Introduction

The gradual depletion of essential minerals and metal primary deposits, combined with the possible interruption of critical metals supplies due to geopolitical situations, make it challenging for electronics industries to maintain or increase future supplies. For instance, according to the Royal Society of Chemistry, smartphones contain around 30 different elements and among these, six are currently in scarce supply and are estimated to exhaust their natural sources within the next 100 years based on the assumptions of current use rate and the available global reserves [1]. Considering these circumstances, increased recycling to recover the embedded high-value, scarce materials from end-of-life (EoL) electronics is an obvious way to ensure future availability. Approximately 53.6 million metric tonnes (Mt) of electronic waste (e-waste) from multiple waste streams was generated across the world in 2019, however, the collection and recycling of only 17% of the available e-waste was reported [2]. The amount of e-waste generation is projected to rise to 70.4 Mt by the year 2030 [2]. In many devices, the concentration of several precious metals (Ag, Au, Pt) and base metals (Cu, Sn, Ni) are significantly higher than in their primary mined ores that are currently processed to extract the metals [3, 4]. Recycling of these metals from secondary resources therefore has multiple benefits such as reducing the carbon emissions associated with new exploration and mining and with the subsequent processing steps which are typically highly energy intensive [5, 6]. Moreover, recycling protects the environment from the potentially harmful effects that result from dumping and landfilling of the rapidly growing e-waste [7].

The major embedded metal values in e-waste materials are present in complex printed circuit boards (PCBs) as plastic-ceramic–metal composites [8,9,10,11]. Several processing routes including physical separation, pyrometallurgical, hydrometallurgical, biohydrometallurgical and supercritical techniques, either individually or in combination, have been proposed and have been extensively investigated to extract value metals from PCBs [12, 13]. However, still there remains significant challenges to recover values at an industrial scale primarily because of the heterogeneity of the materials, the rapid changing of technology and the chemistry and structure of the materials. Pyrometallurgical processing is, therefore, becoming the preferred method for many commercial operations and is currently being adapted by several industries across the world including Umicore, Outotec, and Glencore [14]. In these smelters/refiners, e-waste scrap is co-fed with ores and other wastes, in particular those coming from Cu smelting industries. However, many countries do not have primary Cu smelters and in some large countries with smaller populations such as Australia that have smelters located far from the population centres, additional costs are incurred to collect and transport materials to the remote smelter [15]. In those circumstances, a relatively smaller scale smelting process would be a suitable alternative in the beneficiation of most metals within the alloy stream. The metal-poor slag generated during the smelting of PCBs contains mostly SiO2, Al2O3 and CaO with some other minor oxides [16,17,18]. These three major oxides make the slag system highly refractory requiring high operating temperatures to completely melt the PCBs and allow the separation of the metals from the slag. The high operating temperatures (typically > 1400 °C) may cause difficulties in keeping control of the process and the addition of a suitable flux could potentially reduce the liquidus temperature of the slags generated and make the process operationally viable at a significantly lower temperature. This in turn reduces the energy required to recover the metals from e-waste thereby lowering emissions.

Investigations into the phase equilibria of complex liquid oxide systems have been conducted for many years, particularly in relation to the formation of primary igneous rock types [19]. Similarly, research to investigate metallurgical operations, especially in slag-bearing systems relevant to the smelting of high-value metals such as copper and lead, have also been extensive [20,21,22]. This has led to an increased understanding of the partitioning of metals between solids and liquids in the relevant systems as well as improvements in the design of the types of refractory materials used in metal processing and production. In particular, several studies have focussed on improving the mould flux for continuous casting of steels from blast furnace slags in the CaO–Al2O3–SiO2–MgO system [23, 24]. Since commercial mould fluxes contain around 2–15% CaF2 the introduction of molten liquids releases fluorine which is detrimental to the equipment and creates environmental hazards. Based on these studies steps have been taken to replace CaF2 with other fluxes such as B2O3 [25, 26]. These types of studies have also been important in glass technology industries though they essentially only discuss the physical, mechanical, optical and some electrical properties of glasses rather than the phase equilibria of the systems [27, 28]. The last example involves research in the energy industry where many researchers studied the phase equilibria of aluminosilicate systems, specifically focusing on coal gasification, ash fusibility and viscosity of slags derived from coal ash [29,30,31]. While the study of phase equilibria in slag systems has previously been applied to several areas, a new emerging area is in the application of existing knowledge of phase equilibria data or the development of new slag systems which could potentially apply to the smelting of discarded PCBs at a moderately low temperature to assist in the separation of metal and slag phases and hence increase the recovery of valuable components from existing (at present) waste streams.

Since the CaO–Al2O3–SiO2 (CAS) slag system generated during the pyrometallurgical processing of e-waste PCBs needs elevated temperatures, the addition of a suitable metal oxide fluxing agent could be a way to reduce the liquidus temperature. For example, the effect of minor elements such as TiO2, MnO, Na2O, K2O, CaS, and B2O3 on the liquidus temperatures of blast furnace slags was previously investigated by Chen and co-workers [32]. In experiments within the melilite primary phase field, it was shown that the decrease in liquidus temperature, depended on the addition of particular oxides. The strongest effect on liquidus temperature was by B2O3 where the liquidus was reduced by 136 °C due to fluxing with 4.9 wt% B2O3. However, this study only reported data from a limited composition range (up to 4.9 wt% B2O3) and only within one master slag composition relevant to blast furnace ironmaking i.e., the SiO2–Al2O3–CaO–MgO system. It is not clear how the liquidus changes with further B2O3 addition.

The structure and viscosity of slags in the CaO–SiO2–Al2O3–B2O3 system was also investigated for applications in ironmaking (blast furnace slags) and as a mould flux for continuous casting of steel [25]. Similarly, the conductivity, heat transfer capability, fluidity, and crystallization behaviour have been studied thoroughly [33,34,35]. Further studies have focussed on the physical and mechanical properties of the glass (quenched slag) phase. B2O3 can act as an amphoteric oxide and also as an aluminosilicate network modifier by reducing the relative number of strong bonds in the glass and lowering the melting point and viscosity [36,37,38]. Wang et al. [39] investigated melting properties (i.e., softening, hemispherical and fluidity temperatures) in the CaO–SiO2–Al2O3–B2O3 system with 5–9 wt% B2O3 and a C/S ratio ranging from 0.83 to 1.5. While boron did not change the hemispherical temperature (temperature at which cylindrical flux sample loses its 50% height during hot stage microscopy technique), the viscosity decreased significantly with increasing B2O3. While it is possible to retain B2O3 in the slag originating from the boron-based flame retardants in PCBs, to the best of our knowledge the study of phase equilibria in CaO–Al2O3–SiO2–B2O3 slags relevant to e-waste smelting has not been reported so far.

Experimental

Materials

Three ternary master slags based on the CaO–Al2O3–SiO2 slag system were prepared by melting reagent grade CaCO3 (99.5 wt%, Sigma-Aldrich), Al2O3 (99.5 wt%, VWR Chemicals), and SiO2, (99.9 wt%, Sigma-Aldrich) at 1600 °C in air, at conditions well above the liquidus temperature to ensure complete melting and homogenization of all components [40]. Compositions with CaO/SiO2 (C/S) ratios of 0.3 (series S100), 0.6 (S200) and 1.0 (S300) were prepared, each with an Al2O3 content close to 20 wt%. A platinum crucible was used to hold the materials during melting inside a muffle furnace and for each master slag composition a transparent, glassy state of the quenched pre-melted slags was obtained. B2O3 (99.95wt%, VWR Chemicals) was then added, nominally at levels of 5, 10, 15 and 20 wt% B2O3 to each of the three master slags producing twelve B-doped sub-slag compositions in total. Note, that while the C/S ratio of the quaternary sub-slags were kept the same, the percentage of Al2O3 component decreased proportionally due to the addition of B2O3 (Table 1). The pre-cursor, B-doped sub-slag compositions were then re-melted at 1600 °C, quenched and pulverised to obtain powdered samples that were then screened to < 212 µm in particle size. Platinum foil used for containing the sub-slags in the drop quench experiments was supplied by Cookson Dental, UK.

Table 1 Compositions of the investigated master slags S100, S200 and S300 and the twelve B-doped sub-slags
Full size table

Composition of the Pre-melted Slags

X-ray Fluorescence Spectroscopy (XRF) and Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES) were used to analyse the compositions of the three master slags (S100, S200 and S300) and the twelve B2O3-doped sub-slags. For the XRF measurements, the slag samples were fused with lithium borate to produce a homogenous glass disc to analyse in the XRF instrument (Bruker 4 kW S8 Tiger WDXRF). The measurement accuracy of the XRF results was close to 0.5% relative, based on calibration datasets collected by CSIRO over several years. Since boron is a light element and is difficult to analyse with XRF, a Varian 730-ES ICP-OES was used. To analyse the samples using ICP-OES, the powdered slag samples were digested with HNO3 and HF and heated until the dissolution was complete. Before the analysis with ICP-OES the samples were diluted as required. Certified multi-element standards were used to check the accuracy of the calibrations and analyses. The compositions of the investigated master and sub-slags are shown in Table 1. The resulting compositions were close to the intended target compositions and showed negligible B loss had occurred during melting. Figure 1 shows the composition of the three master slags and their primary phase fields within the CaO–Al2O3–SiO2 ternary phase diagram.

Fig. 1
figure 1

The ternary CaO-Al2O3-SiO2 slag system showing the location of major phase boundaries and the three master slag compositions; redrawn using data from reference [41]

Full size image

Liquidus Determination Experiments

To study the phase equilibria, the prepared slag and B-doped sub-slag powders were equilibrated in air at high temperatures inside a vertical tube furnace. All samples were rapidly quenched in a water bath (room-temperature water) situated at the open bottom end of the furnace. In this way the high temperature structure and phase assemblage of the system was preserved through rapid solidification.

For the liquidus determination experiments, capsules made from thin platinum foil (0.025 mm thickness) were filled with 200–300 mg of the slag mixtures and suspended from the top of a vertical tube furnace by a platinum wire (0.5 mm diameter). The top end of the capsules remained open and exposed to the furnace (air) atmosphere. The capsules were then lowered into a 3–4 cm long hot zone at the centre of the furnace tube. A Type-B thermocouple inserted in an alumina sheath was placed close to the samples to continuously measure the temperature. The thermocouple was calibrated against the melting temperature of pure copper and shown to have an accuracy of ± 5 °C. Both the thermocouple and sample holder were inserted through the top endcap and sealed, while the bottom endcap was closed with a thin polypropylene sheet touching the water surface of the quenching bath. Figure 2a shows a schematic of the furnace arrangement for a drop quench test. A more detailed description of the experimental methods can be found in reference [42].

Fig. 2
figure 2

a Schematic diagram of the drop quench test furnace and experimental setup b thermal profile used for the equilibrium-quenching experiments (after Rait [43])

Full size image

Based on a procedure developed by Rait [43] the slags were heated to an initial temperature 100 °C above the planned experimental temperature, at a rate of 200 °C/h. The samples were held for 30 min at that temperature to completely melt and homogenize the molten slag. To introduce nucleation sites and thus assist crystallization at equilibrium, the temperature was then reduced to 200 °C below the experimental temperature and held for 30 min before being heated to the final experimental temperature. The samples were allowed to equilibrate for 4–24 h at the experimental temperature to complete the reaction. Once the equilibration time was reached, the samples were released from the top holder to drop directly into the quenching water bath to ensure rapid solidification. The temperature profile used for the equilibrium drop-quench experiments is shown as Fig. 2b.

The samples obtained from the drop quench tests were dried immediately to avoid any potential water absorption by the phases, separated from the platinum capsules, and then mounted in epoxy resin for analysis. After grinding and polishing of the resin blocks, the specimens were examined using SEM back-scattered electron (BSE) imaging to determine the phase assemblage and the compositions of the phases were qualitatively analysed using EDS and quantitatively using EPMA. Every new experiment was based on the phase assemblage from of the previous test depending on the equilibrium phases present, to determine if a higher or lower temperature was required. Repeat tests were done above the bracketed liquidus to examine if crystals present and below the temperature to check the absence of primary solid phase.

SEM–EDS Analysis

The mounted quenched slag samples were initially ground using SiC abrasive papers, finely polished with diamond suspensions of different particle sizes (6, 3 and 1 µm). Grinding and polishing of samples were performed in water-less medium to avoid dissolution of any phase and used oil/alcohol-based lubricants (i.e. DP-Lubricant Yellow from Struers). Before examination in the SEM the polished samples were carbon coated to avoid charge build up on the surface. An FEI Quanta 400 SEM equipped with a Bruker EDS system was used to investigate the microstructure and identify the phases present. The SEM analysis was performed using an accelerating potential of 15 kV with a working distance of 10.0 mm. EDS was used to measure the preliminary compositions of the solid and/or liquid phases present and hence identify the crystal phases present in the slag samples.

Electron Probe Microanalysis (EPMA)

The quantitative analysis of the different equilibrium phases was performed using a JEOL (Model JXA-8500F) electron microprobe. The analysis was performed with an accelerating voltage of 12 kV, a probe current of 10 nA and a beam defocus of 5 µm. The standards used for calibration were wollastonite (CaO.SiO2) for Ca and Si, danburite (CaB2Si2O8) for B and spinel (MgAl2O4) for Al. Caution was taken to minimize loss of boron by evaporation due to the excitement by the electron beam by defocussing the beam and employing short counting times (20 s. on the main B peak and 10 s. on the background). In addition, the low-beam current (10 nA) helped reduce potential volatilization. With this arrangement it was observed that there was negligible boron loss during analysis as confirmed by repeat analyses. Since boron is very sensitive and lighter it requires more rigorous precautions to reduce uncertainty in analysis. Before analysis, all samples were carbon coated together with the standards to ensure a similar coating thickness so that no additional variable can come from the differential coating thickness. For each phase at least 10–15 points were taken, and the average composition was reported.

Equilibrium Time

For most of the drop quench tests, 4 h was allowed for equilibrium to be reached between the liquid and primary crystal phases. This time allowance was confirmed by comparing the phase assemblages and their compositions for samples equilibrated at 4 and 24 h. After 24 h there was no significant difference in the phases present and their compositions from the samples equilibrated for 4 h. For example, the master slag S200 was equilibrated for 4 and 24 h at 1300 °C. Both equilibration times produced a slag system consisting of anorthite as the primary phase and liquid. The measured composition of the anorthite crystals and the liquid phase were the same in each case. In both cases the anorthite crystals and liquid phases proved to be homogeneous through checking compositions at many locations in the samples. In some samples close to eutectic compositions and at lower temperatures, a longer equilibration time of 8 h was used. For these samples the longer equilibration time was used to ensure the precipitation of a primary crystal phase.

Results and Discussion

Phase Relations at Equilibrium

Some representative microstructures of the equilibrium phases are shown in Fig. 3a–f. Solid phases such as anorthite, tridymite, pseudowollastonite and gehlenite can be seen in equilibrium with liquid. An example of complete liquid phase is shown in Fig. 3a indicating the typical microstructure of a slag equilibrated above the liquidus temperature for this slag composition. Figure 3b shows anorthite primary phase in equilibrium with liquid for the sample S205B equilibrated at 1230 °C. Here anorthite is the primary phase for this slag composition and the liquidus temperature is bracketed between 1230 and 1245 °C. The composition of the S200 master slag lies within the anorthite primary phase field and the liquidus is predicted to be 1340 °C based on the phase diagram of Rankin and Wright [44]. Upon addition of ~ 5% B2O3, the primary phase field does not change but the liquidus temperature was reduced by approximately 100–110 °C. For the same composition, at lower temperatures, two solid phases (anorthite and pseudowollastonite) were in equilibrium with liquid as shown in Fig. 3c. For some compositions, tridymite was also found in equilibrium with anorthite and liquid at a significantly lower temperature than the liquidus of the slag. Such an example can be represented by Fig. 3d for S110B slag quenched from 1050 °C. Primary gehlenite phase in equilibrium with liquid was observed in the boron-free master slag (S300) composition at 1310 °C is shown in Fig. 3e which agrees with the ternary CaO–Al2O3–SiO2 phase diagram [44]. Unlike anorthite in the S100 and S200 series slags, the addition of boron to the S300 master slag changes the primary phase from gehlenite to pseudowollastonite as shown in Fig. 3f, where the pseudowollastonite primary phase is now in equilibrium with the liquid. It is to be noted that the chemical composition of anorthite is close to the composition of the equilibrium liquid phase in the S100 series of slags, therefore both phases share a similar contrast in BSE imaging and require a careful adjustment in brightness and contrast to distinguish the anorthite primary phase and the liquid matrix. This can be observed in Fig. 3d. For the low-boron doped slags (S305B and S310B), pseudowollastonite crystals were in equilibrium with their liquid at close to their liquidus temperature. As the BSE images show, the brighter pseudowollastonite crystals form in different shapes and sizes in close association with the relatively darker contrast anorthite phase (Fig. 3d). Anorthite in the S200 and S300 series samples appears relatively darker in contrast as opposed to the S100 series. This is due to the comparatively higher CaO content in the liquid phase appearing brighter in the BSE imaging conditions of the SEM. On the other hand, pseudowollastonite, having higher Ca content, appears brighter in backscatter images since Ca is the heaviest element present.

Fig. 3
figure 3

Representative micrographs of S100 series of quenched slags a Liquid of S220B slag at 920 °C b Anorthite of S205B slag at 1230 °C c Anorthite + Pseudowollastonite + Liquid of S205B slag at 1180 °C d Anorthite + Tridymite + Liquid of S110B slag at 1050 °C e Gehlenite + Liquid of S300 at 1310 °C and f Pseudowollastonite + Liquid of S310B at 1195 °C

Full size image

Liquidus Temperature of the Slags

The liquidus temperatures of the 3 master slags and 12 sub-slags containing different boron contents are shown in Figs. 4, 5, 6. For the S100 (C/S = 0.3) slag series, the liquidus temperatures are plotted in Fig. 4. Results show that the liquidus temperature of 1320 ± 15 °C determined for the B-free master slag was in good agreement with the 1345 °C liquidus temperature reported by Rankin and Wright [44]. Upon progressive addition of B2O3 to the master slag, the liquidus temperature dropped significantly. With 5.2 wt% addition of B2O3 (S105B composition), at 1180 ± 20 °C anorthite was observed in equilibrium with liquid, while at 1200 °C the sample contained only liquid phase. This indicates that a temperature of 1200 °C is above the liquidus of the slag while 1180 °C is below the liquidus temperature. Therefore, the liquidus for the S105B slag was bracketed between 1180 and 1200 °C with an uncertainty of 20 °C. This represents a lowering of the liquidus temperature by around 140 °C compared to the B-free master slag composition. An increase in the boron content to 9.2 wt% (S110) reduced the liquidus temperature further to 1100 ± 20 °C. This trend continued with higher boron contents with compositions S115B and S120B having liquidus temperatures of 1040 ± 10 °C and 1035 ± 10 °C respectively. In all samples, the primary phase coexisting with a liquid below the liquidus temperature was anorthite. Although an increase in the amount of B2O3 consistently decreased the liquidus temperature of C/S = 0.3 master slag, the amount by which the temperature was lowered, decreased with increasing boron content. The liquidus lowering effect was greater for boron contents between 5 and 10 wt% while at the higher boron contents (i.e. from above 10–19 wt%), the extent of liquidus reduction is comparatively less steep. At 14 wt% B2O3 the liquidus was 1040 ± 10 °C and with further B2O3 addition up to 19 wt% the liquidus was reduced further by only 15–1025 ± 10 °C. Thus, the lowest liquidus for the S100 (C/S = 0.3) series slags was 1025 ± 10 °C which indicates a total lowering of liquidus by around 300 °C when fluxed with 19 wt% B2O3 doping.

Fig. 4
figure 4

Liquidus temperatures of the S100B series of slags doped with varying B2O3 contents; C/S = 0.3 and 15.6–18.2 wt% Al2O3. Error bar indicates the liquidus uncertainty range; L Liquid, A Anorthite, T Tridymite

Full size image
Fig. 5
figure 5

Liquidus temperatures of the S200B series of slags doped with varying B2O3 contents; C/S = 0.6 and 15.8–18.5 wt% Al2O3. Error bar indicates the liquidus uncertainty range; L Liquid, A Anorthite, W Pseudowollastonite

Full size image
Fig. 6
figure 6

Liquidus temperatures of the S300B series slags doped with varying B2O3 contents; C/S = 1.0 and 15.6–18.2 wt% Al2O3. Error bar indicates the liquidus uncertainty range; L Liquid, A Anorthite, W Pseudowollastonite, G Gehlenite

Full size image

For the S200 (C/S = 0.6) slag series, the liquidus temperatures are plotted in Fig. 5. The boron-free ternary CAS master slag for this composition has a liquidus temperature of 1335 °C. As for the C/S = 0.3 slag series, increasing the B2O3 content caused a significant reduction in the liquidus temperature. At the lowest boron content (S205B, 4.8 wt% B2O3), the slag was a completely liquid phase at 1245 °C while at 1230 °C the slag showed anorthite in equilibrium with liquid. Thus, the liquidus temperature for S205B slag was between 1230 and 1245 °C with 15 °C uncertainty. With a further increase in the amount of B2O3 added, the liquidus temperature dropped to 1080 °C with 9.7 wt% B2O3, 980 °C with 14.1 wt% B2O3 and 900 °C with 18.8 wt% B2O3. In each sample, at temperatures below the liquidus, anorthite was the only primary phase in equilibrium with the liquid. Unlike the S100 slag series however, a linear decline in the slag liquidus temperature was evident upon increasing B2O3 content with the lowest liquidus temperature determined for sample S220B as 900 ± 20 °C. Thus, the S200 series slags (C/S = 0.6) showed a decrease of the liquidus by more than 400 °C from that of the boron-free CAS master slag when doped by 18.8 wt% B2O3. Further experiments are required to determine if the trend continues at higher levels of B2O3 doping.

Figure 6 shows the experimentally obtained liquidus temperatures for the slag series S300 having a C/S ratio of 1.0 and with increasing B2O3 in the quaternary slags. According to the ternary CAS phase diagram the liquidus for the master slag S300 was estimated as 1310 °C and was located in the gehlenite primary phase field [44], although it should be noted that the composition is in close proximity to the gehlenite-anorthite-pseudowollastonite ternary eutectic. Overall, this slag series experienced significant lowering of the liquidus temperature with increasing B2O3 content in slags containing up to 14.9 wt% B2O3 after which the liquidus began to rise when the B2O3 content rose above this level. Addition of 5.2 wt% B2O3 (S305) caused a reduction in the liquidus temperature to 1220 °C, while additions of 9.7 wt% and 14.9 wt% saw the liquidus reduce even further to 1195 °C and 1140 °C, respectively. With the further addition of B2O3 to 18.8 wt% the liquidus increased to 1170 °C for slag S320B. For the S300 B-doped series therefore, the lowest liquidus temperature was 1140 °C with 14.9 wt% B2O3 representing an overall 170 °C decline in the slag liquidus due to the B-doping. In contrast to the S100 and S200 series, the addition of increased amounts of boron caused a shift in the primary phase assemblage. In the samples with boron levels of less than 10 wt% the primary crystalline phase switched from gehlenite (in the B-free system) to that of pseudowollastonite indicating that the ternary eutectic had shifted due to the presence of boron causing an expansion of the pseudowollastonite primary phase field. A further increase in the B content (i.e., 14.9 and 18.8 wt%) saw the appearance of anorthite and wollastonite coexisting with the liquid probably because of the expansion of pseudowollastonite and anorthite primary phase field in the quaternary slag system resulting from the doping with B2O3.In the ternary CAS system this equilibrium could be represented by the phase boundary between pseudowollastonite and anorthite primary phases. It is seen that liquidus temperature increases with increasing B2O3 above 14.9 wt% while there is no change in the phase assemblage (Fig. 6). This can happen when the equilibrium composition falls in the cotectic curve between pseudowollastonite and anorthite primary phase field along which these two solid phases are in equilibrium with the liquid in the pseudoternary CaO–Al2O3–SiO2 slag system doped with B2O3. When B2O3 content increases and proportionally the other components decrease the equilibrium liquid composition reaches the highest temperature point on the cotectic boundary which is indicated by the intersecting point on the cotectic curve by the anorthite–gehlenite–pseudowollastonite compatibility triangle.

The results from the three slag series examined have shown that doping with B2O3 reduces the slag liquidus significantly, but the reduction in temperature varies according to the composition of the initial master slag (C/S ratio) and the primary phase field that can be seen from the liquidus trend in Figs. 4, 5, 6. The liquidus reduction is linear when the primary phase was anorthite and did not change with progressive B-doping within the master slags (C/S = 0.3–0.6) although for C/S = 0.3 the trend reached a plateau beyond 14.1 wt% B2O3. However, for the master slag having a C/S = 1.0 upon B2O3 doping the primary phase field changes from gehlenite to pseudowollastonite and eventually falls on the boundary between pseudowollastonite and anorthite having these solid phases in equilibrium with the liquid. These phase transitions cause the non-linear changes in the liquidus temperatures of this series of slags. Among the three slags series, the S300 series showed the least effect of boron in lowering the liquidus temperature. In contrast, the highest reduction in liquidus temperature was in S200 series slags with an overall 435 °C reduction in the liquidus (composition S220B) while the S100 series experienced a maximum decline of 295 °C (composition S120B). Similar findings regarding the effect of boron on liquidus temperatures have been reported by other researchers where the effect of adding boron to a CAS-MgO slag was investigated using a similar technique of high temperature equilibration and drop-quench followed by EPMA [32]. They showed that the liquidus temperature dropped by 136 °C when doped with 4.9 wt% B2O3 in a slag with a C/S of ratio 1.1, and 8 wt% MgO and 16 wt% Al2O3.

Effect of C/S Ratio on the Liquidus Temperature of Slags

A comparison of the effect of C/S ratio on the liquidus temperature of B-doped slags shows that at lower B2O3 contents (5.0–9.5 wt%), for a given B2O3 content, the liquidus temperature increases slightly with increasing C/S ratio (Fig. 7). An overall increase of 40 °C was noted for the change of C/S ratio from 0.3 to 1.0 at 5 wt% B2O3 while at the increased B2O3 content (9.5 wt%) the liquidus temperature rose by 105 °C from 1080 °C for C/S ratio 0.3 to 1185 °C for C/S ratio 1.0. In contrast, at higher B2O3 contents (14 and 19 wt%), the liquidus initially began to reduce between C/S ratios 0.3 and 0.6 before sharply rising at C/S = 1.0 forming a V-shape pattern of the liquidus variation (Fig. 7). From the liquidus of samples with C/S ratio = 1.0 for 14.9 wt% and 18.8 wt% it is seen that increasing B2O3 content with this composition of slags increases the liquidus. A comparative analysis of the results showed that the best effect of reducing liquidus temperature with B2O3 doping was obtained in the S200 slag series with C/S ratio 0.6 while the S100 series (C/S = 0.3) also showed significant reduction of the slag liquidus. On the other hand, for the S300 slag series (C/S = 1.0), liquidus temperatures decreased at the lower boron contents but increased again beyond 14 wt% B2O3. Also, the degree of liquidus lowering in the C/S = 1.0 slags was smaller compared to the other two slag series. This suggests the optimum C/S ratio is 0.6. Similar results were found for CaO–Al2O3–SiO2–Na2O system where a sharp increase in the liquidus was observed with Na2O contents above 9 wt% in larnite primary phase field [42]. The results indicate, therefore, that within the investigated slag compositions the S200 slag series would be the best to flux with B2O3 to obtain relatively lower practicable liquidus temperatures. The optimum composition would be S220B with a C/S ratio of 0.6 and 18.3 wt% B2O3 which would result in a liquidus temperature of 920 ± 20 °C.

Fig. 7
figure 7

Variation of the liquidus temperature with C/S ratio for the different B2O3 contents (wt%) in the slags

Full size image

Comparison with CAS-Na2O Slags

In a recent study [42], the effect of Na2O on the liquidus temperature in CAS slags was studied. Overall, the addition of B2O3 was found to be more effective in lowering the liquidus temperature of CAS slags compared to the similar amount of Na2O as can be seen in Fig. 8. The largest decline in the liquidus temperature with boron doping was 435 °C for the S200 slag series (C/S = 0.6) while for the Na2O doped slags the maximum reduction of liquidus was reported as 225 °C within the same master slag. For the S100 (C/S = 0.3) composition, the maximum reduction in liquidus temperature caused by B2O3 doping was 285 °C while it was only 150 °C for Na2O doping. On the contrary, for the S300 series, the liquidus decreased by 170 °C to 1140 °C with 14.6 wt% B2O3 and increased to 1170 °C at 18.8 wt% B2O3. In general, adding Na2O in the CAS slag lowers the liquidus in the quaternary CAS–Na2O slags in a like manner to the CAS–B2O3 system. However, Na2O levels above 9 wt% increased the liquidus of S300 (C/S = 1) slags. A similar pattern was observed for the same series of slags above 14.6 wt% B2O3. A sharp fall in the liquidus temperature was observed in Na2O containing slag series due to the primary phase field changes, which was not seen in the CAS–B2O3 slags. The greater extent of liquidus reduction in blast furnace slags by B2O3 compared to Na2O was also reported by Chen et al. [32] which agrees with the findings of this current study.

Fig. 8
figure 8

Comparison of the effect of B2O3 with Na2O on the liquidus temperature for CAS slags. The Na and B-free slag liquidus temperatures were taken from Rankin and Wright [44]

Full size image

While in the S100 and S200 master slags the progressive doping by Na2O changed the primary phase field from anorthite (Na-free slag) to pseudowollastonite, in the B2O3 doped system adding a similar flux content did not change the phase assemblages from that of the boron-free master slags. On the other hand, Na2O doping of the S300 (C/S = 1.0) composition changed the primary phase from gehlenite to larnite, while boron doping changed the phase assemblage to anorthite and pseudowollastonite in equilibrium with liquid. Within a very narrow temperature bracket the two solid phases remain in equilibrium with liquid. Therefore, it was difficult to ascertain which one is the primary phase in this case. The equilibration experiments were repeated several times to ensure the reproducibility of the results. Every time the same phase assemblages were obtained.

Application of the Liquidus and Phase Equilibria Data

The experimentally determined liquidus temperature and phase equilibria data of slags in the CAS–B2O3 system can be used to design smelting operations to recover various metals from e-waste materials. This study has determined the liquidus temperature of 15 slags relevant to the composition of slag generated in electronic PCBs smelting. One of the objectives of boron doping in CAS slag was to lower the liquidus temperature of the highly refractory oxide CAS system. Overall, this objective was achieved as results indicated it was possible to lower the liquidus by more than 400 °C by B2O3 doping. However, the degree of reduction in the liquidus temperature was not the same for all the slags examined, being highest in slag S200 series with a C/S ratio of 0.6 and with doping by 18.8 wt% B2O3. In contrast, for some slags and at higher B2O3 on the slags the liquidus temperature increased with the increasing B2O3 flux content. Therefore, the precise liquidus data presented in this study will play important role to identify the best slag and flux composition based on the feed materials’ compositions. The experimental findings of the liquidus will guide the selection of the optimum slag compositions required to obtain the lowest practicable liquidus temperature. This will in turn reduce the energy required to recover valuable metals from e-wastes and bring operational flexibility. In addition, this information will be useful in modelling more complex slags using thermochemical calculation packages.

Conclusion

Phase equilibria of the quaternary system CaO–Al2O3–SiO2–B2O3 was studied to investigate the effect of B2O3 on the liquidus temperature in slag compositions with C/S ratios of 0.3–1.0 and 15.6–19.1 wt% Al2O3. The phase assemblages of the slags were identified at or close to their respective liquidus temperature, and the equilibrium phase composition was precisely determined with EPMA. Anorthite, pseudowollastonite, gehlenite and tridymite solid phases were found in equilibrium with liquid. An iterative approach was followed to estimate the liquidus temperature of slags within an error between 10 to 20 °C. Overall, doping with B2O3 decreased the liquidus temperature relative to the boron-free CaO–Al2O3–SiO2 based master slags but an increase in case of C/S ratio 1.0 at higher B2O3 doping. The highest degree of liquidus reduction was obtained in the slags with C/S ratio 0.6 and with increasing B2O3. A decline of liquidus by 435 °C was obtained with 18.8 wt% B2O3 doping in S220 slag. The lowest liquidus temperature within the investigated slags was 900 °C while the same undoped boron-free slag had the liquidus of 1335 °C. In contrast, the liquidus increased due to increasing B2O3 doping in S300 (C/S = 1.0) series slags at higher boron content. A comparison with similar Na2O doping in the CAS slag shows boron has a more pronounced effect in lowering the slag liquidus. Based on the liquidus temperature determined in this research CAS-B slags could be used in smelting waste PCBs at lower operating temperatures and B2O3 can be considered as flux to reduce the liquidus with C/S ratio 0.3 and 0.6. However, it might not be helpful to use B2O3 when the slag composition is close to C/S ratio 1.0.