sodium silicate composites

Construction and Building Materials 93 (2015) 230–240

Contents lists available at ScienceDirect

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

Novel mechanical behaviour of perlite/sodium silicate composites Md Arifuzzaman ⇑, Ho Sung Kim Mechanical Engineering, School of Engineering, Faculty of Engineering and Built Environment, The University of Newcastle, Callaghan, NSW 2308, Australia

h i g h l i g h t s  A novel mechanical behaviour of perlite/sodium silicate composites is studied.  Dehydration behaviour of sodium silicate as binder is characterised.  Mechanical properties are discussed in relation with manufacturing parameters.  Manufacturing parameters and volume fractions of constituents are correlated.  A rule of mixtures is proposed for a constant compaction ratio.

a r t i c l e

i n f o

Article history: Received 24 March 2015 Received in revised form 15 May 2015 Accepted 17 May 2015

Keywords: Expanded perlite Sodium silicate Foam Density Compaction Composite Compressive strength Compressive modulus Rule of mixtures

a b s t r a c t A novel mechanical behaviour of perlite/sodium silicate composites is studied with the benefits of a new manufacturing method based on the perlite particle buoyancy. The objective was to develop perlite composites and to understand their quantitative relations between manufacturing parameters, volume fractions of constituents, and properties. For the composites development, sodium silicate dehydration behaviour was characterised with phases formed during dehydration i.e. liquid, gel, and solid phases. The water loss–time curve for dehydration was found to have three distinctive parts – linear part at an early stage for liquid phase, followed by non-linear part during a period between commencements of gel and hydrated solid phase formations, and then another linear part for hydrated solid phase. Foams as composites were manufactured with diluted sodium silicate binder for a density range of 0.2–0.5 g/cm3. One of practical milestones achieved for composite properties without reinforcement was a density of 0.3 g/cm3 at a compressive strength of 1 MPa. Manufactured perlite/sodium silicate composites are analysed/discussed for understanding from three different perspectives i.e. manufacturing parameters (i.e. binder content, compaction pressure, and compaction ratio), properties (i.e. particle size, density, compressive strength, and modulus), and volume fractions of constituents. A rule of mixtures applicable for perlite composites for a constant compaction ratio was developed in comparison with that for particulate composites with non-compaction. It may be a basis for further development for variable compaction ratio in the future. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Perlite is a glassy volcanic rock of rhyolitic composition [1], which can be processed into an expanded form for cellular structure formation [2,3]. The expansion takes place due to the presence of water in perlite when it is heated to about 649–816 °C [4]. The expanded perlite particles are light, environment-friendly [5], and possess good acoustic [6] and insulation properties [7]. Their uses are broadly covered in the literature by Kendall [8]. They have been used as additives or main components for composites, e.g. Portland cement/perlite composites for blocks [9,10], ⇑ Corresponding author. E-mail address: [email protected] (M. Arifuzzaman). http://dx.doi.org/10.1016/j.conbuildmat.2015.05.118 0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.

perlite/sodium silicate boards [11], roof insulation panels made of perlite/fibres/bituminous material [12], fibre reinforced perlite/cement composites [13], building boards made of fibre/asphalt coated perlite [14] or urea–formaldehyde resin/mineral fibres/gypsum/glass fibres [15], fibre reinforced sodium silicate/perlite composite [16], moisture resistant gypsum boards modified with perlite/starch/boric acid/vinyl acetate [5], gypsum/perlite composites [17], and light weight concrete [18]. However, their applications as the main constituent of composites have been limited due to their relatively poor mechanical properties. One of the reasons for this is that the expanded perlite particles are fragile and hence easily damaged during the process of mixing with binder, resulting in a high ratio of density to strength.

231

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

Nomenclature Fp Fb

qpB qpE qpM qpS qf qb mp ¼ mpm mb Mp Mf Mb Mbc

load carrying capacity of perlite particles load carrying capacity of binder perlite bulk density perlite envelop density perlite material density perlite skeletal density density of dry foam density of binder (sodium silicate) mass fraction of perlite in foam mass fraction of binder (sodium silicate) in foam mass of perlite in foam mass of foam mass of binder (sodium silicate) in diluted binder mass of binder (sodium silicate) in foam after compaction 

 Rbd ¼ MV ib ¼ MV icbc

rbond rb

rcf rsf rsb

mass of pure binder per unit volume of diluted

binder bonding strength between binder and particles binder strength compressive strength of perlite/sodium silicate foam shear strength of perlite/sodium silicate foam shear strength of binder

The study on mechanical performance of perlite composites compatible with gypsum boards [19,20] has not much been available in the literature. It is only recently that Shastri and Kim [21,22] studied some selected properties for mechanical behaviour of expanded perlite consolidated with starch for demonstration of a new manufacturing process based on the principle of buoyancy [23–28]. The new process appears to be capable of extending the limitation of perlite application, allowing us to manufacture novel perlite/sodium silicate composites. In the development of perlite composites, selection of binder is another consideration along with manufacturing process. Sodium silicate, which is an inorganic colloidal system, may be one of candidate binders. It has been used as foundry sand binder, fire-retardants, adhesives, and deflocculants among other applications [29] even though the behaviour of sodium silicate is not fully understood [30]. Also, it is non-combustible, water-resistant and sufficiently inexpensive for developing building materials. This paper focuses on the novel mechanical behaviour of expanded perlite/sodium silicate composites developed using the new manufacturing process [21,22].

2. Constituent materials and characterisation 2.1. Expanded perlite Commercial grades of expanded perlite particles were obtained from Australian Perlite Pty Limited. Expanded perlite particles were sieved using a vibratory sieve shaker (Analysette 3 SPARTAN) into three different particle size ranges i.e. sizes between 1 and 2 mm, 2 and 2.8 mm, and 2.8 and 4 mm, which will be referred to as Size 1–2, Size 2–3, and Size 3–4, respectively. Four different perlite densities measured and listed in Table 1. The density terminology is based on ASTM D 3766-08 and illustrated in Fig. 1. For bulk density measurement, an initial volume of 100 cm3 of expanded perlite particles was poured into a glass measuring cylinder with a 28 mm diameter fitted to a manual tapper with a tapping stroke height of 5 mm, and then tapping was conducted for 300 times. For envelope density measurement, a

rsb0 rsP tcpore topore tipp tpp ttotal

vb v bi

v bs v pe v pm v tv v vi v v i0 Vi V ic

shear stress of binder shear strength of perlite particles volume fraction of closed pores in perlite particles volume fraction of open pores in perlite particles inter-particle porosity in bulk volume particle porosity total porosity in bulk volume volume fraction of binder in foam volume fraction of binder in foam between perlite particles volume fraction of binder in skeletal volume of perlite particles volume fraction of perlite envelope in foam volume fraction of perlite material in foam volume fraction of voids in foam including both interand intra-particle voids volume fraction of inter-particle voids in foam volume fraction of inter-particle voids in foam without binder total volume of diluted binder total volume of diluted binder in foam after compaction

volume of about 4 cm3 of expanded perlite particles was poured into molten paraffin wax in an aluminium container (37 mm in diameter and 13 mm height), ensuring it was fully submerged and each particle was fully wetted before wax solidification. The enveloped volume of perlite was determined by the difference in wax volume before and after submersion of perlite. Particle skeletal and material (true) densities were measured using a gas pycnometer (AccuPyc 1330). For the material density sample preparation, expanded perlite particles were crushed into fine powder using a ball mill (8000D Mixer/Mill SPEX) for at least 5 min to remove the closed pores before volume was measured in pycnometer. It was visually confirmed using an optical microscope (Olympus SZ-CTV) that the closed pores were removed. Various porosities defined below were obtained and listed in Table 2. The total porosity (ttotal ) in bulk volume is defined as

ttotal ¼ 1 

qpB qpM

! ð1Þ

where qpB is the perlite bulk density and qpM is the perlite material density; the volume fraction of open pores in perlite particles (topore ) as

q topore ¼ 1  pE qpS

! ð2Þ

where qpE is the perlite envelop density and qpS is the perlite skeletal density; the volume fraction of closed pores in perlite particles (tcpore ) as Table 1 Densities of expanded perlite particles. Perlite particle size

Bulk density (qpB ),

Particle envelope density (qpE ),

Particle skeletal density (qpS ),

g/cm3

g/cm3

g/cm3 Size 1–2 Size 2–3 Size 3–4

0.089 0.091 0.100

Material density (qpM ), g/cm3

0.140 0.160 0.152

1.466 1.309 1.207

2.46 2.46 2.46

232

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

Bulk volume

2.2. Sodium silicate solution and dehydration behaviour

Envelope volume

Intra-particle closed pore/void

Inter-particle void Open pore Intra-particle open pore/void

(a)

(b)

Dense material Skeletal volume

(c)

(d)

Fig. 1. Various parts of expanded perlite for definitions: (a) bulk volume enclosed by rectangle; (b) envelope volume enclosed by black line; (c) skeletal volume enclosed by black line; and (d) material volume shown in grey.

tcpore ¼

qpE ðq  qpS Þ; qpM qpS pM

ð3Þ

the total particle porosity (tpp ) as

!

q tpp ¼ 1  pE ; qpM

ð4Þ

and inter-particle porosity in bulk volume (tipp ) as

!

q tipp ¼ 1  pB : qpE

ð5Þ

Also, volume fractions of the following: binder (v b ), perlite material (v pm ), and perlite envelope (v pe ) can be obtained from:

v b ¼ qf

mb

qb

mpm

v pm ¼ qf

ð6Þ

;

qpM

¼ qf

mp

qpM

ð7Þ

;

and

v pe ¼ qf

mpe

ð8Þ

qpE

where qf is the dry density of foam, qb is the density of binder (sodium silicate), qpM is the true density of perlite particles, and qpE is the envelope density of perlite particles. Table 2 Various porosities of perlite particles. Perlite particle size

Size 1–2 Size 2–3 Size 3–4

Sodium silicate solution (ChemSupply) with a density range of 1.37–1.40 g/cm3, a solid content range of 37.10–38.00% (by mass), and a weight ratio of silica to sodium oxide (SiO2/Na2O) range of 3.16–3.22 was used as binder. Sodium silicate solution was dehydrated at 80 °C to obtain a solid sample, and then was ball-milled (8000D Mixer/Mill, SPEX) for 30 min into powder for material density measurement. The density was measured using a gas pycnometer (AccuPyc 1330) and found to be 2.17 g/cm3. For dehydration behaviour, sodium silicate solution (SSS) was diluted with water as specified in Table 3. The diluted samples in a container of approximately 40 mm diameter were placed in an electric fan forced air oven (Lebec Oven BTC-9090) at 65 °C and subsequently mass loss was recorded every 10 min. Three phases were identified from diluted sodium silicate solution as dehydration progressed. The first phase was of liquid, gel in the second phase, and solid in the third phase (the gel is a cohesive substance consisting of colloidal particles [29]). The commencement and completion points of the gel formation and other phases are indicated with dashed lines in Fig. 2 and listed with water contents in Table 4. Some of phase transitional points appear to be practically distinguishable on the water loss-time curve. The first linear portion of the curve corresponds to liquid phase, non-linear portion corresponds to a stage where gel phase started to form from liquid phase until the two phases (liquid and gel) become fully gel prior to solidification, and the other linear part corresponds to hydrated solid phase. Free water contents at the completion of gel formation from (gel + liquid) were measured to be approximately constant ranging from 58.71% to 59.73% as listed in Table 4. Further drying of the gel leads to the hydrated solid phase. However, the water (regarded as ‘chemically bonded water’ according to Owusu [29]) contents as ranged from 29% to 44% in hydrated solid phase are not as much constant as those at the gel completion point. 3. Manufacturing process of samples and mechanical tests The manufacturing process consists of different stages i.e. dilution of sodium silicate binder, mixing of binder and perlite in a container, flotation of wet-mix, moulding and compaction, demoulding and drying as detailed elsewhere [21]. The dilution of sodium silicate was made in drinkable tap water. The perlite was poured into a mixing container containing the prepared binder, followed by stirring/tumbling of the mixture for phase separation consisting of top phase of perlite and binder, and bottom phase of binder. When the top phase was formed, it was transferred into a mould for compaction. The compaction was conducted for various densities at a crosshead speed of 10 mm/minute on a universal testing machine (Shimadzu 5000). Compression tests for manufactured foam specimens were conducted on a universal testing machine (Shimadzu 5000) at a crosshead speed of 5.0 mm/min and at an ambient temperature range

Table 3 Samples of sodium silicate solution (SSS) used for dehydration behaviour.

Total porosity in bulk volume (ttotal ), %

Volume fraction of open pores in perlite particles (topore )

Volume fraction of closed pores in perlite particles (tcpore )

Total particle porosity (tpp ), %

Interparticle Porosity in bulk volume (tipp ), %

96.37 96.31 95.92

0.9045 0.8778 0.8741

0.0386 0.0572 0.0641

94.31 93.50 93.82

36.43 43.13 34.21

a

Samples

Mass of SSS (g)

Mass of diluted SSS, (g)

Water mass content in dilution (%)

Sample 1 (control) Sample 2 Sample 3 Sample 4

13.85

13.85

62.45a

6.93 4.61 3.46

11.93 11.28 10.96

78.21 84.66 88.14

Provided by the manufacturer (ChemSupply).

233

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

Fig. 2. Water loss during dehydration at 65 °C as a function of time. The first filled data point on each sample indicates gel formation commencement.

Table 4 Water contents and times at different stages for dehydration behaviour at 65 °C. Samples

Commencement time (min) of gel formation from liquid

Water content at commencement of gel (%)

Completion time (min) of gel formation from liquid

Water content at gel formation (%)

Water (%) of initial water content

Hydrated solid commencement time (min)

Water content at hydrated solid commencement (%)

Sample 1 (control) Sample 2 Sample 3 Sample 4

30

61.12

50

59.44

(11.92)

670

43.92

120 150 160

63.06 66.31 68.12

150 180 190

58.71 59.58 59.73

(14.62) (11.46) (10.74)

560 520 500

37.14 29.12 30.76

16–21 °C. A Hounsfield compression cage was used, in which the platens were lubricated (engine oil SAE 15-40) to minimise the friction between test samples and platens. Cylindrical test samples of 35 mm high and 35 mm in diameter were compressed 10–15% of the initial heights, which was sufficient to obtain the results for characteristic stages of stress–strain curve. Compressive strength was calculated using the original cross-sectional area and compressive modulus was calculated from the tangent to the most linear portion or inflection point of the initial stress–strain curve. 4. Experimental results and discussion Data sets collected in the course of foam manufacturing for compaction pressure versus compaction ratio, c (=top phase volume in mixing container divided by compacted volume in mould) for given ranges of perlite particle sizes and binder contents (Rbd) are given in Fig. 3. They appear to be not much dependent of binder content but somewhat dependent on particle size – the higher particles size the higher compaction pressure. A reason for the tendency may be that the larger particles of the initial top phase volume in the mixing container may have smaller inter-particle surface to surface distances and, as a result, the larger particles would be subjected to more fragmentation of particles requiring higher compaction pressures whereas the smaller particles would

be subjected to more repositioning than fragmentation in the course of the volume reduction process. Data sets collected for dry foam density (of perlite-sodium silicate) versus applied compaction pressure for given binder contents and particle sizes are shown in Fig. 4. Individual plots for different sodium silicate contents are given in Fig. 4(a)–(c) to see particle size effects and a combined plot for different sodium silicate contents without distinguishing particle sizes is given in Fig. 4(d) for further comparison purposes. For any given binder content, foam density tends to increase linearly with increasing compaction pressure for all the particle sizes. Also, a clear trend is found that, as expected, foam density increases with increasing sodium silicate content (Fig. 4(d)). The Pearson correlation coefficients (r) for the individual plots in Fig. 4(d) were found to be 0.9691, 0.9682, and 0.9841 for sodium silicate contents (Rbd), 0.35 g/ml, 0.20 g/ml, and 0.05 g/ml, respectively, indicating foam densities are not sensitive to particle size although there is a weak tendency that the smaller particle size the higher foam density, and a little higher compaction pressure is required for larger particle sizes to achieve a certain density. It is also found as expected that a higher compaction pressure is required to achieve a certain foam density as the binder content becomes lower. A reason why the initially smaller particles tend to have higher foam densities may be due to a higher binder retention of large surface areas, as will be also discussed below (Fig. 6), given that there is not much difference in

234

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

Fig. 3. Compaction pressure versus compaction ratio for producing perlite foams with ranges of particle sizes (mm) and sodium silicate contents in dilution (g/ml). The compaction ratio is defined as the top phase volume in mixing container divided by compacted volume in mould. The least square line (y) and Pearson correlation coefficient (r) were found to be: y = 1.1848x  1.6095 with r = 0.9938 for Size 3–4; y = 0.9644x  1.3268 with r = 0.9916 for Size 2–3; and y = 0.8605x  1.1607 with r = 0.9848 for Size 1–2.

fraction of open pores between different particle sizes as indicated in Table 2. Data sets obtained for wet density of perlite-sodium silicate foam versus applied compaction pressure before drying for different particles sizes and binder contents (Rbd) were plotted in Fig. 5 for understanding the manufacturing process. Individual plots for different sodium silicate contents are given in Fig. 5(a)–(c) and a combined plot is given in Fig. 5(d). As expected, it is seen that wet density is more sensitive to particle size than dry density (Fig. 4) because its fraction of liquid binder retained in foam is higher than that of dry one for a constant bulk volume. However, it is found to be less sensitive to sodium silicate content in water than dry density (Fig. 4) because the difference in binder content of wet foam between different dilution levels is relatively small compared to dry foam. Various volume fractions in perlite foams at a compaction ratio range of 3.55–3.66 were calculated and are shown in Fig. 6 as a function of volume fraction of sodium silicate in diluted binder (VFSSB). The volume fraction of perlite material in foam (v pm ) (VFPMF) (Fig. 6(a)) appears to be not sensitive to VFSSB but the larger particle size tends to have a higher VFPMF probably because the fragmentation of larger particles during compaction did not leave much space for inter-particle voids compared to small particles. Volume fraction of total voids in foam (v tv ) (Fig. 6(b)), as expected, decreases as VFSSB increases but without much sensitivity of particle size effect indicating that the smaller particles tend to retain more binder due to more inter-particle spaces and surface areas for smaller particles. On the other hand, the volume fraction of SS in foam (v b ) (Fig. 6(c)) is found to be highly proportional to VFSSB as expected. The Pearson correlation coefficients (r) with a forced intercept at zero were found to be 0.985, 0.983, and 0.977 for Size 1–2, Size 2–3, and Size 3–4 respectively (Fig. 6(c)). Compressive strength and specific compressive strength of perlite-silicate foams are plotted as a function of dry foam density in Fig. 7. It is seen that they expectedly increase with increasing foam density for all the different contents of sodium silicate in diluted binder. The least square lines and Pearson correlation coefficients (r) for compressive strength were found to be: y = 8.3442x  1.7088 with r = 0.9907, y = 5.5094x  0.7902 with r = 0.9647, and y = 2.9833x  0.3331 with r = 0.9836 for respective sodium silicate contents (Rbd) of 0.35, 0.20, and 0.05 g/ml. Also,

Fig. 4. Dry foam density versus applied compaction pressure for different binder contents and particle sizes. Pure binder mass per unit diluted binder volume (g/ml): (a) 0.05 g/ml; (b) 0.20 g/ml; (c) 0.35 g/ml; and (d) combined plot from all particle size groups. Pearson correlation coefficients (r) were found to be 0.9691 for 0.35 g/ ml, 0.9682 for 0.20 g/ml, and 0.9841 for 0.05 g/ml.

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

235

Fig. 6. Various volume fractions in perlite-sodium silicate foam as a function of volume fraction of sodium silicate in diluted binder for compaction ratio c = 3.55– 3.66: (a) volume fraction of perlite material (v pm ) excluding pores in foam; (b) volume fraction of total voids in foam (v tv ); and (c) volume fraction of sodium silicate in foam (v b ). Pearson correlation coefficients (r) with forced intercept at zero – 0.985 for Size 1–2, 0.983 for Size 2–3, and 0.977 for Size 3–4.

Fig. 5. Foam wet density versus compaction pressure for various binder contents: (a) 0.05 g/ml; (b) 0.20 g/ml; (c) 0.35 g/ml; and (d) combined plot of ‘(a)’, ‘(b)’, and ‘(c)’ without distinguishing the particles sizes.

those for specific compressive strength were found to be: y = 11.925x  0.899 with r = 0.9743, y = 8.7068x + 0.0935 with r = 0.9031, and y = 5.9992x + 0.032 with r = 0.9482 for respective sodium silicate contents of 0.35, 0.20, and 0.05 g/ml. Each data set for the least square line includes three different particle size ranges (i.e. Sizes 1–2, 2–3, and 3–4) appearing in a form of small cluster for similar foam density values but clearly distinguishable in Fig. 7. The high correlation coefficients, though, indicate that the particle size effect on the compressive strength is not

236

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

Fig. 7. (a) Compressive strength as a function of dry foam density for various sodium silicate contents (g/ml) with the least square line (y) and Pearson correlation coefficient (r): for 0.35 g/ml, y = 8.3442x  1.7088 with r = 0.9907; for 0.20 g/ml, y = 5.5094x  0.7902 with r = 0.9647; and for 0.05 g/ml, y = 2.9833x  0.3331 with r = 0.9836. (b) Specific compressive strength as a function of foam density for various sodium silicate contents: for 0.35 g/ml, y = 11.925x  0.899 with r = 0.9743; for 0.20 g/ml, y = 8.7068x + 0.0935 with r = 0.9031; and for 0.05 g/ml, y = 5.9992x + 0.032 with r = 0.9482. [Compaction ratio was varied within a range of 1.5–3.5 for all the sodium silicate contents.]

significant. It is noted that, given that the foam density for a given condition is a function of two independent variables i.e. sodium silicate content and compaction ratio (c = 1.5–3.5), an optimum combination of the variables may exist for practical manufacturing purposes. If we choose a value of 1 MPa, for example, for compressive strength with a density of 0.3 g/cm3, there may be two different choices i.e. sodium silicate contents, 0.35 g/ml with a compaction ratio (c) of about 1.5 (not shown on graph), and 0.20 g/ml with a compaction ratio of 2.5 (not shown on graph). Now, it may be useful to compare the current results with some other materials in the literature. A gypsum compressive strength of 1 MPa for a density of 0.87 g/cm3 was reported by Colak [19]. Also, a specific compressive strength range of 0.8– 5.37 MPa/(g/cm3) in Fig. 7(b) may be compared with a range of 0.62–2.03 MPa/(g/cm3) from gypsum by Colak [19], a range of 1.1–3.1 MPa/(g/cm3) from foamed gypsum by Skujans et al. [20] or 1–3.86 MPa/(g/cm3) from gypsum/perlite composites by Vimmrova et al. [17]. Compressive modulus and specific compressive modulus as a function of dry foam density (qf ) for various binder contents (but without distinguishing particle sizes) are given in Fig. 8. As expected, they increase with increasing foam density, and more rapidly increase with increasing binder content but with relatively high scatters compared with those for compressive strengths as indicated by Pearson correlation coefficients (r) for compressive modulus (i.e. 0.8509, 0.6430, and 0.9413 for respective sodium silicate contents (Rbd) of 0.35, 0.20, and 0.05 g/ml). For specific

Fig. 8. (a) Compressive modulus and (b) specific compressive modulus as function of foam density. [Compaction ratio was varied within a range of 1.5–3.5 for all the sodium silicate contents.]

compressive modulus, Pearson correlation coefficients (r) were found to be 0.3940, 0.0911, and 0.7875 for respective sodium silicate contents of 0.35, 0.20, and 0.05 g/ml. Compressive strength and modulus were plotted as a function of compaction ratio (c) for each binder content (Rbd) in Fig. 9 to find relations of properties with a manufacturing parameter. They appear to increase with increasing compaction ratio expectedly. The least square lines and Pearson correlation coefficients (r) for compressive strength were found to be y = 0.8854x  0.6907 with r = 0.9707, y = 0.5454x  0.4307 with r = 0.9476, y = 0.27 38x  0.2801 with r = 0.9832, and y = 0.1527x  0.15 with r = 0.8765 for respective sodium silicate contents of 0.35, 0.20, 0.05, and 0.00 g/ml. For compressive modulus, they were also found to be: y = 44.095x + 14.938 with r = 0.8423, y = 28.6 16x + 10.851 with r = 0.6130, y = 18.135x  18.474 with r = 0.9428, and y = 4.0795x + 2.6842 with r = 0.6854 for respective sodium silicate contents of 0.35, 0.20, 0.05, and 0.00 g/ml. Thus, the Pearson correlation factor (r) values appear to be higher when compaction ratio (c) is used than when density (qf ) is used, as the independent variable for a constant binder content (Rbd ), for compressive modulus as a function. Compressive strength and modulus (Fig. 10) were again plotted for further information as a function of another manufacturing parameter, binder content (Rbd ) for each compaction ratio. They also appear to increase with increasing compaction ratio expectedly. The least square lines and Pearson correlation coefficients (r) for compressive strength were found to be: y = 5.9348x + 0.3831 with r = 0.9756, y = 3.3964x + 0.2341 with r = 0.9931, and y = 1.8039x + 0.0702 with r = 0.9240 for respective compaction ratio 3.5, 2.5, and 1.5. They were also found to be for compressive modulus: y = 396.15x + 31.085 with r = 0.8339,

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

237

Fig. 9. (a) Compressive strength, and (b) compressive modulus: both as function of compaction ratio for various binder contents in diluted binder before mixing (Rbd = 0.00, 0.05, 0.20, and 0.35 g/ml) without distinguishing particle sizes.

Fig. 10. Mechanical properties: (a) compressive strength; and (b) compressive modulus, as a function of pure binder content (Rbd) for various compaction ratios (c = 1.5, 2.5, and 3.5) without distinguishing particle sizes.

y = 400.12x + 3.0419 with r = 0.9614, and y = 209.55x + 5.534, r = 0.9489 for compaction ratios c = 3.5, 2.5, and 1.5, respectively. Again, the Pearson correlation factor (r) values appear to be higher when compaction ratio (c) is used than when density (qf ) is used, as the independent variable for a constant compaction ratio (c), for compressive modulus as a function. The empirical approach attempted above to find correlations of compressive strength and modulus with three parameters (i.e. qf , c, and Rbd ) seems to have limitations for generalisation due to the difficulty involving multiple parameters even though it may be useful for intuitive analyses and partial description of results. The understanding of mechanical behaviour may be enhanced by a volume fraction analysis as will be given in Section 5.

5. Rules of mixtures and discussion Some representative stress–strain curves of perlite/sodium silicate foams with images captured for failures of specimens used were given in Fig. 11 for various binder contents and compaction ratios. Each sample in picture of Fig. 11 is coded for binder content (Rbd ) and compaction ratio (c). For example, S_0.35_1.57 represents a sample made with a sodium silicate content of 0.35 g per unit ml of diluted binder and a compaction ratio of 1.57. The foams made of low binder contents (0.05 g/ml) for all compaction ratios display

a continuous smooth decrease in stress after cracking at the peak of each curve, indicating that no densification stage exists during compression. Also, it was observed in this case that failure modes are of vertical splitting/shear. In contrast, curves for foams made of higher binder contents (Rbd = 0.35 and 0.25 g/ml) tend to be undulated, and, as the compaction ratio increases, they tend to have lower slopes after the initial peak, indicating that some densification of foam has taken place. Since the failure images indicate failures are dominantly in shear mode (or ‘cup and cone’ mode), the densification of foam during compression at the high binder contents may not be as significant as those with a ‘layered crush’ mode of syntactic foam reported by Kim and Plubrai [24]. However, the curves and failure modes at the high binder contents appear to be similar to some of those for starch/perlite foams reported by Shastri and Kim [21]. Since the foam density is a function of two independent variables (i.e. binder content and compaction ratio), mechanical properties as functions of density (Fig. 7) do not provide a compositional understanding. Hence, the compressive strength and modulus are re-plotted in Fig. 12 as a function of volume fraction of sodium silicate, v b for various compaction ratios. The compressive strength appears to increase linearly with increasing volume fraction of binder (sodium silicate) in foam with an exception in compressive modulus for a high compaction ratio (c) of 3.5. For compressive strength, the following least square lines were found: y = 9.9399x + 0.1022

238

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

with a Pearson’s correlation coefficient (r) of 0.9714 for a compaction ratio (c) of 1.5; y = 17.251x + 0.2979 with r = 0.9818 for c = 2.5; y = 25.311x + 0.5630 with r = 0.9526 for c = 3.5. For a rule of mixtures equation (RME) development associated with data in Fig. 12, it is necessary to find a relative property condition of constituent materials [23]. It is difficult, though, to measure the shear strengths of sodium silicate binder and perlite particles individually to establish the relative property condition. The difficulty, however, may be overcome by knowing the fact that perlite cellular structure consists of very thin walls [21] whereas sodium silicate binder is of solid structure. This leads to a reasonable assumption for a condition for both silica based materials (binder and perlite) that an average shear strength of perlite parti    cles rsP is lower than that of sodium silicate binder rsb . Also, sodium silicate volume fraction is very small due to dilution, providing an approximate range of binder volume fractions applicable for RME and that load carrying capacity (F P ) of perlite is higher than that of sodium silicate binder (F b ) leading to a useful condition [F P > F b and rsP < rsb ] for RME. Here, another consideration for developing a RME, in general, is that compressive strength of   foam rcf may be considered to be a function of three variables i.e. volume fraction of perlite envelope (v pe ), volume fraction of binder between perlite particles (v bi ), and volume fraction of inter-particle voids without binder (v v i0 ). It is possible to reduce the number of variables if we find a relation between two of the variables. To this end, we may follow the assumptions: v v i0 / v bi and v bi / v b for estimation, where v v i ¼ 1  v pe  v bi and v bi ¼ 1  v pm  v tv  v bs (see Fig. 2), given that the higher volume fraction of voids the more capable of holding binder due to capillary action after compaction. Accordingly, inter-particle void fraction when binder is present (v v i ) which is a function of compaction ratio (c) is given by

v v i0 ¼ k1 v bi

ð9aÞ

and

v bi ¼ k2 v b

ð9bÞ

where k1 (>1) and k2 (<1) are proportional constants for a given compaction ratio and

v pe ¼ 1  v v i  v bi ¼ 1  v v i 

v v i0 k1

ð10Þ

:

Then, we may be able to adopt and modify one of RMEs developed   by Kim and Islam [23] for syntactic foam compressive strength rcf when v v i ð¼ 1  v pe  v bi Þ is constant under the condition we found above [F P > F b and rsP < rsb ]: 0

rcf ¼ 2rsf ¼ 2rsb v bi þ 2rsP v pe

ð11Þ 

0

 s

where rsb is the shear stress lower than shear strength rb of binder due to the fact that volume fraction of binder ðv bi Þ is very low, rsP is the average perlite (or microsphere in the case of syntactic foam) shear strength. For a constant compact ratio (c), both v v i and v pe become constants, and if we ignore the strengthening effect of binder inside the intra-particle open pores of perlite particles for small volume fraction of binder, then 0



rsf ¼ rsb k2 v b þ rsP 1  v v i  Fig. 11. Engineering stress versus engineering strain for different compaction ratios: (a) c = 1.5, (b) c = 2.5, and (c) c = 3.5. Three different binder contents are given for each compaction ratio.

v v i0 k1

 :

ð12Þ

Fig. 13 graphically illustrates Eqs. (11) and (12). The term  

rsP 1  v v i  vkv1i0 in Eq. (12) is an intercept indicating an average shear stress in foam due to mechanical inter-locking between

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

239

Fig. 12. Properties as a function of volume fraction of sodium silicate in foam: (a) compressive strength; (b) compressive modulus for all particle sizes collectively for each compaction ratio (c = 1.5, 2.5, and 3.5); and (c) a replot for compaction ratio (c) of 3.5 with different perlite particle sizes to show the limit of increase of modulus occurring at the largest perlite particle Size 3–4 – correlation coefficients (r) were found to be 0.9594 and 0.9378 for combined data of Sizes 1–2 and 2–3, and the initial part of Size 3–4, respectively.

Fig. 13. Graphical illustration by solid lines of Eq. (12) for condition [F p > F b and

perlite particles without binder. The slope of Eq. (12) is high compared to that of Eq. (11) because the slope of Eq. (12) is equal to that of the first term of Eq. (11). The least square lines already shown in Fig 12(a) are represented by Eq. (12). As the compaction ratio (c) or compaction pressure decreases, fragmentation and mechanical inter-locking of perlite particles during compaction may less and less take place but more and more detachment/debonding between particles during compression

rsp < rsb ] and a constant compaction ratio.

testing rather than shear failure would take place, involving bond   strength rsbond between perlite particles. The bond strength  s  rbond may not be explicitly defined because of the complexity of deformation/failure mechanism of debonding, given the fact that foam is macroscopically of shear deformation/failure as observed but the tensile failure mode ultimately at a smaller scale may take place as well as shear deformation. Also, the bonding between particles are not necessarily by one single solid point contact but

240

M. Arifuzzaman, H.S. Kim / Construction and Building Materials 93 (2015) 230–240

multiple points to have the condition rsP > rsbond resembling the failure/deformation mechanism of syntactic foams as already described in reference [23]. Accordingly, the rule of mixtures for low compaction ratios under another condition [F P > F b and rsP > rsbond ] may be given by

rsf ¼ C rsbond v bi

ð13Þ

where C is a constant. The data obtained at the lowest compaction ratio (c) of 1.5 where the intercept is close to zero (Fig. 12(a)) supports Eq. (13). For compressive modulus (Fig. 12(b)), the following least square lines were found: y = 1070.2x + 11.2220 with r = 0.9247 for c = 1.5; y = 2084.5x + 9.2211 with r = 0.9743 for c = 2.5; y = 160 3.1x + 45.466 with r = 0.7730 for c = 3.5. The low r value (0.7730) for compressive modulus for the highest compaction ratio c = 3.5 is obviously due to the flattened region of the data points in which the role of binder for modulus has seemingly reached its limit at around a volume fraction of 0.04. The transitional behaviour at the high compaction ratio (c = 3.5) may be possibly due to: (a) the reasons that the binder volume fraction increase has reached its limit earlier than those at other compaction ratios due to the reduced inter-particle porosity represented by v v i0 ; and (b) the compressive modulus does not require as much binder/solid material as the compressive strength to reach the limit. To validate these reasons, another graph is plotted in Fig. 12(c) for individual perlite particle Sizes. It shows that the flattened region only occurs for the highest perlite particle size (Size 3–4), indicating that other small Sizes have not reached the limit yet. This observation is consistent with the discussions above that the smaller perlite particles tend to contain more binder than the larger perlite particles, supporting the reasons suggested here. Accordingly, the constant k1 in Eq. (9a) may be valid up to the transitional point, allowing us to derive a limited rule of mixtures for compressive modulus. For shear modulus of the foam (Gf ), when the condition of [F P > F b and rsP < rsb ] is applied, both perlite particles and binder contribute to foam shear modulus (Gf ) at the beginning of deformation. Thus, both sides of Eq. (12) can be divided by foam shear strain (cf ) so that, if we ignore the contribution binder within particles for the region where the transitional point has not reached and relatively high compaction ratios, Gf becomes:

  v v i0 Gf ¼ k2 Gb v b þ Gp 1  v v i  k1

ð14Þ

where Gb is the binder shear modulus and Gp is the perlite shear modulus. Also, the foam shear modulus (Gf ) may be converted into compressive modulus (Ef ) if the Poisson’s ratio (m) is constant

Ef ¼ 2ð1 þ mÞGf :

ð15Þ

6. Conclusion Sodium silicate as binder has been characterised for developing perlite foams. Perlite foams as composites with sodium silicate have been manufactured for a foam density range of 0.2– 0.5 g/cm3. One of practical milestones achieved for the composite properties without reinforcement was a density of 0.3 g/cm3 at a compressive strength of 1 MPa. Manufactured perlite/sodium silicate composites have been analysed/discussed for understanding from three different perspectives i.e. manufacturing parameters (i.e. binder content, compaction pressure, and compaction ratio), properties (i.e. particle size, density, compressive strength, and compressive modulus), and volume fractions of constituents. A rule of mixtures applicable for perlite composites for a constant compaction ratio has been developed in comparison with those for

particulate composites with non-compaction. It may be a basis for further development of a rule of mixtures with a variable compaction ratio in the future. Acknowledgments The authors gratefully acknowledge the Newcastle University International Postgraduate Research Scholarship (NUIPRS) and The University of Newcastle Research Scholarship (UNRSC 50:50) provided for Md Arifuzzaman. References [1] Le Maitre RW, Streckeisen A, Zanettin B, Le Bas BM, Bonin B, Bateman P, et al. Igneous rocks: a classification and glossary of terms, recommendations of the international union of geological sciences, Sub-commission of the Systematics of Igneous Rocks. 2nd ed. NY: Cambridge University Press; 2002. [2] Singh M, Garg M. Perlite-based building materials – a review of current applications. Constr Build Mater 1991;5:75–81. [3] Burriesci N, Carmelo A, Antonucci P. Physico-chemical characterization of perlite of various origins. Mater Lett 1985;3(3):103–10. [4] Johnstone SJ, Johnstone MG. Minerals for the chemical and allied industries. 2nd ed. London: Chapman and Hall; 1961. [5] Luongo JS. Strengthened lightweight wallboard and method and apparatus for making the same. US Patent No 6,251,979 B1. 2001. [6] Yilmazer S, Ozdeniz M. The effect of moisture content on sound absorption of expanded perlite plates. Build Environ 2005;40:311–8. [7] Dube WP, Sparks LL, Slifka AJ. Thermal conductivity of evacuated perlite at low temperatures as a function of load and load history. Cryogenics 1991;31:3–6. [8] Kendall T. No sign of the bubble bursting – perlite uses and markets. Ind. Miner. 2000:51–9. [9] Rodsky B. Building Material. US Patent No 2,858,227. 1958. [10] Gray B. Building material. US Patent No. 4,042,406. 1977. [11] Shepherd PB, Dolin RL. Lightweight building material board. US Patent No. 5,256,222. 1993. [12] Hill JH. Perlitic insulation board. US Patent No. 4,126,512. 1978. [13] Aglan H, Morsy M, Allie A, Fouad F. Evaluation of fiber reinforced nanostructured perlite-cementitious surface compounds for building skin applications. Constr Build Mater 2009;23:138–45. [14] Miscall J, Rahr CE. Building board of fiber and asphalt coated perlite. US Patent No 2,626,864. 1953. [15] Sherman N, Cameron JH. Method of manufacturing improved mineral board. US Patent No 4,297,311. 1981. [16] Seybold HG. Wallboard composition and method of making same. US Patent No 2,705,198. 1955. [17] Vimmrova A, Keppert M, Svoboda L, Cerny R. Lightweight gypsum composites: design strategies for multi-functionality. Cement Concr Compos 2011;33:84–9. [18] Topcu IB, Isikdag B. Effect of expanded perlite aggregate on the properties of lightweight concrete. J Mater Process Technol 2008;204:34–8. [19] Colak A. Density and strength characteristics of foamed gypsum. Cement Concr Compos 2000;22:193–200. [20] Skujans J, Vulans A, Uldis I, Aboltins A. Measurements of heat transfer of multilayered wall construction with foam gypsum. Appl Therm Eng 2007;27:1219–24. [21] Shastri D, Kim HS. A new consolidation process for expanded perlite particles. Constr Build Mater 2014;60:1–7. [22] Kim HS. Method of forming syntactic foams. US Patent No 2014/0033953 A1. 2014. [23] Kim HS, Islam MM. Syntactic foams as building materials consisting of inorganic hollow microspheres and starch binder. In: Cornejo DC, Haro JL, editors. Building materials: properties and performance and applications. Nova Publishers; 2009. p. 1–56 [chapter 1]. [24] Kim HS, Plubrai P. Manufacturing and failure mechanisms of syntactic foam under compression. Compos A Appl Sci Manuf 2004;35:1009–15. [25] Islam MM, Kim HS. Novel syntactic foams made of ceramic hollow microspheres and starch – theory, structure and properties. J Mater Sci 2007;42:6123–32. [26] Islam MM, Kim HS. Manufacture of syntactic foams: pre-mold processing. Mater Manuf Processes 2007;22:28–36. [27] Islam MM, Kim HS. Manufacture of syntactic foams using starch as binder: post-mold processing. Mater Manuf Processes 2008;23:884–92. [28] Islam MM, Kim HS. Pre-mould processing technique for syntactic foams: generalised modelling, theory and experiment. J Mater Process Technol 2011;211:708–16. [29] Owusu YA. Physical–chemical study of sodium silicate as a foundry sand binder. Adv Colloid Interface Sci 1982;18:57–91. [30] Karger-Kocsis J. Editorial corner – a personal view: water glass – an alternative precursor for sol-gel derived silica nanofiller in polymer composites? Express Polym. Lett. 2014;8(12). 880-880.