is collected on a weekly basis.  6.2.3 Mixed Flow Reactor (MFR) experimental design The mixed flow reactor experiments placed a 5 g waste rock subsample in a 50 mL Teflon flow-through cell (FIALabs, Bellevue, WA), with a stirrer bar at the bottom to ensure well-mixed conditions. A Dionex ICS-3000 pump was used to continuously introduce the eluent (DIW or HNO3) at a constant flow rate of 0.05 mL min-1, resulting in an eluent residence time of 1000 min (16.6 hours). Effluent samples were collected over a period of 150 minutes (resulting in a   155 7.5 mL sample), followed by 700 minutes of effluent sent to waste using a Foxy® R2 fraction collector equipped with a diverter valve (Teledyne ISCO). The initial 700 minutes of effluent generated during an experiment (including after restarting an experiment after an accidental shutdown) was sent to waste. The MFR experimental design is shown in Fig. 6.3.  Two separate pH measurement approaches were used in the MFR experiments: (a) For the initial DIW MFR experiment, the effluent passed through a flow-cell containing an in-line pH electrode before being sent to the fraction collector for periodic sample collection. pH was continuously measured using a Bio-Rad DuoFlow in-line pH electrode (Catalogue number 760-2040) connected via a Dionex UCI-100 universal chromatography interface to a PC (Dionex Chromeleon™ 6.80 software). Due to the continuous data recording nature of this set-up, pH data collected during fraction collection was averaged over the period of time when fractions were collected. For calibration, 3 commercially available pH buffers were used (pH 4, 7 and 10). The pH electrodes can be deemed accurate to within 0.1 pH units. (b) For the follow-up DIW MFR and the acidic MFR experiments, individual fractions were measured offline, using a Beckman Coulter 350 pH meter was used at the UBC Department of Chemistry. The pH meter was calibrated on a weekly basis using commercially available pH buffers (pH 4, 7 and 10), and is deemed accurate to within 0.1 pH units.    156  Figure 6.4 Experimental MFR design. Eluent (blue arrow) is continuously pumped through a 50 mL mixed flow reactor in which the sample is placed (convex bottom, minimizing sample grinding due to rotating stirrer bar), after which the effluent (red arrows) passes through a 0.2 µm syringe filter (preventing sample loss), and continues via an in-line pH sensor prior to periodic fraction collection. When not collecting fractions, effluent is sent to waste.  6.2.4 MFR experiments MFR experiments were carried out using both dilute nitric acid (referred to as the acidic MFR experiment), as well as deionized water (DIW) eluent (referred to as the DIW MFR experiment) for approximately 6 months each, at room temperature (20 – 22ºC). The DIW eluent bottle was in equilibrium with ambient 𝑝𝐶𝑂!, with pH 5.7. The dilute nitric acid experiment used a 5 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! HNO3 solution, resulting in an eluent pH of 2.3.  During the DIW eluent experiment, an accidental flow stoppage occurred after 65 days, which was resolved 17 days later. No accidental stoppages occurred during the acidic MFR experiment.    157 6.2.5 MFR effluent elemental and sulphate measurements Elemental analysis was carried out using an autosampler connected to an Agilent 7700× quadrupole ICP-MS in He-mode. 23Na, 24Mg, 27Al, 39K, 44Ca, 55Mn, 56Fe, 63Cu, 88Sr, 115In and 138Ba were measured for 0.3 seconds per mass (0.6 seconds for 115In). 5 scans were carried out per mass and counts per second (cps) averaged. 115In was used as an internal standard to correct for instrumental drift. Standard solutions (Al, Na, Mg, Ca, Mn, Fe, Cu, Sr, Ba: 0 – 50 ppb; Si: 0 – 26 ppb, K: 0 – 500 ppb) were made gravimetrically using a microbalance. Elemental concentrations were calculated using Agilent MassHunter workstation software version B.01.01. The range in absolute and relative errors on concentrations (95% confidence interval) from highest to lowest concentration measured is expressed in % in Table 6.3.   Acidic experiment DIW experiment Element [High] [Low] [High] [Low] 27Al 62.7 (3.3)   15.2 (2.1) 1.2 (25.8)  0.6 (51.2) 23Na 121 (3.3)   3.4 (2.3) 53.6 (3.9)  1.0 (15.3) 24Mg 52.5 (2.1)  2.4 (1.9) 4.8 (14.8)  0.8 (25) 28Si 44.4 (2.8)  1.6 (1.5) 17.2 (5.6) 2.9 (11.5) 39K 155 (3.4)  1.5 (1.2) 145 (3.7) 23.27 (3.5) 44Ca 73.2 (3.8)  2.6 (5.5) 1.0 (5.1) 0.7 (35.1) 55Mn 5.2 (2.0) 0.04 (2.6)  0.4 (15.6) 0.6 (216) 57Fe 50.1 (4.1) 8.8 (1.2)  0.7 (29.5) 0.7 (194) 63Cu 21.3 (2.5) 0.08 (1.6)  0.1 (4.0) 0.7 (450) 88Sr 2.9 (1.9) 0.05 (0.9)  0.1 (6.9) 0.7 (398) 138Ba 27.6 (3.3) 0.1 (0.7) 1.3 (2.9) 0.22 (16.3) Table 6.3 Absolute error, expressed in ppb (relative error on concentrations, expressed in %).    158  Sulphate concentrations were measured using a Metrohm IC instrument, using a 150 mm Metrohm metrosep A Supp 5 – 150/4.0 IC column. In order to preserve IC column quality, samples obtained as part of this study were diluted in DIW (where applicable) to raise pH to a minimum of 3. A high sulphate standard of approximately 100 mg L-1 was made using Na2SO4(s). This high standard was diluted to generate 6 standards, covering a sulphate range of approximately 0.7 – 50 mg L-1. Measurement precision ranged from 0.5 to 2.5 % from high to low concentrations. Replicate analyses were carried out on a middle standard solution during each experimental run, revealing a relative standard deviation (RSD) ranging from 4 – 25 %. This relatively poor reproducibility can be attributed to a deteriorating IC column.   6.3 Results 6.3.1 Trickle leach experiments 6.3.1.1 pH Effluent pH in the trickle leach experiment stayed mildly acidic throughout the experiment, ranging from 4.5 to 6.0 (Figure 6.1). A gradual increase in effluent pH can be observed, with a small drop occurring at t = 700 days. No simultaneous changes in metal or sulphate concentrations could be observed.  6.3.1.2 Elemental and sulphate concentrations Concentrations in the trickle leach solution vary as a function of time following various patterns for different elements:   159 • Ca, Mg, K, Sr, Cu and Mn essentially co-vary, decreasing by about 10-fold during the initial 64 days, followed by a 2 – 3 fold increase peaking at t = 260 days before gradually decreasing to their lowest values by the end of the experiment (Fig. 6.5A, 6.5B); • Al concentration shows a similar pattern to t = 400 days (including the maximum at t = 260 days) and stays quasi constant to the end of the experiment (Fig. 6.5B). An initial high Al release is also identified during the initial 20 days; • Na concentration decreases more than 100-fold during the first 400 days and stays constant to the end of the experiment (Fig. 6.6); • Si concentrations stays quasi constant during the initial 300 days, then gradually decreases for the remainder of the experiment (Fig. 6.6); • Fe concentration decreases 100-fold during the first 200 days and stays roughly constant albeit with sporadic increases (Fig. 6.7); • Sulphate concentration decreases quasi exponentially during the course of the experiment (Fig. 6.7); • Ba initially decreases during the initial 36 days, and then gradually increases for the remainder of the experiment (Figure 6.8).   160  Figure 6.5 Trickle leach experiment effluent concentrations, expressed in moles L-1. A: Ca (initial, highest value not shown but indicated), K and Mg concentrations; B: Cu, Mn, Sr, Al (initial, highest value not shown but indicated).    161  Figure 6.6 Trickle leach experiment effluent Na (connected filled black circles, left y-axis) and Si (connected filled light blue circles, right y-axis) concentrations, expressed in moles L-1.  Na Si Na concentration (moles L-1)00.0010.0020.0030.0040.0050.0060.0070.0080.0090.0100.0110.012Si concentration (moles L-1)0.00060.00070.00080.00090.00100.00110.00120.00130.0014Time (days)0 200 400 600 800 1000  162  Figure 6.7 Trickle leach experiment effluent sulphate (connected filled black circles, left y-axis) and Fe (connected filled red circles, right y-axis) concentrations, expressed in moles L-1.    163  Figure 6.8 Trickle leach experiment effluent Ba concentrations, expressed in moles L-1.  6.3.2 MFR experiments 6.3.2.1 pH Effluent pH associated with the DIW eluent MFR experiment was not obtained during the initial 125 days, due to unanticipated drift of the pH probe calibration, which was not monitored. During the final 81 days, the pH probe was calibrated on a weekly basis (Fig. 6.9, Table 6.4). In order to calculate pH, calibration curve slope and intercept values were interpolated to the fraction collection times collected. To establish whether a pH-buffering phase might have been present at the start of the DIW MFR experiment, an identical experiment was conducted on a separate subsample – albeit only for a month – focusing solely on pH measurements. This   164 experiment revealed increased pH conditions during the initial 2 days, followed by relatively constant pH values close to the incoming eluent pH (5.7± 0.1).  Effluent pH in the acidic MFR experiment (measured in the fraction as opposed to in-line) had a constant pH of 2.3± 0.1, consistent with the expected pH of the effluent (5 × 10!! 𝑚𝑜𝑙𝑒𝑠 𝐿!! HNO3). This indicates that sample dissolution did not affect pH beyond the precision of the measurements (± 0.1 pH units).  Day Slope Intercept Oct. 8th 2013 -0.0179 2.648 Oct. 15th 2013 -0.0164 3.189 Oct. 22nd 2013 -0.0169 2.897 Oct. 29th 2013 -0.0164 4.022 Nov. 5th 2013 -0.0165 3.346 Nov. 12th 2013 -0.1662 3.333 Nov. 19th 2013 -0.0167 3.247 Nov. 26th 2013 -0.0169 3.005 Dec. 4th 2013 -0.0166 3.070 Dec. 10th 2013 -0.0166 3.080 Dec. 17th 2013 -0.0169 2.933 Dec. 23rd 2013 -0.0170 2.884 Dec. 30th 2013 -0.0167 2.870 Jan. 3rd 2014 -0.0167 3.255 Table 6.4 pH probe calibration data DIW MFR experiment, where 𝑝𝐻 = (𝑠𝑙𝑜𝑝𝑒 × 𝑚𝑉) + 𝑖𝑛𝑡𝑒𝑟𝑐𝑒𝑝𝑡. The coefficient of determination (R2) was always greater than 0.99. Note maximum shift in calibration intercept occurred between Oct. 22nd and Oct. 29th 2013   165  Figure 6.9 Original DIW MFR experiment effluent pH (connected black filled circles) and follow-up DIW MFR experiment (connected red filled circles). Incoming eluent pH of 5.7 ± 0.1 is shown in between dashed lines. Error shown associated with original DIW MFR experiment represents error associated with pH calibration drift, which equals the difference in pH calculated by applying two consecutive pH calibration curves.  6.3.2.2 Elemental and sulphate concentrations 6.3.2.2.1 DIW experiment As in the trickle leach experiment, concentrations in the MFR effluent varied as a function of time following various patterns for different elements.  However, the patterns observed with the two experimental settings are different: • As for the trickle leach experiment, Ca, Mg, Sr, Cu and Mn show similar pattern (Fig. 6.10). However, after a brief decrease in concentration at the beginning of the experiment, concentrations increase sharply to reach a maximum value after about 15 DIW MFR experiment Follow-up DIW MFR experiment pH3.03.54.04.55.05.56.06.57.0Time (days)0 50 100 150 200  166 days and either gradually decreases (Ca, Sr, Cu) or stay roughly constant (Mg, Mn) to the end of the experiment (Fig. 6.10); • Unlike in the trickle leach experiment, Ba shows a similar pattern (Fig. 6.11) but K shows a different pattern, with a gradual decrease over the first 60 days (Fig. 6.11). Si decreases during the initial 10 days and stays roughly constant thereafter (Fig. 6.11); • Although Al concentrations are low and noisy, the pattern that emerges is a decrease during the first 10 days followed by a gradual increase (Fig. 6.12); • As in the trickle leach experiment, Na concentration decreases more than 100-fold but during the first 10 days and stays roughly constant thereafter (Fig. 6.12); • As in the trickle leached experiment, Fe concentration decreases 100-fold but during the first 10 days before staying roughly constant, with a clear step increase at t = 120 days.  albeit with sporadic increases (Fig. 6.13); • Sulphate concentration decreases rapidly during the first 10 days and remains within a factor of 2 later on (Fig. 6.13).   The gap in data between day 65 and day 82 is due to an accidental pump shutdown. The noise associated with some of the [Al, Cu, Mn, Sr, Fe] data can be explained by the low measured concentrations (sub-ppb range, reflected in the low measurement precision at low concentrations, see Table 6.3).    167  Figure 6.10 DIW MFR experiment effluent Ca, Mg (connected filled blue and red circles, respectively; scale on left y-axis) and Sr, Cu and Mn concentrations (connected filled green, orange and black circles, respectively; scale on right y-axis), expressed in moles L-1. Gap in data from day 65 – 82 due to accidental pump shutdown.    168  Figure 6.11 DIW MFR experiment effluent K, Si (connected filled purple and light blue circles, respectively; scale on left y-axis) and Ba concentrations (connected filled black circles; scale on right y-axis), expressed in moles L-1. Gap in data from day 65 – 82 due to accidental pump shutdown.      169  Figure 6.12 DIW MFR experiment effluent Na (connected filled black circles; scale on left y-axis) and Al concentrations (connected filled gray circles, respectively; scale on right y-axis), expressed in moles L-1. Gap in data from day 65 – 82 due to accidental pump shutdown.     170  Figure 6.13 DIW MFR experiment effluent sulphate (connected filled black circles; scale on left y-axis) and Fe concentrations (connected filled red circles; scale on right y-axis), expressed in moles L-1. Gap in data from day 65 – 82 due to accidental pump shutdown.    6.3.2.2.2 Acidic experiment In the MFR experiment at pH 2.3, most elements (Ca, Mg, Na, Si, Sr, Cu, Mn, Ba) concentration showed a similar patter, with an initial rapid 10-fold decrease, followed by a more gradual decrease to the end of the experiment (Figs. 6.14, 6.15).  Fe concentrations varied much less but also showed an initial decrease, while Al displayed a brief initial increase before following the same pattern as the other elements (Fig. 6.17). K showed the most striking divergence to the general pattern; an abrupt ten-fold drop, 15 days after the start of the experiment (Fig. 6.14). A calibration error can be ruled out as the samples immediately before   171 and after the drop in K concentration were part of the same sample run. Much higher concentrations in the acidic experiment compared to the DIW experiment were observed for Na, Ba, Ca, Mn, Sr, Mg, Fe and Al (Figs. 6.14, 6.15, 6.16). On the other hand, sulphate concentrations were similar in both the DIW and acidic MFR experiment (Fig. 6.17). The relatively high error associated with sulphate measurements after t = 150 days (relative standard deviation of 25.64 %, based on replicate standard measurements within a single analysis run) was likely due to a deteriorating IC column.    Figure 6.14 Acidic MFR experiment effluent Ca, Si and K concentrations (connected filled blue, light blue, and prurple circles, respectively), expressed in moles L-1.     172  Figure 6.15 Acidic MFR experiment effluent Mg, Na (connected filled red and black circles, respectively; scale on left y-axis) and Ba, Mn, Cu and Sr concentrations (connected filled green, blue, orange and light blue circles, respectively; scale on right y-axis), expressed in moles L-1.     173  Figure 6.16 Acidic MFR experiment effluent Al (connected filled gray circles; scale on left y-axis) and Fe concentrations (connected filled red circles; scale on right y-axis), expressed in moles L-1.   174  Figure 6.17 Acidic MFR experiment effluent sulphate concentrations (moles L-1). Error is based on relative standard deviation between replicate standard analyses within one day of analyses. Missing data can be attributed to sample loss during processing.   6.4 Discussion 6.4.1 pH 6.4.1.1 Trickle leach experiment Assuming that all the sulfate leached during trickle leach is due to jarosite dissolution and jarosite dissolution is the only source of acidity (equation 6.1, Appendix D), the expected eluent pH can be calculated based on 1.5 moles of hydrogen being produced for every sulphate released. This results in a lower expected pH by at least 1.5 to 3 pH units (Fig. 6.18) as compared to the observed pH. Hence, a buffering phase such as calcium carbonate would be required to Sulphate (x 10-5  moles L-1)2.53.03.54.04.55.05.56.0Time (days)0 50 100 150 200  175 consume the hydrogen ions produced. The amount of calcium carbonate required to buffer the pH (2.8 grams) represents a proportionally low amount (< 0.03 % of the sample), which falls below the quantitative x-ray diffraction detection limit. Although it is likely more soluble – unidentified – sulphates as opposed to jarosite will have contributed to the effluent, this simple exercise illustrates how relatively small amounts of a buffering phase are required to affect the overall effluent pH.  𝐾𝐹𝑒!(𝑆𝑂!)!(𝑂𝐻)! (!)  +  𝐻!𝑂(!") →  3𝐹𝑒(𝑂𝐻)! ! +  2𝑆𝑂! (!")!! + 3𝐻(!")! + 𝐾!(!")   (6.1)   Figure 6.18 Measured trickle leach effluent pH (connected black filled circles) and calculated pH based on sulphate effluent data (connected filled red circles). In this calculation, jarosite was assumed to be the sole source of hydrogen ions (as shown in equation 6.1). Measured Calculated pH1.52.02.53.03.54.04.55.05.56.06.5Time (day)0 200 400 600 800 1000  176 6.4.1.2 MFR experiments Due to the anticipated drifting of the in-line pH measurements, no meaningful pH could be recorded for the DIW MFR experiment for the first 4 months (an identical probe did not drift during method development). Using weekly calibrations during the final 81 days (Table 6.4), error in pH measurements were calculated based on the difference in pH based on two consecutive pH calibrations carried out before and after sample pH measurement. Although a likely overestimate as calibration curve slope and intercept were interpolated to the time of pH measurement, given the degree of calibration curve drift this cautionary calculation can be justified. Nevertheless, pH values recorded during the final 81 days (ranging between approximately 4 and 5.5) were of similar magnitude compared to the trickle leach experiment (ranging between approximately 4.5 and 6). The follow-up experiment confirms the absence of any significant pH neutralizing phase, as previously suggested by the trickle leach data (Figs. 6.1, 6.9).  In the acidic MFR experiment, constant pH values of 2.3 confirm the absence of any significant pH buffering mineral phase on the time scale of the residence time, as previously indicated using quantitative XRD (Table 6.2).   6.4.2 Metals (Al, Mg, Fe, K, Na, Ca, Ba), silicon and sulphate release 6.4.2.1 Trickle leach experiment The trickle leach experiment is unable to reach steady state release rates in terms of most metals including Cu, even after a time period of approximately 3 years (e.g. Fig. 6.5). As mentioned in section 6.1.2, the continuing evolution of the leachate does not allow for extrapolation of metal release rates into the future, thereby preventing the quantification of time until depletion. This   177 results in particular challenges for the elevated [Cu] concentrations, considering that Cu might be of environmental concern. When calculating the trickle leach Cu release rate using the final measurement (~ 0.001 mg day-1 kg-1) and the Cu content in the waste rock (0.01%), one obtains a period until depletion of close to 300 years. This depletion period will likely vary spatially within the waste rock pile due to heterogeneous distribution of the sulfate minerals and formation of secondary reaction products, and temporally as weathering conditions will vary over time. Regardless, as acid-generating minerals become more depleted over time – thereby adding less acidity to the system – one can expect Cu release rates to decrease due to incorporation into and adsorption onto secondary minerals.  Since Cu is of possible environmental concern, it is important to establish its mineral associations, so as to better predict its future release. Cu could either be associated with Jarosite (beaverite: 𝑃𝑏𝐶𝑢(𝐹𝑒,𝐴𝑙!)(𝑆𝑂!)!(𝑂𝐻)!) or adsorbed on iron oxides. When considering the saturation index (log Q/K, where Q = reaction quotient of the activities of the reaction products, K = equilibrium constant) for jarosite, excluding the initial few weeks, undersaturated conditions were found throughout the trickle leach experiment (Fig. 6.19). Although jarosite is dissolving, relatively few metals (< 0.01 – 5.82 %), S (3.01 %) are released into solution over the course of the trickle leach experiment (Table 6.5). It has previously been reported that as jarosite weathers to dilute waters with higher pH, it decomposes to ferrihydrite (𝐹𝑒(𝑂𝐻)!(a)), and as dehydration proceeds, goethite (𝐹𝑒𝑂(𝑂𝐻)) and hematite (𝐹𝑒!𝑂!) is formed (e.g. Nordstrom, 1982). As saturation index calculations show supersaturated conditions throughout the trickle leach experiment for minerals including hematite and goethite, and amorphous ferrihydrite was found to be relatively close to equilibrium throughout the trickle leach experiment, it is likely that amorphous ferrihydrite provided a solubility control for Fe(III) (Fig. 6.19). In addition,   178 iron(hydro)oxide amorphous secondary minerals, such as Fe(OH)3 (ferrihydrite), have been documented to form as jarosite dissolves under pH > 3.5 (Madden et al., 2012, Appendix D). The presence of these amorphous secondary minerals is difficult to confirm, as they cannot be identified using x-ray diffraction. It has been previously noted that iron(hydro)oxide minerals such as ferrihydrite serve as a potent sorbent and repository of contaminants (e.g. Hansel et al., 2003), facilitated by the high surface area / site density. This characteristic makes ferrihydrite an extremely reactive mineral phase (e.g. Hiemstra, 2013), playing a dominant role in the attenuation of trace metals.    Figure 6.19 Trickle leach experiment mineral saturation index (log Q/K) associated with jarosite minerals, ferrihydrite (Fe(OH)3). Except for the initial weeks, jarosite minerals remain undersaturated (log Q/K < 0) throughout the course of the trickle leach experiment. Amorphous ferrihydrite is relatively close to saturation Hematite Goethite Fe(OH)3(a) Jarosite-K Jarosite-Na Saturation Index (log Q/K)−10−5051015Time (days)0 200 400 600 800 1000  179 throughout the trickle leach experiment. In comparison, goethite and hematite are supersaturated (log Q/K > 0) throughout the experiment. Calculated using Geochemist’s WorkBench 10.0.4, using WATEQ4F thermodynamic database (Ball and Nordstrom, 1991).  Assuming all effluent sulphate was derived from jarosite, and based on 2 moles of sulphate being released during the dissolution of 1 mole of jarosite (Equation 6.1), the amount of dissolved jarosite can be calculated. Considering the 260 g of jarosite present in the trickle leach sample (2.6 % of 10,000 g, Table 6.2), the amount of dissolved jarosite can be calculated using the integrated S mass (Table 6.5). 21.71 g of jarosite dissolved (~8%) based on integrated S, which can be interpreted as an upper value, as more soluble sulphates (not detected by quantitative x-ray diffraction) are excluded from this calculation. Therefore, incomplete jarosite dissolution can be inferred. When comparing the amount of K to Fe dissolved over the course of the trickle leach experiment, 62 times more K was released compared to Fe (Table 6.4), counterintuitive from what one would expect during archetypal jarosite dissolution (releasing 3𝐹𝑒!!for every 𝐾!). This can be explained by the likely formation of secondary, amorphous Fe-bearing minerals such as amorphous ferrihydrite, more likely to reach saturation as opposed to goethite and / or hematite, which were found to be supersaturated throughout the trickle leach experiment.  When considering potassium to sulphate ratios as an indicator for jarosite dissolution, trickle leach effluent ratios lie close, however somewhat below stoichiometric jarosite dissolution (stoichiometric jarosite dissolution 𝐾: 𝑆𝑂!!! = 0.5, Fig. 6.20A). Considering the dissolution kinetics of all K-bearing minerals (muscovite, K-feldspar, K-jarosite), K-jarosite is by over 3 orders of magnitude the most soluble (Nagy, 1995; Blum and Stillings, 1995; Madden   180 et al., 2012). Hence, either the presence of an unidentified sulphate phase lowered the 𝐾: 𝑆𝑂!!! ratio below 0.5, or other cations substituted in the A, B sites (section 6.1.1, e.g. Welch et al., 2007). Regardless of the assumptions made, the relatively low amount of leached jarosite re-emphasizes the limited degree to which the jarosite dissolves over 3 years under undersaturated conditions, adding evidence to the slow dissolution rates previously associated with this mineral (e.g. Madden et al., 2012). The initially rapid decrease in sulphate concentrations during the initial 150 days can likely be attributed to flushing of resident pore water and the dissolution of a more soluble form of sulphate. Possible candidates include gypsum (CaSO4), barite (BaSO4) and mirabilite (Na2SO4). The integrated amounts of Ba2+ and Ca2+ released during the initial 150 days are not sufficient by an order of magnitude to account for the amount of sulphate released. However, when integrating the amount of Na2+ released during the initial 150 days (0.031 moles), the expected amount of associated sulphate (0.015 moles, as 2 moles of Na are released for every sulphate released during mirabilite dissolution) may represent some (approximately 17%) of the sulphate dissolved as part of the trickle leach experiment.   𝐾𝐹𝑒!(𝑆𝑂!) !(𝑂𝐻)! (!)  +  𝐻!𝑂(!") →  3𝐹𝑒(𝑂𝐻)! ! +  2𝑆𝑂! (!")!! + 𝐾(!")! + 3𝐻(!")!    (6.1)  The pattern of Cu leaching during the trickle leach experiment follows that of K, Ca, Mg and Sr, suggesting a possible association with jarosite whose dissolution would generate H+ releasing alkaline earth cations. In fact, the K:Cu molar ratio stabilizes after 200 days, possibly indicating Cu substitution in the jarosite (Fig. 6.20B). Assuming this K:Cu molar ratio (approximately 21) accurately reflects Cu substitution in the jarosite, the expected amount of K and sulphate    181 associated with jarosite can be calculated using the integrated amount of Cu leached (Table 6.5). The expected amount of associated sulphate based on this calculation (0.0234 moles of sulphate, or 7.82 × 10!! moles of S) is less than the measured amount, but accounts for approximately 27 % of the observed sulphate release (Table 6.5).  Using the latter 2 calculations, we can account for approximately half of the observed sulphate release, with the sources for the remaining soluble sulphate to be determined.   Total (moles) Leached  (moles) Leached (%) Al 27.72 4.77 × 10!! ≪ 0.01 Ba 0.12 2.67 × 10!! 0.02 Ca 1.42 2.05 × 10!! 1.44 Cu 0.02 5.47 × 10!! 3.20 Fe 3.09 1.43 × 10!! ≪ 0.01 K 9.00 8.88 × 10!! 0.01 Mg 1.39 6.90 × 10!! 0.49 Mn 0.01 3.87 × 10!! 5.82 Na 2.30 3.87 × 10!! 1.68 S 0.96 2.90 × 10!! 3.01 Sr 0.01 9.88 × 10!! 0.63 Table 6.5 Integrated moles in four-acid digested sample (first column), trickle leach experiment (second column) and % leached during trickle leach experiment (third column).     182  Figure 6.20 A: 𝐾: 𝑆𝑂!!!molar ratio associated with the trickle leach experiment. Stoichiometric jarosite dissolution (𝐾: 𝑆𝑂!!! = 0.5) is indicated as a dashed black line; B: trickle leach experiment effluent K:Cu molar ratio.    183 6.4.2.2 Comparison between DIW trickle leach and MFR experiments Meaningful comparison between both experimental designs is facilitated when considering the flow rates as well as the volume of rock through which the eluent passes. Although the flow rates are quite similar in both experimental designs (approximately 43 mL day-1 in the trickle leach experiment versus approximately 72 mL day-1 in the MFR experiment), the volume of rock through which the eluent passes is vastly different (10 kg in the trickle leach experiment versus 5×10!! kg in the MFR experiment). Nevertheless, the total amount of Cu leached in the DIW MFR experiment is of similar magnitude (2.88% of Cu dissolved, Table 6.6) compared to the trickle leach experiment (3.2%, of Cu dissolved, Table 6.5). Hence, the DIW MFR experiment leaches a similar fraction of the total Cu in a shorter period of time (approximately 6 months) as compared to the trickle leach experiment (approximately 3 years). This reflects the 6 times higher Cu release rate using the MFR experimental design (Figs. 6.21A, 6.21B), which are obtained by flowing orders of magnitude more pore volumes of eluent through the sample as compared to the trickle leach experiment (Fig. 6.21C).  During both the trickle leach, as well as the DIW MFR experiments, relatively little (<< 0.01 and 0.01%, respectively) Fe dissolved. Similar to the trickle leach experiment, Fe-bearing minerals such as goethite and hematite were supersaturated throughout the experiment (Fig. 6.22). Amorphous ferrihydrite was found to be more undersaturated during the DIW MFR experiment (Fig. 6.22) in comparison to the trickle leach experiment, as is expected due to the smaller sample volume and shorter residence time. Hence, a mineral with a solubility between ferrihydrite and goethite is likely providing a Fe(III) solubility control.  When comparing the Cu release rate per unit mass of material subjected to dissolution, we find a similar temporal pattern in both set up but compressed by about 5-fold with the MFR (Figs.   184 6.21A, 6.21B), suggesting that MFR is not only reproducing, but also accelerating the trickle leach experiment.  0 50 100 150 200Time (days)0.4900.050.100.150.20DIW MFR experiment Trickle leach experiment 10-10.2Cu release (mg week-1  kg-1)00.0050.0100.0150.020Time (days)0 200 400 600 800 1000Cu release (mg week-1  kg-1)110-210-3# pore volume replacements101 102 103 104110-110-2Trickle leach experiment DIW MFR experiment ABC  185 Figure 6.21 Cu release rate (mg week-1 kg waste rock-1) in A: the DIW MFR experiment (initial, highest value not shown but indicated) plotted versus time; B: the trickle leach experiment (initial, highest value not shown but indicated) plotted versus time number and C: both experiments plotted versus number of pore volume replacements, shown on a logarithmic x, y scales. Note (1) close to order of magnitude higher Cu release rates associated with DIW MFR experiment, compared to the trickle leach experiment, both carried out using deionized water eluent; (2) order(s) of magnitude higher pore volume replacements in the MFR experimental design as compared to the trickle leach experimental design. Pore volume replacement calculated using (𝑓𝑙𝑜𝑤 𝑟𝑎𝑡𝑒 𝑄 (𝑣𝑜𝑙𝑢𝑚𝑒 𝑝𝑒𝑟 𝑢𝑛𝑖𝑡 𝑡𝑖𝑚𝑒) × 𝑡 𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 𝑡𝑖𝑚𝑒 ) 𝑠𝑎𝑚𝑝𝑙𝑒 𝑣𝑜𝑙𝑢𝑚𝑒  × 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦.    Figure 6.22 DIW MFR experiment mineral saturation index (log Q/K) associated with hematite, goethite, jarosite minerals as well as ferrihydrite (Fe(OH)3). Saturation indices during the initial 125 days could not be calculated due to absence of pH data. Jarosite minerals remain undersaturated (log Q/K < 0) throughout experiment. Similarly, ferrihydrite is quite undersaturated during the final 81 days of the experiment. In contrast, hematite and goethite are Hematite Goethite Jarosite-K Jarosite-Na Fe(OH)3(a) Saturation Index (log Q/K)−25−20−15−10−5051015Time (days)120 125 130 135 140 145 150 155 160 165 170 175 180 185 190 195  186 supersaturated throughout the experiment. Calculated using Geochemist’s WorkBench 10.0.4, using WATEQ4F thermodynamic database (Ball and Nordstrom, 1991)   Table 6.6 Total moles, calculated using four-acid digestion data (first column), DIW MFR experiment (second column), % leached during DIW MFR experiment (third column), acidic MFR experiment (fourth column) and % leached during acidic MFR experiment (fifth column).   6.4.2.3 Comparison between DIW, acidic MFR experiments Assuming jarosite dissolves at similar rates at pH 2.3 and 5.7, as previously reported by Madden et al. (2012, log 𝑟 ≈  −8.75 𝑚𝑜𝑙𝑒𝑠 𝑚!! 𝑠!!), mainly the stability of other minerals, including secondary minerals present in the sample will be affected by the difference in eluent acidity between both MFR experiments.   Total (moles) LeachedDIW  (moles) LeachedDIW (%) Leachedacid (moles) Leachedacid  (%) Al 1.39 × 10!! 8.77 × 10!! 0.01 8.84 × 10!! 0.64 Ba 6.15 × 10!! 1.69 × 10!! 2.75 6.86 × 10!! 11.16 Ca 7.11 × 10!! 2.50 × 10!! 0.35 2.01 × 10!! 2.83 Cu 8.52 × 10!! 2.46 × 10!! 2.88 4.34 × 10!! 50.95 Fe 1.55 × 10!! 9.47 × 10!! 0.01 1.86 × 10!! 11.99 K 4.50 × 10!! 3.99 × 10!! 0.01 8.90 × 10!! 1.98 Mg 6.99 × 10!! 6.79 × 10!! 0.97 1.13 × 10!! 16.11 Mn 3.32 × 10!! 2.79 × 10!! 8.39 1.10 × 10!! 33.00 Na 1.15 × 10!! 1.68 × 10!! 1.46 1.04 × 10!! 9.00 S 4.81 × 10!! 1.81 × 10!! 37.67 1.58 × 10!! 32.89 Sr 7.86 × 10!! 1.29 × 10!! 1.64 1.05 × 10!! 13.39   187 As elevated effluent Cu concentrations were measured in the trickle leach experiment (section 6.1.2), it is of interest to identify the host mineral phase. If the bulk of the Cu originated from an isostructural mineral related to jarosite as opposed to archetypal jarosite (e.g. beaverite: 𝑃𝑏𝐶𝑢(𝐹𝑒,𝐴𝑙!)(𝑆𝑂!)!(𝑂𝐻)!), effluent Cu concentrations would be expected to dissolve following a similar pattern as archetypal jarosite, assuming this mineral is characterized by similar dissolution kinetics as archetypal jarosite. However, depending on MFR eluent conditions, reaction products may be adsorbed onto secondary minerals (as believed to occur under DIW conditions, see sections 6.4.2.1 and 6.4.2.2 above) or stay in solution (as expected under acidic eluent conditions). When comparing DIW and acidic eluent MFR effluent trace element concentrations, order of magnitude higher Cu concentrations were recorded in the acidic MFR experiment compared to the DIW MFR experiment (Figs. 6.10, 6.15). Also, over three orders of magnitude more Fe was mobilized in the acidic MFR experiment compared to the DIW MFR experiment (Figs. 6.13, 6.16), dissolving close to 12 % of the total sample Fe content over the course of the acidic MFR experiment (Table 6.6). In parallel, over half (~51%) of the total sample Cu content was dissolved (Table 6.6), resulting in a sample residue with approximately 53 ppm Cu, which is close to crustal abundance (e.g. Mason, 1952).  In terms of saturation indices, both jarosite, as well as amorphous ferrihydrite were undersaturated in the acidic MFR experiment (Fig. 6.23), suggesting that either mineral phase may have hosted Cu. However, quantitative x-ray diffraction analysis carried out on the sample residue of the acidic experiment indicates the presence of jarosite (2.0%, Appendix D Table D.1), suggesting the bulk of Cu must have be absorbed onto, or co-precipitated with some form of amorphous iron(hydro)oxide secondary mineral, such as Fe(OH)3 (ferrihydrite). In fact, Fe oxide minerals were identified in the original sample (section 6.1.1) and a considerable amount   188 of Fe was leached under acidic conditions (Table 6.6).  Further evidence dismissing jarosite as the main host for Cu is provided by the effluent 𝐾: 𝑆𝑂!!! stoichiometry (Fig. 6.24), which lies either above (due to significant K release during the initial 15 days) or below stoichiometric jarosite dissolution (0.5). Finally, when considering the Cu/Fe molar ratio, a relatively high Cu, Fe mineral phase initially dissolves over the course of approximately 50 days, followed by a lower Fe, and restricted Cu concentration range phase for the remainder of the experiment (Fig. 6.25). The amount of Cu dissolved over the initial 50 days represents over 75% of the total leached Cu over the duration of the entire acidic MFR experiment, indicating the relative importance of this initial, relatively high Cu-bearing mineral phase.    Figure 6.23 Acidic MFR experiment mineral saturation index (log Q/K) associated with jarosite minerals, ferrihydrite (Fe(OH)3). Note gaps in jarosite data at 60, 115 and 140 – 155 days due to absence of sulphate data. Hematite Goethite Jarosite-K Jarosite-Na Fe(OH)3(a) Saturation Index (log Q/K)−20−15−10−50510Time (days)0 50 100 150 200  189 Jarosite minerals remain undersaturated (log Q/K < 0) throughout the course of the experiment. Ferrihydrite is undersaturated to a greater degree as compared to the DIW MFR experiment (Fig. 6.19). Although hematite and goethite remain supersaturated throughout the experiment, they are to a smaller degree compared to the DIW MFR experiment. Calculated using Geochemist’s WorkBench 10.0.4, using WATEQ4F thermodynamic database (Ball and Nordstrom, 1991).   Figure 6.24 𝐾: 𝑆𝑂!!!molar ratio associated with the acidic MFR (connected red filled circles) and DIW MFR (connected blue filled circles) effluent. Stoichiometric jarosite dissolution (𝐾: 𝑆𝑂!!! = 0.5) is indicated as a dashed black line. acidic MFR experiment DIW MFR experiment K : SO2- 4 molar ratio00.51.01.52.02.53.03.54.04.55.05.56.0Time (days)0 50 100 150 200  190  Figure 6.25 Acidic MFR experiment effluent Cu: Fe molar ratio.  6.4.2.4 MFR sulphate release, comparison to trickle leach experiments The total amount of sulphate leached for the duration of the MFR based experiments design is considerably higher compared to the trickle leach experiment (approximately 38 and 33% in the DIW and acidic MFR experiments respectively, vs. only 3% in the trickle leach experiment, Table 6.6), regardless of eluent pH. The sulphate release rates achieved using the MFR experimental design are order(s) of magnitude higher as compared to the trickle leach experiment (Fig. 6.26A), which is likely – and similar to the Cu behavior – related to the high number of pore volume replacements (Fig. 6.26B). Assuming all sulphate stems from jarosite dissolution and considering the 0.13 g of jarosite present in the MFR sample (2.6 % of 5 g, Table 6.2), the amount of dissolved jarosite Cu : Fe (molar ratio)00.10.20.30.40.50.60.7Time (days)0 50 100 150 200  191 amounts to 0.12 – 0.13 g of dissolved jarosite under acidic and DIW experiments respectively. This simple calculation confirms that either all jarosite was dissolved, or other minerals containing sulphate are present, in line with previous trickle leach data interpretation. Furthermore, quantitative XRD on the sample residues confirm the presence of jarosite (2.2 % in the DIW MFR experiment and 2.0 % in the acidic MFR experiment, Appendix D Table D.1) and stoichiometric 𝐾: 𝑆𝑂!!! ratios do not indicate jarosite dissolution (Fig. 6.24, section 6.4.2.3) compared to stoichiometric jarosite dissolution. Therefore, the dissolution of an unidentified sulphate mineral phase must be responsible for the bulk of sulphate released.       192  Figure 6.26 Sulphate release rate (mg week-1 kg waste rock-1) plotted versus A: time (days) and B: number of pore volume replacements, shown on a logarithmic scale. Note orders of magnitude higher sulphate release rates associated with MFR experiments, compared to the trickle leach experiment.    193 6.5 Environmental implications Of interest to this study is identifying the source material responsible for the release of Cu. The acidic MFR experiment has revealed that the bulk of Cu is likely associated with a relatively high Cu, Fe mineral phase. An initially high effluent Cu concentration, followed by a graduate decline indicates it is likely the Cu is absorbed onto amorphous iron(hydro)oxide secondary mineral coatings. We speculate that the remainder of Cu is likely contained in a less soluble secondary mineral phase, possibly jarosite. Based on the combined findings associated from the trickle leach and MFR experiments, we can envisage waste rock kinetic drainage conditions. Unless acidic conditions are generated, relatively few metals are expected to dissolve under circumneutral groundwater conditions (assuming the absence of ARD producing rocks). However, under more acidic groundwater conditions, metals such as Cu may be mobilized relatively quickly, as demonstrated by the acidic MFR experiment.  6.6 Key benefits associated with MFR and future work • A benefit associated with the MFR experimental design used in this study is the ability to relatively quickly (here, approximately 6 months) leach a similar % of trace metals under DIW eluent conditions compared to the traditional trickle leach experimental design. A higher number of pore volume replacements (as illustrated in Fig. 6.21) likely play an important role.  • Using an acidic MFR experiment, it has been possible to dissolve approximately half of total amount of Cu, down to a sample concentration close to crustal abundance. Based on the presence of jarosite based on sample residue analysis, and considering the non-  194 stoichiometric jarosite dissolution, this finding suggests the importance of Cu absorption onto amorphous iron(hydro)oxide secondary mineral coatings. • The ability to carry out a series of MFR experiments simultaneously (e.g. covering distinct eluent pH conditions as demonstrated in this study) allows the user to exploit mineral dissolution characteristics in order to identify the relative contribution of dissolving mineral phases to the overall effluent composition.  This type of experimental design provides valuable complementary information to conventional kinetic tests, where mineral saturations may be reached. Future work relevant to this particular study might include leaching a waste rock sample under even lower or higher eluent pH conditions, thereby increasing the jarosite dissolution rates even further. In a broader sense, future work might also involve introducing pH-buffering minerals such as lime (CaO), commonly used as a neutralizing agent in ARD prone waste rock.  • Although release rates obtained during a MFR experiment are not representative of typical weathering conditions, the ability to carry out MFR experiments relatively quickly allows for studying parameters related to the scaling conundrum. MFR experiments can be designed to hold wide a range of waste rock volume, and the characteristics of eluent introduction (flow rate, composition) can be adjusted. This ability to vary a wide range of parameters under controlled laboratory conditions may thus provide valuable insights. • Finally, if mine waste material is transferred to a mixed flow reactor as done in this study, leaching rates are measured in a high liquid : waste rock ratio, which is applicable to e.g. a tailings pond breach. This type of measurement will inform mining companies of potential impacts and water treatment requirements.     195 Chapter 7: Conclusions and future outlook 7.1 Synthesis of thesis accomplishments The overarching aim of this thesis was to evaluate the potential of FT-TRA for several applications ranging from the study of mineral dissolution kinetics, to the analysis of microfossils for paleoceanographic and paleolimnologic applications, and to the prediction of metal leaching from mine waste.   The questions addressed in each chapter were:   I. Chapter 3: can FT-TRA measure mineral dissolution rates and provide accurate estimates of dissolution rate parameters, and if so, has the experimental design advantages over or complementarity with other methods currently used? II. Chapter 4: can FT-TRA be used to identify transport-limitations during mineral dissolution and combined with pore-scale modeling to document the effect of transport limitation on mineral dissolution? III. Chapter 5: with the insights accrued from chapters 3 and 4, can the premise on which FT-TRA data analysis has been based in paleoceanographic studies be substantiated? In particular, can the FT-TRA technique – using the current configuration – be used to distinguish between distinct biogenic carbonate phases? IV. Chapter 6:  can FT-TRA be adapted to provide additional mechanistic and/or quantitative information complementary to conventional methods used to predict the leaching rates of toxic elements from mine wastes?    196 The results obtained indicate: I. Chapter 3: FT-TRA has several advantages compared to the mixed-flow reactors conventionally used to measure mineral dissolution. The mineral dissolution regime (surface versus transport-controlled) can be readily established prior to conducting dissolution experiments, which is important if the goal of the experiment is to determine dissolution rate parameters. Because of the small volume (25 µL) of the flow-through reactor, the time to reach steady state concentration in the effluent is shorter and is limited by the intrinsic properties of the mineral, instead of the residence time of the effluent in the reactor. For minerals reaching steady-state dissolution rapidly, dissolution rate parameters can be established in a few hours, allowing for replications, statistical analysis of the results, and investigation of the underlying causes of variability. The experimental set-up can be adapted for minerals that require longer periods of time to reach steady-state dissolution. In addition, FT-TRA lends itself particularly well for detailed monitoring of dissolution rates and dissolution stoichiometry under transient conditions. The results presented here indicate that FT-TRA with online ICP-MS analysis could be a useful and multifaceted addition to the toolbox available to study mineral dissolution.  II. Chapter 4: FT-TRA can be used to empirically determine the rate-limiting step during mineral dissolution. This utility has been demonstrated using calcite and forsterite, two well-studied minerals with very different dissolution properties. A proportional increase in steady-state effluent [Mg, Si] concentrations with increasing flow-through cell eluent residence times confirms a surface-controlled dissolution regime for forsterite at pH 2.3. In contrast, dependence between flow rates and dissolution rates for calcite at pH 2.3 – 4   197 indicates transport-controlled conditions. In the latter case, the formation of a concentration gradient across a diffusive boundary layer (DBL) is confirmed using a pore-scale model, which reproduces the experimental effluent [Ca] concentrations. Hence, FT-TRA and pore-scale modeling may be combined to determine dissolution rate parameters from mineral dissolution rate measurements conducted under transport-controlled conditions.  III. Chapter 5: When dissolving a mixture of minerals with known differences in solubility (calcite, aragonite), the effluent composition can be calculated by taking into account the proportion of the minerals in the mixture. Hence, using the current configuration, FT-TRA does not sort mineral phases according to susceptibility to dissolution, thereby challenging the existing premise on which paleoceanographic data interpretation has been based. Instead, mineral dissolution within the flow-through cell is dictated by: (1) the intrinsic rate parameters associated with each mineral and exposed surface area (Chapter 3) and (2) the degree of transport-controlled dissolution (Chapter 4).  IV. Chapter 6: By acquiring a 50 mL mixed flow reactor (MFR) and a fraction collector, two 6-month leaching experiments (under DIW, acidic eluent conditions) were carried out on a waste rock sample, another subsample of which had undergone conventional trickle leach kinetic testing. The waste rock contained jarosite, an acid-producing mineral and trickle leach effluent Cu concentrations were of environmental concern. Whereas both experiments carried out using deionized water (DIW MFR and trickle leach experiment) leached very little Cu, the acidic MFR experiment was able to dissolve approximately half of all copper contained within the waste rock subsample. High initial Cu/Fe effluent ratios in the acidic MFR experiment, combined with the presence of jarosite in the acidic   198 MFR residue using x-ray diffraction, and non-stoichiometric dissolution with respect to jarosite suggests the bulk of the Cu to be absorbed onto amorphous iron(hydro)oxide secondary mineral coatings.   Overall, the applications of the FT-TRA technique explored in this thesis provide a solid foundation from which FT-TRA can be developed further in each research area touched upon in this thesis (mineral dissolution kinetics studies, paleoceanographic studies, as well as environmental impact studies).   7.2 Future research directions 7.2.1 Mineral dissolution kinetics studies I identify the following promising areas for future research: I. Use FT-TRA to rapidly build up a database of dissolution parameters in a systematic manner to gain insight on the cause for their variability; II. Since this variability is likely linked to variability in reactive surface sites, combine FT-TRA with AFM as explained below; III. Combine FT-TRA and pore scale modeling to determine dissolution rate parameters from dissolution experiment conducted under transport-controlled conditions. This is particularly important for very soluble minerals (e.g. carbonates under acidic conditions); IV. Use high resolution afforded by FT-TRA to study in greater detail the factors leading to surface passivation and the formation of the surface leached layer. In addition, study the factors that control the rate of exfoliation.    199 The flow-cell used in this thesis can be readily replaced by a flow-through cell compatible with AFM (e.g. Duckworth and Martin, 2003). This flow-through cell could be used to experiment with different mineral cleavage surfaces to take into account structural anisotropy. It could also be modified to produce hydrodynamic conditions that can be modeled on the pore scale to confirm whether dissolution occurs under a surface-controlled dissolution regime (e.g. Compton and Unwin, 1990). Prior to starting the FT-TRA dissolution experiment, the cleaved surface could be examined by AFM. To reach the quantification limits of ICP-MS, the surface subjected to dissolution must be larger than the typical field of view of the AFM. Therefore, multiple areas of the mineral surface will have to be examined to estimate statistically the distribution and nature of the nanoscale surface features. As dissolution with the FT-TRA module proceeds, a representative area of the mineral surface could be monitored to simultaneously document changes in nano-topography (taking into account the transport time of the bulk solution from the dissolution cell to the plasma of the mass-spectrometer). Once steady state dissolution has been achieved, the mineral surface could be examined once again to statistically determine the distribution of reactive sites after dissolution has reached steady state. The multiple data sets produced by this approach could help assessing whether the evolution of reactive surface site during mineral dissolution can be predicted, and whether these changes can be detected with the high temporal resolution and sensitivity afforded by FT-TRA. If so, the data would yield intrinsic dissolution rate parameters normalized to active site distribution, which would now be an intrinsic property of the minerals, and could possibly solve the normalization conundrum.   Assuming FT-TRA is adopted to measure mineral dissolution rates, the use of FT-TRA can be adopted as an empirical tool to confirm the presence of a surface-controlled dissolution regime prior to quantifying dissolution rates. This will limit laboratories from establishing   200 potentially different degrees in transport-controlled mineral dissolution due to differences in experimental design. In turn, this may contribute towards reducing the up to order of magnitude(s) differences in mineral dissolution rates observed (e.g. Rimstidt et al., 2012).  In addition, the combined pore-scale modeling, FT-TRA approach is providing new insights into the effect of transport-limitations on mineral dissolution. The use of an off-the-shelf syringe filter as a flow-through reactor can be considered a first step. A purpose-built flow-through reactor that can be accurately constrained in a pore-scale modeling domain (possibly similar to Compton and Unwin, 1990) represents a logical next step. In parallel, the measurement of additional effluent parameters such as pH will aid in further constraining pore scale modeling results, thereby providing additional means to calibrate dissolution rate parameters. Advances in pH electrode technology, allowing for the measurement of pH within the flow-cell close to the dissolving mineral may be explored (e.g. Batchelor-McAuley, 2013). Then, using the obtained FT-TRA (possibly AFM) experimental data, pore scale model parameter estimation, may allow for the successful extraction of inherent dissolution rate equation parameters (assuming knowledge of an applicable dissolution rate equation). This approach may prove particularly powerful when measuring dissolution rates associated with soluble minerals. From a different perspective, when mineral dissolution rate parameters are well-established, FT-TRA data can be used to calibrate pore scale models.   7.2.2 Paleoceanographic studies While it is clear that the premise on which FT-TRA has been applied to the analysis of microfossil is not supported by the results obtained in this study, there might still be a way forward. In order to extract ontogenetic, gametogenic calcite molar ratios (such as Mg/Ca),   201 different proportions of biogenic calcite need to dissolve over time. Unless both biogenic phases are present in equal amounts and dissolve at equal rates, this is automatically achieved when a sample consisting of different biogenic phases dissolves. Using FT-TRA effluent concentration data, a series of non-linear equations (as outlined in section 5.4.4) can be established, which will result in an over-determined system. Theoretically, solving this over-determined system of non-linear equation will extract the ontogenetic, gametogenic calcite molar ratios.  However, this approach hinges on the assumption that both biogenic calcite phases are homogeneous with respect to the variable being measured (e.g. Mg/Ca). Recent high spatial resolution elemental mapping has revealed the heterogeneous distribution of Mg/Ca, challenging our suggested way forward (e.g. Kunioka et al., 2006; Eggins et al., 2003; Hathorne et al., 2003; Sadekov et al., 2005). Therefore, the future of FT-TRA in paleoceanography is not very promising, especially considering the alternative microanalytical techniques capable of high spatial resolution (including nanoSIMS, LA-ICP-MS, electron microprobe mapping).  7.2.3 Environmental impact studies FT-TRA, when used in its MFR configuration, provides advantages and complementary information to conventional kinetic tests. It generates data faster and different dissolution conditions can be run (potentially in parallel) to provide additional insights into the source and rate of release of specific metals, which may be of environmental concern. The main advantage is the adaptability of the method to specific problems. In the present study conducted here, combining data from trickle leach and MFR experiments indicates that Cu, which was the main concern in this particular case, is partially associated with jarosite and partially adsorbed on iron oxides. At slightly acidic pH, Cu-associated jarosite will be released slowly over centuries, while   202 Cu adsorbed on Fe oxides remains immobile. On the other hand, if pH conditions drop significantly, a large amount of Cu can be released quickly. While this exploratory study used a relatively large sample (5 g), it is also possible to use multiple samples of smaller size (10’s of mg), which could be run rapidly in the FT-TRA configuration used to study mineral dissolution, to acquire ‘intrinsic’ dissolution rate parameters on the various minerals dominating the waste rocks, which could then be used to model metal release.   203 Bibliography  Alkattan, M., Oelkers, E.H., Dandurand, J.-L., Schott, J., 1998. An experimental study of calcite and limestone dissolution rates as a function of pH from 1 to 3 and temperature from 25 to 80ºC. Chemical Geology 151, p. 199– 214. Anand, P., Elderfield, H., Conte, M.H., 2003. Calibration of Mg/Ca thermometry in planktonic foraminifera from a sediment trap time series. Paleoceanography 18, p. 28-31.  Appelo, C.A.J., Postma, D., 2005. Geochemistry, groundwater and pollution. A.A. Balkema Publishers, Leiden, The Netherlands, p. 649  Arnesen, R.T., Iversen, E.R., 1997. The Lokken project – flooding a sulphide ore mine. Proceedings, Fourth International Conference on Acid Rock Drainage. Vancouver, BC 3: 1093-1107. Arvidson, R.S., Ertan, I.E., Amonette, J.E., Lüttge, A., 2003. Variation in calcite dissolution rates: A fundamental problem? Geochimica et Cosmochimica Acta 67, p. 1623–1634. Bachu, S., 2000. Sequestration of CO2 in geological media: criteria and approach for site selection in response to climate change. Energy Conversion and Management 41, p. 953-970. Ball, J.W. and Nordstrom, D.K., 1991. User's manual for WATEQ4F, with revised thermodynamic data base and test cases for calculating speciation of major, trace, and redox elements in natural waters. U.S. Geological Survey Open-File Report 91-183. Barker, S., Greaves, M., Elderfield, H., 2003. A study of cleaning procedures used for foraminiferal Mg/Ca paleothermometry. Geochemistry, Geophysics, Geosystems 4 (9), p. 1-20.   204 Batchelor-McAuley, C., Tschulik, K., Compton, R.G., 2013. Nanotoxicity - an electrochemist's perspective. Portugaliae Electrochimica Acta 31, p. 249-256.  Bé, A.W.H., Anderson, O.R., Faber Jr., W.W., Caron, D.A., 1983. Sequence of morphological and cytoplasmic changes during gametogenesis in the planktonic foraminifer Globigerinoides sacculifer (Brady). Micropaleontology 29 (3), p. 310-325.  Bé, A.W.H., Lott, L., 1964. Shell growth and structure of planktonic foraminifera. Science 145, p. 823-824.  Bé, A.W.H., Morse, J.W., Harrison, S.M., 1975. Progressive dissolution and ultrastructural breakdown of planktonic foraminifera, in: Sliter, W.V., Bé, A.W.H., Berger, W.H. (Eds.), Dissolution of Deep-Sea Carbonates. Cushman Foundation for Foraminiferal Research Special Publication 13. Allen Press, Inc., Lawrence, Kansas. Béarat, H., McKelvy, M.J., Chizmeshya, A.V.G., Gormley, D., Nunez, R., Carpenter, R.W., Squires, K., Wolf, G.H., 2006. Carbon sequestration via aqueous olivine mineral carbonation: role of passivating layer formation. Environmental Science and Technology 40, p. 4802-4808. Bentov, S., Erez, J., 2005. Novel observations on biomineralization processes in foraminifera and implications for Mg/Ca ratio in the shells. Geology 33, p. 841-844.  Benway, H.M., Haley, B.A., Klinkhammer, G.P., Mix, A.C., 2003. Adaptation of a flow-through leaching procedure for Mg/Ca paleothermometry. Geochemistry, Geophysics, Geosystems 4, p. 1-15. Berner, R.A., 1978. Rate control of mineral dissolution under earth surface conditions. American Journal of Science 278, p. 1235-1252.  Bigham, J.M., Nordstrom, D.K., 2000. Iron and aluminum hydroxysulfates from acid sulfate   205 waters, in: Alpers, C.N., Jambor, J.L., Nordstrom, D.K. (Eds.), Sulfate Minerals – Crystallography, Geochemistry and Environmental Significance. Reviews in Mineralogy & Geochemistry 40, p. 351 – 403.  Binnig, G., Quate, C.F., Ginzton, E.L., 1986. Atomic force microscope. Physical Review Letters 56, p. 930-933. Blum, A., and Lasaga, A., 1988. Role of surface speciation in the low-temperature dissolution of minerals. Nature 331, p. 431-433. Blum, A.E., Stillings, L.L., 1995. Feldspar dissolution kinetics, in: White, A.F., Brantley, S.L. (Eds.), Chemical Weathering Rates of Silicate Minerals. Reviews in Mineralogy and Geochemistry 31, p. 291-351. Brantley S.L., 2008. Kinetics of Mineral Dissolution, in: Brantley, S.L., Kubicki, J.D., White, A.F. (Eds.), Kinetics of Water-Rock Interaction. Springer Science, New York, p. 151-210.  Brantley, S.L. and Chen, Y., 1995. Chemical weathering rates of pyroxenes and amphiboles. Reviews in Mineralogy and Geochemistry 31, p. 119-172. Brodie, M.J., Broughton, L.M., Robertson, A.MacG., 1991. A conceptual rock classification system for waste management and a laboratory method for ARD prediction from rock piles. Proceedings of the Second International Conference on Abatement of Acidic Drainage, Montréal, Quebéc 3, p. 119-136.  Brown, S.J., Elderfield, H., 1996. Variations in Mg/Ca and Sr/Ca ratios of planktonic foraminifera caused by postdepositional dissolution: evidence of shallow Mg-dependent dissolution. Paleoceanography 11, p. 543-551.  Brunauer S., Emmett P.H., Teller E., 1938. Adsorption of Gases in Multimolecular Layers. Journal of the American Chemical Society 60, p. 309–319.    206 Busenberg, E., Plummer, L.N., 1986. A comparative study of the dissolution and crystal growth kinetics of calcite and aragonite, in: Mumpton, F.A. (Ed.), Studies in Diagenesis, U.S. Geological Survey Bulletin 1578, p. 139-168.  Cama, J., Ganor, J., 2006. The effects of organic acids on the dissolution of silicate minerals: A case study of oxalate catalysis of kaolinite dissolution. Geochimica et Cosmochimica Acta 70, p. 2191–2209.  Cemič, L., 2005. Thermodynamics in Mineral Sciences. Springer, Berlin Heidelberg, p. 399. Chen, Y. and Brantley, S.L., 2000. Dissolution of forsteritic olivine at 65°C and 2<pH<5. Chemical Geolology 165, p. 267–281.  Chou L., Garrels R., Wollast R., 1989. Comparative Study of the Kinetics and Mechanisms of Dissolution of Carbonate Minerals. Chemical Geology 78, p. 269–282. Chou, L. and Wollast, R., 1984. Study of the weathering of albite at room temperature and pressure with a fluidized bed reactor. Geochimica et Cosmochimica Acta 48, p. 2205–2217. Colombani J., 2008. Measurement of the pure dissolution rate constant of a mineral in water. Geochimica et Cosmochimica Acta 72, p. 5634–5640. Compton R.G., Unwin P.R., 1990. The Dissolution of Calcite in Aqueous Solution at pH < 4: Kinetics and Mechanism. Philosophical Transactions Royal Society A 330, p. 1–45. Covington, M.D., Gulley, J.D., Gabrovšek, F., 2015. Natural variations in calcite dissolution rates in streams: Controls, implications, and open questions. Geophysical Research Letters 42, p. 2836-2843.  Cubillas, P., Köhler, S., Prieto, M., Chaïrat, C., Oelkers, E.H, 2005. Experimental determination of the dissolution rates of calcite, aragonite, and bivalves. Chemical Geology 216, p. 59-77. Daval, D., Hellmann, R., Saldi, G.D., Wirth, R., Knauss, K.G., 2013. Linking nm-scale   207 measurements of the anisotropy of silicate surface reactivity to macroscopic dissolution rate laws: New insights based on diopside. Geochimica et Cosmochimica Acta 107, p. 121-134. De Baere, B., François, R., Mayer, K.U., 2015. Measuring mineral dissolution kinetics using on-line flow-through time resolved analysis (FT-TRA): an exploratory study using forsterite. Chemical Geology. 413, p. 107-118. De Baere, B., Molins, S., Mayer, K.U., François, R., 2015. Empirical determination of mineral dissolution regimes using flow-through time-resolved analysis (FT-TRA). Submitted.  De Giudici, G., 2002. Surface control vs. diffusion control during calcite dissolution: Dependence of step-edge velocity upon solution pH. American Mineralogist 87, p.  1279-1285.  Declercq, J., Bosc, O., Oelkers, E.H., 2013. Do organic ligands affect forsterite dissolution rates? Applied Geochemistry 39, p. 69–77. Doherty, J., 2010. Open-source model-independent parameter estimation and uncertainty analysis software - PEST. Watermark Numerical Computing. Duckworth, O.W., Martin, S.T., 2003. Connections between surface complexation and geometric models of mineral dissolution investigated for rhodochrosite. Geochimica et Cosmochimica Acta 67, p. 1787–1801. Eggins, S., De Deckker, P., Marshall, J., 2003. Mg/Ca variation in planktonic foraminifera tests: implications for reconstructing palaeo-seawater temperature and habitat migration. Earth and Planetary Science Letters 212, p. 291-306.  Eggleton, R.A., 1986. The Relation between Crystal Structure and Silicate Weathering Rates, in: Colman, S.M., Dethier, D.P., Rates of Chemical Weathering of Rocks and Minerals. Academic Press, Inc., Orlando, Florida, p. 21-40.    208 Egiebor, N.O., Oni, B., 2007. Acid rock drainage formation and treatment: a review. Asia-Pacific Journal of Chemical Engineering 2, p. 47-62.  Elderfield, H., Bertram, C.J., Erez, J., 1996. A biomineralization model for the incorporation of trace elements into foraminiferal calcium carbonate. Earth and Planetary Science Letters 142, p. 409-423.  Elderfield, H., Ganssen, G., 2000. Past temperature and delta (super 18) O of surface ocean waters inferred from foraminiferal Mg/Ca ratios. Nature 405, p. 442-419. Elderfield, H., Rickaby, R.E.M., 2000. Oceanic Cd/P ratio and nutrient utilization in the glacial Southern Ocean. Nature 405, p. 305-310.  Fischer, C., Kurganskaya, I., Schäfer, T., Lüttge, A., 2014. Variability of crystal surface reactivity: What do we know? Applied Geochemistry 43, p. 132-157. Frostad, S., 1999. Prediction of Acid Rock Drainage Red Mountain Project. MA Thesis, University of British Columbia.  Gautier, J.M., Oelkers, E.H., Schott, J., 2001. Are quartz dissolution rates proportional to B.E.T. surface areas? Geochimica et Cosmochimica Acta 65, p. 1059-1070.  Giammar, D.E., Bruant, R.G., Peters, C.A., 2005. Forsterite dissolution and magnesite precipitation at conditions relevant for deep saline aquifer storage and sequestration of carbon dioxide. Chemical Geology 217, p. 257-276. Grandstaff, D.E., 1986. The dissolution rate of forsterite olivine from Hawaiian beach sand, in: Rates of Chemical Weathering of Rocks and Minerals, Colmanand, S.M., Dethier, D.P. (Eds.), p. 41-59, New York. Haarman, T., 2012. The use of single foraminiferal shells for recording seasonal temperatures and water column stratification and its applicability to the fossil record. PhD thesis,   209 University of Bremen.  Haley B.A., Klinkhammer G.P., 2002. Development of a flow-through system for cleaning and dissolving foraminiferal tests. Chemical Geology 185, p. 51–69. Haley B.A., Klinkhammer G.P., Mix A.C., 2005. Revisiting the rare earth elements in foraminiferal tests. Earth and Planetary Science Letters 239, p. 79-97.  Haley B.A., Klinkhammer P., 2003. Complete separation of rare earth elements from small volume seawater samples by automated ion chromatography: method development and application to benthic flux. Marine Chemistry 82, p. 197-220.  Hamilton, P.C., Spero, H.J., Bijma, J., Lea, D.W., 2008. Geochemical investigation of gametogenic calcite addition in the planktonic foraminifera Orbulina universa. Marine Micropaleontology 68, p. 256-267.  Hansel, C.M., Benner, S.G., Neiss, J., Dohnalkova, A., Kukkadapu, R.K., Fendorf, S., 2003. Secondary mineralization pathways induced by dissimilatory iron reduction of ferrihydrite under advective flow. Geochimica Et Cosmochimica Acta 67, p. 2977-2992. Hathorne, E.C., Alard, O., James, R.H., Rogers, N.W., 2003. Determination of intratest variability of trace elements in foraminifera by laser ablation inductively coupled plasma-mass spectrometry. Geochemistry, Geophysics, Geosystems 4 (12), doi: 10.1029/2003GC000539. Hathorne, E.C., James, R.H., 2006. Temporal record of lithium in seawater: A tracer for silicate weathering? Earth and Planetary Sciences Letters 246, p. 393-406.  Hecht, A.D., Eslinger, E.V., Garmon, L.B., 1975. Experimental studies on the dissolution of planktonic foraminifera. Cushman Foundation for Foraminiferal Research, Special Publication 13, p. 56-69.   210 Hemleben, Ch., Spindler, M., 1983. Recent advances in research on living planktonic foraminifera. Utrecht Micropaleontological Bulletin 30, p. 141-170.  Hemleben, Ch., Spindler, M., Breitinger, I., Deuser, W.G., 1985. Field and laboratory studies on the ontogeny and ecology of some globorotaliid species from the Sargasso sea off Bermuda. Journal of Foraminiferal Research 15, p. 254-272.  Hiemstra, T., 2013. Surface and mineral structure of ferrihydrite. Geochimica Et Cosmochimica Acta 105, p. 316–325. doi:10.1016/j.gca.2012.12.002. Holdren, G.R., Speyer, P.M., 1986. Stoichiometry of Alkali Feldspar Dissolution at Room Temperature and Various pH Values, in: Colman, S.M., Dethier, D.P., Rates of Chemical Weathering of Rocks and Minerals. Academic Press, Inc., Orlando, Florida, p. 61-81. Hughes, I. and Hase, T., 2010. Measurements and their uncertainties: a practical guide to modern error analysis. Oxford University Press, Oxford. Inskeep, W.P. and Bloom, P.R., 1985. An evaluation of rate equations for calcite precipitation kinetics at less than 0.01 atm and pH greater than 8. Geochimica et Cosmochimica Acta 49, p. 2165–2180. Jarvis, K., Carpenter, R.W., Windman, T., Kim, Y., Nunez, R., Alawneh, F., 2009. Reaction mechanisms for enhancing mineral sequestration of CO2. Environment, Science, and Technology 43, p. 6314-6319.  Jeandel, C., Oelkers, E.H., 2015. The influence of terrigenous particulate material dissolution on ocean chemistry and global element cycles. Chemical Geology 394, p. 50-66.  Kim, Y., Nunez, R., Carpenter, R.W., Chizmeshya, A.V.G., McKelvy, M.J., 2005. The nanoscale mechanism for San Carlos olivine carbonation. Microscopy and  Microanalysis 11, p. 1530-1531.    211 Klinkhammer G.P., Haley B.A., Mix A.C., Benway H.M., Cheseby M., 2004. Evaluation of automated flow-through time-resolved analysis of foraminifera for Mg/Ca paleothermometry. Paleoceanography 19, p. 1-12. Klinkhammer, G.P., Mix, A.C., Haley, B.A., 2009, Increased dissolved terrestrial input to the coastal ocean during the last glaciation. Geochemistry, Geophysics, Geosystems 10 (3), p. 1-11. Kohlmeyer, U., Jantzen, E., Kuballa, J., Jakubik, S., 2003. Benefits of high resolution IC-ICP-MS for the routine analysis of inorganic and organic arsenic species in food products of marine and terrestrial origin. Analytical and Bioanalytical Chemistry 377, p. 6-13.  Kump, L.R., Brantley, S.L., Arthur, M.A., 2000. Chemical weathering, atmospheric CO2, and climate. Annual Review of Earth and Planetary Sciences 28, p. 611-667.  Kunioka, D., Shirai, K., Takahata, N., Sano, Y., Toyofuku, T., Ujiie, Y., 2006. Microdistribution of Mg/Ca, Sr/Ca, and Ba/Ca ratios in Pulleniatina obliquiloculata test by using a NanoSIMS: Implication for the vital effect mechanism. Geochemistry, Geophysics, Geosystems 7 (12), doi: 10.1029/2006GC001280. Lapakko, K., Berndt, M., 2003. Comparison of Acid Production From Pyrite and Jarosite. Proceedings of the 6th International Conference of Acid Rock Drainage, Cairns, Queensland, p. 461–467. Lasaga A.C., 1998. Kinetic Theory in the Earth Sciences. Princeton University Press, Princeton, New Jersey.  Lasaga, A., 1984. Chemical kinetics of water-rock interactions. Journal of Geophysical Research 89, p. 4009-4025.  Lea, D.W., Boyle, E.A., 1993. Determination of carbonate-bound barium in foraminifera and   212 corals by isotope dilution plasma-mass spectrometry. Chemical Geolology 103, p. 73-84.  Lea, D.W., Mashiotta, T.A., Spero, H.J., 1999. Controls on magnesium and strontium uptake in planktonic foraminifera determined by live culturing. Geochimica et Cosmochimica Acta 63, p. 2369-2379.  Levenberg, K., 1944. A method for the solution of certain non-linear problems in least squares. Quarterly of Applied Mathematics 2, p. 164–168. Levich, V. G., 1962. Physicochemical Hydrodynamics. Prentice-Hall, Englewood Cliffs, New Jersey. Liang Y., Baer D.R., 1997. Anisotropic dissolution at the CaCO3 – water interface. Surface Science 373, p. 275–287. Liang, Y., Baer, D.R., McCoy, J.M., Amonette, J.E., LaFemina, J.P., 1996. Interplay between step velocity and morphology during the dissolution of CaCO3 surface. Journal of Vacuum Science and Technology 14, p. 1368-1375.  Liu, Y., Olsen, A.A., Rimstidt, J.D., 2006. Mechanism for the dissolution of olivine series minerals in acidic solutions. American Mineralogist 91, p. 455-458.  Lohmann, G.P., 1995. A model for variation in the chemistry of planktonic foraminifera due to secondary calcification and selective dissolution. Paleoceanography 10, p. 445-457.  Longerich, H.P., Jackson, S.E., Günther, D., 1996. Laser Ablation Inductively Coupled Plasma Mass Spectrometric Transient Signal Data Acquisition and Analyte Concentration Calculation. Journal of Analytical Atomic Spectrometry 11, p. 899-904.  Luce, R.W., Bartlett, R.W., Parks, G.A., 1972. Dissolution kinetics of magnesium silicates. Geochimica et Cosmochimica Acta 36, p. 35-50. Lüttge, A., Winkler, U., Lasaga, A.C., 2003. Dolomite dissolution kinetics studied with vertical   213 scanning white light interferometry. Geochimica Et Cosmochimica Acta 67, 1099-1116.  Lüttge, A., 2004. Crystal dissolution kinetics studied by Vertical Scanning Interferometry (VSI): a brief review and outlook, in: Liu, X.Y., De Yoreo, J. (Eds.), Solid-Fluid interfacial structure to nanostructural engineering, Kluwer Academic Publishers, 209-247.  Lüttge, A., 2005. Etch pit coalescence, surface area, and overall mineral dissolution rates. American Mineralogist 90, p. 1776-1783. Lüttge, A., Arvidson, R.S., Fischer, C., 2013. A stochastic Treatment of Crystal Dissolution Kinetics. Elements 9, p. 183-188.  Mackenzie, F.T., Lerman, A., 2006. Mineralogy, Chemistry and Reaction Kinetics of the Major Carbonate Phases, in: Landman, N.H., Jones, D.S. (Eds.), Carbon in the Geobiosphere – Earth’s Outer Shell -, Springer, Dordrecht, p. 89-116.  Madden, M.E.E., Madden, A.S., Rimstidt, J.D., Zahrai, S., Kendall, M.R., Miller, M.A., 2012. Jarosite dissolution rates and nanoscale mineralogy. Geochimica Et Cosmochimica Acta 91, p. 306-321. Marquardt, D.W., 1963. An algorithm for least-squares estimation of nonlinear parameters. Journal of the Society for Industrial and Applied Mathematics 11, p. 431-441. Marti, O., Drake, B., Hansma, P.K., 1987. Atomic force microscopy of liquid-covered surfaces: atomic resolution images. Applied Physics Letters 51, p. 484-486.  Mason, B., 1952. Principles of geochemistry. John Wiley and Sons, New York. McCurdy E., Woods, G., 2004. The application of collision/reaction cell inductively coupled plasma mass spectrometry to multi-element analysis in variable sample matrices, using He as a non-reactive gas. Journal of Analytical Atomic Spectrometry 19, p. 607-615.  Mekik, F., François, R., 2006. Tracing deep sea calcite dissolution: Globorotalia menardii   214 fragmentation index and elemental ratios (Mg/Ca and Mg/Sr) in planktonic foraminifers. Paleoceanography 21, p. 1-12.  Molins S., Trebotich D., Steefel C.I., Shen C., 2012. An investigation of the effect of pore scale flow on average geochemical reaction rates using direct numerical simulation. Water Resources Research 48, p. 1–11.  Molins S., Trebotich D., Yang L., Ajo-Franklin J.B., Ligocki T.J., Shen C., Steefel C.I., 2014. Pore-Scale Controls on Calcite Dissolution Rates from Flow-through Laboratory and Numerical Experiments. Environmental Science and Technology 48, p. 7453–7460. Morse J.W. and Arvidson R.S., 2002. The dissolution kinetics of major sedimentary carbonate minerals. Earth-Science Reviews 58, p. 51–84. Morse, J.W., 1974. Dissolution kinetics of calcium carbonate in seawater III: A new method for the study of carbonate reaction kinetics. American Journal of Science 274, p. 97-107. Morse, J.W., Arvidson, R.S., Lüttge, A., 2007. Calcium Carbonate Formation and Dissolution. Chemical Reviews 107, p. 342–381. Murphy, W.M., Oelkers, E.H., Lichtner, P.C., 1989. Surface reaction versus diffusion control of mineral dissolution and growth rates in geochemical processes. Chemical Geology 78, p. 357-380.  Nagy, K.L., 1995. Dissolution and precipitation kinetics of sheet silicates, in: White, A.F., Brantley, S.L. (Eds.), Chemical Weathering Rates of Silicate Minerals. Reviews in Mineralogy 31, p. 173-233. Ni, Y., Foster, G.L., Bailey, T., Elliot, T., Schmidt, D.N., Pearson, P., Haley, B., Coath, C., 2007. A core top assessment of proxies for the ocean carbonate system in surface-dwelling foraminifers. Paleoceanography 22, doi:10.1029/2006PA001337.     215 Nordstrom, D.K., 1982. Aqueous pyrite oxidation and the consequent formation of secondary minerals. In: Acid Sulfate Weathering, Chapter 3. (Eds. Kittrick, J.A., Fanning, D.S., Hosner, L.R.), p. 37-56.  Nürnberg, D., Bijma, J., Hemleben, C., 1996. Assessing the reliability of magnesium in foraminiferal calcite as a proxy for water mass temperatures. Geochimica et Cosmochimica Acta 60, p. 803-814.  Oelkers, E.H., Schott, J., Devidal, J.L., 1994. The effect of aluminum, pH, and chemical affinity on the rates of aluminosilicate dissolution reactions. Geochimica et Cosmochimica Acta 58, p. 2011-2024.  Olsen, A.A. and Rimstidt, J.D., 2008. Oxalate-promoted forsterite dissolution at low pH. Geochimica et Cosmochimica Acta 72, p. 1758-1766. Olsen, A.A., 2007. Forsterite Dissolution Kinetics: applications and implications for chemical weathering. Ph.D. thesis, Virginia Polytechnic Institute and State University. Olsson, J., Bovet, N., Makovicky, E., Bechgaard, K., Balogh, Z., Stipp, S.L.S., 2012. Olivine reactivity with CO2 and H2O on a microscale: Implications for carbon sequestration. Geochimica et Cosmochimica Acta 77, p. 86-97. Parkhurst, D.L. and Appelo, C., 1999. User's Guide to PHREEQC (version 2) - A Computer Program for Speciation, Batch-Reaction, One-Dimensional Transport and Inverse Geochemical Calculations. U.S. Department of the Interior, U.S. Geological Survey, Denver, Colorado. Pleskov Y.V., Filinovskii V.Y., 1976. The rotating disc electrode. Consultants Bureau, New York.  Plummer L.N., Pakhurst D., Wigley T.M.L., 1979. Critical review of the kinetics of calcite   216 dissolution and precipitation, in: Jenne, E.A. (Ed.) Chemical Modeling in Aqueous Systems, American Chemical Society 93, p. 537–573. Plummer L.N., Wigley T., Parkhurst D.L., 1978. The kinetics of calcite dissolution in CO2-water systems at 5 to 60 degrees C and 0.0 to 1.0 ATM CO2. American Journal of Science 278, p. 179–216. Pokrovsky O.S., Golubev S.V., Schott J., 2005. Dissolution kinetics of calcite, dolomite and magnesite at 25 ºC and 0 to 50 atm pCO2. Chemical Geology 217, p. 239–255. Pokrovsky, O.S. and Schott, J., 2000a. Kinetics and mechanism of forsterite dissolution at 25°C and pH from 1 to 12. Geochimica et Cosmochimica Acta 64, p. 3313–3325. Pokrovsky, O.S. and Schott, J., 2000b. Forsterite surface composition in aqueous solutions: A combined potentiiometric, electrokinetic, and spectroscopic approach. Geochimica et Cosmochimica Acta 64, p. 3299-3312. Pokrovsky, O.S., Golubev, S.V., Schott, J., Castillo, A., 2009. Calcite, dolomite and magnesite dissolution kinetics in aqueous solutions at acid to circumneutral pH, 25 to 150ºC and 1 to 55 atm pCO2: New constraints on CO2 sequestration in sedimentary basins, Chemical Geology 265, p. 20-32. Price, W.A., 2009. Prediction Manual for Drainage Chemistry from Sulphidic Geologic Materials. B.C. Ministry of Energy and Mines, MEND Report 1.20.1, p. 579.  Price, W.A., Kwong, Y.T.J., 1997. Waste rock weathering, sampling and analysis: observations from the British Columbia ministry of employment and investment database. Proceedings of the 4th International Conference on Acid Mine Drainage. Vancouver, BC, p. 31-45.  Putnis, C.V. and Ruiz-Agudo, E., 2013. The Mineral–Water Interface: Where Minerals React with the Environment. Elements 9, p. 177–182.    217 Rickard D., Sjöberg E.L., 1983. Mixed control of calcite dissolution rates. American Journal of Science 283, p. 815–830. Rimstidt, J.D. and Dove, P.M., 1986. Mineral/solution reaction rates in a mixed flow reactor: Wollastonite hydrolysis. Geochimica et Cosmochimica Acta 50, p. 2509–2516.  Rimstidt, J.D., 2015. Diffusion control of quartz and forsterite dissolution rates. Applied Geochemistry, 61, p. 99-108.  Rimstidt, J.D., Brantley, S.L., Olsen, A.A., 2012. Systematic review of forsterite dissolution rate data. Geochimica et Cosmochimica Acta 99, p. 159-178. Rosenthal, Y., Lohmann, K.C., Sherrell, R.M., 2000. Incorporation and preservation of Mg in Globigerinoides sacculifer: Implications for reconstructing the temperature and 18O/16O of seawater. Paleoceanography 15, p. 135-145.  Rosenthal, Y., et al., 2004. Interlaboratory comparison study of Mg/Ca and Sr/Ca measurements in planktonic foraminifera for paleoceanographic research. Geochemistry, Geophysics, Geosystems 5 (4), p. 1-29. Rosso, J.J. and Rimstidt, J.D., 2000. A high resolution study of forsterite dissolution rates. Geochimica et Cosmochimica Acta 64, p. 797–811. Ruiz-Agudo, E., Putnis, C.V., 2012. Direct observations of mineral-fluid reactions using atomic force microscopy: the specific example of calcite. Mineralogical Magazine 76, p. 227-253.  Russell, A.D., Hönisch, B., Spero, H.J., Lea, D.W., 2004. Effects of seawater carbonate ion concentration and temperature on shell U, Mg, and Sr in cultured planktonic foraminifera. Geochimica et Cosmochimica Acta 68, p. 4347-4361.  Sadekov, Y.S., Eggins, S.M., De Deckker, P., 2005. Characterization of Mg/Ca distributions in planktonic foraminifera species by electron microprobe mapping. Geochemistry,   218 Geophysics, Geosystems 6 (12), doi: 10.1029/2005GC000973. Sadekov, Y.S., Eggins, S.M., Klinkhammer, G.P., Rosenthal, Y., 2010. Effects of seafloor and laboratory dissolution on the Mg/Ca composition of Globigerinoides sacculifer and Orbulina universa tests – A laser ablation ICPMS microanalysis perspective. Earth and Planetary Science Letters 292, p. 312-324.  Sapsford, D.J., Bowell, R.J., Dey, M., Williams, K.P., 2009. Humidity cell tests for the prediction of acid rock drainage. Minerals Engineering 22, p. 25-36.  Simate, G.S., Ndlovu, S., 2014. Acid mine drainage: Challenges and opportunities. Journal of Environmental Chemical Engineering 2, p. 1785-1803.  Sjöberg E.L., 1978. Kinetics and mechanisms of calcite dissolution in aqueous solutions at low temperatures. Stockholm Contributions in Geology 32, p. 1–94. Sjöberg E.L., Rickard D.T., 1983. The influence of experimental design on the rate of calcite dissolution. Geochimica et Cosmochimica Acta 47, p. 2281–2285. Sjöberg E.L., Rickard D.T., 1984. Calcite dissolution kinetics: surface speciation and the origin of the variable pH dependence. Chemical Geology 42, p. 119–136. Spero, H.J., 1988. Ultrastructural examination of chamber morphogenesis and biomineralization in the planktonic foraminifer Orbulina universa. Marine Biology 99, p. 9-20. Steefel, C.I., et al., 2015. Reactive Transport Codes for Subsurface Environmental Simulation, Computers and Geosciences 19, p. 445–478, doi: 10.1007/s10596-014-9443-x. Swoboda-Colberg, N.G., Drever, J.I., 1993. Mineral dissolution rates in plot-scale and laboratory experiments. Chemical Geology 105, p. 51-69.  Welch, S.A., Christy, A.G., Kirste, D., Beavis, S.G., Beavis, F., 2007. Jarosite dissolution I - Trace cation flux in acid sulfate soils. Chemical Geology 245, p. 183-197   219 White, A.F., Brantley, S.L., 1995. Chemical Weathering Rates of Silicate Minerals. Reviews in Mineralogy 31, Mineralogical Society of America, Washington, D.C.  White, A.F., Claasen, H.C., 1979. Dissolution kinetics of silicate rocks – Application to solute modeling. American Chemical Society Symposium Series 93, p. 447-473. Wogelius, R.A. and Walther, J.V., 1991. Olivine dissolution at 25ºC: Effects of pH, CO2, and organic acids. Geochimica et Cosmochimica Acta 55, p. 943-954.  Wolery, T.J., and Daveler, S.A., 1992. EQ3/6: A software package for geochemical modelling of aqueous systems: Package overview and installation guide (version 7.0). Technical Report UCRL-MA-110662 PT IV, Lawrence Livermore National Laboratory, USA. Wollast, R., 1967. Kinetics of the alteration of K-feldspar in buffered solutions at low temperature. Geochimica et Cosmochimica Acta 31, p. 635-648.  Yu, Y., Elderfield, H., Hönisch, B., 2007. B/Ca in planktonic foraminifera as a proxy for surface seawater pH. Paleoceanography 22, doi:10.1029/2006PA001347.     220 Appendices Appendix A: chapter 3 [Mg, Si] concentrations and Mg/Si molar ratios associated with samples 4, 5, 7 and 8 are provided in Fig. A.1. The leached layer thickness associated with these samples is provided in Fig. A.2. An alternative approach to quantify mineral dissolution rate parameters using model parameter estimation is discussed here (conceptual diagram provided in Fig. A.3), which may prove particularly powerful when studying mineral dissolution kinetics associated with more complex rate equations. Although we initially used this approach to quantify forsterite rate parameters, for simplicity we excluded this from the thesis chapter. Nevertheless, the potential use of inverse modeling is illustrated here using forsterite as a proof-of-concept study. Using forsterite as an example, in order to estimate the dissolution rate constant that fits best the entire data set generated during a run, we assumed that dissolution of forsterite follows the established rate law (Rimstidt et al., 2012): 𝑅! =  𝐴 × 𝑘!!× 𝑎!!! . This equation is based on a large empirical database which indicates that, at pH < 6 and for a given temperature, surface normalized forsterite dissolution rates depend only on the hydrogen ion activity of the contact solution. A forward reactive transport model developed with the geochemical software PHREEQC (version 3.1.3, Parkhurst and Appelo, 1999) in conjunction with the WATEQ4F thermodynamic database (Ball and Nordstrom, 1991) was combined with a parameter optimization program (Parameter ESTimation (PEST), Doherty, 2010). Using the SOLUTION, RATES, KINETICS, and TRANSPORT keywords, the dissolution of forsterite was simulated. The parameters relevant to forsterite dissolution under the experimental conditions used in this study (𝑟𝑓!! and 𝑛!!) were defined using BASIC language within the RATES keyword (Table A.1). Each following simulation represents a time step in the flow-through experiment, determined by the residence   221 time of the eluent in the flow-cell (2.1 seconds), and contains compositional details of the infilling solution. Given the large number of simulations required to model a single flow-through experiment, Python™ programming language was used to efficiently generate the infilling solution simulation data blocks (Table A.2). Parameters that are typically adjusted include: “total number of steps” (depending on length of experiment), “N-values” (representing the various nitric acid concentrations at different times over the course of a FT-TRA experiment), “N_steps” (representing the various times at which the nitric acid concentrations change) and transport time step (residence time of the eluent in the flow-through cell). Here, the python script allows for the programming of up to five time-dependent linear increases or decreases in acidity.  RATES     Forsterite -start 10 k1 = TBD 20 area = MEASURED 30 n = TBD 40 rf1 = ((M/M0)^(2/3))*area*k1*(ACT("H+")^n) 50 PUT(rf1, 1) 60 rf = rf1 70 PUT(rf, 2) 80 moles = rf *TIME 90 SAVE moles -end Table A.1 RATES keyword defining the forward rate equation (here: forsterite) as used in the PHREEQC geochemical forward model. 𝑘! and n are to be determined (TBD) as part of the parameter optimization process.         222 import sys def block(temp, pH, pe, redox, units, density, N, water, transport_shifts, transport_time_step, transport_lengths, selected_output):     s = "SOLUTION 0\n"    #for each time step in a flow-through dissolution      if temp: s += "  temp\t%d\n"      if pH: s += "  pH\t%s\n"      if pe: s += "  pe\t%s\n"      if redox: s += "  redox\t%s\n"      if units: s += "  units\t%s\n"      if density: s += "  density\t%s\n"      if N: s += "  N(5)\t%f\n"      if water: s += "  -water\t%s\n"      s += "TRANSPORT\n"     if transport_shifts: s += "  shifts\t%d\n"      if transport_time_step: s += "  time_step\t%s\n"      if transport_lengths: s += "  lengths\t%s\n"      s += "PRINT\n"     if selected_output: s += "        -selected_output\t%s\n"      s += "EQUILIBRIUM_PHASES 0\n"     s += "CO2(g)    -3.5 10\n"   #ambient conditions include atmospheric [CO2]      s += "END"     return s  def main():     total_steps = TBD      #number of times fow-through cell is refilled     N_begin_value = TBD    #starting [HNO3] (= 0 under DIW), expressed in M     N_value_1 = TBD         N_value_2 = TBD     N_value_3 = TBD     N_value_4 = TBD     N_value_5 = TBD     N_end_value = TBD     N_begin_step = TBD     #shift at which [HNO3] starts to ramp up linearly     N_step_1 = TBD     N_step_2 = TBD     N_step_3 = TBD     N_step_4 = TBD     N_step_5 = TBD     N_end_step = TBD     N_increment_1 = (N_value_1 - N_begin_value)/((N_step_1 - N_begin_step)*1.0)     N_increment_2 = (N_value_2 - N_value_1)/((N_step_2 - N_step_1)*1.0)     N_increment_3 = (N_value_3 - N_value_2)/((N_step_3 - N_step_2)*1.0)     N_increment_4 = (N_value_4 - N_value_3)/((N_step_4 - N_step_3)*1.0)     N_increment_5 = (N_value_5 - N_value_4)/((N_step_5 - N_step_4)*1.0)     #N_increment_6 = (N_end_value - N_value_5)/((N_end_step - N_step_5)*1.0)     for i in range(1, total_steps+1):         # Default values         temp = 25         pH = "7 charge"         pe = "4"         redox = "pe"         units = "mol/kgw"         density = "1"         N = None      # [HNO3], defined below    223         water = "1 # kg"         transport_shifts = 1 # the number of shifts happening         transport_time_step = " TBD # seconds" # residence time of eluent in flow-through cell         transport_lengths = "1e-008"         selected_output = "t"         if i < N_begin_step: N = N_begin_value # [HNO3] are dependent on the step number         elif i >= N_begin_step and i <= N_step_1: N = N_begin_value + N_increment_1*(i-N_begin_step)         elif i >= N_step_1 and i <= N_step_2: N = N_value_1 + N_increment_2*(i-N_step_1)         elif i >= N_step_2 and i <= N_step_3: N = N_value_2 + N_increment_3*(i-N_step_2)         elif i >= N_step_3 and i <= N_step_4: N = N_value_3 + N_increment_4*(i-N_step_3)         elif i >= N_step_4 and i <= N_end_step: N = N_value_4 + N_increment_5*(i-N_step_4)         elif i > N_end_step: N = N_end_value                # generate output file          print block(temp, pH, pe, redox, units, density, N, water, transport_shifts, transport_time_step, transport_lengths, selected_output)  main() Table A.2 Python script used to automatically generate a series of infilling solutions, simulating a typical FT-TRA experiment (comprised of SOLUTION 0, TRANSPORT and EQUILIBRIUM_PHASES keywords). Values to be determined (TBD) must be adjusted to FT-TRA experimental conditions.  In order to optimize dissolution rate parameters, model (PHREEQC) generated effluent concentrations must be compared to measured concentrations. This process occurs in an iterative manner using freely available Parameter ESTimation software (PEST). PEST extracts model generated flow-cell elemental concentrations after each model run in order to optimize pre-defined model parameters. In order to extract this information from the PHREEQC model output file, Matlab scripts were developed to isolate the relevant information (Table A.3) and interpolate this data to the temporal resolution of the experimental data (Table A.4).         224 clear all data = importdata('selected.out');    # raw model output data is imported data = data.data; a = -99;        # conditions are defined which indicate rows condition1=data(:,4)==a;     of irrelevant information data(condition1,:)=[];     # rows to which condition applies are  b = 1;        cleared condition2=data(:,4)==b; data(condition2,:)=[]; dlmwrite('cleaned', data);     # cleaned model output data is saved exit Table A.3 Matlab script used to isolate relevant PHREEQC output file data. Rows of model data output containing irrelevant information are deleted and the column of data containing the flow-cell elemental concentrations is saved.  clear all data = importdata('cleaned'); time = data(:,3);      # model time is imported Ca = data(:,14); # model output concentrations are  model = zeros(length(time),2);     imported model(:,1) = time; model(:,2) = Ca; dlmwrite('model_out_raw', model) exp_time = importdata('interpoldata');   # experimental time is imported time_exp = exp_time(:,1); Ca_int_exp = interp1(time,Ca,time_exp);   # model concentrations are interpolated model_Ca_interpol = zeros(length(Ca_int_exp),1);  to experimental time-scale model_Ca_interpol(:,1) = Ca_int_exp; dlmwrite('Ca.out', Ca_int_exp)    # interpolated concentrations are saved exit Table A.4 Matlab script used to linearly interpolate model data to experimental temporal resolution. As the temporal resolution of the model output equals one eluent residence time between two consecutive data points and the temporal resolution associated with the measured data was dictated by time-resolved ICP-MS settings, this script was necessary to align both temporal resolutions.  The automated framework (PYTHON – PHREEQC – MATLAB – PEST) was used to quantify the two dissolution parameters (dissolution rate constant 𝑘!! and reaction order 𝑛!!) that best fit the Mg or Si concentration data generated by the flow-through apparatus either over an entire run (sample 4 – 8) or during the steady state concentration “plateaus” (sample 2). Mg and Si concentration data (excluding the first DIW section) associated with samples 4 – 8 (Fig.   225 A.4) and sample 2 (Fig. A.5) was used to quantify the dissolution rate constant (𝑘!!) and the reaction order (𝑛!!). Optimized dissolution rate parameters are provided in Table A.5.   226    Mg Si Mg / Si (molar ratio) 051015202530354045Mg/Si (molar)1.41.61.82.02.22.42.62.83.03.2[Mg, Si] (x 10-7  moles L-1)051015202530Mg/Si (molar)1.01.52.02.53.03.5051015202530Mg/Si (molar)1.41.61.82.02.22.42.62.83.03.2[Mg, Si] (x 10-7  moles L-1)051015202530Mg/Si (molar)1.52.02.53.03.5Time (s)0 2000 4000 6000 8000[Mg, Si] (x 10-7  moles L-1)[Mg, Si] (x 10-7  moles L-1) A: sample 4B: sample 5C: sample 7D: sample 8  227 Figure A.1: [Mg, Si] concentrations (5-point running average in black and grey respectively) and Mg/Si molar ratio (red) measured in the effluent from the dissolution of samples 4, 5, 7 and 8. The dashed lines indicate the stoichiometric Mg / Si ratio (1.81 ± 0.07). Leached layer thickness associated with samples 4, 5, 7 and 8 are provided in Fig. A.2.   228  Mg/Si (molar ratio) Leached layer thickness (Å)Mg / Si (molar ratio)1.41.61.82.02.22.42.62.83.03.2Leached layer thickness (Å)024681012140 2000 4000 6000 8000Mg / Si (molar ratio)1.01.52.02.53.03.54.04.5 Leached layer thickness (Å)02468101214160 2000 4000 6000 8000Mg / Si (molar ratio)1.41.61.82.02.22.42.62.83.03.2Leached layer thickness (Å)0123456780 2000 4000 6000 8000Mg / Si (molar ratio)1.52.02.53.03.54.0Leached layer thickness (Å)024681012Time (s)0 2000 4000 6000 8000A: sample 4B: sample 5C: sample 7D: sample 8  229 Figure A.2 Leached layer thickness calculated using difference in Mg and Si leached layer thickness (Mg – Si thickness, expressed in Å, solid red line) associated with samples 4, 5, 7 and 8 (as indicated). The dashed red lines represent range in Mg – Si leached layer thickness associated with error on sample stoichiometry (Mg/Si = 1.81 ± 0.07). Mg/Si molar ratio (5-point running average) is shown in black with running average shown in black. Leached layer thickness was calculated by multiplying time-resolved cumulative moles of Mg (stoichiometry corrected), Si dissolved by forsterite molar volume (43.79 cm3 mol-1, sourced from Cemič, 2005) and dividing by BET surface area.    Figure A.3 Conceptual diagram showing the automated dissolution rate parameter optimization process. A reactive transport model simulates the dissolution of a mineral during a typical FT-TRA experiment, and modeled effluent composition can be exported (during model output post-processing). Automated parameter estimation software compared model output to experimental data and optimizes model parameter values in an iterative way until dissolution rate parameters have been optimized.    230  Figure A.4 Mg concentrations* obtained by inverse modeling (green) and measured by ICP-MS (red) for samples 4 – 8. *Initial ~30 minutes of data obtained under DIW conditions was excluded from the inverse modeling process, given the lack of steady state conditions.   231  Figure A.5 Sample 2 experimental and inverse modeling results. (A): Steady state Mg concentrations obtained by inverse modeling (green) and measured by ICP-MS (red). (B): Steady state Si concentrations obtained by inverse modeling (green) and measured by ICP-MS (red).         232  Sample Optimized 𝑘!! rate constant  (moles m-2 s-1) Optimized reaction order   𝑛!!  Optimized 𝑘!! rate constant  (moles m-2 s-1) Optimized reaction order   𝑛!!   Based on Mg Based on Si 4 6.05 ± 0.06(!)  × 10!! 0.50 ± 0.05(!)  6.49 ± 1.32(!)  × 10!! 0.50 ± 0.08(!)  5 5.45 ± 0.75(!)  × 10!! 0.50 ± 0.07(!)  5.36 ± 0.19(!)  × 10!! 0.50 ± 0.02(!)  6 4.20 ± 0.12(!)  × 10!! 0.52 ± 0.01(!)  4.07 ± 0.59(!)  × 10!! 0.50 ± 0.07(!)  7 3.26 ± 0.10(!)  × 10!! 0.50 ± 0.02(!)  3.50 ± 0.07(!)  × 10!! 0.54 ± 0.02(!)  8  3.97 ± 0.27(!)  × 10!! 0.50 ± 0.03(!)  3.97 ± 0.23(!)  × 10!! 0.50 ± 0.02(!)   Average   4.6 ± 1.0(!)  × 10!!  0.50 ± 0.01(!)   4.68 ± 1.09(!)  × 10!!  0.51 ± 0.02(!)  2 2.91 ± 0.04(!)  × 10!! 0.50 ± 0.01(!)  2.83 ± 0.07(!)  × 10!! 0.50 ± 0.01(!)  Table A.5 PEST optimized dissolution rate constant 𝑘!! and reaction order 𝑛!!based on measured Mg and Si concentrations.  (a) values represent PEST generated 95 % confidence intervals.  (b) values represents 2 standard error based on results from replicate samples 4 – 8.          233 Appendix B: chapter 4  An additional series of calcite dissolution experiments were carried out using deionized water eluent (pH 5.6) under varying flow rates (Table B.1). In parallel, pore-scale modeling was also carried out under these conditions (Figs. B.1, B.2). Pore scale modeling results reveal multiple mechanisms to control the dissolution of calcite, depending on the flow rate: • At flow rates higher than 2 × 10!! mL s-1, transport limitations exist pertaining to 𝑘! dependence. • At flow rates between  3 × 10!! → 2 × 10!! mL s-1, surface limitations exist pertaining to 𝑘! dependence. • At flow rates lower than 3 × 10!! mL s-1, transport limitations exist pertaining to equilibrium (supersaturation) dependence. As a result, calcite dissolution rates decrease as flow rates decrease. Comparing the pore scale modeling findings to our experimental results, disagreement was found in values and qualitative trends (Figure B.2). Further work is required to address this issue. Nevertheless, when comparing the highest measured calcite dissolution rate (at Q = 0.125 mL s-1) to calcite dissolution rates calculated based on previously reported rate constants, close agreement can be found. Measured calcite dissolution rates fall within a factor of at least 3 compared to Chou et al., 1989, Plummer et al., 1978, Sjöberg, 1978, and Busenberg and Plummer, 1986. This finding suggests that FT-TRA is close to achieving surface-controlled dissolution (assuming previously reported dissolution rate constants were recorded under a surface-controlled dissolution regime) under high flow rates at eluent pH 5.6. Future work involving higher flow rates and/or the development of optimized flow-cell design may allow for   234 the establishment of constant flow rates across a range of flow rates, thereby establishing an empirically demonstrated surface-controlled dissolution regime.   Q (cm3 s-1)  1.25E-01 8.33E-02 6.67E-02 5.00E-02 3.33E-02 1.67E-02 1.17E-02 5.83E-03 4.17E-03 1.33E-03 1.00E-03 5.00E-04 3.33E-04 2.50E-04 1.67E-04 Table B.1 DIW eluent (pH = 5.6) experimental flow rates.   Fast flow rate Slow flow rate     235 Figure B.1 Model domain cross-section (simulations are 3-dimensional), showing transport-controlled calcite rhomb (represented as white square) dissolution during highest (left) and lowest (right) flow rate conditions under DIW eluent conditions. Color gradient within flow-cell indicates pH. Eluent velocity direction and magnitude indicated by arrow direction and length (scaled in each plot), respectively.   Figure B.2 Calcite dissolution rates on logarithmic scale from experimental data (black filled circles, error falls within symbol size), and pore scale modeling (black diamonds) under incoming DIW eluent (pH = 5.6). Pore scale modeling data were obtained using k1, k2 from Chou et al. (1989), and k3 from Busenberg and Plummer (1986).  Modeled effluent pH data is also shown (black filled triangles, axis shown on right).     ExperimentalModel pH log r calcite (moles cm-2 s-1)−10.4−10.2−10.0−9.8−9.6−9.4−9.2pH5.56.06.57.07.58.08.59.09.510.0log flow rate (mL s-1)−4 −3 −2 −1  236 Appendix C: chapter 5  Sample 1 Sample 2 Sample 3          Figure C.1: SEM imagery associated with calcite grains (samples 1 – 3).      237 Sample 4 Sample 5       Figure C.2: SEM imagery associated with aragonite grains (samples 4 – 5).   238 Exploring the added value of FT-TRA to measure Mg/Ca, Sr/Ca in ostracod shells Introduction  Due to the complex structure of foraminifera involving multiple biogenic calcite phases (ontogenetic versus gametogenic), we explore the added value of using FT-TRA on ostracods, small crustaceans that are common in most types of aquatic environments. Ostracods secrete bivalved shells, which consist of low-Mg calcite. The key benefit associated with ostracod shells is that they are secreted over a very short time (e.g. Turpen and Angell, 1971), and therefore – unlike foraminifera – do not build up incrementally. It has been established for a while that trace elements co-precipitate with the calcite in ostracod shells (e.g. Cadot et al., 1972; Kesling, 1951; Sohn, 1958), thereby providing a means to study past ambient water composition. In downcore studies, the use of e.g. Mg/Ca, Sr/Ca molar ratios are used as indicators of relative water temperature and chemistry, rather than quantitative values requiring calibrations (Griffiths and Holmes, 2000). Here, we explore using the newly developed FT-TRA approach to measure Mg/Ca, Sr/Ca molar ratios in ostracod shells.   Study goals I. Develop an FT-TRA leaching protocol for ostracod shells. II. Asses the effect of ostracod cleaning on Mg/Ca, Sr/Ca measured using FT-TRA. III. In both a recent outcrop sample and downcore samples, compare Mg/Ca and Sr/Ca molar ratios obtained using (a) laser ablation and (b) FT-TRA.  Sample origin Ostracod shell samples from a recent outcrop at the northwestern shore of lake Tangra YumCo (central Tibetan Plateau) were obtained. This sample was chosen because abundant adult valves   239 of one species were available from the same outcrop sample, facilitating method development. The ostracod species Leucocytherella sinensis (discovered by Huang in 1982) was selected as it is the most abundant species in lakes on the central Tibetan Plateau and it represents a small and lightweight ostracod species (ca. 700 µm long and weighs approximately 7 µg). A meaningful signal would therefore represent a promising start towards studying other species. For the downcore comparison study, sediment core NC 08/01, with a length of 10.4 m, was collected in the eastern part of lake Nam Co (central Tibetan Plateau, 30.73º N, 90.78º E) at a depth of 93 m. Samples were sieved using 200 and 1000 µm mesh size sieves, and rinsed with deionized water. From the 200 – 1000 µm size fraction adult L. sinensis valves were picked and – for a subset of the sample – cleaned with a fine brush under a low magnification stereoscopic microscope using deionized water and ethanol. Both male and female valves were picked, as no chemical differences between males and females or right and left valve exist (e.g. Dwyer et al., 2002; Marco-Barba et al., 2012; Morishita et al., 2007).  FT-TRA leaching methodology FT-TRA leaching was carried out on 8 method calibration samples (S1 – S8) as well as 7 downcore samples (S9 – S15). The continuous leaching sequence used in this work involves an initial rinse with deionized water followed by gradually increasing acid conditions in the form of nitric acid. After initial exposure to DIW for 10 minutes, the sample is gradually exposed to a maximum acidity of 50 mM HNO3 (pH = 1.3) until the entire sample is dissolved (determined by on line monitoring of the Ca counts per second), after which the acidity is increased to 155 mM HNO3 (pH = 0.8). This leaching sequence results in an initial release of any water-soluble phases, followed by the dissolution of the main carbonate (resulting in a Ca peak), and ends in   240 the dissolution of more acid-resistant contaminant mineral phases such as clays (if present). Although chapter 5 has shown the inability of FT-TRA to separate mineral phases based on susceptibility to dissolution, this work was carried out based on the assumption ostracods largely consist of a single biogenic carbonate phase.   Laser ablation methodology Laser ablation analyses were carried out both at the Max Planck Institute for Chemistry (MPC) in Mainz (downcore samples S9 – S15), as well as locally at the Pacific Centre for Isotopic and Geochemical Research (PCIGR), housed at the University of British Columbia (calibration samples S16, S17). Laser ablation methodology used to analyze ostracod shells at MPC Mainz has previously been documented by Yang et al. (2014).  At PCIGR, an Excimer 193 nm (Resonetics M-50) laser module was used, coupled to an Agilent 7700x inductively coupled plasma mass spectrometer housed at the University of British Columbia. Spot analyses (17 µm in diameter, energy density 4 J cm-2, repetition rate of 5 Hz) were carried out for 60 seconds, which was found to be sufficient to ablate through the thickness of the ostracod shell. Line-scan measurements were carried out using the above settings, with a scan speed of 17 µm sec-1. The carrier gas consisted of a mixture of He (900 mL min-1) and N2 (2 mL min-1). In the ICP-MS, an Ar flow of 0.53 mL min-1 was used. Data reduction was carried out using gas blank (25 second washout period) corrected count rates of the analyzed isotopes (24Mg, 25Mg, 27Al, 43Ca, 55Mn, 88Sr) relative to the internal standard (43Ca). Calibration was carried out using the NIST SRM 612 and 610 reference glasses. Oxide formation was monitored using 248ThO/232Th, and was found to be below 0.2%.     241 Results and Discussion 1. Development of an FT-TRA leaching protocol for ostracod shells The FT-TRA leaching protocol developed as part of this study allows for the dissolution of a single ostracod shell over the period of approximately an hour or less (Figs. C.3, C.4). This allows for studying the geochemical heterogeneity between multiple, single-shell samples. Time intervals below the main Ca peak were selected (indicated by bold black lines in Figs. C.3, C.4), and Mg/Ca, Sr/Ca below this peak were averaged. Hence, the assumption is made that the ratios measured below this Ca peak represents the biogenic carbonate associated with the ostracod shell.  Within a single ostracod shell, Sr/Ca seems to be relatively homogeneously distributed throughout the ostracod shell as compared to Mg/Ca (Figs. C.3, C.4). This finding is in line with limited Sr/Ca variability reported by Morishita et al. (2007). Most pronounced is the over 2-fold Sr/Ca variability between multiple single-shell samples (e.g. sample S4 versus S5, Table C.1). Mg/Ca varied much less between multiple single-shell samples (Table C.1), and was affected by the presence / absence of manual cleaning (section 2). Most importantly, the absence of constant Mg/Ca molar ratios associated with the main Ca peak (Fig. C.3) suggests the presence of Mg/Ca heterogeneity within a single ostracod shell. Geochemical Mg/Ca heterogeneity within an ostracod shell has recently been studied using an electron microprobe analyzer by Morishita et al. (2007) and has been attributed to changes in physiological activity.  The analysis of a single ostracod shell has previously been noted to likely lead to misleading stratigraphic interpretations (e.g. Griffith and Holmes, 2000). Therefore, a number of studies have analyzed between three and five single ostracod shells per interval in stratigraphic   242 studies (e.g. Chivas et al., 1985; De Deckker et al., 1988). Reasons put forward include (1) seasonal changes in water temperature and composition; (2) inter-annual variability in water temperature and composition and (3) spatial variability in water temperature and composition.  2. The effect of ostracod cleaning on Mg/Ca, Sr/Ca measured using FT-TRA As part of the calibration dataset, average Mg/Ca molar ratios associated with manually cleaned ostracod shells (n = 5) were 1.61± 0.04  × 10!!, whereas untreated ostracod shells (n = 3) had Mg/Ca molar ratios of 1.98± 0.22  × 10!!, illustrating manual brush cleaning seems to be effective at removing Mg-containing contaminants (Fig. C.3, Table C.1). Hence, manual brush cleaning seems appropriate and effective at removing Mg contamination.  In comparison, brush cleaning does not seem to affect Sr/Ca ratios (Fig. C.4, Table C.1), as cleaned ostracod shell Sr/Ca amounts to 1.41± 0.52  × 10!! whereas untreated ostracod shell Sr/Ca amounts to 1.80± 0.17  × 10!!.  3. Laser ablation results 3.1 Laser ablation and FT-TRA comparison of Mg/Ca and Sr/Ca in recent outcrop sample In order to further study the geochemical Mg/Ca heterogeneity within the calibration sample used in sections 1 and 2, in-house laser ablation experiments were carried out. 2 samples from the same recent outcrop sample underwent laser-ablation analysis: sample S16 underwent track analysis (Fig. C.5), whereas sample S17 was ablated at a single spot (Fig. C.6). Using this two-fold approach, geochemical heterogeneity could be assessed across the shell surface (track analysis), as well as across the shells depth (single spot ablation). Results are provided in Figs. C.5. and C.6, and reveal no distinct patterns. Averaged Mg/Ca, Sr/Ca molar ratios for both   243 samples are provided in Table C.3. The Sr/Ca molar ratios associated with samples S16, S17 lie within the range measured using FT-TRA (Fig. C.7). Similarly, Mg/Ca associated with the track analysis measurement (S16) falls close to the Mg/Ca range measured using the FT-TRA measurements. In particular, the Mg/Ca molar ratio associated with the single spot ablation revealed somewhat elevated Mg/Ca molar ratios (2.49± 0.10  × 10!!) relative to FT-TRA based Mg/Ca measurements (highest Mg/Ca molar ratio measured in sample S8: 2.19±0.029  × 10!!).   3.2 Laser ablation and FT-TRA comparison of Mg/Ca and Sr/Ca in downcore samples. Given the promising results obtained in terms of agreement between in-house laser ablation and FT-TRA based Mg/Ca, Sr/Ca measurements, downcore samples (previously analyzed using laser ablation at the Max Planck Institute for Chemistry in Mainz) were analyzed using FT-TRA to measure Mg/Ca, Sr/Ca molar ratios of more fossil ostracod shells (Figs. C.8, C.9). Non-steady state Mg/Ca and Sr/Ca molar ratios were found associated with the main Ca peaks, suggesting the presence of geochemical heterogeneity. Mg/Ca molar ratios obtained using laser ablation at the Max Planck Institute for Chemistry in Mainz lie consistently higher (except for agreement found at 2915 yrs BP) compared to FT-TRA derived molar ratios (Fig. C.10, Table C.4). Nevertheless, a similar downcore pattern was obtained. The reported Mg/Ca molar ratios obtained using LA-ICP-MS represents the average of 2 – 4 spot analyses and 5 – 8 track analyses, which is assumed to represent the composition of the whole sample (Yang et al., 2014). Given the fact that only the average molar ratios are available, we can only speculate that geochemical heterogeneity may have contributed to these elevated Mg/Ca molar ratios. Similar to the calibration samples, no   244 steady Mg/Ca molar ratios were measured below the main Ca peak in the downcore samples (Fig. C.8). The in-house laser ablation work carried out in on the calibration samples did not reveal significant geochemical heterogeneity using LA-ICP-MS, however this may have been related to the nature of the sample. The calibration ostracod shells were sourced from a recent outcrop sample, having undergone minimal diagenetic alteration. This downcore comparison study is based on fossil material, which may have undergone more exhaustive geochemical alteration over time. Hence, the more substantial difference may be attributable to diagenetic alteration.  In comparison, Sr/Ca molar ratios are in good agreement between both analytical techniques (Table C.4, Fig. C.10). Hence, in terms of Sr/Ca molar ratios, post-depositional alteration does not affect the surface of the ostracod shell (obtained by laser ablation) to a greater degree than the bulk composition of the shell (obtained by FT-TRA).   Preliminary conclusions and future research directions I. FT-TRA can be used as a tool to quantify elemental molar ratios in ostracod shells. However, it must be noted that FT-TRA does not provide additional information related to the geochemical heterogeneity found within a single ostracod shell, as previously reported by for example Morishita et al. (2007) and Cadot and Kaesler (1977).  II. 2-fold Sr/Ca variability was found between multiple, single shell ostracod samples sampled from a recent outcrop. This finding further highlights the need to consider multiple shells to yield meaningful records of long-term variability of past water conditions.    245 III. Based on ostracod shells sampled from a recent outcrop, manual brush cleaning can be considered an effective tool to remove Mg-contaminants, which is reflected in lower Mg/Ca molar ratios as quantified during FT-TRA.  IV. No patterns in geochemical Mg/Ca, Sr/Ca heterogeneity was found using spot and track laser ablation analysis carried out on a recent outcrop sample. V. In downcore samples, when comparing Mg/Ca obtained using FT-TRA to laser ablation analysis, similar trends were found, with somewhat lower Mg/Ca molar ratios measured using FT-TRA. This may be attributable to post-depositional alteration affecting the LA-ICP-MS measurement, which must be addressed in future studies. In contrast, very similar downcore values were found for Sr/Ca, suggesting minimal post-depositional Sr/Ca alteration.     246  Figure C.3 Core-top sample flow-through [Ca], Mg/Ca molar ratio observations. Mg/Ca molar ratios (5-point running average) shown in green. Ca concentrations expressed in 10-6 moles L-1. Bold black lines indicate time interval used below main Ca peak to quantify Mg/Ca ratios.   247  Figure C.4 Core-top sample flow-through [Ca], Sr/Ca molar ratio observations. Sr/Ca molar ratios (5-point running average) shown in green. Ca concentrations expressed in 10-6 moles L-1. Bold black lines indicate time interval used below main Ca peak to quantify Sr/Ca ratios.      248 Sample Preparation Weight  (mg) Time interval (s) Mg / Ca (moles / mole) Sr / Ca (moles / mole) S1 Untreated(1) 0.004 1000 – 1600 1.81 ± 0.012(!) × 10!! 1.65 ± 0.009(!) × 10!! S2 Untreated (1) 0.004 900 – 1200 1.93 ± 0.022(!) × 10!! 1.80 ± 0.013(!) × 10!! S3 Cleaned(1) 0.007 900 – 1500  1.57 ± 0.015(!) × 10!! 1.90 ± 0.012(!) × 10!! S4 Cleaned(1) 0.008 1000 – 1450 1.57 ± 0.013(!) × 10!! 1.86 ± 0.015(!) × 10!! S5 Cleaned(1) 0.007 850 – 1350 1.60 ± 0.016(!) × 10!! 0.78 ± 0.007(!) × 10!! S6 Cleaned(1) 0.007 900 – 1350 1.68 ± 0.031(!) × 10!! 1.73 ± 0.016(!) × 10!! S7 Cleaned(2) 0.04 4400 – 4850 1.64 ± 0.016(!) × 10!! 0.79 ± 0.005(!) × 10!! S8 Untreated (2) 0.03 1000 – 1450 2.19 ± 0.029(!) × 10!! 0.94 ± 0.011(!) × 10!! Average  Cleaned + Untreated   1.75 ± 0.15(!) × 10!! 1.56 ± 0.34(!) × 10!! Average  Clean   1.61 ± 0.04(!) × 10!! 1.41 ± 0.52(!) × 10!! Average Untreated   1.98 ± 0.22(!) × 10!! 1.80 ± 0.17(!) × 10!! Table C.1 Overview of core-top samples (1) Sample consists of individual ostracod shell (2) Sample consists of 5 ostracod shells (3) 95% confidence interval (2𝜎) derived from single sample (4) 95% confidence interval (2𝜎) derived from all samples measured (5) 95% confidence interval (2𝜎) derived from all cleaned samples measured (S3, S4, S5, S6, S7).  (6) 95% confidence interval (2𝜎) derived from all untreated samples measured (S1, S2 and S8)    249  Figure C.5 Track laser ablation Mg/Ca (upper panel) and Sr/Ca (lower panel) analysis of a core-top ostracod shell (sample S16). Average values are provided in Table C.3.     250  Figure C.6 Single spot laser ablation analysis (Sample S17). Mg/Ca molar ratio shown in upper panel, Sr/Ca shown in lower panel. Average values are provided in Table C.3.  Sample Preparation n Mg / Ca (moles / mole) Sr / Ca (moles / mole) S16 Untreated 27(1) 1.53 ± 0.033(!) × 10!! 0.78 ± 0.011(!) × 10!! S17 Untreated 1(3) 2.49 ± 0.10(!) × 10!! 0.81 ± 0.020(!) × 10!! Table C.3 Overview of laser ablation samples (1) Number of ablations across laser ablation track (2) 95% confidence interval (2𝜎) derived from all ablation spots measured (3) Single spot surface ablation (4) 95% confidence interval (2𝜎) associated with measurements from single ablation spot.    251  Sample (yrs BP) Method Weight  (mg) Time interval (s) Mg / Ca (× 10!! molar) Sr / Ca (× 10!!molar) S9 (525) FT-TRA 0.003 800 – 1300  1.34 ± 0.031(!) 13.8 ± 0.160(!)  LA-ICP-MS  N/A 3.99 11.6 S10 (16824) FT-TRA 0.002 750 – 950  0.27 ± 0.017(!) 2.68 ± 0.025(!) LA-ICP-MS  N/A 0.49 2.44 S11 (2915) FT-TRA 0.002 750 – 850  0.62 ± 0.023(!) 2.75 ± 0.073(!) LA-ICP-MS  N/A 0.58  2.28 S12 (6685) FT-TRA 0.004 750 – 950  0.44 ± 0.011(!) 2.05 ± 0.030(!) LA-ICP-MS  N/A 0.92 1.59 S13 (9825) FT-TRA 0.002 1325 – 1400 0.53 ± 0.021(!) 2.48 ± 0.085(!) LA-ICP-MS  N/A 0.84 2.62 S14 (11280) FT-TRA 0.005 875 – 1350  1.06 ± 0.014(!) 19.55 ± 0.116(!) LA-ICP-MS  N/A 1.90 16.35 S15 (13165) FT-TRA 0.001 850 - 1050 0.24 ± 0.008(!) 2.25 ± 0.004(!) LA-ICP-MS   2.12 1.67 Table C.4 Overview of downcore samples. All FT-TRA samples consisted of a single ostracod shell, which were brush-cleaned prior to analysis.   (1) 95% confidence interval (2𝜎) derived from single sample     252  Figure C.7 Summary of core-top sample Mg/Ca (upper panel) and Sr/Ca (lower panel) molar ratios measured using FT-TRA. Data provided in Table C.1. Errors represent 95% confidence interval based on data envelope selected below Ca peak. Data measured using laser ablation (LA-ICP-MS) also shown for comparison.   253  Figure C.8 Downcore sample flow-through [Ca], Mg/Ca molar ratio observations. Mg/Ca molar ratios (5-point running average) shown in green. Bold black lines indicate time interval used below main Ca peak to quantify Mg/Ca ratios.    254  Figure C.9 Downcore sample flow-through [Ca], Sr/Ca molar ratio observations. Sr/Ca molar ratios (5-point running average) shown in green. Bold black lines indicate time interval used below main Ca peak to quantify Mg/Ca ratios.    255  Figure C.10 Comparison of downcore results obtained by FT-TRA (filled black circles connected by solid line) and conventional laser ablation approach (filled red circles connected by dotted line). A: Mg/Ca molar ratios; B: Sr/Ca molar ratios. Error bars associated with FT-TRA results represent minimum and maximum molar ratios measured below main Ca peak.          256 References cited Cadot, H.M., Kaesler, R.L., 1977. Magnesium content of calcite in carapaces of benthic marine Ostracoda. Paleontological contributions, p. 1–23. Cadot, H.M., Van Schmus, W.R., Kaesler, R.L., 1972. Magnesium in calcite of marine ostracoda. Geological Society of America Bulletin 83, p. 3519–3522. Chivas, A.R., De Deckker, P., Shelley, J.M.G., 1985. Strontium content of ostracods indicates lacustrine palaeosalinity. Nature 316, p. 251–253. De Deckker, P., 1988. An account of the techniques using ostracodes in palaeolimnology in Australia. Palaeogeography, Palaeoclimatology, Palaeoecology 62, p. 463–475. Dwyer, G.S., Cronin, T.M., Baker, P.A., 2002. Trace elements in marine ostracodes. Geophysical Monograph Series 131, p. 205–225. Griffiths, H.I., Holmes, J.A., 2000. Non-marine Ostracods and Quaternary Palaeoenvironments. Issue 9 of Quaternary Research Association technical guide.  Kesling, R.V., 1951. The morphology ostracod molt stages. Illinois Biological Monographs 21, p. 1–126. Marco-Barba, J., Ito, E., Carbonell, E., Mesquita-Joanes, F., 2012. Empirical calibration of shell chemistry of Cyprideis torosa (Jones, 1850) (Crustacea: Ostracoda). Geochimica Et Cosmochimica Acta 93, p. 143–163. Morishita, T., Tsurumi, A., Kamiya, T., 2007. Magnesium and strontium distributions within valves of a recent marine ostracode, Neonesidea oligodentata: Implications for paleoenvironmental reconstructions. Geochemistry Geophysics Geosystems 8, p. 1–11. Sohn, I.G., 1958. Chemical constituents of ostracodes; some applications to paleontology and paleoecology. Journal of Paleontology 32, p. 730–736.   257 Turpen, J.B., Angell, R.W., 1971. Aspects of Molting and Calcification in the Ostracod Heterocypris. Biological Bulletin 140, p. 331–338. Yang, Q., Jochum, K.P., Stoll, B., Weis, U., Börner, N., Schwalb, A., Frenzel, P., Scholz, D., Doberschütz, S., Haberzettl, T., Gleixner, G., Mäusbacher, R., Zhu, L., Andreae, M.O., 2014. Trace element variability in single ostracod valves as a proxy for hydrochemical change in Nam Co, central Tibet, during the Holocene. Palaeogeography, Palaeoclimatology, Palaeoecology 399, p. 225–235.                   258 Appendix D: chapter 6  The oxidation of pyrite in aqueous systems consists of a series of biogeochemical processes involving a number of redox reactions as well as microbial catalysis. If pyrite is oxidized by O2, and all Fe precipitates as ferric hydroxide, every mole of FeS2 generates 4 moles of H+ (equation D.1):  𝐹𝑒𝑆!(!) + !"!  𝑂!(!") + !!  𝐻!𝑂(!) →   𝐹𝑒(𝑂𝐻)! (!) +  2 𝑆𝑂!(!") + 4 𝐻 (!")!       (D.1)  However, this amount of acid will only be generated if all the Fe precipitates as 𝐹𝑒(𝑂𝐻)! (!) or if 𝐹𝑒(𝑂𝐻)! (!")!  or 𝐹𝑒(𝑂𝐻)! (!") form (Lapakko and Berndt, 2003). If the concentration of Fe3+ and pH are low enough that the solubility of ferric hydroxide is not exceeded, and if aqueous hydroxide species are not formed, then only ¼ of the acid is generated (equation D.2). The remaining ¾ will be generated further downstream (in a more diluted form) wherever the ferric hydroxide is able to precipitate or form as an aqueous species (equation D.3).  2 𝐹𝑒𝑆!(!) +  !"!  𝑂!(!") +  𝐻!𝑂(!") →  2 𝐹𝑒(!")!! +   4 𝑆𝑂!(!") +  2𝐻(!")!        (D.2) 𝐹𝑒(!")!! +  3 𝐻!𝑂(!")  →   𝐹𝑒(𝑂𝐻)! ! + 3𝐻(!")!        (D.3)  However, below pH 4.5, 𝐹𝑒(!")!!  becomes the dominant oxidant (Williamson and Rimstidt, 1994), generating acidity without the formation of ferric hydroxide (equation D.4).     259 𝐹𝑒𝑆!(!) +  𝐹𝑒(!")!! +  8 𝐻!𝑂(!")  →  𝐹𝑒(!")!! +  2 𝑆𝑂!(!")!! + 16 𝐻(!")!        (D.4)  Under sufficiently dry conditions, the dissolved ferrous and sulfate ions first reach saturation with respect to melanterite (𝐹𝑒𝑆𝑂! ⋅ 7𝐻!𝑂). Upon melanterite dissolution and Fe oxidation without 𝐹𝑒(𝑂𝐻)! ! formation, acid is consumed (equation D.5).  2 𝐹𝑒!! + 2 𝑆𝑂! (!")!! + !!𝑂!(!") + 2 𝐻(!")!  → 2𝐹𝑒(!")!!  + 2 𝑆𝑂! (!")!!    + 𝐻!𝑂(!")   (D.5)  But, when ferric hydroxide (also referred to as ferrihydrite) forms, acidity is released (equation D.6).  2 𝐹𝑒!! + 2 𝑆𝑂! (!")!! + !!𝑂!(!") + 5 𝐻!𝑂(!")  →   2 𝐹𝑒(𝑂𝐻)! ! + 2𝑆𝑂! (!")!!  + 4𝐻(!")!   (D.6)  Under low pH, high sulphate conditions, jarosite (𝐾𝐹𝑒!(𝑆𝑂!) !(𝑂𝐻)!) saturation can be reached, as shown in equation D.7.   3𝐹𝑒𝑆!(!) + !"!  𝐻!𝑂(!") + !"!  𝑂!(!") + 𝐾!(!") →  𝐾𝐹𝑒!(𝑆𝑂!) !(𝑂𝐻)! (!)  +  4 𝑆𝑂!(!") + 9 𝐻 (!")!  (D.7)  Compared to equation D.1, less acid is released during jarosite formation and the difference is stored in the jarosite mineral, subject to be released upon dissolution (and ferric hydroxide is formed) as illustrated in equation D.8.   260  𝐾𝐹𝑒!(𝑆𝑂!) !(𝑂𝐻)! (!)  +  𝐻!𝑂(!") →  3𝐹𝑒(𝑂𝐻)! ! +  2𝑆𝑂! (!")!! + 3𝐻(!")! + 𝐾!(!")   (D.8)  Alternatively, if ferric hydroxide doesn't precipitate (i.e. when the jarosite dissolves very slowly and/or pH is sufficiently low) acidity will be generated downstream wherever Fe3+ forms ferric hydroxide (solid or aqueous). Jarosite dissolution rates are minimal at pH 3.5 (Madden et al., 2012). As a result, jarosite reaches saturation under acidic conditions, temporarily storing acid, which is gradually released over time.  Mineral Ideal Formula Original sample DIW exp. residue Acidic exp. residue Quartz  SiO2 45.2 46.7 45.0 Clinochlore (Mg,Fe2+)5Al(Si3Al)O10(OH)8 1.8 1.5 0.9 Kaolinite Al2Si2O5(OH)4 1.4 1.2 1.9 Muscovite KAl2AlSi3O10(OH)2 29.8 29.5 28.7 Plagioclase NaAlSi3O8 – CaAl2Si2O8 10.7 10.3 10.5 K-feldspar KAlSi3O8 8.5 8.4 10.9 Jarosite KFe3(SO4)2(OH)6 2.6 2.2 2.0 Pyrite FeS2  0.1 < 0.1 Total  100.0 100.0 100.0 Table D.1 Sample (column 2B) composition expressed in percentage based on x-ray diffraction  Rietfeld analysis.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0220867/manifest

Comment

Related Items