Use of recycled glass as a raw material in the manufacture of Portland cement

Materials and Structures/Matériaux et Constructions, Vol. 35, September-October 2002, pp 510-515 Use of recycled glass as a raw material in the manufacture of Portland cement Z. Xie and Y. Xi Department of Civil, Environmental and Architectural Engineering, University of Colorado at Boulder, CO 80309, USA Paper received: February 15, 2001; Paper accepted: April 25, 2002 A B S T R A C T Scrap glass is a solid waste from daily recycling. Most of the waste glass is sodium-lime-silicate glass which has, more or less, similar chemical compositions to clay, a raw material in cement manufacturing. Therefore, we utilize the solid waste in cement raw mix by replacing part of the clayey component. In this study, the effects of the glass in cement raw mix on clinker burning were investigated. The experimental results show that the addition of the glass into cement raw mix (1) results in the formation of more liquid phase between 950 C to 1250 C compared with conventional raw meals; (2) decreases C 3 S content in the clinker; and (3) increases NC 8 A 3 content, which leads to flash setting and poor strength development of the cement. Therefore, it is necessary to increase the SG value [SG = SO 3 100% /(1.292 K 2 O + 0.85 Na 2 O)] of the clinker when the glass is present in the raw mix. R É S U M É Le verre est l un des grands déchets de la vie quotidienne. Une analyse des déchets de verre révèle que ceux-ci sont constitués essentiellement de silicate de soude et de calcium. Dans nos essais, des déchets de verre sont mélangés à la pâte de ciment en se substituant à sa composante argileuse, et l'effet de leur présence dans le ciment sur le brûlage dans le clinker a été examiné. Les essais montrent que l'adjonction de verre dans la pâte de ciment stimule la formation de la phase liquide entre 950 C et 1 250 C par comparaison avec à sa forme habituelle, diminue le taux de C 3 S dans le clinker et augmente celui de NC 8 A 3 qui engendre une prise rapide et un faible développement de la résistance mécanique du ciment. Il faudrait augmenter de façon appropriée le SG [SG = SO 3 100% /(1,292 K 2 O + 0,85 Na 2 O)] du clinker en présence de verre. 1. INTRODUCTION Currently, only a small fraction of the post-consumer glass is recycled directly to the primary market - the bottling and container industry. The problem is particularly serious in major metropolitan areas. The total amount of waste glass is projected to be doubled within about three years. The literature review has shown that there are many potential secondary uses of waste glass. The most important ones are glasphalt, fiberglass, clean fill, and drainage. Analysis indicates that most of waste glass from daily recycling is sodium-lime-silicate glass which has, more or less, similar chemical compositions to clay, a raw component in cement manufacturing. Therefore, it must be viable to use the waste glass in cement raw mix by replacing a small portion of the clayey component. However, our literature review shows that there has not been any attempt to re-utilize the waste glass in cement manufacturing. The most important factor that must be considered in the use of the glass as a raw material for the cement manufacturing is the alkali components in the glass, which result in an increase of alkalis, especially Na 2 O, in the cement. It is well known that alkalis have various adverse effects on the production and applications of the cement. This may be the reason as to why the waste glass has not been used in cement production. Compared with other recycling techniques, using the waste glass as one of the raw materials for making Portland cement has some advantages: 1359-5997/02 RILEM 510

Xie, Xi No major investment is required, only some modifications on the current design of raw cement mix are needed; Due to the large amount of consumption of portland cement in the construction industry, the potential for the re-utilization of the waste glass is unlimited; All sizes of the crushed glass including very fine particles can be used as the raw mix for cement manufacturing since the raw materials must be grounded into fine particles for cement manufacturing anyway; The total energy for cement production will be reduced by the reduced amount of energy needed to burn off the water in the clay. 2. EFFECTS OF ALKALIS ON PROPERTIES OF PORTLAND CEMENT Usually, about 50% of alkalis present in the raw feed in cement kilns are volatilized between 800 C and 1000 C [7]. The volatilized alkalis partially condense in the cooler parts of the kiln and return to the high temperature zone. The enrichment of alkalis in the kiln system leads to the formation of rings and a coating on the lining of the kiln and preheaters, thus causing shutdown of the system. A direct linear relationship between the density of the alkali-containing raw materials and the extent of volatilization has been observed [2, 11]. It appears that in the clinker SO 3 makes the most of demand on the alkalis [5]. The alkali sulfates most commonly formed are: arcanite (K 2 SO 4 ), aphthitalite (Na 2 SO 4 3 K 2 SO 4 ), and calcium langbeinite (2 CaSO 4 K 2 SO 4 ). It has been reported that the addition of gypsum in alkali containing raw materials has a positive effect on the formation of clinker minerals, and that the presence of alkali sulfates results in well developed alite and belite crystals [2]. It is also known from the literature that aluminates and ferrites accommodate about half or more of the available alkalis [6]. Free alkali reacts with C 3 A to form Na 2 O 8CaO 3Al 2 O 3 (NC 8 A 3 ) and free lime [9]. NC 8 A 3, with its high reactivity, could cause setting problems to the cement [5]. Alkalis enter the structure of silicates to form KC 23 S 12 and NC 23 S 12, both are kinds of C 2 S stabilized by alkalis and hardly form further into C 3 S [3]. When alkali-containing cement is mixed with water, the alkali metal ions readily go into the liquid phase of the hydrating system and affect the rate of cement hydration. This, in turn, affects the strength and other engineering properties of the hardened cement paste. It was found that most of the alkali cations remained in the pore solution, only a small portion is incorporated in the solid hydration products [2]. The ph value of the pore solution in a high alkali cement paste may reach 12.9 within two minutes, and 13.7 or even higher after 28 days [4]. The presence of alkali ions depresses the solubility of Ca 2+. The decreased solubility of Ca(OH) 2 varies the rate of nucleation and crystallization of hydration products, and therefore alters the setting as well as hardening processes of the cement. A high alkali level in Portland cement (e.g. over 0.8 % Na 2 O) affects early strength of the cement paste, especially when the alkali presents as alkali sulfates in the cement. In this case, an increase by about 10% in early age strength is associated with a decrease by about 10-15% in 28 day strength [5]. According to the experience of the German cement industry, by controlling the SG value, which is expressed as: SG = 100% SO 3 /(1.292 K 2 O + 0.85 Na 2 O), the cement with a high alkali content could attain normal properties. For the Na 2 O-riched clinker, the proper range of the SG value should be from 90% to 100%; for K 2 O-riched clinker, the range should be from 60% to 70% [1]. In addition to the effect of high alkali on the early strength development of concrete, there is another major adverse effect from using high alkali cements, that is, the damage due to the so-called alkali-aggregate reaction (AAR) in matured concrete, which is a long-term durability problem. AAR has been a research topic in the concrete industry for long time, and it will not be discussed here in detail. From above review, it is clear that high alkalis in the raw mix may cause problems in cement manufacturing, and high alkalis in Portland cement may lead to various adverse effects on the properties of the cement and the concrete made of the cement. Therefore, a great caution must be taken when using the waste glass as one of the raw components. In the present study, we systematically investigated the methods that can be used to avoid the adverse effects of the glass on properties of cement as well as on the formation of cement clinker, and find the maximum rate to replace the clay by the waste glass without any major impact on the resulting Portland cement. 3. EXPERIMENTAL RESULTS AND DISCUSSION 3.1 Preparation of cement clinkers All of the raw materials used in the present study were from a local cement plant in Colorado. The results of chemical analysis for the raw materials are given in Table 1. The raw materials were separately ground in a jar mill into powders of fineness below 75 µm. The raw materials were then mixed with or without the waste glass according to given formulas. Cement raw meals used in the present study have similar chemical compositions as those commonly used in the Portland cement plants. The raw meals were then mixed with a small amount of water and pressed into tablets of 25.4 mm in diameter and 4 mm in height. After dried in an oven under 80 C, the tablets were put into a chamber furnace and burnt at different temperatures. After burnt for a half hour, the clinkers were taken out of the furnace and cooled down in the air. 511

Materials and Structures/Matériaux et Constructions, Vol. 35, September-October 2002 Table 1 Chemical analyses of the materials used in the study SiO 2 Al 2 O 3 Fe 2 O 3 CaO MgO SO 3 Na 2 O K 2 O LOI Total Limestone 6.32 1.55 0.89 49.45 0.48 0.30 0.04 0.34 35.45 94.82 Sandstone 82.76 3.39 0.96 2.99 0.26 0.25 0 0.17 9.05 99.76 Low grade limestone 16.97 5.96 1.77 39.27 0.90 0.39 0 1.12 33.64 100.02 Iron ore 2.54 0.58 97.93 1.09 0 0.15 0 0.02 0 102.31 Glass 69.36 3.46 2.91 11.69 0.83 0 11.16 0.53 0.85 100.79 could be satisfied. The other factor is that the glass has high content of Na 2 O. If a raw mix contains 10 percent of glass, it would take into the clinker as high as about 1.8 percent of Na 2 O. Table 2 Chemical compositions of the clinkers studied Weight ratios of raw materials in formulation Calculated oxide composition of the clinkers (%) Sand- Iron Low grade Lime-stone glass SiO 2 Al 2 O 3 Fe 2 O 3 CaO MgO SO 3 Na 2 O K 2 O Sum stone Ore lime-stone #1 6.2 2.0 63.29 77.14 0 20.80 5.19 3.83 63.21 0.96 0.50 0.03 0.98 95.50 #2 3 2 62.53 76.68 4 20.77 5.17 3.89 63.06 0.97 0.48 0.44 0.96 95.74 #3 0 2 63.29 74.36 7 20.35 5.18 3.94 62.47 0.98 0.47 0.78 1.00 95.17 #4 0 2 52.74 82.11 10 21.13 4.77 3.91 62.50 0.95 0.46 1.12 0.92 95.76 Table 2 (continued) Chemical compositions of the clinkers studied Calculated Modulus of Clinker Potential Mineral Composition (Bogue Method) LSF SM IM C 3 S C 2 S C 3 A C 4 AF #1 0.956 2.31 1.36 57.39 11.67 7.28 11.64 #2 0.955 2.29 1.33 57.17 11.89 7.13 11.82 #3 0.963 2.23 1.31 57.81 10.2 7.07 11.98 #4 0.940 2.43 1.22 54.83 15.37 6.03 11.88 Table 3 Free lime contents in the clinkers (%) Formula 1350 C 1400 C 1450 C #1 0.22 0.17 0.17 #2 0.13 0.17 0.00 #3 0.17 0.13 0.26 #4 0.22 0.00 0.00 3.2 Glass contents and clinker compositions A typical cement raw mix is composed of limestone, clayey materials and iron materials. Two factors limit the maximum substitution of clayey material by glass. One is that the scrap glass from recycling usually has an approximately composition of SiO 2 :Na 2 O:CaO = 75.0 : 12.5 : 12.5, while the natural clayey materials have approximate composition of SiO 2 : Al 2 O 3 : other oxides = 65 : 20 : 15 (on ignition basis). It can be seen that the glass is characterized by high SiO 2, high Na 2 O and low Al 2 O 3 contents. When too much glass is added, Al 2 O 3 content would be low although SiO 2 content in the raw mix Therefore, the glass can only be added into cement raw mixes in low percentages. With the raw materials available in our laboratory, the calculated results based on our formulation are given in Table 2, in which four different clinkers named formulas #1, #2, #3, and #4 are listed. Formula #1 represents conventional raw meal without any addition of the glass. Formulas #2, #3, and #4 are the mixes with various amounts of the glass. In the formula #2, a part of the sandstone was replaced by the glass, while in the formulas #3 and #4, all sandstone was replaced by two different amounts of glass. The potential mineral compositions of the four formulas calculated by Bogue method are also included in Table 2. 3.3 The effect of the waste glass on the formation of clinker In the present study, raw meals corresponding to formula #1, #2 and #3 were burnt at 1300 C, 1350 C and 1400 C, respectively. The clinkers obtained are referred to as 1-1300, 1-1350, 1-1400, 2-1300, and so on. The clinkers were then tested by XRD analysis. The parameters of the XRD were: Cu K-a radiation, operated at 40 kv and 25 ma, 0.02 degree per step. We took the peak at 2θ = 29.45 (d = 3.03) as the characteristic peak of C 3 S. The heights (cps) of the peaks are summarized in Fig. 1 and listed also in Table 5. From Fig. 1 the following tendencies can be observed: 1. When the glass is added in the raw meals, the intensities of the characteristic peak for C 3 S in all clinkers of the 2-series and the 3-series are decreased. This indicates a decrease of C 3 S formation in the clinkers after the glass is added. 2. With increasing burning temperature, the amounts of C 3 S formed in the clinkers of the 2-series and the 3-512

Xie, Xi Fig. 1 Comparison of the heights of XRD peaks of C 3 S in the clinkers. Table 4 Setting times and compressive strengths of the cements Cement from SO 3 W/C Setting times (hr:min) Compressive strength (MPa) clinker (%) (%) Initial Final 3 days 7 days 28 days 1-1400 2.1 32 0:55 1:46 27.4 36.9 52.7 2-1400 2.1 38 0:04 -- 9.5 13.4 19.3 3-1400 2.1 38 0:02 -- 8.2 11.2 16.8 min., the samples were taken out of the furnace and measured for the change in the height. In principle, when liquid phase appears in the conoids, the solid particles in the mix will get compacted due to surface tension of the liquid phase, which results in a macroscopic shrinkage of the testing specimens. Therefore, the linear shrinkage of the conoids can be used as an indicator to show if there is any liquid phase in the mix or not. The experimental results are shown in Fig. 2, in which the shrinkage behaviors of the specimens can be analyzed in three different temperature ranges. In the temperature range below 950 C, all specimens with and without the glass expand at about the same magnitude. Expansion is shown as positive in Fig. 2, which implies that there is no liquid phase in the specimens. In the temperature range above 950 C, all specimens start to shrink, which is an indication for the appearance of liquid phases. Comparing the four different specimens in Fig. 2, one can see that, from 950 C to 1250 C, the shrinkages of #2, #3 and #4 (with the glass) increase much faster than the increase of #1 (without the Table 5 Quantity of C 3 S formed at different temperatures (Expressed in cps) 1300 C 1350 C 1400 C #1 55.41 80 62 #2 30.19 51.5 51.6 #3 37.34 38.63 54.15 series increase, but the amounts are still less than those in the 1-series burnt at the same temperatures. Therefore, in order to increase C 3 S content in clinker, it is necessary to increase the burning temperature for the raw meals with the glass. 3. The C 3 S contents in the clinkers shown in Fig. 1 are comparable with the calculated amounts based on Bogue method as shown in Table 2. The C 3 S content of #1 (without the glass) is higher than the value predicted by Bogue method, while those of #2 and #3 are lower. 3.4 The effect of glass on the eutectic of the raw mix Since most of sodium-lime glass has a melting point below 1000 C, it was expected that the addition of the waste glass into cement raw mix would lead to a reduced eutectic of the materials. In order to examine the effect of waste glass on the eutectic of the system, the linear shrinkage of the conoid made of raw meals was tested after burnt under different temperatures. The raw meal was cast into conoids of 15.2 mm in bottom diameter and about 38 mm in height. After burnt under various temperatures ranging from 600 C to 1300 C for 15 Fig. 2 Linear length change of the conoids after burning. glass). This suggests that, in this temperature range, liquid phases appear earlier and chemical reactions accelerate in the specimens with the glass. Above 1250 C, the shrinkage of #1 (without glass) increases sharply and the final shrinkage of #1 is even higher than other specimens with the glass. This means that, in this temperature range, the clinker #1 has higher rate of liquid phase formation and chemical reactions. 513

Materials and Structures/Matériaux et Constructions, Vol. 35, September-October 2002 meals based on formula #3 were burnt under 1300 C, 1350 C and 1400 C for 30 min., respectively, and then analyzed for loss of alkalis by an atomic absorption spectrometer. The results are shown in Fig. 3. It is evident from Fig. 3 that the Na 2 O coming from the glass has high volatility. At the normal burning temperature above 1400 C for cement manufacturing, it is expected that the evaporation rate of Na 2 O will excess 40%. It is also noticed in Fig. 3 that the K 2 O, brought into the raw meals by other raw materials, has even higher volatility. The volatilization of alkalis increases with burning temperature, but no much difference at temperatures above 1350 C. Fig. 3 Volatility of alkalis in the raw meals containing the glass. 3.5 The effect of alkalis in the glass on properties of flux As observed earlier, the addition of the glass results in a decrease of C 3 S in clinkers, which makes calcium oxide over supplied. The redundant CaO remains in clinkers as free lime. If C 3 S decreases by 10 percent (its weight fraction in a clinker would be dropped from 57% as estimated to 51.3%), 1.45% of free lime would form. However, from the XRD data, we cannot find any increase in the content of free lime in the clinkers with various glass additions. Then, the free lime content in the clinkers was analyzed by wet chemical analysis method. The results are listed in Table 3. It can be seen from Table 3 that the addition of the glass does not cause any significant increase in free lime content in the clinkers. This contradiction can only be explained by an assumption that the redundant CaO in the raw meal is captured by the flux during the cooling process and remained in amorphous phase together with other oxides. The Na 2 O in the flux, introduced by the glass in raw mixes, increases the alkalinity of the liquid phase, and thus Al 2 O 3 and Fe 2 O 3 exist in the form of AlO 4 5- and FeO 4 5-. Moreover, the glass also introduces SiO2 into the flux. As we know, all AlO 4 5-, FeO4 5-, and SiO4 4- are in the form of tetrahedroid, which has strong bonds between the elements and helps to increase the viscosity of the flux [10]. In such a liquid phase, it is difficult for the minerals (including free CaO) to crystallize. The decrease in C 3 S contents in the clinkers due to the addition of the glass may be partly attributed to the increase in the flux viscosity, and partly attributed to the intrusion of alkalis into the lattice of C 2 S, inhibiting the formation of C 3 S. 3.6 Volatility of alkalis during the burning process In general, the volatilization-condensation cycling of alkalis in kiln endangers the stable operation of kilns. Therefore, the volatility of the alkalis in cement raw meal was carefully examined in the present study. The raw 3.7 Properties of the cement made from raw mixes with the glass The clinkers 1-1400, 2-1400 and 3-1400 were mixed with gypsum separately and ground in the jar mill into cement with fineness less than 75 µm. Vicat needle test was used to examine the setting times. The compressive strength of the cement paste was examined with the specimens cast in polyethylene bottles. The dimension of the specimens was 16.5 mm in diameter by 25 mm in height. The test data are shown in Table 4. It can be seen in Table 4 that the cement made from clinker 1-1400 has normal setting and strength development. However, the cements made from clinker 2-1400 and 3-1400 exhibit flash setting within 2-4 minutes. These cements have much higher water demand than the cement made from 1-1400. The extremely low strength in mixes 2-1400 and 3-1400 may be due to the flash setting that makes the molding and compacting very difficult. The SG values of the three clinkers 1-1400, 2-1400 and 3-1400 are 60.0%, 34.67% and 25.83%, respectively. The alkali in the clinker 1-1400 is dominated by K 2 O, and this clinker has a proper SG value and the cement also has normal setting and strength properties. For clinker 2-1400 and 3-1400, the SG values are too low, and in these cases, the solid solution of C 3 A- NC 8 A 3 (KC 8 A 3 ) causes the flash setting of the cements right after mixed with water. Therefore, for these two clinkers, SO 3 -bearing materials must be added into raw mixes in order to increase the SG values of the clinkers. The raw materials used in the experiment have relatively high K 2 O content, and it is difficult to increase SO 3 content and thus the SG value. This is because that in the case of high K 2 O content and high SG value, the double salts 2CaSO 4 K 2 SO 4 would appear in the cement, which would also result in fast setting and poor strength property of the cement. Therefore, when the waste glass is used in a cement raw mix, other raw materials added in the mixes must contain alkalis as low as possible. 514

Xie, Xi 4. CONCLUSIONS The waste glass (sodium-lime-silicate glass) from daily recycling can be used to replace a portion of clay in cement manufacturing. In this study, the effects of the glass in cement raw mixes on clinker burning were investigated. 1. XRD analysis indicates that the addition of the glass into cement raw mix does not result in the formation of new minerals in the clinker. The glass can be added into cement raw mixes in a small fraction. 2. The presence of the glass in cement raw mixes may results in the formation of more liquid phase than conventional raw meals in the temperature range of 950 C to 1250 C. This would have an adverse effect on the stable operation of the kilns with a cyclone preheating system. 3. The addition of the glass decreases C 3 S content in the clinker. 4. The addition of the glass tends to cause fast set of the cement. It is necessary to properly increase the SG value of the clinker when the glass is present in the raw mix. 5. The glass as a raw material has the same effects on the burning process of clinker as other high-alkali raw materials. The properties of the cement made with the waste glass are expected to be the same as ordinary high alkali cements. ACKNOWLEDGEMENT Technical supports (including the supply of raw materials and chemical analysis of the clinkers) provided by Holnam Cement Company is gratefully acknowledged. REFERENCES [1] Hu, H. et al., Manufacture and Applications of Cement, (Shandong Keji Press, Jinan, 1993), 173. [2] Jawed, I. and Skalny, J., Alkalies in cement: a review - II. Effects of alkalies on hydration and performance of Porland cement, Cement Concrete Research (8) (1978) 37-52. [3] Lea, F.M., The Chemistry of Cement and Concrete, (Chem. Publ. Corp., 2 nd Edition, New York, 1971). [4] Locher, F.W., Setting of cement. Part I: Reaction and development of structure, Zement-Kalk-Gips, (29) (1976) 435-442. [5] Peter, J.J., Lea s Chemistry of Cement and Concrete, (4 th Edition, edited by P.C. Hewlett, London, 1998) 65. [6] Pollitt, H.W. and Brown, A., The distribution of alkalis in Portland cement clinker, Proc. 5 th Int l Symp. Chemistry of Cement, Tokyo, Vol. I (1968) 322,. [7] Schwiete, H.E., The influence of clay minerals on the formation of dust in the burning of clinker, Zement-Kalk-Gips (9) (1956) 351-357. [8] Suzukawa, Y., The alkali phase in Portland cement, Zement- Kalk-Gips (9) (1956) 345-351. [9] Taylor, H.F.W., Cement Chemistry, (2 nd edition, Thomas Telford, 1997). [10] Yanev, I.P. and Mitcheva, L., Kinetics of hydration of cements containing alkaline clinker minerals, Proceedings 6 th International Congress of Chemistry of Cement, Moscow, 1973, Supplemental paper, 170. [11] Zhang, X. and Tong, D., Principle of Cement Technology, (China Jiangong Press, Beijing, 1979) 51. 515