Data quality assessment framework for critical raw materials. The case of cobalt

Graphical abstract


Applications
The description given in this section is specific for the applications studied in the research.

Portable batteries
Portable batteries are defined as sealed batteries that can be carried by hand, excluding industrial and vehicle batteries (European Commission, 2006). Cobalt is found in lithium-ion (Li-ion), nickel-cadmium (NiCd) and nickel-metal hydride (NiMH) batteries, being Li-ion batteries the most commonly used (Cobalt Institute, 2018). Li-ion batteries based on lithium cobalt oxide (LiCoO2, LCO) present the higher amount of cobalt (approximately 14%), while in NiCd, NiMH and in Li-ion batteries based on lithium nickel manganese cobalt oxide (NMC) and lithium nickel cobalt aluminum oxide (NCA), the metal is present only in minor amounts . The three type of batteries are rechargeable, and can be found in mobile phones, laptops, tablets, cameras, cutting tools, and other household products.
In 2017, around 23% of the global consumption of Co was in portable batteries (Darton Commodities Limited, 2018).

Mobility batteries
Mobility batteries are found in electric vehicles and hybrid electric vehicles (xEV), electric trains, electric buses, and electric bikes (Cobalt Institute, 2018). The battery stores electrical energy that the electric motor uses to power the vehicle (European Environment Agency, 2016). Batteries containing cobalt used in these devices are NMC, NCA, and NiHM batteries (Al-Thyabat et al., 2013;Darton Commodities Limited, 2018).
In 2017, 29% of the global consumption of Co was in mobility batteries. Around 1% was used in ESS (energy storage systems) (Darton Commodities Limited, 2018).

Hydroprocessing catalyst
Hydroprocessing catalysts are used to produce clean fuels and other usable oil feedstocks and products, upgrading oil fraction through the removal of impurities such as sulphur and nitrogen. By adding hydrogen, they also improve the properties and performance of the products (Darton Commodities Limited, 2018). The activity of the catalyst decreases in time, and when it drops below an economically determined level, the catalyst is removed from the reactor. The three main reasons for the loss of activity are coke deposition, poisoning, and sintering of the active phase. Depending on the type of deactivation, the catalyst can be regenerated, recycled, or sent out for disposal .
In 2017, 3% of the global consumption of Co was in hydroprocessing catalysts (Darton Commodities Limited, 2018).

Hydroformylation and GTL process catalysts
Cobalt catalysts are used in the production of different hydrocarbons, through the hydroformylation reaction and the gas to liquid (GTL) process. In hydroformylation, an alkene reacts with CO and H2 to produce aldehydes, which are then used in the production of alcohols (used in detergents and plasticizers), and carboxylic acids (used in pharmaceuticals) (De et al., 2013). The GTL process is commonly performed using Fischer-Tropsch synthesis (FTS), where syngas (CO and H2) react to produce liquid hydrocarbons, which are further processed into liquid fuels (e.g. gasoline, jet fuel, and diesel) (Dry, 1999).
Hydroformylation uses homogeneous catalysts (same phase as the reactants), while FTS involves heterogeneous catalysts (different phase from the reactants) (De et al., 2013;Jahangiri et al., 2014). In hydroformylation, the catalyst is recycled to the process. However, it tends to build-up on the walls of the reactor and on the tower packing, which makes it necessary to periodically remove it and send it to recycling (National Research Council, 1983;Hebrard and Kalck 2009). In FTS, the catalyst can be regenerated to be used in the same process or recycled for metal recovery (Brumby et al., 2005;Jahangiri et al., 2014).
In 2017, 0.5% of the global consumption of Co was in hydroformylation and GTL process catalysts (Darton Commodities Limited, 2018).

PET precursor catalyst
Cobalt-manganese-bromide (CMB) and cobalt-manganese-acetate (CMA) are liquid catalysts used for the oxidation of para-xylene (PX), in the synthesis of terephthalic acid (TPA) and di-methyl terephthalate (DMT). These compounds are raw materials in the production of polyester fibres articles, such as polyethylene terephthalate (PET) bottles, films, and paints (Joo et al., 2016;Darton Commodities Limited, 2018). CMB and CMA are homogeneous catalysts, which are recycled to the process. (National Research Council, 1983;Joo et al., 2016).
In 2017, 2.6% of the global consumption of Co was in PET precursors catalysts (Darton Commodities Limited, 2018).

Dissipative uses
Dissipative uses include applications of cobalt chemicals, from which the metal is not recovered. This category comprises pigments and drying agents; and agricultural, nutritional, and medical uses. Pigments and drying agents are applied in glasses, ceramics, refractories, driers, paints, and varnishes. In agriculture and nutrition, chemicals are used to correct cobalt deficiencies in soils and in animals. Medical uses includes the treatment of certain types of anemia, and as an antidote in cyanide poisoning (Donaldson and Beyersmann, 2012).
In 2017, around 8% of the global consumption of Co was in dissipative uses (Darton Commodities Limited, 2018).

Hard metals
Hard metals or cemented carbides are composite materials consisting of hard tungsten carbide (WC) particles bonded together by a metallic binder. This metallic binder is usually Co, although it can also be Fe, Ni, and other metallic phases (Freemantle et al., 2014;Cobalt Institute, 2018). Its uses comprises cutting tools and wear-resistant components in metalworking, mining, oil drilling, and construction industries. Cobalt is also used in diamond tools, applied as a binding agent to hold together wearresistant particles (in this case diamonds) .
The two main methods for the recycling of hard metals are chemical processes and the zinc process . The former considers the use of acids, electrochemistry or chemical modification techniques (Freemantle et al., 2014); the latter consists of the addition of Zn, which dissolves the binder phase of the cemented carbide without changing the composition of the material. Chemical processes produce material equivalent to virgin material, contrary to the Zn process that produces material to be used in the production of new hard metals .
In 2017, 7% of the global consumption of Co was in hard metals (Darton Commodities Limited, 2018).
In 2017, 3% of the global consumption of Co was in magnets (Darton Commodities Limited, 2018).

Other metallic uses
This category includes semi-conductors and integrated circuits, tool steels, and hardfacing and coatings. Semi-conductors and integrated circuits are contained in all modern electronic devices or systems (e.g. televisions, laptops, cameras, cell phones) (Cobalt Institute, 2018). Tools steel are used to work, cut and form metal components, for which they require high hardness and strength (International Molybdenum Association, 2018). Hardfacing and coating is the process where harder material is put onto to a base metal. It is used to increase the wear resistance of metallic components, or to refurbish a surface that is worn-down on used parts (A&A Coatings, 2018) In 2017, around 7% of the global consumption of Co was in other metallic uses (Darton Commodities Limited, 2018).

Superalloys
Superalloys are Ni, Fe-Ni, or Co based alloys, usually used at temperatures above 540 ºC (Donachie and Donachie 2002). High melting temperatures, and excellent creep, corrosion, and oxidation resistance characterize this type of material (Srivastava et al., 2014). Due to these properties, superalloys are used in a number of applications such as aircraft, rocket, and gas turbine engines; heat exchanger tubing; and nuclear reactors. Superalloys scrap is partly recycled for Co recovery; the rest is downcycled for steel production .
In 2017, 16% of the global consumption of Co was in superalloys (Darton Commodities Limited, 2018).

Parameters definition
Processing/manufacturing yield Ratio of usable output from the processing/manufacturing process to the input quantity, expressed as a percentage.
Processing/manufacturing scrap recovery Ratio of the processing/manufacturing scrap that is recycled to the process to the total scrap produced, express as a percentage.
Processing/manufacturing downcycled scrap Ratio of the processing/manufacturing scrap that is downcycled (for the production of low-end applications, such as steel) to the total scrap produced, express as a percentage.

Shape parameter (Weibull distribution)
The Weibull distribution is a continuous probability distribution, commonly used to assess product reliability and survival analysis, analyse life data, and model failure times. The function is characterised by three parameters: the scale parameter (normally denoted as α), the shape parameter (normally denoted as β), and the location parameter (normally denoted as τ). When τ is zero the function become the two-parameter distribution. The shape parameter determines the appearance or shape of the distribution (Lai et al., 2006).

Hoarding rate
Hoarding or hibernation refers to the dead storage of a product that is no longer in use anymore (Wilson et al., 2017). Here is understood as the hoarding of end-of-service (EoS) products.
This parameter is understood as the ratio of hoarded EoS products to the total EoS products produced in a year, expressed as a percentage.

Hoarding time
Hoarding time refers to the period in which the EoS product is hoarded. It is understood as the time between the EoS of the product until its collection (for recycling or disposal).

Non-selective collection rate
Non-selective collection rate is related to the misplacement of EoL products in waste bins.
This parameter is understood as the ratio of not-selectively collected EoL products to the total EoL products produced in a year, expressed as a percentage.

Collection rate
Waste collection is defined as "collection of solid waste from point of production (residential, industrial commercial, institutional) to the point of treatment or disposal" (Hoornweg and Bhada-Tata, 2012). However, the UNEP defined collection rate as the ratio of EoL products collected for recycling to the total EoL products produced in a defined period (UNEP, 2011). In the case of WEEE, the EU defined it as "the volumes collected divided by the average sum of EEE (electrical and electronic equipment) put on the market in the previous three years" (European Parliament, 2006).
In the research, collection rate is understood as the ratio of EoL products collected for recycling to the total EoL products produced in a year, expressed as a percentage.
It is noteworthy to distinguish between collection rate and recycling rate. The EU and the UNEP define the latter as the collection rate multiplied by the rate of recycling at the treatment facilities, assuming that the total amount of collected WEEE is indeed sent to treatment/recycling facilities (European Parliament, 2006;UNEP, 2011).

Pre-treatment efficiency
Pre-treatment covers several operations such as separation, sorting, physical processes, and chemical processes.
In the research this parameters is understood as the ratio of usable output from any pre-treatment operation (or a set of them) to the input quantity, expressed as a percentage.

Distribution to recycling processes
Waste and scrap can be recycled through different processes. This parameter refers to the distribution of waste or scrap to the different processes (how much of the total waste is recycled by one or another process), expressed as a percentage.

Recycling efficiency
Ratio of usable output from the recycling processes to the input quantity, expressed as a percentage.

DQA methods description
In 1996, Weidema and Wesnaes proposed a DQA method called Pedigree-matrix, which consisted of five independent DQIs: Reliability, Completeness, Temporal correlation, Geographical correlation, and Further technological correlation (hereafter referred to as Technological correlation). The indicator Reliability was related to the assessment of the sampling methods and verification procedures. The indicator Completeness defined how complete the available datum was in function of its statistical representativeness, the number of measurements in the sample, and the time periods for data collection. The indicators Temporal and Geographical correlation described the representativeness of the datum regarding its year of generation and the intended geographical area, respectively. The indicator Technological correlation evaluated the congruence of the available data and the targeted data with respect to technology, product, etc. In the matrix, each indicator was described by a score from 1 to 5, with 1 for the highest quality and 5 for the lowest quality. This matrix was established to be used in data quality management for life cycle inventories (LCI). Manfredi et al. (2012), and Edelen and Ingwersen (2016) established their own modified Pedigreematrix, to be applied likewise on LCI. Manfredi and colleagues established six DQIs: Completeness, Methodological appropriateness and consistency, Time representativeness, Technological representativeness, Geographical representativeness, and Parameter uncertainty. Edelen and Ingwersen defined DQIs at flow level (making a distinction between reliability and representativeness) and at process level. At flow level five DQIs were given: Reliability, Temporal correlation, Geographical correlation, Technological correlation, and Data collection methods. The latter four were linked to representativeness. For processes, two DQIs were defined: process review and process completeness. Laner et al. (2015) based their work on the Pedigree-matrix, applying the same five indicators, although modifying their definition to be applicable to MFA studies. Furthermore, an additional DQI was added, termed Expert estimate. This indicator was used alone, as a replacement of the other five. In this work, the DQIs were described by a score from 1 to 4, with 1 for the highest quality and 4 for the lowest quality. Table A1 present a summary of the comparison between the four described methods.

List of consulted experts
Fifteen experts from academia; 18 associations, groups or societies; and more than 110 companies (producers, users, and recyclers) were contacted for collection and/or validation of data. Of these, 51 gave a general answer. Table A14 lists the institutions that provided an expert insight about Co flows, and/or information used in data collection or validation. Some of these institutions and/or experts asked not to reveal their identity.