Report No. 422 DEVELOPMENT of an ANALYTICAL METHOD FOR

Report No. 422

DEVELOPMENT OF AN ANALYTICAL METHOD FOR TASTE AND ODOR COMPOUNDS COMMONLY FOUND IN DRINKING WATER SOURCES

By Detlef R.U. Knappe and Anjali Viswakumar

Department of Civil, Construction, and Environmental Engineering North Carolina State University 208 Mann Hall, 2501 Stinson Drive Raleigh, North Carolina 27695-7908

UNC-WRRI-422

DEVELOPMENT OF AN ANALYTICAL METHOD FOR TASTE AND ODOR COMPOUNDS COMMONLY FOUND IN DRINKING WATER SOURCES Detlef R.U. Knappe and Anjali Viswakumar

Department of Civil, Construction, and Environmental Engineering North Carolina State University 208 Mann Hall, 2501 Stinson Drive Raleigh, North Carolina 27695-7908

The research on which this report is based was supported by funds provided by the North Carolina General Assembly through the North Carolina Department of Environment and Natural Resources.

Contents of this publication do not necessarily reflect the views and policies of the WRRI, nor does mention of trade names or commercial products constitute their endorsement by the WRRI or the State of North Carolina.

This report fulfills the requirements for a project completion report of the Water Resources Research Institute of The University of North Carolina. The authors are solely responsible for the content and completeness of the report.

WRRI Project No. 70239 June 7, 2011

ABSTRACT

DEVELOPMENT OF AN ANALYTICAL METHOD FOR TASTE AND ODOR COMPOUNDS COMMONLY FOUND IN DRINKING WATER SOURCES An analytical method was developed to detect and quantify 19 compounds commonly associated with taste and odor (T&O) problems in drinking water. The method can be used by utilities during T&O episodes to quickly and reliably detect T&O compounds and determine their concentrations. Knowledge about the identity and concentration of T&O compounds will greatly aid utilities in the selection of appropriate treatment strategies. Head space solid phase microextraction (SPME) was used to concentrate T&O compounds, and gas chromatography (GC) followed by tandem mass spectrometry (MS/MS) was used to separate, detect, and quantify the T&O compounds. Method development included (1) determination of parent ion mass and retention time, (2) optimization of the ion trap MS/MS instrumental parameters, (3) development of calibration curves, and (4) determination of the limits of quantitation (LOQs) for each compound. For 12 of the 17 targeted T&O compounds with known odor threshold concentrations (OTCs), LOQs were below the OTC. This result suggests that the developed method is capable of detecting developing T&O problems for these 12 compounds and allow utilities to implement treatment strategies before consumers can detect objectionable tastes and odors in their water. The developed method was tested by analyzing three water samples from North Carolina ponds and lakes that experienced algae/cyanobacteria blooms. In addition, Raleigh source and tap water samples were analyzed. Of the 19 targeted T&O compounds, all but 2,4,6 tribromoanisole were detected in the collected samples. In the bloom samples, geosmin, β-ionone, and trans-2,cis-6-nonadienal most frequently occurred at concentrations that exceeded their OTCs, sometimes by a factor of >100. A comparison of results for non-filtered and filtered (0.45 µm membrane) samples suggests that many T&O compounds were predominantly present inside algae/cyanobacteria cells. Only geosmin and β-cyclocitral were present at measurable levels in Raleigh source and tap water. Concentrations in the raw and tap water samples were similar and at levels that were at or below the OTCs of geosmin and β- cyclocitral.

ACKNOWLEDGMENTS

The authors of this report would like to thank the following people and organizations for their cooperation, participation, and support of this project: • The Water Resources Research Institute (WRRI) of the University of North Carolina for funding this project • The WRRI staff for project management • David Black of the NC State Environmental Engineering Research Laboratory for his support throughout this research • Elle Allen and JoAnn Burkholder of the NC State University Center for Applied Aquatic Ecology for cyanobacteria identification • Staff at the E.M. Johnson Water Treatment Plant in Raleigh and at the Twin Lakes Community in Burlington for their support during sampling

1. Introduction and Objectives

The presence of taste and odor (T&O) compounds in drinking water is an important issue for many utilities. In general the public perception is that the aesthetic quality of water is a good indicator of how safe the water is to drink. Thus water utilities meeting primary drinking water standards will not be able to guarantee consumer satisfaction unless their water is also taste and odor free (McGuire, 1995). Considerable financial resources are spent by the water industry to prevent and control T&O problems. For example, a T&O survey of about 800 US and Canadian water utilities showed that water utilities on average spent 4.5% of their total budget to control T&O problems (Suffet et al., 1996). The causes of T&O problems have been largely attributed to microbial byproducts, disinfectants and disinfection byproducts, and distribution system materials (AwwaRF and Lyonnaise des Eaux, 1995; Whelton and Dietrich, 2004). Most T&O compounds implicated in consumer complaints have been microbial metabolites formed in surface waters (Suffet et al., 1999; Peter et al., 2009). Actinomycetes, cyanobacteria, algae and even fungi have been associated with the occurrence of T&O compounds (Gerber, 1979; Watson, 2003: Zaitlin and Watson, 2006). T&O problems commonly occur when “culture-like conditions” (a 2- to 5- week period with little or no rain, average flows less than 6500 cfs and water temperatures of 17oC or greater) exist in reservoirs (Raschke et al., 1975). Such conditions may occur with increasing frequency in North Carolina, which experienced extreme drought conditions in 2002, 2007 and 2008 in >30% of the state (ncwater.org). Apart from the water source, T&O compounds may also form during treatment and distribution. Therefore, it is important for water utilities to screen their source and finished waters as well as distribution system samples for T&O compounds to determine the identity and origin of these compounds. The principal objective of this study was to develop an analytical method that can be used by drinking water utilities to simultaneously detect and quantify 19 compounds commonly associated with T&O problems in drinking water. To assure sensitivity and specificity, gas chromatography tandem mass spectrometry (GC-MS/MS) was used. Sample preconcentration was achieved by head space solid phase microextraction (SPME) which has emerged as an ideal preconcentration technique with such advantages as small sample volume, solvent-free extraction, and adaptability to automation. Additional objectives of this study were to develop standard curves and to identify the limit of quantitation (LOQ) for each compound. The final objective was to apply the developed method to identify and quantify T&O compounds in several North Carolina water samples including a drinking water source, tap water, and water from lakes that were experiencing algae/cyanobacteria blooms.

2. Background

2.1 Classification of taste and odor The Taste and Odor Wheel (Figure 1) was developed over the last twenty years (AwwaRF and Lyonnaise des Eaux, 1987, Suffet at al., 1999, 2004) and organizes T&O compounds according to their specific organoleptic characteristics. It includes eight classes of odors, four tastes and a mouth feel/nose feel category. The inner circle (Figure 1) describes the primary T&O groups, the second circle defines the primary T&O groups as described by trained sensory panelists, and finally the outer circle shows chemicals that have been confirmed in (denoted with a *), or are “representative” standards or “substitutes” (standards from natural odorous material that can be consistently prepared) for the various T&O categories. The main goal for the development of the T&O wheel was to 1) develop a common language for T&O sensory panelists and drinking water professionals, and 2) to present to the water industry the existing knowledge about the organoleptic characteristics of drinking water. The following is a short description of common T&O compounds using the primary odor descriptors in the wheel as a guide for their classification. Table 1 provides a summary of common T&O compounds together with their CAS number, chemical properties, and structure.

Earthy/musty/moldy Geosmin. With its name derived from the Greek word “ge”= earth and “osme”= odor, geosmin or trans-1,10-dimethyl-trans-9-decalol was first isolated from actinomycetes (Gerber and LeChevalier, 1965). As its name suggests, geosmin has an earthy odor with an odor threshold concentration (OTC) of <10 ng/L (AwwaRF and Lyonnaise des Eaux, 1987). Frequently, an OTC of 4 ng/L is reported for geosmin (Malleret et al., 2001). Cyanobacteria are regarded to be the major producers of geosmin in surface water (Watson, 2004). Geosmin is a saturated tertiary alcohol that is quite stable in water and quite resistant to oxidation by conventional oxidants used in water treatment (Gerber, 1979; Peter and von Gunten, 2007). Granular and powdered activated carbon, ozonation, and high pressure membrane filtration processes can be effective for the removal of geosmin (Cook and Newcombe, 2004).

Figure 1. The Drinking water Taste and Odor wheel “2000” (Suffet et al., 1999).

Table 1. Common taste and odor compounds, their CAS number, chemical properties and structure.

Vapor Molecular Weight Boiling point Compound Name CAS Number Pressure Chemical Structure OTC (g/mol)* (°C)* (Torr)* (ng/L) <4,000[2] dimethyl disulfide 624-92-0 94.20 109.7±0.0 28.7 H3C S S CH3 4,500[2] 1-hexanal 66-25-1 100.16 127.9±3.0 10.9 H3C O H C 3 ? cis-4-heptenal 6728-31-0 112.17 151.6±9.0 3.64 O 3,000[2] 1-heptanal (heptaldehyde) 111-71-7 114.19 150.4±3.0 3.85 O H3C S S 10[2] dimethyl trisulfide 3658-80-8 126.26 183.1±23.0 1.07 H C S CH 3 3 CH3 1500[5] cis-3-hexenyl acetate 3681-71-8 142.20 174.2±19.0 1.22 O O

CH 3 2,500- H C [3] trans,trans-2,4-heptadienal 4313-03-5 110.15 177.4±9.0 1.04 3 O 5,000

H3C CH3 2-isopropyl-3- 0.20[4] 25773-40-4 152.19 210.8±40.0 0.274 O methoxypyrazine H3C N N

Table 1. Continued

Vapor Molecular Weight Boiling point Compound Name CAS Number Pressure Chemical Structure OTC (g/mol)* (°C)* (Torr)* (ng/L)

H C 3 4.00[6] trans-2,cis-6-nonadienal 557-48-2 138.21 203.3±0.0 0.280 O

CH3 O 2-isobutyl-3- N 1.00[4] 24683-00-9 166.22 210.8±35.0 0.273 methoxypyrazine H3C N CH3 HO CH3 H3C 9.00[1] 2-methylisoborneol 2371-42-8 168.28 208.7±8.0 0.0487 H3C H3C H3C CH3 3,000[6] β-cyclocitral 432-25-7 152.23 212.1±19.0 0.176 O

CH3 H N 300,000[3] 2,3-benzopyrrole (indole) 120-72-9 117.15 253.0±9.0 0.0298

O CH3 ? trans,trans-2,4-decadienal 25152-84-5 152.23 244.6±0.0 0.0300

Table 1. Continued

Vapor Molecular Weight Boiling point OTC Compound Name CAS Number Pressure Chemical Structure (g/mol)* (°C)* (ng/L) (Torr)* Cl

0.03[1] 2,4,6-trichloroanisole 87-40-1 211.47 246.0±0.0 0.0436 O Cl H3C Cl Cl

7.00[1] 2,3,6-trichloroanisole 50375-10-5 211.47 254.1±35.0 0.0281 O H3C Cl Cl CH3 4.00[1] geosmin 19700-21-1 182.30 252.4±8.0 3.01E-3 OH CH3 CH H3C CH3 3 7.00[2] β-ionone 79-77-6 192.30 254.8±0.0 0.0169 O

CH3 Br

0.03[1] 2,4,6-tribromoanisole 607-99-8 344.83 298.0±0.0 2.31E-3 O Br H3C Br *Source: SciFinder Scholar (ACS, 2009). Calculated using Advanced Chemistry Development (ACD/Labs) Software V9.04 for Solaris (© 1994-2009 ACD/Labs) @760 Torr and 25oC. [1] Malleret et al. 2001, [2] Cotsaris et al. 1995, [3] Young and Suffet 1999, [4] Young et al. 1996, [5] Khiari et al. 1995, [6] Rashash et al. 1996

2-methylisoborneol (MIB). The compound 1,2,7,7-tetramethylbicyclo[2.2.1]heptan-2-ol, commonly known as 2-methylisoborneol or more succinctly MIB, was first isolated from actinomycetes by Gerber (1969). MIB is produced by some actinomycetes that also produce geosmin (Gerber, 1978). MIB is more volatile than geosmin and has a higher boiling point (Pirbazari et al., 1992). MIB is a methylated monoterpene and has similar chemical stability as geosmin and is typically more difficult to remove from water than geosmin. A musty odor is identified with 2-MIB and its OTC is in the 9-42 ng/L range (Krasner et al., 1983) with 9 ng/L often being the reported value (Malleret et al., 2001). Trihalogenated anisoles. The compounds 2,4,6-trichloroanisole (2,4,6 TCA); 2,3,6- trichloroanisole (2,3,6 TCA) and 2,4,6-tribromoanisole (2,4,6 TBA) commonly form via biomethylation of trichlorophenols in distribution systems. Chlorophenols can form during the free chlorine disinfection of water sources containing phenol. Halogenated anisoles can have OTCs in the sub-ng/L range (Malleret et al., 2001, Diaz et al., 2005); e.g. the 2,4,6 isomers have an OTC of 30 pg/L. A study involving a T&O episode in Paris, France established that 2,4,6 TBA was released from the cementitious coating of a storage tank and was the principal cause of the musty odor of the water (Malleret and Bruchet, 2002). 2-isopropyl-3-methoxypyrazine (IPMP). IPMP is of major significance in the flavor of peas (Murray et al., 1970) and is responsible for the earthy aroma of potatoes (Buttery and Ling, 1973), the musty aroma of sterile milk (Gordon and Morgan, 1972), and the musty potato-like odor of some spoiled and chilled fish (Miller et al., 1973). In a survey conducted of 25 raw vegetables, 18 contained measurable quantities of IPMP (Murray and Whitfield, 1975). It has been described as having a “potato-bin/musty” odor by FPA panelists and has an OTC of 2 ng/L (Cotsaris et al., 1995, Young et al., 1996).

Marshy/swampy/septic/sulfurous Dimethyl disulfide and dimethyl trisulfide. Alkyl sulfides can be produced by actinomycetes (e.g. Pseudonocardia thremophilia, Wilkins, 1996), by the cyanobacteria Microcystis (Juttner, 1984) and Anabaena (Juttner et al, 1983), and by the golden-brown alga Synura petersenii (Collins and Kalnins, 1965). Dimethyl disulfide is said to produce a decaying vegetation odor and dimethyl trisulfide a swampy/septic odor (Khiari et al., 1997). The OTC of dimethyl disulfide has been reported as 0.2-5 µg/L (Van Gemert and Nettenbreijer, 1977), and < 4 µg/L (Cotsaris et al., 1995) while that of dimethyl trisulfide is 10 ng/L (Buttery et al., 1976, Cotsaris et al., 1995). 2,3-benzopyrrole (indole). Indole has been associated with septic odors. It has been described as having a “fecal and stench” odor at high concentration and a pleasant odor at lower levels (Khiari et al., 1997). Its OTC has been reported as 300 µg/L (Van Gemert and Nettenbreijer, 1977, Young and Suffet, 1999).

Fishy/Rancid Fishy odors have been associated with byproducts of algal growth and decomposition (Malleviale and Suffet, 1987). In a study by Watson et al. (1999), golden-brown algae (Uroglena americana, Dinobryon spp., and Synura petersenii) were isolated from ponds and

cultured to produce polyunsaturated fatty acid (PUFA)-derived aldehydes associated with fishy/rancid off flavors. Both the trans,cis (biological) isomers and the trans,trans derivatives of the compounds 2,4,7-decatrienal, 2,4-heptadienal and 2,4-decadienal (from Dinobryon) have been shown to cause fishy odors during an episode of high biomass algal growth under an ice- covered Canadian reservoir (Watson et al., 2001). trans,trans-2,4-heptadienal. The compound t,t-2,4-heptadienal has been obtained from Uroglena americana cultures (Yano, 1988; Watson et al., 1999) and confirmed to be responsible for the fishy/swampy type odors in some water supplies (Khiari et al, 1995). It can also originate from decaying grass in water (Khiari et al., 1995b). Young and Suffet (1999) determined its OTC to be 2.5-5 µg/L by FPA analysis. 1-heptanal (heptaldehyde). A flavor profile analysis (FPA) of 1-heptanal by Rashash et al., (1997) described its odor as “oily, slight citrus” (7 µg/L), ”oily” (21 µg/L) and “strong cilantro” (>80 µg/L), while Fabrellas et al. (2004) described 1-heptanal as having a rancid oil + fish, fruity (orange, mandarin) odor. It has been identified as an algal metabolite along with 1-hexanal in an isolate of Synedra rumpens Kutz from Lake Biwa (Japan) (Kikuchi et al., 1974). The OTC of 1- heptanal has been reported to be 3 µg/L (Cotsaris et al., 1995). trans,trans-2,4-decadienal and trans,cis,cis-2,4,7-decatrienal. Decadienal and decatrienal are lipooxygenase-mediated cleavage products of Chyrosphyceae (golden-brown algae) and diatoms, respectively (Juttner, 1995). The trans,cis and trans,trans isomers of 2,4-decadienal have been identified in Poterioochromonas malhamensis (Juttner and Hahne, 1981), in Synura uvella (Juttner, 1981), and in Dinobryon isolates (Juttner et al., 1986). 2,4,7-decatrienal has been identified in the diatom Asterionella formosa (Juttner and Muller, 1979). The odor descriptor for trans,trans-2,4-decadienal and trans,cis,cis-2,4,7-decatrienal is “fishy, oily” and the OTC of trans,cis,cis-2,4,7-decatrienal is reported as 1.5 µg/L (Watson, 2004). No OTC has been reported for trans,trans-2,4-decadienal. Grassy/hay/straw/woody 1-hexanal. Initially given a fishy odor descriptor, 1-hexanal (or n-hexanal) was identified as a metabolite of the diatom Synedra rumpens Kutz isolated from the Lake Biwa (Japan) (Kikuchi et al., 1974). The presence of 1-hexanal has also been associated with the algae Scendesmus subspicatus and Asterionella formosa (Cotsaris et al., 1995). Aldehydes are also known to be byproducts of ozonation (Bruchet et al., 1992) and low molecular weight aldehydes have been considered to be the cause of odors in drinking waters (Hrudey et al., 1988). Hexanal was given the descriptor of green, grass and earthy (<80 µg/L) and green apple, fruity, candy (> 80µg/L) in a recent study (Fabrellas et al., 2004). The OTC of 1-hexanal is 4.5 µg/L (Cotsaris et al., 1995). cis-3-hexen-1-ol and cis-3-hexenyl acetate. Cis-3-hexenyl acetate is used as a reference standard for the grassy descriptor in the T&O wheel (Khiari et al., 1995) and both cis-3-hexen-1-ol and cis-3-hexenyl acetate have been identified in drinking water supplies (Khiari et al, 1999). Cis-3-hexenyl acetate is first released by fresh grass which is followed by chemical reaction (hydrolysis, biodegradration) to produce cis-3-hexen-1-ol. The acetate has been shown to be about 50 times as odorous as the alcohol (Khiari et al., 1995). Cis-3-hexen-1-

ol has also been shown to be a product of the green alga Scendesmus suspicatus (Costaris et al., 1995) and is thought to contribute to the tomato aroma in some plants (Eskin, 1979). The OTC of cis-3-hexenyl acetate is 1.5 µg/L (Khiari et al., 1995). trans-2,cis-6-nonadienal. This metabolite is produced by the cyanobacterium Synura petersenii (Rashash et al., 1996) and has been confirmed as the cause for “cucumber” odor in water (Burlingame et al., 1991b). Trans-2,cis-6-nonadienal can be formed by the oxidation of linolenic acid, and algae that form the acid also have the potential to form this conjugate aldehyde (Rashash et al., 1997). The compound has also been found in cucumbers (Boelens et al., 1987; Schieberle et al., 1990). The FPA panel of the Philadelphia Water Department found the trans,trans configuration of the compound to have more rancid and oily organoleptic characteristics as compared to the green, cucumber characteristics of the trans,cis form (Burlingame et al., 1992). The FPA also found that for the trans,cis form at its OTC (4 ng/L, Rashash et al., 1996) the odor was sweet, green vegetation, at 60 ng/L it was a distinctive fresh green cucumber odor, while at higher concentrations the odor became more waxy, oily, and rancid.

Fragrant: vegetable/fruity/flowery 2-isobutyl-3-methoxypyrazine (IBMP). The compound IBMP is similar to IPMP in structure and like IPMP, IBMP is present in numerous vegetables. It has been described by FPA panelists as having a “vegetable” and “bell pepper” like odor and an OTC of 1-2 ng/L (Khiari et al., 1992; Khiari et al., 1997; van Gemert and Nettenbreijer, 1977, Young et al., 1996). 6-methyl-5-hepten-2-one. This compound is produced by cyanobacteria such as Synechococcus strains (Henatsch and Juttner, 1983) and Anabaena cylindrica (Juttner et al., 1983) as well as by the golden-brown alga Synura uvella (Juttner, 1981). Its odor descriptor is fruity and ester-like, and it has an OTC of 50 µg/L (Watson, 2004). β-cyclocitral. The nor-carotenoid 2,6,6-trimethyl-1-cyclohexane-1-carboxaldehyde (or β- cyclocitral) may be a degradation product of carotenes and has been observed at significant levels by Juttner (1976, 1983, 1984a, 1987) in Microcystis wesenbergii, Microcystis viridis and Microcystis aeruginosa. The odor character of β-cyclocitral changes with concentration. Between 2 and 20 µg/L in distilled water it imparts a hay/woody odor (Young et al., 1999) while at higher concentrations it has a “tobacco-like” odor (Juttner, 1976, Slater and Block, 1983b). Based on the results of Rashash et al. (1996), the OTC of β-cyclocitral is 3 µg/L. β-ionone. Also a nor-carotenoid, β-ionone has been associated with Cyanidium caldarium and Synura uvella (Juttner, 1983), Scendesmus subspicatus (Cotsaris et al., 1995) and Planktothrix rubescens (Peter et al., 2009). It is a product of carotene oxidation (Watson, 2003) and has been identified in natural waters (Khiari et al., 1997). The odor descriptor for β-ionone is “flowery, violet,” and it has an OTC of 7 ng/L (Cotsaris et al., 1995).

2.2 Causes of taste and odor The causes of taste and odor problems have been largely attributed to microbial byproducts, to disinfectants and disinfection byproducts, and to distribution system processes and materials (Whelton and Dietrich, 2003).

Microbial byproducts Of all the T&O compounds related to microbial byproducts, MIB and geosmin are the most extensively researched. Both MIB and geosmin were originally associated with the heterotrophic actinomycete bacteria (Gerber, 1968; Medsker, 1968). It was after the study by Tabachek and Yurkowski (1976) that the photoautotrophic cyanobacteria were also implicated with the production of T&O compounds and became recognized to be the dominant source of these compounds (Juttner and Watson, 2007). The genus Streptomyces (actinomycetes) has been reported to produce not only MIB and geosmin but also other musty/earthy smelling compounds (e.g. IBMP and IPMP, dimethyl disulfide, dimethyl trisulfide) (Zaitlin and Watson, 2006). Among the cyanobacteria, a long list of genera (e.g. Anabaena, Aphanizomenon, Microcystis, Oscillatoria, Phormidium) is associated with the occurrence of MIB, geosmin and other T&O compounds (Juttner and Watson, 2007). Table 2 lists some common T&O compounds and the algae/cyanobacteria/actinomycetes groups that produce them. In addition, soil fungi in the genera Penicillium, Aspergillus and Chaetomium can produce MIB and geosmin (Borjesson et al., 1993; Bjurman and Kristensson, 1992; Kikuchi et al., 1983). Geosmin and MIB occur as both cellular (bound) and dissolved fractions (extracellular products) and the differentiation between these fractions is essential for effective water quality analysis and management (Juttner and Watson, 2007; Rashash et al., 1996). Suffet et al. (1999) point out that a presumptive test (identifying an organism considered to be the cause of a particular T&O problem and quantifying its population density) will not be sufficient to confirm that the microorganism under question is the true cause for the T&O problem. They describe the “rules-of-evidence” method that must be satisfied to validate a cause-and-effect relationship with either one of the following steps: 1. Correlating the odor produced by an organism isolated by culture techniques to the original odor problem in the water as determined by a sensory panel. 2. Isolating microbial metabolites causing the odor from the culture and confirming the compounds by GC/MS. 3. Isolating the microbial metabolites causing the odor from natural water samples. 4. Developing a FPA quantification (employing either the Weber-Fechner model or Stevens Power law model) of the compound in actual water samples.

Table 2. Common taste and odor compounds and the major algae /actinomycetes associated with them

Compound Major algae group and dominant taxa Actinomycete taxa Odor descriptors Source

Anabaena, Aphanizomenon, Fischerella, Gerber(1979),Zaitlin and Geosmin Lyngbya, Oscillatoria, Streptomyces earthy-corn-musty Watson, 2006 Phormidium,Schizothrix, Symploca

Oscillatoria, Phormidium, Pseudanabaena, Gerber(1979),Zaitlin and 2-methylisoborneol (MIB) Streptomyces earthy-musty Synechococcus Watson, 2006

Watson et al., 2000, Anabaena circinalis, Aphanizomenon sweet-fruity- Young et al., 1999; β-cyclocitral flosaquae Microcystis wesenbergii, Microcystis chocolate-pipe Juttner, 1984; Juttner et viridis and Microcystis aeruginosa tobacco al., 1983

2,4-heptadienal, 2,4-decadienal, Chrysosphaerella, Dinobryon sociale, Synura Juttner et al., 1995, fishy 2,4,7-decatrienal petersenii Watson et al., 2000

trans,trans-2,4-heptadienal Uroglena Americana fishy Yano et al., 1988

Chrysosphaerella, Dinobryon sociale, Synura trans-2,cis-6-nonadienal cucumber Watson et al., 2000 petersenii

cis-3-hexen-1-ol, Scenedesmus subspicatus grassy Khiari et al., 1999 cis-3-hexenyl acetate

2-isobutyl-3-methoxypyrazine, Septic, decaying Gerber(1979),Zaitlin and Streptomyces 2-isopropyl-3 methoxypyrazine vegetation Watson, 2006

Streptomyces,Pseudonocardia Wilkins(1996); Zaitlin dimethyl disulfide, , Saccharomonospora, Septic, decaying and Watson, 2006; Microcystis, Anabaena dimethyl trisulfide Thermoactinomyces, vegetation Juttner, 1984; Juttner et Thermomonospora al., 1983

Fruity, flowery, Wilkins(1996); Juttner 6-methyl-5-hepten-2-one Synechococcus, Anabaena,Synura Thermoactinomyces tobacco (1981,1983)

Disinfectants and disinfection products Chlorinous T&O problems are the primary cause of consumer complaints at utilities in the U.S (Suffet et al., 1996) and in France (Bruchet, 1999). Such T&O episodes are usually associated with the detection of chlorinated, brominated and iodinated disinfection byproducts (DBPs) (Bruchet, 1999). Halogenated anisoles impart an earthy-musty odor and have OTCs in the pg/L range. For example, brominated anisoles found in treated waters have an OTC of 30 pg/L (Malleret et al., 2001). The chlorinous odor is pH dependent (odor intensity decreases as pH is raised from 5.5 to 8.5) and so is the formation of odorous chloro- and bromophenols (optimum formation at 8< pH < 9) (Bruchet, 1999; Suffet et al., 1999). Table 3 summarizes odorous compounds associated with disinfectants and DBPs, their method of formation, and the primary T&O group descriptor.

Table 3. Disinfectants and disinfection byproducts

Odorous compounds Formation method Odor References descriptor

Free Chlorine, Chloramines Chlorination Chlorinous Krasner and Barret, 1984

Chlorophenols, Bromophenols Free chlorine + phenol Medicinal Khiari et al., 1999b + (Bromide)

Brominated/Iodinated Haloforms Free chlorine + NOM Medicinal Cancho et al., 1999 +bromide/iodide

Halogenated anisoles (e.g. 2,4,6- Biomethylation of Earthy-musty Malleret et al., 2001, trichloroanisole, 2,3,6-trichloroanisole halogenated phenols Diaz et al., 2005 and bromoanisoles)

Distribution system About 65% of U.S utilities surveyed by AWWA in 1989 identified the distribution systems as the primary cause of their T&O problems (Suffet et al., 1996). The major causes leading to T&O episodes in the distribution system are either (1) of biological origin, (2) related to disinfection or oxidation byproducts, or (3) associated with leaching of chemicals from distribution system materials. The first two causes were discussed above. Regarding T&O occurrences related to the materials used in distribution systems and premise plumbing, Tomboulian et al. (2004) identified the following scenarios: a) Leaching of chemicals from additives and coatings in the system; reaction of these chemicals with disinfectants or other additives; biotransformation of the leachates. After the rehabilitation of a reservoir in Rennes, France, consumer complaints describing “medicinal” odors in drinking water led to further investigation (Rigal and Danjou, 1999). A non-inert PVC net had been used along with cement-based products to renew 13

the reservoir. It was discovered that under high pH conditions, triphenylphosphate leached from the PVC and reacted with the chlorine residual to form an odorous trichlorophenol hydrolysis product. b) Non-compliance of rules during application of coatings and additives, which results in adverse T&O episodes, especially after any new installation or maintenance work. A T&O episode in eastern France with customer complaints describing phenol/styrene-type odors occurred in May 1994 (Khiari et al., 1999). On investigation, it was discovered that one of the reservoirs had recently been lined with acrylic coating reinforced with fiberglass. Laboratory experiments demonstrated that phenol was leaching from the coating material, which led to the formation of iodinated by-products (known to produce medicinal odors) that were found in the sample. A defective batch of the material used for lining the tanks was determined to be the cause of the problem. c) Periods of stagnation in the distribution system leading to increases in trace level contamination. A T&O complaint in Barcelona, Spain, was related to the occurrence of medicinal odors. Stagnation exacerbated the problem, and it was postulated that either slow formation kinetics of brominated compounds or the dissipation of chlorine residual, which could have masked the odor initially, were the cause (Khiari et al., 1999). The treated water had been standing in the distribution system for longer than usual due to maintenance. Tomboulian et al. (2004) give a comprehensive list of materials typical of most distribution systems and the possible undesirable organoleptic characteristics that can be associated with them. These compounds were not targeted in this study.

2.3 Taste and odor compound detection and quantification A distinct odor as perceived by humans is often the result of a complex mixture of volatile compounds, and the concentration-response relationship for every compound is unique. The odor threshold concentration (OTC) is defined as the lowest concentration at which the odor is perceived. The OTC is determined from a flavor profile analysis (FPA) as the concentration of a compound that is needed in a FPA sample to evoke an average intensity rating of 2 from panelists when the sample temperature is 45oC (Rashash, 1994). Generally the psychometric function (concentration-response) of the compounds is a sigma plot. There is a gradual increasing response to concentration until the OTC is reached. After the OTC has been reached small changes in concentration are easily perceived. Finally at higher concentrations, changes in concentration are not easily perceived (Delahunty et al., 2006). Thus the analysis of an odor is complicated and subject to many variables such as the concentration-response range analyzed, difficulty in separating a mixture of compounds associated with a particular odor, and associating the source of a given odor with an individual compound. For these reasons, methods for T&O compound detection and quantification comprise both instrumental and sensory techniques.

Sensory methods Sensory methods involve qualitative description of tastes or odors and rating the quality on an intensity scale. Sensory analysis is typically conducted by a FPA panel of trained professionals. The OTCs for many T&O compounds are below instrumental detection limits and hence sensory methods are often more sensitive than instrumental methods (Suffet et al., 2004). However sensory methods also have disadvantages such as: a) Sensory methods require highly trained panelists and are labor intensive. For example the FPA directs panelists to record descriptors quantitatively on a 7-point scale for all T&O compounds. b) Sensory methods are also highly subjective to the individuals on the panel. Sensory panels relate the chemical concentration in water samples to the odor intensity using either the Weber-Fechner model or the Stevens’s power model. Alternatively, the TON (threshold odor number) is defined as the highest dilution of the water sample at which more than one individual can perceive an odor. This is a much simpler method than the FPA and was originally used as a standard method in T&O analysis (APHA et al., 2005). Overall sensory methods have an important place for the early detection of T&O events during the regular monitoring of source and finished waters. Also, new and simpler FPA versions are in use today. The simpler methods include (1) the attribute rating test; (2) the 2-out-of-5 odor test; and (3) the rating method for evaluating distribution system odors in comparison to a control (Dietrich et al., 2004). Instrumental methods Instrumental methods of analysis have the advantage of objectivity, repeatability, and accuracy. Instrumental methods are able to confirm the identity and quantity of the compound responsible for the T&O problem. However, instrumental methods also have some disadvantages. E.g., because the OTCs of many T&O compounds are in the ng-pg/L range, instrumental detection limits are a limiting factor, and a preconcentration step is needed for the analysis. Another potential disadvantage is that the instrument is set to detect and quantify only the compounds you are looking for, and the compound causing the T&O problem might not be in the analyte list of the method. The need for low detection limits necessitates the use of highly sophisticated instrumentation like gas chromatography/mass spectrometry for separation/detection in addition to a preconcentration step. Gas chromatography/mass spectrometry. Gas chromatography (GC) is based on the principle of partitioning of the analyte between a gaseous mobile phase and a liquid phase immobilized on the surface of an inert solid (Skoog and Leary, 1992). A mass spectrum (MS) is obtained by the conversion of the components of a sample into rapidly moving gaseous ions and separating them based on their mass-to-charge (m/z) ratios (Skoog and Leary, 1992). In ion trap MS, eluates from a chromatographic column are ionized, stored in a radio-frequency field, and periodically ejected to the electron multiplier detector so that ion scanning on the basis of mass/charge is possible (Skoog and Leary, 1992). Tandem mass spectrometry (MS/MS) involves carrying out two mass selective processes in series. In the first mass-selective operation, the parent ion of each compound is isolated; in the second operation the mass/charge ratio of parent ion fragments

are determined (March, 1997). GC-MS methods have emerged as the method of choice in T&O detection and analysis for three primary reasons: (1) high sensitivity and high resolving power; (2) ability to yield a spectral signature for each compound that can be used for simultaneous quantitation and structural confirmation of the compounds, and (3) convenience of bench-top GC-MS systems that have become available in many laboratories (Suffet et al., 1995). Table 4 summarizes analytical methods that have been used for each primary odor group (Suffet et al., 1995). Chemical properties like polarity, volatility, hydrophobicity, boiling point, and molecular weight of the compounds in the analyte list are all important when deciding on the analytical method for T&O compounds. Sensory GC or GC-olfactometry is the technique of using human assessors to detect and evaluate the volatile compounds eluting from a GC column. GC-olfactometry has found many applications in the food and beverage industry (Delahunty et al., 2006). It is often used as a complement to chemical analysis of T&O compounds (Anselem et al., 1985; Peter et al., 2009) with the objectives of characterizing the odor profile of a sample and quantifying the relative importance of various compounds to a particular odor (Delahunty et al., 2006). Disadvantages of this method of analysis include: (1) sensory GC detects compounds that are present at concentration levels lower than their OTCs in water as they are volatilized into pure states in the GC column, and (2) sensory GC does not account for the synergistic effect of compounds that yields a particular odor. However, sensory GC can be used to screen potential T&O candidates in a sample that result from a GC-MS analysis (Suffet et al., 1995).

Table 4. Analytical methods of choice for common odor categories (adapted from Suffet et al., 1995)

Odor descriptor Target analytes Analytical method preconcentration, separation- detection* Earthy-musty 2-MIB, geosmin, trichloroanisole CLSA,GC-MS; CLSA,GC-ECD Medicinal Iodinated THMs, phenols, chlorophenols LLE(CH2Cl2),GC-MS

Swampy Sulfides CLSA,GC-MS; Purge and trap, GC-MS; Headspace, GC, FPD Hydrocarbons Hydrocarbons LLE,GC,FID; LLE,GC-MS Chemical-miscellaneous LLE,GC-MS; Purge and Trap, GC-MS; Direct aqueous injection, GC-MS Chlorinous Residual oxidants DPD methods** Cucumber trans-2,cis-6-nonadienal CLSA,GC-MS

Fruity Aldehydes ≥ C7 CLSA,GC-MS * CLSA - closed loop stripping analysis; ECD - electron capture detector; LLE - liquid-liquid extraction; FPD - flame photometric detector; FID - flame ionization detector. ** Standard method 4500, APHA et al. (2005).

Preconcentration techniques. Preconcentration techniques in use for the detection of T&O compounds include (1) closed loop stripping analysis (CLSA) (Standard method 6040, APHA et al., 2005), (2) simultaneous distillation extraction (SDE) (Khiari et al., 1995), (3) purge and trap, (4) open stripping analysis (OSA), (5) liquid-liquid extraction (LLE), (6) liquid-liquid microextraction (LLME), (7) continuous liquid-liquid extraction (CLLE), (8) solid-phase extraction (SPE) with different adsorbents, and (9) solid-phase micro-extraction (SPME). The LLE, SDE, CLSA and OSA methods have been shown to give low detection limits but require a large sample size, require expensive equipment, and are time consuming and labor intensive. The SPME method developed by Pawliszyn et al. (1990) has found application as a preconcentration technique across various fields and different matrices such as drinking water (Pan and Pawliszyn, 1997), wastewater (Huang et al., 2004), foods and beverages (Yang and Pepard, 1994), environmental samples (Chai et al., 1993; James and Stack, 1996), and biological samples (e.g. human blood and urine) (Lee et al., 1996). The major advantages of SPME as compared to other preconcentration techniques are: (1) small sample volume requirement, (2) elimination of organic solvents, and (3) suitability for automation. Furthermore, head space SPME (HSPME) eliminates the need of the fiber to be immersed in the sample and thus can be effectively used in matrices with suspended matter (Bao et al., 1999; Huang et al., 2004; Watson et al., 2000).

2.4 Treatment techniques for mitigating taste and odor problems The biggest challenge in treating T&O compounds in drinking water is related to their presence and olfactory detection at low concentrations (pg/L-µg/L). Their removal is required from a complex mixture that contains organic matter and inorganic matter at mg/L levels. These constituents can also form odorous compounds during chemical oxidation or biological reactions (Bruchet and Duget, 2004). Therefore the control of T&O problems at the source is recognized to be a far less expensive option than treatment. Source control strategies include (1) control of nutrient influx and availability in reservoirs to prevent eutrophication, (2) algae control in reservoirs or presedimentation basins through the use of algicides like copper sulfate, potassium permanganate, and chlorine, (3) artificial destratification or mixing to prevent anoxic conditions, (4) hypolimnetic aeration so that deep, cold water is available for withdrawal and T&O, color, iron and manganese problems are avoided, and (5) biological control by using biological control agents (e.g. viruses, fungi, protozoa and zooplankton) (AwwaRF and Lyonnaise des Eaux,1995). Various treatment processes from conventional oxidation methods to the latest advances in membrane treatment have been shown to control a variety of off-flavours. Table 5 summarizes different treatment options, their advantages and disadvantages, and their applicability to particular odor classes. Table 5 was adapted from McGuire (1999) and Bruchet and Duget (2004). The sequence of treatment processes also plays an important role in the control of T&O. Hoehn and Mallevialle (AwwaRF and Lyonnaise des Eaux, 1995) discuss the various possible sequences that utilities can employ to control T&O problems.

Table 5. Treatment methods, target T&O categories, and advantages/disadvantages of treatment methods

Treatment Method Target T&O odor categories Advantages Disadvantages Sources

Oxidation - Chlorine, chloramines, Earthy/musty; Chlorinous; a) Both disinfection and oxidation a) Ineffective for MIB and Reckhow, 1996; chlorine dioxide Grassy/woody; Swampy; Fragrant; in one step geosmin Kirmeyer et al., Fishy; Medicinal; Chemical 1993; Lyn et al., b) Chlorine and chlorine dioxide b) Impart chlorinous type 1995 oxidize many T&O compounds odor and some DBPs impart medicinal/earthy musty odor c) Chloramines reduce chlorinous odor but cannot oxidize many T&O compounds

d) Tendency to mask other odors Oxidation - Potassium Compound-selective oxidant; a) Inexpensive a) A weaker oxidant Lalezary et al., permanganate effective for β-cyclocitral, linolenic compared to ozone and 1986; Ficek and acid, and trans-2,cis-6-nonadienal. chlorine Waer, 1993; Dietrich et al., b) Selective oxidant - does 1995 not target many compounds, including MIB and geosmin Oxidation - Ozone/Advanced Earthy/musty; Chlorinous; a) Powerful oxidant effective for a) Forms aldehydes and Smith and Moss, oxidation processes (AOPs) - Grassy/woody; Swampy; Fragrant; MIB and geosmin ketones as byproducts 1993; Coffey et Fishy; Medicinal; Chemical creating fragrant/fruity odor al., 1995; b) Ozone/GAC combination most Andrews et al., 1996; Lang et al., efficient for T&O removal. b) Need “promoter” e.g. H2O2 or NOM for generation of OH 1996 c) Also removes color and many radicals micropollutants c) Expensive compared to a) Both disinfection and oxidation chlorination/potassium in one step permanganate

Table 5. Continued

Treatment Method Target T&O odor categories Advantages Disadvantages Sources Adsorption - GAC, PAC Earthy/musty; Medicinal; Chemical; a) “Tried and true” method that a) Expensive; not effective DeMarco, 1994; Fishy/Swampy. removes the compound from for all T&O categories. Smith and Pettit, water instead of just 1996; Graham et transforming it. al., 1996; Simpson et al., 1997; Krasner et al., 2004. Biological Treatment Earthy/musty; Fishy; Swampy; a) Combination of ozone/GAC a) Less commonly Huck, 1996; Medicinal or ozone and any granular media implemented in the U.S. Norton et al., filter helps produce biologically 1997; Rice and stable water, less chlorine Hansen, 1994; demand and fewer T&O Croll, 1996 problems in distribution system. Membranes Salty; many T&O compounds a) Can control a variety of off- a) Expensive Anon, 1997; flavours. Baudin et al., 1993; Anselme et b) Advances in membrane b) Concentrate disposal al., 1997 materials has broadened application for T&O removal.

3. Materials and Methods 3.1 Chemicals and materials Table 6 summarizes information about the T&O compounds and other chemicals used in this study. Apart from their name, CAS number, purity, and source, Table 6 lists the stock standard solution concentration if the compound was purchased in methanol or the density of the neat compound if the pure substance was purchased. Deionized water was used as a laboratory blank to account for background corrections and to prepare the calibration standards. A 1-cm divinylbenzene/ carboxenTM/ polydimethylsiloxane stableflexTM (DVB/CAR/PDMS) SPME fiber (Supelco 57329-U) was used for sample preconcentration by headspace SPME. Every new fiber was conditioned at 270oC for 2 hours before sample analysis. Twenty-mL clear glass vials (Supelco SU860097) fitted with open top screw caps and PTFE-faced silicon septa (Supelco SU860101) were used to hold samples for GC-MS/MS analysis. Sodium chloride was baked at 450oC for 4 hours prior to use as a salting out agent. The 20-mL sample vials were washed in soap, rinsed three times with DI water, and baked at 550oC for 2 hours. Screw caps and PTFE-faced silicone septa (Sigma Aldrich, Inc.) were rinsed three times with DI water and baked at 105oC. The 5, 10, 100 and 1000 µL microsyringes used to prepare the standards were rinsed with methanol 10-15 times between the use of different solutions. 3.2 Instrumentation A Varian (Palo Alto, CA, USA) Saturn 2200 tandem mass spectrometer (MS/MS) connected to a Varian 3800 gas chromatograph (GC) and a Combi PALTM SPME autosampler was used for method development and sample analysis. The Combi PALTM autosampler (CTC Analytics AG, Zwingen, Switzerland) had a built-in agitator that enabled heated agitation of the sample during the SPME process. The GC was fitted with a capillary split/splitless injector (Varian 1177). The capillary column used was a FactorFour VF-5ms column - 30m x 0.25 mm, 0.25 µm film thickness (Varian Inc., Palo Alto, CA, USA). The ion trap MS was operated in the positive chemical ionization (CI) MS/MS mode to identify and quantify the targeted analytes. The trap temperature was set at 180oC, the manifold at 60oC, and the transfer line at 250oC. LC-MS grade methanol was used as the CI reagent. The mass spectrometer default settings are summarized in Table 7, and the settings for the individual MS segments are discussed in detail in Chapter 4.

Table 6. Names, properties, and sources of compounds used in this study.

Stock solution Density of neat concentration material Compound Name CAS Number Purity (%) Vendor in methanol (g/cm3) (µg/ml) dimethyl disulfide 624-92-0 99.9 1000 Supelco hexanal-d12 99.5 atom%D 0.890 CDN Isotopes 1-hexanal 66-25-1 98 0.814 Aldrich cis-4-heptenal 6728-31-0 ≥98 0.850 SAFC 1-heptanal (heptaldehyde) 111-71-7 ≥95 0.818 Fluka dimethyl trisulfide 3658-80-8 99.9 1000 Supelco cis-3-hexenyl acetate 3681-71-8 ≥98 0.897 SAFC trans,trans-2,4-heptadienal 4313-03-5 90 0.881 Aldrich 2-isopropyl-3-methoxypyrazine 25773-40-4 99.9 100 Supelco trans-2,cis-6-nonadienal 557-48-2 95 0.866 Aldrich 2-isobutyl-3-methoxypyrazine 24683-00-9 99.9 100 Supelco methylisoborneol-d3 135441-89-3 >99 atom%D CDN Isotopes 2-methylisoborneol 2371-42-8 99.9 100 Supelco β-cyclocitral 432-25-7 ≥90 0.943 SAFC 2,3-benzopyrrole (indole) 120-72-9 99.9 2000 Supelco trans,trans-2,4-decadienal 25152-84-5 85 0.857 Aldrich 2,4,6-trichloroanisole 87-40-1 99.1 100 Supelco 2,3,6-trichloroanisole 50375-10-5 99.9 Sigma-Aldrich geosmin 19700-21-1 99.8 100 Supelco β-ionone 79-77-6 96 0.945 Aldrich 2,4,6-tribromoanisole 607-99-8 99 Aldrich sodium chloride 7647-14-5 Certified ACS Fisher Scientific methanol (LC-MS chromasolv) 67-56-1 99.9 Fluka acetone(GC resolv) 67-64-1 99.5 Fisher Scientific

Table 7. Default settings for mass spectrometer

MS/MS parameter Default value Brief explanation of select parameters

Emission current 15 µA Will affect the number of ions formed (higher peak area)

Multiplier Offset 300 V

Scan time 0.600 seconds

Ionization mode CI-auto

Ion preparation technique MRM Multiple reaction monitoring - MS/MS technique; user- selected ions are transmitted through the first MS stage and fragments arising from these ions are measured by the second MS stage.

Reagent gas Methanol

CI storage level 19.0 m/z

Ejection Amplitude 15.0 m/z This voltage ejects all ions not falling within the acquisition mass range

Background mass 55 m/z

Maximum ionization 2500 microseconds time

Maximum reaction time 128 milliseconds

Target TIC 5000 counts Total ion count

Prescan ionization time 200 microseconds

Ionization parameters

Ionization storage level 48.0 m/z

Ejection amplitude 20.0 V This voltage ejects all ions not falling within the acquisition mass range

Isolation window 3.0 m/z The amount of parent ions in the trap is a function of this parameter.

Low-edge offset 6 steps

High-edge offset 2 steps

High-edge amplitude 30.0 V

Table 7. Continued

MS/MS parameter Default value Brief explanation of select parameters

Isolation time 5 milliseconds Duration of collision-induced dissociation (CID) waveform

Dissociation parameters

Waveform type Resonant

Modulation range 2 steps The number of digital to analog conversion steps during resonant excitation

Modulation rate 3000 microseconds/step

Number of frequencies 1

CID frequency offset 0 Hz

Excitation time 20 milliseconds

3.3 Methods

Preconcentration Solid phase microextraction (SPME) of T&O compounds from the headspace of a water sample was carried out at 65ºC for 30 minutes using 1-cm DVB/CAR/PDMS SPME fibers. To 10 mL of sample, about 2.5 grams of sodium chloride was added in 20-mL glass vials. Salt addition decreases the solubility of the compounds in water and along with stirring has been shown to accelerate the mass-transfer from the aqueous phase into the headspace of the vial (Bao et al., 1999). The salt-sample mixture was placed in the autosampler tray of the Combi PALTM that was programmed to perform the following sequential steps: (1) preheat sample in the agitator at 300 rpm for 30 minutes at 65ºC to ensure that the sample is at 65ºC before extraction begins, (2) insert SPME fiber through the septum and adsorb T&O compounds from the headspace for 30 minutes in the agitator (fiber depth at 22 mm from the bottom of the vial, agitator set at 250 rpm, agitator rotation stopped for 2 seconds every 30 seconds), and (3) desorb at 250ºC for 5 min into the GC injector (with 0.8-mm glass liner) that operated initially in the splitless mode. The injector was switched to the split mode with a split ratio of 20:1 after 3.00 minutes of desorption. The carrier gas was helium at a flow rate of 1.5 mL/min. The Combi PALTM settings and the GC- injector settings are summarized in Table 8.

Table 8. SPME conditions and GC-injector settings

Combi PALTM settings

Incubation Temperature 65oC

Agitator On time 30 seconds

Agitator Off time 2 seconds

Vial Penetration 22 mm

Extraction time 30 minutes

Injection penetration 54 mm

Desorption time 5 minutes

Injector (Varian Type 1177)

Oven temperature 250 oC

Split ratio initial On at 20:1

Split ratio at 0.01 minutes Off

Split ratio at 3.00 minutes On at 20:1

Column flow 1.5 mL/min

Separation Separation of T&O compounds was achieved with a FactorFour VF-5ms column with the following temperature program: 35ºC for 23 minutes, ramp to 139ºC at a rate of 4ºC/min, ramp to 301ºC at a rate of 27ºC/min, and hold at 301ºC for 5 minutes. Many T&O studies have utilized the same or similar columns, and the temperature program was previously shown to be effective for a method developed for 12 fishy, swampy, and grassy odor compounds (Sclimenti and Krasner, 2003). The optimal method run time was determined to be seventy minutes and twenty seconds by the Varian Software.

Detection/Quantification Compounds were identified by comparing mass spectra to those obtained in previous studies (Young and Suffet, 1999; Sclimenti and Krasner, 2003; and application notes from Varian Inc.). Compound-specific mass spectrometer settings were determined with the help of the automated methods development (AMD) feature of the Varian Software (MS Workstation ver 6.9). Most importantly, the optimum voltage for resonant excitation was determined. This voltage fragments the parent ion into daughter ions while the parent ions are injected into the ion trap of the MS.

The dominant daughter ion for each compound was chosen as the quantitation ion. Detailed descriptions of the AMD results are provided in Chapter 4. Preparation of stock standard and intermediate standards Stock standard solutions purchased in methanol were kept at -17oC and were used prior to the listed expiration date. For compounds purchased in neat form, stock standard solutions were prepared by dissolving a certain volume (in µL: 2 / density of the neat compound) in 2 mL of methanol to obtain a 1000 mg/L stock standard solution. E.g. to prepare 1000 mg/L of 1-hexanal stock, 2 /0.814 = 2.5 µL of the neat compound was added to 2 mL methanol. Intermediate standards were prepared from stock standard solutions and stored at -17oC for up to two months. Table 9 summarizes the stock standard solution concentrations for each compound and provides details on the preparation of intermediate standards and calibration stock mixtures.

Calibration mixtures and standards Based on instrument responses of preliminary sample analyses, compounds were grouped into a high and a low sensitivity bin. Group 1 compounds exhibited high sensitivity while Group 2 compounds exhibited lower sensitivities. Accordingly, two calibration mixtures were prepared as follows: a high-concentration calibration mixture (12.5 mg/L) for Group 2 compounds and a low-concentration calibration mixture (125 µg/L) for Group 1 compounds. The calibration curves were obtained with seven aqueous calibration standards. Serial dilutions of the calibration mixtures (12.5 mg/L and 125 µg/L) yielded intermediate solutions from which the seven calibration standards were prepared (Table 10). MIB-d3 and hexanal-d12 were the internal standards for the Group 1 and Group 2 compounds, respectively (Table 9). The internal standards were spiked at 10 ng/L (MIB-d3) and 500 ng/L (hexanal-d12) in every calibration standard.

Isotope dilution method Quantification of the compounds was accomplished using an isotope dilution procedure. The concept behind the isotope dilution method is that an isotopically labeled analog of the target analyte will have a similar extraction recovery, ionization response and retention time as the compound under study. It has been shown to account for extraction inefficiency and loss of analytes because of inter-sample variability in many studies and was also shown to be effective across different sample matrices (e.g., Stanford and Weinberg, 2007). Deuterium-labeled MIB (MIB-d3) and hexanal (hexanal-d12) served as the two internal standards in this study. Their retention times nearly overlapped with those of their respective non-labeled counterparts, and the same MS segments were used to capture both the labeled and non-labeled parent ions in this method. More details about the isotope dilution method are provided in Chapter 4.

Table 9. Concentrations of stock standard solutions and procedures used to obtain intermediate standards and calibration standard mixtures.

Stock standard solution Volume of stock standard solution added Volume of stock/intermediate standards Compound Name concentration to acetone to prepare 2 mL of to make calibration mixtures (µL) in methanol (mg/L) intermediate standard (µL) Calibration Mixture 1 Calibration stock mixture containing 125 Intermediate Standards (5mg/L each) (Group 1) µg/L of each compound in 2 mL acetone 2-isopropyl-3-methoxypyrazine 100 100 50 2-isobutyl-3-methoxypyrazine 100 100 50 2-methylisoborneol 100 100 50 β-cyclocitral 1000 10 50 2,4,6-trichloroanisole 100 100 50 2,3,6-trichloroanisole 1308 7.6 50 geosmin 100 100 50 β-ionone 1000 10 50 2,4,6-tribromoanisole 100 100 50 No intermediate standards. Calibration stock Calibration Mixture 2 Calibration stock mixture containing 12.5 mixture prepared directly from stock (Group 2) mg/L of each compound in 2 mL acetone standard solutions dimethyl disulfide 1000 25 1-hexanal 1000 25 cis-4-heptenal 1000 25 1-heptanal (heptaldehyde) 1000 25 dimethyl trisulfide 1000 25 cis-3-hexenyl acetate 1000 25 trans,trans-2,4-heptadienal 1000 25 trans-2,cis-6-nonadienal 1000 25 2,3-benzopyrrole (indole) 2000 12.5 trans,trans-2,4-decadienal 1000 25 Second intermediate standard of 100 µg/L Internal Standards in 2mL acetone for MIB-d3 methyl-d3-isoborneol 1120 8.9 40 hexanal-d12 2657 3.8

Table 10. Concentrations and preparation of calibration standards.

Group 1 Compounds Group 2 Compounds Internal Standards

Concentration Concentration Concentration to Volume added Concentration of stock Volume Concentration of stock Volume be achieved in 10- Concentration of to 10-mL calib. to be achieved mixture from added to to be achieved mixture from added to mL calibration stock mixture standard (µL) Calibration in 10-mL which 10-mL in 10-mL which 10-mL standard (ng/L) Points calibration calibration calib. calibration calibration calib. MIB- Hex- standard standard was standard standard standard was standard MIB- Hex- MIB- Hex- d3 d12 (ng/L) prepared (µL) (ng/L) prepared (µL) d3 d12 d3 d12 (µg/L) (mg/L) (µg/L) (mg/L)

1 0.1 1.25 0.8 10 0.125 0.8 10 500 100 5 1 1

2 0.5 1.25 4.0 50 0.125 4.0 10 500 100 5 1 1

3 1 12.5 0.8 100 1.25 0.8 10 500 100 5 1 1

4 5 12.5 4.0 500 1.25 4.0 10 500 100 5 1 1

5 10 62.5 1.6 1000 6.25 1.6 10 500 100 5 1 1

6 25 62.5 4.0 2500 6.25 4.0 10 500 100 5 1 1

7 50 125 4.0 4000 12.5 3.2 10 500 100 5 1 1

4. Results and Discussion

To meet the objectives of this study, the following tasks were completed: 1) Validate suitable conditions for effective headspace SPME preconcentration and separation of the 19 T&O compounds on the GC column. 2) Determine optimal MS/MS settings. 3) Develop calibration curves for the 19 T&O compounds. 4) Determine the limit of detection (LOD) and limit of quantitation (LOQ) for each T&O compound. 5) Test the developed method with different water samples from NC: algal bloom samples, drinking water and distribution system samples were analyzed using the method after calibration curves were developed. The results of tasks 1 and 2 are described in section 4.1 while the results of tasks 3-5 are described in sections 4.2 to 4.4 of this chapter.

4.1 Method development Preconcentration and compound separation To ensure effective preconcentration and separation of the 19 target analytes, a previously used headspace SPME method and GC temperature program for the analysis of 12 T&O compounds (Sclimenti and Krasner, 2003) was used as a starting point. After preliminary testing, the same SPME conditions and GC temperature program were chosen in this study because they effectively concentrated and separated the 19 T&O compounds targeted in this study. For compounds with similar retention times, the ion trap MS/MS method used here had the capability to individually detect and quantify two compounds in the same MS/MS segment. Ion trap - MS/MS optimization Tandem mass spectrometry is the process of carrying out two mass selective operations in series. The first step isolates the parent ion while the second isolates a parent ion fragment known as the product or quantitation ion. The collision-induced dissociation (CID) or fragmentation is effected by resonant/non-resonant excitation of the selected parent ion in the ion trap. To achieve a high fragmentation efficiency, the resonant/non-resonant excitation conditions need to be optimized (March, 1997). Some common instrumental parameters that are optimized to improve CID performance efficiency are (1) parent ion isolation window, (2) excitation amplitude, (3) CID time, (4) acquisition mass range, (5) broadband amplitude, (6) CID bandwidth, (7) modulation range, (8) filament current and (9) ion trap temperature (Larrazabal et al., 2004; Focant et al., 2001; Verenitch et al., 2007; Kuchler and Brzezinski, 2000).

Table 7 summarized default MS values that were not adjusted during method development. The acquisition mass range, optimal excitation amplitude, and the excitation storage level were determined for each compound using the automated method development (AMD) feature of the Varian software. The acquisition mass range was determined from the parent ion and quantitation ion mass-to-charge ratios (m/z) as described in detail below.

1) The first step in the determination of the ion trap-MS/MS settings was identification of the parent ion mass for each compound. To ensure effective identification of the parent ion mass and retention time, analyses of individual T&O compounds were initially conducted in single-stage CI-MS mode to obtain the total ion chromatogram (TIC, capturing all ions in the 40-650 m/z range). Analyses were conducted following headspace SPME of 10-mL samples containing 1, 10 or 50 µg L-1 of an individual T&O compound and 2.5 g NaCl. The TIC was used to identify both a first estimate of retention time and the most abundant ion, which was considered to be the parent ion. Figure 2 depicts a representative chromatogram, from which the retention time and parent ion mass for trans,trans-2,4-heptadienal were determined to be 34.96 minutes and m/z = 110.8, respectively. Parent ion chromatograms and mass spectra for all 19 T&O compounds targeted in this study are shown in the Appendix, and parent ion masses are summarized in Table 11.

2) Next, the captured parent ion in the ion trap needed to be effectively fragmented by CID to obtain the quantitation ion. The AMD feature of the Varian Software helps determine the resonant excitation voltage that will optimally fragment the parent ion into the quantitation ion. As illustrated for trans,trans-2,4-heptadienal in Figures 3 to 5, increasing the resonant excitation voltages from 0.0 to 0.4 and 0.6 V led to the appearance of the quantitation ion (m/z = 92.8) and the disappearance of the parent ion (m/z = 110.8). To identify the optimal resonant excitation voltage, AMD tests were conducted in two stages. In the first stage, a wider range of resonant excitation voltages was evaluated, and voltage increments were broader. In the second stage, AMD analysis was done with smaller voltage increments to pinpoint the best voltage for optimal fragmentation. Figures 6 and 7 illustrate example results of the first and second stage AMD analyses, respectively. Figure 7 shows that 0.60 volts was the optimal resonant excitation voltage for trans,trans-2,4-heptadienal because it maximized the production of the quantitation ion (m/z = 92.8). Quantitation ions and retention times obtained from the first- and second- stage AMD analyses are summarized in Table 12, and AMD voltage scan figures for each compound are shown in the Appendix. Retention times in Table 12 were used to define start and end times of the MS/MS segment for each target analyte (Table 13). The AMD results were used to calculate the excitation storage level for each compound; for this purpose, the molecular weight of each compound was used, and the qz value was set to 0.40 in the Varian Software. The qz value is defined as the stability factor in the z-direction by the Mathieu equation, which accounts for the ion motion in the trap (March, 1997). Once all resonant excitation voltages and excitation storage levels were determined, the following parameters were entered for each MS/MS segment using the MS method editor window (in the method builder option of the MS workstation software): parent ion mass, acquisition mass range, optimal excitation voltage, and duration of each segment. Table 13 summarizes individual MS/MS segment parameters.

kCounts 3-5-2009 t,t-2,4,hep t&o-ci-spme.SMS Ions: 111.0 40:650 1A 150

125

100

10 20 30 40 50 minutes

Spectrum 1A BP: 110.8 34.960 min, Scan: 3518, 40:650, Ion: 288 us, RIC: 166691, BC 110.8 58412 100%

75%

152.9 31217

50%

25%

66.9 4740 154.0 3784

100 200 300 400 500 600 m/z Figure 2. Parent ion chromatogram (top) and mass spectrum (bottom) of trans,trans-2,4- heptadienal.

Table 11. Molecular weight and parent ion for each target analyte.

Concentration Molecular weight Parent ion Compound (µg/L) (Da) (m/z)

dimethyl disulfide 50 94.2 95 hexanal-d12 10 112.23 92.8 1-hexanal 10 100.16 82.8

cis-4-heptenal 50 112.17 95 1-heptanal (heptaldehyde) 10 114.19 96.7 dimethyl trisulfide 10 126.26 127 cis-3-hexenyl acetate 50 142.2 83 trans,trans-2,4-heptadienal 50 110.15 110.8 2-isopropyl-3-methoxypyrazine 1 152.19 153 trans-2,cis-6-nonadienal 50 138.2 121 2-isobutyl-3-methoxypyrazine 1 166.22 167 methylisoborneol-d3 1 171.30 153.8 2-methylisoborneol 1 168.3 151 β-cyclocitral 1 152.23 153 2,3-benzopyrrole (indole) 50 117.15 118 trans,trans-2,4-decadienal 10 152.23 153 2,4,6-trichloroanisole 1 211.5 211 2,3,6-trichloroanisole 1 211.5 211 geosmin 1 182.3 165 β-ionone 1 192.3 193 2,4,6-tribromoanisole 10 344.83 345

MCounts 6-13-2009 t,t,2,4,hepta-1 t&o-ci-amd17.SMS 111.0>80:150 [0.00V] 111.0>80:1501A [0.00V] 5

10 20 30 40 50 minutes Spectrum 1A 30.188 min, Scan: 2936, 111.0>80:150 [0.00V], Ion: 75 us, RIC: 5.161e+6, BC 110.8 4.842e+6 100%

75%

50%

25%

80 90 100 110 120 130 140 150 m/z Figure 3. AMD result for 50 µg/L trans,trans-2,4-heptadienal at a resonant excitation voltage of 0 volts. The parent ion m/z (m/z = 110.8) dominates at the lower voltages.

MCounts 6-13-2009 t,t,2,4,hepta-1 t&o-ci-amd17.SMS 111.0>80:150 [0.40V] 111.0>80:150 [0.40V] 3.5

1A 3.0

2.5

2.0

1.5

1.0

0.5

0.0

10 20 30 40 50 minutes Spectrum 1A 30.208 min, Scan: 2938, 111.0>80:150 [0.40V], Ion: 75 us, RIC: 3.586e+6, BC 110.8 2.261e+6 100%

75%

50%

92.8 809700

25%

111.8 177514

80 90 100 110 120 130 140 150 m/z Figure 4. AMD result for 50 µg/L trans,trans-2,4-heptadienal at a resonant excitation voltage of 0.40 volts. The quantitation (product) ion m/z (m/z = 92.8) first appears at this voltage.

MCounts 6-13-2009 t,t,2,4,hepta-1 t&o-ci-amd17.SMS 111.0>80:150 [0.60V] 2.0 111.0>80:1501A [0.60V]

1.5

1.0

0.5

0.0

10 20 30 40 50 minutes Spectrum 1A 30.218 min, Scan: 2939, 111.0>80:150 [0.60V], Ion: 75 us, RIC: 1.971e+6, BC 92.8 956967 100%

75%

50% 110.8 384696

90.9 300131

25%

111.8 82.7 111461 86246

80 90 100 110 120 130 140 150 m/z

Figure 5. AMD result for 50 µg/L trans,trans-2,4-heptadienal at a resonant excitation voltage of 0.60 volts. The quantitation (product) ion m/z (m/z = 92.8) dominates at this voltage. 34

1.2E+06

1.0E+06

8.0E+05

6.0E+05

4.0E+05 Ion countsIon(m/z)

2.0E+05

0.0E+00 0 0.5 1 1.5 2 Excitation Voltage (volts)

Figure 6. Duplicate first-stage AMD results for trans,trans-2,4-heptadienal by varying the resonant excitation voltage from 0 to 2 volts.

9.0E+05 8.0E+05

7.0E+05

6.0E+05 5.0E+05 4.0E+05

3.0E+05 Ioncounts (m/z) 2.0E+05 1.0E+05 0.0E+00 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Excitation Voltage (volts)

Figure 7. Duplicate second-stage AMD results for trans,trans-2,4-heptadienal by varying resonant excitation voltage from 0.2 to 1.4 volts.

Table 12. Parent and quantitation ions as well as retention time range of peak base for each target analyte.

Retention time (min) Retention time (min) Compound Parent ion (m/z) Quantitation ion (m/z)

1st AMD analysis 2nd AMD analysis 80.7 dimethyl disulfide 95 4.211 - 4.349 4.063 - 4.147 hexanal-d12 92.8 59.8 6.182 - 6.281 6.041 - 6.143 1-hexanal 82.8 54.8 6.503 - 6.59 6.339 - 6.459 cis-4-heptenal 95 66.8 14.935 - 15.064 14.663 - 14.809 1-heptanal (heptaldehyde) 96.7 68.7 15.179 - 15.295 15.053 - 15.136 dimethyl trisulfide 127 92.7 24.589 - 24.707 24.382 - 24.560 cis-3-hexenyl acetate 83 54.8 29.925 - 30.115 29.816 - 29.939 trans,trans-2,4-heptadienal 110.8 92.8 30.158 - 30.272 30.079 - 30.185 2-isopropyl-3-methoxypyrazine 153 120.8 34.961 - 35.058 34.928 - 35.023 trans-2,cis-6-nonadienal 121 92.8 38.158 - 38.256 38.111 - 38.196

2-isobutyl-3-methoxypyrazine 167 124.8 39.071 - 39.163 38.999 - 39.085 97.8 methylisoborneol-d3 153.8 39.389 - 39.487 39.348 - 39.439 2-methylisoborneol 151 94.8 39.477 - 39.569 39.411 - 39.504 β-cyclocitral 153 108.8 40.796 - 40.884 40.796 - 40.884 2,3-benzopyrrole (indole) 118 90.8 43.592 - 43.686 43.604 - 43.688 trans,trans-2,4-decadienal 153 134.8 44.586 - 44.681 44.521 - 44.610 2,4,6-trichloroanisole 211 196 44.703 - 44.768 44.626 - 44.709 2,3,6-trichloroanisole 211 176 46.176 - 46.26 46.126 - 46.221 geosmin 165 108.8 47.683 - 47.778 47.666 - 47.751

β-ionone 193 174.8 49.545 - 49.63 49.545 - 49.633 265.8 2,4,6-tribromoanisole 345 51.16 - 51.245 51.163 - 51.245

Table 13. MS/MS segment settings for each target analyte.

Segment Compound Emission Multiplier Low High Parent Excitation Excitation Scan time Ionization Ion Prep boundaries (Molecular current offset mass mass ion mass storage level amplitude (sec)** Mode Technique (min)* , + weight) (µA)** (volts)** (m/z)* (m/z)* (m/z)*** (m/z)* (volts)*

dimethyl disulfide 3.65-5.50 15 300 0.600 70 110 CI Auto MRM 95.0 41.6 1.70 (94.20)

hexanal-d12 5.50-7.50 15 300 0.600 40 110 CI Auto MRM 93.0 40.7 0.60 (112.23)

1-hexanal CI Auto MRM 83.0 36.3 0.50 (100.16)

cis-4-heptenal 13.25-16.25 15 300 0.600 50 110 CI Auto MRM 95.0 41.6 0.50 (112.17)

heptaldehyde CI Auto MRM 97.0 42.5 0.60 (114.19)

dimethyl trisulfide 22.25-24.00 15 300 0.600 80 150 CI Auto MRM 127.0 55.8 0.40 (126.26)

cis-3-hexenyl 28.75-31.50 15 300 0.600 40 120 CI Auto MRM 83.0 36.3 0.50 acetate (142.20)

t,t-2,4-heptadienal CI Auto MRM 111.0 48.7 0.60 (110.15)

Table 13. Continued

2-isopropyl-3- 33.75-35.00 methoxypyrazine 15 300 0.600 100 200 CI Auto MRM 153.0 67.3 0.60 (152.19)

trans-2,cis-6- 37.30-38.40 nonadienal 15 300 0.600 80 150 CI Auto MRM 121.0 53.1 0.40 (138.21)

2-isobutyl-3- 38.40-38.88 methoxypyrazine 15 300 0.600 100 200 CI Auto MRM 167.0 73.5 0.50 (166.22)

MIB-d3 38.88-40.15 15 300 0.600 80 200 CI Auto MRM 154.0 67.7 0.60 (171.30)

MIB CI Auto MRM 151.0 66.4 0.60 (168.28)

β-cyclocitral 40.15-41.00 15 300 0.600 90 200 CI Auto MRM 153.0 67.3 0.65 (152.23)

indole 42.75-43.80 15 300 0.600 80 150 CI Auto MRM 118.0 51.8 0.60 (117.15)

t,t-2,4-decadienal 43.80-45.00 15 300 0.600 120 230 CI Auto MRM 153.0 67.3 0.60 (152.23)

2,4,6- trichloroanisole CI Auto MRM 211.0 92.9 0.50 (211.47)

Table 13. Continued

Segment Compound Emission Multiplier Low High Parent Excitation Excitation Scan time Ionization Ion Prep boundaries (Molecular current offset mass mass ion mass storage level amplitude (sec)** Mode Technique (min)*, + weight) (µA)** (volts)** (m/z)* (m/z)* (m/z)*** (m/z)* (volts)*

2,3,6- 45.00-46.20 trichloroanisole 15 300 0.600 150 230 CI Auto MRM 211.0 92.9 0.55 (211.47)

geosmin 46.70-47.50 15 300 0.600 100 200 CI Auto MRM 165.0 72.6 0.50 (182.30)

β-ionone 48.80-49.60 15 300 0.600 150 210 CI Auto MRM 193.0 85.0 0.70 (192.30)

2,4,6- 50.80-51.40 tribromoanisole 15 300 0.600 250 350 CI Auto MRM 345.0 152.2 0.80 (344.83)

* Determined from AMD

+ Segment boundaries were adjusted throughout the course of the project because the GC column was cut periodically for maintenance purposes. As a result, segment boundaries in Table 13 do not align perfectly with the retention times shown in Table 12.

** Indicates default setting

*** Determined from single-stage CI-MS analysis

Cells in grey indicate that the same segment was used for the analyses of two compounds.

4.2 Calibration curves Calibration curves were prepared with seven calibration standards. Concentrations in the calibration standards ranged from 0.1 - 50 ng/L for Group 1 compounds and from 10 - 4000 ng/L for Group 2 compounds (see also Table 10). The internal standards MIB-d3 (10 ng/L) and hexanal-d12 (500 ng/L) were spiked into each calibration standard to obtain the response factors (RF) for the Group 1 and Group 2 compounds, respectively. The response factor was obtained using the formula: Response factor (RF) = Peak area of the compound / Peak area of the internal standard A signal-to-noise (S/N) ratio of at least 10:1 was used as the criterion for quantifying compound concentrations. Peak areas of quantitation ions obtained for each target analyte and internal standard were determined using the MS workstation software (Varian, Inc.). The RF was plotted against the respective concentration of each compound to obtain a set of calibration curves for all 19 T&O compounds targeted in this study. Calibration curves were developed every time a new SPME fiber was used. The lower ends of calibration curves were based on the limit of quantitation (LOQ) values for each compound. The calibration curves were linear on a log-log plot (Table 14), and power equations (RF= a* Cb) were used to describe the calibration data. At least five points were used to generate the calibration curve for each compound, and Tables 15 and 16 summarize the regression equations that relate the RF to the concentration of each compound for two different SPME fibers. Calibration curves for each compound are shown in the Appendix. As indicated by the coefficient of determination (r2) values in Tables 15 and 16, the calibration equations effectively described the trends in the calibration data (the lowest r2 value was 0.950). Non- linear calibration curves were also obtained by Sclimenti and Krasner (2003) for dimethyl disulfide and 1-hexanal.

Table 14. Representative calibration curves, regression models, and compounds with calibration curves similar to the one shown.

Calibration curve Regression Model Compounds

IPMP 100.00 Group 1 10.00 Low concentration (0.1 - 50 ng/L):

1.11 1.00 RF = 1.08*C RF R² = 0.992 Power Law IPMP, IBMP, MIB, 0.10 β-cyclocitral, 0.01 2,4,6-TCA, 2,3,6-TCA, 0.1 1.0 10.0 100.0 geosmin, β-ionone, 2,4,6-TBA Concentration (ng/L)

trans-2,cis-6-nonadienal Group 2 100.00 High concentration 10.00 (10 - 4000 ng/L):

1.00 dimethyl disulfide, hexanal,

RF 1.32 RF = 0.000411*C Power Law cis-4-heptanal, heptaldehyde, R² = 0.991 0.10 dimethyl trisulfide, indole, cis-3-hexenyl acetate, 0.01 trans,trans-2,4-heptadienal, 10 100 1000 10000 trans-2,cis-6-nonadienal, Concentration (ng/L) trans,trans-2,4-decadienal

Table 15. Calibration curves for 19 T&O compounds: applicable concentration ranges, regression equations, and r2 values. Obtained with SPME fiber used in the sampling campaign for Meredith College Pond (Raleigh, NC).

Compound Concentration range (ng/L) Regression equation r2

dimethyl disulfide 50-4000 y = 1.56E-05x1.77 0.977

hexanal 50-4000 y = 4.54E-03x0.883 0.9704

cis-4-heptenal 50-4000 y = 0.485E-03x1.27 0.991

heptaldehyde 100-4000 y = 2.53E-03x1.05 0.987

dimethyl trisulfide 100-4000 y = 2.86E-07x2.33 0.976

cis-3-hexenyl cetate 100-4000 y = 1.51E-03x1.15 0.963

trans,trans-2,4-heptadienal 100-4000 y = 0.265E-03x1.509 0.992

2-isopropyl-3-methoxypyrazine 0.5-50 y = 1.08x1.11 0.992

trans-2,cis-6-nonadienal 100-4000 y = 0.411E-03x1.32 0.991

2-isobutyl-3-methoxypyrazine 0.1-50 y = 0.704x1.03 0.995

2-methylisoborneol 1-50 y = 0.207x1.07 0.992

β-cyclocitral 0.1-50 y = 2.11x0.891 0.979

indole 100-4000 y = 0.105E-03x1.42 0.996

trans,trans-2,4-decadienal 100-4000 y = 0.382E-03x1.408 0.992

2,4,6-trichloroanisole 0.5-50 y = 0.159x1.12 0.987

2,3,6-trichloroanisole 0.5-50 y = 0.366x1.09 0.994

geosmin 1-50 y = 0.259x1.04 0.988

β-ionone 0.1-50 y = 1.05x0.802 0.962

2,4,6-tribromoanisole 1-50 y = 0.015x1.51 0.994

Table 16. Calibration curves for 19 T&O compounds: applicable concentration ranges, regression equations, and r2 values. Obtained with SPME fiber used in the sampling campaign for Twin Lakes Pond (Burlington, NC).

Compound Concentration range (ng/L) Regression equation r2

dimethyl disulfide 50-4000 y = 3.39E-03x1.08 0.987

hexanal 50-4000 y = 4.83E-03x0.902 0.974

cis-4-heptenal 50-4000 y = 2.96E-03x1.28 0.993

heptaldehyde 100-4000 y = 1.34E-03x1.107 0.986

dimethyl trisulfide 100-4000 y = 9.02E-05x1.56 0.986

cis-3-hexenyl cetate 100-4000 y = 2.28E-03x1.07 0.988

trans,trans-2,4-heptadienal 100-4000 y = 0.522E-03x1.43 0.994

2-isopropyl-3-methoxypyrazine 0.5-50 y = 1.14x1.09 0.997

trans-2,cis-6-nonadienal 100-4000 y = 0.3304E-03x1.29 0.997

2-isobutyl-3-methoxypyrazine 0.1-50 y = 3.04x0.854 0.973

2-methylisoborneol 1-50 y = 0.187x1.08 0.999

β-cyclocitral 0.1-50 y = 2.52x0.849 0.988

indole 100-4000 y = 0.179E-03x1.42 0.970

trans,trans-2,4-decadienal 100-4000 y = 0.748E-03x1.33 0.996

2,4,6-trichloroanisole 0.5-50 y = 0.242x0.843 0.957

2,3,6-trichloroanisole 0.5-50 y = 0.781x0.804 0.962

geosmin 1-50 y = 0.274x1.03 0.998

β-ionone 0.1-50 y = 1.45x0.779 0.994

2,4,6-tribromoanisole 1-50 y = 8.08E-03x1.57 0.961

4.3 Limits of detection and quantitation One of the simplest and widespread approaches to determine the detection and quantification limits in GC methods is based on a comparison of the sample signal to the baseline noise obtained with a blank sample (Vial and Jardy, 1999). Using that approach here, the limits of detection and quantification were based on the peak-to-peak noise (N) measured on the base line close to an analyte’s chromatographic peak. As shown in the Table 17 the limit of detection (LOD) was based on the concentration that gave a chromatographic peak signal (S) that was at least 3 times the noise (N). Similarly, the limit of quantification (LOQ) was the concentration that gave a signal (S) that was at least 10 times the noise (N).

Table 17. LOD and LOQ regions (adapted from MacDougall et al., 1980)

Signal to noise ratio (S/N) Inference Nomenclature used

S/N* < 3 Analyte not detected (C < LOD) N/D

3 < S/N <10 Analyte detected but not quantifiable N/Q* (LOD < C < LOQ)

S/N > 10 Analyte detected at quantifiable levels Concentrations are (C > LOQ) provided

The LOQ of each compound is summarized in Table 18 and compared with its OTC. Of the 17 compounds with known OTCs, 12 compounds had LOQs below the OTCs. The LOQs of the following compounds were considerably lower than their OTCs: dimethyl disulfide, hexanal, heptaldehyde, cis-3-hexenyl acetate, trans,trans-2,4-heptadienal, β-cyclocitral and indole. In contrast those for MIB, 2,3,6-TCA, IBMP, geosmin and β-ionone were similar to their OTCs. In its current form, the developed GC-MS/MS method can detect but is not sufficiently sensitive to quantify dimethyl trisulfide, trans-2,cis-6-nonadienal, IPMP, 2,4,6-TCA, and 2,4,6-TBA at levels near the OTC.

Table 18. Comparison of the LOQs and OTCs of the targeted analytes.*

Compound LOQ (ng/L) OTC (ng/L)

dimethyl disulfide 50 <4,000

hexanal 50 4,500

cis-4-heptenal 50 ?

heptaldehyde 100 3,000

dimethyl trisulfide 100 10

cis-3-hexenyl acetate 100 1500

trans,trans-2,4-heptadienal 100 5,000

2-isopropyl-3-methoxypyrazine 0.5 0.2

trans-2,cis-6-nonadienal 100 4

2-isobutyl-3-methoxypyrazine 0.1 1

2-methylisoborneol 1 9

β-cyclocitral 0.1 3,000

indole 100 300,000

trans,trans-2,4-decadienal 100 ?

2,4,6-trichloroanisole 0.5 0.03

2,3,6-trichloroanisole 0.5 7

geosmin 1 4

β-ionone 0.1 7

2,4,6-tribromoanisole 1 0.03

* Shaded cells indicate the compounds with LOQs lower than their OTCs.

4.4 Analysis of North Carolina water samples Samples from surface waters experiencing algae blooms were collected at three locations in North Carolina. In addition, one drinking water source and distribution system was sampled. GC- MS/MS analyses were conducted to identify which target analytes were present in the samples and to determine the concentrations of the quantifiable compounds.

Meredith College pond (Raleigh, NC) During a bloom event, samples were collected on October 1, 2009 at two different locations from a pond on the Meredith College campus in Raleigh, NC. Both non-filtered and filtered (0.45 µm PVDF membrane) samples were analyzed. T&O compounds originating from algae are a result of two major biosynthesis patterns: (1) compounds that are produced during growth and (2) compounds that are released during cell senescence, death or mechanical damage (Watson, 2003). With the developed method, there is a strong possibility that cell damage will occur as a result of (1) NaCl addition (25% w/w) and/or (2) the heated agitation of the samples during the preconcentration procedure (T = 65oC). Hence from the comparison of results obtained for non- filtered and filtered samples, an estimate of the intracellular T&O compounds in the samples can be made. For the non-filtered samples, Figure 8 illustrates that the following compounds were present at quantifiable levels: 1-hexanal, cis-4-heptenal, trans,trans-2,4-heptadienal, trans-2,cis-6- nonadienal, trans,trans-2,4-decadienal, β-cyclocitral, geosmin and β-ionone. It should be noted that the concentrations of β-cyclocitral, geosmin and β-ionone greatly exceeded the upper limit of the calibration range; therefore, their concentrations were determined from 1:80 and 1:50 dilutions of samples A and B, respectively. In the filtered samples, 1-hexanal, trans-2,cis-6- nonadienal, β-cyclocitral, and β-ionone were present at quantifiable levels. Geosmin was present at >50 ng/L (upper limit of calibration curve) in the filtered sample but could not be quantified more exactly because dilutions of the filtered samples were not analyzed. In addition, the following compounds were detected at levels that were too low to be quantifiable in the filtered sample: 1-heptaldehyde and indole. Dimethyl disulfide, heptaldehyde and indole could be detected in the non-filtered samples but their concentrations were too low to be quantifiable.

10,000

1,000 1-hexanal cis-4-heptenal t, t-2,4-heptadienal t-2, cis-6-nonadienal > 50 ng/L > 50 ng/L t, t-2,4-decadienal β-cyclocitral* Concentration (ng/L) Concentration 100 geosmin*

β-ionone*

N/D N/Q

N/D

N/Q

N/D N/D 10 N/Q A A (filtered) B B (filtered)

* concentrations determined from diluted samples (see text)

Figure 8. Taste and odor compounds in water collected at two sampling points from Meredith College pond, Raleigh, NC.

A comparison of the results obtained with non-filtered and filtered samples (Table 19) suggests that a large percentage of β-cyclocitral and β-ionone was present in intracellular form. E.g., the concentration of β-cyclocitral in the non-filtered samples was about 40 times that in the filtered samples. Similarly, the β-ionone concentration in non-filtered samples was about 65 times that in the filtered samples. For geosmin, the peak areas of non-diluted samples were compared, and the comparison suggests that the geosmin concentration in the non-filtered sample was about 2 to 3 times that in the filtered samples. Relative to their OTCs geosmin (OTC = 4 ng/L) and β-ionone (OTC = 7 ng/L) were present at concentrations that exceeded the respective OTCs by factor of more than 100. In addition, trans-2,cis-6-nonadienal (OTC = 4 ng/L) was present at levels that exceeded its OTC by a factor of 30. It is likely that these compounds most strongly contributed to objectionable odor in the sample. A formal FPA analysis can help confirm which of these compounds produced odors and also determine their odor characteristics in this sample.

Table 19. Comparison of total T&O compound concentration and extracellular fraction (C non- filtered/C filtered) as well as comparison of measured and odor threshold concentrations (C measured/OTC) for T&O compounds detected in Meredith College pond samples.

Compound C non-filtered/C filtered C measured/OTC Sample A Sample B Sample A Sample B 1-hexanal 10 6 0.76 0.30 trans,trans-2,4-heptadienal 0.04 trans-2,cis-6-nonadienal 2.01 1.52 39 33 β-cyclocitral 36 36 0.34 0.32 geosmin 266 177 β-ionone 86 48 185 109

Twin Lakes pond (Burlington, NC) T&O compounds were identified and quantified in two samples collected on September 6, 2009 during a bloom event in an ornamental lake in Burlington, NC. The bloom was dominated by cyanobacteria of the genus Anabaena. A photo of the bloom and the dominant Anabaena cells are shown in Figure 9. Both non-filtered and filtered samples were analyzed. As shown in Figure 10, eleven compounds were present at measurable concentrations in non-filtered sample A. The concentrations of dimethyl disulfide, 1-hexanal, β-cyclocitral and β-ionone were determined from 1:5 dilutions of non-filtered sample A. Upon filtration of sample A (Figure 10, Table 20), the concentrations of three compounds (dimethyl trisulfide, trans,trans-2,4- heptadienal and indole) decreased to non-quantifiable levels (< 100 ng/L), and cis-3-hexenyl acetate was no longer detectable (S/N <3). This result suggests that these six compounds were primarily present inside the algal/cyanobacterial cells. Filtration also lowered the concentrations of dimethyl disulfide and 1-hexanal, but to a lesser extent. In contrast, the concentrations of MIB, β-cyclocitral, geosmin and β-ionone were essentially unaffected by filtration, suggesting that they were primarily present in the aqueous phase (i.e. as extracellular compounds).

(a)

(b)

Figure 9. Photos of (a) bloom at Twin Lakes pond and (b) dominant Anabaena species

10,000

DMDS

1-hexanal

1,000 cis-4-heptenal

> 250 ng/L DMTS

t,t,heptadienal

MIB 100 β-cyclocitral

indole

Concentration (ng/L) Concentration geosmin

10 β-ionone

cis-3-hexenyl acetate

N/Q

N/D

N/Q N/Q N/Q

N/Q

N/Q N/Q 1 N/Q

A A(filtered) B B(filtered)

Figure 10. Taste and odor compounds in water collected at two sampling points from Twin Lakes pond, Burlington, NC.

Results obtained with sample B (Figure 10) were largely similar to those obtained with sample A with the following exceptions: (1) the concentrations of dimethyl disulfide and 1-hexanal were lower, (2) cis-3-hexenyl acetate was not detectable in the non-filtered sample, and (3) the β- ionone concentration was high (> 250 ng/L) and not quantifiable in a 1:5 dilution of the non- filtered sample. Filtration did lead to measurable decreases in β-cyclocitral, geosmin and β- ionone concentrations. In contrast, MIB and 1-hexanal concentrations appeared to be unaffected by filtration (Figure 10, Table 20). The concentrations of the compounds in the non-filtered samples were compared to the corresponding OTCs to determine which compounds could contribute to the T&O characteristics of the sample (Table 20). The following compounds could all have contributed: dimethyl trisulfide was present at about 60 times its OTC (10 ng/L) in sample A and at about 9 times its OTC in sample B. The concentrations of MIB, geosmin, and β-ionone in sample A were greater than or similar to their respective OTCs (Table 20). Sample B had similar concentrations of MIB and geosmin, but β-ionone was present at much higher concentrations (> 250 ng/L).

Table 20. Comparison of total T&O compound concentration and extracellular fraction (C non- filtered/C filtered) as well as comparison of measured and odor threshold concentrations (C measured/OTC) for T&O compounds detected in Burlington pond samples.

Compound C non-filtered/C filtered C non-filtered/OTC Sample A Sample B Sample A Sample B dimethyl disulfide 2.5 0.39 0.05 1-hexanal 2.1 0.90 0.17 0.09 dimethyl trisulfide 59 8.9 cis-3-hexenyl acetate 0.11 trans,trans-2,4-heptadienal 0.09 2-methylisoborneol 1.2 1.4 1.3 1.4 β-cyclocitral 2.1 3.6 0.04 0.08 indole 0.003 0.0004 geosmin 2.2 5.5 8.7 8.6 β-ionone 5.6 >6.5 22 >35

Harris Lake (New Hill, NC) Figure 11 shows the compounds found in a bloom sample collected on October 4, 2009 from Harris Lake in New Hill, NC. The sample was analyzed after it was filtered to remove the floating plant material in the samples (visually identified to be Duckweed). Six compounds were present at measurable concentrations (Figure 11). In addition 1-hexanal was present at concentrations > 4,000 ng/L; i.e., the response factor obtained for the sample exceeded the upper limit of the standard curve. Apart from the compounds shown in Figure 11, the following compounds were detected (S/N>3), but were present at non-quantifiable levels: 1-heptanal, cis-3- hexenyl acetate, cis-4-heptenal, trans,trans-2,4-heptadienal, trans,trans-2,4-decadienal, IPMP, IBMP, 2,4,6-TCA and 2,3,6-TCA. The concentration profile showed the saturated and 51

unsaturated aldehydes to be present at ~100 times the concentrations of geosmin, β-cyclocitral and β-ionone. The odor characteristic of this sample was likely dominated by trans-2,cis-6- nonadienal, which was present at a concentration that exceeded its OTC by a factor of nearly 500 (Table 21). Geosmin and β-ionone were present at concentrations similar to their OTCs.

10,000 > 4,000 ng/L

1-hexanal

t-2, cis-6-nonadienal

1,000 β-cyclocitral

indole

t, t-2,4-decadienal

geosmin

β-ionone 100

Concentration (ng/L) Concentration

1 Harris Lake (filtered)

Figure 11. Taste and odor compounds in water collected from Harris Lake, New Hill, NC.

Table 21. Comparison of measured and odor threshold concentrations (C/OTC) for taste and odor compounds detected in Harris Lake sample.

Compound C/OTC 1-hexanal trans-2,cis-6-nonadienal 530 β-cyclocitral 0.002 indole 0.003 trans,trans-2,4-decadienal geosmin 1.2 β-ionone 1.7

E. M. Johnson drinking water treatment plant raw water and distribution system samples Falls Lake serves as the source for the E. M. Johnson drinking water treatment plant in Raleigh, NC. On September 17, 2009, a raw water sample was collected at the E. M. Johnson water treatment plant. In addition, a sample from a tap in the Environmental Engineering Research Laboratory at NC State University was collected on the same day to determine the effects of treatment and distribution on the identity and concentration of T&O compounds. The tap water sample was dechlorinated with 0.1N sodium thiosulfate immediately after sampling (about 500 μL 0.1N sodium thiosulfate solution was added to a 250-mL sample). As shown in Figure 12, only β-cyclocitral and geosmin were present at measurable concentrations in raw and tap water samples. In addition, the following compounds were detected (S/N > 3) in the source water at levels that were too low to be quantified: 1-hexanal, heptaldehyde, trans-2,cis-6-nonadienal, MIB, indole and trans,trans-2,4-decadienal. Apart from indole and trans,trans-2,4-decadienal, the same compounds were also detected in the finished water (S/N > 3). Geosmin was present at a concentration similar to its OTC (4 ng/L) while β-cyclocitral was present at a concentration much lower than its OTC of about 3,000 ng/L (Rashash et al., 1996). Geosmin and β-cyclocitral concentrations in the raw water and distribution system samples were similar, which suggests that the processes at the E. M. Johnson treatment plant did not effectively remove β-cyclocitral and geosmin. Furthermore, target analytes that were not detected in the source water did not appear at detectable levels in the finished water. Therefore, it can be concluded that at the time of sampling, none of the targeted taste and odor compounds were formed in the distribution system.

4 β-cyclocitral geosmin

3 Concentartion(ng/L) 2

0 Falls lake Distribution system

Figure 12. Taste and odor compounds in Falls Lake water (raw water for the EM Johnson water treatment plant, Raleigh, NC) and in Raleigh tap water.

5. Conclusions

 A GC-MS/MS method was developed with which 19 common T&O compounds can be analyzed in a single analysis.

 The calibration curves developed had high coefficients of determination (r2 > 0.950). The calibration data of both Group 1 and Group 2 compounds were best described with a power law.

 Limits of quantitation (LOQs) were approximately 1 ng/L for Group 1 compounds and 100 ng/L for Group 2 compounds. For 12 of the 17 compounds with known odor threshold concentrations (OTCs), the LOQs were at or below the OTC. Consequently, the developed analytical method is capable of detecting developing T&O problems for 12 common T&O compounds before consumers can detect objectionable tastes and odors in their water. Results from the developed analytical method therefore allow utilities to develop source water management and treatment strategies in a more proactive manner.

 To test the developed analytical method, algae/cyanobacteria bloom samples from three North Carolina ponds/lakes were analyzed. Of the compounds targeted in this study all compounds except 2,4,6-tribromoanisole were detected in the bloom samples. Geosmin, β-ionone and trans-2,cis-6-nonadienal were present at concentrations that greatly exceeded their respective OTCs. Effective geosmin and trans-2,cis-6-nonadienal removal during drinking water treatment can be accomplished by activated carbon adsorption and/or ozonation processes. Chlorination alone is also effective for trans-2,cis-6- nonadienal removal. The removal of β-ionone can be accomplished by ozonation, and the effectiveness of other treatment processes for β-ionone is not known to date.

 A comparison of results obtained for non-filtered and 0.45-µm filtered samples showed that many T&O compounds in the bloom samples were present as intracellular compounds. Conventional treatment (coagulation, flocculation, sedimentation) can be effective for the removal of intracellular compounds, and utilities should be careful with the application of preoxidants that can lead to the release of intracellular compounds that can otherwise be removed by conventional treatment.

 For Raleigh raw and tap water samples, there was no significant difference in the measured concentrations of β-cyclocitral and geosmin, suggesting no significant effect of both the treatment processes and distribution system on these compounds. Geosmin was present at a concentration that was near its OTC while the β-cyclocitral concentration was well below its OTC.

6. Recommendations for Future Work

In the developed analytical method, deuterated MIB (MIB-d3) and hexanal (hexanal-d12) were used as the internal standards to obtain the response factors. Future work should consider including additional stable isotope analogs of the targeted T&O compounds. The use of additional internal standards that more closely match compound classes that were not well represented by the current two internal standards may enhance accuracy and precision for such compounds as dimethyl disulfide, dimethyl trisulfide, and the dienals. Furthermore, in the IT-MS/MS optimization procedure, only the excitation amplitude was experimentally optimized. The optimization of other instrumental parameters such as the parent ion isolation window, CID time, broadband amplitude, CID bandwidth, modulation range, filament current and ion trap temperature should therefore be considered in future work. Finally, the use of different GC columns (e.g. columns designed for polar compounds) and a 2- cm DVB/CAR/PDMS SPME fiber should be evaluated in future work to see whether one or both can lower the LOQs and MDLs of those compounds, for which current analytical limits are above their OTCs.

7. References

American Public Health Association. 2005. Standard Methods for the Examination of Water and Wastewater, 21st edition. Arthur, C. and Pawliszyn, J. 1990. “Solid phase microextraction with thermal desorption using fused silica optical fibers.”Anal. Chem. 62(19):2145. AwwaRF and Lyonnaise des Eaux. 1995. Advances in Taste-and-Odor Treatment and Control. I.H. Suffet, J. Mallevialle, and E. Kawczynski, eds. American Water Works Association Research Foundation, Denver, CO. AwwaRF and Lyonnaise des Eaux. 1987. Identification and Treatment of Tastes and Odors in Drinking Water. J. Mallevialle and I.H. Suffet, eds. American Water Works Association Research Foundation, Denver, CO. Bjurman, J., Kristensson, J. 1992. “Production of volatile metabolites by the soft rot fungus Chaetomium globosum on building materials and defined media.” Microbios 72(290):47. Boerjesson, T., Stoellman, U.M., Schnuerer, J.L. 1993. “Off-odorous compounds produced by molds on oatmeal agar: Identification and relation to other growth characteristics.” J. Agric. Food Chem. 41(11):2104. Bruchet, A. 1999. “Solved and unsolved cases of taste and odor episodes in the files of inspector Cluzeau.” Water Science and Technology 40(6):15. Bruchet, A. and Duguet, J. 2004. “Role of oxidants and disinfectants on the removal, masking and generation of tastes and odours.” Water Science and Technology 49(9):297. Buttery, R.G and Ling, L.C. 1973. ”Earthy aroma of potatoes.” J. Agric. Food. Chem 21(4):745. Buttery, R.G, Guadagni, D.G, Ling, L.C. 1978 . “Volatile aroma components of cooked artichoke.” J. Agric. Food Chem. 26 (4):791. Buttery, R.G, Guadagni, D.G, Ling, L.C, Seifer R.M. and Lipton W. 1976. “Additional volatile components of cabbage, broccoli, and cauliflower.” J. Agric. Food Chem. 24(4):829. Cancho, B., Ventura F., Galceran M. 1999 . “Behavior of halogenated disinfection by-products in the water treatment plant of barcelona, spain.” Bull. Environ. Contam. Toxicol. 63(5):610. Collins, R.P. and Kalnins, K. 1965 . “Volatile constituents produced by the alga Synura petersenii - II. alcohols, esters and acids.” Air Water. Pollut. 9(7-8):501. Cook, D. and Newcombe, G. 2004. “Can we predict the removal of MIB and geosmin with PAC by using water quality parameters?” Water, science & technology: water supply 4(4):221.

Cotsaris, E., Bruchet, A., Mallevialle J., Bursill D.B. 1995. “The identification of odorous metabolites produced from algal monocultures.” Water Science and Technology 31(11):251. Delahunty, C.M., Eyres, G., Dufour, J.P. 2006. “Gas chromatography-olfactometry.” Journal of Separation Science 29(14):2107. Diaz, A, Fabrellas, C, Ventura, F. 2005. “Determination of the odor threshold concentrations of chlorobrominated anisoles in water.” J. Agric. Food Chem. 53(2):383. Dietrich, A.M., Hoehn, R.C., Dufresne, L.C., Buffin, L.W., Rashash, D.M.C. and Parker, B.C. 1995. “Oxidation of odorous and nonodorous algal metabolites by permanganate, chlorine, and chlorine dioxide.” Water Sci. Tech. 31(11): 223-228. Fabrellas, C., Cardenoso, R., Devesa, R., Flores, J., Matia, L. 2004. “Taste and odor profiles (off- flavors) in the drinking waters of the Barcelona area (1996-2000).” Water Science and Technology 49(9):129. Gerber, N.N. 1979. “Volatile Substances from Actinomycetes: Their Role in the Odor Pollution of Water.” Crit. Rev. Microbiol. 7 (3):191.

Gerber, N.N. 1969. “A volatile metabolite of actinomycetes, 2-methylisoborneol.” J. Antibiot. 22(10):508. Gerber, N.N. 1968. “Geosmin from microorganisms is trans-1,10-dimethyl-trans-9-decalol.” Tetrahedron Lett. (25):2971. Gerber, N.N. and Lechevalier, H.A.1965.“Geosmin, an earthy-smelling substance isolated from actinomycetes.” Appl. Microbiol. 13(6):935. Gordon, D. T. and Morgan, M. E. 1972. “Principal Volatile Compounds in Feed Flavored Milk.” J. Dairy Sci. 55(7):905. Guadagni, D.G and Buttery, R.G. 1978. “Odor threshold Of 2,3,6-trichloroanisolein water.” J. Food Sci. 43(4):1346. Henatsch, J. J. and Jüttner, F. 1983. “Volatile odorous excretion products of different strains of Synechococcus (cyanobacteria).” Water Science and Technology 15(6-7):259. Juttner, F. 1995. “Physiology and biochemistry of odorous compounds from freshwater cyanobacteria and algae.” Water Science and Technology 31(11):69. Juttner, F. 1984. “Characterization of Microcystis strains by alkyl sulfides and beta cyclocitral.” Zeitschrift für Naturforschung C 39(9-10):867. Juttner, F. 1984. “Dynamics of the Volatile Organic Substances Associated with Cyanobacteria and Algae in a Eutrophic Shallow Lake.” Appl. Environ. Microbiol. 47(4):814. Juttner, F. 1983. “Volatile odorous excretion products of algae and their occurrence in the natural aquatic environment.” Water Science and Technology 15(6-7):247.

Juttner, F. 1981. “Detection of Lipid Degradation Products in the Water of a Reservoir During a Bloom of Synura uvella.” Appl. Environ. Microbiol. 41(1):100. Juttner, F. 1976. “Beta-cyclocitral and alkanes in Microcystis (Cyanophyceae).” Zeitschrift für Naturforschung C 31(9-10):491. Juttner, F. and Watson S.B. 2007. “Biochemical and ecological control of geosmin and 2- methylisoborneol in source waters.” Appl. Environ. Microbiol. 73(14):4395. Juttner, F. and Hahne, B. 1981. “Volatile excretion products of Poterioochromonas malhamensis. Identification and formation.” Zeitschrift für Pflanzenphysiologie 103(5):403. Juttner, F. and Muller, H. 1979. “Excretion of octadiene and octatrienes by a freshwater diatom.” Naturwissenschaften 66(7):363. Juttner, F, Höflacher, B. and Wurster, K. 1986. “Seasonal analysis of volatile organic biogenic substances (Vobs) in fresh-water phytoplankton populations dominated by Dinobryon, Microcystis and Aphanizomenon.” J. Phyco. 22(2):169. Khiari D., Barrett S.E., Suffet I. 1997. “Sensory GC analysis of decaying vegetation and septic odors.” American Water Works Association.Journal 89(4):150. Khiari, D., Suffet, I.H., Barrett, S.E. 1995. “Extraction and identification of chemicals causing grassy odors in flavor profile analysis (FPA) reference standards.” Water Science and Technology 31(11):93-8. Khiari D, Suffet I.H., Barrett S.E. 1995. “The determination of compounds causing fishy/swampy odors in drinking water supplies.” Water Science and Technology 31(11):105-12. Khiari, D., Brenner, L., Burlingame, G.A. and Suffet, I.H. 1992. “Sensory gas chromatography for evaluation of taste and odor events in drinking water.” Water Science and Technology 25(2):97. Khiari D., Bruchet A., Gittelman T., Matia L., Barrett S.E. and Suffett I.H. 1999. “Distribution- generated taste-and-odor phenomena.” Water Science and Technology 40(6):129. Khiari, D, Young, C.C, Amah, G., Ye, Q., Atasi K., Huddleston, J., Suffet, I.H. 1999. “Factors affecting the stability and behavior of grassy odors in drinking water.” Water Science and Technology 40(6):287-92. Kikuchi, T., Mimura, T., Moriwaki, Y., Ando, M., Negoro, K. 1974. “Metabolites of a diatom, Synedra rumpens Kutz, isolated from water in Lake Biwa. Identification of odorous compounds, n-hexanal and n-heptanal, and analysis of fatty acids.” Chem. Pharm. Bull 22(4):945. Krasner,S., Hwang, C.J. and McGuire, M.J. 1983. “A standard method for quantification of earthy-musty odorants in water, sediments and algal cultures.” Water Science and Technology 15(6-7):127. 59

Malleret L and Bruchet A. 2002. “A taste and odor episode caused by 2,4,6-tribromoanisole.” American Water Works Association Journal 94(7):84. Malleret L., Bruchet A., Hennion M. 2001. “Picogram determination of "earthy-musty" odorous compounds in water using modified closed loop stripping analysis and large volume injection GC/MS.” Analytical Chemistry 73(7):1485. March, R.E. 1997. “An introduction to quadrupole ion trap mass spectrometry.” Journal of Mass Spectrometry 32(4):351. MacDougall, D., Amore, F.J., Cox, G.V., Estes, F.L. 1980. “Guidelines for data acquisition and data quality evaluation in environmental chemistry.” Anal. Chem. 52(14):2249. McGuire, M. 1999. “Advances in treatment processes to solve off-flavor problems in drinking water.” Water Science and Technology 40(6):153. Miller III, A., Scanlan, R. A., Lee J. S. and Libbey L. M. 1973. “Identification of the Volatile Compounds Produced in Sterile Fish Muscle (Sebastes melanops) by Pseudomonas fragi.” Appl. Microbiol. 25(6):952. Murray, K.E., and Whitfield, F.B. 1975. “The occurrence of 3-alkyl-2-methoxypyrazines in raw vegetables.” J. Sci. Food. Agric. 26(7):973. Pan, L. and Pawliszyn, J. 1997. “Derivatization/solid-phase microextraction: New approach to polar analytes.” Anal. Chem. 69(2):196. Peter, A. and von Gunten, U. 2009. “Taste and odour problems generated in distribution systems: A case study on the formation of 2,4,6-trichloroanisole.” Aqua 58(6):386. Peter, A. and Von Gunten, U. 2007. “Oxidation kinetics of selected taste and odor compounds during ozonation of drinking water.” Environ. Sci. Tech. 41(2):626. Peter, A., Koster O, Schildknecht A, von Gunten U. 2009. “Occurrence of dissolved and particle- bound taste and odor compounds in Swiss lake waters.” Water Res. 43(8):2191. Pirbazari, M. , Borow, H.S. Craig, S., Ravindran, V. and McGuire, M.J. 1992. “Physical chemical characterization of five earthy-musty-smelling compounds.” Water Science and Technology 25(2):81. Raschke, R.L., Carroll, B. and Tebo, L.B. 1975. “The relationship between substrate content, water quality, actinomycetes and musty odours in the broad river basin.” J. Appl. Ecol. 12(2):535-60. Rashash, D.M.C., Dietrich, A.M., Hoehn R.C. 1997. “FPA of selected odorous compounds.” Jounral American Water Works Association 89(4):131. Rashash, D.M.C., Dietrich, A.M.,Hoehn, R.C. and Parker, B.C. 1995. “The influence of growth conditions on odor-compound production by two chrysophytes and two cyanobacteria.” Water Science and Technology 31(11):165.

Rashash, D.M.C.; R.C. Hoehn; A.M. Dietrich; T.J. Grizzard; and B.C. Parker. 1996. Identification and Control of Odorous Algal Metabolites. American Water Works Association Research Foundation, Denver, CO. Rigal, S. and Danjou, J. 1999. “Tastes and odors in drinking water distribution systems related to the use of synthetic materials.” Water Science and Technology 40(6):203. Schierberle, P., Ofner, S., Grosch, W. 1990. “Evaluation of Potent Odorants in Cucumbers (Cucumis sativus) and Muskmelons (Cucumis melo) by Aroma Extract Dilution Analysis.” J. Food Sci. 55(1):193. Sclimenti, M.J. and S.W. Krasner. 2003. “The development of a solid-Phase microextraction technique for determining fishy, swampy, or grassy odors in drinking water.” In Proc. 2003 AWWA WQTC, Philadephia, PA. Skoog, D.A., and Leary, J.J. 1992. Principles of Instrumental Analysis, 4th edition, Saunders College Publishing. Suffet,I.H., Schweitze,L., Khiari, D. 2004. “Olfactory and chemical analysis of taste and odor episodes in drinking water supplies.” Rev. Env. Sci. Biotech. 3(1):3. Suffet, I.H., Khiari,D., Bruchet, A. 1999. “The drinking water taste and odor wheel for the millennium: Beyond geosmin and 2-methylisoborneol.” Water Science and Technology 40(6):1 Suffet, I. H., Corado, A., Chou, D., McGuire, M. Butterworth, S. 1996. “AWWA Taste and Odor Survey.” Journal AWWA 88(4): 168. Tabachek, J. and Yurkowsk,i M. 1976. “Isolation and identification of blue-green algae producing muddy odor metabolites, geosmin, and 2-methylisoborneol, in saline lakes in Manitoba.” Journal of the Fisheries Research Board of Canada 33(1):25. Tomboulian, P., Schweitzer, L., Mullin, K., Wilson, J. and Khiari D. 2004. “Materials used in drinking water distribution systems: contribution to taste-and-odor.” Water Science and Technology 49(9):219. U.S. Environmental Protection Agency, EPA method 40 CFR - Part 136 (Appendix B). 2003. Van Gemert, L.J. and Nettenbreijer, A.H. 1977. Compilation of Odour Threshold Values in Air and Water. National Institute of Water Supply, Voorburg, The Netherlands. Vial, J. and Jardy, A. 1999. “Experimental Comaprison of the different approaches to estimate LOD and LOQ of an HPLC method.” Anal. Chem. 71(14):2672. Watson, S. 2004. “Aquatic taste and odor: A primary signal of drinking-water integrity.” Journal of Toxicology and Environmental Health. Part A 67(20-22):1779. Watson, S. 2003. “Cyanobacterial and eukaryotic algal odour compounds: Signals or byproducts? A review of their biological activity.” Phycologia 42(4):332.

Watson, S., Satchwill, T., Dixon, E., McCauley, E. 2001. “Under-ice blooms and source-water odour in a nutrient-poor reservoir: Biological, ecological and applied perspectives.” Freshwater Biol. 46(11):1553. Watson, S., Brownlee, B., Satchwill, T. and McCauley E. 1999. “The use of solid phase microextraction (SPME) to monitor for major organoleptic compounds produced by chrysophytes in surface waters.” Water Science and Technology 40(6):251. Wilkins, K.. 1996. “Volatile metabolites from actinomycetes.” Chemosphere 32(7):1427. Wisconsin Department of Natural Resources. 1996. Analytical Detection Limit Guidance. Yano, H., Nakahara, M. and Ito, H. 1988. “Water blooms of Uroglena americana and the identification of odorous compounds.” Water Science and Technology 20(8-9):75. Young, W.F.; H. Horth; R. Crane; T. Ogden; and M. Arnott. 1996. “Taste and odour threshold concentrations of potential potable water contaminants.” Water Res. 30(2): 331-340. Young, C.C. and I.H. Suffet. 1999. “Development of a standard method-analysis of compounds causing tastes and odors in drinking water.” Water Sci. Tech. 40(6): 279-285. Zaitlin, B., Watson, S., Ridal, J., Satchwill, T., Parkinson, D. 2003. “Actinomycetes in lake ontario: Habitats and geosmin and MIB production.” Journal American Water Works Association 95(2):113. Zaitlin, B. and Watson, S.B. 2006. “Actinomycetes in relation to taste and odour in drinking water: Myths, tenets and truths.” Water Res. 40(9):1741-53.

Appendix 1 – Abbreviations

2,3,6-TCA 2,3,6-trichloroanisole 2,4,6-TBA 2,4,6-tribromoanisole 2,4,6-TCA 2,4,6-trichloroanisole ACD Labs Advanced Chemistry Development ACS American Chemical Society AMD automated methods development AOP Advanced Oxidation Processes APHA American Public Health Association AWWA American Water Works Association AwwaRF American Water Works Association Research Foundation CAS Number Chemical Abstracts Service registry number CI chemical ionization CID collision-induced dissociation CLLE continuous liquid-liquid extraction CLSA closed loop stripping analysis DBPs disinfection byproducts DI deionized water DPD diethyl-p-phenylene diamine DVB/CAR/PDMS divinylbenzene/ carboxenTM/ polydimethylsiloxane ECD electron capture detector EI electron ionization FID flame ionization detector FPA flavor profile analysis FPD flame photometric detector GAC granular activated carbon GC-MS/MS gas chromatography – tandem mass spectrometry HSPME headspace solid-phase microextraction IBMP 2-isobutyl-3-methoxypyrazine IPMP 2-isopropyl-3-methoxypyrazine IS internal standard IT-MS/MS ion trap tandem mass spectrometry LC-MS liquid chromatography – mass spectrometry LLE liquid-liquid extraction LLE liquid- liquid extraction LLME liquid-liquid microextraction LODs limits of detection LOQs limits of quantitation m/z mass/charge ratio 63

MIB 2-methylisoborneol MRM multiple reaction monitoring MS mass spectrometer, mass spectrometry N/D not detectable, S/N < 3 N/Q not quantifiable, 3 < S/N <10 NOM natural organic matter OSA open stripping analysis OTC odor threshold concentration PTFE polytetrafluoroethylene = Teflon PUFA polyunsaturated fatty acid PVC polyvinyl chloride PVDF polyvinylidene fluoride qz Stability factor r2 coefficient of determination RF Response factor = Peak area of compound / Peak area of internal standard S/N signal-to-noise ratio SDE simultaneous distillation extraction SPE solid-phase extraction SPME solid-phase microextraction T&O taste and odor TIC total ion count TON threshold odor number

Appendix 2 – Publication

Viswakumar, A. 2010. Development of a gas chromatography-tandem mass spectrometry method for the simultaneous analysis of 19 taste and odor compounds. M.S. Thesis, NC State University, Raleigh, NC.

Appendix 3 – Mass Spectra and Calibration Curves for Taste and Odor Compounds

kCounts 3-5-2009 DMDS t&o-ci-spme.SMS Ions: 95.0 Spectrum 1A 60 Ionization Off 1A40:650 BP: 58.9 5.018 min, Scan: 304, 40:650, Ion: 178 us, RIC: 1.072e+6, BC

20% 50

15% 40

30 10% 94.8 59419

20 5% 92.8 72.9 24814 18404

10 0%

5 6 7 8 80 90 100 110 120 130 140 minutes m/z

DMDS-1 DMDS-2 100.00 y = 0.00339x1.08247

1.0E+05 R² = 0.98668

10.00

RF Ion Counts Ion 5.0E+04 1.00

0.10 0.0E+00 10 100 1,000 10,000 0.9 1.1 1.3 1.5 1.7 1.9 Concentration (ng/L) Excitation amplitude (volts)

Figure A.1 Parent ion chromatogram and mass spectrum for 50 µg/L dimethyl disulfide. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples. 67

kCounts 3-5-2009 1-hexanal t&o-ci-spme.SMS Ions: 83.0 Spectrum 1A 40:650 1A BP: 82.8 6.190 min, Scan: 425, 40:650, Ion: 136 us, RIC: 1.336e+6, BC 82.8 538449 100% 500

400 75%

300

50%

200 100.5 149460 25%

100 55.8 64740

0 0%

6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 100 200 300 400 500 600 minutes m/z 5.0E+03 hexanal1 y = 0.00483x0.90150 10.00 4.0E+03 hexanal2

R² = 0.97368

3.0E+03

Ion Counts Ion 1.00

2.0E+03 RF

1.0E+03 0.10 0.0E+00 10 100 1,000 10,000 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Excitation amplitude (volts) Concentration (ng/L)

Figure A.2 Parent ion chromatogram and mass spectrum for 50 µg/L 1-hexanal. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples. 68

kCounts 3-5-2009 cis-4-heptenal t&o-ci-spme.SMS Ions: 95.0 Spectrum 1A 40:650 1A BP: 94.9 14.831 min, Scan: 1377, 40:650, Ion: 203 us, RIC: 295068, BC 94.9 200 141634 100%

150 75%

100 50%

50 25%

54.8 13464

0 0%

12.5 15.0 17.5 20.0 22.5 25.0 27.5 100 200 300 400 500 600 minutes m/z

cis-4-hep1 5.0E+05

cis-4-hep2 1,000.0

4.0E+05 y = 0.00296x1.27746 100.0 R² = 0.99333 3.0E+05

RF 10.0 Ion Counts Ion 2.0E+05 1.0

1.0E+05 0.1 0.0E+00 10 100 1,000 10,000 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Concentration (ng/L) Excitation amplitude (volts) Figure A.3 Parent ion chromatogram and mass spectrum for 50 µg/L cis-4-heptenal. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-5-2009 heptaldehyde t&o-ci-spme.SMS Ions: 97.0 Spectrum 1A 40:650 1A BP: 96.7 14.720 min, Scan: 1360, 40:650, Ion: 141 us, RIC: 623980, BC 500 96.7 192988 100%

400

75%

300

54.8 50% 94646

200

114.5 25% 44690 100 97.8 22979

0 0%

12.5 15.0 17.5 20.0 22.5 25.0 27.5 100 200 300 400 500 600 minutes m/z

6.0E+04 hep1 100.00

5.0E+04 hep2

10.00

4.0E+04

3.0E+04 1.00 RF Ion Counts Ion y = 0.00134x1.10697 2.0E+04 0.10 R² = 0.98628

1.0E+04 0.01 10 100 1,000 10,000 0.0E+00 0 0.5 1 1.5 2 Concentration (ng/L) Excitation amplitude (volts) Figure A.4 Parent ion chromatogram and mass spectrum for 50 µg/L heptaldehyde. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-5-2009 DMDS t&o-ci-spme.SMS Ions: 127.0 Spectrum 1A 40:650 1A BP: 126.6 23.294 min, Scan: 2265, 40:650, Ion: 119 us, RIC: 1.882e+6, BC 126.6 752938 100% 700

600

75% 500

400 50%

300

200 25%

125.3 117994 100 63.8 204.5 45850 48843

0 0%

22.5 25.0 27.5 30.0 100 200 300 400 500 600 minutes m/z

100.00

DMTS1 2.0E+04 10.00

DMTS2

1.00 RF

Ion Counts Ion y = 9.015E-05x1.559E+00 1.0E+04 0.10 R² = 9.856E-01

0.01 10 100 1,000 10,000 0.0E+00 0 0.2 0.4 0.6 0.8 1 Concentration (ng/L) Excitation amplitude (volts)

Figure A.5 Parent ion chromatogram and mass spectrum for 50 µg/L dimethyl trisulfide. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

MCounts 3-5-2009 cis-3-hex-ace t&o-ci-spme.SMS Ions: 83.0 Spectrum 1A 40:650 1A BP: 82.7 29.675 min, Scan: 2938, 40:650, Ion: 79 us, RIC: 3.720e+6, BC 82.7 1.535e+6 1.50 100%

1.25

75%

1.00

0.75 50%

66.9 575161

0.50

25% 54.8 275288 0.25

83.8 97483 0.00 0%

27.5 30.0 32.5 35.0 37.5 100 200 300 400 500 600 minutes m/z 9.0E+04 cis-3-hex1 8.0E+04 100.00 cis-3-hex2 7.0E+04 6.0E+04 10.00 5.0E+04

Ion Counts Ion 4.0E+04 1.00 RF 3.0E+04 y = 0.00228x1.06754 2.0E+04 0.10 R² = 0.98812 1.0E+04 0.0E+00 0.01 0 0.2 0.4 0.6 0.8 1 1.2 1.4 10 100 1,000 10,000 Excitation amplitude (volts) Concentration (ng/L) Figure A.6 Parent ion chromatogram and mass spectrum for 50 µg/L cis-3-hexenyl acetate. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-5-2009 t,t-2,4,hep t&o-ci-spme.SMS Ions: 111.0 Spectrum 1A 40:650 1A BP: 110.8 34.960 min, Scan: 3518, 40:650, Ion: 288 us, RIC: 166691, BC 150 110.8 58412 100%

125

75%

100

152.9 31217 75 50%

25%

66.9 4740 154.0 3784

0 0%

10 20 30 40 50 100 200 300 400 500 600 minutes m/z 6.0E+05 t,t,2,4hep1 1,000.00 5.0E+05 t,t,2,4hep2 y = 0.000522x1.431107 100.00 4.0E+05 R² = 0.994080 10.00

3.0E+05

RF Ion Counts Ion 1.00 2.0E+05 0.10 1.0E+05 0.01 0.0E+00 10 100 1,000 10,000 0 0.2 0.4 0.6 0.8 1 1.2 Excitation amplitude (volts) Concentration (ng/L)

Figure A.7 Parent ion chromatogram and mass spectrum for 50 µg/L trans,trans-2,4-heptadienal. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-9-2009 2-IP-3-MP t&o-ci-spme.SMS Ions: 153.0 Spectrum 1A 40:650 1A BP: 152.8 34.789 min, Scan: 3624, 40:650, Ion: 288 us, RIC: 186313, BC 152.8 97638 100%

100

75%

50% 50

25% 25

154.0 13215

0 0%

32.5 35.0 37.5 40.0 42.5 45.0 100 200 300 400 500 600 minutes m/z

7.E+04 100.00 IPMP1 1.0878 6.E+04 y = 1.139x IPMP2 10.00 R² = 0.9967 5.E+04

4.E+04 1.00 RF Ion Counts Ion 3.E+04 2.E+04 0.10 1.E+04 0.01 0.E+00 0.1 1.0 10.0 100.0 0.2 0.7 1.2 Excitation amplitude (volts) Concentration (ng/L) Figure A.8 Parent ion chromatogram and mass spectrum for 1 µg/L isopropyl methoxypyrazine. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-5-2009 nonadienal t&o-ci-spme.SMS Ions: 121.0 Spectrum 1A 40:650 1A BP: 167.0 38.687 min, Scan: 3902, 40:650, Ion: 254 us, RIC: 375622, BC 167.0 105465 100%

100

75%

50%

120.8 36549

25% 25 92.8 17891 168.0 14280 96.8 8310

0 0%

37.5 40.0 42.5 45.0 47.5 100 200 300 400 500 600 minutes m/z

2.8E+05 nona1 100.00 2.4E+05 nona2 y = 3.304E-04x1.289E+00 R² = 9.974E-01 2.0E+05 10.00 1.6E+05

1.00 Counts Ion

1.2E+05 RF

8.0E+04 0.10

4.0E+04 0.01 0.0E+00 10 100 1,000 10,000 0.1 0.3 0.5 0.7 0.9 1.1 1.3 Concentration (ng/L) Excitation amplitude (volts)

Figure A.9 Parent ion chromatogram and mass spectrum for 50 µg/L trans-2,cis-6-nonadienal. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-9-2009 2-IB-3-MP t&o-ci-spme.SMS Ions: 167.0 Spectrum 1A 40:650 1A BP: 166.9 39.496 min, Scan: 4180, 40:650, Ion: 359 us, RIC: 82212, BC 70 166.9 12764 100%

50 75%

50%

25% 168.0 2297 63.7 1821 10 191.8 532.8 1321 412.5 1375 105.8 322.8 1093 825 832

0 0%

37.5 40.0 42.5 45.0 47.5 100 200 300 400 500 600 minutes m/z

IBMP1 6.0E+05 1,000.00 IBMP2

5.0E+05 100.00 y = 3.0426x0.854 4.0E+05 R² = 0.9728 10.00

3.0E+05 Ion Counts Ion

RF 1.00 2.0E+05 0.10 1.0E+05

0.0E+00 0.01 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0.1 1.0 10.0 100.0

Excitation amplitude (volts) Concentration (ng/L)

Figure A.10 Parent ion chromatogram and mass spectrum for 1 µg/L isobutyl methoxypyrazine. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-10-2009 2-MIB t&o-ci-spme.SMS Ions: 151.0 Spectrum 1A 40:650 1A BP: 150.8 39.060 min, Scan: 4159, 40:650, Ion: 294 us, RIC: 375817, BC 150.8 132533 100% 125

100 75%

94.9 50% 65425

25%

151.7 54.8 13148 10818

0 0%

39.0 39.5 40.0 40.5 100 200 300 400 500 600 minutes m/z 100.00 MIB1 MIB2 1.0805 10.00 y = 0.1868x

2.0E+04 R² = 0.9989

1.00

RF Ion Counts Ion

1.0E+04 0.10

0.01 0.0E+00 0.1 1.0 10.0 100.0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Concentration (ng/L) Excitation amplitude (volts)

Figure A.11 Parent ion chromatogram and mass spectrum for 1 µg/L 2-methylisoborneol. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples. 77

kCounts 3-10-2009 b-cyclocitral t&o-ci-spme.SMS Ions: 153.0 Spectrum 1A 40:650 1A BP: 152.8 41.092 min, Scan: 4394, 40:650, Ion: 445 us, RIC: 84470, BC 152.8 33979 70 100%

75% 50

50%

20 25%

10 153.8 3847 64.2 2056

0 0%

40.0 42.5 45.0 47.5 100 200 300 400 500 600 minutes m/z

4.0E+05 cyclo1 100.00 cyclo2 0.8487 3.0E+05 y = 2.5148x

10.00 R² = 0.988

RF Counts Ion 2.0E+05 1.00 1.0E+05

0.10 0.0E+00 0.1 1.0 10.0 100.0 0.35 0.45 0.55 0.65 0.75 0.85 0.95 1.05 Concentration (ng/L) Excitation amplitude (volts)

Figure A.12 Parent ion chromatogram and mass spectrum for 1 µg/L β-cyclocitral. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-6-2009 indole t&o-ci-spme.SMS Ions: 118.0 Spectrum 1A 40:650 1A BP: 117.9 44.200 min, Scan: 4493, 40:650, Ion: 345 us, RIC: 159447, BC 117.9 72947 100% 70

75%

50%

20 25%

10 119.0 7599 58.8 4216

0 0%

43 44 45 46 47 48 100 200 300 400 500 600 minutes m/z 4.0E+05 indole 100.00

3.0E+05 indole 10.00

2.0E+05

RF 1.00 Ion Counts Ion y = 1.792E-04x1.422E+00 1.0E+05 0.10 R² = 9.694E-01

0.0E+00 0.01 0 0.2 0.4 0.6 0.8 1 1.2 1.4 10 100 1,000 10,000 Excitation amplitude (volts) Concentration (ng/L)

Figure A.13 Parent ion chromatogram and mass spectrum for 50 µg/L indole. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples. 79

MCounts 4-3-2009 t,t-2,4,deca-50 t&o-ci-spme.SMS Ions: 153.0 Spectrum 1A 40:650 1A BP: 152.7 44.084 min, Scan: 4638, 40:650, Ion: 35 us, RIC: 4.331e+7, BC 152.7 9.033e+6 100%

7.5

75%

5.0

50%

2.5 151.8 25% 2.101e+6 52.9 1.741e+6

153.8 53.8 1.088e+6 857090

0.0 0%

43 44 45 46 47 100 200 300 400 500 600 minutes m/z

t,t deca1 7.0E+05 100.00 t,t, deca2

6.0E+05 10.00 5.0E+05

4.0E+05

1.00

RF Ion Counts Ion 3.0E+05 1.3345115 2.0E+05 0.10 y = 0.0007485x R² = 0.9962344 1.0E+05 0.01 0.0E+00 10 100 1,000 10,000 0 0.2 0.4 0.6 0.8 1 Concentration (ng/L) Excitation amplitude (volts)

Figure A.14 Parent ion chromatogram and mass spectrum for 50 µg/L trans,trans-2,4-decadienal. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples. 80

kCounts 3-10-2009 2,4,6 TCA t&o-ci-spme.SMS Ions: 211.0 Spectrum 1A 40:650 1A BP: 210.8 44.684 min, Scan: 4799, 40:650, Ion: 551 us, RIC: 128280, BC 210.8 30 30319 100%

212.8 27068

75%

15 50%

10 214.7 8564

25%

5 211.8 64.5 3633 2776 354.9 2048

0 0%

43 44 45 46 47 48 100 200 300 400 500 600 minutes m/z 1.2E+05 246TCA1 100.00 0.8429 1.0E+05 246TCA2 y = 0.242x 10.00 R² = 0.9569 8.0E+04

6.0E+04 1.00 RF Ion Counts Ion 4.0E+04 0.10 2.0E+04 0.01 0.0E+00 0.1 1.0 10.0 100.0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Concentration (ng/L) Excitation amplitude (volts)

Figure A.15 Parent ion chromatogram and mass spectrum for 1 µg/L 2,4,6-trichloroanisole. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-10-2009 2,3,6 TCA t&o-ci-spme.SMS Ions: 211.0 Spectrum 1A 40:650 1A BP: 210.8 46.102 min, Scan: 4915, 40:650, Ion: 349 us, RIC: 254834, BC 210.8 72911 70 100%

212.8 63645 60

75% 50

50%

20 214.8 18394 25%

10 211.8 8674

0 0%

45 46 47 48 49 50 100 200 300 400 500 600 minutes m/z 2.0E+05 100.00 236TCA1 0.8039 236TCA2 y = 0.781x 10.00 R² = 0.9617

1.0E+05 1.00

Ion Counts Ion RF

0.10

0.0E+00 0.01 0 0.2 0.4 0.6 0.8 1 1.2 0.1 1.0 10.0 100.0 Excitation amplitude (volts) Concentration (ng/L) Figure A.16 Parent ion chromatogram and mass spectrum for 1 µg/L 2,3,6-trichloroanisole. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-10-2009 geosmin t&o-ci-spme.SMS Ions: 165.0 Spectrum 1A 40:650 1A BP: 164.8 47.490 min, Scan: 5037, 40:650, Ion: 186 us, RIC: 1.083e+6, BC 164.8 287911 100%

250

75% 200

150

50% 108.9 124960

100 54.9 90003

25% 94.9 53432 50 165.8 40594 66.9 25871

0 0%

46 47 48 49 50 100 200 300 400 500 600 minutes m/z 100.00 1.5E+04 geosmin1 y = 0.2736x1.0318 geosmin2 10.00 R² = 0.998 1.0E+04 1.00

RF Ion Counts Ion

5.0E+03 0.10

0.01 0.0E+00 0.1 1.0 10.0 100.0 0 0.5 1 1.5 2 Concentration (ng/L) Excitation amplitude (volts) Figure A.17 Parent ion chromatogram and mass spectrum for 1 µg/L geosmin. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples.

kCounts 3-10-2009 b-ionone t&o-ci-spme.SMS Ions: 193.0 Spectrum 1A 40:650 1A BP: 192.9 50.269 min, Scan: 5251, 40:650, Ion: 321 us, RIC: 183825, BC 192.9 64437 100% 80

75% 60

40 50%

30 175.0 18680

20 25%

193.9 10 7239

0 0%

49 50 51 52 53 100 200 300 400 500 600 minutes m/z

4.0E+04 ionone1 100.00 ionone2 3.0E+04 y = 1.4459x0.7789 10.00 R² = 0.9936

2.0E+04

Ion Counts Ion 1.00 RF

1.0E+04 0.10

0.01 0.0E+00 0.1 1.0 10.0 100.0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Excitation amplitude (volts) Concentration (ng/L)

Figure A.18 Parent ion chromatogram and mass spectrum for 1 µg/L -ionone. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples. 84

kCounts 3-10-2009 2,4,6 TBA t&o-ci-spme.SMS Ions: 345.0 Spectrum 1A 40:650 1A BP: 344.8 51.306 min, Scan: 5352, 40:650, Ion: 111 us, RIC: 3.491e+6, BC 344.8 622599 600 100%

500

75%

400

300 50% 78.8 282472 342.8 248973

200

25% 52.8 100 101925 345.8 82589

266.0 37229 0 0%

50 51 52 53 54 55 100 200 300 400 500 600 minutes m/z

7.0E+05 246 TBA1

6.0E+05 246 TBA2 10.00

5.0E+05 y = 0.00808x1.57169 1.00

4.0E+05 R² = 0.96102

Counts Ion

3.0E+05 0.10 RF

2.0E+05 0.01 1.0E+05

0.0E+00 0.00 0.5 0.7 0.9 1.1 1.3 1.5 0.1 1.0 10.0 100.0 Excitation amplitude (volts) Concentration (ng/L)

Figure A.19 Parent ion chromatogram and mass spectrum for 1 µg/L 2,4,6-tribromoanisole. Excitation amplitude scan for determining the optimal voltage to fragment the parent ion mass. Calibration curve obtained with SPME fiber used to analyze Twin Lakes samples. 85

MCounts Hexanal-d12 6-15-2009 6-03-25 PM T&O-CI-SPME.SMS Ions: 93.0 40:650 1A

2.5

2.0

1.5

1.0

0.5

0.0

6.00 6.25 6.50 6.75 7.00 7.25 7.50 minutes Spectrum 1A BP: 92.8 6.301 min, Scan: 441, 40:650, Ion: 61 us, RIC: 8.547e+6, BC 92.8 2.714e+6 100%

75%

50%

93.9 1.108e+6

112.8 716159 25%

62.0 428838

100 200 300 400 500 600 m/z

Figure A.20 Parent ion chromatogram and mass spectrum for 50 µg/L hexanal-d12 internal standard.

MCounts MIB-d3 6-15-2009 7-21-17 PM T&O-CI-SPME.SMS Ions: 154.0 40:650 1A

1.50

1.25

1.00

0.75

0.50

0.25

0.00

39.3 39.4 39.5 39.6 minutes Spectrum 1A BP: 153.8 39.417 min, Scan: 4056, 40:650, Ion: 81 us, RIC: 5.307e+6, BC 153.8 1.581e+6 100%

75%

50%

98.0 483391

25% 54.9 321263 154.8 224271 52.8 150291

0% 100 200 300 400 500 600 m/z

Figure A.21 Parent ion chromatogram and mass spectrum for 1 µg/L MIB-d3 internal standard.