Sintered glass ceramic articles from plasma vitrified asbestos containing waste

Introduction

Ceramics have a huge potential for environmental clean-up since they may offer, with proper compositions, the possibility of immobilising pollutants and producing new engineering materials.1 The formation of at least partially crystalline materials, by direct sintering of inorganic waste with additives, i.e. ‘ceramication’,2^–4 and the development of glasses, by melting of inorganic waste with additives, i.e. ‘vitrification’,5,6 have been generally considered as two separate options. This paper aims at providing a combination of the two approaches, starting from a specific type of waste, i.e. asbestos containing materials, vitrified by application of plasma technology.

Thermal plasma melting is one of the most promising technologies for the treatment of hazardous waste due to many favourable features, such as flexibility, efficiency in destroying organic compounds, lack of waste products and compactness of plants.7^–11 One fundamental drawback of plasma technology is represented by the generally high costs that could be justified when treating very hazardous waste, like the just mentioned asbestos containing materials, i.e. asbestos reinforced concrete, asbestos panels, etc.12 It is well known that the hazardousness of asbestos is due to the particular fibrous morphology that may be destroyed upon treatment at high temperatures.13 Plasma technology is particularly suitable for the treatment of asbestos containing waste since the melting undoubtedly ensures the destruction of asbestos fibres; in addition, the high achievable temperatures allow vitrification without composition corrections.14

Asbestos vitrification by means of plasma technology could be additionally promoted if the high processing costs could be at least partially compensated by the manufacturing of useful products from the obtained glasses, such as glass ceramics. Glass ceramics have been proposed as an application for waste derived glasses since the early 1960s.5,15,16 Moreover, asbestos and asbestos containing materials provide glasses of CaO–MgO–Al₂O₃–SiO₂ system,17 which is known to be the reference for many glass ceramics produced starting from waste.5,15,16

This paper reports the manufacturing of sintered glass ceramic articles from vitrified asbestos (VA) by not only exploiting the relatively high crystallisation tendency of the obtained waste glass, but also considering fast heating cycles applied in order to configure low cost treatments, comparable to those used for traditional ceramics. More specifically, asbestos derived glasses were applied to monolithic and cellular sintered (or ‘sinter crystallised’)18 glass ceramics as well as to new type of stoneware ceramics, presented in recent papers, based on the replacement of conventional feldspar fluxes with glass, undergoing crystallisation upon firing.19^–21 All the investigated glass ceramic articles, due to the achieved mechanical properties and microstructures, could find applications as construction materials.

Experimental

Starting vitrified waste

The investigated waste material consisted of a mass of VA containing materials, provided by Inertam,22 known with the trade name Cofalit. The composition is reported in Table 1.23 The material was partially crystalline since the vitrified waste had been originally subjected to uncontrolled and slow cooling (see later discussion).

OPEN IN VIEWER

Table 1

Chemical composition of starting materials/wt-%

Oxide	Vitrified asbestos from Inertam	Soda lime glass	Industrial clay
SiO2	35·7	71·1	64·8
Al2O3	7·4	2·2	21·9
TiO2	1·5		1·3
Fe2O3	7·0	0·8	0·9
CaO	35·1	9·9	0·5
MgO	12·6	2·2	0·5
K2O	0·1	0·9	1·9
Na2O	<0·05	12·9	0·5
Others	0·6 (LOI*)	5·3	6·6 (LOI*)

*LOI: loss on ignition.

The waste material was remelted in a platinum crucible at 1400°C for 1 h and quenched by pouring in cold water with the formation of a number of coarse fragments. The fragments were subsequently dry ball milled (repeated cycles of 30 min at 500 rev min⁻¹) and sieved to powders with a maximum dimension of 37 μm. A fraction with a maximum dimension of 75 μm was also considered. Dilatometric analysis (402 E; Netzsch Gerätebau GmbH, Selb, Germany) was performed on a glass fragment, while differential thermal analysis (DTA/TGA, STA 409; Netzsch Gerätebau GmbH, Selb, Germany), operated at 10°C min⁻¹ heating rate, was performed on both coarse and fine powders (with a maximum dimension of 75 or 37 μm). Dilatometric and DTA plots are both reported in Fig. 1.

OPEN IN VIEWER

Figure 1

Dilatometric and DTA plots for VA glass

Results and discussion

As shown in Fig. 1, the DTA plot for fine fowders (<37 μm) features an exothermic crystallisation peak T_C at 900°C, well below that for coarse powders (<75 μm), confirming the susceptibility of VA glass to surface crystallisation. From the dilatometric plot, reported in the same figure, it is evident that softening (properly ‘dilatometric softening’ T_D) occurs at a temperature of ∼100°C lower than T_C. Such gap was thought to be enough to determine dense ‘sinter crystallised’ glass ceramics since remarkable viscous flow sintering may be found from 100–150°C above T_D.20,25

Figure 2a, concerning the phase evolution of discs of pressed powders sintered at 900°C after heating at 10°C min⁻¹, confirms that ‘sinter crystallisation’ is a rapid method for glass ceramic production since the intensity of the diffraction peaks does not exhibit particularly significant changes with increasing holding time (above 30 min); the amorphous character of VA after remelting is confirmed for treatment at 900°C with a 0 min holding time.

OPEN IN VIEWER

Figure 2

Phase compositions and densities of sintered samples at different sintering conditions: a phase evolution at 900°C with different sintering times (heating rate of 10°C min⁻¹); b density evolution with sintering time; c comparison in XRD patterns of vitrified waste in as received state and samples sintered at 900°C for 30 min

The trend of apparent density of disc samples, reported in Fig. 2b, confirms the rapidity of sinter crystallisation since after 30 min there is no increase (the density is stationary at ∼2·85 g cm⁻³, at least within the experimental errors). The application of direct heating did not alter the apparent density significantly but changed the crystallisation sequence. From Fig. 2c, it is evident that the precipitation of crystal phases is much different in sintered glass ceramics compared to the VA containing waste in the as received condition. Since there was no sudden cooling after plasma vitrification, and the melt was subjected to a quite slow cooling in large steel containers, the starting material was actually crystalline. Such crystallisation is not surprising due to the relatively low content of silica compared to that of network formers in the glass formulation.15 The main phase formed in this condition corresponds to a melilite solid solution (most probably 2CaO.0·75MgO.0·25Al₂O₃.1·75SiO₂, PDF no. 79-2424, having an intermediate composition between akermanite and gehlenite end members),15 as recognised by the Match! program package used for semiautomatic identification. Remelting and sintering (with conventional heating of 10°C min⁻¹) gave rise to a secondary phase consisting of pyroxene with an intermediate composition among those of augite (0·82CaO.0·79MgO.0·27FeO.0·21Al₂O₃.1·751SiO₂, PDF no. 71-0721) and diopside (CaO.MgO.2SiO₂, PDF no. 83-1819), recognised as the most probable phases again by the Match! program package. Interestingly, pyroxene, featuring a remarkably lower CaO content than melilite solid solution, is practically the only crystal phase after sintering at 900°C with direct heating.

Figure 2c reports also the XRD pattern of a sample sintered (in the form of rectangular tiles, later cut into several bars) at 900°C, for 30 min, adopting a 40°C min⁻¹ heating rate. Such a heating rate was conceived as a solution to favour industrial production since the adopted rate is in the order of magnitude of that used in furnaces for the manufacturing of economic traditional ceramics, such as porcelain stoneware. It may be observed that the last sample maintains the features of samples from direct heating in terms of phase distribution and even density (the apparent density in this case was 2·92±0·02 g cm⁻³).

Figure 3 illustrates some microstructural details of glass ceramic samples subjected to different treatments. In Fig. 3a, the vitrified waste is shown in the as received state, i.e. after uncontrolled crystallisation: large pores, associated to the trapping of gasses during the melting stage or upon casting into steel containers, are visible. The crystallinity, revealed by the above discussed X-ray diffraction analysis, is confirmed in the high magnification detail of Fig. 3b. Apart from the macropores visible in Fig. 3a, the porosity is quite limited, and melilite solid solution is in the form of quite large, elongated crystals (light grey colour). The sintered samples, shown in Fig. 3c–f, are much different from the sample in the as received state. First, both samples from slow (10°C min⁻¹) and fast (40°C min⁻¹) heating feature many micropores; the amount of porosity (inferred from image analysis) is ∼5 vol.-%, similar to that of many valuable traditional ceramics.26 Second, both high magnification details (Fig. 3d and f) comprise a number of fine crystals, much smaller than those present in the vitrified waste in the as received state. Beyond these similarities, the different heating rates caused differences in the distribution of crystals: fast heating promoted the formation of many submicrometre crystals, very close to each other (see centre and right part in Fig. 3f), together with more isolated, bigger ones. To our opinion, the ‘crowded’ zones could be justified by intensive surface nucleation, originating from the rapid thermal treatment at 900°C, at which crystallisation is maximised.

OPEN IN VIEWER

Figure 3

a vitrified waste in as received state and b–f polished sections (SEM images) (b vitrified waste; c, d sample sintered at 900°C for 30 min at 10°C min⁻¹; e, f sample sintered at 900°C for 30 min at 40°C min⁻¹)

The microstructural features of the glass ceramic obtained by rapid heating are associated to very promising mechanical properties, as summarised in Table 2. Owing to the bending strength, >100 MPa, the obtained sintered glass ceramic may be compared to the strongest materials for tiles application. It could be observed that the bending bars were small, but it must be considered also that the bending tests were conducted in a four-point bending configuration, i.e. in a much more severe condition than three-point bending, employed for analogous determinations.11 We did not apply Weibull statistics due to the limited number of samples, but the Weibull modulus m can be roughly estimated from the standard deviation, from the literature.27 For the investigated sample, m is ∼10, so that in applying scaling equations28 the hypothetical three-point strength of samples of standardised geometry (bars with cross-section of 4×3 mm, tested on a span of 40 mm) should be very close to that determined in a four-point configuration with the actual samples. The Vickers’ microhardness (6·4±0·2 GPa), in addition, is well above the values associated to the best traditional ceramics;19 like the best traditional ceramics, the water absorption, as determined by applying the current norms,29 is negligible (<0·5%).

OPEN IN VIEWER

Table 2

Physical and mechanical properties of investigated glass ceramic articles*

Type of glass ceramic article	Sintering temperature/°C	Density†/g cm−3	WA/%	Young's modulus/GPa	Strength/MPa
Sintered glass ceramic (40°C min−1)	900	2·92±0·02	<0·5	92·5±9·5	107·5±11·3 [Bend]
Cellular glass ceramic	900	0·81±0·02			4·1±0·7 [Comp]
GCS (60%VA)	850	2·26±0·05		13·1±0·2	9·8±1·6 [Bend]
	900	2·32±0·05		25·6±1·1	22·8±3·5 [Bend]
	950	2·38±0·04		33·1±0·8	31·1±2·9 [Bend]
	1050	2·35±0·05		36·8±0·6	33·8±2·6 [Bend]
	1150	2·45±0·06	11·1	61·5±1·3	63·0±6·9 [Bend]
GCS (54%VA–6%SL)	1050	2·33±0·05	12·1	45·1±0·5	35·1±3·8 [Bend]
	1150	2·50±0·04	2·0	70·5±0·9	69·7±3·5 [Bend]

*[Bend]: data from four-point bending tests; [Comp]: data from compressive tests.

†Apparent density (except for cellular sample).

The microstructure of a cellular glass ceramic sample is reported in Fig. 4. It may be observed that the adopted processing, based on the usage of sacrificial PE sphere templates, led to a quite uniform cellular structure; the occurrence of sinter crystallisation limited any significant viscous collapse (Fig. 4a). The contact points between PE spheres led to the main openings; secondary, much smaller pores are visible at higher magnifications. The struts, in fact, are not dense but are determined by the joining of several granules (Fig. 4b and c). Such granules are, in turn, porous, and they may be originated by partial foaming of glass provided by gaseous decomposition products of the silicone resin used as binder. The resulting great roughness of cell walls is thought to be promising in view of applications in filters or catalyst supports. The compressive strength of the foam is substantial, if we think that the porosity is mainly open, and in the substantial amount of 74%, estimated from the comparison of bulk and true density. In fact, if we consider the well known Ashby's equation30 ruling the dependence of compressive strength of an open celled foam on the bending strength of the solid phase multiplied by an exponential scaling function, based on the relative density, σ_comp≈0·2σ_bend(ρ_rel)^1·5, where σ_comp is the compressive strength of foam, σ_bend is the bending strength of the solid material and ρ_rel is the relative density), the bending strength of the solid phase, to justify the observed compressive strength, should exceed 100 MPa, in good agreement with the bending tests on dense samples. Like in the case of dense samples, there was an intensive surface nucleation testified by the formation of a number of submicrometre crystals (Fig. 4c and d).

OPEN IN VIEWER

Figure 4

Microstructural details of cellular glass ceramic sample: a optical stereomicroscopy image; b–d SEM images

The analysis of the XRD patterns of the ‘glass ceramic stoneware’, reported in Fig. 5, underlines that the mixing of VA glass (60 wt-%) with industrial clay (40 wt-%) had a great impact on crystallisation in the presence of a rapid heating rate of 40°C min⁻¹. The temperature of 900°C, as shown in Fig. 5a, is confirmed to be a threshold for substantial crystallisation since in the sample sintered below (850°C) we can notice practically only quartz, PDF no. 83-0539, coming from industrial clay. Another phase, not previously found, consists of calcium feldspar (i.e. anorthite, CaAl₂Si₂O₈, PDF no. 85-1560), appearing as a consequence of the dissolution of clay minerals within the glass. In fact, like in previous works,19^–21 this phase could be seen as the product of the interaction between clay minerals and Ca²⁺ ions from glass (the direct reaction between metakaolinite, Al₂Si₂O₇, provided by the dehydration of kaolinite upon firing, and CaO, gives exactly anorthite). The most significant change, however, is that the main crystal phase from glass devitrification at 900°C was no longer pyroxene, like in the other cases of rapid heating, but melilite solid solution, which was previously promoted for slow thermal processes. Such condition is quite surprising since the clay/Ca²⁺ interaction should favour the phase with a lower CaO content, i.e. pyroxene, rather than melilite, as occurring from 1050°C. Additional investigations will probably be needed, but other phases (quartz, anorthite) likely had a specific nucleating action on melilite.

OPEN IN VIEWER

Figure 5

X-ray diffraction patterns of ‘glass ceramic stoneware’: a phase evolution with increasing sintering temperature (sample with 60 wt-%VA glass); b variations in balance among crystal phases associated with different formulations and thermal treatments

The different phase assemblages of ‘glass ceramic stoneware’ samples compared to sintered glass ceramics is coupled with poor densification. As inferable from the data reported in Table 2, the apparent density is much lower than that of sintered glass ceramic (in Fig. 6a and b, the sample sintered at 900°C can be seen as a not coherent mass), with consequently poor mechanical properties, except for the sample sintered at 1150°C (Fig. 6c and d). This sample, although possessing superior properties, still features very high water absorption values due to partially open porosity; from image analysis, the total porosity is ∼11%.

OPEN IN VIEWER

Figure 6

Microstructural details of selected ‘glass ceramic stoneware’ samples: a, b sintering at 900°C, 60%VA glass–40% clay; c, d sintering at 1150°C, 60%VA glass–40% clay; e, f sintering at 1150°C, 54%VA glass–6%SL glass–40% clay

The partial replacement of VA glass with glass (i.e. SL glass) not prone to crystallisation was chosen as a further development in order to improve the viscous flow. The sample resulting from a new formulation (Fig. 6e and f), with 10% of the VA glass fraction replaced by SL glass, is only slightly less porous, with the total porosity being ∼10%. However, the replacement led to a lower water absorption. The supposed enhanced viscous flow ability of a ‘not crystallising’ glass had probably a ‘sealing’ effect at the surface of the samples owing to the temperature gradients caused by the high heating rate applied. Both types of ‘glass ceramic stoneware’, without or with SL glass added, feature substantial crystallisation (Fig. 6d and f), even if the crystals, owing to the much higher sintering temperature than that adopted for pure VA glass, are quite coarser (as an example, Fig. 3c and e has the same magnification as Fig. 6d and f, but the crystals are hardly visible).

Replacing >10% of the VA glass fraction with SL glass would probably cause more significant variations (it can be seen from Fig. 5b that also the phase balance is almost unchanged), but it must be observed that the last glass ceramic stoneware sample possesses interesting properties, justified by the remarkable, although coarse, crystallisation (Fig. 6f). The density and bending strength values are in the order of those presented by ordinary stoneware tiles (the same observations about strength and dimension of samples, made above for sintered glass ceramics, may be considered in this case); the water absorption, in the order of 2%, for sintering at 1150°C may be accepted in applications wherein tiles are positioned vertically, e.g. in ventilated facades.31,32 As an alternative, it could be considered that the application of a glaze is advantageous also to modify the colour (green–brown for both ‘glass ceramic stoneware’ and for sintered glass ceramics).

In conclusion, ‘glass ceramic stoneware’, although generally possessing lower mechanical properties than sintered glass ceramics from VA glass, may be seen as a valid solution since its production conditions are closer to those employed for traditional ceramics, favouring the utilisation of pre-existing plants.

As a final remark, it is important to note that all the glass ceramic materials were developed from VA containing materials in an amorphous state. It is suggested to use the glass after drastic cooling of the melt to control the crystallisation during secondary treatment. This could be possible after a slight modification of the vitrification plant, i.e. by applying cold lamination of the melt, as in an analogous plant for the inertisation of municipal solid waste fly ash.33 The absence of any annealing of glass would undoubtedly favour the subsequent grinding.

Conclusion

The ceramic materials that could be obtained from VA glass, all with potential applications in the building industry, may be divided into two categories, with distinct features: