Masahiro
Nagao
a,
Kazuyo
Kobayashi
a,
Yongcheng
Jin
b,
Ippei
Maruyama
ac and
Takashi
Hibino
*a
aGraduate School of Environmental Studies, Nagoya University, Nagoya 464-8601, Japan. E-mail: hibino@urban.env.nagoya-u.ac.jp
bSchool of Materials Science and Engineering, Ocean University of China, Qingdao 266100, China
cGraduate School of Engineering, The University of Tokyo, Tokyo 113-8656, Japan
First published on 27th July 2020
Few studies have reported the ionic conductive and photoactive properties of cementitious materials, owing to the compositional and structural variations of cement paste. The present report focuses on calcium silicate hydrate (C-S-H) and tetracalcium aluminoferrite (C4AF), prepared individually from pure constituent elements. C-S-H is an ionic conductor with a conductivity ranging from 10−4 to 10−3 S cm−1 in humidified air at temperatures between 60 and 100 °C. The hydroxide ions present in the mesopores of C-S-H as basic O–H groups were determined to be the major charge carriers, and the faradaic efficiency for the water oxidation reaction was estimated to be approximately 100%. C4AF is a mixed ionic n-type electronic conductor, particularly at high oxygen partial pressures. The conductivity of this compound was several times lower than that of C-S-H under the same humidity and temperature conditions, not only because of the bulk resistance but also the grain boundary resistance. C4AF and the hydrates exhibited photovoltaic and photocatalytic responses for water splitting and methylene blue degradation, respectively. These results indicate the important contribution of C-S-H and hydrated C4AF to the electrochemical and photochemical properties of cement paste.
It has been demonstrated that the ionic conduction in a hardened cement paste is driven by the humidification of a thin cement-based membrane.11 When the cement paste was moistened, the electrical resistance decreased by two orders of magnitude, the extent of which became more significant as the temperature increased. However, since there are many constituent cations and anions in cement paste, it cannot be determined which ions conduct charge carriers in the solid. Calcium silicate hydrate (C-S-H), the main hydration product of Portland cement, as well as sodium and potassium hydroxides (NaOH and KOH), are believed to be ion-conductive materials. However, calcium hydroxide (Ca(OH)2) is a minor ionic conductor, because of its low solubility product. If the ionic conductivity of C-S-H is comparable to those of alkali metal hydroxides, then the pH of the solution in the pores of the cement paste would be determined not only by the amounts of alkali metal hydroxides but also the amount of C-S-H, thus affecting the corrosion behavior of steel.
Iron-based oxides such as Fe3O4, CuFe2O4, and ZnFe2O4 have been studied as photocatalysts.12,13 Unlike the commonly known titanium dioxide (TiO2), these metal oxides show visible-light-induced catalytic activity, which enables the use of white LED lamps as the light source for various photoinduced reactions. Hirabayashi et al. reported the photocatalytic activity of brownmillerite-type Ca2Fe2O5, which is similar in structure and composition to tetracalcium aluminoferrite ((CaO)4·Al2O3·Fe2O3; C4AF), one of the main components of Portland cement.14 C4AF undergoes hydration to form ettringite (katoite) in the presence (absence) of gypsum (CaSO4).15,16 It is also reported that C4AF exists as Fe-siliceous hydrogarnet during hydration in the coexistence of silicate.17,18 It is meaningful to investigate the photochemical properties of these products; however, we consider it as future work, because the crystalline structure and solid solution state of the hydration products cannot be precisely controlled in the synthesis, thus making it difficult to obtain pure and reproducible material at this moment. (Even if a certain product is obtained, it may not be accepted as a satisfactory representative of various products of C4AF.) According to the literature, the degree of hydration of C4AF remains in the 0.6–0.8 range 1 year after the start of hydration.19 Moreover, as will be discussed later, a C4AF-rich cement clinker can be used to enhance the photocatalytic activity of the cement paste, which would further increase the amount of unreacted C4AF in the concrete. For these reasons, an easily synthesizable and highly pure C4AF was used as a research target in this study.
The goal of this study was to investigate the electro- and photo-chemical properties of cementitious materials. C-S-H and C4AF were prepared from pure constituent elements. The experimental conditions were adjusted by controlling the temperature and humidity of the sample during the measurements. In the electrochemical measurements, the alternating current (AC) conductivity was compared with the direct current (DC) conductivity to separate the electrical resistance components. Various galvanic cells were constructed to inspect the electrochemical behavior of the conductive ions in the solid. In the photochemical measurements, photovoltaic water splitting and photocatalytic degradation of methylene blue (MB) were performed under visible light irradiation. The relationships between the function and structure of C-S-H and C4AF were discussed.
![]() | ||
Fig. 1 (a) XRD patterns for C-S-H and C4AF. The PDFs of Ca1.5SiO3.5·xH2O and Ca2FeAlO5 are included for reference. (b) TEM images of C-S-H and C4AF. |
Fig. 1a also shows the XRD pattern for the C4AF sample, which is indexed to brownmillerite (Ca2FeAlO5; PDF 01-071-0667),28 with an average crystallite size of 446 Å. This pattern is almost identical to that refined previously by Rietveld analysis,11 where the octahedral and tetrahedral layers, occupied by Fe3+ and Al3+ ions, respectively, were alternately connected in the brownmillerite structure. It appears from the TEM image in Fig. 1b that the C4AF particle is a coarse crystal with microdomains, rather than an aggregate of crystallites. The C4AF particle size of approximately 2–4 μm is comparable to the values observed for Portland cement clinker.29
N2 adsorption/desorption isotherms of the C-S-H and C4AF samples were measured (see Fig. 2a). The C-S-H sample showed a type-IV sorption behavior, with an H3-type hysteresis loop, which is characteristic of slit-like pores, giving rise to capillary condensation of the adsorbates.30,31 The BET specific surface area and average pore diameter were calculated to be 125 m2 g−1 and 19 nm, respectively. The extent of N2 adsorption on the surface of the C4AF sample was significantly lower. Similarly, the adsorption volume obtained for this sample from the Kr sorption isotherm was smaller (see Fig. 2a), confirming that the surface area of C4AF available for gas adsorption was small. These results suggest that C-S-H is composed of aggregates of plate-like particles, while C4AF is a bulky, non-porous material. This agrees with the structural data obtained by XRD and TEM observations (Fig. 1).
The degrees of hydration of the C-S-H and C4AF samples were compared by thermal analysis (see Fig. 2b). The TGA curve for C-S-H indicated a weight loss of approximately 25% up to 600 °C, while the DTA curve exhibited two endothermic peaks at ca. 70 and 500 °C, attributable to dehydration and dehydroxylation, respectively.32,33 No weight loss was found above 600 °C, which is indicative of the absence of CaCO3 in the C-S-H sample. In contrast, the weight of the C4AF sample remained almost unchanged during heating from room temperature to 1000 °C, indicating that the as-synthesized sample was scarcely hydrated.
The results of FTIR analysis in Fig. 2c reveal the surface structures of the C-S-H and C4AF samples. The peak positions of all the bands observed for the C-S-H sample were in close agreement with the values reported by Lodeiro et al.34 The broad and strong peak at 3440 cm−1 is assigned to the stretching vibration mode of O–H groups. The peak at 1650 cm−1 is correlated to the bending vibration mode of O–H groups. These peaks are attributed to the O–H groups of both C-S-H and Ca-rich phase. The peaks of the carbonate group at 1458 and 1419 cm−1 correspond to the stretching vibrations of C–O bond groups. The appearance of these peaks is attributed to the adsorption of atmospheric CO2 on the C-S-H surface occurring during the measurements. The four peaks at 1000–450 cm−1, typical of silicates, are ascribed to the asymmetric stretching vibration of Si–O–Si bonds for the first three peaks and to their bending vibration for the last peak. Meanwhile, the C4AF sample showed only two peaks at 778 and 615 cm−1 corresponding to internal modes of AlO4 tetrahedra and FeO6 octahedra in the brownmillerite structure,35,36 which demonstrates the absence of any functional groups on the C4AF surface.
To gain further insight into the surface structure, TPD experiments were performed using CO2 as the probe molecule: the samples were loaded with CO2 at room temperature, followed by purging with Ar until no more CO2 desorption was detected. Afterwards, the temperature was increased linearly up to 900 °C. In Fig. 2d, two large distinctive peaks are observed for the C-S-H sample, while no peak is found for the C4AF sample. This difference is due not only to the different specific surface areas of the two samples, but also to their relative basicities toward CO2. (The pH of a suspension of 0.2 g of C-S-H in 200 mL deionized water was found to be 11.9, which was higher than that of a suspension of C4AF in deionized water measured under the same conditions.) The FTIR analysis results indicate that CO2 is easily adsorbed on the C-S-H surface, but this adsorption likely does not occur on the inert C4AF surface. It is also apparent from the TPD results that C-S-H has at least two types of O–H groups that vary in terms of basicity. This strongly suggests that the Ca-rich phase present in C-S-H has O–H groups, which function as a second adsorption site for CO2.
σ = L/(R × S) | (1) |
Electrochemical impedance spectroscopy (EIS) is another powerful technique for measuring the ohmic resistance component of an electrolyte. The impedance spectra for cells using C-S-H and C4AF membrane electrolytes obtained at 60 °C are presented in Fig. 4a. At high frequencies, only the end section of the semicircle was observed for C-S-H, while an almost complete semicircle was obtained for C4AF. The high-frequency semicircle is often related to the ohmic resistance and the interfacial capacitance, rather than to the charge transfer resistance and the double layer capacitance, which will be discussed later. The ohmic resistance of the electrolyte was determined through equivalent circuit models, consisting of a resistor in parallel with a constant phase element (CPE) for C-S-H and a series resistor and a resistor-CPE parallel combination for C4AF, both of which were accompanied by a Warburg impedance element. The ohmic resistance of C-S-H was computed to be 602 Ω cm−2, which is in agreement with the ohmic resistance of this membrane obtained by the current interruption method. Comparatively, the ohmic resistance of C4AF was resolved into bulk and interfacial contributions: a bulk resistance of 845 Ω cm−2 and an interfacial resistance of 1900 Ω cm−2. The sum of these values, 2745 Ω cm−2, is close to the DC ohmic resistance of this membrane. Thus, the semicircle observed at high frequencies is attributable to the interfacial resistance, presumably owing to the grain boundary contribution. It is also possible that this resistance is much smaller for C-S-H than for C4AF, because C-S-H possesses a far higher specific surface area and a far larger amount of O–H groups than C4AF. Here, it should be noted that the Ca-rich phase was scattered on the C-S-H surface, as seen from the TEM image (see Fig. 1b). Thus, the Ca-rich phase does not contribute significantly to the reduction of the grain boundary resistance, due to a large gap between the phases. The ohmic resistances of these membranes above 60 °C were measured and estimated by similar procedures, and the AC conductivities were calculated viaeqn (1). The results are presented in Fig. 4b, all of which are in good agreement with the DC conductivities obtained under the same conditions. Accordingly, the higher sum of the bulk and grain boundary conductivity of C-S-H as compared to C4AF can be explained in terms of its lower grain boundary resistance. The ionic or electronic current generally passes through the grains and across the grain boundaries in polycrystalline materials (bulk conduction mechanism).37 However, recent studies suggest that ion carriers, especially protons, can migrate along the grain boundaries in nanomaterials, whose resistance is affected sensitively by the amount of adsorbed water, grain size, and porosity (interfacial conduction mechanism).38–40 Thus, these two paths for ion transport need to be considered on the basis of the conduction mechanism. In the impedance spectrum of C4AF, the grain boundary resistance was higher than the bulk resistance, and these components appeared individually. On the other hand, the impedance spectrum of C-S-H showed a single semicircle with a smaller diameter. These results, along with the characterization of the two materials, suggest that carrier transport occurs by the bulk conduction mechanism in C4AF and the interfacial conduction mechanism in C-S-H.
![]() | ||
Fig. 4 (a) Impedance spectra of C-S-H and C4AF at 60 °C. (b) Temperature dependence of AC conductivity for C-S-H and C4AF. |
Fig. 4b also includes the AC conductivity of the hydrated cement paste recorded in our previous study.11 A somewhat higher conductivity was obtained for the cement paste, with a slightly lower activation energy (0.57 eV) than the values for C-S-H. This difference occurs because cement paste contains alkali metal ions such as Na+ and K+, which bond to the silanol groups (Si–OH) present on the C-S-H surface.41 The effect of K+ ions on electrical conductivity will be further investigated in the next section.
H2O + 1/2O2 + 2e− → 2OH− | (2) |
In the oxygen gas concentration cell, the H2 electrode showed opposite polarity potentials at all the tested temperatures, which can be explained through the following reaction in the hydroxide ion conductor:45,46
H2 + 2OH− → 2H2O + 2e− | (3) |
A decrease in the EMF value with decreasing membrane thickness was observed in the oxygen gas concentration cell (Fig. 5b), which was due to gas leakage and the resultant fuel loss. Thus, the observed EMF values were lower than the theoretical values calculated by the Nernst equation, depending on the membrane state. To obtain more direct evidence of hydroxide ion conduction, water vapor electrolysis was performed using the C-S-H membrane electrolyte at 100 °C. The galvanostatic electrolysis characteristics are presented in Fig. 5c, along with the voltage–time curve at each current density. H2 was confirmed to be produced at the cathode, indicating that the H2 evolution reaction proceeded as follows:
2H2O + 2e− → H2 + 2OH− | (4) |
It was also confirmed that the amount of H2 formed roughly agreed with the theoretical values calculated by Faraday's law, strongly supporting hydroxide ion conduction in C-S-H. Nevertheless, degradation of H2 production was observed at a current density of 60 mA cm−2 in Fig. 5c. This was due to the application of high voltage to the cell (see inset in Fig. 5c), which may have caused decomposition of the electrolyte or corrosion of the electrode. Note that unlike the C-S-H formed in cement pastes, the C-S-H examined in this study does not contain any alkaline metal ions, thus ruling out these ions as charge carriers. However, the contribution of Ca2+ ions released from the Ca-rich phase to ionic conduction cannot be neglected.
From a practical standpoint, it is also important to determine the ionic conduction in C-S-H under alkaline conditions, because hydrated cement pastes contain alkaline metal ions.47,48 The impedance spectra of the C-S-H membrane were measured by adding KOH solutions with various pH values to the cell at 100 °C. The results are presented in Fig. 5d. All impedance spectra consisted of three components in series: ohmic resistance (intercept of the impedance line and real axis), grain boundary resistance (high-frequency part of arc), and electrode reaction resistance (low-frequency part of inclined line). Both the ohmic and grain boundary resistances decreased with increasing pH. These behaviors differ from those observed in the impedance spectrum for the alkaline-metal-free C-S-H membrane (see Fig. 4a). The difference is further emphasized by the much lower activation energy for ionic conduction (ca. 0.1 eV) estimated for the C-S-H membrane operated at a pH of 12.8, as compared to that operated in humidified air (see inset of Fig. 5d). This was due to the replacement of adsorbed water with KOH solution for the ion channels, which changes the interfacial conduction mechanism to vehicular diffusion or the Grotthuss mechanism for ionic conduction.37
These results demonstrated hydroxide ion conduction in C-S-H, which is useful for interpreting the electrochemical behavior of hydrated cement pastes. Given its relatively high ion conductivity, C-S-H can function as an electrolyte for a variety of galvanic cells formed at the cement–steel interface in reinforced concrete structures, resulting in corrosion of the steel due to self-discharge when a local cell is established. In this case, the corrosion rate is directly affected by the humidity and basicity at the cement–steel interface. Importantly, such a local cell is formed even in the absence of alkali metal and alkaline earth metal hydroxides in the cement paste. Also important is the mesoporous structure formed in C-S-H, which enhances the water adsorption capacity of this material through capillary condensation. Thus, the hydroxide ion conductivity of C-S-H is maintained high even under low relative humidity conditions, which is one of the causes of increased corrosion under such conditions.
The electrode characteristics of C4AF as a photoanode were first validated by experimental demonstration of photovoltaic water splitting in a two-electrode cell consisting of a C4AF membrane anode, a Pt cathode, and a NaOH solution electrolyte. The CV profiles for this cell obtained under light and in the dark are presented in Fig. 6b, along with the corresponding impedance spectra recorded at a bias voltage of +1 V. Irradiation-induced enhancement effects were observed in both the cyclic voltammograms and impedance spectra, which are attributable to the decrease in the charge-transfer resistance due to irradiation. (In the impedance spectra, both the ohmic and grain boundary resistances were almost unchanged, regardless of the irradiation.) Next, the photocatalytic activity of C4AF was determined by degrading MB in deionized water under irradiation. The change in the absorbance of MB with irradiation time is shown in Fig. 6c. The peaks in the 600–700 nm range, which correspond to the conjugated π-system, and the peak at 291 nm, assignable to the aromatic ring, decreased with time. Also, the color of the MB solution changed to light purple after irradiation for 120 h. These results indicate that C4AF shows photocatalytic activity for the degradation of MB. From a plot of the natural logarithm of the relative absorbance A0/A against irradiation time (see inset of Fig. 6c), the first-order rate constant was calculated to be 3 × 10−4 min−1, which is comparable to that of α-Fe2O3 (3 × 10−4 min−1) and about half that of brownmillerite Ca2Fe2O5 (7 × 10−4 min−1).14 Note that the specific surface area of C4AF was 0.8 m2 g−1, which is lower than those of Fe2O3 (5.9 m2 g−1) and Ca2Fe2O5 (1.8 m2 g−1). We cannot further compare the photocatalytic activities of different materials, since only the rate constant and BET specific surface area have been reported in the literature. Thus, an attempt was made to increase the specific surface area of C4AF by ball milling. The particle size of C4AF decreased with grinding time, as characterized by the decrease in the XRD peak intensity and in the average crystallite size (Fig. 6d). After a 24 h grinding run, the specific surface area of C4AF increased to 2.7 m2 g−1. (No additional effect on the specific surface area was observed past 24 h.) Fig. 6e compares the rate constants for MB degradation in the case of unground and ground C4AF. The rate constant estimated for ground C4AF was 5 × 10−4 min−1, which is higher than that of Fe2O3, but lower than that of Ca2Fe2O5, despite the larger surface area of C4AF. One possible reason for the difference in photoactivity between C4AF and Ca2Fe2O5 is the occupation of the tetrahedral sites of the brownmillerite structure by Al3+ ions, which may reduce the changeable valence state of Fe3+ ions in the octahedral sites.51
The occurrence of the photocatalytic reaction on the C4AF surface is not negligible in cement clinkers, but extremely rare in cement pastes, because portions of C4AF react with gypsum (CaSO4) to form ettringite [Ca6(Al,Fe) (SO4)3·26H2O].52 However, if C4AF photocatalysis is actively applied to the removal of soil and pollutants from concrete surfaces, then it is required to increase the C4AF content of the cement clinker so as to form katoite [Ca3(Al,Fe)2(OH)12],52 which has a higher potential as a photocatalyst than ettringite, because the present XRD results showed the coexistence of the brownmillerite and hydrogarnet (katoite) phases in the photoactive sample during MB photodegradation measurements (data not shown). If realized, this may eliminate the need to coat concrete surfaces with a photocatalyst such as TiO2 for self-cleaning purposes.
This journal is © The Royal Society of Chemistry 2020 |