Science of the Total Environment 649 (2019) 739–748

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Corrosion of reinforcing steel in concrete sewers

Yarong Song a,b, Elaine Wightman c, Yimei Tian a, Kevin Jack d, Xuan Li b, Huiyun Zhong b,
Philip L. Bond b, Zhiguo Yuan b, Guangming Jiang b,⁎
a
School of Environmental Science and Engineering, Tianjin University, Tianjin 300350, China
b
Advanced Water Management Centre, The University of Queensland, St. Lucia, QLD 4072, Australia
c
Sustainable Minerals Institute, Julius Kruttschnitt Mineral Research Centre, The University of Queensland, Indooroopilly, QLD 4068, Australia
d
Center for Microscopy and Microanalysis, The University of Queensland, St. Lucia, QLD 4072, Australia

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• A first thorough investigation of rebar

corrosion in sewers
• Rusts at steel/concrete interface charac-
terized by mineral analytical techniques
• H2S and sulfuric acid are not directly in-
volved in sewer rebar corrosion.
• Chloride initiates rebar corrosion in
sewers and forms a chloride-enriched
layer.
• A conceptual model about corrosion
processes of sewer rebar is established.

a r t i c l e i n f o a b s t r a c t

Article history: Hydrogen sulfide is a controlling factor for concrete corrosion in sewers, although its impact on sewer rebar cor-
Received 14 June 2018 rosion has not been investigated to date. This study determined the corrosion mechanism of rebar in sewers by
Received in revised form 25 August 2018 elucidating the roles of chloride ions, apart from the effects of hydrogen sulfide and biogenic sulfuric acid. The na-
Accepted 25 August 2018
ture and distribution of rusts at the steel/concrete interface were delineated using the advanced mineral analyt-
Available online 27 August 2018
ical techniques, including mineral liberation analysis and micro X-ray diffraction which is the first-ever use in
Editor: Zhen (Jason) He such studies. The corrosion products were found to be mainly iron oxides or oxy-hydroxides. H2S and biogenic
sulfuric acid did not directly participate in the product formation of steel partly covered by concrete or directly
Keywords: exposed to sewer atmosphere. Instead, chloride ions played an important role in initiating steel corrosion in
Sewer sewers, supported by a thin chloride-enriched layer at the steel/rust interface. Away from the chloride-
Corrosion enriched layer, iron oxides accumulated on both sides of the mill-scale to form a corrosion layer and corrosion-
Rebar filled paste respectively. The corrosion layer around rebar circumference was non-uniform and the rust thickness
Chloride with respect to polar coordinates followed a Gaussian model. These findings support predictions of sewer service
Hydrogen sulfide
lifetime and developments of corrosion prevention strategies.
Mineral liberation analysis
© 2018 Elsevier B.V. All rights reserved.

1. Introduction

Sewer networks are essential urban infrastructures for collecting

⁎ Corresponding author. and safely transporting wastewater to treatment facilities. Hydrogen
E-mail address: [email protected] (G. Jiang). sulfide (H2S), which is produced in sewers especially under anaerobic

https://doi.org/10.1016/j.scitotenv.2018.08.362
0048-9697/© 2018 Elsevier B.V. All rights reserved.
740 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748

conditions, is well recognized for shortening the service lifetime of these been made to identify the reactions. Other studies on H2S-induced
installations due to the microbial-induced concrete corrosion (Lin et al., steel corrosion usually employ accelerated tests using high voltages or
2017; Parker, 1947; Pikaar et al., 2014). The corrosion causes loss of con- excess H2S concentrations (Idriss et al., 2001; Parande et al., 2006), of
crete mass, cracking of concrete cover and ultimately, the structural col- which the conditions would be quite different from those in real sewers,
lapse. The rehabilitation and replacement of corroded sewers is such as H2S concentrations, relatively humidity, pH, and temperature.
estimated to cost several billion dollars annually worldwide (Sterling When steel corrosion is initiated, corrosion products at the steel/
et al., 2010), and the cost is expected to increase as the aging sewers concrete interface are found to cause expansive stresses on the sur-
continue to fail (EPA, 1991). rounding concrete by obtaining around 2–6 times the volume of the
Concrete pipes used in wastewater systems can be of plain concrete original steel (EPA, 1991; Lu et al., 2018). Consequently, the main struc-
or steel reinforced. Usually, for pipes with diameters larger than 0.60 m, tural damage is not the loss of steel cross-sections, but the concrete
it is common to have a sufficient number of steel bars designed to with- cracking. A two-stage model is proposed for the steel corrosion-
stand a combination of internal pressure and external loads (Nnadi and induced concrete cracking which involves: (1) rust growing, migrating
Lizarazo-Marriaga, 2012; Grengg et al., 2018; Joshi, 2015). A recent and filling the porous zone around the steel/concrete interface, and
study commented that reinforced concrete pipes, with wider ranges of (2) rust causing expansive pressure on the surrounding concrete
sizes (12–72 in) and relatively low corrosion rates compared to metallic (Wong et al., 2010; Zhao et al., 2013; Zhao et al., 2012b). It was reported
pipes, are prevalent among ductile iron, vitrified clay, polyvinyl- that the rust migration is greatly increased in the presence of chloride
chloride (PVC), composite fiber reinforced polymer (FRP) and high den- ions. However, in sewers affluent in H2S, whether the chloride-
sity polyethylene (HDPE) pipes (Vahidi et al., 2016). Corrosion of con- facilitated rust migration occurs is largely unknown. To better under-
crete sewers involves a range of abiotic and biotic processes. The stand correlations between the steel corrosion and concrete cracking
process proceed by initial acidification by carbonate and sulfide and in sewers, a more detailed characterization of rust distribution at the
then progress to more damaging chemical reactions between the bio- steel/concrete interface is highly required.
genic produced sulfuric acid and the cementitious materials In the present study, a long-term exposure experiment (45 months)
(Noeiaghaei et al., 2017; Parker, 1945a, b; Parker, 1947). As the concrete was conducted to investigate the corrosion mechanism of rebar in
corrosion develops, the reinforcing steel that is originally protected by sewers when the concrete cover was gradually removed due to the
concrete matrix also suffers from possible corrosion. Factors contribut- sulfide-induced concrete corrosion. This was performed using the rein-
ing to steel corrosion could be summarized as follows: the reduced alka- forced concrete samples exposed to simulated sewer conditions in a lab-
linity of surrounding concrete, of which the pH could reach 2–4 due to oratory corrosion chamber with well-controlled conditions of H2S
the biogenic sulfuric acid produced in the corrosion process (Jahani concentrations, relative humidity and temperature. In particular, the
et al., 2001; Jiang et al., 2014b); the ingress of H2S, CO2, oxygen and nature and distribution of corrosion products at the steel/concrete inter-
wastewater through concrete cracks to rebar surface (Parande et al., face were delineated using various advanced mineral analytical tech-
2006); the direct exposure of rebar by regional concrete scaling or niques and modeling. As a first thorough investigation of rebar
spalling that may occur in the later stages of service life (Fandrich corrosion in sewers, this study provides significant improvement to
et al., 2007). These factors can enhance steel corrosion in a short time the understanding of the corrosion processes in concrete sewer
period. However, most studies on sewer deterioration to date concen- systems.
trate only on the concrete corrosion (Islander et al., 1991; Joseph et al.,
2012; Parker, 1945a, b). There is poor understanding of the fundamental
aspects of reinforcing steel bar corrosion in the sewer environment. The 2. Material and methods
corrosion mechanism, i.e. the detailed reactions involved, and the na-
ture and distribution of corrosion products, remains to be elucidated. 2.1. Reinforced concrete coupons
In spite of this lacking knowledge of the sewer rebar corrosion, there
are many studies reporting rebar corrosion as a major cause of durabil- The concrete coupons were prepared using corroded reinforced con-
ity problems in other reinforced concrete structures such as concrete crete obtained from a main trunk sewer of Sydney Water Corporation,
buildings, bridge decks, marine structures, slabs and floors (Lu et al., Australia. Generally, those pre-corroded concrete consisted of quartz
2018). The electrochemical process causes the dissolution of iron to (aggregates and sand particles) and calcium silicate accounting for
form various corrosion products, such as iron oxides, hydroxides and around 13.1% and 22.2%, respectively. Details of the coupon preparation
hydrated oxides, the nature of the products being dependent on the ex- were described elsewhere (Jiang et al., 2014a; Jiang et al., 2015b; Jiang
posure environment (Bertolini et al., 2013; Lu et al., 2018). In these et al., 2014b). The coupons were cut to a size that was approximately
studies, chloride ions are considered as one of the major causes of the 100 mm (length) × 70 mm (width) × 70 mm (thickness). Each coupon
steel corrosion (Sanchez-Silva and Rosowsky, 2008). The dissolved contained a section of plain round steel bars with a nominal diameter of
chlorides from the external environment could permeate through 12 mm, and the concrete cover above the steel bar surface was between
sound concrete or cracks to the steel surface thus inducing corrosion 5 and 20 mm. There was no evidence of existing rebar corrosion in the
of the steel bar (Grengg et al., 2018). In sewers, however, gaseous H2S concrete coupons.
and the biogenic sulfuric acid, produced in corrosion biofilms (Parker, After cutting, the coupons were washed in fresh water to remove the
1945a, b), are commonly accepted to be the corrosive substances rather existing 2–4 mm corrosion layer of concrete and any surface contamina-
than chlorides (Fernandes et al., 2012). In the active corrosion regions of tion. Washed coupons were immediately dried in an oven (Thermotec
sewers, such as the pipe crown area (Jiang et al., 2014a; Jiang et al., 2000, Contherm) at 60 °C for 3 days to achieve a stable initial water con-
2015b), the roles and interactions of chloride, H2S and sulfuric acid in tent and then were embedded in stainless steel frames using epoxy (FGI
the rebar corrosion process are unknown. R180 epoxy & H180 hardener). The stainless steel frame had outside di-
It is reported that the direct contact with H2S can encourage steel mensions of 230 mm (length) × 110 mm (width) × 60 mm (thickness).
corrosion to a level of 12.8 mm/year (Tiller, 1990), and sulfides are in- After concrete coupons were placed inside the frames, the epoxy resin
tensive corrosion stimulators (Heitz et al., 1996). However, steel corro- was injected until the level of the epoxy resin reached the upper rim
sion in concrete is likely more complicated due to impediment of H2S of the frames. Once the resin was hardened, a flat surface would form
diffusion, the protective and heterogeneous concrete layers, which are at the surface of the epoxy resin and the upper rim of the frames. Be-
subjected to degradation themselves. Although a study (Sanchez-Silva cause of the thickness difference between the coupon and the stainless
and Rosowsky, 2008) proposes that H2S produced in the sewer could steel frame, some blank gaps would appear between the resin and the
diffuse to the metal surface and form FeS, no rigorous attempts have coupon.
 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748 741

2.2. Corrosion chamber and exposure conditions surface (Fig. S2) to produce a series of 10 mm-thick cross-sectional
slices, until no evident corrosion could be observed at the steel/concrete
To achieve controlled environments simulating those of real sewers, interface. Noting that the rebar corrosion in Coupon A was not so exten-
corrosion chambers were constructed using glass panels of 4 mm thick- sive, only the front slice was prepared (Fig. S2a). In the front slice of
ness. The chamber design and operation details were described previ- Coupon A, the rebar cross-section was located above the upper rim of
ously (Joseph et al., 2012). In brief, the gas tight chambers had the stainless steel frame and the surface of the epoxy resin, thus was ex-
dimensions of 550 mm (L) × 450 mm (D) × 250 mm (H) (Fig. S1), posed directly to the gas and corroded severely (Fig. S2a). For Coupon B,
and contained 2.5 L of domestic sewage (Table 1) that was collected 7 slices were obtained (Fig. S3).
from a local sewer pumping station and replaced every two weeks. To characterize the rust composition and distribution for distinct
The analytical methods for the sewage parameters were presented in rebar environments when there was a slight concrete cover or the
Table SI-1. The coupons were exposed to the gas phase within the rebar was completely exposed, the front slice of Coupon A and three
chambers with the exposed coupon surface facing downwards, about slices of Coupon B (Fig. S3) were selected. Then these four slices were
110 mm above the sewage surface. This arrangement simulated the po- further cut into smaller sections (30 mm (length) × 40 mm (width)
sition of the sewer pipe crown, a location highly susceptible to sulfide × 10 mm (thickness)) to extract the rebar while keeping the surround-
induced corrosion (Mori et al., 1992; Vollertsen et al., 2008). ing concrete intact (Fig. S4). In particular, Sections 2 and 3 were ob-
Three key environmental factors of sewer corrosion, gaseous H2S tained from the regions closer to the rear of Coupon B where the steel
concentration, temperature and relative humidity, were controlled in was covered by a cracking concrete layer (3–4 mm in thickness),
the corrosion chamber (the oxygen concentration was relatively con- while Section 4 was obtained closer to the front where the rebar was ex-
stant at a level similar to that in air). To avoid extreme conditions and posed directly.
artificial acceleration, 5 ppm of gaseous H2S was selected as the expo- For the Mineral Liberation Analyzer (MLA), Energy Dispersive X-ray
sure condition, which is also quite typical in real gravity sewers Spectroscopy (EDS) and Micro X-ray Diffraction (Micro-XRD) analyses,
(Cayford et al., 2017; Li et al., 2017). The constant H2S gaseous concen- the four sections above were further ground, polished and trimmed to
tration (5 ppm) was achieved by injecting Na2S solution into a HCl acid smaller blocks, with the rebar, steel/concrete interface and surrounding
solution (13% HCl), controlled by a programmable logic controller, and concrete preserved. The low-viscosity epoxy resin was used for casting
this was monitored using a H2S gas detector (Oda-Log Type2) with a to minimize any artificial damage that may be incurred during the treat-
range between 0 and 200 ppm (App-Tek International Pty Ltd., ment. Additionally, Section 1 was further cut along the rebar axis
Brendale, Australia). The temperature of the corrosion chamber was (Fig. S5), to investigate the depth profiles of mineral structures and
controlled at 20 °C by installing the chamber into a temperature con- rust distributions. A fixed diamond lap and the 600 grit silicon carbide
trolled cabinet. The sewage temperature (23 °C) was controlled to be were used for grinding. Polishing was carried out using 3 μm diamond
slightly warmer than the gas phase by re-circulating the temperature- slurry (with an aliphatic hydrocarbon base as a suspension media) on
controlled water through glass tubes immersed in sewage. The relative a ceramic lap, and 1 μm diamond slurry with a textile cloth. Prior to
humidity was controlled at approximately 100%. Two resistance tem- the mineral analysis, sections were also coated with a layer of carbon
perature detector (RTD) probes were used to monitor humidity, while approximately 25 nm thick, using a JEOL JEE-420 vacuum evaporator.
another RTD probe was employed to monitor the temperature of A copper standard was used to determine the thickness of the carbon
sewage. coating.
After 45 months of exposure, significant concrete corrosion products
were observed, together with rust accumulation on the exposed steel 2.4. Visual inspection and X-ray tomography
bar of which the concrete cover was completely corroded away. Then
two representative coupons, one with approximately 10 mm concrete The macro-level corrosion morphology of each section was
cover left above steel bar (Coupon A) and another with steel exposed di- photographed for visual inspections. Section 1 was also observed by
rectly (Coupon B), were retrieved from the chamber (Fig. S2). The com- using an optical microscope (Leica DMS, Germany) connected to a dig-
parison of the coupons before and after the exposure showed that ital camera. Further, a three-dimensional reconstruction of the rust
1–5 mm of the concrete cover was corroded away, thus the rebar in layer above steel bar was established by X-ray tomography (Skyscan
Coupon B with relatively thin concrete cover (5 mm) before exposure 1172, Bruker Instruments), which is produced by collecting multiple
was exposed eventually. However, the rebar in Coupon B was still par- images at a series of steps through a full rotation. Details of the X-ray to-
tially covered by a thin cracking concrete layer (3–4 mm) shown at mography was described previously (Jiang et al., 2014a; Jiang et al.,
the top-right area (Fig. S2), which was used for investigating the initia- 2015b; Jiang et al., 2014b).
tion of rebar corrosion with the progression of concrete corrosion.
2.5. MLA, EDS and Micro-XRD
2.3. Preparation of coupon sections
To determine the minerals and their depth distribution on the cross-
Thin slices of coupons were prepared using an automatic abrasive sections, advanced mineral techniques involving MLA, EDS and Micro-
cutter and diamond cut-off blade, designed for hard brittle materials, XRD were applied. MLA combines backscattered electron (BSE) inten-
and cutting was performed at a low speed of b0.3 mm/s. Each coupon sity and X-ray analysis to identify minerals present in particles in the
was sectioned perpendicularly to the rebar axis from the front side polished section. The BSE image is used to identify individual particles
and define boundaries of mineral phases; then the X-ray collected
Table 1 from each segmented phase was used to determine the mineral or
Typical characteristics of the domestic sewage. phase based on the chemical composition. The MLA measurement en-
ables a comprehensive analysis of the mineral compositions and distri-
Parameters Value Parameters Value
bution. It results as a BSE image, a false-color classified particle map and
pH 6.9–7.0 Total suspended solids 145–545
a database containing adequate particle and grain information, such as
(mg/L)
Chemical oxygen demand 350–850 Total phosphorus (mg P/L) 10–15 modal mineralogy, elemental deportment, mineral grain size and asso-
(mg/L) ciation and mineral liberation.
Dissolved oxygen (mg/L) 1–3 Sulfate (mg S/L) 10–20 EDS measurement was carried out on the coupon sections used for
Total Kjeldahl nitrogen (mg/L) 65 Dissolved sulfide (mg S/L) 15–25 MLA analysis. The EDAX software (EDAX, AMETEK Inc) was utilized, at
Temperature (°C) 20–22
a frame resolution of 1024 × 800, with a dwell time of 200 s/frame, to
742 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748

collect 16 frames for each region of interest. Each image related to the For Sections 2–3, thin light-yellow corroded concrete layers
intensity of the Kα peak of the element selected over the region of inter- (1–2 mm) with uniform corrosion front were observed (Fig. S4), indi-
est with intensity related to the concentration of the element. cating that the concrete corrosion progressed uniformly (Jiang et al.,
Micro-XRD was employed to further differentiate the crystallized 2014b). When the cover-penetrated cracks occurred in the concrete
compositions of rust at selected spots. A Rigaku SmartLab micro diffrac- matrix (Section 2) or the acidic concrete corrosion front reached the
tion system with a rotating Cu anode (9 kW) and micro-focus optics was steel surface (Section 3), significant rust accumulation was observed.
used to carry out point mineralogical analysis on the polished cross- The rusts at the steel/concrete interface were not uniformly distributed
sectional surface, especially the rust at the steel/concrete interface. X- around the steel bar, but accumulated mainly in the region that was
rays were generated at 45 kV and 200 mA with 0.798 s/step as scanning nearer to the exposed coupon surface. It confirmed that rebar corrosion
speed, 4499 counts as scanning step, and 2Θ = 20–110° as scanning is spatially variable due to the non-uniform distance to the exposed con-
range. The diameters of collimators for sampling spots were 100 μm crete surface (Zhao et al., 2012b). Besides, no serious rust accumulation
for Section 1, and 50 μm for Sections 2–4, due to differences in the thick- was observed in Section 4 possibly due to the rust dissolution induced
ness of the rust layer (50–500 μm) (Fig. 2). Diffraction patterns were by the biogenic sulfuric acid.
processed using the Rigaku PDXL2 data processing software (version
2.3.1.0). Mineralogical phase identifications were carried out using 3.2. Composition and distribution of corrosion products
Gade software (version 5) and pdf-2 crystal structure database.
3.2.1. Mineral compositions of the rebar-concrete section
2.6. Measurements and modeling of the corrosion layer thickness MLA mapping of Section 1 (Fig. S5), with the top edge as the ex-
posed surface, shows distinguished boundaries of the intact con-
MLA allowed for the accurate measurement of the rebar corrosion crete, the corroded concrete layer, the intact rebar and its rust
layer thickness. The thickness of the corrosion layer for Sections 2–3 layer (Fig. 1). The thickness of the concrete corrosion products
was measured at 16 locations around the perimeter of the rebar. How- was only 0.25 mm. This low level of corrosion most likely resulted
ever, no rust thickness was determined for Section 4 due to the rust dis- from the low H2 S concentration at 5 ppm and also that it was the
solution that was evident. Also, the area of the corrosion layer for freshly cut surface at the time of the coupon preparation
Sections 2–4 was measured. Then the loss of rebar cross-section area (Section 2.1). In contrast, a much thicker rust layer was observed
was calculated by dividing the corrosion layer area by the volume ex- above the steel bar. The cracks and voids within the rust layer
pansion ratio of rust, which was assumed to be 2 in this study (Lu were evident, consistent with the observations by X-ray tomogra-
et al., 2018). The degree of rebar corrosion was expressed as n, i.e. the phy (Fig. S7).
percentage of the original area of the rebar cross-section. The weight percentages of the mineral compounds were deter-
To further investigate the non-uniform rust distribution around the mined for the rebar-concrete section above (Fig. S8). The rebar
circumference of rebar, polar coordinates were established. To focus rusts were mainly consisted of various iron oxides, and iron chlo-
on the rust distribution, the maximum rust thickness was retained at a rides (0.2%) were scattered along the steel/rust interface (Fig. 1). Ex-
fixed position, i.e. θ = π (Fig. S17). The model based on a Gaussian func- cept for the large proportion of the steel bar (66.5%), the section
tion (Zhao et al., 2011b) was used to describe the variations in the rust were dominated by the intact concrete majorly composed of calcium
thickness: silicate (9.7%) as the primary hydrated cement, and quartz (8.3%) as
aggregates and sand particles. However, gypsum and ettringite,
 2 well-known as the major corrosion products of concrete sewers
a1 ffi2a
− pθ−π (Jiang et al., 2014b; Parande et al., 2006; Parker, 1945a, b), were
T r ¼ pffiffiffiffiffiffi e 2 þ a3 ð1Þ only at 0.08% and 0.03% respectively. Carbonate was determined to
a2 2π
be 0.6% and that implies significant carbonation which would agree
with this being at the onset of concrete corrosion (Jiang et al.,
where Tr is the rust thickness at coordinate θ (excluding the thickness of 2014b; Sun et al., 2016).
the mill-scale delineated in Section 3.4); a1, a2 and a3 are fitting param-
eters, i.e. the non-uniform coefficient, the spread coefficient, and the 3.2.2. Compositions of rebar corrosion products
uniform coefficient of the rust layer, respectively. The rust compositions for the four sections were determined using
Micro-XRD spot analysis. To our knowledge, this is the first time that
3. Results and discussion Micro-XRD, accurate for the identification of localized compositions
(Chitty et al., 2005), has been employed to characterize such thin corro-
3.1. Visual inspection and X-ray tomography of the coupons sion layers formed in sewers. Typically, studies are confined to elemen-
tal analysis using EDS (Jiang et al., 2014b; Wong et al., 2010; Zhao et al.,
The rebar corrosion in Coupon A was not so extensive, so that only 2013; Zhao et al., 2012b) or X-Ray powder diffraction which destroys
the exposed rebar cross-section at the front of the coupon accumulated the layers during the sample preparation. Based on the MLA and BSE im-
the thick rust. Likely, this less extensive steel corrosion inside of the con- ages, six sampling spots for Section 1 (Fig. S9) and 2–3 spots for Sections
crete was attributed to the concrete cover that had remained and the big 2–4 (Fig. 2) were selected for spot analysis. The results showed no iron
aggregates that inhibited the ingress of corrosive agents. The visual in- sulfide or sulfate in the rusts; the main products were iron oxides or
spection showed that the major color of rusts in Section 1 was a dark oxy-hydroxides, which are typical of corrosion of constructional rein-
brownish-red dotted with a brighter red-hue, indicating the formation forcing steel bars (Wong et al., 2010). Also, higher contents of iron chlo-
of iron oxides or hydroxides (Fig. S6). X-ray tomography shadow im- rides were observed at the steel/rust interface, confirming the findings
ages demonstrated that cracks within the rust propagated as the rust of the MLA analysis (Fig. 1).
developed (Fig. S7), likely resulting from the rust expansion which is re- Three typical regions in the rust layers can be distinguished based on
ported to be 2–6 times the volume of the original steel (Zhao et al., the rust compositions. These regions are the steel/rust interface, the in-
2011). Particularly, an evident crack, 0.3–0.5 mm in width, was ob- tact rust area and the cracking or external rust layer. Rusts at the steel/
served at the rust surface (Fig. S6). This crack seemed to cause the asym- rust interface of Section 1 (points 2–3) were primarily composed of Fe,
metric rust migration into the adjacent concrete cover. A previous study Fe3O4, FeCl2 and its hydrated compounds. The formation of Fe3O4 was
concludes that cracking is an inevitable step of rust propagation (Zhao likely resulting from the restricting surrounding concrete and the low
et al., 2012a). availability of oxygen. The FeCl2 products corresponded well to the
 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748 743

Fig. 1. MLA mapping of primary mineral components on the 25 mm × 10 mm vertical rebar-concrete surface of Section 1.

Fig. 2. BSE images of the rust layers and the representative Micro-XRD spot analysis for Section 1 (a), Section 2 (b), Section 3 (c) and Section 4 (d). For Section 1, other spot analyses (points
1, 2 and 5) were shown in Fig. S9.
744 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748

chloride-enriched layer detected by MLA mapping. The relatively less 3.3. Microstructure analysis at the steel/concrete interface
evident peaks of FeCl2 might be attributed to the extremely thin FeCl2
layer, which could diminish the intensity of X-ray signal needed for Microstructure analysis at the steel/concrete interface is essential to
the XRD analysis. However, this data and the results from MLA and understand the development of rebar corrosion with the progression of
EDS measurements (Figs. 1, 3 and 4) together would make the FeCl2 concrete degradation. MLA mappings of Sections 2–4 with different
layer more convincing. In the region of relatively intact rusts (point 4), depths of concrete cover were determined (Fig. 3). Similar rust distribu-
the formation of Fe3O4, α‑FeOOH and γ‑FeOOH indicated further oxida- tions were observed in Section 2 and 3; Section 4 showed less rust on
tion that might be related to the rust cracking characteristics. As higher the exposed rebar surface, although at this point metal loss due to cor-
oxygen levels would exist at the more external regions of points 5–6, rosion was evident. To further visualize the elemental compositions
further dehydration and crystallization would occur leading to the for- and rust distribution, BSE and EDS mapping were performed on the se-
mation of γ‑Fe2O3. Similar layered rust structures are detected in an- lected areas (Fig. 3), i.e. areas 1, 2 and 3 of Section 2 (Fig. S10, Fig. 4,
cient reinforced buildings using Micro-XRD and Raman techniques Fig. S11, respectively) and other areas of Sections 3–4 (Figs. S12–16).
(Chitty et al., 2005). Collectively, two types of interfacial conditions, i.e. non-corroded and
The results for Sections 2–4 were fairly consistent, with FeCl2, Fe3O4 corroded, were identified.
and γ‑Fe2O3 being the main rust components. However, the rust layers
located between the steel and concrete were thinner and more con- 3.3.1. Non-corroded steel/concrete interface
densed, which was reported as the direct cause of concrete cracking BSE imaging and elemental mapping, including Fe, O, Cl, S, Ca, Si and
due to the expansive pressure exerted on the surrounding concrete Al, were determined at the steel/concrete interface in area 1 (Section 2)
(Zhao et al., 2012a). where no obvious corrosion occurred (Fig. S10). A uniform layer,
termed as mill-scale (MS), was observed with a depth around 30–35
μm. This was similar to the mill-scale average thickness detected in a
previous study (Zhao et al., 2013). A sharp contrast was particularly dis-
tinct for the distributions of Fe, O, Ca, and Si, where Fe was mainly ob-
served in the steel and mill-scale, while O, Ca, and Si were detected in
the concrete matrix. The evident separation implied no formation and
migration of iron rust at the non-corroded steel/concrete interface.

3.3.2. Corroded steel/concrete interface

BSE imaging and EDS mapping in area 2 (Section 2), the corroded re-
gion, showed a layered structure at the steel/concrete interface (Fig. 4).
The mill-scale (MS) was much thinner (10–20 μm), showing signs of
deterioration and further oxidation. Also, it detached away from the
steel surface to the outer edge of the steel/concrete interface, demon-
strating the growth and expansion of the corrosion layer (CL) from
the reaction zone. Indeed, thick rusts were observed between the
rebar and the mill-scale. At the steel/rust interface, a chloride-
enriched layer was evident in EDS mapping (Fig. 4) which was aligned
with the results of MLA and Micro-XRD analyses. Into the corrosion
layer (CL), away from the chloride-enriched layer, the rust was mainly
composed of Fe and O, consistent with the rust compositions described
in Section 3.2.2.
On the other side of the mill-scale (MS), the surrounding concrete
was partially brighter than the concrete matrix farther away (Fig. 4,
BSE image). This suggests an increase in the local mean atomic number
and backscattering coefficient, perhaps due to ingress of a heavy ele-
ment (e.g. Fe) into the paste (Cardoso et al., 2018; Lu et al., 2018). Ac-
cording to the EDS mapping, apart from higher concentrations of Fe in
steel (S), corrosion layer (CL) and mill-scale (MS), relatively high con-
tents of Fe were also detected in this region together with abundant
levels of O. This affected area was called the corrosion-filled paste (CP)
(Wong et al., 2010). EDS mapping also showed that the corrosion-
filled paste was depleted of Ca, indicating the dissolution of calcium hy-
droxide and/or calcium silicate hydrate likely caused by the enhanced
acidification process near the steel surfaces. However, the corrosion-
filled paste was not uniform or ubiquitous for all regions, which might
be influenced by the nature, amount and distribution of aggregates,
and also porosities of cements.
A thicker corrosion layer was evident at area 3 (Section 2) and the
mill-scale was a minor layer (Fig. S11). These characteristics might
have occurred through oxidative conditions by oxygen penetrating the
channels and micro cracks. As the concrete cover was progressively cor-
roded, relatively abundant oxygen could be available at the steel sur-
face. Due to the variation in oxygen availability, different shaded
Fig. 3. MLA mapping of primary mineral components on a 20 mm × 20 mm cross-section bands were evident in the corrosion layer (Fig. S11, BSE image),
of Section 2 (a), Section 3 (b) and Section 4 (c). BSE and EDS mapping were performed on where the lower Fe/O atomic ratios accounted for the dark bands and
the selected areas in red squares. the higher Fe/O ratios referred to the lighter bands.
 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748 745

Fig. 4. BSE image and EDS element mapping of Fe, O, Cl, S, Ca, Si and Al respectively for the area 2 marked with red square in Fig. 3a.

These results agree with the two-stage model implying that the cor- steel/concrete interface. To our knowledge, this is the first report that
rosion layer and corrosion-filled paste form simultaneously (Zhao et al., these layered structures are present in reinforced concrete sewers. Sim-
2013). Interestingly, it is worth to note that although the corrosion layer ilar structures have been found in reinforced concrete specimens ex-
in area 3 was much thicker than that in area 2, less evident corrosion- posed to alternate wetting and drying cycles with chloride spraying
filled paste was observed in area 3. This indicated that the expansive (Zhao et al., 2013).
pressure created by the corrosion layer might even start earlier than
the process of rust penetration into the surrounding concrete. Indeed,
at some area, the corrosion-filled paste was even not evident 3.4. Gaussian model to describe the non-uniform corrosion layer
(Fig. S16). In such, the time to initiate cracking of the covering concrete
might be reduced as the time needed for corrosion products to fill the Polar coordinates were established for Sections 2 and 3 (Fig. 3) for
cement paste becomes shorter. the Gaussian modeling. It was seen that the measured variations of
Collectively, five regions (i.e. steel, corrosion layer, mill-scale, the rust thickness around rebar circumference against the polar coordi-
corrosion-filled paste and concrete) were identified at the corroded nate were fitted fairly well by the Gaussian regression lines (Fig. 5). This

Fig. 5. The measured rust thickness around the rebar perimeter, the regression line of the Gaussian model and predicted lines with 95% confidence interval for Section 2 (a) and Section 3
(b).
746 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748

suggests that the corrosion initiates where the cracks penetrate and concrete through cracks. Indeed, according to Fig. S11, a small part of
then it expands uniformly at both directions on the rebar surface. iron oxide at the outer edge of the corrosion layer seemed to be con-
The values estimated for parameters a1, a2, a3 (Table 2) are all within verted to iron sulfide where H2S could reach through the adjacent
reasonable ranges (Sutrisno et al., 2017; Xi et al., 2018), and the deter- cracks (area 3 in Fig. 3a). Nevertheless, no iron sulfide could be detected
mination coefficient R2 shows values above 0.83, indicating the good by MLA or micro–XRD test, indicating that only a small amount of H2S
fitting of the model. The rebar corrosion values n are also presented could reach the rust layer. Thus H2S might only play a minor role in
(Table 2). The low values of a3 further confirm that the rebar corrosion the steel corrosion compared to oxygen, only when evident cracks
in Sections 2–3 did not develop across the entire rebar surface but oc- occur inside concrete. Additionally, for rebar that loses the concrete
curred mainly on the interface towards the exposed surface of concrete cover and is directly exposed to sewer atmosphere, the absence of sul-
and penetrating cracks. The greater value of a1 for Section 3 than for fide in the rusts might be attributed to the weak acidity of H2S (pKa1
Section 2 suggests that the steel corrosion was less even for Section 3, = 6.9; pKa2 = 11.96, 18 °C) (Ellis and Giggenbach, 1971) and low sol-
of which the rebar corrosion value n was also higher. Additionally, a2 ubility of H2S in the water film. H2S will unlikely be able to diffuse
for Section 3 was also higher, inferring that the rust spread more exten- through the passive layer of iron oxides produced, further limiting its
sively and to a relatively flatter shape. Compared to these covered slices diffusion to the steel surface. Overall, the key corrosion process of
(Sections 2 and 3), the rebar exposed Section 4 was subjected to heavier rebar in concrete sewers likely conforms to the corrosion mechanisms
corrosion demonstrated by higher value of n, probably due to loss of shown in the reactions below:
cracked rust over time.
4Fe þ 3O2 þ 2H2 O→4FeOOH ð4Þ
3.5. The corrosion mechanism of reinforcing steel in sewers
2FeOOH→Fe2 O3 þ H2 O ðhigh oxygenÞ ð5Þ
The sewer environment is corrosive mainly due to the presence of
H2S that gets converted into sulfuric acid in humid conditions (Jiang 2Fe þ O2 þ 2H2 O→2FeðOHÞ2 ð6Þ
et al., 2015a). It is conventionally postulated that the direct contact of
steel with H2S or sulfuric acid induces FeS formation, as shown below 6FeðOHÞ2 þ O2 →2Fe3 O4 þ 6H2 O ðlow oxygenÞ ð7Þ
(Heitz et al., 1996; Sanchez-Silva and Rosowsky, 2008):

4Fe þ SO4 2− þ 4H2 O→3FeðOHÞ2 þ FeS þ 2OH− ð2Þ While the sulfate- or sulfide-induced reactions have not been iden-
tified in the study, a potential role of these reactions cannot be
Fe2þ þ H2 S→FeS þ 2Hþ ð3Þ completely ruled out. There are also possibilities that H2S would act as
a catalyst and doesn't participate directly in the product formation. Fur-
Besides, if the biogenic sulfuric acid penetrates to reach the rebar di- ther studies of concrete rebar corrosion at various H2S concentrations
rectly, the formation of iron sulfate might also occur. However, these re- and also without H2S should be carried out to verify the hypothesis.
actions and corrosion products have not been verified in the real rebar Apart from the iron oxides as main rust components, higher contents
corrosion situation in sewers. of iron chlorides were observed at the steel/rust interface. Although the
Against the traditional speculations, the results above show no evi- chloride-induced steel corrosion is well recognized, the pattern in
dent iron sulfide or sulfate products in the rusts, indicating that H2S which chlorides distribute in the corrosion layer and promote corrosion
and sulfuric acid may not directly participate in the product formation. is unknown. This study for the first time used advanced mineral analyt-
The possible reasons are clarified as follows according to different ical techniques to visualize the accumulation and non-uniform distribu-
rebar corrosion conditions. tion of chloride at the steel/rust interface, suggesting a significant role of
For rebar still covered by intact concrete layers, H2S cannot pene- chloride in rebar corrosion in sewers. The findings support a new model
trate deep easily because it will be oxidized to elemental sulfur and/or for steel corrosion in sewers based on the dissolution, migration and ac-
sulfate (sulfuric acid) by the biofilm present on the concrete, or react cumulation of chloride (Fig. 6).
with the alkaline constituents. The sulfuric acid might then penetrate In the electrochemical reactions, the anode and cathode are located
further, but the alkaline environment inside concrete makes it easy to separately (Fig. 6). The anodic region is the steel surface at the steel/cor-
neutralize the sulfuric acid. Unlike the chloride attack, the acid pene- rosion layer interface, where electrons are produced during the iron ox-
trates and reacts with concrete simultaneously; therefore, the penetra- idation process. The cathodic region is the mill-scale/corrosion-filled
tion away from the affected area is very small. Indeed, it was reported paste interface, especially the concrete porous area where the water
that sulfuric acid and H2S attack can be considered a superficial phe- containing dissolved oxygen is available. The theory that concrete adja-
nomenon, in which the acid attack destroys the concrete cover first cent to the steel/concrete interface acts as the cathodic region has been
(Nnadi and Lizarazo-Marriaga, 2012). In this study, the abundance of frequently reported in literature (Zhao et al., 2013, 2014). The concrete
iron oxides in the rust layer strongly suggests that the rebar corrosion matrix is heterogeneous with air voids and macro pores dispersed in-
in sewers is more likely influenced by the oxygen available. This could side (Song et al., 2016). Once the air voids are occupied by water due
be related to the fact that the oxygen concentration in sewer air (similar to the condensation process (Li et al., 2017), oxygen likely dissolves
to atmosphere) is 3–4 orders higher than that of H2S (typically in the and is utilized as the electron acceptor in the cathodic reactions. Be-
range of a few ppm to hundreds of ppm). This means that the diffusion tween the anodic and cathodic regions, the electrons are able to flow
rate of oxygen would be much higher than that of H2S. However, there through the conductive particles in the corrosion layer and the mill-
are possibilities that a small amount of gaseous H2S penetrates the scale to reach the cathodic regions where oxygen and water are accessi-
ble. Although the concrete material is not conductive itself, the water-
filled porous microstructures make the electron flow possible. The
Table 2 exact anodic and cathodic reactions (with giving and accepting elec-
Values of a1, a2, and a3 obtained from fitting of experimental data and values of rebar cor- trons) are as follows:
rosion n (no rust thickness modeling for Section 4 due to the rust dissolution).

a1 a2 a3 R2 n (wt%) Anode: Fe→Fe2þ þ 2e− ð8Þ

Section 2 0.6795 0.3961 0.0301 0.8673 1.6
Section 3 1.7851 0.6446 0.0569 0.8314 4.4 1
Section 4 – – – – 5.1 Cathode: H2 O þ O2 þ 2e− →2ðOHÞ− ð9Þ
2
 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748 747

Fig. 6. A conceptual model describing the migration of ions and chemical reactions at the steel/concrete interface. Box 1, corrosion reactions close to steel; Box 2, corrosion reactions close to
concrete.

Between the concrete and steel, soluble species can move through the total chloride content (by weight of cement) of the concrete are varied in
mill-scale at the steel/concrete interface, leading to rust formation on different countries (Ahmad, 2003). In this study, the maximum chloride
both sides of the mill-scale, forming the corrosion layer and corrosion- content of cement was 0.19%, which may have contributed to the iron
filled paste. The anions, Cl− and OH− formed by the reduction of O2 in chlorides occurring in the corrosion products. Although the decreasing
the cathodic region (concrete porous area at the MS/CP interface), regulatory chloride value, such as 0.06 wt% for common Portland cement
would be released and penetrate through the mill-scale, reacting with in China (Chinese Standard, 2007) may reduce the chlorides in cement,
Fe2+ produced by the iron oxidation near anodic region (steel surface at chlorides can also derive from other concrete ingredients including
the S/CL interface). Thus ferrous hydroxides and ferrous chlorides some admixtures and impurities in aggregates (Australian Standard,
would be formed in the corrosion layer (Fig. 6, Box 1). Once the soluble 2007). Further, chlorides from the external environment, i.e. wastewater,
intermediary ‘green complexes’ (represented by FeCl2) encounters OH−, are also non-negligible through the exposure during the long service life
it would undergo further oxidation to form solid precipitates and release (50–100 years). External chlorides were reported to be able to permeate
Cl− near the steel. The Fe2+cations produced at the anodic region would through sound concrete or reach the steel through cracks thus inducing
diffuse away from the steel surface to the adjacent concrete (Fig. 6, Box corrosion afterwards (Cardoso et al., 2018).
2) which simultaneously induces similar reactions. However, the amount This study suggested that, due to the relatively low levels of H2S
of ferrous chlorides would be lower in the corrosion-filled paste. Although compared to oxygen in sewer air and the protective concrete covers,
the chloride ions carried by FeCl2 can also move into the corrosion-filled H2S and the biogenic sulfuric acid may not directly contribute to the
paste, the lower potential of the cathode compared to the anode would product formation of rebar corrosion in sewers, although its contribu-
favor Cl− to primarily move from the corrosion-filled paste towards the tion to concrete corrosion would reduce the protective concrete layer.
rebar. Additionally, because of the small volume and high ion activity, Instead, oxygen and chloride are elements of crucial importance in
chloride ions are likely capable of penetrating the protective oxide film rebar corrosion in sewers. However, further studies in real sewer condi-
(mill-scale layer and corrosion layer) easier than do other ions and tions are required to verify these laboratory findings and to incorporate
reach the steel surface, leaving it vulnerable to corrosion (Joshi, 2015). higher and more diverse H2S levels.
In that case, the area where chloride ions accumulate and act as a corro-
sion stimulator is the steel surface, which is the inner most layer at the 4. Conclusions
corroded steel/concrete interface. Furthermore, the chloride accumula-
tion at the steel/rust interface is a dynamic process. The thickness of the The corrosion mechanism of reinforcing steel in concrete sewers
chloride-enriched layer depends on the chloride amount released from were comprehensively investigated using the advanced mineral analyt-
concrete or the external environment, the migration of the anions ical techniques including mineral liberation analyzer, energy dispersive
powered by the potential difference, and the corrosion reactions near spectroscopy and Micro X-ray diffraction, which were applied for the
the steel surface. The possible sources of chlorides are clarified as follows. first time to the analysis of rebar corrosion in sewers. This has led to
Chloride ions could reach to the rebar level either through the con- the following key findings:
crete ingredients or from the external environment. Previous studies pro-
pose that chloride-containing cement may potentially contribute to rebar • H2S and the biogenic sulfuric acid may not directly participate in cor-
corrosion (Portland Cement Association, 2002). The limits imposed on the rosion reactions of rebar covered by concrete or exposed completely
748 Y. Song et al. / Science of the Total Environment 649 (2019) 739–748

Jahani, F., Devinny, J., Mansfeld, F., Rosen, I., Sun, Z., Wang, C., 2001. Investigations of sul-
to sewer atmosphere. Instead, oxygen is more likely responsible for furic acid corrosion of concrete. I: modeling and chemical observations. J. Environ.
rebar corrosion in sewers. When evident cracks occur inside concrete, Eng. 127 (7), 572–579.
Jiang, G., Keller, J., Bond, P.L., 2014a. Determining the long-term effects of H2S concentra-
H2S may reach the rebar and play a role in the steel corrosion process tion, relative humidity and air temperature on concrete sewer corrosion. Water Res.
by converting the iron oxides. 65, 157–169.
• Chloride ions play a significant role in initiating rebar corrosion in con- Jiang, G., Wightman, E., Donose, B.C., Yuan, Z., Bond, P.L., Keller, J., 2014b. The role of iron
in sulfide induced corrosion of sewer concrete. Water Res. 49, 166–174.
crete sewers, resulting as a thin chloride-enriched layer at the steel/
Jiang, G., Sun, J., Sharma, K.R., Yuan, Z., 2015a. Corrosion and odor management in sewer
rust interface. systems. Curr. Opin. Biotechnol. 33, 192–197.
• Away from the chloride-enriched layer, iron oxides can accumulate on Jiang, G., Sun, X., Keller, J., Bond, P.L., 2015b. Identification of controlling factors for the ini-
tiation of corrosion of fresh concrete sewers. Water Res. 80, 30–40.
both sides of the mill-scale to form a corrosion layer and corrosion-
Joseph, A.P., Keller, J., Bustamante, H., Bond, P.L., 2012. Surface neutralization and H2S ox-
filled paste respectively. The corrosion layer around rebar circumfer- idation at early stages of sewer corrosion: influence of temperature, relative humidity
ence is non-uniform and the rust thickness with respect to polar coor- and H2S concentration. Water Res. 46 (13), 4235–4245.
dinates follows a Gaussian model. Joshi, N., 2015. Concrete Pipes and Pipelines: Unreinforced and Reinforced. Alpha Science.
Li, X., Kappler, U., Jiang, G., Bond, P.L., 2017. The ecology of acidophilic microorganisms in
• MLA and Micro-XRD were demonstrated to be effective and accurate the corroding concrete sewer environment. Front. Microbiol. 8.
in determining various corrosion products for both rebar and con- Lin, H.-W., Lu, Y., Ganigué, R., Sharma, K.R., Rabaey, K., Yuan, Z., Pikaar, I., 2017. Simulta-
crete, which in particular, helped to characterize the rusts at the neous use of caustic and oxygen for efficient sulfide control in sewers. Sci. Total En-
viron. 601, 776–783.
steel/concrete interface. Lu, Y.-y., Hu, J.-y., Li, S., Tang, W.-s., 2018. Active and passive protection of steel reinforce-
ment in concrete column using carbon fibre reinforced polymer against corrosion.
Electrochim. Acta 278, 124–136.
Notes Mori, T., Nonaka, T., Tazaki, K., Koga, M., Hikosaka, Y., Noda, S., 1992. Interactions of nutri-
ents, moisture and pH on microbial corrosion of concrete sewer pipes. Water Res. 26
(1), 29–37.
The authors declare no competing interests. Nnadi, E.O., Lizarazo-Marriaga, J., 2012. Acid corrosion of plain and reinforced concrete
sewage systems. J. Mater. Civ. Eng. 25 (9), 1353–1356.
Acknowledgments Noeiaghaei, T., Mukherjee, A., Dhami, N., Chae, S.-R., 2017. Biogenic deterioration of con-
crete and its mitigation technologies. Constr. Build. Mater. 149, 575–586.
Parande, A., Ramsamy, P., Ethirajan, S., Rao, C., Palanisamy, N., 2006. Deterioration of Re-
The authors acknowledge the financial support provided by the inforced Concrete in Sewer Environments. Thomas Telford Ltd, pp. 11–20.
Australian Research Council through the ARC linkage project Parker, C., 1945a. The corrosion of concrete 1. The isolation of a species of bacterium as-
sociated with the corrosion of concrete exposed to atmospheres containing hydrogen
(LP150101337). Guangming Jiang is the recipient of an Australian Re- sulphide. Aust. J. Exp. Biol. Med. Sci. 23 (2).
search Council DECRA Fellowship (DE170100694). Parker, C., 1945b. The corrosion of concrete 2. The function of Thiobacillus concretivorus
(nov. spec.) in the corrosion of concrete exposed to atmospheres containing hydro-
gen sulphide. Aust. J. Exp. Biol. Med. Sci. 23 (2).
Appendix A. Supplementary data Parker, C.T., 1947. Species of sulphur bacteria associated with the corrosion of concrete.
Nature 159 (4039), 439–440.
Supplementary data to this article can be found online at https://doi. Pikaar, I., Sharma, K.R., Hu, S., Gernjak, W., Keller, J., Yuan, Z., 2014. Reducing sewer corro-
sion through integrated urban water management. Science 345 (6198), 812–814.
org/10.1016/j.scitotenv.2018.08.362. Portland Cement Association, 2002. Concrete Information: Types and Causes of Concrete
Deterioration. Portland Cement Association, Skokie, IL, USA.
References Sanchez-Silva, M., Rosowsky, D.V., 2008. Biodeterioration of construction materials: state
of the art and future challenges. J. Mater. Civ. Eng. 20 (5), 352–365.
Ahmad, S., 2003. Reinforcement corrosion in concrete structures, its monitoring and ser- Song, Y., Tian, Y., Zhao, X., Guo, H., Zhang, H., 2016. Corrosion process of ductile Iron with
vice life prediction––a review. Cem. Concr. Compos. 25 (4–5), 459–471. cement mortar linings as coatings in reclaimed water. Int. J. Electrochem. Sci. 11,
AS 1379-2207, Standard Australian, 2007. Specification and Supply of Concrete 7031–7047.
(Australia). Sterling, R., Simicevic, J., Allouche, E., Condit, W., Wang, L., 2010. State of Technology for
Bertolini, L., Elsener, B., Pedeferri, P., Redaelli, E., Polder, R.B., 2013. Corrosion of Steel in Rehabilitation of Wastewater Collection Systems. Rep. EPA/600/R-10/078. US Envi-
Concrete: Prevention, Diagnosis, Repair. John Wiley & Sons. ronmental Protection Agency, Washington, DC (Mar. 25, 2012).
Cardoso, J.L., Prado, A.P.A., de Araujo, A., Pacheco, M.S., Ribeiro, J.L.S., Panossian, Z., 2018. Sun, X., Jiang, G., Chiu, T.H., Zhou, M., Keller, J., Bond, P.L., 2016. Effects of surface washing
Stainless Steel Rebar for Marine Environment: Study of Galvanic Corrosion with Car- on the mitigation of concrete corrosion under sewer conditions. Cem. Concr. Compos.
bon Steel Rebars Used in the Same Concrete Structure. 68, 88–95.
Cayford, B.I., Jiang, G., Keller, J., Tyson, G., Bond, P.L., 2017. Comparison of microbial com- Sutrisno, W., Suprobo, P., Wahyuni, E., 2017. Investigation of Non-uniform Rust Distribu-
munities across sections of a corroding sewer pipe and the effects of wastewater tion and Its Effects on Corrosion Induced Cracking in Reinforced Concrete. EDP Sci-
flooding. Biofouling 33 (9), 780–792. ences, p. 02013.
Chitty, W.-J., Dillmann, P., L'Hostis, V., Lombard, C., 2005. Long-term corrosion resistance Tiller, A., 1990. Biocorrosion in civil engineering. Microbiol. Civ. Eng. 24–38.
of metallic reinforcements in concrete—a study of corrosion mechanisms based on Vahidi, E., Jin, E., Das, M., Singh, M., Zhao, F., 2016. Environmental life cycle analysis of pipe
archaeological artefacts. Corros. Sci. 47 (6), 1555–1581. materials for sewer systems. Sustain. Cities Soc. 27, 167–174.
Ellis, A., Giggenbach, W., 1971. Hydrogen sulphide ionization and sulphur hydrolysis in Vollertsen, J., Nielsen, A.H., Jensen, H.S., Wium-Andersen, T., Hvitved-Jacobsen, T., 2008.
high temperature solution. Geochim. Cosmochim. Acta 35 (3), 247–260. Corrosion of concrete sewers—the kinetics of hydrogen sulfide oxidation. Sci. Total
EPA, U, 1991. Hydrogen Sulphide Corrosion in Wastewater Collection and Treatment Environ. 394 (1), 162–170.
System. Wong, H., Zhao, Y., Karimi, A., Buenfeld, N., Jin, W., 2010. On the penetration of corrosion
Fandrich, R., Gu, Y., Burrows, D., Moeller, K., 2007. Modern SEM-based mineral liberation products from reinforcing steel into concrete due to chloride-induced corrosion.
analysis. Int. J. Miner. Process. 84 (1), 310–320. Corros. Sci. 52 (7), 2469–2480.
Fernandes, I., Pericao, M., Hagelia, P., Noronha, F., Ribeiro, M., Maia, J., 2012. Identification Xi, X., Yang, S., Li, C.-Q., 2018. A non-uniform corrosion model and meso-scale fracture
of acid attack on concrete of a sewage system. Mater. Struct. 45 (3), 337–350. modelling of concrete. Cem. Concr. Res. 108, 87–102.
GB175, Chinese Standard, 2007. Common Portland Cement. Chinese Standard Publishing, Zhao, Y., Ren, H., Dai, H., Jin, W., 2011. Composition and expansion coefficient of rust
Beijing, China. based on X-ray diffraction and thermal analysis. Corros. Sci. 53 (5), 1646–1658.
Grengg, C., Mittermayr, F., Ukrainczyk, N., Koraimann, G., Kienesberger, S., Dietzel, M., Zhao, Y., Yu, J., Hu, B., Jin, W., 2012a. Crack shape and rust distribution in corrosion-
2018. Advances in concrete materials for sewer systems affected by microbial in- induced cracking concrete. Corros. Sci. 55, 385–393.
duced concrete corrosion: a review. Water Res. 134, 341–352. Zhao, Y., Yu, J., Wu, Y., Jin, W., 2012b. Critical thickness of rust layer at inner and out sur-
Heitz, E., Flemming, H.-C., Sand, W., 1996. Microbially Influenced Corrosion of Materials: face cracking of concrete cover in reinforced concrete structures. Corros. Sci. 59,
Scientific and Engineering Aspects. Springer-Verlag. 316–323.
Idriss, A.F., Negi, S.C., Jofriet, J.C., Hayward, G.L., 2001. SE—structures and environment: Zhao, Y., Wu, Y., Jin, W., 2013. Distribution of millscale on corroded steel bars and pene-
corrosion of steel reinforcement in mortar specimens exposed to hydrogen sulphide, tration of steel corrosion products in concrete. Corros. Sci. 66, 160–168.
part 1: impressed voltage and electrochemical potential tests. J. Agric. Eng. Res. 79 Zhao, Y., Ding, H., Jin, W., 2014. Development of the corrosion-filled paste and corrosion
(2), 223–230. layer at the steel/concrete interface. Corros. Sci. 87, 199–210.
Islander, R.L., Devinny, J.S., Mansfeld, F., Postyn, A., Shih, H., 1991. Microbial ecology of
crown corrosion in sewers. J. Environ. Eng. 117 (6), 751–770.