... CO 80235-3098 and. HDR Engineering, Inc.,(4) SAUR,(5) and SAUR Services(6) ...... co-precipitate NOM, the precipitate was redissolved in acetic acid, co-precipitated sulfate ...... Christman, R.F., D.L.Norwood, Y.Seo, F.H.Frimmel. 1989.
ISOLATION, FRACTIONATION AND CHARACTERIZATION OF NATURAL ORGANIC MATTER IN DRINKING WATER
ii
ISOLATION, FRACTIONATION AND CHARACTERIZATION OF NATURAL ORGANIC MATTER IN DRINKING WATER
Prepared by: Jean-Phillipe Croue(1), Gregory V. Korshin (2), Jerry Leenheer(3) and Mark Benjamin(2) (1)
Laboratoire de Chimie de l’Eau et de l’Environnement, UPRESA CNRS 6008 Ecole Supérieure d’Ingénieurs de Poitiers, Université de Poitiers 86022 Poitiers Cedex, France (2)
Department of Civil Engineering, Box 352700, University of Washington Seattle, WA 98195-2700 (3)
U.S. Geological Survey Denver, Colo.
Jointly Sponsored by: AWWA Research Foundation 6666 Quincy Avenue Denver, CO 80235-3098 and HDR Engineering, Inc.,(4) SAUR,(5) and SAUR Services(6) Seattle Wash.(4); Maurepas, France(5); West Sussex, England(6) Published by: AWWA Research Foundation and American Water Works Association
DISCLAIMER
This study is funded by the AWWA research Foundation (AWWARF). AWWARF assumes no responsibility for the content of the research study reported in this publication or for the opinions or statements of fact expressed in the report. The mention of trade names for commercial products does not represent or imply the approval or endorsement by AWWARF. This report is presented solely for informational purposes.
iv
Copyright ©1998 by AWWA Research Foundation and American Water Works Association Printed in the U.S.A.
ISBN0-00000-000-0
v
CONTENTS TABLES .......................................................................................................................... xiii FIGURES....................................................................................................................... xviii Foreward ........................................................................................................................ xxvi acknowledgments.......................................................................................................... xxvii EXECUTIVE SUMMARY ......................................................................................... xxviii ABBREVIATIONS ..................................................................................................... xxxix CHAPTER 1 Introduction..................................................................................................1 NOM Fractions and Water Quality Issues ...............................................................5 CHAPTER 2 Literature Review ........................................................................................4 Overview..................................................................................................................4 Preliminary characterization ....................................................................................5 Techniques for Concentration, Isolation and Fractionation of NOM......................7 Techniques Intended Primarily for Concentration of NOM ........................7 Evaporation ......................................................................................7 Freeze concentration and freeze-drying...........................................8 Membrane technologies ...................................................................8 Sorption Techniques for Concentrating, Isolating, and Fractionating NOM11 Quantitative Analysis of the Use of Adsorptive Columns for NOM Isolation and Fractionation ................................................12 vi
Sorption onto Al and Fe oxides and ion exchange resins ..............16 Sorption onto hydrophobic sorbents (XAD and similar resins) ....18 Desalting ....................................................................................................21 Ion exchange ..................................................................................22 Precipitation and Co-precipitation .................................................22 Desalting with XAD resins ............................................................23 Losses of NOM During Processing ...........................................................25 NOM Characterization...........................................................................................26 13C and H-NMR spectroscopy..................................................................27 FTIR spectroscopy .....................................................................................30 Pyrolysis-GC-MS.......................................................................................32 Elemental Analysis ....................................................................................35 UV / Visible and Fluorescence Spectrometry............................................37 Nature of Light Absorption by NOMError! Bookmark not defined. The Three-Band Theory of UV Spectra of NOMError! Bookmark not defined. Nature of Light Emission by NOMError! Bookmark not defined. Compound Class Identification .................................................................41 CHAPTER 3 Materials and Methods...............................................................................46 Sample Collection..................................................................................................46 NOM isolation protocols .......................................................................................49 vii
Membrane-Based NOM Isolation Protocols..............................................50 European Water Samples...............................................................50 Pacific Northwest Water Samples .................................................53 NOM Isolation Using Adsorption and Elution Processes .........................55 European Water Samples...............................................................55 Pacific Northwest Water Samples .................................................56 NOM desalting...........................................................................................59 Experimental Procedures for Characterizing NOM reactivity ..............................60 Chlorination Studies ..................................................................................61 Coagulation - Flocculation Study ..............................................................61 Analytical Methods................................................................................................62 Inorganic Species .......................................................................................62 Inorganics in Source Waters ..........................................................62 Elemental Analysis of NOM Isolates ............................................64 Total Organic Halides (TOX) ........................................................65 Trihalomethanes.............................................................................66 Haloacetic Acids ............................................................................66 Total Dissolved Amino Acids....................................................................67 Total Dissolved Carbohydrates..................................................................69 Pyrolysis-GC-MS.......................................................................................70 viii
NMR and FTIR Spectra .............................................................................71 UV Absorbance and Fluorescence Emission Spectra................................72 CHAPTER 4 NOM Concentration, Isolation and Fractionation .....................................74 Overview................................................................................................................74 Case Studies with the Goal of Maximal Recovery and Fractionation...................76 Suwannee River .........................................................................................76 South Platte River ......................................................................................82 Comparison of NOM Recovery from the Suwannee and S. Platte Rivers90 Case Studies Focusing on Maximal NOM Recovery without Fractionation.........92 Membrane Processes for NOM Recovery from European Waters............92 Salt Concentration and Recovery ..............................................................94 Use of XAD-8 /XAD-4 for NOM Recovery from European Waters ........98 Combination of RO with XAD Resin for NOM Recovery from European Waters ..........................................................................................100 NOM Concentration and Isolation of Judy Reservoir and Tolt River Water By Reverse Osmosis ....................................................................103 Comparison of NOM Concentration and Isolation Using IOCS, IOCO, and XAD Resins for Judy Reservoir and Tolt River Water ...............105 Competition Between Sulfate and NOM for Oxide Adsorbents .............112 Recovery of Various NOM Fractions with IOCO ...................................113
ix
Summary and Comparison of NOM Concentration-Isolation-Fractionation Techniques ...................................................................................114 CHAPTER 5 NOM Characterization: Suwannee and South Platte Rivers ...................122 Introduction..........................................................................................................122 Bulk Fractionation and Elemental analysis .........................................................123 FTIR Characterization .........................................................................................129 13
1
C-NMR Characterization ..................................................................................134
H-NMR Characterization ...................................................................................138
Dissolved Amino Acids and Carbohydrates ........................................................138 Pyrolysis - gas chromatography - mass spectrometry .........................................145 UV Spectra...........................................................................................................150 Relationship between various structural and spectral properties of NOM fractions160 Comparison of NOM Characteristics between the Suwannee and South Platte Rivers .......................................................................................................169 Reactivity of NOM ..............................................................................................171 Coagulation-Flocculation.........................................................................171 Removal of DOC and A254 .........................................................171 Chlorination of NOM: Reactions of Different Fractions and Changes in NOM Characteristics ....................................176 DBP formation potentials ....................................................................176 Relationships between DBPFP and Spectral Properties......................181 x
Summary and Comparison of Suwannee River and South Platte River NOM Characterizations .....................................................................................186 CHAPTER 6 Characteristics of NOM Captured by Different Techniques ...................191 General Characterization of NOM in the Blavet River .......................................191 Elemental Analysis and SUVA of the NOM isolates from the Blavet River ......195 Total dissolved Amino acids and Carbohydrates ................................................198 13
C-NMR and FTIR Data for the Blavet River NOM Isolates.............................201
Pyrolysis - Gas Chromatography - Mass Spectrometry for the Blavet River NOM Isolates .....................................................................................................204 Chlorination of the Blavet River NOM Isolates ..................................................207 UV and fluorescence spectra of the Blavet NOM concentrates ..........................210 Correlations between spectral and structural characteristics of NOM in Blavet River fractions..........................................................................................213 Conclusions..........................................................................................................222 CHAPTER 7 Correlations between Data From Structure-Sensitive Methods and UV and Fluorescence Spectroscopy..............................................................................................225 Introduction..........................................................................................................225 The Major Parameters of UV and Fluorescence Spectra of NOM ......................227 Correlations between 13C-NMR spectroscopy and the major parameters of UV and fluorescence emission spectra ...........................................................227 Correlations between Pyr-GC-MS spectroscopy and the major parameters of UV and fluorescence emission spectra...........................................................232 xi
Conclusions..........................................................................................................234 CHAPTER 8 Summary, Conclusions and Recommendations.......................................236 Comparison of NOM concentration methods ......................................................236 Comparison of Chemical Composition of NOM Isolates....................................240 Desalting of NOM Isolates ..................................................................................242 Variability of NOM .............................................................................................244 Seasonal Changes of NOM..................................................................................248 Effects of NOM Isolation and Concentration on DBP Precursors ......................248 Comparison of Ex Situ and In Situ Methods for NOM Characterization............249 13
C CPMAS NMR....................................................................................249
Fourier Transform IR Spectroscopy (FTIR) ............................................250 Pyrolysis GC-MS .....................................................................................251 Total Dissolved Amino Acids and Carbohydrates ..................................252 Elemental Analysis ..................................................................................253 UV Spectroscopy .....................................................................................253 Fluorescence ............................................................................................255 Relationship Between Data From in Situ and ex Situ Analyses..............256 Appendix..........................................................................................................................257 references .........................................................................................................................258
xii
TABLES Table 1.1 Major chemical classes of compounds included into NOM and associated water quality problems ..........................................................................................................7 Table 1.2 Association between specific types of disinfection by-products and major chemical classes of NOM ............................................................................................1 Table 2.1 Some literature values for DOC rejection and recovery efficiencies by RO.......9 Table 2.2 Structural assignments for 13C-NMR spectra ....................................................28 Table 2.3 Structural assignments for 1H-NMR spectra .....................................................29 Table 2.4 Infrared frequency bands for biomolecular structures in NOM isolates ...........31 Table 2.5 Characteristic infrared spectral peaks of inorganic solutes (in KBr pellets) .....32 Table 2.6 Origin of biopolymers and their respective specific pyrolysis fragments .........33 Table 2.7 Elemental analysis of hydrophobic acids and transphilic acids isolated from surface waters ............................................................................................................36 Table 2.8 Average elemental analysis of humic substances (with or without fractionation to humic and fulvic acids) and transphilic acids isolated from surface waters..........37 Table 2.9 Concentrations of amino acids and total dissolved sugars in various surface waters .........................................................................................................................43 Table 2.10 Amino acid and sugar concentrations in various NOM fractions....................44 Table 3.1 Water quality characteristics of untreated water samples..................................49 Table 3.2 Characteristics of the membranes used to process European samples ..............51
xiii
Table 3.3 Characteristics of the composite iron oxide-based media used for NOM adsorption...................................................................................................................57 Table 3.4 Operation parameters for concentration of NOM by adsorption for Pacific Northwest waters .......................................................................................................58 Table 3.5 Acceptable concentration ranges for elemental analysis ...................................64 Table 3.6 TDAA content of the IHSS standard Suwannee River fulvic acid....................68 Table 3.7 TDCA content of the IHSS standard Suwannee River fulvic acids ..................70 Table 4.1 Fraction name conventions and their correlation with the previously used terminology ................................................................................................................75 Table 4.2 Yields and recovery of dissolved natural organic matter (NOM) fractions from the Suwannee River ...................................................................................................81 Table 4.4 Yields and recovery of NOM fractions from the second South Platte River sample ........................................................................................................................89 Table 4.5 Volume and DOC content of the solutions collected from reverse osmosis and nanofiltration processing of five surface waters ........................................................92 Table 4.6 Isolation of NOM: Efficiency of RO and NF membranes.................................93 Table 4.7 Concentration of anions in permeate and concentrate produced by reverse osmosis and nanofiltration .........................................................................................95 Table 4.8 Anion recoveries for reverse osmosis and nanofiltration ..................................96 Table 4.9 Recovery of anions in the concentrated water produced by RO treatment of a synthetic solution .......................................................................................................98 Table 4.10 DOC content of the XAD-8 and XAD-4 permeates of the three water sources99 Table 4.11 XAD-8/XAD-4 DOC distribution and DOC recoveries..................................99 xiv
Table 4.12 Purification of RO-concentrated water: RO and XAD resin desalting in series102 Table 4.13 Results for the fractionation of IOCO effluent for treatment of Judy Reservoir water.........................................................................................................................114 Table 5.1 Elemental analysis of the Suwannee River NOM isolates ..............................126 Table 5.2 Elemental analysis of the South Platte River NOM isolates ...........................126 Table 5.3 C/O, C/H and C/N ratios of the Suwannee River NOM isolates.....................128 Table 5.4 C/O, C/H and C/N ratios of the South Platte River NOM isolates..................128 Table 5.5 Integrated areas of 13C-NMR Spectra of Suwannee River NOM fractions.....135 Table 5.6 Integrated areas of 13C-NMR spectra of South Platte River NOM fractions from first ...........................................................................................................................136 Table 5.7 Integrated areas of
13
C-NMR Spectra of South Platte River NOM Fractions
from second sampling ..............................................................................................136 Table 5.8 TDCA and TDAA in the Suwannee River NOM isolatesError! Bookmark not defined.
Table 5.9 TDCA and TDAA in the South Platte River and its isolates fractionError! Bookmark not defined Table 5.10 Major compounds identified by Pyr-GC-MS analysis of NOM fractions.....146 Table 5.11 Relative proportions of biopolymers in South Platte River NOM isolates ...149 Table 5.12 Relative proportions of biopolymers in Suwannee River NOM isolates ......149 Table 5.13 UV and fluorescence parameters for Suwannee River NOM fractions.........152 Table 5.14 UV and fluorescence parameters for South Platte River NOM fractions......158
Table 5.15 Aromatic carbon content and SUVA of the Suwannee River NOM isolatesError! Bookmark not
xv
Table 5.16 Aromatic carbon content and SUVA254 of the South Platte River NOM isolates .......................................................................Error! Bookmark not defined. Table 5.17 DOC and A254 removals of Suwannee River NOM isolates during coagulation-flocculation with aluminum at pH 6.5. ................................................172 Table 5.18 DOC and A254 removals of South Platte River NOM isolates during coagulation-flocculation with aluminum at pH 6.5. ................................................172 Table 5.19 Chlorine demand and disinfection by-products formation potentials of the Suwannee River NOM isolates................................................................................177 Table 5.20 Chlorine demand and DBP formation potentials of the South Platte River and its isolated NOM fractions.......................................................................................178 Table 5.21 DBP formation potentials of the Suwannee River NOM isolates .................179 Table 5.22 DBP formation potentials of the South Platte River and its NOM isolated fractions ...................................................................................................................180 Table 6.1 TDCA, TDAA, and BDOC in the Blavet River ..............................................194 Table 6.2 Apparent molecular weight distribution of the NOM of the Blavet River ......195 Table 6.3 Elemental analysis of the Blavet River NOM isolates ....................................196 Table 6.4 C/O, C/N and C/H ratios with SUVA of the Blavet River NOM....................198 Table 6.5 TDAA content of the Blavet River NOM isolates (winter sample) ................199 Table 6.6 Integrated areas of 13C-NMR spectra of the Blavet River NOM isolates .......203 Table 6.7 Relative proportions of biopolymers in the Blavet River NOM isolates ........204 Table 6.8 Some chlorinated DBPs formed by chlorination of the Blavet River and its NOM isolates ...........................................................................................................208
xvi
Table 6.9 UV and fluorescence parameters for NOM samples concentrated from the Blavet River .............................................................................................................211 Table 7.1 Summary of results from 13C-NMR, UV absorbance and fluorescence emission analysis for 27 samples of concentrated or fractionated NOM from the Blavet, South Platte, Suwannee and Tolt Rivers. ...........................................................................228 Table 7.3 Ranges of major parameters of Pyr-GC-MS, UV absorbance and fluorescence emission analyses of 13 samples from three sources (Blavet, South Platte and Suwannee Rivers) ....................................................................................................232 Table 7.4 External and internal correlations of the UV and fluorescence spectral parameters with Pyr-GC-MS results for 13 samples from the Blavet, South Platte and Suwannee Rivers...............................................................................................233 Table 8.1 Comparison of NOM concentration methods..................................................237 Table 8.2 Comparison of distribution of biopolymers in RO and XAD-8/XAD-4 samples, based on UV and Pyr-GC-MS .................................................................................241
xvii
FIGURES Figure 2.1 Conceptual representation of electronic transitions caused by the absorbance of light for benzene and NOM ...................................................................................39 Figure 2.2 Summation of three composite absorption bands and formation of unconvoluted UV absorbance spectrum of NOM......................................................40 Figure 3.1 Schematic for NOM concentration using reverse osmosis and adsorption......54 Figure 4.1 Flow chart for fractionation and isolation of hydrophobic and transphilic NOM fractions from the Suwannee River .................................................................77 Figure 4.2 Flow chart for fractionation and isolation of hydrophilic NOM (k' = 5-100) from the Suwannee River...........................................................................................79 Figure 4.3 Flow chart for fractionation and isolation of ultra-hydrophilic NOM fractions (k' < 5) from the Suwannee River..............................................................................80 Figure 4.4 Flow chart for fractionation and isolation of ultra-hydrophilic NOM fractions from the South Platte River .......................................................................................85 Figure 4.7 Breakthrough curves for treatment of concentrated NOM solutions with XAD-4 resin.............................................................................................................101 Figure 4.8 DOC breakthrough curves for treatment of water from Judy Reservoir using IOCO, IOCS and XAD-8.........................................................................................105 Figure 4.9 A254 breakthrough curves for treatment of water from Judy Reservoir using IOCO, IOCS and XAD-8.........................................................................................106 Figure 4.10 Cumulative DOC retention efficiency for treatment of Judy Reservoir water using IOCO, IOCS and XAD-8 ...............................................................................107
xviii
Figure 4.11 Cumulative A254 retention efficiency for treatment of Judy Reservoir water using IOCO, IOCS and XAD-8 ...............................................................................108 Figure 4.12 DOC breakthrough curves for processing of Tolt River water using IOCO, IOCS and XAD-8 adsorbents...................................................................................109 Figure 4.13 A254 breakthrough curves for processing of Tolt River water using IOCO, IOCS and XAD-8 adsorbents...................................................................................110 Figure 4.14 Cumulative DOC retention efficiency by IOCO, IOCS and XAD-8 processing Tolt River water.....................................................................................111 Figure 4.15 Cumulative A254 retention efficiency by IOCO, IOCS and XAD-8 processing Tolt River water.....................................................................................112 Figure 4.16 Sulfate breakthrough curves for IOCS mediumError! Bookmark not defined. Figure 4.17 Relation between initial SUVA and XAD-8 resin adsorbability for several surface waters ..........................................................................................................116 Figure 4.18 Relative amounts of NOM sorbed onto XAD-8 and XAD-4 resins in series117 Figure 4.19 Relationship between the adsorption of DOC and SUVA using XAD-8 and XAD-4 resins in series versus XAD-8 alone ...........................................................118 Figure 4.20 Relationship between DOC concentration efficiency using reverse osmosis and SUVA for surface waters ....................................Error! Bookmark not defined. Figure 5.1 DOC distribution of the Suwannee River.......................................................124 Figure 5.2 DOC distribution of the South Platte River....................................................125 Figure 5.3 FTIR spectra of hydrophobic and transphilic Suwannee River NOM fractions130 Figure 5.4 FTIR spectra of hydrophilic and ultrahydrophilic NOM fractions from the Suwannee River .......................................................................................................130 xix
Figure 5.5 FTIR spectra of hydrophobic and transphilic NOM fractions from the first sampling of the South Platte River ..........................................................................132 Figure 5.6 FTIR spectra of hydrophilic and ultrahydrophilic NOM fractions from the first sampling of the South Platte River ..........................................................................133 Figure 5.7 FTIR spectra of neutrals NOM fractions isolated from the second South Plate River sampling .........................................................................................................134 Figure 5.8
13
C-NMR spectra of the hydrophobic and transphilic fractions of Suwannee
River NOM ................................................................Error! Bookmark not defined. Figure 5.9
13
C-NMR spectra of the hydrophilic and ultrahydrophilic fractions of
Suwannee River NOM...............................................Error! Bookmark not defined. Figure 5.10 13C-NMR spectra of the hydrophobic and transphilic fractions of South Platte River NOM from the first sampling event.................Error! Bookmark not defined. Figure 5.11 13C-NMR spectra of the hydrophilic and ultrahydrophilic fractions of South Platte River NOM from the first sampling event.......Error! Bookmark not defined. Figure 5.12
13
C-NMR spectra of the NOM fractions of South Platte River from the
second sampling event ...............................................Error! Bookmark not defined. Figure 5.13 1H-NMR spectra of NOM fractions from the Suwannee RiverError! Bookmark not defined. Figure 5.14 TDCA and TDAA contents of the Suwannee River NOM isolates .............139 Figure 5.15 TDCA and TDAA contents of the South Platte River and its NOM isolates140 Figure 5.16 TDCA distribution of the South Platte River and its NOM isolates ............142 Figure 5.17 TDCA distribution of the Suwannee River and its NOM isolates ...............143 Figure 5.18 TDAA distribution of the South Platte River and its NOM isolates ............144 Figure 5.19 TDAA distribution of the Suwannee River NOM isolates...........................144 xx
Figure 5.20 Pyrolysis GC/MS chromatograms of hydrophilic acid (HPIA) and hydrophilic neutral (HPIN) fractions of NOM from Suwannee RiverError! Bookmark not defined. Figure 5.21 Pyrolysis GC-MS chromatogram of hydrophilic base (HPIB) fraction of NOM from Suwannee River ......................................Error! Bookmark not defined. Figure 5.22 Pyrolysis GC-MS chromatograms of hydrophobic acid (HPOA) and
hydrophobic neutral (HPON) fractions of NOM from South Platte RiverError! Bookmark not defined Figure 5.23 Pyrolysis GC-MS chromatograms of transphilic acid (TPHA) and transphilic neutral (TPHN) fractions of NOM from South Platte RiverError! Bookmark not defined. Figure 5.24 Specific UV absorbance (SUVA) spectra of Suwannee River NOM fractions.151 Figure 5.25 Correlation between SUVA254 and half-width of the ET band for Suwannee River NOM concentrates .........................................................................................153 Figure 5.26 Set of selected fluorescence emission spectra of Suwannee River NOM fractions .....................................................................Error! Bookmark not defined. Figure 5.27 Normalized fluorescence emission spectra of selected Suwannee River NOM fractions ...................................................................................................................154 Figure 5.28 Comparison of the fluorescence yield of the Suwannee River NOM fractions155 Figure 5.29 Relationship between the half-width of the ET band in the UV absorbance spectra of the Suwannee River NOM fractions and the position of the maximum in the fluorescence emission spectra............................................................................156 Figure 5.30 Specific UV absorbance spectra of South Platte River NOM fractions.......157 Figure 5.31 Normalized fluorescence emission spectra of South Platte River NOM fractions ...................................................................................................................159 Figure 5.32 Tentative correlation between nitrogen content and TDAA content for NOM fractions from the Suwannee and South Platte Rivers.Error! Bookmark not defined. xxi
Figure 5.33 Relation between the relative proportion of proteins and TDAA content in NOM fractions from the Suwannee and South Platte Rivers. .................................161 Figure 5.34 Relationship between the relative proportion of polysaccharides and the anomeric carbon content in NOM fractions from the Suwannee and South Platte Rivers. ......................................................................................................................162 Figure 5.35 Correlation between aromatic C and SUVA in NOM fractions from the Suwannee and South Platte Rivers. .........................................................................163 Figure 5.36 Correlation between the relative proportion of PHA and SUVA254 in NOM fractions from the Suwannee and South Platte Rivers.............................................164 Figure 5.37 Correlation between the values of SUVA254 and λmax and the percentage of
nitrogen in the South Platte River fractions estimated using elemental analysis.Error! Bookmark not d Figure 5.38 Correlation between the aromaticity of the Suwannee River NOM fractions and ∆ET. ..................................................................................................................165 Figure 5.39 Dependence of the fluorescence yield of the South Platte River NOM fractions vs. the concentration of tyrosine and phenylalanine.Error! Bookmark not defined. Figure 5.40 Correlation between the aromaticity of the Suwannee River NOM fractions and the position of the maximum in the fluorescence emission spectra..................166 Figure 5.41 Correlation between the position of the emission maxima in the fluorescence spectra of the Suwannee River NOM fractions and the aromatic carbon content of the sample, based on Pyr-GC-MS data. ...................................................................167 Figure 5.42 Correlation between the position of maximum in the fluorescence emission spectra and the content of proteinaceous and polyhydroxyaromatic carbon (based on Pyr-GC-MS analysis) in the South Platte River NOM fractions .............................168 Figure 5.43 Comparison of FTIR spectra of transphilic neutral fractions between Suwannee River and South Platte River. .................................................................170 xxii
Figure 5.44 Relationship between DOC or A254 removals and SUVA254....................175 Figure 5.45 Relationships between DBPFPs and SUVA254 for NOM fractions from the Suwannee and South Platte Rivers ..........................................................................184 Figure 5.46 Relationship between TOXFP and THMFP in Suwannee and South Platte River NOM fractions ...............................................................................................186 Figure 5.47 Relationship between TCAAFP and DCAAFP for the Suwannee River and South Platte River NOM............................................Error! Bookmark not defined. Figure 5.48 TOXFP versus destruction of A254 by chlorination for Suwannee and South Platte NOM fractions .................................................Error! Bookmark not defined. Figure 6.1 Bromide concentrations in the Blavet River during a one-year sampling period192 Figure 6.2 DOC and A254 in the Blavet River during a one-year sampling period .......193 Figure 6.3 SUVA254 in the Blavet River during a ne-year sampling period..................193 Figure 6.4 TDAA distribution of the Blavet River (winter sample)................................199 Figure 6.5 TDCA distribution of the Blavet River (winter sample)................................200 Figure 6.6 TDAA distribution of the Blavet River (summer sample) .............................200 Figure 6.7 TDCA distribution of the Blavet River (summer sample) .............................201 Figure 6.8 FTIR spectra of NOM fractions isolated from Blavet River (winter sample) using XAD resins.......................................................Error! Bookmark not defined. Figure 6.9 FTIR spectra of NOM fractions isolated from Blavet River (summer sample) using XAD resins.......................................................Error! Bookmark not defined. Figure 6.10
13
C-NMR spectra of NOM fractions isolated from Blavet River (winter
sample) using XAD resins .........................................Error! Bookmark not defined.
xxiii
Figure 6.11
13
C-NMR spectra of NOM fractions isolated from Blavet River (winter
sample) using membranes with and without further desalting proceduresError! Bookmark not defined Figure 6.12
13
C-NMR spectra of NOM fractions isolated from Blavet River (summer
sample) using XAD resins .........................................Error! Bookmark not defined. Figur 6.13
13
C-NMR spectra of NOM fractions isolated from Blavet River (summer
sample) using membranes with and without further desalting proceduresError! Bookmark not defined Figure 6.14 Pyrolysis GC-MS chromatogram of XAD-8/XAD-4 mixture NOM isolate from Blavet River (winter sample) ............................Error! Bookmark not defined. Figure 6.15 Pyrolysis GC-MS chromatograms of RO and NF NOM isolates (without desalting) from Blavet River (winter sample) ...........Error! Bookmark not defined. Figure 6.16 Pyrolysis GC-MS chromatograms of RO and NF NOM isolates with further desalting from Blavet River (winter sample).............Error! Bookmark not defined. Figure 6.17 Pyrolysis GC-MS chromatogram XAD-8/XAD-4 mixture and RO NOM fractions isolated from Blavet River (summer sample)Error! Bookmark not defined. Figure 6.18 Pyrolysis GC-MS chromatogram of XAD-8/XAD-4 mixture and RO (without desalting) NOM isolates from Blavet River (summer sample)Error! Bookmark not defined. Figure 6.19 Set of UV spectra of NOM concentrates from the Blavet River winter sampling period........................................................................................................210 Figure 6.20 Correlation between SUVA254 and ∆ΕΤ for NOM concentrates from the Blavet River .............................................................................................................214 Figure 6.21 Selected fluorescence emission spectra of NOM concentrates from the Blavet River.........................................................................................................................215 Figure 6.22 Normalized fluorescence emission spectra of selected NOM concentrates from the Blavet River winter sample .......................................................................216 xxiv
Figure 6.23 Correlation of the ET band half-width and the wavelength of maximum fluorescence emission intensity ...............................................................................217 Figure 6.24 Correlation between the percentage of carbon in low-ash NOM concentrates and percentages of aromatic carbon as determined through
13
C-NMR and
Pyr-GC-MS data ......................................................................................................218 Figure 6.25 Correlation between the values of SUVA254 and the percentage of polyhydroxyaromatic and total aromatic carbon .....................................................220 Figure 6.26 Correlation between ∆ET and the percentage of polyhydroxyaromatic and total aromatic carbon in the Blavet NOM fractions based on Pyr-GC-MS data .....221 Figure 6.27 Correlation between the position of maximum in the fluorescence emission spectra and the percentage of aromatic carbon based on 13C-NMR data ................222 Figure 7.1 Correlation between the aromaticity of NOM estimated using
13
C-NMR
spectroscopy and corresponding values of SUVA254 and ∆ET .............................229 Figure 7.2 Correlation between the aromaticity of NOM estimated using
13
C-NMR
spectroscopy and corresponding λmax values ...........................................................230
xxv
FOREWARD
xxvi
ACKNOWLEDGMENTS The authors are grateful to the following water utilities and individuals for their support, cooperation and participation in this project: Skagit County Public Utility District, Mt. Vernon, Wash., Greg Peterka and Greg Hamilton Seattle Public Utilities, Seattle, Wash., David Hilmoe and Bryan Hoyt A great deal of the experimental work was conducted by Dr. Laurence LabouyrieRouillier and David Violleau (MS student) at the Université de Poitiers (France) and Dr. Chi-wang Li at the University of Washington. Dr. George Aiken of the USGS (Denver, Colo.) carried out the XAD-8 and XAD-4 isolations of the water from the South Platte River. The experimental effort contributed by these three individuals, as well as their input to help interpret the experimental results, are greatly appreciated. The advice of the Project Advisory Committee (PAC) – including Erika Hargesheimer, City of Calgary, Calgary, Alb.; Peter Huck, University of Waterloo, Waterloo, Ont.; Gayle Newcombe, Australian Water Quality Centre, Adelaide, South Australia; Douglas Owen, MalcolmPirnie, Inc., Carlsbad, Calif.; and Earl Peterkin, Philadelphia Water Dept., Philadelphia, Pa. – and of AWWARF project officer Jeff Oxenford, are also appreciated.
xxvii
EXECUTIVE SUMMARY
BACKGROUND The material referred to as natural organic matter (NOM) is an extremely complex mixture of organic compounds found in all potable water sources. NOM is typically dominated by humic substances generated by biological activity both in the watershed surrounding a water source (allochthonous NOM) and within the water source itself (autochthonous NOM). In addition to humic substances, proteins, polysaccharides, and other classes of biopolymers also contribute to NOM. Monomeric species such as simple sugars and amino acids are also present in water sources, but they are less abundant because they are subject to relatively rapid biodegradation. Although the name implies that NOM is of natural origin, as a practical matter the molecules collected as NOM from any water source include many organic compounds contributed by human activities. NOM is important in many different reactions and processes that affect water quality. For instance, it provides precursor material for most halogenated and oxygenated disinfection by-products, substrate for biogrowth in water treatment and distribution systems, and complexation sites for binding of heavy metals, and it affects the behavior of colloidal matter by binding to the colloids’ surfaces. To understand these processes better and to control their effects on drinking water quality, it is necessary to understand the chemistry of NOM. However, due to the diversity of the molecules that constitute NOM and the relatively low concentration of NOM in potable water sources (typically, 2 to 10 mg/L when quantified as dissolved organic carbon (DOC)), methods are needed that can either characterize NOM in dilute solutions containing a variety of other chemicals (i.e., in situ) or that can isolate NOM without altering its properties. The current research project addressed these issues.
xxviii
APPROACH A three-fold research approach was used. First, the performance of several NOM isolation methods was compared. The techniques investigated included evaporation, reverse osmosis (RO), nanofiltration (NF), and adsorption onto several adsorbents, including XAD-4 and XAD-8 resins and iron-oxide-coated sand and olivine (IOCS and IOCO). These techniques were compared with respect to the efficiency with which the NOM was collected and separated from other constituents of solution and the type of NOM that was lost (i.e., that could not be collected efficiently) during the process. Next, NOM from two water sources was fractionated based on the hydrophobicity and acidity of the molecules, and the various fractions were compared using several stateof-the-art analytical methods. These methods included
13
C-NMR spectroscopy;
pyrolysis-gas chromatography-mass (Pyr-GC-MS) spectrometry; Fourier transform infrared (FTIR) spectroscopy; analyses of elemental composition and total dissolved carbohydrates and amino acids (TDCA and TDAA); and ultraviolet (UV) and fluorescence spectroscopy. This part of research shed light on the variability of molecules contributing to NOM, its site-specificity, and, to a limited extent, the effects of seasonal processes on NOM composition. It also highlighted the potential value and limitations of the analytical methods employed in NOM research. Finally, the effects of NOM isolation procedures on the chemical composition and reactivity of the isolates was investigated. This part of the study used several water sources from Europe and the Pacific Northwest of the United States, but the report focuses on the results from the source that was studied most intensively: the Blavet River in France. In the report, the feasibility and adequacy of in situ and ex situ techniques for studying NOM are addressed. Overall, this project represents a concerted and consistent effort to gain more insight into the intrinsic properties of NOM and to develop practical methods to isolate and study it.
xxix
RESULTS AND CONCLUSIONS NOM Concentration Methods. Isolation of 100% of the NOM in a water sample is an admirable, but unattainable goal. No technique currently available is capable of efficiently isolating low molecular weight, low aromaticity organic compounds. Evaporation is a very efficient technique for concentrating the non-volatile NOM in a sample. However, this technique requires a great deal of time and effort, and it does not isolate the NOM from the inorganic matrix. This concentration approach might be adequate for subsequent analysis of the NOM by Pyr-GC-MS, but not for analysis by 13
C-NMR or FTIR spectroscopy, or for determination of elemental composition. RO and NF afford relatively convenient ways to concentrate NOM from large
volumes of water. RO allows isolation and recovery of 90 to 95% of the DOC and virtually 100% of the UV absorbance in a sample. NF membranes are usually 5 to 15% less efficient than RO with respect to recovery of both DOC and UV absorbance. Due to the high surface area of RO and NF membranes, it is not uncommon for up to 10% of the NOM to be trapped within the cartridge. At least part of this NOM can be eluted by rinsing the membranes with deionized water or a sodium hydroxide solution, but the efficiency of such a step and the compatibility of the NaOH with the membrane depend on the particular water and membrane under study. Like evaporation, RO and NF concentrate the inorganic salts in the sample along with the NOM. In all these cases, the inorganic (ash) content of the material isolated by RO or NF depends on the inorganic content of source water. The XAD series of macroreticular resins has been used extensively in the past to concentrate and isolate certain fractions of NOM. Passage of a sample through XAD-8 and XAD-4 resins in series collects 60 to 80% of the NOM and yields NOM concentrates of high purity (ash HPO > TPI > HPI, where HA denotes the fraction of humic acids that has a mixed colloidal/polymeric nature. A similar trend applies in South Platte River NOM fractions, but the SUVA values are much lower in the South Platte than for the corresponding fractions from the Suwannee River. For the Suwannee River NOM fractions, fluorescence emission yields increase and the peak in the emission spectrum shifts to lower wavelengths (it is ‘blue-shifted’) as the hydrophilic character of the NOM increases. This trend probably reflects a decrease in average molecular weight as hydrophilic character increases. Fluorescence data for South Platte River NOM fractions suggest that the average molecular weight in this NOM is less than for the corresponding Suwannee River NOM fractions, but that molecular weight tends to increase as the hydrophilic character of the fraction increases. The protein-rich fractions derived from both NOM sources exhibit strong fluorescence that is blue-shifted compared with that of highly aromatic fractions. Moderate correlations were found between SUVA and coagulability of the fraction. Acid fractions were more efficiently coagulated with alum than were neutral or base fractions. The coagulability of the Suwannee NOM fractions was higher than that of the South Platte NOM fractions. Seasonal changes of NOM were studied using NOM extracted from the Blavet River during summer and winter sampling periods. The hydrophobic and transphilic NOM comprised 57 and 21%, respectively, of the NOM in the summer sample, and 79 and 11% in the winter sample. The most dramatic differences between the summer and xxxiii
winter samples were the higher concentrations of TCAA and TDAA in the summer sample. The properties of the humic components of the NOM were not significantly different in the two seasons. Ex Situ and In Situ NOM Characterization. A combination of methods is required to explore the chemistry of NOM and to probe its reactions in potable water systems. Among the structure-sensitive methods investigated, 13C-NMR is notable for its ability to quantify the abundance of organic carbon types. This analysis requires dry, desalted NOM samples containing at least 50 mg of organic carbon. The precision of 13
C-NMR in estimating the fraction of the carbon that is associated with aromatic and
carbonyl or carboxyl functional groups in NOM is limited, so
13
C-NMR data should be
considered as semi-quantitative in this regard. Pyr-GC-MS requires a few milligrams of dry sample, but unlike the case for 13
C-NMR analysis, the sample need not be desalted. Pyr-GC-MS is a semi-quantitative
analytical tool that yields information about the distribution of molecules belonging to various biopolymer classes. The interpretation of NOM pyrochromatograms is more subjective than that of NMR spectra, since only a fraction of the pyrolysis fragments is used for the interpretation. The interpretation may be further complicated by the presence of secondary reactions of pyrolysis fragments. Pyr-GC-MS is one of the newest analytical techniques available for NOM characterization, and more effort is needed to establish and verify approaches for interpreting the output. However, it is clear that this technique provides a unique and distinct NOM fingerprint that can be a valuable adjunct to other information for characterizing NOM. FTIR analysis also requires only a few milligrams of dry sample and is useful as a monitoring tool for ascertaining the inorganic composition of NOM isolates, since it can detect the presence of ammonium, bicarbonate, carbonate, nitrate, silicate, and sulfate (but not that of inorganic halides). It was used for that purpose in the current study. FTIR is a qualitative spectrometric tool that is most valuable as a source of supplementary structural information about inorganic and organic components of the sample, in conjunction with more quantitative methods such as xxxiv
13
C-NMR and elemental analyses.
For example, the carbonyl group of acids, amides, and esters is not resolved in 13C-NMR spectra, but there is sufficient resolution of these groups in FTIR spectra to indicate whether an NOM isolate is predominately an acid (as in humic substances), an ester (as in certain tannins), or an amide (as in proteins and amino sugars). Complementary information on aliphatic hydrocarbon and aliphatic alcohol composition of NOM isolates is also provided by FTIR spectrometry. The sensitivity of FTIR spectrometry to inorganic constituents is both a useful feature and a problem: it is useful because the FTIR signal can indicate whether or not the sample has been desalted successfully, but it is problematic because, if the sample has not be efficiently desalted, the signal from the organic constituents cannot be discerned. Analyses for elemental composition and specific classes of chemical species in NOM can complement structure-sensitive analytical techniques. Elemental analysis can be reliably conducted only on dry, low-ash NOM isolates. Such analyses provide the most direct measure of the efficiency of protocols for both concentrating and desalting NOM. Although elemental analysis does not appear to be a very specific characterization tool, it does provide important information, such as elemental ratios that allow various NOM fractions isolated from the same source to be compared. For instance, the C/O ratio is an indicator of the oxygenated functional group content in NOM, the C/N ratio indicates the content of nitrogenous functional groups, and the C/H ratio indicates the degree of unsaturation of the NOM. The C/N ratio might also be a good indicator of the origin (autochthonous versus allochthonous) of the NOM. Analysis for total dissolved amino acids and carbohydrates can be conducted on natural or treated waters or on NOM isolates. Typically, amino acids and carbohydrates comprise a few percent of the DOC of surface waters. The distribution of monomeric species in TDAA and TDCA is not very site-specific or indicative of the NOM generation processes. The only exception is ornithine, which might be a good indicator of microbial (algal, rather than microbial**? JP) activity in natural waters. Given the effort required to carry out these analyses, the evaluation of TDAA and TDCA content does not merit high priority for structural characterization of NOM. On the other hand, these xxxv
analyses are certainly useful in NOM biodegradability studies, since the analytes represent a significant part of the biodegradable DOC (BDOC) of the NOM. Techniques that can be used on unaltered samples typically require much less time and effort per analysis than do the techniques described above. The techniques meeting this criterion that were utilized in the current study were UV and fluorescence spectroscopy. Only carbon that is in aromatic moieties has been shown unambiguously to affect the UV absorbance spectrum of NOM. The SUVA value at 254 nm (SUVA254) can be used as a reasonably good indicator of the aromaticity of NOM (as quantified by either
13
C-NMR or Pyr-GC-MS). Determination of SUVA254 is much easier and less
expensive than analysis of NMR or Pyr-GC-MS spectra, but even SUVA254 requires some off-line analysis (for DOC). An alternative parameter that seems to provide comparable information is the width of the so-called electron-transfer band (∆ET) in the UV absorbance spectrum, which can be calculated using the ratio of absorbances at 350 and 280 nm (A350/A280) without any off-line analysis. ∆ET is correlated to both NOM aromaticity and, presumably, its molecular weight. The dependence of ∆ET on NOM properties should be investigated in more detail. Fluorescence spectroscopy is an extremely sensitive method that permits NOM to be studied in solution at DOC concentrations 90% of the DOC in most natural waters (Thurman 1985). Most dissolved humic substances are thought to have molecular weights of a few hundred to a few thousand atomic mass units (amu) (McIntyre et al. 1997, Remmler et al. 1995, Wershaw and Aiken 1985). Humic molecules contain aromatic, carbonyl, carboxyl, methoxyl, and aliphatic units (Stevenson 1982, Christman et al. 1989, Perdue 1985, Gjessing 1976), with the phenolic and carboxylic functional groups providing most of the protonation and metal complexation sites. As opposed to synthetic polymers and many biological polymers (e.g., proteins), humic molecules are not comprised of unique, highly reproducible monomeric building blocks (Christman et al. 1989, Hess and Chin 1996, Pompe et al. 1992). Rather, a group of similar building blocks is probably present in many humic molecules, but the sequence and frequency of occurrence of the building blocks, and the exact structure of the regions between adjacent building blocks, is probably different in every humic molecule. For this and other reasons, the chemistry of humic substances is distinct in many ways from that of conventional polymers. Previous investigations of NOM from a wide variety of sources has led to some generalizations about the characteristics of NOM molecules in different environments. For instance, environments in which water is exposed to mineral surfaces that complex and adsorb NOM contain low concentrations of dissolved NOM, especially humic substances. NOM in lakes and reservoirs of moderate to high trophic status is often dominated by material generated in the water body (autochthonous material), whereas low-order rivers and streams usually carry more NOM that is generated exterior to the water body (allochthonous NOM). Allochthonous NOM has large C/N ratios (near 3
100:1), is highly colored, and has significant aromatic carbon content, whereas autochthonous NOM has lower C/N ratios (near 10:1), is almost colorless, and has low aromatic carbon content (Aiken, McKnight et al. 1991). No attempt to fractionate NOM can be completely successful, for both practical and theoretical reasons. For instance, separation of the humic fraction from the nonhumic fraction is impeded by the fact that molecules from the two groups can form complexes or otherwise associate with one another in solution (Lytle and Perdue 1981, Leenheer et al. 1989, Boerschke et al. 1996, Cook and Langford 1998, Volk et al. 1997, Jahnel and Frimmel 1996). Nevertheless, conventional operational approaches have been developed for carrying out these separations. The most common approach for distinguishing between hydrophobic and hydrophilic dissolved NOM is to define them as the organic matter that is adsorbable and non-adsorbable, respectively, on XAD-8 resins. Hydrophobic NOM is often further separated into humic acids (defined as the fraction that coagulates at pH t breakthrough ⎪ ⎭
(Equation 2.4)
Consider a group of NOM molecules that have strong and approximately equal affinities for the adsorbent, so that they break through the column as a square wave. After breakthrough, the concentration of these NOM molecules in solution throughout the column is Cinfluent. The concentration of NOM adsorbed to the media after breakthrough is also constant throughout the column and can be represented as the product of a constant and Cinfluent:
C ads = k C influent
(Equation 2.5)
where Cads is the concentration of adsorbed NOM in mg of organic carbon per liter of adsorbent media, and k is the conditional adsorption equilibrium constant (applicable for the given influent composition), sometimes called a distribution coefficient. The adsorbed concentration can also be expressed as the mass of organic carbon adsorbed per liter of packed bed ( C 'ads ) by multiplying both sides of Equation 2.5 by the bed porosity ε. Defining k' as k ε, Equation 2.5 becomes:
14
C 'ads = C ads ε = k εC influent = k' C influent
(Equation 2.6)
k' is called the column capacity factor. It is the parameter used in the chromatography literature to characterize the affinity of adsorbents for dissolved molecules. At the time of breakthrough, the total mass of the molecules of interest that are in the column includes εVbedCinfluent in solution plus Vbed k' Cinfluent bound to the media, for a total of (ε + k')VbedCinfluent. The total mass of such molecules applied to the column up to the time of breakthrough is VinfluentCinfluent. Assuming 100% retention up to that time, these latter two terms can be equated to yield: (ε + k')VbedCinfluent = VinfluentCinfluent
ε + k' =
(Equation 2.7)
Vinfluent = BVbreakthrough Vbed
(Equation 2.8)
where BVbreakthrough is the number of bed volumes treated before breakthrough. Typically, k′>>ε, so Equation 2.8 can be approximated as: k' =
Vinfluent = BVbreakthrough Vbed
(Equation 2.9)
Equation 2.9 makes the point mathematically that is described in text above, that the column capacity factor (which is used to describe sorption in chromatography columns) is directly related to the number of bed volumes of water that can be processed prior to breakthrough (which is used to describe sorption in columns used for water treatment). In the literature, adsorptive columns are sometimes described as being operated ‘at a column capacity factor of x’. What this really means is that the amount of sample being applied to the column corresponds to the amount that would cause molecules with a column capacity factor (k') equal to x to just barely break through the column. Put another way, if a column is operated at a column capacity factor of x, the number of bed volumes of water being processed equals x. Under these conditions, molecules with k' equal to or greater than x would be almost completely sorbed (assuming a square wave
15
breakthrough pattern), and molecules with k' values less than x would break through the column before the processing is complete and would be incompletely captured. The lower the value of k' for a group of molecules relative to x, the less efficiently the molecules will be captured by the adsorbent.
Sorption onto Fe oxides and ion exchange resins
Both iron and aluminum oxides are used to collect NOM and aid in its removal from drinking water in coagulation processes. The mechanisms by which the coagulants react with NOM are essentially identical. In this section, the focus is on NOM interactions with iron oxides, since iron oxides were used as NOM adsorbents in the research. The ability of iron oxides to adsorb many inorganic ions and NOM molecules is well established (Dzombak and Morel 1990; Levashkevich 1966; Parfitt et al. 1977; Parfitt and Russell 1977; Tipping 1981; Tipping and Cooke 1982; Loder and Liss 1982; Gu et al. 1994, 1995; Korshin et al. 1996). NOM binds to iron oxides via specific chemical interactions, as indicated by the fact that the oxide surface acquires a negative charge when sufficient NOM sorbs to it (Tipping and Cooke 1982, Loder and Liss 1982). Many of the specific chemical interactions are believed to involve replacement of surface-coordinated H2O or OH− groups by carboxyl functional groups of the NOM molecules (Parfitt and Russell 1977; Gu et al. 1994, 1995) in reactions that can be represented generically as follows: R-COO− + ≡FeOH ↔ ≡FeOOC-R + OH−
(Reaction 2.1)
Consistent with Reaction 2.1, adsorption of NOM on hydrous iron oxides generally decreases with increasing pH, with sorption typically becoming negligible at pH > 8 (Tipping and Cooke 1982, Loder and Liss 1982). On the other hand, pH values considerably higher than 8 are sometimes required to desorb NOM from oxides once it has adsorbed at a lower pH. The above reaction also makes evident the importance of
16
acidic groups in NOM molecules for the performance of iron oxide-based adsorbent media and explains why acidic fractions of NOM adsorb preferentially to such media. Solution pH also affects NOM adsorption by altering the degree of ionization of carboxyl groups in the NOM. At low pH’s, the degree of ionization of acidic groups in NOM molecules decreases, and the overall charge on the molecules becomes less negative or neutral. As a result, electrostatic repulsion between adsorbed NOM molecules decreases and their maximum surface density (the apparent adsorption capacity of the oxide surface) increases. Simultaneously, the charge reduction neutralizes the electrostatic repulsion between ionized functional groups within individual molecules, causing them to form more compact and more hydrophobic structures. This conformational change can decrease the affinity of the molecules for cationic surface sites on hydrophilic surfaces (e.g., for Fe sites on iron oxides), while enhancing NOM adsorption onto hydrophobic surfaces (such as XAD-8 and XAD-4 resins). The net result of the competing processes described above is that the adsorption of NOM onto oxide surfaces increases as pH decreases from around 8 to 4 or 5, and then either remains approximately constant or decreases as the pH is lowered further. Although NOM removal efficiency can be controlled to a substantial extent by adjusting solution pH and the concentration of adsorbent, a portion of the DOC cannot be adsorbed even at very high oxide concentrations. The residual (non-adsorbable) NOM typically includes lower-MW, less polar, and less acidic NOM molecules (Sinsabaugh et al. 1986a, Semmens and Staples 1986, Korshin et al. 1997b). The ability of iron oxide-based adsorbent media to retain NOM fractions may be substantially compromised by competition from naturally occurring inorganic anions. The major ion of concern in this respect is sulfate, which can be present in natural waters at concentrations exceeding 100 mg/L. Sulfate interference with NOM adsorption is expected to be most severe in high-sulfate, low-NOM waters, but even in low-sulfate waters, sulfate ions may be problematic if they adsorb and subsequently elute with the NOM. In such cases, additional desalting steps may be necessary before certain analyses
17
can be carried out on the NOM (e.g., analyses for elemental composition and IR absorption characteristics). Interactions of ion exchange media with NOM molecules bear some similarities to those of iron oxides, although the different adsorbents might target somewhat different groups of molecules, and the sorption of NOM onto iron oxides generally involves more specific interactions. Ion exchange has been used by a number of researchers to collect NOM (along with accompanying salts) and fractionate it into basic and acidic fractions. Basic NOM can be efficiently adsorbed on hydrogen-form, strong acid cation exchange resins and then can be eluted with ammonium hydroxide. However, reactions of ammonia with ketone, ester, and quinone groups might alter the NOM in the process (Thorn et al. 1992). Acidic NOM is difficult to elute from strong-base anion exchange resins, but high recoveries are obtainable using weak-base anion exchange resins to collect the NOM, using sodium hydroxide as an eluent. As with iron oxide adsorbents, inorganic anions are often co-eluted with the NOM from ion exchange media and may present a problem for subsequent processing or analysis.
Sorption onto hydrophobic sorbents (XAD and similar resins)
Leenheer (1984) provided an extensive review of the use of hydrophobic sorbents to isolate and fractionate NOM as of the mid-1980’s. Although activated carbon has high sorption efficiencies for NOM and was used in a number of early studies as an NOM sorbent, only half to two-thirds of the NOM could be eluted, and there was some evidence that eluted NOM was chemically altered. Various polyamide sorbents have been used as NOM adsorbents, but problems with NOM desorption have also limited their use. Presently, the procedure used most extensively for isolating aquatic NOM is sorption onto XAD-type resins. These and similar non-ionic, macroporous resins combine relatively high adsorption affinities for NOM with high elution efficiencies and very low affinities for inorganic salts. The resins adsorb NOM through a combination of 18
non-polar (hydrophobic) and polar (hydrogen bonding and aromatic π-electron) interactions. They have better sorption efficiencies and better stability when exposed to acidic and basic eluents than the C-18 silica based sorbents that are popular for isolating various contaminants from water, and they can be used with higher processing rates. Although the hydrophobicity of the NOM molecules is the major determinant of their tendency to sorb to XAD resins, the molecules’ acid-base characteristics and size play an important role as well, since these characteristics have a strong effect on the hydrophobic-hydrophilic nature of the molecules. (Charged molecules, whether acidic or basic, are much more hydrophilic than their uncharged conjugate species.) Molecular size also helps determine the extent to which NOM molecules bind to the resins, since much of the binding capacity is in internal pores that might not be accessible to large molecules. Over two decades ago, Leenheer and Huffman (1976) proposed using XAD and ion exchange resins in a hierarchical fractionation procedure that characterizes NOM molecules based on their hydrophobic-hydrophilic and acid-base properties. This approach was developed further in subsequent years, and a version of the protocol, first proposed by researchers at the USGS in 1981 (Leenheer 1981, Thurman and Malcolm 1981) has become virtually a reference method for the isolation of humic and fulvic acids. The approach has been further expanded and modified since that time (Aiken and Leenheer 1993, Leenheer 1981, Leenheer and Noyes 1984, Leenheer 1997). It still provides the basic framework for most fractionation studies and is the procedure against which alternative fractionation approaches are measured. In any version of the basic procedure, NOM is separated into two fractions, which are referred to as humic and non-humic by some researchers and as hydrophobic and hydrophilic, respectively, by others. The humic fraction comprises the NOM molecules that adsorb at acidic pH onto an XAD-8 resin column using a column capacity factor (k') on the order of 50 to 100 (different protocols use slightly different k' cut-offs), while the non-humic fraction passes through the column. Each fraction is often further fractionated into acidic, basic, and neutral fractions by selective elution and/or subsequent sorption 19
and elution procedures. Generally, more than 90% of the material adsorbed on XAD-8 can be eluted with base and is therefore identified as hydrophobic acids. Typically, the DOC in surface waters is approximately evenly split between the XAD-8 adsorbable (humic or hydrophobic) and non-sorbable (non-humic or hydrophilic) fractions (Leenheer and Huffman 1976, Leenheer 1981, Thurman and Malcolm 1981, Martin-Mousset et al. 1997). Martin-Mousset et al. (1997) reported that the hydrophobic fraction is generally slightly more abundant in reservoir water (51 to 62% for four water sources) than in river water (41 to 50% for four water sources), perhaps due to the adsorption of hydrophobic NOM onto river-borne sediments. Similar results were obtained by Semmens and Staples (1986) and Collins, Amy and Steelink (1986). Leenheer (1981) proposed a protocol for isolating and fractionating the NOM that is not sorbed by XAD-8, but this protocol has not been used as extensively as the method for isolating the adsorbed (hydrophobic) fraction. Recently, Aiken et al. (1992), Malcolm and MacCarthy (1992), Croue et al. (1993a), and Andrews and Huck (1994) have used XAD-4 resins at acidic pH to collect a portion of the NOM that does not sorb to XAD-8 under the conditions used to collect the hydrophobic NOM. These authors reported that 25 to 30% of the DOC of the raw water, corresponding to 50 to 60% of the NOM passing through the column packed with XAD-8, adsorbs onto the XAD-4. About 75 to 80% of the NOM that sorbs onto the XAD-4 resin can be eluted with base. Although previous authors have referred to the NOM that sorbs to XAD-4 and is released by elution with base as the ‘hydrophilic acid’ fraction or the ‘XAD-4 acids’, this NOM is designated as ‘transphilic’ in the current report, to distinguish it from more hydrophilic NOM that does not sorb to either XAD-8 or XAD-4 under the specified conditions. Processing samples as described above typically collects about 50% of the NOM if only an XAD-8 column is used, and about 75% of the NOM if both XAD-8 and XAD-4 columns are used. In more sophisticated isolation and fractionation procedures, additional steps can be employed to collect more of the NOM. As noted in Chapter 1, these steps typically offer diminishing returns: more and more effort is required to collect progressively smaller incremental amounts of NOM. Nevertheless, in cases where the 20
goals of the study justify these efforts, procedures have been proposed to capture more of the NOM. For instance, Aiken et al. (1992) used XAD-8 and XAD-4 resins in series to isolate essentially all hydrophobic and transphilic NOM molecules with column capacity factors k' > 100, respectively. Recoveries from the XAD-4 resin can be improved by using vacuum evaporation to concentrate the effluent from the XAD-4 resin to the point of salt saturation and then passing the concentrated solution through another column packed with XAD-4 to isolate the NOM molecules with k' between 5 and 100. Some molecules sorb in the second exposure to XAD-4 but not the first exposure because fewer bed volumes of sample are passed through the column in the second exposure (corresponding to the lower k' value, per Equation 2.9) (Aiken and Leenheer 1993). That is, in the first exposure step, enough sample is applied to the column to allow weakly binding molecules to break through, whereas in the second exposure, the processing is stopped when these same molecules have not yet broken through the column. In this report, the NOM that adsorbs in the second exposure to XAD-4 is referred to as ‘hydrophilic’, and the NOM that does not adsorb on the XAD-4 resin at this point as ‘ultra-hydrophilic’. An even more comprehensive NOM fractionation and isolation scheme using XAD-8 resin to fractionate hydrophobic NOM with selective elution into hydrophobic acid, base, and neutral fractions, and usng ion exchange resins to fractionate hydrophilic NOM into acid, base, and neutral fractions has been incorporated by the USGS into a mobile field laboratory. This system has been described by Leenheer and Noyes (1984).
Desalting
Inorganic salts interfere with many analytical procedures for characterizing NOM. The need to remove these salts and the best approach for doing so depend on the nature of the salt species and NOM. This section describes various approaches that have been used to remove salts from raw water or, more often, from water that has been processed to concentrate NOM and in which the salts have been concentrated as well.
21
Ion exchange
Anion exchange resins have been used by a number of researchers to concentrate, fractionate, and isolate NOM acids (Leenheer 1984). The two major problems associated with the use of anion exchange resins are: (1) NOM is difficult to elute from most anion exchange resins, and (2) inorganic anions are also concentrated on the resins and are often eluted with the NOM acids. The elution problem can be overcome by using a weakbase anion exchange resin with a phenol-formaldehyde matrix that reverses its charge from positive at low pH to negative at high pH (Leenheer 1981). One such resin was used to concentrate organic hydrophilic NOM acids in this project. Cation-exchange resins in the hydrogen-form and chelating resins in the cupric ion form have been used to isolate amino acids and peptides from fresh and salt waters (Degens and Reuter 1964, Siegel and Degens 1966). Amino acids and peptides can be eluted from these resins with ammonium hydroxide, which does not displace inorganic cations. Amino acids and peptides usually comprise only a very small percentage of the NOM in water samples.
Precipitation and Co-precipitation
Leenheer (1984) has reviewed several methods that have been developed for co-precipitating NOM with metal hydroxides (Al3+, Cu2+, Fe3+, Mg2+, Mn2+, and Pb2+). These methods are selective for the acid fraction of the NOM and work especially well on the acids that form complexes with metals. Concentrations of metal ions from about 30 to 50 mg/L have been found to remove about half of the DOC from fresh and sea water samples. None of the early studies (1960’s and 70’s) using co-precipitation attempted to recover and purify NOM from the precipitate. However, Aiken and Leenheer (1993) recently described a process in which cupric hydroxide was used to co-precipitate NOM, the precipitate was redissolved in acetic acid, co-precipitated sulfate was removed by precipitation with barium chloride, and barium and copper were
22
removed by ion exchange, so that NOM could be isolated as a residue after vacuum evaporation and freeze-drying of the ultimate solution.
Desalting with XAD resins
Hydrophobic NOM can be separated from salts dissolved in the sample by sorption onto and subsequent elution from XAD-8 resins, and transphilic NOM can desalted in an analogous way with XAD-4 resins. These processes can be applied either to the raw water or to water in which the NOM has been pre-concentrated, e.g., by reverse osmosis, vacuum evaporation, or ion exchange. Pre-concentration of NOM facilitates its recovery, especially for transphilic NOM, because a significant amount of NOM can be collected while processing a smaller volume of sample, thereby facilitating collection of NOM molecules with lower column capacity factors (k') (Aiken and Leenheer 1993). The practical minimum value for k' is 5 for desalting on XAD-4 resin because of band spreading of the salt peak. NOM that is not adsorbed by XAD-8 and XAD-4 resins (i.e., the ultrahydrophilic fraction) is much more difficult to desalt. Although substantial desalting of this fraction is possible, it often requires several steps, each of which targets a few specific salt ions. For instance, zeotrophic distillation (fractional distillation of two miscible solvents that do not form an azeotrope) of a solution containing water and acetic acid can lead to precipitation of sodium chloride and calcium sulfate as acetic acid is enriched during distillation. Therefore, this process can efficiently separate these salts from ultra-hydrophilic NOM (which remains soluble) during the distillation. However, calcium and magnesium chloride, nitrate salts, phosphate salts, silicic acid, and boric acid are soluble in acetic acid and therefore are not separated from the NOM by this process (Audrieth and Kleinberg 1953). Boric acid can be removed as volatile trimethyl borate by dissolving the isolate in methanol and heating to dryness three times (Aiken and Leenheer 1993).
23
Calcium and magnesium can be removed by cation exchange (exchanging these ions for H+), and HCl can then be evaporated from the NOM by azeotrophic evaporation with acetonitrile (Aiken and Leenheer 1993). Alternatively, calcium and magnesium hydroxide can be precipitated with sodium hydroxide at pH 12. Some of the NOM co-precipitates with magnesium and calcium hydroxides in this procedure. However, the NOM can be redissolved and separated from the calcium and magnesium by suspending the solids in a solution of sulfuric and acetic acids and then applying zeotrophic distillation. During this step, calcium and magnesium sulfate precipitate and are separated from the NOM, which remains in solution. Phosphates can be precipitated as magnesium ammonium phosphate (Aiken and Leenheer 1993) or as tri-lithium phosphate (Leenheer 1997). The majority of the nitrate in the sample can be removed by adding a 10-fold molar excess (relative to nitrate concentration) of barium chloride to the ultra-hydrophilic NOM concentrate and removing barium nitrate by zeotrophic distillation in acetic acid. Traces of nitrate can be removed by evaporation with anhydrous formic acid that reduces nitric acid to gaseous nitrogen dioxide, but it is not advisable to use this procedure with large amounts of nitric acid because of the potential for oxidation and nitration of NOM by nitric acid. Separation of silica from ultra-hydrophilic NOM is particularly problematic, because silicic acid polymerizes with itself and with the NOM to form a silica-NOM ester gel when taken to dryness. This problem can be overcome by hydrolyzing the ester by gentle heating with 0.01 N HCl, a process that dissolves the NOM but not the silica gel, so the latter can be removed by centrifugation. Thus, if one knows which salts must be separated from the NOM, methods are probably available to carry out the separation. However, each procedure is often complicated and applicable to only one or a limited suite of salts. Analysis of inorganic constituents is therefore a key prerequisite for designing a sensible and effective concentration, fractionation, and isolation scheme for ultra-hydrophilic NOM constituents in a particular sample.
24
Losses of NOM During Processing
NOM recoveries can be estimated at various stages of an isolation-concentrationfractionation procedure, or for the overall procedure, by a mass balance on DOC, as long as organic reagents are not used in the procedure. The goal of comprehensive NOM isolation is, of course, 100 percent recovery of desalted NOM fractions. However, practical considerations limit the recovery currently achievable to 70 to 90% for freshwater samples and less than 50% for seawater samples. The major limitation is imposed by the presence of salts and is quantitatively related to the salt:DOC concentration ratio. For a soft freshwater sample such as the Suwannee River, this ratio is about 0.2 mg/mg, while for the Mississippi River at New Orleans, it is about 50 mg/mg (Hem 1971**JL (year given as 1970 in refs), Leenheer et al. 1995), and for seawater it is 18,000 mg/mg (Hem 1971, Benner et al. 1992). If all of the salts remain in solution when the hydrophobic NOM is extracted from the sample, the salt:DOC ratio in the remaining (transphilic and hydrophilic) fractions typically increases by a factor of two or more. Therefore, efficient desalting techniques are particularly important if one is to avoid losses of hydrophilic NOM during the processing steps. Removal of salt by zeotrophic distillation, ion exchange, or selective precipitation techniques invariably results in both non-selective and selective losses of NOM. Nonselective losses (i.e., those that decrease the mass of NOM recovered but do not alter its composition) are primarily because of mechanical operations. For instance, as discussed previously, some NOM might be lost by sticking to the walls of freeze-drying vessels. Also, losses of NOM by accidents, such as spills and breakage of containers, occur with increasing frequency as the complexity and difficulty of the isolation procedure increases. Selective losses of certain classes of NOM can occur during various steps that are intended to target other molecules. For instance, some NOM might sorb onto ion exchange resins intended to remove salts from solution, some might volatilize or be lost as foam during vacuum evaporation steps, and some might form precipitates at various
25
stages in the concentration procedures. In addition, organic colloids might be lost at various points in a procedure.
NOM CHARACTERIZATION
Due to the complex, ‘multi-dimensional’ nature of NOM described earlier, a variety of experimental methods has been used to characterize it. For purposes of discussion, these methods can be separated into four tiers. The first tier addresses the chemical identities of individual species included in the NOM pool, such as amino acids and carbohydrates. The second tier addresses the nature and abundance of structural units in the NOM molecules. The relevant methods include elemental composition analysis, 13
C- and 1H- nuclear magnetic resonance (NMR) spectroscopy, Fourier Transform
Infrared (FTIR) spectroscopy, and pyrolysis - gas chromatography - mass spectrometry (Pyr-GC-MS). The techniques associated with the first two tiers provide chemically specific information but require substantial sample preparation and cannot be used in situ. The third tier of methods addresses issues related to the chemical behavior of NOM, often focusing on the polymeric nature of NOM molecules. Techniques to probe the molecular size distribution, acid-base and hydrophobic-hydrophilic properties, and reactivity of NOM are included in this group. This tier of the experimental methods generally requires less sample preparation than the first two. The fourth group of methods comprises those that do not explicitly probe the chemical identities of functional groups or molecules, but measure a spectral signature of NOM in toto, such as UV absorbance and fluorescence spectroscopy. Due to the exceptional sensitivity and experimental simplicity of these methods, they can be used to probe NOM in situ, without preconcentration and with minimal or no pre-treatment. However, interpretation of the data generated by these techniques is usually less direct than for techniques in the other three analytical tiers.
26
13
C and H-NMR spectroscopy
The theory and application of solution and solid-state NMR spectroscopy to obtain quantitative and qualitative information on organic carbon structural distributions has been reviewed by Nanny et al. (1997), Wershaw and Mikita (1987), and Wilson (1987). Solution-state 1H-NMR provides information on non-exchangeable structural proton distributions in NOM isolates. The acquisition and interpretation of 1H-NMR spectra is discussed in a recent report by Leenheer et al. (1987). Carbon (13C), hydrogen (1H), nitrogen (15N), and phosphorus (31P) nuclei in NOM can be probed by NMR spectroscopy; however, only
13
C and 1H have sufficient natural
abundance and relative receptivities (sensitivity) to allow routine determinations of NMR spectra. NMR signals are generated by the absorption and emission of radio frequency (RF) signals of spinning nuclei that are precessing about an axis in a magnetic field. When the RF frequency matches the precession frequency during an RF pulse, RF energy is absorbed and the nuclear spin is shifted to a different energy level. When the nuclei relax to the ground state energy level between RF pulses, RF energy is emitted, and it is this energy that is detected as the NMR signal. The electron field around a nucleus is determined by its chemical structure and affects the NMR signal to various degrees. The magnitude of this effect is defined as the “chemical shift”. Chemical shifts are measured as parts per million frequency differences relative to a standard compound. NMR signals generated in the time domain are converted by Fourier Transform mathematics to frequency domain information that graphs the intensity of an NMR signal (ordinate) to the chemical shift (abscissa). The orientation of NMR nuclei relative to the applied magnetic field must be random to achieve useful NMR signals. This randomization is accomplished either by dissolving the sample (which randomizes the nuclei through molecular Brownian motion) or by spinning a solid sample at a “Magic Angle” relative to the applied magnetic field. The latter approach is referred to as cross-polarization magic angle spinning (CPMAS).
27
Both solution-state and solid-state capabilities are desirable for characterizing NOM structure. Solid-state NMR is generally not as quantitative as solution-state NMR (Thorn et al. 1989), assuming complete solution of NOM fractions at high concentrations. The cost of NMR spectrometers is high ($300,000-$500,000), so NOM isolates are frequently sent to institutions that run the samples. Approximately 50 mg of desalted sample is required to run the analysis. Possible structural assignments for
13
C-NMR and 1H-NMR spectra of NOM
isolates are given in Tables 2.2 and 2.3, respectively. Both tables indicate considerable overlap in the chemical shifts for various types of structures. In addition, in 1H-NMR spectra, there is an intense and sometimes broad peak near 4.6 ppm from exchangeable hydroxyl hydrogen that obscures some of the structural hydrogen groups near this region. For these reasons, quantitative determinations of various structures in NOM is a somewhat subjective exercise that depends on the judgment of the analyst. Table 2.2 Structural assignments for 13C-NMR spectra Chemical linkage
Compound type
Chemical shift range (ppm)
C-H
Hydrocarbon
0-55
C-N
Amines, amides, proteins
40-55
O-CH3
Methoxy groups in tannins and lignins
55-60
C-O
Aliphatic alcohols, ethers, and esters
60-90
O-C-O
Anomeric carbon in carbohydrates, lactols
90-110
φ
Aromatic carbon
95-165
φ-O
Aromatic esters, ethers, and phenols
135-165
O=C-O,O=C-N
Carboxylic acids, esters ,amides
160-190
O=C-C=C
Flavones, quinones
170-200
O=C-C
Aliphatic and aromatic ketones
190-220
28
Table 2.3 Structural assignments for 1H-NMR spectra Chemical Linkage
Compound Type
Chemical Shift Range (ppm)
R-CH3
Aliphatic Hydrocarbons
0.6-0.9
-CH2-
Aliphatic Hydrocarbon Chains
0.9-1.4
O=C-C-CH3
α-Methyl ketones, carboxylic acids
0.9-1.2
O-C-C-H
α-Oxy alcohols, ethers, esters
1.4-1.8
Aliphatic, alicyclic hydrocarbon
1.4-1.8
O=C-CH3
Methyl ketones, carboxylic acids
1.9-2.1
O=C-C-H
α-Methylene and methine ketones, acids 2.13.2
φ-C-H
Aliphatic CH on aromatic rings
2.1-3.2
O-C-H
Alcohols (carbohydrates), ethers, esters
3.2-5.2
O=C-CH-C=O
β-Keto carboxylic acids, ketones
O=C-CH-φ
β-Aryl carboxylic acids, ketones
3.5-5.5
C=C-H
Olefinic hydrogen
5.0-6.4
H-ortho phenols, phenol ethers, phenol esters
6.4-7.0
Aromatic protons in general
6.4-8.5
H-ortho carboxylic acids, ketones
7.6-9.5
O H H
O
H
13
C-NMR with CPMAS may have significant limitations that affect its precision
in estimating the contribution of aromatic and carbonyl or carboxyl carbon in NOM and overemphasize the contribution of other types of carbon in the sample. Variable contact time studies by Alamany et al. (1983) indicate that in the
13
C-NMR cross-polarization
experiments, an optimal signal-to-noise ratio can be achieved at a 1 ms contact time. 29
However, these experimental conditions compromise the precision of the method in estimating the contributions from different structural groups. Comparison of CPMAS 13
C-NMR spectra acquired with 1 ms contact times with highly-precision quantitative
liquid-state
13
C-NMR of aquatic NOM isolates indicate that aromatic carbon is
underestimated by 20 to 40% (phenolic by 50%) and carbonyl (carboxyl, ester, amide) carbon is underestimated by 30 to 50%, and that the aliphatic carbon is overestimated by corresponding amounts. This problem exists because the aromatic rings in humic substances have low and remote protonation. Using a 5 ms contact time gives results much closer to quantitative liquid-state spectra, but a 1 ms contact time has become the standard because the lower field instruments for which the original methods were developed needed the sensitivity that 1 ms contact time provided. Due to this limitation, inferences about the structural distribution of carbon derived from
13
C-NMR data should be considered as semi-
quantitative, similar to those from Pyr-GC-MS data (discussed below). In particular, estimates of NOM aromaticity obtained using 1 ms contact times should be considered as minimum estimates.
FTIR spectroscopy
FTIR spectrometry detects various molecular vibrations (rotations and stretches). When the frequency of the infrared radiation entering a solution or crystal of an organic compound corresponds to the frequency of a molecular motion in the organic compound, radiation is absorbed. A plot of percent transmission versus frequency indicates the relative amounts of molecular stretching and bending vibrations of various atoms in the molecule. Fourier transform mathematics converts the data into a spectrum showing transmission or absorbance of infrared radiation on the ordinate versus the infrared frequency on the abscissa. Infrared spectrometry is especially useful for qualitative identification of oxygen and nitrogen functional groups in NOM.
30
Comprehensive interpretation of FTIR spectra of pure compounds is complex because so many absorption bands are generated. Paradoxically, the complexity of fractionated NOM simplifies interpretation of the spectra because only the strongest bands can be identified and associated with the predominant structures. For interpretation of the spectra of pure compounds, the reader is referred to Pouchert (1985), and for analysis of complex biomolecular structures and humic substances, to Bellamy (1975**JL (year given as 1960 in refs) and Stevenson (1982), respectively. Table 2.4 lists characteristic IR frequency bands for some complex biomolecules typically found in NOM isolates. Table 2.4 Infrared frequency bands for biomolecular structures in NOM isolates Biomolecule
Frequencies (cm−1) and Structure
Carbohydrates
3400-3300 (O-H); 1100-1000 (C-O)
Fulvic Acid
3400-3300 (O-H); 2700-2500 (COOH); 1760 (COOR); 1720 (COOH); 1660-1630 (φ-C=O); 1280-1150 (φ-O; COOH)
Hydrocarbons
2960 (CH3); 2940 (CH2); 1460 (CH2); 1380 (CH3)
Proteins
1660 (Amide-1 band; N-C=O); 1550 (Amide-2 band; N=C-O)
FTIR characterization of isolates can be particularly useful for identifying the proteinaceous component of NOM, which is difficult to identify using
13
C-NMR. In
addition, the relative abundances of hydrocarbons (hydrophobic) and carbohydrate (hydrophilic) moieties indicated by the IR spectrum can be used as an indicator of the hydrophobic-hydrophilic nature of NOM isolates. FTIR spectrometry can also serve as an assay of the purity of NOM fractions because it allows bicarbonate, carbonate, nitrate, phosphate, silicate, and sulfate salts in the sample to be readily detected. Table 2.5 gives characteristic peaks that can be used to identify inorganic contaminants in NOM fractions. By the same token, inorganic salts (with the exception of chloride salts) can be a major interference, so purification
31
requirements are significant for FTIR spectrometry to be a useful tool for NOM characterization. Table 2.5 Characteristic infrared spectral peaks of inorganic solutes (in KBr pellets) Inorganic solute
Characteristic IR peaks (cm−1)
Boric acid
3212, 2260, 1450, 1194, 548
Sodium bicarbonate
2541, 1920, 1695, 1618, 1307, 1000, 837, 696
Sodium carbonate
1440, 880
Sodium nitrate
1385, 838
Phosphoric acid
1007, 490
Disodium hydrogen phosphate
1159, 1074, 950, 860, 544, 521
Silicic acid
1093, 964, 798, 468
Sulfuric acid
1288, 1176, 1071, 1012, 889, 852, 617, 577, 455
Sodium hydrogen sulfate
1251,1182, 1046, 865, 607,577, 481
Sodium sulfate
1122, 640, 608
Pyrolysis-GC-MS
If NOM is degraded thermally, macromolecules that were originally synthesized from natural biopolymers (e.g., polysaccharides, proteins, amino sugars or polyhydroxyaromatics) tend to produce fairly specific by-products. Pyrolysis followed by gas chromatography and mass spectrometry (Pyr-GC-MS) can separate and identify those by-products. While Pyr-GC-MS cannot be used as a strictly quantitative analytical technique, it can provide a fingerprint of the NOM that is quite specific. For instance, it has been used to distinguish among humic substances isolated from various natural waters (Peschel and Wildt 1988 **JP – spelling is Pischel in ref list) and between fulvic and humic acids isolated from the same water source (Bruchet et al. 1986). It is also useful for following the evolution of NOM as a function of season or during drinking water treatment processes (Bruchet et al. 1990, and Bornick. 1996). As pointed out by Saiz32
Jimenez (1994), one major advantage of the technique is that it does not require prior hydrolysis, purification or fractionation of the organics. Table 2.6 presents information about the presumed origin of some natural biopolymers and their pyrolysis by-products. Based on the types and diversity of fragments produced by Pyr-GC-MS, humic acids are reported to be structurally more heterogeneous than fulvic acids and to contain carbohydrates as the most prevalent class of constituents (Bruchet et al. 1986). Humic acids also appear to have a larger phenolic and unsubstituted aromatic content than fulvic acids based on this technique (Gadel et al. 1992). Pyr-GC-MS analysis of NOM suggests that, despite their high specific UV absorbance (SUVA) values, humic acids are highly aliphatic (Gadel and Bruchet 1987, Bruchet et al. 1986), and that proteins and carbohydrates are much more substantial components of NOM than is suggested by other types of analyses. Table 2.6**MB landscape Origin of biopolymers and their respective specific pyrolysis fragments Type
Origin
Pyrolysis by-products
Polysaccharides
Aquagenic (algae and
Hexoses and pentoses: furan,
(stored, e.g., starch, and
bacteria) and pedogenic
furfural, levoglucosenone
structural, e.g., cellulose)
(plant residuals)
Proteins
Aquagenic (algae and
Pyridines, pyrroles, indoles,
phytoplankton)
nitriles; phenol, paracresol (equal quantities, from tyrosine); toluene, styrene, phenylacetonitrile (from phenylalanine); indole from tryptophane
Amino sugars
Cell walls of bacteria and
Amides (acetamide,
fungi
ethanamide)
Polyhydroxy-aromatics
Plants (lignin) and aquatic
Phenolic compounds (e.g.,
(PHA)
(algae, diatoms, animals)
methoxyphenols)
Source: Adapted from Bruchet et al. 1986; Gadel et Bruchet 1987; Bruchet et al. 1990; Biber et al. 1996.
33
The amounts of saturated and aromatic hydrocarbons generated by Pyr-GC-MS of NOM might be useful as indicators of terrestrial organics in the NOM. For instance, Schulten and Plage (1991) reported that benzene and allylbenzene are the major thermal degradation by-products of humic acids isolated from soils. However, this category of by-products can also be generated by pyrolysis of fatty acids, aromatic acids, aryl aliphatics and alcohols (Saiz-Jimenez 1994, Göbbels and Püttmann 1997). Croue et al. (1993b) published Pyr-GC-MS chromatograms of fulvic and transphilic acids isolated from a French reservoir. The chromatogram of fulvic acids gave clear evidence of aromatic structures (large peaks of phenol and cresol). These fragments were present at lower concentrations in the chromatogram of the transphilic acids, which is consistent with the observation that the fulvic acids had a higher SUVA than the transphilic acids. By contrast, transphilic acids contained a higher proportion of sugars and amino sugars (based on large peaks of furfural, methyl furfural, levoglucosenone and acetamide), a fact that is undoubtedly related to their higher hydrophilic character. Harrington et al. (1996) reported that, although phenol is generally the major peak in the pyrochromatogram of hydrophobic NOM (XAD-8 isolates), the relative proportions of the four biopolymer classes identified in Table 2.6 depend on the origin of the humic materials. For the five isolates studied, they found a strong correlation between the phenolic or polyhydroxy aromatic content (based on Pyr-GC-MS) and aromatic carbon content (based on
13
C-NMR spectra). Somewhat surprisingly, the correlation
between amino sugars and aliphatic carbon content was also strong. Using the same approach, Martin (1995) established relationships between polyhydroxy aromatics and the aromatic carbon content and between TDAA content (based on HPLC analysis) and proteins. Further development and calibration of the Pyr-GC-MS technique might allow it to become an important tool for NOM characterization.
34
Elemental Analysis
Elemental analysis is generally among the first approaches that researchers use to the characterize NOM and its isolates. The elements analyzed commonly include carbon, hydrogen, oxygen, nitrogen, and sulfur; the non-oxidizable element content is also usually characterized and reported as ‘ash’. Phosphorus and halogens are analyzed in some cases, but more rarely than the elements listed above. Results are typically given in percent by weight, and some specific ratios (e.g., C/H, C/O and C/N ratios) are reported and used as indicators of particular characteristics of NOM. The elemental analysis data base available in the literature mainly includes hydrophobic acid fractions (with or without fractionation into humic and fulvic acids), along with some results on transphilic acid fractions. Table 2.7 provides a comparison of elemental analyses of humic, fulvic and transphilic acids isolated from three different water sources. In this table, the hydrophobic acid fraction isolated by Aiken et al. (1992) from the Yakima river can be considered comparable to the fulvic acid fraction reported for the other waters, since fulvic acids generally comprise the dominant portion of the hydrophobic acid fraction.
35
Table 2.7 Elemental analysis of hydrophobic acids and transphilic acids isolated from surface waters Source
Fraction
C
H
N
O
S
Ash
Contribution to Fraction Mass (%) Yakima river*
Hydrophobic acids §
56.1
4.95
2.2
35.5
0.97
1.1
Transphilic acids
50.5
4.4
3.0.
40.6
1.2
3.9
Humic acids
55.8
3.58
0.96
36.9
0.32
1.18
Fulvic acids
54.2
3.96
0.56
39.3
0.24
0.33
Transphilic acids
50.2
4.0
0.97
43.8
0.51
0.85
Humic acids
48.1
4.9
3.04
36.1
2.58
4.7
Fulvic acids
49.7
4.9
2.14
39.5
1.88
1.5
Transphilic acids
41.1
4.4
3.1
41.1
1.6
nd**
Lake Skjervatjern †
Apremont reservoir ‡
*Aiken et al. 1992 † Malcolm et al. 1993**not in refs ‡ Martin 1995 § Fulvic acids generally account for 90% of the hydrophobic acids; **MB nd : not determined
For the three water sources described in Table 2.7, the transphilic acids contained less carbon and more oxygen than did the humic and/or fulvic acids from the same origin. These results indicate that oxygenated functional groups are more abundant in the transphilic acids than the hydrophobic acids, as expected (since oxygen-containing functional groups cause NOM molecules to be more hydrophilic). For all sources, fulvic acids had the lowest proportion of nitrogen, while the nitrogen content of transphilic and humic acids was similar. Table 2.8 gives typical elemental analyses for various NOM fractions based on literature data.
36
Table 2.8**MBlandscape Average elemental analysis of humic substances (with or without fractionation to humic and fulvic acids) and transphilic acids isolated from surface waters Fraction
C
H
O
N
S
C/O
C/N
C/H
n
12
Contribution to Mass (%) Humic acids
53.1
4.5
37.4
2.1
1.5
1.4
28.3
12.1
±2.9
±0.6
±2.1
±0.7
±0.9
±0.1
±11.1
±2.3
53.2
4.8
38.3
1.4
0.8
1.4
43.7
11.2
and HPOA *
±2.5
±0.7
±2.0
±0.6
±0.4
±0.1
±18.5
±1.6
Transphilic
45.8
4.4
43.9
2.5
1.0
1.0
24.1
10.6
acids
±3.6
±0.5
±2.1
±1.1
±0.5
±0.1
±13.1
±2.0
Fulvic acids
24
10
Source: Adapted from Reckhow et al. 1990, Aiken et al. 1992, Martin 1995. *HPOA = hydrophobic acids
In a literature review prepared in 1985, Thurman suggested that significant differences could be observed between humic substances isolated from ground waters, surface waters and soils, and that these differences could be related to the physical and chemical characteristics of the media. Nevertheless, only minor differences were observed between humic substances that have a similar origin, indicating that elemental analysis was not specific enough to distinguish among NOM samples isolated from similar types of sources. Table 2.7, which presents more recent data, supports this conclusion. However, the range of values for each element (minimum and maximum values) does indicate that there can be significant differences in the elemental composition of NOM from different sources. **JP add a sentence and reference Perdue’s work.
UV / Visible and Fluorescence Spectrometry
The absorption of both visible and ultraviolet (UV) light by surface waters is widely attributed to the aromatic chromophores (light-absorbing sub-units) present in dissolved NOM, primarily in humic molecules. Humic molecules are also thought to be 37
largely responsible for the fluorescence of natural waters. As a result, the energy (related to the wavelength, λ) and intensity of light absorption and/or emission can be used to infer structural information about the NOM molecules. UV absorbance is attractive analytically because it is simple to carry out, the required instrumentation is relatively inexpensive, and minimal sample preparation is required. Absorbance of ultraviolet light by NOM in the wavelength range from 200 to 400 nm is easier to assess than absorbance of visible light because few inorganic species present in natural fresh waters absorb substantial amounts of light at λ>200 nm (in some waters, bromide and nitrate absorb enough light to be problematic at wavelengths up to 230 nm (Ogura and Hanya 1966, 1968; Mrkva 1969). The aromatic content of NOM, as found by
13
C-NMR, has been reported to
correlate well with UV absorbance at 272 nm (A272), with regression coefficients in the range from 0.70 to 0.94 (Traina et al. 1990, Novak et al. 1992). Resorcinol, catechol, and benzoic, hydroxybenzoic and vanillic acids have all been suggested as model aromatic chromophores that are likely to be incorporated into the structure of NOM (Christman et al. 1989); other aromatic units undoubtedly contribute as well. Because humic species are likely to be the predominant organic reactants in reactions with disinfectants and coagulants, studies of the humic part of NOM can provide important insights into these processes (Korshin et al. 1996, 1997c). In these cases, UV spectrophotometry, which is inexpensive and virtually universally available, can contribute substantially to our ability to predict and monitor the reactions of interest. One drawback of using UV spectroscopy for studying NOM is that the spectra are typically broad and nearly featureless (Ghosh and Schnitzer 1979, Wang et al. 1990). Only minor peaks have occasionally been reported (see, for example, Baes and Bloom 1990), the number of possible types of chromophores is high, and none of the chromophores possesses an easily distinguishable spectrum. Although the UV absorbance spectrum of NOM from any given source could, in theory, be deconvoluted into separate spectra contributed by distinct chromophores, such 38
an approach is not a practical possibility. As a result, the potential value of UV spectroscopy in the study of NOM has remained unrealized. Most researchers have limited their data collection to monitoring the absorbance at 254 nm, using these values as a rough indicator of the overall NOM concentration. The value of SUVA at 254 nm (SUVA254) is also often calculated and used as an indicator or the aromatic, hydrophobic character of the NOM (Traina et al. 1990, Novak et al. 1992). Korshin et al. (1996, 1997c) recently proposed modeling the UV absorbance spectrum of NOM as a composite of three absorption bands, each of which in turn represents a composite of the absorbance from certain electronic transitions in aromatic chromophores in NOM molecules. They suggested that the transitions were similar to those identified as the local excitation (LE), benzenoid (Bz), and electron transfer (ET) bands in simple aromatic compounds (Figure 2.1).
A
εmax>45,000 εmax=7,400
Local excitation (LE) band, max. @ 180 nm (6.88 eV)
B LE band Bz band
εmax=204
Benzenoid (Bz) band, max. @ 203 nm (6.11 eV)
ET band
Electron transfer (ET) band, max. @ 253 nm (4.90 eV)
Ground electronic state
Ground electronic state
Figure 2.1. (A) Conceptual representation of electronic transitions caused by the absorbance of light for benzene (B) Conceptual representation of composite lightabsorption bands for NOM. Source: Adapted from Jaffee and Orchin (1962) and Scott (1964). They represented the absorbance intensity of each model spectrum as a Gaussian function of the corresponding absorption energy. Typically, the three bands have peaks near and 180, 203, and 253 nm, respectively. When the three model spectra are superimposed and summed, the resultant spectrum closely matches the experimental one (Figure 2.2).
39
0.9 Absorbance
unconvoluted spectrum tailing part of LE band
0.6
Bz band ET band
0.3
0.0 190
210
230
250
270
290
310
330
Wavelength, nm
Figure 2.2. Summation of three composite absorption bands and formation of unconvoluted UV absorbance spectrum of NOM Korshin et al. (1996, 1997c) proposed that the absolute and relative intensities of the three model bands, their peak wavelengths (λi,max), and their widths (∆i) provide useful structural information about the NOM molecules, and that alterations in these parameters when the NOM is subjected to various physico-chemical processes could provide information about the NOM reactions in those systems. Fluorescence occurs when optically-excited molecules emit light whose wavelength is longer (‘red-shifted’) than that of the excitation energy. In organic molecules, the emission occurs in functional groups called fluorophores. Like the chromophores that absorb light, fluorophores in NOM are thought to be associated with phenolic groups in the humic portion of NOM (Ghosh and Schnitzer 1980, 1981; Lochmueller and Saavedra 1986; Vinodgopal and Kamat 1992; Goldberg and Negomir 1989). Aromatic amino acids (e.g., tyrosine, tryptophan and phenylalanine) also fluoresce intensely (Coble 1996) and thus may contribute to the emission (Laane and Koole 1982). It is estimated that 250 nm is Gaussian and that the maximum absorbance of the band is at ~4.90 eV (252 to 254 nm). Using those assumptions, ∆ET can be computed from Equation 2.10 (Korshin et al. 1996, 1997c).
41
∆ ET
⎛ ⎛ A ⎞⎞ = 218 . ⋅ ⎜ ln⎜ 280 ⎟ ⎟ ⎝ ⎝ A350 ⎠ ⎠
−
1 2
(Equation 2.10)
Compound Class Identification
Amino acids and sugars are present in both free and combined form in natural waters. The combined forms, which dominate over the free forms in surface waters (Thurman 1985), include associations with polypeptides, proteins, and polysaccharides or with humic substances. Ittekkot et al. (1982) demonstrated the importance of amino acids and sugars as possible tracers of the different types of NOM transported by water bodies. For instance, an increase in the concentration of arabinose correlates with increasing concentrations of β-alanine and γ-amino-butyric acid and is an indicator of NOM of bacterial origin. Total dissolved amino acids (TDAA) and total dissolved sugars (referred to in this report as total dissolved carbohydrates, TDCA) are typically present in surface waters at mean concentrations of 300 µg/L and 500 µg/L, respectively. These constituents contribute about 2 to 5% and 5 to 10% of the chemical oxygen demand (COD) of such waters. **JP provide a reference? Glutamic acid, glycine, serine and aspartic acid are the major amino acids found in surface waters (Thurman 1985), and glucose is the most abundant sugar. Concentrations of TDAA and TDCA in some surface waters are listed in Table 2.9, and corresponding data for some fractionated NOM samples are presented in Table 2.10.
42
Table 2.9**JP: see note 1
Source
Types Analyzed
Mackenzie River Indus River
Concentration (µg/L)
Contribution to COD (%)
114-566 Total
Orinoco River
347-1213
Ittekkot et al. (1982)
65-284 Total dissolved
300
2-3
Oise River
Total dissolved
462
4.5
Marne River
Total dissolved
372
5
Seine River
Total dissolved
378
5
Zelivka River Vlata River
Reference
Thurman (1985)
Dossier-Berne (1994)
270 Combined
Berounka River
450
Chudoba et al. (1986)
337 Sugars
Mackenzie River
Total
520-1540
13-40
Indus River
Total
267-1141
0.7-11
Orinoco River
Total
103-970
1.7-9
Total dissolved
500
5-10
Thurman (1985)
Ado River
Dissolved neutral
18
2
Ochiai et Nakajima (1998)
Mano River
Dissolved neutral
26
2.3
43
Ittekkot et al. (1982)
Table 2.10**JP see Note 1 Amino acid and sugar concentrations in various NOM fractions Amino Acids Source
Fraction
Suwannee River
Humic acids
Concentration AA (nmol/mg C) (% N)
C/N
Reference
110
-
-
Thurman and Malcolm (1989)
Ohio River
307.5
-
-
Malcolm (1990**JP provide full ref for ref list)
Apremont Reservoir
324
17
15
260
14.5
16
314
22
18
127
-
35
Thoreau’s bog
78.5
-
71
Suwannee River
34
-
-
Thurman and Malcolm (1989)
Ohio River
63.5
-
-
Malcolm (1990)
Lake Fryxell and Lake Hoare
71-98
30.54.9
1722
McKnight et al. (1991)
Apremont Reservoir
133
10
20
120
9
22
205
18
24
198
90.5
12
217
12
14
231
13
13
Mayenne River Shawsheen River
Fulvic acids
Mayenne River Apremont Reservoir
Mayenne River
Hydrophilic acids
44
Martin (1995)
McKnight et al. (1985)
Martin (1995)
Martin (1995)
Fractions Source
Fraction
sugars (µmol/mg C)
Reference
Ohio River
Humic acids
3.9
Malcolm (1990)
Ohio River
Fulvic acids
6.8
McKnight et al. (1985) found that arabinose and mannose account for 74% and 15% of the carbohydrate in hydrolyzed hydrophilic (73.4%) and fulvic (74.6%) acids, respectively. Thurman and Malcolm (1989) demonstrated that, in comparison with fulvic acids, humic acids are enriched in basic, hydroxy-, sulfur-containing, and aromatic amino acids. The major amino acids in fulvic acids are glycine and aspartic acid, and these acids along with hydroxyproline are the dominant ones in humic acids. While the detailed nature of the linkages between humic molecules and amino acids are not well understood, three types of linkages might be important: hydrogen bonds, bonds with metal ions, and covalent bonds. Understanding the nature of the bonds is important both for assessing the biodegradability of these species and for assuring that the analytical procedure used to hydrolyze the combined forms of the compounds is appropriate when the total dissolved concentration is analyzed.
45
CHAPTER 3
MATERIALS AND METHODS
This chapter contains information about the water sources, materials and methods used in the research. Because a major focus of the research conducted was the development of methods for concentration and isolation of NOM (as opposed to the use
and testing of various methods), it is difficult to segregate the experimental methods used in much of the research from the results. This is especially true of the work conducted at the USGS, but is also true to a lesser extent of the work conducted at the other participating laboratories. Therefore, in general, presentation of the methods that were developed as part of the research is combined with the presentation of results in Chapter 4. The material presented here is limited to information about more routine aspects of sample collection, preliminary characterization, and analytical methods.
SAMPLE COLLECTION
Three rivers and one water supply reservoir in the U.S. and five rivers in Europe were studied in this project. Two of the U.S. rivers with very different water quality and NOM characteristics were selected for fractionation case studies: the Suwannee River in southeastern Georgia and the South Platte River in Colorado. The Suwannee is a very soft, “black water” river with low salt content. It contains a high concentration of NOM that is derived principally from terrestrial plants and that has been minimally fractionated by sorption onto soil mineral constituents. This NOM has been extensively characterized (Averett et al. 1995) because of its use as a standard NOM by the International Humic Substances Society. The river was sampled at its origin, at the outlet of the Okeefenokee Swamp, on October 18, 1995. The entire sample (453 L) was filtered in the field through two Balston glass-fiber cartridge filters in series (25-µm and 0.3-µm porosity) and was then shipped in 40-L stainless steel milk cans to Denver. The sample was held in refrigerated storage during processing.
46
The South Platte River, which serves as a major source of drinking water for Denver, Colorado, was sampled in Waterton Canyon below Strontia Springs Reservoir on February 7, 1996, when 408 L was collected, and March 27, 1996, when 440 L was collected. The NOM in each sample was fractionated, but the corresponding fractions from the two sampling events were combined prior to analysis. The river was almost completely covered with ice on February 7, and on March 27, the ice cover was mostly gone but the spring runoff had not yet begun. For reasons described below, the river was re-sampled on November 21, 1996. The sampling point was on the South Fork of the river about 15 miles upstream of the previous sampling point, because a forest fire on the North Fork had caused massive quantities of ash to enter Strontia Springs Reservoir. In contrast to the NOM in the Suwannee, a substantial portion of the NOM in the South Platte is generated in the water itself, i.e., it is autochthonous NOM. The water in the South Platte is moderately hard (**JL- typical value?) with moderate salt content (**JL- typical value?), and the NOM content is low. In addition, there are extensive mineral sediments and soils that act as solubility controls on the NOM content. NOM was also obtained and concentrated from the Tolt River and Judy Reservoir in Washington State, in the U.S. Pacific Northwest. The Tolt River, whose basin is in the Cascade mountains, is a major potable water source for the city of Seattle, WA. Judy Reservoir is the main water supply source for the city of Mt. Vernon, WA. Neither water is subject to significant impacts from industry or agriculture. For both water sources, the total dissolved solids are very low. More information on these waters is provided in the Results section. Five samples were collected from surface waters in Europe (the Thames River in England, and the Vienne, Gartempe, and Blavet Rivers in France). The Blavet was sampled on two occasions. Approximately 1,000 L of each water was sampled and was filtered through a 0.45-µm porosity membrane on the same day. It was then stored at 4°C in a refrigerated tank.
47
The Vienne and Gartempe Rivers are both in the primarily agricultural Vienne region of France. The Gartempe River is a tributary of the Vienne. Only small industries discharge into these rivers, except for a pulp and paper mill located ~50 km upstream of the sampling point on the Vienne. The Blavet River was sampled 500 m downstream of the Kerne Uhel Reservoir, near the water treatment plant of Lanrivin (Côte D’Armor, Brittany region). This reservoir is located in a rural area approximately 20 km from the Atlantic ocean. The reservoir is surrounded by pine trees that were planted several years ago to define the protected zone. The Blavet River was sampled in winter (December 1995) and in summer (July 1996). The Thames River was sampled at Bray, about 40 km west of London. At this location, discharges of wastewater effluents may have already significantly impacted the quality of the river. Some general water quality characteristics of the untreated waters are summarized in Table 3.1.
48
Table 3.1 Water quality characteristics of untreated water samples Vienne River
Blavet River
Blavet River
Thames Gartempe Suwannee river river River
South Platte River
Tolt river
Judy Reservoir
10/12/95
12/6/95
7/18/96
1/21/97
2/17/97
10/18/95
11/21/96
9/96, 10/96
6/96, 7/96
Bellefonds
Kerne Uhel
Kerne Uhel
Bray
Saulge
Okeefenokee Swamp
*Waterton Canyon
DOC (mg/L)
4.9
12.0
6.6
3.9
6.4
46.8
3.0
1-2
3-4.5
SUVA254 (L/mg-m)
3.6
5.1
4.8
3.2
4.4
4.6
2.4
2.9
3.3
pH
7.6
7.0
7.8
7.4
7.9
nd**
nd
6.5-7.1
6.6-7.4
Conductivity
125
133
145
740
90
30-60
400
25
50
Alkalinity (mg/L CaCO3)
48.2
35.0
37.0
191
39.0
nd
nd
4-6
6-10
Chloride (mg/L)
20.2
22.0
19.7
77
12.0
nd
nd
0.7-1.0
1.0-3.0
Bromide (µg/L)
60
80
80
nd
nd
nd
nd
< 0.02
< 0.02
Nitrate (mg/L)
9.7
12.0
14.1
36.3
8.2
nd
nd
0.1-0.5
0.5-1.5
Sulfate (mg/L)
6.3
8.6
7.7
nd
7.7
nd
nd
1.0-4.0
3.0-6.0
Calcium (mg/L)
20.8
12.5
12.0
123
11.6
nd
nd
5-8
4.0-6.0
Magnesium (mg/L)
3.4
4.3
4.9
nd
1.2
nd
nd
0.4-0.6
1.40-1.70
Sodium (mg/L)
10.5
15.0
14.5
nd
nd
nd
nd
nd
nd
Potassium (mg/L)
2.3
nd
2.5
nd
nd
nd
nd
nd
nd
TDAA (µg/L C) (µg/L N)
222
342
664
nd
nd
nd
nd
nd
nd
79
133
277
TDCA (µg/L C)
224
194
338
nd
nd
nd
nd
nd
nd
Sampling period Location
Mt. Mt.Verno Vernon, n not Wash. Carnation, Wash.
(µS/cm)
First sample at Waterton Canyon; second sample 15 miles upstream of Waterton Canyon
NOM ISOLATION PROTOCOLS
49
NOM was concentrated and isolated at each of the participating laboratories. Both the European water samples and those from the Pacific Northwest were subjected to two types of isolation protocols: one based on membrane-based processes and one based on adsorption/ elution processes. The membrane processes were often used in conjunction with various approaches for desalting the solution. Adsorption processes that were investigated used XAD-8 and XAD-4 resins in series for the European waters, and either XAD-8 or an oxide-based adsorbent for the Pacific Northwest waters. These procedures are described next.
Membrane-Based NOM Isolation Protocols
European Water Samples
NOM from European waters was concentrated using one RO and one NF membranes. Throughout this report, these membranes are referred to by their brand names, viz., TW30 and NF70, respectively.1 Both membranes are thin-film composites made of a polyamide. Some characteristics of these membranes are provided in Table 3.2.
1
All manufactured by Film Tek Corp., Minneapolis, MN 55439. 50
Table 3.2 Characteristics of the membranes used to process European samples Property Membrane Identifier
Value
Max. Operating Pressure (psi)
CTAB-2- TW30 (RO) NF70 (NF) 10HF (RO) 125 300 250
Max. Feed Flow Rate (gpm)
0.021
17
16
pH range, Continuous
3 to 8
2 to 11
3 to 9
1 to 12
1 to 11
pH range, Cleaning (30 min) Max. Operating Temp. (°C)
35
45
35
Max. Feed Turbidity (NTU)
5
1
1
5
5
50% efficiency. As a practical matter, it is not possible to sorb and thereby desalt a significant amount of NOM molecules with k' < 5 efficiently in this type of column setup. Retention of HPI NOM by XAD-4 resin in the first processing cycle is also limited by competition with more strongly binding transphilic NOM molecules for adsorption sites. Ultra-hydrophilic NOM was fractionated and isolated as shown in the flow chart of Figure 4.3. The fractionation and recovery data for dissolved NOM from the Suwannee River are shown in Table 4.2.
78
Vacuum-evaporate XAD-8/ XAD-4 effluent (Figure 4.1) at pH 4 to 1.0 L Acidify to pH 1 with HCl, centrifuge solution
solids Disperse solids in 0.1 N NaOH, centrifuge, and discard particulates; Solids that dissolve are humic acids.
XAD-4 k' =5
Acidify with HCl to pH 1 to precipitate humic acid, centrifuge, wash precipitate, and freeze-dry to isolate ultrahydrophilic humic acid
Combine extracts and remove acetonitrile by evaporation
80 mL MSC1H cationexchange
5. Desorb with 1.0 N NaOH
100 mL Duolite A-7 anion exchange
6. Desorb with 1.0 N NaOH
1. Pass sample through 80 mL column and rinse with 200 mL 0.1 N HCl 2. Desorb with 75% acetonitrile/ 25% water
Evaporate to 85% of beginning sample volume, filter, wash, and discard salts Repeat steps six times until sample to column volume ratio is equivalent to k´ = 5 Acidify with HCl to pH 1
Column eluent from step 6* MSC-1H cation exchange
7. Desorb with 75% acetonitrile/ 25% water Freeze dry to isolate hydrophilic acids
XAD-4 k´ = 5
Discard salts MSC-1H cationexchange
Column removes bleed from Duolite A-7 resin
8. Remove acetonitrile by evaporation, freeze-dry to isolate hydrophilic bases
Non-sorbed material to processing as ultrahydrophilic NOM
Freeze-dry hydrophilic neutrals*
* The mass of hydrophilic NOM slightly exceeded the capacity of the Duolite A-7 resin, so the column effluent was recycled through regenerated ion-exchange resins, and a hydrophilic acid 2 fraction was isolated before the isolation of the hydrophilic neutral fraction.
Figure 4.2. Flow chart for fractionation and isolation of hydrophilic NOM (k' = 5-100) from the Suwannee River
79
Adjust pH to 2.0 of ultra-hydrophilic NOM concentrate Remove majority of remaining salts by zeotrophic distillation procedure (Aiken and Leenheer, 1993) in which water from the sample is evaporated from glacial acetic acid; remove inorganic salts by filtration; ultra-hydrophilic NOM remains in solution Remove acetic acid by evaporation to dryness 80 mL MSC-1H cation exchange resin 100 mL Duolite A-7 anion exchange resin 20 mL MSC-1H cation exchange resin
Neutralize with HCl to pH 7.0 Desorb with 1.0 N NaOH
Evaporate to dryness
Column is used to remove bleed from Duolite A-7 resin
Take effluent to dryness and evaporate with methanol to remove boric acid.
Methylate dry residue with 100 mL of 10% acetyl chloride in methanol Evaporate to dryness; Dissolve residue in 50 mL H2O
Re-dissolve residue in 10 mL 0.01 N HCl to hydrolyze silicateorganic esters; filter out silica gel 100 mL XAD-8 resin column
Freeze-dry to isolate ultrahydrophilic neutrals
Elute w/100% acetonitrile
Discard salts in rinse Evaporate acetonitrile eluent from XAD-8 resin to isolate methylated ultrahydrophilic acids
Figure 4.3. Flow chart for fractionation and isolation of ultra-hydrophilic NOM fractions (k' < 5) from the Suwannee River
80
Table 4.2 Yields and recovery of dissolved natural organic matter (NOM) fractions from the Suwannee River Fraction
Hydrophobic Neutrals
Mass recovered
Percent C
DOC recovery
(mg)
in fraction
efficiency (%)
26,421
---
59.92
357
55.8
0.94
Acids
25,969
48.0
58.79
Bases
95
Transphilic
I * (43)
I * (0.19)
6,008
--
12.20
Neutrals
2,523
47.0
5.59
Acids
3,485
40.2
6.61
2,649
--
5.25
90
39.2
0.17
2,351
43.0
4.77
Acids2
160
35.6
0.27
Bases
48
19.5
0.04
--
2.46
Hydrophilic Neutrals Acids
Ultra-hydrophilic
1,140
Neutrals
107
31.8
0.16
Acids
101
45.3
0.22
932
47.4
2.08
36,218
--
79.83
3,077
--
6.86
Acids
32,998
--
72.74
Bases
143
--
0.23
--
--
20.17
(methylated) Humic acid Total Neutrals
Loss •
I = Incomplete data (data in parentheses are estimates).
81
The DOC concentration of the unfractionated sample was 46.8 mg/L, 90% of which adsorbed on the initial passes through XAD-8 and XAD-4 columns in series. However, only about 72% of this DOC was recovered by elution of these columns (the sum of the hydrophobic and transphilic recoveries given in Table 4.2), corresponding to a loss of 18% of the DOC that was in the original sample. Most of this loss probably occurred when a freeze-drying flask containing some of the hydrophobic fraction broke. Of the 4.6 mg/L of the DOC that did not adsorb in these columns, about 80% was recovered by subsequent processing, i.e., in the hydrophilic and ultra-hydrophilic fractions, meaning that an additional 2% of the original DOC was lost in these steps. Therefore, the vast majority (about 90%) of the NOM that was not recovered in any of the fractions was lost in the processing of the hydrophobic fraction. After the elution steps, the color of the XAD-8 resins was similar to that of fresh resins, suggesting that relatively little NOM remained on the resin. Therefore, it is likely that the breakage of the flask noted above was responsible for most of the DOC loss, and the DOC recovered by the fractionation is presumed to be representative of the total DOC distribution. The DOC fractionation data in Table 4.2 show that NOM from the Suwannee River is predominantly hydrophobic and acidic material. Base fractions are almost non-existent. This finding is consistent with other studies of DOC from the Suwannee River (Malcolm et al. 1994**JL this ref not in ref list, Thurman 1985); however, the DOC fractionation of the Suwannee River is not typical of that from other aquatic systems (Thurman 1985, Leenheer 1994**JL this ref not in ref list).
South Platte River
The hydrophobic and transphilic fractions of the NOM in the water samples from the South Platte River were isolated using the same procedures as were used for the Suwannee, except that the total sample volume was different (848 L) and the evaporative concentration step applied to solution that passed through the XAD-8 and XAD-4 columns in series was stopped when the volume had been reduced to 3.1 L. Additionally, no hydrophobic base fraction was isolated from the sample from the South Platte. 82
The process used to isolate the ultra-hydrophilic NOM fractions was somewhat different from that used for the Suwannee sample and is shown schematically in Figure 4.4. The key differences in the procedures were as follows: (1) No ultra-hydrophilic humic acid fraction was isolated from the South Platte, because a large amount of silica gel co-precipitated with this fraction during vacuum evaporation (the first step in the process; see Figure 4.2). (2) Two new procedures were tested to isolate ultra-hydrophilic acids from the South Platte: (a) co-precipitation with calcium and magnesium hydroxides, and (b) adsorption on alumina with subsequent removal of phosphate and sulfate by selective precipitation. These new procedures for isolating the ultra-hydrophilic acid fraction were devised to avoid changing the chemical characteristics of this fraction by methylation, which was part of the isolation procedure for the uHPIA fraction from the Suwannee River. NOM fractionation and recovery data for the South Platte River sample are summarized in Table 4.3.
83
Adjust ultra-hydrophilic NOM concentrate to pH = 2.0 Remove majority of remaining salts by zeotrophic distillation procedure (Aiken and Leenheer 1993) in which water from the sample is evaporated from glacial acetic acid; remove inorganic salts are by filtration; ultra-hydrophilic NOM remains in solution Remove acetic acid by evaporation to dryness Dissolve residue in water, adjust to pH 12 with NaOH, remove solids (Ca(OH)2 and Mg(OH)2) by centrifugation.
Solids
Dissolve in dilute H2SO4
Solution 80-mL column MSC-1H cationexchange resin
Desorb with 1.0 N NaOH
100-mL column Duolite A-7 anionexchange resin
20-mL column MSC-1H cationexchange resin
Take effluent to dryness and evaporate with methanol to remove boric acid Re-dissolve residue in 10 mL 0.01 N HCl to hydrolyze silicateorganic esters; filter out silica gel
Neutralize with HCl to pH 6.0
Zeotrophic distillation with acetic acid to precipitate CaSO4 and MgSO4
Contact with 30 g Al2O3 at pH 6 for 12 hours
Centrifuge to collect Al2O3; discard supernatant; contact Al2O3 with pH 12 solution prepared with LiOH for 1 hour
Evaporate to reduce volume by 90%; filter to remove Li3PO4 Adjust to pH 1 with formic acid, add BaCl2 to precipitate BaSO4
100 mL MSC-1H resin
20 mL MSC-1H resin
(Removes metals)
Evaporate and freezedry to isolate co-precipitated ultrahydrophilic acids
(Removes Al3+ Ba2+, Li+) Vacuum evaporate and freeze-dry to isolate ultrahydrophilic acids
Freeze-dry to isolate ultrahydrophilic neutrals
84
Figure 4.4. Flow chart for fractionation and isolation of ultra-hydrophilic NOM fractions from the South Platte River (**MB keep w/fig) Table 4.3 Yields and recovery of dissolved natural organic matter (NOM) fractions from the South Platte River Fraction
Hydrophobic Neutrals Acids Transphilic
Mass recovered
Percent C
DOC recovery
(mg)
in fraction
efficiency (%)
1,562
---
34.01
110
58.8
2.93
1,452
47.2
31.08
1,450
--
26.02
Neutrals
538
48.0
11.71
Acids
912
34.6
14.31
339
--
6.28
52
39.3
0.93
Acids
274
41.0
5.10
Bases
13
Hydrophilic Neutrals
Ultra-hydrophilic
I* (43)
61
--
0.25 1.11
Neutrals
25
I(40)
0.45
Acids
12
I(40)
0.22
24
I(40)
0.44
(coprecipitated) Acids (alumina) Total
3,412
--
67.42
725
--
16.02
Acids
2,674
--
51.15
Bases
13
--
0.25
--
--
32.58
Neutrals
Loss
* I = Incomplete data, data in parentheses are estimates.
85
The dissolved organic carbon concentration (DOC) of the combined sample was 2.6 mg/L, of which an average of 1.1 mg/L was not adsorbed on the initial passage though the XAD-8 and XAD-4 columns in series. Therefore, this column series adsorbed 57.7% of the DOC, essentially all of which was recovered (computed recovery of 104%). Of the 1.1 mg/L of DOC that did not adsorb to these columns, only 18% was recovered by further processing, so it appears that most of the DOC loss was from the hydrophilic and ultra-hydrophilic NOM fractions. Possible reasons for this loss include the failure to recover colloidal ultra-hydrophilic humic acids, failure to recover NOM that co-precipitated with silica during vacuum evaporation, and/or failure to recover the hydrophobic bases. Because ultrafiltration was not used in the fractionation scheme, it is postulated that much of the unrecovered ultra-hydrophilic humic acid fraction consisted of organic colloids, as was found previously for the Mississippi River and its major tributaries (Leenheer et al. 1995a, 1995b). To further investigate the causes for the loss of hydrophilic NOM, the South Platte River was re-sampled on November 21, 1996, at which time 505 L was collected. The sample was filtered as before, but a 1.0-µm glass fiber filter was used instead of the 0.3-µm filter because the low-porosity filter was no longer commercially available. A four-column resin sorption scheme was devised that did not use vacuum evaporation, thereby avoiding the problem of silica precipitation (Leenheer 1996**JL this ref not in ref list). A flow chart of this NOM fractionation scheme is shown in Figure 4.5.
86
Filtered water sample
Hydrophobic neutral fraction
1 L XAD-8 resin
Base fraction
1 L MSC-1H hydrogen form cation exchange resin
Hydrophobic acid, hydrophilic acid, and ultra-hydrophilic acid fractions
0.5 L AG-MP-1 anion exchange resin in borate form
Hydrophilic neutral fraction
Water to waste
Figure 4.5. Four-column preparative NOM fractionation scheme The hydrophobic neutral fraction was isolated as in the previous samples (Steps 2 and 7 in Figure 4.1). Organic bases were eluted from the MSC-1 resin with a combination of 2 N sodium formate and 2 N formic acid, at pH 3.5. A four-fold excess of sodium to cation-exchange equivalents was used to maximize the elution efficiency. This combination of sodium formate and formic acid was found to be especially effective in eluting highly retained cations such as iron and aluminum that interact with organic amino acids when high pH solutions are used to elute the resin. The column eluent was acidified to pH 1, most of the water and formic acid were removed by vacuum evaporation, and the sample was evaporated to a salt slurry. The precipitated salt (presumably mostly sodium chloride) was filtered in a funnel with a glass-wool plug and was leached with a minimum volume of water until free of color. The bases in the 87
leachate were isolated on XAD-4 resin as in Step 7 of Figure 4.3. The acid fractions were eluted from the Duolite A-7 column with 1.0 N NaOH, and hydrophobic, hydrophilic, and ultra-hydrophilic NOM fractions were isolated from this anion concentrate using the procedures detailed in Figures 4.1, 4.2 and 4.4. The fourth column was added to adsorb carbohydrates by borate complexation with polyalcohol groups. Khym and Zill (1952) studied the retention of a number of simple carbohydrates on a borate-form, strong-base anion exchange resin. They reported that sucrose, the least retained sugar, had a k' of 19 (i.e., it was 50% retained when the system was operated with k' = 19). Most of the other carbohydrates studied had k' values >100 when dissolved in deionized water. These k' values are comparable with the operational k' values used previously for the three-column system for NOM fractionation (Leenheer and Noyes 1984). The hydrophilic neutral fraction was eluted from the fourth column with 20% acetic acid. In this step, both boric acid (formed by protonation of the borate ions that were used to pre-saturate the anion exchange sites) and silicic acid, which adsorbs on the resin from the sample, are co-eluted with the NOM. The water and acetic acid were evaporated, and boric acid was removed as volatile trimethyl borate by evaporation with methanol. The hydrophilic neutrals were solubilized from the silica residue by repeated extractions with 0.01 N HCl, which hydrolyzes the silicate esters. The hydrophilic neutral fraction was recovered after evaporation of water and HCl. The water sample was divided in half and was passed through the four-column system of Figure 4.5 in two portions because the conductivity of the sample (400 µmho/cm) indicated that the ion-exchange capacity would be exceeded if all 505 L was passed through in one portion. The DOC concentration for this sample was 3.0 mg/L. The NOM fractionation and recovery data for the sample are shown in Table 4.4.
88
Table 4.4 Yields and recovery of NOM fractions from the second South Platte River sample Fraction
Hydrophobic neutrals acids Bases Hydrophilic
Mass recovered
Percent C
DOC recovery
(mg)
in fraction *
efficiency (%)
1,082
---
33.94
30
58.8
1.16
1,052
47.2
32.78
85
I † (40)
2.24
402
--
10.82
60
39.3
1.56
342
41.0
9.26
28
40
0.74
1,597
--
47.74
90
--
2.72
acids
1,422
--
43.94
bases
85
--
2.24
--
--
52.26
neutrals acids Ultra-hydrophilic acids Total neutrals
Loss
* Carbon percentages are based on the values of similar fractions from Table 4.3. † I = Incomplete data, data in parentheses are estimates.
The NOM loss during processing of the second sample was greater than that for the first. Since the hydrophobic DOC percentages were very similar for both samples, the NOM loss for the second sample is again probably hydrophilic NOM. Two of the three previously hypothesized reasons for NOM loss were discounted by this second experiment. The base fractions represented only a small percentage of the DOC, so unrecovered bases cannot account for the NOM losses. Further, the silica remaining after extraction of the hydrophilic NOM was destroyed with dilute HF, and the fluorosilicates were removed from the solution by anion exchange. No appreciable hydrophilic NOM was recovered after silica was removed. This finding effectively rules out complexation by soluble silica as a possible sink for the unrecovered NOM. 89
Therefore, the most likely explanation for the loss of hydrophilic NOM is that it was present as colloids that passed through the entire four-column resin adsorption system. The percentage of the NOM that was colloidal in the second sample might have been greater than that in the first sample because of the use of a larger porosity glassfiber filter to pre-filter the second sample. In a previous study of NOM in the Mississippi River, Leenheer et al. (1995a,b) found that colloidal organic carbon concentrations in the main stem of the river and the major tributaries were usually between 0.5 and 1.0 mg/L, while the DOC varied between 3 and 15 mg/L. Electron microscopy showed that most of this colloidal organic carbon was bacterial cells. If colloidal organic carbon concentrations in the South Platte River are in this range and if the colloids all passed through the resin columns, they could account for 30 to 60 percent of the NOM loss. The remaining loss is probably mechanical, resulting from separating large amounts of salt from very small amounts of NOM. Future comprehensive NOM isolation studies should incorporate an ultrafiltration step to quantify and separate the colloidal NOM component for recovery calculations.
Comparison of NOM Recovery from the Suwannee and S. Platte Rivers
A comparison of the DOC recoveries for various NOM fractions for the Suwannee River and South Platte River (first sampling) is shown in Figure 4.6.
90
60
Percent of DOC
50 Suwannee South Platte
40 30 20 10 0 Hydrophobic
Transphilic
Hydrophilic
Ultrahydrophilic
Figure 4.6. Comparison of NOM fractionation in the Suwannee and South Platte Rivers **MB add column ‘unaccounted for’ The figure illustrates dramatically the result shown in Tables 4.2, 4.3, and 4.4 that the bulk of the NOM is contained in the hydrophobic and transphilic fractions. These fractions can be isolated with a modest degree of effort (Figure 4.2). The hydrophilic and ultra-hydrophilic fractions constitute less than 10% of the recovered DOC, and they can be isolated only with a substantial amount of effort (Figures 4.3, 4.4, and 4.5). The South Platte River NOM is significantly more hydrophilic than the Suwannee River NOM, and the base and neutral fractions represent a greater percentage of the NOM in the South Platte than in the Suwannee. This difference in NOM fractionation patterns suggests that the NOM is less degraded (humified) in the South Platte. Such a difference might be related to the winter sampling of the South Platte when microbial degradation rates are slow. Further discussion of these differences is provided after the results from some of the other characterization techniques are presented.
91
CASE STUDIES FOCUSING ON MAXIMAL NOM RECOVERY WITHOUT FRACTIONATION
Membrane Processes for NOM Recovery from European Waters
The efficiency with which NOM could be collected and concentrated by RO and NF was compared directly using four surface waters. Most of the experiments were conducted using a concentration factor (initial volume/final volume) near 10. The results obtained using different membranes are summarized in Tables 4.5 and Table 4.6. Table 4.5 Volume and DOC content of the solutions collected from reverse osmosis and nanofiltration processing of five surface waters Water
Membrane
Raw Water
Concentrate
Permeate
NaOH elution
DOC Volume DOC Volume DOC Volume DOC Volume (mg/L)
(L)
(mg/L)
(L)
(mg/L)
(L)
(mg/L)
L
Gartempe
TW-30 2514
6.4
10
55.5
1
0.6
9
nd *
nd
River
TW-30 4040
6.4
740
39.0
110
0.2
630
10.9
40
Thames River TW-30 40401
3.9
200
17.2
32.5
0.4
167
4.3
38
TW-30 40402
3.9
95
8.7
35
0.3
60
1.7
40
Vienne River TW-30 4040
4.9
345
37.4
30
0.4
315
10.5
40
NF-70 4040
4.9
340
36.2
28
1.2
312
8.8
40
Blavet River TW-30 4040
12.0
400
160
27.5
0.2
372
16.3
40
NF-70 4040
12.0
310
105
28.0
1.0
281
17.3
41
Blavet River TW-30 4040
6.6
200
41
28
0.3
170
nd
nd
6.6
200
37
27
0.2
170
nd
nd
(Winter)
(Summer)
NF-70 4040
* nd : not determined
92
Table 4.6 Isolation of NOM: Efficiency of RO and NF membranes Water
Membrane
Rejection
Concentrate
Permeate
NaOH Elution
Total
DOC Rejection or Recovery (%) Gartempe
TW-30 2514
92
87
8
-
-
River
TW-30 4040
97
90
3
9
102
Thames River TW-30 4040
91
72
9
21
102
TW-30-4040
95
81
5
19
105
Vienne River TW-30 4040
93
65
7
24
96
NF-70 4040
77
60
23
21
104
Blavet River
TW-30 4040
98
91
2
13
106
(Winter)
NF-70 4040
92
79
8
19
106
Blavet River
TW-30 4040
97
87
3
-
-
(Summer)
NF-70 4040
96
76
4
-
-
For all waters, the DOC rejection efficiency ranged from 91 to 98% using RO and from 77 to 96% using NF. Overall DOC recoveries were generally close to 100%, but the distribution of the DOC among the concentrate, permeate, and the membrane surface depended on the water source and the type of membrane used. In the only tests in which the different sizes of units were compared (using the 2514 and 4040 size membranes to process Gartempe water), they gave similar results. The effect of raw water quality on the performance of these systems can be assessed by comparing the results for the five samples obtained using the TW-30 4040 RO membrane or the four samples concentrated using the NF-70 membrane. The DOC recovery in the RO concentrate was similar (87% to 90%) for the Gartempe river and the two samples from the Blavet River. However, for the Vienne and Thames rivers, the NOM recovery in the concentrate was much lower (65% to 81%), and the recovery in the NaOH rinse was correspondingly higher. The DOC concentration in the source water does not seem to be a critical factor controlling the recovery efficiency in the concentrate, since the Gartempe river and the Blavet River (winter sample) yielded similar DOC recoveries despite having a large difference in DOC contents (6.4 and 12.0 mg C/L, 93
respectively). Sun et al. (1995) also obtained good recoveries by RO regardless of the DOC concentration in the source water. The nature of the NOM, and in particular its aromatic character, might be at least partially responsible for the different behavior of the different waters. The SUVA254 value was significantly lower for the Vienne and Thames rivers (3.6 and 3.2 L/mg-m) than for the Gartempe and Blavet Rivers (4.4 to 5.1 L/mg-m), suggesting that hydrophilic NOM was preferentially retained in or on the membrane. The result might also be related to the salt content of the source water.
Salt Concentration and Recovery
When water samples are concentrated by RO or NF, inorganic species are concentrated along with NOM. Because inorganic species may influence NOM properties and can interfere with some analytical characterizations of the NOM, attention was paid to monitoring and controlling the major inorganic ions in the samples. Only anions were taken into account, because most inorganic cations were exchanged for Na+ in the cation exchange resin prior to membrane filtration. Extensive analysis of anions was conducted only on samples from the Vienne and Blavet Rivers. The results of these analyses are presented in Tables 4.7 and 4.8.
94
Table 4.7 Concentration of anions in permeate and concentrate produced by reverse osmosis and nanofiltration Raw water Vienne River
Vol
L
(1) * HCO 3− mg/L
Concentrate
Permeate
Ion Rejection (%)
TW30
NF70
TW30
NF70
TW30
NF70
-
30
28
315
312
-
-
48.2
315
237
2.6
11.6
95
76
Cl−
mg/L
20.2
57.0
36.0
0.6
3.7
97
82
Br−
mg/L
0.06
0.16
0.07
60
NO 3−
mg/L
9.7
23.7
2.5
0.7
2.9
93
70
SO 2− 4
mg/L
6.3
41.5
58
0.1
0.2
98
98
Blavet River
Vol
L
-
27.5
28
372
281
-
-
(Winter)
HCO 3−
mg/L
35.0
220
170
9.0
11.0
74
69
Cl−
mg/L
22
190
120
1.1
4.4
95
80
Br−
mg/L
0.08
0.58
0.4
60
NO 3−
mg/L
12.0
90
38.5
1.8
6.9
85
42.5
SO 2− 4
mg/L
8.6
79
68
98
Blavet River
Vol
L
28
27
170
170
-
-
(Summer)
HCO 3−
mg/L
37.0
170
160
4.4
4.8
88
87
Cl−
mg/L
19.7
96
70
0.3
1.7
98
91
mg/L
0.08
0.5
0.3
60
NO 3−
mg/L
14.1
56
37
0.8
3
94
79
SO 2− 4
mg/L
7.7
51
38
99
Br
−
* as CaCO3
95
Table 4.8 Anion recoveries for reverse osmosis and nanofiltration Concentrate recovery
Permeate recovery
Total recovery
(%)
(%)
(%)
TW30
NF 70
TW30
NF70
TW30
NF70
CF(1)
11.5
12.1
-
-
-
-
HCO 3−
57
40
5
22
63
62
Cl−
27
15
3
17
30
32
23
10
-
-
-
-
NO 3−
21
2
7
28
28
30
SO 2− 4
57
75
2
3
59
78
Blavet River
CF *
14.5
11.0
-
-
-
-
(Winter)
HCO 3−
43
44
24
28
67
72
Cl−
59
49
5
18
64
67
50
45
-
-
-
-
NO 3−
52
29
14
52
66
81
SO 2− 4
63
71
-
-
-
-
Blavet River
CF(1)
7.1
7.3
(Summer)
HCO 3−
64
58
10
11
74
69
Cl−
68
48
1.5
7
69.5
55
Br−
87
51
-
-
-
-
NO 3−
56
35
5
18
61
53
SO 2− 4
93
67
1
1
94
68
Vienne River
Br
Br
−
−
* CF : Concentration factor (initial volume/final volume)
Both RO and NF concentrated the inorganic salts significantly, with RO being more efficient than NF. Chloride and sulfate were the most efficiently rejected anions, with RO rejection coefficients ranging from 95 to 98% and from 98% to >99%, respectively. For NF, the corresponding values were from 80 to 91% and from 98 to >99%. Rejection coefficients for bicarbonates and nitrates were slightly lower.
96
Because of the low concentration of bromide in natural waters and because of the relatively high detection limit (35 µg/L), the Br− rejection efficiency could not be determined exactly. However, a lower limit of 60% could be estimated for both types of membranes. The actual value is probably similar to that for chloride. Overall, anion recoveries for the concentrated waters are higher for the Blavet River sampled in summer (65 to 90%) than in winter (40 to 60%) and are also higher for the Blavet than for the Vienne River (20 to 60%). No explanation for this finding is apparent since the three waters had similar ionic balances. The most significant result of this part of the work is that total recoveries in the concentrate plus permeate never even approached 100% for most of the inorganic species, regardless of the water source. With few exceptions, 30 to 40% (by weight) of the salt content was not recovered in the combined concentrate and permeate. To test for analytical accuracy, the chloride concentration in several samples was determined by addition of AgNO3 and gravimetric analysis of the precipitated AgCl(s). The results confirmed those obtained by ion chromatography. The most reasonable explanation for this finding is adsorption of inorganic species onto the polyamide membrane or precipitation of such species in the membrane module. An additional experiment was conducted on a synthetic solution containing NaHCO3, NaCl, NaNO3, and Na2SO4 plus the XAD-8/XAD-4 mixture of NOM isolated from the Blavet River winter sample. These constituents were all solubilized into MilliQ water to obtain the following concentrations: − Cl− = 21.2 mg/L; NO 3− = 11 mg/L; SO 2− 4 = 43 mg/L; HCO 3 = 7.1 mg/L
pH=5.5 or 8; DOC = 5.3 mg/L Ten liters of this solution was concentrated in the lab-scale RO unit equipped with the TW-30 RO membrane. The concentrate was recirculated into the feed tank until the volume of the concentrate was reduced to 1 L, i.e., until the concentration factor was 10.
97
Table 4.9 gives the recovery efficiencies for the anions in this system, based on ion chromatographic analyses. Table 4.9 Recovery of anions in the concentrated water produced by RO treatment of a synthetic solution Recovery Efficiency (%) Chloride
Nitrate
Sulfate
Carbonate species
pH 5.5
56
72
64
59
pH 8
55
52
54
62
The concentration of anions in the permeate was near the detection limit, which was approximately 1 mg/L for all species. However, the overall recoveries did not exceed 70% for any anion, which supports the hypothesis that some of the anions were retained on or in the polyamide membrane.
Use of XAD-8 /XAD-4 for NOM Recovery from European Waters
XAD-8 and XAD-4 resins were used to collect NOM from the Vienne River and from winter and summer samples from the Blavet River. In each case, approximately 300 L of water was processed in two columns in series containing 10 L of XAD-8 and XAD-4 resin, respectively. The size of these columns allowed the total volume of water to be processed in a single run at a k' of 50. One hundred mL of effluent was collected from each resin column for every 20 L of water processed. The average DOC in the effluent from the whole run is presented in Table 4.10, and the data are presented in various alternative forms in Table 4.11.
98
Table 4.10 DOC content of the XAD-8 and XAD-4 permeates of the three water sources Raw water
XAD-8 Effluent *
XAD-4 Effluent *
Volume (L)
DOC (mg/L)
DOC (mg/L)
DOC (mg/L)
Vienne River
270
4.9
2.0
1.1
Blavet River (winter)
300
12.0
2.5
1.25
Blavet River (summer)
290
6.6
2.8
1.7
* blank subtracted, 0.1 to 0.2 mg DOC/L for XAD-8, 0.2 to 0.3 mg DOC/L for XAD-4
Table 4.11 XAD-8/XAD-4 DOC distribution and DOC recoveries DOC Adsorption (%)
* DOC Recovery (%)
XAD-8
XAD-4
Total
XAD-8
XAD-4
Total
Vienne River
59
18
77
45
13.5
58
Blavet River (winter)
79
10.5
90
55
6
61
Blavet River (summer)
57
21
78
56
17
73
* after column rinse, NaOH elution, and ion exchange.
The DOC adsorption efficiency indicated in Table 4.11 can be appropriately compared with the rejection coefficient for membrane processes, whereas the total DOC recovery efficiency (the sum of the DOC recoveries of the XAD-8 and XAD-4 resins) might appropriately be compared with either the concentrate recovery efficiency or the total recovery efficiency for membrane processes, depending on the intended use of the comparison. For the three surface waters, 77 to 90% of the DOC adsorbed onto the two resins and 58 to 73% of the raw water DOC was recovered in the eluents. As was the case for membrane processes, the highest NOM recovery efficiency was obtained for the Blavet River sampled in the winter, which is also the sample that had the largest SUVA.
99
Combination of RO with XAD Resin for NOM Recovery from European Waters
NOM that had been concentrated from the Blavet River winter sample and the Thames river samples using the larger scale RO unit (TW 4040) was subjected to a second RO concentration step using the smaller RO unit (TW 2514) and was then desalted using XAD resins. Highly concentrated NOM solutions were produced in the second stage of RO. DOC recoveries were similar to those obtained in the first RO concentration step. Any loss of organic matter can be attributed to adsorption or precipitation onto the membrane, since the DOC content of the permeate was always negligible. Extrapolating these results, at least for the conditions tested here, the DOC recovery from a series of RO concentration steps can be roughly estimated as (0.90)n, where n is the number of concentration steps. (This estimate assumes that adsorbed and/or precipitated NOM is not recovered by application of an NaOH rinse.) After acidification to pH 2, the concentrated NOM solutions were filtered through XAD-8 or XAD-4 resin using low k' values in order to optimize NOM retention. Typical breakthrough curves for DOC in the column effluent are shown in Figure 4.7.
100
60 Volume of resin : 250 mL
DOC, mg /L
50
DOCin = 250 mg/L
40 30 20 10 0 0.0
0.5
1.0
1.5
2.0
2.5
DOC, mg/L
Vol. H2O/ Vol. XAD-8, L/L
20 18 16 14 12 10 8 6 4 2 0
Volume of resin : 730 mL DOCin = 272 mg/L
0.0
0.5
1.0
1.5
2.0
Vol. H2O/ Vol. XAD-8, L/L
Figure 4.7. Breakthrough curves for treatment of concentrated NOM solutions with XAD-4 resin
101
The experiments confirmed the large adsorption capacity of both XAD-8 and XAD-4 resins for aquatic NOM (Table 4.12). NOM recovery efficiencies were 87% or higher when k′ was less than 10. As expected based on its greater hydrophilic character, the recovery efficiency for NOM from the Thames river was lower than that from the Blavet. Table 4.12 Purification of RO-concentrated water: RO and XAD resin desalting in series Water source
Procedure
XAD desalting
Membrane concentration
§ Total recovery
(RO TW 2514)
* DOCf (mg/L)
Blavet River
(winter)
Thames River (run #2) * Final
RO + XAD-4 + dilution RO + XAD-4 RO + XAD-8 RO + XAD-4
† DOC Resin Recovery volume (%) (mL)
k'
DOC Recovery efficiency (%)
‡ DOC Isolation yield (%)
DOC %
120
89
250
9
94
74
61
250
89
730
1.7
97
83
69
250
89
250
3.2
92
80
66
52
80
1,000
6.5
87
70
46
DOC content after the second stage of reverse osmosis.
† DOC recovery from the second stage of RO concentration. ‡ DOC recovery calculated based on the elemental analysis of the NOM fraction. § Includes both RO steps.
Following the adsorption step, the columns were rinsed with formic acid to remove salt from the column void volume before the organics were eluted with a mixed acetonitrile (75%), MilliQ water (25%) solution. After rotary evaporation of the majority of the acetonitrile, the eluate was lyophilized. During the lyophilization step, residual formic acid and acetonitrile are removed. As discussed in the following chapter, the ash content of the isolated fractions was low and confirmed the efficiency of this desalting technique. However, the DOC recovery efficiencies ranged from only 70 to 83%, corresponding to 12 to 20% loss of NOM during the desalting procedure. No significant difference in this respect was observed between the XAD-8 resin and the XAD-4 resins. 102
The last column of Table 4.12 shows the total NOM recovery efficiency in these tests. The difference between this value and 100% yields the combined loss of NOM in the two RO concentration steps and the XAD resin desalting step. For the Blavet, total recoveries for the three experiments were similar, ranging from 61 to 69%. These values are close to the DOC recovery achieved by processing the whole water by the XAD-8/ XAD-4 protocol alone. For the Thames river, the total recovery was only 46%. The reason for this poor recovery is unclear. Since the Thames river NOM was probably more hydrophilic and probably had a lower average molecular weight than that from the Blavet, it is possible that some volatile organics were lost during the lyophilization. It is also possible that colloidal NOM was not recovered, including colloids that might have formed in the RO concentrate because of its high concentration of salts.
NOM Concentration and Isolation of Judy Reservoir and Tolt River Water By Reverse Osmosis
The performance of RO for NOM recovery was also tested with water from Judy Reservoir and the Tolt river. The key results can be summarized as follows: •
the DOC concentration in the permeate was in the range 0.2 to 0.5 mg/L for both waters, comparable to the results for the European water sources;
•
the UV absorbance and fluorescence of the permeate were negligible;
•
NOM was prone to become trapped in the RO cartridge during each round of concentration. Most of the trapped NOM could be eluted with deionized water, as opposed to the experience with the European waters, in which case more elaborate techniques were required. The NOM rejection efficiency by RO ( η RO DOC ) can be calculated based on the DOC
concentrations in the raw water and permeate using Equation 4.1:
103
η RO DOC =
( R split + 1) ⋅ DOC raw water − R split ⋅ DOC permeate ( R split + 1) ⋅ DOC raw water
(4.1)
where Rsplit is the RO membrane splitting ratio (Vpermeate/Vretentate). η RO DOC is also the maximum possible NOM recovery efficiency using RO, for a given water and operating conditions; the actual recovery efficiency is typically less than η RO DOC , because some NOM remains bound to the membrane, rather than being recovered in the concentrate. For the operational conditions used, η RO DOC was 87% and 92% for the Tolt River and Judy Reservoir, respectively. The negligible UV absorbance and fluorescence of the permeates signifies that humic species were essentially completely rejected or retained by the RO membrane, so the NOM in the permeate must have been low molecular weight and very low aromaticity fractions of NOM. For typical operational conditions, calculations based on Equation 4.1 show that the DOC retention efficiency by RO is not highly sensitive to the DOC of the permeate or the splitting ratio. For instance, for permeate DOC concentrations from 0.3 to 0.5 mg/L and splitting ratios from 2 to 7, η RO DOC is between 75 and 89% for the Tolt River and between 88 and 95% for Judy Reservoir. Thus, as is commonly observed, high NOM recoveries are relatively easy to achieve using RO. However, the amount of concentrate generated and, correspondingly, the number of RO cycles required to achieve a given concentration factor, do depend strongly on the splitting ratio selected. The retention of Ca, Mg, and other inorganic salts by RO is potentially the biggest problem associated with the use of RO to concentrate NOM from natural waters. These salts can interfere by forming inorganic or mixed organic-inorganic solids in the RO unit, potentially causing irreversible blockage of the pores and/or retention of NOM. Off-line removal of calcium and magnesium by cation exchange (as was done in RO experiments with Judy Reservoir and Tolt River waters) overcomes these problems, but at the risk of simultaneously losing the basic fractions of the NOM. The performance of RO is compared below with that of a few sorption-based techniques.
104
Comparison of NOM Concentration and Isolation Using IOCS, IOCO, and XAD Resins for Judy Reservoir and Tolt River Water
Retention of NOM from the Pacific Northwest waters by IOCS and IOCO was compared with that by XAD-8 resin and RO. Typical DOC and A254 breakthrough curves from columns packed with IOCO, IOCS and XAD-8 and receiving Judy Reservoir water are presented in Figures 4.8 and 4.9.
Effluent DOC (mg/L)
3.0
2.0
1.0 IOCO IOCS XAD-8 0.0 0
50
100
150
200
250
300
350
400
Bed volumes
Figure 4.8. DOC breakthrough curves for treatment of water from Judy Reservoir using IOCO, IOCS and XAD-8 (influent DOC = 3.6 mg/L).
105
0.5 IOCO IOCS XAD-8
A254 (cm-1)
0.4
0.3
0.2
0.1
0.0 0
100
200
300
400
Bed volumes
Figure 4.9. A254 breakthrough curves for treatment of water from Judy Reservoir using IOCO, IOCS and XAD-8 (influent A254 = 0.62 cm−1) Equation 2.3 was used to convert the data to values of the average retention efficiency of DOC and A254, respectively, as a function of the number of bed volumes of water processed. The results are shown in Figures 4.10 and 4.11. Of the three adsorbents, IOCS was most efficient at removing both A254 and DOC from solution. The integrated DOC collection efficiencies of IOCO, IOCS and XAD-8 are similar for the two water sources and are close to 80% for treatment of fewer than ten BVs. However, for treatment of ~50 to a few hundred BVs of water, the efficiency of IOCS remains close to 70%, whereas it decreases to ca. 45% for XAD-8 and to approximately 20% for IOCO. In terms of retention of A254, the performance of IOCS is close to ideal (ηUV near 90%) for processing of several hundred bed volumes of water, while for IOCO and XAD-8, ηUV is approximately 40 and 60%, respectively.
106
DOC retention efficiency
100%
75%
50%
IOCO IOCS XAD-8
25%
0% 0
100
200
300
400
Bed volumes
Figure 4.10. Cumulative DOC retention efficiency for treatment of Judy Reservoir water using IOCO, IOCS and XAD-8
107
UV retention efficiency
100%
75%
50% IOCO IOCS XAD-8
25%
0% 0
100
200
300
400
Bed volumes
Figure 4.11. Cumulative A254 retention efficiency for treatment of Judy Reservoir water using IOCO, IOCS and XAD-8 When water from the Tolt River was processed, the IOCS once again outperformed the other two media in terms of DOC and A254 retention (Figures 4.12 through 4.15). However, in this case, IOCO outperformed XAD-8.
108
Effluent DOC (mg/L)
1.5
1.0
0.5
IOCO IOCS XAD-8
0.0 0
100
200
300
400
Bed volumes
Figure 4.12. DOC breakthrough curves for processing of Tolt River water using IOCO, IOCS and XAD-8 adsorbents (influent DOC = 1.9 mg/L)
109
0.25 IOCO IOCS XAD-8
A254 (cm-1)
0.20
0.15
0.10
0.05
0.00 0
100
200
300
400
Bed volumes
Figure 4.13. A254 breakthrough curves for processing of Tolt River water using IOCO, IOCS and XAD-8 adsorbents (influent A254 = 0.26 cm−1)
110
DOC retention efficiency
100%
75%
50%
IOCO IOCS XAD-8
25%
0% 0
100
200
300
400
Bed volumes
Figure 4.14. Cumulative DOC retention efficiency by IOCO, IOCS and XAD-8 processing Tolt River water
111
UV retention efficiency
100%
75%
50%
IOCO IOCS XAD-8
25%
0% 0
100
200
300
400
Bed volumes
Figure 4.15. Cumulative A254 retention efficiency by IOCO, IOCS and XAD-8 processing Tolt River water
Competition Between Sulfate and NOM for Oxide Adsorbents
The most significant limitation on the use of IOCS for concentrating NOM from natural fresh waters appears to be competition between NOM and sulfate for the IOCS surface sites. NOM concentrates obtained by eluting the IOCS or IOCO columns with sodium hydroxide contained substantial amounts of sulfate, but not nitrate or chloride. Judy Reservoir water contains comparable concentrations of DOC and sulfate (3.4 and 3.7 mg/L, respectively). When this water was applied to the IOCS column, 90% breakthrough of DOC occurred at BV≈850, whereas 90% breakthrough of sulfate occurred at BV≈255. Thus, NOM has an affinity for the IOCS surface that is three to four times as large as that of sulfate. The relative affinities are close enough that competition from sulfate could be a serious impediment to the use of IOCS for NOM concentration, especially for low-NOM, high-sulfate waters. As a rough guideline, IOCS is expected to 112
perform well for SO4/DOC mass ratios below about 2.5, and to perform acceptably for SO4/DOC mass ratios between 2.5 and about 5. For SO4/DOC ratios >5, the presence of sulfate will become a substantial interfering factor. The actual performance of IOCS in high-sulfate waters was not tested in this study. However, it is reasonable to expect that in such waters use of alternative concentration techniques (XAD-8/XAD-4 and/or membranes) would be preferable.
Recovery of Various NOM Fractions with IOCO
Several samples of effluent from IOCO processing of Judy Reservoir water were fractionated using XAD and ion exchange resins. The samples were collected after various numbers of BVs had passed through the IOCO column and had DOC concentrations ranging from 3 mg/L. The volume of sample processed in the fractionation experiments varied from 500 to 2000 mL depending on DOC, such that the total load of organic carbon applied to the fractionation columns was always close to 2 mg. Only the neutral and acidic NOM fractions were analyzed, because previous research suggested that organic bases were present in the raw water in negligible amounts and because, if any bases had been present, they probably would have been partially removed by the ion exchange step applied in processing the sample. The results of the analyses are shown in Table 4.13.
113
Table 4.13 Results for the fractionation of IOCO effluent for treatment of Judy Reservoir water
Range of BVs during sample collection DOC (mg/L)
Effluent #1
Effluent #2
Effluent #3
Effluent #4
raw water
6-8
54-56
198-200
396-398
n/a
1.05
1.65
2.30
2.88
3.40
Percentage of DOC in Given Fraction hydrophobic fractions (XAD-8) transphilic fractions (XAD-4) hydrophilic acids (Duolite A-7) hydrophilic neutrals
31
38
38
38
39
20
24
23
23
22
0
0
0
8
11
50
39
38
30
28
Changes over time in the composition of NOM in the IOCO effluent reflect progressive saturation of the column with adsorbed NOM. At low bed volumes, the hydrophilic neutrals account for 50% of the DOC in the effluent compared to only 28% in the raw water, indicating that the IOCO preferentially adsorbs the hydrophobic and hydrophilic acidic fractions of NOM. At later times, the IOCO seems to be especially efficient at collecting hydrophilic acids while not collecting hydrophilic neutrals. Thus, throughout the run, the IOCO adsorbs acidic fractions of NOM preferentially and efficiently. In all likelihood, the same result would be obtained for IOCS and other iron oxide-based adsorbent media (see Korshin et al. (1997) for additional discussion of the IOCS data).
Comparison of NOM Concentration-Isolation-Fractionation Techniques
This portion of the research evaluated NOM isolation using both traditional and novel methods (RO, NF, XAD-8/XAD-4, iron-oxide-coated adsorbent media). The XAD-8 resin protocol developed by researchers at the U.S. Geological Survey (Leenheer 1981, Thurman and Malcolm 1981) has been widely used and is considered the reference method for the isolation of aquatic NOM. In this procedure, NOM is separated into hydrophobic and hydrophilic parts. The hydrophobic fraction of NOM is separated by 114
adsorption at pH = 2 onto XAD-8 resin packed in a column, using a pre-defined column capacity factor k´ (typically in the range 50 to 100). Most (at least 70%, and often >80%) of the NOM adsorbed on XAD-8 resin can be eluted with 0.1 N NaOH; this fraction is referred to as the hydrophobic acid (HPOA) fraction. Almost all of the NOM remaining on the resin can be recovered with acetonitrile and is referred as hydrophobic neutrals (HPON). Conventionally, hydrophilic NOM is isolated using a similar process with XAD-4 resin. However, in this study, the NOM that is conventionally defined as hydrophilic was collected as three portions, which were defined (in order of increasing hydrophilic character) as the transphilic, hydrophilic and ultrahydrophilic fractions. Elution of XAD-4 resin with NaOH yields the transphilic acid (TPHA) fraction that typically comprises 50 to 70% of the NOM adsorbed by this resin. The subsequent elution of XAD-4 resin with acetonitrile yields the transphilic neutral fraction (TPHN). The NOM that is not adsorbed is processed further and fractionated by sorption reactions into hydrophilic and ultrahydrophilic fractions. Although significant effort is required to clean the resins used in these procedures, the adsorption and elution processes are relatively simple. The literature (Martin 1997) and our data indicate that the percentage of NOM that adsorbs on XAD-8 increases substantially with increasing SUVA254 of the influent (Figure 4.16). For Suwannee and Blavet (winter) NOM that had SUVA254 values close to 5 L/(mg-m), approximately 80% of the DOC adsorbed. By contrast, for South Platte and Tolt NOM, with SUVA values TPH-NOM > HPI-NOM. Similar variations were found in South Platte River NOM fractions, but SUVA was much lower than for corresponding fractions from the Suwannee River. SUVA was well correlated with the width of the ET band for NOM fractions from both water samples. The state and concentration of aromatic carbon in the samples largely define their spectral properties. At least two spectral parameters (SUVA254 and ∆ET) can be used to estimate the aromatic carbon content of the sample.
Fluorescence Spectrometry. For the Suwannee River NOM fractions, fluorescence emission yields increase and the maxima of emission spectra shift to lower wavelengths as the hydrophilic character of the NOM increases. This trend is indicative of a relative increase in lower molecular weight species as hydrophilic character increases. Fluorescence data for South Platte River NOM fractions indicate generally smaller molecular weights than for corresponding Suwannee River NOM fractions. The fluorescence results suggest that the average molecular weight of NOM fractions tends to increase as the hydrophilic character of the fraction increases. The fluorescence spectra also appear to be sensitive to the hydrophilicity and to acidity-basicity of the NOM fractions. These results might reflect an inverse correlation between the average molecular weight and the hydrophilicity of the fractions. 188
8. Characterization Correlations. The carbon content of the NOM fractions investigated correlated inversely with carbohydrate content (based on
13
C-NMR), and oxygen
content was directly correlated with carboxylic acids and/or amides (13C-NMR). Proteins (pyrolysis GC-MS) were correlated with TDAA, but nitrogen content correlations were weak because of multiple sources of nitrogen from both proteins and amino sugars. For Suwannee River NOM fractions, aromatic carbon content (13C-NMR) was directly correlated with SUVA, ∆ET, PHA fragments, and the fluorescence emission maximum; however, for the South Platte River NOM fraction, aromatic carbon content correlated only with PHA fragments, indicating the importance of phenolic aromatics (as compared with total aromatic carbon) on ultraviolet and fluorescence measurements. 9. Overall NOM Characteristics. Suwannee River NOM is derived from allochthonous tannins and lignins that give rise to highly colored humic and fulvic acids of high poly-phenol content and low nitrogen content. Hydrophobic characteristics predominate. South Platte River NOM is derived from both allochthonous inputs in which aromatic and acid constituents are depleted by mineral solubility controls and autochthonous inputs from phytoplankton and bacteria. These autochthonous inputs contribute proteins, polysaccharides, and amino sugars that result in significant hydrophilic character for this NOM. 10. Reactivity of NOM a. Coagulation-Flocculation. Coagulation with alum at pH 6.5 removes acid NOM fractions better than neutral or base fractions. NOM in the Suwannee River fractions was removed more efficiently than that in than South Platte River fractions. Moderate correlations were found between SUVA and DOC removal by coagulation, and better correlations were found between SUVA and A254 removal. b. Chlorination. Chlorinated DBP yields were greater in Suwannee River NOM fraction isolates than in corresponding fractions from the South Platte River. For Suwannee River NOM fractions, THM and TCAA yields are relatively similar; 189
chloroform was the major chlorination by-product for South Platte River NOM fractions. SUVA correlated well (and aromatic carbon somewhat less well) with THMFP, TOXFP, and TCAAFP, but did not correlate with DCAAFP. Extrapolation of the correlation plots suggests that significant amounts of DBPs could be produced even from NOM fractions with negligible SUVA or aromaticity. Overall, the results point to major differences between Suwannee River and South Platte River NOM characteristics and reactivity. In almost all cases, the various NOM characterization methods gave supporting and complementary information rather than contradictory information. Despite some as yet unexplained findings, the qualitative and semi-quantitative approaches used for the interpretation of Pyr-GC-MS chromatograms support previous structural hypotheses for some NOM fractions and confirm the structural differences between hydrophobic and hydrophilic NOM. The combination of
13
C-NMR, FTIR and
Pyr-GC-MS analyses are complementary and can be a powerful tool for NOM characterization.
190
CHAPTER 6
CHARACTERISTICS OF NOM CAPTURED BY DIFFERENT TECHNIQUES
NOM from the Vienne, Gartempe, and Blavet Rivers in France, the Thames river in England, and Judy Reservoir and the Tolt River in Washington State was fractionated and characterized, though not as intensively as that from the Suwannee and South Platte rivers. The intent was to determine whether and how different fractionation techniques alter the NOM and thereby affect the conclusions that one might draw about the NOM in the water source. Although not all fractionation and characterization tools were used for all waters, the results of the testing were qualitatively similar in all cases. Of the waters studied, the Blavet River was investigated most thoroughly. For this reason, and in light of the large amount of data gathered, the discussion in this chapter focuses primarily on the Blavet. Most of the experiments and analyses reported in this chapter were conducted using NOM isolates that were lyophilized and then, in some cases, re-dissolved. However, the SUVA of the RO and NF isolates was analyzed on the concentrated waters prior to lyophilization. Dissolution of the lyophilized sample of non-desalted RO and NF concentrates was often problematic, yielding solutions that contained some suspended particulates that were removed by 0.45-µm membrane filtration.
GENERAL CHARACTERIZATION OF NOM IN THE BLAVET RIVER
The Blavet River was sampled twice during the project, once in summer and once in winter. Inorganic characteristics of the river water were quite similar in the two samples, except that the pH was higher in the summer than winter (7.8 vs. 7.0). The difference was probably due to the algal bloom that was in progress during the summer sampling period (in fact, a few hours after the summer sampling, copper sulfate was added to the reservoir to reduce the algal population). Bromide was analyzed monthly
191
between September 1995 and October 1996, and its concentration varied from 60 to 120 µg/L (Figure 6.1), with the maximum at the beginning of the summer.
Br− Concentration, µ g/L
120 100 80 60 40 20 0 Jul-95
Oct-95
Feb-96
May-96
Aug-96
Dec-96
Date
Figure 6.1. Bromide concentrations in the Blavet River during a one-year sampling period By contrast, the organic composition of the water varied substantially throughout the year. Figure 6.2 presents the DOC and A254 values obtained in a survey conducted by the SAUR laboratory between September 1995 and October 1996 on the Blavet River at the Kerne Uhel reservoir, and Figure 6.3 shows the profile of SUVA254 during the same period. The winter sample contained 12 mg/L DOC, compared with 6.6 mg/L during the summer. A254 was also significantly lower during the summer, causing the SUVA254 values to be similar in the two samples (4.8 and 5.1 L/mg-m, respectively). The high SUVA254 values indicate that the organic material has a pronounced humic nature. The similarity in the SUVA values in summer and winter is somewhat unexpected, in light of the algae bloom that was occurring during the summer sampling period.
192
5 4.5
10
4 3.5
8
3
6
2.5
DOC UV
4
2 1.5 1
2
0.5
0 Oct-96
Aug-96
Jul-96
May-96
Mar-96
Feb-96
Dec-95
Oct-95
Jul-95
Sep-95
0
Date
Oct-96
Sep-96
Aug-96
Jul-96
Jun-96
May-96
Apr-96
Mar-96
Feb-96
Jan-96
Dec-95
Nov-95
5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 Oct-95
SUVA254, m− 1
Figure 6.2. DOC and A254 in the Blavet River during a one-year sampling period
Date
Figure 6.3. SUVA254 in the Blavet River during a one-year sampling period 193
UV254, m− 1
DOC Concentration, mg/L
12
DOC concentration and UV absorbance in the Blavet decreased by ~50% during the first few months of 1996, increased to the earlier values during Spring, 1997, then decreased to reach relatively stable values in the summer. The increase in Spring probably reflects an algal bloom. The DOC content of this reservoir also seems to be strongly related to the local rainfall pattern (greatest in fall and winter), which introduces soluble terrestrial organic materials from runoff. Turnover of the reservoir may also play a role. Other properties of the NOM in the Blavet also differed during the summer and winter sampling periods. For example, the absolute concentrations of TDAA and TDCA were substantially higher in the summer sample (Table 6.1). Because DOC was higher in winter sample, these components comprised a much larger fraction of the DOC in summer than in winter. These results were undoubtedly influenced by the algal bloom noted above. Most of the increase in the TDAA content in the summer sample could be attributed to a large increase in the ornithine concentration, which is a good indicator of intense biological (**JP algal?) activity (Spitzy 1988). The increase in the TDCA concentration was more uniformly distributed among all the carbohydrates detected, with fucose having the largest proportional increase. Table 6.1 TDCA, TDAA, and BDOC in the Blavet River Winter
Summer
TDAA µg C/L
342
664
µg/L N
133
277
µg C/mg C
28
100
TDCA µg C/L
194
338
µg C/mg C
16
51
The apparent molecular weight (AMW) distribution of the NOM (Table 6.2) provided further support that the organics in the summer sample differed significantly from those in the winter. Specifically, the proportion of DOC smaller than 1,000 Daltons
194
increased from 7% in winter to 17% in summer, a change that is consistent with a somewhat more hydrophilic character of the NOM in summer. Table 6.2 Apparent molecular weight distribution of the NOM of the Blavet River Apparent Molecular Weight Distribution Range (amu) 10,000
DOC
A254
DOC
A254
DOC
A254
DOC
A254
(%)
(%)
(%)
(%)
(%)
(%)
(%)
(%)
Winter
3.5
1.2
3.5
3.1
47.8
40.9
45.2
54.8
Summer
6.8
1.1
10.5
8.6
36.5
28.1
47.2
62.2
ELEMENTAL ANALYSIS AND SUVA OF THE NOM ISOLATES FROM THE BLAVET RIVER
Table 6.3 shows the elemental analyses of the NOM isolates from the Blavet River samples. Although samples that were obtained using the same techniques in the two sampling periods do not differ dramatically in their elemental composition, apparent differences in composition are noticeable when dissimilar NOM concentration methods are compared. All the RO and NF NOM isolates had a substantial amount of ash, causing the estimate of carbon content to be low and preventing analysis of oxygen. For the winter samples, attempts were made to remove cations from the RO and NF concentrates by H+-ion exchange prior to lyophilization. This reduced the ash content of the isolates, but not enough to allow an accurate elemental analysis to be conducted. For the summer samples, ash accounted for approximately one-third of the mass of the RO and NF isolates. HCl acidification and CO2 stripping helped reduce the ash contents to 8 to 15%, but this was still too high to obtain good results from the elemental analyses of the organics.
195
Table 6.3 Elemental analysis of the Blavet River NOM isolates Fraction
C (%)
H (%)
N (%)
O (%)
S (%)
Ash (%)
*w
s
w
s
w
s
w
s
w
s
w
s
46.1
44.8
4.5
4.4
2.1
1.7
39.9
41.8
1.1
nd
2.4
3.0
XAD-8 fraction
47.0
46.6
4.6
4.6
2.0
2.1
38.8
38.4
0.5
nd
6.2
2.8
XAD-4 fraction
43.3
41.7
4.6
4.2
2.9
2.5
40.6
44.9
1.6
nd
2.7
3.5
RO isolate
13.0
10.6
1.5
1.2
2.0
2.2
nd
nd
1.8
nd
34.3
33.3
NF isolate
15.6
13.2
1.7
1.4
1.4
1.5
nd
nd
2.7
nd
20.5
32.5
RO + IX
22.5
-
3.7
-
1.3
-
44.8
-
7.9
-
14.5
-
NF + IX
24.9
-
3.9
-
1.4
-
nd
-
6.5
-
13.7
-
RO + XAD-8
35.0
-
3.7
-
1.5
-
31.8
-
0.7
-
2.7
-
RO + XAD-4
46.2
-
4.7
-
2.5
-
39.4
-
0.9
-
3.7
-
RO + HCl
-
6.9
-
0.8
-
2.2
-
nd
-
nd
-
15.2
NF + HCl
-
7.7
-
1.0
-
0.9
-
nd
-
nd
-
8.7
NF + XAD-4
-
45.0
-
4.4
-
2.6
-
40.4
-
nd
-
4.2
XAD-8/XAD-4 mixture
* w: winter sample, s: summer sample **JP is this correct: nd: analysis not reliable due to interferences
By contrast, the XAD-8 and XAD-4 desalting procedures were efficient in removing inorganic species from the membrane-concentrated waters. The elemental analysis of the RO + XAD-4 isolate was almost identical to that of the XAD-8/XAD-4 mixture, and >97% of the NOM mass in the original sample could be accounted for by the masses of the elements recovered. On the other hand, only about 75% of the original mass was recovered in the RO + XAD-8 isolate. No explanation is apparent for this loss of material. For both rounds of sampling, the XAD-4 fraction was richer in oxygen and nitrogen and poorer in carbon than the XAD-8 fraction. In the winter sample, the ash content was significantly lower in the XAD-4 fraction (2.7%) than in the XAD-8 isolate (6%), and sulfur was also more abundant in the XAD-4 fraction.
196
Table 6.4 shows the C/O, C/N and C/H ratios (by weight) and SUVA254 values for the isolates. The elemental ratios support the greater hydrophilic character of the XAD-4 fraction than the XAD-8 fraction. The XAD-4 isolates that have the highest hydrophilicity according to their C/O, C/N and C/H ratios also have the lowest SUVA254 values. The opposite was observed for the (hydrophobic) XAD-8 fraction. The SUVA254 value of the XAD-8/XAD-4 mixture was comparable to that of the XAD-8 fraction, because the XAD-8 fraction comprises up to 90% of the isolate. The XAD-8/XAD-4 mixture and the RO + XAD-4 isolate were also similar with respect to both elemental composition and SUVA254. In the absence of XAD resin desalting, the elemental ratios calculated for the RO and NF isolates are not necessarily meaningful. The RO concentrates that were desalted with XAD-4 and XAD-8 have similar elemental ratios, although this result should be interpreted with caution, given the loss of material in the RO + XAD-8 sample preparation process.
197
Table 6.4 C/O, C/N and C/H ratios with SUVA of the Blavet River NOM C/O
C/N
C/H
SUVA (L/mg-m)
w.
s.
w.
s.
w.
s.
w.
s.
XAD-8/XAD-4 mixture
1.15
1.07
21.75
26.18
10.27
10.29
4.4
4.1
XAD-8 fraction
1.21
1.21
23.73
21.76
10.33
10.17
4.3
5.2
XAD-4 fraction
1.06
0.93
14.77
16.74
9.41
9.95
2.6
3.2
RO isolate
nd
nd
6.45
4.84
8.65
8.66
4.0
5.2
NF isolate
nd
nd
11.27
9.07
9.09
9.32
4.3
4.16
RO + IX
0.50
-
16.95
-
6.17
-
nd
-
NF + IX.
nd
-
17.50
-
6.39
-
nd
-
RO + XAD-8
1.10
-
22.74
-
9.39
-
4.3
-
RO + XAD-4
1.17
-
18.32
-
9.91
-
3.7
-
RO + HCl
-
nd
-
3.07
-
8.35
-
nd
NF + HCl
-
nd
-
8.92
-
7.91
-
nd
NF + XAD-4
-
1.11
-
17.23
-
10.31
-
4.3
TOTAL DISSOLVED AMINO ACIDS AND CARBOHYDRATES
The TDAA and TDCA contents of the Blavet River NOM isolates are presented in Table 6.5, and the distributions of species comprising the TDAA and TDCA are represented in Figures 6.4 through 6.7. Among the three summer isolates analyzed for TDAA and TDCA, the RO isolate had the highest concentration of these compounds. The XAD-8/XAD-4 sample contained less carbohydrates and more amino acids than the XAD-4 fraction. For the winter samples, the TDCA content decreased in the order XAD-8 ≈ NF + IX > XAD-8/XAD-4 > XAD-4 ≈ NF + IX. Although it is possible that the difference between the NF + IX and RO + IX isolates was caused by preferential adsorption of carbohydrates onto the surface of the RO membrane, no such difference 198
was observed for NOM from the Vienne River (data not shown). The difference might have been caused by analytical problems. Table 6.5 TDAA content of the Blavet River NOM isolates (winter sample) TDAA µg C/mg C
TDAA µg N/mg C
TDCA µg C/mg C
winter
summer
winter
summer
winter
summer
25
23
9
9
15
17
XAD-8 fraction
30
-
11.5
-
21
-
XAD-4 fraction
22
18.4
9
7
10
24
RO+H+cat res.
13
-
5.4
-
8
-
NF+ H+cat res
12.5
-
5
-
19
-
RO isolate
-
43
-
16
-
39.4
XAD-8/XAD-4
Amino Acid
Figure 6.4. TDAA distribution of the Blavet River (winter sample) 199
lys
orn
leu
ile
phe
met
tyr
ala
thr
gly
arg
his
ser
glu
10 9 8 7 6 5 4 3 2 1 0 asp
µ g AA carbon / mg DOC
mixture
2.0 1.5 1.0 0.5
fructose
man-xyl
glucose
galactose
glucosamine
arabinose
rhamnose
0.0 fucose
µ g carbohydrate C / mg DOC
2.5
Carbohydrate
Figure 6.5. TDCA distribution of the Blavet River (winter sample)
µ g AA carbon / mg DOC
40 35 30 25 20 15 10 5
Amino Acid
Figure 6.6. TDAA distribution of the Blavet River (summer sample) 200
lys
orn
leu
ile
phe
met
tyr
ala
thr
gly
arg
his
ser
glu
asp
0
12 10 8 6 4 2 fructose
man-xyl
glucose
galactose
glucosamine
arabinose
rhamnose
0 fucose
µ g carbohydrate C / mg DOC
14
Carbohydrate
Figure 6.7. TDCA distribution of the Blavet River (summer sample) The TDAA in the XAD resin fractions did not follow the general trend discussed in the literature (Croue et al. 1993a) or observed in the Suwannee and South Platte Rivers. In this sample, TDAA was higher in the XAD-8 (HPOA) fraction than in the XAD-4 (TPHA) fraction. Furthermore, less TDAA appeared in the RO + IX and NF + IX isolates than in the fractions isolated using XAD resins, a result that is almost certainly attributable to adsorption of amino acids onto the ion exchange resin.
13
C-NMR AND FTIR DATA FOR THE BLAVET RIVER NOM ISOLATES
The FTIR spectra of the Blavet River NOM isolates obtained with XAD resins did not show any trace of the major anions (bicarbonate, nitrate and sulfate) in either winter or summer, indicating that the desalting process was very efficient (Figure in Appendix**MB). Silica could be detected in the XAD-8 and the XAD-8/ XAD-4 winter isolates (peak at 470 cm−1), but not in the spectrum of the XAD-4 sample from the winter or in any of the spectra of the summer isolates. In the FTIR spectra of the winter and 201
summer XAD-8 and NF + XAD-4 isolates, there are features associated with phenolic structures. In the winter samples, these phenolic structures are more prominent in the XAD-8 and XAD-8/XAD-4 isolates. The major peaks in the FTIR spectra of the RO and NF concentrates were caused by the inorganic anions. Not surprisingly, the same is true after each of the concentrates was treated by H+-cation exchange. These results reaffirm that membrane treatment followed by cation exchange is not appropriate if a low-salt NOM concentrate is required. The FTIR spectrum of the RO + XAD-4 isolate had no significant signal from inorganic species except for a small silica signal. In general, it looked similar to the spectrum of the XAD-8/XAD-4 mixture. Partial desalting via H+-cation exchange did not allow complete resolution of the 13
C-NMR spectra of the membrane isolates. As discussed previously, the apparent
increase of the aromatic carbon content and the presence of distinct bands in the C-O and C-C regions in the membrane concentrates might be explained by the presence of sulphonated aliphatic and aromatic structures that were produced by lyophilization in the presence of sulfuric acid. The reaction between sulfuric acid and carbohydrate species can also lead to the production of furans that increase the intensity of the aromatic carbon band. Therefore, the higher aromatic carbon content in the membrane samples might be an artifact caused by interactions between the organic and inorganic components of the isolates. The spectra of the winter RO + XAD-8 and RO + XAD-4 isolates look more like those of ‘normal’ NOM. The major differences between the two spectra are the larger phenolic content of the RO + XAD-8 isolate and the larger C-O band of the RO + XAD-4 isolate. The integrated areas of the 13C-NMR spectra in Figures Error! Reference source not found. and Error! Reference source not found. are reported in Table 6.6. This
semi-quantitative approach could be applied only to the fractions that were desalted on XAD resins. According to the
13
C-NMR integration data for the winter samples, the 202
XAD-4 fraction is the most hydrophilic NOM fraction, with the highest C-O content, highest anomeric carbon content, lowest aromatic carbon content, and high carboxylic acid content. The XAD-8 fraction is the most hydrophobic. The three other NOM fractions have similar carbon distributions that indicate an intermediate hydrophobic (or intermediate hydrophilic) character. Among these three fractions, the RO + XAD-4 isolate was poorest in aromatic carbon content, as expected from its SUVA254. Table 6.6 Integrated areas of 13C-NMR spectra of the Blavet River NOM isolates Integration range (ppm) 0-60
60-90
90-110
110-160
160-190
190-220
Sample
w.
s.
w.
s.
w.
s.
w.
s.
w.
s.
w.
s.
XAD-8/XAD-4
* 34
47
21
20
5
7
21
15
15
10
4
1
XAD-8 fraction
39
41
13
19
4
5
28
20
13
13
3
2
XAD-4 fraction
38
35
24
25
7
7
13
12
16
19
2
2
RO + XAD-8
36
-
17
-
6
-
22
-
16
-
2
-
RO + XAD-4
37
-
21
-
6
-
17
-
16
-
3
-
NF + HCl
-
35
-
21
-
10
-
16
-
17
-
2
NF + XAD-4
-
42
-
20
-
8
-
15
-
13
-
1
RO + HCl
-
38
-
31
-
10
-
9
-
10
-
2
mixture
* Values represent the area under the total integrated area of the spectrum.
13
C-NMR spectral curve in the indicated range as a percentage of
Based on integration of the NMR spectra in the 110 to 160 ppm range, the aromaticity of all the summer NOM isolates is low, exceeding 16% only in the XAD-8 fraction (20%). The aromaticity of the XAD-8 and XAD-8/XAD-4 isolates are notably higher in the winter than the summer. The relatively hydrophilic character of the NF + HCl isolate is confirmed by its high proportion of alcoholic and carboxylic carbon. The carbon distribution of the XAD-4 fraction was similar to that of the NF + HCl isolate, while the NF + XAD-4 isolate was more comparable to the XAD-8/XAD-4
203
mixture. Thus, partial or total desalting of the membrane isolates had a significant impact on the apparent structural characteristics of the NOM.
PYROLYSIS - GAS CHROMATOGRAPHY - MASS SPECTROMETRY FOR THE BLAVET RIVER NOM ISOLATES
The relative proportions of biopolymers in the Blavet River NOM isolates are given in Table 6.7. (This semi-quantitative approach was not utilized on the NF + IX isolate because of the structural changes noted above.) Table 6.7 Relative proportions of biopolymers in the Blavet River NOM isolates Fraction
* PHA
UA
PR
PS
AS
unknown
Fraction of DOC (%) w†
s
w
s
w
s
w
s
w
s
w
s
XAD-8/XAD-4 mixture
31
21.8
21
14.7
14.6
20.5
19.5
21.4
1.7
6.9
14
14.7
RO isolate
35.5
15
8.8
8.7
35
31.6
7
9.5
10.5
33.8
2.6
1.4
NF isolate
33.6
-
14.4
-
35
-
6.9
-
8.5
-
1.5
-
RO + XAD-4
19.7
-
7.2
-
26.8
-
11.4
-
7.2
-
13.7
-
XAD-8 fraction
-
25.9
-
13.4
-
21.3
-
18.7
-
4.7
-
16
XAD-4 fraction
-
14.4
-
16.3
-
22.6
-
18
-
10.2
-
18.5
PHA: Polyhydroxyaromatics, UA: unsubstituted aromatics, PR: Proteins, PS: Polysaccharides, AS: Amino sugars † w: winter, s: summer
Based on these analyses, the summer XAD isolates had higher proportions of polyhydroxyaromatics and polysaccharides than the winter sample. The differences between the XAD-8 and XAD-4 fractions of the Blavet NOM are similar to those that characterize these fractions in the Suwannee and South Platte Rivers (see Chapter 5). Phenol and cresol are the major peaks of the XAD-8 fraction, while acetic acid, 204
acetamide (fragment from amino sugars), acetonitrile and methylpyrrole (produced from proteins),
methyl
furfural
and
levoglucosenone
(fragments
produced
from
polysaccharides) are proportionally more abundant in the pyrochromatogram of the XAD-4 fraction. The XAD-4 fraction was also characterized by its higher proportion of acetamide and lower proportion of polyhydroxyaromatics the XAD-8 sample. Qualitatively, the pyrochromatogram of the winter XAD-8/XAD-4 isolate is also a typical fingerprint of NOM associated with the hydrophobic acid fraction. The pyrolysis data for the XAD-8/XAD-4 isolate correspond to the results obtained for the two XAD resin fractions mixed in a 70%/30% mass ratio. The Pyr-GC-MS chromatograms of the NF and RO isolates were different from those of the XAD-8/XAD-4 samples, having acetonitrile (a protein pyrolysis fragment) and acetamide (an amino sugar pyrolysis fragment) as major constituents. In the winter RO isolate, the signals from acetonitrile and acetamide were higher than even those from phenol and cresol, which are commonly found PHA fragments. Ethyl hexanol, which is generally not found as a pyrolysis fragment of NOM, was also identified in these two chromatograms for the winter RO sample. This was not the case for the summer RO isolate, whose pyrochromatogram contained butanine, benzene and acetamide as the major fragments. Phenol was also an important fragment, but most of the other identified peaks were generated from the pyrolysis of polysaccharides and proteins. The NOM isolated via RO was enriched in proteins and amino sugars as compared to the resinisolated fractions, but it contained lower proportions of polyhydroxyaromatic and polysaccharide moieties, indicating that it is quite hydrophilic. In order to determine the origins of ethyl hexanol in the winter RO sample, a small piece of hexanal tubing that is used in the membrane apparatus was pyrolyzed. The corresponding pyrochromatogram showed the presence of this compound, but only at a trace level that could not explain the large signal from this molecule in the RO isolate. Possible explanations for the presence of ethyl hexanol in the pyrochromatogram include: •
the compound was originally present in the raw water as a contaminant. Ethyl hexanol is a widely used industrial chemical (mainly for the production of poly(vinyl 205
chloride) plasticizers. Ethyl hexanol is widely diffused in the environment and its presence has been detected in natural waters (Vitali and Leoni 1993). •
the compound was released to the water from the membrane or from another part of the membrane unit, or was produced during the pyrolysis of an impurity that came from them. This possibility is considered unlikely, because the RO and NF isolates produced with the same unit did not show any trace of ethyl hexanol in their Pyr-GC-MS spectra.
•
the compound was produced during the pyrolysis of NOM structures that were selectively isolated with the membranes. The final hypothesis seems to be the most plausible, because RO was the most
efficient technique in terms of overall NOM retention, and trace amounts of organic species could be retained by this method and lost by others. Nevertheless, the exact source of the ethyl hexanol in the winter RO concentrate remains unclear. The pyrochromatogram of the winter RO + XAD-4 isolate was also characterized by an abundance of pyrolysis fragments produced from proteins (especially pyrrole) and amino sugars. The identification of sulfur-containing organic compounds in the Pyr-GC-MS chromatogram of the NF + IX isolate seems to confirm the hypothesis that the structure of NOM was modified by lyophilization in the presence of sulfuric acid. The major points established by the table can be summarized as follows: •
with or without desalting, the membrane isolates contain larger proportions of proteins and amino sugars than the XAD-8/XAD-4 mixture does.
•
the XAD-8/XAD-4 mixture contains a larger proportion of polysaccharides than the non-desalted RO and NF isolates and the RO + XAD-4 isolate.
•
the RO + XAD-4 isolate contains a lower proportion of polyhydroxyaromatic structures than the other three NOM fractions investigated, all of which contain similar amounts of this structural unit. 206
This last point is consistent with the SUVA254 values and
13
C-NMR spectra for
these fractions, which support the idea of a preferential loss of high molecular weight organics during the back elution of the column with the acetonitrile/ water solution.
CHLORINATION OF THE BLAVET RIVER NOM ISOLATES
Table 6.8 shows the formation of by-products as a result of chlorination of the Blavet River and its isolates under UFC conditions. The TOX concentrations analyzed in all solutions were similar, except for the summer RO and NF isolates.
207
Table 6.8 Some chlorinated DBPs formed by chlorination of the Blavet River and its NOM isolates XAD-8/XAD-4 mixture
NF isolate
RO isolate
Raw water
w.
s.
w.
s.
w.
s.
w.
s.
Cl2 dose (mg/mgC)
1.4
1.4
1.3
1.1
1.6
1.4
1.6
1.6
Cl2 residual (mg/L)
1.1
0.9
0.8
2.6
1.2
5.3
1.0
0.6
TOX (µgCl/mgC)
143
146
140
80.4
164
78
160
157
CHCl3 (µg/mgC)
42.8
39.3
46.6
25.4
46.6
26.7
47.4
44.1
CHCl2Br (µg/mgC)
0.6
0.9
2.2
2.5
3.2
3.4
4.2
8.7
CHClBr2 (µg/mgC)
nd
Nd
0.05
0.1
0.1
0.2
0.2
0.8
TTHM (µgCl/mgC)
38.7
35.6
42.9
24.4
43.6
24.0
45.1
45.4
DCAA (µg/mgC)
11.5
17.3
8.1
9.2
8.8
9.6
14.2
19.4
TCAA (µg/mgC)
20.2
21.8
9.8
13.4
10.1
13.4
19.7
19.7
DBAA (µg/mgC)
nd
0.01
0.9
0.1
0.6
1.5
0.9
0.2
DCBAA (µg/mgC)
0.1
0.6
0.1
0.7
0.2
0.08
0.1
3.2
DBCAA (µg/mgC)
0.1
Nd
0.8
nd
0.8
nd
0.08
nd
The results obtained with the XAD-8/XAD-4 mixture were similar to those for the unfractionated sample. The addition of bromide did not significantly change the yield and speciation of DBPs. For all samples, chloroform was the major THM species produced during chlorination, accounting for about 30% of the TOX. Except in a few cases discussed 208
below, the yield of chloroform was similar in all the solutions studied, ranging from 41 to 48 µg chloroform generated per mg of DOC in the sample. CHCl2Br and CHClBr2 were detected in all chlorinated solutions, but CHBr3 was below the detection limit. CHCl2Br was the main brominated THM. The concentrations of brominated THMs were lower for the NF and RO isolates than for the raw water, probably because of loss of bromide during the membrane concentration step. The production of brominated THMs was lower in the XAD-8/XAD-4 mixture than in the other solutions, however the values obtained with the mixed XAD-8/XAD-4 solution spiked with bromide were significantly increased and of the same magnitude as the raw water and RO isolate (data not shown? **JP am I right that we are referring to data here that we are not including? (ok with me, I’m just checking)) TCAA and DCAA were the main HAAs detected, and the brominated HAAs were present in relatively minor concentrations. Chlorination of the membrane isolates generated less TCAA and DCAA than did chlorination of the raw water, while the concentrations of these by-products formed by chlorination of the XAD-8/XAD-4 mixture were closer to those in the raw water. As was observed for THMs, the addition of bromide to the mixed XAD-8/XAD-4 solution led to an increase in production of the brominated HAAs after chlorination. The chlorine demand of the summer RO and NF isolates was low, as were the concentrations of DBPs formed in these samples. In fact, the production of TOX, THMs and HAAs was only about half of the corresponding production in the raw water or the XAD-8/XAD-4 mixture. It is unlikely that these results were caused by analytical problems, because the results obtained on the same day for the XAD-8/ XAD-4 mixture were similar to those for the raw water, as expected based on prior work. The most likely explanation for the anomalous results appears to be incomplete dissolution of the lyophilized sample (and low reactivity of the undissolved NOM). Additional evidence for the erroneous DOC value includes a computed SUVA value that is well out of the expected range based on the strong relationship previously established between SUVA and DBP formation or chlorine demand (computed SUVA near 2.0 L/mg-m for both NF and RO isolates, versus an expected value between 4 and 5). This data set highlights the 209
problems with resolubilizing the non-desalted NOM isolates obtained from RO- and NFconcentrated waters.
UV AND FLUORESCENCE SPECTRA OF THE BLAVET NOM CONCENTRATES
UV and fluorescence spectra of the Blavet NOM concentrates were analyzed in order to investigate the influence of concentration and fractionation methods on its spectral parameters. The UV spectra of all samples were recorded at a DOC concentration of 10 mg/L. The UV spectra for selected isolates are presented in Figure 6.8. Only the data for λ>220 nm are shown, because in some samples (including almost all those from the summer sampling period) the absorbance was too high at λ < 220 nm to be measured with sufficient precision, possibly because of the presence of interfering species. The spectral parameters of the samples are compiled in Table 6.9.
3.0 2.5
XAD-8
Absorbance
NF 2.0 RO
1.5 1.0
XAD-4
0.5 0.0 225
250
275
300
325
350
375
400
Wavelength (nm)
Figure 6.8. Set of UV spectra of NOM concentrates from the Blavet River winter sampling period (10 mg DOC/L, cell length 5 cm) 210
Table 6.9 UV and fluorescence parameters for NOM samples concentrated from the Blavet River ∆ET (eV)
λmax (nm)
s.
w.
s.
w.
s.
nd
nd
Nd
nd
nd
nd
4.1
0.39
0.35
2.23
2.11
438
434.5
4.3
5.2
0.40
0.36
2.26
2.14
437
432
XAD-4
2.6
3.2
0.29
0.28
1.97
1.92
424
423
RO
4.0
5.2
0.38
0.35
2.20
2.11
428.5
NF
4.3
4.2
0.39
0.35
2.24
2.13
431
RO + IX
4.5
0.45
2.42
438
NF + IX
4.6
0.44
2.41
434
SUVA254 (L/(mg-m))
A350/A380
Sample
w.
s.
w.
raw water
5.1
4.8
XAD-8/XAD-4 4.4 XAD-8
RO + HCl
4.1
0.43
2.35
NF + HCl.
4.5
0.43
2.35
NF + XAD-4
4.3
0.34
2.08
427
The SUVA254 values for the raw Blavet River water from both the winter and summer sampling periods are remarkably high. In fact, these values are larger than the corresponding value in almost all of the NOM fractions from these samples or the NOM fractions from the Suwannee or South Platte. In light of the widely accepted correlation between the aromaticity of NOM and its SUVA254 value, one would conclude that the aromaticity of the NOM in the Blavet is very high. However, the 13C-NMR data suggest that the aromaticity of the desalted isolates is in the range 21 to 28% for the winter samples and 15 to 20% during the summer (Table 6.6), which are not unusually high. Also, as a point of comparison, the aromaticities of the Suwannee River fractions with similar SUVA254 (the uHA and HPOA fractions) were ~27 and 31%, respectively. Pyr-GC-MS and other compound- and structure-specific analyses described in Chapter 5 and the preceding sections of this chapter suggest that NOM from the Blavet Rivers is very similar to that from the Suwannee. Thus, it seems unlikely that the difference in SUVA254 values of the raw waters can be attributed to any specific group of 211
high-absorbance chemicals that are present in the Blavet and absent in Suwannee. Three explanations for the difference that are considered to be more likely include: (1) the aromaticity of the Blavet NOM is underestimated by integration of the 13C-NMR spectra; (2) the isolation and fractionation procedures altered the NOM so that the raw water could not be represented as the sum of the fractions; or (3) the DOC of the sample was greater than the reported value, due to some analytical problem. With respect to the first of the proposed explanations, it is important to recall the limitations of
13
C-NMR as a quantitative tool for estimating the structural identity of
carbon atoms in NOM, as noted in Chapter 2. Correlations of SUVA254 and other spectral parameters with the NOM aromaticity evaluated using 13C-NMR data acquired at a 5 ms contact time may be considerably better than those derived using a 1-ms contact time (which was used in the current study). The fact that the SUVA254 values for raw Blavet River water prior to NOM concentration were higher than those of virtually all of the sub-fractions isolated from those samples is even more difficult to explain than the anomalously high SUVA254 of the raw water samples. Based on prior studies by the authors and others, the SUVA254 values of the RO and NF concentrates and, even more so, the XAD-8 and XAD-8/ XAD-4 fractions, were expected to be higher than the corresponding values in the raw waters, since low-absorbing NOM molecules are selectively lost during these processing steps. The fact that the opposite result was found for the Blavet concentrates might signify that the concentration and desalting techniques altered the properties of the NOM, a possibility that is supported to some extent by the
13
C-NMR and Pyr-GC-MS data
discussed in the previous sections of this chapter. A third possible explanation for the results is that the measured DOC concentration of the raw water samples was less than the true value, so that the correct value of SUVA254 was smaller than reported. While analytical or human error is always a possibility, it seems odd that such errors should be manifested in the two raw water samples, while the results for all the other samples fell within commonly expected
212
ranges, and that no evidence of such errors were apparent in the QA-QC checks conducted throughout the project. Nevertheless, this possibility cannot be ruled out. NOM concentration and fractionation method(s), however gentle they may be, are intended to alter the chemical composition of a very complex solution. As a result, they are almost certain to alter NOM composition in unforeseen ways that one hopes are not critical to the evaluation being carried out. The techniques that have been developed to date for this purpose seem to be quite successful at meeting this goal. Nevertheless, it is clear that greater attention needs to be paid to development of internal and/or easily performed checks that can be used to provide an ongoing assessment of whether the techniques are achieving the desired result (**MB add to exec summary). One such check should involve careful comparison of characteristics of the NOM (e.g., SUVA254) before and after each processing step. Alternatively, or in parallel, development and validation of in situ NOM characterization methods that do not involve any concentration steps would be extremely valuable. The uncertainty about the source of the anomaly in the SUVA values of the Blavet raw water and fractions is testimony to the complexity of the problem and the need for constant vigilance in all aspects of the isolationfractionation-characterization process.
CORRELATIONS BETWEEN SPECTRAL AND STRUCTURAL CHARACTERISTICS OF NOM IN BLAVET RIVER FRACTIONS
The next section of this chapter focuses on evaluation of correlations between the spectral parameters of the Blavet isolates and their chemical characterization. In particular, various data associated with the fluorescence and UV absorbance spectra of the of NOM from the Blavet samples will be discussed. As noted above, isolation procedures are bound to alter NOM. Analysis of the chemistry of NOM by in situ measurements provides a valuable adjunct to the information that can be acquired by application of sophisticated techniques to the concentrated and isolated samples.
213
Generally, the A350/A280 ratios and ET band half-widths (∆ΕΤ) of the samples follow the same patterns as SUVA254. The correlation between SUVA254 and ∆ET for the samples is shown in Figure 6.9. This correlation (R2 = 0.63) probably reflects the intrinsic interdependence between the aromaticity and molecular weight of NOM, on one hand, and the inter-chromophore interactions on the other. One possible advantage of using ∆ΕΤ rather than SUVA254 as an indicator of these properties is that ∆ΕΤ can be estimated directly from the UV absorbance spectrum of the sample, without simultaneous analysis of the DOC.
2
R = 0.63
ET band halfwidth (eV)
2.40
2.25
2.10
1.95
1.80 2.0
3.0
4.0
SUVA254 (L/mg·m)
Figure 6.9. Correlation between SUVA254 and ∆ΕΤ for NOM concentrates from the Blavet River (data shown are for both the summer and winter sampling periods) The fluorescence emission spectra of the NOM concentrates are sensitive to the concentration method employed (Figure 6.10). The emission intensity of the XAD-4 sample is substantially higher than that of any of the other samples, and the maximum of its emission band is at shorter wavelengths. Similar results have been reported in the literature for the transphilic fractions of NOM (Donard et al. 1987, Ewald et al. 1992). 214
The increased fluorescence intensity of the XAD-4 sample is almost certainly related to its lower molecular weight (compared to the other NOM in the other samples), which decreases the rate of radiationless losses of excitation. 240
Emission intensity
210
XAD-4
180
XAD-8
150
NF
120
RO
90 60 30 0 375
400
425
450
475
500
525
550
Wavelength (nm)
Figure 6.10. Selected fluorescence emission spectra of NOM concentrates from the Blavet River, winter sampling period (10 mg DOC/L, cell length 1 cm, excitation at 320 nm) The emission spectrum of the XAD-4 (transphilic) acids is substantially blueshifted compared with that of the RO isolate, whereas the maximum in the XAD-8 concentrates is slightly red-shifted. This comparison is more obvious when the emission spectra are normalized, as in Figure 6.11. The normalized emission spectra of the RO and NF concentrates are virtually identical. The position of the maximum in the emission spectrum of the different concentrates is correlated with ∆ET in the corresponding UV absorbance spectrum, as shown in Figure 6.12. As the ET band becomes broader, the emission maximum exhibits a red shift.
215
Normalized fluorescence intensity
1.00 XAD-8 0.75 RO,NF 0.50
XAD-4
0.25
0.00 375
400
425
450
475
500
525
550
575
600
Wavelength (nm)
Figure 6.11. Normalized fluorescence emission spectra of selected NOM concentrates from the Blavet River winter sample (10 mg DOC/L, cell length 1 cm, excitation at 320 nm)
216
440
λ max (nm)
435
2
430
R = 0.64
425
420 1.90
2.00
2.10
2.20
2.30
2.40
2.50
ET band halfwidth (eV)
Figure 6.12. Correlation of the ET band half-width and the wavelength of maximum fluorescence emission intensity (NOM samples concentrated from the Blavet River, winter and summer sampling periods) Based on these and literature data, as well as theoretical considerations, it is reasonable to conclude that the UV and fluorescence spectra of NOM isolates are mainly determined by the aromaticities and MW distribution of the humic species comprising the NOM. In the next section, we explore whether chemical properties other than aromatic content might affect the spectral response of NOM in consistent ways. Potential correlations between the elemental composition of NOM and its spectral response were explored first. Only the low-ash samples (see Table 6.1), for which the absolute concentrations of elements and the C/O, C/N and C/H ratios were known most reliably, were included in the analysis. The carbon mass fraction (%C) in these samples was well correlated with the SUVA254 (R2 = 0.78) and even more strongly correlated with ∆ET (R2 = 0.90). The total carbon content of the NOM was also correlated with the percentage of aromatic carbon in the NOM concentrates (as determined by integration of the
13
C-NMR spectra over the range 110 to 160 ppm) and with the PHA carbon (as 217
determined from Pyr-GC-MS analysis) (Figure 6.13). Thus, the correlation between ∆ET for the Blavet concentrates and %C probably reflects the fact that the NOM concentration-isolation methods are likely to be most efficient at collecting high-MW and aromatic molecules; these molecules, in turn, are expected to be less oxygenated and have a high carbon content than the molecules that are collected less efficiently.
35 Pyr-GC-MS
30
13C NMR data 25
25 2
20
R = 0.84 2
R = 0.74
15
15
10 5
110-160 ppm aromatic carbon (%)
Polyhydroxyaromatic carbon (%)
35
5 41
42
43
44
45
46
47
48
Carbon in NOM (%)
Figure 6.13. Correlation between the percentage of carbon in low-ash NOM concentrates and percentages of aromatic carbon as determined through 13C-NMR and Pyr-GC-MS data. Blavet River samples (summer and winter sampling periods) Contrary to the case for carbon content, the nitrogen content of NOM did not correlate well with any of the major spectral parameters (SUVA254, ∆ET, λmax, and fluorescence intensity). Attempts were also made to correlate the spectral parameters of the concentrated NOM samples with the concentrations of specific groups of nitrogencontaining species. These attempts utilized data from all the samples, regardless of their ash content. It was hypothesized, for example, that the intensity of NOM fluorescence might be related to the total concentration of dissolved amino acids, since the amino 218
acids include highly fluorescent species such as tyrosine, tryptophan and phenylalanine. However, the correlation between the TDAA concentration and the intensity of fluorescence was relatively weak (R2 = 0.47), and no correlation could be found between the concentrations of proteins or amino sugars and any spectral parameter. Similarly, little evidence was found linking the spectra of the NOM samples and their carbohydrate contents. Not surprisingly, SUVA254 was correlated with the polyhydroxyaromatic carbon content, and slightly less strongly with the total aromatic carbon content of the NOM (Figure 6.14). The correlation of polyhydroxyaromatic carbon with ∆ET was similar or somewhat stronger than that with SUVA254. The position of the maximum in the emission spectra was sensitive to the percentage of aromatic carbon (Figure 6.16): as the concentration of aromatic carbon decreases, the emission maximum shifted toward shorter wavelengths. This result further supports the assumption that the UV and fluorescence spectra of the Blavet NOM concentrates are largely controlled by the concentration and chemical state of aromatic carbon they contain.
219
40
Aromatic carbon (%)
35 30 25 2
R = 0.50
20 15 10 5 0 2.0
2.5
3.0
3.5
4.0
4.5
5.0
SUVA254 (L/mg·m)
Figure 6.14. Correlation between the values of SUVA254 and the percentage of polyhydroxyaromatic carbon (Pyr-GC-MS data).
220
45
Aromatic carbon (%)
40 35 30 25 2
R = 0.73
20 15 10 5 0 1.90
1.95
2.00
2.05
2.10
2.15
2.20
2.25
2.30
ET band halfwidth (eV)
Figure 6.15. Correlation between ∆ET and the percentage of polyhydroxyaromatic carbon in the Blavet NOM fractions based on Pyr-GC-MS data.
221
Aromatic carbon (%)
30
25
20
15
2
R = 0.64
10
5 420
425
430
435
440
λ max (nm) Figure 6.16. Correlation between the position of maximum in the fluorescence emission spectra and the percentage of aromatic carbon based on 13C-NMR data.
CONCLUSIONS
The data presented in this chapter illustrate some of the complexity of NOM and its changes when it is isolated and fractionated. The conclusions drawn from the data relate to the NOM characteristics in a particular water supply (the Blavet River) and, more generally, to ways that fractionation-isolation may alter NOM. These conclusions are summarized below. NOM from the Blavet River was efficiently bound to XAD-8 resin (79 and 57% for the winter and summer samples, respectively), and its SUVA254 was comparable to or higher than that of the HPOA and uHA fractions of the Suwannee River NOM. NOM from the Suwannee is widely accepted to be highly humified and hydrophobic, perhaps even representing one extreme on a scale of NOM properties. The Blavet NOM appears to be similar to the Suwannee NOM in such properties. The differences in the properties 222
of the Blavet NOM between winter and summer are most notably associated with changes in small, biodegradable constituents such as TCAAs and TDAAs. However, these chemicals represent a minor fraction of the NOM in either season, so the overall similarity between Blavet and Suwannee NOM applies to both samples collected. One possible interpretation is that the highly modified NOM is roughly the same in the two seasons, and the samples differ primarily in the amounts of degradable molecules that ‘dilute’ the more highly modified molecules. The summer sample had a higher BDOC and was more hydrophilic: hydrophobic and transphilic NOM accounted for 57 and 21%, respectively, of the NOM in the summer, and 79 and 11% in the winter. In terms of this distribution, the NOM in the Blavet is more akin to Suwannee NOM in the summer than in the winter; disregarding losses during the fractionation, the hydrophobic and transphilic fractions accounted for roughly 60 and 21%, respectively, of the NOM in the Suwannee. Other conclusions derived from the data presented in this chapter are concerned with the performance of the NOM isolation-concentration methods. As expected, RO was the most efficient concentration and/or fractionation technique for retaining the TCAA and TDCA fractions of NOM. However, carbohydrates tended to adsorb on the RO membranes. The
13
C-NMR data also indicate that RO is probably the best method to
isolate the proteinaceous fraction of NOM. In our study, NF was noticeably less efficient than RO at retaining amino acids and carbohydrates, although paradoxically, it did retain considerable amounts of salts. In many cases, the desalting of RO and NF isolates is necessary before the chemical composition and properties of the collected NOM can be elucidated. However, the desalting and ion-exchange procedures commonly used for this purpose can alter the composition and properties of the NOM. For example, amino acids in the NF isolates were noticeably depleted by ion exchange. As a result, while cation exchange might be necessary for desalting NOM concentrates collected using membranes, it does have drawbacks and should be used only if a clear need for desalting exists.
223
Interactions between the inorganic and organic species in highly concentrated and lyophilized RO and NF isolates might alter the NOM due to the formation of sulfonates and, possibly, furanes that may be detected as aromatic carbon in
13
C-NMR
measurements. The irreversible alteration of NOM by lyophilization is also manifested in the impossibility of completely re-dissolving lyophilized NF and RO isolates.
224
CHAPTER 7
CORRELATIONS BETWEEN DATA FROM STRUCTURE-
SENSITIVE METHODS AND UV AND FLUORESCENCE SPECTROSCOPY
INTRODUCTION
This study indicates that the majority of the NOM in most water sources can be isolated, but that limitations exist on the extent and significance of NOM isolation. Even careful application of the most efficient isolation technique (evaporation) recovers less than 100% of the NOM in a sample. Furthermore, the NOM that can be isolated may be altered by any isolation process. Although approaches have been developed to limit the extent of NOM alteration during processing, the possibility always exists that as yet unrecognized mechanisms of NOM alteration are active in a given situation. Because of the inherent limitations of NOM isolation efficiency and the possibility of NOM alteration during processing, use of characterization methods that do not require NOM to be isolated and concentrated are attractive. However, the sensitivity and sample requirements of many available characterization methods do not allow those techniques to be used on unaltered NOM. A comparison of analytical methods that do and do not require extensive NOM pre-processing permits a researcher to judge whether NOM isolation is necessary for a particular purpose. Thus, one goal of this project was to evaluate the scope of information on NOM structure and properties that can be extracted using various analytical methods, alone and in combination. Given the wide range of situations in which information about NOM might be of interest, it is not possible to recommend a single method of characterization that would best for all applications. It is useful, nevertheless, to critically assess the information that can be obtained from different analytical procedures. Such an assessment is provided here. For the purposes of this discussion, the analytical techniques of interest are grouped into two categories. One category includes structure-sensitive methods such as 225
13
C-NMR and Pyr-GC-MS. In general, these analyses require advanced instrumentation,
skilled personnel, and, in many cases, substantial preconcentration of NOM. 13
C-NMR spectrometry has constituted a benchmark in NOM structural studies,
and there is support for the idea that Pyr-GC-MS spectrometry can become a second one. Interpretation of the data from both of these advanced methods cannot be as unambiguous as it might be for analysis of individual species. The complexity of NOM necessitates the use of aggregate parameters (e.g., estimates of organic functional groups based on the integration of the CPMAS 13C-NMR signal in assigned ranges of chemical shifts, or quantification of PHA, PR, AS and PS by comparing the intensity of peaks of signature functional groups in Pyr-GC-MS spectra). The precision of these integration assignment procedures is limited, and the results are semi-quantitative. The other category of NOM characterization methods includes UV and fluorescence spectroscopy, techniques that are easy to use but whose output is related to the composition and structure of NOM in a more complicated and ambiguous way. These techniques cannot quantify the presence of specific chemical functional groups as is accomplished using 13C-NMR and Pyr-GC-MS. At best, they can probe only some of the aggregate properties of NOM (e.g., aromaticity). However, the spectral measurements are highly precise, rapid and inexpensive, and they can be carried out in situ on unaltered NOM. UV and fluorescence data have been compared with results from
13
C-NMR and
Pyr-GC-MS methods in preceding sections of the report, but the comparisons were carried out separately for NOM samples derived from different sources. In this section, a more general discussion of these correlations is presented, based on pooled data from various sites and fractions. Because NOM is known to be at least somewhat site-specific, the correlations for the pooled data are bound to be weaker than those for more narrow data sets. However, the analysis might be useful for identifying relationships that deserve additional study. It might also provide some insight into the type of information that can be extracted by in situ analyses, as opposed to information that can be obtained only on isolated, concentrated and/or purified NOM. 226
THE MAJOR PARAMETERS OF UV AND FLUORESCENCE SPECTRA OF NOM
Data presented in this report indicate that several parameters of the UV absorbance and fluorescence spectra can be used to characterize NOM. The only spectral parameter that has been used extensively in previous studies is the specific absorbance at 254 nm (SUVA254), which is widely understood to be related to the aromaticity of NOM (e.g., Korshin et al. 1997b, Westerhoff et al. 1998, Edzwald et al. 1985). The half-width of the ET band (∆ET) has been introduced as another parameter of the UV spectra of NOM in this study. The value of this parameter is a complex function of the concentration, extent of activation, and mutual interactions among aromatic chromophores. However, it is easily estimated from the UV absorbance spectrum in a range where interferences from inorganics are minimal (280 - 350 nm), and it seems to be related to important characteristics of the NOM. The fluorescence emission spectra of NOM can also be recorded in a range of wavelengths (λ > 360 nm) where interferences are minimal. Although several parameters might be used to characterize NOM fluorescence emission (e.g., the wavelength the maximum emission intensity (λmax), the fluorescence yield at λmax, the half-width of the fluorescence spectrum, and others), interpretation of many of these parameters is complex and speculative, and only λmax will be discussed here.
CORRELATIONS BETWEEN 13C-NMR SPECTROSCOPY AND THE MAJOR PARAMETERS OF UV AND FLUORESCENCE EMISSION SPECTRA
The ranges of several parameters of interest from 13C-NMR, UV absorbance, and fluorescence analyses NOM from the Suwannee, South Platte, Tolt and Blavet Rivers are presented in Table 7.1. The data ranges shown are from analysis of both concentrated NOM and isolated NOM fractions.
227
Table 7.1 Summary of results from 13C-NMR, UV absorbance and fluorescence emission analysis for 27 samples of concentrated or fractionated NOM from the Blavet, South Platte, Suwannee and Tolt Rivers. Parameter
range
Aromaticity (%)
3 to 34
Total carboxyl carbon (%)
11 to 27
Specific absorbance at 254 nm, SUVA254 (L/(mg-m))
0.6 to 6.2
ET band half-width, ∆ET (eV)
1.52 to 2.51
Position of emission maximum, λmax (nm)
416 to 463
The SUVA254 values for the samples correspond to a range from virtually no light absorbance at 254 nm to highly absorbing samples (that were also highly colored). Values of ∆ET have not been reported previously, so the range for these samples cannot be compared with literature data. The range of emission intensities is wider than those reported in previous research. The relationships between the spectral parameters and aromaticity is shown in Figures 7.1 and 7.2. SUVA254 is correlated with aromaticity, ∆ET, and λmax. The correlation coefficients for these relationships and several others are summarized in Table 7.2.
228
7.0
SUVA254 (L/mg·m)
6.0 5.0
2.0
4.0
1.5 2
3.0
R = 0.73
1.0
2.0 SUVA254 ET band halfwidth
1.0
0.5
0.0
ET band halfwidth (eV)
2.5
2
R = 0.55
0.0 0
5
10
15
20
25
30
35
Aromaticity (%)
Figure 7.1. Correlation between the aromaticity of NOM estimated using
13
C-NMR
spectroscopy and corresponding values of SUVA254 and ∆ET. Data for Suwannee and South Platte River fractions and Tolt and Blavet River NOM concentrates
229
470 460 2
R = 0.61
λ max (nm)
450 440 430 420 410 0
5
10
15
20
25
30
35
Aromaticity (%)
Figure 7.2. Data for Suwannee and South Platte River fractions and Tolt and Blavet River NOM concentrates
230
Table 7.2 External and internal correlations for the UV and fluorescence spectral parameters and the data of CPMAS 13C-NMR spectroscopy for 27 samples from four sources (Blavet, South Platte, Suwannee and Tolt Rivers) 13
UV/fluorescence parameter
Linear R2 value
Aromaticity
SUVA254
0.72
Aromaticity
∆ET
0.57
Aromaticity
λmax
0.61
Total carboxyl carbon
SUVA254, ∆ET, λmax
0.01, 0.06, 0.03
Anomeric carbon
SUVA254, ∆ET, λmax
0.04, 0.04, 0.00
Aromatic/carboxyl ratio
SUVA254
0.65
Aromatic/carboxyl ratio
∆ET
0.56
Aromatic/carboxyl ratio
λmax
0.60
Aromatic/anomeric ratio
SUVA254
0.44
Aromatic/anomeric ratio
∆ET
0.39
Aromatic/anomeric ratio
λmax
0.28
C-NMR parameter
Correlations among CPMAS 13C-NMR parameters Aromaticity
Total carboxylic carbon
0.00
Aromaticity
Anomeric carbon
0.01
Total carboxylic carbon
Anomeric carbon
0.00
Correlations among UV/fluorescence parameters SUVA254
∆ET
0.72
SUVA254
λmax
0.46
∆ET
λmax
0.60
In addition to their association with the aromaticity of NOM, the SUVA254, ∆ET and λmax values are all inter-correlated. This result reflects the fact that all of these values are manifestations of the excitation and relaxation of aromatic units caused by irradiation of NOM with UV or visible light. As a result, any or all of these parameters may be used to predict and monitor the state and alteration of aromatic moieties in NOM. Though they 231
are partially correlated with one another, the values of SUVA254, ∆ET and λmax are also partially independent of one another. Combinations of these types of data can potentially enhance our ability to monitor NOM characteristics and reactions.
CORRELATIONS BETWEEN PYR-GC-MS SPECTROSCOPY AND THE MAJOR PARAMETERS OF UV AND FLUORESCENCE EMISSION SPECTRA.
Thirteen samples were analyzed using Pyr-GC-MS, and the results of these analyses can be compared with the corresponding SUVA254, ∆ET and λmax values. Two samples that were deemed extreme outliers due to their very high concentration of nitrogenous species (the Suwannee River HPIB and South Platte River TPIN fractions) that were excluded from the analysis of
13
C-NMR data are included in this discussion,
since Pyr-GC-MS provides a direct estimate of the concentration of nitrogenous species simultaneously with those for aromatic and carbohydrate units. Ranges of the parameters of interest are summarized in Table 7.3, and the corresponding R2 coefficients are given in Table 7.4. Table 7.3 Ranges of major parameters of Pyr-GC-MS, UV absorbance and fluorescence emission analyses of 13 samples from three sources (Blavet, South Platte and Suwannee Rivers) Parameter
Range
Polyhydroxyaromatic carbon, PHA (%)
7 to 34
Unsubstituted aromatic carbon, UA (%)
8 to 30
Total aromatic carbon, PHA+UA (%)
24 to 54
Polysaccharides, PS (%)
3 to 28
Amino sugars, AS (%)
3 to 34
Total saccharides, PS+AS (%)
15 to 46
Specific absorbance at 254 nm SUVA254 (L/(mg-m))
0.6 to 4.6
ET band half-width, ∆ET (eV)
1.62 to 2.24
Wavelength of emission maximum, λmax (nm)
400 to 438
232
Table 7.4 Correlations of UV and fluorescence spectral parameters with Pyr-GC-MS results for 13 samples from the Blavet, South Platte and Suwannee Rivers Pyr-GC-MS parameter
UV/fluorescence parameter
Linear R2 value
External correlations, aromatic groups PHA
SUVA254
0.37
PHA
∆ET
0.41
PHA
λmax
0.65
External correlations, nitrogen-containing groups Total nitrogenous species
λmax
0.55
AS
λmax
0.46
Proteins
λmax
0.39
Correlations for Pyr-GC-MS vs. 13C NMR data PHA
Aromaticity
0.44
PS
Anomeric carbon
0.03
AS
Anomeric carbon
0.12
Total saccharides
Anomeric carbon
0.00
Several results shown in Table 7.4 are worth noting. First, the SUVA254 and ∆ET values are correlated only with the concentration of PHA units, and then with R2 values of only 0.37 to 0.41. By contrast, the emission of NOM is affected by both PHA and nitrogenous species. The position of the fluorescence emission maximum is related to the percentage of PHA in the NOM (R2 = 0.65) and to the amino sugar (AS) content (R2 = 0.46). The relationship between the position of the emission maximum and the concentration of nitrogenous species that was first noticed in the studies of the South Platte NOM fractions (Chapter 5) is reinforced by the results presented here.
The correlation between the 13C-NMR and Pyr-GC-MS data is generally weak; R2 is >0.2 only for the relationship between the aromaticity and Pyr-GC-MS 233
polyhydroxyaromatic carbon (R2 = 0.44). There is little correlation between the anomeric carbon and the relevant Pyr-GC-MS carbohydrate data. Thus, it appears that
13
C-NMR
and Pyr-GC-MS may probe somewhat different moieties, or that their calibration and validation need to be developed in considerably more detail. Note, however, that the absolute concentrations of various functional groups cannot be directly compared between the two methods, since the integration is done over all organic carbon atoms in the sample with 13C-NMR, but only over an indeterminate portion of the organic carbon with Pyr-GC-MS.
CONCLUSIONS
Data from
13
C-NMR, Pyr-GC-MS, UV absorbance and fluorescence emission
analyses are correlated in important ways, primarily because all of these analytical techniques are sensitive to aromatic carbon in NOM molecules. Specifically, SUVA254, ∆ET, and λmax are all correlated with the aromaticity of NOM quantified by either 13
C-NMR or Pyr-GC-MS. As opposed to SUVA254 and ∆ET (which are affected solely by
the aromatic carbon), λmax is related to both the aromaticity and, in some cases, to the presence of nitrogen-containing species. While the
13
C-NMR aromaticity is thought to be proportional to the total
concentration of aromatic carbon in NOM, the only parameter of the Pyr-GC-MS analysis that correlates strongly with spectral parameters is the polyhydroxyaromatic carbon in the sample. The compatibility and interpretation of these two methods should be investigated in more detail. All the spectral parameters (SUVA254, ∆ET, λmax) can be used to monitor the concentration and transformation of aromatic carbon-rich NOM samples in situ without preconcentration. At present, ∆ET and λmax are not widely used for this purpose, but use of SUVA254 is relatively extensive. Information from analysis of these parameters is somewhat overlapping, but also complementary. The values of ∆ET and λmax can be
234
determined directly from spectral analysis, whereas determination of SUVA254 requires spectral data and analysis for DOC. The strength of the correlations between spectral parameters and the aromaticity (from
13
C-NMR) and/or PHA concentration (from Pyr-GC-MS) ranges from weak to
moderately strong. For example, the R2 value for the PHA (Pyr/GC-MS data) vs. SUVA254 was 0.37, while the R2 for
13
C aromatic carbon vs. SUVA254 was 0.72. Given
the wide range of properties of the NOM samples studied, the amount of scatter observed in these correlations is not surprising. It is possible that, for a series of NOM samples derived from the same source but subjected to the types of physico-chemical alteration expected in water treatment processes, the correlations between the structural and spectral properties of NOM will be substantially stronger.
235
CHAPTER 8
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
COMPARISON OF NOM CONCENTRATION METHODS
This research assessed the efficiency and practicality with which NOM could be concentrated and isolated, using both existing and novel methods (RO, NF, XAD-8/XAD-4, iron-oxide-coated adsorbent media). The concentration methods tested are compared in Table 8.1.
236
Table 8.1 Comparison of NOM concentration methods Evaporation
RO
NF
IOCS
IOCO
XAD-8
XAD-8/ XAD-4
DOC retention efficiency, %
80-90 (100*)
87-98
77-96
70-80
20-50
† 45-50
† 60-75
UV254 retention efficiency (%)
100*
100
90-100
85-95
20-70
50-60
80-90
Benefits
Theoretically complete recovery of non-volatile organic
High efficiency, high speed
High efficiency, somewhat less salts compared with RO, high speed
High efficiency, lack of irreversible adsorption
No pH adjustment is necessary
Traditionally used method, lack of accumulation of salts
Improved efficiency compared with XAD-8, lack of accumulation of salts
Problems
Very laborintensive, precipitation of salts, desalting is necessary
Entrapment/ adsorption of organics by membranes, loss of low molecular weight organics, desalting is necessary
Entrapment/ adsorption of organics by membranes, loss of low molecular weight organics, desalting is necessary
Less efficiency compared with RO, loss of low molecular weight organics, coadsorption of sulfate, desalting may be necessary
Low efficiency, limited pH range, possible co- adsorption of sulfate, desalting may be necessary
Laborious preparation and cleaning, hydrophilic NOM is lost
Laborious preparation and cleaning, irreversible adsorption on XAD-4, loss of low molecular weight organics
* maximum theoretical efficiency † typical data are shown, performance widely varies; in some waters 80% of DOC may be retained by XAD-8 while in others TPH>HPI>uHPI. NOM attributed to each gradation of hydrophobicity can be further divided into acidic, neutral and basic fractions. 244
The acidic fractions of NOM are characterized by slightly lower C/O ratios than the basic and neutral fractions, reinforcing the hypothesis that most of the acidic groups are carboxylic. The C/N and C/H ratios of the neutral fractions are generally lower than those of the acid fractions, consistent with the greater aliphatic character of the neutral NOM fractions. The base fractions typically have the highest nitrogen content, consistent with the expectation that most organic bases are related to amino and amide functionality. In general, the C/O, C/N, and C/H ratios decrease with increasing hydrophilic character for both acid and neutral fractions (the isolated ultrahydrophilic acids do not fit this trend because they were methylated as part of the isolation procedure). Put another way, the higher the hydrophilic character, the higher the proportions of oxygen, nitrogen and hydrogen in the NOM fraction, i.e., the more aliphatic it is. South Platte NOM is derived from both allochthonous inputs in which aromatic and acid constituents are depleted by mineral solubility controls, and autochthonous inputs from phytoplankton and bacteria. These autochthonous processes contribute proteins, polysaccharides, and amino sugars that increase the hydrophilic character of this NOM. The nitrogen content of the TPHN fraction of the South Platte River is very high, consistent with the suggestion, based on its FTIR and 13C-NMR spectra, that this fraction is highly proteinaceous. The Suwannee NOM is predominantly hydrophobic and is thought to be primarily allochthonous. It is derived from tannins and lignins that give rise to highly colored humic and fulvic acids of high poly-phenol content and low nitrogen content. The C/N ratios of the Suwannee River fractions are greater than these ratios in the corresponding fractions in the South Platte River. In almost all cases, the various NOM characterization methods gave supporting and complementary rather than contradictory information. Similarities and differrences between Suwannee River and South Platte River NOM, based on the results of various characterization methods, can be summarized as follows. •
DOC Fractionation. Suwannee River NOM is dominated by hydrophobic acids:
hydrophilic and transphilic species contribute much more to the South Platte River NOM. 245
•
Elemental Analyses. The C/N ratios of the Suwannee River fractions are greater than
these ratios in the corresponding fractions in the South Platte River, reflecting the greater allochthonous character of Suwannee River NOM. C/H, C/N, and C/O ratios decrease as the hydrophilic character of the NOM fraction increases. •
FTIR Spectra. NOM fractions can be isolated that are free of inorganic constituents
except for minor amounts of silica in certain fractions. Almost all of the NOM fractions have carboxylic acid peaks, and amide peaks from proteins and nacetylamino sugars are especially prominent in certain transphilic and hydrophilic NOM fractions from the South Platte River. •
13
C-NMR Spectra. As the hydrophilic character of the NOM fraction increases, the
NMR spectra have smaller aliphatic and aromatic hydrocarbon peaks and larger carbohydrate and acid or amide peaks. Suwannee River NOM fractions have greater aromatic carbon and phenol contents than corresponding fractions from the South Platte River. C-N peaks from amines and amides are evident in base, hydrophilic neutral, and transphilic neutral fractions, especially in South Platte River NOM, and the methyl peak from N-acetyl amino sugars is readily detected. •
Dissolved Amino Acids and Carbohydrates. TDCA and TDAA concentrations are
greater in the South Platte River than in the Suwannee River, but they account for only a small part of the DOC in both waters. TDAA accounts for the majority of nitrogen in the hydrophobic fractions of both waters, but only accounts for 5-20% of the nitrogen in the other fractions. Glucose is the dominate sugar in most NOM fractions. Large ornithine contents distinguish Suwannee River NOM from South Platte River NOM. •
Pyrolysis Gas Chromatography-Mass Spectrometry. Hydrophobic neutral NOM is
characterized by fatty acid fragments; hydrophobic acid NOM is characterized by polyhydroxyaromatic fragments typical of humic substances; and transphilic and hydrophilic NOM fractions are characterized by fragments indicative of carbohydrates, proteins, and amino sugars. Much of the nitrogen in the transphilic and hydrophilic fractions that cannot be attributed to TDAAs is present as amino 246
sugars. The transphilic neutral fraction from the South Platte River is very high in protein content. •
Ultraviolet Spectrometry. Specific UV absorbance (SUVA) varied substantially in
Suwannee River NOM fractions, decreasing in the order HA > HPO-NOM > TPHNOM > HPI-NOM. Similar variations were found in South Platte River NOM fractions, but SUVA was much lower than for corresponding fractions from the Suwannee River. SUVA was well correlated with the width of the ET spectral band for NOM fractions from both water samples. •
Fluorescence Spectrometry. For the Suwannee River NOM fractions, fluorescence
emission yields increase and the wavelength of maximum emission shifts to lower values as the hydrophilic character of the NOM increases. This trend is indicative of a decrease in the average molecular weight species as hydrophilic character increases. Fluorescence data for South Platte River NOM fractions indicate generally smaller molecular weights than for corresponding Suwannee River NOM fractions, but molecular weights tended to increase as the hydrophilic character of the fraction increased. The protein-rich fractions from both NOM sources exhibit strong fluorescence that is blue-shifted compared with that of highly aromatic fractions. •
Coagulability. Acid NOM fractions were better removed by alum at pH 6.5 than
neutral or base fractions. Suwannee River NOM was better removed by coagulation than South Platte River NOM. Moderate correlations were found between SUVA and DOC removal, and better correlations were found between SUVA and A254 removal. •
Characterization Correlations. The carbon content of NOM fractions was inversely
correlated with the carbohydrate content (13C-NMR), and oxygen content was directly correlated with carboxylic acids and/or amides (13C-NMR). The protein content of NOM (based on pyr-GC-MS analysis) was correlated with TDAA content. For Suwannee River NOM fractions, aromatic carbon content (13C-NMR) was directly correlated with SUVA, ∆ET, PHA fragments, and the fluorescence emission maximum; however, for the South Platte River NOM fraction, aromatic carbon content correlated only with PHA fragments which indicates the importance of 247
phenolic aromatics (as compared with total aromatic carbon) for ultraviolet and fluorescence spectral measurements.
SEASONAL CHANGES OF NOM
Seasonal changes of NOM were studied using NOM extracted from the Blavet River during summer and winter sampling periods. The NOM was more hydrophobic in the winter than in the summer (hydrophobic and transphilic NOM accounted for 57 and 21% of the DOC, respectively, in the summer and was 79 and 11% in the winter). In terms of its hydrophobic-transphilic-hydrophilic composition, the Blavet NOM in the summer was comparable with Suwannee River NOM. The seasonal changes in the properties of the Blavet NOM were mainly associated with a considerable increase in TCAA and TDAA in the summer sample. The properties of the humic component of Blavet NOM were not significantly affected by seasonal processes.
EFFECTS OF NOM ISOLATION AND CONCENTRATION ON DBP PRECURSORS
NOM chlorination studies showed that the membrane isolates typically formed less DBPs than the NOM in the raw water did. However, it is possible that this result was an artifact related to the difficulty of re-dissolving lyophilized membrane isolates. XAD isolates of NOM behaved very similarly to the NOM in the raw water in this respect. Experiments with IOCS isolates carried out using the method of differential UV spectroscopy (published in detail elsewhere) showed that in terms of halogenation these isolates were virtually identical with the raw water. Therefore, it is concluded that XAD resins and IOCS are satisfactory media for isolating the majority of the DBP precursors in natural samples. Nevertheless, it must be acknowledged that aliphatic structures, whose retention by non-membrane methods is not necessarily efficient, can affect the formation of halogenated disinfection by-products in some cases.
248
Chlorinated DBP yields were greater in isolated Suwannee River NOM fractions than in the corresponding fractions from the South Platte River. For Suwannee River NOM fractions, THM and TCAA yields were similar, but chloroform was the major byproduct from chlorination of South Platte River NOM fractions. It is hypothesized that the more hydrophilic nature of the South Platte NOM is responsible for this difference. Proteinaceous NOM fractions (e.g., Suwannee HPIB, South Platte HPIN and TPHN) yield considerably more DCAA than do other fractions, showing the importance of nitrogenous units in the formation of this DBP species. SUVA254 correlated well (and aromatic carbon somewhat less well) with THMFP, TOXFP, and TCAAFP, but did not correlate with DCAAFP. Some DBP production can occur by reactions of chlorine with NOM fractions that have negligible SUVA and aromaticity. DBP formation potentials of the NOM fractions are not well correlated with their SUVA254 values. Although DBP yields increase with SUVA254, the correlations are stronger when the data for each source are considered separately. This seems to indicate the importance of NOM origin and its intrinsic properties for halogenation.
COMPARISON OF EX SITU AND IN SITU METHODS FOR NOM CHARACTERIZATION
CPMAS 13C-NMR
CPMAS
13
C-NMR is a powerful analytical tool whose utility is well established
in NOM research. Use of this tool to quantify the abundance of dissimilar types of organic carbon in NOM is well established, and the data obtained using this technique are valuable in any study attempting to probe the reactivity of NOM. However, its use is contingent on the availability of dry, preferably desalted NOM samples from which inorganic carbonates have been removed. The acquisition of high quality spectra requires substantial time and highly sophisticated and expensive equipment. Continued 249
development of NMR spectrometers is increasing their sensitivity, improving spectral quantitation, and decreasing spectral acquisition time and cost. While the value of CPMAS
13
C-NMR data is undeniable, the technique might
have significant limitations that affect its precision in estimating the contribution of aromatic and carbonyl or carboxyl carbon to NOM. Specifically, the contribution of aromatic carbon may be underestimated by 20-40% (phenolic by 50%), and that of carbonyl (carboxyl, ester, amide) carbon by 30-50%. The aliphatic carbon may be overestimated by comparable amounts. Due to these limitations, 13C-NMR data should be viewed as providing semi-quantitative information about the NOM, much as is the case for Pyr-GC-MS data. NOM aromaticity values obtained using a 1 ms contact time should be considered as minimum estimates. Increasing the contact time in CPMAS
13
C-NMR
experiments from 1 to 5 ms is expected to improve the precision of the method. Correlations of SUVA254 and other spectral parameters with the NOM aromaticity evaluated using
13
C-NMR data acquired at a 5 ms contact time may be considerably
better than the corresponding correlations based on contact times of 1 ms.
Fourier Transform IR Spectroscopy (FTIR)
The presence of non-halide inorganic salts and silica (especially important in XAD-8 samples) may by ascertained by FTIR analysis of NOM isolates in KBr pellets. Important organic functional group information, especially the distinguishing of acids, amides, and esters, can be derived from FTIR spectra if interfering inorganic constituents have been removed. FTIR is presently a qualitative spectrometric method, but it may become a semi-quantitative method if it is calibrated with standards that are applicable to NOM composition. Recent developments in spectral software programs can now deconvolute complex and broad peaks that are typical of NOM FTIR spectra. It might be possible to apply these peak deconvolution programs to derive semi-quantitative information for FTIR spectra as was done for UV spectra in this study.
250
Pyrolysis GC-MS
Pyr-GC-MS analysis requires only a few milligrams of dry sample which can be obtained by lyophilization or rotary evaporation. Salts in the sample do not seem to interfere with the pyrolysis process and/or analysis of the fragments, but the available literature on this subject is limited and more studies are needed. Metals might affect the pyrolysis fragmentation. Similarly to
13
C-NMR, Pyr-GC-MS is a semi-quantitative analytical tool. Data
interpretation yields information about the distribution of molecules belonging to various biopolymer classes. The interpretation of NOM pyrochromatograms may be more subjective than that of NMR spectra, since only a portion of the pyrolysis fragments are generally used for the interpretation. Furthermore, the interpretation is complicated by the fact that some pyrolysis fragments (e.g., phenol, cresol) may have several origins, and others can be produced through secondary reactions (e.g., defunctionalization or cyclization). Previous studies of the NOM “fingerprint” defined by the pyrolysis fragments have led to the conclusion that proteins and carbohydrates are major constituents of humic substances. This inference is in conflict with evidence from spectroscopic data and other analyses for specific constituents in NOM, which suggest that proteins and carbohydrates are only minor constituents of NOM. Thus, more work is needed to reconcile these conflicting pieces of data. Despite the limitations imposed by the relative paucity of Pyr-GC-MS data in the literature, this technique is already a valuable tool for understanding of NOM. It provides researchers with a fingerprint of NOM that is distinct from that obtainable by other techniques. Our results show that each NOM fraction isolated from two source waters generated a unique pyrochromatogram, with some resemblance between fractions obtained using the same isolation protocols. In some cases, good correlations were found between NOM characteristics identified by other analytical techniques and Pyr-GC -MS chromatograms. Also, some aspects of the pyrochromatograms reinforce the interpretation of
13
C-NMR, FTIR and UV or fluorescence spectra, in particular with
respect to the presence of nitrogenous moieties in NOM. 251
Total Dissolved Amino Acids and Carbohydrates
Total dissolved amino acids and carbohydrates (TDAA and TDCA, respectively) represent the sum of the monomers analyzed by HPLC after acidic hydrolysis of NOM. The analyses can be conducted on natural or treated waters and on NOM isolates. Stateof-the-art instrumentation requires only a few milligrams of NOM for TDAA and TDCA analyses. To ensure adequate accuracy of trace-level analyses in natural waters, an ion exchange resin column, fluorimetric and pulsed amperometric HPLC detectors and in most cases dedicated equipment need to be used. The TDAA and TDCA analytical methods have been validated using known pure biopolymers. However, since the structure of NOM is not well understood, the reliability of the techniques for analyzing natural samples, especially with regard to the efficiency of the acidic hydrolysis step, is uncertain. Amino acids and carbohydrates comprise a relatively minor fraction of the DOC of surface waters. They exist as free amino acids and monosaccharides and in bound forms (as polypeptides or proteins, polysaccharides and/or as monomer units incorporated into humic substances). The concentrations of free amino acids and monosaccharides are typically much less than those of the corresponding bound forms. As a result, polysaccharides and bound (combined) amino acids represent the major fraction of TDCA and TDAA in natural waters. This research and literature data indicate that the distributions of monomeric species in TDAA and TDCA are not very sitespecific and/or indicative of the NOM generation processes. This was found to be case for the Suwannee and South Platte NOM fractions. The only exception is the concentration of ornithine, which seems to be a good indicator of microbial (**JP algal?) activity. In tandem with other methods such as FTIR, TDCA and TDAA data may provide useful information to support conclusions regarding the molecular structure of selected NOM fractions. For instance, the neutral NOM fractions were found to be richest in sugars, while the basic fractions were richest in amino acids. However, because amino acids and carbohydrates represent a small part of the bulk NOM, and analyses for these 252
components are difficult, the evaluation of TDAA and TDCA should not be a high priority for structural characterization of NOM. These analyses are certainly more useful in NOM biodegradability studies since the associated organic compounds are rapidly assimilated by microorganisms, and amino acids and carbohydrates represent a significant part of the BDOC fraction of NOM.
Elemental Analysis
Elemental analysis can be reliably conducted only on dry NOM ash-free isolates. If the ash content of a dry sample is > 5% by mass, significant errors occur in the evaluation of organic oxygen. Thus, elemental analysis may be a good indicator of the “purity” of NOM and the efficiency of isolation and desalting protocols. Elemental analysis does not appear to be a very specific NOM characterization tool, since the elemental composition of NOM is similar even for highly divergent fractions (e.g., Suwannee River HPOA and South Platte TPHA). Thus, elemental analyses are not sensitive enough to characterize or distinguish among NOM samples from various sources. However this analysis does provide important information such as elemental ratios that allow NOM fractions isolated from the same source to be compared. For instance, the C/O ratio is indicative of the concentration of oxygenated functional groups in the sample, the C/N ratio is indicative of the concentration of nitrogenous functional groups, and the C/H ratio is indicative of the degree of unsaturation of the NOM. The C/N ratio also seems to be a good indicator of the extent to which a given sample of NOM is derived from autochthonous material.
UV Spectroscopy
The research led to several important findings relevant to the use of UV spectroscopy in NOM-related research and in the water treatment practice. In this report, only the data of conventional UV spectroscopy are discussed. The fundamentals and 253
applications of differential UV spectroscopy for studying NOM reactions are discussed in separate publications (Li et al. 1998, Korshin et al. 1997a, 1996). Of all types of organic carbon in NOM, only the aromatic moiety in NOM has been unambiguously shown to affect its UV absorbance. SUVA254 is a good indicator of NOM aromaticity quantified by either CPMAS
13
C-NMR or by Pyr-GC-MS. A
noticeable exception from this correlation was found with the Blavet isolates, which have a high SUVA254 and only moderate aromaticity. The probable reason for this deviation is an underestimate of the aromaticity based on the improvement in the
13
13
C-NMR data. It is expected that
C-NMR data acquisition methods in NOM research will improve
these relationships. An auxiliary hypothesis that nitrogen bases (e.g., purine, pirimidine) and tryptophane associated with algal activity might affect SUVA254 has been proposed but has not been experimentally tested. SUVA254 is not the only indicator of intrinsic NOM properties that can be probed by UV spectroscopy. An alternative parameter is the width of the composite electrontransfer band (∆ET), which is correlated to both NOM aromaticity and, probably, its molecular weight. The manifestations of the latter in UV spectra of NOM need to be investigated in more detail. Other information that can be derived from UV spectral analysis of NOM might also be useful. For instance, the A252/A202 ratio might be indicative of the extent of activation of aromatic units. This and other parameters have been associated with NOM coagulability, and the corresponding data are discussed elsewhere (Korshin et al. 1996). Although UV spectroscopy does not convey information on abundance of specific types of organic carbon of structural units in NOM, it is a powerful tool for predicting NOM reactivity and for probing NOM reactions in situ. It is especially efficient in water with high concentrations of humic species. On the other hand, this analysis is not sensitive to the presence of some important groups of NOM molecules, such as those that are responsible for the BDOC. Further studies of the UV spectra of NOM and their changes in response to various physico-chemical processes used in the potable water industry can enhance the usefulness of UV absorbance to the drinking water community. 254
Advances in this area might require the use of more sophisticated instrumentation for collection and analysis of UV (and fluorescence) spectra than that deemed to be adequate at present.
Fluorescence
NOM fluorescence is an extremely sensitive method that permits investigating NOM in situ at DOC concentrations < 1 mg/L. The current study supported the widely accepted hypothesis that the fluorescence emission of NOM is predominantly governed by aromatic functionality in NOM molecules. However, in several important cases, the nitrogenous fluorescing species associated with proteins and free amino acids also contribute noticeably to the emission. In those cases, the emission is blue-shifted compared with that of predominantly aromatic NOM. Thus, NOM fluorescence may be useful as an in situ probe to investigate the predominance of aromatic or biopolymeric species in NOM and therefore may be employed to monitor and predict biological processes. It is also clear that the fluorescence emission of aromatic units in NOM molecules is substantially affected by their molecular weight and conformation. This fact may allow fluorescence to be used to track the reactions of NOM in water treatment processes or, alternatively, to probe the origin of NOM and to probe mixing processes in water distribution systems, if sources with dissimilar NOM are blended. The amount of high quality data currently available on the relationship between average molecular weight and/or conformation of NOM and the intensity and shape of emission spectra does not permit adequate quantification of this relationship. However, these effects are strong, and they merit further state-of-the-art exploration. There is also no unified mathematical theory of NOM emission. These and other issues need to be addressed in order for this method to be used at its full potential in NOM-related research and practice.
255
Relationship Between Data From in Situ and ex Situ Analyses
Data from
13
C-NMR, Pyr-GC-MS, UV absorbance and fluorescence emission
analyses (parameterized by SUVA254, ∆ET, and λmax) are correlated, since all of the analytical tools are sensitive to aromatic structures in NOM. Correlations between spectral parameters and the 13C-NMR aromaticity and/or PHA concentrations as inferred from Pyr-GC-MS data are noticeable but not exceedingly strong. Given the wide range of the properties of the NOM samples and the semi-quantitative nature of the aromaticity or PHA estimates by 13C-NMR and Pyr-GC-MS, the scatter observed in these correlations is not surprising. We believe that SUVA254, ∆ET, and λmax might be useful for monitoring the concentration and transformations of the aromatic moieties of NOM in situ. At present, ∆ET and λmax are not widely used for this purpose, but use of SUVA254 is extensive. The use of other proposed spectral parameters may augment the predictive capabilities of UV spectroscopy. No functionality other than the aromatic and/or PHA moiety seems to affect the UV spectrum of NOM. However, the fluorescence emission (quantified by λmax) is sensitive to both the aromaticity and, in some cases, the presence of nitrogen-containing species in NOM. The abundance and reactions of fluorescing nitrogenous moieties in NOM may also potentially be tracked in situ by fluorescence spectroscopy, but substantially more research is necessary to achieve this goal. The development of numerical methods to process UV and fluorescence spectra of NOM and improvements in the precision and interpretability of CPMAS
13
C-NMR and Pyr-GC-MS spectra will
enhance the versatility and predictive capacity of both in situ and ex situ methods.
256
APPENDIX
257
REFERENCES
Aiken, G.R., D.M.McKnight, K.A.Thorn, E.M.Thurman. 1992. Isolation of Hydrophilic Organic Acids from Water using Nonionic Macroporous Resins. Org. Geochem., 18: 567-573.
Aiken, G.R., D.McKnight, R.Wershaw, L.Miller. 1991. Evidence for Diffusion of Aquatic Fulvic Acid from the Sediments of Lake Fryxell, Antarctica. In Organic
Substances and Sediments in Water, V. 1, Humics and Soils (R. A. Baker, ed.) Lewis Publishers, Chelsea, Michigan, pp. 75-88. Aiken, G.R., J.A.Leenheer. 1993. Isolation and Chemical Characterization of Dissolved and Colloidal Organic Matter. Chemistry and Ecology, 8: 135-151. Aiken, G.R, E.M.Thurman, R.L.Malcolm, H.F.Walton. 1979. Comparison of XAD Macroporous Resins for the Concentration of Fulvic Acid from Aqueous Solution: Anal Chem. 51 (11): 1799- 1803. Alamany, L.B., D.M.Grant, R.J.Pugmire, T.D.Alger, K.W.Zilm. 1983. Cross Polarization and Magic Angle Spinning NMR Spectra of Model Organic Compounds. 2. Molecules of Low or Remote Protonation. J. Amer. Chem. Soc., 105: 2142-2147. Andrews, S.A., P.M.Huck. 1979. Using Fractionated Natural Organic Matter to Quantitate Organic Byproducts of Ozonation. Ozone Sci. Eng., 16: 1-12. Arora, H., M. LeChevallier, K.L.Dixon. 1997. DBP Occurrence Survey. J. Amer. Water
Works Assoc., 89 (6): 60-68. Audrieth, L.F., J.Kleinberg. 1953. Acetic Acid, In Non-Aqueous Solvents: Applications
as Media for Chemica Reactions: John Wiley and Sons, New York, pp. 148-151. 258
Averett, R.C., J.A.Leenheer, D.M.McKnight, K.A.Thorn. 1994. Humic Substances in the
Suwannee River, Georgia: Interactions, Properties, and Proposed Structures. U.S. Geological Survey Water-Supply Paper 2373. Baes, A.E., Bloom, P.R. 1990. Fulvic Acid Ultraviolet Visible Spectra: Influence of Solvent and pH. Soil Sci. Am. J. 54: 1248-1254. Bellamy, L.J. 1960. The Infra-red Spectra of Complex Molecules. J. Wiley and Sons, London, England. Benjamin, M.M., Yu-Jung Chang, Chi-Wang Li, G.V.Korshin. 1993. NOM Adsorption
onto Iron-Oxide Coated Sand. AWWA Research Foundation and American Water Works Association. Benner, R., J.E.Pakulski, M.McCarthy, J.I.Hedges, P.G.Hatcher. 1992. Bulk Chemical Characteristics of Dissolved Organic Matter in the Ocean. Science, 255: 15611564. Biber, M.V., F.O.Gulacar, J.Buffle. 1996. Seasonal Variations in Principal Groups of Organic Matter in a Eutrophic Lake using Pyrolysis/GC/MS. Environ. Sci.
Technol., 30: 3801-3807. Boerschke, R.C., E.A.Gallie, N.Belzile, R.N.Gedye, J.R.Morris. 1996. Quantitative Elemental and Structural Analysis of Dissolved Organic Carbon Fractions from Lakes near Sudbury, Ontario. Can. J. Chemistry, 74 (12): 2460-2470. Bruchet, A., C.Anselme, O.Marsigny, J.Mallevialle. 1986. THM Formation Potential and Organic Content: a New Analytical Approach. Conférence présentée au
séminaire ‘Substances Humiques’, 15-16 Octobre, Rennes, France, 1-26.
259
Bruchet, A., C.Rousseau, J. Mallevialle. 1990. Pyrolysis GC-MS for Investigating High Molecular Weight THM Precursors and Other Refractory Organics. J. AWWA, 82: 66-74. Brun, G.L., D.L.D.Milburn. 1977. Automated Fluorometric Determination of Humic Substances in Natural Waters. Analyt. Lett., 10: 1209-1219.
Characterization of Natural Organic Matter and its Relationship to Treatability. 1993. American Water Works Association Research Foundation. Denver, CO. Chang, Y., M.M.Benjamin. 1996. Iron Oxide Adsorption and UF to Remove NOM and Control Fouling. J. AWWA, 88 (12): 74-88. Chang, Y., K.H.Choo, M.M.Benjamin, S.Reiber. 1998. Combined Adsorption-UF Process Increases TOC Removal. J. AWWA, 90 (5): 90-102. Chian, E.S.K., W.N.Bruce, H.H.P.Fang. 1975. Removal of Pesticides by Reverse Osmosis. Environ. Sci. Technol., 9: 52-59. Chudoba, J., J Hejzlar, M. Dolezal. 1986. Microbial Polymers in the Aquatic Environment. II : Isolation from River, Potable and Underground Water and Analysis. Water Res., 20 (10): 1223-1227 Clair, T.A., J.R.Kramer, M.Sydor, D.Eaton. 1991. Concentration of Aquatic Dissolved Organic Matter by Reverse Osmosis. Wat. Res., 25: 1033-1037. Clark, M.M. C.Jucker. 1993. Interactions between Hydrophobic Ultrafiltration Membranes and Humic substances. In Proceedings of the AWWA Membrane
Technology Conference, Baltimore, Md. Coble, P.G. 1996. Characterization of Marine and Terrestrial DOM in Seawater Using Excitation Emission Matrix Spectroscopy. Marine Chem., 51: 325-346. 260
Collins, M.R., G.L.Amy, C.Steelink. 1986. Molecular Weight Distribution, Carboxylic Acidity and Humic Substances Content of Aquatic Organic Matter : Implications for Removal during Water Treatment. Environ. Sci. Technol., 20: 1028-1032. Conrad, D.J., P.M.Huck. 1996. Chlorination By-Products from Ferulic Acid (a Lignin Model Compound) and Biologically Treated Pulp Mill Effluents. Wat. Res., 30: 2776-2784. Cook, R.L., C.H.Langford. 1998. Structural Characterization of a Fulvic Acid and a Humic Acid using Solid State Ramp CP-MAS C-13 Nuclear Magnetic Resonance. Env. Sci. Technol., 32 (5): 719-725. Christman, R.F., D.L.Norwood, Y.Seo, F.H.Frimmel. 1989. Oxidative Degradation of Humic Substances from Freszhwater Environments. In Humic Substances II. New York, NY: John Wiley & Sons. Ciavatta, C., M.Govi, L.Sitti, C.Gessa. 1995. Capillary Electrophoresis of Humic Acid Fractions. Comm. Soil Sci. Plant Analysis, 26: 3305-3313. Croué, J.P, B.Martin, A.Deguin, B.Legube. 1993b. Isolation and Characterization of Dissolved Hydrophobic and Hydrophilic Organic Substances of a Reservoir Water. NOM Workshop, Chamonix France, September 19-22, 43-51. Croué, J.P. 1987. Contribution à l’Étude de l’Oxydation par le Chlore et l’Ozone
d’Acides Fulviques Naturels Extraits d’Eaux de Surface. Thèse de Doctorat, Université de Poitiers, n0 d’ordre 89. Croué, J.P., E.Lefebvre, B.Martin, B.Legube. 1993c. Removal of Dissolved Hydrophobic and Hydrophilic Organic Substances during Coagulation/Flocculation of Surfaces Waters. Wat. Sci. Tech., 27: 143-152. 261
Croué, J.P., B.Martin, P.Simon, B.Legube. 1993a. Les Matières Hydrophobes et Hydrophiles des Eaux de Retenues : Extraction, Caractérisation et Quantification.
Water Supply, 11: 51-62. Degens, E.T., J.H.Reuter. 1964. Analytical Techniques in the Field of Organic Geochemistry: Inter. Ser. Monographs Earth Sci., No 15, 377-402. De Leer, E.W.B., C.Erkelens, L. de Galan. 1990. The Influence of Organic Nitrogen Compounds on the Production of Organochlorine Compounds in the Chlorination of Humic Material. Water chlorination : Environmental Impact and Health
Effects (Ed. by Jolley R.L., Condie L.W., Johnson J.D., Katz S., Minear R.G., Mattice J.S. and Jacobs V.A.), 6: 763-781. Ann Arbor Science, Ann Arbor, Mich. De Marini, D.M., A.Abu
Shakra, C.F.Felton, K.S.Patterson, M.L.Shelton. 1995.
Mutation Spectra in Salmonella of Chlorinated, Chloraminated, or Ozonated Drinking Water Extracts: Comparison to MX. Env. Mol. Mutagenesis, 26: 270285. DiGiano, F.A., A.Braghetta, B.Utne, J.Nilson. 1993. Nanofiltration Fouling by Natural Organic Matter and Role of Particles in Flux Enhancement. In Proceedings of the
AWWA Membrane Technology Conference, Baltimore, Md. Disinfection/Disinfection By-Products Database and Model Project. 1991. James M. Montgomery Consulting Engrs. AWWA Report. Denver, CO. Donard, O.F.X., C.Belin, M.Ewald. 1987. Corrected Fluorescence Excitation Spectra of Fulvic Acids. Comparison with the UV/Visible Absorption Spectra. Sci. Total
Env., 62: 157-161.
262
Dossier-Berne, F. 1994. Analyse des Acides Aminés dans les Eaux. Développement et
Application, Thèse de Doctorat, Université de Poitiers. Dycus, P.J.M., K.D.Healy, G.K.Stearman, M.J.M.Wells. 1995. Diffusion Coefficients and Molecular Weight Distributions of Humic and Fulvic Acids Determined by Flow Field Fractionation. Separation Sci. Technol., 30: 1435-1453. Dzombak, D.A., F.M.M.Morel. 1990. Surface Complexation Modeling. Hydrous Ferric
Oxide. John Wiley & Sons, N.Y. Edzwald, J.K. 1993. Coagulation in Drinking Water Treatment : Particles, Organics and Coagulants. Wat. Sci. Tech., 27: 21-35. Edzwald, J.K., W.C.Becker, K.L.Wattier. 1985. Surrogate Parameters for Monitoring Organic Matter and THM Precursors. J. AWWA, 77 (4): 122-132. Ewald, M., C.Belin, J.P.Croue, B.Legube. 1992. Fractionation and Isolation of Dissolved Organic Matter from Reservoir Water. Characterization and Contribution to the Total Fluorescence of the Hydrophilic Fraction. 6th Int. Meet. of Intenat. Humic
Substances Soc., Sept. 20-25, 1992, Monopoli (Bari), Italy, p.256. Fireman, M., R.C. Reeve. 1948. Soil Science Society of America, Proceedings, 13: 494496. Frimmel, F.H., M.U.Kumke. 1998. Fluorescence Decay of Humic Substances (HS) –A Comparative Study. In Humic Substances: Structure, Properties and Uses. Edited by G.Davies and E.Ghabbour. Cambridge, UK: Royal Society of Chemistry.
263
Gadel, F., A.Bruchet. 1987. Application of Pyrolysis-Gas Chromatography-Mass Spectrometry to the Characterization of Humic Substances Resulting from Decay of Aquatic Plants in Sediments and Waters. Wat. Res., 21: 1195-1206. Gadel, F., B.Charriere, L.Serve, l.Comellas. 1992. Caractérisation Chimique des Composés Humiques et de leurs Diverses Classes de Poids Moléculaires dans les Dépôts du Delta du Rhône. Oceanologica Acta, 15: 61-74. Gauthier,T.D., E.C.Shane, W.F.Guerin, W.R.Seitz. 1986. Fluorescence Quenching Method for Determining Equilibrium Constants for Polycyclic Aromatic Hydrocarbons Binding to Dissolved Humic Materials. Env. Sci. Technol, 20: 1162-1166. Ghosh, K., M.Schnitzer. 1979. UV and Visible Absorption Spectroscopic Investigation in Relation to Macromolecular Characteristics of Humic Substances. J. Soil Sci., 30: 735-745. Ghosh, K., M.Schnitzer. 1980. Fluorescence Excitation Spectra of Humic Substances.
Can. J. Soil Sci., 60: 373-379. Ghosh, K., M.Schnitzer. 1981. Fluorescence Excitation Spectra and Viscosity Behavior of a Fulvic Acid and its Copper and Iron Complexes. Soil Sci. Soc. Am. J., 45: 2529. Gjessing, E.T. 1976. Physical and Chemical Characteristics of Aquatic Humus. Ann Arbor Science Publishers, Inc. Ann Arbor, Michigan. Göbbels, F.J., W.Püttman. 1997. Structural Investigation of Isolated Aquatic Fulvic and Humic Acids in Seepage Water of Waste Deposits by Pyrolysis-Gas Chromatography/Mass Spectrometry. Wat. Res., 31: 1609-1618. 264
Goldberg, M.C., P.M.Negomir. 1989. Characterization of Aquatic Humic Acid Fractions by Fluorescence Depolarization Spectroscopy. - Luminescence Applications in Biological, Chemical, Environmental and Hydrological Sciences. American
Chemical Society Simposium series. 383. (M.C.Goldberg, Ed.). ACS, Washington, D.C. 180-205. Gray, K.A., R.M.Bornick. 1996. Use of Pyrolysis to Characterize Natural Organic Material in Artificial Wetlands : Issues Related to Drinking Water Wuality.
Natural Organic Matter Workshop, September 18-19, Poitiers, France. Green, S.A., F.M.M.Morel, N.V.Blough. 1992. Investigation of the Electrostatic Properties of Humic Substances by Fluorescence Quenching. Env. Sci. Technol., 26: 294-302.
Gu, B., J.Schmitt, Z.Chen, L.Liang, J.F.McCarthy.1994. Adsorption and Desorption of Natural Organic Matter on Iron Oxide: Mechanisms and Models. Environ. Sci.
Technol. 28: 38-46. Gu, B., J.Schmitt, J.F.McCarthy. 1995. Adsorption and Desorption of Different Organic Matter Fractions on Iron Oxide. Geochim. Cosmochim. Acta 59: 219-225. Hall, K.J., G.F.Lee. 1974. Molecular Size and Spectral Characterization of Organic Matter in a Meromictic Lake. Water Res., 8: 239-251. Harrington, G.W., A.Bruchet, D.Rybacki, P.C.Singer, 1996. Characterization of Natural Organic Matter and its Reactivity with Chlorine. Water Disinfection and Natural
Organic Matter, R. A.Minear and G.L.Amy, Editors. Chapter 10, 138-158. Hayase,K., H.Tsubota. 1985. Sedimentary Humic Acid and Fulvic Acid as Fluorescent Organic Materials. Geochim. Cosmochim. Acta, 49: 159-163. 265
Hem, J.D. 1970. Study and Interpretation of the Chemical Characteristics of Natural
Water. U.S. Geological Survey Water-Supply Paper 1473, 363 p. Hess, A.N., Y.P.Chin. 1996. Physical and Chemical Characteristics of Polymaleic Acid, a Synthetic Organic Colloid Analog. Colloids and Surfaces A., 107: 141-154. Hongve, D., G.Åkesson, G.Becher. 1989. Comparison of Molecular Weight Distribution and Acid/Base Properties between the IHSS Nordic Fulvic Acid and Whole Water Humic Substances. Sci. Tot. Env., 81/82: 307-314. Horowitz, A. J., K.R.Lum, J.R.Garbarino, G.E.M.Hall, C.Limieux., C.Demas. 1996. Problems
Associated
with
Using
Filtration
to
Define
Trace-Element
Concentrations in Natural Water Samples. Environ. Sci Technol., 30: 954-963 Horth, H. 1989. Identification of Mutagens in Drinking Water. Aqua, 38: 80-100. Hureiki, L. 1993. Étude de la Chloration et de l’Ozonation d’Acides Aminés Libres et
Combinés en Milieu Aqueux Dilué. Thèse de Doctorat, Université de Poitiers, n0 d’ordre 21. Hureiki, L., J.P.Croué, B.Legube, Doré. 1993. Ozonation of Amino Acids : Ozone Demand and Aldehyde Formation. Proceedings of the 11th Ozone World
Congress, August 29-September 3, 1993, San-Francisco (California) USA Ittekkot, V., A.Spitzy, U.Lammerz. 1982. Data on Dissolved Carbohydrates and Amino
Acids in the World Rivers : A Documentation, SCOPE/UNEP Sonderband, Heft 52, pp.575-584. Jadas-Hecart, A., A.El Morer, M.Stitou, P.Bouillot, B.Legube. 1992. Modélisation de la Demande en Chlore d’une eau traitée, Wat. Res., 26, pp. 1073-1084.
266
Jaffe, H.H, M.Orchin 1962. Theory and Applications of Ultraviolet Spectroscopy. John Wiley and Sons, New York. Jahnel, J.B., F.H.Frimmel. 1996. Detection of Glucosamine in the Acid Hydrolysis Solution of Humic Substances. Fresenius J. Anal. Chem., 354 (7/8): 886-888. Kalbitz, K., P.Popp, W.Geyer, G.Hanschmann. 1997. β-HCH Mobilization in Polluted Wetland Soils as Influenced by Dissolved Organic Matter. Sci. Total
Environment, 204: 37-48. Khym, J.X., L.P.Zill. 1952. The separation of Sugars by Ion Exchange, J. Am. Chem.
Soc., 74: 2090-2094. Korshin, G.V., Chi-Wang Li, M.M.Benjamin. 1997a. The Decrease of UV Absorbance as an Indicator of TOX Formation. Water Research, 34 (4): 946-949. Korshin, G.V., M.M.Benjamin, R.S.Sletten. 1997b. Adsorption of Natural Organic Matter on Iron Oxide. Effects of NOM Composition and on the Formation of Organo-Halide Compounds. Water Research, 31: 1643-1650. Korshin, G.V., Chi-Wang Li, M.M.Benjamin 1997c. Monitoring the properties of natural organic matter through UV spectroscopy. Evaluation of a consistent theory. Water
Research 31 (7): 1787-1795. Korshin, G.V., Chi-Wang Li, M.M.Benjamin. 1996. Use of UV Spectroscopy to Study Chlorination of Natural Organic Matter. In Water Disinfection and Natural
Organic Matter. Edited by R.A.Minear and G.Amy. Washington, DC: American Chemical Society.
267
Kuckuk, R., W.Hill, J.Nolte, A.N.Davies. 1997. Preliminary Investigations into the Interactions of Herbicides with Aqueous Humic Substances. Pesticide Science, 51: 450-454.
Kumke, M.U., G. Abbt-Braun, F.H.Frimmel. 1998. Time-Resolved Fluorescence Measurements of Aquatic Natural Organic Matter (NOM). Acta Hydrochim.
Hydrobiol., 26: 73-81. Laane, R.W.P.M., L.Koole. 1982. The Relation between Fluorescence and Dissolved Organic Carbon in the Ems-Dollart Estuary at the Western Wadden Sea. Neth. J.
Sea Res., 15: 217-227. Lapen,A.J., W.R.Seitz. 1982. Fluorescence Polarization Studies of the Conformation of Soil Fulvic Acid. Anal. Chim. Acta, 134: 31-38. Larson, R.A., E.J.Weber. 1994. Reaction Mechanisms in Environmental Organic
Chemistry. Lewis Publishers, Boca Raton, FL Leenheer, J.A., 1997. Fractionation, Isolation, and Characterization of Hydrophilic Constituents of Dissolved Organic Matter in Water: Proceedings of the Natural
Organic Matter Workshop, Sept 18-19, Poitiers, France, pp. 4-1 to 4-5. Leenheer, J. A.. 1984. Concentration, Partitioning, and Isolation Techniques (for Organic Substances in Water), Chapter III in Vol. III of Treatise on Water Analysis (R. Minear, ed.), Academic Press, New York, p. 83-166. Leenheer, J.A. 1981. Comprehensive Approach to Preparative Isolation and Fractionation of Dissolved Organic Carbon from Natural Waters and Wastewaters. Environ.
Sci. Technol., 15 (5): 578-587.
268
Leenheer, J.A., L.B.Barber, C.E.Rostad, T.I.Noyes. 1995a. Data on Natural Organic
Substances in the Dissolved, Colloidal, Suspended-Silt and Clay, and BedSediment Phases in the Mississippi River and Some of its Tributaries, 1991-92: U.S. Geological Survey Water-Resources Investigations Report 94-4191, 47 p. Leenheer, J. A., P.A.Brown, T.I.Noyes. 1989. Implications of Mixture Characteristics on Humic-Substance Chemistry. In Aquatic Humic Substances: Influence on Fate
and Treatment of Pollutants, (I.H. Suffet and P. MacCarthy, eds) Advances in Chemistry Series 219, American Chemical Society, Washington, D.C., pp. 25-40.
Leenheer, J.A., E.W.D.Huffman. 1976. Classification of Organic Solutes in Water by Using Macroreticular Resins. J. Res. U.S. Geol. Surv., 4 (6): 737-751. Leenheer, J.A., R.H.Meade, H.E.Taylor, W.E.Pereira. 1989. Sampling, Fractionation, and Dewatering of Suspended Sediment from the Mississippi River for Geochemical and Trace-Contaminant Analysis. In U. S.Geological Survey Toxic Substances Hydrology Program - Proceedings of the technical meeting, Phoenix, Arizona,
September 26-30, 1988. (Mallard, G. E., and Ragone, S.E., Eds), U.S. Geological Survey Water Resources Investigations Report 88-4220, p. 501-512. Leenheer, J.A., T.L.Noyes. 1984. A Filtration and Column Adsorption System for OnSite Concentration and Fractionation of Organic Substances from Large Volumes of Water. U.S. Geological Survey Water Supply Paper 2230. Leenheer, J.A., T.I.Noyes, R.LWershaw. 1997. (**JL this paper not referred to in report; do you want to find a place to refer to it?). Acquisition and Interpretation of Liquid State 1H NMR Spectra of Humic and Fulvic Acids. In Nuclear Magnetic 269
Resonance Spectroscopy in Environmental Chemistry; Nanny, M.A., Minear, R.A., and Leenheer, J.A., Editors. New York, Oxford University Press, pp. 295303. Leenheer, J. A., T.I.Noyes, P.A.Brown. 1995b. Data on Natural Organic Substances in Dissolved, Colloidal, and Suspended Silt Phases in the Mississippi River and its Major Tributaries, 1987-90: Water-Resources Investigations Report 93-4204.. LeLacheur, R.M., W.H.Glaze. 1996. Reactions of Ozone and Hydroxyl Radicals with Serine. Env. Sci. Technol., 30: 1072-1080. Levashkevich, G.A. 1966. Interaction of Humic Acids with Iron and Aluminum Hydroxides. Sov. Soil Sci., April: 422-427. Levesque, M.1972. Fluorescence and Gel Filtration of Humic Compounds. Soil Science, 113: 346-353. Li, C.W., G.V.Korshin, M.M.Benjamin. 1998. Monitoring Formation of Disinfection ByProducts
with
Differential
UV
Spectroscopy.
J.
Amer.
Water
Work
Association**MB. Lochmueller,C.H., S.S.Saavedra. 1986. Conformational Changes in a Soil Fulvic Acid Measured by Time-Dependent Fluorescence Depolarization. Anal. Chem., 58: 1978-1981. Loder, T.C., P.S.Liss. 1982. The Role of Organic Matter in Determining Surface Charge of Suspended Particles in Estuarine and Oceanic Waters. Thalassia Yugosl. 18: 433-447.
270
Lytle, C.R., E.M.Perdue. 1981. Free, Proteinaceous and Humic-Bound Amino Acids in River Water Containing High Concentration of Aquatic Humus. Env. Sci.
Technol., 15: 224-228 Malcolm, R.L. 1992.
13
C-NMR Spectra and Contact Time Experiment for Skjervatjern
Fulvic and Humic Acids. Environment International, 18: 609-620. Malcolm, R.L. 1989. The Uniqueness of Humic Substances in Each Soil, Stream and Marine Environments. Analytica Chimica Acta, ACA 13199 **JL or **JP we need a more complete reference, 1-12. Malcolm, R. L. 1968. Freeze-Drying of Organic Matter, Clays, and Other Earth Materials: U.S. Geological Survey Professional Paper 600-C, pp C211-C216. Malcolm, R.L., P.MacCarthy. 1992. Quantitative Evaluation of XAD-8 and XAD-4 Resins Used in Tandem for Removing Organic Solutes from Water. Environment
International, 18: 597-607. Martin, B. 1995. La Matière Organique Naturelle Dissoute des Eaux de Surface :
Fractionnement, Caractérisation et Réactivité, Thèse de Doctorat, Université de Poitiers. Martin-Mousset, B., J.P.Croué, E.Lefebvre, B.Legube. 1997. Distribution and Characterization of Dissolved Organic Matter of Surface Waters. Wat. Res., 31: 541-553. McIntyre, C., B.D.Batts, D.R.Jardine. 1997. Electrospray Mass Spectrometry of Groundwater Organic Acids. J. Mass Spectrometry, 32: 328-330.
271
McKnight, D.M., G.R.Aiken, R.L.Smith. 1991. Aquatic Fulvic Acids in Microbially Based
Ecosystems :
Results
from
Two
Desert
Lakes
in
Antartica.
Limnol.Oceanogr., 36 (5): 998-1006. McKnight, D.M., E.M.Thurman, R.L.Wershaw. 1985. Biogeochemistry of Aquatic Humic Substances in Thoreau’bog, Concord, Massachusetts. Ecology, 66 (4): 1339-1352. Miller, J.W., P.C.Uden. 1983. Characterization of Non-Volatile Aqueous Chlorination Products of Humic Substances. Environ. Sci. Technol., 17: 150-157. Mrkva, M. 1969. Investigation of Organic Pollution of Surface Waters by Ultraviolet Spectrophotometry. J. Wat. Pol. Contr. Fed., 41: 1923-1931. Nanny, M.A., R.A.Minear, J.A.Leenheer. 1997. Nuclear Magnetic Resonance
Spectroscopy in Environmental Chemistry, Oxford University Press, New York. Novak, J.M., G.L.Mills, P.M.Bertsch. 1992. Estimating the Percent Aromatic Carbon in Soil and Aquatic Humic Substances Using Ultraviolet Absorbance Spectroscopy.
J. Environ. Qual., 21: 144-147. Ochiai, M., T. Nakajima. 1988. Distribution of Neutral Sugar and Sugar Decomposition Bacteria in Small Stream water. Arch. Hydrobiol., 113 (2): 179-187 Ogura, N., T.Hanya. 1968. Ultraviolet Absorbance as an Index of the Pollution of Seawater. J. Wat. Pol. Contr. Fed., 40: 464-467. Ogura, N., T.Hanya. 1966. Nature of Ultraviolet Absorbance in Seawater. Nature, 212: 758-759. Panais, B. 1994. Contribution à la Mise au Point et à l’Exploitation de Nouveaux Outils
Analytiques Permettant d’Évaluer la Qualité des Eaux dans les Usines de 272
Production d’Eau Potable. Diplôme d’ingénieur du C.N.A.M, Université de Poitiers. Paode, R.D., G.L.Amy, S.W.Krasner, R.S.Summers, E.W.Rice. 1997. Predicting the Formation of Aldehydes and BOM. J. AWWA, 89, June: 79-93. Parfitt, R.L., A.R.Fraser, V.C.Farmer. 1977. Adsorption on Hydrous Oxides. III. Fulvic Acid and Humic Acid on Goethite, Gibbsite and Imogolite. J. Soil Sci. 28: 289296. Parfitt, R.L., J.D.Russell. 1977. Adsorption on Hydrous Oxides. IV. Mechanism of Adsorption of Various Ions on Goethite. J. Soil Sci. 28: 297-305. Pennanen,V., J.Mannio. 1987. A Note on the Use of Dilution to Overcome Quenching Effects when Measuring Fluorescence of Humic Lakes. Sci. Total Env., 62: 163164. Perdue, E.M. 1985. Acidic Functional Groups in Humic Substances. In Humic
Substances in Soil, Sediment and Water. Geochemistry, Isolation and Characterization. Edited by G.R.Aiken, D.M.McKnight, R.L.Wershaw, and P.MacCarthy. New York, NY: John Wiley & Sons. Pischel, G., Th.Wildt. 1988. Humic Substances of Natural and Anthropogeneous Origin.
Wat. Res., 22: 105-108. Pompe, S., M.Bubner, M.A.Denecke, T.Reich, A.Brachmann, G.Geipel, R.Nicolai, K.H.Heise, H.Nitsche. 1992. A Comparison of Natural Humic Acids with Synthetic Humic Acid Model Substances: Characterization and Interaction with Uranium(VI). Radiochim. Acta, 74: 135-140.
273
Pouchert, C.J. 1985. The Aldrich Library of FT-IR Spectra, Volumes 1 and 2: The Aldrich Chemical Company, Milwaukee, Wis, 2958 p. Puchalsky,M.M., M.J.Morra. 1992. Fluorescence Quenching of Synthetic Organic Compounds by Humic Materials. Env. Sci. Technol., 26: 1787-1792. Reckhow, D.A. 1984. Organic Halide Formation and the Use of Pre-Ozonation and
alum Coagulation to Control Organic Halide Precursors. Ph.D. Thesis, Department of Environmental Sciences and Engineering, Chapel Hill, N.C. Reckhow, D.A., P.C.Singer, R.L.Malcolm. 1990. Chlorination of Humic Materials : Byproduct Formation and Chemical Interpretations. Environ. Sci. Technol., 24: 478-482. Remmler, M., A.Georgi, F.-D.Kopinke. 1995. Evaluation of Matrix-Assisted Laser Desorption/Ionization (MALDI) Time-of-Flight (TOF) Mass Spectrometry as a Method for the Determination of the Molecular Mass Distributions of Humic Acids. European. J. Mass Spectrometry, 22: 403-407. Rees, T.F., J.A.Leenheer, J.F.Ranville. 1991. Use of a Single-Bowl Continuous-Flow Centrifuge for Dewatering Suspended Sediments: Effect on Sediment Physical and Chemical Characteristics. Hydrological Processes, 5: 201-214. Roemelt,P.M., W.R.Seitz. 1982. Fluorescence Polarization Studies of Perylene-Fulvic Acid Binding. Env. Sci. Technol., 16: 613-616. Saiz-Jimenez, C. 1994. Analytical Pyrolysis of Humic Substances : Pitfalls, Limitations, and Possible Solutions. Environ. Sci. Technol., 28: 1773-1780. Schulten, H.R., B.Plage. 1991. A Chemical Structure for Humic Substances.
Naturwissenschaften, 78: 311-312. 274
Scott, A.I. 1964. Interpretation of the Ultraviolet Spectra of Natural Products. Pergamon Press, New York. Seitz,W.R. 1981. Fluorescence Methods for Studying Speciation of Pollutants in Water.
Trends Anal. Chem., 1: 79-83. Semmens, M.J., A.B.Staples. 1986. The Nature of Organics Removed during Treatment of Mississippi River Water. J. AWWA, 78(2): 76-81. Serkiz, S.M., E.M.Perdue. 1990. Isolation of Dissolved Organic Matter from the Suwannee River Using Reverse Osmosis. Wat. Res., 24: 911-916. Shapiro, J. 1961. Freezing Out, a Safe Technique for Concentration of Dilute Solutions.
Science, 133: 2062-2064. Siegel. A., E.T.Degens. 1966. Concentration of Dissolved Amino Acids from Saline Waters by Ligand Exchange Chromatography. Science, 151: 1098-1101. Sinsabaugh, R.L., R.C.Hoehn, W.R.Knocke, A.E.Linkins. 1986a. Removal of Dissolved Organic Carbon by Coagulation with Iron Sulfate. J. AWWA, 78 (5): 74-82. Sinsabaugh, R.L., R.C.Hoehn, W.R.Knocke, A.E.Linkins. 1986b. Precursor Size and Organic Halide Formation Rates in Raw and Coagulated Surface Waters. J.
Environ. Eng., 112: 139-153. Smart, P.L., B.L.Finlayson, W.D.Rylands, C.M.Ball. 1976. The Relationship of Fluorescence to Dissolved Organic Carbon in Surface Waters. Wat. Res., 10: 805811. Smeds, A., T.Vartiainen, J.Maki-Paakkanen, L.Kronberg. 1997. Concentrations of Ames Mutagenic Chlorohydroxyfuranones and Related Compounds in Drinking Waters.
Env. Sci. Technol., 31: 1033-1039. 275
Spitzy, A. 1988. Dissolved Organic Matter in Groudwaters from Different Climates. SCOPE/UNEP Sonderband, Hamburg.
Standard Methods for the Examination of Water and Wastewater. 1995. 18th Edition. American Public Health Association, American Water Works Association, Water Pollution Control Federation: Washington, D.C. Stevenson, F.J. 1982. Humus Chemistry, Genesis, Composition, Reactions, Spectroscopic Approaches, Chapter 11. J. Wiley and Sons, New York, pp. 264-284. Stewart,A.J., R.G.Wetzel. 1981. Asymmetrical Relationships between Absorbance, Fluorescence and Dissolved Organic Carbon. Limnol. Oceanogr., 26: 590-597. Stewart,A.J., R.G.Wetzel. 1980. Fluorescence: Absorbance Ratios - a Molecular Weight Tracer of Dissolved Organic Matter. Limnol. Oceanogr., 25: 559-564. Summers, R.S., S.M.Hooper, H.M.Shukairy, G. Solarik, D.Owen. 1996. Assessing DBP Yield : Uniform Formation Conditions. J. AWWA, 88, June: 80-93. Sun, L., E.M.Perdue, J.F.McCarthy. 1995. Using Reverse Osmosis to Obtain Organic Matter from Surface and Ground Waters. Wat Res., 29: 1471-1477. Sweet, M.S., E.M.Perdue. 1982. Concentration and Speciation of Dissolved Sugars in River Water. Environ. Sci. Technol., 16: 692-698. Thorn, K.A., Arterburn, J.B., Mikita M.A. 1992.
15
N and
13
C NMR investigation of
hydroxylamine-derivatized humic substances. Env. Sci. Technol., 26(1): 107-116. Thorn, K.A., D.W.Folan, P.MacCarthy. 1989.
Characterization of the International
Humic Substances Society Standard and Reference Fulvic and Humic Acids by Solution State Carbon-13 (13C) and Hydrogen-1 (1H) Nuclear Magnetic Resonace
276
Spectrometry. U.S. Geological Survey Water-Resources Investigations Report 894196, 93 p. Thurman, E.M. 1985. Organic Geochemistry of Natural Waters. M.Nijhoff and W.Junk, Publishers, Dordrecht. Thurman, E.M., G.Aiken, R.L.Malcolm. 1978a. (**JL or **JP this paper not referred to in report; do you want to find a place to refer to it?). The use of Macroreticular Nonionic Resins to Preconcentrate Trace Organic Acids from Water. In
Proceedings, 4th Joint Conference on Sensing of Environmental Pollutant, 166: 630-634. Thurman, E.M., R.L.Malcolm. 1989. Nitrogen and Amino Acids in Fulvic and Humic
Acids. United States Geological Survey, Open-File Report 87-557, Humic Substances in the Suwannee River, Georgia : Interactions, Properties, and Proposed Structures, Denver, CO. Thurman, E.M., R.L.Malcolm, G.Aiken. 1978b. (**JL or **JP -- this paper not referred to in report; do you want to find a place to refer to it?). Prediction of Capacity Factors for Aqueous Organic Solutes Adsorbed on a Porous Acrylic Resin. Anal.
Chem., 50: 775-779. Thurman, E.M., R.L.Malcolm. 1981. Preparative Isolation of Aquatic Humic Substances.
Environ. Sci. Technol., 15: 463-466. Tipping, E. 1981. The Adsorption of Aquatic Humic Substances by Iron Oxide. Geochim.
and Cosmochim. Acta 45: 191-199.
277
Tipping, E., D.Cooke. 1982. The Effects of Adsorbed Humic Substances on the Surface Charge of Goethite (a-FeOOH) in Freshwaters. Geochim. and Cosmochim. Acta 46: 75-80.
Traina, S.J., J.Novak, N.E.Smeck. 1990. An ultraviolet Absorbance Method of Estimating the Percent Aromatic Carbon Content in Humic Acids. J. Environ.
Qual, 19: 151-153. Velapoldi, R.A, K.D.Mielenz. 1980. A Fluorescence Standard Reference Material:
Quinine Sulfate Dihydrate. National Bureau of Standards. Washington: Dept. of Commerce. Vinodgopal, K., P.V.Kamat. 1992. Environmental Photochemistry of Surfaces. Charge Injection from Excited Fulvic Acid into Semiconductor Colloids.
Env. Sci.
Technol., 26: 1963-1966. Vitali, M., V.Leoni. 1993. Determination of 2-Ethyl-1-hexanol as Contaminant in Drinking Water. J. of AOAC International, 76: 1133-1137. Volk, C.J., C.B.Volk, L.A.Kaplan. 1997. Chemical Composition of Biodegradable Dissolved Organic Matter in Streamwater. Limnology and Oceanography, 42 (1): 39-44. Von Wandruszka, R., C.Ragle, R.Engebretson. 1997. The Role of Selected Cations in the Formation of Pseudomicelles in Aqueous Humic Acid. Talanta, 44: 805-809. Wang,
Z.D.,
B.C.Pant,
C.H.Langford.
1990.
Spectroscopic
and
Structural
Characterization of a Laurentian Fulvic Acid: Notes on the Origin of Color. Anal.
Chim. Acta, 232: 43-49.
278
Wershaw, R. L., G.R.Aiken. 1985. Molecular Size and Weight Measurements of Humic Substances: In Humic Substances in Soil, Sediment, and Water: Geochemistry,
Isolation, and Characterization, (G.R. Aiken, D. M. McKnight, R. L. Wershaw, and P. MacCarthy, eds), J. Wiley and Sons, New York, pp. 477-492. Wershaw, R. L., M.A.Mikita. 1987. NMR of Humic Substances and Coal. Lewis Publishers, Chelsea, Michigan Westerhoff, P., R.G.Song, G.Amy, R.Minear. 1998. NOM's Role in Bromine and Bromate Formation during Ozonation. J. AWWA, 90 (2): 82-94. Wiesner, M., S.Chellam. 1992. Mass Transport Considerations for Pressure-Driven Membrane Processes. J. AWWA, 84, January: 88-95. Wilson, M.A. 1987. NMR Techniques and Applications in Geochemistry and Soil
Chemistry, Pergamon, Sydney, Australia. Xiong, F. 1990. Contribution à l’Étude de l’Ozonation des Acides Fulviques Aquatiques, Thèse de Doctorat, Université de Poitiers, n0 d’ordre 383. Xu, X., H.X.Zou, J.Q.Zhang. 1997. Formation of Strong Mutagen [3-chloro4(dichloromethyl)-5- hydroxy-2(5H)-furanone] MX by Chlorination of Fractions of Lake Water. Wat. Res., 31: 1021-1026.
279
