Deterioration Process of Concrete Exposed to Internal Sulfate Attack

materials

Article Deterioration Process of Concrete Exposed to Internal Sulfate Attack

Weifeng Chen 1,2, Bei Huang 1,2,*, Yuexue Yuan 1,2 and Min Deng 1,2

1 College of Mateials Science and Engineering, Nanjing Tech University, Nanjing 211800, China; [email protected] (W.C.); [email protected] (Y.Y.); [email protected] (M.D.) 2 State Key Laboratory of Materials-Oriented Chemical Engineering, Nanjing 211800, China * Correspondence: [email protected]; Tel.: +86-138-5160-2803

Received: 10 February 2020; Accepted: 13 March 2020; Published: 15 March 2020

Abstract: Damage to concrete structures with gypsum-contaminated aggregate occurs frequently. Aggregates in much of the southern part of China are contaminated with gypsum. Therefore, in this study, the effects of using different quantities of gypsum-contaminated aggregate on the expansion and compressive strength of concrete were investigated over a period of one year. Two groups of concrete were designed with the gypsum-contaminated aggregate containing different parts of fine and coarse aggregate, respectively. The SO3 contents were 0%, 0.5%, 1%, 1.5%, 3%, 5%, and 7% by weight of aggregate. X-ray diffraction (XRD), thermogravimetry (TG), and differential scanning calorimetry (DSC) were used to analyze the change in mineral composition over time. The microstructure was also studied by scanning electron microscopy (SEM) and energy dispersive spectrometry (EDS). The results showed that significant expansion and great loss in compressive strength did not occur in concrete if the content of SO3 lay below 1.5% and 3% in fine and coarse aggregates, respectively. The concentration of sulfate ions in concrete was not enough to form new a phase of gypsum. During the process of internal sulfate attack, the content of gypsum decreased and the content of ettringite increased. Ettringite was the main reason for the expansion damage of concrete. Additionally, the fracture mode of internal sulfate attack on concrete was the crack extension from gypsum to paste; finally, the aggregate separated from the paste.

Keywords: internal sulfate attack; concrete; gypsum; performance

1. Introduction Sulfate attack on concrete is a key durability problem. Internal sulfate attack (ISA) on concrete is a reaction between sulfates present in the original mix of concrete and the calcium aluminate in cement and water, which can show signs of deterioration [1]. In the early stage of hydration, the gypsum reacts with tricalcium aluminate to form ettringite (AFt), which does not damage concrete. It is beneﬁcial and leads to an increase in strength. However, the excess gypsum remains in the hardened concrete, 2 and the gypsum will dissolve sulfate ions. The SO4 − continues to react with the remaining tricalcium aluminate to form AFt, which is harmful to concrete [2]. The harmful AFt may cause a deleterious expansion and strength loss, since large stresses are generated in the hardened cement paste [3]. In the past 30 years, ISA damage due to aggregates with sulfate salts has been found in tunnels and dams in China [4], Middle Eastern countries [5], and some European countries [6]. A large number of engineering disasters have occurred because of the aggregate contained in sulfates. As a result, great economic losses as well as safety problems have arisen. The mechanism of external sulfate attack (ESA) has been studied for many years. Sulfate is supplied by seawater, groundwater, and industrial wastewater. Sulfate ions react with ionic species of the pore solution to precipitate ettringite, gypsum, or thaumasite [7]. The major reaction product of ESA

Materials 2020, 13, 1336; doi:10.3390/ma13061336 www.mdpi.com/journal/materials Materials 2020, 13, 1336 2 of 13 is AFt. The formation of AFt results in expansion and cracking, spalling, and other damaging effects. There is a correlation between the C3A content in Portland cement and the susceptibility of concrete to sulfate attack, and cements with limited C3A content were introduced to the market. These cements are used in the concrete that is exposed to moderate and severe sulfate concentrations [8,9]. The sulfate ions react with portlandite to form gypsum. Even though gypsum formation is generally accepted to be harmful, the specific mechanism by which this occurs is not well established. It is widely accepted that gypsum formation only has a softening effect and causes mass and strength loss. However, Tian demonstrated that gypsum formation during sulfate attack may cause expansion [10]. The development of gypsum is influenced by the pH value and the concentration of sulfate ions. In sulfate solution, the minimum concentration of sulfate ions to form gypsum is 1400 mg/L, at a pH value equal to 12.4 [11]. Normally, gypsum is the first reaction product at a high sulfate ion concentration, especially 2 SO4 − > 8000 mg/L[12]. Thaumasite is usually considered to be formed at a low temperature in the presence of carbonate [13]. The thaumasite form of sulfate attack is different from ettringite formation because it is the C-S-H in hardened Portland cement that is targeted for the reaction and not the calcium aluminate hydrates. However, C-S-H is the primary binding agent in concrete [14]. Compared with the sulfate ions that penetrate into concrete by diffusion for ESA, there is a higher concentration of sulfate ions in concrete for ISA. The hazard is illustrated in the materials which cause higher tensile stress, leading to spalling and disruption of concrete structures [15]. For example, several highway tunnels in China were damaged seriously by ISA in the year after they were built. In order to prevent ISA, Chinese national standards have set the upper limit for the SO3 content in fine and coarse aggregates. It is stipulated in GB/T14684 and GB/T14685 that the SO3 content in fine and coarse aggregates shall not exceed 0.5% and 1%, respectively. Samarai proposed a 0.1% expansion as a safe margin for finding the allowable sulfate level [16]. According to Crammond, the allowable gypsum content in aggregates used in Portland cement mortars would not be much larger than 2.5% by weight of aggregate [17]. Because of the higher concentration of sulfate ions, the types of reaction products and the degradation process of ISA are important problems to be studied. In this paper, the expansion, strength loss, and formation process of reaction products of concrete with different amounts and particle sizes of gypsum-contaminated aggregate were investigated, and we aimed to discuss the feasibility of using gypseous rock as a concrete aggregate for highways in Shanxi province.

2. Research Signiﬁcance In this paper, the inﬂuences of the dosage of gypsum aggregate on the compressive strength and expansion performance of concrete were studied. The types of reaction products in the process of internal sulfate attack were also studied.

3. Materials and Mixes

3.1. Raw Materials P II 52.5 Portland cement conforming to Chinese National Standard GB175 was used in this · study and its chemical composition is shown in Table1. Two types of aggregate were used. One was gypseous aggregate, which was crushed into ﬁne and coarse aggregate, from the WU YAN HE railway tunnel from Shanxi province to Shandong province. Another was normal aggregate with no gypsum; river sand was used as ﬁne aggregate and limestone was used as coarse aggregate. The appearance of gypsum-contaminated aggregate is shown in Figure1. The chemical compositions and XRD patterns of this aggregate are shown in Table2 and Figure2, respectively. The chemical compositions of white part (WP) are shown in Table3. From analysis results, it was found that the WP of this rock was a mix of gypsum and anhydrite, and the rock matrix was dolomite. Materials 2020, 13, x FOR PEER REVIEW 3 of 15 compositions of white part (WP) are shown in Table 3. From analysis results, it was found that the WP of this rock was a mix of gypsum and anhydrite, and the rock matrix was dolomite.

Materials 2020, 13, x FOR PEER REVIEW 3 of 15

Materialscompositions2020, 13 ,of 1336 white part (WP) are shown in Table 3. From analysis results, it was found that3 ofthe 13 WP of this rock was a mix of gypsum and anhydrite, and the rock matrix was dolomite.

Figure 1. Appearance of gypsum-contaminated aggregate.

Table 1. Chemical compositions of P·Ⅱ 52.5 Portland cement.

SiO2 CaO Figure 1. AppearanceAl2O3 of gypsum-contaminatedFe2O3 MgO aggregate. Na2O SO3 19.46 63.23 Figure 1. Appearance4.53 of gypsum2.80 -contaminated2.41 aggregate . 0.13 3.31 Table 1. Chemical compositions of P II 52.5 Portland cement. · Table 1. Chemical compositions of P·Ⅱ 52.5 Portland cement. SiO2 CaO Al2O3 Fe2O3 MgO Na2O SO3 Table 2. Chemical compositions of gypsum-contaminated aggregate. SiO2 19.46CaO 63.23Al2O 4.533 Fe 2.802O3 2.41MgO 0.13Na2O 3.31 SO3 SiO19.462 CaO 63.23 Al2O3 4.53Fe 2O3 MgO2.80 K2O 2.41 Na2O 0.13SO 3 3.31LOI 15.83 17.21 Table 2.6.71Chemical compositions4.64 11.59 of gypsum-contaminated 2.87 0.26 aggregate. 18.03 21.08

SiO2 TableCaO 2. Chemical Al2O3 compositionsFe2O3 MgO of gypsum K-contaminated2O Na2O aggregate SO3 . LOI 15.83Table 17.21 3. Chemical 6.71 compositions 4.64 11.59of the white 2.87 part (WP 0.26) in aggregate 18.03. 21.08 SiO2 CaO Al2O3 Fe2O3 MgO K2O Na2O SO3 LOI

SiO2 CaO Al2O3 MgO Fe2O3 SO3 K2O Na2O TiO2 LOI 15.83 17.21 Table 3.6.71Chemical compositions4.64 of11.59 the white2.87 part (WP) in0.26 aggregate. 18.03 21.08 0.19 35.43 0.08 4.05 0.02 51.36 0.01 0.01 0.01 12.79 SiO2 CaO Al2O3 MgO Fe2O3 SO3 K2O Na2O TiO2 LOI Table 3. Chemical compositions of the white part (WP) in aggregate. 0.19 35.43 0.08 4.05 0.02 51.36 0.01 0.01 0.01 12.79

SiO2 CaO Al2O3 MgO Fe2O3 SO3 K2O Na2O TiO2 LOI 0.19 35.43 0.08 4.05 0.02 51.36 0.01 0.01 0.01 12.79

Figure 2. XRD patterns of gypsum-contaminated aggregates. Figure 2. XRD patterns of gypsum-contaminated aggregates. 3.2. Mix Proportion In our experimental program, there were two types of concrete specimen: 100 mm by 100 mm

by 100 mm and 75 mm by 75 mm by 280 mm. They were demolded a day after casting and cured in water at 20 2 C. The cube samples were used to determine the compressive strength of the concrete ± ◦ Figure 2. XRD patterns of gypsum-contaminated aggregates. with diﬀerent amounts and sizes of gypsum. The prism specimens were used for the ultrasonic and expansion test of concrete at diﬀerent ages. Materials 2020, 13, x FOR PEER REVIEW 4 of 15

3.2. Mix Proportion In our experimental program, there were two types of concrete specimen: 100 mm by 100 mm by 100 mm and 75 mm by 75 mm by 280 mm. They were demolded a day after casting and cured in water at 20 ± 2 °C. The cube samples were used to determine the compressive strength of the concrete Materialswith different2020, 13, 1336amounts and sizes of gypsum. The prism specimens were used for the ultrasonic4 ofand 13 expansion test of concrete at different ages. The gypsum-contaminated aggregate was crushed into gypsum sand (<4.75 mm) and gypsum stoneThe (5 gypsum-contaminated–20 mm). Part of the aggregate normal fine was and crushed coarse into aggregate gypsum sand was (< replaced4.75 mm) with and gypsum- stone (5–20 mm). Part of the normal fine and coarse aggregate was replaced with gypsum-contaminated contaminated aggregate in the mixture of concrete, to get five levels of SO3 content (0, 0.5%, 1.0%, aggregate1.5%, 3.0%, in 5.0%, the mixture and 7% of by concrete, weight toof getaggregate) five levels in ofthe SO concrete.3 content The (0, 0.5%,grain 1.0%,size curves 1.5%, 3.0%,of fine 5.0%, and andcoarse 7% aggregates by weight of are aggregate) shown in in Figure the concrete. 3. The mix The design grain sizeof concretes curves of is fine shown and coarsein Table aggregates 4. are shown in Figure3. The mix design of concretes is shown in Table4.

(a) (b)

Figure 3. The grain size curve of aggregates: (a) ﬁne; (b) coarse. Figure 3. The grain size curve of aggregates: (a) fine; (b) coarse. Table 4. Mix design of concretes (kg/m3). Table 4. Mix design of concretes (kg/m3). SO by Weight Fine Aggregate Coarse Aggregate No. 3 Cement Water of Aggregate/% River SandFine GypsumAggregate Sand LimestoneCoarse Gypsum Aggregate Stone R0 0 420 189 735 0 1110 0 G0.5SO3 by 0.5 Weight 1058 52 No.G1 1Cement Water Gypsum 1006Limestone 104 G1.5of Aggregate/% 1.5 River 954 156Gypsum 420 189 735 0 G3 3Sand 798 312 Stone G5 5Sand 590 520 G7 7 382 728 FG0.5 0.5 683 52 FG1R0 10 420 189 631735 104 0 1110 0 FG1.5 1.5 579 156 420 189 1110 0 FG3 3 423 312 G0.5FG5 50.5 215 520 1058 52 FG7 7 7 728

4. MethodsG1 1 1006 104

4.1.G1.5 Expansion Test1.5 on Concrete 420 189 735 0 954 156 An expansion test was conducted using three specimens per mix. The reference points used to measureG3 the dimensional3 variations of samples were placed on the 75 mm by 280 mm798 sides that were312 in contact with the lateral sides of the molds. As shown in Figure4, two reference points were placed on each side with 250 mm separation. According to Equation (1), the expansion of a specimen at a certain G5 5 590 520 age was calculated by the average of the measurements performed at both sides with a handholding strain gauge with 0.01% precision. In the formula, L0, L, and Lt represent the initial length, eﬀective length of 250 mm, and testing length at diﬀerent ages, respectively.

Lt L P = − 0 100% (1) L × Materials 2020, 13, x FOR PEER REVIEW 6 of 15

Materials 2020, 13, x FOR PEER REVIEW 6 of 15 4. Methods 4. Methods 4.1. Expansion Test on Concrete 4.1. ExpansionAn expansion Test on test Concrete was conducted using three specimens per mix. The reference points used to measureAn e thexpansion dimensional test was variations conducted of samplesusing three were specimens placed on per the mix.75 mm The by reference 280 mm sidespoints that used were to measurein contact the with dimensional the lateral variations sides of the of molds. samples As were shown placed in Fig onure the 4 ,75 two mm reference by 280 mm points sides were that placed were inon contact each side with with the 250lateral mm sides separation. of the molds. According As shown to Equation in Figure (1), 4, twothe expansionreference points of a specimen were placed at a oncertain each ageside waswith calculated 250 mm separation. by the average According of the to measurements Equation (1), the performed expansion at of both a specimen sides with at a a certainhandholding age was strain calculated gauge with by the 0.01% average precision. of the In measurements the formula, L performed0, L, and L att represent both sides the with initial a length, effective length of 250 mm, and testing length at different ages, respectively. handholding strain gauge with 0.01% precision. In the formula, L0, L, and Lt represent the initial length, effective length of 250 mm, and testing length at different ages, respectively. L − L P = t 0 100% (1) L −LL P = t 0 100% (1) Materials 2020, 13, 1336 L 5 of 13

Figure 4. Expansion test on concrete. Figure 4. Expansion test on concrete. Figure 4. Expansion test on concrete. 4.2.4.2. Compressive Strength ofof ConcreteConcrete 4.2. CompressiveThe compressive Strength strength of Concrete of the concrete cubic specimen was examined at the ages of 28, 60, 90, The compressive strength of the concrete cubic specimen was examined at the ages of 28, 60, 90, 150, 210, and 240 days by using a compressive-testing machine with a loading rate 0.5–0.8 MPa/s. The 150, 210,The compressive and 240 days strength by using of athe compressive-testing concrete cubic specimen machine was withexamined a loading at the rate age 0.5–0.8s of 28, MPa60, 90,/s. machine for testing compressive strength is shown in Figure 5. For each test, the average compressive The150, 210, machine and 240 for days testing by using compressive a compressive strength-testing is shown machine in Figurewith a5 loading. For each rate 0.5 test,–0.8 the MPa/s. average The strength of the three cubic concrete samples was used. compressivemachine for testing strength compre of thessive three strength cubic concrete is shown samples in Figure was 5. used.For each test, the average compressive strength of the three cubic concrete samples was used.

. Figure 5. Compressive strength test on cubic concrete.. Materials 2020, 13, x FOR PEER REVIEW 7 of 15 4.3. Ultrasonic Test on PrismFigure Concrete 5. Compressive strength test on cubic concrete. The ultrasonic velocities, which describe the material internal damage, in specimens at different The ultrasonic velocities,Figure which 5. Compressive describe the strength material test internalon cubic concrete. damage, in specimens at different ages4.3. Ultrasonic were tested Test, by on fixing Prism the Concrete launcher and receiver transducer firmly on the two sides of the specimen ages were tested, by fixing the launcher and receiver transducer firmly on the two sides of the specimen (754.3. mmUltrasonic by 75 Testmm) on, in Prism accordance Concrete with the “Technical specification for inspection of concrete defects (75 mm by 75 mm), in accordance with the “Technical specification for inspection of concrete defects by by ultrasonic method” (CECS 21: 2000). The machine used to test the ultrasonic pulse velocity is shown ultrasonic method” (CECS 21: 2000). The machine used to test the ultrasonic pulse velocity is shown in Figure 6. in Figure6.

Figure 6. Ultrasonic pulse velocity test on prism concrete. Figure 6. Ultrasonic pulse velocity test on prism concrete.

4.4. XRD and Thermogravimetry–Differential Scanning Calorimetry (TG–DSC) Analysis on Mortar To get the complementary mineral information, mortar specimens were subjected to XRD and TG– DSC analyses at different ages. The mortars cut from the concrete samples were immersed in absolute ethyl alcohol for 12 h to terminate hydration, before being vacuum dried at 60 °C for 24 h, and then ground into powder. The test powders were passed through an 0.08 mm sieve. The XRD data were collected in the range of 10°–70°, 2θ at a counting time of 15 s/step, and a divergence slit of 1°. TG–DSC data were obtained from 50 to 950 °C at a rate of temperature increase of 10 °C/min in a N2 atmosphere.

4.5. Microstructure Examination on Concrete In order to investigate the morphology of concrete under internal sulfate attack, a fractured fresh piece of concrete was sampled, hydration terminated, and vacuum dried. Then, the dried sample was embedded in a low-modulus epoxy, polished using progressively smaller grids, coated with a gold- palladium coating, and investigated using JSM-7600F SEM.

4.6. Dissolution of WP in Concrete The pore solution in concrete was expressed directly with a compressive-testing machine whose final pressure value was 2000 MPa at a speed of 2 MPa/s. Then, the sulfate concentration in this pore solution was tested by inductive coupled plasma emission spectrometer (ICP).

5. Results

5.1. Expansion Measurements Figure 7a presents the expansion data for the concretes with different amounts of gypsum- contaminated fine aggregate. Figure 7b shows the change in length of concrete with different amounts of contaminated coarse aggregate. The results for the gypsum-contaminated fine aggregate samples show that the mixes with a gypsum content above 3% SO3 presented a significant increase in expansion for 360 days. There was a little expansion when the SO3 content in sand was 1.5%, and the expansions of the specimen was less than 0.1% in 360 days. Meanwhile, specimens containing SO3 less than 1.5% showed no expansion.

Materials 2020, 13, 1336 6 of 13

4.4. XRD and Thermogravimetry–Diﬀerential Scanning Calorimetry (TG–DSC) Analysis on Mortar To get the complementary mineral information, mortar specimens were subjected to XRD and TG–DSC analyses at diﬀerent ages. The mortars cut from the concrete samples were immersed in absolute ethyl alcohol for 12 h to terminate hydration, before being vacuum dried at 60 ◦C for 24 h, and then ground into powder. The test powders were passed through an 0.08 mm sieve. The XRD data were collected in the range of 10◦–70◦, 2θ at a counting time of 15 s/step, and a divergence slit of 1◦. TG–DSC data were obtained from 50 to 950 ◦C at a rate of temperature increase of 10 ◦C/min in a N2 atmosphere.

4.5. Microstructure Examination on Concrete In order to investigate the morphology of concrete under internal sulfate attack, a fractured fresh piece of concrete was sampled, hydration terminated, and vacuum dried. Then, the dried sample was embedded in a low-modulus epoxy, polished using progressively smaller grids, coated with a gold-palladium coating, and investigated using JSM-7600F SEM.

Materials4.6. Dissolution 2020, 13, x FOR of WP PEER in ConcreteREVIEW 8 of 15

ForThe the pore samples solution with in concrete gypsum was-contaminated expressed directly coarse with aggregate, a compressive-testing no significant machine variation whose was observedﬁnal pressure, when value the SO was3 content 2000 MPa in coarse at a speed aggregate of 2 MPa was/s. less Then, than the 3%. sulfate The average concentration expansion in this increased pore obviouslysolution wasas the tested SO3 content by inductive increased coupled to 5% plasma and 7%. emission The final spectrometer measured (ICP). expansions for the specimens containing 5% and 7% SO3 were 0.1401% and 0.3602%, respectively, at 360 days. The expansion data in 5. Results Figure 7a indicated that the smaller aggregate particle size resulted in higher expansion of the specimen, at5.1. the Expansionsame SO3 Measurementscontent.

SeveralFigure 7studiesa presents recommended the expansion 0.1% expansion data for as the a safe concretes margin with for determining di fferent amountsthe maximum of percentagegypsum-contaminated of sulphate than fine aggregate.can be present Figure in7b a showsmixture the without change incausing length any of concrete significant with degradation di fferent [2amounts]. Based on of contaminatedthe results show coarsen in aggregate.Figure 7, FG1.5 The results, FG1, FG0.5 for the, G3 gypsum-contaminated, G1.5, G1, and G0.5 finecomply aggregate with the maximumsamples showexpansion that the suggested mixes with in athe gypsum literature. content Consequently, above 3% SO these3 presented mixtures a significant should not increase show a in risk expansion for 360 days. There was a little expansion when the SO content in sand was 1.5%, and the of damage from a structural point of view, but these SO3 contents3 are bigger than the limit established inexpansions several standards, of the specimen such as wasGB/T14684 less than an 0.1% GB/T14685 in 360 days. in China. Meanwhile, specimens containing SO3 less than 1.5% showed no expansion.

(a) (b)

Figure 7. Expansion curves of concrete with gypsum-contaminated aggregate: (a) gypsum- Figure 7. Expansion curves of concrete with gypsum-contaminated aggregate: (a) gypsum- contaminated fine aggregate; (b) gypsum-contaminated coarse aggregate. contaminated fine aggregate; (b) gypsum-contaminated coarse aggregate. For the samples with gypsum-contaminated coarse aggregate, no significant variation was 5.2.observed, Compressive when Strength the SO3 content in coarse aggregate was less than 3%. The average expansion increased obviouslyFigure as8 theshow SOs3 thecontent compressive increased tostrength 5% and 7%.development The final measured for concrete expansions with various for the specimens amounts of gypsum-contaminated aggregate. The results indicate that all concrete specimens exhibited a reduction in initial compressive strength as SO3 increased. It was evident that with 0.5%–1.5% SO3 in sand, the initial compressive strength of concrete decreased by 25%. Additionally, 3%–7% SO3 in sand led to a loss in the initial compressive strength of samples of about 38%. The compressive strength of all specimens with gypsum-contaminated aggregate increased first and then decreased with the curing age. As shown in Figure 8a, the final strengths of the FG0.5, FG1.0, and FG1.5 specimens at 360 days were higher than the initial strength of the reference sample, which means that an SO3 content in sand of less than 1.5% is relatively safe.

For the samples with SO3 in coarse aggregate, only the strength of the G7 specimen increased— from 30.2 to 33.3 MPa in 28 days—before reducing to 14.2 MPa at 360 days because of the expansion properties shown in Figure 7. The strength curve of the G5 sample showed a significant decrease after 210 days. It was indicated that samples with gypsum incorporated (0.5%–3% SO3) had less effect on the compressive strength of concrete relative to the initial strength of the reference.

Materials 2020, 13, 1336 7 of 13

containing 5% and 7% SO3 were 0.1401% and 0.3602%, respectively, at 360 days. The expansion data in Figure7a indicated that the smaller aggregate particle size resulted in higher expansion of the specimen, at the same SO3 content. Several studies recommended 0.1% expansion as a safe margin for determining the maximum percentage of sulphate than can be present in a mixture without causing any signiﬁcant degradation [2]. Based on the results shown in Figure7, FG1.5, FG1, FG0.5, G3, G1.5, G1, and G0.5 comply with the maximum expansion suggested in the literature. Consequently, these mixtures should not show a risk of damage from a structural point of view, but these SO3 contents are bigger than the limit established in several standards, such as GB/T14684 an GB/T14685 in China.

5.2. Compressive Strength Figure8 shows the compressive strength development for concrete with various amounts of gypsum-contaminated aggregate. The results indicate that all concrete specimens exhibited a reduction in initial compressive strength as SO3 increased. It was evident that with 0.5–1.5% SO3 in sand, the initial compressive strength of concrete decreased by 25%. Additionally, 3–7% SO3 in sand led to a loss in the initial compressive strength of samples of about 38%. The compressive strength of all specimens with gypsum-contaminated aggregate increased first and then decreased with the curing age. As shown in Figure8a, the final strengths of the FG0.5, FG1.0, and FG1.5 specimens at 360 days were higher than the initial strength of the reference sample, which means that an SO3 content in sand of less than 1.5% is relatively safe. For the samples with SO3 in coarse aggregate, only the strength of the G7 specimen increased—from 30.2 to 33.3 MPa in 28 days—before reducing to 14.2 MPa at 360 days because of the expansion properties shown in Figure7. The strength curve of the G5 sample showed a significant decrease after 210 days. MaterialsIt was 2020 indicated, 13, x FOR that PEER samples REVIEW with gypsum incorporated (0.5–3% SO3) had less effect on the compressive9 of 15 strength of concrete relative to the initial strength of the reference.

(a) (b)

Figure 8. Compressive strength of concrete with gypsum-contaminated aggregate: (a) gypsum- Figurecontaminated 8. Compressive ﬁne aggregate; strength (b) gypsum-contaminated of concrete with gypsum coarse-contaminated aggregate. aggregate: (a) gypsum- contaminated fine aggregate; (b) gypsum-contaminated coarse aggregate. 5.3. Ultrasonic Test 5.3. UltrasonicThe internal Test damage of specimens under internal sulfate attack was tested by ultrasonic pulse velocity,The internal and the damage results are of presentedspecimens in under Figure 9internal. The value sulfate of ultrasonic attack was velocity tested is by closely ultrasonic related pulse to velocity,the microstructure. and the result Thes are more presented compact in a structureFigure 9. is, The the value higher of the ultrasonic ultrasonic velocity velocity is obtained closely related will be. to theThe microstructure. results show thatThe themore change compact trend a forstructure the ultrasonic is, the higher velocities the of ultrasonic samples isvelocity consistent obtained with that will of be. Thecompressive results show strength. that the Except change for trend theFG3, for the FG5, ultrasonic FG7, G5, velocities and G7 specimens, of samples the is consisten damage tot with the other that of compressivesamples was strength. small. Except for the FG3, FG5, FG7, G5, and G7 specimens, the damage to the other samples was small.

(a) (b)

Figure 9. Ultrasonic test results of concrete samples: (a) gypsum-contaminated fine aggregate; (b) gypsum-contaminated coarse aggregate.

5.4. Mineral Composition Examination To detect the existence of gypsum in mortar during the internal sulfate attack, while avoiding the effects of quartz and dolomite peaks in the XRD, the XRD results from 5° to 20° are shown in Figures 10 and 11, for the most intensive peak of gypsum within 20° in the XRD patterns.

Figure 10 shows a comparison of the X-ray diffractograms of concrete samples with different amounts of SO3 in sand at 56, 240, and 360 days. The peaks of ettringite (E), portlandite (P), and gypsum (G) are highlighted in the graph. The results indicate that the mortar in concrete had no gypsum when the SO3 content in the sand was less than 1.5% at 56 days. Note that there were apparently no peaks corresponding to gypsum in the FG3 specimen at 240 days. Moreover, for the FG5 and FG7 samples,

Materials 2020, 13, x FOR PEER REVIEW 9 of 15

(a) (b)

Figure 8. Compressive strength of concrete with gypsum-contaminated aggregate: (a) gypsum- contaminated fine aggregate; (b) gypsum-contaminated coarse aggregate.

5.3. Ultrasonic Test The internal damage of specimens under internal sulfate attack was tested by ultrasonic pulse velocity, and the results are presented in Figure 9. The value of ultrasonic velocity is closely related to the microstructure. The more compact a structure is, the higher the ultrasonic velocity obtained will be. The results show that the change trend for the ultrasonic velocities of samples is consistent with that of compressive strength. Except for the FG3, FG5, FG7, G5, and G7 specimens, the damage to the other Materialssamples2020 was, 13 small, 1336 . 8 of 13

Materials 2020, 13, x FOR PEER REVIEW 10 of 15 (a) (b)

thereFigure was also 9. Ultrasonic a certain amount test results of gypsum of concrete in mortar samples: until (a 360) gypsum-contaminated days. It was clear that ﬁne during aggregate; the process of ISA,(bFigure) gypsum-contaminated the 9gypsum. Ultrasonic in concrete test results coarse could aggregate.of concretebe dissolved. samples: (a) gypsum-contaminated fine aggregate; (b) gypsum-contaminated coarse aggregate. 5.4. MineralFigure Composition 11 represent Examinations the X-ray patterns of concrete containing SO3 in coarse aggregate at 360 days. 5.4.It is Mineral worth mentioningComposition thatExamination all concrete with gypsum coarse aggregate incorporated had gypsum for To detect the existence of gypsum in mortar during the internal sulfate attack, while avoiding 360 days during the internal sulfate attack. This provides evidence that the dissolution rate of gypsum the eﬀToects detect of quartz the existence and dolomite of gypsum peaks in mortar in the XRD,during the the XRD internal results sulfate from attack 5 to, while 20 are avoid showning the in in gypsum-contaminated aggregate with bigger particle size is relatively slower.◦ ◦ Figureseffects of10 quartz and 11 and, for dolomite the most peaks intensive in the peak XRD, of gypsumthe XRD within results 20 from◦ in the5° to XRD 20° patterns.are shown in Figures 10 and 11, for the most intensive peak of gypsum within 20° in the XRD patterns.

(a) (b)

(c)

Figure 10. XRD patterns of paste in concrete with gypsum-contaminated ﬁne aggregate of diﬀerent Figure 10. XRD patterns of paste in concrete with gypsum-contaminated fine aggregate of different ages: (a) 56 days; (b) 240 days; and (c) 360 days. ages: (a) 56 days; (b) 240 days; and (c) 360 days.

Figure 11. XRD patterns of paste in concrete with gypsum-contaminated coarse aggregate at 360 days.

Materials 2020, 13, x FOR PEER REVIEW 10 of 15 there was also a certain amount of gypsum in mortar until 360 days. It was clear that during the process of ISA, the gypsum in concrete could be dissolved.

Figure 11 represents the X-ray patterns of concrete containing SO3 in coarse aggregate at 360 days. It is worth mentioning that all concrete with gypsum coarse aggregate incorporated had gypsum for 360 days during the internal sulfate attack. This provides evidence that the dissolution rate of gypsum in gypsum-contaminated aggregate with bigger particle size is relatively slower.

(a) (b)

(c)

Figure 10. XRD patterns of paste in concrete with gypsum-contaminated fine aggregate of different

ages:Materials (a) 202056 days, 13, 1336; (b) 240 days; and (c) 360 days. 9 of 13

Figure 11. XRD patterns of paste in concrete with gypsum-contaminated coarse aggregate at 360 days. Figure 11. XRD patterns of paste in concrete with gypsum-contaminated coarse aggregate at 360 days. Figure 10 shows a comparison of the X-ray diﬀractograms of concrete samples with diﬀerent

amounts of SO3 in sand at 56, 240, and 360 days. The peaks of ettringite (E), portlandite (P), and gypsum (G) are highlighted in the graph. The results indicate that the mortar in concrete had no gypsum when the SO3 content in the sand was less than 1.5% at 56 days. Note that there were apparently no peaks corresponding to gypsum in the FG3 specimen at 240 days. Moreover, for the FG5 and FG7 samples, there was also a certain amount of gypsum in mortar until 360 days. It was clear that during the process of ISA, the gypsum in concrete could be dissolved. Figure 11 represents the X-ray patterns of concrete containing SO3 in coarse aggregate at 360 days. It is worth mentioning that all concrete with gypsum coarse aggregate incorporated had gypsum for 360 days during the internal sulfate attack. This provides evidence that the dissolution rate of gypsum in gypsum-contaminated aggregate with bigger particle size is relatively slower. The XRD results show that, during ISA the gypsum dissolved when the SO3 content was less than 3% in sand. For the higher contents of SO3 in FG5 and FG7, the new gypsum phase would be formed by the dissolution of SO3. To solve these problems, the TG–DSC analysis of FG7 specimen was researched. Figure 12 shows the TG–DSC curves of the FG7 samples at ages of 56, 120, and 240 days. As shown in Figure 12b, the temperatures of 80–120, 120–150, and 390–470 ◦C correspond to the decomposition of AFt (3CaO Al O 3CaSO 32H O), gypsum (CaSO 2H O), and portlandite (Ca(OH) ), respectively. · 2 3· 4· 2 4· 2 2 During heating from 80 to 120 ◦C, 26 mol of water will be removed from AFt [18]. The amount of AFt in the specimen can be calculated by Equation (2). According to the mass loss (at 120–150 ◦C and 390–470 C) caused by the decomposition of CaSO 2H O and Ca(OH) (shown in Figure 12a), ◦ 4· 2 2 the quantities of gypsum and portlandite can be calculated using Equation (3) and Equation (4).

MAFt AFt(%) = WAFt (2) · 26MH

MG G(%) = WG (3) · 2MH

MP P(%) = WP (4) · 2MH

where WAFt,WG, and WP correspond to the mass loss in percentage attributable to AFt, gypsum, and portlandite dehydration, respectively, and MAFt,MG, and MP are the molecular weights of AFt, gypsum, and portlandite, respectively. MH is the molecular weight of H2O. Materials 2020, 13, x FOR PEER REVIEW 11 of 15

The XRD results show that, during ISA the gypsum dissolved when the SO3 content was less than 3% in sand. For the higher contents of SO3 in FG5 and FG7, the new gypsum phase would be formed by the dissolution of SO3. To solve these problems, the TG–DSC analysis of FG7 specimen was researched. Figure 12 shows the TG–DSC curves of the FG7 samples at ages of 56, 120, and 240 days. As shown in Figure 12b, the temperatures of 80–120, 120–150, and 390–470 °C correspond to the decomposition of AFt (3CaO·Al2O3·3CaSO4·32H2O), gypsum (CaSO4·2H2O), and portlandite (Ca(OH)2), respectively. During heating from 80 to 120 °C, 26 mol of water will be removed from AFt [18]. The amount of AFt in the specimen can be calculated by Equation (2). According to the mass loss (at 120– 150 °C and 390–470 °C) caused by the decomposition of CaSO4·2H2O and Ca(OH)2 (shown in Figure 12a), the quantities of gypsum and portlandite can be calculated using Equation (3) and Equation (4).

M AFt AFt(%) =WAFt  (2) 26M H

M G G(%) =WG  (3) 2M H

M P P(%) =WP  (4) 2M H where WAFt, WG, and WP correspond to the mass loss in percentage attributable to AFt, gypsum, and portlandite dehydration, respectively, and MAFt, MG, and MP are the molecular weights of AFt, gypsum,Materials and2020 portlandite,, 13, 1336 respectively. MH is the molecular weight of H2O. 10 of 13

(a) (b)

Figure 12. Thermogravimetry–diﬀerential scanning calorimetry (TG–DSC) curves of the FG7 specimen Figureat the 12. ages Thermogravimetry of 56, 120, and 240– days.differential (a) TG; (b scanning) DSC. calorimetry (TG–DSC) curves of the FG7 specimen at the ages of 56, 120, and 240 days. (a) TG; (b)DSC. Figure 13 shows the estimated quantities of AFt and gypsum at 56, 120, and 240 days. The results showFigMaterialsure that 13 2020 theshow, 13 amount, x sFOR the PEER ofestimated gypsum REVIEW quantities phase decreased of AFt and and the gypsum quantity at of 56, AFt 120, increased and 240 during days. the The12ISA. of results 15 showAdditionally, that the amount the amount of gypsum of portlandite phase decreased stayed relative and stablethe quantity throughout of AFt the wholeincreased process. during the ISA. Additionally, the amount of portlandite stayed relative stable throughout the whole process.

Figure 13. Estimated quantities of ettringite (Aft) and gypsum at 56, 120, and 240 days in concrete. 5.5. WP Dissolution in Concrete Figure 13. Estimated quantities of ettringite (AFt) and gypsum at 56, 120, and 240 days in concrete. 2 Figure 14 shows the concentration of SO4 − in the pore solution in concrete at diﬀerent ages. As5.5. shown WP Dissolution in Figure 14in, Concrete concrete with 5% and 7% SO3 in aggregate showed a higher concentration of sulfate ions as compared to the reference sample at all ages. The sulfate concentration increased with Figure 14 shows the concentration of SO42− in the pore solution in concrete at different ages. As 2 an increase in the gypsum content in specimens. Figure 14 also indicates that the SO4 − dissolved shown in Figure 14, concrete with 5% and 7% SO3 in aggregate showed a higher concentration of sulfate more easily for aggregate with a smaller particle size, which is consistent with the results for the ions as compared to the reference sample at all ages. The sulfate concentration increased with an expansion and compressive strength of samples. Furthermore, for the FG7 specimen, the concentration increase in the gypsum content in specimens. Figure 14 also indicates that the SO42− dissolved more of SO 2 was 0.0408 mol/L, which exceeded the minimum concentration of sulfate ions (1400 mg/L) to easily4 −for aggregate with a smaller particle size, which is consistent with the results for the expansion form gypsum in solution [11]. However, this number is much smaller than 8000 mg/L, the sulfate ion and compressive strength of samples. Furthermore, for the FG7 specimen, the concentration of SO42− concentration required for external sulfate attack to form gypsum [12]. Additionally, as the mineral was 0.0408 mol/L, which exceeded the minimum concentration of sulfate ions (1400 mg/L) to form analysis results showed that portlandite did not decrease, this means that there was no new phase of gypsum in solution [11]. However, this number is much smaller than 8000 mg/L, the sulfate ion gypsum formed. concentration required for external sulfate attack to form gypsum [12]. Additionally, as the mineral analysis results showed that portlandite did not decrease, this means that there was no new phase of gypsum formed.

Figure 14. Concentration of SO42− in the pore solution at different ages.

5.6. Microstructure

Materials 2020, 13, x FOR PEER REVIEW 12 of 15

Figure 13. Estimated quantities of ettringite (AFt) and gypsum at 56, 120, and 240 days in concrete.

5.5. WP Dissolution in Concrete

Figure 14 shows the concentration of SO42− in the pore solution in concrete at different ages. As shown in Figure 14, concrete with 5% and 7% SO3 in aggregate showed a higher concentration of sulfate ions as compared to the reference sample at all ages. The sulfate concentration increased with an increase in the gypsum content in specimens. Figure 14 also indicates that the SO42− dissolved more easily for aggregate with a smaller particle size, which is consistent with the results for the expansion and compressive strength of samples. Furthermore, for the FG7 specimen, the concentration of SO42− was 0.0408 mol/L, which exceeded the minimum concentration of sulfate ions (1400 mg/L) to form gypsum in solution [11]. However, this number is much smaller than 8000 mg/L, the sulfate ion concentration required for external sulfate attack to form gypsum [12]. Additionally, as the mineral analysis results showed that portlandite did not decrease, this means that there was no new phase of gypsum formed. Materials 2020, 13, 1336 11 of 13

2 Figure 14. Concentration of SO4 − in the pore solution at different ages. 5.6. Microstructure Figure 14. Concentration of SO42− in the pore solution at different ages. SEM was applied to observe the microstructure of mortar with 5% SO in sand. Figure 15 shows Materials 2020,, 13,, xx FORFOR PEERPEER REVIEW 3 13 of 15 the5.6 cracking. Microstructure behavior of concrete with 5% SO3 in sand at 120 and 240 days. It is shown that with the developmentSEM ofwas ISA, applied the to fracture observe the firstly microstructure extended of frommortar the with gypsum 5% SO33 inin to sand.sand. the FigFig paste.ure 15 Then,shows the crack the cracking behavior of concrete with 5% SO33 in sand at 120 and 240 days. It is shown that with the extended aroundthe cracking the behavior sand, and of concrete finally, with the 5% sand SO separatedin sand at 120 from and 240 the days paste. It completely.is shown that with Figure the 16 shows development of ISA, the fracture firstly extended from thethe gypsum to thethe paste. Then,, thethe crack crack the crack widthextended change around duringthe sand, the and ISA.finally The,, thethe picturesandsand separatedseparated shows fromfrom that thethe the pastepaste crack completely.completely. width FigFig increasedure 16 show significantly.s Additionally,thethe crackcrack in the widthwidth crack, changechange the duringduring needle thethe ISA ettringiteISA.. The picture crystals showsoriented thatthat thethe crack their width growth increased in significantly. the surface between the paste andAdditionally, aggregate, inin thethe as crack,crack, shown thethe needleneedle in Figure ettringiteettringite 17 . crystalscrystals Thus, orientedoriented the formation theirtheir growthgrowth of inin ettringite thethe surfacesurface was betweenbetween the thethe cause of ISA paste and aggregate, as shown in Figure 17. Thus, the formation of ettringite was the cause of ISA in in the concretepaste and with aggregate, gypsum-contaminated as shown in Figure 17 aggregate.. Thus, the formation of ettringite was the cause of ISA in thethe concreteconcrete withwith gypsumgypsum-contaminated aggregate.

(a) (b)

Figure 15. SEM photographs of cracking behavior of concrete with 5% SO in sand: (a) 120 days; Figure 15. SEM photographs of cracking behavior of concrete with 5% SO33 inin sandsand:: ((a)3) 120 days;; ((b))240 (b) 240 days.days.

(a) (b)

Figure 16.FigureSEM 16.photographs SEM photographs of of crack crack widthwidth in in concrete concrete::: ((a)) 120 (a) days 120;; ( days;(b))240 days. (b)240 days.

Materials 2020, 13, 1336 12 of 13 Materials 2020, 13, x FOR PEER REVIEW 14 of 15

(a) (b)

Figure 17. SEM images of cracking and EDS image of product in crack. (a) SEM; (b) EDS. Figure 17. SEM images of cracking and EDS image of product in crack. (a) SEM; (b) EDS. 6. Conclusions 6. Conclusions TheThe deterioration deterioration process process of of concrete concrete exposed exposed to internal internal sulfate sulfate attack attack was was investigated. investigated. The main The main conclusionsconclusion ares are as follows:as follows:

⚫With With an increasean increase in in the the SO SOcontent3 content in in aggregate, aggregate, the the concrete concrete expansion expansion increased, increased, and the and the • 3 compressivecompressive strength strength decreased. decreased. However, However, if the if the SO3 SO content3 content was lower was lowerthan 1.5% than and 1.5% 3% in and fine 3% in fineaggregate aggregate and and coarse coarse aggregate aggregate,, respectively. respectively. Although, Although, the expansion the expansionwas not obvious was, notand obvious,the andcompressive the compressive strength strength of samples of samples was above was the above initial the strength initial strengthof the reference of the referencespecimen. specimen.The critical values of SO3 content in fine and coarse aggregate were found to be 1.25% and 3%, The critical values of SO3 content in fine and coarse aggregate were found to be 1.25% and 3%, respectively. At values above these harmful expansion and large strength loss appear. respectively. At values above these harmful expansion and large strength loss appear. 2 2− ⚫SO SOdissolved4 dissolved more more easily easily when when the the particle particle size of size aggregate of aggregate was smaller was. When smaller. the SO When3 content the SO • 4 − 3 contentin the in sand the was sand less was than less 1.5% than at 56 1.5% days at, all 56 the days, gypsum all the was gypsum dissolved. was That dissolved. was the reason That why was the the expansion of specimens with a smaller SO3 content (1.25% for fine aggregate and 3% for coarse reason why the expansion of specimens with a smaller SO3 content (1.25% for fine aggregate and aggregate) was not significant. 3% for coarse aggregate) was not significant. ⚫The The concentration concentration of of sulfate sulfate ions ions inin thethe pore solution solution of of concrete concrete was was not notenough enough to form to forma new a new • phasephase of gypsum of gypsum when when the the SO SO3 3content content was less than than 7% 7% in in aggregate. aggregate. For Forthe theinternal internal sulfate sulfate attack,attack, the the main main reaction reaction product product isis ettringite, which which caus causeses the theexpansion expansion of specimens. of specimens. ⚫The The fracture fracture mode mode of of internal internal sulfate sulfate attackattack was the the crack crack extend extendinging from from gypsum gypsum particle particless to to • paste,paste, which which made made the the aggregate aggregate separateseparate from from the the paste. paste. In this paper, only TG–DSC was used to analyze the content of major phases, and no other methods • ⚫ In this paper, only TG–DSC was used to analyze the content of major phases, and no other methods werewere combined. combined. This This paper paper only only studied studied the changes changes in in compressive compressive strength strength and and expansion expansion performanceperformance of concreteof concrete containing containing gypsum gypsum aggregatesaggregates for 360 360 days. days. To To study study whether whether it is it safe is safe to use gypsumto use gypsum aggregates aggregates in concrete, in concrete, other other durability durability tests are are needed. needed. In Inaddition, addition, there there has been has been lacklack of evaluation of evaluation on on methods methods that that can can bebe used to to determine determine the the safe safe dosage dosage of gypsum of gypsum aggregate aggregate in concretein concrete at homeat home and and abroad, abroad, whichwhich will will be be the the focus focus of ofour our follow follow-up-up work. work.

Author Contributions: Conceptualization, W.C. and B.H.; data curation, W.C. B.H. and Y.Y.; writing—original Authordraft Contributions: preparation, WConceptualization,.C.; writing—review and W.C. editing, and B.H.; B.H. dataand M curation,.D. All authors W.C. B.H.have andread Y.Y.;and agreed writing—original to the draftpublished preparation, version W.C.; of the writing—review manuscript. and editing, B.H. and M.D. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by National Natural Science Foundation of China 51608265, and Anhui Key Funding: This research was funded by National Natural Science Foundation of China 51608265, and Anhui Key Laboratory of Advanced Construction Materials at Anhui Jianzhu University JZCL 200604KF. Laboratory of Advanced Construction Materials at Anhui Jianzhu University JZCL 200604KF. Acknowledgments:Acknowledgments:The The authors authors gratefully gratefully acknowledgeacknowledge the the assistance assistance from from Zhongyang Zhongyang Mao Mao,, Huan Huan Yuan,Yuan, Jun Jun WangWang from from NJTECH, NJTECH, and and the the sta staffff from from State State Key Key Laboratory Laboratory of of Materials Materials-Oriented-Oriented Chemical Chemical Engineering. Engineering. ConflictsConflicts of Interest: of Interest:The The authors authors declare declare no no conflict conflict of interest.

Materials 2020, 13, 1336 13 of 13

References

1. Hime, W.G.; Mather, B. “Sulfate attack,” or is it. Cem. Concr. Res. 1999, 29, 789–791. [CrossRef] 2. Skalny, J.; Marchand, J.; Odler, I. Sulfate Attack on Concrete; Spon Press: London, UK, 2002. 3. Leemann, A.; Löser, R. Analysis of concrete in a vertical ventilation shaft exposed to sulfate-containing groundwater for 45 years. Cem. Concr. Compos. 2011, 33, 74–83. [CrossRef] 4. Ma, B.; Gao, X.; Byars, E.A.; Zhou, Q. Thaumasite formation in a tunnel of Bapanxia Dam in Western China. Cem. Concr. Res. 2006, 36, 716–722. [CrossRef] 5. Gesoglu, M.; Guneyisi, E.; Nahhab, A.H.; Yazıcı, H. The effect of aggregates with high gypsum content on the performance of ultra-high strength concretes and Portland cement mortars. Constr. Build. Mater. 2016, 110, 346–354. [CrossRef] 6. Crammond, N. The thaumasite form of sulfate attack in the UK. Cem. Concr. Compos. 2003, 25, 809–818. [CrossRef] 7. Janković,K.; Miliˇcić,L.; Stanković,S.; Šušić,N. Investigation of the mortar and concrete resistance in aggressive solutions. TehniˇckiVjesnik 2014, 21, 173–176. 8. Fu, Y.; Gu, P.; Xie, P.; Beaudoin, J. A kinetic study of delayed ettringite formation in hydrated portland cement paste. Cem. Concr. Res. 1995, 25, 63–70. [CrossRef] 9. Clark, B.A.; Brown, P.W. The formation of ettringite from tricalcium aluminate and magnesium sulphate. Adv. Cem. Res. 2000, 12, 137–142. [CrossRef] 10. Tian, B.; Cohen, M.D. Expansion of Alite Paste Caused by Gypsum Formation during Sulfate Attack. J. Mater. Civ. Eng. 2000, 12, 24–25. [CrossRef] 11. Bellmann, F.; Moser, B.; Stark, J. Influence of sulfate solution concentration on the formation of gypsum in sulfate resistance test specimen. Cem. Concr. Res. 2006, 36, 358–363. [CrossRef] 12. Santhanam, M.; Cohen, M.D.; Olek, J. Sulfate attack research—Whither now? Cem. Concr. Res. 2001, 31, 845–851. [CrossRef] 13. Scholtzová, E.; Kucková, L.; Kožíšek, J.; Palkova, H.; Tunega, D. Experimental and computational study of thaumasite structure. Cem. Concr. Res. 2014, 59, 66–72. [CrossRef] 14. Kohler, S.; Heinz, D.; Urbonas, L. Effect of ettringite on thaumasite formation. Cem. Concr. Res. 2006, 36, 697–706. [CrossRef] 15. Oliveira, I.; Cavalaro, S.H.P.; Aguado, A. New kinetic model to quantify the internal sulfate attack in concrete. Cem. Concr. Res. 2013, 43, 95–104. [CrossRef] 16. Samarai, M.A. The disintegration of concrete containing sulphate-contaminated aggregates. Mag. Concr. Res. 1976, 28, 130–142. [CrossRef] 17. Crammond, N. Examination of mortar bars containing varying percentages of coarsely crystalline gypsum as aggregate. Cem. Concr. Res. 1984, 14, 225–230. [CrossRef] 18. Qoku, E.; Bier, T.; Westphal, T. Phase assemblage in ettringite-forming cement pastes: A X-ray diffraction and thermal analysis characterization. J. Build. Eng. 2017, 12, 37–50. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).