• Open Access

Co-generated fast pyrolysis biochar mitigates green-house gas emissions and increases carbon sequestration in temperate soils


  • Catherine E. Stewart,

    Corresponding author
    1. Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO, USA
    • USDA/ARS, Soil-Plant-Nutrient Research Unit, Fort Collins, CO, USA
    Search for more papers by this author
  • Jiyong Zheng,

    1. State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Northwest A & F University, Yangling, Shaanxi, China
    Search for more papers by this author
  • Jorin Botte,

    1. Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO, USA
    2. Department of Soil and Crop Sciences, Colorado State University, Fort Collins, CO, USA
    Search for more papers by this author
  • M. Francesca Cotrufo

    1. Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO, USA
    2. Department of Soil and Crop Sciences, Colorado State University, Fort Collins, CO, USA
    Search for more papers by this author

Correspondence: Catherine E. Stewart, tel. + 970 492 7270, fax + 970 492 7213, e-mail: catherine.stewart@colostate.edu


Char is a product of thermochemical conversion of biomass via pyrolysis, together with gas (syngas), liquid (bio-oil), and heat. Fast pyrolysis is a promising process for bio-oil generation, which leaves 10–30% of the original biomass as char. Char produced for soil application, is defined biochar (BC), and it may increase soil C storage, and reduce soil emissions of greenhouse gases (GHG), such as N2O and CH4 –potentially making fast pyrolysis bioenergy generation a C-negative system. However, differences in production conditions (e.g., feedstock, pyrolysis temperature and speed, post handling, and storage conditions) influence the chemical properties of BC and its net effect when added to soils. Understanding if fast pyrolysis BC can increase C sequestration and reduce GHG emissions will enable full assessment of the economic value and environmental benefits of this form of bioenergy. We characterized a BC produced by fast pyrolysis for bio-oil generation and examined GHG (CO2, N2O and CH4) efflux, C partitioning using δ13C, and soil C sequestration across four temperate soils and five BC rates; 0%, 1%, 5%, 10%, and 20% w/w. The fast pyrolysis process created a highly aromatic, low N, ash-rich BC with a O : C ratio of 0.01, which we expected to be highly recalcitrant. Across soils, CO2 emissions increased linearly and N2O emissions decreased exponentially with increasing BC addition rates. Despite still being actively respired after 2 years, total BC-derived C-CO2 comprised less than the BC volatile C content (4%). Expressed as CO2 equivalents, CO2 was the primary GHG emitted (97.5%), followed by N2O. All GHG emissions were small compared to the total SOC sequestered in the BC. Fast pyrolysis produced a highly recalcitrant BC that sequestered C and reduced GHG emissions. The recovery and soil application of BC would contribute to a negative carbon balance for this form of bioenergy generation.


The earth is experiencing an unprecedented warming due to the increase in greenhouse gases (GHGs, e.g., CO2, CH4 and N2O) in the atmosphere (IPCC, 2007; Montzka et al., 2001). Mitigation strategies include energy-neutral fuels, such as biofuels, alongside the implementation of land management strategies which increase C sequestration in soils and reduce net CH4 and N2O emissions (Paustian et al., 1997; Follett, 2001; Venterea et al., in press). The use of biochar – the high C solid fraction remaining after thermochemical conversion of biomass – as a soil amendment may be a successful strategy to increase soil C sequestration and to concomitantly reduce net soil emissions of N2O and CH4 (Lehmann et al., 2006; Woolf et al., 2010).

Depending on production parameters, bio-oil generation can produce varying amounts of biochar for soil application, potentially making it a carbon negative form of energy (Lehmann et al., 2006). Thermochemical conversion can either be optimized for energy (e.g., gasification and fast pyrolysis) with low BC yield (10–30%) or for higher BC yields 30–60% (e.g., slow pyrolysis) (Laird et al., 2009). Gasification typically produces ash, with little residual BC (Lehmann 2007). In fast pyrolysis, the proportion of bio-oil, BC and heat products depends on feedstock particle size, reaction temperature, and gas flow rate through the system (Brown, 2009) and consequently the BC produced has a range in chemical characteristics (Brewer et al., 2011; Krull et al., 2009; Spokas, 2010). Fast pyrolysis may result in incomplete biomass conversion producing a BC with more bio-available C (i.e., higher O : C ratios) than slow pyrolysis (Bruun et al., 2011; Spokas et al., 2011). On the other hand, the higher temperatures generally used during fast pyrolysis (500–600 °C for fast-pyrolysis vs. 350–550 °C slow pyrolysis; Brown, 2009) might produce BCs with a higher level of condensation and C-chemical resistance. The proportion of bio-oil vs. BC as well as the chemical composition of the BC produced will impact the economic value of bioenergy production and its environmental benefits.

Production parameters (slow vs. fast, pyrolysis temperature, post-handling, etc.) and feedstock characteristics (nutrient and cation content, lignin content, etc.) combine to produce BCs with a continuum of physical and chemical characteristics that subsequently influence gas flux after addition to soil. Because hemicellulose, cellulose, and lignin all have predictable thermal decomposition trajectories (i.e., Yang et al., 2007), pyrolysis temperature has often been used as a surrogate for BC stability in soil (Spokas, 2010). Increasing pyrolytic temperature generally results in a lower volatile matter content, lower O : C ratios, greater aromaticity and predicted half-lives – resulting in a more recalcitrant BC (Krull et al., 2009; Spokas, 2010; Brewer et al., 2011). Production processes that remove vapors during pyrolysis, such as those found in pyrolysis and gasification systems, can produce BC with smaller amounts of volatile matter compared to those left to cool with vapors intact (Spokas et al., 2011). Woody feedstocks tend to produce BCs with lower ash content and pH as compared with grass BCs (Spokas et al., 2011; Brewer et al., 2011). However, there is a wide range of BC production characteristics and feedstocks, resulting in uncertainty around the net effect of BC addition on GHG fluxes (Scheer et al., 2011).

Biochar and soil properties control the effects of BC amendments on soil C sequestration and GHG emissions, but consistent patters have yet to be elucidated. The addition of BC to soil can decrease, increase, or have no effect on net GHG emissions (Spokas & Reicosky, 2009; Spokas, 2010; Scheer et al., 2011; Karhu et al., 2011; Zheng et al., 2012). Total soil CO2 emissions generally increase after BC addition (Spokas & Reicosky, 2009; Smith et al., 2010; Bruun et al., 2011; Jones et al., 2011; Rogovska et al., 2011; Zheng et al., 2012), with BC produced by slow pyrolysis at low temperature stimulating the highest soil CO2 efflux (Zimmerman, 2010). This increase is somewhat expected due to the addition to soil of a large amount of organic carbon (i.e., biochar) and does not necessarily affect the potential benefit of BC amendments for soil C sequestration. The question to ask is whether or not the addition stimulates the decomposition of native soil organic C [i.e., priming sensu Fontaine et al. (2003)] in excess of the added BC-C, and how long the added C will be stored in the soil. The interaction of BC with other forms of soil organic matter (SOM) results in both priming and stabilization (Wardle et al., 2008; O'Donnell et al., 2009; Zimmerman et al., 2011). The mechanism is not clear, with both the preexisting C status of SOM and the C-chemistry of BC being shown to play a role in the stability and stabilization potential of BC (Kimetu & Lehmann, 2010).

The beneficial effects of BC additions to soil may also result from reducing other GHG emissions, such as N2O and CH4 (Zheng et al., 2012), but not in all cases (Taghizadeh-Toosi et al., 2011; Rogovska et al., 2011). In other studies though, no effects (Scheer et al., 2011), or no consistent effects among different soils (Spokas & Reicosky, 2009; Clough et al., 2010) were observed. The CH4 exchange data from incubation studies of aerobic soils suggest a decrease or no effect in CH4 uptake with BC addition (Spokas & Reicosky, 2009). However, a 9 t ha−1 BC addition to agricultural field soils increased CH4 uptake by 96% (Karhu et al., 2011). Taken together as CO2 equivalents, Zheng et al. (2012) found in an incubation experiment that the magnitude of the total decrease in GHG flux was dependent on N2O under N-fertilized treatments and on CO2 in the other non-fertilized soils. The variety in each GHG response to added BC and the variety in response between systems emphasize the need for evaluation of all GHGs.

The majority of the above studies used BC produced by slow pyrolysis. Biochar co-generated from bio-energy production (e.g., by fast pyrolysis or gasification) needs to be assessed with respect to GHG emission and C sequestration to fully evaluate the environmental benefits of bioenergy (i.e., make the process carbon-negative) and its economic feasibility (see Field et al., this issue). Because GHG flux is also dependent on the soil properties, studies also need to focus on effects on soils of specific interest. The scope of this work was (1) to investigate if BC co-generated from bioenergy production can reduce net GHG emissions and stimulate C sequestration when added to temperate soils, and (2) to determine those relationships over increasing BC addition rate. We hypothesized that fast pyrolysis would generate a chemically recalcitrant BC, resistant to microbial degradation, which would stimulate long-term C sequestration in soil. We anticipated that the above effects would increase with the increasing rate of BC additions, and with decreasing rates of organic carbon in soils. We also hypothesized that increasing BC addition rate would decrease N2O emissions and CH4 uptake.

Materials and methods

Soils and biochar

To examine the environmental benefit of a co-product of bio-oil generation, we used a BC produced at the National Renewable Energy Laboratory in Golden, CO from oak pellets by fast pyrolysis, at 550 °C, with primary vapors undergoing cracking at 850 °C for the production of bio-oil (Baldwin et al., 2012). Because different-sized biochars may have different nutrient contents, pH, and surface area characteristics (A. Gas, personal communication, 2010), the BC was sieved to 250 μm to obtain a fine (<250 μm) and a coarse (>250 μm) BC size fraction (Zheng et al., 2012). Due to resource constraints, this experiment was conducted only on the <250 μm size fraction (Table 1).

Table 1. Physical and chemical properties of the oak-derived biochar (BC) used in this study
 BC < 250 μm
C (%) 56.07
N (%) 0.22
O (%) 0.70
H (%) 0.85
C : N253
O : C 0.012
H : C 0.015
δ13C−27.06 ± 0.15
IC (mg C g−1 BC) 3.12 ± 0.00
pH 10.5
Volatile matter (%) 4.42
Ash (%) 41.93
Surface area (m2 g−1)116.8

To partition the CO2 efflux into a BC-derived and a native SOC derived flux, we used temperate soils with a natural 13C abundance significantly different from the BC (Tables 1 and 2). The four temperate soils differed in texture, C and N content and pH (Table 2). The incubation soils were a Weld silt loam (fine, smectitic, mesic Aridic Argiustoll) from the Central Plains Experimental Research Station (Colorado, USA) (CO); an Oxyaquic Hapludalf sampled near Sioux City, IA (IA); an Aeric Haplaquept from the Saginaw Valley Research Farm, MI Michigan (MI); and a Normania loam (fine-loamy mixed mesic Aquic Haplustoll) from the Lamberton Experimental Research Station (Minnesota, USA) (MN) (Huggins & Fuchs, 1997; Stewart et al., 2008; Zheng et al., 2012). All soils except Colorado had been cropped to continuous corn for at least 15 years before sampling (Stewart et al., 2008). Soil samples were collected from the A-horizon (usually 0–20 cm depth) and were screened to remove large roots and stones. Soils were then air-dried, sieved to a 2 mm mesh size, homogenized, and stored at room temperature, prior to use.

Table 2. Soil properties of the four incubation soils
SiteMATMAPSandSiltClayTOCTICδ13C (‰)Total N (%) 
(°C)a(mm)b(g 100 g soil−1)pH
  1. a

    MAT, mean annual temperature.

  2. b

    MAP, mean annual precipitation.

  3. c

    Zheng et al. (2012).

  4. d

    Stewart et al. (2008).

Colorado (CO) 9.632370.6 9.420.00.68c0.07c−12.660.078.717
Iowa (IA)10.654168.521.9 9.61.14d0.39−
Michigan (MI) 8.578812.−21.640.188.266
Minnesota (MN) 8.966039.927.632.41.97d−15.980.196.663

Soil & BC analyses

Soil elemental composition (%C, %N) and BC C isotope ratio (δ13C) were analyzed using a Europa Scientific automated nitrogen carbon analyzer (ANCA-NT) with a Solid/Liquid Preparation Module (Dumas combustion sample preparation system) coupled to a Europa 20-20 Stable isotope analyzer continuous flow isotope ratio mass spectrometer (Europa Scientific Ltd., Crewe, England). Inorganic C was determined using the pressure-transducer method (Sherrod et al., 2002) and pH was measured in water (5 : 1) on BC and soils, before and after BC addition, on air-dried 2 mm sieved samples.

The BC C, N, O, H, ash, and volatile matter concentrations were determined using proximate and ultimate analyses by Hazen Research, Inc (Golden, CO, USA), and surface area using BET analysis by Pacific Surface Science, Inc (Ventura, CA, USA) (Table 2). In addition, BC chemical composition was characterized using pyrolysis-gas chromatography-mass spectrometry (py-GC/MS) at 700 °C (Frontier pyrolyzer - Shimadzu QP-2010SE) with a SHRIX-5ms column (30 m length × 0.25 mm ID, 0.25 μm film thickness). Tri-tert-butyl benzene was added as an internal standard. The initial column temperature was 40 °C with a 1 min-hold followed by a 7 °C min−1 ramp to 300 °C and a final 5 min hold. The mass spectrometer detection range was 50–600 m/z with the interface temperature held at 300 °C. Compounds were identified using the NIST 2011 mass spectral library.

Experimental design

To determine a wide range of GHG responses, we used a factorial experimental design, with soil type and BC addition rate (0%, 1%, 5%, 10%, and 20% by weight) as factors, and with each treatment in four replicates. These additions represent a field application rate of 24–480 Mg ha−1 (assuming an incorporation depth of 20 cm and a bulk density of 1.2) and the higher addition rates would be above what could be practically applied to agricultural fields. Incubation units consisted of 75 g of soil, amended with 0, 0.75, 3.75, 7.5 and 15 g of BC placed in a plastic specimen cup. To ensure consistent soil water content across all treatments, we determined the water holding capacity (WHC) of soils for each soil × addition rate combination using the Wilcox method (Colman, 1947) and maintained each treatment at 60% of the determined WHC. Water holding capacity differed significantly between soils (P < 0.0001) and addition rates (P < 0.0001, Table 3). Soil + BC were pre-incubated in a refrigerator overnight before being placed in airtight jars and incubated in the dark at 25 °C.

Table 3. Average soil water holding capacity [g H2O (g soil + BC)−1] after five rates of BC addition (0%, 1%, 5%, 10%, and 20% by weight) for four incubation soils
g H2O (g soil + BC)−1
  1. Lowercase letters indicate significant differences between all soil x BC addition treatments.


CO2 efflux measurements and partitioning

The CO2 efflux from each unit was measured periodically (on days 1, 3, 8, 15, 21, 31, 43, 65, 95, 158, 187, 222, 277, 307, 369, 430, 495, 523, 564, 606, 648, and 699) by determining the CO2 concentration accumulated in the jar headspace, during the time elapsed between samplings. Sampling times were chosen to maximize experimental efficiency, while assuring that jars remained well oxygenated. After thoroughly mixing the air in the jar's headspace, 2 ml of air was sampled using a syringe and CO2 concentration measured using infrared gas analyzer (IRGA, LICOR model LI6252, Lincoln, NE). At day 1, 15, 31, 95, 187, 277, 369, 523, 606, and 699, an additional 20 ml subsample was taken from jar headspace and injected into a 12 ml pre-evacuated vial (Exetainer, Labco Limited, High Wycombe, Buckinghamshire, UK) for δ13C-CO2 analysis on a Gas Chromatograph–Isotope Ratio Mass Spectrometer (VG Optima). On a few occasions, samples had a CO2 concentration below the Optima detection limit and were analyzed on a Trace GC Ultra and PreCon-Delta V IRMS (Thermo Scientific, San Jose, CA, USA). Afterwards, sampling jars were opened and well aerated before being closed airtight until the following sampling. Isotope analyses were carried out the day after the gas sampling. CO2 and isotope measurements were calibrated using external calibration curves from gas standards. We used the stable isotope mixing model approach (Balesdent & Balbane, 1992) to partition soil and BC respiration, as follows:

display math

where fBC is the fraction of CO2 derived from the BC; δmix and δsoil are the δ13C of the CO2 emitted by the soil + BC mixtures and by the soil only, respectively, at the same sampling date; and δBC the δ13C of the original BC (Table 1). For the sampling days when δ13C-CO2 was not measured, we estimated it by linear interpolation between the δ13C-CO2 values measured for the same unit at the nearest previous and subsequent sampling day. The fraction (f) of CO2 derived from soil respiration was calculated as follows:

display math

Respiration values are reported (mg CO2 per unit) by adding the CO2 efflux measured at each sampling on an experimental unit base. The total CO2 derived from BC and native SOC were calculated by multiplying fBC and fsoil, respectively, for the CO2 efflux at each sampling, and by summing the obtained values to the end of the experiment (Day 699).

Measurements of N2O and CH4 fluxes

On the same dates of soil respiration measurements, a 20 ml subsample was taken from jar headspace and injected into a 12 ml pre-evacuated vial (Exetainer, Labco Limited, High Wycombe, Buckinghamshire, UK) for N2O and CH4 analysis. These analyses were performed using a fully automated gas chromatograph (Varian Model 3800, Varian Inc., Palo Alto, CA, USA) with a flame ionization detector (FID) to quantify CH4 and an electron capture detector to quantify N2O (Halvorson et al., 2008, 2010). After year 1, the FID method was lengthened to detect ethylene using external standards, but no ethylene was detected in any of the samples. All measurements were calibrated using external calibration curves from gas standards with variation less than 2%.

Data analyses

Biochar and soil effects on daily and cumulative total CO2, BC-derived CO2, SOC-derived CO2, N2O, and CH4 were evaluated using a repeated measures design in PROC MIXED in SAS (Cary, NC, USA) with day, soil, and BC addition rate as fixed factors. Cumulative values were assessed using an ANOVA at the end of the incubation in PROC GLM using soil and BC addition as main effects. Correlation analysis between cumulative N2O, soil pH, percent C, percent N, and C : N ratio was performed using PROC CORR in SAS. Treatments were considered significantly different at a 0.05 level with a Tukey's adjustment. Biochar addition rate effects on CO2, N2O, and CH4 were modeled in Sigmaplot 11 using linear regression or a two- [y = a*e(−b*x))] or three- [y = a*e(b/(x+c)] parameter exponential model. Stepwise regression modeling was performed by site to determine the most parsimonious model that explained cumulative N2O efflux. The best model was determined using AIC criterion to balance model complexity and explanatory power and included variables water holding capacity, soil pH after 1 year, soil + BC C, soil + BC N, soil + BC C : N ratio, and cumulative respired CO2.


Biochar characterization

The fast pyrolysis BC was chemically recalcitrant, as revealed by the high fixed C content, low N and O content (Table 1), and dominantly aromatic composition (Fig. 1). This consisted of single, double, triple, and quadruple C ring structures (i.e., toluene, naphthalene, anthracene, and pyrene) and lacked any pyrolysis products of ligno-cellulose or sugars typically observed in a wide variety of biomass-derived BCs from fast and slow pyrolysis (Kaal & Rumpel, 2009; Fabbri et al., 2012). The biochar recalcitrance is reflected in the very low O : C ratio of 0.012 and low volatile matter 4.42%. The surface area was 116.8 m2 g−1. Increasing rates of BC addition to the soils significantly increased water holding capacity (P < 0.0001, Table 3) and soil pH (P < 0.0001, Table 4).

Figure 1.

Pyrolysis-GC/MS chromatogram from the oak-derived biochar pyrolyzed at 700 °C. 1 = toluene, 2 = ethylbenzene, 3 = styrene, 4 = naphthalene, 5 = phenyl, 6 = IS, tri-tert-butyl benzene, 7 = anthracene, 8 = 3-ring C phenanthene, 9 & 10 = pyrene.

Table 4. Average soil pH after five rates of BC addition (0%, 1%, 5%, 10%, and 20% by weight) for four incubation soils
  1. Lowercase letters indicate significant differences between all soil x BC addition treatments.

0 08.717j8.502i8.266f6.663a
1 08.598cd8.158g8.150g5.998k

CO2 efflux

Soil respiration differed among soils in control units and total CO2 efflux increased from 141 mg CO2-C in CO to 157 mg in MIS, 187 mg in IA and, 199 mg in MN by the end of the experiment. Biochar addition to these soils initially suppressed and then stimulated CO2 efflux throughout the incubation (Fig. 2). In general, a very unusual CO2 efflux dynamic was observed through time, in particular for the 10% and 20% BC addition rates, with three distinct phases: (1) first 60 days, (2) 60–500 days, (3) 500–700 days, each characterized by an initial acceleration in efflux rate followed by some degree of leveling (Fig. 2). This BC effect differed between the soils resulting in a significant day × BC × soil interaction (Fig. 2). Overall, after nearly 2 years of incubation, BC increased cumulative CO2 efflux across the four soils by 8%, 36%, 88%, and 226%, respectively for the 1%, 5%, 10%, and 20% addition rate, as compared to the control (i.e., no BC addition, P < 0.0001).

Figure 2.

Cumulative mg CO2-C and standard deviation (n = 3) for the four soils with five rates of BC addition (0%, 1%, 5%, 10%, and 20% by weight) over the 699 days of incubation. Inlaid boxes show the cumulative CO2-C to 60 days.

Biochar- vs. soil-derived CO2

Total BC-derived CO2 significantly increased with increasing BC addition rate (P < 0.0001) and differed among soils (P < 0.0001; Fig. 3). Unlike total respiration, across all treatments, MN had the lowest total BC-derived CO2, 33.34 mg BC-C respired and IA the greatest, 262.96 mg BC-C respired. Across BC additions, cumulative SOC-derived CO2 differed significantly with BC addition rate (P < 0.0278) and between soils (P < 0.0001) (Fig. 3). Also, their interaction was significant (P = 0.0045) (Fig. 3), due to a decreased SOC-derived CO2 (i.e., negative priming) in CO, IA and MI, but increased (i.e., positive priming) in MN. Across all soils, the 20% BC addition significantly decreased soil respiration compared to the 0%, 5% and 10% BC addition rate by 19%, 20%, and 28%, respectively (P = 0.0278). The priming effect of the 20% BC addition rate was exponentially related to the initial SOC content [y = −74.6 + 0.05*e(4.07*x), R2 = 0.99 d, P = 0.021] over the entire incubation, with high negative priming at low soil C% (e.g., CO), and positive priming at high soil C% (e.g., MN).

Figure 3.

Partitioning of cumulative respired CO2-C between soil-derived and BC-derived for the five rates of BC addition (0%, 1%, 5%, 10%, and 20% by weight) for the four incubation soils.


Biochar addition suppressed N2O efflux between 21.3% and 91.6% throughout the incubation (Fig. 4), but the majority of the decrease occurred within the first 3 months (Table 5). The addition of BC significantly decreased both cumulative (P < 0.0001) and daily N2O efflux (P < 0.0001) from these soils, but differed between soils (soil × BC interaction, P < 0.0001). Cumulative μg N2O-N decreased from 53.9%, 72.4%, 76.3%, and 83.5% for the 1%, 5%, 10% and, 20% BC, respectively, compared to the control soils and were significantly different from one another, except for the 10%. Suppression of the rate of N2O efflux was evident throughout the incubation and still after 2 years, N2O flux from all BC addition rates were significantly lower compared to the control (Table 5, P < 0.0001).

Figure 4.

Cumulative greenhouse gas efflux (mg CO2-C, ng N2O-N and μg CH4-C) for the four soils with five rates of BC addition (0%, 1%, 5%, 10%, and 20% by weight). Carbon dioxide data were fit with a linear model, and N2O data were fit with an exponential model. Data without lines had no significant model fits over BC addition rate.

Table 5. Mean μg N2O-N day−1 for day 3, 187, 369 and 699 after five rates of BC addition (0%, 1%, 5%, 10%, and 20% by weight) for four incubation soils
SoilBCDay 3Day 187Day 369Day 699
μg N2O-N day−1
  1. Lowercase letters indicate significant differences between all soil x BC addition treatments.

CO 02.190c0.086d0.045d0.025d
IA 06.308a0.425a0.217a0.115a
MI 00.931de0.160b0.083b0.045b
MN 00.665ef0.041fg0.022ef0.014ef


Biochar addition (P < 0.0001), soil (P < 0.0001), and their interaction (P < 0.0001) significantly affected cumulative CH4 emissions (Fig. 4). There was no generalizable effect of BC addition on CH4 flux, with 10% BC increasing and 1% BC decreasing CH4 emissions compared to the 0BC addition. The stronger control of cumulative CH4 emissions was from soil type. Both CO and IA were strong CH4 sinks (−29.67 and −11.12 cumulative μg CH4-C) whereas MI and MN were slight sources (0.55 and 0.60 cumulative μg CH4-C, respectively).

Total GHG and soil C sequestration balance

To account for the soil respiration and trace-gas losses and compare them to the net C sequestration, we expressed each treatment as CO2 equivalents using a 100-year time span (Fig. 5). In these unfertilized soils, the dominant GHG component was CO2 (>92.5%), and although the reduction in N2O emissions with BC addition was significant, N2O comprised less than 7.5% of the total GHG emitted. Total CO2 efflux increased significantly with increasing BC addition, but overall comprised less than 10% of total soil C sequestered (expressed as CO2 equivalents) in the 5%, 10%, and 20% BC additions and around 32% in the 1% BC additions.

Figure 5.

Global warming potential (CO2 equivalents) for the five rates of BC addition (0%, 1%, 5%, 10%, and 20% by weight) for all four soils in the incubation.


BC characteristics

The fast pyrolysis process using a woody biomass created a highly aromatic, low N, ash rich BC, which based on its chemical composition was expected to be fairly recalcitrant to microbial degradation. In general, BCs have an O : C ratio under 0.6, and their residence time is inversely related to O : C ratio (Krull et al., 2009; Spokas et al., 2010). The wood-derived fast pyrolysis BC used in this experiment, with an O : C ratio of 0.01 (Table 1), falls in the ‘soot’ category of the BC continuum with a predicted half-life of >1000 years (Spokas et al., 2010; Hedges et al., 2000). The complete lack of identifiable lignocellulose or carbohydrate biomarkers and the presence of aromatic structures (toluene, naphthalene) in the pyrolysis products suggests this BC, although created at a moderate temperature (550 °C) is very altered compared to some characterized to date (Kaal & Rumpel, 2009; Fabbri et al., 2012) and more indicative of a BC that was intensively charred, and consequently entirely aromatic (Knicker et al., 2005). A py-GC/MS analysis of 20 BCs and their respiration rate when added to soil found that BCs with more sugar, protein, and lignocellulosic pyrolysis products tended to have greater CO2 respiration, perhaps from the unaltered or slightly thermally altered biomass structures (Fabbri et al., 2012; Brewer et al., 2011). Those BCs with an O : C less than 0.05 had lower respiration rates, a high degree of aromaticity in the pyrogram, and increased resistance to surface oxidation (Fabbri et al., 2012). This highly recalcitrant BC is the type suggested for long-term C sequestration in soils (Lehmann, 2007; Laird et al., 2009).

Fast pyrolysis with a solid residue of only 10% of the initial biomass, was expected to produce BC with high ash content and consequently, high pH. This was confirmed in our study where BC had a pH of 10.5 (Table 2), and after application to the soil, significantly increased soil pH after 1 year (Table 4). In the field, biochar addition to soils may reduce agricultural lime application (Novak et al., 2009; Laird et al., 2010; Galinato et al., 2011). Biochar addition to soils increases water holding capacity (Novak et al., 2009; Karhu et al., 2011), and here we observed a 10%, 17%, and 25% increase in water holding capacity in the 5%, 10%, and 20% BC additions, which may contribute to increased crop yields (Major et al., 2010). On the basis of chemical characteristics, our results suggest that this fast pyrolysis BC would be best applied to acidic or neutral soils.

BC addition effects on CO2 efflux & priming

Given the chemical recalcitrance of the BC used in this experiment we did not expect it to stimulate soil respiration. Thus, the initial suppression of CO2 efflux with increasing BC addition rate was expected. However, the increase in respired C from 6 months to the second year in all BC-amended soils was surprising (Fig. 2). Even more so is the fact that after nearly 2 years of incubation, the 10% and 20% BC addition rates in IA and MI do not yet appear to have leveled off, and our isotope analyses show that nearly all these CO2 fluxes are derived from BC, implying that BC-C has become the main microbial C source in these systems.

Biochar-derived respiration increased linearly with increasing BC addition rate, and interacted with native SOC respiration (Fig. 3). In the CO, IA, and MI soils BC substituted native SOC as a substrate for microbial decomposition, resulting in negative SOC priming. By contrast in MN, BC was hardly respired by microbes which utilized SOC and resulted in positive SOC priming. Negative (Zimmerman et al., 2011; Zimmerman 2010, Jones et al., 2011; Keith et al., 2011), as well as positive (Keith et al., 2011; Luo et al., 2011) priming of native SOC mineralization has been documented in other laboratory incubations after BC addition to soils. For example, when examining 3 soils and 16 BC additions, Spokas & Reicosky (2009) found little consistent effect on CO2 production, with some types of BC promoting and others inhibiting CO2 production. In addition, BC decomposition was reported to interact with the decomposition of other soil amendments, in general stimulating and being stimulated by labile C substrate additions (Hamer et al., 2004; Calfapietra et al., 2010).

A mechanistic understanding of these interactions is still missing, yet some important factors can be implicated. The initial quantity and possibly the chemical composition of the SOM appear to play an important role (Spokas & Reicosky, 2009; Kimetu & Lehmann, 2010), as indicated also by our study, with BC-C representing a more abundant substrate for microbial degradation in SOC poor soil (e.g., CO), but also a microbial protection/stimulating agent for the degradation of SOC in rich soils (e.g., MN). We found an exponential relationship between initial SOC and cumulative soil C primed by BC addition (R2 = 0.99, P = 0.021). The ability of BC to interact with enzyme activity (Bailey et al., 2011) and to stabilize labile organic matter (Keith et al., 2011) may also play a key role in these interactions, and requires specific mechanistic investigations.

Although BC has been shown to both stimulate and repress decomposition of some SOM in soils, it has been argued that over the long term, repression of SOM mineralization may prevail (Kuzyakov et al., 2009; Zimmerman et al., 2011). Even when BC addition to soil increases, dissolved organic carbon leaching or respiration losses are not sufficiently high to alter C sink benefits from BC additions (Bell & Worrall, 2011). In our study after 2 years of incubation, total BC derived CO2-C corresponded to between 0.3% (MN) and 3.2% (CO, IA, MI) of the initial BC-C content. These amounts are comparable to the volatile matter fraction (4%, Table 1) of our fast-pyrolysis BC, a fraction which is commonly used to assess the ‘lability’ of a BC (Zimmerman, 2010). In our study, the BC-C losses are also comparable to the amount of inorganic carbon (IC) in the BC (3%; Table 1). Inorganic C was also shown to contribute to short-term abiotic CO2 emissions from char (Jones et al., 2011), and it was likely responsible for the high BC-derived CO2 emissions measured at the very beginning of the incubation (data not shown). Our hypothesis of fast-pyrolysis of a woody biomass producing a recalcitrant char, suitable for long-term soil C sequestration was confirmed, but further experiments should be done to quantify the microbial vs. abiotic contributions to CO2 efflux.

BC and trace gas flux

The 21–89% reduction in both cumulative and daily N2O efflux throughout the incubation after BC addition corroborates our hypothesis and the many other studies that show a decrease in N2O emissions after BC addition both in laboratory incubations and in the field (Spokas & Reicosky, 2009; van Zwieten et al., 2009; Singh et al., 2010; Taghizadeh-Toosi et al., 2011; Rogovska et al., 2011). After 414 days incubation, Rogovska et al. (2011) found an 80–88% decrease in N2O emissions in soils amended with 5, 10 and 20 g kg−1 of hardwood charcoal, both with and without added manure. In an analysis of 16 differing BCs in three soils, Spokas & Reicosky (2009) found that all types of BC decreased N2O production, with the exception of a compost-amended, high N BC. An average of 70% N2O reduction from ruminant urine patches was observed in pastures amended with 30 t ha−1 (Taghizadeh-Toosi et al., 2011).

However, other studies have found no significant decrease in N2O production with BC addition. A recent field study using continuous GHG monitoring found that a 10 t ha−1 application of cattle feedlot BC did not decrease N2O emissions from a fertilized pastured Australian Ferrisol (Scheer et al., 2011). A laboratory incubation study documented an initial increase in N2O emissions, but no subsequent decrease after bovine urine application with a 20 Mg ha−1 addition of pine-derived BC (Clough et al., 2010). In the same MN soil used here, Zheng et al. (2012) found no significant reduction of N2O efflux after a 10% BC by weight addition in unfertilized treatments, but over 50% reduction after fertilizer addition.

Decreased N2O after BC addition has been attributed to changes in water, NH4+ and NO3 availability, increased soil pH, and VOC inhibition of nitrification (van Zwieten et al., 2009). The decreased N2O with BC addition in this study is unlikely to have been from VOC inhibition because the BC was very low in volatile matter (Table 1) and had a very small VOC signature obtained from thermal headspace desorption (Spokas et al., 2011). We found no presence of α-pinene, a nitrification inhibitor observed in some BCs (Clough et al., 2010) or ethylene, an ammonium nitrification inhibitor observed in VOC of some types of BC (Spokas et al., 2009).

The soils in this study represent a wide range in C, N, pH, and texture all of which potentially alter N2O fluxes. So, to identify the most important parameters for each site's cumulative N2O flux, we used stepwise regression modeling. Variables predicting N2O efflux were site-specific and included soil and BC, C and N content, pH and WHC (Table 6). The importance of the variables C : N ratio, and soil + BC-C and soil + BC-N suggests that increased BC addition rate decreased N availability, either through NH4+ sorption to BC (Lehmann et al., 2003) or by microbial immobilization of NO3 (Singh et al., 2010; Ippolito et al., 2011; Zheng et al., 2012). Because no additional N was added in this study, organic N would have to be the source of microbial N and decreased NO3 availability is likely to decrease N2O flux.

Table 6. Model selection results and explanatory variables for cumulative μg N2O-N. WHC = water holding capacity
SiteVariablePartial R2Model R2AIC
COCumulative CO20.100.4475.9
 C : N ratio0.090.5374.8
IASoil + BC C0.570.57101.1
MIpH 1 year0.760.7667.1
MNC : N ratio0.230.2326.6
 pH 1 year0.310.5421.8
 Soil + BC N0.090.6321.2
 Soil + BC C0.190.8213.8

Although BC addition increased soil pH in all soils throughout the incubation (Table 4), soil pH was the most important variable explaining variability in N2O flux only in MI (76%). The activity of N2O reductase (the enzyme reducing N2O to N2) is pH dependent (Fujita and Dooley 2007) and could explain the decrease in N2O efflux with increasing BC addition rate. Soil pH and total soil N (soil + BC + fertilizer N) were the two best predictors of N2O fluxes from fertilized and unfertilized incubation soils (Zheng et al., 2012). The negative correlation between N2O and WHC suggests that although soil water increased significantly, BC addition must have also increased pore space. The importance of different soil and biochar variables explaining N2O efflux in this incubation highlight the importance of accounting for C and N availability and soil chemical changes after BC addition. Modeling exercises that account for these soil and BC properties and nutrient transfers, such as DayCent or Roth-C, will be useful to predict future BC effects on GHG emissions.

In this experiment, the dominant driver for cumulative CH4 efflux was soil type, not BC addition rate. Minnesota and MI had very little CH4 efflux and both IA and CO were net CH4 sinks. In an incubation experiment, Zheng et al. (2012) found the same CO soil to be a net CH4 sink and that a 10% BC by weight addition in unfertilized treatments decreased CH4 uptake. In contrast, the MN soil had no significant efflux and no significant effect of BC addition. In a continuous GHG monitoring experiment of Australian pasture soils, Scheer et al. (2011) found no significant effect of a 10 t ha−1 application of cattle feedlot BC on CH4 uptake, but a 9 t ha−1 BC application to agricultural soils in Finland increased CH4 uptake by 96% (Karhu et al., 2011). The complex effect of BC on CH4 efflux is no doubt coupled to changes in soil bulk density, aeration, and C and N availability, and needs further mechanistic investigation.

Total GHG and soil C sequestration balance

The CO2 equivalent data from this experiment emphasize the importance of accounting for all GHGs as well as the BC-C sequestered when assessing BC additions to soils. The CO2 efflux comprised more than 92.5% of GHG emissions from these non-fertilized soils and increased with increasing BC addition rate. This CO2 loss comprised less than 10% of the BC-C sequestered (expressed as CO2 equivalents) in the 5%, 10% and 20% BC addition rates (equivalent to 120, 240 and 480 Mg ha−1). In the 1% addition rate (24 Mg ha−1), CO2 comprised 20–35% of the CO2 sequestered (expressed as CO2 equivalents). Even at the 1% addition rate, this fast-pyrolysis BC is an effective amendment for long-term soil C sequestration, but other, less recalcitrant chars may not be.

Cumulative N2O emissions decreased exponentially, up to 89% with increasing BC addition and N2O flux rates were still lower after nearly 2 years of incubation, but comprised less than 7.5% of the total GHG emitted as CO2 equivalents (Fig. 5). There are few other studies who have expressed GHG flux in terms of CO2 equivalents, but Zheng et al. (2012) also found that in non-fertilized treatments, CO2 was the primary GHG and N2O was the primary GHG in fertilized treatments. When assessing BC as an amendment for soil C sequestration and for GHG reduction, it will be important to consider the detailed chemical composition of the biochar, the C and N status of the soil and biochar, and the rate of biochar addition. These data confirm that biochar can mitigate N2O emissions in a variety of temperate soils, but this result must be confirmed by field studies.


The authors gratefully acknowledge Dr. Kim Magrini at the National Renewable Energy Laboratory for providing the BC. Mary Smith, Melissa Reyes-Fox, Morgan Carey, and Kris Nichols for assistance with GC analyses. Two anonymous reviews provided helpful comment to earlier versions of the manuscript. This work is part of the USDA-ARS GRACEnet project. Funding was provided by State of Colorado Department of Agriculture ACRE contract No. 277777, CSU Clean Energy Supercluster, the Warner College of Natural Resources, and the China Science Fund (41071195). Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. USDA is an equal opportunity provider and employer.