Construction and Building Materials 364 (2023) 129933

Contents lists available at ScienceDirect

Construction and Building Materials

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

Graphene oxide’s regenerative acidity and its effects on the hydration of Type
II Portland Cement
Taimur Mazhar Sheikh a,b ,∗, Mohammed Parvez Anwar b , Kasturi Muthoosamy c ,∗,
Jayaprakash Jaganathan d , Andy Chan b , Abdullahi Ali Mohamed b
a
Department of Civil Engineering, Wah Engineering College, University of Wah, Wah Cantt, Pakistan
b Department of Civil Engineering, University of Nottingham Malaysia, Semenyih, Malaysia
c Nanotechnology Research Group, Centre of Nanotechnology and Advanced Materials, University of Nottingham Malaysia, Semenyih, Malaysia
d School of Civil Engineering, Vellore Institute of Technology, Vellore Campus, Vellore, India

ARTICLE INFO ABSTRACT

Keywords: Incorporation of graphene oxide (GO) has been found to considerably improve the hydration process, strength
Nanomaterials and durability of Portland cement. However, the exact nature of its chemical interactions with the cement are
Graphene oxide unclear. In this study, GO of varying amounts of hydroxyl groups were synthesized to investigate the effects
Concrete
of these interactions on the hydration of an environment-friendly Type II Portland Cement. XPS, Raman, and
Cement
FTIR analysis verified the functional group differences between the GO types, and SEM and AFM observations
MicroCT
Functionalized
illustrated the existence of a hydronium layer coating the high-hydroxyl GO (HGO). The hydronium layer
neutralizes small base additions as measured through titration, and regenerates via protonation of resulting
water by HGO’s hydroxyl groups, confirmed via zeta potential analysis. In cement, HGO shows accelerated
and greater early-age hydration, measured via heat of hydration and XRD. Finer microstructural density of
HGO-cement was also observed from BET and microCT analysis. On the other hand, low-hydroxyl GO (XGO)
cement ‘locks’ C-S-H on the GO sheet, preventing its propagation in the cement microstructure, as observed by
Si-O bond changes during hydration. Lastly, HGO-concrete showed significantly improved workability (>40%),
28-day compressive strength (29%), and 28-day flexural strength (24%) with respect to control. Conversely,
XGO-concrete showed reduced workability (−40%), and smaller 28-day strength improvements (compressive
by 5%, and flexural by 8%). This research leads to new understandings of how GO may improve the strength,
workability, and durability of concrete, with potentially less overall cement consumption and superplasticizer
use.

1. Introduction typically arranged in a hexagonal, honeycombed pattern. Graphene

sheets with added oxygen atoms in hydroxyl (C-OH), epoxy (C-O-
Portland Cement (PC) is widely used in the concrete construction C), carbonyl (C=O), and carboxyl (O=C-OH) formations result in GO.
industry due to its cost-effectiveness, versatility and ease of application. These added functional groups make GO polar and easily dispersible
Its production, however, is not environmentally sustainable, emitting
in water. As cement requires access to water molecules for hydration,
approximately 8% of the world’s carbon dioxide (CO2 ) [1]. PC reacts
GO’s polarity concentrates a large amount of water molecules in its
with water in a process called hydration, forming a binding agent
that holds aggregates together in concrete (curing), resulting in high proximity, providing ‘seeding’ points for cement particles to latch on.
compressive strength and good durability. However, not all PC in This allows quicker and higher overall hydration, and therefore pro-
the structure reacts with water, with certain high-cement concretes duces a cement strengthening effect with denser microstructure [4–6].
only using 30%–50% of the total cement during the 28-day curing Additionally, GO sheets can interlink in alkaline solutions, connected
process, a considerable waste [2]. Thus, cement wastage can be reduced by free-floating divalent calcium (Ca2+ ) ions in aqueous solutions [7].
by maximizing its total hydration potential in structures, and minute Free-floating Ca2+ is also found in cement matrix during the curing
additions of graphene oxide (GO) nanoparticles have shown to improve process, and PC is considerably weak under tensile and flexural loading,
the hydrated cement microstructure significantly [3]. GO is an oxidized
hence interlinking can possibly provide higher flexural resistance to
form of graphene, which is a 2D sheet structure of carbon atoms

∗ Corresponding authors.
E-mail addresses: [email protected] (T.M. Sheikh), [email protected] (K. Muthoosamy).

https://doi.org/10.1016/j.conbuildmat.2022.129933
Received 8 August 2022; Received in revised form 24 November 2022; Accepted 28 November 2022
Available online 12 December 2022
0950-0618/© 2022 Elsevier Ltd. All rights reserved.
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

cement [8,9]. Despite the known advantages of GO-cement, a key issue of ice cubes to keep the temperature below 60 ◦ C. The quantity of
of GO incorporation is the unpredictability of its strength and durability H2 O2 varied for each GO type: OGO required 5 mL of H2 O2 until the
improvements in cement [10], and this research aims to identify the colour turned pale yellow and temperature rise subsided, indicating
chemical interactions between the GO and cement during hydration. oxidation reaction termination. The initial filtrate was brown in colour
To understand how GO is improving cement hydration, its role as and darkened with successive washing. For high-hydroxyl graphene
a ‘seeding’ point should be addressed. GO sheets have typically high oxide (HGO), the oxidative treatment is similar to OGO, except after the
aspect ratios, with narrow sheet thickness (approximately 1 nm) and intercalation period (10 min of stirring of graphite in 9:1 H2 SO4 :H3 PO4
varying lengths from 500 nm to several μm [11]. The hydroxyl and acid mix), where 10 mL of distilled water is added slowly to the
epoxy groups of GO mainly exist on the surface of the sheets, while mix as per Chen et al. [17], followed by KMnO4 (6 wt. equivalent)
carboxyl groups are found at the edges. That is because carboxyl groups and the remaining procedure is similar to OGO. For termination, it
require 3 covalent bonds from each carbon atom, which is only possible is found only 3.5 mL of H2 O2 was required for the mix to turn pale
at the terminal edges of the GO sheets. Hence, it can be inferred that yellow, and subsequent addition of H2 O2 resulted in no colour change
the hydroxyl and epoxy groups are primarily responsible for attracting or rise in temperature. The decontamination and sonication of HGO
the polar water molecules around GO. Indeed, molecular simulations was followed similar to OGO. Noteworthy, HGO was lighter brown in
indicate GO sheets are easily dispersed by hydrogen bond networks colour than OGO, however the solution darkened over a period 1–2
between GO’s surface functional groups and water molecules [12]. First weeks, as did all the GO variants. For low-hydroxyl graphene oxide
principle simulations also show active proton (H+ ) transfer between (XGO), the intercalation and oxidation procedures are similar to OGO,
the GO hydroxyl groups and surrounding water molecules [13], which however after 120 h and prior to termination, 100 ml of water was
leads to an interesting aspect of GO: its acidity in aqueous solutions. GO slowly added and the GO mix was heated to 95 ◦ C and maintained
solutions are shown to have pH between 2 and 3, lower than saturated at that temperature for 15 min [17]. Visually, the mix changed colour
carboxylic acids [14]. As GO typically has few edge carboxyl groups, from deep purple to dark mahogany brown, with pale yellow frothing
the acidity must be from the proton transfer between hydroxyl groups was observed during the heat treatment. After 15 min, the mix was
and water molecules. Moreover, GO tends to undergo decarboxylation allowed to cool and any addition of H2 O2 did not change its colour
in alkaline conditions, which could be due to the vinylogous activity or cause temperature rise. All three GO solutions were allowed to cool
of the GO sheet and hydroxyl groups [15]. A dynamic interchange of to room temperature and decontamination process was initiated, first
H+ is found between GO’s hydroxyl groups and water, leading to an by centrifuging each solution at 7500 rpm for 30 min, after which the
equilibrium of H2 O and protonated hydronium (H3 O+ ) surrounding supernatant was discarded. The filtrate was washed with 1 M HCl and
the GO sheet [14]. The opposite charges of deprotonated hydroxyl centrifuged at 7500 rpm for 30 min (eppendorf 5810R), after which the
groups (C-O− ) and H3 O+ result in a dynamic hydronium layer being filtrate was extracted and subjected to 2 more rounds of washing and
electrostatically attracted to GO, which has possibly been observed as centrifugation with 1 M HCl. After 3 repeats, the filtrate was washed
oxidative debris in earlier research [16]. with distilled water and centrifuged (7500 rpm) for 30 min, and this
It is proposed that this hydronium layer is the primary seeder for step was repeated for 7 times. After final centrifugation, the filtrate was
C-S-H nucleation and growth during cement hydration, as it is the first added to 100 mL distilled water and stirred manually until all visible
point of contact for the dissociating cement particles. GO is exposed particles were mixed evenly. It was then placed in an ultrasonicating
to cement only when the layer is successfully deprotonated, which is bath and sonicated for 20 min (Elma Transsonic 35 kHz, 40% intensity).
an expected outcome due to the alkaline by-products of the hydra- Finally, distilled water was added until total volume reached 500 mL.
tion reaction. As the hydronium layer originates from GO hydroxyl
groups’ reaction with water molecules, it follows that the amount of 2.2. Analytical tests
hydroxyl group on GO’s surface can affect its subsequent contribution
to cement hydration. Hence, using Chen et al.’s [17] findings, three GO For FTIR and SEM, all GO samples were heated to 65 ◦ C for 48 h
with different amounts of hydroxyl groups were synthesized: a high- prior to testing (oven-dried samples). Moreover, separate GO samples
hydroxyl GO (HGO), a low-hydroxyl, partly reduced GO (XGO), and GO were freeze dried for 72 h and stored in a desiccator with silica gel
synthesized using Marcano’s [18] modified Hummer’s method (OGO), until FTIR/SEM was conducted. FTIR (Perkin Elmer Frontier) analysis
which has a moderate amount of hydroxyl groups. Their acidities was carried out with 10 scans from 4000 to 400 cm−1 wavelengths.
are measured, and their effects on the hydration and microstructural The spectra was averaged and base corrected using the Perkin Elmer
development of cement, and the strength and workability of concrete Spectrum 10 software. FEI Quanta 200 3D Dual Beam FIB-SEM machine
is investigated. was used for imaging of GO and GO-cement samples. The Quantera II,
ULVAC-PHI XPS was used with an Al K𝛼 X ray source (25 W, voltage
2. Methodology source 15 kV) with beam size 100 μm at tilt angle of 45◦ for far and near
XPS spectra scan (C1s and O1s). Pass energy was kept at 112 eV, with
2.1. Synthesis of GO a dwell time of 20 ms per step. The 633 nm wavelength was used to
acquire the D and G peaks from Raman testing. 1 mg/mL GO solutions
For ordinary graphene oxide (OGO), as per Marcano et al. [18] were spin-coated on glass cover-slips using a KISKO Specialty Coating
13.4 mL 85% phosphoric acid (H3 PO4 , Chemiz) was added to 120 System 6800 machine (3000 rpm for 60 s) for sheet size measurements
mL of 95%–97% concentrated sulphuric acid (H2 SO4 , Fulltime) (9:1 via SEM and AFM analysis (Hitachi AFM5100N, using Dynamic Force
ratio) and the mixture was allowed to cool to 20 ◦ C. 1 g of graphite Microscopy).
powder (< 20 μm Sigma-Aldrich) was added and the mixture was stirred 50 mL of 1.5 mg/mL concentrations of each GO samples were added
at 300 rpm for 10 min to allow intercalation. After 10 min, 6 g of in a 250 mL beaker and subjected to magnetic stirring at 200 rpm. A
solid potassium permanganate (KMnO4 , Chemiz, 6 wt. equivalent) was Mettler-Toledo FiveEasy pH meter was calibrated and used to measure
added slowly to the intercalated graphite mix over a period of 5 min. pH changes. Nitrogen gas was bubbled through the stirring liquid at
During this period, the mix warmed to approximately 35 ◦ C, after a constant rate. The beaker was sealed with a rubber septum, while
which it was kept under constant stirring in a fume hood at a room holes were made for the pH meter, the gas tube, and an inlet/outlet
temperature of 20 ◦ C for 120 h. The temperature was not increased tube for addition of base/acid and/or extraction for zeta potential mea-
as per Chen et al. findings [17] where keeping a low temperature surements. To facilitate liquid extraction for zeta, it was not feasible to
typically increases the yield of GO. After 120 h, 30% hydrogen peroxide keep the beaker/container completely airtight, however this setup does
(H2 O2 , R&M Chemicals) was added drop-wise alternating with 135 mL allow CO2 expulsion during titration, while keeping the environment

2
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

free from unwanted contamination. 1 M potassium hydroxide (KOH, was set at 0.35 with 0.04% addition of each GO separately. After 14
R&M Chemicals) was added to the solution by pipetting 50 μL each days of curing in a moist temperature controlled environment, these
time, allowing the pH to stabilize before taking measurements and samples were subjected for microCT scans using the Scanco Medical
extracting 0.50 mL for zeta potential measurements. To better observe μCT35 instrument (Energy/intensity at 70 kVp, 114 μA, 8 W), with high
the regenerative ability of the GO-hydronium layer, two separate pH resolution and voxel size of 10 μm. A 7 mm × 7 mm × 2.32 mm rectan-
readings were recorded, once after 10 s of pH stabilization, and another gular block was cropped from the scanned results to ignore anomalies
after 20 s. The pH reading is considered stable when after 10 or 20 on the surface of the cured samples. These scans were processed using
consecutive seconds, the pH stays within ±0.01, respectively. For zeta Slicer software, and the pores were isolated using image thresholding
potential measurements, 0.5 mL of each solution was pipetted into of the scanned slices, ignoring any pores smaller than 50 voxels. 3D
a DTS1070 folded capillary cell which was inserted into a Malvern- segments of each pore were constructed to obtain their volumes for
Panalytical zetasizer nanoZS. The cell was equilibrated for 30 s at 25 ◦ C pore analysis.
and 3 runs were taken, for 10 measurements each. The Smoluchowski The concrete was prepared in a weight ratio of 0.0004:0.009:
equation was used to calculate the zeta potential. 0.43:1:1.96:3.18 of GO:polycarboxylate superplasticizer:water:cement:
fine aggregates:coarse aggregates, respectively. The mix design was
2.3. Cement microstructural analysis, heat and strength tests based on the BRE design method [19] (for control), aimed for a
target strength of 40 MPa with a slump of 20 mm. The high amount
Portland Cement Composite (YTL Castle, CEM II/B-L 32.5N, cer- of polycarboxylate superplasticizer was required to ensure the target
tified MS EN 197-1:2014) was used in all final cement and concrete slump of XGO-concrete. No GO was included in control samples. 4
samples. Fine and coarse aggregates were locally sourced from the mg/mL aqueous GO suspensions were subjected to ultrasonication for
Negeri-Sembilan quarry (river sand) in Malaysia with fineness modulus 15 min, before pouring into the remaining water for concrete mix,
of 2.35, bulk density of 1350 kg/m3 , and specific gravity 2.6. For followed by polycarboxylate superplasticizer addition. Meanwhile, the
BET, FTIR, XRD, and SEM analysis, hydrated cement samples (control required amounts of cement and aggregates were weighed and dry
included) with a w/c ratio of 0.43 and polycarboxylate superplasticizer mixed in a pan mixer. The GO+superplasticizer was mixed for a few
(Sika Viscocrete 1250NT, liquid) amount of 0.9% (by weight of cement) seconds, then poured slowly into the dry mix while the mixer was
were prepared at 3, 6, 12, 24, 72 and 672 (28 days) hours of hydration. switched on. Total wet mixing time was controlled to maximum 10 min
All GOs were added at 0.04% amount (by weight of cement) to distilled for each sample, after which the machine was switched off and the mix
water, followed by the superplasticizer. The resulting solution was was allowed to set. After 10 min the mix was poured (in 3 batches)
sonicated for 5 min before adding to cement for hydration. To terminate into a slump cone (305 mm height, 100 mm top diameter and 200 mm
the hydration process (at the designated hours) as quickly as possible, bottom diameter) and tamped 25 times with a metal rod for each
the samples were manually crushed and freeze-dried for 72 h (Martin poured batch, before lifting the cone and measuring slump. The final
Christ, −11 ◦ C at 0.01 mbar), after which they were grinded (Retsch slump value was recorded when consecutive slump tests for each type
ZM200 ultracentrifugal mill, 0.2 mm sieve) and placed in a desiccator did not vary by more than 1 mm: most sample mixes gave the same
with silica gel until time of testing. The PANalytical X’Pert pro XRD was consecutive slump after 3 tests, with the XGO based concrete mixes
used for analysing the respective samples, over a rotation of 10◦ to 80◦ took 4 slump measurements to show similar consecutive slump. The
with a step size of 0.02◦ and scan time of 0.15 s, using Cu anode with mix was then poured (in 3 batches) into three respective 100 mm
K-𝛼1 setting at 1.54060. Profex-BGMN software was used to perform cubes and one 100 × 100 × 500 prism moulds, and subjected to 10 s
Rietveld Refinement on the XRD scan data. The monoclinic Mumme vibrations via vibrating table for each batch. Once the moulds were
and triclinic Belov structures were used for alite quantification. The fully filled and vibrated, they were left to harden for 24 h, after which
micrometrics ASAP 2020 surface area and porosity analyzer was used they were demoulded and placed in a water pond (pH 11) and water
to conduct BET tests for 24 h freeze-dried samples only. cured for 28 days. The Universal unit testing machine was used to
In the absence of an air-calorimeter (for isothermal calorimetry perform 3-point bending tensile tests on the prism and compressive
of hydrating cement), two DS18B20 (Maxim Integrated, 0.5 ◦ C sen- strength test on the cubes, respectively. It should be noted that flexural
sitivity) water-proof temperature sensors were connected to Arduino and slump tests are imprecise due to their testing methods/equipment
UNO micro-controller. One sensor was submerged in the hydrating and low number of samples, however concrete tests were performed
cement mix to record the temperature changes over an approximately solely to show how microstructural analysis results from cement can
24 h period, while the other was kept outside the sample to record be translated effectively to concrete.
ambient temperature, and measurements were taken at 30 s intervals.
Briefly, the hydrating cement’s w/c ratio was reduced to 0.35, and 3. Results
superplasticizer amount reduced to 0.3% by weight of cement. All GOs
were added at 0.04% amount (by weight of cement) to distilled water, 3.1. GO characterization
followed by the superplasticizer. The wet cement mix was manually
stirred and kept in a plastic cup inside a 500 ml glass beaker, which Analytical characterization tests were performed to verify the chem-
was placed inside a larger 2 L glass beaker, surrounded by insulating ical compositions of HGO, OGO, and XGO. X-ray Photoelectron Spec-
fabric to retain heat and minimize ambient temperature changes. The troscopy (XPS) analysis is used to distinguish between the different
increase in heat was calculated by simply subtracting the external chemical bonds of each GO, and their C1s peaks are shown in Fig. 1(a).
sensor’s temperature reading from the internal sensor, and converted As the bond peaks overlap, peak-fitting is used to separate the spectra,
to joules by 𝑄 = 𝑚1 𝑐1 𝛥𝑇 + 𝑚2 𝑐2 𝛥𝑇 , where the mass, 𝑚 and specific heat and the results can be seen in Fig. 1(b). HGO and OGO show similar
capacities, 𝑐 are of water and cement, respectively and 𝑇 is the recorded peaks in XPS, while XGO-concrete shows significantly fewer counts per
change in internal–external temperature readings. The obtained graphs second. OGO has higher peak counts than HGO, however HGO has a
were smoothed using a central moving average (uniform, rectangular) higher C-O peak than C–C, indicating more hydroxyl group formation.
of 300 points. As such, these reading are meant to illustrate a signif- HGO also has more carbonyl (C=O) groups than OGO. Meanwhile,
icant change in hydration heat between control and GO cements, but XGO shows two separate peaks for C(O)O, indicating high carboxyl
however data is not precise enough for quantifiable analysis, and is not formation. As carboxyl groups are formed at the edges of GO sheets, and
used as such accordingly. based on the poor counts in XPS analysis, XGO seems to be structurally
10 mm diameter × 5 mm height cylindrical cement samples were defective. To better gauge this structural integrity for all GOs, Raman
prepared for X-ray micro-computed tomography analysis. The w/c ratio spectroscopy was performed (see Fig. 1(c)). The graphitic G peaks

3
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

Fig. 1. (a) The C1s XPS scan of HGO, OGO and XGO, with (b) isolated peak fittings for each bonds’ respective binding energy. (c) Raman spectra with D and G peaks of GO
samples, and (d) FTIR spectra of freeze-dried GO. Please note HGO, OGO, and XGO are vertically offset in each graph for clearer peak size comparison.

indicate intact, non-oxidized regions in the GO sheets, while the D reacting with the surface groups on GO [14]. Moreover, this layer can
peaks show ‘defects’, caused by sp3 bonding of O to the C atoms of GO. easily be destroyed by freeze-drying of GO. Hence, slow oven-dried
Hence, the ratio of the intensity of these peaks is calculated as ID /IG , a and freeze-dried GO samples were tested to estimate the removal of
measure of the intactness of the GO sheets. HGO shows the lowest ID /IG hydronium, with the former is prepared by drying the GO without
(0.86), indicating high structural integrity, followed by XGO (0.99) and excessive hydronium removal. Electron Dispersive X-ray (EDX) analysis
OGO (1.07), respectively. However, this test is also inconclusive, as of oven-dried GO samples show C/O ratios of 0.25 (for HGO), 1.42
higher ID /IG of GO can indicate either high oxidation degree or more (OGO), and 1.60 (XGO) (Supplemental Figure S1). HGO’s ratio is not
structural defects in the GO sheet. Furthermore, research has indicated feasible, as there cannot be more than one O atom attached to a C
XPS and Raman analysis can reduce surface functional groups of GO atom in a GO sheet, or more than two O atoms attached to the C
during the measurements itself, either via prolonged exposure to X- atom at the terminating edges of the GO sheet. Therefore, the excess
ray sources, or thermal annealing during sample preparation [20]. The oxygen count must be registering from the hydronium layer around
samples were then analysed using Fourier-transform Infrared (FTIR) HGO. FTIR analysis also shows significantly higher O-H absorption for
spectroscopy to further estimate the presence of functional groups the oven-dried HGO relative to freeze-dried OGO (see Supplementary
(see Fig. 1(d)). The high hydroxyl and epoxide presence is obvious in Figure S2), confirming HGO to have more surrounding hydronium than
HGO, shown by high IR absorption between 1000–1200 cm−1 , which OGO or XGO, and this layer is removed freeze-drying. Atomic Force Mi-
is attributed to C-O and C-O-C bond stretching. High O-H presence can croscopy (AFM) analysis is conducted to get visible confirmation of any
also be seen at 3250 cm−1 bending and 1620 cm−1 stretching dips, surrounding hydronium, shown in Fig. 2(a) and Supplementary Figure
however O-H is also present in any residual hydronium surrounding the S3. 3D profile of the AFM scans (see Fig. 2(a)) highlight uniform bright
GO, hence C-O bond absorption is a better indication of hydroxyl and spots on HGO’s sheets reaching a height of 12 nm. OGO contains far
epoxide presence in HGO. High C–C bond absorption is also observed fewer stacks reaching a height of 5 nm, followed by XGO having almost
for HGO, indicating high structural integrity and verifying the Raman no bright spots at all. This is possibly the hydronium layer, though it
spectral analysis. Thus, HGO is concluded to have high hydroxyl (and may also be some oxidative debris. Furthermore, the structural defects
epoxide) groups, followed by OGO and XGO, respectively. created by XGO’s thermal reduction are explicitly visible as holes and
leaf-like shapes. Noteworthy, well dispersed, smaller GO sheets have
3.2. The hydronium activity of GO shown to contribute higher available surface area for cement hydration,
resulting in a denser and stronger cement structure [21]. Thus, to
The next step is to confirm whether high hydroxyl groups are account for the sheet sizes, Scanning Electron Microscopy (SEM) tests
producing a denser hydronium layer around the GO sheets. This layer is are performed on mica sheets spin coated with GO samples. HGO
challenging to measure, as it exists in a dynamic equilibrium constantly has the largest sheet with varied sizes (1.5–11 μm), followed by XGO

4
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

Fig. 2. (a) The AFM 3D scan and thickness profile of HGO, OGO, and XGO. (b) SEM of HGO shows high water/hydronium stacking on oven dried HGO, and respective bulging
deformation when freeze dried, (c) the pH ranges of all GO samples upon titration, and (d) surface charge increase showing hydronium regeneration for HGO, not prominent in
OGO or XGO.

(2–8 μm) and OGO (1–3 μm), respectively (see Supplementary Figure also required more than twice the amount of [OH]−1 to remove its
S4). SEM imaging shows visible evidence of stacked hydronium on hydronium layer completely.
HGO, which were removed upon freeze-drying, causing bulging and
deformation of the GO sheets (Fig. 2(b)). 3.3. Illustration of hydronium activity on GO

3.2.1. Titration and zeta potential analysis To fully explain the significance of the titration and zeta potential
Based on the above tests, the hydronium layer is the largest in results, the unique sheet structure of GO must be addressed first. The C
HGO, followed by OGO and XGO, respectively. Another way to quantify atoms are typically connected to three other C atoms in the hexagonal
hydronium is by measuring the amount of base addition needed for graphene sheet structure. As carbon has 4 valence electrons, it can form
complete pH neutralization of an aqueous GO solution. Hence, all a maximum of four covalent bonds, hence GO has O atoms grafted to
GO samples were titrated, adding small amounts of 1M potassium these C atoms. As the H+ interchange causes the build-up of hydronium
hydroxide (KOH). However, GO’s dynamic hydronium layer exists in layer, the deprotonated GO’s hydroxyl groups can reform to epoxide
an equilibrium state in water [14], so it follows that addition of base groups, so long as a neighbouring C atom has the available covalent
would disrupt this equilibrium, affecting the amount of GO hydroxyl bonding [13]. Conversely, epoxides open under presence of acid (such
deprotonation to create the hydronium layer. Assuming the change as hydronium) to form two separate hydroxyl groups. This illustrates
in equilibrium would take time to adjust, two sets of titration were how the dynamic hydronium layer may be affecting the types of
performed for each GO sample. The first set recorded the pH value after functional groups on GO’s surface. Furthermore, the hexagonal arrange-
each base addition once it stabilized for 10 consecutive seconds, while ment of C atoms in graphene allows valence electrons to freely travel
the second set recorded the pH value after 20 consecutive seconds of between the atoms, resulting in an aromatic ring structure with high
a stable reading. The reading is considered stable if the pH does not thermal and electrical conductivity. For GO, the oxidized zones hinder
fluctuate by more than 0.01 for the time duration. HGO has a signif- this electron transfer between oxidized and free C atoms, resulting in
icantly bigger pH range between the two titration tests, followed by alternating C–C and C=C bonds through the oxidized sheets. These C=C
OGO and XGO, respectively (Fig. 2(c)). This is remarkable: for instance, bonds can shift through the structure, facilitated by the epoxide forma-
0.005 mM base addition causes pH to rise to 9 (in the 10 s stability tions or openings, called vinylogous activity [14,23,24] (Fig. 3(a)). It
tests), but slowly gets neutralized back to pH 2.5 over 20 s stability is believed that these bond changes are responsible for determining the
readings. In addition, HGO’s initial pH before any base addition is equilibrium condition of the hydronium layer. HGO is more acidic than
closer to 2, while OGO and XGO are approximately 3. This shows that OGO or XGO, however these hydronium/GO interactions only allow
HGO not only has a denser hydronium layer, but its layer can keep some, but not all, of the GO hydroxyl groups to deprotonate. As such,
regenerating significant hydronium to neutralize the added hydroxides. it can be thought that HGO has a lot of reserve hydroxyl groups, that
OGO shows slightly higher pH resistance than XGO, and both show can deprotonate further should the equilibrium condition be shifted.
hydronium regeneration capabilities to a certain extent. Zeta potential This shift is introduced with base additions to the GO solutions, as the
tests are performed to study the net change in GO’s surface charge as existing hydronium is reacting with hydroxide to form water molecules.
the base is added (Fig. 2(d)), and HGO shows an unexpected increase The reduction in hydronium and the addition of new water molecules
(i.e. more positive) in net surface charge as the base is introduced, in GO’s proximity cause a spike in the rate of deprotonation from
before decreasing to the expected −40 mV surface charge [22]. HGO the remaining GO hydroxyl groups, which is the shift in its dynamic

5
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

Fig. 3. Simplified models illustrating (a) the interchange of H+ ions between GO and water, and (b) the regenerative hydronium ability of highly oxidized GO.

equilibrium state. This explains why HGO not only resists a decrease pores in the sample, which was also confirmed by BET (average cement
in surface charge once base is added, but counters it (i.e. approaches pore width with HGO = 18.2 nm, OGO = 22.6 nm, XGO = 24.5 nm, and
0) as recorded in the zeta potential tests. By comparison, OGO offers control = 25.9 nm). Thus, HGO-cement has the densest microstructure
little resistance evidenced by the drop in surface charge, and XGO falls at 24 h of hydration, followed by OGO-cement, XGO-cement and con-
immediately as it has the least hydronium generation capability. This trol, respectively. The 12-hour hydrated cement samples are imaged via
hydronium regeneration of HGO is only possible while hydroxyl groups SEM, to observe how the microstructure is developed mid-hydration.
remain protonated on GO (Fig. 3(b)), and after a certain amount of All GO-cements have finer and more developed C-S-H growth than
[OH]−1 is added, HGO fails to generate any more hydronium and its control, indicating cement dissociation and hydration has been ac-
zeta potential falls to approximately −40 mV, typical of base-washed celerated (Fig. 5(b)). Additionally, large Ca(OH)2 crystals can only
GOs [22]. It should be noted that zeta potential values can be off by be seen in control samples. It is concluded that GO have somewhat
±2 mV, due to the non-spherical shape and potential diffusion of ions neutralized the hydroxides, and minimized their crystal formations. The
through the hydronium layer on GO affecting accuracy of the tests [25]. crystalline composition of the cement samples are needed to confirm
This regenerative ability of GO (specifically, HGO) is expected to this inference, hence X-ray Diffraction (XRD) analysis is performed.
play the primary role in the ‘seeding’ ability of GO during cement
hydration. Calcium hydroxide (Ca(OH)2 ) is a known by-product of PC 3.4.2. XRD and heat of hydration analysis
hydration, and a known limiting factor in the mechanisms of cement PC has various crystalline compounds, with some having multiple
hydration [26]. By neutralizing the [OH]−1 produced during cement crystal phases, resulting in a highly complicated X-ray spectra from
hydration, HGO-cement should optimize cement use and provide a XRD tests. Rietveld Refinement is performed on the XRD data to deter-
stronger concrete structure, as discussed in the following section. mine cement’s crystalline composition, by a semi-quantitative statistical
peak fitting of XRD peaks. The XRD scans of all GO samples at various
3.4. GO-cement performance stages of hydration are showing in Fig. 4. The consumption of alite is
clearly evident in all samples from the diminishing of peaks at 32.5,
Unreacted water in the cement microstructure evaporates during its 39, 41, 42,52, 56.5, and 62.5 degrees, respectively. The vanishing of
lifetime, leaving cracks and voids that leave the structure susceptible the 34◦ also signifies the usage of C4 AF in cement. However, due to
the overlapping of peaks and the decrease in overall signal due to
to chloride penetration and shrinkage. Hence, a finer microstructure is
large peaks of calcite/limestone, Rietveld Refinement was performed to
desirable as it prevents water from evaporating out of the structure.
better ascertain the quantities of each compound in the sample, which
is discussed in the main article.
3.4.1. Surface area analysis via BET tests
C-S-H is difficult to measure in XRD due to its amorphous, triclinic
Brunauer, Emmett and Teller (BET) porosity tests indicate the av-
shape [27], but the cement hydration reaction uses alite (3CaO.SiO2 )
erage pore size for 24-hour hydrated control and GO-cements, during
and belite (2CaO.SiO2 ) in cement to produce C-S-H (3CaO.2SiO2 .3H2 O),
which the microstructure is fully developed, defining how strong the
as outlined in these (approximate) equations:
cement binding would be later in its life [26]. HGO-cement has > 40%
single point surface area than control, followed by OGO-cement and 6CaO.2SiO2 + 6H2 O → 3CaO.2SiO2 .3H2 O + 3Ca(OH)2 (1)
XGO-cement, respectively (Fig. 5(a)). Higher surface area leaves more
open space for cement grains to hydrate in future, and indicates smaller 4CaO.2SiO2 + 4H2 O → 3CaO.2SiO2 .3H2 O + Ca(OH)2 (2)

6
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

Fig. 4. XRD scans of all cement samples from beginning of hydration (time = 0, cement), to 28 days.

Alite dissociates much faster than belite, and is the primary con- control, thus hydration has increased (Fig. 5(d)). OGO and XGO’s heat
tributor to C-S-H during the first 28 days of hydration. Thus, Rietveld gain is also earlier than control, indicating the hydration reaction is
Refinement is used to find the amounts of alite and Ca(OH)2 (port- accelerated, not evident for HGO. However, HGO released significantly
landite) in control and GO-cements over 12 h, 24 h, and 28 days of more heat than OGO, XGO and control, respectively. Hence, it can be
hydration. At 12 h, HGO-cement and OGO-cement show greater alite concluded that the hydronium layer is contributing to the increase in
consumption than control, with XGO-cement the least (see Fig. 3(c)). cement hydration. It should be noted certain aluminates in PC also
However, at 24 h all GO-cements have alite at approximately 20%, far contribute to the heat release, however the cement used in these ex-
lower than control (at 34%). Over 28 days, XGO and OGO have used periments contains significantly lower aluminate amounts. Despite the
up even more alite, while HGO-cement’s alite has stalled. Moreover, increased and accelerated GO-cement hydration, they follow the same
the amount of Ca(OH)2 in 12 h samples was highest in HGO, followed general heat curve of control cement, and XRD Rietveld Refinement
by both OGO and XGO, respectively, and least in control. This indicates analysis did not show any unique crystalline peaks, hence the GOs are
all GO samples have accelerated hydration, as Ca(OH)2 production and acting as chemical catalysts for cement and not altering the cement
alite consumption is higher than control. Noteworthy, at 24 h Ca(OH)2 matrix crystalline structure. The decrease in heat generated after 12 h
is highest in HGO, but these crystals appear lowest out of all samples at occurs due to total consumption of small alite particles, leaving only
28 days. This could be the neutralization of hydroxides due to HGO’s large crystals that do not dissociate completely [31], which cannot be
hydronium layer. The crystallization of Ca(OH)2 takes energy away affected by GO-incorporation to the authors’ knowledge.
from C-S-H growth [28,29], therefore the hydroxide neutralization is
an important factor that can propagate C-S-H growth over a longer 3.4.3. Porosity analysis via MicroCT scans
time. Separately, higher CaCO3 amounts were recorded in all GO- MicroCT scans were performed on control and GO-cements to ob-
cement samples, a consequence of GO sheet destruction in alkaline serve macro- and micro-pore formations, as opposed to nano-porosity
environments, with the removed C atoms forming CO2 in the cement analysis via BET. Fig. 6(a) shows the isolated pore segments in each
samples (Supplementary Figure S5). The CO2 can then react with CaO of the 7 × 7 × 2.32 mm cropped 14-day cement samples. Both HGO-
in the cement to form calcite (CaCO3 ) [30]. This makes determining the cement and XGO-cement show noticeably less pore quantities with
exact alite consumption in GO-cements difficult, as some CaO is being respect to control, and smaller pores in general. In contrast, OGO-
diverted to calcite (not evident in control). Furthermore, SEM images cement possesses large pores similar to control, however the remaining
did not show the presence of Ca(OH)2 crystals in any GO-cements, pore sizes are quite small. A cumulative distribution of pore volumes
despite the high XRD count. These crystals may be significantly smaller show XGO-cement to have the highest number of pores, and the dis-
and harder to observe in the images, or their creation is localized in tribution is quite broad, signalling larger pores in the 0.005 mm3 to
some GO-absent zones on the samples, not observed in our imaging 0.0005 mm3 range (Fig. 6(b)). It is followed by control, HGO-cement,
process. To verify the hypothesis on increased GO-cement hydration, and OGO-cement, respectively. OGO-cement (2066) and HGO-cement
the heat released during the process was measured for the first 24 h. (2182) show significantly lower number of total pores compared to
Cement hydration is an exothermic reaction, and the change in released control (2634) and XGO-cement (2955), with similar slopes of the
heat between the cements can enable us to confirm the XRD results. distribution curves. A comparable trend for all samples is seen for the
All GO-cement samples release greater heat in the first 24 h relative to equivalent pore size distribution (Fig. 6(c)), where control shows the

7
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

Fig. 5. (a) The BET surface area and pore width (at 24 h), (b) SEM (at 12 h), (c) alite and Ca(OH)2 content, and (d) heat of hydration (over 24 h) of control, HGO, OGO and
XGO cement samples.

least smallest pore sizes, and XGO-cement shows the largest number highlights the importance of the intercalcium ions, which is the foun-
of pores close to 40 μm size. Moreover, OGO-cement shows the least dation for this research’s hypothetical model [33]. The model consists
total number of pores, followed by HGO-cement. However, total pore of negatively charged calcium silicate oxide complexes, weakly held
volume is high for OGO-cement (2.55 mm3 ), slightly larger than control together by divalent Ca2+ ions in the hydrate layers, which also consist
(2.50 mm3 ), and smaller for HGO-cement (1.78 mm3 ) and XGO-cement of OH− and water molecules. The weak layer allows C-S-H to grow in
(1.60 mm3 ), respectively (Fig. 6(d)). In summary, HGO-cement shows different directions, making the dense cement matrix binding materials
low total pore volume and less number of cumulative pores, further together. The hydrate layer is also the reason for C-S-H’s amorphous
verifying its improved effects on the cement. Conversely, OGO-cement (triclinic) crystal shape, making it difficult to detect via XRD tests [33].
possessed the least total number of pores, but had considerably higher The inner C-S-H layer is negatively charged due to the anionic Si-O-Ca-
overall pore volume, owing to the existence of a few large sized pores O-Si bridge, balanced by Ca2+ . However, this may cause C-S-H to be
evident in its segmentation 3D image. XGO-cement had highest number electrostatically attracted to the positively charged, acidic hydronium.
of pores, but lowest total pore volume, with larger number of 0.005 to A plausible interaction scenario is shown in Fig. 7(c). Once HGO comes
0.0005 mm3 pores. in contact with a C-S-H bridge unit, OH- ions from C-S-H’s hydrate
layer are neutralized by the hydronium (H3 O+ + OH− → 2H2 O), and
the resulting water molecules are used up by GO’s hydroxyl group to
3.5. Hypothetical model of early-age C-S-H/GO interactions regenerate the hydronium layer. Alternatively, the water can be used
up for further alite dissolution resulting in more C-S-H, as outlined
Based on the findings of the regenerative hydronium layer and GO- in Eq. (1). Meanwhile, the floating hydronium could disrupt the C-S-H
cement microstructure, molecular interactions between C-S-H and GO bridges by donating H+ to the silicates, and as there is no divalent ion
can be hypothesized. The growth of C-S-H on the GO sheet is simply (Ca2+ ) to hold the units together, they will drift away from one another.
highlighted in Fig. 7(a). To further understand the Ca2+ inter-bridging, The drifting C-S-H units allow the binding material to grow longer and
the C-S-H base unit is shown in Fig. 7(b). The Gartner et al.’s model [32] in all directions, away from the GO sheet until there is no hydronium
is applied here as it illustrates the hydrate layers in detail, which is im- in proximity. This explains the ‘hairy’ C-S-H growth on the HGO sheet,
portant for outlining the role of hydronium with cement. Moreover, this as shown in the SEM image in Graphical abstract. The inclusion of GO

8
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

Fig. 6. (a) 3D distribution of pores in all control and GO-cement samples. (b) Pore volume distribution for all CT scanned samples. (c) Equivalent pore diameter size for all CT
scanned samples , and (e) Total pore volume and difference in %age between all samples, respectively.

will also affect the 1–10 nm limits of nanoconfined water between the is also far higher than control and remaining GO samples, verifying
C-S-H layers [34]. Conversely, once the hydronium has been completely higher C-S-H production from alite and water’s reaction (Eq. (1))
depleted in the alkaline environment, any remaining Ca2+ ions can due to the neutralization of hydroxides by the hydronium layer. The
bridge the deprotonated GO hydroxyl groups with the anionic C-S-H decrease in heat evolved after 12 h is sometimes attributed to lack
bridge units. That is the more likely scenario for XGO, where there of available water, reducing C-S-H growth [31]. However, HGO’s heat
is no regenerating hydronium to allow C-S-H drifting and expanding. curve decreases at significantly slower rate than the others, and coupled
The Ca2+ bonds C-S-H to the GO sheet, and also limits C-S-H to grow with lowering Ca(OH)2 amounts after 24 h, confirms that hydronium
perpendicularly from the GO sheet (see Fig. 7(d)). However, due to neutralization is providing water for further cement hydration. A major
the limited direction of growth, C-S-H on GO may be dense but not as discrepancy is HGO-cement’s stalling of alite consumption after 24 h,
‘hairy’, as there is no hydronium to generate water molecules to feed which cannot be due to lack of available water as the hydronium
the C-S-H nucleation process. This is evident in the Graphical abstract, neutralization in HGO-cement continues up to 28 days at least. A more
where XGO sheets have thin and dense C-S-H coatings. However, it sensible reason would be complete consumption of small alite particles,
should be noted that the removal of hydroxyl groups in synthesis of which have hardened the cement matrix around HGO, thus inhibiting
XGO results in lower dispersive ability than HGO, which explains the greater dissolution. Ultimately, this indicates HGO’s effects on C-S-H
stacking of sheets in the image. Typically, C-S-H growth on GO should are quicker and shorter lived than other GO samples.
improve the flexural ability of the cement matrix, as the GO is now
acting as physical reinforcement. However, this is only true if (i) the 3.5.1. Further study of hypothetical model via FTIR analysis
GO is successfully embedded in the hardened cement matrix, and (ii) In research literature, molecular simulations of GO/C-S-H inter-
the GO sheets themselves are strong enough material to substantially actions show formation of ‘cages’, where the hydrate layer ions of
contribute to the flexural strength of cement. Molecular simulations C-S-H are locked between the GO sheet and C-S-H silicate units via
show exfoliation of functional GO groups during any pullout force Ca2+ inter-bridging [38]. One way to verify this is by FTIR analysis,
in the cement matrix, and the resistance to pull-out of GO groups is which can detect the Si-O bond lengths in cement samples. If there is
incumbent on the basal hydroxyl groups [35,36]. Kai et al. [36] further inter-bridging, the Si-O bond length will change, which will shift the
concluded this resistance may carryover to their ability in arresting absorption peak in FTIR spectra. Typically, the Si-O bond from C-S-
crack propagation in the cement matrix, as pure graphene nanosheets H is registered at 970 cm−1 in FTIR analysis [39]. The shift in Si-O
(i.e. with no functional groups) failed to inhibit crack openings or prop- absorption peak over hours of hydration are shown in Fig. 8(a) (and
agations in their simulations. Additionally, Zhang et al. [37] simulated in Supplemental Figure S7), and there is an obvious rapid increase in
the penetration of water between the C-S-H layers, and noted that the the wavelength during the first 24 h of hydration for all GO-cements.
carbonyl bond (Ca and COO− ) were the strongest and most stable, as A higher peak wavelength implies reduced bond length, possibly due
the hydroxyl groups lose structural H-bonds during their interactions to the Ca2+ bridging of C-S-H silicate with GO sheets. The increase
with the increased penetrated water. in wavelength is not uniform, however, possibly following the heat of
As the C-S-H and hydronium interactions have been explained, the hydration curve for HGO, OGO and XGO cements. Hence, based on
results from Figs. 5 and 6 can be further assessed. The micro-pore the FTIR it can be postulated that there is Ca2+ inter-bridging, and
analysis using CT scans show largest number of small pores by XGO- furthermore the hydronium layer on GO is stopping these GO/C-S-H
cement, with HGO-cement and OGO-cement having less overall number cage formations. It should be noted that while bridging GO and C-S-H
of pores. HGO-cement shows smallest total number of pores and less possibly increases GO’s ability to physically reinforce the cement matrix
overall total pore volume out of all the samples. The nano pore size (by better bonding), it also reduces C-S-H’s ability to grow and intercon-
recorded by BET tests are smallest for HGO, then OGO, XGO and nect the entire cement microstructure. Moreover, keeping in mind the
control respectively, possibly due to the increased C-S-H growth and molecular simulations in literature [35–37,40], the dense hydronium
drifting ability from the hydronium on GO. HGO’s heat of hydration layer from HGO may cause an initial decrease in bond formations due

9
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

Fig. 7. Schematic diagrams of (a) how C-S-H grows on GO, (b) Gartner’s C-S-H model unit, (c) how high-hydroxyl HGO regulates C-S-H growth and bridging, and (d) how
low-hydroxyl XGO inhibits C-S-H growth and bridging.

to its location between the C-S-H and GO hydroxyl groups, but reactions 30% stronger in compression, and 25% stronger in flexural strength,
with Ca(OH)2 would rapidly neutralize, resulting in increased cement the strongest overall, followed by OGO-concrete and XGO-concrete, re-
hydration and successful embedment of GO in the C-S-H layers. In FTIR, spectively. The flexural test shows less strength difference between the
HGO-cement shows a smaller Si-O wavelength increase than OGO and GO-concretes, which hints back to the physical reinforcement provided
XGO, respectively, due to its hydronium layer, however all GO-cements by GO as it is bonded in the cement matrix. XGO-concrete was only
shift to a peak wavelength higher than 960 cm−1 at 28 days. For HGO- slightly stronger than control in flexural strength, despite its smaller
cement, this is due to complete depletion of the hydronium layer after GO sheet size (compared to HGO) which is supposed to improve the
being exposed to the alkaline environment, where most of the early- cement microstructure [21]. A possible answer for this is that XGO
age hydration has already taken place by that period, thus the C-S-H is structurally defective, as indicated by Raman (Fig. 1(c)) and AFM
growth was not hindered. As a confirmation, SEM images of 24-hour (Fig. 2(a)) analysis. Thus, it cannot provide strength improvement
cement samples show smaller, more dispersed C-S-H in HGO-cement, to concrete. There is a known compromise of using GO in cement,
where C-S-H is connecting the cement matrix together (see Fig. 8(b)). which is the reduction in workability (or flow) of the cement mix,
attributed to accelerated cement hydration and the interbridging of
3.6. Effects of GO/hydronium-cement interactions on concrete GO sheets in the cement matrix [3]. Slump test was performed on the
fresh mixes of all GO-concretes, and HGO shows almost 50% higher
Finally, to show the strength contributions of all GO samples, 28- slump than control, a remarkable outcome (Fig. 8(e)). By contrast, XGO
day compressive and flexural strength tests were performed on GO- shows a 40% lower slump than control mix. This further reinforces
concretes (Figs. 8(c) and 8(d)). Compared to control, HGO-concrete is how hydronium initially prevents the C-S-H from coming into contact

10
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

Fig. 8. (a) The FTIR peak shifts of the C-S-H Si-O wavelength over hydration periods. (b) SEM images showing how HGO-cement’s C-S-H is better dispersed in the cement matrix.
(c) Bar graphs displaying the 28-day ultimate compressive strength, (d) 28-day ultimate flexural strength, and (e) slump test of fresh control, HGO, OGO, and XGO-concrete,
respectively.

with, and hence being locked on, to the GO hydroxyl groups on sheet a hydronium layer by donating protons in an aqueous solution. This
surface, via Ca2+ ions in C-S-H’s hydrate layer. Of note, the smaller hydronium layer is regenerating, as long as HGO has the ability to
pores in XGO-cement observed from microCT analysis (Fig. 6(a)) did donate protons. As such, the layer is shown to be the first point of
indicate a denser microstructure, but it does not translate into concrete contact with hydrating cement, neutralizing hydroxides in the C-S-
samples due to significant reduction in cement fluidity, resulting in H layer and allowing C-S-H molecules to drift away from the GO
sub-par strength of respective concrete. Conversely, the hydronium’s sheet, improving interconnections in the cement matrix, resulting in
interactions with C-S-H allow greater growth and drifting of particles a stronger, more durable cementitious structure with increased work-
that is attributed to the higher workability of HGO concrete (and OGO ability. It is important to include the hydronium layer in explaining
concrete) compared to control. This also indicates that highly oxidized GO-cement development, as the chemical interactions and workability
GO can be used to optimize cement, leading to stronger and denser differences of GO-cement and GO-concrete cannot be explained without
structures without compromising fluidity of the mix, a highly valued taking it into account. Hence, it is believed the primary contributor to a
property of fresh mixes in the concrete industry. It may also reduce the stronger and denser hydrated cement is the hydronium generation (and
need for additional superplasticizer consumption in concrete. regeneration) ability of GO, which is primarily dictated by its hydroxyl
groups. This research presents a new use of the acidic capabilities of
4. Conclusions GO, showing significant enhancements in hydration, strength, worka-
bility and densification of cement by an easy, practical and economical
By modifying the synthesis of GO, 3 types of GO with varying GO application technique (just add to water and mix with cement ).
hydroxyl groups were characterized to investigate GOs chemical in- However, there are several key interactions not covered in this
teractions with hydrating cement. High-hydroxyl GO (HGO) formed research. Specifically, the reaction of GO with the aluminates in cement

11
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

is not discussed, and late strength and durability of GO-cement also [6] K. Chintalapudi, R.M.R. Pannem, The effects of graphene oxide addition on
needs to be researched, taking into account the hydronium layer. Poly- hydration process, crystal shapes, and microstructural transformation of ordinary
portland cement, J. Build. Eng. 32 (May) (2020) 101551, http://dx.doi.org/10.
carboxylate superplasticizer was incorporated in cement and concrete
1016/j.jobe.2020.101551.
samples to increase workability at low w/c ratios, however the chem- [7] I. Chowdhury, N.D. Mansukhani, L.M. Guiney, M.C. Hersam, D. Bouchard,
ical interactions between the superplasticizer and GO have not been Aggregation and stability of reduced graphene oxide: Complex roles of divalent
investigated in great depth. The durability and corrosion resistance cations, pH, and natural organic matter, Environ. Sci. Technol. 49 (18) (2015)
should logically also be improved due to the denser C-S-H microstruc- 10886–10893, http://dx.doi.org/10.1021/acs.est.5b01866.
[8] A. Peyvandi, P. Soroushian, A.M. Balachandra, K. Sobolev, Enhancement of the
ture, although GO’s acidity may be detrimental to any steel in the
durability characteristics of concrete nanocomposite pipes with modified graphite
structure, thus it remains to be tested. Lastly, it is imperative to define nanoplatelets, Constr. Build. Mater. 47 (2013) 111–117, http://dx.doi.org/10.
how long does GO remain acidic in an aqueous solution, with respect 1016/j.conbuildmat.2013.05.002.
to its hydronium generating capabilities. [9] M. Wang, R. Wang, H. Yao, S. Farhan, S. Zheng, C. Du, Study on the
three dimensional mechanism of graphene oxide nanosheets modified ce-
CRediT authorship contribution statement ment, Constr. Build. Mater. 126 (2016) 730–739, http://dx.doi.org/10.1016/
j.conbuildmat.2016.09.092, URL http://www.sciencedirect.com/science/article/
pii/S0950061816315410?via%3Dihub.
Taimur Mazhar Sheikh: Conceptualization, Investigation, Writing [10] G. Jing, K. Xu, H. Feng, J. Wu, S. Wang, Q. Li, X. Cheng, Z. Ye, The non-
– original draft. Mohammed Parvez Anwar: Supervision, Resources. uniform spatial dispersion of graphene oxide: A step forward to understand the
Kasturi Muthoosamy: Methodology, Funding acquisition, Resources, inconsistent properties of cement composites, Constr. Build. Mater. 264 (2020)
Writing – review & editing. Jayaprakash Jaganathan: Supervision, 120729, http://dx.doi.org/10.1016/j.conbuildmat.2020.120729.
[11] W. Gao, The chemistry of graphene oxide, Graphene Oxide: Reduct. Recipes
Writing – review & editing. Andy Chan: Supervision, Project admin- Spectrosc. Appl. (2015) 61–95, http://dx.doi.org/10.1007/978-3-319-15500-5_3,
istration, Funding acquisition. Abdullahi Ali Mohamed: Writing – arXiv:0811.1860v1.
review & editing. [12] N.V. Medhekar, A. Ramasubramaniam, R.S. Ruoff, V.B. Shenoy, Hydrogen
bond networks in graphene oxide composite paper: Structure and mechanical
Declaration of competing interest properties, ACS Nano 4 (4) (2010) 2300–2306, http://dx.doi.org/10.1021/
nn901934u.
[13] F. Mouhat, F.X. Coudert, M.L. Bocquet, Structure and chemistry of graphene
The authors declare that they have no known competing finan- oxide in liquid water from first principles, Nature Commun. 11 (1) (2020) 1–9,
cial interests or personal relationships that could have appeared to http://dx.doi.org/10.1038/s41467-020-15381-y, arXiv:1911.04987.
influence the work reported in this paper. [14] A.M. Dimiev, L.B. Alemany, J.M. Tour, Graphene oxide. Origin of acidity, its
instability in water, and a new dynamic structural model, ACS Nano 7 (1) (2013)
576–588, http://dx.doi.org/10.1021/nn3047378.
Data availability
[15] X. Fan, W. Peng, Y. Li, X. Li, S. Wang, G. Zhang, F. Zhang, Deoxygenation of
exfoliated graphite oxide under alkaline conditions: A green route to graphene
Data will be made available on request. preparation, Adv. Mater. 20 (23) (2008) 4490–4493, http://dx.doi.org/10.
1002/adma.200801306, arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/
Acknowledgements adma.200801306, URL https://onlinelibrary.wiley.com/doi/abs/10.1002/adma.
200801306.
[16] J.P. Rourke, P.A. Pandey, J.J. Moore, M. Bates, I.A. Kinloch, R.J. Young, N.R.
The main author acknowledges the University of Nottingham
Wilson, The real graphene oxide revealed: Stripping the oxidative debris from
Malaysia for the PhD scholarship. Authors would also like to ac- the graphene-like sheets, Angew. Chem. 123 (14) (2011) 3231–3235, http:
knowledge the financial support from Ministry of Higher Education //dx.doi.org/10.1002/ange.201007520.
Malaysia via research grants FRGS/1/2019/STG05/UNIM/01/1 and [17] J. Chen, Y. Zhang, M. Zhang, B. Yao, Y. Li, L. Huang, C. Li, G. Shi, Water-
FRGS/1/2016/STG07/UNIM/02/1. enhanced oxidation of graphite to graphene oxide with controlled species of
oxygenated groups, Chem. Sci. 7 (3) (2016) 1874–1881, http://dx.doi.org/10.
1039/C5SC03828F, URL http://xlink.rsc.org/?DOI=C5SC03828F.
Appendix A. Supplementary data [18] D.C. Marcano, D.V. Kosynkin, J.M. Berlin, A. Sinitskii, Z.Z. Sun, A. Slesarev, L.B.
Alemany, W. Lu, J.M. Tour, Improved synthesis of graphene oxide, ACS Nano 4
Supplementary material related to this article can be found on- (8) (2010) 4806–4814, http://dx.doi.org/10.1021/nn1006368, arXiv:0410550.
line at https://doi.org/10.1016/j.conbuildmat.2022.129933. The sup- [19] D.C. Teychenné, R.E. Franklin, H.C. Erntroy, Design of normal concrete mixes,
Build. Res. Establishment Ltd 331 (1) (1997) 46, URL https://epdf.pub/design-
plementary data contains additional information pertaining to GO and
of-normal-concrete-mixes-br-331-ci-sfb.html.
GO-cement characterization tests. [20] M. Rogala, P. Dabrowski, P.J. Kowalczyk, I. Wlasny, W. Kozlowski, A. Busi-
akiewicz, I. Karaduman, L. Lipinska, J.M. Baranowski, Z. Klusek, The observer
References effect in graphene oxide - How the standard measurements affect the chemical
and electronic structure, Carbon 103 (2016) 235–241, http://dx.doi.org/10.
[1] J. Lehne, F. Preston, Making Concrete Change Innovation in Low-carbon 1016/j.carbon.2016.03.015.
Cement and Concrete The Royal Institute of International Affairs, Chatham [21] S. Sharma, N.C. Kothiyal, Influence of graphene oxide as dispersed phase in
House Report Series, 2018, p. 138, https://www.chathamhouse.org/sites/ cement mortar matrix in defining the crystal patterns of cement hydrates and
default/files/publications/research/2018-06-13-makingconcrete-c, URL https:// its effect on mechanical, microstructural and crystallization properties, RSC
www.chathamhouse.org. Adv. 5 (65) (2015) 52642–52657, http://dx.doi.org/10.1039/C5RA08078A, URL
[2] M. Rajczakowska, L. Nilsson, K. Habermehl-Cwirzen, H. Hedlund, A. Cwirzen, http://xlink.rsc.org/?DOI=C5RA08078A.
Does a high amount of unhydrated Portland cement ensure an effective auto- [22] T. Szabo, P. Maroni, I. Szilagyi, Size-dependent aggregation of graphene oxide,
genous self-healing of mortar? Materials 12 (20) (2019) http://dx.doi.org/10. Carbon 160 (2020) 145–155, http://dx.doi.org/10.1016/j.carbon.2020.01.022.
3390/ma12203298. [23] A.L. Higginbotham, D.V. Kosynkin, A. Sinitskii, Z. Sun, J.M. Tour, Lower-defect
[3] S. Chuah, Z. Pan, J.G. Sanjayan, C.M. Wang, W.H. Duan, Nano reinforced cement graphene oxide nanoribbons from multiwalled carbon nanotubes, ACS Nano 4
and concrete composites and new perspective from graphene oxide, Constr. Build. (4) (2010) 2059–2069, http://dx.doi.org/10.1021/nn100118m.
Mater. 73 (2014) 113–124, http://dx.doi.org/10.1016/j.conbuildmat.2014.09. [24] P.D. Pancharatna, G. Jhaa, M.M. Balakrishnarajan, Nature of interactions be-
040. tween epoxides in graphene oxide, J. Phys. Chem. C 124 (2) (2020) 1695–1703,
[4] S. Lv, S. Ting, J. Liu, Q. Zhou, Use of graphene oxide nanosheets to regulate the http://dx.doi.org/10.1021/acs.jpcc.9b10262.
microstructure of hardened cement paste to increase its strength and toughness, [25] G.V. Lowry, R.J. Hill, S. Harper, A.F. Rawle, C.O. Hendren, F. Klaessig, U.
CrystEngComm 16 (36) (2014) 8508, http://dx.doi.org/10.1039/C4CE00684D, Nobbmann, P. Sayre, J. Rumble, Guidance to improve the scientific value of zeta-
URL http://xlink.rsc.org/?DOI=C4CE00684D. potential measurements in nanoEHS, Environ. Sci.: Nano 3 (5) (2016) 953–965,
[5] L. Zhao, X. Guo, Y. Liu, Y. Zhao, Z. Chen, Y. Zhang, L. Guo, X. Shu, J. Liu, http://dx.doi.org/10.1039/c6en00136j.
Hydration kinetics, pore structure, 3D network calcium silicate hydrate, and [26] J.W. Bullard, H.M. Jennings, R.A. Livingston, A. Nonat, G.W. Scherer, J.S.
mechanical behavior of graphene oxide reinforced cement composites, Constr. Schweitzer, K.L. Scrivener, J.J. Thomas, Mechanisms of cement hydration, Cem.
Build. Mater. 190 (2018) 150–163, http://dx.doi.org/10.1016/j.conbuildmat. Concr. Res. 41 (12) (2011) 1208–1223, http://dx.doi.org/10.1016/j.cemconres.
2018.09.105. 2010.09.011.

12
T.M. Sheikh et al. Construction and Building Materials 364 (2023) 129933

[27] J.W. Bullard, G.W. Scherer, An ideal solid solution model for C – S – H, 4145, [35] M. Izadifar, J.S. Dolado, P. Thissen, A. Ayuela, Interactions between reduced
2016, pp. 4137–4145, http://dx.doi.org/10.1111/jace.14493. graphene oxide with monomers of (calcium) silicate hydrates: A first-principles
[28] I. Odler, H. Dörr, Early hydration of tricalcium silicate II. The induc- study, Nanomaterials 11 (9) (2021) http://dx.doi.org/10.3390/nano11092248,
tion period, Cem. Concr. Res. 9 (3) (1979) 277–284, http://dx.doi.org/10. URL https://www.mdpi.com/2079-4991/11/9/2248.
1016/0008-8846(79)90119-4, URL https://linkinghub.elsevier.com/retrieve/pii/ [36] M.F. Kai, L.W. Zhang, K.M. Liew, Graphene and graphene oxide in calcium
0008884679901194. silicate hydrates: Chemical reactions, mechanical behavior and interfacial sliding,
[29] P.W. Brown, C.L. Harner, E.J. Prosen, The effect of inorganic salts on tricalcium Carbon 146 (2019) 181–193, http://dx.doi.org/10.1016/j.carbon.2019.01.097.
silicate hydration, Cem. Concr. Res. 16 (1) (1986) 17–22, http://dx.doi.org/ [37] T. Yang, Y. Jia, D. Hou, H. Li, J. Jiang, Molecular dynamics study on the
10.1016/0008-8846(86)90063-3, URL https://www.sciencedirect.com/science/ weakening effect of moisture content on graphene oxide reinforced cement
article/pii/0008884686900633. composite, Chem. Phys. Lett. 708 (August) (2018) 177–182, http://dx.doi.org/
[30] S.A. Yaseen, G.A. Yiseen, Z. Li, Elucidation of calcite structure of calcium 10.1016/j.cplett.2018.08.023.
carbonate formation based on hydrated cement mixed with graphene oxide [38] D. Hou, Z. Lu, X. Li, H. Ma, Z. Li, Reactive molecular dynamics and experimental
and reduced graphene oxide, ACS Omega 4 (6) (2019) 10160–10170, http: study of graphene-cement composites: Structure, dynamics and reinforcement
//dx.doi.org/10.1021/acsomega.9b00042. mechanisms, Carbon 115 (2017) 188–208, http://dx.doi.org/10.1016/j.carbon.
[31] K.L. Scrivener, P. Juilland, P.J. Monteiro, Advances in understanding hydration 2017.01.013.
of Portland cement, Cem. Concr. Res. 78 (2015) http://dx.doi.org/10.1016/j. [39] R. Ylmén, U. Jäglid, B.-M. Steenari, I. Panas, Early hydration and setting of Port-
cemconres.2015.05.025. land cement monitored by IR, SEM and Vicat techniques, Cem. Concr. Res. 39
[32] E. Gartner, I. Maruyama, J. Chen, A new model for the C-S-H phase formed (5) (2009) 433–439, http://dx.doi.org/10.1016/j.cemconres.2009.01.017, URL
during the hydration of Portland cements, Cem. Concr. Res. 97 (2017) 95–106, http://www.sciencedirect.com/science/article/pii/S0008884609000465.
http://dx.doi.org/10.1016/j.cemconres.2017.03.001. [40] L. Lu, Y. Zhang, B. Yin, Structure evolution of the interface between graphene
[33] E. Duque-redondo, P.A. Bonnaud, H. Manzano, Cement and Concrete Research oxide-reinforced calcium silicate hydrate gel particles exposed to high tempera-
A comprehensive review of C-S-H empirical and computational models, their ture, Comput. Mater. Sci. 173 (November 2019) (2020) 109440, http://dx.doi.
applications, and practical aspects, Cem. Concr. Res. 156 (April) (2022) 106784, org/10.1016/j.commatsci.2019.109440.
http://dx.doi.org/10.1016/j.cemconres.2022.106784.
[34] P.A. Bonnaud, Q. Ji, K.J. Van Vliet, Effects of elevated temperature on the struc-
ture and properties of calcium–silicate–hydrate gels: the role of confined water,
Soft Matter 9 (2013) 6418–6429, http://dx.doi.org/10.1039/C3SM50975C.

13