Global emissions of fluorinated greenhouse gases 2005-2050 with abatement potentials and costs

Abstract. This study uses the GAINS model framework to estimate current and future emissions of fluorinated greenhouse gases (F-gases), their abatement potentials, and costs for twenty source sectors and 162 countries and regions, which are aggregated to produce global estimates. Global F-gas (HFCs, PFCs, and SF6) emissions are estimated at 0.7 Pg CO2 eq.  in 2005 with an expected increase to 3.7 Pg CO2 eq.  in 2050 if application of control technology remains at the current level. There are extensive opportunities to reduce emissions using existing technology and alternative substances with low global warming potential. Estimates show that it would be technically feasible to reduce cumulative F-gas emissions from 81 to 11 Pg CO2 eq.  between 2018 and 2050. A reduction in cumulative emissions to 23 Pg CO2 eq.  is estimated to be possible at a marginal abatement cost below 10 EUR t−1 CO2 eq. We also find that future F-gas abatement is expected to become relatively more costly for developing than developed countries due to differences in the sector contribution to emissions and abatement potentials.


S1. Introduction
This is a description by sector of the estimations of global anthropogenic emissions of F-gases (HFC, PFC and SF 6 ) presented in the paper "Global emissions of fluorinated greenhouse gases 2005-2050 with abatement potentials and costs." It provides further insights into the details of the activity data, estimations of emissions, mitigation potentials and associated costs as well as a discussion of the most important sources for uncertainty in the sector estimates.

S2.1 Hydrofluorocarbon (HFC) emissions
In compliance with the Montreal protocol (MP), many sectors that formerly used the highly ozone-depleting substances (ODS) chlorofluorocarbons (CFCs) refrigerants changed rapidly to applications employing hydrochlorofluorocarbons (HCFCs) with lower ozone-depleting effects or hydrofluorocarbons (HFCs) with no ozone-depleting effects (IPCC/TEAP, 2005). Later, amendments to the MP require a complete phase-out of all ODS including HCFCs (UNEP, 2007). In the GAINS model, 14 different sources of HFC or HCFC emissions have been identified, whereof 8 are related to refrigeration and air conditioning. Table S1 presents sub-sectors distinguished in GAINS for HFC or HCFC emissions. Emissions from refrigeration and air conditioning sources are split by emissions from leakage from equipment in use and emissions from scrapping of the equipment at the end-of-life. In addition, for each emission source the fraction of HCFC to HFC in use is identified and modeled following the phase-out schedule of HCFCs in the latest revision of the MP.  Gschrey et al., 2011;Schwarz et al., 2011;Chaturvedi et al., 2015). Emission factors are sector specific with GWPs determined on the basis of the sector-specific shares of different types of HFCs commonly used and their respective GWPs. Table S2 presents GWPs used in GAINS, expressed in CO 2 equivalents over 100 years, as presented in the IPCC Fourth Assessment Report (AR4) (IPCC, 2007b) and now adopted for policy purposes in the Kyoto protocol. GWP's associated with perfluorocarbons (PFCs) and sulfur hexafluoride (SF 6 ) are also presented in Table S2. 23500 Magnesium production and casting Soundproof windows Other SF 6 * Stationary air-conditioning includes both commercial and residential air-conditioning ** Mobile air-conditioning includes buses, cars, light and heavy duty trucks + Foam includes both one component and other foams ++ HCFC-22 production for both emissive and feedstock use Source: (IPCC, 1996;IPCC, 1997;IPCC, 2007b;Gschrey et al., 2011;UNFCCC, 2012;IPCC, 2014)

S2.1.1 Stationary air-conditioning (residential sector)
To estimate emissions from stationary air conditioners (AC's) in the residential sector, we apply a method similar to what has been used in a model described by (McNeil and Letschert, 2007). HFC use for air conditioning depends both on the average HFC consumption per household using air conditioning (kg HFC/unit) and on the fraction of households who own air conditioners (penetration). (1) The number of households was calculated by dividing total population by average household size. Data and scenario values for average household sizes are taken from the UN Global Report on Human Settlements 2005 (UN-Habitat, 2005).
We assume that both energy consumption per appliance and the proportion of households owning air conditioners (penetration) depend on climate and income, being higher in warmer and richer places. Penetration in a certain region is formulated as a function of the climate maximum saturation for that region and of the percentage of the climate maximum saturation achieved at that time in the region (availability). (2) The climate maximum saturation is derived from the assumption that current penetration rates in the USA are the maximum for a climate with a given amount of cooling degree days (CDD's). The relationship between maximum saturation and CDD is exponential, as developed by (Sailor and Pavlova, 2003) and corrected to give a maximum of 100 percent by (McNeil and Letschert, 2007) whose equation we have used here. Availability of air conditioners as a function of income is assumed to develop along a logistic function, with a threshold point beyond 6 which ownership increases rapidly. Using data on present day air conditioner penetration in various countries from McNeil and Letschert (2007) we find availability as a function of income .
. / ( 3) where income is defined as GDP per capita in purchasing power parity (PPP) and converted to constant Euro 2010.
GDP and population data is taken from the GAINS model in consistency with relevant external scenarios, i.e. the PRIMES model for EU-28 countries Capros et al. (2012) and the IEA's Energy Technology Perspective (ETP) for non-EU countries (IEA/OECD, 2012). Data on cooling degree days and household size is taken from (Baumert and Selman, 2003) and (UN-Habitat, 2005), respectively. Once the number of stationary air conditioners is estimated, the HFC consumption is estimated assuming the average size of each appliance is 2.62 kW (Adnot et al., 2003) (Sailor and Pavlova, 2003) and the average refrigerant charge is 0.25 kg per kW (UNEP, 2011). An annual leakage rate of 11 percent and 13 percent is assumed for unitary air-conditioning systems in developed and developing countries respectively . At the end-of-life the scrapped equipment is assumed to be fully loaded with refrigerant which needs recovery with recycling or destruction. Servicing emissions are especially high for the informal servicing sector which constitutes a large part of servicing market in developing countries. For India, due to inadvertent releases during servicing of the air-conditioning equipment a higher annual leakage rate of 25 percent is used in this study (Chaturvedi et al., 2015).
The control options available for this source are different good practice options including leakage control, improved components and end-of-life recollection. These options are assumed to remove 30 percent of emissions banked in equipment in use and almost 88 percent of scrapping emissions (Tohka, 2005). Good practice options are being implemented in the EU as part of the different regulations controlling F-gases (Höglund-Isaksson et al., 2013). In countries with no prior national F-gas regulation in EU, full adoption of good practice options is assumed from 2015 onwards. For substantial further emission reductions, the use of HFC-410A (GWP 100 =2002) and other high GWP blends need to be replaced by an alternative low GWP refrigerant such as HFC-32 (GWP 100 =675) or HC-290 (GWP 100 = 3) pressurized CO 2 (GWP 100 =1).
One of the important features of low-GWP HFC alternatives refrigerants (i. e. HFC-32) is their heat transfer capacity. HFC-32 possesses about 1.5 times higher heat transfer capacity than HFC-410A, which means that its charge volume can be up to 30 percent smaller than existing refrigerants, depending on the model design.  (Rajadhyaksha et al., 2015).
In recent years, companies like Honeywell and Dupont have developed and marketed alternative substances with better performances and very short lifetimes of less than a few months. These are known as HFOs (or unsaturated HFCs). E.g. HFO-1234ze with a GWP 100 of 6 can be used in foam products and HFO-1234yf with a GWP 100 of 4 can be used in mobile air-conditioners. The suitability of these substances for stationary air conditioners has not yet been confirmed and they are therefore currently not applied in GAINS for this source. Another option would 7 be to use other non-HFC substances with low or zero GWP like hydrocarbons, CO 2 , dimethyl ether and other diverse substances used in various types of foam products, refrigeration, air-conditioning and fire protection systems. Switching to these alternatives is typically costly because it involves process modifications (Halkos, 2010), e.g., changing the process type from ordinary to secondary loop systems.

S2.1.2 Stationary air-conditioning (commercial sector)
The GAINS model store data on commercial floor space area for Annex-1 countries (Cofala et al., 2009). The primary data source for this data is the PRIMES model (Capros et al., 2012). For year 2005, the data on commercial floor space area was correlated with GDP/capita as illustrated in Figure S1. Fitting a linear trend line, the following relationship was retrieved: Using GDP per capita as driver, projections for future growth in commercial floor space area were obtained for each country. To estimate the HFC consumption in commercial air conditioning, a sector specific HFC consumption value of 0.02 kg/m 2 was applied (Höglund-Isaksson et al., 2013). Source: PRIMES model.
An annual leakage rate of 11 percent and 13 percent is assumed for unitary air-conditioning systems in developed and developing countries respectively (Gschrey et al., 2013). At the end-of-life the scrapped equipment is assumed to be fully loaded with refrigerant which needs recovery, recycling or destruction. Control options available for this source are similar to the options discussed for residential air conditioning (Section 2.1.1). 8

S2.1.3 Domestic refrigeration
For refrigeration in the domestic sector, growth in activity levels follows growth in number of households. Stock of refrigerators and national end use consumption are driven by population growth and trends in appliance ownership rates. In developed countries the market for refrigerators is saturated, i.e., nearly every household owns a refrigerator. Ownership rates are further increased only by ownership of multiple units of each appliance. In developing countries, however, ownership rates of even basic appliances are dynamic, and depend critically on household income level, degree of urbanization and electrification. In countries experiencing rapid growth in those parameters (e.g. China, India, Brazil etc.), appliance ownership growth is dramatic.
The GAINS model utilizes population forecasts in combination with an income model and econometric parameterization to arrive at the national ownership rate for each year in the forecast. The rate of ownership of refrigerator(s) per household is derived using a function estimated by the PAMS 1 model (2012). The general form of the function for the rate of refrigerator ownership per household is given by: where Sat DOM represents the saturation (rate) of domestic refrigerator ownership, I is the monthly household income given by GDP per household in the country, U is the national urbanization rate, E is the national electrification rate, and t is the year of the projected saturation.
The econometric parameter estimates from the PAMS model were applied to derive the rate of refrigerator ownership per household in GAINS. The number of refrigerators in a country was calculated by multiplying the ownership rate by the number of households in a country (UN-Habitat, 2005). Growth in number of refrigerators is driven by population growth and trends in appliance ownership as estimated above. Once the number of refrigerators is estimated, an average refrigerant charge of 0.1 kg HFC per unit (USEPA, 2010a) is used to estimate the HFC consumption in domestic refrigerators.
As domestic refrigerators are hermetic there is no risk of leakage during use, but there is a risk of emission release during the scrapping phase. At the end-of-life the scrapped equipment is assumed to be fully loaded with refrigerant which needs recovery with recycling or destruction. The control option available for this source is good practice during end-of-life scrapping, which is assumed to remove 80 percent of emissions (Tohka, 2005). The option is already in place in the EU through the F-gas Regulation 2006 . HC-600a (GWP 100 =3) is widely available for domestic refrigeration applications and suitable components (such as compressors) are widely available (UNEP, 2015a). HFOs are not yet used for this application. Compressors optimized for HFO-1234yf or HFO-1234ze in domestic refrigeration appliances are not yet widely available.

S2.1.4 Commercial refrigeration
Commercial refrigeration includes refrigerated equipment found in supermarkets, convenience stores, restaurants, and other food service establishments (Girotto et al. 2004 Figure S2 presents the HFC consumption in commercial refrigeration per unit value added for commercial sector in 2005 as reported by Annex-I countries to UNFCCC. As shown, reported rates vary greatly across countries. As we are not able to fully explain the variations in the reported consumption, e.g., by having access to information on consumption patterns for refrigerated goods, we adopt HFC consumption as reported. Projections for future HFC consumption are driven by growth in service sector value added. other high GWP blends need to be replaced by alternative low GWP refrigerants such as HFC-152a (GWP 100 =124), hydrocarbons and natural refrigerants (i.e. pressurized CO 2 , ammonia etc.). For stand-alone systems, HFO-1234yf and HFO-1234ze are possible alternatives when HCs are restricted by regional safety codes, as they have lower flammability. For condensing units, CO 2 is an option, although getting high efficiency and low capital cost is proving a challenge for condensing units. For new centralized systems, CO 2 is now in widespread use, especially in Europe (UNEP, 2015b).

S2.1.5 Industrial refrigeration
Food processing and cold storage is an important application of industrial refrigeration used for preservation and distribution of food while keeping nutrients intact. On a global scale this application is very significant in size and economic importance (Mohanraj et al. 2009 Starting point for the estimation of emissions from industrial refrigeration in Annex-I countries in GAINS is the HFC consumption reported for this source by member states to the UNFCCC for the years 2005 and 2010. Figure   S3 presents the HFC consumption in industrial refrigeration per unit value added for industrial sector in 2005 as reported by Annex-I countries to UNFCCC. As shown, reported rates vary greatly across countries. As we are not able to explain the variations in the reported consumption, we adopt it as activity data as reported. Projections for future HFC consumption are driven by growth in value added for manufacturing industry. For countries not reporting HFC consumption in this sector, the German consumption per value added has been adopted as default.  (GWP 100 =2088) and other high GWP blends need to be replaced by alternative low GWP refrigerants such as ammonia (NH 3 ) or pressurized CO 2 (Pearson, 2008;Messineo, 2012).

S2.1.6 Refrigerated transport
Refrigerated road transport includes transportation of food products (fresh, frozen or chilled), pharmaceutical products, and plants/flowers. The type of vehicles used for such transportations are trailers, heavy and small trucks, and vans. Refrigerated road transport vehicles have different capacities; vans are typically below 3.5 tonnes, small trucks and trailers vary between 3.5 to 7.5 tonnes, and heavy trucks have a capacity of more than 7.5 tonnes.
In 2010, there were around 4 million refrigerated vehicles in service worldwide (UNEP, 2010), including vans (40%), trucks (30%), semi-trailers or trailers (30%). These units predominantly use HFC-404A and HFC-410A as refrigerants. HFC-134a is also used for chilled distribution only vehicles. It is reported that the emission leakages from transport refrigeration systems are higher than those from stationary refrigeration because the  Figure S4 presents the HFC consumption in refrigerated transport sector per unit freight transportation in 2005 as reported by Annex-I countries to UNFCCC (2012). For countries not reporting HFC consumption specific for this sector, the rate reported for Austria (2.5 kg HFC per million tonne-km of freight transported) is adopted as default.
Projections of HFC consumption in refrigerated transport follow proportionately growth in GDP ( (Tohka, 2005). Further emission reductions 13 from this source can be achieved through switches to alternative refrigerants like hydrocarbons (HC-290, HC-600a, etc.), CO 2 and NH 3 . HFOs (i. e. HFO-1234yf blends) are also under consideration for use across transport refrigeration modes (USEPA, 2015).

S2.1.7 Mobile air-conditioning
A major source of F-gas emissions from the transport sector is emissions from mobile air-conditioners (MAC).
Air In the GAINS model, emissions from MAC are accounted for in cars, light and heavy duty trucks, and buses. The number of vehicle types in different GAINS regions is extracted from the GAINS model and consistent with transport fuel use in respective external energy scenario. The penetration rates for air-conditioners in different vehicle types is extracted from a detailed literature review (Kanwar, 2004;Hu et al., 2004;IPCC/TEAP 2005;CSI 2009;Rhiemeier and Harnisch, 2009;Uherek et al. 2010;Henne et al., 2012;Yan et al. 2014;Su et al., 2015).
Using the average charge size for different vehicle types the HFC consumption in the mobile air-conditioning sector is estimated (Repice and Schulz, 2004;IPCC/TEAP, 2005). Average charge sizes used are 0.6 kg for cars, 1.2 kg for light and heavy duty trucks and 12 kg for buses (Tohka, 2005;Schwarz et al., 2011). The leakage rate assumed from MAC in use is 10 percent (Tohka, 2005) and at the end-of-life the scrapped MAC is assumed to be fully loaded with coolant which needs recovery, recycling or destruction.

Evidence for mobile air conditioners from the B-COOL (2011) project funded by the EU Sixth Framework
Program suggests that the cost of a CO 2 -based AC system is between 1.5 to 2 times the costs of a HFC-134a system. Moreover, CO 2 -based systems show slightly higher fuel consumption at higher thermal load (35 °C) as compared to the HFC-134a system. This is in contrast to the fuel (diesel/gasoline) savings claimed by some CO 2 promoters (e.g., www.r744.com). As a compromise we do not assume any effect on energy consumption when switching to a CO 2 based system in stationary or mobile air conditioners.

Polyurethane one component foam (OC)
Foams became a significant application of HFCs as part of the phasing-out of CFCs under the MP. HFCs are used as blowing agents in a solidifying matrix of a polymer (UNEP, 2006). The main application of polyurethane (PU) 14 one component (OC) foam is to fill cavities and joints when installing inner fixtures in housing constructions.
Since one component foams come in pressurized canisters and cylinders, they are also called aerosol foams. One component blowing agents are typically gaseous and function as propellant for the foam. They volatilize upon application, except for small residues that remain for at most one year in the hardened foam. From early 2003, HFC-365mfc has been commercially produced as a substitute for foam blowing agent HCFC-141b, whose use in Europe has been banned since January 2004 (Stemmler et al., 2007).
To estimate emissions from one component foams we adopt HFC consumption in OC foams as reported by Annex-I countries to the UNFCCC (2012) for year 2005 and 2010. When reporting is missing for this source, the Swiss consumption per unit GDP (6.8 tonne HFC per billion Euro GDP) is adopted as default. Figure   The EU F-gas Regulation requires that all EU member states from 2008 stop using HFCs in OC foam unless this is required to meet national safety standards. The most common current replacement options for HFCs in foams are hydrocarbons and CO 2 . For some applications the performance of CO 2 in foam blowing is limited (UNEP, 2010). In GAINS, the options considered available for replacement of HFCs in one component foams are CO 2 , different hydrocarbons like propane and butane, and HFO-1234ze.

Other foams (OF)
The sector for other foams ( Emissions from foams can be controlled by replacing HFC-134a and other high GWP blends with an alternative blowing agent like CO 2 , water, hydrocarbons like propane or butane. According to Harvey (2007) a water/CO 2 mixture has been used in Europe (with a 10 to 20 percent market share by 2000) for solid PU in building applications. Approximately, 80 percent of XPS board foams in the EU use CO 2 for foam blowing however, CO 2 has some limitations with respect to thermal resistance and product thickness (UNEP, 2010). The remaining 20 percent will therefore need to use some other alternative, e.g., a mix of HFCs, HCs and water could be possible, but also HFO-1234ze is an interesting possible option (UNEP, 2010). In GAINS, the options considered available for replacement of HFCs in OF foams are CO 2 , HFC-152a, different hydrocarbons like propane and butane, and HFO-1234ze.

S2.1.9 Aerosols
HFC is used as propellant for aerosols released from cans and metered dose inhalers, e.g., medical asthma inhalers.   Table S3 for Article 5 (developing) and non-Article 5 (developed) countries. In addition to the phase-out of the use of HCFCs, the MP also requires the production and sales of HCFC-22 for emissive use to end completely by 2040.
In contrast to production of HCFC-22 for emissive use, the production and use   (2007) To calculate HFC-23 emissions from HCFC-22 production, GAINS applies an IPCC default emission factor of 3 percent related to the volume of HCFC-22 production for emissive (HCFC22_E) and feedstock (HCFC22_F) applications (IPCC/TEAP, 2005). Activity data are based on reported production levels for historic years (UNEP, 2012) and UNEP's phase out schedule for HCFC products for future years (UNEP, 2007). Projections of HCFC-22 production for feedstock use are assumed to grow proportionately with value added in manufacturing industry.
HFC-23 emissions from HCFC-22 production can be almost eliminated through post combustion during which HFC-23 is oxidized to carbon dioxide, hydrogen fluoride (HF) and water. The marginal abatement cost for destruction of HFC-23 emissions from HCFC-22 production is very low, less than 1 Euro/tCO2eq (Schneider 2011;IPCC/TEAP 2005). HFC-23 emissions from HCFC-22 production are assumed fully controlled in OECD countries through post-combustion. In this analysis we assume that the impact of CDM on emissions from HCFC-22 production in developing countries remain at the current level in the future (Fenhann, 2014).

S2.1.11 Ground source heat pumps
Geothermal energy is a renewable energy resource that can be used to provide electricity, heating, and cooling of commercial and domestic buildings and other facilities (IPCC, 2011). Geothermal heat pumps or ground source heat pumps (GSHP) are systems combining a heat pump with a ground heat exchanger (closed loop systems) or being fed by ground water from a well (open loop systems). The earth is used as a heat source when operated in heating mode, with a fluid as the medium which transfers the heat from the earth to the evaporator of the heat pump, thus utilizing geothermal energy (Sanner et al., 2003). In cooling mode, heat pumps use the earth as a heat sink. With borehole heat exchangers (BHE), geothermal heat pumps can offer both heating and cooling at virtually any location, with great flexibility to meet demands.
In Europe, the growth in GSHP systems has been accelerated by national policies stimulating installation, e.g., through subsidies, efficiency standards to new buildings and heating demand mandates for heat pumps (Eurobserver. 2009). Many European countries have identified barriers that mirror those seen in the United States, namely higher investment costs, lack of knowledge and awareness among end users, and underdeveloped institutional and financial support (EHPA, 2008). In the EU, Sweden (>320,000) and Germany (>150,000) today show the highest absolute numbers of GSHPs as shown in Figure S7.  For projections, it is assumed that the annual growth in GSHPs using HFCs follows the growth of solar heating 19 in the domestic sector as provided by the PRIMES model for EU-28 countries and IEA/OECD (2012) for non-EU-28 countries, but with the additional assumption that the market is saturated when the number of heat pumps corresponds to 20 percent of the number of households in a country. Growth in solar heating is here used as an approximation for the general growth in renewable energy sources. This is a rather crude assumption, which would be desirable to improve in the future through better information about the possibilities and limitations of expanding year. Emissions can be controlled through good practice options and switching to alternative substances. . GAINS considers HC-290 direct, CO 2 and HFO-1234yf as a key alternatives for HFC-410A use in GSHP.

S2.1.12 Fire extinguishers
Fire extinguisher, or extinguisher, is an active fire protection device used to extinguish or control small fires, often in emergency situations. The extinguishing agent is stored in a container and released in case of fire. Unlike in the refrigeration and air conditioning sector, on site refilling and on site recycling do not take place. After intended release in the event of fire or in case the equipment is malfunctioning (leakage, pressure drop), the containers are returned to the manufacturers. Re-charging, repair work and recovery is always done off site by specialist personnel. As long as the extinguishing agent is contained, it does not get polluted by impurities, and reclamation is not relevant. The industry points out that recovery and recycling of F-gas fire extinguishing agents has been only carried out to a small extent, since HFCs in fire protection have only been in use since the mid-1990s. As the lifetime is 15 years or longer, most systems are still in use .
HFCs were not used in fire protection before the MP. Their current, and growing, usage is a direct result of their adoption as halon alternatives, despite being inferior to halons both in terms of cost and performance (IPCC/TEAP, 2005). To estimate HFC consumption in fire extinguishers, we derive consumption rates per unit of GDP using HFC consumption reported by Annex-I countries to UNFCCC (2012)   . It is observed that HFC-227ea (54%), HFC-23 (23%), HFC-125 (13%) and HFC-236fa (7%) are mostly used for fire extinguishers in Annex-I countries. The majority of emissions will occur when the system is discharged, either when triggered accidentally or during a fire. Emissions may also occur during filling or maintenance of the systems; however these emissions are very small in newer systems, which often have leak detection and alarm systems as standard. Emissions are estimated to range from 1 to 3% of the fixed-system bank and 2 to 6% of the portable extinguisher bank per year (IPCC/TEAP, 2005). Annual leakage from equipment in GAINS is assumed 3.5 percent per year.

Figure S9. Share of HFC/PFC in fire protection sector of Annex-I countries in 2005
Emissions can be controlled through good practice options and switching to alternative substances. In recent years a low GWP alternative fluid (GWP 100 = 1) with equivalent extinguishing properties had been introduced to the 21 market with still growing success (Defra, 2008), the perfluoro-ketone FK 5-1-12 (Novec™ 1230). Its manufacturer and most European specialist equipment distributors rate FK 5-1-12 to be a feasible substitute for almost all applications of HFCs -for both HFC-227ea and HFC-23. GAINS considers FK-5-1-12 a key alternatives for HFC's in the fire extinguisher sector. In EU-28, fixed fire extinguisher systems are fully subject to the measures according to Art 3 and 4 of the F-gas Regulation. In developing countries, the Firefighting Sector phased out use of CFCs under the MP. The conversion technologies used were FM200, ABC powder, CO 2 , etc.

S2.1.13 Solvents
F-gas based solvents are mainly used for degreasing of metal prior to precision coating and in the optics and electronics sector (Defra, 2008). Specific end-user sectors identified by March study (1999) include dry cleaning, metal cleaning, precision cleaning and electronics cleaning. In recent years, HFCs have been developed that are used for this application in sectors such as aerospace and electronics. CFCs were used as solvents in precision cleaning before being replaced by certain HCFCs, namely HCFC-141-b. As an ozone depleting substance, this Recovery for recycling or reclamation of F-gas based solvents is unlikely. Therefore, no further mitigation options beyond the ban on F-gas based solvents are considered necessary to control emissions from solvent sector in GAINS model.

S2.2 Perfluorocarbon compounds (PFC) emissions
There are two major sources for emissions of perfluorocarbon compounds (PFCs); primary aluminium (Al) production and the semiconductor industry. Emissions from these secto1rs have typically very high global warming potentials.

S2.2.1 Aluminium Industry
Primary Al production has been identified as a major emission source of the two PFCs tetrafluoromethane (CF 4 ) with GWP 100 7,390 and hexafluoroethane (C 2 F 6 ) with GWP 100 12,200 times that of CO 2 (IPCC, 2007b). During normal operating conditions, an electrolytic cell used to produce aluminum does not generate measurable amounts of PFC. PFC is only produced during brief upset conditions known as "anode effects". These conditions occur when the level of aluminum oxide drops too low and the electrolytic bath itself begins to undergo electrolysis.
Since the aluminum oxide level in the electrolytic bath cannot be directly measured, surrogates such as cell electrical resistance or voltage are most often used in modern facilities to ensure that the aluminum in the electrolytic bath is maintained at the correct level. The GAINS model uses the production volume of aluminum as the activity driver for calculating emissions from this source. Primary Al production data is taken from external data sources, i.e., from the PRIMES model for the EU countries and from U.S. Geological Survey (USGS, 2013) for non-EU countries. For China and India, primary Al production data is taken from the GAINS Asia project (Amann et al., 2008;Purohit et al., 2010). Four different types of activities are distinguished based on the technology used; point-feeder prebake (PFPB), side-worked prebake (SWPB), vertical stud söderberg (VSS), and center-worked prebake (CWPB) technology. Shares of different Al production technologies were adopted from the Al industry websites, national communications to the UNFCCC (2012) and other publically available literature (Schwarz, 2008;RUSAL. 2009;IAI, 2009;Schwarz et al., 2011;Marks and Rand, 2012;IAI, 2014).  (Harnisch and Hendricks, 2000). Data on mitigation costs is taken from the same source. In Europe, emissions from the primary Al production is regulated under the EU-ETS system, control options with marginal costs falling below the expected ETS carbon price are adopted in the reference scenario. This means that with the natural turn-over of capital, all EU member states will have phased-in PFPB technology by 2020.
The development of inert anodes is sometimes promoted as a promising mitigation option, which could eliminate emissions of PFCs from the electrolysis process (IPCC, 2007a;Kvande and Drabløs, 2014). In the Energy Technology Perspective (ETP) 2010 by the International Energy Agency (IEA/OECD, 2010), deployment of inert anode technologies is expected to start in 2015-2020 with full commercialization by 2030 (Table S4). If realized, 23 inert anode technology would have significant energy, cost, productivity, and environmental benefits for the aluminum industry worldwide (RUSAL, 2010). The technology is expected to eliminate PFC emissions from primary Al production altogether. Despite promising initial results, the technology still needs further development before it can be introduced as a viable alternative to PFPB technology. In GAINS, inert anode technology is assumed available as a mitigation option from 2035 onwards, however, no adoption in the reference scenario is assumed. Source: (IEA/OECD, 2010)

S2.2.2 Semiconductor industry, PFC use in CVD and etching
The semiconductor industry uses HFC-23, CF 4 , C 2 F 6 , octafluoropropane (C 3 F 8 ), carbon tetrafluoride (c-C 4 F 8 ), sulphur hexafluoride (SF 6 ) and nitrogen trifluoride (NF 3 ) in two production processes: plasma etching thin films (etch) and plasma cleaning chemical vapour deposition (CVD) tool chambers (IPCC, 2001). Because PFC is only used by few companies in a country (Tohka, 2005) and because the amount of PFC use allows deriving production volumes, data on PFC use are often confidential. As activity variables for this sector GAINS uses the volume of PFC emissions as reported by Annex-I countries to UNFCCC (2012). For countries not reporting PFC consumption in this sector, the Chinese consumption rate of 6.4 Gg PFC per billion Euro value added in manufacturing sector (Bartos et al., 2008) in 2005 has been adopted as default.

S2.3 Sulphur hexafluoride (SF 6 ) emissions
Sulphur hexafluoride emissions arise from high and mid-voltage switches, magnesium production and casting, soundproof glazing and a variety of other applications using SF 6 . Compared to anthropogenic sources, natural sources of SF 6 are negligible. Although the atmospheric concentration of SF 6 is relatively low, contributing 0.1% of the total anthropogenic radiative forcing, the concentration is growing continuously (Levin et al., 2010;Rigby, et al. 2010) because of the compound's long lifetime of ̴ 3200 years (Ravishankara et al., 1993).

S2.3.1 High and mid voltage switches
The electrical equipment sector is the major emission source of SF 6 through leakage, maintenance, and retiring (IPCC/TEAP, 2005). SF 6 is used as an electrical insulator in the transmission and distribution equipment of electric systems. Most of the SF 6 is stored in gas-insulated switchgears for high and mid-voltage electric networks.
Emissions of SF 6 depend on the age of the gas insulated switchgear since older models leak more than newer ones, as well as on the size of the transmission network and on recycling practices of the old equipment. The GAINS model uses electricity consumption as activity variable for this sector. The emission factor for SF 6 in electricity transmission per unit of electricity consumed is taken from the GHG inventory of California (CEPA, 2010) and applied in a consistent manner to all regions.
Suitable alternatives to SF 6 do not exist for these applications as the oil and compressed air systems, which were used previously, suffer from safety and reliability problems (AEAT, 2003). SF 6 emissions resulting from leaks in electrical equipment can be addressed through leak detection and repair (LDAR) and, for larger leaks, refurbishment. SF 6 emissions can be reduced through the adoption of recycling practices of used SF 6 switchgears.
The EU F-gas Regulation requires end-of-life recollection and recycling from 2010 onwards. Full compliance with this regulation is assumed in GAINS to apply in all EU countries.

S2.3.2 Magnesium production and magnesium casting
Casting and production of primary and secondary magnesium are well known sources of atmospheric emissions of SF 6 . The gas is used as a shielding gas in magnesium foundries to protect the molten magnesium from reoxidizing whilst it is running to best casting ingots (IPCC, 2001). Activity data on historic volumes of processed magnesium are taken from the United States Geological Survey (USGS, 2013), UN statistics and the national communications to UNFCCC (UNFCCC, 2012). An emission factor of 1 kg SF 6 per ton processed metal is taken from Schwartz and Leisewitz (1999) and Tohka (2005). Based on the recently published data, magnesium processing SF 6 consumption factors of 1.65 kg SF 6 /t Mg is used for China (Fang et al., 2013). SF 6 emissions in magnesium production and casting can be substituted by using sulphur dioxide (SO 2 ) as alternative gas.

S2.3.3 Soundproof windows
Some European countries used significant amounts of SF 6 in soundproof windows. where the first term represents the end-of-life emissions from soundproof windows scrapped in year t and the latter term represents the emission leakage from windows still in use.
No further mitigation options beyond the ban included in the F-gas Regulation are considered necessary to control emissions from soundproof windows.

S2.3.4 Other applications
SF 6 have been used in tyres, sports equipment manufacturers in tennis balls and sport shoes. Activity data for these other sources of SF 6 emissions in Annex-I countries are taken from emissions reported by countries to the UNFCCC (2012). From 2006, the F-gas Regulation bans the use of SF 6 in sports equipment and tyres in EU-28.
Energy consumption in non-ferrous metals from IEA/OECD (2012) 16 Semiconductor industry PFC emissions in semiconductor industry (Bartos et al., 2008;UNFCCC, 2012) Growth in industrial value added from IEA/OECD (

EU-wide All GHGs
All non-ETS sectors Decision defines legally binding national GHG emission targets for non-ETS sectors. Target year is 2020, but countries need to comply with a linear emission path between 2013 and 2020.

2013
F-gas regulation (Regulation 517/2014) EU-wide HFCs, PFCs, SF6 All F-gas Limits the total amount of the most important F-gases that producers and importers are entitled to place on the market in the EU from 2015 onwards and phases them down in steps to one fifth of 2014 sales by 2030.
1 Jan 2015 National F-gas regulations Austria HFCs, PFCs, SF6 All F-gas sectors "HFKW-FKW-SF6-Verordnung" is more stringent than EU F-gas regulation in the control of emissions from foams.