RECONSIDERING LEAD CORROSION IN DRINKING WATER:

PRODUCT TESTING, DIRECT CHLORAMINE

ATTACK AND GALVANIC CORROSION

Abhijeet Dudi

Thesis submitted to the Faculty of the

Virginia Polytechnic Institute and State University

In partial fulfillment of the requirements for the degree of

Master of Science

in

Environmental Engineering

Marc Edwards, Chair

John Little

Peter Vikesland

July 22, 2004

Blacksburg, Virginia

Keywords: ANSI/NSF 61 Section 8, Chloramines, lead and copper rule, galvanic

Copyright 2004, Abhijeet Dudi, Marc Edwards

Reconsidering Lead Corrosion In Drinking Water:

Product Testing, Direct Chloramine

Attack and Galvanic Corrosion

Abhijeet Dudi

Abstract

The ban on lead plumbing materials in the Safe Drinking Water Act (1986) and the EPA

Lead and Copper Rule (1991) have successfully reduced lead contamination of potable

water supplies. The success of these regulations gave rise to a belief that serious lead

contamination was an important past problem that had been solved, and that additional

fundamental research was therefore unnecessary. This work carefully re-examined the

lead contamination issue from the perspective of 1) new regulations causing a shift from

chlorine to chloramine disinfectant, 2) assumptions guiding sampling strategies, 3)

existing performance standards for brass, and 4) galvanically driven corrosion of lead

bearing plumbing materals. The results were instrumental in uncovering and

understanding a serious problem with lead contamination in Washington, D.C.

A critical reading of the literature indicates that chloramines can accelerate corrosion of

lead bearing materials and increase lead contamination of water. When a new sampling

protocol was conceived and used in Washington homes to assess the nature of the

problem, hazardous levels of lead were found to be present in some drinking water

samples. Contrary to the conventional wisdom, lead was not always highest in first draw

samples, but often increased with flushing. This has several important implications for

monitoring and public health. For instance, well-intentioned public education materials

were causing consumers to drink water containing very high levels of lead in some

circumstances. Laboratory and field-testing proved that chloramines were causing

serious lead corrosion problems. That testing also discovered that, unbeknownst to

scientists and utilities, free chlorine itself can act as a corrosion inhibitor, reducing lead

solubility and contamination of water. The net result is that changing disinfectant from

free chlorine to chloramine can sometimes trigger serious lead contamination of water.

iii

While the worst problems with lead in Washington, D.C. came from the lead services,

significant levels of lead were occasionally sampled from homes with solders or brass as

the lead source. This prompted re-evaluation of the ANSI/NSF 61, Section 8 standard,

which is relied on to protect public health from in-line brass plumbing devices that might

leach excessive lead to potable water. In-depth study of the standard revealed serious

flaws arising from use of a phosphate buffer in the test waters and a failure to control

carbonate dissolution from the atmosphere. Due to these deficiencies, small devices

made of pure lead could actually pass the performance test. The public therefore has no

assurance that devices passing NSF Section 8 testing are safe and reforms to the standard

are obviously needed.

Other problems arise from connecting copper pipe to lead bearing plumbing in practice.

The copper is cathodic and dramatically accelerates corrosion of the lead anode via a

galvanic current. Corrosion and hydrolysis of released Pb+2 can lower pH near the surface

of the lead and increase its solubility. A similar galvanic effect can arise from cupric ions

present in the water via deposition corrosion mechanism. In cases where part of a lead

service line is replaced by copper pipe, the galvanic corrosion effect can create a serious

long-term problem with lead contamination. Such partial lead service line replacements

are occurring in many US cities and the practice should be stopped.

Lead contamination of potable water is not only a problem of the past but also of the

present. While additional research is necessary before regulators, utilities and

homeowners can anticipate and mitigate such problems with confidence, this work

provides sound fundamental basis for future progress.

iv

Acknowledgements

I would like to thank my advisor Dr. Marc Edwards, for his patience, professional and

personal advice that guided me through this project. I also appreciate the assistance of

Jody Smiley, Betty Wingate, Sherry Burke and Julie Petruska in experimental and

analytical procedures.

I also thank Michael Shock for sharing his expertise knowledge on lead corrosion.

Finally, I would especially like to thank all the members of my research group for

support, help and encouragement provided throughout this research.

v

Author’s Preface

All three chapters are separate manuscripts formatted to the specifications of Virginia

Tech’s journal article formatting. All three chapters have been submitted to Journal

American Water Works Association (JAWWA) for publication. Chapter 1 focuses on

relative effects of chloramines and various other oxidants on lead leaching from brass

devices and lead pipes. This chapter is published in the October, 2004 JAWWA.

Chapter 2 critically evaluates and examines the practical rigor of the NSF 61 section 8

testing procedure that is relied on to protect public health. The results highlight the need

for an improved standard to certify lead bearing brass devices. Chapter 3 describes the

galvanic corrosion of lead and brass materials connected to copper. The adverse effects

of such connections in the context of lead leaching are verified.

vi

Table of Contents Abstract ........................................................................................................................... ii Acknowledgement ............................................................................................................ iv Author’s Preface ................................................................................................................. v Table of Contents............................................................................................................... vi List of Figures .................................................................................................................. viii CHAPTER 1: ROLE OF CHLORINE AND CHLORAMINE IN CORROSION OF

LEAD BEARING PLUMBING MATERIALS ............................................... 1

Abstract ............................................................................................................. 1

Keywords .......................................................................................................... 1

Introduction....................................................................................................... 1

Chloramines. ................................................................................... 3

Nitrate/Ammonia. ........................................................................... 4

Materials and Methods...................................................................................... 6

Results and Discussions.................................................................................... 9

Field sampling................................................................................. 9

Lead solubility in the presence of no oxidant, free chlorine and

chloramines ................................................................................... 12

Lead leaching from pure lead pipes .............................................. 15

Lead leaching from brass devices ................................................. 16

Summary and Conclusions ............................................................................. 19

Acknowledgements......................................................................................... 21

References....................................................................................................... 21

CHAPTER 2: LEAD LEACHING FROM IN-LINE BRASS DEVICES: A CRITICAL

EVALUATION OF THE EXISTING STANDARD ..................................... 37

Abstract ........................................................................................................... 37

Introduction..................................................................................................... 37

Materials and Methods.................................................................................... 41

Results and Discussions.................................................................................. 43

Consideration of Test Water Chemistry. ...................................... 43

Effect of Phosphate on Lead Leaching Propensity of the NSF

pH 5 Water. ................................................................................... 45

vii

Effect of Aeration on Lead Leaching Propensity of the NSF pH

10 Water........................................................................................ 46

Considering the Overall Rigor of the Test. ................................... 47

Retrospective Evaluation and Recommendations......................... 49

Conclusions..................................................................................................... 52

Acknowledgements......................................................................................... 53

References....................................................................................................... 53

CHAPTER 3: GALVANIC CORROSION OF LEAD BEARING PLUMBING

DEVICES........................................................................................................ 69

Abstract ........................................................................................................... 69

Keywords ........................................................................................................ 69

Introduction..................................................................................................... 69

Materials and Methods.................................................................................... 74

Results and Discussion ................................................................................... 76

Effect of Chloride to Sulfate Ratio ............................................... 76

Effect of Galvanic Current on Lead Leaching.............................. 77

pH changes in the pipe rig ............................................................ 78

Effect of pH................................................................................... 79

Effect of gap between metals........................................................ 79

Passive Anodic Protection ............................................................ 80

Deposition corrosion..................................................................... 81

Effect of oxidants.......................................................................... 82

Galvanic relationship between current and lead leached.............. 82

Other effects.................................................................................. 83

Conclusion ...................................................................................................... 84

Acknowledgements......................................................................................... 85

References....................................................................................................... 85

viii

List of Figures:

Figure 1-1: DC WASA 90%’ile lead in LCR Sampling events. .....................................26

Figure 1-2: Comparison of impacts of lead leaching from brass in Potomac water

with free and chloramine. Representative data are from Davis et al.,

1994...............................................................................................................27

Figure 1-3: Drinking water collected at a Washington DC sampling site after a one

hour stagnation time and 1 minute flush time . Error bars during and

after switch to free chlorine are 90% confidence intervals on triplicate

sampling events during the same day. ..........................................................28

Figure 1-4: (a) Lead profile from a tap as a function of flushing time (> 10 hour

stagnation). (b) Lead levels from a tap after the indicated stagnation time

and a fixed 1 minute flush.............................................................................29

Figure 1-5: At pH 5.5 solutions with 48,000 ppb Pb+2 are clear since most of the

lead is soluble (far left). At pH 8.5 a white solid precipitates (middle

left). When chlorine is present at pH 5.5 or pH 8.5, a highly insoluble

red brown solid precipitates (middle right). The red-brown solids are

hard to distinguish from “red water” samples caused by corroding iron

(far right). The filter paper attached to each container illustrates the

color of the captured particles.......................................................................30

Figure 1-6: Lead solubility in three different beaker experiments. Without any

phosphate (a), 1 hour after phosphate was dosed to the experiment

running for 48 hours without phosphate (b), companion experiment

started with phosphorous (c).........................................................................31

Figure 1-7: Representative lead leaching data from lead pipes in different waters.

Error bars indicate 90% confidence intervals. ..............................................32

Figure 1-8: Lead leaching typically decreased exponentially with time. ........................33

ix

Figure 1-9: Lead leached from brass hose bibs after 58th day .........................................34

Figure 1-10: Effect of nitrate on lead leaching in brass hose bibs exposed to

synthesized water in absence oxidants..........................................................35

Figure 1-11: Effect of nitrate on lead leaching from different hose bib types...................36

Figure 2-1: The lead and copper rule requires at the treatment plant (LCR 1) and in

first draw samples from consumers’ tap (LCR 3). Some sampling is

recommended of service lines (LCR 2). NSF section 8 and section 9 are

performance standards that attempt to insure that devices do not have a

tendency to leach high levels of lead to potable water. ................................57

Figure 2-2: NSF Section 8 certified devices....................................................................58

Figure 2-3: Brass hose bib assembly used in experiment................................................59

Figure 2-4: Equilibrium solubility of lead as a function of phosphate concentration

at pH 5.0. Dissolved inorganic carbon is assumed to be that present at

equilibrium with air (0.14 mg/L). .................................................................60

Figure 2-5: Equilibrium solubility of lead as a function of dissolved inorganic

carbon (DIC) at pH 10. .................................................................................61

Figure 2-6: Amount of lead leached after an overnight stagnation after 9 days from

brass hose bibs. Result is average of triplicate samples. Error bars

denote 90% confidence interval....................................................................62

Figure 2-7: Total and soluble lead leached after an overnight stagnation from the

pure lead pipes after 9 days of dump and fill................................................63

Figure 2-8: Amount of copper leached after an overnight stagnation after 9 days of

dump and fill for NSF pH 5 water with and without phosphate...................64

Figure 2-9: Lead and copper leached after an overnight stagnation as a function of

time for brass hose bibs.................................................................................65

x

Figure 2-10: Lead and copper leached after an overnight stagnation as a function of

days for brass hose bibs. ...............................................................................66

Figure 2-11: Lead leached after an overnight stagnation as a function of time for pure

lead pipes. .....................................................................................................67

Figure 2-12: Allowed lead leaching in NSF devices as a function of the water holding

volume of the product. ..................................................................................68

Figure 3-1: Conceptualization of galvanic effects on lead leaching to water. For a

pure lead pipe, leaching is properly viewed as dissolution, with anodic

and cathodic reactions occurring in close proximity over the pipe surface

(top). Galvanic corrosion driven by coupled lead and copper separates

anodic and cathodic reactions, yielding lower pH near the lead surface

(middle). Deposition corrosion via soluble cupric ions can directly

corrode lead (lower left), and can create micro-cells in which metallic

copper catalyzes oxygen reduction reactions and lowers pH near the lead

surface (lower right)......................................................................................88

Figure 3-2: Final pH as a function of the concentration of lead corroded at the

anode. Temp 25oC, I= 0.001, Solid phases considered include

hydrocerussite, cerussite and Pb(OH)2. ........................................................89

Figure 3-3: Soluble lead as a function of total lead corrosion. At low levels of lead

corrosion in the model system, a greater fraction of the lead is soluble at

higher alkalinity (above). In contrast, at higher levels of lead corrosion,

more lead is soluble at lower alkalinity due to greater reduction of pH at

the anode. Temp 25oC, I =0.001, Solid phase hydrocerussite considered

include hydrocerussite, cerussite and lead hydroxide...................................90

Figure 3-4: Experimental set up.......................................................................................91

Figure 3-5: Effect of higher chloride to sulfate ratio (elevated chloride) on lead

leaching .........................................................................................................92

xi

Figure 3-6: Effect of higher chloride to sulfate ratio (elevated chloride) on current in

galvanically active rigs. ................................................................................93

Figure 3-7: Effect of galvanic connection on lead leaching ............................................94

Figure 3-8: pH measured from the cathode and anode side in a galvanically active

rig separated 2 inches apart...........................................................................95

Figure 3-9: Overall change in pH from the water collected from rigs with lead pipes

placed head to head with copper pipes. ........................................................96

Figure 3-10: Effect of water pH on lead leaching from lead pipes....................................97

Figure 3-11: Effect of installing 2 inch wide ball valve on lead concentrations

between copper and yellow brass devices (Not NSF certified). ...................98

Figure 3-12: Effect of installing 2 inch wide ball valve on current values between

copper and yellow brass devices (Not NSF certified). .................................99

Figure 3-13: Synchronized rise and fall of lead leaching and current on discharging

the gap of 2 inches. .....................................................................................100

Figure 3-14: Effect of iron on lead leaching from copper – lead pipe arrangement kept

head to head ................................................................................................101

Figure 3-15: Effect of copper ions on lead concentration in experiment performed in

lab ware with lead granules in water...........................................................102

Figure 3-16: Effect of adaptation of pure lead pipes and passivation of exposed lead

surface on lead leaching..............................................................................103

Figure 3-17: Prolonged effect of chloramines on current................................................104

Figure 3-18: Demonstrating galvanic relationship between predicted (calculated using

current values) vs. actual lead leaching for brass (top) and lead pipes

(bottom).......................................................................................................105

1

CHAPTER 1

ROLE OF CHLORINE AND CHLORAMINE IN

CORROSION OF

LEAD BEARING PLUMBING MATERIALS

Marc Edwards and Abhijeet Dudi Department of Civil and Environmental Engineering Virginia Polytechnic Institute and State University

Blacksburg, VA 24060 USA ABSTRACT: A switch from free chlorine to chloramine disinfectant triggered problems with excessive lead in Washington DC drinking water. Very high levels of lead originated in the service lines but excessive lead was also derived from solder or brass plumbing materials. In many cases the highest lead emerged from the tap after about one minute flushing, which is troublesome given that routine public notification recommended that consumers’ flush for about one minute to minimize lead exposure. Bench scale testing revealed that chlorine reacts with soluble Pb+2 to rapidly precipitate a red-brown colored lead solid that was insoluble even at pH 1.9 for twelve weeks—this solid did not form in the presence of chloramine. Further experiments revealed that chloramines sometimes dramatically worsened lead leaching from brass relative to free chlorine, whereas new pure lead pipe was not strongly impacted. KEYWORDS: lead, brass, chloramine, chlorine, lead and copper rule

INTRODUCTION:

The leaching of lead to potable water from corrosion of lead bearing plumbing materials

has been managed nationwide by the Environmental Protection Agency (EPA) lead and

copper rule (LCR) (Federal Register, 1991). While recent studies have heightened

concerns regarding low lead exposure on cognitive development of children (Bellinger et.

al., 1991; Garavan et. al., 2000; Werner Troeskan, 2003; Weizsaeker, 2003; Fewtrell et.

al., 2004), actual levels of lead in children’s blood have dropped nearly 80% in the last

quarter century due to improved control of lead paint and dust, a national ban on leaded

gas and success of the lead and copper rule (CDC, 2000). The average national

contribution of drinking water to blood lead is currently believed to be on the order of 7-

20% (Shannon and Graef, 1988; Guidotti; 2004). Given these trends, serious problems

with lead contamination of potable water were largely considered historical. Considering

2

that new plumbing materials are expected to have a much lower propensity to leach lead,

further reductions to drinking water lead were anticipated based on momentum of past

efforts.

Washington DC is served water produced by the Washington Aqueduct and transmitted

to consumers by the DC Water and Sewer Authority (DC WASA). Washington is an

example of a city that once had a lead problem in drinking water but which later

successfully met the lead action limit. LCR sampling events through 1999 continually

confirmed that there were no problems with 90%’ile lead (Figure 1-1). However, after

switching to chloramines in November 2000 some serious problems with lead leaching

started to occur.

Certain factors prevented prompt recognition of a problem. In the LCR round that ended

in spring 2001 (Figure 1-1), the majority of samples were collected before the

disinfectant changeover took place. Moreover, 5 samples collected in that sampling

round after the switch to chloramine contained between 31-113 ppb lead, and these

samples were improperly invalidated (Leonnig, 2004; Holder, 2004). The sample

invalidation caused the reported 90%ile lead to meet the action limit, when in reality the

90%’ile lead did exceed the 15 ppb action limit. Therefore, the late 2001 sampling round

is the first time full effects of chloramine are apparent in reported 90%’ile lead.

Sampling rounds after 2003 utilized a dramatically different sampling pool and results are

therefore not shown, but they confirm a very widespread problem with lead throughout

the system that persisted through at least August 2004. Washington DC consumers have

therefore been exposed to relatively high levels of lead for more than 3.5 years.

Since chloramines decay to form ammonia which can then be converted to nitrites and

nitrates during nitrification, a literature review examining effects of these constituents on

lead corrosion was undertaken to understand possible causes for the phenomenon.

3

Chloramines.

There is a long history in the water industry that documents practical problems from

chloramine attack on brass (Larson et. al., 1956; Ingleson et. al., 1949; anonymous,

1951). Brass is an alloy of copper, zinc, lead and other trace constituents. In tests by

Larson using dechlorinated water, brass faucet seats were determined to fail three times

faster in the presence of a chloramine residual. In these early studies of 1949-1951, it

was noted that the breakdown of the brass faucet seats occurred in regions of the metal

where lead had segregated. It is therefore reasonable to expect that chloramine would

increase lead leaching from brass under some circumstances, even though such

measurements were not made in these older investigations.

In a later study of chloraminated versus chlorinated water in Portland (Portland, 1983),

220 feet copper coils were tested with 50/50 Pb-Sn solder joints at 20 feet intervals. Over

an 18 month study, the samples exposed to chlorine leached median lead values of about

10 ppb with only 1 of 19 samples exceeding 50 ppb Pb. In contrast, in the same water

but with chloramine, median lead was about 100 ppb and 13 of 19 samples exceeded 50

ppb Pb. Thus, the use of chloramine versus chlorine increased median lead leaching by

about a factor of 10. The typical pH of this water was about 6.9, and measurements of

pH tended to be lower in the chlorinated water relative to the chloraminated water. If

lead leaching is assumed to increase at lower pH, the relative adverse impact of

chloramine versus chlorine is even more profound than these data indicate. It was

speculated that the chloramine enhanced galvanic corrosion between the copper and

solder in the pipe rig, since weight loss of pure lead was not increased by the presence of

chloramine relative to free chlorine.

A study by Reiber (1993) did not find noteworthy adverse effects from chloramine on

corrosion of copper, brass, bronze and soldered joints relative to free chlorine. However,

the primary focus of that work was on elastomers and the limited work conducted with

metallic plumbing did not measure lead leaching to the water. Water was also

continuously flowing during these tests, which is an obvious and importance difference

when compared to LCR sampling for lead after mandatory stagnation periods.

4

Effects of chloramine on lead corrosion were studied in Potomac River water by Lin et.

al. (1997). The Potomac River also serves as the source water for Washington DC.

Chloramine actually reduced lead leaching when compared to free chlorine in tests using

pure lead materials in a 1 month experiment at pH 7.2-8.5. Likewise, a parallel

experiment demonstrated that chloramine caused less lead leaching than free chlorine

when pure lead was galvanically coupled to copper. However, in the tests with brass (3%

lead), lead leaching was worsened by the presence of chloramine relative to free chlorine

by a typical factor of about 2. The greatest difference was at pH 8.5, where chloramine

caused leaching of about 5 times more lead than did the same water with free chlorine

(Figure 1-2). This occurred despite the fact that overall corrosion rates on brass were

lower in the presence of chloramine; thus, chloramine selectively increased lead leaching

from the brass as was implied from careful reading of earlier research by Larson and

others. Drinking water collected at homes with high lead in Washington DC had pH up

to 8.67, so the Lin et. al. (1997) result was considered highly relevant.

One confounding factor in the Lin (1997) study is that the original report (Davis et. al.,

1994) indicates that the tests were all done at different times. Test with free chlorine at

different pHs were done from September until October, whereas the corresponding tests

with chloramine were done in February until March. The Lin et.al. (1997) and Davis et.

al. (1994) studies also revealed that various phosphate inhibitors might reduce the

adverse effects of chloramines on lead leaching from brass.

Nitrate/Ammonia.

It is well understood that in alkaline solutions, high concentrations of nitrate break down

pure lead passivity and cause pitting (Rehim and Mohamed, 1998). However, all testing

in that work was done at pH 13 or greater and at > 0.01 M nitrate. Thus, its relevance to

the current problem is uncertain, even though the authors’ data suggested that the

problem would get even worse at lower pHs. The direct attack of metallic Pb by nitrate

was reported by Uchida and Okuwaki, 1998:

5

NO3- + Pb NO2

- + PbO

The nitrite that is produced can react further with lead to form ammonia or N2 gas.

NO2- + 3Pb + 2H2O NH3 + 3PbO + OH-

2NO2- + 3Pb +H2O N2 + 3PbO + 2OH-

Clearly, nitrate can be considered corrosive to lead. It is noteworthy that if nitrifying

bacteria were growing on the lead surface, the above abiotic reactions could remove NO2-

produced by the bacteria, regenerating ammonia and alkalinity required for their further

growth. This could be an especially dangerous synergistic reaction under some

circumstances in water systems. Specifically, nitrifying bacteria could derive substantial

benefit by accelerating the corrosion of lead surfaces.

Even even low levels of ammonia can rapidly attack copper alloys. If a brass metal

sample has residual internal tensile stresses, the ammonia can cause a phenomenon

known as “stress corrosion” in which brass is physically cracked (e.g., Guo et. al., 2002).

This problem becomes more likely above pH 7.0 and in one study was worst at pH 11.4.

Implications of stress corrosion cracking have not been studied at levels of ammonia

typical of drinking water, but it most certainly can be involved in degradation of brass by

chloramines.

There is also a suggestion that the combination of ammonia and higher nitrate would

synergistically drive lead corrosion. Specifically, lead weight loss in the 1998 Uchida

study increased with higher concentrations of ammonia and nitrate in the water. A follow

up study (Uchida and Okuwaki, 1999) revealed that the lead corrosion rate in the

presence of nitrate was higher only if ammonia was also present. Scanning electron

microscopy suggested that ammonia was interfering with formation of a passive scale

layer on the pure lead samples. pHs were not reported, the temperatures studied were

6

high (typically above 40° C), and the levels of the ammonia and nitrate constituents were

also at least 10 times higher than levels typically found in drinking water.

Overall, the literature review and LCR monitoring by DC WASA supported the idea that

chloramine dosing might sometimes increase lead leaching to drinking water. Field

experiments were undertaken to characterize the problem in greater detail. Rigorously

controlled bench scale experiments directly examined effects of chlorine and chloramine

on lead solubility, lead pipe and brass corrosion. The important effects of chloramine on

lead leaching from plumbing materials that are galvanically connected to copper pipe are

discussed in a companion study (Dudi, 2004).

MATERIALS AND METHODS:

Laboratory experiments were conducted to understand effects of oxidants on lead

solubility and lead leaching from brasses or pure lead pipe. In all experiments discussed

below except the field tests, the synthesized water contained 82 mg/L CaCl2.2H2O, 89.6

mg/L CaSO4.2H2O and 84.1 mg/L NaHCO3.3H2O. This recipe has been successfully

used in previous research to simulate the corrosivity of Potomac water based on typical

levels of hardness, alkalinity, sulfate and chloride (Rushing and Edwards, 2004). This

solution is termed synthesized Potomac water in subsequent discussion.

For lead solubility experiments, three different types of synthesized Potomac water were

utilized including water without disinfectant, synthesized Potomac water with chlorine

(10 mg/L as Cl2) and synthesized Potomac water with chloramine (10 mg/L as Cl2 and

ammonia at a 1:1 Cl2/N ratio). Each water was further tested at pH 7.5, 8.25 or 9

(adjusted using 1 N HCl acid or 1 N NaOH base) making 9 tests total. To start the test,

all waters were dosed with 5 mg/L Pb (using PbCl2) and continuously stirred at 200 rpm

for 48 hours. Samples were collected for soluble and total lead concentration after 1, 8,

24 and 48 hours. Soluble samples were first passed through a 0.45 µm pore size filter

before analysis. Free chlorine and total chlorine (± 0.1 mg/L-Cl2) and pH (±0.2 units)

7

were also monitored and maintained at the target value. After the 48 hour sampling

event, the test waters were further dosed with 3 mg/L as P (from NaH2PO4), stirred for an

additional 24 hours, and samples for lead were then collected 1 and 24 hours later. For

comparison, the entire experiment was repeated with 3 mg/L as P dosed before the Pb+2

was added.

Another phase of experiments was conducted using 16 mm diameter pure lead pipes.

Each lead pipe was cut into identical 6.5 cm long pieces to hold 13 ml water each. These

lead pipes were thoroughly cleaned twice with de-ionized water, pickled in 1 N HCl for 6

hours, and then thoroughly rinsed with distilled and de-ionized water. Rubber stoppers

were used to plug both ends of the pipes-- QA/QC determined the stoppers did not leach

significant concentrations of inorganic or organic contaminants. The lead pipes were

exposed to five different synthesized waters including: 1) Potomac Water, 2) Potomac

Water + Ammonia, 3) Potomac Water + Chlorine, 4) Potomac Water + Chloramines, 5)

Potomac Water + Chloramines + Orthophosphate. In these tests ammonia was dosed at a

level of 1.37 mg/L NH3-N and chlorine was dosed at a level of 5.13 mg/L as Cl2 from a

solution of sodium hypochlorite. Chloramines were formed by dosing the ammonia to

the water before free chlorine at pH above 8-- measurements using the DPD method

(AWWA, 1998) confirmed that free chlorine was near detection and that the vast

majority of chloramine was present as monochloramine as would be expected given the

1:3.5 mass ratio of N:Cl2. Phosphate was dosed at a level of 1 mg/L as P using NaH2PO4.

Each condition was tested in triplicate using 3 lead pipes, except for the condition with

phosphate which was only tested in duplicate. Water was changed in the pipes using a

dump and fill protocol every morning Monday through Friday, as well as an extra change

on Friday evening. Samples waters were collected after 16 hours stagnation time from

each lead pipe on the 2nd, 9th, 25th and 58th day of exposure.

The final phase of laboratory experiments used brass devices. Since lead leaching from

brass is known to be highly dependent on the alloy (Lytle and Schock, 1996), brass hose

bibs were selected for testing due to the wide range of brass types that could be obtained

and easy adaptation of the device to experiments (Dudi, 2004a). 8 distinct types of brass

8

hosebibs were exposed to waters 1 through 4 (above), with 1 hose bib per water (8 types

of brass x 4 waters = 32 experiments). The same schedule of dump and fill water

changes was used for brass as for pure lead pipes. After 60 days, the experiment was

continued but with a spike of nitrate (10 mg/L as N from NaNO3) to the water.

Soluble lead was determined after filtering the sample through a 0.45 µm pore size filter.

Samples collected for soluble and total lead were preserved with 5% nitric acid and held

for at least 28 hours before analysis using JY Ultrace Induced Coupled Plasma-Emission

Spectroscopy via standard method 3120 B (AWWA, 1998) with a 3 ppb detection limit for

lead. QA/QC confirmed that this method recovered 100% of all types of lead solids

formed in this study, and comparison of sub-samples to lead detected using mass

spectroscopy or graphite furnace was favorable.

Given the obvious temporal link between the onset of lead problems in the WASA

system and the switch from free chlorine to chloramines, field sampling was conducted in

which “profiles” of lead concentration were collected as water ran from a DC WASA

consumer’s tap. To collect a typical profile, water was rinsed through the lines 10

minutes the night before, at which time a water sample was collected just before the

faucet was closed. After allowing the water to sit stagnant in the consumer’s plumbing

for 10 hours, the faucet was opened at time zero and 1 liter samples were collected after

specified “flush” times. The faucet was opened to a maximum rate in order to approach a

plug flow regime as closely as possible. Similar profiles were collected by Britton and

Richards (1982) and on a much smaller time scale by Lytle et. al., 1995 to reveal sources

of lead to potable water. Samples were also collected from sites after various stagnation

times and flushing times as described in text. In field work, lead was quantified using a

field test kit. A comparison of field test results to those obtained using more standard

techniques for lead quantification were favorable as long as a strong nitric acid digestion

was used.

9

RESULTS AND DISCUSSIONS:

Field-testing was first conducted to define the nature of the lead problem as it occurred in

Washington DC. This was followed by laboratory testing to gain fundamental scientific

understanding of potential contributing causes including lead solubility, testing of pure

lead pipes and then brass.

Field Sampling.

Field testing at a single site between 2003 and 2004 provides insight to the cause of the

lead problem (Figure 1-3). Lead concentrations had been relatively high at the test site in

water samples collected after a consistent 1 hour stagnation time and then 1 minute

flushing before spring 2004. However, during routine springtime use of free chlorine as

disinfectant, lead levels dropped by a factor of 7.6 times compared to the previous sample

obtained when chloramine was used. Lead levels then increased by a factor of 13.6 in a

sample collected 10 days after the switch back to chloramine. DC WASA also collected

monitoring data during this time period and reported up to a 10x drop in lead

concentrations from lead service line samples (Cohn, 2004). The order of magnitude

increase in lead with chloramine versus chlorine is consistent with that observed in

Portland for pipe rigs with lead solder (Portland, 1983). While this type of full-scale data

cannot be used to prove cause and effect, it did provide further support for the idea that

the switch from free chlorine to chloramines was key.

Public education materials distributed by EPA and water utilities often recommend that

consumers’ flush water from the tap for time periods between 30 seconds and several

minutes to minimize lead exposure. It is also common to recommend flushing until the

water temperature changes before collecting volumes to brush teeth, drink or cook (e.g.,

EPA, 2002). These well-intentioned instructions are based on the assumption that the

“first draw” lead sample (i.e., sample collected in the first liter flowing from the tap)

usually contains the highest amount of lead and that lead levels will decrease with

flushing. Consumer’s are often further informed (e.g., EPA, 2004) that flushing is

necessary “any time the water in a faucet has gone unused for more than six hours.”

10

In marked contrast to the conventional wisdom, lead levels were not worst in the first

draw samples at some homes in DC WASA, but were sometimes worst after 1 minute of

flushing (Figure 1-4a). One sample collected after 1 minute flushing was off-scale even

after a 1:10 dilution, and the concentration on the graph is reported at 1,250 ppb even

though it is most certainly > 1,250 ppb. Samples of lead collected after flushing and

measured by DC WASA (using standard EPA sample handling and analysis procedures)

were as high as 48,000 ppb (Nakamura, 2004). This provides indirect support for the

high lead values obtained after flushing using the field test kit in Figure 1-4a. As a point

of comparison, lead levels above 5,000 ppb qualify a water sample as hazardous waste.

The concentration of lead in the samples did not always return to below the action limit

even after 10 minutes of flushing.

Another round of experiments was conducted at one sampling site, during which time a 1

liter sample was collected from the tap after following the publicly recommended 1

minute flush time. The time that water was allowed to sit in the pipes between sampling

events was varied (Figure 1-4b). Even after only 1 hour stagnation, lead levels in the

sample had built up to 140 ppb, demonstrating clear problems well before the 6 hour

threshold cited in some public education materials.

These findings were of obvious public health concern for several reasons. First,

consumers following the written guidance to minimize their lead exposure could actually

markedly worsen their exposure to lead. In retrospect, problems with this advice could

have been anticipated. USEPA has issued explicit instructions aimed at collecting

samples from lead service lines: “Each service line sample shall be one liter in volume

and have stood motionless in the pipe for at least six hours.”…“…[A]llowing the water to

run until there is a significant change in temperature” is “indicative of water that has been

standing in a lead service line.” (USEPA, 2002b). In other words, the USEPA

instructions to collect samples with high lead levels are completely indistinguishable

from the USEPA advice given to consumers regarding collection of their drinking water.

11

Clearly, the advice to collect drinking water after flushing briefly or until water

temperature changes should never be distributed to consumers’ in homes with lead

services when a corrosion problem arises-- unless extensive sampling confirms that the

technique is producing water with acceptable levels of lead. Collection of lead profiles

by WASA at the instruction of the author proved the standard USEPA flushing advice

could be harmful in many cases. In late March of 2004, nearly three and a half years

after the switch to chloramine, the utility publicly recommended 10 minutes of flushing

and use of a lead filter before collecting drinking water in homes that had a lead service

line (Nakamura and Goldstein, 2004).

The second public health concern relates to Centers for Disease Control (CDC)

instructions for assessment of childhood lead poisoning cases (CDC, 2004). The CDC

correctly notes that poisoning from water is unlikely if lead concentrations are at or near

the 15 ppb EPA action limit. However, to assess lead in drinking water, CDC

recommends examination of 90 percentile lead values posted by EPA on the internet. If

the listed 90 percentile value in the community is below the action limit, "no additional

testing is necessary, unless no other sources of a child’s elevated blood lead level can be

found." If testing of water is done it is almost always a first draw sample.

CDC is therefore relying on reported EPA 90 percentile lead values to represent the

potential health threat from drinking water in individual homes, which is unacceptable

given that 10% of homes could have first draw lead at any concentration above the action

limit. These higher lead values are of greatest interest in a case of lead poisoning.

Moreover, the first draw sample tested by CDC could contain much less lead than the

water consumers are actually drinking by following recommendations (Figure 1-4).

Consequently, if lead services were to cause a case of childhood lead poisoning, the

published CDC protocol would have little likelihood of identifying drinking water as the

source.

A recent study in Germany (Fertmann, 2004) attempted to measure average lead

concentration in drinking water, and a correlation was established between blood lead

12

and average lead in drinking water. Consumers with drinking water lead concentrations

above 10 ppb and who switched to bottled water reduced their blood lead levels by 37%

during the study. Public health officials in the US should strongly consider collection of

multiple drinking water samples and use of average lead concentration when assessing

cases of childhood lead poisoning. Use of 90 percentile USEPA LCR concentrations to

assess lead exposure from drinking water is highly misleading.

It is also noteworthy that the sampling site depicted in Figure 1-4 did not have a lead

service line. This suggests a significant problem with solder or brass materials in the

system under at least some circumstances. DC WASA spent $36 million replacing

110,000 water meters in the system in 2002 and 2003 with new devices containing

between 5 to 7 percent lead (Nakamura, 2004a). Recent research (Dudi et. al., 2004a;

Dudi et. al., 2004b) suggests that standards for such devices are not always sufficiently

protective of public health. As of May 2004 it was uncertain if the new meters were a

significant source of lead in the water in Washington DC.

Lead Solubility in the Presence of No Oxidant, Free Chlorine and Chloramines

The solubility of lead was studied as a function of time, pH and oxidant type (chlorine

and chloramines) in water. To provide simple photographic illustration of the dramatic

impact of chlorine on lead solubility, some tests were conducted using the highest level

of lead reported in the DC WASA monitoring data (48,000 ppb Pb). In solutions

maintained at pH 5.5-8.5, red-brown colored solids formed when chlorine was dosed (see

Figure 1-5). In contrast, no solid at all formed at pH 5.5 and a white solid formed at pH

8.5 in the systems dosed with no oxidant or with chloramines. The obvious implication is

that a different type of low solubility lead solid was rapidly formed in the presence of

chlorine.

Experimental results for soluble and total lead concentrations at the end of 48 hours were

representative of trends obtained at shorter time periods (Figure 1-6). Soluble lead was

about 4.4 to 6 times lower in the presence of chlorine at pH 7.5 and pH 9.0 than in

13

comparable systems without oxidant or with chloramine (Figure 1-6a). A less significant

effect was observed at pH 8.25, but chlorine still reduced solubility relative to the other

two conditions at this pH. The soluble lead concentrations in the water without chlorine

were relatively consistent with expectations based on carbonate chemistry and Pb+2

precipitation, whereas those in the presence of chlorine were well below those

expectations (Schock, 1989; Schock and Gardels, 1983). Qualitatively, in the presence of

10 mg/L chlorine, the red-brown solids began to form in the chlorinated water at pH 7.5

in the first 24 hours, but required about 36 hours to form in the same water at pH 8.25

and even longer (48 hours) at pH 9.0. Thus, kinetics of the solid formation seemed

slower at the higher pH in this water, perhaps due to predominance of the weaker oxidant

OCl- at higher pH.

After 48 hours phosphate was spiked into all of the samples for which soluble lead was

described in upper Figure 1-6a. Soluble lead was decreased by phosphate dosing in all

systems within 1 hour, in some cases by as much as a factor of 10 (Figure 1-6b). The

greatest reductions were observed in the system originally dosed with chloramine. In the

companion experiment in which phosphorus was dosed at the start (before the red colored

solids formed), soluble lead also dropped, but red brown solids did not form at any pH

during the test, even though the level of chlorine was unchanged (Figure 1-6c). Thus,

transformations of solids from one type to another will obviously depend on water

quality, disinfectant type/dose, and sequence of chemical addition as has been noted

previously for copper solids (Edwards et. al., 2001; Powers et. al., 2000). In the specific

context of lead solubility, Schock hypothesized (Schock et. al., 1996; Schock et. al.,

2001) that very insoluble PbO2 could form under highly oxidizing conditions, and he

further identified the presence of this solid on lead pipes in chlorinated systems. Given

Schock’s results it is considered highly likely that the low solubility red-brown solids

formed in these tests are comprised of Pb(IV) oxides.

While data are not shown, measurements of total lead in the experiments (soluble lead

plus any particulate lead in suspension) indicated that the red brown lead solid had a

strong affinity for polycarbonate labware. For instance, in one case well over 80%

14

percent of the Pb solid that formed in the presence of chlorine stuck to the labware, but

nearly none of the Pb solid that formed without any oxidant or in the presence of

chloramine did so. The dosing of phosphate tended to cause detachment of these lead

solids that were attached to the labware. It is uncertain whether this observation has any

relevance to the ability of these solids to stick to pipe walls, or for that matter, to any

practical aspect of lead contamination of drinking water. However, it is possible that

phosphates could detach and disperse particulates, consistent with trends in particle

stability reported for iron particles in the presence of phosphate (Lytle and Snoeyink,

2002).

There are several other critical observations of interest to utilities and public health

professionals. First, the red-brown lead particulates are difficult to distinguish from iron

particulates (Figure 1-5) that cause “red water.” Some red solids were collected from the

Washington DC system during field-work for this study that were pure lead oxides.

When samples of pure red brown lead solids were acidified to pH < 2.0 as per

conventional EPA procedures before analysis (i.e. see Parks et al, 2004), the red brown

solid did not dissolve. In fact, even after two months of holding time in the presence of

water acidified to pH 1.9 using nitric acid, only 2% of these solids had dissolved.

Given that even 1 ntu of this red-lead solid was usually undetectable by eye, and yet was

responsible for > 1000 ppb of lead in samples, this is an obvious concern since innocuous

“red water” reports invariably attributed to iron might be due to presence of high

concentrations of Pb solids under rare circumstances. Utilities should be vigilant for such

incidents and should not assume red water is harmless, since it may contain high levels of

lead as well as other contaminants (Reiber et al, 1997; Davis et al, 2000; Schock et. al.,

2003).

In the presence of 5% nitric acid or hydroxylamine, the red-brown lead particles were

rendered completely soluble. Strong nitric acid digestions were used throughout this

research, but are not required by EPA for LCR sampling. It is very likely that the

resistance of the undissolved red-brown lead solids to acid dilution would interfere with

15

lead detection during conventional LCR sampling and analysis. To support this idea, in a

few field samples both strong nitric acid digestion and conventional acidification to < pH

2 were conducted, and in some cases lead levels were 500% high if the stronger digestion

was used compared to the conventional EPA procedure. Thus, there is a large potential

to “miss” lead that is actually present in samples. Similar problems have recently been

documented for Cr(III) by Parks et. al.(2004). It is of obvious future interest to

determine if the lead solids formed in the presence of free chlorine might be less

bioavailable to humans.

Lead Leaching from Pure Lead Pipes

For samples collected from the pipes after 2, 9 and 30 days exposure, there were not

differences significant at 90% confidence in total lead leaching from synthetic Potomac

water with no amendments, with ammonia, with chlorine, or with chloramine (Figure 1-

7). The water from the lead pipe receiving chloramine did consistently have a larger

soluble fraction of lead (66%) than did the other waters (3-30%). Additional testing of

one month duration did not show a significant difference between lead leaching in waters

with or without nitrate. Thus, the direct effect of chloramine on lead leaching from pure

lead in short duration experiments in this type of water is relatively small, consistent with

findings of Portland (1983), Davis et. al., (1994) and Lin et. al. (1997).

Duplicate samples were run for pure lead pipes receiving both chloramine and phosphate.

Surprisingly, this combination consistently had the highest total lead leaching of all

waters tested (Figure 1-7). Unfortunately, soluble lead measurements were not made for

this sample. When a paired t-test was conducted, matching data for lead leaching from

water with chloramine + phosphate versus other waters collected at the same time, lead

leached from the water containing both chloramine and phosphate was highest overall at

> 90% confidence. This result seems to be contrary to the trend of Lin et. al. (1997) for

phosphate effects on lead leaching, but there are many differences in methods that could

explain results. If nothing else, it is clear that phosphate is not always beneficial in the

context of mitigating lead leaching. Resolving such discrepancies should be the focus of

future research.

16

These preliminary results suggest that chloramine does not cause serious problems with

leaching from pure lead pipes relative to the testing in the same synthesized water with

chlorine. Three explanations for the discrepancy in this result and observations at DC

WASA are forwarded. The first is that the PbO2 solids discovered on pipes by Schock

(1996, 2001) may be involved, and that the problem in Washington DC may be caused

only in older lead pipe samples that had pre-existing PbO2 solids built up over decades of

exposure to chlorine (Schock, 2004). Such solids may not be present on the new pipes

used in this work or that of Lin et. al. (1997) or Portland (1983). Another hypothesis is

that the high lead values might result from the presence of nitrifying bacteria growing on

lead pipes as per discussion in the literature review. A third possibility is that the worst

problems might only be manifested when lead materials are galvanically coupled to

copper pipe as discussed in Dudi (2004a) and as suggested by Portland (1983). In most

practical situations, lead materials are galvanically connected to copper pipe. A

combination of the above possibilities could also be operative. Given that the water

industry has used free chlorine and lead pipe for more than 100 years, it seems

remarkable that fundamental direct reactions between chlorine and lead were only

discovered by Schock in 1996-2001 and then further revealed in this research. Clearly,

more research on this subject is needed.

Lead Leaching from Brass Devices

The corrosion chemistry of brass is very complex, with a great deal of variation in lead

leaching amongst the alloys and production methods (Lytle and Schock, 1996). To

attempt to capture the range of possible effects, eight different types of brass hose bibs

were studied for lead, copper and zinc leaching. The brass devices in question contained

between 2 to 8% lead. Manufacturers of these devices were contacted to obtain

information regarding alloys and manufacturing practices, but as of August 2004 this

information was not shared with the authors.

Two types of responses were observed among the eight devices for chloramine when

nitrate was not present. In the typical response followed for 7 of 8 samples, lead leaching

17

started out high and decreased exponentially with time over the 58 days of testing (Figure

1-8). Chloramine typically increased lead leaching compared to the same water with

chlorine. However paired t-testing indicated that this difference was only significant at

the 85% confidence level for 2 of the brass samples over the duration of this phase of

study. This low confidence is partly due to the low number of samples collected, since

the tendency was usually very consistent. A different response was observed for sample

6, which sometimes had 33 times more lead than did the other brass samples exposed to

the same water (Figure 1-9). Lead leaching from sample 6 was very erratic even after

two months.

After 58 days, the experiment was continued, but with addition of 10 mg/L NO3-N to all

waters. Sampling after 4 days revealed a very startling result-- lead leaching dramatically

increased for 7 of 8 brass samples in water without any oxidant present (Figure 1-10).

Limited sampling revealed that a significant amount (up to 15%) of the nitrate was being

converted to ammonia during stagnation in the pipes, a trend that is commonly noted for

nitrate oxidation (corrosion) of iron metal (e.g., Westerhoff, 2003). Thus, nitrate can also

oxidize (corrode) brass as would be expected based on results of Uchida and Okuwaki

(1998). Exceptional behavior was noted for sample 6, in which lead leaching decreased

markedly in response to the higher nitrate (Figure 1-10). The net result is that sample 6,

which had been leaching the most lead by far amongst the 8 brass samples, was suddenly

leaching nearly the same amount of lead (Figure 1-10) as the other devices.

In the presence of chloramine, lead leaching also dramatically increased in the presence

of nitrate, as typified by the 500% increase observed for brass sample 1 (Figure 1-11).

But sample 6 was again the exception, as lead leaching plummeted if nitrate was present

in synthesized Potomac water with ammonia, chlorine or chloramine (Figure 1-11). The

long term effects of nitrate may also be different than the short-term effects, in that after

nitrate exposure was continued for two months, water dosed with chloramine often

leached less lead than the same water dosed with chlorine (data not shown). Given the

limited amount of lead in the alloy, lead can be selectively leached from the surface. It is

therefore conceptually possible that for brass, a water producing very high lead leaching

18

short term could eventually produce very low lead leaching long term because of lead

depletion from the surface.

The results confirm the idea that lead leaching from brass devices can be a complex

function of brass type, exposure time, and water quality (Lytle and Schock, 1996).

Nonetheless, ammonia, nitrate and chloramine can clearly exert a strong influence.

Given the early work documenting high rates of brass failure in the presence of

chloramine, more fundamental research is needed.

19

SUMMARY AND CONCLUSIONS:

A switch to chloramine from free chlorine can trigger dramatically higher total lead

leaching in real circumstances. Among the possible mechanisms for this effect.

1) Free chlorine can dramatically reduce lead solubility relative to waters with

chloramine or oxygen as an oxidant.

2) Under at least some circumstances, chloramines can attack brasses and can

significantly increase lead leaching.

3) A galvanic connection between lead pipe/lead solder to copper pipe may be

involved, since effects of chloramine on leaching from pure lead pipe do not

appear to be significant.

4) Bacteria such as nitrifiers may couple biological and chemical reactions to

exacerbate lead leaching.

Some public education materials routinely distributed to consumers in cases of lead

corrosion problems need revision. Specifically, consumers with lead service lines

should not be instructed to collect samples for drinking after flushing until the

temperature changes or for a time period of seconds to a few minutes, especially

when a corrosion problem is occurring. The lead concentration in pipes can increase

to harmful levels in far less than 6 hours stagnation.

The low solubility red brown colored solid produced in the presence of chlorine

complicates the conventional wisdom regarding lead bioavailability, LCR monitoring

and handling of “red water” complaints. Red water can be a serious health concern in

unusual cases, and certain forms of lead in drinking water systems are poorly

quantified using conventional EPA approved sample handling and analytical

procedures.

20

Until additional experience and understanding is obtained, utilities serving systems

with either pure lead services or pure lead pipes in consumers homes are urged to

exercise considerable caution and diligence when switching from free chlorine to

chloramine. Indeed, any change that could change redox in the distribution system

should be carefully tested before full-scale implementation. Monitoring also should

be in place to confirm that problems with brass failures and lead leaching from solder

and brasses are not triggered by disinfectant changeover. Problems with brass may be

harder to detect due to the limited amount of lead in the alloy.

Given the fact that many utilities are have recently switched to chloramine as a

residual disinfectant, or are considering a changeover in the near future, additional

research on chloramines and all types of material degradation is needed. It is clear

that in some waters, few problems will result from such a changeover, but in other

instances there is ample reason to believe that serious problems will occur. In some

cases decades might pass before adverse impacts on infrastructure longevity would be

revealed, unless a utility was proactively evaluating infrastructure degradation.

21

ACKNOWLEDGEMENTS:

The authors acknowledge the financial support of the National Science Foundation under

grant DMI-0329474. Opinions and findings expressed herein are those of the authors and

do not necessarily reflect the views of the National Science Foundation. The assistance

and insights of consumers’ in Washington DC were critical to success of this work.

Finally, the first author acknowledges the wisdom, advice and leadership of Michael

Schock (USEPA)— a true friend even under the most difficult of circumstances.

REFERENCES:

Anonymous. (1951). Review of Current Investigation No 8: Waterworks Fittings.

Research Group Report. Jour. Inst. Water Engrs. (Br.), 5:700.

AWWA; APHA; & WEF. (1998). Standard Methods for the Examination of Water and

Wastewater, 20th ed. Washington, DC.

Bellinger, D.; Sloman, J.; Leviton, A.; Rubinowit, M.; Needleman, H.L.; & Waternaux,

C. (1991). Low-level lead exposure and children’s cognitive function in the

preschool years. Pediatrics 87:219.

Britton, A. & Richards, W. N. (1981). Factors Influencing Plumbosolvency in

Scotland. Journal of the Institute of Water Engineers and Scientists 35:5:349.

CDC. (2000). Blood Lead Levels in Young Children—United States and Selected States,

1996-1999. Moribidity and Mortality Weekly Report 49, no. 50, 1133.

CDC. (2004). Managing Elevated Blood Lead Levels Among Young Children:

Recommendations from the Advisory Committee on Childhood Lead Poisoning

Prevention. Accessed at

http://www.cdc.gov/nceh/lead/CaseManagement/caseManage_chap2.htm#Figure

%202.1. Accessed Aug. 3, 2004.Cohn, D., 2004. Lead in D.C. Water Slashed.

Decline Comes After WASA Resumes Using Chlorine as Disinfectant.

Washington Post. May 21. B1.

22

Davis, A.P.; Lin, N.H.; & Zeinali, M. (1994). Bench-Scale Evaluation of Lead

Corrosion Control in Drinking Water. Final Report Submitted to Washington

Suburban Sanitation Commission.

Davis, J. (2000). Geo-chemical Controls on Arsenic and Phosphate in Aqueous

Systems. Virginia Tech MS Thesis.

Dudi, A. (2004a). Reconsidering Lead Corrosion Control in Drinking Water: Product

Testing, Direct Chloramine Attack and Galvanic Corrosion. Virginia Tech MS

Thesis.

Dudi, A.; Schock, M.; Murray R.; & Edwards, M. (2004b). Lead in Drinking Water

from In-Line Devices: Making the case for a new standard. Presentation at the

2004 AWWA Inorganics Workshop. Reno, Nevada.

Edwards, M.; Powers, K.; Hidmi, L.; & Schock, M.R. (2001). The role of ageing in

copper corrosion by-product release. Water Science and Technology. 1:3:25.

EPA. (2002). Lead Free Drinking Water in Schools and Day Cares. Accessed August 3,

2004 at http://www.epa.gov/safewater/lead/pdfs/v2final.pdf.

EPA. (2004). Health Effects of Lead. Accessed August 3, 2004 at

www.epa.gov/safewater/lcrmr/pdfs/guidanceprint/guidance_lcmr_lead_public_ed

ucation_cwsbrochnolslbuildings.doc .

Federal Register. (1991). 56:110:26460 Lead and Copper. Final Rule.

http://www.epa.gov/safewater/standard/lead&cop.html

Federal Register. (2002). 40 CFR 141.86. Accessed at

http://a257.g.akamaitech.net/7/257/2422/14mar20010800/edocket.access.gpo.gov/

cfr_2002/julqtr/pdf/40cfr141.86.pdf on August 2, 2004.

Fertmann, R. (2004). Lead Exposure by Drinking Water: An Epidemiological Study in

Hamburg, Germany. Int. J. Hyg. Environ. Health. 207:235.

Fewtrell, L.J.; Pruss-Ustun, A.; Landrigan, P.; & Ayuso-Mateos, J.L. (2004). Estimating

the global burden of disease of mild mental retardation and cardiovascular

diseases from environmental lead exposure, Environmental Research, 94:120.

Garavan, H.; Morgan, R. E.; Levitsky, D.A.; Hermer-Vazquez, L.; & Strupp, B. J.

(2000). Enduring effects of early lead exposure: evidence for a specific deficit in

associative ability, Neurotoxicology and Teratology 22:151.

23

Guidotti, T.L. (2004). as quoted in The Washington Post, May 7 2004 Page B01.

Guo, X.; Chu, W.; Gao K.; & Qiao, L. (2002). Relativity between corrosion-induced

stress and stress corrosion cracking of brass in an ammonia solution. Jour. of

University of Science and Technology Beijing, 9:6:431.

Holder, E. (2004). Summary of Investigation Reported to the Board of Directors of the

District of Columbia Water and Sewer Authority. Accessed August 2, 2004 at

http://www.washingtonpost.com/wp-srv/metro/specials/water/wasa071604.pdf.

Ingleson, H.; Sage, A.M.; & Wilkinson, R. (1949). The Effect of the Chlorination of

Drinking Water on Brass Fittings. Jour. Inst. Water Engineers (Br.)., 3:81.

Larson, T.E.; King, R.M.; & Henley, L. (January 1956). Corrosion of Brass by

Chloramine. Jour. AWWA, 84-88.

Leonnig, C.D. (2004a). EPA Concludes WASA Broke Lead Law. Washington Post,

June 18. Page B1.

Lin, N-H.; Torrents, A.; Davis, A.P.; & Zeinali, M. (1997). Lead Corrosion Control

from Lead, Copper-Lead Solder, and Brass Coupons in Drinking Water

Employing Free and Combined Chlorine. Jour. Environ. Sci. Health, A32(4),

865.

Lytle, D.A. & Snoeyink, V.L. (2002). Effect of ortho- and polyphosphate on the

properties of iron particles and suspensions. Jour. AWWA, 94:10:87.

Lytle, D.A. & Schock M.R. (1996). Stagnation Time, Composition, pH and

Orthophosphate Effects on Metal Leaching From Brass (EPA/600/R-96-103).

Lytle, D.A.; Schock, M.R.; & Sorg, T. J. (1995). Investigation on Techniques and

Control of Building Lead and Copper Corrosion by Orthophosphate and Silicate.

Paper 609. NACE Annual Conference.

Nakamura, D. (2004a). WASA Studying Meters for Lead. Washington Post. May 24.

Page A01.

Nakamura, D.; & Timberg, C. (2004b). Two Lead Readings Reach Into Thousands.

D.C. Households Get No Extra Warning. Washington Post. March 4. B1.

Nakamura, D.; & Goldstein, A. (2004c). WASA Starts to Mail Water Filters Today.

March 23, 2004. B1.

24

Parks, J.; & Edwards, M. (2004). Analysis of Chromium in Environmental Samples. In

press, Water Research.

Portland, (1983). Internal Corrosion Mitigation Study Addendum Report. Bureau of

Water Works. Portland, Oregon.

Powers, K.A. & Edwards M. (2001). The Role of Silica and Chlorine in Cupric

Hydroxide Aging. Proceedings of the Australasian Corrosion Association Annual

Conference in Newcastle, Australia. November, 2001.

Rehim, S.S.A.E. & Mohamed, N.F. (1998). Passivity Breakdown of Lead Anode in

Alkaline Nitrate Solutions. Corrosion Science, Vol. 40, N. 11, pp 1883-1896.

Reiber, S. (1993). Chloramine Effects on Distribution System Materials. AWWA.

Denver, CO.

Reiber, S.; Doastal, G.; Onnen, L.; McCafferty, J.; Sund, W.;& Andreasen, L. (1997)

"Final Report-Groundwater Disinfection and Arsenic Concentration in the

Fremont Distribution System.

Rushing, J.C., & Edwards, M. (2004). Effect of Aluminum Solids and Free Cl2 on

Copper Pitting Corrosion. In press, Corrosion Science.

Schock, M.R. & Gardels, M.C. (1983). Plumbosolvency Reduction by High pH and

Low-Carbonate Solubility Relationships. Jour AWWA, 75:2:87.

Schock, M.R. (1989). Understanding Corrosion Control Strategies for Lead,. Jour.

AWWA, 81:88.

Schock, M.R.; Wagner I.; & Oliphant, R.J. (1996). Corrosion and Solubility of Lead in

Drinking Water in Internal Corrosion of Water Distribution Systems. 2nd Edition.

AWWARF/DVGW-Technologiezentrum. AWWA, Denver, CO (1996).

Schock, M.R. (1999) (Fifth ed.). Internal Corrosion and Deposition Control. Ch. 17 In:

Water Quality and Treatment: A Handbook of Community Water Supplies, pp.

17.01-17.109. McGraw-Hill, Inc., New York.

Schock, M.R.; Harmon, S.M.; Swertfeger, J.; & Lohmann, R. (2001). Tetravalent Lead:

A Hitherto Unrecognized Control of Tap Water Lead Contamination, Proc.

AWWA Water Quality Technology Conference, Nov. 11-15, Nashville, TN.

25

Schock, M.R. & Holm, T.R. (2003). Are We Monitoring in the Right Places for

Inorganics and Radionuclides Journal of the New England Water Works

Association, 117:2:102.

Schock, M.R. (2004). Personal communication.

Shannon, M. & Graef, J. (1988). Lead intoxication from lead-contaminated water used

to reconstitute infant formula. Clin Pediatr 28: 380.

Uchida, M., & Okuwaki, A. (1998). Decomposition of nitrate by in situ abrasion of lead

plate. Hydrometallurgy, 49: 297.

Uchida, M. & Okuwaki, A. (1999). The dissolution behavior of lead plates in aqueous

nitrate solutions. Corrosion Science, 41:1977.

Weizsaecker, K. (2003). Lead toxicity during pregnancy, Primary Care Update for

OBGYNS, 10:6:304

Werner Troesken (2003). Lead Water Pipes and Infant Mortality in Turn-of-the-Century

Massachusetts, NBER Working Papers 9549, National Bureau of Economic

Research, Inc.

Westerhoff, P. (2003). Reduction of nitrate, bromate, and chlorate by zero valent iron

(Fe-0). Jour. of Environmental Engineering – ASCE, 129 :1:10.

26

Figure 1-1: DC WASA 90%’ile lead in LCR Sampling events.

0

10

20

30

40

50

60

70

80

Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03

Date

90%

'ile L

ead

Rep

orte

d to

EP

A

Switch to Chloramine

First LCR report considering full effect of chloramine

EPA Limit15 ppb

Invalidated Samples and Most Samples Collected with Chlorine

27

Figure 1-2: Comparison of impacts of lead leaching from brass in Potomac water with free and chloramine. Representative data are from Davis et. al., 1994.

28

Figure 1-3: Drinking water collected at a Washington DC sampling site after a one hour stagnation time and 1 minute flush time . Error bars during and after switch to free

chlorine are 90% confidence intervals on triplicate sampling events during the same day.

020406080

100120140160

Feb-03 Jun-03 Sep-03 Dec-03 Mar-04 Jul-04

Date

Lea

d (p

pb)

ChloramineChlora-mine

Cl2

15 ppb EPA action limit

29

Figure 1-4: (a) Lead profile from a tap as a function of flushing time (> 10 hour stagnation). (b) Lead levels from a tap after the indicated stagnation time and a fixed 1

minute flush.

0

400

800

1200

1600

0 1 2 3 4 5 6 7 8 9 10Flushing Time (minutes)

Drin

king

Wat

er L

ead

(ppb

)

Residence #1

Residence #2

Samples that would have been collected and consumed based on written recommendations to protect public health (0.5 min to a few minutes)

EPA Pb Action Limit = 15 ppb

(a)

0

400

800

1200

1600

0 1 2 3 4 5 6 7 8Stagnation Time (hours)

Drin

king

Wat

er L

ead

(ppb

) .

Lead collected in drinking water from a residence after the indicated stagnation time and 1 minute flushing

EPA Pb Action Limit = 15 ppb

(b)

30

Figure 1-5: At pH 5.5 solutions with 48,000 ppb Pb+2 are clear since most of the lead is soluble (far left). At pH 8.5 a white solid precipitates (middle left). When chlorine is

present at pH 5.5 or pH 8.5, a highly insoluble red brown solid precipitates (middle right). The red-brown solids are hard to distinguish from “red water” samples caused by corroding iron (far right). The filter paper attached to each container illustrates the color

of the captured particles.

31

Figure 1-6: Lead solubility in three different beaker experiments. Without any phosphate (a), 1 hour after phosphate was dosed to the experiment running for 48 hours without

phosphate (b), companion experiment started with phosphorous (c).

0

0.1

0.2

0.3

Lead

(ppm

)

No phosphorous (48 hours)

(a)

0

0.1

0.2

0.3

Lead

(ppm

)

49 hours (48 hours without P and 1 hour with P)

(b)

0

0.1

0.2

0.3

Pot

omac

Wat

erP

otom

acW

ater

+C

hlor

ine

Pot

omac

Wat

er +

Chl

oram

ine

Pot

omac

Wat

erP

otom

acW

ater

+C

hlor

ine

Pot

omac

Wat

er +

Chl

oram

ine

Pot

omac

Wat

erP

otom

acW

ater

+C

hlor

ine

Pot

omac

Wat

er +

Chl

oram

ine

pH 7.5 pH 7.5 pH 7.5 pH8.25

pH8.25

pH8.25

pH 9 pH 9 pH 9

Lead

(ppm

)

With Phosphorous (48 hours)

(c)

32

Figure 1-7: Representative lead leaching data from lead pipes in different waters. Error bars indicate 90% confidence intervals.

(a) After 2 days

0

4

8

12

16

20

Lead

(ppm

)

SolubleTotal

(b) After 9 days

0

4

8

12

16

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlro

amm

ine

Syn

thet

ic W

ater

+ C

hlor

amm

ine

+ P

hosp

horo

us

Lead

(ppm

)

SolubleTotal

33

Figure 1-8: Lead leaching typically decreased exponentially with time.

Brass sample no. 4

y = 0.6572x-0.4566

R2 = 0.9325

y = 1.2839x-0.5899

R2 = 0.9492

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 10 20 30 40 50 60Days

Lead

(ppm

)

SW + Cl2+ AmmoniaSW + Cl2Power (SW + Cl2)Power (SW + Cl2+ Ammonia)

34

Figure 1-9: Lead leached from brass hose bibs after 58th day

Lead Leached from Brass hose bibsAfter 58 days of exposure

0

0.4

0.8

1.2

1.6

2

2 4 6Brass sample number

Lead

(ppm

)

Synthetic WaterSW + AmmoniaSW + Cl2SW + Cl2+ Ammonia

15 mg/L

35

Figure 1-10: Effect of nitrate on lead leaching in brass hose bibs exposed to synthesized water in absence oxidants.

0

0.1

0.2

0.3

0.4

1 2 3 4 5 6 7 8Brass Sample Number

Lead

(ppm

) Before NitrateAfter nitrate addition (10mg/L - N)

36

Figure 1-11: Effect of nitrate on lead leaching from different hose bib types.

(a) Effect of Nitrate on Brass sample number - 1

0

0.05

0.1

0.15

0.2

0.25

Lead

(ppm

)

Before Nitrate AdditionAfter Nitrate Addition

(b) Effect of Nitrate on Brass sample number - 6

0

0.5

1

1.5

2

2.5

3

Syn

thet

icW

ater

+A

mm

onia

Syn

thet

icW

ater

+C

hlor

ine

Syn

thet

icW

ater

+C

hlor

amin

e

Lead

(ppm

)

Before Nitrate AdditionAfter Nitrate Addition

15 mg/L

37

CHAPTER 2 LEAD LEACHING FROM IN-LINE BRASS DEVICES:

A CRITICAL EVALUATION OF THE EXISTING STANDARD

Abhijeet Dudi†, Michael Schock‡ Nestor Murray† and Marc Edwards †

† Department of Civil and Environmental Engineering

Virginia Polytechnic Institute and State University Blacksburg, VA 24060 USA

‡ Environmental Protection Agency Water Supply and Water Resources Division

National Risk Management Research Laboratory USEPA, Cincinnati, OH 45268

ABSTRACT. The ANSI/NSF 61, Section 8 standard is relied on to protect the public from in-line brass plumbing products that might leach excessive levels of lead to potable water. Experiments were conducted to examine the practical rigor of this test. Contrary to expectations, the test was not highly protective. Rather, it was determined that small devices made of pure lead can easily pass the leaching protocol. Reforms are needed to help prevent such unacceptable outcomes in the future. In the meantime, there is no assurance that brass devices passing this test are safe.

INTRODUCTION:

With the recent crisis of confidence arising from excessive lead in potable water of the

nation’s capital (Nakamura, 2004), a comprehensive national review of lead standards is

inevitable. That comprehensive review should consider the adequacy of existing

standards and sampling protocols, regulatory oversight and response, issues of

simultaneous compliance and resolution of knowledge gaps (Edwards, 2004).

It is widely accepted that minimizing the extent of lead leaching to drinking water from

plumbing materials is a worthy public health goal and the EPA Lead and Copper Rule

(LCR) has achieved great progress in this area (Federal Register, 1991). The amount of

lead leached to public drinking water is a function of two key factors that are under

38

societal control. The first factor is the type of plumbing material used—some plumbing

products have a relatively high tendency to contaminate drinking water with lead. The

second factor is the corrosivity of the water supply that contacts plumbing products.

Achieving the goal of very low lead in drinking water requires progress in both areas.

The EPA LCR has been designed to assess the lead leaching tendency of public water

supply in the context of existing plumbing. Lead present in the one liter "first draw" EPA

LCR sample can come from: 1) lead present in the source water, 2) lead leached to water

as it quickly passes through service lines at high flow rates, and 3) lead leached to water

as it contacts the home plumbing system during prolonged stagnation (Figure 2-1). The

EPA LCR also requires monitoring of lead leaving the treatment plant and has very

limited sampling provisions to detect “spikes” of lead that might build up in water from

prolonged stagnation in service lines. If 90% of first draw samples are not below the 15

ppb lead action limit, the LCR prescribes steps to minimize corrosivity of the water. For

example, by manipulating the chemistry of public drinking water through adjustment of

pH, alkalinity, or by addition of corrosion inhibitors such as orthophosphate, it is often

possible to greatly reduce leaching of lead to public water supplies at relatively modest

cost (Schock, 1989a; Schock et al, 1996; Edwards, 2002a). Such corrosion control steps

can often reduce lead leaching to water from many lead-bearing plumbing products

including pure lead pipe, lead solders and brass.

Progress has also been made in phasing out potentially harmful plumbing products. Pure

lead pipe, leaded solders and brass with > 8% lead content were deemed such a potential

hazard, they were explicitly banned in the 1986 SDWA as amended in 1996 (40 CFR

Part 141, §141.43). More recent progress has been driven by Proposition 65 in

California, which has led to use of very low lead brass residential water meters and

required a more stringent standard for lead discharge from consumer faucets. Proposition

65 has also impacted standards for products that are available nationally (Maas et. al.,

2002a-e). However, the definition of “lead free” brass still allows sale of devices

containing up to 8% lead by weight as long as they are in compliance with performance

standards established in accordance with 42 U.S.C. 300g-6(e).

39

To illustrate the potential magnitude of the problem, we note that a typical gate valve has

at least 100 g of brass that could uniformly corrode away before the product failed. At

5% lead content, the lead leached to water from corrosion could contaminate every drop

of water (to the 15 ppb lead action limit) used by a typical family of four in a year. Of

course, this amount of brass corrosion does not occur in a year, but is spread over the

product lifetime. But considering that multiple brass devices are installed between the

water main and the home, there can be an additive effect. There is also potential for

“spikes” of lead in drinking water coming from the tap, attributable to the water that sat

inside these devices during stagnation.

These considerations illustrate a societal need for a test that prevents installation of brass

products that could leach excessive lead to water. What are the desirable characteristics

of a test? On the practical side, those selling plumbing products are correct to argue that

the test should be of reasonable duration and expense. If the test were too expensive or

required too much time, improved plumbing products might be excluded from market,

which is not in the public interest. It is also critical that the test be reasonably

reproducible in different laboratories around the country and at different times. Finally, it

is critical that the test be prospective, such that products leaching problematic levels of

lead can be excluded from the pool of products that is installed in homes (or schools)

before they pose a direct public health threat. Obviously, many of the home or

institutional plumbing products in question are not visible or even readily accessible once

installed, so a “recall” of dangerous products would be practically impossible.

Some have argued that the EPA Lead and Copper Rule (LCR) itself can protect the

public from brass products that may leach excessive lead. This argument logically fails

for many reasons. In the case of new in-line products, this would require installation of

potentially dangerous untested products in consumers’ homes, sampling according to an

as yet undetermined protocol, and exposing consumers to potentially dangerous levels of

lead throughout the test. The EPA LCR would also have to be modified extensively to

test leaching from in-line devices. At present the LCR does not 1) intend to make sure

40

that homes are sampled with every type of plumbing product that has ever been made, 2)

attempt to document the presence of specific types of brass products present in the homes

which are sampled in the EPA LCR, or 3) purposefully detect spikes of lead that might

occur from brass installed near the service lines. In short, the LCR was not designed, nor

can it rationally be used, to protect the public from potentially harmful plumbing

products.

For new brass plumbing products, there are two current standards designed to provide

protection. The first is NSF 61 Section 9, which tests "end point" devices that can

contact water collected in the first draw sample during stagnation (Figure 2-1). Plumbing

products that pass NSF Section 9 have explicit recognition from the EPA in terms of

limiting lead leaching to water (EPA, 2004). All other brass devices fall under NSF 61

Section 8 including backflow preventers, building valves, check valves, compression

fittings, corporation stops, meter couplings, water meters, strainers, pressure regulators

and many other devices. These are termed “in-line” devices. There is presently no

national legal requirement that in-line devices brass devices meet NSF 61 Section 8

standards. However, 44 states had legislation requiring conformance to these standards

by 2001, and the remaining states had intentions to do so (ASDWA, 2001)

While many in the drinking water industry are familiar with the general reputation of

NSF 61 in certifying plumbing products as safe for drinking water usage, detailed

knowledge of the test protocol is not readily accessible nor understood. Few appreciate

that there are important differences between the test protocols for end-point devices such

as faucets versus those for in-line devices such as meters and shutoff valves as detailed

by Hazan (1994). In-line devices are tested in two waters henceforth termed the “NSF

pH 5” and “NSF pH 10” water. After the devices are exposed to the test waters for up to

14 days conditioning, lead leached to the test water is quantified after a 12-16 hour

stagnation period. The measured concentration of lead from the test is then normalized to

“determine the level of contaminants projected “at the tap” based on the level of

contaminants identified during laboratory analysis” (e.g. NSF, 2002). If the normalized

concentration is less than 15 ppb the product passes the protocol and can be NSF

41

certified, as long as the product also contains less than 8% lead by weight as specified by

law (EPA, 2004).

Products that pass the test can be labeled "ANSI/NSF 61-8 Clean Water for Our Future”

in the marketplace and other prominent displays indicate that devices have passed the

standard (Figure 2-2). A purchaser would be further reassured to read NSF literature

which indicates the NSF 61 Section 8 standard was developed by a consortium of: NSF

International, The American Water Works Association Research Foundation, the

Association of State Drinking Water Administrators, The American Water Works

Association with support from The U.S. Environmental Protection Agency under

cooperative agreement #CR-812144 (e.g. NSF, 2000). In the preface on page v of the

original 1988 NSF 61 Standard, it states: “Standard 61 was developed to establish

minimum requirements for the control of potential adverse human health effects from

products which contact drinking water.”

The point of this discussion is that conscientious members of the public and the water

industry have been give every reason to believe that NSF 61, Section 8 certification

protects against products that would leach excessive concentrations of lead to drinking

water. It is recognized that compromises are necessary in developing a test, and that the

actual concentrations of lead measured during the test are controlled by the test protocol.

Key factors in the protocol include chemistry of the test water, duration of the test,

duration of the contact time of test water and material before sampling, normalization

factors, and reproducibility. This research will carefully examine only the chemistry of

the test waters, normalization factors, and whether the existing test is sufficiently

protective.

MATERIALS AND METHODS:

Sixteen hose bibs that were ANSI/NSF 61 Section 8 certified were purchased from a

local hardware store. While hose bibs themselves do not normally require NSF 61

Section 8 testing, as they are not specifically intended to produce water for human

consumption, they were convenient for testing scientific principles related to aggressivity

42

of the test waters. The hosebibs were covered at one end with a threaded PVC cap and at

the other end with a non-threaded cap (Figure 2-3). During testing the hose bibs were set

horizontally with the outlet directed upwards. Twelve identical sections of 12 cm long

1.7 cm inner diameter pure lead pipe were also used to examine the aggressivity of the

test solutions to pure lead. One end of the pipes was plugged with a rubber stopper that

did not leach lead, whereas the other end of the lead was loosely covered to exclude dust

from entering the device.

The leaded samples were rinsed twice with de-ionized water and then washed twice with

exposure solution. Each type of lead bearing material was filled with water from one of

four different solutions:

1) NSF pH 5.0 water: 203.25 mg/L MgCl2 (51.82 mg/L as Mg) and 347.25

mg/L NaH2PO4 (77 mg/L as P) with 2 mg/L as Cl2.

2) NSF pH 5.0 water without phosphate: same recipe as pH 5 water but

without NaH2PO4.

3) NSF pH 10 water: 476.75 mg/L sodium borate (110.4 mg/L-B) at pH 10

dosed with 2 mg/L as Cl2.

4) Aerated NSF pH 10 water: Same recipe as NSF pH 10 water but aerated for

an hour.

For comparison, some leaded materials were exposed to a relatively non-aggressive tap

water containing 30.5 mg/L alkalinity as CaCO3, pH 8.6, 40 mg/L of Cl- and 50 mg/L of

SO4-2. Fresh solutions were prepared weekly. pH was precisely adjusted to the target

value using 1 N HCl or 1 N NaOH base. Water was changed in the devices as per the

ANSI/NSF 61, Section 8 schedule. Samples of water were collected from each plumbing

specimen for analysis on the 2nd, 9th, 25th and 58th day after a 16 hour stagnation. The

same schedule was followed for the pure lead pipes except that the experiments were

terminated after the 25th day. On occasion, samples collected from the pure lead pipes

were passed through a 0.45 µm pore size nylon syringe filter to quantify soluble lead.

43

Another short-term experiment was designed to test the ability of the NSF protocol to

detect plumbing products that would pose an obvious lead hazard. For the test, an “in-

line device” was made of pure lead that held 6.4 mL of water. This device was then

tested according to the NSF protocol with only 2 days of exposure, which is allowable

under NSF 61 but which increases the likelihood a product would fail. Samples were

collected from the pipe after a 13 hour dwell time and were preserved with 2% nitric acid

and held for at least 28 hours before analysis. Samples were analyzed for phosphorous,

lead, copper, and zinc using Induced Coupled Plasma-Emission Spectroscopy (ICP-ES)

per standard method 3120 B (APHA, 1998).

RESULTS AND DISCUSSIONS:

Results are presented in five sections including: 1) considerations of test water chemistry,

2) effect of the phosphate on leaching propensity of the NSF pH 5.0 water, 3) effect of

CO2 influx on leaching propensity of the NSF pH 10 water, 4) ability of the NSF protocol

to detect an obviously flawed product, and 5) retrospective analysis and

recommendations.

Consideration of Test Water Chemistry.

Many believe that the two NSF Section 8 test waters were specifically designed to extract

a very high amount of lead from a product. To the extent this is true, the test would be

more protective, since concentrations of lead detected in the test water would be much

higher than would be encountered in practice. But close examination of the test water

chemistry and methodology reveal potential problems with that belief.

For the NSF pH 5 water, the concern is the presence of 2.5 mM of orthophosphate. This

is a level of orthophosphate expressed, in common units, as 77 parts per million PO4-P or

240 mg/L PO4. Orthophosphate is present in the test solution as a buffer, but it is also

specifically added to drinking water to reduce lead leaching (e.g., Schock et al, 1996).

Nationally, about half of all major public utilities add phosphate corrosion inhibitors to

44

their drinking water (McNeill & Edwards, 2002). The concentration of orthophosphate in

the NSF pH 5.0 water is between 20-100 times higher than the level of phosphate

commonly added to water by utilities to control lead leaching to drinking water.

Solubility models have been useful in predicting changes to lead leaching in response to

phosphate inhibitor dosing. The models predict that even low doses of phosphate

dramatically reduce lead solubility even at pH 5 (Figure 2-4); therefore, it is expected that

the presence of the phosphate would reduce lead leaching to the test solution (Schock,

1989a; Schock el al, 1996; Edwards et. al., 1999 and 2002a&b). For example, a water at

pH 5 in equilibrium with the atmosphere can hold millions of ppb lead in solution, but the

same water with 77 mg/L orthophosphate present can hold only a few hundred of ppb

lead in solution. This suggests that the NSF pH 5 water is not as aggressive as is

commonly believed given the high concentrations of inhibitor that are present.

Considering the NSF pH 10, which is a pure water buffered with borate, off the record

discussions with laboratories that conduct such testing have indicated that this water is of

highly variable aggressiveness. It is generally reported that products fail more easily in

“fresh” solutions, whereas they pass more easily in older or “seasoned” pH 10 waters.

The test protocol itself does not specify age of the solutions to be used, or the type of

containers used (i.e., open or closed to the atmosphere), which could affect CO2

exchange rates with the water. Some labs use fresh waters while others make up large

quantities of water to use over a period of months. If it were proven that this difference

in approach caused any variability in results, it would be problematic since the same

product might fail in one lab and then pass the test in others.

The notion that storage of the test water alters corrosivity could be anticipated based on

potential dissolution of carbon dioxide from air to the alkaline solution. Lead solubility

is extremely sensitive to dissolution of CO2 in this pH range. In fact, over a reasonable

range of CO2 that might dissolve into the water in a period of months, solubility models

predict that influx of even small amounts of CO2 to the NSF pH 10 water would

dramatically decrease lead leaching (Figure 2-5). This is consistent with water utility

45

experiences, in that high pH and low carbonate waters are known to be highly non-

aggressive to lead leaching based on both solubility models and practical data (Schock,

1980; Schock et al, 1983; Dodrill & Edwards, 1995; Edwards et. al., 1999).

Experiments in the following two sections examine the lead leaching ability of the NSF

pH 5.0 water and pH 10 waters.

Effect of Phosphate on Lead Leaching Propensity of the NSF pH 5 Water.

The concentration of copper, zinc and lead leached from the brass hose bibs and pure lead

pipes were studied as a function of time. Results after 9 days of dump and fill exposure

are selected as illustrative. The phosphate (i.e., inhibitor) in pH 5 water markedly

decreased leaching of lead from both brass (Figure 2-6) and pure lead pipe (Figure 2-7).

For instance, on the 9th day of exposure, average release from the water without

phosphate in brass hose bibs was 0.185 mg/L, consistent with the relatively high

aggressiveness of distilled water noted in Gardels & Sorg (1989). In the NSF pH 5 water

but with the orthophosphate, lead leaching was 5 times lower at 0.038 mg/L (Figure 2-6).

This difference is deemed significant at greater than 99% confidence using a t-test

assuming equal variance (P = 0.005). As a further point of comparison, leaching of lead

to the relatively non-aggressive tap water was 0.16 mg/L. Thus, the non-aggressive tap

water actually leached more than 4 times less lead than did the NSF pH 5 water.

Not surprisingly, levels of lead leached from pure lead pipes during the test were orders

of magnitude higher than for brass (Figure 2-7), consistent with the hazardous nature of

lead pipe in practical situations. More than 25 times less lead was leached to the NSF pH

5 water relative to the same water without phosphate. The NSF pH 5 water leached 3.5

times less lead than did the non-aggressive tap water at pH 8.5.

Interestingly, while the presence of phosphate in the NSF pH 5 water decreased lead

leaching, it increased copper leaching by 14 times relative to the same samples for pH 5

without phosphate buffer (Figure 2-8). This is not surprising (Edwards et. al., 1996 and

46

2002a; Lytle & Schock, 1996), since copper phosphate solids are not predicted to form at

this very low pH, and the phosphate buffer tended to reduce the final pH of water at the

end of the test (pH = 5.2) relative to the same water without buffer (pH = 7.77). Since

the final pH was lower and phosphate could exert no inhibiting action on copper

leaching, the phosphate solution is more aggressive for copper than is the same water

without phosphate (Edwards et. al., 2002a).

The effect of test duration was also studied (Figure 2-9). As expected, a higher amount

of lead was leached from each device at the start of experiment, and it gradually

decreased with time as the sample aged. However, in the presence of phosphate buffer,

lead leaching was relatively low even from the very first sample collected and aging was

less of an issue. In fact, in the presence of phosphate buffer, lead leaching was less than

0.05 mg/L in just 9 days. The opposite effect was observed for copper in the presence of

phosphate. Specifically, with phosphate the copper surface did not passivate, whereas in

the absence of phosphate relatively low amounts of copper were leached from the start of

the experiment (Figure 2-9). Leaching of zinc from the brass was relatively insensitive to

the level of phosphorous, and leaching decreased markedly with exposure time (data not

shown).

Effect of aeration on Lead Leaching Propensity of the NSF pH 10 Water.

After one hour of mild aeration, the NSF pH 10 water had a final pH of 9.2 and an

inorganic carbon content of 37 mg/L. This is only about 30% of the inorganic carbon

expected for a water equilibrated at pH 9.2 CO2 with the atmosphere, so the water

remains highly undersaturated and could have held much more CO2 (Lytle et al, 1998).

This aerated solution was deemed a reasonable compromise between a fresh solution and

what might normally be encountered after months of storage in open containers in

practice, since the aeration increased the inorganic carbon in this highly buffered solution

while only slightly lowering the pH. The pH of this water was re-adjusted back to 10

with freshly prepared NaOH in order to isolate the effect of inorganic carbon alone on the

results. The same water prepared without aeration had a pH of 10 and a typical inorganic

47

carbon content of 1.0 mg/L. Solubility curves predict that lead solubility would actually

be minimized at only 4 mg/L inorganic carbon and maximum at 0 mg/L carbon (Figure

2-4).

Inorganic carbon can have complex effects on lead leaching as a function of exposure

time, since carbonate can form soluble complexes with lead in solution but carbonate can

also form highly insoluble lead solids on the surface of the plumbing material. Since it

takes time to form the insoluble lead solids, the presence of inorganic carbon can increase

lead leaching for a time before ultimately decreasing lead leaching (Schock & Gardels;

1983; Schock, 1989b; Schock et. al. 1996; Edwards & McNeill, 2002b). Consistent with

that expectation, the first measurement (Figure 2-10) demonstrated higher lead leaching

in the aerated water (high carbonate) relative to the same water without aeration (lower

carbonate). But by the 9th day of testing the two results were nearly identical, and at

every time thereafter the aerated water with higher carbonate contained much less lead.

In this case copper showed nearly identical trends to lead (Figure 2-10), as would be

expected given short term importance of soluble cupric carbonate complexes relative to

the longer term importance of insoluble cupric hydroxycarbonate solids on the pipe wall

(Edwards et. al., 1996).

As is expected, when pure lead pipes were exposed to aerated and un-aerated water,

aeration increased lead solubility after two days, but at 9 and 25 days soluble lead was

much lower in aerated water relative to the same water without aeration (Figure 2-11).

This is consistent with utility experiences.

Considering the Overall Rigor of the Test.

It is instructive to consider normalization factors and how they determine whether a

device passes the NSF protocol. The normalization factors attempt to take into

consideration dispersion of water sitting stagnant within in-line devices as it travels to the

tap, the likelihood the water is diluted by collection in a one liter container and a time

adjustment factor. For a 13 hour test using a device that holds 6.3 mls, the actual lead

48

measured in the water at the end of the test is multiplied by a normalization factor of 1.98

x 10-3 to project concentrations at the tap.

A test was conducted using pure lead pipes that held 6.3 mls of water. After conditioning

only 2 days, which is allowed by the protocol and can be only expected to make the

product more likely to fail the test when compared to the 14 day exposure, the

concentration of lead leached to the water was 2100 ppb after an overnight stagnation of

13 hours. The normalized lead concentration was therefore 4.2 ppb (= 2100 * 0.00198).

This is far below the threshold of 15 ppb that would fail the device. This pure lead

device would therefore pass the leaching part of the NSF test using the pH 5 water. This

proves that the NSF test sometimes offers the public very little protection, since it cannot

always detect materials that are known to be hazardous.

A closer look at the normalization calculation used by NSF for certification reveals that

the standard is not actually based on concentration, but rather, the total mass of lead that

is leached to the water during the stagnation time. Thus, for a 12 hour stagnation time, 45

µg of Pb can leach to water regardless of the size of the device. For devices holding

small volumes, this means that very high lead concentrations are required to fail the

standard, whereas devices holding larger volumes will fail at lower levels of lead (Figure

2-12).

It is worth considering the level of protection offered by the standard in the context of

modern regulation and concern about lead in drinking water. A typical residential

plumbing system might have 10 in-line brass devices between the water main and the

consumers tap. There is no upper limit. As the normalization calculation shows, each

device can leach up to 45 µg of lead to water in a 12 hour stagnation period and still

achieve NSF certification. California proposition 65 has legal penalties for

manufacturers of devices that could expose consumers to more than 0.5 µg Pb on any

day. It seems highly likely that, in a relatively aggressive tap water, a consumer could

someday draw a sample of water that contacted the devices in question during stagnation,

and therefore be exposed to much more than 0.5 µg/L Pb.

49

Retrospective Evaluation and Recommendations.

Any reasonable test of a product’s lead leaching tendency requires considerable

compromise. Some compromises made in the NSF 61, Section 8 protocol appear to have

a resulted in a test methodology that gives little public protection. It is useful to consider

“what went wrong” with the NSF process and where, so that lessons can be learned and

similar mistakes in the future might be avoided.

As part of the Drinking Water Additives Program, Task Groups were formed by National

Sanitation Foundation in 1986 and 1987 to develop standards for lead and other trace

metal contamination from plumbing devices and materials in contact with drinking water

(McClelland & Gregorka, 1986; McClelland et. al., 1989). One of the most important

issues discussed is the identity and nature of the test water. In an internal memo, Schock

(1987) stated “I think it is possibly quite misleading, if not erroneous, to use a buffer

system consisting of species which can function as major corrosion inhibitors….” and

“..the strategy of pH buffering using salts that have a profound effect on metal leaching

can not give test results that have any meaning when applied to realistic field

situations…” At that time, the effect of phosphate as an important plumbosolvency

inhibitor had already been established by numerous studies (Schock & Wagner, 1985).

Also, during this time period, NSF conducted an internal exposure methods study using

the original pH 5 and pH 10 proposed solutions, plus a pH 8 water. These tests indicated

that the pH 5 water with orthophosphate caused significantly less Pb release upon 24

hours of exposure than either the pH 8 or pH 10 water. Five test waters were later tested

without any phosphate whatsoever (Schock, 1989a).

By early 1989, the Mechanical Plumbing Products Task Group of NSF adopted the high

alkalinity, pH 8 water as the preferred test water for a new “Section 9” for Mechanical

Plumbing Devices. This water met the criteria of being both relatively aggressive

towards lead and copper, and also being easier to control in composition which is

important given that many laboratories would do testing. Interestingly, the February 1,

1989 draft standard for Section 9 had the following proposed coverage:

50

“This section covers the following plumbing devices installed in the water distribution system of a building and components or materials used in devices as follows: kitchen faucets; laundry faucets; lavatory faucets, bar sink faucets; rough, in line control valves; backflow prevention devices; water coolers; drinking fountains; water heaters/storage devices/tanks; water meters and related fittings; supplies and supply stops; drinking water treatment units; ice makers and similar devices as are intended by the manufacturer for drinking and culinary purposes.”

Clearly, it was originally intended that many in-line devices and water meters, now

covered in Section 8, were to have been covered in Section 9. This change was important

given the relatively non-rigorous nature of Section 8 as revealed in this research. Water

industry personnel should be made aware of this potential problem.

It is well known that manufacturers have invested considerable money in producing

devices that can meet NSF certification via surface washes or coatings. However, there is

no guarantee that the techniques reducing lead leaching to a pH 5 water loaded with

phosphate inhibitor will have anything to do with reducing lead leaching to real water,

especially over long term exposures. The NSF testing protocol does not consider the fact

that brass galvanically coupled to copper in practice will be sacrificed (Dudi, 2004;

Schock, 1989), thereby enhancing lead leaching relative to a test on a stand alone product.

Nor does test the likelihood that chloramine is more aggressive than chlorine (Dudi,

2004). At a minimum, the NSF Section 8 protocol needs an overall if it is to meet its

intended purpose.

It would also be desirable to have a standard that is open to public scrutiny and written in

a comprehensible manner. Copies of the standard have numerous typos and directions

are extremely difficult to follow. Requests made of NSF to clarify the simplest issues,

such as whether the pH 5.0 water actually had orthophosphate in it, met with no response.

The current copies of the standard are sold by NSF for several hundred dollars per copy.

Thankfully, previous researchers attempting to quantify the leaching propensity of leaded

brass devices have invariably used a more realistic water regardless of NSF test protocol

specifications (e.g., Maas et. al., 2002a-e; Patch et. al., 1998). The collective work of

51

Maas and Patch (1997-2002) demonstrates a significant leaching potential for “lead-free”

devices, as defined by the Safe Drinking Water Act (EPA, 2004). Given our experiences

with the voluntary NSF standard and the experimental data reported herein, we now

question the ability of NSF to safeguard public health in the context of lead leaching to

potable water. A dramatic reduction in the allowable lead content of brass products

would seem highly desirable relative to experiences to date under the voluntary standards

program. Future research will be necessary to assess the extent of the problem in terms

of lead leaching from water meters and other devices.

52

CONCLUSIONS:

- The use of phosphate buffer in NSF pH 5.0 test waters dramatically reduces lead

leaching from brass devices. Consequently, the phosphate also acts to inhibit lead

leaching to the test water, rendering it less aggressive than many tap waters. The

addition of phosphate to the pH 5 water increases copper leaching relative to the same

water without phosphate, making it more aggressive than many tap waters.

- Influx of carbon dioxide to the pH 10 water can dramatically alter its propensity to

leach lead. Over the short term, preventing CO2 influx increases lead solubility by

preventing formation of soluble lead carbonate complexes. After about one week,

dramatic reductions in lead leaching are observed, most likely due to formation of

lead scales that limit solubility. A failure to specify “freshness” of the solutions

therefore allows unacceptable variability, and may allow certification of devices that

leach excessive lead to water. It has been shown that a small device made of pure

lead can pass the NSF leaching test.

- The NSF 61 Section 8 standard for brass in-line devices does not appear to be

sufficiently protective to public health in the context of modern regulations such as

California’s proposition 65. A reduction in allowable lead content in brass from the

very high 8% level now allowed is now feasible and highly desirable.

53

ACKNOWLEDGEMENTS:

The authors acknowledge the financial support of the National Science Foundation under

grant DMI-0329474. Opinions and findings expressed herein are those of the authors and

do not necessarily reflect the views of the National Science Foundation or of the USEPA.

Any mention of products or trade names does not constitute recommendation for use by

the National Science Foundation or USEPA.

REFERENCES:

ASDWA. (2001). Association of State Drinking Water Administrators Survey on State

Adoption of ANSI/NSF Standards 60/61. Washington, DC.

APHA. (1998). Standard Methods for the Examination of Water and Wastewater. 20th

Edition, United Book Press, Inc; Baltimore, Maryland.

Dodrill, D. & Edwards, M. (1995). Corrosion Control on the Basis of Utility

Experience. Journal AWWA, 87:7:74.

Dudi, A. (2004). Reconsidering Lead Regulations in Drinking Water: Testing of

Products, Chloramination and Galvanic Corrosion. Virginia Tech MS T hesis.

Edwards, M; Schock, M.R.; & Meyer, T.E., (1996). pH, Alkalinity and Copper

Corrosion By-Products. Journal AWWA. 88(3), 81-94.

Edwards, M.; Jacobs, S.; & Dodrill, D. (1999) Desktop Guidance for Mitigating Pb and

Cu Corrosion By-Products. Journal AWWA., 91:5:66-77.

Edwards, M.; Hidmi, L.; & Gladwell, D. (2002b). Phosphate Inhibition of Soluble

Copper Corrosion By-Product Release. Corrosion Science. V. 44 1057-1071.

Edwards, M. & McNeill, L.S. (2002a) Effect of Phosphate Inhibitors on Lead Release

from Pipes. Journal AWWA. V. 94, No. 1, 79-90.

Edwards, M. (March 5, 2004). Testimony to the 108th Congress of the United States.

Lead in DC Drinking Water. House Committee on Government Reform (29

pages).

54

EPA. (2004). Web page accessed 5/24/2004.

http://www.epa.gov/safewater/lead/testing.htm

Federal Register, 56:110:26460 (June 7, 1991) Lead and Copper. Final Rule.

Gardels, M.C. & Sorg, T.J. (1989) A Laboratory Study of the Leaching of Lead from

Water Faucets. Journal AWWA, V. 81, No. 7 101-113.

Hazan, S. S.; Gebhart, A. M.; & Epstein, P. S. (1994) “Evaluation of Mechanical

Products and Plumbing Products for Lead Under ANSI/NSF Standard 61.” Proc.

AWWA, New York, New York, June 19-23.

Lytle, D.A. & Schock, M.R. (1996). Stagnation Time, Composition, pH and

Orthophosphate Effects on Metal Leaching from Brass (EPA/600/R-96-103).

Lytle, D.A.; Schock, M.R.; Clement, J.A.; & Spencer, C.M. (1998) Using aeration of

corrosion control. Journal AWWA V. 90, No. 3 74-88.

Maas, R.P.; Patch, S.C.; Morgan D.M.; & Kawaguchi. H. (1997). Lead Leaching From

Brass Water Meters Under Pressurized Flow Conditions. Proceedings of The

American Water Works Association Annual Conference, pp. 589-602.

Maas, R.P.; Parker A.; & Patch. S.C. (1998). Lead Leaching From In-Service Leaded-

Brass Water Meters. N.C. Journal AWWA, Vol 1:270-279.

Maas, R.P.; Patch, S.C.; Morgan D.M.; & Loucaides, S.C. (1999a). Lead Leaching

From Los Angeles In-Service Brass Water Meters. UNC Asheville Environmental

Quality Institute Technical Report #99-070. 12 p.

Maas, R.P.; Patch, S.C.; Morgan D.M.; & Loucaides, S.C. (1999b). Lead Leaching

From Los Angeles In-Service Brass Curb Valves. UNC Asheville Environmental

Quality Institute Technical Report #99-068. 17 p.

Maas, R.P. (2002a). Estimation of Human Lead Exposure From Leaded Brass Water

Works Parts Installed in Residential Service. UNC Asheville Environmental

Quality Institute White Paper. April 2002. 9 p.

Maas, R.P.; Patch, S.C.; & Parker, A.F. (2002b). An Assessment of Lead Exposure

Potential From Residential Cut-off Valves. Journal of Environmental health,

Vol. 65(1):9-14.

Maas, R.P.; Patch, S.C.; & Lagasse, M. (2002c). Measurements of Lead, Bismuth and

Selenium Discharge From Water Service Valves and Fittings Manufactured From

55

Sebilay (EnvironBrass) Brass Alloy. UNC-Asheville Environmental Quality

Institute Technical Report #02-105. 11p.

Maas, R.P., Patch, S.C.; Berkowitz, J.; & LaGasse, M. (2002d). Comparison of Lead

Discharge From Conventional Leaded Brass Versus No-LeadType Water Service

Valves and Fittings. UNC Asheville Environmental Quality Institute Technical

Report #02-097. 29 p.

Maas, R.P.; Patch, S.C.; LaGasse, M.; & Morgan, D.M. (2002e). Lead Discharge From

Leaded-Brass Residential Gate Valves, Pressure-Reducing Valves and Back-Flow

Preventers. UNC Asheville Environmental Quality Institute Technical Report

#02-098.

McClelland, N. I. & Gregorka, D. A. (1986). “The Drinking Water Additives

Program.” Water Engr. Mgmt., 12:18.

McClelland, N. I.; Gregorka, D. A. & Carlton, B. D. (1989) “The Drinking-Water

Additives Program.” Envir. Sci. & Technol., 23:1:14.

McNeill, L.S. & Edwards, M. (2002) Phosphate Inhibitor Use at US Utilities. Journal

AWWA, V. 94, No. 7, 57-63.

Nakamura, D. (January 31, 2004). Water in DC Exceeds EPA Limit. Page A1.

NSF. 2000. (2000a) Drinking Water System Components-Health Effects. ANSI/NSF 61.

Patch, S.C.; Maas, R.P.; & Pope, J. (1998). Lead Leaching from Faucet Fixtures Under

Residential Conditions. Journal of Environmental Health, 61(3): 18-21.

Schock, M. R. (1980) Response of Lead Solubility to Dissolved Carbonate in Drinking

Water. Jour. AWWA 72:695-704.

Schock, M.R. & Gardels, M.C. (1983) Plumbosolvency Reduction by High pH and Low-

Carbonate Solubility Relationships. Journal AWWA, 75:2:87.

Schock, M.R. & Wagner, I. (1985) “The Corrosion and Solubility of Lead in Drinking

Water.” Ch. 4 in: Internal Corrosion of Water Distribution Systems. AWWA

Research Foundation/DVGW Forschungsstelle, Denver, CO, p.213.

Schock, M.R. (October 19,1987). Personal communication to Tom. S. Grable, NSF

Mechanical Devices Task Group.

Schock, M. R. (1989a) “Extraction Water Justification Report,” internal report prepared

for National Sanitation Foundation, Ann Arbor, MI.

56

Schock, M. R. (1989b) Understanding Corrosion Control Strategies for Lead. Journal

AWWA 81:88-100.

Schock, M.R.; Wagner, I. & Oliphant, R.J. (1996) “Corrosion and Solubility of Lead in

Drinking Water” in Internal Corrosion of Water Distribution Systems. 2nd

Edition. AWWARF/DVGW-Technologiezentrum. AWWA, Denver, CO.

57

Figure 2-1: The lead and copper rule requires at the treatment plant (LCR 1) and in first

draw samples from consumers’ tap (LCR 3). Some sampling is recommended of service

lines (LCR 2). NSF section 8 and section 9 are performance standards that attempt to

insure that devices do not have a tendency to leach high levels of lead to potable water.

58

Figure 2-2: Public advertisement of NSF Section 8 certified devices.

59

Figure 2-3: Brass hose bib assembly used in experiment

60

Figure 2-4: Equilibrium solubility of lead as a function of phosphate concentration at pH

5.0. Dissolved inorganic carbon is assumed to be that present at equilibrium with air

(0.14 mg/L).

61

Figure 2-5: Equilibrium solubility of lead as a function of dissolved inorganic carbon

(DIC) at pH 10.

62

Figure 2-6: Amount of lead leached after an overnight stagnation after 9 days from brass

hose bibs. Result is average of triplicate samples. Error bars denote 90% confidence

interval.

Average Amount of lead leached at the end of 9th day

0

0.05

0.1

0.15

0.2

0.25

NSF pH 5 Water NSF pH 5 Waterwithout P

Tap Water

Lead

leac

hed

(mg/

L)

63

Figure 2-7: Total and soluble lead leached after an overnight stagnation from the pure

lead pipes after 9 days of dump and fill.

Total Lead

0

20

40

60

80

100

120

140

160

180

200

NSF pH 5 NSF pH 5Without

Phosphorous

NSF pH 10 Tap Water

Type of water

Am

ount

of l

ead

leac

hed

(mg/

l)

2.44 mg/L

62.9 mg/L

8.69 mg/L

175 mg/L

Soluble Lead

0

10

20

30

40

50

60

70

NSF pH 5 NSF pH 5Without

Phosphorous

NSF pH 10 Tap Water

Type of water

Am

ount

of l

ead

leac

hed

(mg/

l) .

0.638 mg/L

28.7 mg/L

0 mg/L

53.7 mg/L

64

Figure 2-8: Amount of copper leached after an overnight stagnation after 9 days of dump

and fill for NSF pH 5 water with and without phosphate.

Total Copper

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

NSF pH 5 Water NSF pH 5 Water without P

Tota

l Cop

per (

mg/

L)

65

Total Lead

0

0.5

1

1.5

2

0 10 20 30 40 50 60

Days

Am

ount

of L

ead

Leac

hed

(mg/

L) NSF pH 5 WaterNSF pH 5 Water without Phosphate

Figure 2-9: Lead and copper leached after an overnight stagnation as a function of time

for brass hose bibs.

Total Copper

0

1

2

3

4

0 10 20 30 40 50 60

Days

Am

ount

of C

oppe

r Lea

ched

(mg/

L)

NSF pH 5 WaterNSF pH 5 Water without Phosphate

66

Figure 2-10: Lead and copper leached after an overnight stagnation as a function of days

for brass hose bibs.

Total Lead

0

0.1

0.2

0.3

0.4

0.5

0.6

0 10 20 30 40 50 60Days

Am

ount

of L

ead

leac

hed

(mg/

l)NSF pH 10 Water

NSF pH 10 Water Aerated

Total Copper

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 10 20 30 40 50 60Days

Am

ount

of C

oppe

r Lea

ched

(mg/

L)

NSF pH 10 WaterNSF pH 10 Water Aerated

67

Figure 2-11: Lead leached after an overnight stagnation as a function of time for pure

lead pipes.

NSF pH 10 Water

0

50

100

150

200

250

2 9 25Days

Lead

leac

hed

(mg/

L)

NSF pH 10 Total

NSF pH 10 Soluble

68

Figure 2-12: Allowable lead leaching from NSF devices as a function of the water

holding volume of the product.

69

CHAPTER 3 GALVANIC CORROSION OF LEAD BEARING PLUMBING DEVICES

Abhijeet Dudi and Marc Edwards Department of Civil and Environmental Engineering Virginia Polytechnic Institute and State University

Blacksburg, VA 24060 USA ABSTRACT: Lead contamination of water contacting brass or lead bearing plumbing can increase if there is a galvanic connection to copper. The rate at which lead materials were sacrificed by galvanic action was directly influenced by various factors including chloride and sulfate concentrations, type of disinfectant, pH and insulating gaps between the metals. The galvanic current spatially separates the anodic and cathodic reactions, thereby lowering the pH near the surface of the lead materials in poorly buffered waters and increasing lead solubility. Chloramine caused sustained galvanic currents relative to chlorine. The partial replacement of lead service lines by new copper pipes can be expected to increase lead leaching under at least some circumstances.

KEYWORDS: lead, brass, galvanic, current, pH, chloramines

INTRODUCTION:

A typical plumbing system in buildings will include direct electrical connections between

lead bearing materials and other metals. For instance, leaded solders are in electrical

contact with copper tube at joints (Reiber, 1991), in-line and endpoint leaded brass

devices are directly connected to copper tube, and it is still common to connect new

copper pipe directly to old lead pipe during partial lead service line replacements. While

the issue of galvanic corrosion for dissimilar metals is often discussed, most commonly

from the perspective of premature material failure, there are numerous practical gaps in

understanding. In particular, there is little research work on factors arising from galvanic

corrosion that increase lead contamination of drinking water. This work will address that

deficiency.

Key hypothesis for galvanic phenomenon are pictorially developed in Figure 3-1.

Corrosion of pure lead pipe proceeds via cathodic and anodic reactions occurring in

70

relative close proximity over the pipe surface (Figure 3-1 top). The initial anodic

reaction is:

Pb Pb+2 + 2e- Reaction 3-1

and a representative cathodic reaction is oxygen reduction:

O2 + 4e- + 2H2O 4OH- Reaction 3-2

Dependent on pH, pipe age and other factors, the Pb+2 ions are released and can form

various complexes or solids, yielding the following representative overall corrosion

reactions:

Pb + ½ O2 +H2O 2OH- + Pb+2 Reaction 3-3

Pb + HCO3-+ ½ O2 OH- + PbCO3 Reaction 3-4

Pb + ½ O2 +H2O Pb(OH)2 (s) Reaction 3-5

While the exact final disposition of lead depends on numerous factors, Reaction 3-3 and

Reaction 3-5 define extremes of effects on solution pH under conditions typically

encountered in potable water supplies. That is, reaction 3-3 releases 2 OH- per Pb

corroded, whereas reaction 3-5 neither consumes nor produces OH- during corrosion.

Reaction 3-4 is in between these extremes, leading to production of 1 OH- per Pb

corroded. As a result, the overall electrochemical corrosion reaction will either increase

or have no effect on the pH of water inside a pure lead pipe. To the extent that higher pH

generally indicates lowered lead solubility and propensity for contamination (e.g. Schock

1989; Dodrill et. al., 1995, Lin et. al. 1997), targeting a minimally acceptable pH in the

water served to homes will provide an upper bound to lead contamination, and corrosion

reactions within pipes that further raise pH would further improve the situation.

When two dissimilar metals are electrically connected the anodic and cathodic reactions

are separated. For materials commonly used in water supplies, the galvanic series defines

the typical anode or cathode in such connections:

1) Copper 2) Brass (with lead) 3) Lead 4) Iron 5) Zinc

Cathodic

Anodic

71

Lead and leaded brass are therefore expected to be anodic relative to copper tubing. This

situation can be expected to dramatically alter the danger of corrosion in the context of

lead leaching (Figure 3-1, middle). First, under these circumstances, the rate of lead

corrosion can be increased above that which occurs for a pure lead pipe, whereas the rate

of cathodic reactions such as oxygen reduction on the copper pipe surface is also

increased (Jones 1992). This acceleration to corrosion reactions is termed galvanic or

contact corrosion (Reiber, 1991; Shreir et. al 1995). The second and perhaps more

important implication, is that the representative net reactions occurring near the lead

anode surface are altered to the following:

Pb 2A- + Pb+2 Reaction 3-6

Pb + 2H2O Pb(OH)2 + 2H+ + 2A- Reaction 3-7

where 2A- represents transport of anions in the water (e.g., Cl-) towards the anode to

balance electroneutrality. As before, the above reactions describe a range of anticipated

behavior, in that near the lead surface the net reaction either has no effect (Reaction 3-6)

or decreases (Reaction 3-7) pH as a result of corrosion. We further note that if the pH is

not decreasing (Reaction 3-6), the lead is in a form that will not produce a protective

scale on the surface of the anode. The combined effect of higher corrosion rate and

production of acidity near the anode might sometimes dramatically worsen lead leaching

relative to the situation depicted for pure lead pipe via Reaction 3-3.

Another important yet unappreciated galvanic phenomena is known as deposition

corrosion (Figure 3-1, bottom). Deposition corrosion can occur when soluble ions from a

more cathodic metal are present in the drinking water flowing through a pipe. For

example, cupric ions released from a copper pipe installed upstream of lead pipe, can be

directly deposited and plated onto the lead surface, directly corroding the underlying lead

metal. The newly plated copper metal forms a micro galvanic cell that can catalyze

corrosion of the underlying lead indefinitely (Figure 3-1 bottom right).

72

The general concept of deposition corrosion is well established scientifically there are

numerous warnings to avoid the practice (NACE, 1984). The water industry also has well

documented experiences with deposition corrosion, since the fact that dissolved copper

accelerates corrosion of iron pipe is well understood (Hatch 1955, Cruse 1971, McNeill

et. al. 2001). While we could find no research reporting impacts of cupric ion on lead

pipe or brass corrosion in potable water, Cifuentes (2001) reported that copper ions

increased corrosion of lead anodes in commercial plating operations. Thus, the topic

certainly seems worthy of study given that new copper pipe is sometimes installed

upstream of old pure lead pipe.

There is sufficient reason to believe that impacts will vary markedly from water to water.

For instance, while the galvanic series for metals presented earlier is a good guide, it is

overly simplistic for potable water applications. Under non-standard conditions and in

the presence of scale and other water constituents, the electrochemical potential of copper

(Ecorr) can vary by several hundred millivolts. Indeed, recent work has demonstrated that

in certain waters with relatively high levels of pH, aluminum solid and free chlorine

residuals, the potential of copper can rise to nearly 800 mV versus a AgCl electrode from

more common potentials of approximately 0 mV (Edwards 2002, Edwards et. al. 2003,

Edwards et. al. 2004, Marshall et. al. 2003, Rushing et. al. 2002). Consequently, in

some waters copper surfaces will evolve to a much stronger cathode than in other

systems; indeed, in some situations where leaded materials are electrically connected to

copper, reactions occurring on the copper surface might even be more important than

those on the lead surface since the cathodic reaction is often the rate limiting step in

corrosion. Likewise, there is evidence that the Cl-:SO4-2 ratio can influence lead

contamination of potable water. Oliphant (1983) and Gregory (1990) noted enhancement

to galvanic corrosion currents and lead leaching when Cl-:SO4-2 ratio was high in bench

scale studies of lead leaching. Also, examination of utility monitoring data illustrated

that waters with a high Cl-:SO4-2 ratio had a tendency for higher 90%’ile lead values

(Edwards et. al., 1999).

73

It is instructive to model the potential effect of separated anodic reactions on lead

solubility in water as a function of dissolved inorganic carbon (DIC). Mineql+ was used

to predict the final pH of water at initial pH 8.5 and variable DIC as a function of Pb+2

release from the anode (Edwards and Schecher, 1996) using stability constants described

in Dodrill et. al. (1995). As more Pb+2 is released to water near the anode from

corrosion, the local pH decreases markedly. At DIC of 0.25 mM (12.5 mg/L alkalinity as

CaCO3), leaching of 10 mg/L lead to water as Pb+2 would depress solution pH down to

about 6.7 from an initial pH of 8.5 (Figure 3-2). At higher DIC of 0.010 M (500 mg/L

alkalinity as CaCO3), buffering reduces the impact of 10 mg/L release to the point that

the pH drop would be undetectable. Release of cupric and zinc species from brass anodes

would also tend to produce acid and decrease pH.

It is also useful to consider the ultimate disposition of lead leached from the anode in

Figure 3-2. That is, the lead that is corroded could either remain soluble or react with

carbonate/hydroxide species in the water to form solids. It is possible that precipitated

lead would attach to the pipe whereas soluble lead would undoubtedly contaminate the

drinking water. The first increment of lead corroded is predicted to be 100% soluble

(Figure 3-3a). If a highly insoluble lead oxide phase was forming and controlling

solubility, the water would hold more lead at higher alkalinity (Figure 3-3), since the pH

drop from low levels of lead corrosion is very small. As a result, more lead would be

soluble in a water with 1 mM DIC versus 0.01 mM DIC after 1 µm Pb+2 had leached to

water (Figure 3-3a). However, if much higher amounts of lead were corroded (Figure 3-

3b), the converse becomes true in that lower DIC water is predicted to have a greater

percentage of soluble lead. This is due to the lowered buffering capacity of the low DIC

water and resultant lower pH as discussed earlier. Thus, the net effect of galvanic

corrosion on lead contamination is predicted to be a function of DIC, pH, extent of

corrosion and types of lead solids that form on the pipe surface.

The most recent research on the subject of galvanic corrosion was conducted by Reiber

(1991), who directly examined the electrochemistry of galvanic corrosion for samples of

50:50 lead:tin solder coupled to copper tube. In samples exposed to continuously

74

recirculated and flowing water, galvanic currents at pH 7 dropped to low levels after only

two weeks and were not deemed worrisome. However, the conceptual development of

Figure 3-1 suggests that changes in experimental protocol consistent with residential use

and practice might lead to a markedly different conclusion. For instance, if long periods

of stagnation were imposed on the system, acid production at the lead surfaces could

permanently hinder passivation. In partial support of this hypothesis, Reiber noted that

dropping pH to 5.0 from 7.0 almost completely reversed the passivation of galvanic

corrosion that had occurred. Briton and Richards (1981) also presented practical data

suggesting much higher lead concentrations were in water after stagnation when galvanic

connections were present between lead bearing materials and copper. The authors further

recommended that creation of galvanic connections should be avoided whenever

possible. The effects of chloramine also need consideration (Dudi, 2004), given that 10

fold higher leaching of lead in water with chloramine versus chlorine was suspected to be

due to galvanic reactions (Portland, 1983).

MATERIALS AND METHODS:

Six yellow brass hose bibs, six NSF 61 Section 8 certified red brass hose bibs and 12

sections of hard copper pipe (6’ each) were procured from a local hardware store and

used after rinsing twice with distilled-deionized water. Pure lead pipe sections 6.5 cm

long that had been exposed 2 months to the test waters were also utilized as described in

Dudi, 2004.

Pipe rigs were constructed in different configurations to accomplish various objectives.

One key parameter is the distance or gap between the anodic and cathodic pipe sections

(Figure 3-4). In experiments herein, three different gaps were used, including 1) 2 inches

apart with brass devices separated from copper pipes using a PVC ball valve with an

external grounding strap wire connected if desired, 2) direct connection between the two

dissimilar metals using a rubber clamp instead of solder, and 3) a 2 mm gap between the

metals created with a thin plastic washer (Figure 3-4). In one test to simulate possible

75

impacts of connecting copper to a steel main, a mass of steel wool was placed inside a

PVC tube and connected to copper as depicted in lower Figure 3-4.

Water was sampled from each rig after an 8 hour stagnation time unless specified

otherwise in later text. Galvanic current and potential between the dissimilar metals were

measured with an inexpensive multi-meter (10 Mega Ohm Internal Resistance in parallel

with 2pF capacitance). The currents were measured between the metals by connecting

the multimeter in-line for 5 seconds after disconnecting the wire between the two metals.

Potential measurements were made after 2 minutes in a DC voltage mode with external

grounding wire disconnected. Ecorr was measured using Ag-Ag (silver) electrode.

The synthesized tap water used in the tests contained 82 mg/L CaCl2.2H2O, 89.6 mg/L

CaSO4.2H2O and 84.1 mg/L NaHCO3.3H2O (Ionic strength 3.33 x 10-3 M). Other tests

were run with water that had higher chloride to sulfate ratio, and this water contained 410

mg/L CaCl2.2H2O, 17.9 mg/L CaSO4.2H2O and 84.1 mg/L NaHCO3.3H2O (Ionic

strength 9.38 x 10-3 M). For simplicity, these two waters are henceforth referred to as

“normal chloride” and “higher chloride” water in later text.

Each of these waters was further modified with different types of disinfectant including:

1) Unmodified Synthesized water, 2) Synthesized water + Ammonia (1.37 mg/L as N) 3)

Synthesized water + Chlorine (5.13 mg/L as Cl2), 4) Synthesized water + Chloramines

(5.13 mg/L as Cl2 and 1.37 mg/L as N), 5) Synthesized water + Chloramines (5.13 mg/L

as Cl2 and 1.37 mg/L as N) + Phosphate (1 mg/L as P added as NaH2PO4).

The pH meter was calibrated using pH 7 and pH 10 buffer solutions. pH of the solution

collected was measured after 30 seconds while stirring the solution. Samples collected

for total lead and copper were preserved with 5% nitric acid and held for at least 28 hours

before analysis of lead, copper, and zinc using ICP-ES (Induced Coupled Plasma-

Emission Spectroscopy) per standard method 3120 B (APHA, 1998). QA/QC revealed

this approach recovered 100% of all types of lead solids formed in this study.

76

RESULTS AND DISCUSSION:

Experimental results examined the effects of chloride: sulfate ratio, effect of galvanic

current on lead leaching, change in pH during stagnation, and effects of gap distance

between cathode and anode. The role of water chemistry in effects observed was

emphasized throughout.

All the experiments for brass were performed in two phases. During the conditioning

phase water was changed every 8 hours for 3 weeks, and sampling for lead contamination

was conducted thereafter. The brass and copper were connected with the grounding wire

throughout this exposure. Initial testing was conducted in water with low chloride and

high sulfate, with sampling at 3 weeks, followed by conditioning at elevated chloride to

sulfate ratio with sampling at 6 weeks using the same brass and copper samples. The

lead pipes were conditioned for approximately 2 months using synthesized water with

low chloride without any copper present. Thereafter, testing was initiated when the

conditioned lead was galvanically connected to new copper pipe.

Effect of Chloride to Sulfate Ratio

Higher Cl-:SO4-2 ratio caused increased lead leaching from both yellow brass and red

brass in 11 of 12 waters tested (Figure 3-5). In most cases, the increase in lead leaching

was about 240%, although in waters with phosphate and chloramines the water with

higher chlorides leached an average of 10X more lead. In most cases, the trend was

significant at a confidence level greater than 90% (error bars plotted, Figure 3-5). The

adverse effect of higher chloride to sulfate ratio on lead leaching is probably much more

significant than these data indicate, given that with time the galvanic couple can be

expected to passivate and lead leaching from brass would naturally decrease (Reiber,

1991; Lytle and Schock, 1996). In other words, the adverse effects of higher Cl-:SO4-2

ratio overwhelmed benefits of the extra exposure time. The exceptional sample was a

yellow brass device exposed to synthetic water spiked with chlorine, for which the switch

to the water with higher chloride tended to decrease lead leaching.

77

Higher galvanic current values were present in samples at elevated chloride to sulfate

ratio (Figure 3-6). The average increase in galvanic current in water with higher chloride

was 140 percent (Figure 3-6), which is of the same magnitude as the average increase in

lead leaching (Figure 3-5). All experiments with brass in later sections, which are a

continuation of tests discussed in this section, were conducted using the water with a

higher ratio of Cl-:SO4-2.

Effect of Galvanic Current on Lead Leaching:

By measuring lead leaching from the apparatus with and without the external grounding

wire attached (Figure 3-4), the effects of galvanic currents between brass and copper on

lead leaching could be discerned directly. Three sequential sets of measurements were

made from each rig after 8 hours of stagnation, and representative results for yellow brass

are presented.

There is clearly a large enhancement to lead leaching when previously conditioned pure

lead pipes are first galvanically connected to new copper pipes. In every case tested, lead

leaching was enhanced (upper Figure 3-7). The lead leached from pure lead pipes ranged

from 3-11 mg/L with the galvanic connection versus only 0.45 to 1.5 mg/L in pipes

separated 2 mm apart by a plastic ring. Leaching from pure lead was increased by 2 to 20

times by the galvanic current and the difference was significant at greater than 90%

confidence in every case (error bars, Figure 3-7). The average amount of lead leached

from the brass devices was higher for galvanically connected rigs, however, the typical

effect was “only” to double lead leaching and results were significant at greater than 90%

confidence in 3 of 6 cases (Figure 3-7).

Interestingly, the concentration of copper leaching to the water was not significantly

impacted by galvanic connections in any of the rigs tested (results not shown). The

change brought about in the copper Ecorr with and without the galvanic connection to the

pure lead pipe was -4.2 mV to +6.3 mV. This is essentially within the detection limit of

78

Ecorr measurements. In contrast, Ecorr of the lead increased by 11 to 110 mV when

connected to copper, with an average increase of 39 mV. This is expected given that the

ratio of copper surface area to lead surface area was 19:1. Conceptually, the same

galvanic corrosion current is concentrated on the smaller lead surface but diffused over

the larger copper surface; thus, galvanic current density on the anode is quite high. Thus

galvanic current can have a greater impact on the smaller anodic surface while leaving

the larger cathode relatively unaffected.

pH changes in the pipe rig:

Two different sets of experiments were performed to study pH changes in the pipe rigs

including 1) pH of water contacting a separated anode and cathode, and 2) overall change

of pH in the system. The pH change within the pure lead pipe is representative.

To facilitate the collection of clearly separate samples from the anode and cathode, the

ball valve rig was utilized (Figure 3-4). An exterior conductive wire was used to

galvanically connect the pipes. The initial pH of water was set to 8.6 ± 0.02. After 8

hours of stagnation the ball valve was closed, and samples were then collected from the

lead and copper sides of the rig (Figure 3-8). In the typical case, pH was higher than 8.6

in water from the cathode (i.e. copper) side and lower than 8.6 in water from the anode

(i.e. lead) side. This is expected based on the representative reactions in Figure 3-1.

This behavior was followed for all waters except for Potomac water spiked with

chloramines where pH on the anode (lead pipe) side was not lowered to less than 8.6;

however, pH was still lower than on the cathode side. In general, the pH reduction on the

anode side was relatively small, and was near the detection limit, except in the

synthesized water without amendments where pH dropped 1.0 units during the 8 hour

stagnation. As noted earlier, any decrease in pH is expected to adversely impact lead

leaching to water (Figure 3-3). It is likely that this lowered pH was at least partly

responsible for increased lead leaching when the galvanic connection was in place in

Figure 3-7.

79

There are several practically important effects worth noting in the context of this

phenomena. First, it is instructive to note that the pH differential effect at the anode and

cathode surfaces would be nearly completely masked if a single sample was collected,

and contents of each section were mixed (Figure 3-9). In such cases normal

measurement of pH provides an indication of the overall corrosion reaction, consistent

expectations in upper Figure 3-1 for corrosion in the absence of dissimilar metal.

Secondly, the pH reduction that did occur at the anode in Figure 3-8 is the average

reduction for water within the anode compartment. pH near at the surface of the lead

bearing material would be lower still due to diffusion limitations to mixing. Finally, the

artificial gap of two inches, necessary to separate the anode and cathode compartment,

tends to reduce the galvanic current by virtue of the solution resistance (effect discussed

in the next section). The net result is that actual pH drops near lead surfaces in practice

are probably much greater than are indicated in Figure 3-8.

Effect of pH:

To examine the direct effect of pH on lead leaching, an experiment was conducted at pH

7.5 ± 0.02 at elevated chloride for comparison to results at pH 8.5 in the same water for

pure lead galvanically connected to copper. The lead leached to the pH 7.5 water was

significantly higher than at pH 8.5 water at a confidence level of 90% for 5 of 6

conditions (results not shown). This is consistent with increased lead solubility at lower

pH values. Somewhat surprisingly, water at pH 7.5 dosed with chloramines and

phosphate consistently showed more than 45 mg/L lead in samples collected after 8 hours

of stagnation (90% confidence level error bar plotted, Figure 3-10. The reason why

phosphate and chloramine had such an adverse effect was not followed up on in this

study, although it is deemed worthy of future research.

Effect of gap between metals:

The effect of the artificial gap width on lead leaching has not been previously studied. In

general, it was anticipated that the larger the gap, the lower the galvanic current since

solution resistance is expected to be significant. Lead was quantified using a direct

connection to copper (Figure 3-4 c) versus a 2 inch ball valve in-between (Figure 3-4 a).

After samples were analyzed separately with the ball valve closed from brass and copper

80

end, virtually all of the lead was in the water from the brass side of the ball valve

(>99.99% Confidence interval, results not shown). For purposes of direct comparisons

the plotted lead concentrations from the ball valve apparatus are those measured after

mixing with water from the copper tube. Directly connecting the anode and cathode

increased the lead leaching by a factor of 6 to 20 times relative to that obtained for in the

case of a 2 inch gap (90% confidence level error bars plotted, Figure 3-11).

The galvanic current could not be measured with a direct connection between the brass

and copper, but it was possible to compare the galvanic current from an experiment with

a 2 mm gap to that for the experiment with the 50mm (2”) gap from the ball valve.

Galvanic current increased 3 to 17 times higher for the smaller gap (Figure 3-12). The

increase in lead leaching with a smaller gap was correlated to the increase in galvanic

current. (Figure 3-13).

Passive Anodic Protection:

Since iron or galvanized service lines are sometimes present in a water system, and they

are connected to copper and lead, it was deemed worthwhile to study the effect of

possible anodic protection from the iron. Steel wool was connected to the copper tube as

per the illustration of Figure 3-4. The steel was a distance of 6 feet away from the lead

and 2 inches away from the copper pipe (Figure 3-4). The connection to iron lowered the

average lead leaching, albeit not at 90% confidence (error bars plotted, Figure 3-14). As

may be expected, Ecorr values for the lead pipe were lowered by 28 mV when the steel

wool was galvanically connected to the lead (results not shown). Clearly, a remotely

located iron pipe could impact leaching and corrosion of lead plumbing materials.

In recent years, water utilities have started installing a dielectric between iron and copper

service lines to eliminate galvanic corrosion problems. At first glance, this would tend to

decrease corrosion of the utility owned iron mains, and increase corrosion of the lead and

copper services. However, in many jurisdictions, a wire must be installed around the

dielectric to meet electrical code for grounding. The net benefit of installing a dielectric

that is subsequently shorted with a grounding wire is difficult to understand. The sole

81

advantage we can see is provision of a small gap (0.5 to 2 inches) between the dissimilar

metals, which could reduce internal galvanic corrosion due to solution resistance as per

Figure 3-11 and Figure 3-12. However, the dielectric itself is usually made of a

dissimilar metal. In any case, plumbers and water utility personnel general seem unaware

that installation of the grounding wire to meet electrical code circumvents the intent of

installing the dielectric. Future research should confirm the purported benefits (or lack

thereof) of this change in code.

Deposition corrosion:

An experiment was performed to conceptually prove that dissolved cupric ion can

increase lead corrosion and leaching to water. For the test, granular lead (5 gram) was

added to a beaker along with the synthesized water at pH 8.5. Cupric sulfate was then

dosed to the water and pH was maintained at the target value. The concentration of lead

increased markedly in the presence of cupric ions (Figure 3-15). The trend was

consistent regardless of the type of water the lead granules were exposed to.

Clearly, deposition corrosion can adversely impact lead leaching. If a copper pipe is

installed downstream of lead pipe during a partial lead service line replacement, direct

galvanic corrosion between the lead and copper would occur. Deposition corrosion from

the cupric ions in the water could also be significant. In general, the net effect of both

factors together cannot be predicted. At one extreme, as copper deposits on the lead

surface, its potential rises, reducing the galvanic current between the separated anode and

cathode. On the other hand the entire surface of the lead could be eventually plated with

copper in which case protection is possible. While neither extreme of behavior is

expected in practice, it is useful as a hypothetical construct, and illustrates just how

complicated such effects may be. It is also possible that galvanic corrosion between the

dissimilar pipe materials dominates completely in the short term, and then eventually

deposition corrosion effects become more important in the long term. Research is

urgently needed to study these effects, since the outcome is likely to depend on the water,

and it will control the effectiveness of the partial lead service line replacement program.

82

One week after the copper and lead pipe were galvanically connected, a large current was

measured. In comparison, 3 weeks later the galvanic current had dropped by a factor of

1.3 to 6 times. (Figure 3-16). Since these lead pipes had been conditioned in water for

over 2 months devoid of any copper, it is possible that during three weeks of subsequent

testing copper deposited on the lead surface.

Effect of oxidants:

This research was initially instigated to explain adverse consequences of chloramines on

lead leaching and lead corrosion in practice (i.e. Edwards et. al., 2004). For the most

part the experiments did not show a systematic detriment from chloramine versus free

chlorine. At the end of the study, some testing was conducted to examine the effect of

prolonged stagnation on galvanic currents. Interestingly, in all systems without

chloramine or ammonia, galvanic currents dropped below 1 µA within 8 hours using

brass devices (Figure 3-17). The drop in corrosion current was especially pronounced for

free chlorine, which had the highest galvanic current at the start of stagnation but the

lowest galvanic current after 139 hours. In contrast, in systems with chloramines,

galvanic currents actually increased during the first 24 hours. The net result is that after

24 hours stagnation, galvanic currents were about 2.6 times higher in the presence of

chloramine versus free chlorine, and after 100 hours the currents were about 6.3 times

higher. The implication is that if a stagnation time greater than 8 hours had been used in

this study, a dramatically different outcome between systems with chloramine versus free

chlorine may have been obtained. That remains to be proven in future research, but it is

clear there is a (Figure 3-17) dramatic difference in galvanic behavior in system with

chloramine versus chlorine at longer stagnation time.

Galvanic relationship between current and lead leached.

For every pair of electrons removed from lead or brass by galvanic current, one lead

molecule can be corroded and potentially released to water. The relationship between

current and lead concentrations was examined. Current was measured at the start of the

experiment, and assumed to stay constant during the 8 hours stagnation period, allowing

lead leached from galvanic action to be predicted using the following equation.

83

Maximum Lead Leaching (g) =

Maximum lead leaching versus actual lead concentration was plotted for brass devices

and lead pipes. Hypothetically, lead leaching is directly proportional to the current

flowing between the metals. When results were compared across the wide range of

experiments performed regardless of initial pH and gaps between the metals, actual lead

leaching was equal to or less than predicted by the above equation for brass devices

(Figure 3-18. On the other hand actual lead leached was greater than the predicted lead

for pure lead pipes (Figure 3-18). This is possible since galvanic corrosion supplements

corrosion occurring on the lead surface (Figure 3-1).

Other effects:

To understand the effect of flow on galvanic corrosion, water was pumped through the

rig via a reservoir. The current was demonstrably higher in flowing conditions compared

to stagnant conditions for 5 out of 6 water types studied. The difference exhibited was

significantly higher at 90 % confidence level.

Some unusual alloying effects were also noted. Brass faucets were sometimes cathodic

when galvanically connected to copper pipes. Measurements were made for current and

lead concentrations as per 8 hours of stagnation schedule after a short conditioning phase.

The brass remained cathodic consistently and lead leaching was actually reduced by the

galvanic connection to copper in this arrangement with greater than 90% confidence. It

is striking because of 7 of 8 brass devices of this type were highly anodic to copper, and

only one was cathodic. This is obviously highly desirable in the context of lead leaching,

since the copper material is sacrificed to protect the lead.

⎟⎠⎞

⎜⎝⎛×⎟

⎠⎞

⎜⎝⎛××⎟

⎠⎞

⎜⎝⎛×

⎟⎠⎞

⎜⎝⎛××⎟

⎠⎞

⎜⎝⎛

Pbe2

Pb. of molePb 10023.6 101.6

Pb of molePb of g 207.2 (Sec) T

sec I

2319-

eCoulomb

Coulomb

84

CONCLUSION:

Galvanic connections between copper and lead materials can dramatically worsen lead

leaching under a wide range of circumstances. The effects are exacerbated by higher

chloride to sulfate ratio, reduced gap distance, and may be worsened when new copper

pipe is connected to old lead. The presence of chloramines may produce sustained

galvanic action for a prolonged time period. The effects of deposition corrosion are

important over the short term, but need study over the months and years of exposures

typical in water distribution systems.

85

ACKNOWLEDGEMENTS:

The authors acknowledge the financial support of the National Science Foundation under

grant DMI-0329474. Opinions and findings expressed herein are those of the authors and

do not necessarily reflect the views of the National Science Foundation.

REFERENCES:

APHA. (1998). Standard Methods for the Examination of Water and Wastewater. 20th

Edition, United Book Press, Inc; Baltimore, Maryland.

Britton, A. & Richards, W. N. (1981). "Factors Influencing Plumbosolvency in

Scotland." Journal of the Institute of Water Engineers and Scientists 35(5): 349-

364.

Cifuentes, L. (2001). The Corrosion Behavior of Lead Anodes in CuSO4-H2SO4

Electrolytes. Presented at the 2001 Joint International Meeting - the 200th

Meeting of The Electrochemical Society, Inc. and the 52nd Annual Meeting of

the International Society of Electrochemistry - San Francisco, California.

Cruse, H. (1971). Dissolved-Copper Effect on Iron Pipe. Journal American Water

Works Association, 63:2:79.

Dodrill, D. & Edwards, M. (1995). Corrosion Control on the Basis of Utility

Experience. Journal American Water Works Association. V. 87, No. 7, 74-85.

Edwards, M.; Rushing, J.C.; Kvech S. & Reiber S. (2004). Assessing Copper Pinhole

Leaks in Residential Plumbing. In “Scaling and Corrosion In Water and

Wastewater Systems.” Edited by Simon Parsons, Richard Stuetz, Bruce Jefferson

and Marc Edwards. Water Science and Technology p. 83-90 V. 49, N. 2.

Edwards, M. (2003). Degradation of Drinking Water Treatment Plant Infrastructure by

Enhanced Coagulation. Presented October, 2003 at the AWWA Distribution

System Conference. Portland, OR.

Edwards, M. (July 2002). Initiation and Propagation of Localized Corrosion in Water

Distribution Systems. Invited Lecture at the 2002 Gordon Conference on

Aqueous Corrosion. Colby-Sawyer College.

86

Edwards, M.; Jacobs S. & Dodrill D. (1999). Desktop guidance for mitigating Pb and

Cu Corrosion By-Products. Journal American Water Works Association. V. 91,

No. 5, 66-77.

Edwards, M. & Schecher, W. (1996). Manual for Corrode Software. pub AWWA.

Gregory, R. (1990). Galvanic Corrosion of Lead Solder in Copper Pipework. Journal of

the Institute of Water and Environment Management, 4:112.

Hatch, G.B. (1955). Control of Couples Developed in Water Systems. Corrosion,

11:1:15.

Jones D. A. (1992). Principles and prevention of corrosion. Printice hall, NJ.

Lin, N. H.; Torrents, A.; Davis, A.P.; Zeinali, M.; & Taylor, F.A. (1997). Lead

Corrosion Control from Lead, Copper-Lead Solder, and Brass Coupons in

Drinking Water Emplyoying Free and Combined Chlorine. Journal of

Environmental Science and Health, A32:4:865.

Lytle, D.A. & Schock. M.R. (1996). Stagnation Time, Composition, pH and

Orthophosphate Effects on Metal Leaching from Brass (EPA/600/R-96-103).

Marshall, B.J.; Rushing, J.C.; & Edwards M. (2003). Confirming the Role of Aluminum

Solids in Copper Pitting Corrosion. Proceedings of the American Water Works

Association National Conference in Anaheim, CA. June, 2003. T-7-2. 13 pages.

McNeill, L.S.; & Edwards M. (2001). Iron Pipe Corrosion in Distribution Systems.

Journal American Water Works Association. V. 93, No. 7, 88-100.

NACE (1984). Corrosion Basics. Edited by Delinder V.L.S. NACE Texas Huston.

Oliphant, R.J. (1983). Summary Report on the Contamination of Potable Water by Lead

from Soldered Joints. Water Research Center, ER 125-E.

Reiber, S. (July 1989). Copper plumbing surfaces: An electrochemical study. Jour.

AWWA.

Reiber, S. (1991). Galvanic stimulation of corrosion on lead-tin solder sweated joints.

Journal American Water Works Association V. 83 N. 7, p. 83-91.

Portland. (1983). Internal Corrosion Mitigation Study Addendum Report. Bureau of

Water Works. Portland, Oregon.

87

Rushing, J.C.; & Edwards M. (2002). Role of Chlorine and Aluminum in Pitting

Corrosion of Copper Pipes. Proceedings of the 2002 Water Quality Technology

Conference in Seattle Washington. November, 2002. W-5 15 pages.

Schock, M.R. (July 1989). Understanding Corrosion Control strategies for Lead.

Journal American Water Works Association 81:7:88.

Shreir L.L.; Jarman R.A.; & Burstein G.T. (1995). Corrosion control. Butterworth-

Heinemann Ltd.

88

Figure 3-1: Conceptualization of galvanic effects on lead leaching to water. For a pure lead pipe, leaching is properly viewed as dissolution, with anodic and cathodic reactions occurring in close proximity over the pipe surface (top). Galvanic corrosion driven by

coupled lead and copper separates anodic and cathodic reactions, yielding lower pH near the lead surface (middle). Deposition corrosion via soluble cupric ions can directly

corrode lead (lower left), and can create micro-cells in which metallic copper catalyzes oxygen reduction reactions and lowers pH near the lead surface (lower right).

89

Figure 3-2: Final pH as a function of the concentration of lead corroded at the anode. Temp 25oC, I= 0.001, Solid phases considered include hydrocerussite, cerussite and

Pb(OH)2.

90

Figure 3-3: Soluble lead as a function of total lead corrosion. At low levels of lead corrosion in the model system, a greater fraction of the lead is soluble at higher alkalinity

(above). In contrast, at higher levels of lead corrosion, more lead is soluble at lower alkalinity due to greater reduction of pH at the anode. Temp 25oC, I =0.001, Solid phase

hydrocerussite considered include hydrocerussite, cerussite and lead hydroxide.

91

Figure 3-4: Experimental set up.

92

Figure 3-5: Effect of higher chloride to sulfate ratio (elevated chloride) on lead leaching

Yellow brass devices

0

0.1

0.2

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

Lead

(ppm

)

Normal Chloride

Elevated Chloride

pH 8.52 Inches apart

Red brass (NSF certified) devices

0

0.025

0.05

0.075

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

Lead

(ppm

)

Normal Chloride

Elevated ChloridepH 8.52 Inches apart

93

Figure 3-6 : Effect of higher chloride to sulfate ratio (elevated chloride) on current in galvanically active rigs.

Yellow brass devices

0

1

2

3

4

5

6

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

Cur

rent

(Mic

ro a

mp)

Normal Chloride

Elevated ChloridepH 8.52 Inches apart

94

Figure 3-7 :Effect of galvanic connection on lead leaching

Lead leaching from Lead pipes

0

2

4

6

8

10

12

14

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

Lead

con

cent

ratio

n (p

pm)

Galvanically UnconnectedGalvanically ConnectedpH 8.5

Elevated chlorideHead to Head connection

Lead leaching from brass devices

0

0.02

0.04

0.06

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es +

Pho

spho

rous

Lead

con

cent

ratio

n (p

pm)

Galvanically UnconnectedGalvanically Connected

pH 8.5Elevated ChlorideHead to Head connection

95

Figure 3-8: pH measured from the cathode and anode side in a galvanically active rig separated 2 inches apart.

Lead pipes

7.2

7.6

8

8.4

8.8

9.2

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

pH

Initial pHpH on Cu sidepH on Pb side

Elevated Chloride 2 Inches apart

96

Figure 3-9: Overall change in pH from the water collected from rigs with lead pipes placed head to head with copper pipes.

Lead Pipes

7

7.5

8

8.5

9S

ynth

etic

Wat

er

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

pH

7.5 Unconnected 7.5 Connected8.5 Unconnected 8.5 ConnectedElevated chloride

Head to Head connection

97

Lead leaching from lead pipes for water with chloramine and phosphorous

0

10

20

30

40

50

60

pH 7.5 pH 8.5

Ave

rage

Lea

d C

once

ntra

tion

mg/

L

pH 8.5Elevated chlorideHead to Head connection

Figure 3-10: Effect of water pH on lead leaching from lead pipes.

98

Figure 3-11: Effect of installing 2 inch wide ball valve on lead concentrations between copper and yellow brass devices (Not NSF certified).

Yellow Brass Devices

0

0.4

0.8

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

Lead

(ppm

)

At distance 2 inch apartTouching head to head

pH 8.5Elevated ChlorideGalvanically connected

99

Figure 3-12: Effect of installing 2 inch wide ball valve on current values between copper and yellow brass devices (Not NSF certified).

Yellow brass Devices

0

5

10

15

20

25

30

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

Cur

rent

(Mic

ro A

mpe

re)

2 inches apart2 mm apartpH 8.5

Elevated Chloride

100

Figure 3-13: Relative increase in galvanic current (right axis) and increase in lead leaching (left axis) when gap was decreased. A rise in galvanic current increased the lead

leaching.

0

2

4

6

8

10

12

14

16

18

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amm

ine

+ P

hosp

horo

us

Lead

rele

ase

2mm

/Lea

d re

leas

e 50

mm

0

5

10

15

20

25

Gal

vani

c C

urre

nt 2m

m / G

alva

nic

curre

nt 50

mm

.

Increase in current Increase in lead concentration

pH 8.5Elevated Chloride

101

Figure 3-14: Effect of iron on lead leaching from copper – lead pipe arrangement kept head to head

0

4

8

12

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

Lead

(ppm

)

Pb-CuPb-Cu-Fe

pH 8.5Elevated ChlorideHead to Head connection

102

Figure 3-15 : Effect of copper ions on lead concentration in experiment performed in lab ware with lead granules in water.

Lead in 10 mg/L Chloramine Water

0

0.5

1

1.5

2

2.5

3

0 10 20 30 40 50 60Time (hour)

Tota

l Lea

d (p

pm)

Without CopperWith Copper

pH 8.5Elevated Chloride

Lead in 10 mg/L Chlorine Water

0

0.5

1

1.5

2

2.5

3

3.5

0 10 20 30 40 50 60Time (hour)

Tota

l Lea

d (p

pm)

Without CopperWith Copper

pH 8.5Elevated Chloride

103

Figure 3-16: Effect of aging on pure lead pipes corrosion in the presence of copper.

Pure lead pipes

0

50

100

150

200

250

Syn

thet

ic W

ater

Syn

thet

ic W

ater

+ N

itrat

e

Syn

thet

ic W

ater

+ A

mm

onia

Syn

thet

ic W

ater

+ C

hlor

ine

Syn

thet

ic W

ater

+ C

hlor

amin

es

Syn

thet

ic W

ater

+ C

hlor

amin

es+

Pho

spho

rous

Cur

rent

(mic

ro a

mp)

After 3 weeksAfter 1 week

pH 8.5Elevated chlorideHead to Head connection

104

Figure 3-17: Prolonged effect of chloramines on current

NSF certified devices

0

1

2

3

4

5

0 20 40 60 80 100 120 140Time (hours)

Cur

rent

(mic

ro a

mp)

SWSW + NitrateSW + AmmoniaSW + ChlorineSW + ChloraminesSW + Chloramines + Phosphorous

pH 8.52 inches apartElevated Chloride

105

Lead Pipes

0

2

4

6

8

10

12

14

16

18

20

0 2 4 6 8 10 12 14

Maximum lead leaching (mg/L)

Act

ual C

once

ntra

tion

(mg/

L) .

Predicted vs Actual

1 : 1 line for Relationship

Figure 3-18: Demonstrating galvanic relationship between predicted (calculated using current values) vs. actual lead leaching for brass (top) and lead pipes (bottom)

Yellow Brass Devices

0

0.3

0.6

0 0.2 0.4 0.6 0.8Maximum lead leaching (mg/L)

Act

ual C

once

ntra

tion

(mg/

L) .

Predicted vs Actual

1 : 1 line for Relationship