sign in sign up
<<
Home , Biomineralization, Exoskeleton

Establishing a physical and chemical framework for Amorphous (ACC)

Sebastian T. Mergelsberg

Dissertation submitted to the faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of

Doctor of Philosophy In Geosciences

Patricia M. Dove, chair F. Marc Michel J. Donald Rimstidt Michael F. Hochella, Jr. Biswarup Mukhopadhyay

April 27, 2018 Blacksburg, VA

Keywords: Biomineralization, ACC, , , magnesium

© Sebastian T. Mergelsberg, 2018

Establishing a physical and chemical framework for Amorphous Calcium Carbonate (ACC) biomineralization

Sebastian T. Mergelsberg

ABSTRACT

Recent advances in high-resolution analytical methods have brought about a paradigm shift in our understanding of how crystalline materials are formed. The scientific community now recognizes that many earth materials form by multiple pathways that involve metastable intermediates. Biogenic calcium carbonate are now recognized to develop by aggregating molecules or clusters to form amorphous phases that later transform to one or more crystalline polymorphs. Amorphous calcium carbonate (ACC) is now recognized as a precursor to CaCO3 biominerals in a wide variety of natural environments. Recent studies suggest an ACC pathway may imprint a different set of dependencies from those established for classical growth processes. Previous ACC studies provided important insights, but a quantitative understanding of controls on ACC composition when formed at near-physiological conditions is not established. The Mg content of ACC and is of particular interest as a minor element that is frequently found in final crystalline products in calcified . This three-part dissertation investigated biological and well-characterized synthetic ACC using high-energy x-ray methods, Raman spectroscopy, and mechanical tests. The findings establish chemical and physical properties of ACC in the of crustaceans and show Mg and P levels are tuned in the component to optimize exoskeleton function that could be sensitive to ecological or environmental conditions. Calcite and crystallinity exhibit a similar body-part-specific pattern that correlates directly with the mechanical strength of the exoskeleton. Insights from this study suggest precise biological control of ACC chemistry in the to regulate exoskeleton properties. Laboratory measurements using quantitative methods and compositions that approximate the physiological conditions of crustaceans, demonstrate at least two types of ACC are formed by controlling Mg concentration and alkalinity. We also find temporal changes in the short-range ordering of ACC after precipitation that is dependent upon carbonate content. The findings from this study provide a quantitative basis for deciphering relationships between ACC structures, solution chemistry, and the final transformation products under biologically relevant conditions.

Establishing a physical and chemical framework for Amorphous Calcium Carbonate (ACC) biomineralization

Sebastian T. Mergelsberg

PUBLIC ABSTRACT

With the development of new imaging methods for nano-scale materials, scientists across diverse disciplines have recognized that many earth materials can form complex shapes by the formation and aggregation of nanocrystals or structureless (amorphous) particles. Biological minerals, such as shells and skeletons, are well-documented to form CaCO3 via both of these attachment pathways, particularly amorphous calcium carbonate (ACC). However, little is known about the ACC properties and the factors that determine the final composition of skeletal minerals. This three-part dissertation focuses on ACC and calcite in the of crustaceans to understand how form composite exoskeletons of calcium carbonate minerals. This knowledge is important because CaCO3 minerals are the primary component of the shells and skeletons of many economically important marine species. These minerals are also prevalent in the geological record as roadmaps for the evolutionary record. Amorphous and crystalline forms of CaCO3 are also used as inert 'filler' materials for pharmaceutical products. By designing a series of experiments to characterize ACC in exoskeletons from and , one part of the dissertation shows relationships between chemical composition and physical behavior of the materials. Building on this biomineral information, a separate experimental study synthesizes ACC under near-physiological conditions to show how amorphous CaCO3 forms under controlled conditions. The findings have far-reaching consequences for understanding the complex chemistry that underlies the formation of calcium carbonate as a component of shells and skeletons, and what physical properties are optimized by the composition of these materials.

TABLE OF CONTENTS

ABSTRACT ...... ii PUBLIC ABSTRACT ...... iii TABLE OF CONTENTS ...... iv LIST OF FIGURES ...... vii LIST OF TABLES ...... ix PREFACE ...... x CHAPTER 1. INTRODUCTION ...... 1 1.1 MOTIVATION FOR STUDY ...... 1 1.2 THE EXOSKELETON – PATTERNS IN COMPOSITION ...... 2 1.3 THE CRUSTACEAN EXOSKELETON – PHYSICAL PROPERTIES ...... 3 1.4 LITTLE IS KNOWN ABOUT ACC PROPERTIES ...... 4 1.5 THE TWO TYPES OF ACC ...... 5 1.6 ORGANIZATION OF THIS DISSERTATION ...... 5 1.7 REFERENCES ...... 7 CHAPTER 2. TOWARD AN UNDERSTANDING OF COMPOSITION SYSTEMATICS IN THE EXOSKELETON OF THE AMERICAN , HOMARUS AMERICANUS ...... 9 ABSTRACT ...... 9 2.1 INTRODUCTION ...... 10 2.2 MATERIALS AND METHODS ...... 14 2.2.1. and Sample Preparation...... 14 2.2.2. Dissolution of the organic and mineral fractions...... 15 2.2.3. Composition Analysis...... 15 2.2.4. Pair Distribution Function (PDF) analysis...... 16 2.3 RESULTS ...... 17 2.3.1. Mineral component is primarily ACC...... 17 2.3.2. Mineral fraction contains variable Mg, P, and Ca levels with constant ratios...... 18 2.3.3. Average composition of the mineral fraction...... 20 2.3.4. Low levels of Mg P, and Ca in the organic fraction...... 21 2.3.5. Bulk exoskeleton composition dominated by the mineral fraction...... 24 2.4 DISCUSSION ...... 24 2.4.1. Chemistry of mineral component is highly regulated...... 24 2.4.2. No evidence for a separate phase...... 26

iv

2.4.3. Enigma of a P/Mg signature...... 27 2.4.4. Broader pattern for multiple crustacean species? ...... 29 2.4.5. Physical basis for taphonomic bias in skeletal preservation...... 30 2.5 CONCLUSIONS ...... 31 2.6 REFERENCES ...... 33 CHAPTER 3. RELATIONSHIPS OF α-CHITIN AND COMPONENTS REVEAL BIOMINERALIZATION STRATEGIES OF CRUSTACEAN EXOSKELETONS ...... 37 ABSTRACT ...... 37 3.1 INTRODUCTION ...... 38 3.2 MATERIALS AND METHODS ...... 41 3.2.1 Animals ...... 41 3.2.2 Three-point flexural and mechanical analysis ...... 42 3.2.3 Raman spectroscopy ...... 43 3.2.4 Synchrotron X-ray Diffraction ...... 44 3.3 RESULTS AND DISCUSSION ...... 46 3.3.1. High-energy X-ray Diffraction Reveals Disparate Structural Trends ...... 46 3.3.2 Raman Spectroscopy Reveals Heterogeneity ...... 48 3.3.3 Steric strain is a proxy for chitin crystallinity ...... 50 3.3.4 Raman Spectroscopy independently measures chitin crystallinity ...... 51 3.3.5 Mechanical testing reveals chitin crystallinity controls flexural rigidity ...... 53 3.3.6 Comparison to other materials ...... 55 3.3.7 Ecological Basis for Exoskeleton Structure ...... 56 3.4 CONCLUSIONS AND IMPLICATIONS ...... 58 3.5 REFERENCES ...... 60 CHAPTER 4. ESTABLISHING THE MG-DEPENDENT SOLUBILITY AND LOCAL STRUCTURE(S) OF AMORPHOUS CALCIUM CARBONATE (ACC)...... 62 ABSTRACT ...... 62 4.1 INTRODUCTION ...... 63 4.2 MATERIALS AND METHODS: ...... 65 4.2.1 Experimental Design ...... 65 4.2.2 ACC synthesis and partial dissolution ...... 66 4.2.3 Isolation of ACC from suspension ...... 67 4.2.4 Titration ...... 67 4.2.5 Element analysis ...... 68

v

4.2.6 Thermogravimetric Analysis ...... 68 4.2.7 X-ray Diffraction experiments and Rietveld Refinement of transformed samples .... 68 4.2.8 Scanning Electron Microscopy ...... 69 4.2.9 Raman Spectroscopy ...... 69 4.2.10 Total X-Ray Scattering and Pair Distribution Function Analysis ...... 70 4.3 RESULTS AND DISCUSSION ...... 70 4.3.1 SEM reveals two independent morphologies ...... 70 4.3.2 TGA confirms presence of two amorphous products...... 72 4.3.3 In situ PDF analysis quantifies structural differences ...... 73

4.3.4 Evolution of ACC is sensitive to total aCO3/aCa ...... 75 4.3.5 Solubility is dependent on Mg content for both types of ACC ...... 77 4.3.6 Stoichiometry predicts solubility trend for each type of ACC ...... 79 4.3.7 Thermodynamic constraints on ACC solubility ...... 80 4.4 CONCLUSIONS ...... 81 4.5 REFERENCES ...... 83 APPENDIX A. SUPPLEMENTARY INFORMATION FOR CHAPTER 2 ...... 85 A1. SUPPLEMENTARY FIGURES ...... 85 A2. COMPOSITIONAL DATA...... 86 APPENDIX B. SUPPLEMENTARY INFORMATION FOR CHAPTER 3 ...... 89 B1. REFERENCES FOR X-RAY DIFFRACTION AND RAMAN SPECTROSCOPY..... 89 B2. SUPPLEMENTARY FIGURES ...... 90 APPENDIX C. MECHANICAL ANALYSIS OF CRUSTACEAN EXOSKELETONS ..... 96 C1. DEFINITION OF TERMS AND CONDITIONS ...... 96 C2. DATA ANALYSIS ALGORITHM ...... 98 C3. EXAMPLES OF MECHANICAL ANALYSES...... 105 C4. SUMMARY OF MECHANICAL DATA ...... 106 APPENDIX D. SUPPLEMENTARY INFORMATION FOR CHAPTER 4 ...... 108 D1. SUPPLEMENTARY FIGURES ...... 108 D2. DATA SUMMARY...... 113 D3. ESTIMATION OF SOLUBILITY VALUES FROM ENTHALPY VALUES ...... 116 D4. SUMMARY OF MEASURED AND CALCULATED VALUES FOR ALL EXPERIMENTS...... 117 APPENDIX E. CONSISTENT STRUCTURE OF ACC AT EACH RESIDENCE TIME 122

vi

LIST OF FIGURES

Fig. 2.1. The seven body parts of H. americanus investigated in this study...... 14 Fig. 2.2. Characteristic PDF analysis determined for reference standards and the bulk exoskeleton...... 17 Fig. 2.3. Analysis of the mineral fraction shows element concentrations are covariant ...... 19 Fig. 2.4. Molar ratios of elements contained in the mineral fraction of exoskeleton body parts are conserved ...... 22 Fig. 2.5. Composition analysis of the bulk exoskeleton...... 23 Fig. 2.6. Ratios of P/Ca and Mg/Ca are covariant in the lobster bulk exoskeleton and available exoskeleton data for other animals are compared to show a similar interspecies correlation ...... 28 Fig. 3.1 The four body parts investigated in this study ...... 39 Fig. 3.2 Calcite/ACC ratio vs. chitin crystallinity shows two different strategies of exoskeleton reinforcement ...... 45 Fig. 3.3 Raman spectroscopy reveals sample heterogeneity ...... 49 Fig. 3.4 Comparison to high-energy x-ray data identifies novel spectroscopic measure of chitin crystallinity ...... 51 Fig. 3.5 Exoskeleton flexural rigidity co-varies with an increase chitin crystallinity ...... 53 Fig. 3.6 Difference in calcite content between claw and cephalothorax as a function of species shows more mobile species exhibit smaller discrepancy in calcium carbonate crystallinity than more sessile animals ...... 56 Fig. 4.1 Morphology of ACC reveals two distinct particle sizes produced under different conditions...... 70 Fig. 4.2 TGA weight loss curves of 8 minute old samples confirm differences in morphology correspond to two different ranges of ACC composition...... 71 Fig. 4.3 In situ PDF profiles of ACC at high and low carbonate activities after 4 minutes residence time...... 73 Fig. 4.4 ACC evolves to an end-member with a more stable Mg-O short-range order over time ...... 75 Fig. 4.5 Solubility of ACC as a function of Mg content describes a strong linear dependence between 10 and 70 mol% MgCO3 ...... 77 Fig. 4.6 Comparison to previous study reveals solubility is controlled by the less metastable type of ACC...... 80

Fig. A1 Composition of the organic matrix fraction of the exoskeleton ...... 84 Fig. B1 High energy x-ray scattering of exoskeleton allow for the calculations of chitin crystallinity ...... 89 Fig. B2 PDF-determined abundance of calcite shows little correlation to sample thickness in 5 ...... 90 Fig. B3 Raman spectra of the crustacean exoskeletons from three species ...... 91 Fig. B4 Shifts in Amide I and III peak positions relative to differences in CaCO3 vibrational energies...... 92 Fig. B5 Flexural modulus is dependent on sample thickness ...... 93 Fig. B6 The correlation of common mechanical properties ...... 94 Fig. C1 Examples of mechanical analysis performed on all samples ...... 104

vii

Fig. D1 Characterization of solution and ACC composition reveals Mg sensitivity to chemical environment...... 107 Fig. D2 Expanded vertical scale PDF profiles show the short-range order of all conditions aligns over time...... 108 Fig. D3 Short-range order of solutions is distinctly different from in situ ACC...... 109 Fig. D4 Comparison of ACC types to other Ca and Mg reveals distinct short- range order of amorphous phases that feature composite atom pair distances. ...110 Fig. D5 Estimated dependence of apparent solubility on magnesium content for the two end-member compositions...... 111 Fig. E1 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition A ...... 121 Fig. E2 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition B ...... 122 Fig. E3 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 residence time, condition C...... 123 Fig. E4 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition D ...... 124 Fig. E5 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition E ...... 125 Fig. E6 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition H ...... 126 Fig. E7 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition I ...... 127 Fig. E8 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition J ...... 128 Fig. E9 Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 minute residence time, condition K ...... 129

viii

LIST OF TABLES

Table 2.1. Summary of studies reporting the P and Mg content of the and other decapod exoskeletons ...... 13 Table 3.1. Summary of all animals used in this study, sorted by defense strategy...... 55

Table A1 Summary of concentrations and molar ratios of Ca, Mg, and P in all body parts investigated in this study ...... 85 Table A2 Summary of previously reported Mg/Ca and P/Ca contained in the bulk exoskeletons of other crustaceans ...... 86 Table A3 Statistical analysis of linear regression models for each figure...... 87 Table B1 D-spacing references for chitin and calcite...... 88 Table B2 Raman peak reference table for common chitin and calcite vibrations ...... 88 Table C1 Definition of parameters used for the mechanical analyses...... 96 Table C2 Summary of all mechanical analyses performed for this study ...... 105 Table D1 TGA and ICP-OES analysis give overall compositions of ACC produced after 8- minute residence times...... 112 Table D2 Transformation experiments reveal ACC from the tested conditions produce a large number of polymorphs...... 113 Table D3 Chemical and calorimetric data from Radha et al., 2012...... 114 Table D4 Data for all ACC synthesis experiments ...... 116

ix

PREFACE

The dissertation is organized into four chapters. Chapter 1 provides an introduction to field of carbonate biomineralization and outlines the research objectives. The subsequent three chapters are presented as manuscripts in various stages of publication or preparation. Chapter 2 is presently submitted to Geochimica et Cosmochimica Acta. Chapters 3 and 4 are in advanced preparation to be submitted for publication to the Proceedings of the National Academy of

Sciences. All chemical composition, mechanical, and x-ray data are presented in a series of 5

Appendices (A-E).

x

CHAPTER 1. INTRODUCTION

1.1 MOTIVATION FOR STUDY

Recent advances in high-resolution analytical methods have brought a paradigm shift to the scientific community with the realization that many earth materials can form by multiple pathways. Evidence from laboratory syntheses and biological studies show calcium carbonate minerals nucleate and grow in many settings via non-classical pathways that involve metastable intermediates. The biominerals produced by a large

number of calcifying organisms are now known to begin as Amorphous Calcium

Carbonate (ACC) followed by transformation to calcite (Gong et al., 2012),

(Mass et al., 2017), or stabilized ACC (Hild et al., 2008). The large body of work on

classical growth suggests CaCO3 composition is greatly influenced by short-range

order in the crystal (Rodriguez-Blanco et al., 2014), mineral growth rate (Lemarchand et

al., 2004), and the interplay of trace element concentrations (Mucci et al., 1989). Despite the prevalence of many mineralization pathways in natural and synthetic systems, little is known about ACC reactivity and the evolution of mineral composition en route to final products.

The goal of this dissertation is to quantify the chemical and physical properties of natural ACC samples of crustacean exoskeletons with the objective of establishing the

role of amorphous intermediates as biominerals in highly controlled biological

environments. These measurements of exoskeleton biomineral properties provide the

basis for the final project in this dissertation – to quantify ACC properties in highly

controlled synthetic environments that are closely modeled after biological systems. In

addition to providing a better understanding of biological ACC, the findings from this

1 dissertation reconcile the apparent differences reported by the numerous in vitro observations of ACC evolution.

1.2 THE CRUSTACEAN EXOSKELETON – PATTERNS IN COMPOSITION

Often described as complex biocomposites, crustacean exoskeletons are composed of organic matrix (60-80%) and CaCO3 minerals (20-40%) as ACC and calcite

(Roer and Dillaman, 1984). The organic matrix is predominantly chitin with the remainder as (≈5-7%) and small organic molecules. The mineral fraction is ≈80 to

95% ACC and this metastable polymorph persists without transforming to crystalline products for an entire year between molting cycles (Hayes and Armstrong, 1961). A number of studies report the compositions of exoskeletons from Malacostraca animals, but a consistent picture of biomineral chemistry has not emerged (Addadi et al., 2003).

This lack of a chemical framework for interpreting, and thus predicting, exoskeleton composition has limited the usefulness of these animals as indicators of environmental and physiological conditions.

The lobster provides an ideal model for investigating biogenic ACC properties and relationships to the function of crustacean exoskeletons because of its large size and long intermolt period. Although the data are limited, this study measured the exoskeleton composition of the American lobster and found that specific elemental ratios are dependent upon the type of body part. There may also be an underlying relationship between Mg, P, and Ca for exoskeletons of multiple crustacean species. For example, the larger crustaceans (e.g. lobsters and crabs) generally show higher Mg and P levels than (Huner et al., 1976; Vijayan and Diwan, 1996). Insights from this

2 study combined with advances in understanding chemical controls on ACC stability

(Blue et al., 2017), led us to hypothesize that the lobster exoskeleton has body part- specific compositions that may reflect broader systematics. The distinctive patterns in the

biomineral composition of the American lobster raised two questions: 1) do the chemical

differences between body parts relate to physical properties? and 2) Do similar

systematics occur in the exoskeletons from other crustacean species?

1.3 THE CRUSTACEAN EXOSKELETON – PHYSICAL PROPERTIES

The second investigation focused on determining the structure and flexural

rigidity of exoskeletons from multiple crustacean species. The large crustaceans of the

order Malacostraca – relatives of the lobster – exhibit extreme morphological and functional differences in their exoskeleton and thus are ideal model organisms (Raabe et al., 2006; Boßelmann et al., 2007). Using high-energy x-ray scattering and Raman spectroscopy, we determined two separate strategies to reinforce crustacean cuticle using the structure of inorganic and organic components. The crystallinities of chitin and calcium carbonate show a distinct co-variance for exoskeletons, whereas the lobster cuticle is relatively constant in terms of structure, despite the differences in chemistry that

I identified in the first study.

Measurements of the exoskeleton rigidity reveal these two strategies of

reinforcement produce similar results in homologous body parts. Ultimate toughness of

the exoskeleton, however, is proportional primarily to chitin crystallinity. Other publications have suggested chitin to be the primary component involved in cuticle reinforcement due to its structural variability (Al-Sawalmih et al., 2008). This study

3 proposes that biomineralization is likely the primary strength-conferring structure in the crustacean exoskeleton, but calcium carbonate crystallinity is correlated with species-specific differences and reinforcement strategies. The variation in exoskeleton physical properties identified by the second study, along with compositional patterns established for the lobster exoskeleton, suggest a species-specific optimization of biomineral formation and chemical/physical properties. To establish a basis for calcium carbonate synthesis in crustacean exoskeletons, however, a quantitative framework of

ACC properties at near-physiological conditions was needed.

1.4 LITTLE IS KNOWN ABOUT ACC PROPERTIES

The third investigation focused on establishing the properties of synthetic ACC under controlled conditions. A major challenge in measuring evolution of chemical and physical properties of ACC is the precise control over reaction conditions (Wang et al.,

2012; Blue and Dove, 2015). Numerous qualitative studies have been reported, but there is currently no clear or framework for understanding the relationship between initial conditions and final ACC products. In vitro studies of amorphous intermediates have shown ACC to evolve by multiple distinct forms (Lam et al., 2007; Radha et al., 2010;

Cartwright et al., 2012). Most of these studies have identified differences in hydration state and ACC metastability between the different amorphous intermediates that are observed in both biological and synthetic samples.

Results from synthetic ACC studies suggest an additional type of amorphous carbonate exists under high-Mg conditions that contains greater amounts of Mg than Ca

(Radha et al., 2012; Rodriguez-Blanco et al, 2015). Further quantitative understanding of

4 this complex energetic landscape and the associated chemical and physical properties of

ACC requires synthetic environments that are easily controlled and modeled closely after conditions identified at sites of biomineralization.

1.5 THE TWO TYPES OF ACC

Using an improved synthesis method that controls magnesium and carbonate chemistry in parallel with high-resolution analyses (Blue et al., 2017), this study confirmed the presence of two types of ACC. Further, we established and quantified the morphology, composition, structure, and solubility of both types of ACC. Measurements show that most conditions produce ACC that evolves over the course of several minutes to form heterogenous mixtures of the two different types of ACC. However, bulk solubility was shown to be controlled by one type of ACC up to very high magnesium concentrations. Results from this study provide a quantitative basis for deciphering relationships between ACC structures, solution chemistry, and the final transformation products in the presence of Mg.

1.6 ORGANIZATION OF THIS DISSERTATION

This dissertation addresses three core objectives:

1. Establish composition systematics in the exoskeleton of the American lobster,

Homarus americanus. Relationships of Mg, Ca, and P distribution suggest a

means of optimizing composition to meet functional requirements of the mineral

fraction in exoskeletons. The findings suggest lobsters hold promise as a novel

class of animals that record composition systematics within their CaCO3

5 biominerals and similar trends are noted across species in later decapods and

malacostracans. Resulting trends and implications are described in Chapter 2 and

the data are provided in Appendix A. This manuscript is currently submitted to

Geochimica et Cosmochimica Acta.

2. Reveal disparate strategies for the biomineralization of crustacean

exoskeletons via α-chitin and carbonate mineral crystallinity. This study

builds on the previous investigation by quantifying the relationship between

biomineral structure and mechanical properties for several crustacean species. The

findings and implications suggest a broad relationship between

ecology/evolutionary history and exoskeleton reinforcement strategies. This study

is presented in Chapter 3 and the corresponding data are summarized in

Appendices B and C. The manuscript is currently in preparation for publication.

3. Determine the structure(s) and solubility of Amorphous Calcium

Carbonate(s) (ACC) under controlled conditions. Using chemical parameters

relevant to crustacean biomineralization, experiments from this study reveal two

distinct types of ACC are formed under controlled conditions of Mg/Ca and

CO3/Ca. The two types exhibit different morphologies, composition ranges, short-

range ordering, and lifetimes. However, solubility is controlled by the metastable

equilibrium between the two phases, as described in Chapter 4. The data are

provided in Appendices D and E. This manuscript is currently in preparation for

publication.

6

1.7 REFERENCES

Addadi L., Raz S. and Weiner S. (2003) Taking advantage of disorder: Amorphous calcium carbonate and its roles in biomineralization. Adv. Mater. 15, 959–970. Al-Sawalmih A., Li C., Siegel S., Fabritius H., Yi S., Raabe D., Fratzl P. and Paris O. (2008) Microtexture and Chitin/Calcite Orientation Relationship in the Mineralized Exoskeleton of the American Lobster. Adv. Funct. Mater. 18, 3307–3314. Blue C. R. and Dove P. M. (2015) Chemical controls on the magnesium content of amorphous calcium carbonate. Geochim. Cosmochim. Acta 148, 23–33. Blue C. R., Giuffre A., Mergelsberg S., Han N., De Yoreo J. J. and Dove P. M. (2017) Chemical and physical controls on the transformation of amorphous calcium carbonate into crystalline CaCO 3 polymorphs. Geochim. Cosmochim. Acta 196, 179–196. Boßelmann F., Romano P., Fabritius H., Raabe D. and Epple M. (2007) The composition of the exoskeleton of two crustacea: The American lobster Homarus americanus and the edible crab . Thermochim. Acta 463, 65–68. Cartwright J. H. E., Checa A. G., Gale J. D., Gebauer D. and Sainz-Díaz C. I. (2012) Calcium carbonate polyamorphism and its role in biomineralization: How many amorphous calcium carbonates are there? Angew. Chemie - Int. Ed. 51, 11960– 11970. Gong Y., Killian C., Olson I., Appathurai N. P., Amasino A., Martin M., Holt L., Wilt F. and Gilbert P. (2012) Phase transitions in biogenic amorphous calcium carbonate. Proc. Natl. Acad. Sci. U. S. A. 109, 6088–6093. Hayes D. K. and Armstrong W. D. (1961) The Distribution of Mineral Material in the Calcified Carapace and Claw Shell of the American Lobster, Homarus americanus, Evaluated by Means of Microroentgenograms. Biol. Bull. 121, 307–315. Hegdahl T., Gustavsen F. and Silness J. (1978) The Structure and Mineralization of the Carapace of the Crab (Cancer pagurus L.) 2. The exocuticle. Zool. Scr. 6, 101–105. Hild S., Marti O. and Ziegler A. (2008) Spatial distribution of calcite and amorphous calcium carbonate in the cuticle of the terrestrial crustaceans and Armadillidium vulgare. J. Struct. Biol. 163, 100–108. Huner J. V., Kowalczuk J. G. and Avault J. W. (1976) Calcium and Magnesium Levels in the Intermolt (C4) Carapaces of Three Species of Freshwater Crawfish (Cambarida: Decapoda). Comp. Biochem. Physiol. 55A, 183–185. Lam R. S. K., Charnock J. M., Lennie A. and Meldrum F. C. (2007) Synthesis-dependant structural variations in amorphous calcium carbonate. CrystEngComm 9, 1226. Lemarchand D., Wasserburg G. J. and Papanastassiou D. A. (2004) Rate-controlled calcium isotope fractionation in synthetic calcite. Geochim. Cosmochim. Acta 68, 4665–4678. Mass T., Giuffre A. J., Sun C.-Y., Stifler C. A., Frazier M. J., Neder M., Tamura N., Stan

7 C. V., Marcus M. A. and Gilbert P. U. P. A. (2017) Amorphous calcium carbonate particles form skeletons. Proc. Natl. Acad. Sci., 201707890. Mucci A., Canuel R. and Zhong S. (1989) The solubility of calcite and aragonite in sulfate-free and the seeded growth kinetics and composition of the precipitates at 25°C. Chem. Geol. 74, 309–320. Raabe D., Romano P., Sachs C., Fabritius H., Al-Sawalmih A., Yi S. B., Servos G. and Hartwig H. G. (2006) Microstructure and crystallographic texture of the chitin- protein network in the biological of the exoskeleton of the lobster Homarus americanus. Mater. Sci. Eng. A 421, 143–153. Radha A. V., Fernandez-Martinez A., Hu Y., Jun Y.-S., Waychunas G. A. and Navrotsky A. (2012) Energetic and structural studies of amorphous Ca1−xMgxCO3·nH2O (0 x 1). Geochim. Cosmochim. Acta 90, 83–95. Radha A. V., Forbes T. Z., Killian C. E., Gilbert P. U. P. A. and Navrotsky A. (2010) ⩽ ⩽ Transformation and crystallization energetics of synthetic and biogenic amorphous calcium carbonate. Proc. Natl. Acad. Sci. 107, 16438–16443. Rodriguez-Blanco J. D., Shaw S., Bots P., Roncal-Herrero T. and Benning L. G. (2014) The role of Mg in the crystallization of monohydrocalcite. Geochim. Cosmochim. Acta 127, 204–220. Rodriguez-Blanco et al (2015) A Route For The Direct Crystallization Of Dolomite. Am. Minerol. 100, 1172–1181. Roer R. and Dillaman R. (1984) The Structure and of the Crustacean Cuticle. Am. Zool. 24, 893–909. Vijayan K. K. and Diwan A. D. (1996) Fluctuations in Ca, Mg and P levels in the hemolymph, muscle, midgut gland and exoskeleton during the moult cycle of the Indian white , Penaeus indicus (Decapoda: Penaeidae). Comp. Biochem. Physiol. - A Physiol. 114, 91–97. Wang D., Hamm L. M., Giuffre A. J., Echigo T., Rimstidt J. D., Yoreo J. J. De, Grotzinger J., Dove P. M., De Yoreo J. J., Grotzinger J. and Dove P. M. (2012) Revisiting geochemical controls on patterns of carbonate deposition through the lens of multiple pathways to mineralization. Faraday Discuss. 159, 371.

8

CHAPTER 2. TOWARD AN UNDERSTANDING OF COMPOSITION SYSTEMATICS IN THE EXOSKELETON OF THE AMERICAN LOBSTER, HOMARUS AMERICANUS

Sebastian T. Mergelsberg, Robert N. Ulrich, Shuhai Xiao, Patricia M. Dove (submitted) Geochimica et Cosmochimica Acta.

ABSTRACT

Exoskeletons are unique biocomposites that provide protection and stability. Previous studies of these structures for lobsters and other crustaceans report variable compositions for disparate body parts but systematic relationships have not been established. Anecdotal evidence, however, suggests underlying patterns may exist. We test this idea by designing a protocol to separately extract the mineral (amorphous calcium carbonate plus calcite) and organic (chitin plus protein) fractions of the exoskeleton. The fractions were analyzed by wet chemistry methods to quantify Mg, Ca, and P contents of the bulk, mineral, and organic matrix. Applying this approach to the exoskeleton for seven body parts of the American lobster, Homarus americanus, we quantify the chemical composition of each fraction. The measurements show Mg, P, and Ca concentrations in lobster exoskeletons are highly variable as previously reported. However, the ratios of Mg/Ca and P/Ca in the mineral fraction are constant for all parts, except the chelae (claws), which are offset to higher values. By normalizing concentrations to obtain P/Ca and Mg/Ca, we show that all body parts conserve P/Mg to 1.27±0.30. The findings suggest lobsters hold promise as a novel class of animals that record composition systematics within their CaCO3 biominerals. Parallel structural analyses of the bulk samples confirm a large proportion of amorphous calcium carbonate relative to calcite in the mineral fractions for each body part using high-energy X-ray diffraction and PDF analysis. There is no evidence for a phosphate phase. We compare our findings to previously reported compositions for other marine (crab, lobster, and marine shrimp) and terrestrial (pillbug) crustaceans. Although the available data are limited, the published measurements indicate similar Mg/Ca and P/Ca patterns. The relationships suggest a means of optimizing composition to meet functional requirements of the mineral fraction in exoskeletons. Solubility differences associated with compositional variability may provide a thermodynamic basis for the taphonomic bias observed in the record.

9 2.1 INTRODUCTION

The American lobster (Homarus americanus) is a large decapod crustacean with a close evolutionary relationship to crabs, both of which belong to the class Malacostraca.

All malacostracans have an exoskeleton that is molted at regular intervals. Often described as a complex biocomposite, the exoskeleton is composed of organic matrix

(60-80%) and CaCO3 minerals (20-40%) as amorphous calcium carbonate and calcite

(Roer and Dillaman, 1984). In this study, we collectively refer to the amorphous and crystalline phases of calcium carbonate as the mineral component. The organic matrix is predominantly chitin with the remainder as protein (≈5-7%) and small organic molecules.

The primary function of the organic matrix is to provide a framework during the intermolt period while the calcium carbonate is postulated to add rigidity and the ability to withstand higher impact forces. These phases may have additional functions such as maintaining osmotic balance or inhibiting infection (Glynn, 1968; Greenaway, 1985;

Kunkel et al., 2012). Compared to other that use chitin exclusively, such as the and chelicerates, crustaceans can grow larger, highly diverse body structures including claws, and swimming tails. The consistent constitution of the exoskeleton, despite variable body size, morphology, and function across crustacean species (Roer and

Dillaman, 1984), suggests a highly regulated biomineralization process.

With the realization that many taxa utilize amorphous calcium carbonate (ACC) during biomineralization, the lobster exoskeleton is a biomineral of particular interest.

The mineral fraction is ≈80 to 95% ACC and this metastable polymorph persists without

transforming to crystalline products for an entire year between molting cycles (Hayes and

Armstrong, 1961). In contrast, most animals use ACC during biomineralization as a

10 short-lived intermediate that transforms within a few hours to days (Aizenberg et al.,

1996; Addadi et al., 2003). With its large size and long molting cycles, the lobster thus provides an ideal model organism to determine biogenic ACC properties and relationships to the function of crustacean exoskeletons.

Previous studies that investigated origins of ACC metastability have focused on

Mg and P (from phosphate) as elements that are known to modify CaCO3 properties and

are abundant in biological systems (Levi-Kalisman et al., 2002; Akiva-Tal et al., 2011;

Sato et al., 2011). The ability of Mg to increase the lifetime of ACC is well documented

(Loste et al., 2003; Blue et al., 2017), but the role of P is not well understood. In vitro

studies show P stabilizes ACC and slows its transformation into crystalline polymorphs

for several hours (Reddy, 1977; Clarkson et al., 1992; Bentov et al., 2010). In contrast,

ACC transforms to crystalline products within seconds to minutes in pure CaCO3 compositions (Sawada, 1997; Blue et al., 2017). To achieve the longevity of ACC that is observed in many crustaceans, these observations suggest both Mg and P have stabilizing roles.

ACC can also be stabilized by interactions with biomacromolecules (Aizenberg et al., 1996). Traditionally, the scientific community has assumed the aspartic and glutamic acid sidechains of are the primary moieties involved in lowering the energetic barrier to ACC nucleation (Addadi et al., 2006). Recent work, however, shows that other functional groups may also promote ACC precipitation and prevent particle dissolution

(Arias and Fernandez, 2008; Qi et al., 2014). An interplay of cations and macromolecules to stabilize ACC is also plausible. For example, Mg is often involved in ionic stabilization of protein structures (Glusker, 1991), while phosphate groups are common

11 and often covalently attached to serine and threonine sidechains (Bentov et al., 2010).

These observations raise the question whether P is primarily associated with the mineral

or organic components of the exoskeleton. It is possible that P is concentrated in the

protein component, yet still able to influence the stability of mineral fraction.

A number of studies report the compositions of exoskeletons from Malacostraca

animals (Table 2.1), but a consistent picture of biomineral chemistry has not emerged.

This lack of a chemical framework for interpreting, and thus predicting, exoskeleton

properties has limited the usefulness of these animals as indicators of environmental and

physiological conditions. There are several obstacles to advancing this knowledge. First,

most studies report the composition of the bulk exoskeleton, an average of contributions

from mineral and organic constituents. It is possible that the large organic fraction masks

unrecognized systematics in mineral composition. To our knowledge, one study reports

compositions of the separated mineral fraction (ACC + calcite) for the lobster

cephalothorax and shows Mg and P are concentrated in the ACC component (Levi-

Kalisman et al., 2002). This raises the second point that most investigations are limited to

specific body parts, not all of which are homologous amongst all crustaceans. Third, most

studies did not quantify concentrations of all three elements. Finally, many studies of

macrofauna necessarily use small sample sizes to control for variations in water

temperature, diet, etc. (e.g. Vigh and Dendinger, 1982; Boßelmann et al., 2007; Al-

Sawalmih et al., 2009).

Although the data are limited, Table 2.1 contains anecdotal evidence that exoskeleton compositions are specific to the type of body part and there may be an underlying relationship between Mg, P, and Ca. For example, the larger crustaceans (e.g.

12 lobsters and crabs) generally show higher Mg and P levels than shrimp. These published

bulk measurements, combined with advances in understanding chemical controls on ACC

stability, lead us to hypothesize that the lobster exoskeleton has body part-specific

compositions that may reflect broader systematics. To test this idea, we designed a

protocol to separately extract the organic and mineral fractions of seven body parts.

These fractions were analyzed by wet chemistry methods to quantify the Mg, Ca, and P

content of the bulk, mineral, and organic matrix. A parallel structural analysis of the

mineral samples determined the proportions of ACC and calcite.

Table 2.1. Summary of studies reporting the P and Mg content of the American lobster and other decapod exoskeletons. Organism Bulk Mg Bulk P Body Reference and notes wt% wt% parts American Lobster, 1.46% 2.41% Chela Clarke and Wheeler, 1922 (a) H. americanus 1.11% 1.65% Cthx Clarke and Wheeler, 1922 (a) N/A 0.88 –1.82% Cthx Travis, 1963 (b) 10%* 12%* Cthx Levi-Kalisman et al., 2002 (c) 1.2% 0.65% Chela Boßelmann et al., 2007 (d) 1.0% 2.1% Cthx Boßelmann et al., 2007 (d) 0.60% 0.99% Chela This study 0.37% 0.52% Cthx This study 0.45% 0.67% This study 0.47% 0.63% Ceph This study 0.39% 0.62% Abd This study 0.33% 0.57% Upd This study Blue crab, 2.2% N/A Cthx Vigh and Dendinger, 1982 C. sapidus Red rock crab, 1.4% 0.9% Chela Boßelmann et al., 2007 (e) C. pagurus 1.2% 1.1% Cthx Boßelmann et al., 2007 (e) California brown shrimp, 1.23% 1.93% Rost Huner et al., 1979 (f) P. californiensis 0.99% 1.86% Cthx Huner et al., 1979 (f) 0.61% 1.32% Abd Huner et al., 1979 (f) Indian white prawn, 1.2% 1.2% Av. Vijayan and Diwan, 1996 (g) P. indicus 2.8% <1.2% Rost Vijayan and Diwan, 1996 (g) 0.4% >1.2% Swrt Vijayan and Diwan, 1996 (g) , 0.71% 0.69% Cpce Becker et al., 2005 P. scaber Pillbug, 0.72% 0.59% Cpce Becker et al., 2005 A. vulgare (a) Averages for 3 animals; (b) Highest and lowest concentrations recorded throughout molt cycle; (c) ACC fraction only; (d) Highest concentration of P found in cephalothorax; (e) Red rock crab, C. pagurus. This species has a very thick exoskeleton; (f) Most comprehensive bulk and mineral data; (g) Quantified elemental distribution across body parts during the molt cycle. Cthx: cephalothorax, Ceph: cephalon, Abd: abdomen, Upd: uropod, Rost: rostrum, Av: average, Swrt: swimmerets, Cpce: carapace. See Table A1, Appendix A for average concentrations for each body part.

13 2.2 MATERIALS AND METHODS

2.2.1. Animal and Sample Preparation.

Live specimens of H. americanus used in this study were captured at the same

location by LobsterGuy (Point Judith, Rhode Island), shipped to our laboratory, and flash

frozen for 30 minutes at -80°C within one hour of receipt. The animals were adult males

in the C4 stage of intermolt and collected during early summer. Sample preparation

began by thawing both animals at room temperature for ≈2 hours and dissecting them in

air. Exoskeleton samples were prepared from two animals in sets of three (sampling

triplets) for each body part of interest (Fig. 2.1), a total of 6 samples for each body part.

Figure 2.1. The seven body parts of H. americanus investigated in this study. Three or more samples were collected from each location while also distinguishing between the distal and proximal sites on the chela for a total of 30 exoskeleton samples per animal.

Each of these samples was subsequently analyzed in triplicate to give a total of 18 measurements per body part. The exoskeleton samples were then rinsed for 5 s in 18.2

MΩ·cm ultrapure water and methanol and dried in air. When processing occurred on another day, the exoskeleton samples were put into sealed polystyrene containers and stored in a freezer at -20°C. Each sample was frozen in liquid nitrogen and ground to a homogenous powder in a mortar and pestle in preparation for component dissolutions.

14 2.2.2. Dissolution of the organic and mineral fractions.

The experimental design used three extraction procedures to isolate the mineral (ACC

plus calcite) fraction from the organic (separate chitin and protein) fraction. To dissolve

the mineral fraction, each bulk powder was individually weighed and decalcified in 10%

acetic acid at 4°C for 24 hours. The supernatant was collected via vacuum filtration on

polyethersulfone (PES) filters (< 30 nm; Sterlitech, Kent, WA) and stored at 4°C until

analysis. The remaining solids were rinsed in 18.2 MΩ·cm ultrapure water for 25

seconds, then rinsed for 25s in methanol, dried in air, and weighed.

The water-soluble proteins in the exoskeleton were dissolved by adding the residual

decalcified sample to a solution of 6 M urea and 0.06 M Tris at pH 6.3. Dissolution

occurred for 24 hours at 4˚C, followed by separation by vacuum filtration. Each sample

was collected and stored at 4°C for later analysis. Remaining solids were rinsed in 18.2

MΩ·cm ultrapure water for 25 s, for 25 s with methanol, dried in air, and weighed.

In a final step, the powders were added to a 10 M HCl solution for a minimum of 24

hours to dissolve the chitinous samples completely. Bulk concentrations were calculated

using the sum of elements removed by each dissolution procedure divided by the initial

weight of each sample. Element concentrations are reported only for the total organic

matrix because of the analytical errors associated with measurements of the small amount

of protein relative to chitin. Table A1, Appendix A summarizes the wt% of the mineral

fraction for each body part of the exoskeleton.

2.2.3. Composition Analysis.

Solutions from the three separation procedures were analyzed via ICP-OES to measure the total concentrations of Ca, Mg, and P in the collected solutions. Analyses

15 were performed using a Spectro® ARCOS SOP instrument using an yttrium internal standard. Reference solutions were prepared independently to exclude impurities from the solvents used in each extraction. For each sample, the resulting solutions were split into thirds to provide instrumental triplets, such that 18 measurements were made for each body part. Detection limits for Ca and Mg were 0.033 mg/L and 0.079 mg/L for P.

Error estimates for these analyses are reported as the 1σ standard error from triplicate analyses.

2.2.4. Pair Distribution Function (PDF) analysis.

The predominance of ACC in the mineral fraction (Wood and Russell, 1987) was confirmed by characterizing the crystallinity using x-ray diffraction at the Advanced

Photon Source (Argonne, IL USA) using beamline 11-ID-B (Rütt et al., 2001) and an incident photon energy of ~58.6 keV (λ = 0.2114 Å). Samples of bulk exoskeleton were mounted between two opposing pieces of polyimide (Kapton®) tape and the scattered radiation was measured in transmission mode using an amorphous-Si detector system manufactured by Perkin ElmerTM (2048 × 2048 pixels, 200 × 200 μm2 pixel size). A

CeO2 standard (NIST diffraction standard set 674a) was used to calibrate the sample-to- detector distance and the non-orthogonality of the detector relative to the incident beam path. Conversion of scattering data from 2D to 1D was performed using the program

Fit2D (Hammersley et al., 1995). A polarization correction was applied during integration of the data. Direct subtraction of the sample holder was accomplished by the independent measurement of the true background intensity of a blank polyimide tape.

PDF analysis profiles were calculated using PDFgetX2 (Qiu et al., 2004). The proportion

16 of crystalline and amorphous calcium carbonate as calcite and amorphous calcium

carbonate (ACC) were determined using linear analysis to fit synthetic standards (Michel

et al., 2008) to sample PDF profiles.

2.3 RESULTS

2.3.1. Mineral component is primarily ACC.

PDF analyses of the high-energy x-ray scattering data confirm the mineral fraction of the exoskeleton contains ACC and calcite in variable proportions between body parts (Hayes and Armstrong, 1961; Al-Sawalmih et al., 2009). For example,

Figure 2.2. Characteristic PDF analysis determined for the reference standards (Michel et al., 2008) and the bulk fraction of the cephalothorax and dominant chela. The similar profiles of ACC and the dominant chela indicate the predominance of ACC. Linear analysis of the PDF data estimate the mineral component of the cephalothorax and dominant claw is ≈85% and ≈97% ACC respectively, with the remainder as calcite.

17

the cephalothorax is ≈86% ACC in contrast to the chelae with ≈97% ACC. The short-

range order of the biogenic ACC is consistently ≈7Å and calcite particle diameters are

well above 1.5 nm (Fig. 2.2). There is no structural evidence for a separate phosphate or other amorphous or crystalline phases in any of the body parts (see later discussion).

2.3.2. Mineral fraction contains variable Mg, P, and Ca levels with constant ratios.

The Mg concentration in the mineral fraction (ACC plus calcite) varies across the

exoskeleton with differences of up to ≈5X between the seven body parts (Fig. 2.3A). The legs and abdomen exhibit the largest variability, while the uropods and cephalothorax have relatively uniform compositions. Despite differences in total concentrations, Mg and

Ca are covariant across all body parts, except the chelae, by the relation:

(2.1A)

The near zero intercept of Eq. 2.1A allows us to assume the slope of 0.087 is approximately equal to the Mg/Ca of the exoskeleton for all body parts. To put this into perspective, the seawater habitat of the lobster has Mg/Ca ≈ 3 (by weight). Chelae compositions are offset to greater Mg concentrations to give:

(2.1B)

where R2 =0.896 and n=12 (e.g., Table A1, Appendix A). Statistics for these

relationships and the entire body are reported in Table A3, Appendix A.

18

Figure 2.3. Analysis of the mineral fraction shows element concentrations are covariant; chelae compositions are offset to higher wt% Mg and wt% P. (A) wt% Ca vs. wt% Mg; (B) wt% Ca vs. wt% P; (A) wt% Mg vs. wt% P. Regression fits are determined for the data from all body parts except the chelae. Detailed statistics are presented in Table A3, Appendix A.

Total phosphorous concentration is also variable between body parts, but again, P levels are covariant with Ca (Fig. 2.3B). For all body parts except the chela, the relationship between P and Ca concentrations is given by:

(2.2)

19 The phosphorous contents of the chelae are offset to higher values by ≈30% (for three

claw samples) without an apparent trend (Table A1 and Table A3, Appendix A); except for greater P levels measured in the distal part of the claw (toward the tip). P concentrations in the proximal parts of both claws are similar to other body parts.

As expected from Fig. 2.3A and 2.3B, the relationship between P and Mg in Fig.

2.3C is conserved for all body parts, except the chelae:

(2.3A)

A similar dependence is determined for the ensemble of all body parts:

(2.3B)

where R2 =0.825 and n=30. To our knowledge, a P/Mg correlation is not previously

reported in carbonate biominerals, but the statistical analyses show high certainty with

very low standard errors and p-values (Table A3, Appendix A). By themselves, the

chelae do not exhibit a clear relationship between Mg and P.

2.3.3. Average composition of the mineral fraction.

Recognizing that the biomineral is a polymorphic mixture of ACC and calcite that

cannot be physically separated, we estimate an average composition. The mineral fraction

in the cephalothorax exhibits a solid-solution with the average composition:

(2.4A)

Chelae (for both claws) contain more Mg and P relative to Ca as given by:

(2.4B)

These compositions were determined using all of the mineral fraction analyses in this

study. We also assume: (1) all measured elements are associated with a single phase, (2)

20 mineral composition is charge-balanced by substitution of expected monovalent cations

+ + + 3- (Na , K , H ), and (3) P occurs as PO4 . The speciation of anions and concentrations of

monovalent cations were not measured, thus the of the charge-balance substitution

is unknown. Previous studies assume the presence of hydroxide (Becker et al., 2005;

Kunkel et al., 2012), but a high concentration of this anion seems unlikely given that

local conditions during exoskeleton deposition are not expected to exceed pH 9 (Ziegler,

2008).

We assume the amorphous polymorph contains a greater proportion of Mg and P

relative to the calcite (Dai et al., 2008; Sato et al., 2011; Blue et al., 2017). This is

consistent with structural (Fig. 2.2) and chemical (Fig. 2.3AB) data determined for the

chelae versus cephalothorax.

2.3.4. Low levels of Mg P, and Ca in the organic fraction.

The organic matrix contains very low concentrations of Mg, P, and Ca compared

to the mineral fraction (compare axis values in Fig. A1, Appendix A and Fig. 2.3ABC).

All linear regression models show a statistically significant correlation between element concentrations (p<0.05), but the low Mg values in the organic fraction are within the error of the larger concentrations measured for the mineral fraction (Fig A1A and analysis in Table A3, Appendix A). Similarly, the organic portion of most exoskeleton parts contains low wt%P (less than 0.1, Fig A1B, Appendix A), which also has a minor impact on bulk composition relative to the mineral fraction. The very high ratios of P/Ca and P/Mg in the organic fraction are likely due to contributions by phosphorylated peptides.

21

Figure 2.4. Molar ratios of elements contained in the mineral fraction of exoskeleton body parts are conserved relative to the wide range of ratios determined for the corresponding organic fraction. Dashed lines give average values of (A) (Mg/Ca)mineral = 0.084 ±0.001; (B) (P/Ca)mineral = 0.093 ±0.003; and (C) (P/Mg)mineral =1.110 ±0.030. Gray bands illustrate the 2σ standard error (95% confidence interval). Detailed statistics are given in Tables A1 and A3, Appendix A.

The absence of a distinct composition signature in the organic fraction is illustrated in Fig. 2.4A. Mg/Ca ratios determined for the organic matrix are widely

22

Figure 2.5. Composition analysis of the bulk exoskeleton. (A) wt% Ca vs. wt% Mg; (B) wt% Ca vs. wt% P; (C) wt% Mg vs. wt% P. The 1:1 relationship between element ratios in the mineral fraction versus bulk exoskeleton demonstrates bulk composition is dominated by the contributions from the mineral. (D) Mg/Ca; (E) P/Ca; and (F) Mg/P. Detailed statistics are given in Table A3, Appendix A.

dispersed relative to the narrow range of values in the mineral fraction (gray band). We acknowledge the disparate values obtained for the organic fraction include analytical errors associated with measurements near the detection limit of our analysis. The P/Ca

23 (Fig. 2.4B) and P/Mg (Fig. 2.4C) ratios show similarly wide dispersions for the organic fraction compared to the mineral.

2.3.5. Bulk exoskeleton composition dominated by the mineral fraction.

Although the organic fraction dominates the volume and mass of the bulk

exoskeleton, bulk composition trends mimic those of the mineral fraction (compare Fig.

2.5ABC and Fig. 2.3ABC). Similarities include offsets of the chelae to higher Mg and P

concentrations. This is supported by the conserved Mg/Ca and P/Ca ratios in the mineral

fraction of the chelae and the rest of the body, respectively (Fig. 2.4AB). Absolute

concentrations are lower due to the ‘dilution’ effects of the large organic fraction. Our

finding that the mineral fraction is the driver to bulk composition is illustrated by

comparing the normalized elemental ratios of the bulk and mineral components (Fig.

2.5DEF). All ratios exhibit a 1:1 dependence with high statistical confidence (Table A3,

Appendix A). The findings confirm the assumption of previous studies that bulk

exoskeleton composition can be used to infer biomineral composition for the elements

investigated herein. However, element ratios are more useful as indicators of composition

for bulk and mineral fractions because of the variable concentrations that occur both

within and between body parts.

2.4 DISCUSSION

2.4.1. Chemistry of mineral component is highly regulated.

Our finding that Mg/Ca and P/Ca are conserved in the exoskeleton (Fig. 2.4AB)

is consistent with previous studies implicating these ions in optimizing CaCO3 properties.

24 The data also indicate the lobster joins other biomineralizing animals, such as

and mollusks in their ability to produce characteristic biomineral

compositions. For example, the Mg/Ca ratio is well-established for shells of many

bivalves (Klein et al., 1996; Freitas et al., 2006).

There are three plausible explanations for why regulating Mg and P

concentrations confers physical and chemical advantages to exoskeleton function. First,

studies show that increasing concentrations of Mg and P increase the lifetime of the

metastable amorphous polymorph optimize proportions of ACC to calcite for the duration

of the molt cycle. Also, in vitro studies show Mg and P also progressively modify ACC

morphology (Xu et al., 2005; Blue and Dove, 2015). In contrast to the more uniform size

and shape of impurity-free ACC particles, incorporation of Mg or P produces more

variably-sized aggregated spheroids (Xu et al., 2005; Blue and Dove, 2015). It is

plausible that ACC allows for greater morphological and functional diversity within the

composite material.

Second, ACC is proposed to function as a chemical buffer (Kunkel et al., 2012).

The exoskeleton has a critical role in maintaining the osmotic and ionic balance between

the haemolymph of the animal and sea water (Glynn, 1968; Greenaway, 1985). An abundance of ACC particles may maintain this balance throughout the molt cycle. As a metastable phase, ACC also provides readily accessible buffering capacity in the event of exoskeletal damage (Kunkel et al., 2012).

Finally, previous studies suggest Mg and P promote bonding with the organic matrix to increase physical strength (Neues et al., 2007; Cribb et al., 2009). For example,

P distribution at the submicron scale through the ACC may reflect differences in protein

25 identity, degree of crosslinking, and the presence of glycoproteins (Marlowe et al., 1994;

Coblentz et al., 1998). The skeletal structures of diverse animals show evidence for impurity-enhanced physical properties such as the Sr-rich aragonite in claws of the blue crab (38) and Mg and P enriched carbonate mineral(s) from the puparial cuticle of the fly species, M. autumnalis and M. domestica (Grodowitz et al., 1987). In the latter example,

Mg and P are thought to partially replace the sclerotization of organic cuticle components

(Roseland et al., 1985). These studies suggest high concentrations of Mg and P are associated with body parts that require additional reinforcement. Other investigations report that greater levels of Mg, P, and other trace elements are found in high stress locations of specialized structures (Becker et al., 2005; Boßelmann et al., 2007). In these cases, however, local composition shifts are mostly attributed to the presence of other minerals.

2.4.2. No evidence for a separate phosphate phase.

Previous studies debate the presence of amorphous and crystalline phosphate minerals in lobster exoskeletons. For example, high P concentrations associated with localized features, such as organule canals, are interpreted to be carbonated

(Kunkel et al., 2012). Yet, other studies are unable to identify specific non-carbonate mineral phases using high-resolution spectroscopy and Thermo-Gravimetric Analysis

(Becker et al., 2005; Boßelmann et al., 2007; Al-Sawalmih et al., 2009). Arguments for a separate phosphate phase that persists throughout the molting cycle are made for another malacostracan, the giant prawn M. rosenbergii (Soejoko and Tjia, 2003).

26 In this quantitative study, structural analyses of the cephalothorax and chelae

detect only ACC and calcite, without evidence for additional inorganic phases (e.g. Fig.

2.2). Moreover, a simple sensitivity analysis suggests the probability of a separate phase

is low. First, stoichiometric P/Ca ratios of common minerals are 0.6 to

2.0 for monocalcium phosphate Ca(HPO4)2 to the family Ca5(PO4)3(OH), respectively. If one assumes all P is contained in a single and separate phase, the mineral fraction would need to contain ≈5 to 16% of this phase to give the P/Ca of 0.093±0.003 determined in this study (Fig. 2.4B). This amount is well above the 1 to 2% resolution of the PDF analysis for phosphate minerals. However, the limitations of current analytical methods prevent an unambiguous conclusion. For example, if P is distributed in multiple minor crystalline or amorphous phases, the detection of additional phase(s) in the total profile is further decreased. Knowledge of the exact structure(s) is also critical to detection.

2.4.3. Enigma of a P/Mg signature.

To our knowledge, the covariance of Mg and P in the exoskeleton of the main body and

the chelae is not previously reported in biominerals. The relationships in Fig. 2.3AB and

2.4AB suggest an interplay between the concentration of these elements and calcium. To

test this idea, we evaluate Mg and P concentrations, normalized to Ca. By comparing

Mg/Ca and P/Ca ratios, Fig. 2.6A shows the compositions of all body parts, including the

chelae, collapse onto a single trend given by:

(2.5)

27

Figure 2.6. (A) Ratios of P/Ca and Mg/Ca are covariant in the lobster bulk exoskeleton. Compositions indicate a linear trend that includes all body parts and also resolves the cephalon (head), main body, and claw compositions into distinctive ranges. Linear regression (Eq. 2.5) finds P/Mg is conserved at 1.27 ±0.30 across all body parts. (B) The available exoskeleton data for other Malacostraca animals (Table A2, Appendix A) are compared with measurements from this study (one averaged data point per body part) and show a similar interspecies correlation. Dashed lines correspond to the 2σ standard error (95% confidence interval) about the regression (solid line). Detailed statistics are given in Table A2, Appendix A. *(Clarke and Wheeler, 1922); †(Boßelmann et al., 2007); ‡ this study; §(Huner et al., 1979); ¶(Becker et al., 2005)

28 As seen in Fig. 2.6A, the relationship resolves the cephalon, posterior parts of the body,

and chelae into distinct compositional regions. Rearranging Eq. 2.5 and substituting the

bulk Ca/Mg value for all body parts (10.86±0.29; Table A1, Appendix A), we obtain the

mean P/Mg signature of 1.27 ±0.30 for the entire exoskeleton. This value is internally

consistent with compositions (4A) and (4B) within the errors of the estimate.

Recalling that Mg and P have probable roles in increasing the lifetime of ACC, it is plausible the Mg/P signature has geochemical origins. For example, the relationship in

Fig. 2.6A could be explained by charge balance requirements. Owing to the relatively

2+ 2+ 2- high charge density of Mg , a coupled substitution between Mg and HPO4 may occur

2- in the mineral fraction with phosphate groups substituting for CO3 . Alternatively, the

size difference between the small Mg2+ and larger Ca2+ may be the primary driver for

2- 2- substituting a portion of CO3 by the larger HPO4 . Arguments against a size-based

interpretation include the fact that most of the mineral fraction is amorphous and

therefore the typical rules for ion substitution should not apply. However, our structural

analysis shows short-range ordering within 7 Å, suggesting the possibility of crystalline

domains.

2.4.4. Broader pattern for multiple crustacean species?

The relationships in Fig. 2.6A lead us to return to the reported bulk compositions

(Table 2.1) and ask if the exoskeletons of other crustacean species contain similar Mg/Ca

and P/Ca systematics. We first assume bulk compositions of exoskeletons from other

species are dominated by the mineral fraction as demonstrated in this investigation of the

lobster. For those few studies that also report Ca data, we calculate bulk Mg/Ca and P/Ca.

29 Fig. 2.6B suggests the exoskeleton of three marine animals (crab, lobster, and marine shrimp) and one terrestrial species (pillbug) exhibit a similar composition trend for n=17 measurements:

(2.6)

We recognize the available data are limited (Table A2, Appendix A) and that we cannot account for modifying factors such as the dependence of composition on the age of an individual animal (Richards, 1951). We also note the calcite-enriched chelae of the red rock crab (Boßelmann et al., 2007) plot below the trend but this offset is consistent with our finding that lobster body parts with a greater fraction of calcite are associated with lower P/Ca. Rewriting Eq. 2.6 and substituting the ratio of Ca/Mg = 11.40 ±1.0 for all animals (Table A1 and A2, Appendix A), we obtain a P/Mg signature = 1.12 ± 0.04.

The relationship suggests Mg and P levels may be co-optimized in the exoskeletons of each species to meet functional needs and indicates crustacean exoskeletons share previously unrecognized composition patterns.

2.4.5. Physical basis for taphonomic bias in skeletal preservation.

The fossil record of decapods is dominated by isolated chelae (Hyžný et al., 2015) and modern taphonomic studies conducted in the field and the laboratory indicate the chelae are sometimes the only skeletal elements to survive significant degradation and fragmentation (Allison, 1986; Mutel et al., 2008). Our measurements showing that chelae have a high P/Ca may provide a thermodynamic basis for empirical observations of a preservation bias for chelae compared to other skeletal elements— Substitution of P into the carbonate solid at concentrations near those observed in exoskeletons is known to

30 significantly reduce the solubility and growth of carbonate phases (Avnimelech, 1983;

Busenberg and Plummer, 1985). This reduced solubility may provide a thermodynamic basis for the taphonomic bias observed in the fossil record. As discussed previously, P interactions between mineral and organic matrix may also promote durability of this composite.

Findings from this study may also explain the observation that decapod are preferentially preserved in phosphatic concretions (Tsujita, 2003). Ca and P from the degrading decapods could at least partially sustain phosphatization, as suggested by numerous studies of and (Martill, 1990; Briggs and

Kear, 1994; Briggs and Wilby, 1996). There is some evidence to the contrary (Allison,

1986), but the findings of this study provide a path forward.

2.5 CONCLUSIONS

This quantitative investigation determines the composition of seven exoskeleton body parts and finds the lobster exoskeleton can contain chemical systematics in Mg/Ca and P/Ca. The measurements also identify a strong covariance between Mg and P concentrations for all body parts and we show the ratio of these elements is conserved.

The patterns provide quantitative evidence for claims that the composition of the mineral fraction is highly regulated. By applying our finding that bulk concentration of Ca, Mg, and P in the lobster exoskeleton is dominated by the mineral fraction, we return to published data and show other Malacostraca animals may share similar Mg/Ca and P/Ca systematics. The relationships suggest underlying commonalities in the biomineralization processes of diverse marine and terrestrial crustaceans, perhaps through their shared

31 evolutionary roots. For example, the lobster is an early decapod species whose biomineralization pathways pre-date those of crabs and isopods (pillbugs, etc.). Similar trends are noted across species in later decapods and malacostracans. The composition patterns suggest the diverse structure-function requirements of exoskeletons are optimized, at least partially, through composition.

32

2.6 REFERENCES

Addadi L., Joester D., Nudelman F. and Weiner S. (2006) Mollusk shell formation: A source of new concepts for understanding biomineralization processes. Chem. - A Eur. J. 12, 980–987. Addadi L., Raz S. and Weiner S. (2003) Taking advantage of disorder: Amorphous calcium carbonate and its roles in biomineralization. Adv. Mater. 15, 959–970. Aizenberg B. J., Lambert G., Addadi L. and Weiner S. (1996) Stabilization of Amorphous Calcium Carbonate. Adv. Mater. 8, 222–226. Akiva-Tal A., Kababya S., Balazs Y. S., Glazer L., Berman A., Sagi A. and Schmidt A. (2011) In situ molecular NMR picture of bioavailable calcium stabilized as amorphous CaCO3 biomineral in gastroliths. Proc. Natl. Acad. Sci. U. S. A. 108, 14763–14768. Al-Sawalmih A., Li C., Siegel S., Fratzl P. and Paris O. (2009) On the stability of amorphous minerals in lobster cuticle. Adv. Mater. 21, 4011–4015. Allison P. A. (1986) Soft-bodied animals in the fossil record: The role of decay in fragmentation during transport. Geology 14, 979–981. Arias J. L. and Fernandez M. S. (2008) and Proteoglycans in Calcium Carbonate-based Biomineralization. Chem. Rev. 108, 4475–4482. Avnimelech Y. (1983) and calcium carbonate solubilities in Kinneret. Limnology 28, 640–645. Becker A., Ziegler A. and Epple M. (2005) The mineral phase in the of two species of Crustacea consists of magnesium calcite, amorphous calcium carbonate, and amorphous calcium phosphate. Dalt. Trans., 1814. Bentov S., Weil S., Glazer L., Sagi A. and Berman A. (2010) Stabilization of amorphous calcium carbonate by phosphate rich organic matrix proteins and by single phosphoamino acids. J. Struct. Biol. 171, 207–215. Blue C. R. and Dove P. M. (2015) Chemical controls on the magnesium content of amorphous calcium carbonate. Geochim. Cosmochim. Acta 148, 23–33. Blue C. R., Giuffre A., Mergelsberg S., Han N., De Yoreo J. J. and Dove P. M. (2017) Chemical and physical controls on the transformation of amorphous calcium carbonate into crystalline CaCO3 polymorphs. Geochim. Cosmochim. Acta 196, 179–196. Boßelmann F., Romano P., Fabritius H., Raabe D. and Epple M. (2007) The composition of the exoskeleton of two crustacea: The American lobster Homarus americanus and the edible crab Cancer pagurus. Thermochim. Acta 463, 65–68. Briggs D. E. G. and Kear A. J. (1994) Decay and Mineralization of . Palaios 9, 431. Briggs D. E. G. and Wilby P. R. (1996) The role of the calcium carbonate-calcium phosphate switch in the mineralization of soft-bodied fossils. J. Geol. Soc. London. 153, 665–668. Busenberg E. and Niel Plummer L. (1985) Kinetic and thermodynamic factors controlling the distribution of SO32− and Na+ in calcites and selected . Geochim. Cosmochim. Acta 49, 713–725. Clarke F. W. and Wheeler W. C. (1922) The inorganic constituents of marine

33 / by Frank Wigglesworth Clarke and Walter Calhoun Wheeler., U. S Gov’t. Print. Off., Washington. Clarkson J. R., Price T. J. and Adams C. J. (1992) Role of metastable phases in the spontaneous precipitation of calcium carbonate. J. Chem. Soc. Faraday Trans. 88, 243. Coblentz F. E., Shafer T. H. and Roer R. D. (1998) Cuticular proteins from the blue crab alter in vitro calcium carbonate mineralization. Comp. Biochem. Physiol. - B Biochem. Mol. Biol. 121, 349–360. Cribb B. W., Rathmell A., Charters R., Rasch R., Huang H. and Tibbetts I. R. (2009) Structure, composition and properties of naturally occurring non-calcified crustacean cuticle. Arthropod Struct. Dev. 38, 173–178. Dai L., Cheng X. and Gower L. B. (2008) Transition Bars during Transformation of an Amorphous Calcium Carbonate Precursor. Chem. Mater. 20, 6917–6928. Freitas P. S., Clarke L. J., Kennedy H., Richardson C. A. and Abrantes F. (2006) Environmental and biological controls on elemental (Mg/Ca, Sr/Ca and Mn/Ca) ratios in shells of the king scallop Pecten maximus. Geochim. Cosmochim. Acta 70, 5119–5133. Glusker J. P. (1991) Structural Aspects of Metal Liganding to Functional Groups in Proteins. In pp. 1–76. Glynn J. P. (1968) Studies on the ionic, protein and phosphate changes associated with the moult cycle of Homarus vulgaris. Comp. Biochem. Physiol. 26, 937–946. Greenaway P. (1985) Calcium Balance and in the Crustacea. Biol. Rev. 60, 425–454. Grodowitz M. J., Roseland C. R., Hu K. K., Broce A. B. and Kramer K. J. (1987) Mechanical properties of mineralized and sclerotized puparial cuticles of the flies Musca autumnalis and M. domestica. J. Exp. Zool. 243, 201–210. Hammersley A. P., Svensson S. O., Thompson A., Graafsma H., Kvick Å. and Moy J. P. (1995) Calibration and correction of distortions in two-dimensional detector systems. Rev. Sci. Instrum. 66, 2729–2733. Hayes D. K. and Armstrong W. D. (1961) The Distribution of Mineral Material in the Calcified Carapace and Claw Shell of the American Lobster, Homarus americanus, Evaluated by Means of Microroentgenograms. Biol. Bull. 121, 307–315. Huner J. V., Colvin L. B. and Reid B. L. (1979) Postmolt Mineralization of the of Juvenile California Brown Shrimp,Penaeus Californiensis (Decapoda:Penaeidae). Comp Biochem Physiol 62A, 889–893. Hyžný M., Schlögl J., Charbonnier S., Schweigert G., Rulleau L. and Gouttenoire M. (2015) Intraspecific variation and taphonomy of a new erymid lobster (Crustacea: Decapoda) from the Middle of Belmont (Beaujolais, France). Geobios 48, 371–384. Klein R. T., Lohmann K. C. and Thayer C. W. (1996) Bivalve skeletons record sea- surface temperature and d18 O via Mg/Ca and 18-O/16-O ratios. Geology 24, 415– 418. Kunkel J. G., Nagel W. and Jercinovic M. J. (2012) Mineral Fine Structure of the American Lobster Cuticle. J. Res. 31, 515–526. Levi-Kalisman B. Y., Raz S., Weiner S., Addadi L. and Sagi I. (2002) Structural Differences Between Biogenic Amorphous Calcium Carbonate Phases Using X-ray

34 Absorption Spectroscopy **. Adv. Fun 12, 43–48. Loste E., Wilson R. M., Seshadri R. and Meldrum F. C. (2003) The role of magnesium in stabilising amorphous calcium carbonate and controlling calcite morphologies. J. Cryst. Growth 254, 206–218. Marlowe R. L., Dillaman R. M. and Roer R. D. (1994) Lectin Binding by Crustacean Cuticle: The Cuticle of Callinectes sapidus Throughout the Molt Cycle, and the Intermolt Cuticle of Procambarus clarkii and Ocypode quadrata. J. Crustac. Biol. 14, 231–246. Martill D. M. (1990) Macromolecular resolution of fossilized muscle tissue from an elopomorph fish. Nature 346, 171–172. Michel F. M., MacDonald J., Feng J., Phillips B. L., Ehm L., Tarabrella C., Parise J. B. and Reeder R. J. (2008) Structural characteristics of synthetic amorphous calcium carbonate. Chem. Mater. 20, 4720–4728. Mutel M. H. E., Waugh D. A., Feldmann R. M. and Parsons-Hubbard K. M. (2008) Experimental Taphonomy Of Callinectes Sapidus And Cuticular Controls On Preservation. Palaios 23, 615–623. Neues F., Ziegler A. and Epple M. (2007) The composition of the mineralized cuticle in marine and terrestrial isopods: A comparative study. CrystEngComm 9, 1245–1251. Qi C., Zhu Y. and Chen F. (2014) Microwave Hydrothermal Transformation of Amorphous Calcium Carbonate Nanospheres and Application in Protein Adsorption. ACS Appl. Mater. Interfaces 6, 4310–4320. Qiu X., Thompson J. W. and Billinge S. J. L. (2004) PDFgetX2: A GUI-driven program to obtain the pair distribution function from X-ray powder diffraction data. J. Appl. Crystallogr. 37, 678. Reddy M. M. (1977) Crystallization of calcium carbonate in the presence of trace concentrations of phosphorus-containing anions. I. Inhibition by phosphate and glycerophosphate ions at pH 8.8 and 25°C. J. Cryst. Growth 41, 287–295. Richards A. G. (1951) The integument of arthropods: the chemical components and their properties, the and development, and the permeability, University of Minnesota Press, Minneapolis. Roer R. and Dillaman R. (1984) The Structure and Calcification of the Crustacean Cuticle. Am. Zool. 24, 893–909. Roseland C. R., Grodowitz M. J., Kramer K. J., Hopkins T. L. and Broce A. B. (1985) Stabilization of mineralized and sclerotized puparial cuticle of muscid flies. Biochem. 15, 521–528. Rütt U., Beno M. A., Strempfer J., Jennings G., Kurtz C. and Montano P. A. (2001) Diffractometer for high energy X-rays at the APS. Nucl. Instruments Methods Phys. Res. Sect. A Accel. Spectrometers, Detect. Assoc. Equip. 467–468, 1026–1029. Sato A., Nagasaka S., Furihata K., Nagata S., Arai I., Saruwatari K., Kogure T., Sakuda S. and Nagasawa H. (2011) Glycolytic intermediates induce amorphous calcium carbonate formation in crustaceans. Nat. Chem. Biol. 7, 197–9. Sawada K. (1997) The mechanisms of crystallization and transformation of calcium carbonates. Pure Appl. Chem. 69, 921–928. Soejoko D. S. and Tjia M. O. (2003) Infrared spectroscopy and X ray diffraction study on the morphological variations of carbonate and phosphate compounds in giant prawn (Macrobrachium rosenbergii) skeletons during its moulting period. J. Mater. Sci. 38,

35 2087–2093. Travis D. F. (1963) Structural Features of Mineralization From Tissue To Macromolecular Levels of Organization in the Decapod Crustacea. Ann. N. Y. Acad. Sci. 109, 177–245. Tsujita C. J. (2003) Smothered scampi: taphonomy of lobsters in the Upper Cretaceous Bearpaw Formation, southern Alberta. J. Taphon. 1, 187–206. Vigh D. A. and Dendinger J. E. (1982) Temporal Relationships of Postmolt Deposition of Calcium, Magnesium, Chitin and Protein in the Cuticle of the Atlantic Blue Crab, Callinectes Sapidus Rathbun. Comp Biochem Physiol 72. Vijayan K. K. and Diwan A. D. (1996) Fluctuations in Ca, Mg and P levels in the hemolymph, muscle, midgut gland and exoskeleton during the moult cycle of the Indian white prawn, Penaeus indicus (Decapoda: Penaeidae). Comp. Biochem. Physiol. - A Physiol. 114, 91–97. Wood S. and Russell J. D. (1987) On the Nature of the Calcium Carbonate in the Exoskeleton of the Woodlouse Oniscus asellus L . ( , Oniscoidea ). Crustaceana 53, 49–53. Xu A. W., Yu Q., Dong W. F., Antonietti M. and Cölfen H. (2005) Stable amorphous CaCO3 microparticles with hollow spherical superstructures stabilized by phytic acid. Adv. Mater. 17, 2217–2221. Ziegler A. (2008) The cationic composition and pH in the moulting fluid of Porcellio scaber (Crustacea, Isopoda) during calcium carbonate deposit formation and resorption. J. Comp. Physiol. B 178, 67–76.

36

CHAPTER 3. RELATIONSHIPS OF α-CHITIN AND CARBONATE MINERAL COMPONENTS REVEAL BIOMINERALIZATION STRATEGIES OF CRUSTACEAN EXOSKELETONS

Sebastian T Mergelsberg, F Marc Michel, Scott Case, Patricia M Dove (in prep.)

ABSTRACT

Animals in the Class Malacostraca include well-known species, such as crabs, isopods, shrimp, and the American lobster. Their exoskeletons are comprised of calcite and/or amorphous calcium carbonate, as well as chitin and proteins such as present in other arthropods, insects, and . These structures are unique natural biocomposites that provide protection, locomotion, and stability. This study was designed to determine the structural and mechanical properties of exoskeletons from three Malacostraca species— two crabs and the American lobster. Using high-energy synchrotron x-rays and PDF, we found that the crystallinity of chitin (% CrI130) and CaCO3 (calcite/total CaCO3) are covariant between the body parts of the crab species. Crab claws have greater chitin crystallinity and a higher proportion of calcite than the main body part, as expected per the biological function of these body parts. In contrast, the compositions of the claws and main body of the lobster are similar and parts contain little calcite relative to amorphous CaCO3. Parallel Raman spectroscopic analyses of exoskeleton cross-sections independently confirm the covariant crystallinities. Crystallinity within the cross-section is highly variable in the lobster exoskeleton. The findings led us to test the hypothesis that chitin crystallinity, not mineral crystallinity, is the primary control on exoskeleton toughness and rigidity. By determining the stress-strain relationship for exoskeleton samples, we found that greater rigidity in the exoskeleton is correlated with chitin crystallinity. That is, lower crystallinity correlates with more flexible structures. There is no relationship between exoskeleton strength and % calcite, but strength is correlated with exoskeleton thickness. The measurements suggest chitin is the primary contributor to exoskeleton reinforcement for these crustaceans and challenge the long-standing assumption that CaCO3 contributes strength to the exoskeleton. Rather, the data suggest chitin is the key structural component that confers toughness and rigidity to the exoskeletons of all arthropods, including crustaceans.

37

3.1 INTRODUCTION

Malacostraca is a morphologically diverse class of crustaceans that includes well-

known species, such as the American lobster, all crabs, isopods, and shrimp. While little

is known about the exoskeleton or cuticle of the earliest crustaceans due to poor

preservation in the sedimentary record (Klompmaker et al., 2015), all modern arthropods

share a homologous, chitinous structure. Crustacean exoskeletons are unique

biocomposites that contain inorganic minerals, as well as the chitin and proteins present

in insects and arachnids. While the primary functions of the exoskeleton are understood

to be protection, locomotion, and stability, little is known about the molecular and

nanoscale structure (Raabe et al., 2006).

The most abundant component of the cuticle by volume is the polysaccharide

chitin, which forms an ordered framework of anti-parallel strands, referred to as α-chitin

that is organized in a twisted plywood structure (Bouligand, 1972; Fabritius et al., 2009).

Surrounding these fibers is a protein layer composed of up to 50 different peptides that

contain motifs commonly associated with chitin- or calcite-binding functions (Skinner et

al., 1992; Compère et al., 2002). Interspersed throughout these organic materials is the

primary mineral component, calcium carbonate, occurring as a biomineral in most

crustaceans (Roer and Dillaman, 1984; Cribb et al., 2009). Crystalline calcite and

amorphous calcium carbonate (ACC) typically occur in ratios that vary from species to

species, but are oriented along the underlying organic framework (Hegdahl et al., 2008;

Al-Sawalmih et al., 2008). With the exception of several species of crab, which are enriched in halogen elements (chlorine and bromine) in the tips of very specialized claw structures (Cribb et al., 2005; Cribb et al., 2009), current studies suggest a highly

38 conserved inorganic chemistry with 2 wt% of magnesium and approximately 4 wt% of phosphorous throughout the exoskeleton (Boßelmann et al., 2007).

Malacostracans exhibit extreme morphological and functional differences in their exoskeleton. Most commonly, this discrepancy is explained by an implied difference in the expression of cuticular biomolecules, despite very poor characterization and understanding of these compounds (Roer and Dillaman, 1984). Proteins are generally thought of as the nucleation site for the calcium carbonate minerals, as well as the

‘adhesive’ that provides strength to the chitin framework. Recent studies have found evidence in modern skeletal structures for the participation of polysaccharides in the nucleation and stabilization of biominerals, raising the question if nano-scale heterogeneities in the chitin fibers may serve as sites of CaCO3 precipitation and growth

(Ehrlich, 2010). Rather than functionalized side-chains, polysaccharides can have variable three-dimensional structures conducive to the formation of crystallization sites at specific intervals along the fiber. In other biomineralization systems, many studies have suggested, polysaccharides provide adequate biological and chemical control to explain the formation and texture of the minerals. Unlike proteins, chitin is more consistently available in all crustaceans in a highly conserved structure.

Evidence from Raabe et al. (2006) suggests that overall structure and orientation of calcium carbonate and α-chitin are tightly linked on the nano-scale in the cuticle of the

American lobster. Using high-energy X-ray diffraction beams, they were able to demonstrate that orientations of organic and inorganic components were aligned in relation to location along the body, resulting in specific ranges of arrangements for each body part. The crystalline calcite only represents a fraction of the inorganic particles,

39 however, raising the question if there are any body part specific structural interactions

between ACC and the organic matrix. According to Raman imaging studies by Hild et al.

(2008) both components are easily resolved with the spectroscopic technique, even in the

thin cuticles of terrestrial isopods. Development of these chemistry-sensitive

spectrographic analyses offers the opportunity to establish a physical chemical picture of

the hierarchy and organization between chitin and calcium carbonate.

Figure 3.1. The four body parts investigated in this study as they compare between lobster and crab bodies. Three or more samples were collected from each location of three animals for a total of up to 36 samples for each species.

This study hypothesizes chitin and CaCO3 assume separate structural roles within

this unique and complex biocomposite of four homologous body parts (Fig. 3.1). Using a

well-established synchrotron-based X-ray analysis method, we establish that across three

Malacostraca, two crab species and the American Lobster, polymeric order of calcium

carbonate and of the polysaccharide framework linked between body parts of single

organisms, but not across species. To confirm this finding, we design a novel proxy for

chitin crystallinity using shifts in the Raman spectrum of the exoskeleton. This allows for

40 the comparison of CaCO3 crystallinity with the chemistry and structure of chitin. We also measure the rigidity of each body part from each of the three species to establish the relationship between structure and mechanical properties of the exoskeleton.

Using this method, we observe two separate strategies to reinforce crustacean cuticle using the inorganic and organic components. In crabs, crystallinity of chitin and calcium carbonate are elevated in claws relative to the cephalothorax, whereas the lobster cuticle only showed increases of CaCO3 crystallinity. Further, results show that

crystallinity of the polysaccharide chitin controls the rigidity of exoskeletons from all

three animals. Recognizing that most other arthropods lack the mineral component in the

exocuticle, the presence of calcium carbonate in crustacean exoskeletons may not serve a

mechanical purpose. Comparison of crystallinity measurements from six species suggests

the percentage of calcite vs. ACC may be an indicator of crustacean ecology/evolution and thus suggests chitin and CaCO3 are independently- optimized components of the

exoskeleton.

3.2 MATERIALS AND METHODS

3.2.1 Animals

O. asellus and A. spp were collected from mossy ground near Portland Oregon

(USA). Animals were kept in separate terraria on a leafy substrate and were fed a diet of

leaves and vegetables. H. americanus, M. magister, and C. productus were purchased live

from online food retailers. Several aquarium retailers provided living specimens of U.

uca, N. maculatus, L. amboinensis, S. hispidus, P. cadenati. All animals were living male

adults in the C4 stage of intermolt at the beginning of sample preparation. Dissection

41 occurred at room temperature and cuticles were rinsed twice for 1s each in 18.2 MΩ·cm

ultrapure water and 100% methanol and air dried before storage at -20° C.

3.2.2 Three-point flexural test and mechanical analysis

Fresh exoskeleton samples were cut into ≈5 cm long and ≈1.3 cm wide strips

directly after dissection. Each sample was carefully cut to be approximately flat—all

surface features (e.g. bumps and ridges) and inherent curvature were photographically

documented to assess any geometric bias in the data. During testing, the exoskeletons

were kept in ice-cooled beakers to prevent changes in moisture. The width and thickness

of each strip was determined at six locations using a digital caliper and all measurements

are reported as the arithmetic mean. Exoskeletons were then inserted concave side down

into a custom three-point fixture with a span of 40 mm between supports and the

crosshead was zeroed to begin testing. The diameter of the cylindrical loading nose and

supports measured 3 mm. The load was applied using an Instron Model 4204 Universal

Testing machine retrofitted with a MTS Systems ReNew controller and measured using a

1 kN Instron cell 2518-806. Testing speed was set at 1.0 mm/min for crosshead

movement and load vs. deflection data were collected in the Testworks 4 software.

A customized script was developed for data analysis in the statistical programming language R (R Development Core Team, 2008) to calculate stress and strain values, estimate the modulus, and to determine toughness at the initial and ultimate points of failure (Appendix C2). Specific terms and definitions are summarized in

Appendix C1. Final values are reported in Table C2, Appendix C. In general, all analysis follows ASTM D2344/D2344M-16 and ASTM D790-17. Only the slope of the

42 initial linear section of the stress-strain curve was considered in the calculation of the

flexural modulus. The point of initial failure (σi) was defined as the first significant drop in σ that was followed by a measurable change in slope. If a change in slope was

measured at least 0.3% from ultimate failure (σult), initial failure was defined as the point

at which tangents to both slopes (initial and final) intersect. The ultimate failure (σult) is only reported for samples where the exoskeleton failed. If no failure occurred, maximum stress values are reported as σmax. Toughness values were calculated using a trapezoidal

approximation to the area under the curve to initial or ultimate failure points.

3.2.3 Raman spectroscopy

Dried cuticle samples were fractured and mounted on 2 cm long polyethylene

substrates using cyanoacrylate adhesive (Toagosei, Japan) to expose each saggital cross

section. Polished surfaces were prepared using steel blades, glass knives, and a diamond

knife (DiATOME, Biel, Switzerland). For each sample, the fractured surface was first

leveled using a glass knife. The sample was then polished using a diamond knife by

advancing the blade 15 times at increments of 100nm, 70nm, 40nm, and 30nm.

A confocal Raman microscope (Horiba, Kyoto, Japan) was used to acquire

spectra at approximately 10 locations per sample. For each location, five full Raman

spectra (100 to 3750cm-1) were collected using an integration time of 50s each. To minimize local effects of polarization and orientation, a 50x Nikon objective was used.

The fluorescent background was removed by using the baselineWavelet function (Zhang et al., 2009) in the statistical programming language R (R Development Core Team,

2008). Absolute and relative peak positions were refined using a Voigt fit in OriginPro

43 (OriginLab, Northampton, MA). Reference wavelengths of peaks specific to chitin and

CaCO3 are summarized in Table B2, Appendix B.

3.2.4 Synchrotron X-ray Diffraction

X-ray diffraction experiments were performed at beamline 11-ID-B (Rütt et al.,

2001) of the Advanced Photon Source (Argonne, IL USA) using an incident photon

energy of ~58.6 keV (λ = 0.2114 Å). The samples were mounted between two opposing

pieces of polyimide (Kapton®) tape and the scattered radiation was measured in

transmission mode using an amorphous-Si detector system manufactured by Perkin

TM 2 Elmer (2048 × 2048 pixels, 200 × 200 μm pixel size). A CeO2 standard (NIST diffraction standard set 674a) was used to calibrate the sample-to-detector distance and the non-orthogonality of the detector relative to the incident beam path. Conversion of scattering data from 2D to 1D was performed using the program Fit2D (Hammersley et al., 1995). A polarization correction was applied during integration of the data. Direct subtraction of the sample holder was accomplished by the independent measurement of the true background intensity of an empty polyimide tape.

Reflections of both chitin and calcite peaks in a typical averaged 2D diffraction pattern were acquired by synchrotron X-ray diffraction roughly perpendicular to the cuticle surface (Fig. B1, Appendix B). Calcite reflections were weak and highly variable between samples, indicating a high proportion of ACC for samples from the American lobster and the banded coral shrimp. Chitin reflections are generally more discernable for all samples, despite the broad nature of its diffuse scattering (Table B1, Appendix B).

44 The Chitin Crystallinity Index (CrI) was determined as the ratio of the measured x-ray intensity along [020] to that along [130] in the α-chitin structure. α -Chitin exhibits an increasingly strong reflection along its [020] plane with an increase in disordered fibers, which allows for the determination of relative crystallinity— a method that has been well-established for synthetic chitin samples (Kumirska et al., 2010):

CrI130 = (I020/I130) ×100 (3.1) where I020 denotes the intensity of the diffraction peak of [020], and I130 that of the diffraction peak of [130]. Peak positions used in this calculations were determined using a Voigt function fit in OriginPro.

45 3.3 RESULTS AND DISCUSSION

3.3.1. High-energy X-ray Diffraction Reveals Disparate Structural Trends

Analysis of calcite and chitin reflections acquired from the synchrotron x-ray method

(Fig. B1, Appendix B) provides a unique opportunity to evaluate cuticle structure. Focusing on multiple body parts from three animals, Fig. 3.2 shows the data could be analyzed for broad trends suggesting relationships between the relative abundance of calcite in the CaCO3 particles derived from Pair Distribution Functions (PDFs) and the Chitin crystallinity index (CrI). Neither determination reflects cuticle hardness or thickness (Fig. B2, Appendix B), but both exhibit a clear trend when compared to body part function.

Figure 3.2. Calcite/ACC ratio vs. chitin crystallinity shows covariant polymeric ordering in reinforced cuticles of crabs. Samples from the American Lobster indicate a smaller, inverse relationship between Scherrer size for more reinforced structures. Crystallinity and size measurements of chitin and calcite do not reflect cuticle hardness. Samples from body parts of each organism, in decreasing order of reinforcement:  chela (claw), ■ cheliped (), ♦ abdomen (tail), and  cephalothorax (main body cavity). N=14.

46

First, measurements of body parts from the show covariation of both chitin crystallinity and the fraction of calcite present in the mineral component of the cuticle.

Measurements of α-chitin crystallinity show the differences between body parts are expressed as increases in the crystallinity index of almost 10% from the leg to the cephalothorax to the claw.

The RRC exhibits a similar trend, although with a less significant spread in both quantities.

Differences in α-chitin crystallinity are less pronounced compared to calcite content of appendices. A positive correlation of these measures for animals of taxon Brachyura (crabs) suggest a higher degree of order in both the mineral and polysaccharide components in reinforced body structures.

In contrast, the lobster cuticle maintains a narrow variation in α-chitin crystallinity index

(3%), while more significant differences between body parts are primarily contained in the inorganic component (20%). Increased chitin crystallinity in the thickened claw cuticle provides a similar progression compared to the trend observed for Brachyura. Unlike the crabs, the mineral component in the lobster exoskeleton appears to be most crystalline in the cephalothorax.

Differences in exoskeleton reinforcement suggest chitin and calcite ordering may be taxon-specific, without underlying first-order homologous or physico-chemical trends. Our findings for chitin using the synchrotron x-ray approach are consistent with observations made by several material studies (Raabe et al., 2006; Al-Sawalmih et al., 2008). This demonstrates the promise of this analytical method for quantifying chitin and calcite structure within cuticles.

Differences in exoskeletons between Brachyura organisms and the lobster have previously only been documented as discrepancies in physical properties. Palmer et al., 1999 showed that animals of genus Cancer (relatives of the RRC) were able to achieve biting forces

47

with their claws that exceeded that of the lobster by an order of magnitude. Accordingly, the

breaking force of the claw exoskeleton is approximately two orders of magnitude higher for

Cancer organisms than it is in the case of the lobster (Chen et al., 2008)

Due to analytical restrictions, such as mounting limitations and beam size, high-energy x-

ray diffraction yields results integrated over large volumes of material. Measurements detailed in

Fig. 1 show averages, without any information of variation or co-localization of cuticle order.

Possible physico-chemical relationships and interactions of cuticle components are lost to

variations in cuticle thickness and orientation. Furthermore, anisotropy created by co-orientation

of inorganic crystallites along the chitin matrix greatly increases errors in PDF acquisition.

3.3.2 Raman Spectroscopy Reveals Cuticle Heterogeneity

To further resolve the chitin crystallinity and the proportion of calcite in each body part,

the samples were analyzed in cross-section using Raman spectroscopy, a toolkit sensitive to

bonding energies within the composite. To quantify the structure and chemical basis for

differences in chitin crystallinity, Raman spectra were obtained for the cuticle of the three

organisms (Fig. B3). Typical spectra for the cuticle from the Red Rock Crab (green), Dungeness

Crab (blue), and American Lobster (red) illustrate the peak position of the vibrations for the

mineral and chitin components (Fig. B3).

For the mineral component, we focus on the υ1-CO3 Raman shift that occurs between

1079 and 1088 cm-1. Raman shifts closer to 1079 cm-1 indicate a higher proportion of ACC,

whereas peaks at 1088 cm-1 correspond to crystalline calcite. Note that Raman peaks of other

carbonate compounds are in this region as well, so high-energy x-ray scattering was used to confirm the CaCO3 polymorphs.

48

The chitin polysaccharide has a number of major vibrational modes (Fig. 3.2A), of which the CHx modes are usually the focus of most synthetic studies (Kumirska et al., 2010). The

-1 -1 amide and CH2 vibrations within the 1150 cm and 1660 cm region are the most sensitive for natural samples (Fig. 3.2B). to quantify local differences in chitin crystallinity between body parts of the three animals, Raman peak positions for Amide I were treated as follows:

(3.2)

-1 -1 where x denotes the peak position of the Raman shifts near 1660cm and 1624cm and δAmide I represents the ratio between them. This ratio is relatively disperse in the cephalothorax (main body cavity) of the crabs, compared to the more narrow distribution in the claw at higher values of ν1-CO3 (Fig. B3A). There is no apparent trend between δAmide I and ν1-CO3, which is expected because chitin fibers are not thought to directly interact with the mineral.

Similarly to δAmide I, the ratio δAmide III can be used to quantify the difference in amide vibrations caused by intramolecular hydrogen bonding as follows:

(3.3)

Again, there is no apparent trend between δAmide III and ν1-CO3 (Fig. B3B). Despite showing a highly consistent degree of chitin crystallinity in the X-ray analysis, lobster samples appear to vary highly in the degree to which the fibers are consistently hydrogen bonded.

Bonding of chitin fibers in the cephalothorax is more consistent compared to the claw, despite an observed increase in CaCO3 crystallinity. There likely is no direct hydrogen bonding between chitin fibers and the mineral component, as predicted by Raabe et al. (2006). Other measures of chitin vibrational states may be more conclusive if there is no direct association between the polysaccharide and the mineral.

49

3.3.3 Steric strain is a proxy for chitin crystallinity

To quantify steric strain of the chitin polymorph, the quantity δCH2 was defined as

follows:

(3.4)

This ratio represents the vibrational energy of an intramolecular mode to that of a rocking

motion between fibers. In samples from the crabs, a significant drop in this ratio is observed

between samples of the main body and those on the more reinforced claws (Fig. 3.3A). Lower

average values in δCH2 correspond to more regularly structured fibers with a higher polymeric

order.

Figure 3.3. Raman spectroscopy reveals sample heterogeneity. Shifts in CH2 vibrational -1 energies of peaks located at 1376, 1414, and 1449 cm relative to differences in CaCO3 crystallinity. A) The difference in vibrational strain is higher in samples from the cephalothorax for all three organisms when compared to claw material. Raman peaks at 1376 and 1414 cm-1 are very sensitive to vibrations of sidechains relative to the ring structure. B) Similar trends as in A) are observed when comparing the Raman peaks at 1414 and 1449 cm-1. Unlike the previous example, δCH2 is a more direct measure of intramolecular ring strain, without significant contributions from sidechains.

50

Similarly to the amide quantities, a more narrow spread is observed in samples from the reinforced structure. A new quantity, δδCH2 was defined, as follows:

(3.5)

As expected, a similar reduction in the average value of δδCH2 is observed for all three organisms in the claw versus the cephalothorax (Fig. 3.3B) without a significant drop in the degree of distribution. Reinforced structures of all three organisms are thus more likely to have more strained chitin molecules, meaning they are present in a more condensed linear structure.

The ratios δCH2 and δδCH2 thus indicate consistently longer fibers, without a corresponding increase in the order between fibers.

3.3.4 Raman Spectroscopy independently measures chitin crystallinity

Comparisons between Raman and X-ray methods show that the two are in relatively good agreement for the nanostructure of samples from the three organisms (Fig. 3.4A). As one would

-1 expect from the discussion above, wavenumbers around 1079 cm indicate CaCO3 spherules of a very short-range structure (ACC), while wavenumbers closer to 1088 cm-1 correspond to crystalline calcite minerals (dotted line; Hild et al., 2008; Kumirska et al., 2010). Because X-ray analyses were performed perpendicular to the sample, the % calcite value represents that of the bulk material, causing the Raman data to plot with a slight offset for thinner layers within the exoskeleton cross section. This highlights the large variability of CaCO3 crystallinity within the lobster claws, which cannot be resolved by diffraction. Crab cuticles exhibit a narrower spread in

CO3 vibrational energies, suggesting a more homogenous biocomposite. This is very pronounced for RRC claws and the DC cephalothorax. Offsets in RRC cephalothorax and DC chela values

51

from the dotted line are likely due to measurement along slightly different structures within body parts, as they both have severe modifications (longitudinal ridges along RRC carapace and seta on the underside of DC claws).

Figure 3.4. Comparison to high-energy x-ray data identifies novel spectroscopic measure of chitin crystallinity. Correlation between crystallinity measures obtained PDF analysis and Raman spectroscopy for the American Lobster, Dungeness Crab, and Red Rock Crab. A) Relative abundance of percent calcite covaries with the wavenumber measured for the ν1-CO3 vibration roughly along a line corresponding to 0% calcite at 1079 cm-1 and 100% calcite at 1088 cm-1 (dotted line). B) A similar comparison can be performed for chitin when comparing the Raman -1 peaks at 1414 and 1449 cm (δδ-CH2) to chitin crystallinity (CrI130). Separation between Raman peaks decreases with increasing chitin crystallinity, suggesting a more compressed chitin structure. Dashed lines correspond with the 2σ standard error (95% confidence interval).

Similarly, measurements of chitin crystallinity show negative correlation with the quantity δδ-CH2, defined as the ratio between the Raman peaks at wavenumbers 1414 and 1449 cm-1 (Fig. 3.4B). This suggests an increase in the ring strain of the chitin molecular units as the overall order of the polymer (CrI) increases. Unlike the inorganic structure, polysaccharide order exhibits a relatively constant variability between organisms and body parts, indicative of a

52

framework structure rather than a more adaptable, fine-tuned component (Raabe et al., 2006).

However, Raman spectroscopy is able to differentiate between body parts more easily due to a

more variable chitin structure in the Brachyura cephalothorax compared to the claws. In contrast,

claws from the AL appear less consistent in α-chitin structure when compared to that of the

cephalothorax – similar to observations made for the inorganic component (Fig. 3.4A).

Raman spectroscopy thus confirms the more ordered nature of chitin fibers in the cuticle

of the Red Rock Crab compared to that of the Dungeness Crab. Qualitatively, this model is able

to resolve some of the molecular-scale structures recorded by the chitin crystallinity index

observed in Fig.1, though they are observed independently of the CaCO3 component. To resolve

the nano-scale structure of the cuticle for each of the separate strategies of reinforcement, a more

comprehensive model of molecular vibrations is needed.

3.3.5 Mechanical testing reveals chitin crystallinity controls flexural rigidity

Using three-point flexural testing geometry, the flexural rigidity of each sample was determined. This quantity describes the amount of energy required to bend the exoskeleton.

When compared to the proportion of calcite in the of each body part, no consistent trend emerges (Fig. 3.5A). Flexural rigidity is approximately equal in both body parts of the

Dungeness Crab. Samples from the red rock crab and the American lobster show opposite dependence of rigidity on the calcite content and the claw exoskeleton is very variable in its mechanical properties. The flexural modulus of the samples exhibits some sensitivity to calcite content (Fig. B5A); however, plotting these data as a function of chitin crystallinity produces a similar trend (Fig. B5B). This may be due to variations in sample thickness, as both the strain at initial failure εi and the flexural modulus Ef exhibit some dependence on exoskeleton thickness

(Fig. B5 C and D).

53

Figure 3.5. (A) The flexural rigidity of exoskeletons varies as a function of the proportion calcite to ACC. Comparison of the lobster exoskeleton to that of the red rock crab reveals opposite trends of rigidity vs. differences in CaCO3 crystallinity between body parts. (B) If compared to chitin crystallinity, however, flexural rigidity co-varies with an increase in ordering. This confirms observations previously made on synthetic polymer-calcite composites (Wegst and Ashby, 2004).

The dependence of flexural rigidity on chitin crystallinity exhibits a more discernable trend (Fig. 3.5B) compared to the proportion of calcite. Returning to the literature, several studies indicate chitin is interconnected network of fibers separated from the mineral component by a layer of proteins (Raabe et al., 2006; Nagasawa, 2012). Results from this study suggest the chitin polysaccharide framework controls the rigidity of the exoskeleton. This challenges the current paradigm that CaCO3 has a reinforcing and strengthening role in the biocomposite

(Becker et al., 2005; Huber et al., 2015). Further, it raises the question why crustaceans are the only taxon amongst the arthropods that consistently incorporate a mineral component in the exoskeleton, despite the energetic cost of biomineralization. To answer this question, we first

54

compare the mechanical properties of the crustacean exoskeleton to that of other natural

materials.

3.3.6 Comparison to other materials

Comparison of the crustacean exoskeleton to mechanical properties of other natural

samples suggests its properties are similar to more flexible bio-composite and biominerals

(Ashby et al., 1995; Wegst and Ashby, 2004; Wegst et al., 2014). Specifically, Fig. B6A shows a

range of 10 to 200 in the initial failure strength, 1 to 50 for the flexural modulus, and a ratio of

σ/E = 0.01 (resilience), which is approximately equal to previous measurements of cuticular

materials. Its modulus is higher than more flexible natural materials, such as tendon, wool, and

other protein-based biocomposites. Initial failure is exceeds most polysaccharide-based

materials, such as most woods and timber. Other biomineral composites, such as , enamel,

and the mollusk shell, only show higher maximum values in one of the two quantities because of

much lower values of resilience. In other words, the crustacean exoskeleton more easily recovers

from large deflections compared to biomineral composites of other invertebrates.

Another material property that may be optimized in the crustacean exoskeleton is the

toughness or resistance to initial fracture, Ui. For most samples this value was between 0.1 and

1.5 kJm-2, which is a similar range as that for mollusk shells, enamel, and dentine (Fig. B6B).

This means that the exoskeleton fractures more easily compared to bone but is several orders of magnitude tougher than most crystalline materials. In summary, the crustacean exoskeleton is a highly optimized biocomposite that functions as both a strong and tough skeleton (protection against predators), as well as a resilient outer layer, akin to skin (allows for flexing due to breathing). Further, the crustacean exoskeleton is several orders of magnitude less flexible than the cuticle from other arthropods (i.e. insects and arachnids). Despite differences in thickness,

55

the large difference in mechanical properties establishes crustacean cuticle as a unique

biocomposite. However, no further measurements exhibited a dependence on the proportion of

calcite in the structure. This raises the question whether the mineral component is indicative of

any other exoskeleton properties.

3.3.7 Ecological Basis for Exoskeleton Structure

Careful examination of data presented in Fig. 1 leads to the definition of a new quantity,

the difference in amount of calcite between the chela and cephalothorax, for each of the species

in question (Fig. 3.6). In addition, three further species are included in this dataset to examine

possible influences on the selection of reinforcement strategies (Table 3.1).

Banded Coral Shrimp of order Stenopodidea is lobster-like in morphology, while the Fiddler and

Porcelain Crabs represent more crab-like body types. Despite their smaller size, these animals

exhibit differences in calcium carbonate crystallinity most similar to the larger animal of

comparable body type.

Table 3.1. Summary of all animals used in this study, sorted by defense strategy. Defense Strategy Common Name Species Name Taxon Fight Porcelain Crab Neopetrolisthes maculatus Anomura Red Reef Paguristes cadenati Dungeness Crab Metacarcinus magister Brachyura Red Rock Crab Cancer productus Fiddler Crab Uca uca Flight American Lobster Homarus americanus Astacidea Pacific Cleaner Shrimp Lysmata amboinensis Caridea Banded Coral Shrimp Stenopus hispidus Stenopodidea

Body morphology often corresponds to particular defense strategies that animals

commonly exhibit – a more sessile “fight” response is commonly associated with crab-like species, while shrimp- and lobster-like organisms often swim or run away. The latter strategies require a much more flexible and lighter body structure capable of withstanding short bursts of

56

energy. Crabs on the other hand, rely on a flexible body that is able to withstand the pressure

from a predator, while possessing strong claws to inflict damage to attackers, as well as prey.

Figure 3.6. Difference in calcite content between claw and cephalothorax as a function of species shows more mobile species exhibit smaller discrepancy in calcium carbonate crystallinity than more sessile animals. This suggests a possible ecological driving force for exoskeleton reinforcement and its underlying nano-structure. Independent of animal size, the more streamlined AL and BCS both exhibit a difference in calcite crystallinity of around 10% across the body. In contrast, crab-like body morphologies show an average difference in CaCO3 crystallinity of almost twice the magnitude.

Because of this tight linkage between morphology and mode of defense, it is often difficult to assess which may link more tightly to the different reinforcement strategies, as expressed by difference in calcite content. In addition, the animals are presented in decreasing divergence times from their closest relatives. BCS and AL and many organisms of similar

57

morphology represent older taxa that are estimated to have diverged 397 Ma and 211 Ma from

other Malacostraca, respectively (Qian et al., 2011). Brachyura and Anomura on the other hand,

are estimated to have developed 194Ma and 199 Ma (Dixon et al., 2003).

Regardless of deciding factor, however, we present our findings to indicate that difference in

structural reinforcement may present a proxy for evolutionary and/or ecological development

within Malacostraca. Due to rapid decomposition of cuticular material in most natural

environments, this method may become best suited in the analysis of extant organisms in an

attempt to establish additional lines of evidence for phylogenetic and behavioral relationships.

3.4 CONCLUSIONS AND IMPLICATIONS

Crustaceans of class Malacostraca exhibit highly variable nano-scale structures of

exoskeletal components, as determined by synchrotron and Raman analyses. Discrepancies in

polymeric order appear to correlate with a degree of reinforcement for certain species, while

other animals exhibit differences in localized structural homogeneity instead. These

measurements were established to be independent of chemistry and cuticle thickness, which may

have a large influence on averaged bulk analyses, such as those produced by high-energy x-ray

diffraction. Further, mechanical analysis revealed rigidity is controlled by chitin crystallinity and exoskeleton properties are independent of CaCO3 polymorph. As such, we can demonstrate that

lobsters and crabs use fundamentally different biomineralization strategies that may be linked to

evolutionary and ecological pressures.

This raises additional questions about biomineralization in other arthropods, as well as

closely related phyla. Unlike crustaceans, the insects, chelicerates, and myriapods

only mineralize chitin, with a few exceptions in Coleoptera (; Leschen and Cutler, 1994).

58

While this may shed light on the highly directed α-chitin deposition in crustaceans, it poses the

question: Who are the closest CaCO3 biomineralizers to arthropods and could they share a common genetically-controlled mechanism? Within the Protostomia, both and mollusk genera are amongst the most well-known biomineralizers specializing in calcium carbonate. Both of these phyla, however, also contain a large number of species with chitinuous extra-cellular

substances. In these are often present as amorphous chitin membranes, while some

mollusks appear to mineralize beaks, matrices, and primitive using the α-chitin

polymorph. This suggests that calcium carbonate biomineralization may be closely linked to the

evolution of polysaccharide support structures among invertebrates.

59

3.5 REFERENCES

ASTM D2344/D2344M-16 Standard Test Method for Short-Beam Strength of Polymer Matrix Composite Materials and Their Laminates, ASTM International, West Conshohocken, PA, 2016. ASTM D790-17 Standard Test Methods for Flexural Properties of Unreinforced and Reinforced Plastics and Electrical Insulating Materials, ASTM International, West Conshohocken, PA, 2017. Al-Sawalmih A., Li C., Siegel S., Fabritius H., Yi S., Raabe D., Fratzl P. and Paris O. (2008) Microtexture and Chitin/Calcite Orientation Relationship in the Mineralized Exoskeleton of the American Lobster. Adv. Funct. Mater. 18, 3307–3314. Ashby M. F., Gibson L. J., Wegst U. and Olive R. (1995) The Mechanical Properties of Natural Materials. I. Material Property Charts. Proc. R. Soc. A Math. Phys. Eng. Sci. 450, 123–140. Becker A., Ziegler A. and Epple M. (2005) The mineral phase in the cuticles of two species of Crustacea consists of magnesium calcite, amorphous calcium carbonate, and amorphous calcium phosphate. Dalt. Trans., 1814. Boßelmann F., Romano P., Fabritius H., Raabe D. and Epple M. (2007) The composition of the exoskeleton of two crustacea: The American lobster Homarus americanus and the edible crab Cancer pagurus. Thermochim. Acta 463, 65–68. Bouligand Y. (1972) Twisted Fibrous Arrangements in Biological Materials and Cholesteric Mesophases. Tissue Cell 4, 189–217. Chen P.-Y., Lin A. Y.-M., McKittrick J. and Meyers M. A. (2008) Structure and mechanical properties of crab exoskeletons. Acta Biomater. 4, 587–596. Compère P., Jaspar-Versali M. F. and Goffinet G. (2002) Glycoproteins from the cuticle of the Atlantic shore crab Carcinus maenas: I. Electrophoresis and western-blot analysis by use of lectins. Biol. Bull. 202, 61–73. Cribb B. W., Rasch R., Barry J. and Palmer C. M. (2005) Distribution of calcium phosphate in the exoskeleton of larval Exeretonevra angustifrons Hardy (Diptera: Xylophagidae). Arthropod Struct. Dev. 34, 41–48. Cribb B. W., Rathmell A., Charters R., Rasch R., Huang H. and Tibbetts I. R. (2009) Structure, composition and properties of naturally occurring non-calcified crustacean cuticle. Arthropod Struct. Dev. 38, 173–178. Dixon C. J., Ahyong S. T. and Schram F. R. (2003) A new hypothesys of decapod phylogeny. Crustaceana 76, 935–975. Ehrlich H. (2010) Chitin and as universal and alternative templates in biomineralization. Int. Geol. Rev. 52, 661–699. Fabritius H.-O., Sachs C., Triguero P. R. and Raabe D. (2009) Influence of Structural Principles on the Mechanics of a Biological Fiber-Based Composite Material with Hierarchical Organization: The Exoskeleton of the Lobster Homarus americanus. Adv. Mater. 21, 391– 400. Hammersley A. P., Svensson S. O., Thompson A., Graafsma H., Kvick Å. and Moy J. P. (1995) Calibration and correction of distortions in two-dimensional detector systems. Rev. Sci. Instrum. 66, 2729–2733. Hegdahl T., Gustavsen F. and Silness J. (1978) The Structure and Mineralization of the Carapace of the Crab (Cancer pagurus L.) 2. The exocuticle. Zool. Scr. 6, 101–105. Hild S., Marti O. and Ziegler A. (2008) Spatial distribution of calcite and amorphous calcium

60

carbonate in the cuticle of the terrestrial crustaceans Porcellio scaber and Armadillidium vulgare. J. Struct. Biol. 163, 100–108. Huber J., Griesshaber E., Nindiyasari F., Schmahl W. W. and Ziegler A. (2015) Functionalization of biomineral reinforcement in crustacean cuticle: Calcite orientation in the partes incisivae of the mandibles of Porcellio scaber and the supralittoral species Tylos europaeus (Oniscidea, Isopoda). J. Struct. Biol. 190, 173–191. Iconomidou V. A., Chryssikos G. D., Gionis V., Willis J.H., Hamodrakas S. J. (2001) “Soft”- cuticle protein secondary structure as revealed by FT-Raman, ATR FT-IR and CD spectroscopy. Insect Biochem. Mol. Biol. 31(9), 877-885. Klompmaker A. A., Hyžný M. and Jakobsen S. L. (2015) Taphonomy of decapod crustacean cuticle and its effect on the appearance as exemplified by new and known taxa from the Cretaceous–Danian crab Caloxanthus. Cretac. Res. 55, 141–151. Kumirska J., Czerwicka M., Kaczyński Z., Bychowska A., Brzozowski K., Thöming J. and Stepnowski P. (2010) Application of Spectroscopic Methods for Structural Analysis of Chitin and . Mar. Drugs 8, 1567–1636. Leschen R. A. B. and Cutler B. (1994) Cuticular Calcium in Beetles (Coleoptera: Tenebrionidae: Phrenapetinae). Ann. Entomol. Soc. Am. 87, 918–921. Nagasawa H. (2012) The crustacean cuticle: structure, composition and mineralization. Front. Biosci. 4, 711–720. Palmer A. R., Taylor G. M. and Barton A. (1999) Cuticle strength and the size-dependence of safety factors in Cancer crab claws. Biol. Bull. 196, 281–294. Qian G., Zhao Q., Wang A., Zhu L., Zhou K. and Sun H. (2011) Two new decapod (Crustacea, Malacostraca) complete mitochondrial genomes: Bearings on the phylogenetic relationships within the Decapoda. Zool. J. Linn. Soc. 162, 471–481. Raabe D., Romano P., Sachs C., Fabritius H., Al-Sawalmih A., Yi S. B., Servos G. and Hartwig H. G. (2006) Microstructure and crystallographic texture of the chitin-protein network in the biological composite material of the exoskeleton of the lobster Homarus americanus. Mater. Sci. Eng. A 421, 143–153. Roer R. and Dillaman R. (1984) The Structure and Calcification of the Crustacean Cuticle. Am. Zool. 24, 893–909. Rütt U., Beno M. A., Strempfer J., Jennings G., Kurtz C. and Montano P. A. (2001) Diffractometer for high energy X-rays at the APS. Nucl. Instruments Methods Phys. Res. A 467–468, 1026–1029. Skinner D. M., Kumari S. S., Brien J. J. O. and Brien J. J. O. (1992) Proteins of the Crustacean Exoskeleton. Am. Zool. 32, 470–484. Wegst U. G. K. and Ashby M. F. (2004) The mechanical efficiency of natural materials. Philos. Mag. 84, 2167–2186. Wegst U. G. K., Bai H., Saiz E., Tomsia A. P., Ritchie R. O., Ortiz C., Boyce M., Wegst U. G. K., Bai H., Saiz E., Tomsia A. P. and Ritchie R. O. (2014) Bioinspired structural materials. Nat. Mater. 14, 23–36. Zhang Z.-M., Chen S., Liang Y.-Z., Liu Z.-X., Zhang Q.-M., Ding L.-X., Ye F. and Zhou H. (2009) An intelligent background-correction algorithm for highly fluorescent samples in Raman spectroscopy. J. Raman Spectrosc. 41, 659–669.

61

CHAPTER 4. ESTABLISHING THE MG-DEPENDENT SOLUBILITY AND LOCAL STRUCTURE(S) OF AMORPHOUS CALCIUM CARBONATE (ACC)

Sebastian T Mergelsberg, James De Yoreo, Robert Ulrich, Marc Michel, Donald Rimstidt, (in prep.)

ABSTRACT This experimental study quantified the chemical and physical properties of ACC and its evolution to final products. We synthesized ACC under controlled chemical conditions using a flow-through reactor developed by our group (Blue et al., 2017, GCA). The experimental design varied Mg concentration and total alkalinity at an environmentally relevant pH. ACC solubility in the synthesis solution was measured at specific time points from super- and under-saturated conditions. The subsequent evolution of the ACC was monitored by small-angle x-ray scattering (SAXS). Parallel experiments characterized the composition and structure of the corresponding amorphous products using Thermogravimetric Analysis (TGA) and in situ pair distribution function (PDF) analyses. The measurements demonstrate at least two types of ACC are produced by adjusting Mg concentration and alkalinity. Each phase exhibits a different short-range order with distinct structural properties. TGA independently confirms two structure-dependent compositions. We also find temporal evolution of ACC structure and morphology after precipitation. This suggests composition of the final crystalline product is dependent on pathway/evolution of the amorphous intermediate. The findings hold promise for quantifying chemical and structural properties of ACC and reconciling discrepancies in the literature.

62

4.1 INTRODUCTION

Calcium carbonate minerals are an essential component in the skeletons of many invertebrates, such as sea urchins, mollusks, and crustaceans. Many of these animals have produced polymorphs of CaCO3 as biominerals over large periods of geologic time, thus recording patterns in both the evolution of the organism, as well as the chemical and isotopic composition of the environment and biological processes. The formation of minerals that comprise these skeletal structures often begins by precipitating an unstable, amorphous, intermediate phase as amorphous calcium carbonate (ACC). Understanding ACC properties — both chemical and physical — is critical to establishing the processes that govern formation, stabilization, preservation, and transformation pathways of CaCO3 phases during biomineralization. Recent advances in our understanding of crystallization via multistep pathways (De Yoreo et al., 2015) underline this need for the characterization of ACC and other metastable intermediates in crystallization.

Investigations of ACC show this amorphous intermediate evolves by multiple distinct forms (Politi et al., 2006; Lam et al., 2007; Radha et al., 2010; Cartwright et al., 2012; Gong et al., 2012; Radha et al., 2012). Several studies have identified differences in hydration state and

ACC metastability in both biological and synthetic samples. For example, photoelectron emission spectromicroscopy (PEEM) on growing sea urchin larval spicules reveals initial deposition of short-lived hydrated ACC that evolves to an anhydrous phase before ultimately transforming to calcite (Gong et al., 2012). Experiments on synthetic ACC in the pure CaCO3 system report multiple intermediate phases with different stabilities (Radha et al., 2010) and structures (Cartwright et al., 2012). However, most biogenic minerals form in the presence of

Mg, which is known to affect calcium carbonate stability. Results from Radha, 2012 suggest that

63

under Mg-bearing conditions, an additional type of amorphous carbonate is formed that contains

greater amounts of Mg than Ca. This Mg-phase is more short-lived and disappears after several

minutes to an hour (Lam et al., 2007).

Understanding the conditions that precipitate and stabilize different types of ACC in

biological conditions requires a basic framework of physical and chemical properties. The

solubility and short-range structure of the mineral is of particular interest to establish the

composition of the biological environment at the time of precipitation. The most widely cited

-6.4 value of Ksp,ACC = 10 at 25°C (Brečević and Nielsen, 1989) is about one hundred times

-8.48 greater than that of calcite, Ksp,calcite = 10 . A number of recent experimental studies suggest this number is too high with evidence that suggests a lower solubility form of ACC is also possible (Blue et al., 2013; Kellermeier et al., 2014).

Toward the broader goal of establishing the physical basis of CaCO3 biomineralization,

the precipitation and evolution of ACC via multiple intermediates is a driving question.

Ultimately, this knowledge is critical to establish the energetic and kinetic pathways that

regulation fractionation of elemental and isotopic signatures found in biominerals. Those CaCO3

precipitates that are Mg-dominated, hydrated, or have a more variable short-range order are of

particular interest (Mavromatis et al., 2017). For example, the fractionation of carbon and

isotopes is highly dependent on Mg content (Tarutani et al., 1969) and pH (Dietzel et al.,

2009) for calcite and aragonite. It is plausible that similar effects also modify isotope fractionation in biogenic ACC.

The purpose of this study is to establish the structural and chemical properties of Mg- stabilized ACC. Experiments were designed to focus on biologically relevant conditions of pH (8 to 9.5), ratios of CO3/Ca (2 and 8), and Mg/Ca (1 and 8). The relationships between structure

64

and chemistry were further explored by examining temporal changes over an interval of 30

seconds to 8 minutes. The experimental design used a mixed-flow reactor (MFR) for ACC synthesis under controlled conditions, followed by ultrafiltration. By conducting the synthesis in parallel with high energy x-ray scattering experiments, we establish the short-range order, composition, and solubility of two types of ACC. Analysis of the data suggests ACC evolution between types is dependent on the solution CO3/Ca ratio. Comparisons to a the calorimetric

study by Radha et al., (2012) indicate ACC solubility is controlled by both types of ACC.

4.2 MATERIALS AND METHODS:

4.2.1 Experimental Design

All syntheses were performed in a temperature-controlled glove-chamber (Coy, Grass

Lake, MI USA) at 25.0°C (±0.5°C) using a mixed-flow reactor (MFR) methodology adapted

from Blue et al. 2017, GCA. Advantages of the MFR over previous synthesis methods include:

(1) synthesis at constant, quantified supersaturation under controlled chemical conditions, (2)

continuous formation of the amorphous phase at steady- state conditions, and (3) production of

large amounts of ACC with reproducible compositions for characterization by complementary

methods, and (4) partial dissolution of ACC precipitates in secondary MFR to approach the

system from undersaturated conditions.

Two 100 mL reactant solutions were prepared with: (1) CaCl2·2H2O and MgCl2·6H2O,

and (2) NaHCO3 (all ACS reagent grade from Sigma-Aldrich, Saint-Louis, MO USA) using 18.2

MΩ·cm ultrapure water. Bicarbonate solutions were adjusted to pH 9.5 using 1.0 M NaOH

(Sigma-Aldrich). Exact concentrations and proportions are recorded in Table D4. The

65

experimental design was optimized to hold the final pH within 1.5 units of the initial value and the supersaturation with respect to ACC (Brečević and Nielsen, 1989) between 10 and 70:

(4.1) where x is the mole fraction of Mg in ACC and ai denotes the activity of each species in the initial solution. The solubility product K’sp is defined as:

(4.2) where at denotes the activity of each species in the solution after equilibrium has been reached.

Supersaturation as σ is also used as a measure of supersaturation throughout this study:

(4.3)

4.2.2 ACC synthesis and partial dissolution

To synthesize ACC, a syringe pump delivered 100 mL of each solution into the 25 mL reactor at flow rates to give average residence times of 30, 60, 120, and 240 seconds. Solutions within the reactor were continuously stirred at 800 rpm to ensure the system was properly mixed.

Effluent solutions were collected after 2.5 to 3 residence times, when ACC products and the solutions were assumed to have reached steady-state from supersaturaion (Jensen, 2001). To measure solubility from undersaturation, the outflow from the reactor was redirected into a secondary MFR together with 20% 18.2 MΩ·cm ultrapure water. The length of the connective tubing between reactors was adjusted to approximately double the total residence time for partial dissolution experiments. Stirring speed in both reactors was the same, and solutions were collected after another 2.5 to 3 residence times.

66

4.2.3 Isolation of ACC from suspension

After collecting 14 mL of the suspension, the ACC was separated from the solution by

centrifugal ultrafiltration using Corning Spin-X® UF 20 poly-ether sulfonate (PES) filters

(Corning, Corning, NY USA) at 8,000 x g for 10 minutes to achieve an approximate molecular

weight cut-off (MWCO) of 5kDa. Previous studies report this corresponds to particle sized of

around 1.1 to 5 nm in diameter (Karnik et al., 2009; Erickson, 2009). The pH of permeates was

measured immediately before titrating the solution. Solutions were stored in an incubator at 25 C

and titrations were conducted within 20 minutes of separation. ACC precipitates were washed

from the filter using ethanol and washed three times in an additional 50 mL ethanol. After each

washing, the ethanol-ACC suspension was centrifuged at 12,000 x g for 8 minutes and the

supernatant was discarded. Final powdered samples were air-dried for 3 hours. To produce an

acidified solution for elemental analysis, 2 mL of 0.25M HNO3 and 8 mL 18.2 MΩ·cm ultrapure

water were added to approximately 1 mg of dried ACC. All ion activities were calculated using

Geochemist’s Workbench (Aqueous Solutions LLC, Champaign, IL USA).

4.2.4 Titration

To measure the concentration of the bicarbonate/carbonate anion in solution, a 5 mL

sample of each permeate was titrated to pH=1.7 with 0.1 M HCl using a Titrando system

(Metrohm, Herisau, Switzerland) in a stirred, temperature-controlled vessel at 25 °C. The system

was configured to add titrants in increments of 10 μL, then wait a minimum of 1 to 38 s until the pH electrode stabilized to 20.0 mV/min after each addition. Titration curves were analyzed in

OriginLab Pro (Northampton, MA USA) and total carbonate concentrations were calculated using the volume of acid added between the two equivalence points.

67

4.2.5 Element analysis

Effluents from all filtrates and the acidified ACC samples were analyzed via ICP-OES to measure the total concentrations of Ca, Mg, and Na in the collected solutions. Analyses were performed using a Spectro® ARCOS SOP instrument using an yttrium internal standard.

Reference solutions were prepared independently to exclude impurities from HCl and HNO3.

Samples were analyzed in randomized order to avoid propagating any systematic experimental

error. Measured detection limits for Ca, Mg, and Na were 0.011 mg/L, 0.012 mg/L, and 0.027

mg/L respectively. The 1σ standard errors of the analyses calculated for each condition are

reported (Table D4).

4.2.6 Thermogravimetric Analysis

To synthesize the larger quantities of ACC required for TGA analysis, output from 5 one-

minute residence times were combined, washed three times in 50 mL ethanol and air-dried for 3

hours. One 6 mg sample from each condition was analyzed by TGA Q500 thermogravimetric

analyzer (New Castle, DE USA) at a rate of 10 C/min from room temperature to 860 C. For two

conditions, production of H2O and CO2 was confirmed with a Nicolet iS 10 FT-IR spectrometer connected to the TGA instrument.

4.2.7 X-ray Diffraction experiments and Rietveld Refinement of transformed samples

To produce transformed samples, the entire 200 mL output from a one-minute residence time experiment was collected and stirred at 250 rpm until transformation was confirmed (Table

D2). Each solid precipitate was analyzed by XRD for the presence of transformed crystalline

68

polymorphs. Small amounts of powder were scanned in a Rigaku Microflex II system from 20 to

60 2θ using Cu K-α emission. At regular intervals, pH of the solution was measured and 10 mL were centrifuged (8 min at 12,000 x g), washed three times with ethanol, and air-dried. All powders were scanned in a Rigaku Microflex II system from 20 to 60 2θ using Cu K-α emission.

Mineral composition was determined using Rietveld Refinement using Profex (Doebelin and

Kleeberg, 2015). Related statistics are listed in Table D2.

4.2.8 Scanning Electron Microscopy

ACC morphology was imaged by placing a small amount of precipitate from the solubility and transformation experiments onto tape and coating with 3-7 nm iridium using a Leica EM ACE600 sputter coater. All micrographs were acquired using a LEO 1550 field-emission SEM (Zeiss) with an acceleration voltage of 2 kV and a working distance of approximately 4.5 mm.

4.2.9 Raman Spectroscopy

Raman spectra are an additional line of evidence to confirm which compounds are lost when heating the sample to each of the temperatures. Subsamples of the dried ACC produced for

TGA analysis were placed in furnaces at 125, 300, and 500 C for 13, 30, and 50 minutes, respectively. Raman spectra from each of these heated samples and room-temperature ACC were compared acquired using a 514 nm Argon laser on a Horiba JobinYvon LabRam HR instrument.

69

4.2.10 Total X-Ray Scattering and Pair Distribution Function Analysis

Structural information was gathered using high-energy x-ray diffraction and Pair

Distribution Function (PDF) analysis. All experiments were performed at Beamline 11-ID-B

(Rütt et al., 2001) of the Advanced Photon Source (Argonne, IL USA) using an incident photon energy of ~58.6 keV (λ = 0.2114 Å). Samples were produced using the MRF method from section 2.2 with residence times of 4 and 8 minutes. Reactor output passed through approximately 30 cm of Tygon® tubing before entering a polyimide (Kapton®) capillary and the scattered radiation was measured in transmission mode using an amorphous-Si detector system manufactured by Perkin ElmerTM (2048 × 2048 pixels, 200 × 200 μm2 pixel size). The collection of total scattering data commenced <10 s after the sample passed in front of the detector and continued for approximately 25 to 45 minutes, until the reactant solutions were exhausted. A

CeO2 standard (NIST diffraction standard set 674a) was used to calibrate the sample-to-detector distance and the non-orthogonality of the detector relative to the incident beam path. Conversion of scattering data from 2D to 1D was performed using the program Fit2D (Hammersley et al.,

1995). A polarization correction was applied during integration of the data. Direct subtraction of the sample holder was accomplished by the independent measurement of the true background intensity of a water-filled polyimide capillary. Pair distribution profiles were calculated using the xPDF suite (Juhás et al., 2015).

4.3 RESULTS AND DISCUSSION

4.3.1 SEM reveals two independent morphologies

Observations of ACC morphology from 58 samples (total N=70) show two populations of particle sizes that differ in size by ≈1 to 2 orders of magnitude (Fig. 4.1). No intermediate particle sizes are observed, which likely indicates there are two distinct types of precipitate.

70

Particle size appears qualitatively to be independent of Mg content above 30 mol% MgCO3.

Careful examination of Table D4 reveals ACC compositions of most experiments enrich in Mg over time and with a progression toward an increasing amount of the small type of ACC.

However, quantitative analysis of particle size was not possible with this dataset, so a quantitative correlation cannot be drawn between the two characteristic sizes.

Figure 4.1. (A-D) Morphology of ACC reveals two distinct particle sizes produced under different conditions. (A) Very small ACC particles are observed in older samples and ACC that is above 30 mol% MgCO3. (B) Most 30-second experiments produce larger ACC particles that are observed most frequently other studies. (C) The majority of all experiments produce a mixture of morphologies, though the proportion of each particle size was not determined. (D) Experiments that produce large ACC spheres often accumulate smaller particles with increasing residence times. Scale bars are 1μm, scale bar in inset A is 200 nm. A detailed breakdown of morphology for each experiment is presented in Table D4.

71

4.3.2 TGA confirms presence of two amorphous products.

Weight loss measurements of 8-minute old samples show two distinct decomposition

temperatures at which CO2 is produced (Fig. 4.2). ACC that reaches a metastable equilibrium

with excess Ca2+ (in blue) exhibits only one major decomposition temperature (600-680 °C) for

CO2 and shows the presence of structural water. Note this structural water is released at a much higher temperature (190-240 °C) than adsorbed water (30-105 °C). ACC stoichiometry is given by:

(4.4)

Figure 4.2. TGA weight loss curves of 8-minute old samples confirm differences in morphology correspond to two different ranges of ACC composition. Samples that reach steady-state at conditions where calcium activity outweighs carbonate activity (blue) contain about 50 mol% H2O. ACC that evolves in carbonate-dominated solutions (red) does not contain strongly bound water and loses most of its mass at 400 °C.

72

In contrast, ACC that formed in solutions with aCa2+ > aCO32- (blue curve) appear mostly

2- anhydrous and show a high proportion of CO3 decomposition takes place at a much lower temperature. The bulk composition of ACC produced at these conditions is:

(4.5)

Previous studies (Rodriguez-Blanco et al., 2012; Radha et al., 2012) attribute the anhydrous phase to the presence of a separate amorphous magnesium carbonate (AMC). Measurements from this study confirm a higher proportion of Mg in these samples, but IR and Raman spectroscopy also show the presence of some bicarbonate.

Given that ACC formed under these conditions is a physical mixture of two types, the isolated composition of a Mg-ACC phase could not be measured. Assuming Eq. 4.4 corresponds to the pure hydrated ACC stoichiometry, however, the anhydrous ACC composition can be estimated by substituting Eq. 4.4 into Eq. 4.5 to give:

(4.6)

All TGA results are summarized in Table D1. Only those samples prepared at the largest aCO32-

/aCa2+ ratio produced what appears to be mostly hydrated ACC.

4.3.3 In situ PDF analysis quantifies structural differences

The short-range order of the two types of ACC is also different (Fig. 4.3). The sharpest peaks in the PDFs that occur in the 1.3 to 2.4 Å range contain information on the first neighbor coordination environments around carbon∼ and calcium. The PDF profile for material precipitated from high carbonate solutions (red) indicates a precipitate with a highly regular Ca-O short-range ordering at 2.36 Å, but almost no regular recurring Mg-O distance that would be expected at

≈2.13 A (Fig. 3.3A). Other features that define the ordering of this ACC are mostly above 3 Å,

73

which describe cation-cation pairs, as well as more complex Ca-O pairs. In contrast, the profile

generated for ACC that forms in a solution with excess Ca2+ (blue) exhibits a high density of Ca-

O and Mg-O distances at 2.36 Å and 2.13 Å. Further, there is a large feature at 3.2 Å due to

secondary Mg-O pairing. Although the coherence of these precipitates is only weakly visible

beyond 7 Å, vertically exaggerated profiles reveal further differences between the two types of

ACC even at these distances. Amorphous materials inherently contain disorder, so it is expected

these features are less well resolved in the PDF profiles and may be artifacts related to the

Fourier transformation. Atom pair distances for all other samples are summarized in Fig. D2.

Figure 4.3. (A and B) In situ PDF profiles of ACC at high (red) and low (blue) carbonate activities after 4 minutes residence time. (A) At this condition, more Mg is incorporated into the ACC, but the small peak at 2.1 Å indicates it is not coordinated in a consistent manner. (B) Despite lower Mg content, ACC under these conditions exhibits a short-range order that features a reoccurring Mg-O distance. Insets show an expanded vertical scale and reveal some mid-range features that are not consistent between the two samples (7 to 15 Å). Dashed lines highlight distinct features of (B).

74

Because all measurements are made in suspension, it is necessary to confirm the PDF profiles contains information of the solids and not just ion pairs in solution (Fig. D3). In general, we observe the red profile bears some resemblance in terms of distance-distribution to bicarbonate solution. Similarly, the blue profile contains similar features to the Ca/Mg chloride solution. This concurs with the observation that the corresponding ions are the dominant species in solution after the precipitation. However, the profiles from concentrated solutions exhibit distinctly different atom-pair distances that are offset from the ACC distances by several picometers. The mismatch in feature resolution and peak positions confirms that the ACC profiles describe an amorphous solid and not simply the solution surrounding it.

Previous studies have reported that certain short-range features of ACC can align with atom-atom pairs in crystalline calcium carbonate polymorphs (e.g. Cartwright et al., 2012). In this study, final transformation products were determined to be calcite, aragonite, and nesquehonite (Table D2). Comparison of ACC PDF profiles to calculated standards does not reveal any short-range features that are present in the polymorphs. Instead, the sharp features in the short-range order of ACC bear resemblance to Ca-O and C-O pairs in calcite and aragonite, but are decidedly offset (Fig. D4). Broader features in the PDF of the amorphous intermediates around 6, 10 and 12 Å align more generally to several features in the , which may be more indicative of general mid-range ordering present in many carbonates (Michel et al., 2008).

4.3.4 Evolution of ACC is sensitive to total aCO3/aCa

TGA and PDF analyses confirm the presence of two distinct types of ACC in almost all experiments. Their main distinguishing features are magnesium content and hydration state. All samples evolve to a mixture of ACC types. In most cases, a Ca-dominated phase precipitates first, which evolves to a mixture that has intermediate Mg/Ca-O distance or two distinct

75

Figure 4.4. (A and B) ACC evolves to an end-member with a more stable Mg-O short-range order over time. (A) The Ca-O distance at 2.36 Å evolves to a shorter distance as more Mg is regularly incorporated in the short-range order. The 8-minute sample is similar in ordering to the low-carbonate condition (B). Few differences are observed in these second samples over time. Similar differences are visible for mid-range features at an expanded vertical scale (7 to 15 Å). Dashed lines highlight distinct features of (B).

coordination environments (Fig. D2). The Ca-limited condition (red) exhibits a highly regular

Ca-O pair distance at 4 minutes residence time that is observed closer to an intermediate value

between Ca-O and Mg-O after a total of 8 minutes residence time (Fig. 4.4A). In general, the

76

PDF profile evolves to feature ion-ion pairs similar to the PDFs of the mixed ACC condition

(blue). Certain mid-range features disappear altogether in favor of shorter average distances.

Under conditions with excess Ca2+, profiles after 4- and 8-minute residence times are

almost indistinguishable (Fig 4B). This suggests the mixture of ACCs with both regular Ca-O

and Mg-O pairs – or a single Mg-rich ACC – is the most metastable over time. This is further

supported by comparing Figs. D2A and D2B. Conditions B, C, and E all reach equilibrium at Ca-

limited conditions and precipitate Ca-dominated ACC after 4 minute residence time. At 8-minute

residence times, however, all samples have reached very similar short-range ordering. Mid-range

ordering is fairly consistent, as well, suggesting there are no ordered domains evolving at this

time.

4.3.5 Solubility is dependent on Mg content for both types of ACC

Apparent ACC solubility shows a strong dependence on Mg content in the solid (Fig 4.5

A) and summary in Table D4). Based on solubility of crystalline calcium carbonates, this is

expected (Berner, 1975). Extrapolation predicts the solubility of pure ACC (0% Mg) is similar to

-6.4 the previous measurement of Ksp = 10 reported by (Brečević and Nielsen, 1989). The linear trend continues to > 50 mol% MgCO3 and suggests one continuous solubility between the pure

CaCO3 and MgCO3 endmembers. Similarly, the pH and residence time of each experiment does

not affect apparent solubility in a significant manner and is approximately constant for all tested

conditions (Fig. D1).

77

Figure 4.5. (A) Solubility product of ACC as a function of Mg content describes a strong linear dependence between 10 and 70 mol% MgCO3. Measurements from super- and undersaturation are in good agreement from all initial compositions and MFR residence times. (B) Comparing the Mg partitioning between experiments reveals Mg content of ACC is independent of Mg in solution under certain conditions. N=70.

However, the previous discussion shows there are two ACC phases with distinct

morphologies, stoichiometries, and structures (section 4.3.2). Thus, it is expected the chemistry

of the two end-member phases will be distinct, as well. The partitioning of Mg into the solid, for

example, does not fall onto a single trend (Fig 4.5B). Rather, there appears to be a horizontal feature in this figure that suggests Mg/Casolution is constant for the majority of ACC precipitates, regardless of Mg content. Samples that reach a metastable state where the activity of carbonate exceeds that of calcium (data points in red) are very high in Mg with a very narrow Mg/Ca ratio in solution. On the other hand, ACC that contains less Mg exhibits a narrow range of Mg/Casolid values at which the solution chemistry varies greatly. It follows then that there are chemical differences between the different types of ACC, as predicted by differences in short-range order and composition (sections 4.3.2 and 4.3.3). The increased partitioning of Mg into some of the

78

solids does not cause a change in the apparent solubility, however, because Mg/Casolution appears

to be buffered for a large number of conditions. These observations raise the question whether

the solubility product of ACC is either (1) similar for both types of ACC or (2) controlled by one of the two types.

4.3.6 Stoichiometry predicts solubility trend for each type of ACC

To answer this question, we estimate the expected differences in apparent solubility

between the two types of ACC based on end-member stochiometries (section 4.3.2). For the

hydrated ACC end-member (Eq. 4.4), the apparent solubility K’sp is given by:

(4.7)

where ai denotes the activity of each species at steady-state. Similarly, the composition estimated

by Eq. 4.6 can be used to calculate the apparent solubility of the anhydrous ACC phase:

(4.8)

The apparent solubility as a function of magnesium content for each experiment is summarized

for both types of ACC in Fig. D5, Appendix D. Estimated solubilities of the two types of ACC

are tightly distributed and differ by approximately half an order of magnitude.

Comparison to Fig. 4.5A reveals the K’sp of the bulk precipitate may be that of a physical

mixture of end-member solubilities. For example, at low Mg content, the solubility of the

hydrated ACC at low-Mg is very similar to that of the bulk ACC (Fig. D5A). Similarly,

solubility of the anhydrous ACC at high Mg contents is approximately log K’sp = -5.7, the same

range of values determined in Fig. 4.5A. Thus, it is likely that the bulk solubility simply reflects

the physical mixture of two ACC phases. However, recognizing that ACC in most experiments

79

evolved from a one-phase to a two-phase system without a significant difference in apparent

solubility between reaction times, it is unlikely the proportion of ACC type alone controls

solubility. Variability in phase stoichiometry may account for the near-constant solubility values

as the ACC evolves, but it raises the question whether there is a thermodynamic basis for the

observed solubility values.

4.3.7 Thermodynamic constraints on ACC solubility

Returning to the literature, we find Radha et al. (2012) measured the enthalpies of

solution of ACC with similar compositions and short-range order compared to the solids

produced in this study. Fig. 4.6A shows that the energy of the single-phase ACC and the two-

phase ACC are different and exhibit distinct relationships to Mg content. Dissolution of the

hydrated, Ca-dominated ACC (purple) is less exothermic than dissolution of a mixture of phases.

Further, the authors note that the slope of the dependence changes around 45 mol% MgCO3,

indicating there is a specific point at which the mixture of ACC types is more stable than the

single phase. To compare this data with our measurements of solubility, we estimated the

apparent solubilities associated with the enthalpies of solution (See Table D3).

Comparison of the calculated solubilities to measurements of apparent ACC solubility

from this study independently confirm that the less metastable single-phase ACC controls the

relationship between K’sp and Mg content (Fig. 6B). Despite a small offset, the slope of the calculated regression is approximately equal to the measurements in this study. Even though

ACC will eventually evolve to the more metastable two-phase ACC condition, any initially formed Ca-dominated phase will control solubility. Furthermore, in the range of our

80

Figure 4.6. Comparison to previous study independently confirms solubility is controlled by the less metastable type of ACC. (A) Measurements from Radha et al. (2012) show energetic stabilities of the homogenous and heterogenous ACC to be different (N = 22). (B) Results from this study suggest the measured solubility is controlled by the initially formed ACC. Further, the dependence of solubility on magnesium content is similar between the two types of ACC and measurements cannot resolve a possible change in slope. The data for this figure is presented in Table D3, Appendix D. Shaded regions correspond to the 2σ standard error.

experiments, the dependence of solubility on magnesium content is similar between the two

types of ACC and measurements cannot resolve a possible change in slope. It is possible that the

solubility of ACC with more than 70 mol% MgCO3 would be approximately 10-5.5, no such

conditions were measured in this study, however.

4.4 CONCLUSIONS

This quantitative investigation establishes the structure, composition, and morphology of two different types of ACC in the presence of Mg. Measurements also identify an evolution between the two phases, from a Ca-dominated single-phase ACC to a mixture of both phases.

81

The relative stability of the two types is likely correlated to the activity of the carbonate anion.

Apparent solubility of ACC, however, is determined to be linearly dependent on Mg content of the solid mixture, regardless of which type of ACC is present. The fractionation of Mg between the two phases is indicative of a dynamic partitioning between the two types of ACC. A comparison of our data to previously published results suggests solution chemistry is controlled by the less metastable Ca-dominated ACC. Results from this study provide a quantitative basis

for deciphering relationships between ACC structures, solution chemistry, and the final

transformation products in the presence of Mg. The higher metastability of the anhydrous ACC

identified in this investigation has been noted in biological systems by several studies, for

example in echinoderms (Gong et al., 2012) and (Mass et al., 2017). This suggests findings from this study may hold promise for reconciling different types of ACC identified by previous studies and in different compositional regimes.

82

4.5 REFERENCES

Berner R. A. (1975) The role of magnesium in the crystal growth of calcite and aragonite from sea water. Geochim. Cosmochim. Acta 39. Blue C. R., Giuffre A., Mergelsberg S., Han N., De Yoreo J. J. and Dove P. M. (2017) Chemical and physical controls on the transformation of amorphous calcium carbonate into crystalline CaCO3 polymorphs. Geochim. Cosmochim. Acta 196. Blue C. R., Rimstidt J. D. and Dove P. M. (2013) A Mixed Flow Reactor Method to Synthesize Amorphous Calcium Carbonate Under Controlled Chemical Conditions. In Methods in Enzymology Elsevier Inc. pp. 557–568. Brečević L. and Nielsen A. E. (1989) Solubility of amorphous calcium carbonate. J. Cryst. Growth 98, 504–510. Cartwright J. H. E., Checa A. G., Gale J. D., Gebauer D. and Sainz-Díaz C. I. (2012) Calcium carbonate polyamorphism and its role in biomineralization: How many amorphous calcium carbonates are there? Angew. Chemie - Int. Ed. 51, 11960–11970. De Yoreo J. J., Gilbert P. U. P. A., Sommerdijk N. A. J. M., Penn R. L., Whitelam S., Joester D., Zhang H., Rimer J. D., Navrotsky A., Banfield J. F., Wallace A. F., Michel F. M., Meldrum F. C., Colfen H. and Dove P. M. (2015) Crystallization by particle attachment in synthetic, biogenic, and geologic environments. Science 349, 498. Dietzel M., Tang J., Leis A. and Köhler S. J. (2009) Oxygen isotopic fractionation during inorganic calcite precipitation ― Effects of temperature, precipitation rate and pH. Chem. Geol. 268, 107–115. Doebelin N. and Kleeberg R. (2015) Profex: A graphical user interface for the Rietveld refinement program BGMN. J. Appl. Crystallogr. 48, 1573–1580. Erickson H. P. (2009) Size and Shape of Protein Molecules at the Nanometer Level Determined by , Gel Filtration, and Electron Microscopy. Biol. Proced. Online 11, 32–51. Gong Y., Killian C., Olson I., Appathurai N. P., Amasino A., Martin M., Holt L., Wilt F. and Gilbert P. (2012) Phase transitions in biogenic amorphous calcium carbonate. Proc. Natl. Acad. Sci. U. S. A. 109, 6088–6093. Hammersley A. P., Svensson S. O., Thompson A., Graafsma H., Kvick Å. and Moy J. P. (1995) Calibration and correction of distortions in two-dimensional detector systems. Rev. Sci. Instrum. 66, 2729–2733. Jensen J. N. (2001) Approach to steady state in completely mixed flow reactors. J. Environ. Eng. 127, 13–18. Juhás P., Farrow C. L., Yang X., Knox K. R. and Billinge S. J. L. (2015) Complex modeling: A strategy and software program for combining multiple information sources to solve ill posed structure and nanostructure inverse problems. Acta Crystallogr. Sect. A Found. Adv. 71, 562–568. Karnik B. S., Baumann M. J., Corneal L. M., Masten S. J. and Davies S. H. (2009) TEM characterization of -oxide-coated ceramic membranes. J. Mater. Sci. 44, 4148–4154. Kellermeier M., Picker A., Kempter A., Cölfen H. and Gebauer D. (2014) A Straightforward Treatment of Activity in Aqueous CaCO 3 Solutions and the Consequences for Nucleation Theory. Adv. Mater. 26, 752–757. Lam R. S. K., Charnock J. M., Lennie A. and Meldrum F. C. (2007) Synthesis-dependant structural variations in amorphous calcium carbonate. CrystEngComm 9, 1226. Mass T., Giuffre A. J., Sun C.-Y., Stifler C. A., Frazier M. J., Neder M., Tamura N., Stan C. V.,

83

Marcus M. A. and Gilbert P. U. P. A. (2017) Amorphous calcium carbonate particles form coral skeletons. Proc. Natl. Acad. Sci., 201707890. Mavromatis V., Purgstaller B., Dietzel M., Buhl D., Immenhauser A. and Schott J. (2017) Impact of amorphous precursor phases on magnesium isotope signatures of Mg-calcite. Earth Planet. Sci. Lett. 464, 227–236. Michel F. M., MacDonald J., Feng J., Phillips B. L., Ehm L., Tarabrella C., Parise J. B. and Reeder R. J. (2008) Structural characteristics of synthetic amorphous calcium carbonate. Chem. Mater. 20, 4720–4728. Politi Y., Levi-Kalisman Y., Raz S., Wilt F., Addadi L., Weiner S. and Sagi I. (2006) Structural Characterization of the Transient Amorphous Calcium Carbonate Precursor Phase in Sea Urchin Embryos. Adv. Funct. Mater. 16, 1289–1298. Radha A. V., Fernandez-Martinez A., Hu Y., Jun Y.-S., Waychunas G. A. and Navrotsky A. (2012) Energetic and structural studies of amorphous Ca1−xMgxCO3·nH2O (0 x 1). Geochim. Cosmochim. Acta 90, 83–95. Radha A. V., Forbes T. Z., Killian C. E., Gilbert P. U. P. A. and Navrotsky A. (2010)⩽ ⩽ Transformation and crystallization energetics of synthetic and biogenic amorphous calcium carbonate. Proc. Natl. Acad. Sci. 107, 16438–16443. Rodriguez-Blanco J. D., Shaw S., Bots P., Roncal-Herrero T. and Benning L. G. (2012) The role of pH and Mg on the stability and crystallization of amorphous calcium carbonate. J. Alloys Compd. 536, S477–S479. Rütt U., Beno M. A., Strempfer J., Jennings G., Kurtz C. and Montano P. A. (2001) Diffractometer for high energy X-rays at the APS. Nucl. Instruments Methods Phys. Res. Sect. A Accel. Spectrometers, Detect. Assoc. Equip. 467–468, 1026–1029. Tarutani T., Clayton R. N. and Mayeda T. K. (1969) The effect of polymorphism and magnesium substitution on oxygen isotope fractionation between calcium carbonate and water. Geochim. Cosmochim. Acta 33, 987–996.

84

APPENDIX A. SUPPLEMENTARY INFORMATION FOR CHAPTER 2

A1. SUPPLEMENTARY FIGURES

Figure A1. Composition of the organic matrix fraction of the exoskeleton with respect to: (A) wt% Ca vs. wt% Mg; (B) wt% Ca vs. wt% P; and (C) wt% Mg vs. wt% P. Detailed statistics are given in Table A3, Appendix A.

85

A2. COMPOSITIONAL DATA

Table A1. Summary of concentrations and molar ratios of Ca, Mg, and P in all body parts investigated in this study. The mineral fraction (last column) is defined as wt% of the bulk sample mass. The error values correspond to the 1σ (68%) standard error of the mean. Mineral Ca Mg P Mg/ P/ P/ fraction (wt%) (wt%) (wt%) Ca Ca Mg (wt%) Dominant chela (n=36) 33.6 (±2.1) bulk 8.957 ±0.277 0.605 ±0.009 1.062 ±0.035 0.113 ±0.002 0.156 ±0.007 1.378 ±0.042 organic 0.293 ±0.109 0.009 ±0.002 0.113 ±0.043 0.093 ±0.013 0.824 ±0.135 8.969 ±0.830 mineral 26.871 ±1.399 1.824 ±0.068 2.975 ±0.082 0.114 ±0.002 0.149 ±0.008 1.300 ±0.041 Non-dom. chela (n=36) 31 (±1.2) bulk 9.425 ±0.273 0.602 ±0.013 0.925 ±0.031 0.106 ±0.002 0.131 ±0.009 1.219 ±0.057 organic 0.038 ±0.012 0.002 ±3.7E-04 0.022 ±0.004 0.175 ±0.026 1.579 ±0.245 9.760 ±1.015 mineral 30.591 ±1.086 1.952 ±0.051 2.957 ±0.106 0.106 ±0.002 0.130 ±0.009 1.202 ±0.057 Legs (n=36) 32.5 (±1.8) bulk 8.474 ±0.677 0.449 ±0.036 0.674 ±0.060 0.087 ±3.9E-04 0.101 ±0.002 1.162 ±0.021 organic 0.185 ±0.051 0.005 ±0.001 0.094 ±0.021 0.072 ±0.008 0.919 ±0.082 13.939 ±0.933 mineral 25.829 ±2.152 1.376 ±0.115 1.883 ±0.157 0.088 ±4.8E-04 0.094 ±0.001 1.073 ±0.015 Cephalon (n=18) 41 (±8.4) bulk 10.265 ±0.483 0.474 ±0.023 0.634 ±0.022 0.076 ±0.001 0.080 ±0.001 1.056 ±0.013 organic 0.075 ±0.020 0.006 ±0.002 0.040 ±0.003 0.121 ±0.010 0.955 ±0.145 8.424 ±1.295 mineral 26.457 ±1.882 1.221 ±0.093 1.600 ±0.135 0.076 ±0.001 0.078 ±0.001 1.024 ±0.010 Cephalothorax (n=18) 18.2 (±0.1) bulk 6.915 ±0.527 0.368 ±0.020 0.520 ±0.037 0.089 ±0.002 0.098 ±0.001 1.102 ±0.016 organic 0.095 ±0.021 0.003 ±3.5E-04 0.046 ±0.009 0.061 ±0.006 0.663 ±0.024 11.507 ±1.026 mineral 37.560 ±2.840 2.005 ±0.111 2.646 ±0.164 0.089 ±0.002 0.092 ±0.001 1.033 ±0.007 Abdomen (n=18) 32.4 (±17.6) bulk 11.933 ±0.027 0.589 ±0.003 1.103 ±0.004 0.081 ±3.0E-04 0.120 ±2.3E-04 1.469 ±0.003 organic 0.032 ±2.7E-04 0.003 ±0 0.049 ±3.8E-05 0.177 ±0.002 1.961 ±0.018 11.108 ±0.009 mineral 27.944 ±0.063 1.377 ±0.006 2.521 ±0.009 0.081 ±3.0E-04 0.117 ±2.3E-04 1.436 ±0.003 Uropods (n=18) 42.2 (±0.4) bulk 5.941 ±0.012 0.289 ±0.001 0.489 ±0.001 0.080 ±2.7E-04 0.107 ±2.6E-04 1.328 ±0.001 organic 0.020 ±7.0E-05 0.002 ±0 0.016 ±5.4E-06 0.124 ±4.3E-04 1.012 ±0.003 8.167 ±0.003 mineral 22.086 ±0.043 1.072 ±0.004 1.766 ±0.005 0.080 ±2.7E-04 0.103 ±2.6E-04 1.292 ±0.001

86

Table A2. Summary of previously reported ratios of Mg/Ca and P/Ca contained in the bulk exoskeletons of other crustaceans (e.g. Fig. 2.6B). Organism Mg/Ca P/Ca Body parts References Am. Lobster 0.140 0.182 Claw Clarke and Wheeler, 1922 H. americanus 0.108 0.126 Cephalothorax 0.107 0.046 Claw Boßelmann et al., 2007 0.095 0.065 Cephalothorax Red rock crab 0.098 0.050 Claw C pagurus 0.090 0.065 Cephalothorax Calif. brown shrimp 0.145 0.178 Rostrum Huner et al., 1979 P. californiensis 0.131 0.193 Cephalothorax 0.115 0.196 Abdomen Woodlouse 0.048 0.037 Carapace (Becker et al., 2005) P.scaber Pillbug 0.040 0.026 Carapace A vulgare

87

Table A3. Statistical analysis of linear regression models for each figure. The rows in bold font correspond to linear fits that are shown in the corresponding plots. Figure trend n slope error (1σ) intercept error (1σ) R2 Adj. R2 F statistic p-value 2.3 A whole body 30 0.0921 0.0079 0.0846 0.2202 0.8353 0.8292 136.951 4.40E-12 body w/o chelae 18 0.0865 0.0036 -0.044 0.0976 0.9745 0.9728 573.7475 2.28E-13 chelae 12 0.0728 0.0078 1.0219 0.2289 0.8964 0.886 86.493 3.08E-06 2.3 B whole body 30 0.0884 0.0166 0.5777 0.4639 0.5127 0.4947 28.4095 1.26E-05 body w/o chelae 18 0.0865 0.0052 0.1184 0.1399 0.949 0.9456 278.887 4.23E-11 chelae 12 0.0093 0.0289 3.5707 0.8455 0.0102 -0.0888 0.1033 0.755 2.3 C whole body 30 1.1126 0.0986 0.1093 0.2643 0.8251 0.8187 127.4073 9.94E-12 body w/o chelae 18 1.0007 0.0403 0.1597 0.0928 0.9763 0.9747 616.7277 1.34E-13 chelae 12 0.3514 0.3612 2.7438 1.1349 0.0864 -0.0049 0.9461 0.354 2.4 A whole body 30 0.0131 0.0389 0.093 0.0053 0.004 -0.0315 0.1134 0.739 body w/o chelae 18 -0.0485 0.0249 0.089 0.0028 0.1913 0.1407 3.7844 0.070 chelae 12 -0.0178 0.0337 0.1124 0.0055 0.0272 -0.0701 0.2796 0.608 2.4 B whole body 30 -0.0027 0.0081 0.1148 0.0113 0.0038 -0.0318 0.1075 0.745 body w/o chelae 18 0.0027 0.0041 0.0902 0.0054 0.027 -0.0338 0.4435 0.515 chelae 12 -0.0098 0.0124 0.1513 0.0186 0.0592 -0.0348 0.6296 0.446 2.4 C whole body 30 -0.0142 0.0073 1.3218 0.0857 0.1199 0.0884 3.8138 0.061 body w/o chelae 18 4.E-04 0.0072 1.1053 0.091 2.E-04 -0.0623 0.0025 0.961 chelae 12 -0.0307 0.0143 1.5391 0.1447 0.3161 0.2477 4.6212 0.057 2.5 A whole body 30 0.0544 0.0069 0.0228 0.0608 0.6885 0.6773 61.876 1.44E-08 body w/o chelae 18 0.0475 0.0027 0.0253 0.0228 0.9508 0.9477 308.9847 6.93E-12 chelae 12 0.0314 0.0077 0.315 0.0709 0.6281 0.5909 16.8854 0.002 2.5 B whole body 30 0.0743 0.0194 0.1327 0.1699 0.3445 0.3211 14.7159 6.51E-04 body w/o chelae 18 0.072 0.0096 0.0321 0.0809 0.7788 0.765 56.3481 1.25E-06 chelae 12 -0.0564 0.0378 1.5123 0.3505 0.1819 0.1001 2.2234 0.167 2.5 C whole body 30 1.6788 0.1792 -0.052 0.0906 0.7582 0.7496 87.7998 3.97E-10 body w/o chelae 18 1.5502 0.1581 -0.0205 0.0673 0.8573 0.8484 96.1521 3.61E-08 chelae 12 -0.0582 1.0543 1.0288 0.6386 3.E-04 -0.0997 0.003 0.957 2.5 D whole body 30 1.0167 0.0142 -0.0012 0.0013 0.9946 0.9944 5161.311 2.66E-33 body w/o chelae 18 1.0391 0.0246 -0.0031 0.0021 0.9911 0.9906 1786.809 7.59E-18 chelae 12 0.9957 0.05 0.0011 0.0055 0.9754 0.973 397.0707 2.22E-09 2.5 E whole body 30 0.9797 0.033 -0.0025 0.004 0.9692 0.9681 881.7727 1.04E-22 body w/o chelae 18 0.8556 0.0686 0.009 0.0068 0.9067 0.9008 155.4391 1.18E-09 chelae 12 0.9635 0.0678 9.00E-04 0.0101 0.9529 0.9482 202.1558 5.84E-08 2.5 F whole body 30 0.8987 0.0708 0.0669 0.0876 0.852 0.8467 161.1731 3.90E-13 body w/o chelae 18 0.8822 0.092 0.0748 0.1085 0.8518 0.8426 91.9953 4.89E-08 chelae 12 0.8629 0.1259 0.1307 0.1659 0.8244 0.8068 46.9459 4.45E-05 2.6 A whole body 30 2.028 0.2063 -0.0742 0.0196 0.7754 0.7674 96.68 1.4E-10 body w/o chelae 18 1.2449 0.5167 -0.006 0.0435 0.2662 0.2204 5.608 0.028 chelae 12 3.3103 0.4923 -0.2184 0.0541 0.8189 0.8008 45.22 5.21E-05 2.6 B all species 18 1.5811 0.3057 -0.0403 0.0309 0.6258 0.6024 26.7525 9.26E-05 A1 A whole body 30 0.0353 0.0057 0.0032 9.E-04 0.5929 0.5772 37.8606 1.66E-06 body w/o chelae 18 0.0313 0.0076 0.0038 0.0014 0.5289 0.4975 16.8424 9.39E-04 chelae 12 0.0574 0.0059 0.0014 6.E-04 0.9132 0.9036 94.7149 4.48E-06 A1 B whole body 30 0.4953 0.0386 0.0168 0.0059 0.8637 0.8584 164.7274 9.31E-13 body w/o chelae 18 0.5054 0.0454 0.024 0.0082 0.8919 0.8847 123.7987 1.21E-08 chelae 12 0.3025 0.0258 0.016 0.0024 0.9388 0.932 137.9756 9.24E-07 A1 C whole body 30 8.585 1.5412 0.0075 0.0131 0.5441 0.5266 31.0299 7.51E-06 body w/o chelae 18 9.1263 2.1734 0.0134 0.0208 0.5403 0.5097 17.6328 7.74E-04 chelae 12 4.7633 0.692 0.0113 0.0045 0.8404 0.8227 47.3858 7.20E-05

88

APPENDIX B. SUPPLEMENTARY INFORMATION FOR CHAPTER 3

B1. REFERENCES FOR X-RAY DIFFRACTION AND RAMAN SPECTROSCOPY

Table B1. D-spacing references for chitin and calcite. d (Å) Material Reflection 10.309 Chitin 020 7.098 Chitin Amorphous 5.122 Chitin 120 4.580 Chitin 110 3.827 Calcite 012 3.389 Chitin 130 2.480 Calcite 104 2.271 Calcite 2.084 Calcite 202 1.594 Calcite 122 1.432 Calcite 300 1.241 Calcite 220

Table B2. Raman peak reference table for common chitin and calcite vibrations Raman Shift (cm-1) Bond 281 T(Ca, CO3) 711 ν4-Symmetric CO3 deformation 1085 ν1-Symmetric CO3 stretching 1206 Amide III 1264 Amide III 1326 Amide III 1372 Rocking CH2 1414 δ CH2 1448 δ CH2 1624 Amide I 1660 Amide I 2883 CHx stretching 2937 CHx stretching 2965 CHx stretching

89

B2. SUPPLEMENTARY FIGURES

Figure B1. High energy x-ray scattering of exoskeleton allow for the calculations of chitin crystallinity (brown). The chitin amorphous region is a reflection along the [021] plane and is invisible at this scale, as it only enriches at very low crystallinity values.

90

Figure B2. PDF-determined abundance of calcite shows little correlation to sample thickness in 5 organisms. While all members of Brachyura reveal an increased calcite content in their claws, the lobster and shrimp samples exhibit opposite trends. Thus, exocuticle thickness contains little information about nanostructure and order. As observed in Fig. 2.1, polymeric order is independent of between orders of Malacostraca. Samples from body parts of each organism are divided into:  chela (claw) and  cephalothorax (main body cavity).

91

Figure B3. A and B) Raman spectra of the endocuticle from the chela (claw) of the American Lobster (red), the Dungeness Crab (purple), and the Red Rock Crab (green). A) Unlike the endocuticle of the lobster, that of the Rock Crab contains crystalline calcite as indicated by sharp peaks at 281 cm-1 and 711 cm-1. B) Chitin crystallinity varies greatly between organisms depending on bonding and orientation of the polysaccharide chain. Differences in Amide I and III vibrational energies correspond to a variance in hydrogen bonding between and within chitin molecules. Large differences in wavenumber between the 1620 cm-1 and the 1659 cm-1 peaks correlate with a more crystalline polysaccharide (Iconomidou et al., 2001).

92

Figure B4. Shifts in Amide I and III peak positions relative to differences in CaCO3 vibrational energies. A) Amide I peak energies are more divergent with an increase in the ratio of ACC to calcite, as seen in the main body cavity. Consistent hydrogen bonding between polysaccharide chains implies higher degree of chitin crystallinity in claw exoskeletons. B) Intramolecular hydrogen bonds, represented by Amide III vibrational energies, indicate similar patterns of increased heterogeneity in samples with a more amorphous inorganic precipitate.

93

Figure B5. Flexural modulus is dependent on sample thickness. A) Unlike flexural rigidity, the flexural modulus is higher for samples with a higher proportion of calcite for all three animals. (B) Similar observations can be made for chitin crystallinity for exoskeletons from crabs. The lobster exoskeleton exhibits an opposite trend. (C) Measurements of initial failure strain are independent of exoskeleton thickness. (D) However, the flexural modulus exhibits much higher values for thin samples.

94

Figure B6. (A) The correlation of the initial failure strength to flexural modulus shows a slope of ≈0.01, which compares to the mechanical properties of similar natural composites, such as mollusk shells and corals (Wegst and Ashby). (B) Exoskeleton toughness as a function of the flexural modulus shows no correlation, but the values compare to those of other natural polymers and polymer composites, such as bone (Ashby et al, 1995).

95

APPENDIX C. MECHANICAL ANALYSIS OF CRUSTACEAN EXOSKELETONS

C1. DEFINITION OF TERMS AND CONDITIONS

For all samples, the radius of curvature was determined to be at least one order of

magnitude larger than the sample thickness, meaning all samples were assumed to be near-flat

beams. Despite minor surface topography (i.e. bumps and ridges), the cross section was

approximately rectangular along the length of each sample. Using this information, we can

define and calculate the following terms for each sample:

The second moment of area of the rectangular cross-section can be calculated using measured thickness and width values:

(C1)

The bending moment at the center of a flat beam is calculated for each point along the load vs. deflection curve:

(C2)

Stress can be calculated for each load:

(C3)

For this geometry, strain is defined as:

(C4)

96

The ratio of stress to strain yields the flexural modulus, Ef:

(C5)

Table C1. Definition of parameters used for the mechanical analyses. Term Definition Notes a Sample width (mm) Average of 6 measurements. b Sample thickness (mm) Average of 6 measurements. I The second moment of area (mm4) L Support span (mm) Fixed at 40 mm for all analyses. F Load applied to the center of the sample (N) Measured value. D Deflection at the center of the sample (mm) Measured value. 2 Mc The bending moment at the center of the sample (N·mm ) σi Stress in outer fibers at initial failure (MPa) σult Stress in outer fibers at ultimate failure (MPa) σmax Maximum stress in outer fibers when no failure occurred (MPa) εi Strain in outer surface at initial failure (%) εult Strain in outer surface at ultimate failure (%) εmax Maximum strain in outer surface when no failure occurred (%) Ef Flexural modulus of elasticity (GPa) Calculated as initial slope of stress-strain curve. Ef×I Flexural rigidity (kJ·m) -3 Ui Toughness up to initial failure (kJ·m ) Defined as the area under the stress-strain curve from the origin to the point of initial failure. -3 Uult Toughness up to ultimate failure (kJ·m ) Defined as the area under the stress-strain curve from the origin to the point of ultimate failure.

97

C2. DATA ANALYSIS ALGORITHM function(disp.data,sample.info){ setwd("~/figs3") require(pracma) x<-as.numeric(readline("Where would you like to start? ")) y<-as.numeric(readline("How many samples? ")) z<-x+y-1 results.i <- list() for (i in as.numeric(x:z)){ a<-2*i-1 b<-2*i delta<-disp.data[,a] P<-disp.data[,b] plot.id<-data.frame(delta,P) colnames(plot.id)<-c("delta","P") plot.id.diff<-plot.id[c(1:200),] change<-loess(P ~ delta, data = plot.id.diff) p.change<-predict(change, newdata = plot.id.diff$delta) plot.id.diff$diff<-c(0,diff(p.change)) p.min<-as.numeric(which.max(plot.id.diff$diff)) del.min<-as.numeric(which.max(plot.id.diff$diff)) p.min<-plot.id.diff[p.min,2] del.min<-plot.id.diff[del.min,1] plot.id2 <- subset(plot.id, delta > del.min) del.max<-max(plot.id2$delta, na.omit=TRUE) plot.id2$delta <- plot.id2$delta - plot.id2[1,1] plot.id2<-na.omit(plot.id2) plot.ext<-plot.id2[c(1:100),] ext.lm <- lm(delta ~ P, data = plot.ext) ext.rs<-residuals(ext.lm) ext.rs<-as.numeric(which.max(ext.rs<0)) ext.rs<-plot.ext[ext.rs,2] ext<-seq(0, ext.rs, length = 200) ext.pred<- predict(ext.lm,newdata = list(P = ext)) plot.id25<-subset(plot.id2, delta > max(ext.pred)) ext.head <- data.frame(ext.pred,ext) colnames(ext.head) <- c("delta","P") plot.id25 <- rbind(ext.head,plot.id25) plot.id25$delta <- plot.id25$delta - plot.id25[1,1] plot.id25<-na.omit(plot.id25) plot(plot.id,type="l", xlab= expression(delta ~ " (mm)"), ylab="P (N)", col="black", main = sample.info[i,1]) lines(plot.id25, col="red") u<-as.logical(readline("Does that look alright? (T/F) ")) if(u == FALSE){ print("Select first point in linear section.") plot(plot.id, type="l", xlab= expression(delta ~ " (mm)"), ylab="P (N)", col="black", main = sample.info[i,1]) range.lower <- locator(n=1, type='p') range.del <- range.lower[[1]]

98

range.del <- which(plot.id$delta > range.del)[1] range.del <- seq(from = range.del, by = 1, length.out = 30) plot.ext <- plot.id[range.del,] ext.lm <- lm(delta ~ P, data = plot.ext) ext.rs<-residuals(ext.lm) ext.rs<-as.numeric(which.max(ext.rs<0)) ext.rs<-plot.ext[ext.rs,2] ext<-seq(0, ext.rs, length = 200) ext.pred<- predict(ext.lm,newdata = list(P = ext)) plot.id25<-subset(plot.id, delta > max(ext.pred)) ext.head <- data.frame(ext.pred,ext) colnames(ext.head) <- c("delta","P") plot.id25 <- rbind(ext.head,plot.id25) plot.id25$delta <- plot.id25$delta - plot.id25[1,1] plot.id25 <- na.omit(plot.id25) plot(plot.id,type="l", xlab= expression(delta ~ " (mm)"), ylab="P (N)", col="black", main = sample.info[i,1]) lines(plot.id25, col="red") } plot.id2 <- na.omit(plot.id25) L <- as.numeric(sample.info[i,2]) Width <- as.numeric(sample.info[i,3]) thickness <- as.numeric(sample.info[i,4]) sigma <- plot.id2$P*((3*L)/(2*width*(thickness^2))) epsilon <- plot.id2$delta*((6*thickness)/(L^2)) I <- (width*(thickness^3))/12 plot.id3 <- data.frame(epsilon, sigma) sig<-plot.id3[plot.id3$sigma == max(plot.id3$sigma),] ult<-sig[1,] plot(plot.id3, type="l", xlab= expression(epsilon), ylab= expression(sigma ~ "(MPa)"), col="black", main = sample.info[i,1]) points(ult, col="red", pch=16, cex=2) u <- as.logical(readline("F for Fail, T for reTurns. ")) if(u == TRUE){ sig.max <- ult ult <- data.frame(c(NA),c(NA)) notes.i <- as.numeric(0) } else{ sig.max <- data.frame(c(NA),c(NA)) } u<-as.logical(readline("There is no sig.i. (T/F) ")) if(u == TRUE){ sig.i<-data.frame(c(NA), c(NA)) } else{ u<-as.logical(readline("Is sig.i defined by a change of slope? (T/F) ")) if(u == TRUE){ print("Okay, select low (1) and high (2) tangent points") 99

plot(plot.id3, type="l", xlab= expression(epsilon), ylab= expression(sigma ~ "(MPa)"), col="black", main = sample.info[i,1]) coords <- locator(n=2, type='p') tan.one <- max(which(plot.id3$epsilon <= coords$x[1])) tan.two <- max(which(plot.id3$epsilon <= coords$x[2])) tan.low <- plot.id3[seq(from = tan.one - 10, to = tan.one + 10, by = 1),] tan.high <- plot.id3[seq(from = tan.two - 10, to = tan.two + 10, by = 1),] lm.tan.low <- lm(sigma ~ epsilon, data = tan.low) lm.tan.high <- lm(sigma ~ epsilon, data = tan.high) m.tan <- summary(lm.tan.low)$coefficients[2,"Estimate"] a.tan <- summary(lm.tan.low)$coefficients[1,"Estimate"] n.tan <- summary(lm.tan.high)$coefficients[2,"Estimate"] b.tan <- summary(lm.tan.high)$coefficients[1,"Estimate"] tan.p <- (b.tan - a.tan)/(m.tan - n.tan) sig.i <- plot.id3[max(which(plot.id3$epsilon <= tan.p)),1:2] sig.i <- data.frame(epsilon = sig.i[,1], sigma = sig.i[,2]) notes.i<-as.numeric(5) } else{ u<-as.logical(readline("Is sig_ult equal to sig_i? (T/F) ")) if(u == TRUE){ sig.i<-ult notes.i<-as.numeric(1) } else{ m<-sd(sigma)/sqrt(length(sigma)) n<-c(abs(diff(sigma)),0) o<-data.frame(epsilon,sigma,n) o<-o[o$n>m,] o<-o[1,c(1:2)] plot(plot.id3, type="l", xlab= expression(epsilon), ylab= expression(sigma ~ "(MPa)"), col="black", main = sample.info[i,1]) points(o, col="red", pch=16, cex=2) u<-as.logical(readline("Is this sig_i? (T/F) ")) if(u == TRUE){ sig.i<-o[1,] notes.i<-as.numeric(2) } else{ plot.id4<-plot.id3[plot.id3$sigma >= sigma[1],] 100

id4.lm <- lm(sigma ~ epsilon, data = plot.id4) id4.rs <- residuals(id4.lm) q<-data.frame(epsilon,sigma) p<-as.numeric(match(max(id4.rs),id4.rs)) q<-q[p,] plot(plot.id3, type="l", xlab= expression(epsilon), ylab= expression(sigma ~ "(MPa)"), col="black", main = sample.info[i,1]) points(q, col="red", pch=16, cex=2) u<-as.logical(readline("Is this sig_i? (T/F) ")) if(u == TRUE){ sig.i<-q[1,] notes.i<-as.numeric(3) } else{ print("Okay, select point of first failure.") plot(plot.id3, type="l", xlab= expression(epsilon), ylab= expression(sigma ~ "(MPa)"), col="black", main = sample.info[i,1]) coords <- locator(n=1, type='p') sig.i.eps<-coords[[1]] sig.i.sig<-coords[[2]] sig.i<-data.frame(sig.i.eps,sig.i.sig) notes.i<-as.numeric(5) } } } } } if(is.na(sig.i[1,1])){E.point <- sig.max} else{E.point <- sig.i} plot.E<-plot.id3[plot.id3$epsilon < E.point[,1],] E.lm <- lm(sigma ~ 0 + epsilon, data = plot.E) E.pred <- predict(E.lm, newdata = list(epsilon=plot.E$epsilon)) E.pred <- data.frame(plot.E$epsilon,E.pred) E.f <- summary(E.lm)$coefficients[,1] E.I <- E.f * I plot(plot.id3, type="l", xlab= expression(epsilon), ylab= expression(sigma ~ "(MPa)"), col="black", main = sample.info[i,1]) lines(E.pred, col="red", lwd = 1.5) u<-as.logical(readline("Does E.f need adjustment? (T/F) ")) if(u == TRUE){ print("Okay, select highest point at which initial slope changes.") plot(plot.id3, type="l", xlab= expression(epsilon), ylab= expression(sigma ~ "(MPa)"), col="black", main = sample.info[i,1]) coords <- locator(n=1, type='p') E.p <- max(which(plot.id3$epsilon <= coords$x[1])) plot.E <- plot.id3[1 : E.p,] 101

E.lm <- lm(sigma ~ 0 + epsilon, data = plot.E) E.pred <- predict(E.lm, newdata = list(epsilon=plot.E$epsilon)) E.pred <- data.frame(plot.E$epsilon,E.pred) E.f <- summary(E.lm)$coefficients[,1] } E.f <- summary(E.lm)$coefficients[,1] E.f.err <- coef(summary(E.lm))[, "Std. Error"] E.f.rsquared <- summary(E.lm)$r.squared E.f.F <- summary(E.lm)$fstatistic E.f.p <- pf(E.f.F[1],E.f.F[2],E.f.F[3],lower.tail=F) attributes(E.f.p) <- NULL tough.xy <- plot.id3[plot.id3$epsilon <= sig.i[1,1],] tough <- trapz(tough.xy[,1], tough.xy[,2]) tough <- round(tough/0.01)*0.01 if(!is.na(ult[[1]])){ tough2.xy <- plot.id3[plot.id3$epsilon <= ult[1,1],] tough2 <- trapz(tough2.xy[,1], tough2.xy[,2]) tough2 <- round(tough2/0.01)*0.01 } else{tough2<-NA} nam <- sample.info[i,1] file.nam1<-paste(nam,".pdf", sep="") eps.limit <- ceiling(max(plot.id3$epsilon)/0.002)*0.002 sig.limit <- ceiling(max(plot.id3$sigma)/10)*10 pdf(file = file.nam1, width=5, height=5, useDingbats = FALSE) par(mar = c(3.5,3.5,2,2), mgp = c(2,0.5,0)) plot(1, type = 'n', xlab= expression(epsilon), ylab= expression(sigma ~ " (MPa)"), xlim=c(0, eps.limit), ylim = c(0, sig.limit), xaxs='i', yaxs='i', tck=0.02, cex.lab=1, cex.axis=1, cex.main=1, cex.sub=1) if(!is.na(tough2)){ polygon(c(tough2.xy[,1], ult[1,1]), c(tough2.xy[,2], 0), col = "#add8e6", border = NA) } polygon(c(tough.xy[,1], sig.i[1,1]), c(tough.xy[,2], 0), col = "#e3e3e3", border = NA) lines(plot.id3, col="black", lwd = 1.5) lines(E.pred, col="red", lwd = 1.5) axis(1,labels=F,tck=0.02, lwd=1.25, lwd.ticks=1.25) axis(2,labels=F,tck=0.02, lwd=1.25, lwd.ticks=1.25) axis(3,labels=F,tck=0.02, lwd=1.25, lwd.ticks=1.25) axis(4,labels=F,tck=0.02, lwd=1.25, lwd.ticks=1.25) if(!is.na(sig.max[1,1])){ points(sig.max, col="black", pch=16, cex=1.5, lwd = 2) text(sig.i[,1] - (0.1*eps.limit), sig.i[,2], labels = expression(sigma[max])) tough<-c(NA) } else{ if(!(is.na(ult[1,1])) && sig.i[1,1] == ult[1,1]){

102

points(ult, col="blue", bg="red", pch=21, cex=1.5, lwd = 2) text(ult[,1] - (0.1*eps.limit), ult[,2], labels = expression(sigma[ult] ~ "&" ~ sigma[i])) } else{ points(sig.i, col="red", pch=16, cex=1.5) points(ult, col="blue", pch=16, cex=1.5) text(sig.i[,1] + (0.05*eps.limit), sig.i[,2], labels = expression(sigma[i])) text(ult[,1] - (0.05*eps.limit), ult[,2], labels = expression(sigma[ult])) } } text(0.05*eps.limit, 0.8*sig.limit, labels = bquote(E[f] %.% I ==.(round(E.I/10)*0.01) ~ "kJ" %.% m), cex=0.75, adj = c(0,1)) text(0.05*eps.limit, 0.75*sig.limit, labels = bquote(E[f]==.(round(E.f/10)*0.01) %+-% .(round(E.f.err)*0.001) ~ "GPa"), cex=0.75, adj = c(0,1)) if(!(is.na(tough))){ text(0.7*sig.i[,1], 0.5*sig.i[,2], labels = bquote(U[T]==.(tough) ~ "kJ" %.% m^-3), cex=0.75) } if(!(is.na(tough2))){text(0.8* max(tough2.xy[,1]), 0.7* max(tough2.xy[,2]), labels = bquote(U[T]==.(tough2) ~ "kJ" %.% m^- 3), cex=0.75)} legend(0.05*eps.limit, 0.95*sig.limit, legend=c("data",expression("fit for" ~ E[f])), col=c("black", "red"), lwd=c(2,2), bg='#e3e3e3', bty="o", box.col="#e3e3e3", cex=0.75, x.intersp=0.5) dev.off() eps.ult<-round(ult[[1]]/0.00001)*0.001 sig.ult<-round(ult[[2]]/0.001)*0.001 eps.i<-round(sig.i[[1]]/0.00001)*0.001 sig.i<-round(sig.i[[2]]/0.001)*0.001 eps.max<-round(sig.max[[1]]/0.00001)*0.001 sig.max<-round(sig.max[[2]]/0.001)*0.001 E.I <- round(E.I/10)*0.01 E.f <- round(E.f/10)*0.01 E.f.err<-round(E.f.err*1.96)*0.001 E.f.rsquared<-round(E.f.rsquared/0.0001)*0.0001 E.f.p<-round(log10(E.f.p)/0.1)*0.1 all.data <- numeric(15) all.data <- c(nam,eps.ult,sig.ult,eps.i,sig.i,eps.max,sig.max,E.I,E.f,E.f.err,E.f. rsquared,E.f.p,tough,tough2,notes.i) results.i[[i]] <- all.data } results <- do.call("rbind",results.i) results <- data.frame(results) colnames(results)<- c("sample",'eps.ult','sig.ult','eps.i','sig.i','eps.max','sig.max','Mo 103

d.I','app.Mod','Mod.err','R2','p- value','toughness.i','toughness.ult','notes') results }

104

C3. EXAMPLES OF MECHANICAL ANALYSES.

Figure C1. Four examples of mechanical behavior of the crustacean exoskeleton. (A) Common brittle fracture with some slight departure from linearity in the initial slope; initial and ultimate stresses coincide. (B) Exoskeleton samples for the lobster cephalothorax often exhibit changes in initial slope that extend beyond 0.2% strain, so initial failure was defined as the point where the initial slope intersects with the tangent to the curve. (C) Samples from crab claw exoskeleton showed brittle behavior until initial point of failure and a separate point of ultimate failure. (D) Samples without points of ultimate failure were only assigned a maximum stress value.

105

C4. SUMMARY OF MECHANICAL DATA

Table C2. Summary of all mechanical analyses performed for this study. See Appendix C1 for explanation of terms.

Organism Body thickness width εi σi εult σult εmax σmax Ef x I Ef Ui Uult CrI130 Calcite part (mm) (mm) (%) (MPa) (%) (MPa) (%) (MPa) (kJ·m) (GPa) (kJ·m-3) (kJ·m-3) (%) (%) Am. lobster chelae 1.81 13.92 ------3.96 49.5 9.42 1.6 -- -- 67.5 2.1 H. americanus 1.18 12.48 2.55 86.0 2.55 86.0 -- -- 5.97 3.54 1.15 1.15 67.5 2.1 1.11 11.77 1.65 48.3 -- -- 1.95 69.5 4.15 3.09 -- -- 67.5 2.1 1.28 8.83 ------2.33 67.4 5.14 4.19 -- -- 67.5 2.1 1.05 12.16 2.44 66.3 2.44 66.3 -- -- 3.23 2.75 0.83 0.83 67.5 2.1 1.59 16.17 ------3.06 44.5 8.72 1.76 -- -- 67.5 2.1 1.67 16.40 1.48 48.7 1.48 48.7 -- -- 20.31 3.18 0.35 0.35 67.5 2.1 1.44 15.28 1.91 39.3 -- -- 3.11 51.4 8.33 2.2 -- -- 67.5 2.1 2.16 14.14 1.78 27.4 -- -- 5.74 47.1 19.19 1.63 -- -- 67.5 2.1 1.82 14.76 2.08 118 2.08 118 -- -- 41.78 5.65 1.2 1.2 67.5 2.1 1.84 14.73 1.25 47.9 1.44 50.9 -- -- 29.8 3.88 0.3 0.4 67.5 2.1 3.01 15.96 2.07 44.1 7.28 65.3 -- -- 82.33 2.28 0.49 3.51 67.5 2.1 1.72 13.38 0.99 27.0 -- -- 3.59 59.6 16.16 2.85 -- -- 67.5 2.1 1.07 11.32 1.73 51.4 -- -- 2.68 57.6 3.66 3.14 -- -- 67.5 2.1 1.35 10.77 2.59 110 -- -- 3.12 117 8.89 4.07 -- -- 67.5 2.1 1.39 12.20 1.37 87.9 1.65 89.9 -- -- 17.6 6.49 0.61 0.85 67.5 2.1 1.14 9.72 1.22 42.8 -- -- 1.88 56.0 4.08 3.42 -- -- 67.5 2.1 1.07 9.75 ------1.73 49.9 3.06 3.09 -- -- 67.5 2.1 1.32 11.62 1.77 117 2.18 129 -- -- 15.33 6.97 1.1 1.61 67.5 2.1 1.15 13.39 1.41 132 1.81 148 -- -- 16.4 9.62 0.95 1.52 67.5 2.1 1.11 10.85 ------3.01 81.7 3.37 2.76 -- -- 67.5 2.1 Cthx. 0.65 16.86 0.55 159 0.97 203 -- -- 11.65 30.62 0.47 1.27 66.5 13.2 0.66 13.71 0.99 111 -- -- 1.55 137 3.83 11.92 -- -- 66.5 13.2 0.69 15.91 0.84 114 -- -- 1.52 149 6.41 14.6 -- -- 66.5 13.2 0.71 16.86 ------1.23 175 8.23 20.07 -- -- 66.5 13.2 0.79 12.42 ------0.71 58.1 4.16 8.22 -- -- 66.5 13.2 0.69 15.85 1.13 136 1.13 136 -- -- 6.11 16.86 0.95 0.95 66.5 13.2 0.73 14.69 0.94 113 -- -- 1.47 132 6.15 12.75 -- -- 66.5 13.2 0.68 13.01 ------1.12 125 4.26 12.38 -- -- 66.5 13.2 0.63 11.37 0.91 122 0.91 122 -- -- 3.36 14.52 0.62 0.62 66.5 13.2 0.65 13.57 0.66 133 1.08 160 -- -- 6.42 21.17 0.46 1.1 66.5 13.2 0.63 11.53 ------1.40 121 2.25 9.47 -- -- 66.5 13.2 0.66 11.94 1.15 103 1.15 102 -- -- 2.49 8.61 0.55 0.55 66.5 13.2 0.62 11.07 ------1.28 159 2.86 13.16 -- -- 66.5 13.2 0.72 10.80 0.82 90.3 0.82 90.3 -- -- 3.68 11.06 0.38 0.38 66.5 13.2 0.69 8.92 ------1.44 149 2.44 10.23 -- -- 66.5 13.2 0.64 9.70 ------1.77 131 1.57 7.61 -- -- 66.5 13.2 0.67 9.39 ------0.94 104 2.63 11.03 -- -- 66.5 13.2 0.71 11.58 ------1.39 79.3 1.86 5.45 -- -- 66.5 13.2 legs 0.68 11.95 1.08 86.0 1.28 91.9 -- -- 2.6 8.23 0.47 0.65 -- -- D. crab chelae 0.86 9.87 ------1.66 65.5 2.19 4.69 -- -- 63.9 52.15 C. magister 0.75 10.18 0.90 50.2 1.16 63.8 -- -- 1.89 5.31 0.21 0.36 63.9 52.15 0.87 10.15 1.15 94.8 1.15 94.8 -- -- 4.78 8.59 0.58 0.58 63.9 52.15 0.80 10.03 0.32 50.2 0.76 116 -- -- 7.02 16.41 0.09 0.38 63.9 52.15 0.96 12.16 1.99 71.7 1.97 71.9 -- -- 3.94 6.16 0.91 0.89 63.9 52.15 0.90 11.94 1.14 83.4 1.34 90.7 -- -- 5.65 7.87 0.52 0.7 63.9 52.15 0.92 9.97 ------1.28 83.9 4.5 6.91 -- -- 63.9 52.15 0.87 10.37 1.34 130 1.34 130 -- -- 5.8 10.37 0.94 0.94 63.9 52.15 Cthx. 0.98 11.41 2.30 56.2 2.30 56.2 -- -- 2.43 2.76 0.75 0.75 59.1 6.3 1.09 11.79 2.36 71.6 2.32 71.9 -- -- 4.53 4.02 1.01 0.98 59.1 6.3 0.87 12.21 0.63 21.9 1.73 55.2 -- -- 2.28 3.47 0.07 0.49 59.1 6.3 1.03 9.32 ------2.06 71.9 3.19 4.1 -- -- 59.1 6.3 1.26 12.47 1.40 48.6 2.96 50.1 -- -- 7.46 3.61 0.36 1.08 59.1 6.3 1.17 11.43 2.04 70.5 2.04 70.5 -- -- 5.63 3.69 0.78 0.78 59.1 6.3 1.01 11.00 1.58 69.9 1.58 69.9 -- -- 4.51 4.82 0.62 0.62 59.1 6.3 1.05 12.07 1.49 33.7 2.38 38.8 -- -- 2.83 2.44 0.28 0.58 59.1 6.3 1.06 12.39 1.90 75.5 1.90 75.5 -- -- 5.21 4.6 0.78 0.78 59.1 6.3 0.92 11.58 ------2.28 52.0 1.85 2.43 -- -- 59.1 6.3 1.08 10.86 1.64 75.4 1.64 75.4 -- -- 5.45 4.81 0.65 0.65 59.1 6.3 1.06 12.18 1.73 79.9 2.16 84.4 -- -- 5.7 4.76 0.71 1.06 59.1 6.3

106

Organism Body thickness width εi σi εult σult εmax σmax Ef x I Ef Ui Uult CrI130 Calcite part (mm) (mm) (%) (MPa) (%) (MPa) (%) (MPa) (kJ·m) (GPa) (kJ·m-3) (kJ·m-3) (%) (%) 0.76 12.45 0.99 61.5 0.99 61.5 -- -- 2.95 6.55 0.32 0.32 59.1 6.3 0.87 9.63 ------1.88 79.9 2.47 4.68 -- -- 59.1 6.3 1.05 12.64 0.92 54.1 1.91 90.9 -- -- 7.15 5.9 0.25 0.97 59.1 6.3 0.94 11.64 ------1.93 92.3 4.23 6.7 -- -- 59.1 6.3 0.74 10.95 1.73 71.8 1.73 71.8 -- -- 1.59 4.28 0.64 0.64 59.1 6.3 0.92 10.07 ------2.40 84.1 2.51 4.3 -- -- 59.1 6.3 legs 1.07 11.47 1.49 117 1.49 117 -- -- 9.82 8.34 0.93 0.93 -- -- 0.84 7.25 1.76 107 1.76 106 -- -- 2.16 6.05 0.94 0.94 -- -- 0.88 13.58 ------2.63 100 3.62 6 ------0.91 11.57 2.55 107 2.53 106 -- -- 3.49 5.55 1.6 1.58 -- -- Red rock crab chelae 1.39 9.79 0.73 104 1.05 125 -- -- 32.25 14.81 0.4 0.77 69.5 100 C. pagurus 1.08 7.98 0.21 28.3 0.82 107 -- -- 11.26 13.56 0.03 0.44 69.5 100 1.44 8.22 0.45 80.1 0.45 80.1 -- -- 36.18 17.87 0.18 0.18 69.5 100 Cthx. 1.02 11.58 0.89 59.6 1.12 65.7 -- -- 6.88 6.71 0.27 0.41 67.2 74.6 1.06 13.31 0.94 24.1 1.38 38.1 -- -- 3.33 2.5 0.11 0.24 67.2 74.6 1.20 12.29 1.10 35.1 1.10 35.1 -- -- 5.68 3.25 0.2 0.2 67.2 74.6 1.12 9.70 0.83 71.4 1.24 96.6 -- -- 10.04 8.79 0.31 0.65 67.2 74.6 0.90 10.41 1.28 97.3 1.63 113 -- -- 5 7.99 0.66 1.04 67.2 74.6 0.88 10.33 0.55 28.3 1.38 70.3 -- -- 3.07 5.18 0.08 0.49 67.2 74.6 0.94 10.73 1.69 125 1.69 125 -- -- 5.7 7.68 1.1 1.1 67.2 74.6 0.97 11.26 1.52 89.2 1.52 89.2 -- -- 5.32 6.21 0.72 0.72 67.2 74.6 1.11 11.78 0.97 45.5 1.29 53.7 -- -- 6.49 4.8 0.22 0.39 67.2 74.6 1.05 11.67 1.96 99.2 1.96 99.2 -- -- 6.06 5.62 1.06 1.06 67.2 74.6 1.15 12.65 2.40 103 2.40 103 -- -- 7.52 5.11 1.38 1.38 67.2 74.6 1.15 13.04 0.78 66.9 0.78 66.9 -- -- 14.73 8.87 0.27 0.27 67.2 74.6 0.95 15.45 1.36 76.7 1.36 76.7 -- -- 6.44 5.84 0.55 0.55 67.2 74.6 0.96 12.10 1.23 72.8 1.23 72.8 -- -- 5.29 5.88 0.44 0.44 67.2 74.6 0.96 9.58 2.06 126 2.06 126 -- -- 4.48 6.3 1.36 1.36 67.2 74.6 0.92 14.06 1.31 94.0 1.31 94 -- -- 6.71 7.28 0.63 0.63 67.2 74.6 1.00 12.30 2.16 109 2.16 110 -- -- 5.52 5.42 1.28 1.28 67.2 74.6 0.92 9.84 2.23 92.5 2.23 92.5 -- -- 2.7 4.3 1.08 1.08 67.2 74.6 legs 0.45 9.82 ------0.60 123 1.47 19.48 ------0.41 10.02 0.44 96.7 0.43 98.8 -- -- 1.47 25.09 0.25 0.24 -- -- 0.58 9.80 0.79 113 0.79 113 -- -- 2.24 13.92 0.44 0.44 -- -- 0.86 11.37 1.39 73.3 1.39 73.3 -- -- 3.31 5.49 0.53 0.53 -- -- 0.63 8.83 0.60 81.0 0.60 80.9 -- -- 2.44 13.4 0.24 0.24 -- -- 0.74 9.27 0.80 75.8 0.81 75.8 -- -- 3 9.49 0.31 0.31 -- -- 0.61 8.63 ------1.47 234 2.29 12.93 ------

107

APPENDIX D. SUPPLEMENTARY INFORMATION FOR CHAPTER 4

D1. SUPPLEMENTARY FIGURES

Figure D1. (A-B) Characterization of solution and ACC composition reveals Mg sensitivity to chemical environment. (A) Solubility of ACC as a function of time is mostly consistent. Neither high nor low carbonate conditions exhibit a significant evolution. (B) Apparent solubility as a function of pH is a more significant correlation, yet not very consistent.

108

Figure D2. (A and B) Expanded vertical scale PDF profiles show the short-range order of all conditions aligns over time. (A) At 4 minutes, conditions B, C, and E produce ACC with more consistent Ca-O coordination. (B) After 8 minutes, this type has largely disappeared and more developed Mg-O pairs appear.

109

Figure D3. Short-range order of solutions is distinctly different from in situ ACC. The highly consistent Mg/Ca-O distance at 3.2 A of the ACC in blue is very similar in length and coordination to the Mg/Ca-O(-H2) distance in a Mg/Ca chloride solution.

110

Figure D4. Comparison of ACC types to other Ca and Mg carbonates reveals distinct short- range order of amorphous phases that feature composite atom pair distances.

111

Figure D5. Estimated dependence of apparent solubility on magnesium content for the two end- member compositions. (A) Hydrated, Ca-rich ACC and (B) anhydrous, Mg-rich ACC both exhibit narrow solubility ranges, but are separated by a significant difference in solubility of approximately half an order of magnitude.

112

D2. DATA SUMMARY

Table D1. TGA and ICP-OES analyses give overall compositions of ACC produced after 8- minute residence times.

Sample mol% MgCO3 composition Anhydrous phase

A 20.9 Ca0.79Mg0.21CO3·0.59H2O <5%*

B 15 Ca0.85Mg0.15CO3·0.60H2O <5%* C 10.4 Ca0.90Mg0.10CO3·0.63H2O <5%* D 13.9 Ca0.86Mg0.14CO3·0.42H2O <1%

E 38.3 Ca0.62Mg0.38CO3 51% H 27.7 Ca0.72Mg0.28CO3·0.51H2O 15% I 26.5 Ca0.73Mg0.27CO3·0.37H2O 28%

113

Table D2. Transformation experiments reveal ACC from the tested conditions produce a large 2 number of polymorphs. Statistics from the Rietveld Refinement are given by Rwp, Rexp, and χ . The presence of ACC was confirmed using scanning electron microscopy (SEM).

2 pHi pH0 Time pHf mol% MgCO3 composition Rwp Rexp χ ACC (min) A 9.502 8.18 1181 8.115 5.7 Monohydrocalcite 14.56% 9.95% 2.141 yes 2602 8.117 4.4 Monohydrocalcite 10.48% 6.31% 2.758 no B 9.507 8.41 1150 8.435 14.2 Monohydrocalcite 12.47% 9.69% 1.656 yes 2586 8.482 12.0 Monohydrocalcite 15.11% 6.43% 5.522 no C 9.502 8.55 1122 8.597 2.5 Monohydrocalcite (74.9%), 13.63% 11.78% 1.339 yes Aragonite (25.1%) 2567 8.683 4.5 Monohydrocalcite (61.2%), 13.16% 6.86% 3.680 yes Aragonite (38.8%) D 9.533 8.212 1010 7.027 1.3 Aragonite 9.2% 6.34% 2.106 no E 9.505 8.507 991 8.02 71.0* Nesquehonite (77.6%), 8.52% 6.09% 1.957 yes Monohydrocalcite (22.4%) 2453 8.063 71.2* Nesquehonite (81.8%), 8.43% 6.41% 1.730 yes Monohydrocalcite (18.2%) H 9.49 8.702 1233 8.4 9.2 Aragonite (93.1%), 7.27% 6.78% 1.150 yes Monohydrocalcite (6.9%) I 9.578 8.761 1251 8.619 34.1 Monohydrocalcite (66%), 12.85% 8.97% 2.052 yes Aragonite (34%) 1530 8.627 8.1 Monohydrocalcite (62.4%), 15.45% 9.16% 2.845 yes Aragonite (37.6%) J 9.559 8.871 1228 8.418 12.5 N/A N/A N/A N/A N/A 1509 8.474 4.2 Aragonite (98.8%), 10.93% 8.1% 1.821 no Calcite (1.2%) K 9.542 8.929 1120 8.808 39.8 Aragonite (62%), 10.95% 12.56% 0.760 yes Monohydrocalcite (38%) 1398 8.803 13.5 Aragonite (76.2%), 11.45% 8.77% 1.705 yes Monohydrocalcite (23.8%)

* The dominant phase is Nesquehonite (MgHCO3·3H2O), a magnesium carbonate.

114

Table D3. Chemical and calorimetric data from Radha et al., 2012. All data was collected at 26°C. Log K’sp was calculated using the thermodynamic approximation explained in Appendix D3. Solubility values of the reference standards for ACC: Brečević and Nielsen (1989), for calcite: Benjamin et al. (2002), and for magnesite: Plummer and Busenberg (1982).

mol% MgCO3 Water content ΔHsoln ΔHcryst Calculated log K’sp Notes 100 1.28 -72.08 -35.65 -4.73 Mg-ACC 87.5 1.18 -65.71 -30.26 -5.14 Two-phase ACC 86 1.68 -58.06 -22.72 -5.64 Two-phase ACC 78.1 1.07 -54.59 -19.86 -5.87 Two-phase ACC 68.8 1.08 -54.72 -20.71 -5.86 Two-phase ACC 59.7 1.05 -50.66 -17.36 -6.12 Two-phase ACC 58 1.5 -47.18 -14.01 -6.35 Two-phase ACC 53 1.31 -48.15 -15.36 -6.29 Two-phase ACC 53.7 1.49 -47.42 -14.57 -6.33 Two-phase ACC 51.1 1.8 -41.73 -9.08 -6.70 Two-phase ACC 50.6 1.76 -41.64 -9.03 -6.71 Two-phase ACC 47.2 1.75 -42.18 -9.83 -6.67 Two-phase ACC 41.6 1.3 -48.81 -16.90 -6.24 One-phase ACC 32.8 1.36 -47.31 -16.07 -6.34 One-phase ACC 25.4 1.26 -45.84 -15.18 -6.44 One-phase ACC 23.1 1.14 -45.25 -14.77 -6.47 One-phase ACC 17.8 1.39 -40.71 -10.63 -6.77 One-phase ACC 14 1.17 -46.30 -16.48 -6.41 One-phase ACC 8.4 1.141 -42.31 -12.99 -6.66 One-phase ACC 5 0.25 -42.50 -13.31 -6.65 One-phase ACC 3.5 1.09 -40.36 -11.39 -6.79 One-phase ACC 1.9 0.98 -41.18 -12.33 -6.74 One-phase ACC 0 1.36 -46.38 -17.09 -6.4 ACC reference 0 0 -28.70 -8.5 Calcite reference

100 0 -36.40 -8.0 Magnesite reference

115

D3. ESTIMATION OF SOLUBILITY VALUES FROM ENTHALPY VALUES

In general, the solubility product of a given phase (Ksp) can be approximated by the following

equation:

(D1)

-1 -1 where ΔGsol is the free energy of solution, R is the gas constant (0.008314 kJmol K ), and T is

the temperature (299.15°K).

Taking the logarithm base 10 of both sides of equation D1 yields:

(D2)

Note that ΔGsol can be calculated from the enthalpy of solution ΔHsol, the temperature T,

and the change in entropy ΔS:

(D3) Substituting Eq. D3 into D2 yields the following expression for estimating the solubility constant

Ksp:

(D4)

The entropy expression in Eq. D4 can be approximated for ACC using the measurements

of Ksp and ΔHsol for the ACC reference standard from table D3. Assuming the entropy term ΔS is similar for both types of ACC, solubility constants can then be calculated for all values of

ΔHsol. Note that the error in this assumption may be quite large and resulting Ksp values are only to be used as approximate points of reference. However, the slope of the dependence of solubility on Mg content is likely conserved by this approximation, as enthalpy and temperature

have been measured.

116

D4. SUMMARY OF MEASURED AND CALCULATED VALUES FOR ALL EXPERIMENTS.

Table D4. Summary of measured and calculated chemical parameters of ACC and solutions for all experiments.

(Table starts on next page.)

117

experimental set-up calculated from measured values (Geochemist’s Workbench) measured directly

log log log log log log I aCO32-/ condition H2O* time (s) σACC σcalcite pH i Mg/Casolution pH f mol% MgCO3 morphology aCa2+ aMg2+ aNa+ aCO32- aHCO3- K’sp (molal) aCa2+ A 30 2.80 7.84 9.52 -3.015 -1.845 -1.362 -3.487 -1.347 -6.16 0.307 0.337 0.337 8.20 29.3 large

Initial concentration at x 60 2.81 7.82 9.52 -3.034 -1.876 -1.396 -3.466 -1.391 -6.17 0.294 0.370 0.370 8.27 28.4 >70% large mixing (200 mL; μM) 60 2.83 7.81 9.51 -3.009 -1.845 -1.377 -3.502 -1.364 -6.18 0.305 0.322 0.322 8.21 28.1 >70% large [MgCl2]: 125 x 120 2.82 7.84 9.51 -3.029 -1.860 -1.388 -3.480 -1.390 -6.17 0.298 0.354 0.354 8.25 29.2 large [CaCl2]: 25 120 2.81 7.83 9.53 -3.112 -1.896 -1.355 -3.407 -1.327 -6.17 0.299 0.508 0.508 8.26 28.8 large [NaHCO3]: 100 x 240 2.82 7.79 9.53 -3.006 -1.851 -1.398 -3.492 -1.410 -6.19 0.298 0.326 0.326 8.26 26.8 large [NaOH]: 3.1 240 3.04 7.89 9.52 -3.311 -2.048 -1.460 -3.334 -1.275 -6.24 0.271 0.947 0.947 8.28 31.8 >70% large x 480 2.85 7.76 9.52 -3.010 -1.851 -1.398 -3.496 -1.411 -6.21 0.298 0.327 0.327 8.26 25.4 --

B 30 2.86 7.97 9.48 -3.322 -2.287 -1.396 -3.142 -1.317 -6.13 0.169 1.515 1.515 8.52 32.7 Large x 60 2.87 7.97 9.48 -3.337 -2.333 -1.446 -3.118 -1.343 -6.13 0.161 1.657 1.657 8.57 32.2 >70% large

[MgCl2]: 50 60 2.76 7.84 9.50 -2.939 -2.119 -1.427 -3.372 -1.355 -6.14 0.181 0.370 0.370 8.33 20.9 >70% large

[CaCl2]: 25 x 120 2.94 7.85 9.50 -3.347 -2.331 -1.409 -3.093 -1.348 -6.21 0.163 1.793 1.793 8.60 22.3 >70% large

[NaHCO3]: 100 120 2.92 7.86 9.50 -3.341 -2.287 -1.370 -3.100 -1.336 -6.20 0.170 1.743 1.743 8.58 22.9 >70% large [NaOH]: 3.0 x 240 2.84 7.91 9.50 -3.364 -2.311 -1.398 -3.068 -1.364 -6.14 0.164 1.977 1.977 8.64 27.4 >70% large 240 2.83 7.87 9.51 -3.364 -2.313 -1.386 -3.041 -1.283 -6.16 0.172 2.099 2.099 8.59 23.4 >70% large

x 480 2.51 8.17 9.51 -3.343 -2.305 -1.401 -3.071 -1.366 -5.89 0.164 1.870 1.870 8.64 50.7 >70% small

C 30 3.25 7.76 9.48 -3.353 -2.597 -1.413 -3.133 -1.292 -6.39 0.128 1.660 1.660 8.50 12.8 large x 60 3.16 7.77 9.48 -3.305 -2.562 -1.373 -3.141 -1.329 -6.34 0.129 1.459 1.459 8.53 13.8 large

[MgCl2]: 25 60 3.38 7.77 9.52 -3.532 -2.686 -1.442 -3.031 -1.290 -6.44 0.124 3.167 3.167 8.60 14.4 large

[CaCl2]: 25 x 120 3.21 7.77 9.52 -3.460 -2.642 -1.406 -3.027 -1.325 -6.37 0.125 2.709 2.709 8.64 14.6 >70% large

[NaHCO3]: 100 120 3.13 7.78 9.51 -3.320 -2.550 -1.355 -3.130 -1.296 -6.33 0.134 1.547 1.547 8.51 15.7 >70% large [NaOH]: 3.0 x 240 3.10 7.78 9.51 -3.320 -2.572 -1.374 -3.113 -1.332 -6.32 0.129 1.610 1.610 8.56 15.4 >70% large 240 2.98 7.82 9.53 -3.367 -2.541 -1.330 -3.081 -1.290 -6.25 0.136 1.928 1.928 8.55 24.4 >70% large

x 480 3.10 7.78 9.53 -3.360 -2.578 -1.367 -3.084 -1.337 -6.31 0.129 1.888 1.888 8.60 16.4 --

D 30 2.13 7.03 9.50 -2.633 -1.680 -1.576 -3.770 -1.655 -6.22 0.325 0.073 0.073 8.23 19.2 -- x 60 2.08 7.55 9.50 -2.656 -1.712 -1.604 -3.752 -1.681 -5.97 0.312 0.080 0.080 8.27 46.4 --

[MgCl2]: 125 60 2.15 7.08 9.52 -2.640 -1.684 -1.585 -3.775 -1.665 -6.20 0.322 0.073 0.073 8.23 22.2 --

[CaCl2]: 25 x 120 2.14 7.08 9.52 -2.662 -1.717 -1.616 -3.750 -1.686 -6.20 0.309 0.082 0.082 8.28 21.9 --

[NaHCO3]: 50 120 2.01 7.98 9.52 -2.646 -1.695 -1.589 -3.759 -1.651 -5.75 0.319 0.077 0.077 8.24 68.5 --

118

[NaOH]: 1.2 x 240 2.16 7.02 9.52 -2.658 -1.721 -1.627 -3.757 -1.698 -6.24 0.307 0.080 0.080 8.28 19.1 -- 240 2.12 7.07 9.51 -2.627 -1.682 -1.581 -3.775 -1.644 -6.20 0.325 0.071 0.071 8.21 21.5 --

x 480 2.13 7.07 9.51 -2.642 -1.699 -1.602 -3.762 -1.675 -6.20 0.316 0.076 0.076 8.26 21.3 --

E 30 3.21 8.58 9.50 -3.845 -2.171 -1.226 -2.921 -1.023 -6.01 0.313 8.392 8.392 8.45 45.0 large x 60 3.20 8.58 9.50 -3.851 -2.185 -1.246 -2.907 -1.061 -6.01 0.303 8.794 8.794 8.50 44.9 large

[MgCl2]: 125 60 3.28 8.63 9.51 -3.879 -2.170 -1.219 -2.955 -1.108 -6.02 0.299 8.409 8.409 8.50 47.4 >70% large

[CaCl2]: 25 x 120 3.18 8.62 9.51 -3.907 -2.200 -1.235 -2.876 -1.072 -5.99 0.302 10.726 10.726 8.54 46.7 >70% large

[NaHCO3]: 200 120 3.17 8.61 9.50 -3.842 -2.175 -1.220 -2.917 -1.020 -5.99 0.314 8.403 8.403 8.45 46.3 >70% large [NaOH]: 6.9 x 240 3.17 8.60 9.50 -3.836 -2.182 -1.243 -2.914 -1.058 -5.99 0.304 8.356 8.356 8.49 45.8 >70% large 240 3.12 8.69 9.50 -3.836 -2.176 -1.228 -2.928 -1.019 -5.92 0.313 8.082 8.082 8.43 50.6 >70% large

x 480 3.14 8.63 9.50 -3.796 -2.173 -1.230 -2.939 -1.064 -5.96 0.305 7.188 7.188 8.47 47.6 ≈1:1

H 30 2.58 7.96 9.52 -3.228 -2.156 -1.620 -3.203 -1.628 -6.01 0.197 1.057 1.057 8.77 39.4 >70% large x 60 2.62 7.82 9.52 -3.223 -2.174 -1.622 -3.202 -1.668 -6.09 0.194 1.050 1.050 8.81 32.3 large

[MgCl2]: 100 60 2.53 7.71 9.51 -3.218 -2.157 -1.608 -3.164 -1.561 -6.10 0.201 1.130 1.130 8.74 26.8 >70% large

[CaCl2]: 20 x 120 2.52 8.02 9.51 -3.218 -2.184 -1.629 -3.178 -1.629 -5.96 0.195 1.096 1.096 8.80 42.4 >70% large

[NaHCO3]: 100 120 2.54 8.10 9.50 -3.192 -2.149 -1.624 -3.223 -1.592 -5.93 0.199 0.930 0.930 8.71 46.5 ≈1:1 [NaOH]: 1.3 x 240 2.53 7.96 9.50 -3.194 -2.169 -1.625 -3.199 -1.631 -5.99 0.196 0.989 0.989 8.78 39.5 ≈1:1 240 2.63 8.29 9.53 -3.313 -2.259 -1.703 -3.169 -1.549 -5.89 0.189 1.393 1.393 8.72 56.4 >70% small

x 480 2.41 8.21 9.53 -3.216 -2.181 -1.624 -3.151 -1.582 -5.83 0.197 1.163 1.163 8.78 51.9 small

I 30 2.87 7.80 9.49 -3.272 -2.205 -1.615 -3.190 -1.627 -6.21 0.180 1.206 1.206 8.78 23.7 >70% small x 60 2.80 8.29 9.49 -3.262 -2.231 -1.633 -3.198 -1.687 -5.97 0.175 1.159 1.159 8.83 47.9 ≈1:1

[MgCl2]: 90 60 2.76 8.32 9.52 -3.276 -2.206 -1.612 -3.186 -1.623 -5.93 0.180 1.230 1.230 8.78 49.4 >70% small

[CaCl2]: 18 x 120 2.51 8.49 9.52 -3.314 -2.246 -1.634 -3.057 -1.557 -5.75 0.181 1.808 1.808 8.84 58.0 ≈1:1

[NaHCO3]: 144 120 2.60 8.34 9.49 -3.277 -2.215 -1.606 -3.115 -1.525 -5.85 0.185 1.455 1.455 8.75 50.8 >70% small [NaOH]: 1.3 x 240 2.72 8.37 9.49 -3.288 -2.246 -1.639 -3.149 -1.597 -5.90 0.177 1.379 1.379 8.79 52.0 ≈1:1 x 480 2.87 8.21 9.48 -3.261 -2.237 -1.649 -3.219 -1.663 -6.03 0.174 1.102 1.102 8.79 44.0 ≈1:1

J 30 1.92 7.53 9.55 -3.199 -2.306 -1.829 -3.217 -1.664 -5.91 0.114 0.959 0.959 8.79 56.6 small x 60 1.83 7.51 9.55 -3.256 -2.354 -1.851 -3.125 -1.674 -5.88 0.110 1.351 1.351 8.89 55.8 small

[MgCl2]: 50 60 1.73 7.47 9.57 -3.216 -2.290 -1.809 -3.135 -1.628 -5.86 0.117 1.206 1.206 8.84 53.5 small

[CaCl2]: 10 x 120 2.02 7.45 9.57 -3.162 -2.277 -1.838 -3.292 -1.920 -5.99 0.108 0.742 0.742 8.97 52.7 small

[NaHCO3]: 50 120 2.85 7.35 9.50 -3.154 -2.265 -1.852 -3.660 -1.838 -6.39 0.108 0.312 0.312 8.52 47.5 -- [NaOH]: 0.5 x 240 2.33 7.29 9.50 -3.198 -2.285 -1.857 -3.402 -1.925 -6.20 0.105 0.625 0.625 8.87 44.2 small

119

x 480 1.73 7.11 9.51 -3.261 -2.339 -1.840 -3.073 -1.548 -6.01 0.117 1.544 1.544 8.82 35.4 --

K 30 1.65 7.87 9.51 -3.428 -2.504 -1.704 -2.824 -1.494 -5.65 0.113 4.021 4.021 9.01 65.5 small x 60 1.99 7.94 9.51 -3.352 -2.458 -1.721 -3.025 -1.725 -5.76 0.102 2.119 2.119 9.04 68.8 small

[MgCl2]: 45 60 1.77 7.89 9.52 -3.384 -2.465 -1.704 -2.917 -1.597 -5.69 0.108 2.930 2.930 9.02 66.5 small

[CaCl2]: 9 x 120 1.84 7.89 9.52 -3.409 -2.493 -1.728 -2.917 -1.627 -5.72 0.105 3.103 3.103 9.05 66.6 small

[NaHCO3]: 72 120 1.76 7.67 9.48 -3.369 -2.457 -1.703 -2.917 -1.514 -5.78 0.112 2.827 2.827 8.94 55.4 small [NaOH]: 0.8 x 240 2.00 7.78 9.48 -3.347 -2.450 -1.733 -3.039 -1.654 -5.84 0.104 2.031 2.031 8.96 60.8 small 240 1.83 7.86 9.54 -3.388 -2.474 -1.710 -2.936 -1.599 -5.73 0.107 2.836 2.836 9.01 64.9 small

x 480 1.81 7.96 9.54 -3.390 -2.488 -1.724 -2.917 -1.625 -5.68 0.105 2.973 2.973 9.05 69.7 small

120

*x indicates the addition of 20% water to measure solubility from undersaturation conditions.

121

APPENDIX E. CONSISTENT STRUCTURE OF ACC AT EACH RESIDENCE TIME

Figure E1. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition A. ACC short-range order is approximately constant over the course of the experiment.

122

Figure E2. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition B. ACC short-range order is approximately constant over the course of the experiment.

123

Figure E3. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 minute residence time, condition C. ACC short-range order is approximately constant over the course of the experiment.

124

Figure E4. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition D. ACC short-range order is approximately constant over the course of the experiment.

125

Figure E5. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition E. ACC short-range order is approximately constant over the course of the experiment.

126

Figure E6. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition H. ACC short-range order is approximately constant over the course of the experiment.

127

Figure E7. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition I. ACC short-range order is approximately constant over the course of the experiment.

128

Figure E8. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 and 8 minute residence times, condition J. ACC short-range order is approximately constant over the course of the experiment.

129

Figure E9. Comparison of ACC PDF profiles averaged over 5 minutes for experiments with 4 minute residence time, condition K. ACC short-range order is approximately constant over the course of the experiment.

130