Professional Documents
Culture Documents
Hale 2016
Hale 2016
Hale 2016
Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere
Review
h i g h l i g h t s g r a p h i c a l a b s t r a c t
a r t i c l e i n f o a b s t r a c t
Article history: The sorption strength of neutral organic compounds to charcoal, also called biochar was reviewed and
Received 4 May 2015 related to charcoal and compound properties. From 29 studies, 507 individual Freundlich sorption co-
Received in revised form efcients were compiled that covered the sorption strength of 107 organic contaminants. These sorption
28 July 2015
coefcients were converted into charcoal-water distribution coefcients (KD) at aqueous concentrations
Accepted 18 August 2015
Available online xxx
of 1 ng/L, 1 mg/L and 1 mg/L. Reported log KD values at 1 mg/L varied from 0.38 to 8.25 across all data.
Variation was also observed within the compound classes; pesticides, herbicides and insecticides, PAHs,
phthalates, halogenated organics, small organics, alcohols and PCBs. Five commonly reported variables;
Keywords:
Charcoal
charcoal production temperature T, surface area SA, H/C and O/C ratios and organic compound octanol
Biochar ewater partitioning coefcient, were correlated with KD values using single and multiple-parameter
Sorption linear regressions. The sorption strength of organic compounds to charcoals increased with increasing
Surface area charcoal production temperature T, charcoal SA and organic pollutant octanolewater partitioning co-
Temperature efcient and decreased with increasing charcoal O/C ratio and charcoal H/C ratio. T was found to be
Hydrophobicity correlated with SA (r2 0.66) and O/C (r2 0.50), particularly for charcoals produced from wood
Elemental composition feedstocks (r2 0.73 and 0.80, respectively). The resulting regression: log KD (0.18 0.06)
log Kow (5.74 1.40) log T (0.85 0.15) log SA (1.60 0.29) log OC (0.89 0.20)
log HC (13.20 3.69), r2 0.60, root mean squared error 0.95, n 151 was obtained for all
variables. This information can be used as an initial screening to identify charcoals for contaminated soil
and sediment remediation.
2015 Elsevier Ltd. All rights reserved.
* Corresponding author.
E-mail address: sarah.hale@ngi.no (S.E. Hale).
http://dx.doi.org/10.1016/j.chemosphere.2015.08.047
0045-6535/ 2015 Elsevier Ltd. All rights reserved.
66 S.E. Hale et al. / Chemosphere 144 (2016) 65e74
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2. Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.1. Scope of the review article . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.2. Charcoal-water distribution coefficients (KD values) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.3. Single-parameter and multiple-parameter linear regression analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.1. Summary of previously reported charcoal properties and their inter-correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.2. Summary of previously reported charcoal-water distribution coefficients (KD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.3. Single-parameter regression analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.3.1. Relationship between KD and charcoal physicochemical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.3.2. Relationship between KD and organic pollutant KOW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.4. Multiple-parameter linear regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.5. Biases in the single and multiple-parameter linear regressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4. Future work and implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
charcoal and the organic pollutant can act as either electron donors 2011; Sun et al., 2011; Ahmad et al., 2012; Chen et al., 2012;
or acceptors. For example, graphene units of charcoal can act as Graber et al., 2012; Sun et al., 2012b; Gomez-Eyles et al., 2013; Li
donors or other surface functional groups such as polycarboxylated et al., 2013; Qiu et al., 2013; Sun et al., 2013; Wang et al., 2013b; Xie
aromatics, charged heterocyclic amines and quinones can act as et al., 2013; Zhang et al., 2013; Zhao et al., 2013; Xiao et al., 2014)
acceptors. The strength of pep EDA interactions are a function of i) were used in this review. In the majority of cases the reported data
degree of condensation, ii) polycondensate sheet size and iii) extent was in the form of Freundlich sorption isotherms in which
to which sheet edges are substituted with electron withdrawing Freundlich sorption coefcients (KFr (mg/kg) (mg/L)n) and isotherm
oxygen containing functional groups (Zhu and Pignatello, 2005; nonlinearity were reported:
Sun et al., 2012b). H-bond interactions can also occur in charcoal-
n
organic pollutant system in cases where polar organic pollutants Ccharcoal KFr CW
like phenols, anilines, nitrocompounds and mono-
ourocompounds, which contain highly electronegative N, O, and F where Ccharcoal (mg/kg) is the concentration of pollutants on the
atoms and are able to form bonds with H donor functional groups charcoal at equilibrium, n () is the Freundlich exponent and
on the surface of charcoal (Endo et al., 2009; Sun et al., 2012b). CW (mg/L) is the concentration of pollutants in the water phase at
Isolating and quantifying sorption mechanisms for all types of equilibrium. In order to allow a comparison between reported data
charcoals in use today is very challenging and involves many and to remove the dependence of sorption on the nonlinearity of
different mechanisms that are affected by charcoal production the isotherms, average Freundlich sorption coefcients were con-
methods and biomass feedstocks. The rst step needed in this area verted to charcoal-water distribution coefcients (KD) by calcu-
is to gain an empirical overview about the variation in sorption lating them from the reported sorption isotherms at three
behaviour amongst different types of charcoals, as well as how this particular values of CW (where X 1 ng/L, 1 mg/L and 1 mg/L):
variability may be accounted for by key parameters describing the
Ccharcoal at CW X
charcoal. Thus, the state of current knowledge related to the KD at CW X ;
sorption of neutral organic compounds to charcoals is the focus of CW X
this data synthesis. The inuence of charcoal physical characteris- In the majority of cases experiments were carried out at the
tics and organic pollutant properties on the charcoal-water distri- mg/L concentration range and therefore we chose one concen-
bution coefcient were explored empirically through single and tration that was higher and one that was lower to investigate
multiple-parameter linear regressions. As such this synthesis whether relationships were applicable over this wide concen-
builds upon that reported by Ahmad et al. (2014) in which dis- tration range. A complete list of all KD values at 1 mg/L, 1 mg/L and
cussion around this began, but charcoal-water distribution co- 1 ng/L is presented in Table S1 of the Supplementary material
efcients were not reported. Such knowledge is important for the (SM). Data is presented by referring to calculated KD values at
practical use of charcoal in applications such as the remediation of these concentrations.
contaminated soils and sediments. This review is not intended to Four charcoal parameters (charcoal production temperature, SA,
provide an in depth mechanistic review of sorption mechanisms, O/C, H/C) and one organic compound parameter (the octanole-
but a synthesis and compilation of data and relationships between water partitioning coefcient, KOW) were chosen for correlation
parameters. analysis with KD values. These parameters were selected as i) there
are mechanistic links, albeit indirect, between these parameters
2. Materials and methods and KD, ii) these parameters are among those most often reported
in the literature, iii) they are also most commonly used in subse-
2.1. Scope of the review article quent data interpretation of charcoal KD values and, iv) are those
that can be controlled in charcoal production in order to produce a
Peer-reviewed literature about the sorption of neutral organic desirable product. As sorption of neutral organic compounds is
compounds to charcoal was reviewed and a comprehensive list of controlled by the surface functionality of the charcoal, H, C and O
charcoal-water distribution coefcients (KD values) was compiled. content is important, and also by the pores available for sorption,
Commonly reported variables related to charcoal production pro- SA is also a relevant parameter. Temperature of charcoal production
cess, charcoal physical properties and organic compound physico- is an underlying factor able to control these variables. In order to
chemical properties were also taken from these literature. Single avoid data bias and ensure full coverage and environmental rele-
and multiple-parameter linear regressions were constructed to vance all data was used in regressions (thus wide temperature
identify whether these commonly measured parameters could ranges, SA and elemental ratios were assured). For surface area
account for the variability of sorption behaviour. Note that this measurements, studies using N2 as the probe gas were considered
review does not focus on the sorption of ionic/ionizable compounds as there was only one study that reported CO2 surface area mea-
or metals to charcoal, (Beesley et al., 2011), charcoals formed from surements (Sun et al., 2013). This is noteworthy, as limited access of
hydrothermal carbonization (i.e., hydrochars (Berge et al., 2011)) or N2 to pores at the experimental temperature of 77 K means that the
the effects of other competing processes such as bi/multi-solute CO2 surface area measured at room temperature is likely more
systems and the presence of soil (and thus pore blocking and representative of the charcoal surface area that organic pollutants
sorption attenuation (Kwon and Pignatello, 2005)). can sorb to (Pignatello and Xing, 1996).
in order to identify the degree to which individual property pa- respectively). The parameters SA and O/C were correlated with a
rameters correlated with KD values. Correlations were carried out coefcient of determination of 0.31. The coefcient of determina-
on log transformed data as explained in the SM on page S2. In tion for other combinations (each of the ve parameters were
addition, the data was checked to see if it was better described as checked for correlation with all other parameters) of charcoal pa-
normally distributed or log-normally distributed data using Pear- rameters were much lower (r2 0.23 or less).
son's moment coefcient of skewness, g (see SM).
Multiple-parameter regression analysis was also carried out to 3.2. Summary of previously reported charcoal-water distribution
identify the degree to which all of the parameters could collectively coefcients (KD)
be used to predict KD. In addition, the inter-correlation between
parameters was investigated, the data was checked for skewness Tables S1 and S2 in the supplementary material provide a
and the multiple-parameter linear regressions were tested using a summary of the average KD values (thus considering the standard
validation set comprising 75% of the data set that was randomly deviation across studies), normalised to a water phase concentra-
selected. Regression analysis was conducted in Microsoft Excel tion of 1 mg/L for each organic compound where literature existed.
2013 using the Analysis ToolPak Add-on. Further details related to Reported KD values varied over eight orders of magnitude, from
all of the statistical methods and regressions are given in the sup- 0.38 (MTBE on maize charcoal) to 8.25 (PCB 149 123 on acai pit
plementary material. The discussion below is primarily focused on charcoal), depending on the compound and the charcoal. The range
log KD values normalised to a water concentration of 1 mg/L as the of average log KD values for different compound classes was from
majority of studies were carried out in this concentration range. All 3.40 to 5.69 for pesticides, herbicides and insecticides, from 5.26 to
results for log KD values normalised to 1 ng/L and 1 mg/L are given 7.16 for PAHs, from 4.47 to 5.41 for phthalates, from 4.49 to 6.38 for
in the SM. halogenated organics, from 3.14 to 4.18 for small organics, from 3.03
to 5.63 for alcohols and from 4.43 to 7.27 for PCBs. The compounds
for which the greatest amount of data points have been measured
3. Results and discussion
were phenanthrene (range of reported values 2.02e7.71, n 42),
and naphthalene (range of reported values 2.01e7.87, n 29). The
3.1. Summary of previously reported charcoal properties and their
range in this PAH data indicates that sorption to charcoal can vary
inter-correlation
by up to 5 orders of magnitude, though tends to cluster around a
median as in the case of phenanthrene (median: 5.82, interquartile
An overview of compiled charcoal properties and compound
range: 5.23e6.21).
properties specic to the KD values is presented in Table 1. The
reported H/C data ranged over three orders of magnitude (from
0.001 to 1.26), O/C data over two orders of magnitude (from 0.02 to 3.3. Single-parameter regression analysis
1.0), SA over three orders of magnitude (from 0.25 to 669 m2/g),
charcoal production temperature T from 373 to 1123 K and organic Eqs. (1)e(5) in Table 2 show the single-parameter regressions
compound KOW values from 1.0 to 4.1. The distribution of these for log KD values normalised to 1 mg/L for the complete data set
variables was investigated in order to ascertain if the reported data versus; log charcoal production temperature (Eq. (1)), log charcoal
distributed normally or log-normally (see SM). H/C and O/C ratios SA (Eq. (2)), log charcoal H/C (Eq. (3)), log charcoal O/C (Eq. (4)) and
were distributed more log-normally than normally, though SA and log organic compound KOW (Eq. (5)). Table S4a and b show the
T were distributed more normally. Log values were subsequently resulting single-parameter regressions for KD values calculated at
used for all parameters based on fundamental thermodynamic 1 ng/L and 1 mg/L. Coefcients of determination for Eqs. (1)e(5)
theory (Goss and Schwarzenbach, 2001), as explained in the SM. ranged from 0.05 for log O/C to 0.32e0.34 for log SA and log T.
With the exception of H/C ratio, all parameters exhibited substan- Fig. 1 shows the correlations of charcoal log KD with production
tial skewing (see the supplementary material). This skewing is due temperature T (n 349), charcoal SA (n 477), charcoal H/C ratio
to more data being available for some of the charcoals than others. (n 187), charcoal O/C ratio (n 184) and organic pollutant
As is evident from Table S2, the data for the variables cluster around octanolewater partitioning coefcient (n 507), plotted according
a median value for each compound, and the inter quartile ranges to feedstock materials (wood, grass and crop waste and other
are much narrower for particular compounds than the total ranges waste). Fig. S1 shows the corresponding data without such a
within a compound class. classication.
A regression analysis was also carried out to investigate if any of
the charcoal parameters were correlated with each other, the re- 3.3.1. Relationship between KD and charcoal physicochemical
sults of which are presented in Table S3aed in the SM. The pa- properties
rameters exhibiting the most correlation were charcoal production Charcoal production temperature and surface area were the two
temperature and SA (coefcient of determination, r2 0.66) and variables that showed the strongest correlation with log KD,
charcoal production temperature and O/C ratio (r2 0.50), partic- although r2 values were low (log KD 7.93 0.59
ularly for charcoals from wood feedstocks (r2 0.73 and 0.80, log T 17.37 1.70, r2 0.34, Eq. (1), Table 2 and
Table 1
Summary of previously reported charcoal and organic compound properties.
Table 2
Summary of statistical analysis for the single-parameter linear regression (Eqs. (1)e(5)) and the multiple-parameter linear regression (Eqs. (6)e(9)) for Log KD normalized to a
water concentration of 1 mg/L.
All log KD (7.93 0.59) log T (17.37 1.70) (1) 0.34 1.10 350
All log KD (0.91 0.06) log SA (3.67 0.13) (2) 0.32 0.97 472
All log KD (1.85 0.22) log HC (2.62 0.29) (3) 0.28 1.29 187
All log KD (0.78 0.26) log OC (4.47 0.20) (4) 0.05 1.46 184
All log KD (0.23 0.03) log Kow (4.34 0.14) (5) 0.13 1.11 507
All log KD (0.18 0.06) log Kow (5.74 1.40) 0.60 0.95 151
log T (0.85 0.15) log SA (1.60 0.29)
log OC (0.89 0.20) log HC (13.20 3.69)(6)
Grass and crop waste log KD (0.06 0.08) log Kow (5.91 1.76) 0.67 0.81 65
log T (0.09 0.19) log SA (0.49 0.44)
log OC (1.37 0.26) log HC (13.44 4.67)(7)
Wood log KD (0.04 0.10) log Kow (11.70 2.79) 0.81 0.78 55
log T (1.22 0.25) log SA (3.14 0.67)
log OC (0.01 0.45) log HC (28.10 7.05)(8)
Other feedstocks log KD (0.70 0.21) log Kow (5.83 4.05) 0.64 0.93 25
log T (1.07 0.77) log SA (1.00 0.61)
log OC (3.54 1.12) log HC (13.29 10.07) (9)
log KD 0.91 0.06 log SA 3.67 0.13, r2 0.32, Eq. (2), Table 2). Chen et al., 2012; Sun et al., 2012b; Xiao et al., 2014). The result-
On a statistical level, this is partly accounted for by the good degree ing single-parameter linear regression for this selected data set was
of correlation between log SA and log T as discussed below. Log KD 9.46 log T- 21.68 and the r2 value was 0.53, indicating a
Charcoal production temperature has been previously identied stronger correlation for these selected studies than within the
as a key variable affecting the resulting charcoal's physicochemical whole data set.
properties and therefore in turn the sorption of organic compounds For two of these studies (Chen et al., 2008, 2012) a continuous
(Chen et al., 2012; Rutherford et al., 2012; Xiao et al., 2014). During increase in KD with production temperature over the complete
pyrolysis, a series of complex reactions take place that convert temperature range (up to 700 C) was observed. For the remaining
aliphatic components contained in the charcoal precursor biomass four studies (Chen and Chen, 2009; Sun et al., 2011, 2012b; Xiao
to aromatic components in the resultant charcoal. The reactions et al., 2014) an increase in KD values was seen up to temperatures
proceed via the: i) degradation of hemicellulose and partial of around 400e600 C, where they decreased, before increasing
destruction of alkyl moieties (220e315 C), ii) degradation of cel- again. These trends are mirrored by the correlation of KD with SA
lulose (315e400 C) and iii) chain fragmentation of lignin to release (Chen and Chen, 2009; Sun et al., 2012b; Xiao et al., 2014). Several
monomeric phenol units (>400 C) (Chen et al., 2012; Uchimiya reasons were given for the increase, decrease and further increase
et al., 2013; Wang et al., 2013a; Mimmo et al., 2014). As the amor- in KD values by the authors of the studies, related both to the
phous aromatic components within charcoal begin to arrange into charcoal SA itself and to the properties of the organic compound.
turboclastic crystallites, the SA of the charcoal increases dramatically For example, the adsorption of phthalates to charcoals produced
(Chun et al., 2004; Chen et al., 2012; Wang et al., 2013a). Several from grass and wood (Sun et al., 2012b) was dependant on the size
authors have noted a peak in SA at around a production temperature of the phthalates which showed variable increases in adsorption
of 600 C due to a structural ordering and micropore coalescence relative to SA. The sorption strength of naphthalene and 1-naphthol
around this temperature (Mimmo et al., 2014) or due to a restruc- was dependent on the amount of available sorption sites which was
turing and onset of an ash melting process (Brown et al., 2006). not only affected by the absolute SA, but also by the overlap be-
Corresponding to this, a pore selective sorption mechanism (based tween the molecular geometry of the organic pollutants and the
upon the size of the charcoal pores and the organic compounds sorption sites (Chen and Chen, 2009).
themselves) often occurs for charcoals made at higher temperatures Charcoal elemental composition is also directly inuenced by
(Chen et al., 2008; Chen and Chen, 2009) and a polarity selective charcoal feedstock and production temperature as it is related to
mechanism (based upon the polarity of charcoal surface functional the reactions that take place as the biomass is converted to
groups and organic compounds that are sorbing to the charcoal) at charcoal. As pyrolysis temperature and time of heating increase,
lower temperatures. the amount of C increases, the amounts of H and O decrease and
The previously reported literature contained data for charcoals this leads to lower H/C and O/C ratios (Lehmann and Joseph, 2009;
produced between 100 and 850 C, thus providing a diverse data Xiao et al., 2014). This inter-relationship was observed in our
set and including materials dominated by both aliphatic and aro- study, with logelog correlation r2 values between T and H/C and
matic carbon (Sun et al., 2012b). Whilst the inclusion of charcoals O/C being 0.23 and 0.31, respectively. Within the data set
produced at temperatures as low as 100 C (5 data points) in the compiled here, the elemental H/C ratio was able to describe log KD
complete data set can be questioned, they were included for to a greater extent than the elemental O/C ratio (r2 0.28 and
completeness and to avoid bias. Fig. 2a shows the correlation of log r2 0.05, respectively); indicating that the loss of H from the
KD with charcoal production temperature for seven selected studies charcoal can be ascribed to a greater change in KD and is thus a
in which the charcoals were produced at minimally four different better (though still weak), indicator of increasing sorption ca-
temperatures from 100 to 800 C (Chen et al., 2008; Chen and Chen, pacity than the loss of O.
2009; Chen and Yuan, 2011; Sun et al., 2011, 2012b; Chen et al.,
2012; Xiao et al., 2014). Fig. 2b shows the correlation of log KD 3.3.2. Relationship between KD and organic pollutant KOW
and SA for six of these same studies where SA data was reported As well as those of the charcoal, the properties of the organic
(Chen et al., 2008; Chen and Chen, 2009; Chen and Yuan, 2011; pollutants (e.g. molecular weight and size, hydrophobicity,
70 S.E. Hale et al. / Chemosphere 144 (2016) 65e74
Fig. 1. The correlation of log KD vs. a) Log SA, b) Log T, c) Log H/C, d) Log O/C and e) log organic compound KOW for the data set split according to biomass feedstock (wood ( ), grass
and crop waste ( ) and other waste ( )).
polarity) can also affect the strength of sorption. The relationship information about size it does not provide information about
between Log Kow and Log KD is given in Eq. (5) in Table 2 spatial arrangement which is critical when pore lling takes place.
(log KD 0.23 0.03 log Kow 4.34 0.14, r2 0.13). The corre- Of the literature surveyed, only three studies contained data for
lation with log Kow was not as good as with log SA and log T, which the sorption of three or more compounds with varying hydro-
is indicative that the SA and T inuence the charcoal KD more than phobicity to the same charcoal (Chen et al., 2008; Lian et al., 2011;
the hydrophobicity of the contaminant itself. While it can be Sun et al., 2012b) and these are shown in Fig. 2c. The sorption of
questioned how applicable a relationship with KOW is for some of diethyl phthalate, dibutyl phthalate and butyl benzyl phthalate to
these compounds, KOW is an easily accessible parameter and pro- grass and wood charcoals was correlated with compound hydro-
vides a commonly used basis for correlation with other partitioning phobicity, but was also inuenced by charcoal physicochemical
coefcients. The molecular volume of a compound would also properties (Sun et al., 2012b). According to these authors, the
provide useful information with regards to binding to charcoal, strongest sorption combination was observed for dibutyl phthalate
however this parameter is rarely reported in sorption studies and (log Kow 4.8) to charcoals produced at 500 C that had amorphous
also suffers shortcomings. Whilst molecular volume provides carbon properties (and thus a possible additional absorptive
S.E. Hale et al. / Chemosphere 144 (2016) 65e74 71
Fig. 2. The correlation of log KD with a) log T for selected studies which contain sorption data for a charcoal produced at at least four temperatures, b) log SA for selected studies that
contain sorption data for a charcoal produced from the same feedstock, but at different temperatures (and therefore having variable SA) and c) log organic compound KOW values for
selected studies that compare the sorption of at least three compounds to the same charcoal.
Fig. 3. The correlation between log KD measured and log KD predicted using the correlation coefcients in Eq. (6) in Table 2 for KD values normalised to water concentrations of
1 mg/L, 1 mg/L and 1 ng/L. MLR Multiple Linear relationship (Table 2 Eqs. (6)e(9)).
mechanism), rather than for charcoals produced at the highest set divided by feedstock (Eqs. (6)e(9), Table 2 for log KD at 1 mg/L
production temperatures. and in Table S5aed for KD at 1 ng/L and 1 mg/L). Note that this type
In order to achieve maximum sorption, the molecular di- of multiple-parameter linear regression can only be carried out for
mensions of the organic pollutant must match those of the pores in isotherms in which all parameter data are available. The resulting
the charcoal. The pollutants themselves may have restrictive overall regression equation for the complete data set was;
shapes and therefore contact with the charcoal surface may be log KD 0.18 0.06 log Kow 5.74 1.40 log T 0.85 0.15
limited due to steric hindrance (Endo et al., 2009) and the distance log SA 1.60 0.29 log OC 0.89 0.20 log HC 13.20 3.69
the pollutant can penetrate into a given pore be reduced (Pignatello (Eq. (6), Table 2), with substantially improved coefcients of
and Xing, 1996). Size exclusion becomes more severe as sorption determination compared to the single-parameter linear free energy
volume becomes dominated by small micropores (Lattao et al., relationships (r2 0.60). It should be emphasized that this equation
2014) and thus larger compounds are not able to access sorption is not meant to describe all of the mechanistic interactions occur-
sites, as many pollutants have sizes in the same order of magnitude ring during sorption, but rather the best possible way of estimating
as charcoal nanopores (5e10 nm). As compound hydrophobicity this charcoal KD from commonly available data of relevance to
generally increases with increasing compound molecular diameter, sorption, and therefore provides a tool that can be used for
for a charcoal dominated by small micropores the sorption of larger screening purposes. Fig. 3 shows the correlation between measured
compounds would theoretically be impaired (Amstaetter et al., log KD values and modelled log KD values using the coefcients in
2012; Lattao et al., 2014). However, for the literature surveyed Eq. (6) and plotted according to feedstock. The multiple linear
here, pore size distribution of the charcoals was rarely reported. regression ts most of the data within a factor 30 (1.5 orders of
magnitude). For KD at 1 mg/L, 9% of data lie outside this factor (3% for
3.4. Multiple-parameter linear regression KD at 1 mg/L and 30% for KD at 1 ng/L). The present data were within
a similar variability range as that previously reported for contam-
Multiple-parameter linear regressions based on all ve param- inated sediments (Arp et al., 2009). The data points that lie furthest
eters were determined for the complete data set and for the data from the 1:1 line are mainly charcoals produced at temperatures
72 S.E. Hale et al. / Chemosphere 144 (2016) 65e74
below 350 C, as marked in Fig. 3, which is around the temperature was the strongest (r2 0.81), followed by grasses and crop waste
where amorphous aromatic components within charcoal begin to (r2 0.67), and then other feedstocks (r2 0.64). It is likely that the
arrange into turboclastic crystallites. Below 350 C charcoals may relative homogeneity of the wood charcoals (when compared to the
contain complex amorphous subdomains that exhibit decreased or other freedstocks) lead to the improved correlations. The other
increased sorption compared to carbonaceous fractions in the feedstock category contained a broad mixture of feedstocks sug-
charcoal, depending on their structure (e.g. humic materials sorb gesting the need to further separate charcoal feedstock in order to
less than charcoals but tars can sorb more strongly than charcoal) improve the prediction of sorption strength.
(Arp et al., 2009). It is also interesting to note that within the data
there is a greater difference in KD values for the same compound 3.5. Biases in the single and multiple-parameter linear regressions
produced at different temperatures than for different compounds
produced at the same temperature. For example the log KD for As mentioned above, the shortcoming of these multiple-parameter
naphthalene for orange peel biochar produced at 150 C was 2.1 linear regressions are that some of the assumed independent variables
compared to 7.3 for the same biochar and compound but with a are not completely independent of each other, such as SA and T. In
production temperature of 700 C. However the log KD value for 1- addition, the calibration data set showed non-normality of all pa-
naphthol for the same biochar produced at 150 C was 2.1 and was rameters except H/C, likely because some charcoals and compounds
6.7 for the same biochar produced at 700 C. were represented within the data set more than others. Thus, despite
The correlation statistics, in terms of r2 and rmse, differ from one the large data set, these biases will have some impact on the re-
another depending on the concentration the KD values are gressions themselves, regardless of the mechanistic basis. As dis-
normalized to, be it 1 mg/L (r2 0.46, rmse 0.71), 1 mg/L cussed in the SM, the inter-correlation between SA, T and OC, indicates
(r2 0.60, rmse 0.95) or 1 ng/L (r2 0.56, rmse 1.55). The best that not all of these parameters may be needed to make screening level
correlation, as quantied by the r2, is for 1 mg/L normalized data, estimations of charcoal log KD values. In Table 3, multiple linear re-
and the best model performance for predicting KD values, as gressions excluding either just log SA or both log SA and log O/C are
quantied by rmse, is for 1 mg/L normalized data. A likely expla- presented. As can be seen, the quality of the ts (as indicated by the
nation for the best r2 at 1 mg/L is that the majority of experiments coefcients of determination) are not as good as when considering all
were carried out in the mg/L water concentration range, and thus input parameters in the majority of cases, however the differences are
any errors in the reported Freundlich n values will lead to positive quite small. For example for wood charcoals removing O/C ratio from
or negative biases in log KD values. As evident from Fig. 3, the the statistical analysis does not change the quality of t, thus sug-
overall scatter of KD values is the smallest at 1 mg/L, which accounts gesting in this case a knowledge of T and H/C allows a prediction about
for having the best rmse despite potential extrapolation biases. sorption strength to be made.
Splitting the dataset up into different feedstocks (Eqs. (7)e(9), Multiple-parameter linear regressions for the complete data set
Table 2) improved the correlations compared to considering all were compared with those calculated based on 75% of the data set
feedstocks together (Eq. (6); r2 0.60). The correlation for wood (selected randomly) in order to tests the agreement between the
Table 3
Multiple-parameter linear regressions after considering multicollinearity (Eqs. (10)e(17)) as well as the multiple-parameter linear regressions using all variables (Eqs. (6)e(9)
from Table 2).
All log KD (0.18 0.06) log Kow (5.74 1.40) 0.60 0.95 151
log T (0.85 0.15) log SA (1.60 0.29)
log OC (0.89 0.20) log HC (13.20 3.69) (6)
All log KD (0.17 0.06) log Kow (9.66 1.06) 0.53 1.01 177
log T (1.44 0.26) log OC (1.05 0.20)
log HC (23.35 2.81) (10)
All log KD (0.19 0.06) log Kow (5.87 0.83) 0.47 1.09 182
log T (1.11 0.21) log HC (13.66 2.24)(11a)
Grass and crop waste log KD (0.06 0.08) log Kow (5.91 1.76) 0.67 0.81 65
log T (0.09 0.19) log SA (0.49 0.44)
log OC (1.37 0.26) log HC (13.44 4.67) (7)
Grass and crop waste log KD (0.03 0.08) log Kow (6.26 1.57) 0.65 0.80 75
log T (0.41 0.41) log OC (1.44 0.24)
log HC (14.30 4.17) (12)
Grass and crop waste log KD (0.02 0.08) log Kow (4.99 0.94) 0.65 0.80 76
log T (1.54 0.23) log HC (11.03 2.63) (13)
Wood log KD (0.04 0.10) log Kow (11.70 2.79) 0.81 0.78 55
log T (1.22 0.25) log SA (3.14 0.67)
log OC (0.01 0.45) log HC (28.10 7.05) (8)
Wood log KD (0.09 0.11) log Kow (18.35 2.50) 0.69 0.95 66
log T (2.46 0.75) log OC (0.07 0.42)
log HC (44.95 6.47) (14)
Wood log KD (0.07 0.11) log Kow (11.50 1.44) 0.66 1.00 71
log T (0.04 0.44) log HC (27.06 3.55) (15)
Other feedstocks log KD (0.70 0.21) log Kow (5.83 4.05) 0.64 0.93 25
log T (1.07 0.77) log SA (1.00 0.61)
log OC (3.54 1.12) log HC (13.29 10.07) (9)
Other feedstocks log KD (0.61 0.18) log Kow (1.04 2.61) 0.63 0.95 30
log T (1.44 0.43) log OC (3.15 1.00)
log HC (2.48 6.37) (16)
Other feedstocks log KD (0.80 0.17) log Kow (0.17 1.89) 0.42 1.13 35
log T (1.14 0.92) log HC (0.06 4.81) (17)
S.E. Hale et al. / Chemosphere 144 (2016) 65e74 73
James, G., Sabatini, D.A., Chiou, C.T., Rutherford, D., Scott, A.C., Karapanagioti, H.K., Sun, K., Kang, M., Zhang, Z., Jin, J., Wang, Z., Pan, Z., Xu, D., Wu, F., Xing, B., 2013.
2005. Evaluating phenanthrene sorption on various wood chars. Water Res. 39, Impact of deashing treatment on biochar structural properties and potential
549e558. sorption mechanisms of phenanthrene. Environ. Sci. Technol. 47,
Jonker, M.T.O., Koelmans, A.A., 2002. Sorption of polycyclic aromatic hydrocar- 11473e11481.
bons and polychlorinated biphenyls to soot and soot-like materials in the Sun, K., Keiluweit, M., Kleber, M., Pan, Z., Xing, B., 2011. Sorption of uorinated
aqueous environment mechanistic considerations. Environ. Sci. Technol. 36, herbicides to plant biomass-derived biochars as a function of molecular
3725e3734. structure. Bioresour. Technol. 102, 9897e9903.
Kasozi, G.N., Zimmerman, A.R., Nkedi-Kizza, P., Gao, B., 2010. Catechol and humic Teixido, M., Hurtado, C., Pignatello, J.J., Beltran, J.L., Granados, M., Peccia, J., 2013.
acid sorption onto a range of laboratory-produced black carbons (biochars). Predicting contaminant adsorption in black carbon (biochar)-amended soil for
Environ. Sci. Technol. 44, 6189e6195. the veterinary antimicrobial sulfamethazine. Environ. Sci. Technol. 47,
Khan, S., Chao, C., Waqas, M., Arp, H.P.H., Zhu, Y.-G., 2013. Sewage sludge biochar 6197e6205.
inuence upon rice (Oryza sativa L) yield, metal bioaccumulation and greenhouse Uchimiya, M., Orlov, A., Ramakrishnan, G., Sistani, K., 2013. In situ and ex situ
gas emissions from acidic paddy soil. Environ. Sci. Technol. 47, 8624e8632. spectroscopic monitoring of biochar's surface functional groups. J. Anal. Appl.
Kleineidam, S., Schuth, C., Grathwohl, P., 2002. Solubility-normalized combined Pyrolysis 102, 53e59.
adsorption-partitioning sorption isotherms for organic pollutants. Environ. Sci. Wang, T., Camps-Arbestain, M., Hedley, M., 2013a. Predicting C aromaticity of bio-
Technol. 36, 4689e4697. chars based on their elemental composition. Org. Geochem. 62, 1e6.
Koelmans, A.A., Jonker, M.T.O., Cornelissen, G., Bucheli, T.D., Van Noort, P.C.M., Wang, X.L., Xing, B.S., 2007. Sorption of organic contaminants by biopolymer-
Gustafsson, O., 2006. Black carbon: the reverse of its dark side. Chemosphere derived chars. Environ. Sci. Technol. 41, 8342e8348.
63, 365e377. Wang, Y.-S., Shan, X.-Q., Feng, M.-H., Chen, G.-C., Pei, Z.-G., Wen, B., Liu, T., Xie, Y.-N.,
Kong, H., He, J., Gao, Y., Wu, H., Zhu, X., 2011. Cosorption of phenanthrene and Owens, G., 2009. Effects of copper, lead, and cadmium on the sorption of 2,4,6-
mercury(II) from aqueous solution by soybean stalk-based biochar. J. Agric. trichlorophenol onto and desorption from wheat ash and two commercial
Food. Chem. 59, 12116e12123. humic acids. Environ. Sci. Technol. 43, 5726e5731.
Kwon, S., Pignatello, J.J., 2005. Effect of natural organic substances on the surface Wang, Y., Wang, L., Fang, G., Herath, H.M.S.K., Wang, Y., Cang, L., Xie, Z., Zhou, D.,
and adsorptive properties of environmental black carbon (char): pseudo pore 2013b. Enhanced PCBs sorption on biochars as affected by environmental fac-
blockage by model lipid components and its implications for N-2-probed sur- tors: humic acid and metal cations. Environ. Pollut. 172, 86e93.
face properties of natural sorbents. Environ. Sci. Technol. 39, 7932e7939. Warnock, D.D., Lehmann, J., Kuyper, T.W., Rillig, M.C., 2007. Mycorrhizal responses
Lattao, C., Cao, X., Mao, J., Schmidt-Rohr, K., Pignatello, J.J., 2014. Inuence of mo- to biochar in soil e concepts and mechanisms. Plant Soil 300, 9e20.
lecular structure and adsorbent properties on sorption of organic compounds to Xiao, L., Bi, E., Du, B., Zhao, X., Xing, C., 2014. Surface characterization of maize-
a temperature series of wood chars. Environ. Sci. Technol. 48, 4790e4798. straw-derived biochars and their sorption performance for MTBE and ben-
Lehmann, J., 2007. A handful of carbon. Nature 447, 143e144. zene. Environ. Earth Sci. 71, 5195e5205.
Lehmann, J., Joseph, S., 2009. Biochar for Environmental Management: Science and Xie, M., Lv, D., Shi, X., Wan, Y., Chen, W., Mao, J., Zhu, D., 2013. Sorption of mono-
Technology. Earthscan, London. aromatic compounds to heated and unheated coals, humic acid, and biochar:
Li, J., Li, Y., Wu, M., Zhang, Z., Lu, J., 2013. Effectiveness of low-temperature biochar Implication for using combustion method to quantify sorption contribution of
in controlling the release and leaching of herbicides in soil. Plant Soil 370, carbonaceous geosorbents in soil. Appl. Geochem. 35, 289e296.
333e344. Xu, C., Liu, W.P., Sheng, G.D., 2008. Burned rice straw reduces the availability of
Lian, F., Huang, F., Chen, W., Xing, B., Zhu, L., 2011. Sorption of apolar and polar clomazone to bamyardgrass. Sci. Total Environ. 392, 284e289.
organic contaminants by waste tire rubber and its chars in single- and bi-solute Yamato, M., Okimori, Y., Wibowo, I.F., Anshori, S., Ogawa, M., 2006. Effects of the
systems. Environ. Pollut. 159, 850e857. application of charred bark of Acacia mangium on the yield of maize, cowpea
Liang, B., Lehmann, J., Solomon, D., Kinyangi, J., Grossman, J., O'Neill, B., and peanut, and soil chemical properties in south Sumatra, Indonesia. Soil Sci.
Skjemstad, J.O., Thies, J., Luizao, F.J., Petersen, J., Neves, E.G., 2006. Black carbon Plant Nutr. 52, 489e495.
increases cation exchange capacity in soils. Soil Sci. Soc. Am. J. 70, 1719e1730. Yanai, Y., Toyota, K., Okazaki, M., 2007. Effects of charcoal addition on N2O emis-
Loganathan, V.A., Feng, Y., Sheng, G.D., Clement, T.P., 2009. Crop-residue-derived sions from soil resulting from rewetting air-dried soil in short-term laboratory
char inuences sorption, desorption and bioavailability of atrazine in soils. Soil experiments. Soil Sci. Plant Nutr. 53, 181e188.
Sci. Soc. Am. J. 73, 967e974. Yang, Y., Ligouis, B., Pies, C., Achten, C., Hofmann, T., 2008. Identication of carbo-
Mimmo, T., Panzacchi, P., Baratieri, M., Davies, C.A., Tonon, G., 2014. Effect of py- naceous geosorbents for PAHs by organic petrography in river oodplain soils.
rolysis temperature on miscanthus (Miscanthus x giganteus) biochar physical, Chemosphere 71, 2158e2167.
chemical and functional properties. Biomass Bioenerg. 62, 149e157. Zhang, P., Sun, H., Yu, L., Sun, T., 2013. Adsorption and catalytic hydrolysis of carbaryl
Pignatello, J.J., Xing, B.S., 1996. Mechanisms of slow sorption of organic chemicals to and atrazine on pig manure-derived biochars: impact of structural properties of
natural particles. Environ. Sci. Technol. 30, 1e11. biocharse. J. Hazard. Mater. 244, 217e224.
Qiu, Y., Wu, M., Jiang, J., Li, L., Sheng, G.D., 2013. Enhanced irreversible sorption of Zhao, X., Ouyang, W., Hao, F., Lin, C., Wang, F., Han, S., Geng, X., 2013. Properties
carbaryl to soils amended with crop-residue-derived biochar. Chemosphere 93, comparison of biochars from corn straw with different pretreatment and
69e74. sorption behaviour of atrazine. Bioresour. Technol. 147, 338e344.
Rondon, M.A., Lehmann, J., Ramirez, J., Hurtado, M., 2007. Biological nitrogen xa- Zheng, R.-L., Cai, C., Liang, J.-H., Huang, Q., Chen, Z., Huang, Y.-Z., Arp, H.P.H., Sun, G.-
tion by common beans (Phaseolus vulgaris L.) increases with bio-char additions. X., 2012. The effects of biochars from rice residue on the formation of iron
Biol. Fertil. Soils 43, 699e708. plaque and the accumulation of Cd, Zn, Pb, As in rice (Oryza sativa L.) seedlings.
Rutherford, D.W., Wershaw, R.L., Rostad, C.E., Kelly, C.N., 2012. Effect of formation Chemosphere 89, 856e862.
conditions on biochars: compositional and structural properties of cellulose, Zheng, W., Guo, M., Chow, T., Bennett, D.N., Rajagopalan, N., 2010. Sorption prop-
lignin, and pine biochars. Biomass Bioenerg. 46, 693e701. erties of greenwaste biochar for two triazine pesticides. J. Hazard. Mater. 181,
Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic 121e126.
Chemistry. Wiley-Interscience, New York. Zhu, D.Q., Pignatello, J.J., 2005. Characterization of aromatic compound sorptive
Spokas, K.A., 2010. Review of the stability of biochar in soils: predictability of O: C interactions with black carbon (charcoal) assisted by graphite as a model. En-
molar ratios. Carbon Manag. 1, 289e303. viron. Sci. Technol. 39, 2033e2041.
Sun, K., Gao, B., Ro, K.S., Novak, J.M., Wang, Z., Herbert, S., Xing, B., 2012a. Assess- Zimmerman, J.R., Werner, D., Ghosh, U., Millward, R.N., Bridges, T.S., Luthy, R.G.,
ment of herbicide sorption by biochars and organic matter associated with soil 2005. Effects of dose and particle size on activated carbon treatment to
and sediment. Environ. Pollut. 163, 167e173. sequester polychlorinated biphenyls and polycyclic aromatic hydrocarbons in
Sun, K., Jin, J., Keiluweit, M., Kleber, M., Wang, Z., Pan, Z., Xing, B., 2012b. Polar and marine sediments. Environ. Toxicol. Chem. 24, 1594e1601.
aliphatic domains regulate sorption of phthalic acid esters (PAEs) to biochars.
Bioresour. Technol. 118, 120e127.