You are on page 1of 7

Construction and Building Materials 23 (2009) 19151921

Contents lists available at ScienceDirect

Construction and Building Materials


journal homepage: www.elsevier.com/locate/conbuildmat

Effects of powder concentration and type of resin on the performance of marble composite structures
C. Borsellino a, L. Calabrese b, G. Di Bella c,*
a

Department of Ingegneria Civile, University of Messina, Italy Department of Chimica Industriale e Ingegneria dei Materiali, University of Messina, Italy c Department of Tecnologia Meccanica Produzione e Ingegneria Gestionale, University of Palermo, Viale delle Scienze, 98166 Palermo, Palermo, Italy
b

a r t i c l e

i n f o

a b s t r a c t
In this work, we will expose the behaviour of a composite structure reinforced with marble powder. The attention was focused on the effects produced on the physical and mechanical properties due to different kinds of matrix (epoxy and polyester resins) and ller amounts (60%, 70%, 80%). Rheological, static exural and Izod impact tests have been carried out. An experiment designed was dened for the mechanical tests, in order to quantify the effects of both the resin and the marble percentage on the material properties, as well as to identify a possible correlation between these factors. Moreover, water absorption and stain resistance have been evaluated to study the surface properties of the panels. 2008 Elsevier Ltd. All rights reserved.

Article history: Received 17 July 2007 Received in revised form 4 September 2008 Accepted 5 September 2008 Available online 11 October 2008 Keywords: Marble powder Composite Mechanical properties Water absorption Stain resistance

1. Introduction Marble is a metamorphic rock, composed mainly of calcite (a crystalline form of calcium carbonate, CaCO3) and other minerals, such as muscovite, sericite and chlorite. It has a very low content of quartz (0.2%), and its density is 2.7 g/cm3 and its porosity is approximately 0.4% [1]. It is extensively used in the architecture as a building material, in the arts for sculpture and in many other applications [2]. Because the word marble is colloquially used to refer to many other stones that are capable of taking a high polish showing different composition, origin and commercial properties, the UNI 8458 standard dened marble as: every crystalline rock constituted mainly by minerals with a Mohs hardness equal to 34. Marble results from regional or at times contact metamorphism of sedimentary carbonate rocks, either limestone or dolostone. This metamorphic process causes a complete recrystallization of the original rock into an interlocking mosaic of calcite and/or dolomite crystals. The temperatures and pressures necessary to form marble usually destroy any fossils and sedimentary textures present in the original rock [24]. In the recent years the marble industry reached an important development. As a consequence, the volume of scraps produced

* Corresponding author. Tel.: +39 0903977286/3494628299; fax: +39 0903977464. E-mail addresses: c.borsellino@ingegneria.unime.it (C. Borsellino), calabrese@ ingegneria.unime.it (L. Calabrese), gdibella@ingegneria.unime.it (G. Di Bella). 0950-0618/$ - see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.conbuildmat.2008.09.005

has increased, i.e., in this period about 2 million tons of scraps have been produced from granite and marble production (70%/ 75% of whole extract material). The rock fragments, solid, not degradable and unsolvable, are accumulated in quarry dumps, called ravaneti. These are near the extraction areas and they are classied in historic and present. The ravaneti are the result both of excavation of marble with use of explosive devices (i.e. historic; this technique is now obsolete and it is replaced by more upto-date techniques) and of new scraps deposits (i.e. present). The present ravaneti covers 5 millions square meters. Their presence hinders the exploitation of about 33% of the marble quarries. It also increases the landslide risk. As a consequence, it needs to build the retaining structures. The processing mud is often accumulated in those quarry dumps that blocks the vegetation growth. Moreover, all ravaneti have also been used as a source of both inert material and calcium carbonate, without any planning or regulation. The historic ravaneti has become an integral part of the landscape: their whiteness characterises the mountainous massif (i.e. Apuan Alpes). Moreover, these ravaneti symbolize the past extractive activity. Sometimes they show traces of quarrying work (i.e. the vie di lizza, rocky chutes used by quarry workers to move the marble blocks). They slow down the rainwater course towards the watercourses in order to protect habited centres against oods. For this fact the historic ravaneti are protected [5]. Actually the marble can be pulverised to obtain calcium carbonate for industrial applications, i.e., paint production, paper industry. It is also possible to obtain a thinner powder to produce

1916

C. Borsellino et al. / Construction and Building Materials 23 (2009) 19151921 2.2.4. Water absorption According to UNI 10444, the surface/volume ratio of the samples was equal to 0.10.2 mm1. These were initially heated in an oven at 105 C until their mass reaches a constant value M0. Then they were fully immersed in the water. Finally the mass Mi was measured by varying the immersion time ti (i.e. 1, 81, 24, 48, and 312 h). Five samples were tested for each resin with 70% of marble. It is possible to evaluate the water absorption by the following equation:

concretes [6]. This technique (i.e. using the process scraps) can allow to reduce the environmental impact of the new ravaneti. In this work, the marble powder has been used to make a composite material, with a low cost technology. This should be commercially promoted in order to increase the market of ravaneti marble. In fact, the powders are used directly to obtain very expensive materials (i.e. okite) for civil applications. These are also employed as rigid llers [7]. In the rst case, the material shows very high properties (i.e. powder fraction to 93%) but for the high production costs the distribution is limited. In the second one the powder is used to increase the elastic modulus of a polymeric composite. The attention is focused on the effect of the powder amount, the shape of particles, and the relative modulus. The llers concentration is lower to 30% or 55% [7,8]. The aim of this work is to allow the exploitation of the marble powder to make a structural material characterised both by low costs and by a high concentration level however, with good properties. To evaluate the performances of this material rheological, static exural and Izod impact tests have been carried out. A design of experiments for the mechanical tests has been dened. Moreover, water absorption and stain resistance have been evaluated to study the surface properties of the panels.

C i 100

Mi M0 M0

where Ci is the percentage of absorbed water 2.2.5. Stain resistance According to UNI EN 423, the samples had a surface of at least 3000 mm2. For the liquid substances (i.e. coca cola, detergent, wine, lemon, coffee, oil), a small amount was poured on the specimen. Then it was pressed under a glass to create a stain with a surface varying between 300 and 400 mm2. For the doughy substances (i.e. mayonnaise, butter, ketchup), a small volume was spread on a surface varying between 100 and 300 mm2. After 2 h the stains were cleaned. Then the samples surface was analysed by a distance of about 800 mm with an angle of 45, rotating slowly the observations plane. If the stain was visible the test was repeated for 30 min with the same procedure. A graduated scale allows to classify the alteration degree (Table 2). 2.3. DOE In order to verify the inuence of several parameters on both exural and impact tests, the DOE (design of experiments) method was applied. The safety measures, adopted in the phase of sample preparation, have allowed to eliminate such process factors, as marble grains sizes and resin curing time. Two parameters were considered: the powder concentration and the resin type, adopting three and two variability levels, respectively (i.e. three values of concentration and two types of resins). The experimental results were analysed through a two way analysis of variance.

2. Experimental setup 2.1. Materials The marble powder, used to make the composite samples, comes from Massa Carrara. It is constituted mainly by calcite. The properties both of the polyester resins and of the epoxy one, used as matrix, are reported in Table 1. In the next these are identied respectively with PE and Epo. The panels, having sizes 200 100 10 mm, were made in a wood mould after a homogeneous mixing of powder/resin. To avoid marble deposits on a specimen side, the mould was in rotation until full cure of the matrix occurs. Samples with three different concentrations of marble (i.e. 60%, 70% and 80%) were realised. 2.2. Experimental tests 2.2.1. Rheological test The rheological properties during curing were measured by a parallel plate, stress controlled, rheometer, SR5 supplied by Rheometric Scientic. A liquid resin volume, with different marble powder concentrations, of about 0.51.0 cm3 was placed on an aluminium plate (diameter, 25 mm). Isothermal tests were carried out with a stress of 1000 Pa at 1 Hz frequency. 2.2.2. Static exural test According to UNI 10443, the samples (100 10 5 mm) were placed on two supports with a span length of 90 mm. The test was performed at room temperature using a LLOYD Instruments LR10K UTM with a 10 kN load cell. The cross-head speed was 1 mm/min. Five samples for each concentration and for each resin were tested. 2.2.3. Izod impact test This test was carried out using an impact tester (Resil Impactor 50 by Ceast). According to ASTM D256, the sample sizes were 11.5 7.5 5 mm, the notch length was 2.5 mm. Five impact tests for each concentration at varying the resin were conducted at room temperature evaluating the impact energy.

3. Results 3.1. Rheological tests In Fig. 1 the viscosity trends are reported for both resins with varying the marble concentration. For the sample PE, it is possible to observe three different zones: i. initially the viscosity slightly decreases, this reduction is due to the resins rheological behaviour. As a consequence the system ows with lower difculty by increasing the frequency; ii. the viscosity values reach a plateau; iii. nally the system is unstable and the measures are scattered. With increasing the marble concentration the viscosity is higher because the powder presence reduces the ow capability. As a consequence, the workability of the system is lower. For the sample Epo, the trend is always constant because the high percentage of catalyst increases the ow capability. As for samples PE, for high frequency, the system is unstable. Also, for this resin the rheological parameter increases with the marble concentration. Comparing the graphs the difference between the two systems is evident: the viscosity for the resin PE is always higher than the Epo one.

Table 1 Resins properties Resin Catalyst Catalyst amount [%] Aspect Viscosity (25 C37 s1) [poise] Stability (in the dark 20 C) [months] Tensile stress [MPa] Young modulus [MPa] Break elongation [%] Polyester Butanox M50 2 Blue 5.3 6 57 3700 1.8 Epoxy SD 8822 31 Yellow 7.0 24 70 3000 3.8 Table 2 Classication of the stains Index 0 1 2 3 4 Alteration degree No alteration Very slight Slight Alteration Heavy

C. Borsellino et al. / Construction and Building Materials 23 (2009) 19151921

1917

Fig. 2. Tan d trend for PE resin.

3.2. Static exural test In Fig. 3 the stress/strain curves for both resins are reported with varying the marble concentration. For the samples Epo it is possible to observe three regions: i. the trend is linear denoting a perfectly elastic behaviour of the material; ii. the curve changes. It is always linear but with a lower slope due to both the cracks start and their propagation in the region around the pin; iii. nally the break occurs catastrophically.
Fig. 1. Viscosity for both resins: (a) polyester; (b) epoxy.

Due to this rheological behaviour the workability of the samples with resin Epo is higher. Moreover a sedimentation effect can be observed, according to the Stokes lags:

2R2 q q0 g

where R is the marble particles radius; q and q0, respectively, the marble and resin density; g the gravity acceleration and g the system viscosity. For this reason it is possible to obtain two different structures with different properties: in the samples Epo it is possible to nd regions with high marble density mixed with pure resin due to both the high workability and the sedimentation effect. In the samples PE the marble particles are more uniformly distributed in the resin. In Fig. 2 the trend of tan d for the resin PE is reported with varying the frequency. It is possible to observe that tan d is always higher than 1. Then the loss modulus G00 prevails on the storage modulus G0 . Moreover two regions are identied: i. at low frequencies, there is an evident effect of the marble percentage; i.e. the loss modulus decreases with increasing the marble concentration. The solid particles presence allows the creation of a more stiff structure; ii. at high frequencies the behaviour is more elastic and the effect of the marble percentage is less evident. It is possible to identify a slight reversal because a higher amount of powder creates a continuous phase reducing the previous elastic behaviour. The resin Epo is characterised by a similar behaviour.
Fig. 3. Stress/strain curves with varying the marble concentrations: (a) epoxy; (b) polyester.

1918

C. Borsellino et al. / Construction and Building Materials 23 (2009) 19151921

By increasing the marble concentration the material behaviour varies. In fact the slope of the rst region (see Table 3) slightly increases. The stiffness of the samples improves. Moreover it is possible to observe that both the point of cracks start and the slope of the second region are strongly modied. Also the maximum stress and strain are different, as reported in Table 3. This fact occurs because a high amount of powder increases the defectivity. As a consequence, despite the modulus initial increase, the cracks start at lower strain values reducing the performance of the structure. The resin PE shows a different behaviour: the curve increases almost linearly until the catastrophic rupture of the sample occurs. The cracks produce the break without an evident propagation. Also in this case by varying the concentration the slope slightly increases, but the maximum stress and strain are lower (see Table 3). Then by increasing the marble amount the composite properties are evidently reduced due to the higher amount of defects, created in the production phase (i.e. mixing and curing). It is evident from the analysis of the fracture surfaces, i.e., for a sample with a 70% concentration (see Fig. 4). Also for the resin PE the stiffness of the samples slightly increases with the marble concentration. The sample Epo shows both a non-catastrophic break and a higher strain than the PE ones due to its structure. At the same time the maximum resistance is lower than the samples with the other resin due to the mixing between two different phases (resin/powder). The sample PE is less workable due to the higher viscosity. It presents a structure with evident empty spaces that produce the immediate rupture as the cracks start. Moreover, the mobility of marble particles, during mixing, is reduced. As a consequence the structure is more homogenous reaching the higher stresses. The empty spaces are characterised by higher sizes for high marble percentage. Then by increasing the marble amount the performances are strongly reduced. 3.3. Analysis of variance The rst step of the analysis of variance is to look for a correlation between the variables (resin, concentration). In Fig. 5 an interaction is not evident (i.e. the straight lines connecting the data are parallel). But performing a two way analysis of variance with interaction (Table 4), it is possible to observe that the interaction is signicant (P-value is lower than 0.05). Analysing the effects separately, Table 4 also reports that both the resin and the concentration are signicant factors (P < 0.05).

Fig. 5. Stress/marble% curves.

Table 4 Statistical parameters (Flexural test) Source Resin % Marble Interaction Error Total DF 1 2 2 24 29 SS 286.32 822.56 24.75 28.76 1162.39 MS 286.319 411.280 12.373 1.198 F 238.90 343.17 10.32 P 0.000 0.000 0.001

This fact is also evident from the analysis of means. Fig. 6 shows that an interaction effect exists. In fact the mean stress values are outside of the condence range. Also the effects of both the resin and the treatment are signicant. In conclusion (see Fig. 7) it is not possible to identify two different populations because the interaction effect is not negligible. 3.4. Izod impact test Fig. 8 reports the impact energy for unit cross-section area by varying the powder concentration for both resins. For the resin Epo the impact strength reaches the maximum value (0.261 J/ cm2) with 60% of marble, it decreases upon increasing the concentration to 70% (0.232 J/cm2 with a reduction of 11%) and 80% (0.193 J/cm2 with a reduction of 26%). This behaviour is due to the powder distribution in the matrix. In fact at 80% the structure is defective.

Table 3 Flexural properties % Marble Young modulus [MPa] Epo 60 70 80 4861 5060 8145 PE 7333 7980 9079 Maximum stress [MPa] Epo 22.2 17.7 10.6 PE 30.7 21.8 16.6 Maximum strain Epo 0.0070 0.0066 0.0059 PE 0.0054 0.0038 0.0025

Fig. 4. Fracture surfaces for PE70.

Fig. 6. Analysis of means (a = 0.05).

C. Borsellino et al. / Construction and Building Materials 23 (2009) 19151921

1919

Fig. 7. Boxplot for the stress. Fig. 9. Analysis of means (a = 0.05).

outside the condence range. Both the resin and the concentration act as signicant factors. 3.6. Water absorption Fig. 10 reports the water absorption percentage Ci as a function of the time. It is possible to dene two regions: absorption and stabilisation, according to the Ficks law. Initially the absorption increases with a constant slope until the values reach a plateau at about 0.16% for the resin Epo and 0.25% for the resin PE. The trends are similar for both resins with the same slope for the rst region and the same time of plateau start. But the sample PE shows a higher absorption due to the higher absorption of the polyester resin than the epoxy one. As a consequence after the maximum immersion time of the samples PE are completely degraded.
Fig. 8. Impact strength by varying the marble concentration for both resins.

3.7. Stain resistance Fig. 11a reports the alteration degree for a sample Epo with 70% of marble. After the rst treatment (2 h) different stains are clearly visible on the surface (detergent, oil, wine and butter). The last ones produce a slight alteration. Repeating the procedure for these stains in a second treatment (30 min) only the oil and the butter stains are clearly visible but the alteration is very slight. The sample Epo with 60% of powder shows a high stain resistance. In fact the investigated substances do not produce an alteration on the surface. In this case the higher resistance is directly

The behaviour of the resin PE is similar, but the values are lower. This fact is evident analyzing the area below the exural stress/ strain curves (Fig. 3). 3.5. Analysis of variance To understand this different behaviour and to evaluate how the parameters (resin and concentration) inuence the impact material properties, the analysis of variance is carried out. Fig. 8 shows that actually an interaction is not present (i.e. the straight lines connecting the data are not parallel). Performing a two way analysis of variance with interaction (Table 5), it is possible to observe that the interaction is signicant (P-value is lower than 0.05). Analysing the effects separately, Table 5 shows that resin and concentration are signicant factors (P < 0.05). This is also evident by the analysis of means. Fig. 9 shows that an interaction effect exists. In fact the mean stresses values are

Table 5 Statistical parameters (Impact test) Source Resin % Marble Interaction Error Total DF 1 2 2 24 29 SS 0.0183027 0.0114355 0.0024264 0.0010788 0.0332434 MS 0.0183027 0.0057177 0.0012132 0.0000449 F 407.18 127.20 26.99 P 0.000 0.000 0.000

Fig. 10. Absorbed water in the time.

1920

C. Borsellino et al. / Construction and Building Materials 23 (2009) 19151921

Fig. 12. Lemon effect on the sample PE80.

is signicant producing a surface damage. Fig. 12 reports the alteration of the lemon on the sample PE80.

4. Conclusions From experimental study it is possible to draw out that:


Fig. 11. Alteration degree: (a) sample Epo70; (b) sample PE70.

due to the resin stain resistance. The samples Epo with 80% of marble are more altered than the previous ones. For some substances (wine and coca cola) the alteration is signicant. High defectivity reduces the stain resistance. Fig. 11b reports the behaviour of the sample PE with 80% of resin. In this case after the rst treatment, lemon, oil and ketchup produce a slight alteration. Only for the wine the surface remains with a very slight alteration after the second treatment. For the samples PE with 60% of marble the stain resistance is high. Whereas, with 80% of powder there is a higher alteration (i.e. signicant for the lemon). Table 6 reports the alteration values for the samples Epo and PE after the rst treatment and the second one when it is required. It is possible to observe that the stain resistance decreases with increasing the marble concentration. Only for some substances (coca cola and wine for Epo80 and lemon for PE80) the alteration

The viscosity of the system resin/marble is higher for the PE samples than the Epo ones. It increases with the powder concentration. The PE samples are then characterised both by a low workability and by a high sedimentation effect. In the exural test the Epo samples show an elasticplastic behaviour. The PE ones are characterised by an elastic trend with a brittle fracture. The maximum stress values (22.2 for Epo and 30.7 for PE) decrease with the marble concentration. The exural strength is inuenced signicantly both by the resin and by the marble concentration. Moreover an interaction between these factors exists. The impact behaviour is similar for the two systems. The samples with resin Epo are characterised by a higher resistance. There is a signicant inuence on the impact behaviour both by the resin and by the concentration factors. The interaction of resin/concentration is evident. The samples Epo shows a lower water absorption than the PE ones due to the higher absorption of the polyester resin than the epoxy one. The stain resistance decreases with the marble concentrations. The alteration is signicant for the samples with 80% of marble (coca cola and wine for Epo80 and lemon for PE80). The samples, realised with epoxy resin and with a percentage of marble equal to 60%, are characterised by higher properties than the monolithic marble [9]. For the last one with a low cost technology a structural material, characterised by good performances, has been realised. References
[1] Maurigiannakis S, Mitritsakis N, Agioutantis Z, Anastasiadis C, Triantis D, Stavrakas I. Investigation and quantication of damage in geomaterials with the technique of dielectric spectroscopy. In: Proceedings of the 9th European conference on NDT. Berlin; 2006. Poster 222.

Table 6 Stain resistance level Substances Epo60 I Coca cola Coffee Detergent Lemon Oil Wine Butter Ketchup Mayonnaise 0 0 0 0 0 0 0 0 0 II Epo70 I 1 0 2 0 2 2 2 0 0 II 0 0 1 0 1 Epo80 I 3 2 0 1 0 3 0 2 0 II 2 2 0 2 1 PE60 I 0 1 0 1 0 1 0 0 0 II 0 0 0 PE70 I 0 0 1 2 2 1 0 2 0 II 0 0 0 1 0 PE80 I 2 2 0 3 2 1 1 2 0 II 1 2 1 1 1 0 2

C. Borsellino et al. / Construction and Building Materials 23 (2009) 19151921 [2] Ashurst J, Dimes F. Conservation of building and decorative stone. Oxford: Butterworth Heinemann Ltd.; 1990. [3] SME and NIOSH of Pittsburgh. Rock and mineral. Pittsburgh; 2006. [4] Marble Institute of America. The beauty of natural stone elegant, enchanting, enduring and more affordable than ever before. Architectural Record 2002;4: 2015. [5] Damato Avanzi G, Verani M. Le valenze ambientali ed economiche dei ravaneti delle Alpi Apuane (Toscana, Italia). Mem Soc Geol Italiana 1998;53:489501. [6] Corinaldesi V, Moriconi G, Naik TR. Characterisation of marble powder for its use in mortar and concrete. In: Proceedings of the CANMET/ACI international

1921

symposium on sustainable development of cement and concrete: Toronto; 2005. Report CBU-2005-06-09. [7] Goyanes SN, Marconi JD, Konig PG, Martin MD, Mondragon I. Dynamical properties of epoxy composites lled with quartz powder. J Alloys Compd 2000;310:3747. [8] Valeva V, Hristova J. Numerical prediction of the strain state of thermoset matrix/mineral ller composite plates. Mech Compos Mater 2005;41(2): 97104. [9] Cardani G, Meda A. Behaviour under monotonic and cycling loading in tension. Constr Build Mater 2004;18:41924.