Testing the Bromide penetration resistance of concrete: substitution of NaCl by NaBr in Rapide Chloride/Bromide Migration Test (RCM/RBM)

. The chloride migration coefficient of concrete describes the resistance of concrete against chloride ingress. This important material variable can be determined with the rapid chloride migration test (RCM). During the test, chloride ions penetrate concrete by means of an applied electrical field. If the concrete has already been exposed to chlorides before the test starts, the test set up has to be adapted to achieve reliable results. Due to repeated exposure to the same salt (e.g. NaCl), there is no direct possibility to distinguish the chloride ions from the test procedure and the exposure before this, within the initial test procedure. Therefore, the RCM-test setup has been adapted by replacing NaCl through NaBr in the same molar amount. Consequently, this adapted test setup is called the rapid bromide migration test (RBM). The indicator silver nitrate AgNO 3 forms sparingly soluble silver salts (AgCl, AgBr) in the presence of halogen ions (e.g. Cl - , Br - ), which form a white precipitate in areas with Cl - or Br - ions. The penetration depths are measured at two split cylinder halves. In case the RBM-specimen has previously been exposed to NaCl, an analyzation technique has to be available for the two elements Cl and Br to enable a differentiation. As the white precipitate areas - a result of the formation of AgCl and AgBr - have a quite similar visual appearance, a differentiation cannot be achieved and other analytical techniques have to be considered. With laser-induced breakdown spectroscopy (LIBS), it is possible to determine the depth-dependent ion quantity (wt.-% of Cl or wt.-% of Br). In this study, it is shown by test results that it is possible to exchange NaCl by NaBr in the RCM test, since the penetration behavior of both ions (Cl - , Br - ) is very similar in the migration test


Introduction
The conventional rapid migration test (RCM test = Rapid Chloride Migration test) according to [10] is used to determine the chloride penetration resistance of noncracked concrete that has not yet been exposed to chlorides.Due to the relatively short test duration, the test is extensively used for research purposes.The derived material parameter from this test is the chloride migration coefficient, which is an input variable in service life calculations.These calculations enable a prediction of the service life of reinforced concrete elements linked to reinforcement corrosion.
In the area of research, it may also be of interest to define this important material variable for concrete, which has already been exposed to chlorides.By replacing the salt NaCl (sodium chloride) with NaBr (sodium bromide) in combination with the application of an analytical system such as laser-induced breakdown spectroscopy it becomes possible to investigate the ion penetration resistance of concrete already been exposed to NaCl on a laboratory scale.
Compared to [9], in which NaCl is replaced by NaI (sodium iodide), in this study a more economical possibility of substituting NaCl with NaBr in the (RBM test = Rapid Bromide Migration Test) is presented.
The concrete studied was not exposed to any salt before the test started to investigate the ion transport behaviour for each NaCl and NaBr.
By using this procedure, the time dependent development of the chloride migration resistance can be investigated in more detail also for in-situ concrete.

Chemical properties of Cl and Br
To adapt the test method, it is necessary to replace chlorides by bromides.Therefore, some chemical properties of chloride and bromide are being discussed.Table 1 lists selected characteristics of chlorine and bromine that may have an influence on the penetration behaviour of the ions in the migration test.The comparison of the chemical properties shows some very similar characteristics [11].The solvation number of chloride and bromide is given in a range from 2 to 5 depending on the test methods [6].The same source specifies the solvation number of the compound NaCl as 7.1 and for NaBr as 6.5.The solvation number describes the number of H2O molecules accumulating around the ion.The ionic radius of Br is about 8 % larger compared to chlorine.Since the ionic radii show little difference, it can be assumed that the number of accumulated H2O molecules around chlorine and bromine is similar.

Experimental set up of RCM/RBM tests
The execution of the test procedure for determining the chloride penetration resistance of concrete can be taken from the standard DIN EN 12390-18 [3].In Fig. 1 the experimental setup is shown für two series with each n = 3 for the RCM and RBM test.In addition to the test procedure with 10 wt.% NaCl in 0.2 N KOH, it is possible to reduce the salt content according to the standard.The results and interpretations shown here are based on migration tests carried out with a salt content of 3 wt.%NaCl in 0.2 N KOH.Compared to the conventional rapid chloride migration test with NaCl, the RBM test uses the salt NaBr in the same concentration.The other test parameters and the other procedure remain unchanged.The NaCl concentration of 3 wt.%NaCl in 0.2 N KOH is approx.0.5 mol/l according to equation (1) [1].
where is : Since the atomic masses of bromine and chlorine are different (Table 1), the weighed quantity adjusted for NaBr is calculated by changing equation ( 1) to m(NaBr) in equation (1-1) to obtain the same concentration of c(NaBr) = c(NaCl) = 0.5 mol/l in 0.2 N KOH.

Concrete mixtures, samples and measurement
For investigations of the ion penetration behaviour (Cl -or Br -) in the rapid migration test with NaCl or NaBr in a 0.2 N KOH solution, concretes with four different binders were tested.The binders CEM I, CEM I+SF (f=1.0,silica fume content = 5.8 wt.%), CEM II/B-S and CEM III/A were tested.Furthermore, for the concretes produced with CEM I and CEM I+SF, the water-binder value were varied, cp.Table 2.
Due to the high specific surfaces of silica fume and the rather low w/b values, the binder content had to be increased to ensure proper workability.The aim of the concrete mixtures was to achieve reliably a consistency class of F3 with a low use of plasticizer.The basis of the following measurements is the migration test according to DIN EN 12390-18 [3].After the test, the sample (cylinders) are splitted in two halves.For the presented investigations, the penetration depth was measured with the indicator (0.1 N AgNO3) only on onehalf (instead of both) of each sample.This deviates from the outlined procedure in the standard, where this measurement has to be performed on both halves of the cylinder.The other halves were used for the LIBS analysis to measure the elements chlorine and bromine quantitatively.Here concrete surfaces without the indicator applied had been used to avoid influences from the silver salt formation on the measurement results.By using LIBS the penetration depth of chlorine and bromine can be measured simultaneously.These investigations are not part of this publication.
For the investigations presented here, the penetration depth was determined optically according to the standard.The basis for this procedure is the formation of white reaction products leading to white areas on the concrete surface.These poorly soluble silver salts AgCl or AgBr are developed within the reaction of the indicator (0.1 N AgNO3) and the chlorides and/or bromides.The boundary between the white and brown areas is basically understood as the penetration depth and was determined on the fracture surface of each sample at nine measuring points of three samples.

Correlation of test results
In order to test whether the experimental results of the migration tests with NaCl or NaBr show a linear correlation NaCl has been replaced by NaBr in the test solution of the migration test, the confidence band and the probability with which the null hypothesis [4] is rejected were calculated using a linear regression model.The observations were carried out at a significance level of α = 5%.
If the data points consisting of the arithmetic mean values of the penetration depths from the RCM and RBM tests are within the confidence bands [5], it can be assumed with 95% probability that the unknown sample values of the penetration depths are within the lower and upper confidence limits.The narrower the confidence band, the less the individual results scatter around the expected value.Fig. 1 shows that all mean values of the samples are within the confidence band and that the confidence band is relatively narrow, which indicates low scatter.
Furthermore, it was tested whether the null hypotheses for the slope and the offset of the ordinate (y-axis intercept) of the transfer function f(x) in Fig. 2 are rejected.If the null hypothesis is not accepted, the slope and the offset are not equal to 0 with the probabilities given in fig. 1 and can thus be used for the transfer function of the mean penetration depths of AgBr on AgCl.The coefficient of determination R² = 0.995 also proves the good correlation of the migration experiments with NaCl and NaBr.

Probability density function and student test (t-Test)
The following figures Fig. 3 and Fig. 9 show the probability density function and the corresponding histograms of the measured penetration depths of AgCl and AgBr, respectively, after the indicator (AgNO3) was sprayed on the splitted surfaces of the specimens.The probability density functions have been calculated twosided at a significance level of α = 5%, i.e. approx.95% of the random variables (measuring points) lie within the interval of µ ± 2-σ with a scatter of 2-σ.
The expected values µCl and µBr (arithmetic mean of three sample halves of nine measuring points each; n=27), the standard deviations σCl and σBr and the corresponding ranges in which probably 5% of the measured values are not located are plotted.
Fig. 3 and Fig. 4 show the probability density functions of the concrete with CEMI with the w/b values 0.4 and 0.5.After the student test, in which the equality of two mean values is tested, the null hypothesis is rejected with a probability of p=0.4 for the concrete with CEMI, w/b=0.40 and p = 2.7 for the concrete with CEMI, w/b=0.50.This means that in these two samples the mean values are not equal.This means that there is no equality of the mean values of the penetration depths of chloride (AgCl) and bromide (AgBr) in these two samples with the previously stated probability at a siginificance level of α = 5%.The following three figures Fig. 5 to Fig. 7 of the concretes with the binder CEMI+SF fig. 1 to fig. 7 show an increasing dispersion (flattening normal distributions) depending on increasing w/b-values (0.4, 0.5, 0.6).
Due to the increasing water content at the same binder content, it can be assumed that the pore system has increasingly more capillary pores and thus the ion penetration (Cl-or Br-) becomes more uneven.The null hypotheses of the t-tests of all three concretes with different w/b values must be accepted.
The probability that the null hypothesis is true for the concrete CEMI+SF with w/b = 0.4, i.e. that the expected values of the averaged ion penetration depths (Cl, Br) are equal, is p = 17.5% for w/b = 0.6 is p = 7.0% and for w/b = 0.6 is p = 15.3%.In the following figure Fig. 8 of the concrete produced with the binder CEMII/A-S (Portland cement clinker and a blastfurnace slag content between 21-35 wt.-% according to [4]), the mean values hardly differ and the ttest with a probability of p=79.2% states that the penetration depths are likely to be similar, but the individual measuring points scatter strongly.In the following figure Fig. 9 of the concrete with the binder CEMIII/A (Portland cement clinker and a granulated blastfurnace slag content between 66-80 wt.% according to [4]), the scatter lies in a moderate range.The t-test shows with a probability of p=23.3% that the expected values of the penetration depths of AgCl and AgBr are equal.

Comparison of the colours AgCl and AgBr
To give an impression of the intensity of the "white discolouration" after the application of the indicator AgNO3, which leads to the formation of poorly soluble silver salts (AgCl or AgBr) in the presence of AgCl or AgBr, exemplary pictures are shown below for the concrete produced with the binder CEMI+SF and a w/bvalue of 0.50.The white intensities between AgCl and AgBr do not appear to differ from each other.The white intensities between AgCl and AgBr do not seem to differ.

Current flow rates
The amount of electrical charge introduced can be derived from the current flow rates [12].The more charge is introduced, the less resistance the corresponding concrete has to the chloride or bromide ions.
In order to investigate only the parameter of salt influence, all concretes were tested at a voltage of 30 volts for approx.24 hours, regardless of the initial current.The current flow decreases linearly in all concretes after an initial increase in current flow until the end of the test.In order to compare the charge quantities of chloride and bromide, the time-dependent integrals of the current flows were calculated using the trapezoidal method and compared with each other.Based on the standard deviations (SD) of the current flow rates of the tested three test specimens as a function of the time intervals in which the current flows were recorded, an approximately equal behaviour of the two salts NaCl and NaBr in the migration test can also be observed here.There is no consistent trend that the charge quantity of the tests carried out with NaCl always show a slightly larger integral and thus larger electric charge quantities.
The comparable magnitudes of the integrals, however, suggest a similar electric charge quantity.This additionally proves the possibility of the exchangeability of the salt NaCl by NaBr in the migration experiment according to [3].
Since the current flow rates are also dependent on the specimen geometry and thus on the heights of the cylinders used in the migration test, the determined integrals are subject to manufacturing-related fluctuations, which can also lead to minor differences in the electrical charge quantities.The tolerances according to [3] for the heights of the cylinders are ± 2 mm.

Summary
The investigations have shown that the replacement of the salt NaCl by NaBr in the conventional rapid migration test is possible.The equality of the penetration depths of the poorly soluble silver salts AgCl and AgBr, which form on the split test specimens after the migration test, are predominantly comparable both in their appearance and in the penetration depths.The coefficient of determination R² in the comparison of the averaged penetration depths of the samples of the concretes CEMI, CEMI+SF, CEMII/A-S and CEMIII/B examined here is 99.5%.The transfer function of the penetration depths of bromide to chloride in concrete after the migration test may be applied according to statistical investigation.Likewise, the scatter of the individual measurements of the penetration depths behaves comparably, so that it cannot be assumed that the substitution of NaCl by NaBr leads to a larger scatter.
With suitable analytical methods such as laserinduced breakdown spectroscopy, concrete that has already been exposed to NaCl it is probably possible to investigate the bromide penetration resistance in the migration test with NaBr (RBM test) [2].Further investigations have been [8] and have to be made to use the RBM test getting more information about concrete which has already been exposed to NaCl to implement this additional information in the service life considerations [7].

Fig. 1 .
Fig. 1.Experimental setup for RCM (I, above) and RBM (I, bellow) test and data logger for voltage, current and temperature (II)

Fig. 12 Fig. 12 .
Fig. 12. Example of the current flow rates and electric charge calculated by the time depending integral of a Rapid Migration Test with NaCl and NaBr of concrete with binder CEMI+SF, w/b = 0.50

Table 1 .
Chemical properties of Chlorine and Bromine

Table 3
lists the time-dependent integrals of the electrical charge quantities entered in the migration test with NaCl or NaBr.

Table 3 .
Integrals of electric charges of a Rapid Migration Test with NaCl and NaBr of selected concretes