MIAMI UNIVERSITY The Graduate School
CERTIFICATE FOR APPROVING THE DISSERTATION
We hereby approve the Dissertation
Of
Jennifer Christine Anderson
Candidate for the Degree:
Doctor of Philosophy
Andre Sommer (Advisor)
Gary Lorigan (Reader)
Stacey Lowery-Bretz (Reader)
Neil Danielson (Committee Chair)
John Rakovan (Graduate School Representative)
ABSTRACT
QUANTITATIVE AND QUALITATIVE INVESTIGATIONS INTO URINARY CALCULI USING INFRARED MICROSPECTROSCOPY
by Jennifer Anderson
This dissertation encompasses research focused on both qualitative and quantitative techniques for the analysis of loose urinary calculi (often referred to as renal stones) as well as biopsied tissue sections containing urinary calculi material. Due to high rates of renal calculi misdiagnosis, qualitative techniques that are efficient, accurate, and lack long sample preparation are desperately needed in clinics and hospitals around the world. The techniques presented in this dissertation rely on the use of infrared molecular microspectroscopy and infrared spectral maps for unbiased, fast, and accurate methods of both urinary calculi and tissue analysis. Utilizing infrared reflectance techniques, samples of varying size, shape, thickness, and consistency are easily and quickly analyzed. Additionally, infrared reflectance techniques are non-destructive, allowing the analysis of delicate samples without their contamination or destruction. Infrared spectral maps pin-point and visually differentiate urinary calculi components in a matter of minutes, resulting in an unbiased and accurate diagnosis.
Not only are qualitative techniques and results detailed in this dissertation, but quantitative results are included as well. Using infrared attenuated total internal reflectance (ATR) techniques, detection limits of urinary calculi components are investigated. Both concentration detection limits and particle size detection limits are important in the quantification of urinary calculi components. ATR analysis techniques allow the quantitative analysis of particles as small as 10 μm in size, as well as the analysis of concentrations of urinary calculi components as low as 1% by weight.
Finally, an appendix is included outlining the procedure, sample types, and results of an EPA internship lasting 900 hours. The focus of this internship was the analysis of corrosion samples from public and private drinking water systems. The main capacity of this research was confirmatory in nature, using Raman microspectroscopy to confirm or deny the presence of certain lead and copper compounds.
QUANTITATIVE AND QUALITATIVE INVESTIGATIONS INTO URINARY CALCULI USING INFRARED MICROSPECTROSCOPY
A DISSERTATION
Submitted to the Faculty of
Miami University in partial fulfillment
of the requirements
for the degree of
Doctorate of Philosophy
Department of Chemistry and Biochemistry
By
Jennifer Christine Anderson
Miami University
Oxford, Ohio
2007
Dissertation Advisor: Dr. Andre Sommer
Table of Contents Page Chapter 1 Introduction 1 1.0 Introduction 2 1.1 Research Outline 2 1.2 Characteristics of Renal Stones 3 1.2.1 Kidney Disease 3 1.2.2 Suspected Stone Nucleation and Growth 4 1.2.3 Types of Stones 13 1.2.4 Causes and Statistics 15 1.2.5 Analysis 16 1.2.6 Treatment 16 1.3 Instrumentation 18 1.3.1 Infrared Spectroscopy 18 1.3.1.1 Transmission 18 1.3.1.2 Reflection 20 1.3.1.3 Molecular Microspectroscopy and the Infrared Microscope 28 1.3.1.4 Infrared Molecular Imaging 35 1.3.2 Raman Analysis 36 1.3.2.1 Principles of Raman Microspectroscopy 36 1.3.2.2 Raman Microspectroscopy 38 1.3.2.3 Raman Analysis of Pathological Samples 38 1.4 Conclusions 39
Chapter 2 Analysis of Urinary Calculi Using an Infrared Microspectroscopic Surface Reflectance Imaging Technique 40 2.0 Overview 41 2.1 Introduction 41 2.2 Materials and Instrumentation 44 2.3 Results and Discussion 45
ii 2.3.1 Surface Reflectance 46 2.3.2 Polished vs. Unpolished Sample Surfaces 55 2.3.3 Method Comparison 60 2.3.4 Method Reproducibility 63 2.4 Conclusions 66
Chapter 3 Determination of the Detection Limit of Homogeneously Dispersed Calcium Oxalate Monohydrate in a Protein Matrix and Band Fitting of the 1618 cm-1 Oxalate Band 67 3.0 Introduction 68 3.1 Experimental 70 3.1.1 Materials and Instrumentation 70 3.1.2 Sample Preparation 70 3.1.2.1 Calcium Oxalate Monohydrate in a Gelatinous Protein Matrix 70 3.1.2.2 Calcium Oxalate Monohydrate in a Thin Film of Gelatin 70 3.1.2.3 Calcium Oxalate Monohydrate in High Viscosity Protein 71 3.1.2.4 Calcium Oxalate Monohydrate in a Protein Slurry 71 3.2 Results and Discussion 72 3.2.1 COM/Protein Mixtures 72 3.2.2 Band Fitting of the 1618 cm-1 Calcium Oxalate Monohydrate Absorption 78 3.2.3 Reproducibility 85 3.3 Conclusions 86
Chapter 4 Determination of the Minimum Identifiable Quantity of Calcium Oxalate Monohydrate Particles in a Protein Matrix 88 4.0 Introduction 89 4.1 Experimental 93 4.1.1 Materials 93 4.1.2 Instrumentation 93 4.2 Results and Discussion 94
iii 4.3 Conclusions 107
Chapter 5 Non-Linear Calibration Curves of Renal Stone Components Analyzed by ATR-IR 108 5.0 Introduction 109 5.1 Experimental 112 5.1.1 Materials 112 5.1.2 Instrumentation 112 5.1.3 Mixture Preparation 113 5.1.4 Peak Areas and Analysis Method 113 5.2 Experimental Calibration Curves 118 5.2.1 Theoretical Considerations 114 5.2.2 Experimental Calibration Curves 106 5.3 Particle Sizes 128 5.3.1 Surface Area and Particle Size 138 5.3.2 Depth of Penetration and Particle Size 146 5.3.3 Other Considerations 154 5.3.3.1 Quadratic Equations 154 5.3.3.2 Component Crushing 157 5.4 Reproducibility 157 5.5 Determining Accuracy using Non-Linear Calibration Curves 159 5.5.1 Artificial Sample 159 5.5.2 Real-World Sample 160 5.6 Conclusions 160
Chapter 6 Investigating the Interface between Tissue and Mineral Portions of a Renal Biopsy using Infrared Microspectroscopy 161 6.0 Introduction 162 6.1 Stone Former Characteristics 164 6.2 Experimental 169 6.2.1 Materials 169
iv 6.2.2 Instrumentation 169 6.3 Discussion 170 6.4 Conclusions 191
Conclusions to Renal Stone Research 193
Possible Future Research Pertaining to the Study of Renal Stones 194
Appendix EPA Internship 195 A.1 Introduction 196 A.1.1 Sample Preparation 197 A.1.2 Sampling Procedures 202 A.1.3 Instrumentation Parameters 202 A.2 Standards 203 A.3 Results 205 A.4 Discussion 213 A.5 Conclusions 219
References 220
v List of Tables Page Table 2.1: Method Reproducibility Data 64 Table 5.1: Renal Stone Components and their Linearity 119 Table 5.2: Particle Size and Linearity Matrix 129 Table 5.3: Uncertainty in COM Measurements 158 Table A.1: EPA Standards 204 Table A.2: EPA Results 206
vi List of Equations Page Equation 1.1: Beer-Lambert Equation 18 Equation 1.2: Absorbance 20 Equation 1.3: Snell’s Law 21 Equation 1.4: Critical Angle 26 Equation 1.5: Depth of Penetration 28 Equation 1.6: Spatial Resolution 31 Equation 1.7: Energy 37 Equation 2.1: Absorbance and Reflectance 65 Equation 3.1: Beers Law 72 Equation 3.2: Propagation of Uncertainty 77 Equation 5.1: Equation for HAP/COM ratios 114 Equation 5.2: Surface Area 138
Equation 5.3: ns Calculation for Depth of Penetration 146 Equation 5.4: Quadratic Equation 156
vii List of Figures Page Figure 1.1: Evolution of Stone in Kidney 6 Figure 1.2: Molecular Image of Stone 8 Figure 1.3: Visible Image of Stone/Tissue Interface 10 Figure 1.4: Cross-sectioned Renal Stone 12 Figure 1.5: Various Visual Images of Renal Stones 14 Figure 1.6: Infrared Transmission 19 Figure 1.7: Diffuse Reflectance 22 Figure 1.8: The Reflection/Absorption Process 25 Figure 1.9: IRE/Sample Interaction 27 Figure 1.10: Infrared Microscope Diagram 30 Figure 1.11: Reflectance Mode of the Infrared Microscope 32 Figure 1.12: Spatial Discrimination Using the Infrared Microscope 34 Figure 2.1: Visible Image of Cross-Sectioned Renal Stone 47 Figure 2.2: Infrared Images of Stone Cross-Section 49 Figure 2.3: Stone Nucleation Vein 51 Figure 2.4: Infrared Images of the Two Hydrates of Calcium Oxalate 54 Figure 2.5: Comparison Between Bone Cross-section and Stone Cross-section 57 Figure 2.6: Comparison Between Polished and Unpolished Bone Cross-sections 59 Figure 2.7: Comparison Between Surface Reflectance, DRIFTS, and ATR 61 Figure 3.1: Theoretical COM//Protein Spectra 74 Figure 3.2: Experimental Data of COM/Protein Spectra 76 Figure 3.3: GRAMs Output for 5 wt% COM 79 Figure 3.4: GRAMs Output for 70 wt% COM 80 Figure 3.5: Theoretical Band Fitting Data 82 Figure 3.6: Experimental Band Fitting Data 84
Figure 4.1: Stained Tissue Section, Tissue Biopsy between BaF2 Windows 91 Figure 4.2: Molecular Image of COM in Protein 95 Figure 4.3: Particle Spectra 98 Figure 4.4: Calibration Curve of COM on Bare KCl 99
viii Figure 4.5: Calibration Curve of HAP on Bare KCl 100 Figure 4.6: Spectra of COM in Protein Matrix 102 Figure 4.7: Calcium Sulfate in Protein Spectra 104 Figure 4.8: Spectra of HAP on Protein 106 Figure 5.1: Theoretical Spectra of COM and HAP 115 Figure 5.2: Theoretical COM in HAP Calibration Curve 117 Figure 5.3a: 778 cm-1 Absorption of COM in HAP 121 Figure 5.3b: 1318 cm-1 Absorption of COM in HAP 122 Figure 5.3c: 1618 cm-1 Absorption of COM in HAP 123 Figure 5.4: COM in KCl 126 Figure 5.5: HAP in KCl 127 Figure 5.6: SEM Image of COM 130 Figure 5.7: SEM Image of HAP 131 Figure 5.8: Latex Spectrum 133 Figure 5.9: COM and Latex (Latex band) 134 Figure 5.10: COM and Latex (COM band) 135 Figure 5.11: HAP and Latex (Latex band) 136 Figure 5.12: HAP and Latex (HAP band) 137 Figure 5.13: Unequal Distribution of Particles (ATR-IRE) 140 Figure 5.14: Small Particles Outnumbering Large Particles (ATR-IRE) 141 Figure 5.15: COM in UA 143 Figure 5.16 HAP in UA 144 Figure 5.17: Evanescent Wave into an HAP Sample 148 Figure 5.18: HAP Stretch as HAP Concentration Increases in COM 151 Figure 5.19: Depth of Penetration of COM in HAP 153 Figure 5.20: Quadratic Trend Lines 155 Figure 6.1: Randall’s Plaque 165 Figure 6.2: Diagram of the Ducts of Bellini and the Thin Loops of Henle 167 Figure 6.3: Areas of Interest in the Kidney 168 Figure 6.4: Visual Image of Tissue-mineral Boundary 171 Figure 6.5: Visual Image of Tissue-mineral Boundary 172
ix Figure 6.6: View of Thin Epithelium Layer in the Kidney 174 Figure 6.7: Molecular Image of the Tissue/Stone Interface 176 Figure 6.8: FT-IR Spectra of Tissue, Interface, and Stone 179 Figure 6.9: Layers of Protein and Stone Material 181 Figure 6.10: Heated and Unheated HAP 183 Figure 6.11: Possible Amorphous Stone Center 185 Figure 6.12: Visual Image of Tissue Section 187 Figure 6.13: Molecular Image Showing Calcium Oxalate 188 Figure 6.14: Spectra from Molecular Image 190 Figure A.1: Intact Pit Cap 198 Figure A.2: Pit Cap Removed 199 Figure A.3: Cross-sectioned Pipe Corrosion 201 Figure A.4: Copper Acetate Standard Spectra 216 Figure A.5: Spectrum of Iron Media 218
x Dedication
To my beloved husband: for being so understanding, supportive, and encouraging.
“When I came here, I was confused on this topic. Now I am still confused, but at a higher level.” — Enrico Fermi
xi Acknowledgements
Wow! It was a long but great 5.5 years. Depending on when you talked to me and what stage I was going through, I either was loving graduate school or else I couldn’t wait to get out; but I guess that’s how it’s supposed to be, isn’t it? Now that it’s finished, I am amazed at how naïve I was when I came in, and how much I have learned over the past several years. I feel I am leaving with more questions than answers sometimes, but that is all a part of research, as well as a part of gaining wisdom as opposed to just intelligence. Thank you: First to God, who saw me through my very tough times personally and professionally, as well as my joys and successes. Second to Nathan, who was never short of encouragement and always had words of wisdom (even when I really didn’t want to hear them). I never could have done this without you my love! If you can stick with me through this, you can stick with me through anything. You are my greatest gift, and I love you. Then to my parents, who let me learn valuable lessons early in life so that I’d have the strength of character and personal responsibility to get through everything later on. They are the best parents anyone could ever hope for (even if I did realize that they were great later on in life). Kathy, the continuing glue of the lab who always made sure everyone’s birthday was celebrated with an Entenmens Caramel Nut Twist roll and a card, and never failed to get us registered for conferences even when the deadline was passed. My co-workers, both past and present: Dr. Lou Tisinger (Captain Lou), Dr. Brian Patterson (invaluable advice), Luis Lavalle (never ending source of fun), Adam Lanzarotta (yes, he does talk every now and then, believe it or not), and Heather Stahl (she likes to read poetry in the middle of classes I hear). Friends: Meghan, my closest friend and adventure-buddy. Some day we’ll do that cage diving with sharks thing I know you’re dying to try. And thanks especially for our yearly paddle- boating excursion and going with me on Black Fridays at 4 a.m. to fight crowds and shop! And for our 4-hour runs… Yes, especially the 4-hour runs! Kerry, my roommate and dear sister in Christ, thanks for loving to clean and our Tuesday night talks. Tracy, who was my marathon buddy early on—thanks for coming back at the State-to-State for me! And so many others— thank you for being people I can talk to, confide in, and enjoy lazy days with. I really appreciate you all. What will I do without each of you in my life? It will be so boring! Thanks to my committee, Drs. Rakovan, Lorigan, Bretz, and Danielson, for all of your comments, encouragement, guidance, and toughness at times when I’m sure I needed it (but rarely appreciated it). And finally to Andy, who never let me eat smelly foods in lab without bringing it to my attention that I may very well be working at McDonalds if I kept it up. And who taught me so much about the differences between spectral and spatial resolution, optics, making up my own words, and that just about everything can wait until the day it’s due. Sometimes even the day after! Thanks for taking a chance on me. I hope that I contributed in some way towards the goals of both the research and the lab group. Favorite memories: Nathan cooking and brining me dinner on the nights I was at lab for far too long. Andy’s wooden Santa (it’s just not Christmas without it). Meghan and I skipping a day of work to go paddle boating ever summer (don’t worry, it was just one day every summer, not all the time…). Girls nights, with Desperate Housewives and Grey’s Anatomy (and wine!). Our annual Christmas candy-making day. Kicking people awake during CHM600 seminar. And last but not least: GRADUATION!!!! That hood looks awesome…
Thanks!!!!
xii
CHAPTER 1
Introduction
1 1.0 Introduction
This dissertation encompasses research focused on both qualitative and quantitative techniques for the analysis of loose urinary calculi (often referred to as renal stones) as well as biopsied tissue sections containing urinary calculi material. Due to high rates of renal calculi misdiagnosis, qualitative techniques that are efficient, accurate, and lack long sample preparation are needed in clinics and hospitals. The techniques presented in this dissertation rely on the use of infrared molecular microspectroscopy and infrared imaging for unbiased, fast, and accurate identification of both urinary calculi components and tissue analysis. Utilizing infrared reflectance techniques, samples of varying size, shape, thickness, and consistency are easily and quickly analyzed. Additionally, infrared reflectance techniques are non-destructive, allowing the analysis of delicate samples without their contamination or destruction. Infrared imaging visually differentiates urinary calculi components in a matter of minutes, resulting in an unbiased and accurate diagnosis. Not only are qualitative techniques and results detailed in this dissertation, but initial attempts at quantitative analysis are investigated. Using infrared attenuated total internal reflectance (ATR) techniques, detection limits of urinary calculi components are investigated. Both concentration detection limits and particle size detection limits are important in the quantitation of urinary calculi components.
1.1 Research Outline Chapter 2 describes in detail the surface reflectance studies undertaken in order to collect molecular images of cross-sectioned renal stone surfaces. Chapter 3 focuses on limits of quantitation for tissue models containing finely dispersed renal stone components. Band fitting procedures are also investigated for the oxalate 1618 cm-1 absorption in the presence of the overlapping amide I band of protein. Chapter 4 focuses on providing an in-depth explanation of particle-size detection limits using ATR-IR microspectroscopy. Chapter 5 discusses non-linear trends in calibration curves of renal stone components, as well as possible optical and concentration effects responsible for these non-linear trends. Chapter 6 describes ongoing research contributing towards the
2 work of Drs. A. Evan, F. Coe and J. Williams investigating the interface existing between tissue and mineralized portions of the renal stone. Finally, an appendix is included outlining the procedure, sample types, and results of an EPA internship lasting 900 hours. The focus of this internship was the analysis of corrosion samples from public and private drinking water systems. The main capacity of this research was confirmatory in nature, using Raman microspectroscopy to confirm or deny the presence of certain lead and copper compounds.
1.2 Characteristics of Renal Stones Urinary calculi (renal stones) are insoluble mineral deposits, the majority of which contain calcium oxalate monohydrate (COM), hydroxylapatite (HAP), or calcium oxalate dihydrate (COD) within their makeup, though other compounds such as uric acid (UA) are also common. [1, 2] Renal stones containing 82 components have previously been discovered. [3] 1.2.1 Kidney Disease The term kidney disease encompasses several ailments, and is a broad term applicable to damaged tissue or cells in the kidney. Alternate terms often associated with kidney disease are renal disorder, nephropathy, and nephrosis. The development of renal stones, which are mineral aggregations residing in the kidney, is often termed a type of renal or kidney disease due to the damage incurred by the papillary tissue. [4-6] Though the exact mechanism of renal stone formation is currently unknown, several stages of renal stone evolution have a tendency to cause tissue damage. The first stage in renal stone formation is the presence of renal tissue containing finely dispersed mineral components. It is unknown whether homogeneous mineral disbursement throughout the tissue is the result of a change in urine chemistry or whether this state induces a change in urine chemistry. The second stage in renal stone formation is the creation of mineral deposits in interstitial openings (open pockets) in the renal tissue, causing tissue damage as well as nucleation points for larger renal stone growth. The final step is the growth of large renal stones that necessitate treatment.
3 Renal stones thrive in environments where concentrated urine is present. The concentrated urine contains increased levels of minerals and salts that have the ability to interact with nucleation sites in the kidney tissue. The formed renal stones are capable of detaching from the tissue wall and becoming lodged in the urinary tract. [7] The chemical makeup of renal stones varies with physiological chemistry, and can be altered as their surroundings change. The nucleation and growth of renal stones remains a diverse field currently under intense study. 1.2.2 Suspected Stone Nucleation and Growth Papillary renal tissue can contain interstitial sites inside of which mineral deposits can develop. Interstitial sites, or voids in tissue where a fluid or air pocket resides, are ideal sites of plaque formation, commonly termed Randall’s plaque. (Figure 4.1) Randall’s plaque, discovered by Alexander Randall, is a layered calcium-based mineral deposit present over epithelial tissue and is located in interstitial openings on the internal surfaces of the kidney. In the late 1930’s, Dr. Randall examined the kidneys of cadavers, finding a thick mineral plaque in some areas that he believed was the precursor of renal stones. [8] The plaque found in these interstitial sites of tissue sections as well as on the surface of the renal tissue is predominantly hydroxylapatite, even in calcium oxalate stone formers such as idiopathic calcium stone formers (ICSF). [9-11] The majority of samples analyzed in this dissertation are formed by ICSF’s. ICSF’s are characterized by the formation of Randall’s plaque in interstitial sites and form, nearly exclusively, calcium oxalate renal stones. [11, 12] It is thought that ICSF’s form calcium oxalate renal stones in a unique manner. As Randall’s plaque increases in surface area, a hard mineral inclusion is eventually formed. A thin layer of epithelial cells separates the Randall’s plaque deposit from the urine in the kidney. Eventually, as the thin layer of epithelial tissue is worn away, the mineral deposit (plaque) is exposed directly to the urine and begins a reaction that is still under investigation. More detail on this mechanism is provided in Chapter 6 of this dissertation. The environment of the urine allows the stone to mature, coating layer upon layer until the stone eventually becomes large. Figure 1.1 details one suspected evolution of the mineral inclusion (plaque occupying an interstitial site) into a renal stone. [13] In this example of the formation of an ICSF renal stone, the site of nucleation is off-center and
4 closer to the side of the stone rather than the center.
5 Figure 1.1: The hypothesized evolution of the stone in the kidney.
A-Mineral inclusions/Randall’s B-The epithelial wall is C-Mineral components plaque (black spheres) are removed or worn away, and begin to amass and present in tissue but shielded the mineral inclusion is build layers around the from the urine environment by a exposed to the urine exposed plaque thin layer of epithelial cells environment. inclusion. (black stripes). Urine Environment Tissue Environment
Interstitial Sites Epithelial Layer (plaque)
D-Over time, the E-Eventually, the large layers are built up stone breaks away from the around the original nucleation site, generating a mineral inclusion (the loose renal stone and nucleation site), leaving the initial mineral gradually forming a inclusion formed from larger and larger stone. Randall’s plaque behind and exposed to the urine environment once more.
6 Figure 1.1 A represents the interstitial sites (black portions) of mineral deposits originating from Randall’s plaque beneath a thin layer of epithelial cells (black lines). In the next stage of evolution, Figure 1.1 B illustrates how these mineral deposits are exposed to the environment of the kidney once the thin layer of epithelial cells are removed. In Figure 1.1 C, deposits of renal stone material are built up in concentric layers around the opening in the epithelial cells with the mineral deposit as the nucleation site. Figure 1.1 D illustrates the growth of the renal stone in the kidney, while Figure 1.1 E displays a detached renal stone with the nucleation site off-center on the outer edge of the stone. Notice that a portion of the initial interstitial mineral deposit remains embedded in the tissue section. Figure 1.2 provides a molecular image of a calcium oxalate renal stone that is believed to have been formed via this mechanism. The stone composition in Figure 1.2 is primarily calcium oxalate, with a small section of hydroxylapatite located in the lower left portion of the stone. This area may correspond to the site of attachment for the stone to the tissue wall, and may be the result of Randall’s plaque exposed to the urine.
7 Figure 1.2: A molecular image of a renal stone with a possible nucleation site originating from Randall’s plaque, possibly evolving from the reaction taking place in Figure 1.1. Left—areas of high concentration of calcium oxalate. Right—areas of high hydroxylapatite concentration.
8 It is known that there is interaction between the Randall’s plaque and the urine, however, the interface between the kidney tissue and the mineralized material has been rarely investigated, but could hold the key to kidney stone formation. [14] Using infrared molecular microspectroscopy (elaborated upon in section 1.3.1.4), the interface between the tissue and the renal stone has recently been explored. Figure 1.3 below is a visible image of an interface between tissue and renal stone material. The purple portion is stained tissue, while the white crystalline portion is part of a renal stone.
9 Figure 1.3: Visible image of the interface between tissue (purple) and renal stone material (white crystalline portion). Taken at 50x magnification by J. Anderson.
10 Observing infrared spectra collected from the tissue section (Figure 6.8), one can track the orthophosphate stretch of HAP at 1018 cm-1 as the path of analysis travels from the tissue across the interface. It was observed that the absorption belonging to hydroxylapatite becomes broad and undefined at the interface compared to the sharper HAP band present in the mineralized portion of the sample. This observation is possibly indicative of disordered mineral material, and will be detailed further in Chapter 6. However, as the site of analysis moves from the interface to the crystalline renal stone material, the sharpness of the absorption band increases, implying a possible increase in crystalline order. Another possibility for the increase in broadness of the HAP band is that the interface consists of nano-crystals of HAP. Chapter 6 discusses this hypothesis in further detail. In order to determine if nano-crystals are responsible for band broadening, other techniques may be needed to study the interface between the renal tissue and the crystalline renal stone. These alternate techniques may include AFM and SEM. Using these techniques, a determination can be made as to the hydroxylapatite crystal size present at the interface, allowing one to conclude whether the band broadening at the interface is due to disordered crystal structure or simply due to the presence of nano- crystals. In ICSF’s, the interaction between the Randall’s plaque interface material and the urine environment produces renal stones with a HAP nucleation point and CaOx body. However, other renal stones exhibit formation indicative of alternate mechanisms. For example, renal stones such as that illustrated in Figure 1.4 have been analyzed in this laboratory and appear to have been layered much like a pearl (apparent inward to outward growth). However, the formation could also be indicative of a geode-like formation (outward to inward growth). The precise method of renal stone formation is as of yet unknown.
11 Figure 1.4: A cross-sectioned renal stone with concentric layers. The superimposed box corresponds to the designated area of interest for full spectral maps.
12 Figure 1.4 is a common renal stone configuration. It is suspected that the layering of renal stone components in stones such as Figure 1.4 occurs in one of two mechanisms. The first resembles the layering of a pearl inside an oyster. Over time, a loose nucleation point (particle of renal stone material) is evenly covered with deposited mineral material until eventually the stone grows in size and is removed or passed. This would be an example of an alternative point of nucleation other than Randall’s plaque. The second mechanism may model geode formation. [15, 16] As the interstitial site fills with Randall’s plaque, a stone begins to form. In this model, the oldest layers would be on the outside surface of the renal stone, while the center of the stone would be the last portion to form. Essentially, the renal stone is formed from the outside toward the center. Here, the nucleus is the outside layer. Regardless of the mechanism of renal stone formation, analysis of an entire cross- sectioned stone will result in a more accurate diagnosis than if just the surface of the stone was analyzed. 1.2.3 Types of Stones Renal stones can be formed into various morphologies, depending upon the growth environment and the compounds present in the stone. Figure 1.5 illustrates various morphologies of calculi.
13 Figure 1.5: Visual images of various categories of renal stones. Taken from Louis C. Harring and Company, Analytical and Consulting Chemists, Orlando, FL. http://www.herringlab.com/photos/
COM Brushite Struvite Cystine COM Uric Acid
COM Struvite COM COD Silica Struvite
Cholesterol Cystine Carbonate apatite Brushite Uric Acid Carbonate apatite
Apatite COM Struvite COM COD/COM COM/Apatite
Sulfamethoxazole Metabolite Ciprofloxacin Metabolite Oxypurinol
14 The most common varieties of calculi examined in our laboratory, calcium oxalate monohydrate (COM) and hydroxylapatite (HAP), have the potential to take on prickly, tentacle-like forms, generating a painful experience for the patient. However, the majority of stones examined are smooth and ovular. Other common components of calculi include calcium oxalate dihydrate (COD), brushite (a calcium phosphate), struvite (magnesium ammonium phosphate hexahydrate), uric acid (2, 6, 8-Trihydroxypurine), carbonate hydroxylapatite, and cystine (β,β'-Diamino- β,β'-dicarboxydiethyl disulfide), as displayed in Figure 1.5. In addition, drug metabolites have been found in the nucleus of certain calculi in patients suffering from specific medical conditions. Daudon and Estepa have documented this phenomenon in an in-depth yet concise manner. [17] The bottom three photos of Figure 1.5 illustrate drug-related renal stones. 1.2.4 Causes and Statistics The causes behind renal stone formation and recurrence vary, making a single broad statement of cause inaccurate. In addition to common renal stone formation, there exist several examples of documented renal stone growth in uncommon circumstances, such as space travel and as the result of gastric bypass surgery. In space travel, the effects of microgravity on the human body have been found to increase desorption of calcium salts from the bones, generating a urinary state of saturation; [18] urinary volume decreases and pH is more acidic during space flight, sharply increasing the potential to form renal stones. [19-23] The conditions resemble the mechanism taking place in dehydrated individuals, though dehydration during space travel may or may not be involved. As longer space missions become routine, understanding the chemistry and conditions for renal stone development in microgravity situations will become critical. Renal stone growth has also been studied in conjunction with obesity and its common treatment, gastric bypass surgery. [24-27] In this process, the stomach is surgically reduced to a much smaller size, thereby promoting drastic weight loss. Unfortunately, approximately 14% of the patients that have undergone gastric bypass surgery produce renal stones of a unique variety. [28, 29]
15 Another suspected cause of renal stone formation, genetic pre-disposition, is debated in urological circles. However, evidence does suggest that the inability of one’s body to properly process oxalate, cystine, or excess calcium can lead to renal stones. [4, 30-36] Additionally, excess weight has also been linked to renal stone development. [37- 46] 1.2.5 Analysis The current analysis of renal stones relies heavily upon visual methods, though x- ray diffraction (XRD) and diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) techniques are becoming more accepted and routine. However, XRD and DRIFTS require experienced investigators, and expensive instrumentation, both of which are costly investments. Because proper training and instrumentation can be costly, the majority of clinics and hospitals rely on the judgment of trained urologists who utilize microscopy and personal experience in the diagnosis of renal calculi. Unfortunately, visual diagnoses can be incorrect up to 40% of the time, presenting difficulty for patients requiring treatment for recurring bouts of renal calculi. [47, 48] Because a treatments effectiveness is specific for different renal stone types, accurate diagnosis is imperative in the early stages of the affliction. The preparation of renal stones for visual diagnosis includes the dehydration of the stone with subsequent microscopic examination to determine crystal forms and sizes. Generally no staining is required for loose stone sections, though it is a common practice for embedded mineral deposits in tissue sections (often the nucleation or attachment site of the stone). [43, 49] Visual diagnosis is often made based on the crystalline structure and color of the crystal. However, several different compounds may share the same crystallinity or birefringence. [1] For example, COM and sodium acid urate are both monoclinic, though sodium acid urate can also be triclinic.[1] Therefore, misdiagnosis using visual methods alone are foreseeable. 1.2.6 Treatment It is estimated that the total cost associated with evaluation, treatment, and outpatient care due to renal stones in the United States is approximately $1.83 billion annually [50]. The treatment and removal of renal calculi is often dependant upon the
16 composition and hardness of the stone, as well as stone size and placement in the urinary tract. Though the main components of renal stones tend to be calcium oxalate, calcium phosphate (hydroxylapatite), and uric acid, calculi with up to 82 components have been previously analyzed. [51] Treatment options vary depending on the placement of the stone in the renal system. The most preferable and least invasive treatment is, of course, natural passing, though this option is often the most uncomfortable for the patient. The option of natural passing is dictated by the size of the renal stone, since blockage of the urethra will occur with renal stones larger than approximately 6 mm in diameter. [52] Intractable renal stones are treated through several different techniques, two of the more well-known methods being extracorporeal shockwave lithotripsy (ESWL) and surgical removal. Unfortunately, ESWL causes renal tissue trauma and hemorrhaging with virtually every treatment [53], while surgical methods are invasive and require longer recovery times. ESWL subjects the patient to a series of intense sound pulses directed at the stone position in the urinary tract. The depth of the pulse that passes through the body is determined by the scope of the pulse: broad pulses have a larger field of effectiveness but are not as powerful, while a more focused pulse is more powerful but also more difficult to direct to the exact location of the renal stone. In addition, the more narrow pulses tend to injure internal tissue faster and more dramatically than the broader pulses. [47] There are recent concerns as to the long term side-effects of lithotripsy and the damage caused to the renal system. [53-55] Current research indicates that in addition to soft tissue damage, the development of brushite renal stones is directly related to treatment by ESWL, where HAP, COM, or UA stones are removed but brushite stones are formed in their place. [56] Percutaneous nephrolithotripsy (the shattering of stones) and percutaneous nephrolithotomy (the removal of stones) involves the insertion of a narrow endoscope through the back of the patient into the kidney in order to break up or remove the stone, respectively. This procedure is often used when ESWL has failed and is usually attempted prior to open surgery, which is considered a last resort. Ureteroscopy is similar
17 to percutaneous nephrolithotomy, but in this case, the endoscope is inserted through the urethra instead of through the back of the patient.
1.3 Instrumentation 1.3.1 Infrared Spectroscopy The source of energy for an infrared system is a broadband source, meaning that a large range of wavelengths are incident upon the sample during analysis. Infrared wavelengths, from 2.5 to 25 μm, may be absorbed by molecules in the sample. As the incident electromagnetic radiation impinges upon the sample, specific wavelengths are absorbed. These wavelengths correspond to specific harmonic vibrations of covalent bonds within the molecule, producing a unique fingerprint for each molecule. This absorption is measured by the detector and is proportional to the Beer-Lambert equation (Equation 1.1), stating that the absorbance is a function of the extinction coefficient ε (A cm2g-1), the path length of radiation through the sample, b (cm), and the density of the sample, ρ (g/cm3). [57] The absorption, A in Equation 1.1, is also defined as the log of the power of the incident radiation divided by the power of the transmitted radiation.
A=εbρ Equation 1.1
1.3.1.1 Transmission
When an incident beam of radiation Io is transmitted through a medium, the energy of the beam is attenuated as a direct result of several factors. Energy is lost both
as radiation enters and exits the sample in a reflection (IR) off phase boundaries; this loss results from differences in the refractive indices of the medium and its surroundings, usually air. Energy is also lost due to the scattering and absorption that takes place by the sample itself. These losses of radiation combine to yield a transmitted beam (I) of energy less intense than the incident beam. Ideally however, absorption by the sample should be the only mechanism responsible for loss of radiant intensity. Great care must be taken to ensure that the other components (reflection, scattering) are in most cases negligible, should one want to attempt a quantitative analysis. [58] Figure 1.6 illustrates these principles of infrared transmission spectroscopy.
18 Figure 1.6: Sample undergoing an infrared transmission process. The incident radiation, Io, enters the sample. Due to a phase boundary between the air and the sample surface, some infrared radiation is reflected in a specular manner (IR), while other radiation is diffusely reflected (ID) at the second phase boundary. The resulting transmitted radiation contains a much smaller energy than the original beam of radiation. The resultant radiation (I) has been attenuated, though the loss of energy does not originate completely with absorbance by the sample.
ID IR
Io I
IR
ID
19 In transmission mode, the sample must be freestanding or placed on a non
infrared-absorbing substrate, such as a KBr, KCl, or BaF2. Ideally, the path length of the radiation is considered to be the sample thickness since radiation enters the sample at or near normal incidence. Transmittance (T), related to absorbance (A) through Equation 1.2, is the usual manner in which infrared spectra are displayed for qualitative comparisons.
A=-log (T) Equation 1.2 %T=T*100
Above, T is equal to the ratio of the power of the transmitted energy to the incident energy, or I/Io. In any infrared analysis, it is desirable to have the absorbance bands registering between 80 and 20 % T. An absorption band located within this range is termed photometrically accurate, and can be used for quantitative as well as qualitative purposes. Bands producing an absorbance lower than 20 % T indicate that the sample is too thick, while bands less than 80 % T indicate that the sample is too thin. In general, the ideal thickness of a sample for transmission analysis is approximately 6 μm. [43] However, this guideline is for visual reference and qualitative analysis only. All quantitative measurements are conducted in Absorbance, not %T. Examples of transmission analysis for biological specimens are many, and include lung and cervical tissue specimens associated with cancer. [59-61] Transmission spectra can yield excellent results both qualitatively and quantitatively. [43, 58, 62-64] 1.3.1.2 Reflection When determining which reflectance technique is best suited to a specific sample, it is helpful to know the samples optical details. Sample characteristics such as particle or crystal size as well as the sample smoothness are important considerations when evaluating the use of reflectance techniques such as specular, diffuse, or surface reflectance, and sample thickness is important when considering reflection/absorption techniques.
20 Five types of reflection techniques are referred to in this dissertation: specular, surface, diffuse, reflection/absorption (R/A), and attenuated total internal reflection (ATR). All of these methods are currently employed for the analysis of biological samples. However, each method requires that the sample have specific properties. For specular reflectance, the surface of the stone must be highly polished and optically flat. [65] In specular reflectance, the angle of the incident radiation (θi) equals that of the reflected radiation (θR) according to Snell’s Law.
sinθ i = sinθ R Equation 1.3
For most materials, only 5-10% of the energy is reflected from the surface. Specular reflectance spectra exhibit absorptions resembling the first-derivative of a normal transmission absorption band, though computer corrections can be applied to the spectrum to make it appear as a normal transmission spectrum. Diffuse reflectance requires that the sample be a highly scattering powder with individual crystal sizes that are smaller or approach that of the wavelength of radiation being employed in the analysis. Radiation enters the sample and undergoes a complex path, which can involve reflection, refraction, scattering, and absorption, as is shown in Figure 1.7 [66]; there is no angular relationship between the incident and diffusely scattered electromagnetic radiation. As the crystal size increases, the surface of the sample appears more flat and smooth, increasing the amount of specular reflection.
21 Figure 1.7: Diffuse Reflectance
22 Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) is a technique often used in the pharmaceutical industry to determine the enantiomeric purity and homogeneity of active drugs dispersed in benign excipients. [67-79] The field is well established, and several quantitative methods using this technique have been approved by the Food and Drug Administration (FDA). DRIFTS has been used as a tool for the analysis of urinary calculi with some success. [80] Though the method produces quantitative results, the preparation of the calculi samples is time consuming. In a previous report, the calculus was dissected layer by layer; this process often results in small amounts of stone for analysis, and in the event of error, may not provide enough material for an accurate study. [80] Additionally, the concentric rings of material surrounding the nucleus of the renal stone may not be uniform in either thickness or consistency, presenting opportunity for both qualitative and quantitative errors. In order to obtain photometrically correct data, the sample has been diluted with KBr. Several problems potentially exist when a sample is ground and diluted with KBr (the most common non-infrared absorbing medium used in conjunction with infrared techniques, though KCl is also commonly used) for DRIFTS analysis. Some of these complications can include bromine ion exchange with sample components, or water absorption by the KBr, which is hygroscopic to a certain degree. [80] Surface reflectance is very similar to specular reflectance in that it is strictly a surface technique with penetration of the sample by the infrared radiation being minimal. However, the resulting surface reflectance spectra of cross-sectioned renal stone samples are more DRIFTS-like in appearance than specular. Unlike specular reflectance, surface reflectance spectra are relatively free of derivative-shaped. Surface reflectance spectra resemble DRIFTS spectra due to the characteristics of the cross-sectioned renal stone sample. It is a fortunate coincidence that the apparent characteristics of renal stones permits DRIFTS-like spectra to be obtained without significant sample preparation like common DRIFTS. For examples of papillary renal tissue research containing mineral inclusions, a technique termed reflection/absorption (R/A) analysis has been previously used. [43, 46] This method is also referred to as reflection-absorption or trans reflectance. In the R/A
23 method, a reflective substrate or low-energy (low-e) glass slide is utilized in order to allow the transmission of radiation through the sample. However, this radiation is consequently reflected off the surface of the substrate after being transmitted through the sample. When a sample is mounted on the reflective substrate, the path of the incident radiation proceeds through the tissue twice and reflects off the surface of the substrate once. This process allows the path length of electromagnetic radiation through the sample to be approximately doubled, which can enhance the absorbance for thin films or other transmissible materials, such as tissue sections. A diagram of the R/A process is given in Figure 1.8. A doubling of the path length results in a larger absorbance, theoretically increasing the signal while maintaining approximately the same level of noise. However, for highly absorbing materials, a path length that is too long can generate absorption bands that are photometrically inaccurate.
24 Figure 1.8: Infrared electromagnetic radiation as it enters a sample is refracted and reflected and exits the sample in a reflection/absorption (R/A) process.
25 The final reflectance method discussed in this section, attenuated total internal reflection (ATR) analysis, involves the use of an internal reflection element (IRE), and an evanescent wave penetrating the sample to a depth that is related to the wavelength of the radiation, the IRE and sample index of refraction, and the angle at which radiation is incident through the IRE. This immersion method of analysis allows for photometrically accurate spectra to be collected with minimal sample preparation. However, it also carries with it the slight potential to damage the sample or contaminate the IRE since the IRE must be in intimate contact with the sample.
Typical values of the depth of penetration (dp) using a Ge IRE for calcium oxalate monohydrate samples range from 0.25 μm at small wavelengths (2.5 µm=4000 cm-1) to 1.4 μm at large wavelengths (14 µm=714 cm-1). Because of the small penetration depth, the ATR technique is ideal for highly absorbing samples and for surface sensitive measurements. ATR spectroscopy is a technique in which a single crystal IRE of a high refractive index material is placed into intimate contact with a sample. (Figure 1.9) IR radiation is passed through the crystal and is nearly totally internally reflected at the sample/IRE interface. The radiation internally reflects inside the crystal as long as it is at an angle
greater than the critical angle of the IRE, θc, which is given by Equation 1.4.
⎛ n ⎞ ⎜ s ⎟ θ c = arcsin⎜ ⎟ Equation 1.4 ⎝ nc ⎠
In the above equation, nc is the index of refraction of the crystal while ns is the refractive index of the sample.
26 Figure 1.9: The path of infrared electromagnetic radiation as it interacts with the sample and is internally reflected inside the germanium (Ge) IRE.
27 The small fraction of energy that is not internally reflected, termed the evanescent wave, penetrates into the sample to a depth dp, which is similar to the path length b discussed in the previous section. This penetration depth is determined by Equation 1.5.
λ d p = Equation 1.5 2 1/ 2 ⎡ ⎛ n ⎞ ⎤ 2πn ⎢sin 2 φ − ⎜ s ⎟ ⎥ c ⎜ n ⎟ ⎣⎢ ⎝ c ⎠ ⎦⎥
Here it can be seen that the depth of penetration increases with the wavelength of radiation, but is reduced by increasing the angle of the incident beam φ. In addition, the depth of penetration is increased as the refractive index of the sample (ns) approaches that of the crystal (nc). [81]
Though ATR spectra resemble transmission spectra, there can be ambiguities at longer wavelengths. As the wavelength increases, the depth of penetration dp (b in Equation 1.1) also increases, thereby increasing the absorbance (A in Equation 1.1). This increase in absorbance requires the spectrum to be adjusted for direct comparison to a transmission spectrum. This process is often referred to as the ATR Correction. [81] The ATR technique is more popular due to reduced sample preparation or sample restrictions of smoothness or size. Minimal sample preparation is needed in order to obtain quantitative and qualitative information, and there is no restriction on sample thickness due to the limited depth of penetration. 1.3.1.3 Molecular Microspectroscopy and the Infrared Microscope Infrared spectroscopy is termed a molecular technique as opposed to an elemental technique because of its abilities to identify specific molecular groups present in a sample. Molecular spectroscopy differs from atomic (elemental) spectroscopy in many respects. In atomic spectroscopy, a determination of the atoms present is the goal; information on how the atoms are oriented or linked to each other can be difficult to determine. Molecular spectroscopy yields data concerning molecular structure. Molecular microspectroscopy is the coupling of molecular spectroscopy with a
28 microscope, generating a powerful analytical tool useful in countless areas of science and research. In the past, standard optical microscopy techniques have been used to prepare biological or histological samples, define the area of interest, and determine parameters to be included in the analysis. Such microscopy techniques involve dissection of the sample, staining, and physical isolation of areas of interest in the sample makeup. Spectroscopy was subsequently used to determine the chemical composition and form of the analyte. In microspectroscopy, however, these two methods are indistinguishable; microspectroscopic imaging can be used to isolate an area of the analyte in question using an adjustable aperture, and areas of interest can be studied by obtaining a molecular image of the specimen J.A. Reffner writes that microscopy is “…the science of generating, recording and interpreting magnified images, while analytical spectroscopy is the science of emission, reflection, and absorption of radiant energy to determine the structure and composition of matter.” [82] Microspectroscopy, a complementary combination of microscopy and spectroscopy, has the advantages of both techniques with few limitations. A diagram of a standard infrared microscope is presented in Figure 1.10.
29 Figure 1.10: A schematic of an infrared microscope.
MCT
Detector Optics
Primary Image Plane
Objective
Sample
Condenser
Source
30 The infrared microscope can be used in all infrared sampling modes: transmission, diffuse reflection, specular reflection, surface reflection, and ATR. The instrument specifications and sampling modes will be discussed further in each section of this dissertation relative to the analysis of different sample types. As with all microscopes, the infrared microscope is limited by diffraction through the wavelength of energy that is employed for the analysis. The smallest sample size (spatial resolution) that can be analyzed without minimal spectral artifacts or contamination from near neighbors is given by Equation 1.6 below.
1.22λ d = Equation 1.6 n sinθ
In Equation 1.6, the sample size or beam waist is represented by d, n is the refractive index of the medium in which the measurement is conducted (usually air, where n=1), and λ is the wavelength used for analysis. The value of θ, the most extreme angle of the incident radiation from the normal, is set by the manufacturer. For the Perkin Elmer Spectrum Spotlight used in this research, sinθ = 0.6, with θ=37o. For transmission and reflectance measurements, the beam waist approximates 2λ and 4λ, respectively. Reflectance measurements using the infrared microscope utilize half of the objective to introduce radiation onto the sample surface, and the other half of the objective to allow radiation to exit the system to the detector. This path of radiation is illustrated in Figure 1.11. By using mirrors and a half-filled aperture (as opposed to a beam splitter), no loss of energy occurs since the entire energy from the source is transmitted to the sample and the detector both. However, there is a slight loss of spatial resolution since the aperture is only half-filled and only one-half of the objective is being utilized. Fortunately, the loss of spatial resolution can be compensated for by using a drop-down IRE, which offers a measure of magnification associated with the refractive index of the crystal as described in section 1.3.1.2 of this dissertation.
31 Figure 1.11: Reflectance mode of the infrared microscope.
MCT
Detector Optics
Primary Image Plane
Mirror
Source
Sample
32 Using an infrared microscope, small samples on the order of micrometers in size can be spatially isolated from their surroundings and analyzed. This analysis can be performed in any of the modes available, and is termed confocal microspectroscopy. Using spatial isolation on the x-y plane of the sample surface via the instrument aperture, the amount of sample information analyzed by the instrument can be increased or decreased according to the size and shape of the site of interest. An example of spatial discrimination could include the analysis of copier ink on a piece of paper (cellulose), or an impurity in a plastic vehicle component. In both cases, it is necessary to isolate the analyte (ink or impurity) from a matrix (cellulose or plastic) that is capable of yielding complex spectra where the analyte may be overlooked or the analyte and matrix spectra may be overlapping.
33 Figure 1.12: Spatial discrimination on the X-Y plane of the sample using the infrared microscope. The regions labeled A, B, and C represent different apertures, while the patterned areas represent differing areas of interest on the surface of a sample. A B C
Apertures A, B, and C
Lens
Lens
C B A
34 Apertures set in the instrument (Aperture line A, B, and C in Figure 1.12) define the beam of radiation impinging on the sample, such that areas of interest (patterned spheres on sample surface) are spatially isolated from the surrounding materials and are individually analyzed. 1.3.1.4 Infrared Molecular Imaging Infrared imaging refers to the generation of maps that are specific for molecular absorptions. This process is often referred to as molecular mapping or imaging, and can be performed in both transmission and reflection modes. The two detectors available on the infrared microscope used in this research are serial (point-by-point) and parallel in character. The point-by-point mode involves a singe-point detector that collects infrared spectra in a serial fashion over the entire designated area of the sample. This process yields high signal to noise data, but is significantly slower than the parallel imaging mode. In parallel imaging, a linear or 2 dimensional detector collects radiant information (i.e. spectra) from multiple points on the sample simultaneously. For the linear array on the Perkin Elmer Spectrum Spotlight, 16 spectra can be simultaneously collected. Infrared mapping in reflection mode has recently been performed with excellent results, and has specific applicability in the biological and pathological sciences. [43-46] Current examples of infrared mapping include examinations of dermal and sub dermal tissue, as well as cancerous tissue samples. However, reflection molecular imaging is presently a qualitative method of analysis rather than quantitative. ATR is also capable of producing a molecular image using the point-by-point mode previously discussed. [83] The drop-down Ge IRE comes into intimate contact with the sample, with the spatial resolution being set by the aperture of the instrument and the refractive index of the IRE. The ATR method of imaging produces possibly the best and most photometrically correct set of spectra, with the least amount of sample preparation of any infrared method. However, there remains the possibility that particles from the sample may adhere to the IRE, contaminating other sections of sample and yielding false or increased absorbance bands. Off-axis ATR imaging alleviates this problem, and has been performed previously on soft samples with some success. Lewis and Sommer have
35 successfully mapped soft samples of laminates using an ATR germanium hemisphere [83], and Patterson et al have used off-axis ATR imaging to analyze silicone based foam samples. [84] 1.3.2 Raman Analysis
Raman spectroscopy, a scattering technique that was observed by Indian physicist C.V. Raman in 1928, holds both similarities as well as vast differences from infrared analysis. Like infrared analysis, Raman spectroscopy provides information regarding molecular bonding groups present in a sample. Each compound has a unique Raman fingerprint just as it has a unique infrared fingerprint. However, how that fingerprint is generated is very different from infrared analysis. 1.3.2.1 Principles of Raman Spectroscopy In order for a sample to be termed Raman active, the molecules in a sample must undergo a dipole moment, either naturally or induced. [58, 85] The interaction of the monochromatic energy of the laser with the sample causes a dipole moment in the sample molecules, vibrating the molecules with a characteristic frequency. This characteristic frequency is associated with a harmonic vibration of covalent bonds within the molecule, and is one of the principles behind Raman spectroscopy. [86] A molecule that is Raman active will also undergo a nuclear displacement, where a change in polarizability must occur. Polarizability, or a measure of the ease with which electrons in the molecule are displaced, is a key component in Raman spectroscopy. The polarizability of the molecule relates the energy resulting from interactions of excitation radiation with the sample to the change in electron displacement occurring with a change in nuclear displacement. [86, 87] If the polarizability of a sample remains unaltered during a nuclear displacement, the sample is Raman inactive. When the excitation radiation strikes the molecule, energy can either be added to the incident beam (emission of a photon by the molecule: Eresultant=hvexcitation+hvvibration), removed from the excitation beam (absorption by the molecule: Eresultant=hvexcitation-
hvvibration), or elastically scattered (remaining the same, no interaction: Eresultant=hvexcitation- 0). [86] The relationship between the increase or decrease in the wavelength as a function of the energy is described in Equation 1.7 below, a re-arranged variation of Planck’s Equation.
36
hc λ = Equation 1.7 E
Incident radiation induces an oscillation in the molecule, and if the radiation is elastically scattered (no change in wavelength), the term is known as Rayleigh scattering. If, during this oscillation, the molecule interacts with the incident radiation to alter the incident wavelength, then the resultant effect is called Raman scattering. [58] When incident radiation is changed to a longer wavelength, yielding Stokes lines, it means that energy has been absorbed from the incident beam, decreasing the resultant energy. When energy is added to the incident beam in the form of an emitted photon, anti-Stokes lines result. Stokes lines are more intense than anti-Stokes lines due to the Boltzmann distribution of energy in vibrational states within the molecule. The Boltzmann distribution is the distribution of energy between the excited vibrational state and ground vibrational state of the molecule, with the majority of energy (99%) in the ground vibrational state and only 1% present in the excited state. Due to this distribution, the sample has a higher probability of absorbing energy than emitting energy as the radiation is impinged upon the sample. Therefore, Stokes lines, corresponding to the absorption of energy and the shortening of the wavelength, are more intense than anti-Stokes lines and are analytically more useful. A principal drawback to Raman analysis is fluorescence of the sample, which arises from an excited electronic state of the molecule. Because infrared does not have sufficient energy (wavelength) to excite electrons to fluoresce, infrared does not suffer from the effects of fluorescence. Fluorescence of the sample occurs when the incident beam (a visible excitation source) initiates both electronic transitions that produce fluorescence as well as vibrational transitions that produce Raman signal. [86] In non- FT-Raman systems, a technique termed photo-bleaching is often used, where the incident laser is allowed to excite the sample for an extended period of time (3-5 minutes) in order to reduce fluorescence to a level conducive to analysis. [88] Often, this technique
37 provides enough photo-bleaching of the sample such that an adequate Raman spectrum can be obtained. Additionally, in Raman spectroscopy, both heteronuclear and homonuclear molecules can be analyzed. Raman spectroscopy is sensitive to symmetric bonds and
vibrations, such as CH4. Infrared spectroscopy is sensitive to asymmetric bonds and vibrations, such as those exhibited by water. Therefore, though both techniques appear to visually yield similar data, the techniques are complementary in character, providing a full analysis of the sample. 1.3.2.2 Raman Microspectroscopy Raman microspectroscopy is a coupling of Raman spectroscopy with optical microscopy, similar to the coupling of infrared spectroscopy with optical microscopy as previously mentioned. Like infrared microspectroscopy, Raman microspectroscopy is also capable of performing confocal operations in the x-y plane using spatial filtering and a small, focused laser spot. Since the diameter of the laser spot in Raman is related to the wavelength of the radiation (Equation 1.6), smaller particles (approximately 10x smaller) are capable of being analyzed. [86] As in Equation 1.6, the beam waist (d) is directly proportional to the wavelength of energy used and indirectly proportional to the square of the numerical aperture (NA). For a NA value of 0.6 and a wavelength of 633 nm, the beam waist (d) is 2.15 μm. 1.3.2.3 Raman Analysis of Pathological Samples Presently, Raman spectroscopy is used for biological samples such as those found in tissue and cell research. [89-91] Raman analysis has been successfully used on a variety of biological samples, including membranes [92], nucleic acids [93, 94], virology studies [95], and biological warfare agents. [96] Unfortunately, biological samples have a tendency to fluoresce, and several adaptations of Raman spectroscopy have had to be made. [92, 97, 98] In some cases, surface enhanced Raman spectroscopy (SERS) is used in order to overcome the problem of high fluorescence in biological samples. [90] In addition to the generic Raman spectroscopy commonly used, highly sophisticated techniques are currently being developed in order to perform specific analyses. These include coherent anti-Stokes Raman spectroscopy (CARS) for the analysis of biological membranes [99], hole-enhanced Raman scattering (HERS) on thin
38 gold films to simulate molecular functional groups [100], and tip-enhanced Raman spectroscopy (TERS) for the analysis of bacterial surfaces. [101]
1.4 Conclusions In conclusion, this dissertation will provide research results for both qualitative and quantitative renal stone analysis. Using the samples, instrumentation, and methods outlined in this Introduction, this dissertation will show that accurate qualitative measurements can be quickly and efficiently obtained from the surface of cross-sectioned renal calculi, that detection limits of homogeneous mixtures of renal stone components that model finely dispersed compounds in tissue can be determined, and that tissue/stone interfaces can be successfully analyzed to aid in the elucidation of the chemistry of formation for renal stones. Additionally, information on non-linear calibration curves is provided, as well as research in particle size detection limits for modeled mineral inclusions.
39
CHAPTER 2
Analysis of Urinary Calculi Using an Infrared Microspectroscopic Surface Reflectance Imaging Technique
40 2.0 Overview
This investigation highlights the use of infrared microspectroscopy for the structural analysis of urinary stones. The research presented here has utilized the reflectance mode of an infrared microscope for use in generating chemically specific maps of cross-sectioned renal calculi surfaces, precisely showing the location of renal stone components in the calculus sample. The method has been applied to renal stones of both single and multiple components consisting primarily of hydroxylapatite, calcium oxalate monohydrate and calcium oxalate dihydrate. Factors discussed include the photometric accuracy of the spectra obtained, a comparison of the surface reflectance method with existing methods such as diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) and attenuated total internal reflection (ATR) analysis, and the influence of specular reflectance between polished and unpolished sample spectra. Results: Infrared images of cross-sectioned renal stones provided positive localization of components using qualitatively accurate spectra similar in appearance to DRIFTS spectra. Unlike ATR and DRIFTS spectra, surface reflectance spectra lack photometric accuracy and are therefore not quantitative at present; however, spectra are suitable for qualitative analysis. It was found that specular reflectance increases minimally with a highly polished stone cross-section surface, though qualitative data is not affected. Conclusions: Surface reflectance imaging of sections of renal stones is useful for spatially mapping the identity of stone components.
2.1 Introduction The qualitative and quantitative analysis of intact urinary calculi (stones) is challenging for several reasons, including the size and fragility of the calculi and the spatial domains of their structural configuration. Many modern methods of analysis destroy the structure of the calculus when the sample is prepared for introduction into the instrument. Maintaining the structural integrity of the renal calculi is important for the elucidation of the chemistry of formation and the etiology of the calculi in the urinary system. Understanding the etiology of urinary stones requires analysis not only of their composition, but also of the architectural arrangement of minerals and organic materials
41 in each stone. Architectural information provides important indicators of the mechanisms of initial formation and subsequent growth of the stone. [102-104] Recent studies have clearly shown that there is more than one pathological pattern for the nucleation of calcium based stones,[11] and the three-dimensional architecture of early stones is important for understanding these different processes. [105] Twenty-four hour urine tests, often used to determine and track urine chemistry in a patient, are helpful for the determination and treatment of urine/renal system ailments being suffered from at the time of collection. [106-108] However, 24-hour urine tests do not indicate past urine chemistry, and are thus unreliable in determining the chemical etiology or architecture of the renal stones being produced. In order for 24-hour urine chemistries to be applied to the long-term determination of renal stone formation, urine chemistry would have to be collected daily over a period of months or years in order to track the changes occurring in the renal system. Additionally, the practitioner would have to possess preemptive knowledge that a renal stone would be formed for that particular patient within the time constraints the 24-hour urine test was being undertaken. Taking these facts into consideration, the study of the architecture of renal stones collected from idiopathic calcium stone formers remains the best method for determining the chemistry of formation occurring in the renal system. Some of the best work on the architecture of stones has utilized thin sections of stones, which are studied by x-ray or optical methods. [109, 110] Such studies have elegantly shown the nature of the progressive addition of layers to stones, and have also attempted to identify the nucleus, or initial nidus of the stones. Related work has also been done using microdissection with analysis of samples by infrared (IR) spectroscopic methods. [104] Additionally, several reports have been published on the comparison of infrared techniques to wet chemical methods for renal stone and other biological analyses, though these can be somewhat outdated. [3, 111, 112] Recent infrared imaging studies of biological samples have focused on isolated areas of calcified tissue and bone sections. The majority of these studies have employed transmission methods of analysis which require the sample to be present as thin sections approximately 6 micrometers thick. [113-117] Several of these reports study thin sections of bone, which are structurally more robust than thin sections of urinary calculi. [117,
42 118] Thin sections of reproducible thickness are difficult to obtain with urinary stones because of the fragile nature of the material. [109, 110] In a report previously published by this laboratory, [43] we demonstrated the feasibility of using reflection/absorption infrared (RAIR) and attenuated total internal reflection (ATR) microspectroscopy as a technique for the qualitative analysis of tissue sections containing embedded mineral deposits. In addition to the qualitative analysis of these mineral deposits, reflectance infrared microspectroscopic techniques can also be used to determine the composition of cross-sectioned urinary calculi—more durable than thin sections—which is the topic of this report. Reflectance spectroscopy includes both specular and diffuse measurements; however, each method requires that the sample have specific properties. For specular reflectance, the surface of the sample must be similar to that of a dielectric material (e.g. plastics or porcelain). [65] The spectra collected typically exhibit absorptions resembling the first-derivative of a normal absorption band. These spectra can be corrected for such asymmetric features using the well known Kramers-Kronig transformation. Thus a specular reflectance spectrum can be transformed into an apparent transmission spectrum. [119] Although specular reflectance infrared microspectroscopy could be employed, this method also requires significant sample preparation and specific sample properties. [120] Diffuse reflectance spectroscopy requires that the sample be a highly scattering powder with individual particle sizes that approach the wavelength of radiation being employed in the analysis. Radiation enters the sample and re-emerges after traveling a complex path through the sample. During the residence time in the sample, the radiant energy experiences a myriad of phenomena which can include diffraction, reflection, refraction, scattering, and absorption. [66] Theoretically, as the particle size increases relative to the wavelength of light, the sample surface appears more mirror-like, increasing the amount of specular reflection. Currently, DRIFTS is widely used in urological analysis. Unfortunately, this method requires the sample to be ground to a fine powder and diluted with potassium bromide thereby destroying the stone. Although DRIFTS can yield qualitative and quantitative results, the preparation of the calculi sample is time consuming and difficult. For the analysis of a renal stone, each individual layer must be physically separated from
43 the main body of the stone. This method has the potential to introduce significant error since the layers are often not uniform with respect to composition or their coverage of the calculus and can often be quite thin. [109] With these considerations in mind, this chapter presents a facile method of analysis for isolated cross-sectioned urinary calculi, either unmounted or stabilized in resin. A cross-sectioned stone, regardless of thickness, can be analyzed using the reflectance mode of an infrared microscope, yielding spectra that are qualitatively accurate and are similar in appearance to DRIFTS spectra. Surface reflectance techniques are strictly qualitative at this time due to a lack of photometric accuracy. As will be discussed in detail, the path length of the infrared radiation using the surface reflectance method is much shorter, generating a lower absorbance and non-photometrically accurate spectra. In contrast, ATR and DRIFTS spectra are photometrically accurate, allowing quantitative data to be obtained. Though surface reflectance is not as quantitative as ATR methods, this chapter will show that surface reflectance data is much more reproducible than ATR. In addition to the qualitative infrared microspectroscopic analysis of urinary calculi components in cross-sectioned renal stones, this chapter also presents studies on the effects of surface roughness on infrared reflectance spectra and the general reproducibility of the method. This chapter differs from previous work performed by others in that the structural integrity of the renal stone section is maintained. Using the surface reflectance method presented here, elucidation of the stone components present in the stone cross section is performed in a manner conducive to subsequent analysis by either infrared, or other techniques such as Raman spectroscopy or x-ray fluorescence (XRF). Counter to the common method of infrared analysis using DRIFTS techniques [67, 70, 121, 122], the method of surface analysis contains fewer preparation steps and less time, while providing more detail as to the general structure of the renal stone.
2.2 Materials and Instrumentation Intact human urinary calculi were obtained as discards from a stone analysis laboratory, and were embedded in methyl methacrylate. Calculi cross-sections 1-2 mm
44 thick were cut using a diamond wire saw (Well Saw, Delaware Diamond Knives, DE) and were mounted onto low-E glass slides for subsequent analysis. Rabbit femur cross-sections were obtained as discards from appropriately approved animal work performed by other laboratories. Serial cross sections were alternately embedded in methyl methacrylate or left unmounted. Both embedded and loose bone sections were polished using a series of silicon carbide abrasive paper possessing a grit size from 16 μm to 0.05 μm. This procedure yielded samples ranging from unpolished wire-cut surfaces to highly polished surfaces. The DRIFTS spectra of ground reference materials presented in this chapter were collected using the Perkin-Elmer Spectrum Spotlight 300 infrared imaging microscope equipped with a single point HgCdTe (MCT) detector, and represent the average of 32 individual scans collected at a spectral resolution of 4 cm-1. A 25 X 25 μm confocal aperture was employed to isolate the sample region of interest for both the DRIFTS and reflectance experiments when single point detection was used. Background spectra were collected using ground KBr as the reference material. Images of sectioned calculi and bone were collected on the Perkin-Elmer Spectrum Spotlight 300 infrared imaging microscope using the linear array MCT detector. Spectra collected and extracted from false-color images using this detector represent the average of 8 scans at a spectral resolution of 8 cm-1. Each spatial element on this detector represents 6.25 x 6.25 μm on the sample. Background spectra were collected using the reflective side of a low-E slide. ATR spectra were collected on the Perkin-Elmer Spectrum Spotlight 300 infrared imaging microscope using the drop-down Ge internal reflection element (IRE) and the single point MCT detector. Spectra collected represent the average of 32 scans with a spectral resolution of 4 cm-1. Background spectra were collected using a KCl pellet as the reference material.
2.3 Results and Discussion A renal stone sample, encased in resin and sliced to a thickness of several millimeters, is utilized by medical researchers due to the fact that this sample configuration retains its physical features for analysis. By studying the radial projection
45 of cross sectioned urinary calculi, it is possible to determine the chemistry from nucleation to some arbitrary point in time. The nucleation and growth process of urinary calculi are thought to be similar to those of either a pearl-like growth mechanism, or geode chemistry, as explained in Chapter 1. Infrared imaging of cross sectioned urinary calculi allows one to precisely obtain molecular information as a function of position and therefore time enabling one to determine the chemistry of calculi formation. 2.3.1 Surface Reflectance
Due to their importance and relative abundance, renal stones containing calcium oxalate and/or hydroxylapatite are the focus of this laboratory. Figure 2.1 illustrates a 6.5 x 9.0 mm cross-sectioned calculus initially diagnosed to be composed solely of hydroxylapatite. The physical features of the stone include the radial distribution of the component layers in irregular patterns and varying colors. The varying colors in the radial distribution are not necessarily indicative of component differences, though quite often this is the case. [123] The outlined box in Figure 2.1 contains several bands of varying color as well as a portion of an innermost region. This sampled area of the renal stone is representative of a radial projection of the entire stone.
Upon analysis, it was determined that the nucleus consists of hydroxylapatite. The calcium oxalate monohydrate mixed with oxalate dihydrate encircles the hydroxylapatite center. The alternating layers of different components form concentric rings, ending in an outer shell of calcium oxalate.
46 Figure 2.1: Visible image of a cross-sectioned renal stone containing radial projections of differing components. The area of interest is outlined by a box containing a varied sampling of the components present in the cross-section.
47 Figure 2.2 illustrates false color infrared image of the selected area, generated from ~130,000 spectra collected over a nearly 7 hour time frame. The spectra were collected using a linear array detector with a pixel size corresponding to a 6.25 μm by 6.25 μm spatial element on the sample. Background spectra were collected every 5 points to maintain good atmospheric compensation during the course of the nearly seven hour analysis period. To reduce the time period required for analysis, spectral resolution can be lowered slightly, or the area selected for analysis diminished in size. The slight lowering of the spectral resolution is acceptable due to the purely qualitative nature of the method.
48 Figure 2.2: Infrared image of the outlined area in Figure 2.1 and associated spectra of a cross-sectioned renal calculus. Top: hydroxylapatite. Bottom: calcium oxalate. Blue/green indicates the presence of the component, while red/yellow indicates a lack of the specified component.
%R
%R
49 Each spatial element in the image is associated with a complete infrared spectrum ranging from 4000 to 720 cm-1 that can be viewed separately or combined as a molecular (component) specific image. For example, the spectra illustrated in Figure 2.2 were used to generate false color images using the total absorbance of specified spectral features. The top panel shows a spectrum characteristic for hydroxylapatite; the main feature -1 located near 1020 cm is due to the asymmetric stretch (vas ) of the orthophosphate group in hydroxylapatite.
The false color image in the top panel shows the component specific image obtained using the integrated band intensity near 1020 cm-1. Areas with a greater abundance of hydroxylapatite are indicated by the blue/green colors corresponding to a higher integrated band intensity; areas shaded red/yellow indicate the absence of hydroxylapatite and/or the presence of another material. The image demonstrates that the hydroxylapatite core of this stone does not contain calcium oxalate monohydrate or dihydrate, however, several other layers of mixed oxalates are present.
Notice also in the top image of Figure 2.2, there exists a vein through the center of the hydroxylapatite core. This vein, upon examination of the spectra, consists of hydroxylapatite, though the spectrum looks slightly different as can be seen in Figure 2.3. The inner vein, denoted by an arrow, produces a hydroxylapatite band much lower in intensity and broader than that of the rest of the core (denoted by a star).
50 Figure 2.3: Inner hydroxylapatite vein spectrum (solid—denoted by arrow) compared to regular hydroxylapatite spectrum (dashed—denoted by star). Molecular image is a magnified view of the core from Figure 2.2.
*
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 cm-1
51 Though the inner core vein consists of hydroxylapatite just as the rest of the core, this vein is the true nucleus and is indicative of the chemistry occurring in the renal system at the time of renal stone formation. It can be seen from Figure 2.3 that the vein has a lower intensity of hydroxylapatite band than the bulk of the rest of the core. The spectrum in the vein of the stone may be indicative of either disordered hydroxylapatite crystals, or of a previously proteinacious material that has been destroyed during the dehydration process undergone by the stone during embedding. Future studies will focus on samples such as that exhibited in Figure 2.1, attempting to isolate, analyze, and explain the nucleation process for stones not originating from a Randall’s plaque process such as suggested in Figure 1.1.
The composition in the outer extremity of the stone can be determined by extracting a spectrum from this region, and was found to correspond to calcium oxalate.
The lower panel of Figure 2.2 illustrates a spectrum characteristic of calcium oxalate. Features observed in this spectrum are due to the asymmetric stretch of the oxalate anion -1 (vas) located near 1624 cm , the symmetric stretch of the oxalate anion (vs) C=O stretch at 1321 cm-1 and the C-O bend located near 785 cm-1. A map of the calcium oxalate content of this section is shown in the right of the lower panel, based on the integrated intensity of the peak near 1624 cm-1. Again, blue/green indicates a high prevalence of the calcium oxalate material, while red/yellow indicates its absence.
A closer inspection of the calcium oxalate spectra reveals that two different hydrates are present in this stone. These two hydrates consist of calcium oxalate monohydrate (COM) and calcium oxalate dihydrate (COD). The two materials are readily distinguished by the position of the asymmetric oxalate anion stretch. For calcium oxalate monohydrate (COM) this band is located at 1624 cm-1, and for the dihydrate (COD), it is located near 1680 cm-1. The 1321 cm-1 and the 785 cm-1 bands remain the same for both oxalate species. Figure 2.4 displays spectral image maps of COM (top panel) and COD (bottom panel) for this region of the stone section, indicating the variability in hydrate content in different layers of the stone. Most importantly, these differences are detected using the 56 cm-1 wavenumber difference between COM and COD and demonstrate the chemical specificity of infrared spectroscopy. The contrast
52 detail in the images is surprising in light of the fact that they are based on integrated peak areas alone. Better contrast should be achievable in future studies using a band ratio method, which will make the contrast between calcium oxalate monohydrate and calcium oxalate dihydrate much higher. Blue/green areas designate a higher absorbance of the selected compound.
53 Figure 2.4: Infrared images and associated spectra of the two hydrates of calcium oxalate. Top: Calcium oxalate monohydrate. Bottom: Calcium oxalate dihydrate. Blue/green areas designate regions of high absorbance by the designated compound.
%R
%R
54 The compositional complexity of the stone section shown in Figures 2.1-2.4 is a common finding in renal stones,[104, 109] and the general lack of recognition of such complexity is a possible reason for the generally poor performance of stone analysis facilities. A standardized test system in the United Kingdom has shown an average error rate of 30-40% in standard samples analyzed by commercial laboratories.[48] Looking at the stone composition based on molecular information presented in Figures 2.2-2.4, it can be seen how the portion of the stone taken for analysis could easily yield a result of hydroxylapatite or calcium oxalate monohydrate or dihydrate, dependant upon the portion analyzed. Or, a single sampling could yield a result that would indicate a mixture of all three minerals. Therefore it is imperative that multiple sites be analyzed throughout the sample stone, providing a full understanding of the stones chemistry.
It is also easy to see from the molecular images in Figures 2.2-2.4 that the evolution of renal stones is much more complex than the outward appearance would suggest. Several phases of exposure to differing urinary tract environments over an extended period of time generate the potential to produce a complex layering system. Each layer fully obscures the material underneath, generating a specific need for renal stones to be analyzed in cross sections as opposed to whole samples.
The molecular images in Figures 2.2-2.4 clearly show the distribution of calcium oxalate and hydroxylapatite components within the renal stone. The mineralogical chemistry of the patient appears to have been modulated from hydroxylapatite to oxalate and back again as the stone grew, generating the ringed layers so clearly seen. Future studies aim to correlate patients histories (e.g. blood and renal chemistry, dietary intake) to these alternating structures in an attempt to better define the chemistry of stone formation.
2.3.2 Polished vs. Unpolished Sample Surfaces
The composition of consecutive layers of mineral in a renal stone can be determined using IR reflectance imaging. Two important questions concerning these results are: 1. Would better sample preparation help and 2. Would another measurement mode be preferable?
55 It is known that polished surfaces, such as those of a dielectric sample, often result in greater specular reflectance than unpolished surfaces. [120, 124] However, in the case of qualitative identification, a large contribution of specular reflectance mixed into the DRIFTS-like spectra may inhibit spectral interpretation. If a spectrum is purely specular in character, a Kramers-Kronig transformation can be applied to produce a transmission-like spectrum that is easily interpretable. But since the spectra obtained using the surface reflectance technique employed here would not be purely specular in nature, Kramers-Kronig transformations would be futile. Similarly, if the spectrum were purely diffuse in character (DRIFTS), Kubelka-Monk transforms could be employed to yield quantitative data. However, since both diffuse and specular characteristics are present in surface reflectance spectra to some degree, neither set of mathematical transforms are applicable.
In an attempt to determine if the data collected would be improved with respect to specular reflectance, individual renal stones were gently hand polished using abrasive materials varying in grit size from 16μ-0.05μ. However, polishing the surface of the fragile renal calculi frequently led to their disintegration. As an alternative, bone cross sections were polished to varying degrees utilizing the same range of abrasive media in order to address this first question. Bone was chosen as an acceptable substitute since, spectroscopically, the intensity and band shape of the main hydroxylapatite band at 1020 cm-1 for bone and renal calculi were observed to be nearly identical, as shown in Figure 2.5.
56 Figure 2.5: Comparison between spectra taken from a bone cross-section (solid line) and a renal stone cross-section (dashed line). Observed and collected under identical surface reflectance conditions.
23.3
22
20
18
16
14 %R 12
10
8
6
4
2
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 cm-1
57 Figure 2.5 displays spectra corresponding to hydroxylapatite from both bone (solid) and a cross-sectioned renal stone (dashed). As can be seen, the spectra are qualitatively similar, and both spectra show low peak-to-peak noise. Spectra of bone, polished using 0.05μ grit sanding paper and embedded in resin, showed a peak-to-peak noise of 0.66 whereas that of the wire-cut stone was 0.53. Polished and unpolished bone sections were imaged using the reflectance mode of the infrared microscope in a manner identical to that of the renal calculi samples.
Spectra of highly polished (0.05 μm surface roughness) bone sections and unpolished, wire-cut bone sections are compared in Figure 2.6. The results show minimal differences between the spectra with respect to reflected intensity and peak-to-peak noise. The absorption band intensity of the polished cross section (dashed spectrum) is slightly larger than that of the unpolished section (solid spectrum) by approximately 10%. This slight intensity difference is due to the lack of scattering from the polished surface compared to the more diffuse nature of the surface from the wire-cut bone sample. The data suggest that improvements with additional polishing are negligible when considering the increased sample preparation time and the assumption that a polishing procedure could be developed for urinary calculi.
58 Figure 2.6: Comparison between highly polished (dashed line) and unpolished/wire-saw cut (solid line) bone cross-sections. Observed and collected under identical surface reflectance conditions. Polishing occurred using 0.05μm grit sand paper.
28
26
24
22
20
18
16 %R
14
12
10
8
6
4
2
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 cm-1
59 2.3.3 Method Comparisons
To answer the second question posed in section 2.3.2, which pertains to sample analysis techniques, a comparison of COM spectra obtained with ATR, surface reflectance and diffuse reflectance infrared microspectroscopy using identical conditions other than measurement mode is shown in Figure 2.8. The spectra show minimal differences with regard to both band shape and positioning, though the signal to noise ratio is much higher for DRIFTS (top) and ATR (middle) than for the surface reflectance method (bottom). Upon close inspection of the surface reflectance spectrum in Figure 2.7, it can be seen that the group of bands from 3400 to 3000 cm-1 attributed to the symmetric and asymmetric OH stretch of crystallized water, [125] and two weaker bands at 884 cm-1 and 945 cm-1 attributed to OH deformation [126] of calcium oxalate monohydrate are either absent or greatly diminished in both the surface reflectance and ATR spectra of calcium oxalate monohydrate. Additionally, the shoulder at 1375 cm-1 is also greatly reduced in the surface reflectance spectrum. If observed closely, it is observed that the shoulder is present, however, specular reflectance at the base of the 1318 cm-1 band as well as increased noise inhibits interpretation of the 1375 cm-1 band area.
60 Figure 2.7: Comparison between DRIFTS (top), ATR (middle), and surface reflectance (bottom), spectra of calcium oxalate monohydrate.
4000.0 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 cm-1
61 The absence of these bands in the surface reflectance spectrum are due to a combination of factors, including the crystalline structure of the stone but most importantly the optical characteristics (e.g. refractive index) of the material. The low absorbance observed in the surface reflectance spectrum is an indicator that the probe beam penetrates the surface only slightly resulting in a reduced optical path length. On the other hand, the high absorbance associated with the bands in the diffuse reflectance spectrum result from a greater optical path length. For diffuse reflectance, a small amount of sample is mixed with an infrared transparent sample such as potassium bromide (KBr). Due to the nature of the sample, multiple scattering events take place, which increases the optical path length for a given sample. ATR possesses an intermediate optical path length, with the optical path length increasing from the high- energy end of the spectrum to the low-energy end. At the high-energy end of the spectrum, the ATR optical path length is less than 0.25 μm, but increases to 1.4 μm toward the low-energy end. This difference in path length of over 5x is why the features located near 3200 cm-1 (higher energy end of spectrum) are greatly diminished in the ATR spectrum, but those at 945 and 884 cm-1 (low energy end of spectrum) are clearly visible.
For the present research, surface reflectance and ATR methods were utilized in a complementary method, with each possessing certain strengths. Analysis utilizing the surface reflectance microspectroscopic technique is fast and requires limited sample preparation allowing a diversity of samples to be analyzed. However, due to scattering, reflection, refraction and sample topography, quantitative information may be difficult to extract. Under most conditions, ATR produces photometrically accurate bands capable of yielding quantitative data with a high signal to noise ratio.[127] Unfortunately, this method requires sample contact with the internal reflection element (IRE). This contact could potentially damage the sample, and in the case of samples that are capable of adhering to the IRE after analysis has been completed, the IRE may need to be cleansed in between successive sample points. Therefore the process of generating a spectral map of a sample using ATR is time consuming and requires constant involvement by the operator.
62 The method most applicable and successful for sample analysis is to use the two techniques in conjunction with one another. A molecular image can quickly be collected from the sample, highlighting features of interest and determining general homogeneity using the surface reflection method of analysis. Earlier research by this laboratory has confirmed that subsequent analysis using ATR microspectroscopy yields photometrically accurate data suitable for quantitative analysis. [43]
2.3.4 Method Reproducibility
In order to ensure the accuracy of the methods utilized for this research, reproducibility was determined using both ATR and surface reflectance techniques. A specific site on a renal stone section of known composition was marked using computer software such that each measurement was collected from precisely the same site. This site was re-visited ten times, with backgrounds taken between each scan. Both ATR and surface reflectance used 16 scans and 4 cm-1 resolution. The integrated band intensity from 1300-1345 cm-1 range of calcium oxalate monohydrate was utilized. The apertures for both methods, however, differed slightly. Due to the 4x magnification of the ATR IRE, a 100x100 μm aperture was used for ATR analysis, which translates into 25x25 μm on the surface of the sample. For surface reflectance, a 25x25 μm aperture was used, translating to 25x25 μm on the surface.
The results conclude that the surface reflectance data is more reproducible than ATR with standard deviations in the integrated band intensity of 0.02 and 0.97, respectively. This lower level of reproducibility in the ATR data may be due to the slight indentation generated in the sample when the IRE is placed into sampling position. The surface of the sample becomes slightly imprinted, making it difficult to take subsequent data from the exact point as the previous site of analysis. Table 2.1 displays data with respect to the measurements obtained.
63 Table 2.1: Reproducibility data from both ATR and surface reflectance techniques under the same conditions for the same stone. Aperture: 100 μm ATR (translates to 25 μm on the sample surface) and 25 μm surface reflectance; scans: 16; resolution: 4 cm-1. Solid contact was achieved in all ATR data.
64 Looking at the data in Table 2.1, it would appear that either the value of 0.66 or 3.55 could possibly be an outlier for the ATR reproducibility data. Using the Q test on
the 0.66 point, Qexp=0.147, while Qcrit for 90% confidence is 0.437. Therefore, 0.66 is
not an outlier. For the point 3.55, Qexp=0.111, while Qcrit=0.437. Therefore, 3.55 is also a valid data point. There are no outliers in either data set.
In order to compare the data directly, some slight calculations were performed. Surface reflectance is measured in percent reflectance (%R), while ATR is measured in absorbance (A). In order to directly compare the absorbance of ATR with surface reflectance, the standard equation
A=-log T or A=-log R Equation 2.1
must be used. Therefore, each surface reflectance spectrum was divided by 100 in order to convert from %R to R. The integrated band intensity from 1300-1345 cm-1 was taken and subsequently used in Equation 2.1 in order to determine the corresponding absorbance.
The results in Table 2.1 indicate that ATR has 7.13 times the absorbance area than surface reflectance for the same wavenumber range. These results correspond to the assertion in Figure 2.8 that ATR spectra contain a larger absorbance and are photometrically accurate, while surface reflectance has too low of an absorbance to be photometrically accurate. The higher uncertainty associated with the ATR measurements is most likely due to the small imprint created on the surface of the sample by the IRE after each measurement. Though each measurement was attempted at the same site on the sample, the placement of the IRE most likely differed by a few micrometers and came into contact with the small indent created by the previous measurement. This slight alteration in sample morphology is most likely responsible for somewhat skewing the ATR results.
65 2.4 Conclusions This chapter has presented a facile method for qualitative spatial analysis of the mineral layers in a section of urinary stone using instrumentation and software that are easy to use and readily available. Using surface reflectance methods, layer-by-layer analysis of concentric rings of renal stone material can be analyzed in order to produce molecular specific images of high quality and high detail. Not only can components varying greatly in their composition be determined, but components such as calcium oxalate monohydrate and dihydrate can be differentiated both visually and spectrally as well. The method presented in this research is less susceptible to error or misdiagnosis than common methods of stone analysis, and bypasses the thin sectioning and/or grinding steps that have been required for analysis of urinary calculi in the past. Not only has this chapter provided evidence that an unpolished sample yields similar information to that of polished sections, but this chapter has also provided evidence that both embedded and loose samples provide similar data. These findings further support the assertion that no extensive sample preparation is needed to obtain high quality spectral images using the infrared microspectroscopic reflection technique presented in this research. Unfortunately, the above research is not quantitatively valid at this time due to the lack of photometrically accurate absorption bands. It may be a regretful fact that surface reflectance proves at this point to be solely qualitative in nature without the aid of subsequent ATR analysis. Future research in our laboratory is focusing on quantitative methods of cross-sectioned stone analysis utilizing off-axis ATR methods that should yield better photometric accuracy. This chapter has been accepted for publication and will be published in 2007 in Urological Research under the title Analysis of Urinary Calculi Using an Infrared Microspectroscopic Surface Reflectance Imaging Technique with the authors in the following order: Jennifer C. Anderson, James C. Williams, Jr., Andrew P. Evan, Keith W. Condon, and André J. Sommer.
66
CHAPTER 3
Determination of the Detection Limit of Dispersed Calcium Oxalate Monohydrate in a Protein Matrix and Band Fitting of the 1618 cm-1 Oxalate Band
67 3.0 Introduction The focus of this chapter is the determination of detection limit (DL) for calcium oxalate monohydrate (COM) dispersed in a protein matrix. This work differs from similar publications in that the mixtures are produced from readily available components, as well as analyzed using a single bounce infrared ATR system. This detection is performed in order to monitor dispersed renal stone constituent levels as they increase or decrease across regions of the kidney. Similar studies monitoring concentrations of a designated analyte in kidney tissue have been successfully performed for cancerous samples [128, 129], though COM was not a component of interest. This research is important because high concentrations of renal stone components dispersed in papillary renal tissue may gradually evolve into plaques and embedded mineralized deposits. [45] It is from these mineralized deposits that a stone can nucleate, grow, and detach, leaving the crystallized site of nucleation behind for the nucleation of future renal stones. [130] COM is one of the most prevalent components in renal stones. [8, 34, 131, 132] The quantitative measurement of COM via infrared (IR) methods has been previously investigated. Volmer, using partial least squares (PLS) regression and transmission techniques, successfully quantified COM down to 10 wt% with an uncertainty of 8.8%RSD. [133] However, no detection limit was provided. A study by Maurice-Estepa has investigated COM in the presence of other renal stone components using a first- derivative calculation method referred to as IR Zero-Point Crossing, quantifying COM down to 40 wt%. [132] Unfortunately, only the mention of a calibration curve was included in the report, stating that the calibration curve was made of mixtures from 10-90 wt% COM; no information was provided on detection limits of the method other than a ±7.5 wt% reproducibility. Cohen-Solal et.al. have also published on the quantitative determination of renal stone components, focusing on calcium oxalate dihydrate (COD) and carbapatite. [134] Using a mathematically generated calibration line from four initial data points, Cohen-Solal et.al. were able to quantitatively measure COD and carbapatite down to 1 wt% in some samples. However, as stated earlier, this calibration curve was generated from only 4 initial data points, not from a series of measured standards. The same report stated that the technique was applied to COM, providing a LOQ of ≥10 wt%.
68 The sample preparation for DRIFTS analysis undertaken for these reports involved an extensive amount of time, requiring first visual analysis to determine both the type of material present and the thickness of the individual layers, followed by grinding of the stone and mixing with KBr. An ATR method is much faster, requiring minimal sample preparation and preserving the integrity of the stone structure for future analysis. Using ATR, the integrated band intensity of the sample is not dependant upon sample
thickness if the thickness is larger than dp of the evanescent wave, so thick cross-sections of stone can be analyzed without extensive preparation. Additionally, ATR has a magnification factor associated with the refractive index of the IRE, allowing for better spatial discrimination than other infrared techniques. In general, literature pertaining to the infrared detection limit or limit of quantification of COM is difficult to locate, especially IR COM concentrations in the presence of protein. However, it is probable that research similar to that presented here has taken place among other research groups but has remained unpublished. Differing concentrations of renal stone components in epithelial tissue indicate different and changing chemical environments in the renal system. An ability to accurately quantify stone material may lead to insight into the chemical processes occurring. It is therefore beneficial to measure COM in the presence of protein in order to imitate a realistic setting. Since the environment of stone nucleation is protein, the spectral absorptions arising from the protein must be accounted for. In order to successfully model tissue, proteinacious material (gelatin) was mixed with powdered COM in varying weight percent ratios. ATR-IR was employed to generate calibration curves relating integrated band intensity for a given component to its concentration. Additionally, band fitting techniques are applied to theoretical and experimental spectra of COM mixed with protein. Due to the close proximity of the amide I -1 absorption of protein at 1633 cm to the C=O vas oxalate ion absorption of COM at 1618 cm-1, there can be great difficulty in determining the integrated band area of COM. Previous studies have successfully employed band fitting techniques for oxalate determination, though the other components were mineral in character as opposed to organic. [11] Axe found, using infrared spectra, band fitting techniques could not appropriately determine the area of the oxalate absorption in the presence of a competing
69 component. In the present research, band fitting is investigated for both theoretical and experimental data in order to determine a limit of detection of COM in the presence of protein.
3.1 Experimental 3.1.1 Materials and Instrumentation COM (Acros Organics) was manually ground prior to analysis in order to produce uniform particle sizes (ground particle size d=150-200 µm). Protein in the form of unflavored gelatin powder (Kroger Company, Ohio) was used in both dried form for weight % ratios and in gelatin form prepared according to package directions. The mixtures were analyzed using the Harrick Split Pea ATR accessory interfaced to a Perkin Elmer Spectrum 2000 infrared spectrometer. Individual spectra collected represent the average of 32 scans collected at a spectral resolution of 4 cm-1. A load of 0.5 kg was used to ensure good contact between the IRE and the sample. Five sites from each sample were analyzed and averaged into a single spectrum for quantitative purposes. 3.1.2 Sample Preparation Several different types of sample preparation were investigated in order to determine which method yielded the most uniform samples. 3.1.2.1 Calcium Oxalate Monohydrate in a Gelatinous Protein Matrix Protein (gelatin) was prepared according to package directions. Dried protein was dissolved into ~400 mL of boiling water. After complete dissolution of the protein, the appropriate amount of COM was added to the liquid in order to make the ratios 3, 5, and 10 wt% COM. The mixture was then stirred continuously and allowed to cool to room temperature. Once the liquid mixture was at room temperature, it was transferred to a refrigerator and allowed to set overnight. Unfortunately, once stirring ceased, undissolved COM settled to the bottom of the container. This method was not used since it produced heterogeneous mixtures. 3.1.2.2 Calcium Oxalate Monohydrate in a Thin Film of Gelatin The protein/COM was prepared as before. When the mixture reached room temperature, droplets were pipetted onto a glass substrate and allowed to cool completely.
70 The resulting films were stored in a desiccator and allowed to dry. Unfortunately this method also produced heterogeneous samples. 3.1.2.3 Calcium Oxalate Monohydrate in High Viscosity Protein Dried protein and COM were combined in a weighing dish such that a 5 wt% COM to protein ratio was produced. A minimum amount of water was added such that a viscous mixture was formed. This mixture was then stirred thoroughly in order to achieve even distribution of the components and was subsequently placed in a desiccator and allowed to dry. Average drying time was 4 days in order to obtain equal dryness over the entire sample. It was found using IR-ATR measurements from different points on the sample that the sample was heterogeneous with respect to COM concentration. Additionally, the amount of time needed in order to create the sample was deemed too long. 3.1.2.4 Calcium Oxalate Monohydrate in a Protein Slurry Dried protein and COM were combined in a weighing dish in weight percent ratios from 1-99 wt% COM. A minimum amount of water was added such that thick pastes were formed. These pastes were then mixed in order to achieve even distribution of the components and spread upon glass plates in order to generate thick films. This sample configuration yielded thick films in which the distribution of COM particles visually appeared to be homogenous. Homogeneity was tested by determining the integrated band intensities from five different sites on a 70 wt% COM sample. Measurements were taken over the 1300-1345 cm-1 range, providing integrated band intensities of 0.0720, 0.2250, 0.1575, 0.1820, and 0.1806. Observing the values, the 0.0720 would appear to be an outlier. Using the Q-test, which is used pervasively in the
literature for the evaluation of biological results, [136-139] Qcrit=0.642 at the 90%
confidence limit, and Qexp=0.710. Therefore, only 4 of the 5 measurements could be used to determine an average corrected area of 0.19±0.02. Using a Student t-test [58] inclusive of all five values, an average of 0.16±0.03 was determined at the 90% confidence limit (α=0.10). Due to both the ease of creation and relative homogeneity of this sample configuration, this method was utilized for all future sample preparations.
71 3.2 Results and Discussion 3.2.1 COM/Protein Mixtures Because the environment of the kidney changes over time, a measurement of renal stone materials and their dispersion in the renal system may aid in the determination of chemical etiology. However, in order to accurately determine mineral concentration, a method of measuring the mineral in the presence of protein is necessary. Therefore, for the ideal analysis of mineral deposit samples or their models, it is necessary to analyze the tissue matrix simultaneously with the component of interest. It is possible that finely dispersed renal stone components play an important role in the overall etiology of renal stone and mineral inclusion formation. Therefore it is of interest to attempt to quantify the concentrations of these components in tissue prior to, during, and after the formation of larger embedded mineral species. The distribution of COM particles in a protein matrix attempts to model tissue containing finely dispersed COM in the above situations. The correlation between component concentration and peak area is ideally linear and obeys Beers Law, which is given in Equation 3.1 below.
A=εbc Equation 3.1
Here, A is the absorbance, ε is the extinction coefficient of a pure solid, b is the path length of radiation through the sample, and c is the concentration of the analyte. [58] The path length of radiation averages 1 μm, with an effective sampling depth [140] of approximately 3 μm due to the evanescent wave from the IRE, and ε=1.49 (obtained from literature [141, 142] ) for COM. Therefore, the concentration and the absorbance are the only two variables of interest if COM and protein do not interact chemically. Certain assumptions are made when using Beers Law; Beers Law assumes that the path length of radiation is not changing, and that each component interacts with the absorbed radiation independently from the other components present. [58, 85] Realistically, this is not the case since the path length of the radiation of the evanescent wave will change with wavelength when using an ATR technique (Equation 1.5), and the absorptions of protein overlap slightly with those of COM for some peaks. [45, 143]
72 Others have observed a linear relationship between COM absorption and concentration in solid mixtures using infrared spectroscopy. Salvadori quantified COM and Gypsum from mural paintings using an external standard, and was able to quantitate COM at 0.4 wt%. [22] A detection limit (DL) or MIQ was not mentioned. Maurice- Estepa quantitatively measured COM in the presence of calcium oxalate dihydrate (COD) from renal stone samples using first-derivative calculations, however, 40 wt% COM was the smallest amount measured. [7] No mention of a MIQ or DL was made. Cohen-Solal has quantitatively measured COD in the presence of carbapatite, generating a quadratic calibration curve from four sample points. [10] This technique was carried over to include COM, as well as other renal stone components. Cohen-Solal was able to quantitate COM at ≥10 wt%. In order to determine a best case scenario for the ATR detection limit of COM in protein, a theoretical calibration curve was generated. The theoretical spectra were generated by combining the pure COM spectrum with a pure protein spectrum in varying ratios as described below: (0.X)*(neat COM spectrum)+(0.Y)*(neat protein spectrum)=theoretical spectrum Where (0.X)+(0.Y)=1.0 The resultant calculated DL represents the optimal conditions and parameters for the set resolution (4 cm-1) and measurement time (32 scans taking approximately 1.5 minutes). One major spectral interference occurs due to the close proximity of the 1633cm-1 -1 amide I absorption of protein to the 1618 cm vas C=O stretching absorption of oxalate. -1 -1 Therefore, the O-C-O bend of oxalate at 778 cm , and the vs C=O stretch at 1318 cm are used for most quantitative measurements in this chapter. Figure 3.1 displays several theoretical spectra of varying concentrations of COM in protein, with special emphasis placed on the 778 cm-1 and the 1318 cm-1 absorptions. In order to determine the theoretical DL, 3x the root-mean-square (RMS) noise (0.0014) is divided by the slope of the calibration curve. Using this criteria, a DL of 0.10±0.02 wt% COM is determined. A similar approach using the 1318 cm-1 absorption of oxalate gave a DL of 0.040±0.006 wt% COM. All R2 values equal 1, which is expected since the spectra were mathematically generated.
73 Figure 3.1: Theoretical spectra of COM in a protein matrix. Here, the black spectrum is pure protein, red is 1 wt% COM, purple is 3 wt% COM, and blue is 5 wt% COM. Window A exhibits the entire spectral range from 720-4000 cm-1. Windows B and C display magnified examples of the linear increase in area for the 778 cm-1 and 1318 cm-1 absorptions, respectively.
0.160
0.15 C 0.14
0.13
0.12 A
0.11
0.10
A 0.09 1355 1345 1335 1325 1315 1305 1295 1283 cm-1 0.08
0.07 B
0.06
0.05
0.04
0.030 4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 600 cm-1
800 796 792 788 784 780 776 772 768 764 760 cm-1
2.5 Theoretical Calibration Curves
-1 2 770-790 cm y = 0.04x + 0.0961 1300-1345 cm-1
1.5
1 y = 0.0145x + 0.0079 Integrated PeakIntensity 0.5
0 0 102030405060
% COM
74 An experimental value was subsequently determined using dried mixtures of protein and COM in ratios ranging from 1 to 75 wt% COM. An upper limit of 75 wt% COM was used since, above this concentration, loose COM was found statically adhered to the surface of the sample instead of being intrinsically incorporated into the sample, making the true concentration uncertain. Figure 3.2 illustrates the correlation between the 778 cm-1 and 1318 cm-1 peak areas of COM and the increasing COM concentration.
75 Figure 3.2: Experimental data showing an overall linear relationship between the growth of the 770-790 cm-1 and 1300-1345 cm-1 absorptions in relation to increasing COM concentration. Standard deviation ranges from 12-66% RSD and 19-68% RSD for the 770-790 and 1300-1345 cm-1 regions, respectively.
Experimental Data
770-790 cm-1 range 1300-1345 cm-1 range 2
1.5
y = 0.0172x + 0.1191 2 1 R = 0.9269
0.5 y = 0.0074x + 0.0381
Integrated BandIntensity Integrated 2 R = 0.9287 0 01020304050607080
% COM in Protein -0.5
76 Three times the RMS noise divided by the slope in Figure 3.2 determined an ATR DL of 0.20±0.03 wt% COM for the 778 cm-1 absorption area. The experimental value is slightly larger than the theoretical DL value of 0.10±0.02 wt % COM. This data implies that tissue containing small quantities of COM can be semi-quantitatively analyzed at best. Using these same principles, a DL of 0.09±0.02 wt% COM was determined for the 1318cm-1 absorption, which is also larger than the theoretical value of 0.040±0.006 wt% obtained from theoretical calculations. Uncertainty in the experimental DL was determined by using the propagation of uncertainty equation (Equation 3.2) below; [145]
s 1 1 (y − y') 2 s = y + + Equation 3.2 x m k n m 2Σ(x − x') 2
sx is the uncertainty in the experimental DL, sy is the standard deviation in the integrated
band intensity over the range (sy770-790=0.14, sy1300-1345=0.35), k is the number of measurements in the unknown (k=5), n is the number of data points total over the 0-100 wt % range (n=15), m is the slope of the line (778 cm-1, m=0.0074; 1318 cm-1, m=0.0172), y and y’ are the true value and averaged total value for the integrated band intensity (0.0468 and 0.3039 for 778 cm-1 and 0.0960 and 0.734 for 1318 cm-1, respectively), and x and x’ are the true value and averaged total value for the data point(s) (0.1 and 35.4 for 778 cm-1 and 0.04 and 35.4 for 1318 cm-1, respectively). Using the above data, an uncertainty of 25% RSD was determined for the 778 cm-1 DL, meaning the DL is 0.20±0.03 wt%. Uncertainty for the 1318 cm-1 DL is 23% RSD, yielding a DL of 0.09±0.02 wt%. These inconsistencies and poor standard deviations are due to a lack of overall homogeneity in the sample. However, the R2 values show a reasonable fit to a linear line (0.9287 and 0.9269 for the 778 and 1318 cm-1 regions, respectively), indicating that although sample precision may be lacking (large uncertainty), the average of the data set for each point is accurate with respect to following Beers Law (Equation 1.1). The homogeneity of the pellets was determined by testing an arbitrary sample at 10 different sites. Using the 35 wt% COM sample, the relative standard deviation of the 10
77 measurements was found to be 54% and 61% for the 778 and 1318 cm-1 absorptions, respectively. Reproducibility is discussed in-depth in section 3.2.3. 3.2.2 Band Fitting of the 1618 cm-1 Calcium Oxalate Monohydrate Absorption Because the amide I absorption of protein lies in close proximity to the 1618 cm-1 absorption of the oxalate ion, it is difficult to easily determine integrated peak intensity of the 1618 cm-1 absorption of COM. In infrared spectroscopy, spectral absorptions often overlap to form complicated and unresolved spectra. For these situations, a technique known as band fitting is applied in order to better quantify specific peaks within the region of interest. Using Gaussian curve-fitting software in addition to an approximate location and full-width at half the maximum height (FWHM) of the absorptions, general positions and band contributions can be determined. Previous work by Axe has also included band fitting to determine the area of oxalate absorptions, though oxalate was determined in the presence of other mineral species, not protein. [135] Band fitting requires the investigator to know an approximate location of the peak center and approximate FWHM for the peak under analysis. Because the sample is in a condensed phase, the measured band width at the sample FWHM will be larger than that set by the instrument resolution. [146] In this research, both the amide I and COM peak centers and FWHM are known with some certainty from reference spectra of the neat materials. Once the peaks are entered into the software, the software uses set algorithms to determine the peak shapes and areas required to produce the final spectrum. Figures 3.3 and 3.4 illustrate the output from the GRAMs (a spectral program aiding in the determination of peak area, FWHM, center, etc.) spectral program (Galactic) used to determine the band areas of protein and COM.
78 Figure 3.3: GRAMs band fitting output of 5 wt% COM in protein. Below, the fitted spectrum (red) is nearly identical to the experimental spectrum (blue). The dotted peaks indicate the estimated absorption bands of the components (COM and protein), while the highlighted purple peak is the selected peak for which information is supplied in the upper right corner. The green trace on the bottom, near the baseline, is the residual difference between the original spectrum and the fit result. The information provided includes the band area, peak center, height, and FWHM (width column).
Red: original spectrum
Blue: fitted spectrum
Dotted: program- estimated peaks
Purple: peak for which data is provided
79 Figure 3.4: GRAMs band fitting output of 70 wt% COM in protein. Below, the fitted spectrum (red) is nearly identical to the experimental spectrum (blue). The dotted peaks indicate the estimated absorption bands of the components (COM and protein), while the highlighted purple peak is the selected peak for which information is supplied in the upper right corner. The green trace on the bottom, near the baseline, is the residual difference between the original spectrum and the fit result. The information provided includes the band area, peak center, height, and FWHM (width column).
80 A measure of the “goodness of fit”, χ2, provides a relative guide as to whether the fitted placement of peak center positions and widths create a good approximation to the 2 actual spectrum. A large χ implies that the mathematically determined peaks are in poor agreement with the actual spectrum, while a small χ2 implies good agreement. The goal is to minimize χ2. By fitting all 5 suspected bands (as in Figures 3.3-3.4), the value χ2 is 3.193 for the 5 wt% COM, while χ2=74.484 if only 4 peaks are fitted. For the 70 wt% COM, the χ2 values are 7.712 and 362.398 for 5 and 4 peaks, respectively. Therefore it was determined that the spectrum is more accurately reproduced using 5 theoretical peaks as opposed to 4. Since fitting more peaks to the proposed spectrum will usually improve the χ2 value, apriori knowledge of the sample and its associated number of peaks is imperative. Band fitting using theoretical spectra of mixtures from 0-100 wt% COM indicate a steady increase of the COM peak area using GRAMs software. (Figure 3.5) Using the equation provided by the data in Figure 3.5, a theoretical DL of 0.03±0.01 wt% COM was determined. It should be noted that, in theoretical calculations, COM peak areas below 10 wt% COM were indistinguishable using the available software. Reproducibility in the theoretical calculation was determined by fitting peak areas from the same spectrum five times in the same manner. The uncertainty in reproducibility ranged from 4-23% RSD. The poor fit of the amide I band of protein indicated by the error bars in Figure 3.5 will be addressed shortly.
81 Figure 3.5: Theoretical calculations for band areas of the 1633 and 1618 cm-1 band centers. Purple squares: amide I band. Blue diamonds: COM band.
Band fitting of Theoretical
1633cm-1 1618cm-1 25 y = 0.046x + 12.209 R2 = 0.2603 20
15
10
y = 0.0543x - 0.1578 5 R2 = 0.9686 Integrated Band Intensity Band Integrated
0 0 102030405060708090100 % COM
82 Band fitting calculations were repeated on the experimentally generated ATR spectra of dried mixtures ranging from 1-75 wt% COM. Figure 3.6 illustrates the experimental trends associated with both the amide I absorption of protein (pink squares) and the oxalate ion (blue diamonds) as the concentration of COM increases. Again, the poor fit of the amide I absorption of protein will be addressed shortly.
83 Figure 3.6: Experimental band fitting calculations for integrated band intensities of the 1633 and 1618 cm-1 absorptions.
Experimental Band Fitting Data
7
6
y = 0.0044x + 3.5168 2 R = 0.0141 5
4
3 IntegratedIntensity Band
2
1
y = 0.0164x + 0.5943 2 R = 0.6011 0 0 1020304050607080 % COM
84 Uncertainty for the experimental data points range from 3-19 % RSD for the 1618 cm-1 peak area of COM, and from 5-27 % RSD for the amide I peak area. All spectra were collected and analyzed in an identical manner using the same software during the same time period. Though the amide I band centered at 1633 cm-1 displays inconsistent values for
both the theoretical and the experimental data (Figures 3.5 and 3.6), the vas C=O stretch of oxalate increases monotonically with concentration. It is unclear as to why the spectral software has difficulty in determining the definitive placement of the amide I absorption. One major reason as to the variability of the protein measurements may include the heterogeneity and particle sizes of the sample. Though if this were the case, it would be reasonable to conclude that the oxalate ion measurements would display similar scattering and a similarly poor R2; however, the R2 for the Amide I band illustrates a far worse fit than that for the oxalate ion (0.0141 vs. 0.6011, respectively). A more likely conclusion is that the COM 1618 cm-1 absorption is easier to determine because of its higher extinction coefficient; pure COM has a much higher extinction coefficient than protein, yielding a much more intense absorbance. Because the COM peak grows in height and integrated peak intensity much faster and more definitively than the protein absorbance decreases, it is easier for the spectral software to determine COM absorption over the much weaker protein absorption. Therefore, there is less uncertainty in all aspects of the COM absorption than in the protein absorption and COM is easily detectible down to low concentrations. A wt% DL for COM using the 1618 cm-1 absorption was determined by adding 3x the RMS noise divided by the slope. It was determined that oxalate absorption has a detection limit of 0.09±0.01 wt% COM. Using the Student t-test [27], the 90% confidence limit corresponds to 0.09±0.01 wt% COM.
3.2.3 Reproducibility In order to determine the reproducibility of the instrumental response, the same site on an arbitrarily chosen COM/protein sample (35 wt% COM) was analyzed ten times. Using the absorption at 778 cm-1, which encompasses the O-C-O bend of oxalate, an average integrated peak intensity of 0.255 was found possessing a relative standard
85 deviation of 3.09%. Using the 1318 cm-1 absorption, the average integrated peak intensity was 0.360 with a relative standard deviation of 2.67%. Therefore it was established that instrumentally, reproducible experimental results are attainable. However, sample homogeneity was found to be lacking. Using the same 35 wt% COM sample analyzed at ten different sites, the average integrated peak intensity for the 778 cm-1 absorption was found to be 0.452, with a relative standard deviation of 54%. The average integrated peak intensity for the 1318 cm-1 absorption was 0.566 with a relative standard deviation of 64%. Plugging the experimental area of a 35 wt% COM sample (0.452) into the 778 cm-1 calibration curve equation of the line for the theoretical data (Figure 3.1), it was found that the experimental 35 wt% COM corresponds to the theoretical 31 wt% COM. One possible cause as to the lack of sample homogeneity could pertain to the physical sample size. The manner in which the thick films were prepared made films 1.25 in x 0.75 in. These viscous liquids dispersed themselves upon the glass slide prior to drying due to the amount of water added in an attempt to produce homogeneity. Perhaps, though, the water added was too much, and this made pockets of higher and lower concentrations of material as the film spread on the slide and particles aggregated due to their high mobility in the liquid matrix. Had the amount of water added been decreased, perhaps better homogeneity would result over the 1-90 wt% COM range made (though only up to 75 wt% COM was analyzed as previously discussed). Particle size variability, discussed in-depth in Chapter 5 of this dissertation, may also cause variations when using the ATR. Due to aggregation of the particles, the probe beam may be completely masked by the COM particles in the dried protein matrix, leading to irreproducible results.
3.3 Conclusions This research has been conducted in order to determine if ATR can quantitatively measure COM in the presence of protein. It was established that although linear relationships exist between the integrated peak areas and concentration of COM, the uncertainty in the DL’s is far too high to determine COM in a quantitative manner. The determined DL for these thick films is 0.20±0.03 wt% and 0.09±0.02 wt% COM, using
86 the 778 cm-1 and the 1318 cm-1 absorptions, respectively. The source of uncertainty originates in the lack of homogeneity in the sample, decreasing measurement precision. In addition, band fitting procedures were presented in an attempt to use the 1618 cm-1 absorption of COM in a quantitative manner. Using GRAMs spectral software, an experimental DL of 0.09±0.01 wt% COM in protein was determined.
87
CHAPTER 4
Determination of the Minimum Identifiable Quantity of Calcium Oxalate
Monohydrate Particles in a Protein Matrix
88 4.0 Introduction
The study of mineralized deposits forming from tissue containing finely dispersed renal stone components has increased in past decades. [3, 33, 104, 147] However, the progression of finely dispersed particles evolving into mineral inclusions and later renal stones continues to be under investigation. One of the keys to elucidating the chemistry of formation for specific renal stone types may lie in the analysis of embedded mineral deposits found in renal tissue samples. Mineral inclusion deposits such as those shown in Figure 4.1a offer valuable information as to the origins of the renal calculus produced, as well as the circumstances of calculus development and detachment. Since embedded mineral deposits are often only micrometers in size, the ability to detect, identify, and quantify these sites in a tissue matrix is a practical and pertinent goal. The first step in this process is the determination of the minimum identifiable quantity (MIQ), which is the focus of this report. The MIQ, according to the EPA document Method 8430, Section 9.2.1, is defined as “The minimum quantity that must be [analyzed] to result in a spectral match that has the correct compound identification in the top 5 spectral matches. The MIQ will vary depending on instrument sensitivity and sample matrix effects.” [148, 149] This definition assumes the user has access to a spectral library containing a standard spectrum of the analyte for comparison. Plainly, MIQ is the spectroscopic identification limit. Many of the tissue samples currently analyzed using infrared (IR) techniques do not contain embedded mineral deposits, however, infrared techniques have been previously employed on various types of tissue sections with very positive results. [61, 150-158] Additionally, the majority of tissue samples previously analyzed are either
stained or immobilized between two infrared-transparent BaF2 windows, such as that exhibited in Figure 4.1b. Such preparation techniques require the tissue to be sectioned to a thickness suitable for transmission spectroscopy (i.e., less than 6 μm). [43, 46] Estepa et.al. have successfully analyzed crystalline species in kidney tissues using FTIR microspectroscopy, however, their preparation required the thin slicing of tissue sections mounted upon infrared-transparent windows, which requires significant preparation time and produces fragile samples. [51] Estepa provides case studies, but no quantitative
89 information as to a detection limit for crystallized deposits or the minimum identifiable quantity.
90 Figure 4.1a: A stained tissue section mounted upon a low-e glass slide for analysis. Dark areas indicate stained mineral inclusions with sizes ranging from approximately 3 μm to 10 μm. Photo used in publication [43].
Figure 4.1b: A 4 μm sliced tissue biopsy held between two BaF2 windows for infrared microanalysis. Dark areas indicate stained mineral inclusions. Photo used in publication [43].
91 Previous studies of mineral inclusions in thin papillary tissue sections have been conducted using several thorough and intricate techniques, the majority of which are macroscopic in nature. [51, 154, 156, 159-161] Unfortunately, a macroscopic approach to mineralized inclusion analysis results in the loss of spatial information. This loss is exacerbated when multiple components are present in the sample. On the contrary, a microscopic approach allows for the analysis of individual inclusions, with resulting data contributing to the overall chemical etiology of the stone. However, detailed microscopic analyses can take several hours, as demonstrated in Chapter 2 using a surface reflectance technique. An example of infrared microspectroscopic analysis of embedded mineral species has been previously published by this laboratory, demonstrating the feasibility of using reflection/absorption infrared (RAIR) microspectroscopy for the qualitative analysis of thin tissue sections containing embedded mineral deposits. [43, 46] In contrast to this previous work, the present research focuses on coupling infrared microspectroscopy and ATR in order to provide information pertaining to the minimum identifiable quantity in addition to qualitative analysis. Molecular imaging using infrared microspectroscopic techniques, such as those presented in this chapter, yield infrared images of a sample that are easily and quickly obtained. Though the molecular images are qualitative in nature, they aid the detection of microscopic components often overlooked with the naked eye. With a spatial resolution on the order of several micrometers, the visual contrast of a false-color image offers researchers the ability to isolate and pursue specific features on a given sample. [43] In order to model the mineral inclusions in tissue sections, particles of calcium oxalate monohydrate (COM) and hydroxylapatite (HAP) were individually dispersed onto thin layers of protein applied evenly upon a low-e glass substrate. A low-e glass substrate is a standard glass slide over which three thin metallic layers have been deposited. This layering allows the transmission of visible light while reflecting 100% of the infrared radiation. Therefore, low-e slides are conducive to both optical microscopy as well we infrared reflectance microspectroscopy. [162] The particles were dispersed heterogeneously over the film in order to imitate pockets of mineral buildup in a protein (tissue) matrix. The individual particles were
92 subsequently analyzed using ATR techniques as well as infrared imaging in reflectance mode. The research presented in this chapter addresses the minimum particle sizes that can be identified using infrared microspectroscopy. Other research corresponding to the minimum identifiable quantity (MIQ) using infrared analysis pertains to the analysis of zirconium stationary phase performance for thin layer chromatography using infrared DRIFTS and transmission spectroscopy as a detector. [163] However, to the best of our knowledge, no published research details the identification of COM particles (or other renal stone components) in a protein matrix using infrared techniques.
4.1 Experimental 4.1.1 Materials COM (Acros Organics) was manually ground using a mortar and pestle prior to analysis in an attempt to obtain a range of particle sizes, while hydroxylapatite, because its particle size is smaller than a micrometer, (Spectrum Chemical, New Jersey) was used unaltered from the bottle. Protein in the form of unflavored gelatin powder (Kroger Company, Ohio) was used in both dried form for wt/wt ratios and in gelatin form prepared according to package directions. To construct a protein layer suitable for particle size analysis, liquid gelatin was prepared by dissolving one packet of gelatin in ~400 mL of boiling water according to packet instructions. After dissolution of the solid, the gelatin (protein) was evenly applied to a low-e substrate via the draw-down method to generate a thin film. Upon drying, this process was repeated until a substantial protein layer of uniform thickness had been built. Once the final protein layer had been placed upon the low-e slide, COM or HAP was dispersed over the surface of the liquid by gently tapping dust-like particles from the tip of a micro spatula. The slide was then allowed to dry for 24 hours in a desiccator. 4.1.2 Instrumentation The location of particles randomly dispersed in a protein matrix was confirmed by using a Perkin Elmer (PE) Spectrum Spotlight 300 infrared imaging microscope. The array detector is a 16x1 linear array, and each pixel is a 30x30 μm mercury cadmium
93 telluride (MCT) element. Spectra collected represent the average of 8 scans with a spectral resolution of 4 cm-1. Linear array background spectra were taken on the clean surface of a low-e substrate. Particle analysis via ATR was also performed on the PE Spotlight 300 using the drop-down Ge internal reflection element (IRE) and the 100x100 μm single point MCT detector. Spectra collected represent the average of 64 scans with a spectral resolution of 4 cm-1. ATR background spectra were collected using a KCl pellet as the reference material. ATR aperture size was adjusted according to individual particle size.
4.2 Results and Discussion The detection of individual component particles in the presence of protein contributes towards an understanding of the formation of stone nucleation points within tissue. For the analysis of particles modeling mineralized deposits in a protein matrix, the reflectance mode of the Perkin Elmer (PE) Spectrum Spotlight 300 infrared imaging microscope was used to produce false color images of the dried protein layer. (Figure 4.2) This false color image, containing nearly 33,000 individual spectra, yields the exact location of dispersed components in the tissue matrix and can allow the subsequent identification of these components.
94 Figure 4.2: Molecular image of COM particles dispersed in a protein matrix. Particles range in size from 10 μm to 40 μm.
95 The use of ATR infrared microspectroscopy as a tool for the identification of minute renal stone components is advantageous over reflectance microspectroscopy for several reasons. In ATR analysis, a shaped germanium IRE is placed in contact with the sample, and the size of the sampled area is dependent upon both the refractive index of the IRE and the size of the confocal aperture. [164] For the same aperture size as used in reflectance analysis, the spatial resolution is four times better in the ATR analysis due to the 4-fold magnification associated with the refractive index of the germanium IRE
(nGe=4.0). [165] Therefore, a 40 μm aperture translates into a 10 μm spot size on the sample surface. There is only a 2-fold improvement for ATR over transmission techniques since only half of the objective is used, reducing the numerical aperture from 0.6 to 0.3. This 2-fold loss coupled with the 4-fold gain from the IRE refractive index provides a 2-fold net improvement. Further discussion about the instrumental differences between reflectance and transmission microspectroscopy can be found in Chapter 1. Another benefit of the ATR method is that the evanescent wave penetrating into the sample is often less than 1 μm. Therefore, the optical path through the sample is independent of sample thickness for samples thicker than this penetration depth (dp). [43] As a result, spectra can be obtained from extremely thick sample sections. For the analysis of mineral deposit models, the protein is analyzed simultaneously with the component of interest. Unfortunately, the identification of COM in a protein matrix is complicated due to the close proximity of the 1633 cm-1 amide I absorption of -1 protein to the 1618 cm vas C=O stretching absorption of oxalate. Therefore, the O-C-O -1 -1 absorption of oxalate at 778 cm , and the vs C=O stretch of COM at 1318 cm , are used for identification and subsequent measurements. In order to determine the MIQ of renal stone components without the presence of a protein matrix, KCl substrates containing either COM or HAP particles dispersed upon the surface were analyzed utilizing the drop-down ATR. Spectroscopically, COM can be identified down to a 10 μm particle (Figure 4.3), while HAP can be identified to a 20 μm particle. Corresponding spectra are illustrated in Figure 4.3, while the calibration curves are displayed in Figures 4.4 and 4.5. Calibrations curves such as those in Figures 4.4 and 4.5 were generated in order to approximate a detection limit or limit of quantitation (LOQ) of the compounds, since knowledge of a LOQ may be useful for future
96 quantitative studies. As can be observed in the following calibration curves, a linear relationship can be established between particle size and absorbance, indicating that future quantitative studies are possible.
97 Figure 4.3: Spectra of HAP and COM particles on KCl.
20 um COM
10 um COM
A 20 um HAP
10 um HAP
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 580 cm-1
Figure 2: Spectra of HAP and COM particles on KCl.
98 Figure 4.4: Calibration curves of COM particles on bare KCl substrates.
-1 COMCOM on on KCL KCl 1300-1330 1318 cm cm Absorption-1
7 y = 0.164x - 3.5729 2 5 R = 0.976
3
1 Integrated Band Intensity Band Integrated -1 15 25 35 45 Particle size (µ)
COM onCOM KCL on KCl778 730-820cm-1 Absorption cm-1
8
6 y = 0.1894x - 4.031 2 4 R = 0.9821
2
0
Integrated Band Intensity Band Integrated 15 25 35 45 -2
Particle Size (µ)
99 Figure 4.5: Calibration curve of HAP particles on bare KCl substrates. 10 µm particles are not determinable using the MIQ (Figure 4.3).
-1 HAP on KCL 1020 cm Absorption
40
35 y = 0.7153x - 15.782 R 2 = 0.9117 30
25
20
15 10
5 Integrated Band Intensity 0 0 102030405060 -5
-10
-15 particle size (u)
100 It is theoretically expected that the integrated band intensity (absorbance) of a single particle is directly proportional to the sample thickness, and should obey Beers
Law. If a particle, larger than the diffraction limited diameter, is thicker than the dp of the evanescent wave, the resultant absorption should remain constant. If, however, the particle is thinner than the dp of the evanescent wave, the collected absorption will vary with thickness. Additionally, if the position of the IRE is off-center, the IRE will come into contact with more of the protein surrounding the particle, detecting a smaller absorption for the measured size of the particle and generating non-linear points in the calibration curve. In the research presented here, the thickness of the sample is unknown, adding to the uncertainty of the averaged data points. Though the actual diameter of HAP particles is 0.5-0.25 µm, aggregation of particles can produce larger “particle sizes”. Likewise, though the diameter of COM is 150-200 µm, smaller fractions of a single crystal can occasionally be found, though finding extremely small fractions as those exhibited here is difficult. As can be seen from the previous Figures, the uncertainty in the data points is large. When using small apertures in the ATR such as 40 and 80 μm (translating to 10 and 20 μm on the sample surface, respectively), the overall absorption reproducibility is poor. For instance, for an 80 μm aperture (20 μm particle), the averaged area of four measurements is 0.1821, while the standard deviation of the four measurements is 0.1291; this equals a 70% RSD. The poor reproducibility arises from the slight uncertainty in the placement of the IRE upon the surface of the sample. Using a 25 µm latex particle mounted on a KCl substrate, 10 ATR measurements were collected using a 100 µm aperture. Using the same bead for each measurement, the relative standard deviation of the measurements was determined to be 30 %, displaying the error associated with IRE placement. Using both the 778cm-1 and the 1318 cm-1 absorptions, particles of COM as small as 40 μm could be identified when dispersed upon a tacky protein layer. Spectra are displayed in Figure 4.6.
101 Figure 4.6: Spectra of COM in a protein matrix. The first visual appearance of COM occurs with a 40μm particle. Bottom: 1318 cm-1 band of COM.
Pure COM
Pure Protein
40 um COM
A
30 um COM
25 um COM
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 580 cm-1 0.357 0.34 Pure COM
0.32
0.30 30 um COM particle 0.28 40 um COM particle 0.26
0.24
0.22
A 25 um COM particle 0.20 Pure Protein 0.18
0.16
0.14
0.12
0.10
0.08
0.063 1355 1350 1345 1340 1335 1330 1325 1320 1315 1310 1305 1300 1295 1290 1285 1280 1276 cm-1
102 Hydroxylapatite, dispersed upon the tacky protein layer in the same manner as COM, yields interesting and unexpected results. It is suspected that the orthophosphate group of hydroxylapatite centered at 1018 cm-1 reacts with a sulfate group in the tacky protein layer, resulting in the appearance of calcium sulfate. Figure 4.7 displays three spectra: neat protein, neat hydroxylapatite, and the calcium sulfate spectrum obtained from the sample.
103 Figure 4.7: Spectra belonging to neat protein (black solid line), neat hydroxylapatite (red dashed line), and the calcium sulfate spectrum associated with the sample dispersed in protein (green dash-dot line).
0.449
0.40
0.35
0.30
0.25
0.20 A
0.15
0.10
0.05
0.00
-0.052 4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 580 cm-1
104 -5 o The observation of calcium sulfate (Ksp=2.4x10 at 25 C) [166] is interesting -59 since CaSO4 appears much more soluble than HAP—Ca10(PO4)6(OH)2 (Ksp=2.0x10 at 37o C). [167] However, this assumption can be misleading. If we solve for the actual solubility of the components, we see that their solubility is similar.
2+ 2- CaSO4↔Ca + SO4 2+ 2- -5 Ksp=[Ca ][SO4 ]=X*X=2.4x10 2 -5 -3 2+ 2- X =2.4x10 X=4.9x10 =[Ca ]=[SO4 ]
2+ 3- - Ca10(PO4)6(OH)2↔10 Ca +6 PO4 +2 OH 2+ 10 3- 6 - 2 10 6 2 -59 Ksp=[Ca ] [PO4 ] [OH ] =10X *6X *2X =2.0x10 1010*66*22*X18=2.0x10-59 X18=1.7x10-74 -5 2+ -4 3- -4 - -4 X=7.8x10 [Ca ]=7.8X10 [PO4 ]=4.7X10 [OH ]=1.6X10
The calculations prove that CaSO4 is more soluble than HAP, however, the difference in solubility is not as large as one would initially expect. It is unclear why the
more soluble CaSO4 previals over phosphate in the presence of this particular protein. It was determined that experiments involving HAP and this particular protein source must be conducted while both components are dry. In light of this side reaction, a slightly altered experimental technique was used for the MIQ detection of HAP in the presence of protein. HAP, loosely dispersed upon the surface of a dried protein film, circumvents any possible side reactions. Using this technique, ATR methods produced a particle MIQ of 15 μm for HAP (Figure 4.8)
105 Figure 4.8: Spectra of HAP particles on protein. MIQ of HAP is 15 µm.
Pure HAP
15 um HAP
A
10 um HAP
Pure Protein
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 580 cm-1
106 4.3 Conclusions This chapter has demonstrated the ability of ATR microspectroscopy to identify COM particles down to 10 μm in size and HAP particles down to 15 µm in size. Additionally, it is questionable as to whether HAP can be used in the presence of this particular brand of liquid gelatin protein due to a side reaction of the calcium with sulfate groups present in the protein. Therefore it may be necessary to use HAP with dried protein in tissue and mineral inclusion models.
107
CHAPTER 5
Non-linear Calibration Curves of Renal Stone Components Analyzed by ATR-IR
108 5.0 Introduction Analytical calibration, a method of determining the relationship between instrument response and analyte concentration, is necessary for the accurate measurement of an analyte in a substance. [168] The calibration curves produced in this research result primarily from the relationship between calculi components and their respective infrared absorbance. The goal of these calibration curves is to elucidate a useful equation for the determination of specific compounds in mixed-component renal stone samples by modeling a mixed composition. By knowing the respective concentrations of renal stone material in mixed-component stones, the urine chemistry and time duration of different urine environments can be determined, and patient health cycles can be treated. The majority of renal calculi are not pure in composition; it has been shown by Estepa that approximately 40% of renal stones have more than 3 components. [42, 51, 159, 169-173] For example, Carmona has discovered renal stones with as many as 82 components [3], and Daudon has studied mixed renal stones containing foreign components, such as pharmaceutical compounds. [17, 174] Since a preponderance of one component over another may influence the patients prescribed treatment, a method of accurately determining the amount of components present in a renal stone is a fundamental need. An ideal technique for the analysis of renal stones and the calibration of renal stone standards is attenuated total internal reflectance (ATR) infrared (IR) microspectroscopy due to the lack of required sample preparation and the fast analysis time. ATR techniques can yield photometrically accurate absorbance bands in a matter of seconds, allowing qualitative as well as quantitative analysis to be conducted with high sample throughput. ATR microspectroscopy uses a high refractive index crystal, referred to as an internal reflection element (IRE), in intimate contact with a sample for surface analysis. [127] A complete discussion of ATR is presented in Chapter 1 of this dissertation. Since the penetration depth of the evanescent wave (Equation 1.5) is approximately 1 μm, the sample thickness is not a determining factor in the collection of a spectrum. As such, sample preparation time is greatly diminished. In addition, the small penetration depth of the infrared radiation into the sample is ideal for the analysis of strongly infrared absorbing materials. [175]
109 Another advantage of ATR spectroscopy is increased spatial resolution for certain instrument configurations. Due to the refractive index of the IRE, a magnification factor is associated with ATR. [165] This increased spatial resolution can be a drawback, however, when samples are not homogeneous and larger spatial averaging would provide a more accurate analysis of the material (as demonstrated in Chapter 3). Helmy analyzed solid pharmaceutical tablets in order to determine the homogeneity of the drug excipients as well as the polymorphs present in the sample. [176] Helmy found a limit of detection of 3 % by weight with an uncertainty as high as 28 % by weight for data points at low concentrations. This reference is an example of the quantitative abilities of ATR for compact samples (such as the pellets used in the current research), as well as the ability of ATR to quantitate relatively small amounts of components using linear calibration curves. Though linear calibration curves generated for the measurement of components is considered the norm, non-linear calibration curves have also been proven analytically useful. Estepa and Daudon have experienced the phenomenon of non-linear calibration curves for renal stone components using experiments run in DRIFTS and transmission modes as opposed to ATR. [51, 177] In a recent publication, Estepa et.al. published non- linear calibration curves resulting from several renal stone component mixtures obtained through patient procedures. [51] However, these curves were published without detection limits, calibration equations, experimental proof, or reasons justifying or hypothesizing their non-linear nature. In addition, the calibration curves are displayed in separate sections of less than and greater than 50 wt% of the component, making visual continuity of the curves difficult. Another interesting investigation of Cohen-Solal and Daudon et al is the automated identification and quantitative measurement of renal stone components using a DRIFTS/KBr pellet process. [134] In their report, spectral addition of carbapatite and calcium oxalate dihydrate (COD) was employed to generate theoretical calibration curves in a manner identical to that demonstrated in this chapter. Comparisons to their work, with exception of the simulated calibration curve, are difficult since only three real samples were used to generate the calibration curve (30, 50 and 70% carbapatite in COD). All other points in the curve were generated mathematically. Additionally,
110 Cohen-Solal and Daudon relied on the expertise of urologists for visual confirmation of renal stone components present in samples, however, it has been shown that human examination can be inaccurate up to 40% of the time. [45, 48] The current chapter investigates both methods: the generation of a calibration curve from spectra of COM and HAP by spectral simulation as well as the generation of a calibration curve from renal stone standards. The focus of the calibration curves studied in this chapter is the relationship between the two most common renal stone components: calcium oxalate monohydrate (COM) and hydroxylapatite (HAP). Calibration curves were developed from standards in varying ratios and were evaluated via ATR-IR microspectroscopy. The integrated band inintensities for specific components were then plotted against the analyte concentration in weight percent to generate the calibration curves. A concern that arises during the generation of these calibration curves is the particle size difference between the components, and their resultant non-linear calibration curves. One hypothesis examined in this research is the idea that large particles aggregate in the sample, leading to non-linear calibration curves. Wadayama et.al. has noticed an increase in integrated band intensity using ATR on aggregated silver nanoparticles. [178] Unfortunately, no calibration curves are presented in the article. The limit of quantification (LOQ) is the minimum amount of substance that can be quantitatively measured via a given instrumental method, and is determined by multiplying the standard deviation of the noise of the spectrum by ten, dividing by the slope of the calibration curve, and adding it to the average “background” signal for the wavenumber range in question. [58] For the purposes here, the background is the spectrum of the non-analyte component. For instance, if COM in HAP is being examined, and the LOQ for the 1300-1345 cm-1 range of COM is in question, then 10x the standard deviation of the noise of the spectrum divided by the slope is added to the signal of 100 % HAP for the range 1300-1345 cm-1. [58] Any value over this calculated value can theoretically be measured quantitatively. LOQ is slightly different from the detection limit (DL), which is 3x the standard deviation of the noise divided by the slope. In order to approximate an “ideal” LOQ, theoretical calibration curves were generated using spectra of COM and HAP that were mathematically produced to imitate
111 weight percents of the components. Previous precedence for this technique has recently been published by Cohen-Solal, who used a theoretically generated calibration curve for the analysis of renal stone components. [134] Using the LOQ criteria detailed below, a theoretical minimum of 1.3±0.3 wt%COM can be quantitatively measured in the presence of HAP. A theoretical calibration curve indicates a response range of the instrument for the given analyte under ideal conditions, and is an important parameter to determine prior to experimental data collection. Experimental data was subsequently collected using weight percent ratios; this procedure is well established and used often in the pharmaceutical industry. [176, 179] Though the LOQ for COM was determined, it is very different from the theoretical value, indicating that the non-linear calibration curves generated should not be used for quantitative measurements.
5.1 Experimental 5.1.1 Materials The following compounds were acquired for the generation of ATR calibration curves: calcium oxalate monohydrate (COM)(Acros Organics, NJ), hydroxylapatite (HAP) (Spectrum Chemical, Gardina, CA), L-cystine (Aldrich, Milwauee, WI), potassium chloride (KCl) powder (Fisher, Fair Lawn, NJ), anionic uniform latex beads (0.108±0.01μ, Magsphere, Pasadena, CA), and uric acid (UA)(Aldrich, Milwauee, WI). 5.1.2 Instrumentation ATR spectra were collected using a Harrick Split-pea ATR accessory interfaced to a Perkin Elmer 2000 Fourier transform infrared spectrometer. This accessory employs a single bounce Si (n=3.4) IRE and the standard deuterium triglycine sulfate (DTGS) detector on the Spectrum 2000 macro bench. The IRE is 3 mm in diameter with an infrared active region of 200 µm in diameter. Spectra collected represent the average of 32 individual scans with a spectral resolution of 4 cm-1. The samples were brought into intimate contact with the IRE using a loading of 0.5 kg. Scanning electron microscopy (SEM) data for particle sizes was obtained using a Zeiss Supra 35 FEG-VP scanning electron microscope. Dried particles were dispersed on a carbon tab and subsequently coated with a thin layer of gold. Images were collected
112 using a range of magnifications. Particle sizes were subsequently measured visually at various magnifications using the calibrated scales available on the instrument display. By measuring an assortment of particles, an approximate size range was determined. 5.1.3 Mixture Preparation Mixtures were made using weight percent ratios with a total mass of ~1.0g. Several approaches of combining materials were attempted in order to determine an optimal method for making homogeneous mixtures of pure materials. The first method involved the bulk grinding of individual powdered components by hand, subsequently weighing out and combining components to produce the proper ratios. This method resulted in excess ground material. The second method consisted of weighing out the necessary amount of raw powdered components for a given ratio, combining the components, and subsequently grinding the powdered mixture together. The second method was found to be time consuming. The third method was a combination of the first two methods: individual ground components were weighed and combined and subsequently reground in the mixture. This method was also time consuming. The final product for all methods resulted in a homogeneous powder, as tested using ATR and described in section 5.4. Therefore the first method was employed for all calibration curves, regardless of components. The following mixtures were made: COM/KCl, COM/HAP, COM/latex, COM/UA, HAP/latex, HAP/KCl, HAP/UA and L-Cystine/KCl. Ratios of components were made in the following amounts for most mixtures: 0, 1, 3, 5, 7, 10, 20, 30, 40, 50, 60, 70, 80, 90, 93, 95, 97, 99 and 100%. Sample pellets were pressed using a manual pellet press (Parr Instruments, Moline, IL). 5.1.4 Peak Areas and Analysis Method Spectra were analyzed in absorbance mode using the peak height/area function of the Spectrum v.5.0.1 software (Perkin-Elmer, Shelton, CT). Integrated peak intensities of -1 the C-O bend from 770-790 cm , the vs C=O stretch of the oxalate ion from 1300-1345 -1 -1 cm , and the vas C=O stretch of the oxalate ion from 1500-1700 cm of calcium oxalate monohydrate were determined, as well as the asymmetric stretch from 970-1125 cm-1 of the orthophosphate group of hydroxylapatite. Integrated peak intensity was used for all absorptions. Data from the 1618 cm-1 absorption of oxalate is used in this research,
113 however, it should be noted that the 1618 cm-1 absorption overlaps with the protein amide I absorption at 1633 cm-1 when protein is present, as explained in Ch.3. Five pellets per concentration were analyzed and subsequently averaged together to generate the calibration curves.
5.2 Results and Discussion Pharmaceutical analysis often involves the qualitative and quantitative analysis of powdered composite samples. [180-183] Pharmaceutical tablets are checked for homogeneity of the active drug and dispersing excipient, often called binding agents. These measurements ensure the controlled release of the drug into the body required by the FDA. [184] Calibration curves using ATR-IR methods are common and well established, most often generating linear calibration curves. [185, 186] Like the use of infrared techniques for pharmaceutical analysis, renal calculi containing dispersed components are seemingly ideal candidates for ATR-IR quantitation. Unfortunately, however, the stone models studied in this research often produced non-linear calibration curves for reasons not obvious upon initial examination. 5.2.1 Theoretical Considerations In order to model the calibration curves and determine a theoretical LOQ, a spectrum obtained on a neat calcium oxalate monohydrate (COM) pellet was mathematically added to a similar spectrum of hydroxylapatite (HAP) with different weighting factors in order generate different standards. The equation used, Equation 5.1, illustrates this method.
(0.X*neat COM spectrum)+(0.Y*neat HAP spectrum)=Ratioed spectrum Equation 5.1 Where 0.X+0.Y=1.0
Using Equation 5.1, spectra were synthetically generated over a concentration range of 0-99% by weight COM in HAP. Figure 5.1 displays an example of these spectra.
114 Figure 5.1: Theoretical spectra calculated using a spectral calculator and the parameters defined in Equation 5.1.
1318 cm-1 1020 cm-1 778 cm-1
1400 1300 1200 1100 1000 900 800 700 cm-1
115 The theoretical calibration curves (Figure 5.2) arising from the data are linear for all peaks analyzed, as is expected. The LOQ for the 1618 cm-1 absorption (1500-1700 cm-1 range) is 1.3±0.2 wt% COM, the LOQ for the 1318 cm-1 absorption (1300-1345 cm- 1 range) is 7±2 wt% COM, and the LOQ for the 778 cm-1 absorption (770-790 cm-1 range) is 33±10 wt% COM. Ten times the standard deviations is 0.03, 0.3 and 0.5, respectively, while the 100% HAP values for the 1618 cm-1, 1318 cm-1 and 778 cm-1 absorbance bands are 1.1949, 0.0002 and 0.0468, respectively. The uncertainty is 3x the standard deviations divided by the slope [58] and is associated with the differences in the reproducibility of the neat spectra of COM and HAP prior to mathematical addition.
116 Figure 5.2: Theoretical integrated band intensities for the main peaks in the COM/HAP mixture. Uncertainty is 3x the standard deviation divided by the slope [58] and is associated with the differences in the reproducibility of the neat spectra of COM and HAP prior to mathematical addition.
1300-1345 cm-1
770-790 cm-1 Integrated Peak Intensity (Theoretical) 970-1125 cm-1
-1 30 1500-1700 cm
25 y = 0.229x + 1.195 y = -0.2044x + 19.843 20
15
10 Integrated Peak Intensity
5 y = 0.0411x - 0.0159
y = 0.0146x - 0.0016 0 0 102030405060708090100
-5 % COM
117 5.2.2 Experimental Calibration Curves The mixtures of renal stone components in specific ratios resulted in non-linear calibration curves for many binary combinations attempted. These results are summarized in Table 5.1. Also tabulated in Table 5.1 are the mixtures associated linearity, number of breaks in the calibration curve, and the percent of the component at which the break appears. In Table 5.1, linearity is assigned to calibration curves having R2>0.95.
118 Table 5.1: A matrix describing the binary associations of renal stone components and their resulting linearity. COM, HAP and UA, three very common components, are included in this research. Additionally, latex beads were used as part of a control experiment pertaining to particle size. Linearity implies an R2>0.95.
119 The first binary mixture made contained HAP and COM. Figures 5.3a-5.3c are the calibration curves resulting from the integrated band intensities of COM located at 778, 1318, and 1618 cm-1, respectively. Each curve exhibits a definitive break present at approximately 20% COM in HAP. For all calibration curves presented in this chapter, the curve with and without the trend line is provided so that the character of the calibration curve can be observed without a trend line.
120 Figure 5.3a: The calibration curve of the binomial mixture containing COM and HAP. 778 cm-1 band. Uncertainties range from 4-18% RSD due to slight inhomogeneity in the sample. Particle sizes: COM—200 µm; HAP—0.25 µm; a ratio of 1:800 COM:HAP
778 cm-1 COM in HAP
1.2
1.0
0.8 y = 0.0088x + 0.269 R2= 0.9952 0.6
0.4 Integrated Peak IntegratedIntensity Peak 0.2 y = 0.0205x + 0.0314 R2= 0.977 0.0 0 102030405060708090100 % CaOx in HA
1.2
1.0
0.8
0.6
0.4 Integrated Peak IntegratedIntensity Peak 0.2
0.0 0 102030405060708090100 % CaOx in HA
121 Figure 5.3b: 1318 cm-1 COM band. Uncertainties range from 1-16% RSD due to slight inhomogeneity in the sample. Particle sizes: COM—200 µm; HAP—0.25 µm; a ratio of 1:800 COM:HAP
0.90 -1 0.80 1318 cm COM in HAP
0.70
0.60
0.50
0.40 Corrected Area Corrected 0.30
0.20 Integrated Peak Intensity Intensity Peak Integrated
0.10
0.00 0 102030405060708090100 % COM in HAP 0.90
0.80
0.70 y = 0.0051x + 0.2754 2 0.60 R = 0.9888
0.50
0.40 Corrected Area Corrected
Integrated Peak Intensity Intensity Peak Integrated 0.30 y = 0.0189x + 0.0192 2 0.20 R = 0.9839
0.10
0.00 0 102030405060708090100 % COM in HAP
122 Figure 5.3c: 1618 COM cm-1 band. Uncertainties range from 3-19% RSD due to slight inhomogeneity in the sample. Particle sizes: COM—200 µm; HAP—0.25 µm; a ratio of 1:800 COM:HAP
-1 1618 1500-1700cm COM cm in-1 HAP
20
15
10
5
Integrated Peak Intensity Peak Integrated 0 0 102030405060708090100 % COM
18 16 14 12 y = 0.1393x + 3.3394 10 y = 0.3158x + 0.4883 R2 = 0.9833 8 R2 = 0.9917 6 4
Integrated Peak Intensity Peak Integrated 2 0 0 102030405060708090100 % COM
123 The LOQ for the data in Figure 5.3 was found with 10x the standard deviation of the noise of 0.5, 0.03 and 0.003 for the 778, 1318 and 1618 cm-1 absorptions, respectively. The LOQ for the lower segment of the 1318 cm-1 absorption is equal to 1.6±0.5 wt%, which is smaller than the theoretical value of 7±2 wt%. LOQ’s for the bottom segments of the 778 cm-1 (background= -0.0016) and 1618 cm-1 (background=1.1949) absorbances are 24±6 wt% and 1.2±0.3 wt%, respectively; both of these values are smaller than their respective theoretical values of 33±10 and 1.3±0.2 wt % COM. The upper calibration portions present LOQ’s of 1.2±0.3, 0.6±0.2 and 57±15 wt% COM, respectively, for the 1618, 1318 and 778 cm-1 absorbances. While one can utilize the 778 cm-1 absorption for quantitative analysis, it is analytically ideal to use the most intense absorption in order to determine the lowest LOQ possible. Using the most intense absorption provides a first indicator that the analyte is present and measurable. However, both weak and strong absorptions yield apparently inaccurate values for the calibration curves generated. Observing the calibration curves resulting from the mixture of COM and HAP, one is unable to determine what is responsible for the non-linearity. Estepa has encountered non-linear calibration curves in previous research similar to those presented here. [51] However, no mention is made as to the theory behind the non-linearity or the accuracy of the curves. Instead, he states there is no need for accurate quantitation of stone components in a clinical setting. In experimental trials, Estepa found the curves to produce reasonable values. Daudon has also experienced non-linear calibration curves, and has stated that he is uncertain as to their origin, [177] hypothesizing that particle size may be a contributing factor. [187] When using ATR, particle size is the main concern. The penetration depth of the infrared radiation and the refractive index of the sample are also considerations, however, in the case here, the particle size differential appears to be paramount. The depth of penetration by the evanescent wave and the refractive index difference between the renal stone components are nearly inconsequential when a large particle component is present, since (as will be shown in section 5.3.1) the effective sampling depth does not traverse deeper than the large particle diameter. This lack of equal sampling causes a deviation from Beers Law.
124 Linearity is a function of Beers Law (Equation 1.1), and while Beers Law generally produces a region of linearity at lower analyte concentration, Beers Law often fails at higher concentrations. [57] Additionally, interactions of the analyte with the solvent can hinder analysis. However, for the purposes of this research, either COM or HAP is the solvent, and no interaction should occur since both components are in solid form. A larger concern is the heterogeneous sampling due to unequal particle sizes. The larger component, COM, is nearly the same size as the probe beam diameter exciting the IRE. [140] This concept will be expanded in section 5.3.1. Briefly, the large size differential causes the evanescent wave to come into contact with more COM than HAP. Since the ATR method is micro in character, size differentials can be extremely important. It is possible that macro analysis of the sample pellets would produce a linear relationship adhering to Beers Law. In order to observe the behavior of each component individually, HAP and COM were separately mixed with KCl. KCl has particle sizes ranging from 0.8-8 µm in diameter. While the presence of KCl should not influence the measurement of the analyte response, KCl can influence data indirectly by affecting analyte particle spacing. Because of KCl’s relatively small size compared to COM, the KCl/COM (ratio of 1:25) calibration curve is strikingly similar to that of the HAP/COM (ratio of 1:800) calibration curve (Figures 5.4 and 5.5). However, in the KCl/HAP mixture, the size discrepancy is only a single order of magnitude, meaning that a more homogeneous mixture results. Therefore the resulting calibration curve is a measurement of particle spacing (concentration) of the analyte, where KCl is evenly dispersed in the analyte and may reduce aggregation of the HAP.
125 Figure 5.4: COM in KCl. Uncertainties range from 1-27% RSD due to inhomogeneity in the sample. Particle sizes: COM—200 µm; KCl—~8 µm; ratio COM: KCl is 1:25. 1318 cm-1 COM in KCl 7
6
5
4 Corrected Area Corrected 3
2
Integrated Peak Intensity Intensity Peak Integrated 1
0 0 102030405060708090100 % COM in KCl 7
6
5 y = 0.0211x + 4.2369 2 R = 0.8465
4
Corrected Area Corrected 3
2 y = 0.1556x + 0.4485 R2 = 0.9936 1 Integrated Peak Intensity Intensity Peak Integrated
0 0 102030405060708090100 % COM in KCl
126 Figure 5.5: HAP in KCl. Uncertainties range from 2-16% RSD due to slight inhomogeneity in the sample. Particle sizes: HAP—0.25 µm; KCl—~8 µm; ratio of HAP:KCl is 32:1.
1020 cm-1 HAP in KCL 30
25
20
15
Corrected Area Corrected 10
Integrated Peak Intensity Intensity Peak Integrated 5
0 0 102030405060708090100 30 % HAP in KCl
y = -0.282x + 42.953 2 25 R = 0.9892 y = 0.2844x + 10.17 2 R = 0.9655
20 y = -0.0201x + 22.208 2 R = 0.92
Corrected Area Corrected 15
Integrated Peak Intensity Intensity Peak Integrated 10
y = 1.0783x + 1.19 R2 = 0.9971 5
0 0 102030405060708090100
% HAP in KCl
127 COM mixed with KCl (Figure 5.4) has a break occurring between 20 and 30% COM. Additionally, HAP mixed with KCl (Figure 5.5) has multiple breaks in linearity, one of which occurs at 80% HAP, which coincides with 20% COM in the COM/HAP mixture displayed in Figure 5.3. These results indicate that particle sizes possibly contribute to the break in linearity in the mixed component system. Therefore, particle sizes were investigated more thoroughly.
5.3 Particle Sizes Using scanning electron microscopy, the average particle diameters of HAP and COM were observed to vary greatly, with HAP ranging from 0.1-0.25 μm, and COM ranging from 150-200 μm. These are illustrated in Table 5.2. Figures 5.6 and 5.7 illustrate the associated SEM images of COM and HAP.
128 Table 5.2: Particle size and linearity matrix.
129 Figure 5.6: SEM images of COM.
COM before grinding (250x)
COM after grinding (250x)
130 Figure 5.7: HAP before and after grinding.
HAP before grinding (66600x)
HAP after grinding (65000x)
131 Due to concerns that the particle sizes of the renal stone components play a role in the non-linear characteristics of the calibration curves, COM and HAP were individually combined with latex beads of known size (0.108 μm in diameter) in order to evaluate their response. In this experiment, the latex beads act as a size standard in order to gauge the response of the other components. Figure 5.8 below displays the spectrum of pure latex and its associated absorption assignments. [188] Figures 5.9-5.10 display the results of the COM and latex mixtures, while Figures 5.11-5.12 display the HAP and latex results.
132 Figure 5.8: Pure polystyrene latex spectrum with associated band assignments.
90
88 86 84
82 3025.74 1601.27 80 1583.79 78 2850.71 1069.65 2921.18 76 1492.86 1028.31 1451.91 74 754.49 72 70 %T 68 66 64 62 60 58 56 54 52 696.41
49 4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 cm-1
133 Figure 5.9: COM and Latex (Latex band). Uncertainties range from 2-26% RSD due to slight inhomogeneity in the mixing of the sample. Particle sizes: COM—200 µm; Latex—0.108µm. Ratio of COM:Latex is 1:1851.
-1 0.9 1451 cm Latex Band 0.8 0.7
0.6
0.5 0.4 0.3
0.2
0.1 Integrated Peak Intensity Intensity Peak Integrated 0 0 102030405060708090100 -0.1 % Latex in COM
y = 0.0101x - 0.1334 0.9 2 0.8 R = 0.992 0.7 0.6 0.5 0.4 y = 0.0011x + 0.1296 2 0.3 R = 0.8691 0.2 Integrated Peak Intensity Intensity Peak Integrated y = 0.0177x - 0.0362 0.1 2 R = 0.9943 0 0 102030405060708090100 -0.1 % Latex in COM
134 Figure 5.10: COM and Latex (COM band). Uncertainties range from 1-18% RSD due to slight inhomogeneity in the sample. Particle sizes: COM—200 µm; Latex—0.108µm. Ratio of COM:Latex is 1:1851.
1318 cm-1 COM band in Latex 6
5
4
3 Corrected Area Corrected 2
1 Integrated Peak Intensity Intensity Peak Integrated
0 0 102030405060708090100 % Latex in COM 6
5
y = 0.0536x + 0.2562 4 2 R= 0.9927
3
Corrected Area Corrected 2
1 Integrated Peak Intensity Intensity Peak Integrated
0 0 102030405060708090100 % Latex in COM
135 Figure 5.11: HAP and Latex (Latex band). Uncertainties range from 0.1-2 % RSD due to slight inhomogeneity in the mixing of the sample. Particle sizes: HAP—0.25 µm; Latex—0.108µm. Ratio of HAP:Latex is 1:2.3.
1451 cm1425-1465-1 Latex cm-1 BandLatex Band
20
18
16
14
12
10
8
6 Integrated Band Intensity Band Integrated 4
2
0 0 102030405060708090100 20 % Latex in HAP
y = 0.2095x - 0.2764 2 15 R = 0.9974
10
5 Integrated Band Intensity Band Integrated
0 0 102030405060708090100
-5 % Latex in HAP
136 Figure 5.12: HAP and Latex (HAP band). Uncertainties range from 6-30% RSD. Particle sizes: HAP—0.25 µm; Latex—0.108µm. Ratio of HAP:Latex is 1:2.3.
25 1020 cm-1 HAP band in Latex
20
15 Corrected Area 10 Integrated Peak Intensity Intensity Peak Integrated
5
0 20 0 102030405060708090100 %Latex in HAP 18
16
14 y = -0.2441x + 17.849 R2 = 0.9818 12
10 Corrected Area 8 y = -0.0733x + 9.9067 6 R2 = 0.9718 Integrated Peak Intensity Intensity Peak Integrated 4
2
0 0 102030405060708090100 %Latex in HAP
137 The 1451 cm-1 latex absorption was chosen due to its lack of interference with COM or HAP absorptions. Latex absorptions residing at lower wavenumbers are in close proximity to the 778 cm-1 absorption of COM or the 1020 cm-1 absorption of HAP, while latex absorptions at higher wavenumbers are in close proximity to the 1618 cm-1 absorption of COM. Figures 5.9 and 5.11 depict the same spectral region of latex absorption for two different mixtures. The curve pertaining to a large particle/small particle mixture is non- linear, with three separate areas of linearity (Figure 5.9). The small particle/small particle mixture (Figure 5.11) is linear, with an R2=0.9568. A hypothesis could be formed stating that large particle/small particle mixtures will result in a non-linear calibration curve, while same-size mixtures result in linear calibration curves. This hypothesis is investigated below. In examining this hypothesis, surface area, particle size, and the associated depth of penetration are all considered. 5.3.1 Surface Area and Particle Size The surface area of the particles in question is important in order to account for relative sizes, more so than simply taking into account the diameters of said particles. Additionally, increased surface area possibly provides increased reactivity in the renal system with urine in real-world samples. It is therefore important to consider these measurements in the standards analyzed here for applicability towards renal crystals in stone formation. Looking at the surface area of the two main components of interest, a large discrepancy is observed. A COM particle, ranging from 150-200 μm in diameter, has a surface area ranging from 70,686-125,663 μ2, while HAP (diameter ~0.25 μm) has a surface area of ~0.196 μ2. Surface area was calculated using Equation 5.2 below.
Surface Area= 4πr2 Equation 5.2
Where r is the radius of the particle. Calculations show that approximately 360,000 HAP particles are necessary to equal the surface area of a single COM particle. The conservative estimate of 360,000 was determined by dividing the surface area of HAP (0.196 μ2) into that of COM (70,686 μ2). The values of the surface areas assume all
138 particles of HAP have a diameter of 0.25 μm, and all COM particles have a diameter of 150 μm. Therefore, these values are merely approximations. However, the point to be made is that the surface area of COM is drastically larger than that of HAP. For sample pellets containing a large diameter component (such as COM), the evanescent wave from the IRE (a 200 μm diameter surface, similar in size to COM) may be detecting only the larger component, leading to inaccurate data. Figures 5.13 and 5.14 depict the interaction of the IRE with mixtures containing HAP and COM at differing concentrations of large and small components. Wadayama has documented this phenomenon by observing aggregated silver nanoparticles (~15 nm radius) using a Ge IRE. [178] In his research, the concentration of octadecanethiol infused into the presence of silver nanoparticles was altered, changing the aggregation of silver nanoparticles and proving that large areas of a single component—whether due to particle size or aggregation—increases the integrated peak intensity when using ATR.
139 Figure 5.13: The unequal distribution of large particles in a large particle/small particle mixture.
HAP (0.05- 0.25μ) or KCl (0.5-8μ)
COM (150-200μ) IRE Surface
CaOx 75x larger than HAP CaOx 75x larger than HAP
140 Figure 5.14: A small particle/large particle mixture where small particles greatly outnumber the larger particles.
HAP (0.05- 0.25μ) or KCl (0.5-8μ) CaOx (150- IRE Surface 200μ)
141 If mixtures containing a large surface area (such as COM) particle are present, the evanescent wave of the IRE may sample a non-representative portion of the mixture. [178] This factor could help explain the non-linearity observed in previous calibration curves. For this hypothesis to hold true, the following trends should be observed. All mixtures consisting of only large particles (>100 µm) should exhibit non-linear characteristics due to heterogeneous dispersion. Realistically, any diameter over 5 µm
may be considered large since the effective dp (Section 5.3.2) is less than 5 µm, meaning any additional increase in diameter does not increase absorbance. Mixtures of large particle (both greater than 100 µm) components should appear homogeneous from a macro perspective, but are in reality heterogeneously mixed. The IRE sampling
accessory is micro in character, meaning that the dp subsequently comes into unequal contact with particles that may absorb the entire evanescent wave, and that the calibration curve should be non-linear. Alternately, all mixtures consisting of large particles (>100 µm) and small particles (<5 µm) should also be non-linear because of the large-particle characteristics and heterogeneous character of the mixture. Even the smallest concentration of a large particle component can interfere with the absorption of the evanescent wave by the small particle component. Lastly, all mixtures where both components are less than 5 µm in size should be linear due to homogeneous mixing and equal dispersion in the path of the evanescent wave. Figures 5.15 and 5.16 illustrate the results of COM and HAP combined with uric acid (1-20 μm), considered large particle/large particle and small particle/large particle mixtures, respectively.
142 Figure 5.15: COM and UA (large particle/large particle). Uncertainties range from 2- 17% RSD due to slight inhomogeneity in the sample. Particle sizes: COM—200 µm; UA—~10 µm. Ratio of COM:UA is 1:20.
2.5 715-750 cm-1 UA in COM
2
1.5
1 Corrected Area Corrected 0.5
Integrated Peak Intensity Intensity Peak Integrated 0 0 102030405060708090100 % UA in COM
2
1.5 y = 0.0249x - 0.3 R2= 0.9965 Corrected Area Corrected 1 Integrated Peak Intensity Intensity Peak Integrated 0.5
0 0 102030405060708090100 % UA in COM
-0.5
143 Figure 5.16: HAP and UA (small particle/large particle). Uncertainties range from 9- 77% RSD (77% RSD at point 10% UA) due to large inhomogeneity in the sample. HAP—0.25 µm; UA—~10 µm. Ratio of HAP:UA is 40:1.
2.500 715-750 cm-1 UA in HAP
2.000
1.500
1.000 Corrected Area Corrected
Integrated Peak Intensity Intensity Peak Integrated 0.500
0.000 10 20 30 40 50 60 70 80 90 100 2.500 %UA in HAP
2.000
y = 0.022x + 0.0043 R2 = 0.9804 1.500
Corrected Area 1.000
0.500 Integrated Peak Intensity Intensity Peak Integrated 0.000 0 102030405060708090100 %UA in HAP
144 As can be seen, both calibration curves are linear and therefore do not support the hypothesis, however, the large uncertainty in the HAP/UA mixture may be the result of the particle diameter discrepancy leading to more heterogeneous mixtures than other samples. The first mixture, COM and UA, should exhibit non-linear characteristics due to the sizes of the particles, which it does not. However, the linearity of the HAP/UA mixture does support the above hypothesis. For both mixtures, 5 samples per data point were used. Mixtures that do support this hypothesis are COM/KCl, COM/HAP, COM/Latex, and HAP/UA. All four mixtures, all large particle/small particle, contain non-linear characteristics. Mixtures detracting from this hypothesis include HAP/KCl, HAP/Latex, and COM/UA, which are small particle/small particle, small particle/small particle, and large particle/large particle mixtures, respectively. The small particle/small particle mixtures would ideally depict a linear calibration curve implying that the infrared radiation is in uniform contact with the proper amount of each component. However, the small particle/small particle mixtures have non-linear characteristics. The large particle/large particle calibration curve of COM/UA is expected to be non-linear, though in reality it is linear, with an R2 value of 0.9965. The accuracy of the particle sizes was confirmed using the SEM. However, the method of disbursing the analyte onto the metal SEM stub included the dry dusting of the powder onto a carbon tab. A gold coating was subsequently sprayed onto the dried particles in order to provide conductivity of the sample for analysis. This method of SEM sample preparation has recently been termed inaccurate due to particle aggregation. [189] Therefore an alternate method of sonicating and suspending the particles in alcohol prior to deposition onto the carbon tab will be used in the future, in order to relieve particle aggregation. Using the above data, it can be seen that particle size alone is not responsible for the non-linear characteristics exhibited by the COM/HAP calibration curve.
145 5.3.2 Depth of Penetration and Particle Size
The depth of penetration (dp) by the evanescent wave was briefly mentioned in the last section, and has been functionally discussed in Chapter 1. Equation 1.5
illustrates that dp is dependant upon the wavelength, angle of incidence, and the refractive
index of the crystal. Additionally, dp is also dependant upon the refractive index of the
sample (ns). However, the ns in Equation 1.5 cannot be used without modification due to the fact that each component in the sample has a different refractive index. In order to estimate the penetration depth of the evanescent wave into mixed component samples, adjusted refractive indices of the samples were calculated. To
determine ns for the COM/HAP mixture, the weight concentration of each component was multiplied by the refractive index of that component and added together. This concept is explained by Equation 5.3 below:
(wt% HAP*1.64)+(wt% COM*1.48)=ns Equation 5.3
Using Equation 5.3, the refractive index of the sample as a whole is altered to reflect the
weight concentration equivalents present in the sample ratio. nCOM=1.49, nHAP=1.64.
Using Equation 1.5 for a 1% COM in HAP mixture, a dp of 0.68 μm is determined -1 for the 1618 cm absorption. The effective sampling depth is equal to 3*dp, or a value of -1 -1 2.04µm. [127, 140] For the 778 cm absorption, 3dp=4.2μm, and for the 1318 cm
absorption, 3dp=2.49μm. Using the value of the effective sampling depth as 2.04 μm for an example, several conclusions about the evanescent wave and its interaction with the sample can be drawn. When the effective sampling depth is divided by the diameter of an HAP particle (2.04 μm/0.25 μm), it can be concluded that 8.16 layers of HAP are needed in order to fully eclipse the 150 µm-diameter COM present in the mixture. (Figure5.17) Using the density of HAP as 3.15g/cm3 [190], it is determined that 8.16 layers of 0.25 µm-diameter HAP corresponds to only 5.1 % HAP by weight, which appears to be a very small percentage. However, when the effective sampling depth is divided by the diameter of a COM particle (2.04 μm/150 μm), it can be concluded that 0.014 layers of COM (1/70 of a single COM particle) are necessary in order to eclipse all HAP particles. 0.014 layers of
146 150 µm-diameter COM corresponds to 0.03 wt% COM using a density of 2.02 g/cm3. Considering the calculations presented, it would be expected that even the smallest presence of a large particle component can eclipse small-diameter component contribution.
147 Figure 5.17: The penetration of the evanescent wave through over 8 layers of HAP.
HAP (0.05- 0.25μ) or KCl (0.5-8μ)
COM IRE Surface (150- 200μ)
dp
148 If the hypothesis pertaining to the dp is correct, the uncertainty in the calibration points for mixtures of large and small components should be larger than the uncertainty in the small/small component mixtures. The % RSD of the integrated band intensity for a COM/HAP mixture ranges from 1-16%, and for the COM/Latex mixture, the % RSD is from 3-24%. However, for the HAP/Latex mixture (small-small particle), the % RSD ranges from 0.1-2%, which is much smaller than that of the COM/Latex mixture and COM/HAP measurements. Therefore, it appears that the penetration depth of the evanescent wave in conjunction with the sizes of the particles present is partially responsible for the non-linear character of the calibration curves observed. Supporting the fact that penetration depth is only partially responsible for the non- linear nature of the data are observations by Kortüm. In Kortüm’s discussion on the effects of particle size on calibration curves, he states that as the particles decrease in size, the penetration depth of the infrared radiation in a reflectance process diminishes, lessening the intensity. [191] His conclusion is based upon the presence of monodispersed spherical particles such as HAP, which were included in his study. Kortüm’s observations were conducted using particle sizes from 1 µm to 200 µm in diameter; unfortunately, all particles in the samples analyzed were approximately the same diameter, so no mixtures of particle sizes were measured. Also, measurements were not made with respect to analyte concentration. Instead, his study of non-linear curves focused on the changes in the actual spectrum as the total particle size of the sample decreased. Observing Figure 5.6 both before and after grinding, it can be seen that HAP is approximately spherical in character. However, a drawback to the applicability of Kortüm’s hypothesis to the data presented here is that his conclusion is drawn primarily for particles undergoing transmission or DRIFTS-type analysis, since ATR was not a common technique or well established at the time of his writing [192], though the fundamentals are the same. Using the example of HAP and COM, Kortüm’s hypothesis would imply that an increase in HAP concentration should decrease the penetration depth by the evanescent wave, lessening the absorbance. Though the particle sizes of the individual components is constant, the “average” particle size of the mixture will be changing as the concentrations
149 change. Therefore, the absorbance of the overall mixture will change with concentration. Following Kortüm’s hypothesis, the absorbance intensity will decrease as the HAP concentration increases, generating a more gradually sloping curve. Observing the data provided in Figures 5.3a-c, Kortüm’s hypothesis appears to be an accurate statement. Looking strictly at the COM absorption band, when the concentration of HAP is increased (percent of COM is decreased), the absorption of the oxalate band decreases. However, since we are observing the COM band as COM concentration decreases, it is a natural occurrence for the absorption band to decrease. For Kortüm’s statement to hold true, the HAP band must experience a decrease in slope as the concentration of HAP increases in the presence of COM, indicating a change in extinction coefficient ε. Figure 5.18 illustrates the 1020 cm-1 absorbance band corresponding to the orthophosphate stretch of hydroxylapatite as HAP concentration increases in the presence of COM.
150 Figure 5.18: Absorbance of the HAP orthophosphate stretch as HAP increases in concentration for a HAP/COM mixture. Uncertainties range from 3-36% RSD due to size and concentration inhomogeneity in the sample.
-1 1020 cm HAP in COM 12
10
8
6 Corrected Area Corrected 4
Integrated Peak Intensity Intensity Peak Integrated 2
0 0 102030405060708090100 %HAP in COM 12
10
y = 0.0876x + 1.215 8 2 R = 0.9407
6 y = 0.0626x + 0.1504 Corrected Area R2= 0.9821
Integrated Peak Intensity Intensity Peak Integrated 4
2
0 0 102030405060708090100 %HAP in COM
151 The plot shows that not only does the absorbance grow linearly, but the slope of the linear absorption increases slightly towards the highest concentrations of HAP. Therefore absorption related to the small particle concentration is not solely responsible for the non-linear calibration curve character of the data. The refractive index is related to the depth of penetration also. In order to test whether the changing refractive index of the sample initiates a non-linear penetration depth, leading to subsequent non-linear calibration curves for binary mixtures, calculations were performed using Equation 1.5. The thought would be that as the concentrations of HAP and COM change, the refractive index also changes according to Equation 5.3. If this change in refractive index causes a non-linear character in the penetration depth in the region of a major absorbance band, then the non-linearity may be more readily accounted for. Using the refractive index of the Si crystal as 3.4 and refractive indices of COM and HAP as 1.49 and 1.64 (respectively), depth of penetration charts were produced for specific wavelengths over the concentration range of COM in HAP. Figure 5.19 displays the depth of penetration for the 1318 cm-1 band of COM over a range of concentrations of COM in HAP.
152 Figure 5.19: Depth of penetration for the 1318 cm-1 COM band in HAP.
dp for COM in HAP
0.84
0.83
0.82 y = -0.0008x + 0.8251 0.81 R2 = 0.9974
0.80
0.79 dp
0.78
0.77
0.76
0.75
0.74 0 102030405060708090100 % COM in HAP
153 As can be seen from Figure 5.19, the penetration depth is linear with respect to the 1318 cm-1 absorption of COM. Therefore, the change in refractive index of the sample is not solely responsible for the non-linear characteristics. Due to these results, it was concluded that factors other than the depth of penetration, refractive index, and particle size must be included in the non-linear calibration curve analysis. 5.3.3 Other Considerations 5.3.3.1 Quadratic Equations In the investigation of Cohen-Solal and Daudon et al, automated identification and quantitation of renal stone components using a DRIFTS/KBr pellet process is performed using COD and carbapatite, generating a theoretical calibration curve from 3 data points. [134] The mathematical generation of their data provided a non-linear calibration curve possessing a quadratic character. The generated curve used the height of the carbapatite band divided by the combined heights of the carbapatite and the COD plotted against the percent of carbapatite. By doing this, the sensitivity of the method for carbapatite peak height is measured as the concentration is increased, effectively yielding a calibration curve comparing relative peak height to concentration. However, this measurement was carried out using a spectral calculator, not true data points. When this same principle is applied to the data obtained in this research, Figure 5.20 is produced. Here, the experimental area of the COM 1318 cm-1 absorption is divided by the experimental area of the combined HAP 1020 cm-1 and COM 1318 cm-1 absorptions, generating a quadratic calibration curve using Microsoft Excel trend line fitting software with an R2 value of 0.9754 (red triangles). When a quadratic equation is fitted to the data in Figure 5.3b, an R2 value of 0.9754 results (blue diamonds). The data from Figure 5.3b looks strikingly similar to that of Cohen-Solal et.al.
154 Figure 5.20: Quadratic trend lines applied to the data.
1318 cm-1 COM/(COM+HAP) 1318 cm-1 COM/ (COM+HAP) area
0.9 2 COM/(COM+HAP) y = -5E-05x + 0.0118x + 0.0864 2 -1 R = 0.9754 0.8 COM 1318 cm from 5.3b
0.7
0.6
0.5
0.4
Integrated Band Intensity 0.3 Integrated Peak Intensity Intensity Peak Integrated
0.2 2 y = -2E-06x + 0.0004x + 0.003
0.1 2 R = 0.9754
0 0 102030405060708090100
% COM in HAP
155 The general form of a quadratic equation is y = ax 2 + bx + c , where y is the known area of the sample and x is the wt% COM. Using the quadratic formulas provided from Figure 5.20 and the COM area value (y) of the artificially generated sample described further in Section 5.5.1 of this chapter (0.8257 for the 1318 cm-1 absorption, and 0.3284 cm-1 for the 778 cm-1 absorption), x values (wt% COM) were calculated using Equation 5.4.
− b ± b 2 − 4ac x = Equation 5.4 2a
Initially, the generated quadratic equation must be set to 0 by either adding or subtracting the y value (0.8257 and 0.3284) to the right side of the equation. Doing this for the quadratic equation relating the integrated band intensity of the COM to the wt% COM (blue graph in Figure 5.20, obtained from Figure 5.3b) for y=0.8257 (1318 cm-1 absorption), the equation becomes: 0= -5x10-5x2+0.0118x -0.7393 Substituting the values into Equation 5.4, the term (b2-4ac)1/2 becomes negative, and the square root of a negative number is imaginary. Using the same principle for the 778 cm-1 absorption of COM (Figure 5.3a), the equation y=-5x10-5x2+0.0158x +0.0769 (R2=0.9915) is observed. Using y=0.3284, a value of x= 658 wt% COM is determined. Therefore it is determined that a quadratic fit to the Figure 5.3 data of integrated band intensity vs. wt% COM is not a viable solution. Using the quadratic equation associated with the 1318 cm-1 COM intensity divided by the [1318 cm-1 COM+ 1020 cm-1 HAP] areas modeled after Cohen-Solal [134] (Figure 5.20, red, where y=0.8257) and Equation 5.4, an imaginary value results for the wt % of COM in HAP. When a similar quadratic equation is generated for the 778 cm-1 absorption divided by the [778 cm-1 COM+ 1020 cm-1 HAP] areas, a quadratic equation of y=0.0001x2-0.0046x+0.0601 is observed (R2=0.9596). Using y=0.3284, a value of x=173.5 wt% COM is determined. Therefore it is determined that a quadratic fit to the Cohen-Solal style data of COM integrated band intensity divided by added COM and HAP integrated band intensities is not a viable solution for an accurate wt % of
156 COM. While there is visual similarity between the data generated here and that of Cohen-Solal, visual inspection is misleading, and the current data is deemed inaccurate. 5.3.3.2 Component Crushing The report of Cohen-Solal investigates the amount of sample crushing necessary in order to generate spectra with reproducible peak heights. [134] It was found that 10 separate grindings and pressings on the same pellet were necessary in order to achieve reproducible spectra. Unfortunately, no particle sizes either pre- or post-crushing are given for the components investigated, therefore, it is impossible to draw parallels in size between the compounds used in the published report and those presented in this chapter. It is, however, Dr. Daudon’s assertion that the particle size is the primary limitation to achieving a linear calibration curve. [187] He has stated that, once small particles are combined with other small particles, the slope of the calibration curve becomes reproducibly linear. However, the mixture of HAP (d=0.25 μm) and latex beads (d=0.108 μm) exhibited in Figures 5.11 and 5.12 produces non-linear calibration curves, raising questions concerning Dr. Daudon’s statement. It is unclear if such a method, though perhaps true for DRIFTS measurements, would hold true for other such infrared techniques. Future work in this laboratory will focus on either proving or disproving Dr. Daudon’s hypothesis.
5.4 Reproducibility The reproducibility of the ATR and the sample preparation method were tested using two methods. The first method observed the % RSD of five different pellets analyzed within each ratio. Table 5.3 presents the data concerning the COM/HAP mixture.
157 Table 5.3: Uncertainty within the COM measurements, observing the corrected area of the 778 cm-1 absorption band of COM.
158 As can be seen, the % RSD varies from 1-11%. Overall, the homogeneity of the samples appears to be fairly consistent. The second method involved the creation of three batches of approximately 3% (3.04%, 3.06%, and 3.01%) COM in an HAP matrix. Six pressed pellets from each batch were analyzed, generating an average, standard deviation, and relative standard deviation. For each analysis, every attempt was made to ensure that the pellets were equal in size and compactness. Here, the reproducibility between batches was tested in an effort to qualify the method of sample preparation. Using the COM stretch at 778 cm-1, a relative standard deviation of 6.94% between the separate batches was determined. The 778 cm-1 absorption was chosen for these analyses because in real-world samples, the 778 cm-1 absorption is well isolated from the amide I and amide III bands of protein. However, using data from both the 1318 cm-1 and the 1618 cm-1 absorptions, averages and standard deviations were similar.
5.5 Determining Accuracy using the Non-Linear Calibration Curves 5.5.1 Artificial Sample In order to determine if the non-linear calibration curves generated in this research are accurate for quantitative measurements of samples, an unknown ratio of HAP and COM was mixed and analyzed. Three pellets were pressed, analyzed, and averaged from the unknown sample. Using the upper calibration segment of Figure 5.3b, a value of 31 % COM with an RSD% of 78% was determined. Using the lower calibration segment, a value of 21% COM was determined, containing an RSD% of 29%. The true percentage was 27% COM, calculated using the weights of each component in the sample after analysis had occurred. It was observed from the varied values and large RSD% that precision as well as accuracy are lacking for unknown samples. This method of testing the accuracy of the generated calibration curve has also been employed by Cohen-Solal and Daudon [134], though in the published report, it was used as a test of the of the accuracy for the theoretical calibration curve. The quadratic calibration curves used in Section 5.3.3 modeled after Daudon’s work produced imaginary percent COM values due to taking the square root of a negative number.
159 5.5.2 Real-World Sample Using the renal stone exhibited in Figure 1.2, a real world sample was attempted for analysis. Three separate points on the renal stone were analyzed and averaged. The upper calibration curve of Figure 5.3b was utilized since visual and molecular image examination of the renal stone indicated a COM percentage above 20 %. The upper calibration equation yielded a value of 107 % COM. The true value is unknown, but is thought to be near 100 % COM from visual inspection of the molecular image (Figure 1.2). It was therefore concluded that the non-linear calibration curves generated in this research are not quantitative in character but instead are merely very rough approximations of the true value.
5.6 Conclusions Mixed-component renal stones are the majority of renal stones seen in this laboratory. In an effort to generate models for mixed-component analysis, homogeneous mixtures of powdered renal stone components were made and analyzed via ATR-IR microspectroscopy. Several non-linear calibration curves were the result of this analysis. Hypotheses for the non-linear character of renal stone calibration curves include contributions from the particle size, refractive index, and depth of penetration by the evanescent wave. Unfortunately, however, no single explanation appears to exist to account for the non-linear nature of both single and mixed-component systems. The fact that these calibration curves are not linear remains unaccounted for. It has also been established that the non-linear calibration curves produced in this research are not quantitative and should not be used in a quantitative manner. Further models and calibration experiments are needed in order to fully assess the optical and non-linear characteristics of the homogeneous mixtures presented in this chapter. Though the issue of non-linear calibration curves has been mildly addressed in the literature, the majority of analyses pertain to transmission spectroscopy or DRIFTS methods—not ATR.
160
CHAPTER 6
Investigating the Interface between Tissue and Mineral Portions of a Renal Biopsy using Infrared Microspectroscopy
161 6.0 Introduction The evolution of renal stones remains unexplained by the scientific and medical community. From nano-bacteria [6, 193, 194] to extracorporeal shockwave lithotripsy (ESWL) treatment [53-55, 195-197], the suspected causes of renal stone formation vary by circumstance, region, diet, and genetics. [17, 33-35, 106, 198-203] Recently, evidence has been presented for one specific method of renal stone formation: idiopathic calcium stone formers (ICSF). [130] Though characteristics of ICSF’s such as diet, Randall’s plaque coverage, and stone composition have been studied previously [11, 204-207], never before has an in-depth study been performed on the mechanism of stone formation as that currently being pursued. [130] The research presented in this chapter is complementary to studies led by Drs. A. Evan, J. Williams, and F. Coe, examining the microscopic interactions between Randall’s plaque, tissue, and urine, and providing ground breaking data on the evolution of calcium oxalate renal stones formed from Randall’s plaque. The infrared data provided in this chapter aids in the determination of the components present, as well as raises several new questions and observations. Drs. Evan, Williams and Coe have, over the past decade, published numerous works on the chemical interactions and crystal morphologies associated with renal stones and their development. [9, 11, 28, 33, 35, 56, 105, 106, 130, 208-212] One of the seminal papers details an investigation into the formation of Randall’s plaque via a unique metabolic pathway for idiopathic calcium stone formers (ICSF’s), and the resultant calcium oxalate stones generated. [11, 28] Evan et. al. have found that ICSF’s develop Randall’s plaque (explained further in section 6.1) in the thin loops of Henle, which are thin tubules in the kidney. Randall’s plaque, a phosphate-based mineral deposit, does not normally form in the thin loops of Henle for renal stone formers. On the contrary, renal stone patients that form stones as a result of bypass surgery, for example, generate physical crystals (as opposed to plaque) in a region of the kidney termed the Ducts of Bellini, and non-stone formers generate Randall’s plaque, but not in the thin loops of Henle. Therefore, the formation of Randall’s plaque in the thin loops of Henle evolving to calcium oxalate renal stones is a unique occurrence. [28]
162 Of special interest is the interaction occurring between the phosphate-based suburothelial plaque and the produced calcium oxalate stones. [130] Though hydroxylapatite plaque is prevalent in the thin loops of Henle, no calcium oxalate is observed, and the mechanism by which calcium oxalate renal stones are formed from a hydroxylapatite initiation point remains uncertain. [28] Therefore, recent studies have included the structurally fragile interface between the tissue containing Randall’s plaque and the mineralized portion of an oxalate renal stone. In this thin region, optical microscopy has been used to observe the interactions occurring in the urine environment [130], while infrared analysis is used to make a qualitative diagnosis of the present materials and is presented in this chapter. The methodology used to acquire infrared data of the tissue/mineral interface for the current research is similar to the examination of mineral inclusions embedded in tissue sections. [43, 46] Using reflection/absorption infrared (RAIR) microspectroscopy in order to produce an infrared image of the sample surface, general qualitative determinations can be made regarding the composition and morphology of the mineral component(s) present. Line scans, traversing from the tissue portion of the kidney to the mineral portion of the stone, yield data that display changes as the tissue is exposed to, and interacts with, the urine. The methods utilized in this chapter have been previously utilized in research from this laboratory. [43, 46] For the current research, however, the focus of the technique is oriented towards the determination of a specific region of sample: the delicate and complex interface existing between the proteinacious tissue and the mineralized stone sections of a biopsy. One of the results of this research is the observation of band broadening at the interface between the tissue and crystal material. This band broadening effect has been observed elsewhere in the literature, though rarely. [213] Aizenberg found that organic
species inhibit the formation of crystalline CaCO3, causing IR band broadening. Currently it remains unclear as to the cause of the band broadening observed in the present research, however, a plausible explanation is the existence of non-crystalline, amorphous hydroxylapatite due to the presence of organic material. This hypothesis is examined in subsequent sections.
163 6.1 Stone Former Characteristics Idiopathic calcium stone formers (ICSF’s) make up the majority of renal stone patients around the world. An ICSF is generally identified as a recurring stone former who produces—almost exclusively—calcium oxalate (COM) stones originating from Randall’s plaque. Randall’s plaque, a hydroxylapatite (HAP) mineral layer formed in renal tissue, is displayed in Figure 6.1. An average person has approximately 1-2% of their renal papillary tissue surface covered in Randall’s plaque. However, ICSF patients have 10-20% coverage. [12, 214] As opposed to other types of recurrent renal stone formers, the Ducts of Bellini (Figures 6.2 and 6.3) are not plugged with mineral material in ICSF patients.
164 Figure 6.1: Randall’s Plaque on papillary renal tissue (stars) near renal stones (arrows) composed of calcium oxalate. Randall’s plaque forms in the Thin Loops of Henle, shown in Figure 6.2. Obtained from Evan, et. al.[11]
165 Other calcium stone formers, such as renal bypass patients (Chapter 1), form calcium oxalate stones without the presence of Randall’s plaque, but plug the Ducts of Bellini (Figure 6.2) with mineral material. Calcium phosphate stone formers (forming brushite and hydroxylapatite renal stones) form both Randall’s plaque as well as mineral plugs in the Ducts of Bellini.
166 Figure 6.2: Diagram of the Ducts of Bellini and the Thin Loops of Henle. (Non- copyrighted public domain material. Originally from Grey’s Anatomy of the Human Body,[52])
167 Figure 6.3: Orientation of areas of interest (Figure 6.2) in the kidney. Artwork from Brenner & Rector’s The Kidney, 2004. [215]
168 6.2 Experimental 6.2.1 Materials Tissue biopsy samples were obtained from the Indiana University Medical School with the informed consent of the patients and were prepared by a certified histologist. The patients varied in age, symptoms and backgrounds and are described in a recent study by Evan et.al. [28] The unstained tissue sections on the low-E substrates were imaged using the visible CCD camera and frame grabber on the Spectrum Spotlight. A serial section stained with Yasue silver replacement stain was employed as a control to confirm the location of the calcium. [216] This serial section was reviewed using a standard visible microscope in an attempt to visually determine the general areas of interest. Tissue sections from patients with idiopathic calcium oxalate kidney stone disease were analyzed in the present study. Papillary biopsies were obtained at the time of percutaneous nephrolithotomy from a group of well-characterized idiopathic calcium oxalate stone formers. 6.2.2 Instrumentation All samples were analyzed using a Perkin-Elmer Spectrum Spotlight 300 infrared imaging microscope equipped with an array detector for the rapid acquisition of infrared images and a single point detector for the acquisition of high signal to noise spectra. Both detectors are based on the well-established mercury cadmium telluride (MCT) technology. The minimum sample size that can be analyzed using either detector is approximately two wavelengths, 5-28 µm corresponding to 2.5-14 µm wavelengths, respectively. [165] Many of the spectra presented in this report were collected using the single point detector and represent the average of 64 individual scans collected at a spectral resolution of 4 cm-1. The microscope can be operated in transmission, reflection or attenuated total internal reflection (ATR) modes. In this latter mode, a drop-down Ge internal reflection element (IRE) was employed. A confocal aperture was employed to isolate the sample region of interest for the transmission and reflection modes. The same aperture was employed for the ATR mode, however, the Ge IRE provides an additional 4X magnification resulting in a sampling area four times smaller on the surface of the sample
169 than the actual aperture size. For example, a 40 μm aperture translates into a 10 μm spot size on the surface of the sample. Infrared images of sectioned tissue/mineral junctions were collected on the Perkin-Elmer Spectrum Spotlight 300 infrared imaging microscope using the linear array MCT detector. Spectra collected and extracted from false-color images using this detector represent the average of 8 scans at a spectral resolution of 8 cm-1. Each spatial element on this detector represents 6.25 x 6.25 μm on the sample. Background spectra were collected using the reflective side of a low-E slide. The substrate used in this investigation was a low-E glass slide (Kevley Technologies, Chesterland, Ohio). [162] The slides have the same dimensions as those of standard glass microscope slides, making the mounting of tissue samples routine for histologists. Visual microscopy was performed on an Olympus BH-2 microscope interfaced to a JVC TK-128 OU video camera. The objective most often used was the Olympus 50x objective, though 10x, 20x, and 40x were also utilized.
6.3 Discussion A papillary tissue biopsy containing in-tact mineral components of a renal stone still attached to tissue is a rare sample type due to the delicacy of the procedure and the fragility of the sample. For the research conducted in this chapter, several ICSF patient tissue biopsies were obtained and cross-sections of these biopsies analyzed at the interface between mineral and tissue components in order to determine the components present. Figures 6.4 and 6.5 illustrate visible images of a tissue-mineral boundary. Purple/darker areas are stained tissue portions, while the white crystalline substance is the stone.
170 Figure 6.4: A visual image taken in this laboratory of a tissue-mineral boundary at 50x magnification. The purple areas are indicative of tissue, while the white areas indicate renal stone or mineral inclusions.
171 Figure 6.5: A visual image of a tissue-mineral boundary at 50x magnification. The purple areas are indicative of tissue, while the white areas indicate renal stone or mineral inclusions.
172 The extremely thin boundary between the mineral and tissue portions of the sample contains damaged epithelial walls. It is in these areas that plaque becomes exposed to the urine environment, and an interaction takes place that is still under investigation. (Figure 1.1) Figure 6.6, courtesy of Dr. A. Evan, illustrates this epithelial degradation and formation of mineral layers. Here, the epithelium is indicated by arrows. Breaks in the epithelium can be seen in both Figures A and B. In both cases, where the epithelium has been thinned or lost, Randall’s plaque can be seen interacting with the urine.
173 Figure 6.6: Microscopic view of the thin epithelium layer of the kidney (arrows). Where the epithelium layer has been lost (A) or thinned (B), Randall’s plaque can be seen breaking through to interact with the urine. Photo courtesy of Dr. A. Evan.
174 As Figures 6.4-6.6 and the previous discussion suggest, there is an interaction occurring at the boundary between the formed renal stone and the papillary tissue. An infrared image (Figure 6.7) of a similar sample type yields spectra useful in the determination of this boundary interaction.
175 Figure 6.7: Absorbance infrared image of the tissue/stone interface based on the 1020 cm-1 absorption of HAP.
Slide
Renal Stone Material
Renal Stone / Tissue Interface Tissue
176 Hydroxylapatite spectra obtained from the interface portion of the biopsy look slightly different from hydroxylapatite spectra collected from the mineral portion of the sample. As can be seen in Figure 6.8, the interface spectrum has a much broader orthophosphate band at 1020 cm-1 than the HAP obtained from the stone portion (red and green areas of Figure 6.7). The thin boundary, not even a micrometer in width in certain places, is difficult to collect a pure spectrum from. This difficulty arises due to the achievable spatial resolution of the infrared microscope; because of the diffraction-limited spatial resolution (Equation 1.6—2λ/0.6= ~33 µm for 1020 cm-1) of the instrument, the spectrum from the thin boundary region will nearly always have the influence of the protein. Fortunately, the presence of protein does not alter the location of the interface absorptions since HAP and protein have no overlapping spectral absorptions. Protein will, however, alter the shape, as demonstrated by Aizenberg. The broad features of the interface could originate from the presence of disordered crystalline hydroxylapatite, where disordered—often termed amorphous— crystals of HAP are present due to the presence of organic material such as protein, increasing the width of the absorption. Aizenberg has documented this mechanism in
CaCO3, finding that the presence of protein inhibits the growth of crystalline CaCO3
(making amorphous CaCO3) and broadens IR bands. [213] Upon her examination of
skeletal tissue, Aizenberg found that amorphous CaCO3 was formed prior to the
formation of crystalline CaCO3, implying that the evolution of crystals begins in an amorphous state. This same explanation pertains to the tissue biopsies observed in this research, where HAP would be in an amorphous form as it interacts with the proteinacious tissue at the urine/tissue boundary, subsequently forming crystalline HAP that eventually grows into a larger renal stone. Figure 6.8 displays hydroxylapatite spectra obtained from the tissue portion (top), the interface portion (middle), the stone portion (3rd) of the biopsy, and (bottom) a crystalline HAP sample from a commercial source. Notice a general narrowing of the HAP orthophosphate absorption centered at 1020 cm-1 as the site of analysis migrates from the interface to the stone portion. Spectra were collected using the R/A process. The orthophosphate absorption of the interface was measured to be more than twice the
177 FWHM of the HAP obtained from the stone portion, with FWHM’s of 299 and 116 cm-1, respectively, using GRAMs spectral software (Galactic).
178 Figure 6.8: FT-IR spectra of the (top) tissue, (middle) interface, (3rd) stone portion, and (bottom) a commercial sample of HAP. The top three spectra were obtained from a molecular image of a renal tissue biopsy similar to that in Figure 6.7.
Tissue
%T Interface
Stone
Crystalline HAP (Commercial Source)
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 cm-1
179 The narrowing of the HAP absorption as the site of analysis moves away from the interface portion in the sample is most likely due to the physical state of the HAP. Though the interface between the tissue and the renal stone is difficult to observe, high-resolution optical microscopy provides a magnified view of the tissue/stone interface. Optical microscopy will always have better spatial resolution than infrared microspectroscopy since optical microscopy uses shorter wavelengths than infrared. Figure 6.9 illustrates several portions of material where epithelium cells have been removed and Randall’s plaque has been exposed to the environment of the urine. A model of this process is shown in Figure 1.1 of Chapter 1. These images, collected by Dr. A. Evan, Dr. F. Coe, and Dr. J. Williams, Jr., depict the interface as has never been viewed before. [130]
180 Figure 6.9: Five layers of protein and stone material where epithelium cells have been removed and Randall’s plaque has been exposed to urine. Collected and presented at the International Urolithiasis Research Symposium by Drs. A. Evan, F. Coe and J. Williams, Indianapolis, 2006. Used with permission. A and B—A complex ribbon of urine material and plaque material. Arrows in A and B—Randall’s plaque as the outer ribbon layer and crystals in the urine that are attracted to the surface of the ribbon.
181 The alternating bands of protein (white) and mineral (black) indicate several changes in urine chemistry, and are reminiscent of the alternating concentric layers seen in many renal stones. On the outside of the tissue/Randall’s plaque area in Figure 6.9, several crystal structures can be seen. It is thought that these crystals, made of hydroxylapatite, are failed attempts at renal stone formation. [130] A more ordered crystalline sample will yield sharper spectral bands. [213]
Observing Aizenbergs data for CaCO3, decreased crystallinity was manifest as an increase in the FWHM of the main absorption located near 1420cm-1. Evidence for the sharpening of the main hydroxylapatite absorption in the current research arises from FT- IR studies performed on pre- and post-heated commercially available HAP standard samples. Prior to heating, the IR spectrum appears to have a larger FWHM (top— FWHM=54.7±8.21) than after heating; post heating for 5 days at 950 C, the HAP spectrum is sharper and well-defined (bottom—FWHM=50.9±7.64). This effect is illustrated in Figure 6.10. As can be observed from the overlap of the FWHM ranges of the two samples, there is no statistical difference between the widths of the peaks. However, resolution is greatly impacted, as is observed with the increased visibility of -1 3- the shoulder located near 1090 cm , defined by Daudon as the vas (PO4) stretch [51]
(also known as the v3 stretch [217]); the entire band envelope increases in resolution, indicating an increase in crystallinity, and includes the absorptions near 970, 1020 and 1090 cm-1. (Note—uncertainties were determined by using GRAMs software to calculate the FWHM for each spectrum five times and their associated standard deviation.)
182 Figure 6.10: FT-IR spectra of heated and unheated HAP standard. Materials originate from the same commercial source.
Heated HAP
%T
Unheated HAP
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 700 cm-1
183 HAP, in a disordered form, becomes more crystalline upon heating as seen in Figure 6.10. This increase in crystallinity is because an ordered crystal state is at a lower energy level (more energetically favorable) than a disordered state. However, for crystal structures to become ordered, energy must be put into the system and an energy barrier must be overcome. [218] When the crystal system has no energy input, the crystals will not become more ordered, according to the second law of thermodynamics which states that a closed system will tend toward disorder. However, the renal system is not a closed system. Therefore, conditions are constantly in flux, meaning that under the correct conditions (i.e., the absence of protein), amorphous HAP may become crystallized in the renal system. In addition to decreasing the entropy of the HAP sample shown in Figure 6.10, heating can drive off excess water and other impurities, allowing the substance to become purer in nature. [219, 220] This is observed for the asymmetric stretch of water at 1650 cm-1 in Figure 6.10. Also due to heating, excess carbonate is driven off, and the absorbance of those bands at 870 and 1400 cm-1 is greatly decreased. A disordered HAP intermediate would be a plausible conclusion in the evolution of Randall’s plaque to a renal stone. Examples of disordered natural crystalline hydroxylapatite are difficult to attain. However, some examples do exist in dental literature. [221] No disordered hydroxylapatite appears to be present on the bone samples examined in our laboratory. However, on the interior portion of a stone section analyzed in this laboratory (Chapter 2), a possible site of disordered hydroxylapatite was found. Figure 6.11 illustrates the molecular image associated with the stone nucleus, as well as two spectra. The arrow in Figure 6.11 points to the proposed disordered crystalline nucleus (green vein), while a star denotes the ordered hydroxylapatite layer surrounding it (blue). The dashed spectrum, taken from the blue area denoted by the star, exhibits a strong, sharp hydroxylapatite peak; the solid spectrum, taken from the green vein pointed to by the arrow, exhibits a much weaker hydroxylapatite signal that appears much less defined than the dashed spectrum.
184 Figure 6.11: Possible site of disordered crystalline hydroxylapatite in a renal stone analyzed in this laboratory. Dashed spectrum: hydroxylapatite from the blue portion of the stone (star designation). Solid spectrum: possible disordered crystalline hydroxylapatite from the green vein (arrow designation).
*
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 cm-1
185 Further studies of the interface of tissue and stone materials conducted using Raman microspectroscopy at 514 nm, 633 nm, and 1064 nm have proven unsuccessful due to a lack of signal strength and/or interfering fluorescence. In addition to the discovery of hydroxylapatite at the tissue-mineral interface, calcium oxalate has also been found. Figure 6.12 illustrates a visual image of a tissue section containing mineral deposits, while Figure 6.13 displays a molecular image corresponding to a selected region of the upper right portion of Figure 6.12.
186 Figure 6.12: Visual image of a portion of tissue from the tissue-mineral interface. Purple areas are stained tissue, while white areas are mineral in character. Dashed box is Figure 6.13 area, though the optical resolution is better than that achievable using IR, so the large void (star designation) appears closed in in Figure 6.13.
Tissue
Mineral
*
187 Figure 6.13: Infrared image of a selected region of Figure 6.12. CaOx CaOx
HAP HAP
CaOx * Mixed Mixed
188 This sample, from an ICSF, was previously diagnosed as containing only hydroxylapatite. The discovery of calcium oxalate raises the question of additional interactions between the urine, hydroxylapatite, and calcium oxalate. Figure 6.14 presents spectra obtained from the molecular image in Figure 6.13.
189 Figure 6.14: Spectra obtained from the molecular image in Figure 6.13. Top: Pure protein. 2nd: HAP in a crystal. 3rd: possibly disordered HAP. 4th: CaOx from tissue edge with protein present. 5th: CaOx in crystallized form physically separate from tissue section. Bottom: CaOx and HAP co-existing, possibly disordered.
Protein
HAP Sharp
HAP Broad
%T CaOx w/ protein
CaOx crystal away from tissue
CaOx and HAP w/ protein
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 cm-1
190 All calcium oxalate spectra are termed CaOx (as opposed to COM or COD) due to the fact that their distinguishing absorptions between 1620 and 1660 cm-1 are overlapping with the amide I absorption of protein. Therefore it is not possible to state at this point in time whether calcium oxalate monohydrate or dihydrate is present since both share the same 778 and 1320 cm-1 absorptions. As can be seen from Figure 6.14, calcium oxalate is definitively determined by the presence of the 778 cm-1 absorption, as well as the 1320 cm-1. Both the 2nd and 3rd HAP spectra have been previously detailed (Figure 6.8). The 4th spectrum was obtained from the upper edge of Figure 6.13, still incorporated into the main body of the tissue. The 5th spectrum was obtained from a crystal separated from the main body of the tissue, in the upper right portion of Figure 6.13. The bottom spectrum exhibits interesting characteristics. Not only is CaOx present, but HAP and protein are present as well. Additionally, HAP and CaOx may possibly be in a disordered crystalline form due to the increased FWHM of the peaks. The presence of HAP and CaOx existing simultaneously at a mineral-tissue interface is important in the evolution of ICSF renal stones. It has been known that Randall’s plaque (hydroxylapatite) is the precursor for calcium oxalate stones in ICSF’s, however, the mechanism or urine environment necessary for the evolution of such stones remains undetermined. [11-13, 28, 105, 130, 222] Figure 6.13 demonstrates the ability of CaOx to co-exist with HAP as well as protein, suggesting possibly the first indication of the formation of CaOx stones from an HAP (Randall’s plaque) nucleation point.
6.4 Conclusions Though the interface between papillary tissue and renal stones formed from Randall’s plaque is difficult to visualize and interpret, progress has recently been made. The analysis of this interface using infrared microspectroscopy as well as optical microscopy has produced the possibility that the interface may consist of disordered hydroxylapatite, indicated by the breadth of the infrared absorptions. Additionally, calcium oxalate has been detected, which could indicate a transformation from hydroxylapatite to calcium oxalate as stone evolution occurs.
191 Optical microscopy has provided detailed structural information as to the physical characteristics of the interface, while infrared microspectroscopy illuminates the chemical composition present between the proteinacious tissue and the mineralized stones in the environment of the urine.
192 Conclusions to Renal Stone Research Several aspects of renal stone research have been addressed in this dissertation. Qualitative studies of renal stones, tissue sections, embedded mineral deposits, and renal stone models have been presented using available instrumentation and techniques. Qualitative analysis has been successfully performed on cross-sectioned renal stones using infrared surface reflectance techniques. Specifically, spatial resolution is high enough to define the placement of slightly differing components such as the two hydrates of calcium oxalate. MIQ and quantitative analyses using ATR have been attempted on models of tissue containing finely dispersed mineral components, as well as models of mineral inclusions embedded in tissue sections. Results demonstrate that, in homogeneous dispersions of COM in a tissue matrix, low concentrations of COM are detectable, specifically, COM has an detection limit of 0.09±0.02 wt% COM in a tissue matrix. Additionally, the MIQ of COM particles is approximately 10 μm, while hydroxylapatite particles down to approximately 20 μm can be identified. The quantitative measurement of homogeneously mixed renal stone components results in a number of non-linear calibration curves. Evidence indicates that non-linear calibration curves arise from several sources, mostly concerning the sizes of the particles involved in the mixture. However, the true source of the non-linearity remains undetermined. Finally, the tissue-mineral interface of idiopathic calcium stone formers has been analyzed using infrared reflection/absorption microspectroscopy. It was found that disordered hydroxylapatite crystals are present at the interface between the stone and tissue. An additional observation was that hydroxylapatite and calcium oxalate co-exist in regions where renal stones have not yet formed, lending insight into the transition between phosphate-based Randall’s plaque and the final product of a calcium oxalate renal stone. In conclusion, progress has been made in both qualitative and quantitative aspects of renal stone analysis. However, many future endeavors are possible that will continue to further this understanding and knowledge, as well as increase the applicability of the techniques reported here for clinical and commercial use.
193 Possible Future Research Pertaining to the Study of Renal Stones Though the qualitative analysis portion of this research has concluded, more quantitative data is required in order to develop a useable method of analysis. Off-axis ATR analysis may be a viable approach for the fast, accurate ATR mapping of renal stone cross-sections without many of the hindrances current single-bounce ATR possesses. Off-axis ATR is currently being pursued in other areas of analysis, most pharmaceutical in nature. [83, 223] Applications focusing on renal stone analysis are pending in this laboratory and will be patterned after Patterson, et. al. [223] Using off- axis ATR, increased spatial resolution will be achieved over typical surface reflectance due to the refractive index of the IRE. The investigation of renal stone components using a planar array infrared (PAIRS) microscope holds the advantage of being able to determine chemical composition in real-time. In the future, it would be beneficial to attempt the active spectral monitoring of renal stone formation in a reaction cell using the PAIRS instrument. The transition from calcium oxalate dihydrate to calcium oxalate monohydrate that is speculated to occur can be monitored in real-time, as well as the shift from hydroxylapatite to calcium oxalate compounds with a change in environmental factors. [173] Another suggested research endeavor is based off a recent study by Yano et al that successfully established infrared indicators for specific types of hereditary cancer. [59] This research sets a precedence that would be interesting to follow up and expand. It is claimed by some that genetic factors are included in the formation of certain types of renal stones.[31, 33, 35, 36] If these assertions are correct, it would be reasonable to believe that there is a genetic marker correlation between a patient currently experiencing renal stone formation and those of the same family that have yet to develop large renal stones requiring treatment. Using Yano’s research as a guide [59], an experiment could be devised to determine if infrared markers are present for the genetic proclivity to produce renal stones. Perhaps the use of genetically altered mouse kidney sections at selected intervals of renal stone development would be a wise place to begin the suggested research.
194
APPENDIX
Environmental Protection Agency Internship from May 2005 to September 2006
195 A.1 Introduction The overall focus of this internship was the determination and/or confirmation of components contained in environmental water samples and public water systems. This sample set included samples of pipe sections from water fountains and homes, residue from fire hydrants, samples of purifying agents used in water treatment plants, residue from residential water heater systems, and water filters used on a community level. Pinhole leaks, a common name for the spot-corrosion of copper pipes, were the primary focus of this research. Leaks and corrosion in drinking water pipes pose both environmental and health risks. Not only are leaky pipes wasteful monetarily, but it is estimated that up to 6 billion gallons of treated water are lost daily. [224] In addition, pipes that contain leaks and corrosion produce increased levels of copper and lead at initial daily use, meaning that the first use of water after the initiation of flow from a faucet contains dramatically higher levels of metal ions and related compounds than subsequent tests. [225] It has been shown that high levels of copper and lead in drinking water contribute to damage to the brain, blood cells, kidneys, stomach, and intestines in children, as well as lead to lower IQ’s and attention deficit disorder. In adults, the symptoms are less severe due to a lack of critical growth, however, blood pressure can be affected. [226] Additionally, colored or odorous water may result from pipe corrosion or its by-products. [227, 228] At this point in time, there is limited knowledge as to the mechanisms behind the formation of pinhole leaks. [229-231] Current forms of analysis for pinhole leak material and the associated corrosion include X-ray diffraction (XRD) and ion chromatography, among others.[229, 231-234] Infrared techniques and Raman analysis are used to a lesser extent.[235-238] Nagata used XRD in conjunction with IR and Raman techniques to analyze Japanese water pipe corrosion, finding the main component present was a zinc silicate. [235] Purcell used Raman spectroscopy to quantify the two competing corrosion processes simultaneously occurring in drilling pipe. [239] In order to understand the corrosion processes occurring inside metal pipe, model pipe sections are used in differing aqueous environments in order to mimic the circumstances of corrosion. [240] Izumi created a test situation to determine the corrosion mechanism for oxygenated boiler water treatment using alkali volatile treatment (AVT)
196 in a test loop. [236] It has only been in recent years that these attempts have been reproduced successfully. [229, 234, 241, 242] The internship with the Cincinnati office of the Environmental Protection Agency (EPA) began in May of 2005 and was set to last for 600 hours. In December 2005, the project was extended for an additional 300 hours at the request of the internship director, Dr. Darren Lytle. The initial internship outline focused on the microspectroscopic analysis of arsenic compounds in drinking water pipe sections. However, the focus was quickly altered to the analysis of corrosion products and pipe sections in general since the arsenic samples provided proved difficult to analyze. The original intent of the research was to analyze samples via Raman microspectroscopy alone, though after some discussion this was expanded to include infrared (IR) attenuated total internal reflectance (ATR) analysis as well. Using both Raman and infrared microspectroscopy, samples in a variety of forms were interrogated. For the majority of the 87 samples presented, the analysis performed in this laboratory was simply a confirmation or clarification of data obtained via XRD or chromatography. Only for a minority of the samples brought to our attention was the actual chemical composition in question. A.1.1 Sample Preparation Several sample morphologies were interrogated over the course of this internship. The following information summarizes the steps taken to ensure complete and competent analysis of the samples. For whole pipe sections containing a pit or pit cap, a small 1x1 cm area containing the point of interest was cut from the section in order to make the sample a manageable size. The act of cutting the sample into much smaller pieces also allowed the analysis of “flat” portions of the sample, where the curvature was minimized at one particular point. Figure A.1 displays a small pipe section containing an attached pit-cap composed of iron hydroxide, while Figure A.2 displays a pit in a copper pipe, also composed of iron hydroxide.
197 Figure A.1: Pit-cap on copper pipe. Pit-cap is composed of calcium hydroxide.
198 Figure A.2: Pit in copper pipe, cap removed. Inner brown layer is an interface membrane separating the pipe from the corroding material.
199 Several samples that were cross-sectioned and encased in resin allowed a side- profile view of the corrosion on the pipe. In these samples, all layers of corrosion are visible, allowing the interface layers between the corrosion and the pipe to be analyzed. (Figure A.3)
200 Figure A.3: Cross-section view of corrosion in pipe. Going from top to bottom are the following layers: top right green layer—corrosion on outside of pipe; gold band— uncorroded pipe; brown/teal material—thin meniscus layer containing oxo-anions and organic polymers; brown material—metal oxide; green layer at bottom left—organic polymer with sulfate ion; yellow inclusions in green layer—organic material.
201 Several samples arrived as intact pit caps detached from the pipe section. For these samples, the pit cap was analyzed in one of two ways: if a site on the pit cap was found to be of suitable flatness for analysis, the pit cap was studied as received. If no suitable flat surface could be located, the pit cap was ground into a fine powder and manually pressed into a homogeneous pellet. In either case, several sites on the sample were analyzed in order to determine overall composition. For samples that arrived in powder form, a manual pellet press was used to make a pellet for analysis. No additional grinding was required.
A.1.2 Sampling Procedure After the sample had been appropriately prepared for analysis, it was subsequently viewed at 10x magnification using the microscope of the Raman instrument in order to determine areas of interest. Areas of differing color or texture were analyzed in order to compile a complete record of sample components. Analysis occurred utilizing the 50x objective of the microscope. The 50x objective collects information from a 2μ diameter area on the surface of the sample, effectively allowing specific points on a sample to be isolated and analyzed regardless of their proximity to other sites of interest. The ability to isolate and analyze specific points of interest is important due to the fact that there exists a thin layer of material at the interface between the pipe and the pit or pit cap. This layer is several microns thick, necessitating the use of a high spatial resolution technique in order to ensure there is no influence from the surrounding materials.
A.1.3 Instrumentation Parameters Raman molecular microspectroscopy was used in the analysis of environmental standards and samples for work pertaining to the EPA. Each sample was analyzed using a Renishaw 2000 confocal Raman microprobe. Samples were excited with both a Spectra Physics HeNe (632nm) laser and a Lexel tunable Argon ion laser operating at 514nm and 488nm. These sources were focused on to the sample using the 50x (0.85 N.A.) objective which produces an approximate beam diameter of two micrometers on the sample surface. Spectra were collected at 4 cm-1 resolution and represent the average of 20
202 individual scans. The interrogation time for each spectral element was 30 seconds. The instrument was operated using the cosmic ray blocking program. In addition to Raman analysis, ATR-IR spectroscopy was also often utilized for samples not embedded in resin. Pressed pellets were analyzed using the Harrick Split Pea ATR attachment to the Perkin Elmer Spectrum 2000 infrared imaging system macro bench. Individual spectra collected represent the average of 32 scans with a spectral resolution of 4 cm-1. A load of 0.5 kg was used to ensure good contact between the IRE and the sample.
A.2 Standards Table A.1 is a complete list of standards run on the Renishaw 2000 in order to act as references.
203 Table A.1: List of standards for EPA internship
204 In addition, outside references such as internet searches and internet databases for the identification of some peaks were used. [232, 243, 244] All standard materials were prepared and analyzed under identical condition as previously described.
A.3 Results Table A.2 displays the results of over 900 hours of collection on 87 samples and sample sites. The initials DL, standing for Darren Lytle, begin most file names of the samples received (left-hand column). The Raman Results column details the results of Raman analysis using the 488, 514 and 633nm wavelengths. The Infrared Results column lists ATR results obtained on the Harrick Split Pea. The degree of certainty uses the following scale: Certain—98% sure it is the said component(s); Probable, 80-98%; Fairly Certain, 70-80%, and Possible, under 70% certain the said components are present. The term inconclusive implies that there was not enough data or there was too much fluorescence to determine the components present in the sample. The term unknown implies that peaks were present, however, the identity remains unknown. The term N/A implies that that procedure was not undertaken for that sample. Some samples, labeled Cu(XXX) were previously thought to be copper acetate. However, recent information has corrected that initial analysis. Currently, these compounds are believed
to be a copper carbonate, but not CuCO3.
Table A.2: Results of EPA analysis on corroded pipe sections and similar samples.
Sample Name Raman Infrared Degree of Notes Results Results Certainty DL000298 Inconclusive Inconclusive N/A DL000300 Bands are broad and Fe(NO3)2 Nitrates Fairly certain noise is high, but peak centers correspond 488nm wavelength is only supporting DL000301 Pb3O4 and Slight nitrates Possible evidence of Pb3O4. 633 Fe(NO3)2 and 514 are indicative of Fe(NO3)2. Vivianite, Only 514nm yielded DL000302 information, though Fe3(PO4)2*8 Phosphates Fairly certain vivianite is supported H2O by IR. No Raman bands were Phosphates, obtained for any DL000303 wavelength for this Inconclusive possibly N/A sample. IR bands were sulfates broad and inconclusive. Only one peak was DL000374 Inconclusive— present at any Green and one peak at Inconclusive N/A wavelength for Raman -1 and only on peak for Yellow 1543cm IR. DL000219 α-Fe O and Only ID’d by the Blue Bulk 2 3 N/A Fairly certain SO 2- 514nm wavelength Material 4 Fairly certain, Other materials are DL000219 present in addition to CuCO3 N/A with other Clear Layer CuCO3, however, these materials remain unidentified. DL000219 Metal oxide, Brown possibly Cr or N/A Fairly certain Ribbon in Mg Bulk DL000219 Silver Ribbon Cu(XXX) N/A Possible in Brown Ribbon DL000219 Blue Green No wavelengths Inconclusive N/A N/A responded to this Interface with sample. Pipe Only one peak was Building 441 exhibited, but it CuO2 N/A Possible corresponds to the expected component of CuO2. DL000246 PbO mixed Several peaks matched 2 N/A Possible the reference sample, Green Matter with organic but there appears to be
206 Sample Name Raman Infrared Degree of Notes Results Results Certainty in Bulk Cap material unidentified materials present as well. Oxo-anion, 488nm yielded peaks 2- probably belonging to DL000246 probably SO4 , a polymeric material, Blue Bulk in addition to Sulfates Probable possibly the encasing resin. IR confirms the Material organic Raman 633nm material evaluation of sulfates. DL000246 Metal oxide, Most likely MgO. Brown 488nm yielded only possibly Cr or Inconclusive Probable Interface bands. No info from Mg IR, 514 or 633nm. Layer DL000246 SO 2- in Green in 4 , 514 yields organic addition to resin bands, while Brown N/A Probable organic 633nm yields oxo- Interface anion information. material Layer DL000246 633nm yielded several Green strong peaks, none of Material Unknown N/A N/A which correspond to any of our known Outside of standards Pipe DL000440A Phosphates, Inconclusive N/A No Raman data. C=O IR suggests the DL000440B presence of hydrated Inconclusive Inconclusive N/A sulfate groups with strong broad peaks at 1158 and 1225cm-1. DL000440C Inconclusive Inconclusive N/A No Raman data.
Since phosphorous DL000440D Sulfur or isn’t indicated in CuS Possible Raman, the IR is phosphorus probably due to sulfur groups. Since phosphorous DL000440E Sulfur or isn’t indicated in CuS Possible Raman, the IR is phosphorus probably due to sulfur groups. DL000339 Two bands correspond Fe(NO3)2 Orthophosphate Possible to Fe(NO3)2 in the Raman spectrum DL000340 Inconclusive Inconclusive N/A
IR markers for CaCO3 DL000355 CaCO , are positively ID’d Inconclusive 3 Certain suggests sulfur while the presence of sulfur is suggested.
IR markers for CaCO3 DL000356 CaCO , are positively ID’d Inconclusive 3 Certain suggests sulfur while the presence of sulfur is suggested.
207 Sample Name Raman Infrared Degree of Notes Results Results Certainty DL000443 Confirmed by Raman CuSO4 CuSO4 Certain and IR. DL000442A Five Raman bands α-FeOOH Inconclusive Probable matched α-FeOOH to the standard. DL000442B Strong orthophosphate Inconclusive Orthophosphate Probable band at 1010cm-1.
DL000439 Strong orthophosphate Unknown Orthophosphate Probable band at 1016cm-1. DL000445 No Raman or IR data of any quality. (white and Inconclusive Inconclusive N/A Possible S influences green areas) in IR. CuSO4*X DL000253 H O, possibly 2 N/A Fairly Certain Blue Inclusion magnetite and organics DL000253 Blue Main Organics N/A Probable Body of Pit Cap DL000253 Uncertain; N/A “Cap side” of Possible Organics interface DL000253 Uncertain; “Pipe side” of N/A Possible Organics interface The Raman markers DL000214 Possible for Fe O are pretty Fe O Probable 3 4 Black Mass 3 4 phosphate consistent with the standard. α-FeOOH The Raman markers DL000214 Possible for α-FeOOH are (Goethite), Probable Orange Mass phosphate pretty consistent with Organics the standard. Organics, Raman information DL000312 As reveals only organics Carbon (C-C, Organics (Ring present for certain, removal Probable C=C),Carbon stretching) though some Fe may media be implied in the rings 514nm spectrum. Expected to be Cu(OH) . IR supports DL000334 Cu2SO3(OH)2* Intermolecular x Fairly Certain the presence of *H2O H2O H-bonding with strong broad H- bonding DL000329 Thought to be CuO. Black and Too much Inconclusive Unknown N/A fluorescence in both White pipe and pellet form to Residue determine content. DL000328 IR and Raman confirm CuCO3 CuCO3 Certain the presence of Green CuCO3.
208 Sample Name Raman Infrared Degree of Notes Results Results Certainty Residue
DL000273
Cold Water Piping Inconclusive N/A N/A Raman provided unuseable data. (Brown Residue) DL000273 488nm and 633nm Cold Water yielded no results, CuCO3 N/A Certain however, all peaks in Piping (Green 514nm corresponded Residue) nicely to CuCO3. DL000273 Cold Water Piping Unknown N/A N/A (Brown Residue) DL000293 Water Heater Unknown Inconclusive N/A Precipitates A
DL000293 Majority of peaks in Water Heater Pb3O4 Inconclusive Probable 633nm spectrum Precipitates B correspond to Pb3O4. DL000293 Only one firm peak in 633nm and 514nm. Water Heater Unknown N/A N/A Not enough to pass Precipitates C judgment on. DL000293 Only 633nm yields Water Heater Alum Unknown Possible peak information. Precipitates D DL000293 Four of 6 peaks in the 633nm spectrum Water Heater Ni(Acetate) Inconclusive Fairly Certain correspond; 514 and Precipitates E 488nm yield no peaks. DL000350 Fe bulk material. Both Very detailed, 633 and 514 Orange α-FeOOH Probable but Unknown correspond nicely to α- Residue FeOOH DL000350 Fe bulk material. Fe3O4 Unknown Possible 488nm possibly Silver Residue corresponds to Fe3O4. Thought to contain Pb Aging PbO2 and filter PbO2 Certain PbO2, confirmed by IR residue and Raman. DL000272 Unknown N/A N/A DL000348 Unknown N/A N/A Cu Morphis IR and Raman confirm CuO, CuSO4 CuO, P=O, OH Certain the presence of CuO and CuSO4
209 Sample Name Raman Infrared Degree of Notes Results Results Certainty 10mg/L Cu Phosphates, Strong evidence in Organics Fairly certain both IR and Raman 3mg/L PO4 CuO, organics DL000219 Alumina with Raman bands are characteristic of Corrosion Cr impurity, CaCO3 Fairly certain Alumina with Cr Cap organics impurity.
IR bands confirm the DL000219 Alumina with presence of CaCO3, Exterior Cr impurity, CaCO Certain while Raman is 3 characteristic of Corrosion organics Alumina with Cr impurity. Phosphates, Cu pipe Phosphates, Generally confirmed organics, Fairly certain organics by both IR and Raman amide A lot of fluorescence Pb pipe Organics, in Raman spectra, but FeOOH Fairly certain Metal oxide at least 5 bands correspond to FeOOH. Several Raman bands DL000253 correspond to FeOOH, Organics? but this is not Unexposed FeOOH Possible confirmed in IR or (fresh) pit cap with additional Raman bands. Several Raman bands DL000253 correspond to FeOOH, but this is not Exposed (Old) FeOOH Organics? Possible confirmed in IR or pit cap with additional Raman bands. IR is positive for DL000304 CaCO , and Raman is FeOOH, Cu, S CaCO , S Probable 3 Loose pit cap 3 a likely match to FeOOH IR is negative for DL000304 CaCO3, but is positive CuxSO4yOHz, Attached pit FeOOH, Cu, S Probable for CuxSO4yOHz, and S cap Raman is a likely match to FeOOH IR is negative for DL000304 Pit CaCO3, but is positive CuxSO4yOHz, cap/ pit FeOOH, Cu, S Probable for CuxSO4yOHz, and S interface Raman is a likely match to FeOOH DL000275 Possible S Inconclusive N/A group DL000278 Possible S Inconclusive N/A group Arsenic is much more DL000280 M (AsO )OH, difficult to determine x 4 M (AsO )OH Possible where M=Fe? x 4 than we thought it’d be. Arsenic is much more DL000283 M (AsO )OH, difficult to determine x 4 M (AsO )OH Possible where M=Fe? x 4 than we thought it’d be. DL000329 Cu Inconclusive Inconclusive N/A Too much
210 Sample Name Raman Infrared Degree of Notes Results Results Certainty pipe with fluorescence in Raman to get data; IR is white residue unremarkable. DL000312 As Arsenic is much more Organic Possible difficult to determine removal Possible material CaCO than we thought it’d media 3 be. CuCO3, not CuCO3 matches the DL000328 sample precisely. The Malachite, Green standard may have Cu2(CO3)(OH)2 OH bonding Certain been Cu2(CO3)(OH)2, Homogeneous in which case the as was Layer sample corresponds to suspected this as well. The Raman spectrum DL000334 has several OH pH9 Cu SO (OH) * corresponding peaks, 2 3 2 intermolecular Probable as well as the fact that DIC=100mg/L H O IR indicates strong 2 bonding Cu=10mg/L intermolecular OH bonding. Strong evidence in the DL000214 Fe3O4 and Raman spectra for N/A Probable Fe O , weaker Orange Mass Fe(NO3)2 3 4 evidence for Fe(NO3)2. Strong evidence in the DL000214 α-FeOOH and Raman spectra for N/A Probable FeOOH , weaker Black Mass PbO2 evidence for PbO2. DL000253 Raman spectra Main body of CuSO4 N/A Probable correspond nicely. pit cap
DL000253 Majority of peaks in Blue inclusion Ni(Ac) N/A Certain all three wavelengths in pit cap correspond to Ni(Ac) DL000253 633nm is only spectrum not “Pipe-side” of Ni(Ac) N/A Certain influenced by interface fluorescence DL000253 633nm strongly Cu(XXX), supports the presence “Cap-side” of N/A Fairly Certain of Ni(AC), while Ni(Ac) 514nm confirms interface Cu(XXX). DL000304 Cu(XXX) CuCO Fairly Certain Loose pit cap 3
DL000304 IR contains no bands Attached pit Cu(XXX) Inconclusive Possible corresponding to cap CuCO3. DL000304 Pb3O4 is supported in Pit/Pit cap Cu(XXX) Pb3O4? Possible IR, but not in Raman interface as it should be DL000253 Cu(XXX), Several Raman bands Exposed pit N/A Fairly Certain Pb O correspond to each cap 3 4
211 Sample Name Raman Infrared Degree of Notes Results Results Certainty DL000253 Cu(XXX), N/A Fairly Certain Several Raman bands Unexposed pit Pb3O4 correspond to each cap
DL000120 Cu IR and Raman agree pipe from CuCO3 CuCO3 Certain on the presence of North Dakota CuCO3 Pb Pipe Correspondence of Raman peaks to Section α-FeOOH N/A Certain standards. Raman peaks are very intense.
212 A.4 Discussion The appearance of pitting and corrosion in household and industrial water piping is not uncommon. Pinhole leaks, as they are often known, can form in a matter of months, causing great property damage. Due to sedentary layers formed in the bottom of the pipes and the stop-and-flow nature of domestically piped water, copper, lead, and galvanized steel pipes remain exposed to water treatment chemicals and other compounds for extended periods of time. Residue, stationary in the bottom of the pipe, is an ideal home to microorganisms, which are often suspected as the initiators of corrosion. [229, 230] Copper pitting, the adjusted primary focus of this internship, is fundamentally electrochemical corrosion. [229] Reduction and oxidation take place in the pipe, with the site of corrosion being focused at the anodic site. When the thin, protective coating present on new pipe is worn away in a specific site, metal oxides and other elements can combine under the proper conditions to induce oxidation/reduction reactions and form pinhole leaks from corrosion. This is termed “non-uniform corrosion”, as opposed to uniform corrosion, which takes place over the entire length of the piping system and can lead to high concentrations of copper and lead in drinking water.[229] Once a site has begun corroding, metal ions coming off the site can precipitate as solids, building up into a mound termed a pit cap in this dissertation. Underneath this pit cap, a microenvironment is formed that excludes oxygen and accelerates corrosion. The research performed for this internship served to identify the materials present in the pipe, pit, pit cap, and the membrane layer at the interface of the reacting species. At the interface between the pit and pit cap, as shown in Figure A.2, there exists a thin membrane layer serving to separate the uncorroded pipe from the corroding material. This membrane layer, often containing large amounts of organic material, is on the order of 3-4 μm thick. Compounds on the inside surface of the pipe undergo oxidation/reduction reactions depending on the pH, temperature, and components present. Contrary to popular belief which states that corrosion occurs in more acidic environments, Edwards has found that alkaline environments enhance corrosion and accelerate its effects. [234, 245, 246] In addition to pH changes, corrosion is also accelerated by the presence of
213 aluminum. [229] This information is supported by the findings of polishing media in several cross-sectioned samples observed in this laboratory. Though the actual ion-exchange mechanism and methods of formation responsible for the corrosion of pipe surfaces remains largely unknown, [233] the circumstances under which pipes corrode are well-documented. [231, 235-238, 241, 247- 250] A general listing of copper species perpetuated as a result of corrosion is given by Martens et. al., [237] with an expanded library of corrosion spectra of various metal species available from Bouchard et. al..[232] Interestingly, pipe corrosion and archaeological corrosion share many of the same compounds, as Bouchard points out. As opposed to elemental techniques, Raman and infrared techniques provide molecular information. Though X-ray diffraction, a common technique used for the interrogation of corrosion samples, is also molecular in nature, Raman is non-destructive and maintains the integrity of the sample for further analysis. In Raman microspectroscopy, the organic or inorganic ligands that are bound to the metals are known determined, providing the oxidation state of the attached metal. For example, because of the Raman bands at 251, 301, 394, 470 and 531 cm-1, we know that the Fe compound present is Goethite, also known as α-FeOOH. α-FeOOH is easily
distinguishable from other Fe compounds because of its peak placements; Fe(NO3)2, for example, has key Raman bands at 1067, 669 and 283 cm-1, which vary widely from those of α-FeOOH. Only rarely has Raman analysis been used for the determination of corrosion products in water pipe sections. [239, 251-255] More often, Raman analysis is applied to industrial steel samples, such as building frames and construction equipment, such as the work by Ezawa, examining soft steel sheets. [256] In the research presented here, Raman analysis has been applied to a variety of water pipe sections, including the common copper and lead sections found in most residential and industrial facilities. Additionally, outside references were used to confirm the identity of copper species. Figure A.4 displays an example spectrum from a published reference [257] as well as a copper acetate spectrum (488 nm) collected from the research conducted here. It should be mentioned that the range for the published spectrum is 100-2000 cm-1, while the range for the experimentally collected spectrum is 200-2000 cm-1. Therefore, the
214 intense peak in the published spectrum at 179 cm-1 will not be seen in the experimental data.
215 Figure A.4: Top: Verdigris (copper acetate monohydrate) Raman spectrum.[257] Bottom: Raman spectrum of copper acetate collected experimentally (488nm excitation laser).
-1 -1 948 cm 321 cm
1439 cm-1 1416 cm-1
-1 702 cm-1 1358 cm
216 It would appear that an impurity may also be present in the copper acetate standard sample used in this research. This is because in addition to the several peaks corresponding to the reference (top) in Figure A.4, three other peaks are present at 1358, 1416, 1439 cm-1, possibly corresponding to a sulfate. [258] In an interesting finding, phosphates were found in many of the pipe samples analyzed in our laboratory. Upon further investigation, it was found that as a result of pipe pitting, orthophosphate is often used in pipe treatment as a corrosion inhibitor in order to extend pipe lifetime. [229] This fact is corroborated by many of the samples analyzed here where IR analysis revealed strong orthophosphate peaks near 1020cm-1. Others have found substantial evidence for phosphate-containing copper compounds in this wavenumber region. [259] Arsenic compounds, originally the primary focus of this research, proved to be difficult sample material in the form presented. Though iron and the organic carbon previously believed present could be detected, the presence of arsenic was continually questioned by Raman spectroscopy. This is because referenced arsenic compounds display peaks in the range of 900-85 cm-1, while the samples provided for this research had only two Raman peaks at 1600 and 1332 cm-1, with no other peaks present (Figure A.5). Andrikopoulos found arsenic sulfide Raman peaks at approximately 350 cm-1, [260] while Beck found arsenic compounds to have several peaks, however, all were below 500 -1 3- -1 cm . [261] Griffith found (AsO4) peaks in the low to mid 800 cm region, as well as -1 FeAsO4*H2O at 819 cm . [262] Since the arsenic-collecting media presented for analysis is iron, it would be reasonable that some sort of arsenic coordinate compound would present peaks near 800 cm-1. However, as can be seen, no peaks are present. It remains unclear as to the reason for such persistent difficulties.
217 Figure A.5: Iron media supposedly containing arsenic. Peaks from left to right correspond to 1600 and 1350 cm-1, representing the organic components previously suspected to be present in the Fe pellets.
218 A.5 Conclusions The use of Raman microspectroscopy, in conjunction with infrared ATR or alone, is a viable technique for the elucidation of chemical information for samples pertaining to water treatment, transport, and storage. Though unsuccessful for several samples listed in Table A.2, Raman microspectroscopy serves overall as a useful tool for the confirmation of materials present in pipe samples of varying origin. The ability to analyze copper and lead pipe sections by both infrared and Raman microspectroscopy allows the complete scope of material present to be determined. Of the 87 samples supplied, only 18 were unable to be analyzed. Of those 18, 6 were arsenic removal media samples. Therefore, the methods presented here have a success rate of 86% for non-arsenic containing samples. Unfortunately, the ability to determine arsenic in the sample form supplied for analysis is unexpectedly poor.
219
1. Dion, R., The Microscopic Examination of Urine. 2005, College de Rosemont: quebec, Canada. 2. Herring, L.G. and Co., Analytical and Consulting Chemists. www.herringlab.com. 3. Carmona, P., J. Bellanato, and E. Escolar, Infrared and Raman Spectroscopy of Urinary Calculi: A Review. Biospectroscopy, 1997. 3: p. 331-346. 4. Sperling, O., Hereditary renal hypouricemia. Molecular Genetics and Metabolism, 2006. 89(1-2): p. 14-18. 5. Roeling, J., et al., HIV-associated renal diseases and highly active antiretroviral therapy-induced nephropathy. Clinical Infectious Diseases, 2006. 42(10): p. 1488-1495. 6. Jeske, T. and A. Denys, Nanobacteria -the role in human infectious pathology. International Review of Allergology & Clinical Immunology, 2005. 11(4): p. 151- 152. 7. Berg, W., et al., Submicroscopic Investigations on Calcium Oxalate Stone Genesis. European Urology, 1979. 5(2): p. 136-43. 8. Randall, A., The Origin and Growth of Renal Calculi. Ann. Surg., 1937. 105: p. 1009-1027. 9. Evan, A.P., et al., Apatite plaque particles in inner medulla of kidneys of calcium oxalate stone formers: osteopontin localization. Kidney International, 2005. 68(1): p. 145-154. 10. Evan, A.P., et al., Randall's plaque of patients with nephrolithiasis begins in basement membranes of thin loops of Henle. Journal of Clinical Investigation, 2003. 111(5): p. 607-616. 11. Evan, A., et al., Randall's plaque: Pathogenesis and role in calcium oxalate nephrolithiasis. Kidney International, 2006. 69(8): p. 1313-1318. 12. Kim, S.C., et al., Stone formation is proportional to papillary surface coverage by Randall's plaque. Journal of Urology (Hagerstown, MD, United States), 2004. 173(1): p. 117-119. 13. Evan, A.P., Personal Communication, J. Anderson, Editor. 2006: Oxford. 14. Evan, A., P., et al., To Be Announced. Nature, 2006. 15. Gilg, H.A., et al., Genesis of amethyst geodes in basaltic rocks of the Serra Geral Formation (Ametista do Sul, Rio Grande do Sul, Brazil): a fluid inclusion, REE, oxygen, carbon, and Sr isotope study on basalt, quartz, and calcite. Mineralium Deposita, 2003. 38(8): p. 1009-1025. 16. Garcia-Guinea, J., et al., Formation of gigantic gypsum crystals. Journal of the Geological Society (London, United Kingdom), 2002. 159(4): p. 347-350. 17. Daudon, M. and P. Jungers, Drug-Induced Renal Calculi: Epidemiology, Prevention and Management. Drugs, 2004. 64(3): p. 245-75. 18. Zerwekh, J.E., Nutrition and Renal Stone Disease in Space. Nutrition, 2002. 18(10): p. 857-863. 19. Whitson, P.A., et al., The Risk of Renal Stone Formation During and After Long Duration Space Flight. Nephron, 2001. 89(3): p. 264-270. 20. Whitson, P.A., et al., Alterations in Renal Stone Risk Factors After Space Flight. Journal of Urology, 1993. 150(3): p. 803-807.
220 21. Whitson, P.A., R.A. Pietrzyk, and C.Y.C. Pak, Renal Stone Risk Assessment During Space Shuttle Flights. Journal of Urology, 1997. 158(6): p. 2305-2310. 22. Whitson, P.A., R.A. Pietrzyk, and C.F. Sams, Space Flight and the Risk of Renal Stones. Journal of Gravitational Physiology, 1999. 6(1): p. p87-p88. 23. Hoyer, J.R., et al., Effects of Microgravity on Urinary Osteopontin. Journal of the American Society of Nephrology, 1999. 10(11, Suppl. 14): p. S389-S393. 24. Dash, A., et al., Ureteroscopic Treatment of Renal Calculi in Morbidly Obese Patients: A Stone-Matched Comparison. Urology, 2002. 60: p. 393-397. 25. Manoharan, M. and P. Schwille, Oxypurines, Protein, Glucose and the Functional State of Blood Vasculature are Markers of Renal Calcium Stone-Forming Processes? Observations in Men with Idiopathis Recurrent Calcium Urolithiasis. Clinical Chemistry and Laboratory Medicine, 2002. 40(3): p. 266-277. 26. Andreoni, C., et al., Flexible Ureteroscopic Lithotripsy: First-Line Therapy for Proximal Ureteral and Renal Calculi in the Morbidly Obese and Superobese Patient. Journal of Endourology, 2001. 15(5): p. 493-498. 27. Buchholz, N.-P., Three-Dimensional CT Scan Stone Reconstruction for the Planning of Percutaneous Surgery in a Morbidly Obese Patient. Urologia Internationalis, 2000. 65: p. 46-48. 28. Evan, A., et al., Randall's Plaque of Patients With Nephrolithiasis Begins in Basement Membranes of Thin Loops of Henle. Journal of Clinical Investigation, 2003. 111(5): p. 607-616. 29. Nelson Wayne, K., et al., Enteric hyperoxaluria, nephrolithiasis, and oxalate nephropathy: potentially serious and unappreciated complications of Roux-en-Y gastric bypass. Surgery for obesity and related diseases : official journal of the American Society for Bariatric Surgery, 2005. 1(5): p. 481-5. 30. Barbey, F., et al., Carbonic anhydrase inhibitors and calcium phosphate stones. 2004, Division de nephrologie, Centre hospitalier universitaire vaudois, Lausanne, Suisse.: Switzerland. p. 169-72. 31. Bushinsky, D.A., et al., Calcium oxalate stone formation in genetic hypercalciuric stone-forming rats. Kidney International, 2002. 61(3): p. 975-987. 32. Bushinsky, D.A. and J.R. Asplin, Thiazides reduce brushite, but not calcium oxalate, supersaturation, and stone formation in genetic hypercalciuric stone- forming rats. Journal of the American Society of Nephrology, 2005. 16(2): p. 417-424. 33. Coe, F.L., A. Evan, and E. Worcester, Kidney stone disease. Journal of Clinical Investigation, 2005. 115(10): p. 2598-2608. 34. Moe, O.W., Kidney stones: pathophysiology and medical management. Lancet, 2006. 367(9507): p. 333-344. 35. Evan, A.P., et al., Calcium oxalate crystal localization and osteopontin immunostaining in genetic hypercalciuric stone-forming rats. Kidney International, 2004. 65(1): p. 154-161. 36. Jaeger, P., Genetic Versus Environmental Factors in Renal Stone Disease. Current Opinion in Nephrology and Hypertension, 1996. 5(4): p. 342-6. 37. Emmerson, B.T. and P.J. Ravenscroft, Abnormal Renal Urate Homeostasis in Systemic Disorders. Nephron, 1975. 14(1): p. 62-80.
221 38. Hall, J., et al., Abnormal Kidney Function as a Cause and a Consequence of Obesity Hypertension. Clinical and Experimental Pharmacology and Physiology, 1998. 25: p. 58-64. 39. Hall, J., The Kidney, Hypertension, and Obesity. Hypertension, 2003. Part II(March 2003): p. 625-633. 40. Hall, J., et al., Obesity-Associated Hypertension and Kidney Disease. Current Opinion in Nephrology and Hypertension, 2003. 12: p. 195-200. 41. Sasatomi, Y., et al., Obesity Associated with Hypertension or Hyperlipidemia Accelerates Renal Damage. Pathobiology, 2001. 69: p. 113-118. 42. Ahmed, J. and N. Khalid, Infrared Spectroscopic Studies of Secondary Renal Calculi. Science International (Lahore), 1992. 4(3): p. 273-4. 43. Anderson, J.C., et al., A Concerted Protocol for the Analysis of Renal Mineral Deposits in Biopsied Tissue Using Infrared Microanalysis. Urological Research, 2005. 33: p. 213-219. 44. Anderson, J.C., et al., Analysis of Urinary Calculi Using an Infrared Microspectroscopic Surface Reflectance Imaging Technique. Urological Research, 2006. Accepted 2006. 45. Anderson, J., et al. Quantitative Analysis of Oxalate Salts in Renal Biopsies via ATR Infrared Microspectroscopy and its Relevance to Kidney Disease. in Microscopy and Microanalysis. 2006. Chicago, Illinois: Cambridge University Press. 46. Anderson, J., Infrared Microspectroscopic Techniques for the Qualitative Analysis of Cross-Sectioned Renal Calculi and Embedded Mineralized Deposits, in Chemistry. 2004, Miami University: Oxford. p. 124. 47. Evan, A.P., Personal Communication, J. Anderson, Editor. 2003: Oxford, OH. 48. Kasidas, G.P., C.T. Samuell, and T.B. Weir, Renal stone analysis: Why and how? Annals of Clinical Biochemistry, 2004. 41(2): p. 91-97. 49. Kiernan, J., Histological & Histochemical Methods: Theory & Practice. 1999, Oxford: Butterworth Heinemann. 50. Clark, J., I. Thompson, and S. Optenberg, Economic Impact of Urolithiasis in the United States. Journal of Urology, 1995. 154: p. 2020-2024. 51. Estepa, L. and M. Daudon, Contribution of Fourier Transform Infrared Spectroscopy to the Identification of Urinary Stones and Kidney Crystal Deposits. Biospectroscopy, 1997. 3: p. 347-369. 52. Grey, H., Anatomy of the Human Body. 1918, Philadelphia: Lea & Ferbiger. 53. Lingeman, J.E., Bioeffects of Extracorporal Shock Wave Lithotripsy: A Worry or Not? Journal of Urology, 1992. 148: p. 1025. 54. Shao, Y., et al., Morphological Changes Induced in the Pig Kidney by Extracorporeal Shockwave Lithotripsy: Nephron Injury. Anatomical Record, 2003. 275A(1): p. 979-89. 55. Lingeman, J., Extracorporal Shockwave Lithotripsy--What Happened? Journal of Urology, 2003. 169(1): p. 63. 56. Evan, A.P., et al., Crystal-associated nephropathy in patients with brushite nephrolithiasis. Kidney International, 2005. 67(2): p. 576-591. 57. Skoog, D.A., D.M. West, and F.J. Holler, Fundamentals of Analytical Chemistry. 7th ed. 1996, Fort Wort: Saunders College Publishing.
222 58. Strobel, H. and W. Heineman, Chemical Instrumentation: A Systematic Approach. 1989, New York: Wiley-Interscience Publication. 59. Yano, K., et al., Direct Measurement of Human Lung Cancerous and Noncancerous Tissues by Fourier Transform Infrared Microscopy: Can an Infrared Microscopy be Used as a Clinical Tool? Analytical Biochemistry, 2000. 287: p. 218-225. 60. Chiriboga, L., et al., Infrared Spectroscopy of Human Tissue. III. Spectral Differences Between Squamous and Columnar Tissue and Cells From the Human Cervix. Biospectroscopy, 1997. 3(4): p. 253-257. 61. Chiriboga, L., H. Yee, and M. Diem, Infrared Spectroscopy of Human Cells and Tissue. Part VI: A Comparative Study of Histopathology and Infrared Microspectroscopy of Normal, Cirrhotic and Cancerous Liver Tissue. Applied Spectroscopy, 2000. 54(1): p. 1-8. 62. Oka, T., T. Koide, and T. Sonoda, Estimation by Infrared Spectrophotometer of the Calcium Oxalate Dihydrate to Calcium Oxalate Monohydrate Ratio. Journal of Urology, 1985. 134(4): p. 813-17. 63. Schaeberle, M.D., I. Levin, and E.N. Lewis, Biological Vibrational Spectroscopic Imaging, in Infrared and Raman Spectroscopy of Biological Materials, H.-U. Gremlich and B. Yan, Editors. 2001, Marcel Dekker, Inc: New York. 64. Schaeberle, M., I. Levin, and E. Lewis, Clinical Applications of Near- and Mid- Infrared Spectroscopy, in Infrared and Raman Spectroscopy of Biological Materials, H.-U. Gremlich and Y. B;, Editors. 2001, Marcel Dekker: New York. p. 231-258. 65. Bergin, F.J., Some Novel Applications of an Infrared Microscope. Applied Spectroscopy, 1989. 43(3): p. 511-15. 66. Blitz, J.P., Diffuse Reflectance Spectroscopy, in Modern Techniques in Applied Molecular Spectroscopy, F.M. Mirabella, Editor. 1998, Wiley-Interscience Publication: New York. p. 185-219. 67. Aboul-Enein, H.Y., A.A. Bunaciu, and S. Fleschin, Analysis of Mebendazole Polymorphs by Fourier Transform IR Spectrometry Using Chemometric Methods. Biopolymers, 2002. 67(1): p. 56-60. 68. Agatonovic-Kustrin, S. and R. Alany, Application of Diffuse Reflectance Infrared Fourier Transform Spectroscopy Combined With Artificial Neural Networks in Analyzing Enantiomeric Purity of Terbutaline Sulfate Bulk Drug. Analytica Chimica Acta, 2001. 449(1-2): p. 157-165. 69. Agatonovic-Kustrin, S., et al., Determination of Polymorphic Forms of Ranitidine-HCl by DRIFTS and XRPD. Journal of Pharmaceutical and Biomedical Analysis, 2001. 25(5-6): p. 741-750. 70. Beckstead, H.D., G.A. Neville, and H.F. Shurvell, Differentiation of Solvated Spironolactone Samples by FT-Raman and FT-IR Diffuse Reflectance Spectroscopy. Fresenius' Journal of Analytical Chemistry, 1993. 345(11): p. 727- 32. 71. Bunaciu, A.A., S. Fleschin, and H.Y. Aboul-Enein, Analysis of Mebendazole Polymorphs by Fourier Transform Infrared Spectrometry Using Chemometric Methods. Spectroscopy Letters, 2001. 34(5): p. 527-536.
223 72. Bunaciu, A.A., S. Fleschin, and H.Y. Aboul-Enein, Analysis Improvement of Drugs Using Modern Methods. Roumanian Biotechnological Letters, 2002. 7(2): p. 647-654. 73. Chan, T.Y., et al., High Throughput On-Bead Monitoring Of Solid Phase Reactions by Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS). Tetrahedron Letters, 1997. 38(16): p. 2821-2824. 74. Gordon, K. and T. Rades, Spectroscopy and Solid State Properties of Drugs: Will Femtochemistry Meet Pharmacy? Chemistry in New Zealand, 2002. 66(4): p. 43- 47. 75. Neville, G.A., H.D. Beckstead, and H.F. Shurvell, Utility of Fourier Transform- Raman and Fourier Transform-Infrared Diffuse Reflectance Spectroscopy for Differentiation of Polymorphic Spironolactone Samples. Journal of Pharmaceutical Sciences, 1992. 81(12): p. 1141-6. 76. Sertsou, G., S. Agatonovic-Kustrin, and T. Rades, The Use of DRIFTS to Analyze Binary Mixtures of Two Polymorphic Forms of Ranitidine-HCl Using Only One Specific Peak. Internet Journal of Vibrational Spectroscopy, 1999. 3(5): p. No pp. given. 77. Suzuki, E.M. and W.R. Gresham, Forensic Science Applications of Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS): II. Direct Analysis of Some Tablets, Capsule Powders, and Powders. Journal of Forensic Sciences, 1986. 31(4): p. 1292-313. 78. Vippagunta, S.R., et al., Solid-State Characterization of Nifedipine Solid Dispersions. International Journal of Pharmaceutics, 2002. 236(1-2): p. 111-123. 79. Yang, H. and J. Irudayaraj, Rapid Determination of Vitamin C by NIR, MIR and FT-Raman Techniques. Journal of Pharmacy and Pharmacology, 2002. 54(9): p. 1247-1255. 80. Berthelot, M., et al., Diffuse Reflectance Technique for Infrared Analysis of Urinary Calculi. Clinical Chemistry, 1987. 33(6): p. 780-3. 81. Ingle, J.D. and S.R. Crouch, Spectrochemical Analysis. 1988, New Jersey: Prentice Hall. 590. 82. Reffner, J. and P. Martoglio, Uniting Microscopy and Spectroscopy, in Practical Guide to Infrared Microspectroscopy, H. Humecki, Editor. 1995, Marcel Dekker, Inc.: New York. p. 472. 83. Lewis, L.L. and A.J. Sommer, Attenuated Total Internal Reflection Infrared Mapping Microspectroscopy of Soft Materials. Applied Spectroscopy, 2000. 54(2): p. 324-330. 84. Patterson, B.M. and G.J. Havrilla, Attenuated total internal reflection infrared microspectroscopic imaging using a large-radius germanium internal reflection element and a linear array detector. Applied Spectroscopy, 2006. 60(11): p. 1256-1266. 85. Skoog, D., A;, F.J. Holler, and T.A. Nieman, Principles of Instrumental Analysis. fifth ed. 1998, Philadelphia: Saunders College Publishing. 849. 86. Sommer, A., Raman Microspectroscopy, in Modern Techniques in Applied Molecular Spectroscopy, F.M. Mirabella, Editor. 1998, John Wiley & Sons, Inc.: New York. p. 410.
224 87. Pemble, M.E., Vibrational Spectroscopy from Surfaces, in Surface Analysis: The Principle Techniques, J.C. Vickerman, Editor. 1997, Wiley: New York. 88. Andersen, M.E. and R.Z. Muggli, Microscopical techniques in the use of the molecular optics laser examiner Raman microprobe. Analytical Chemistry, 1981. 53(12): p. 1772-7. 89. Hallmark, V.M., et al., FT-Raman spectroscopy of biological molecules. Microchimica Acta, 1988. 95(1-6): p. 215-218. 90. Kiefer, W., et al., Biological Applications of Micro-Raman Spectroscopy. Microscopy and Microanalysis, 2003. 9(Suppl 2): p. 1056-1058. 91. Ling, J., Toward Raman Spectroscopy of Biological Tissue for Cancer Diagnosis. 2003, Southwest Reserach Institute. 92. Wallach, D.F.H., S.P. Verma, and J. Fookson, Application of laser Raman and infrared spectroscopy to the analysis of membrane structure. Biochimica et Biophysica Acta, Reviews on Biomembranes, 1979. 559(2-3): p. 153-208. 93. Malt, R.A., Feasibility of obtaining Raman spectra from nucleic acid constituents. Biochimica et Biophysica Acta, Biophysics Including Photosynthesis, 1966. 120(3): p. 461-5. 94. Lord, R.C. and G.J. Thomas, Jr., Raman studies of nucleic acids. II. Aqueous purine and pyrimidine mixtures. Biochimica et Biophysica Acta, Nucleic Acids and Protein Synthesis, 1967. 142(1): p. 1-11. 95. Lambert, P.J., et al., Raman spectroscopy: the gateway into tomorrow's virology. Virology Journal, 2006. 3: p. No pp given. 96. Pearman, W.F. and A.W. Fountain, III, Classification of chemical and biological warfare agent simulants by surface-enhanced Raman spectroscopy and multivariate statistical techniques. Applied Spectroscopy, 2006. 60(4): p. 356- 365. 97. Romero-Torres, S., et al., Raman spectroscopy for tablet coating thickness quantification and coating characterization in the presence of strong fluorescent interference. Journal of Pharmaceutical and Biomedical Analysis, 2006. 41(3): p. 811-819. 98. Durig, J.R. and D.J. Gerson, Historical survey of the infrared and Raman spectroscopic study of biological molecules. NATO Advanced Study Institutes Series, Series C: Mathematical and Physical Sciences, 1979. 43(Infrared Raman Spectrosc. Biol. Mol.): p. 35-43. 99. Rodriguez, L.G., S.J. Lockett, and G.R. Holtom, Coherent anti-stokes Raman scattering microscopy: a biological review. Cytometry, Part A, 2006. 69A(8): p. 779-791. 100. Bahns, J.T., et al., Hole-enhanced Raman scattering. Applied Spectroscopy, 2006. 60(9): p. 989-993. 101. Neugebauer, U., et al., On the way to nanometer-sized information of the bacterial surface by tip-enhanced Raman spectroscopy. ChemPhysChem, 2006. 7(7): p. 1428-1430. 102. Sperrin, M.W. and K. Rogers, The architecture and composition of uroliths. British journal of urology, 1998. 82(6): p. 781-4. 103. Daudon, M., Analysis and classification of calculi: contribution to the etiology of calculous disease. Revue medicale de la Suisse romande, 2004. 124(8): p. 445-53.
225 104. Daudon, M., C.A. Bader, and P. Jungers, Urinary calculi: review of classification methods and correlations with etiology. Scanning Microscopy, 1993. 7(3): p. 1081-106. 105. Williams, J., C., Jr., et al., Calcium Oxalate Calculi Found Attached to the Renal Papilla: Preliminary Evidence for Early Mechanisms in Stone Formation. Journal of Endourology, 2006. In Press. 106. Williams, C.P., et al., Why oral calcium supplements may reduce renal stone disease: report of a clinical pilot study. Journal of clinical pathology, 2001. 54(1): p. 54-62. 107. Stitchantrakul, W., et al., Effects of calcium supplements on the risk of renal stone formation in a population with low oxalate intake. Southeast Asian Journal of Tropical Medicine and Public Health, 2004. 35(4): p. 1028-1033. 108. Chai, W., et al., Oxalate absorption and endogenous oxalate synthesis from ascorbate in calcium oxalate stone formers and non-stone formers. American Journal of Kidney Diseases, 2004. 44(6): p. 1060-1069. 109. Murphy, B.T. and L.N. Pyrah, The composition, structure, and mechanisms of the formation of urinary calculi. British journal of urology, 1962. 34: p. 129-59. 110. Cifuentes, D.L., Study of calculi structure using thin mineralogical sections. Journal d'urologie et de nephrologie, 1977. 83 Suppl 2: p. 592-6. 111. Gault, M.H., et al., Comparison of infrared and wet chemical analysis of urinary tract calculi. Clinica Chimica Acta, 1980. 104(3): p. 349-59. 112. Volmer, M., J. DeVries, and H. Goldschmidt, Infrared Analysis of Urinary Calculi by a Single Reflection Accessory and a Neural Network Interpretation Algorithm. Clinical Chemistry, 2001. 47(7): p. 1287-1296. 113. Ouyang, H., et al., Infrared Microscopic Imaging of Bone: Spatial Distribution of 2- CO3 . Journal of Bone and Mineral Research, 2001. 16(5): p. 893-900. 114. Paschalis, E., et al., Spectroscopic Characterization of Collagen Cross-Links in Bone. Journal of Bone and Mineral Research, 2001. 16(10): p. 1821-1828. 115. Gadaleta, S., et al., Polarized FT-IR Microscopy of Calcified Turkey Leg Tendon. Connective Tissue Research, 1996. 34(3): p. 203-211. 116. Mendelsohn, R., et al., IR Microscopic Imaging of Pathological States and Fracture Healing of Bone. Applied Spectroscopy, 2000. 54(8): p. 1183-1191. 117. Mendelsohn, R., E. Paschalis, and A. Boskey, Infrared Spectroscopy, Microscopy, and Microscopic Imaging of Mineralizing Tissues: Spectra-Structure Correlations from Human Iliac Crest Biopsies. Journal of Biomedical Optics, 1999. 4(1): p. 14-21. 118. Boskey, A.L., et al., Fourier Transform Infrared Microspectroscopic Analysis of Bones of Osteocalcin-Deficient Mice Provides Insight Into the Function of Osteocalcin. Bone, 1998. 23(3): p. 187-196. 119. Kubelka, P. and M.P. Munk, Z. Tech. Phys., 1931. 12: p. 593. 120. Rintoul, L. and P.M. Fredericks, Infrared microspectroscopy of bauxitic pisoliths. Applied Spectroscopy, 1995. 49(11): p. 608-16. 121. Kociba, K.J. and P.K. Gallagher, A Study of Calcium Oxalate Monohydrate Using Dynamic Differential Scanning Calorimetry and Other Thermoanalytical Techniques. Thermochimica Acta, 1996. 282/283: p. 277-296.
226 122. White, R.L. and J. Ai, Variable-Temperature Diffuse Reflectance Infrared Studies of Calcium Oxalate Monohydrate. Applied Spectroscopy, 1992. 46(1): p. 93-9. 123. Kleeberg, J., et al., Studies on the morphology of human uric acid stones. Urological Research, 1981. 9(6): p. 259-61. 124. Lippert, R.J., B.D. Lamp, and M.D. Porter, Specular Reflection Spectroscopy, in Modern Techniques in Applied Molecular Spectroscopy, F.M. Mirabella, Editor. 1998, John Wiley-Interscience: New York. p. 83-126. 125. Tomar, V.S. and H.D. Bist, Infrared Spectrum of Potassium Oxalate Monohydrate at Liquid Nitrogen Temperature. Applied Spectroscopy, 1970. 24(6): p. 598-601. 126. Bellamy, L.J., The Infra-red Spectra of Complex Molecules. 1966, New York: John Wiley & Sons, Inc. 425. 127. Mirabella, F.M., Attenuated Total Reflection Spectroscopy, in Modern Techniques in Applied Molecular Spectroscopy, F.M. Mirabella, Editor. 1998, Wiley- Interscience Publication: New York. p. 127-183. 128. Fernandez, D.C., et al., Infrared Spectroscopic Imaging for Observer-invariant Histopathology. Nature Biotechnology, 2005. 23: p. 469-474. 129. Levin, I.W. and R. Bhargava, Fourier Transform Infrared Vibrational Spectroscopic Imaging: Integration of Microscopy and Molecular Recognition. Annual Review of Physical Chemistry, 2005. 56: p. 429-474. 130. Evan, A., P., et al., Ultrastructural Characteristics of the Kidney-Stone Junction. Science, 2007. Submitted. 131. Anderson, J.C., et al., Analysis of Urinary Calculi Using an Infrared Microspectroscopic Surface Reflectance Imaging Technique. Urological Research, 2007. Accepted. 132. Maurice-Estepa, L., et al., Advantage of zero-crossing-point first-derivative spectrophotometry for the quantification of calcium oxalate crystalline phases by infrared spectrophotometry. Clinica Chimica Acta, 2000. 298(1-2): p. 1-11. 133. Volmer, M., et al., Partial least-squares regression for routine analysis of urinary calculus composition with Fourier transform infrared analysis. Clinical Chemistry (Washington, DC, United States), 1993. 39(6): p. 948-54. 134. Cohen-Solal, F., et al., Automated Fourier Transform Infrared Analysis of Urinary Stones: Technical Aspects and Example of Procedures Applied to Carbapatite/Weddellite Mixtures. Applied Spectroscopy, 2004. 58(6): p. 671-678. 135. Axe, K., M. Vejgarden, and P. Persson, An ATR-FTIR spectroscopic study of the competitive adsorption between oxalate and malonate at the water-goethite interface. Journal of Colloid and Interface Science, 2006. 294(1): p. 31-37. 136. Kordek, A., W. Podraza, and R. Czajka, Reliability of semiquantitative determination of procalcitonin serum concentrations in neonates. Diagnostic Microbiology and Infectious Disease, 2006. 56(1): p. 31-34. 137. Xu, W. and Z. Yi, Changes of bone mineral density and metabolism in patients with type 2 diabetes mellitus of different traditional Chinese medicine syndrome types. Zhongguo Linchuang Kangfu, 2005. 9(47): p. 182-184. 138. Fabrizi, F., V. Dixit, and P. Martin, Meta-analysis: anti-viral therapy of hepatitis B virus-associated glomerulonephritis. Alimentary Pharmacology and Therapeutics, 2006. 24(5): p. 781-788.
227 139. Turkun, L.S. and B.O. Aktener, Twenty-four-month clinical evaluation of different posterior composite resin materials. Journal of the American Dental Association, JADA, 2001. 132(2): p. 196-203. 140. Harrick, N.J., Internal Reflection Spectroscopy, ed. N.J. Harrick. 1987, Ossining. 327. 141. Yernault, J.C., et al., "Solid" mediastinal bronchogenic cyst: mineralogic analysis. 1986: United States. p. 73-4. 142. Van den Kerkhof, A.M. and U.F. Hein, Fluid Inclusion Petrography. Lithos, 2001. 55: p. 27-47. 143. Anderson, J., et al., Quantitative Analysis of Renal Stone Particles using ATR Infrared Microspectroscopy and their Corresponding Size Detection Limits. Applied Spectroscopy, 2007. To be submitted. 144. Salvadori, B., et al., Evaluation of Gypsum and Calcium Oxalates in Deteriorated Mural Paintings by Quantitative FTIR Spectroscopy. Spectroscopy Letters, 2003. 36(5 & 6): p. 501-513. 145. Harris, D.C., Quantitative Chemical Analysis. 7th ed, ed. J. Fiorillo. 2007, New York: W. H. Freeman and Co. 146. Roberts, S.T., J.J. Loparo, and A. Tokmakoff, Characterization of Spectral Diffusion from Two-dimensional Line Shapes. Journal of Chemical Physics, 2006. 125(8): p. 084502/1-084502/8. 147. Capasso, G., et al., Uric acid and the kidney: Urate transport, stone disease and progressive renal failure. Current Pharmaceutical Design, 2005. 11(32): p. 4153- 4159. 148. EPA, U., Method 8430. 1996, Environmental Protection Agency. p. Section 9.2.1. 149. Krock, K. and C. Wilkins, Determination of Minimum Idetnifiable Quantities for Target Analysis of Semivolatile Organic Compounds. Applied Spectroscopy, 1992. 46(11): p. 1621-1625. 150. Diem, M., et al., Infrared Microspectroscopy of Cells and Tissue: Infrared Spectra and Infrared Spectral Maps of Human Tissues. Inst. Phys. Conf., 2000. Ser. No 165(Symposium 1): p. 77-78. 151. Diem, M., L. Chiriboga, and H. Yee, Infrared Spectroscopy of Human Cells and Tissue. VIII: Strategies for Analysis of Infrared Tissue Mapping and Applications to Liver Tissue. Biospectroscopy, 2000. 57: p. 282-290. 152. Diem, M., et al., IR Spectra and IR Spectral Maps of Individual Normal and Cancerous Cells. Biopolymers (Biospectroscopy), 2002. 67: p. 349-353. 153. Diem, M., S. Boydston-White, and L. Chiriboga, Infrared Spectroscopy of Cells and Tissues: Shining Light onto a Novel Subject. Applied Spectroscopy, 1999. 53(4): p. 148A-161A. 154. Aparicio, S., et al., Optimal Methods for Processing Mineralized Tissues for Fourier Transform Infrared Microspectroscopy. Calcified Tissue International, 2002. 70: p. 422-429. 155. Mendelsohn, R., et al., Infrared Microspectroscopic Imaging Maps the Spatial Distribution of Exogenous Molecules in Skin. Journal of Biomedical Optics, 2003. 8(2): p. 185-190.
228 156. Marcott, C., et al., Infrared Microspectroscopic Imaging of Biomineralized Tissues using a Mercury-Cadmium-Telluride Focal-Plane Array Detector. Phosphorus, Sulfur and Silicon, 1999. 144-146: p. 417-420. 157. Chiriboga, L., H. Yee, and M. Diem, Infrared Spectroscopy of Human Cells and Tissues. Part VII: FT-IR Microspectroscopy of DNase- and RNase-Treated Normal, Cirrhotic, and Neoplastic Liver Tissue. Applied Spectroscopy, 2000. 54(4): p. 480-485. 158. Ouyang, H., et al., Fourier Transform Infrared Microscopic Imaging: Effects of Estrogen and Estrogen Deficiency on Fracture Healing in Rat Femurs. Applied Spectroscopy, 2004. 58: p. 1-9. 159. Kandil, S.H., et al., Computerized Infrared Analysis of Renal Stone Calculi. Computer Enhanced Spectroscopy, 1986. 3(4): p. 171-7. 160. Haka, A., L. Kidder, and E.N. Lewis, A Combined Spectroscopic Imaging and Chemometric Approach for Automatically Partitioning Tissue Types in Human Prostate Tissue Biopsies. Proceedings of SPIE, 2001. 4259: p. 47-55. 161. Estepa, L., et al., Fourier Transform Infrared Microscopy Identification of Crystal Deposites in Tissues. American Journal of Clinical Pathology, 1996. 105: p. 576- 582. 162. Cohen, J. and C. Virnelson, FT-IR Spectrometer Window and Method, in USPTO. 1992, Tremco Incorporated: US. 163. Danielson, N., et al., Zirconium Oxide Stationary Phase for Thin-Layer Chromatography with Diffuse Reflectance Fourier Transform Infrared Detection. Applied Spectroscopy, 1992. 64(18): p. 2183-2186. 164. Patterson, B., N. Danielson, and A. Sommer, Attenuated Total Internal Reflectance Infrared Microspectroscopy as a Detection Technique for High- Performance Liquid Chromatography. Analytical Chemistry, 2003. 75: p. 1418- 1424. 165. Sommer, A.J., et al., Attenuated Total Internal Reflection Infrared Mapping Microspectroscopy Using an Imaging Microscope. Applied Spectroscopy, 2001. 55(3): p. 252-256. 166. Brown, T., et al., Chemistry: The Central Science. 9th ed. 2003: Pretice Hall. 167. Amis, E. and J. Rumble, Certificate of Analysis, SRM 2910, in Standard Reference Material Data Sheet, N.I.o.S.a. Technology, Editor. 2003, U.S. Department of Commerce NIST. 168. Osborne, C., Statistical Calibration: A Review. International Statistical Review, 1991. 59(3): p. 309-336. 169. Bellin, M.-F., et al., Helical CT evaluation of the chemical composition of urinary tract calculi with a discriminant analysis of CT-attenuation values and density. European radiology, 2004. 14(11): p. 2134-40. 170. Peuchant, E., et al., Discriminant Analysis of Urinary Calculi by Near-Infrared Reflectance Spectroscopy. Clinica chimica acta; international journal of clinical chemistry, 1992. 205(1-2): p. 19-30. 171. Sutor, D.J., Difficulties in the Identification of Components of Mixed urinary Calculi Using the X-Ray Powder Method. BJU International, 1968. 40: p. 29-32. 172. Thongboonkerd, V., T. Semangoen, and S. Chutipongtanate, Factors determining types and morphologies of calcium oxalate crystals: Molar concentrations,
229 buffering, pH, stirring and temperature. Clinica Chimica Acta, 2006. 367(1-2): p. 120-131. 173. Trinchieri, A., et al., Calcium stone disease: a multiform reality. Urological Research, 2005. 33(3): p. 194-198. 174. Daudon, M. and P. Jungers, Drug-induced lithiasis. Revue medicale de la Suisse romande, 2004. 124(8): p. 483-90. 175. Kazarian, S.G. and K.L.A. Chan, Applications of ATR-FTIR spectroscopic imaging to biomedical samples. Biochimica et Biophysica Acta, Biomembranes, 2006. 1758(7): p. 858-867. 176. Helmy, R., et al., Characterization and Quantitation of Aprepitant Drug Substance Polymorphs by Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy. Analytical Chemistry, 2003. 75(3): p. 605-611. 177. Daudon, M., Personal Communication, J.C. Anderson, Editor. 2004: Oxford, OH. 178. Wadayama, T., et al., In situ UV-visible and IR absorption study of the aggregation of silver colloidal particles via chemisorption of octadecanethiol. Surface and Interface Analysis, 2004. 36(7): p. 613-618. 179. McArdle, P., et al., Determination of the polymorphic forms of bicifadine hydrochloride by differential scanning calorimetry-thermogravimetric analysis, X-ray powder diffraction, attenuated total reflectance-infrared spectroscopy, and attenuated total reflectance-near-infrared spectroscopy. Applied Spectroscopy, 2005. 59(11): p. 1365-1371. 180. Chan, K.L.A., N. Elkhider, and S.G. Kazarian, Spectroscopic imaging of compacted pharmaceutical tablets. Chemical Engineering Research and Design, 2005. 83(A11): p. 1303-1310. 181. Boyer, C., et al., Direct determination of niflumic acid in a pharmaceutical gel by ATR/FTIR spectroscopy and PLS calibration. Journal of Pharmaceutical and Biomedical Analysis, 2006. 40(2): p. 433-437. 182. Lin, Z., et al., Real-time endpoint monitoring and determination for a pharmaceutical salt formation process with in-line FT-IR spectroscopy. Journal of Pharmaceutical and Biomedical Analysis, 2006. 41(1): p. 99-104. 183. Wartewig, S. and R.H.H. Neubert, Pharmaceutical applications of Mid-IR and Raman spectroscopy. Advanced Drug Delivery Reviews, 2005. 57(8): p. 1144- 1170. 184. Chan, K.L.A., S.V. Hammond, and S.G. Kazarian, Applications of Attenuated Total Reflection Infrared Spectroscopic Imaging to Pharmaceutical Formulations. Anal. Chem., 2003. 75(9): p. 2140-2146. 185. Sastre, T., J. and S.H.C. van Hattum, Quantitative analysis of active compounds in pharmaceutical preparations by use of attenuated total-reflection Fourier transform mid-infrared spectrophotometry and the internal standard method. Fresenius' Journal of Analytical Chemistry, 2001. 371(4): p. 532-535. 186. Zhou, L., et al., Determination of the enantiomeric excess of an M3 antagonist drug substance by chemometric analysis of the IR spectra of different guest-host complexes. Chirality, 2006. 18(5): p. 306-313. 187. Daudon, M., Personal Communication, J. Anderson, Editor. 2006: Indianapolis. p. Particle Size.
230 188. Bhutto, A.A., D. Vesely, and B.J. Gabrys, Miscibility and interactions in polystyrene and sodium sulfonated polystyrene with poly(vinyl methyl ether) PVME blends. Part II. FTIR. Polymer, 2003. 44(21): p. 6627-6631. 189. Rydzak, J., SEM analysis, A. Sommer, J., Editor. 2006: Oxford. p. Phone conversation. 190. Practical Handbook of Physical Properties of Rocks and Minerals, ed. R.S. Carmichael. 1989, Boca Raton: CRC Press. 741. 191. Kortüm, G., Reflectance Spectroscopy. 1969, New York: Springer-Verlag. 192. Salari, A. and R.E. Young, Application of Attenuated Total Reflectance FTIR Spectroscopy to the Analysis of Mixtures of Pharmaceutical Polymorphs. International Journal of Pharmaceutics, 1998. 163(1-2): p. 157-166. 193. Kajander, O. and N. Ciftcioglu, Nanobacteria: An Alternative Mechanism for Pathogenic Intra- and Extracellular Calcification and Stone Formation. Proceedings of the National Academy of Sciences, USA, 1998. 95: p. 8274-8279. 194. Sommer, A. and O. Kakander, Nanobacteria-Induced Kidney Stone Formaiton: Novel Paradigm Based on the FERMIC Model. Crystal Growth and Design, 2002. 2(6): p. 563-565. 195. Connors, B., A., et al., Renal Nerves Mediate Changes in Contralateral Renal Blood Flow After Extracorporeal Shockwave Lithotripsy. Nephron, 2003. 95(4): p. p67-75. 196. Costa-Bauza, A., et al., An experimental study on residual lithiasis after shock wave lithotripsy. Urological Research, 2005. 33(1): p. 51-56. 197. Parks, J.H., et al., Clinical implications of abundant calcium phosphate in routinely analyzed kidney stones. Kidney International, 2004. 66(2): p. 777-785. 198. Curhan, G.C., et al., Comparison of dietary calcium with supplemental calcium and other nutrients as factors affecting the risk for kidney stones in women. Annals of Internal Medicine, 1997. 126(7): p. 497-504. 199. Costa-Bauza, A., et al., Factors affecting the regrowth of renal stones in vitro: a contribution to the understanding of renal stone development. Scandinavian Journal of Urology and Nephrology, 2005. 39(3): p. 194-199. 200. Barbey, F., et al., Infective lithiasis. 2004, Division de Nephrologie, CHUV, Lausanne.: Switzerland. p. 465-9. 201. Hess, B., Acid-base metabolism: implications for kidney stosne formation. Urological Research, 2006. 34(2): p. 134-138. 202. Parks, J.H., et al., A Single 24-hour Urine Collection is Iinadequate for the Medical Evaluation of Nephrolithiasis. Journal of Urology, 2002. 167(4): p. 1607- 12. 203. Ward, M., et al. Probing Adhesion Forces and Macromolecule Binding at Calcium Oxalate Single Crystal Surfaces: Toward an Understanding of Kidney Stone Formation. in 227th ACS National Meeting. 2004. Anaheim, CA, United States,. 204. Voss, S., et al., Intestinal oxalate absorption is higher in idiopathic calcium oxalate stone formers than in healthy controls: measurements with the [13C2]oxalate absorption test. Journal of Urology (New York, NY, United States), 2006. 175(5): p. 1711-1715.
231 205. Nicolaidou, P. and A. Papadopoulou, Idiopathic hypercalciuria in children. Current Pediatric Reviews, 2006. 2(1): p. 93-98. 206. Favus, M.J. and F.L. Coe, Idiopathic hypercalciuria and nephrolithiasis. Vitamin D (2nd Edition), 2005. 2: p. 1339-1354. 207. Schwille, P.O., A. Schmiedl, and M. Manoharan, Is calcium oxalate nucleation in postprandial urine of males with idiopathic recurrent calcium urolithiasis related to calcium phosphate nucleation and the intensity of stone formation? Studies allowing insight into a possible role of urinary free citrate and protein. Clinical Chemistry and Laboratory Medicine, 2004. 42(3): p. 283-293. 208. Evan, A., et al., Sequential Analysis of Kidney Stone Formation in the Aprt Knockout Mouse. Kidney International, 2001. 60: p. 910-923. 209. Evan, A.P., et al., Renal crystal deposits and histopathology in patients with cystine stones. Kidney International, 2006. 69(12): p. 2227-2235. 210. Williams, J., C., et al., Calcium oxalate calculi found attached to the renal papilla: preliminary evidence for early mechanisms in stone formation. J Endourol FIELD Full Journal Title:Journal of endourology / Endourological Society, 2006. 20(11): p. 885-90. 211. Williams, J., C., Jr., et al., High Resolution Detection of Internal Structure of Renal Calculi by Helical Computerized Tomography. Journal of Urology, 2002. 167(1): p. 322-6. 212. Coe, F.L., Uric Acid and Calium Oxalate Nephrolithiasis. Kidney International, 1983. 24: p. 392-403. 213. Aizenberg, J., et al., Factors Involved in the Formation of Amorphous and Crystalline Calcium Carbonate: A Study of an Ascidian Skeleton. Journal of the American Chemical Society, 2002. 124(1): p. 32-39. 214. Coe, F., L., Personal Communication, J. Anderson, H. Stahl, and A. Sommer, Editors. 2006: Indianapolis. p. Conference Presenatation. 215. Brenner and Rector, The Kidney. 7th ed. 2004: W.B. Saunders Company. 216. Yasue, T., Histochemical Identification of Calcium Oxalate. Acta Histochem. Cytochem., 1969. 2(3): p. 83-95. 217. Bernath, P.F., Laboratory Astrophysics and Moleuclar Astronomy of Pure Carbon Molecules. Advanced Space Research, 1995. 15(3): p. 15-23. 218. Huheey, J.E., E.A. Keiter, and R.L. Keiter, Inorganic Chemistry. 4th ed. 1993: Harper Collins. 219. Chernov, A.A., et al., Step and kink dynamics in inorganic and protein crystallization. MRS Bulletin, 2004. 29(12): p. 927-933. 220. Asta, M., F. Spaepen, and J.F. van der Veen, Solid-liquid interfaces: molecular structure, thermodynamics, and crystallization. MRS Bulletin, 2004. 29(12): p. 920-922. 221. Hayashi, Y., High resolution electron microscopy of a small crack at the superficial layer of enamel. Journal of electron microscopy, 1994. 43(6): p. 398- 401. 222. Sohnel, O. and F. Grases, Calcium Oxalate Monohydrate Renal Calculi. Formation and Development Mechanism. Advances in Colloid and Interface Science, 1995. 59: p. 1-17.
232 223. Patterson, B. and G. Havrilla, Attenuated Total Internal Reflection Infrared Microspectroscopic Imaging Using a Large Radius Germanium Internal Reflection Element and a Linear Array Detector. Applied Spectroscopy, 2006(accepted). 224. Stephens, T., Listen Closely on how to Find Leaky Pipes. 2005, http://www.djc.com/news/ae/11173377.html: A&E Perspectives. 225. Mast, D., Personal Communication, J. Anderson, Editor. 2006: Oxford. 226. EPA, Lead and Copper Rules, in EPA 816-F-05-030. 2005, Environmental Protection Agency Office of Water. 227. Schock, M.R., et al., Replacing polyphosphate with silicate to solve lead, copper, and source water iron problems. Journal - American Water Works Association, 2005. 97(11): p. 84-93. 228. Lytle, D.A., P. Sarin, and V.L. Snoeyink, The effect of chloride and orthophosphate on the release of iron from a cast iron pipe section. Journal of Water Supply: Research and Technology--AQUA, 2005. 54(5): p. 267-281. 229. Dalton, L.W., Pinhole Leaks Plague Copper Pipes. Chemical and Engineering News, 2003. 81(33): p. 51-53. 230. Energy, U.D.o., Copper Pipe Corrosion. Ask a Scientist, 2004. 231. Marshall, B., J. Rushing, and M. Edwards, Confirming the role of aluminum solids and chlorine in copper pitting corrosion. Proceedings - Annual Conference, American Water Works Association, 2003: p. 1414-1426. 232. Bouchard, M. and D.C. Smith, Catalogue of 45 Reference Raman Spectra of Minerals Concerning Research in Art History or Archaeology, Especially on Corroded Metals and Coloured Glass. Spectrochimica Acta Part A, 2003. 59: p. 2247-2266. 233. CorrosionSource.com, Pipe Corrosion And Its Growing Threat To Office Building and Plant Operations. Corrosioneering, 2002. Online(www.corrosionsource.com/corrosioneering/journal). 234. Edwards, M., M.R. Schock, and T.E. Meyer, Alkalinity, pH, and copper corrosion byproduct release. Journal - American Water Works Association, 1996. 88(3): p. 81-94. 235. Nagata, H., M. Matsunaga, and K. Hosokawa, IR analysis of crystallization process of corrosion products in galvanized steel pipe for water service. Proceedings - Electrochemical Society, 1991. 91-7(Proc. Symp. Appl. Surf. Anal. Methods Environ./Mater. Interact., 1990): p. 538-47. 236. Izumi, J., et al., Chemical state analyses of metal oxides at oxygenated water treatment. Interact. Iron-Based Mater. Water Steam, Proc. Int. Conf., 1993(TR- 102101): p. 22/1-22/19. 237. Martens, W., et al., Raman spectroscopic study of the basic copper sulphates- implications for copper corrosion and \"bronze disease\". Journal of Raman Spectroscopy, 2003. 34(2): p. 145-151. 238. Sridhar, N., et al., Effects of water and gas compositions on the internal corrosion of gas pipelines - modeling and experimental studies. Corrosion (Houston, TX, United States), 2001. 57(3): p. 221-235.
233 239. Purcell, F. and R. Heidersbach, The use of the Raman microprobe for failure analysis of oil field drill pipe. Microbeam Analysis (San Francisco), 1985. 20th: p. 57-9. 240. Lytle, D.A. and V.L. Snoeyink, The effect of pH and dissolved inorganic carbon on the properties of iron colloidal suspensions. Journal of Water Supply: Research and Technology--AQUA, 2003. 52(3): p. 165-180. 241. Edwards, M. and A. Dudi, Role of chlorine and chloramine in corrosion of lead- bearing plumbing materials. Journal - American Water Works Association, 2004. 96(10): p. 69-81. 242. Edwards, M., Controlling corrosion in drinking water distribution systems: a grand challenge for the 21st century. Water science and technology : a journal of the International Association on Water Pollution Research, 2004. 49(2): p. 1-8. 243. Nakamoto, K., Infrared and Raman Spectra of Inorganic and Coordination Compoundss. 4th ed. 1986, New York: Wiley-Interscience. 484. 244. Ross, S.D., Inorganic Infrared and Raman Spectra. 1972, Maidenhead: McGraw- Hill. 414. 245. Padival, N.A., W.A. Kimbell, and J.A. Redner, The use of iron salts to control dissolved sulfide in districts' trunk sewers. Critical Issues in Water and Wastewater Treatment, Proceedings of the National Conference on Environmental Engineering -- Boulder, Colo., July 11-13, 1994, 1994: p. 743-50. 246. Meyer, T.E. and M. Edwards, Effect of alkalinity on copper corrosion. Critical Issues in Water and Wastewater Treatment, Proceedings of the National Conference on Environmental Engineering -- Boulder, Colo., July 11-13, 1994, 1994: p. 9-16. 247. Morton, S.C., Y. Zhang, and M.A. Edwards, Implications of nutrient release from iron metal for microbial regrowth in water distribution systems. Water Research, 2005. 39(13): p. 2883-2892. 248. Rushing, J.C. and M. Edwards, Effect of aluminum solids and chlorine on cold water pitting of copper. Corrosion Science, 2004. 46(12): p. 3069-3088. 249. Rushing, J.C. and M. Edwards, The role of temperature gradients in residential copper pipe corrosion. Corrosion Science, 2004. 46(8): p. 1883-1894. 250. Sun, Y., S.C. Morton, and M. Edwards, Equilibrium and kinetic aspects of anaerobic iron corrosion in the presence of phosphate. Proceedings - Water Quality Technology Conference, 2002: p. 2379-2396. 251. Dick, L.F.P. and L.M. Rodrigues, Influence of humic substances on the corrosion of the API 5LX65 steel. Corrosion (Houston, TX, United States), 2006. 62(1): p. 35-43. 252. Qin, Z., et al., Localized dissolution of millscale-covered pipeline steel surfaces. Corrosion (Houston, TX, United States), 2004. 60(10): p. 906-914. 253. Imai, H., In-situ analysis of metal corrosion products in aqueous solution. Hyomen, 2003. 41(6): p. 185-201. 254. Brown, C.W., In situ analysis of corrosive and passive surfaces by laser-excited Raman spectroscopy. 1984, Dep. Chem.,Univ. Rhode Island,Kingston,RI,USA. p. 145 pp.
234 255. Bolotov, A.V., N.P. Zaretskaya, and V.N. Musolin, Analysis of corrosion films by Raman spectroscopy. Zhurnal Prikladnoi Khimii (Sankt-Peterburg, Russian Federation), 1984. 57(3): p. 490-3. 256. Ezawa, M., et al., Examination of soft steel sheets, and products using soft steel sheets examined by the method. 1996, (Hitachi Ltd, Japan). Application: JP. 257. Wehling, B., et al., Investigation of pigments in medieval manuscripts by micro- Raman spectroscopy and total reflection x-ray fluorescence spectrometry. Mikrochimica Acta, 1999. 130(4): p. 253-260. 258. Silverstein, R.M. and F.X. Webster, Spectroscopic Identification of Organic Compounds. 6th ed. 1998, New York: John Wiley and Sons. 259. Frost, R.L., et al., Raman spectroscopy of the basic copper phosphate minerals cornetite, libethenite, pseudomalachite, reichenbachite and ludjibaite. Journal of Raman Spectroscopy, 2002. 33(4): p. 260-263. 260. Andrikopoulos, K.S., et al., Pressure dependence of the Boson peak in glassy As2S3 studied by Raman scattering. Journal of Non-Crystalline Solids, 2006. 352(42-49): p. 4594-4600. 261. Beck, J., S. Schlueter, and N. Zotov, The polycationic main group element clusters (As3S5)+ and (As3Se4)+ - syntheses from chloroaluminate melts, crystal structures, and vibrational spectra. Zeitschrift fuer Anorganische und Allgemeine Chemie, 2005. 631(12): p. 2450-2456. 262. Griffith, W.P., Raman studies on rock-forming minerals. II. Minerals containing MO3, MO4, and MO6 groups. Journal of the Chemical Society [Section] A: Inorganic, Physical, Theoretical, 1970(2): p. 286-91.
235
This Page Intentionally Left Blank In Order To Practice For Government
Bureaucracy.
The End.