Comparative Analysis of Supply Risk-Mitigation Strategies for Critical

Impacts of Combined Cooling, Heating and Power Systems, and Rainwater Harvesting on Water Demand, Carbon Dioxide, and NOx Emissions for Atlanta...
1 downloads 0 Views 2MB Size
Subscriber access provided by READING UNIV

Policy Analysis

Comparative Analysis of Supply Risk Mitigation Strategies for Critical Byproduct Minerals – A Case Study of Tellurium Michele Bustamante, Gabrielle Gaustad, and Elisa Alonso Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b03963 • Publication Date (Web): 07 Nov 2017 Downloaded from http://pubs.acs.org on November 12, 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

Comparative Analysis of Supply Risk Mitigation Strategies for Critical Byproduct

2

Minerals – A Case Study of Tellurium

3

Michele L. Bustamantea, Gabrielle Gaustadb,*, Elisa Alonsoc

4

Dr. Michele Bustamante, Post-doctoral Associate aMaterials Systems Lab, Massachusetts

5

Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139 USA

6

Dr. Gabrielle Gaustad, Associate Professor bGolisano Institute for Sustainability, 190 Lomb

7

Memorial Drive, Rochester Institute of Technology, Rochester, NY 14623 USA

8

Dr. Elisa Alonso, Consultant cElisa Alonso LLC, Towson, MD 21022 USA

9

*Corresponding Author: [email protected]

10

Abstract

11

Materials criticality assessment is an analytical framework increasingly applied to identify

12

materials of importance to stakeholders who face scarcity risks. Although criticality assessment

13

studies highlight materials for the implicit purpose of informing future action, the aggregated

14

nature of the studies’ results make them poorly suited for use as guidance in nuanced strategy

15

development and implementation. As a first step in the selection of mitigation strategies, the

16

present work proposes a modeling framework and accompanying set of metrics to directly

17

compare strategies by measuring effectiveness of risk reduction as a function of features of

18

projected supply-demand balance over time. The work focuses on byproduct materials whose

19

criticality is particularly important to understand because their supplies are less responsive to

20

market balancing forces, i.e. price feedbacks. Tellurium, a byproduct of copper refining, critical

21

to solar photovoltaics, is chosen as a case study and three commonly discussed byproduct-

22

relevant strategies are selected: dematerialization of end-use product, byproduct yield

23

improvement, and end-of-life recycling rate improvement. Results suggest that dematerialization

24

will be nearly twice as effective at reducing supply risk as the next best option, yield

25

improvement. Finally, due to its infrequent current use and dependence on long product

26

lifespans, recycling end-of-life products is expected to be the least effective option, despite

27

offering potentially other benefits (e.g. cost savings, environmental impact reduction).

28 29

1.0 Introduction

30

1.1 Critical Materials & Criticality

1 ACS Paragon Plus Environment

Environmental Science & Technology

Page 2 of 28

31

Meeting the needs of a growing global population with limited shared resources is one of

32

the most fundamental challenges of sustainability. Especially in cases where those resources are

33

non-renewable, such as fossil fuels, metals, and other minerals, ensuring their continued supply

34

at reasonable costs in the future may require intentional management of consumption over time.

35

This is because rapid increase in demand, such as those resulting from technological paradigm

36

shifts, or unexpected restrictions in supply, for example due to regional conflicts, natural

37

disasters, or regulatory action, can lead to extreme price spikes and market disruptions, all of

38

which can have damaging impacts on affected stakeholders. A number of historical examples

39

have been studied showing the challenges that can be created for manufacturing firms and the

40

national security concerns that can be raised when such conditions occur1–3.

41

In order to better predict and prevent similar future supply shortages, several efforts have

42

been made at the company4,5, regional6,7, and international8 levels to identify materials that are

43

particularly vulnerable. These materials, often referred to as critical or strategic materials, are

44

characterized as such because they are essential to key economic sectors but also have significant

45

scarcity concerns. Metrics used in these studies include a wide variety of factors, such as limited

46

natural abundance, international trade restrictions, recycling rates, national import reliance and

47

per capita consumption. Commonly recognized critical materials include the rare earth

48

elements6,9, particularly neodymium and dysprosium10,11, platinum group metals12,13, as well as

49

indium and tellurium, particularly in the context of solar clean energy technologies9,14.

50 51

1.2 Byproduct Mining: Obstacles to Market Balancing Feedbacks

52

Conventional economic wisdom dictates that resource scarcity can be managed

53

independently by inherent feedback mechanisms of the free market. That is, increased scarcity

54

drives up price, which in turn may encourage use of less expensive substitutes by consumers,

55

lowering demand. Simultaneously, higher prices also incentivize exploitation of previously sub-

56

economic resources by producers, therefore reducing scarcity through increased supply. This

57

mechanism is generally quite reliable for major metal markets, such as copper, iron, and

58

aluminum, where new mines are prepared to open because they form relatively abundant ore

59

deposits and are produced independently or as the main product of mining operations. However,

60

the same cannot always be assumed for more minor metals, particularly those produced via

61

byproduct mining15. 2 ACS Paragon Plus Environment

Page 3 of 28

62

Environmental Science & Technology

Byproduct mining is a general term for the supply of metals and minerals recovered

63

predominantly as minor coproducts of major metal mining and refining operations. Byproduct

64

minerals, like tellurium, selenium, indium, and gallium, are not found in adequately high

65

concentrations to generate sufficient revenue to cover their full mining costs. Rather, their

66

separation occurs as a natural consequence of the main product’s processing into a purer form.

67

They do not command independent costs to mine from ore (only to refine further, which is

68

typically performed by a separate agent) and instead are treated as a credit to mining costs,

69

recovered in the quantities yielded naturally for sale to refiners. Despite the relatively limited

70

attention paid to byproduct supply risk in the existing criticality literature, byproduct supply is

71

quite common; one quarter of all naturally occurring elements are supplied over 50% from

72

byproduct mining16,17.

73

Because the quantity of byproducts are typically small compared to the main product,

74

even large price increases in the byproduct market are seldom sufficient to justify the additional

75

costs of processing more ore. Further, these materials are supplied via byproduct mining to

76

begin with because they largely lack viable or economic alternatives for direct mining. This

77

means that strong market signals, even sustained for a few years, may not be able to affect

78

change from a supply perspective in the same way for byproduct metals as for major metals. As a

79

result, it may become necessary to intervene or support changes in these markets to mitigate risk

80

for highly critical byproduct materials if their continued use is considered desirable.

81

Unfortunately, criticality assessment studies, while effective at communicating relative risk for

82

different materials, are poorly suited to inform decision making regarding mitigation strategy

83

selection to apply to those materials markets.

84 85

1.3 Limitations to Criticality Assessment: Need for Supplemental Analyses to Respond

86

Strategically

87

Review of the criticality assessment literature reveals that results are often presented in a

88

highly aggregated fashion. While this format is quite effective as a method of simple

89

comparison among many materials, they provide little direct insight about how to resolve the

90

risk being identified6,8,10,11. In many ways, this is a practical limitation to the scope of criticality

91

studies, which already require significant time and effort to collect data for comparison across

92

many materials; data that are often uncertain18. This limitation however can become 3 ACS Paragon Plus Environment

Environmental Science & Technology

Page 4 of 28

93

problematic when there is no guidance for actors looking to interpret these results into

94

meaningful action for reducing risk.

95

Different agents, including policy makers and firm-level strategists, are faced with

96

virtually endless potential approaches to resolve identified risk; some of these strategies are

97

listed in Table 1 organized by key leverage area. Currently decision makers can only look to the

98

risk indicator scores in published criticality studies, if available, and then try to compare them to

99

one another to determine which lever to pull. However, if supply risk for a particular material is

100

scored very highly because recycling rate is zero, byproduct portion of supply is 100%, and

101

supplier geographical concentration is high, what should be done? Should decision-makers

102

develop recycling technology or infrastructure? Should they reduce import reliance by levying

103

tariffs to allow domestic suppliers to compete? Should they stop using the material altogether

104

and develop novel substitutes? The best response is not immediately clear from the raw scores

105

alone because it is not possible to directly compare the three disparate sources of risk driving this

106

criticality. Each approach will have its own costs and benefits, so further analysis is needed to

107

quantify these tradeoffs in pursuit of equitable comparison.

108 109

Table 1. Critical material risk mitigation strategies (non-comprehensive)19–21 Primary Supply

Secondary Supply

Demand Efficiency

General

trade, diplomacy

regulating collection and

government purchase

fundamental R&D

use of waste

quotas

development financing

import/export policies for

R&D for technologies and

government information

for mines

waste

innovation

sharing (e.g. USGS)

R&D in mining and

funding recycling

substitution

transparency regulations

extraction technology

infrastructure

(e.g. increase yield)

development

stockpile purchase

incentives to secondary

guarantees

processors

subsidize domestic

restrict landfilling

for firms in supply chain

dematerialization

reporting standards for trade (HS codes)

product lifetime extension

international collaboration

mining

110 111

Additionally, the byproduct nature of critical materials is not often considered when

112

evaluating mitigation response feasibility. For example, the potential effectiveness of

113

substitution is often overstated for byproducts because proposed substitutes are often also 4 ACS Paragon Plus Environment

Page 5 of 28

Environmental Science & Technology

114

byproducts or coproducts of the same restricted material system. Examples of this cited in

115

previous literature include fellow copper byproduct selenium as a substitute for tellurium in solar

116

cells22, fellow rare earth terbium as a substitute for dysprosium as a dopant in neodymium

117

magnets23, and fellow platinum group metal palladium for platinum in autocatalysts24. Further,

118

other common strategies applicable to major metals are often not equally applicable to

119

byproducts. For example, existing mines’ production volumes cannot directly be increased in

120

response to greater demand for the byproduct because of the additional costs cannot be justified

121

in most cases. However, byproduct yield can be improved by refiners through the use of higher

122

efficiency extraction processes in a somewhat analogous way, allowing for greater extraction of

123

byproduct from what becomes available through mining wastes.

124

The present work aims to expand upon the existing criticality literature to address

125

challenges of evaluating strategies for supply risk mitigation of critical byproduct materials. The

126

proposed framework draws upon previous work by the authors in byproduct supply and demand

127

modeling25 to systematically evaluate and compare the effectiveness of different potential

128

mitigation strategies using a small set of key metrics. This approach seeks to strike a balance

129

between conflicting objectives, representing a mid-level modeling solution that is more

130

informative than raw criticality results but much less complex and resource-intensive than

131

detailed, agent-based or general equilibrium market models. Further, by grounding the analysis

132

in dynamic material flow analysis, temporal aspects of different mitigation options’ strengths and

133

weaknesses can be considered. Three commonly discussed firm-level strategies will be used to

134

test the framework with the goal of identifying which response is most effective overall at

135

reducing supply risk for case study material, tellurium. Tellurium is a byproduct of copper

136

production considered to be critical to cadmium telluride solar photovoltaics.

137 138 139

2.0 Methods To evaluate how supply risk can be reduced through different mitigation strategies, a set

140

of scenarios is developed. First, a scenario representing the baseline supply and demand in the

141

absence of intervention is created to establish risk associated with existing and expected future

142

production and consumption trends. Then, additional scenarios representing alternate supply

143

and/or demand expected to result from implementation of each mitigation strategy are created for

144

comparison to the baseline as a basis for measuring supply risk reduction potential. Finally, a set 5 ACS Paragon Plus Environment

Environmental Science & Technology

145

of novel metrics are proposed to describe the different ways strategies may reduce risk using

146

features of the modified supply and demand projections and how they change upon

147

implementation. Benefits of different strategies relative to one another are then assessed using

148

sensitivity of those change metrics to different levels of mitigation possible.

Page 6 of 28

149 150

2.1. Creating a baseline market scenario

151

Independently derived models of future byproduct supply and demand are constructed

152

using updated versions of dynamic material flow analysis methods previously described by the

153

authors25. These methods use carrier metal production data to estimate byproduct supply in light

154

of limited data availability and the crucial role of carrier metal supply-chain dynamics in driving

155

byproduct availability. The authors were inspired to develop this method by emerging trends in

156

copper production with negative implications for tellurium supply; i.e. increasing use of non-

157

tellurium-yielding production techniques, i.e. hydrometallurgical solvent extraction and

158

electrowinning (SX-EW) and steady use of secondary copper as a share of total production. For

159

a more detailed description of this approach and its implications for tellurium, refer to the

160

previous publication. A basic overview is provided in the following paragraphs with additional

161

data and discussion of certain parameters included in the supplemental information files

162

(Appendix A). Because approach of using carrier metal proxy data to estimate future supply is

163

one that can be useful beyond this case study of tellurium production, generalized description of

164

parameters is used when possible and then further clarified for its specific manifestation in this

165

case.

166

In this approach, primary supply of the byproduct is modeled as a function of its

167

associated major carrier metal production rate, C [i.e. tonnes of copper], the share of total carrier

168

metal production yielding the byproduct, α [i.e. % of copper produced via electrolytic refining

169

rather than SX-EW or secondary] , byproduct ore grade associated with the carrier metal, x [i.e.

170

kg tellurium available in ore relative to each tonne of copper produced], and byproduct yield per

171

unit of major metal produced, γ [i.e. % of available tellurium ultimately extracted and refined].

172

All of this is then offset by a factor, f, representing the share of total byproduct production

173

associated with the predominant carrier metal as opposed to other means [i.e. tellurium produced

174

from copper anode slimes as opposed to lead slag skimmings]. Each of these parameters may

175

have dynamic character in reality, however, only total copper production, C, and tellurium6 ACS Paragon Plus Environment

Page 7 of 28

Environmental Science & Technology

176

yielding share, α, are evolved over time in the base case scenario due to their crucial role and

177

superior available knowledge of historical trends to project forward for these parameters.

178

Alternate assumptions regarding these static values are tested for sensitivity and reported upon in

179

the supplemental (Appendix B).

180

For the case of tellurium, the predominant carrier metal is copper, estimated to supply

181

over 90% of tellurium26. The byproduct-yielding share of copper production, α, refers to

182

electrolytic refining of copper concentrates, which has been observed to be falling steadily as

183

share of total copper production for several decades from over 80% before 1987 to below 65%

184

by 201427. In the baseline scenario, it is assumed to continue to decline according to historically

185

observed trends of about 0.4% share per year. This decline is driven by increasing use of

186

hydrometallurgical (SX-EW) copper mining and recycling of copper scrap, neither of which

187

offer meaningful potential for tellurium by-production28. Conversely, although yield is a

188

parameter that may evolve in the future with direct implications for tellurium supply, in the

189

baseline scenario it is assumed to be constant over time and remain at a conservative estimate of

190

35% 29,30. Rather than evolve this parameter over time and assert how that happens within the

191

base scenario, this fact became a major motivation for selecting yield improvement as a

192

mitigation strategy for evaluation by the framework being developed. Total copper refinery

193

production is assumed to continue growing at a historical rate of about 3% per year in the base

194

case. Given rapid consumption in the face of declining ore grades, some recent studies suggest

195

this may be an optimistic estimate beyond mid-century where primary mining may peak31,32. As

196

a result, alternate growth rates are explored and presented in the supplemental (Appendix B). It

197

is possible that this assumption is too strong, which would result in the underestimation of

198

criticality risk for tellurium in the base case, however, in a major commodity market such as

199

copper techno-optimists would argue there is potential for self-correction by miners beyond what

200

is currently used to characterize these peak forecasts (e.g. scarcity drives prices up, high prices

201

create incentive for more mines to open, development of better extraction technology, and

202

increased scrap recovery, substitution). In fact, the relative weakness of these feedback

203

mechanisms in byproduct material production systems is the reason they require the additional

204

strategizing to manage criticality at the heart of this work.

205 206

Secondary supply of the byproduct is also included in the supply model. The current approach focuses on old scrap from recycling of end-of-life PV since it is the most materially 7 ACS Paragon Plus Environment

Environmental Science & Technology

Page 8 of 28

207

intensive tellurium-consuming end use product. Secondary supply from PV is modeled as a

208

function of recycling rate, R [e.g. % of Te in CdTe PV panels in stock collected and recycled]

209

and previous year’s demand, DPV(t-T) [e.g. kg Te in CdTe PV panel stock], which is assumed to

210

become available according to a normally distributed product lifetime, where L is the average

211

with standard deviation σ in years. Other distributions, such as Weibull, have been used

212

elsewhere in the literature on solar PV lifetime but with limited empirical justification as to its

213

applicability for CdTe PV specifically; for simplicity, a normal distribution is used to establish

214

the base case as in Eq. 1. Alternate treatment using a Weibull distribution is included in the

215

supplemental to test sensitivity to this assumption, but it was shown to have negligible impact on

216

the overall conclusions (Appendix B). PV recycling rate is set to a low, but non-zero, initial

217

value of 0.1% in the base scenario to represent the fact that there is currently no secondary

218

supply of tellurium26 without introducing potential calculation errors when relative change in this

219

parameter is used later to help describe risk reduction effectiveness. Secondary supply from

220

non-PV sources, such as thermoelectric devices, could also be included (third term in numerator

221

of Eq. 2), but in this base case, is assumed to be equal to zero for all time periods. Total supply

222

is the sum of primary and secondary supply (Eq. 2).

223 t D (t −T) (T −L)2  EOL PV = ∑  PV exp 4σ 2  T=1  σ 2π

224 225

S(t) =

α(t)⋅γ(t)⋅x⋅C(t)

+

R(t)⋅EOLPV + R'(t)⋅EOLnon-PV f

(1) (2)

226 227

As the predominant and critical end-use of tellurium demand is modelled explicitly as a

228

function demand for cadmium telluride solar panels (CdTe PV). Demand from solar is predicted

229

from external projections of future solar energy demand, estimation of CdTe’s share of the solar

230

energy market, the material intensity of CdTe panels, and replacement needs as described in Eqs

231

3 and 4. Annual demand for CdTe solar power is modelled using a logistic growth form to

232

predict installed capacity (stock) because there is significant competition in the solar market and

233

exponential growth like that observed may not continue for this particular technology (Eq 3).

234

The eventual plateau in demand characteristic of these “S-shaped” functions is controlled

235

by parameter, K, set at 150 GW in the baseline scenario. Fitting parameters β0 and β1 are used to 8 ACS Paragon Plus Environment

Page 9 of 28

Environmental Science & Technology

236

match the observed historical growth, approximated using installations reported by First

237

Solar33,34, the major manufacturer or these panels, to reach the steady state demand plateau

238

previously described by 2035. This base scenario assumes moderate levels of solar growth

239

predicted by the US Energy Information Administration’s International Energy Outlook 2016

240

under the reference scenario35, paired with an aggressive short-term outlook for the technology to

241

grow to represent up to one third of installed solar generation capacity by 2030. This assumption

242

leads to 2020 sales of 8 GWp/y in line with the range of 10-36 GWp/y reported in the literature

243

for the entire thin film market (including CIGS and amorphous silicon PV)36 and cumulative

244

installed CdTe capacity in line with a more modest share (5-10%) of the total solar market as

245

predicted by optimistic groups, like the International Renewable Energy Agency (IRENA) and

246

the International Energy Agency’s Photovoltaic Power Systems Program (IEA-PVPS)37. The

247

aggressive ramp up scenario evaluated here is followed by an assumed transition to a more

248

diverse renewable energy future. CdTe sourced power demand eventually plateaus as other gen-

249

II and gen-III PV technologies, like CIGS, multijunction III-V devices, dye-sensitized organic

250

PV, and inorganic perovskites, currently in development and early stages of market penetration,

251

begin to compete commercially and ramp up in market share as well. However, the base

252

scenario assumes some share of the market will remain best served by CdTe for its combination

253

of moderately low levelized cost, strong high-temperature performance, and quick energy-

254

payback time. Key candidates for this technology to persist include lower-income Subsaharan

255

African and South Asian nations where high-temperature yield advantages are maximized and

256

users would benefit from relatively low cost, yet mature technology38,39.”

257

Material intensity of the panels [e.g. tellurium mass per unit power] is modelled as a

258

function of panel area, A, cadmium telluride active layer film thickness, and density, τ and ρ,

259

tellurium mass fraction, n, and panel power rating, Π. Inherent within panel power rating is a

260

measure of panel efficiency; for the baseline scenario, 160 W/m2 rated panels are assumed,

261

reflecting the average 16% efficiency for newly produced modules from leading CdTe

262

manufacturer First Solar as of 201540,41. Finally, total demand for tellurium from PV is

263

calculated considering extra capacity installation needed to replace end-of-life panels as in Eq 4.

264 265

StockPV (t) =

A τρnK Π1 + exp( −β0 −β1 t ) 

(3)

9 ACS Paragon Plus Environment

Environmental Science & Technology

266

DPV (t) = Stock(t) -Stock(t-1) + EOLPV (t) D(t) = DPV (t) + Dnon-PV (t)

267

Page 10 of 28

(4) (5)

268 269

However, because solar is not the only end-use of tellurium, estimation of total market

270

balance for the commodity must include some estimation of non-PV demand as well (Eq 5).

271

Other applications include thermo-electrics, ferrous, and non-ferrous alloying42, although it is

272

said that solar is the predominant use consuming 40% of tellurium demand43. Less is published

273

about non-PV tellurium demand growth expectations, so this component was treated

274

simplistically in this framework demonstration by assuming all other sectors grow

275

proportionately to the solar sector. Although this approach represents a significant simplifying

276

assumption, more detailed modeling is beyond the scope of the current work and alternate

277

assumptions are tested in the Appendix for reference. Ideally, where available for other

278

markets, similarly detailed approximations of non-critical application demand should be added to

279

this formulation of total demand. In the present form, non-PV demand is represented using the

280

following approximation: Dnon-PV (t) = (1-η) η ⋅ DPV (t) , where η is share of tellurium demand from

281

solar, fixed at 40% in the base case.

282

Note that market clearing price-related feedback mechanisms are intentionally excluded

283

from this framework because the goal is not to predict the actual future supply and demand, but

284

rather to frame a discussion of risk reduction in terms of deviation from the current path.

285

Although this choice more than likely prevents the modeled supply and demand from reflecting

286

realistic future conditions, it allows for the communication of powerful insights regarding supply

287

risk while also eliminating significant computational complexity and data collection

288

requirements necessary to fully model price dynamics. In the case of a byproduct material like

289

tellurium, this data collection and market modeling burden is doubled because both the carrier

290

metal market and byproduct market would need to be described, making the trade-off with this

291

approaches limitations more appealing than it might be otherwise for other more simply

292

structured commodity markets.

293 294 295 296

2.2. Measuring baseline supply risk Once baseline supply and demand scenarios are established, it becomes possible to identify features that communicate different information regarding risk; in this case, of potential 10 ACS Paragon Plus Environment

Page 11 of 28

Environmental Science & Technology

297

market imbalance. Market imbalance is of concern at any point in time where demand outpaces

298

supply. As a result, it becomes clearest to assess this risk by mapping the difference between

299

demand and supply over time. When the market imbalance is positive, conditions for supply gap

300

formation are in place and supply risk is high. When market imbalance is negative, supply risk

301

is low.

302

However, more detail is made available regarding supply risk than just whether or not

303

supply gap conditions emerge. Specifically, the following three features show promise: year at

304

onset of supply gap conditions, degree of peak market imbalance, and duration of supply gap

305

condition persistence. Mathematical descriptions of each of these metrics is shown in Eqs 6-8.

306

Onset : t where S(t) = D(t) and

307

d D(t) - S(t) ) > 0 dt (

(6)

308

Imbalance: max {D(t)−S(t)} subject to t > Onset and D(t) − S(t) > 0

(7)

309

Duration : t − Onset where S(t) = D(t) and

d D(t) − S(t) ) < 0 dt (

(8)

t

310

Onset of supply gap conditions, or rather the number of years from model start until the onset of

311

supply gap conditions, is useful to characterize supply risk because it communicates the time

312

frame available for development of mitigation strategies (e.g. alternate supply resources or for

313

product innovations to reduce demand while still allowing demand to grow at present rate); the

314

sooner the onset, the faster the action is needed if maintaining the current course is desired.

315

Similarly, the peak market imbalance is useful to characterize supply risk because it can

316

communicate the quantitative deviation necessary to avoid supply gaps (e.g. additional supply

317

capacity needed to open or the amount of material that needs to be stockpiled for future use); the

318

greater the market imbalance, the larger the change in supply and/or demand needed to avoid gap

319

conditions. Finally, the duration of the supply gap condition is also useful to characterize supply

320

risk because it can communicate the degree of permanence needed from a solution (e.g. is a

321

temporary ramp up in marginal production sufficient, if possible, or is investment in developing

322

totally new direct mining resources required?) Together these metrics can communicate a great

323

deal of information directly useful for the development of mitigation strategies designed to

324

reduce supply risk.

325 326

2.3 Selecting Strategic Response Mechanisms 11 ACS Paragon Plus Environment

Environmental Science & Technology

327

Page 12 of 28

Comprehensive analysis of all mitigation strategies outlined in Table 1 would be ideal to

328

produce a hierarchy of potential solutions in order of effectiveness, however this is outside the

329

scope of this work. Therefore, three mitigation strategies were selected to serve as commonly

330

discussed, illustrative examples suited for byproduct materials in this analysis: byproduct yield

331

improvement, dematerialization via film thickness reduction, and recycling collection rate

332

improvement. These represent one from each of the three key mitigation categories: primary

333

supply, secondary supply, and demand efficiency. Further, these approaches were strategically

334

selected to focus on responses available at the firm level and not requiring broader legislation or

335

diplomacy efforts that may be out of the purview of a company. In each instance, the associated

336

parameter values in the supply or demand models – yield, film thickness, and recycling rate,

337

respectively – are varied through a range dictated by the previous literature to create multiple

338

scenarios. A unique supply or demand scenario is generated for several degrees of improvement

339

for each strategy and used to determine risk reduction effectiveness as described in the following

340

section.

341 342

2.4. Measuring Risk Reduction Effectiveness

343

Effectiveness of individual mitigation strategies is probed by, first, identifying the model

344

parameters which are affected by each approach, then using it to perturb the model in a way that

345

represents the effect these strategies would have on the system. The results are supply and/or

346

demand scenarios for the byproduct that differ from the baseline, creating shifts in the previously

347

identified risk metrics – (1) onset of supply gap conditions, (2) degree of market imbalance, and

348

(3) duration of supply gap – as well. Risk reduction efficacy of the strategies are then assessed

349

using three companion metrics: (1) delay of supply gap onset, (2) reduction in peak market

350

imbalance, and (3) reduction in supply gap duration; all calculated as percentage change of the

351

new scenario value from its baseline value. However, there are also at least 3 different ways to

352

describe effectiveness of each risk reduction strategy (equations listed in SI Table 2).

353

First, there is the maximum benefit, which is the largest reduction observable using the

354

strategy. This represents an upper bound on risk mitigation potential and can be communicated

355

as a raw result (e.g. x years delay of supply gap onset) or relative to the baseline risk level (e.g. y

356

more years delay of supply gap onset than in baseline scenario). Another potential measure is

357

the average benefit, which is calculated as the change in the raw risk metric, such as years of 12 ACS Paragon Plus Environment

Page 13 of 28

Environmental Science & Technology

358

delay in supply gap onset, over the percentage change in the relevant model input parameter,

359

such as yield, from its baseline. This allows for comparison across the different risk reduction

360

strategies, each of which is based upon different parameters with different units and ranges.

361

Finally, the marginal benefit from implementing a strategy, can be calculated as the percentage

362

change in the risk reduction measure over the percentage change in the input parameter.

363

Marginal benefit results in a unitless measure, which allows for comparison between the risk

364

metrics for a given risk reduction strategy, which have different units, as well as across the

365

different strategies. Ultimately this enables the calculation of a single measure of risk reduction

366

effectiveness for each strategy that encompasses the impact to each of the three risk metrics.

367 368

3.0 Results & Discussion

369

3.1. Updated Baseline

370

Figure 1 shows the baseline scenario against which all strategic response scenarios will

371

be compared. Onset of supply gap conditions is expected in 2020 with imbalance peaking at 999

372

tonnes in 2024 before declining after total of 8 years. The periodic form observed for the market

373

imbalance is caused by the interaction of a plateauing logistic PV stock, lagging replacement

374

needs, and relatively slow-growing supply potential (see Appendix Fig S1 for underlying supply

375

and demand series). It is important to note that this result is not intended to be a prediction of

376

actual future market conditions because it lacks key market clearing feedback mechanisms.

377

However, it serves as an effective basis for this analysis because it communicates the level of

378

risk implied by the current path of supply and demand behavior, which is valuable to know for

379

those looking to reduce risk and understand the relative difficulties of doing so.

380

13 ACS Paragon Plus Environment

Environmental Science & Technology

Page 14 of 28

381 382

Figure 1. Baseline tellurium market balance scenario showing onset of supply gap conditions,

383

peak market imbalance, and gap duration.

384 385 386

3.2. Dematerialization Dematerialization was modeled as a reduction of the thin film layer of CdTe. In reality,

387

efficiency improvement may also play a large role. The baseline scenario assumes layer

388

thickness of 3 microns, but here a range from 5 microns to 0.67 microns is considered for

389

potential scenarios. Although it is traditionally considered that 1 micron is the limitation of film

390

thickness without deterioration of PV cell electrical performance, reduction to 0.67 micron was

391

proposed as a goal for supply risk reduction44. Figure 2 shows market balance scenarios

392

representing the full range of film thicknesses found in the literature as compared to the baseline

393

in black. Interestingly, the results indicate that for film thicknesses less than 2 micron, supply

394

gap conditions are prevented from emerging altogether, even in the face of slowing growth of

395

electrolytic copper production and corresponding tellurium supply. This possibly provides

396

further motivation to push toward what has traditionally been considered the limitation of film

397

thickness reduction. Further, note that the dematerialization scenarios diverge during years with

398

growing demand but converge again during years with shrinking demand. This is due to the fact

399

that demand is directly proportional the film thickness.

400 14 ACS Paragon Plus Environment

Page 15 of 28

Environmental Science & Technology

401 402

Figure 2. Market scenarios for CdTe PV dematerialization as an approach to tellurium risk

403

reduction.

404 405

To quantify risk reduction potential, impact of changing strategy relevant input

406

parameters – in this case film thickness – on risk metrics must be calculated. Figure 3a shows

407

results for delay of supply gap onset relative to the baseline scenario of 3 micron film thickness,

408

i.e. how effective dematerialization would be to delaying supply issues. To the right, where there

409

is higher film thickness, supply gap is projected to emerge earlier, as soon as 2018 for fleet

410

average film thickness of 5 microns and 2019 for thickness of 4 microns. When film thickness is

411

reduced below 3 microns assumed in the baseline, it continues a through a linear period of

412

improvement before approaching infinity below 2 micron thickness. Therefore, if such film

413

thickness reduction can be achieved without tradeoffs due to efficiency loss, even aggressive PV

414

growth like what has been modeled in the baseline scenario can proceed without inducing supply

415

gap conditions at all.

416

Next, the reduction in peak market imbalance is shown in Figure 3b. A maximum

417

decrease in peak market imbalance of about 1700 tonnes can be achieved, which is actually 75%

418

larger than the baseline imbalance value. This is because in the film thickness scenarios below 2

419

microns, the entire baseline imbalance is eliminated and a supply surplus is projected. By

15 ACS Paragon Plus Environment

Environmental Science & Technology

Page 16 of 28

420

contrast with gap onset delay, market imbalance reduction maintains a linear form throughout the

421

entire range of scenarios.

422

Finally, reduction in gap duration is shown in Figure 3c. Results show a slightly less

423

linear trend than for reduction of market imbalance due to this metric’s greater dependence upon

424

the non-linear shape of the entire market balance curve, creating reductions in the gap from both

425

directions through later onset and swifter recovery. A maximum reduction in gap duration of 8

426

years is achievable, which corresponds to the length of the entire baseline gap, again because of

427

the complete avoidance of gap conditions below 2 micron film thickness. The difference in

428

linearity of effects underscores the need to examine a range of risk metrics.

429

430

16 ACS Paragon Plus Environment

Page 17 of 28

Environmental Science & Technology

431 432

433 434

Figure 3. (a) Effectiveness of dematerialization as a means to delay supply gap onset. (b)

435

Effectiveness of dematerialization to reduce peak market imbalance. (c) Effectiveness of

436

dematerialization to reduce duration of supply gap conditions.

437 438

3.3. Yield Improvement 17 ACS Paragon Plus Environment

Environmental Science & Technology

439

Page 18 of 28

Similarly, yield improvement was modeled by creating scenarios that represent a

440

reasonable range of values for fraction of tellurium that gets recovered from initial content in

441

copper ore. Because this definition of yield is broader than the efficiency of a single extraction

442

process, it is not reasonable to expect near perfect recovery could be possible. The concept here

443

is intended to capture losses across all stages of mining, refining, and byproduct extraction, and

444

the carrier metal processing that precedes tellurium extraction from the anode slime is not

445

optimized to reduce tellurium losses. A reasonable range of 35 to 80% discussed in previous

446

literature45 was ultimately selected (see Appendix A4 for more detail) (Figure 4).

447

Unlike how, in the dematerialization scenario set, the alternate film thickness cases scaled

448

proportionally in all time periods, yield improvement benefits amplify over time because this

449

parameter scales with growing supply and downward trending periodic demand. Notably, if

450

overall yield of 55% or above can be achieved as is sometimes reported in the literature, supply

451

gap conditions can be avoided by this technique as well and can lead to even more favorable

452

availability conditions, where relatively more excess supply capacity is expected.

453

454 455

Figure 4. Effectiveness of byproduct yield improvement for reducing supply risk, showing all

456

scenarios considered.

457 458

3.4. Recycling Collection Rate Improvement

18 ACS Paragon Plus Environment

Page 19 of 28

459

Environmental Science & Technology

Finally, the introduction of mandates or at least goals for recycling rate of solar panels are

460

modeled (Figure 5). Because CdTe solar panels represent the highest tellurium density end-use

461

product for recovery, and are already modeled to calculate demand, the recycling program is

462

intended to apply to only this application. Currently, Austria, as part of broader e-waste

463

legislation, has set goals for future solar cell recycling rates between 70-80% and making it the

464

producers’ responsibility to collect and bear the costs of recycling46. First Solar also has a

465

recycling program in place that currently recycles the company’s fabrication scrap. However,

466

for the baseline, a near zero global average recycling rate of 0.1% is assumed to reflect the

467

current reality that most panels are still in their useful life and currently appear to lack a take-

468

back infrastructure47. The range for scenarios was developed all the way through 100% to reflect

469

a full recycling mandate; in reality, some losses in collection and material recovery are

470

inevitable.

471

The most striking difference between the recycling scenario evolution and the previous

472

strategies discussed is the lack of any benefit to delay and duration of gap onset. This is due to

473

the early emergence of the gap conditions relative to the long product lifetime before secondary

474

material becomes available. End-of-life recycling does impact the degree of market imbalance

475

expected; although in a very minor way, reducing the measure by a maximum of 1 tonne as

476

compared to several thousand by the other approaches reviewed. These are important results

477

because, for as much discussion as surrounds recycling of PV, the approach will have little value

478

in the short to mid-term future in terms of stabilizing tellurium supply risk. However, greater

479

impact could be expected through utilization of prompt scrap recycling from PV manufacturing

480

inefficiencies because this would have a much shorter residence time before it could be used to

481

supplement virgin supply.

482

19 ACS Paragon Plus Environment

Environmental Science & Technology

Page 20 of 28

483 484

Figure 5. Effectiveness of CdTe recycling rate for reducing supply risk for tellurium, showing all

485

scenarios considered.

486 487 488

3.5. Comparing Strategies A key goal of performing this assessment is to compare supply risk mitigation strategies

489

to one another utilizing the metrics proposed. However, it is necessary first to determine, if,

490

how, and why the choice of evaluation metric matters to decision making. For example, does use

491

of average benefit lead to the different conclusions about strategy effectiveness as marginal

492

benefit does? In this case the answer is no as indicated by the consistent ranking of

493

dematerialization over yield improvement over recycling rate improvement regardless of risk

494

metric using each effectiveness measurement approach, as shown in Figure 6. However, it is

495

important to emphasize that despite having the same units, results across risk metric categories in

496

terms of average benefit cannot be compared due to their differing baseline and/or units, so the

497

normalized marginal benefit metric should be used if comparison is the goal.

498

20 ACS Paragon Plus Environment

Page 21 of 28

Environmental Science & Technology

499 500

Figure 6. Comparison of results across the proposed measures of effectiveness: a) maximum

501

benefit, b) average benefit, and c) marginal benefit (graph), including detail on each contributing

502

risk reduction factor (column cluster) and mitigation strategy (column color).

503 504

An additional interesting result is the ability to generate a single measure of risk

505

reduction effectiveness for each mitigation strategy by summing the normalized metrics (Figure

506

7). The results confirm the greater effectiveness of dematerialization under current conditions,

507

followed by yield improvement, and then recycling. Further, they communicate in a single

508

measure of how much better it is in terms of supply risk reduction potential: in this case,

509

dematerialization provides on average 65% greater benefit than yield improvement for the same

510

relative degree of mitigation parameter change. These results provide a foundation for strategic

511

stakeholders to determine where they will get the greatest marginal benefit in terms of supply

512

risk reduction, suggesting which interventions could be most effective.

513

Despite already suggesting the greatest effectiveness among the strategies considered in

514

this study, the effect of dematerialization and yield improvement on supply risk reduction are

515

both still likely understated using the single score approach alone. They not only delay onset of

516

supply gap conditions and reduce gap duration but completely eliminate it in several of their

517

alternate value cases considered. Marginal benefit is calculated excluding points corresponding

518

to gap avoidance, so it may not fully communicate the value of the approaches that can achieve 21 ACS Paragon Plus Environment

Environmental Science & Technology

Page 22 of 28

519

this. Instead, a comparison of the maximum benefit metric to the baseline risk metrics could be

520

used to reveal this advantage (Figure 6a); however, this would suggest no difference between

521

dematerialization and yield.

522

523 524 525

Figure 7. Total risk reduction potential (measured as marginal benefit) for each of the mitigation

526

strategies: dematerialization, yield improvement, and recycling.

527 528 529

4.0 Discussion This work presents an approach to quantitatively evaluate potential effectiveness of

530

supply risk mitigation strategies informed by previously published scenario modeling techniques

531

[25]. The approach proposes three metrics to measure risk, three metrics to measure risk

532

reduction, and three metrics to measure effectiveness of risk reduction by each strategy from

533

comparison of possible supply-demand imbalance scenarios. This method is one step toward a

534

better design of specific strategies in response to criticality assessment by focusing on

535

parameters that can be leveraged by firms and broader industries. Because the present approach

536

requires relatively in-depth, dynamic modeling of individual materials’ supply chains, it is not

537

recommended that this method be incorporated directly into traditional criticality assessment

538

frameworks. Rather, it could be thought of as a supplemental tool for use once the most critical

539

materials have been identified but before more detailed market simulation need be pursued. The 22 ACS Paragon Plus Environment

Page 23 of 28

Environmental Science & Technology

540

goal is to standardize the ways in which decisions are made regarding criticality mitigation by

541

creating a tractable, quantitative basis for comparison between different mitigation strategies.

542

In the case of tellurium, the results consistently present a strong case for pursuing both

543

dematerialization and yield improvement strategies because each is shown to be able to prevent

544

gap conditions from emerging, suggesting significant lowering of supply risk for the otherwise

545

short-term highly critical material. However, results suggest that incremental progress toward

546

dematerialization (via film thickness reduction) might be significantly more effective than the

547

same degree of incremental improvement in yield: about 65% more so as indicated by comparing

548

the marginal benefit scores for these techniques. Finally, despite much research on recycling as a

549

solution to critical materials supply, in the case of materials like tellurium where gap short-term

550

criticality is the key problem and the major application has a long product lifetime with no

551

existing recycling infrastructure, the benefits of end of life recycling is not expected to be

552

effective at preventing or mitigating supply gap conditions.

553

The model presented here has several limitations. A strategy’s effectiveness is logically

554

quite sensitive to when it is implemented, but the current approach does not assume a slow

555

dynamic transition, rather a discrete jump to the alternate scenarios beginning in a particular year

556

(model year 1 for the baseline). This may not necessarily be the most realistic scenario but it can

557

be thought to set expectations and creates goals by presenting case an upper bound of

558

effectiveness via an immediate change in the system as compared to a longer-term phase in

559

solution. Several other simplifying assumptions exist in the current approach to modeling supply

560

and the demand. For example, in the specific case study assessed, tellurium demand is assumed

561

to increase at the same rate for other end-uses as for PV and similarly supply is assumed to grow

562

at the same rate for non-copper byproduction streams of tellurium as for copper. Some

563

assumptions skew the presented analysis in a conservative manner while others may be overly

564

optimistic. A less optimistic total copper production growth rate and a higher CdTe PV adoption

565

rate (both of which may be likely) will exacerbate the supply issue studied here, whereas

566

assuming a continuing gradual shift toward non-tellurium-yielding SX-EW copper production as

567

share of total and the continuation of rapid adoption of CdTe PV would alleviate baseline

568

risk. These assumptions could be adjusted if more detailed data is obtained. However, in the

569

interest of keeping data collection burden moderately low, sensitivity of the case study findings

570

to alternate assumptions are explored instead in the supplemental (Appendix B). The authors 23 ACS Paragon Plus Environment

Environmental Science & Technology

571

believe the results represent a reasonable basis for demonstration of the proposed mitigation

572

strategy evaluation methods and provide heightened level of insights into key levers for the

573

tellurium market to respond to criticality risk.

574

Page 24 of 28

Whether at the firm level or national level, translating information into action, even at the

575

level presented here, will still be challenging. This framework is intended to inform one

576

component of a more complex, multifaceted evaluation of candidate strategies’ benefits and

577

weaknesses likely to be necessary for institutional decision-making. For example, future work

578

could include additional analysis to evaluate complex tradeoffs in terms of other practical

579

considerations for each technique (e.g. marginal cost of improvement, environmental benefit of

580

improvement). Overall, the exploration of multiple effectiveness indicators for a diverse

581

sampling of mitigation strategies should serve as a reminder that there will be no universal

582

solution for all materials nor for a given materials critical applications. Instead, optimal results

583

should be pursued through targeted and temporally-relevant strategic development. As such,

584

these findings on strategy effectiveness should be regarded as specific to the tellurium market

585

and not a general conclusion on their merit for other commodity systems.

586 587

Acknowledgements

588

This work has been funded by the National Science Foundation (NSF) (CBET #1454166), the

589

Golisano Institute for Sustainability (GIS), and Rochester Institute of Technology (RIT). Special

590

thanks to Dr. Callie Babbitt for her helpful feedback and guidance in developing this work.

591 592

Supporting Information

593

The supporting information contains: A1) additional literature review, A2) model parameters,

594

A3) risk reduction metrics, A4) mitigation strategy parameters, B1) baseline sensitivity analysis,

595

and B2) alternate risk reduction sensitivity analysis.

596 597

TOC Art

24 ACS Paragon Plus Environment

Page 25 of 28

Environmental Science & Technology

598 599

References

600

(1)

601 602

chains: Current status, constraints and opportunities. Resour. Policy 2014, 41 (1), 52–59. (2)

603 604

Alonso, E.; Gregory, J.; Field, F.; Kirchain, R. Material availability and the supply chain: Risks, effects, and responses. Environ. Sci. Technol. 2007, 41 (19), 6649–6656.

(3)

605 606

Golev, A.; Scott, M.; Erskine, P. D.; Ali, S. H.; Ballantyne, G. R. Rare earths supply

Duclos, B. S. J.; Otto, J. P.; Konitzer, D. G. Design in an Era of Constrained Resources. A.S.M.E. Mech. Eng. 2010, No. September, 36–40.

(4)

Ku, A. Y.; Dosch, C.; Grossman, T. R.; Herzog, J. L.; Maricocchi, A. F.; Polli, D.; Lipkin,

607

D. M. Addressing Rare-Earth Element Criticality: An Example from the Aviation

608

Industry. JOM 2014, 66 (11), 2355–2359.

609

(5)

Lloyd, S.; Lee, J.; Clifton, A.; Elghali, L.; France, C. Ecodesign through Environmental

610

Risk Management : A Focus on Critical Materials. In Proceedings of EcoDesign 2011

611

International Symposium; 2012; pp 1–6.

612

(6)

National Research Council. Minerals, critical minerals, and the US economy; 2007.

613

(7)

European Commission. Report on critical raw materials for the EU; 2014.

614

(8)

Graedel, T. E.; Barr, R.; Chandler, C.; Chase, T.; Choi, J.; Christofferson, L.; Friedlander,

615

E.; Henly, C.; Jun, C.; Nassar, N. T.; et al. Methodology of Metal Criticality

616

Determination. Environ. Sci. Technol. 2012, 46 (59), 1–58.

617 618

(9)

Moss, R. L.; Tzimas, E.; Kara, H.; Willis, P.; Kooroshy, J. Critical Metals in Strategic Energy Technologies: Assessing Rare Metals as Supply-Chain Bottlenecks in Low-Carbon 25 ACS Paragon Plus Environment

Environmental Science & Technology

619

Energy Technologies; European Commission Joint Research Centre Institute for Energy

620

and Transport, 2011.

621

(10)

622

Page 26 of 28

Bauer, D.; Diamond, D.; Li, J.; Sandalow, D.; Telleen, P.; Wanner, B. US Department of Energy: Critical Materiasl Strategy, December 2010. Agenda 2010, No. December, 1–166.

623

(11)

European Commission. Critical raw materials for the EU. 2010, p 84.

624

(12)

Graedel, T. E.; Nuss, P. Employing Considerations of Criticality in Product Design. Jom

625 626

2014, 66 (11), 2360–2366. (13)

627 628

Panousi, S.; Harper, E. M.; Nuss, P.; Eckelman, M. J.; Hakimian, A.; Graedel, T. E. Criticality of Seven Specialty Metals. J. Ind. Ecol. 2016, 20 (4), 837–853.

(14)

Viebahn, P.; Soukup, O.; Samadi, S.; Teubler, J.; Wiesen, K.; Ritthoff, M. Assessing the

629

need for critical minerals to shift the German energy system towards a high proportion of

630

renewables. Renew. Sustain. Energy Rev. 2015, 49, 655–671.

631

(15)

632 633

mineral supply. Miner. Depos. 2012, 47 (7), 713–729. (16)

634 635

(17)

(18)

Achzet, B.; Helbig, C. How to evaluate raw material supply risks-an overview. Resour. Policy 2013, 38 (4), 435–447.

(19)

640 641

Nassar, N. T.; Graedel, T. E.; Harper, E. M. By-product metals are technologically essential but have problematic supply. Sci. Adv. 2015, 1 (3), e1400180–e1400180.

638 639

Redlinger, M.; Eggert, R. Volatility of by-product metal and mineral prices. Resour. Policy 2016, 47, 69–77.

636 637

Wellmer, F. W.; Dalheimer, M. The feedback control cycle as regulator of past and future

Goe, M.; Gaustad, G. Identifying critical materials for photovoltaics in the US: A multimetric approach. Appl. Energy 2014, 123, 387–396.

(20)

Bustamante, M. L.; Gaustad, G. Materials Research to Enable Clean Energy: Leverage

642

Points for Risk Reduction in Critical Byproduct Material Supply Chains. REWAS 2016

643

Towar. Mater. Resour. Sustain. 2016, No. Table I, 193–201.

644

(21)

645 646

Allwood, J. M.; Ashby, M. F.; Gutowski, T. G.; Worrell, E. Material efficiency: A white paper. Resour. Conserv. Recycl. 2011, 55 (3), 362–381.

(22)

Nassar, N. T.; Barr, R.; Browning, M.; Diao, Z.; Friedlander, E.; Harper, E. M.; Henly, C.;

647

Kavlak, G.; Kwatra, S.; Jun, C.; et al. Criticality of the Geological Copper Family.

648

Environ. Sci. Technol. 2012, 46 (2), 1071–1078.

649

(23)

Nassar, N. T.; Du, X.; Graedel, T. E. Criticality of the Rare Earth Elements. J. Ind. Ecol. 26 ACS Paragon Plus Environment

Page 27 of 28

Environmental Science & Technology

650 651

2015, 19 (6), 1044–1054. (24)

652 653

Nassar, N. T. Limitations to elemental substitution as exemplified by the platinum-group metals. Green Chem. 2015, 17 (4), 2226–2235.

(25)

Bustamante, M. L.; Gaustad, G. Challenges in assessment of clean energy supply-chains

654

based on byproduct minerals: A case study of tellurium use in thin film photovoltaics.

655

Appl. Energy 2014, 123, 397–414.

656

(26)

657

US Geological Survey (USGS). Mineral Commodity Summaries 2016. United States Geological Survey 2016.

658

(27)

ICSG. World Copper Factbook 2015. 2015.

659

(28)

Bustamante, M. L.; Gaustad, G. The evolving copper-tellurium byproduct system: A

660

review of changing production techniques & their implications. In TMS Annual

661

Meeting; 2014.

662

(29)

663

Guilinger, J. Assessment of Critical Thin-Film Resources. RAF-9-29609. TELLURIUM. In World Industrial Minerals; Golden, CO, 1999.

664

(30)

Goldfarb, R. Tellurium —The Bright Future of Solar Energy; 2015.

665

(31)

Northey, S.; Mohr, S.; Mudd, G. M.; Weng, Z.; Giurco, D. Modelling future copper ore

666

grade decline based on a detailed assessment of copper resources and mining. Resour.

667

Conserv. Recycl. 2014, 83, 190–201.

668

(32)

Sverdrup, H. U.; Ragnarsdottir, K. V.; Koca, D. On modelling the global copper mining

669

rates, market supply, copper price and the end of copper reserves. Resour. Conserv.

670

Recycl. 2014, 87, 158–174.

671

(33)

First Solar. Q1 2012 Earnings Call & Five Year Plan. 2012.

672

(34)

DeJong, T. Manufacturing Update. In First Solar 2016 Analyst Meeting; 2016.

673

(35)

U.S. Energy Information Administration. International Energy Outlook 2016; 2016; Vol.

674

0484(2016).

675

(36)

IEA. Technology Roadmap: Solar Photovoltaic Energy; 2014.

676

(37)

IRENA; IEA-PVPS. End-of-Life Management: Solar Photovoltaic Panels; 2016.

677

(38)

First Solar. Sustainability Report; 2017.

678

(39)

Mittal, A. K.; Singh, S. Solar Edge - First Solar’s energy advantage in Indian climatic

679 680

conditions. Thin Film Technol. 2017, No. March, 35–38. (40)

Wesoff, E. First Solar Reaches 16.3% Efficiency in Production PV Modules. 27 ACS Paragon Plus Environment

Environmental Science & Technology

Page 28 of 28

681

(41)

First Solar. First Solar Series 4 PV Module v3. 2016, p 2.

682

(42)

US Geological Survey (USGS). Tellurium End-Use Statistics. In Historical Statistics for

683

Mineral and Material Commodities in the United States; 2005.

684

(43)

Selenium-Tellurium Development Association (STDA). SE & TE.

685

(44)

Zweibel, K. The Impact of Tellurium Supply on Cadmium Telluride Photovoltaics.

686 687

Science (80-. ). 2010, 328 (5979), 699–701. (45)

Bustamante, M. L.; Gaustad, G. Challenges in assessment of clean energy supply-chains

688

based on byproduct minerals: A case study of tellurium use in thin film photovoltaics.

689

Appl. Energy 2014, 123.

690

(46) (2014). Verordnung: Änderung der Elektroaltgeräteverordnung (EAG-VO-Novelle 2014).

691

Austria. Rechtsinformationssystem (RIS), http://www.solarwaste.eu/in-your-

692

country/austria/. Verordnung 193.

693

(47) Marwede, M.; Reller, A. Estimation of life cycle material costs of cadmium telluride- and

694

copper indium gallium diselenide-photovoltaic absorber materials based on life cycle

695

material flows. J. Ind. Ecol. 2014, 18 (2), 254–267.

696

28 ACS Paragon Plus Environment