Environmental Assessment and Risk Analysis Element

Research Project Summary

March, 2003

Abstract

This is a summary report of a detailed investigation in which the analysis of synthetic organic chemicals by Gas Chromatogra-phy (GC) and Liquid Chromatography (LC) was conducted on raw and finished water samples collected from public watersupplies using ground water as a source of drinking water. All water systems sampled are known to be contaminated byvolatile organic chemicals except for one (the “control” system). This work investigated the potential presence of non-volatileand semi-volatile organic chemicals in those water supplies. Five bottled waters were also sampled. Several generalizationscan be made: 1) water serving systems impacted by identified hazardous waste sites have distinct and sometimes uniqueTICs associated with them; 2) TICs are generally low in concentration, most being estimated at a concentration below a partper billion (microgram per liter, mg/L); and 3) many organic chemicals reported as TICs were not actually in the watersampled but were found in the analysis due to sampling and/or laboratory contamination.

The Characterization of Tentatively Identified Compounds (TICs) in WaterSamples Collected from Public Water Systems in New Jersey

Eileen Murphy1 , Brian Buckley2 , Lee Lippincott1, Ill Yang2, Bob Rosen3

Introduction

Presently, certain conventional analytical methods foranalyzing drinking water samples from public water suppliesfor specific, or targeted, organic chemical contamination arerequired by the NJ Safe Drinking Water Act. For the mostpart, this routine testing is adequate for the determination ofcommonly occurring volatile organic chemicals (VOCs). Itwas always known that VOCs, which are the currentregulatory focus of analysis for organics in drinking water,may serve as markers for the presence of mostly unregu-lated non- and semi-volatile contaminants in addition tobeing significant in their own right. In situations whereimpacted water is being used as a potable source, this issueis very important. In the past, reliable analytical methodswere not available to determine the presence or the nature ofmany non-volatile (e.g., some pharmaceuticals, dyes, inks)and semi-volatile (e.g., plasticizers, fragrances, somecomponents of fuel oils) contaminants, with the exception ofcertain types of semi-volatiles (i.e., some pesticides andplasticizers).

A volatile compound is defined chemically as one with arelatively low boiling point. That is, a volatile compound“evaporates” readily into the air. Whereas, a non-volatilecompound evaporates much more slowly or not at all. Asemi-volatile compound falls in between. Thus, due to thehistorical focus on VOCs, the full picture of exposure andhealth risk from contaminated drinking water may not havebeen adequately determined. With the emergence of moresensitive analytical capabilities for non- and semi-volatileorganic contaminants, a more complete assessment of thisadditional contamination, if and where it exists, can be made,and appropriate steps can be taken to protect public health.

This study was able to address the potential detection ofhundreds of chemicals because the instrumentation was setup to screen for tentatively identified compounds (TICs). ATIC is a compound that can be seen by the analytical testingmethod, but its identity and concentration cannot beconfirmed without further analytical investigation. An analogyis when a photograph is taken of a subject. The picture alsocaptures the information in the background, and often thisinformation is fuzzy, but the focus of the picture is thesubject. The subject (i.e., target item) is clear, but thebackground components (i.e., the tentatively identifieditems), while captured in the picture, are fuzzy.

Objectives

There were three related objectives to this multi-year project.

1. Tentatively identify and possibly quantify chemicalspresent in raw and treated water samples collected fromwater supply systems impacted by hazardous wastesites.

2. In instances where chemicals are present in the rawwater, determine if existing water treatment is effective atremoving them.

3. Characterize the types of unregulated compoundspresent in water samples due to sampling and labora-tory contamination.

Methods

Data on organic analyses from public community watersystems that use ground water as their water source wasgenerated and delivered to the project investigators by theNJ Department of Environmental Protection’s (NJDEP)

Bureau of Safe Drinking Water (BSDW). Review of this datashowed which systems had historical organic contaminationabove appropriate maximum contaminant levels (MCLs) andwhich systems had water treatment technologies in place toremove the contamination. This became the candidate listfrom which systems were selected for participation in thestudy. There were 96 individual facilities (points-of-entry tothe water distribution system), serving approximately 54community water systems, identified in 1997 where volatilecontamination above MCLs occurred in the untreated sourcewater and where some type of water treatment was in placeto remove the contamination before the water was distrib-uted to customers.

Using the candidate list, the investigators selected appropri-ate water systems to sample as part of this study. Therewere several exceptions. Two of the water systems werevery small (one is a church and the other is a school) with nowater treatment. They were included as part of the studybecause historical results showed the presence of unusualorganic chemical contamination, according to BSDWrecords. One surface water system was selected forcomparison purposes, and one system used both surface aswell as ground water. One ground water system wasselected as a control system (i.e., no known impacts and notreatment to remove organic contamination). Ultimately, 21water systems were sampled during this study. The samplebottles for one system (NJ American – Atlantic City) werebroken in the laboratory. No resample was collected.Therefore, results for 20 systems are reported here.

Conventional Analytical MethodsAll water samples were sent to the NJ Department of Healthand Senior Services (NJDHSS) laboratory for analysis byconventional USEPA Methods 524.2 (84 target volatilechemical analytes) and 525.2 (42 target semi-volatilechemical analytes) and for arsenic and mercury.

General Nonconventional MethodsNonconventional analytical methods were developed at theNJ Environmental and Occupational Health SciencesInstitute (EOHSI) and the NJ Center for Advanced FoodTechnology (CAFT) of Rutgers University. The EOHSImethod utilized gas chromatography to analyze for volatileand semi-volatile compounds. The CAFT method utilizedhigh pressure liquid chromatography to analyze for non-volatile compounds. All water samples were run throughthese methods at least once to screen for classes ofchemical compounds present.

ResultsCONVENTIONAL ANALYSIS:The volatile organic compounds most frequently detectedabove maximum contaminant levels in the raw waters weretrichloroethylene, tetrachloroethylene, and 1,1,1-trichloroethane. Water samples collected after the airtreatment systems indicated that these compounds wereremoved to levels below the MCLs, mostly below methoddetection levels. However, several instances, levels oftrichloroethylene and tetrachloroethylene were detected atlevels close to their MCLs of 1.0 µg/L. The results of theconventional analysis validated the historical data collected atthe water systems, indicating that raw water is contaminated

with volatile organic chemicals and that air treatment installedto remove these contaminants is effective at removing them.Also, trihalomethanes and other types of disinfection by-products were detected in chlorinated water samples atlevels greatly below appropriate existing MCLs.

NONCONVENTIONAL ANALYSIS:High Pressure Liquid Chromatography (HPLC)During the first two years of the study, no non-volatilecompounds were detected using HPLC. During the thirdyear of the study, larger sample sizes were analyzed. Twoliters of sample were used instead of one to improve theoverall method detection limit. Even with the larger samplesize, almost all of the samples reported a non-detect result.

Gas Chromatography Over the course of the 4 year study,approximately 600 TICs were detected in 199 water samplescollected: 108 raw water samples (3 raw surface water and105 raw ground water), 51 finished water samples, 35 blanksamples, and 5 bottled water samples. Of the 600 TICs, 112were detected frequently among the types of samplescollected and among the systems sampled. For instance,butylated hydroxytoluene was detected in raw, finished,bottled and blank water samples, so it is included under“more than one category”. The remaining 488 TICs weredetected in the distribution shown in Figure 1, whichdelineates TIC distribution in raw, finished, raw & finished,bottled and blank samples. The distribution of TICs by watersupply system and type of sample is shown in Figure 2.TICs unique to water supply systems are shown in Figure 3.

43%

15%

9%

1%

13%

19% Raw Water (266)

Finished Water(88)Raw & Finished(51)Bottled Water (6)

Blanks (77)

More Than OneCategory (112)

Figure 1. Distribution of TICs (unique to the group) by type of water sample.

0

50

100

150

# to

tal T

ICs

in w

ater

sa

mpl

es

Fairl

awn

Mer

chan

tvill

e-El

izab

etht

own

Bottl

edU

WTR

-Dov

erR

ocka

way

Rid

gew

ood

Roc

kaw

ayR

ahw

ayG

arfie

ldPa

rk R

idge

Perth

Am

boy

Flem

ingt

nW

aldw

ick

Sea

Girt

New

ton

Sale

mR

ocky

Hill

Ros

emon

tE.

Am

wel

lC

hris

t Chu

rch

R&F pairs

Finished

Raw

Figure 2. TICs in Water Samples (not including TICs that were detected in blanks)

Flem

ingt

on

2

Raw & Finished Sample PairsIn general, when TICs were detected in both raw andfinished water from the same system, it was in systemswhere carbon treatment was not present. This is notsurprising because air-based treatment technologies aredesigned specifically to remove volatile organic chemicals(and many are capable of removing some semi-volatilecompounds). Carbon treatment technologies can removemany types of volatile, semi-volatile and non-volatile chemi-cals.

Finished Drinking Water SamplesFinished drinking water contained some TICs that werenever detected in raw water samples, indicating that theymay enter the distribution system through the treatmentprocess, chemical transformation of other compoundsduring treatment, or addition of disinfection reagents.

While 88 unique TICs were detected in finished watersamples, only 8 of these appeared in more than one finishedwater sample. The appearance of compounds in finishedwater is not unusual in and of itself – the conventionalanalyses showed disinfection by-products appearing infinished water samples and not in raw water samples.Similarly, the treatment of water by air, carbon, or theaddition of disinfectants may introduce compounds thatwould not necessarily be present in raw water.

Raw water samplesOf the 600 TICs detected in this study, 338 were detected inraw water samples (and not in blanks). Of these 338, asubset of 266 TICs were detected in raw water samples only,and not in finished water samples or any other category. Thewells sampled as part of this study were selected becausethey had historical volatile chemical contamination. Anothercriteria for selection was proximity of the wells to knowncontaminated sites. In several instances, the contaminatedsite influencing the wells had been identified and, in fact, theresponsible party paid for installation and maintenance of thetreatment technology at the water system. It was notsurprising, therefore, to see that semi-volatile compoundswere present in the raw water samples, as these samplesalso contained the highest numbers of and highest concen-trations of volatile organic chemicals of the groups.

Figure 3. Number of TICs Found by System

0

20

40

60

80

100

120

140

160

180

Fairl

awn

Blan

ks

Tom

s R

iver

M

erch

antv

ille-P

enns

auke

n G

arfie

ld

Rah

way

Pa

rk R

idge

El

izab

etht

own

Rid

gew

ood

Roc

kaw

ay T

owns

hip

Sea

Girt

Pe

rth A

mbo

yBo

ttled

wat

ers

Roc

kaw

ay B

oro

Flem

ingt

onN

ewto

n W

aldw

ick

E. A

mw

ell s

choo

lC

hris

t Chu

rch

Sale

m

Roc

ky H

illR

osem

ont-

cont

rol

All TICs in the system

TICs exclusive to thesystem

Bottled Water SamplesThirty-two TICs were found in the five bottled water samples.Twenty-four of these were not detected in the correspondingblank water samples. There were six (6) TICs unique tobottled water. Only one of the six TICS unique to bottledwater was detected in more than one bottled water and notdetected in the blanks. It was 3,5-di-tert-butyl-4-hydroxybenzyl alcohol.

Blank SamplesBecause of the prevalence of TICs in blank water, it is difficultto interpret their presence in environmental water samples.However, several patterns emerge when investigating thedata of TICs in blank water samples: there is a population ofTICs that occur only in blanks; there are TICs that occurfrequently in blanks and environmental samples; and thereare TICs that sometimes appear in a blank and sometimesappear in an unrelated environmental sample. What itimplies is that when regulators look at TIC information fromenvironmental samples, it is vital that they also look at thecorresponding blank sample information. The detection of aTIC in a water supply sample does not directly imply thatthere is an environmental contamination problem if the TICalso appears in the field or trip blank. Similarly, the fact that acompound appears in a blank does not preclude its pres-ence in an environmental sample. The data need to beevaluated side-by-side in order for an assessment to bemade about the actual occurrence of a contaminant in asample.

Discussion

Source Water Assessment Areas (SWAs) have beendelineated and are available for several of the water systemssampled as part of this study. One is shown in Figure 4.When looking at the map, it is clear that there are potentialsources of contamination near some of the communitysupply wells sampled. The NJDEP should continue its workon assessing the potential impacts from hazardous wastesites to drinking water sources in the state. As a result of thisstudy, the NJDEP may want to consider more intensivescrutiny of the inventory of chemicals reported by hazardouswaste site operators. Currently, the site inventories are verybroad. It may be useful to have site operators generatemore specific types of waste lists in order for NJDEP staff todetermine if there is the potential for contaminants to reachdrinking water wells. This study shows that contamination byhazardous waste sites may not be limited to volatile organicchemicals and that treatment to remove volatile chemicalsmay not be sufficient to remove semi- and non-volatilechemicals. Evidence from the control system shows thatwater systems not impacted by sources of contamination donot appear to contain unique TICs.

3

Next Steps

Further work is underway to definitively identify and quantifysome of the TICs seen in this study. While it is impossible topursue positive identifications for all 600 TICs reported, it ispossible to cull the list and focus on a more manageablenumber of TICs. The criteria for selection of which TICs topursue include: availability of an analytical standard, fre-quency of occurrence in water samples not likely to bepresent due to sampling or laboratory contamination, andpotential human toxicity. A report is expected on this work in2003.

The study described herein focused on the occurrence ofTICs in water samples from water supply systems usingground water as their water sources. Presently, watersamples collected from surface water systems are beingcollected and will be analyzed using the same GC-MSscreening methods described here. A report on this work isexpected by the spring of 2004.

A report on the preliminary assessment of health informationon the TICs found in this study is expected in 2003. Re-searchers from the University of Medicine and Dentistry, NJhave reviewed the available literature to find chemical,industrial and health information on as many of the TICs aspossible. Further, a preliminary assessment of potentialhuman risk based on toxicity data (when available) will beincluded.

1 NJDEP, Division of Science, Research & Technology,Trenton, New Jersey2 Environmental and Occupational Health Sciences Instituteof NJ, Piscataway, New Jersey3 Center for Advanced Food Technology at Rutgers Univer-sity, New Brunswick, New Jersey

Funding

Funding for this multi-year project was from the NJ A-280Safe Drinking Water Research Fund, administered by theDivision of Science, Research & Technology for the Bureauof Safe Drinking Water. Dr. Judy Louis and Gail Carter, bothof DSRT, provided the map of the Source Water Assessmentin Figure 4. Thanks also to the myriad reviewers whosupplied important comments on this report to make itreadable to both a technical and lay audience.

Prepared By

4

$

$

$

$

$ $

$

$

$

$

$

$$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

##

#

#

#

#

#

#

#

#

#

##

#

#

#

N

0.4 0 0.4 0.8 Miles

Figure 6 - Source Water Assessment Areas for Fairlawn Water Department

J.Louis, DSRT, Sept. 2002

N

Source Water Assessment Areas2 yr. time of travel5 yr. time of travel12 yr. time of travel

Classification Exception Areas$ Known Contaminated Sites# CWS Wells

N

4

STATE OF NEW JERSEYJames E. McGreevey, Governor

Department of Environmental ProtectionBradley M. Campbell, Commissioner

Division of Science, Research & TechnologyMartin Rosen, Director

Environmental Assessment & Risk Analysis ElementDr. Eileen Murphy, Assistant Director

Please send comments or requests to:Division of Science, Research and TechnologyP.O.Box 409, Trenton, NJ 08625Phone: 609 984-6070Visit the DSRT web site @ www.state.nj.us/dep/dsr

RESEARCH PROJECT SUMMARY