Semiconductor Process Emissions
A semiconductor fab emits a few tonnes of fluorinated gas a year — a rounding error by mass against almost any other industrial process. Converted to CO₂-equivalent, those few tonnes can outweigh thousands of tonnes of ordinary combustion.
When one kilogram of a gas is worth 24,300 kilograms of CO₂e, the global warming potential is the methodology.
Semiconductor process emissions are high-GWP fluorinated gases (CF₄, C₂F₆, NF₃, SF₆ and others) released from etching and chamber-cleaning. Calculate them as gas fed × (1 − utilization) × (1 − abatement) × GWP, adding any byproduct gases formed — never from gas purchased alone.
This methodology covers CO₂-equivalent process emissions of fluorinated compounds (FCs) from semiconductor, flat-panel display, and photovoltaic manufacturing, plus fluorinated heat-transfer fluids, under IPCC 2006 Guidelines Volume 3, Chapter 6 (Electronics Industry Emissions). It sets out the gases and why their global warming potentials dominate the result, the four sub-sectors, the fed-gas mass balance, the Tier 1/2/3 ladder, the utilization, byproduct-formation and abatement terms, a complete six-gas worked example, and the edge cases and verification rules that determine whether the figure survives audit.
A methodology page is the execution layer. It takes the scope concept from the glossary, the calculation rules from the IPCC electronics chapter, and the global warming potentials from the data layer, and tells you exactly how to turn gas-consumption records into a defensible fluorinated-gas inventory. If you already have your gas-by-gas consumption and abatement data, jump to the Tier 2 mass balance or the worked example.
What Counts as a Semiconductor Process Emission
Semiconductor, display and photovoltaic manufacturing use fluorinated compounds in two process steps: plasma etching, which carves circuit patterns into the wafer, and chamber cleaning (CVD chamber clean), which strips deposited residue between production runs. Both feed gases such as CF₄, C₂F₆, NF₃ and SF₆ into a plasma. Some of the gas reacts as intended; the unreacted remainder, plus any new fluorinated species formed in the plasma, leaves the chamber and — unless destroyed by an abatement device — is emitted to atmosphere.
These are Scope 1 fugitive process emissions: they occur at the fab’s own equipment and would not exist without the process. They are entirely distinct from two other large lines a fab reports:
- Scope 2 electricity. A fab’s purchased-electricity footprint is typically far larger in absolute CO₂e than its process FC emissions. It is calculated from kWh × grid factor and reported on the Scope 2 line — never mixed with the process-gas line.
- Scope 1 combustion. Fuel burned in on-site boilers or abatement-device burners is stationary combustion, calculated from fuel quantity, and reported separately from the process FC line even though both are Scope 1.
Process FC emissions are trivial by mass and easy to under-prioritise — a fab might emit under 100 kg of SF₆ in a year. But SF₆ has a global warming potential of 24,300 [GreenCalculus gwp.SF6.ar6_100 · IPCC AR6 WGI Ch 7 Table 7.SM.7 (2021) — AR6 GWP-100 · v2026.284] on the AR6 100-year basis, so that sub-100 kg becomes thousands of tonnes of CO₂e. The entire methodology is built around getting the gas-by-gas mass right and applying the correct GWP — the mass balance and the GWP conversion are where the result lives, not the fuel or the electricity.
Provider directory
Need help keeping this up all year?
See who does this work. Every listing names the standards it works to, and paid placements are labelled. Including The Carbon Trust and Attain Zero.
Browse 6 carbon accounting & inventory providers →Do this work? A listing is US$390 a year. Get listed →
The Gases and Why GWP Dominates
Six fluorinated process gases carry IPCC default emission factors in this methodology, spanning three chemical families. What unites them is extreme atmospheric persistence and correspondingly enormous global warming potentials.
2.1 The fluorinated families
- Perfluorocarbons (PFCs) — CF₄ (tetrafluoromethane, PFC-14), C₂F₆ (hexafluoroethane, PFC-116) and C₃F₈ (octafluoropropane, PFC-218). Used in etch and chamber clean; extremely long-lived in the atmosphere.
- Fluorinated methane — CHF₃ (trifluoromethane, also designated HFC-23). An etch gas and a known byproduct of other processes.
- Nitrogen trifluoride and sulphur hexafluoride — NF₃, the dominant modern chamber-clean gas, and SF₆, used in etch and some clean steps. SF₆ has the highest GWP of any gas in routine industrial use.
2.2 Global warming potentials (live values)
| Gas | Designation | AR6 GWP-100 | AR5 GWP-100 | MasterBrain key |
|---|---|---|---|---|
| CF₄ | PFC-14 | 7,380 | 6,630 | gwp.PFC_CF4 |
| C₂F₆ | PFC-116 | 12,400 | 11,100 | gwp.PFC_C2F6 |
| C₃F₈ | PFC-218 | 9,290 | 8,900 | gwp.PFC_C3F8 |
| CHF₃ | HFC-23 | 14,600 | 12,400 | gwp.HFC_23 |
| NF₃ | Nitrogen trifluoride | 17,400 | 16,100 | gwp.NF3 |
| SF₆ | Sulphur hexafluoride | 24,300 | 23,500 | gwp.SF6 |
GWP-100 values: IPCC AR6 WGI Ch 7 Table 7.SM.7, surfaced live from the MasterBrain gwp.* family. Full dataset at the AR6 GWP values data page.
2.3 AR5 vs AR6 — which basis to use
Use AR6 GWP-100 for corporate GHG reporting: IFRS S2 and ESRS call for the latest IPCC values, the GHG Protocol and CDP recommend them, and it is the default for every GWP shortcode on this page. AR5 values are retained only for historical comparison or where a specific regulatory regime still mandates AR5. The two bases differ materially for these gases: CF₄ moved from 6,630 (AR5) to 7,380 [GreenCalculus gwp.PFC_CF4.ar6_100 · IPCC AR6 WGI Ch 7 Table 7.SM.7 (2021) — AR6 GWP-100] (AR6), and CHF₃ from 12,400 to 14,600 [GreenCalculus gwp.HFC_23.ar6_100 · IPCC AR6 WGI Ch 7 Table 7.SM.7 (2021) — AR6 GWP-100]. Mixing AR5 and AR6 values within a single inventory total is a reporting error.
In most emission methodologies the emission factor and the activity data carry the result, and the GWP is a minor closing step. Here it is inverted. The mass of gas emitted is small and the GWPs span four orders of magnitude — from CF₄ at 7,380 [GreenCalculus gwp.PFC_CF4.ar6_100] to SF₆ at 24,300 [GreenCalculus gwp.SF6.ar6_100]. A gas-allocation error that moves mass between CF₄ and SF₆ changes the result by a factor of more than three even though the kilograms are identical. Track each gas separately through to its own GWP; never sum gases by mass and apply an average.
The Four Sub-Sectors
IPCC Vol 3 Ch 6 publishes Tier 1 default consumption factors for four electronics sub-sectors, each on its own substrate-area basis. The MasterBrain stores them under sub-industry keypaths, so the calculation reads the factor that matches the product being made.
Semiconductors (integrated circuits)
semiconductors.tier1.*Logic and memory wafer fabs. Default consumption factors per m² of silicon processed — the largest and most data-rich sub-sector, covering CF₄, C₂F₆, NF₃ and SF₆.
Flat-panel display (TFT-FPD)
tft_fpd.tier1.*Thin-film-transistor display manufacturing. Factors per m² of glass substrate — generally lower per-area consumption than IC fabs, expressed in grams rather than kilograms per m².
Photovoltaic cells (PV)
pv_cells.tier1.*Solar-cell manufacturing. Factors per m² of silicon. A growing FC source as PV capacity scales, with chamber-clean gases the dominant contributor.
Heat-transfer fluids (HTF)
heat_transfer_fluids.tier1.*Fluorinated liquids used for temperature control and testing. Emitted by evaporative loss rather than plasma reaction; the MasterBrain carries a C₆F₁₄-equivalent proxy factor per m² of silicon.
These factors live in a MasterBrain section whose internal name differs from the keypath you query. Reads go through the sub-industry slug — semiconductors.tier1.cf4, tft_fpd.tier1.cf4, pv_cells.tier1.cf4, heat_transfer_fluids.tier1.c6f14_proxy — not through the section name. Querying the section name returns no value. Pick the keypath that matches the product type before reading the factor.
The Emission Mechanism
The emitted quantity of any process gas is not the quantity fed into the chamber — it is what survives the process and any abatement, plus any new fluorinated species the plasma creates. Four terms govern the chain.
- Utilization (U) — the fraction of fed gas consumed (destroyed or incorporated) by the process itself. A high-utilization step emits little of its fed gas; a low-utilization step emits most of it. U varies by gas, tool and process step.
- Byproduct formation (B) — the mass of a new fluorinated gas created per unit of fed gas. The dominant case is CF₄ formed from C₂F₆ and C₃F₈ dissociation. Byproduct CF₄ is emitted in addition to any directly fed CF₄.
- Abatement (A) — the destruction or removal efficiency (DRE) of any downstream abatement device (combustion, plasma or catalytic). A = 0 for an unabated tool; modern point-of-use abatement reaches 0.90–0.95 or higher for many gases, though NF₃ and CF₄ are harder to destroy than others.
- GWP — each emitted gas, including byproducts, is converted to CO₂e at its own AR6 GWP-100.
Choosing a Tier
IPCC Vol 3 Ch 6 offers a tier ladder of increasing precision and data demand. The choice is driven by what gas data the fab can produce.
semiconductors.tier1.* etc.), then by GWP. A gross screening estimate that bakes in default utilization and assumes no abatement.
Widest uncertainty; over-states for any fab with real abatement. Use for screening or first-year only.
The Tier 1 substrate factors are gross consumption proxies — they assume default utilization and no abatement. For a modern abated fab, a Tier 1 estimate can be an order of magnitude above the Tier 2 mass-balance result. Tier 1 is a screening ceiling, not a cross-check on Tier 2: a large gap between them is expected and is not evidence of a Tier 2 error. Report on the tier your data supports, and do not “reconcile” an abated Tier 2 figure back to an unabated Tier 1 default.
Tier 1: Substrate-Area Method
The Tier 1 method estimates gas consumption from the area of substrate processed, then converts to CO₂e. It is the only method available when gas-purchase records are unavailable, and it serves as a conservative screening ceiling.
7,380 for CF₄
| Sub-sector / gas | MasterBrain key | Default factor | Unit |
|---|---|---|---|
| Semiconductors — CF₄ | semiconductors.tier1.cf4 | 0.9 | kg per m² silicon |
| Semiconductors — C₂F₆ | semiconductors.tier1.c2f6 | 1 | kg per m² silicon |
| Semiconductors — NF₃ | semiconductors.tier1.nf3 | 0.04 | kg per m² silicon |
| Semiconductors — SF₆ | semiconductors.tier1.sf6 | 0.2 | kg per m² silicon |
| Flat-panel display — CF₄ | tft_fpd.tier1.cf4 | 0.5 | g per m² glass |
| Photovoltaic cells — CF₄ | pv_cells.tier1.cf4 | 5 | g per m² silicon |
| Heat-transfer fluids — C₆F₁₄ proxy | heat_transfer_fluids.tier1.c6f14_proxy | 0.3 | kg C₆F₁₄-eq per m² silicon |
IPCC 2006 Guidelines, Vol 3, Ch 6 Tier 1 defaults, surfaced live from the MasterBrain semiconductors.tier1.*, tft_fpd.tier1.*, pv_cells.tier1.* and heat_transfer_fluids.tier1.* families. Watch the unit column — TFT-FPD and PV factors are in grams per m², the others in kilograms per m².
The semiconductor and heat-transfer-fluid factors are in kilograms per m²; the flat-panel-display and photovoltaic factors are in grams per m². Reading a gram-basis factor as a kilogram-basis one overstates that sub-sector by a factor of 1,000. Confirm the unit from the field="unit" string before multiplying, and convert to a common mass basis before applying GWP.
Tier 2: Gas-by-Gas Mass Balance
The Tier 2 method is the industry standard and the one nearly all reporting fabs use. It works from actual gas consumption and applies the four mass-balance terms gas by gas.
7.1 The master equation
Ebyproduct = Σ ( Fsource × Bsource→bp ) × (1 − Abp)
ECO₂e = Σ ( Egas × GWPgas )
17,400 for NF₃, 24,300 for SF₆, and so on
7.2 Utilization and byproduct factors
Utilization and byproduct formation are gas- and process-specific. Tier 2a uses the IPCC default factors from Vol 3 Ch 6; Tier 2b and 2c replace them with tool-specific or process-specific values where the fab has measured them. Utilization is generally higher for chamber-clean gases like NF₃ than for some etch gases, and the C₂F₆→CF₄ and C₃F₈→CF₄ byproduct pathways must always be included where those gases are fed — omitting them understates CF₄.
7.3 The abatement credit
Abatement is the single most powerful term in the equation and the most scrutinised at audit. A correctly substantiated DRE can remove 90% or more of a gas’s emission; an abatement claim with no destruction-efficiency evidence is treated as A = 0 by verifiers. Crucially, DRE is gas-specific — a device that destroys 95% of C₂F₆ may destroy far less CF₄ or NF₃ — so a single blanket abatement percentage applied across all gases is a common and material error.
Because abatement directly reduces the reported figure, verifiers require destruction/removal-efficiency evidence for each gas and each device — commissioning tests, periodic re-tests, and uptime records. An unevidenced abatement claim is set to A = 0. On a fab whose true emissions assume 90% abatement, claiming that abatement without DRE evidence is the difference between a defensible figure and one roughly ten times larger — see the error traps for the magnitude.
A full six-gas fab is worked end-to-end in the worked example, including the CF₄ byproduct branch and the abatement credit.
Worked Example
A complete Tier 2 gas-by-gas calculation for a fictional fab, computed end-to-end with all arithmetic shown. Values are fixed — this is an audit record at the snapshot date of this methodology revision, reconciling stated inputs to the stated result. The utilization, byproduct and abatement parameters are IPCC-default-style illustrative values; a real fab substitutes its own measured Tier 2b/2c factors.
Each gas: fed mass (kg), utilization
U, abatement DRE A, and CF₄ byproduct factor B (kg CF₄ per kg fed).CF₄ 1,200 kg (U 0.20, A 0.90) · C₂F₆ 900 kg (U 0.40, A 0.90, B 0.10) · C₃F₈ 300 kg (U 0.40, A 0.90, B 0.08)
CHF₃ 250 kg (U 0.60, A 0.90, B 0.07) · NF₃ 600 kg (U 0.80, A 0.95) · SF₆ 150 kg (U 0.40, A 0.90)
GWP basis: AR6 GWP-100.
| Gas | Fed (kg) | 1−U | 1−A | Emitted (kg) | GWP (AR6) | tCO₂e |
|---|---|---|---|---|---|---|
| CF₄ | 1,200 | 0.80 | 0.10 | 96.00 | 7,380 | 708.4800 |
| C₂F₆ | 900 | 0.60 | 0.10 | 54.00 | 12,400 | 669.6000 |
| C₃F₈ | 300 | 0.60 | 0.10 | 18.00 | 9,290 | 167.2200 |
| CHF₃ | 250 | 0.40 | 0.10 | 10.00 | 14,600 | 146.0000 |
| NF₃ | 600 | 0.20 | 0.05 | 6.00 | 17,400 | 104.4000 |
| SF₆ | 150 | 0.60 | 0.10 | 9.00 | 24,300 | 218.7000 |
| Direct subtotal | 193.00 | 2,014.40 |
Two features of the result are worth drawing out. First, SF₆ contributes 218.7 tCO₂e from just 9 kg emitted — more than CHF₃ or NF₃ — because its GWP is the highest in the set. Second, the byproduct CF₄ adds 4.6% to the total from gases that were not fed as CF₄ at all; a calculation that fed those gases but omitted the byproduct branch would understate the inventory by that margin.
FabCo’s purchased-electricity emissions — likely tens of thousands of tonnes of CO₂e for a 300mm logic fab — are reported on the Scope 2 line, calculated from kWh × grid factor, and never added to this 2,111 tCO₂e process figure. The process FC methodology produces only the fluorinated-gas number; the fab’s largest climate line sits elsewhere in the inventory.
Edge Cases
Five situations recur in fab inventories and each has a defined treatment.
- NF₃ remote-clean vs in-situ clean. NF₃ dissociated in a remote plasma source upstream of the chamber achieves very high utilization (often above 0.95), so little is emitted; the same NF₃ used for in-situ chamber clean has lower utilization and emits more. Apply the utilization factor that matches the actual clean architecture, not a blanket NF₃ default.
- The C₂F₆/C₃F₈ → CF₄ byproduct. CF₄ formed as a byproduct is real emission and must be added to directly-fed CF₄. It is the most commonly omitted term in the whole method because the fab never purchased that CF₄ — but the plasma created it. Always include the byproduct branch for any fab feeding C₂F₆ or C₃F₈.
- Heat-transfer fluid evaporative loss. Fluorinated HTFs emit by evaporation and handling loss, not plasma reaction. They are accounted on a consumption/top-up basis (purchases minus inventory change), converted at the fluid’s GWP — the Tier 1 substrate proxy is a screening fallback only.
- Reclaimed and recycled gas. Gas captured, purified and returned to use is not emitted at the point of capture. Net fed gas is purchases plus inventory drawdown minus quantities reclaimed on-site; double-counting reclaimed gas as both emitted and refed overstates the inventory.
- Partial-year abatement. Where an abatement device is installed, fails, or is bypassed for part of the year, a single annual DRE is wrong. Time-weight the abatement: apply the device’s DRE only to the gas that passed through it while operating, and A = 0 to gas emitted during downtime. Uptime records are the evidence.
Reporting: F-Gas, ETS, CBAM and CSRD
Several regimes interact with semiconductor process emissions, and their scope differs — getting which applies right matters as much as the calculation.
- EU F-gas Regulation and the Kigali Amendment. The EU F-gas Regulation and the Kigali Amendment govern the phase-down of certain fluorinated gases. Semiconductor process gases such as NF₃ and the PFCs sit in the broader fluorinated-gas policy landscape, though the specific controls and reporting thresholds differ from those on refrigerant HFCs. Confirm which gases and uses are in scope for your jurisdiction.
- EU ETS. The EU Emissions Trading System covers PFC process emissions from specific listed activities, most notably aluminium smelting. Semiconductor manufacturing is not currently a listed ETS activity in the way aluminium is — so a fab’s process FC emissions are reported in its voluntary or regulatory GHG inventory under the GHG Protocol, not surrendered as ETS allowances. Verify current scope rather than assuming inclusion.
- EU CBAM. The Carbon Border Adjustment Mechanism currently covers a defined set of goods (iron and steel, cement, fertilisers, aluminium, hydrogen, electricity). Semiconductors are not a CBAM good at present, so imported chips do not currently carry a CBAM embedded-emissions obligation. Treat any future extension as a separate development.
- CSRD / ESRS E1 and CDP. For in-scope companies, process FC emissions feed the Scope 1 disclosure under the GHG Protocol as carried into ESRS E1 and CDP. The gas-by-gas breakdown and the GWP basis (AR6) must be disclosed; high-GWP gases attract particular scrutiny because a small mass error is a large CO₂e error.
It is tempting to assume that because aluminium PFCs are in the EU ETS, semiconductor PFCs are too — they are not, under current scope. Overstating regulatory obligation is its own compliance risk. The reliable position today: semiconductor process FCs are a Scope 1 GHG-inventory line under the GHG Protocol, disclosed through CSRD/CDP where the company is in scope, and governed for gas-handling purposes by F-gas policy. Confirm any ETS or CBAM extension against the live regulation before reporting under it.
Uncertainty and Tier Guidance
| Tier | Method | Typical uncertainty | When acceptable |
|---|---|---|---|
| Tier 1 | Substrate area × default consumption factor × GWP; no abatement credit | Very wide; systematic over-statement for abated fabs | Screening; first-year; no gas-purchase data |
| Tier 2a | Gas-by-gas fed mass × IPCC default U/B factors × (1 − A) × GWP | ±25–50% | Routine reporting where tool-specific factors are unavailable |
| Tier 2b / 2c | As 2a but with tool- or process-specific measured U/B factors | ±15–30% | Material fabs; better abatement and utilization evidence |
| Tier 3 | Direct stack/tool emission measurement (FTIR/QMS) scaled to annual operation | ±10–20% | Large emitters; reasonable-assurance audits |
Unlike combustion, where the activity data dominates uncertainty, here the U, B and A factors do. A fab on Tier 2a default factors carries wide uncertainty mainly because its real utilization and abatement differ from the generic defaults. Measuring tool-specific utilization and substantiating per-gas DRE — the move from Tier 2a to 2b/2c — is what narrows the band, and it is exactly what a reasonable-assurance audit examines. Plan that measurement programme alongside the abatement-installation programme, not after.
Verification and Assurance
- Gas reconciliation. Auditors reconcile fed-gas mass to purchase invoices and inventory-change records, gas by gas. The fed mass that enters the mass balance must tie to procurement — an unsupported fed figure undermines the entire calculation.
- Abatement DRE substantiation. The most scrutinised element. Verifiers require per-gas, per-device destruction/removal-efficiency evidence — commissioning tests, periodic re-tests, and uptime/availability records. Abatement claimed without this evidence is set to zero, which can multiply the reported figure several-fold.
- Byproduct inclusion. Verifiers check that CF₄ byproduct from C₂F₆ and C₃F₈ has been included. Its omission is a known, recurring under-statement.
- GWP basis disclosure. The AR6 GWP-100 basis must be stated, and no AR5 values mixed into the same total. Because the gases span four orders of magnitude of GWP, the basis and the gas-by-gas mapping are checked explicitly.
The audit framework follows the GHG Protocol Corporate Standard for boundaries and the IPCC Vol 3 Ch 6 method for calculation criteria, with the 2019 Refinement applied where it updates a default.
Error Traps with Magnitudes
Each error below produces a specific, quantifiable distortion. Magnitudes are shown for the FabCo example (2,111.447 tCO₂e total) to make validation and disclosure risk concrete.
| Error | What happens | Magnitude (FabCo) | How to avoid |
|---|---|---|---|
| Claim abatement without per-gas DRE evidence | 90% abatement assumed in the figure but unsubstantiated; verifier sets A = 0. | Figure jumps from ~2,111 to ~22,328 tCO₂e Roughly 10× — the unabated gross. The largest single error in the method. |
Hold per-gas, per-device DRE test and uptime records. Unevidenced abatement is zero. |
| Omit GWP conversion | Emitted gas mass reported as if it were tonnes of CO₂e. | Reports ~207 “tonnes” vs 2,111 tCO₂e Off by roughly 10,000× — the gases average thousands in GWP. |
Always multiply each emitted gas by its AR6 GWP-100 before summing. |
| Sum gases by mass and apply one GWP | All gases pooled and converted at a single average GWP instead of gas by gas. | Arbitrary error up to several-fold CF₄ (7,380) vs SF₆ (24,300) differ >3×; any mass shift between them mis-states the total. |
Track each gas to its own GWP. Never average GWPs across a mixed-gas total. |
| Use AR5 instead of AR6 GWPs | Inventory built on AR5 values where the corporate framework requires AR6. | Understates by ~205 tCO₂e (−9.6%) FabCo AR5 total 1,919.34 vs AR6 2,111.447. |
Use AR6 GWP-100 for corporate reporting; disclose the basis; never mix bases. |
| Forget the CF₄ byproduct | C₂F₆/C₃F₈/CHF₃ fed but the CF₄ they form omitted. | Understates by 97.047 tCO₂e (−4.8%) 13.15 kg byproduct CF₄ × 7,380. |
Include the byproduct branch whenever C₂F₆ or C₃F₈ is fed. |
| Read a g/m² Tier 1 factor as kg/m² | TFT-FPD or PV factor (grams) applied as kilograms. | Overstates that sub-sector 1,000× The TFT-FPD and PV factors are per-gram; semiconductors and HTF are per-kilogram. |
Confirm the unit via field="unit" and convert to a common mass basis first. |
| Apply one blanket abatement % across all gases | A device’s C₂F₆ DRE applied to CF₄ and NF₃, which it destroys less effectively. | Understates hard-to-abate gases CF₄ and NF₃ are more abatement-resistant than C₂F₆; a blanket figure flatters them. |
Use gas-specific DRE values. A single abatement percentage across all gases is not defensible. |
Methodology Metadata — for GHG Inventory Documentation
Copy into your GHG inventory methodology statement for inventory-transparency compliance. Adjust the tier and abatement lines to match what you actually applied.
Quick converters: turn a single process-gas mass into CO₂e — C₃F₈, C₄F₁₀, c-C₄F₈, C₅F₁₂ and C₆F₁₄.
Frequently Asked Questions
Because the gases are extraordinarily potent per kilogram. SF₆ has a 100-year global warming potential of 24,300 and NF₃ of 17,400, so reporting their mass alone would massively understate their climate impact. Converting each gas to CO₂-equivalent at its AR6 GWP-100 puts them on a comparable footing with CO₂ and with every other line in the inventory. The conversion is the most consequential step in the whole calculation — a small mass becomes a large CO₂e figure.
From the gas you emit, which is less than the gas you buy. The Tier 2 mass balance takes the fed mass and reduces it by the fraction consumed in the process (utilization) and the fraction destroyed by any abatement device, then converts the remainder to CO₂e. It also adds any byproduct gas the plasma creates — most importantly CF₄ formed from C₂F₆ and C₃F₈. Reporting purchased gas as if it were all emitted overstates the inventory; reporting only fed gas without the byproduct branch understates it.
As much as you can substantiate with per-gas, per-device destruction/removal-efficiency evidence — commissioning tests, periodic re-tests, and uptime records. Modern point-of-use abatement reaches 90–95% or more for many gases. But the credit is gas-specific: a device that destroys 95% of C₂F₆ may destroy considerably less CF₄ or NF₃, so a single blanket percentage across all gases is not defensible. And abatement claimed without DRE evidence is treated as zero at verification, which can multiply the reported figure several-fold.
When C₂F₆ and C₃F₈ are used in a plasma, some of each dissociates and re-forms as CF₄ — a gas the fab never purchased but nonetheless emits. This byproduct CF₄ is added to any directly fed CF₄, is subject to the same abatement, and is converted at CF₄’s GWP. It is the most commonly omitted term in the method precisely because it does not appear in purchase records. Any fab feeding C₂F₆ or C₃F₈ must include the byproduct branch, or it understates its CF₄ emission.
Not under current scope. The EU ETS covers PFC process emissions from specific listed activities such as aluminium smelting, but semiconductor manufacturing is not a listed ETS activity in the same way, and semiconductors are not currently a CBAM good. A fab’s process FC emissions are therefore reported in its GHG inventory under the GHG Protocol and disclosed through CSRD/ESRS E1 and CDP where the company is in scope, rather than surrendered as ETS allowances. Always confirm against the live regulation before reporting under any of these regimes, as scope can change.
Use the Tier 2 gas-by-gas mass balance wherever you have gas-consumption records — which most fabs do. Tier 1 substrate factors assume default utilization and no abatement, so for an abated fab they can be an order of magnitude too high; they are a screening estimate for the case where gas-purchase data is unavailable, not a cross-check on Tier 2. A large gap between the two is expected and is not evidence of a Tier 2 error. Report on the tier your data supports.