Use, Storage, and Disposal of Electronic Equipment in Switzerland

Mar 15, 2017 - Devices included are mobile phones, desktop and laptop computers, monitors, cathode ray tube and flat panel display televisions, DVD pl...
0 downloads 0 Views 1MB Size
Subscriber access provided by University of Newcastle, Australia

Article

Use, storage and disposal of electronic equipment in Switzerland Esther Thiébaud, Lorenz M. Hilty, Mathias Schluep, and Martin Faulstich Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b06336 • Publication Date (Web): 15 Mar 2017 Downloaded from http://pubs.acs.org on March 16, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28

Environmental Science & Technology

1

Use, storage and disposal of electronic equipment in

2

Switzerland

3

Esther Thiébaud*1), Lorenz M. Hilty1),2),3), Mathias Schluep4), Martin Faulstich5)

4

1) Empa, Swiss Federal Laboratories for Materials Science and Technology, Technology and

5

Society Laboratory, Lerchenfeldstrasse 5, CH-9014 St. Gallen, Switzerland

6

2) University of Zürich, Department of Informatics, Binzmühlestr. 14, CH-8050 Zürich, Switzer-

7

land

8

3) KTH Royal Institute of Technology, Centre for Sustainable Communications (CESC), Lind-

9

stedtsvägen 5, S-10044 Stockholm, Sweden

10

4) World Resources Forum (WRF), Lerchenfeldstrasse 5, CH-9014 St. Gallen, Switzerland

11

5) Clausthal Environmental Institute (CUTEC), Leibnizstraße 21+23, D-38678 Clausthal-

12

Zellerfeld, Germany

13

ACS Paragon Plus Environment 1

Environmental Science & Technology

Page 2 of 28

14 15

TOC Art:

16 17

ACS Paragon Plus Environment 2

Page 3 of 28

Environmental Science & Technology

18 19

Abstract

20

For an efficient management of these resources, it is important to know where the devices are lo-

21

cated, how long they are used and when and how they are disposed of. In this article, we explore

22

the past and current quantities of electronic devices in the in-use stock and storage stock in Swit-

23

zerland and quantify the flows between the use, storage and disposal phase with dynamic materi-

24

al flow analysis (MFA). Devices included are mobile phones, desktop and laptop computers,

25

monitors, cathode ray tube and flat panel display televisions, DVD players, and headphones. The

26

system for the dynamic MFA was developed as a cascade model dividing the use phase in first,

27

second and further use, with each of these steps consisting of an in-use stock and a storage stock

28

for devices. Using a customized software tool, we apply Monte Carlo simulation to systematical-

29

ly consider data uncertainty. The results highlight the importance of the storage stock, which ac-

30

counts for 25% (in terms of mass) or 40% (in terms of pieces) of the total stock of electronic de-

31

vices in 2014. Reuse and storage significantly influence the total lifetime of devices and lead to

32

wide and positively skewed lifetime distributions.

33 34 35 36

Keywords: In-use stock; storage stock, disposal pathways, lifetime, dynamic material flow analysis

Electronic devices contain important resources, including precious and critical raw materials.

ACS Paragon Plus Environment 3

Environmental Science & Technology

Page 4 of 28

37 38

Introduction

39

secondary resources. Over the past twenty years, waste electronic equipment (e-waste) has been

40

collected in Switzerland and sent for recycling, where, among other materials, base and precious

41

metals are recovered. Several rare technical metals such as indium, gallium, tantalum, or the rare

42

earth metals are not recycled. Existing efforts to improve recycling processes are hampered by

43

missing information on the location and quantities of these metals, the complex structure of elec-

44

tronic equipment (EE), low concentrations per device, the metallurgical limits of recovery pro-

45

cesses, as well as lack of economic incentives.1–5

Electronic devices we use or store at our homes can be viewed as an anthropogenic stock of

46

Stocks and flows of EE and the critical raw materials they contain6 have been subject to vari-

47

ous recent investigations. The flows entering collection and recycling schemes are often mod-

48

elled quantitatively by combining sales or stock data with estimated product lifetimes, using a

49

dynamic material flow analysis (MFA) approach.7–17 Most of these studies are limited to one or a

50

few electronic device types and present results either for EE flow only, or together with qualita-

51

tive or quantitative recovery potentials of critical raw materials in specific types of EE.

52

Many of these studies have shown discrepancies between high sales flows and low collection

53

flows or long phase-out periods of technologies that are no longer sold for certain device

54

types.10,11,17–21 These findings suggest that there are other significant disposal pathways than the

55

official e-waste collection. Other possible disposal pathways are, for example, export for reuse or

56

recycling, waste incineration (via the municipal solid waste collection pathway), and illegal

57

dumping. Additionally, as we will show in the following, e-waste generation can be significantly

58

slowed down due to reuse or storage of obsolete equipment, which is often not explicitly taken

59

into account. Only preliminary investigations in this direction can be found in literature so far.

60

Parajuly et al.7 include reuse and storage of IT and telecommunication equipment and consumer

ACS Paragon Plus Environment 4

Page 5 of 28

Environmental Science & Technology

61

electronics in their MFA system, but do not quantify related stocks and flows. Similarly, Yoshida

62

et al.16 mention both storage and reuse of computers, but only quantify domestic reuse flows.

63

Chung et al.22 evaluate the amount of stored televisions (TVs) and computers in China. The

64

amount of stored TVs and e-waste in general in US households is analyzed by Milovantseva et

65

al.23 and Saphores et al.24. Polák and Drápalová14 include storage times for mobile phones and

66

Sabagghi et al.25, as well as Williams und Hatanaka26 report storage times of personal computers

67

in their studies. Steubing et al.15 include reuse and storage times in their MFA of computers and

68

monitors, and quantify related flows. To the best of the authors' knowledge, there is, however, no

69

dynamic MFA research that includes reuse and storage times and differentiates between in-use,

70

reuse and storage stocks and the related flows of EE.

71

In an attempt to close this research gap, we conducted a survey on the service lifetime, storage

72

time, and disposal pathways of EE in Switzerland between 2014 and 2016. The findings high-

73

light the influence of storage and reuse on the time a product remains in the use phase27. In this

74

article, we present a dynamic MFA of nine different types of EE in Switzerland. We develop a

75

model to identify past and current in-use stocks and storage stocks of EE and quantify in detail

76

the flows between and from the use, storage and disposal phases. Based on the in-depth 'bottom-

77

up' data we are able to differentiate between new and (re)used devices. Such a model is a signifi-

78

cant step towards a deeper understanding of the behavior of device types in the use phase, such

79

as the above mentioned discrepancy between high sales flows and low collection flows. It facili-

80

tates detecting leakages where devices and the incorporated resources leave the system, as well

81

as determining their final sink. In combination with assumed or extrapolated sales flows, the

82

model can also serve to forecast the composition of future recycling flows, a capability which

83

enables recycling system managers to provide appropriate and tailored recycling capacities and

ACS Paragon Plus Environment 5

Environmental Science & Technology

Page 6 of 28

84

technologies.27 The present article will introduce the model and the results in terms of mass

85

flows; the implied flows of critical raw materials incorporated in EE (with indium, neodymium

86

and gold as examples) will be published in a planned subsequent article.

87

To implement the model, we developed an open source software tool in Python 3, allowing for

88

the calculation of dynamic MFAs in a generic and flexible way. The tool implements 'dynamic

89

probabilistic material flow analysis', a combination of dynamic material flow modeling with

90

probabilistic modeling, as proposed by Bornhöft.28 The probabilistic approach allows to system-

91

atically dealing with data uncertainty in our model.

92 93

Method

94

as proposed in Müller et al.29

95 96

Overview The purpose of the dynamic MFA model is to quantify in-use stocks and storage stocks of nine

97

types of EE with a high content of indium, neodymium or gold, so that they cumulatively cover

98

around 90% of these three metals in private Swiss households30 (see Table 1) and to map and

99

evaluate in detail the flows between the use, storage and disposal phases.

100

This chapter is structured according to the ODD (overview, design concepts, details) protocol

Table 1: Electronic device types included in the MFA. Device type Description

UNU -Key

Fraction containing Indium

Neodymium

Gold

Magnets in hard disk drive (HDD), optical PWB34 0302 drive30,31 / printed wiring board (PWB)32,33

Desktop

Desktop computer (incl. all-in-one computer, excl. peripherals)

Laptop

Laptop computer/ Notebook

Monitor

Flat panel display (FPD) monitor LCD35

Mobile

Conventional mobile phone

Liquid crystal Magnets in HDD, optical drive, loudspeak- PWB34 0303 display (LCD)35 er30,31 / PWB32,33

LCD35

PWB32,33

PWB34 0309

Magnets in loudspeaker, vibration alarm30 / PWB34 0306

ACS Paragon Plus Environment 6

Page 7 of 28

Environmental Science & Technology

PWB32

phone Smart phone Smart phone Headset CRT TV

Cathode ray tube (CRT) TV

FPD TV

Flat panel display (FPD) TV

Magnets in loudspeaker, vibration alarm30 / PWB34 0306 PWB32 Magnets in Loudspeaker30,36

Headphones / Headset

DVD player DVD player / Blu-ray player

101 102

LCD35

PWB LCD35

32,33

0401 PWB

34

0407

PWB32,33

PWB34 0408

Magnets in optical drive37 / PBW32,33

PWB34 0404

Unlike Thiébaud et al.27, we do not include loudspeakers, as no sales data are available for such devices.

103

The MFA system includes the process 'use phase', divided into 'active use' and 'storage' (Figure

104

1). The 'active use' and 'storage' are further divided into the subprocesses 'first use', 'second use'

105

and 'further use' as well as 'first storage', 'second storage and 'further storage'. Each subprocess

106

includes an in-use stock or storage stock. The system boundary corresponds to the Swiss national

107

border. The temporal extent is different for each device type. It starts when an electronic device

108

type was first put on the market and ends with the year 2014. The temporal scale is one year.

109

Based on the empirical results of our survey, we developed a cascade model, with each step

110

consisting of an in-use stock and storage stock. The first step includes the first use and storage of

111

new devices, the second step the second use and storage of second-hand devices. The third and

112

final step summarizes all possible further uses and storages. The cascade model enables the rep-

113

resentation of different service lifetimes, storage times and disposal pathways both for new and

114

for used devices.

115

The system (see Figure 1) has one inflow corresponding to the sales of new devices and four

116

outflows, one for each disposal option. Internal flows include 'reuse' flows from in-use stock and

117

storage stock to the next in-use stock and 'storage' flows from in-use stock to storage.

ACS Paragon Plus Environment 7

Environmental Science & Technology

Page 8 of 28

118 119

Figure 1: Cascade model of the process 'use phase', divided into 'active use' and 'storage'.

120 121

Design concepts

122

from the net flow by using the balance of masses (equation (1)), with the constant sampling rate

123

T = 1 year.29

The model employs a retrospective top-down approach, deriving the stock S[n] at a time n

   

  ∙    1 124 125

(1)

The outflows of the in-use stock and the storage stock are calculated according to equation (2).29 

       ∙ 

(2)



126

The model applies different lifetime distribution functions f[m] for the service lifetimes and

127

storage times of new and used devices. The service lifetime is defined as the time of active use of

128

a device. The storage time is defined as the time between the end of the active use of a device and

ACS Paragon Plus Environment 8

Page 9 of 28

Environmental Science & Technology

129

its final disposal or transfer to a different user.27 The flows from the different stocks to the in-

130

use stock, storage, or disposal pathways are determined by transfer coefficients. All data include

131

changes over time provided that time series are available.

132

We accounted for data uncertainty by modelling inflows, transfer coefficients and mass of de-

133

vices as probability distributions. We chose either normal or triangular distributions, based on

134

the characteristics of the available data. The dependent variables are calculated by Monte Carlo

135

simulation and again provided as probability distributions.28,38 The uncertainty of the lifetime

136

distribution parameter was taken into account by conducting a sensitivity analysis with a lower

137

and upper lifetime distribution scenario.

138

Details

139 140

Initial condition In our model, stocks of EE are calculated from inflow data and lifetime distribution functions

141

(equation (1) and (2)), starting from a stock = 0. For sales data, the model thus demands infor-

142

mation back to the year when a device type was first brought onto the market. In cases where

143

such data was not available, we extrapolated existing sales figures.

144 145

Model input data Inflow data for desktops, laptops and flat screen monitors were taken from annual ICT market

146

reports for Switzerland, which include both the business and the home segment.39. Sales data of

147

television sets (CRT and FPD TVs) and DVD players were collected by the Swiss Consumer

148

Electronics Association40 and the market research company GfK41. GfK additionally reports

149

sales data for mobile phones and smartphones.41,42 We further assumed that the number of sold

150

headsets corresponds to the sum of all sold portable audio and video devices, smart phones and

151

mobile phones. The uncertainties of inflow data were modeled as normal distributions. The

ACS Paragon Plus Environment 9

Environmental Science & Technology

Page 10 of 28

152

standard deviation (SD) was estimated at 10% of the inflow value for time series with available

153

data from Weissbuch, GfK or SCEA.43 For extrapolated time series, we assumed a SD of 20%.

154

As the inflow data for headsets are based on rough estimations, we assumed a SD of 30%.

155

Data regarding the service lifetime, the storage time, and the disposal pathways of EE were

156

taken from Thiébaud et al.27 They are provided as histograms of the service lifetimes and storage

157

times of new and second hand devices of ten different EE types. The temporal change of the ser-

158

vice lifetime of new devices was included where appropriate by dividing the histograms into

159

sales year groups as defined in Thiébaud et al.27. For the temporal change of the storage time of

160

new devices, we calculated for each device the year it was put to storage and subsequently divid-

161

ed the histograms into storage year groups. For the third step of the cascade model, no data on

162

the service lifetime or storage time are available, although the survey data confirms that for most

163

device types, there are some second-hand devices that are further used.27 Due to the lack of data,

164

we assumed that the same service lifetimes and storage times as for the second use apply for all

165

further uses in total (modelled as the third step of the cascade model). This assumption is tested

166

by a sensitivity analysis.

167

In order to estimate the lifetime distribution functions for the dynamic MFA, we fitted Weibull

168

distribution functions to the normalized histograms of the service lifetimes and storage times of

169

new and second-hand devices. Modeling positively skewed distributions of product lifetimes is

170

typically done with Weibull distributions, as they can adopt many different shapes.44,45

171

The uncertainty of the service lifetime and storage time is at least ±1 year for all histograms.27

172

The tool developed for the simulation of the dynamic MFA supports probabilistic modelling of

173

the inflows and transfer coefficients, but not of lifetime distribution parameters. In order to eval-

174

uate the sensitivity of our model to the Weibull distribution parameters as well, we shifted the

ACS Paragon Plus Environment 10

Page 11 of 28

Environmental Science & Technology

175

histograms provided by Thiébaud et al.27 by ±1 year and fitted the Weibull distribution functions

176

again. The model was then run additionally with the resulting lower and upper lifetime distribu-

177

tion scenarios for each device type.

178

Thiébaud et al.27 further provide transfer coefficients of the flows between the processes 'active

179

use' and 'storage' as well as to the different disposal pathways: 'donation', 'collection', 'municipal

180

waste' and 'unknown'. We modelled the 'donation' option as an export flow, as most donation

181

projects for used electronics are situated abroad.46,47 Due to the limited amount of available data

182

sets, the temporal change of transfer coefficients was analyzed for CRT TVs, mobile phones,

183

laptops and desktops only. For the third step of the cascade model, the transfer coefficients for

184

second-hand devices were adapted by splitting the reuse flows proportionally to the other dispos-

185

al pathways. As the uncertainty of the transfer coefficients increases with smaller sample sizes

186

per process, we introduced a triangular distribution for each transfer coefficient, varying

187

with ±



√

, with n being the total number of devices transferred from a specific process.

188

To simulate the distributions of total lifetime, that is, the time devices remain in the use phase

189

from sales to disposal, we run an impulse response analysis. For each device type, we fed an im-

190

pulse of 100,000 devices in one year into the cascade model and simulated the succeeding peri-

191

ods of 50 years. This includes all different service lifetimes and storage times, as well as the per-

192

centage of products that enter reuse and storage according to the transfer coefficients provided by

193

Thiébaud et al.27 The response at the outflow of the overall use phase, corresponding to the sum

194

of all four disposal pathways and adding up to the 100,000 devices, is equivalent to the total life-

195

time distribution function.

196

The average mass per device of a given type was determined by own measurements and litera-

197

ture data.9,17,34,48,49 We introduced a triangular distribution for the mass of each device type, de-

ACS Paragon Plus Environment 11

Environmental Science & Technology

Page 12 of 28

198

termined by the weighted mean and the lowest and highest value. The lowest and highest values

199

were extended by taking into account the SD of the SD, as suggested in JCGM,

200

lower and upper limit of the triangular distribution should lie within the 95% confidence interval.

201

For the mass per device type, available data did not show any significant temporal change.

50

so that the

202

More information on the inflow data, an overview of all fitted Weibull distribution functions

203

including details on the sales year and storage year groups, the sensitivity analysis of the third

204

step of the cascade model, the transfer coefficients, the impulse response analysis, and the uncer-

205

tainty distributions of mass per device type can be found in the Supporting Information (SI).

206 207

Model output data and evaluation Model output data comprise stochastic time series of all stocks and flows occurring in Figure

208

1. The distribution of the time series is visualized by the 10th percentile, mean and 90th percentile.

209

Additionally, the lower, mean and upper lifetime distribution scenarios as described above are

210

presented. With the exception of Sankey diagrams, the in-use stocks 1, 2 and 3 and storage

211

stocks 1, 2, and 3 are aggregated to the total in-use stock and total storage stock.

212

The output can be compared to statistical data from various sources for model validation. From

213

the Swiss Federal Statistical Office (FSO), time series data on the in-use stocks of desktops, lap-

214

tops, CRT TVs, FPD TVs and DVD players are available from 2006 to 2013.51 The Swiss Feder-

215

al Office of Communications (OFCOM) provides in-use stock data for mobile phones from 1990

216

to 201452. Swico Recycling, the collection system for EE in Switzerland, publishes numbers and

217

tonnages of desktops, FPD monitors, laptops, mobile phones, CRT and FPD TVs they collected

218

from 2010 to 2014.48 The uncertainty of these statistical data is not quantified by their providers

219

but it is assumed to be considerably high. These data are compared to the simulated stocks and

220

outflows to collection.

ACS Paragon Plus Environment 12

Page 13 of 28

Environmental Science & Technology

221 222

Detailed model description The conceptual model, as described in Figure 1, was implemented using our own MFA simula-

223

tion tool called 'pymfa' (written in Python 3, using the numpy, scipy, and matplotlib libraries),

224

which supports the calculation of dynamic MFAs in a flexible way. The core of the tool can be

225

used as a Python library and provides the necessary functionality to run analyses through a

226

command line interface and an interactive web application that can also be run locally. Model

227

descriptions are specified using a simple data-driven approach: the researcher creates a spread-

228

sheet where each row represents a link in the MFA. Links can represent inflows, unit conver-

229

sions, transfer rates, delays or the split of a fraction entering a process into multiple fractions

230

leaving the process. For each link, time series of data including information on the probabilistic

231

distribution have to be provided. As such, model descriptions can simply be stored and forward-

232

ed as spreadsheets or CSV files. Model description source files can be uploaded through the web

233

interface, upon which the tool calculates the time series of resulting stocks and flows by means

234

of Monte Carlo simulation, visualizes them and offers the resulting data as a CSV download. A

235

screenshot of the web interface as well as an example of a model description source file and out-

236

put plots can be found in the SI.

237

For each device type, the model was simulated for numbers of devices and for physical mass,

238

producing results in pieces and in kg, respectively. In addition to the standard scenario, we run a

239

lower and upper lifetime scenario and an impulse scenario for each device type. Each simulation

240

run was repeated 10000 times, which we considered a sufficient sample size for the Monte Carlo

241

simulation

ACS Paragon Plus Environment 13

Environmental Science & Technology

Page 14 of 28

242

Results and discussion

243 244

Total lifetime

245

to the median service lifetime of new devices, the median total lifetime, derived from the im-

246

pulse response analysis, increases, for example, from 3 to 7 years for conventional mobile

247

phones and smartphones, from 5 to 8 and 9 years for desktops and laptops, respectively and from

248

9 to 11 years for CRT TVs. Existing studies from the Netherlands53 and Switzerland54 show

249

similar results regarding the average or median total lifetime. The median and average total ser-

250

vice lifetimes as well as a comparison to existing studies are listed in Table S6 in the SI.

Reuse and storage significantly extend the time a device remains in the use phase. Compared

251

The total lifetime distribution functions, compared to the service lifetime distribution functions

252

of new devices, show a shifted mode towards longer lifetimes, and a wider and more positively

253

skewed distribution. An example of four device types is illustrated in Figure 2 (for the remaining

254

distributions see Figure S24 in the SI). Only few sources present Weibull parameter for the total

255

lifetime of a large range of EE17,53.As we assume that Dutch usage patterns are similar to those of

256

the Swiss, we compare our results to Wang et al.53 Their Weibull distribution functions for the

257

year 2005 match well our findings for most device types. Exceptions are laptop computers, with

258

a Weibull distribution function more similar to our service lifetime distribution, and mobile

259

phones, with a sharply and monotonically decreasing Weibull distribution function. The blue and

260

the red curve for mobile phones have similar means (9.6 and 8.5 years, respectively) but different

261

medians (4.5 and 7.0 years, respectively). As both laptops and mobile phones are often stored

262

and reused in Switzerland, our findings of wider and more positively skewed distributions seem

263

to be more realistic.

ACS Paragon Plus Environment 14

Environmental Science & Technology

CRT TV SL

0.2

TL TL Wang et al.

0.1 0

Relative Frequency [-]

0

10

20

Laptop computer

0.3

0.2

TL Wang et al.

0 0 0.3

0.1

0.1

10 20 30 Lifetime [years]

TL

0.1

SL 0.2 TL TL Wang et al.

0

b)

SL

30

c)

Desktop computer

0.3

0.2

0

264

a)

Relative Frequency [-]

0.3

Relative Frequency [-]

Relative Frequency [-]

Page 15 of 28

10 20 30 Mobile/Smartphone d) SL TL TL Wang et al.

0 0

10 20 30 Lifetime [years]

265

Figure 2: Comparison of own results of service lifetime (SL) with total lifetime (TL) and TL of

266

Wang et al.,53 for a) cathode ray tube television (CRT TV), b) desktop computer, c) laptop com-

267

puter and d) mobile phone/smartphone.

268

The results show that data on mean or median lifetimes do not provide conclusive information

269

on the actual lifetime distribution function. They further highlight the importance of including

270

flexible lifetime distribution functions in dynamic MFA models of products that are reused and

271

stored. Assuming average, median or normally distributed lifetimes may, in case of technology

272

change, often result in underestimated phase-out times. Whether lifetime distributions functions

273

are favorably divided into different service lifetime and storage time distribution functions or

274

merged to total lifetime distribution functions depends on the question to be answered and the

275

required level of detail.

276 277

Stocks and flows

278

3 for nine device types between 1980 and 2014.

The cascade model produces a cumulated in-use stock and storage stock as illustrated in Figure

ACS Paragon Plus Environment 15

10000

0

5000

Year 250

c)

200

150

100

50

2014

2012

2010

2008

2006

d)

200

150

100

50

2012 2014

2008 2010

2002 2004 2006

1998 2000

1992 1994 1996

1986 1988 1990

1982 1984

1980

2012 2014

2010

2008

2004 2006

2002

2000

1996 1998

1994

1990 1992

1988

1986

1982 1984

0 1980

0

Year

Year

279

2004

Year Cumulated storage stock [1'000 toones]

Cumulated in-use stock [1'000 tonnes]

250

2002

1980

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

0 2000

5000

15000

1998

10000

20000

1996

15000

25000

1994

20000

30000

1992

25000

35000

1990

30000

40000

1988

35000

Page 16 of 28

b)

45000

1986

40000

50000

1984

a)

45000

1982

50000

Cumulated storage stock [1'000 pieces]

Cumulated in-use stock [1'000 pieces]

Environmental Science & Technology

CRT TV

FPD TV

Mobile phone

Smartphone

Desktop

Laptop

Headset

DVD player

10th percentile

90th percentile

Monitor

280

Figure 3: a) Cumulated in-use stock in 1000 pieces, b) cumulated storage stock in 1000 pieces,

281

c) cumulated in-use stock in 1000 tonnes and d) cumulated storage stock in 1000 tonnes, for nine

282

device types between 1980 and 2014

283

The cumulated in-use stock has grown in the past 20 years to around 44 million devices or

284

220 000 tonnes in 2014. In terms of devices, the 10th and 90th percentiles, resulting from the

285

Monte Carlo simulation, account for ±10% or a range of roughly 9 million devices. As the mass

286

of each device type increases uncertainty, the 10th and 90th percentile for the in-use stock in

287

tonnes accounts for ±12%. The storage stock comprises 30 million ±8% devices or 75 000 ±9%

288

tonnes in 2014. The storage stock thus accounts for 40% of the total stock in terms of number of

ACS Paragon Plus Environment 16

Page 17 of 28

Environmental Science & Technology

289

devices and 25% in terms of mass. Both in-use stock and storage stock are dominated by small

290

devices such as mobile phones, smartphones and headsets. Regarding mass, however, these de-

291

vices types are insignificant and CRT as well as FPD TVs and desktop computers are prevalent.

292

The in-use stock of CRT TVs and conventional mobile phones is declining as these device types

293

are replaced by FPD TVs and smartphones, respectively. The stock of smartphones is increasing

294

with a growth rate of 20 – 30%, while the in-use stocks of most other device types are slowly ap-

295

proaching or have already reached saturation. Thus, the overall growth rate of the in-use stock

296

and storage stock in pieces as well as the storage stock's mass have been declining in the past 10

297

years. Mostly due to the replacement of CRT TVs with lighter FPD TVs, the in-use stock's mass

298

is declining.

299

The simulated stocks and flows to, from and within the cascade model are depicted in Figure 4

300

for mobile phones and FPD TVs in 2014. For the sake of clarity, we refrain from including un-

301

certainty ranges here. The Sankey diagrams for the remaining seven device types can be found in

302

the SI (Figures S25-S31). The disaggregation of the total stock to the in-use stock and storage

303

stock of new devices (cascade step 1), second-hand devices (cascade step 2) and all the rest (cas-

304

cade step 3) shows large discrepancies among device types. Mobile phones, CRT TVs and head-

305

sets have a total storage stock that is larger than or close to the size of the total in-use stock. The

306

smallest total storage stock, compared to the total in-use stock, have FPD TVs and smartphones.

307

The in-use stock 2 of DVD players, CRT TVs and mobile phones accounts for over 15% of the

308

total stock in 2014, while for smartphones and headsets, it is below 5%. Various trends can be

309

derived from these findings: device types that are phased out, such as CRT TVs, mobile phones

310

but also DVD players (replaced by streaming services) are more frequently stored or used as se-

311

cond-hand devices. New device types such as smartphones or FPD TVs are still located mostly

ACS Paragon Plus Environment 17

Environmental Science & Technology

Page 18 of 28

312

in the in-use stock 1. Furthermore, small devices such as mobile phones and headsets are more

313

frequently stored than large devices.27 The storage stock 2 compared to the total stock are largest

314

for monitors, mobile phones and desktops. The decision to store a second-hand device therefore

315

does not seem to be influenced by the size of the device, but more by the prospective utility. The

316

small in-use stock 3 and storage stock 3 show that further uses occur, but are irrelevant.

317

Flows to collection are largest for all device types. Flows to municipal waste incineration

318

(MWI) are only relevant for headsets. With 20-30% of the total outflow, the share of FPD TVs,

319

mobile phones and smartphones entering unknown disposal pathways is high. Unknown flows

320

include, for example, lost or stolen devices and devices disposed of by others than the survey re-

321

spondents. It is thus possible that material reaching unknown disposal pathways still ends up in

322

the collection system. Export flows are highest for laptops and mobile phones. Total outflows

323

compared to sales flows are by a factor 3 to 7 higher for mobile phones and DVD players. For

324

desktops, monitors and headsets, sales and outflows are of the same magnitude. For laptops, FPD

325

TVs and smartphones, sales are still larger by a factor of 1.2, 2 and 5, respectively. Despite long

326

delays due to reuse and storage, this comparison thus reveals for which device types the system

327

is currently saturated. These delays, however, result in still growing outflows, with an average

328

growth rate of 5% over the last 20 years. Figure S32 in the SI shows the cumulated flows to col-

329

lection between 1980 and 2014. In total, the nine device types result in 24 000 tonnes ±10% that

330

reached the collection system in 2014, dominated by CRT TVs, FPD TVs and desktop comput-

331

ers.

332

From a resource point of view, it is not important how long products are delayed in the use

333

phase, as long as they reach the collection system in the end, which is the case for 80% of all de-

334

vices on average. However, if we would aim for more reuse or refurbishment, it would be im-

ACS Paragon Plus Environment 18

Page 19 of 28

Environmental Science & Technology

335

portant that devices are not stored for too long, as newer devices are more in demand on the se-

336

cond-hand market.

a)

337

b)

338 339

Figure 4: Stocks and flows of a) mobile phones and b) flat panel display televisions (FPD TVs)

340

in 2014 in tonnes. Mobile phone flows below 5 t and FPD TV flows below 100 t are not labeled.

341 342

Model sensitivity and validation

343

most device types (Figure 3). If the mass per product with its associated uncertainty is included

The 10th and 90th percentiles of the in-use stock indicate an uncertainty of around ±10% for

ACS Paragon Plus Environment 19

Environmental Science & Technology

Page 20 of 28

344

in the model, the uncertainty range increases up to ±12%. The lower and upper lifetime distribu-

345

tion scenarios of the sensitivity analysis add a further ±10% deviation from the standard lifetime

346

distribution scenario. Considering the 10th and 90th percentiles of the lower and upper scenario,

347

the maximum deviation from the mean in-use stock of the standard scenario amounts to approx-

348

imately ±30% (Figure 5). The storage stock is less sensitive with ca. ±20% maximum deviation.

349

The sensitivity of the flows is presented using the collection flows as an important example.

350

As longer service lifetimes and storage times lead to smaller collection flows, the lower scenario

351

accounts for the highest flows and vice versa. The deviations between the means of the standard,

352

lower and upper lifetime distribution scenario are similar to the deviations of the 10th and 90th

353

percentiles of the standard scenario (ca. ±10%). The maximum deviations between the mean of

354

the standard scenario and the 10th and 90th percentiles of the lower and upper scenario account

355

for ±20%.

ACS Paragon Plus Environment 20

Page 21 of 28

Environmental Science & Technology

Total collection flow [1000 pieces/year]

Total in-use stock [1000 pieces]

60'000

356

a)

50'000 40'000 30'000 20'000 10'000 0 4'000 1998 2000 2002 2004 2006 2008 2010 2012 2014

b)

3'000

2'000

1'000

0 1998 2000 2002 2004 2006 2008 2010 2012 2014 Year Standard Standard 90th percentile Lower 10th percentile Upper mean

Standard 10th percentile Lower mean Lower 90th percentile Upper 10th percentile

Upper 90th percentile

Data FSO/OFCOM/Swico

357

Figure 5: a) Total in-use stock in 1000 pieces and b) total collection flow in 1000 pieces/year

358

based on the standard, lower and upper lifetime distribution scenarios for desktops, laptops, mo-

359

bile phones, CRT TVs, FPD TVs and DVD players (in-use stock) and FPD monitors, desktops,

360

laptops, mobile phones, CRT TVs and FPD TVs (collection flow), compared to statistical data

361

available from FSO, OFCOM and Swico Recycling. 48,51,52

362

By comparing the total in-use stock and the total collection flow with available data from FSO

363

and Swico Recycling, we evaluated to what extent the cascade model is able to reproduce the

ACS Paragon Plus Environment 21

Environmental Science & Technology

Page 22 of 28

364

behavior of the real system. As can be seen in Figure 5, the model seems to overestimate the total

365

in-use stock. The sum of FSO and OFCOM data lies in the vicinity of the 10th percentile of our

366

standard scenario and between the mean and the 90th percentile of our lower scenario. A poten-

367

tial explanation is that the service lifetimes and storage times we assumed in the model are long-

368

er than those in the real system. However, the comparison of collection flows shows that fewer

369

devices are collected compared to the modeled collection flow, which could be, in contrast, an

370

argument for longer service lifetimes and storage times. Again, the sum of devices collected by

371

Swico Recycling lies in the vicinity of the 10th percentile of our standard scenario, but now also

372

between the mean and the 90th percentile of our upper scenario. Breaking this down to single de-

373

vice types, the discrepancies between modelled and reported collection flows are largest for mo-

374

bile phones, with 60% less collected than what the model calculates, even though we included

375

long reuse and storage times. As our model for mobile phones overestimates the in-use stock by

376

about 20%, we argue against extending the service lifetime in our model. Whether we assumed

377

too few mobile phones being stored for a too short time, or a large amount of them is exported,

378

could not be investigated within this study. Lange19 or Huisman et al.18, among others, report

379

transboundary shipment of EE to Eastern Europe or Africa from Germany and the Netherlands.

380

Although Switzerland has higher EE collection rates than most EU Member States, it might still

381

be possible for small devices such as mobile phones to be formally or informally exported with-

382

out our knowledge. It is well known that mobile phones are the most difficult devices for collec-

383

tion schemes to get hold of, with already various studies investigating potential causes.14,55,56

384

Taking into account the large model uncertainty of up to ±30%, the cascade model is able to

385

map reasonably well the total in-use stock and the total collection flows. This shows that the

386

mixture of detailed bottom-up information on service lifetimes, storage times and transfer coeffi-

ACS Paragon Plus Environment 22

Page 23 of 28

Environmental Science & Technology

387

cients combined with a top-down approach to calculate stocks and flows is an adequate approach

388

to enhance dynamic MFAs. The cascade model provides important details regarding reuse and

389

storage stocks and flows that have been often neglected in dynamic MFA studies. It further un-

390

veils different behavior of device types within the use phase, for example with regard to technol-

391

ogy changes or phase out periods.

392

The model structure can be applied to any other product that is potentially reused and stored

393

after its first service life. However, in many cases, data required for a cascade model will be dif-

394

ficult to collect. Furthermore, from a resource point of view, it is not important to know exactly

395

the status of products in the use phase, as long as it can be safely assumed that they will reach the

396

collection system in the end. The use phase could thus be modeled as a 'black box' as in earlier

397

studies, but by adapting the total lifetime distribution function, eventual reuse and storage should

398

be taken into account. Such a simplified model can still be useful to simulate outflows, their

399

composition (mix of devices and materials) and the share reaching collection more precisely.

400 401

Associated Content

402

functions, transfer coefficients and the uncertainty distributions of mass per device type. It fur-

403

ther includes more information on the software tool used to implement the model, run the simu-

404

lations and visualize the output. Additional results on the total lifetime distributions as well as

405

stocks and flows by device type are provided as well. This material is available free of charge at

406

http://pubs.acs.org.

407 408

Author Information

409

*Corresponding author phone: +41 58 765 78 44; e-mail: esther.thié[email protected]

The Supporting Information provides more detail about the inflow data, Weibull distribution

Corresponding Author

ACS Paragon Plus Environment 23

Environmental Science & Technology

Page 24 of 28

410

Notes

411

The authors declare no competing financial interest.

412 413

Acknowledgments

414

designing, implementing and refining the software tool. We further thank the two anonymous re-

415

viewers for their valuable input. The project was funded by Empa.

416

References

417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442

(1)

We thank Carol Alexandru, Rolf Badat and Dimitri Kohler from the University of Zurich for

(2)

(3)

(4) (5) (6)

(7) (8) (9)

(10)

Gerst, M. D.; Graedel, T. E. In-Use Stocks of Metals: Status and Implications. Environ. Sci. Technol. 2008, 42 (19), 7038–7045. Graedel, T.E., Allwood, J., Birat, J.-P., Buchert, M., Hagelüken, C., Reck, B.K., Sibley, S.F., Sonnemann, G. What do we know about metal recycling rates? J. Ind. Ecol. 2011, 15 (3), 355–366. UNEP. Metal Recycling: Opportunities, Limits, Infrastructure,; A Report of the Working Group on the Global Metal Flows to the International Resource Panel. Reuter, M. A.; Hudson, C.; van Schaik, A.; Heiskanen, K.; Meskers, C.; Hagelüken, C., 2013. van Schaik, A.; Reuter, M. A. Dynamic modelling of E-waste recycling system performance based on product design. Miner. Eng. 2010, 23 (3), 192–210. Reuter, M. A. Limits of design for recycling and “sustainability”: A review. Waste Biomass Valorization 2011, 2 (2), 183–208. European Commission. Critical Raw Materials for the EU. Report of the Ad-hoc Working Group on defining critical raw materials; European Commission: Brussels, Belgium, 2014. Parajuly, K.; Habib, K.; Liu, G. Waste electrical and electronic equipment (WEEE) in Denmark: Flows, quantities and management. Resour. Conserv. Recycl. 2016. Zeng, X.; Gong, R.; Chen, W.-Q.; Li, J. Uncovering the Recycling Potential of “New” WEEE in China. Environ. Sci. Technol. 2016, 50 (3), 1347–1358. Ueberschaar, M.; Rotter, V. S. Enabling the recycling of rare earth elements through product design and trend analyses of hard disk drives. J. Mater. Cycles Waste Manag. 2015, 17 (2), 266–281. Kalmykova, Y.; Patrício, J.; Rosado, L.; Berg, P. E. Out with the old, out with the new – The effect of transitions in TVs and monitors technology on consumption and WEEE generation in Sweden 1996–2014. Waste Manag. 2015, 46, 511–522.

ACS Paragon Plus Environment 24

Page 25 of 28

443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481

Environmental Science & Technology

(11)

(12)

(13) (14) (15)

(16) (17) (18)

(19)

(20)

(21)

(22) (23)

(24)

Duygan, M.; Meylan, G. Strategic management of WEEE in Switzerland— combining material flow analysis with structural analysis. Resour. Conserv. Recycl. 2015, 103, 98–109. Lau, W. K.; Chung, S.-S.; Zhang, C. A material flow analysis on current electrical and electronic waste disposal from Hong Kong households. Spec. Themat. Issue Urban Min. Urban Min. 2013, 33 (3), 714–721. Yoshimura, A.; Daigo, I.; Matsuno, Y. Global substance flow analysis of indium. Mater. Trans. 2013, 54 (1), 102–109. Polák, M.; Drápalová, L. Estimation of end of life mobile phones generation: The case study of the Czech Republic. Waste Manag. 2012, 32 (8), 1583–1591. Steubing, B.; Böni, H.; Schluep, M.; Silva, U.; Ludwig, C. Assessing computer waste generation in Chile using material flow analysis. Waste Manag. 2010, 30, 473– 482. Yoshida, A.; Tasaki, T.; Terazono, A. Material flow analysis of used personal computers in Japan. Waste Manag. 2009, 29 (5), 1602–1614. Oguchi, M.; Kameya, T.; Yagi, S.; Urano, K. Product flow analysis of various consumer durables in Japan. Resour. Conserv. Recycl. 2008, 52 (3), 463–480. Huisman, J.; Botezatu, I.; Herreras, J.; Liddane, M.; Hintsa, J.; Luda di Cortemiglia, V.; Leroy, P.; Vermeersch, E.; Mohanty, S.; van den Brink, S.; et al. Countering WEEE Illegal Trade (CWIT) Summary Report, Market Assessment, Legal Analysis, Crime Analysis and Recommendations Roadmap; Lyon, 2015. Lange, U. Evaluation of informal sector activities in Germany under consideration of electrical and electronic waste management systems. Dissertation, Technische Universität Dresden: Dresden, 2013. Sinha-Khetriwal, D. Diffusion, Obsolescence and Disposal of Consumer Durables: Models for Forecasting Waste Flows of End-of-Life Consumer Durables. Dissertation, University of St.Gallen HSG: St. Gallen, 2012. Sander, K.; Schilling, S. Transboundary shipment of second-hand Electric and Electronical Equipment/e-waste - Analysis and proposal for optimising of material flows; OEKOPOL – Institute for Environmental Strategies: Hamburg, Germany, 2010; p 116. Chung, S.; Lau, K.; Zhang, C. Generation of and control measures for, e-waste in Hong Kong. Waste Manag. 2011, 31 (3), 544–554. Milovantseva, N.; Saphores, J.-D. Time bomb or hidden treasure? Characteristics of junk TVs and of the US households who store them. Waste Manag. 2013, 33 (3), 519–529. Saphores, J.-D. M.; Nixon, H.; Ogunseitan, O. A.; Shapiro, A. A. How much e-waste is there in US basements and attics? Results from a national survey. J. Environ. Manage. 2009, 90 (11), 3322–3331.

ACS Paragon Plus Environment 25

Environmental Science & Technology

482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519

(25)

(26)

(27)

(28) (29)

(30)

(31) (32)

(33)

(34)

(35)

(36)

Page 26 of 28

Sabbaghi, M.; Esmaeilian, B.; Raihanian Mashhadi, A.; Behrendt, S.; Cade, W. An investigation of used electronics return flows: A data-driven approach to capture and predict consumers storage and utilization behavior. Waste Manag. 2015, 36, 305–315. Williams, E.; Hatanaka, T. Residential Computer Usage Patterns, Reuse and Life Cycle Energy Consumption in Japan. In Panels of the 2005 ACEEE Summer Study on Energy Efficiency in Industry; European Council for an Energy Efficient Economy: Stockholm, 2005; Vol. 4, pp 146–157. Thiébaud (-Müller), E.; Hilty, L. M.; Schluep, M.; Widmer, R.; Faulstich, M. Service lifetime, storage time and disposal pathways of electronic equipment: a Swiss case study. J. Ind. Ecol. 2017, n/a-n/a. Bornhöft, N. A.; Sun, T. Y.; Hilty, L. M.; Nowack, B. A dynamic probabilistic material flow modeling method. Environ. Model. Softw. 2016, 76, 69–80. Müller, E.; Hilty, L. M.; Widmer, R.; Schluep, M.; Faulstich, M. Modeling Metal Stocks and Flows: A Review of Dynamic Material Flow Analysis Methods. Environ. Sci. Technol. 2014, 48 (4), 2102–2113. Böni, H.; Wäger, P.; Thiébaud, E.; Du, X.; Figi, R.; Nagel, O.; Bunge, R.; Stäubli, A.; Spörry, A.; Wolfensberger-Malo, M.; et al. Projekt e-Recmet - Rückgewinnung von kritischen Metallen aus Elektronikschrott am Beispiel von Indium und Neodym. Schlussbericht [Project e-Recmet – Recovery of critical metals from e-waste, the example of indium and neodymium. Final report]; Empa: St. Gallen, 2015. Sprecher, B.; Kleijn, R.; Kramer, G. J. Recycling Potential of Neodymium: The Case of Computer Hard Disk Drives. Environ. Sci. Technol. 2014, 48 (16), 9506–9513. Wäger, P.; Widmer, R.; Restrepo, E.; Fernandes, M. Wäger, P., Widmer, R. Restrepo, E. und Fernandes, M. (2014) Seltene Metalle in Fraktionen aus der maschinellen Aufbereitung von WEEE. Bern: Bundesamt für Umwelt.; Federal Office for the Environment: Bern, Switzerland, 2014. Blaser, F.; Castelanelli, S.; Wäger, P.; Widmer, R. Seltene Metalle in Elektro- und Elektronikaltgeräten - Vorkommen und Rückgewinnungstechnologien; Bundesamt für Umwelt: Bern, 2011. Chancerel, P.; Meskers, C.; Hagelüken, C.; Rotter, V. S. Assessment of Precious Metal Flows During Preprocessing of Waste Electrical and Electronic Equipment. J. Ind. Ecol. 2009, 13 (5), 791–810. Licht, C.; Peiró, L. T.; Villalba, G. Global Substance Flow Analysis of Gallium, Germanium, and Indium: Quantification of Extraction, Uses, and Dissipative Losses within their Anthropogenic Cycles. J. Ind. Ecol. 2015, 19 (5), 890–903. Westphal, L.; Kuchta, K. Permanent Magnets from Small Waste Electrical and Electronic Equipment. In Proceedings Sardinia 2013; Sardinia, Italy, 2013.

ACS Paragon Plus Environment 26

Page 27 of 28

520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559

Environmental Science & Technology

(37)

(38)

(39) (40) (41) (42) (43) (44)

(45) (46) (47)

(48) (49)

(50)

(51)

(52)

Habib, K.; Schibye, P. K.; Vestbø, A. P.; Dall, O.; Wenzel, H. Material Flow Analysis of NdFeB Magnets for Denmark: A Comprehensive Waste Flow Sampling and Analysis Approach. Environ. Sci. Technol. 2014, 48 (20), 12229–12237. Gottschalk, F.; Scholz, R. W.; Nowack, B. Probabilistic material flow modeling for assessing the environmental exposure to compounds: Methodology and an application to engineered nano-TiO2 particles. Environ. Model. Softw. 2010, 25 (3), 320–332. Weiss, R. Weissbuch 2015; Männedorf, 2015. SCEA. Verkäufe von Unterhaltungselektronik in der Schweiz; Swiss Consumer Electronics Association: Zürich, 2005. GfK Switzerland. Verkäufe von Informations- und Kommunikationstechnologien und Unterhaltungselektronik in der Schweiz; GfK Switzerland AG: Hergiswil, 2015. GfK Retail and Technology. Telecom, Total Market Estimations Switzerland, 2008. GfK Switzerland. Personal communication with Senior Market Consultant, 2016. Dahlström, K.; Ekins, P.; He, J.; Davis, J.; Clift, R. Iron, Steel and Aluminium in the UK: Material Flows and Their Economic Dimensions, Final Project Report, March 2004; CES Working Paper 03/04; Centre for Environmental Strategy: University of Surrey, Surrey, 2004. Melo, M. T. Statistical analysis of metal scrap generation: the case of aluminium in Germany. Resour. Conserv. Recycl. 1999, 26 (2), 91–113. Labdoo. Labdoo.org. Tag a laptop, spread education around the world https://www.labdoo.org/ (accessed Oct 26, 2016). Swisscom AG. Donate to Swisscom Mobile Aid https://www.swisscom.ch/en/about/company/sustainability/mobile-aid.html (accessed Oct 26, 2016). SWICO. Swico Tätigkeitsberichte 2006-2015 (Swico Activity Reports 2006-2015); SWICO Recycling: Zürich, 2016. Chancerel, P. Substance flow analysis of the recycling of small waste electrical and electronic equipment. An assessment of the recovery of gold and palladium, Technische Universität Berlin, Institut für Technischen Umweltschutz, ITUSchriftenreihe, 2010: Berlin, 2010. JCGM. JCGM 100:2008. Evaluation of measurement data — Guide to the expression of uncertainty in measurement (GUM); Joint Committee for Guides in Metrology, 2008. FSO. Haushalte und Bevölkerung - IKT-Ausstattung (Households and Population ICT equipment) http://www.bfs.admin.ch/bfs/portal/de/index/themen/16/04/key/approche_globa le.indicator.30103.2.html (accessed Sep 29, 2016). OFCOM. Number of mobile telephony customers https://www.bakom.admin.ch/bakom/en/homepage/telecommunication/facts-

ACS Paragon Plus Environment 27

Environmental Science & Technology

560 561 562 563 564 565 566 567 568 569 570 571 572 573 574

(53)

(54)

(55)

(56)

Page 28 of 28

and-figures/statistical-observatory/mobile/number-of-mobile-telephonycustomers.html (accessed Oct 27, 2016). Wang, F.; Huisman, J.; Stevels, A.; Baldé, C. P. Enhancing e-waste estimates: Improving data quality by multivariate Input-Output Analysis. Waste Manag. 2013, 33 (11), 2397–2407. Stocker, R.; Bleiker, S.; Allibone, A.; Albiez, T.; Kurer, A.; Streicher, M. Stocks und Lifespans elektronischer Geräte [Stock and lifespans of electronic devices]; Fachhochschule Nordwestschweiz, Hochschule für Technik: Brugg, Switzerland, 2013. Ylä-Mella, J.; Keiski, R. L.; Pongrácz, E. Electronic waste recovery in Finland: Consumers’ perceptions towards recycling and re-use of mobile phones. Waste Manag. 2015, 45, 374–384. Yin, J.; Gao, Y.; Xu, H. Survey and analysis of consumers’ behaviour of waste mobile phone recycling in China. J. Clean. Prod. 2014, 65, 517–525.

ACS Paragon Plus Environment 28