USF Libraries
USF Digital Collections

UV laser and LED induced fluorescence spectroscopy for detection of trace amounts of organics in drinking water and wate...

MISSING IMAGE

Material Information

Title:
UV laser and LED induced fluorescence spectroscopy for detection of trace amounts of organics in drinking water and water sources
Physical Description:
Book
Language:
English
Creator:
Sharikova, Anna V
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Water fluorescence
Laser-induced fluorescence
Reverse osmosis
Water quality monitoring
Online real time reagentless system
Dissertations, Academic -- Physics -- Doctoral -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: A UV Laser Induced Fluorescence (LIF) system, previously developed in our laboratory, was modified and used for a series of applications related to the development and optimization of UV LIF spectroscopic measurements of trace contaminants in drinking water and other water sources. Fluorescence spectra of a number of water samples were studied, including those related to the reverse osmosis water treatment and membrane fouling, domestic and international drinking water, industrial toxins, bacterial spores, as well as several fluorescence standards. Of importance was that the long term detection of the trace level of Dissolved Organic Compounds (DOC) was measured, for the first time to our knowledge, over a one week period and with a time resolution of 2.5 minutes. A comparison of LIF emission using both 266 nm and 355 nm excitation was also made for the first time. Such real-time and continuous measurements are important for future water treatment control.The LIF system was modified to accommodate UV Light Emitting Diodes (LED) as alternative excitation sources, and tested for the detection of trace organic species in water. In addition, a compact system using LED excitation and a spectrometer was xviii developed and underwent initial testing. The original LIF system had two laser sources, 266 nm and 355 nm. The additional sources incorporated in the system were UV LEDs emitting at 265 nm, 300 nm, 335 nm and 355 nm. The LED spectral emission was studied in detail, in terms of spectral variability and power output. It was found that all LEDs had some emission in the visible spectrum, and an optical filter was used to remove it. The signal-to-noise ratio for the LED-based systems was determined and compared with that of the LIF system. The fluorescent signal of the LED-based system was smaller by 1 to 2 orders of magnitude, despite the fact that the LED pulse energy was 2 to 3 orders of magnitude less than the laser's.As such, the fluorescent signal from the LED was greater than expected. Therefore, a UV LED may be a compact and much cheaper optical source for future water measurement instruments.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2009.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Anna V. Sharikova.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 232 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002317271
oclc - 659877966
usfldc doi - E14-SFE0003013
usfldc handle - e14.3013
System ID:
SFS0027330:00001


This item is only available as the following downloads:


Full Text

PAGE 1

UV Laser and LED Induced Fluorescence Spect roscopy for Detection of Trace Amounts of Organics in Drinking Water and Water Sources by Anna V. Sharikova A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Physics College of Arts and Sciences University of South Florida Major Professor: Dennis K. Killinger, Ph.D. Myung K. Kim, Ph.D. Nicholas Djeu, Ph.D. Hariharan Srikanth, Ph.D. Date of Approval: May 21, 2009 Keywords: water fluorescence, laser-induced fl uorescence, reverse osmosis, water quality monitoring, online real ti me reagentless system Copyright 2009 Anna V. Sharikova

PAGE 2

DEDICATION To my mother

PAGE 3

ACKNOWLEDGEMENTS I would like to express my gratitude to Dr. Dennis K. Killinger, my advisor, for his guidance and patience throughout this endeavor. I also thank Dr. Myung K. Ki m, Dr. Nicholas Djeu and Dr. Hariharan Srikanth for serving on my dissertation committee and Dr Don Morel for chairing the examination. I thank Dr. Robert Carnaha n, Miles Beamguard, and Dhanan jaya Niriella from the College of Engineering for collaboration on reverse osmosis studies, and Dr. Audrey Levine and Panagiotis Amitzoglou from the College of Engineering for collaboration on total organic carbon studies. I thank the staff of the Department: Mary Ann, Daisy, Kimberly, Phil, and Robert for their help. I would also like to thank Anali for colla boration on DPA and other samples, and Marek for help with initial LED electronics. To Anali, Avis and Dennis, thanks for friendship and intere sting discussions. Finally, to Alex goes my eternal thankf ulness for companionship and inspiration. This work was supported in part by the US Army Research Office under grant # W911NF-05-1-0431.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES vi LIST OF FIGURES vii ABSTRACT xvii CHAPTER ONE: INTRODUCTION 1 CHAPTER TWO: BACKGROUND IN FORMATION ON WATER QUALITY AND MEASUREMENT TECHNIQUES 6 2.1 Classes of Contaminants in Drinking Water 6 2.2 Approved Water Testing Techni ques for Organic Compounds 10 2.2.1 Chromatographic Methods 10 2.2.2 Other Methods 12 CHAPTER THREE: BACKGROUND IN FORMATION ON LASER-INDUCED FLUORESCENCE, ABSORPTION AND OPTICAL SPECTROSCOPY OF WATER AND TRACE SPECIES, INCL UDING PREVIOUS LIF WATER STUDIES AT UNIVERSITY OF SOUTH FLORIDA 16 3.1 Scattering, Absorption a nd Fluorescence of Light 16 3.1.1 Rayleigh, Mie and Raman Scattering 17 3.1.2 Absorption of Optical Radiation 18 3.1.3 Fluorescence versus NonRadiative Energy Transfer 19 3.1.3.1 Relaxation Processes in Complex Molecules 19 3.1.3.2 Characteristics of Fluorescence 23 3.2 Use of Fluorescence Measurement in Water Studies 24

PAGE 5

ii 3.2.1 Methods and Examples of Fluorescence Measurements 25 3.2.2 Fluorescence of Organic Compounds 25 3.3 Previous LIF Water Studies at University of South Florida 27 CHAPTER FOUR: UV LIF SY STEM MEASUREMENTS OF CONTAMINANTS IN DRINKING AND REVERSE OSMOSIS PROCESSED WATER 38 4.1 Portable LIF System 38 4.1.1 System Modifications 42 4.2 LIF Measurements of Reverse Osmosis Processed Water 46 4.2.1 Reverse Osmosis 46 4.2.1.1 Principles of Reverse Osmosis System 46 4.2.1.2 RO Units at USF 47 4.2.2 LIF Measurements of Ground Water Treated by Reverse Osmosis 47 4.2.3 Membrane Fouling and RO Efficiency 50 4.2.3.1 Effect of Scaling on Water Treatment 53 4.2.3.2 Effect of Clay Part icles in RO System with Scaling 53 4.3 LIF Measurements of Various Water Samples 58 4.3.1 Domestic and Intern ational Drinking Water 58 4.3.2 Bacterial Spores in Distilled Water 65 4.3.3 Malathion in Distilled Water 68 4.4 General Observations Regarding Deep UV LIF Spectra 68 CHAPTER FIVE: UV LIF LONG-TERM CONTINUOUS MONITORING OF TAP WATER 71 5.1 Experimental System for LongTerm Continuous Monitoring of Tap Water 71

PAGE 6

iii 5.2 Flowing Tap Water Monitoring 71 5.2.1 LIF Signal of Flowing Ta p Water over Twelve Hours 73 5.2.2 Peak Fluorescence Channel versus Raman Channel 78 5.2.3 Comparison with Recirc ulated Tap Water Sample 81 5.2.4 Week-Long Tap Water Monitoring 81 5.2.4.1 Running Tap Water 84 5.2.4.2 Recirculated Tap Water 91 5.3 Other Long-Term Experiments 101 5.3.1 Distilled Water 104 5.3.2 Tannic Acid Representing TOC 104 5.3.3 Chlorinated Tannic Acid 113 CHAPTER SIX: RESEAR CH AND DEVELOPMENT OF LED DRIVER 125 6.1 Initial Drivers for UV LEDs 126 6.1.1 Initial Continuous Wave Mode Operation 126 6.1.2 Initial Pulsed Mode Operation 128 6.2 Motivation for LED Driver Optimization 128 6.3 Optimized Current Driver for LEDs 129 6.3.1 Driver Design 131 6.3.1.1 Optimized Driver Schematic 131 6.3.2 Testing Current Driver Performance with a Dummy Load 135 6.3.3 LED Performance with Optimized Current Driver 135 6.3.3.1 Continuous Mode 139 6.3.3.2 Pulsed Mode 139

PAGE 7

iv CHAPTER SEVEN: OPTICAL CHAR ACTERIZATION OF UV LEDS 144 7.1 Specifications for the UV LEDs 144 7.2 Experimental Systems for Optic al Characterization of UV Light Emitting Diodes 146 7.3 LED Emission Spectra 146 7.3.1 Out-of-Band Emission 152 7.3.2 Spectra with Visi ble-Blocking Filter 152 7.4 LED Emission: Electrical Modul ation versus Mechanical Chopping 158 7.5 LED Emission in Three Different Modes of Operation 158 7.5.1 Unfiltered Spectra 162 7.5.2 Spectra with Visi ble-Blocking Filter 162 7.6 LED Emission: Comparison of Two Pulsed Regimes 162 7.6.1 Output Intensity versus Driving Current 171 7.6.2 Pulse Power Output 171 CHAPTER EIGHT: COMPARISON OF FLUORESCENCE EXCITATION BY LED IN CW MODE WITH LASER INDUCED FLUORESCENCE USING A LABORATORY BENCH-TO P LED-IF/LIF SYSTEM 179 8.1 Experimental Setup for Comparison of Fluorescence Excitation by CW LEDs with Laser I nduced Fluorescence 179 8.2 Comparison of LED and Laser Output 181 8.3 Measurements of LIF and LED-IF in Drinking and Natural Water 181 8.3.1 Lake Water 185 8.3.2 Tap Water 192 8.3.3 Distilled Water 192 8.4 Detection of Chemicals Using LIF and LED-IF 197 8.4.1 Tonic with Quinine 197

PAGE 8

v 8.4.2 Coumarin Laser Dye 199 8.4.3 Spores of Bacillus Subtilis 199 8.4.4 Terbium-Doped Dipicolinic Acid 199 CHAPTER NINE: COMPARISON OF FLUORESCENCE EXCITATION BY LED IN PULSED MODE WITH LASER INDUCED FLUORESCENCE USING THE PORTABLE LIF SYSTEM 203 9.1 Experimental Setup for Comparison of Fluorescence Excitation by Pulsed LEDs with Laser Induced Fluorescence 203 9.2 Comparison of LED and Laser Output 206 9.3 Measurements of LIF and LED-IF in Lake Water 206 CHAPTER TEN: CONCLUSION S AND FUTURE WORK 212 REFERENCES 215 APPENDICES 219 Appendix A: CALCULATIONS FOR LED DRIVER DESIGN 220 Appendix B: DATA COLLECTION AND PROCESSING SOFTWARE 223 Appendix C: SPECTROMETER CORRECTION CURVES 229 ABOUT THE AUTHOR End Page

PAGE 9

vi LIST OF TABLES Table 2.1 Water properties and appropriate laboratory measurement techniques. 7 Table 2.2 Water contaminants and detection methods. 8 Table 3.1 Absorption and emission wavele ngths of some organic substances. 28 Table 4.1 Specifications for the 266 nm and 355 nm microchip lasers. 40 Table 4.2 Specifications for the bandpa ss interference and cutoff absorption filters in LIF system and PMT efficiency. 41 Table 4.3 Specifications for the Hamamatsu PMT. 43 Table 4.4 Typical settings for the LIF experiments. 44 Table 6.1 Nominal values of resistances and other parameters of the optimized LED driver. 134 Table 7.1 Specifications and output ch aracteristics for the UV-TOP LEDs. 145 Table 7.2 Specifications for the ST-2000 sp ectrometer (Ocean Optics, Inc.). 148 Table 8.1 Specifications for the USB-2000 spectrometer (Ocean Optics, Inc.). 184

PAGE 10

vii LIST OF FIGURES Figure 2.1 Schematic diagram of a liqui d chromatography apparatus (adapted from Nollet (2000)). 13 Figure 2.2 Chromatogram of tap wate r measured using high performance liquid chromatography with fast-scanning fluorescence spectrometer (reprinted with permission from Beltran, Guiteras, and Ferrer (1998), Copyright 1998 American Chemical Society). 14 Figure 3.1 Optical absorpti on spectrum of pure water (a dapted from Pope and Fry (1997)). 20 Figure 3.2 Jablonski diagram of energy levels and competing processes in a complex molecule (adapted from Albani (2007)). 21 Figure 3.3 Fluorescence of organic com pounds in water: (a) river water; (b) sea water; (c) photobleached sea wa ter ((a) and (c) are reprinted with permission from Coble ( 2007), Copyright 2007 American Chemical Society; (b) is repr inted from Coble (1996), Copyright (1996), with permission from Elsevier). 26 Figure 3.4 Schematic diagram of the la boratory LIF setup developed at USF (from Sivaprakasam and Killinger (2003, JOSA B), reprinted with permission from Optical Society of America). 29 Figure 3.5 LIF of natural water samples recorded with laboratory setup (from Killinger and Sivaprakasam (2003, Appl. Opt.), reprinted with permission from Optical Society of America). 30 Figure 3.6 LIF of drinking water sample s recorded with laboratory setup (from Killinger and Sivaprakas am (2006), reprinted with permission from Optical Society of America). 31 Figure 3.7 LIF of bisphenol A at diff erent concentrations recorded with laboratory setup (from Killinger and Sivaprakasam (2006), reprinted with permission from Optical Society of America). 33 Figure 3.8 Schematic diagram of the porta ble LIF system developed at USF. 34

PAGE 11

viii Figure 3.9 Photograph of the portable LIF system developed at USF. 35 Figure 3.10 Linearity and se nsitivity of the portable LIF system developed at USF (from Sivaprakasam, Sha nnon, Luo, Coble, Boehme, and Killinger (2003), reprinted with permission from Optical Society of America). 36 Figure 3.11 LIF of water in the Gulf of Mexico recorded with the portable LIF system developed at USF (from Sivaprakasam, Shannon, Luo, Coble, Boehme, and Killinger ( 2003), reprinted with permission from Optical Society of America). 37 Figure 4.1 Schematic diagram of the modified portable LIF system. 39 Figure 4.2 Photograph of the modified portable LIF system: (a) optics box; (b) electronics box. 45 Figure 4.3 Diagram illustrating the princi ple of a reverse osmosis system. 48 Figure 4.4 Photographs of reverse osmo sis systems at USF: (a) small SEPA unit, (b) E-Series unit (Cour tesy of Dr. Carnahan, USF Engineering Department). 49 Figure 4.5 LIF of ground water before and after RO treatment; 266 nm excitation; distilled water shown for comparison (May 13, 2005). 51 Figure 4.6 LIF of ground water before and after RO treatment; 355 nm excitation; distilled water shown for comparison (May 13, 2005). 52 Figure 4.7 LIF of water with mineral content (to model me mbrane fouling) before and after RO treatment; ex citation 266 nm (deionized water shown for comparison). 54 Figure 4.8 LIF of water with mineral content (to model me mbrane fouling) before and after RO treatment; ex citation 355 nm (deionized water shown for comparison). 55 Figure 4.9 LIF of water with mineral content and clay particles (to model membrane fouling) before and after RO treatme nt; excitation 266 nm (deionized water shown for comparison). 56 Figure 4.10 LIF of water with mineral content and clay particles (to model membrane fouling) before and after RO treatm ent; excitation 355 nm (deionized water shown for comparison). 57 Figure 4.11 LIF of clay in distilled wa ter at different c oncentrations; 266 nm excitation. 59

PAGE 12

ix Figure 4.12 LIF of clay in distilled wa ter at different c oncentrations; 355 nm excitation. 60 Figure 4.13 LIF of various water sample s; 266 nm excitation (June 1, 2005). 61 Figure 4.14 LIF of various water sample s; 355 nm excitation (June 1, 2005). 62 Figure 4.15 LIF of international and do mestic drinking water samples; 266 nm excitation (June 11, 2007). 63 Figure 4.16 LIF of international and do mestic drinking water samples; 355 nm excitation (June 11, 2007). 64 Figure 4.17 LIF of Bacillus Subtilis spores in distilled water; 266 nm excitation. 66 Figure 4.18 LIF of Bacillus Subtilis spores in distilled water; 355 nm excitation. 67 Figure 4.19 LIF of malathion in dist illed water; 266 nm excitation. 69 Figure 4.20 LIF of malathion in dist illed water; 355 nm excitation. 70 Figure 5.1 Schematic diagram of wa ter flow for continuous tap water monitoring. 72 Figure 5.2 LIF of flowing tap water for 12 hours continuous monitoring; 266 nm excitation (Sept. 11, 2006). 74 Figure 5.3 LIF of flowing tap water for 12 hours continuous monitoring; 355 nm excitation (Sept. 11, 2006). 75 Figure 5.4 LIF of flowing tap water at selected wavelengths for 12 hours continuous monitoring; 266 nm excitation (Sept. 11, 2006). 76 Figure 5.5 LIF of flowing tap water at selected wavelengths for 12 hours continuous monitoring; 355 nm excitation (Sept. 11, 2006). 77 Figure 5.6 Ratio of LIF (451 nm) to Ra man (291 nm) signals of flowing tap water for 12 hours continuous monitoring; 266 nm excitation (Sept. 11, 2006). 79 Figure 5.7 Ratio of LIF (451 nm) to Ra man (400 nm) signals of flowing tap water for 12 hours continuous monitoring; 355 nm excitation (Sept. 11, 2006). 80 Figure 5.8 LIF of recirculated tap wate r for 4 hours continuous monitoring; 266 nm excitation (Oct.10, 2006). 82

PAGE 13

x Figure 5.9 LIF of recirculated tap wate r for 4 hours continuous monitoring; 355 nm excitation (Oct.10, 2006). 83 Figure 5.10 LIF of flowing tap water fo r 7 days continuous monitoring; 266 nm excitation (Nov. 2006). 85 Figure 5.11 LIF of flowing tap water fo r 7 days continuous monitoring; 355 nm excitation (Nov. 2006). 86 Figure 5.12 Rayleigh scattering (266 nm ) of flowing tap water for 7 days continuous monitoring; 266 nm excitation (Nov. 2006). 87 Figure 5.13 Rayleigh scattering (355 nm ) of flowing tap water for 7 days continuous monitoring; 355 nm excitation (Nov. 2006). 88 Figure 5.14 LIF of flowing tap water at selected wavelengths for 7 days continuous monitoring; 266 nm excitation (Nov. 2006). 89 Figure 5.15 LIF of flowing tap water at selected wavelengths for 7 days continuous monitoring; 355 nm excitation (Nov. 2006). 90 Figure 5.16 Ratio of LIF (451 nm) to Ra man (291 nm) signals of flowing tap water for 7 days continuous m onitoring; 266 nm excitation (Nov. 2006). 92 Figure 5.17 Ratio of LIF (451 nm) to Ra man (400 nm) signals of flowing tap water for 7 days continuous m onitoring; 355 nm excitation (Nov. 2006). 93 Figure 5.18 LIF of recirculated tap wate r for 6 days continuous monitoring; 266 nm excitation (Dec. 2006). 94 Figure 5.19 LIF of recirculated tap wate r for 4 days continuous monitoring; 355 nm excitation (Dec. 2006). 95 Figure 5.20 Rayleigh scattering (266 nm) of recirculated tap water for 6 days continuous monitoring; 266 nm excitation (Dec. 2006). 97 Figure 5.21 Rayleigh scattering (355 nm) of recirculated tap water for 4 days continuous monitoring; 355 nm excitation (Dec. 2006). 98 Figure 5.22 LIF of recirculated tap wate r at selected wavelengths for 6 days continuous monitoring; 266 nm excitation (Dec. 2006). 99 Figure 5.23 LIF of recirculated tap wate r at selected wavelengths for 4 days continuous monitoring; 355 nm excitation (Dec. 2006). 100

PAGE 14

xi Figure 5.24 Ratio of LIF (451 nm) to Raman (291 nm) signals of recirculated tap water for 6 days continuous monitoring; 266 nm excitation (Dec. 2006). 102 Figure 5.25 Ratio of LIF (451 nm) to Raman (400 nm) signals of recirculated tap water for 4 days continuous monitoring; 355 nm excitation (Dec. 2006). 103 Figure 5.26 LIF of distilled water fo r 7 hours continuous monitoring; 266 nm excitation. 105 Figure 5.27 LIF of distilled water fo r 7 hours continuous monitoring; 355 nm excitation. 106 Figure 5.28 Raman scattering (266 nm ) of distilled water for 7 hours continuous monitoring; 266 nm excitation. 107 Figure 5.29 Raman scattering (355 nm ) of distilled water for 7 hours continuous monitoring; 355 nm excitation. 108 Figure 5.30 LIF of distilled water at selected wavelengths for 7 hours continuous monitoring; 266 nm excitation. 109 Figure 5.31 LIF of distilled water at selected wavelengths for 7 hours continuous monitoring; 355 nm excitation. 110 Figure 5.32 LIF of tannic acid for 7 hours continuous monitoring; 266 nm excitation. 111 Figure 5.33 LIF of tannic acid for 7 hours continuous monitoring; 355 nm excitation. 112 Figure 5.34 Raman (291 nm) signal of tannic acid for 7 hours continuous monitoring; 266 nm excitation. 114 Figure 5.35 Raman (400 nm) signal of tannic acid for 7 hours continuous monitoring; 355 nm excitation. 115 Figure 5.36 LIF of tannic acid at select ed wavelengths for 7 hours continuous monitoring; 266 nm excitation. 116 Figure 5.37 LIF of tannic acid at select ed wavelengths for 7 hours continuous monitoring; 355 nm excitation. 117 Figure 5.38 LIF of tannic acid and chlori ne for 5 hours continuous monitoring; 266 nm excitation. 118

PAGE 15

xii Figure 5.39 LIF of tannic acid and chlori ne for 5 hours continuous monitoring; 355 nm excitation. 119 Figure 5.40 Raman (291 nm) signal of ta nnic acid and chlorine for 5 hours continuous monitoring; 266 nm excitation. 120 Figure 5.41 Raman (400 nm) signal of ta nnic acid and chlorine for 5 hours continuous monitoring; 355 nm excitation. 121 Figure 5.42 LIF of tannic acid and chlorine at selected wavelengths for 5 hours continuous monitoring; 266 nm excitation. 123 Figure 5.43 LIF of tannic acid and chlorine at selected wavelengths for 5 hours continuous monitoring; 355 nm excitation. 124 Figure 6.1 Schematic diagrams of the in itial LED drivers: (a) CW mode; (b) pulsed mode. 127 Figure 6.2 Schematic diagram of the optimized LED driver. 130 Figure 6.3 LED driver photo: (a) inside view; (b) enclosure and connectors. 132 Figure 6.4 Dummy load characteristics of the LED driver: collector current vs. power supply voltage in CW mode. 136 Figure 6.5 Dummy load characteristics of the LED driver: collector current vs. load resistance in CW mode. 137 Figure 6.6 Dummy load characteristics of the LED driver: collector current vs. adjustable emitter resistance in CW mode. 138 Figure 6.7 Characteristics of the LED dr iver in CW mode: LED V-I curves. 140 Figure 6.8 Characteristics of the LED driver in CW mode: LED current vs. adjustable emitter resistance. 141 Figure 6.9 Characteristics of the LED dr iver in pulsed mode: LED V-I curves. 142 Figure 6.10 Characteristics of the LED dr iver in pulsed mode: LED current vs. adjustable emitter resistance. 143 Figure 7.1 Schematic diagram of the a pparatus used to study the spectral properties of the LED emission. 147 Figure 7.2 Schematic diagram of the a pparatus used to study the temporal characteristics of the LED emission. 149 Figure 7.3 Photograph of a UV LED in operation. 150

PAGE 16

xiii Figure 7.4 Normalized emission spectra of all LEDs; CW mode, 10 mA. 151 Figure 7.5 Emission spectra of LED265 for different CW currents. 153 Figure 7.6 Emission spectra of LED320 for different CW currents. 154 Figure 7.7 Transmission of UV-transmitting visible-blocking CG-UG-11 filter (from CVI Laser catalog). 155 Figure 7.8 Emission spectra of LED265 fo r different CW currents with CGUG-11 filter. 156 Figure 7.9 Emission spectra of LED320 fo r different CW currents with CGUG-11 filter. 157 Figure 7.10 Emission spectra of electr ically modulated LED265 for different pulse repetition rates; 50% duty cycle. 159 Figure 7.11 Emission spectra of CW LED265 with an external chopper for different chopper frequenc ies; 50% duty cycle. 160 Figure 7.12 Waveforms of LED265 emi ssion in electrically pulsed and externally chopped regimes, 100 Hz rate. 161 Figure 7.13 Emission spectra of LED265 in CW, 5 ms pulsed and 10 s pulsed modes. 163 Figure 7.14 Emission spectra of LED265 w ith CG-UG-11 filter (losses 52%) in CW, 5 ms pulsed and 10 s pulsed modes. 164 Figure 7.15 Emission spectra of LED300 in CW, 5 ms pulsed and 10 s pulsed modes. 165 Figure 7.16 Emission spectra of LED300 w ith CG-UG-11 filter (losses 20%) in CW, 5 ms pulsed and 10 s pulsed modes. 166 Figure 7.17 Emission spectra of LED335 in CW, 5 ms pulsed and 10 s pulsed modes. 167 Figure 7.18 Emission spectra of LED335 w ith CG-UG-11 filter (losses 17%) in CW, 5 ms pulsed and 10 s pulsed modes. 168 Figure 7.19 Emission spectra of LED355 in CW, 5 ms pulsed and 10 s pulsed modes. 169 Figure 7.20 Emission spectra of LED355 w ith CG-UG-11 filter (losses 25%) in CW, 5 ms pulsed and 10 s pulsed modes. 170

PAGE 17

xiv Figure 7.21 LED355 emission in 5 ms, 100 Hz pulsed mode as a function of LED current. 172 Figure 7.22 Waveforms of modulating voltage and LED355 emission in 5 ms, 100 Hz pulsed mode. 173 Figure 7.23 LED355 emission in 10 s, 500 Hz pulsed mode as a function of LED current. 174 Figure 7.24 Waveforms of modulati ng voltage and LED355 emission in 10 s, 500 Hz pulsed mode. 175 Figure 7.25 LED spectral peak power de nsity in 10 mA, 5 ms, 100 Hz pulse mode (lower curves are for the LED with CG-UG-11 filter). 177 Figure 7.26 LED spectral peak power density in 50 mA, 10 s, 500 Hz pulse mode (lower curves are for the LED with CG-UG-11 filter). 178 Figure 8.1 Schematic diagram of the CW laboratory bench-top LED-IF / LIF system. 180 Figure 8.2 Photograph of the CW labor atory bench-top LED-IF system. 182 Figure 8.3 Transmission spectrum of the BK-7 Schott glass. 183 Figure 8.4 Lake water fluorescence; 266 nm laser excitation (Sept. 14, 2006). 186 Figure 8.5 Lake water fluorescence; 265 nm and 320 nm LED excitation (Sept. 14, 2006). 187 Figure 8.6 Lake water fluorescence; 266 nm laser excitation; UV-blocking filter (Sept. 23, 2006). 188 Figure 8.7 Lake water fluorescence; 265 nm and 320 nm LED excitation; UVblocking filter (Sept. 23, 2006). 189 Figure 8.8 Lake water fluorescence; 266 nm laser excitation; UV-blocking filter and correction / sm oothing (Sept. 23, 2006). 190 Figure 8.9 Lake water fluorescence; 265 nm and 320 nm LED excitation; UVblocking filter and correction / smoothing (Sept. 23, 2006). 191 Figure 8.10 Tap water fluorescence; 266 nm laser excitation; UV-blocking filter and correction / sm oothing (Sept. 23, 2006). 193 Figure 8.11 Tap water fluorescence; 265 nm and 320 nm LED excitation; UVblocking filter and correction / smoothing (Sept. 23, 2006). 194

PAGE 18

xv Figure 8.12 Distilled water fluorescence; 266 nm laser excitation; UV-blocking filter and correction / smoothing. 195 Figure 8.13 Distilled water fluorescen ce; 265 nm and 320 nm LED excitation; UV-blocking filter and correction / smoothing. 196 Figure 8.14 Fluorescence of tonic with qui nine in distilled water; 265 nm and 320 nm excitation; UV-blocking filter and correction / smoothing. 198 Figure 8.15 Fluorescence of Coumarin 540A dye in distilled water; 265 nm and 320 nm excitation. 200 Figure 8.16 Fluorescence of B. Subtilis bacterial endospores in distilled water; 265 nm and 320 nm excitation. 201 Figure 8.17 Fluorescence of Tb-DPA in distilled water; 265 nm excitation. 202 Figure 9.1 Schematic diagram of the pulsed LIF/LED-IF system. 204 Figure 9.2 Photograph of the pulsed LIF/LED-IF system. 205 Figure 9.3 Fluorescence of lake water; pulsed 265 nm excitation (March 12, 2008). 207 Figure 9.4 Fluorescence of lake water; pulsed 355 nm excitation (March 12, 2008). 208 Figure 9.5 LED-IF of lake water; pu lsed excitation at 265, 300, 335 and 355 nm (March 12, 2008). 210 Figure 9.6 Transmission curves of cutoff filters in LIF / LED-IF system. 211 Figure A.1 Schematic diagram of the multifunctional LED. 221 Figure B.1 Front panel of the LED-IF data collection LabVIEW program. 223 Figure B.2 Diagram of the LED-IF da ta collection LabVIEW program. 223 Figure B.3 Front panel of the Save_Data.vi. 224 Figure B.4 Diagram of the Save_Data.vi. 224 Figure C.1 ST2000 spectrometer detector sensitivity. 229 Figure C.2 ST2000 spectrometer grating efficiency. 229 Figure C.3 ST2000 spectrometer fiber transmission. 230

PAGE 19

xvi Figure C.4 ST2000 spectrometer overall correction. 230 Figure C.5 USB2000 spectrometer detector sensitivity. 231 Figure C.6 ST2000 spectrometer grating efficiency. 231 Figure C.7 ST2000 spectrometer fiber transmission. 232 Figure C.8 ST2000 spectrometer overall correction. 232

PAGE 20

xvii UV LASER AND LED INDUCED FLUORE SCENCE SPECTROSCOPY FOR DETECTION OF TRACE AMOUNTS OF ORGANICS IN DRINKING WATER AND WATER SOURCES Anna V. Sharikova ABSTRACT A UV Laser Induced Fluorescence (LIF) sy stem, previously developed in our laboratory, was modified and used for a series of applications related to the development and optimization of UV LIF spectroscopic measurements of trace contaminants in drinking water and other water sources. Fluorescence spectra of a number of water samples were studied, including those related to the reverse osmosis water treatment and membrane fouling, domestic and international drinking water, industrial toxins, bacterial spores, as well as several fluorescence standa rds. Of importance was that the long term detection of the trace level of Dissolved Organic Compounds (DOC) was measured, for the first time to our knowledge, over a one week period and w ith a time resolution of 2.5 minutes. A comparison of LIF emission us ing both 266 nm and 355 nm excitation was also made for the first time. Such realtime and continuous measurements are important for future water treatment control. The LIF system was modified to acco mmodate UV Light Emitting Diodes (LED) as alternative excitation sources and tested for the detecti on of trace organic species in water. In addition, a compact system us ing LED excitation and a spectrometer was

PAGE 21

xviii developed and underwent initial testing. The original LIF system had two lase r sources, 266 nm and 355 nm. The additional sources incorporated in the system were UV LEDs emitting at 265 nm, 300 nm, 335 nm and 355 nm. The LED spectral emission was studied in de tail, in terms of spectral variability and power output. It was found that a ll LEDs had some emission in the visible spectrum, and an optical filter was used to remove it. The signal-to-noise ratio for the LED -based systems was determined and compared with that of the LIF system. Th e fluorescent signal of the LED-based system was smaller by 1 to 2 orders of magnitude, despite the fact that the LED pulse energy was 2 to 3 orders of magnitude less than the laser' s. As such, the fluorescent signal from the LED was greater than expected. Therefore, a UV LED may be a compact and much cheaper optical source for future water measurement instruments.

PAGE 22

1 CHAPTER ONE INTRODUCTION Fluorescence is widely used for the detection of trace levels of various chemicals (Christian & Callis, 1986; Lakowicz, 2006; Nollet, 2000). It offers several advantages over conventional methods of water analysis, such as continuous online monitoring and no need for reagents or sample preparation. Th ese features might play a critical role in a situation where quick recognition of a problem might be necessary, for example, at the facilities providing drinki ng water to populated areas. Using lasers as excitation sources improve s the sensitivity of detection by several orders of magnitude compared to broadband sources, e.g. UV lamps. In this case, the detection limit for Dissolved Organic Compoun ds (DOCs) in water below the parts-perbillion level have been reported (Sivapraka sam, 2002). In particular, differences between the spectra of tap, distilled and bottled brands of water were observed (Sivaprakasam & Killinger, 2003). These st udies showed the importance of using a deep-UV laser or optical exci tation source with wavelength s near 250 to 350 nm, since most previous LIF studies of water have used lasers with wavelengths in the 400 to 500 nm range. Longer wavelengths were not as e fficient in exciting th e DOC fluorescence or in the selectivity of the emission spectrum compared to other background interfering species.

PAGE 23

2 Recently there has been considerable pr ogress in extending the wavelength range of Light-Emitting-Diodes (LEDs) into the deep -UV. These recent advances in UV light emitting diode manufacturing have made it possible to perform sensitive fluorescence measurements with these compact, inexpensive devices. Several vi sible or moderate UV wavelength LEDs have been studied, including the use of a LED whose output is directly controlled by the current passing through the device and used in the design of a phasemodulation spectrofluorometer (I wata, Kamada, & Araki, 2000). However, very little spectrofluorometric studies have been conduc ted in the deep-UV be cause the appropriate LEDs were not available. Our research was designed to pursue such a deep-UV system expecting that a UV LED based system fo r fluorescence detection might be more attractive due to its small size, afford ability and reduced energy consumption. This dissertation builds upon earlier research done in our laboratory. A custommade UV laser induced fluorescence (LIF) syst em was modified and optimized for realtime and continuous measurements of water spectr a. First, the fluorescence spectra of a number of water samples were recorded a nd analyzed for the first time using both 266 nm and 355 nm laser excitation, including t hose related to the reverse osmosis water treatment and membrane fouling, domestic and international drinking water, as well as several fluorescence standards. It should be noted that water samples for the reverse osmosis water treatment studies were provided by Prof. Carnahan's laboratory (USF College of Engineering), while Total Orga nic Carbon (TOC) samples were provided by Prof. Levine's laboratory (USF College of E ngineering). Second, the LIF system was modified to accommodate UV light emitting di odes and extensively tested for the detection of trace organic speci es in water. The changes included hardware adjustments,

PAGE 24

3 completely redesigned software for data acq uisition, development and optimization of a current driver for the UV LEDs, and design of the optical system to optimize the LIF signal-to-noise ratio. A comparison between the laser and LED excitation regarding the measured signal-to-noise of the fluorescence sign al was also quantified for the first time. Thirdly, a compact spectrofluorometer wate r measuring system using LED excitation and a compact spectrometer was developed a nd underwent initial te sting. These last results were very encouraging and demons trated the high potential for the future development and potential on-line use of such a device for water monitoring. Some of the more unique results of this research involve the use of the deep-UV LED. The driver for the UV LEDs was ma de to accommodate both continuous wave (CW) and pulsed regimes, and a slow turn-on was used to avoid transient currents when the unit was powered up. A time cons tant reduction option for short (a few microseconds) pulses was also implemented. Th e driver provided a stable and adjustable CW current for the LED. In the pulsed mode, the LED current was modulated according to an external input up to the maximum current set by the driver controls. The LED emission was studied in detail, in terms of spectral variabilit y as a function of the driving current, and in terms of power output. It was found that all LEDs had some emission in the visible spectrum, and a special optical filt er was used to correct that problem. The signal-to-noise ratio for the LED-based system s was determined and compared with that of the LIF system. Depending on the LED operating regime and the output wavelength, the fluorescent signal of the LED-based system was down only by a factor of 7 to 70, despite the fact that the LED average power and pulse energy were 2 to 3 orders of magnitude less than that of the laser, and the LED peak power was 6-7 orders of

PAGE 25

4 magnitude lower. This makes UV LED a promising optical exc itation source in water fluorescence measurements, especially for inexpensive, compact devices. The presentation of this research is orga nized as following. Chapter Two reviews the issues of water quality and the measurement techniques presently approved for water testing. Chapter Three provides the b ackground information regarding absorption, fluorescence and scattering of light occurring in complex molecules, as it relates to water fluorescence spectroscopy. It also discusses previous LIF water studies in our group at USF. Chapter Four starts with a description of the LIF a pparatus and initial modifications introduced in it by the author. Then it details the measurements of LIF for reverse osmosis (RO) studies and other water sa mples. In Chapter Five, the results of long-term monitoring of tap water are presented and discussed. Chapter Six describes the development of an LED driver to power and modulate the output of UV LEDs used in later expe riments. Chapter Seven presents the investigation of UV LED output in different operating re gimes, both in spectral and temporal domains. In Chapter Eight, a comparison between laser and CW LED performance as excitation sources for fl uorescence measurements is made based on diverse water samples, and a compact LED spectrometer apparatus is developed and experimental measurements and results are presented. Chapter Ni ne describes the LIF system setup incorporating UV LEDs, and measurements done to compare the performance of pulsed LEDs with that of lasers in this system. Conclusions and future work are discussed in Chapter Ten. Th e Appendices contain additional information concerning the LED driver calculations, La bVIEW data acquisition and Matlab data

PAGE 26

5 processing programs written by the author, and interpolated correction curves for fiber/spectrometer combinations used in the experiments.

PAGE 27

6 CHAPTER TWO BACKGROUND INFORMATION ON WATER QUALITY AND MEASUREMENT TECHNIQUES This chapter provides the background information regarding the types of contaminants typically encountered in dri nking water, and the standard laboratory methods used to monitor water contamination. 2.1 Classes of Contaminants in Drinking Water Water quality is a rather complex issue. Drinking water has to satisfy a variety of standards, both in terms of its physical a nd chemical aspects (Ame rican Public Health Association [APHA], Ameri can Water Works Association [AWWA], & Water Pollution Control Federation [WPCF], 1989; Hunt & W ilson, 1986; Crompton, 2000). Therefore, there is no single method to determine the quality of a water sample. Rather, based on a property or contaminant being tested, a partic ular technique is selected to analyze the sample. Table 2.1 lists various water properties and methods for their examination, while Table 2.2 shows the typical classes of c ontaminants that can be present in drinking water and are a health hazard, together w ith most common detection methods (APHA, AWWA, & WPCF, 1989). These tables are not meant to be exhaustive, but are

PAGE 28

7 Table 2.1 Water properties and appropriate laboratory measurement techniques. Property Examination methods Color Visual comparison method, spectrophotometric method, tristimulus filter method Turbidity Nephelometric method Odor Threshold odor test Taste Flavor threshold test, flavor rating assessment, flavor profile analysis Acidity, alkalinity Titration method Hardness Hardness by calculation, EDTA titrimetric method Conductivity Laboratory method Salinity Electrical conduc tivity method, density method Temperature Thermometer, thermistor (for depth temperature measurements)

PAGE 29

8 Table 2.2 Water contaminants and detection methods. Class of contaminant Examples Detection methods Metals Aluminum through zi nc Absorption spectrometry, inductively coupled plasma Inorganic nonmetallic constituents B, Br-, CO2, CN, Cl, Cl-, ClO2, F-, H+, I, I-, N, NH3, NO2, NO3, O, O3, P, Si, S2-, SO3 2-, SO4 2Ion chromatography, varied Organic constituents Volatile organics, aromatic organics, halocarbons, phenols etc. Total organic carbon (TOC) Gas chromatography/mass spectrometry (GCMS), high-performance liquid chromatography (HPLC), varied Combustion-infrared method, persulfateultraviolet oxidation method, wet-oxidation method Radioactivity Cesium, iodine, radi um etc. Precipitation method, varied Microbiological Bacteria, viruses, fungi Varied

PAGE 30

9 presented to illustrate the great variety of te st parameters and methods. The issue is further complicated by the fact that the same element or substance might exhibit different properties depending on its physico-chemical fo rm (known as speciation), thus the total concentration of a particular element does not ne cessarily predict what effect it will have on the environment. Micronutrient elemen ts, such as nitrogen, phosphorus and silicon, are the classical examples of this phenomenon (Hunt & Wilson, 1986). The type of water also influences what method is used. Drinking and ground waters, being among the cleanest, require th e most sensitive techniques. They are followed by sea waters and other surface waters in terms of purity. Waste waters, trade effluents and sludge tend to have higher con centration of contamin ants, and, therefore, proper handling of these subs tances becomes important. Organic substances in water, particularly in potable water, ha ve attracted serious attention in recent years. Although many organic substances occur naturally in surface waters, an increasing number of these compounds are human made, and their effects on the environment and health are poorly understo od. Apart from direct contamination, organic compounds are formed as a result of chemical and biological processes from inorganic substances, e.g., halo forms are produced as a side effect of water chlorination. According to Crompton (2000), up to 2000 organi c substances can be present in river waters, and many might remain even af ter processing in the waterworks. One example is the plastic monomer and pl asticizer bisphenol A (BPA). This chemical has one of the highest levels of worl dwide production. As a result, it has been found in varying amounts in the environment, food, water and biological tissues. BPA is a known endocrine disruptor, and although specific mechanisms and results of continuous

PAGE 31

10 low level exposure are still being studied, the fa ct that it was found in over 90% of tested individuals is sobering (Va ndenberg, Hauser, Marcus, Olea, & Welshons, 2007). Our group has recently detected BPA in water at the sub-ppb levels (Sivaprakasam & Killinger, 2003). Since there are many substances that can be harmful even in small concentrations, it is necessary to have methods capable of detecting trace amounts of such compounds, both to be able to study their effects and to take preventative measur es where necessary. Another recent example is the Associated Press report about medi cation, particularly antibiotics, found in municipa l drinking water of several cities (Mendoza, March 2009; Donn, Mendoza, Pritchard, April 2009). 2.2 Approved Water Testing Te chniques for Organic Compounds The majority of currently approved water testing techniques require substantial sample preparation, often involving reagen ts (APHA, AWWA, & WPCF, 1989). As a result, the testing is rarely performe d more than once every few days. 2.2.1 Chromatographic Methods Chromatographic methods allow determin ation of individual organic compounds at low concentrations. A large number of variations of the technique exist, but two methods Gas Chromatography/ Mass Sp ectrometry (GCMS) and High Performance Liquid Chromatography (HPLC) are the most useful. Due to the complexity of the technique, only qualified analysts can prope rly perform the experime nts and interpret the data.

PAGE 32

11 Chromatography is a method of physical separation of components between two phases: stationary (a column packing or a capillary column coating) and mobile (usually gas or liquid) (Nollet, 2000; APHA, AWWA, & WPCF, 1989). Depending on the mobile phase, the technique is divide d into gas chromatography or liquid chromatography. In gas chromatography, the sample is injected into the column, compounds are vaporized, and the carrier ga s (nitrogen, argon-methane, he lium or hydrogen) transports them through the column. The rate at which they move depends on the differences in partition coefficients between the phases. Therefore, compounds can be distinguished based on the retention time in the column. Several types of detectors can be used with the system, with the choice de pending primarily on the kind of analyte. The most common detectors are electro lytic conductivity detector (for compounds containing halogen, nitrogen, sulfur and nitrosamine), electron capture detector (for molecules containing electronegative groups), flame i onization detector (for all organic carboncontaining compounds), photoionization dete ctor (organic and some inorganic compounds), and mass spectrometer (classifies fragments based on their charge-to-mass ratio) (APHA, AWWA, & WPCF, 1989). This technique is particularly suited for the detection of volatile compounds, with a boiling point up to 250 C (Crompton, 2000). In High Performance Liquid Chromatogra phy (HPLC), the sample is transported through the column by a liquid mobile phase. As the sample reacts with the liquid stationary phase, individual compounds ar e retained and separated. A post column reactor can be used to enhance previ ously undetectable co mpounds by attaching a chromophore by means of a chemical reaction. A combination of a grating with a

PAGE 33

12 photodiode array, recording the absorbance sp ectrum of the sample illuminated with a UV-VIS source, is typically used to detect th e signal. Alternatively, a detector can be used to record fluorescence of the compounds excited by a monochromatic source. The latter method is highly sele ctive and very sensitive. Fig. 2.1 shows a schematic diagram of a liquid chromatography instrument (Nollet, 2000, p. 850). A chromatogram of tap water containing polycyclic aromatic hydrocarbons is presented in Fig. 2.2 (B eltran, Guiteras, and Ferrer, 1998). 2.2.2 Other Methods Titration procedures are us ed to determine anionic and cationic surfactants, various organic acids and othe r ionized organics (Crompton, 2000). The ions present in the tested substance react with a standard alkali or a standard acid. A titration curve is constructed based on the sample pH after each increment in the amount of titrant (APHA, AWWA, & WPCF, 1989). Titration can be performed using fully automated instruments; however, it is not very sensitiv e and tends to be used for heavily polluted waters (Crompton, 2000). Polarography is suitable for compounds that can undergo an oxidation reduction, such as polyaromatic hydrocarbons, aldehydes, am ides, etc. In this method, a voltage ramp is applied during a controlled drop gr owth with a standard dropping mercury electrode. The current is integrated over a fe w tens of milliseconds at the end of the drop, or at several points of a volta ge signal superimposed on th e voltage ramp. Equipment for polarographic measurements is also automated and easy to use (Crompton, 2000).

PAGE 34

Figure 2.1 Schematic diagram of a liquid chromatography apparatus (adapted from Nollet (2000)). 13

PAGE 35

Figure 2.2 Chromatogram of tap water m easured using high performance liquid chromatography with fast-scanning fluor escence spectrometer (reprinted with permission from Beltran, Guiteras, and Ferrer (19 98), Copyright 1998 American Chemical Society). 14

PAGE 36

15 Atomic absorption is applied to determ ine metals in water (APHA, AWWA, & WPCF, 1989). In this techni que, the sample is sprayed into a flame, and the resulting vapor is illuminated by light of an appropriate wavele ngth. Absorption spectrum characteristic of the atoms present is then us ed to determine metal concentrations. The sensitivity of this method is low due to large background interfer ences (Crompton, 2000). Inductively coupled plasma is employed for determination of mercury, silicon, phosphorus and boron. The sample is introduced into the plas ma made of a suitable gas by application of an RF magnetic field. Th e emission spectrum of the excited sample vapor is then recorded (Crompton, 2000). It should be noted that there are seve ral commonly used terms in different disciplines that are related to the carbon content of a sample For example, Total Organic Carbon (TOC) is a measure of all carbon atoms covalently bonded in organic molecules. Dissolved Organic Carbon (DOC) is a fraction of TOC remaining after filtering out all particulates of greater size than 0.45 m (APHA, AWWA, & WPCF 1989). Dissolved Organic Matter (DOM) is a term used in marine science to denote the material of biological origin, containing both carbon a nd hydrogen, from which particulates have been removed. Colored Dissolved Organi c Matter (CDOM) is a component of DOM that absorbs light in the UV and visibl e parts of the spectrum (Coble, 2007). As can be seen from these past examples there is a wide variety of different instrumentation techniques that are used to measure selected species in water.

PAGE 37

16 CHAPTER THREE BACKGROUND INFORMATION ON LASER-INDUCED FLUORESCENCE, ABSORPTION AND OPTICAL SPECTROSCOPY OF WATER AND TRACE SPECIES, INCLUDING PREVIOUS LIF WATER STUDIES AT THE UNIVERSITY OF SOUTH FLORIDA This chapter discusses the process of li ght absorption, fluorescence and scattering occurring in complex molecules, and th eir application to water fluorescence spectroscopy. It also discusses pr evious LIF water studies at USF. 3.1 Scattering, Absorption and Fluorescence of Light Light impinging on a particle or molecule can be either scattered, which results primarily in the change of the photon's di rection, or it can be absorbed (Sharma & Shulman, 1999). Absorption of light by a molecule brings it into an excited state due to the addition of energy to the system. Return to the ground (equilibrium) state is achieved by one or more of the following processes: emi ssion (fluorescence), ra diationless internal conversion, excitation transfer to another molecule, conversi on to the triplet state, and photochemical reactions. The path taken by the system to reestablish its equilibrium is determined by a combination of quantum mechan ical rules, such as symmetry properties

PAGE 38

17 of the upper and lower states, and therm odynamics, i.e. relaxation to the lowest rotational/vibrational level because th e levels' separation is less than kT (Brand & Johnson, 1997), the later being affected by envi ronmental factors, su ch as interaction with a solvent. 3.1.1 Rayleigh, Mie and Raman Scattering Elastic scattering is called Rayleigh if the size of the scattering particles is much less than the wavelength of light. The part icle is considered to be in a homogenous electric field, and behaves as an electri c dipole (Born & Wolf, 1999). For example, experiments involving liquids (e.g. water) have a large Rayl eigh scattering peak at the excitation wavelength due to the H2O molecules even if the sample contains no other substances. When the size of the particle is comparable to the wave length of light, Mie elastic scattering occurs. The angular distributi on of the Mie scattered light is complex and strongly depends on the ratio of the particle size and the incident wavelength (Born & Wolf, 1999). Therefore, the presence of contaminants will generally increase the scattering signal at the excitation wavelength. Inelastic scattering of the incident light is known as Raman scattering. Although the frequency of Raman scattered light is diffe rent from the frequency of the excitation light within the energy gap of the rotational and/ or vibrational levels of the molecule, this phenomenon does not have the resonant dependence characteristic of the absorption/fluorescence processes. Raman s cattering at shorter wavelengths than the incident light (called Anti-St okes scattering) is rarely obs erved at room temperature,

PAGE 39

18 since it corresponds to transitions from larg ely unpopulated upper rotational/vibrational levels. The Stokes component, occurring at wavelengths longer than the excitation wavelength, is usually the only one present (B orn & Wolf, 1999). In aqueous media, Raman scattering is caused by OH bonds and its peak wavelength, Ram, in nm, depends on the excitation wavelength, Ex, in nm, as (Albani, 2007) 1 /Ram = 1/Ex 0.00034 (3.1) In practical terms, it is important to be able to separate or dis tinguish Raman scattering from fluorescence. The choice of excitati on wavelength should take possible overlap between these phenomena into account. Usua lly, the shorter excita tion wavelengths (in the UV) produce a greater separation betw een the Raman emission and the fluorescence emission. 3.1.2 Absorption of Optical Radiation Optical radiation is absorbed by a molecu le if the photon energy is equal to the energy difference between the upper (excited ) and lower (usuall y, ground) states. Transitions associated with electrons from p and d orbitals (creating bonds in molecules), as well as those re lated to nonbonding (already paired) n electrons in the valence shell, correspond to the region of UV to near IR, 200 1500 nm (Sharma & Schulman, 1999). Absorption within a transition band obeys the Lambert-Beer law, and, therefore, is representative of the concentration of the absorbing molecules in the sample. The time scale of the electronic transi tion for the majority of molecules is on the order of 10-15 s (Sharma & Schulman, 1999).

PAGE 40

19 The absorption spectrum of pure water is shown in Fig. 3.1 (Pope & Fry, 1997). The absorption coefficient is less than 0.1 m-1 for the near UV and visible region of the spectrum up to 580 nm, and below 1.0 m-1 between 580 and 700 nm. 3.1.3 Fluorescence versus Non-Radiative Energy Transfer The Frank-Condon principle states that atomic nuclei do not significantly alter their positions during th e electronic transition. Therefor e, right after the excitation the molecule retains the same geometry and envi ronment as in the ground state; that is, the structure of the vibrational and rotational leve ls is generally similar for the excited and ground electronic states (Lakowicz, 2006; Ch ristian & Callis, 1986; Sharma & Schulman, 1999). Since other processes are longer than 10-15 s, absorption is treated as instantaneous, and relaxation of the molecule by various m eans is subsequent to that. The electronic transitions diagram, al so called a Jablonski diagram (Fig. 3.2), illustrates the processes occurring in a molecule after absorption. 3.1.3.1 Relaxation Processes in Complex Molecules Vibrational relaxation, i.e., the loss of vibrational energy through collisions or heat radiation, occurs within 10-14 to 10-12 s. This process brings the excited molecule to the lowest vibrational level of the excite d electronic state (S harma & Schulman, 1999). Internal conversion is nonr adiative relaxation to the lo wer electronic level through thermal equilibration or tunneling. Its like lihood increases with gr eater overlap between vibrational levels of the higher and lower electronic stat es, and typically dominates

PAGE 41

0.001 0.01 0.1 1 10 350400450500550600650700750 Wavelength, nmAbsorption coefficient a, 1/m Figure 3.1 Optical absorption spectrum of pure water (ada pted from Pope and Fry (1997)). 20

PAGE 42

Figure 3.2 Jablonski diagram of energy leve ls and competing processes in a complex molecule (adapted from Albani (2007)). 21

PAGE 43

22 transitions between higher excited electronic states. Internal conversion occurs on the timescale of 10-12 s (Sharma & Schulman, 1999). Intersystem crossing from a singlet to triple t state is a radiati onless transition that can occur between the first excited singlet stat e and the triplet state. This transition requires the change in electron spin and is electric dipole-forb idden; therefore the lifetime of such a process is of the order of 10-8 s. In molecules containing heavy atoms or transition-metal ions the spin-orbit coupling can increase the efficiency of intersystem crossing. Fluorescence is a radiative transition, typi cally occurring from the lowest level of the first excited singlet state to one of the vibrational sublevels of the ground state. The emitted photon usually belongs in the visible or UV range. The lifetime of fluorescence is 10-10 to 10-7 s. Fluorescence competes with inte rnal conversion and intersystem crossing as a deactivation process for the first excited state (Lakowicz, 2006; Sharma & Schulman, 1999). Phosphorescence is a radiative transition from the lowest triplet to the ground state. As a forbidden process, it has lifetimes from 10-5 s to a few seconds. It is unlikely to occur in liquid media at room temperatur e because other deexci tation processes would take place first (Sharma & Schulman, 1999). Delayed fluorescence is a result of repopul ation of the first ex cited singlet state through a reverse intersystem crossing that can occur at a very low temperature in solid or viscous media. It has the lifetim es typical of phosphorescence and emission frequencies of a fluorescent transition (S harma & Schulman, 1999; Christian & Callis, 1986).

PAGE 44

23 Quenching of fluorescence describes a number of mechanisms resulting in a decreased fluorescence. Collisional quenching is a deactivation of the fluorescing molecule where its energy is transferred to a nother molecule in a solution. Reabsorption of fluorescence by other particles is another form of quenching. In case of static quenching, fluorophore and quencher form a nonf luorescent complex (Lakowicz, 2006; Dewey, 1991). Generally, these processe s only become significant for large concentrations of the quenching molecule. 3.1.3.2 Characteristics of Fluorescence Fluorescence is subject to the Stokes shift, i.e. emitted light is of longer wavelength than the absorbed light. This is primarily due to vibrational relaxation. Emission spectra do not depend on the ex citation wavelength, as long as it belongs in the absorption band of the fluorophor e (Kasha's rule). Fast relaxation of vibrational levels is th e reason (Lakowicz, 2006). Emission spectra often look like mirror images of absorption spectra. An absorption spectrum reflects the structure of the excited state, while the emission spectrum is representative of the ground state configuration. Similarity between these spectra is a consequence of the Frank-Condon rule (Lakowicz, 2006; Christian & Callis, 1986). The fluorescence quantum yield, f is the ratio of fluor escence rate constant of the first excited singlet state, kf to the sum of all rate constants of the processes participating in the deactivation of that state, where kd represents nonradiative deexcitation processes (Sharma & Schulman, 1999):

PAGE 45

24f = kf / (kf + kd) (3.2) Quantum yield can also be e xpressed through the radiative (f 0) and mean (f) lifetimes of the same level (Sharma & Schulman, 1999): f = f / f 0. (3.3) On the other hand, fluorescence yield is a fr action of absorbed light that resulted in fluorescence (f = If / Ia). Using the fact that the in cident light inte nsity should be equal to the sum of absorbed and transmitted intensities (I0 = Ia +It), and taking into account the Lambert-Beer law ( It / I0 = 10-cl), the fluorescence intensity can be expressed as (Sharma & Schulman, 1999; Albani, 2007): If = f I0 (1 10-cl) 2.3 f I0 c l (3.4) where is the molar absorptivity, also called molar extinction coefficient; c is the concentration of abso rbing molecules, and l is the thickness of the sample. The approximation is valid for low optical densities, so that Ia << I0. In actual measurements, the concentration of the unknow n sample is deduced from the standard sample concentration and the ratio of thei r intensities, under the assumption that other factors remain unchanged (Sharma & Schulman, 1999). An examination of the fluorescence lifetim e and anisotropy can provide additional information about the fluorescing substance. A variety of techniques exists that employ these properties of fluorescence. 3.2 Use of Fluorescence Measu rement in Water Studies Fluorescence is often used in marine science for the detection of Colored Dissolved Organic Matter (CDOM) (Coble, 2007). It is also empl oyed for analysis of

PAGE 46

25 polycyclic carbons (petroleum) (Fetzer, 2000; Vo-Dinh, 1989) and in terrestrial and ocean remote sensing (Muttiah, 2002; Bunkin & Voliak, 2001). 3.2.1 Methods and Examples of Fluorescence Measurements CDOM measurements are usually inves tigated by excitation-emission matrix (EEM) spectroscopy, where the fluorescence is recorded for a range of excitation wavelengths. The resulting three-dimensiona l plots (Fig. 3.3) have maxima at specific excitation/emission wavelengths characteristic of a particular substance, such as humic or fulvic compounds (Coble, 1996; Yoshioka et al, 2007; Hudson et al, 2007). LIF-EEM has also been employed for hydrocarbon cont amination detection in subsurface soils (Balshaw-Biddle, Oubre, & Ward, 2000). The composition of diverse crude oil samples has been differentiated by fluorescence lifetime analysis (Ryder, 2004). Th is approach was also useful in remote sensing of the ocean to distinguish oil spills from the chlorophyll fluorescent signal (Bunkin & Voliak, 2001). For many applications in water analysis, HPLC with fluorescence detection is currently the method of choice (Crompton, 2000). 3.2.2 Fluorescence of Organic Compounds Fluorescing organic substances have a wide range of ch emical configurations. However, they typically include aromatic or heteroaromatic groups. They can be generally separated into the following types: aromatic hydrocarbons and derivatives (such

PAGE 47

(a) (b) (c) Figure 3.3 Fluorescence of organic compounds in water: (a) river water; (b) sea water; (c) photobleached sea water ((a) and (c) are reprinted with permission from Coble (2007), Copyright 2007 American Ch emical Society; (b) is reprinted from Coble (1996), Copyright (1996), with permission from Elsevier). 26

PAGE 48

27 as polyphenyl hydrocarbons); compounds with fiveor six-membered heterocycles; compounds with a carbonyl group; compounds with several fluorophors; metallic complexes with organic ligands (Krasovitskii & Bolotin, 1988). Table 3.1 lists peak absorption and emissi on wavelengths for several natural and man-made organic compounds (Coble, 2007; Hudson et al, 2007; Fetzer, 2000; Sivaprakasam, Shannon, Luo, Coble, Boehme, & Killinger, 2003). 3.3 Previous LIF Water Studies at University of South Florida A series of measurements primarily fo cused on natural and ocean water studies have been performed at our Physics Laser Re mote Sensing lab at USF, in collaboration with the College of Marine Science. Two LIF systems were developed for that purpose: one, a laboratory system with a tunable dye laser and a Nd:YAG mi crochip laser (266 nm) excitation, and a spectrometer (Acton Re search Corporation, Model SpectraPro-150) with a cooled CCD detector (Princeton In struments, Model RTE-CCD-128) (Fig. 3.4); the other, a portable LIF system with two Nd:YAG microchip lase rs excitation (266 and 355 nm) and the optical inte rference filters/PMT detector (Sivaprakasam, 2002). Fluorescence spectra of various natural water samples recorded with this laboratory setup are shown in Fig. 3.5. Rayleigh scattering at the excitation wavelength of 266 nm, as well as Raman scattering at 290 nm, were observed, together with a broad fluorescence peak due to CDOM (DOC) with a maximum at 450 nm. The river water had the largest fluorescence intensity, and dist illed water the smallest. In Fig. 3.6, LIF of different drinking water samples taken with the same apparatus is plotted. The wide

PAGE 49

28 Table 3.1 Absorption and emission wavele ngths of some organic substances. Fluorophore Peak absorption wavelength Peak emission wavelength Source/Type Hydrophobic acid fraction, A 237 260 nm 400 500 nm Humic-like M 290 310 nm 370 410 nm Marine humiclike; Anthropogenic, wastewater & agriculture C 300 370 nm 400 500 nm Humic-like; Terrestrial, anthropogenic, agriculture 260 nm 430 nm, 540 nm Fluorescent whitening agent Naphtalene 220 230 nm 340 370 nm XOM landfill leachate Chlorophyll a 431 nm 670 nm Chlorophyll b 435 nm 659 nm Bisphenol A 230 nm 310 nm, 410 nm Some large polycyclic aromatic hydrocarbons (LPAH) 350 400 nm 400 500 nm 250 nm 504 nm Fulvic acid

PAGE 50

Figure 3.4 Schematic diagram of the labor atory LIF setup deve loped at USF (from Sivaprakasam and Killinger (2003, JOSA B), reprinted with permission from Optical Society of America). 29

PAGE 51

Figure 3.5 LIF of natural water samples recorded with laboratory setup (from Killinger and Sivaprakasam (2003, Appl. Opt.), re printed with permission from Optical Society of America). 30

PAGE 52

Figure 3.6 LIF of drinking water samples recorded with laboratory setup (from Killinger and Sivaprakasam (2006), reprinted with permission from Optical Society of America). 31

PAGE 53

32 peak at 450 nm is present for some samples; brand #1 has a peak at 330 nm associated with plasticizer residue. Figure 3.7 demonstr ates LIF spectra of bi sphenol-A plasticizer in distilled water at differe nt concentrations. Presen ce of the compound down to ten parts-per-billion was detected (Sivaprakasam & Killinger, 2003). A schematic diagram and a photograph of the portable LIF system are given in Figs. 3.8 and 3.9. Linearity and sensitivity lim its of the apparatus in terms of quinine sulfate in sulfuric acid and bisphenol-A in distilled water are shown in Fig. 3.10. The portable system was used during a two-day cruise in the Gulf of Mexico. As can be seen from Fig. 3.11, CDOM fluorescence in the harbor (start and end point s of the recording) far exceeds that of the blue water offshore, but even there the system detected a noticeable LIF signal (Killinger & Sivaprakasam, 2006).

PAGE 54

Figure 3.7 LIF of bisphenol A at different c oncentrations recorded with laboratory setup (from Killinger and Sivaprakasam (2006) reprinted with permission from Optical Society of America). 33

PAGE 55

Figure 3.8 Schematic diagram of the por table LIF system developed at USF. 34

PAGE 56

Figure 3.9 Photograph of the portabl e LIF system developed at USF. 35

PAGE 57

Figure 3.10 Linearity and sensit ivity of the portable LIF syst em developed at USF (from Sivaprakasam, Shannon, Luo, Coble, Boehme, and Killinger (2003), reprinted with permission from Optical Society of America). 36

PAGE 58

Figure 3.11 LIF of water in the Gulf of Mexi co recorded with the portable LIF system developed at USF (from Sivaprakasam, Shannon, Luo, Coble, Boehme, and Killinger (2003), reprinted with permission from Optical Society of America). 37

PAGE 59

38 CHAPTER FOUR UV LIF SYSTEM MEASUREMENTS OF CONTAMINANTS IN DRINKING AND REVERSE OSMOSIS PROCESSED WATER This chapter describes the custom ma de laser-induced fluorescence system, including initial modifications to it, and measurements related to reverse-osmosis water processing, as well as spectra from various drinking water samples, and biologically and chemically contaminated water. 4.1 Portable LIF System Fluorescence measurements of water we re performed using a LIF system, a schematic diagram of which is shown in Fig. 4.1. The system had two interchangeable microchip laser sources, 266 nm and 355 nm (JDS Uniphase, Models NU-10110-100 and NV-10110; see Table 4.1 for output characteristics of the micr ochip lasers). A silicon APD photodetector (New Focus, Model 1621) was used to trigger data acquisition. The laser beam passed through th e optical quartz cell (Spectro cell Inc., Model RF-3010-F) multiple times, for which two plane mirrors on two sides of the cell were used. Fluorescence from a sample was collected at 90 to the excitation beam by a concave mirror (Optosigma, Model 035-0130) and a fu sed silica lens (Optosigma, Model 0140490), and went through one of the bandpass op tical filters (Table 4.2) spanning the UV

PAGE 60

Figure 4.1 Schematic diagram of the portable LIF system. 39

PAGE 61

40 Table 4.1 Specifications for the 26 6 nm and 355 nm microchip lasers. Parameter 266 nm Microchip Laser 355 nm Microchip Laser Wavelength 266 nm 355 nm Pulse width 0.4 ns 0.4 ns Pulse energy (calculated) 0.4 J 0.3 J Repetition rate 7.1 kHz 12.1 kHz Peak power (calculated) 1 kW 0.8 kW Average power 3 mW 4 mW Polarization (H:V) 100:1 100:1 Beam size 1 mm 1 mm

PAGE 62

41 Table 4.2 Specifications for the bandpa ss interference and cutoff absorption filters in LIF system and PMT efficiency. Wave length Bandwidth of the interference filters Transmission of the interference filters PMT quantum efficiency Cutoff filter 1: 280 nm Cutoff filter 2: 305 nm Cutoff filter 3: 365 nm Cutoff filter 4: 320 nm Cutoff filter 5: 420 nm 265 8.9656 17.4 27.74124 0.15436 0.17057 0.01448 0.02574 0.00799 291 12.485 16.6 28.30633 0.64624 0.28537 0.01448 0.02574 0.00799 314 12.307 19 28.20746 0.84005 0.78088 0.01448 0.25157 0.01192 334 9.7469 28.5 27.31394 0.87414 0.87601 0.01965 0.74778 0.01110 355 10.854 33.06 26.19718 0.88398 0.89133 0.09585 0.86434 0.00987 370 10.351 35.3 26.57143 0.88591 0.89023 0.47729 0.87470 0.00489 400 13.794 38.6 26.57143 0.88921 0.89394 0.83569 0.88339 0.23688 420 7.4717 39.9 23.40816 0.88571 0.89065 0.87808 0.88225 0.55332 436 7.8814 45.1 21.93971 0.88819 0.89339 0.89068 0.88515 0.69273 451 8.7577 58.2 20.42445 0.88904 0.89043 0.89378 0.88482 0.75991 470 9.9137 47.9 18.46809 0.89109 0.89424 0.89708 0.88623 0.80259 490 10.384 58.838 15.54519 0.89791 0.90023 0.90576 0.89454 0.83444 514.5 8.3283 53.267 14.28849 0.88955 0.89454 0.90161 0.88829 0.84087 540 10.682 49.069 11.31746 0.89261 0.89484 0.90205 0.88873 0.85074 560 9.1175 55.159 10.28061 0.88508 0.88334 0.88870 0.87966 0.84394 590 8.9994 52.74 9.15738 0.88626 0.87970 0.87656 0.86193 0.83741 630 9.5846 48.695 7.17007 0.89328 0.89899 0.88459 0.87518 0.84904 660 11.817 61.76 5.77056 0.89910 0.89806 0.88396 0.88894 0.85125 685 9.7083 54.7 4.13764 0.89778 0.90110 0.88032 0.89175 0.85224

PAGE 63

42 and visible region before be ing focused by another lens onto a PMT (Hamamatsu, Model H6780-03; see Table 4.3). Furthermore, one of several absorption cu t-off filters (CVI Laser, Models CG-WG-280-2.00-2 and CG-WG295-2.00-2) was used to block Rayleigh and Raman scattered light. All filters were mounted on three motorized filter wheels (CVI Laser, Models AB-302 and AB-304). The PMT signal was processed by a gated integrator and boxcar averager unit (Stanf ord Research System, Model SR-250). The process of data collection was handled by LabVIEW software through a computer interface unit (Stanford Research System, Model SR-245). The typical settings during LIF measurements are given in Table 4.4. It should be noted that while a 355 nm laser shown in Fig. 4.1 was installed previously, it was primarily used for absorpti on measurements in earlier experiments. The 355 nm LIF measurements to be descri bed in the following sections are new. 4.1.1 System Modifications The LIF apparatus was modified by the au thor in several ways. First, a precision adjustable mount was devised fo r the 355 nm laser to simplify system alignment. Then, absorption detectors were removed from the optical box to make space for the LED installation. Th e details of LED mounting inside the LIF system are given in Chapter Nine Moreover, the LabVIEW program controlling data acquisition was comple tely rewritten to accommodate the differences in pulse repetition frequencies and wavelengths (and, therefore, filter choices) between lasers and LEDs, and to al low flexibility in data collection. A photograph of the LIF system including th ese modifications is given in Fig. 4.2.

PAGE 64

43 Table 4.3 Specifications for the Hamamatsu PMT. Parameter Hamamatsu PMT Model H6780-03 specifications Effective area 8 mm Supply voltage 11.5 15.5 V Control voltage 0.25 0.95 V Min wavelength 185 nm Max wavelength 650 nm Rise time 0.78 ns

PAGE 65

44 Table 4.4 Typical settings for the LIF experiments. Parameter Portable LIF system setting Gate width 100 ns Boxcar averaging 300 pulses Sensitivity 200 mV Number of sample points per measurement 3000 Flow rate 0.3 L/min Linear flow rate through sample cell 5 cm/s

PAGE 66

(a) (b) Figure 4.2 Photograph of the modified por table LIF system: (a) optics box; (b) electronics box. 45

PAGE 67

46 4.2 LIF Measurements of Reverse Osmosis Processed Water The LIF system was used to analyze the samples of Reverse Osmosis (RO) treated water in collaboration with Prof. R obert Carnahan's water treatment and RO processing lab (USF College of Engineering). In addition to testing water before and after RO treatment, the effects of membrane fouling by scaling, as well as the influence of clay particles on the RO process in the presence of scaling, were investigated. Samples of ground water before and after RO processing were provided by Miles Beamguard; water samples related to membra ne fouling were provided by Dhananjaya Niriella. 4.2.1 Reverse Osmosis To treat water by RO, feed water is for ced through a semi-permeable membrane under pressure. The method is used to elim inate about 90% of contaminants, but is not particularly effective for biological pat hogen removal (APHA, AWWA, & WPCF, 1989; Nollet, 2000). 4.2.1.1 Principles of Re verse Osmosis System The process of regular osmosis is based on the principle that two solutions of different concentrations, separated by a se mi-permeable membrane, will have a water flow in the direction of lower concentration. This process stops when the concentration on both sides becomes the same (Crompton, 2000). However, if pressure (40 psi or more) is applied to the more concentrated solution, then the water flow occurs from the more concentrated to the less concentrated

PAGE 68

47 side. Thus, the less concentrat ed solution becomes even more diluted (purified). This is the process of reverse osmosis. Water en tering the system is called the feed, purified water is the permeate, and the more concen trated solution after the RO treatment is known as the concentrate. Figure 4.3 illustrates the RO process. RO is primarily used for desalinization a nd removal of particulates. The major drawback is a large amount of waste water, since recovery rates are relatively low. 4.2.1.2 RO Units at USF Prof. Carnahan's laboratory at USF had several RO units that were operational and being used for water treatment testing (Fig. 4.4). One unit, used in most experiments described below, was a small laboratory SEPA CF II Membrane Cell System (GE Osmonics SEPA, 2003). Its output was about 4 li ters per day, and the recovery rate was below 1%. Another was a large E-4H model capable of purifying 2.6 6.8 cubic meters per hour (16,200 43,200 gallons per day), w ith a maximum recovery of 75% (GE Osmonics E-Series, 2003). 4.2.2 LIF Measurements of Ground Water Treated by Reverse Osmosis The LIF system was used to analyze a nd measure the LIF spectra of the water used in the RO system. The samples of f eed and permeate water were circulated through the LIF system using a quartz flow cell connect ed by plastic tubing to a peristaltic water pump (Cole-Palmer Instruments Company, MasterFlex L/S Model 7554-90) situated

PAGE 69

Figure 4.3 Diagram illustrating the prin ciple of a reverse osmosis system. 48

PAGE 70

(a) (b) 49 Figure 4.4 Photographs of reverse osmosis sy stems at USF: (a) small SEPA unit, (b) ESeries unit (Courtesy of Prof. Carn ahan, USF College of Engineering).

PAGE 71

50 outside the system. Ground water from a USF shallow well was used as the feed (input) for RO treatment. The spectra were collected by averaging 300 laser pulses with a gate width of 100 ns and sensitivity of 200 mV/V. These se ttings were used for the majority of measurements. The data was corrected for the cutoff filter transmission, interference filter transmission and bandwidth, as well as the PMT quantum efficiency (Table 4.2). Fluorescence spectra of the shallow well ground water before and after RO processing, obtained with exci tation at 266 nm and 355 nm, are shown in Figs. 4.5 and 4.6, respectively. The spectrum of distille d water is present for comparison. The ground water exhibited a broad fluorescence peak around 400 550 nm typical for the Colored Dissolved Organic Matte r signature, likely due to fluorophores A and C as listed in Table 3.1 (Coble, 2007). The spectra demonstrate significant fluorescence reduction in the range of 370 500 nm, associated w ith various organic compounds, after the RO processing. In Fig. 4.5, a peak at 515 nm, associated with fulvic acid-like compounds (Coble, 2007), is still present, although reduced roughly in half The same peak is not apparent in Fig. 4.6, due to the diffe rent excitation wa velength of 355 nm. 4.2.3 Membrane Fouling and RO Efficiency The presence of foreign material in the feed water affects th e performance of the RO system. With time, it accumulates on the membrane and drastically reduces the efficiency of the unit. This process is known as membrane fouling. Two types of foulants were studied in our experiments: scaling (mineral precipitates) and colloidal matter in the form of kaolin clay.

PAGE 72

0 200 400 600 800 1000 1200 1400 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsRO input: ground water from USF shallow well RO output Distilled water Figure 4.5 LIF of ground water before and after RO treatment; 266 nm excitation; distilled water shown for comparison (May 13, 2005). 51

PAGE 73

0 200 400 600 800 1000 1200 1400 250300350400450500550600650700 Wavelength, nmFluorerscence intensity, arb. unitsRO input: ground water from USF shallow well RO output Distilled water Figure 4.6 LIF of ground water before and after RO treatment; 355 nm excitation; distilled water shown for comparison (May 13, 2005). 52

PAGE 74

53 4.2.3.1 Effect of Scaling on Water Treatment To imitate membrane fouling by scali ng, water with mineral content (sodium carbonate and calcium chloride) was used as th e feed for the RO system. Figures 4.7 and 4.8 show the LIF spectra of the input (feed) and output (permeate) samples for the two excitation wavelengths afte r a 6 hour scaling run. The fluorescence of the feed water had a broad peak, but the signal around 450 550 nm was stronger than at 400 450 nm, compared to the shallow well ground wa ter spectra in Figs. 4.5 and 4.6. Both graphs show a reduction in contaminant fluorescence for the RO treated water, particularly for the 450 nm signal. 4.2.3.2 Effect of Clay Particles in RO System with Scaling Next, kaolin clay was added to the input (feed) water in the amounts of 50 250 mg/L. LIF spectra of the feed and perm eate samples for 266 nm and 355 nm excitation wavelengths are presented in Figs. 4.9 and 4.10. One can see that there was little difference in the amount of fluorescence betw een the input and output samples after the clay was introduced. Comparing the spectra with and without clay material (Fig. 4.7 and Fig. 4.9; Fig. 4.8 and Fig. 4.10), it is reasonable to conclude th at the presence of clay has reduced the performance of the RO system dramatically. In addition, in the case of 266 nm excita tion, it even appears that the permeate has more fluorescence at the longer wa velengths than the feed water. Since the later effect was likely due to the fluorescence quenching by colloidal particles (clay), the fluorescence of clay suspended in distilled water at different concentrations was

PAGE 75

0 10 20 30 40 50 60 70 80 90 100 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsRO input: water with mineral materials RO output Deionized water Figure 4.7 LIF of water with mineral conten t (to model membrane fouling) before and after RO treatment; excitation 266 nm; deionized water shown for comparison. 54

PAGE 76

0 10 20 30 40 50 60 70 80 90 100 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsRO input: water with mineral materials RO output Deionized water Figure 4.8 LIF of water with mineral conten t (to model membrane fouling) before and after RO treatment; excitation 355 nm; deionized water shown for comparison. 55

PAGE 77

0 10 20 30 40 50 60 70 80 90 100 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsRO input: water with mineral materials and clay RO output Deionized water Figure 4.9 LIF of water with mineral conten t and clay particles (to model membrane fouling) before and after RO treatme nt; excitation 266 nm; deionized water shown for comparison. 56

PAGE 78

0 10 20 30 40 50 60 70 80 90 100 250300350400450500550600650700 Wavelength, nmFluorescence intens ity, arb. unitsRO input: water with mi neral materials and clay RO output Deionized water Figure 4.10 LIF of water with mineral conten t and clay particles (to model membrane fouling) before and after RO treatme nt; excitation 355 nm; deionized water shown for comparison. 57

PAGE 79

58 investigated. Figures 4.11 and 4.12 show the fluorescence signals as a function of clay concentration. Based on these plots, it is clear that quenchi ng began to affect fluorescence detection at clay concentrations exceeding 150 mg/L. 4.3 LIF Measurements of Various Water Samples The LIF system was used to study fluores cence spectra of a wide range of water samples. As can be seen from Figs. 4.13 and 4.14, Hillsborough River water (Tampa, FL) had the strongest fluorescence with a broad peak between 400 and 550 nm, corresponding to the large amount of Dissolv ed Organic Matter (DOM), or Dissolved Organic Compounds (DOC). Ground water from the shallow well at USF had a similar, but weaker, fluorescence. Dr inking water from the fountai n and tap water in the USF Physics building had, as expected, much sm aller fluorescence signal, with the maximum of the broad peak around 450 nm. Distilled water (Zephyrhills brand) had the smallest amount of fluorescence. However, its features were similar to the drinking water spectra. Despite overall similarities in the spectr a, in case of 266 nm excitation, the river and shallow well ground water had the stronge st signal at 515 nm (fulvic-like peak), while 450 nm (humic-like) signal domina ted in the drinking water samples. 4.3.1 Domestic and International Drinking Water The spectra of drinking water from seve ral sources in Tampa, FL, USA and St. Petersburg, Russia are shown in Figs. 4.15 and 4.16.

PAGE 80

0 10 20 30 40 50 60 70 80 90 100 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsDistilled water 50 mg/L clay in distilled water 150 mg/L clay in distilled water 250 mg/L clay in distilled water Figure 4.11 LIF of clay in di stilled water at different concentrations; 266 nm excitation. 59

PAGE 81

0 10 20 30 40 50 60 70 80 90 100 250300350400450500550600650700 Wavelength, nmFluirescence intensity, arb. unitsDistilled water 50 mg/L clay in distilled water 150 mg/L clay in distilled water 250 mg/L clay in distilled water Figure 4.12 LIF of clay in di stilled water at different concentrations; 355 nm excitation. 60

PAGE 82

0 200 400 600 800 1000 1200 1400 250300350400450500550600650700 Wavelength, nmFluirescence intensity, arb. unitsHillsborough River water USF shallow well ground water USF Physics drinking fountain water USF Physics tap water Disilled water Figure 4.13 LIF of various water sample s; 266 nm excitation (June 1, 2005). 61

PAGE 83

0 200 400 600 800 1000 1200 1400 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsHillsborough River water USF shallow well ground water USF Physics drinking fountain water USF Physics tap water Distilled water Figure 4.14 LIF of various water sample s; 355 nm excitation (June 1, 2005). 62

PAGE 84

0 50 100 150 200 250 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsTap water from apartment in St. Petersburg, Russia Tap water from apartment in Tampa, FL Tap water from apartment in Tampa, filtered by PUR system Ground water from countryside near St. Petersburg, Russia Tap water from USF Physics building Distilled Figure 4.15 LIF of international and do mestic drinking water samples; 266 nm excitation; distilled water shown for comparison (June 11, 2007). 63

PAGE 85

0 50 100 150 200 250 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsTap water from apartment in St. Petersburg, Russia Tap water from apartment in Tampa, FL Tap water from apartment in Tampa, filtered by PUR system Ground water from countryside near St. Petersburg, Russia Tap water from USF Physics building Distilled Figure 4.16 LIF of international and do mestic drinking water samples; 355 nm excitation; distilled water shown for comparison (June 11, 2007). 64

PAGE 86

65 It is interesting to note the similarity (both in pattern and intensity) between the fluorescence of tap water taken from apartments in these two cities. These spectra had a broad peak in 400-550 nm region, with a di p around 435 nm. The likeness between these spectra suggests very similar composition of the samples, and might be due to the analogous water treatment processes. Another notable resemblance was betw een the LIF spectra of the Tampa apartment water filtered by a commercially available PUR brand on-tap system, the ground water from the St. Petersburg (Russia) countryside, and the tap water from the USF Physics building. These spectra had le ss than a half of the fluorescence intensity compared to the apartment tap LIF, and in the case of 266 nm excitation, the strongest signal shifted from 470 nm to 450 nm. Otherwise, the spectr a had the same features as the apartment tap water spectra. 4.3.2 Bacterial Spores in Distilled Water As part of a related but se parate research study into LI F of bacteria endospores (Makoui & Killinger, February 2009; Makoui & Killinger, March 2009), we had an opportunity to study the LIF of selected b acterial spores. As such, the natural fluorescence of Bacillus Subtilis (also known as Bacillus Globigii ) endospores at 3.5 g/L in distilled water was measured a nd is shown in Figs. 4.17 and 4. 18. Bacillus Subtilis is an anthrax simulant. These spectra were obtained without doping by terbium typically used for such measurements. As can be s een, they feature a broa d peak between 400 and 550 nm representative of organic fluorescence.

PAGE 87

0 200 400 600 800 1000 1200 1400 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsBacillus Subtilis in distilled water Distilled water Figure 4.17 LIF of Bacillus Subtilis spores in distilled water; 266 nm excitation. 66

PAGE 88

0 200 400 600 800 1000 1200 1400 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsDistilled water Bacillus Subtilis in distilled water Figure 4.18 LIF of Bacillus Subtilis spores in distilled water; 355 nm excitation. 67

PAGE 89

68 The greatest difference between B. Subtilis spectra and typical water fluorescence was seen with 266 nm excitation. The bacter ial spectrum had a str onger signal at 370 nm, relative to the rest of fluorescence emission, than any na tural or drinking water samples. 4.3.3 Malathion in Distilled Water Fluorescence of an insecticide malathion (a VX simulant) was measured and is presented in Figs. 4.19 and 4.20. The sample wa s made by dilution of 4 mL of Dexol in 1 L of distilled water. The spectrum excited by 266 nm light has a distinct peak at 370 nm that might be characteristic of the substa nce. However, much more studies would be required to better quantify the spectrum for future sensing. 4.4 General Observations Regarding Deep UV LIF Spectra Overall, comparison between 266 nm a nd 355 nm excitation wavelengths for water fluorescence have shown that the shorte r excitation wavelength usually allowed to observe more features in th e spectrum, mostly due to the separation between Raman scattering and fluorescence signals. That is, the deeper UV excitation wavelength was more likely to result in distinguishable spectra from various water sources. These results were reported in an IJHSES pa per (Sharikova & Killinger, 2007). The spectra measured for the natural (river, lake) water samples were in agreement with similar spectra, for the same excitation wavelengths, as reported by Coble (1996, 2007), Hudson et al (2007), and other re searches in the field (see Fig. 3.3).

PAGE 90

0 50 100 150 200 250 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsMalathion in distilled water Distilled water Figure 4.19 LIF of malathion in di stilled water; 266 nm excitation. 69

PAGE 91

0 50 100 150 200 250 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsMalathion in distilled water Distilled water Figure 4.20 LIF of malathion in di stilled water; 355 nm excitation 70

PAGE 92

71 CHAPTER FIVE UV LIF LONG-TERM CONTINUOUS MONITORING OF TAP WATER In this chapter, the continuous, long-term measurements of water fluorescence using the portable LIF system, particularly the online monitoring of tap water, are described. 5.1 Experimental System for Long-Term Continuous Monitoring of Tap Water The portable LIF system from Chapter F our was connected to the drinking water supply in our laboratory as shown in Fig. 5.1. The overflow contro l container and the external peristaltic pump (Col e-Palmer Instruments Compa ny, MasterFlex L/S, Model 7554-90) were employed to make sure the cons tant flow of tap wa ter through the flow cell was not influenced by pressure changes in the water pipes. A flow rate of 0.3 liters per minute (setting "2" on the pump) was maintained throughout the experiment. 5.2 Flowing Tap Water Monitoring Initially, the fluorescence of the running tap water was observed for several hours. The changes in the signal over time were compar ed to the typical sta ndard deviation of a recorded measurement (which was itself an average of 3000 sampled points), and to the changes in a control sample: ta p water continuously recirculat ed through the system in a

PAGE 93

Figure 5.1 Schematic diagram of water flow for continuous tap water monitoring. 72

PAGE 94

73 closed loop under the same conditions as the running water. These measurements have shown that the signal uncertainty due to the instrumentation was less than 1 %, and more often about 0.1%. After it was clear that the changes obse rved in fluorescence of the running tap water exceed both the measurement uncertainty and variations over time in the control sample, a longer-term experiment was performe d. The LIF system recorded the running tap water fluorescence continuously for a week. To confir m that the periodic changes observed in the signal were not related to long-term drifts in the system itself, a six-day experiment with recirculated ta p water was carried out as well. 5.2.1 LIF Signal of Flowin g Tap Water over Twelve Hours The LIF of running tap water for all channe ls as a function of time is shown as a 3D plot in Fig. 5.2 (266 nm excitation) a nd Fig. 5.3 (355 nm excitation). A twodimensional view of several fluorescence ch annels and of Raman scattering (291 nm for 266 nm excitation; 400 nm for the excitati on of 355 nm) is given in Figs. 5.4-5.5. The peak fluorescence for both excitation wavelengths was at 451 nm. Fluorescence from the organic species (370 nm to 560 nm) was changing in a similar fashion for all these channels, dipping in the first 4 hours a nd trending upwards after that (Figs. 5.4-5.5). The Raman peak's behavior for the second half of the experiment was quite different: it was either decreasing ( 266 nm excitation) or remained unchanged (355 nm excitation). For the 266 nm excitation, the average re lative uncertainty of the signal at a particular wavelength (average of standard deviations of the measurements in a given

PAGE 95

Figure 5.2 LIF of flowing tap water fo r 12 hours continuous monitoring; 266 nm excitation (Sept. 11, 2006). 74

PAGE 96

Figure 5.3 LIF of flowing tap water fo r 12 hours continuous monitoring; 355 nm excitation (Sept. 11, 2006). 75

PAGE 97

0 20 40 60 80 100 120 140 024681 0 Time, hoursFluorescence intensity, arb. units291 nm 314 nm 420 nm 451 nm 560 nm 685 nm 1 2 515 nm Figure 5.4 LIF of flowing tap water at se lected wavelengths for 12 hours continuous monitoring; 266 nm exc itation (Sept. 11, 2006). 76

PAGE 98

0 20 40 60 80 100 024681 0 Time, hoursFluorescence intensity, arb. units400 nm 420 nm 451 nm 560 nm 685 nm 1 2 515 nm Figure 5.5 LIF of flowing tap water at se lected wavelengths for 12 hours continuous monitoring; 355 nm exc itation (Sept. 11, 2006). 77

PAGE 99

78 channel over the average of those measurements) was 0.01% for the Raman peak at 291 nm, below 0.1% for the fluorescence channels starting at 370 nm, and below 1% for 314 nm and 355 nm. Changes in the signal at a given wavelength over time (standard deviation of the signal in a given channel over the average of that signal) was 3% at 291 nm, 9% for 314 nm (plastics), 5% to 10% for wavelengths 370 660 nm, and 17% for 685 nm (chlorophyll). For the 355 nm excitation, the relative uncertainty was below 0.06% for all channels from 355 to 590 nm, and below 0.3% for the remaining ones, while signal variation was 4% for the Raman peak at 400 nm, 7-10% for 420 to 630 nm, and reached 27% at 685 nm. In all these cases, the signal variation over time was about 100 times greater than the measurement uncertainty, indicating that the changes in fluores cence were due to variations in the organic c ontent of the tap water and not the instrument error. 5.2.2 Peak Fluorescence Channel versus Raman Channel To distinguish the actual variation of the fluorescence signal from other parameters affecting the experiment, one can compare the LIF signal with the water Raman signal. The ratio of the maximum fluorescence signa l to the Raman peak as a function of time for the two excitation wavelengths is show n in Figs. 5.6 and 5.7. In both cases, the ratio increased by about 5% during 12 hours.

PAGE 100

0 0.2 0.4 0.6 0.8 1 1.2 024681 0 Time, hoursMax LIF (451 nm) to Raman (291 nm) ratio1 2 Figure 5.6 Ratio of LIF (451 nm) to Raman (291 nm) signals of flowing tap water for 12 hours continuous monitoring; 266 nm excitation (Sept. 11, 2006). 79

PAGE 101

0 0.2 0.4 0.6 0.8 1 1.2 1.4 024681 0 Time, hoursMax LIF (451 nm) to Raman (400 nm) ratio1 2 Figure 5.7 Ratio of LIF (451 nm) to Raman (400 nm) signals of flowing tap water for 12 hours continuous monitoring; 355 nm excitation (Sept. 11, 2006). 80

PAGE 102

81 5.2.3 Comparison with Recirculated Tap Water Sample Three-dimensional plots of fluoresce nce variation over four hours for a recirculated tap water sample are presented in Fig. 5.8 (266 nm excitation) and Fig. 5.9 (355 nm excitation). Maximum fluorescence was at 451 nm for both excitation wavelengths. Unlike in the running tap water case, LIF signals grew very slowly and monotonously over time. In particular, for th e 266 nm excitation, the Raman signal (291 nm) increased faster than the organic peak at 451 nm, and the signal at 314 nm (plastics) more than doubled due to leaching from the soft tubing. Such leaching from soft plastic or rubber tubing has been measured before by our group (Sivaprakasam, 2002). Overall, the fluorescence signal uncertainty was similar to or lower than that of the running water for the 4-hour period for most channels, but had somewhat greater values at 685 nm (0.3% and 0.5% for the two excitation wavelengths). For the organic fluorescence channels, the changes in the sign al were 1.5 3 times smaller than for the running tap water. However, for the plas ticizer fluorescence channels (314 334 nm) the signal growth was two times greater than in the case of running water. These results are consistent with our group's previous measurements of long-term plastic tubing leaching and particulate generation in a cl osed recirculating cell and pump reservoir configuration (Sivaprakasam, 2002). 5.2.4 Week-Long Tap Water Monitoring A week-long continuous observation of the running tap water from the laboratory drinking water supply was compared to the si x-day experiment invol ving a recirculated sample from the same source.

PAGE 103

Figure 5.8 LIF of recirculated tap water for 4 hours continuous monitoring; 266 nm excitation (Oct. 10, 2006). 82

PAGE 104

Figure 5.9 LIF of recirculated tap water for 4 hours continuous monitoring; 355 nm excitation (Oct. 10, 2006). 83

PAGE 105

84 5.2.4.1 Running Tap Water The results of a seven-day long (Frida y through Thursday) monitoring of running tap water LIF are shown in Figs. 5.10 and 5.11 for 266 nm and 355 nm excitation respectively. As can be seen, large period ic variations in fluo rescence were occurring simultaneously for both excitation waveleng ths. The peaks correspond roughly to midday on Saturday and Monday, Tuesday through Wednesday, and Thursday noon. The Rayleigh scattering as a function of time for the two excitation wavelengths is plotted in Figs. 5.12 and 5.13. The observe d decrease in the s cattering signal with time may be indicative of a reduced amount of Mie scattering from particulates present in the infrequently used pipes. The observe d spikes in Fig. 5.12 are probably due to surface/particle enhanced scattering and have been reported before by Sivaprakasam and Killinger (2003, J. Opt. Soc. Am. B). Several fluorescence channe ls and the Raman signal as a function of time are shown in Fig. 5.14 (266 nm excitation) and Fi g. 5.15 (355 nm excitation). While all organic fluorescence channels have similar be havior, with repeated maxima and minima as described above, the Raman channel either does not have the same pattern (266 nm excitation), or reflects it to a far lesser extent, likely due to the overlap between Raman and fluorescence signals at 400 nm (355 nm exci tation). Channels at 314 nm (plastics) and 685 nm (chlorophyll) do not demonstrate the periodic behavior either. The signal uncertainty typically was below 0.3%. Overall, changes in the signal were 12% at 291 nm (Raman scattering), 20% at 314 nm fluorescence, 10 to 13% in 370 nm 660 nm channels, and 20 % at 685 nm for

PAGE 106

Figure 5.10 LIF of flowing tap water for 7 days continuous monitoring; 266 nm excitation (Nov. 2006). 85

PAGE 107

Figure 5.11 LIF of flowing tap water for 7 days continuous monitoring; 355 nm excitation (Nov. 2006). 86

PAGE 108

0 500 1000 1500 2000 0 20 40 60 80100120140160 Time, hoursRayleigh scattering (266 nm) intensity, arb. units Figure 5.12 Rayleigh scattering (266 nm) of flowing tap water for 7 days continuous monitoring; 266 nm excitation (Nov. 2006). 87

PAGE 109

0 500 1000 1500 2000 2500 0 20 40 60 80100120140160 Time, hoursRayleigh scattering (355 nm) intensity, arb. units Figure 5.13 Rayleigh scattering (355 nm) of flowing tap water for 7 days continuous monitoring; 355 nm excitation (Nov. 2006). 88

PAGE 110

0 20 40 60 80 100 120 140 0 20 40 60 80100120140160 Time, hoursFluorescence intensity, arb. units291 nm 314 nm 420 nm 451 nm 560 nm 685 nm 515 nm Figure 5.14 LIF of flowing tap water at se lected wavelengths for 7 days continuous monitoring; 266 nm excitation (Nov. 2006). 89

PAGE 111

0 20 40 60 80 100 0 20 40 60 80100120140160 Time, hoursFluorescence intensity, arb. units400 nm 420 nm 451 nm 560 nm 685 nm 515 nm Figure 5.15 LIF of flowing tap water at se lected wavelengths for 7 days continuous monitoring; 355 nm excitation (Nov. 2006). 90

PAGE 112

91 266 nm laser. For 355 nm excitation, they were 6% at 400 nm Raman scattering, 9 to 14% in 420 nm 660 nm channels, and 24% at 685 nm. For 266 nm excitation, the correlation coeffi cient between the individual spectral channels was found to be highest between the peak fluorescence at 451 nm and surrounding 370 660 nm channels (exceeding 0.98 for 370 590 nm). Correlation between the 291 nm Raman peak and 314 nm plasticizer channels was 0.94, and 0.79 between the Raman and 685 nm chlorophyll chan nels. For 355 nm excitation, the same pattern was observed; the co rrelation coefficient was a bove 0.98 between the 420 560 nm and the 451 nm fluorescence channels; the correlation coefficient for the 400 nm Raman and 685 nm chlorophyll channels was 0.48. This suggests that the composition of organic matter (the relative intensities of the different fluorescence peaks) in the running tap water did not change significantl y during the experiment, however the overall amount did vary to a moderate degree. The peak fluorescence to Raman signal ratio is plotted in Figs. 5.16 5.17. Both graphs show the periodic variation, but for 266 nm excitation it is superimposed on an upward sloping baseline. Because of that, of the overall change in this ratio for the 266 nm excitation (6%) is greater than for 355 nm excitation (4%). 5.2.4.2 Recirculated Tap Water Fluorescence of a recirculated tap water sample was monitored continuously for six days. The 3D plots are shown in Fi g. 5.18 (266 nm excitation) and in Fig. 5.19 (355 nm excitation; in this latter case, the laser stopped working after four days). The

PAGE 113

0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 20 40 60 80100120140160 Time, hoursMax LIF (451 nm) to Raman (291 nm) ratio Figure 5.16 Ratio of LIF (451 nm) to Raman (291 nm) signals of flowing tap water for 7 days continuous monitoring; 266 nm excitation (Nov. 2006). 92

PAGE 114

0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 20 40 60 80100120140160 Time, hoursMax LIF (451 nm) to Raman (400 nm) ratio Figure 5.17 Ratio of LIF (451 nm) to Raman (400 nm) signals of flowing tap water for 7 days continuous monitoring; 355 nm excitation (Nov. 2006). 93

PAGE 115

Figure 5.18 LIF of recirculated tap water for 6 days continuous monitoring; 266 nm excitation (Dec. 2006). 94

PAGE 116

Figure 5.19 LIF of recirculated tap water for 4 days continuous monitoring; 355 nm excitation (Dec. 2006). 95

PAGE 117

96 fluorescence signal increased dur ing the first 20 30 hours, th en leveled off; no periodic fluctuations were observed in the recirculated sample. Variations in the Rayleigh signal and se lected fluorescence channels over the same period of time are shown in Figs. 5.20 .21 and Figs. 5.22 5.23, respectively. The drop-off and subsequent gradual increase in the Rayleigh-Mie scattering in the first hours of observation are probably due to the settling and partial re-agitation of the particles present in the sample. Fluores cence and Raman signals mostly grow for the first 20-30 hours, and then slow down, without fluctuations that have been seen in the running water experiment. The signal at the 314 nm (plastics) channel increased 3 times from its initial value. The average signal uncertainty for most channels was below 0.1%. Changes over time in the Raman peak were 2 3% for either excitation wavelength. For 266 nm excitation, they we re 17% at 314 nm, stayed at or below 1% for 370 nm 630 nm and 685 nm fluorescence ch annels, and were 6% at 660 nm. For 355 nm laser excitation, they were below 4% for 420 560 nm, and within 6% for 590 685 nm. Comparing changes in the LIF signal over time between the running and recirculated water, the greatest difference was seen for the DOC fluorescence channels (370 630 nm) with 266 nm excitation, where it was about 10 times greater for the running water than for the recirculated water. The correlation coefficient was higher betw een the peak fluorescence at 451 nm and other channels than between the Ra man signal and fluorescence. For 266 nm excitation, it exceeded 0.99 for 370 560 nm, and was 0.96 for 314 nm and 0.71 for

PAGE 118

0 200 400 600 800 1000 0 20 40 60 80 100 120 140 Time, hoursRayleigh scattering (266 nm) intensity, arb. units Figure 5.20 Rayleigh scattering (266 nm) of recirculated tap water for 6 days continuous monitoring; 266 nm excitation (Dec. 2006). 97

PAGE 119

0 500 1000 1500 2000 2500 01 02 03 04 05 06 07 08 Tme, hoursRayleigh scattering (355 nm) intensity, arb. units0 Figure 5.21 Rayleigh scattering (355 nm) of recirculated tap water for 4 days continuous monitoring; 355 nm excitation (Dec. 2006). 98

PAGE 120

0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Time, hoursFluorescence intensity, arb. units291 nm 314 nm 420 nm 451 nm 560 nm 685 nm 515 nm Figure 5.22 LIF of recirculated tap wate r at selected wavelengths for 6 days continuous monitoring; 266 nm excitation (Dec. 2006). 99

PAGE 121

0 20 40 60 80 100 120 01 02 03 04 05 06 07 08 Time, hoursFluorescence intensity, arb. units400 nm 420 nm 451 nm 560 nm 685 nm 0 515 nm Figure 5.23 LIF of recirculated tap wate r at selected wavelengths for 4 days continuous monitoring; 355 nm excitation (Dec. 2006). 100

PAGE 122

101 685 nm. Correlation between the Raman 291 nm peak and other channels always remained smaller, reaching maximum of 0.96 at 400 nm. For 355 nm excitation, the 451 nm correlation coefficient was a bove 0.99 for 420 560 nm, but the Raman correlation coefficient was only slightly less. The peak fluorescence to Raman signal ratios are plotted in Figs. 5.24 and 5.25. For 266 nm excitation, the graph shows a slight decline in the first few hours and slow growth in the last few, staying mostly c onstant throughout. For the 355 nm laser, the ratio increased slowly and monotonously. The character of this dependence is very different from the periodic fluctuations obs erved in the running water. For 266 nm excitation, the ratio changed by 3% over time, and for 355 nm excitation it changed by 7% due to the upward slope. 5.3 Other Long-Term Experiments The LIF system was used to study long-t erm changes in the fluorescence of Total Organic Carbon (TOC) and the influence of ch lorine on TOC fluorescence. Tannic acid is often used to represent TOC in water analys is measurements. Distilled water and trace solutions of organic compounds we re recirculated in the portable LIF system for several hours. Tannic acid representing Total Organic Carbon (3 mg/L in deionized water) and chlorinated tannic acid (3 mg/L TOC and 12 mg/L Cl in deionized water) samples were prepared by Mr. Panagiotis Amitzoglou from the water processing la boratory of Prof. Audrey Levine (USF College of Engineering) The solutions were produced by dilution of pure chemicals in nanopure deio nized water. Concentrati ons of TOC and chlorine in the last sample were typical of those in drinking water.

PAGE 123

0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 20 40 60 80 100 120 140 Time, hoursMax LIF (451 nm) to Raman (291 nm) ratio Figure 5.24 Ratio of LIF (451 nm) to Raman (291 nm) signals of recirculated tap water for 6 days continuous mon itoring; 266 nm excitation (Dec. 2006). 102

PAGE 124

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 01 02 03 04 05 06 07 08 Time, hoursMax LIF (451 nm) to Raman (400 nm) ratio0 Figure 5.25 Ratio of LIF (451 nm) to Raman (400 nm) signals of recirculated tap water for 4 days continuous mon itoring; 355 nm excitation (Dec. 2006). 103

PAGE 125

104 5.3.1 Distilled Water Fluorescence of distilled Zephyrhills water (Figs. 5.26 5.27) was recorded for seven hours to observe changes in a pure samp le over time. Fluorescence due to the organic materials (370 560 nm) was somewh at different from the broad spectrum observed in the tap water with 266 nm excitati on. Here, the spectrum had three peaks of nearly identical intensity, at 370, 420 and 451 nm, separated by valleys at 400 and 436 nm. In Figs. 5.28 and 5.29, Raman signals over time are shown. For 266 nm excitation, the peak increased monotonously ; for 355 excitation, it stayed constant throughout the experiment. The fluorescen ce signals, plotted in Fig. 5.30 for 266 nm and in Fig. 5.31 for 266 nm excitation, grew w ith time to various degrees. Notably, the plasticizer-related signal at 314 nm was three times str onger than the largest CDOMrelated signal at 451 nm. The cause of the step-like nature of the signal growth in Fig. 5.30 is not well understood, but might be related to in teractions between various compounds or to changes in the particulate levels in the samp le. The steps typically occur after a period of an hour or so, and do not happen at the same time in all channels (the latter seems to eliminate the possibility of particles or othe r common influences as a cause). Further chemical studies are required to better understand this behavior. 5.3.2 Tannic Acid representing TOC The LIF spectra of tannic acid as a f unction of time were measured and are depicted in Fig. 5.32 for 266 nm excitation a nd in Fig 5.33 for 355 nm excitation. As

PAGE 126

Figure 5.26 LIF of distilled water fo r 7 hours continuous monitoring; 266 nm excitation. 105

PAGE 127

Figure 5.27 LIF of distilled water fo r 7 hours continuous monitoring; 355 nm excitation. 106

PAGE 128

0 20 40 60 80 100 120 140 01234567 Time, hoursRaman scattering (291 nm) intensity, arb. units Figure 5.28 Raman scattering (291 nm) of distilled water for 7 hours continuous monitoring; 266 nm excitation. 107

PAGE 129

0 20 40 60 0123456 Time, hoursRaman scattering (400 nm) intensity, arb. units7 Figure 5.29 Raman scattering (400 nm) of distilled water for 7 hours continuous monitoring; 355 nm excitation. 108

PAGE 130

0 5 10 15 20 25 0123456 Time, hoursFluorescence intensity, arb. units314 nm 420 nm 451 nm 560 nm 515 nm 370 nm 7 Figure 5.30 LIF of distilled water at selected wavelengths for 7 hours continuous monitoring; 266 nm excitation. 109

PAGE 131

0 2 4 6 8 10 0123456 Time, hoursFluorescence intensity, arb. units420 nm 451 nm 560 nm 515 nm 7 Figure 5.31 LIF of distilled water at selected wavelengths for 7 hours continuous monitoring; 355 nm excitation. 110

PAGE 132

Figure 5.32 LIF of tannic acid for 7 hours c ontinuous monitoring; 266 nm excitation. 111

PAGE 133

Figure 5.33 LIF of tannic acid for 7 hours c ontinuous monitoring; 355 nm excitation. 112

PAGE 134

113 can be seen, the fluorescence in Fig. 5.32 had the strongest peak at 370 nm, the secondstrongest at 420 nm, and a weaker peak at 451 nm. This made it quite distinct from the typical natural fluorescence of water, fo r example Fig. 4.15, which had a broad peak maximized at 451 nm. Figures 5.34 and 5.35 show that the Raman signal over time increased by about 15% for each excitation wavelength. Selected fluorescence channel signals as a function of time are plotted in Fig. 5.36 (266 nm exc itation) and Fig. 5.37 (355 nm excitation). These signals increased very slowly, only 314 nm (plastics) peak increased by 12 15%. A step-like behavior was also observed in the 291 nm Raman signa l and the channels associated with organic matter fluorescence. For 266 nm excitation, the ratio of the st rongest fluorescence peak (370 nm) to the Raman signal (291 nm) decreased by about 10%. For 355 nm excitation, the 451 nm fluorescence peak grew with respect to the 400 nm Raman signal by about 20%. 5.3.3 Chlorinated Tannic Acid Fluorescence of a solution containing both tannic acid and chlorine is shown in Figs. 5.38 5.39 for 266 nm and 355 nm exc itation respectively; the solution was recirculated for about 5 hours. As can be seen, the spectrum for the 266 nm excitation was significantly different from that of ta nnic acid alone. The peak at 370 nm was completely suppressed, and the 420 nm peak was reduced drastically, so that the strongest fluorescence was observed at 451 nm. The Raman signal of tannic acid with chlorine as a function of time is shown in Figs. 5.40 5.41. The changes in the Raman ch annel were within 5% of the average

PAGE 135

0 20 40 60 80 100 120 012345678 Time, hoursRaman scattering (291 nm) intensity, arb. units Figure 5.34 Raman (291 nm) signal of tannic acid for 7 hours continuous monitoring; 266 nm excitation. 114

PAGE 136

0 20 40 60 012345678 Time, hoursRaman scattering (400 nm) intensity, arb. units Figure 5.35 Raman (400 nm) signal of tannic acid for 7 hours continuous monitoring; 355 nm excitation. 115

PAGE 137

0 10 20 30 40 50 012345678 Time, hoursFluorescence intensity, arb. units314 nm 420 nm 451 nm 560 nm 685 nm 515 nm 370 nm Figure 5.36 LIF of tannic acid at select ed wavelengths for 7 hours continuous monitoring; 266 nm excitation. 116

PAGE 138

0 10 20 012345678 Time, hoursFluorescence intensity, arb. units420 nm 451 nm 560 nm 685 nm 515 nm Figure 5.37 LIF of tannic acid at select ed wavelengths for 7 hours continuous monitoring; 355 nm excitation. 117

PAGE 139

Figure 5.38 LIF of tannic acid and chlorine for 5 hours continuous monitoring; 266 nm excitation. 118

PAGE 140

Figure 5.39 LIF of tannic acid and chlorine for 5 hours continuous monitoring; 355 nm excitation. 119

PAGE 141

0 20 40 60 80 100 120 0123456 Time, hoursRaman scattering (291 nm) intensity, arb. units Figure 5.40 Raman (291 nm) signal of tannic acid and chlorine for 5 hours continuous monitoring; 266 nm excitation. 120

PAGE 142

0 20 40 60 0123456 Time, hoursRaman scattering (400 nm) intensity, arb. units Figure 5.41 Raman (400 nm) signal of tannic acid and chlorine for 5 hours continuous monitoring; 355 nm excitation. 121

PAGE 143

122 value for either excitation wavelength. The LIF intensity at selected channels as a function of time is shown in Fig. 5.42 fo r 266 nm excitation and Fig. 5.43 for 355 nm excitation. All signals were seen to increase over time to a varying degree. For example, fluorescence at 314 nm (plastics) grew by nearly 30% in the first hour and remained almost unchanged for the next three. The presence of chlorine in the tannic acid solution might have affected both the appearance and streng th of the solution's fluorescence. Further studies are needed to confirm whether the interactions between chemicals resulted in formation of new compounds, and if these processes were the cause of the unusual step-like changes in th e fluorescence of distilled wa ter (Figs. 5.30 and 5.31) and tannic acid solution (Figs. 5.36 and 5.37).

PAGE 144

0 5 10 15 20 25 0123456 Time, hoursFiuorescence intensity, arb. units314 nm 420 nm 451 nm 560 nm 685 nm 515 nm 370 nm Figure 5.42 LIF of tannic acid and chlorine at selected wavelengths for 5 hours continuous monitoring; 266 nm excitation. 123

PAGE 145

0 2 4 6 8 10 12 14 16 18 20 0123456 Time, hoursFluorescence intensity, arb. units420 nm 451 nm 560 nm 685 nm 515 nm Figure 5.43 LIF of tannic acid and chlorine at selected wavelengths for 5 hours continuous monitoring; 355 nm excitation. 124

PAGE 146

125 CHAPTER SIX RESEARCH AND DEVELOPMENT OF LED DRIVER The UV LIF system described previously was modified to use a compact and inexpensive ($200) UV Light-Emitting-Diode (L ED) to replace the expensive ($10,000) UV microchip laser. The early studies invol ved the use of several UV LEDs working at different wavelength for fluores cence measurements. This chapter primarily covers the early findings that the LED optical emissi on in the pulsed mode was not sufficiently stable or useful for a spectroscopic instru ment due to the limitations of the initial electrical driver used to power the LED. As such, this chapter discusses the development of a new multifunctional driver / power supply for the UV LED. It also describes testing the LED driver performance and presents its electri cal characteristics. It should be noted that five LEDs, at wavelengths of 265, 300, 320, 335, and 355 nm from Sensor Electronic Technology, Inc. we re initially used. However, one (320 nm) failed due to transient currents resulti ng from the poor power supply. This was a major motivation for the development of a better driver. Later chapters will cover the spectroscopic and optical performance of the LIF system using the UV LEDs, but this will be presented after the LED power supply driver studies and results are presented in this chapter.

PAGE 147

126 6.1 Initial Driv ers for UV LEDs The schematic diagrams of the electrical drivers used initially with the UV LEDs are shown in Fig. 6.1. These diagrams and supp lies represent a baseline type of system. Their primary advantage was the ease of impl ementation, particularly in the case of the CW mode driver. These driver s were used to collect the data in the early testing of our UV LED performance and in some LED -induced fluorescence experiments. However, the early drivers had a number of shortcomings, and were eventually replaced by an optimized multifunctiona l LED driver which was designed and constructed by the author, and are di scussed in the following sections. 6.1.1 Initial Continuous Wave Mode Operation The initial CW LED driver (Fig. 6.1(a)) consisted of an adjustable +/15 Volt dual-tracking DC power supply (Micronta, Model 22-121), and a 1 k resistor with the LED connected in series. The maximum setting of VPSmax = 15 V on the power supply resulted in approximately 10 mA current fl owing through the LED. The 10 mA current was 1/3 of the maximum rated value for CW mode and was deliberately chosen to prolong the LED life. The maximum possible current through the ci rcuit (for a shortcircuited LED) was 15 mA, which was half of the maximum rating. The current in the circuit was periodically measured by a multi meter (Cen-Tech digital multimeter, Model 90899).

PAGE 148

(a) (b) Figure 6.1 Schematic diagrams of the ini tial LED drivers: (a) CW mode; (b) pulsed mode. 127

PAGE 149

128 6.1.2 Initial Pulsed Mode Operation The circuit for the pulsed LED driver (Fig. 6.1(b)) was suggested by the textbook "The Art of Electronics" (H orowitz & Hill, 1989) and disc ussions with Marek Merlak from the USF Physics Department. The driv er used an adjustable dual power supply (Micronta, Model 22-121) that provided up to 15 V (up to 30 V in tracking mode) to the collector branch of a Darlington NPN transistor (STMicroelectronics, Model TIP120) consisting of a resistor R2 = 224 and an LED, and a pulse generator (Stanford Research Systems, Model DG535) conne cted through a resistor R1 = 1.63 k to the base of the transistor. The driver c ould provide pulses up to 40 mA. The signal generator could produce pulses varying from 5 picoseconds to 900 seconds, with repetition rates adjustable from 0.001 Hertz to 1 MHz. The current through the LED was determined by measuring the voltage drop across the R2 resistor during on and off times using a Tektronix TDS 210 oscilloscope connected between the R2 resistor and the LED. 6.2 Motivation for LED Driver Optimization A major drawback of the early drivers was the fact that they were voltagecontrollers, while an LED is a current-controll ed device. An LED output is determined by the current flowing through it. As a cons equence of voltage cont rol, the pulsed mode could not be operated at higher currents due to the danger of burni ng out the LED, since minor adjustments to the power supply voltage resulted in drastic changes in the LED current. Moreover, the same voltage across different LEDs does not necessarily result in the same current, and any changes in the LED resistance during operation lead to variations in the LED output even when the power supply voltage is unchanged. For that

PAGE 150

129 reason, driver settings had to be adjusted when switching from one LED to the next, and any drift in the LED current during operati on had to be manually compensated. Another problem of the initial CW driver wa s its inefficiency. To make sure that the LED was not damaged during the turn-on or voltage adjustment (as LED resistance drops rather precipitously wh en it is in a conducting stat e), and to keep the output fluctuations low, the resistor in such a circ uit has to be the element dissipating most of the power. For example, to guarantee that the current in the circ uit does not change by more than 10%, the resistor must dissipate 90% of the power. Fo r the typical forward LED voltage VLED of 5.5V and a 10 mA current, this condition require s a power supply providing 55V, and a resistor of 5 k The additional shortcoming of the pulsed driver turned out to be the distorted pulse shape it created. Unlike a clean square wave of the signal generator, the output pulse of the driver had noisy spikes at the st art and the end of the pulse, and a step in the pulse itself, also accompanied by spikes. These features contributed to the LED deterioration and lifetime failure. Furtherm ore, the driver increased the length of the pulse by about 5 s, which was significant for 10 s pulses. Finally, it was inconvenient to disconnect equipment and cables every time when switching from the CW to the pu lsed mode. Moreover, as a part of a compact LEDbased fluorescence detection system, the driv er had to be small and largely selfcontained. For these reasons, a new LED driver was designed and implemented. 6.3 Optimized Current Driver for LEDs The new driver (Fig. 6.2) was based on tr ansistor current source schematics

PAGE 151

Figure 6.2 Schematic diagram of the optimized LED driver. 130

PAGE 152

131 with pulsed mode operation implemented by m eans of a resistor-equipped transistor (RET), as described in "The Art of Electr onics" (Horowitz & Hill, 1989, p. 73) and in Philips PSSI2021SAY datasheet (Philips Electronics, 2004). 6.3.1 Driver Design The current-source driver (photograph in Fi g. 6.3) was powered by an AC adapter rated at 9VDC, modified to be used with a BNC cab le. It was equipped with an on/off switch and a knob to slowly turn on the power (a gua rd against transients ). Another knob was used to adjust the LED current between 10 mA and 50 mA. The transition between CW and pulsed operation was accomplished by another switch. The pulsed regime required an external source, such as a signal genera tor or a strobe signal of an Ocean Optics spectrometer. A special feature reducing the time constant associated with an LED capacitance could be used for the short (a few s) pulses. The LED current was monitored by connecting an oscilloscope between the LED cathode and ground, and reading the voltage across a 100 resistor placed in series with the LED. The driver circuit was enclosed in a 20 cm 10 cm 5 cm box with controls and connectors. 6.3.1.1 Optimized Driver Schematic The basis of the driver was a transistor current source, where the collector (load) current IC (total current flowing th rough the elements connected to the collector of the PNP transistor (STMicroelectronics, Model TIP125)) was determined by the emitter resistance RE (total resistance of the R3, R4 and RV2 combination), and did not depend on the collector voltage (Horowitz & Hill, 1989):

PAGE 153

(a) (b) Figure 6.3 LED driver photo: (a) inside view; (b) enclosure and connectors. 132

PAGE 154

133 Ic IE = VE/RE = (VB + 0.6V)/RE (6.1) where IE is the emitter current, VE is the emitter voltage, and VB is the base voltage. The base voltage was maintained at a consta nt offset from the power supply voltage VPS by a series of forward-biased diodes: VB = VPS Ndiodes 0.6V where Ndiodes is the number of diodes in series. The base resistor RB was chosen to bring the diodes into conduction. Equation 6.1 becomes invalid when the tr ansistor saturates (collector voltage VC becomes close to the emitter voltage VE). The last condition sets a limit on the load current IC. The resistor-equipped transistor (RET) c onsisted of a TIP120 transistor, a base resistor R1, and a resistor R2 set between the base and the emitter. When the input from the signal generator was high, the TIP120 emitter was in a conducting regime, and the effect of the RET on the rest of the circuit was as if RB was grounded; when the input was low, the TIP120 emitter was reverse-biased, a nd the effect was that of an open circuit below RB. When the LEDs were tested in a short (10 s) pulsed mode, due to the LED capacitance, the LED voltage signal had an exponentially decaying character after the driving pulse was turned off; this effect did not occur with a dumm y load. Since it would affect the fluorescence and also contribute to the LED degrada tion, a resistor in parallel with the LED was installed, chos en to be small enough to reduce the time constant to a few microseconds, but large e nough to draw less than 2% to 3% of the collector current at 50 mA. The parameters of the optimized LED driv er are given in Table 6.1, and detailed calculations are provid ed in Appendix A.

PAGE 155

134 Table 6.1 Nominal values of resistances and other parameters of the optimized LED driver. Turn-on: DC power supply, Vcc AC adaptor, 9 VDC 500 mA output, BNC Switch on/off 2-way RV1, trim-pot 0-50 kOhm, slow turn-on Current source: Transistor 1 TIP125 (PNP) Diodes Vd 0.6 V, 4 in series RB 6.8 kOhm nominal (7.29 kOhm measured) R emitter, variable: R4 18 Ohm nominal (18.9 Ohm measured) RV2, trim-pot 0-1000 Ohm, fine current setting R3 150 Ohm nominal (155 Ohm measured) Collector: LED, parameter range 0-60 mA, 0-7.5 V RCONTR, Iled monitoring 100 Ohm nominal (99.9 Ohm measured) Time constant switch Used as 2-way RPAR, time const reduction 4.7 kOhm nominal (4.7 kOhm measured) RET (resistor-equipped transistor) for pulsed mode: Transistor 2 TIP120 (NPN) R1 22 kOhm nominal (23.3 kOhm measured) R2 47 kOhm nominal (44.9 kOhm measured) Pulse/CW switch 2-way Pulse input, V pulse TTL, BNC

PAGE 156

135 6.3.2 Testing Current Driver Performance with a Dummy Load The driver operation in the CW regime was tested using a 670 resistor as a dummy load, close to the typical LED resist ance at 10 mA current. The goal was to check the driver performance and to determin e the limits for the variable parameters in the circuit. The load (collector) current IC as a function of the power supply voltage VPS (Fig. 6.4) was measured with RE set to 120 The current increased linearly up to approximately 10 V on the power suppl y before beginning to saturate. To investigate the collector current IC as a function of the load resistance (Fig. 6.5), the power supply was set to 15 V, and RE to 120 A "Dial-an-Ohm" device (General Resistance, Inc., Model DA 75-3X) was used as a load resistance. The current remained constant within 3% up to 1.1 k and decreased rapidly after that. Dependence of the collector current IC on the emitter resistance RE (the adjustable parameter in the driver circuit) was perfor med with a power supply voltage set to 15 V (Fig. 6.6) and the "Dial-an-Ohm" device was used as an emitter resistance. The current stayed within 8% of the initial value until emitter resistance reached 70 after which it declined swiftly. 6.3.3 LED Performance with Optimized Current Driver Electrical characteristics of the LEDs powered by the new driver were determined to make sure to follow the proper limits for operation. V-I curves and the LED current as a function of emitter resistance for four LEDs were recorded in CW mode and in pulsed (10 s, 500 Hz) mode. The power supply was kept at 15 V, a "Dial-an-Ohm"

PAGE 157

0 2 4 6 8 10 12 14 0 5 10 15 20 25 30 Power supply voltage, VCollector current, mA Figure 6.4 Collector current vs. power suppl y voltage for the LED driver with dummy load (CW mode). 136

PAGE 158

0 2 4 6 8 10 12 100 1000 10000 100000 1000000 Load resistance, OhmCollector current, mA Figure 6.5 Collector current vs. load resist ance the LED driver with dummy load (CW mode). 137

PAGE 159

0 2 4 6 8 10 12 14 16 18 20 1 10 100 1000 10000 100000 Emitter resistance, OhmCollector current, mA Figure 6.6 Collector current vs. adjustable emitter resistance for the LED driver with dummy load (CW mode). 138

PAGE 160

139 device was used as an emitter resistance, and the LED current was deduced from the voltage across resistor RCONTR through Ohm's law. 6.3.3.1 Continuous Mode In CW regime, the LED current had to be kept below 30 mA to avoid breakdown. V-I curves shown in Fig. 6.7 demonstrate the advantage of using a current-source driver. Each LED has a different turn-on voltage, so that to get 10 mA through the LEDs the voltage would have to be adjusted from 3.5 V (LED355) to 6 V (LED265). However, as can be seen from Fig. 6.8, the same value of the emitter resistance ensures the same current through all four resistors. 6.3.3.2 Pulsed Mode In the pulsed mode with a duty cycle of 1% or less, the maximum LED current must not exceed 200 mA. Fig 6.9 shows the V-I curves of the four LEDs driven by 10 s pulses at 500 Hz. As in the CW case, th e voltage current dependence is different for each LED. In Fig. 6.10, the differences in the LED current for the same value of emitter resistance are seen because of the time constant-reducing resistor RPAR connected in parallel with the LED. When the LED cu rrent is low, its resistance starts being comparable with the RPAR, and a greater portion of the colle ctor current is drawn through the resistor. It affects different LEDs to a greater or lesse r extent depending on their V-I curve. However, for the short, large pulses (10 s, 50 mA) the RPAR was designed for, the differences in the LED current are ne gligible, and for CW or long (millisecond) pulses it is not needed.

PAGE 161

0 5 10 15 20 25 30 0123456 LED voltage, VLED current, mALED355 LED335 LED300 LED265 7 Figure 6.7 LED V-I curves (CW mode). 140

PAGE 162

0.1 1 10 100 10 100 1000 10000 Emitter resistance, OhmLED current, mALED265 LED300 LED335 LED355 Figure 6.8 LED current vs. adjustable emitter resistance (CW mode). 141

PAGE 163

0 50 100 150 200 012345678 LED voltage, VLED current, mALED355 LED335 LED300 LED265 9 Figure 6.9 LED V-I curves (pulsed mode). 142

PAGE 164

0.1 1 10 100 1000 1 10 100 1000 10000 Emitter resistance, OhmLED current, mALED355 LED335 LED300 LED265 Figure 6.10 LED current vs. adjustable emitter resistance (pulsed mode). 143

PAGE 165

144 CHAPTER SEVEN OPTICAL CHARACTERIZATION OF UV LEDS In this chapter, optical ch aracteristics of the UV LED in terms of their spectral properties and power output are presented. Differences in the LED performance under three modes of operation are discussed. In particular, the output ch aracteristics of the LED were shown to be stable under CW ex citation, but the pulsed mode was not, and required modification of the pulse driver circuits as outlined in the pr evious chapter. As such, the pulsed LED characteristics after th e driver modification ar e presented in this chapter. 7.1 Specifications for the UV LEDs UV LEDs studied in this chapter were research grade AlGaN/GaN LED chips encapsulated in a metal-glass package fr om Sensor Electronic Technology, Inc., UVTOP series (Sensor Electronic Technology, In c., 2004). Their peak output ranged from 265 nm to 355 nm. A UV-transparent he mispherical lens installed on top of the package produced a nearly-collimated beam approximately 0.5 mm in diameter. LED specifications and measured parameters are listed in Table 7.1. In this text, the LEDs are denoted by their nominal emission wavelength.

PAGE 166

145 Table 7.1 Specifications and output characteristics for the UV-TOP LEDs. Parameter Value Maximum DC power dissipation, mW 150 Maximum forward DC current, mA 30 Maximum pulse forward current, mA (1kHz, 1%duty cycle) 200 Maximum reverse voltage, V 6 Maximum reverse current, A 100 Forward voltage, V 5.5 7.5 Spectrum half-width, nm 12 Output window Hemispheric lens LED output in 50 mA, 10 s, 500 Hz mode (without UG-11 filter) LED265 LED300 LED335 LED355 Pulse energy, J 7.14E-09 6.70E-09 6.80E-09 2.20E-08 Peak power, W 7.14E-04 6.70E-04 6.80E-04 2.20E-03 Ave. power, W 3.57E-06 3.35E-06 3.40E-06 1.10E-05 LED output in 10 mA, 5 ms, 100 Hz mode (without UG-11 filter) LED265 LED300 LED335 LED355 Pulse energy, J 3.97E-07 5.32E-07 4.98E-07 2.28E-06 Peak power, W 7.94E-05 1.07E-04 9.96E-05 4.57E-04 Ave. power, W 3.97E-05 5.32E-05 4.98E-05 2.28E-04 LED output in 10 mA, CW mode LED265 LED300 LED335 LED355 Ave. power, W 7.94E-05 1.07E-04 9.96E-05 4.57E-04 Ave. power after UG-11 filter, W 3.94E-05 8.89E-05 8.71E-05 2.99E-04 Average loss due to UG-11 filter, all modes, % 52 14 13 34

PAGE 167

146 7.2 Experimental Systems for Optical Characterization of UV Light Emitting Diodes The LED emission was measured and char acterized primarily in terms of its spectral properties, for which the setup shown in Fig. 7.1 was used. The light from the LED, attenuated by a neutral density filter when necessary, was collected by an Ocean Optics, Inc. UV-VIS fiber and analyzed by a spectrometer (Ocean Optics, Inc., Model ST2000) connected to a PC. The spectromete r specifications are given in Table 7.2. Data acquisition parameters such as integrati on time and spectra averaging were adjusted to obtain the best signal. To analyze the temporal properties of the LED emission in different modes of operation, the apparatus shown in Fig. 7.2 was used. The LED emission was separated into spectral components by a monochromator (Thermo Jarrel Ash, Model 82-415) and collected by a UV-sensitive detector (Hamamatsu PMT, Model H6780-04 or UDT Photodetector, Model UV-005). The signal wa s observed on an oscilloscope (Tektronix Model TDS 210), transmitted to the PC through a GPIB module, and recorded by LabVIEW software (Appendix B). 7.3 LED Emission Spectra A UV LED on a custom-made mount is shown in Fig. 7.3 (a). The LED illuminating a quartz sample cell is shown in Fig. 7.3 (b). The LED emission was invisible, but it induced a blue gl ow due to the sample fluorescence. The emission spectra of five UV LEDs collected using the apparatus from Fig. 7.1 are plotted in Fig. 7.4. Here, the spectra are no rmalized to the peak of each output curve,

PAGE 168

Figure 7.1 Schematic diagram of the apparatus used to study the spectral properties of the LED emission. 147

PAGE 169

148 Table 7.2 Specifications for the ST2000 spectrometer (Ocean Optics Inc.). Parameter Channels 1 & 2 Channel 3 Effective range 200 800 nm 530 1100 nm Grating spacing 600 lines/mm Blaze angle 400 nm 750 nm Entrance slit 25 m No Order sorting filter No Yes Fiber diameter 400 m (UV/VIS) 50 m (UV/VIS) Detector collection lens Yes No Detector 2048 element linear silicon CCD array Estimated sensitivity 86 photons/count (for 1 s integration) Integration time 3 ms to 30 s

PAGE 170

Figure 7.2 Schematic diagram of the a pparatus used to study the temporal characteristics of the LED emission. 149

PAGE 171

(a) (b) Figure 7.3 Photograph of a UV LED: (a) on a custom-made mount; (b) illuminating a sample in a quartz cell. 150

PAGE 172

0 0.2 0.4 0.6 0.8 1 1.2 250 300 350 400 Wavelength, nmNormalized LED emission, arb. unitsLED265 LED300LED320LED335 LED355 Figure 7.4 Normalized emission spectra of all LEDs; CW mode, 10 mA. 151

PAGE 173

152 because the calibration of the UV detector was not known for wavelengths below 400 nm. The peak emission wavelengths and Full-Width-Half-Maximum (FWHM) values are within manufacturer's specifications for all LEDs except the LED355, which has a side peak at 370 nm and a FWHM of 20 nm. 7.3.1 Out-of-Band Emission Figures 7.5 and 7.6 represent the emission sp ectra at several driving currents for the LED265 and LED320, respectively. The se mi-log scale was used to emphasize the out-of-band emission observed in each case. The secondary peak was on the order of 0.5% 1% of the main peak's intensity. B ecause the secondary emission occurred in the same spectral region as the fluorescence of organic compounds, it co uld contribute to the background noise in the fluorescence experi ments, and had to be eliminated. For that purpose, a colored glass UV-transmitting VIS-blocking filter (CVI, CGUG-11-1.00-1) was purchased and placed in front of the LED. The transmission of the UG-11 filter is shown in Fig. 7.7 as a dash-and-dot line. 7.3.2 Spectra with Visible-Blocking Filter The output spectrum of the LED265 a nd LED320, recorded with the VISblocking filter, is plotted in Fig. 7.8 and Fig. 7.9, respectively. It is clear that the filter completely blocked the out-of-band emi ssion for the LED320, while for the LED265 it was eliminated for wavelengths above 400 nm. However, the downside of using the filter was partial reduction in the UV light transmission.

PAGE 174

1 10 100 1000 10000 200 300 400 500 Wavelength, nmIntensity, arb. unitsPeak LED265 emission 10 m A 5 mA 1.6 mA Dark Out-of-band emission Figure 7.5 Emission spectra of LED 265 for different CW currents. 153

PAGE 175

1 10 100 1000 10000 200 300 400 500 600 Wavelength, nmIntensity, arb. units Peak LED320 emission 10 mA 5 mA 1.6 mA Dark Out-of-band emission Figure 7.6 Emission spectra of LED 320 for different CW currents. 154

PAGE 176

Figure 7.7 Transmission of UV-transmitting visible-blocking CG-UG-11 filter (from CVI Laser catalog). 155

PAGE 177

1 10 100 1000 10000 200 300 400 500 Wavelength, nmIntensity, arb. unitsPeak LED265 emission 10 m A 5 mA 1.6 mA Dark Figure 7.8 Emission spectra of LED265 for different CW currents with CG-UG-11 filter. 156

PAGE 178

1 10 100 1000 10000 200 300 400 500 600 Wavelength, nmIntensity, arb. units Peak LED320 emission 9 mA 5 mA 1.5 mA Dark Figure 7.9 Emission spectra of LED320 for different CW currents with CG-UG-11 filter. 157

PAGE 179

158 7.4 LED Emission: Electrical Modul ation versus Mechanical Chopping To compare the electrical modulation of a LED with a CW LED modulated by a mechanical chopper, the system from Fig. 7.2 was used. Frequencies of 100, 200 and 500 Hz with 50% duty cycle were studied. As can be seen from Figs. 7.10 7.11, both kinds of modulation resulted in nearly identical spectra. Typical waveforms recorded during the measurements are shown in Fig. 7.12, where the gradual geometrical opening of the mechanical chopper is evident. These results confirmed that electrical modula tion of the LEDs did not significantly change the optical properties of the LED emission compared to the CW regime. Therefore, electrical modulation was selected for the pulsed LED operation in later experiments. One major advantage of electrical modulation over CW LED / mechanical chopper combination was the opport unity to use higher pulse currents (50 mA instead of 10 mA) if the duty cycle was decreased below 1%. Space and power requirements of the system were cut as well. 7.5 LED Emission in Three Different Modes of Operation LED spectra for three operating modes: 10 mA CW; 10 mA pulses of 5 ms at 100 Hz (50% duty cycle); and 50 mA pulses of 10 s at 500 Hz (0.5% duty cycle) were recorded using the setup from Fig. 7.1. The same spectra were also taken with the VISblocking CG-UG-11 filter placed between the LED and the spectrometer fiber.

PAGE 180

0 1 2 3 4 250 260 270 280 290 300 Wavelength, nmIntensity, arb. units100 Hz 200 Hz 500 Hz Figure 7.10 Emission spectra of electrically modulated LED265 for different pulse repetition rates; 50% duty cycle. 159

PAGE 181

0 1 2 3 4 250 260 270 280 290 300 Wavelength, nmIntensity, arb. units100 Hz 200 Hz 500 Hz Figure 7.11 Emission spectra of CW LED265 wi th an external chopper for different chopper frequencies; 50% duty cycle. 160

PAGE 182

-2 -1 0 1 2 3 -0.015 -0.010 -0.005 0.000 0.005 0.010 0.015 Time, sPMT voltage, VPulsed Chopped Figure 7.12 Waveforms of LED265 emission in electrically pulsed and externally chopped regimes; 100 Hz rate. 161

PAGE 183

162 7.5.1 Unfiltered Emission Spectra The measurement of unfiltered (apart fr om an ND filter) emission spectra of the LED265, LED300, LED335 and LED355 are shown in Figs. 7.13, 7.15, 7.17 and 7.19, respectively. The spectra were normalized to compensate for the differences in the duty cycle, since the spectrometer in tegrated the sign al continuously. The LED335 had the same spectra for all three operation regimes. LED265 and LED355 performance was the same in CW and 5 ms 100 Hz modes (pulsed at 50% duty cycle), while 10 s 500 Hz mode (0.5% duty cycl e) appeared to have some reduction in the out-of-band emission. LED300 was the only one to exhibit variation in the main emission peak, with a wider longwavelength wing in the 5 ms 100 Hz mode, and narrower one in the 10 s 500 Hz regime. 7.5.2 Spectra with Visible-Blocking Filter Spectra of the same LEDs with the VI S-blocking filter are shown in Figs. 7.14, 7.16, 7.18 and 7.20. As can be seen, the out-of-band emission was effectively eliminated in all four cases. The loss in the peak emission, from comparison of the LED energy output data for filtered a nd unfiltered spectra, was 52% for LED265, 14% for LED300, 13% for LED335, and 34% for LED355. These results are consistent with the transmission curve of the CG-UG-11 filter. 7.6 LED Emission: Comparis on of Two Pulsed Regimes The functional dependence between the LED driving current and its output intensity was studied using a modified vers ion of the apparatus in Fig. 7.2. The

PAGE 184

0.001 0.01 0.1 1 10 200 300 400 500 Wavelength, nmNormalized intensity, arb. unitsCW @ 10 mA 10 s, 500 Hz @ 50 mA 5 ms, 100 Hz @ 10 mA Figure 7.13 Emission spectra of LED265 in CW, 5 ms pulsed and 10 s pulsed modes. 163

PAGE 185

0.001 0.01 0.1 1 10 200 300 400 500 Wavelength, nmNormalized intensity, arb. units5 ms, 100 Hz @ 10 mA CW @ 10 mA 10 s, 500 Hz @ 50 mA Figure 7.14 Emission spectra of LED265 with CG-UG-11 filter in CW, 5 ms pulsed and 10 s pulsed modes. 164

PAGE 186

0.001 0.01 0.1 1 10 200 300 400 500 Wavelength, nmNormalized intensity, arb. unitsCW @ 10 mA 10 s, 500 Hz @ 50 mA 5 ms, 100 Hz @ 10 mA Figure 7.15 Emission spectra of LED300 in CW, 5 ms pulsed and 10 s pulsed modes. 165

PAGE 187

0.001 0.01 0.1 1 10 200 300 400 500 Wavelength, nmNormalized intensity, arb. units5 ms, 100 Hz @ 10 mA CW @ 10 mA 10 s, 500 Hz @ 50 mA Figure 7.16 Emission spectra of LED300 with CG-UG-11 filter in CW, 5 ms pulsed and 10 s pulsed modes. 166

PAGE 188

0.001 0.01 0.1 1 10 200 300 400 500 600 Wavelength, nmNormalized intensity, arb. units5 ms, 100 Hz @ 10 mA CW @ 10 mA 10 s, 500 Hz @ 50 mA Figure 7.17 Emission spectra of LED335 in CW, 5 ms pulsed and 10 s pulsed modes. 167

PAGE 189

0.001 0.01 0.1 1 10 200 300 400 500 600 Wavelength, nmNormalized intensity, arb. units5 ms, 100 Hz @ 10 mA CW @ 10 mA 10 s, 500 Hz @ 50 mA Figure 7.18 Emission spectra of LED335 with CG-UG-11 filter in CW, 5 ms pulsed and 10 s pulsed modes. 168

PAGE 190

0.001 0.01 0.1 1 10 200 300 400 500 600 Wavelength, nmNormalized intensity, arb. unitsCW @ 10 mA 10 s, 500 Hz @ 50 mA 5 ms, 100 Hz @ 10 mA Figure 7.19 Emission spectra of LED355 in CW, 5 ms pulsed and 10 s pulsed modes. 169

PAGE 191

0.001 0.01 0.1 1 10 200 300 400 500 600 Wavelength, nmNormalized intensity, arb. units5 ms, 100 Hz @ 10 mA CW @ 10 mA 10 s, 500 Hz @ 50 mA Figure 7.20 Emission spectra of LED355 with CG-UG-11 filter (losses 34%) in CW, 5 ms pulsed and 10 s pulsed modes. 170

PAGE 192

171 monochromator was replaced with a combin ation of a VIS-blocking CG-UG-11 and ND filters, to record the total LED emission in the UV. The data was collected for LED currents up to 10 mA in the 5 ms 100 Hz mode, and 60 mA in the 10 s 500 Hz mode. Here, the nomenclature of 5 ms 100 Hz is used to indicate 100 Hz repetition rate with a 5 ms pulse length (a 50% duty cycle). The LED energy per pulse was measured using a LaserProbe Universal Radiometer (Model RM-3700) with a LaserP robe Silicon Energy Probe (Model RjP-465) averaged over 2000 pulses. The data was us ed to calculate the peak power within the LED pulse. 7.6.1 Output Intensity versus Driving Current The measured emission of the LED355 as a function of its drive current in 5 ms 100 Hz regime is plotted in Fig. 7.21. T ypical waveforms of modulated pulses across RCONTR and the measured LED emission as reco rded by a UV detector are presented in Fig. 7.22. As can be seen, the dependence of the LED emission on the LED current from 5 to 10 mA is well approximated by a linear fit, with a regression coefficient R2 = 0.995. For the 10 s 500 Hz mode, the LED355 emission as a function of current is shown in Fig. 7.23, and typical waveforms are de monstrated in Fig. 7.24. A linear fit to the data yielded an R2 of 0.997 for the LED currents from 10 to 60 mA. 7.6.2 Pulse Power Output The LED output energy per pulse at 10 mA in the 5 ms 100 Hz mode was measured for the LED265, LED300, LED335 and LED355, with and without a CG-UG

PAGE 193

0.01 0.02 0.03 0.04 0.0040.0050.0060.0070.0080.0090.010.011 Current, AIntensity, arb. units Figure 7.21 LED355 emission in 5 ms, 100 Hz pulsed mode as a function of LED current. 172

PAGE 194

-0.8 -0.4 0 0.4 0.8 -0.015-0.01-0.005 0 0.005 0.01 0.015 Time, sV contr, V-0.04 -0.03 -0.02 -0.01 0 0.01 Modulating voltage (V contr) LED emission (V det) V det, V Figure 7.22 Waveforms of modulating volta ge and LED355 emission in 5 ms, 100 Hz pulsed mode. 173

PAGE 195

0 0.05 0.1 0.15 0.2 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 Current, AIntensity, arb. units Figure 7.23 LED355 emission in 10 s, 500 Hz pulsed mode as a function of LED current. 174

PAGE 196

-4 -2 0 2 4 -0.00002-0.0000100.000010.000020.000030.000040.00005 Time, sV contr, V-0.16 -0.12 -0.08 -0.04 0 0.04V det, V Modulating voltage (V contr) LED emission (V det) Figure 7.24 Waveforms of modulati ng voltage and LED355 emission in 10 s, 500 Hz pulsed mode. 175

PAGE 197

176 11 filter (see Table 7.1). The relative measurement uncertainty was from 0.1% to 0.3%. The corresponding power for the unfilte red LEDs ranged from 0.079 mW for the LED265 to 0.46 mW for the LED355. In the 50 mA, 5 ms 100 Hz operating regime power output without a VIS-blocking filter varied from 0.71 mW for LED265 to 2.2 mW for LED355. The disproportional power output of the LED355 may have been due to its double-peak emission. Using these values, the normalized relative spectral output curves given in Fig. 7.4 were scaled to the output power and the area under each spectral curve. The resultant spectral output peak power curves are shown in Figs. 7.25 (for 10 mA, 5 ms, 100 Hz pulses) and 7.26 (for 50 mA, 10 s, 500 Hz pulses). As can be seen in Fig. 7.25 and 7.26, the peak power within the pulse for the 10 s pulse was about five times greater than that for the 5 ms pulse. However, the average output power over a 1 s interval was much less due to the differences in the allowed duty cycle, because of thermal limitations of the LED. For completeness, the LED power was calculated from the pulse energy measurements as in the following example. For the LED265 in 50 mA, 10 s, 500 Hz regime: P peak = E peak / t peak = 7.138 10-9J / 10 10-6s = 7.138 10-4W, P average = E peak / f PRF = 7.138 10-9J 500 Hz = 3.569 10-6W.

PAGE 198

0.00E+00 1.00E-05 2.00E-05 3.00E-05 250 300 350 400 Wavelength, nmSpectral power density, W/nmLED265 LED300 LED335 LED355 Figure 7.25 LED spectral peak power dens ity in 10 mA, 5 ms, 100 Hz pulse mode (lower curves are for the LED with CG-UG-11 filter). 177

PAGE 199

0.00E+00 5.00E-05 1.00E-04 1.50E-04 250 300 350 400 Wavelength, nmSpectral power density, W/nmLED265 LED300 LED335 LED355 Figure 7.26 LED spectral peak power density in 50 mA, 10 s, 500 Hz pulse mode (lower curves are for the LED with CG-UG-11 filter). 178

PAGE 200

179 CHAPTER EIGHT COMPARISON OF FLUORESCENCE EXCI TATION BY LED IN CW MODE WITH LASER INDUCED FLUORESCE NCE USING A LABORATORY BENCHTOP LED-IF/LIF SYSTEM This chapter describes the results of fl uorescence experiments with the UV LEDs operating in the CW mode as compared with using the UV laser for excitation using a laboratory bench-top LED-IF/LIF system. A compact spectrometer was used to collect and analyze most of the data. 8.1 Experimental Setup for Comparison of Fluorescence Excitation by CW LEDs with Laser Induced Fluorescence A schematic diagram of our laboratory bench-top LED-IF/LIF system used to compare the CW LED and laser-induced fluor escence is shown in Fig. 8.1. A microchip 266 nm laser or an UV LED equipped with a VIS-blocking CG-UG-11 filter were used interchangeably to illuminate a sample c ontaining quartz cell. Fluorescent emission passed through a UV-blocking filter to elimin ate the second-order peak of the scattered excitation wavelength, and was collected at 90 by an optical fiber connected to the compact spectrometer (Ocean Optics, Inc., USB2000). The spectrometer output was sent to the PC via an USB cable. A photogr aph of the laboratory bench-top LED-IF/LIF

PAGE 201

Figure 8.1 Schematic diagram of the CW laboratory bench-top LED-IF / LIF system. 180

PAGE 202

181 system is presented in Fig. 8.2. The transm ission of a BK-7 Schott glass used as a UVblocking filter is given in Fig. 8.3. Table 8.1 lists specifications for the spectrometer. 8.2 Comparison of LED and Laser Output To compare the optical output of the laser and LEDs in this experiment, the average power should be considered, because th e detector integrated the signal for a set period of time. The longest setting for the in tegration time, three seconds, was used with most signals. The UV laser produced 0.4 J pulses at the repetition rate of 7 kHz, which corresponds to an average power of 3 mW. For an integratio n time of 3 s, the total laser energy was 9 mJ. The CW LED power output at 10 mA driving current for LED265 and LED320 was about 0.08 mW. This corresponds to 0.24 mJ total LED energy over a 3 s integration period. Thus, the laser optical output during a 3 s integration time was approximately 40 times greater than that of the LEDs. 8.3 Measurements of LIF and LEDIF in Drinking and Natural Water Fluorescence of natural and drinking water samples was recorded using the system shown in Fig. 8.1 to compare the signa l-to-noise ratio with the laser and LED excitation, and to determine the feasibility of CW LED sources in this compact setup. Lake water with its strong C DOM fluorescence was tested first, and then examples of drinking water were studied as well.

PAGE 203

Figure 8.2 Photograph of the CW la boratory bench-top LED-IF system. 182

PAGE 204

Figure 8.3 Transmission spectrum of the BK -7 Schott glass (fro m CVI Melles Griot catalog). 183

PAGE 205

184 Table 8.1 Specifications for the USB-2000 spectrometer (Ocean Optics Inc.). Parameter USB2000 specifications Effective range 380 850 nm Grating spacing 600 lines/mm Blaze angle 500 nm Entrance slit 200 m Fiber diameter 600 m (VIS/NIR) Detector collection lens Yes Detector 2048 element linear silicon CCD array Estimated sensitivity 86 photons/count (for 1 s integration) Integration time 3 ms to 30 s Optical resolution 10 nm FWHM

PAGE 206

185 8.3.1 Lake Water A sample of USF lake water was placed in the quartz cell of the system in Fig. 8.1. It was illuminated by the 266 nm lase r, LED265 and LED320, and the average of 3 spectra, each integrated for 3 s, was recorded with each excitation source. Figure 8.4 shows the measured spectrum of the lake water fluorescence acquired with 266 nm laser excitation, and without the UV-blocking filter. In Fig. 8.5, the spectra obtained with LED265 and LED320 under the same conditions are shown. The peaks seen around 530 nm were identified as the se cond-order peaks of the Rayleigh scattered 265 266 nm emission line, and the peaks at about 580 nm were due to the second-order Raman scattering. To eliminate these bac kground contributions, a UV-blocking filter of BK-7 glass was used in subsequent experiments. In Figs. 8.6 and 8.7, the LIF and LED-Indu ced-Fluorescence (LED-IF) spectra of lake water taken with the UV-blocking filte r, but otherwise under the same conditions, are plotted. As expected, the second -order scattering peaks were gone. The raw spectra were corrected for fibe r transmission, grating efficiency and detector sensitivity (Appendix C), as well as sm oothed over nine data points. The results for LIF and LED-IF are plotted in Fig. 8.8 a nd Fig. 8.9, respectively. However, even after the correction for the fiber, grating a nd detector spectral characteristics was applied, the CDOM peak appeared to have a maxi mum at the longer wavelength (around 500 nm) compared to the 450 nm measured with the portable LIF system and reported in other sources. This discrepancy might be due to inaccuracies in the provided intensity calibration curves for the spectrometer, or seasonal variation in water samples, and changes in water composition, as suggested by data in Fig. 3.3 for different water types.

PAGE 207

0 500 1000 1500 2000 350 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units Figure 8.4 Lake water fluorescence; 266 nm laser excitation (Sept. 14, 2006). 186

PAGE 208

0 50 100 150 200 350 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units LED265 LED320 Figure 8.5 Lake water fluorescence; 265 nm and 320 nm LED excitation (Sept. 14, 2006). 187

PAGE 209

0 1000 2000 3000 350 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units Figure 8.6 Lake water fluorescence; 266 nm la ser excitation; UV-blocking filter (Sept. 23, 2006). 188

PAGE 210

0 100 200 350 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units LED265 LED320 Figure 8.7 Lake water fluorescence; 265 nm and 320 nm LED excitation; UV-blocking filter (Sept. 23, 2006). 189

PAGE 211

0 1000 2000 3000 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units Figure 8.8 Lake water fluorescence; 266 nm laser excitation; UV-blocking filter and correction/smoothing (Sept. 23, 2006). 190

PAGE 212

0 100 200 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units LED265 LED320 Figure 8.9 Lake water fluorescence; 265 nm and 320 nm LED excitation; UV-blocking filter and correction/smoothing (Sept. 23, 2006). 191

PAGE 213

192 8.3.2 Tap Water Fluorescence of tap water from our USF la boratory was integrated for 3 s, and the average of 3 spectra was recorded. The laser-induced fluorescen ce is shown in Fig. 8.10, and the LED-induced one in Fig. 8.11. The spectra were corrected for the fiber, grating and detector spectral characteristics, and smoothe d over 9 points. As can be seen in the data, the LIF peak is well defined, while the LED-IF signals are approaching the background noise level. 8.3.3 Distilled Water LIF and LED-IF spectra of Zephyrhills br and distilled water were obtained and processed in the same way as the lake and the tap water samples. As can be seen from the laser and LED induced fluorescence spec tra (Fig. 8.12 and Fig. 8.13 respectively), the CDOM peak is indistinguishable from the background noise. The signal-to-noise ratio (SNR) for the lake and tap water samples was found as the difference between the peak fluorescence signal (around 500 nm ) and the background (distilled water at the same wavelength) over the double standard deviation at the offpeak range (approximately 20 nm bandwidth in the region of a flat signal baseline). SNR for the lake water with LIF was 242, and 36 and 19 for the LED-IF using the LED265 and LED320, respectively. For a tap wa ter sample, it was 180 with LIF, and 7 and 5 for the LED-IF using the LED265 and LED320, respectively. Thus, for the tap water sample, the ratio between the laser and LED signal-tonoise ratio was 26 36, close to th e ratio of their power outputs (40). For the lake water

PAGE 214

0 200 400 600 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units Figure 8.10 Tap water fluorescence; 266 nm la ser excitation; UV-blocking filter and correction/smoothing (Sept. 23, 2006). 193

PAGE 215

0 50 100 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units LED265 LED320 Figure 8.11 Tap water fluorescence; 265 nm and 320 nm LED excitation; UV-blocking filter and correction smoothing (Sept. 23, 2006). 194

PAGE 216

0 200 400 600 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units Figure 8.12 Distilled water fluorescence; 266 nm laser excitation; UV-blocking filter and correction/smoothing. 195

PAGE 217

0 50 100 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units LED265 LED320 Figure 8.13 Distilled water fluorescence; 265 nm and 320 nm LED excitation; UVblocking filter and correction/smoothing. 196

PAGE 218

197 sample, the laser SNR was 7 13 times that of the LEDs, i.e. LEDs had a SNR about 3 to 5 greater than that expect ed for their power levels. 8.4 Detection of Chemicals Using LIF and LED-IF Several chemical and biological substances were tested for fl uorescence detection with the system. Tonic water with quinin e (similar to a fluorescence standard quinine sulfate) was diluted in distilled water at 1 mL /L. Content of quini ne in the tonic water was 5% according to the manufacturer's label. A fluorescent dye, Coumarin 540A, was also used at 1 mL/L. Endospores of bacteria Bacillus Subtil is were used at a rather considerable concentration of 3.5 g/L. Terb ium and dipicolinic acid (DPA) in distilled water were at 50 M/L each. 8.4.1 Tonic Water with Quinine The fluorescence spectra of quinine in tonic water were measured for 266 nm laser, and 265 nm and 320 nm LED excita tion, as shown in Fig. 8.14. The signal integration time was 0.2 s. The signal wa s corrected for the system parameters (the fiber, grating and detector spectral characteristics) and smoothed over 9 points. As can be seen, the LED-IF signal due to LED320 ex citation was very str ong, about two-thirds of the LIF peak, while the one due to the LED265 was around one-f ifth of the laserinduced signal. Considering that the smallest SNR was 170 (LED265 excitation), the system could have easily detected a signal fr om a much less concentrated solution, such as 10 L/L.

PAGE 219

0 1000 2000 3000 4000 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units Laser266 LED265 LED320 Figure 8.14 Fluorescence of tonic with quini ne in distilled water; 265 nm and 320 nm excitation; UV-blocking filter and correction/smoothing. 198

PAGE 220

199 8.4.2 Coumarin Laser Dye The spectra of laser dye Coumarin 540A are plotted in Fig. 8.15. Integration time was set to 3 s, and averaging over 3 spectra was used. The peak fluorescence wavelength for 265 nm excitation was observed around 550 nm, and for 320 nm excitation it was at about 500 nm. As can be seen, the LIF signal was approximately 10 times greater than the LED-IF signal. 8.4.3 Spores of Bacillus Subtilis Figure 8.16 shows the natural fluorescence of bacterial endospores with LIF and LED-IF excitation (uncorrected). The signal was integrated for 3 s, and 3 spectra were averaged. As seen, the LIF signal was approximately 10 times stronger than the LED265 fluorescence, and exceeded the LED320 fluorescence by about 5 times. 8.4.4 Terbium-Doped Dipicolinic Acid In Fig. 8.17, the fluorescence spectra of Tb-DPA complex are plotted for the 266 nm laser excitation and LED265 excitation with and without the CG-UG-11 filter. This dataset was collected using a monochromator (CVI Digikrom 240) with a PMT detector (Hamamatsu, Model H6780-03) instead of a USB2000 spectrometer, utilizing another LIF system that was available in our laboratory which used a more sensitive detector. Integration time was 0.5 s, averaging 10 spectra Four distinct peaks were observed at 490 nm, 545 nm, 585 nm and 620 nm. The lase r-induced fluorescence was about 8 times stronger than the unfiltered LED-IF signal, and about 20 times stronger than the peaks due to the LED filter combination.

PAGE 221

0 400 800 1200 350 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units0 50 100 150 Laser266 (scale on left) LED265 (scale on right) LED320 (scale on right) Figure 8.15 Fluorescence of Coumarin 540A dye in distilled water; 265 nm and 320 nm excitation. 200

PAGE 222

0 1000 2000 3000 350 400 450 500 550 600 Wavelength, nmFluorescence intensity, arb. units0 500 1000 Laser266 (scale on left) LED265 (scale on right) LED320 (scale on right) Figure 8.16 Fluorescence of B. Subtilis bacterial endospores in distilled water; 265 nm and 320 nm excitation. 201

PAGE 223

0 1000 2000 3000 4000 450 500 550 600 650 700 Wavelength, nmFluorescene Intensity, arb. units0 100 200 300 400 500 600 Laser266 (scale on left) LED265 (scale on right) LED265 CG-UG-11 filter (scale on right) Figure 8.17 Fluorescence of Tb-DPA in distilled water; 265 nm excitation. 202

PAGE 224

203 CHAPTER NINE COMPARISON OF FLUORESCENCE EX CITATION BY LED IN PULSED MODE WITH LASER INDUCED FLUO RESCENCE USING THE PORTABLE LIF SYSTEM In this chapter, the results of fluorescence measurements with pulsed LED excitation are presented and compared to the LIF data. The portable LIF system with modifications to accommodate a LED s ource was used in the experiments. 9.1 Experimental Setup for Comparison of Fluorescence Excitation by Pulsed LEDs with Laser Induced Fluorescence Figures 9.1 and 9.2 show a schematic diagram and a photograph of the pulsed LIF/LED-IF system. The absorption detector s in the portable LIF system were removed to incorporate the LED source in the system, due to the space restricti ons inside the optics box. The LED and a front-surface mirror were placed immediately before and after the sample cell, along the laser beam path, to maximize the fluorescence excitation. Both the LED and the mirror were mounted in such a way that these objec ts could be pivoted out of the path of the laser beam to allow LIF measurements. The mirror improved fluorescence excitation by 25 30%.

PAGE 225

Figure 9.1 Schematic diagram of the pulsed LIF/LEDIF system. 204

PAGE 226

Figure 9.2 Photograph of th e pulsed LIF/LED-IF system. 205

PAGE 227

206 9.2 Comparison of LED and Laser Output Since the pulsed regimes of the laser a nd LED sources were of different duration and repetition frequency, the comparison of their output wa s based on the energy of the pulse gated by the gated integrator and boxcar averager in each case. The 266 nm and 355 nm lasers operated at 7 kHz and 12 kHz re spectively, with 0.4 J (0.3 J for 355 nm laser) in a 0.4 ns pulse. Th e gate was set to 100 ns to cover the entire fluorescence peak. All LEDs were operated in the 10 s, 500 Hz mode at 10 mA pulse current. The gate was set to the maximum possible value of 15 s. Their measured pulse energy in this mode of operation was 0.76 nJ for LED265, 1.1 nJ for LED300, 0.99 nJ for LED335 and 4.7 nJ for LED355. This means that the lasers produced approximately 525 (LED265) to 80 (LED355) times more energy per pulse than the LEDs. For both types of excitation, 1000 pulses were averaged to record one data point. Sensitivity was set to 200 mV. 9.3 Measurements of LIF and LED-IF in Lake Water The fluorescence of USF lake water with laser and LED excitation of 266 (265) nm was measured and plotted in Fig. 9.3. The peak LIF signal was approximately 70 times stronger than that of the LED-IF, which means that the LED induced fluorescence had a signal 7.5 times greater than could be expected from the LED output energy. The correlation coefficient for the la ser and LED fluorescence was 0.94. In Fig. 9.4, the fluorescence spectra of the laser and LED excita tion at 355 nm are shown. The laser-induced fluorescence peak was about 20 times greater than the LED-

PAGE 228

0 200 400 600 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. units0 4 8 12 Laser266 LED265 Figure 9.3 Fluorescence of lake water; pulsed 265 nm excitation (March 12, 2008). 207

PAGE 229

0 200 400 600 350400450500550600650700 Wavelength, nmFluorescence intensity, arb. units0 10 20 30 40 Laser355 LED355 Figure 9.4 Fluorescence of lake water; pulsed 355 nm excitation (March 12, 2008). 208

PAGE 230

209 induced peak signal, that is, the LED-IF signal was 4 times stronger than could be expected from the LED output energy. Co rrelation coefficient fo r the entire 355 685 nm range was only 0.10 because of the LED355 secondary peak at 370 nm. For the fluorescence region of 420-685 nm, the correlation coefficient was 0.90. The fluorescence spectra excited with a ll four LEDs are given in Fig. 9.5. Because the cut-off (Fig. 9.6) and bandpass filt ers in the LIF system predated the LEDs, it was not possible to use an optimum opti cal filter combination for LED300 and LED335 to efficiently block Rayleigh and Rama n scattering while allowing the desired fluorescence signal to reach the de tector without attenuation. The new LED driver with current-contr olled pulses of 50 mA and corresponding fivefold increase in the energy output is expe cted to produce an improvement in the LED induced fluorescence signal of about 5 times. Among the possible reasons why the LED-IF signal is stronger than might be expected from the LED/laser power output ratio are: (1) the LED emission might have a better overlap with CDOM absorption spectra due to the wider bandwidth of the LEDs; (2) not all of the laser-illuminated volume is in the field-of view of the collection optics for particular parts of the multi-pass system; and (3) short-pulsed, narrow laser beam might be causing saturation in the central part of the beam's cross-section, although previous measurements (Sivaprakasam, 2002) have shown that photobleaching did not occur if the sample was continuous ly pumped through the LIF system.

PAGE 231

0 5 10 15 20 250300350400450500550600650700 Wavelength, nmFluorescence intensity, arb. unitsLED265 LED300 LED335 LED355 Figure 9.5 LED-IF of lake water; pulsed excitation at 265, 300, 335 and 355 nm (March 12, 2008). 210

PAGE 232

0% 20% 40% 60% 80% 100% 250300350400450500550600650700 Wavelength, nmTransmissionFilter 1 (280 nm) Filter 2 (305 nm) Filter 3 (365 nm) Filter 4 (320 nm) Filter 5 (420 nm) Figure 9.6 Transmission curves of cutoff filters in LIF / LED-IF system. 211

PAGE 233

212 CHAPTER TEN CONCLUSIONS AND FUTURE WORK Fluorescence measurements of drinking and natural water sources were made using laser excitation at 266 and 355 nm. Thes e studies demonstrated that these spectra can be used to discriminate between various water samples, incl uding those containing trace amounts of organic contaminants. In particular, differences between the samples before and after reverse osmosi s treatment of ground water were clearly seen. Similarly, tap water samples with and without on-tap filter treatment were differentiated. Interestingly, it was shown that drinking wate r from international sources had essentially the same spectra and levels of fluorescence. Initial detection of toxic chemicals and biological contaminants was also demonstrated. The portable LIF system was used to m onitor running tap water continuously for a week. Observed variations in the peak fluorescence signal exceeded both the measurement uncertainty and ch anges in a control sample of continuously recirculated tap water. It was surmised that periodic variations in the running water signal were related to the daily usage pattern of water on campus. Studies of pure and chlorinated tannic acid were also performe d with the intent to observe chemical changes in the solution as a result of interactions between chlorine and organic substances, but results were not conclusive.

PAGE 234

213 To study the feasibility of using much cheaper, smaller and less energyconsuming UV LEDs as an alternative exci tation source, a multifunctional LED driver was developed and extensively tested. Vari ous electrical and optical characteristics of several LEDs emitting in the 265 355 nm ra nge were studied as well, including V-I curves, optical spectra under different ope rational conditions, and optical power output. Comparison between the laser and CW LED (10 mA) excitation was performed using a compact spectrometer. For the 266 (265) nm excitation of the tap water fluorescence, the signal-to-noise ratio was 180 with LIF and 7 with LED-IF. The ratio between the laser and LED signalto-noise ratio was about the same as the ratio of their power outputs. Therefore, a compact CW LED-IF system should be viable for the detection of the trace levels of c ontaminants typical for tap water. Pulsed (10 mA, 10 s) LED operation was studied usi ng the portable LIF system. The pulse length was limited by th e maximum gate setting of 15 s. For the 266 nm excitation, the peak LIF signal was approxima tely 70 times stronger than that of the LED-IF, while the laser energy per pulse was approximately 525 times more than the LED energy. For 355 nm excitation, the la ser-induced fluorescence peak was about 20 times greater than the LED-induced peak si gnal, while the laser energy per pulse was 80 times more than the LED energy. This means that the LED induced fluorescence had a signal 4 7 times greater than should have b een expected from the LED output energy. This increase may be due to a better overlap of the LED emission with CDOM absorption spectra, the saturation in the center of the short-pulsed, narrow laser beam, or the possibility that not all of the laser-illumina ted volume is in the field-of view of the collection optics for parts of the multi-pass system.

PAGE 235

214 Future research includes pulsed LED-IF m easurements with the new LED driver, comparison of CW laser and LED excitati on using a lock-in am plifier for signal integration, and improvements to the data acquisition software. Finally, it is important that a comparison be made of our LIF results for drinking water with that of more conventional standa rd water analysis tec hniques, such as TOC measurements or HPLC analysis. Such an in-depth study would help quantify our new LIF water analysis approach, and might explain temporal ch anges in the drinking water fluorescence during long-term monitoring.

PAGE 236

215 REFERENCES Albani, J. R. (2007). Principles and applications of fluorescence spectroscopy Oxford: Blackwell Publishing. American Public Health Association, Am erican Water Works Association, & Water Pollution Control Federation. (1989). Standard methods for the examination of water and wastewater (17th ed.). Washington: Authors. Balshaw-Biddle, K., Oubre, C. L., & Ward, C. H. (Eds.). (2000). Subsurface contamination monitoring using laser fluorescence Boca Raton: CRC Press, Inc. Born, M., & Wolf, E. (1999). Principles of optics: El ectromagnetic theory of propagation, interference, and diffraction of light Cambridge: Cambridge University Press. Brand, L., & Johnson, M. L. (Eds.). (1997). Fluorescence spectroscopy. New York: Academic Press. Beltran, J. L., Guiteras, J., & Ferrer, R. (1998). Three-way multivariate calibration procedures applied to high-performance liq uid chromatography coupled with fastscanning fluorescence spectrometry detection. Determination of polycyclic aromatic hydrocarbons in water samples. Analytical Chemistry, 70 (9), 1949-1955 Bunkin, A., & Voliak, K. (2001). Laser remote sensing of the ocean: Methods and applications New York: John Wiley & Sons, Inc. Christian, G. D., & Callis, J. B. (Eds.). (1986). Trace analysis: Spectro scopic methods for molecules. New York: John Wiley & Sons, Inc. Coble, P. G. (1996). Characterization of mari ne and terrestrial DOM in seawater using excitation-emission matrix spectroscopy. Marine Chemistry 51, 325-346. Coble, P. G. (2007). Marine optical biogeoc hemistry: the chemistry of ocean color. Chemical Review 107, 402-418. Crompton, T. R.(2000). Determination of organic compounds in natural and treated waters. London: E & FN Spon.

PAGE 237

216 Dewey, T. G. (Ed.). (1991). Biophysical and biochemical aspects of fluorescence spectroscopy New York: Plenum Press. Donn, J., Mendoza, M., & Pritchard, J. (Apr il 19, 2009). AP IMPACT: Tons of released drugs taint US water. Associated Press Fetzer, J. (2000). Large (C>=24) polycyclic aromatic hydrocarbons: Chemistry and analysis New York: John Wiley & Sons, Inc. GE Osmonics. (2003). E-series general i ndustrial equipment USA: Author. GE Osmonics. (2003). SEPA CF II membrane cell system USA: Author. Horowitz, P., & Hill, W. (1989). The Art of Electronics (2nd ed.). Cambridge: Cambridge University Press. Hudson, N. J., Baker, A., & Reynolds, D. ( 2007). Fluorescence analysis of dissolved organic matter in natural, waste and polluted waters a review. Rivers Research 23, 631-649. DOI: 10.1002/rra.1005. Hunt, D. T. E., & Wilson, A. L. (1986). The chemical analysis of water (2nd ed.). Cambridge: The Royal Society of Chemistry. Iwata, T., Kamada, T., & Araki, T. (2000) Phase-modulation fluorometer using an ultraviolet light-emitting diode. Optical Review 7 (6), 495-498. Killinger, D. K., & Sivaprakasam, V. (2006). How water glows: Water monitoring with laser fluorescence. Optics and Photonics News 17 (1), 34-39. Krasovitskii, B. M., & Bolotin, B. M. (1988). Organic luminescent materials Germany: VCH. Lakowicz, J. R. (2006). Principles of fluorescence spectroscopy (3rd ed.). New York: Springer. Makoui, A., & Killinger, D. K. (2009). Fluor escence lifetime and in tensity of terbiumdoped dipicolinic acid in water, HCl, and sodium acetate buffer solutions. Applied Optics, 48 (4), B111-B118. Makoui, A., & Killinger, D. K. (2009). Tran sient fluorescence spectroscopy of terbium doped dipicolinic acid: a fluorescen ce lifetime measurement technique. Journal of the Optical Society of America B 26 (4), 691-698. Mendoza, M. (March 25, 2009). Study: Range of pharmaceuticals in fish across US. Associated Press.

PAGE 238

217 Muttiah, R. S. (Ed.). (2002). From laboratory spectroscopy to remotely sensed spectra of terrestrial ecosystems Netherlands: Kluwer Academic Publishers. Nollet, L. M. L. (Ed.). (2000). Handbook of water analysis New York: Marcel Dekker, Inc. Philips Electronics. (2004). PSSI2021SAY Constant Current Source in SOT353 Package Product Data Sheet Netherlands: Author. Pope, R. M., & Fry, E. S. (1997). Absorption spectrum (380 nm) of pure water. II. Integrating cavity measurements. Applied Optics 36 (33), 8710-8723. Ryder, A. G. (2004). Time-resolved fluorescen ce spectroscopic study of crude petroleum oils: Influence of chemical composition. Applied Spectroscopy 58, 613-623. Sensor Electronic Technology, Inc. (2004). UVTOP-365 UV LED in TO-39 Package Product Data Sheet USA: Author. Sharikova, A. V., & Killinger, D. K. (2007). LIF detection of trace species in water using different UV laser wavelengths. International Journal of High Speed Electronics and Systems 17 (4), 689-695. Sharma, A., & Schulman, S. G. (1999). Introduction to fluorescence spectroscopy New York: John Wiley & Sons, Inc. Sivaprakasam, V. (2002). UV laser induced fluorescence sp ectroscopic studies and trace detection of dissolved plastics (bis phenol-A) and organic compounds in water Doctoral Dissertation. Univer sity of South Florida. Sivaprakasam, V., & Killinger, D. K. (2003). Ef fect of polarization and geometric factors on quantitative laser-induced fluorescenceto-Raman intensity ratios of water samples and a new calibration technique. Journal of the Optical Society of America B 20 (9), 1980-1989. Sivaprakasam, V., & Killinger, D. K. ( 2003). Tunable ultraviolet laser-induced fluorescence detection of trace plastic s and dissolved organic compounds in water. Applied Optics 42 (33), 6739-6746. Sivaprakasam, V., Shannon, R. F., Luo C., Coble, P. G., Boehme, J. R., & Killinger, D. K. (2003). Development and initial calibration of a portable laser-induced fluorescence system used for in situ measurements of trace plastics and dissolved organic compounds in seawater and the Gulf of Mexico. Applied Optics 42 (33), 6747-6756. Vandenberg, L. N., Hauser, R., Marcus, M., Olea, N., & Welshons, W. V. (2007). Human exposure to bisphenol A (BPA). Reproductive Toxicology 24, 139.

PAGE 239

218 Vo-Dinh, T. (Ed.). (1989). Chemical analysis of polycyclic aromatic compounds New York: John Wiley & Sons, Inc. Yoshioka, T., Mostofa, K. M. G., Konohira E., Tanoue, E., Hayakawa, K., Takahashi, M., Ueda, S., Katsuyama, M., Khodzher, T., Bashenkhaeva, N., Korovyakova, I., Sorokovikova, L., & Gorbunova, L. (2007). Distribution and characteristics of molecular size fractions of freshwater-d issolved organic matter in watershed environments: its implication to degradation. Limnology 8, 29-44.

PAGE 240

219 APPENDICES

PAGE 241

220 APPENDIX A. CALCULATIONS FOR LED DRIVER DESIGN This Appendix contains the details of th e multifunctional LED driver (Fig. A.1) design and calculations. Base resistance to keep diodes in conduction: RB = (VPS ND VD)/IB, where VPS = 15 V, the number of diodes ND is 4, VD = 0.6 V, and IB = 2 mA. Limits for the variable emitter resist ance based on the load current range: RE = (ND VD VEB)/ILOAD = [(ND 2) VD]/ILOAD, where VEB = 1.2 V. Variable emitter resistance details (consists of R3, R4, and RV2). Solving a system of two linear equations: 1/R3 +1/R4 = 1/RMIN 1/(R4+RV2) + 1/R3 = 1/RMAX, One gets R3 = RV2 { [1 + 4RMAXRMIN/RV2(RMAX RMIN)]1/2 1}/2, and R4 = (1/RMIN 1/R4). For RMIN = 15 RMAX = 130 and RV2 = 1000 one finds R3 = 17 and R4 = 150

PAGE 242

APPENDIX A (Continued) Figure A.1 Schematic diagram of the multifunctional LED. 221

PAGE 243

222 APPENDIX A (Continued) Time constant reduction: = RTOT CLED, where RTOT = [1/(RLED + RCONTR) + 1/RPAR]-1. Without RPAR, is 0.5 ms; with RPAR of 4.7 k is 0.5 s. It follows that RLED is of the order of 4 5 M and CLED is about 0.1 nF. Maximum possible load current ILOAD (load includes LED, RCONTR and RPAR): ILOAD MAX = [VPS VD (ND 1)]/RLOAD.

PAGE 244

APPENDIX B. DATA COLLECTION AND PROCESSING SOFTWARE This appendix contains data collection software in LabVIEW and data processing MATLAB script. Examples of VIs written in LabVIEW for the LIF/LED-IF system data collection are shown below: Figure B.1 Front panel of the LED-IF data collection LabVIEW program. Figure B.2 Diagram of the LED-IF data collection LabVIEW program. 223

PAGE 245

224 APPENDIX B (Continued) Figure B.3 Front panel of the Save_Data.vi. Figure B.4 Diagram of the Save_Data.vi.

PAGE 246

225 MATLAB scripts written for the data processing are shown below: 3D plot script % 3D plots of sample fluorescence vs wavelength vs time clear; % get user input about the data file and prepare relevant variables % open window to select directory filepath = uigetdir( '\\file1\home\ashariko\Research\Portable LIF Data 2006', 'Select directory with (first) data file' ); % open window to select wavelength and # days prompt = { 'Plot multiple days? (y for yes)' 'Excitation wavelength (355/266)' }; dlg_title = 'Specify file parameters:' ; '2 options.Resize= 'on'; nswer = inputdlg(prompt,dlg_title,num_lines,def); %# consecutive days to plot waveindex = answer{2}; %excitation wavelength if (strcmp(waveindex, '266')) %use default 266 nm file excwave = 266; %wavelength colskip = 12; %# columns before wavelength/fluorescence colwave = 19; %# wavelength/fluorescence columns fluread = '%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s %f%f%f%f%f%f%f%f%f%f%f%f%f%f%f%f%f%f%f %*[^\n]' ; % fluor reading format elseif (strcmp(waveindex, '355'))%read 355 nm file excwave = 355; %wavelength colskip = 11; %# columns before wavelength/fluorescence colwave = 14; %# wavelength/fluorescence columns fluread = '%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s %f%f%f%f%f%f%f%f%f%f%f%f%f%f %*[^\n]' ; % fluor reading format else msgbox( 'Invalid wavelength' ); end; \d*' 'match' ); %extract date info msgbox( 'Invalid file path' ); end; date = date{1,1}; %extract regular string from cell filename = sprintf( 'avg_%s_%dnm.txt' date, excwave); %form file name pathname = strcat(filepath, '\', filename); %form full path to the file date = regexprep(date, '_', '); %replace underscore with space APPENDIX B (Continued) num_lines = 1; def = { '', 66'}; a multdays = answer{1}; date = regexp(filepath, '\w\w\w_\d\d_ if isempty(date)

PAGE 247

226% open fi checkname = isdir(pathname); datafile = fopen(pathname, 'r'); %open for reading textscan(datafile, '%*s', colskip, 'Delimiter' '\t'); %skip starting columns wave = textscan(datafile, '%f' elimiter' '\t'); %read wavelengths (datafile, '', 1, 'HeaderLines' ,1); %skip 1st line ad miter' '\t', %extract regular matrix fluorescence from cell sts rm path path ened files (debugging) h directory data 'r'); %open next file for reading kip 1st line comment = [comment; temp{1,2}]; %add new comments frewind(datafile); %return to beginning of the file textscan(datafile, '', 1, 'HeaderLines' ,1); %skip 1st line APPENDIX B (Continued) le and read data colwave, 'D wave = wave{1,1}; %extract regular array from cell datafile); %return to beginning of the file frewind( tscan tex temp = textscan(datafile, '%*s %f %s %*[^\n]' 'Delimiter' '\t'); %re e & comments tim time = temp{1,1}; %assign time time = time-time(1,1); %start time count at zero comment = temp{1,2}; %assign comments clear temp; %delete temporary variable frewind(datafile); %return to beginning of the file textscan(datafile, '', 1, 'HeaderLines' ,1); %skip 1st line fluor = textscan(datafile, fluread, 'Deli 1); %read fluorescence 'CollectOutput' fluor = fluor{1,1}; format fclose(datafile); % handle multiple files files should be merged if ~isempty(multdays) %multiple day = sscanf(date, '%*s %d' ); %read day as a number merged % for k = 1:(multdays-1) %repeat for all days to be xi while isdir(filepath) %check if this directory e day = day+1; %increment day %convert to string if day < 10 newday = sprintf( '_0%d_' day); %d_' day); else newday = sprintf( '_ end; y); %fo pathname = regexprep(pathname, '_\d\d_' newda for next file filepath = regexprep(filepath, '_\d\d_' newday); %form xt file for ne % msgbox(filepath); %list op if ~isdir(filepath) %no suc break %exit loop end; % open next file and read datafile = fopen(pathname, textscan(datafile, '', 1, 'HeaderLines' ,1); %s temp = textscan(datafile, '%*s %f %s %*[^\n]' 'Delimiter' '\t'); %read next time & comments ce temp{1,1} = temp{1,1} + time(end,1); %continue time sequen time = [time; temp{1,1}]; %add new time

PAGE 248

227ale, fluread, 'Delimiter' '\t', fluorescence canf(date, '%*s %d %d' ); %read day and year day, temp(2,1)); ip # last wavelengths' 'Skip # for automatic)' }; ble(answer{1}); %# first wavelength channels to skip e; %starting wavelength channel r{2}); %# last wavelength channels to skip ave; %ending wavelength channel ski le tle = sprintf( 'LIF of %s, %d nm excitation, velength on the left, x),fluor(wavmin:wavmax,timin:tiAPPENDIX B (Continued) temp = textscan(dat fi 'CollectOutput' ,1); %read next fluor = [fluor; temp{1,1}]; %add new fluorescence fclose(datafile); end; month = sscanf(date, '%s', 1); %read month temp = ss date = sprintf( '%s %d %d %d' month, temp(1,1), %insert last day for graph title clear temp; %delete temporary variable end; % get user input about settings and plot 3D graph prompt = { 'Skip # first wavelengths' 'Sk first spectra' 'Skip # last spectra' 'Graph title (a dlg_title = 'Specify plot parameters:' ; num_lines = 1; def = { '2', '0', '0', '0', ''}; options.Resize= 'on'; answer = inputdlg(prompt,dlg_title,num_lines,def); startwave = str2dou min = 1+startwav wav endwave = str2double(answe wavmax = size(wave,1)-endw starttime = str2double(answer{3}); %# first time channels to skip timin = 1+starttime; %starting time channel endtime = str2double(answer{4}); %# last time channels to p timax = size(time,1)-endtime; %ending time channel plottitle = answer{5}; %user title if ~isempty(plottitle) % nonempty tit if plottitle == 'a' %assign name automacally, 1st plotted comment ti & excitation wavelength plotti comment{1+starttime,1}, excwave); end; plottitle = [plottitle date]; %add date to the title end; fluor = fluor'; %transpose fluo ght rescence data: wa time on the ri mesh(time(timin:timax),wave(wavmin:wavma max)); %plot a 3D mesh graph set(gca, 'YDir' 'reverse' ); %reverse wavelength axis for better view shading interp ; %various graph setting colormap(jet); %alternatives: gray,hsv,cool,hot,copper,winter,autumn,white title(plottitle); xlabel( 'Time, hours' ); ylabel( 'Wavelength, nm' ); zlabel( 'Fluorescence intensity, arb. units' );

PAGE 249

228ocessing?' 'Yes', 'No'); % import? perform spectrum correction ; correction rve usingialog box ng correction curve' 'OK'); % friendly correction curve od' rve?' 'Yes', 'No'); % normalize? 1 % yes rmalize tion to one ;APPENDIX B (Continued) Interpolation of correction cu rve for spectrometer data points % Interpolate a curve for the ( 'Import spectrum for pr spectrometer data points imp = menu if imp == 1 % yes %clear spectrum; % legacy of clear all % import spectrum for processing using dialog box menu( 'Select file containing spectrum for correction' 'OK'); % friendly reminder = uiimport( '-file' ); % select file to s on spectrum = s.data; % assign spectrum for correction clear s ; % clear import structure end imp = menu( 'Import correction curve?' 'Yes', 'No'); % import? if imp == 1 % yes %clear curve; % legacy of clear all d % import cu menu( 'Select filcontaini e reminder s = uiimport( '-file' ); % select file containing curve = s.data; % assign correction curve clear s ; % clear import structure end; % correction curve inter/extrapolation: spectrum(:,3) int_method = { 'nearest' 'linear' 'spline''cubic' }; % cell array listing interpolation methods choose_method = menu( 'Select correction curve inteolation meth rp 'nearest' 'linear' 'spline' 'cubic' ); spectrum(:,3) = interp1(curve(:,1),curve(:,2),spectrum(:,1),int_method{choose_method}); plot(curve(:,1), curve(:,2), spectrum(:,1),spectrum(:,3)); % correction curve % corrected spectrum: spectrum(:,4) imp = menu( 'Normalize correction cu if imp == spectrum(:,3) = spectrum(:,3)/max(spectrum(:,3)); % no correc end spectrum(:,4) = spectrum(:,2)./spectrum(:,3); % correct the spectrum figure; %new figure window plot(spectrum(:,1), spectrum(:,2), spectrum(:,1), spectrum(:,4)); % original and corrected spectra

PAGE 250

229 APPENDIX C. SPECTROMETER CORRECTION CURVES This a SB2000 e ppendix contains plots of the Ocean Optics, Inc. ST2000 and U correction curves. The detector, grating and fiber spectral curves were ob tained from th manufacturer's website (www.oceanoptics.com). ST2000 spectrometer Typical spectral sensitivity of a Sony ILX511A detector 1 No data available below 400 nm 0.9 0.8 0.7 0 250 300 350 400 450 500 550 600 650 700 Wavelength, nm 0.1 0.2 0.3 0.4Spec0.5tral sens0.6itivity at 25C 00 spectrometer detector sensiti Figure C.1 ST20 vity. Efficiency of grating #2: 600 lines/mm, 200-800 nm 80 70 40ng effi50 60ciency, %20 30Relative grati0 10 250 300 350 400 450 500 550 600 650 700 Wavelength, nm Figure C.2 ST2000 spectrometer grating efficiency.

PAGE 251

APPENDIX C (Continued) Known and calculated transmission of a 2 m long UV/VIS fiber used with ST200070 90 0 10 20 30 40 50 60 80 250 300 350 400 450 500 550 600 650 700 Wavelength, nmTransmission, %Transmission calculated using T = 10^(-aL), % Transmission (specs), % Figure C.3 ST2000 spectrometer fiber transmission. Normalized correction for ST2000 spectrometer0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 400450500550600650700750800850900 Wavelength, nmNormalized correctionCorrection assumed Sony ILX511A detector, 600 lines/mm 200-800 nm grating #2, and UV/VIS solarization resistant 2 m fiber with known transmission Detector extrapolation no data below 400 nm Figure C.4 ST2000 spectrometer overall correction. 230

PAGE 252

APPENDIX C (Continued) 231 USB2000 spectrometer Typical spectral sensitivity of a Sony ILX511B detector 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 3 50 400 450 500 550 Wavelength, nm 600 650 700Spectral sensitivity at 25CNo data available below 400 nm igure C.5 USB2000 spectrometer detector sensitivity. F Efficiency of grating #3: 600 lines/mm, 350-900 nm 0 10 20 30 40 50 60 70 80 350 400 450 500 550 600 650 700 Wavelength, nmRelative grating efficiency, % Figure C.6 ST2000 spectrometer grating efficiency.

PAGE 253

APPENDIX C (Continued) Transmission and attenuati on of a 0.25 m long VIS/NIR fiber used w ith USB20000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 350 400 450 500 550 600 650 700 Wavelength, nmAttenuation, dB/m0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%Relative transmission Transmission T = 10^(-a L/10), % Transmission T = 10^ ( -a L ) % Attenuation a, dB/m Figure C.7 ST2000 spectrometer fiber transmission. Normalized correction for USB2000 spectrometer0 0.2 0.4 0.6 0.8 1 1.2 400450500550600650700750800850900 Wavelength, nmNormalized correctionCorrection assumed Sony ILX511B detector, 600 lines/mm 350-900 nm grating # 3, and VIS/NIR 0.25 m fiber of known attenuation [u sed T10^(-aL)] Figure C.8 ST2000 spectrometer overall correction. 232

PAGE 254

ABOUT THE AUTHOR Anna V. Sharikova obtained a Bach elor of Science degree in Laser Physics/Optical Engineering from the St. Pe tersburg Institute of Fine Mechanics and Optics (Technical University), St. Petersbur g, Russia in 1996, and a Master of Science degree in Physics from the University of S outh Carolina, USA in 2000. She entered the Ph.D. program in Applied Physics at the University of South Florida, USA in 2003. She has completed an industrial practicum in Laser-Induced Breakdown Spectroscopy at the Ocean Optics, Dunedin, FL as part of the Applied Physics Ph.D. training. She published a paper with Prof. D. K. Killinger called LIF detection of trace species in water using diffe rent UV laser wavelengths ( IJHSES 17 (4), 689-695, 2007), later reprinted in Spectral Sensing Research for Wa ter Monitoring Applications and Frontier Science and Technology for Chemical, Biological and Radi ological Defense (2008). She has presented the results of her research at se veral technical conferences, including CLEO, OSA, SPIE, and ISSSR.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200385Ka 4500
controlfield tag 001 002317271
005 20100831180502.0
007 cr cnu|||uuuuu
008 100831s2009 flua ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003013
035
(OCoLC)659877966
040
FHM
c FHM
d FHM
049
FHMM
090
QC21.2 (Online)
1 100
Sharikova, Anna V.
0 245
UV laser and LED induced fluorescence spectroscopy for detection of trace amounts of organics in drinking water and water sources
h [electronic resource] /
by Anna V. Sharikova.
260
[Tampa, Fla.] :
b University of South Florida,
2009.
500
Title from PDF of title page.
Document formatted into pages; contains 232 pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2009.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
520
ABSTRACT: A UV Laser Induced Fluorescence (LIF) system, previously developed in our laboratory, was modified and used for a series of applications related to the development and optimization of UV LIF spectroscopic measurements of trace contaminants in drinking water and other water sources. Fluorescence spectra of a number of water samples were studied, including those related to the reverse osmosis water treatment and membrane fouling, domestic and international drinking water, industrial toxins, bacterial spores, as well as several fluorescence standards. Of importance was that the long term detection of the trace level of Dissolved Organic Compounds (DOC) was measured, for the first time to our knowledge, over a one week period and with a time resolution of 2.5 minutes. A comparison of LIF emission using both 266 nm and 355 nm excitation was also made for the first time. Such real-time and continuous measurements are important for future water treatment control.The LIF system was modified to accommodate UV Light Emitting Diodes (LED) as alternative excitation sources, and tested for the detection of trace organic species in water. In addition, a compact system using LED excitation and a spectrometer was xviii developed and underwent initial testing. The original LIF system had two laser sources, 266 nm and 355 nm. The additional sources incorporated in the system were UV LEDs emitting at 265 nm, 300 nm, 335 nm and 355 nm. The LED spectral emission was studied in detail, in terms of spectral variability and power output. It was found that all LEDs had some emission in the visible spectrum, and an optical filter was used to remove it. The signal-to-noise ratio for the LED-based systems was determined and compared with that of the LIF system. The fluorescent signal of the LED-based system was smaller by 1 to 2 orders of magnitude, despite the fact that the LED pulse energy was 2 to 3 orders of magnitude less than the laser's.As such, the fluorescent signal from the LED was greater than expected. Therefore, a UV LED may be a compact and much cheaper optical source for future water measurement instruments.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: Dennis K. Killinger, Ph.D.
653
Water fluorescence
Laser-induced fluorescence
Reverse osmosis
Water quality monitoring
Online real time reagentless system
690
Dissertations, Academic
z USF
x Physics
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.3013