Content uploaded by Michele Prevost
Author content
All content in this area was uploaded by Michele Prevost on Feb 08, 2016
Content may be subject to copyright.
DESHOMMES ET AL | 102:10 • JOURNAL AWWA | PEER-REVIEWED | OCTOBER 2010 91
Lead removal from tap water
using POU devices
Common point-of-use (POU) devices that met NSF/ANSI-53 certification standards for lead
removal before July 16, 2007, were evaluated for the reduction of lead and other trace
metals in drinking water. Systematic testing and evaluation of various POUs focused on
particulate and dissolved lead removal efficacy, under conditions different from those
addressed in the NSF-53 testing protocol (particle type, water tested, lead levels). Tap-
mounted and under-the-sink POUs showed, globally, adequate removal performance,
although the NSF threshold value for lead of 10 µg/L was occasionally slightly exceeded.
Pour-through POUs failed to remove particulate lead, decreased pH, did not reduce
turbidity, and were the least efficient for chlorine removal. Copper removal was effective
for any concentration tested, and silver was detected in effluent water. Nitrification did
not increase the dissolution of lead particles trapped in the filters.
Point-of-use (POU) devices can contribute to the improvement of water
quality and safety (USEPA, 2006a). A recent nationwide survey on the
household domestic water use patterns in the United States conducted
by the US Environmental Protection Agency (USEPA) showed an
increase in the use of water treatment devices between 1995 and 2002
(37% of the household respondents in 2002; USEPA, 2003). POU devices are
commercially available in various forms, including under-the-sink, tap-mounted,
and pour-through devices. The most widely used POU devices are the pour-
through (pitcher-style filter) and tap-mounted treatment systems. Under-the-sink
devices are more expensive and not easy to install (USEPA, 2006b). POU devices
are widely used to remove taste and odor compounds from drinking water. They
can also be used as a temporary remediation strategy to reduce lead (Pb) exposure
during the establishment of adequate corrosion control treatment, or during lead
service line replacement. In Canada, some provincial public health authorities
have issued advisory notices recommending that pregnant women and children
under six years old use bottled water or NSF/ANSI-certified filtration devices if a
lead service line is in use or when total or partial replacement does not result in
lead levels below the Canadian drinking water threshold of 10 µg/L total lead
(WDGPH, 2007; MPHD, 2006). Tens of thousands of filters were also distrib-
uted to residents of Washington, D.C., to mitigate hazards from high lead levels
in drinking water (Edwards et al, 2009).
Of the NSF/ANSI certifications, standards NSF-53 and NSF-42 specifically
concern POU and point-of-entry (POE) devices. NSF-53 covers POU/POE devices
that reduce health-related contaminants, including lead in drinking water, whereas
NSF-42 covers POU/POE devices that remove aesthetic and nonhealth-related
contaminants (chlorine, particulate matter, turbidity). In 2007, in response to
ELISE DESHOMMES,
YAN ZHANG,
KARINE GENDRON,
SÉBASTIEN SAUVÉ,
MARC EDWARDS,
SHOKOUFEH NOUR,
AND MICHÈLE PRÉVOST
2010 © American Water Works Association
92 OCTOBER 2010 | JOURNAL AWWA • 102:10 | PEER-REVIEWED | DESHOMMES ET AL
reports of particulate lead occurrence in tap water (Trian-
tafyllidou et al, 2007; NSF International, 2006; McNeill
& Edwards, 2004), changes were made to the lead-testing
protocol for the NSF-53 certification of water treatment
units. The protocol before 2007 focused on soluble lead
and did not test for the removal of lead particles. In the
NSF-53 protocol, two types of water are used to verify
manufacturers’ claims. The first is low-alkalinity aggres-
sive water (pH 6.5, alkalinity 10–30 mg/L calcium carbon-
ate) and the second is more mineralized water (pH 8.5,
alkalinity 100 mg/L calcium carbonate). It is expected that
the demonstration of claims using these two waters will
be sufficient to predict performance in a wide range of
natural waters, as indicated by the capacity not to exceed
10 µg /L lead in the filtered water (up to 120% of the
service lifetime of the POU filter). The revised protocol
(2007) includes both particulate lead of different sizes and
soluble lead in water at pH 8.5, specifically: 0.15 mg/L of
total lead, consisting of at least 20%, 0.1–1.2-µm particu-
late lead, and 30±10% total particulate lead. The lead
colloids/particles used are generated from an insoluble lead
stock solution made from soluble lead nitrate [Pb(NO3)2;
NSF International, 2007]. Most of the pour-through POU
devices failed to meet the standards set out in the revised
protocol and subsequently lost their certification for total
lead removal (Renner, 2007).
POU devices are based on various treatment tech-
nologies, including ion exchange resin (IX), granular
activated carbon (GAC), solid block activated carbon
(SBAC), reverse osmosis (RO), and distillation (Molloy
et al, 2008; USEPA, 2006a). Some of these technologies
applied to POU filters have been identified by the USEPA
as small system compliance technologies for lead: cationic
exchange (CX) resins, and RO (USEPA, 2006a, 2006b).
CX resins are the IX resins commonly used for water
softening, but they can be formulated to preferentially
exchange metallic cations, including lead. They are, how-
ever, subject to fouling and channeling (Chen et al, 2006).
GAC is often found in POU filters mixed with IX resin.
GAC has high and nonuniform porosity and can present
a high surface area. It is particularly well adapted for the
removal of organic compounds, chlorine, and taste and
odor compounds. However, as with IX, water can easily
channel through the relatively large filter media. SBAC
is composed of tiny particles of activated carbon that are
fused together, resulting in a block of uniform pore size.
They present smaller pore sizes (0.5–1.0 µm) than GAC
and can intercept pathogenic protozoa (Giardia cysts and
Cryptosporidium oocysts) as well as inorganic particles.
They are effective in removing some organic compounds
and chlorine. Carbon can also be formulated to remove
lead (USEPA, 2006b). Carbon-based filters are prone to
bacterial regrowth because of accumulated organic mat-
ter; however, this does not appear to pose a human health
concern (USEPA, 2006a). To help reduce microbial
regrowth, silver (Ag) is often added as a bactericide in
the filter material (Molloy et al, 2008). Nitrification can
also occur in these filters, especially in chloraminated
systems, which decreases pH and alkalinity. This can
result in increasing lead release from lead-containing
plumbing materials downstream of the devices if such
devices are installed upstream of a faucet or drinking
water fountain (Zhang et al, 2008).
POU filtration devices have been extensively used for
the removal of arsenic, copper (Cu), fluoride, and nitrate
and have been shown to be generally effective, although
breakthrough in a POU filter can often occur earlier than
predicted (EQP, 2008; Abdo et al, 1999). In Missouri
schools, more than 90% of the POU devices installed in
drinking water fountains met the lead action level (AL;
USEPA, 2006a). In schools in Seattle, Wash., under-the-
sink filters installed in drinking water fountains were
highly effective at removing lead (effluent < 1 µg/L),
cadmium, iron, and turbidity, under normal use, intermit-
tent flow, or without filter preconditioning (Boyd et al,
2005). Influent lead concentrations were, in general,
relatively low (< 10 µg/L), but higher concentrations were
also tested at lab scale (up to 136 µg/L). The same kind
of POU devices showed good results in a school reme-
diation program; however, materials downstream of the
POU device could release lead (Boyd et al, 2008b). In
Australia, Gulson and colleagues (1997) studied lead
removal by pour-through POU filters for various water
qualities. Removal varied widely (25–95%) because of
differences in initial lead content in the feedwater and the
ions and lead species present. Low efficiency was attrib-
uted to low initial lead levels (< 4 µg/L). In lab-scale
testing, lead removals decreased from 80 to 60% during
the filter lifetime.
This study provides insight into the performance and
reliability of POU devices when applied to typical tap
water from a distribution system. POU units tested
included under-the-sink and tap-mounted devices, which
currently meet the revised NSF-53 protocol for lead
reduction. Water quality and influent lead levels could
affect the operation of POU devices and influence their
efficiency. There is also concern that the colloids formed
from an insoluble lead solution made from soluble
Pb(NO3)2 may not be representative of particulate lead
in tap water, and the particles trapped in the POU filter
may release lead after a period of stagnation. Validation
of certified POU devices appears justified because of
problems reported in the earlier literature and because of
legal concerns associated with advisories regarding rec-
ommended use of POU devices by health authorities to
protect vulnerable populations such as pregnant women,
infants, and young children. Also, pour-through POU
filters that were NSF-53-certified for lead removal before
2007 were tested to better understand their malfunction
and whether or not they could still be used in some cases.
Indeed, these devices are widely used and had been rec-
ommended by some health authorities to limit exposure
2010 © American Water Works Association
DESHOMMES ET AL | 102:10 • JOURNAL AWWA | PEER-REVIEWED | OCTOBER 2010 93
to lead at the tap before July 2007. The objectives of this
study were to: (1) compare common POU devices that
were NSF-53-certified before 2007 for dissolved and
particulate lead removal in water from a distribution
system, (2) evaluate the behavior of some POU devices
for the removal and release of other metals, (3) measure
the effect of POU devices on water quality, and (4) eval-
uate whether nitrification in the filter could increase lead
release from particles trapped in the filter.
MATERIALS AND METHODS
Four experiments were carried out (1, 2, 3, and 4).
Figure 1 shows the POU devices tested and their charac-
teristics. They are widely used brands typically available
in local stores (four brands, seven stores). Tables 1 and
2 summarize the various experimental conditions for
these POU devices. All labware was thoroughly washed
with a nonphosphate detergent, soaked in 10% nitric
acid (HNO3) for at least 24 h, rinsed six times in distilled
FIGURE 1 Schema and characteristics of each POU device tested
Removable lid
Filter: GAC and CX
Tap water reservoir
Product water reservoir
Cold tap water
Kitchen tap
Product water
Cold tap water
SBAC filter
Product water reservoir
Cold tap water
Kitchen tap
SBAC filter
Tubing connecting
filter to the tap
(removable)
Cold tap
water Product
water
Separate tap; mounted
close to the kitchen tap
SBAC filter
CX—cationic exchange, GAC—granular activated carbon, Pb—lead, POU—point of use, SBAC—solid block activated carbon
Under-the-sink device:
• Devices A and B (test 4)
• NSF-53 certified for Pb before and after 2007
• NSF-42 certified for particulate class 1 (0.5– < 1 µm)
• Treatment technology: SBAC (A: ~180 cm3, B: ~650 cm3)
• Capacities: 2,840 L (A) and 1,515 L (B)
• Recommended lifetime: approximately six to 12 months
• Recommended water temperature: 4–38°C
• Pb removal (claimed): > 99 %
• Filtration method: filtration under pressure
• Service flow: ~2 L/min
• Separate tap in chromate brass (A) or plastic (B)
Pour-through device:
• Devices X (tests 1 and 3), Y (test 2), G, and H (test 4)
• NSF-53-certified for Pb before 2007
• Device H: NSF-42-certified for particulate class 5 (30–< 50 µm)
• Treatment technology: GAC and CX resin (~200 cm3)
• Capacity: 150 L
• Recommended lifetime: approximately two months
• Recommended water temperature: 2–30°C
• Pb removal (claimed): > 95%
• Filtration method: gravity
Tap-mounted device, type 1:
• Devices C and D (test 4)
• NSF-53-certified for Pb before and after 2007
• NSF-42-certified for particulate class 1 (0.5– < 1 µm)
• Treatment technology: SBAC (~125 cm3)
• Capacities: 378 L (D) and 757 L (C)
• Recommended lifetime: approximately three to four months
• Recommended water temperature: 1–38°C
• Pb removal (claimed): > 98%
• Filtration method: filtration under pressure
• Service flow: 2.2 L/min
Tap-mounted device, type 2:
• Device F (test 4)
• NSF-53-certified for Pb before and after 2007
• NSF-42-certified for particulate class 1 (0.5– < 1 µm)
• Treatment technology: SBAC (~100 cm3)
• Capacity: 150 L
• Recommended lifetime: approximately two months
• Recommended water temperature: < 34°C
• Pb removal (claimed): > 99%
• Filtration method: filtration under pressure
• Service flow: 3.8 L/min
2010 © American Water Works Association
94 OCTOBER 2010 | JOURNAL AWWA • 102:10 | PEER-REVIEWED | DESHOMMES ET AL
water and distilled–deionized water, dried in clean air,
and stored in sealed plastic bags until use. Samples were
taken in polypropylene bottles and vials. Analyses were
conducted in different laboratories, except for tests 1 and
3. The four tests are described in the following sections.
Test 1: Pour-through POU devices, lab-scale study on the
reduction of Pb and other metals. Four identical pour-through
POU devices (devices X) were tested for total Pb, chro-
mium (Cr), Cu, cobalt (Co), nickel (Ni), and Ag before
and after filtration at time 0 and after 40, 80, 120, and
150 L (POU recommended lifetime) in duplicate and over
three days in the laboratory. The delay between each batch
of water put through the filter was as recommended by
NSF-53 (15–60 s). Time 0—referring to time 0 of the POU
device lifetime—does not correspond exactly to 0 L. In
fact, the manufacturer recommended not drinking the first
two batches of filtered water because of possible initial
carbon release from the filter (not included in the POU
device recommended lifetime). Municipal tap water 1
(Table 2) taken directly from two different laboratory
faucets was tested as-is and spiked with Pb, Cr, and Cu
(faucet 1: spiked/not spiked, faucet 2: spiked/not spiked).
The tap water used was the first morning flush after over-
night stagnation to maximize metal concentrations. Influ-
ent and effluent waters were sampled as in Gulson and
colleagues (1997), using the following procedure: (1) the
first morning flush was sampled in a 500-mL bottle, (2)
the sample was mixed, (3) two aliquots of 100-mL were
taken in two 100 mL bottles (tap water: influent), (4) the
remaining water was poured into the POU device reservoir,
and (5) two aliquots of 100 mL of product water (filtered
tap water, effluent) were taken in two 100-mL bottles (at
least one unit volume is recommended by NSF, i.e., ≥ 200
mL for the pour-through POU devices tested). Spiked tap
water was prepared with two concentrated solutions of
500 mL, the first with Cr only and the second with Pb and
Cu. The 500-mL concentrated solutions were prepared
using 0.389 g of chromium(III) chloride (CrCl3) × 6H2O,
0.020 g of lead(II) chloride (PbCl2), and 4.008 g of
copper(II) chloride (CuCl2) × 2H2O. Five millilitres of each
concentrated solution were diluted in 15 L of tap water.
Final spiked concentrations poured into the POU device
reservoir did not exceed maximum acceptable concentra-
tions (MAC) for each metal: 10 µg/L Pb, 1,000 µg/L Cu,
50 µg/L Cr (Canadian levels). Spiked waters were filtered
and sampled as for tap water. Samples were acidified to
pH < 2 with ultrapure HNO3, and stored at 4°C for a
minimum of 16 h before analysis according to method
200.8 (USEPA, 1994). Samples were analyzed for Pb, Ni,
Cu, Cr, Co, and Ag by an inductively coupled plasma mass
spectrophotometer (ICP/MS),1 and the quantitation limits
were: 0.3 µg/L Pb, 0.6 µg/L Ni, 0.6 µg/L Cu, 1 µg/L Cr,
0.1 µg/L Co, and 0.3 µg/L Ag (three standards and blanks
for 20 samples, five duplicates for 100 samples). Metals
other than lead were chosen for the following reasons: Cu
is regulated in the United States with an AL of 1.3 mg/L
under the Lead and Copper Rule (LCR), and for aesthetic
objectives at 1 mg/L in Canada and is released from copper
pipes and brass materials that are frequently found in
premise plumbing. Cr is regulated (US MCL of 0.1 mg/L,
Canadian MAC of 0.05 mg/L) and can be leached from
faucets and brass devices. Ni is a concern in Europe (it is
Certification
for Pb Water Quality Parameters Measured
Device; Number Before After Spiking in Temp
Test Tested, Study Type 2007 2007 Duration Feedwater pH Metals NH3-N Cl2 oC Turbidity
1 Pour-through (X), Yes Three days, up to Part A: dissolved Pb, Cr,
4, lab scale* 100% of capacity Pb, Cr, Cu Cu, Co,
(150 L total, 50 L/d) Part B: no addition Ni, Ag
2 Pour-through (Y), Yes 12 batches of 1.6 Particulate Pb Yes Pb Yes 4 and
3, laboratory L filtered over 23 (Pb–Sn solder and 20
nitrification† days (19.2 L) PbO2), NH3-N
3 Pour-through (X), Yes Nine weeks ( 150 L) Pb, Cu,
3, domestic use Cr, Co,
pattern* Ni, Ag
4 Pour-through Yes Yes Six days, up to 120% Dissolved Pb and Yes Pb Yes 10 Yes
(G, H), 2; tap- (except of POU device particulate Pb
mounted (C, D, F); for G capacities (PbO2)
under-the-sink and H)
(A, B), 2, all lab
scale*
Ag—silver, Cl2—free chlorine, Co—cobalt, Cr—chromium, Cu—copper, NH3-N—ammonia nitrogen, Ni—nickel, Pb—lead, PbO2—lead dioxide, POU—point of use,
Sn—tin, Temp—temperature
*Distribution system 1 water
†Distribution system 2 water
TABLE 1 Summary of testing conditions
NSF-53
2010 © American Water Works Association
DESHOMMES ET AL | 102:10 • JOURNAL AWWA | PEER-REVIEWED | OCTOBER 2010 95
a suspected carcinogen) and may be regulated in the near
future because it is released from taps and fittings (World
Health Organization [WHO] provisional guideline value
of 0.02 mg/L). Ag was analyzed because of its potential
release from filter material. Finally, Co was analyzed for
comparison with the removal of other metals (selectivity
of the IX resin).
Test 2: Pour-through POU devices, lab-scale study on
particulate Pb reduction and effect of nitrification. Three
identical pour-through POU devices (devices Y) were
challenged with tap water 2 (Table 2) containing added
lead particles. Treated water was monitored for total lead,
pH, and ammonia nitrogen at 4 and 20°C (duplicates).
Before filtering, the water was dechlorinated, 1 mg/L of
ammonia was added, and the pH was adjusted to 7.9. The
filters were inoculated with water from a GAC filter in
which nitrification had been established for more than a
year. Particulate lead (lead dioxide [PbO2] and Pb–tin [Sn]
solder, ~57 mg Pb) was added to the first batch of water
(day 0), and then 12 batches of 1.6 L of water with unde-
tectable lead (< 1 µg/L Pb) were filtered over 23 days. All
particles were passed through a medium-sized aerator
screen, as described in Triantafyllidou et al (2007). Filtra-
tion was conducted on days 2, 6, 7, 9, 13–16, and 20–23
(one batch per day). All the treated water was collected
and subjected to heated and strongly acidic digestions that
were proven (Triantafyllidou et al, 2007) to recover all
the lead during later analysis with ICP/MS (blank and
standard every five samples, quantitation limit: 0.03 µg/L,
detailed in Zhang et al, 2008). The lead concentrations
used, although high, are representative of the “worst-
case” samples collected by a water utility in Washington,
D.C., (48,000 µg/L) in 2003 (Edwards & Dudi, 2004).
The oxidation–reduction potential was not maintained,
so PbO2 oxides were inherently unstable, as in the case of
Washington, D.C., after a switch from chlorine to
chloramines (Lytle & Schock, 2005). Nitrification was
tracked with mass balances on nitrate, nitrite, and ammo-
nia, as detailed elsewhere (Zhang et al, 2008).
Test 3: Pour-through POU devices, domestic-use-pattern
study on the reduction of Pb and other metals. Three identi-
cal pour-through POU devices (devices X as in test 1)
were tested for Pb, Cr, Cu, Ni, and Ag in three homes,
during real-time household use. Water from the kitchen
tap (Table 2, distribution system 1) was tested before and
after filtration at time 0 POU device lifetime, and after
three, six, and nine weeks of normal domestic use in
duplicate (POU device recommended lifetime). Residents
were instructed to filter about 2 L/d if possible and to
note each batch in order to comply with the POU device
capacity (150 L) after nine weeks of use. The first morn-
ing flush water was filtered in order to maximize metal
levels. Sampling procedure, methods, and analysis were
the same as in test 1.
Test 4: Three types of POU devices, lab-scale study on
dissolved and particulate Pb reduction (Pbdiss, Pbpart). Three
types of POU devices were tested: two under-the-sink
devices (A and B), three tap-mounted devices (C, D, and
F), and two pour-through devices (G and H). POU devices
were fed with Pb-spiked municipal tap water 1 (Table 2),
tested for lead (particles and soluble) over six days and
up to 120% of their capacity, and left to stagnate over-
night (> 8 h). The Pb-spiked feedwater was prepared
daily by adding either 46 or 231 mg of PbO22 sieved with
a standard screen to a fraction > 0.45 µm directly into a
mixing tank (~2,000 L) plus a lead-concentrated solution
containing 156 mg of Pb(NO3)2.3 Target concentrations
were ~20 µg/L or ~100 µg/L of Pbparticulate and ~50 µg/L
of Pbdissolved. Water in the mixing tank was pumped to a
pilot-testing POU device (Figure 2) with a configuration
approximately similar to the test apparatus prescribed by
the NSF-53 protocol. In the presence of lead particles in
test water with such an apparatus, calculated particulate
removal (Pbparticulate effluent/Pbparticulate influent) is not
as precise as calculated dissolved lead removal because
of nonhomogeneous mixing in the water supply tank.
However, calculated particle removals give an idea of
POU device efficiency. Influent water was sampled for
Turbidity Alkalinity Ca2+ Mg2+ Hardness P
Test Utility pH ntu mg/L as CaCO3 mg/L mg/L mg/L as CaCO3 mg/L
1, 3, 4 1 7.5–8.2 0.09–1.63 78–86 29–33 7.5–8.5 111–121 0.005–0.012
(7.7)† (0.24) (82) (32) (8.0) (116) (0.008)
2 2 7.9‡ 0.03–0.08 32‡ 12‡ 5‡ 51‡ 0.5‡
NSF-water 1 6.5±0.25 < 1 10–30 10–30 < 0.5
NSF-water 2 8.5±0.25 100±10% 100±10%
CaCO3—calcium carbonate, Ca2+—calcium cation, Mg2+—magnesium cation, P—phosphates
*Data from annual reports on drinking water quality
†Numbers in parentheses indicate mean value
‡Data from measurements during test 2
TABLE 2 Distribution system water quality characteristics*
2010 © American Water Works Association
96 OCTOBER 2010 | JOURNAL AWWA • 102:10 | PEER-REVIEWED | DESHOMMES ET AL
Pbdissolved and Pbparticulate twice daily (the first 1 L in the
morning and 1 L at noon), and effluent water from the
POU devices was sampled at the same time. Effluent
water was analyzed from day 1 to day 4 for total lead
only, and the last two days for Pbdissolved and Pbparticulate.
When analysis for both Pbdissolved and Pbparticulate was
required, 50 mL of the mixed 1-L sample were collected
and filtered on a 0.45-µm filter4 using a 60-mL dispos-
able syringe5. Samples were acidified to pH < 2 with
ultrapure HNO3 and stored at 4°C for at least 16 h
before analysis, according to protocol 200.8 (USEPA,
1994). Lead was analyzed using an ICP/MS6 with a
quantitation limit of 0.06 µg/L (blank and standard every
29 samples, duplicates every 24 samples). The pH,7 tur-
bidity,8 temperature, and chlorine content9 were mea-
sured for each sample.
RESULTS
Total lead removal. Figure 3, part A, summarizes the
results for total lead removal for pour-through POU filters
in test 1 (pour-through X, lab-scale), test 2 (pour-through
Y, lab-scale, and initial addition of high Pbparticulate), and
test 3 (pour-through X, household). Overall, significant
removals of lead by pour-through POU devices were
observed, ranging from 68 to 99%. For the same device
(X) and water (from distribution system 1; tests 1 and 3,
Figure 3, part A), total lead levels in filtered water
decreased and remained well below 10 µg/L, the reference
level for POU NSF-53 certification for lead, regardless of
the usage pattern. Effluent lead levels were greater for
test 1 (maximum = 3.5 µg/L) than for test 3 (maximum
= 1.1 µg/L); however, lead concentrations were higher in
influent 1 (10–22 µg/L versus 1.0–7.4 µg/L for influent
3, Figure 3, part A). When challenged with an initial
batch of water containing a high mass of lead particles
(57 mg, i.e., 36,000 µg/L Pb), pour-through Y retained a
portion of the lead particles. However, the lead levels in
the filtered water were extremely high and persisted in
the filtered water of the 12 batches of water without
spiked lead (up to 10,470 µg/L; Figure 3, part A, test 2).
The cumulative lead release was estimated using the fol-
lowing equation
Pb
released, mg = i = 1…12 (volume of filtered wateri, L
× Pb concentration in filtered wateri, mg/L) (1)
About 19 mg of lead was released at 4°C (34% of
Pbt = 0), and 2–7 mg was released at 20°C (3–12% of
Pbt:0). The mass of lead trapped in the POU filter was
calculated (Pb at t = 0: cumulative Pb released or “influ-
ent Pb”), and, despite high lead levels in effluent water
of the 12 batches cited previously, the calculated removal
percentage was high; removal < 95% was observed in
only two samples.
A comparison of total lead removal by various types of
POU devices challenged with water containing dissolved
and particulate lead is shown in Figure 3, part B (test 4,
54–160 µg/L of Pbtotal; Pbtotal = Pbparticulate + Pbdissolved).
Tap-mounted POU device
Agitated tank for
influe nt water
(2,000-L volume
for one day)
Water
evacuation
Addition of
water and lead
every day
Tap for filling
pour-through POU
devices a nd
tap- mounte d device F
Under-the-sink POU device
FI
G
URE 2
C
onfiguration of the pilot test designed for test 4
POU—point of use
2010 © American Water Works Association
DESHOMMES ET AL | 102:10 • JOURNAL AWWA | PEER-REVIEWED | OCTOBER 2010 97
Median
10– 90 %
Minimum /m aximum
n = 1
(particulate
Pb at t = 0)
n = 36
Pb released
(23 days )
n = 20
n = 20
n = 12
n = 12
Total Pb— µg/ L
Total Pb— µg/ L
0.04
0.07
0.10
0.40
0.70
1.00
4.00
7.0 0
10.00
40.00
70.00
100.00
400.00
700.00
1,000.00
4,000.00
7,000.00
10,000.00
40,000.00 A B
–10
0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
Tap-mounted
POU
Removal:
80–99%*
n = 36
n = 12 Under-the-
sink POU
Removal:
82–99%
n = 24
Pour-
through P OU
Removal:
35–97%
n = 24
Test 1 Test 2 Test 3 Test 4
FIGURE 3 Total lead before and after filtration by pour-through POU devices X and Y (A) and POU devices A–D and F–H (B)
E—effluent, I—influent, n—number of samp les, Pb—lead, POU —point of us e, t—time
*Not including the outlier at 43 µg/L
Test parameters: test 1—lab scale, water spiked and not, pour-through device, 83–97% removal; test 2—lab scale, particulate
Pb at t = 0, pour-through device, 71–99% removal; test 3 —hous ehold, pour-through device, 68 –99 % removal; test 4—lab scale,
seven POUs comprising three types, dissolved and particulate Pb
Dotted line indicate s the NSF Int ernational limit for tota l Pb.
I E I E I E I E
Days 0–6: Effluent Tested for Total Pb
Influent levels: 54–160 µg/L Pbtotal, 16–54 µg/L Pbdissolved, 1.5–144 µg/L Pbparticulate (3–90% Pbtotal)
Type/Name of POU Device
Under-the-Sink (n = 24) Tap-mounted (n = 36) Pour-through (n = 24)
Effluent Pbtotal A B C D F G H
Pbtotal 10 µg/L 10–11 (3) (0) 10–12 (5) 10–11 (4) 43 (1) 12–63 (8) 13–38 (8)
Pbtotal < 10 µg/L 0.6–9.9 (9) 1.3–9.4 (12) 1.9–9.8 (7) 0.4–9.7 (8) 0.1–6.5 (11) 4.0–9.0 (4) 8.9–9.7 (4)
Days 5 and 6: Effluent Tested for Dissolved and Particulate Pb
Influent Levels: 60–121 µg Pbtotal/L, 40–54 µg Pbdissolved/L, 7.8–78 µg Pbparticulate/L (12–66 % Pbtotal)
Pbdissolved 10 µg/L (0) (0) 10–11 (2) 11 (1) (0) (0) (0)
Pbdissolved < 10 µg/L 3.3–9.6 (4) 2.9–8.7 (4) 4.9–5.0 (2) 5.5–9.8 (3) 3.2–6.2 (4) 2.7–6.5 (4) 5.0–6.8 (4)
Pbparticulate 10 µg/L (0) (0) (0) (0) 38 (1) 27–57 (2) 23–32 (2)
Pbparticulate < 10 µg/L 0.03–0.74 (4) 0.05–0.75 (4) 0.05–0.90 (4) 0.01–0.6 (4) BDL–0.3 (3) BDL–5.3 (2) 4.7–7.9 (2)
BDL—below detection limit, n—number, Pb—lead, POU—point of use
Values in parentheses inside the table indicate number of samples.
TABLE 3 Number and ranges of values below or above 10-µg/L Pb in the effluent water of POU devices in test 4
2010 © American Water Works Association
98 OCTOBER 2010 | JOURNAL AWWA • 102:10 | PEER-REVIEWED | DESHOMMES ET AL
Two under-the-sink POU devices (A, B), three tap-mounted
POU devices (C, D, F), and two pour-through POU devices
(G, H) were tested. Total lead concentrations in effluent
water below or above 10 µg/L and their ranges are shown
in Table 3. Variable and sometimes very low removals
(35–97%) were observed for the pour-through POU
devices tested, whereas the others showed global ade-
quate removal efficiency (80–99%). Total lead concentra-
tions from the under-the-sink POU devices showed lim-
ited variability (mean = 6.5 µg/L, median = 7.7 µg/L,
standard deviation [SD] = 3.26), and levels rarely
exceeded 10 µg/L (three values of 10–11 µg/L, number
[n] = 24). For tap-mounted POU devices, similar and
stable lead concentrations in the filtered water were
observed (mean = 7.5 µg/L, median = 6.1 µg/L, SD = 7.0),
and levels barely exceeded 10 µg/L, except for one outlier
at 43 µg/L (nine values of 10–12 µg/L, n = 36). Lead
concentrations in the filtered water from the two pour-
through POU filters were much higher and variable, with
16 of 24 values > 10 µg/L up to 63 µg/L Pb (mean = 19
µg/L, median = 15 µg/L, SD = 13). The minimum observed
for these devices was nearly half the target level of 10
µg/L (4 µg/L Pb). Overnight stagnation was not found to
increase lead levels or any filter breakthrough.
Dissolved and particulate lead removal. The removal
of dissolved and particulate lead by the devices in test
4 (seven POUs, three types) are shown in Figure 4 and
Table 3. Test 4 was conducted by adding dissolved
and particulate lead to the feedwater. Figure 4 shows
measurements taken on the last two days of the experi-
ment because the POU effluents were only analyzed for
dissolved and particulate lead on those days (total lead,
days 1–4). Dissolved lead removals were consistently
high for all the devices tested. However, dissolved lead
concentrations in three out of 12 effluent samples for
the tap-mounted POU devices were somewhat above 10
µg/L (10–11 µg/L, influent range = 40–54 µg/L; Figure
4, part A, Table 3). In terms of dissolved lead remov-
als, pour-through POU units performed slightly better
(effluent 6.8 µg/L Pb) than the under-the-sink POU
devices (effluent 9.6 µg/L). In terms of particulate
lead, the removal pattern was considerably different
(Figure 4, part B, Table 3). With an influent range of
7.8–78 µg/L Pbparticulate, particulate lead removal for
all under-the-sink and tap-mounted devices tested
was high and remarkably stable, with low concentra-
tions of particulate lead in the filtered water ( 0.9
µg/L) except for one outlier, at 38 µg/L. This outlier
0
10
20
30
40
50
60
Under-the-
sink POU
82–94% removal
Tap-mounted POU
80–92% removal
Pour-through POU
87–92 % removal
Under-the-sink
POU
98– 99 % removal
Tap-mounted POU
96–99% removal*
Pour-through POU
26– 99% removal
n = 4
n = 4
n = 4
n = 4
Median
10– 90 %
Minimum /m aximum
–10
0
10
20
30
40
50
60
70
80
90
n = 4
n = 4
n = 4
n = 4
n = 4
n = 4
n = 4
Dissolved Pb—µg/L
Particulate Pb—µg /L
I A B C D F G H I A B C D F G H
n = 4 n = 4 n = 4 n = 4 n = 4
A B
FIGURE 4 Dissolved lead (A) and particulate lead (B) before and after filtration by POU devices in test 4
I—influent, n—numb er of samples, Pb—lead, POU— point of use
*Not including the outlier at 38 µg/L
Test parameters: lab scale, seven POUs comprising three types
Dotted lines indicat e NSF International limit for total Pb.
Device Name Device Name
2010 © American Water Works Association
DESHOMMES ET AL | 102:10 • JOURNAL AWWA | PEER-REVIEWED | OCTOBER 2010 99
corresponded to the total lead outlier cited previously
(43 µg/L). Inconsistent and sometimes low removals
(26–99%) were observed for the pour-through POUs.
The two pour-through devices (G, H) were ineffective
at removing particulate lead, and elevated particulate
lead remained in the filtered water (up to 57 µg/L).
POU behavior for other metals (Cu, Cr, Co, Ni, and Ag).
Table 4 shows the results for other metals (tests 1 and
3). For different testing conditions and initial metal
levels, good and stable copper removal was observed:
all removals exceeded 88%, except for a single event
at 68% (t = 0 week, house number 1). Copper concen-
trations in effluent water were higher for Cu-spiked
influent water (43–130 µg/L) than for water that was
not Cu-spiked (4.1–51 µg/L). Other elements showed
variable results: removal ranges were 9–99% for Ni
and 0–100% for Cr and Co. Metal levels in effluent
water versus litres of filtered water (because test 3
examined household use, metals effluent levels were
studied versus time [0–9 weeks] rather than the num-
ber of filtered litres) were studied for each of the seven
identical POUs (X) sampled in tests 1 and 3 (1-not
spiked-lab, 2-not spiked-lab, 1-spiked-lab, 2-spiked-
lab, and 1-house, 2-house, 3-house). Metal levels in
effluent water between t = 0 and t = 150 L (or nine
weeks) increased for all the POU devices for Ni (maxi-
mum +1.3 µg/L) and Cr (maximum +19 µg/L, nil for
two POU devices). Such increases were not systematic
for Co (+0 µg/L), Cu (–47 to +85 µg/L), and Ag (–11 to
+10 µg/L). No clear tendency was noted for Cr, Co, Ag,
or Cu effluent concentrations versus filtered litres (or
time). Ni levels increased linearly with filtered volumes
within the POU filter lifetime for all the POU devices,
except for two house-tested POU devices (1-house,
2-house, nonlinear). The best relationship was obtained
for the 1-not spiked-lab POU device (Figure 5). Silver
release was detected in the effluent of these devices.
Maximum metal levels measured at the effluent were
130 µg/L Cu, 38 µg/L Cr, 1.5 µg/L Ni, 0.09 µg/L Co,
and 43 µg/L Ag (Table 4).
Effects of POU filters on water quality. Figure 6 shows the
effect of POU devices on water quality parameters (tests 2
and 4). The pH was relatively stable for most of the
devices, except for the pour-through POU filters; a decrease
of 1.2–2.0 pH units (based on the median values, Figure
6, part A) was observed for pour-through devices G, H,
and Y. The decrease was greatest for Y. Also, pH values as
low as 4.4 (test 4) and 5.2 (test 2) were observed in the
TABLE 4 Concentration and removal ranges of other metals measured in water before and after filtration
with pour-through POU devices
Test Condition/Element
Lab Scale, Spiked (n = 10)
Parameter Nickel Copper* Chromium* Cobalt Silver
Test 1, part A
Influent—µg/L 1.6–3.0 890–1,242 42–52 ND–0.11 ND–0.05
Effluent—µg/L 0.22–1.5 43–130 19–38 ND–0.06 23–34
Removal—% 37–86 88–95 21–53 0–85
Effluentt = 150 L–Effluentt = 0 L—µg/L 0.37–0.53 37–85 7.5–19 0.02–0.04 0.2–1.2
Lab Scale, Not Spiked (n = 10)
Test 1, part B
Influent—µg/L 1.9–3.5 223–433 0.06–2.2 ND–0.10 ND–0.02
Effluent—µg/L 0.13–1.4 11–27 ND–6.0 ND–0.06 11–43
Removal 49–94 92–96 0–100 35–100
Effluentt = 150 L–Effluentt = 0 L—µg/L 0.73–1.3 (–0.6)–(–2.0) 0.38–3.4 0.0 (–11)–4.4
Household, Not Spiked (n = 12)
Test 3
Influent—µg/L 1.1–24 81–732 0.17–4.7 0.05–0.16 ND–0.04
Effluent—µg/L 0.02–1.5 4.1–51 0.19–3.2 0.01–0.09 12–30
Removal 9–99 68–98 0–63 3–82
Effluentt = 9 weeks–Effluentt = 0—µg/L 0.46–1.1 (–47)–0.28 (–0.3)–3.0 0–0.1 3.8–10
n—number of samples, ND—not detected, POU—point of use
*Spiked metals
Test 1, part A—lab-scale testing with Cr and Cu spiking; test 1, part B—lab-scale testing without any spiking; test 3—household usage without any spiking
2010 © American Water Works Association
100 OCTOBER 2010 | JOURNAL AWWA • 102:10 | PEER-REVIEWED | DESHOMMES ET AL
effluent of pour-through POU filters (Figure 6, part A). As
expected with carbon filters, chlorine was almost totally
reduced by all the devices (Figure 6, part B). Some trace
chlorine remained for devices G and H (up to 0.62 mg/L
free chlorine [Cl2]), whereas the effluent levels for the
other POU filters remained < 0.05 mg/L Cl2. SBAC POUs
did not efficiently decrease turbidity, with median values
decreasing from 0.8 to 0.4–0.5 ntu (Figure 6, part C). An
extreme value of 6.5 ntu corresponded to a single break-
through of particulate lead: 38 µg/L (device F). For pour-
through POU devices, median values decreased slightly
(0.8–0.6 ntu), but the maximum values and the 90th per-
centile were close to the influent water values, especially
for device G. Ammonia measured in the effluent suggests
that nitrification can take place during 24 h of stagnation
(test 2; Figure 6, part D). The pH dropped in all instances
after filtration during test 2, and, as expected, water tem-
perature influenced the activity of the nitrifying bacteria
(because reductions of ammonia were lower at 4°C than
at 20°C). However, no impact of nitrification on lead
release was observed because Pb release (19 mg) was
higher at 4°C, i.e., when nitrification was less active.
DISCUSSION
Total Pb removal by pour-through POU devices: Effect of
influent Pb concentrations (Figure 3, part A). Removals for
tests 1, 2, and 3 (68–99%) were lower than those indicated
by the manufacturers (> 95%, Figure 1). In fact, removals
are affected by influent concentrations (Gulson et al, 1997),
which, in this study, were variable (1.0-36,000 µg Pb/L)
and different from those of NSF-challenged water (150 µg
Pb/L). The low removal efficiency of test 3 could be associ-
ated with low influent lead levels (< 10 µg/L); however, the
lead concentrations in the filtered water remained low and
were not considered to be hazardous ( 1.1 µg/L). Further-
more, effluent lead concentrations increased with influent
lead concentrations. Therefore, a high removal rate may
not be a reliable indicator of a low lead level in effluent
water if the influent lead concentration is not considered.
As shown in test 2, despite substantial lead removal (71–
99%), lead levels in the effluent water were extremely high
(maximum 10,470 µg/L).
Total lead removal by pour-through POU devices: Effect
of dissolved Pb in influent water (Figure 3; Figure 4, part A;
Table 3). The composition of the total lead influenced
the efficiency of pour-through POU devices. When chal-
lenged with total lead containing significant levels of
particulate lead in test 4, pour-through POU devices G
and H were ineffective because total lead in effluent
water exceeded the 10-µg/L NSF threshold in most of
the samples collected. In contrast, lead levels in the
effluent of pour-through device X tested in tests 1 and
3 were always 3.5 µg/L. This is surprising, because
POU X had the same brand name and the same media
filter as POU G, which had shown inconsistent removal.
The efficiency gap can be partly explained by the pre-
dominance of soluble lead in challenge water in tests 1
and 3. Indeed, the tap water used in these two tests was
taken from a distribution system (system 1) showing
moderate levels of particulate lead (random daytime
sampling in 45 homes with lead service lines: median =
0.39 µg/L, average = 1.1 µg/L; Deshommes et al, 2010).
Also, the spiked water in test 1 was prepared with
soluble lead. Thus, the majority of total lead in test
waters 1 and 3 was soluble, and the pour-through POU
filters appeared effective in this case. Furthermore, for
the devices evaluated in test 4, the dissolved lead con-
centrations measured in the effluent water over two
days never exceeded 6.8 µg/L (dissolved Pb in influent:
40–54 µg/L). Thus, pour-through POU devices seem
effective for dissolved lead removal from tap water.
Moreover, pour-through filters were composed of GAC
and CX resin, the latter being recognized as a small
system compliance technology for lead removal (USEPA,
2006a). The limited efficiency observed in the author’s
tests may be attributable to the nature of CX resins,
which are designed to remove ions and not particles
(Chen et al, 2006). These results support cancellation
of the NSF-53 certification for pour-through POU filters
for lead removal in 2007, following the addition of lead
particles to the test water (Renner, 2007).
Particulate lead removal by pour-through POU devices
(Figure 3; Figure 4, part B; Table 3). The pour-through
POU filters tested were not capable of reliably remov-
ing particulate lead. In test 2, a high concentration of
0 50 100 150
Filtered Water—L
Nickel in Effluent Water—µg/L
0.0
0.2
0.4
0.6
0.8
1.0
1.2
y = 0.0046x + 0.29
R2 = 0.98
FIGURE 5 Best relationship between nickel concentrations
in the effluent of pour-through POU devices
and the number of filtered litres
POU—point of use
2010 © American Water Works Association
DESHOMMES ET AL | 102:10 • JOURNAL AWWA | PEER-REVIEWED | OCTOBER 2010 101
4
5
6
7
8
9
10
pH
Under-
the-sink POU
Pour-
through P OU
Tap-mounted
POU
n = 12
n = 15 n = 15
n = 15
n = 15
n = 15
n = 45
n = 15
n = 16 n = 16
–0.1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Under-the-
sink POU
Tap-mounted
POU
Pour-through
POU
n = 12
Median
10– 90 %
Minimum /m aximum
A B
CD
–1.0
–0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
Turbidity—ntu
Under-the-
sink POU
Pour-through
POU
Tap-mounted
POU
n = 12
n = 12
n = 12
n = 12n = 12
n = 30
n = 15
n = 15
–0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
E 4°C E 20°C
I A B C D F G H I
I
A B C D F G H
IABCDFGH
n = 12 n = 12 n = 12 n = 12
n = 12
n = 12
n = 12
Chlorine—mg/L
NH3-N—mg/ L
n = 12 n = 12
n = 12
FIGURE 6 Water quality parameters before and after POU filtration for pH in tests 2 and 4 (A), chlorine in test 4 (B),
turbidity in test 4 (C), and nitrification in test 2 (D)
E—effluent, I—influent, n—number of samples, NH3-N—ammonia-nitrogen, POU —point of use
Device Name
Device Name
Device Name
Test 2 Test 4
I E
2010 © American Water Works Association
102 OCTOBER 2010 | JOURNAL AWWA • 102:10 | PEER-REVIEWED | DESHOMMES ET AL
particulate lead added to the first batch of influent
water for pour-through device Y resulted in high lead
levels continuously released into the filtered water. No
lead particles were added after t = 0, indicating that
the particulate lead added initially was eventually
released sporadically. Extremely high particulate lead
concentration, as used in the water for test 2 (t = 0:
36,000 µg/L) considerably
exceeded the lead levels in
NSF test challenge water
(150 µg/L). Pour-through
device Y had been certi-
fied for the removal of
about 150 µg/L of soluble
lead but not for such
extreme concentrations of
particulate lead. However, high lead levels can occur,
as demonstrated in Washington, D.C. (Edwards &
Dudi, 2004), when lead–tin particles from solder are
released into the water (Triantafyllidou et al, 2007),
or following partial lead pipe replacement (Sandvig et
al, 2008; Boyd et al, 2004). Numerous samples con-
taining more than 10,000 µg Pb/L were collected after
partial lead service line replacements in Washington,
D.C. (Edwards et al, 2009), and, to counter this risk,
consumers were advised about potential elevated lead
levels in tap water under these circumstances (USEPA,
2000). More recently, the Centers for Disease Control
and Prevention has determined that children living in
homes with partial lead service line replacements have a
400% higher likelihood of elevated blood lead (CDC,
2010), and, in response, the use of filters to mitigate lead
spikes is under consideration. In view of the results of
test 2, pour-through POU devices—even if certified for
lead before 2007—might no longer be considered in the
effort to reduce lead exposure at the tap because par-
ticulate lead can potentially be released into the filtered
water. The breakthrough of particulate lead and the
resulting high total lead in filtered water also occurred
with lower influent levels of particulate lead in test 4.
Particulate lead influent levels in test 4 were closer to
those of the NSF-revised protocol but still different:
particulate lead (PbO2 ≥ 0.45 µm) varied between 1.5
and 144 µg/L and represented 3–90% of total lead
(median = 37 µg/L, 42%), whereas concentrations used
in the revised NSF protocol are 150 µg/L Pbtotal, includ-
ing at least 20% of fine lead particles (0.1–1.2 µm) and
20–40% of total Pbparticulate. As stated previously, these
results support cancellation of the NSF-53 standard for
lead in 2007. Pour-through POU devices that were NSF-
53-certified for lead before 2007 should not be used for
water containing particulate lead.
The inconsistent removal of particulate lead by
pour-through POU devices can be partly explained by
the composition of their type of filter. For example,
pour-through POU filters made of GAC and CX resin
have a tendency to develop channels (USEPA, 2006b)
that provide a pathway for particles to pass through
and contaminate the water, as was observed in the tests
conducted in this study. Moreover, PbO2 particles, as
used in tests 2 and 4, are fragile and could break into
smaller particles. Particles smaller than 30 µm are likely
to pass directly into the filtered water, which is why
NSF-42 certified pour-
through POU devices are
not rated for the removal
of particulate lead. The
highest level of NSF-42
certification observed for
particle removal was class
5, i.e., removal of parti-
cles > 30–50 µm (Figure
1). The significant release of lead measured in test 2
could also be the result of the low pH in the device
during stagnation; the pH decreased in that test from
7.9 to 5.9 (median values), as shown in Figure 6, part
A. Fresh solder particles (Pb–Sn) without any protective
coating are pH-sensitive and could partially dissolve.
Long stagnation times for the particles in the filter
might also contribute to the dissolution of lead particles
(Reiber, 1991).
Until now, few pour-through POU devices have
passed the revised NSF-53 certification for lead reduc-
tion (NSF International, 2008). These can be found on
the NSF consumer website (www.NSF.org/consumer)
under “search for certified products” category. After
some POU devices lost their certification in 2007,
labels using a variety of indirect references to certi-
fication were found, which might mislead consumers
to falsely conclude that a device is NSF-certified for
lead removal with the revised testing protocol. For
example, notes such as “Laboratory tests follow the
NSF protocol,” and “Effluent lead levels were under
the NSF protocol prescribed limit,” were used on the
package labels. No indication of NSF certification of
these devices was found on the NSF website at the
time of their purchase.
Comparison of various types of POU devices for Pb
removal (Figure 4, Table 3). Pour-through POU filters (G
and H) removed slightly more dissolved lead than the
other types of POU filters and with notably more con-
sistency in test 4. Thus, the CX resin removes dissolved
lead more effectively than the SBAC filters, which are
used in the other POU filters, and results in lower con-
centrations of dissolved lead in the filtered water ( 6.8
µg/L for G and H, 11 µg/L for the other devices
tested). The concentration of dissolved lead in the fil-
tered water from the tested tap-mounted devices some-
what exceeded the 10-µg/L Pb NSF threshold on three
occasions, indicating a possible lack of consistency. In
addition, some dissolved lead values in the effluent of
tap-mounted and under-the-sink POU filters were close
Until now, few pour-through point-of-use
devices have passed the revised NSF-53
certification for lead reduction.
2010 © American Water Works Association
DESHOMMES ET AL | 102:10 • JOURNAL AWWA | PEER-REVIEWED | OCTOBER 2010 103
to 10 µg/L (e.g., 8.7, 9.6, and 9.8 µg/L), and thus could
slightly exceed this limit when combined with low par-
ticulate lead levels in the effluent (0.01–0.9 µg/L). As
mentioned previously, values in excess of the 10-µg/L
NSF threshold limit were not high, but present (10–12
µg/L Pbtotal, an outlier at 43 µg/L Pbtotal because of 38
µg/L Pbparticulate), although the water quality tested was
typical tap water.
All nonpour-through
devices were very effective at
removing particulate lead.
The tap-mounted and under-
the-sink POU devices con-
tained SBAC filters, and
their performances were
higher for particulate lead
(96–99%, effluent Pbparticulate 0.9 µg/L) than for dis-
solved lead removal (80–94%, effluent Pbdissolved 11
µg/L). Only one particulate lead value in the filtered water
significantly exceeded the 10-µg/L NSF threshold (device
F, 38 µg/L). This outlier seemed to be linked to particulate
lead leaching from particles trapped in the filter, as par-
ticulate lead concentration was higher in the effluent than
in the influent (the reason why the removal percentage
could not be calculated). The combined capabilities of
tap-mounted and under-the-sink POU filters to adsorb
dissolved lead and capture particulate lead explain their
recent certification by the revised NSF-53 protocol.
When comparing the under-the-sink devices tested,
it appears that total lead in effluent water for POU
device A slightly exceeded 10 µg/L on three occasions
(10–11 µg/L), whereas POU device B maintained levels
9.4 µg/L. As stated previously, dissolved lead con-
tributed to these levels. A plausible explanation is that
the brass faucet (not NSF-61 certified for Pb) included
in POU device A could have contributed some lead,
whereas the plastic faucet supplied with POU device B
did not. Indeed, Boyd and colleagues (2008a; 2008b)
showed those elements downstream of POU units
could contribute to lead levels in tap water.
Regarding total lead removals for POU devices that
passed the revised NSF-53 protocol (under-the-sink and
tap-mounted), tap-mounted POUs failed to produce
filtered water with lead levels consistently < 10 µg/L
(exceeded 10 times). Although they exceeded 10 µg/L
only marginally, this performance was surprising, given
that the lead levels tested (dissolved plus particulate)
were lower (≤ 160 µg/L, mean = 97 µg/L) compared
with the NSF challenge water (150 ± 30 µg/L for single
point, mean = 150 ± 15 µg/L). However, many param-
eters differed from those of the NSF testing protocol:
the percentage of particulate lead (3–90% for test 4,
20–40% for NSF) and its size (≥ 0.45µm for test 4, at
least 20% between 0.1–1.2 µm for NSF), the type of
particles [PbO2 for test 4, Pb particles from insoluble
solution of Pb(NO3)2 for NSF], and the type of water
used (tap water for test 4, two synthetic waters for
NSF). Values of the under-the-sink POU devices in test
4 that were slightly in excess of the 10-µg/L NSF thresh-
old limit were attributed to lead leaching from the tap
sold with the device. Thus, under-the-sink POU filters
seem more efficient than tap-mounted ones. Results
obtained for under-the-sink devices agree with those
presented by Boyd et al
(2005), showing satisfac-
tory global total lead
removal efficiency for
under-the-sink devices
containing SBAC filters.
Finally, for all of the
POU devices examined in
test 4, stagnation was not
found to create any POU filter breakthrough. However,
the authors believe that the accumulation of lead par-
ticles in POU filters and the effect of stagnation warrant
further investigation. Few studies are available on the
efficiency of these devices for particulate lead removal.
The use of prefilters might be considered in the case of
high-particulate-lead concentrations (Boyd et al, 2008a,
2008b). POU devices may represent cost-effective short-
and long-term solutions for large buildings facing lead
issues resulting from long stagnations and large volumes
of internal plumbing.
Removal/addition of other metals by pour-through POU
filters (Figure 5, Table 4). Overall, metal concentrations
remained below most drinking water guidelines and
standards: the USEPA action level of 1,300 µg/L for Cu
(maximum observed: 130 µg/L) and the WHO provi-
sional guideline of 20 µg/L for Ni (maximum observed:
1.5 µg/L). Residual Cr reached 38 µg/L, which is below
the USEPA MCL of 100 µg/L but close to the Canadian
MAC of 50 µg/L. However, this high value corresponded
to Cr-spiked influent water. Cu was removed effectively
for all the influent concentrations tested and did not
show any decrease in efficiency during the entire POU
lifetime, in accordance with Gulson et al’s observations
(1997) for similar devices. The trend of Ni levels versus
filtered water was not observed for other metals, sug-
gesting that either Ni is more sensitive to saturation of
the filter material or it is desorbed from the filter mate-
rial. The affinity of CX resin for a given cation depends
on its valence, hydrated radius, complex formation,
functional group reactivity, and the ionic strength of the
solution. The resulting expected order of selectivity for
the cations measured is Pb2+ > Ni2+ > Cu2+ > Co2+ (no
data for Cr), making Pb the preferred cation and Co the
less-readily retained (Chen et al, 2006). Ni cations
should have higher affinity than Cu cations; however,
the opposite trend was observed. These differences
probably reflect the presence of complexes that change
the order of selectivity, especially for Cu, Ni, and Co
(Chen et al, 2006), resulting in higher affinity for Cu by
Few studies are available on the efficiency
of point-of-use devices for particulate lead
removal.
2010 © American Water Works Association
104 OCTOBER 2010 | JOURNAL AWWA • 102:10 | PEER-REVIEWED | DESHOMMES ET AL
the resin. This could also be caused by Cu levels in the
influent water, which were higher than those of Ni. Ag
release detected in filtered water (11–43 µg/L) is in
agreement with a previous study on similar POU devices.
Gulson et al (1997) found variable Ag levels in the same
range (first filtration 80 µg/L, subsequent filtrations
40 µg/L), although the current study did not find
higher Ag release at the beginning of the POU lifetime.
However, filtered water was analyzed in tests 1 and 3
after two batches of water had passed through the filter,
in accordance with the manufacturer’s recommenda-
tions; Gulson and co-workers analyzed the first batch of
filtered water. Ag is not considered hazardous to health;
however, a secondary MCL of 0.10 mg/L has been set
by the USEPA. Ag levels measured in the effluent water
were 2–10 times lower than this MCL and so did not
cause any concern.
Effects of POU filters on water quality (Figure 6). Passage
of water through the pour-through POU devices and
stagnation resulted in significantly lower pH levels. This
study suggests that the pH decrease is mostly the result
of CX resin action, simply the exchange of ions that
control pH. In fact, CX is acidic and can exchange its
hydrogen ions with cations from the water (Chen et al,
2006). Carbon dioxide incorporation is unlikely to cre-
ate such a high decrease, given the short filtration time,
the absence of cartridge opening, and the absence of any
stagnation of the sampled water during test 4. The
additional pH depletion in test 2 could result from
nitrification. Residual chlorine was removed by all POU
devices, increasingly so with greater carbon content in
those using SBAC, as expected. Turbidity removal was
not as high as expected for SBAC devices with NSF-42
certification for class 1 particles (0.5–1.0 µm). However,
turbidity can be created by particles smaller than 0.5–
1.0 µm. The unchanged turbidity for pour-through POU
filters is explained by the absence of a physical barrier
for particles < 30 µm in size (NSF-42, class 5). More-
over, the highest turbidity values were observed in efflu-
ent water of device G, which is not NSF-42-certified for
particle removal (Figure 1). Finally, nitrification created
in the filters was not found to affect lead release. Fur-
ther research is needed because GAC combined with
chloraminated distribution systems can cause nitrifica-
tion and a consequent decrease in pH and alkalinity,
which could result in lead release into the filtered water
(Zhang et al, 2008).
CONCLUSION
Pour-through POU devices were the most efficient
for removing dissolved lead because of the presence of
CX resin. However, if lead particles are retained by the
filter media, they can be released or the filter media can
leach them into the filtered water. Other POU devices
tested consisting of an SBAC filter, overall, reduced
dissolved lead effectively and removed particulate lead
extremely well. Some of these devices slightly exceeded
the NSF threshold of 10 µg/L; however, the testing
conditions differed from those of the NSF. The accu-
mulation of the tested particles (PbO2) in POU filters
using SBAC was not associated with lead release from
the devices. Copper was removed effectively by pour-
through POU devices for any concentration tested.
Silver levels detected in effluent water were below the
secondary MCL of 0.1 mg/L. Pour-through POU devices
decreased the pH, maintained turbidity, and removed
chlorine less efficiently than the other devices.
ACKNOWLEDGMENT
The authors acknowledge the support of the city of
Montréal, Quebec (Jean Houle, Laurent Laroche), the
city’s Public Health Department (Monique Beausoleil),
personnel from École Polytechnique de Montréal (Annie
Bernier, Yves Fontaine), and McGill University (Hélène
Lalande), and funding from the Canadian Water Network
and the Natural Sciences and Engineering Research Coun-
cil (NSERC) of Canada Industrial Chair on Drinking
Water partners.
ABOUT THE AUTHORS
Elise Deshommes is a doctoral student
at École Polytechnique of Montréal,
Department of Civil Engineering, and
is employed by the Natural Sciences
and Engineering Research Council
(NSERC) Industrial Chair in Drinking
Water, École Polytechnique of
Montreal, 2900 Boulevard Édouard-
Montpetit Campus de l’Université de Montréal, 2500
Chemin de Polytechnique, Montréal; elise.
deshommes@polymtl.ca. She holds a bachelor’s degree
in civil engineering from École Spéciale des Travaux
Publics in Paris. She began a master of applied science
program on particulate lead in drinking water at the
NSERC Industrial Chair in Drinking Water in 2007.
She has been pursuing her doctoral degree on the same
subject since 2009. Yan Zhang is an environmental
engineer for DXV Water Technologies in Tustin, Calif.
Karine Gendron is an environmental chemistry student
in the department of chemistry at the University of
Montréal, Québec. Sébastien Sauvé is an associate
professor of environmental chemistry at the University
of Montréal. Marc Edwards is a professor in the
Department of Civil and Environmental Engineering at
Virginia Polytechnic Institute and State University in
Blacksburg, Va. Shokoufeh Nour is a research associate
for the NSERC Industrial Chair in Drinking Water.
Michèle Prévost is a professor and senior chairholder
of the NSERC Industrial Chair in Drinking Water.
Date of submission: 11/01/09
Date of acceptance: 05/10/10
2010 © American Water Works Association
DESHOMMES ET AL | 102:10 • JOURNAL AWWA | PEER-REVIEWED | OCTOBER 2010 105
REFERENCES
Abdo, G.N.; Keller, B.J.; & Rupp, G., 1999. An Investigation of Flush-
ing and a Point-of-use Device to Control Copper Corrosion in
Two Small Montana Public Water Supplies. Proc. 1999 AWWA
Ann. Conf., Chicago.
Boyd, G.R.; Pierson, G.L.; Kirmeyer, G.J.; Britton, M.D.; & English, R.J.,
2008a. Lead Release From New End-use Plumbing Components
in Seattle Public Schools. Jour. AWWA, 100:3:105.
Boyd, G.R.; Pierson, G.L.; Kirmeyer, G.J.; & English, R.J., 2008b. Lead
Variability Testing in Seattle Public Schools. Jour. AWWA,
100:2:53.
Boyd, G.R.; Kirmeyer, G.J.; Pierson, G.L.; Hendrickson, S.L.; Kreider,
D.; & English, R.J., 2005. Testing of Point-of-use Filters at Seat-
tle Schools Drinking Fountains. Proc. 2005 AWWA WQTC, Qué-
bec City, Que.
Boyd, G.R.; Shetty, P.; Sandvig, A.M.; & Pierson, G.L., 2004. Pb in Tap
Water Following Simulated Partial Lead Service Pipe Replace-
ments. Jour. Envir. Engrg., 130:10:1188.
CDC (Centers for Disease Control and Prevention), 2009. Important
Update: Lead-based Water Lines. www.cdc.gov/nceh/lead/
waterlines.htm (accessed Feb. 2010).
Chen, J.P.; Yang, L.; Ng, W.-J.; Wang, L.K.; & Thong, S.-L., 2006. Ion
Exchange. Advanced Physicochemical Treatment Processes
(L.K. Wang, Y.-T. Hung, and N.K. Shammas, editors). The
Humana Press Inc., Totowa, N.J.
Deshommes, E.; Laroche, L.; Nour, S.; Cartier, C.; & Prévost, M., 2010.
Source and Occurrence of Particulate Lead in Tap Water.
Water Res., 44:12:3734
Edwards, M.; Triantafyllidou, S.; & Best. D., 2009. Elevated Blood
Lead Levels in Young Children Due to Lead-contaminated
Drinking Water: Washington, DC, 2001–2004. Envir. Sci. & Tech-
nol., 43:5:1618.
Edwards, M. & Dudi, A., 2004. Role of Chlorine and Chloramines in
Corrosion of Lead-bearing Plumbing Materials. Jour. AWWA,
96:10:69.
EQP (Environmental Quality Program), 2008. Final Report on the Per-
formance of Two Point-of-use Drinking Water Arsenic Treat-
ment Systems. Environmental Services Division, Department of
Health and Human Services and Municipality of Anchorage,
Alaska.
Gulson, B.L.; Sheehan, A.; Giblin, A.M.; Chiaradia, M.; & Conradt, B.,
1997. The Efficiency of Removal of Lead and Other Elements
From Domestic Drinking Waters Using a Bench-top Water Fil-
ter System. Sci. Total Envir., 196:3:205.
Lytle, D.A. & Schock, M.R., 2005. Formation of Pb(IV) Oxides in Chlo-
rinated Water. Jour. AWWA, 97:11:102.
McNeill, L.S. & Edwards, M., 2004. Importance of Pb and Cu Particu-
late Species for Corrosion Control. Jour. Envir. Engrg.,
130:2:136.
Molloy, S.L.; Ives, R.; Hoyt, A.; Taylor, R.; & Rose, J.B., 2008. The Use
of Copper and Silver in Carbon Point-of-use Filters for the Sup-
pression of Legionella Throughput in Domestic Water Systems.
Jour. Appl. Microbiol. 104:4:998.
MPHD (Montréal Public Health Department), 2006. Lead In the Tap
Water of Some Homes on the Island of Montréal. Agence de la
Santé, et des Services Sociaux de Montréal, Québec.
NSF International, 2006. Personal communication between Robert
Ferguson and Marc Edwards.
NSF International, 2007. Drinking Water Treatment Units—Health
Effects (NSF/ANSI 53-2007). NSF International, Ann Arbor,
Mich.
NSF International, 2008. NSF Certified First Pour-through Filtration
Device to New Lead Reduction Protocol. www.nsf.org/
business/newsroom/press_releases/press_release.asp?
p_id=16013 (accessed Sept. 21, 2010).
Reiber, S., 1991. Galvanic Stimulation of Corrosion on Lead–Tin
Solder-Sweated Joints. Jour. AWWA. 83:7:83.
Renner, R., 2007. Water-pitcher Filters Miss Lead Particles. Envir.
Sci. & Technol., 41:18:6321.
Sandvig, A.; Kwan, P.; Kirmeyer, G.; Maynard, B.; Mast, D.; Trussell,
R.R.; Trussell, S.; Cantor, A.; & Prescott, A., 2008. Contribution of
Service Line and Plumbing Fixtures to Lead and Copper Rule
Compliance Issues. AwwaRF, Denver.
Triantafyllidou, S.; Parks, J.; & Edwards, M., 2007. Lead Particles in
Potable Water. Jour. AWWA. 99:6:107.
USEPA (US Environmental Protection Agency), 2006a. Point-of-use
or Point-of-entry Treatment Options for Small Drinking Water
Systems. Washington, Office of Water.
USEPA, 2006b. Investigation of the Capability of Point-of-Use/
Point-of-Entry Treatment Devices as a Means of Providing
Water Security. National Homeland Security Research
Center.
USEPA, 2003. Analysis and Findings of the Gallup Organization’s
Drinking Water Customer Satisfaction Survey. Washington,
Office of Groundwater and Drinking Water.
USEPA, 2000. Notification and Reporting Requirements for Partial
Lead Service Line Replacement Under the Lead and Copper
Rule. Washington.
USEPA, 1994. Method 200.8: Determination of Trace Elements in
Waters and Wastes by ICP–MS. Revision 5.4. Environmental
Monitoring System Laboratory, USEPA, Cincinnati, Ohio.
Wellington-Dufferin-Guelph Public Health (WDGPH), 2010. Lead:
Your Health and Your Drinking Water. www.wdghu.org/
page.cfm?id=1679 (accessed Sept. 30, 1010). Ontario, Canada.
Zhang, Y.; Triantafyllidou, S.; & Edwards M., 2008. Effect of Nitrifica-
tion on GAC Filtration on Copper and Lead Leaching in Home
Plumbing Systems. Jour. Envir. Engrg.,134:7:521.
FOOTNOTES
1Ultramass 700 ICP/MS, Varian Inc., Palo Alto, Calif.
2Lead (IV) oxide (PbO2), Alfa Aesar, Ward Hill, Mass.
3Lead (II) nitrate [Pb(NO3)2], Sigma Aldrich, St. Louis, Mo.
4Durapore HV, Millipore, Billerica, Mass.
5B-DTM, Cole Parmer, Vernon Hills, Ill.
67500a ICP-MS, Agilent Technologies, Santa Clara, Calif.
7Accumet AB15 pH-meter, Fisher Scientific, Pittsburgh, Pa.
82100 AN Turbidimeter, Hach Co., Loveland, Colo.
9Cary 100 Scan Spectrophotometer, Varian Inc., Palo Alto, Calif.
2010 © American Water Works Association
