How permanent is biochar? H/C, random reflectance and the EU CRCF

Expert article · Smart Carbon Farming

How long does the carbon in biochar really stay stored? This determines how many carbon removal units a batch may generate – and whether they will still stand in 20 years. This article summarises the state of science and of the EU framework and shows how we at tgo AG derive conservative balances from it. It is based on our technical dossier for the Smart Carbon Farming group (Interreg NWE) on the IBI webinar by Prof. Hamed Sanei (Aarhus University) in September 2026, complemented by a review of literature and standards.

The result in one sentence

Traffic light	Point	Assessment
🟢	H/C below 0.5 → 100/200-year balance	All models are close together (73–94% after 100 years). Use the lowest value.
🔴	H/C 0.5–0.7 without Rₒ measurement	Models differ by up to 80 percentage points (10% to 91%). Do not claim long permanence.
🟢	Rₒ + reactive fraction → 1,000-year proof	Directly measured inertinite fraction; provided as a separate method in the CRCF (Delegated Regulation (EU) 2026/285).
🟡	Unified model (Sanei et al. 2026)	Coherent, links lab and soil trials – but new and not yet independently confirmed.
🟡	Rₒ for every batch	Expensive and slow. Better: proxy measurement on site and Rₒ as a spot check.

How the Earth itself locks away CO₂

Biochar mimics one of the two geological pathways by which the Earth binds CO₂ over millions of years: plants take up CO₂, the biomass is charred – for example by wildfires – and carried into sediments. There the carbon survives over geological time as so-called inertinite. Pyrolysis corresponds to the charring, application to soil to the start of burial.

An important thought for any monitoring: if biochar moves from the field into waters and sediments through erosion, that is not a “leak” but exactly the natural storage pathway. What should be assessed is actual decomposition to CO₂ – not physical relocation.

Method 1: H/C ratio and incubation trials

Biochar is incubated with soil in a closed vessel for one to several years, the CO₂ released is measured and extrapolated to 100 years. The results are plotted against the molar hydrogen-to-carbon ratio (H/C_org): the lower the H/C, the more carbonised and the more stable. Two models are widely used:

• Woolf et al. 2021: F_perm(100) = 1.0445 − 0.6351 × H/C_org (R² = 0.33) – a straight line.
• Azzi et al. 2024: BC₁₀₀ = 93.8 − 18.0 × (H/C)^4.85 (R² = 0.91) – almost flat at 92–94% up to about H/C 0.6, then a steep drop.

The weaknesses: in Woolf, H/C explains only a third of the variation; if you take only well-carbonised samples (H/C < 0.5), the relationship almost disappears. For highly carbonised char, incubation mainly measures the decomposition of tar condensates on the surface – the extrapolation to 100 years is then largely an assumption.

The temperature-dependent 200-year variant

Registries use a further development that takes mean annual soil temperature into account (warmer soil = faster decomposition). Example: Isometric module “Biochar Storage in Agricultural Soils” v1.1, permitted only up to H/C_org < 0.5 and O/C_org < 0.2: F_durable(200) = 1 − [c + (a + b × ln T_soil) × H/C_org] with a = −0.383, b = 0.350, c = −0.048.

H/Corg	8 °C	10 °C	14 °C	21 °C
0.3	94%	92%	89%	85%
0.4	91%	88%	83%	79%
0.5	88%	84%	78%	72%

Method 2: random reflectance and the inertinite benchmark

Random reflectance (Rₒ) comes from coal and petroleum geology, where it has been ISO-standardised for decades. The sample is embedded in resin, polished and examined under a reflected-light microscope; what is measured is how much light each individual particle reflects – around 500 particles per sample. The more ordered the carbon structure, the brighter. A dedicated ISO standard for biochar is in preparation.

Above Rₒ = 2%, biochar behaves like natural inertinite. Three independent measurements support this: electrical conductivity rises sharply, the share of volatile matter levels off, and micro-Raman spectra match those of natural inertinite. Rule of thumb: below a production temperature of about 550 °C the char stays at the edge of the threshold; from about 600 °C it is safely above it. In the study by Sanei et al. (2024), 76% of the commercial biochars examined were above the threshold.

The calculation: inertinite fraction

Part of the sample is re-pyrolysed up to 650 °C – what escapes is the reactive fraction F_react. Under the microscope, the share of particles above Rₒ 2% is determined (F_Rₒ>2). This gives: F_inert = F_Rₒ>2 × (1 − F_react).

Example	FRₒ>2	Freact	Finert
Char A: mean Rₒ 3%, H/C 0.3, uniformly carbonised	1.00	0.03	97%
Char B: mean Rₒ 3%, H/C 0.3, but two particle groups	0.67	0.05	64%

Sanei explains this with an image: a char particle consists of concrete (inertinite, permanently stable), peanut butter (poorly carbonised particles below Rₒ 2%) and jam (reactive carbon in pores and on the surface). 100% inertinite does not exist. The reactive fraction correlates so closely with H/C that it can be calculated from H/C – saving a laboratory analysis.

H/C (elemental analysis)	Random reflectance (microscope)
+ fast, cheap, easy to standardise	− trained staff, slower, more expensive
− only one average for the whole sample	+ distribution of around 500 individual values
− can be distorted by additives (minerals, binders)	+ measures only charred particles
− cannot distinguish from fossil coal	+ proof of authenticity against lignite or petroleum coke; also measurable in soil and composite samples

Method 3: the unified model (Sanei et al. 2026)

In August 2026, “A unified framework for biochar carbon permanence” was published (International Journal of Coal Geology 324, 105075). Different materials – from raw wood through biochars to inertinite coal and graphite – are oxidised under accelerated conditions, the half-lives calibrated against soil trials and linked by a two-step curve. According to the paper: raw wood about 2 years, medium-carbonised biochars 4,380–9,150 years, inertinite coal about 125,000 years.

The key insight: the model explains the drop at H/C 0.5–0.7 that the linear models do not capture. Below H/C 0.5 it sits at 90–97%; at H/C 0.7 – exactly the EBC quality limit – only about 10% remains after 100 years. To put this in context: the curve rests on few calibration points, has not yet been independently confirmed and assumes 20 °C – so for cool soils in north-west Europe it tends to be cautious.

Which value applies? Our accounting rule

H/Corg	Woolf	Azzi	Sanei (unified)	Range	Balance value (minimum)
0.3	85%	94%	~95%	10 pts	85%
0.4	79%	94%	~90%	15 pts	79%
0.5	73%	93%	~81%	20 pts	73%
0.6	66%	92%	~39%	53 pts	39%
0.7	60%	91%	~10%	81 pts	10%

How to read the table: at H/C 0.4 it hardly matters which model you use – with the lowest (79%) you are on the safe side. At H/C 0.7 one model says 91%, another 10%. Anyone selling 90% here without an Rₒ measurement risks the sink later turning out to be largely non-existent.

Worked examples: 1 t of biochar with 70% organic carbon, soil temperature 10 °C

Maximum storable: 0.70 t C × 44/12 = 2.567 t CO₂

Case	Calculation	Share	t CO₂	Period
A: H/C 0.40, H/C method only	1.048 − 0.423 × 0.40	87.9%	2.26	200 years
B: same char, Rₒ shows two groups (S = 0.67; N = 500; Freact 0.05)	(0.67 − 0.021) × 0.95	61.7%	1.58	1,000 years
C: uniformly carbonised (S = 1.0; Freact 0.03)	1.00 × 0.97	97.0%	2.49	1,000 years
D: H/C 0.60 (e.g. temperature too low)	Minimum of the three models	~39%	~1.00	100 years

When does an Rₒ measurement pay off? When revenue from 1,000-year tonnes exceeds revenue from 200-year tonnes plus laboratory costs. For uniformly well-carbonised char (case C) almost always – more and higher-value tonnes. For mixed char (case B) only when 1,000-year tonnes fetch at least 1.43 times the price (2.26 ÷ 1.58). And in case D, better process control helps more than any measurement. So Rₒ is not just an upgrade tool – it also uncovers batches that are worse than their H/C value suggests.

What we do with it

Situation	Recommended method	Why
Industrial plant, ≥ 600 °C, H/C < 0.4	H/C per batch + Rₒ spot check; with a good distribution, 1,000-year tonnes	All models agree; Rₒ yields more and higher-value tonnes
Small plant / group, H/C 0.4–0.5	H/C + conductivity, 200-year method; Rₒ as a spot check	Cheap, permitted, the spot check catches outliers
Kon-Tiki / TLUD, H/C 0.5–0.7	Rₒ mandatory – or use only as a soil improver without credits	Models differ by 50–80 points
Blended products (char + minerals / compost)	H/C of the pure char before blending + Rₒ in the final product	H/C in the blend is distorted
Building materials / composites	Rₒ	The only method that measures the char within the composite
Field monitoring after years	Rₒ in soil samples	Detects the char and its condition in the soil

• Conservative accounting rule: always use the lowest value of the three models.
• Process window for pilot plants: ≥ 600 °C, H/C_org < 0.4, O/C_org < 0.2 – then model uncertainty disappears.
• Three-stage measurement concept: H/C per batch, continuous proxy measurement on site (e.g. electrical conductivity), Rₒ as a random spot check.
• Proof of authenticity: Rₒ microscopy rules out admixture of lignite or petroleum coke.

Regulatory framework (as of October 2026)

• EU CRCF – Delegated Regulation (EU) 2026/285, in force since 7 May 2026: two parallel routes – a 200-year decay function depending on H/C_org and the mean annual temperature at the application site, and the random reflectance method for the stable fraction (~1,000 years). A reconciliation between the two is still missing.
• Isometric – option 1: Woolf model with soil temperature (200 years, only H/C < 0.5); option 2: Rₒ > 2%, at least 500 measurements (1,000 years). The options may not be combined.
• Global Biochar C-Sink – distinguishes persistent aromatic carbon (millennia) and a semi-persistent fraction; recommends Rₒ and hydropyrolysis.
• EBC – H/C_org < 0.7 as a quality criterion. This limit says nothing about permanence of 100 years or more.

Note: values read from presentation slides are marked “~”. Please check details of the Delegated Regulation against the legal text before using them in applications.

Sources

• Sanei, H. (2026): IBI webinar “Measuring Biochar Permanence: H/C, Random Reflectance and Long-term Carbon Storage”, Aarhus University (LOC), September 2026.
• Sanei, H. et al. (2024): Assessing biochar’s permanence: An inertinite benchmark. International Journal of Coal Geology. pub.geus.dk
• Sanei, H. et al. (2026): A unified framework for biochar carbon permanence. International Journal of Coal Geology 324, 105075.
• Woolf, D. et al. (2021), Environmental Science & Technology; Azzi, E. S. et al. (2024), Geoderma.
• Isometric: Biochar Storage in Agricultural Soils, module v1.1.
• Delegated Regulation (EU) 2026/285; Regulation (EU) 2024/3012 – EUR-Lex