Waste Glass in Porcelain

ln this work, lhe fe1dspar used as a fiux in typical porcelain was replaced by waste glass , and lhe resulting batch composition contained 50% kaolin, 25% quartz and 25% waste glass. The properties oflhis porcelain wilh glass powder (GP) were compared to traditional porcelain (TP), which utilizes feldspar. The samples were fired at different temperatures, ranging from 1200 °C to 1400 °C. Technical parameters, such as water absorption, modulus of rupture and K1c, phase analysis by XRD, and microstructure by SEM were analysed. The results indicated lhat lhe lower fracture toughness and MOR of GP is a consequence of lhe peculiar lnicrostructure of lhis porcelain. The reduction in firing temperature and lhe use ofa cheaper substitute for feldspar makes GP also an attractive economical altemative.


Introduction
The choice of recyclable materiais to replace conventional raw materiais must be made carefully.The first step is to analyze lhe decomposition of lhe materiais during lhe firing process due to lhe possible release of poisonous gases.Thís will probably elilninate lhe use of polymeric wastes such as tires, plastíc bottles, etc. ln~ organic materiais may release heavy metais, fiuorine and sulphur, whích are very harmful to human heallh.An expensive alternative would be lhe treatrnent of lhe fiue gas; however, lhese emissions can also cause serious darnages on kiln refractory bricks.The gas elnission levels, which wíll probably be even more restrícted in lhe future, since lhey are controlled by environmental standards, must be considered too.The second step is to make sure lhat lhe cerarnic product is inert during usage .This is especially important in lhe case of dinnerware, sanitary ware, and so on.Leachíng can be a serious problem for fioor tiles and roof tiles and also límíts lhe choice of materiais lhat can be used in body composition.Thus, lhe use of recycled colorless glass in lhe substitution of feldspar in lhe porcelain production, as proposed in lhis work, seems to be very suítable and envíronmentally safe.
As an initial críterion, and prior to make any experimental tests, lhe availabílity of recycled glass powder to lhe industry was checked.ln Brazil, 890,000 tons of glass per year are produced; so, lhere is a hu ge potential of recycled glass powder availabilíty.The total amount of glass collected in lhe cities for recycling exceeds 339,000 tons.Thís could be significantly increased by governmental incentives.The quantity of glass on lhe urban waste in lhe United States is 6.6% of lhe total weight, and is equivalent to 11.3 lnillion tons of glass 1 .The aim oflhis work was to evaluate lhe qualíty oflhe porcelain tllat can be made using glass powder.Samples were produced and fired in laboratory scale.Technical pararneters such as water absorption, shrinkage, modulus of rupture, apparent porosíty, densíty, weight loss and fracture toughness were deterlnined.The phases present in lhe lnicrostructure were evaluated using Scanning Electron Microscopy and X~ray diffractometry.Glass powder was used in lhe second batch replacing feldspar, which is normally used in traditional compositions.The main observations showed tlle role oflhe glass powder as a fiuxing agent, whích could be easily compared to TP.Table 1 shows lhe chemical analyses of lhe raw materiais.

Experimental
The glass powder was obtained from hígh transparency soda~lime glass pots and bottles wilhout colouring oxides.The glass was ball lnilled and sieved to 270 mesh ( < 53 f.llll) .The crystalline phases in raw materiais were deterlnined by XRD.For kaolin, lhe results showed lhe presence of kaolinite (Al 2 Si 2 05 (OH) 4 ) as tlle main crystalline phase, and muscovite ((K Al 2 (Al Si 3 0 10 ) (OH)) plus íllite (K Al 3 Si 3 0 10 (OH) 2 ) as secondary crystalline phases.Kaolin was sieved to pass 325 mesh (<45 f.llll).For feldspar, whích was sieved to pass 270mesh (<53 f.llll), main crystalline phases were lnicrocline (KA1Sip 8 ) and albíte (NaA1Si 3 0 8 ) .Quartz was sieved to 325 mesh (<45 f.llll), whích is lhe parti ele size used in most Brazílian industries.The raw materials se1ected were dry milled in a ba11 mill during 5 minutes in order to homogeníze the mixture.The millíng time necessary to achíeve the optimum samp1es properties was previous1y deterrníned.Eíght percent of forrníng water was then added and the mixture sieved to 20 mesh prior to pressing.The dímensions of the samp1es were 8 mm x 20 mm x 60 mm for techníca1 characterization and 8 mm x 1 O mm x 60 mm for K 1 c and modu1us of rupture (four poínts bendíng strength) deterrnínation.The samp1es were dried for 48 hours in aír and then at 11 O oc for 24 hours in an e1ectric fumace.
No further change in weight occurred during this time, since the specímens were already dry.The dried pieces were then weighed and measured to obtaín weightloss and dry densíty.The firíng took p1ace in an e1ectríc fumace at dífferent temperatures between 1200 oc and 1420 oc with 150 K.h-1 heating rate and soakíng time of 30 minutes.ln order to obtaín comparative results, samp1es with the sarne dry densíty (1.80 g/cm 3 ) were used in both batches.Such densíty was easíly obtaíned for both formu1ations by changíng the quantity of forrníng water and forrníng pressure.

Results and Discussion
G1ass powder is a strong fiuxing agent and has the capacíty to forrn a 1ower me1ting point sílica te.Therefore the firing temperature range was re1ative1y narrow and dífferent from that of tradítional porce1aín.Tab1e 2 and Figure 1 show the technical characteristics of the fired pieces.
From the results shown in Tab1e 2 and Figure 1, ít is possib1e to consider the temperature of 1240 oc as the best firing temperature for g1ass porce1aín and 1340 oc for tradítional porce1aín, considering linear shrinkage and bu1k densíty values and a very 1ow value of water absorption.
It can be seen in Figure 1 that the use of g1ass powder resulted in a 1ower firing temperature for GP, approxímate1y 1 00 oc 1ower than for TP, but within a shorter temperature range to tire GP.Tab1e 2 shows that 0.39% of water absorption and 8.8% oflinear shrinkage occurred for the 1240 oc GP samp1es, whí1e the 1340 oc TP samp1es attaíned values of0.34% water absorption and 12.2% oflínear shrinkage.The 1ower firíng temperature is an advantage for GP and confirms that the glass powder is a strong fiux.But the short temperature range for firing GP indícates that more accurate temperature contro1 is required during the firíng process of this porce1aín.Microstru ctural ana1ysis revea1ed which factors were more important to govern the final shrinkage and densíty of the píeces (Figure 2).The presence of a 1arger quantity of gas bubb1es trapped in glassy phase compared with previous temperature had a stronger infiuence.TP samp1es had the sarne behaviour, (Figure 3), butin thís case, b1oating occurred at higher temperature and a 1arger firing range is possib1e for thís composition.
The highest obtained densíties were 2. 28 glcm 3 for G P and 2.48 g/cm 3 for TP at the best firing temperatures of 1240 oc and 1340 oc, respective1y.These were hígher than the densities obtaíned at 1ower temperatures.The decrease in densíty values with increase in temperature -over firing -cou1d be exp1aíned by the presence of trapped gas in the g1assy phase as mentioned before.Thus , densíty, water absorption and apparent porosíty a11 show the sarne tendency, regardíng b1oating, with increasing firing temperature .The hígher bubb1e díarneter on GP samp1es exp1aíns the 1ower densíty of this composition when compared to TP.The resu1ting microstructures dueto firing at these temperatures (1280 oc and 1420 oc, for GP and TP, respective1y) are characteristic of an over fired porce1aín and the excess of heat exp1aíns the physical properties showed in Tab1e 2.
The EDX analysis was taken símu1taneous1y with microstructure observations, whích supported the phases identification.The results of thís analysis are shown in Figures 3 and 4. B ased on phase chernícal analysis and phase morpho1ogy, w hích are well documented in literaturel-6 , ít was possib1e to identify the phases named in Figure 4 and 5.
XRD analysis is showed in Figure 6, for the optimum firing temperature of each porce1aín.
The maín dífferences between the two porce1aíns are the presence of anorthíte (CaA1 2 Sip 8 ) in GP (Figure 6) and the presence

Tradítional Porce1aín
G1ass powder porce1aín of secondruy mullite in TP (Figure 5).The other crystalline phases, mullite (2Si0 2 .3Alp 3 )andquartz(Si0z)werepresent in thestmcture for both fotmulations.The highest peak confumed the existence of a greater amount ofthemullite phase in TP (Figure 6).The peak height of quartz is approximately the sarne in the GP aJJd TP samples, aJJd it means that a significaJJt amount of residual quartz did not suffer dissolution.Image aJJalysis, performed according to the method suggested by VaJJ Vlack 7 , was used to quantitatively evaluate phase composition.It confumed that the amount of residual qurutz was high, indicating approximately 20% of qua1tz for the trad itional fotmulation and 16% of quartz for glass porcelain.
Anmthite is an unusual phase in porcelain.The crystallization of th is phase within GP composition is a consequence of the presence o f CaO in powdered soda-lime glass.Porcelain usually containsquartz, mullite aJJd a glassy phase, as shown in Figures 5 and 6 for TP.
The importance ofthe glass phasecomposition, as well as the low aluminurn content in themelt, must be highlighted as a peculiru•ity of the use of soda-lime glass instead of feldsparo As a consequence of that, secondary mullite could not crystallize in the melt, as usually occurs with feldspar porcelaino However, the crystallization of sec-ondaJY mullite from primary mullite could be occasionally observed, as shown in Figure 70 3.lo Fracture toughness and bending strength The mechaJJical behaviour of the ceramic bodies were aJJalysed in this work in terms of fracture toughness and modulus of rupture (four points bending strength)o Fracture toughness aJJd fracture energy pru•ameters aJJd the criticai fiaw size were calculated according to Hübner 8 o The meaJJ value of~c was 1.6 MPa m 1 n for the TP samples aJJd 1.3 MPa m 1 n for GP samples (fables 3 and 4)0 These are acceptable values for porcelaino Porcelain usually has a ~c raJJge of between 1 aJJd 2 MPa m 112 90 The modulus of mpture (bending   strength) was high enough for porcelain, since they are suitable even for the porcelain stoneware (minimum strength 35 N/mm 2 according to the ISO 13006-B1a) 10 • When comparing the values for the two porcelains it can be seen that TP has a higher K 1 c and a higher MOR.However, the mean crack length was almost similar (200 j.l.m) for both porcelains.Therefore, there is another factor besides crack length that has a strong inftuence on strength, which is responsible for the higher energy necessary to rupture TP.This factor is fracture energy ('\'), which has avalue of 16.4 J/m 2 for TP and 10.7 J/m 2 for GP.Fractw•e energy is related to the energy conswned dw•ing crack propagation.This energy conswnption is due to the presence of barriers that increase the crack path until the complete rupture of the sample.This phenomenon is widely explained in literature 11 • 12 • The presence of a higher content of crystalline phases, mainly qua1tz, primary and seconda~y mullite !.:? (as indicated in themicrostructure and X-ray diffractogram analysis), and the higher density ofTP account for the higher strength ofTP.The origin ofthis ftaw is explained bythe size of quartzparticles used in the batch formulation.The stresses released due to la~•ge quartz particles and matrix the1mal expansion mismatch cause this fiaw.The size of the criticai fiaw is explained by the linking ofthe cracks originated around the qua~tz particles, as stated in the literature 13 -17  and showed here in Figure 8.

Conclusions
The comparison between glass porcelain (GP) and traditional porcelain (fP) shown that the use of recycled soda-lime glass powder as a ftuxing agent replacing feldspar in a porcelain formulation is possible.Porcelain with excellent material properties was produced.Small differ•ences in material properties were observed when compared to a traditional porcelain.
The lower firing temper•ature of GP (1240 °C) than that of TP (1340 °C) was the most important advantage of GP.This could cer-   tainly bring a cost reduction in the production process and make the use of glass powder an economical attractive altemative.The cwve of water absorption and shrinkage versus firing temperature indicated that glass powder behaved similarly to a strong fiux.This has an advantageous e ffect of lowering the firing temperature, but decreases the temperature range for sintering.Glass porcelain demonstrated to be very susceptible tobloating, whathaddeleterious effects to glass porcelain properties in higher firing tem per atures.
XRD analysis revealed the presence of anorthite in GP, as well as the mullite and quartz phases also found in TP porcelain.The crystallization of anorthite was a consequence ofthe calcium oxide within the soda-lime composition.
A fracture toug hness ~c) of 1. 6 MPa m 112 anda fiexural strength of59N/mm 2 wereobtained fortheTP samples fired at 1340 oc.For the GP samples fired at 1240 °C a K 1 c value of 1.3 MPa m 112 and a MOR of 49 NfmmZ we re obtained.These results are good values for a fine ceramic.The lower fracture toughness and MOR of GP is a consequence of the peculiar m icrostructure of this porcelain.

Figure 1 .
Figure 1.Water absorption and shrinkage as a function of peak temperature , s howing the difference in optimum firing temperature between g1ass porce1ain (GP) and traditional porcelain (TP).

Tab1e 2 .
Water absorption, bulk density and shrinkage.Comparison between traditional porcelain (TP) and glass powder porcelain (GP) related to firing temperature.

Figure 2 .
Figure 2. SEM photomicrograph of glass porcelain fired at 1240 oc and 1280 °C.B loating as a consequence of temperature increasing.

Figure 6 .
Figure 6.X-ray diffractogram of GP fired at 1240 oc and TP fired at 1340 °C.

Figure 8 .
Figure 8. SEM photomicrograph of unpolished and non-etched glass porcelain surface, show ing that crack bridging can occur dueto the stresses released from large quartz parti eles.
Remark: "c'" was obtained with a diamond disc."c" results from theoretical calculations.