USF Libraries
USF Digital Collections

Estimation of the particle and gas scavenging contributions to wet deposition of organic and inorganic nitrogen

MISSING IMAGE

Material Information

Title:
Estimation of the particle and gas scavenging contributions to wet deposition of organic and inorganic nitrogen
Physical Description:
Book
Language:
English
Creator:
Calderón, Silvia Margarita
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
DON
DIN
Rain scavenging
UV photolysis
Atmospheric chemistry
Dissertations, Academic -- Chemical Engineering -- Doctoral -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Atmospheric deposition of nitrogen species represents an additional nutrient source to natural environments, and can alter the nitrogen cycle by increasing nutrient levels beyond the requirements of organisms. In Tampa Bay, atmospheric deposition of dissolved inorganic nitrogen species (DIN) has been found to be the second largest nitrogen source, but little is known about dissolved organic nitrogen species (DON). The research goal was to improve the dry and wet deposition estimates by inclusion of the DON contribution. In the atmospheric chemistry field a standard method to measure DON in atmospheric samples has not been agreed upon. This research proposes the use of the ultraviolet (UV)-photolysis method and presents the optimal settings for its application on atmospheric samples. Using a factorial design scheme, experiments on surrogate nitrogen compounds, typically found in the atmosphere, indicated that DON can be measured with no biases if optimal settings are fixed to be solution pH 2 with a 24 hr irradiance period.DIN species (NH₄□□₊, NO₂□□₋, NO₃□□₋) and DON concentrations were determined in fine (PM₂.₅) and coarse particles (PM₁₀₋₂.₅) as well as in rainwater samples collected at Tampa Bay. The estimates of wet despotition fluxes for NH₄□□₊, NO₃□□₋ and DON were 1.40, 3.18 and 0.34 kg-N ha⁻¹yr⁻¹, respectively. Hourly measured gas concentrations and 24-hr integrated PM10 concentrations were used in conjunction with a below-cloud scavenging model to explain DIN and DON concentration in rainwater samples. Scavenging of aerosol phase DON contributed only 0.9±0.2 percent to rainwater DON concentrations, and therefore gas scavenging should be responsible for 99 percent. These results confirmed the existence of negative biases in the dry and wet deposition fluxes over Tampa Bay. There is increasing interest in simulating wet deposition fluxes, and the proposed below-cloud scavenging model offers a new computational approach to the problem.It integrates the typical gas and particle collection functions and the concept of the deposition-weighted average concentrations. The model uses mass balance to describe the time-dependent cumulative contribution of all droplets in the rain spectrum to the rainwater concentration, giving predictions closer to experimental values and better estimations than those reported in the literature for similar cases.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Silvia Margarita Calderón.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 188 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 - 001795154
oclc - 146061424
usfldc doi - E14-SFE0001538
usfldc handle - e14.1538
System ID:
SFS0025856:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Estimation of the Particle and Gas Scav enging Contributions to Wet Deposition of Organic and Inorganic Nitrogen by Silvia Margarita Caldern A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Chemical Engineering College of Engineering University of South Florida Co-Major Professor: Scott W. Campbell, Ph.D. Co-Major Professor: Noreen D. Poor, Ph.D. Abdul Malik, Ph.D. Aydin K. Sunol, Ph.D. John T. Wolan, Ph.D. Date of Approval: March 27th, 2006 Keywords: DON, DIN, rain scavengi ng, UV photolysis, atmospheric chemistry Copyright 2006, Silvia Margarita Caldern

PAGE 2

DEDICATION To my beloved husband, Edinzo. His example of hard work, honesty and patience has encouraged me in the hardest times. During my journey, his love is the guiding star and the shining light that makes my sky brighter. To my family, Hayde, Silvina, Davi d and Carolina. Their love goes in my blood as the sage on the trees. With every breath I fight for their happiness as they fight for mine. To my friends. You are the family give n to me by God. There are not words to describe how close to my heart you are. Thank you for making me a better human being. To my home country, Venezuela. You ar e the place were my dreams were born and to you they come back. I was lucky to be born in your lands where our parents sing for us the national an them as a lullaby. To God. You are all over around, as the voice in my head, the beat in my heart, the sunshine in my face, the wind blowing my hair.

PAGE 3

ACKNOWLEDGMENTS I would like to express my deepest gr atitude to Dr. Scott Campbell and Dr. Noreen Poor. They gave to me not only th e academic advice but all the support to overcome the difficulties of being far from home. Their dedication to do Science with the highest quality will be forever the example to follow. They planted in my heart the seeds of curiosity and perseverance to l ook for the truth. They will grow to produce more seeds for my future students. I also would like to thank all th e USF members of the Bay Regional Atmospheric Chemistry Experiment (BRACE). They taught me so many things that a simple “thank you” does not represent my gratitude. Special thanks go to all the committee members and faculties of the USF Chemical Engineering Department. They we re my professors and offered to me numerous tools to be a better scientist. Finally, I would like to thank the Univer sity of Los Andes in Venezuela and its Chemical Engineering School for the opportuni ty to increase my academic formation. This research was sponsored by the Florida Department of Environmental Protection (AQ-156) under the dir ection of Dr. Thomas Atkeson.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES vi LIST OF FIGURES ix LIST OF ACRONYMS xii LIST OF VARIABLES xiv LIST OF PAPERS xvi ABSTRACT xvii 1. INTRODUCTION 1 Air Quality and Eutrophication in Tampa Bay 1 Goals and Structure of the Dissertation 5 Research Contributions 9 2. ORGANIC NITROGEN IN ATMOSPHERIC SAMPLES 13 3. INVESTIGATION OF THE UV PHOTOLYSIS METHOD FOR THE DETERMINATION OF DISSOLVED ORGANIC NITROGEN IN ENVIRONMENTAL SAMPLES 17 Comparison of the UVPhotolysis With Othe r Methods to Determine DON 17

PAGE 5

ii Materials and Methods 23 Chemicals 23 Chromatographic Analyses 24 DON Analyses 25 Field Sampling and Sample Processing 26 Blanks and General Performance 27 pH and Irradiance Period 22 Factorial Experiments 29 Proportions of Ammonium, Nitrite, and Nitrate Formed from DON 36 Application of the UV Photolysis Method to Aerosol Samples 39 Effect of DON Mixtures a nd DIN/DON Ratio on Conversion Efficiency 43 Summary 47 4. PARTICLE AND GAS SCAVENGING PROCESSES 49 Introduction 49 Characteristics of the Rain Droplet Distribution 51 Deposition-Weighted Average Concentration 56 Gas Scavenging 58 Gas Scavenging With Reversible Absorption at Constant Gas Concentration 61

PAGE 6

iii Gas Scavenging With Irreversible Absorption at Constant Gas Concentration 64 Gas Scavenging With Irreversible Absorption and Exponential Decay of th e Gas Concentration 66 Particle Scavenging 70 Particle Scavenging at Consta nt Particle Concentration 79 Particle Scavenging With Exponential Decay of the Particle Concentration 80 Scavenging Coefficients Used in Community-Scale and Mesoscale Air Quality Models 81 5. ESTIMATION OF THE GAS AND PARTICLE CONTRIBUTIONS TO WET DEPOSITION OF DISSOLV ED ORGANIC NITROGEN 83 Introduction 83 Experimental Methods 84 Field Sampling and Sample Processing 84 Laboratory Analyses 85 DIN and DON Concentrations and Wet Deposition Fluxes 86 Estimation of Dimethylamine (DMA) Gas Concentrations 91 Particle and Gas Scavenging Contributions 95 Summary 101

PAGE 7

iv 6. ESTIMATION OF THE PART ICLE AND GAS SCAVENGING CONTRIBUTIONS TO WET DE POSITION OF INORGANIC NITROGEN 102 Introduction 102 Experimental Methods 109 Field Sampling and Sample Processing 109 Aerosols and Rainwater Concentrations 111 Scavenging Coefficients 130 Summary 131 7. ANALYSIS OF PARTICLE FORMATION PROCESSES THROUGH THE EFFECT OF METEOROLOGI CAL CONDITIONS ON ORGANIC AND INORGANIC NITROGEN CO NCENTRATIONS IN ATMOSPHERIC AEROSOLS 133 Introduction 133 Statistical Analyses 134 Multi-Linear Regressions 150 Summary 156 8. CONCLUSIONS AND FURTHER RESEARCH 157 REFERENCES 162 APPENDICES 176 Appendix A. Anova Results From 22 Factorial Design 177

PAGE 8

v Appendix B. Configuration of th e R&P Dichotomous Air Sampler 180 Appendix C. Study of Error Propagation for the DON Concentration 183 ABOUT THE AUTHOR End Page

PAGE 9

vi LIST OF TABLES Table 3-1. p-values from Analyses of Variance of the Conversion of Organic Nitrogen Solutions under UV-Photolysis. 35 Table 3-2. Average Percentage of In organic Nitrogen Produced Per Species After UV-Photoxidation at pH=2 an d 24-hr Irradiance Period at Concentrations Under 100 M-N. 37 Table 3-3. DIN and DON Concentrations (nmol-N m-3) in PM10 Samples Collected at the Gandy Bridge Monitoring Site, Tampa, FL. 39 Table 3-4. Conversion Efficiencies of Urea, Amino Acids, and Methylamine Solutions After UV Photolysis at pH 2 for 24-hr. 44 Table 3-5. Effect of the DIN DON-1 Ratio on the Conversion of a 10 M-N Urea Solution. 45 Table 4-1. Parameters for Log-normal Size Distribution for PM10 Samples Collected at the Gandy Bridge Monitoring Site (Campbell, 2005). 77 Table 5-1. Experimental Concentrations of 24-hr Integrated Aerosol (PM10) and Rainwater Samples Collected at the Gandy Bridge Monitoring Site, Tampa, FL. 87 Table 5-2. Composition of PM10 Samples from the Gandy Bridge Monitoring Site, Tampa, FL, July-September 2005. 90 Table 5-3. Antoine’s Constant to Calculat e Vapor Pressure of Pure Substances ( T in oC, P in mmHg) (Gmehling et al. 1977). 94

PAGE 10

vii Table 5-4. Average Dissolved Organi c Nitrogen (DON) and Dimethylamine (DMA) Concentrations in Rainwa ter Concentrations Predicted by Particle Scavenging Using the Aver age of Three Different Drop Size Distributions. 99 Table 6-1. Aerosol (PM10) and Rainwater Concentrations and Average Meteorological Conditions at the Gandy Bridge Monitoring Site. 112 Table 6-2. Hourly precipitation rates and gas concentrations during rain events at the Gandy Bridge monitoring site, Tampa, FL. 114 Table 6-3. Contribution of Particle and Gas Scavenging to Rainwater Concentrations of Samples Collected at the Gandy Bridge Monitoring Site, FL. 115 Table 6-4. Contribution of Particle and Gas Scavenging to Rainwater Concentrations of Samples Collected at the Gandy Bridge Monitoring Site, FL consideri ng Constant Particle and Gas Concentrations. 117 Table 6-5. Scavenging Coefficients from the Below-Cloud Scavenging Model. 131 Table 7-1. Average Meteorological Cond itions at the Gandy Bridge Monitoring Site. 136 Table 7-2. Average DIN and DON Particle Contents in Samples from the Dry and Wet Period. 138 Table 7-3. Average DIN and DON Con centrations at the Gandy Bridge Monitoring Site. 139 Table A-1. Conversion of 80 M-N Amino Acids to Inorganic Nitrogen After UV Photolysis. 177 Table A-2. Conversion of 80 M-N Urea to Inorganic Nitrogen After UV Photolysis. 178

PAGE 11

viii Table A-3.Conversion of 80 M-N Methylamine to Inorganic Nitrogen After UV Photolysis. 179 Table C-1. Standard Deviation Associat ed With Variables Involved in the DON Determination. 185

PAGE 12

ix LIST OF FIGURES Figure 1-1. Map of the Tampa Bay Area and the Gandy Bridge AIRMoN Site, Identifying the Locations of the Largest Local NOX and NH3 Emissions Sources (Strayer, H. 2005). 2 Figure 1-2. Contribution of Nitrogen Sour ces to the Total Annual Loading to Tampa Bay (1992-1994 average) (TBNEP, 1996). 3 Figure 2-1. Contribution of Organic Nitr ogen to the Total Dissolved Nitrogen Deposited in Areas Around the World (Cornell et al. 2003). 14 Figure 3-1. Photochemical Chamber R eactor Model RPR-100 (Southern New England Co, Inc.). 26 Figure 3-2. Inorganic Nitrogen Distribu tion of the Organic Nitrogen Fraction After UV-Photolysis of Aerosol Samples. 42 Figure 4-1. Important Colle ction Mechanisms for Particles During Rain Scavenging. 71 Figure 4-2. Collision Efficiency Function in Terms of the Particle and Droplet Diameter. 74 Figure 4-3. Contributions of the Most Im portant Mechanisms to the Collision Efficiency. 74 Figure 4-4. Collision Volume for Particle/Droplet Systems. 75 Figure 5-1. Composition of Rainwater a nd Aerosol Samples Collected on the Gandy Monitoring Site During July-August 2005. 89

PAGE 13

x Figure 5-2. Dissolved Organic and Inorga nic Nitrogen Concentrations from PM2.5 Samples Collected at the Gandy Br idge Monitoring Site Between November (2004) to September (2005). 96 Figure 6-1. Instrument Schematic Used for Gas Measurement (Hartsell et al ., 2006). 122 Figure 6-2. Particle Scavengi ng Coefficients at Differe nt Precipitation Rates for Three Drop Size Distributions: LLog-Normal, MM-Massambani and Morales, MP-Marshall-Palmer. High po (precipitation rate) =8.9 mm hr-1. Low po=0.8 mm hr-1. 123 Figure 6-3. Effect of the Drop Size Dist ribution on the Varia tion of Ammonium PM10 Concentration After One-hr Rain Event at 2.5 mm hr-1 (L-Lognormal, MM-Massambani and Morales, MP-Marshall-Palmer). 124 Figure 6-4. Effect of the Drop Size Dist ribution on the Varia tion of Nitrate PM10 Concentration After One-hr Rain Event at 2.5 mm hr-1 (L-Log-normal, MM-Massambani and Morales, MP-Marshall-Palmer). 125 Figure 7-1. Variation of the Average Wi nd Speed and Direction for the Dry and Wet Period on the Gandy Monitoring Site, Tampa, FL. 137 Figure 7-2. Variation of the NH4 + Concentrations in PM10 With Respect to Meteorological Conditions. 142 Figure 7-3. Variation of the NO3 Concentrations in PM10 With Respect to Meteorological Conditions. 145 Figure 7-4. Variation of the DON Concentrations in PM2.5 With Respect to Meteorological Conditions. 148 Figure 7-5. Variation of the DON Concentrations in PM10 With Respect to Meteorological Conditions. 149 Figure 7-6. Goodness of the Fitting for the Log-Transformed NH4 + Concentration With Meteorological Conditions. 152

PAGE 14

xi Figure 7-7. Goodness of the Fitting for the Log-Transformed NO3 Concentration With Meteorological Conditions. 153 Figure 7-8. Goodness of the Fitting for the Log-Transformed DON Concentration in PM2.5 With Meteorological Conditions. 155 Figure B-1. Schematic Flow Diagram fo r the Dichotomous Air Sampler. 181 Figure B-2 Dichotomous Air Sampler Characteristics 182

PAGE 15

xii LIST OF ACRONYMS Variable referred to as Acronym Dissolved Inorganic Nitrogen DIN Dissolved Organic Nitrogen DON Total Dissolved Nitrogen TDN Particulate Matter Under 10 m PM10 Coarse Particles PM10-2.5 Fine particles PM2.5 Wind Speed WS Wind Direction WD Standard Deviation of th e Wind Direction WDirSTD Relative Humidity RH Dry Bulb Temperature DBT Particle Size Distribution PSD Droplet Size Distribution DSD Logarithmic Droplet Size Distribution L-DSD Morales-Massambani Droplet Size Distribution MM-DSD

PAGE 16

xiii Marshall-Palmer Droplet Size Distribution MP-DSD Methylamine MA Dimethylamine DMA Trimethylamine TMA Urea U Amino acids AA NRTLNon Random Two Liquid M odel to determine activity of species in solution NRTL Aerosols Inorganic Model 2 AIM-2

PAGE 17

xiv LIST OF VARIABLES Variable Symbol Units Time t s-1 Vertical distance measured from cloud base z m Droplet diameter Dp m Particle diameter dp m Pollutant gas concentration Cg(z,t) mol m-3 Pollutant aqueous concentration or droplet concentration C(z,t) mol L-1 Pollutant rainwater conc entration at ground level Crainwater(h,t) mol L-1 Droplet Settling velocity Ut m s-1 Total falling distance per droplet h m Rain intensity or precipitation rate po mm hr-1 Gas-phase mass transfer coefficient of pollutant Kc m s-1 Gas diffusivity of pollutant Dg m2 s-1 Particle diffusivity DAerosol m2 s-1 Air density g kg m-3 Air viscosity g N m-2 s-1 Particle mass concentration nM mol m-3 Geometric mean of the particle diameter used in the logarithmic particle size distribution dgm m Standard Deviation of the particle diameter used in the logarithmic particle size distribution p m Mass concentration of pollutant in PM10 M mol m-3

PAGE 18

xv Henry’s law coefficient as the ratio of the molar pollutant concentration in the aqueous and gas phases Hc Dimensionless Reynolds number Re Dimensionless Schmidt number Sc Dimensionless Stokes number St Dimensionless Total Particle-droplet collision efficiency E Dimensionless Particle-droplet collision efficiency by Brownian diffusion ED Dimensionless Particle-droplet collision efficiency by interception EN Dimensionless Particle-droplet collision efficiency by impaction EM Dimensionless Ratio of particle diamet er to droplet diameter Dimensionless Ratio of air viscosity to water viscosity Dimensionless Critical Stokes number S* Dimensionless Particle relaxation time s Gas scavenging coefficient s-1 Particle scavenging coefficient s-1 Mean mass particle scavenging coefficient m s-1 Particle Scavenging coefficient normalized by precipitation rate m po -1 (s mm/hr) -1 Gas Scavenging coefficient normalized by precipitation rate po -1 (s mm/hr) -1 Droplet Number size dist ribution as number of droplets per unit volume of air and size category ND Number of droplets m-3 m-1 One-hour rain interval ts hr Particle mass scavenged per rain interval p scavengedW mol m-3 Pollutant Wet deposition flux of droplets with size Dp p WD t F mol m-2 s-1 Rain intensity of of droplets with size Dp pD t I m s-1

PAGE 19

xvi LIST OF PAPERS This thesis is based on the following pape rs referred to by their Roman numerals throughout. Papers are reprinted with pe rmission from the copyright holders. I. Caldern, S.M., Poor, N.D. and Campbell, S.W. Investigation of the UV photolysis method for the determinati on of dissolved organic nitrogen in environmental samples. Journal of the Air and Waste Management Association, 2006, In Press. II. Caldern, S.M., Poor, N., and Campbell, S.W. (2006) Analysis of particle formation processes through the effect of meteorological conditions on organic and inorganic nitrogen concentr ations in atmospheric aerosols. In 86th American Meteorological Society Annual Meeting. 8th conference on Atmospheric Chemistry, Paper no. 105312. American Meteorological Society (AMS), Atlanta, GA. III. Caldern, S. M.; Poor, N. D.; Campbell, S. W. (2006) Estimation of the particle and gas scavenging contribut ions to wet deposition of organic nitrogen; Atmos. Environ ., 2006, In review. IV. Caldern, S. M.; Poor, N. D.; Campbell, S. W. Estimation of the particle and gas scavenging contributions to wet deposition of inorganic nitrogen; Atmos. Environ 2006, In review.

PAGE 20

xvii ESTIMATION OF THE PART ICLE AND GAS SCAVENGING CONTRIBUTIONS TO WET DEPOSITI ON OF ORGANIC AND INORGANIC NITROGEN Silvia Margarita Caldern ABSTRACT Atmospheric deposition of nitrogen speci es represents an additional nutrient source to natural environments, and can alter the nitrogen cycle by increasing nutrient levels beyond the requirements of organism s. In Tampa Bay, atmospheric deposition of dissolved inorganic nitrogen species (DIN) has been found to be the second largest nitrogen source, but little is known about dissolved organic nitrogen species (DON). The research goal was to improve the dry and wet deposition estimates by inclusion of the DON contribution. In the atmospheric chemistry field a standard method to measure DON in atmospheric samples has not been agreed upon. This research proposes the use of the ultraviolet (UV)-photolysis method and presents the optimal settings for its application on atmospheric samples. Using a factoria l design scheme, experiments on surrogate nitrogen compounds, typically found in the atmosphere, indicated that DON can be

PAGE 21

xviii measured with no biases if optimal settings are fixed to be solution pH 2 with a 24-hr irradiance period. DIN species (NH4 +, NO2 -, NO3 -) and DON concentrations were determined in fine (PM2.5) and coarse particles (PM10-2.5) as well as in rainwater samples collected at Tampa Bay. The estimates of wet deposition fluxes for NH4 +, NO3 and DON were 1.40, 3.18 and 0.34 kg-N ha-1yr-1, respectively. Hourly-measured gas concentrations and 24-hr integrated PM10 concentrations were used in conjunction with a below-cloud scavenging model to explain DIN and DON concentration in rainwater samples. Scavenging of aerosol-phase DON contri buted only 0.9 0.2% to rainwater DON concentrations, and therefore gas scavengi ng should be responsible for 99%. These results confirmed the existence of negative bi ases in the dry and wet deposition fluxes over Tampa Bay. There is increasing interest in simulating wet deposition fluxes, and the proposed below-cloud scavenging model offers a new computational approach to the problem. It integrates the typi cal gas and particle collection functions and the concept of the deposition-weighted average concentra tions. The model uses mass balance to describe the time-dependent cumulative contribu tion of all droplets in the rain spectrum to the rainwater concentration, giving predic tions closer to experimental values and better estimations than t hose reported in the liter ature for similar cases.

PAGE 22

1 1. INTRODUCTION Air Quality and Eutrophication in Tampa Bay The effect of air pollutio n goes beyond the reduction in the air quality and extends to the reduction in th e water and soil qual ities. After deposition, gases, particles and rain droplets constitute external sources of chemicals to the ecosystems and in many cases alter their ability to sustain life. As an example, atmospheric deposition of nitrogen species represents an additional nutrient source to natu ral environments, and can unbalance the nitrogen cy cle by increasing nutrient le vels beyond the requirements. This is referred as nitrogen saturation, a situation in which the supply of nitrogen compounds from total nitrogen input ex ceeds the demand by plants and microorganisms (Stoddard, 1994). Nitrogen saturati on has negative consequences such as eutrophication, a phenomenon associated with high phytoplankton concentrations, decrease in sea grass coverage reduction in light penetratio n; decrease of oxygen levels, poor water quality and fish kills (Paerl, 1988). In the late 1970s, eutrophi cation affected coastal areas of the USA, especially Tampa Bay, and elevated phytoplankton concentr ations caused a decrease in the light penetration with reduction in the water quality and a decl ine in sea grass meadows. Although the Tampa Bay area is enriched in another nutrient, phosphorus (phosphorus

PAGE 23

2 mining in the area offers 20 % of the world production), water quality models have predicted a strong response of phytoplankton co ncentrations in Tampa Bay to changing nitrogen inputs and have positively correlated phytoplankton concentrations to total nitrogen levels in water (Wang et al. 1999). As an evidence of the negative eutrophication effects, high ni trogen levels were linked to a 46 % loss of the sea grass coverage between 1950 and 1982 (Tomasko et al. 2005). Pinellas Co. Manatee Co. Hillsborough Co. Gandy Bridge AIRMoN Site. 1 NOX TECO Big Bend Power Plant 2 NOX TECO Gannon Power Plant 3 NOX, NH3 Florida Power Manatee Power Plant 4 NOX, NH3 Florida Power Weedon Is. Power Plant 5 NOX Pinellas Co. 110th Ave. Power Plant 6 NH3 CF Industries, Inc., Plant City Phosphate 7 NH3 Nitram, Inc. (now defunct) 8 NH3 Cargill Fertilizer, Inc. 1 2 3 4 5 6 7 8 Figure 1-1. Map of the Tampa Bay Area and the Gandy Bridge AIRMoN Site, Identifying the Locations of the Largest Local NOX and NH3 Emissions Sources (Strayer, H. 2005). Although in 1991 runoff water was consid ered as the main N source and atmospheric deposition was assumed neg ligible, for 1992-1994 it was estimated that

PAGE 24

3 29% of the total nitrogen loading was cau sed by atmospheric deposition of dissolved inorganic nitrogen species (DIN), mostly deposited as ammonia/ammonium from fertilizer industries and agricultural emissions or as nitric acid/nitrate after transformation of NOx from power plants around the area (Figure 1-1) and vehicle emissions (TBNEP, 1996). Additional nitrogen sources are presented in Figure 1-2. For 1996-1999 this number slightly decreased to 22 %, being equivalent to 760 metric tonsN yr-1, 58 % of this value as ammonia/ammonium and 42 % as nitric acid/nitrate with a larger flux coming from wet deposition vs. dry deposition (56 % of the total-N flux) (Poor et al. 2001). Average Annual Loadings of Nitrogen to Tampa Bay, FL(1992-1994) 29% 13% 7% 5% 46% Stormwater Runoff Atmospheric Deposition Industrial & Municipal Point Sources Fertilizer Material Losses Springs & Groundwater Figure 1-2. Contribution of Nitrogen Sources to the Total Annual Loading to Tampa Bay (1992-1994 average) (TBNEP, 1996).

PAGE 25

4 Once the link between high nitrogen le vels and the change s in the ecological equilibrium of the bay was confirmed, correc tive actions have focused on the reduction of anthropogenic nitrogen loadings thr ough the improvement in the treatment of wastewater effluents an d re-powering power plants from coal to natural gas. In recent years, overall water quality on the bay has in creased and the sea gr ass coverage is 20 % higher than in 1982 but still 65 % of the 1950 coverage (Tomasko et al. 2005). However, all loadings estimates, inve ntories and corrective actions did not consider possible sources of organic nitrog en species or dissolved organic nitrogen (DON). Although DON species such as amino acids, aliphatic amines and urea have been identified in atmospheric sample s, the DON fraction represents a poorly understood element of the nitrogen atmospheric deposition. Therefore it is an important piece for the complete estimation of depos ition fluxes over natu ral environments; especially those affected by or susceptib le to be affected by nitrogen saturation phenomena, as the Tampa Bay Estuary. In such coastal waters, DON species act as a new source of nutrients. After photolysis by the solar radiation, DON generates bioavailable nitrogen mainly in the form of NH4 + and contributes potentially to the productivity of coastal waters (Vahatalo and Zepp, 2005; Wang et al. 2000b; Tarr et al. 2001; Tarr et al. 1999). The importance of the constant monitori ng and regulation of all anthropogenic nitrogen loadings to the bay becomes eviden t when it is realized that it is Florida’s largest open-water estuary. Tampa Bay Estuar y stretches 398 square miles at high tide, and as a natural system plays an important ro le in the life and survival of many plants

PAGE 26

5 and animal species. Due to the mixing of fr esh water from the rivers and sea-water, Tampa Bay Estuary serves as an adequate z one for the early stages of development of many aquatic species, such as fish, shellfish, and crustaceans. It has been estimated that 70 % of the commercially important species of fish depends on estuaries for their development. In addition, the water quality of Tampa Bay influences the quality of life of more than 2 million people living ar ound its coastal areas (TBNEP, 1996). Goals and Structure of the Dissertation In the last section it was established that nitrogen levels in Tampa Bay guide not only its recovery from eutrophication but also influence the successful development of the flora, fauna and human population livi ng in/around the area. It was also well established that the correct estimation of th e atmospheric deposition contribution to the bay nitrogen loadings would contribute to better regulation and monitoring plans. Due to the lack of knowledge about DON species a nd their significance in the nitrogen cycle of Tampa Bay, the main goals of the research were to: I. Estimate the DON and DIN concentration in atmospheric aerosols and rainwater samples collected in Tampa Bay. II. Identify the phase (gas or particle) with the biggest contribution to the DIN and DON concentrations in rainwater. III. Improve the nitrogen deposition estimates to Tampa Bay by inclusion of the organic nitrogen contribution to wet deposition fluxes. IV. Identify possible DON sources missing from current inventories and meteorological effects leading to seasonal patter ns in DON and DIN concentrations.

PAGE 27

6 All these goals were successf ully reached and the results were comprised in this document in eight chapters. The main findings about DON in atmospheric samples reported in the literature were presented as a review in the second chapter. In this chapter the fundamental basis that motiv ated this research are established. Because there is not a standard method to measure DON in atmospheric samples, the research started with the selection of a method and its assessment. The UV-photolysis was selected from among the available options, mainly for its advantages of minimal sampling handling a nd low cost in the equipment. The method was tested to identify possible biases in the DON measurements due to inadequate operational parameters. The optimal method settings were found through a 22 factorial design in terms of the effect of the sa mple pH and the irradiance period on the photochemical conversion of selected organic ni trogen species to their inorganic forms. Once the optimal settings of the UVphotolysis were found, the method was applied to determine DON concentrations in atmospheric aerosols collected at the Tampa Bay. A sampling campaign was conducted at the Gandy Bridge monitoring site (Figure 1-1) to collect samples of particulate matter (PM) under 10 m of aerodynamic diameter or PM10. Samples were collected in two fr actions, fine particles (PM under 2.5 m of aerodynamic diameter or PM2.5) and coarse particles (PM under 10 m and over 2.5 m of aerodynamic diameter or PM10-2.5). Details about the method optimization, sampling, sample preparation, analysis and c oncentrations were pr esented in the third chapter.

PAGE 28

7 Results of the first sampling campai gn provided motivation to examine the possible existence of DON species in rainwater. PM10 and rainwater samples were collected simultaneously to examine the natu ral connections between particle and rain droplets concentrations give n by scavenging processes. PM10 concentrations were used in conjunction with a below-cloud scave nging model to estimate the DON and DIN rainwater concentrations. The below-cloud scavenging model used was described in detail in the fourth chapter. All the theoreti cal basis of the wash-out process in terms of the reversibility of the gas-absorption into th e rain droplets were comprised in detail in this section of the document. To reach the second research goal, the simplest case of the below-scavenging model (constant particle and gas concentrations during rain events) was applied to estimate the contribution of DON and DIN particle scavenging to their observed rainwater concentrations. Because the presence of Dimethylamine (DMA) was detected in PM2.5 samples, its concentrations in the gas phase were estimated and used the scavenging model to quantify its contribution to DON rainwater concentrations. All results were reported in the fifth chapter. In a similar way, the below-cloud scave nging model was used to predict the DIN wet deposition flux using the DIN particle concentrations and DIN gaseous species measured at the monitoring site. The a mmonia/ammonium and nitric acid/nitrate contributions to the DIN wet deposition flux were estimated for the two extreme scavenging cases, constant gas and particle concentrations and time-exponential decay of the concentrations during rain events. Th e goal of this approach was to bracket the

PAGE 29

8 results between the maximal and minimal removal rates. This allowed quantification of the effect of emissions and gas-to-particle conversions during rain events, processes not directly accounted for in the model. A descri ption of this section of the research was compiled in the sixth chapter. The information collected during the two sampling campaigns was used to satisfy the requirements of the fourth rese arch goal. Both sampling periods were done under different meteorological conditions; a nd this allowed studying possible seasonal effects over the DON and DIN concentrations in atmospheric aerosols as well as evaluating possible particle formation pro cesses leading to the observed DON and DIN particle levels. Simple statistical analyses we re carried out to iden tify significant effects of meteorological conditions on DIN and DON concentrations Multi-linear regression analyses between DIN and DON concentra tions and the average meteorological conditions were used to look for evidence a bout possible particle formation processes. These results were comprised into the seventh chapter. Conclusions and recommendations for future research were included in the eight and final chapter.

PAGE 30

9 Research Contributions I. In the atmospheric chemistry community a standard method to measure DON in atmospheric samples has not been agreed upon. The UVphotolysis method was selected to overcome this issue and its optimal settings were determined to avoid biases in the measurement of DON concentrations in atmospheric samples. The results showed that the UV-photolysis can be adopted to analyze atmospheric samples with DON concentrations under 33 M-N using an irradiation period of 24-hr and adjusting to 2 the solutio n pH. It was also found that underestimations in DON concentrations can occur if the UV-photolysis method is applied at less than optimal irradiation periods and solu tion pH, but DON could be overestimated by 15 % if the DIN/DON ratio is higher than 4. II. The use of the factorial design technique he lped to reduce the num ber of experiments necessary to test the UV-Phot olysis performance, with th e selected surrogate organic nitrogen compounds. This appr oach is innovative in the atmospheric chemistry field where numerous experiments are required to test the performance of the analytical procedures over the complex matr ices of atmospheric samples. III. UV-photolysis and ion chromatography analys es showed that DON species represent 10 % of the total dissolved nitrogen present in aerosol and rainwater samples collected at Tampa Bay. This suggested that atmospheri c nitrogen loading to surface waters are biased low because of unaccounted for DON contributions to wet and dry deposition fluxes. These findings also indicated th e existence of missing sources in the inventories of nitrogen deposited on Tampa Bay.

PAGE 31

10 IV. In the months of July-August 2005, concentra tions in rainwater samples showed that the DON wet deposition flux is 0.34 kg-N ha-1 yr-1 and contributes 7% to the total nitrogen loading from wet deposition, which was equal to 4.91 kg-N ha-1 yr-1. DIN species showed to be the biggest contribu tors to this value, with 1.40 and 3.18 kg-N ha-1 yr-1 from the wet deposition fluxes of NH4 +, NO3 respectively. This confirmed the existence of biases in previous estimations for Tampa Bay. V. DON in fine particles represented 75 19 % of the total DON found in PM10 samples. This finding added to the existence of strong correla tions between DON and DIN in fine particles and meteorological conditions suggested that gas-to-particle conversions are likely to be responsible for the TDN c oncentration seen in atmospheric aerosols collected at Tampa Bay. VI. Dimethylamine (DMA) showed to be present in PM2.5 samples at concentrations between below detection and 1911 pmol-N m-3, with an average value of 688 615 pmol-N m-3. DMA represented by itself an averag e contribution of 12.8 6.7 % to the total DON concentration measured in the same particle fraction. VII. Particle scavenging of DON in PM10 is responsible just for 0.9 % 0.2% of the DON concentrations in rainwater. Gas scavengi ng of DON is assumed to be responsible for 99% of the DON wet deposition flux. Gas-pha se organic nitrogen, likely contributes to the dry deposition of nitrogen to Tampa Bay. VIII. Using DMA concentrations in fine particles and estimated DMA gaseous concentrations it was determined that DM A is responsible for 0.4 0.7 % of all DON

PAGE 32

11 present in rainwater samples. DMA particle scavenging accounted for 11.7 7.1 % of the contribution from D ON-particle scavenging. IX. Below-cloud scavenging calculations confir med that DIN concentrations in gas and particles can explain their concentrations in rainwa ter. The wet deposition of NH4 + is mainly caused by gas scavenging of ammonia, while the NO3 wet deposition flux is mainly consequence of particle scavenging. X. Results from scavenging calculations showed that the scavenging process is closely modeled to real conditions when it is assumed that gas and particle concentrations remain constant instead of decaying expone ntially with time during rain scavenging. This could indicate that continuous amm onia emissions and nitrate from gas-toparticle conversion help to counteract the decrease in their con centration during rain scavenging. XI. DON gaseous species could be responsible for a dry deposition nitrogen flux higher than the estimated values for Tampa Bay. These results highlight the need for a method to measure gas-phase organic nitrogen gases. XII. The below-cloud model proposed in this rese arch offers a new computation approach for the simulation of wet deposition processes, a problem of increasing interest to the atmospheric chemistry community. By integr ation of the typical gas and particle collection models and the concept of deposit ion-weighted average concentrations, the model is able to predict rainwater sample concentrations closer to the experimental values than other models reported in the literature for the same cases.

PAGE 33

12 XIII. The study of the contribution of DON species to atmospheric concentrations is of substantial value to many researchers around th e world trying to quantify its role in ecosystems growth limitations, especial ly with the increasing anthropogenic emissions. Due to the possible existence of natural background c oncentrations after the DON production by living organisms, conc lusions from this research can be extended to similar geographic areas and can be used to identify and estimate possible sources when other information is not available. XIV. This research offers information about the composition of the organic compound fraction in the atmospheric samples, a vari able of increasing interest and not well understood by scientists. DON species in partic ulate matter and rain droplets, as any other organic compound, can force change s in the climate by influencing cloud condensation nuclei processes; and, if toxic, also can impact human health after inhalation into the respiratory tract.

PAGE 34

13 2. ORGANIC NITROGEN IN ATMOSPHERIC SAMPLES DON species have shown to be present in atmospheric samples collected around the world and can contribute to the total n itrogen loadings by offering a new “fresh” nitrogen source, readily available to aquati c species. However since two decades ago, DON atmospheric deposition was rarely associat ed with the unbalance of the nitrogen cycle in ecosystems. Considerable effort s have been made to determine DON in atmospheric samples, due not only to the cont ribution of nitrogen species to the total nitrogen deposition process, but also due to its critical role of reactive nitrogen compounds in photochemistry (Gorzelska et al. 1992; Timperley et al. 1985). In 1993 Milne and Zika presented a summary of 12 se parate studies from four continents and covering a 14-year time span, in which the ratios of deposition rates of dissolved total organic nitrogen to total inorganic forms in precipitation varied from 13 to 800%. In a more recent review (Cornell et al. 2003), it was shown that th e percentage of nitrogen organic over the total deposited nitrogen in world areas could vary from 11 to 56 % (Figure 2-1). Atmospheric organic nitrogen have been cl assified in bacterial, particulate, reduced organic N and oxidized organic N. Bact erial N, usually in particles larger than 0.2 m, comes from atmospheric bacteria. Particulate N comes from atmospheric dust

PAGE 35

14 and organic debris in particles between 0.45 m and 1 m in the form of organic nitrates or amine N species. Reduced organic N, mainly as amine N in particles smaller than 0.45 m, is present in oceanic aerosols or terrestrial emissions from agricultural sites. Oxidized organic N is mainly a product of chemical formation of nitric acid esters and diesters and hydroxyl ni tric acid esters, peroxyn itric acid esters and peroxycarboxylic nitric anhydrides (PAN) (Neff et al. 2002). The dissolved fraction of atmospheric organic nitrogen (DON) has been related with bacterial, particulate and reduced organic N forms. ON (%) of TDN 0102030405060 Region 1 2 3 4 5 6 7 Islands, 30% Antarctica, 11% Oceania, 56% South-East Asia, 41% South-Central America, 29 % Europe, 23% North America, 38% Figure 2-1. Contribution of Orga nic Nitrogen to the Total Dissolved Nitrogen Deposited in Areas Around the World (Cornell et al. 2003).

PAGE 36

15 DON has been found in the form of aliphatic amines, amino acids and urea in aerosol, rainwater, fog wate r and sea water samples ar ound the world ( Spitzy, 1990; Gorzelska et al. 1992; Cornell et al. 1998; Chirico et al. 1999; Gibb et al. 1999b; Sommerville and Preston, 2001; Mace et al. 2003a; Mace et al. 2003b; Mace et al. 2003c). Besides its contribution to dry and we t deposition fluxes by itself, it can also offer a secondary source of NOx when is transformed by photochemical reactions under sunlight (Milne and Zika, 1993; Zhang and Anastasio, 2003a). DON comes most likely from direct inj ection from sources including oceanic, agricultural and biomass burning and dust emission, the latter carrying pollen and vegetation debris. Oceanic emissions are a consequence of aero sol generation after bubble bursting with seawater containing amino acids (Neff et al. 2002) and amines from degradation, excretion and metabolis m by marine animals and bacteria (Yang et al. 1994). Agricultural emissions are a consequence of fertil izer use and disposal, e.g. urea (Mace et al. 2003b), and aliphatic amines from animal husbandry operations (Schade and Crutzen, 1995). Biomass burning accounts for emission of a variety of amino acids (Mace et al. 2003a), e.g. humic acid-like compounds such as fulvic acids produced are photolyzed to release disso lved free-amino compounds after biomass burning (Matsumoto and Uematsu, 2005; Chan et al ., 2005). L-glycine, L-serine, Lglutamine, L-alanine, L-threonine, L-arginine L-asparagine were reported as the most common amino acids found during bi omass burning emissions (Chan et al. 2005). Between aliphatic amines the most commonly found are methylamine (MMA), dimethylamine (DMA) and trimethylamine (TMA ), which are expected to have natural

PAGE 37

16 background levels in the air because they ar e produced by a variety of living organisms and are very vol atile (Gronberg et al. 1992). Phytoplankton and zooplankton, for example, produce MA and DMA during grazi ng, ingestion, digestion and excretion and/or by intracellular degradat ion of quaternary amine osmoly tes. Seawater analyses of these amines showed concentration levels in the Arabian Sea from 0-66 nM with higher values for coastal waters, with MMA and DMA being the largest contributors (Gibb et al. 1999b). Estuarine waters from Flax P ond, NY, USA showed that DMA was the most abundant species with concentrations ranging from 25 nM to 180 nM and varying seasonally from lower values at winter to higher values at summer (Yang et al. 1994). Amines could be also linked to domestic wastewater discharges because they are present in human and animal urine (Mitche ll and Zhang, 2001), e.g. DMA is the most prevalent amine in human urine ( Teerlink et al. 1997; Mitchell and Zhang, 2001). Real time measurements using time-of-flight mass spectrometry have found strong amine signals when air is collec ted around busy freeways (Angelino et al ., 2001), rural areas (Beddows et al. 2004) and large feedlots (Murphy et al by Beddows et al ., 2004). Industrial applications of DMA includes synt hesis of agricultural insecticides, insect attractants, and fungicides; vul canization accelerators for sulfur-cured rubber; softeners and lubricants; textile wa ter-proofing agents; tanni ng and dehairing; cationic surfactants; pharmaceuticals; detergents and soaps.

PAGE 38

17 3. INVESTIGATION OF THE UV PHOTOLYSIS METHOD FOR THE DETERMINATION OF DISSOLVED ORGANIC NITROGEN IN ENVIRONMENTAL SAMPLES Comparison of the UVPhotolysis Wi th Other Methods to Determine DON The DON determination is based on thos e methods specifically designed to analyze seawater and some of them have b een used since a centur y ago (Putter cited by Sharp et al. 2002). Methods have evolved in an a ttempt to improve the accuracy of the measurements. The differences between the methods are in the sample treatment and the final form in which nitrogen is detected and measured. Methods can be grouped in two classes: wet oxidation and high temperature oxidation or catalytic oxida tion methods (Cornell and Jickells, 1999). Wet oxidation methods imply the degradation by oxidation of nitrogen organic compounds present in aqueous solution to the ionic nitrogen forms (NH4 +, NO3 and/or NO2 -) by chemical oxidants such as persulfate or peroxide, or by ultraviolet photolysis. High temperature oxidation or catalytic oxidation carries out a thermal combus tion of the aqueous sample and the N-organic compounds are transformed to CO2 and NO which is detected by chemiluminiscence. Sharp et al. (2002) propose a differe nt classification in UV oxidation (UV), persulfate oxidation methods (PO) and high temperature combustion (HTC). This classification agrees with the one proposed by (Bronk et al. 2000).

PAGE 39

18 It is important to highlight that ther e are few comparison studies for seawater samples (Walsh, 1989; Chen and Wangersky, 1993; Bronk et al. 2000; Sharp et al. 2002). Lately an increasing number of studies have appeared for atmospheric samples assuming that the same procedures defined for seawater samples can be extended to atmospheric samples : rainwater, f og water, and aerosols (Scudlark et al. 1998; Cornell and Jickells, 1999; Zhang and Anastasio, 2001; Zhang et al. 2002; Mace et al. 2003a; Mace et al. 2003b; Zhang and Anastasio, 2003a). None of the studies has examined whether the difference in the nature of organic matter found in both natural matrices can a ffect the final conversion to the oxidized forms and therefore the efficiency and accuracy of the methods. As an example of the influence of the variability and complexity of air organic matrix, rare DON losses that led to negative DON values (negative result of subtraction of TDN from DIN) were found for aerosol samples using PO and UV wet oxidation methods and never have been seen in seawater samples (Cornell and Jickells, 1999; Mace and Duce, 2002). No work has been done to adjust or adapt the marine procedures to the complexity, especially in solubility, of the organic matrix present in atmospheric samples. The advantages and disadvantages of each method are summarized here from the few inter-comparison studies done. The goal is to involve as much information as possible in the selection process of the method to measure DON. According to the literature, the oldest method, Kjeldahl dige stion (1883) has not been included in any of the other groups. Ho wever, due to its recent use in the study of rainwater samples from the Tampa Bay area (Hendrix-Holmes, 2002) some literature-

PAGE 40

19 based comments about the efficiency of the method and the reasons for which it was avoided for DON determination are presented. The Kjeldahl method consists of three steps: digestion, distillation and titrati on. Digestion comprises the conversion to NH4 + of primary and secondary organic amine co mpounds by the action of the digestion reactant, usually concentrated sulfuric acid under high temperatures for a period of time. Post-digestion, the resulting “s ludge” is diluted and its pH is raised above 9.5 by the addition of NaOH, followed i mmediately by distillation of the evolved ammonia gas. The distilled ammonia is abso rbed in a boric acid solution. Titration for absorbed NH3 is done with sulfuric acid solution in pr esence of a colored indicator (Tomar, 1999) alternately, the trapped ammonium can be quantified with ion chromatography. The biggest disadvantage of the method lie s in the low final conversion achieved by the digestion process due to the incomplete oxidation of some nitrogen substances such as urea. Nitrogen present in forms other than amines or amides are not efficiently converted to NH4 +, but may be released instead as nitrogen oxides or N2, which are not kept in solution (Cornell et al. 2003; Zhang and Anastasio, 2003a). Bronk et al. (2000) based on seawater analyses, affirm that the Kjeldahl method has high blanks and resulting low precision. Gra sshoff (1983) (cited by Walsh, 1989) also affirms that the Kjeldahl digestions are tedious, time consuming and subject to contamination, exhibiting poor precision. As one of the most common wet oxidati on methods, persulfate oxidation (PO) (Solorzano and Sharp, 1980) oxi dizes organic matter to NO3 and NO2 ions by reaction with potassium persulfate in a strong alkali ne solution, which is autoclaved for a period

PAGE 41

20 of time. After that, hydrochloric acid is added to dissolve any precipitate and the solution is buffered at a pH equal to 8.5. In order to measure the NO3 concentration, NO3 is reduced to NO2 by passage of the sample through a column of copperactivated cadmium, and NO2 is analyzed colorimetrically after NO2 reacts with a mixture of sulfanilamide and N-(1-n aphthyl)-ethylenediamine (Parsons et al ., 1984 cited by Cornell and Jickells, 1999). Despite the fact that th e PO method has shown the highest percentages of recovery of most N-organic compounds for standard samples in comparison studies of PO, UV and HTC methods (Bronk et al. 2000) and greater oxidati on efficiency relative to UV oxidation of rainwater samples (Scudlark et al. 1998), its most inconvenient and undesirable disadvantage is the extensiv e sample handling. Fairly high blanks, difficulties in chromatographic ion measurem ents, chances for additional mistakes and variability due to the NO2 colorimetric analysis, and copper-cadmium NO3 reduction are additional disadvantages to the PO method. The PO method has a high contamination potential due to the large oxi dizing reagent addition (2 ml to 15 ml sample), but requires simple lab skills to be applied, allows a number of processed samples limited only by the available h eat source and also has the lowest instrumentation cost (Bronk et al. 2000). The PO method was discarded for the proposed WSON analyses considering the evident relation betw een the extensive handling of the samples, the DON losses registered by Scudlark et al. (1998) and Cornell and Jickells, (1999) for atmospheric samples, less efficient oxidation of aerosol organic matter, and losses of oxidized nitrogen gases reported by the last cited author.

PAGE 42

21 Armstrong et al. (1966) proposed the UV photolys is method as an improvement of the digestion processes us ed for the dissolved organi c matter determination in seawater. The method implies the photolysis of the aqueous-phase organic nitrogen to a mixture of NH4 +, NO2 and NO3 ions by exposing the sample to UV irradiation from mercury lamps during certain periods of time (usually from 18-36 hours) under a controlled ambient temperature. The resulti ng inorganic nitrogen can be detected and measured using ion chromatography. Different arrays have been employed for the batch photolysis reactor, for example, a single lamp placed on the center of the reactor (Armstrong et al. 1966; Walsh, 1989), or multiple lamps attached to the inside reactor wall (Zhang et al. 2002). Some oxidation reagents su ch as persulfate and hydrogen peroxide have been added to the sample (Bronk et al. 2000) without a conclusive improvement of the oxidation efficiency but with higher blank values (Manny et al. 1971 ; Scudlark et al. 1998; Walsh, 1989). UV photolysis offers low conversion for sulfur-containing amino acids (Cornell et al. 2003) and substances with N-N and N=N bonding, but a better conversion of substances with heterocyclic-N and HC=N bonding in relation with the PO method (Scudlark et al. 1998; Walsh, 1989). Bronk et al (2000) considered that the high co st and effort of building the UV photolysis reactor were important disadvant ages, too. Variable values for conversion were related with deficient UV lamp func tioning, therefore, (B ronk and Ward, 2000) recommended monitoring of the UV lamp spectr a to assure it agrees with the expected spectrum.

PAGE 43

22 The High Temperature Combustion method transforms all organic compounds present in aqueous samples to CO2 and NO by combustion in a pyro-reactor furnace where ultra pure oxygen is used for combusti on. Inside the reactor a chemiluminescent reaction occurs between ozone and nitric oxide (NO) yielding n itrogen dioxide (NO2) and oxygen. This reaction produces light at inte nsity proportional to the mass flow rate of NO. The released energy, is then pr oportional to the mass of chemically bound nitrogen in the sample(Walsh, 1989). The HT C method exhibits ex cellent linearity and good precision for total dissolved nitrogen dete rmination in seawater, and according to an inter-comparison study (Walsh, 1989) s howed a high and comparable recovery of standard compounds in rela tion to the UV photo-oxidation method. Advantages that must be highlighted are the minimal sample handling that eliminates almost completely any chance of contamination, a nd the non-existent in terference of ioni c substances in the final N measurement. Other advantages can be related to the small volume required to make the measurement and the large num ber of samples that can be processed. Disadvantages are a possible loss of accuracy due to the variati on in the ozone flow rate, high cost of the equipment and high degr ee of lab skills needed to perform the method (Bronk et al. 2000). A single reference of the application of the HTC method for atmospheric samples was found, affirming that the highest results for rainwater DON were generated by the HTC method in a comparison study, implying that PO and UV methods underestimated DON quantities. (Cape et al ., 2001 cited by Cornell et al. 2003). Adding to these disadvantages is the fact that no reference for DON determination in atmospheric particles by the HTC method was found. Considering that

PAGE 44

23 the main conclusion extracted by (Sharp et al. 2002) from an inter-laboratory study was that the three methods give comparable results with no systematic difference for TDN concentration for seawater samples from PO, UV and HTC methods, the HTC method was discarded. The minimal sampling handling, the signi ficantly higher DON recoveries from aerosol particles in relation with those provi ded by the PO method (Cornell and Jickells, 1999) as well as the consistent data re ported by (Zhang and Anastasio, 2001) for fog water (Zhang et al ., 2003) and for fine atmo spheric particles (Zhang et al. 2002) were used as criteria for the selection of the UV-photolysis method to measure DON in atmospheric samples. Materials and Methods Chemicals Urea standard solutions were prepared from a 40% (w/w) urea solution (SigmaAldrich, Inc.) and solid urea (ACS reagent, ACROS, Inc.), and amino acid standard solutions from an L-amino acids and glyc ine kit (Sigma-Aldrich, Inc.) containing individual solid portions of 22 standards. Methylamine standard solutions were made from a 40% (w/w) methylamine solution (ACS reagent, Sigma-Aldrich, Inc.). Calibration curves were made from dilutions of indi vidual 1000 mg l-1 NH4 +, NO2 -), and NO3 standard solutions (UltraScientific, In c), while check standards were prepared

PAGE 45

24 from 100 mg l-1 anion (Cl-, NO2 -, NO3 -, PO4 -, and SO4 -2) and cation (Na+, NH4 +, K+, Mg+2, and Ca+2) mixtures (Inorganic Ventures, Inc. and Certiprep, Inc., respectively). All solutions were prepared with pure reagent and doubl e de-ionized water (DDW) to reach DIN concentrations be low the detection limit of the ion chromatographic (IC) method. Double de-ionized water (> 18.0 M -cm and total organic carbon content (TOC ) less than 5 ppb) was obtained from a Nanopure DiamondTM Analytical Ultrapure Water System Model D11901 (Barnstead, Inc.). The resin cartridge used was ultra-low organi cs, Type I, with de-ionized water feed. Solutions were kept in the refrigerator (4 oC) protected from light. Chromatographic Analyses NH4 +, NO2 and NO3 that comprise the DIN fraction, and methylamine, were determined using a Dionex 600 ion chromatograph (IC) equipped with a CD25A conductivity detector and an AS50 autosampler. Anions were separated by an IonPac AS9-HC analytical column (4 x 250 mm) and ASRS-Ultra 4-mm suppressor using sodium carbonate/bicarbonate solution as an eluent. Isocratic separation or gradient separation for methylamine (Rey, 2001) of cations was done by an IonPac CS16 analytical column (4 x 250 mm) kept at 40 oC and a CSRS-Ultra II 4-mm suppressor. The eluent for cations was a methanesulfonic ac id solution and its flow rate was fixed to be 1 mL min-1. The injection volume was set to be 100 L.Check standards were run after every ten samples and were within 5 % of their prepared con centrations (within 10 % for concentrations near the LOD).

PAGE 46

25 DON Analyses Solutions and laboratory bla nks of acidified DDW were irradiated in a Rayonet Photochemical Chamber Reactor Model RPR-100 (Figure 3-1) with a 36-sample merry-go-round unit (Southern New England C o, Inc.). The reactor was equipped with 16 mercury low-pressure lamps (RPR-253.7 nm ), each one with a power of 35 W. The samples were rotated at a speed of 5 rp m during the UV photolysis process and kept approximately 3 cm from the lamps. For sa fety reasons, the reactor was placed in a temperature-controlled environment under th e laboratory hood duri ng operation. The temperature of the air surrounding the sample s was monitored using a Type T (copper – constantan) thermocouple. Solutions and laboratory bla nks were acidified to pH 2 in batches immediately before the irradiation process with less than 20 l of concentrated sulfuric acid. The solution pH was measured using an Accu met AR50 dual channel pH/ion/conductivity meter (Fisher Scientific, Inc.) and a potassi um chloride electrode (Corning, Inc.). To eliminate sample volume as a sour ce of variation, 5.0 ml aliquots were delivered to 8.0 ml quartz tubes. Quartz tubes were sealed with customized PTFE stoppers wrapped with PTFE tape. The weight of the quart z tubes filled with sample was measured before and after irradiation to account for possible evaporative losses. Proper precautions regarding contamination of all glassware, es pecially the quartz tubes, were taken according standard methods.

PAGE 47

26 Nitrogen atoms present in organic molecule s can be released as inorganic forms of NH4 +, NO2 or NO3 -; therefore the measured DON is the difference between the molar sum of the three ions before and after irradiation (DONmeasured). Figure 3-1. Photochemical Chamber Reactor Model RPR-100 (Southern New England Co, Inc.). Field Sampling and Sample Processing The sampling campaign was conducted at an atmospheric deposition monitoring site adjacent to Tampa Bay at the eastern end of the Gandy Bridge in Tampa, Florida (27.78 oN, 82.54 oW, Figure 1-1). A Rupprecht and Pa tashnick Dichotomous PartisolPlus Model 2025 sequential air sampler (Appe ndix B) was used to collect 24-hr

PAGE 48

27 integrated PM2.5 and PM10–2.5 ambient air samples at a to tal flow rate of 16.7 l min-1 onto PTFE 2 m pore size, 47 mm diameter filter s (Whatman, Inc.). Two magazines (i.e., precision-manufactured filter holders) with filters for samples and field blanks were transported daily to and from the site without refrigeration. To account for contamination from filter storage, handling, a nd transport, field blanks were stacked in the magazines but unexposed to air flow during sampling. Samples, field blanks, and lab blanks were extracted under sonication in an isothermal bath at 40 oC with 20 ml of DDW into amber glass bottles without headspace. Extraction bottles were carefu lly sealed with Teflon tape to avoid contamination from the water inside the s onicator bath and evaporative losses during storage. Extracts were stored in the dark at 4 oC and brought up to room temperature prior to analysis. Field blank concentrations were determined every two days to check for possible contamination from sampling handling. Field blanks were processed identical to the samples. Nitrogen detected in un-irradiated extracts of field blanks was subtracted from DIN concentrations, while nitr ogen measured in irradiated extracts was subtracted from TDN. DON concentrations were calculated as the difference between TDN and DIN after the blank corrections. Blanks and General Performance To account for possible organic nitrogen contaminants, acidified blanks of double deionized water (DDW) were irradiated a nd analyzed as the rest of the samples. The average DON in blanks was 0.9 0.5 M-N (n = 30), which agreed with those previously reported (Cornell and Jickells, 1999;Zhang and Anastasio, 2001; Mace and

PAGE 49

28 Duce, 2002). The detection limit (LOD) was 1.6 M-N, defined as three times the standard deviation of the DON concentr ation measured in the DDW blanks. The influence of temperature during phot olysis has not been well established. Experiments for atmospheric samples have been done using temperatures under 50 oC (Zhang and Anastasio, 2001; Zhang et al. 2002), between 50-55 oC (Cornell et al. 2001), between 75-85 oC (Carrillo et al. 2002), at 85 oC (Mace and Duce, 2002) and under 90 oC (Scudlark et al. 1998). The maximum temperature reached during sample illumination is a function of the UV lamp pow er and type (low-, mediumor highpressure) and the type of cooling system us ed. Higher temperatures have been used to reduce irradiance periods and increase conve rsion rates; however, lower temperature and longer irradiance periods were chosen to reduce the evaporati on and loss of volatile species in samples (e.g. if a 87 M-N methylamine solution was left exposed to clean air at 20 oC, volatilization of the amine will cause a 11 % reduction on the initial concentration. This is especially importa nt for very volatile species such as methylamine. At the sample extraction a nd solution irradiation temperature of 40 oC, the vapor pressures of water and methylamin e are 7.4 kPa and 580 kPa, respectively). By controlling the room temperat ure and air flow around the reactor, temperature fluctuations inside the reactor were kept to a minimum. The steady state temperature measured in the reactor near the samples was approximately 44 oC. Water evaporation losses were minimal: 0.03 0.05 % (n=14). Methylamine evaporation losses were within 5 %, based on IC analyses of non-irradiated cont rol samples held at room temperature on the IC auto sampler.

PAGE 50

29 pH and Irradiance Period 22 Factorial Experiments The conversion by UV photolysis of organi c molecules to their inorganic form can be affected by temperature, irradiance period, pH, and concentration of inorganic and organic species (Gustafsson, 1984; Wals h, 1989). To measure the extent of this conversion, the conversion efficiency () was defined as the molar ratio of the DONmeasured to the initial organic nitrogen concentration of the standards ( DONstandard), as shown in Equation 3-1. standard measureDON DONd Equation 3-1 To find the optimal parameters for the use of UV photolysis with DON species, a 22 factorial experiment was used with samp le pH and irradiance period as the factors for solutions containing urea, am ino acids (alanine, aspartic ac id, glycine and serine), or methylamine. pH and irradiance period were assumed to be the most important operational parameters of the method. For each substance the factorial design is based on two and three replicates per combination, and there was no need of extra replicates because the standard deviation of the convers ion efficiency values was not higher than 2.9 %, showing good reproducibility for the analyses. The pH of the treated solution has show n to be an important factor in the conversion of species such as urea. The role of pH in photolysis can be associated with its influence on the formation rate of hydroxyl radicals, candidates responsible for the oxidation of organic matter (Arakaki et al. 1999). It has been re ported that pH values between 7 and 9 allow the decomposition of almost all organic nitrogen species, but lower pH values between 4 and 6 help decom pose more refractory species such as urea

PAGE 51

30 (Golimowski and Golimowska, 1996). The pH had only a slight effect during UV photolysis with peroxodisulpha te of urea solutions (Roig et al. 1999), but could increase the hydrolysis rates through whic h urea is normally transformed to carbon dioxide and ammonia (Cornell et al. 1998). It was found that for the duration of the experiment, urea hydrolysis rates were neglig ible: for standard ur ea solutions of 10, 20, and 72 M-N, acidified urea samples did not releas e ammonium, nitrite, or nitrate in the absence of UV light. For the treatment of atmospheric samples, irradiance periods of 2 hr (Mace and Duce, 2002), 3 hr (Carrillo et al. 2002), 12 hr (Scudlark et al. 1998), 18-22 hr (Walsh, 1989), 24 hr (Zhang and Anastasio, 2001; Zhang et al. 2002) have been used with good conversions. Another factor influencing conversion duri ng photolysis is the concentration of organic nitrogen. Incomplete conversions ha ve been found for sta ndard solutions of urea at concentration le vels higher than 50-60 M-N (Cornell and Jickells, 1999); (Mace and Duce, 2002) and for 2-75 M-N glycine, EDTA, antipyrine (1,2-Dihydro1,5-dimethyl-2-phenyl-3H-pyrazol-3-one), ur ea, and humic acid in an organic-rich solution (Bronk et al. 2000). The same decreases in th e conversion of urea, EDTA and antipyrine were seen in seawater samples when their spiked concentrations were increased from 10 to 40 M (Walsh, 1989). Irradiation power may change over the lifetime of the lamp and depends on such fact ors as the spectrum of the UV lamp, the bulb temperature and the arc current. It was assu med that this variable did not change in the performed experiments. The temperature inside the reactor was checked constantly

PAGE 52

31 in order to detect any change in the i rradiation power. Decreases on the conversion efficiency have been associated with changes in the lamp spectrum (Bronk et al. 2000). Between the DON selected species, urea was chosen because it has shown a strong refractory behavior under UV photolys is ( Golimowski and Golimowska, 1996; Bronk et al. 2000) and because of its presence in rainwater and aerosol samples (Timperley et al. 1985; Cornell et al. 1998; Cornell et al. 2001; Mace et al. 2003b). Amino acids were included because of their ubiquity in atmospheric samples (Scheller, 2001; Zhang and Anastasio, 2003b). Alanine, aspa rtic acid, glycine, and serine were selected to represent an important frac tion of the free-amino compounds as found in aerosol samples (Mace et al. 2003a; Mace et al. 2003c; Zhang and Anastasio, 2003b). Methylamine as found in aerosol an d fog water samples (Van Neste et al. 1985; Gorzelska et al. 1992; Zhang and Anastasio, 2003b). It is the simplest substance of the amine family and although an optimizati on of its photochemical conversion does not guarantee conversion of more complex amines, its higher volatility made possible quantifying evaporative losses during the experiments. To fix the effect of concentration of DON for the 22 factorial experiments, solutions were prepared to a target concentration of approximately 80.0 M-N, and final concentrations were: urea, 71.4 M-N; amino acids, 83.6 M-N; and methylamine, 87.0 M-N. Concentrations for individual amino acids were L(+)-alanine, 25.1 M-N; Laspartic acid, 18.5 M-N; glycine, 21.5 M-N; and L-serine, 18.4 M-N. The method optimization occurred prior to field sampling, so target concentrations were chosen in attempt to bracket the averag e aerosol concentrations re ported in the literature (e.g.,

PAGE 53

32 extracts from bulk aerosols samples from the Eastern Mediterranean had an average DON equivalent to 58 M-N (Mace et al. 2003c), while extracts from California PM2.5 samples had an average DON equivalent to 127 M-N (Zhang et al. 2002). Although afterwards much lower levels were found in aerosol samples, the complete conversion of DON species at higher levels made unlik ely the presence of negative biases in the presented results. For the 22 factorial experiments factor levels were chosen with a high separation between them to maximize the effect of the variables on the conversion and thus improve the identification of their influence. The factor levels for the irradiance period were 8 hr and 24 hr for amino acids and methylamine solutions, and 16 and 24 hr for urea solutions. For urea, the lower irradiance period was increased to assure measurable conversion efficiencies. The le vels of irradiance period repr esented a trade-off between higher sample conversion efficiencies and the loss of volatile consti tuents. The selected pH levels were the original sa mple pH (from 5 to 9), and pH 2 after sample acidification with concentrated sulfuric acid (94-98 %). Below pH 2, the SO4 -2 chromatography peak overwhelms the NO3 peak and thus interferes with NO3 determination. For urea solutions, maximum and minimu m conversion values were 72 % and 8 %, respectively. The maximum conversion valu e reached for a solution of pH 2 agreed with the value presented by (Bronk et al. 2000) for a urea solution in de-ionized water with 75 M-N. A conversion efficiency of approximately 70 % revealed the recalcitrance of urea to UV photolysis (W alsh, 1989; Golimowski and Golimowska, 1996; Bronk et al. 2000).

PAGE 54

33 The resistance of urea to UV photolysis has been observed since the first application of the method to DON determination, Armstrong reported back in 1966 that urea was partially oxidized after irradiance at conditions found adequate for other substances (Armstrong et al. 1966). Urea is an amide co mpound, and amide functional groups have shown to be specially resist ant to both photoche mical and microbial degradation (Buffam and McGlathery, 2003). Although a reaction mechanism to explain the special behavior of this substance is not reported in the l iterature, it has been suggested that urea decomposes with an oxygen-related mechanism which is slowed by the decrease in the relative oxygen concentra tion that occurs during irradiance. Urea solutions increase their conversion efficien cy values when irradiance was continued with the addition of hydrogen peroxide (Walsh, 1989). For urea solutions at prepared at higher concentrations (714.3 M-N), urea disappearance under UV-photolysis was monito red using a colorimetric method (the reaction with diacetyl monoxime in the pres ence of thiosemicarbazide and ferric ions) (Hassan et al. 1995; Goeyens et al. 1998). The detection level of 71.4 M-N equivalent to 1000 g-N l-1 for this colorimetric technique was too high to use for the chosen experimental urea concentrations. For amino acids, maximum and minimum conversions were 97 % and 69%, respectively; the difference between these valu es indicates smaller factor effects, and also a less refractory behavior of amino aci ds by UV photolysis as compared with urea. The same behavior was observed for methylamine, for which maximum and minimum conversions were 84 % and 5 %, respectively.

PAGE 55

34 At pH 2 and 24 hr irradiance period a ll the solutions of the DON selected species showed maximum values for the conv ersion efficiency when analyzed with UV photolysis. Even when this could had been us ed to define the optimal settings for the application of the method, the direction of the factor effect over the conversion efficiency was confirmed through analyses of variance (ANOVA) and multi-linear regression. Evidence of the significance of the factor effects were studied using p-values per factor as presented in Table 3-1 and the linear regressions pr esented in Equations 3-2 to 3-4. It was found a significant effect (p < 0.05) of the solution pH on the conversion of urea, amino acids, and methylamine by UV phot olysis. The irradiance period effect was significant over the conversion efficiency of all selected DON species except for urea. The irradiance period did not show a significan t effect over the conve rsion efficiency of urea because its lower level was chosen to be 16 hr instead of 8 hr. The time effect was not caught due to the proximity of the fact or levels. The interaction of pH and irradiance period (pH time) had a significant infl uence on the conversion of methylamine. In all cases, the regression coefficient of pH was negative meaning that the lower the pH the more complete was the c onversion efficiency. Re gression coefficients for irradiance period were positive meani ng that the longer the irradiance period the more complete was the conversion efficien cy for amino acids and methylamine.

PAGE 56

35 Table 3-1. p-values from Analyses of Vari ance of the Conversion of Organic Nitrogen Solutions under UV-Photolysis. p-values* Factor Urea (71.4 M-N )Amino acids (83.6 M-N)Methylamine (87.0 M-N) pH < 0.0001 0.0032 < 0.0001 Time 0.4893 0.0010 < 0.0001 pH Time 0.1545 0.4483 < 0.0001 *The effect is significant when p<0.05 for a confidence level =95%. Urea: time pH time pH 1 0 Equation 3-2 Amino acids: time pH time pH 09 0 4 5 0 88 Equation 3-3 Methylamine: time pH time pH time pH 3 0 9 2 6 3 3 35 Equation 3-4 The pH of the solution proved to be the most important operational parameter of the UV photolysis method but its role on reac tion mechanisms cannot be defined by this study. Transformation of DON to DIN can occur by indirect photo-oxidation as a result of photo-formed reactive species (Anastasio and McGregor, 2000) and/or by reaction with hydroxyl radicals, singlet molecular oxyge n, or peroxyl radica ls (Anastasio and McGregor, 2001; McGregor and Anastasio, 200 1). Reaction mechanisms are not the same for all species and are not availabl e for the DON selected species. A mechanism proposed for the decomposition of glycine to NH4 +, one of the selected amino acids, was driven by hydroxyl radicals (Berger et al. 1999; Karpel Vel Leitner et al. 2002). If

PAGE 57

36 that were the case, it is known that in acidi c conditions, the glycine decomposition starts with the OH radical attack on the glycine an ions and ends with main production of ammonium ions. The OH reaction with the glycin e anionic form is faster at pH levels lower than the glycine pKa. Vel Leiner et al reported that at pH 8 the OH radical attack is 2.8 times faster than at of the neutral form (pKa=9.8). Glycine decomposition by direct photolysis is negligible(Karpel Vel Leitner et al. 2002). Another possible explanation for the pH role was found in th e literature from experiments of conversion of fogwater and aerosol or ganic nitrogen samples to NH4 +, NO3 and NOX under simulated sunlight. Although there was a non significant pH effect on the NH4 + formation rate, rates for N(III) destruction and NO3 formation at pH 3.2 were more than 6 times faster than at pH 6.4. Authors attr ibuted the effect to the increase in HNO2 concentration, a species more phot o-chemically reactive than NO2 (Zhang and Anastasio, 2003a). Proportions of Ammonium, Nitrite, and Nitrate Formed from DON The average distribution of inorganic nitrogen after th e irradiation of organic nitrogen solutions under 200 M-N is presented in Table 32. The consistency of these results for urea, amino acids, and methylamin e offered an opportunity to predict the final distribution of inorganic nitrogen fo r known mixtures of these organic nitrogen compounds.

PAGE 58

37 Table 3-2. Average Percentage of Inorga nic Nitrogen Produced Per Species After UVPhotoxidation at pH=2 and 24-hr Irradi ance Period at Concentrations Under 100 M-N. Substance n NH4 + (%)NO3 (%) Urea 1751.7 5.348.2 5.6 Methylamine 1096.1 2.03.9 2.0 Amino acids 9 96.8 3.23.2 3.2 Amine Mixture2 96.0 0.64.0 0.6 For the experiments, inorganic nitrogen formed by the UV photolysis of urea, amino acids and methylamine was dominated by NH4 +. NO3 was produced in an appreciable quantity when urea was irradiated by UV at low pH. NO3 also appeared when methylamine was irradiated for only 8 hr at high pH. This shows the existence of different decomposition mechanisms: the pho tolysis of amino acids and methylamine released mainly NH4 +, but urea released a near equimolar ratio of NH4 + and NO3 -. Evidence supporting the decomposition to NH4 + of amino acids, amines and other similar species such as humic acids can be found elsewhere (Gustafsson, 1984; Berger et al. 1999; Tarr et al. 2001; Wang et al. 2000a; Zhang and Anastasio, 2003b). Methylamine is the simplest amine of the aliphatic family; therefore mixtures of other amines commonly found in atmospheric samp les were later tested. A mixture of 75.4 M-N of dimethylamine, 26.6 M-N of diethanolamine and 97.8 M-N methylamine generated after UV photolysis more than 95% of NH4 + following the same tendency of pure methylamine solutions.

PAGE 59

38 If all the organic nitrogen molecules decompose with the same conversion value and produce the same ammonium and nitrate ra tio, then the fraction of each inorganic nitrogen species generated from a mixtur e could be predicted by Equation 3-5, where [Uo], [AAo] and [MAo] are the initial concentratio ns of urea, amino acids, and methylamine, respectively, in the mixture; and yiU, yiAA and yiMA are the fractions of the inorganic nitrogen species i produced by photol ysis. Differences between the predicted and actual concentrations of organic nitrogen were with in 7 % at approximately 10 MN but less than 1 % from the approximately 100 M-N level. The consistency of the product to reactant ratios fo r mixtures of these com pounds treated under the same laboratory conditions is remarkable An interpretation of these results, applicable to just the specific set of six organi c nitrogen compounds, is that each substance was converted from its organic to its inorga nic nitrogen form with the same efficiency and mechanism whether as a single solute or in a mixture. Wh ile these results are not directly applicable to complex atmospheric samples, this approach can be used to eval uate organic-organic or organic-inorganic interactions, which may cause variations in the UV photolysis method performance (Cornell and Jickells, 1999). 100 % o o o iMA o iAA o iU o iMA AA U y MA y AA y U DIN Equation 3-5 Even though the UV photolytic appro ach is purely operational, these observations may provide comparative experi mental data for researchers involved in determining mechanistic pathways of photolytic and photo-assisted DON decomposition.

PAGE 60

39 Application of the UV Photolysis Method to Aerosol Samples Between November 2004 and April 2005, 32 samples of particulate matter (PM2.5 + PM10-2.5 = PM10) were analyzed for DIN and DON showing average concentrations of 78.1 29.2 nmol-N m-3 and 8.3 4.9 nmol-N m-3 for DIN and DON, respectively. The ratio between DON and to tal dissolved nitrogen (TDN=DIN+DON) was 10.1 5.7 %. Although concentrations were higher during winter months compared with spring months, the contribu tion of the organic fraction to the total nitrogen in PM10 was approximately 10 % over the sampling period (Table 3-3). Table 3-3. DIN and DON Concentrations (nmol-N m-3) in PM10 Samples Collected at the Gandy Bridge Monitoring Site, Tampa, FL. Sampling Day PM10-DONPM10-DIN-NH4 +PM10-DIN-NO3 PM10-DIN 4-Nov-2004(outlier) 29.8 36.9 23.1 62.4 5-Nov-2004 8.5 59.4 8.6 68.1 6-Nov-2004 11.5 73.4 14.2 87.5 7-Nov-2004 13.4 64.6 9.9 75.1 8-Nov-2004 9.9 94.2 22.6 116.8 25-Jan-2005 4.1 34.1 18.0 52.2 26-Jan-2005 5.9 38.3 25.2 63.5 28-Feb-2005 6.2 28.5 11.3 39.8 1-Mar-2005 5.0 16.5 16.9 33.4 2-Mar-2005 8.8 21.6 5.0 26.7 12-Mar-2005 6.6 20.7 20.0 40.8 13-Mar-2005 5.2 58.1 43.9 102.0

PAGE 61

40 Table 3-3. (continued) Sampling Day PM10-DONPM10-DIN-NH4 +PM10-DIN-NO3 PM10-DIN 14-Mar-2005 14.8 134.0 28.0 162.5 30-Mar-2005 9.9 47.7 41.1 88.8 31-Mar-2005 8.7 62.4 32.2 94.5 1-Apr-2005 12.9 53.2 33.6 86.7 3-Apr-2005 3.4 39.9 15.1 55.0 4-Apr-2005 9.5 74.4 25.9 100.3 5-Apr-2005 5.2 57.8 27.9 85.7 6-Apr-2005 5.9 52.5 40.7 93.2 9-Apr-2005 11.1 114.4 23.0 137.3 10-Apr-2005 6.9 43.0 46.3 89.3 11-Apr-2005 7.5 28.3 36.9 65.2 12-Apr-2005 7.1 25.5 37.7 63.2 13-Apr-2005 5.8 30.0 27.1 57.1 14-Apr-2005 9.3 54.1 18.5 72.6 15-Apr-2005 5.6 36.7 31.2 67.9 16-Apr-2005 2.2 26.4 32.7 59.0 17-Apr-2005 5.6 29.7 32.1 61.8 18-Apr-2005 5.2 34.3 47.9 82.2 19-Apr-2005 6.4 55.8 44.8 100.7 20-Apr-2005 6.6 48.7 60.0 108.7 Average 8.3 49.8 28.2 78.1 Standard deviation 4.9 26.5 13.0 29.2

PAGE 62

41 The predominance of NH4 + and NO3 in the fine and coarse particle fractions, respectively, agreed with the results of size-segregated particle measurements previously made at the Gandy Bridge monito ring site. By comparison of experimental and modeled data,(Campbell et al. 2002) predicted the formation of solid phase ammonium sulfate (mainly in the fine mode ) and sodium chloride, sodium nitrate and gypsum (mainly in the coarse mode). Organic nitrogen concentrations for the PM2.5 fraction were low, relative to the 18.9 13.6 nmol-N m-3 in the same fraction observed in northern California (Zhang et al. 2002). Compared with measurements made in Hawaii with a cascade impactor (0.2 10 m diameter particles), the Tampa Bay PM10 samples were higher in DON than the average of 3.3 2.0 nmol-N m-3 seen under trade wind conditions and lower than the average of 28.5 25 nmol-N m-3 collected without th e trade winds (Cornell et al. 2001). PM10 samples collected in the Amazon Basi n, Brazil, were richer in DON during the dry season with an average co ncentration of 61 67 nmol-N m-3 (below detection to 280 nmol-N m-3) and poorer in DON in the wet season with concentrations of 3.5 4.6 nmol-N m-3 (below detection to 18.0 nmol-N m-3) (Mace et al. 2003a). On the other hand, a much lower average con centration of 3.6 5.7 nmol-N m-3 (below detection to 18.8 nmol-N m-3 ) was detected in bulk aerosol s from Tasmania, Australia (Mace et al. 2003b). It was found that the DON PM2.5 fraction contributed on the average 79.1 18.2 % of the total organic nitrogen found in the aerosol samples. The higher concentration

PAGE 63

42 of organic nitrogen on the fine mode filter i ndicates gas to particle transformations or subsequent enrichment of the small partic les with gas-phase or ganic nitrogen. A higher contribution of the PM10-2.5 would have been associated w ith natural sources of organic nitrogen such as plant detrit us or sea spray, for example. Evidence for these hypotheses was found by statistical analys is of 24-hr integrated DI N and DON concentrations in PM10 samples and average meteorological conditions using multi-linear regression techniques (Caldern et al. 2006a). The distribution of inorganic nitroge n produced by exposure of the aerosol samples to UV light showed an average NH4 + fraction of 73.3 11.5 % and 57.6 34.2 % and NO3 fraction of 27.3 11.2 % and 21.0 18.7 % in PM2.5 and PM10-2.5 fractions, respectively (Figure 3-2). Figure 3-2. Inorganic Nitrogen Distribution of the Organic Nitrogen Fraction After UVPhotolysis of Aerosol Samples.

PAGE 64

43 As NH4 + is the highest contributor, th is could indicate organic nitrogen compounds similar to amino acids and amines. The presence of dimethylamine in PM2.5 extracts was confirmed by ion chromatography (Caldern et al. 2006b). The existence of NO3 could also indicate a contribution from urea. NO3 from the UV photolysis of aerosol organic nitrogen was c onsistently present in the PM2.5 fraction at concentrations above the LOD. Percentages higher than 100 % are generated by negative numbers related to the disappearance of NO2 after irradiation (lower concentrations after irradiance leading to negative concentrations and therefore negativ e ratios with total DON) due to possible conversion to NO3 by photochemical reactions. The ratio between the photo-produced NH4 + and NO3 is relatively constant during all the sampling period and could indicate dominant so urces of DON in the area, for example, the sea surface. Effect of DON Mixtures and DIN/DO N Ratio on Conversion Efficiency After field sampling, the method appli cability was tested at DON and DIN concentrations similar to those found in sa mples. The experiments were done with replicate equimolar mixtures (n = 5) of these compounds at 10.1 M total organic nitrogen, as well as with replicate solutions of 19.9 M-N urea (n = 5), 9.9 M-N urea (n = 5), 5.4 M-N amino acids (n = 4), and 10.1 M-N methylamine (n = 5). The potential for organic – organic interactions was studied with replicate mixtures (n = 9) of 34 M-N urea, 33 M-N amino acids (the amino acid concentrations were: L(+)alanine, 10 M-N; L-aspartic acid, 7 M-N; glycine, 9 M-N; and L-serine, 7 M-N), and 34 M-N methylamine, or 101 M total organic nitrogen. To account for the

PAGE 65

44 variation of experimental conditions and te st the method reproducibility, two blocks of 100 M-N mixtures were irradiated 7 d ap art. Each block was analyzed immediately after the solutions were removed from the reactor. Every replicate came from the same batch of solution and was treated in the same way in both blocks. Conversion values per block were not significan tly different (two-sided unpaired t-test, p=0.16, n=9). Mixtures at two con centration levels 10 M-N and 101 M-N gave average conversion efficiencies of 103 2 % and 100 2 %, respectively (Table 3-4). These values confirmed that the total UV-photon fl ux during the irradiance period is enough to assure complete conversion of organic nitrogen under 33 M-N. Concentrations for each organic nitrogen compound in mixtures were under individual limits for complete conversion found during single-substance expe riments. The low standard deviations confirm that adequate precision was seen for individual compounds. Table 3-4. Conversion Efficiencies of Urea, Amino Acids, and Methylamine Solutions After UV Photolysis at pH 2 for 24-hr. Solution NH4 + ( M-N)* NO3 ( M-N)* DON ( M-N) (%) 19.9 M-N Urea (n=5) 10.2 0.2 9.6 0.03 19.7 0.3 99.2 1.3 10 M-N Urea (n=5) 5.2 0.1 4.8 0.1 10.0 0.2 100.3 1.7 5 M-N Amino Acids (n=4) 5.0 0.3 0.4 0.1 5.4 0.3 100.7 5.1 10 M-N Methylamine (n=5) 9.2 0.3 0.5 0.1 9.7 0.3 96.4 3.3 10 M-N Mixture (n=5) 9.2 0.2 1.2 0.04 10.4 0.2 103.2 2.2 101 M-N Mixture (n=10) 85.8 2.0 18.4 0.8 104.0 2.0 100.3 2.0 These are the concentrations of the DIN species generated after photolysis

PAGE 66

45 Finally, the effect of the DIN DON-1 ratio on the conversion efficiency was studied with replicate mixtures of 10 M-N urea (n=3) with five DIN concentration levels changing from 0.3 to 104 M-N (Table 3-5). For these experiments, the standard solutions were adjusted to pH 2 and irradiated for 24 hr. The effect of the DIN DON-1 ratio on the conversion efficiency of the 10 M-N urea solutions was found to be non significant (p=0.065) at 95 % confidence level, but did increase the conversion efficiency beyond 100 % with the largest DIN DON-1 ratios showing the highest conversion efficiencies. The average conversion for all replicates and factors was 108 5 %. Because the Tampa Bay PM10 samples showed an average DIN DON-1 ratio of 9 (DON TDN-1 in PM10 is 10.1 5.7 %) they ar e likely biased high. Table 3-5. Effect of the DIN DON-1 Ratio on the Conversion of a 10 M-N Urea Solution. DIN DON-1 DIN-NH4 + ( M-N)* DIN-NO3 ( M-N)* DON ( M-N) (%) 0.03 0.1 0.2 10.5 105.0 0.03 0.1 0.2 10.7 106.6 0.03 0.1 0.2 10.6 106.0 0.6 0.6 5.5 10.6 106.0 0.6 0.7 5.3 10.6 106.4 0.6 0.6 5.4 10.6 106.0 3.9 5.0 33.9 10.5 104.6 3.8 4.6 33.4 10.4 103.6 3.8 4.8 33.4 10.5 104.9 These are the DIN concentrations pr esent before irradiance of solutions

PAGE 67

46 Table 3-5. (Continued) DIN DON-1 DIN-NH4 + ( M-N)* DIN-NO3 ( M-N)* DON ( M-N) (%) 10.4 12.0 91.8 12.2 122.4 10.4 12.5 91.5 10.7 106.7 10.4 12.9 91.2 11.3 113.1 10.4 12.0 91.8 12.2 122.4 These are the DIN concentrations pr esent before irradiance of solutions Plausible explanations for conversions greater than 100 % are the propagation of random errors given by IC quantification errors The standard deviation associated with the conversion efficiency is presented in Equation 3-6. The standard deviation associated with the DONmeasured was calculated as 5 % of the NH4 +, NO2 -, and NO3 concentrations measured for calibrated sta ndard solutions tested during IC analyses. The standard deviation associated with the DON concentration in the standard solution prepared to measure the conversion efficiency ( DONstandard) is related to the dilution errors calculated through th e scale precision (0.0001 g) and the reactant/solution weight values. 2 tan 2tan% % dard s DON measured DONDON DONdard s measured Equation 3-6 At 10 M-N the standard deviation of the c onversion efficiency varied from 3.5 % from 55.0 % when the DIN/DON ratio of the solution changed from 0.3 to 10.4 (Table 3-5). With 10 M-N of DON in sample a deviation from the true value can be

PAGE 68

47 expected changing from 0.35 M-N to 5.5 M-N when the DIN/DON ratio changes from 0.03 to 10.4. Such uncertainty confirms the existence of positive biases in the measured DON concentrations because of the high DIN DON-1 ratios observed in aerosol samples. Similar DON uncertainties are reported for the method in the literature (Cornell and Jickells, 1999; Mace and Duce, 2002), and constitute the biggest disadvantage of all methods that dete rmine DON by difference of TDN and DIN concentrations. Attempts to reduce the method uncertainty through the reduction of DIN concentrations by dialysis pretreatment (Lee and Westerhoff, 2005) or catalytic reduction of nitrate contents (Ambonguilat et al. ) before DON determination have only been partially successful. Summary Factorial design experiments tested the optimal settings of the UV-photolysis method for the DON determination in aerosol samples. Complete conversion of organic nitrogen in solutions of urea, methylamin e and amino acids (alanine, aspartic acid, glycine and serine) at total organic nitrogen concentration levels under 33 M-N were found at pH 2 for a 24-hr irradiance period. The results agreed with those reported in the literature, but offer additional information about decomposition products and interactions of the selected DON species. The existence of DON in atmospheric aerosols revealed an unaccounted for nitrogen flux to Tampa Bay. Although DON con centrations in the coarse mode were

PAGE 69

48 smaller compared to the fine mode, their co ntribution to the dry deposition flux may be larger because of their higher deposition rate The presence of DON in fine particles could indicates gas-to-parti cle conversion as the correla tion of DON and meteorological conditions. Gas-phase or ganic nitrogen would contribute to the wet and dry deposition by rainwater scavenging and direct mass transf er from the air to the water surface, respectively.

PAGE 70

49 4. PARTICLE AND GAS SCAVENGING PROCESSES Introduction By absorption, rainwater delivers to th e Earth’s surface particulate and gaseous species dissolved in the atmosphere th rough in-cloud and below-cloud scavenging of gases and particles. These processes define an important part of the wet deposition fluxes, especially in tropical places where snow and fog events do not occur or are not so common. Below cloud scavenging processes ar e guided by momentum, mass and energy transfer between gas, liquid and solid phases changing constantly in the four dimensional time-space continuum. Under nor mal conditions all transfer processes between the dispersed and th e continuous phase are bidire ctional or reversible. The complexity of these phenomena is comm only approached by assuming a dispersed system of particles (liquid and/or solid mixtures) suspe nded in a mass of “clean” air, usually referred as the continuous phase, with traces of gas pollutants. Given the vastness of the continuous phase and in orde r to simplify its modeling, it is assumed to be a compartment of infinite volume, ma inly comprised by air. Air conditions are defined by meteorological conditions such as relative humidity, dry bulb temperature and so on.

PAGE 71

50 Gas pollutants, such as sulfuric and nitric acid as well as ammonia, are usually represented by their time-dependent concentr ations in a one-dimensional space defined by the distance between the base cloud height and the ground. It is common to assume that their concentrations are horizontally uniform in the atmosphere. Particles and droplets are considered as chemically hom ogeneous and size distri buted with properties dependent on time, height and particle si ze. Their chemical composition, mass, volume and number concentrations are the results of their physical and chemical interactions with trace gases and/or the normal atmospheric species under isothermal and well mixed conditions. Rainwater droplets are formed after wate r super-saturation of the atmosphere and are precipitated when they reach a si ze of around 1 mm after growing mainly by vapor condensation and droplet coalescen ce (Seinfeld and Pandis, 1998). During or after its formation in the cloud, droplets can dissolve chemical species by particle nucleation scavenging or absorp tion of gases in the intersti tial cloud air (Seinfeld and Pandis, 1998). These processes define the init ial concentration in droplets before they start to scavenge ga ses and particles from the belo w cloud atmosphere. The complete description of the process implies a two-step process with in-cloud and below-cloud scavenging. In this chapter the approaches most commonly used to describe the rain scavenging process are shown to establish the theoretical ba sis of the below-scavenging model used to estimate the contribution of particle and gas s cavenging to the wet deposition fluxes of DIN and DON over Tampa Bay.

PAGE 72

51 Characteristics of the Ra in Droplet Distribution For both particles and gas transport, s cavenging rates depend more strongly on the number concentration/droplet size rela tion defined by the dr op size distribution (DSD). DSDs have been measured and fitted to several analytical functions in terms of the precipitation rate po(mm hr-1), which logically determines the total number of raindrops for a rain event. A raindrop dist ribution defines the number concentration or number of droplets N(Dp) with an specific droplet diameter Dp (mm) suspended in a volume of air (m-3) per unit size ca tegory (mm). To quantify the DSD effect on scavengi ng rates and in the absence of better information about DSDs on the Gandy sampli ng site, three of the most common DSDs were selected: the modified MarshallPalmer (MP), the Gamma distribution by Massambani and Morales (MM) (1991) (Goncalves et al ., 2000) and the lognormal distribution (L) (Mircea et al ., 2000). The Marshall-Palmer one-parameter expone ntial size distributi on is one of the most widely used due to its simplicity (E quation 4-1). Its drawbacks are related with failures in the description of instantaneous spectra and high over prediction of the smallest and largest drops (Feingold and Levin, 1986). Later studies tried to improve its applicability by changing the parameter de pendence (Seinfeld and Pandis, 1998), those changes are usually referred as the modi fied Marshall-Palmer distribution (MP*) (Equation 4-2), even though this distribution is assumed to be stri ctly applicable for drop limit size of 1.2 mm.

PAGE 73

52 mm p mm m N D N D No p o p 21 0 31 4 8000 exp Equation 4-1 mm p mm m p N D N D No o p o p 14 0 3 37 08 3 7000 exp Equation 4-2 N represents the total nu mber of droplets suspende d in a volume of air (m-3) for a certain precipitation rate po. It has been reported that e xponential distribu tions such as MP and MP* cannot describe well the skew -symmetric type distributions observed from experimental measurements. They cannot adjust quickly enough to the decrease in the number of drops at the small end of the spectrum due to their collection by the largest ones during their fall (F eingold and Levin, 1986). A first solution to this problem wa s proposed by Massambani and Morales (MM) (1991) (Goncalves et al.,2000 ) using the Gamma fu nction based of the precipitation rate (Equation 4-3). 2 287 0 2 0524 041 1 exp 86 290p o p o pD p D p D N Equation 4-3 A second solution to the problem is the l ognormal size distribution (L). This is a three parameter distribution based on the tota l number of drops, the geometric mean of drop diameter and the standard deviation of the drop diameter (Equation 4-4). Authors claim that this distribution is easier to understand due to the physical meaning of its

PAGE 74

53 parameters and because the higher moments of the distribution are normally distributed if the original number of drops per unit volume per unit size is lognormal. The parameters used to describe d the total number of droplets N the geometric particle diameter and the standard de viation of the particle di ameter are related to the precipitation rate po(mm hr-1) and were calculated to fit m easured drop size distributions associated with frontal and postfrontal conve ctive rain events. This distribution proved to give a better squared error to the obse rved data than the gamma or the exponential distribution (Feingold and Levin, 1986). o o g o gm p p pp mm p D mm m p N D D D N D N4 23 0 3 22 0 2 210 3 43 1 72 0 172 ln 2 ln exp ln 2 Equation 4-4 Droplet diameter ranges for each distribu tion were chosen to be within the recommended values found in the literature. Lower and upper limits for droplet size are respectively, 1.2 and 6 mm for the MP distri bution (Equation 4-2), 0.3 and 6 mm for the MM distribution (Equation 43) and 0.127 mm and 6 mm for the L distribution (Equation 4-4). Although each DSD has its own advantages and disadvantages, the LDSD had a smaller squared error than the Ga mma or the lognormal distributions after fitting of experimental data (Feingold and Levin, 1986). To simplify the problem it was also assu med that DSDs do not change with time and are not affected by splitting, break-up events or other phenomena (Goncalves et al.

PAGE 75

54 2000). Once the droplet is formed it was assu med that evaporati on does not occur and the droplet diameter remains constant until it reaches ground level (Seinfeld and Pandis, 1998). Another assumption related with the cont act time during scavenging is that each droplet reaches its terminal velocity ( Ut) quickly and travels at this speed until reaches the ground level (Kumar, 1985; Se infeld and Pandis, 1998). The drop settling velocity is the velocity reaches by the particle/droplet when buoyancy-corrected gravitational force and the drag force balance one to each other. Although rain drop normal sizes settle at high Reynolds number ( Re ),the droplet settling velocity was calculated using a routine that allows the three possible motion regimes of particles under grav itational settling: Stokes’ regime ( Re <1), transitional regime (11000). This feature allowed us to use the same routine to calculate the particle settling characteristics. Due to the internal codependence of the settling velo city and the Reynolds number, it was necessary to iterate between Re and Ut values using as a first assumption that particle velocities are in the transition regime. In this regime Ut can be calculated using only the particle diameter by an empirical expression of the product of the Reynolds number ( Re ) and the drag coefficient ( CD), CDRe2 (Equation 4-5). This relation agr ees with tabulated data within 3% for 1< Re <600 and within 7% for 0.5
PAGE 76

55 g t g p g p g t g g p p DU D J J D U g D C J Re 0178 0 9935 0 070 3 exp 3 4 ln Re ln 1000 Re 12 2 3 2 Equation 4-5 The drag coefficient CD accounts for drag forces acti ng on the particle during its settling and therefore depends on the particle’s shape, size and relative velocity as well as on the air’s density and viscosity. It is equal to 24/Re at Re <1 and constant at 0.44 for Re >1000 and can be calculated iteratively from empirical expression s such as the one presented by (Hinds, 1999). Expressions for Ut (Equation 4-6) for each regime can be found elsewhere (Hinds, 1999). g D p p t g c p p tC g D U gC D U 3 4 1000 Re 18 1 Re Equation 4-6 To adjust the routine for the particles’ analyses, the slip correction factor ( Cc ) was included. The CC or Cunningham factor corrects for deviation from Stokes’ law for particles smaller than 0.1 m and was calculated using its empirical expression depending only on the drop diameter and the air mean free path ( ) both measured in m (Equation 4-7) (Jung et al. 2003).

PAGE 77

56 p p p CD D D C 435 0 exp 84 0 493 2 1 Equation 4-7 Properties of the continuous phase were a ssumed to change just with changes of temperature at isobaric cond itions. Viscosity and density of the air were found using Equation 4-8 assuming valid the idea l gas at atmospheric conditions. 15 293 15 273 2 1 15 293 15 273 1 15 293 15 298 15 273 10 81 1 15 293 15 273 1 15 2933 74 0 2 5 74 0T m Kg T atm K T T m s N T atm K Tg g g Equation 4-8 Deposition-Weighted Average Concentration During a rain event, the raindrop concen tration changes with the precipitation rate and the atmospheric concentration of pollutants for each time instant. The bulk concentration of the rainwater collected is the result of the cumulative contribution of the wet deposition flux of all droplets in th e DSD that have fallen to the ground during the time interval considered. The wet deposition flux for a ll droplets with diameter Dp Fw(t,Dp) (Equation 49) should be equal to th e droplet concentration C(h,t,Dp) at the ground level at a certain time and the intensity of rain for droplets of size Dp. p p p wD t I D t h C D t F , Equation 4-9

PAGE 78

57 The intensity of rain I(t,Dp) (Equation 4-10) for droplets of size Dp is given by the number of droplets NDdDp from the DSD, the water vo lume per spherical droplet Vp, and their settling velocity Ut(Dp). This parameter depends exclusively on the precipitation rate defining the DSD’s parameters. p p D p t p pdD D t N D U D D t I 6 ,3 Equation 4-10 At any time the pollutant concentration in the bulk rain collected at ground level (z=h) is given by the deposition-weighted average concentration (Equation 4-11) over the raindrop size distri bution (Hales, 2002). 0 0, , ,p p p p p rainwaterdD D t I dD D t I D t h C t h C Equation 4-11 The sum of all I(t,Dp) is equivalent to the total precipitation rate (Equation 412) (Seinfeld and Pandis, 1998). 0 3 0, 6 ,p p D p t p p p odD D t N D U D dD D t I t p Equation 4-12 t p dD D N D U D D t h C t h Co p p D p t p p rainwater0 36 , Equation 4-13 If the rain continues for a time interval ts at a constant precipitation rate then the bulk rain concentration (Equa tion 4-14) is given by:

PAGE 79

58 t p dD D N D U D D t h C dD D I dD D I D h C h Co p p D p t p p p p p p p rainwater 0 3 0 06 , Equation 4-14 s o t p p D p t p p t o t p W rainwatert p dt dD D N D U D D t h C dt t p dt D t F h Cs s s 00 3 0 06 , Equation 4-15 Gas Scavenging The reversible absorption/desorption of trace gas pollutants on droplet surfaces is commonly modeled using the film-theory approach, where the pollutants transfers through infinitesimal liquid and gaseous f ilms at the interface limit establishing thermodynamical equilibrium. The resistance de fining the mass transfer rate is assumed to be concentrated in the gas phase, where the transport occurs primarily by diffusion and convection (Hales, 1972). The liquid-phase resistance is negligible compared to this due to well mixing conditions inside th e drop given by shear forces on its surface. The overall mass transfer coefficient ( KC) based on the gas-phase driving force is determined by the Frssling equation (Kum ar, 1985). The mass transfer coefficient ( KC) defined for a droplet in motion by the Fr ssling equation (Fr ssling (1938) cited by Kumar, 1985):

PAGE 80

59 3 36 0 2 Re 6 0 2g g g g p g t g p CD U Sc D K Sh D D Equation 4-16 Sh is the Sherwood number and depends on the gas-phase mass transfer coefficient KC, the particle diameter and the gas diffusivity Dg. Sc is the Schmidt number given in terms of the kinematic viscosity (g / g) and the diffusivity Dg of the gas. The Fuller-Schettler-Giddings correlati on (Equation 4-17) was used to calculate gas diffusivity values in air. This corr elation is recommended for non-polar gases in binary mixtures at low pressure and it ha s an associated error of 5.4% (Perry, 1999). 23 1 3 1 2 1 4 7001 0B A AB AB gv v P M T D D Equation 4-17 T is the temperature in K, MAB is the molecular weight of gas A in B, P is the pressure of gas B and is the atomic diffusion volume increment in cm-3 mol. The molar volume of A is considered additive in terms of the atoms comprising the molecule. If the aqueous concentration of th e absorbed gas is represented by C(z,t,Dp) the one at the gas phase by Cg(z,t) and the gas concentration at the gas-liquid interface is Cgas-liq(z,t) then a mass balance over an horizon tal plane of infinitesimal thickness z (Kumar, 1985) is given by Equation 4-18: t z C t z C K D z D t z C D U t D t z C t z C t z C K D z D t z C D U t D t z C Dliq gas g C p p p t p liq gas g C p p p t p p, 6 , , , , , 62 3 Equation 4-18

PAGE 81

60 At the gas-liquid interface, it is assu med that the concentration of the gas pollutant reaches equilibrium according to Henry’s law coefficient ( Hc), and can therefore be expressed in te rms of the dissolved gas con centration on the aqueous phase. When the dissolved dissociates in solution, Hc can be substituted by effective values of Henry’s constant ( Hc* ). The Henry’s law coefficient to be used in these equations must be in its dimensionless form, the ratio betw een the concentration of the species in the gas and liquid phases at equilibrium. c g C p p p t pH t z C t z C K D z D t z C D U t D t z C , 6 , , Equation 4-19 Equation 4-19 describes a reversible droplet-gas transfer driven by the concentration gradient between the droplet surface and the bulk gas concentration. A gas is considered reversible if it can be absorbed or desorb ed at any time from the rain droplet because Cg(z,t) ~ Caq(z,t)/Hc. This is applicable to all gases with low to moderate Henry’s constants. In the case of a gas fo r which absorption depends on pH, such as a weak acid or base, Hc can be substituted by its effective value ( Hc* ). From here, four possible appr oaches were considered: I. The gas behaves as an reversible gas and the gas concentration is uniform and time-independent, meaning it remain s constant during the rain event and does not change with altitude ( Cg(z,t) ~ Cg(z) ~ Cgo) II. The gas behaves as a reversible gas a nd the gas concentration is uniform but decreases exponentially w ith time during the rain event following a firstorder rate.

PAGE 82

61 III. The gas behaves as an irreversible ga s and the gas concentration is uniform and time-independent, meaning it remain s constant during the rain event and does not change with altitude ( Cg(z,t) ~ Cg(z) ~ Cgo) IV. The gas behaves as an irreversible gas and the gas concentration is uniform, but decreases exponentially with time dur ing the rain event following a firstorder rate. All of these cases are approximations for the atmospheric real processes occurring during rain scavenging. During rain events, emissions do not stop and then the atmospheric system receives a constant i nput of the gaseous chemicals that is not accounted for in any of the four cases be fore. At the same time, gas-to-particle conversions could occur and, if so, an additi onal particle input shoul d be accounted for. The case II was not considered in the perf ormed calculations. In the absence of hourly gas concentrations of reversible gases at the Gandy monitoring site, no calculations could be done to describe gas scav enging under exponential decay of the gas concentration. Gas Scavenging With Reversible Absorpti on at Constant Gas Concentration If the pH and size of drople t is assumed to be constant during its fall from clouds the concentration on the aqueous pha se for a droplet with diameter Dp can be obtained by rearrangement of Equati on 4-19 under the assumption of an homogeneous and constant gas concentration:

PAGE 83

62 t z C H C H D K z t z C U t t z Cc g c p c t, 6 , Equation 4-20 Boundary and initial conditions are 0 0 t C or solute-free droplets, and 0 0 z Ccorresponding to a zero c oncentration with a zero contact time at any distance for a single droplet size. Applyi ng a Laplace Transform on Equation 4-20 we have: z C s H C D H K dz z C d U z C z C sc g p c c t6 0 Equation 4-21 z C s H C dz z C d U z C z C sc g t0 Equation 4-22 s U H C z C U s dz z C dt c g t Equation 4-23 s s H C z U s C z Cc g texp1 Equation 4-24 Substituting the boundary condition (0,t) equa l to zero or solute-free droplets: z U s s s H C z Ct c g exp 1 Equation 4-25 Separating in simple fractions and expanding the exponential term: z U z U s s s H C z Ct t c g exp exp 1 1 1 Equation 4-26

PAGE 84

63 Applying the inverse Laplace Transform: 1 exp exp exp 1 ,t t t c gU z t U z t U U z t H C t z C Equation 4-27 tU z t U is the step function at is equal to zero for tU z t and equal to one when tU z t at which the concentration becomes: z H U D K H C z Cc t p c c g6 exp 1 Equation 4-28 When the droplet has fallen all the way to the ground from the base cloud at height h, its con centration is given by: h H U D K H C D h Cc t p c c g p6 exp 1 Equation 4-29 At any time the pollutant concentration in the bulk rain collected at ground level (z=h) is given by the deposition-weighted average concentration over the raindrop size distribution (Hales, 2002).

PAGE 85

64 Gas Scavenging With Irreversible Absorption at Constant Gas Concentration In the case of very soluble gases such as ammonia or nitric acid, Hc is so large that c gH t z C t z C , Such is the affinity of the gas to the aqueous phase that once the gas is absorbed; it dissociates to b ecome a solvated species staying in solution if the pH level is kept under/over certain values. If this is the case the mass-transfer is said to be irreversible. Mass-transfer from an irreversible gas to the surface of a droplet of size Dp can be modeled by: t z C D K D z D t z C D U t D t z Cg p C p p p t p, 6 , , Equation 4-30 If the gas absorption proceeds on each droplet with no interference from droplets on the surroundings, and the gas concentration is assumed to be constant with changes in time and height during the droplet fall from the base cloud, the variation on the concentration of a droplet with diameter Dp can be described by: g C p tC K D z t z C U t t z C 6 , Equation 4-31 Initial and boundary conditions were assume d to be C(0,t) equal to zero due to the lack of information about in clo ud scavenging and C(z,0) equal to zero corresponding to a concentration with a zero contact time at any distance.

PAGE 86

65 The linear partial differential equation of first order can be solved in two steps by Laplace transformation. and -1 are the symbols to represent the Laplace operator and the inverse operator of the Laplace transf orm. To solve Equation 4-31 we apply the Laplace transform: g C p tC K D z t z C U t t z C 6 , Equation 4-32 s D C K dz z C d U z C z C sp g C t1 6 0 Equation 4-33 s U D C K z C U s dz z C dt p g C t1 6 Equation 4-34 s U z s D C K z Ct p g Cexp 1 1 62 Equation 4-35 The final solution is then found by appl ying the inverse La place operator as follows: s U z s D C K z Ct p g Cexp 1 1 62 1 1 Equation 4-36 t t p g CU z t U z t U t D C K t z C 6 Equation 4-37 The tU h t Uis the step function that is equal to zero at all times tU h t and equal to one when tU h t The gas concentration in this case is the measured concentration at the time the rain started and the final concentration of the rainwater

PAGE 87

66 sample collected at the end of the rain even t due the gas scavenging of the gas would be given by the time and deposition-weighted averag e concentration as is explained in the next section. Gas Scavenging With Irreversible Absorption and Exponential Decay of the Gas Concentration The gas concentration changes with time, because as soon as the rain starts, the pollutant is removed from the air, and if th ere are not additional inputs from emissions, its concentration should decrease with time after it is ab sorbed into the rain drops. The variation of the trace gas pollutant concentration CAg during the scavenging can be described applying the conti nuity equation (Equation 4-38) In the case of purely advective-diffusive transport, the first term on the right describes the net rate of addition in moles of A per unit of volume by convecti on, the second term describes the addition by diffusion, the third one w is the time rate of gain of pollutant mass in the liquid phase by scavenging and the last one represents the production of A by chemical reactions (Hales, 1972; Bird et al., 2002). A A A g Ar w J v C t Cg Equation 4-38 In the absence of chemical production of A and if gas concentration is assumed to be variable just in one-dimension z (the height from the clouds to the ground), the Equation 4-38 can be written as:

PAGE 88

67 2 0, ,, ,g g g gAA p A zgpCpDpAp cCztC CDxt CvDKDNDCztdD tzzzH DEquation 4-39 KC and g Dare respectively the gas-diffusion coefficient and the overall masstransfer coefficient based on the gas-phase driving force, where C(Dp,z,t) is concentration of the aqueous phase and Hc is the dimensionless Henry’s constant based on molar concentrations. If the convective tran sfer is assumed to be negligib le, the diffusion coefficient is assumed to be constant and the gas is considered as an irreversible gas c gH t z C t z C , and the equation becomes: 2 2 2 0, ,gg gAA g ApCppCztC CztDKDdD tz D Equation 4-40 The term inside the integral is defi ned as the gas scavenging coefficient It depends only on the characteristics of the drop size distribution and the transport properties of the trace gas pollutant. 2 2,, ,gg gAA gACztCzt Czt tz D Equation 4-41 Initial and boundary conditions commonl y assumed are represented by a onedimensional gas concentration profile, and negligible fluxes of the pollutant at the system boundaries:

PAGE 89

68 0 0 ,0 0 0 h z A z A g g Az C z C C z C z Cg g g Equation 4-42 The gas concentration determined by the solution of Equation 4-41 was presented by Kumar (1985) as: h g n h g Adz h n z z C h zn t h n dz z C t h t z Cg0 0 1 2 2 2 0 0' cos cos exp exp 1 Equation 4-43 If the initial condition is a uniform initial-gas-phase concentration profile then the solution can be simplified to: t C Cg g A* exp0 Equation 4-44 The concentration of the aqueous phase can be found by solving the Equation 419 after the introduction of the exponential decay for the gas concentration. t C D K D z D t z C D U t D t z Cg p C p p p t p exp 6 , ,0 Equation 4-45 Initial and boundary conditions were assumed to be C(0,t) equal to zero due to the lack of information about in cloud scavenging and C(z,0) equal to zero corresponding to a concentration with a zero contact time at any distance. For a single droplet diameter, the soluti on of the equation can be found applying Laplace Transform and then solving the fi rst order linear differential equation. t C K D z t z C U t t z Cg C p texp 6 ,0 Equation 4-46

PAGE 90

69 s D C K dz z C d U z C z C sp g C t 06 0 Equation 4-47 s D U C K z C U s dz z C dp t g C t06 Equation 4-48 s s D C K z U s C z Cp g C t0 26 exp Equation 4-49 0 6 00 2 s s D C K C Cp g C Equation 4-50 s s D C K Cp g C0 26 Equation 4-51 z U s s s D C K z Ct p g Cexp 1 60 Equation 4-52 z U s s s D C K z Ct p g Cexp 1 1 1 60 Equation 4-53 z U s s s D C K z Ct p g Cexp 1 1 1 60 1 1 Equation 4-54 The concentration on a droplet with diameter Dp is: t t p g C p g CU z t U z t U D C K t D C K t z C exp 1 6 exp 1 6 ,0 0 Equation 4-55

PAGE 91

70 As it was said before, the tU h t Uis the step function that is equal to zero at all times tU h t and equal to one when tU h t The contribution of the gas scavenging to the final concentration of the rainwater sample collected would be given by the time and deposition-weighted average con centration, and the concentration Cg0 considered per interval is the gas concentrati on measured at the correspondent hour. Particle Scavenging The mass gained by droplets during part icle scavenging depends exclusively on the number of particles dissolved by each dr oplet during its fall from the base cloud. Particle collection can occur by different m echanisms, but the most important ones are impaction, interception and Brownian diffusi on. All three mechanisms are depicted in Figure 4-1. In order to simplify collision m odeling, all mechanisms are considered to be independent from each other. Becaus e not all particles surrounding a droplet can collide with its surface, rates of particle collection are described using the collision efficiency E(Dp,dp) which is defined as the ratio of total number of droplet/particle collisions to the total number of particles in an area equivalent to the droplet’s effective cross-sectional area The collection model used in the partic le scavenging calcula tions was proposed by (Slinn, 1977), and is based on the assump tion that acceptable laws of physics are homogeneous in all dimensions (Weisstein, 1999) and can be described using a few number of dimensionless groups. All particle /droplet interactions can be described

PAGE 92

71 through the following dimensi onless groups: the Reynolds ( Re ), Schmidt ( Sc ), Stokes ( St ) numbers, the interception parameter () (droplet/particle diameter ratio) and the is the ratio of air to water visc osity. The collision efficiency ( E ) can be seen as the sum of independent and additive effects from each mechanism (Equation 4-56). St E E Sc E St Sc EM N DRe, Re, Re, , Re, Equation 4-56 ED, EN and EM are the collision efficiency for Brownian diffusion (Equation 457), interception (Equation 4-58) and imp action (Equation 4-59). Brownian diffusion prevails over the other mechanisms when pa rticles have an aerodyna mic diameter < 0.2 m, while impaction dominates for particles larger than 1 m (Figure 4-3). Rain Droplet Dp+dpParticle Interception Particle Impaction Particle Brownian Diffusion dp Figure 4-1. Important Co llection Mechanisms for Particles During Rain Scavenging.

PAGE 93

72 During the droplet gravitational settling, the air establishes a flow field around the droplet which streamlines influence the part icle motion. If the particle with diameter Dp has enough inertia to overcome the flowing forces on the streamlines, it will hit the drop surface and will be collected by the dr oplet. This collection process is called impaction and is a function of the Stokes number ( Stk ), defined as the ratio between the particle stopping distance and the droplet diameter. If the particle has negligible inertia, settling and Brownian motion, and it is moving following the streamlines in the interval p pd D very close to the particle surface, it will intercept th e particle surface and will be collected. This is the intercep tion mechanism and it depends mainly on the ratio between the droplet and particle diameter (). It is the only one does not depend on flow velocity. The Brownian diffusion mech anism occurs when particles collide with the droplet surface during their random m ovement caused by bombardment of gas molecules (Hinds, 1999). The semi-empirical expression for ED assumes Stokes’ flow for a “solid” spherical droplet suspended in incompressibl e air where particles of negligible mass move toward their surface according the conve ctive diffusion equation. This expression is more applicable when 60 Re2 1 (Slinn, 1977). Sc Sc Sc Sc EDRe 16 0 Re 4 0 1 Re 4 Re,3 Equation 4-57 During interception a collision can occu r only if the particle reaches the stagnation point or the droplet/particle axis of symmetry 2 2 2 p pD d but the droplet boundary layer moves this contact po int in a fraction approximate to 2 1Re 2 ~ 2pD ;

PAGE 94

73 therefore all collisions occurs in a di stance from the droplet equivalent to 2 1Re 2 1 (Slinn, 1977). Thus, for a droplet with internal circulation EN can be defined as: Re 2 1 1 4 NE Equation 4-58 EM was based on the solution of dimensionless forms in terms of Stokes number for the continuity and momentum equation ( no diffusion) for a particle immersed in potential flow. The main characteristic of the EM model is the Critical Stokes number or S* below which there is no inertial impaction (Slinn, 1977). Special considerations must be done for systems with possible Stokes numbe rs over its critical va lue. The performed calculations assumed EM equal to zero for all collisions with Stokes numbers smaller than S*. Re 1 ln 1 Re 1 ln 2 1 12 1 3 2 * S S St S St EM Equation 4-59 Dimensionless numbers are defined as: 2 Re tptp tpggp w g gaerosolppgUDud UDd ScSt DD D Equation 4-60 The collision efficiency depends then on the droplet and the particle diameter. Smaller particles are more efficiently re moved by smaller droplets than by bigger droplets. This can be seen in the deep slope of the surface shown as Figure 4-2.

PAGE 95

74 Figure 4-2. Collision Efficiency Function in Terms of the Particle and Droplet Diameter. 10-3 10-2 10-1 100 101 10-6 10-5 10-4 10-3 10-2 10-1 100 Particle Diameter ( d p m ) E(Dp,dp) Total Impaction Interception Brownian Figure 4-3. Contributions of the Most Im portant Mechanisms to the Collision Efficiency.

PAGE 96

75 All possible collisions will occur in side a space called “collision” volume (Figure 4-4), delimited by the maximum c ontact volume swept by a particle/droplet system with diameter p pd D moving at a velocity of p t p td u D U The collision volume is then equal to p t p t p pd u D U d D 2. Figure 4-4. Collision Volume for Particle/Droplet Systems.

PAGE 97

76 If after a collision the particle is comple tely dissolved into the droplet, then the rate of mass collected by a droplet falling in side the control volume will be the mass of all particles suspended in it times the coll ision efficiency (Equation 4-61). 0 2, ,p p p p M p t p t p p pdd d D E t d n d u D U d D dt D dW Equation 4-61 nM is the particle mass number describi ng the mass concentrated in particles with diameter (dp) per unit of volume of air (m3) per unit size category (mm). Assumptions commonly made are that 2 2~p p pD d D and that p t p td u D U (Seinfeld and Pandis, 1998). In the absence of information a bout changes in the particle and droplet distribution during the rain event, it is necessary to assume that both distributions remain constant in time during the rain event. The concentration in rainwater is then expressed in terms of the water content collected for the entire rain event and the total particle mass scavenged by all droplets in the distribution. For rain events longer than 1 hr, this equation can be used approximating the whole event to a single one with the average precipitation rate: 0 3 000 26 , *p p D w p ts p D p p p p M p t p s l scavenged p p rainwaterdD D N D dt dD N dd d D E t d n D U D t W W C Equation 4-62 Equation 4-62 can be used for any substance because particle scavenging is mainly treated as a mechanical process.

PAGE 98

77 To describe nM the lognormal size distribution was used (Equation 4-63). M represents the total aero sol mass collected as PM10, dgm is the geometric mean diameter and is the standard deviation. Lower and upper limits for particle sizes were chosen to be 0.01 and 10 m. Based on experimental evidence from the Gandy Bridge site (Campbell, 2005) 2 2ln 2 ln exp ln 2gm p p p Md d d M d n Equation 4-63 NH4 +, NO3 -, Cland Na+ contents for samples from the Gandy Monitoring site have shown to follow lognormal distributions in function of the particle diameter. Prediction of particle size distributions of these species were successfully made by using a set of general parameters for dg m and determined by fitting experimental data (Table 4-1) (Poor et al. 2006; Campbell, 2005). Table 4-1. Parameters for Log-normal Size Distribution for PM10 Samples Collected at the Gandy Bridge Monitoring Site (Campbell, 2005). Substance Geometric Mean of Particle Diameter ( m) Standard Deviation of Particle Diameter ( ) NO3 3.5 1.90 NH4 +~ DON ~DMA 0.38 1.99 Na+ 4.0 1.90 SO4 -2 0.48 2.4 Cl4.80 1.80

PAGE 99

78 The scavenging process can also be treated as the sum of many first-order processes describing the particle collection fo r all possible particle diameters. In this case, the total particle removal due to interactions with droplets is determined by the scavenging coefficient, this defines the rate (time-1) at which particles with diameter dp are removed for all droplets in a rain event with a precipitation rate po. p p D p p p t p pdD D N d D E D U D d 42 0 Equation 4-64 Because the particle collection by droplets is just seen as a mechanical process, all equations described here are applicable to nitrate and ammonium particle contents. For a particle with diameter dp, its scavenging by rainwater can be approximated to a first-order linear process with a constant rate equal to (dp). All possible droplets in the DSD that can effectively collide with the particle dp cause a decrease in its mass concentration according to this relation: t d n d dt t d dnp M p p M, Equation 4-65 If the precipitation rate po remains constant during the rain event the scavenging coefficients do not change. This expre ssion indicates two possible cases, timeindependent or variable particle mass distri butions (particle concentration), as it was seen for gases.

PAGE 100

79 Particle Scavenging at Consta nt Particle Concentration If even when the particles are removed from the air by droplets, gas-to-particle conversion processes and particle emission s were able to keep constant particle concentrations during the rain event, the total particle scavenging rate would be defined by all first-order scavenging processes of all particles in the particle size distribution. The rate of particle removal or decrease in the aerosol concentration would be equal to: p p M o p aerosoldd d n p d dt dW 0, Equation 4-66 If it assumed that the precipitation rate remains constant during each one-hour rain interval then the particle scavenging coefficients remain constant and the total aerosol mass lost in an time interval 0t t ts where 00 t becomes equal to the mass scavenged. This is equal to: s p p M o p aerosol p scavengedt dd d n p d t dt dW W 0, Equation 4-67 The rainwater concentration due to th e particle scavenging for one rain event with n one-hour intervals should be equal to the total mass scavenged and the total water volume of all droplets collected at the ground level. n i p p D p n i i s p p M i o p n i p p D p n i i p scavenged rainwater pdD D N D t dd d n p d dD D N D W C1 0 3 1 0 1 0 3 16 6 Equation 4-68

PAGE 101

80 Particle Scavenging With Exponential Decay of the Particle Concentration The previous case considers that particle concentration remains constant during the whole rain event, but in the absence of gas-to-particle conversion or other particle sources, particle concentrations should d ecrease in time according to the exponential decay defined by the scavenging coefficient ra te for each particle size of the particle size distribution. For a specific particle size dp, the exponential decay of its particle concentration during a rain event of duration 0t t ts duration where 00 t is given as: s o p M p Mt p d t n t d n exp ,0 Equation 4-69 This equation also defines how many particles of diameter dp remain suspended in the air after a scavenging process of 0t t ts duration where 00 t at a removal rate of o pp d The total rate of loss of aerosol mass that is equal to the mass gained by rainwater droplets is the sum of decrements in the concentration of particles with all possible aerodynamic diameters: p t s o p M o p aerosol p scavengeddd dt t p d n p d t dt dW Wo 00* exp Equation 4-70 p s o p M p scavengeddd t p d n Wo 0* exp 1 Equation 4-71 For a rain event longer than 1 hour wi th different precipitation rates per hour, calculations were done in 1-hr interval using the particle concentration remaining as the initial particle concentration of the period. The total decrease in the particle mass by

PAGE 102

81 scavenging after all n rain intervals is de termined adding all differential changes as follows: p n i s i o p M total p scavengeddd t p d n Wi i 1 0* exp 11 Equation 4-72 The concentration of the rainwater sample collected at the end of the rain event is equal to the volume weighted concentration in terms of the total volume of water droplets collected for each period of rain. n i p i o p D p p n i s i o p M p rainwaterdD p D N D dd t p d n Ci i1 0 3 1 0, 6 exp 11 Equation 4-73 Scavenging Coefficients Used in Comm unity-Scale and Mesoscale Air Quality Models In air quality models, all the variables related with mass-transfer and chemical reactions must be averaged for the range of variation of the meteorological conditions. Average scavenging coefficients, determin ed for a specific precipitation rate for example, are used to describe the contri bution of the gas and particle scavenging processes in events with different precipitation rates. The variables more commonly used in the models are the gas scavenging coefficient, the mean mass particle scavengi ng coefficient, and the normalized particle scavenging coefficient. Equations 4-74 a nd 4-75 show the definition of the gas

PAGE 103

82 scavenging coefficient and mean particle scavenging coefficient, the latter accounts for the size effect on the partic le collection process. The problem in the use of average valu es for scavenging properties is their dependency on the precipitation rate. Th at difficulty can be circumvented by normalizing the scavenging coefficients by the precipitation rate. The Equation 4-76 shows the average normalized par ticle scavenging coefficient. 0 2p D c pdD N K D Equation 4-74 p p M p p p M mdd d n dd d d n 0 0 Equation 4-75 p p M o p p p M o mdd d n p dd d d n p 0 0 Equation 4-76

PAGE 104

83 5. ESTIMATION OF THE GAS AND PARTICLE CONTRIBUTIONS TO WET DEPOSITION OF DISSOLVED ORGANIC NITROGEN Introduction As it was shown in previous chapters dissolved organic nitrogen (DON) is present in particulate matter (PM2.5 + PM10-2.5 = PM10) and it was found that DON contributed 10.1 5.7% of the tota l dissolved nitrogen (TDN=DIN+DON). DON recovered from fine particles represented an average of 79.1 18.2% of the total organic nitrogen found. The existence of DON in particles suggested the DON presence in rainwater samples. To improve estimates of wet depositi on fluxes over Tampa Bay, DON and DIN concentrations were measured in PM2.5, PM10-2.5, and rainwater samples collected at a bayside monitoring site between July-Septe mber 2005. Samples collected during rain episodes were used to estimate particle an d gas scavenging rates of nitrogen species. The simultaneous collection of samples aids in understanding the phase partitioning of species and in evaluating possible sources th rough the presence of similar species in both samples. Although a complete chemical characte rization of the DON was not available, it was possible to identify the presence of dimethylamine (DMA) in the aqueous extracts from the PM2.5 samples. Under the assumption of vapor-liquid equilibrium between the

PAGE 105

84 atmosphere and the liquid content of partic les, the DMA gaseous concentrations were estimated and used to estimate the cont ribution of the gas scavenging to the DON rainwater concentrations. Experimental Methods Field Sampling and Sample Processing Aerosol and rainwater were collected during rain events on July-September 2005 at an atmospheric deposition monitoring site adjacent to Tampa Bay at the eastern end of the Gandy Bridge (Figure 1-1) in Tampa, Florida (27.78 oN, 82.54 oW). 24-hr integrated PM2.5 and PM10–2.5 samples were collected by a Ruppretch-Patashnick Dichotomous Partisol-Plus Model 2025 sequ ential air sampler (Appendix B) following the same procedures described in the second chapter. Rainwater samples were collected using a precipitation-only rainwater collector (Aerochem Metrics, Inc.) and accumulated fo r the 24-hr period of aerosol collection. Field and lab blanks consisted of DDW from a clean unexposed co llection vessel taken to the site or prepared in the lab and p oured into amber glass bottle after volume measurement. All rainwater and aerosol samples and blanks were stored at 4 oC until analysis. Details about sample and blank preparation can be found in the second chapter. Precipitation rates, time of occurrence and duration of rain events were found by simple averaging of hourly data per samplin g day measured at the site and provided by

PAGE 106

85 the Environmental Protection Commission of Hillsborough County (Florida). Base cloud height and other meteorological data were obtained as ho urly surface weather observations from the Tampa International Ai rport available at the National Climatic Data Center website ( www.ncdc.noaa.gov/servlets/ULCD ). Laboratory Analyses NH4 +, Na+, Cl-, NO2 -, and NO3 were determined using ion chromatography with the same procedures explained in the s econd chapter. Dimethylamine (DMA) was determined using a gradient separation meth od define in the litera ture (Rey,2001) in a IonPac CS16 analytical column (4 x 250 mm) kept at 40 oC and a CSRS-Ultra II 4mm suppressor with a methanesulfonic acid (MSA) aqueous solution as eluent. Calibration curves were made from dilu tions with DDW of individual 1000 mg l-1 NH4 +, Na+, Cl-, NO2 -, and NO3 standard solutions (UltraScientific, Inc), while check standards were prepared from 100 mg l-1 anion (Cl-, NO2 -, NO3 -, PO4 -, SO4 -2) and cation (Na+, NH4 +, K+, methylamine, dimethylamine, trimethylamine, diethanolamine, triethanolamine, Mg+2, and Ca+2) mixtures (Inorganic Ventures Inc.). Check standards were run after every ten samples and were w ithin 5% of their prep ared concentrations (within 10% for concentrations near the DL). Dissolved organic nitrogen was determ ined using the UV-photolysis method at the optimal conditions defined in the second chapter. The reactor performance was checked by irradiation of mi xtures of cations and/or am ines at known concentration

PAGE 107

86 levels much higher than those expected fr om samples. Conversion values ranged from 75 to 98%. NO2 concentrations were under its detection limit for all samples. For all ionic species, including the protonated DMA, the detection limit (DL) was defined according to the method detection level (MDL) defin ition. The method uses a solution with the matrix of interest prepared to be in the ra nge of one to five times the calculated MDL. The solution concentration was measured seve n times randomly over a period of at least 3 days, and the MDL is defined as the standa rd deviation of the replicates times the tvalue from a one-side t-dist ribution with six degrees of freedom at a 99% confidence level (APHA, 1998). Detection limits were 7 ppb, 9 ppb, 7 ppb and 100 ppb for NH4 +, NO2 -, NO3 and DMA, respectively. The DON detection limit was 1.6 M-N and was defined as three times the standard devi ation of DON levels in blanks (Caldern et al ., 2006a). DIN and DON Concentrations and Wet Deposition Fluxes Precipitation events during July-September 2005 at the Gandy Bridge monitoring site were variable with an av erage precipitation rate of 7.0 10.7 mm hr-1. The average dry bulb temperature was 27.9 1.0 oC. There were 34 rain events recorded at the Tampa Intern ational Weather Station during the months from July to September 2005 with an averag e rate of 0.1 0.06 mm hr-1. At the Gandy Bridge monitoring site, 12 precipitation events were studied and showed an average precipitation rate of 7.0 10.7 mm hr-1. The volume of rainwater samples was variable ranging from 40 to 2000 ml, and there was ex pected to see sample evaporation because

PAGE 108

87 samples were kept in the collector until the daily pick-up. Evaporation may have concentrated the analytes for small sample volumes. Concentrations for Na+, Cl-, NO3 -, NH4 +, DIN (DIN= NH4 + + NO3 -+ NO2 -) and DON) in rainwater and PM10 from 24-hr integrated samples collected during rain events are shown in Table 5-1 and Figure 5-1. Table 5-1. Experimental Concentrati ons of 24-hr Integrated Aerosol (PM10) and Rainwater Samples Collected at the Gandy Bridge Monitoring Site, Tampa, FL. Rainfall Concentration ( M-N) PM10 Concentration (nmol-N m-3) RAIN EVENTS ClNH4 + NO3 DIN DONClNH4 + NO3 DIN DON 07/20/2005 36.2 15.6 34.0 49.7 7.0 49.539.0 14.4 53.3 5.1 07/24/2005 34.6 11.9 24.1 36.0 10.2 47.2116.1 13.7 129.77.3 08/06/2005 55.6 8.2 19.1 27.4 3.2 42.415.2 12.3 27.5 9.5 08/07/2005 23.2 5.5 12.4 17.9 B.D. 34.412.6 9.6 22.2 6.5 08/08/2005 26.0 8.1 15.1 23.2 B.D. 27.343.5 9.3 52.8 5.6 08/09/2005 17.6 8.4 26.0 34.4 2.5 2.9 155.4 10.6 166.011.5 08/22/2005 58.5 25.0 105.3130.37.2 15.3167.9 11.3 179.27.1 08/23/2005 17.2 19.4 48.7 68.1 2.7 10.981.7 11.8 93.5 6.4 08/24/2005 61.0 20.4 127.1147.44.1 28.517.9 16.0 33.9 2.9 08/28/2005 59.1 11.5 15.2 26.7 2.7 34.325.1 10.9 36.0 5.0 09/01/2005 45.5 11.9 28.6 40.5 3.1 24.354.6 14.1 68.8 3.0

PAGE 109

88 After analyses, it was found that NH4 + represented 29.0 7.8% and NO3 71.0 7.8 % of the DIN measured in rainwater sa mples. DIN and DON average concentrations were 54.7 44.0 M-N and 4.7 2.7 M-N, and DON represented 8.9 5.8 % of the total dissolved nitrogen (TDN = DIN + DON). In contrast, NH4 + represented 75.8 15.4 % and NO3 24.2 15.4 of the DIN in PM10 samples. In aerosols, the DON represented 10.3 7.3 % of the TDN. Wet deposition flux (Fwet,i, kg-N ha-1 yr-1) was calculated as the product of the volume-weighted average concentration (Cvwa,i, mg l-1) and the average rainfall amount (D, mm day-1) (Equation 5-1), where 10-2 is a units conversion factor. VWA concentrations for NH4 +, NO3 and DON were equal to 8.9, 20.3 and 2.16 M-N, respectively; and the average rainfall amount for July and August 2005 was 3.06 mmd-1. 2 ,10 day mm L N mg i day ha N kg i wetD ion Concentrat VWA F Equation 5-1 Estimated wet deposition fluxes for NH4 +, NO3 and DON were 1.40, 3.18 and 0.34 kg-N ha-1 yr-1, respectively, for individual cont ributions of 28%, 65% and 7% to the total nitrogen loading from wet deposition, which was equal to 4.91 kg-N ha-1 yr-1. Poor et al (2001) estimated for 1996-1999 average annual fluxes of 1.74 and 2.40 kg-N ha-1 yr-1 for NH4 + and NO3 -, respectively. In aerosol samples (PM10) collected during rain events, NH4 + constituted 75.8 15.4 % of the DIN, and average concentrations of DIN and DON were 78.5 56.2 nmol-N m-3 and 6.3 2.6 nmol-N m-3, respectively. In particles, DON represented approximately 10.3 7.3 % of TDN.

PAGE 110

Sample 0246810121416 Conc(uM) 0 50 100 150 200 250 300 RWW-Na + ( M) RWW-Cl ( M) RWW-NO 3 ( M) RWW-NH 4 + ( M) RWW-DON ( M) Rainwater Composition T(oC) / po (mm hr-1) 0 10 20 30 40 po(mm hr-1) T(oC) PM10-2.5 Composition Sample 0246810121416 Conc(nmol m-3) 0 50 100 150 200 PM10-2.5Na+ PM10-2.5ClPM10-2.5NO3 PM10-2.5NH4 + PM10-2.5DON T(oC) / po (mm hr-1) 0 10 20 30 40 po(mm/hr) T(oC) PM 10-2.5 Composition Sample 0246810121416 Conc(nmol m-3 ) 0 50 100 150 200 PM2.5Na+ PM2.5ClPM2.5NO3 PM2.5NH4 + PM2.5DON PM 2.5 Composition 0 10 20 30 40 po(mm/hr) T(oC) T(oC) / po (mm hr-1) 0246810121416 0 50 100 150 200 250 300 RWW-NO3 (M) RWW-NH4 +(M) RWW-DON (M) 0 10 20 30 40 po(mm/hr) T(oC) DIN / DON Rainwater CompositionT(oC) / po (mm hr-1)Conc(uM)Sample Figure 5-1. Composition of Rainwater and Aerosol Samples Coll ected on the Gandy Monitoring Site During July-August 2005.

PAGE 111

90 Details about the composition of fine and coarse particles are shown in Table 52. Aerosol samples revealed that on average, 97.0 2.8 % of the total NH4 + was in the fine fraction, and 80.7 3.4 % of the NO3 was in the coarse particle fraction. Of the DON measured in aerosols, 68.0 17.0% was in the fine particle fraction. Average concentrations of DIN in PM2.5 and PM10-2.5 samples were 77.9 63.4 nmol-N m-3 and 12.3 5.2 nmol-N m-3, respectively; and average values for DON in the same fractions were 5.1 2.9 nmol-N m-3 and 2.6 0.8 nmol-N m-3. Measurements at the same site from March-April 2005 showed similar tendencies (Caldern et al. 2006c). These integrated measurements also agree with si ze-fractionated measur ements made at the same site (Campbell et al. 2002; Evans et al. 2004). Table 5-2. Composition of PM10 Samples from the Gandy Bridge Monitoring Site, Tampa, FL, July-September 2005. Average Concentration Standard Deviation PM2.5 PM10-2.5 Na+(nmol m-3) (n=19) 7.8 8.5 30.8 28.9 Cl-(nmol m-3) (n=19) 1.0 2.0 27.7 23.9 DIN-NH4 +(nmol-N m-3) (n=19) 75.4 64.6 1.5 1.0 DIN-NO3 -(nmol-N m-3) (n=19) 2.5 1.7 10.8 5.1 DIN(nmol-N m-3) (n=19) 77.9 63.4 12.3 5.2 DON(nmol-N m-3 ) (n=19) 5.1 2.9 2.6 0.8 DMA(pmol m-3) (n=17) 688 615 b.d. DMA/DON (%) (n=17) 12.8 6.7

PAGE 112

91 Dimethylamine (DMA) was observed in aqueous extracts from PM2.5 and PM102.5 samples, but could be quantitatively meas ured only in fine particles. Minimum and maximum values were, respectively, b.d. and 1911 pmol-N m-3, with an average concentration of 688 615 pmol-N m-3 (n=17), representing an average contribution of 12.8 6.7 % to the total DON concentration m easured in the same particle fraction. DMA concentrations were under the detection limit for all rainwater samples. Although the gradient concentration used in ion chro matographic analyses allowed separation of methylamine, dimethylamine, trimethylamine and diethanolamine, only DMA was detected. Because DMA is a very volatile substance and it was found in an appreciable quantity on fine particles, it was assumed to represent an important fraction of DON in the gas phase. Estimation of Dimethylamine (DMA) Gas Concentrations In the absence of experimental data, DMA gas concentrations were estimated by assuming vapor-liquid equilibrium between th e aqueous film on fine particles and the enveloping gas. Thermodynamic data and va por-liquid-solid equili brium relations for the NH4 +/SO4 -2/NO3 -/H+/H2O system are integrated in the AIM 2 model (Clegg et al. 1998). NO3 was excluded from the equilibrium calculations because experimental studies have found that in Tampa, FL, nitrat e is present as a coar se particle (Campbell et al. 2002; Evans et al. 2004). In the simple calculation option, AIM 2 is able to calculate composition and mass of all possible liquid, solid and gas phas es for given values of relative humidity, temperature and number of moles of nitrat e, hydrogen ions, sulfate and ammonium

PAGE 113

92 present in a unit of air volume (Clegg et al. 1998). In this case, measured concentrations of ammonium and hydronium ions were input to the model; sulfate concentration was assumed to be half of sum of the ammonium plus hydronium concentration to balance el ectrical charges. The H+ concentration was determined using the pH of PM2.5 extracts and the volume of sampled air. The formation of solids was avoided in order to describe better those meta-stable c onditions of supersaturated aqueous aerosols expected to exist at high relative humidity values such as those occurring just before and during a rain event. DMA concentrations in the liquid phase (wat er in fine particles) were calculated using the AIM-2 predicted water content for a certain set of conditions. As a rough estimate, activity coefficients for DMA and water were assumed to be independent of electrolyte concentrations and were de termined using the Non-Random Two-Liquid (NRTL) model with temperature independent parameters. Considering the gamma-phi approach the vapor-liquid equilibrium is reached when at the same temperature and pressure, th e fugacity of each species is the same in both phasesL i v if fˆ ˆ Fugacities on vapor and liquid phases can be written as: P y f f x fi V i V i L i i i L i ˆ ˆ Equation 5-2 Where i is activity coefficient of species i in solution, xi is the molar fraction of species i in the liquid phase, fi L is the fugacity of pure species i in the liquid phase, V i is the fugacity coefficient for sp ecies i in the vapor phase; yi is the molar fraction of species i in the vapor phase, and P is the total pressure. When the pressure is low (under

PAGE 114

93 3 atm in most of the cases), the fugacity coefficient V i is approximately equal to one, and L ifcan be approximated to the vapor pre ssure of the pure co mpound; therefore the molar concentration of species i in the gas phase can be written: RT P x RT P y RT P Csat i i i i i gi Equation 5-3 RT A RT g g RT A RT g g G G G x x G G x x G x G x x G G x x G x21 11 21 21 12 22 12 12 21 21 21 12 12 12 2 12 1 2 21 21 12 1 2 12 12 2 1 2 2 21 2 1 12 12 21 2 1 21 21 2 2 1 exp exp ln ln Equation 5-4 NRTL parameters for the dimethylamine (1)water (2) system at 298.15 K have been estimated to be A12=-615.7161, A21=518.7172 and 12=0.3062 for R=1.98721 cal/mol/K. Vapor pressure of pure subs tances was calculated using Antoine’s expression (Equation 5-5) with consta nts indicated in Table 5-3 (Gmehling et al. 1977). c T b a Psat log Equation 5-5

PAGE 115

94 Table 5-3. Antoine’s Constant to Calculat e Vapor Pressure of Pure Substances ( T in oC, P in mmHg) (Gmehling et al. 1977). Antoine’s constanta b c Region (oC) Dimethylamine 7.0812 960.242221.667[-71; 7] Water 8.071311730.63233.426[1; 100] Predicted gas-phase concentrations ranged from 17.3 to 502.9 pmol-N m-3, for an average of 107.4 176.9 pmol-N m-3. The estimated and measured values for DMA aerosol and gas concentrations are on the same order of magnitude as those measured for atmospheric samples collected over the Arabian Sea. For this maritime environment, the minimum and maximum aeros ol concentrations in the first sampling period were equal to 23 and 96 pmol m-3, respectively, with an average of 45 pmol m-3; and 81 and 379 pmol m-3 with an average of 241 pmol m-3 in the second sampling period. Minimum and maximum gas concentr ations were 16 and 65 pmol m-3 in the first sampling period, and 50 and 870 pmol m-3 for the second sampling period. Average gaseous concentrations were 39 and 196 pmol m-3 for first and second sampling periods, respectively (Gibb et al. 1999a). Van Neste et al (1987) observed similar magnitudes of DMA gaseous concentrations on atmosphe ric samples from Island, USA and Hawaii, USA, in which the average concentrations were 93 51 and 240 40 pmol m-3. Air samples in southern Sweden showed a total amine concentration range from 160-2,800 pmol m-3 with DMA values ranging from 23 to 310 pmol m-3 and MA and TMA as the

PAGE 116

95 largest contributors with concentr ations from 46 to 1,500 pmol m-3 (Gronberg et al. 1992). Aliphatic amines are short-range transp ort species with a s hort lifetime of 1-5 days (Angelino et al. 2001; Neff et al. 2002) due to their very high reactivity towards the OH radical (Schade and Crutzen, 1995). This suggests that the presence of DMA in the samples is due to local sources. Particle and Gas Scavenging Contributions To gain an understanding a bout droplet/particle intera ctions and their influence on wet deposition fluxes, particle and gas scavenging rates fo r species were calculated using DON and DMA concentrations for sa mples from the Gandy Bridge monitoring site. Gas scavenging rates for dimethylam ine were calculated using a value of Hc of 57 M/atm at 298.15K and H/R (K) equal to 4000 K (Sander, 1999). DSDs for the inorganic species were a ssumed to follow lognormal distributions as a function of the particle diameter. Us ing data from the same monitoring site, prediction of particle size distributions of these speci es were successfully done by fitting experimental data to find the best parameter values of dgm and (Table 4-1) (Campbell, 2005). It was assumed that the DON size distribution was identical to the one for NH4 +, because fine particle DON and NH4 + were well correlated (r= 0.831, n=45, Figure 5-3).

PAGE 117

96 Figure 5-2. Dissolved Organic and Inor ganic Nitrogen Concentrations from PM2.5 Samples Collected at the Gandy Bridge Mon itoring Site Between November (2004) to September (2005). DMA was considered as a reversible ga s with a Henry’s constant of 57 M atm-1 at 298.15 K. In the absence of better info rmation about DMA gas concentrations, the gas scavenging contribution was calculated a ssuming constant gas concentration during the rain events. Non significant differences between the particle scavenging results were found when the particle concentr ations were considered cons tant or variable during the rain events. Low particle s cavenging coefficients for NH4 +-enriched particles caused this result. Calculation routines to solve equations for scavenging rates were programmed using Matlab as the programming language. Double integrals were solved using the

PAGE 118

97 Simpson-composite rule with dp=(dpmax-dpmin) 1000-1, while simple integrals were solved using the quadl (Lobbatto-quadrature) function with a to lerance level of 10-16. Non-significant changes were seen for results when the tolerance level and dp values were decreased from the selected values. By using the lognormal distribution (L), the Massambani-Morales gamma distribution (MM) and the modi fied Marshall-Palmer distribu tion (MP), the effect of the DSD on the final estimated rainwater concentrations was quantified. ANOVA studies showed a non-significant effect of DSD over estimated concentrations for all species, likely because particle scavenging depends mo re on the droplet/particle size ratio that influences collision efficiencies and in the particle concentration than on the number and size of droplets available to collect particles. Particle scavenging of DON explained 0.9 0.2% of the nitr ogen in rainfall (Table 5-3). Fine particles are poorly scavenged due to low collision efficiencies, even in the presence of a high number of large droplets. Th is is especially evident for species such as NH4 + and DON which were assumed to be concentrated on particles with a diameter mode of 0.4 m. Collection efficiencies are minimal for particles between 0.1 and 1 m in diameter (Seinfeld a nd Pandis, 1998). Because 99% of DON concentrations in rainwater remained unexplained, it was a ssumed that the organic nitrogen wet deposition flux is the consequence of gas scavenging. Due to the presence of DMA in extracts from fine particulate matter, it is suggested that an important fraction of DON could be comp osed by aliphatic amines of low molecular weight. DMA is not released individually from natural sources and is

PAGE 119

98 usually found in correlati on with other very volatile aliphatic amines such as methylamine, trimethylamine, diethanolamin e (Schade and Crutzen, 1995), therefore, it is logical to assume that these other species were present in atmo spheric concentrations not detectable by the performed method. None of these species could be identified in rainwater samples. Scavenging calculati ons for DMA showed predicted rainwater concentrations were below the defi ned analytical detection levels. To quantify the contribution of DMA, gas and particle DMA scavenging rates were estimated, assuming that DMA fo llowed the same size distribution as NH4 +. The average modeled DMA contribution to DON con centration in rainwater was 0.4 0.7%. DMA particle scavenging accounted for 11. 7 7.1% of the contribution from DONparticle scavenging. Beca use MMA and TMA have chemical properties (Henry’s constant, gas diffusivity, etc.) similar to DMA, they likely have similar gas scavenging rates, especially if the particle-gas system is at vapor-liquid equilibrium. A simple and logical conclusion is that gaseous organic nitrogen species, different from aliphatic amines, are responsible for 99.5% of DON rainwater concentrations.

PAGE 120

Table 5-4. Average Dissolved Organic Nitrogen (DON) and Dimet hylamine (DMA) Concentrations in Rainwater Concentrations Predicted by Particle Scavenging Using the Aver age of Three Different Drop Size Distributions. DON DMA Day ts (hr) po (mm/hr) T(oC) Rainwater concentration explained by particle scavenging ( mol-N L-1) % Concentration explained by particle scavenging Rainwater concentration explained by particle and gas scavenging ( mol-N L-1) % DON concentration explained by DMA particle and gas scavenging %DMA Contribution to DONRainwater concentration explained by particle 07/20/2005 1 8.9 29.0 0.021 0.001 0.30 0.01 07/24/2005 3 3.0 27.2 0.036 0.005 0.35 0.05 08/06/2005 3 7.6 26.6 0.041 0.002 1.29 0.06 0.01020.0005 0.3220.013 20.7 08/07/2005 1 38.1 26.1 0.00140.0001 0.1260.011 08/08/2005 4 8.6 27.2 0.002810.0001 2.1890.082

PAGE 121

Table 5-4 (continued) DON DMA Day ts(hr) po (mm/hr) T(oC) Rainwater concentration explained by particle scavenging ( mol-N L-1) % Concentration explained by particle scavenging Rainwater concentration explained by particle and gas scavenging ( mol-N L-1) % DON concentration explained by DMA particle and gas scavenging %DMA Contribution to DONRainwater concentration explained by particle 08/09/2005 1 2.5 28.4 0.058 0.009 2.29 0.34 0.00510.0007 0.2020.029 6.7 08/22/2005 3 0.6 28.2 0.050 0.015 0.58 0.14 0.00600.0019 0.0950.026 10.5 08/23/2005 1 2.5 28.2 0.030 0.005 1.21 0.18 0.00220.0003 0.0830.012 5.18 08/24/2005 1 0.8 29.0 0.017 0.004 0.42 0.11 08/28/2005 1 2.3 28.5 0.026 0.004 0.94 0.15 09/01/2005 2 2.5 28.7 0.015 0.002 0.49 0.07 0.00320.0005 0.1020.015 15.9

PAGE 122

101 In the case of DIN, particle scave nging calculations (under the assumption of constant particle concentration) explaine d 3.02 2.9 % and 85.9 38.9 % of the NH4 + and NO3 concentrations in rainwater, respectively. The average precipitation rate per rain event was used instead of the hourly-val ues of these variables. In the case of NO3 it is interesting to highlight that particle scavenging is responsible for almost all its concentration in rainwater samples. This agre es with the fact that nitric acid from NOx transformations tends to react with NaCl from sea spray to form NaNO3 leaving then only a small fraction of free gas to be scavenged by raindrops. (Caldern et al.,2006c ) showed that gas scavenging rates for NH3 and HNO3 can be added to NH4 + and NO3 particle scavenging rates to explain ~100% of their ionic concentr ations in rainwater samples. Summary DON represented 8.9 5.8% of the total nitrogen concentration in rainwater samples and its estimated annual flux was 0.34 kg-N ha-1 yr-1 or ~7% of the total rainfall nitrogen flux. Dimethylamine (DMA) was found in fine particulate matter and represented ~13% of DON found in this fraction. Its concen tration in par ticles and the corresponding equilibrium gas concentrati on explained through s cavenging processes just ~0.4% of the DON measured in rainfall. Gas scavenging of DON is assumed to be responsible for 99% of its wet deposition flux, and gas-phase organic nitrogen, including DMA, likely contri butes to the dry deposition of nitrogen to Tampa Bay.

PAGE 123

102 6. ESTIMATION OF THE PART ICLE AND GAS SCAVENGING CONTRIBUTIONS TO WET DEPOSITION OF INORGANIC NITROGEN Introduction The main forms of DIN species found in rainwater are NO3 and NH4 + ions, and this part of the study is focused in explaini ng their concentrations in rainwater using gas and particle scavenging rates. The goal is to understand better th e contribution of gas and particle nitrogen species to the measured wet deposition flux of nitrogen, to explore the uncertainty associated with common assumptions applied to below-cloued scavenging models, and to obtai n scavenging coefficients fo r use with the air quality models. Hourly-measured gas concentrations and 24-hr integrated PM10 concentrations were used in conjunction with a belowcloud scavenging model to explain ammonium and nitrate concentration in rainwater samp les collected at the Gandy Bridge monitoring site, FL. All rain events were treated as if they were composed by one-hour intervals with a precipitation rate constant an d equal to its hourly average value. During scavenging, gas and particle con centrations not only change due to the interactions with the droplets but also by direct emissions from industrial sources, gasto-particle conversion, chemical reactions (e.g. NOx transformation to HNO3), diffusive

PAGE 124

103 and convective transport, etc. Many models have been proposed to describe such a complicated atmospheric process in orde r to predict and/or explain rainwater concentrations. Air pollutant scavenging mode ls vary in complexity from the simple model based on the ratio of rainwater to air concentration (Samara and Tsitouridou, 2000) to the two-dimensional description of the in-cloud and below-cloud scavenging with detailed cloud, rain and particle microphysic s (Wurzler, 1998). The model applicability depends on the availabl e data and computational tools. Below-cloud scavenging models can be classified into four groups by complexity. A first group of models uses th e concept of scavenging ratio (SR) or the ratio between the concentration of a chemical species in rainwater (Crain) to its concentration in the particle phase (Cparticle). SR is usually defined as air particle rainC C SR a dimensionless quantity where air is the air density. The implicit assumption is that particle and rainwater are directly and exclusively related (Samara and Tsitouridou, 2000). Scavenging rati os can help to identify those species that are preferentially rem oved from the atmosphere and can be used to predict concentration in one of the phases in terms of the other. Because of their simplicity and the lack of agreement in the requirements for the determination of the concentrations, SRs are poor predictors of actua l rainfall concen trations(Casado et al. 1996; Samara and Tsitouridou, 2000). SRs can vary several orders of magnitude even for the same species during the same rain event at th e same site; these variations are the consequences of changes not considered in side the SR-model such as changes in the rain intensity, different particle size di stribution (PSD) and dr op size distributions

PAGE 125

104 (DSD), convective and diffusive transport, c oncentration vertical profile etc. In the absence of strong gas scavenging contributi ons to the rainwater concentration, SRs should higher for water-soluble species present in coarse particles such as Ca, Na and K than for SO4 -2 and NO3 usually present in fine particles (Casado et al. 1996; Samara and Tsitouridou, 2000). SRs are useful as qua litative indicators of the characteristics and contributions of the scavenging process occurring at a certain monitoring site, but they cannot be used to predict accu rately rainwater concentrations. At the next level of complexity, models include experimental determination of scavenging coefficients that are depende nt on both the scaven ged species and the rainfall intensity (Mizak et al ., 2005; Maria and Russell, 2005; Pandey et al. 2002; Minoura and Iwasaka, 1997). They require continuous rainwater, gas and aerosol sampling to determine the variation of the chemical composition in the rainwater at small time intervals after the rain has started. Exclusionary criteria such as high or gusty winds or low rainfall rates have been app lied to experimental data to improve the overall validity of the derive d scavenging coefficients. Measurements are selected from the experimental set and used in the calculation if they satisfy the requirements of stable atmospheric conditions under whic h the effects of convective and diffusive transport as well as direct emissions from sources ar e much smaller and do not significantly influence the concentration changes in the ra in droplets. Stable atmospheric conditions have been set to be rain intensity below 15 mm hr-1, wind speed below 2.5 m s-1 and width of variation of wind direction of less than 90o (Minoura and Iwasaka, 1997), or steady wind speeds of less than 3 m s-1, constant wind directi on and rain intensity

PAGE 126

105 greater than 0.4 mm hr-1 for longer than 0.5 hr (Maria and Russell, 2005). Under stable conditions the gas/particle scavenging proce ss is assumed to be a first-order process with a rate of change equal to the empiri cal scavenging coefficient. Particle and gas scavenging coefficients are ca lculated using the chemical evolution of the aerosol/gas concentration during precipitation. Scavengi ng coefficients tend to increase as rain intensity increases and have been descri bed with both linear (M inoura and Iwasaka, 1997) and power law models (Asman, 1995; Okita et al ., 1996; Pandey et al ., 2002). Minoura and Iwasaka (1997) f ound a linear correlation betw een empirical particle scavenging coefficients for NO3 and SO4 -2 and rain intensities from 0 to 12 mm hr-1; however a potential function of the kind b op a where po is the rain intensity (0.1 to 10 mm hr-1) has also been found and used for SO4 -2 (Okita et al. 1996; Pandey et al. 2002). Even when the particle scavenging is usua lly assumed to be just mechanical and independent on the particle chemical co mposition, empirical particle scavenging coefficients have shown to depend on the water solubility of the scavenged species. Under stable conditions the NO3 scavenging coefficient proved to increase more with rain intensity increases and was determined to be 5.2 10-4 s-1, twice the SO4 -2 coefficient, equal to 2.2 10-4 s-1 (Minoura and Iwasaka, 1997). This situation has also been studied by normalization of the empirical particle scavenging coefficients with the rain intensity in order to isolate the rain intensity effect. Variations in the normalized coefficients have been rela ted not only with composition bu t also with the mixing state and age of the aerosol particles. A ssuming monodisperse PSD and DSD and

PAGE 127

106 exponential decay of particle concentration, sulfate and NH4 + scavenging coefficients were determined to be three times larger than those for organic matter and silicate species, showing that water-soluble inorgani c species are preferentially scavenged. The same information was obtained by analyzi ng the mass fractions of species in the residual aerosols after rain (Maria and Russell, 2005). Ce rtain organic compounds such as long length organic alcohols, acids, aldehydes, esters, ke tones and amines, can act as surfactants on the particle surface and change their ability to be dissolved onto rain droplets or after dissolution can al ter the droplet’s ability to ab sorb or desorb gases in its interior (Gill et al. 1983). Both effects certainly cause preferential scavenging according the relationship between the rain droplet, gas and particle chemical compositions. Models at the third level of complexity re quire as input a deta iled description of the PSD and DSD. Particle scavenging is studied as a mechanical process depending mainly on the collision effici ency of particle-droplet en counters (Seinfeld and Pandis, 1998; Slinn, 1977), while gas scavenging is de pendent on the interfacial mass transfer and absorption/desorption of th e gaseous species into raindrops. The changes in the droplet concentration due to these interactions are integrated over the DSD and PSD to predict the rainwater concentrations. Variati on in the gas and particle concentrations during rain events are usually modeled as exponential decay and removal rates are proportional to the scavenging coefficients. Th e rain intensity effect on the scavenging is accounted for by the DSD and the number of droplets per size category ( Oberholzer et al. 1993; Volken and Schumann, 1993; Asman, 1995 ; Goncalves et al. 2000;

PAGE 128

107 Mircea et al. 2000; Andronache et al. 2002; Goncalves et al. 2003; Chate et al ., 2003; Chate and Pranesha, 2004; Mircea et al. 2004; Zhao and Zheng, 2006). Special cases of this group are the Eule rian model by Kumar (1985) which employs a monodisperse DSD to determine the dynamic evolution of the gas-phase depleti on and aqueous-phase accumulation during rainfall scavenging; and a below-cloud scavenging model by Chate (2005) that includes particle collecti on by thermophoresis, diffusiophoresis, and electrical forces during thunde rstorms. Kumar (1985) also includes expressions for the changes in the rain droplet and gas concentration with time and height. The most complex models keep the deta iled descriptions of the DSD and PSD but include multidimensional transfer phenomena and microphysics. Wurzler (1998), for example, proposed a model to describe the in-cloud and below-cloud scavenging of HNO3. The main model assumptions are well-m ixed droplets, exponential decay with height of particle concentr ations, and acidic nuclei for the droplet formation. The changes in the PSD are assumed to be cause d by activation of particles to droplets, diffusional growth (by condensation or eva poration) and impaction scavenging. The changes in the DSD are due also to activa tion of particles to droplets and diffusional growth, as well as to the collision, coales cence and break-up of drops, and to the gasuptake and particle-uptake. A ll the equations describing the change in the concentration of the particle, rain droplets and gas incl ude the effects of advective transport and turbulent mixing. Modeling with detailed ra ther than parameterized microphysics for cloud processing yielded higher nitrogen wet deposition rates (Wurzler, 1998).

PAGE 129

108 This is the more detailed and complex se t of models to describe the scavenging process, however its applicability to pred ict rainwater concentrations is limited by information about all the variables involve d in the model at the monitoring site. The research goal is to use a belowcloud scavenging model for the evaluation of the contributions of gas and particle scavenging to rainwa ter concentration of samples collected at Tampa Bay. Because the aerosol samples were not always collected at stable atmospheric conditions, tw o different cases were studied, the first one assuming constant particle and gas concentra tions occurring during all rain intervals; and the second one assuming an exponential decay with time of the concentrations per rain interval in terms of the particle a nd gas scavenging coefficients defined as the correspondent precipitation rate In the first case, it is assumed that even when the concentrations decrease by scavenging th ey are compensated by convective and/or diffusive transport, direct emissions from sources, or gas-to-particle conversion. The second case describes better th e scavenging process under st able conditions where all the factors mentioned be fore are negligible. A below-cloud scavenging model was c hosen from among the available ones in the third group with particle scavengi ng rates defined by the Slinn’s collision efficiency model (1977) presented by Seinfeld and Pandis (1998) with the particular use of the Kumar’s approach (1985) to determine the changes in the droplet concentrations after gas scavenging. Model assumptions inherent in our be low-cloud scavenging model included negligible contributi ons of in-cloud scave nging, irreversible HNO3 and NH3 gas

PAGE 130

109 transfer, no change in the DSD during th e rainfall event and uniform vertical concentration profiles of gases and part icles as represented by near surface measurements. Hourly gas concentrations were used to determine the initial level of the trace pollutants at the time the rain started for the first case and the co rrespondent values for all rain intervals. In the ab sence of continuous measurements of particle concentrations, we fixed them to be those from 24-hr in tegrated samples and their exponential decay was estimated to be dependent to the precipit ation rate measured for the time interval. As a test of the model performance, s cavenging rates were also calculated for sodium (Na+), as Na+ is a non-volatile species and its co ncentration in rainwater should be completely explained by particle scav enging when there are stable atmospheric conditions. Experimental Methods Field Sampling and Sample Processing Aerosol and rainwater sample collection we re explained in detail in the last chapter. All procedures and analyses were the same. Continuous HNO3 concentrations were obtained from a dual-channel chemiluminescence analyzer based on the Thermo Environmental 42C-TL NOx instrument modified with dual 350 oC molybdenum catalysts by Atmospheric Research and Analysis, Inc. Both channels redu ced all oxidized nitrogen species (NOy) to nitrogen oxide (NO). In the second cha nnel, a potassium chloride-coated denuder selectively removed the HNO3 ahead of the converter. The HNO3 concentration was

PAGE 131

110 operationally defined as the differenc e between the denuded and un-denuded NOy. A schematic of this system is shown in Figure 6-1. The converters for the chemiluminescent system were housed in a weatherproof enclosure mounted on a tower at a height of 10 m. The inlet line for the undenuded cha nnel was constructed of PFA Teflon tubing with a 0.05 sec residence time to minimize line losses of HNO3; the open end of the denuder served as inlet for the denuded channel. The method detection level of this system to HNO3 is 2 nmol m-3 with 60-min resolution (Arnold et al ., 2006)Continuous NH3 concentrations were obtained from a dual-channel chemiluminescence analyzer based on the Thermo Environmental 42C-TL NOx instrument. The instrument design is nearly identical to the HNO3 system described above, except that each ch annel has an additional 550oC platinum catalyst to oxidize ammonia/ammonium (NHy) to NO, followed by a 350oC molybdenum catalyst to reduce any NO2 to NO. The inlets of both channels have a 142-mm Na2CO3 denuder to remove HNO2 and HNO3. In one of the two channels, a phosphoric acid-coated denuder selectively removes the NH3 ahead of the converter. The NH3 concentration was operationally defined as the difference be tween the denuded and un-denuded N (e.g. N = NH3+ NH4 + + NO+ NO2 + other reduced and oxidized nitrogen species not removed by the Na2CO3 denuder). The open end of the denuders served as inlet for both channels. The converters for the chemilu minescence system were housed in a weatherproof enclosure mounted on the top of a trailer at a height of 3.5 m. Inlet lines were short and constructed of PFA Teflon tubing with a 0.05 sec residence time to

PAGE 132

111 minimize line losses of NH3. The method detection leve l of this system for NH3 is 2.0 nmol m-3 with 60-min resolution. NH3 and HNO3 concentrations obtained w ith the chemiluminescence method were compared with measurements made with the collocated annular denuder system and show reasonable agreement (Arnorld et al ., 2006; Hartsell et al ., 2006) Aerosols and Rainwater Concentrations Table 6-1 shows the average values for NH4 + and NO3 concentrations as well as dry bulb temperature (T, oC) during all 24-hr integrated sampling periods. Values for the base cloud height and the total duration and precipitation rate of the rain event are also registered. Table 6-2 shows details about the precipit ation rates and NH3 and HNO3 concentrations for all 1-hr periods composi ng each rain event. The base cloud height (h) was obtained at the starting time of the rain event and read as the cloud height of the first layer in ascending order from the ground.

PAGE 133

Table 6-1. Aerosol (PM10) and Rainwater Concentrations an d Average Meteorological Conditions at the Ga ndy Bridge Monitoring Site. Sampling Day NH4 +-PM10 (nmol-N m-3) NH4 +-RWW ( M-N) NO3 --PM10 (nmol-N m-3) NO3 -RWW ( M-N) CNH3 (nmol-N m-3) CHNO3 (nmol-N m-3) h(m) ts (hr) pototal (inches) T (oC) 7/20/2005 39.0 15.6 14.4 34.1 148.4 14.0 792.5 1 0.35 29.0 7/24/2005 116.1 12.0 13.7 24.1 71.3 6.8 853.4 3 0.35 27.2 8/06/2005 15.2 8.2 12.4 19.1 59.2 6.3 792.5 3 0.9 26.6 8/07/2005 12.6 5.5 9.7 12.4 37.2 27.6 823.0 1 1.5 26.1 8/08/2005 43.5 8.1 9.3 15.1 79.1 18.3 914.4 4 1.35 27.2 8/09/2005 155.4 8.4 10.6 26.0 29.7 25.7 762.0 1 0.1 28.4 8/14/2005 94.0 46.9 11.5 185.6 27.1 26.4 792.5 1 0.02 27.7

PAGE 134

Table 6-1. (continued) Sampling Day NH4 +-PM10 (nmol-N m-3) NH4 +-RWW ( M-N) NO3 --PM10 (nmol-N m-3) NO3 -RWW ( M-N) CNH3 (nmol-N m-3) CHNO3 (nmol-N m-3) h(m) ts (hr) pototal (inches) T (oC) 8/22/2005 167.9 25.1 11.3 105.3 96.6 33.7 914.4 2 0.07 28.2 8/23/2005 81.7 19.4 11.8 48.7 34.3 13.6 914.4 1 0.1 28.2 8/24/2005 17.9 20.4 16.0 127.1 89.8 36.0 853.4 1 0.03 29.1 8/28/2005 25.2 11.6 10.9 15.2 136.6 6.9 914.4 1 0.09 28.5 9/01/2005 54.6 11.9 14.1 28.6 32.0 16.0 762.0 2 0.2 28.7

PAGE 135

Table 6-2. Hourly precipitati on rates and gas concentrations dur ing rain events at the Gandy Br idge monitoring site, Tampa, F L. Sampling Day po (mm hr-1) CNH3 (nmol-N m-3) CHNO3 (nmol-N m-3) 7/20/2005 8.9 148.4 14.0 7/24/2005 3.0 3.0 3.0 71.3 50.9 117.4 6.8 5.6 11.2 8/06/2005 7.6 2.5 12.7 59.2 31.4 65.5 6.3 5.8 8.3 8/07/2005 38.1 37.2 27.5 8/08/2005 26.7 2.5 2.5 2.5 79.1 56.0 83.2 77.0 18.3 7.6 14.6 15.0 8/09/2005 2.5 29.7 25.7 8/22/2005 0.9 0.9 96.6 303.3 33.7 31.9 8/23/2005 2.5 34.3 13.6 8/24/2005 0.8 89.8 36.0 8/28/2005 2.3 136.6 6.9 9/01/2005 2.5 2.5 32.0 231.3 16.0 14.6

PAGE 136

Table 6-3. Contribution of Par ticle and Gas Scavenging to Rainwater Concentra tions of Samples Collect ed at the Gandy Bridge Monitoring Site, FL. NH4 + NO3 Na+ Sampling Day %Concentration explained by particle scavenging %Concentration explained by gas scavenging %Concentration explained by total scavenging %Concentration explained by particle scavenging %Concentration explained by gas scavenging %Concentration explained by total scavenging %Concentration explained by particle scavenging 7/20/2005 10.1 55.97.9 55.97.9 31.65.8 20.3 33.65.5 95.218 7/24/2005 4.40.6 71.916.6 71.916.6 423.5 2.70.7 44.72.9 51.34.4 8/06/2005 0.70.04 40.75.4 40.75.4 24.85.9 1.80.3 26.65.7 38.89.4 8/07/2005 0.70.1 13.61.1 13.61.1 22.210 4.20.6 26.39.5 53.724.7 8/08/2005 1.80.1 51.48.9 51.48.9 17.55.6 4.50.6 22.15.4 4715.2 8/09/2005 91.3 36.39.5 36.39.5 55.71.4 7.72.1 63.42.2 186.35

PAGE 137

Table 6-3. (continued) NH4 + NO3 Na+ Sampling Day %-Concentration explained by particle scavenging %Concentration explained by gas scavenging %Concentration explained by total scavenging %Concentration explained by particle scavenging %Concentration explained by gas scavenging %Concentration explained by total scavenging %Concentration explained by particle scavenging 8/22/2005 3.80.9 145.370.8 149.171.7 16.60.4 4.22.1 20.82 86.82.5 8/23/2005 20.3 21.75.7 23.85.9 33.20.9 2.60.7 35.81 121.83.3 8/24/2005 0.50.1 79.641.3 80.141.5 22.91 3.72 26.72.9 165.86.7 8/28/2005 1.10.2 151.742.8 152.743 101.21.9 4.41.3 105.62.2 154.83 9/01/2005 2.20.3 11429.7 116.230 50.42.6 4.21.2 54.51.8 95.25.3

PAGE 138

Table 6-4. Contribution of Par ticle and Gas Scavenging to Rainwater Concentra tions of Samples Collect ed at the Gandy Bridge Monitoring Site, FL considering Consta nt Particle and Gas Concentrations. NH4 + NO3 Na+ Sampling Day %Concentration explained by particle scavenging %Concentration explained by gas scavenging %Concentration explained by total scavenging %Concentration explained by particle scavenging %Concentration explained by gas scavenging %Concentration explained by total scavenging %Concentration explained by particle scavenging* 7/20/2005 1.050.04 93.321.9 94.321.9 76.21.1 2.90.7 79.10.8 236.43.3 7/24/2005 4.820.64 91.427.7 96.228.3 111.85.8 3.21 1156.7 75.72.8 8/06/2005 0.790.04 69.716.9 70.516.8 118.62 2.70.6 121.31.7 827.5 8/07/2005 0.780.08 49.524.4 50.324.4 127.58.2 11.75.8 139.310.9 325.220.6 8/08/2005 2.160.1 101.923.3 104.123.2 108.33 8.22.3 116.51.7 219.124.2 8/09/2005 9.341.38 45.214.9 54.516.3 80.74.6 93 89.77.5 273.515.4

PAGE 139

Table 6-4. (continued) NH4 + NO3 Na+ Sampling Day %Concentration explained by particle scavenging %Concentration explained by gas scavenging %Concentration explained by total scavenging %Concentration explained by particle scavenging %Concentration explained by gas scavenging %Concentration explained by total scavenging %Concentration explained by particle scavenging* 8/22/2005 3.930.95 16386 166.986.9 22.41.9 4.52.4 274.2 101.84.1 8/23/2005 2.130.31 278.9 29.19.2 48.12.8 31 51.23.7 178.810.1 8/24/2005 0.530.13 88.449.1 88.949.2 26.42.3 42.2 30.44.5 191.816.3 8/28/2005 1.120.18 185.864.8 186.965 142.48.6 5.11.8 147.510.4 220.513.1 9/01/2005 2.320.34 141.846.9 144.147.2 97.35.6 4.91.6 102.17.1 136.63.3 *: Considering different particle concentr ations per hour of rain according the exp onential decay but constant concentration du ring scavenging calculations

PAGE 140

119 The average wind speed per rain event was equal to 2.1 1.3 m s-1 ranging from 0.5 m s-1 to 4.04 m s-1, the average wind direction was 209.9 136.3 deg ranging from 14.5 to 341.7 deg and the average standard deviation was 40.9 36.6 deg ranging from 5 to 94 deg. Between all the sampling days, only the days 07/20/2005, 07/24/2005, 08/22/2005 and 08/24/2005 satisfied all the requirements of stable atmospheric conditions. DIN gases such as ammonia and nitric acid are very soluble gases and their effective values of the Henry’s constants are so large that they absorb irreversibly into the rain droplets. For example, ammonia has effective values of the Henry’s coefficient that are higher than 1*106 at pH levels below 5 (Mizak et al. 2005). For the gas scavenging calculations presente d, pH on rainwater samples (n=16) showed an average value of 4.56 0.65, and therefore ammonia and nitric acid were assumed to behave as irreversible gases. Two set of equations were used to calc ulate the gas scavenging contribution to the wet deposition fluxes of NH4 + and NO3 -. The first one assumed constant gas concentrations and particle concentrations and the sec ond one assumed an exponential decay of these variables. Integrals on de position and volume weighted concentrations were solved by the Simpson-composite rule with dp=(dpmax-dpmin) 1000-1 and the quadl (Lobbatto-quadrature) Matlab -built-in function with a tolerance level of 10-16. All routines were programmed using Matlab. Table 6-3 shows the fraction of th e rainwater concentration of NH4 +, NO3 and Na+ explained using particle and gas scav enging rates. The standard deviations

PAGE 141

120 presented per sampling day were calculated fr om the results of the three different drop size distributions (DSD) used to approximate the rain spectra on the site. The standard deviations of the contributions of gas s cavenging were bigger for those of particle scavenging because gas absorption rates are strongly dependent on the droplet surface and this parameter is very different from one DSD to the other. Figure 6-2 shows how the particle scave nging changes with the particle size and precipitation rate. At high precipitation rates the particle scavenging coefficients from the log-normal drop size dist ribution (L-DSD) tend to be very similar to those calculated using the Marshall-Palmer dr op size distribution (MP-DSD), while the Massambani-Morales or gamma-distributi on (MM-DSD) produced lower values. At low precipitation rates the diffe rences between the three DS Ds were more evident for particles under 2 m in diameter. When the three DSD are used in conjunction with the particle concentrations to determine the contribution of particle scavenging to rainwater concentrations it was found for the measured range of precipitation rates that they do not differ from each other. ANOVA results showed a non significant difference at 95% confidence level between the mean values of each variable for each DSD in each scavenging regime (particle or gas). This result disagree s with those found previously in sensitivity test for HNO3 scavenging coefficients with respect to the variation on the DSD by Mircea et al .(2004). They used analytical expressions to determine HNO3 scavenging coefficients with the lognormal DSD and a gamma-DSD and found a variation up to 50 % in them due to change s in the DSD. They concluded that the function chosen to describe the DSD has a marked influence on the values of the gas

PAGE 142

121 scavenging coefficient (Mircea et al. 2004). When the gas scavenging coefficients for HNO3 were compared, ANOVA results did not s howed significant differences on the mean values at 95 % confidence level, bur for the case 2 or exponential decay the pvalue was 0.06, very close to the limit of 0.05. Causes of the disagr eement between this study and the one by Mircea et al .(2004) could be just on th e reduced number of rain events studied in this analysis. Additional di fferent between the two studies could be in the expressions used to determ ine droplet settling velocity. Figures 6-3 and 6-4 show cha nges to the initial PSDs of NH4 + and NO3 after one hour of rainfall at a precipit ation rate of 2.5 mm hr-1. The change in initial PSD and decrease in particle concentration is much more evident for NO3 with a dgm of 3.5 m than for NH4 + with dgm of 0.38 m, which reflects the low collection efficiency of the particles with diameters between 0.1 and 1. 0 m (Seinfeld and Pandis, 1998).

PAGE 143

122 Figure 6-1. Instrument Schematic Used for Gas Measurement (Hartsell et al ., 2006).

PAGE 144

123 10-3 10-2 10-1 100 101 10-5 10-4 10-3 10-2 10-1 100 101 Droplet Diameter (m)Scavenging Coefficient (hr -1)Variation of Particle Scavenging Coefficient L-DSD MM-DSD MP-DSD data4 data5 data6 101 100 Low po High po Figure 6-2. Particle Scavengi ng Coefficients at Different Precipitation Rates for Three Drop Size Distributions: L-L og-Normal, MM-Massambani and Morales, MP-MarshallPalmer. High po (precipitation rate) =8.9 mm hr-1. Low po=0.8 mm hr-1.

PAGE 145

124 Figure 6-3. Effect of the Drop Size Distribution on the Variation of Ammonium PM10 Concentration After One-hr Rain Event at 2.5 mm hr-1 (L-Log-normal, MMMassambani and Morales, MP-Marshall-Palmer).

PAGE 146

125 Figure 6-4. Effect of th e Drop Size Distribution on th e Variation of Nitrate PM10 Concentration After One-hr Rain Event at 2.5 mm hr-1 (L-Log-normal, MMMassambani and Morales, MP-Marshall-Palmer). Particle scavenging helped to explain just an average of 2.6 2.6 % of the experimental NH4 + rainwater concentration when constant particle concentrations were assumed (case 1) and 2.5 2.5 % when the exponential decay was considered (case 2). In contrast, particle scavenging explained 87.2 39.4 % of the NO3 concentration in the rainwater samples and 38.0 24.1 % for the second case. In the case of sodium, particle scavenging explained an aver age of 185.6 78.5% of its rainwater

PAGE 147

126 concentration in the case 1 and 93.2 53.2 % in the case 2. Sodium was included in the calculations to test the valid ity of the assumptions done in the particle scavenging model. Sodium is a non-volatile substan ce emitted only from particle sources and therefore only under stable atmospheric condi tions only the particle scavenging should be able to explain its concentration in ra inwater. These results showed satisfactory results especially for the case 2 with exponent ial decay of the particle concentration. Over predictions for the first case were e xpected because sodium-enriched particles have larger particle diameters and therefore faster scavenging rates. The assumption of constant particle concentration does not describe well this situation, while the exponential decay with time works better. This means that in this case, sodium particle concentrations really decreased during scave nging and there were negligible effects of particle generation from sea spray during ra in events with high wind speeds. Sodiumenriches particles comes from sea spray comes produced by breaking waves at wind speeds higher than 3-4 m s-1 (Andronache et al. 2002). In conclusion, the particle scavenging c ontribution to rainwate r concentrations is mainly affected by the particle size distri bution, and it will be responsible for an important fraction of the rainwater concentra tion only for those substances concentrated in particles outside the Greenfield gap. Gas scavenging proved to be more important for NH4 + than for NO3 concentrations in rainwater. Even when th e precursor gases of their ionic forms, NH3 and HNO3, can be both considered as irreversible gases and they have high tendencies to be dissolved in aqueous media, the s cavenging coefficients for both gases at the

PAGE 148

127 range of precipitation rates and temperatur e of this study differed from each other by 33.2 0.4 %. At the average dry bulb temperature for the sampling period the diffusivities of both gases are different by 45.9 %, with the higher values of NH3. In the first case of constant gas and pa rticle concentrations, the contribution of gas scavenging to NH3 concentrations in rainwater was in average 96.1 59.5 % and equal to 71.1 52.2 % for the second case w ith exponential decay. Similar results were seen using gas and aerosol measurements between two sampling sites at different altitudes. Gas scavenging of NH3 was more important than particle scavenging and was found to be responsible for 58-88 % of th e increase in the predicted rainwater concentration between the middle (1030ma sl) and the bottom (430 masl) station (Oberholzer et al. 1993). For NO3 -, the contributions of HNO3 scavenging are smaller than for NH3 and equal to 5.4 3.5 % and 3.8 1.9 % for th e cases 1 and 2 respectively. The contrast between the NH3/ NH4 + and the HNO3/ NO3 below-cloud scavenging processes agrees with the results of depo sition studies for 1996-1999 over the Tampa bay area where it was found that NH3 is responsible for the largest fr action of dry deposition rates while NO3 represents the biggest contribu tion to wet deposition rates (Poor et al. 2001). In general, the proposed below cloud s cavenging explained the 92.6 40.6 % of NO3 rainwater concentrations in the first case and 41.8 24.7 % in the second case; while for NH3 concentrations, the firs t case gave an average explanation of 98.7 59.3 % and 73.6 52.3 % for the second case. Although, the difference between the predicted and the measured rainwater con centrations for both scavenging cases are

PAGE 149

128 smaller for the NH3/ NH4 + set than for the HNO3/ NO3 set; the predicted values offer a close approximation to the real atmospheri c processes that define aerosol and gas concentrations at the site are between these extremes. Both cases proved to be applicable to describe the rain scave nging process for the average atmospheric conditions reported in the rain events. 24-hr integrated measurements were used to define particle concentrations and it is possible that this value was smaller than the real concentration at the starting time of rain event. Generally then, under predictions are expected to occur especially for NO3 -. It was also assumed a negligible influence of the in-cloud scavenging and the initial rain droplet concentrations were assumed to be e qual to zero, this coul d be not true in the Gandy Bridge monitoring site, especially befo re rain events when the relative humidity reaches maximum values at the cloud base and it is expected to have in-cloud gas scavenging. Cloud water concentrations in samples from South-Western China were found to be between 9 to 12 eq L-1. This value could be much higher than the expected in-cloud concentrations because while Tamp a Bay is a relatively unpolluted area, the monitored areas in China are very polluted by the use of high sulf ur-content fuels and unfavorable topography and climate (Tanner et al ., 1997). By assuming constant particle concentrations for NO3 scavenging the predicted concentrations were closer to the measured rainwater concentrations. This suggests that gas-to-particle convers ion processes or other particle sources participate during rain events and help to keep particle concentratio ns at higher levels than those expected if a exponential decay with time was occurring. The mean in the ratios between the

PAGE 150

129 contribution of particle and gas scavenging to NO3 concentrations in rainwater is significantly different for both cases (p-value <0.0001), while it is not significantly different in the case of NH4 +. For NH4 + in rainwater, the firs t case without exponential decay in the gas and particle concentration, also gave results closer to the observed values indicating that during rain events inputs from emissions counteract the decrease in gas concentration due to scavenging by raindrops. Particle concentration were assumed to remain constant in the vertical profile and equal to those measured at ground level, this could be causi ng over-predictions in the proposed below-cloud scavenging model, esp ecially in days with long rain events and strong precipitation rates. It has been found that ioni c particle concentrations decrease from the ground le vel to the cloud base height following different profiles including the exponential decay with height (Goncalves et al. 2003). Higher overpredictions are expected to occur for species concentrated in coarse particles, such as NO3 and Na+, because they have higher dry deposition velocities and should have a steeper decrease in concentration from the ground level to the cl oud base compared to species in fine particles (NH4 +). In the case of the gas ver tical profile, the model could be under estimating the gas scavenging rate s because their concentrations at ground level are lower than at higher altitudes. Gonalves et al (2003) used an approach very si milar to one used in this study, but they included both in-cloud and belowcloud scavenging rates and did not use the deposition-weighted average conc entrations to predict the NO3 and SO4 -2 rainwater concentrations in samples from the Amap State(Brazil). Using their model they

PAGE 151

130 calculated NO3 rainwater concentrations ranging fr om 15.5% to 58.2 % of the observed values, showing the same under-prediction tenden cy seen in this study especially for the case 2. The same under predictions were determined by a below-cloud scavenging model more similar to the one in this study for rainwate r samples collected at the Cubato region (Brazil) (Goncalves et al. 2000). Scavenging Coefficients From the results of the below-cl oud scavenging model, in Table 6-4 are presented the gas scavenging coefficien ts (Equation 4-74), mean mass particle coefficients (Equation 4-75), and rainfall-nor malized particle scavenging coefficients (Equation 4-76) for use in community-scale and mesoscale air quality models. For comparison, the USEPA-recommended mesos cale model CALPUFF has at its default rainfall scavenging coefficients for HNO3 and NO3 -, respectively, 6 10-5 and 1 10-4 s-1 (Scire et al ., 2000); a below-cloud scavenging model for soluble gases by Asman (1995) yielded a scavenging coefficient for NH3 of 3 10-4 s-1 for a 7 mm hr-1 precipitation rate; and Chate (2005) provided scavenging co efficients for submicron particles of 1.1 10-5 to 7.6 10-4 s-1, two orders of magnitude higher than our scavenging coefficient for NH4 +.

PAGE 152

131 Table 6-5. Scavenging Coefficients fr om the Below-Cloud Scavenging Model. Species Mean Scavenging Coefficient (s -1) Normalized Mean Scavenging Coefficient (s mm/hr) -1 NO3 1.44 10-4 1.96 10-4 2.64 10-5 0.67 10-5 NH4 + 3.16 10-7 3.90 10-7 6.27 10-8 2.43 10-8 Na+ 1.79 10-4 2.44 10-4 3.28 10-5 0.82 10-5 HNO3 1.56 10-4 1.85 10-4 2.78 10-5 1.81 10-5 NH3 2.17 10-4 2.57 10-4 4.60 10-5 1.91 10-5 Summary The below-cloud scavenging model presented allowed explanation of the measured concentration of NH4 +, NO3 and Na+ in rainwater samples collected at the Gandy bridge site, FL. Gas scavenging of NH3 contributed the most to rain deposited NH4 +, while scavenging of coarse particle NO3 and Na+ accounted for a subs tantial fraction of the rainfall NO3 and all of the Na+. The three DSDs used in this study did not give significantly different scavenging rates for the range of precipitation rates measured during the rain events studied. For rainfall NH4 + and NO3 -, scavenging calculations assuming constant gas and particle concentra tions gave predictions closer to observed rainwater concentrations, but for Na+, modeled results were closer to measured concentrations for the assumption of expone ntial decay. These results revealed the inherent weakness in a model that does not include in-cloud pro cessing and advection

PAGE 153

132 terms, for example, and the need for improved highly resolved time and space pollutant and meteorological measurements to understand better the species-dependent scavenging processes.

PAGE 154

133 7. ANALYSIS OF PARTICLE FORMATION PROCESSES THROUGH THE EFFECT OF METEOROLOGICAL CONDITIONS ON ORGANIC AND INORGANIC NITROGEN CONCENTRATIONS IN ATMOSPHERIC AEROSOLS Introduction The research goal in this study is to improve the knowledge about DON species by revealing how they integrate into atmospheric particles. Particle formation processes are driven by mass transfer and chemi cal reactions and therefore should show correlations with meteorological data. Meteorological parameters, such as relative humidity, temperature, wind speed and direction, drive the formation, interaction, size and deposition of particles as well as ga s absorption rates and salt formation into droplets on the atmosphere. Temperature, for example, is a driving force for chemical reactions, as well as for mass-transfer dur ing evaporation, abso rption and desorption processes. To evaluate this idea, 24-hr integr ated DIN and DON concentrations in PM10 samples and average meteorological conditions for two different periods (dry and wet) were analyzed using multi-linear regressi on techniques. Results were compared to identify seasonal effects and normal backgr ound conditions that defi ne the air quality and the nitrogen atmospheric deposition flux over Tampa Bay.

PAGE 155

134 Statistical Analyses For each sampling day, values of relativ e humidity, dry bulb temperature, and wind speed and wind direction were obtained as hourly surf ace weather observations from the Tampa International Airport (8 km north from the site, Figure 1) available at the National Climatic Data Center website ( www.ncdc.noaa.gov/servlets/ULCD ). Simple averaging of the hourly data pe r sampling day was used for relative humidity (RH, %), dry bulb temperature (DBT, oC), inverse of dry bulb temperature (INVDBT, oC-1), wind speed (WS, knots). For wind direction (WD, deg) and its st andard deviation (WDirSTD, deg), the east-west and the north south wind component s (knots) were calcu lated using vector computation with measurements of the horiz ontal component and th e azimuth angle of the wind vector using the equations presen ted by the (U.S. Environmental Protection Agency, 2000). The standard deviation of the wind direction was used as a measure of mixing in the atmosphere. It was determined using the Yamartino method with hourly wind direction values (U.S. Environmental Protection Agency, 2000). Equations 7, 8 and 9 define the average wind direction and its standard deviation. Values of the variable are chosen between (0, 1, 3/2) to place the resultant wind vector in the right quadrant indicated by the signs of its components. n is the number of hourly data points included to describe the variable variation of the sampling period.

PAGE 156

135 360 2 cos 360 2 sin *1 1 n i i i n i i iWD WS WD WS arctg WDAve Equation 7-1 2 360 ) 0.1547 + (1 ) ( sin arc WDirSTD3 Equation 7-2 2 1 2 1360 2 cos 360 2 sin 1 n WD n WDn i i n i i Equation 7-3 The persistence, or the ra tio between the module of th e average wind vector and the simple average wind speed per day, wa s calculated to evaluate the effect of sustained winds in a certain direction on daily measurements and its relation with possible fixed sources. However it was impo ssible to find significant correlations between this variable and the DIN and DON concentrations using Pearson Correlation Coefficients and Multi-linear Regression Analyses. DIN and DON concentrations were measur ed in two different seasons, the dry period with samples taken between Novemb er-2004 and April-2005 and the wet period between July and September-2005. The lat itude of Tampa Bay is near enough to the Equator not to show strong s easonal tendencies in temperature for example. There are significant variations in daily precipitati on rates during the hurricane season (wet period) and the rest of the y ear (dry period). During the samp ling period, there were just two rainstorms associated with Hurricane Ka trina but no aerosol samples were collected during those events.

PAGE 157

136 Average values for meteorological cond itions are reported on Table 7-1 and were individually compared by a t-test (t wo-tailed), after comparison of the data variances by an F-test. When the variance s were different from each other, a new number of degrees of freedom was calcula ted. The p-values indicated significant differences (at 95% confidence level) be tween the DBT, RH, WS and precipitation means for both periods. Table 7-1. Average Meteorological Conditi ons at the Gandy Bri dge Monitoring Site. Average s Dry Period: Nov, 2004 – Apr, 2005 (n=30) Wet Period: Jul-Sep, 2005 (n=25) p-value ( =0.05) Dry Bulb Temperature (oC) 19.9 3.1 28.2 0.9 0 Relative Humidity (%) 63.5 10.5 76.8 3.7 0 Wind Speed (knots) 7.1 2.1 4.9 1.9 0 Wind Direction (deg) 194.4 113.1 131.4 106.8 0.06 Standard Deviation of Wind Direction (deg) 47.9 24.8 59.3 25.7 0.13 Persistence (%) 70.5 25.3 57.9 28.4 0.06 East-West Wind Component (knots) 0.9 3.6 -0.1 3.4 0.24 North-South Wind Component (knots) -0.9 4.4 0.2 2.4 0.35 Precipitation (mm/day) 0.4 1.8 7 11 0.002

PAGE 158

137 The dry period was characterized by lo wer temperature, humidity and daily precipitation (few and short rain events) w ith higher wind speeds, especially when winds were blowing from the north and north-west (Figure 7-1). The wet period can be considered as isothermal due to the sm all standard deviations of the dry bulb temperature. Another important characteristic of this period is the constantly high water content of the air or high relative humidity values. 90 0 180 150 120 210 240 270 300 330 60 30 10 8 6 4 2 2 4 6 8 10 WS(dry) vs WD(dry) WS(wet) vs WD(wet) 90 0 180 150 120 210 240 270 300 330 60 30 10 8 6 4 2 2 4 6 8 10 90 0 180 150 120 210 240 270 300 330 60 30 10 8 6 4 2 2 4 6 8 10 90 0 180 150 120 210 240 270 300 330 60 30 10 8 6 4 2 2 4 6 8 10 WS(dry) vs WD(dry) WS(wet) vs WD(wet) Figure 7-1. Variation of the Average Wind Speed and Direction for the Dry and Wet Period on the Gandy Monitoring Site, Tampa, FL. (Dots identify the wind speed (knots) in the radial axis and the wind direction in the angular axis)

PAGE 159

138 Table 7-2. Average DIN and DON Particle C ontents in Samples from the Dry and Wet Period. Average s (%) Dry Period: Nov, 2004 – Apr, 2005 (n=30) Wet Period: Jul-Sep, 2005 (n=25) PM2.5 : NH4 +/DIN 86.2 9.4 92.6 8.7 PM2.5: DON/TDN 10.3 3.2 8.2 5.6 PM10-2.5 : NO3 -/DIN 93.3 5.0 87.5 7.5 PM10-2.5 : DON/TDN 8.3 13.3 17.3 10.9 PM10 : DON/TDN 9.5 4.1 9.9 6.0 PM10 : DON2.5 /DON1079.7 19.6 68.6 17.6 Table 7-2 shows the contribution of all ni trogen species to the general chemical composition of fine and coarse particles. DIN in fine particles is mainly ammonium, while in coarse particles is mainly nitrat e. DON represents approximately 10 % of the TDN in both fractions, but 70-80 % of the total organic nitrogen in PM10 is in the fine particles. Table 7-3 shows the average values of DIN concentrations as well as DON concentrations. Using the same comparison procedure applied to the meteorological values, it was determined that both periods have just significant differences (at 95% confidence level) in the NO3 content of fine and coarse particles.

PAGE 160

139 Table 7-3. Average DIN and DON Concentratio ns at the Gandy Bridge Monitoring Site. Average s (nmol m-3) Dry Period: Nov, 2004 – Apr, 2005 (n=30) Wet Period: Jul-Sep, 2005 (n=25) p-value ( =0.05) PM2.5 DINNH4 + 49.1 27.0 70.5 57.3 0.17 PM2.5 DINNO3 6.3 3.7 2.4 1.6 0 PM2.5 DIN 55.4 26.1 72.9 56.2 0.28 PM2.5 DON 6.1 2.6 4.9 2.7 0.05 PM10-2.5 DINNH4 + 1.6 1.2 1.5 1.0 0.48 PM10-2.5 DIN NO3 22.3 10.8 11.4 6.0 0 PM10-2.5 DIN 23.9 11.5 13.0 6.2 0 PM10-2.5 DON 1.7 1.7 2.4 1.6 0.08 A possible explanation for these differenc es can be found in the rain scavenging coefficients. NO3 particle contents were lower du ring the wet period because they are efficiently removed by rain droplets and th en do not remain suspended in air during sampling periods with rain events. NO3 --enriched particles have high scavenging coefficients caused by their average si ze (average particle diameter = 4 m (Poor et al. 2006; Campbell, 2005). NO3 --enriched particles are formed after th e nitric acid removal via reaction with sea salt (NaCl). Even if nitric acid concentrations are high, NO3 -particles are removed

PAGE 161

140 by rain as soon as they are formed. This makes impossible to maintain the same DINNO3 particle concentrations seen in the atmosphere on the dry period. A very different tende ncy is observed for NH4 + particle contents. They do not show significant difference even when temper ature and relative humidity are different for each period. Because it was already shown that DON and NH4 + are strongly correlated and mainly present in fine partic les, it was expected to have non-significant differences in DON particles contents from both periods. Before the application of multi-linear regression techniques to the data, we looked for possible deviations from the norma l distribution using th e Shapiro-Wilk test provided by the UNIVARIATE co mmand on SAS (SAS Institut e Inc, 2004). Data for DBT, RH and WDirSTD from both periods and WS of the wet period showed non significant deviations from th e normal distributions. Deviatio ns were seen just for WD and WS of the dry period, NH4 + particle concentrations fo r both periods, and nitrate and DON concentrations on the wet and dry period s, respectively. A natural logarithmic transformation was applied to all concentrati on values to correct such deviations. The log transformation was also intended to correct wedge patterns in residual vs. fitted plots seen in the first regressions caused by heteroscedasticity or unequal variance of the data. Improvements in the normality of data and residuals from regressions, as well on th e correlation coefficient r, were found for all vari ables after the log transformation in almost all cases. The log-transformation does not alter any distribution moments.

PAGE 162

141 Similar transformations have been found in the literature and were applied to reach the same goal: improve f it and achieve constant variance in the models (Robarge et al. 2002). Because NH4 + and NO3 are contained mainly in pa rticles with very different particle sizes respectively, it was assumed they are formed by different processes. Thus, their concentrations in the fine and coarse mode were added and studied them as single data distributions. To evaluate individual effects of meteorologi cal variables on DIN and DON concentrations, the Pearson correlation coefficients were calculated for all possible binary combinations between variable s. They are significantly different from zero (at 90% confidence level) when p-valu es are lower than or equal to 0.1. During the dry period, DIN-NH4 + concentrations were corr elated with almost all meteorological conditions except with wind direction (Figure 7-2). The strongest positive correlation is seen with the relative humidity. At high water contents the gaseous precursors of ammonium sulfate, ammonia and sulfuric acid, have more absorptive media and should be effectivel y removed from the atmosphere. At higher relative humidity values, particles covered by water films and/or water droplets could more effectively scavenge inorganic nitrogen gases, such as amm onia and nitric acid. The higher the NH4 + and SO4 -2 contents in liquid phase, the higher the rate formation of their salt. As is also expected, NH4 + particle contents are pos itively correlated with temperature. The higher the temperature, th e higher are the gas absorption rates, the reagents concentrations, and th e rate of chemical reactions leading to the formation of ammonium salts.

PAGE 163

142 Figure 7-2. Variation of the NH4 + Concentrations in PM10 With Respect to Meteorological Conditions.

PAGE 164

143 There is a negative correlation with wind speed, which means that the higher the wind speed the lower the DIN-NH4 + concentration. High wind speeds are related to higher mass fluxes of air. If the whole mass of air blowing over the site was homogeneous in composition, then it w ould be expected to have higher DIN concentrations at higher wind speeds. But for cleaner air masses blowing at the higher wind speeds, it is expected to have lower c oncentration of inorga nic nitrogen gases and therefore lower absorption to particles due to the dilution effect. In conclusion, this suggests that ammonium is diluted not de livered for stronger winds. The opposite was expected for NO3 as higher winds deliver more sea salt and with this more reactive surface area. Another possible explanation coul d be that higher wind speeds transport ammonia emissions from local sources adj acent to the monitoring site, reducing their possible transformation into particles. In the wet period, DIN-NH4 + concentrations showed ne gative correlations with the speed and the direction of the wind. Both variables are positively correlated (r=0.42, p=0.04, Figure 7-1). Stronger winds were m easured when the winds came from northwest and south-west directions. The sa me possible explanations given for the correlations in the dry period are applicable. DIN-NO3 concentrations do not show the same behavior seen for ammonium in the dry period (Figure 7-3). NO3 particle concentrations do not have significant correlation with relative humidity. Inst ead they showed a very strong positive correlation with temperature (DBT). NO3 in particles is likely to be present as sodium nitrate, and the equilibrium constant for its chemical formation decreases approximately

PAGE 165

144 9 % with an increment of 5 oC from 20 oC. A decrease of 39 % in the nitric acid dissociation constant is al so associated with a temperature increase from 20 oC to 25 oC (Seinfeld and Pandis, 1998). Those tendenc ies intimately related with the particle formation of sodium nitrate do not match with what was seen during the dry period. The temperature effect must be related then w ith the increase in the rate of chemical reactions leading to the nitric acid production from NOx emissions present in the atmosphere.

PAGE 166

145 Figure 7-3. Variation of the NO3 Concentrations in PM10 With Respect to Meteorological Conditions. In addition, DIN-NO3 concentrations show a moderate correlation with the standard deviation of the wind direction. This variable is a measure of mixing in the atmosphere, and it showed to be negatively co rrelated in the dry period with wind speed (r = 0.51 p=0.003). The higher the standard de viation of the wind direction, the lower

PAGE 167

146 is the wind speed. This could indicate more st able conditions in the atmosphere, better mixing and more contact time between pollutant s. Also, variations in the wind direction could benefit the sea spray formation and w ith this increase, NaCl concentrations available to form sodium nitrate. Alt hough atmospheric stability does not depend exclusive on wind conditions, it has b een associated with them (Robarge et al. 2002). In the wet period DIN-NO3 concentrations correlated significantly with only one meteorological variable, the standard devi ation of the wind direction. In contrast with the dry period, the corre lation coefficient is negative indicating decreases in the NO3 concentrations when the standard deviati on of the wind directi on was high. In this case, the better mixing effect could cause th e increase in the cont act time between rain droplets and NO3 --enriched particles and therefore the increase in the particle scavenging rates. At high values of the sta ndard deviation of the wind direction, if the particle formation rate does not increase fast enough, the scavenging rate could lead to a significant reduction in NO3 particle contents, as it was seen in this case for the wet period. Because 80 % of the DON is in fine particles it was decided to study the correlations of this fraction with meteorol ogical conditions (Fi gure 7-4). As for DIN concentrations, DON contents increase when relative humidity and temperature increase; and they decrease when wind speed increases. This beha vior could indicate that DON concentrations in particles depend also on the quantity of water available to absorb organic nitrogen gases or to serve as a media for aqueous pha se reactions leading to DON formation, as was seen for ammonium sulfate and sodium nitrate formation.

PAGE 168

147 Evidence suggesting that DON in particles could come from such gas-to-particle conversion processes was already found from samples collected at the same site. DON is concentrated on fine particles and their DON-fraction is 14% dimethylamine (DMA). DMA is a very volatile substa nce (Vapor pressure at 20 oC= 1.2 atm) and its presence in particles could indicate two things, first ve ry high gaseous DMA concentrations and second high retention of the DMA contents of air by the acidic pa rticles. DMA when diluted in water shows a basic character and is efficiently retained after protonation when the solution is acidic (Caldern et al. 2006b). However, DON showed a stronger correlation with the DIN concentrations in fine particles (~90 % ammonium). This could suggest that the same processes leading to ammonium sulfate production drive the DON formation or transference to the particles. This tendency was also observed for the wet period, where DON and DIN contents in fine particles were correlated with a moderate coefficient. The variation of temp erature and relative humidity during the wet period is almost negligible when it is comp ared to conditions on the dry period. This explains why no significant co rrelations were seen between these variables and all DIN and DON concentrations. Whatever is th e influence of those variables on the concentrations; it remained constant dur ing the whole sampling period and does not help to explain the variance in the concentrations.

PAGE 169

Figure 7-4. Variation of the DON Concentrations in PM2.5 With Respect to Mete orological Conditions. 148

PAGE 170

149 As was seen in the first period, DON in fine particles correlates positively with their DIN content; and due to the DIN be havior, increases in wind speed and wind direction should produce a decr ease in DON concentrations. This is confirmed by the negative correlations coefficients report ed for those variables in Figure 7-4. Figure 7-5. Variation of the DON Concentrations in PM10 With Respect to Meteorological Conditions.

PAGE 171

150 DON concentrations in PM10 showed only a moderate and negative correlation with the standard deviation of the wind di rection (Figure 7-5), while they kept the positive correlation with the DIN content of fine particles. This was expected due to the fact that 80 % of the DON is found in fine part icles and is strongly co rrelated with their DIN content. Multi-Linear Regressions All regressions were done using the stepwise method w ith significance levels for variable addition and retention in the model of 0.5 and 0.1 respectively. The method adds progressively to the model those i ndependent variables that make significant contributions to the description of the de pendent variable variance according to the confidence level for addition. Once a new variab le is added, the signi ficance test is done for the remaining of the vari ables and those which do not of fer significan t contribution according to the retention confidence leve l are discarded until all the non-included variables are tested at the addition level or have been already discarded (SAS Institute Inc, 2004). Outliers were identified and extracted from the data set when they had higher values than critical values fo r at least two of these variab les: studentized residuals, hratio, DFFITS and DFBETAS statistics, and second, if their presence decreased the multivariable correlation. Probability-probability plots were used to check normality of the model residuals. Collinearity diagnostics were also performed on each data set to avoid

PAGE 172

151 unstable estimates and high standard errors due to linear dependen ce between predictor variables. All conditions indices for all variables were under 25. Collinearities are expected to exist when condition indices are high er than 200 (SAS Institute Inc, 2004). The final chosen set of independent vari ables could explain a large fraction of the variance for the dependent variable. A ll regression models showed strong multiple correlation coefficients (r >0.6) for the op timal weighted linear combination of the predictor variables includ ed in this study. Before presenting the regression results, it is necessary to say that even when statistical analyses could reveal very impor tant variable dependencies, the information obtained from them is not conclusive. They do not substitute experimental or modeled data. Tendencies in variable dependences we re used here as indicators for possible processes. The goodness of the fitting for DIN-NH4 + concentrations in the dry and wet periods can be seen on Figure 7-6. Equa tions 7-4 and 7-5 showed that higher temperatures and lower wind speeds (more stab le atmosphere) lead to higher particle contents of DINNH4 +. The variation of thos e variables helped to explain an important fraction of the variance of the NH4 + concentrations in both periods. Additional influences of relative humidity and wi nd direction were caught by the model. The relationship between temperat ure and the log-transformed NH3 concentrations was also observed by Robarge et al 2002. In this study, wind speed also showed a negative correlati on with log-transformed concentrations and helped to explain 7% of their variance. Authors cl aimed that low wind speeds can help to

PAGE 173

152 accumulate trace gases close to boundary layer increasing their concentrations in the vicinity of sources. On the other hand, wind direction helped to explain ~15 % of the variance and together, temperature, wind di rection and wind speed explained 76% of the variance in the log-transformed ammonium concentrations (Robarge et al. 2002). Because ammonia is one of the main precursors of NH4 + in particles, these conclusions could be extended to the current conclusions. 28 0.72 R Adj 76 0 88 12 019 0 07 0 72 3 log4 n R DBT RH WS NH DIN Equation 7-4 20 86 0 89 0 4 275 003 0 2 0 01 15 log4 n R Adj R DBT WD WS NH DIN Equation 7-5 Figure 7-6. Goodness of the Fit ting for the Log-Transformed NH4 + Concentration With Meteorological Conditions.

PAGE 174

153 In the case of NO3 particle concentrations, ev en though Figure 7-7 shows two distinctive groups of points with lower c oncentrations during the wet period, both regressions showed a negative effect of the wind direction. Winds blowing from north and south west were associated with decrease s in the nitrate particle contents. As was expected according to the bina ry correlation coefficient, temperature showed a positive effect on DIN-NO3 concentrations. This effect coul d not be caught for the wet period because the DBT variance was very small. 27 0.88 R Adj 89 0 5 41 002 0 002 0 7 5 log3 n R DBT WDirSTD WD NO DIN Equation 7-6 19 0.51 R dj 62 0 005 0 0006 0 06 0 21 3 log3 n A R WDirST D WD WS NO DIN Equation 7-7 Figure 7-7. Goodness of the Fit ting for the Log-Transformed NO3 Concentration With Mete orological Conditions.

PAGE 175

154 So far, the effect of the meteorological conditions on the main two inorganic nitrogen species is as expected. However, a more interesting be havior is seen for organic nitrogen in fine particles. The optimal multiple regressions are shown in Equations 7-8 and 7-9, and the goodness of the fitting in Figure 7-8. The variance in DON concentrations in fine particles for both periods can be well explained using only their DIN contents. As was suggested before, two situations are likely to occur. First, it is possible that DIN and DON are fed into particles by the same processes, or second, they come from similar sources; therefore the higher the DIN the higher the DON. The DIN behavior in fine particles led to assumption of absorption of gases as the main process involved in particle formation; then DON could come from absorption of organic nitrogen gases in the atmosphere. This is the most likely process that can explain the DMA levels seen on fine particles. On the ot her hand, it is possible that inorganic nitrogen species and water present in particles could develop a film with a high chemical affinity for organic nitroge n species. With a high fine particle concentration, more surf ace area for the DON-absorption became available and DON concentration increases. 27 0.88 R Adj 89 0 log 8 0 38 1 log5 2 5 2 n R DIN PM DON PM Equation 7-8 21 0.79 R Adj 80 0 log 5 0 6 0 log5 2 5 2 n R DIN PM DON PM Equation 7-9

PAGE 176

155 Figure 7-8. Goodness of the Fitting for th e Log-Transformed DON Concentration in PM2.5 With Meteorological Conditions. For DON in coarse particles the optimal linear combinations of the predictor variables could not catch as mu ch data variance as the ot her models (Equations7-10, 711). It is important to say that after includi ng DON concentration in coarse particles, the good correlation with meteorologi cal conditions for the fine particles was lost. This could indicate that DON sources for coarse pa rticles are different from the ones for fine particles. External effects, not related w ith the chosen meteorological set, such as vegetation abundance or polle n concentrations could be responsible for this. 28 0.54 R Adj 56 0 log 3 0 8 0 log4 10 10 n R NH DIN PM DON PM Equation 7-10

PAGE 177

156 20 0.60 R Adj 0.65 log 8 0 002 0 02 4 log3 10 10 n R NO DIN PM WD DON PM Equation 7-11 The variables included in the regressi on of DON concentrations from the wet period do not match those from the dry period, and therefore any a dditional information can be found from compari ng seasonal effects. Summary Results from multi-linear regression of DON and DIN concentrations with meteorological data suggested as expected th at gas-to-particle conve rsions are the main source of nitrogen in particles collected over Ta mpa Bay. This agrees with results from previous studies, and highlights the need for gas measurements that can offer more information about a possible and very importa nt role of the gas phase on dry and wet deposition fluxes of organic nitrogen. As gases, they can be very well distributed in the atmosphere, can be removed by rainwater scav enging and/or direct mass transfer from the air to the water surface, offering a more global and continuous effect. If they had longer life times they could represen t a very important source of NOx after degradation by sunlight. In regard to DON, regressi on models showed a secondary source or formation process for DON in coarse particles, possibly related to external factors such as vegetation debris or suspended pollen.

PAGE 178

157 8. CONCLUSIONS AND FURTHER RESEARCH The UV-photolysis method proved to be e fficient to measure DON concentrations in atmospheric samples with equiva lent concentrations lower than 33 M-N if the operational parameters are set to be the op timal irradiation period of 24-hr and the optimal solution pH of 2. DON concentrations could be overestimated by 15 % if the DIN/DON ratio is higher than 4. The succe ssful application of the method to the analysis of field samples proved that unde r its optimal settings, the UV-Photolysis method can be adapted as the standard method to measure DON in atmospheric samples. The factorial design scheme, used for the UV-Photolysis optimization, helped to reduce the number of experiments with the surrogate organic nitrogen compounds. This approach is innovative in the atmospheric chemistry field where numerous experiments are required to test the performan ce of the analytical procedures. DON species proved to be present in aer osol and rainwater and represents approximately 10 % of the total dissolved nitrogen measured in samples collected at Tampa Bay. In Tampa Bay, concentrations of DON in fine and coarse particles are 5.3 2.6 and 2.1 1.6 nmol m-3 (n=55) respectively, and DON concentrations in rainwater samples are 3.3 3.1 M-N (n=13). Estimated wet deposition fluxes for NH4 +, NO3 and DON were 1.40, 3.18 and 0.34 kg-N ha-1 yr-1, respectively, for individual

PAGE 179

158 contributions of 28 %, 65 % and 7 % to the total nitrogen loading from wet deposition, which was equal to 4.91 kg-N ha-1 yr-1. This indicated that atmospheric nitrogen loading previously estimated for surface waters are biased low due to the exclusion of the DON contributions to wet a nd dry deposition fluxes. The existence of DON in atmospheric samp les also indicated missing sources in the inventories of nitrogen deposited on Tampa Bay, e.g. DMA showed to be present in PM2.5 samples and represented an average co ntribution of 12.8 6.7 % to the total DON concentration measured in the same pa rticle fraction. DMA could come from oceanic emissions or evaporation for wast ewater effluents charged to the bay. The importance of this result is seen when it is considered that DON in fine particles represented the 75 19 % of the total DON found in PM10 samples. Furthermore the strong correlations seen for the DON and DIN c oncentrations in fine particles and also with the meteorological conditi ons suggested that gas-to-par ticle conversions are likely to be responsible for the T DN concentration seen in atmo spheric aerosols collected at Tampa Bay. These findings about DON atmospheric concen trations are of substantial value to many researchers focused on the study of its role in ecosystems growth limitations, especially in areas with increasing anthropogenic emi ssions. Due to the possible existence of natural background concentrations of species such as DMA, conclusions from this research can be extended to similar geographic areas and can be used to identify and estimate possible sources when other information is not available.

PAGE 180

159 The gas-phase role in the atmospheric deposition of nitrogen became more important after the application of a belo w-cloud scavenging model. According to the calculations, gas scavenging of DON is res ponsible for 99% of the DON wet deposition flux. Particle scavenging of DON in PM10 is responsible only fo r 0.9 % 0.2% of the DON concentrations in rainwater reason. DM A particle scavenging represented 11.7 7.1 % of this value. Below-cloud scavengi ng calculations also confirmed that DIN concentrations in gas and particles can expl ain their concentrations in rainwater. The wet deposition of NH4 + is mainly caused by gas scavenging of ammonia, while the NO3 wet deposition flux is mainly conseq uence of particle scavenging. The below-cloud model proposed in this re search was able to predict rainwater sample concentrations closer to the experime ntal values than other models reported in the literature for the same cases. Its distinct ive characteristic is the integration of the typical gas and particle collection models to the concept of deposition-weighted average concentrations. The model implementation offers a new solution to a problem of increasing interest to the atmospheric ch emistry community, the simulation of wet deposition fluxes. This research offers information about the composition of the organic compound fraction in the atmospheric samples, a vari able of increasing interest and not well understood by scientists. DON species in partic ulate matter and rain droplets, as any other organic compound, can force change s in the climate by influencing cloud condensation nuclei processes; and, if toxi c, also can impact human health after inhalation into the respiratory tract.

PAGE 181

160 According to this, DON gaseous species co uld be responsible for a dry deposition nitrogen flux higher than the estimated valu es for Tampa Bay. These results highlight the need for a method to measure gasphase organic nitrogen gases. As a recommendation to future research it is ne cessary to test and implement a method to measure gaseous DON concentrations. During the course of this research, an annular denude r system with a methanesulfonic acid methanol-based co ating solution was tested. When the UVphotolysis method was applied to th e extracts from exposed denuders, DON concentrations became negative due to lower TDN levels compared to initial DIN measured concentrations. Even though amm onia concentrations were increased after irradiance, as occurred during the orga nic-breakdown of DON in PM samples or standard solutions; the nitrat e concentrations in the irra diated samples were under the method detection level. Field blanks for this method satisfied the requirements for DON concentrations. A second option to measure DO N in the gas phase was also tested with failed results. Glass bubble tubes or impingers gases fi lled with DDW or DDW and XAD-16 resin were used to promote the ab sorption of gases pumped from the atmosphere. The system configuration and th e field blanks failed and gave very high DON concentrations that made it impossible to determine the sample DON content. Future research should concentrate on the complete chemical composition, particle size distribution and gas phase concentrations of DON species. A column filled with C18 resin (such as the Ultra-Clean SPE column from Alltech Associates, Inc) could be used to adsorb DON gases pumped from the ai r. A single lab study conducted in this

PAGE 182

161 research showed that DMA vapors produced from bubbling air into a concentrated solution were adsorbed into the resin. I on chromatographic analyses confirmed the existence of DMA in extracts from the re sin elution with DDW. Due to the volatile nature of DMA and the possible presence of more short-chain aliphatic amines in the air, it is possible that the application of this method could reveal information about a large fraction of the DON gaseous species.

PAGE 183

162 REFERENCES Ambonguilat, S., Gallard, H., Garron, A., Epr on, F., and Philippe Croue, J. Evaluation of the catalytic reduction of nitrate for the determin ation of dissolved organic nitrogen in natural waters. Water Research In Press, Corrected Proof. Anastasio, C. and McGregor, K.G. (2000) Photodestruction of dissolved organic nitrogen species in fog waters. Aerosol Science and Technology 32, 106-119. Anastasio, C. and McGregor, K.G. (2001) Chemistry of fog waters in California's Central Valley: 1. In situ photoformation of hydr oxyl radical and singlet molecular oxygen. Atmospheric Environment 35, 1079-1089. Andronache, C. (2003) Estimated variability of below-cloud aerosol removal by rainfall for observed aerosol size distributions. Atmospheric Chemistry and Physics 3, 131-143. Angelino, S., Suess, D.T., and Prather, K.A. (2001) Formation of aerosol particles from reactions of secondary a nd tertiary alkylamines: Ch aracterization by aerosol time-of-flight mass spectrometry. Environmental Science and Technology 35, 3130-3138. APHA (1998) Standard Methods for the examina tion of Water and Wastewater 20th ed. Arakaki, T., Miyake, T., Hirakawa, T ., and Sakugawa, H. (1999) pH dependent photoformation of hydroxyl radical and absorbance of aqueous-phase N(III) (HNO2 and NO2 -). Environmental Science and Technology 33, 2561-2565.

PAGE 184

163 Armstrong, F.A.J., Williams, P.M., and Stri ckland, J.D.H. (1966) Photo-oxidation of organic matter in sea water by ultra-vi olet radiation, analytical and other applications. Nature 211, 481-484. Arnold, J. R., Hartsell, B. E., Luke, W.T., Ullah, SM. R., Dasgupta, P. K., Huey, L.G., Tate, P. (2006) Field testi ng of four methods for gasphase ambient nitric acid, Atmospheric Environment, In Review. Asman, W. A. H., 1995. Parameterization of below-cloud scavenging of highly soluble gases under convective conditions. Atmospheric Environment 29: 1359-1368. Beddows, D.C.S., Donovan, R.J., Harrison, R.M., Heal, M.R., Kinnersley, R.P., King, M.D., Nicholson, D.H., and Thompson, K.C. (2004) Correlations in the chemical composition of rural bac kground atmospheric aerosol in the UK determined in real time using time-of-flight mass spectrometry. Journal of Environmental Monitoring 6, 124-133. Berger, P., Leitner, N.K.V., Dore, M., and Legube, B. (1999) Ozone and hydroxyl radicals induced oxidation of glycine. Water Research 33, 433-441. Bird, R.B., Stewart, W.E., and Lightfoot, E.N. (2002) Transport Phenomena 2nd. ed. edn. John Wiley and Sons, Inc., USA. Bronk, D.A., Lomas, M.W., Glibert, P.M., Schukert, K.J., and Sanderson, M.P. (2000) Total dissolved nitrogen analysis: comp arisons between the persulfate, UV and high temperature oxidation methods. Marine Chemistry 69, 163-178. Bronk, D.A. and Ward, B.B. (2000) Magnitude of dissolved organic nitrogen release relative to gross nitrogen upt ake in marine systems. Limnology and Oceanography 45, 1879-1883. Buffam, I. and McGlathery, K.J. (2003) Eff ect of ultraviolet ligh t on dissolved nitrogen transformations in coastal lagoon water. Limnology and Oceanography 48, 723734.

PAGE 185

164 Caldern, S.M., Poor, N., and Campbell, S.W. (2006a) Analysis of particle formation processes through the effect of mete orological conditions on organic and inorganic nitrogen concentr ations in atmospheric aer osols. In 86th American Meteorological Society Annual Meeti ng. 8th conference on Atmospheric Chemistry.,Paper no. 105312. Ameri can Meteorological Society (AMS), Atlanta, GA. Caldern, S.M., Poor, N.D., and Campbell, S. W. (2006b) Estimation of the particle and gas scavenging contributions to we t deposition of organic nitrogen. Atmospheric Environment In review. Caldern, S.M., Poor, N.D., and Campbell, S.W. (2006c) Investigation of the UV photolysis method for the determinati on of dissolved organic nitrogen in environmental samples. Journal of the Air and Wast e Management Association In Press. Campbell, S. (2005) Fitting of the parameters of the lognormal distribution for particle size distribution. Tampa, FL. Campbell, S.W., Evans, M.C., and Poor, N. D. (2002) Predictions of size-resolved aerosol concentrations of ammonium, chlori de and nitrate at a bayside site using EQUISOLV II. Atmospheric Environment 36, 4299-4307. Carrillo, J.H., Hastings, M.G., Sigman, D.M ., and Huebert, B.J. (2002) Atmospheric deposition of inorganic and organic n itrogen and base cations in Hawaii. Global Biogeochemical Cycles 16, 24-(1-16). Casado, H., Encinas, D., and Lacaux, J.P. (1996) Relationship between the atmospheric particulate fraction and th e ionic content of prec ipitation in an area under influence of a waste incinerator loca ted in the Basque Country (Spain). Atmospheric Environment 30, 1537-1542. Chan, M.N., Choi, M.Y., Ng, N.L., and Chan C.K. (2005) Hygroscopicity of watersoluble organic compounds in atmosphe ric aerosols: Amino acids and biomass burning derived organic species. Environmental Science and Technology 39, 1555-1562.

PAGE 186

165 Chate, D.M. and Pranesha, T.S. (2004) Fiel d studies of scavengi ng of aerosols by rain events. Journal of Aerosol Science 35, 695-706. Chate, D.M., Rao, P.S.P., Naik, M.S., Momin, G.A., Safai, P.D., and Ali, K. (2003) Scavenging of aerosols and their chemical species by rain. Atmospheric Environment 37, 2477-2484. Chen, W.H. and Wangersky, P.J. (1993) A High-Temperature Catalytic-Oxidation Method for the Determination of Mari ne Dissolved Organic-Carbon and Its Comparison with the Uv Photooxidation Method. Marine Chemistry 42, 95106. Chirico, R.D., Knipmeyer, S.E., Nguyen, A., and Steele, W.V. (1999) Thermodynamic properties of the methylpyridines. Part 2. Vapor pressures, heat capacities, critical properties, derived thermodynami c functions between the temperatures 250 K and 560 K, and equilibrium isomer distributions for all temperatures >= 250 K. Journal of Chemical Thermodynamics 31, 339-378. Clegg, S.L., Brimblecombe, P., and Wexler A.S. (1998) Thermodynamic model of the system H+-NH4 +-SO4 -2-NO3 --H2O at tropospheric temperatures. Journal of Physical Chemistry A 102, 2137-2154. Cornell, S., Mace, K., Coeppicus, S., Duce, R., Huebert, B., Jickells, T., and Zhuang, L.Z. (2001) Organic nitrogen in Hawaiian rain and aerosol. Journal of Geophysical Research-Atmospheres 106, 7973-7983. Cornell, S.E. and Jickells, T.D. (1999) Wa ter-soluble organic nitr ogen in atmospheric aerosol: a comparison of UV and persulfate oxidation methods. Atmospheric Environment 33, 833-840. Cornell, S.E., Jickells, T.D., Cape, J.N., Ro wland, A.P., and Duce, R.A. (2003) Organic nitrogen deposition on land and coastal environments: a review of methods and data. Atmospheric Environment 37, 2173-2191. Cornell, S.E., Jickells, T.D., and Thor nton, C.A. (1998) Urea in rainwater and atmospheric aerosol. Atmospheric Environment 32, 1903-1910.

PAGE 187

166 Evans, M.C., Campbell, S.W., Bhethanabotla, V., and Poor, N.D. (2004) Effect of sea salt and calcium carbonate in teractions with nitric acid on the direct dry deposition of nitrogen to Tampa Bay, Florida. Atmospheric Environment 38, 4847-4858. Feingold, G. and Levin, Z. (1986) The Lognormal Fit to Ra indrop Spectra from Frontal Convective Clouds in Israel. Journal of Climate and Applied Meteorology 25, 1346-1363. Gibb, S.W., Mantoura, R.F.C., and Liss, P. S. (1999a) Ocean-atmosphere exchange and atmospheric speciation of ammonia and me thylamines in the region of the NW Arabian Sea. Global Biogeochemical Cycles 13, 161-177. Gibb, S.W., Mantoura, R.F.C., Liss, P.S., and Barlow, R.G. (1999b) Distributions and biogeochemistries of methylamines and ammonium in the Arabian Sea. DeepSea Research Part Ii-Topical Studies in Oceanography 46, 593-615. Gill, P.S., Graedel, T.E., and Weschler, C.J. (1983) Organic Films on Atmospheric Aerosol-Particles, Fog Droplets, Clo ud Droplets, Raindrops and Snowflakes. Reviews of Geophysics 21, 903-920. Gmehling, J., Onken, U., and Rarey-Nies, J.R. (1977) Vapor-Liquid Equilibrium Data Collection DECHEMA. Goeyens, L., Kindermans, N., Abu Yusuf, M., and Elskens, M. (1998) A Room Temperature Procedure for the Manual Determination of Urea in Seawater. Estuarine, Coastal and Shelf Science 47, 415-418. Golimowski, J. and Golimowska, K. (1996) UV-photooxidation as pretreatment step in inorganic analysis of environmental samples. Analytica Chimica Acta 325, 111133. Goncalves, F.L.T., Andrade, M.F., Forti, M. C., Astolfo, R., Ramos, M.A., Massambani, O., and Melfi, A.J. (2003) Preliminar y estimation of the rainfall chemical composition evaluated through the scav enging modeling for north-eastern Amazonian region (Amapa State, Brazil). Environmental Pollution 121, 63-73.

PAGE 188

167 Goncalves, F.L.T., Massambani, O., Beheng, K.D., Vautz, W., Schilling, M., Solci, M.C., Rocha, V., and Klockow, D. (2000) Modelling and measurements of below cloud scavenging processes in the highly industrialized region of Cubatao-Brazil. Atmospheric Environment 34, 4113-4120. Gorzelska, K., Galloway, J.N., Watterson, J., and Keene, W.C. (1992) Water-soluble primary amine compounds in rural continental precipitation. Atmospheric Environment 6, 1005-1018. Gronberg, L., Lovkvist, P., and Jonsson, J.A. (1992) Measurement of Aliphatic-Amines in Ambient Air and Rainwater. Chemosphere 24, 1533-1540. Gustafsson, L. (1984) Interferences in the de termination of total nitrogen in natural waters by photo-oxidation to nitrate-nitrite mixture. Talanta 31, 979-986. Hales, J.M. (1972) Fundamentals of the Theory of Gas Scavenging by Rain. Atmospheric Environment 6, 635-659. Hales, J.M. (2002) Wet removal of polluta nts from gaussian pl umes: Basic linear equations and computational approaches. Journal of Applied Meteorology 41, 906-918. Hartsell, B., Poor, N., Tate, P., Breckner, F. (2006) A comparison of continuous and integrated atmospheric ammonia measurements, in preparation. Hassan, S.M., Salem, F.B., and Elsalam, N. A. (1995) Colorimetric Determination of Urea in Potable Water Using Diacetyldioxime and Different Oxidizing-Agents. Journal of the Indian Chemical Society 72, 73-75. Hendrix-Holmes, H. (2002) Can bulk deposit ion measurements reasonably estimate ammonia nitrogen deposition rate s? Master Thesis, Univer sity of South Florida, Tampa, FL.

PAGE 189

168 Hinds, W.C. (1999) Aerosol Technology: Properties, Behavior and Measurement of Airborne Particles. 2nd. ed. edn. John Wiley and Sons, Inc., USA. Jung, C.H., Kim, Y.P., and Lee, K.W. ( 2003) A moment model for simulating raindrop scavenging of aerosols. Journal of Aerosol Science 34, 1217-1233. Karpel Vel Leitner, N., Berger, P., and Le gube, B. (2002) Oxidation of Amino Groups by Hydroxyl Radicals in Re lation to Oxidation Degr ee of the Alpha-Carbon. Environmental Science and Technology 36, 3083-3089. Kumar, S. (1985) An Eulerian Model fo r Scavenging of Pollutants by Raindrops. Atmospheric Environment 19, 769-778. Lee, W.T. and Westerhoff, P. (2005) Disso lved organic nitrogen measurement using dialysis pretreatment. Environmental Science and Technology 39, 879-884. Mace, K.A., Artaxo, P., and Duce, R.A. (2003a) Water-soluble organic nitrogen in Amazon Basin aerosols during the dry (biomass burning) and wet seasons. Journal of Geophysical Research-Atmospheres 108, Art. No. 4512. Mace, K.A. and Duce, R.A. (2002) On the use of UV photo-oxidation for the determination of total nitrogen in rainwater and water-extracted atmospheric aerosol. Atmospheric Environment 36, 5937-5946. Mace, K.A., Duce, R.A., and Tindale, N.W. (2003b) Organic nitrogen in rain and aerosol at Cape Grim, Tasmania, Australia. Journal of Geophysical ResearchAtmospheres 108, Art. No. 4338. Mace, K.A., Kubilay, N., and Duce, R.A. (2003c) Organic nitrogen in rain and aerosol in the eastern Mediterranean atmosphere: An association with atmospheric dust. Journal of Geophysical Research-Atmospheres 108, Art. No. 4320. Manny, B.A., Miller, M.C., and Wetzel, R.G. (1971) Ultraviolet Combustion of Dissolved Organic Nitrogen Compounds in Lake Waters. Limnology and Oceanography 16, 71-85.

PAGE 190

169 Maria, S.F. and Russell, L.M. (2005) Or ganic and inorganic aerosol below-cloud scavenging by suburban New Jersey preci pitation. Environmental Science and Technology 39, 4793-4800. McGregor, K.G. and Anastasi o, C. (2001) Chemistry of fog waters in California's Central Valley: 2. Photochemical tran sformations of amino acids and alkyl amines. Atmospheric Environment 35, 1091-1104. Milne, P.J. and Zika, R.G. (1993) AminoAcid Nitrogen in Atmospheric Aerosols Occurrence, Sources and P hotochemical Modification. Journal of Atmospheric Chemistry 16, 361-398. Minoura, H. and Iwasaka, Y. (1997) Ion c oncentration changes obs erved in drizzling rains. Atmospheric Research 45, 165-182. Mircea, M., Stefan, S., Facchini, M.C., and Fu zzi, S. (2004) Analytical formulas for the below-cloud scavenging coefficient of an irreversibly soluble gas: a quantitative evaluation for HNO3. International Journal of Environment and Pollution 21, 547-565. Mircea, M., Stefan, S., and Fuzzi, S. ( 2000) Precipitation scav enging coefficient: influence of measured aerosol and raindrop size distributions. Atmospheric Environment 34, 5169-5174. Mitchell, S.C. and Zhang, A.Q. ( 2001) Methylamine in human urine. Clinica Chimica Acta 312, 107-114. Mizak, C.A., Campbell, S.W., Luther, M.E., Carnahan, R.P., Murphy, R.J., and Poor, N.D. (2005) Below-cloud ammonia scave nging in convective thunderstorms at a coastal research site in Tampa, FL, USA. Atmospheric Environment 39, 15751584.

PAGE 191

170 Neff, J.C., Holland, E.A., Dentener, F.J., McDowell, W.H., and Russell, K.M. (2002) The origin, composition and rates of organic nitrogen deposition: A missing piece of the nitrogen cycle? Biogeochemistry 57, 99-136. Oberholzer, B., Volken, M., Collet Jr, J.L ., Staehelin, J., and Waldvogel, A. (1993) Pollutant concentrations and below-cloud s cavenging of selected N (-III) species along a mountain slope. Water, Air and Soil Pollution 68, 59-73. Okita, T., Hara, H., and Fukuzaki, N. (1996) Measurements of atmospheric SO2 and SO42-, and determination of the wet scav enging coefficient of sulfate aerosols for the winter monsoon season over the sea of Japan. Atmospheric Environment 30, 3733-3739. Paerl, H.W. (1988) Nuisance P hytoplankton Blooms in Coastal, Estuarine, and Inland Waters. Limnology and Oceanography 33, 823-847. Pandey, J.S., Khan, S., Joseph, V., and Kumar, R. (2002) Aerosol scavenging: Model application and sensitivity anal ysis in the Indian context. Environmental Monitoring and Assessment 74, 105-116. Perry, R. (1999) Perry's chemical engineers' handbook 7th ed. edn. McGraw-Hill, New York. Poor, N., Pollman, C., Tate, P., Begum, M., Evans, M.S.C., and Campbell, S.W. (Accepted) Nature and magnitude of at mospheric fluxes of total inorganic nitrogen and other inorga nic species to the Tampa Bay watershed, FL, USA. Water, Air and Soil Pollution Poor, N., Pribble, R., and Greening, H. (2001) Direct wet and dry deposition of ammonia, nitric acid, ammonium and n itrate to the Tampa Bay Estuary, FL, USA. Atmospheric Environment 35, 3947-3955. Poor, N., Tamanini, T., Tate, K., Tran, T ., Stevens, R., and Atkeson, T. (2000) Comparison of collocated Dichotomous Air Samplers a PM2.5 FRM Air Sampler. In Presentations from PM2.5M onitoring, Quality Assurance, and Data Analysis Workshop

PAGE 192

171 Office of Air Quality, Pla nning and Standards. Envir onmental Protection Agency. United States. May 22-25, 2000. Robarge, W.P., Walker, J.T., McCulloch, R.B., and Murray, G. (2002) Atmospheric concentrations of ammonia and ammoni um at an agricultural site in the southeast United States. Atmospheric Environment 36, 1661-1674. Roig, B., Gonzalez, C., and Thomas, O. (1999) Measurement of dissolved total nitrogen in wastewater by UV photooxidatio n with peroxodisulphate. Analytica Chimica Acta 389, 267-274. Rupprecht and Patashnick Co., I. ( 1999). Dichotomous Partisol-Plus Model 2025 Sequential Air Sampler. Oper ating Manual.Revision A. Rupprecht and Patashnick Co., Inc., Albany, USA. Samara, C. and Tsitouridou, R. (2000) Fine and coarse ionic aerosol components in relation to wet and dry deposition. Water Air and Soil Pollution 120, 71-88. Sander, R. (1999) Compilation of Henry's Law Constants for I norganic and Organic Species of Potential Importance in E nvironmental Chemistry., pp. 107, Mainz, Germany. SAS Institute Inc (2004) SAS/STAT 9.1: User's guide SAS institute Inc., Cary, NC. Scire, J. S., Stimaitis, D. G., Yamartino, R. J., 2000. A User's Guide for the CALPUFF Dispersion Model, Version 5, Earth Tech, Inc., Concord, MA, pp. 521. Schade, G.W. and Crutzen, P.J. (1995) Em ission of Aliphatic-Amines from Animal Husbandry and Their Reactions Po tential Source of N2o and Hcn. Journal of Atmospheric Chemistry 22, 319-346. Scheller, E. (2001) Amino acids in de w origin and seasonal variation. Atmospheric Environment 35, 2179-2192.

PAGE 193

172 Scudlark, J.R., Russell, K.M., Galloway, J. N., Church, T.M., and Keene, W.C. (1998) Organic nitrogen in precipi tation at the mid-Atlan tic US coast Methods evaluation and preliminary measurements. Atmospheric Environment 32, 17191728. Seinfeld, J.H. and Pandis, S.N. (1998) Atmospheric Chemistry and Physics: From Air Pollution to Climate Change. John Wiley and Sons, Inc., USA. Sharp, J.H., Rinker, K.R., Savidge, K.B., Abell, J., Benaim, J.Y., Bronk, D., Burdige, D.J., Cauwet, G., Chen, W.H., Doval, M.D., Hansell, D., Hopkinson, C., Kattner, G., Kaumeyer, N., McGlathery, K.J., Merriam, J., Morley, N., Nagel, K., Ogawa, H., Pollard, C., Pujo-Pay, M., Raimbault, P., Sambrotto, R., Seitzinger, S., Spyres, G., Tirendi, F., Walsh, T.W., and Wong, C.S. (2002) A preliminary methods comparison for measur ement of dissolved organic nitrogen in seawater. Marine Chemistry 78, 171-184. Slinn, W.G.N. (1977). Precipitation Scav enging: Some problems, approximate solutions, and suggestions for future research. In Precipitation Scavenging (eds R.G. Semonin and R.W. Beadle), pp. 843. Technical Information Center Energy Research and Development Admini stration., Champaing, IL, USA. Solorzano, L. and Sharp, J.H. (1980) Dete rmination of Total Di ssolved Nitrogen in Natural-Waters. Limnology and Oceanography 25, 751-754. Sommerville, K. and Preston, T. (2001) Ch aracterisation of dissolved combined amino acids in marine waters. Rapid Communications in Mass Spectrometry 15, 12871290. Spitzy, A. (1990). Amino acids in marine aerosol and rain. In Facets of Modern Biogeochemistry pp. 433. Springer-Verlag Berlin Heidelberg New York, Berlin, Germany. Stoddard, J.L. (1994) Long-term changes in watershed re tention of nitrogen: its causes and aquatic consequences American Chemical Society, Washington DC. Tanner, P.A., Lei, H., Huang, M. and Sh en Z. (1997) Acid Rain and Below-Cloud Scavenging in South-Western China. Journal of Atmospheric Chemistry 27, 7178.

PAGE 194

173 Tarr, M.A., Wang, W.W., Bianchi, T.S., a nd Engelhaupt, E. (2001) Mechanisms of ammonia and amino acid photoproduction from aquatic humic and colloidal matter. Water Research 35, 3688-3696. Tarr, M.A., Wang, W.W., Engelhaupt, E., and Bianchi, T.S. (1999) Ammonium and amino acid photoproduction from aquatic humic and colloidal matter. Abstracts of Papers of the American Chemical Society 218, U650-U650. TBNEP (1996) Charting the Course: The Comp rehensive Conservation and Management Plan for Tampa Bay. Tampa Bay National Estuary Program, Saint Petersburg, FL. Teerlink, T., Hennekes, M.W.T., Mulder, C ., and Brulez, H.F.H. (1997) Determination of dimethylamine in biological samples by highperformance liquid chromatography. Journal of Chromatography B 691, 269-276. Timperley, M.H., Vigorbrown, R.J., Kawash ima, M., and Ishigami, M. (1985) Organic Nitrogen-Compounds in Atmospheric Pr ecipitation Their Chemistry and Availability to Phytoplankton. Canadian Journal of Fisheries and Aquatic Sciences 42, 1171-1177. Tomar, M. (1999) Quality Assessment of water and wastewater Lewis Publishers, USA. Tomasko, D.A., Corbett, C.A., Greening, H. S., and Raulerson, G.E. (2005) Spatial and temporal variation in seagrass coverage in Southwest Florida: assessing the relative effects of an thropogenic nutrient load reduc tions and rainfall in four contiguous estuaries. Marine Pollution Bulletin 50, 797-805. U.S. Environmental Protection Agency (2000) Meteorological Monitoring Guidance for Regulatory Modeling Applications North Carolina, USA. Vahatalo, A.V. and Zepp, R.G. (2005) P hotochemical mineralization of dissolved organic nitrogen to ammonium in the Baltic Sea. Environmental Science and Technology 39, 6985-6992.

PAGE 195

174 Van Neste, A., Duce, R.A., and Lee, C. (1985) Methylamines in the Marine Atmosphere. Geophysical Research Letters 14, 711-714. Volken, M. and Schumann, T. (1993) A Critical-Review of Below-Cloud Aerosol Scavenging Results on Mt Rigi. Water Air and Soil Pollution 68, 15-28. Walsh, T.W. (1989) Total Dissolved Nitroge n in Seawater a New High-Temperature Combustion Method and a Compar ison with Photo-Oxidation. Marine Chemistry 26, 295-311. Wang, G.S., Hsieh, S.T., and Hong, C.S. ( 2000a) Destruction of humic acid in water by UV light Catalyzed oxidati on with hydrogen peroxide. Water Research 34, 3882-3887. Wang, P.F., Martin, J., and Morrison, G. (1999) Water quality and eutrophication in Tampa Bay, Florida. Estuarine Coastal and Shelf Science 49, 1-20. Wang, W.W., Tarr, M.A., Bianchi, T.S ., and Engelhaupt, E. (2000b) Ammonium photoproduction from aquatic humic and colloidal matter. Aquatic Geochemistry 6, 275-292. Weisstein, E.W. (1999) Buckingham's Pi Theorem. In MathWorld--A Wolfram Web Resource., Vol. 2005. Wolfram Research, Inc. Wurzler, S. (1998) The scavenging of n itrogen compounds by clouds and precipitation: : Part II. The effects of cloud micr ophysical parameterization on model predictions of nitric acid scavenging by clouds. Atmospheric Research 47-48, 219-233. Yang, X.H., Scranton, M.I., and Lee, C. ( 1994) Seasonal-Variations in Concentration and Microbial Uptake of Methylamines in Estuarine Waters. Marine EcologyProgress Series 108, 303-312.

PAGE 196

175 Zhang, Q. and Anastasio, C. (2001) Chemis try of fog waters in California's Central Valley Part 3: concentrations and speci ation of organic and inorganic nitrogen. Atmospheric Environment 35, 5629-5643. Zhang, Q. and Anastasio, C. (2003a) Conve rsion of fogwater and aerosol organic nitrogen to ammonium, nitrate, and NO x during exposure to simulated sunlight and ozone. Environmental Science and Technology 37, 3522-3530. Zhang, Q. and Anastasio, C. (2003b) Free and combined amino compounds in atmospheric fine particles (PM2.5) and fog waters from Northern California. Atmospheric Environment 37, 2247-2258. Zhang, Q., Anastasio, C., and Jimemez-Cruz M. (2002) Water-solub le organic nitrogen in atmospheric fine particles (P M2.5) from northern California. Journal of Geophysical Research-Atmospheres 107. Zhao, H., Zheng, C. (2006) Monte Carlo solution of wet re moval of aerosols by precipitation. Atmospheric Environment 40, 1510-1525.

PAGE 197

176 APPENDICES

PAGE 198

177 Appendix A. Anova Results From 22 Factorial Design Table A-1. Conversion of 80 M-N Amino Acids to Inorganic Nitrogen After UV Photolysis. Solution pH Exposure Time, hr (%) Average s NH4 + (%) NO2 (%) NO3 (%) 2 8 87.8 88.3 0.6 100.0 0.0 0.0 2 8 88.7 100.0 0.0 0.0 2 8 72.9* 100.0 0.0 0.0 2 24 96.5 95.3 1.7 99.1 0.0 0.9 2 24 93.3 99.6 0.0 0.4 2 24 96.0 99.1 0.0 0.9 5 8 70.9 69.1 2.9 99.8 0.0 0.2 5 8 65.7 100.0 0.0 0.0 5 8 70.6 99.9 0.0 0.1 5 24 83.9 85.7 2.5 99.1 0.0 0.9 5 24 88.6 99.3 0.0 0.7 5 24 84.8 98.8 0.9 0.3 *Outlier

PAGE 199

178 Appendix A: (continued) Table A-2. Conversion of 80 M-N Urea to Inorganic Nitrog en After UV Photolysis. Solution pH Exposure Time, hr (%) Average s NH4 + (%) NO2 (%) NO3 (%) 2 16 72.1 71.5 0.9 57.0 1.8 41.2 2 16 70.8 61.3 1.6 37.1 2 24 72.3 70.4 2.6 62.4 1.1 36.5 2 24 68.5 64.0 1.1 34.5 6 16 8.2 7.9 1.1 83.6 13.6 2.9 6 16 7.2 86.5 10.2 3.2 6 24 9.7 10.3 0.8 89.5 10.5 0.0 6 24 10.9 86.3 9.1 4.6

PAGE 200

179 Appendix A: (continued) Table A-3. Conversion of 80 M-N Methylamine to Inor ganic Nitrogen After UV Photolysis. Solution pH Exposure Time, hr (%) Average s NH4 + (%) NO2 (%) NO3 (%) 2 8 45.3 46.8 1.7 95.6 0.0 4.4 2 8 46.5 95.7 0.0 4.3 2 8 48.6 95.7 0.0 4.3 2 24 85.5 84.4 2.0 95.1 0.0 4.9 2 24 85.5 95.4 0.0 4.6 2 24 82.1 95.4 0.0 4.6 9 8 6.0 5.4 0.5 67.8 4.2 28.0 9 8 5.2 74.5 3.5 22.1 9 8 5.0 62.0 7.4 30.6 9 24 10.6 11.1 1.4 79.6 1.9 18.5 9 24 12.7 83.0 1.9 15.1 9 24 10.0 74.5 2.5 23.0

PAGE 201

180 Appendix B. Configuration of th e R&P Dichotomous Air Sampler The Rupprecht and Patashnick (R&P) dichotomous Partisol-Plus model 2025 sequential air sampler collects the particulate matter in two modes: fine (PM2.5) and coarse (PMcoarse). The dichotomous sampler operates at ~16.7 l min-1 in a split flow configuration that allows a 1.7 l min-1 for coarse particles collec tion and a flow of 15 l min-1 for fine particles collection, for a total averag e volume of sampling air of 21 and 2.3 m-3. The dichotomous air sampler has a cyclone inle t to separate the biggest particles of the coarse fraction, those with particle size larger than 10 m. The pre-cleaned stream passes through a virtual impact or that divides the particle s according to their size and concentrates them into the PM2.5 and PM10 fractions; while a pair of flow controllers keeps constant the two ai r flow rates at 15 l min-1 and 1.67 l min-1, respectively. The virtual impactor uses the inertia of th e particles to separate those larger than the cutoff diameter onto the coarse collecti on filter. The high flow makes the smaller particles move radially inside the impactor until they are collected in the filter for the fine fraction. The cutoff diameter of the impactor is 2.5 m, therefore all particles with a size equal to 2.5 m will be separated with 50% efficiency. PMcoarse fraction can contain between 5-10% of the particles smalle r than the cutoff size in addition to those larger than the cutoff size. (Hinds, 1999). (Figure B-1)

PAGE 202

181 Appendix B: (continued) Figure B-1. Schematic Flow Diagram for the Dichotomous Air Sampler. Some advantages in the use of dic hotomous air samplers are: a 16-filter sequential collection feature that allows for unattended operation for up to two weeks between site visits, with automatic exchange s of filters according to a user-defined sampling schedule, dual-flow configur ation to collect simultaneously PM2.5 and PM10 or PM2.5 in filters made of different material a nd a second flow contro ller for each stream to maintain a second flow at the constant vo lumetric flow rate required for sampling.

PAGE 203

182 Appendix B: (continued) The selection of this equipment was guide d by the fact that the dichotomous air sampler design satisfies the requirements for measuring PM2.5 and PM10 of important environmental agencies such as USEPA (United States Environmental Protection Agency) (Rupprecht and Patashnick Co ., 1999). A recent study done by the Hillsborough County Environmental Protec tion Commission in Tampa, Florida, affirmed that they can meet a 10% relative bias and 10% relative precision target for ambient air particulate measurements, and re veals the potential fo r the dichotomous air sampler to combine FRM PM2.5 and PM10 measurements in one sampler (Poor et al. 2000). The main characteristics of the dichotom ous air sampler can be seen in figure B2. Figure B-2 Dichotomous Ai r Sampler Characteristics

PAGE 204

183 Appendix C. Study of Error Prop agation for the DON Concentration The DON concentration in a sample or DONmeasured is the difference between the inorganic nitrogen concentration before UV-photolysis and after UV-photolysis. The conversion efficiency ( ) was defined as the molar ratio of the final net ammonium, nitrite and nitrate nitr ogen concentrations ( DONmeasured) to the initial organic nitrogen concentration of the standards ( DONstandard), as shown in Equations 3-1 and C-1. 3 2 43 3 2 2 4 41000 1000 1000NO beforeUV L mg afterUV L mg NO beforeUV L mg afterUV L mg NH beforeUV L mg afterUV L mg measuredMW NO NO MW NO NO MW NH NH N M DON Equation C-1 100 *standard measureDON DONd Equation 3-1 The propagation of the systematic errors as sociated with these variables is given by equations C-2 and C-3. 2 2 23 3 2 2 4 41000 1000 1000 NO L mg NO NO L mg NO NH L mg NH DONMW MW MW N Mmeasured Equation C-2

PAGE 205

184 Appendix C: (continued) 2 tan 2tan% % dard s DON measured DONDON DONdard s measured Equation C-3 All molecular weights and conversion f actors are considered constant and therefore do not represent an error source. The DON concentration in the standard solutions is mainly given by systematic errors associated with the dilution process. Volumetric instruments, such as flask or pipette, were used in the solution preparat ion. The solute and the final solution mass were determined dir ectly by weighting. The standard deviation associated with the weight determination is the scale precision. The standard deviati on associated with the ion concentrations is determined using the detection limit (DL). The DL wa s defined according to the method detection level (MDL) definition. The method uses a solu tion with the matrix of interest prepared to be in the range of one to five times the calculated MDL. The solution concentration is measured seven times randomly over a peri od of at least 3 da ys, and the MDL is defined as the standard deviation of the re plicates times the t-value from a one-side tdistribution with six degrees of freedom at a 99% confidence leve l (American Public Health Association (APHA), 1998. Standard Method for the examination of Water and Wastewater, 20th ed.). The correspondent values are presented in Table C-1.

PAGE 206

185 Appendix C: (continued) Table C-4. Standard Deviation Associat ed With Variables Involved in the DON Determination. Variable Standard Deviation associated with the variable Weight (g) 0.0001 g Ammonium concentration0.007 mg L-1 Nitrite concentration 0.009 mg L-1 Nitrate concentration 0.007 mg L-1 Using the DL for all ions, the standard deviation associated with the DON measured is equal to: N MmeasuredDON 45 0 62 007 0 1000 46 009 0 1000 18 007 0 10002 2 2 Equation C-4 For the determination of the error propag ation in the solution preparation it is necessary to start from the preparation of the mother solution. The molarity of a solution is defined as:

PAGE 207

186 Appendix C: (continued) solution solute solute L molV MW W M Equation C-5 If the solution volume is determined i ndirectly through the solution weight and all quantities are measured in grams th en, the molarity can be written as: mol N mol solution solute solution solute L N molx W MW W M 1000 Equation C-6 The solution density can be appr oximated by the water density at 20oC because it is a diluted system. The molecular weight and the water density are considered as constants, and therefore the sy stematic error associated w ith the solution molarity is given by: 2 2 solution W solute W MW W Msolution solute Equation C-7 gsolution soluteW W0001 0 For a solution of urea 0.1024 g were we ighted and water was added until the final solution weight was 50.0788 g. The solution molarity is: L N mol mol g L mL mL gmol N mol g g M 06797 0 2 0788 50 056 60 1000 9982 0 1024 0 The standard deviation of this variable is: N N mol N N mol Mg g g g 00007 0 0788 50 0001 0 1024 0 0001 0 06797 02 2

PAGE 208

187 Appendix C: (continued) A series of dilutions is d one to adjust the solution c oncentration to the required levels. In a dilution, the concentration of a solution 2 changes according to the added mass of the solution 1 as given: 2 11 2 S SW W M M The standard deviation of the new concentration is then: 2 2 2 1 2 22 2 1 1 1 S W S W MW W M M MS S Using the mother solution to prepare a ~1000 m-N solution, the concentration and its error range is: L N mol L N molg g M 001030 0 5888 51 7820 0 06797 02 L N mol L N mol L N mol L N molE g g g g M 06 1 5888 51 0001 0 7820 0 0001 0 06797 0 00007 0 001030 02 2 2 2 L N mol L N molE g g M 03 0997 0 2197 50 8622 4 001030 03 L N mol L N mol L N mol L N molE g g g g E E M 07 1 2197 50 0001 0 8622 4 0001 0 001030 0 06 1 03 0997 02 2 2 3 L N mol L N mol ME M 1 0 7 99 07 1 99733 L N mol L N molE g g E M 06 98 9 0170 25 5030 2 03 0997 04 Appendix C: (continued)

PAGE 209

188 L N mol L N mol L N mol L N molE g g g g E E E M 08 1 0170 25 0001 0 5030 2 0001 0 03 0997 0 07 1 06 98 92 2 2 4 L N mol L N mol ME M 01 0 98 9 08 1 99844 For a 10 m-N urea solution the conversion efficiency has an associated error of: % 50 4 98 9 01 0 00 10 45 0 % 02 1002 2 % 5 4 % 0 100 For a 100 m-N urea solution the conversion effi ciency has an associated error of: % 45 0 7 99 1 0 31 93 45 0 % 36 932 2 % 5 0 % 4 93

PAGE 210

ABOUT THE AUTHOR Silvia Caldern earned her bachelor’s degree in Chemical Engineering in 1998 graduating cum laude at the University of Los Andes (ULA), Venezuela. In 1999 earned a position as a faculty member in the Chemical Engineering School at ULA. In 2002 she graduated from a Master’s degree in Applied Engineering Mathematics from ULA. In the same year, ULA awarded her w ith a scholarship to study in the doctoral program on Chemical Engineering at the University of South Florida (USF). During her doctoral studies at USF she received numerous awards and scholarships by the Air and Waste Manage ment Association (AWMA) including the Milton Feldstein award as recognition of th e work in air quality issues. The Chemical Engineering Department at USF gave her an honorary award as Ou tstanding Research Assistant on 2005. Since 2003 she has been a member of the Bay Regional Atmospheric Chemistry Experiment (BRACE) and her research was funded by the Florida Department of Environmental Protection.


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 2200397Ka 4500
controlfield tag 001 001795154
005 20080428130218.0
007 cr mnu uuuuu
008 070628s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001538
035
(OCoLC)146061424
040
FHM
c FHM
049
FHMM
090
TP145 (ONLINE)
1 100
Caldern, Silvia Margarita.
0 245
Estimation of the particle and gas scavenging contributions to wet deposition of organic and inorganic nitrogen
h [electronic resource] /
by Silvia Margarita Caldern.
260
[Tampa, Fla] :
b University of South Florida,
2006.
520
ABSTRACT: Atmospheric deposition of nitrogen species represents an additional nutrient source to natural environments, and can alter the nitrogen cycle by increasing nutrient levels beyond the requirements of organisms. In Tampa Bay, atmospheric deposition of dissolved inorganic nitrogen species (DIN) has been found to be the second largest nitrogen source, but little is known about dissolved organic nitrogen species (DON). The research goal was to improve the dry and wet deposition estimates by inclusion of the DON contribution. In the atmospheric chemistry field a standard method to measure DON in atmospheric samples has not been agreed upon. This research proposes the use of the ultraviolet (UV)-photolysis method and presents the optimal settings for its application on atmospheric samples. Using a factorial design scheme, experiments on surrogate nitrogen compounds, typically found in the atmosphere, indicated that DON can be measured with no biases if optimal settings are fixed to be solution pH 2 with a 24 hr irradiance period.DIN species (NH, NO, NO) and DON concentrations were determined in fine (PM.) and coarse particles (PM.) as well as in rainwater samples collected at Tampa Bay. The estimates of wet despotition fluxes for NH, NO and DON were 1.40, 3.18 and 0.34 kg-N hayr, respectively. Hourly measured gas concentrations and 24-hr integrated PM10 concentrations were used in conjunction with a below-cloud scavenging model to explain DIN and DON concentration in rainwater samples. Scavenging of aerosol phase DON contributed only 0.90.2 percent to rainwater DON concentrations, and therefore gas scavenging should be responsible for 99 percent. These results confirmed the existence of negative biases in the dry and wet deposition fluxes over Tampa Bay. There is increasing interest in simulating wet deposition fluxes, and the proposed below-cloud scavenging model offers a new computational approach to the problem.It integrates the typical gas and particle collection functions and the concept of the deposition-weighted average concentrations. The model uses mass balance to describe the time-dependent cumulative contribution of all droplets in the rain spectrum to the rainwater concentration, giving predictions closer to experimental values and better estimations than those reported in the literature for similar cases.
502
Dissertation (Ph.D.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 188 pages.
Includes vita.
590
Adviser: Scott W. Campbell, Ph.D .
653
DON.
DIN
Rain scavenging.
UV photolysis.
Atmospheric chemistry.
690
Dissertations, Academic
z USF
x Chemical Engineering
Doctoral.
773
t USF Electronic Theses and Dissertations.
949
FTS
SFERS
ETD
TP145 (ONLINE)
4 856
u http://digital.lib.usf.edu/?e14.1538