Recovery of indium and yttrium from Flat Panel Display waste using solvent extraction

Separation and Purification Technology 166 (2016) 117–124

Contents lists available at ScienceDirect

Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur

Review

Recovery of indium and yttrium from Flat Panel Display waste using solvent extraction Jiaxu Yang ⇑, Teodora Retegan, Britt-Marie Steenari, Christian Ekberg Department of Industrial Materials Recycling, Institute of Chemical Engineering, Chalmers University of Technology, Kemigarden 4, SE-412 96 Gothenburg, Sweden

a r t i c l e

i n f o

Article history: Received 26 October 2015 Received in revised form 12 April 2016 Accepted 14 April 2016 Available online 20 April 2016 Keywords: Indium Rare earth elements FPD Solvent extraction Recycling DEHPA Cyanex 923

a b s t r a c t Oxides of indium and yttrium are two of the key components in Flat Panel Displays (FPDs). In recent years the need to recycle these metal oxides from waste FPDs has been growing. In this work a process to recycle indium and yttrium based on acid leaching and solvent extraction was proposed and studied. Solid waste was leached by acid at S/L ratio = 0.1 g/ml, HCl was found to be more effective than HNO3, possibly due to the formation of soluble metal chloride complexes. The extraction of indium using Cyanex 923 and yttrium using DEHPA from chloride media studied using lab-scale mixer-settlers at a flow rate of 3 ml/ min. Leachate of real FPD waste was used as aqueous feed. Good separation between indium, yttrium and other impurities such as iron, copper and aluminum could be achieved by extraction from 1 M HCl with 0.25 M Cyanex 923 diluted in kerosene, followed by stripping with 1 M HNO3 and further purification with 0.2 M DEHPA diluted in kerosene. Ó 2016 Elsevier B.V. All rights reserved.

Contents 1. 2.

3.

4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Acid leaching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Batch solvent extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Mixer-settler tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Sample analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Leaching experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1. Composition of soluble metals in the FPD waste . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2. Acid leaching behavior of HCl and HNO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.3. Optimal leaching acid and conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Solvent extraction process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Batch extraction experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. Mixer-settler tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Third phase and crud formation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Solvent extraction process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

⇑ Corresponding author. E-mail address: [email protected] (J. Yang). http://dx.doi.org/10.1016/j.seppur.2016.04.021 1383-5866/Ó 2016 Elsevier B.V. All rights reserved.

118 118 118 118 118 118 119 119 119 119 120 120 120 122 123 124 124 124 124

118

J. Yang et al. / Separation and Purification Technology 166 (2016) 117–124

1. Introduction During the last decade, the global shipment of Flat Panel Displays (FPDs), especially Liquid Crystal Displays (LCDs) has surpassed Cathode Ray Tube (CRT) displays and has become the most popular type of displays on the market [1]. A typical LCD contains approximately 0.7 g indium per m2 in the form of Indium-Tin Oxide (ITO) [2]. In addition, small amounts of Rare Earth Elements (REEs) can be found in the backlight units and loudspeakers. These metals are considered to be critical raw materials by the European Commission due to their economic importance, availability and supply risk [3]. Therefore in recent years the recovery of indium and REEs from secondary sources such as production waste and End-of-Life (EoL) products has been gaining interest. Previous studies have shown that many different methods can be used to recover indium from waste LCDs, such as chloride vaporization [4], supercritical CO2 extraction [5] and solvent extraction. In addition, ionic liquids have also been tested for separation of REEs [6]. In this study solvent extraction was chosen due to its simplicity in terms of material, equipment and process conditions required, and its efficiency in metal separation. Numerous previous studies have shown that indium can be recovered from aqueous solutions using methods such as solvent extraction, and extensive studies on organic molecules suitable for indium extraction have been carried out [7]. Many different extractants, especially organophosphate compounds such as Tributyl Phosphate (TBP) [8], (Cyanex 272) [9], bis-2,2-ethylhexyl phosphate (DEHPA) [10,11], 2-ethylhexyl phosphonic mono-2-ethylhexyl ester [12] and Cyanex 923 (a mixture of four trialkyl phosphates) [13] have been tested. Results of these earlier studies showed that apart from Cyanex 272, other extractants were all able to selectively extract indium from many other elements such as transition metals from either acidic chloride or sulfate media. With respect to the extraction of REEs from aqueous solutions, earlier studies have shown that Cyanex 923 can be an effective extractant for the extraction of REEs from e.g. phosphor powders of EoL fluorescent lamps and electrode materials of Nickel Metal Hydride (NiMH) batteries [14,15]. In addition, there have been a number of previous studies on the recovery of indium from leachate of LCD waste [12,2,16]. However, since the FPD waste material studied in the present work contain both indium and yttrium (>10 ppm), modification of the previously designed method for indium recycling [2] was necessary. In this work, a recycling method based on acid leaching followed by solvent extraction was explored and presented. In the acid leaching studies, composition of soluble metals in the waste and optimal leaching conditions were determined. Based on a previous study [17], Cyanex 923 and DEHPA were used as the extractants in the design of a laboratory-scale solvent extraction process for the separation and recovery of indium and yttrium from the acidic leachate containing different metal ions. Furthermore, a problem encountered in the solvent extraction process regarding crud formation was discussed and a few possible solutions were proposed.

liquid ratio was 0.1 g/ml. A simple test on the effect of increased temperature on leaching efficiency was performed by leaching the samples at 22 and 80 °C. The acid solutions used for leaching were diluted from concentrated HNO3 (65%, Sigma Aldrich) or HCl (>37%, Sigma Aldrich) with de-ionized water. It was calculated from the difference in pH of acid before and after leaching that acidity after leaching was reduced by approximately 15% from 1 M. Due to this reduction in acidity had little effect on metal extraction, the leachate was used as aqueous feed in subsequent solvent extraction processes.

2.2. Batch solvent extraction Batch extractions were performed using the following procedure: 1.5 ml of each organic and aqueous phase was added to a 3.5 ml glass vial (O/A = 1). The extractions were performed at room temperature (21 ± 1 °C). For batch tests, an extraction kinetics study was performed and it was found that metal extraction using Cyanex 923 reached equilibrium after approximately 5 min of mechanical shaking (IKA VIBRAX VXR basic) at 1500 rpm. A contact time of 5 min was therefore used. The organic diluent studied was kerosene (Solvent 70, Statoil), and the extractants were Cyanex 923 (93%, Cytec) and DEHPA (97%, Sigma Aldrich). For the construction of a McCabe-Thiele diagram (calculation of number of ideal stages), the volume ratio of the organic and aqueous phases (O/A) varied from 1:5 to 5:1.

2.3. Mixer-settler tests Mixer-settler experiments were performed using laboratory scale mixer-settlers in counter-current arrangement as described in an earlier work [18]. The mixer-settlers were made of PVDF, for both indium and yttrium extraction experiments, the volumes of the mixing and settling chambers were 100 ml and 500 ml, respectively. The pump rate was set to approximately 2 ml/min for both organic and aqueous flows. The rate of stirring was set to approximately 1000 rpm, and samples of both phases were taken after 600 ml (2 stages) or 900 ml (3 stages) of solution had been pumped through the system, in order to ensure that equilibrium was reached. Leachate of the solid waste was used as aqueous feed (HCl with an initial concentration of 1 M). The organic feed was 0.25 M Cyanex 923 diluted in kerosene, with 1% toluene. The separation and recovery process of indium and yttrium from acidic leachate was divided into four major steps: (1) separation of indium and yttrium with Cyanex 923. (2) back-extraction of indium from loaded organic phase to another aqueous phase. (3) Extraction of yttrium from the aqueous phase in step (1) with DEHPA. (4) back-extraction of yttrium from the loaded organic phase from step 3.

2. Experimental

2.4. Sample analysis

2.1. Acid leaching

Metal concentrations in the acidic leachate and the aqueous phase of solvent extraction experiments were determined using Inductively Coupled Plasma with Optical Emission Spectrometer (Thermo iCAP-6000). Aqueous concentrations of metals were quantified by calibration with 1000 ppm external standards (Sigma-Aldrich). The detection limit of the instrument for the metals of interest is in the range of parts per billion (<0.1 lM), which is low enough for the quantification to be reliable. More precise values of distribution ratios below 0.01 were not presented due to measured values being near the detection limit of the instrument.

The solid waste sample was in the form of powders (<1 mm) and was one of the fractions obtained from a FPD waste shredding line. All leaching experiments were performed in 120 ml plastic beakers. The rate of stirring was approximately 700 rpm with 6
20 mm stirring magnets. Aqua regia was used to determine the total metal composition in the solid waste. For other studies the initial concentration of HCl or HNO3 was 1 M, samples were taken between 1 h and 4 days after leaching started, the solid to

119

J. Yang et al. / Separation and Purification Technology 166 (2016) 117–124 Table 1 Metal content of the solid waste samples used in the experiments. Concentration (mg/L)

Al Co Cu Eu Fe In Mo Nd Ni Sn Y Zn

1030 ± 50 2 ± 0.4 370 ± 22 31 ± 20 2200 ± 800 30 ± 5 11 ± 1 19 ± 4 41 ± 9 72 ± 4 319 ± 35 370 ± 85

3.1. Leaching experiments 3.1.1. Composition of soluble metals in the FPD waste Metal composition of the solid waste is shown in Table 1. The content of yttrium is the highest among the three valuable metals (In, Y and Nd). Moreover, other metals such as Al, Fe, Ni, Cu, Sn, and Zn were detected in much larger concentrations than those of In, Y and Nd. The presence of these metals in such quantities is expected to be the main impurities in the subsequent separation process. 3.1.2. Acid leaching behavior of HCl and HNO3 The data presented in Figs. 1 and 2 shows the amount of Y and In leached between 1 h and 80 h. The leaching kinetics for indium in 1 M HCl are much faster than in 1 M HNO3. In 1 M HCl indium

35

45 40 35 30 25 20 15 10 5 0

Y

0

20

40

60

Leaching Efficiency (%)

Leaching Efficiency (%)

Metal

3. Results and discussion

30 25 20 15 5 0

80

In

10

0

20

Time (h)

40

60

80

Time (h)

100 90 80 70 60 50 40 30 20 10 0

Y

0

20

40

60

Leaching Efficiency (%)

Leaching Efficiency (%)

Fig. 1. The effect of contact time on the leaching behavior of Y and In when using 1 M nitric acid at 20 ± 2 °C.

80

100 90 80 70 60 50 40 30 20 10 0

In

0

20

Time (h)

40

60

80

Time (h)

100 90 80 70 60 50 40 30 20 10 0

100

Al Ni Fe Zn

Leaching Efficiency (%)

Leaching Efficiency (%)

Fig. 2. The effect of contact time on the leaching behavior of Y and In when using 1 M hydrochloric acid at 20 ± 2 °C.

80 60

Al

40

Fe

0 0

20

40 Time (h)

60

80

Ni

20

Zn 0

20

40 Time (h)

60

80

Fig. 3. The effect of contact time on the leaching behavior of Al, Fe, Ni, Zn in (left) 1 M nitric acid (right) 1 M hydrochloric acid at 20 ± 2 °C and S/L = 0.1 g/ml.

-5

100 90 80 70 60 50 40 30 20 10 0

20 C 80 C

5

15

25

Leaching Efficiency (%)

J. Yang et al. / Separation and Purification Technology 166 (2016) 117–124

Leaching Efficiency (%)

120

-5

100 90 80 70 60 50 40 30 20 10 0

20 C 80 C

5

Time (h)

15

25

Time (h)

Fig. 4. Leaching efficiency of In and Y at 20 ± 2 °C ( ) and 80 ± 2 °C ( ) in 1 M HCl. S/L = 0.1 g/ml.

120

100 80 60

80 C

40

20 C

20 0

0

10

Leaching Efficiency (%)

Leaching Efficiency (%)

120

100 80 60

Time (h)

20 C

20 0

20

80 C

40

0

10

20

Time (h)

Fig. 5. Leaching efficiency of In and Y at 20 ± 2 °C ( ) and 80 ± 2 °C ( ) in 1 M HNO3. S/L = 0.1 g/ml.

concentration reached equilibrium after 24 h of leaching, while in 1 M HNO3 indium concentration did not reach equilibrium after three days. Compared to the results of a previous study on leaching of LCD glass [2], the leaching efficiency of metals in this case was observed to be much slower. This could be the result of much higher content of metals and metal oxides in the FPD waste investigated in the present study compared to earlier studies. The effect of HNO3 concentration on leaching efficiency of Al, Fe, Ni and Zn is shown in Fig. 3. It can be observed that the leaching efficiency of Al, Fe, Ni and Zn are faster than for In, Y and Eu, therefore it could be possible to selectively leach Al, Fe, Ni and Zn first using HNO3 at concentrations between 1 M and 0.1 M. Figs. 4 and 5 show the leaching efficiency in 1 M HCl and 1 M HNO3 at 20 ± 2 °C and 80 ± 2 °C plotted for the metals of interest. The leaching of Y in 1 M HCl reached equilibrium after 48 h at 20 ± 2 °C, while at 80 °C equilibrium was reached in 10 h. In addition, leaching of In required 4 days to reach equilibrium in 1 M HNO3 at 20 ± 2 °C, while approximately 8 h were sufficient at 80 ± 2 °C. This could be explained by the difference between the standard enthalpy of formation for In2O3 (925 kJ/mol) [19], Y2O3 (1905 kJ/mol) [19] and H2O (241 kJ/mol) [19], this indicates that the leaching of In2O3 and Y2O3 by acid, forming aqueous metal ions and water is an endothermic process. 3.1.3. Optimal leaching acid and conditions By comparing between 1 M HCl and 1 M HNO3, it was concluded that HCl was more efficient and therefore more suitable acid for leaching. This difference could be related to the fact that the stability constants of indium chloride complexes being greater than those of indium nitrate complexes. The main disadvantage with leaching at higher temperature, other than the extra energy requirement, was the increased solubility of other materials such as silica in the waste material by

Table 2 Metal extraction in the indium extraction step. The aqueous phase was 0.85 M HCl; kerosene was used as the diluent.

Cu Eu Fe In Mo Nd Sn Y Zn

1 M Cyanex 923 (%)

0.25 M Cyanex 923 (%)

96 ± 1 6±3 47 ± 2 97 ± 1 98 ± 1 6±2 >99 6±2 99 ± 1

4±2 5±3 30 ± 3 91 ± 1 96 ± 1 14 ± 4 >99 3±2 87 ± 1

reaction (1) [20], which can give rise to the formation of crud in subsequent solvent extraction processes [21].

SiO2 ðsÞ þ 2H2 OðlÞ ! H4 SiO4

ð1Þ

3.2. Solvent extraction process Metal composition in the HCl leach liquor used as aqueous feed for solvent extraction experiments is similar to the values in Table 1. The initial acidity of the aqueous feed is approximately 0.85 M. 3.2.1. Batch extraction experiments Before mixer-settler experiments was performed, a series of batch extraction experiments were done to determine suitable starting conditions for mixer-settler experiments, such as extractant concentration, appropriate stripping agents and their concentrations. Table 2 shows the distribution ratio of In, Y, Eu and several

121

J. Yang et al. / Separation and Purification Technology 166 (2016) 117–124

600

Table 3 Distribution ratio of metals in the stripping step using different aqueous phase. The organic phase was 0.25 M Cyanex 923 diluted in kerosene. 1 M HNO3

1 M H2SO4

77 ± 13% >99% >99% 13 ± 1% 4 ± 2% <1% >99%

30 ± 10% >99% 78 ± 2% 16 ± 1% 5 ± 2% <1% >99%

58 ± 22% >99% 87 ± 1% <1% 11 ± 2% <1% 67 ± 1%

500 Org conc (mg/L)

Cu Fe In Mo Nd Sn Zn

6 M HNO3

400 300 200 100

Table 4 Metal concentration of the HCl aqueous raffinate. The uncertainties were calculated as the standard deviation of triplicate samples. Metal

Metal content (lg/ml)

Cu Eu Fe Nd Y Zn

350 ± 21 29 ± 19 1510 ± 550 16 ± 3 300 ± 33 48 ± 11

Cu (%)

Eu (%)

Fe (%)

Nd (%)

Y (%)

Zn (%)

2±1 <1 4±2

0.77 ± 0.08 2±1 4±1

12 ± 2% 23 ± 3% 28 ± 3%

3±2 3±2 4±3

>99 94 ± 2 46 ± 1

52 ± 1 <1 2±1

Table 6 Metal distribution ratio in the yttrium stripping step, the organic phase is 0.2 M DEHPA diluted in kerosene after extraction from 1 M HCl leachate.

2 M HCl 3 M HCl

Cu (%)

Eu (%)

Fe (%)

Nd (%)

Y (%)

Zn (%)

91 ± 7 92 ± 6

94 ± 2 92 ± 2

18 ± 3 50 ± 5

2±1 2±1

>99 98 ± 1

98 ± 1 >99

other metals between Cyanex 923 diluted in kerosene and the aqueous feed. With 1 M Cyanex 923, over 98% of indium was extracted to the organic phase in one stage. However, in this case the co-extraction of other metals such as Zn and Cu is significant

100 90

200

300

400

500

(>95% extracted), and co-extraction of Eu and Y is also slightly higher with 1 M Cyanex 923 compared to 0.25 M Cyanex 923. Another drawback of a higher extractant concentration is the need for more concentrated acid when metals are to be back-extracted or stripped into another aqueous phase. Due to the aforementioned reasons, 0.25 M Cyanex 923 diluted in kerosene was chosen as the organic phase to separate indium and REEs. Stripping of indium from the organic phase was tested with HNO3 and H2SO4 at different concentrations, as shown in Table 3. Less than 15% of indium extracted into the organic phase was stripped into 1 M H2SO4, whereas 1 M HNO3 and 6 M HNO3 was able to strip 70% and 99% of extracted indium in one stage, respectively. This difference is  likely due to [SO2 4 ] (Ka2 = 0.01) being much lower than [NO3 ] (Ka  20) at 1 M acid concentration, leading to the reduced formation of soluble indium complexes. Despite the fact that 6 M HNO3 was more efficient for stripping indium than 1 M HNO3, the separation of indium from co-stripped iron and copper in the subsequent extraction step would become more difficult with e.g. DEHPA. Therefore 1 M HNO3 was considered to be the more suitable aqueous phase for stripping. The results for yttrium extraction and stripping can be seen in Tables 5 and 6, respectively. Concentrations of metals in the HCl raffinate after extraction are shown in Table 4. Data for In, Mo and Sn are not shown in Table 6 because they were quantitatively extracted in the previous step with Cyanex 923. With 1 M DEHPA diluted in kerosene the extraction of Y was quantitative over one extraction stage. However, at such high extractant concentrations

100

70

Cu

60

80 Extracon (%)

Org Conc (mg/L)

100

Aq conc (mg/L)

80

50 40 30 20

Eu

60

Fe

40

In Mo

20

10 0

0

Fig. 7. McCabe-Thiele diagram of yttrium. Aqueous phase is 1 M HCl, organic phase is 0.2 M DEHPA diluted in kerosene, the diagonal line is the operating line with h = 1.

Table 5 Extraction efficiencies in the yttrium extraction step, after extraction of 0.85 M HCl leachate with 0.25 M Cyanex 923 diluted in kerosene.

1 M DEHPA 0.2 M DEHPA 0.1 M DEHPA

0

Nd

0 0

20

40

60

80

100

Aq Conc (mg/L) Fig. 6. McCabe-Thiele diagram of indium. Aqueous phase is 1 M HCl, organic phase is 0.25 M Cyanex 923 diluted in kerosene, the diagonal line is the operating line with h = 1.

0

1

2

stage number

3

Sn Y

Fig. 8. Extraction efficiencies of mixer-settler stages in the indium extraction step. Aqueous phase is 1 M HCl leachate on solid FPD waste, organic phase is 0.25 M Cyanex 923 diluted in kerosene, stage 0 is the aqueous feed.

122

J. Yang et al. / Separation and Purification Technology 166 (2016) 117–124

100

120 Cu

80

Eu

80

Fe

60

In

40

Strip (%)

Strip (%)

100

Mo

20 0

1

2

3

Eu

40

Fe Sn

20

Nd

0

60

0

Sn

Y 0

Y

stage number

1

2 3 stage number

4

Fig. 9. Strip efficiencies of mixer-settler stages in the indium stripping step. Aqueous phase is 1 M HNO3, organic phase is 0.25 M Cyanex 923 diluted in kerosene.

Fig. 11. Strip efficiencies of mixer-settler stages in the yttrium stripping step. Aqueous phase is 2 M HCl, organic phase is 0.2 M DEHPA diluted in kerosene.

the co-extraction of Zn and Eu became significant. Therefore an extractant concentration of 0.2 M was chosen, since at this concentration yttrium extraction was still high, while very little Eu and Zn were co-extracted. Distribution ratios of the stripping stage after extraction with 0.2 M DEHPA in kerosene is shown in Table 6. It can be seen that both 2 M and 3 M HCl could quantitatively strip Y from the organic phase. However, due to the stripping of Fe being less in 2 M HCl than 3 M HCl, the former was considered to be the better stripping agent. 1 M H2SO4 and 1 M HNO3 were also tested, but their results were not shown because the percentage stripping of yttrium was low (<50%).

Table 7 Extraction and strip efficiencies of different options to further purify the aqueous raffinate solution containing yttrium. Option 1: after yttrium stripping step, extract iron from the 2 M HCl raffinate with 0.25 M Cyanex 923 in kerosene. Option 2: use 3 M HNO3 instead of 2 M HCl for yttrium stripping.

3.2.2. Mixer-settler tests 3.2.2.1. Ideal stages calculation. McCabe-Thiele diagrams for indium and yttrium extraction were constructed in order to determine the number of minimum counter-current mixer-settler stages required for each step of the process. The results are shown in Fig. 6 for indium and in Fig. 7 for yttrium. It can be seen in Fig. 6 that when indium concentrations in the aqueous feed are between 20 and 40 lg/L, one ideal stage was nearly sufficient to obtain quantitative extraction of indium. However two stages would ensure the quantitative extraction of indium. In Fig. 7 the results show that if yttrium concentration in the feed solution is less than 300 lg/L, quantitative extraction of yttrium can be obtained with 2 ideal stages. Based on these results, two counter-current stages were used for the extraction and stripping of indium, and three stages were used for the extraction and stripping of yttrium. 3.2.2.2. Extraction and stripping of indium. The results presented in Fig. 8 show that due to rapid extraction kinetics, >90% of In and

Extraction (%)

100

Option 1 Option 2

Eu (%)

Fe (%)

Nd (%)

Sn (%)

Y (%)

2 >99

>99 5

2 65

75 7

4 2

>99% of Sn in the feed solution was extracted to the organic phase after the first stage. Changes in concentrations of REEs were small over the mixer-settler stages. However, it was observed that by adding a second extraction stage, larger amounts of Fe, Zn and Cu were extracted into the organic phase, e.g. the concentration of Fe decreased from 2100 lg/L in the feed to 1660 lg/L after the first stage and 1260 lg/L after the second stage. This was expected, as according to the results shown in Table 3, the distribution ratio of Fe in this extraction system is approximately 0.4, which corresponds to 30% extraction in one stage. The organic phase after extraction was stripped using 1 M HNO3. It can be seen that >90% of extracted indium was recovered into the 1 M HNO3. Data in Fig. 9 also shows that small amount of REEs were present in the strip acid. Other quantitatively stripped metals were Fe and Zn, and their concentrations in the raffinate of the strip stages are more than one order of magnitude higher than indium. Further separation of indium from the strip raffinate can be achieved by e.g. extraction with 0.2 M DEHPA diluted in kerosene, followed by stripping with 1 M HCl. Comparisons between results of an earlier work [2] and the results in Figs. 6 and 8 showed that the theoretical prediction based on the McCabe-Thiele diagrams agreed well with the experimental results; indium was quantitatively extracted in 2 mixer-settler stages, and most of the indium extraction occurred in the first stage.

80 60

Eu

40

Fe Sn

20 0

Y 0

1

2 3 stage number

4

Fig. 10. Extraction efficiencies of mixer-settler stages in the yttrium extraction step. Aqueous phase is 1 M HCl, organic phase is 0.2 M DEHPA in kerosene, and stage 0 is the aqueous feed, which is the aqueous HCl raffinate after extraction with 0.25 M Cyanex 923 in kerosene.

3.2.2.3. Extraction and stripping of yttrium. Mixer-settler results of the extraction and stripping of yttrium are shown in Figs. 10 and 11. Data for some of the metals were omitted in these two figures as their concentrations in the aqueous phase was nearly constant. It can be seen from the figures that although most of the yttrium in the aqueous feed can be extracted and stripped in three stages, this was more than was predicted from Fig. 7. It can also be seen from Fig. 11 that the concentrations of several metals such as Eu and Nd increased more than Y in the third strip stage. Therefore, with respect to yttrium purity, it could be more beneficial to reduce the number of strip stages from three to two. Furthermore, since iron is the most significant impurity in the strip raffinate, several options were explored to reduce the concentration of Fe in the raffinate solution containing yttrium. One of the

J. Yang et al. / Separation and Purification Technology 166 (2016) 117–124

123

3.3. Third phase and crud formation

Fig. 12. SEM image of the crud material collected.

Table 8 Material composition of the crud material determined by EDX. Positions of the scanned locations labelled as ‘‘Spectrum 1–5” are shown in Fig. 12. Position scanned

O (%)

Al (%)

Si (%)

Ca (%)

Fe (%)

Y (%)

Sn (%)

Spectrum Spectrum Spectrum Spectrum Spectrum

59 52 56 59 58

5 1 1 1 1

30 32 31 26 31

2 1 1 2 1

0 1 1 1 1

0 1 1 1 0

1 1 1 0 1

1 2 3 4 5

options was to use Cyanex 923 diluted in kerosene to extract Fe from the 2 M HCl raffinate containing iron and yttrium. The other option was to use 3 M HNO3 instead of 2 M HCl in the yttrium stripping step of the process. The results are shown in Table 7. Both options effectively increased the purity of yttrium with respect to iron. Due to the much higher separation factor between iron and yttrium, the first option was concluded to be the better choice.

It was noticed during mixer-settler operations that third phase and crud could form and accumulate at the phase interface. This is detrimental to the solvent extraction process as it can result in the loss of solvent, and can make phase separation more difficult [22]. Several previous works [22,23] have studied the formation of third phase in extraction systems using organophosphate compounds similar to Cyanex 923 and concluded that the formation of third phase was proportional to the concentration of inorganic  ligands (e.g. NO 3 , Cl ) and extractable metal ions. In a comprehensive review on cruds in solvent extraction processes [21], the formation cruds were attributed to several factors, such as the presence of suspended solids, organic materials, silica, Al and Mo etc. From batch extraction experiments it was seen that the formation of third phase could be prevented by the addition of 5 vol% of TBP, cyclohexanone, and 1-decanol. However, after the addition of these phase modifiers a small amount of cruds (10 mg per 1 ml aqueous phase) remained at the aqueous/organic interface. From Table 1 it can be seen that the content of several metals such as Fe and Al in the waste are quite high, and since the waste material studied was FPD waste in the form of powders, substances such as suspended solid, organic material and silica could exist in the acidic leachate and contribute to the formation of cruds, even after filtration with a filter pore size of 0.45 lm. Samples of cruds formed during extraction were collected and characterized using Energy Dispersive X-ray (EDX). The results are shown in Fig. 12 and in Table 8. It was seen that the crud material mainly consisted of O and Si in an approximate ratio of 2:1, indicating that the crud is mostly silica, with minor amounts of other elements such as Ca and Fe. Washing of the FPD waste with 1 M NaOH, ethanol or acetone before leaching with 1 M HCl could reduce the amount of crud formed to some extent. Filtration of the leachate with filter pore size <45 lm can prevent the formation of cruds, indicating that the presence of fine suspended solids could be a major cause of crud in this case, as was suggested by Ritcey [21]. However, it could be costly to perform such filtration at industrial scale.

Fig. 13. Flowchart for the indium and yttrium recovery process using Cyanex 923 and DEHPA as extractants.

124

J. Yang et al. / Separation and Purification Technology 166 (2016) 117–124

Table 9 Recovery of metals in the aqueous raffinate at process steps outlined in Fig. 13, converted as % of feed concentration. Data was obtained from laboratory scale mixersettler tests.

Y In Zn Fe

Indium extraction (%)

Indium stripping (%)

Yttrium extraction (%)

Yttrium stripping (%)

Yttrium purification (%)

94.5 5.0 0.7 59.5

0.3 54.4 10.9 24.6

3.7 5.6 0.4 38.4

91.3 5.6 0.4 4.3

87.8 5.6 0.4 0.1

3.4. Solvent extraction process A preliminary solvent extraction process for the recovery of yttrium is shown in Fig. 13. In the step labeled ‘‘indium extraction”, indium, iron, copper, tin and molybdenum were extracted into an organic phase consisting of 0.25 M Cyanex 923 in kerosene. REEs such as yttrium, neodymium, and europium remained in the aqueous phase. In the step ‘‘indium stripping”, indium, copper, iron and zinc are stripped with 1 M HNO3. Yttrium could be extracted by 0.2 M DEHPA diluted in kerosene followed by stripping with 2 M HCl. After indium was stripped from 0.25 M Cyanex 923 in kerosene with 1 M HNO3, it was possible to use this organic phase to separate iron and yttrium, as shown in the ‘‘yttrium purification” step. Concentrations of indium, yttrium and major impurity species such as iron and zinc at aqueous feed and each extraction step outlined in Fig. 13 are shown in Table 9. It can be observed that the proposed process still has large room for improvement with respect to the amount and purity of yttrium and indium recovered. For instance, indium in the raffinate from the ‘‘indium stripping” step as shown in Fig. 13 can be further separated in the process presented in [2], where >95% indium purity can be achieved. Furthermore, more extraction stages can be added to the ‘‘indium stripping” and ‘‘yttrium purification” part of the process as shown in Fig. 13 in order to further reduce the amount of iron and indium impurities in the yttrium stream. 4. Conclusion Compared to previous studies [10,2], the FPD waste analyzed in this study was more complex with respect to the metal composition in the waste. Several modifications to the previously proposed process were therefore made. Leaching kinetics of metals in 1 M HNO3 and 1 M HCl were studied at both 20 ± 2 °C and 80 ± 2 °C. Results showed that the latter could be more time-efficient for leaching. In addition, improving mechanical separation and pretreatment processes could be extremely beneficial for the subsequent solvent extraction process by separating the solid waste into different fractions containing less mixed materials. Studies using laboratory scale mixer-settler methods showed that a combination of DEHPA in kerosene and Cyanex 923 in kerosene was able to separate indium into 1 M HNO3 and yttrium could be recovered in 2 M HCl, as shown in Fig. 12. The indium stream can be further purified using the same method proposed in previous works [10,2], i.e. by extraction with 0.2 M DEHPA in kerosene followed by back extraction with 1 M HCl. Contacting the yttrium stream with 0.25 M Cyanex 923 in kerosene can selectively extract the majority of the Fe impurities into the organic phase. The results of the experiments have shown through solvent extraction, yttrium and indium can be separated from other impurities to a purity of 95%. Higher purity could be achieved by optimization of

the solvent extraction, or by additional processes such as electrolysis. In addition, it was observed that cruds could form during the extraction process. Several methods, such as washing the solid waste with 1 M NaOH before leaching with 1 M HCl, can drastically reduce the formation of cruds. Further study could be performed in this area, however, to gain a better understanding of both the formation and prevention of crud. Acknowledgements The authors would like to thank Recyval-Nano, a project in the EU Seventh Framework Program (No. 310312) for financial support. References [1] DisplaySearch, Quarterly Global TV Shipment and Forecast Report, Q2 2012, 2012 [Online], Available: http://www.displaysearch.com/cps/rde/xchg/ displaysearch/ hs.xsl/quarterly_global_tv_shipment_and_forecast_report.asp (Accessed 16 7 2012). [2] J. Yang, T. Retegan, C. Ekberg, Indium recovery from discarded LCD panel glass by solvent extraction, Hydrometallurgy 137 (2013) 68–77. [3] European Commission, Critical Raw Materials for the EU, European Commission, 2010. [4] T. Kunihiko, S. Atsushi, G. Dodbiba, J. Sadaki, N. Sato, T. Fujita, Recovering indium from the liquid crystal display of discarded cellular phones by means of chloride-induced vaporization at relatively low temperature, Metall. Mater. Trans. A 40 (4) (2009) 891–900. [5] H.-M. Liu, C.-C. Wu, Y.-H. Lin, C.-K. Chiang, Recovery of indium from etching wastewater using supercritical carbon dioxide extraction, J. Hazard. Mater. 172 (2009) 744–748. [6] K. Larsson, K. Binnemans, Separation of rare earths by split-anion extraction, Hydrometallurgy 156 (2015) 206–214. [7] A. Pavia, Recovery of indium from aqueous solutions by solvent extraction, Sep. Sci. Technol. 36 (7) (2001) 1395–1419. [8] K. Anil, K. Asit, Solvent extraction and separation of gallium(III), indium(III), and thallium(III) with tributylphosphate, Talanta 14 (6) (1967) 629–635. [9] B. Gupta, N. Mudhar, I. Singh, Separations and recovery of indium and gallium using bis(2,4,4-trimethylpentyl)phosphinic acid (Cyanex 272), Sep. Purif. Technol. 57 (2007) 294–303. [10] S. Virolainen, D. Ibana, E. Paatero, Recovery of indium from indium-tin-oxide by solvent extraction, Hydrometallurgy 107 (1–2) (2011) 56–61. [11] T. Nakamura, A. Sakai, S. Nishihama, K. Yoshizuka, Solvent extraction of indium gallium and zinc ions with acidic organophosphates having bulky alkyl groups, Solvent Extr. Ion Exch. 27 (2009) 501–512. [12] H.N. Kang, J. Lee, J. Kim, Recovery of indium from etching waste by solvent extraction and electrolytic refining, Hydrometallurgy 110 (2011) 120–127. [13] B. Gupta, A. Deep, P. Malik, Liquid–liquid extraction and recovery of Indium using Cyanex 923, Anal. Chim. Acta 513 (2) (2004) 463–471. [14] K. Larsson, C. Ekberg, A. Ødegaard-Jensen, Using Cyanex 923 for selective extraction in a high concentration chloride medium on nickel metal hydride battery waste, Hydrometallurgy 129–130 (2012) 35–42. [15] K. Larsson, C. Ekberg, A. Ødegaard-Jensen, Using Cyanex 923 for selective extraction in a high concentration chloride medium on nickel metal hydride battery waste: Part II: Mixer-settler experiments, Hydrometallurgy 133 (2013) 168–175. [16] S. Virolainen, D. Ibana, E. Paatero, Recovery of Indium from LCD screens, in: ISEC 2011, Santiago, Chile, 2011. [17] J. Yang, C. Ekberg, T. Retegan, Recovery of In, Y, Nd and Eu from WEEE by solvent extraction, in: International Solvent Extraction Conference (ISEC), Würzberg, Germany, 2014. [18] J. Lilejin, G. Persson, I. Svantesson, S. Wingefors, Experience from Cold Tests of the CTH Actinide Separation Process, in: ACS Symposium Series 161 – Transplutonium Elements – Production and Recovery, 1981, pp. 203–221. [19] D. Wagman, W. Evans, V. Parker, R. Schumm, I. Halow, S. Bailey, K. Churney, R. Nuttall, The NBS tables of chemical of chemical thermodynamic properties, J. Phys. Chem. Ref. Data 11 (2) (1982). [20] J. Rimstidt, H. Barnes, The kinetics of silica-water reactions, Geochim. Cosmochim. Acta 44 (1980) 1683–1699. [21] G.M. Rticey, Crud in solvent extraction processing – a review of cause and treatment, Hydrometallurgy 5 (1980) 97–107. [22] T.S.A. Suresh, P.V. Rao, Parameters influencing third phase formation in the extraction of Th(NO3)4 by some trialkyl phosphates, Solvent Extr. Ion Exch. 27 (2) (2009) 132–158. [23] R. Jha, K. Gupta, P. Kulkarni, P. Gurba, P. Janardan, R. Changarani, P. Dey, P. Pathuk, V. Manchanda, Third phase formation studies in the extraction of Th (IV) and U(VI) by N, N-dialkyl aliphatic amides, Desalination 232 (2008) 225– 233.