LICENTIATE T H E S I S
Luleå University of Technology
Department of Chemical Engineering and Geosciences, Division of Mineral Processing
:
Mechanical Recycling of Consumer Electronic Scrap
Jirang Cui
Mechanical Recycling of Consumer Electronic Scrap
Jirang Cui
Division of Mineral Processing
Department of Chemical Engineering and Geosciences Luleå University of Technology,
SE-971 87, Luleå, Sweden
May 2005
ABSTRACT
Consumer electronic equipment (brown goods), such as television sets, radio sets, and video recorders, are most common. However, recycling of consumer electronic scrap is only beginning.
Characterization of TV scrap was carried out by using a variety of methods, such as chemical analysis, particle size and shape analysis, liberation degree analysis, thermogravimetric analysis, sink-float test, and IR spectrometer. A comparison of TV scrap, personal computer scrap, and printed circuit boards scrap shows that the content of non-ferrous metals and precious metals in TV scrap is much lower than in personal computer scrap or printed circuit boards scrap. It is expected that recycling of TV scrap will not be cost-effective by utilizing conventional manual disassembly.
The result of particle shape analysis indicates that the non-ferrous metals particles in TV scrap formed as a variety of shapes, it is much more heterogeneous than for plastics and printed circuit boards. The results of sink-float tests demonstrate that a high recovery of copper could be produced by an effective gravity separation process.
Identification of plastics shows that the major plastic in TV scrap is high impact polystyrene. Gravity separation of plastics may encounter some challenges in separation of plastics from TV scrap because of specific density variations.
Furthermore, Mechanical recycling of TV scrap oriented to recovery of non-ferrous metals is highlighted by using several techniques, such as air table, eddy current separation, and optical sorting. The separation results reveal that air table separation is an effective technology to recover metals from consumer electronic scraps. By using a DGS table, approximately 90% of non-ferrous metals were recovered in the heavy product with a purity of 40%. Printed circuit boards and cables in TV scrap cause metals loss due to the fact that metals in printed circuit boards and cables are not liberated from plastics and ceramic materials. The study shows that eddy current separation and optical (metal) sorting process provide alternatives to recover metals from TV scraps.
At last, new developments of eddy current separation, such as wet eddy current separation and Magnus separation are discussed in the thesis. A comparison of eddy current separation and Magnus separation on aluminum recovery shows that wet eddy current separation is more effective for recovery of fine non-ferrous particles.
Keywords : WEEE; Consumer electronic scrap; Recycling; Characterization;
Mechanical separation; Materials recovery; Eddy current separation; Air table
separation; Optical sorting; Magnus separation
ACKNOWLEDGEMENTS
I would like to express my sincere gratitude to my supervisor, Professor Eric Forssberg for his guidance, encouragement and invaluable discussions. I am very grateful to Dr. Peter Rem at Faculty of Civil Engineering and Geosciences, Delft University of Technology (TU Delft), the Netherlands, for his creative instructions on eddy current separation. I also thank Professor M.A. Reuter at Faculty of Civil Engineering and Geosciences, TU Delft for providing an opportunity of doing three- month research work at TU Delft.
I would very much like to express my deep gratitude to Professor Shouci Lu at University of Science and Technology Beijing for his encouragement, help and discussions concerning work and life in the past ten years.
I am also indebted to Professor Bo Bjökman, Director of the Minerals and Metals Recycling Research Centre (MiMeR) for providing me an opportunity to conduct this subject. Thanks also go to staff and colleagues at the Department of Chemical Engineering and Geosciences, Lulea University of Technology (LTU), for a friendly and cooperative academic environment, particularly Dr. Yanmin Wang for his suggestions on my experimental work, Dr. Bertil Pålsson for his prompt assistance to solve computer problems and an introduction of the image process system, Dr.
Nourreddine Menad for his conducting TGA analysis and invaluable comments on my papers, Dr. Hamid-Reza Manouchehri for his conducting optical sorting experiments, Ms. Siv T. Berhan and Lic. Eng. Mia Tossavainen for their care and help.
I would like to appreciate invaluable contributions from Ms. Lenka Muchova and Mr.
Bo Zhou at Faculty of Civil Engineering and Geosciences, TU Delft for their support and help during my stay in TU Delft.
I am very grateful to Mr. Sverker Sjölin at Stena Technoworld AB for providing the TV scrap sample and the identification of plastics, Mr. Johan Petersson at Draka Kabel Sverige AB for supplying the cable samples, Mr. Istvan Lukacs, OVAKO Steel AB for the chemical analysis of samples. Financial support from MiMeR. LTU, Sweden is gratefully acknowledged.
Thanks go to all Chinese friends for their help, particularly Dr. Qixing Yang for the translation of Swedish mails, Lic. Eng. Mingzhao He for his suggestions of my work, and Dr. Hongyuan Liu for his help.
Last but not least, I thank my Rong’er, parents, grandmother, brother and sister for
there love, continuous encouragement, patience and support.
LIST OF PAPERS
This thesis is based on five papers referred to in the text by roman numbers:
I. Mechanical recycling of waste electric and electronic equipment: a review Jirang Cui, Eric Forssberg
Journal of Hazardous Materials, B99 (2003) 243-263
II. Characterization of consumer electronic scrap oriented to materials recovery
Jirang Cui, Eric Forssberg submitted to Waste Management
III. Mechanical separation of consumer electronic scrap Jirang Cui, Eric Forssberg, Hamid-Reza Manouchehri to be submitted to Waste Management
IV. Eddy current separation for fine particles Jirang Cui, Eric Forssberg
to be submitted to Journal of Hazardous Materials
V. A comparison of Magnus separation and wet eddy current separation Jirang Cui, Lenka Muchova, Peter Rem, Eric Forssberg
to be submitted to Resources Conservation and Recycling
Paper related to, but not included in the thesis:
Recycling of consumer electronic scrap Jirang Cui, Eric Forssberg
Accepted by the 4
thColloquium of SORTING: Innovations and Applications,
Berlin, Germany, October 2005
CONTENTS
1. Introduction...1
1.1. Management of waste electric and electronic equipment ... 1
1.2. Mechanical recycling processes ... 2
1.3. Objectives of the present work... 4
2. Materials and methods ...5
2.1. Materials... 5
2.2. Methods... 6
3. Characterization of consumer electronic scrap ...11
3.1. Chemical analysis ... 11
3.2. Size and metal distribution of TV scrap... 11
3.3. Particle shapes of materials in TV scrap ... 12
3.4. Liberation degree of copper ... 13
3.5. Sink-float test ... 13
3.6. Quantification of plastics by thermogravimetric analysis... 16
3.7. Identification of plastics by FT-IR spectrometer ... 17
4. Mechanical separation of consumer electronic scrap...21
4.1. Ferromagnetics recovery... 21
4.2. DGS Table separation ... 21
4.3. Eddy current separation ... 22
4.4. Optical sorting... 22
5. New developments of eddy current separation for fine particles ...25
5.1. Theory ... 25
5.2. Traditional eddy current separation for fine particles ... 28
5.3. Preliminary study of Magnus separation and wet eddy current separation... 29
6. Conclusions ...35
References:...37
Paper I- V
1. Introduction
1.1. Management of waste electric and electronic equipment
The production of electric and electronic equipment (EEE) is increasing worldwide.
Both technological innovation and market expansion continue to accelerate the replacement of equipment leading to a significant increase of waste electric and electronic equipment (WEEE). In west Europe, 6 million tonnes of WEEE were generated in 1998, the amount of WEEE is expected to increase by at least 3-5% per annum (European Commission, 2000).
Due to their hazardous material contents, WEEE may cause environmental problems during the waste management phase if it is not properly pre-treated. Many countries have drafted legislation to improve the reuse, recycling and other forms of recovery of such wastes so as to reduce disposal (European Parliament and Council, 2003; Silicon Valley Toxic Coalition, 2002).
Recycling of WEEE is an important subject not only from the point of waste treatment but also from the recovery aspect of valuable materials. The U.S.
Environmental Protection Agency (EPA) has identified seven major benefits when scrap iron and steel are used instead of virgin materials. Using recycled materials in place of virgin materials results in significant energy savings (as shown in Table 1 and 2) (ISRI, 1996).
Table 1
Recycled materials energy savings over virgin materials
Materials Aluminum Copper Iron and steel Lead Zinc Paper Plastics
Energy savings, %
95 85 74 65 60 64 >80
Table 2
Recycled materials energy savings over virgin materials
Materials Aluminum Copper Iron and steel Lead Zinc Paper Plastics
Energy savings, %
95 85 74 65 60 64 >80
Currently, recycling of WEEE can be broadly divided into three major stages:
x Disassembly (dismantling): Selective disassembly, targeting on singling out hazardous or valuable components, is an indispensable process.
x Upgrading: Using mechanical/physical processing and/or metallurgical processing to upgrade desirable materials content, i.e. preparing materials for refining process.
x Refining: In the last stage, recovered materials return to their Life Cycle.
Consumer electronic equipment (brown goods), such as television sets, radio sets, and video recorders, are most common. However, recent work on recycling of waste electric and electronic equipment primarily focused on personal computer and printed circuit boards scraps (Zhang et al., 2000; Macauley et al. 2003; Li et al., 2004; Veit et al., 2005).
The European Directive (2002/96/EC) on waste electric and electronic equipment (WEEE) has to be implemented into national legislation by 13 August 2004 (European Parliament and Council, 2003). According to the WEEE directive, member states shall ensure that, by 31 December 2006, producers meet the following targets:
x The rate of recovery for consumer electronic equipment shall be increased to a minimum of 75% by an average weight per appliance;
x Component, material and substance reuse and recycling for consumer electronic equipment shall be increased to a minimum of 65% by an average weight per appliance.
In order to meet the above targets, disassembly and mechanical recycling of consumer electronic scraps are of concern in European member states due to the fact that they are oriented to towards full materials recovery including plastics (Zhang and Forssberg, 1997; Langerak, 1997; Matsuto et al., 2004). In the practice of recycling of WEEE, selective disassembly (dismantling) is an indispensable process because it aims to remove hazardous or high value components (Stuart and Christina, 2003;
Basdere and Seliger, 2003; Torres et al., 2004). However, a study of potential future disassembly of electronic scraps indicated that full automation disassembly of consumer electronic scraps will not be economically attractive by 2020 (Boks and Tempelman, 1998). As a consequence, a mechanical process is of interest for upgrading metal content of consumer electronic scraps because it can yield high material recovery.
1.2. Mechanical recycling processes 1.2.1. Magnetic separation
Magnetic separators, in particular, low-intensity drum separators are widely used for the recovery of ferromagnetic metals from non-ferrous metals and other non-magnetic wastes. Over the past decade, there have been many advances in the design and operation of high-intensity magnetic separators, mainly as a result of the introduction of rare earth alloy permanent magnets capable of providing very high field strengths and gradients (Schubert, 1991).
1.2.2. Density-based separation
Several different methods are employed to separate heavier materials from lighter
ones. The difference in density of the components is the basis of separation. Gravity
concentration separates materials of different specific gravity by their relative
movement in response to the force of gravity and one or more other forces, the latter
often being the resistance to motion offered by a fluid, such as water or air (Wills,
1988). The motion of a particle in a fluid is dependent not only on the particle’s
density, but also on its size and shape, large particles being affected more than smaller
ones. In practice, close size control of feeds to gravity processes is required in order to
reduce the size effect and make the relative motion of the particle specific gravity
dependent.
The use of air to separate materials of differing density has long been known and is typified by the winnowing of grain using an air current to remove the chaff. Air tables have been used to eliminate a host of small problems in the food industry and in applications such as separating abrasive grains in the cleaning of foundry sand and removing metals from crushed slag (Fuerstenau and Han, 2003). In recent years, it also has been developed and implemented in a few electronic scrap recycling plants.
1.2.3. Electric conductivity-based separation
Electric conductivity-based separation separates materials of different electric conductivity (or resistivity). There are three typical electric conductivity-based separation techniques: (1) eddy current separation, (2) corona electrostatic separation, and (3) triboelectric separation (Meier-Staude and Koehnlechner, 2000; Schubert and Warlitz, 1994; Higashiyama and Asano, 1998; van Der Valk et al., 1982; Stahl and Beier, 1997).
In the past decade, one of the most significant developments in the recycling industry was the introduction of eddy current separators whose operability is based on the use of rare earth permanent magnets. When a conductive particle is exposed to an alternating magnetic field, eddy currents will be induced in that object, generating a magnetic field to oppose the magnetic field. The interactions between the magnetic field and the induced eddy currents lead to the appearance of electrodynamic actions upon conductive non-ferrous particles and are responsible for the separation process.
The separators were initially developed to recover non-ferrous metals from shredded automobile scrap or for treatment of municipal solid waste (Wilson et al., 1994;
Dalmijn and van Houwelingen, 1995; Gesing et al., 1998; Norrgran and Wernham, 1991), but is now widely used for other purposes including foundry casting sand, polyester polyethylene terephthalate (PET), electronic scrap, glass cullet, shredder fluff, and spent potliner (Hoberg, 1993; Dalmijn and van Houwelingen, 1996; Meyer et al., 1995; Wernham et al., 1993; Schubert, 1994; Mathieu et al., 1990). Currently, eddy current separators are almost exclusively used for waste reclamation where they are particularly suited to handling the relatively coarse sized feeds. However, the number of waste streams containing fine metal particles is foreseen to grow substantially in the near future (Rem et al. 2000). In recent years, there have been some developments of eddy current separation processed designed to separate small particles (Zhang et al. 1999, Rem et al. 2000).
1.2.4. Optical sorting process
With the fast development of Charge-Coupled Device (CCD) sensor, computing, and software technology, optical sorting process has been developed in both recycling and mineral processing industry (Kattentidt et al. 2003; Harbeck, 2001; Sötemann, 2000)).
In addition, recording more and better data with sensors improves the separation performance of automated sorting equipment. The measuring of particle properties like color, texture, morphology, conductivity and others allows high quality sorting of mixed materials into almost pure fractions. Multi-sensor systems by using two or more different sensors were of concern in the past years (Kattentidt et al. 2003).
An automatic sorting device named “CombiSense 1200” was developed by Separation
Systems Engineering (SSE), Wedel, Germany (Schäfer et al. 2003). This type of
sorting is a combined opto-electronical system which is operating with a belt width of
600 mm or 1200 mm. It combines the special characteristics of an optical system incorporating a high speed camera with a 1 billion colors recognition and a special conductivity sensor permitting the identification of a variety of metals. The CombiSense can handle mass streams of up to 10 tons/h for instance in the size classes 5-50 or 10-100 mm.
1.3. Objectives of the present work
As discussed above, recycling of waste electric and electronic equipment is an important subject not only from the point of waste treatment but also from the recovery aspect of valuable materials. However, recent work on recycling of waste electric and electronic equipment primarily focused on personal computer and printed circuit boards scraps. Recycling of consumer electronic scrap is only beginning.
It is of great importance to characterize consumer electronic scrap in order to develop
a cost effective and environmentally friendly recycling system. In the present study,
one of the major objectives is to investigate the characteristics of television scrap by
using a variety of methods, such as chemical analysis, particle size and shape analysis,
liberation degree analysis, thermogravimetric analysis, sink-float test, and IR
spectrometer. Mechanical processing technology has been widely utilized in recycling
industry. As a consequence, it is also the objective to develop an improved separation
system to separate valuable materials. Since eddy current separation plays a critical
role in recovery of non-ferrous metals from waste steams, an investigation of new
developments of eddy current separation is another objective.
2. Materials and methods 2.1. Materials
2.1.1. TV scrap sample
Television scrap sample was provided by Stena Technoworld AB, Bräkne-Hoby, an electronic recycling corporation in Sweden. End-of-life TVs of any model and brand with plastic houses that were collected primarily from Sweden were pre-dismantling to remove the cathode ray tubes, CRTs. Then the scraps were shredded into -12 mm particles. An approximately 30 kg of the TV scrap sample was procured and packed for the laboratory study. A detailed description of the sample preparation was given in paper 2.
A powdered sample was prepared by means of a turborotor grinder developed by Görgens Engineering GmbH, Germany, which is capable of grinding metallic materials and plastics. Before the grinding, ferrous metals were removed by a magnetic separator. This powdered sample was used for thermogravimetric analysis (TGA). The size distribution of the powdered sample analyzed by a Cilas 1064 Liquid instrument was shown in paper 2.
2.1.2. Pure material samples
A wide range of materials, such as copper, aluminum, plastics, glass, and stone was produced by cutting or grinding pure materials. Copper wires were provided by Draka Kabel Sverige AB, Sweden. The dimensions and shapes of materials to be investigated are presented in Table 3.
Table 3.
Dimension and shape of test materials Dimension and shape L uWuT (mm)
(sheet)
T uS (mm) (cylinder)
Size range, (mm) (Granulated Particles)
Material
14u14u2 20 u10u2 40u5u2
Al 3u3u2
34 u0.5 12u1.5 8u2.5
3 u6
Cu
5u5u2 Cu, PVC
2-6 Al, Glass, Stone
L: length, W: width, T: thickness, S: section area, PVC: polyvinyl chloride
2.2. Methods
2.2.1. Sampling standard deviation
In order to find out whether or not the test results are consistent, the weight of each specimen amounts up to 1.5 kg, and 2 or 3 specimens were analyzed for the chemical analysis and particle size analysis. The sample standard deviation, S is defined as followings (Montgomery, 2001):
2 / 1 2
1 _
)) 1 /(
) ) (
(( ¦ y y n S
n
i
i
(1) where, S denotes sample standard deviation, n is the number of samples to be studied, y
irepresents a sample, Cy indicates the sample mean.
2.2.2. Chemical analysis
Chemical analyses were carried out in the laboratory of OVAKO Steel AB, Hofors, Sweden. Samples were ground to powder and treated with aqua regia for dissolution of the metal. The plastic was then filtrated and the remaining solution analyzed with ICP/AES (inductively coupled plasma/ atomic emission spectroscopy) and ICP/MS (inductively coupled plasma/mass spectroscopy).
2.2.3. Particle size analysis
The specimens prepared for size analysis were initially dried up at 105qC for 12 hours. Subsequently, the samples were screened by employing an ASTM Retsch testing sieve series with square openings that were shaken off by a RO-TAP testing sieve shaker for 30 minutes.
2.2.4. Particle shape analysis
An image process system, produced by Kronton Elektronik GmbH, Germany, was utilized for particle shape analysis. The quantitative criterion is expressed in terms of FCIRCLE defined as follows (KRONTON, 1991):
FCIRCLE=4 SAREA/PERIM
2(2) PERIM=PERIMX+PERIMY+PERIMXY 2 (3) where AREA, is defined as the number of pixels multiplied by the scaled pixel area, PERIM is the perimeter of the object, PERIMX, PERIMY is the length of perimeter in x and y direction, respectively, PERIMXY is the length of perimeter having direction of 45 and 135 degrees to x-axis. In this case, holes in the object will contribute to the perimeter.
Eq. (2) shows that the values of circularity shape factor, FCIRCLE range between close to 0 for very elongated or rough objects and 1 for circular objects.
2.2.5. Liberation degree analysis Liberation degree can be simply expressed as:
LD=N
f/(N
f+N
l) (4)
where, LD is liberation degree, N
frepresents the number of free particles of the
desired material, and N
lindicates the number of locked particles of the same material.
In the present study, up to 2 kg sample was analyzed and the liberation degree of copper was calculated by Eq. (4).
2.2.6. Sink-float test
Sink-float test is an effective method to determine the density of characteristics sample. The heavy liquids that were used in the laboratory test were presented in Table 4.
Table 4
Heavy liquids and their densities employed in the sink-float test Heavy
liquids
H
2O NaCl+
H
2O NaCl+
H
2O
NaCl+
H
2O
CaCl
2+ H
2O
CaCl
2+ H
2O
Acetone+
TBE
Acetone+
TBE
Tetrabrome- ethane (TBE) Density,
g/cm
31.0 1.02 1.06 1.13 1.23 1.41 2.00 2.44 2.97
The densities of the liquids were detected by using a 25 ml volumetric flask and following equation:
D=(W
t-W
f)/25.00 (5) where D denotes the density of liquid, W
tis the total weight of liquid and the volumetric flask, W
fis the weight of the volumetric flask.
2.2.7. Quantification and identification of plastics
Thermogravimetric analyses (TGA) were performed by using NETZSCH STA 409 in both argon and air atmosphere to quantify the amount of plastics in TV scrap. In this test, the samples of 100 mg were heated linearly at a heating rate of 10 qC/min from 25 qC to 1200 qC with a gas flow rate of 100 ml/min.
Identification of plastics in the products of sink-float test was carried out by using the Perkin Elmer System 2000 FT-IR spectrometer, coupled with one FT-IR microscope.
Plastics pieces from sink-float test were also identified by using an industry-scale online infrared technique in Stena Technoworld AB, Sweden.
2.2.8. Magnetic separation
A low intensity drum magnetic separator, Mörsell Separator, was employed for removing ferrous metals from the sample (as shown in Fig. 1). In the present study, the drum peripheral speed is 2 m/s.
2.2.9. DGS Table separation
Air table separation was carried out by using a DGS-Sort 300D in MinPro AB, Stråssa, Sweden. The separator was developed by Fren Erschliessungs-und Bergbau GesmbH, Austria.
2.2.10.Eddy current separation
The eddy current separation experiments were conducted with a rotating drum eddy current separator, BM 29.710/18, developed by Bakker Magnetics, the Netherlands.
The BM 29.710/18 rotor has 9 pairs magnetic poles, the magnetic induction at the belt
surface is 0.32 T, and the dimension of the magnetic rotor is 300 mm.
Fig. 1. Flowsheet of magnetic separation
The separability of pure material sample was characterized by their distribution in an array of the collectors that were placed in front of the conveyor belt pulley (as shown in Fig. 2). Twelve collectors, each with dimensions of 500 u85u100 (lengthuwidthuheight) mm, were used. The material distribution was analyzed by its percent weight in each collectors such that:
% 100 ) /(
) (
12
1
¦ u
j ij ij
ij
W W
PW (6) where (PW)
ijis the percent weight of the ith material in the jth collector, and W
ijis the weight of the ith material in the jth collector.
Fig. 2. Illustration of rotating eddy current separation A: Magnetic drum rotates in a Forward mode B: Magnetic drum rotates in a Backward mode
No.12 ... No.1 Collectors
Belt Feed
Non-ferrous metals
A B
Scrap sample
Ferrous metals Non-ferrous metals
and non-metals
2.2.11.Optical (metal) sorting
The optical (metal) sorting process was performed by a Clara All-metal Separator (Scan & Sort GmbH, Wedel, Germany). As demonstrated in Fig. 3, the optical (metal) sorting appliance consisting of electromagnetic sensors and/or color line-cameras identifies the material on the belt and transmits the corresponding information to a high performance computer in milliseconds. A pneumatic ejection system with up to 256 valves shoots the selected material out of the product stream by air pressure.
Fig. 3. Demonstration of optical (metal) sorting system
2.2.12.Magnus separation
The Magnus separator (Fig. 4) was developed by Delft University of Technology, the Netherlands. At the present study, the magnetic rotor speed is 1000 rad/s for the dipole rotor.
Fig. 4. Schematic draws of the Magnus separator Water lever
Non-ferrous metals Feeder
Splitter
Non-metals
Magnetic rotor
2.2.13.Hand picking
Hand picking method was used in the evaluation of separation for qualitative and
quantitative analysis of products. Approximately 1 kg of each product sample was
separated by a chute riffling for hand picking. Subsequently, metals, printed circuit
boards and cables (PCBs), and plastics were separated from each other by hand.
3. Characterization of consumer electronic scrap
3.1. Chemical analysis
Table 5 shows the multi-element analysis result of TV scrap sample. From the result, it can be seen that TV scrap contains very low-grade of non-ferrous metals and precious metals, 1.2% Al, 3.4% Cu, 7 ppm gold, 20 ppm silver, and less than 6 ppm platinum and palladium. A comparison of TV scrap, personal computer scrap (Legarth et al., 1995), and printed circuit boards scrap (Zhang and Forssberg, 1997) is given in Table 6. It is apparent that the content of non-ferrous metals and precious metals in TV scrap is much lower than that of in personal computer or printed circuit boards scrap. From the point of view of recycling industry, the major economic drive force to process those scraps is recovery of non-ferrous metals and precious metals. Therefore, it is expected that recycling of TV scrap will not be economically viable by using conventional manual dismantling. Mechanical processing techniques may provide an alternative to separate copper and different plastics.
Table 5
Multi-element analysis of TV scrap samples
Al Cu Pb Zn Cr Mo Ni V Ag Au Pt Pd
% ppm
Assay 1.2 3.4 0.2 0.3 90 13 380 7 20 <10 <2 <2
Note: These results are the average obtained from two samples.
Table 6
Comparison of TV scrap, personal computer scrap, and printed circuit boards scrap
Al Cu Pb Zn Ni Ag Au
% ppm
TV scrap 1.2 3.4 0.2 0.3 0.038 20 <10 PC scrapa 2.8 14.3 2.2 0.4 1.1 639 566 Assay
PCBs scrapb 7.0 10.0 1.2 1.6 0.85 280 110
a
data source: Legarth et al. (1995),
bdata source: Zhang and Forssberg (1997)
3.2. Size and metal distribution of TV scrap
Fig. 5 gives the size cumulative distribution of TV scrap sample. From the figure, it can be seen that approx. 90% of particles is present in +5 mm size range; median size of the sample (d
50) is about 9 mm.
A cumulative oversize distribution of copper for TV scrap sample is presented in
Figure 6. We can see that approximately 90% of Cu is widely distributed in +2.36mm
fraction. This indicates that mechanical processing techniques, such as eddy current
separation, air table, jigging, and sink-float separation, may be employed in this size
range to recover copper. But this wide size range (2mm to 15mm) is also a challenge
for those mechanical separation techniques.
Particle size, mm
Cumulative undersize, %
0.1 1 10 100
0 20 40 60 80 100
Fig. 5. Size cumulative weight of TV scrap sample
0 10 20 30 40 50 60 70 80 90 100
1 10 100
Size range, mm
Fig. 6. Cu distribution in screening products
3.3. Particle shapes of materials in TV scrap
Fig. 7 shows images of non-ferrous metals (a), plastics (b), and printed circuit boards (PCBs) (c) separated from TV scrap sample. It is evident that non-ferrous metals are extremely heterogeneous, formed as wide variety of particle shapes such as, straight and bent bars, bent plates, cable and wire bundles. Furthermore, it can be seen that almost all of the plastics in TV scrap is black in color (Fig. 7 (b)). Therefore, with the
Cumulative distribution of Cu, %
fast development of CCD (Charge-Coupled Device) sensor technology, optical sorting process may provide a good choice to separate black plastics.
An image process system introduced by Kronton Elektronik was used to quantify particle shape factor, FCIRCLE (as shown in Figure 8). It is obvious from Figure 8 that the frequency distribution of FCIRCLE for non-ferrous particles varies to a large range (0.1-0.9); the frequency distributions of FCIRCLE for plastics and PCBs are mainly in the range of 0.6 to 0.9. This result indicates that non-ferrous metals particles in TV scrap sample form in a variety of shapes, much more different than that of plastics and printed circuit boards. The separation processes will be significantly influenced by the particle shape for recovery of non-ferrous metals.
It should be pointed out that shape separation techniques, primarily developed to control properties of particles in powder industry provide an alternative to separate non-ferrous metals from TV scrap (Cui and Forssberg, 2003). Shape separation by tilted plate and sieves is the most basic method that has been utilized in recycling industry. An inclined conveyor and inclined vibrating plate were used as a particle shape separator to recover copper from electric cable waste (Koyanaka et al., 1997).
3.4. Liberation degree of copper
It is well-known that the liberation of values in scraps is of primary importance for mechanical processing. The liberation degree of copper in TV scrap was quantified (as shown in Table 7). From the result, we can see that it is difficult to achieve complete liberation, since in this particle size copper in printed circuit boards and cables is almost impossible to liberate. This result indicates that printed circuit boards and cables in TV scrap may cause copper loss or low quality of copper product in mechanical processing.
Table 7
Liberation degrees of Copper in TV scrap
3.5. Sink-float test
The result of the sink-float test is given in Fig. 9 and Fig. 10. It is obvious that a high recovery of copper is obtained by using a sink-float process. For +1.4 g/cm
3fraction, the recovery of Cu is up to 88.4% with an assay of 42.4%. In addition, it must be pointed out that approximately 18% of the copper is distributed in –2.0+1.23 g/cm
3fraction with an assay of only 7%. As discussed in the liberation degree section, this is because copper in printed circuit boards is not liberated from plastics and ceramic materials.
Size range, mm Weight, % Liberation degree of Cu, %
+12.5 22.9 0.0
+9.5 25.7 0.0
-9.5+6.7 27.6 36.4
-6.7+4.75 14.3 54.3
-4.75+3.35 3.1 74.4
-3.35+2.36 3.5 73.4
-2.36+1.65 1.5 51.1
-1.65 1.4 n.d.
Fig. 7. Images of non-ferrous metals (a), plastics (b), and printed circuit boards (c) separated
from TV scrap sample (+2.36mm)
Fig. 8. FCIRCLE analysis of non-ferrous metals (a), plastics (b), and printed circuit boards (c)
separated from TV scrap sample (+2.36mm)
0 20 40 60 80 100
1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8 3,0 Density, g/cm
3C u m u la ti v e w ei ght of s inks , %
Fig. 9. Cumulative weight of sinks versus specific density for TV scrap (-9.5+1.65mm)
0 20 40 60 80 100
0,5 1 1,5 2 2,5 3
Density, g/cm
3Cumulative assay, % assay
distribution
Fig. 10. Cumulative data of copper for sinks versus specific density for TV scrap (-9.5+1.65mm)
3.6. Quantification of plastics by thermogravimetric analysis
In the present study, the sink-float test is oriented not only to evaluate the separability of copper but also to estimate the separability of different plastics. The plastics employed in TV set are primarily HIPS (high impact polystyrene), ABS (acrylonitrile butadiene styrene), PC (polycarbonate), and POM (Polyoxymethylene) with densities of 1.03-1.17, 1.03, 1.15-1.22, and 1.4, respectively (Menad et al., 1998; APC, 2000;
APME, 2001).
Thermogravimetric analysis (TGA) is widely utilized to quantify and identify plastics (Menad et al, 1998; Jakab, 2003; Braun and Schartel, 2004; Levchik et al., 2000;
Wang et al., 2003). In the present test, a HIPS particle from TV scrap was also analyzed in air atmosphere as a reference. Fig. 11 gives the TG/DTG/DTA curves of powdered TV scrap sample in air atmosphere (a), powdered TV scrap sample in argon atmosphere (b), and HIPS sample in air atmosphere (c). It can be seen from the curves that:
x The apparent reaction of powdered TV scrap occurs starting at the temperature of about 210 qC in both air (Fig. 11 (a)) and argon (Fig. 11. (b)) atmosphere.
The complete degradation of TV scrap sample takes place at approx. 924 qC.
At this temperature, the weight losses of samples are 86% and 78%, respectively. The difference of weight loss between air and argon atmosphere is because part of char is oxidized by oxygen at air atmosphere.
x Thermal decomposition of powdered TV scrap (Fig. 11 (a)) is much more complicated than that of pure HIPS (Fig. 11 (c)). From the DTA/DTG curves of Fig. 11 (a), we can see that at least three steps of decomposition of powdered TV scrap sample undergo with characteristic decomposition temperature of 268 qC, 432 qC, and 590 qC, respectively. Otherwise, HIPS sample decompose in one major step with characteristic decomposition temperature of 440 qC (Fig. 11 (c)).
Flame retardants are widely used in plastics to prevent or delay a developing fire in electronic equipment (Levchik et al., 2000; Braun and Schartel, 2004; Jakab et al., 2003; Hamm et al., 2001; Imai et al., 2003; Yamawaki, 2003; Riess et al., 2000). A detailed discussion of Flame retardants in electronic scrap is shown in paper 2.
3.7. Identification of plastics by FT-IR spectrometer
In order to evaluate the separability of plastics in TV scrap using density-based processes, plastics pieces in products of sink-float test were identified by a FT-IR spectrometer. Fig. 12 shows the spectra of plastics with the density range of – 1.02+1.0 g/cm
3, -1.06+1.02 g/cm
3, -1.23+1.13 g/cm
3, respectively.
It is obvious that similar spectra are obtained for plastic samples, which are distributed in various density ranges. In comparison with the spectrum of a commercial HIPS (as shown in Fig. 13) (Sidwell, 1997), the absorption bands at a3010, a2956, a1600, a1500, a1458, and a758cm
-1, are indications of HIPS contributed by aromatic ring and -CH
2-. The absorption bands at a1739 cm
-1can be recognized as characteristic absorption of ester that is common as flame retardants additive in plastics (Braun and Schartel, 2004; Carlsson et al., 2000; Imai et al., 2003;
Levchik et al., 2000; Sjödin et al., 2001).
In addition, identification of plastics in products from the sink-float test also carried out by using an industry scale infrared instrument in Stena Technoworld AB, Sweden.
From the results (Table 8) we can see that plastic in this scrap sample primarily is
HIPS, besides some ABS, PC, and POM. It can be seen that HIPS is widely present
from –1.0g/cm
3fraction to –1.23g/cm
3fraction. This specific density variation of the
same material is due to variations of additives of plastic and from enclosed cavities
and inclusions of other materials. Gravity separation of plastics may encounter some
challenges because of specific density variation of same material.
Fig. 11. Thermogravimetric analysis of a) powdered TV scrap in air atmosphere, b) powdered TV scrap in argon atmosphere, c) HIPS in air atmosphere
c) b) a)
590qC
Fig. 12. FT-IR spectra of plastics from the products of sink-float test
Table 8
Identification of plastics for the products of sink-float test (size range –9.5+1.65mm) Specific density,
g/cm
3-1.0 -1.02 +1.0
-1.06 +1.02
-1.13 +1.06
-1.23 +1.13
-1.41 +1.23
+1.41
Identification of plastics
HIPS HIPS HIPS HIPS, SAN HIPS PC, POM -
Fig. 13. Infrared spectrum of a commercial HIPS
4. Mechanical separation of consumer electronic scrap
4.1. Ferromagnetics recovery
Table 9 shows the chemical assay of ferromagnetics from the TV scrap. It is clear that a high grade of ferromagnetics product can be produced by employing a low intensity magnetic separator. It must be pointed out that due to the high contamination levels of Cu, Al, and Pb, this ferromagnetics fraction may not correspond to the requirements of iron and steel smelters.
Table 9
Chemical assay of ferromagnetics from the TV scrap
Chemical Assay, %
Weight, %
Fe Cu Al Ni Pb Ag Au Ferromagnetics 22.1 90.10 5.70 0.900 2.000 0.960 0.000 0.000
4.2. DGS Table separation
Fig. 14 gives the separation results of DGS table separation. It can be seen that 70% to 90% of metals are recovered in the heavy product with metal content between 40%
and 60%. In addition, printed circuit boards and cables in the sample are difficult to separate from plastics by the DGS table. The result indicates that DGS table separation is effective and efficient for recovery of metals from consumer electronic scraps. Printed circuit boards and cables should be dismantled before further mechanical separation. A number of parameters must be optimized on DGS table separation. Paper 3 gives a detailed discussion of those parameters.
0 10 20 30 40 50 60 70
60 70 80 90 100
Recovery, %
Grade, %
Metals
0 10 20 30 40
10 20 30 40
Recovery, %
Grade, %
PCBs
Fig. 14. Grade-Recovery of metal and printed circuit boards in the heavy product from the
DGS table separation
4.3. Eddy current separation
The separation of non-ferrous metals from the -9.5+6.7 mm fraction and -3.35+1.65 mm fraction of shredded TV scrap performed after an optimization of the operating conditions by using a rotating drum eddy current separator. As shown in Table 10, more than 75% of non-ferrous metals were recovered, while maintaining a purity of 27% in a single pass for the large particle size fraction. However, only 45% of non- ferrous metals can be separated for the small particle size fraction. This result indicates that application of traditional eddy current separation in recycling of consumer electronic scraps may encounter a problem because the limitation of particle size. New development of eddy current separation for recovery of fine particles is required.
Table 10
Eddy current separation result of TV scrap
Particle size, mm Products Weight, % Metal content, % Recovery, %
Non-ferrous metals 34 27 77
-9.5+6.7 Waste 66 4 23
Total 100 12 100
Non-ferrous metals 19 39 45
-3.35+1.65 Waste 81 11 55
Total 100 16 100
4.4. Optical sorting
The optical (metal) sorting experiments by using color and/or metal sensors were carried out in Scan & Sort GmbH, Wedel, Germany. Two samples with particle size of +9.5 mm and -9.5+4.6 mm, were processed respectively (as shown in Fig. 15).
Table 11 and 12 give the results of optical (metal) sorting of TV scrap. It is evident that 90% of metals can be recovered in metallic product by utilizing optical sorting system.
Table 11
Optical sorting result of TV scrap (+9.5mm)
Weight, % Metal content, % Recovery, %
White fraction 37 75 60
Metallic product from dark fraction
32 40 32
Non-metallic product from dark fraction
31 1 8
Total 100 41 100
Table 12
Optical sorting result of TV scrap (-9.5+4.6 mm)
Weight, % Metal content, % Recovery, %
Metallic product 55 47 90
Non-metallic product 45 6 10
Total 100 29 100
Fig. 15. Flowsheet of optical (metal) sorting process of TV scraps White product
TV scraps (+9.5 mm)
Color sorting
Dark product
Non-metallic product Metallic product
Metal sorting
Non-metallic product Metallic product
TV scraps (-9.5+4.6 mm)
Color sorting
White product Dark product
Metal sorting Metal sorting
Metallic product
5. New developments of eddy current separation for fine particles
5.1. Theory
5.1.1. Magnetic interaction
A magnet rotor with k pairs of magnet poles and a magnetic induction b
mat the radius R
mof the outer shell surface produces a magnetic induction outside the shell (r>R
m):
B= ¸¸
¹
·
¨¨ ©
§
¸
¹
¨ ·
©
§
¸ ¸
¹
·
¨ ¨
©
§
) ( sin
) (
1
cos
t k
t k r
b R B B
m m k
m m r
Z M
Z M
M
(7) where (r, M) are cylindrical coordinates with respect to the axis of the rotor, t is time and Z
mis the angular velocity of the rotor.
The expression shows that a stationary particle at some point (r, M) experiences a magnetic induction of constant magnitude B
b
m(R
m/r)
k+1revolving at angular velocity -kZ
m(Fig. 16.). If the particle itself is spinning with some angular velocity :, it perceives a field of the same size as a stationary particle but now rotating at an apparent angular velocity -k Z
m- :. The magnetic torque makes the particle spin in the same direction as the magnetic field.
Fig. 16. Magnet rotor (left) produces a rotating magnetic field B inducing eddy currents in a particle (right) resulting in a particle magnetic moment M.
For particles of simple geometries, such as spheres, thin disks and long cylinders, with a size that is small with respect to the magnetic wavelength 2 SR
m/(k+1) of the rotor, the theory of eddy current separation (Rem, 1999) provides an expression for the particle magnetic dipole moment M in a rotating magnetic field:
M=
» »
¼ º
« «
¬ ª
¸¸ ¹
·
¨¨ ©
§ :
¸
¸
¹
·
¨ ¨
© : §
r m
r
m
B
d B k
B I d B k
V R
II
V Z P V
Z
P ( P ( ) ) ( ( ) )
2 0
2 0
0
(8) N
N N
N S
S
S
S M
B
where V and V are the volume of the particle and its electrical conductivity, respectively, and R([) and I([) are dimensionless functions, for which approximations in terms of rational functions are tabulated in Table 13 (Rem et al., 2002; Fraunholcz et al., 2002).
Table 13.
Parameters defining the magnetic interaction for particles of several shapes and parallel (ӝ) or perpendicular ( ՚) orientations of their axis of symmetry with respect to the axis of the rotor
Shape (R([), I([)) D c
mSphere 21([
2, 42[)/20(1764+[
2) D 1/40
Cylinder ӝ 3([
2, 24[)/2(576+[
2) D 1/16
Cylinder ՚ 9([
2, 24[)/8(576+[
2) D 3/64
Disk ӝ ([
2, 12[)/(144+[
2) G 1/12
Disk ՚ (0.6SG[
2/D, 16[)/4(256+(0.6SG)
2[
2/D
2) D 1/64 D: diameter, G: thickness.
As a consequence, the torque T
mon the particle from its magnetic moment is given by (Rem et al., 2002; Fraunholcz et al., 2002):
T
m=M uB= ( )
0 2
P I [ V B
e
z(9) the direct magnetic force F
mcan be written by:
F
m=MB= ¸¸
¹
·
¨¨ ©
§
) (
) ) (
1 (
0 2
[ [
P I
R r
V B
k
(10)
For conductive particles with d less than 10 mm, the factor I in T
mreduces to a linear function of Z
m:
V d B k
c
T
m m( Z
m: )
2V
2(11) where, the coefficient c
mdepends on the shape and orientation of the particles (Table 13).
5.1.2. Magnus effect
It is known that a spinning particle moving through a fluid experiences a force perpendicular both to its direction of motion and to the axis of rotation. This phenomenon is called the Magnus effect (Massey, 1989).
As shown in Fig. 17, the trajectory of a spinning particle falling in a fluid can be
analyzed to the forces of drag, lift and drag torque (Reynolds number Re>300) (Rem
et al., 2002; Fraunholcz et al., 2002):
F
LI :
F
DV Gravity-buoyancy
Fig. 17. Force diagram for a particle that rotates at an angular velocity : while settling with a linear velocity v with respect to a fluid
F
D=c
DUv
2A/2 (12) F
L=c
LUv
2A/2 (13)
: U : D
5c
T
d T(14) where c
D, c
L, and c
Trepresent the coefficients that depend on the shape and orientation of the particle (Table 14), U is the density of fluid, v is the particle velocity, A is the characteristic area of the particle, D is the characteristic dimension of the particle, : is the angular velocity of the particle (assuming that : is always perpendicular to v).
The speed of rotation : of the conductive particles in a Magnus separation is found by integration of the balance of angular momentum:
J:=T
m-T
d(15) Eq. (15) implies that within the size ranges indicated, the particle spin in a Magnus separation does not depend on the particle size, but only on its shape and orientation, since J, T
m, and T
dare all proportional to the fifth power of the particle size.
Table 14
Measured valuesor the drag torque coefficient for particles of several shapes
Particle definition c
TRough sphere (Re=300-700) 0.007
Smooth sphere (Re=3 u10
6) 0.0008
Rough cylinder (Re=500-700, L/D=3) 0.008 L/D
Smooth cylinder (Re=2u10
6, L/D=5) 0.0012 L/D
Disk (Re=300-30000, D/ G=3.5-4) 0.03
D
r h 5.1.3. Wet eddy current separation
In order to simplify the calculation, we assume a spherical particle with diameter D that is connected to a surface by a cylindrical mass of water (as shown in Fig. 18). For a completely wettable solid particle, the adhesion work Wa between particle and water is much higher than the cohesion work of water, W
C(Lu et al., 2005). As a result, the energy between a wettable solid particle and water can be written by:
E=2 SrhW
C(16) where r and h are the radius and height of the water cylinder. Geometrical analysis shows that radius r= h ( D ) h | Dh (h<<D). Additionally, the work of cohesion W
Cis expressed as:
W
C=2J
gl(17) here, the surface tension of water J
gl=73u10
-3J/m
2.
By putting the Eq. (17) to Eq. (16), the force gluing the particle to the belt surface is given as:
Dh dh
dE
F / | 6 SJ
gl(18)
For instance, if D=3 mm and h=0.2 mm, the force F=1.1u10
-3N, which is about the same order as the gravity force on a stone particle with a same particle size.
Fig. 18. Geometry of wet bond
Although the adhesive force is strong enough to keep most of the non-metal particles glued to the belt surface, the eddy current torque can easily provide the force to break the water bond for the non-ferrous metal particles. As discussed above, the magnetic torque is expressed as Eq. (11). The non-ferrous metal particle is able to break loose if the torque is of the order FD/2. For a typical water layer, h=0.2 mm, and on a traditional rotating drum eddy current separator, B=0.3 T, Z=150 rad/s, this criterion is met for well-conducting metals if D>1 mm, whereas for metals like solder and lead it is realized for D>2 mm (Table 15).
5.2. Traditional eddy current separation for fine particles
Fig. 19 demonstrates the material distribution for large particle size. It is obvious from
Fig. 19. a) that, when the eddy current separator run in the forward mode, almost all
the aluminum particles is distributed in the collectors of No. 1 to No. 4, otherwise
PVC particles are distributed in the collectors of No. 6 to No. 8. Analysis of the
material distribution indicates that it is easy to separate large aluminum particles from non-metals, when the magnetic drum rotates in the forward mode. It can be seen from Fig. 19. b) that, when the eddy current separator run in the backward mode, aluminum particles are widely distributed in collectors of No. 1 to No. 10. This result indicates that it is difficult to separate large non-ferrous metals from non-metals when the magnetic drum rotates in the backward mode.
Table 15
Electrical conductivity of some metals and alloys
Alloy Conductivity V, (1/m)
Aluminum 3003 27u10
6Copper 56u10
6Zinc 17u10
6Yellow brass 15u10
6Lead 5u10
6Solder 50-50 7u10
6Fig. 19 also shows the effect of particle shape on eddy current separation. It is clear that, in the forward mode, the deflections of square plates of Al are larger than those of the rectangular sheets since a square plate is more conducive to eddy-current induction than a rectangular sheet.
The material distribution for fine particles is presented in Fig. 20. It can be seen that fine conducting particles like copper are either mixed up with the non-metals ones or distributed in the collectors that are closer to the magnetic drum. The results indicate that it is difficult to separate fine non-ferrous metals from non-metals selectively, when the magnetic drum rotates in the forward mode. It has been found that if the magnetic drum rotates in the backward mode, separation of fine conducting particles from non-conducting ones is improved drastically. It is shown in Fig 20 that more than 80% of copper particles are distributed in the collectors of No. 1 to No. 6.
Separation of copper wires demonstrated in Fig. 20 shows that fine copper cable and wires can be recovered by traditional rotating drum eddy current separator in a backward mode.
5.3. Preliminary study of Magnus separation and wet eddy current separation
5.3.1. Effect of splitter position
The effect of splitter position on wet eddy current separation of aluminum is
demonstrated in Fig 21. It is observed that the recovery of Al is decreasing slowly, as
the splitter moving from 300 mm to 335 mm. In the meanwhile, the grade of Al
product increases from 26% to 63%. In order to ensure maximum the aluminum
recovery, the splitter position for the rest test was set to 335 mm horizontally away (x)
from the axis of the rotor.
1
2 3
4 5 6 7 8 9 10 11 12
40*5*2
20*10*2
14*14*2
PVC
0 20 40 60 80 100
Weight, %
collectrorNo.
Particle size, mm a)
1 2
3 4
5 6
7 8
9 10
11 12
40*5*2 20*10*2 14*14*2 PVC 0 10 20 30 40 50 60 70 80 90
Weight, %
collectror No.
Particle size, mm b)
Fig. 19. Material distribution for large particle size
(volume of Al particle=400 mm
3, a) forward mode, b) backward mode)
1 2
3 4
5 6
7 8
9 10
11 12
3*3*2 3*6 8*2.5 12*1.5 34*0.5 PVC 0 20 40 60 80 100
Weight, %
collectror No.
Particlesize, mm a)
1 2
3 4
5 6
7 8
9 10
11 12
3*3*2 3*6 8*2.5 12*1.5 34*0.5 PVC 0 10 20 30 40 50 60 70 80 90
Weight, %
collectrorNo. Particlesize, mm
b)
Fig. 20. Material distribution for fine particle size
(volume of Cu particle=18 mm
3, a) forward mode, b) backward mode)
5.3.2. Effect of rotor speed
The effect of rotor speed on wet eddy current separation of aluminum is exhibited in Fig. 22. As can be seen in Fig.22, the grade of Al product is slightly decreasing as the rotor speed increasing from 1000 rpm to 1500 rpm due to a drastic particle-particle interaction. However, the rotor speed from 1000 to 2000 rpm insignificantly influences the recovery of aluminum. It indicates that a high rotor speed that is widely used in traditional rotating drum eddy current separation is dispensable in wet eddy current separation. This result is sufficiently consistent with the preliminary study by Settimo et al. (2004).
0 20 40 60 80 100
250 300 350 400 450
Splitter position, mm
%
Grade Recovery
Fig. 21. Effect of splitter position on eddy current separation of Al
(rotor speed=1500 rpm, belt speed=1 m/s, moisture content=10%, particle size=4-6 mm).
0 20 40 60 80 100
500 1000 1500 2000 2500
Rotor speed, rpm
% Grade
Recovery