1、 Hydration and properties of novel blended cements based on cement kiln dust and blast furnace slag Maria S. Konsta-Gdoutos , 1, , Surendra P. Shah Center for ACBM, Department of Civil Engineering, Northwestern University, Evanston IL 60208, USA http:/dx.doi.org/10.1016/S0008-8846(03)00061-9, How to
2、 Cite or Link Using DOI Permissions Hydration; Compressive strength; Cement kiln dust; GGBFS 1. Introduction The environmental concerns related to Portland cement production, emission and disposal of cement kiln dust (CKD), is becoming progressively significant. CKD is a fine-grained, particulate ma
3、terial readily entrained in the combustion gases moving through the kiln. It is composed primarily of variable mixtures of calcined and uncalcined feed materials, fine cement clinker, fuel combustion by-products, and condensed alkali compounds 1. The generation of CKD is responsible for a significan
4、t financial loss to the cement industry in terms of the value of raw materials, processing, energy usage, and dust collection and disposal. Cement manufacturing plants generate approximately 30 million tons of CKD worldwide per year 2. The U.S. cement industry generates about 4.1 million tons of CKD
5、 every year, 3.3 million tons of which is landfilled and only 0.75 million tons enter commerce as by -products 3. The relatively high alkaline content of CKD is the predominant factor preventing its recycling in cement manufacturing. All CKDs frequently contain alkalis (Na2O, K2O) and sulfates in mu
6、ch higher percentages than those in Portland cement. CKD often contains a large amount of free lime, thus making it a substitute for fertilizers and lime in stabilizing wastewater streams. Furthermore, studies have shown that CKD can be effectively used in soil and sludge stabilization. It has also
7、been successfully used as inorganic filler in bituminous paving and asphaltic roofing 4. Because the characteristics of CKD vary from plant to plant, only a small amount of CKD (15% cement replacement) is used in the cement and concrete industry. However, its high alkali and sulfate content makes it
8、 an excellent activator for pozzolanic materials. The dissolution rate of materials with latent pozzolanic properties such as blast furnace slag generally depends on the alkali concentration of the reacting system5. Latent hydraulic materials develop pozzolanic activity and act as hydraulic cements
9、once their glass network disintegrates when attacked by OH ions. The solubilities of Si, Ca, Al, and Mg are functions of the pH. At a pH lower than 11.5, the equilibrium solubility of silica is low and slag does not dissolve. As a result, more Ca+2 and Mg+2 enter into the solution and an impermeable
10、 aluminosilicate coating covers the surfaces of the slag grains, which inhibits further hydration 6. A chemical activator is required for further hydration of slag. Activators generally include all alkali hydroxides and salts, with the least soluble salts being the most effective 7 and 8. Because al
11、l salts are neutral solutions they exhibit only a weak activating effect; however, when they are combined with lime they undergo an exchange reaction with Ca+2 and hydroxide ions. Solutions of sodium or potassium hydroxides are then formed, which attack and disintegrate the aluminosilicate glass. Th
12、e presence of sulfate ions, supplied either by gypsum or alkali sulfates, accelerates the dissolution of slag by reducing the concentrations of Ca+2 and Al+3 in the mix to form ettringite. Limited information on the pertinence of CKD as an activator for pozzolanic materials is available. Sprouse 9ap
13、plied and received a patent for a cement that combined ground blast furnace slag with CKD. Amin et al. 10studied the effect of kiln dust content and the calcination temperature on the hydration of granulated slag. Hydration products such as CSH, C4AH13, C2AHX, Ca(OH)2, hydrogarnet, and calcite were
14、identified using the DTA technique. Their results indicated that the crystallinity and the content of the cementitious phases, especially -C2S, increased with calcination temperature. They concluded that activation of slags increases with increasing the content and the calcination temperature of the
15、 CKD. El-Didamony et al. 11 evaluated the effect of washed and calcined kiln dust with anhydrite. They found that the activation of the slag increases with firing temperature of the kiln dust and the amount of added anhydrite. Kiln dust calcined at 1300 C with 15% anhydrite was found to be suitable
16、for the production of supersulfated cement. Recently, the production of active -C2S, C12A7 and C2AS cements using different proportions of a low-CaO fly ash and a lime-rich CKD as raw materials was proposed 12. The objective of the present work is to understand the interaction of CKD with slag in or
17、der to explore the feasibility and the approaches to development of a new generation of more durable CKD-activated slag blends. Ground granulated blast furnace slag (GGBFS) has successfully been used with Portland cement to produce high-performance cement blends that are more economical and environm
18、entally friendly. The use and effectiveness of CKD as an activator for slag depends upon its physical and chemical characteristics, most importantly, the alkali and free lime content, and the amount of carbonates and sulfates. The effectiveness of the alkali activation of slag will depend on the alk
19、alinity provided by the CKD. It is expected that the high free lime content of the CKD will improve the hydration process by accelerating hydration and forming more crystalline products of hydration. Sulfate ions provided either by alkali salts or anhydrite will expedite the hydration process and ac
20、celerate the pozzolanic reaction through the formation of ettringite. The experimental approach in this work includes the characterization of binary blends containing 50% slag and 50% CKD in terms of the rates of heat evolution and strength development, hydration products, and time of initial settin
21、g. A study of the effects of the influencing factors of four different CKDs in terms of soluble alkali content, particle size, and free lime content was undertaken. The results indicate the dependence of the dissolution rate of slag on the alkalinity of the system and the importance of the optimum l
22、ime content on the rate of strength development. 2. Materials Four different types of CKDs and a granulated blast furnace slag were selected for this study. The chemical composition and physical properties of the materials are given in Table 1. Compositions of the individual components were determin
23、ed by X-ray fluorescence (XRF) spectroscopy. The slag meets the classification requirements of ASTM C 989 for Grade 100. The specific gravity is 2.90 and the Blaine specific surface is 4460 cm2/g. Results from X-ray diffraction (XRD) analysis (Fig. 1) indicated that the slag consisted mainly of a gl
24、assy phase. The small sharp peaks are due to traces of crystalline phases of merwinite (C3MS2) and akermanite (C2MS2). Table 1. Results of oxide analyses of CKD and GGBFS by XRF Compound (wt.%) CKD (E) CKD (P) CKD (A) CKD (X) GGBS Gr 100 SiO2 14.67 15.7 11.5 8.96 31.96 Al2O3 5.06 5.061 4.68 2.667 10
25、.31 Fe2O3 3.46 2.58 2.04 2.21 1.42 CaO 56.99 45.5 50.2 48.6 45.98 MgO 1.04 2.33 1.34 3.141 7.02 SO3 8.45 5.68 16.7 7.23 2.13 K2O 6.073 5.33 5.86 6.81 0.31 Na2O 0.6 0.89 1.015 0.54 0.26 Alkali equivalent 4.60 4.40 4.871 5.021 0.38 Water-soluble alkali equivalenta 2.25 1.69 2.59 3.30 NA P2O5 0.102 0.0
26、77 0.085 0.02 Cl 0.532 1.802 0.732 1.4 fCaO 8 5.1 13.9 26.9 LOI 15.10 25.5 9.56 17.92 0.2 a Measured by atomic absorption. Table options Fig. 1. XRD of the GGBFS used. Figure options Results from the XRD analysis of the four CKD samples are shown in Fig. 2. Calcite (CaCO3) was identified as the prev
27、ailing phase for all four samples. The CKD samples, identified as E, P, A, and X, were representative of various plant operating conditions affecting dust composition and reactivity, such as the feed raw materials, kiln type, and dust -collection systems. CKDs (E) and (P) were generated from wet pro
28、cess kilns and were collected by electrostatic precipitators and baghouses, respectively. They contained large amounts of calcite, expressed as high LOI, and small amounts of free lime. CKDs exhibiting high levels of carbonation and lacking free lime are usually stockpiled and considered less reacti
29、ve, since the calcite is essentially nonreactive. CKD (P) also contained traces of dolomite CaMg(CO3)2. Dolomitic dusts are characterized by higher MgO contents, which are usually considered more slowly reactive. Fig. 2. XRD of the four raw cement kiln dusts P, E, A, and X. CC=calcite, C=lime, D=dol
30、omite, Q=quartz, An=anhydrite, 1=alkali salts (apthitalite, arcanite), S=sylvite. Figure options CKD (X) is also a dolomitic dust obtained from a dry process kiln with a precalciner and a preheater, and collected by a baghouse. CKD (A) was removed from a long dry kiln with electrostatic precipitator
31、s. Both CKDs contained small amounts of calcite and high amounts of free lime indicating that more Ca+2 will be available for a pozzolanic reaction. CKD (A) contained significant amounts of anhydrite. Traces of portlandite were identified in CKD (X) implying that the CKD has been exposed to moisture
32、. Alkali sulfate salts were present in all four CKDs. Aphtithalite K3Na(SO4)2 was identified in CKDs (X) and (E), while CKD (E) and (A) contained arcanite (K2SO4). CKDs (P), (A), and (X) also contained sylvite (KCl). An important physical characteristic of the CKDs is their particle size, which is s
33、hown in Fig. 3. The finest material used is CKD (A) with a specific surface of 8270 cm 2/g and a mean equivalent spherical particle diameter of 9 m. CKD (A) is also the most uniform: 57% fall into the narrow range of 110 m and 93% are smaller than 45 m. CKD (E) is the coarsest one with a mean diamet
34、er of 60 m and is evenly divided between 45 and 100 m. In CKDs (P) and (X), 75% of their particles are in the range between 1 and 45 m. They both have a very similar particle size distribution, with a mean particle size of approximately 18 m, and specific surfaces of 5410 and 6140 cm2/g, respectivel
35、y. Fig. 3. Particle size distribution of the materials used. Figure options 3. Experimental program 3.1. Heat of hydration The heat of hydration of CKDslag pastes was measured using an adiabatic DSS Qdrum calorimeter. Blended material and deionized water were mixed according to ASTM C 305 using a Ho
36、bart mixer. After mixing, the paste was cast in a 24 in. (50100 mm) plastic cylinder mold. A temperature probe inserted into the cylinder reco rded the adiabatic heat of hydration for 72 h. The time of initial setting of the CKDslag pastes was determined using a Vicat apparatus according to ASTM C19
37、1. 3.2. Compressive strength The compressive strength of cylindrical mortar specimens of 36 in. (75150 mm) was determined according to the ASTM C 39-94 at 7, 28, and 56 days, and was calculated as the average of three specimens with consistent results. Mortars were mixed in accordance with EN 196-1:
38、1987 with 450 g of binder blend, 1350 g of natural sand, and 225 g of water. After demoulding, specimens were cured on shelves in a moist curing room until testing. All test specimens were capped with a high-sulfur capping compound before testing, according to ASTM C 617. 3.3. X-ray diffraction XRD
39、was employed to identify the phases present in hydration products, using standard monochromatic CuK radiation and operating at 40 kV and 20 mA. Scanning was performed at 0.05 2 step size, at 2 s per step, between 5 and 50. Samples were prepared by crushing paste specimens into fine powder. To stop h
40、ydration at the desired ages, the crushed materials were slurried with 200-proof ethanol. The powder was vacuum filtered, air-dried, and reground before mounting in the XRD sample holders. 4. Results and discussion 4.1. Time of setting The time at which the needle penetrates 25 mm into the pastes at
41、 room temperature was taken to define the initial setting, following ASTM 191. The time of initial setting for all blends is shown in Fig. 4. OPC reached initial set in 4.5 h, whereas for the OPCslag blend an increase was observed and, as expected, time of setting was extended for 50 min. Time of se
42、tting was much shorter than the control blends for the blend with CKD (A), which may be due to the formation of ettringite; CKD (A) contains a very high amount of sulfate (17%). Significant retardation was observed for the slag blends activated with CKDs (E), (P), and (X). Initial setting time for t
43、he blends with CKD (P) and (X) were very similar, around 7 h and 30 min for the CKD (P)slag blend, and almost 9 h for the blend with CKD (X). Slag activated with CKD (E) took more than 14 h to reach the initial set, which coincides with the end of the dormant period and the beginning of the accelera
44、tion period of the rate of heat evolution curve. However, a gradual stiffening of the paste and a change in rigidity was observed after the first 5 h of curing for all three blends activated with CKD (E), (P), and (X), which is believed to be caused by the formation of AFm phases throughout the fresh paste. Fig. 4. Comparison of the time of initial set between the slag blends activated with CKDs A, P, E, and X and the control OPCslag blend. Figure options
