Thiébaud E.(a), Schluep M.(b), Böni H.(a), Hilty L. M.(a,c) and Faulstich M.(d)
a) Empa, Swiss Federal Laboratories for Materials Science and Technology, St. Gallen, Switzerland
b) World Resources Forum, St. Gallen, Switzerland
c) University of Zürich, Department of Informatics, Zürich, Switzerland
d) Clausthal Environmental Institute (CUTEC), Clausthal-Zellerfeld, Germany

Keywords: service lifetime; storage time; dynamic material flow analysis; indium.

Abstract: Electronic devices contain many important resources, including critical chemical elements such as indium or neodymium. For an efficient management of these resources, it is important to know where the devices are located, how long they are used for and when and how they are disposed of. This article presents a dynamic material flow analysis of devices with liquid crystal displays; i.e. flat screen televisions and monitors, laptops and mobile phones, and the subsequent indium flows in Switzerland. The stock of the use phase within the material flow system has been split into an in-use stock and a storage stock. The outflows have been modelled by applying two lifetime distribution functions, one for the service lifetime and one for the storage time. Results highlight the importance of the storage time, being for flat screen TVs of 2.6 years, and for monitors, laptops and mobile phones of over 4 years. For monitors and laptops, the storage stock accounts for around 20%, and for mobile phones 35% of their total stock. These devices in Switzerland represent an indium stock of over 1’800 kg, an indium inflow with new products of 200 kg/year and an outflow with discarded devices of only 90 kg/year. Outflows of the model that includes storage time are significantly lower and show better congruence with actually measured flows. This shows that the storage time slows down the reintegration of secondary resources into the material cycles and therefore increases the stock of resources.

Introduction

Stocks and flows of electronic devices and the critical materials they contain have been subject to various recent investigations (Böni & Wäger, 2015; Buchert et al., 2012; Buchert et al., 2009; Sander et al., 2012; Yoshimura, et al., 2013). Devices with liquid crystal displays (LCDs) are an important field of application of indium tin oxide (ITO) (Erdmann & Behrendt, 2011). Efforts to specifically recover indium from LCDs are only just beginning and are hampered by small concentrations of indium per display, low indium prices that make recycling unattractive and limited knowledge on stocks and flows of LCDs.

Dynamic material flow analysis (MFA) is often used to analyse the development of material cycles over time. The product lifetime, which is essential for computing past and future stocks and outflows from inflow data, has been modelled in dynamic MFAs mostly by assuming average lifetimes or lifetime distribution functions for devices in the in-use stock (Müller et al., 2014). It has been discussed in many MFA studies that the modelled stocks and flows are highly sensitive to the chosen lifetime distribution functions and their parameters (Chen & Graedel, 2012; Liu et al., 2011; Müller et al., 2006). This poses various challenges: There are many different definitions of what is covered by the product lifetime (Murakami et al., 2010). Field data regarding how long the various electronic devices are used and stored is scarce and often based on rough estimations or expert opinions. Comparisons of modelled and actual flows of devices have shown that products are often much older than what models predict (Stocker et al., 2013). This highlights that models often neglect or underestimate storage time, which is considered as the time between the active use of a device and its final disposal or its passing to a different user. In this article, we present a dynamic MFA of devices with LCDs, i.e. flat screen televisions (TVs) and monitors, laptops and mobile phones, and the subsequent indium flows in Switzerland. In order to meet the challenges of finding the lifetime distribution functions and their parameters, we conducted an online survey on the service lifetime (which corresponds to the time of active use) and storage time of electronic devices in Switzerland.

Method

Model development

The MFA system includes the process ‘use phase’. Based on the results of the survey, the use phase was split into the two sub processes: ‘active use’ and ‘storage’, which both store material and build the in-use stock and the storage stock. The system has one inflow which corresponds to sales of new devices and one outflow as the final disposal of obsolete items. Internal flows include the flow from ‘active use’ to ‘storage’ and vice versa as well as the direct reintegration into the in-use stock (Figure ).

MFA system

The model employs a retrospective top-down approach, deriving the stock S[n] at a time n from the net flow by using the balance of masses (equation 1), with the constant sampling rate T = 1 year (Müller et al., 2014).

?[?] = (??????[?] − ???????[?]) ∙ ? (1) + ?[? − 1]

The outflows of the in-use stock and the storage stock were calculated according to equation 2 (Müller et al., 2014). The model applies two different lifetime distribution functions f[m], one for the service lifetime and one for the storage time.

Outflow[n] = ∑ ??????[? − ?] ∙ ?[?]      (2)

The pathways of the outflows to final disposal, storage or back to the in-use stock for a second use were determined by transfer coefficients. In order to calculate the indium flows, the stocks and flows of the dynamic MFA were multiplied with the respective indium content per device.

Data collection

Sales data for laptops and flat screen monitors from 1998 to 2013 were taken from an annual ICT market report for Switzerland (Weiss, 2013). For flat screen TVs and mobile phones, sales data from 1998 and 2003, respectively, up to the year 2005 were available from the Swiss Consumer Electronics Association (SCEA, 2005), and subsequently up to 2013 were obtained from the market research institution GfK. (GfK, 2013; GfK Retail and Technology, 2007, 2008).
The lifetime distribution functions and transfer coefficients were derived from the results of an online survey that was conducted between January and May 2014 in Switzerland. It included questions on devices that were still in use, devices that were stored and already disposed of devices. For each item, information was collected on:

  • the year the device was put on the market
  • the condition of the device when it was
  • purchased by the current user (new/second hand)
  • the total service lifetime (for devices still in use including the years the user intends to continue to use it)
  • for second hand devices the division of the service lifetime into first and second use
  • the storage time (for devices still stored including the years the user intends to continue to store it)
  • for second hand devices the storage time after the second use
  • the disposal pathway (final disposal, storage or reintegration into the in-use stock).

With a Swiss population of 8,000,000 a confidence level of 95% and a margin of error of 5%, we needed a sample size of at least 385 people. In total, we had 439 valid responses to our survey, resulting in 981 data sets for laptops, 349 for monitors, 351 for TVs and 1,690 for mobile phones. In order to derive the lifetime distribution functions, we fitted a two- parameter Weibull distribution function to the relative frequency histograms of the total service lifetime as well as the storage time using the Origin Software (OriginLab, 2014). From our survey, we had no service lifetime and storage time data for devices that are reintroduced into the in-use stock directly after use or after storage. We assumed that they had a similar service lifetime as the second use of the second hand devices covered by our survey. Therefore we fitted a Weibull distribution function to the normalized histograms of the service lifetime of the second use. Likewise, for devices that are stored after the reintroduction into the in-use stock, we considered the storage time after the second use from the survey data.

Figure shows an example of the resulting Weibull distribution functions for mobile phones.

Different Weibull distribution functions

Indium is not only contained in LCDs but also found in printed circuit boards. The indium content was taken from literature (MoE and METI, 2010; Sander et. al, 2012) and own measurements (Böni & Wäger, 2015; Wäger et al., 2014). For devices with various data available, the average indium content was taken.

Results

Service lifetime and storage time

The mean total service lifetime of the resulting Weibull distribution functions ranges from 4 years for mobile phones to 8.7 years for TVs. The mean storage time is shortest for TVs with 2.6 years and longest for mobile phones with 4.7 years (Table ). The longest storage time amounts to 7 years for TVs, 9 years for monitors and 16 years for laptops and mobile phones.

Total service lifetime and storage time

Inflow

The sales of the considered devices with LCDs have all been declining in the past years, partly due to market saturation, but partly also due to a change in the electronics market to smaller, more flexible systems such as tablet computers. These new devices, though not included in our study, show a large increase in the past 4 years (Figure ).

Sales

Stock

The in-use stock and storage stock calculated by the dynamic MFA are illustrated in Figure and Figure . Mobile phones represent with 46% in 2013 the largest share of the in-use stock in terms of number of devices, followed by laptops with 23%. Measured in tonnes, it is obviously the opposite and TVs form the largest stock, followed by monitors (63% and 20%, respectively, of the total in-use stock in 2013). The in-use stock growth has declined for all devices in the last 3 years, mostly due to the decrease in sales.

The stored mobile phones represent 35% of all mobile phones in the use phase, for laptops and monitors the share of stored devices is around 20% and stored TVs account for 3% of all TVs. The storage stock is still growing linearly with the highest growth rate for mobile phones. Regarding the number of devices, the total storage stock accounts for 26% of the total stock of the use phase, regarding the mass, the total storage stock only represents 10%.

Outflow

The share of outflowing devices that are stored after their active use ranges from 31% for TVs up to 66% for mobile phones. The share of reuse and disposal is highest for TVs with 27% and 42%, respectively.

Table lists all transfer coefficients of the outflow of the in-use stock. The outflows from the in-use and the storage stock to final disposal calculated by the dynamic MFA are illustrated in Figure . For mobile phones, the flow to disposal has recently been dominated by outflows of the storage stock. For TVs it is the opposite as outflows are mainly coming directly from the in- use stock. For laptops and monitors, outflows of the in-use stock and the storage stock are of similar size. The outflows are still growing for all devices.

Indium

The considered devices in Switzerland represent an indium stock of 1800 kg, an indium inflow with new products of 200 kg/year and an outflow with discarded devices of only 90 kg/year in 2013. The stock is dominated by TVs, monitors and laptops. Mobile phones with their smaller displays only account for 6% of the total indium stock. The total storage stock adds up to 290 kg or 15% of the total indium stock (Figure ).

Cumulative in-use stock

Cumulative storage stock

transfer coefficients

Outflows from storage

Cumulative indium

Discussion

The total service lifetime is highly influenced by the service a device provides. For TVs and monitors, producers and content provider ensure through backward compatibilities that older devices are still able to display current media streams. New devices basically provide the same service (among others), with higher resolution and probably on larger screens. Therefore, TVs and monitors have the longest mean total service lifetime. In contrast, laptops and mobile phones and their software are subject to fast innovation cycles, so that new devices are bought due to their additional functionality. In addition, these mobile devices become obsolete more frequently due to hardware failure such as insufficient battery life or broken displays.

The transfer coefficients after the active use to storage, reuse and disposal demonstrate that the smaller a device, the more often it is stored after its use. However, besides the size, an important reason that mobile phones and laptops are rather stored than disposed of immediately after use might be their ability to store personal data such as photos, files or messages. Instead of transferring these data to a new device or a cloud service, the device itself is often stored. Mobile phones are also stored as a backup for the actively used phone that could get broken or lost. The reintegration into the in-use stock is highest for TVs, probably due to the above mentioned fact that older devices are still able to provide the required service. Monitors, where the same reasoning would apply, have become so cheap that most likely rather new devices are purchased than old ones reused.

The storage time is again mostly influenced by the size but probably also by the personal attachment to a device (Remy & Huang, 2015). Therefore, mobile phones and laptops are kept for the longest time.

If we compare the number of devices in the in- use stock with stock data of Swiss Statistics (BFS, 2012a, 2012b), the model seems to overestimate the in-use stock for all devices.

This could be due to the fact that our survey only covers the private use of devices. Lifetime and disposal pathways of electronic devices in business use differ greatly from private use as devices are often replaced faster. The transfer coefficients to storage, the storage time and the resulting size of the storage stock, even if it might also be overestimated, highlights the importance of taking into account storage in MFA. If storage was neglected, the calculated outflows would for example be 46% higher for mobile phones and 42% higher for TVs. Compared with actual flows in the Swiss e- waste system (Swico, SENS, SLRS, 2013), one sees that, e.g., for mobile phones, the computed flows are still too high but for TVs the flows correspond quite well (Figure 8).

Comparison of outflows

The reason might be that disposed of mobile phones reach other disposal pathways such as municipal waste, donations for exports etc. whereas TVs are rather disposed of in the official channel. Devices that reach the official Swiss e-waste system are manually and mechanically dismantled and sorted into different material fractions such as metals, plastics, printed circuit boards etc. When recycling LCDs, great importance is attached to the environmentally sound removal and disposal of the mercury containing backlights. Critical metals such as indium are not recovered. The LCD panels containing indium are either stored or incinerated with indium lost to the slag. Current projects aim at finding technologies and financing mechanisms for recycling indium in LCDs (Böni & Wäger, 2015).

This would enable the reintegration of the current indium stock of about 1,800 kg into the material cycle.

Conclusions and outlook

The modelling of product lifetimes is an essential part of dynamic MFA. Product lifetimes are mostly defined as the service lifetimes of products. However, many products are stored after their active use and before they are finally disposed of. In this article, we proposed a model that takes into account consecutively both the service lifetime and the storage time. We investigated the effect of such an extended model on stocks and flows of devices with LCDs. Our results show that especially for small devices, such as mobile phones, which contain valuable resources, storage time should always be considered when product lifetime is explored. Outflows of the model that includes storage time are significantly lower and show better congruence with actually measured flows. However, the presented model tends to overestimate stocks, partly because it does not distinguish between private and business use. Future research should therefore explore and include the service lifetime and storage time of business use. It should also investigate different disposal pathways, since for some devices, outflows still seem too high compared to actually collected flows. Parameters such as Weibull parameters or transfer coefficients could also be further optimized for better correspondence of the model results with actually measured stocks and flows. And finally, as online surveys often have a selection bias, the representativeness of the empiric data should be further verified.

The storage time slows down the waste generation as well as the reintegration of secondary resources into the material cycles and therefore increases the stock of resources. It does not contribute to product longevity and the resource efficiency of electronic devices, since products are replaced once they become obsolete, regardless of whether they are stored or disposed of. However, for an efficient resource management, it is primarily important that devices are brought to collection after use or storage, in order to ensure adequate recycling. Our modelling approach contributes to the understanding of current stocks and flows of devices with LCDs, which is an important basis for further measures towards resource efficiency and waste reduction.

Acknowledgments

We would like to thank all the participants of the survey for sharing information about their electronic devices.

References

BFS. (2012a). Haushalte und Bevölkerung – IKT- Ausstattung (Households and Population – ICT Infrastructure). Abgerufen 23. November 2012, Retrieved from: http://www.bfs.admin.ch/bfs/portal/de/index/theme n/16/04/key/approche_globale.indicator.30103.301 .html

BFS. (2012b). Informationsgesellschaft-Indikatoren: Haushalte und Bevölkerung Telefoninfrastruktur. Abgerufen 18 July 2013, Retrieved from: http://www.bfs.admin.ch/bfs/portal/de/index/theme n/16/04/key/approche_globale.indicator.30101.301 .html

Böni, H., & Wäger, P. (2015). Rückgewinnung von kritischen Metallen wie Indium und Neodym aus Elektronikschrott auf der Stufe der manuellen und mechanischen Vorbehandlung. In Recycling und Rohstoffe (Bd. 8). Neuruppin: Thomé-Kozmiensky, K.; Goldmann, D.

Buchert, M., Manhart, A., Bleher, D., & Pingel, D. (2012). Recycling critical raw materials from waste electronic equipment. Oeko-Institut e.V.

Buchert, M., Schueler, D., & Bleher, D. (2009).

Critical metals for future sustainable technologies and their recycling potential. Freiburg: Öko-Institut.

Chen, W.-Q., & Graedel, T. E. (2012). Dynamic analysis of aluminum stocks and flows in the United States: 1900-2009. Ecological Economics. 81, 92- 102.

http://doi.org/10.1016/j.ecolecon.2012.06.008

Erdmann, L., & Behrendt, S. (2011). Kritische Rohstoffe für Deutschland (Abschlussbericht, inkl. Anhang). Berlin: KfW Bankengruppe.

GfK. (2013). Verkäufe von Informations und Kommunikationstechnologien und Unterhaltungselektronik. In der Schweiz. Hergiswil, Schweiz: GfK Switzerland AG.

GfK Retail and Technology. (2007). IT-Market Switzerland, Total Market Estimations. Hergiswil, Schweiz.

GfK Retail and Technology. (2008). Telecom, Total Market Estimations Switzerland. Hergiswil, Schweiz.
Liu, G., Bangs, C. E., & Müller, D. B. (2011). Unearthing potentials for decarbonizing the U.S. aluminum cycle. Environmental Science and Technology, 45(22), 9515–9522. http://doi.org/10.1021/es202211w

MoE and METI. (2010). Report of the Study Group for Less Common Metal Recovery from Used Small Electric Appliances 2009. Ministry of Environment and Ministry of Economy, Irade and Industry.

Müller, D. B., Wang, T., Duval, B., & Graedel, T. E. (2006). Exploring the Engine of Anthropogenic Iron Cycles. Proceedings of the National Academy of Sciences, 103(44), 16111–16116. http://doi.org/10.1073/pnas.0603375103

Müller, E., Hilty, L. M., Widmer, R., Schluep, M., & Faulstich, M. (2014). Modeling Metal Stocks and Flows: A Review of Dynamic Material Flow Analysis Methods. Environmental Science & Technology, 48(4), 2102–2113. http://doi.org/10.1021/es403506a

Murakami, S., Oguchi, M., Daigo, I., & Hashimoto, S. (2010). Lifespan of commodities, part I: The creation of a database and its review. Journal of Industrial Ecology, 14(4), 598–612. http://doi.org/10.1111/j.1530-9290.2010.00250.x

OriginLab. (2014). Origin 9.1. Northampton, MA.

Remy, C., & Huang, E. M. (2015). Addressing the Obsolescence of End-User Devices: Approaches from the Field of Sustainable HCI. In L. M. Hilty & B. Aebischer (Hrsg.), ICT Innovations for Sustainability. Springer International Publishing. 257–267.

Sander, K., Schilling, S., Marscheider-Weidemann, F., Wilts, H., von Gries, N., & Hobohm, J. (2012). Abfallwirtschaftliche Produktverantwortung unter Ressourcenschutzaspekten, Meilensteinbericht August 2012. Hamburg, Germany: Ökopol GmbH, Fraunhofer Gesellschaft ISI, TU Darmstadt, TU Hamburg-Harburg.

SCEA. (2005). Verkäufe von Unterhaltungselektronik in der Schweiz. Zürich, Schweiz: Swiss Consumer Electronics Association.

Stocker, R., Bleiker, S., Allibone, A., Albiez, T., Kurer, A., & Streicher, M. (2013). Stocks and Lifespans elektronischer Geräte. Brugg, Switzerland: Fachhochschule Nordwestschweiz, Hochschule für Technik.

Swico, SENS, SLRS. (2013). Fachbericht 2013. Zürich.

Wäger, P., Widmer, R., Restrepo, E., & Fernandes, M. (2014). Seltene Metalle in Fraktionen aus der maschinellen Aufbereitung von WEEE. Unveröffentlicht. Bern, Schweiz: Bundesamt für Umwelt.

Weiss, R. (2013). Weissbuch 2013. Männedorf.

Yoshimura, A., Daigo, I., & Matsuno, Y. (2013).
Global substance flow analysis of indium. Materials

Transactions, 54(1), 102–109.
http://doi.org/10.2320/matertrans.M2012279


Sign up for plate updates