Chloride penetration resistance of ­micron-nano materials modified mortar used as a concrete coating

1 Introduction

In the field of construction, the 21st century is considered to be the century of the ocean [1]. In recent decades, with the steady development of the marine economy, the construction scale of ocean engineering became larger, the fields expanded, and scores of seaport buildings, coastal power plants, embankments, marine bridges and tunnels, land terminal and treatment facilities for offshore oil and gas fields as well as installation of subsea wires, pipelines and equipment are still being built. According to the State Oceanic Administration, the building industry of marine engineering continued to maintain the momentum of rapid growth in 2015, reaching 504.4 billion RMB, approximately 7.8 % of the total ocean output value.

In the marine environment, reinforced concrete structures suffer from the superimposition of wind, fog, waves, currents, tidal range, ice, storm surge and organism corrosion, which cause corrosion damage and performance degradation and shorten the service life of materials and buildings [2], not only resulting in enormous waste of materials and energy, but also leading to a lot of sudden-onset disasters and triggering leakage of oil and gas, environmental pollution, as well as a great number of personal injuries. According to the survey of American corrosion loss, the direct cost to the United States in 2002 was about $ 276 billion, accounting for about 3.1 % of its annual GDP [3]. In the earthquake in Turkey in 1999, 15 000 people were killed and 50 000 houses collapsed, whereby the use of unwashed sands was the main cause of costly construction damage [4]. In terms of the corrosion loss accounting for 3 % of the national economic output [5], the corrosion loss was over 2 trillion RMB in 2015, a large proportion of which was for marine corrosion.

Professor Sitter’s “The Law of Fives” visually reveals the seriousness of the corrosion [6]. This means that if 1 dollar of steel protection costs was saved, 5 dollars would be spent at the beginning of the corrosion time, an additional 25 dollars for concrete cracking and 125 dollars for serious damage. This terrible amplification effect meant that governments and enterprises around the world should spend a lot more money for the research into concrete structure durability and safeguards. So the study of corrosion prevention materials and technology have a profound importance for extending the security service life of marine engineering structures as well as for guaranteeing the rapid development of the marine economy.

The major failure modes of reinforced concrete structures in a marine environment are the rebar corrosion induced by chloride that damages the concrete structures. A lot of research has shown that coating can form a barrier layer on the concrete surface, preventing the chloride ion from seeping into the concrete and improving its ability to counter chloride diffusion [7, 8]. So far, various surface coatings are used to prevent the chloride ions from penetrating into the internal part of the concrete [9-11]. Most of them consist of organic coatings using volatile organic compounds. However, such organic coatings could have the decisive shortcoming of being air-polluting during the manufacturing process as well as during the coating work. Moreover, in a natural environment, the aging-resistant performance of the polymer coating cannot meet the requirements of durability, especially under the conditions of dry-wet circulation and ultraviolet light, the cracking and peeling phenomena occur much more easily, and thus the protection effects are severely reduced and the coating requires continuous repair, which brings numerous practical difficulties to ocean engineering construction.

As the research moved along, more attention was paid to the application of superfine powders and nano-technology in cement-based materials. Especially in the last ten years, scientists have done quite a lot in terms of changing the microstructure and properties of hardened pastes by using superfine powder and nano-technology. Superfine particles like fly ash microspheres that can fill up the interstitial voids in cement, effectively increase the packing density and decrease the voids volume [12]. Silica fume can refine the pore diameter and make the matrix densification, it can also be used as pozzolanic materials to react with calcium hydroxide, and optimize the interface transition zone between cement and aggregates [13]. Nano-particles have the best potential as accelerators for cement hydration as a result of the nucleation effect and they make the microstructure and interface transition zone of hardened pastes much denser, resulting in lower permeability [14]. All in all, the micro-aggregate effect and the nucleation effect of micron-nano-sized particles promoting the formation of the hydration product help strengthen the permeability-resistance and further increase the durability of cement-based materials. In addition, the durability of cement pastes mixed with nano-silica with high pozzolanic activity was found to be much higher and this is attributed to the increase of high stiffness C-S-H gel in the pastes, which has been well documented [15-17]. Although a lot of research has been done on cement-based materials modified with micro-nano-sized materials, the lack of clear results for the effect of inorganic coating incorporation with superfine powders on the durability of concrete leads to the necessity to perform systematic research on performance evaluation of surface coating materials. It can be deduced that a compact surface mortar layer can not only significantly reduce the chloride ion diffusion coefficient, improving the concrete durability, but also compensate for the disadvantage of organic coatings at the same time.

This study is intended to assess the durability of concrete with denser surface mortar layer modified with micron-nano materials. The diffusivity of chloride has been evaluated using concrete specimens under steady state, and X-ray diffraction (XRD), differential scanning calorimetry (DSC), scanning electron microscope (SEM), and mercury intrusion porosimeter (MIP) are used to explain the mechanism of a denser surface mortar layer modified with micron-nano materials to improve the chloride–penetration resistance of concrete.

2 Experiment

2.1 Materials

Ordinary Portland cement (OPC) CEM I 42.5 has been used with a specific gravity of 3.13 and a surface area of 350 m²/kg. Fly ash (FA) conforming to GB1596–91 [18] was obtained from the Yangluo power plant. Silica fume (SF) with a specific gravity of 2.31 and a surface area of 21 m²/g according to the national standard of GB/T 27690-2011 [19]. Nano-SiO₂ (NS) with the average particle size of 20 nm is purchased from the Degussa Company. The chemical composition of cement, FA and SF are summarized in Table 1. XRD patterns of cement, FA, SF and NS are shown in Figure 1. The dispersing X-ray diffraction peaks of SF and NS show that they are amorphous in structure.

Fine aggregates are medium sand from the Xiang­jiang River in the Hunan Province of China with a fineness modulus of 2.5 and a maximum size of 5 mm, and the coarse aggregate is limestone rubble ranging in size from 5 to 25 mm that is obtained from the China Construction Third Engineering Division Co. Ltd.. They have been utilized in all mix proportions. Liquefied SDS (sodium dodecyl sulfate)-type polycarboxylate superplasticizer at a solid content of 10 % has been admixed to secure entrained air and reduce the specific volume of water.

2.2 Mix proportion

and sample preparation

In order to evaluate the resistance to penetration of chloride, a concrete specimen has been mixed with a water-binder ratio of 0.472, a fly ash-cement ratio of 0.286, a cement-fine aggregate ratio of 0.341 and a sand-ratio of 0.441. Its mixture proportion is given in Table 2.

Cement mortars have been mixed at a sand-binder ratio of 1.5 and a water-binder ratio of 0.35. Different dosages of FA (20 wt% of binders), SF (5 wt% of binders) or NS (from 1 wt% to 5 wt% of binders) were added to the mixes. Before the mix preparation, NS was added to deionized water, stirred and dispersed by ultrasonication at 325 W for 30 min to obtain the uniform suspensions. Their mixture proportions are given in ­Table 3. In order to prepare the samples for SEM and MIP, the small fragments obtained from the middle part of mortars cured to the test age were put into acetone solution for 3 days, and then were dried at 80 °C for 8 h.

All specimens for XRD and DSC were molded into a square with a size of 40 mm × 40 mm × 40 mm using cement pastes, which have the same ratio as mortars except without sand. After curing to the test age, the hydration of the pastes was stopped by submerging the center part of the crushed samples in acetone solution, the samples were oven-dried at 80 °C for 4 h and hand ground in the agate mortar so as to pass the 100 mesh size sieve.

Besides the control sample, other samples for the rapid chloride migration (RCM) test were all covered with a mortar coating. The specimen with coatings was produced as follows. Firstly, concretes and mortars mixed well, respectively, then the mortars were poured into the bottom of molds, and finally the concretes were placed above the mortars. The fabrication of specimens adopted molds of Φ100 mm × 100 mm. After molding, the specimens were put into a standard curing room, and 24 hours later the form stripping was carried out, then the specimens were immersed in the sink of the standard curing room. The specimens should be cut into pieces, with a diameter of 100±1 mm and a height of 50±2 mm 7 days before the rapid chloride migration (RCM) test, and then continued curing to the test age of 28 days. The control sample is pure concrete with the size of Φ100 mm × 50 mm. With the other samples, the concrete thickness is 45 mm, and the mortar coating is 5 mm.

2.3 Test procedure

2.3.1 RCM test

RCM-installation is shown in Figure 2. Before the specimens were installed in the tester, the practical diameter and height of these standard specimen was first measured. The cathode is 0.2 mol/L KOH with 5 wt% sodium chloride solution as dissolvent while the anode is 0.2 mol/L KOH solution without sodium chloride and the inside and outside liquids are at the same level. The RCM tester was powered with a direct current of 30 V at a constant temperature of 25 °C. The initial temperature and current of the anode solution should be measured, and the conduction time is determined by the initial current. At the end of experiment, the final temperature of the anode solution should be measured.

2.3.2 Chloride diffusion depth test

By the end of RCM test, the specimens were removed from the rubber buckets and all of them were split into two halves by the press machine. After the specimens were dried in the air, the 0.1 mol/L AgNO₃ solution was spray-painted on the split surface. About 15 minutes later, white silver chloride precipitation could be observed in areas with the chloride ion. Vernier caliper was used to measure the distance from the boundary of with and without white precipitation to the bottom of the specimen. For each specimen, the test was run ten times for different locations to eradicate any discrepancies and the mean was the chloride diffusion depth.

The chloride diffusion coefficient (D_RCM) of ­concrete was calculated using the following Equation (1) [20]:

     0.0239(273 + T)L                      (273+T)LX_d

D_RCM =                              (X_d – 0.0238√⇥

              U – 2 t                                     U – 2

Where T represents the average values of the initial and final temperatures of the anode solution, L represents the height of the specimen, X_d represents the chloride diffusion depth, t represents the electrolytic time, and U represents the absolute value of voltage used in the test.

2.3.3 XRD analysis

The mineral composition of the samples was confirmed using a Bruker D8 Advance XRD device with a Cu k_α X–ray source at 40 kV and 40 mA. During data collection the step-length was 0.02°, scanning rate was 2°/min and 2θ range was 5–70°.

2.3.4 DSC analysis

DSC analysis was conducted with a STA449c/3/G thermal analysis instrument at a heating rate of 15 °C/min from 20 °C to 800 °C in the atmosphere of nitrogen.

2.3.5 SEM analysis

Quanta 200 FEG-SEM from FEI Company in low vacuum mode was used to determine the morphology of the hardened cement pastes.

2.3.6 MIP analysis

Quantachrome Autoscan-60 mercury intrusion porosimetry (MIP) was used for testing the pore structures of the cement mortars. The measurable aperture ranged from 3 nm to 360 μm and in the measurement process the highest pressure was 300 MPa and the contact angle was set to 130°. After testing, Excel software was used to process the data.

3 Result and discussion

3.1 RCM analysis

The concrete with no protective coating was selected as the control sample and the chloride diffusion coefficient (D_RCM) of concretes with different mortar coatings are shown in Figure 3.

It is observed that D_RCM of the control sample was 15.12×10^-12m²/s, which was the highest in all samples. When the protective coatings were added, the D_RCM decreased significantly. For example, the D_RCM of sample C, FA20, SF5, NS1, NS3 and NS5 were 6.54×10^-12 m²/s, 5.05×10^-12 m²/s, 4.47×10^‑12 m²/s, 3.51×10^-12 m²/s, 2.81×10^-12 m²/s and 1.86×10^‑12 m²/s, respectively, which were 56.7 %, 66.6 %, 70.4 %, 76.8 %, 81.4 % and 87.7 % lower than that of the control sample.

According to the measured D_RCM, the relative resistance to chloride ion of concretes with different coatings was outlined below, from best to worst: sample NS (including NS1, NS3 and NS5), sample SF5, sample FA20 and sample C. For coating with 5 % NS weight fraction, the ability to counter chloride diffusion achieved its best result. It is obvious that, the chloride resistance of concrete with mortar coatings were better than that of the control sample, which is chiefly because the water binder ratio of the mortar surface coating was relatively small, and the compactness of the coatings was better than that of the concrete. As for the comparison of chloride resistance of concrete coated with FA, SF and NS modified mortar coatings, the mechanism will be discussed in the section below by XRD, DSC, SEM and MIP.

In addition, according the reference standard of Tang et al. [21], the resistance chloride ion performance of sample NS5 is categorized very good (D_RCM < 2×10^-12m²/s), sample NS3, NS1, SF5, FA20 and C as good (2×10^-12m²/s < D_RCM < 8×10^-12m²/s), and the control sample as mediocre (8×10^-12m²/s < D_RCM < 16×10^-12m²/s).

3.2 XRD analysis of the hydrated cement pastes

The influences of additives on the resistance chloride ion performance of cement-based materials can be reflected by the hydration products and microstructure changes. Figure 4 shows the XRD spectra of different cement pastes at 28 days.

As can be seen from Figure 4, the hydration products of all samples are basically the same, including ettringite (AFt), calcium hydroxide (CH), calcium carbonate (CaCO₃) and unhydrated tricalcium silicate (C₃S) and dicalcium silicate (C₂S). Simultaneously, it is also found that the diffraction peak intensity of CH significantly reduced with the addition of 20 % FA, 5 % SF, 1 % NS, 3 % NS and 5 % NS. In addition, the change tendencies of the diffraction peak intensity of C₃ S and C₂ S were similar to that of CH.

For samples with SF and NS, the decrease of CH, C₃ S and C₂ S diffraction peaks was related to pozzolanic activity and heterogeneous nucleus effect of SF and NS. The superfine particles of SF and NS can provide a lot of heterogeneous nucleus sites to promote the cement hydration in varying degrees. Moreover, the high pozzolanic activity of SF and NS accelerated the rate of secondary hydration, consuming a lot of CH and accordingly the content of C-S-H increased. All of these speeded up the clinker depletion, decreasing the content of C₃ S and C₂ S. The increased hydration products introduced a dense hardened paste, which would effectively obstruct the invasion of chloride ions and improve the chloride resistance of concrete, echoing the result of the chloride diffusion coefficient in 3.1. In addition, one can still see that the decline of the CH peak was especially evident with the addition of NS, and as the amount increased, the downtrend was more prominent. Although the CH peak could be decreased by mixed SF, the reduce degree was not as obvious as the samples with NS. This is due to the fact that the reaction activity of the smaller NS particles was higher than that of SF.

For the sample with FA, the contents of CH, C₃ S and C₂ S were lower than that of the control sample, but higher than that of sample SF5, NS1, NS3 and NS5, for reasons that the CaO content in FA is far lower than that in cement, and accordingly, using 20 % FA partial replacement of cement would decrease the CH and unhydrated clinker (C₃ S and C₂S) content in the mixture. In addition, plenty of network ions ([SiO₄]^4-, [AlO₄]^5-) in FA vitreous may lead to the formation of a structure with strong stability, therefore, the pozzolanic activity of FA at 28 days is relatively low [22], going against to the consumption of CH, which result in more CH content in sample FA than that in sample SF5, NS1, NS3 and NS5. Although the addition of FA would lead to a reduction in the amount of hydration products compared with that of sample C, the good micro–aggregate filling effect [23] of FA caused by its smaller particle size than that of cement and the smooth surfaces was able to reduce the pore space between the hardened pastes, resulting in the structure densification, which plays a positive role in hindering the chloride erosion. This is also consistent with the aforementioned results of the chloride diffusion coefficient.

3.3 Thermal analysis of the hydrated cement pastes

In general, quantitative analysis of C-S-H cannot be achieved by XRD because amorphous C-S-H is not able to be directly reflected by XRD patterns. In order to get the information of the main hydration product C-S-H intuitively, the DSC test was done and the results are shown in Figure 5. Because the addition amount has no effect on the acting mechanisms of NS, so only sample C, FA20, SF5 and NS5 were selected for testing.

It can be seen from this figure that the four samples have virtually identical hydration products, including C-S-H gels (about 100 °C), AFt (about 150 °C), CH (about 450 °C) and CaCO₃ formed by the carbonation of CH (about 700 °C).

According to DSC curves, the first endothermic valleys caused by dehydration of C-S-H gel of the four kinds of sample varied with the addition of admixture. The first valleys of the sample with SF and NS were deeper than that of sample C, mainly due to the high pozzolanic activity and nucleation effect of SF and NS promoting the generation of hydration products of C-S-H gel. In addition, it is obvious that more C-S-H gel generated by the addition of NS, indicating smaller NS particles with high surface activity is more likely to accelerate the cement hydration. However, the first endothermic valley of sample FA20 are lower than that of sample C, which is largely driven by that FA partial replacement of cement decreased the content of clinker that could react to form C-S-H gels and the pozzolanic activity of FA did not emerge at 28 days.

For the third endothermic valleys caused by dehydration of CH, that of sample C is the deepest, sample FA20 is next, sample SF5 is the third and sample NS5 is the shallowest, indicating that the addition of NS and SF decreased the content of CH, which is attributable to their high pozzolanic activity. Also, because the smaller NS particle has higher chemical activity, so the pozzolanic reaction of NS could consume more CH, resulting in the shallowest endothermic valley. In terms of sample FA20, high addition of FA greatly reduced the clinker content in binders, which caused the decreasing CH content compared with sample C. All of these are also certified by the result of XRD.

The decrease of CH and the increase of C-S-H caused by SF and NS would help enhance the compaction degree of the hardened cement paste, which could effectively obstruct the penetration of corrosive substances to some extent and improve the ability to resist chloride erosion. This result is consistent with the conclusion derived from the chloride diffusion coefficient test. However, although the hydration products of sample FA20 were fewer than that of sample C, it showed more resistance to chloride corrosion, which is probably because the smaller and smooth FA particles could fill the space between hardened pastes, resulting in higher density.

3.4 SEM analysis

In order to intuitively observe the microstructure of samples modified with different admixtures, separate SEM analyses were carried out on sample C, FA20, SF5 and NS5 at 28 days, as shown in Figure 6.

With comparison and analysis, it can be found out that there were many lamellar CH crystals and some unhydrated cement particles involved in the hardened cement pastes of sample C and lots of micron dimension spaces existed between the lamellar CH crystals, all of which would reduce the density degree of the cement matrix. When FA was added, some hexagonal lamellar CH crystals were present, and the amount of CH also reduced. Besides that, a round FA particle with a relatively clean and smooth surface could be remarkably observed, indicating that the FA particle did not show its pozzolanic activity perfectly at 28 days and it simply acted as a fine aggregate to fill the space between the hardened pastes. While for sample NS and SF, the hydration products appeared gel structure, and the distinct CH crystals could not be found. The gelatiniform hydration products were not independent and dispersed distribution, presenting integration structure, which made their structures more compact. In addition, a small amount of AFt could be detected in Figure 6 (a) and (b), but AFt was not found in Figure 6 (c) and (d), which is possibly because the AFt crystals in sample SF5 and NS5 were too small that covered by the mass of C-S-H gels..

SEM photographs show that the density degree of the hardened cement paste varied with the addition of different admixtures. Accordingly, the higher is the hardened paste density, the stronger is the ability to resist chloride erosion, which coincides with the data from the chloride diffusion coefficient test.

3.5 Pore structure analysis of hardened cement paste

The pore structure is one of the important characteristics of mortars, which has a great influence on the physical properties and permeability of mortars. Mercury intrusion porosimetry (MIP) is one commonly used method to characterize the pore structure. The influence of different additives on the pore structure of mortars was studied by mercury injection experiment and the parameters are shown in Table 4. In general, mortars with higher porosity have poorer permeability resistance. The data from Table 4 show that, compared with sample C, the porosities decreased by 13.7 %, 50.6 % and 69.8 % when the content of FA, SF and NS was 20 %, 5 % and 5 %, respectively. The average pore size is another important parameter of pore structure, and is used to characterize the overall condition of the pore structure. There is a strong correlation between the average pore diameter and the chloride ion permeability coefficient. Some research has shown that, the chloride ion permeability coefficient increased with the average pore size, which was closely related to the penetration property of mortars [24]. The incorporation of FA, SF and NS significantly reduced the average pore size of mortars, and thus improved the performance of their resistance to chloride ion penetration, among which NS5 had the best function of reducing the average pore size, and thus had the greatest resistance to chloride ion penetration, SF5 was the second one and FA20 was the third one. The performance of cement-based material is not only related to the parameters such as porosity and pore diameter, but also closely connected with the pore size distribution, which is also another important factor that affects the physical mechanical properties and durability of cement-based material. The integral curves of pore size distribution of hardened cement pastes are shown in Figure 7. Based on the hazardous property for the physical and mechanical performance, pores can be classified into four grades, including harmless pores (smaller than 20 nm), less harmful pores (between 20 and 100 nm), harmful pores (between 100 and 200 nm) and more harmful pores (bigger than 200 nm). However, in all types of pores that influence the chloride ion diffusivity of mortars, the size ranging from 5 to 200 nm has the most adverse effect [25], because in these pores, not only the phenomenon of capillary condensation can occur, enhancing the hygroscopicity of pores, but also larger capillary pressure and osmotic force are generated, increasing the autogenous shrinkage stress of mortars [26], and simultaneously accelerating the rate of surface and normal pressure permeation, which make the overall reduction of surface and normal pressure anti–permeability. In terms of Figure 7, the pore volume of sample C and FA increased quickly with the aperture ranging from 0 to 200 nm, specifically expressing the slope of this part of the curves increased sharply, which indicates that the apertures of the mortars were mainly distributed in that range. While for sample SF5 and NS5, the pore size was mainly distributed in 0~50 nm. Compared with sample C, when the sample was mixed with FA, the amount of capillary pores decreased, which shows that the addition of FA could reduce the porosity and improve the pore-size distribution of the mortars, increasing the ability of the mortars to hinder the chloride penetration, thus dramatically improving the resistance to chloride ion permeability of the matrix. When the sample is mixed with SF, the amount of the capillary pores further reduced, making the mortar more resistant to chloride ion permeability. The greatest reduction of capillary pores was with the sample mixed with NS, which had the greatest resistance to chloride ion permeability. All of these are consistent with the aforementioned results of RCM analysis, XRD analysis, thermal analysis and SEM analysis.

4 Conclusion

In this research, on the basis of the obtained experimental results, the main results are as follows:

(1) Chloride diffusion coefficients of concretes varied with different mortars coating from high to low the order was: concrete with no coating, concrete with pure mortars coating, concrete with FA modified mortars coating, concrete with SF modified mortars coating and concrete with NS modified mortars coating. And the ability to counter chloride diffusion increased along with the dosage of NS.

(2) The diffraction peak intensity of CH decreased along with the addition of 20 % FA, 5 % SF, 1 % NS, 3 % NS and 5 % NS.

(3) The sample with NS has the highest amount of C-S-H at 28 days, and its endothermic valley of CH is the shallowest. While the sample with FA has the lowest amount of C-S-H at 28 days.

(4) There were many lamellar CH crystals and some unhydrated cement particles involved in the hardened cement pastes of the sample without additives and lots of micron dimension spaces existed between the lamellar CH crystals. For the samples mixed with NS, the gelatiniform hydration products appeared integration structure, and the distinct CH crystals could not be found.

(5) Compared with pure cement mortar, the porosities decreased by 13.7 %, 50.6 % and 69.8 % when the content of FA, SF and NS was 20 %, 5 % and 5 %, respectively. The average pore size had similar changing trends. Also, the greatest reduction of capillary pores ranged from 5 to 200 nm was among sample mixed with NS.

(6) The micro-aggregate filling effect of FA reduced the chloride diffusion coefficient of the concrete with FA modified mortar coating. Three effects of SF and NS, including the heterogeneous nucleus effect, high pozzolanic activity and the micro-aggregate filling effect, together affect the resistance to the chloride ion permeability of the cement-based materials.

5 Acknowledgement

The authors would like to acknowledge the National Natural Science Foundation of China (No.51378408), and Science and Technology Support Program of Hubei Province (2015BAA084) and the 13th Five-Year Plan of National Key ­Research and Development (2016YFC0701003-05) for supporting this research and providing the materials tested.