Chemical behaviour of ground waste glass when used as partial cement replacement in mortars

Construction and Building Materials 44 (2013) 74–80

Contents lists available at SciVerse ScienceDirect

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

Chemical behaviour of ground waste glass when used as partial cement replacement in mortars A. Khmiri, M. Chaabouni, B. Samet ⇑ Laboratoire de Chimie Industrielle, Ecole Nationale d’Ingénieurs de Sfax, BP W, 3038 Sfax, Tunisie

h i g h l i g h t s  The pozzolanic activity of finely sized glass was examined.  The main pozzolanic reaction products are calcium silicate and calcium aluminosilicate hydrates and hydrated sodium carbonate.  Formation of hydrated sodium carbonate explains size stability of the prepared mortars.  The 20 lm class ground glass exhibited a good pozzolanic behaviour.

a r t i c l e

i n f o

Article history: Received 25 November 2012 Received in revised form 10 February 2013 Accepted 25 February 2013 Available online 9 April 2013 Keywords: Ground waste glass Pozzolan Mortar Compressive strength Pozzolanic reaction products

a b s t r a c t This paper deals with the pozzolanic activity of finely ground waste glass, when used as partial cement replacement in mortars. The behaviour of this by-product was examined through two sets of tests: a lime–glass test to assess and explain the pozzolanic phenomena and a compressive strength test carried out to monitor the strength development. Analyses by DSC, XRD and SEM of samples containing 25% Ca(OH)2 and 75% of ground glass demonstrated that this admixture induced the formation of calcium silicate and calcium aluminosilicate hydrates at the expense of lime as well as hydrated sodium carbonate. The formation of the latter phase allowed us to explain the size stability of prepared mortars by hydrated sodium silicate formation. Compressive strength of mortars containing ground glass with particles size in the range 100–80 lm; 80–40 lm; <40 lm and < 20 lm finenesses indicated that the 20 lm class exhibited a good pozzolanic behaviour. The corresponding strength activity indexes were 82%, 95% and 102% at 7, 28 and 90 days respectively. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction The non-recyclable waste glass constitutes a problem for solid waste disposal since glass is a non biodegradable material. The use of the ground glass in the construction industry is among the most attracting option to valorise this waste, because it can consume a significant quantity of these materials and does not ask for very high conditions of quality [1]. An attempt was put for a long time to test the use of the waste glass like aggregate in the concrete [2–5]. Due to the strong reaction between the alkali in cement and the reactive silica in glass, the use of glass in concrete as part of the coarse aggregate was not satisfactory because of the marked strength regression and excessive expansion. Recently some experiments were made to use this waste like a raw material for the production of Portland cement [6,7]. However, in this case, it was proved that the alkali content in cement increased which could result in flash setting of

⇑ Corresponding author. Tel./fax: +216 74 676 908. E-mail address: [email protected] (B. Samet). 0950-0618/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.conbuildmat.2013.02.040

concrete due to the formation of 2CaSO4K2SO4 [7]. Thus the value addition of glass in concrete is best achieved if it is used as cement replacement material owing to its pozzolanic activity. Recall that such admixtures lead to an increase of the compressive strength; an improvement of durability [8] and a decrease of its free lime content. Glass is amorphous and high silica content, which is the primary requirement for a pozzolanic material. Few studies have investigated the potential of finely ground waste glass powder as pozzolanic material [9–11]. It was demonstrated that the properties which influence the pozzolanic behaviour of waste glass, and most pozzolans in general, are fineness and composition. The pozzolanic properties of glass are first notable at particle sizes below approximately 300 lm. Below 100 lm, glass can have pozzolanic reactivity which is greater than that of fly ash at low percent cement replacement levels and after 90 days curing [9,12,13]. Based on observed compressive strengths, Meyer et al. [14] postulated that below 45 lm, glass may become pozzolanic. In the studies treating the pozzolanic behaviour of ground waste glass, nobody has succeeded to identify the products of the pozzolanic reaction nor to propose its mechanism.

75

A. Khmiri et al. / Construction and Building Materials 44 (2013) 74–80

This paper presents a study on the assessment of the pozzolanic activity of waste glass, ground to different finenesses, as a partial replacement of cement in mortars and its chemical behaviour when acting as a pozzolan material. To explain the pozzolanic behaviour of this waste glass, samples of Ca(OH)2 were mixed with finely ground glass and first tested in compression, then analysed by XRD, DSC and SEM in order to prove the calcium hydroxide consumption and to identify the reaction products. 2. Materials and methods 2.1. Materials

2.1.2. Silica fume A commercial silica fume, conform to the requirement of the specification of ASTM C 1240-03a, is used as a reference pozzolan. The chemical composition of silica fume, as supplied by the manufacturer, is summarized in Table 1. 2.1.3. Waste glass The waste glass used in this study is a typical soda–lime glass obtained from recycled containers. Its chemical composition is indicated in Table 1. The glass powder is obtained by crushing and dry milling of 50 collected bottles in a jar mill, and by sieving the ground glass to the desired particle size by using 100 lm, 80 lm, 40 lm and 20 lm sieves. To study particle size effect, four different classes are used: A1: A2: A3: A4:

ground ground ground ground

Finally, SEM observation are carried out on fragments of the compression test, placed in acetone (to stop hydration) then dried in an oven at 60 °C for 2 h, by using a Philips XL 30 apparatus.

3. Results and discussion

2.1.1. Portland cement A typical local commercial type I 42,5N Portland cement (CEM I 42,5N) that complies with the requirements of specification ASTM C 150-04 is used as a testing cement. Its chemical composition is indicated in Table 1.

   

ments for these mortar mixes are summarized in Table 2. The water-to-cement ratio for pure Portland cement (reference) is 0.485 as specified in ASTM C311. The water-to-cement ratio for other mixtures containing silica fume and ground glass are tested to reach the same flow than that of the reference.  Physical test: pastes composed of 25% wt Ca(OH)2, 75% wt ground glass finer than 20 lm with a water/binder ratio equal to 0.60 were prepared in order to quantify the remaining Ca(OH)2 [16], by differential scanning calorimetry (NETSCH TA DSC 204, under helium with a heating rate of 10 °C/min) after 7, 28 and 90 days curing.

glass glass glass glass

having having having having

particle particle particle particle

size size size size

between 80 and 100 lm. between 40 and 80 lm. lower than 40 lm. lower than 20 lm.

2.2. Methods The X-ray fluorescence analysis is conducted on an ARL 8400 apparatus to determine the chemical composition of cement, glass and silica fume. The particle size distribution of these raw materials is measured by using a laser particle size analyser (Mastersizer 2000 Ver. 5. 12F) and their specific surface by Blaine method. The crystallinity of the glass is examined using X-ray diffraction technique. The X-ray diffraction (XRD) analysis is carried out using a Philips X’Pert Pro System. The generator settings are 45 kV and 40 mA and the wavelength (k) is 1.5418 Å (Cu Ka). A scanning rate of 1° (2h)/min from 3° to 70° is considered. The pozzolanic activity is assessed using two methods:  Mechanical test (pozzolanic reactivity index): the pozzolanic activity of glass with different finenesses is evaluated according to ASTM C311 test method on mortar bars [15]. In the test mixtures, 20% of the mass of the cement is replaced by glass powder or silica fume. The mix proportions and water require-

3.1. Material characterization The chemical compositions of the raw materials determined by X-ray fluorescence are reported in Table 1. The chemical analysis of the glass is characteristic of a soda lime one. Compared to silica fume, the waste glass has higher CaO, Na2O and Al2O3, but much lower SiO2 content. In agreement with ASTM C618-02 [17], which requires a sum of SiO2 + Al2O3 + Fe2O3 higher than 70% for good pozzolan, the sum for the investigated sample of glass is 73%. The studied glass sample presents satisfactory chemical composition. It can be classified as Class N natural pozzolan and therefore, it is likely to produce good pozzolan. The chemical composition, however, should not be used as the only criterion for prediction of the pozzolanic activity. The amorphous state is also required. It has been confirmed for waste glass by XRD as shown in Fig. 1. Indeed, no peaks attributed to any crystallized compound can be identified except a broad diffraction halo, which is attributed to the glassy phase. In order to assess the effect of fineness on the pozzolanic activity, four granulometric classes (particle size: <20; <40; 40–80; 80– 100 lm) were prepared. The particle size ranges of the four ground glasses as well as that of Portland cement and silica fume are plotted in Fig. 2. The comparison between the particle size distributions of the four glass fineness classes, shows that those of class A1, A2 and A3 contain grains of large sizes comparatively with that of cement, while, the size of the particles of A4 class and the silica fume is finer than that of cement. This result is confirmed by the values of specific surface determined by Blaine method. The specific surface of finest glass powder (WG < 20 lm) is of 4480 cm2/g, while that of cement and silica fume is respectively 3800 cm2/g and 5429 cm2/g.

Table 1 Chemical composition of raw materials. %wt

SiO2

Al2O3

Fe2O3

CaO

MgO

Na2O

K2O

SO3

TiO2

Cr2O3

Waste glass Silica fume Cem I 42.5 N

71.44 93.17 20.59

1.70 0.32 6.62

0.37 1.02 3.54

10.81 0.58 63.61

1.65 0.48 1.39

13.24 0.3 0.13

0.36 1.17 0.61

0.16 0.12 1.94

0.024 – –

0.19 – –

Table 2 Mixture proportions for mortars containing 20% mineral additives. Batch

%PC

%WG

%SF

Water/cement

Sand/cement

Reference mortar (RM) Mortar with A4 waste glass (MA4) Mortar with A3 waste glass (MA3) Mortar with A2 waste glass (MA2) Mortar with A1 waste glass (MA1) Mortar with silica fume (MSF)

100 80 80 80 80 80

0 20 20 20 20 20

0 0 0 0 0 20

0.485 0.502 0.485 0.465 0.465 0.515

2.75 2.75 2.75 2.75 2.75 2.75

76

A. Khmiri et al. / Construction and Building Materials 44 (2013) 74–80

compressive strength of the mortar. Strength activity index at a given age as prescribed by ASTM C 311 [15] is:

Strength Activity Index ðSAIÞ ¼ ðA=BÞ  100

Fig. 1. X-ray spectrum of waste glasses.

The particle size and the particle shape of the ground glass were analysed using the scanning electron microscope. Typical micrographs are shown in Fig. 3. The ground glasses exhibited angular shapes. For each glass, the specified particle size was dominant, although a large number of fines is present in the 80- and 40-lm glass. 3.2. Evaluation of the pozzolanic activity by mechanical method on mortars Compressive strength tests were carried out to study the strength development of mortar containing the ground glass at early and late ages. The cement replacement by the ground waste glass in the mortar was targeted at 20% by mass. The mortars containing the ground glass were compared to those having the same percent replacement of cement by silica fume, as well as to the control mortar without any mineral additives. The mix proportions for the prepared mortars mixes are summarized in Table 2. The water to cement ratio is fixed so that it ensures the normal plasticity of the mortar. As expected, as the fineness of the waste glass increases, the mixture water requirement for constant plasticity increases. Silica fume, which is the finest additive, shows the highest water demand. Nine moulds, (4040160 mm), were cast for each batch. After 24 h of curing in moulds, the samples were demoulded and placed in a water bath at room temperature for moisture curing. Three moulds from each batch were tested at 7, 28, and 90 days to obtain

Percentage passing(%)

120

where A = average compressive strength of blended cement mortar bars. B = average compressive strength of control mortar bars. This gives a measure of how close the values are to the control. The compressive strength of each batch at a particular age was an average of three tests. The maximum standard deviation of all the tests was 10%. The comparison of mortars containing ground waste glass with control mortar and silica fume is shown in Fig. 4. All batches of glass mortars and silica fume mortar had lower strengths than the control at the ages of 7, 28. The strength of the mortar containing 20-lm glass exceeded that of control by 2% after 90 days of curing compared to 6% for those containing 20% silica fume. This indicated that 20-lm ground glass can replace cement in mortar without affecting the mechanical proprieties of the mortar. The strength activity indexes of all the five mortars containing 20% mineral additives are presented in Fig. 5. Recall that ASTM C618 recommends that a pozzolan have a minimum strength activity index of 75% (dashed line on the Fig. 5). As a consequence, the activity index of 100 lm glass and 80 lm glass did not satisfy this criterion. In contrast to this, 40 and 20 lm glasses were not only very suitable, but directly comparable to silica fume mortar at all ages. The relatively higher early strength index of 20 lm glass concrete could be attributed to the high content of Na2O in glass. It was generally accepted that alkalis in concrete could act as catalysts in forming calcium silicate hydrate at early age, and promoting early strength development [18]. Silica fume mortar showed a superior performance. It was the submicron particle size and the high silica content in silica fume that played a critical role in strength development. In order to understand the reaction mechanism between cement and ground waste glass, a simple system composed by waste glass and lime was studied. 3.3. Identification of the pozzolanic reaction products The tests of lime–glass were carried out according to J. Ambroise method [19]. Five batches of samples containing lime and one additive were prepared silica fume, 100 lm glass, 80 lm glass, 40 lm glass and 20 lm glass. Silica fume batch is used as control. The mixture proportions are: additive/Ca(OH)2 = 3/1 and water/solid = 0.60. Each mixture was cast in 4020 mm micro-test-tube moulds, wrapped by wet burlap, sealed by plastic bag, and cured at 21 °C in water for 7, 28 and 90 days. At least three samples were

WG<40 µm

40
80
Portland Cement

WG<20µm

Silica Fume

100 80 60 40 20 0 0,3

1

3,5

8

12

18

24

40

63

80

120

150

200

particle size (microns) Fig. 2. Particle size distributions of glass powders, cement and silica fume, (WG: waste glass).

77

A. Khmiri et al. / Construction and Building Materials 44 (2013) 74–80

(a)

(b)

Acc.V Spot Magn Exp 20.0 kV 5.0 1029x BSE 9.7

20µm V-20

Acc.V Spot Magn Exp 20.0 kV 5.0 1047x BSE 10.0

40µm V-40

(c)

Acc.V Spot Magn Exp 10.0 kV 4.8 501x BSE 10.6

50µm V-80

Fig. 3. MEB analysis: particle size and shape of ground waste glass. (a) 20-lm glass; (b) 40-lm glass and (c) 80-lm glass.

Compressive Strength (Mpa)

70 60 50 40 30 20 10 0 RM

MA4

MA3

7 days

MA2

28 days

MA1

MSF

90 days

Fig. 4. Compressive strength of concretes containing 20% ground glass.

Strength activity index

1,2 1 0,8

0,75

0,6 0,4 0,2 0 RM

MA4

MA3 7 days

MA2 28 days

MA1

MSF

90 days

Dashed line indicates the limit of stength activity index as prescribed by standard ASTM C 618 Fig. 5. Strength activity index of mortars containing 20% mineral additives.

tested and averaged for each batch. Before analysing the obtained samples by XDR, DSC and MEB, the hydration was stopped by addition of acetone 3.3.1. Analysis by XRD The XRD patterns of the lime–glass paste prepared with fraction lower than 20 lm (Fig. 6) shows the presence of remaining

Ca(OH)2 and CaCO3 formed by the contact of lime with atmospheric CO2. The intensities of both phases decrease when curing age increases. By the spectra between 2h 28 and 35° (Fig. 6) we detected mainly calcium silicate hydrates (Ca4(Si6O16)(OH)2; Ca5(SiO4)2 (OH)2; Ca2SiO4 3H2O; Ca4,5 Si6O15 (OH)3 2H2O; Ca2SiO4 0,5H2O; Ca4Si5O13.5 (OH)2) and aluminosilicate hydrates (Ca12Al2Si18O51

78

A. Khmiri et al. / Construction and Building Materials 44 (2013) 74–80

2+3+4 4+6 7 1 44 + 4 5 4 7 44

3

4 7 46+ + 6 7 4 44 3

5

4 + 4 7

7

1

4

7

90 days

7 days 28

33

1:Ca(OH)2 ; 2: CaCO3; 3: NaCO3.10H2O; 4: xCaO SiO2 zH2O; 5: MgSiO3; 6: Ca7 ( Si6O18 ) ( CO3 ) ·2H2 O; 7: x’CaO y’Al2O3 SiO2 z’H2O. Fig. 6. XRD paterns of lime-20 lm waste glass sample at 7et 90 days between 2h and 28–35.

contact with a solution of Ca(OH)2, hydrolysis of SiAOASi bond, which is favoured by high pH, leads to the formation of the SiOH causing silica dissolution [21] as follows:

18H2O; Ca (Al2Si4O12) 6H2O; CaAl2Si3O106H2O) (Al coming from the waste glass). These phases which are the main hydration products are especially abundant at 90 days. Finally sodium calcium carbonate (Na2CO310H2O) was present following the carbonation of sodium initially present in glass which was dissolved by lime in the hydration solution. The presence of this phase is very interesting because it proves that sodium (network modifier) does not react with silica to form expansive sodium silicate hydrates. For this reason no expansive alkali–silica reaction was detected in the studied system.

H2 O ! OH þ Hþ BSiAO þ Hþ ! BSiAOH þ

BSi þ OH ! BSiAOH The third phenomenon observed in DSC thermograms (Fig. 7) situated at 400–500 °C, is attributed to the decomposition of remaining Ca(OH)2. We noticed that the quantity of the remaining Ca(OH)2 decreases between 7 and 28 days and disappears at 90 days which proves the evolution of the pozzolanic reaction. This result is in accordance with that of XRD data.

3.3.2. Differential scanning calorimetry analysis Differential scanning calorimetry was carried out on samples containing ground glass finer than 20 lm after 7, 28 and 90 days curing in order to quantify the remaining Ca(OH)2 and to identify the hydration products. The thermograms obtained are presented in Fig. 7. The thermograms show three phenomena: the first one between 70 and 120 °C is endothermic; it is related to the removal of both physically bound water and the dehydration of calcium silicate hydrate which is the principal product of the pozzolanic reaction [20]. At early ages (7 and 28 days) this phenomenon is subdivided in two peaks. At 90 days, the first peak is transformed to a shoulder. The intensity of this phenomenon increased with the age. The second phenomenon which is exothermic around 300 °C is attributed to glass hydrolysis. Its intensity, which is higher in the pastes at 7 days, decreases with curing age. When silica is kept in

3.3.3. SEM–EDS analyses EDX analysis (Fig. 8a) demonstrates that 20 lm ground glass has reacted with lime, producing C–S–H gels. At 28 days the glass particle appeared more active with lime. This phenomenon is indicated by the apparition of needles of CASAH (Fig. 8b). At 90 days, the hydration products become denser (Fig. 8c) which indicates a continuation of the pozzolanic reaction. 4. Conclusions This research study proves that waste glass, if ground finer than 20 lm, exhibits a pozzolanic behaviour. The strength activity in-

DSC(mW/mg)

1,10

0,80

0,50

90 days

0,20

28 days 7 days -0,10 10

110

210

310

410

510

Temperature (°C) Fig. 7. Differential scanning calorimetry analysis of lime-20 lm waste glass sample.

A. Khmiri et al. / Construction and Building Materials 44 (2013) 74–80

Fig. 8a. SEM–EDS analyses of lime-20 lm waste glass paste at 7 days.

Fig. 8b. SEM–EDS analyses of lime-20 lm waste glass paste at 28 days.

Fig. 8c. SEM–EDS analyses of lime-20 lm waste glass paste at 90 days.

79

80

A. Khmiri et al. / Construction and Building Materials 44 (2013) 74–80

dexes of the mortar with 20% cement replaced by 20-lm glass are 82, 95, and 102% at 7, 28, and 90 days, respectively. Analyses by DSC, XRD and SEM of simple system composed by 75% of ground waste glass and 25% Ca(OH)2 mixed with water prove that the hydrolysis of SiAOASi bonds occurs at high pH and led to the formation of insoluble calcium silicate and calcium aluminosilicate hydrate at the expense of lime. The XRD confirmed in addition, the presence of sodium carbonate hydrate, formed by carbonation of sodium. This can explain the stability to expansion of the prepared mortars generally caused by hydrated sodium silicate formation. Finally SEM observations coupled with EDS analysis confirmed the presence of CASAH and the increase of its content with curing age.

References [1] Chesner WH, Coollins RJ. Mackay M.H., User guidelines for waste and byproduct materials in pavement construction. US Department of transportation, federal highway administration, Publication No. FHWA-RD; 1997. p. 97–148. [2] Meyer C, Egosi N, Andela C. Concrete with waste glass as aggregate. In: Recycling and reuse of glass cullet. In: Proceedings of the international symposium. Concrete technology unit of ASCE and University of Dendee; 2001. [3] Topcu IB, Cambaz C. Properties of concrete containing waste glass. Cem Concr Res 2004;34:267–74. [4] Sangha CM, Alani AM, Walden PJ. Relative strength of green glass cullet concrete. Mag Concr Res 2004;56:293–7. [5] Park SB, Lee BC, Kim JH. Studies on mechanical properties of concrete containing waste glass aggregate. Cem Concr Res 2004;34:2181–9.

[6] Chen G, Lee H, Young KL, Yue PL, Wong A, Tao T, et al. Glass recycling in cement production-an innovative approach. Waste Manage 2002;22:747–53. [7] Xie Z, Xi Y. Use of recycled glass as a raw material in the manufacture of Portland cement. Mater Struct 2002;35:510–5. [8] Matos AM, Sousa-Coutinho J. Durability of mortar using waste glass powder as cement replacement. Constr Build Mater 2012;36:205–15. [9] Shi C, Wu Y, Riefler C, Wang H. Characterization and pozzolanic reactivity of glass powders. Cem Concr Res 2005;35:987–93. [10] Dyer TD, Dhir RK. Chemical reactions of glass cullet used as cement component. J Mater Civ Eng 2001;13:412–7. [11] Shayan A, Xu A. Performance of glass powder as a pozzolanic material in concrete: a field trial on concrete slabs. Cem Concr Res 2006;36:457–68. [12] Schwarz N, Cam H, Neithalath N. Influence of fine glass powder on the durability characteristics of concrete and its comparison to fly ash. Cem Concr Comp 2008;30:486–96. [13] Pereira-de-Oliveira LA, Castro-Gomes JP, Santos PMS. The potential pozzolanic activity of glass and red-clay ceramic waste as cement mortars components. Constr Build Mater 2012;31:197–203. [14] Meyer C, Baxter S, Jin W. Potential of waste glass for concrete masonary blocks. In: Proceedings of the 4th Materials engineering conference, Washington; 1996. p. 666–673. [15] American Society for Testing and Materials, ASTM C 311-02; Standard test methods for sampling and testing fly ash or natural pozzolans for use in Portland-cement concrete. [16] Karamberi A, Moutsatsou A. Participation of coloured glass cullet in cementitious materials. Cem Concr Comp 2005;27:319–27. [17] American society for testing and materials, C 618–02. Standard specification for Coal fly ash and raw or calcined natural pozzolan for use in concrete. [18] Jawed I, Skalny J. Alkalies in cement: A review. Cem Concr Res 1978;8:37–51. [19] Ambroise J, Martin Calle S, Péra J. Pozzolanic behavior of thermally activated Kaolin. 4th Int. Conf Istambul Turkey SP; 1992. p.132–140. [20] Moisés F, Joseph G. Influence of MK on the reaction kinetics in MK/lime and MK-blended cement system at 20°C. Cem Concr Res 2001;31:519–27. [21] Greenberg SA. Reaction between silica and calcium hydroxide solutions. I. Kinetics in the temperature range 30 to 85 °C. J Phys Chem 1961;65:12–6.