1. Introduction
Information and Communication Technology and ‘Entertainment and Media’ are two of the fastest growing industries and a future is foreseen where almost all electronic devices are connected [
1,
2]. The annually shipped number of mobile devices can be counted in billions. As such smartphone sales are currently around 1500 million units and tablets around 350 million [
1]. Smartphones usually have a relatively short operating lifetime of 2 to 4 years which negatively influences eco-environmental impacts such as abiotic resource depletion (ARD) and climate change (CC) [
3].
Life Cycle Assessment (LCA) is a systematic analytical method and model by which eco-environmental effects for product systems can be estimated with a precision of around an order of magnitude [
4]. The applicability of LCA for electronics has been questioned due to numerous constraints like the complexity of the manufacturing chain and short innovation cycles, making data collection difficult [
5]. However, several attempts to address the issues have been made for machine tools and laptops [
6,
7,
8,
9]. The overarching challenge addressed here is how to perform LCAs of millions of different products—in this case mobile phones—using streamlined approaches with good precision. There is an increasing global interest in developing and using Environmental Product Declarations (EPDs) to communicate life cycle based environmental performance of products [
10]. LCA is a cornerstone of any EPD. The common wisdom is that simplified LCA approaches do not have enough precision compared to Full LCA (FLCA) when applied to the rather complex life cycle of smartphones and tablets. The Product Environmental Footprint (PEF) method is expected to be the state-of-the-art for FLCA [
11]. PEF has very strict data quality requirements as product comparisons need good quality data. Ojala et al. [
12] argued that PEF Category Rules (PEFCR) developers should devote time to finding the most appropriate methodological choices.
1.1. Review of Prior Knowledge Observations
FLCA studies of smartphones are common [
4,
13,
14,
15,
16] so the major hot spots—e.g., for the Global Warming Potential indicator (GWPI) are well known as: integrated circuit (IC) production, screen production, use, and distribution. Depending on the midpoint indicator, the ARD indicator (ARDI) scores for mobile phones, tablets etc. is significantly influenced by indium if the products contain more than 1 mg of this metal [
17].
Studies comparing FLCA with simplified LCA have been published [
18,
19,
20,
21,
22], showing high usability of the streamlined approaches. Teehan and Kandlikar [
23] developed a parameterized formula for GWPI of production of electronics based on FLCAs. The formula is based on the masses of main parts such as battery, printed circuit board assembly (PCBA), and screen. It is uncertain whether this model can reflect the characteristics of quickly developing smartphone designs well enough, or if more complex forms of simplified LCA models are called for.
Further, Moberg et al. compared an FLCA for one mobile phone with five different simplification strategies for streamlined LCA of the same phone [
24]. They recommended that simplified LCAs of mobile phones should focus the data collection on energy use in production and use, raw material acquisition of specific metals, air transport, and key components such as ICs [
24].
Andrae and Vaija attempted to quantify the difference between two FLCA approaches for the same smartphone [
4]. They also mentioned that a previous version of eco-rating LCA (Section 15.3.1 in [
25]) produced a very similar GWPI score as the FLCA by Vaija for the smartphone at hand. However, recent FLCAs of smartphones are not comparable to older versions of eco-rating LCA [
25] because they use different GWPI data for several unit processes—e.g., screen production and use—than the present eco-rating LCA, discussed in
Section 1.2.
Andrae presented a short review of 14 LCA tools for consumer electronics [
26]. He concluded that the previous eco-rating LCA methods are fast, have a relatively high precision for GWPIs of mobile phones, give comparable results, give a specific result for each mobile phone, and they could be improved to mimic product category rules (PCR). Andrae claimed that eco-rating LCA has a high precision for GWPI scores compared with FLCAs [
26].
The International Telecommunication Union (ITU) published a baseline assessment framework and defined a minimum set of criteria to be considered when assessing the environmental performance of mobile phones [
27]. Especially Annex V in the ITU report [
27] discusses upstream life cycle based metrics of relevance for the present research.
1.2. Open Eco Rating LCA
Open Eco Rating (OER) is a recent rating of sustainability credentials for mobile phones [
28] aligned with the ITU Recommendation on eco-rating program for mobile phones [
27]. Product environmental impact is one important part of OER and a simplified LCA approach (OLCA) is used for the GWPI and ARDI calculations. The OLCA model equations are based on a large number of FLCAs performed by Orange since 2010. These LCAs were either carried out on entire products (in order to identify the key sub-assemblies) or on specific sub-assemblies such as cameras or PCBAs. The reasons for limiting the OLCA to include GWPI and ARDI are communication friendliness towards consumers and to some degree data availability. As the main purpose of OER and OLCA is customer information, commonly known indicators are considered appropriate. In this context, the French Environment and Energy Management Agency (ADEME) set up a web platform, Base IMPACTS
® [
29], being the official database of the French program for the environmental labeling of consumer goods. For mobile phones GWPI and ARDI were selected as they were deemed to be the most relevant, reliable and implementable by companies.
OLCA uses the midpoint indicators for CC and ARD as prescribed by the International Life Data System 2011 Midpoint+ version 1.08 (ILCD) [
30]. The similarities between OLCA and the SENSE tool for the food industry are obvious as SENSE is “a web-based tool which allows the environmental impact calculation for a food product in a simplified way” [
31].
OLCA is similarly intended to give a very good indication of the FLCA GWPI and ARDI scores of mobile phones. Further it is judged that tablets could also be estimated with OLCA after some development which reflects tablet product characteristics.
OER and OLCA are not analyzed much in the literature however Ercan et al. [
13] mentioned that there is an increasing interest in eco-rating as a way to provide product related sustainability information to customers. Andrae and Vaija [
4] suggested that eco-rating LCA methods [
25] show similar results as further simplified methods. Andrae et al. [
3] explained how eco-rating is linked to eco-design and argued that OLCA facilitates a balance between improved material efficiency and linearly increasing GWPI scores.
The most important product metrics in eco-rating LCA were defined as a result of FLCAs [
4] and therefore the GWPI scores obtained by OLCA are likely to have a high precision. In this research, the precision will be investigated quantitatively.
As product category rules (≈simplified LCA) is a trend in eco-design product policy [
32,
33], the findings from this research are important for decision-making on policy for smartphones and tablets.
In summary, no study is found that compares specific FLCAs with OLCA presenting details of GWPI and ARDI.
1.3. Objectives
This study researches to which degree those FLCA approaches with higher resolution and flexibility—using state-of-the-art LCA tools, ISO and ETSI standardized frameworks, primary LCI data and state-of-the art secondary LCI databases such as Ecoinvent and GaBi—differ from a rather simplified approach based on product metrics OLCA.
The difference is studied for five smartphones on the GWPI and for a tablet and one of the smartphones on the ARDI. The tablet has been included to understand development potentials for the OLCA.
It is expected that FLCA will score significantly higher than OLCA, especially on the ARDI of the tablet and one of the smartphones.
The overall question of this research is: How well do GWPI and ARDI results calculated with OLCA, based on metrics for products—here specifically for smartphones and a tablet—coincide with published GWPI and ARDI results based on FLCA for the same products?
An overall objective is also to show many details on how the OLCA model can be applied and developed further.
1.4. Hypotheses
Hypothesis 1 (H1). The FLCA scores are >50% higher than OLCA scores for Phones A–E for GWPI.
Hypothesis 2 (H2). The FLCA score is >50% higher than the OLCA score for Phone A for ARDI.
Hypothesis 3 (H3). The ARDI score based on metal contents of the tablet is >50% higher than the OLCA ARDI.
A comprehensive review of prior knowledge is given in
Section 1. Full disclosure of methods, data and other relevant information is given in
Section 2 as well as the validity and reliability of the data used, and the validity of the methods used. The conclusions provided in
Section 5 are consistent with the evidence.
4. Discussion
This research discusses the precision of a simplified approach—OLCA—for estimation of the carbon and antimony footprints of smartphones and tablets. On one hand individual practitioners want and need freedom to produce detailed and advanced FLCAs based on the ISO and ETSI LCA standards [
34], but on the other hand these FLCAs should not be compared by consumers. The main reasons are that not all assumptions are transparent and might differ, and that “apples are not compared to apples” in a strict sense.
While comparability is the ultimate aim of the PEF FLCA method, it will require very high data quality. Here instead OLCA is suggested for smartphones which will lead to comparability with enough data quality even before any PEF legislation has been enacted. The reason for this insight is that all assumptions are controlled within OLCA and that the FLCA and OLCA scores for GWPI and ARDI do not differ beyond recognition.
Ultimately comparability is sought by many smartphone consumers. Usually the price, performance and design of the phones are compared before the decision. Can the GWPI and ARDI scores of smartphones also be compared? The ETSI FLCA standard [
34] allows comparisons of GWPI and ARDI scores (and other LCIA scores for other impact indicators) between two external FLCAs if these four requirements are fulfilled:
- (i)
same calculation rules are used,
- (ii)
same LCA tool is used,
- (iii)
same LCI databases are used,
- (iv)
and a third party review is conducted.
OLCA fulfils all of these requirements as a common web tool is used with embedded calculation rules, the same GWPI and ARDI intensities are used for all unit processes, as well as the same algorithms for all, and finally the OLCA method is third party reviewed.
OLCA is not yet designed for tablets—this becomes apparent by the absence of a cobalt parameter. Currently tablets use relatively large amounts of cobalt in their batteries. Indeed, the current selection of metals in the OLCA method is based on the material content of two mobile phones determined from material content declarations from electronic component manufacturers and X-ray Fluorescence analysis. Cobalt was identified in this analysis to be just outside the top five contributors to ARDI.
Furthermore, OLCA merely estimates two midpoint categories and PEF compliant FLCAs need to estimate at least 15 [
11]. Prescribed weighting will be an interesting solution to this dilemma of multidimensional midpoint categories [
11,
47].
Besides, the suitability of OLCA for working out clear-cut ecodesign recommendations is somewhat dubious as specific supply chains are not analyzed. Be that as it may, general ecodesign suggestions are possible to extract, especially since the whole OER [
28] is much wider than OLCA.
By inserting all Equations of OLCA—e.g., Equation (10)—as parameters and intensities into the LCA tool SimaPro 8.2.3.0, and adding appropriate uncertainty ranges for each (
Table 4 and
Table 5), a suggestion of the 95% confidence interval is obtained via 100,000 Monte Carlo simulations as 33 to 43 kg CO
2—eq. for Phone B GWPI score in OLCA (
Figure 12). This suggests that the FLCA score of Phone B for GWPI has a high probability of being equal (41.8 kg) to the OLCA score for Phone B for GWPI.
It can be argued that the intensities and parameters/metrics of OLCA are ‘wrong’ compared with those in FLCA (e.g., I,ScreenA and Siarea). However, OLCA can conveniently be updated with the best practice to enhance its (already good) precision compared to FLCA. The updates will be crucial due to the rapid changes in material and processing technology that might occur. In that sense the camera model was developed and introduced in the OLCA system to improve its precision. The next evolution of the OLCA model could be to assess the environmental footprint of fingerprint scanners. If their impact is significant, a model could be designed with inputs such as the scanner technology (e.g., capacitive or ultrasonic e.g.,) or the sensor’s die area. Indeed a quick review of 20 recent mobile phones conducted within Orange showed that this component area ranges from about 20 to 80 mm², thus it might be possible to use it as a key parameter in an impact assessment model.
Furthermore, if e.g., thallium, is somehow started to be used in large quantities (for infrared optics or glass for e.g.,), and the OLCA is not updated, the error for ARDI will become unnecessarily large as thallium’s characterization factor for ARDI is about 80 times higher than the one for gold.
Additionally, within the 10 next years, Indium Tin Oxide (ITO) used in screens could be replaced by metal mesh, silver nanowires, silver nanoparticles, carbon nanotubes, Poly(3,4-ethylenedioxythiophene) (PEDOT), or graphene. Then indium might have to be removed from the ARDI part of OLCA and be replaced by other metals. Anyway, the smartphone industry can agree on which parameters OLCA should contain and also the most appropriate intensity values for the most important parts such as screens, batteries, cameras, and chargers. OLCA development is as such similar to the development of PEFCR or PCR.
OLCA is a natural predecessor to PEF, or even its replacement if PEF for some reason never happens for smartphones. The reason is that PEF LCA presumably has to be performed for all smartphones sold in the EU which will require streamlined approaches and PEFCR. Those PEFCR could, after some adjustment, be agreed as the OLCA.
Moreover, the GWPI result of Phone D [
15] shows a very strong similarity to OER—59.6 kg compared to 59.8 kg in OLCA—based on neither knowing
Siarea nor
I,
Siarea,
I,
ScreenA or
I,
Use. This comparison underlines the good precision of OLCA compared to FLCA for GWPI. However, Phone E diverges 20% from OLCA (
Figure 10) as the
Siarea is not transparent and the IC production has to be approximated with
SiNAND which is less precise than
Siarea.
The difference between the FLCA ARDI score for Phone B and the OLCA ARDI can partly be explained by metal recycling, which benefits of which are not yet considered in OLCA. Moreover, the characterization factors used for ARDI are different.
The FLCA ARDI score for Phone A also includes metal recycling [
13]. As shown in
Supplementary Materials Section 10, OLCA includes collection of 100% of all phones, transports, sorting and pyrometallurgical treatment and incineration of plastics and the GWPI scores of all. The credits for metal recycling and landfill emissions are not yet considered, neither are the storage/use of phones at the customers’ homes. So far, no clear way of including the recycling credits has been agreed for OLCA.
The present OLCA is global for two midpoint indicators and cannot mimic local FLCAs as well as it resembles global FLCAs. An OLCA, intended for the Chinese market, might have to focus more on end-of-life-treatment and other impact categories such as particulate matter.
The mass of tin used by Phone A is not possible to identify with Equation (8), but tin likely contributes to “Others” (see Figure 5 in [
13]). If the mass of tin used by Phone A is around 1200 mg, the OLCA ARDI score will rise to 1.36 g Sb—eq., however still far from 2. It is logical that ARDI deviation is higher than it is for GWPI. The reason is that the FLCA score is based on life cycle emissions whereas ARDI in OLCA is just based on the metal content, including neither losses in the upstream nor recycling in the downstream.
OLCA adopts the principle of “equal error margin for all” instead of “random error margins by all.”
FLCA is usually performed for one phone by one manufacturer, giving one result. FLCAs are also carried out at a relatively high cost and performed quite slowly. OLCA is performed for many phones by mobile network operators at a relatively low cost and fast manner. This research shows that the precision for OLCA is still quite high, at least for GWPI.
It can be argued that the sample of five smartphones and one tablet is too small to draw reliable conclusions, however, the quality of the included FLCA is judged to be relatively high.