Radiation shielding and mechanical properties of mullite-zirconia composites fabricated from investment-casting shell waste

Low-cost zirconia-toughened mullite composites were prepared from investment casting shell waste and alumina. The inﬂuence of sintering temperature on the composites ' properties, including crystalline phase composition, microstructure, degrees of densiﬁca-tion, mechanical properties, and radiation shielding characteristics, was investigated. The results show that the higher the sintering temperature, the higher the degree of densiﬁ-cation is, improving the mechanical and radiation shielding properties of prepared composites. The mullite-zirconia composite sintered at 1600 (cid:1) C presents a good mechanical strength, with ﬂexural and compressive strength values of 190 and 308 MPa, respectively. These values are comparable to or even better than mullite ceramics prepared from other waste materials. Furthermore, the composites ' gamma-ray and neutron attenuation characteristics suggest that they can be promising as radiation shields. ©


Introduction
The growing global economy and fierce industrial competition are leading to severe environmental deterioration and the unsustainable exploitation of natural resources.To achieve a greener and more sustainable economy, shifting toward sustainable production patterns with low ecological footprints is essential.Investment casting, a lost wax casting, is one of the most widely used manufacturing processes for fabricating complex-shaped metal components.In this process, molten metal is poured into a disposable ceramic mold and then solidified.Afterward, ceramic shell molds are broken off, and the casted metals are collected [1].Accordingly, the investment casting industry generates millions of tons of refractory investment casting shell waste annually [2].
Ceramic shell molds are typically made of valuable refractory materials, including silica, mullite, and zircon.Nevertheless, due to the stringent requirements of the casting process, waste shell molds cannot be recycled or reused.Therefore, waste ceramic shell molds are usually disposed to landfills.Hence, developing an economically feasible

Characterization of CW and experimental procedures
Large chunks of CW were obtained from a local plant in Turkey's Marmara province and broken in a jaw crusher to approximately 1 mm in size.Then, the crushed CW was ground via a mortar grinder (Pulverisette 2, Fritsch, Germany) for 5 min to obtain a particle size lesser than 63 mm.The images of the CW before and after grinding and particle size distributions of the ground powder are presented in Fig. 1a.As shown in Fig. 1c, ground CW powder has an angular morphology, and its grain size distribution is consistent with particle size measurement results.SEM/EDS elemental mapping images show that CW powder mainly consists of Si, Al, Zr, and O, but small amounts of Ti and Fe are also present.The chemical compositions of the CW as determined by XRF analysis (Rigaku RSX, Japan) are listed in Table 1.The results of XRD are in agreement with those of XRF.The main phases in CW include silica, mullite, and zircon (Fig. 1b).
Mullite-zirconia composites were synthesized as follows.According to the above chemical compositions (Table 1) and to obtain 3:2 mullite (3Al 2 O 3 $2SiO 2 ) in the final composite, CW powder and Al 2 O 3 (Merck, 45 mm mean particle size and !99.9% purity) were proportioned in mass ratios of 100:82.Thereby, the following reactions may occur during heat treatment: (1) Three other compositions were also formulated by adding 1 wt % MgO, TiO 2 , or V 2 O 5 sintering additives to the powder mixture.The powder mixtures prepared were mixed in distilled water via a planetary ball mill at a speed of 400 rpm for 2 h using a zirconia jar and high-purity alumina grinding media, then dried overnight at 80 C and crushed in an agate mortar with a pestle.The obtained powder mixtures were mixed with both 1 wt% polyvinyl alcohol and stearic acid, which serve as a binder and internal lubricant, respectively.First, PVA was dissolved in distilled water at 70 C and stirred for 20 min before being added to the powder mixture and continued the stirring process for 2 h, then dried overnight at 70 C and crushed in an agate mortar with pestle.The same procedure was applied for stearic acid, but acetone was used as a solvent.Rectangular (50 mm Â 20 mm x 8 mm), discshaped (12 mm Â 5 mm), and cubic (12 mm Â 12 mm x 12 mm) green bodies obtained by uni-axial pressing by applying a pressure of 200 Mpa, followed by debinding at 500 C for 2 h with the heating rate of 2 C min À1 in an electrical furnace.Then, the sintering process was conducted in air at 1450 C, 1500 C, 1550 C, and 1600 C. The heating and cooling rate was 5 C min À1 , and the dwelling period at the maximum temperature was 4 h.The sample codes, batch compositions, and their sintering temperatures are given in Table 2.
Phase compositions of samples analyzed by X-ray powder technique with a D8 Advance X-ray diffractometer (Bruker, Germany) using CuKa radiation (l ¼ 1.5406 A).The microstructure of samples was analyzed with scanning electron microscopy (Quanta Feg 250, FEI, USA).Linear shrinkage of samples was determined by measuring the diameter of green and sintered bodies.Density and open porosity were measured by the Archimedes method.The compression and three-point bending tests were performed in a RetroLine testing machine (Zwick/Roell, Ulm, Germany) using cube (10 mm Â 10 mm x 10 mm) and bar (20 mm Â 8 mm x 8 mm) shaped sintered bodies respectively, and a cross head speed of 0.5 mm/min applied.EpiXS and PhyX/PSD programs were used to theoretically explore the radiation shielding ability of manufactured ceramics.More information on the computational methods can be found in the literature [21e23].

Results and discussion
The XRD patterns of the samples heated at 1450 C are shown in Fig. 2. It can be seen that the CW1450 sample, sintered without adding any sintering additives, is composed of mullite and zircon (ZrSiO 4 ) crystalline phases, evidencing the mullite formation reactions among the silica and alumina in the starting materials have been completed after the heat treatment at 1450 C.
The dissociation of zircon into ZrO 2 and SiO 2 usually starts at 1673 ± 10 C [24].Nonetheless, impurities may promote zircon dissociation at lower temperatures [24e26].However, no zircon dissociation is observed neither in CW1450 sintered without additives nor in CW1450-T and CW1450eV samples containing TiO 2 and V 2 O 5 additives, respectively.Contrarily, the XRD patterns of the CW1450-M sample containing 1 wt% of MgO additive show characteristic XRD peaks belonging to mullite, zircon, tetragonal ZrO 2 , and monoclinic ZrO 2 phases.These results indicate that the MgO additive promotes zircon (ZrSiO 4 ) dissociation.
It's worth noting that zircon dissociates into silica and tetragonal zirconia at elevated temperatures.Indeed, the tetragonal phase is the stable phase of zirconia above 1170 C, and it might be transformed into the monoclinic phase during the cooling from 1170 C [27,28].Nevertheless, the presence of the tetragonal phase in CW1450-M might be attributed to the stabilization effect of the additive MgO [29].As the sintering temperature is increased from 1450 C to 1500 C, the zircon content significantly decreases (Fig. 3).By further increasing  the sintering temperature to 1550 C, the XRD peaks corresponding to the zircon phase completely disappeared.The zircon decomposition reaction is reversible, and zirconia and silica phases recombine to form zircon during cooling.However, no zircon peaks are detectable in the XRD patterns of the CW1550-M and CW1600-M samples.This observation can be attributed to the alumina additive, which presumably consumes the silica to yield mullite.It is also worth pointing out that CW1450-M contains monoclinic and tetragonal zirconia phases, whereas the samples sintered at higher temperatures contain no tetragonal zirconia.This is likely related to the growth of zirconia grains with increasing sintering temperature.Indeed, the earlier investigations on zirconia ceramics' tetragonal to monoclinic phase transformation have confirmed that increasing the grain size favors this transformation.In contrast, sub-micron grain size is often required to obtain tetragonal phase zirconia [28].
The linear shrinkage, bulk density, and apparent porosity of the sintered composites are shown in Fig. 4. A noticeable linear shrinkage of 15.2% is observed for the CW1500-M.The linear shrinkage increases with sintering temperature, reaching 19.4% for the sample sintered at 1600 C.This trend is consistent with bulk density and apparent porosity results.When the sintering temperature increases from 1500 to 1550 C, the apparent porosity decreases from ~5.4 to ~2.3%, while the bulk density adversely increases from 2.62 to 2.66 g cm À3 , as shown in Table 3. Increasing the sintering temperature to 1600 C results in an enhanced bulk density of 2.87 g cm À3 .
The SEM micrographs of the fractured and polished surfaces of samples are represented in Fig. 5.As shown in Fig. 5a, the CW1500-M sample sintered at 1500 C comprises consolidated angular grains.Increasing the sintering temperature from 1500 C to 1550 C leads to the formation of elongated mullite crystals with curved edges and corners (Fig. 5c).On the other hand, the CW1600-M composite shows similar microstructural development as CW1550-M, but the grain size of mullites is larger due to the Ostwald ripening (Fig. 5e).Notably, the formation of elongated mullite crystals is usually accompanied by the consumption of the surrounding matrix.As a result, mullite crystals were generally encapsulated by closed porosities (Fig. 5e and 5f-inset).
It is now well established that the elongated morphology is a characteristic of mullite crystals grown in a vitreous phase.Furthermore, the curved edges and corners of mullite grains indicate their formation occurred in the liquid phase.The formation of elongated mullite grains can be attributed to the presence of impurity TiO 2 and Fe 2 O 3 phases in the investment-casting shell waste or to the MgO additive, as they all reported to form transient liquid phases that promote the growth of mullite grains in elongated morphologies [20,30,31].The micrograph of the samples with polished surfaces reveals that they consist of irregular zirconia grains with bright contrast distributed in a mullite matrix exhibiting a grey contrast (Fig. 5b).The size of zirconia grains is in the submicron range, approximately up to 5 mm.
The compression and flexural strength values of the asprepared composites and the relevant research on the mullite ceramics reported in the literature are summarized in Table 4.As the compression and flexural strength were enhanced by increasing the sintering temperature, a correlation between mechanical properties and density can be inferred.Besides, such enhancement in the mechanical properties might also be attributed to the formation of elongated-shaped mullite grains with interlocking structures and the toughening effect of zirconia via crack deflection and microcracking mechanisms [32e35].The highest compression and flexural strengths are determined for the CW1600-M composite, with approximately 308 and 190 MPa, respectively.These values are comparable to or even better than most mullite ceramics and mullite-containing composites fabricated from recycled raw materials.Indeed, the flexural strength of developed ceramic composites delivered from investment casting shell waste is considerably higher than its counterparts.This is probably due to the formation of interlocking elongated mullite structures, which are more beneficial to flexural strength than compression strength.However, the mechanical properties are slightly lower than some mullite-zirconia ceramics prepared from fine and pure starting powders due to the closed porosities and impurities in the composites.It is worth stressing that green bodies here were prepared by uniaxial pressing.Besides, the raw materials were coarse; hence, zirconia in these composites was monoclinic.If isostatic or hot pressing is applied and zirconia can be stabilized in the tetragonal phase, then larger strength values could definitely be obtained.
The attenuation of gamma rays occurs through the interaction of the gamma radiation with matter's electrons via multiple absorption and scattering processes.The total crosssection is an important parameter of radiation shielding materials, and it is a measure of the probability of photons interacting with the shielding material.As seen in Fig. 6a, all samples exhibit nearly identical cross-section variations due to their similar chemical compositions.The total crosssection drastically decreases in the low-energy region as photon energy increases to approximately 80 keV.Then, the degree of decrease rate reduces in the photon energy range of 80e6000 keV region.Nevertheless, the cross-section values slowly rise as photon energy further increases.Such drastic changes in the total cross-section values originate from the processes involved in gamma ray-matter interactions in different photon energy regions.
Fig. 6bed illustrates the dominant photon-matter interaction mechanisms for varying energy regions.It can be seen that the photoelectric effect, incoherent scattering (Compton scattering), and pair production are the main processes involved in gamma-ray-matter interactions for all samples.Photoelectric absorption proceeds through the interaction of gamma-ray with an inner shell electron of the matter, which results in the total transfer of gamma-ray energy to the exciting electron.Therefore, elements with higher atomic numbers provide better gamma shielding through the photoelectric absorption mechanism.For the fabricated composites, the photoelectric absorption mechanism dominates the gamma attenuation process in the low-energy region up to 80 keV.As seen in Table 3, CW1600-M and CW1550-M contain slightly larger zirconium than CW1500-M.Zirconium   possesses a higher atomic number than aluminum and silicon.Therefore, its abundance in CW1600-M and CW1550-M provides better gamma attenuation relative to CW1500-M (inset of Fig. 6a).From 80 to 6000 keV, incoherent scattering becomes dominant.When a photon interacts with a charged particle, most frequently an electron, it scatters incoherently and loses energy.This phenomenon is also called the Compton effect and involves the partial transfer of the photons' energy to the recoiling electron [52].Fig. 6d indicates that the CW1600-M sample can attenuate gamma rays up to approximately 20,000 keV via the incoherent scattering mechanism.
When the photon's energy exceeds 20,000 keV, a high-energy pair production phenomenon becomes a dominant attenuation mechanism.In this phenomenon, high-energy photons interact with the nucleus, forming subatomic particle pairs such as an electron and a positron [52].Linear attenuation coefficient (m) is a constant that describes the fraction of attenuated gamma rays per unit thickness of matter.The linear attenuation coefficient can be expressed by Beer-Lambert law as fallow; m (cm À1 ) is the material's linear attenuation, x (cm) is the thickness of the material, I 0 is the incident beam intensity, and I is the intensity after attenuation [53,54].Linear attenuation coefficient (m) can be further converted into a material constant called the mass attenuation coefficient (m r ; cm 2 /g) by simply dividing it by the density of the shielding material.The linear and mass attenuation coefficients of the fabricated composites in the energy region between 1 keV and 1000 MeV are presented in Fig. 7.The results show that increasing sintering temperature led to an increase in the samples' linear attenuation coefficients.Indeed, the CW1600-M sample demonstrates a relatively higher linear attenuation coefficient than both CW1500-M and CW1550-M samples in the whole photon energy region.The probability of an interaction between the radiation and the absorber's atoms decreases with increasing porosity [55].Therefore, it is reasonable that a lowdensity absorber provides less attenuation than a highdensity absorber.Nevertheless, all samples exhibit similar   mass attenuation coefficient variations due to their similar chemical compositions, as seen in Fig. 7b.The mean free path (MFP), the average distance a photon can travel before interacting with the shielding material, can be determined from the linear attenuation coefficient using equation ( 4) [21].The MFP variation results of fabricated ceramics are presented in Fig. 8a.It has been found that MFP values for all ceramic samples remain almost zero, up to roughly 30 keV photon energy.This means that photons with low energy photons cannot be penetrated deeply into the fabricated ceramics.Although the MFP values steadily rise with increasing photon energy from 30 keV to 100 keV, they remain lower than 2 cm for all ceramic composites.Nonetheless, a sharper increase is observed beyond the photon energy of 100 keV.Meanwhile, the difference between the MFP values of the ceramic samples is enlarged.The maximum MFP values for all composites were noticed against photons with ~20 MeV energy.
The half-value layer (HVL) is another important parameter for radiation shielding materials, that is, the required material thickness to reduce the radiation intensity by 50%.HVL can be expressed using the following equation [21].
As shown in Fig. 8b, HVL values demonstrate a similar trend as the MFP values for all ceramic composites.The The effective atomic number (Z eff ) and electron density (N eff ) variations of fabricated ceramic composites are also investigated.The results indicate that Z eff and N eff reach their maximum values at around 23 keV photon energy due to a Kshell absorption edge (Fig. 8ced).Following the maxima, a gradual decrease in both Z eff and N eff is observed with increasing photon energy.Because CW1550-M and CW1600-M have the same chemical composition, their Z eff and N eff variations are expected to be similar.On the other hand, due to its lower zirconium content, the CW1500-M sample shows slightly lower Z eff and N eff values in the whole photon energy range.
The variations of the energy absorption buildup factor (EABF) and exposure buildup factor (EBF) at various penetration depths are depicted in Fig. 8e and f, respectively.These figures demonstrate that EABF and EBF values initially rise with increasing photon energy.They both reach their peak value at 300 keV and then decline.The observed trend is due to changing dominant photon-matter interaction mechanisms in different energy regions [56,57].Lower EABF and EBF values indicate a greater radiation shielding performance.Figs. 6 and  7 show that the photoelectric effect is dominating in the lowenergy region.Because the photoelectric effect is a highly efficient gamma attenuation mechanism, EABF and EBF values remain fairly low until the photon energy of approximately 100 keV for all ceramic composites.
Nonetheless, multiple scattering events occur at higher photon energies when Compton scattering dominates attenuation, where EABF and EBF values reach their peaks.With a further increase in photon energy, the pair production mechanism becomes dominant in the attenuation process, which results in a reducing trend in the EABF and EBF values [56,57].Moreover, there is a dramatic increase in the EABF and EBF values when the penetration depth increases from 1 MFP to 40 MFP.This can be ascribed to increased scattering events within the ceramics.It is also worth mentioning that there is a direct correlation between total cross-sections and the EABF and EBF values.
To better assess the radiation shielding performance of the fabricated ceramic composites, the mass attenuation coefficient and fast neutron removal cross-section results of the CW1600-M sample are compared with commonly used radiation shielding materials.As seen in Fig. 9a, the CW1600-M sample exhibits greater mass attenuation coefficient values than that of the Portland cement and water for the entire photon energy region.Meanwhile, it demonstrates comparable mass attenuation coefficient results to high alumina cement for the whole energy range.
Fig. 9b displays the fast neutron removal cross-sections of samples.It can be seen that the CW1600-M sample provides a relatively higher fast neutron removal cross-section (SR ¼ 0.093 cm À1 ) than other samples due to its higher density.Furthermore, the CW1600-M sample also outperforms two extensively used neutron shielding materials, graphite (SR ¼ 0.077 cm À1 ) and paraffin (SR ¼ 0.0773 cm À1 ), whereas its fast neutron removal cross-section is comparable to that of concrete (SR ¼ 0.094 cm À1 ) [58,59].

Conclusion
This investigation aimed to assess the feasibility of investment casting shell waste as a raw material in producing highperformance mullite-zirconia ceramics.The following conclusions can be drawn from the present study.
(1) XRD analysis revealed that investment casting shell waste consists of mullite, silica, and zircon, providing the required raw materials to produce mullite-zirconia ceramic composites.(2) The silica in the investment casting shell waste reacted entirely with alumina to yield mullite at sintering temperatures higher than 1450 C. (3) Since zirconia phases can be formed in the ceramic composites through the decomposition of the zircon phase in the investment casting shell waste, three sintering additives, including TiO 2 , V 2 O 5 , and MgO, were tested to promote the zircon dissociation.Only the MgO additive could effectively complete the zircon dissociation, but at a high sintering temperature of at least 1550 C when 1 wt% MgO was used.(4) The mechanical properties of the composites were enhanced with increasing sintering temperature from 1500 C to 1600 C. Indeed, the sample sintered at 1600 C for 4 h exhibited higher bulk density, flexural strength, and compressive strength than other samples.The bulk density, flexural strength, and compressive strength of CW-1600 M were 2.87 g cm À3 , 190, and 308 MPa, respectively.These values are relatively better than most mullite ceramics fabricated from other waste materials.The relatively good mechanical properties of the composites were attributed to the formation of elongated-shaped mullite grains with interlocking structures and the toughening effect of zirconia via the crack deflection and microcracking mechanisms.(5) Theoretical investigations demonstrated that density strongly influences the samples' gamma-and neutronshielding properties.Indeed, the sample sintered at 1600 C displayed greater shielding parameters than others due to its higher density.Its gamma-and neutron-shielding properties are comparable to commonly used radiation shields.
Accordingly, this study demonstrates a plausible way to utilize investment casting shell waste.

Fig. 1 e
Fig. 1 e (a) The particle size distributions of the ground CW powder.Inset: photographs of the CW before and after grinding and the D10, D50, and D90 values.(b) XRD pattern of CW powder (+ ¼ zircon, C ¼ mullite, -¼ cristobalite, ▢ ¼ quartz, and ⬠ ¼ TiO 2 ).(c) SEM and EDS mapping images of the ground CW powder.

j o u r
n a l o f m a t e r i a l s r e s e a r c h a n d t e c h n o l o g y 2 0 2 3 ; 2 4 : 5 8 8 3 e5 8 9 5
j o u r n a l o f m a t e r i a l s r e s e a r c h a n d t e c h n o l o g y 2 0 2 3 ; 2 4 : 5 8 8 3 e5 8 9 5

Fig. 6 eFig. 7 e
Fig. 6 e (a) The relative variations of the total cross section for the CW1500-M, CW1550-M, and CW1600-M samples vs. photon energy.The relationship between the photon energy and cross-section of (b) CW1500-M, (c) CW1550-M, and (d) CW1600-M in the energy range from 1 to 10 6 keV.

j o u r
n a l o f m a t e r i a l s r e s e a r c h a n d t e c h n o l o g y 2 0 2 3 ; 2 4 : 5 8 8 3 e5 8 9 5

Fig. 9 e
Fig. 9 e (a) Comparison of the mass attenuation coefficient variations of commonly used radiation shielding materials with CW1600-M (b) The fast neutron removal cross-sections of CW1500-M, CW1550-M, and CW1600-M samples computed using the PhysX software.

Table 1 e
Elemental composition (wt-%) of CW powder determined from XRF analysis.

Table 2 e
Sample codes, batch compositions, and sintering temperatures of the ceramic bodies.

Table 3 e
Measured density and the weight fraction of constituent compounds of the CW1500-M, CW1550-M, and CW1600-M samples.

Table 4 e
The mechanical properties of mullite/zirconia composites prepared in this work compared to different mullites reported in the literature.