• Open Access

A method for screening the relative long-term stability of biochar


Correspondence: A. Cross, tel. +44(0)131 651 7165, e-mail: andrew.cross@ed.ac.uk


Biochar is being actively explored as a tool for long-term soil carbon sequestration. However, in order for this to be effective the long-term environmental stability of biochar must be assured. Here, we define and test an accelerated ageing method that seeks to reflect the oxidative nature of biochar degradation in soil. The method was applied to a systematic set of biochar samples produced from sugarcane bagasse, and a set of biochar samples produced from four different biomass sources. The stability of carbon in these samples was found to range between 41.6% and 76.1%, loosely correlating with biochar O : C ratio (r = 0.73). Increasing intensity of oxidative treatment eliminated more carbon. It also increased surface O : C ratio in a manner reported for naturally aged charcoal samples. The method effectively discriminated biochar produced under contrasting pyrolysis conditions and could be used as a proxy for environmental ageing of approximately 100 years under temperate conditions.


Biochar is increasingly recognized as a valuable tool for carbon abatement, with pyrolysis–biochar systems potentially offering greater carbon-equivalent gain than bioenergy only (Hammond et al., 2011). As biochar may offer additional benefits to soil fertility (Sohi et al., 2010), there is clear potential to integrate pyrolysis–biochar into systems such as sugarcane for bioethanol (Gaunt & Lehmann, 2008).

However, these analyses depend on assumptions of the absolute and relative stability of different biochar products in soil. The critical level of stability is context specific, and for climate change mitigation may be the component that remains in the soil after at least 100 years (Shackley & Sohi, 2010). Irrespective of the quantitative specification of ‘stable’ (e.g. hundreds as opposed to thousands of years), the influence of both feedstock and production conditions on the physico-chemical properties of biochar and the properties of the receiving soil will be relevant (Lehmann et al., 2009). To develop and inform strategy for using biochar as a carbon sequestration tool at the gigatonne scale (Lehmann, 2007), it is important that the relative stability of different biochar can be directly assessed.

Existing approaches to assessing biochar stability have focused either on extrapolation from short-term decomposition studies (Kuzyakov et al., 2009; Zimmerman, 2010), or on investigations of archaeological or environmental charcoal (Cheng et al., 2008). However, there are some fundamental limitations with these approaches concerning the heterogeneity apparent from real-time decomposition studies. Short-term carbon loss clearly does not follow simple first-order decay, so extrapolation of residual long-term loss must be bound by large error (Sohi et al., 2010). Information derived from 13C nuclear magnetic resonance (Baldock & Smernik, 2002), proximate analysis (Joseph et al., 2009) and elemental analysis (Spokas, 2010) have also been proposed as indicators of stability. Although instructive, these approaches remain qualitative (Spokas, 2010). None of the approaches provide a direct analogue for degradative environmental processes over the long-term, or an aged residue that can be used for calibration with reference to natural materials (such as wildfire charcoal). The methods may not properly account for the influence on biochar carbon stability of physical microstructure or mineral constituents.

We propose a practical, laboratory-based method to reliably discriminate different types of biochar according to their long-term stability, based on a controlled, accelerated ageing that is also amenable to calibration and tuning. Our objective is to provide an analogue for the slow, oxidative (abiotic and enzyme-mediated) degradation of recalcitrant carbon substrates that occur in soil (Willmann & Fakoussa, 1997; Hofrichter et al., 1999; Hamer et al. 2004; Nguyen et al., 2010).

Approaches to discriminate other carbon fractions in soil based on this principle have been previously described (Loginow et al., 1987; Skjemstad et al., 1993; Blair et al., 1995; Schmidt et al., 2001; Eusterhues et al., 2005). In this Technical Advance we seek to explore the potential to compress soil-equivalent oxidation of biochar on decadal to centennial timescale, into a treatment period of less than 1 week.

Materials and methods

Accelerated ageing was conducted using a combination of thermal and chemical oxidation, applied to 19 samples of biochar created at the same continuous-flow slow pyrolysis facility (Pacific Pyrolysis, Somersby, Australia) at contrasting temperatures and kiln residence times (with one sample activated by steam injection). The majority of samples were derived from sugarcane (Saccharum officinarum) bagasse. For broader comparison, the procedure was then applied to biochar from sugarcane trash (unharvested leaves) and other samples from different feedstocks made in the same facility and studied elsewhere (e.g. van Zwieten et al., 2010; Lin et al., 2012). These include Triticum spp. (wheat straw) chaff, Eucalyptus spp. (oil mallee) wood, urban green waste and poultry manure. Biochar properties and production characteristics for all samples are summarized in Table 1 and Table S1 in Supporting Information. Materials providing extreme reference points with respect to C stability in soil and the wider environment, commercially produced graphite and humic acid were also tested.

Table 1. Pyrolysis parameters used in the production of biochar from sugarcane biomass for the study and four other biochar samples produced in the same equipment also at 550 °C
Sample numberFeedstockProduction parameters
Heating rate ( °C min−1)Final temperature ( °C)Kiln residence time (mins)
1Sugarcane bagasse 5–1035020
2Sugarcane bagasse5–1035040
3Sugarcane bagasse5–1035080
4Sugarcane bagasse5–1045040
5Sugarcane bagasse5–1055020
6Sugarcane bagasse5–1055040
7Sugarcane bagasse5–10 (activated)55040
8Sugarcane bagasse10–2055040
9Sugarcane bagasse5–1055080
10Sugarcane trash5–1055040
11Wheat chaff5–1055045
12Oil mallee5–1055045
13Green waste5–1055045
14Poultry manure5–1055045

Determination of stability

To provide a measure of minimum stability, the constraints provided to oxidation by physical macrostructure (that in the natural environment could also be short term) were removed prior to the ageing of all samples by milling to a fine powder in a ball mill (Retsch MM200, Retsch, Haan, Germany, oscillating at 25 s−1 for 3 min). Samples were dried overnight at 80 °C to eliminate moisture and thereafter stored in a desiccator. The concentration of C (and H) in each sample was determined using an elemental analyser (Carlo Erba NA 2500, Carlo Erba, Rodano, Italy). An amount of each milled sample containing 0.1 g C was then weighed (4 d.p.) into a glass test tube. 0.01 mol of H2O2 (Certified analytical reagent, Fisher Scientific, Loughborough, U.K.) in a solution of 7 ml deionized water was then added to the test tube, while agitating the test tube to ensure all biochar was in suspension. Tubes were then heated to 80 °C for a thermal oxidation step, during which they were agitated regularly (2–3 times daily). After 2 days of heating, the solution had been evaporated to dryness and the tubes were then placed into an oven at 105 °C overnight. After cooling in a desiccator the tubes were reweighed to determine mass loss. As residual mass and carbon content must be correlated and the variance in carbon content in line with that of repeated analysis of a single sample (see Supporting Information, Table S3), we made carbon determination postageing on a combined residue. The %C result was used in conjunction with the mass loss measured for each replicate to express biochar carbon stability (Æ) as the proportion of initial carbon remaining after treatment:

display math

where Br is the residual mass of biochar following oxidation and BrC its %C content, Bt the initial mass of biochar prior to treatment and BtC the %C content of the biochar.

Treatment with 0.01 mol of H2O2 had previously been found to eliminate 1.7% of charcoal (Supporting Information, Table S1), produced in a traditional manner from hardwood, heating in a kiln at approximately 400 °C for 24 h (Case et al., 2012). Assuming similarity between this sample and natural wildfire charcoal which is typically produced at temperatures of 300 to 600 °C (Swift et al., 1993) and based on the study of Lehmann et al. (2008) in a subtropical environment (mean annual temperature 27 °C) this loss would have occurred between 25 and 161 years, based on a range of low and high mean residence times, respectively (see Supporting Information Table S4), similar to a range of residence times observed in other studies of charcoal stability (e.g. Ohlson et al., 2009; de Lafontaine & Asselin, 2011).

Statistical analysis

Comparisons of C stability between biochar samples were made using one-way analysis of variance with Tukey's post hoc test, with the threshold confidence level set at 0.1%. Statistics were carried out using SigmaPlot software (Windows Version 11, London, UK).

Elemental ratio and stability

To compare the results of accelerated ageing against a proposed predictor of biochar stability, O : C molar ratio (Spokas, 2010), the O content of the unoxidized biochar samples were estimated by difference (excluding ash content), from the results of elemental analysis (for C and H) described above.

Ageing kinetics and XPS analysis

To understand the kinetics of the accelerated ageing method with respect to increasing quantities of oxidant, one selected sample (midproduction temperature, sample 4) was treated as above but applying 0.005, 0.010, 0.020 or 0.030 mol of H2O2 (in each case made up to 7 ml with deionized water). A charcoal sample was similarly subjected to increasing treatment intensities, with the residue examined for the surface properties reported for charcoal increasingly aged in the environment (Cheng et al., 2008). To reflect the kiln charcoal relevant to the study of Cheng et al. (2008), retort charcoal was used. This was produced from mixed hardwood (oak, ash and sycamore) at a maximum temperature of 800 °C with a burn cycle of approximately 12 h. Subsamples of 0.2 g (0.15 g C) were ground as stated earlier and treated with either 0.005, 0.010, 0.020 or 0.030 mol of H2O2. Surface O : C ratio was determined for each residue using X-ray photoelectron spectroscopy (XPS). Samples were mounted on to stainless steel sample holders and then gently pressed onto adhesive tape. XPS spectra were obtained using a VG Sigma Probe (VG Scientific Ltd., East Grinsted, UK) and Al Kα radiation (1486.6 eV). During the analysis, the pressure in the test chamber was maintained at approximately 1.33 × 10−6 Pa. Survey spectra were recorded with the analyser pass energy set to 80 eV. Approximately 20 single region scans were recorded per element, with the pass energy of the analyser set to 10 eV. The spectra were corrected for charging by referencing the aliphatic C 1 s peak of hydrocarbons to 284.6 eV. Using Casaxps software (Casa Software Ltd., Teignmouth, UK), and surface composition was determined as the area under elemental peaks applying built-in sensitivity factors provided. Spectra were fitted using Gaussian-Lorentzian peak shapes with a ratio of 70 : 30. A Shirley background was subtracted for the quantitative analysis.

Results and discussion

Results from accelerated ageing of the systematic sample set produced from sugarcane bagasse (1–9; Table 1) show a clear relationship between key pyrolysis parameters and biochar C stability (Fig. 1). Stability increased significantly with pyrolysis temperature with all samples produced at 550 °C considerably more stable (68.7–77.4%) when compared with those produced at 350 °C (41.6–48.3%). Increasing kiln residence time to 80 min significantly increased stability for the sample produced at 350 °C (sample 1). However, there was no significant effect of residence time at higher pyrolysis temperatures (550 °C), suggesting an interaction between pyrolysis temperature and kiln residence time with respect to biochar C stability.

Figure 1.

The effect of two pyrolysis parameters on the stability of carbon in biochar produced from sugarcane bagasse; heating rate was 5–10 °C min−1 unless otherwise stated. Error bars (where visible) are one standard error of the mean (n = 3).

Carbon in biochar produced from other feedstock displayed a range of stability (65.1–76.1%) that was similar to the bagasse-derived samples produced at the same 550 °C pyrolysis temperature (Fig. 2). Chicken manure provided the least stable biochar, biochar from sugarcane trash the most stable. Biochar produced from green waste, sugarcane bagasse and oil mallee showed intermediate levels of stability with no significant differences between them; biochar produced from wheat chaff was slightly less stable than that from sugarcane trash. The application of the ageing method to unpyrolysed reference materials indicated stability close to 100% for graphite and close to 80% for commercially produced humic acid. We are not fully able to explain the partial stability of the two unpyrolysed feedstock materials treated in our experiments. However, we propose that physical access of the oxidant was impeded at the cellular scale by intact cell walls. A different dilution or heating regime might be required to eliminate substrates where physical access to structure has not been opened by pyrolysis, or possibly a more destructive preparation step (e.g. grinding in liquid nitrogen).

Figure 2.

The stability of biochar samples produced from a range of feedstock pyrolysed at 550 °C with a residence time between 40 and 45 min. Error bars (where visible) are one standard error of the mean (n = 3). Means headed by the same letter denote no significant difference.

Across all samples tested, i.e. the systematic set and the additional samples, the coefficient of variation associated with the method was approximately 1.0%. This high level of precision enabled the method to establish even the relatively small differences in stability resulting from pyrolysis time, as well as the larger effect of kiln residence time and heating rate (Fig. 1).

Treatment with 0.01 mol of H2O2 eliminated 1.7% of the carbon in a sample of wood charcoal, an extent of loss calculated to occur for the natural analogue (wildfire charcoal) over 25–161 years with a mean annual temperature of 27 °C (Lehmann et al., 2008; see Supporting Information). Using the ‘most likely’ figure within this range (45 years) and applying the Q10 commonly used in soil carbon modelling (Q10 = 2.0), our method would exert ageing equivalent to 92 years and 187 years under mean annual temperatures of 17 and 7 °C respectively. In fact, climosequence work has suggested that charcoal may have a higher Q10 (Cheng et al., 2008), which if adopted would lead to a higher estimate of recalcitrance in these cooler environments (See Supporting Information, Table S4).

The proportion of stable C decreased as more oxidant was used in the ageing procedure (Fig. 3). This relationship is important to scope how further loss of biochar might occur over greater longer periods of time in the environment. However, the incremental effect on biochar degradation diminished with additional oxidant, implying a set of heterogeneous subcomponents. If this is confirmed in further work, then a projection of a residence time for biochar C using linear extrapolation of the nominal 50–200 year loss provided by the standard oxidation treatment (above) is likely to be conservative with respect to its capacity to store carbon long term. The progressive degradation of biochar with increased oxidant concentration was, for a charcoal sample, also associated with increasing surface oxidation (Fig. 4). The trend in surface O : C was commensurate with that observed for charcoal in the climosequence study of Cheng et al. (2008). In the Cheng et al. (2008) study, charcoal samples had oxidized in soil for approximately 130 years under strongly contrasting environmental conditions.

Figure 3.

The relationship for sample no. 4 between the amount of oxidant used in accelerated ageing and the proportion of biochar carbon that remained stable. Error bars (where visible) are one standard error of the mean (n = 3).

Figure 4.

The relationship between increasing intensity of accelerated ageing of charcoal and surface O : C ratio measured by X-ray photoelectron spectroscopy.

It has been previously shown that biochar produced at lower pyrolysis temperatures has a less-ordered microstructure (resembling amorphous C), with lower cluster sizes and more reactive microsites than biochar created at high temperature, displays lower stability over a 1 year incubation (Nguyen et al., 2010). At the micro- and nanoscale, degradation that does occur likely begins at accessible sites, for example particle surfaces and the edges of individual carbon layers or sites with defect structures in carbon layers, primarily due to the greater energy state around edges (Pierson, 1993; Nguyen et al., 2010). Stability could be to a large extent determined by the degree of aromatic linked sheet development, and the incidence of these defect structures, both of which are also influenced by pyrolysis production temperature (Leifeld, 2007; Spokas, 2010). It is probable that the accelerated ageing method developed and tested in this Technical Advance captures well these physico-chemical factors.

As the preferential elimination of oxygen (and hydrogen) is a feature of the pyrolysis process and the reason for the high carbon content of biochar, the ratio of O : C has been proposed as an indicator and defining threshold for the relative stability of biochar. However, although there is correlation between stable biochar carbon and O : C ratio, the relationship is weak (r = 0.73) and not predictive (Fig. 5). We propose that the accelerated ageing method captures the physical constraints to degradation that are not reflected in elemental ratios, i.e. the micromorphology of biochar arising from the feedstock, or the composition and physical distribution of its mineral constituents.

Figure 5.

The relationship between the stability of biochar, biochar feedstock and two reference materials and the molar ratio of carbon and oxygen.

It is also possible that residues of the oxidation approach established in this Technical Advance could be used to understand interactions that might occur between biochar, soil and plant roots in the environment, or the influence of physical occlusion and protection of biochar in soil microaggregates (Eusterhues et al., 2005; Nguyen et al., 2008). The results obtained using the method remain independent of site-specific conditions that may apply in the locations where biochar might be deployed, but enables screening of biochar products for their relative stability against which other factors such as soil, climate and management can be superimposed. The potential to extend the quantitative calibration of the stability assessment is considerable, as the surface properties of the oxidized residue can be compared against that of charcoal naturally aged in the environment, over a range of timescales.

In Conclusion, the method presented in this Technical Advance may provide an analogue for the long-term, oxidative degradation (ageing) of biochar in soil. It offers a rapid, practical and precise measurement that is able to discriminate the relative stability of biochar products produced from a range of different feedstock at the same pyrolysis temperature, the same feedstock pyrolysed at different temperatures and even the same feedstock pyrolysed at the same temperature but for contrasting periods of time. It reveals patterns distinct from molar O : C ratio and, importantly, holds potential for calibration against patterns of ageing observed in the natural environment.


This work was supported by funding from Shell Global Solutions and we thank Christian Davies and Gordon Lethbridge for their contribution to overseeing this research, Maria Borlinghaus and Clare Peters for their assistance with sample analysis and Ronald Brown for XPS analysis. Biochar samples were produced by Pacific Pyrolysis Pty., Somersby, Australia, with the systematic set procured by Shell Global Solutions and other samples provided by Lukas van Zwieten. The authors would also like to acknowledge the contribution of two anonymous reviewers for their contribution to the refinement of this manuscript.