Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Cement and Concrete Research 131 (2020) 106034

Contents lists available at ScienceDirect

Cement and Concrete Research


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

Long-term in situ performance of geopolymer, calcium aluminate and T


Portland cement-based materials exposed to microbially induced acid
corrosion
Cyrill Grengga, , Neven Ukrainczykb, Günther Koraimannc, Bernhard Muellerd, Martin Dietzela,

Florian Mittermayre
a
Institute of Applied Geosciences, Graz University of Technology, Rechbauerstraße 12, 8010 Graz, Austria
b
Institute of Construction and Building Materials, Technische Universität Darmstadt, Franziska-Braun-Straße 3, 64287 Darmstadt, Germany
c
Institute of Molecular Biosciences, University of Graz, Humboldstraße 50, 8010 Graz, Austria
d
Institute of Analytical Chemistry and Food Chemistry, Graz University of Technology, Stremayrgasse 9/II, Graz, Austria
e
Institute of Technology and Testing of Building Materials, Graz University of Technology, Inffeldgasse 24, 8010 Graz, Austria

ARTICLE INFO ABSTRACT

Keywords: This contribution presents an extensive performance evaluation of metakaolin-based geopolymer and calcium
Durability aluminate mortars during a long-term field exposure to a wastewater system exhibiting intensive microbially
Alkali activated cement induced acid corrosion (MIAC) conditions. Results were compared to a high-performance OPC concrete, typi-
Calcium aluminate cement cally used for manhole fabrication. Field conditions were obtained by monitoring relative humidity, tempera-
Long-term performance
ture, H2S and CO2. Microstructural degradation and elemental distributions within each binder type over time
Corrosion
Microstructure
were recorded by electron probe microanalysis, X-ray diffraction and pH-imaging techniques, together with
Biocorrosion specimen's mass and surface-pH characterization. Microbiome and epifluorescence analyses indicated exposure-
time dependent alternating bacterial community structures within the materials. Clear differences regarding
bacterial species distribution and biodiversity were observed for the different binder types. Estimated corrosion
rates were the lowest (1.4 mm/a) and the highest (13.3 mm/a) for the tested geopolymer formulations, pointing
out that only well designed geopolymer materials may provide an improved MIAC resistance.

1. Introduction established due to chemical and microbial processes during sewage


transportation [2–4]. In this matter, microbially induced acid corrosion
The establishment of modern underground sewer networks as we (MIAC) of concrete and metals is one of the central processes leading to
know them today represents a major step in the development of sani- sewer network degradation globally, with both economic and health-
tary standards of modern society. Thereby, not only the rising quan- related impact. For example, MIAC accounts for 10% of all possible
tities of sewage produced due to growing populations and increasing damage types in overall wastewater drains and pipe infrastructure in
urbanisation within the last two centuries have been efficiently man- Germany [5]. As 38% of this infrastructure is made from concrete based
aged, but also public health related aspects, such as water-borne dis- materials, it was estimated that MIAC and abrasion accounts roughly
eases were averted. Additionally, the hazardous impact of potential 26% (calculated as 10%/0.38) of damages in concrete wastewater in-
wastewater contamination for the adjacent environments was limited frastructure. The age of the average German pipe infrastructure is ap-
[1]. However, the World Health Organization (WHO) reported in 2005 proaching 40 years, while about one fifth of it has reached damage that
that around 2.6 billion people are still lacking basic sanitation. This must be rehabilitated within the next 4 years. Moreover, rapid urba-
clearly indicates the still growing demand for efficient, safe, and cost- nisation worldwide requires substantial pipe infrastructure to be newly
effective collection, transport and treatment of sewage worldwide, to- constructed. Therefore, an increasing necessity exists to design in-
gether with the need to maintain the long-term performance of existing novative and highly durable materials for wastewater infrastructure,
infrastructure. Frequently reported insufficient material durability either newly constructed or to rehabilitate the existing damaged ones
performances are mainly caused by chemically aggressive conditions for substantial extension of their service life. To date, no sustainable


Corresponding author.
E-mail address: cyrill.grengg@tugraz.at (C. Grengg).

https://doi.org/10.1016/j.cemconres.2020.106034
Received 25 November 2019; Received in revised form 20 February 2020; Accepted 5 March 2020
Available online 13 March 2020
0008-8846/ © 2020 Published by Elsevier Ltd.
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Fig. 1. Exposure conditions, together with a schematic illustration of the analytical approach conducted during each sampling point.

building material exists, which meets the long-term requirements in extensive field campaign over the duration of 18 months in which
such aggressive and corrosive environments [6–8]. Reasons therefore metakaolin-based geopolymer mortars (GPM) were tested with respect
are multifaceted and were recently summarized by Grengg et al. [6]. In to their overall performance within a sewer system exhibiting highly
recent years, much research focused on the development of calcium bio-chemically aggressive conditions. In this study, the performance of
aluminate cement (CAC) based materials, demonstrating their superior geopolymer mortars was evaluated in comparison to a high-perfor-
performance in MIAC environments. Up to 6 times better performance mance OPC concrete typically used for manholes and CAC mortars ty-
was reported in systems using CAC compared to OPC materials [9–11]. pically used for repair of biodegraded concrete. This study focuses on
Reasons for their superior performance are attributed to the acid buffer relevant material properties and their evolution over time documenting
capacity of aluminum hydroxide (Al(OH)3) and the formation of a pore progressive degradation. Monitored parameters comprise surface pH,
filling Al(OH)x gel under acid attack that exhibits better cohesion of the pH cross-section mappings, mass balancing, optical observations and
degraded layer, as well as due to potential antimicrobial properties of microstructural alterations, as well as a detailed characterization of
Al [10,12]. biofilm-forming bacteria.
Ordinary Portland cement-based (OPC) materials form mainly cal-
cium silicate hydrate (C-S-H) phases and calcium hydroxide (CH) from
the hydration of the clinker phases. OPC hydration products are ther- 2. Materials & methods
modynamically unstable at pH values below 10 and consequently dis-
solve or alter (decalcify) [13]. Accordingly, there is serious growing 2.1. Environmental system characteristics
concern about the suitability of OPC-based materials in such environ-
ments. This raises the demand for alternative types of materials with The system chosen for the field study lies in Austria close to the city
similar physiochemical, but different microstructural and chemical of Leoben and is part of the regional communal wastewater network.
properties, ensuring chemical resistance. In this perspective, geopo- Therein, specimens were placed in a power main outlet storage basin,
lymer concretes (GPC) represent a promising group of materials, po- heavily affected by MIAC. The resulting severe damage to the existing
tentially suitable for, but not exclusively, MIAC environments. Geopo- OPC based concrete structures requires integral remediation action
lymer (GP) technology avoids the formation of Ca-rich acid-soluble latest every 2–10 years. Various state-of-the-art repair materials have
hydration products present in OPC and CAC binders as well as in alkali- been applied in the past, ranging from standard OPC-based solutions to
activated binders rich in calcium. Material properties of hardened GPC different lining and sealing solutions, all with little to no success. For
are reached via polycondensation of alumino-silicates, thereby ex- instance, an epoxy-based coating, applied during the last renovation in
hibiting improved acid resistance [14–16]. GP is made from one or 2017 had to be completely renewed after < 1.5 years in spring 2019.
several aluminosilicate materials (e.g. calcined clays, volcanic rocks or Accordingly, this system provides ideal in-situ conditions to investigate
fly ash (FA)) mixed with alkaline reagent solutions, such as soluble the accelerated behavior of novel building materials under highly ag-
sodium or potassium silicates (i.e. water glass (WG)), water and ag- gressive MIAC conditions. Additionally, the storage basin contains an
gregates to form a construction material with similar mechanical elevated pedestal platform with the dimensions of roughly 3 m2, de-
properties compared to OPC- and CAC-based concretes [14,17]. Re- signed for operational service purposes, which was used for the place-
search on alkali aggregate reaction [14,18] demonstrated that meta- ment of the specimen-holder boxes, avoiding any direct contact to the
kaolin-based geopolymer mortars, although having an order of mag- wastewater (see Fig. 1 left side).
nitude higher alkali contents than PC-based mortars, exhibit negligible In order to evaluate the prevailing environmental system char-
swelling and loss of rigidity. acteristics, concentrations of hydrogen sulfide (H2S) and temperature
To date, data on field performances of GP in sewer environments, were monitored permanently, using a myDataSens1000_H2S gas
affected by MIAC are scarce [19]. The present contribution presents an monitor. The relative humidity (r.h.) of the sewer atmosphere was re-
corded by a Voltcraft DL-121TH data logger over the period of several

2
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Table 1
Chemical composition (wt%) of the binders/raw materials.
Material SiO2 Al2O3 CaO Fe2O3 TiO2 MgO Na2O K2 O SO3 H2O

PC (CEM I) 18.1 3.1 51.9 4.75 0.17 1.57 0.17 0.3 2.0
CAC 5.2 51.4 36.7 2.5 1.9 0.5 0.1 0.2
Metakaolin K1 47.6 38.3 1.4 2.0 0.1 0.1
Metakaolin K2 66.2 21.3 2.5 4.1 1.0 0.5 0.1 0.5
Trass 53.4 19.4 3.4 4.9 0.6 1.2 4.5 4.0 0.2
Metakaolin + TrassK1T 49.3 32.6 1.0 2.5 1.6 0.4 1.4 1.3 0.1
Waterglass (WG) 22.0 23.0 55.0

months. Over several days, carbon dioxide (CO2) concentrations were 10 MPa/min), on the previously dried sample with an evacuation
captured utilizing CORA, an energy-efficient NDIR-device with an pressure of 50 μm Hg during 5 min. The pore diameter was calculated
overall accuracy of ± 3% [20]. using the Washburn equation and following mercury properties: density
13.53 g/cm3, surface tension 0.485 N/m, and contact angle 130°.
The open porosity was obtained from the water absorption mea-
2.2. Material characteristics surements using halves of the specimen prisms (40 × 40 × 80 mm,
cured for 28 days). Therefore, the dry mass (dried for 24 h at 105 °C),
In this study, three different groups of mineral-based binder systems the water saturated mass and the sample volume (Archimedes method)
were tested. Two types of geopolymer, one CAC, and one OPC material. were measured on at least three replicates.
Chemical composition of the raw materials is given in Table 1. Po- Chemical compositions of the mortars were calculated based on
tassium based waterglass (alkali silicate solution) had a solid content of chemical composition of individual components, which were evaluated
45 wt%, viscosity of 20 mPa s, pH of 13.5, and SiO2/K2O molar ratio of by X-ray fluorescence analysis (XRF) using a wavelength dispersive X-
1.5 (according to manufacturer information). Geopolymers were man- ray fluorescence spectrometer (PANalytical 2404).
ufactured using two types of metakaolin, a high purity white one (K1) An OPC concrete C55/67 (exposure classes: XC4/XD2/XF3/XA2L/
and a quartz-rich rust brown iron-rich one (K2), as well as Trass (nat- XA2T) according to EN 206 was used as a reference. This concrete was
ural pozzolan). Quantitative XRD using Rietveld refinement (10 wt% fabricated with a CEM I 42.5 R SR0 (EN 197-1), a w/c ratios of 0.43,
spiked corundum) allocated the amorphous (~metakaolin) and quartz predominantly siliceous aggregates (maximum size 16 mm) and about
mass content to be 81 wt% and 10 wt% respectively in sample K1, while 4% of entrained air. Cubes of 150 mm3 were casted and cured according
46 wt% amorphous and 40 wt% quartz in K2 ([21] data not shown). to ÖNORM B4710-1. The w/b of the reference concrete is 0.43, which is
The Trass was obtained as a commercial product according to DIN a typically used concrete in Austria for sewers according to ÖN B4710
51043, exhibiting a LOI = 5.40 wt% and a density of 2.65 g/cm3. The (Table 2). The 28-days compressive strength was 84 N/mm2.
compressive strength of Trass/Ca(OH)2 mortars (DIN 51043) was 150 × 40 × 40 mm3 concrete specimens were cut out of the cube with
7.8 MPa. The Calcium aluminate cement used contained CaAl2O4 as the an angle grinder [22].
main clinker phase and the following minor phases (CaO)2(Al2O3) For the field exposure, all hardened mortar prisms were cut in half
(SiO2), (CaO)(TiO2) and (CaO)12(Al2O3)7 (see appendix for chemical resulting in dimensions of 80 × 40 × 40 mm (75 × 40 × 40 in case of
compositions). CEN standard sand following DIN EN 196-1 was used the reference OPC concrete). Prior to exposing, the specimens were
(grading 0.08–2 mm, conforming to the standard size distribution), dried for 24 h at 40 °C. Subsequently, each specimen was weighted and
having at least 95% quartz and following minor phases; feldspar, albite, surface pH on four randomly distributed points was measured, using an
anorthite, microcline and muscovite as identified by XRD. The mix Extech PH100; ExStik® pH Meter. Thereafter, four (or five) replicates of
design is presented in Table 2, which for the GP and CAC mortars was each material were placed in open plastic boxes with a voided bottom
chosen by fixing the workability spread at 17 ± 1 cm. In geopolymer (Fig. 1), in order to avoid the accumulation of aggressive solutions at
mixture GP-K1T, 30 wt% of metakaolin (K1) was replaced by Trass. the bottom over time, and placed on the elevated platform in the power
Mortar standard specimens (size of 160 × 40 × 40 mm) were cast main outlet basin.
according to EN 196-1. Workability spread and compression tests were
done according to EN 1015-(3 and 11). Specimens were cured for
minimum of 28 days at 100% r.h. and 20 °C. After 28 days, CAC-based 2.3. Analytical methods on exposed specimens
mortars were additionally sealed and cured at 45 °C for one day to
convert the metastable phases to the stable ones. Specimens were removed from site after 3, 6, 12 and 18 months. On
On each hardened mortar, mercury intrusion porosimetry (MIP) site, a photo documentation of each specimen was conducted.
measurements were carried out using a Pascal 440 from Thermo- Thereafter, one side of each specimen was carefully scraped and the
Scientific. The measurements were performed up to 400 MPa with resulting sample was collected in a sterile tube, immediately trans-
gradual reading for each applied pressure (increase speed of about ported to the laboratory and stored at −20 °C until further use (see

Table 2
Mixture design (g) and proportioning of the specimens (mass ratio).
Sample CEM Ia CAC MK-K1 MK-K2 TRASS WG Aggregates Sandd W (H2O) W/Bc MK(+T)b/WG

OPC 450 Silicious dmax = 16 mm 194 0.43


CAC 450 1350 225 0.45
GP-K1T 315 135 404 1350 0.32c 1.11b
GP-K2 450 356 1350 0.31c 1.27

a
CEM I 42.5 R SR0 (EN 197-1).
b
30 m.% replacement of Metakaolin by Trass.
c
Comprises the powder binder as well as the solid content of the waterglass (WG), i.e. W/B = 0.45 WG/(0.55 WG + powder).
d
CEN standard sand acc. EN 196-1.

3
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

below section on DNA extraction). In the lab, specimens were photo- the V3V4 region of bacterial 16S rDNA and sequenced on Illumina
graphed using a digital camera and dried at 40 °C for 24 h. Thereafter, MiSeq with the 2 × 300 bp paired-end read module. Demultiplexed,
the mass of each specimen was determined and four randomly dis- primer-clipped raw sequences in FASTQ format were further analyzed
tributed surface pH measurements were conducted, using an Extech using the Qiime 2 bioinformatics pipeline running on a MacBook Pro
PH100; ExStik® pH Meter. computer.
Analysis of NGS 16S sequence data was done using Qiime 2, version
2.3.1. Microstructural and mineralogical analyses 2019.1 [25]. FASTQ forward and reverse sequences were imported,
From one side, if present, soft deterioration layers were scraped subsequently denoised and filtered using the DADA2 plugin. To remove
from the specimen's surfaces dried and subsequently powdered for low quality bases, forward and reverse sequences were truncated to 250
qualitative X-ray diffraction (XRD) analyses. XRD patterns were re- and 240 bases, respectively. Alpha diversity in the samples was esti-
corded over the range of 4–85°2Θ with a step size of 0.008°2Θ and a mated using a sampling depth of 25,000 sequences. In order to assign
count time of 40 s/step, using a PANalytical X'Pert PRO diffractometer. taxonomy to ASVs (amplicon sequence variants) - formerly called OTUs
The diffractometer was equipped with a Co-tube (40 kV and 40 mA), a (operational taxonomic units) - the SILVA [26] 132 SSU database
spinner stage, 0.5° divergence and anti-scattering slits, and a Scientific (V3V4, 341–805) was used as a classifier.
X'Celerator detector. Mineral identification and quantification were To visualize the presence of microbial activity within the sampled
carried out using the PANalytical X'Pert HighScore software (version material, freshly cut cross-section surfaces were stained using fluor-
2.2e) and pdf-2 database. escent nucleic acid stains SYTO9 and propidium iodide (PI) using a
In order to guarantee stability of the remaining corrosion layers, protocol essentially described in Grengg et al. [25]. Images with both
samples were impregnated with a low viscosity epoxy resin (Buehler; dyes were conducted since PI has a higher DNA/RNA binding capacity
EpoThin 2 Hardener) under vacuum conditions. Subsequently, cross- but cannot penetrate living cells, whereas SYTO9 is able to penetrate
sections with the thickness of around 5 mm were cut from the middle of living cells with lower binding efficiency. From these images, multi-
the longitudinal side of the samples using a diamond saw (see Fig. 1). channel images were prepared by overlaying the image obtained with
On these freshly cut cross-section surfaces, optical pH imaging the fluorescence microscope prior to staining (no stain, red color)
measurements were conducted using the analytical method described in against the image obtained after staining (SYTO9/PI, green color).
[22,23]. A sensor foil exhibiting an apparent pKa value of about 4.5 was
chosen, enabling a sensitive range of between 3.4 and 6.4. The opposite 3. Results
freshly cut surface was used for Epifluorescence analyses (see below).
Thereafter the cross-sections were polished for further micro- 3.1. Material characteristics
structural analyses. (Semi)-quantitative elemental distribution images
of aluminum, calcium, magnesium, silicon, potassium and sulfur were The properties of the mortars (Tables 1 and A2; Fig. A1 in Appendix)
recorded by electron probe microanalysis (EPMA) using a JEOL show that by fixing the workability spread at 17 cm, the capillary
JXA8530F Plus Hyper Probe (JEOL, Tokyo, Japan), equipped with a porosity was maintained around 12.5 ± 0.6% (dried at 105 °C). The
field emission gun. The wavelength-dispersive analytical mode with highest value was for the GP-K1T sample (13.1%) and the lowest for
15 kV acceleration voltage and a beam current of 10 nA was used. A GP-K2 (12.0%), which is in agreement with the lower amount of wa-
dwell time of 15 mS and step size of 3 × 3 μm was chosen. The terglass used (water to reactive solid content shown in Table 2). Al-
quantification of the individual element mappings in wt% were per- though being more porous, the GP-K1T sample exhibited the highest
formed against mineral standards from SPI (Microcline for Al, K and Si; compressive strength, as it was prepared with purer metakaolin (K1),
Dolomite for Ca and Mg; Barite for S; Ilmenite for Fe). The dimensions requiring more waterglass to fix the workability (Table 2). This is also
of the mappings were chosen in a way to image the entire deterioration in agreement with pore size distribution results (Fig. A1), where the
horizons, the transition zones, as well as parts of the interior intact amount of gel pores (< 10 nm) decreases at the expense of capillary
cementitious matrix. In order to better characterize the mineralogical pores by lowering the amount of reactive components (i.e. GP-K1T >
changes, element profiles from the same mapping areas, but only re- GP-K2, acc. to mix proportioning Table 2). The higher amount of the
cognizing micro-regions corresponding to the cementitious matrix of reactive components, i.e. both the reactive metakaolin and waterglass
the materials were extracted. For this, elemental cut-off concentrations content, resulted in the highest amount of gel pores. The drying
were determined to ensure disregard of pixels attributed to siliceous shrinkage micro-crack formation increased in following order: GP-
aggregates. K2 < GP-K1T < OPC < CAC, as detected by MIP results in the
micro-scale pore range 0.1–5 μm. In general, geopolymer mortars ex-
2.3.2. DNA extraction, quality control and 16S bacterial community hibit finer pore sizes compared to the OPC and CAC based materials,
analyses where CAC has the highest amount of coarse pores due to the conver-
To extract the DNA from the collected samples (scraped samples, see sion reactions [27] (Fig. A1).
above) the DNeasy PowerLyzer PowerSoil Kit (Qiagen) was used, fol- Results of compressive strength show a rapid strength development
lowing the manufacturer's protocol. For the bead beating option, a for geopolymers. For geopolymers, the strength results do not correlate
PowerLyzer 24 homogenizer (Qiagen) with 45 s and 2500 rpm for with the open porosity (Table A2), but are in semi-quantitative agree-
disrupting cells was used. The quality and amount of the isolated total ment with the amount of the gel pores (i.e. amount of geopolymer gel)
DNA was determined by DNA agarose gel electrophoresis and by (Fig. A1).
measuring absorbance at 280 nm in a NanoDrop microvolume spec-
trophotometer. In order to verify the presence of bacterial DNA, PCR 3.2. Study site characteristics
reactions were performed with primers specific for 16S rDNA covering
variable regions V5 and V6 as described in Grengg et al. [24]. The re- Four (or five) replicates of each casted specimen were placed into a
sulting DNA amplicons were separated by DNA agarose gel electro- sewer system, strongly affected by MIAC. Average H2S concentrations
phoresis. Only DNA isolated from OPC, CAC and GP that was suffi- were 65 ppm with regular maximum concentrations of up to 722 ppm,
ciently pure and abundant, and yielded the expected PCR product, was corresponding to the pumping pulses of the system every few of hours
further analyzed by 16S profiling. (Fig. 2A). Relative humidity (r.h.) ranged between 69 and 100%
DNA samples in a volume of 15 μL (total amount of DNA was 0.3 to (Fig. 2B). A positive correlation between temperature (T) and H2S
2 μg) were sent to Eurofins Genomics (Ebersberg, Germany) next gen- concentrations was observed, accompanied by a corresponding de-
eration sequencing (NGS) service. Amplicons were generated targeting crease in r.h. Results of CO2 measurements, conducted over the period

4
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

25
700 A H2S (ppm)
T(°C)

600
20

500

15
H2S (ppm)

400

T(°C)
300 10

200
5
100

0 0
0 50 100 150 200 250 300 350
time (days)
100 7000
r.h. (ppm)
T(°C) B 17
CO2 (ppm)
T(°C) C 16
95 6000
16 14
90
5000
15 12
CO2(ppm)
r.h.(%)

T(°C)
85
T(°C)

4000
14 10
80
3000
13 8
75
12 2000 6
70
11 1000 4
65
0 5 10 15 20 25 30 17 Nov 18 Nov 19 Nov 20 Nov 21 Nov
time (days) time

Fig. 2. Average concentration of H2S and temperature (T) over the period of one year (A), together with typical humidity levels (B). CO2 concentrations ranged
between 800 and 6500 ppm over the period of 4 days (C).

of several days, indicate high concentrations between 800 and between 0 and 6 months, followed by a strong drop in pH between
6500 ppm (Fig. 2C). months 6 and 12, from 9.5 ± 0.1 (GP-K1T), 8.9 ± 0.2 (GP-K2) and
8.3 ± 0.3 (CAC) to below 4.5. Between months 12 and 18, only a slight
pH reduction was observed on the CAC (3.1 ± 0.4) and GP-K2
3.3. Optical, mineralogical and microstructural observations (3.9 ± 0.1) while the pH of GP-K1T dropped to around 2.0 ± 1.0. By
contrast, the pH of the OPC material exhibited a strong drop from initial
Optical observation of the specimens extracted after 6, 12 and 11.4 ± 0.6 to 8.4 ± 0.5 after only 3 months of exposure, followed by
18 months of exposure showed high variations (Fig. A2). The OPC gradual decrease in surface pH between months 3 and 9 to 6.6 ± 0.6,
specimens exhibit strong surface alterations including swelling, before again a steep drop was observed to 2.0 ± 1.0 after 18 months.
cracking and spalling, with strongly increasing intensities over time. No mass loss was observed on OPC and CAC materials, but small
The GP-1KT showed strong signs of deterioration after 12 month of mass gains after 18 months of 3.9 wt% and 1.0 wt%, respectively,
exposure, such as cracking, efflorescence and leaching, accompanied by which corresponds to the secondary precipitation of sulfate salts con-
the precipitation of potassium‑aluminum-sulfate (K-Alum: KAl firmed by XRD and EPMA analyses (Fig. 3). Both GP materials exhibited
(SO4)2·12H2O) salt. After 18 months of exposure, major (outer) portions a small gain in mass until 12 months (GP-K1T: 4.3 wt%) and 9 months
of the GP-K1T specimen completely lost its structural integrity and high of exposure (GP-K2: 0.34 wt%), which correlates with the precipitation
deterioration rates were observed. In contrast to OPC and GP-K1T, of potassium alum (see XRD analyses). Between 12 and 18 months, a
substantially less deterioration was observed on the CAC and the GP-K2 strong mass loss of 40.0 wt% was observed on the GP-K1T specimen.
specimens. After 18 months of exposure, low degree of swelling was Contrarily, only a relatively small decrease in mass of 5.2 wt% was
observed on the CAC specimens corresponding to the precipitation of measured for the GP-K2 in the period between 9 and 18 months (Fig. 3).
expansive sulfate salts such as gypsum (see XRD and EPMA measure- Qualitative XRD bulk analyses of the corrosion layers, scraped from
ments), while a low degree of surface cracking, efflorescence and the one side of each specimen surface after 12 and 18 month of exposure,
precipitated K-alum salt was observed on the surface (layer) of GP-K2 confirmed the dissolution of the Ca-(Si)-hydrate rich cementitious ma-
specimen. trix of the OPC material and the precipitation of high amounts of ex-
The evolution of the surface pH of all materials showed a strong pansive gypsum (Gp) and elemental sulfur (S0) (Table A3). Within the
reduction from the initial 11.4 ± 0.6 (OPC), 10.0 ± 0.1 (CAC), deterioration layers of the CAC materials the initial main phases of the
10.3 ± 0.1 (GP-K1T) and 10.4 ± 0.2 (GP-K2), to as low as 2.0 ± 1.0 cement matrix (namely, C3AH6, C2AS and AH3) were largely dissolved
(OPC), 3.1 ± 0.4 (CAC), 2.0 ± 1.0 (GP-K1T) and 3.9 ± 0.1 (GP-K2) resulting in the formation of high quantities of secondary Gp (with
after 18 months of exposure (Fig. 3). Interestingly, the CAC and both GP increasing amounts with time) and K-Alum (Fig. A3). Additionally, a
materials exhibit a similar pH trend showing a gradual pH reduction

5
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Fig. 3. Surface pH evolution of outsourced specimens with time (A), together with the change in mass in % (B). The yellow shaded area (in A) corresponds to the
measurements conducted on strongly deteriorated specimen surfaces after 18 months (OPC and GP_K2), where highly uneven surfaces and high water absorptivity
prevented stable pH measurements, highlighted by high pH variations of up to 2 pH units. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)

strong increase of the amorphous content with time was observed. average concentrations compared to the intact GP framework. Within
Mineralogical characterization of the GP materials revealed that the this area, K decrease flattens out, while no changes in Si concentrations
corrosion products were elemental sulfur (S0) and high quantities of are observed. The end of this zone, close to the surface, is defined by
potassium alums (K-Alum) (Table A3, Fig. A4). Within the GP-K2 ma- strong drop of K and Al concentrations and an opposing slight increase
terial, low amounts of gypsum (Gp) were formed, corresponding to the of Si concentrations. There is no change within the Fe concentrations
higher Ca concentrations of the metakaolin K2 (1.6 wt%) compared to between intact GP framework and the deterioration zones. In addition,
K1T (0.8 wt%). no S incorporation was detected within the GP-K2 material, while low
Element distribution images (Figs. 4 and 5) and element profiles degree accumulation zones were observed within the thick deteriora-
(Appendix Figs. A5–A7) showed strong mineralogical and micro- tion layer of the GP-K1T material after 12 month. Interestingly, higher
structural changes within the deterioration layers of the different ma- Si concentrations in respect to the intact GP network were observed
terials after 12 and 18 months of exposure. Within the OPC material, within these deterioration layers. On the GP-K1T specimen high cor-
complete dissolution of the Ca-rich cement matrix in the micro- rosion rates were obtained, namely around 6.0 mm after only
structurally deteriorated cement layers, as well as an intensive new 12 months of exposure, leading to a complete deterioration and dis-
precipitation of Ca sulfate salts progressed. After 18 months of ex- solution of large parts of the specimen after 18 months (as described up
posure, penetration of biogenic H2SO4 led to swelling of the specimen, to 40% mass loss), preventing the fabrication of a reliable EPMA
combined with crack formation and a complete destabilization of the sample. Contrarily, the GP-K2 specimen displayed the lowest dete-
microstructure to a depth of approx. 3.9 mm after 12 months, and rioration depths of 0.2 mm after 12 and 2.1 mm after 18 months
9.5 mm after 18 months of exposure. The transition from intact to (Fig. 5).
strongly corroded concrete is marked by an increase in Mg, Fe and, to a Optical pH images were taken from the same cross-section and are
lesser extent, Al concentrations followed by a strong decrease of the displayed in Fig. 6. The core regions of the OPC, CAC and GP-K2 spe-
latter elements and Ca, as well as by the counter-rotating successive cimen's cross-sections obtained pH values above 6.4 (grey color; out of
incorporation of S (Figs. 4a, A5). The corrosion layers consist almost calibration), whereas the measured pH throughout the strongly dete-
exclusively of the newly formed calcium sulphates, recognizable by the riorated GP-K1T specimen did not exceed this value. The near surface
corresponding Ca and S accumulation areas in combination with less regions of all specimens exhibited low pH values between 3.6 and 6.0,
dominant layered amorphous silicon horizons and relicts of undissolved although average pH values within these deterioration layers of the
siliceous aggregates (Fig. 4a). OPC material were on average about one pH unit higher, compared to
The BSE and elemental images conducted on the CAC specimen the CAC and GP-K2 material. The average depth of these low pH areas is
revealed a similar corrosion pattern, although significantly lower cor- in good agreement with the thickness of the deterioration layers ob-
rosion depths of around 0.5 mm after 12 months and 3.2 mm after tained from EPMA. A sharp transition exists between the high pH core
18 months (Figs. 4a, A6). Again, the transition to the intact cement areas and the low pH near-surface deterioration regions, which is in
matrix is characterized by a strong depletion of Ca and Al and con- agreement with the observed mineralogical transition zone between
current concordant areas of Ca and S incorporation, as well as layered intact cementitious matrix and the deterioration layers. In this context,
Si accumulation horizons. In contrast to the OPC material, no accu- it is important to point out that due to the development of thick, soft
mulation zones of Mg or Fe were detected. deterioration layers on the OPC material featuring extremely low me-
The depth of microstructural alteration within the GP materials is chanical strength, two sides were entirely scraped off for microbiome
characterized by a depletion of K and Al and associated (micro)crack and XRD analyses. Accordingly, the pH image obtained is missing for
formation (Fig. 5). At the transition between intact and altered GP these two sides (left and lower side of OPC in Fig. 6). This loss of ma-
framework, a sequence of elemental changes was observed (Fig. A7). terial is quantified by lower cross-section dimensions of roughly
Beginning from the sound inner matrix (right side of Fig. A7), the in- 32 × 32 mm on average (initial specimen dimensions 40 × 40 mm). In
nermost transition area with a width of roughly 0.3 mm is characterized contrast, the CAC and GP-K2 still exhibit their initial dimensions
by strongly decreasing K concentrations, while Al and Si keep the (40 × 40 mm) and shape due to higher structural integrity. While the
concentration levels of the intact GP framework. Next, an Al accumu- decrease in pH within the CAC material seems to proceed quite con-
lation zone of roughly 0.5 mm thickness was observed, having higher sistently from all sides, the pH evolution within the GP-K2 material

6
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

12 month 18 month
OPC CAC OPC CAC

BSE

500 µm 500 µm 500 µm 500 µm

high high high high

low low low low

Ca

500 µm 500 µm 500 µm 500 µm

high high high high

low low low low

500 µm 500 µm 500 µm 500 µm

high high high high

low low low low

Al

500 µm 500 µm 500 µm 500 µm

high high high high

low low low low

Si

500 µm 500 µm 500 µm 500 µm

intact corrosion intact corrosion intact corrosion intact corrosion


microstructure layers microstructure layers microstructure layers microstructure layers
deterioration 9.5 mm 3.2 mm
3.9 mm 0.5 mm
depth

Fig. 4. Displays BSE images, together with the element distribution images of Al, Ca, Si and S throughout the deterioration layers, the transition zone and the intact
cement matrix of the OPC and CAC material after 12 and 18 month of exposure. Microstructural alteration depths were characterized reaching values of 3.9 mm after
12 months and 9.5 mm after 18 months for the OPC material, while low depths of 0.5 mm after 12 months and 3.2 mm after 18 months were determined for the CAC
material.

proceeds non-uniformly, where one side (lower side in Fig. 6) shows measured within the deterioration layers of the OPC specimen with
significantly thicker, up to roughly 5 mm, lower pH ranges. The ma- apparent values around 6, as well as with the expected anoxic to
terial GP-K1T shows throughout the entire cross-section extremely low anaerobic conditions within the thick deterioration layers observed
pH values, corresponding to the intensive degree of deterioration. [28,29].
Epifluorescence images obtained of the materials after 18 months of
3.4. Microbial community structures exposure showed that the biofilm expanded throughout the entire de-
terioration layers of all materials (Figs. 8, A8). Thereby the depth of
Microbial community analysis of 16S rRNA Illumina MiSeq am- biofilm penetration always corresponds to the individual micro-
plicon sequence data of CAC, GP-K1T and GP-K2 revealed microbial structural deterioration depths of each material, seen within the ele-
colonization patterns that are overall consistent with the current sci- mental profiles (Figs. A5–A7) and images (Figs. 4 and 5) as well as
entific view regarding the succession of bacteria during progression of within the pH images (Fig. 6). These results are in agreement with
MIAC [28–30]. Autotroph, acidophilic sulfur oxidizing bacteria previous findings on OPC reported by Grengg et al. [29].
(ASOBs) are dominant in all samples analyzed (Fig. 7, dark blue color). In order to be able to better evaluate the results of bacterial pro-
However, although the indicated specimens were collected at the same filing, more time intervals, namely 6 months for OPC and GP-K1T;
time point (12 months) and from the same location in the sewage 9 months for CAC; and 18 months for GP-K2, were investigated and
system, the detailed bacterial community compositions on different compared to the bacterial community composition after 12 months.
materials exhibit striking variations. While the biofilm of the GP-K1T Fig. 9 shows a heat-map of microbial taxa that reached a relative
consists almost entirely of described ASOBs (At. thiooxidans 48.9%, At. abundance of > 10% in at least one sample. Microbial communities
ferrooxidans 31.4%), within the biofilm of the CAC and GP-K2 high found on OPC and GP-K1T materials at 6 months are dominated by
abundances of heterotroph, acidophilic Acetobacteraceae were found neutrophilic SOB (NSOBs) which almost exclusively belong to the genus
(Fig. 7). A strongly dissenting microbial community structure was Halothiobacillus. Sanger sequencing of 16S rDNA clones (variable re-
found on the OPC material. Interestingly, the dominant autotrophs in gions V5V6), simultaneously conducted on all materials after 6 month
the OPC sample mainly belong to Halothiobacillaceae which are neu- of exposure revealed a similar dominance of Halothiobacillus sp. with 5/
trophilic, together with high abundances of anaerobically growing 10 (OPC), 3/10 (CAC), 7/10 (GP-K1T), 9/10 (GP-K2) clones, respec-
heterotrophs from the Firmicutes phylum (Peptostreptococcaceae and tively (data not shown). The biofilm extracted from the CAC material
Clostridiaceae). This is in accordance with the higher pH values after 9 months consist almost entirely of acidophilic bacteria

7
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

12 month 18 month
GP_K1T GP_K2 GP_K1T GP_K2

BSE

500 µm 500 µm 500 µm

high high high

low low low

loss of outermost deterioration layers


Al

measurement not possible


500 µm 500 µm 500 µm

high high high

low low low

Si

500 µm 500 µm 500 µm

high high high

low low low

500 µm 500 µm 500 µm

high high high

low low low

500 µm 500 µm 500 µm

high high high

low low low

Ca

500 µm 500 µm 500 µm

intact corrosion intact corrosion intact corrosion


microstructure layers microstructure layers microstructure layers
deterioration 2.1 mm
6.0 mm 0.2 mm roughly 20.0 mm
depth

Fig. 5. BSE and elemental distribution images of Al, Ca, K, Si and S showing microstructural alterations occurring within the GP-K1T and GP-K2 material after 12 and
18 months of exposure. Alteration horizons reached 0.2 mm after 12 months and 2.1 mm after 18 months within the GP-K2 material, while 6.0 mm within the GP-
K1T material. Due to the fast progressing deterioration of the GP-K1T material and associated loss of the deterioration layers after 18 months of exposure no stable
measurement was possible.

(Acidithiobacillaceae and Acetobacteraceae), with a shift from Acid- 4. Discussion


ithiobacillus thiooxidans towards Acidithiobacillus ferrooxidans between 9
and 12 months. On the GP-K2 material a strong shift from acidophilic While the overall better performance of CAC materials in MIAC
dominated biofilm towards a similar community patterns as described environments has been reported within several studies
within the OPC (43.2% Peptostreptococcaceae) material between 12 and [7,10,11,31–33], to date, data on field performances of geopolymer (or
18 months was observed. AAM) materials within MIAC environments are sparse [19]. However,
Shannon Indices (H index), statistically calculated as a measure for numerous lab studies and empirical tests have shown their potential
the diversity of the different biofilms, revealed overall reducing diver- ability to perform well within various acidic (abiotic) environments
sities with time of exposure, except for the CAC material which shows a [16,17,34,35].
counter-rotating trend (Fig. 9). After 12 month of exposure the highest
H index of 4.14 and 4.02 was observed on the GP-K2 and OPC material,
respectively, while significantly lower values of 2.44 and 2.42 were 4.1. Corrosion mechanisms
obtained for the CAC and GP-K1T materials. Thereby, the CAC material
generally exhibits the lowest biodiversity. Due to the fundamental chemical and microstructural differences
between the OPC and CAC compared to the GP materials, two generally
different corrosion patterns were observed. Thereby the deterioration of

8
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

6.4 materials are dominated by the dissolution of the calcium rich, hydrate-
6.0 based cementitious matrix and concurrent precipitation of expansive
5.6 corrosion products (mainly gypsum) due to biogenic acid production.
On the OPC material, the strong precipitation of secondary sulfate
5.2
salts led to swelling of the specimen, accompanied by an overall gain in
OPC

4.8 mass, resulting in the formation of up to 10 mm thick corrosion layers


4.4 after 18 month of exposure. These thick corrosion layers might be the
4.0 reason for the development of anaerobic, heterogenic dominated bio-
5mm
3.6 film observed after 12 months (Fig. 7). Therein, the initially developed
autotroph biofilm dominated by a succession of NSOBs and with de-
6.4 creasing pH by ASOBs is entrapped by secondary formed gypsum and
6.0 amorphous silica precipitates. Epifluorescence images of the dete-
5.6 rioration layers confirmed biofilm development within the entire cor-
rosion layers of the material (Fig. 8), which is in concordance with
5.2
findings of field observations previously reported by Grengg et al. [29].
CAC

4.8 With increasing thickness, oxygen penetration is limited and overall


4.4 anoxic to anaerobic conditions are established, triggering the coloni-
4.0 zation of heterotrophic bacteria, growing on the autotroph biofilm.
5mm 3.6 Within these late stages of corrosion, high corrosion rates have been
reported [2,24,39]. Accordingly, with gravitationally driven loss of
6.4 material and associated decrease in deterioration layer thickness, po-
6.0 tentially cycles of alternating autotroph and heterotroph dominated
5.6 biofilms established over time. Further extensive research on this topic
is necessary to confirm this hypothesis.
GP-K1T

5.2
Within recent years, several hypotheses were proposed explaining
4.8 the superior resistance of CAC materials in MIAC environments. While
4.4 several studies argue for combined solid state specific (formation of Al
4.0 (OH)x gel lowering porosity) and bacteriostatic effects of Al3+ [10,31],
5mm
3.6 more recent findings trace the superior performance of CAC vs. OPC
mainly back to the intrinsic resistance of phases initially present or
6.4 precipitated during the acid-driven deterioration process [33,40,41]. In
6.0 contrast, the bacteriostatic effects of Al3+ on the growth of relevant
5.6 ASOB could not be verified [41,42]. These new findings are largely in
agreement with the results obtained within this study [33,40]. EPMA
GP-K2

5.2
elemental mappings confirmed the dissolution of the CAC cementitious
4.8
microstructure within the low pH deterioration layers at low rates.
4.4 Interestingly, XRD results confirmed the presence of C3AH6 as a residue
4.0 of the initial CAC microstructure within the same areas in low con-
5mm
3.6 centrations. Furthermore, the formation of significant amounts of sec-
ondary X-ray amorphous phases within the deterioration layers with
Fig. 6. pH images of the OPC, CAC, GP-K1T and GP-K2 recorded using t-DLR time was observed, corresponding to the formation of Al(OH)x gel and
imaging technique after 18 month of exposure (right column), together with an amorphous Si-gels. The extent of Al(OH)x gel formation within the pore
optical picture of the same surfaces. spaces could not be fully verified. However, it is likely that this newly
formed gel effectively reduces open porosity and preserves cohesion in
the OPC and CAC materials comply with the well-described 3 step the CAC material, resulting in lower acid diffusion rates. These findings
corrosion model, first described by Islander et al., and in recent years correspond to field observations conducted by Khan et al. within a
adopted and extended by various studies [2,6–8,29,36]. Since the scope sewer system with similar aggressive environmental conditions over the
of this study is not to review these well-established models, the fol- period of 24 months [33]. The biofilm observed on the CAC materials
lowing discussion will focus on novel findings obtained from the pre- within this study was almost entirely composed of acid producing
sent study. In this context, the synergy of prevailing environmental ASOBs and expands throughout the entire deterioration horizons
parameters represents a central aspect controlling the duration of the (Fig. 8). At the same time, it exhibited by far the lowest biodiversity of
initial, (predominantly) abiotic phase of corrosion. While the impact of all tested materials (Fig. 9). Especially after 9 month of exposure, a low
H2S concentrations on initial surface pH reduction has been described H index of 1.85 was measured. This implies that Al3+ might have
as a key for MIAC [37,38], minor relevance has been attributed to CO2 certain inhibiting effects on the growth of symbiotic microorganisms,
and r.h. so far. To the knowledge of the authors, the first field data on such as heterotrophs and fungi, potentially effecting biogenic metabo-
CO2 concentrations over the durations of several days within MIAC lism mechanisms in the long term [43].
environments are presented herein. The analyzed high CO2 concentra- To the knowledge of the authors, the data presented in this study are
tions (regularily up to 6500 ppm), periodically low r.h. of roughly 70% the first one on Low-Ca- geopolymer/AAM binder systems tested in
and associated higher gas diffusion rates indicate their relevance for the MIAC field conditions. There exists one other study describing a long
strong surface pH reduction observed on the materials. The impact of term field exposure of high Calcium FA-based AAMs [19]. However,
CO2 is not only on carbonation or on acidity, but should be emphasized due to the very different chemistry and associated microstructure, as
as important on the initial development and continuing growth and well as due to the critical role of Calcium regarding dissolution and neo-
activity of biofilms on material surfaces. Therein, CO2 is an integral formation of secondary sulfate salts a direct comparability may be in-
compound used by the autotroph bacteria during chemosynthesis. The admissible. The observed deterioration mechanisms on the metakaolin-
second and third stages of corrosion observed on the OPC and CAC based GP materials can be characterized as follows: (i) Initial lowering
of the surface pH due to leaching of alkali (K+) cations (initially being

9
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

A 100 Other B 100


Pseudomonadaceae
Moraxellaceae
relative abundance (%)

relative abundance (%)


Halothiobacillaceae
Enterobacteriaceae
Burkholderiaceae
50 Shewanellaceae 50
Rhizobiaceae
Caulobacteraceae Other
Erysipelotrichaceae Acidithiobacillaceae
Peptostreptococcaceae Acetobacteraceae
Lachnospiraceae Peptostreptococcaceae
0 Clostridiaceae 1 0 Parachlamydiaceae
OPC CAC

C 100 D 100
Other
Rhodanobacteraceae
relative abundance (%)

relative abundance (%)


Pseudomonadaceae
Enterobacteriaceae
Burkholderiaceae
50 50 Acidithiobacillaceae
Other Rhizobiaceae
Acidithiobacillaceae Acetobacteraceae
Acetobacteraceae Peptostreptococcaceae
Erysipelotrichaceae Clostridiaceae 1
Peptostreptococcaceae Staphylococcaceae
Clostridiaceae 1 Corynebacteriaceae
0 0
GP_K1T GP_K2

Fig. 7. 16S bacterial communities after 12 months of in situ exposure of the indicated material samples. Bacterial communities were determined as described in
Materials and methods, the relative abundance of bacteria at the Family level is given. Families below 1% are grouped as “Other” (grey). Brown hues: Firmicutes;
pink, purple and blue hues: Proteobacteria; dark blue: autotrophs (neutrophilic SOB: Halothiobacillaceae, acidophilic SOB: Acidithiobacillaceae). (For interpretation of
the references to color in this figure legend, the reader is referred to the web version of this article.)

excess in pore solution), CO2 and H2S penetration inwards triggering precipitates as Si-rich gel [48]. EPMA mappings conducted within this
carbonation and acid-base reactions within the first months of exposure study support these findings, since higher Si concentrations, in respect
(Fig. 3). These reactions are expected to cause microstructural altera- to the intact GP framework, were measured within the deterioration
tions, such as decalcification of the partially formed C-A-S-H type layers of the GP materials (Fig. 5). This indicates that progressing acid
phases through formation of different calcium carbonate polymorphs penetration with time has led to stepwise/gradual congruent (com-
[44]. However, the degree of microstructural alteration within the GP plete) dissolution of the GP framework and immediate re-precipitation
materials used in this study is most likely limited to K+-leaching due to of Si-rich gel out of the strongly oversaturated solutions within these
low Ca contents of 0.8 and 1.6 wt% for the GP-K1T and GP-K2, re- boundary layers, in agreement with partial de-alumination of GP-ma-
spectively (Table A1) [45]. (ii) Reaching suitable conditions for mi- trix. In this context, the potassium-silicate solution content of the GP
crobial colonization, biogenic sulfuric acid diffusion led to the (more mixture, and thus excess of potassium and water (i.e. porosity, Table
excessive than in (i)) leaching of the charge-compensating alkalis (K+, A2), most likely plays a crucial role regarding the durability of the GP
part of GP structure) and subsequent replacement by hydronium material. Due to the (2.5 wt%) higher potassium content within the GP-
(H3O+) or other cations (e.g. aluminum species). Additionally, dis- K1T material (Table 1), increased amount of K leaching resulted in the
solution and transport of excess Si-monomers and dimers - not actively formation of higher quantities of K-Alum (Fig. A1). Resulting cracking
participating in the GP framework - occurs, which results in the pre- and associated higher porosity prevented the effective formation of the
cipitation of amorphous Si-gel within the low pH ranges of the mate- protective amorphous Si–rich (aluminosilicate) gel within the GP-K1T
rials. Thereby, the presence and thus the entailed amount of pre- material, as observed within the GP-K2 material. The development of
cipitates of such non-framework Si species is controlled by the Si/Al such amorphous, relatively low porosity Si–rich gel layers within the
ratio of the GP mixture [15,46]. (iii) With increasing acid production GP-K2 material are in accordance with the observed shift in micro-
and associated decreasing pH to below 4.0, de-alumination of the biological community structure towards anaerobic dominated bacterial
geopolymer framework microstructure due to the hydrolysis of the strains (Peptostreptococcaceae, At. ferrooxidans) between 12 and
Si–O–Al bonds proceeded within the outermost corrosion layers 18 months of exposure (Fig. 9). Gevaudan et al. hypothesized that the
(Fig. 5). This resulted in a reduction of the integrity and structural released supplementary cations (e.g. Ca2+ and Fe3+, and other heavy
strength of the geopolymer framework within these areas [47]. The metals) may act as replacements for the initial charge-balancing cations
detailed reaction mechanisms (e.g. congruent vs. incongruent dissolu- (here primarily K+), and may have a key share in inhibiting electro-
tion) leading to the development of these Al-depleted, Si-rich corrosion philic attack of GP gels by hydronium cations [47]. Therefore, the
layers are still the subject of discussion. Sturm et al. proposed that at higher iron and calcium content (Table A1) in the GP-K2 compared to
least a part of the silica is released into solution and subsequently GP-K1T -originating mainly from the different clay raw material used

10
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

white multichannel white multichannel

SYTO9
OPC CAC

1 cm
1 cm
PI

OPC CAC

1 cm
1 cm
SYTO9

GP-K1T GP-K2

1 cm 1 cm
PI

GP-K1T GP-K2

1 cm 1 cm

Fig. 8. Multichannel biofilm distribution images of OPC, CAC, GP-K1T and GP-K2 specimens after 18 months of exposure obtained from dividing the fluorescence
microscope image prior to staining (no stain, white) against the image obtained after staining (multichannel). Biofilm expands throughout the entire deterioration
layers of the different materials as indicated by green to orange color codes. (For interpretation of the references to color in this figure legend, the reader is referred to
the web version of this article.)

(Table 1) - could also explain the improved resistance of the GP-K2. microbial growth of relevant symbiotic heterotrophs and fungi; how-
More research is needed to answer these open questions, which are ever, its impact on material durability needs further systematic in-
raised by discussion of the results of the current research. vestigations.
The reasons for the large variation in resistance, observed between
4.2. Overall performance evaluation the two metakaolin-based GP materials, are multifaceted. Firstly, due to
the high Si content within the GP-K2, the presence of non-framework
After 18 months of exposure within a sewer system exhibiting building excess Si-monomers and dimers positively contributed to the
highly aggressive MIAC, strong variations in durability between the formations of described Si-rich gels. Secondly, the high amounts of fine
tested OPC, CAC and GP materials were observed. Average corrosion (partially) unreacted (quartz) particles in the K2 clay (values given in
depths of 9.5 mm (OPC), 3.2 mm (CAC), 20.0 mm (GP-K1T) and section on Material characteristics), are reported to act as a filler ma-
2.1 mm (GP-K2) were verified by BSE, elemental mappings and pH terial within the GP structure [49]. This enabled the GP-K2 synthesis
imaging (Fig. 4a, b, 5). After 12 months of exposure, all materials ex- with reduced amount of waterglass for same workability as in GP-K1T,
hibit highly acidic conditions suitable for the growth of ASOBs, which reducing open porosity from 13.1% to 12%, respectively. Both of these
were confirmed by microbiome analyses (Fig. 7). Thus all materials material properties effectively reduce the diffusion of corrosive sub-
reached the highest level of biological and chemical aggressiveness stances, such as the diffusion of acids. Additionally, the variations in
within the first year of exposure [2,6,7]. Accordingly, deterioration waterglass contents led to 2.5 wt% higher potassium concentrations
rates observed between 12 and 18 months are suitable to be used as an within the GP-K1T, compared to the GP-K2 mixture, and associated
indication for long-term durability predictions of the tested materials, higher K-leaching and K-Alum precipitation rates. This increased K-
assuming consistent (i.e. highly aggressive) environmental conditions. Alum precipitation is seen in the observed mass gain of GP-K1T of up to
Following this concept, average annual corrosion rates of 11.2 mm/a 4.3 wt% within the first 9 months of exposure, followed by a strong loss
for OPC, 5.4 mm/a for CAC, 28 mm/a for GP-K1T and 3.8 mm/a for GP- in mass due to increased cracking and acid penetration rates (Fig. 3).
K2 material were estimated. The high resistance of CAC material Accordingly, the effective formation of a protective layer of amorphous
against MIAC was mainly attributed to physiochemical and miner- Si-rich gels, clogging porosity, did not proceed. Additionally, higher
alogical material characteristics (see Corrosion mechanisms), which is SiO2/Al2O3 ratios within the GP-K2 mixture (3.7 compare to 2.3 in GP-
in accordance with [41,42]. Low biodiversity observed on the CAC K1T) (Table A1), resulted in associated higher framework stability
specimens potentially indicates a certain bacteriostatic effect of Al on under acid attack. Due to the amphoteric behavior of Al, exhibiting high

11
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Fig. 9. Distribution of dominant bacterial species on the investigated field samples. Bacterial communities were determined as described in Materials and methods,
the relative abundance of identified bacterial species is given. Only species that were found at a relative abundance of > 10% in at least one sample, are shown in the
heatmap matrix. White: no OTU/ASV present; blue: < 1%; green: 1–5%; yellow: 5–20%; red: > 20%. As an estimation for bacterial alpha-diversity, the Shannon
index (H´) is given. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

solubility within strongly acidic conditions, its concentrations within fungi.


the GP framework is most likely one of the key aspects determining GP • Geopolymer framework deterioration under MIAC conditions leads
durability in MIAC environments. to an almost complete dissolution of the Al and large parts of the Si
polymer framework and outwards diffusion followed by re-pre-
5. Conclusion cipitation of amorphous Si-rich gels at low pH degraded zones.
• The higher resistance of the Geopolymer GP-K2 material (compared
This study developed an interdisciplinary field approach including to GP-K1T) is attributed to the lower potassium silicate pro-
optical, chemical, microstructural, mineralogical and microbiological portioning and associated stability of the geopolymer framework
methodologies to comparatively assess the overall in-situ performance due to the higher SiO2/Al2O3 ratio, lower K+ leaching, higher iron
of CAC, GP and OPC materials within a strongly chemically aggressive content, as well as lower initial porosity.
sewer system. The main findings are summarized as follows:
CRediT authorship contribution statement
• High CO concentrations of up to 6500 ppm indicate its central role
2
in MIAC environments not only in respect to initial carbonation Cyrill Grengg: Conceptualization, Methodology, Investigation,
reactions, but also as a source for microbial metabolic reactions Validation, Writing - original draft, Visualization, Project adminis-
within later stages of corrosion. tration, Supervision. Neven Ukrainczyk: Validation, Resources,
• Strong shifts towards anaerobic biofilms over time were observed, Writing - original draft. Günther Koraimann: Validation,
wherein heterotroph bacteria within the thick deterioration layers of Investigation, Visualization, Writing - review & editing. Bernhard
the OPC material dominated the biofilms. This indicates alternating Mueller: Methodology, Visualization. Martin Dietzel: Writing - re-
microbial community structures within the later stages of corrosion, view & editing. Florian Mittermayr: Conceptualization,
regulated by impeded oxygen supply due to the formation of thick Methodology, Investigation, Writing - review & editing.
corrosion layers. A similar trend was observed within the microbial
community structures of geopolymer GP-K2, coinciding with the Declaration of competing interest
formation of low porosity due to the formation of amorphous Si-rich
gel layers, which are expected to reduce oxygen availability. There arises no conflict of interests by publishing this work.
• Obtained annual corrosion rates in descending order are: 28 mm/a
for the GP-K1T, 11.2 mm/a for the OPC, 5.4 mm/a for the CAC, and Acknowledgement
3.8 mm/a for the GP-K2 materials.
• Low corrosion rates of the CAC material were attributed to the in- The authors gratefully thank the Graz University of Technology
trinsic resistance of Al-rich phases initially present and precipitated (Austria) scientific grant program and TU Darmstadt for financial sup-
during the deterioration process and partial formation of Al(OH)x port. Financial support was greatly appreciated by Peter Rappold and
gel. Bacteriostatic effects on ASOBs, as sometimes reported in the the Department of Water Resources Management, Styria, as well as by
literature due to the high Al concentrations, were not observed, Heinz Lackner and the Department of Energy, Residential Constructions
however, the low biodiversity measured suggest certain effects on and Technology, Styria. Special thanks to Walter Fussi for providing the
the growth of symbiotic microorganisms such as heterotrophs and multiple access to the exposure site.

12
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Appendix A

Table A1
Chemical composition of mortar specimens obtained by XRF analyses (in wt%).

Mortar SiO2 Al2O3 CaO Fe2O3 TiO2 MgO Na2O K2O SO3

OPC 63.0 1.5 18.7 1.8 0.1 * 0.4 0.9 0.5


CAC 5.5 52.6 35.7 2.5 2.3 0.4 0.1 0.2 *
GP_K1T 53.5 23.4 0.8 1.8 0.8 0.4 0.8 11.8 *
GP_K2 60.7 16.3 1.6 3.0 0.7 0.3 0.2 9.2 *

Values below the detection limit.

Table A2
Compressive strength after 7 and 28 days, together with the open porosity of mortars after 28 days.

Material Compressive strength [MPa] Open porosity [%]

1 day 7 days 28 days 28 days

CAC 47.9 50.2 56.3 13.0


GP-KT1 50.1 63.8 68.4 13.1
GP-K2 48.7 61.5 60.4 12.0

Table A3
Shows the mineralogy of the deterioration layers (Qtz - quartz, Fsp – feldspar, Gp – gypsum, Katoite – C3AH6). Relative concentrations are indicated from x (trace
concentrations) to xxxx (main constituent).

Sample ID Qtz Fsp Gp Katoite Sulfur K-Alum

OPC_12m xxx x xxxx x


OPC_18m xxx x xxxx
CAC_12m xxxx x xx x
CAC_18m xxx x xxx x xx
GP-K1T_12m xxx x xxxx
GP-K1T_18m xxx x xxx
GP-K2_12m xxx x xxxx
GP_K2_18m xxx x xxxx

Fig. A1. Pore size distributions of the OPC, CAC, GP-K1T and GP-K2 materials analyzed by MIP.

13
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Fig. A2. Optical observation of exposed specimen after 6, 12 and 18 months of exposure.

Fig. A3. Change in mineralogical composition of the intact CAC paste in comparison with the deterioration layers after 12 and 18 months of exposure. Please note
that for demonstration purposes peaks are cut at an intensity level of 2500 a.u.

14
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Fig. A4. Change in mineralogical composition of the intact GP paste of GP-K1T and GP-K2 in comparison with the deterioration layers after 12 and 18 months of
exposure. Please note that for demonstration purposes peaks are cut at an intensity level of 2500 a.u.

15
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Fig. A5. Quantitative element concentration profiles of Al, Ca, Mg, Fe and S, calculated from the elemental distribution mappings of the OPC material after
18 months of exposure. The transition from strongly deteriorated areas (left hand side) to intact microstructure (right hand side) is at a depth of roughly 9 mm,
marked by strongly decreasing S and increasing Ca concentrations, as well as by accumulation zones of Al, Mg and Fe.

Fig. A6. Quantitative element concentration profiles of Al, Ca, Mg, Fe and S, calculated from the elemental distribution mappings of the CAC material after
18 months of exposure. Decreasing Al and Ca concentrations and contemporaneous increasing S concentrations at a depth of roughly 3 mm mark the transition
between intact microstructure (right hand side) and the deterioration layers (left hand side).

16
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Fig. A7. Quantitative element concentration profiles of Al, Ca, Fe, K and Si, calculated from the elemental distribution mappings of the GP-K2 material after
18 months of exposure. From the surface of the material (left hand side) until a depth of roughly 2.5 mm strongly decreasing concentrations of Al, K and Ca are
observed, together with a slight increase in Si concentrations. No change in Fe concentrations was observed.

17
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

18
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

Fig. A8. Epifluorescence images obtained of the materials after 18 months of exposure using either SYTO9 or PI as nucleic acid staining fluorescent dyes. False color
images for each sample reflecting the indicated fluorescence intensities are shown (no stain vs. SYTO9 or PI). The multichannel image represents an image created by
overlaying the fluorescence image prior to staining (no stain, red color) against the image obtained after staining (SYTO9/PI). Bright green areas indicate the
presence of bacterial biofilms within deterioration zones. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of
this article.)

References C.T. Brown, B.J. Callahan, A. Mauricio, C. Rodríguez, J. Chase, E. Cope, R. Da Silva,
P.C. Dorrestein, G.M. Douglas, C. Duvallet, C.F. Edwardson, M. Ernst, J. Fouquier,
J.M. Gauglitz, D.L. Gibson, A. Gonzalez, G.A. Huttley, S. Janssen, A.K. Jarmusch,
[1] T. Hvitved-Jacobsen, J. Vollertsen, C. Yongsiri, A.H. Nielsen, Sewer microbial B.D. Kaehler, K. Bin Kang, C.R. Keefe, P. Keim, S.T. Kelley, R. Ley, E. Loftfield,
processes, emissions and impacts, Sewer Process, Networks, Paris, 2002, p. 13. C. Marotz, B. Martin, D. Mcdonald, L.J. Mciver, V. Alexey, J.L. Metcalf,
[2] B.R.L. Islander, J.S. Devinny, A. Member, F. Mansfeld, A. Postyn, H. Shih, Microbial S.C. Morgan, J.T. Morton, A.T. Naimey, QIIME 2 : Reproducible, Interactive,
ecology of crown corrosion in sewers, J. Environ. Eng. 117 (1991) 751–770. Scalable, and Extensible Microbiome Data Science, (2018), https://doi.org/10.
[3] T. Hvitved-Jacobsen, J. Vollertsen, A.H. Nielsen, Sewer Processes - Microbial and 7287/peerj.preprints.27295.
Chemical Process Engineering of Sewer Networks, 2nd ed., CRC Press, London, [26] C. Quast, E. Pruesse, P. Yilmaz, J. Gerken, T. Schweer, P. Yarza, J. Peplies,
2013. F.O. Glöckner, The SILVA ribosomal RNA gene database project: improved data
[4] H.S. Jensen, PhD Thesis; Hydrogen Sulfide Induced Concrete Corrosion of Sewer processing and web-based tools, Nucleic Acids Res. 41 (2013) 590–596, https://doi.
Networks, Aalborg University, 2009. org/10.1093/nar/gks1219.
[5] C. Berger, C. Falk, F. Hetzel, J. Innekamp, S. Roder, J. Ruppelt, Zustand der [27] D.A. Strauss Rambo, N. Ukrainczyk, F. de Andrade Silva, E. Koenders, R.D. Toledo
Kanalisation in Deutschland: Ergebnisse der DWA-Umfrage 2015, Sonderdruck Filho, O. da Fonseca Martins Gomes, Calcium-aluminate mortars at high tempera-
Korrespondenz Abwasser, Abfall, 2016, https://doi.org/10.3242/kae2011/01.001. tures: overcoming adverse conversion effects using clinker aggregates, Cem. Concr.
[6] C. Grengg, F. Mittermayr, N. Ukrainzcyk, G. Koraimann, S. Kienesberger, Compos. 96 (2019) 212–224, https://doi.org/10.1016/j.cemconcomp.2018.12.
M. Dietzel, Advances in mineral based construction materials for sewer systems 002.
affected by microbiological induced concrete corrosion: a review, Water Res. 134 [28] S. Okabe, M. Odagiri, T. Ito, H. Satoh, Succession of sulfur-oxidizing bacteria in the
(2018) 341–352, https://doi.org/10.1016/j.watres.2018.01.043. microbial community on corroding concrete in sewer systems, Appl. Environ.
[7] M. Peyre-Lavigne, A. Bertron, C. Botanch, L. Auer, G. Hernandez-Raquet, A. Cockx, Microbiol. 73 (2007) 971–980, https://doi.org/10.1128/AEM.02054-06.
J.-N. Foussard, G. Escadeillas, E. Paul, Innovative approach to simulating the bio- [29] C. Grengg, F. Mittermayr, G. Koraimann, F. Konrad, M. Szabó, A. Demeny,
deterioration of industrial cementitious products in sewer environment. Part II: M. Dietzel, The decisive role of acidophilic bacteria in concrete sewer networks: a
validation on CAC and BFSC linings, Cem. Concr. Res. (2015), https://doi.org/10. new model for fast progressing microbial concrete corrosion, Cem. Concr. Res.
1016/j.cemconres.2015.10.002. (2017), https://doi.org/10.1016/j.cemconres.2017.08.020.
[8] M. Peyre Lavigne, A. Bertron, L. Auer, G. Hernandez-Raquet, J.-N. Foussard, [30] X. Li, G. Jiang, U. Kappler, P. Bond, The ecology of acidophilic microorganisms in
G. Escadeillas, A. Cockx, E. Paul, An innovative approach to reproduce the biode- the corroding concrete sewer environment, Front. Microbiol. 8 (2017) 683, https://
terioration of industrial cementitious products in a sewer environment. Part I: test doi.org/10.3389/FMICB.2017.00683.
design, Cem. Concr. Res. 73 (2015) 246–256, https://doi.org/10.1016/j.cemconres. [31] J. Herisson, E.D. van Hullebusch, M. Moletta-Denat, P. Taquet, T. Chaussadent,
2014.10.025. Toward an accelerated biodeterioration test to understand the behavior of Portland
[9] A.M. Goyns, M. Alexander, Performance of Various Concretes in the Virginia and calcium aluminate cementitious materials in sewer networks, Int. Biodeterior.
Experimental Sewer over 20 Years, Balkema, 2014. Biodegradation 84 (2013) 236–243, https://doi.org/10.1016/j.ibiod.2012.03.007.
[10] J. Herisson, M. Guéguen-Minerbe, E.D. van Hullebusch, T. Chaussadent, Influence [32] M. Alexander, A. Goyns, C. Fourie, Experiences with a full-scale experimental sewer
of the binder on the behaviour of mortars exposed to H2S in sewer networks: a long- made with CAC and other cementitious binders in Virginia, South Africa, Proc.
term durability study, Mater. Struct. 50 (2017) 8, https://doi.org/10.1617/s11527- Centen. Conf. Calcium Aluminate Cem, IHS, Avignon, 2008, pp. 279–292.
016-0919-0. [33] H.A. Khan, A. Castel, M.S.H. Khan, A.H. Mahmood, Durability of calcium aluminate
[11] J. Herisson, M. Guéguen-Minerbe, E.D. van Hullebusch, T. Chaussadent, Behaviour and sulphate resistant Portland cement based mortars in aggressive sewer en-
of different cementitious material formulations in sewer networks, Water Sci. vironment and sulphuric acid, Cem. Concr. Res. 124 (2019) 105852, , https://doi.
Technol. 69 (2014) 1502–1508, https://doi.org/10.2166/wst.2014.009. org/10.1016/j.cemconres.2019.105852.
[12] M. Valix, A.W.H. Cheung, J. Sunarho, H. Bustamante, The impact of calcium alu- [34] B. Drugă, N. Ukrainczyk, K. Weise, E. Koenders, S. Lackner, Interaction between
minate cement and aggregates on conversion and on filed performance in sewers, wastewater microorganisms and geopolymer or cementitious materials: biofilm
Calcium Aluminates, IHS, 2014, p. 663. characterization and deterioration characteristics of mortars, Int. Biodeterior.
[13] J. Stark, B. Wicht, Dauerhaf tigkeit von Beton, 2nd ed., Springer, 2013, https://doi. Biodegrad. 134 (2018) 58–67, https://doi.org/10.1016/j.ibiod.2018.08.005.
org/10.1007/978-3-642-35278-2. [35] P. Duxson, J.L. Provis, G.C. Lukey, S.W. Mallicoat, W.M. Kriven, J.S.J. Van
[14] F. Pacheco-Torgal, J. Labrincha, C. Leonelli, A. Palomo, P. Chindaprasit, Handbook Deventer, Understanding the relationship between geopolymer composition, mi-
of Alkali-activated Cements, Mortars and Concretes, 1st Editio, Woodhead crostructure and mechanical properties, Colloids Surfaces A Physicochem. Eng. Asp.
Publishing, 2014. 269 (2005) 47–58, https://doi.org/10.1016/j.colsurfa.2005.06.060.
[15] J. Davidovits, Geopolymer Chemistry and Applications, 4th ed., Springer, 2013. [36] G. Jiang, X. Sun, J. Keller, P.L. Bond, Identification of controlling factors for the
[16] B. Singh, G. Ishwarya, M. Gupta, S.K. Bhattacharyya, Geopolymer concrete: a re- initiation of corrosion of fresh concrete sewers, Water Res. (2015), https://doi.org/
view of some recent developments, Constr. Build. Mater. 85 (2015) 78–90, https:// 10.1016/j.watres.2015.04.015.
doi.org/10.1016/j.conbuildmat.2015.03.036. [37] X. Li, L. O’Moore, Y. Song, P.L. Bond, Z. Yuan, S. Wilkie, L. Hanzic, G. Jiang, The
[17] J.L. Provis, J.S.J. van Deventer, Geopolymers. Structures, Processing, Properties rapid chemically induced corrosion of concrete sewers at high H2S concentration,
and Industrial Applications, (2009), https://doi.org/10.1533/9781845696382. Water Res. 162 (2019) 95–104, https://doi.org/10.1016/j.watres.2019.06.062.
[18] R. Pouhet, M. Cyr, Alkali–silica reaction in metakaolin-based geopolymer mortar, [38] A.P. Joseph, J. Keller, H. Bustamante, P.L. Bond, Surface neutralization and H2S
Mater. Struct. Constr. 48 (2014) 571–583, https://doi.org/10.1617/s11527-014- oxidation at early stages of sewer corrosion: influence of temperature, relative
0445-x. humidity and H2S concentration, Water Res. 46 (2012) 4235–4245, https://doi.
[19] H.A. Khan, M.S.H. Khan, A. Castel, J. Sunarho, Deterioration of alkali-activated org/10.1016/j.watres.2012.05.011.
mortars exposed to natural aggressive sewer environment, Constr. Build. Mater. 186 [39] M. Kiliswa, Composition and Microstructure of Concrete Mixtures Subjected to
(2018) 577–597, https://doi.org/10.1016/j.conbuildmat.2018.07.137. Biogenic Acid Corrosion and Their Role in Corrosion Prediction of Concrete Outfall
[20] M. Luetscher, F. Ziegler, CORA – a dedicated device for carbon dioxide monitoring Sewers, University of Cape Town, 2016.
in cave environments, Int. J. Speleol. 41 (2011) 275–283. [40] M.W. Kiliswa, K.L. Scrivener, M.G. Alexander, The corrosion rate and micro-
[21] O. Vogt, N. Ukrainczyk, E.N. Koenders, Geopolymere als Spezialbaustoff, structure of Portland cement and calcium aluminate cement-based concrete mix-
Nachrichten Aus Der Chemie 65 (2017) 1198–1202 doi:12. tures in outfall sewers: a comparative study, Cem. Concr. Res. 124 (2019), https://
[22] C. Grengg, B. Mueller, C. Staudinger, F. Mittermayr, J. Breininger, B. Ungerböck, doi.org/10.1016/j.cemconres.2019.105818.
S. Borisov, T. Mayr, M. Dietzel, Optical pH imaging of concrete exposed to che- [41] A. Buvignier, C. Patapy, M.P. Lavigne, E. Paul, A. Bertron, Resistance to biodeter-
mically corrosive environments, Cem. Concr. Res. 116 (2019) 231–237, https://doi. ioration of aluminium-rich binders in sewer network environment: study of the
org/10.1051/matecconf/201819902007. possible bacteriostatic effect and role of phase reactivity, Cem. Concr. Res. 123
[23] B. Müller, C. Grengg, V. Schallert, M. Sakoparnig, C. Staudinger, J. Breininger, (2019) 105785, , https://doi.org/10.1016/j.cemconres.2019.105785.
F. Mittermayr, B. Ungerböck, S. Borisov, M. Dietzel, T. Mayr, Wide - range optical [42] A. Buvignier, M. Peyre-Lavigne, O. Robin, M. Bounouba, C. Patapy, A. Bertron,
pH imaging of cementitious materials exposed to chemically corrosive environ- E. Paul, Influence of dissolved-aluminum concentration on sulfur-oxidizing bac-
ments, RILEM Tech. Lett. 3 (2018) 39–45, https://doi.org/10.21809/rilemtechlett. terial activity in the biodeterioration of concrete, Appl. Environ. Microbiol. 85
2018.72. (2019) 1–12, https://doi.org/10.1128/aem.00302-19.
[24] C. Grengg, F. Mittermayr, a. Baldermann, M.E. Böttcher, A. Leis, G. Koraimann, [43] T. Mori, T. Nonaka, K. Tazaki, M. Koga, Y. Hikosaka, S. Noda, Interactions of nu-
P. Grunert, M. Dietzel, Microbiologically induced concrete corrosion: a case study trients, moisture and pH on microbial corrosion of concrete sewer pipes, Water Res.
from a combined sewer network, Cem. Concr. Res. 77 (2015) 16–25, https://doi. 26 (1992) 29–37 (doi:0043-1354’92 $5,00 + 0.00).
org/10.1016/j.cemconres.2015.06.011. [44] S.A. Bernal, X. Ke, M. Criado, S. Mundra, J.L. Provis, Factors controlling carbona-
[25] J.R. Rideout, M.R. Dillon, N.A. Bokulich, C. Abnet, G.A. Al Ghalith, H. Alexander, tion resistance of alkali-activated materials, Proc. 10th ACI/RILEM Int. Conf. Cem.
E.J. Alm, M. Arumugam, Y. Bai, J.E. Bisanz, K. Bittinger, A. Brejnrod, J. Colin, Mater. Altern. Bind. Sustain. Concr, 2017.

19
C. Grengg, et al. Cement and Concrete Research 131 (2020) 106034

[45] S.A. Bernal, J.L. Provis, B. Walkley, R. San Nicolas, J.D. Gehman, D.G. Brice, cobalt improve the acid resistance of alkali-activated cements, Cem. Concr. Res. 115
A.R. Kilcullen, P. Duxson, J.S.J. Van Deventer, Gel nanostructure in alkali-activated (2019) 327–338, https://doi.org/10.1016/j.cemconres.2018.08.002.
binders based on slag and fly ash, and effects of accelerated carbonation, Cem. [48] P. Sturm, G.J.G. Gluth, C. Jäger, H.J.H. Brouwers, H.C. Kühne, Sulfuric acid re-
Concr. Res. 53 (2013) 127–144, https://doi.org/10.1016/j.cemconres.2013.06. sistance of one-part alkali-activated mortars, Cem. Concr. Res. 109 (2018) 54–63,
007. https://doi.org/10.1016/j.cemconres.2018.04.009.
[46] P.J. Gevaudan, K.M. Campbell, T.J. Kane, R.K. Shoemaker, W.V. Srubar, Cement [49] O. Vogt, N. Ukrainczyk, C. Ballschmiede, E. Koenders, Reactivity and micro-
and concrete research mineralization dynamics of metakaolin-based alkali-acti- structure of metakaolin based geopolymers: effect of fly ash and liquid/solid con-
vated cements, Cem. Concr. Res. 94 (2017) 1–12. tents, Materials (Basel) 12 (2019) 1–21, https://doi.org/10.3390/ma12213485.
[47] J.P. Gevaudan, A. Caicedo-Ramirez, M.T. Hernandez, W.V. Srubar, Copper and

20

You might also like