Microstructure of cement pastes with residual rice husk ash of low amorphous silica content

Microstructure of cement pastes with residual rice husk ash of low amorphous silica content

Construction and Building Materials 80 (2015) 56–68 Contents lists available at ScienceDirect Construction and Building Materials journal homepage: ...

3MB Sizes 2 Downloads 4 Views

Construction and Building Materials 80 (2015) 56–68

Contents lists available at ScienceDirect

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

Microstructure of cement pastes with residual rice husk ash of low amorphous silica content J.H.S. Rêgo a,⇑, A.A. Nepomuceno a, E.P. Figueiredo b, N.P. Hasparyk c a

Environmental and Civil Engineering Department, University of Brasília, Campus Universitário Darcy Ribeiro, Asa Norte, 70910-900 Brasília, DF, Brazil School of Civil Engineering, Federal University of Goiás, Praça Universitária s/n, Setor Universitário, 74605-22 Goiânia, Goiás, Brazil c Department of Technical Support and Control, Furnas Centrais Elétricas, Rodovia BR 153/Km 510, Vila São Pedro, 74001970 Aparecida de Goiânia, Goiás, Brazil b

h i g h l i g h t s  Residual rice husk ash (RHA) produced in Brazil was characterized.  Pastes with RHA with low amorphous silica content have strong consumption of Ca(OH)2.  There is a refinement of the pore structure of the pastes with this kind of RHA.  RHA with low amorphous silica content, when finely divided, shows adequate pozzolanic activity.

a r t i c l e

i n f o

Article history: Received 10 July 2014 Received in revised form 15 October 2014 Accepted 24 December 2014

Keywords: Portland cement Residual husk ash rice Pozzolanic activity Microstructure

a b s t r a c t Residual rice husk ash (RHA) has been frequently suggested as possible mineral addition in cements. However, the characteristics of residual RHAs produced by different manufacturers and the effect of residual RHA with low amorphous silica content in addition in cement pastes are poorly understood. This paper aims at characterizing the different varieties of residual RHA produced in Brazil and at investigating the microstructure of cement pastes with 20% replacement of ordinary cement by RHAs which presents high and low amorphous silica content. A broad array of techniques was employed in the characterization, such as X-ray diffraction (XRD), thermogravimetric analysis (TG), scanning electron microscopy (SEM), X-ray microanalysis, and porosimetry by mercury intrusion (PMI). The results demonstrate that RHA with low amorphous silica content, finely divided, reduces the calcium hydroxide content and the Ca/Si ratio of calcium silicate hydrate (C–S–H), and also refines the porous structure of the pastes. Residual RHAs with low amorphous silica content thus proved to be suitable mineral addition for cements. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Rice husk ash (RHA) produced by controlled burning processes has a high amorphous silica content and is used as a mineral addition to the in cements, with properties similar to silica fume. Rice processing industries also use rice husks as fuel to generate the heat or electricity required by the various internal processes. As the combustion is carried out without control of the burning process, it generates large amounts of residual RHA by-products, whose proper disposal represents an important environmental issue [1,2]. Residual RHA is produced in many countries, and studying its properties is becoming increasingly important [3–5]. The residual RHA usually contains residual carbon and partly crystallized silica [6,7]; its main difference from the RHA produced by controlled burning is a lower content of amorphous silica. Several ⇑ Corresponding author. E-mail address: [email protected] (J.H.S. Rêgo). http://dx.doi.org/10.1016/j.conbuildmat.2014.12.059 0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.

authors have suggested the use of finely divided residual RHA as a mineral addition in cements [1–21]. However, only few studies have evaluated the characteristics of residual RHAs produced by different burning processes, and compared the effects of adding RHAs with high (>80%) and low (<20%) amorphous silica content to the cement, assessing the pozzolanic activity and the microstructure of the pastes [17]. Isaia [8] criticized the preconceptions against RHA produced without control of the burning process, noting: ‘‘how important is the presence of these crystalline phases in pozzolans, analyzed separately, for the performance of concrete or mortar? In fact, what matters is the final result that the specific pozzolan or mix with another pozzolan can provide when one or more levels of substitution are studied, in relation to the reference mixture.’’ When dispersed in the paste, the small pozzolan particles generate a large number of nucleation sites for precipitation of cement hydration products. Additionally, the microfiller (physical) effect of the finer grains ensures a denser packing within the cement paste and

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

reduces the wall effect in the paste-aggregate transition zone. This reinforces the effect of the pozzolanic reaction between the amorphous silica present in the pozzolan and the calcium hydroxide produced by cement hydration reactions. These behaviors highlight the synergy between Portland cement hydration, pozzolanic reaction, and, mainly, microfiller effect, whose combination leads to a improvement of the compressive strength compared to Portland cement mixtures [11]. Reducing the particle size of RHAs with low amorphous silica content influences their pozzolanic activity [7,8] and increases the their Ca(OH)2 consumption [4,21]. The ultrafine grinding of residual RHA, that reduces the average particle diameter below 10 lm, was found to be sufficient for generating high pozzolanic activity [1,12]. Concrete with up to 25% replacement of the cement by finely divided residual RHA exhibited improved mechanical properties, compared to a reference concrete [16,18], although the improvement was more significant for RHAs produced by controlled burning [13]. The use of residual RHA and RHA produced by controlled burning has a positive effect on autogenous shrinkage, owing to their porous cellular structure. RHA probably acts as an internal reservoir, providing a local source of water to the paste volume [14]. Depending on the level of substitution in the cement, the finely divided residual RHA can inhibit or enhance the alkaliaggregate reaction in concrete [6]. The partial replacement of the cement by residual RHA improves the decreased expansion, the weight loss and the loss of compressive strength resulting from acetic acid and nitric acid attack [19]. Sensale (2010) [17] concluded that the finely divided residual RHA is effective for producing more durable concrete, compared to the reference concrete. The RHA produced without control of the burning process was deemed adequate for the production of self-compacting concrete with normal resistance, containing up to 40% RHA [20]. In the present paper we characterized residual RHAs produced in Brazil. Furthermore, we analyzed the compressive strength of mortars and microstructures of cement pastes with 20% replacement of ordinary cement by RHAs with high and low amorphous silica content, through X-ray diffraction (XRD), thermogravimetric (TG) analysis, scanning electron microscopy (SEM) with X-ray microanalysis, and porosimetry mercury intrusion (PIM). 2. Materials and methods 2.1. Materials The different burning procedures used in industries producing residual RHAs in several states of Brazil were analyzed. We selected nine residual RHAs to represent the main burning processes. The residual RHAs 1–5 and 9 are produced by rice processing industries from different regions of the country, that use rice husk as fuel to generate power and heat for the rice drying and/or parboiling processes. These residual RHAs are burned in furnaces or boilers without control of the burning process, and are often dumped on the roadside or into rivers in close proximity to the industries. The residual RHAs 6, 7, and 8 were produced by thermal power plants using rice husk as fuel, again without control of the burning process. These power plants generate large amounts of residues, without proper disposal. The RHA 10 was produced in a power plant with a controlled burning process. Burning was performed in a furnace at a temperature of 700 °C for 2 h, with slow cooling at room temperature. The milling process of the 10 RHAs was carried out for 5 h in a ball mill specified for the Los Angeles abrasion test in coarse aggregates (NBR NM 51/2001 [22]). Twelve (12) steel balls (6.0 kg) were used as abrasive weight, to process a total of 2.5 kg RHA. This specific procedure was adopted to ensure milling of all RHAs with the same energy, although this milling procedure is not the most suitable for RHA. The chemical characterization of 10 RHAs was performed by Energy Dispersive X-ray Fluorescence (ED XRF) and Loss on Ignition analysis. The ED XRF data were obtained using a Shimadzu spectrometer (EDX-720) in high vacuum, that analyzes the range of elements from sodium (11Na) to uranium (92U) with a rhodium target X-ray tube. The particle size distribution of RHAs was measured using a laser diffraction particle size analyzer (Mastersizer S Standard Bench – Malvern Instruments) with ethyl alcohol as dispersant and ultrasonic agitation for 60 s. The BET specific surface area was measured using a nitrogen adsorption apparatus (Autosorb 1 – Quantachrome Nova2200). The X-ray diffraction (XRD) data were obtained by a Geigeflex T/Max RIGAKU diffractometer, operating with copper

57

radiation (CuKa = 1.5406 Å) and a nickel filter at 40 kV and 20 mA. The scanning speed was 2°/min in a 2h range from 2° to 70°. The RHAs specific mass was obtained through the NBR NM 23 (2000) standard [23]. The Pozzolanic activity index (PAI) with cement of RHAs was obtained by NBR 5752/2012 [24]. In order to evaluate the effect of the amorphous silica content on the RHA pozzolanic activity, we compared the properties of two RHA samples with high (>80%) and low (<20%) amorphous silica content. It is important for RHAs with high and low contents of amorphous silica to have the same average diameter in order to facilitate the evaluation and comparison of the microstructures of pastes. A suitable grinding procedure was formulated to produce RHA grains of less than 10 lm average diameter and approximately equal size. The RHAs with high and low content of amorphous silica (2.5 kg) were ground in a ball mill for 7 and 6 h, respectively, with a load of 40 steel spheres giving a total weight of 10.50 kg. Ordinary Portland Cement (OPC) without mineral additions (similar ASTM Type I), obtained directly from a cement factory in a single batch, was employed as reference. The chemical characterization of the RHA of high amorphous silica content and RHA of low amorphous silica content and of OPC was performed by ED XRF and Loss on Ignition analysis. The particle size distribution of the RHAs was measured as described above; their amorphous silica content of the RHAs was measured by a rapid analytical method developed by Payá et al. [25], and the Ca(OH)2 consumption of the RHAs was determined through the modified Chapelle method based on Raverdy et al. work [26]. For a material to be classified as pozzolanic, a minimum consumption of 330 mg CaO/g is required. 2.2. Methods for evaluating the microstructure of paste binders Three binders were used in the study: OPC (employed as the reference binder) and two composite binders – RHA with a high amorphous silica content and RHA with a low amorphous silica content – each of which replaced 20 mass% of the reference OPC. According to Ganesan et al. [27], replacement levels of up to 30 mass% are considered appropriate for the production of concrete without any adverse effects on properties such as strength and durability compared to those of pure cement. The binders were homogenised in a ceramic container. The compressive strength of the binders at 3, 7, 28, and 91 days was determined in mortars by NBR 7215/1996 [28] with a water to binder (w/binder) ratio of 0.48. This test was used to determine the compressive strength of binders. A 2 L planetary mixer was used for the preparation of pastes with a w/binder ratio of 0.50. This specific w/binder ratio was used since it is close to the w/binder ratio that is used for determining the compressive strength of mortars. The binders were mixed with water for 2 min at a low rotation speed; subsequently, there was a pause of 30 s, and then, high-speed mixing was performed for another 2 min. Cylinders with dimensions of 5 cm  10 cm were prepared in moulds, covered with plastic wrap, and immersed in water-saturated Ca(OH)2 for 91 days at 20 °C. The samples were then removed from the cylinders, and the hydration reaction was stopped by immersing the samples in acetone for 30 min and drying them in an oven at 100 °C for 1 h. Subsequently, the samples were packaged with soda lime and silica gel to prevent hydration and carbonation. The samples prepared for XRD and thermogravimetric (TG) tests at 3, 7, 28 and 91 days were cracked with a hammer and then milled in a porcelain crucible. TG experiments were performed in an instrument for simultaneous thermal analysis (DSC-TG, SDT 2960 model, TA Instruments). As portable samples, an aluminum crucible and an aluminum oxide crucible were used as references for analysis. All measurements were performed at a heating ratio of 20 °C/min, from 25 °C to 1000 °C under continuous N2 flux (110 cm3/min). For the SEM analysis, including X-ray microanalysis, two test samples were removed at 91 days for preparation with polished section. All samples analyzed by SEM were subjected to a prior procedure of deposition of a thin layer of conductive material (Au) on their surfaces. The Ca/Si (C/S) ratio of C–S–H was determined by a LEICA S440i Scanning Electron microscope. The different phases were identified using backscattered electron detectors (BSE). For semiquantitative microanalysis of the C/S ratio of C–S–H, a spectrometry by X-ray scattered energy (XSE) was performed with an OXFORD Link device with 20 kV voltage. The microanalyses were amplified to 8000x, and 25 microanalyses were randomly determined in regions of possible C–S–H presence for each sample analyzed. The microanalysis instrument was calibrated with pure cobalt standard, with a 100 s acquisition time for each microanalysis. The Mercury Intrusion Porosimetry was performed using a Quanta Chrome PoreMaster mercury intrusion porosimeter, with 0.485 N/m and 130° taken as the surface tension and the contact angle between mercury and pore wall, respectively. After 91 days of curing in a controlled-moisture room and drying at 45 °C until constant mass, samples of about 1 cm3 were cut from the mid-portion of the cylinders using a diamond cutter with 50 mm diameter and 100 mm height.

3. Results and discussion 3.1. Characterization of the residual RHAs produced in Brazil It is important to determine the characteristics and properties of the residual RHAs produced by different burning processes in the industry, in order to enable their use as mineral addition in

58

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

Table 1 Results of characterization tests of 10 RHAs milled for 5 h. Properties

Residuals RHAs

Specific mass (g/cm3) Chemical components (%)

Loss of ignition Magnesium oxide (MgO) Silicon dioxide (SiO2) Iron oxide (Fe2O3) Aluminum oxide (Al2O3) Calcium oxide (CaO) Total alkalis Sodium oxide (Na2O) Potassium oxide (K2O) Equiv. alkaline

Pozzolanic activity index (PAI) with cement (%) Laser granulometry

Average particle diameter (lm) Maximum particle diameter below which 10% of particles are found (lm) Maximum particle diameter below which 90% of particles are found (lm)

BET specific surface area (m2/g)

RHA 1

RHA 2

RHA 3

RHA 4

RHA 5

RHA 6

RHA 7

RHA 8

RHA 9

RHA 10

2.24

2.21

2.20

2.12

2.12

2.16

2.11

2.21

2.13

2.11

3.38 1.31 89.08 1.19 0.65 1.83 0.17 1.17 0.94

6.02 1.52 85.00 1.27 0.15 1.55 0.19 2.06 1.55

2.39 0.86 90.25 0.92 0.14 1.48 0.19 1.89 1.43

13.78 0.56 8.62 0.10 0.51 0.91 0.21 0.15 0.31

11.86 1.21 83.12 0.40 0.08 1.05 0.13 0.14 0.22

12.03 1.11 81.58 0.44 0.37 0.98 0.07 0.84 0.62

15.20 0.61 79.13 0.30 0.69 0.84 0.15 0.58 0.53

4.85 0.66 89.20 0.42 0.50 0.91 0.04 0.89 0.63

13.27 1.01 81.30 0.44 0.15 0.91 0.10 0.36 0.34

9.81 0.40 84.95 0.32 0.45 0.84 0.21 0.50 0.54

80.1

79.9

81.6

71.5

75.2

77.9

76.4

85.6

80.1

81.9

12.22 2.01

9.43 1.49

12.61 2.07

17.19 2.66

17.39 2.54

15.05 2.29

17.05 2.51

13.38 1.99

15.87 2.33

15.80 2.37

39.41

26.61

39.61

49.60

48.25

44.71

48.88

43.25

46.80

48.14

6.71

4.01

5.10

21.20

21.33

9.99

14.23

5.30

13.68

15.97

Fig. 1. Particle size distribution of RHAs after 5 h of milling.

cements. The results of the characterization of RHAs 1–10 milled for 5 h are shown in Table 1. The size distribution of the 10 RHA particles is shown in Fig. 1. The specific mass of residual RHA samples was between 2.10 g/cm3 and 2.30 g/cm3. The Loss on Ignition measurements showed lower values 15%. The residual RHAs showed high levels of silica, between 80% and 90%. The main impurities found were magnesium and potassium oxides, with values generally below 1.5% and 2.0%, respectively. The equivalent alkaline values were below 1.5%. The average diameter of the RHA particles was between 12 and 17 lm. The BET specific surface area of the residual RHAs ranged between 4 and 21 m2/g, showing a wide variation in the RHAs specific surface, which depends on the combustion process used. RHA burned at lower temperature showed a cellular and porous structure, with a high specific surface. When the burning temperature increases, packing of the cell structure leads to a decrease in the specific surface area [3,29]. The X-ray diffractograms of the 10 RHAs are shown in Fig. 2. Most RHAs produced by uncontrolled burning showed diffraction peaks denoting the presence of crystalline silica. The main crystalline silica phase detected was cristobalite. The residual RHAs milled for 5 h generally demonstrated pozzolanic activity index (PAI) with cement between 75% and 85%, which according to NBR 12653/ 2012 [30] classifies them as pozzolanic materials.

The selected RHA with high amorphous silica content was RHA 10, produced by a thermoelectric power plant with controlled burning process, in order to obtain an amorphous material, as confirmed by the diffractogram of Fig. 2. The selected RHA with low content of amorphous silica was RHA 8. The reasons for this choice were:  The results of the characterization tests for this RHA lies within the range of values of the other RHAs produced without control of the burning process, which makes RHA 8 representative of all the other RHAs.  RHA 8 showed the highest peaks in the X-ray diffractogram, indicating a more marked crystalline structure (Fig. 2).  A large amount of this residue is produced in thermoelectric power plants. According to the industry, around 30 tons/day of rice husk ash of low specific mass are produced, that require new disposal routes. The RHAs 10 and 8 were used, respectively, to represent RHA with high and low amorphous silica content, in the characterization of the microstructure of pastes containing 20% of RHAs, whose results are shown in Table 2, together with those for OPC. The RHAs 10 and 8 contained approximately 87% and 89% silica. The Loss on

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

59

Fig. 2. X-ray diffractograms of the ten RHAs: (a) RHA 1 (characteristic peaks of cristobalite), (b) RHA 2 (characteristic peaks of cristobalite), (c) RHA 3 (characteristic peaks of cristobalite), (d) RHA 4 (amorphous band with small amount of cristobalite), (e) RHA 5 (characteristic peaks of cristobalite), (f) RHA 6 (amorphous band with small cristobalite amount), (g) RHA 7 (amorphous band with small cristobalite amount), (h) RHA 8 (characteristic peaks of cristobalite), (i) 9 RHA (characteristic peaks of cristobalite), (j) RHA 10 (amorphous band with small amount of cristobalite).

60

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

Table 2 Results of characterization tests of RHAs 10 and 8 and OPC. Properties

RHA 10

RHA 8

OPC

Specific mass (g/cm3)

2.12

2.22

3.10

8.10 0.41 86.72 0.23 <0.01 0.70 0.13 1.27 0.97

5.26 0.40 88.88 0.59 <0.01 1.05 0.08 1.38 0.99

1.63 4.24 19.19 3.32 4.42 60.60 0.30 0.88 0.88

Chemical components (%)

Laser granulometry

Loss on ignition Magnesium oxide (MgO) Silicon dioxide (SiO2) Iron oxide (Fe2O3) Aluminum oxide (Al2O3) Calcium oxide (CaO) Total alkalis

Sodium oxide (Na2O) Potassium oxide (K2O) Equiv. alkaline

Average particle diameter (lm)

8.48

8.51



Content of amorphous silica (%)

81.30

17.66



Modified method Chapelle (mg/g CaO)

657

450



Fig. 3. Evolution of the compressive strength of mortars at 3, 7, 28 and 91 days.

Ignition test, which determines the amount of material not burned, yielded 8.00% and 5.50% for RHA 10 and 8, respectively. The mean diameters of the two RHAs were almost the same, around 8 lm. The content of amorphous silica in RHA 10 milled for 7 h was 81.30% and that in RHA 8 milled for 6 h was 17.66%. These results confirm the differences in the amorphous silica content between RHA 10 and RHA 8 suggested by the diffractograms in Fig. 2. The results of the modified Chapelle method shows that RHA 10 and 8 qualify for use as pozzolans. It can be noted that the RHA 8 with low amorphous silica content showed significant consumption of Ca(OH)2, indicating pozzolanic behavior. According to Cordeiro et al. [4], the increase of consumption of Ca(OH)2 in pastes with residual RHA has a strong correlation with the reduction of particle size of RHA and this behavior is explained by the increased external specific surface area. Rêgo et al. [21] corroborated this explanation and added that in the case of RHAs with a low content of amorphous silica, the reduction of the particle diameter by ultrafine grinding increases their amorphous silica content and changes the local structure of the silicon atoms, which in turn increases the consumption of Ca (OH)2. 3.2. Microstructure and properties of the paste binders 3.2.1. Compressive strength of mortars The compressive strength of the mortar is influenced by the synergic action of the hydration of the cement, the pozzolanic reaction,

and the microfiller effect of RHAs incorporated in binders [11]. The compressive strength of the mortars of the three binders measured at 3, 7, 28 and 91 days is shown in Fig. 3. In the case of RHA 10, the figure denotes an increase in compressive strength at all ages with respect to the reference mortar, reaching up to 29% for 20% RHA content after 91 days. The incorporation of 20% of RHA 8, despite of a decrease of compressive strength at 3 days compared to the reference mortar, led to an increase in compressive strength of approximately 9% at 91 days. The positive effect on the compressive strength of concrete was also observed by Sensale [13] and by Cordeiro et al. [16] who reported, respectively, a 6% and 8% increase in the compressive strength at 91 days in concretes with of 20% residual RHA, compared to the reference concrete. 3.2.2. X-ray diffraction (XRD) X-ray diffraction was used to identify the main crystalline compounds present in paste binders of different ages. The analysis was conducted for each binder at 4 different fixed ages, in order to evaluate the progress of hydration with time. Fig. 4 shows the X-ray diffraction patterns of pastes of 100% OPC, 20% RHA 10 and 20% RHA 8 with w/binder ratio = 0.50, after 1, 7, 28 and 91 days. The XRD patterns of pastes with 100% OPC showed no significant differences in the (Ca(OH)2) peaks with increasing hydration time beyond 7 days. A decrease in the characteristic peaks of portlandite, especially between 7 to 91 days, was observed in the X-ray patterns of pastes with 20% of RHA10, indicating pozzolanic

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

61

Fig. 4. X-ray diffractograms of paste binders with w/binder ratio = 0.50 after 1, 7, 28 and 91 days: (a) 100% OPC, (b) 20% RHA 10 (showing decreasing Ca(OH)2 peak with increasing age of hydration), (c) 20% RHA 8 (showing decreasing Ca(OH)2 peak with increasing age of hydration, as well as the characteristic peak of cristobalite).

62

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

Fig. 5. TG-DTG curve of pastes with 100% OPC and w/binder ratio of 0.50 that were subjected to 1 day of hydration.

Fig. 6. Evolution of the CH index of the binders (100% OPC, 20% RHA 10 and 20% RHA 8) with increasing age of hydration.

Table 3 C/S average ratio, standard deviation and confidence interval of C–S–H of binders pastes after 91 days of hydration. Binders

C/S average ratio

Standard deviation

Confidence interval

100% OPC 20% RHA 10 20% RHA 8

2.46 1.60 2.08

0.64 0.48 0.52

2.19–2.73 1.40–1.80 1.86–2.30

reaction between Ca(OH)2 and the amorphous silica present in RHA 10. An increase in the characteristic peak of Ca(OH)2 was observed in pastes with 20% of RHA 8 between 1 and 7 days, followed by a decrease of the same peak during the following 7 to 91 days period. This indicates that the pozzolanic reaction with Ca(OH)2 occurred to a greater extent during this period. Cristobalite was found in all pastes with 20% of RHA 8. No reduction of the peak intensity of cristobalite with increasing hydration time

was detected, indicating that crystalline silica remained unreacted during the hydration of the paste. The presence of the characteristic peak of cristobalite and the reduction of the Ca(OH)2 peak with increasing age of hydration in pastes with 20% residual RHA was also observed by Cordeiro [12]. 3.2.3. Thermogravimetric analysis The calcium hydroxide (CH) contents of pastes with 100% OPC and a w/binder ratio of 0.5 that were subjected to 1 day of hydration were obtained from the TG curve by using the differential thermogravimetric analysis (DTG) curve, as shown in Fig. 5. From the DTG curve, it is possible to determine the beginning and end of each step represented by the change in the slope of the TG curve. This approach is used to measure the percentage of mass loss for each jump characteristic of the curves obtained by TG-DTG. The mass loss corresponding to the pitch between approximately 420 °C and 550 °C in Portland cement pastes is

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

63

Fig. 7. (a) Scanning electron microscopy results of pastes with 100% OPC. The arrow indicates the point at which the X-ray microanalysis of C–S–H was performed. (b) X-ray microanalysis results of C–S–H of binder with 100% OPC at the indicated point.

due to the decomposition of Ca(OH)2. Between these temperatures, Ca(OH)2 decomposes to CaO and H2O. The mass loss reported by thermogravimetry is a result of volatilization of water. Thus, it is possible to determine the CH content of particular paste of hydrated Portland cement by using the values obtained by the TG curve and stoichiometric calculations. The content of calcium hydroxide in the paste binders, determined by TG, is indicative of the development of pozzolanic activity [31]. The CH index was obtained by dividing the percentage of Ca(OH)2 of the paste containing a specific binder by the percentage of Ca(OH)2 of a 100% OPC paste. The results are shown in Fig. 6. The figure highlights a strong logarithmic correlation (R2 = 0.94) between the CH index and the age of hydration for the paste with 20% of RHA 10. In the case of RHA 8 significant logarithmic

correlation (R2 = 0.86) between age of hydration and CH index also emerges. The CH index of pastes with 20% of RHAs 10 and 8 reached values of 33.81% and 44.82%, respectively, after 91 days, indicating the consumption of calcium hydroxide by the pozzolanic reaction of RHAs 10 and 8. Tuan et al. [32] measured a CH index of 35% in cement pastes with 20% substitution of amorphous RHA, after 91 days. 3.2.4. Scanning electron microscopy (SEM) with X-rays microanalysis The analysis of the C/S ratio of the C–S–H formed by hydration of the binders (Table 3) assists in the understanding of the formation process and of the morphology of the hydration products. Fig. 7 shows the SEM images of backscattered electrons of the cement paste with 100% OPC, indicating the point where it

64

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

Fig. 8. (a) Histogram frequency vs. C/S ratio of C–S–H of pastes with (a) 100% OPC, (b) 20% RHA 10, and (c) 20% RHA 8.

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

65

Fig. 9. (a) Scanning electron microscopy images of backscattered electrons of pastes with 20% RHA 10. The arrows indicate regions that contain unreacted particles of RHA 10. (b) X-ray microanalysis results of one of the regions indicated by arrows in (a) containing unreacted RHA 10.

determined the X-ray microanalysis of C–S–H and the X-ray microanalysis of the point indicated. The histograms of the C/S ratio of C–S–H for the 3 analyzed pastes are shown in Fig. 8, which also displays a standard normal distribution fit (solid line) to the observed data. Table 3 shows the sample mean and standard deviation of the C/S ratio data along with the 95% confidence interval for the mean C/S ratio based on a t-distribution with 24 degrees of freedom.

Binders with 100% OPC showed the highest C/S ratio (2.46) and standard deviation (0.64). Here, the 95% confidence interval for the mean C/S ratio of C–S–H was in the range of 2.19 and 2.73. In these pastes, C–S–H formation only occurred through OPC hydration. However, in pastes with 20% of RHA 10, containing a greater amount of amorphous silica, a significant decrease in the C/S ratio of C–S–H to about 1.60 was observed. A lower standard deviation was also observed for these pastes, compared to OPC. Here, the

66

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

Fig. 10. (a) Scanning electron microscopy images of backscattered electrons of pastes with 20% RHA 8. The arrows indicate regions that contain unreacted particles of RHA 8. (b) X-ray microanalysis results of one of the regions indicated by arrows in (a) containing unreacted RHA 8.

95% confidence interval for the mean C/S ratio of C–S–H was in the range of 1.40 and 1.80. Khan et al. [33], Uchikawa [34] and Yu et al. [35] also observed a decrease of the C/S ratio of C–S–H with the replacement of cement by RHAs with high amorphous silica content. Fig. 9 shows the results of the X-ray microanalysis of pastes with 20% RHA 10, in which the regions containing unreacted particles of RHA 10 are indicated by arrows. Fig. 9 also shows the X-ray microanalysis results of one of the regions indicated by the arrows. This microanalysis reveals the presence of silica in RHA 10. The RHA that does not chemically react with calcium hydroxide acts as microfiller in concrete and mortar [2].

In the pastes containing 20% of RHA 8, the formation of C–S–H with C/S ratio 2.08, intermediate between pastes with OPC (2.46) and pastes with 20% of RHA 10 (1.60), was observed. Here, the 95% confidence interval for the mean C/S ratio of C–S–H was in the range of 1.86 and 2.30. This behavior can be explained by the lower amount of amorphous silica formation compared to the RHA 10. Fig. 10 shows the X-ray microanalysis results of pastes with 20% RHA 8, in which the regions containing unreacted particles of RHA 8 are indicated by arrows. Fig. 10 also shows the X-ray microanalysis results of one of the regions indicated by the arrows. In this case, a large part of the silica was in crystalline form

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

Fig. 11. Total porosity (%) and porosity of pores larger than 5  10

(cristobalite), which does not contribute to the development of pozzolanic reaction, but acts by microfiller effect in the paste. The reduction in the standard deviation of the C/S ratio of C–S–H of pastes with 20% of RHAs 10 and 8 compared to the reference OPC paste indicated that the consumption of Ca(OH)2 by pozzolanic reaction of RHAs affected both the C–S–H formed by the hydration of Portland cement and the C–S–H formed by the pozzolanic reactions. According to Richardson [36], the decrease of the C/S ratio determines a change in the morphology of C–S–H, from fibrillar for high C/S ratio, to ‘‘foil-like’’ for lower ratios. The ‘‘foil-like’’ morphology of C–S–H is more efficient to fill the voids, decreasing the interconnected capillary pores and contributing to the refinement of the pore structure of the system.

3.2.5. Porosimetry by mercury intrusion The porous structure of cement pastes has great influence on their permeability and mechanical strength. Fig. 11 shows the total porosity (%) and that of pores larger than 5  10 2 lm (%) in each of the three pastes, with w/binder ratio = 0.50 and after 91 days of hydration. The total porosity increased in pastes with 20% replacement of both RHAs. However, a refinement of the porous structure of these pastes was observed. With 20% of RHAs 10 and 8, respectively, just 1.67% and 3.71% of the volume of the paste was comprised of pores greater than 5  10 2 lm, while this fraction increased to 12.86% in pastes with OPC. The pore structure refinement following OPC replacement by RHAs 10 and 8 is probably due to several synergic effects of the highly reactive pozzolan on the microstructure of the pastes: the consumption of calcium hydroxide to form C–S–H (pozzolanic effect), the filling of the pores by very fine pozzolan (microfiller effect), and the larger availability of nucleation sites for the growth of hydration products (effect of nucleation sites). These effects act together to reduce the pore size of the pastes. The change in the morphology of C–S–H with the reduction of the C/S ratio also contributes to this effect [36]. Cordeiro [12] observed refinement of pore structure in pastes with 20% replacement of cement by residual RHA at 90 days. Sensale [17] confirmed the efficiency of the residual RHA in improving the durability of concrete.

4. Conclusions Based on the present experimental results, the following conclusions can be drawn:

2

67

lm (%) in pastes with w/binder ratio = 0.50 after 91 days of hydration.

The 9 residuals RHAs produced in Brazil’s rice processing industries show low variations in chemical composition but large variations in the specific surface BET. These RHAs have silica content between 80% and 90%, and often contain significant amounts of crystalline silica. RHA 8 adequately represents residual RHAs with low content of amorphous silica. Pastes with 20% of RHA with high (>80%) and low (<20%) amorphous silica content show strong Ca(OH)2 consumption, due to the pozzolanic reaction. This behavior is due to both the reduction in Ca(OH)2 content in the pastes, and the reduction in the C/S ratio of the formed C–S–H. The replacement of OPC by RHAs with high and low amorphous silica content leads to a refinement in the pore structure of the paste. This microstructural change positively influences the mechanical strength and durability. RHAs with low amorphous silica content, when finely divided, show adequate pozzolanic activity, which makes them suitable for use as mineral addition. Acknowledgments The authors wish to acknowledge Furnas Centrais Elétricas, CAPES (Coordination of Improvement of Higher Education Staff), CNPq (National Council for Scientific and Technological Development), and ANEEL (National Electric Energy Agency) for financial support to the present research. References [1] Rêgo JHS. As cinzas de casca de arroz (CCAs) amorfas e cristalinas como adição mineral ao cimento – Aspectos da microestrutura das pastas [Amorphous and crystalline Rice husk ashes (RHAs) used as mineral admixtures to cement – Aspect of the paste microstructures] [Ph.D. thesis]. Postgraduate Program in Structures and Construction/University of Brasilia. Brasilia, Brazil; 2004. [2] Jamil A, Kaish ABMA, Raman SN, Zain MFM. Pozzolanic contribution of Rice husk ash in cementitious system. Constr Build Mater 2013;47:588–93. [3] Muthadi A, Kathandaraman S. Optimum production conditions for reactive rice husk ash. Mater Struct 2010;43(9):1303–15. [4] Cordeiro GC, Toledo Filho RD, Tavares LM, Fairbain EMR, Hempel S. Influence of particle size and specific surface area on the pozzolanic activity of residual rice husk ash. Cement Concr Compos 2011;33(5):529–34. [5] Xu W, Lo TY, Menon ESA. Microstructure and reactivity of rice husk ash. Constr Build Mater 2012;29:541–7. [6] Zerbino R, Giaccio G, Batic OR, Isaia GC. Alkali–silica reaction in mortars and concretes incorporating natural rice husk ash. Constr Build Mater 2012;36:796–806. [7] Antiohos SK, Papadaki VG, Tsima S. Rice husk ash (RHA) effectiveness in cement and concrete as a function of reactive silica and fineness. Cem Concr Res 2014;61–62:20–7.

68

J.H.S. Rêgo et al. / Construction and Building Materials 80 (2015) 56–68

[8] Isaia GC. Efeito de Misturas Binárias e Ternárias de Pozolanas em Concreto de Elevado Desempenho: Um estudo de Durabilidade com Vistas à Corrosão da Armadura [Effect of binary and ternary mixtures of Pozzolanas in concrete high performance: a study of durability with views in reinforcement corrosion] [Ph.D. thesis]. São Paulo, Brazil: Polytechnic University of São Paulo; 1995. [9] Ismail MS, Waliuddin AM. Effect of rice husk ash on high strength concrete. Constr Build Mater 1996;10(7):521–6. [10] Rêgo JHS. Viabilidade Técnica da Utilização da Cinza de Casca de arroz produzida sem controle da temperatura como adição mineral ao cimento [Technical Feasibility of using Rice Husk Ash produced without temperature control as mineral addition to the cement] [M.S. thesis]. Postgraduate Course in Civil Engineering/Federal University of Goiás. Goiânia, Brazil; 2001. [11] Isaia GC, Gastaldini ALG, Moraes R. Physical and pozzolanic action of mineral additions on the mechanical strength of high-performance concrete. Cement Concr Compos 2003;25(1):69–76. [12] Cordeiro GC. Utilização de cinzas ultrafinas do bagaço de cana-de-açúcar e da casca de arroz como aditivos minerais em concreto [Use of sugarcane bagasse and rice husk ashes as mineral admixture in concrete] [Ph.D. thesis]. Coordination of Programs Postgraduate Engineering (COPPE)/Federal University of Rio de Janeiro, Rio de Janeiro, Brazil; 2006. [13] Sensale GR. Strength development of concrete with rice-husk ash. Cement Concr Compos 2006;28(2):158–60. [14] Sensale GR, Ribeiro AB, Gonçalves A. Effects of RHA on autogenous shrinkage of Portland cement pastes. Cement Concr Compos 2008;30(10):892–7. [15] Cordeiro LNP. Análise da variação do Índice de amorfismo da cinza de casca de arroz sobre a atividade pozolânica. [Analysis of variation of the index of amorphization of the rice husk ash on the pozzolanic activity] [M.S. thesis]. Postgraduate Program in Civil Engineering, Porto Alegre, Brazil: Federal University of Rio Grande do Sul; 2009. [16] Cordeiro GC, Toledo Filho RD, Fairbairn EMR. Use of ultrafine rice husk ash with high-carbon content as pozzolan in high performance concrete. Mater Struct 2009;42(7):983–92. [17] Sensale GR. Effect of rice-husk ash on durability of cementitious materials. Cement Concr Compos 2010;32(9):718–25. [18] Zerbino R, Giaccio G, Isaia GC. Concrete incorporation Rice-husk ash without processing. Constr Build Mater 2011;25:371–8. [19] Chatveera B, Lertwattanaruk P. Evaluation of nitric and acetic acid resistance of cement mortars containing high-volume black rice husk ash. J Environ Manage 2014;133:365–73. [20] Rahamn ME, Muntohar AS, Pakrashi V, Nagaratnam BH, Sujan D. Self compacting concrete from uncontrolled burning of rice husk and blended fine aggregate. Mater Des 2014;55:410–5. [21] Rêgo JHS, Nepomuceno AA, Figueiredo EP, Hasparyk NP, Borges LD. Effect of particle size of residual rice husk ash (RHA) in consumption of Ca(OH)2. J Mater Civil Eng 2014, http://dx.doi.org/10.1061/(ASCE)MT.1943-5533. 0001136.

[22] ASOCIACIÓN MERCOSUR DE NORMALIZACIÓN (AMN). Agregado graúdo – Ensaio de abrasão ‘‘Los Ángeles’’ (Small-size coarse aggregate – Test method for resistance to degradation by Los Angeles machine). NBR NM 51. Rio de Janeiro, Brazil; 2001. [23] AMN. Cimento portland - Determinação e massa específica (Portland cement – determination of density). NBR NM 23. Rio de Janeiro, Brazil; 2000. [24] ASSOCIAÇÃO BRASILEIRA DE NORMAS TÉCNICAS (ABNT). Materiais Pozolânicos – Determinação de atividade pozolânica com cimento Portland – Índice de atividade pozolânica com cimento Portland – Método de ensaio. (Pozzolans – determination of pozzolanic activity with Portland cement — Test method). NBR 5752. Rio de Janeiro, Brazil; 2012. [25] Payá J, Monza J, Borrachero MV, Mellade A, Ordonez LM. Determination of amorphous silica in rice husk ash by a rapid analytical method. Cem Concr Res 2001;31(2):227–31. [26] Raverdy M, Brivot M, Paillere AM, Bron R. Appreciátion de l’activite pouzzolanique des constituants secondaires [Assessment of activity pozzolanic of the secondary constituents]. In: Proceedings of 7th int. congress on the chemistry of cement. Paris, France; 1980. p. 36–41. [27] Ganesan K, Rajagopal K, Thangavel K. Rice husk ash blended cement: assessment of optimal level of replacement for strength and permeability properties of concrete. Constr Build Mater 2008;22:1675–83. [28] ABNT. Cimentos Portland – Determinação da resistência à compressão – Método de Ensaio (Portland cement – determination of compressive strength). NBR 7215. Rio de Janeiro, Brazil; 1996. [29] Bartha P, Huppertz EA. Structure and crystallization of silica in rice husk. In: Proceedings of international conference on the utilization of rice by-products. IATA. Valencia, Spain; 1977. p. 89–98. [30] ABNT. Materiais Pozolânicos – Requisitos (Pozzolans — requirements). NBR 12653. Rio de Janeiro, Brazil; 2012. [31] Subalakshimi P, Sarrawathy V, Sivasankaran SK. Hydraulicity of mineral admixtures in cement blends. Constr Build Mater 2012;36:648–53. [32] Tuan NV, Ye G, Breugel K, Copuroglu O. Hydration and microstructure of ultra high performance concrete incorporating rice husk ash. Cem Concr Res 2011;41(11):1104–11. [33] Khan MH, Mohan K, Taylor HFW. Pastes of tricalcium silicate with rice husk ash. Cem Concr Res 1985;15(1):89–92. [34] Uchikawa, H. Effect of blending components on hydration and structure formation. In: Proceedings of 8th international congress on cement chemistry, vol. 1, Rio de Janeiro; 1986. p. 249–280. [35] Yu Q, Sawayama K, Sugita S, Isojima Y. The reaction between rice husk ash and Ca(OH)2 solution, and the nature of its product. Cem Concr Res 1999;29(1):37–43. [36] Richardson IG. The nature of C–S–H in hardened cements. Cem Concr Res 1999;29(8):1131–47.