Accepted Manuscript

Gold recovery from thin film deposition facilities: environmental aspects of a novelmethod

Antonio Andrea Gentile, Claudio Rocco, Eng. Ph.D. Salvatore Modeo, Tecla Romano

PII: S0959-6526(14)00805-1

DOI: 10.1016/j.jclepro.2014.07.077

Reference: JCLP 4569

To appear in: Journal of Cleaner Production

Received Date: 31 March 2014

Revised Date: 28 July 2014

Accepted Date: 29 July 2014

Please cite this article as: Gentile AA, Rocco C, Modeo S, Romano T, Gold recovery from thin filmdeposition facilities: environmental aspects of a novel method, Journal of Cleaner Production (2014),doi: 10.1016/j.jclepro.2014.07.077.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

1

Title: Gold recovery from thin film deposition facilities: environmental aspects of a novel method Keywords: Palette Modular Device, Maintenance of Physical Vapour Deposition, Thin Film Deposition, Gold Recovery, Life Cycle Assessment Proposed by: Antonio Andrea Gentile, Claudio Rocco, Salvatore Modeo, Tecla Romano

University of Salento, Department of Innovation Engineering Campus ECOTECKNE, Via per Monteroni 73100 Lecce (LE), ITALY

Phone: +39 0832 297941

E-mail addresses: claudio.rocco@unisalento.it antonio.gentile@unisalento.it Corresponding Author: Claudio Rocco (Eng., Ph.D.) Word Count: 7031 NOTE Dear Board Members, The paper discusses a new original patented system to be used for maintenance of Physical Vapour Deposition (PVD) processes in order to support gold recovery activities. We also provide analyses of efficiency, compared to traditional techniques, in terms of operations to be carried out and environmental impacts. Considering the interest of many industries regarding material recovery methods and the implementation of novel technologies, the authors have selected your journal as both matching the scope of their paper, as well as addressing an audience of potentially interested readers. Authors hope their work will be considered a good proposal for publication. Best Regards

The authors

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

2

Abstract This paper discusses the application of a novel system for metal recovery from Physical Vapour Deposition (PVD) production waste. The recovery process enabled by the “Palette Modular Device” (PMD) avoids the overlapping of different film materials during the production phase, thus limiting the need for chemical separation and refinement of the wasted fraction of the film layers. A description of the new methodology is provided focusing on the particular scenario of multi-layer films (including a gold layer), which are commonly employed in microelectronic contexts. Furthermore, a standard recovery practice and the novel one are compared using a Life Cycle Assessment (LCA) methodology. LCA starts from the definition of the objectives and comes to inventories of all material and energy flows related to the different process stages. Moreover, it provides an impact assessment, which identifies and quantifies potential effects on the environment that is essential information for the interpretation of the results. This research demonstrates how the novel process exhibits interesting performances on different environmental impact indicators, giving a perspective of a "green" technology.

1. Introduction Electrical and electronic devices are composed of a wide range of components such as metals, plastics and other substances. Some of these have a significant economic value (precious metals), others are very harmful to the environment and human health (such as Mercury and Cadmium), and still others may present both of these characteristics (e.g. Lead). For example, a mobile phone can contain more than 40 metallic elements including precious metals (e.g. Silver, Gold, Palladium), rare earths (e.g. Cobalt, Indium, Antimonium) and other metals (e.g. Copper, tin). In particular, if we take into account only precious metals and copper, a mobile phone can contain: 250 mg of Ag, 24 mg of Au, 9 mg of Pd, and 9 g of Cu (Hagelüken and Corti, 2010). At first glance, these values may not seem particularly significant; however, if worldwide sales of electronic devices are considered, these quantities become remarkable. Results of a study conducted in 2011 for market research showed how worldwide shipments of personal computers reached 353 million units (Gartner, 2012). Therefore, starting from an average content of gold in a single PC of 220 mg (Hagelüken and Corti, 2010), 77 tons ca. of gold can be estimated to be contained in all computers sold in the year 2011. Analyzing these data by an environmental point of view, the production of 1 ton of gold, platinum or palladium can generate more than 10 thousand tons of CO2 during the extraction processes, whereas the production of 1 ton of copper produces the emission of 3.4 tons of CO2 (Hagelüken and Meskers, 2010). Furthermore, in metal mining processes, gold requires large amounts of sulphuric acid (already in 1991 the amount was 816.7 tons (U.S. EPA, 1995)) that triggers the phenomenon of acid rain. It is also useful to remember that gold is an example of a non-renewable raw-material and its extraction may become unsustainable in the near future. Electronic integrated circuits are very complex and the requirement of high reliability has enhanced the use of gold for its high thermal and electrical conductivity combined with its resistance to weathering, oxidation or corrosion (lack of these properties has limited the use of silver and copper for the same purposes). Moreover, gold has characteristics difficult to be replaced for specific purposes; in fact it is the material primarily used in ohmic contacts and connectors especially when they are subjected to low currents and voltages. Figure 1 shows the wide usage of gold as a key material for industrial and electronic applications in the last four years.

Figure 1: Tons of gold used in technology (World Gold Council, 2014)

In this research, gold is studied for two reasons: i) it is a “critical” raw material used in electronics for its unique properties and ii) its extraction is a good example of the polluting mining industry (Kumah, 2006). In addition, research programs focused on the use of this material (e.g. the use of gold for nanotechnology (World Gold Council, 2010) and a

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

3

strategic use of gold (due to innovative applications in industrial environments (World Gold Council, 2014)) can be observed. In this context, the present research attempts to contribute by evaluating the environmental impacts of the Gold recovery process from production wastes of the thin-film industry (a sector which has observed an ever growing number of novel applications involving gold) according to a LCA perspective. LCA studies about production wastes deserve to be analysed, in fact scientific literature focuses often on the materials extraction phase (Coelho et al., 2011) and on the recycling of electronic goods and scrap (Hagelüken and Corti, 2010). In particular, fundamental steps of a standard hydrometallurgical refinement process and a technology recently introduced1 are compared. The paper is organized as follows: in Section 2, a review of the background literature is presented, focusing on prior research of gold recovery. The production process addressed in this study (i.e. Physical Vapour Deposition) is also introduced; Section 3 is dedicated to describing the research question and explaining the methodology used for carrying out the study; Section 4 is focused on the case study, where a standard gold recovery process from production scrap is compared with a novel one by discussing their global environmental impacts; and Section 5 is dedicated to conclusions and future outlooks.

2. Background: Physical Vapour Deposition Film deposition is the superimposing of a layer of a material on the surface of another (“substrate”). A thin film has a thickness of a few microns at most (Mattox, 2010). Different kinds of films can be deposited, each one with certain characteristics that make it suitable for specific applications. The objective of the thin film deposition is to take advantage of the surface properties of a target material (such as electrical conductance/insulation, abrasion resistance, etc.), using it as a coating and, at the same time, keeping the properties of the substrate material (such as robustness, low cost, etc.). Among available techniques for thin film deposition, the “atomistic overlay” is widely used. Here, the target material is deposited by successive small quantities without penetrating deeply into the substrate. The atomistic deposition can be implemented according to different procedures (Mattox, 2010):

- PVD (Physical Vapour Deposition); - CVD (Chemical Vapour Deposition); - Electrolytic/Galvanic deposition; - Other processes with limited industrial adoption (Arc Deposition, Pulsed Laser Deposition, etc.).

In particular, PVD employs physical transport (ballistic or diffusive) of the species to be deposited and it is the method considered for this study. PVD is particularly appropriate to produce thin films of metallic materials in a rapid and relatively cheap way, but also without the use of complex chemical precursors (Bunshah, 1994). For this reason, it is widely used for the deposition of metal tracks in microelectronics (Rossnagel, 1999). Evaporation and sputtering are the most common practices for PVD. The main differences are the source of energy that vaporizes the target material and the type of environment where the process takes place. In evaporative techniques, the emission from the target material occurs by thermal runaway. The need to minimize unwanted contaminations and to have a sufficiently long mean-free-path of the emitted particles requires the use of chambers in Ultra High Vacuum.

Figure 2: Schema of a general PVD process

Within the same chamber, several depositions of different materials may occur successively, activating a single target material for each step, or sometimes even simultaneous targets (co-evaporation). Sputtering, instead, requires the aid of

1 Recently patented by MRS, a R&D company

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

4

plasmas for the atomistic extraction from the target, upon which ions of gas (usually inert) collide. Then, the process of extraction and deposition occurs at low pressure. Most of the energy associated to the sputtering process results in a heating of the target (the thermal agitation of the atoms of the target increases). The remaining part, instead, is transferred to the motion of atoms ejected from the target lattice. The material ejected from the target will condense on the substrate, which is placed in an appropriate position to collect most of the emitted particles.

2.1 Limits of PVD technique Deposition techniques mentioned above are subjected to some limitations. First of all, the uniformity of the film improves with the increasing of the distance between the target material and the substrate (long “throw”) and according to the ratio between their “lateral dimensions”. Moreover, in order to avoid overheating of the substrate, a minimum throw distance of a few centimetres is required (Mattox, 2010). Finally, in techniques of evaporation, the deposition process is not completely directional and it is very complicated to focus beams of neutral particles emitted (e.g. in “directed sputtering” gas flows or auxiliary facilities honeycomb are used). The first two limitations push towards the adoption of chambers with large dimensions (in general, evaporative chamber dimensions are, at least, five times bigger than the largest dimension of the substrate). In parallel, the a-directionality of the process leads to cover partially the walls of the chamber with deposited films (Mattox, 2010) (Bunshah, 1994), which are characterized by a material thickness of the same order of final film on the substrate2. As it is well known, this combination of effects produces a high percentage of target material deposited on the walls of the process chamber. In fact, each evaporation process is characterized by the loss of a great fraction of the raw input material (i.e. it can rise more than 90% during e-beam PVD (Fuke et al., 2005)) and an efficiency that rarely exceeds the 40% in industrial practice3. Therefore, severe recovery procedures are required to reduce production wastes and to avoid “flaking” phenomena, which may compromise successive depositions (by contamination from previously employed target materials (Mattox, 2010)).

3. Methodology

3.1 Physical vapour deposition maintenance by Palette Modular Device State-of-art maintenance of PVD facilities involves the adoption of static screens: shields made of vacuum-compatible materials that are mounted inside the deposition chamber. These collect most of the deposited material, which does not condense on the substrate, thus protecting chamber walls. If static screens are adopted standard maintenance involves abrasive removal techniques4 for screen cleaning (more difficult maintenance is involved when no screen is used since the walls are exposed to the flux of particles). Whenever the chamber is used for multi–material depositions, these techniques produce a powder, which is a solid mixture of those same metals composing the overlapped film layers on the substrate. The difficulty to recover this “wasted” fraction is mainly due to the separation of these metals, which requires complex chemical procedures. The Palette Modular Device (PMD) (Gentile and Modeo, 2012) is a patented system5, which tries to overcome the main drawbacks of current maintenance techniques for PVD equipment. In particular, its main characteristics are:

- Removable rotating palettes - exposed to the deposition materials - that screen and protect the surface of the Vacuum chamber;

- Non-abrasive removal techniques to clean screen surfaces: this reduces or even avoids the need to recondition the chamber after maintenance procedures;

- Good adhesion between screening surfaces and deposited films, in order to reduce spontaneous unrolling and flaking of deposited layers due to mechanical stress;

- Preservation of appropriate standards for use in UHV (Ultra High Vacuum). The PMD is able to avoid the formation of the metal mixture6 through the adoption of a dynamic screening: its removable palettes are able to collect different materials used during the PVD process selectively. In fact, different materials are collected on different sides of palettes7 uniformly. Moreover, each side is covered by a layer of “sacrificial polymer”, which can be easily removed by the action of light solvents, when palettes are removed and cleaned, avoiding any abrasive procedure. This facilitates operations for the recovery of valuable materials. Further features of this system can be observed in Figure 3a and 3b (Gentile and Modeo, 2012).

2 Precisely, thicknesses follow a cosine-like law in evaporation, and slightly more complicated patterns in sputtering facilities (Corti, 2002) 3 As emerged by private communications of the authors with technical staff from few PVD equipment suppliers. 4 E.g. sandblasting, bead-blasting, etc. 5 Patented with ID US 20130008375 A1 by MRS™ 6 See par. 3.2 7 For example, in the following of this paper, we will make the hypothesis that gold forms a uniform layer on one of the palette sides.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

5

Figure 3a: Prototype of PMD integrated in a chamber Figure 3b: CAD model of the PMD system

It has been calculated how the adoption of this system can outperform traditional screens, collecting about 70% of the material currently wasted8 and enabling a different strategy for recovery. Furthermore, the new technology may reduce the risk of inhalation of dangerous particulate for operators who perform abrasive cleaning activities on site. 3.2 Gold recovery processes In this study, the test case analysed for the maintenance process presents the following set up:

- The chamber is protected by a screening surface of 1 m2, which is a typical dimension; - Screen maintenance is required after thirty production cycles; - The screen has a lifetime of 1000 maintenance cycles before its replacement (i.e. Iron in traces has been

considered in the maintenance outputs, due to the erosion of the screen); - The chamber is employed for multi-layer depositions, involving Si and GaAs/GaN substrates (Oktyabrsky and

Ye, 2010); - Therefore, materials used mainly in deposition processes are: Gold and Aluminium as contact metals

(respectively 35% and 8% of “wasted material” on the screen), Titanium and Chromium as adhesion layers (respectively, 30% and 25%) (Baenard et al., 1996) (Sze and Ng, 2006), and negligible traces of other metals.

Remarkably, this is a typical set-up for evaporation chambers used in R&D and pre-production facilities of microelectronics industries. Based on a literature review and several discussions with a number of PVD adopters9 (further references and considerations are included afterwards in the paper), the “standard scenario” is characterized by the following phases. The first phase involves the sandblasting of the static screen surface (by means of silica). After the blasting operations, powders obtained are fused in a melting pot, which is heated by flame (LPG/methane) or by induction in small recovery facilities. Subsequently, the molten metal is poured into a tank of water, where metal flakes are formed. At this point, the refining steps begin and numerous different methods can be adopted, each with their own set of advantages and disadvantages (Corti, 2002) including: cost, safety, pollution and health issues for employees at their workplaces (Pitt, 1999). For the standard recovery processes, the dissolution with aqua regia is a widely used technique (Sheng and Etsell, 2007). This substance is a mixture composed of one part of nitric acid and three parts of concentrated hydrochloric acid that oxidizes Gold and dissolves it as a chloride. The practice suggests to dissolve scraps of gold alloy by means of a series of little additions of aqua regia in order to use only a small excess of acid and without leaving residues of undissolved gold. When all the gold is dissolved, the solution has a greenish-yellow colour (chloroauric acid), while the other metals remain on the bottom together with the non-metallic materials such as abrasive grains. At this point, the process of recovery proceeds with the filtering of the solution and the cementation of the metals deposited. The Gold in solution is subjected to precipitation with sodium bisulphite (Brug and Heidelberg, 1974) or other selective salts/resins. The precipitation of gold must be done with mechanical ventilation under a fume hood in order to allow the evacuation of sulphur dioxide generated during the process. Then, the process continues with the filtration of the solution under

8 E.g. the PMD may recover 1.2 g ca. of material within a single deposition process that produces 300 nm film-thickness on the substrate in a standard R&D evaporative chamber (lateral dimensions ~ 0.5 m). 9 See footnote 4

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

6

vacuum and the successive siphonage. The gold remained on the filter is subjected to several washings with hot and cold water, before being sent to the final fusion in ingots, that are characterized by a nominal purity of 99.9 %. Focusing on the workflow related to the new technology, instead, it is enabled by the installation of a PMD in the deposition facility. Gold is deposited only on one side of each palette (side “A”), while the rest of used metals are settled on the back (side “B”) due to the rotation of palettes during the deposition. We suppose to adopt again a blasting process for side B of each palette, where the reduction of the total thickness to be removed allows an estimated decrease of 35% of the time necessary for this activity. Metals produced by this step are considered wastes as in the previous case. The gold recovery is performed by etching the sacrificial polymer (previously applied on the side A of the palettes) with an acetone bath, using a sonicator10 to make its dissolution more rapid and complete.

Figure 4a: Gold deposited on palettes Figure 4b: Employment of sonicator during etching

Figure 5: Gold recovered from etching phase

After the gold recovery procedure, palettes have to be “regenerated”, so a new layer of sacrificial polymer must be applied.

4 Simplified LCA for the recovery process In this section, processing steps where critical environmental impacts may occur are identified, along with information needed to implement feasible improvements. In addition, it should be considered that the focus of this research is mainly on the activity of gold recovery and the analysis is carried out at “process level” (the boundaries are defined in section 4), using a simplified approach. Generally, a “simplified LCA” does not take into account the whole lifecycle of a product/process/service, but offers a sufficient detail for a good analysis. This approach has the same purpose and the same structure of a detailed LCA, but simplifications are made to reduce significantly the time required to perform the study, in fact only selected impact categories have been considered in the case study (Goglio and Owende, 2009; Todd and Curran, 1999; Arena et al., 2013). Although simplifications have to be made carefully, because these may compromise the validity of the research, the streamlined methodology applied in this research is expected to be sufficient for the description and comparison of

10 The sonicator is a device that generates ultrasonic waves in a liquid medium, creating microscopic vacuum bubbles, which rapidly expand and compress themselves. This phenomenon is called cavitation, and acts as high speed miniature brushes.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

7

the scenarios taken in consideration. This is due to the equivalence of process phases preceding and following the boundaries set in this study and it allow us to focus on few operational steps neglecting lifecycle analyses about other steps related to the production equipment.

4.1 Definitions of boundaries and scope As emphasized in the introduction, the goal of this LCA study is to compare two practices for gold recovery from PVD shields maintenance scrap. In particular, one practice is the aqua regia dissolution (that is traditionally adopted by SME recovery industries), whereas the other is an innovative method enabled by the adoption of a PMD (as described in sections 3.1 and 3.2). The final purpose is to investigate which approach has a greater overall impact on the environment among these two. This case study is a small step of a rather complex context and it presents some difficulties in the collection of those data needed for a global study. Therefore, the definition of process boundaries focuses exclusively on the recovery of a single material (Gold) from the maintenance of a chamber, instead of the whole deposition process. In this way, we are allowed to exclude all those steps and factors that can be supposed identical (or rather similar) for the two recovery practices considered. These include: extraction, refinement and transportation of raw materials; operating impacts; production, disposal of deposition facilities11 and of the goods embedding the thin-film structures produced. The focus on the PVD shield maintenance has been further refined, excluding those few steps having intuitively a very low environmental impact12 or equivalent for the two modalities (such as: the scrap transport or the fusion of gold to be recycled). Figures 7a and 7b specify the boundaries of the analysed steps for both maintenance practices.

Figure 7a: Boundaries of standard recovery scenario Figure 7b: Boundaries for recovery by PMD

The boundaries are consistent with the purposes of this study: the environmental comparison of two recovery systems instead of a full LCA analysis of a production process. The functional unit (referred to the process set-up as described in Par 3.2) is defined as follows:

11 With or without the PMD system embedded 12 E.g. consider the unmounting of the static shields/PMD, which can be done by hand.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

8

- Standard CASE: a screen of 1 m2 surface13 covered with a thickness of deposited material of 10 µm, corresponding to 100 g ca.;

- PMD CASE: a shield surface of 1 m2 covered by a total thickness of deposited material as in the standard case, although the material is separated in two parts: about 30% on the A side of palettes (where most of the gold is deposited, according to the operating principles of the PMD system) and the remaining 70% on the B side.

4.2 Inventory analysis This stage tries to define the flow of material input, energy used and emissions in the environment of each step of the recovery methods that are examined. Please, note that only emissions into the atmosphere were taken into account for both scenarios, excluding effects of any exhausted chemical reactants improperly disposed. This is why we focused on the environmental impacts that are more strictly related to the health of operators who carry out the process of chamber maintenance and gold recovery activities. Emissions in the atmosphere are the most dangerous from this point of view. Even if occurring under a fume hood, in fact, it is possible for some vapours/gases to be dispersed in the environment. Furthermore, the hoods normally provide filtering and dispersion of the vapours/gases collected, but catalytic decomposition or transformation is very rare in the field. This is true especially with the basic components (i.e. CO2, NO2, SO2, Acetone, etc.) which are the main outputs considered for this study.

4.2.1 Inventory for the standard recovery scenario

SANDBLASTING The activity of sandblasting is performed using silica sand. Data were provided by Norblast14. Standard equipment provided by the company is able to treat areas of 1 m2 in 30 minutes, using 1.5 kg of sand. Blasting allows operators to clean a screen from superposed films and to get the mixed powders to carry out the recovery. From the analysis of data by US EPA15, 1.5 kg of silica sand produces 20 g of PM10 and 2 g of PM2.5 (U.S. EPA, 1998).

FUSION The sand contaminated by metal powders is subjected to a pyrometallurgical process in order to separate metals from inert materials. This process can take place in induction furnaces or with the use of fuels such as methane or LPG. In this study, we assume that the fusion is accomplished in a LPG oven. As an example of small gas ovens used in the jewellery industries, we have used data supplied by the Legor Group16. This furnace may load up to a maximum of 6 kg of material and it is equipped with a burner (28000 kcal/h power) with a maximum consumption of 2.3 kg/h of LPG. Clearly, the fusion process generates ashes mainly composed of silica (1.45 kg) that are considered completely inert, in the absence of reliable information. The fused metal is poured into tanks of cold water to form flakes. This last operation has no significant environmental impact and, therefore, it will be omitted.

AQUA REGIA DISSOLUTION At this point, flakes of gold are dissolved in aqua regia according to the following chemical reaction (Syed, 2012):

Au(s) + HNO3(l) + 4HCl(l) → AuCl4-(aq.) + H+(aq.) + NO(g) + 2H2O(l)

the aqua regia loses effectiveness quite quickly, so it is necessary to use a quantity much higher than the stoichiometric one. Therefore, a safety factor17 of “2” was taken into account to calculate the quantity of reactants for a complete dissolution of gold in the powders.

PRECIPITATION The next step is the extraction of gold from the solution by a selective chemical reduction. There are many suitable reducing agents for this purpose: ferrous sulphate, sulphur dioxide gas, oxalic acid, hydrazine, formaldehyde, sulphite and sodium bisulphate and so on (Corti, 2002). In this study, as an example, we examine the precipitation with sodium bisulphite, according to the following reaction (Yannopoulos, 1991) (Pacławski and Fitzner, 2004):

2HAuCl4 + 2NaHSO3 → 2Au + 8HCl + Na2SO4 + SO2

Also for this step, we will include in the final reports only the contribution from pollutants releases as gases or vapours (i.e. hydrogen chloride and sulphur dioxide), whereas waste solutions are not included in the boundaries of the scenario.

13 (typical size of the inner surface of an evaporator Bazers, class BAK 600 or compatible used in small and medium size companies) 14 “Norblast” is an Italian enterprise that produces sandblasting and beadblasting equipment 15 Data adapted by table 13.2.6-1 Particulate Emission Factors, AP-42 Supplement D, US EPA 1998 16 Catalogue M&T http://www.legor.com 17 That is, the same factor recommended by reports from the recovery of gold in the jewelry industry

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

9

Table 1: Inventory of standard recovery scenario

4.2.2 Inventory for PMD scenario

SANDBLASTING The case involves the blasting operation only on the palette side where there is no presence of gold. The operation, in fact, is considered only for the maintenance of the PVD chamber, but it plays no role in the recovery of the deposited material. As already mentioned above, the PMD allows a reduction of 35% of the time required by the standard blasting step, due to the smaller thickness of material to be removed, which results in a corresponding reduction of abrasive agents and energy required.

ETCHING with ACETONE and SONICATOR The A-side of a palette is rather subjected to the procedure for recovery of gold through a process of etching in acetone with the use of a sonicator. From data provided by MRS and considering the particular functional unit, preliminary tests register the employment of 3 litres ca. of acetone, which have to reach a temperature range of 70-80 °C. This step requires 0.21 MJ of energy and it can be aided by the usage of a sonicator. In particular, according to the specifications provided by the Scientific Instrument Services, a sonicator characterized by a tank of 6.5 litres, a power of 200 W, operating for 2 hours, requires 1.44 MJ of energy. Therefore, the etching stage provides a total consumption of 1.65 MJ of electricity, which corresponds to the production of about 197 g of CO2. Acetone is a solvent characterized by a high evaporation rate (vapour pressure is 24.46 kPa at 20 °C) and this factor, which is associated to the high temperatures to which it is subjected, leads to the production of harmful vapours, corresponding to approximately 70% of the total quantity of acetone used. For this reason, the high-temperature phase of the etching occurs in a closed recipient and the whole process is carried out under a hood using forced aspiration. However, as in the case of the precipitation with bisulphite in the standard scenario, eventual catalytic treatment of emissions will not be considered in this study.

PALETTE REGENERATION In this phase, the preparation of the sacrificial polymeric layer on the palettes is analyzed. A sacrificial layer of a commercial photopolymer18 is considered for this study19. The quantity of polymer required is approximately 242 g for the functional-unit in order to produce a coating with a thickness in the range 30 - 50 μm. Volatility considerations (they are based on the vapour pressure characteristics of each component) allow estimating the portion that will evaporate during the preparation of the sacrificial layer. In particular, at room temperature (20 °C) dimethyl ether, naphtha and acetone have a rather high vapour pressure (respectively: 531 kPa (Hart, 2007), 585.2 kPa (Imperial MSDS), 24.7 kPa

18 The photopolymer is supplied by RS Components Ltd. 19 Given the trade secret of MRS about its sacrificial layer proprietary formulations, we have here made use of a commercial equivalent, exhibiting similar behavior in the etching process, as a reference.

SANDBLASTING FUSION AQUA REGIA DISSOLUTION PRECIPITATION

Input Input Input Input

Sand SiO2 1.500 kg Contaminated Sand 1.600 kg Metal flakes 0.150 kg Gold in Solution

(HAuCl4) 0.061 kg

Energy 42 MJ LPG 0.150 kg HCl 0.104 kg NaHSO3 0.038 kg

HNO3 0.045 kg

Output Output Output Output

Products Products Products Products

Clean Screen 1 m2

Melted metal 0.15 kg

Gold in

Solution

(HAuCl4)

0.061 kg

Gold 0.035 kg

Contaminated

Sand

1.600

kg Solid waste 0.090 kg Wasted Solution 0.019 kg

Water 0.013 kg Na2SO4 0.013 kg

Emissions Emissions Emissions Emissions

CO2 5.040 kg CO2 0.450 kg NO2 0.017 kg SO2 0.006 kg

PM10 0.020 kg Hashes 1.450 kg Cl2 0.038 kg HCl 0.026 kg

PM2.5 0.002 kg NO2 0.016 kg

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

10

(Howard, 2011)). Therefore, we can consider a full evaporation of these components already in the preparation phase. Relying on the low vapour pressure of the other components, instead, it can be supposed that they will contribute to form a solid layer, which will be etched in acetone when palettes are exhausted and replaced. Regenerated palettes will be rearranged in the vacuum chamber at the end of the maintenance. Input and output quantities of the present step are displayed in table 2:

Table 2: Inventory of PMD recovery scenario

4.3 Impact Assessment The phase of the life cycle inventory analysis allows researchers to identify flows of energy and resources, occurring within the system boundaries, and to provide data for the phase of “Impact Assessment”. Before assessment step, categories of impact to be considered must be defined. On the basis of the purposes of this study, only emissions in the atmosphere are selected. For these impact categories, “CML 2001” method is used in the following as main reference. This characterization was developed by the University of Amsterdam-Leiden and it focuses on the following impact categories: consumption of abiotic resources, acidification, eutrophication, climate change, depletion of the stratospheric ozone layer, human toxicity, eco-toxicity and photochemical smog (Hischier et al., 2010). From a first qualitative evaluation, the following CML 2001 categories were selected as the most important for the processes considered in this study:

• Global Warming Potential (GWP); • Acidification Potential (AP); • Human Toxicity Potential (HTP); • Photochemical Ozone Creation Potential (POCP).

In the following, we will focus on the calculation of relative impact indicators for each of these categories. Notice that the substances involved in the inventory play a negligible role in most other categories of the CML 2001 framework (i.e. marine and freshwater EcoToxicity, terrestrial EcoToxicity, Ozone Depletion, Abiotic Resources Depletion20 and Eutrophication21). This is the reason we decided to exclude these additional categories from the study presented here.

4.3.1 Classification Table 3 summarizes a classification of substances involved in the inventory according to different impact groups.

Emission GWP AP HTP POCP

CO2 X

NO2 X X X

SO2 X X X

20 For these categories, CML 2001 database does not give any impact related to the substances in Table 3. 21 The only contribution for this category is given by NO2, in particular the eutrophication potential is 0.0043 (calculated in kg of PO4 eq.) in the “standard scenario” and 0 in the “PMD scenario”. Therefore, the difference among the two scenarios is negligible also for this category with respect to impacts in Tables 4 and 5.

SANDBLASTING ETCHING (ACETONE) -

SONICATOR PALETTE REGENERATION

Input Input Input

Sand SiO2 1 kg Acetone 0.949 kg Cleaned Palettes 1m2

Energy 27.3 MJ Energy 1.520

MJ Polymer 0.242 kg

Polymer 0.041

kg

Output Output Output

Product Product Product

Cleaned Screen 1 m2

Cleaned Palette 1 m2 Regenerated

Palettes 1 m

2

Contaminated Sand 1.040 kg Gold 0.035 kg

Emissions Emissions Emissions

CO2 3.276 kg CO2 0.183 kg Dimethyl ether

vapours

0.100 kg

PM10 0.013 kg Acetone vapours 0.047 kg Acetone vapours 0.100 kg

PM2.5 0.001 kg Acetone-polymer 0.943 kg Naphtha 0.001 kg

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

11

HCl X X

PM10 X

PM2.5 X

Acetone X

Dimethyl ether X

Table 3: Classification of substances according to environmental impact categories

4.3.2 Characterization This step, which is known as “Characterization”, quantifies impacts of substances in the inventory for each category.

STANDARD SCENARIO Analyzing Table 4, most of the emissions occur during the stages of screen sandblasting and fusion of contaminated sands. The contribution to Global Warming of these processes is obtained as equivalent emissions of carbon dioxide, as specified by the characterization factors in the CML 2001. Nitrogen dioxide, sulphur dioxide and hydrochloric acid are the main substances emitted and responsible for acidification. Among these, the first one is produced during the dissolution with aqua regia, whereas the other two are produced during the precipitation with sodium bisulphite.

Emission Mass (kg) GWP100

(kg CO2 eq.)

AP

(kg SO2 eq.)

HTP

(kg 1,4-dichlorobenzene eq.)

POCP

(kg C2H4 eq.)

CO2 5.4900 5.4900

NO2 0.0330

0.0165 0.0396 0.0009

SO2 0.0058

0.0069 0.0006 0.0003

HCl 0.0260

0.0229 0.0130

PM10+PM2.5 0.0215

0.0176

Total 5.4900 0.0463 0.0707 0.0012

Table 4: Characterization of environmental impact related to the standard scenario

PMD SCENARIO Here, instead, the reduction of the time necessary for the blasting (-35%) allows to decrease the carbon dioxide emission. However, in this scenario, the residual sandblasting continues to be the step that most affects Global Warming and also Human Toxicity Potential due to the emission of particulate. In addition, focusing on the acidification, the process enabled by the PMD does not produce any emission that contributes to this category: this is another advantage of an aggressive solvents removal. However, heavy emissions contributing to the POCP are detected (specifically, they are related to acetone and dimethyl ether outputs from the etching and regeneration of palettes). Table 5 shows the overall values for emissions matched with impact categories.

Emission Mass (kg) GWP100

(kg CO2eq.)

HTP

(kg 1,4-C6H4Cl2eq.)

POCP

(kg C2H4eq.)

CO2 3.4587 3.4587

Acetone vapours 0.1470 0.0735 0.0014 0.0267

Dimethyl ether 0.1000 0.1000 <0.0001 0.0263

PM10+PM2.5 0.0143

0.0117

Total 3.6322 0.0131 0.0530

Table 5: Characterization of environmental impact related to the PMD process

4.4 Interpretation of results The graphs in Figures 8 and 9 provide the impact indicators for selected impact categories. As suggested by results, the standard scenario presents higher environmental impacts in most categories considered in the study, and specifically: GWP, AP and HTP indicators. Analyzing this result in detail, the following considerations can be made:

• Better results for the GWP indicator are a direct consequence of the less energy required during the etching activity, which is introduced in the PMD scenario, compared to the abrasive removal via sandblasting;

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

12

• Reduction of acidification and human toxicity potentials are related, instead, to the usage of “light” solvents as the acetone, which replaces strong acids involved in the aqua regia dissolution;

• The worse performance attained by the PMD scenario and related to the ozone formation is associated to the intense usage of acetone to dissolve the sacrificial layer. Now, this point should be taken carefully, as the POCP indicator used prudentially in these analyses for acetone (0.096 kg C2H4 eq./kg) has been under debate in the scientific community22.

We recall once again, how the eventual introduction of emission reduction systems (e.g. catalytic converters) has been explicitly excluded from the analysis. This has avoided a detailed discussion about available systems for emission treatment and their impact on final results, that is, a comparison of the two scenarios from the same perspective of “primary emissions of pollutants”. Finally, in the PMD scenario, the last process step allows a recovery of 99.86 % pure Gold, whereas the complete standard scenario provides an expected purity of 99.99% (after filtering, washing and final fusion step). Therefore, the output purities of the recovered gold can be considered equivalent in both systems, focusing on the boundaries defined in figures 7a and 7b.

Figure 8: GWP results in both scenarios

Figure 9: AP, HTP and POCP results in both scenarios

22 Starting from the work of Carter (Carter et al., 1993) several results against the influence of low-concentrations of acetone vapors on the air quality (e.g. the study commissioned by: Eastman Chemical Company, Hickory Springs Manufacturing Company and the Chemical Manufacturers Association in 1993) has lead to the recent decision by US EPA to declassify acetone as “volatile organic compound (VOC) of negligible reactivity”.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

13

5. Conclusion and future outlooks This research has analyzed actual and potential environmental impacts of two different techniques currently available for gold recovery process from PVD equipment maintenance. The study was developed according to the methodological guidance provided by ISO standards 14040-43 and it was articulated according to the four steps defined in the LCA methodology in order to maximize standardization and readability of results. In the paragraph dedicated to “goal and scope definition”, boundary conditions of the research and the functional unit were declared explicitly. A full description of the inventory for all processes included in the system boundaries and methods for data processing were defined, calculating inputs and outputs respectively. The research continued with the assessment of impacts using the “CML 2001” method. The authors finally provided a first quantitative interpretation of the results. This last phase showed that the process enabled by the PMD has (globally) a lower environmental impact than the traditional method of recovering gold. However, some critical points of this novel scenario have been highlighted. Among unexpected findings, we emphasized how the PMD scenario contributes the most photochemical ozone formation during the etching phase, compared to the traditional method. This is due to the use of acetone as a solvent in a step of the process at high temperature and in an open container. This conclusion outlines future directions of research, to improve impacts in the POCP. The PMD proposed by MRS, in fact, is at a proposal/prototypal stage for industrial use, therefore revisions of critical issues can be easily implemented. We envisage how a major investigation will involve the possibility to replace the acetone with less impacting solvents (e.g. formic acid) or even to substitute the current sacrificial layer with biodegradable ones.

Acknowledgements This work has been developed thanks to the collaboration with the MRS team, which is an Italian R&D enterprise that has focused its business on the practice of waste reduction in technologies widely used in microelectronics.

References Arena, M., Azzone, G., Conte, A., 2013. A streamlined LCA framework to support early decision making in vehicle

development. Journal of Cleaner Production, 41, pp. 105-113.

Baenard, W. O., Myburg, G., Auret, F. D., Goodman, S. A., Meyer, W. E., 1996. Metal contacts to gallium arsenide. Journal of Electronic Materials, 25(11), pp. 1695-1702.

Brug, J., Heidelberg, E., 1974. Patent nr. 3.856.507. US.

Bunshah, R. F., 1994. Handbook of deposition technologies for films and coatings: science, technology and applications. William Andrew.

Carter, W. P., Luo, D., Malkina, I. L., Pierce, J. A., 1993. An experimental and modelling study of the photochemical ozone reactivity of acetone. Final Report to Chemical Manufacturers Association Contract No. KET-ACE-CRC-2.0.

Coelho, P., Teixeira, J., Goncalves, O., 2011. Mining activities: Health impacts. Encycl. Environ. Health, vol. 3, pp. 788-802.

Corti, C. W., 2002. Recovery and refining of gold jewellery scraps and wastes. World Gold Council. London.

Fuke, I., Prabhu, V., Baek, S., 2005. Computational model for predicting coating thickness in electron beam physical vapour deposition. Journal of manufacturing processes, 7.2, pp. 140-152.

Gartner.com, 2012. Gartner Says Worldwide PC Shipments in Fourth Quarter of 2011 Declined 1.4 Percent; Year-End Shipments Increased 0.5 Percent. (Retrieved 2013-06-17).

Gentile, A. A., Modeo, S., 2012. Palette Modular Device for collection and recovery of metals in thin film deposition equipment. ID: US 20130008375 A1.

Goglio, P., Owende, P. M., 2009. A screening LCA of short rotation coppice willow (Salix sp.) feedstock production system for small-scale electricity generation. Biosystems engineering, 103 (3), pp. 389-394.

Hagelüken, C., Corti, C. W., 2010. Recycling of gold from electronics: Cost-effective use through Design for Recycling. Gold Bulletin, vol. 43, 3, pp. 209-220.

Hageluken, C., Meskers, C. E. M., 2010. Complex Life Cycles of Precious and Special Metals. In: Linkages of Sustainability, T. Graedel, E. van der Voet (eds.), The MIT Press, Cambridge, MA.

Hart, J., 2007. A case study of Volatile Aliphatic Ethers. QSAR Foundation.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

14

Hischier, R., Weidema, B., Althaus, H. J., Bauer, C., Doka, G., Dones, R., Nemecek, T., 2010. Implementation of life cycle impact assessment methods. Ecoinvent report.

Horowitz, L., 2006. Mining and sustainable development. Journal of Cleaner Production 14 (3-4), pp. 307 - 308.

Howard, W.L., 2011. Acetone. In: Kirk-Othmer Encyclopedia of Chemical Technology. Wiley. On line version

Imperial MSDS 85225. http://www.imperialsupplies.com

Kumah, A., 2006. Sustainability and gold mining in the developing world. Journal of Cleaner Production, 14, pp. 315-323.

Mattox, D.M., 2010. Handbook of physical vapor deposition (PVD) processing. William Andrew.

Oktyabrsky, S., Ye, P. (eds.), 2010. Fundamentals of III-V Semiconductor MOSFETs, ISBN: 978-1-4419-1546-7 (Print) 978-1-4419-1547-4 (Online).

Pacławski, K., Fitzner, K., 2004. Kinetics of gold (III) chloride complex reduction using sulfur (IV). Metallurgical and Materials Transactions B, 35(6), pp. 1071-1085.

Pitt, M. J., 1999. In Bretherick’s Handbook of Reactive Chemical Hazards; Butterworth-Heinemann Ltd: Oxford, Vol. 2, pp. 307-312.

Rossnagel, S. M., 1999. Sputter deposition for semiconductor manufacturing. IBM Journal of Research and development, 43, 1-2, pp. 163-179.

Sheng, P. P., Etsell, T. H., 2007. Recovery of gold from computer circuit board scrap using aqua regia. Waste management & research, 25.4: pp. 380-383.

Syed, S., 2012. Recovery of gold from secondary sources - A review. Hydrometallurgy, 115, pp. 30-51.

Sze, S. M., Ng, K. K., 2006. Physics of semiconductor devices. John Wiley & Sons.

Todd, J. A., Curran, M. A., 1999. Streamlined Life-Cycle Assessment: A Final Report From the SETAC North America Streamlined LCA Workgroup, Society of Environmental Toxicology and Chemistry (SETAC).

U.S. EPA, 1998. Supplement D to compilation of air pollutant emission factor. U.S. EPA.

U.S. EPA Office of Compliance Sector, 1995. Notebook Project: Profile of the Metal Mining Industry, EPA.

World Gold Council, 2014. Gold Demand Trends Full Year 2013. www.gold.org.

World Gold Council, 2010. White Paper: Gold for good – gold and nanotechnology in the age of innovation. www.gold.org.

Yannopoulos, J. C., 1991. Recovery of Secondary Gold. In The Extractive Metallurgy of Gold, Springer US, pp. 137-139.