Ultramarine blue pigment degradation in cementitious materials: a new approach to the phenomenon

2.2.1. Study of pigment deterioration in different solutions

In the initial phase of the study, the process of colour loss of the pigment in solution was addressed, to evaluate the influence of different characteristic parameters of the cementitious medium in the degradation process (such as pH, alkali content and Ca²⁺ ion content).

Dissolution studies were carried out based on a modification of the method developed by Chinchón-Payá et al. (2424. Chinchón-Payá S, Aguado A, Chinchón S. 2012. A comparative investigation of the degradation of pyrite and pyrrhotite under simulated laboratory conditions. Engineer. Geol. 127:75-80. https://doi.org/10.1016/j.enggeo.2011.12.003.
) for the study of the oxidation of pyrites and pyrrhotites. Three different solutions were used, with a constant O₂ contribution, to study the above indicated parameters: a neutral solution (NS, medium only with water); a saturated solution of Ca(OH)₂, named SSC; and a solution that simulated the aqueous phase of the pores, named SSP (using KOH 0.6 mol/l; NaOH 0.2 mol/l; and Ca(OH)₂ until saturation). The solid / liquid ratio used was 1/20 (20 g of pigment per 400 ml of solution). The test duration was 60 days, during which the pH of the solution was periodically monitored. Likewise, samples of solution were extracted at 1, 3, 17 and 60 days for chemical analysis by ion chromatography and inductively coupled plasma - optical emission spectrometry (ICP-OES).

Through ion chromatography, the concentration of sulphates in the different media was monitored. The increase in sulphate concentration is directly related to the pigment degradation. As this occurs, the chromophore can get rid of the structure and oxidise. The following elements were analysed by ICP-OES: Na, K, Ca, Fe and Al. This allowed the analysis of the variation of its concentration as a function of the reaction time for the different media.

At the end of the test, the remaining solids were characterised by XRD, XRF and infrared spectroscopy and the results were compared with the analyses performed on the solid samples before beginning the dissolution analysis. Colorimetric analyses were also performed to determine the colour change in the samples.

2.2.2. Study of pigment deterioration in cement pastes

In the second stage of the experimental campaign, the evolution of the hydrated phases and the pigment in cement pastes were studied, using point and continuous DRX measurements. In order to do this, cement pastes were manufactured incorporating the pigments studied at 4% on the weight of the cement (w_c). In all the pastes, a fixed cement content (300 g) and a water/cementitious material ratio of 0.35 (corresponding to the amount of water necessary to obtain a normal consistency paste) were used.

2.3.1. X-Ray Diffraction measurement

Two types of XRD analysis were performed: point measurements and continuous measurements. For the point measurements, a PANalytical X’Pert PRO MPD Alpha1 was used, working in reflection with Bragg-Brentano geometry. CuKα radiation (λ = 1.5418 Å) filtered by Ni and an X’Celerator detector (active length of 2,122°), operating at 40 kV and 45 mA were used. The analysis conditions were between 4° 2θ to 80° 2θ, using an angular pitch of 0.017° 2θ and a step time of 50 seconds, with a variable divergence slit.

Continuous XRD measurements were performed following the same methodology used by Salvador et al. (2525. Salvador RP, Cavalaro SHP, Segura I, Figueiredo AD, Pérez J. 2016. Early age hydration of cement pastes with alkaline and alkali-free accelerators for sprayed concrete. Constr. Build. Mater. 111:386-398. https://doi.org/10.1016/j.conbuildmat.2016.02.101.
) to study the evolution of the hydration of cement pastes with setting accelerators. To this end, cement pastes were manufactured shortly before their measurement with an w/c ratio of 0.4, 4% w_c pigment. An internal alumina standard (NIST) was incorporated at 10% w_c. To avoid water evaporation during the measurements and carbonation of the samples, the sample holders were covered with a Kapton® film, with a thickness of 7.5µm. The temperature remained constant during the measurement process at 26 ^oC. The measurements were made in a PANalytical X’Pert PRO MPD θ / θ device, working in reflection with Bragg-Brentano geometry. CuKα radiation (λ = 1.5418 Å), filtered by Ni, and a PIXcel detector (active length of 3,347°) were used, operating at 40 kV and 45 mA. Cylindrical sample holders, 32 mm in diameter and 3 mm deep, were used, which amounted to the use of approximately 3.7 g of paste for each test. The analysis conditions were between 5° 2θ and 70° 2θ, using an angular pitch of 0.026° 2θ and 60 seconds of passage time, with a fixed divergent slit of 0.5. A diffractogram was recorded every hour for the different samples, obtaining the first diffractogram 6 minutes after mixing the cement with the water. During the first 48 hours after mixing the cement with the water, diffractograms were obtained at the following times: 6 minutes, 1 hour, 12 hours, 24 hours and 48 hours. Once the continuous monitoring was completed, the samples were extracted and kept in containers with water, to facilitate the hydration process and to accelerate the possible degradation process. The samples, thus preserved, were maintained for a maximum of 24 days, obtaining diffractograms at 3, 5, 7, 10, 14, 17, 21 and 24 days. At the end of the study, all samples had lost the blue hue. The different diffractograms obtained were analysed qualitatively using X’Pert High Score Plus software.

2.3.2. Chemical analysis methods

The ICP analyses were performed with a Perkin Elmer Optima 4300DV Inductively Coupling Plasma emission spectrometer. The ion chromatography was carried out in a Dionex DX500 ion chromatography System. Likewise, quantification of the oxide content was carried out by X-ray fluorescence, in a sequential X-ray spectrometer (PHILIPS MAGIX PRO) equipped with a rhodium tube and beryllium window.

The characterisation of the solids before and after the experiment was carried out by infrared (IR) spectroscopy using an FT-IR Spectrometer (Bruker IF66 / S).

2.3.3. Colorimetric analysis

The analysis was conducted using an X-rite spectrophotometer (model RM200QC) with a measurement area of 8 mm in diameter, independent tri-directional 25 LED light source, D65 illuminant and an observer angle of 10^o. Colour was expressed in the CIE L*a*b* space (2626. Teichmann G. 1990. The use of colorimetric methods in the concrete industry. Beton. Fertig.-Tech./Concr. Precast. Plant Technol. 10:58-73.
, 2727. López A, Guzmán GA, Di Sarli AR. 2016. Colour stability in mortars and concretes. Part 1: Study on architectural mortars, Constr. Build. Mater. 120:617-622. https://doi.org/10.1016/j.conbuildmat.2016.05.133.
). The parameters measured were the lightness L* (varying from black to white: 0 and 100, respectively), a* (which varies from +a* and -a*, red and green, respectively) and b* (with ranges from +b* and -b*, yellow and blue, respectively). The total colour change ΔE^* ₇₆ was computed using successive measurements performed at different times:

where DL*= L*_t - L*₁; Da*= a*_t - a*₁ and Db*= b*_t - b*₁, which are the colour differences between age 1 and age t, according to the comparison of the colour before and after the exposure, respectively. For the purposes of the evaluation of the total color change, the criteria indicated in Table 2 were adopted, according to (2626. Teichmann G. 1990. The use of colorimetric methods in the concrete industry. Beton. Fertig.-Tech./Concr. Precast. Plant Technol. 10:58-73.
, 2828. López A, Di Sarli AR. 2020. Measurements number in cementitious mixtures to define the colour and its homogeneity. Constr. Build. Mater. 238:117636. https://doi.org/10.1016/j.conbuildmat.2019.117636.
, 2929. López A, Tobes JM, Giaccio G, Zerbino R. 2009. Advantages of mortar-based design for coloured self-compacting concrete. Cem. Concr. Compos. 31(10):754-761. https://doi.org/10.1016/j.cemconcomp.2009.07.005.
).

3.1.1. pH evolution

Figure 2 shows the evolution of pH with exposure time, depending on the type of sample and the environment studied. The medium with the highest pH value is the one that simulates the aqueous phase of the pores (SSP). In general, there is a decrease in pH in all media regarding values at early ages, although there are differences regarding the speed of change.

The neutral solution (NS) and the saturated solution of calcium hydroxide (SSC) stabilised at a pH between 8.5 and 9.0, by the second or third days after starting the studies. In this sense, there are no significant differences in behaviour between the two pigments studied. The rapid and significant variation of pH in the SSC medium was equal to the pH values of the NS medium. The effect of CO₂ (carbonation) on pH decrease is clear in both SSC and SSP mixtures. The medium that simulates the aqueous phase of the pores (SSP) maintains high pH values for longer time due to larger amount of OH ions (Ca(OH)₂ has, by contrast, low solubility product). In this case stabilisation occurs around 20 days with a pH value close to 10.5. In this medium, behavioural differences are observed between the two pigments studied, the main one being the fact that it took twice as long for the UB1 sample to achieve pH stabilisation in relation to the UB0 sample (16 days and 8 days, respectively).

3.1.2. Ion chromatography

The sulphate concentration in the different media is monitored by ion chromatography. Figure 3 shows this variation depending on the type of sample, medium and time of sampling. Each value represented in Figure 3 is the mean of the results of 5 samples, the coefficients of variation obtained by a basic statistical analysis of the results in all cases were less than 1%.

In all of the samples studied, an increase in the sulphate content occurs with the exposure time. However, this increase is more noticeable in the UB1 pigment samples and in the samples exposed to the solution that simulates the aqueous phase of the pores (SSP). These results show the different behaviour of both pigments, as well as the different aggressiveness of the solutions under study.

3.1.3. ICP-OES

The following elements have been analysed using the ICP-OES technique: Na, K, Ca, Fe and Al. Table 3 shows the variation of its concentration as a function of the reaction time. It is worth noting that Fe ion was not detected in any of the solutions.

First, the variation of the calcium ion content will be analysed. The variations obtained differ significantly depending on the medium, so they will be analysed separately. On the one hand, the samples contained in the neutral solutions (NS) hardly show any increase in the calcium content, with only minor increases occurring for the UB0 sample. On the other hand, saturated solutions of calcium hydroxide (SSC) show significant differences in calcium ion content between both pigments. Considering that it initially started from a concentration of 1850 ppm of calcium in solution (saturated medium), the decrease in calcium content in solution may be due to both environmental carbonation and the incorporation of the calcium ions into the pigment structure. The differences obtained between the pigment samples show a reaction of the calcium in solution with the pigment, which was much more significant in the UB0 sample. Finally, in the SSP medium, the observed behaviour is anomalous a priori. For both samples studied, during the first few days, it was not possible to detect calcium in solution; subsequent measurements showed an increase in its concentration and a subsequent disappearance or decrease.

In the case of the variation of the potassium ion content, the variations are significantly different depending on the type of medium. In the neutral solution and that saturated with calcium hydroxide there are very low contents of this ion (below 13 ppm in all cases). The variations observed in these cases are within the detection limit of the technique and cannot be attributed to specific behaviours. The solution that simulates the aqueous phase of the pores (SSP) shows an increase in the potassium ion content in solution throughout time, without appreciating significant differences between both pigments.

Studies carried out by Booth et al. (88. Booth DG, Dann SE, Weller MT. 2003. The effect of the cation composition on the synthesis and properties of ultramarine blue. Dyes Pigmen. 58(1):73-82. https://doi.org/10.1016/S0143-7208(03)00037-8.
) revealed that the modification in cation type and content in the pigment structure leads to an increase in the size of the unit cell (larger structures), and also to interactions between the cations and the chromophore that can lead to a colour change. Based on the data in Table 1, the potassium content in the pigments was less than 1%. Likewise, the initial content of potassium ion in solution was 24,000 ppm (0.6 mol / L). The K concentration increase in SSP along time, evidencing a possible ion-exchange phenomenon between the pigment and the solution.

The variation in sodium ion content shows similar trends for the different samples and media studied, as can be seen in Table 3. In all cases, a continuous increase in the sodium ion content is observed throughout time, both in the UB0 sample and in UB1.It is noteworthy that the ion content that is present in solution differs significantly between the samples exposed to the medium that simulated the aqueous phase of the pores (SSP) and the rest of the samples (NS and SSC). The initial sodium ion content in that medium was approximately 4,600 ppm, which shows the ion exchange process that is taking place.

Finally, by analysing the aluminium ion content in a general way, in the different samples and media studied, it can be seen that its variation is minor, with the exception of the samples exposed to the medium that simulates the aqueous phase of the pores (SSP). In this medium, both UB0 and UB1 show an increase in the content of aluminium ions in the first 3 days subsequently, show a reduction. In the case of UB0 it goes from a concentration of 203 ppm to one of 11 ppm. This abrupt variation in the content of the aluminium ions may be related to some precipitation or reaction process.

The results presented show that:

3.1.4. Solids analysis

Different samples of solids were taken at the same times as the aliquots for the solutions analyses (at 1, 3, 18 and 60 days). The results obtained by XRD and XRF were compared to analyses carried out on the solid samples before beginning the dissolution analyses. Similarly, infrared spectroscopy analyses were performed on both non-degraded pigments and the final solids of each of the degradation experiences (after 60 days of exposure). The results obtained are presented and discussed below.

Diffractograms obtained for both pigments preserved in the neutral solution (NS), at the different ages studied, did not show significant differences, being consistent with the results of the analysis of the solutions, where no significant changes were evident.

Diffractograms of the samples of both pigments exposed to the lime saturated solution (SSC) showed some differences compared to the untreated sample. The main difference found was the appearance of new peaks at different angular spacings, in the 29 to 30 ° 2θ and 43 to 50 ° 2θ intervals in the UB0 pigment, and in the first interval only in the UB1 pigment. These corresponded to the presence of calcite and could be induced during sample preparation of UB0 for the XRD; in pigment UB1 they were less noticeable than in UB0.

Diffractograms of the two pigments that were exposed to the SSP medium, showed greater differences. The main variations were in relation to the intensity losses in the peaks (displacement to 2θ lower values and an increase in the width of the peaks) as the exposure time of the pigments in solution progresses. The effect of the degradation process is shown in Figure 4, with the variation of the intensity and spacing of the peak located at 24 ° 2θ.

The decrease in the intensity of the peak shows that the crystalline structure is changing, while the modification of spacing d indicates that an increase in the size of the unit cell (and its distortion) is taking place.

As the degradation process progresses, the structure degrades, as shown in Figure 4b . This modification leads to destabilisation and oxidation of the S₃- chromophore and, therefore, to loss of pigment colour.

The analysis of both parts of the figure shows that the degradation phenomenon occurs more significantly in UB0 than in UB1, although both samples show evidence of modifications in their crystal structure. The modification that takes place is a distortion of the lazurite-like structure of the pigment and a loss of crystallinity, this may be associated with the formation of a new phase.

Figure 5 shows the diffractograms of UB0 and UB1 after times 0 and 60 days in the SSP medium, showing the coincidences with the hauynite phase (showed as dashed lines on the diffractograms). Previous results presented by Giraldo (2323. Giraldo C, Tobón JI, Restrepo OJ. 2012. Ultramarine blue-pigment: A non-conventional pozzolan. Constr. Build. Mater. 36:305-310. https://doi.org/10.1016/j.conbuildmat.2012.04.011.
) also evidenced the formation of this phase. The new phase that is formed could be related to hauynite, since the displacement of the peaks coincides with that of this phase and has the chemical formula: K_1,6Ca_2,4Na_4,32(Al₆Si₆O₂₄)(SO₄)_1,52. It is also an alumino-silicate, but has incorporated calcium and potassium into its structure.

The degradation of the pigment structure is observed in the X-ray diffractograms. Peak modifications have been noted, as well as distortion of the unit cell. According to (33. Landman AA. 2003. The role sulphur plays in the formation of ultramarine blue, in: University of Pretoria ed., Aspects of solid-state chemistry of fly ash and ultramarine pigments, PhD These, Chapter 4, Pretoria, 65-76.
), the diffractograms of the samples depend fundamentally on the structure of the aluminosilicate. Ultimately, it is this modification of the aluminosilicate structure that favours the loss of colour of the pigment.

The chemical composition of the solids obtained after 60 days of treatment was determined by means of X-ray fluorescence and compared with the untreated samples of the pigments. Likewise, the loss on ignition (LOI) of the different samples was determined. The results obtained are shown in Table 4, for the different environments studied.

It is worth noting that the pigments were mainly composed of silicon, sodium and sulphur, as shown in Table 1, and that they hardly showed any trace of calcium or potassium in their composition.

Firstly, the variation of aluminium and silicon content was analysed, since it may show modifications in the structure of the zeolite that contains the chromophore. In both cases, the observed variations do not clearly show the effect of the degradation process.

It must be taken into account that X-ray fluorescence provides contents in percentages, so the variation of one of the components can affect the others. In this sense, the analysis of the Si/Al ratio in the pigment can provide more information on the degradation phenomenon, as can be seen in Table 4. This relationship hardly varies in the UB0 samples while there is a slight variation in the UB1 samples exposed to SSC and SSP environments, showing a slight loss of aluminium in the structure.

The analysis of the other major pigment component, sodium, shows that, in both pigments, the samples exposed to the SSP environment show significant reductions in its content. In the analysis of the major components of the pigments, it is worth noting the variation in the sulphur content. In this sense, a noticeable decrease in the sulphur content is observed for all samples and studied environments, although in SSP this decrease was the smallest. In ultramarine pigments, sulphur can be found in the form of a chromophore (S₃-) or in the form of a sulphate ion (77. Climent-Pascual E, Sáez-Puche R, Gómez-Herrero A, Romero J. 2008. Cluster ordering in synthetic ultramarine pigments. Micropor. Mesopor. Mater. 116(1–3):344–351. https://doi.org/10.1016/j.micromeso.2008.04.026.
), the former providing the blue colour to the pigment. The sizes of the two ions differ markedly, and probably this difference in size allows ion exchange of the sulphate anions to take place in all of the media studied.

Another factor to consider is the loss on ignition (LoI), which makes evident the presence of hydrated or carbonated compounds in the pigments. The results obtained show increases in these in all environments, being more remarkable in the samples exposed to the SSP environment. This fact could be the result of the reaction or incorporation of calcium, in the form of portlandite, to the pigment and its subsequent carbonation with ambient CO₂. The modification of the aluminosilicate structure that makes up the pigment cell, has been confirmed by the decrease in the contents of the major components studied in the pigments.

After this, the content variation of the minority elements (calcium, potassium and phosphorus) was analysed. Regarding calcium, there is a remarkable increase in content in samples exposed to the medium saturated with lime (SSC). This fact may be related to the incorporation of this ion into the pigment structure. It is noteworthy that this increase is not seen in the samples exposed to the medium that simulated the aqueous phase of the pores (SSP).

Potassium also shows a remarkable increase in content, in the samples exposed to the medium that simulated the aqueous phase of the pores (SSP); the samples exposed to the other environments do not show significant variations in the content of this element compared to the untreated sample UB0. As previously showed by Booth et al. (88. Booth DG, Dann SE, Weller MT. 2003. The effect of the cation composition on the synthesis and properties of ultramarine blue. Dyes Pigmen. 58(1):73-82. https://doi.org/10.1016/S0143-7208(03)00037-8.
), the ion-exchange phenomenon between the SSP solution and the pigment may lead not only to an increase in the size of the unit cell (larger structure), but interactions between the cations and the S^3- chromophore may also take place, which at the end can lead to the colour change that is experimentally observed.

The presence of phosphorus in the UB1 pigment samples is interesting. The diffractograms of the untreated samples, shown in Figure 1, evidenced the presence of some type of inorganic phosphate in these samples. Table 4 shows a decrease in the phosphorus content in the UB1 pigment samples exposed to the environment that simulated the aqueous phase of the pores (SSP), compared to the original samples; the possible degradation of the coating pigment protector is evident.

Figures 6a and 6b show the different IR spectra obtained for samples UB0 and UB1 before and after exposure to the different environments evaluated. To facilitate interpretation, only the information contained between wavelengths 500 to 1200 cm^-1 is presented.

Figure 6a clearly shows that the spectra obtained from the UB0 pigment are characteristic of aluminosilicates. The characteristic peaks of the S₃- chromophore are located at 542 and 582 cm-¹ (3030. Afremow LC, Vanderberg JT. 1966. High resolution spectra of inorganic pigments and extenders in the mid-infrared region from 1500 cm^-1 to 200 cm^-1. J. Paint. Technol. 38(495):169-202.
, 3131. Sancho JP, Restrepo OJ, García P, Ayala J, Fernández B, Verdeja LF. 2008. Ultramarine blue from Asturian “hard” kaolins. Appl. Clay Sci. 41(3-4):133-142. https://doi.org/10.1016/j.clay.2007.10.005.
); in this pigment only the peak at 582 cm-1 is observed. Similar results have been obtained (3232. Landman A, de Waal D. 2004. Fly ash as a potential starting reagent for the synthesis of ultramarine blue. Mater. Res. Bull. 39(4-5):655-667. https://doi.org/10.1016/j.materresbull.2003.12.002.
), showing an intense band in that location. In this area, a certain modification in the IR spectrum can be seen in the Figure. Samples exposed to neutral and lime saturated solutions do not undergo significant modifications. However, the sample exposed to the solution that simulated the aqueous phase of the pores shows a shift of the chromophore peak to 578 cm^-1.

The peaks that appear in the 650-700 cm^-1 band and the intense peak that appears around 1000 cm^-1 are characteristic of the Si-O-Al bonds of the aluminosilicates (3333. Jakobsson S. 2002. Determination of Si/Al ratios in semicrystalline aluminosilicates by FT-IR spectroscopy. J. Appl. Spectros. 56(6):797-799. https://doi.org/10.1366/000370202760077559.
). In NS and SSC media, there are few modifications in these bands but slight variations in intensity in the 1000 cm^-1 band, which increases in intensity in the SSC medium and moves to values close to 990 cm^-1. This may show the incorporation of Ca²⁺ ions in the aluminosilicate structure. The most significant changes are seen in the sample exposed to the SSP environment. In addition to an outstanding increase in the intensity of the band at 1000 cm^-1 and a shift to values close to 970 cm^-1, an outstanding distortion of the band 650-700 cm^-1 is observed in Figure 6a. These differences reveal alterations in the Si-O-Al bonds, due to the pigment degradation process.

When analysing the spectra obtained for the UB1 sample (see Figure 6b), no significant differences were observed between the different media. There are hardly any differences between the samples exposed to the NS and SSC media, regarding the starting sample. The sample submitted to the SSP medium presents a series of differences that are similar to those found in the UB0 sample, although less intense. The different changes observed are related to the distortion of the structure of the chromophore and of the Si-O-Al bonds of the aluminosilicates.

The shoulder seen around 1100 cm^-1 (marked with an arrow in Figure 6b is related to Si-O-Al bonds, probably due to the amorphous or polymeric silica used in the coating of the UB1 sample. These results agree with previously published results (1717. Li S, Liu M, Sun I. 2011. Preparation of acid-resisting ultramarine blue by novel two-step silica coating process. Industr. Engineer. Chem. Res. 50(12):7326-7331. https://doi.org/10.1021/ie200343k.
).

3.1.5. Colorimetric analysis

The results of the colorimetric analysis of the samples of pigments UB0 and UB1 as well as those of the final solids of all the degradations studied in solution, are presented in Table 5 where the values obtained by the CIE L* a* b colour system are collected, and then represented in Figure 7. In Table 5, it can be clearly seen that the major colour differences DE^* ₇₆ occur in pigment UB0 exposed to the medium that simulates the aqueous phase of the pores (SSP), where ΔE^* ₇₆ >> 12.

Pigment	Legend	Solution	L*	a*	b*	ΔE*76	Visual assessment Of ΔE*76
UB0		Initial	51.16	15.75	-39.02
		NS	50.27	14.64	-35.45	3.84	Very obvious
		SSC	50.80	15.34	-37.34	1.77	Obvious
		SSP	47.45	12.17	-26.89	13.18	Very large
UB1		Initial	53.24	15.16	-40.23
		NS	51.15	14.17	-35.32	5.43	Very obvious
		SSC	51.62	13.77	-35.54	5.15	Very obvious
		SSP	49.60	14.49	-33.17	7.90	Large

3.2.1 Evolution in early and long-term stages

First of all, the evolution of the samples in early stages was analysed during the first 48 hours after mixing the material. The following study intervals were selected: 0.10 hours (first measurement performed), and then 1, 12, 24 and 48 hours. Figure 8 shows the evolution of the hydration of the cement pastes with the two pigments studied at the indicated ages. The main modifications in the diffractograms appear in the range 5 ° to 30 ° 2θ, in which the main detected phases have been indicated in the cited figure: Et - Ettringite, M - Monocarboaluminate, G-Gypsum, P - Portlandite, C - Calcite, L - Lazurite, A - Alumina and Q - Quartz. From this angle, the variations correspond mainly to the unreacted cement.

As the hydration process progresses, as is clearly seen in Figure 8a, portlandite, ettringite and monocarboaluminate appear as the main products of cement hydration. Likewise, and as one would expect, the disappearance of the gypsum is observed between 1 and 12 hours after manufacturing. The other phases present are calcite, alumina (used as an internal standard) and quartz. Figure 8b shows the evolution of the hydration of the cement paste with the UB1 pigment in the initial stages. The phases identified are the same as those found in the paste with UB0 pigment. For at least up to 48 hours, higher ettringite contents are not appreciated, as stated in the bibliography (2222. Restrepo J, Restrepo O, Tobón O. 2009. Evaluación del desempeño mecánico del cemento blanco coloreado con pigmento azul ultramar. DYNA 76(157):225-231. Retrieved from https://revistas.unal.edu.co/index.php/dyna/article/view/9571/11501
, 2323. Giraldo C, Tobón JI, Restrepo OJ. 2012. Ultramarine blue-pigment: A non-conventional pozzolan. Constr. Build. Mater. 36:305-310. https://doi.org/10.1016/j.conbuildmat.2012.04.011.
).

3.2.2. Long-term evolution

Once the continuous monitoring was completed, the pastes were kept in containers with water, as indicated in Section 2.3.2. In relation to the long-term evolution Figure 9 shows diffractograms of the samples made with pigments UB0 and UB1, indicating the main phases detected.

In the samples made with the UB0 pigment (as shown in Figure 9a), the abundant presence of ettringite stands out. The low intensity of the portlandite peak is also noteworthy, especially in the younger samples (3 days) that have been less exposed to environmental carbonation. Finally, it is worth highlighting that the evolution of the pigment degradation process between 10 and 14 days, is not possible to practically detect the peak with the highest intensity. The diffractograms of the samples made with the UB1 pigment present a similar scenario, as shown in Figure 9b. High ettringite contents and low portlandite contents can be repeatedly seen in all of the analysed stages. It is noteworthy that the intensities of the ettringite peaks are higher in the samples with UB1 pigment than in the samples with UB0 pigment. The peak corresponding to the pigment is appreciable for a longer time, although at 24 days its intensity is practically negligible, revealing that degradation has taken place. As a reference element, the diffractograms of the long-term pigment-free cement pastes are shown in Figure 10. Of the mentioned figure, two aspects should be highlighted: the absence of peaks corresponding to ettringite and the portlandite content.

Regarding ettringite, and in light of previous comments on the cement hydration process, it disappears 3 days after the sample is manufactured, as can be seen in Figure 10. Instead, we found the presence of various phases of monosulfoaluminate and monocarboaluminate. As sample curing progresses, the intensity of the peaks in these phases increases. Therefore, the diffractograms obtained at 21 days for the reference sample contrast with those obtained for pastes with different pigment samples. In all the pigmented samples we found high ettringite contents and the intensity of the peak of the monosulfoaluminate / monocarboaluminate phases is lower than in the reference samples.

The intensity of the portlandite peak at 3 days stands out compared to that observed in the pastes with the different pigment samples. The intensity observed in Figure 10, as well as its evolution over time, responds to the normal process of cement hydration and carbonation of the sample as it is preserved in laboratory conditions and exposure to environmental CO₂.

3.2.2 Semi-quantitative analysis

Complementary to the results shown in the previous section, and to facilitate the interpretation of the results, a semi-quantitative analysis of the evolution of the content of each phase was performed. For this, the internal standard added in each sample was used, in a known quantity. As this is an inert phase, its intensity must be constant and vary according to the experimental conditions of the diffractograms.

In this way, it is possible to obtain relative intensities. Table 6 shows the main data from the different phases analysed. The calculation of the relative intensities of the different phases has been performed according to Equation [2], in which I _i is the intensity of the main peak of the phase concerned (in calculation) and is the intensity of the main peak of the internal alumina standard (in calculation).

Figure 11 shows the variation of the relative intensities of the phases of interest (portlandite, ettringite, calcite and pigment) for pastes containing the two studied pigments.

The results corresponding to the pastes with pigment UB0 (Figure 11a), show that, firstly, the ettringite content increases continuously during the study time, stabilising between 7 and 10 days. On the contrary, the content of portlandite increases during the initial period, up to its maximum at 48 hours, then it decreases drastically until it is practically undetectable on the diffractogram.

Figure 11a shows that the content of the UB0 pigment decreases continuously from the beginning, reaching the minimum between 2 and 3 days. Subsequently it remains at a level close to the detection limit.

These results contrast with those that can be seen for the samples with the pigment UB1, whose variation of relative intensities over time is presented in Figure 11b. The most remarkable differences are found in the ettringite and portlandite contents.

Regarding the first, the contents in early ages are similar to those observed in the sample with pigment UB0. However, although there is an increase in the ettringite content over time, it is not as remarkable as the one observed in the sample with pigment UB0. An increase in the ettringite content is only observed in the sample after 14 days, reaching values much higher than those observed in the sample with UB0 pigment. This phenomenon reveals the formation of ettringite from monosulfoaluminate phases and at an age in which the rigidity of the material is high. It should be borne in mind that the formation of ettringite from these phases is an expansive process, due to the difference in volume between the two phases. This could become a durability problem for materials containing this pigment.

Regarding portlandite, its contents are observed in the early ages and are much higher than those found in the sample with UB0 pigment, which shows the different reactivity of both pigments. Finally, regarding the pigment content, Figure 11b shows only a slight variation of the same. It should be noted that the starting intensity of the pigment peak was already lower, which influences the subsequent detection process. At the end of the measurement process, it was practically impossible to detect the pigment peak.