Low Carbon Concrete: Cements, EPDs and How to Specify It

Low carbon concrete is not a separate material. It is conventional structural concrete whose cement, binder quantity or cement accounting has been changed to cut greenhouse-gas emissions, and the documents we read concentrate on the cement and on how much of it a mix uses. This profile covers Portland-type concrete made with reduced-clinker cements and supplementary cementitious materials (SCMs), and cements whose declarations allocate carbon-capture reductions. It shows how a requirement can be written as a limit on global warming potential (GWP), what the standards and exposure classes allow, and which trade-offs the documents describe. Geopolymer and other alkali-activated concretes, which do not rely on Portland-type cement, are not covered; a separate profile is planned.

We read environmental product declarations (EPDs), one national specification, US federal requirements, catalogue pages of standards (not the paid texts) and research papers, and each section says which. For earth-based walls and finishes see rammed earth, compressed earth blocks and lime plaster.

What counts as low carbon concrete?

We found no single definition. The Norwegian Concrete Association's publication NB37 (in the June 2025 consultation draft we read in full; see below on later editions) defines low carbon concrete as concrete produced in accordance with the NS-EN 206+NA rules in which measures have been taken to limit greenhouse-gas emissions, expressed in CO₂-equivalents weighted by GWP.^[9] The abstract of Scrivener, John and Gartner's 2018 paper, which presents the conclusions of a working group initiated by the UN Environment Programme's Sustainable Building and Climate Initiative, states that Portland-based cement approaches will dominate in the near future and names two product-based approaches for further reductions: more use of low-CO₂ supplementary materials as partial replacements for clinker, and more efficient use of clinker in mortars and concretes. We read the abstract only.^[19]

A third lever needs no new material. NB37 notes that choosing a higher strength class than necessary can matter nearly as much as the choice of low carbon class: its Table 1 gives 188 kg CO₂e/m³ for B30 in class 20R and 180 for B25 in class 10R, which it reads as meaning that stepping down a strength class can cut emissions about as much as a better recipe.^[9] It also advises weighing low carbon concrete against reducing the concrete volume.^[9]

Which cements and additions replace clinker?

EN 197-1:2011 is the European standard for common cements. BSI's catalogue page lists it as current and says it defines 27 common cements, 7 sulfate-resisting common cements and low early strength blast furnace cements, each defined by the proportions of its constituents, in nine strength classes. We read the catalogue page, not the standard.^[14] Genorma lists a draft revision, prEN 197-1 ("Common cement - Performance assessment and declaration"), with 1 October 2026 as the close of voting; its summary lists, among other combinations, clinker with up to 95 % blast furnace slag (CEM II/A-S, CEM II/B-S, CEM III/A, III/B, III/C) and composite types CEM II/A-M, B-M and C-M. We read only that summary.^[15] The cement table further down shows the constituent shares printed in four cement declarations.

NB37 describes how each addition behaves in concrete (Norwegian practice, under its national annex):^[9]

• Fly ash. It gives slower strength development, and a longer setting time that can affect power floating and steel trowelling; this is usually offset with an accelerator, a warmer mix or covering after finishing. Fly ash added at the concrete plant counts towards the binder with efficiency factors (k) of 0.4, 0.7 or 1.0 depending on cement type and durability class. Concrete with more than 30 % fly ash is also used as low-heat concrete.
• Slag. It is used mainly inside CEM II and CEM III cements and allows low cement use for a given workability. NB37 says many producers using moderate-slag CEM II cement supply its class 20R as a standard product. High-slag CEM III cement can give a viscous, less compactable mix, so trial pours are advised. Above 50 % slag in the cement, strength development is slow at low curing temperatures, and slag-cement concrete can show relatively high autogenous shrinkage.
• Silica fume. It is normally added at the concrete plant (cements with interground silica fume exist in Europe but, NB37 says, none are available in Norway) and, combined with pure Portland cements, counts with k = 2 in the chloride durability classes, which NB37 notes also helps the carbon result.
• Limestone. It appears as a constituent of the CEM II/A-LL cement in the Heidelberg Materials UK declaration (10.4 %).^[5]

Calcined clay and LC3

Limestone calcined clay cement (LC3) replaces part of the clinker with calcined clay and limestone. We did not open a manufacturer document for an LC3 product, so we cite no product, EPD or standard designation for it. A 2025 open-access bibliometric review of LC3 durability (full text read) says that LC3 can reduce CO₂ emissions by 30 to 50 % because 50 % of the clinker is replaced, that calcined clay is heated to nearly 800 °C, and that pore-solution pH is reported around 13.2 for LC3 against 13.8 for ordinary Portland cement, which suggests lower carbonation resistance that may be handled with more concrete cover. These are statements the review takes from earlier studies, not its own measurements. The review analysed 21 articles and notes research gaps for freeze-thaw cycles, alkali-silica reaction, elevated temperatures and combined exposures.^[20]

How do you specify low carbon concrete by performance?

The route in every document we read is the same: set a numerical limit for GWP per cubic metre, tied to the strength class, and ask for an EPD that shows the mix meets it. The details differ.

The EPD basis

EN 15804:2012+A2:2019 gives the core rules for Type III environmental declarations of construction products; BSI lists it as current and under review (catalogue page read).^[16] The concrete EPDs we read from Hungary and Greece also cite the concrete product rule c-PCR-003, based on EN 16757;^[1][3] the Norwegian one cites NPCR 020:2021 Part B.^[4] All four cement EPDs refer to the cement product rule EN 16908 (Schwenk, OYAK and Heidelberg Materials Sement Norge as their basis); the Heidelberg Materials UK one states that it is based on the IBU product category rules for cement of 1 August 2021 and lists EN 16908 among its references.^[6][7][8][5] GSA's US requirements name the NSF product category rule for concrete (version 2.1, August 2021).^[10]

Three ways limits are written

• NB37 (Norway, consultation draft). Table 1 gives the maximum GWP-total for modules A1 to A3, in kg CO₂e per m³, for strength classes B20 to B65. For B30 the 2022 industry reference is 235; class 10R is 212, 20R 188, 30R 165, 40R 141, 50R 118 and 60R 94, the class number being the percentage reduction from the reference. Classes 70R (71) and 80R (47) are described as reachable only with capture-reduced cements. The limits exclude transport to site, and the EPD must give the value for one kilometre of transport so projects can add their own. NB37 says its earlier class limits are not directly comparable with these because of the change in how cement GWP is calculated.^[9] The file we read is the June 2025 consultation draft (høringsutkast; its foreword refers to the August 2025 revision). Norsk Betongforening's news pages (opened 12 October 2026) say the revised edition was released on 23 October 2025, which we did not open because it is sold through the association's shop, and that a further revision, NB37-2026, went to consultation in September 2026.^[21] In that 2026 consultation draft, which we opened, the B20 to B65 limits for the classes above are unchanged and the classes are renamed Lavkarbon 10 to 80 (LK10 to LK80).^[21]
• GSA (United States, interim requirements of 16 May 2023). EPD-reported GWP per m³ by strength class in psi, in three tiers (Top 20 %, Top 40 %, Better Than Average); for 4,000 psi the limits are 284, 326 and 352 kg CO₂e/m³, for 6,000 psi 319, 374 and 407. 30 % is added where high-early-strength mixes are needed for technical reasons and approved in writing. The EPD must follow the NSF concrete PCR and ISO 14025 and 21930, with facility-specific data where feasible.^[10] The document and its FAQ say GSA may update the requirements; the FAQ also refers to a statutory deadline of 30 September 2026 for obligating the funds, which has passed. We could not open later versions (the December 2023 PDF and GSA's current web pages returned errors).
• prEN 206-1:2024, clause 5.4 (draft, as described by GCCA). The Global Cement and Concrete Association's note says the draft classes concrete by the percentage by which its A1 to A3 GWP falls below that of a reference concrete of the same performance made with CEM I and no additions, in 10 % steps from GWR 0 (under 10 %) to GWR 9 (90 % or more), and that national provisions may define other systems.^[11] We did not read the draft or the final text.

Documentation rules worth copying into a specification

NB37 requires a product-specific EPD or a third-party verified climate declaration (CPD) for the actual mix; average EPDs for a product group are not accepted for classification, and a mix stays documented while its GWP changes by no more than 5 % from the declared value. Published EPDs at EPD Norge are valid for five years, and project-specific EPDs refer to a published one.^[9] It also asks the specifier to check local availability before fixing a class, since the class reachable depends on available binders, aggregate quality, transport distances and the supplier's experience.^[9]

Gross versus net: a trap in the numbers

NB37 explains that Scandinavian cement EPDs counted emissions from burning fossil-derived waste fuels (gross) until 2025, and that net reporting, which excludes them, was introduced in Norway from 15 June 2025; it says this lowers reported A1 to A3 cement GWP by 10 to 25 %.^[9] In the cement declarations we read, Schwenk prints 640 (gross) and 399 (net) kg CO₂e per tonne for one cement,^[6] Heidelberg Materials UK reports net 636 and, in a remark, gross 747.8,^[5] and the Norwegian prospective EPD prints both.^[8] A specification should say which value it limits.

The standard is changing

Genorma lists EN 206:2013+A2:2021 as withdrawn (effective 4 March 2026), revised by EN 206-1:2026 and EN 206-2:2026, with a third part in development.^[13] DIN Media lists the German edition DIN EN 206-1:2026-07 (published July 2026) as partially replacing DIN EN 206:2021-06.^[12] The two catalogue pages give different dates (a CEN withdrawal date on one, a German publication date on the other), and only DIN Media describes the replacement by Part 1 as partial; we read neither standard and cannot say how the content is split between the parts. Check which edition and national annex your contract documents cite.

How do exposure classes and EN 206 limit the mix?

Exposure classes are set by the environment, not chosen to suit the carbon target, and each one brings composition limits that decide how much clinker can be replaced. The clearest text we read is NB37's guidance on the Norwegian annex to EN 206 (national rules differ; for the UK we read only the Concrete Centre's page for its 2025 guide to specifying with BS EN 206 and BS 8500, not the guide or BS 8500-1 itself):^[9][18]

• The water/binder ratio is calculated with an efficiency factor, m = w / (c + k·p), where c is cement, p the addition and k its factor. The factor varies by material and by durability class. For fly ash it is 0.4 in class M60 (the carbonation-related class), because fly ash is reported to have an unfavourable effect on carbonation when added at the plant, and 0.7 in the chloride and chemical-attack classes.
• In M60 the maximum water/binder ratio is under 0.60 for pure Portland cement, under 0.55 for cements with up to 20 % interground fly ash and under 0.50 for up to 35 %.
• Routinely specifying XC1 to XC4 for interior elements that belong in exposure class X0 triggers durability class M60 and, NB37 says, a B30 strength class; many such elements could be specified as B25 in M90 with a considerable carbon gain. Combining a low strength class with a demanding durability class (its example is B30 M40) can force more binder.
• In the frost classes MF45 and MF40 the annex gives no rules for some binder combinations used in classes 30R to 50R; NB37 says the annex's Table NA.14 allows frost resistance to be documented for such combinations.
• Departing from the standard's composition rules extends the producer's responsibility, is to be agreed with the client and is to be stated on the delivery note.

What changes for curing, early strength and site work?

NB37 says fly ash and slag give slower strength development, which can matter for short production cycles in precast. Where the addition is ground into the cement the grinding is often made finer to compensate; when it is added at the concrete plant, compensating usually means lowering the water/binder ratio and therefore adding binder, which it says is normally out of the question when the aim is low carbon concrete.^[9] Fly ash concrete is less temperature-sensitive at low temperatures than ordinary structural concrete. Classes 10R and 20R can normally be cast in winter with ordinary measures, 30R may need extra insulation, and 30R to 60R achieved with fly ash or slag need special winter measures such as insulation plus heating. NB37 adds that any heating needed belongs in the project's carbon account, and that an accelerator changes the mix's EPD.^[9]

The Concrete Centre's GGBS page says the 7-day strength of GGBS concrete is typically 55 to 65 % of the 28-day strength against 70 to 80 % for a similar CEM I concrete, that similar 28-day strength is normally reached with up to 50 % GGBS at the same cementitious content, and that striking times for up to 50 % GGBS do not normally increase enough to matter; with higher levels, one-day strength may not suffice to remove vertical formwork, particularly at low temperatures, low cementitious contents and in thin sections.^[17] The page is undated, and it recommends agreeing a carbon range or limit where a project has low carbon aspirations.^[17]

What do capture-reduced cements change in an EPD?

NB37 describes cements from Heidelberg Materials' Brevik plant in Norway as available at three GWP levels: without capture, "evoBuild" (the capture reduction spread over all clinker produced) and "evoZero" (all clinker emissions offset), tracked in a "carbon bank", and says capture technology does not change the cement's technical properties.^[9]

We read the prospective EPD (P-EPD) NEPD-15839-19832 for evoBuild Low Carbon Anleggsement FA SR, CEM II/A-V 42,5 N, issued 10 June 2026 by Heidelberg Materials Sement Norge. Its A1 to A3 results per tonne, net, break down into 45.1 kg CO₂e for cement production excluding direct emissions and capture, 435 for direct emissions at the plant and minus 237 for carbon capture and storage, which sum to 243; gross values are 559 for direct emissions and 367 overall. The document states that it is prospective: the capture data derive from design and first-year capture rates, no operational data from the capture facility or storage were available, and results carry uncertainty. It defines the evoZero variant as the first column of the breakdown (45.1 net) and says sales of capture-reduced cement are subject to third-party assurance of the carbon bank, which we did not open (the assurance PDF returned a web page).^[8] These are accounting allocations with the direct emissions shown alongside; we do not net them into a statement about the product.

The accounting is still moving. The NB37-2026 consultation draft says that new mass-balance requirements in ISO 22095-2:2026 have led Heidelberg Materials to limit the reduction it allocates by mass balance to 95 % of the relevant direct clinker emissions instead of 100 %, which it puts at an increase of about 50 % in the GWP values of the evoZero cements; as a result, it says, a Lavkarbon 90 class reachable under the 2025 edition can no longer be reached and Lavkarbon 80 is the lowest-emission class reachable with evoZero. The same draft says reductions from mass-balanced capture can be used for NB37 classification but not in climate accounts under rules that accept only EN 15804 EPDs, such as NS 3720.^[21] We did not open ISO 22095-2 or a revised evoZero document; the 243 net figure above is what the June 2026 P-EPD states.

How do the EPDs we read compare?

The tables show eight declarations from seven manufacturers, exactly as each document states them; nothing was estimated or converted. Concrete is declared per cubic metre and cement per tonne, so they sit in two tables. Programme and PCR, strength class, declared unit and the life-cycle modules behind the figure are kept in separate columns.

Ready-mixed concrete, declared per 1 m³

Source notes. Duna-Dráva is the only pair from one manufacturer under one PCR, in the same year and strength class: the CEM III mix declares 189 and the CEM II mix 295 kg CO₂e/m³, which is 106 lower (our subtraction), and GWP-fossil is 165 against 252. The content declarations give 374 kg of cement per m³ for the CEM III mix and 348 for the CEM II mix, and neither document names the CEM II or CEM III sub-type or the share of additions. Both PDFs print 19 February 2026 as the version date; the registry collection page and both registry entries list 20 February 2026, status valid, and validity to 18 February 2031 (checked 12 October 2026).^[1][2] The Attica EPD covers 18 ready-mix products as an average across the Attica plants and lists CEM II/B-M (P-W-L), CEM II/A-L and CEM I cements without saying which one is in the C30/37 mix; the cement share for this mix is 13.5 to 14.5 % by weight at 2,390 kg/m³. Its registry page shows status valid, version date 20 June 2024 and validity to 5 December 2026 (checked 12 October 2026).^[3] The Ribe Betong declaration (approved 13 January 2023, valid to 13 January 2028, study year 2021) lists 312.07 kg of cement and 59.44 kg of SCM per m³ and does not define B30 or M60; the EPD Norge registry search returned an access error, so we have not checked its status there.^[4]

Cement, declared per 1 tonne

Source notes. The CEM III/A cement declares 667, above the net 636 of the CEM II/A-LL and the net 399 of the CEM II/B-M (V-LL). The documents differ in plant, country, year, PCR and accounting basis, and OYAK's states that its Iskenderun plant is not an integrated cement facility and that the clinker is supplied from other plants, including OYAK's Adana plant, with each supplier's clinker modelled separately.^[7] Our reading is that a cement's CEM II or CEM III label does not by itself tell you its GWP; the declaration does. Heidelberg Materials UK's EPD is valid from 19 July 2024 to 18 July 2029.^[5] Schwenk's was issued 27 January 2025 and is valid to 27 January 2030; the gross and net values are as printed, with the first value gross.^[6] OYAK's PDF cover shows a version date of 13 January 2025 and the registry page shows version :002 dated 21 January 2025, status valid, both with validity to 12 January 2030 (checked 12 October 2026).^[7] The evoBuild declaration is prospective and was issued on 10 June 2026.^[8]

EPDs from different programmes or built on different product category rules are not directly comparable. EPD International's own text in the Duna-Dráva documents says EPDs in the same product category from different programmes may not be comparable, and asks for the same PCR including the first-digit version number, identical declared units, the same life-cycle stages and the same assessment method.^[1] By that wording the Duna-Dráva EPDs (PCR version 2.0.1) and the Attica EPD (printed version 1.11) would not be comparable even though both come from the same programme. We rank no product and draw no conclusion about which is lower from the tables.

When to specify it, and when not

Where the documents support it

• Mainstream structural concrete. NB37 says its class 10R can usually be reached with ordinary fly-ash and slag cements and small recipe measures, 20R often needs special recipe measures, and 60R is roughly the most that generally available binders allow when capture-credited cements are left out.^[9]
• Thick elements and mass concrete. NB37 says more than 30 % fly ash, or often more than 50 % slag of the binder, gives low-heat concrete, which helps limit peak temperature and restrained cracking.^[9] OYAK's EPD lists mass concrete requiring low hydration heat among the uses of its CEM III/A (manufacturer statement).^[7]
• Chloride and sulfate exposure. The Concrete Centre's GGBS page lists high resistance to chloride ingress and to sulfate attack among the benefits (no figures given),^[17] and OYAK's EPD lists marine structures and sulfate and chloride exposure among the uses of its CEM III/A (manufacturer statement).^[7]
• Over-specified strength or volume. NB37 says stepping down a strength class can cut emissions about as much as recipe measures for a better class, and that low carbon concrete should be weighed against other measures such as reducing the concrete volume.^[9]

Where it needs care or may not suit

• Short production cycles, cold weather and thin sections. Slower early strength is documented by NB37 and the Concrete Centre; high replacement levels and low temperatures are where it bites.^[9][17] If heating or accelerators are used to compensate, NB37 says they belong in the carbon account.^[9]
• Carbonation exposure. NB37 gives fly ash a low efficiency factor in the carbonation-related class because of its reported effect on carbonation, which limits how much can be used there,^[9] and the LC3 review says its lower pore-solution pH may reduce carbonation resistance unless cover is increased.^[20]
• Frost classes. The Norwegian annex gives no rules for some binder combinations in MF40 and MF45, so extra documentation is needed.^[9]
• Finishing. Fly ash can lengthen setting time, which affects power floating and steel trowelling; NB37 recommends a trial pour where surface finishing is in doubt, and for CEM III mixes.^[9]
• Special concretes. The June 2025 NB37 draft says traditional measures are strongly limited for special concretes such as underwater and lightweight concrete; it says capture-reduced cement can bring underwater concrete into the lowest-emission classes, while for lightweight concrete the lightweight aggregate's emissions largely decide the class.^[9]
• Newer routes. We opened no LC3 product document, and the 2025 review notes gaps in freeze-thaw, alkali-silica and elevated-temperature data.^[20] The capture-reduced cement we read is documented by a prospective EPD that rests on design data,^[8] and the NB37-2026 draft reports a change in how evoZero reductions are allocated.^[21]
• Documents that are moving. EN 206 has a new edition, prEN 197-1 is a draft, the NB37 text we read is the June 2025 consultation draft (a revised edition was released in October 2025 and a 2026 revision is in consultation) and the GSA values are from the May 2023 interim version.^{[12][13][15][9][21][10]} Quote the edition in the specification.

We read no independent test report on the strength, durability or fire performance of any product named here, and the documents give no basis for choosing between the named manufacturers. The list of suppliers below shows whose documents we read, not who to buy from.

Suppliers whose documents we read

Frequently asked questions about low carbon concrete

The bottom line on low carbon concrete

In the documents we read, low carbon concrete is specified as a limit, not a product: a GWP figure for each strength class, a named set of life-cycle modules, a gross or net basis and a product-specific EPD. The documents we read agree on that and on the main trade-offs, slower early strength, winter measures, carbonation and frost classes, but differ in limits, PCR versions and editions. We did not read the texts of EN 197-1, EN 206, EN 15804 or BS 8500-1, we found no LC3 product document, and we read no independent test report; the figures above are what the named documents state.

Related materials