Activated Carbon Filter for Commercial Kitchen Exhaust: Removal, Limits and Specification

Updated September 2026

Activated carbon filter for commercial kitchen exhaust is a gas-phase adsorption stage that polishes cooking odor and vapor after particulate removal, not a grease filter and not a range-hood charcoal pad. The precipitator upstream collects grease aerosol and smoke; the carbon bed downstream holds the molecules that pass any precipitator: gases, not droplets. A generous bed still underperforms when the exhaust is hot, wet or grease-laden, and a spent bed looks like a fresh one. The worked figures below are illustrative, not Polygee test data or quotations.

This guide is for consultants, contractors and facilities engineers who specify a commercial kitchen exhaust treatment train for a United States or United Kingdom project. It covers what the bed removes, what passes through, the inputs that size it, how exhaustion shows itself and how to write a clause a supplier can test. It deliberately does not repeat the model range, MOQ, lead time or quotation content on the Polygee activated carbon filter product page. The first question is what the word on the module label promises.

What a Carbon Filter Is, and What the Label Does Not Tell You

A carbon filter is a packed bed of adsorbent granules or carbon-filled panels placed in the exhaust after particulate removal, and the label on the module describes the media, not the odor outcome at the discharge.

Activated carbon is carbon processed to open an internal network of pores, so that a small mass carries a very large surface. Vapor-phase molecules that touch that surface are held by physical adsorption, not by a chemical reaction, which is why the hold is reversible: the bed can be exhausted, and what it holds can be displaced or driven off by heat. In a kitchen exhaust the bed is a consumable, not a permanent component.

The retail listings that fill the search results for this phrase sell recirculating range-hood charcoal pads: a thin mat of carbon-loaded fabric behind a domestic hood. A commercial carbon box is a duct-mounted cabinet of drawer-type carbon panels with a stated media depth, placed in a discharged exhaust behind a precipitator. The two share a media name, not a duty.

Datasheets carry four numbers a specifier will encounter. ASTM D4607-14(2021) covers the iodine number, a laboratory index of micropore volume measured in solution; its scope does not rate a bed against cooking vapor, and a high value does not prove that the bed will hold a fatty-acid fragment at kitchen temperature. ASTM D3467-04(2020) covers carbon tetrachloride activity, a capacity index taken with one test vapor; it does not translate into an odor-removal percentage for any menu. ASTM D2862-16(2022) covers particle size by sieve and ASTM D2866-11(2025) covers ash content; both describe the lot, not the bed built from it.

When a quotation says “high-grade carbon” without any of these numbers, the phrase is a label, not lot evidence. Ask for the grade sheet, the base material and the test method behind each figure; set a value against another supplier’s only once the method behind each is confirmed.

Decision rule: specify the carbon stage by residence time, media depth and a breakthrough check at the discharge, not by the adjectives on the media datasheet. The media numbers qualify the carbon; the geometry and the site check qualify the stage.

The label says what was poured into the module; only the geometry and the exhaust conditions say what the bed will do on that roof.

How Adsorption Turns Cooking Vapor Into a Loaded Bed

The bed works as a moving front, not as a fixed filter: the inlet layers load first, the front travels toward the outlet, and the day it arrives is the day the odor returns.

1. Remove the droplets first — the precipitator upstream takes out grease aerosol and smoke particulate so that the carbon receives gas, not a coating of oil.
2. Pass the gas through the bed — the airflow, the bed face area and the media depth together set how long each molecule is in contact with the carbon.
3. Let the molecules diffuse into the pores — heavier, less volatile molecules are held more readily; small, very volatile ones are held weakly and can be displaced by heavier arrivals.
4. Watch the front move — the zone of active adsorption travels from inlet to outlet as the upstream layers saturate; temperature and humidity change how fast it moves.
5. Recognize breakthrough — when the front reaches the outlet face, the outlet concentration climbs toward the inlet concentration; nothing visible changes.

The US EPA describes volatile organic compounds as gases emitted from certain solids or liquids, the physical state the bed is built for; that page addresses indoor air and sets no discharge limit for a kitchen exhaust.

Two conditions push the front faster than any sizing sheet assumes. Warm exhaust lowers what the media can hold, because adsorption releases heat and reverses as temperature rises. Moist exhaust fills pore volume with water, so a bed downstream of a dishwash line loads with water before it loads with odor. Neither effect is a media defect.

For the precipitator side of the train, see our what is an electrostatic precipitator guide.

Which Inputs Decide Whether the Carbon Stage Earns Its Place?

A carbon stage is justified by the discharge condition and the exhaust condition together, not by the fact that a kitchen produces odor.

Six inputs decide whether the bed is worth its pressure drop and media cost:

In the United Kingdom the Defra guidance on the control of odour and noise from commercial kitchen exhaust was withdrawn on 15 September 2017; its four-part risk score survives in local authority guidance, but a withdrawn method is council practice, not current government policy, and citing it does not tell a supplier what the authority will accept. How the score is read is in our commercial kitchen odour control UK guide.

Where the discharge condition is loose and the exhaust is hot and wet, the honest answer may be that a carbon stage adds cost without a result; the consultant, not this article, owns that call.

The 6-Variable Carbon Filter Window

The 6-Variable Carbon Filter Window is the set of six values that must exist before any bed can be sized rather than picked.

The two ratios built from this table, Q/A and L/(Q/A), are defined in the sizing section below. Every other datasheet number sits behind these two, and both are checkable with a tape measure and the fan schedule.

A window with one value missing is a guess, not a specification; the supplier who quotes anyway is quoting a module, not a result.

What the Carbon Filter Removes, and What It Leaves Behind

The bed sorts molecules by size, volatility and chemistry, and a kitchen exhaust contains every category at once.

The matrix below is a sorting aid, not a removal-efficiency table:

The bed earns its place on rows three and five, holds row four only weakly, holds row six only when the media is impregnated for it, tolerates row seven, is damaged by rows one and two, carries row eight as a second duty, and does nothing useful on row nine. A specification that ignores the damaging rows buys a bed replaced early, not a bed that removes more.

Why the ESP Comes First: Grease Load and Bed Blinding

The order of the train is a protection decision, not a preference: particulate removal first, photolytic or oxidizing treatment second where used, adsorption last.

Grease leaves a cooking line as droplets small enough to pass a baffle filter. A precipitator charges them and collects them on plates; whatever passes the plates is what the carbon receives. When the precipitator is off, bypassed or running on fouled cells, the droplets coat the granules from the inlet face inward, and airflow does not reverse the coating. For a contractor, the failure risk is a bed grease-loaded early because a cell was left out after a wash, not a media defect.

Where a UV stage is fitted, it sits between the precipitator and the carbon, and the carbon bed must sit last. Its lamps and the precipitator’s corona produce ozone; residual ozone that survives the duct run after the lamps is decomposed on the carbon, which uses some capacity, while the UV stage’s oxidation of grease vapor reduces the gas-phase load reaching the bed. Which effect dominates depends on the train, so the carbon is specified for the measured inlet, not for odor alone. The three-stage layout is on the kitchen ecology unit page; the wider choice of stages is compared in the kitchen exhaust filtration guide.

The order also decides what the operator keeps buying. The precipitator cell is washed and refitted, so its cost repeats as labor; the carbon is used up and repeats as a purchase. On a drawer-type box the media is bought by the panel, so the carbon frame count on the model sheet is the reorder unit, not a performance figure; this guide gives no change interval for any model, because the load, not the airflow, decides it. Which stages are washed and which are consumed is in our ecology unit filters guide.

The standards that govern the train say little about how the bed is sized; the one that names carbon attaches a condition to it. NFPA 96 covers ventilation control and fire protection for commercial cooking operations, from hoods to the cleaning of the grease path; its scope does not set an odor limit or rate an adsorbent. In the United States, IMC 2021 Chapter 5 covers commercial kitchen hoods, grease ducts and exhaust equipment; the chapter does not certify a device. An authority having jurisdiction may also ask whether a pollution control unit carries a UL 8782 listing, which covers unit assemblies installed in line with commercial grease ducts; the listing addresses construction and fire behavior, not odor, and this page claims no such listing for any equipment. In the United Kingdom, BESA DW/172 Section 24 treats pollution control as project-specific: clause 24.6 names the precipitator as one way to reduce visible smoke, and clauses 24.7 and 24.8 name activated carbon as the adsorber for gas-phase odor and the recommended means of adsorbing surplus ozone where a UV stage is used, on condition of a high level of upstream protection, with the note that local authorities increasingly do not accept carbon as the primary means of odor control; under 24.11 the added resistance of whichever stages are chosen goes into the fan selection. It describes a system, not a tested performance of any module. How that clause is read is in our DW/172 kitchen ventilation specification guide.

Caution: a loaded carbon bed in a grease path is a fire-load question for the fire consultant and the adopted code. ASTM D3466-06(2023) covers the ignition temperature of granular activated carbon in flowing air; it tests fresh media, not a grease-laden spent bed. This page is not a compliance opinion, a hazard analysis or an approval.

The precipitator earns its place ahead of the carbon by keeping droplets off the media; the carbon earns its place behind the precipitator by taking what the plates cannot.

Sizing Language: Residence Time, Face Velocity and Bed Depth

Three numbers turn a module into a specification, and all three can be checked on site.

Face velocity is the airflow divided by the bed face area. Residence time is the media depth divided by the face velocity. Bed depth is the media thickness in the direction of flow, not the housing depth on the outline drawing.

Suppose a design airflow of 10,000 m³/h, which is 2.78 m³/s, meets a bed face of 2.0 m² with a media depth of 0.30 m. Face velocity is 2.78 divided by 2.0, about 1.39 m/s; residence time is 0.30 divided by 1.39, about 0.22 s. If the fan turns down to 5,000 m³/h the residence time doubles to about 0.43 s; at 15,000 m³/h it falls to about 0.14 s. Teaching example only — not Polygee test data, not a product claim, not an acceptance threshold.

Device-level test methods also appear on quotations. ISO 10121-1 covers gas-phase air cleaning media and ISO 10121-2 covers full-size gas-phase air cleaning devices for general ventilation; ASHRAE Standard 145.2 covers a laboratory test method for gas-phase air-cleaning devices. Each uses named challenge gases at stated conditions; none reproduces a particular kitchen’s mixture, temperature or humidity, and a report to any of them does not imply that the quoted module was tested to it. A supplier figure such as “removal efficiency 95%” belongs to the challenge gas, concentration and airflow of that one test, not to this kitchen.

Pressure drop belongs in the same clause: the bed adds resistance to the fan static, and the value rises as the inlet face loads. The ventilation contractor, not this article, owns the fan static allowance.

Residence time is a specification number; a removal percentage is a marketing number until the challenge gas, concentration, temperature, humidity and airflow are written beside it.

How Do You Read Carbon Exhaustion Without Guessing?

Only a paired sample, taken upstream and downstream of the bed on the same service, shows exhaustion; every other symptom has a competing cause, so rule those out one check at a time and act on the check that does not clear the bed.

Exhaustion has no visible signature, so the 5-Signal Breakthrough Screen pairs each symptom with its competing causes and the one check that separates them:

An optional differential-pressure gauge on a drawer-type carbon box feeds rows three and four of the 5-Signal Breakthrough Screen; it reads resistance across the whole module, which rises as the inlet face loads, not the vapor capacity behind it. Where the two samples in the first row differ only slightly, the bed is near its end; where they match, the bed is through. The change-out decision belongs to the operator’s record, not to the calendar or the module’s appearance.

Write a Carbon Filter Specification the Supplier Can Test

A testable clause names the media, the geometry, the evidence and the acceptance check in that order, and the 4-Layer Odor Evidence Ladder keeps the four apart.

1. Media identity — grade name, base material, iodine number to ASTM D4607, carbon tetrachloride activity to ASTM D3467, mesh size to ASTM D2862, ash to ASTM D2866, and any impregnation named with its purpose. This layer proves what was supplied, not what it will do.
2. Bed geometry — face area, media depth, media mass per module, face velocity, residence time and clean pressure drop at design airflow. This layer is what a contractor can measure on delivery.
3. Device evidence — any test report offered, with the method, challenge gas, concentration, temperature, humidity and airflow stated. A report records that laboratory’s conditions, not a transferable result for this kitchen.
4. Site acceptance — the sampling point downstream of the bed, the comparison point upstream, the method, who samples, who signs, and what reading triggers change-out. The consultant, not this article, owns the acceptance limit and the sampling method.

Supplier language is where clauses go soft. “Odor-free discharge” is a wish, not a test; “long-life carbon” is a claim until the duty hours, the inlet condition and the change-out trigger are written beside it. Do not borrow a removal figure from one kitchen for another; one site’s result says nothing about a different site’s bed.

The 4-Layer Odor Evidence Ladder is a coordination device, not a test method.

What Belongs in the Supplier RFQ

Copy these fields into the request before comparing quotations:

Polygee can help compare a proposed bed against the 6-Variable Carbon Filter Window and the 10-Field Carbon Filter Enquiry Sheet, but it should not overrule the consultant who owns the acceptance limit. The activated carbon filter product page carries the eight CB carbon box models, selected by airflow, flange size and carbon frame count, and the commercial terms; the one term worth repeating is that carbon media is excluded from the one-year unit warranty as a consumable. Please do not use this guide as a price or model list.

Frequently Asked Questions