1.Introduction

With the increasing use of concrete under early-age marine exposure conditions such as the construction of slip-formed caissons or cast-in-situ quay walls, the importance of studying chloride ingress under early-age exposure is becoming more evident. Standardized test methods are available for the determination of chloride diffusion coefficients of concrete under natural diffusion conditions, including ASTM C1556, Standard Test Method for Determining the Apparent Chloride Diffusion Coefficient of Cementitious Mixtures by Bulk Diffusion (www.astm.org), and NT BUILD 443, Concrete, Hardened: Accelerated Chloride Penetration (www.nordicinnovation.org). These methods involve immersion of specimens in chloride solutions for the intended time interval and subsequent powder grinding at various depths to determine chloride concentration profiles. Fitting the error function solution of Fick's second law of diffusion to the experimental data points yields the apparent diffusion coefficient for the time interval considered. It should be noted that such tests are laborious, expensive, and time-consuming.

Accelerated tests involving the application of electric potential for speeding up chloride ingress into concrete have also been developed and standardized, including ASTM C1202, Standard Test Method for Electrical Indication of Concrete's Ability to Resist Chloride Ion Penetration (www.astm.org), NT BUILD 492, Concrete, Mortar and Cement-based Repair Materials: Chloride Migration Coefficient from Non-Steady-State Migration Experiments (www.nordicinnovation.org), and AASHTO T357, Standard Method of Test for Predicting Chloride Penetration of Hydraulic Cement Concrete by the Rapid Migration Procedure (www.transportation.org). Among these methods, the Rapid Chloride Migration (RCM) test offers a rapid, cost-effective, and relatively straightforward procedure. After subjecting the specimens to chloride penetration under an applied electric potential, the specimens are split and the penetration depth is revealed by a colorimetric procedure. By using test parameters such as the applied potential, test duration, chloride concentrations at the exposed surface, and the color change front, the chloride diffusion coefficient can be derived. Despite being easier to perform, rapid migration tests involve an application of electric potential and differ fundamentally from natural diffusion mechanisms [1, 2]. Furthermore, comprehensive reviews and phenomenological models on service life prediction and the influence of mineral additives on chloride diffusion [3–5] primarily focus on hardened concrete under standard exposure conditions, rather than early-age exposure scenarios. Consequently, these tests give instantaneous values of the diffusion coefficient and cannot be used reliably for the prediction of the long-term performance of concrete exposed to chlorides from an early age.

Extensive reviews have recently summarized the key factors influencing chloride diffusion in hardened concrete and the relevant service-life prediction models. Xu et al. [6] examined the effects of water-binder (w/b) ratio, additives, cracks, and environmental conditions on chloride ingress, noting that silica fume can reduce the diffusion coefficient by up to 50 % at a 5–10 % cement replacement. Wang et al. [4] reviewed chloride penetration in recycled aggregate concrete and showed that silica fume improves chloride resistance by 20–65 % depending on dosage. Khan and Alhasan [5] compared mechanistic, stochastic, and empirical service-life models for chloride-induced corrosion and highlighted the diffusion coefficient as the most critical input parameter. Although these studies demonstrate the well-established influence of silica fume and the importance of diffusion-based service-life modelling, they do not address the scenario where concrete is exposed to chlorides immediately after casting. Consequently, the impact of very early-age exposure – when hydration is far from complete, and the pore structure is rapidly evolving – on long-term chloride transport remains poorly understood. This gap is especially relevant for marine structures built by continuous slip-forming or cast-in-situ methods, where fresh concrete can come into contact with seawater within hours of placement.

Previous work on the effects of early-age exposure of concrete to chlorides has been rather limited. Caballero et al. [7], using the RCM test, studied early-age diffusion coefficients of mortar mixes based on ordinary Portland cement and found that the results for the age of 1 day were about 10 times higher than the 28-day values. They also used the natural diffusion test for determining the apparent diffusion coefficient for exposure intervals of 1.5 to 4 days and 7 to 16 days, and found that the value for the earlier time interval was twice that of the latter. Fraj et al. [8], using Portland cement-based mortars (including slag-blended mixes), found that the apparent diffusion coefficient for the exposure interval from the age of 1 day to 28 days was four times the value for the exposure interval from the age of 28 to 56 days. Although the results of both studies show a substantial effect for early-age exposure in increasing the diffusion coefficients, the exposure durations considered were not long enough to show the effect of early-age exposure on longer-term diffusion coefficients.

Bagheri et al. [9, 10] used the RCM test method to determine diffusion coefficients at the ages of 1 day and 180 days and found that the early-age value was about 10 times higher than the 180-day value. These researchers also used the natural diffusion tests for determining the apparent diffusion coefficient of concrete mixes and found that the value for concrete exposed from the age of 1 day for a duration of 180 days was only 17 % higher than the one exposed from the age of 28 days for a similar duration. These findings are significant and show that although diffusion coefficients for early-age time intervals are many times higher than those of the later-age exposures, at longer durations of exposure, the difference between the early and late-age exposure diminishes significantly. It has therefore become apparent that the appropriate way for studying the effect of early-age exposure on chloride diffusion performance of concrete is by using natural diffusion tests with long-term durations. However, powder grinding and chloride concentration profile determinations involved in this test make it laborious and expensive, and a simpler test would be advantageous in this regard.

Colorimetric methods under natural diffusion conditions appear a viable alternative in this regard. These tests have been used by various researchers for studying concrete performance under chloride environments [11–15]. The test is performed without the application of an electrical potential and therefore provides data under natural diffusion conditions. The test can also be conducted at any required time interval in the same manner as the immersion tests. Although some researchers have used the penetration depth in the colorimetric method as a means for comparing the relative performance of various mixes [16–20], it has also been used for the determination of chloride diffusion coefficients [12, 21–24]. In this test, the penetration front is determined by a colorimetric method, and by knowing the surface chloride concentration value, the diffusion coefficient for the exposure interval can be calculated. The other advantage of the colorimetric method is that it provides data directly on the free chloride diffusion coefficient, which is more relevant than the total chloride profiles determined in the usual immersion tests.

The research object of this study is ordinary and silica-fume-blended Portland cement concrete subjected to chloride solutions immediately after casting. The aim is to evaluate the effect of exposure from very early ages on chloride diffusion characteristics using a simple, natural-diffusion colorimetric method, and to determine whether silica fume can mitigate the detrimental impact of such early exposure. The specific objectives are:

to measure free-chloride penetration depths under natural diffusion conditions for exposure intervals starting from the time of casting up to several days and weeks, and to compare these with the standard 28-day exposure regime;

to derive the corresponding apparent chloride diffusion coefficients from the colorimetric data;

to quantify the influence of partial cement replacement by silica fume on the early-age diffusion behavior.

This approach provides a practical alternative to the laborious immersion/powder-grinding tests while yielding data that can be directly incorporated into service-life predictions.

2.Methods

The colorimetric method under natural diffusion conditions was used to study the chloride diffusion into the concrete mixes with water-cement (w/c) ratios of 0.5, 0.4, and 0.3 at various exposure ages and for different exposure durations. The exposure intervals considered included, exposure from the time of casting to the ages of 1,3,7,28 and 180 days. For exposure age of 1 day, exposure intervals of 1–3, 1–7, and 1-80 days, and for exposure age of 3 days, exposure intervals of 3–7 and 3–180 days were considered. For the exposure age of 7 days, intervals of 7–28 and 7–180 days, and for the exposure age of 28 days, exposure interval of 28–180 days were studied. The colorimetric test was also used to study the effect of partial cement replacement by 10 % silica fume under various exposure ages and exposure durations. In order to obtain further insight into early age exposure performance of the mixes studied, additional tests including electrical resistivity, development of compressive strength, and the RCM tests at various ages were also carried out.

2.1.Materials Used

A type 2 Portland cement complying with the requirements of ASTM C150 was used for the study. The chemical composition of the cement is given in Table 1. The physical properties of the cement are presented in Table 2. Silica fume in powder form complying with the requirements of ASTM C1240 was obtained from the Azna ferrosilicon plant (Iran). The chemical and physical characteristics of the silica fume utilized, are presented in Tables 1 and 2, respectively. A polycarboxylic ether type superplasticizer was used to keep the workability of all mixes studied within the slump range of 100–150 mm. Silver nitrate of chemical grade produced by Merck Company (Germany) was used for determination of chloride penetration front in the colorimetric tests. The aggregates used for the production of concrete mixes included natural river sand with a saturated surface dry density of 2.50 g/cm3 and absorption of 4.7 %, and a crushed coarse aggregate with a maximum nominal size of 19 mm, saturated surface dry density of 2.57 g/cm3 and absorption of 2.1 %. Water of drinkable quality was used for the production of the mixes.

Table 1. Chemical analysis of the cement and silica fume.

K2O

Na2O

SO3

MgO

CaO

Fe2O3

Al2O3

SiO2

Composition %

0.64

0.37

1.86

1.82

62.87

3.94

5.38

21.15

Cement

1.32

0.42

–

0.87

0.49

0.70

1.10

94.30

Silica fume

Table 2. Physical properties of the cement and silica fume.

Cement

Silica fume

Average diameter (μm)

–

0.15

Specific surface area (cm2/g) (blain method)

3170

–

Specific surface area (cm2/g) (BET method)

–

192000

Specific gravity (g/cm3)

3.16

2.21

L.O.I (%)

–

0.1

Percent retained on 45 mm sieve

–

0.3

Accelerated pozzolanic strength activity index (%)

–

145

Setting time (min) (initial)

105

–

Setting time (min) (final)

150

–

Compressive standard mortar at 28 days (MPa)

41.3

–

2.2.Mixes Considered

Three concrete mixes with 420 kg/m3 of a type two Portland cement at w/c ratios of 0.5, 0.4, and 0.3 were used in the study. In order to study the effect of incorporation of silica fume on chloride diffusion under early age exposure, a further three mixes, using 10 % silica fume as partial replacement of cement, at the above-mentioned w/c ratios were also considered in the study. Required amounts of the superplasticizer were used to keep the workability of all mixes within the 100–150 mm slump range. The mix ingredients are presented in Table 3.

Table 3. Mixture proportions of concrete mixtures.

Mixes

Cementitious materials (kg/m3)

w/b ratio

Superplasticizer to Cementitious materials (%)

Coarse aggregate (kg/m3)

Fine aggregate (kg/m3)

Cement (kg/m3)

Silica fume (kg/m3)

Temperature (°C)

0.5-Ref

420

0.5

–

813

813

420

–

20.0

0.5-SF10

420

0.5

0.21

806

806

378

42

19.9

0.4-Ref

420

0.4

0.47

867

867

420

–

20.1

0.4-SF10

420

0.4

0.54

860

860

378

42

20.3

0.3-Ref

420

0.3

1.88

920

920

420

–

19.1

0.3-SF10

420

0.3

2.18

913

913

378

42

20.1

2.3.Tests Carried Out

2.3.1. Electrical resistivity

The electrical resistance of various mixes was determined at various ages according to the method described by Tang [25] using an alternating current to avoid errors due to concrete capacitance.

2.3.2. Compressive strength

The development of compressive strength of the mixes studied with time was determined from the age of 1 day up to 1 year, using 10 cm cubical specimens according to the method described in BS EN 12390 part 3.

2.3.3. Rapid Chloride Migration test

This test was carried out in accordance with the NT BUILD 492 standard. Cylindrical specimens (100 mm in diameter and 50 mm in thickness) were used. One surface was exposed to a 10 % NaCl solution, while the opposite surface was in contact with a 0.3 N NaOH solution. Chloride ingress was accelerated by applying a direct electrical potential. The voltage magnitude was selected based on the initial current measured through the specimen. The test setup is shown in Fig. 1.

Figure 1. The Rapid Chloride Migration test a) the test set up; b) chloride penetration depth of split specimens after the test.

The test duration was predetermined according to the applied potential. After completion, specimens were split axially. The freshly exposed surfaces were sprayed with a 0.1 N AgNO3 solution. Silver nitrate reacts with chloride ions to form a white AgCl precipitate. This clearly marks the chloride penetration front. The non-steady-state migration coefficient was then calculated using Equation (1).

                              (1)

where  is the non-steady-state migration coefficient 10–12 m2/s,  is the applied voltage (V),  is the average temperature of the anolyte solution (°C),  is the specimen thickness (mm),  is the average chloride penetration depth (mm), and  is the test duration (hour). Equation (1) is adopted directly from the NT BUILD 492 protocol. It is derived from the analytical solution of the simplified Nernst–Planck equation. The numerical coefficients in the equation are dimensional. They combine fundamental physical constants with unit conversion factors. This formulation allows direct input of practical laboratory units while ensuring the final result is correctly scaled to 10–12 m2/s.

2.3.4. Natural diffusion test and colorimetric determination of chloride penetration front

Natural diffusion tests, commonly performed following principles outlined in documents such as ASTM C1556 and NT BUILD 443, are widely used to determine the apparent chloride diffusion coefficient of concrete. In these tests, concrete specimens are fully immersed in a chloride solution of known concentration (typically 16 % NaCl by mass, equivalent to ~2.8 mol/L) for a predetermined exposure period ranging from several weeks to several months. After exposure, powder samples are collected at incremental depths by dry grinding. The chloride content of each powder sample is determined by chemical analysis (e.g., potentiometric titration). The resulting concentration profile is then fitted to the error-function solution of Fick's second law of diffusion to calculate the apparent diffusion coefficient. However, the sequential steps of powder collection, chemical analysis, and curve fitting make this approach labor-intensive and costly.

To address these limitations, colorimetric methods have been developed and validated in numerous studies [11, 12, 21–24]. This technique eliminates the need for powder grinding and chemical titration. Instead, a silver nitrate (AgNO3) solution (typically 0.1 mol/L) is sprayed onto a freshly split concrete surface. The reaction between silver ions and free chlorides produces a visible white precipitate of silver chloride (AgCl), clearly marking the chloride penetration front. The derivation of the diffusion coefficient in this method relies on assuming a fixed threshold chloride concentration  at the visible color-change boundary. This threshold is widely reported in the literature as approximately 0.07 mol/L [11, 26]. Given the known surface concentration  of the exposure solution, the analytical solution to Fick's second law is applied as follows:

                                                    (2)

where  is the chloride concentration at depth  and time (mol/L),  is the threshold concentration at the color front (mol/L),  is the apparent diffusion coefficient (m2/s),  is the penetration depth (m),  is the exposure duration  and erfdenotes the Gaussian error function.

Rearranging Equation (2) to solve for  yields the general form:

                                                          (3)

In the present study, the exposure solution concentration was 2.8 (mol/L). Substituting  = = 2.8 (mol/L) and  = 0.07 (mol/L) into the equation above, the term  evaluates to a constant value of approximately 10.05. This constant incorporates both the inverse error function calculation for the specified concentration ratio and the necessary unit conversion factors to accommodate practical laboratory units. Thus, for the specific conditions of this study, the simplified expression for the non-steady-state diffusion coefficient  is given by:

                                                                      (4)

where  is the measured color-change depth (mm) and  is the exposure duration (h). This formulation directly yields  in units of 10–12 m2/s.

Fig. 2a shows a specimen subjected to the chloride solution from the very early age and in Fig. 2b, a split specimen after being sprayed with silver nitrate, showing the color change depth  is presented.

b

a

Figure 2. Very early age exposure to chloride solution and colorimetric determination of chloride ingress: a) specimen under exposure from the time of casting; b) split specimen after being sprayed with silver nitrate.

3.Results and Discussions

Electrical Resistivity

In Fig. 3, time variations of electrical resistivity of the mixes studied are presented. As shown for all mixes during the very early periods, i.e., up to about five hours, there is a slight decrease in electrical resistivity, which is due to the dissolution of ions into pore water. However, with the onset of setting and initial phases of the formation of microstructure, the resistivity increases. As expected, the mixes show higher resistance values. Partial replacement of cement with silica fume does not have a significant effect on concrete resistivity up to the age of 7 days. This is due to the opposing effects of the incorporation of silica fume on the ionic concentration of pore liquid during this period. These include the dilution effect due to partial replacement of cement, and the acceleration of cement hydration due to the nucleation effect of silica fume particles [26–29]. However, with the onset and progress of pozzolanic activity of silica fume and the resulting pore refinement and densification of microstructure the resistivity of silica fume mixes substantially exceed those of the control mixes.

Figure 3. Electrical resistivity of concrete mixes up to 360 days.

Compressive Strength

The compressive strength of the mixes studied at various ages from 1 to 360 days is presented in Table 4. As expected, the lowering of the w/c ratio has had a considerable influence on increasing the compressive strength of all ages. Partial replacement of cement with silica fume has also had a considerable influence in increasing concrete strength which is due to mechanisms including, pore size refinement, microstructure densification, and improved paste aggregate transition zone [27–29].

Table 4. The results of the compressive strength test.

Mixture

ID

Compressive strength (MPa)

–

1 day

3 day

7 day

28 day

90 day

180 day

360 day

0.5-Ref

Mean

9.9

24.5

34.6

48.0

52.9

57.1

60.1

COV*%

1.7

1.2

4.6

2.2

1.4

5.2

3.1

0.5-SF10

Mean

7.7

25.7

36.7

58.5

68.3

69.3

69.5

COV*%

1.5

1.4

1.9

3.7

3.3

1.8

4.2

0.4-Ref

Mean

11.7

28.6

41.6

55.4

61.6

68.1

69.0

COV*%

2.5

0.4

0.8

5.8

2.2

1.7

3.3

0.4-SF10

Mean

13.9

33.0

47.1

71.0

80.7

83.5

84.8

COV*%

1.7

1.3

3.9

4.2

2.2

3.3

3.1

0.3-Ref

Mean

14.6

38.3

54.1

62.7

75.1

78.9

82.2

COV*%

3.4

1.4

2.1

4.9

1.6

3.7

2.5

0.3-SF10

Mean

18.9

46.0

65.3

92.8

104.4

105.7

114.6

COV*%

4.1

2.1

1.2

2.6

2.5

4.1

0.4

* Coefficient of Variation

Rapid Chloride Migration Test

Results of the RCM test are presented in Table 5 and Fig. 4. As can be seen, the diffusion coefficients at early ages are substantially higher than the later age values. For the reference mixes with w/c ratios of 0.5, 0.4, and 0.3, the values at the age of 1 day are, 8, 11, and 12 times higher than their respective 180-day values. The incorporation of silica fume has not had a considerable influence on the diffusion coefficients at the ages of 1, 3, and 7 days. The opposing effects of dilution and acceleration due to partial replacement of cement with silica fume appear to have canceled each other out at these ages. However, with the onset and progress of pozzolanic reactions of silica fume and the concomitant improvement in the microstructure, drastic reductions in diffusion coefficients of these mixes are observed. For silica fume mixes with w/c ratios of 0.5, 0.4, and 0.3, the values at the age of 1 day are 36, 40 and 55 times higher than their respective 180-day values.

Table 5. Results of the RCM test.

Mix ID

Diffusion coefficient (10–12 m2/s)

–

1 day

3 day

7 day

28 day

90 day

180 day

360 day

0.5-Ref

Mean

132

47

34

21

19

16

15

COV*%

6.2

4.6

2.8

5.9

7.1

3.2

5.7

0.5-SF10

Mean

142

68

40

15

7

4

4

COV*%

4.7

5.3

5.4

5.2

3.2

0.9

1.1

0.4-Ref

Mean

118

41

26

17

13

11

10

COV*%

3.9

7.4

2.5

5.8

8.1

7.1

2.3

0.4-SF10

Mean

120

42

26

8

4

3

3

COV*%

7.1

1.4

2.4

7.9

6.5

4.6

2.3

0.3-Ref

Mean

107

32

19

15

11

9

7

COV*%

5.2

9.2

5.8

6.2

7.9

8.1

9.0

0.3-SF10

Mean

109

35

20

8

4

2

2

COV*%

1.3

8.1

5.9

4.4

9.3

5.7

8.5

* Coefficient of Variation

Figure 4. Diffusion coefficient (RCMT) at various ages for the mixes with w/c ratios of: a) 0. 5, b) 0.4, and c) 0.3.

Colorimetric Test under Natural Diffusion Conditions

The results of the colorimetric determination of the chloride penetration depth under natural diffusion conditions for the mixes studied are presented in Table 6. The results presented cover a large range of exposure intervals including; from the time of casting to the ages of 1, 3, 7, 28 and 180 days and also exposure intervals of 1–3, 3–7, 1–7, 7–28, 1–180, 3–180, 7–180 and 28–180 days. The diffusion coefficients derived from the colorimetric tests for the reference mixes at w/c ratios of 0.5, 0.4, and 0.3 are presented in Fig. 5.

Table 6. The results of colorimetric determination of the chloride penetration depth for various exposure intervals (t1–t2) days.

Mix ID

Depth of chloride penetration (mm)

0-1

0-3

0-7

0-28

0-180

1-3

3-7

1-7

7-28

1-180

3-180

7-180

28-180

0.5-Ref

24

27

23

28

54

18

13

19

19