The exposure of concrete to high temperatures induces severe microstructural changes and concrete loses its strength and durability  . Concrete approximately loses half of its strength at a temperature of 600˚C and loses most of its compressive strength above 800˚C  . The literature that deals with the effect of high temperatures on the properties of concrete can be divided into two categories. The first category investigates the degree of deterioration on the mechanical properties of concrete exposed to high temperatures and evaluates the fire damaged concrete by using nondestructive tools such as ultrasonic and resonance vibration techniques   . The influence of high temperatures on the mechanical properties of concrete depends on physico-chemical properties of the concrete constituents   , temperature of exposure  , size of the concrete structure  , external applied loadings  and cooling conditions  . The second category investigates the microstructural changes induced in concrete exposed to high temperatures.
The deterioration of concrete exposed to high temperatures is attributed to:
1) The thermal mismatch due to the expansion of siliceous aggregate and shrinkage of cement paste matrix  ,
2) The decomposition of hydrates due to the decomposition of portlandite and C-S-H  ,
3) The coarsening of pore structure due to void formation as a result of loss of bound water  ,
4) The pore pressure effects due to buildup of vapor pressure inside closed pores  and,
5) Cracking that develops as a result of the rehydration of lime  .
In literature, the following investigations were performed on cement paste after exposure to high temperatures.
1) Assessment of the structural properties of cement paste such as stiffness of cement paste by absorption capacity  , pore structure by backscattered electron microscopy/mercury intrusion porosimetry/gas permeability techniques  , and crack density by digital microscope  .
2) Evaluation of the microstructural changes by XRD, DSC/TGA/DTG/DTA and SEM/EDX techniques  -  .
3) Monitoring of the phase transformations by instrumental techniques    .
4) Estimation of the cementitious phases     .
5) Estimation of the dehydration mechanisms and kinetics for decomposition of portlandite and C-S-H  .
The major changes that occur within cement paste due to exposure to high temperatures include the decomposition of portlandite and calcium silicate hydrates  . There is a contradiction in the literature about the nature and types of the decomposition products of C-S-H gel which decomposes over a wide range temperature (about from 105˚C to 1000˚C  ) as a result of its amorphous nature. Zhang and Ye calculated the activation energy (Ea) of the dehydration of C-S-H and portlandite by Arrhenius equation from TGA/DTG results. Eaportlandite is about 151.82 kJ/mol whereas EaC-S-H varies from 83.69 to 371.93 kJ/mol. Accordingly, the dehydration of portlandite is a single-step reaction, whereas the dehydration of C-S-H is a multi-step reaction because of its complex structure  .
Many researchers examined the dehydration of C-S-H gel in cement paste exposed to high temperatures. According to Alonso and Fernandez, C-S-H gel dehydrates above 200˚C to nesosilicate phase with CaO/SiO2 ≈ 2, assimilated to a structure of C2S but with less crystalline structure and the dehydration was completed at 750˚C  . According to Stepkowska, C-S-H gel dehydrates in the temperature range of 110˚C - 450˚C and transforms to C2S and C3S above 600˚C  . According to Heikal, C-S-H gel dehydrates in the temperature range of 100˚C - 400˚C and transforms to C2S and C3S at 800˚C  . According to Peng, C-S-H gel started to decompose at 560˚C and transforms to C2S and C3S at 800˚C  .
According to Ghosh et al., the rehydration of dehydrated phases that take place in the fired cement pastes after cooling at moist air, that together with the changes in volume, and mass may lead to an additional increase in the porosity of cement paste and enhances the formation of additional cracks  . Many researchers reported that there are many factors that control the extent of deterioration caused by CaO rehydration such as the rate of water absorption  and the initial mineralogy of Portland cement, (i.e. primarily the amount alumina phases  ). The estimation of rehydrating portlandite can be used as tracers for determining the heating history of concrete exposed to high temperatures  . Many researchers investigated the effect of high temperatures and rehydration on the mechanical properties of blended cement pastes. According to Vysvaril et al., the cement paste with the addition of granulated blast furnace slag has the best resistivity to high temperatures; however, the limestone cement paste shows better mechanical properties after rehydration process  . In contrast, the rehydration process promotes the reformation of C-S-H gel from the nesosilicate phase as well as the recrystallization of calcite, portlandite and ettringite phases  . In the other hand, it was reported that the rehydration of the anhydrous phases is accompanied by partial recovery of the mechanical properties of the fire-exposed concrete structure  and  . The rehydration process displays cementitious behavior similar to that of the calcium silicates present in Portland cement, developing strength and thus potentially enabling the production of construction materials. According to Shui et al., the addition of fly ash and Portland cement to preheated recycled concrete can significantly enhance its rehydrated strength  . Serpell and Lopez developed reactivated cementitious materials from recycling of hydrated cement paste wastes as alternative binders of high environmental value  .
The aim of this contribution is to monitor the microstructure and phase changes of cement paste exposed to high temperatures up to 750˚C by XRD, FTIR, TGA/DTA and SEM techniques and to correlate the phase changes with the physico-mechanical properties of the thermally damaged cement paste. What differentiate this work from that of others is that the approximate percentage of residual calcium hydrates (C-S-H, ettringite and calcium aluminate hydrates) in cement pastes fired at a given temperature relative to that in unfired cement paste was estimated.
2. Materials and Experimental Techniques
Cement paste was prepared from OPC [CEM I 42.5] using water/cement ratio equals to 0.27. Cement paste was cast in 2 cm3 stainless steel cubic moulds at about 100% relative humidity for 24 hours. Cement paste was cured for 28 days under water to enhance hydration of the cement paste before being subjected to the fire experiment. The compressive strength of the cement pastes was measured using a manual compressive strength machine according to ASTM designation  .
The loss in compressive strength (L) was calculated according to  :
where Sf is the compressive strength after exposure to high temperature, and S is the compressive strength before exposure to high temperature.
Termination of the hydration of cement paste was performed using a domestic microwave oven  . The bulk density of unfired cement pastes was measured according to the Archimedes principle  . Bulk density was determined as the weight of the cement paste divided by its volume. Where the volume of cement paste is calculated from the weight difference between cement in air and suspended in water. Each measurement was carried out on three similar specimens of the same testing time. The total porosity of unfired cement pastes was measured as described elsewhere  . The total porosity was calculated from the values of bulk density as well as the free and total water contents of the cement paste. Cement pastes were dried at 105˚C for 24 hours and fired at 300˚C, 450˚C, 600˚C and 750˚C for 2 hours in a muffle furnace with a heating rate of 10˚C/min. The heating curve was illustrated Figure 1. The desired temperature was maintained for 2 hours. At the end of the heating period, the specimens were removed from the furnace and were cooled down to room temperature in a desiccator. The weight loss of fired cement pastes was measured as the percentage of the weight loss based on weight after firing. The bulk density and total porosity of fired cement pastes were measured according to the ISO 5018-1983  . XRF analysis was performed by Philips PW 1606 x-ray fluorescence spectrometer. XRD analysis was performed by Philips x-ray diffractometer PW 1370 with Ni filtered CuKα radiation (1.5406 Å). The semi-quantitative phase determination was estimated by measuring the areas of representative peaks, using X’Pert HighScore Plus software and JCPDS-ICDD data base. The standard deviation was ±5%. FTIR analysis was performed by spectrometer Perkin Elmer FTIR System Spectrum X in the range 400 - 4000 cm−1. SEM analysis was performed by Jeol-Dsm 5400 LG apparatus. TGA/DrTGA/DTA were performed by Shimadzu Corporation thermo analyzer with DTG-60H detector with 10˚C/min heating rate from the room temperature up to 1000˚C, under air
Figure 1. The muffle furnace and heating profile.
atmosphere at 40 ml/min flow rate, the hold time at the appropriate temperature is zero.
The percentage of portlandite (P) in the unfired and fired cement pastes was calculated for the following reaction (Ca(OH)2 → CaO + H2O) as  :
where; F is the mole ratio Ca(OH)2/H2O equal 4.11 and, Δwt is the weight loss in TGA curve due to the dehydration of portlandite.
It not possible to calculate the percentage of calcium hydrates (C-S-H, ettringite and calcium aluminate hydrates) by the above method because calcium hydrates has unknown structure and able to incorporate guest ions  . Alternatively, it was assumed that the relative proportions of calcium hydrate species (C-S-H, ettringite and calcium aluminate and aluminosilicate hydrates) are the same for a particular cement paste under the same condition. Hence, the rate of dehydration of calcium hydrates in the fired cement pastes will be constant. Accordingly, the approximate percentage of residual calcium hydrates in cement pastes fired at a given temperature relative to that in unfired cement paste can be estimated as follows:
Relative percentage of calcium hydrates (r) = (Δwtf/Δwt) × 100
Percentage of calcium hydrates decomposition = r − 100
where: Δwt is the weight loss in TGA curve corresponding to dehydration of calcium hydrates for unfired cement paste, and Δwtf is the weight loss corresponding to dehydration of calcium hydrates for cement paste fired at a given temperature.
3. Results and Discussion
Figure 2 illustrates the weight loss of cement pastes that were fired at 300˚C - 750˚C. Weight loss increases with temperature. Weight loss that occurs at 300˚C may be due to the evaporation of capillary water from macro capillary pores, evaporation of gel water from gel pores of cement paste and partial dehydration of C-S-H  . Weight loss that occurs at 450˚C may be due to the dehydration of portlandite as evidenced from DTA results (Figure 3). Weight loss that occurs at 600˚C - 750˚C is due to the dehydration of residual portlandite and C-S-H as well as decomposition of calcite  .
Figure 4 illustrates the bulk density and total porosity of unfired cement paste and that were fired at 300˚C - 750˚C. Bulk density sharply decreases at 300˚C then decreases very slowly at the higher temperatures. Total porosity decreases at 300˚C then increases at the higher temperatures. With rising temperature up to 300˚C, the vapor produced from the evaporation of capillary water and dehydration of calcium hydrates accumulates while the total porosity decreases. Hence, an internal vapor pressure builds up inside the low permeable cement paste. Under this hydrothermal condition, vapor pressure encourages the hydration of unhydrated cement phases and formation of additional hydration products as well as fills some of open pores of cement paste  . The decrease of bulk
Figure 2. Weight loss of cement pastes fired at 300˚C - 750˚C.
Figure 3. DTA thermograms of cement pastes fired at 300˚C - 750˚C.
Figure 4. Bulk density and total porosity of cement pastes fired at 300˚C - 750˚C.
density leads to the suggestion that the hydration products that was formed under this hydrothermal condition have low density compared to that formed at ambient temperatures. With rising temperature above 300˚C, the decomposition of C-S-H and portlandite opens the pore system of cement paste and increases the total porosity  .
Figure 5 illustrates the strength loss of cement paste fired at 300˚C - 750˚C. The loss in compressive strength may be considered as an important measure of
Figure 5. Strength loss of cement pastes fired at 300˚C - 750˚C.
the degree of damage to concrete exposed to high temperatures. The strengthening of unfired cement paste results from the coexistence of the amorphous and crystalline C-S-H that binds particles into cohesive mass  . With rising temperature, a negligible and limited strength loss was observed at 300˚C and 450˚C then the strength loss markedly increases at the higher temperatures. This indicates that beyond the critical temperature of 450˚C the decomposition of C-S-H markedly progresses  .
Figure 6 illustrates the FTIR spectra of cement paste fired at 300˚C - 750˚C. In the case of OPC, the wide band at around 923 - 930 cm−1, which is attributed to the ν3 (Si-O) asymmetric stretching vibrations of alite was shifted to higher wavenumbers (around 978 cm−1) as cement hydration progressed. In the case of the unfired cement sample, the formation of C-S-H gel was accompanied by a gradual weakening of the ν4 (O-Si-O) absorption band at 520 cm−1 which is attributed both to unhydrated alite and belite. The bands at 1144 and 1107 cm−1 which are attributed to ν3 asymmetric stretching vibrations of the sulfate group ( ) in gypsum were replaced with a new band at around 1115 cm−1 that attribute to ν3 vibrations of in ettringite  . In the case of fired cement pastes, the FTIR results can be utilized to monitor the phase changes due to exposure to high temperatures. Table 1 summarizes the relative intensity of the characteristic absorption bands given in Figure 6. C-S-H progressively decreases due to its decomposition above 450˚C. This agrees with the results of strength loss. C3S and C2S increase with rising temperature as a result of the decomposition of C-S-H. Portlandite decreases due to its dehydration around 450˚C. Traces of portlandite were detected at 600˚C - 750˚C due to the rehydration of lime  . Calcite decomposes at 750˚C.
Figure 7 illustrates the TGA thermograms of cement paste fired at 300˚C - 750˚C. The first weight loss in the temperature range 25˚C - 320˚C is attributed to the dehydration of C-S-H, ettringite and calcium aluminate hydrates  . The temperature at which these compounds lose water depends upon the CaO/SiO2 ratio in the hydrated cement matrix  . The second and third weight loss in the temperature ranges 350˚C - 550˚C and 650˚C - 800˚C are attributed to the dehydration of portlandite and decarbonation of calcite respectively  . A significant amount of calcite was detected in all of the hydrated samples is attributed to the limestone filler commonly added to OPC  . The relative
Figure 6. FTIR spectra of cement pastes fired at 300˚C - 750˚C. (where: A alite, B belite, C calcite, E ettringite, G gypsum, P portlandite, W water), (vs very strong, s strong, m medium, w weak, - not found).
Figure 7. TGA thermograms of cement pastes fired at 300˚C - 750˚C.
percentage of residual portlandite and calcium hydrates in cement pastes fired at 300˚C - 750˚C are given in Table 2. These results give a crucial picture about the decomposition of amorphous calcium silicate hydrates which can not be detected by XRD analysis.
The percentage of portlandite that was estimated for the unfired cement paste (28.49%) nearly agrees with that typically reported in the literature (i.e. portlandite occupies 20% - 25% of the hydrated cement paste)  . With rising temperature up to 300˚C, the percentage of portlandite slightly increases due to hydration of the OPC grains which was encouraged under the hydrothermal condition. Above 300˚C, the percentage of portlandite slightly diminishes due to dehydration of portlandite to lime. Whereas above 450˚C, the percentage of
Table 1. Relative intensity of characteristic absorption bands given in Figure 6.
Table 2. The relative percentages of residual calcium hydrates and portlandite in cement pastes fired at 300˚C - 750˚C.
portlandite sharply diminishes but does not fall to zero value in cement pastes that were fired at 600˚C - 750˚C due to the partial rehydration of lime. The percentage of calcium hydrates was lowered to its half value with rising temperature up to 450˚C. A severe and complete loss of calcium hydrates content occurs when cement paste was subjected to 600˚C and 750˚C respectively. This confirms that the dehydration of calcium hydrates is a multi-step reaction because of its complex structure  and occurs over a wide range temperature (i.e. from 105˚C to 1000˚C)  and  . According to Jennings, the C-S-H gel consists of many small globules with a disordered layered structure, hence, water molecules release over a wide range temperature  .
Figure 3 illustrates the DTA thermograms of cement paste fired at 300˚C - 750˚C. DTA results are useful to justify the dehydration process of portlandite. The endothermic peak located at about 450˚C in the case of unfired cement paste and that was fired at 300˚C - 450˚C corresponds to the dehydration of the primary portlandite that forms during the hydration of cement in the case of unfired cement paste and remaining portlandite that present in cement paste fired up to 450˚C.
The endothermic peak of portlandite was shifted to lower temperature range (about 410˚C) in the case of cement pastes fired at 600˚C - 750˚C corresponds to the dehydration of the secondary portlandite that forms as a result of the rehydration of lime in the case of cement pastes fired above 450˚C. This result agrees with the previous results which illustrate that portlandite formed during the rehydration of fired cement paste has an onset dehydration temperature lower than the original portlandite  . The environments and formation condition of the primary and secondary portlandite is quite different, accordingly, their rate of dehydration are not equal. The secondary portlandite dehydrates at lower temperatures range may be due to the critical condition in which it forms which affect its surface area and particle size. The experimental results confirm that the decomposition of hydrates is affected by surface area and particle size  .
Figure 8 illustrates the XRD patterns of cement pastes fired at 300˚C - 750˚C. The dominant crystalline phases that were detected in OPC are; alite, belite and calcite. The dominant crystalline phases that were detected in the unfired cement paste are calcite filler and portlandite arising from the hydration process as well as small proportions of the unhydrated alite and belite. XRD patterns comprise some changes in the phase compositions of cement paste with rising temperature. The proportion of portlandite diminishes above 450˚C  . A significant proportion of belite and alite appears at 450˚C and progressively increases at 600˚C and 750˚C. This proves that the decomposition of C-S-H reaches a significant rate at 450˚C and enhances with rising temperature  . This result agrees with FTIR and DTA results. According to Peng et al., it is difficult to differentiate between the peaks of C2S and C3S in XRD diagrams of cement pastes exposed to high temperatures. Hence, peaks of C2S and C3S can be used to
Figure 8. XRD patterns of cement pastes fired at 300˚C - 750˚C. (A alite, B belite, C calcite and P portlandite).
represent both of them  . Figure 9 illustrates the SEM micrographs of cement paste fired at 300˚C - 750˚C. The results monitor the change in microstructure and phase compositions due to the hydration of OPC and the dehydration of cement paste fired at 300˚C - 750˚C. The micrograph of unfired cement shows OPC grains loaded with the hexagonal portlandite plates  and amorphous C-S-H formed during hydration of OPC. The microstructure of cement paste that was fired at 300˚C has not changed significantly. Whereas, some of unhydrated cement grains and residues of hexagonal portlandite plates appear in the microstructure of cement paste that was fired at 450˚C. The micrograph of cement paste that was fired at 600˚C shows the presence of distorted cubic calcite crystals  . The micrograph of cement paste that was fired at 750˚C may show the presence of dehydration products. Accordingly, the decomposition of C-S-H significantly starts at 450˚C and progresses with rising temperature.
Figure 10 illustrates the crack propagation in cement paste fired at 300˚C - 750˚C. Minor cracks (about 0.1 mm in width) appeared on the surface of cement paste that was fired at 300˚C. The cracks widen and deepen with rising
Figure 9. SEM micrographs of cement pastes fired at 300˚C - 750˚C.
Figure 10. Crack propagation in cement paste fired at 300˚C - 750˚C.
temperature. Thin shallow linear superficial cracks (about 0.4 mm in width) appeared on the surface of cement paste that was fired at 450˚C. Thick grid-like deep cracks (about 0.8 mm in width) appeared on the surface of cement paste that was fired at 750˚C. The cracks form as a result of the following factors; 1) the internal pore pressure caused by water vapor that was accumulated after the dehydration of calcium hydrates in case of relatively low permeable cement paste accompanied by loss of mechanical strength, 2) the thermal shock that arises as a result of cooling of the heated cement paste, 3) the rehydration of lime that is accompanied by a significant volume increase  .
FTIR, XRD, TGA/DTA and SEM techniques were used to monitor the microstructural and compositional changes within cement paste induced with rising temperature. Accordingly, the main conclusions are:
1) The microstructure of cement paste that was fired at 300˚C has not changed significantly.
2) A negligible and limited strength loss was observed before 450˚C attributed to the dehydration of C-S-H, ettringite and calcium aluminate hydrates.
3) 450˚C is a critical temperature in the thermal damage history of cement paste, around which the percentage of portlandite decreases due to its dehydration and the percentage of calcium hydrates was lowered to its half value.
4) Beyond 450˚C, the percentage of portlandite sharply diminishes and decomposition of C-S-H into C2S and C3S markedly progresses until complete loss of calcium hydrates content occurs at 750˚C.
5) These deleterious changes in the composition of cement paste are accompanied by an increase of the total porosity, severe loss in mechanical strength and propagation of harmful cracks.
6) The thermal shock that arises as a result of cooling of the heated cement paste and the rehydration of lime that is accompanied by a significant volume increase enhance the propagation of harmful cracks.