Corrosion Science 48 (2006) 322–342
www.elsevier.com/locate/corsci
Characteristics of iron corrosion scales
established under blending of ground, surface,
and saline waters and their impacts on iron
release in the pipe distribution system
Zhijian Tang a, Seungkwan Hong a,b,*, Weizhong Xiao a,
James Taylor a
a
b
Department of Civil and Environmental Engineering, University of Central Florida,
P.O. Box 162450, Orlando, FL 32816, USA
Department of Civil and Environmental Engineering, Korea University, 1, 5-ka, Anam-dong,
Sungbuk-ku, Seoul, 136-701, Republic of Korea
Received 12 February 2004; accepted 3 February 2005
Available online 9 April 2005
Abstract
Interior scales on PVC, lined ductile iron (LDI), unlined cast iron (UCI) and galvanized
steel (G) were analyzed by XRD, RMS, and XPS after contact with varying water quality
for 1 year. FeCO3, a-FeOOH, b-FeOOH, c-Fe2O3, Fe3O4 were identified as primary UCI corrosion products. No FeCO3 was found on G. The order of Fe release was UCI > G P
LDI > PVC. For UCI, Fe release decreased as % Fe3O4 increased and as % Fe2O3 decreased
in scale. Soluble Fe and FeCO3 transformation indicated FeCO3 solid was controlling Fe
release. FeCO3 model and pilot data showed Fe increased as alkalinity and pH decreased.
2005 Elsevier Ltd. All rights reserved.
Keywords: Iron corrosion; Water blending; XRD; RMS; XPS; Siderite
*
Corresponding author. Address: Department of Civil and Environmental Engineering, Korea
University, 1, 5-ka, Anam-dong, Sungbuk-ku, Seoul, 136-701, Republic of Korea. Tel.: +82 2 3290
3322; fax: +82 2 928 7656.
E-mail address: [email protected] (S. Hong).
0010-938X/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2005.02.005
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
323
1. Introduction
Significant population growth, in conjunction with the depletion of groundwater
resources, has prompted development of alternative sources for drinking water supply. Tampa Bay Water (TBW) has developed surface water and desalination facilities
to augment future drinking water supply [1,2]. The finished water quality of these
alternate supplies is different from finished groundwater and will interact with the
existing distribution scale and have adverse effects on distribution water quality.
The disruption of existing scales on pipe surfaces is of great concerns and can lead
to release of several undesirable substances to drinking water.
Various pipe materials, such as iron, galvanized steel, concrete, PVC, and copper,
have been extensively used in the drinking water distribution systems. Unlined iron
pipes are very prone to electrochemical corrosion and their internal surfaces can become severely corroded over time. Typical iron corrosion products formed on aged
pipe surfaces can be described using three different layers: (a) a macroporous layer of
black rust (magnetite, Fe3O4) in contact with the metal, (b) a microporous film of
mixture of Fe2+ and Fe3+ species that covers the macrolayer, and (c) a top layer
of red rust (mainly goethite, a-FeOOH and hematite, a-Fe2O3). The exact composition and structure of iron corrosion scales, however, varies significantly with water
qualities as well as flow properties [3,4].
In past decades, numerous studies have been conducted to investigate iron corrosion to elucidate fundamental mechanisms responsible for iron release, which often
causes red water. Alkalinity, pH, chloride and sulfate are primary water quality
parameters affecting iron corrosion [2,5–10]. Kuch [11] proposed that red water
was caused by reduction and dissolution of formed corrosion products (e.g. cFeOOH) when oxygen was lacking during stagnation. A study by Sontheimer
et al. [12] suggested that the formation of siderite (FeCO3) might play an important
role in iron corrosion. Despite the extensive research efforts, adverse impacts of
blending different source waters on iron release have not been investigated systematically. In particular, characteristics of chemical films established for different blended
waters in the same pipe are not understood, and as a result, their impacts on iron
release are largely unknown.
A pilot-scale pipe distribution study was conducted to determine the effect of
blending different source waters on distribution system water quality. In this study,
finished ground, surface, and saline waters, were blended in varying ratios and discharged to pilot distribution systems (PDSs). Water quality parameters and scale
characteristics were monitored for 12 months and correlated to pipe material and
water quality. Aged pipes taken from distribution systems were used to build the
PDSs and consisted of unlined cast iron, galvanized steel, lined ductile iron and
PVC pipes. Pipe coupons were placed in cradles connected to the PDS. XRD
(X-ray diffraction), RMS (Raman spectroscopy), and XPS (X-ray photoelectron
spectroscopy) analyses were performed on the coupon corrosion scales. A thermodynamic equilibrium model based on solid phases identified by surface characterization
was developed to predict iron release as a function of blended water qualities.
324
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
2. Experimental work
2.1. Source waters
Seven pilot treatment facilities were constructed to supply finished water to the
pilot distribution systems. These pilot plants were designed to simulate the existing
and future TBW water treatment processes. All processes included chloramination
and pH stabilization for post treatment [1].
This study focused on three principal sources: G1, S1, and RO. The G1 source is
conventionally treated groundwater (i.e. aeration) that distribution systems had received for decades. The S1 represents enhanced surface water treatment consisting
of ferric sulfate coagulation, flocculation, settling, filtration, ozonation, biologically
activated carbon filtration. The RO source simulates desalted seawater by reverse
osmosis membrane processes.
2.2. Pipe materials
Four different aged pipe materials were used in this study: PVC, lined ductile iron
(LDI), unlined cast iron (UCI) and galvanized steel (G). The aged pipes were obtained from actual TBW distribution systems in Pinellas, Pasco, and St. Petersburg
Counties, Florida. Based on surface morphology and chemistry studied by SEM/
EDS and XRD, no significant difference was found in terms of corrosion products
for the same pipe material from different locations.
2.3. Pilot pipe distribution systems
A total of eighteen pilot distribution systems were assembled with aged pipe materials that had received treated groundwater similar to the G1 source. Fourteen of
these lines, referred to as hybrid lines, were made of PVC, LDI, UCI and G pipes
in that order. The four remaining lines were made of single pipe material. The nominal length of each line was approximately 30 m. Pipe diameters were 15 cm with the
exception of galvanized which was 5.6 cm.
The pilot systems were operated with a five-day and a two-day hydraulic residence times (HRT). The flow velocities were 0.25 and 0.625 m/h. The pilot systems
were operated to reflect dead zones or worst case conditions in actual distribution
systems. In order to mimic the hydraulic conditions in a full-scale system, PDSs
were flushed weekly with five pipe volumes at 0.3 m/s. Influent and effluent samples were monitored for dissolved and total iron, pH, alkalinity, temperature,
chloride, sulfate, calcium, magnesium, sodium, silica, dissolved oxygen, and chlorine residual. Dissolved iron was measured after sample filtration using 0.45 lm
filter. Total iron was determined without filtration and included particulate and
dissolved iron.
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
325
2.4. Pipe coupon study
Individual cradles were developed for every pipeline for the pipe coupon study.
Each cradle consisted of 10 cm PVC pipe housing and 8.8 cm PVC pipe coupon
holders. The length of the cradles was approximately 3 m. Each cradle was connected in series to a pilot distribution system so that equivalent coupon, PDS water
quality and flow velocity were maintained. The pipe coupon holders were easily removed from the cradle and replaced after each experimental phase for bio-film and
chemical deposit analysis.
The pipe coupons of four different materials were cut into two rectangular coupons (10 · 7.5 cm): initial and final samples. The back of each pipe coupon was
wrapped with Teflon tape to prevent exterior contamination. The pipe coupons were
mounted and fixed on coupon holders. The coupons were removed from the cradles
after a 3-month incubation period.
2.5. Pipe surface characterization
Various surface characterization techniques have been employed to identify the
corrosion products on iron pipe materials [13–23]. XRD, RMS, and XPS analyses
were performed in this study. XRD is a versatile, non-destructive analytical technique
for identification of the various crystalline phases present in unknown powdered and
solid samples. RMS provides information about molecular vibrations that can be
used for sample identification and quantification. The technique allows the identification of homogeneous materials on the basis of their molecular vibrational spectra, obtained by excitation with visible laser light. Surface analysis by XPS, on the other
hand, involves irradiating a solid in vacuum with monoenergetic soft X-ray and analyzing the emitted electron by energy. Because the mean free path of electron in solids
is very small, the detected electrons originate from only the top few atomic layers,
making XPS a unique surface-sensitive technique for chemical analysis.
The XRD analysis was carried out on powder, which had been previously
scratched from the corrosion layer on the coupons and grounded in a mortar. The
equipment used consisted in a 3720 X-ray diffractometer with a fine structure airinsulated X-ray tube with a copper anode (Cu Ka1 5406 Å). Full X-ray diffraction
patterns were recorded for the scan angles (2h) from 10 to 80 to identify iron oxides. Using the ICDD-PDF database, individual crystalline phases were identified
from their observed XRD patterns.
RMS analysis was done by using micro-Raman system (Renishaw, Model: RM
1000 B). A 25 mW argon-ion laser (514 nm) was incident on the sample surface
and the Raman scatters were collected. Raman spectrum was obtained in the range
of 100–1200 cm1 as the expected corrosion products of iron and steel generally
show their peaks in this range. Calibration is done with two standard samples: pure
alumina silicon wafer.
For XPS analysis, 4 · 4 mm samples were cut from pipe coupons and stored in the
desiccators to remove adsorbed moisture. The XPS scan was performed using a 5400
PHI ESCA. The base pressure during analysis was 109 Torr. Mg-Ka X-radiation
326
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
(1253.6 eV) was used at 350 W. Both survey and high-resolution narrow spectra were
recorded with electron pass energy of 44.75 eV and 35.75 eV, respectively, to achieve
the maximum spectral resolution. Any charging shift produced by the samples was
carefully removed by using a BE scale referred to C(1s) BE of the hydrocarbon part
of the adventitious carbon line at 284.6 eV. Non-linear least squares curve fitting was
performed using a Gaussian/Lorentzian peak shape after background removal.
3. Results and discussion
3.1. Identification of corrosion products on aged pipe materials
Surface morphology of the test coupons of four different aged pipe materials was
first examined to determine the surface features of the historical corrosion products
on the pipe surface which was in practical equilibrium with conventionally treated
groundwater (G1) in the distribution systems. Representative optical micrographs
showing reddish brown corrosion products are presented in Fig. 1. There were no
Fig. 1. Micrograph pictures of interior corrosion products on aged (a) unlined cast iron; (b) galvanized
steel; (c) lined ductile iron; (d) PVC pipes taken from distribution systems receiving conventional
groundwater.
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
327
significant visual differences among coupons of the same material. The UCI coupons
exhibited very thick corrosion layers, while thinner corrosion layers were found for
G coupons. As expected, the corrosion deposits were significantly less on LDI and
PVC coupons.
The XRD patterns of the corrosion products on cast iron, galvanized steel, and
lined ductile iron is shown in Fig. 2. a-FeOOH, b-FeOOH, c-Fe2O3, Fe3O4, FeCO3
and SiO2 crystalline compounds were found on the cast iron coupons. The primary
crystalline compounds on the G coupon were a-FeOOH, b-FeOOH, c-Fe2O3,
Fe3O4, and ZnO as shown in Fig. 2(b). Major peaks for b-FeOOH and ZnO were
observed in Fig. 2(b). No clear crystalline phase was found on the internal pipe surface of the LDI coupons because only small iron corrosion deposits were found on
the internal pipe surfaces. These deposits probably were deposited prior to extraction
and came from unlined metal surfaces in the actual distribution system.
The XPS survey spectra showed the primary UCI scale elements were C, O, Fe, Si,
and N. A significant amount of Zn was found in the G scale by the XPS survey scan.
High resolution scan of Zn(2p3/2) peak indicated the presence of ZnO and Zn(OH)2
on the surface of the corrosion products, verifying the XRD results shown in Fig.
2(b). As expected, for LDI and PVC coupons, no significant iron signals were
observed in the XPS scan spectra.
3.2. Iron release from aged pipe materials
The average water quality of G1, S1 and RO for one year (12/15/01 to 12/14/02) is
summarized in Table 1. G1 had the highest alkalinity, while S1 and RO exhibited the
highest sulfate and chloride, respectively. The high S1 sulfate was due to ferric sulfate
coagulation. The high RO chlorides came from the addition of inorganic sea salt,
which simulate Cl concentrations in RO permeate produced from sea water and
match the finished water quality of the desalination plant in Tampa, Florida. The
average values of total iron in the PDSs receiving G1, S1 and RO finished waters
were 0.06 mg/L, 0.63 mg/L and 0.44 mg/L respectively. More than 90% of the total
iron released was in the particulate form. As shown by these and other results, iron
release from UCI and G pipe increased significantly with decreasing alkalinity [1,2].
Effluent PDS total iron concentration from single material lines were monitored
over one year in four 3-month phases. The blend of influent PDS finished waters
was changed every three months to vary PDS water quality. Table 2 shows the average water quality for each phase. The hydraulic retention time (HRT) was 5 days
during Phases 1 to 3, and 2 days from the end of Phase 3 to the end of the project.
The HRT was changed to maintain residual above 25 C. The source water was a
blend of G1/S1/RO, specifically 23/45/32% for Phases 1 and 3, and 60/30/10% for
Phases 2 and 4. The results are presented in Fig. 3 as a function of PDS material
for each phase. As shown, the order of PDS effluent total iron release by material
was UCI > G > LDI > PVC, which was expected. The greater iron release from cast
iron pipes in Phases 1 and 3, relative to Phases 2 and 4, was attributed to low alkalinity and was caused by a smaller G1 percentage in the PDS feed water and
decreased HRT.
328
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
Fe 3 O4
γ -Fe 2 O3
Fe 3 O 4
FeCO 3
300
α -Fe
γ-Fe 2O 3
Fe3O 4
FeCO 3
400
SiO 2
α -FeOOH
Intensity (counts)
500
∋
600
-Fe
β -FeOOH
700
200
100
0
10
20
30
40
(a)
50
60
80
β -FeOOH
1600
ZnO
1400
Fe3 O4
Fe-C
Fe2 O3
ZnO
γ -Fe O
2 3
400
CaCO 3
600
α -FeOOH
Zn5(CO 3) 2(OH)6
800
Fe-C
α -Fe
1000
γ -Fe 2 O3 /ZnO/Fe 3 O4
1200
Intensity (counts)
70
2-Theta (deg.)
200
0
10
20
30
40
(b)
50
60
70
80
60
70
80
2-Theta (deg.)
140
Intensity (counts)
120
100
80
60
Fe-C
20
γ -Fe 2 O3
CaCO 3
40
0
10
(c)
20
30
40
50
2-Theta (deg.)
Fig. 2. Representative XRD patterns of corrosion products on the interior of aged (a) unlined cast iron;
(b) galvanized steel; (c) lined ductile iron pipes exposed to conventionally treated groundwater.
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
329
Table 1
Average water quality of primary water sources
Water quality parameter
G1
S1
RO
Alkalinity (mg/L as CaCO3)
pH
Chloride (mg/L)
Sulfate (mg/L)
Sodium (mg/L)
Calcium (mg/L)
SiO2 (mg/L)
Total dissolved solids (mg/L)
Dissolved oxygen (mg/L)
Temperature (C)
UV-254 (cm1)
207
7.87
28.9
26.1
18.0
84.8
13.7
421
6.53
24.2
0.060
60
7.92
37.1
190.2
48.7
56.5
10.0
423
6.31
24.0
0.024
69
8.06
91.7
5.8
52.3
28.7
3.5
267
5.15
24.1
0.028
G1 is conventional groundwater treated by aeration, S1 is surface water treated by enhance ferric sulfate
coagulation, and RO is saline water desalted by RO membrane. Water quality data were collected from 12/
15/01 to 12/14/02.
Table 2
Operating conditions and water quality for single material PDSs
Phase
Phase
Phase
Phase
Phase
1
2
3
4
Blending
(G1/S1/RO%)
HRT
(days)
pH
Alkalinity
(mg/L as CaCO3)
Chloride
(mg/L)
Sulfate
(mg/L)
Temperature
(C)
23/45/32
60/30/10
23/45/32
60/30/10
5
5
5
2
8.06
7.94
7.82
7.89
110
158
96
125
56
42
59
39
109
76
77
56
18.8
26.1
26.7
18.5
The pilot study was conducted in four three-month phases (12/15/01–12/14/02). Note that water quality
varied slightly under the identical blending ratios (e.g. Phases 1 and 3 or Phases 2 and 4) because of the
seasonal changes in raw source water qualities.
Phase 1: 12/15/01–03/14/02, Phase 2: 03/15/02–06/14/02, Phase 3: 06/15/02–09/14/02, Phase 4: 09/15/02–12/
14/02.
Fig. 4(a) shows the XRD patterns of the corrosion products on UCI coupons
after a 3-month exposure to G1/S1/RO (23/45/32%). The crystalline compounds
on the UCI coupons were a-FeOOH, b-FeOOH, c-Fe2O3, Fe3O4, FeCO3, and
SiO2. The c-Fe2O3, Fe3O4 and FeCO3 on the UCI coupons increased during incubation. More background noise was noticeable in these scans, which implied more
amorphous phases were in the corrosion scales following a 3-month exposure to
new water. In other words, the corrosion process was ongoing.
The XRD pattern of corrosion products on G coupons is presented in Fig. 4(b)
and shows a-FeOOH, b-FeOOH, Fe2O3, Fe3O4, ZnO and Zn5(CO3)2(OH)6 were
the primary corrosion products. As seen in Fig. 4(b), a-FeOOH and b-FeOOH decreased, and Fe2O3 and Fe3O4 became the major corrosion products after a 3-month
exposure to G1/S1/RO water. Unlike UCI, FeCO3 was not found in the XRD analysis of G, and ZnO and Zn5(CO3)2(OH)6 were observed. The XRD analyses of LDI
and PVC coupons showed no obvious crystalline compounds, with exception of
small amounts of Ca and Fe precipitates such as CaCO3 and c-Fe2O3. Amorphous
330
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
Fig. 3. Effect of pipe materials on total iron release. The detailed operating conditions and water qualities
are summarized in Table 2 for each phase. The total iron data were obtained from single material lines in
order to avoid any possible contaminations caused by iron release from the different pipe materials in the
upstream.
phases were present in the samples. Consequently, background noises in the XRD
spectra were high and covered minor iron phase peaks.
The existence of these iron phases was confirmed by Raman spectroscopy (RMS).
A representative Raman spectrum of a UCI coupon is presented in Fig. 5. The band
285/740/1083 revealed the existence of significant amounts of FeCO3. The a-FeOOH,
b-FeOOH, c-Fe2O3 and Fe3O4 were identified by the band of 208/250/290/394 cm1,
310/390/705 cm1, 250/390/692 cm1 and 224/550/670 cm1, respectively. It is clear
that RMS spectra verified the results of XRD analysis.
Lastly, the XPS analysis performed on four different pipe coupons of single material lines showed the results similar to those of XRD analysis except that no FeCO3
was detected, which is due to FeCO3 instability.
3.3. Correlation between surface characteristics and iron release
Blended finished waters were also fed to the hybrid PDSs which consisted of PVC,
LDI, UCI and G pipes in series, and coupons cradles identical to the single material
PDSs. The coupons were incubated for three months, harvested before phase
changes and analyzed by XRD and XPS. Ninety percent of the iron found in the
coupon scales came from the UCI pipes. The remaining 10% of the iron came from
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
331
γ -Fe2 O3
Fe3O4
150
Fe3O4
FeCO3
FeCO3
200
Fe-C
FeCO3
250
γ -Fe2 O3
FeCO 3
α -FeOOH
300
CaCO3
Intensity (counts)
350
FeCO 3
400
β -FeOOH
450
100
50
0
10
20
30
40
50
60
80
Fe 2 O3
1600
ZnO
Fe3O4
ZnO
ZnO
Fe-C
Fe3O4
Fe-C
β -FeOOH
600
α -FeOOH
800
Zn5 (CO3)2(OH)6
1000
Fe3 O4
CaCO 3
1200
α -FeOOH
β -FeOOH
ZnO
1400
Intensity (counts)
70
2-Theta (deg.)
(a)
400
200
0
10
(b)
20
30
40
50
60
70
80
2-Theta (deg.)
Fig. 4. Representative XRD patterns of corrosion products on (a) unlined cast iron; (b) galvanized steel
pipe coupons following three months of exposure to G1/S1/RO (23/45/32%) water in single material lines.
the G pipes [2]. Thus, iron release from UCI and G pipes was investigated using PDS
influent water quality created by blends of G1 (100%), S1 (100%), RO (100%), G1/S1
(55/45%), G1/RO (68/32%) and G1/S1/RO (62/27/11%).
The UCI XRD patterns of coupons exposed to these six different blended waters
showed little difference, except that G1 exhibited the lowest FeCO3 content as indicated by the relative FeCO3 peak intensity. The XRD spectra of the G coupons also
showed no significant variation. Since the XRD was done on powdered samples, the
change following three-months of incubation might have occurred only in the first 1–
3 lm of the surface scale and was masked by sample preparation.
332
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
FeCO3
FeCO3
Intensity
α -FeOOH,γ -Fe2O3
Fe3O4
FeCO3
Fe3O4
Fe3O4
γ -Fe2O3, β -FeOOH
β -FeOOH
α -FeOOH
150
350
550
750
950
-1
Raman Shift (cm )
Fig. 5. Representative RMS patterns of corrosion products on unlined cast iron pipe coupons following
three months of exposure to G1/S1 (55/45%) water.
The XPS technique is very surface sensitive and provided information on chemical
changes that occurred at the interface between water and inorganic film on pipe surface due to dissolution and precipitation of corrosion products caused by changes in
water quality. Standard peak positions (i.e., 710.4 eV for Fe3O4, 710.8 eV for Fe2O3,
and 711.8 eV for FeOOH) and FWHM (i.e., 3.6 eV for Fe3O4, 3.8 eV for Fe2O3 and
FeOOH) of iron species were determined by scanning pure samples and comparison
to the NIST XPS Database. The differences of binding energy of various iron phases
are significant (>0.3 eV) and can be differentiated. Any samples charging shifts were
carefully removed by using a BE scale referred to as C(1s) BE of the hydrocarbon
part of the adventitious carbon line at 284.6 eV. Non-linear regression based on minimization of least squares of the error was used to correlate the iron content in the
solid phases of the iron corrosion products to a Gaussian/Lorentzian curve for peak
shape. Following this procedure, the content accuracy was within a ±0.1a (a is the
peak area percentage given by Fe peak deconvolution). Fig. 6(a) presents the detailed scan and the deconvolution of Fe2p3/2 peak of unlined cast iron aged coupon
exposed to G1(100%). The deconvolution of Fe(2p3/2) peak revealed four separate
peaks, which corresponded to Fe3O4, Fe2O3, FeOOH and the minor peak. Minor
peaks shown here were in the range from 713 eV to 715.5 eV. It might have two
sources: (1) the minor Fe2+(2p3/2) peaks from FeSO4 and/or the Fe2+ portion of
Fe3O4; and (2) the Fe3+(2p3/2) peak of Fe2(SO4)3. FeSO4 and Fe2(SO4)3 came from
the deposition of soluble iron sulfate species after the UCI coupons were taken from
the high sulfate finished waters produced by Fe2(SO4)3 coagulation. The deconvolution of Fe(2p3/2) peak of G coupons is presented in Fig. 6(b) and also shows the
presence of Fe2O3, Fe3O4, and FeOOH. The curve fit summary of the deconvolution
of Fe(2p3/2) peak is listed in Table 3 for UCI and G coupons exposed to these six
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
333
Min: 0 Max: 2345
Fe3O4
N(E)
Minor Peak
Fe2O3
FeOOH
740
736
732
728
724
720
716
712
708
704
700
Binding Energy (eV)
(a)
Min: 0 Max: 4127
Fe2O3
N(E)
Minor Peak
Fe3O4
FeOOH
740
(b)
736
732
728
724
720
716
712
708
704
700
Binding Energy (eV)
Fig. 6. Representative high-resolution XPS scan and the deconvolution of Fe(2p3/2) peak of corrosion
products formed on (a) cast iron; (b) galvanized steel pipe coupons following 3-month exposure to
conventionally treated groundwater.
water environments. As shown, the peak area percentage, which stands for the relative amount of corrosion products in the scale, varied with water quality.
The correlation (R2) between Fe(2p3/2) peak area percentage of UCI coupons and
the total and dissolved iron levels from the hybrid lines are schematically presented
in Fig. 7. A linear relationship between Fe3O4 and total and dissolved iron release
from the hybrid PDS is shown in Fig. 7(a). The slopes are negative, which means
iron release decreased as Fe3O4 in the corrosion scale increased. Stability of Fe3O4
in the scale reduces iron release. Another correlation was found between peak area
percentages of Fe2O3 and pilot iron release data. Fig. 7(b) showed that dissolved and
total iron release increased with increasing Fe2O3. Fe2O3 plays an important role on
iron release in water. But Fe2O3 solubility is very low and actual dissolved iron concentration was much higher than predicted by Fe2O3 equilibrium, which indicates
that Fe2O3 should not be the controlling solid phase of iron release.
334
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
Table 3
Curve fit summary of the deconvolution of Fe(2p3/2) peak of cast iron and galvanized steel coupons
exposed to six different blended water environments for 3 months
Cast iron
Galvanized steel
G1 (100%)
Fe3O4
Fe2O3
FeOOH
Minor peak
58.3%
22.5%
6.2%
13.0%
Fe3O4
Fe2O3
FeOOH
Minor peak
23.9%
56.3%
3.4%
16.4%
S1 (100%)
Fe3O4
Fe2O3
FeOOH
Minor peak
0.0%
47.3%
30.9%
21.8%
Fe3O4
Fe2O3
FeOOH
Minor peak
19.5%
51.9%
10.4%
18.2%
RO (100%)
Fe3O4
Fe2O3
FeOOH
Minor peak
47.1%
32.1%
4.6%
16.2%
Fe3O4
Fe2O3
FeOOH
Minor peak
39.0%
44.7%
4.3%
12.0%
G1/S1 (55/45%)
Fe3O4
Fe2O3
FeOOH
Minor peak
29.8%
45.1%
8.8%
16.4%
Fe3O4
Fe2O3
FeOOH
Minor peak
24.4%
24.4%
34.6%
16.6%
G1/RO (68/32%)
Fe3O4
Fe2O3
FeOOH
Minor peak
41.6%
34.9%
7.0%
16.5%
Fe3O4
Fe2O3
FeOOH
Minor peak
46.0%
32.4%
0.7%
20.9%
G1/RO/S1 (62/27/11%)
Fe3O4
Fe2O3
FeOOH
Minor peak
45.5%
29.6%
12.4%
12.5%
Fe3O4
Fe2O3
FeOOH
Minor peak
11.8%
66.2%
6.9%
15.1%
In this study, FeCO3 was detected by XRD and RMS, but not by XPS. HeuerÕs
study [17] showed that FeCO3 transformed to Fe2O3 completely after exposure to air
at room temperature for 4 h. In our study, iron coupons were taken from pilot distribution systems, stored in dessicators for 2 weeks, prepared for analysis and analyzed. In the long storage period, FeCO3 transformation to Fe2O3 occurred in the
surface layer prior to XPS analysis as shown by the Fe2O3 peak area found by
XPS. The positive linear correlation between FeCO3 in the surface layer and iron
concentration indicates FeCO3 is the controlling solid phase for iron release in
UCI in drinking water distribution systems.
Correlations between surface characteristics of aged G pipes and iron release were
also examined. Table 4 summarizes the coefficients of determination (R2) for regression XPS and pilot iron data. No clear relationship was found between different iron
species and iron release levels. All R2 values were less than 0.4. This is not surprising
since elevated iron levels were primarily due to the disruption and releases of chemical films formed on cast iron pipes.
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
0.7
Fe3O4 vs. Dissolved Iron
0.05
0.6
0.04
0.5
R2 = 0.68
0.4
0.03
0.3
0.02
0.2
2
R = 0.75
0.01
0.1
0
0
(a)
5
10
20
25
30
35
40
45
50
55
0
65
60
Fe3O4 (%)
0.8
0.06
Fe2O3 vs. Total Iron
0.7
Total Iron (mg/L)
15
Fe2O3 vs. Dissolved Iron
0.05
0.6
0.04
0.5
R2 = 0.47
0.4
0.03
0.3
0.02
0.2
R2 = 0.88
0.1
Dissolved Iron (mg/L)
Total Iron (mg/L)
0.06
Fe3O4 vs. Total Iron
Dissolved Iron (mg/L)
0.8
335
0.01
0
0
20
25
30
(b)
35
40
45
50
Fe2O3 (%)
Fig. 7. Fe3O4 and Fe2O3 versus actual total and dissolved iron released from hybrid pilot distribution lines
receiving various blended waters (Phase 3).
Table 4
Correlation (R2) between total and dissolved iron release from hybrids lines and relative peak areas of
different iron corrosion products on galvanized steel pipe internal surfaces determined by XPS iron peak
analysis
R2
Total iron
Dissolved iron
Fe3O4
Fe2O3
FeOOH
Minor peak
0.002
+0.005
0.001
+0.002
0.318
0.001
+0.353
+0.328
Note that + and signs stand for positive and negative correlations, respectively.
3.4. Kinetic effects on iron release and surface characterization
The change of scale formation and iron release with respect to time was investigated in this work. Mutoti et al. [26] found that iron release from UCI pipe came
from the interior wall of the pipe and hence, could be accurately described by a zero
order kinetic model. Imran et al. [26] found that iron release from UCI and G pipe
over a five day HRT was accurately predicted by a steady state water quality model.
336
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
The models developed by Mutoti and Imran were published in the AwwaRF Final
Report ‘‘Effects of Blending on Distribution System Water Quality’’, Taylor et al. [26]
and showed total iron release was dependent on pipe material, water quality, geometry, and velocity. Dietz [27] found that the time to reach steady state total iron in the
bulk water following a change in water quality was approximately one month and
was accurately described by a first order model, which was dependent on pipe material, water quality and time. These results found total iron release came from the
interior pipe wall and steady state total iron release was dependent on pipe material,
geometry and water quality; however the concentration of total iron released in the
bulk water decreased with velocity.
The change in surface characterization over time was investigated by exposing
coupons to G1, S1, RO and a G1/S1/RO blend for one, three and six month periods
and analyzing the coupons by XPS. As shown in Fig. 6, the results indicate FeOOH,
Fe2O3, and Fe3O4, were the major corrosion products in the top surface layer. Previously, Fe2O3 was shown to correlate well with total and dissolved iron. Because the
detected Fe2O3 might have been transformed from FeCO3, and the observed dissolved iron concentrations were much higher than Fe2O3 solubility, the correlation
of Fe2O3 with iron release might imply the correlation of FeCO3 with iron release.
Fig. 8 shows that Fe2O3 peak area percentage does not significantly change for a
one, three, or six months incubation time, which implies the Fe2O3 (FeCO3) content
in the surface layer reached equilibrium in one month, which was consistent with
Dietz [27]. The results also show that even though the Fe2O3 peak area % changed
to some degree with time and source waters, all of them follow the trend: S1 >
RO > G1/S1/RO > G1. Higher percentages of Fe2O3 (FeCO3) area corresponded
to higher iron release.
3.5. Thermodynamic equilibrium iron release model
The soluble iron release from aged pipe surfaces may be predicted using thermodynamic calculations based on equilibrium constants associated with solid phases in
60
G1
S1
RO
G1/S1/RO
Fe2O3 Peak Area (%)
50
40
30
20
10
0
1
2
3
4
5
6
7
Elapsed Time (Months)
Fig. 8. Effect of incubation time on surface characteristics (Fe2O3 peak area %).
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
337
corrosion scales. Such models are inherently flawed as particle release is not considered; however these models may provide insight into release mechanisms. Based on
XRD and XPS analyses of UCI coupons following incubation, siderite (FeCO3) was
assumed to be the controlling solid phase for soluble iron. Pourbaix (potential–pH)
diagrams were created and also suggested that FeCO3, Fe(OH)2 or Fe(OH)3 existed
within the free energy and pH conditions of our pilot study.
The thermodynamic model assumes all soluble iron is in equilibrium with all solid
phases considered for model development. Concentrations of dissolved complexes
(e.g. carbonate, chloride and sulfate) were estimated after consideration of related
reactions. The summation of all dissolved iron species is total bulk water iron and
is shown in following equation.
FeT ¼ ½Fe2þ þ ½FeOHþ þ ½FeðOHÞ02 þ ½FeðOHÞ
3
þ
0
0
þ ½FeHCOþ
3 þ ½FeCO3 þ ½FeSO4 þ ½FeCl 1010:89 ½Hþ 109:5 1020:6 1032
1
þ
þ
þ 102 Alka
þ
½Hþ Alka 1010:3
½Hþ 2 ½Hþ 3
!
1010:3
4:38
2:25
2
0:9
þ 10 Alka þ 10 ½SO4 þ 10 ½Cl ½Hþ ¼
Table 5
Chemical reactions used for thermodynamic modeling [24,25]
Reactions
2+
pK
+
+
Fe + H2O M FeOH + H
Fe2þ þ 2H2 O $ FeðOHÞ02 þ 2Hþ
þ
Fe2þ þ 3H2 O $ FeðOHÞ
3 þ 3H
þ
$
FeHCO
Fe2þ þ HCO
3
3
0
Fe2þ þ CO2
3 $ FeCO3
2
2þ
Fe þ CO3 $ FeCO3ðSÞ
Fe2+ + 2H2O M Fe(OH)2(S) + 2H+
FeCO3ðSÞ þ 2H2 O $ FeðOHÞ2ðSÞ þ 2Hþ þ CO2
3
0
Fe2þ þ SO2
4 $ FeSO4
2+
+
Fe + Cl M FeCl
Fe3+ + H2O M Fe(OH)2+ + H+
þ
Fe3þ þ 2H2 O $ FeðOHÞþ
2 þ 2H
4þ
3þ
2Fe þ 2H2 O $ Fe2 ðOHÞ2 þ 2Hþ
Fe3þ þ 3H2 O $ FeðOHÞ03 þ 3Hþ
þ
Fe3þ þ 4H2 O $ FeðOHÞ
4 þ 4H
þ
3Fe3þ þ 4H2 O $ Fe3 ðOHÞ5þ
þ
4H
4
3+
Fe + 3H2O M (am) Fe(OH)3(S) + 3H+
þ
Fe3þ þ SO2
4 $ FeSO4
2
3þ
Fe þ 2SO4 $ FeðSO4 Þ
2
Fe3+ + Cl M FeCl2+
Fe3þ þ 2Cl $ FeClþ
2
Fe3þ þ 3Cl $ FeCl03
9.50
20.60
32.00
2.00
4.38
10.89
12.90
23.79
2.25
0.90
3.05
6.31
2.91
13.80
22.7
5.77
3.20
4.04
5.38
1.48
2.13
1.13
338
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
Alka is an abbreviation for alkalinity, which is in mol/L. Note that the model is
developed based on 0.38 V and 25 C. These are reference conditions for waters in
the pilot study. The chemical reactions involved in the model development are summarized in Table 5. According to the ORP values measured in field, Fe2+ exceeded
Fe3+. Thus ferric species were ignored in the model development.
The siderite model was used to predict the iron release for the expected distribution system water quality. Fig. 9 shows the predicted and actual iron concentrations
as a function of pH and alkalinity for a two and five day HRT. The actual total iron
release exceeds the predicted total iron release and the actual total iron release for
five day HRT exceeds the total iron release for a two day HRT, as predicted by
Fig. 9. Actual and predicted total and dissolved iron versus pH and alkalinity using the siderite model.
Data taken from hybrid pilot distribution lines receiving blended waters over one year.
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
1.0
Total Iron
339
Predicted
Iron (mg/L)
0.8
0.6
0.4
0.2
0.0
0
50
100
150
200
250
Alkalinity (mg/L as CaCO3)
Fig. 10. Predicted and actual total iron versus alkalinity.
Mutoti et al. model [26]. As ninety percent or more of the average total released iron
was in the particulate form, the equilibrium model would not be expected to accurately predict total iron release. Dissolved iron release as predicted by siderite model
exceeded the actual release of dissolved iron but was less than actual total iron
release. Although the siderite model did not consider particulate release, the siderite
model predicts iron will decrease as pH and alkalinity increase, which was observed
with the actual data and illustrates the utility of equilibrium models for identification
of trends. This trend is also clearly observed with actual five day HRT data with
alkalinity as shown in Fig. 10.
Other equilibrium models were developed assuming Fe(OH)2 or Fe(OH)3 as the
controlling solid phase. The predicted values of Fe(OH)2 model were much higher
(1000 times) than actual TBW pilot total iron data [2]. The Fe(OH)3 model grossly
under-predicts iron release and the predicted values are 200,000 times lower than actual TBW pilot dissolved iron data. Thus, the FeCO3 model produced the least error
was the best equilibrium model for predicting total iron, which indicates that total
iron release was significantly affected by the FeCO3 solid phase as has been suggested
by others [12]. However, these results and other references associated with this project [26,27] show that total iron release was primarily particulates and was most
accurately described by empirical models based on pipe water quality, geometry,
hydraulics and time.
4. Conclusions
Primary inferences from this study are as follows:
• A pilot pipe distribution study conducted over one year under various blended
waters (i.e. blends of ground, surface and saline waters) showed that iron
340
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
release from aged pipes varied with blended water qualities as well as aged pipe
materials. Specifically, total iron concentration increased significantly when aged
pipes were exposed to water quality that was different from historical groundwater, such as treated surface water and desalinated seawater water quality. Aged
pipe materials also affected iron release: the order of aged pipe material on
total iron release was unlined cast iron > galvanized steel lined ductile iron >
PVC, which demonstrated the importance of pipe surface characteristics on iron
release.
• Based on XRD and RMS results, FeCO3, a-FeOOH, b-FeOOH, c-Fe2O3, Fe3O4
were identified as primary corrosion products of unlined cast iron pipes. Similar
solid phases were identified for galvanized steel pipes except that no FeCO3 was
present. Significant amounts of zinc oxides (e.g. ZnO) were detected in corrosion
scales of galvanized steel. No or very small iron corrosion deposits were found on
the internal surface of aged lined ductile iron and PVC pipes.
• No significant FeCO3 in the corrosion scales of unlined cast iron pipes was
detected by XPS analysis because FeCO3 is very unstable and would transform
into Fe2O3 on contact with air. XPS showed iron release decreased as less soluble
Fe3O4 increased in the corrosion scales, and that iron released increased as the
percentage of the more Fe2O3 increased in scale. Dissolved iron concentrations
were much higher than the predicted by Fe2O3 solubility, suggesting that iron
released in distribution systems is controlled by the solubility of FeCO3 which
could transform into Fe2O3 when exposed to air.
• A thermodynamic siderite (FeCO3) model was developed assuming FeCO3 is the
controlling solid phase for dissolved iron. The siderite model as any equilibrium
model does not consider particulate iron, which accounted for more than ninety
percent of total iron and did not accurately predict total iron release. However,
the model did predict that total iron release decreased as alkalinity and pH
decreased as observed in pilot study, which illustrated the utility of equilibrium
models for identification of trends.
Acknowledgments
Support for this research was provided by Tampa Bay Water (TBW), and
AWWA Research Foundation (AwwaRF). The authors specially acknowledge
Roy Martinez, AwwaRF Senior Account Officer, who was the Project Officer, and
Chris Owen, TBW Quality Assurance Officer. The TBW Member Governments:
Pinellas County, Hillsborough County, Pasco County, Tampa, St. Petersburg, and
New Port Richey; and the AwwaRF Project Advisory Committee are recognized
for their review and recommendations. Pick Talley, Robert Powell, Dennis Marshall
and Oz Wiesner from Pinellas County, and Dr. Luke Mulford from Hillsborough
County are also specifically recognized for their contributions. UCF Environmental
Engineering graduate students, especially Jorge Arevalo and faculty who worked on
this project are recognized for their efforts.
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
341
References
[1] J.D. Dietz, J.S. Taylor, L.A. Mulford, C.J. Cullen, C.A. Owen, Assessment of source water blends on
distribution system water quality, in: Proceeding of Water Quality Technology Conference, American
Water Works Association, 2002.
[2] J.S. Taylor, J.J. Dietz, S.K. Hong, A.A. Randall, Distribution system water quality following
blending surface, ground and saline sources, in: Proceeding of ACE Conference, 2003.
[3] G. Butler, H.C.K. Joon, Corrosion and Its Prevention in Waters, Leonard-Hill, London, 1966.
[4] O.H. Tuovinen, K.S. Button, A. Vuorinen, L. Carlson, D.M. Mair, L.A. Yut, Bacterial, chemical,
and mineralogical characteristics of tubercles in distribution pipelines, J. AWWA 72 (1980) 626.
[5] S. Price (Camp Dresser & McKee, Inc.), T.F. Jefferson, Corrosion control strategies for changing
water supplies in Tucson Arizona, J. NEWWA 111 (1997) 285.
[6] L.J. Hem, E.A. Vik, A. Bjornson-Langen, Water treatment to reduce internal corrosion in the
drinking water distribution system in Oslo, Water Sci. Technol.: Water Supply 1 (2001) 91.
[7] K.E. Shull (Philadelphia Suburban Water Co, Bryn Mawr, Pa), Experimental approach to corrosion
control, J. AWWA 72 (1980) 280.
[8] S.J. Oh, S.-J. Kwon, J.-Y. Lee, J.-Y. Yoo, W.-Y. Choo, Oxidation of Fe2+ ions in sulfate- and
chloride-containing aqueous medium, Corrosion 58 (2002) 498.
[9] L.S. McNeill, M. Edwards, Iron pipe corrosion in distribution systems, J. AWWA 93 (2001) 88.
[10] Internal Corrosion of Water Distribution Systems, second ed., AWWA Research Foundation,
DVGW-Technologiezentrum Wasser, 1996.
[11] A. Kuch, Investigations of the reduction and re-oxidation kinetics of iron (III) oxide scales formed in
waters, Corros. Sci. 28 (1988) 221.
[12] H. Sontheimer, W. Kolle, V.L. Snoeyink, Siderite model of the formation of corrosion-resistant
scales, J. AWWA 73 (1981) 572.
[13] S. Seal, K. Sapre, A. Kale, V. Desai, M. Gopal, P. Jepson, Effect of multiphase flow on corrosion of
C-steel in presence of inhibitor: A surface morphological and chemical study, Corros. Sci. 42 (2000)
1623.
[14] J.F. Marco, M. Gracia, J.R. Gancedo, M.A. Martin-Luengo, G. Joseph, Characterization of the
corrosion products formed on carbon steel after exposure to the open atmosphere in the Antarctic and
Easter Island, Corros. Sci. 42 (2000) 753.
[15] J.J.S. Rodriguez, F.J.S. Hernandez, J.E.G. Gonzalez, XRD and SEM studies of the layer of
corrosion products for carbon steel in various different environments in the province of Las Palmas
(The Canary Islands, Spain), Corros. Sci. 44 (2002) 2425.
[16] Ph. Refait, S.H. Drissi, J. Pytkiewicz, J.-M.R. Genin, Anionic species competition in iron aqueous
corrosion: role of various Green Rust compounds, Corros. Sci. 39 (1997) 1699.
[17] J.K. Heuer, J.F. Stubbins, An XPS characterization of FeCO3 films from CO2 corrosion, Corros. Sci.
41 (1999) 1231.
[18] P. Dillmann, R. Balasubramaniam, G. Beranger, Corros. Sci. 44 (2002) 2231.
[19] Z. Wang, R.C. Moore, A.R. Felmy, M.J. Mason, R.K. Kukkadapu, A study of the corrosion
products of mild steel in high ionic strength brines, Waste Manage. 21 (2001) 335.
[20] J. Lin, M. Ellaway, R. Adrien, Study of corrosion material accumulated on the inner wall of steel
water pipe, Corros. Sci. 43 (2001) 2065.
[21] P. Sarin, V.L. Snoeyink, J. Bebee, W.M. Kriven, J.A. Clement, Physico-chemical characteristics of
corrosion scales in old iron pipes, Water Res. 35 (2001) 2961.
[22] S. Haupt, H.-H. Strehblow, Corrosion, layer formation, and oxide reduction of passive iron in
alkaline solution: a combined electrochemical and surface analytic study, Langmuir 3 (1986) 873.
[23] G. Cai, et al., Study of the reaction mechanism and composition analysis of multiple-component
mixed minerals DTA/T/EGD/GC on-line coupled simultaneous technique, J. Thermal Anal. 45 (1995)
167.
[24] V.L. Snoeyink, D. Jenkins, Water Chemistry, John Wiley & Sons Inc., 1980.
[25] M.M. Benjamin, Water Chemistry, McGraw-Hill Science/Engineering/Math, 2001.
342
Z. Tang et al. / Corrosion Science 48 (2006) 322–342
[26] J.S. Taylor, J.D. Dietz, A.A. Randall, S.K. Hong, C.D. Norris, L.A. Mulford, J.M. Arevalo, S.
Imran, M. Le Puil, S. Liu, I. Mutoti, J. Tang, W. Xiao, C. Cullen, R. Heaviside, A. Mehta, M. Patel,
F. Vasquez, D. Webb, Effects of blending on distribution system water quality, in press, AwwaRF
Final Report, AwwaRF, Denver CO, 2005.
[27] J.D. Dietz, Iron release, controlling distribution water quality in an changing environment, AwwaRF
Technology Transfer Workshop, Tampa, Florida, May 13–14, 2003.