One of the first questions a new carbon accounting project runs into is which emission factor database to use. The answer matters more than most companies expect when they start. Using EPA eGRID factors for a Singapore facility's electricity is factually wrong. Using IPCC AR6 combustion factors for a Scope 3 lifecycle calculation produces numbers that will not hold up under a verifier's methodology review. Using an outdated Ecoinvent version when a newer one changes the key material factors by a meaningful percentage undermines year-over-year comparability.
These three databases serve different purposes. Understanding what each is for, what it covers, and where its limits are is a prerequisite for defensible emissions accounting.
IPCC AR6: combustion emission factors
The Intergovernmental Panel on Climate Change publishes its assessment reports periodically. The most recent full report, AR6, includes emission factors for combustion of stationary and mobile fuel types. These are the default factors for Scope 1 combustion calculations: natural gas, diesel, LPG, fuel oil, coal, biomass, and other fuel categories.
IPCC AR6 combustion factors express the quantity of CO2, CH4, and N2O emitted per unit of energy content (typically kg per gigajoule or kg per tonne of fuel). They are internationally accepted reference values derived from extensive review of combustion chemistry literature. The GHG Protocol explicitly references IPCC factors as appropriate for Scope 1 combustion.
The practical use case is straightforward: you have a natural gas consumption figure from your utility bill in cubic meters or therms, you convert to energy content using the net calorific value for your fuel grade, and you apply the IPCC AR6 CO2, CH4, and N2O factors to get the greenhouse gas quantities. You then apply GWP values from the same IPCC report to aggregate to CO2e.
What IPCC AR6 does not provide: electricity grid factors (that is a national registry question), material lifecycle factors (that is Ecoinvent's domain), and spend-based economic sector factors (that is EEIO model territory). Using IPCC combustion factors for anything other than direct combustion is a category error.
EPA eGRID: US electricity grid emission factors
The US Environmental Protection Agency's eGRID database provides emission factors for US electricity generation by sub-regional grid and by state. It is updated annually and is the standard reference for Scope 2 calculations for US facilities under the location-based method.
eGRID factors vary significantly by US grid subregion. The northeastern US has a different emission intensity than the Midwest because of the different generation mix: a grid with more natural gas and nuclear versus one with more coal produces less CO2 per kilowatt-hour. Using the correct subregion factor for a facility's utility account is important; using a national average will misstate the location-based figure.
For Singapore-based manufacturers, eGRID is not the relevant database for any part of the emissions inventory. eGRID covers the continental US, Hawaii, and Alaska; that is it. A manufacturer in Singapore using eGRID for its electricity calculation is using the wrong geographic database entirely. The correct source for Singapore electricity emission factors is the Energy Market Authority's annual publication of the grid emission factor for the national grid.
eGRID is also not appropriate for fuel combustion calculations (use IPCC AR6) or for Scope 3 material production factors (use Ecoinvent or similar lifecycle databases). It is one data source for one calculation type in one geography.
Ecoinvent: lifecycle inventory database for Scope 3
Ecoinvent is a comprehensive lifecycle inventory database developed by a Swiss not-for-profit consortium. It contains several thousand datasets covering the production and transformation of materials, energy carriers, chemicals, and industrial processes. Each dataset provides an emission profile (in kg CO2e per unit of output) that reflects the full upstream chain of inputs, not just the final production step.
This makes Ecoinvent particularly relevant for Scope 3 Category 1 calculations where you have physical quantity data: if you know you purchased 12 tonnes of primary aluminum, you can apply an Ecoinvent factor for aluminum production to get a supply chain emissions estimate that accounts for the smelting process, the electricity input to smelting, and the upstream mining and refining.
Ecoinvent is versioned. The database is updated regularly as new lifecycle studies are completed and existing datasets are revised. Ecoinvent 3.9, 3.10, and later versions may show different factors for the same material if the underlying production data has been revised. This creates a methodology documentation obligation: you need to record which version you used and keep it consistent across reporting periods, or explicitly document and justify any change of version.
The limitation of Ecoinvent for corporate GHG reporting is that it represents average industry production, not your specific supplier's process. A primary aluminum supplier using hydroelectric power for smelting has a dramatically lower emission profile than one using coal-fired grid electricity. Ecoinvent's factor for aluminum production will not distinguish between them. Primary supplier data remains the gold standard for Scope 3 accuracy; Ecoinvent-based estimates are the best available approximation in the absence of primary data.
Where the databases diverge and why it matters
The risk of database confusion is most acute in three scenarios.
The first is applying a combustion factor to an electricity calculation. A company that uses IPCC combustion factors for its electricity (perhaps because a template has a combustion tab and an electricity tab, and someone accidentally copies the wrong row) will produce a Scope 2 figure that is completely incorrect in both magnitude and units.
The second is using eGRID for non-US electricity. This happens when a company uses a template built for a US context and applies it globally without updating the electricity factor source. The eGRID factor for a mid-Atlantic US subregion will not approximate the Singapore grid factor in any meaningful way, and a verifier will flag it immediately.
The third is using IPCC combustion factors for Scope 3 material purchases. IPCC AR6 provides factors for what happens when you burn a fuel; it does not provide lifecycle factors for the production of steel, plastics, chemicals, or other purchased goods. Applying a combustion factor to a raw material purchase quantity produces a number that is methodologically incoherent.
The version problem
Each of these databases is updated on its own schedule, and the updates matter. IPCC moved from AR5 to AR6, and the GWP values for methane and nitrous oxide changed in ways that affected total CO2e figures even when activity data was constant. Ecoinvent updates certain material factors as new production data becomes available. EMA updates Singapore's grid emission factor annually.
The version you use is part of your methodology documentation. If you switch versions between reporting years, you need to either restate the prior year on the new factor basis or explicitly disclose the change and quantify its effect. This is not a theoretical concern; it is a real question that comes up in third-party verification reviews when year-over-year numbers show unexpected movement.
The consistent principle is: document which database, which version, and which year of that version. Do not let your methodology documentation say only "IPCC emission factors" when the difference between AR5 and AR6 GWP values for CH4 is material to your Scope 1 totals.
What this means for manufacturers in practice
For a mid-market Singapore manufacturer doing emissions accounting, the relevant combination is typically: IPCC AR6 for Scope 1 combustion, EMA grid factor for Scope 2 electricity (with any relevant regional factors for overseas facilities), and Ecoinvent or similar LCA database for Scope 3 material quantities where physical activity data is available, with EEIO spend factors where only spend data exists.
Using the wrong database does not just affect accuracy in a continuous way. It can produce a figure that a third-party verifier will not accept at all, because the mismatch between factor type and activity type is not a precision question but a category error. Getting the database selection right from the start is worth the clarity it provides later.
At Zevero, we maintain current versions of the relevant factor databases and apply the correct database to the correct emission source category automatically during the factor matching step. The methodology documentation records which version was applied to each entry in the ledger, which is what a verifier needs to trace the calculation back to source.