Role of the filler on Portland cement hydration at early ages

Construction and Building Materials 27 (2012) 82–90

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Role of the filler on Portland cement hydration at early ages V. Rahhal a, V. Bonavetti a, L. Trusilewicz b, C. Pedrajas c, R. Talero c,⇑ a

Civil Engineering Department, Faculty of Engineering, UNCPBA Avda. del Valle 5737, B7400JWI Olavarría, Argentina Escuela Universitaria de Ingeniería Técnica Industrial, UPM, Ronda de Valencia 3, Madrid 28012, Spain c ’’Eduardo Torroja’’ Institute for Construction Sciences, CSIC, C/Serrano Galvache, 4, 28033 Madrid, Spain b

a r t i c l e

i n f o

Article history: Received 9 February 2011 Received in revised form 30 June 2011 Accepted 18 July 2011 Available online 20 October 2011 Keywords: Portland cement hydration Crystalline mineral additions Quartz Limestone Filler

a b s t r a c t The effects of mineral additions, quartz (Q) and limestone (C) fillers, on Portland cement (PC) hydration are ultimately reflected in the mechanical behavior and durability of the resulting concrete. The physical and chemical interactions involved may expedite or retard the hydration rate. The present paper describes the hydration mechanism in Portland cements containing crystalline mineral additions (fillers, non-pozzolanic), based on the reaction rates and amount of products formed. The mineralogical composition of the Portland cements used determines their differential behavior when exposed to sulfate or attacked by chloride, separately. Ground quartz, Q, and limestone, C, were the mineral additions chosen. The results show that direct and non-direct stimulation of the hydration reactions increase with the replacement ratio and, obviously, as a result of the concomitant physic dilution effect. Mixing water would be responsible for direct stimulation. non-direct stimulation, in turn, would occur very early in PC hydration, for the positive and negative electrostatic charge acquired by the particles of fíller during grinding and/or initial mixing, and subsequently the zeta potential is generated as PC hydration progresses. On the other hand, it has also been demonstrated that both of the fillers interact chemically as specified below. Hence, Q filler is influenced by the portlandite of any PC, OPC and/or SRPC (due to its randomly inner texture totally compact, but above all, to its acid chemical character) to originate CSH-gels, although in a very slowly way, and for this reason, its chemical interaction is specific for the portlandite and generic for any PC of C3A significant content. Whereas, C filler interacts chemically mainly with the C3A to originate different carboaluminate types, and for this reason, its chemical interaction is specific for the C3A only. Consequently, overall system behavior varies depending on the crystalline mineral addition, the type of PC and the replacement ratio. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Although perfections continue to be introduced in the existing qualitative and quantitative models for Portland cement paste hydration mechanisms [1–3], the main outlines are in place [4]. Moreover, there are models describing the hydration mechanisms for tricalcium silicate and tricalcium aluminate in the presence of pozzolanic additions [5,6]. Pozzolanic additions are natural or artificial materials with capacity to interact chemically with hydration products from Portland cement (pozzolanic activity), as a consequence the capillary pore network becomes segmented and a particle size reduction of the calcium hydroxide is produced [7]. The use of such mineral additions contribute to less consumption of

⇑ Corresponding author. Tel.: +34 913020440; fax: +34 913020700. E-mail addresses: vrahhal@fio.unicen.edu.ar (V. Rahhal), lidia.trusilewicz@ upm.es (L. Trusilewicz), [email protected] (R. Talero). 0950-0618/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.conbuildmat.2011.07.021

natural resources entailed and a lower amount of energy needed to manufacture blended cements. The development of cement manufacture with pozzolanic additions (i.e., amorphous and/or vitreous) has led to the gradual inclusion of crystalline mineral additions known as fillers, some of which interact physically and/or chemically with Portland cement or its reaction products, while others do not [8]. In light of the lack of consensus among researchers about the physical and/or chemical interaction of siliceous and/or calcareous fillers on Portland cement hydration at early ages in this case, this additional research has been carried out with the aim of clearing up their respective behavior. Initially, such siliceous and/or calcareous fillers were used in response to the need to control bleeding in concretes with low cement content. Their effect on early age hydration of the Portland cement to which they were added was subsequently analyzed and studied [9] but not completed. The main conclusion drawn from these studies is that they stimulate hydration in the initial

V. Rahhal et al. / Construction and Building Materials 27 (2012) 82–90

stages when their particles are moistened by the mixing water. This hydration mechanism has been termed as direct stimulation [9] to differentiate it from the non-direct one whose explanation and justification is an aim of this new study. Finally, the use of fillers has become so widespread that they have been standardized as well as their blended cements too [10]. On the other hand and with regard to the filler, the particle surfaces are positively or negatively charged during grinding with Portland clinker and gypsum and/or when mixing process with water and aggregates, consequently and respectively attracting OH and Ca2+ ions [11] very at the start of the hydration. This first layer of anions or cations in turn attracts a second cluster of Ca2+ or OH ions, respectively. As the ionic layer thickens, the electrostatic force of the particles declines [12]. Besides this, the following much more important consideration has to be also taken into account when the hydration moves forward: all inorganic particles assume a charge when dispersed in water. In the case of crystalline silica (Q filler, in this work), this is due to surface silanol, Si–OH, groups losing a proton. The aqueous phase becomes slightly acidic (since it receives protons) whilst the silica surface becomes negative (due to the formation of Si–O ). The charged particle surface then attracts a layer of counter-ions (ions of the opposite charge) from the aqueous phase. In the case of silica, positive ions (Ca2+ mainly, in this work) will crowd the surface. Due to ionic radii considerations, the strongly adsorbed counter-ions will not fully offset the surface charge. A second layer of more loosely held counter-ions then forms. At a certain distance from the particle surface, the surface charge will be fully balanced by counter-ions. Beyond this point, a bulk suspension with a balance of negative and positive electrolyte exists. The size of the double layer will depend firstly on the amount of charge on the particle surface. A large charge, whether positive or negative, will result in a large double layer that stops particles getting close to each other because of the electrostatic repulsion between those particles carrying the same electrical charge. This situation is typical of stable (deflocculated) suspensions having a low viscosity. Conversely, a low surface charge requires fewer counter-ions and smaller double layers. Accordingly, particles then tend to flocculate which leads to high viscosity suspension. The zeta potential (mV) can be related to the energy needed to shear the particle and its inner layer of counter-ions away from the outer layer/bulk medium. This phenomenon has been illustrated in Fig. 1. In short, as it was mentioned earlier that particle charge influences the double layer size and so the zeta potential. Furthermore, the surface charge of calcium carbonates is scarcely changed by the change in pH under the conditions that the activity of CO23 or Ca2+ is kept constant. Consequently, the

83

hydrogen ion is not the surface charge controlling ion. It has also been reported by Sawada [13] that the surface charge of the calcium carbonate shifts to positive side with the decrease in pH. These experiments were conducted under the conditions that the equilibrium concentrations of lattice ions are not kept constant. Thus, the change in the potential is the result of the decrease in the carbonate ion concentration, [CO23 ], caused by its protonation, i.e., the increase in the [Ca2+] concentration. In addition, the surface properties of crystals are significantly altered by the conditions of the way of preparation and history of the crystals. Particularly, the crystal surface created by crush is very rough and active (both fíllers, the Q and the C specially, in this work). The rate of crystal growth on rough surface is much higher than that on the smooth surface and that surface is more favorable for adsorption. Consequently, attention is called for the comparison of the thermodynamic and kinetic data between the crystals from different origins and treatment even if they have the same polymorph. Finally and with regard to the calcite, CaCO3, (ground limestone or C fíller in this work) it presents three polymorphous [14]: the thermodynamically most stable, calcite (solubility product: Ksp = 10–8.48 at 25 °C), less stable, aragonite (Ksp = 10–8.34), and the most unstable polymorph, vaterite (Ksp = 10–7.91). Sawada [13] has demonstrated by different concentrations of phosphate, that the chemical species adsorbed on the vaterite surface are negatively charged one, [Ca2+
PO34 ], whereas, the chemical species adsorbed on the calcite surface are electrically neutral one: low phosphate concentration [(Ca2+)3(
PO34 )2], and high phosphate concentration [(Ca2+)(
PO24 )], (note that the portlandite is also non-charged species, i.e. electrically neutral one). In other words, calcite requires the adsorption of electrically neutralized species. Conversely, the vaterite surface is positive and thus the adsorption of negatively charged species is reasonably explained. For all these reasons, this new research has been conducted, therefore, to determine the optimal replacement ratio for fillers depending on Portland cement, mortar or concrete requirements. In addition, the present paper describes a possible mechanism for their interaction with Portland cement based on previous studies that support its validity [15]. 2. Objective The present paper aims: – to determine type of interaction between fillers (siliceous and/or calcareous) and Portland cement, during its hydration at early ages, and in the former case, whether it is of physical – by direct

Fig. 1. Electrostatic phenomenon in a solution for a charged particle. Graphical description of the Zeta potential.

84

V. Rahhal et al. / Construction and Building Materials 27 (2012) 82–90

way [9] – and chemical type – by non-direct way–, or of physical type only, and – to describe and propose a hydration mechanism for Portland cements containing filler by fitting their behavior to quantitative findings.

3. Experimental 3.1. Materials and methodology 3.1.1. Raw materials selection and characterization Two Portland cements (PC) with widely different mineralogical compositions were chosen to ensure that the results would be extensive to any type of PC. The first one, denominated PC1 and characterized by its high C3A(%) content, whose mineralogical composition was found to be followed: 51.0% C3S, 16.5% C2S, 14.0% C3A and 5.5% C4AF, a density of 3.08, a Blaine specific surface, BSS, of 319 m2/kg and a loss on ignition of 1.6%. On the other hand, the second Portland cement, PC2, was selected due to its low C3A(%) and high C3S(%) contents; its detail mineralogical composition was: 79.5% C3S, 2.5% C2S,  0.0% C3A (<1.0%) and 10.0% C4AF, a density of 3.21, a BSS of 329 m2/kg and a loss on ignition of 1.1%. Finally, the crystalline mineral additions (i.e. non-pozzolanic) chosen as fillers were as follows: – a quartz one, Q (ground ASTM C 778-92a sand [16]), with a SiO2 content of over 99%, a density of 2.70, a BSS of 395 m2/kg and a loss on ignition of 0%, being therefore additionally, siliceous in nature and acid in chemical character; and the other, – a limestone one, C (ground Spanish limestone routinely used to manufacture cement concretes, and mortars) with a CaCO3 content of over 95%, a density of 2.71, a BSS of 362 m2/kg and a loss on ignition of 42.5%, being therefore additionally, calcareous in nature and basic in chemical character. Both of the crystalline additions respond with their respective granular composition to ASTM C 595M-95 standard [17], physical requirement (amount retained when wet-sieved on No. 325–45 lm – sieve, max. = 20%).

Finally and in order to examine the moisture absorption of the crystalline additions, the equal amounts of 1.0 g of the each (C and Q) was separately placed in porcelain vials and then stored in a desiccator (with water distilled at bottom – instead of silica gel – to guarantee the RH P 95% conditions). The experiment was carried out at 23 ± 1.7 °C and the findings are present in Table 3.

3.1.2. Specimens manufacture and operating procedure Since, in general, three replacement ratios were used, (20, 30 and 40% wt.), a total of twelve blended cements were prepared and denominated as follows: 80/20, 70/30 or 60/40 of PC1/ or PC2/Q or C. A 100/00 ratio denotes the two pure PC used as a control. The trials and analyses conducted in this study have included: /1/ times of setting and normal consistency [18], /2/ pozzolanicity [19], /3/ chemically combined water [20] determinations, and /4/ X-ray diffraction analysis. Table 1 gives the setting times and the water amount needed for normal consistency determined in accordance with EN 196-3 standard [18]. According to the practical reasons, the tests were carried out for 10 cements only: two plain PC, PC1 and PC2, and 8 blended cements, 80/20 and 60/40 only, since the 70/30 results are expected to be placed in the middle of them both. On the other hand, the results of the 2-day pozzolanicity and chemically combined water trials are given in Table 2 and were determined for the complete number of 14 cements (twelve blended and two PC 100/00 specimens), since these findings posses a great importance (greater than in the former case) for the aims of this study. On the other hand, the amount of chemically combined water [20] was determined gravimetrically on the basis of loss of mass in heated samples whose hydration had been arrested at the age of 2 days. In the calculus, the following observations have been taken into consideration: AFt and AFm phases lose their water up to 200 °C, calcium hydroxide is decomposed and CSH water molecules are lost up to 500 °C, to be finally eliminated over 500 °C, [21]. Thus, the specimens were heated from 40 up to 925 °C and the value obtained was used to estimate the progress of the hydration reaction, assuming that both the ground limestone and the ground quartz were hydraulically inactive. The results are given in Table 2. Finally, semi-quantitative X-ray diffraction analyses were conducted on Philips PW Cu Ka diffractometer fitted with a graphite monochromator and set for 40 kV and 20 mA. All the samples were prepared, handed and analyzed in the same form and time, to facilitate their reliable semi-quantitative analysis.

Table 1 Setting times of the PC1 and PC2 and their POZC 80/20 and 60/40 with Q and C mineral additions (fillers).

a

Cements

Initial setting (h:m)

Final setting (h:m)

Setting time (h:m)

Water for normal consistency (ml)a

PC1 100/00 PC1/Q 80/20 PC1/Q 60/40 PC1/C 80/20 PC1/C 60/40 PC2 100/00 PC2/Q 80/20 PC2/Q 60/40 PC2/C 80/20 PC2/C 60/40

3:20 2:45 3:15 1:00 0:50 4:30 5:05 5:00 2:55 2:40

5:10 4:30 4:45 2:15 2:10 6:15 7:00 8:00 4:35 4:40

1:50 1:45 1:30 1:15 1:20 1:45 1:55 3:00 1:40 2:00

155.0 160.0 160.0 165.0 165.0 140.0 145.5 150.0 150.0 145.0

Water amount for 500 g of cement.

Table 2 Pozzolanicity (Frattini test) and chemically combined water. Age: 2 days. Cements

Concentration

Chemically combined water (gwater/gPC)

Absolute (mM/l)

Relative (%)

[OH ]

[CaO]

[OH ]

[CaO]

PC1 100/00 PC1/Q 80/20 PC1/Q 70/30 PC1/Q 60/40 PC1/C 80/20 PC1/C 70/30 PC1/C 60/40

72.5 65.0 60.0 57.0 65.0 61.0 55.0

7.6 9.5 10.3 11.6 9.0 9.8 10.5

100.0 112.1 118.2 131.0 112.1 120.2 126.4

100.0 156.3 193.6 254.4 148.0 184.2 230.3

0.139 0.152 0.164 0.180 0.144 0.151 0.158

PC2 100/00 PC2/Q 80/20 PC2/Q 70/30 PC2/Q 60/40 PC2/C 80/20 PC2/C 70/30 PC2/C 60/40

42.5 44.5 44.0 43.5 42.5 43.5 44.0

21.5 21.9 22.2 22.3 20.1 20.3 20.6

100.0 130.9 147.9 170.6 125.0 146.2 172.5

100.0 127.3 147.5 172.9 116.9 134.9 159.7

0.093 0.110 0.128 0.145 0.101 0.106 0.113

85

V. Rahhal et al. / Construction and Building Materials 27 (2012) 82–90

4. Results and discussion 4.1. Influence in the times of setting First of all, it may be inferred from Table 1 that the presence of both fillers have had a tangible effect on the setting times of both PC1 and PC2, which have always been substantially shortened (except for the PC2/Q samples), even though less with Q filler than with C filler and more with PC1 than with PC2. The observed influences are found to be related to the randomly inner texture of these materials which is different. Hence, the C filler texture is very granulated and with very sub-microscopic and microscopic particles as a result of its geochemical origin [22,23]. That is to say, it is not totally compact like for Q filler and, therefore, improves the moisture absorption capacity of C filler which is higher than it occurs for Q filler. This particular observation was confirmed in the laboratory conditions (Table 3). Beyond this, the results justify also that the humidity adsorption values for C filler have been, from the first day onward, higher than for Q filler, and in addition, it also justifies by the way, afterwards, its better behavior as an inner moist closed, which finds reflection in its water for normal consistency (Table 1) and the water needed for its storage values (additional chemical reason will be seen later on). Hence, the set of its PC is as easier as faster, where its C3A(%) wt. is always more receptive, logically, than its corresponding C3S(%) wt. content. In short and confirming our prior experience [9], the higher shortening of the setting times has once again been found to be provoked by the stimulation effect of the corresponding PC fraction hydration by both, Q and C fillers, but by direct stimulation [9], prompting more intense C3A than C3S hydration of PC, and, as a consequence, greater reduction of the PC setting times. Further proof lies in the observation that the BSS (362 m2/kg) of C filler is nonetheless lower than the BSS (395 m2/kg) of Q filler.

4.2. Hydration stimulation (by direct, non-direct and indirect way) First of all, it is also necessary to note (Table 2) that the 2-day pozzolanicity trial shows that as the replacement ratio grew, more calcium hydroxide CH (=portlandite) per gram of Portland cement was generated. The inference is that the fillers particles of the two crystalline mineral additions, Q and C, directly stimulated the hydration reactions [9], acting also at the same time as nucleation and precipitation centers, respectively, for calcium hydroxide. Therefore, this latter behavior should constitute an additional stimulus, initially due to the static electricity acquired by the PC and filler particles during grinding process and/or when mixing process, as explained in detail in Introduction. However, the second mechanism based on non-direct stimulation (which fundamentals were magnificently described by Sawada [13]), generated as the hydration moves forward, would have to be added in an overlaid way to the direct-way stimulation [9] caused previously by both fillers. It would also explain why, when the proportion of filler particles rises, hydration is stimulated more actively. The probable reason is that the larger number of nucleation centers (in case of Q filler) or of precipitation centers (in case of C filler), would strengthen initially the overall electrostatic force. Consequently, the zeta potential originated later on would finally result in the generation of more portlandite or calcium hydroxide, CH, per gram of PC, as observed in the pozzolanicity trial (Table 2). The above findings are corroborated by the amount of chemically combined water results (Table 2), which increases with the PC replacement ratio for both types of the applied mineral additions: Q and C fillers. As in the preceding trial [9], semi-quantitative XRD analysis was conducted on the hydrated cements up to the age of 2 days

Table 3 Humidity absorption for Q y C fillers. Crystalline mineral additions (fillers)

Q C

Humidity absorption (%) Age (days) 1

3

7

14

21

28

0.41 2.07

0.49 2.53

1.38 3.35

1.76 3.90

2.75 6.67

2.75 6.67

(Fig. 2). From the general point of view, none of two types of filler is chemically involved in formation of the hydration products contributing to system hydraulicity at such early age. However and by contrast, the most reactive pozzolanic additions, as silica fume (SF), nanosilica, metakaolin, some natural pozzolans and fly ashes, etc., are always deeply involved forming their own hydration products and contributing to the system hydraulicity. This phenomenon is called indirect stimulation of PC hydration by these very reactive pozzolanic additions [15,24–27]. According to the XRD patterns achieved and to the calcium hydroxide, CH, peaks development particularly, a new pathway for hydration stimulation (in fillers presence) is proposed, which is denominated from now as non-direct stimulation and explained hereafter. The XRD patterns of Fig. 2a–d show that the 2h = 18.02° CH peak is more intense in PC1 than in PC2 (433 and 135 a.u., respectively), despite of the higher C3S content in the latter. Taking into consideration that all the samples were prepared, handed and analyzed in the same form and time, this observation may be explained by the mineralogical composition of these two PC, as well as by the small w/c ratio = 0.5 used. Therefore, since C3A is known to have a higher hydration rate than C3S, the PC1 (14.0% of C3A and 51.0% C3S) must have used more mixing water from the very beginning of hydration than the PC2 (0.0% of C3A and 79.5% of C3S). The foregoing hypothesis is supported by the behavior of the 60/ 40 PC blended cements elaborated separately with both Q and C fillers. Since the filler does not possess pozzolanic properties, the mixing water that initially moistens its particles must prompt greater direct stimulation of the respective PC fractions. This effect would be enhanced by non-direct stimulation mentioned above. Comparing two pure PC, the PC2 fraction must benefit more than the C3A of PC1 one from the mixing water initially moistening the fillers particles. Support for this assertion is provided by a comparison of Fig. 2a,c,e,b,d,f, which shows that both crystalline mineral additions, Q and C, when blended with PC2, consistently generate more CH (369 and 454 a.u. for 40% Q and C, respectively), and when blended with the respective PC1 (253 and 259 a.u. for 40% Q and C, respectively). Further proof lies in the observation that as more effectively the filler stimulates hydration of the 60% PC2 fraction directly and/or non-directly, that greater is the amount of CH generated by this fraction. The last one is also demonstrated by the relative value of the chemically combined water (Table 2). In the present study, in fact, that behavior is observed in C filler specially which, due to its mentioned already randomly inner granulated texture [22,23], absorbs relatively more moisture than Q filler (characterized by its totally compact inner texture), Table 2. This would mean that from the outset, when C filler was mixed with PC to form its 60/40 blend, its 40% fraction must have adsorbed more of the initial mixing water (w/c = 0.5) than Q in the same case, where the 60% PC2 fraction would have benefited more than the 60% PC1 fraction. In short, the non-direct stimulation of hydration in the 60% PC fraction caused by the respective 40% crystalline mineral addition fraction of Q or C, must have been more intense for PC2 than it was for PC1 (to which the effect of direct stimulation would have to be admitted [9]). Conversely, the dilution effect was greater in the 60% PC fraction of PC1 than it occured for PC2.

86

V. Rahhal et al. / Construction and Building Materials 27 (2012) 82–90

Intensity (a.u.)

(a)

500 CHHC

(b)

CP1

500 CP2 PC2

PC1

400

400

300

300

200

200

100

HC CH

100

AFt AFt

0

0

5

(c)

10

15

20

2 θ, º

25

5

(d)

500

60/40 PC1/Q 40% Q

Intensity (a.u.)

400 300

20

25

500

60/40 PC2/Q 40% Q HC CH

300

HC CH Q

Q

200

200

AFt

100 AFt

0

0

5

10

15

20

2 θ, º

(f)

500

60/40

5

25

40% C PC1/C

400

Intensity (a.u.)

15

2 θ, º

400

100

(e)

10

10

15

20

2 θ, º

500

HC CH

60/40 PC2/C

25

40% C

400

300

300

CH HC

200

200 C

100

AFt

C

100

AFt

0

0

5

10

15

2 θ, º

20

25

5

10

15

2 θ, º

20

25

Fig. 2. XRD diffractograms of the PC1 or PC2/Q or C blends at 2 days of hydration. Note: AFt = AFt phase or ettringite, CH = portlandite or calcium hydroxide, Q = quartz, C = calcite.

Finally and according to the pozzolanicity test results (Table 2), some opposite relation are also found comparing the [CaO] values for Q and C fillers to the CH content present in their respective cement paste at 2 days-age, exhibited by XRD analysis, Fig. 2a–f. This findings point out that the C particles must behave more and better like ‘‘seed crystals’’ than Q filler in the same conditions. In addition, the C particles can never react chemically with portlandite, since the limestone is a salt of a feeble acid, the carbonic acid, CO3H2 (the carbonic acid is a diprotid acid and, for this reason, it has two constants of dissociation: K1 = 3.00  10 7 whose pK1 is = 6.523, and K2 = 4.00  10 11 whose pK2 = 10.398, i.e., the first dissociation is enough higher than the second one), and of a very strong base, Ca(OH)2, which suffers hydrolysis phenomena giving rise to a very weak basic solution. Moreover, its solubility is really low, 0.014 g/l at 25 °C, and it does not substantially modify the strong level basic originated in the same liquid phase mainly, by portlandite from PC hydration origin. However, the Q particles are expected to do so, even though much later and slowly [28].

In other words, Q filler is specific only for the portlandite from any PC hydration origin, i.e. Q filler is specific for the portlandite and generic for any PC (OPC and/or SRPC). In contrast, C fíller is specific for C3A of OPC mainly, and as a consequence, it is not generic for any PC; for instance, for the CEM I-SR 0 type of the new PrEN 197-1:2007 FOR DECISION standard [29] (of C3A 0% wt. content), it is neither generic nor specific. As far as here, it can be seen that Q and C fillers are not intrinsically hydraulic. However, both of them do not behave totally as inert. For this reason precisely, another additional explanation in base of the XRD results can be formulated, answering the following interesting question: are they, Q and C fillers, chemically inert indeed? This question has risen in the light of already reported works [30–36] about calcareous filler which is proved to react with the PC aluminate phase to give rise to calcium hemicarboaluminate (Hc), calcium monocarboaluminate (Mc) and calcium tricarboaluminate (Tc). In addition, it has been also demonstrated that during the hydration of silicate phases (C3S and C2S), some small amounts

87

V. Rahhal et al. / Construction and Building Materials 27 (2012) 82–90

of filler are incorporated to CSH gel [30], as well as carbonated hydrated calcium silicate compound is formed [31]. However, the latter possibility is actually believed to be more likely and, in fact, was already confirmed by calorimetric and XRD results, as it can be seen in the previous work [9]. This study provides some additional XRD patters of different hydration stages, from the calorimetric point of view, their respective troughs (calorimetric curves correspondent minima) for the same PC pastes (with and without fillers) which are present in Fig. 3. According to the early hydration results, the PC1 pastes with C and Q (60/40) present intensity of CH peak similar to the pure

Intensity (a.u.)

(b)

CH

PC1

Ts Ms

Ts

rd

Ms Ts

Ts

3 trough

CH

PC2

Intensity (a.u.)

(a)

PC1 paste, although the calcium silicates (C3S, C2S) content in the formers is reduced in 40% wt. On the contrary, the intensity of CH peak for PC2 pastes with both C and Q fillers is stronger that in their equivalent pure PC2 paste. Resuming, the XRD results confirm the same stimulation effect of the fillers on PC hydration reactions which is previously indicated by the pozzolanicity test and the non-evaporable water findings (as well as shown in the calorimetric curves and the total amount of heat dissipated at 2 days-age, both threaten in the preceded study [9]). On the other hand, the formation of the tri-substituted aluminum ferrite compounds was once again [30–36]

2nd trough

1st trough

2nd trough 1st trough

5

10

15

20

5

25

10

Intensity (a.u.)

(d)

CH

60/40 PC1/C

Ts Ms rd

3 trough

C Mc CS Ts

Ts Ms

Mc

25

CH

60/40 PC2/C

2nd trough

C 2nd trough

Tc

Tc

1st trough

5

10

Tc

15

20

1st trough

5

25

10

CH

60/40 PC1/Q

15

20

25

2 θ (°)

2 θ (°)

(e)

20

Intensity (a.u.)

(c)

15

2 θ (°)

2 θ (°)

(f)

Q

CH

60/40 PC2/Q

Ts 3rd trough

Ts

Ms

Ts

Ts

Ms Ms

2nd trough

Intensity (a.u.)

Intensity (a.u.)

Q

2nd trough

1st trough

1st trough

5

10

15

2 θ (°)

20

25

5

10

15

20

25

2 θ (°)

Fig. 3. XRD analysis at the calorimetric curves trough stages [10] for early hydration of PC1 (three troughs) and PC2 (two troughs) and their 60/40 blends with crystalline mineral additions, Q or C. Note: Q = quartz; C = calcite; CH = portlandite, Ts = tri-sulphate aluminum ferrite hydrate, Tc = tri-carbonate aluminum ferrite hydrate, Ms = monosulphate aluminum ferrite hydrate, Mc = mono-carboaluminate ferrite hydrate, CS = carbonated calcium silicate hydrated.

88

V. Rahhal et al. / Construction and Building Materials 27 (2012) 82–90

demonstrated (AFt: Ts or Tc), and also its transformation to the mono-substituted aluminum ferrite compound phase (AFm: Ms or Mc) for samples based on PC1 corroborated, as observed in Fig. 3a,c,e. Nonetheless and according to the XRD results of Fig. 3c, the Ts amount is greater and more consistent (48 a.u. for the 1st trough and 50 a.u. for the 3rd trough) than it occurs for Tc amount (22 a.u. and 15 a.u., respectively). The reason may be related perhaps to the gypsum solubility in water parameter which is  172-fold higher than that for calcite (in any case, the amount of hydration heat released has to be regarded as being considerable [9]). However, in the case of the pure PC2 and its PC2/filler blends (Fig. 3d,e,f), the absence of the tri- and mono-substituted aluminum ferrite phases (Ts and Ms) was verified in correspondence with the absence of C3A in the PC2, and for the same reason, the absence of the corresponding tri- and mono-carboaluminate phases, Tc and Mc, respectively, was verified as well, as logical. The formation of Tc and Mc is especially interesting in blends with the C filler originated from the C3A derived from the PC1 only. Hence, C filler is specific mainly for the C3A of PC and, as a consequence, C filler is not generic for any PC. For instance, its influence for the CEM I-SR 0 type (by the new PrEN 197-1:2007 FOR DECISION standard [29]), of 0% wt. C3A content, is neither generic nor specific. It does not mean in any case and nonetheless that C filler possesses intrinsic hydraulic characteristics. In addition, these compounds (Tc and Mc) have been already found before in other experiments [30–36]. Similarly, the formation of carbonate hydrated calcium silicate (CS) has been detected previously as well [30,31]. It could also be argued, nevertheless, that at these early ages of hydration, the calcite part which reacts with C3A of PC1 to give rise

PORTLAND CEMENT CRYSTALLINE MINERAL ADDITION (filler) MIXING WATER

to Tc comes from portlandite (previously transformed in calcite), and not from C filler origin. However, this hypothesis is difficult to verify. It could be only deduced perhaps indirectly that is not feasible, due to that Tc has not been detected in the corresponding XRD pattern of the POZC PC1/Q 60/40, Fig. 3e, for none of the examined calorimetric troughs [9]. Furthermore, the calcite solubility in water solution containing Ca(OH)2 must be lower than in water itself, likewise gypsum (2.05 g CaSO4/l at 20 °C which equals to 2.3 g CaSO4
2H2O/l), having moreover tendency to insignificant super-saturation, and when dispersed in Ca(OH)2, its solubility turns to  1.2 g CaSO4/1000 g of solution. In sum up, it can be seen that both of the crystalline mineral additions, Q and C, stimulate the PC hydration by direct and non-direct way, being these stimulations intensified as the filler dosage (PC replacement) increases. If all these concurrences appear together, it is always to detriment of final durability of their concretes and mortars. Nevertheless, if the fine aggregate is replaced by Q filler (instead of Q filler being added) the resultant durability may not be diminished but, oppositely, it increases. In contrast, the C filler should always reduce it in case of the proportions used in this work, due to its randomly inner texture which is very granulated and with very sub-microscopic and microscopic particles (a result of its geochemical origin [22,23]), and also due to its basic chemical character. Finally and in base of the conducted analysis and studies, a flow chart depicting the physical–chemical interactions in the ‘‘Portland cement–crystalline mineral addition–water system’’ is proposed in Fig. 4. Note in particular the decline in anhydrous PC to the benefit of calcium hydroxide (CH), hydrated calcium silicate (CSH) and AFt and AFm phases’ precipitation. According to the XRD results, it can be said that, at 2 days age, neither portlandite nor PC paste (exempt from C3A% wt. content) exhibit chemical interaction with C filler. Furthermore, the following reaction mechanism for the ‘‘Portland cement–crystalline mineral addition–water system’’ is suggested: If the particles of both crystalline mineral additions, Q and C fillers in this work, replace partially any PC, various influence in its hydration is produced. First of all, they tend to stimulate their hydration reactions directly [9], due to the mixing water that moistens the surface of their particles during initial mixing. At the same time, either Q and C fillers act as nucleation and precipitation centers, respectively, for the calcium hydroxide (=portlandite) generated once the 60% PC fraction is hydrated, behaving likewise as nucleation centers and ‘‘seed crystals’’, correspondently. This other type of hydration stimulation may take place in an overlaid way:  due, in the very beginning, to the positive and negative electrostatic charge acquired by the particles during grinding and/or mixing, which would attract the OH and Ca2+ ions in the calcium hydroxide molecule until their entire surface is covered; then, this first layer of anions or cations in turn attracts a second cluster of Ca2+ (or OH ) ions, respectively, and so on until till the whole electrostatic force of the original ‘‘pole of attraction’’ (each particle of the crystalline mineral addition) is depleted, and later on;  due to the zeta potential development specially, originated when the hydration moves forward [13].

CSH GEL ETTRINGITE CALCIUM HYDROXIDE

MONOSULFOALUMINATE

Fig. 4. Scheme for Portland cement hydration with crystalline mineral additions (Q and C fíllers in this work).

Thus, raising the PC replacement ratio the number of ‘‘attraction poles’’ is naturally increased, but above all, the zeta potential originated and the precipitation rate of CH crystals, further stimulating the PC hydration reactions. In order to differentiate this stimulation effect from the known direct stimulation, it might be simply termed as non-direct stimulation, since indirect stimulation denominates already the very reactive pozzolanic additions action

V. Rahhal et al. / Construction and Building Materials 27 (2012) 82–90

(silica fume [24], nanosilica, metakaolins [24,25], some fly ashes [26] and natural pozzolans [27]). 5. Summary To summarize, early hydration of PC component minerals in the crystalline mineral additions (fillers) presence, is stimulated both directly and non-directly. Both of these phenomena are intensified as the PC replacement ratio increases and, obviously, as a result of concomitant physical dilution effect. In addition, both crystalline mineral additions, Q and C fillers, interact chemically with portlandite and with PC pastes of high C3A% wt. content, respectively, already at 1 and 2 days age. They give a place to new hydration products formation: CSH gels [28] and different carboaluminate types, respectively as well, although in insignificant amounts. Their contribution to enhance hydraulicity system of PC fraction origin, exclusively, has now been more and better interpreted, cleared up and understood. On the contrary, a PC exempt of C3A% wt. content does not interact chemically at 1 and 2 days age with crystalline mineral additions of calcite type. However, those of quartz type do interact chemically but with portlandite only, originated by PC prior hydration. The chemical interaction of C filler is specific, as it reacts chemically with the C3A of OPC mainly. On the other hand, the Q filler behavior is specific as well since it reacts only with portlandite from any PC origin, (OPC and/or SRPC, including the SRPC of 0% wt. C3A content [29]). Nevertheless and in any case of PC replacement or mineral additions type applied, the mechanical strengths and durability aspects of the final building material must have always the final say on the matter. These particular issues will be addressed in detail in future papers. As a final observation, it is argued that all these results and conclusions drawn about the Q and C fillers role and their influence on PC hydration at early ages, are also wholly applicable to fine and coarse aggregates used commonly to manufacture cement concretes and mortars, even though only from the qualitative point of view.

89

in front of the mixing water amount for normal consistency and setting times, is different as well, as it is demonstrated in this research. Furthermore, in case of volume stability, mechanical strength (compressive and flexural) and facing different chemical attacks (sulfates, chlorides, sea water, carbonation, lixiviation, ASR and ACR) and hydration heat released, the behaviors of these blended cements must be different and dependent on the filler type incorporated. These questions are already addressed in other parallel studies whose results will be present in future papers. Finally, the adequate replacement by Q filler (acid chemical character and very compact its randomly inner texture) must take into consideration the fine and/or coarse aggregate of the PC concrete (above all, whether its granulometric curve is non-continuous and the PC concrete is self–compacting concrete) than with the PC itself. As the most indicative criteria, material workability, mechanical strengths and durability most of all have to be maintained. In any case, the use of plasticizer or super-plasticizer admixture in appropriate quality and quantity, may be needed in order to the water/binder ratio does not increase too much. In contrast, any C filler (basic chemical character and non-compact but very granulated its randomly inner texture) amount contribution calculated and referred to either fine and/or coarse aggregate and/or to the PC itself, does not have to provide suitable durability in front of sulfate attack, for instance. This fact has been already verified with Portland limestone cement mortars by several researchers, i.e. Tosun et al. [37], Talero et al. (mortars and pastes) [38] and Calleja and Aguanell [39] with concretes of limestone coarse (gravel: 1210 kg/m3) and fine (510 kg/m3) aggregate and PC CEM I-SR 3 type [29] (C3A content  2% < 3%). For the foregoing, it should come as no surprise that neither existing Spanish Code RC-08 [40] for the Acceptance of Cements nor European PrEN 197-1:2007 standard [29] consider them to be ‘‘sulfate-resistant’’ or even ‘‘sea water-resistant’’, even when the matrix PC is of SRPC CEM 1-SR 0 type. Fortunately, no other type of aggressive chemical attack decays concrete as rapidly as sulfates. This has made it possible to confide sufficiently in both siliceous and limestone aggregate and filler in concrete manufacture, with no need to disregard the role of each in front of the various types of aggressive chemical attack. The results of the rest of chemical attacks will be the object of forthcoming articles.

6. Deductions and technical consequences It can be seen that according to the obtained conclusions, both of the fillers, Q and C, are characterized by and also differ in the following properties: – randomly inner texture: totally compact for Q filler and nontotally compact but very granulated and with very sub-microscopic and microscopic particles for C filler as a result of its geochemical origin [22,23], – nature: siliceous for Q filler and calcareous for C filler, and – chemical character: acid for Q filler and basic for C filler. Hence and taking into consideration the last mentioned, their respective physicochemical interactions with portlandite and/or with C3A of any PC, respectively, are very different as well: – Q filler: is specific for the portlandite only from any PC hydration, and for this reason, it is generic also but for any PC, and – C filler: is specific for the C3A of OPC mainly, and as a consequence, it is not generic for any PC, for instance, for the CEM ISR 0 type of the new PrEN 197-1:2007 FOR DECISION standard [29] (of C3A 0% wt. content), it is neither generic nor specific. For these reasons precisely and being all the circumstances the same, the behaviors of their, Q and C, respective blended cements

Acknowledgments We would like to thank the Fundación Rotaria, the Universidad Nacional del Centro de la Provincia de Buenos Aires, for financial support, and the Instituto de C.C. ‘‘Eduardo Torroja’’-CSIC from Spain as well, for having provided the authors with necessary cementing materials and some analytical and experimental techniques. References [1] Double DD, Thomas NL, Jameson DA. The hydration of Portland cement. Evidence for an osmotic mechanism. In: 7th Int congress on the chemistry of cement, proceedings, vol. II, Paris; 1980. p. II-256–60. [2] van Breugel K. Modelling of cement-bases systems–the alchemy of cement chemistry. Cem Concr Res 2004;34:1661–8. [3] Bentz DP. A review of early-age properties of cement-based materials. Cem Concr Res 2008;38:196–204. [4] Mindess S, Young J. Concrete. New Jersey: Prentice-Hall, Inc., Englewood Cliffs; 1981. [5] Takemoto K, Uchikawa H. Hydratation des ciments pouzzolaniques. In : 7th International congress on chemistry of cement, vol. IV, Paris; 1980. p. II-1–29. [6] Uchikawa H, Uchida S. Influence of pozzolan on the hydration of C3A. In: 7th International congress on chemistry of cement, vol. III, Paris; 1980. p. IV-24– 35. [7] Mehta PK, Monteiro P. Concrete structure, properties and materials. New Jersey. USA: Prentice Hall Inc.; 1993. [8] Gutteridge WA, Dalziel JA. Filler cement: the effect of the secondary component on the hydration of portland cement: part I. A fine non-hydraulic filler. Cem Concr Res 1990;20(5):778–82.

90

V. Rahhal et al. / Construction and Building Materials 27 (2012) 82–90

[9] Rahhal V, Talero R. Early hydration of Portland cement with crystalline mineral additions. Cem Concr Res 2005;5(7):1285–91. [10] EN 197-1:2000 Standard. Cement. Part 1: composition, specifications and conformity criteria for common cements. AENOR, Calle Génova, 6, 28004 – MADRID. [11] Talero R. Is the clay exchange capacity concept wholly applicable to pozzolans. Mater Constr 2004;54(276):17–36. 2005;55(277):82. [12] Bombled JP. Etude rhéologique des pâtes crues de cimenterie. Revue de Matériaux de Construction 1970;609:229–38. [13] Sawada K. The mechanism of crystallization and transformation of calcium carbonates. Pure Appl Chem 1997;69(5):921–8. [14] Plummer LN, Bunsengerg E. The solubility of calcite, aragonite and vaterite in CO2–H2O solutions between 0 and 90 °C, and an evaluation of the aqueous model for the system CaCO3-CO2-H2O. Geochim Cosmochim Acta 1982;46(6): 1011–40. [15] Rahhal V. Characterization of pozzolanic additions by conduction calorimetry. PhD Thesis, Universidad Politécnica de Madrid-Spain; December 12 2002 (in Spanish). [16] ASTM C 778–92a Standard. Standard specification of standard sand. Annual Book of ASTM Standards, Section 4 Construction, vol. 0.4.01 Cement; Lime; Gypsum; 1995. [17] ASTM C 595M – 95 Standard: Standard specification for blended hydraulic cements. Annual Book of ASTM Standrads, Section 4, Construction, vol. 04.01, Cement; Lime; Gypsum; 1995. p. 291–6. [18] EN 196-3 Standard. Methods of testing cement. Part 3. Times of setting and volume stability determinations (Le Chatelier´s needles). AENOR. [19] EN 196-5:1996 Standard. Methods of testing cement. Pozzolanicity test for pozzolanic cements. AENOR. [20] Power TC. The non-evaporable water content of hardened portland cement paste. ASTM Bull 1949;158:68–76. [21] Taylor HFW. La Química de los Cementos (Chemistry of Cements). Urmo, Ed., Calle Espartero, 10, Bilbao (Spain); 1978 (in Spanish). [22] Melgarejo JC (editor) Atlas de Asociaciones Minerales en Lámina Delgada. Universidad de Barcelona, Fundación Folch, Reedición: Publicaciones UB. , vols. I and II. Publicaciones de la Univ. De Barcelona, Adolfo Florensa, s/n, 08028– Barcelona; (Spain) 2003 (in Spanish). [23] Bastida F. Geología: Una visión moderna de las ciencias de la tierra, vol. I. Ediciones Trea, SL, María González la Pondala, 98, Nave D, 33393–Gijón (Spain); 2005 [in Spanish]. [24] Talero R, Rahhal V. Calorimetric comparison of Portland cement containing silica fume and metakaolin: Is silica fume, like metakaolin, characterized by pozzolanic activity that is more specific than generic? J Therm Anal Cal 2009;96(2):383–93. [25] Talero R. Expansive synergic effect of ettringite from pozzolan (metakaolin) and from OPC, co-precipitating in a common plaster-bearing solution. Part II:

[26] [27] [28]

[29]

[30]

[31] [32] [33]

[34]

[35]

[36] [37]

[38]

[39] [40]

Fundamentals, explanation and justification. Constr Build Mater 2011;25: 1139–58. Rahhal V, Talero R. Fast physics-chemical characterization of fly ash. J Therm Anal Cal 2009;96(2):369–74. Rahhal V, Talero R. Fast physics-chemical characterization of natural pozzolans and other aspects. J Therm Anal Cal 2010;99:479–86. Delgado A. A contribution to the analysis and study of the aggregate-Portland cement paste interface. PhD Thesis, next upcoming defence at the Chemical engineering department of the university of Castilla-La Mancha, Ciudad Real – Spain (in Spanish). PrEN 197-1:2007 FOR DECISION Standard: Cement – Part I: composition, specifications and conformity criteria for common cements. CEN/TC 51/WG 6 Draft Report, National specifications for sulphate resisting cement, European Committee for Standardization, date 2007–2008; Secretariat NBN, CRIC, Rue Volta 10 – B 1050 Bruxelles-Belgium; December 2003, 13 pp. Ramachandran VS, Zhang Ch. Cement with calcium carbonate additions. In: Proceeding of the 8th international congress on the chemistry of cement. vol. II, Río de Janeiro, Brasil; 1986, p. 178–87. Pera J, Husson S, Guilhot B. Influence of finely ground limestone on cement hydration. Cem Concr Comp 1999;1:99–105. Sharma RL, Pandey SP. Influence of mineral additives on the hydration characteristics of ordinary portland cement. Cem Concr Res 1999;9(9):1525–9. Ingram KD, Daugherty KE. Limestone additions to portland cement: uptake, chemistry and effects. In: Proceeding of the 9th international congress on the chemistry of cement, , vol. V, India; 1992. p. 180–6. Tezuka Y, Gomes D, Martins JM, Djanikian JG. Durability aspects of cements with high limestone filler content. In: Proceeding of the 9th international congress of cement chemistry, vol. V, India; 1992. p. 53–9. Vernet, Noworyta C. Mechanisms of limestone fillers reactions in the system {C3A-CSH2-CH-CC-H}: Competition between calcium monocarbo- and monosulfo-aluminate hydrates formation. In: Proceedings of the 9th international congress of cement chemistry, vol. V, India; 1992. p. 430–6. Bonavetti VL, Rahhal VF, Irassar EF. Studies on the carboaluminate formation in limestone filler blended cements. Cem Concr Res 2001;31(6):853–9. Tosun K, Felekoglu B, Baradan B, Akin Altun I. Effects of limestone replacement ratio on the sulphate resistance of Portlad limestone cement mortars exposed to extraordinary high sulphate concentrations. Constr Build Mater 2009;23: 2534–44. Talero R, Pedrajas C, Delgado A, Rahhal V. Re-use of incinerated agro-industial waste as pozzolanic addition. Comparison with Spanish silica fume. Mater Constr 2009;59(296):53–89. Calleja J, Aguanell M. Considerations about the Le Chatelier-Ansttet test and the behaviour of cements in front of sulphates. Mater Constr. 1980;179:39–48. Instrucción para la Recepción de Cementos RC-08 (Real Decreto 956/2008 de 6 de junio; BOE núm. 28 de 19 junio; 2008).