PII: S0043-1354(00)00337-7

Wat. Res. Vol. 35, No. 4, pp. 11001105, 2001

# 2001 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
0043-1354/01/$ - see front matter

RESEARCH NOTE
EFFECT OF ADDING PHOSPHATE TO DRINKING WATER
ON BACTERIAL GROWTH IN SLIGHTLY AND HIGHLY
CORRODED PIPES
B. M. R. APPENZELLER1, M. BATTE2, L. MATHIEU2, J. C. BLOCK1*M,
V. LAHOUSSINE3, J. CAVARD4, D. GATEL5M
1
LCPE, UMR Universite-CNRS 7564, Faculte de Pharmacie, Pole de leau, 15 avenue du Charmois,
54500 Vandoeuvre, France; 2 SERES, Faculte de Medecine, av. de la Foret de Haye, 54500 Vandoeuvre,
France; 3 AESN, 51 rue Salvadore Allende, 92027 Nanterre, France; 4 SEDIF, 14 rue saint Beno t, 75006
Paris, France and 5 GE, 32 place ronde, 92982 Paris-La Defense, France

(First received 28 January 2000; accepted in revised form 5 June 2000)

Abstract}The effect of phosphate addition in drinking water was tested under static conditions as batch
tests and under dynamic conditions using continuously fed reactors. Phosphate supplements in batch tests
from 0.1 to 2 mg PPO4 L 1 did not show any relationship between bacterial growth and phosphate
concentration. Dynamic tests in slightly corroded reactor (stainless steel) treated at 1 mg PPO4 L 1
showed only a moderate improvement in the growth of microorganisms. On the contrary, phosphate
treatment applied to the highly corroded reactor (unlined cast iron) led to an immediate, drastic drop in
iron oxide release and bacterial production. Phosphate uptake by the reactor wall was less than 14% with
the stainless-steel reactor and 7090% with the corroded cast iron reactor. Moreover, about 5% of the
phosphate associated to corroded iron pipe walls was released for 20 days after the end of treatment.
# 2001 Elsevier Science Ltd. All rights reserved
Key words}bacterial growth, phosphate, drinking water, distribution systems, corrosion

INTRODUCTION

Phosphorus is a key element of bacterial cells

(around 1% of the dry weight), where it is part of
biopolymers (DNA, RNA, phospholipids, ATP,
phosphoryl groups of signal molecules), or accumulated into the cytosol as inorganic polyphosphates
(Hardoyo et al., 1994; Kulaev, 1979). Bacteria
constantly tend to maintain a high intracellular
concentration of inorganic phosphates of around
10 mM (Rao et al., 1994; Wanner, 1994). They
therefore combine several specific uptake systems
for efficiently trapping low concentrations of inorganic and organic phosphates from their aquatic
environment (Torriani-Gorini, 1994).
Phosphate has been reported to be the factor
limiting bacterial growth in certain drinking waters
(Miettinen et al., 1997; Sathasivan et al., 1997). The
risks associated with phosphate-based anticorrosion
treatment of distribution systems for lead and iron
oxide control have been therefore recently questioned
and the potential negative effect of adding phosphate
to drinking water has been highlighted. Few reports
have shown that the addition of phosphate leads to a

*Author to whom all correspondence should be addressed.

Tel.: +33-383-59-62-56; fax: +33-383-59-62-60; e-mail:
block@pharma.u-nancy.fr

significant increase in heterotrophic bacterial plate

counts in water (Miettinen et al., 1997; Sathasivan et
al., 1997), or to concomitant legionellosis outbreaks
among consumers (Crespi and Ferra, 1997). In
contrast, Lohmann et al. (1997) and Batte et al.
(1999) did not observe any effect of phosphate
addition on heterotrophic plate counts, and other
authors (Clement and Sandwig, 1996; LeChevallier et
al., 1993; Olson, 1996; Rosenzweig, 1997; Smith et
al., 1989) reported decreases in the occurrence of
coliforms and/or heterotrophic plate counts after
treatment of corroded distribution systems. These
apparently contradictory results may be due to
differences in the environmental situations tested:
the presence or absence of corroded surfaces
(laboratory tests in glass flasks versus distribution
system corroded pipes), the total concentration and
bioavailability of phosphate, which can be modified
by humic substances (De Hann et al., 1990; Francko
and Health, 1979) and the affinity of indigenous
bacterial species for phosphate, etc.
This study was therefore carried out to differentiate between some of the parameters governing the
impact of adding phosphate to drinking water on
bacterial growth. Operating tests were run under
static conditions in laboratory flasks and under
dynamic conditions in continuously fed, highly
corroded (cast iron) and slightly corroded (stainless

1100

Effect of adding phosphate to drinking water on bacterial growth

steel) reactors (Propella reactors). The assays were

carried out by adding phosphoric acid (up to 5 mg P
PO4 L 1) to drinking water having their indigenous
bacterial populations.

MATERIALS AND METHODS

Bacterial counting
The total number of bacteria was determined by DAPI
staining. Aliquots of sample (pure or diluted, 9 mL) were
placed in sterilized glass tubes containing 1 mL of DAPI
(0.5 mg mL 1) (Sigma no. D9542) and 1 mL of Triton X-100
(0.1%) (Prolabo no. 28817295) (all reagents were filter
sterilized through cellulose nitrate filters; pore size, 0.2 mm).
The tubes were mixed for 30 s and left to stand for 10 min.
The samples were then filtered through black polycarbonate
filters (DMF no. 111156; pore size, 0.2 mm). The filters were
rinsed twice with bacterium-free distilled water, dried in hot
air, hydrated with a drop of buffered glycerin (Diagnostics
Pasteur no. 74921), and covered with a coverslip. The filters
were examined under UV light (excitation filter BP330-385,
barrier filter BA420) with an epifluorescence microscope
(Olympus BX60) and an oil immersion objective (Uplan-FI
100  oil immersion universal objective). The bacteria in 30
fields per slide were counted and the results expressed as
number of cells per milliliter.
The viable bacteria were determined by their ability to
grow on R2A agar medium (Difco, Ref. 1826-17-1). Diluted
or undiluted water samples were filtered through celluloseacetate filters (Millipore no. HAWG 047S1; pore size,
0.45 mm). The filters were placed on agar and incubated for 7
days at 22  28C. Colonies were then counted and the results
expressed as colony forming units per milliliter (CFU
7 d mL 1).

1101

Anions in water (SO24 NO3 , Cl ) were measured using

ionic chromatography method (Standard methods, 1998).
Results are expressed as mg L 1.
Batch tests
The batch bacterial growth tests were carried out in 1 L
glass flasks closed by a screw cap. Six flasks were filled up to
1 L with tap water and supplemented with a H3PO4 solution
to obtain final concentrations of 0, 0.1, 0.3, 0.5, 1.0 or
2.0 mg PPO4 L 1. The flasks were then placed under gentle
shaking in the dark at 25  0.58C. Total and cultivable
bacteria in each flask were measured once a day. Three types
of water (W1, W2 and W3) were tested (Table 1). W1 was a
ground water treated only with chlorine. W2 and W3 were
surface water samples subjected to aluminium coagulation
floculation, sand filtration, ozone, activated carbon filtration, pH adjustment with lime and final chlorination. The
characteristics of the surface and ground water used in the
batch tests are given in Table 1.
TM

Dynamic tests in Propella

reactors
TM

The laboratory reactors Propella (Fig. 1) were made of

a 100 mm diameter water distribution pipe, 500 mm long
made of unlined cast iron or stainless steel (ref.: 361L). The
water velocity was controlled with a marine propeller which
pushed the water through an inner cylinder, giving a flow
parallel to the
pipe wall. The water velocity was 0.2 m s 1.
TM
The propella reactors were continuously supplied with tap
water dechlorinated through an activated carbon filter and
filtered through a 2 mm pore size filter at 120 mL h 1
(hydraulic residence time of 24  2 h). The reactors were
supplied with a 50 mg PPO4 L 1 phosphoric acid solution
to give an initial concentration of 1 or 5 mg PPO4 L 1. The
characteristic of the tap water used in all the dynamic
experiments are given in Table 2.
RESULTS

Chemical analysis
Phosphate concentration was measured using the stannous chloride spectrophotometric method (Standard methods, 1998). Samples of 5 mL were diluted to 100 mL with
ultra pure water. 1 mL glycerol/SnCl2 solution and 4 mL
ammonium molybdate were added. The samples were
mixed, left to stand for 10 min and the absorbance was
measured at 690 nm against a blank without phosphate
(spectrophotometer ATI UNICAM-UV/visible Software
V3.01). Results are expressed as mg PPO4 L 1.
Total iron was measured by atomic absorption (Standard
methods, 1998). The results are expressed as mg Fe L 1.
Dissolved organic carbon (DOC) was measured with OI.
corporation analyser (Model 700 TOC) calibrated with a
potassium phthalate solution (Merck, Ref. 4874). Inorganic
carbon was removed using 5% phosphoric acid (Merck,
Ref. 100573) and nitrogen bubbling. Organic carbon was
then oxidized to CO2 with sodium persulphate and the
resulting CO2 was measured by infra red absorbance.
Results are expressed as mg C L 1.

Bacterial growth tests were performed under batch

conditions on three types of drinking water which
differed in their origin (ground water W1, or surface
waters W2 and W3) and some of their chemical and
biological characteristics (the treated surface water
being the richest in dissolved organic carbon and
heterotrophic bacteria) (Table 1). The phosphate
concentrations were below the limit of sensitivity of
the method in all cases (517 mg PPO4 L 1).
A typical bacterial growth was observed in W1
W3 samples with or without added phosphate,
with lag phase of 1 to 2 days, and exponential
growth between days 2 and 4 (data not shown). The
higher numbers of bacteria (total cells counted by
microscopy after DAPI staining, and CFU counted
on agar) were not related to phosphate addition, and

Table 1. Chemical and bacteriological characteristics of the three types of water used in batch tests
Parameters
1

DOC (mg L )
SO24 (mg L 1)
NO3 (mg L 1)
PO34 (mg P L 1)
Cl (mg L 1)
Total suspended bacteria (cells mL 1)
Cultivable suspended bacteria (CFU 7d mL 1)

W1 (Mexy)

W2 (Nancy)

W3 (Nancy)

0.52
22.1
27.1
517
13.6
1.38  104
2.5  101

1.19
30.5
3.7
517
15.6
1.86  105
1.25  103

1.85
39.2
4.6
517
18.1
6.78  104
2.2  103

1102

B. M. R. Appenzeller et al.

there was no doseeffect relationship between added

phosphate and bacterial counts (Fig. 2).
The dynamic tests were run for 50 to 100 days with
continuously fed Propella reactors simulating drinking water circulating in water pipes and allowing
biofilm to accumulate on pipe walls. Bacterial growth
occured without phosphate addition both in the
stainless-steel reactor (i.e. with very few corroded
spots) and in the highly corroded cast iron reactor.
However, the bacterial production, calculated as the
difference in the bacterial concentration at the outlet
and the inlet, was about 10 times higher in corroded
cast iron Propella (2  106 cells mL 1) than in the
stainless steel one (1.8  105 cells mL 1) (Fig. 3). A

3-day continuous injection of phosphate (1 mg P

PO4 L 1) produced only a very slight increase
in bacterial cell concentration (from 5.8 to
8.5  105 cells mL 1 after three days) in stainlesssteel reactors. This was yet not more significant than
the variations observed during the 48 days prior to
the treatment. The corroded cast iron reactor treated
with phosphate behaved in a completely different
way. The phosphate added continuously to the water
caused an immediate, drastic drop in total iron (from
7.5  2.3 to 2.2  0.26 mg L 1), and bacteria counts
whose concentration at the outlet of the reactor was
lower than or equivalent to that of the inlet. In such
conditions, bacterial production was negative
( 3.8  104 cells mL 1) (Fig. 3).
Surprisingly, stopping adding phosphate to the
cast iron Propella reactor after 42 days led to an
immediate but transient and unexplained peak of
bacteria (upto 2  106 cells mL 1) (data not shown)
followed by a return 1 day later to the previous
situation, i.e. a lower concentration of bacteria at the
outlet of the reactor than at the inlet.
The consumption of phosphate in the stainless
steel reactor (added concentration 1 mg PPO4 L 1)
was relatively low (around 12%) (Fig. 4), while the
corroded cast iron Propella consumed a large
fraction of the added phosphate (88% for 1 mg P
PO4 L 1 and 77% for 5 mg PPO4 L 1). However,
the residual phosphate in the bulk water was always
high enough to satisfy bacterial needs for growing on
dissolved organic carbon. There was 230 mg phosphate accumulated in the cast iron Propella reactor
after 1.5 months of treatment (1 or 5 mg PPO4 L 1).
Part of this phosphate (10 mg PPO4 or 5% of the
total) was released by continuous washout during
the month after stopping anticorrosion treatment
(Fig. 5).

DISCUSSION

TM

Fig. 1. Schematic representation of a Propella

(scale drawing 1/7).

reactor

The internal corrosion of distribution pipes

is a major problem faced by utilities, as it
results unaesthetic red water and the leaching
of toxic metals like lead, decreases the hydraulic
capacity of the distribution systems, gives an
environment favouring bacterial growth, causes
higher chlorine demand, and hence less effective
disinfection of biofilm bacteria (Frateur et al., 1999;

Table 2. Characteristics of the tap water used during dynamic Propella

Parameters
1

DOC (mg L )
SO24 (mg L 1)
NO3 (mg L 1)
PO34 (mg P L 1)
Cl (mg L 1)
Total suspended bacteria (cells mL 1)
Cultivable suspended bacteria (CFU 7d mL 1)

TM

reactor experiments

Average

Lower value

Higher value

68
4
4
}
4
46
12

1.41
31.5
4.4
}
18.0
8.22  105
4.62  104

0.96
28
3.6
517
15.6
2.24  105
4.00  103

1.97
39.2
5.4
43
19.1
2.79  106
7.00  105

Effect of adding phosphate to drinking water on bacterial growth

1103

Fig. 3. Produced bacteria (=expressed as the difference

between the outlet and the inlet of the reactor) in stainlesssteel reactor and in corroded cast iron reactor with and
without 1 mg PPO4 L 1 addition (n 4210 according to
the assays).

Fig. 4. Phosphate consumption and free residual phosphate

in stainless steel and cast iron Propella reactor (the numbers
on the x-axis indicate the level of PPO4 treatment).

Fig. 2. Total bacteria ( ) and cultivable bacteria ( ) at the

end of the exponential growth phase (about 100 hs) for three
drinking waters W1, W2 and W3 complemented or not with
phosphate.

Holt et al., 1998; LeChevallier et al., 1993; Rompre

et al., 1999).
Adding orthophosphate is one of the common
processes used to reduce corrosion by-products in

drinking waters (i.e., iron oxides, lead) (Cordonnier

and Barbier, 1994). The high affinity of phosphate
anions for corroded surface metals leads to the
formation of a stable phosphatemetal complex
(Hiemstra and Van Riemsdijk, 1996; Persson et al.,
1996) which limits further corrosion. Such treatment
also leads to an accumulation of phosphate on
the surface of the pipes and keeps free phosphate in
the bulk water below 5 mg P2O5 L 1 (around
2.2 mg P L 1), as required by EU drinking water
standards. The negative effect of such more or less
bioavailable phosphate was recently pointed out. In
this study, the absence of doseeffect relationships in
batch tests shows that phosphate in the water tested
was not the factor limiting bacterial growth or
cultivability. Indeed, the types of water tested by

1104

B. M. R. Appenzeller et al.

estimate the response of biomass to adding phosphate.

Phosphate treatment of corroded networks has a
positive effect. It not only reduces the release of iron
oxide, it also limits the proliferation of heterotrophic
bacteria in pipes by modifying the properties of the
corrosion products.
The high phosphate consumption by iron oxide
also implies that a defined phosphate level in tap
water requires adding more phosphate during the
processing of networks with corroded sections.
Acknowledgements}This work was carried out as part of a
larger research programme (Biofilm V) coordinated by the
Centre International de IEau de Nancy (NANCIE-France),
and funded by the Compagnie Generale des Eaux (ParisFrance), the Communaute Urbaine du Grand Nancy, the
Syndicat des Eaux dIle de France (Paris-France), the Office
National de IEau Potable (ONEP-Maroc), the Agence de
IEau Seine-Normandie (Paris-France), and NANCIE.
Fig. 5. Phosphate release from a corroded cast iron reactor
after phosphate treatment stopping (the slope of the
theoretical washout is calculated as the ratio: volume of
the reactor, V to the flow rate, Q).
REFERENCES

Miettinen et al. (1997) and Sathasivan et al. (1997)

were to a certain extent very particular (i.e. with a
relatively high C/P ratio and a low phosphorus
content), and not representative of most drinking
waters. Thus, in the case of suspicion of anticorrosion adverse effect, phosphate bioavailability
testing, as suggested by Lehtola et al. (1999) and
Sathasivan and Ohgaki (1999), is the only way to
predict the biomass response.
The hypothesis that biofilms would benefit actively
from adding phosphate was not observed in the
slightly corroded stainless-steel pipes, as the changes
in bacteria in the reactor were moderate. This
situation is definitely very different in corroded pipes,
as (i) iron corrosion reactions may promote bacterial
growth (Caccavo et al., 1997; LeChevallier et al.,
1993; Little et al., 1999; Lovley, 1995), and (ii)
phosphate treatment could modify the pipe surface
reactivity and consequently limit bacterial colonization. This was expressed in the experiments by an
immediate drop in the release of iron-oxide particles
and in bacteria following the addition of phosphate.
This is readily explained by the high affinity of
phosphate for iron oxide (Hiemstra et al., 1996;
Persson et al., 1996). The conservative character of
treatment for more than 1.5 months after phosphate
addition was stopped, also suggests a relatively stable
ironphosphate complex, even if there was a low
release of phosphorus for several weeks.

CONCLUSION

The risk of bacterial bloom due to phosphate

anticorrosion treatment appears to occur only in
some specific water resources. The testing of phosphate bioavailability seems to be the best way to

Batte M., Koudjonou B., Prevost M., Hartemann Ph. and

Mathieu L. (1999) Approaches to evaluate the impact of
inhibition of corrosion with polyphosphates on disinfection of a drinking water biofilm. Abstract D39. Proceedings of the Conference on Biofilm Systems. IAWQ, New
York.
Caccavo Jr. F., Schamberger P. C., Keiding K. and Nielsen
P. H. (1997) Role of hydrophobicity in adhesion of the
dissimilatory Fe(III)-reducing bacterium Shewanella alga
to amourphous Fe(III) oxide. Appl. Environ. Microbiol.
63(10), 38373843.
Clement, J. and Sandwig, A. (1996) Experience with
corrosion control. In National Conference on Integrating
Corrosion Control and Other Water Quality Goals.
Boston, MA.
Cordonnier J. and Barbier J. P. (1994) Lead limitation in
tap water: testing technical measures on site. T. S. M. 3,
131136 (In French).
Crespi S. and Ferra J. (1997) Outbreak of legionellosis in a
tourist complex in Lanzarote concomitant with a treatment of water system with megadoses of polyphosphates.
Wat. Sci. Tech. 35(1112), 307309.
De Haan R., Jones I. and Salonen K. (1990) Abiotic
interactions of iron and phosphate in humic lake water
revealed by double-isotope labeling and gel filtration.
Limnol. Oceanogr. 35, 491497.
Francko D. A. and Health R. T. (1979) Functionally
distinct classes of complex phosphorus compounds in
lake water. Limnol. Oceanogr. 24, 463473.
Frateur I., Deslouis C., Kiene L., Levi Y. and Tribollet B.
(1999) Free chlorine consumption induced by cast iron
corrosion in drinking water distribution systems. Wat.
Res. 33(8), 17811790.
Hardoyo, Yamada K., Muramatsu A., Anbe Y., Kato J.
and Ohtake H . (1994) Molecular genetics of polyphosphate accumulation in Escherichia coli. In Phosphate in
Microorganisms: Cellular and Molecular Biology, eds A.
Torriani-Gorini, E. Yagil and S. Silver, pp. 209214.
ASM Press, Washington, DC.
Hiemstra T. and Van Riemsdijk W. H. (1996) A surface
structural approach to ion adsorption: the charge
distribution (CD) model. J. Colloid Interface Sci. 179,
488508.
Holt D. M., Gauthier V., Merlet N. and Block J. C. (1998)
Importance of disinfectant demand of materials for

Effect of adding phosphate to drinking water on bacterial growth

maintening residuals in drinking water distribution
systems. Water Supply 16, 181191.
Kulaev I. S. (ed.) (1979) The Biochemistry of Inorganic
Polyphosphates, 255pp. John Wiley and Sons, Chichester.
LeChevallier M. W., Cheryl C. D., Lee R. G. and Gibbon
D. L. (1993) Examining the relationship between iron
corrosion and the disinfection of biofilm bacteria. J. Am.
Wat. Wks Assoc. 85, 111123.
Lehtola M., Miettinen I. T., Vartiainen T. and Martikainen
P. J. (1999) A new sensitive bioassay for determination of
microbially available phosphorus in water. Appl. Environ.
Microbiol. 65(5), 20322034.
Little B. J., Ray R. I., Jones-Meehan J., Lee C. C. and
Mansfeld F. (1999) Spatial relationships between marine
bacteria and localized corrosion on polymer coated steel.
Biofouling. 13(4), 301321.
Lohmann R., DeMarco J., Hartman D. J., Metz D. H.,
Suzurikawa J. H., Harms L. and Pierson G. (1997) Pilot
study to determine secondary water quality impacts of
corrosion inhibitors. In Proceedings of the Water Quality
Technology Conference, 912 November 1997. American
Water Works Association, Denver.
Lovley D. R. (1995) Microbial reduction of iron, manganese, and other metals. Adv. Agron. 54, 175231.
Miettinen I. T., Vatiainen T. and Martikainen P. J. (1997)
Phosphorus and bacterial growth in drinking water. Appl.
Environ. Microbiol. 63(8), 32423245.
Olson S. C. (1996) Phosphate based corrosion inhibitor
effects on distribution system regrowth. In National
Conference on Integrating Corrosion Control and Other
Water Quality Goals. Boston, MA.
Persson P., Nilsson N. and Sjoberg S. (1996) Structure
and bonding of orthophosphate ions at the iron
oxide-aqueous interface. J. Colloid Interface Sci. 177,
263275.
Rao N. N., Kar A., Roberts M. F., Yashphe J. and
Torriani-Gorini A. (1994) Phosphate, phosphorylated

1105

metabolites, and the Pho regulon of Escherichia coli. In

Phosphate in Microorganisms: Cellular and Molecular
Biology, eds A. Torriani-Gorini, E. Yagil and S. Silver,
pp. 2229. ASM Press, Washington.
Rompre A., Prevost M., Coallier J., Brisebois P., Lavoie J.
and Lafrance P. (1999) Impacts of implementing a
corrosion control strategy on biofilm growth. Abstract
D48, Proceedings of the Conference on Biofilm Systems.
IAWQ, New York.
Rosenzweig W. D. (1997) Influence of phosphate corrosioncontrol compounds on bacterial growth. EPA CR-81161301-0. USEPA, Cincinnati, Ohio.
Sathasivan A. and Ohgaki S. (1999) Application of new
bacterial regrowth potential method for water distribution system}a clear evidence of phosphorus limitation.
Wat. Res. 33(1), 137144.
Sathasivan A., Ohgaki S., Yamamoto K. and Kamiko N.
(1997) Role of inorganic phosphorus in controlling
regrowth in water distribution systems. Wat. Sci. Tech.
35(8), 3744.
Smith D. B., Hess A. F. and Opheim D. (1989) Control of
distribution system coliform regrowth. Proceedings 1989
AWWA-WQTC. San Diego, CA.
Standard Methods for the Examination of Water and
Wastewater (1998) 20th edition, American Public Health
Association, American Water Works Association/Water
Environment Federation, Washington DC, USA.
Torriani-Gorini A. (1994) The Pho regulon of Escherichia
coli. In Phosphate in Microorganisms: Cellular and
Molecular Biology, eds A. Torriani-Gorini, E. Yagil and
S. Silver, pp. 14. ASM Press, Washington, DC.
Wanner B. L. (1994) Multiple controls of Escherichia coli
Pho regulon by the Pi sensor PhoR, the catabolite
regulatory sensor CreC, and acetyl phosphate. In
Phosphate in Microorganisms: Cellular and Molecular
Biology, eds A. Torriani-Gorini, E. Yagil and S. Silver,
pp. 2229. ASM Press, Washington.