Gate-to-Gate vs Cradle-to-Gate LCA: Where Should You Draw the Boundary?

Key differences at a glance

• One boundary is nested inside the other. A facility-only result is always a subset of the wider one, so it is always the smaller number. The gap is a real burden you have chosen not to count, not an error.
• “Gate-to-gate” means two different things. The strict form counts only direct on-site emissions. The usual form also charges energy used on site. On a single-use CHO process the strict form reads exactly zero.
• How much the facility view captures depends on the process, not on scale or titer. 95.6% for a 2,000 L CHO fed-batch, 90.1% for Pichia, 79.0% for a 10,000 L E. coli fed-batch, at any titer.
• It reverses the single-use versus stainless verdict. Inside the fence single-use is 2.9% lower. From the cradle stainless is 0.24% lower, and the verdict differs in 12 of 26 US grid regions.
• Declarations need the wider boundary. Product carbon footprints and EPDs for intermediate products are cradle-to-gate as a minimum. The facility view belongs in internal process work.

Side-by-side comparison

Both figures come from one inventory and one open factor set at a 400 g CO₂e/kWh grid. The gate-to-gate figure is the energy-inclusive form: electricity plus on-site steam.

What gate-to-gate means

A gate-to-gate study draws the system boundary around a single facility, or a single unit operation inside one. Inputs are counted as they arrive at the receiving dock. Outputs are counted as they leave. What happened to a bag of media before it reached you, and what happens to a spent bioreactor bag after the waste contractor collects it, sit outside the study.

The term is used in two ways, and the difference matters far more than it sounds. In the strict form, only emissions physically released on site count: a gas-fired steam boiler, an emergency generator, process CO₂ from the bioreactor off-gas if biogenic carbon is being tracked. In the energy-inclusive form, which is how most process engineers use the phrase, the electricity and fuel consumed inside the fence are also charged with the emissions of generating them, while material inputs are still counted only as masses. The first is roughly a Scope 1 inventory. The second is roughly Scope 1 plus location-based Scope 2, which our guide to Scope 1, 2 and 3 emissions in biomanufacturing maps in detail.

Gate-to-gate inventories also have a second, more respectable life as building blocks. Jiménez-González, Kim and Overcash (2000) set out a methodology for developing gate-to-gate life cycle inventory information: plant-level modules from which a longer chemical supply chain can be assembled. Used that way, a gate-to-gate inventory is not a short cut. It is one link in a longer study, and it is only meaningful once the other links are attached.

When gate-to-gate wins

When the question is about the process you control and the answer has to be measured rather than estimated. A site energy study, a change to cleanroom HVAC setback, a steam trap survey or a comparison of two campaigns in the same suite all live entirely inside the fence. The data comes from your own meters, the uncertainty is small, and nothing outside the gate changes between the options. That last condition is the one to check before relying on it.

What cradle-to-gate means

A cradle-to-gate study follows every input back to raw material extraction and stops when the product leaves the factory. For a biologics process that means the production burden of glucose, amino acids and salts, of the polyethylene, EVA and polypropylene in single-use assemblies, and of the water supply, as well as the treatment of the effluent and the incineration of spent consumables. The downstream life of the product, from fill-finish and cold chain to administration and disposal, is excluded; that further extension is cradle-to-grave, covered in our worked life cycle assessment example.

This is the conventional boundary for drug substance and other intermediate products, for a practical reason: the manufacturer cannot know what happens after the sale, but the buyer needs the whole burden of what they are purchasing. Jiménez-González and colleagues (2004) applied it to pharmaceutical compounds, and the biologics comparisons that followed, such as the single-use versus conventional study by Pietrzykowski et al. (2013) and the streamlined assessment of single-use technologies by Budzinski et al. (2022), necessarily account for the consumables that arrive at the facility, because that is the only way to compare disposable and reusable equipment at all.

The price is secondary data. You cannot meter a supplier's resin plant, so every material input needs an emission factor from a database or an environmental product declaration, and each carries its own uncertainty. Our guide to emission factors for bioprocessing lists the open sources and their confidence tiers.

When cradle-to-gate wins

Whenever the options being compared differ in what they buy or throw away. Single-use against stainless, a chemically defined medium against a hydrolysate, a disposable depth filter against a reusable centrifuge: each of these moves burden across the facility fence, and only a boundary that reaches past the fence can see both sides of the trade. It also wins whenever the number leaves the building, because a customer, an auditor or a product category rule will ask for it.

One batch at three boundaries

Definitions are easier to judge with numbers attached. This is the 2,000 L CHO fed-batch case the rest of this cluster is built on: 80% working volume, 3 g/L titer, 70% downstream yield, 14-day batch, 90% success rate, single-use assemblies, cleanroom HVAC at 80 kW, a 400 g CO₂e/kWh grid. It yields 3.024 kg of product per batch.

Three things stand out.

The strict form says this process emits nothing. This single-use CHO inventory contains no on-site combustion, so a direct-emissions boundary returns zero. That is a true statement about Scope 1 and a useless one about the product. Anyone quoting a “gate-to-gate” figure without saying which form they mean could be handing you this number.

The usual form captures 95.6%. Electricity is almost the entire footprint, and 93.5% of that electricity is cleanroom HVAC. For this process the energy-inclusive facility view is a good approximation, and it is why a hotspot study of a mammalian plant rarely needs anything wider to find the biggest lever.

What it misses is mostly plastic. The 532.2 kg CO₂e per batch outside the fence, 176 kg per kg of product, is 84% single-use resin and its incineration. Media, a large mass flow, is only 0.29% of the carbon. That split is specific to this process, and the next two sections show why it matters.

Where the boundary reverses the answer: single-use vs stainless

Run the matched stainless steel twin through the same three boundaries. It is identical in volume, titer, yield, batch length, success rate, cleanroom and grid; it differs only in cleaning regime (1,400 kg of steam, 26,000 L of CIP water against 2,000 L) and polymer inventory (9.5 kg against 82 kg). The full environmental case is in our single-use vs stainless environmental impact comparison; here the only question is what the boundary does to it.

kg CO₂e per kg product, 400 g CO₂e/kWh.

Both differences are small. The point is not that either option is clearly better; our comparison page concludes they are effectively tied. The point is that the facility view announces a winner, and it announces the wrong one, for a reason that is entirely structural.

The two technologies keep their extra burden on opposite sides of the fence. Per batch, the stainless premium is 193.7 kg CO₂e of steam plus 144.0 kg from distilling the extra 24,000 L of water, both inside the gate, plus 20.6 kg of extra water supply and effluent treatment outside it. The single-use premium is 170.1 kg of extra resin production plus 216.6 kg of extra incineration, and all 386.7 kg of it is outside the gate. A gate-to-gate study sees 337.7 kg of stainless penalty and none of the single-use penalty. A cradle-to-gate study sees both, and the single-use side is marginally the larger.

Under the facility view the verdict cannot move with the grid. Stainless uses more electricity and burns steam, so at any grid intensity it is the higher of the two. From the cradle, the verdict turns on the grid instead, crossing at 479.2 g CO₂e/kWh: below it stainless is lower, above it single-use is. Twelve of the 26 US eGRID subregions sit below that crossover, from upstate New York (NYUP) to Arizona and New Mexico (AZNM). In those 12, the two boundaries give opposite answers to the same question.

Tightening the boundary further makes it worse. Remove cleanroom HVAC as well, the equipment-only boundary sometimes used for unit-operation studies, and the facility view puts stainless 44.9% above single-use (360.2 against 248.6 kg CO₂e/kg) while the cradle view has stainless 2.2% below (415.2 against 424.6). The narrower the boundary, the larger the apparent gap and the more confidently it points the wrong way. Gate-to-gate captures only 58.6% of the single-use footprint once HVAC is gone.

The general rule is simple. If two options differ in what they buy or discard, the boundary must reach past the fence on both sides. A process mass intensity comparison has the same blind spot for the opposite reason: it counts the masses crossing the fence and gives no weight to what making them cost.

When is gate-to-gate close enough?

For a single process, rather than a comparison, the facility view may be an acceptable approximation. How good depends almost entirely on how energy-intensive the process is relative to its material inputs. Every preset in the free LCA calculator, at the same 400 g CO₂e/kWh grid:

Footprints in kg CO₂e per kg product. “Within 10%” counts the eGRID subregions where gate-to-gate captures at least 90% of cradle-to-gate.

The ranking has a clear physical logic. Mammalian processes spend two weeks in a heated, filtered, pressurised cleanroom, so electricity swamps everything else. Microbial processes finish in a day or two with modest HVAC but consume tonnes of glucose, yeast extract and peptone, and the carbon of growing and refining those ingredients sits entirely outside the fence. On the E. coli case, media production alone is 16.3% of the footprint. A small single-use pilot sits in between for a third reason: its resin inventory is large relative to the product it makes.

The closed form, and why titer does not matter

The captured share has a simple structure. Write E for batch electricity in kWh, g for grid intensity, S for on-site fuel emissions and R for everything outside the fence. Then:

captured share = (E·g/1000 + S) ÷ (E·g/1000 + S + R)

R does not depend on the grid, so the absolute gap is fixed and only its share changes. For the single-use CHO batch the gap is 176 kg CO₂e per kg in every region, which is 12.1% of the cradle figure in upstate New York (1,277 against 1,453) and 2.4% in MRO East (7,245 against 7,421). Solving for a 10% miss gives g = (9R − S) × 1000 ÷ E: 166.6 g CO₂e/kWh for single-use CHO, but 1,020.5 g CO₂e/kWh for E. coli, which is dirtier than any US subregion. The cleaner your grid, the less the facility view captures. A site that has decarbonised its electricity has, by the same act, made its upstream supply chain the larger share of what is left.

Titer, by contrast, cancels. Every term inside and outside the fence scales with the same product mass, so the single-use CHO batch captures 95.58% at 1, 3, 5 and 10 g/L alike (gate-to-gate 11,412, 3,804, 2,282 and 1,141 kg CO₂e/kg). The adequacy of a facility boundary is a property of the process type and the grid, not of how well the process runs.

Pros and cons

Which boundary should you use?

If in doubt, build cradle-to-gate and report the facility subtotal alongside it. The facility figure costs nothing extra once the stage breakdown exists, and publishing both removes the temptation to quote whichever looks better.

Real-world use cases

Beyond the gate: cradle-to-grave and modular studies

The two boundaries on this page are the inner two of four. The wider ones are worth placing so the vocabulary is complete.

• Gate-to-gate: one site or process step.
• Cradle-to-gate: adds everything upstream of the site and the treatment of its wastes.
• Cradle-to-grave: adds distribution, use and end of life of the product. For a biologic that means fill-finish, cold chain, administration and disposal of the device and packaging.
• Cradle-to-cradle: replaces end-of-life disposal with recovery into a new product. Rarely applicable to a drug, occasionally to its packaging.

The inner boundaries also stack. A cradle-to-gate footprint for a monoclonal antibody drug substance is itself a gate-to-gate module from the point of view of the fill-finish site that buys it, which adds its own facility burden and passes the total downstream. This is the modular logic behind gate-to-gate inventories, and it only works if every module declares its boundary and its functional unit, because a module expressed per batch cannot be chained to one expressed per kilogram.

Two other choices interact with the boundary and are easy to confuse with it. Whether you use average or marginal data is a modelling approach, not a boundary, and our attributional vs consequential LCA comparison shows it can matter as much. Which impact categories you report is an impact assessment choice, covered in the life cycle impact assessment guide. A defensible study states all three.

What the standards and the software say about boundaries

None of the major frameworks forbids a narrow boundary. They all require that it is declared and justified, and the ones built for disclosure set a minimum.

• ISO 14044 — requires the system boundary to be defined in the goal and scope, consistent with the goal, and any omission of life cycle stages, processes or flows to be stated and justified through cut-off criteria. It names no mandatory boundary.
• GHG Protocol Product Standard — requires cradle-to-grave for final products and permits cradle-to-gate partial footprints for intermediate products whose use is unknown, which is where drug substance normally falls.
• ISO 14025 and the International EPD System — Type III declarations are governed by product category rules, which fix the boundary so that declarations in the same category can be compared.
• EU Product Environmental Footprint — category rules again fix the boundary, and intermediate products are assessed cradle-to-gate.
• openLCA, SimaPro and Brightway — the boundary is whatever you link into the product system. A process with unlinked inputs is gate-to-gate for those inputs whether you intended it or not, which is the most common way a facility-only result ends up labelled as something wider.

That last point is the practical trap. In general-purpose software, a missing supplier link does not raise an error; it silently narrows the boundary. Our comparison of SimaPro, GaBi and openLCA covers the packages themselves. The bioprocess LCA tool avoids the trap by construction, because every media, water and polymer input is charged with a production factor by default, and the stage breakdown shows exactly which terms sit inside and outside the fence.

Related calculators and tools

Related articles

Resources and references

• Jiménez-González, Kim & Overcash (2000), The International Journal of Life Cycle Assessment 5:153–159. Methodology for developing gate-to-gate life cycle inventory information. The reference for gate-to-gate inventories as chainable modules.
• Jiménez-González, Curzons, Constable & Cunningham (2004), The International Journal of Life Cycle Assessment 9:114–121. Cradle-to-gate life cycle inventory and assessment of pharmaceutical compounds. The boundary applied to pharmaceutical manufacture.
• Pietrzykowski et al. (2013), Journal of Cleaner Production 41:150–162. An environmental life cycle assessment comparison of single-use and conventional process technology for the production of monoclonal antibodies.
• Budzinski et al. (2022), New Biotechnology 68:28–36. Streamlined life cycle assessment of single use technologies in biopharmaceutical manufacture.

Further reading

• GHG Protocol Product Life Cycle Accounting and Reporting Standard. Boundary requirements for final and intermediate products.
• US EPA eGRID. Public-domain source of the 26 subregion grid factors behind the regional counts on this page.