Accepted Manuscript Interactions of triazine herbicides with biochar: Steric and electronic effects Feng Xiao, Joseph J. Pignatello PII:
S0043-1354(15)00277-8
DOI:
10.1016/j.watres.2015.04.040
Reference:
WR 11269
To appear in:
Water Research
Received Date: 9 January 2015 Revised Date:
1 April 2015
Accepted Date: 13 April 2015
Please cite this article as: Xiao, F., Pignatello, J.J., Interactions of triazine herbicides with biochar: Steric and electronic effects, Water Research (2015), doi: 10.1016/j.watres.2015.04.040. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
Interactions of triazine herbicides with biochar: Steric and electronic
2
effects
4
Feng Xiao and Joseph J. Pignatello*
5
SC
6
RI PT
3
Department of Environmental Sciences, The Connecticut Agricultural Experimental Station, 123
8
Huntington St., P.O. Box 1106, New Haven, Connecticut 06504-1106, United States
M AN U
7
9
11 12
15 16
corresponding author contact information: Phone: +1-203-974-8518; fax: +1-203-974-8502; e-mail
EP
14
*
address: [email protected]
AC C
13
TE D
10
1
ACCEPTED MANUSCRIPT
ABSTRACT: We studied the adsorption of triazine herbicides and several reference
18
heteroaromatic amines from water onto a temperature series of hardwood biochars (300 – 700 oC,
19
labelled B300 – B700). Adsorption on biochars correlated poorly with pyrolysis temperature,
20
H/C, O/C, mean minimum fused ring size, surface area (N2 or CO2), microporosity, and
21
mesoporosity, but correlated well with a weighted sum of microporosity and mesoporosity.
22
Steric effects were evident by the negative influence of solute molecular volume on adsorption
23
rate. Adsorption rate maximized for the biochar with the greatest mesoporosity-to-total-porosity
24
ratio, suggesting that mesopores are important for facilitating diffusion into pore networks. The
25
cationic forms of amines adsorb more slowly than the neutral forms. To further probe steric and
26
electronic effects, adsorption on a biochar (B400) was compared to adsorption on graphite—a
27
nonporous reference material with an unhindered, unfunctionalized graphene surface—and in
28
comparison with reference compounds (benzene, naphthalene, pyridine, quinoline and 1,3-
29
triazine). Relative to benzene, the surface area-normalized adsorption of the triazine herbicides
30
was disfavored on B400 (favored on graphite) by 11−19 kJ/mole, depending on concentration. It
31
is estimated that steric suppression of B400 adsorption comprises 6.2 kJ/mol of this difference,
32
the remainder being the difference in polar electronic effects. Based on the behavior of the
33
reference amines, the difference in polar effects is dominated by π−π electron donor–acceptor
34
(EDA) interactions with sites on polyaromatic surfaces, which are more electropositive and/or
35
more abundant on graphite. Overall, our results show that mesoporosity is critical, that
36
adsorption rate is a function of solute molecular size and charge, that steric effects in the solute
37
largely suppress equilibrium adsorption, and that π−π EDA forces play a role in triazine polar
38
interactions with the biochar.
AC C
EP
TE D
M AN U
SC
RI PT
17
39
2
ACCEPTED MANUSCRIPT
40
Keywords: Herbicides; Soil contamination; Leaching; Groundwater; Solid waste conversion;
41
Biochar
42
RI PT
43 44 45
SC
46 47
M AN U
48 49 50 51 52
TE D
53 Abstract word count: 279
55
Text word count: 5104
56
Figures (6): 5 × 300 (Figs. 1−4 and 6) + 1 × 600 (Fig. 5) = 2100
57
Table (1): 1 × 300 = 300
58
Total word count: 7783 = Abstract (279) + Text (5104) + Figures (2100) + Table (300)
AC C
59 60
EP
54
3
ACCEPTED MANUSCRIPT
Graphical Abstract
SC
RI PT
61 62 63
M AN U
64 65 66
AC C
EP
TE D
67 68
4
ACCEPTED MANUSCRIPT
1. Introduction
70
Triazine herbicides are used world-wide in agriculture to control broadleaf and grassy weeds.
71
The triazines have low to moderate soil sorption coefficient, moderate solubility in water, and
72
low volatility, making them vulnerable to leaching. In the U.S., triazine herbicides, especially
73
atrazine, simazine and prometon, are frequently detected in ground and surface waters at
74
concentrations up to several hundred micrograms per liter (Gilliom, 2007, Hoffman et al., 2000,
75
Kolpin et al., 1998). Exposure to some triazine herbicides can lead to adverse human health and
76
ecosystem effects, including hormone disruption (atrazine) (Chapin et al., 1996) and damage to
77
testes, kidneys, and thyroid (simazine) (USEPA, 2002; 2005). Effective approaches for
78
mitigating the leaching of triazine herbicides from agricultural fields are clearly needed. Biochar,
79
which is made by anaerobic pyrolysis of biomass waste materials, has attracted attention as soil
80
amendments for reducing the physical and biological availabilities of soil-borne chemicals,
81
among other functions (Lehmann and Joseph, 2015 (in press)). Biochar addition to soils
82
increases adsorption of certain herbicide compounds (Cabrera et al., 2014, Graber and Kookana,
83
2015, Loganathan et al., 2009, Spokas et al., 2009), which can reduce leaching (Delwiche et al.,
84
2014, Graber and Kookana, 2015, Jones et al., 2011, Lu et al., 2012) but at the same time
85
increases the application rate needed to achieve the same level of weed control (Graber and
86
Kookana, 2015). Chars also present intrinsically in soil as the charred products of biomass
87
burning.
SC
M AN U
TE D
EP
AC C
88
RI PT
69
The objective of the present study was to better understand the interaction of triazines
89
with biochar at a fundamental level. Published studies with this objective in mind are few and
90
were seldom carried out by varying solutes and sorbents in a systematic manner. Hao et al. found
91
that the adsorption of atrazine on corncob biochars increases with decreasing H/C and (O + N)/C
5
ACCEPTED MANUSCRIPT
92
ratios, stating that “aromatic C and hydrophobic structures” are the major adsorption sites (Hao
93
et al., 2013). Cao et al. proposed that the adsorption of atrazine by low-temperature (200 and 350
94
◦
95
2009). Zhang et al. found that decreasing the solution pH from ~9.7 to ~3.7 led to a 1.5-fold
96
increase in the adsorption of atrazine and simazine on biochar made from a mixture of hardwood
97
chips and barks, and believed that the increase was due to the coulombic attraction between the
98
triazine cations and the negatively charged surfaces (Zheng et al., 2010). In spite of the insights
99
gained from these studies, our mechanistic understanding of the adsorption of triazines and other
RI PT
SC
agrochemicals to biochar is still rudimentary.
M AN U
100
C) biochars is due to partitioning into an uncarbonized biochar organic matter phase (Cao et al.,
First, the effect of biochar pore geometry on the adsorption is not clear. Biochar and
102
many pyrogenic carbonaceous materials (PCMs) are generally nanoporous, with pores having
103
apertures within the range of solute molecular diameters. Several rate models based on pore-
104
diffusion and pore-filling mechanisms have been developed for the adsorption on porous PCMs
105
(activated carbon) (Ahn et al., 2005, Crittenden et al., 1991, Kleineidam et al., 2002). However,
106
little attention has been paid so far to steric effects that could result in molecular exclusion from
107
pores or hindered diffusion within pore networks. Some studies of simple neutral aromatic
108
compounds have shown that adsorption on biochar becomes hindered with increasing degree of
109
substitution on the ring and increasing fused ring size (Lattao et al., 2014, Pignatello, 2013, Zhu
110
et al., 2005). However, the role of biochar geometry, especially the relative importance of
111
microporosity and mesoporosity of biochar, in relation to the size of solute molecules has not
112
been well characterized.
AC C
EP
TE D
101
113
Second, the influence of the electronic properties of heteroaromatic amines on adsorption
114
is also incompletely understood. Previous studies have shown that π-electron rich polyaromatic
6
ACCEPTED MANUSCRIPT
surfaces of biochar and graphite can interact with certain cyano- or nitroaromatic compounds
116
that are π-electron poor to form π–π electron donor-acceptor (EDA) complexes (Lattao et al.,
117
2014, Zhu and Pignatello, 2005). The π–π EDA force results primarily from attraction between
118
opposing quadrupoles. We recently showed that cationic triazine herbicides (ametryn and
119
prometon) and polyaromatic surfaces undergo π+–π EDA interactions, which are a combination a
120
π−π EDA force and a cation−π force (Xiao and Pignatello, 2015). Zhang et al. suggested that
121
uncharged atrazine undergoes π−π EDA interactions on manure-derived biochars (Zhang et al.,
122
2013). Since the neutral form of the triazines predominates in the soil under most conditions, it is
123
of interest to establish beyond speculation whether neutral triazine herbicides can associate
124
through π–π EDA interactions with the polyaromatic surfaces of PCMs including biochar.
M AN U
SC
RI PT
115
The effect of biochar amendment on the leaching of herbicides also strongly depends on
126
the adsorption rate; however, reports of adsorption rates of organic compounds including
127
herbicides by biochar are rare (Chen et al., 2012). Chen et al. state that the diffusion of organic
128
compounds in an uncarbonized ‘partition’ phase is slower than diffusion in the carbonized
129
phases of biochar; hence, naphthalene adsorption was slower in low-temperature biochars (200
130
and 300 oC) than in high-temperature biochars (> 400 oC), which had a higher degree of
131
carbonization (Chen et al., 2012).
EP
The objective of this study was to systematically probe key physical-chemical properties
AC C
132
TE D
125
133
of the biochar and key electronic and steric properties of the triazine herbicides that govern the
134
overall adsorption rate and extent. The test set included five triazine herbicides―ametryn,
135
atrazine, prometon, simazine, and terbutryn. The study was aided by the use of reference
136
compounds and sorbents. The reference compounds included three heteroaromatic amines
137
(pyridine, quinoline and 1,3-diazine). Quinoline is a steric homolog of pyridine; 1,3-diazine is
7
ACCEPTED MANUSCRIPT
isosteric with pyridine and an electronic homolog of pyridine. Pyridine and quinoline are used in
139
a variety of industrial, commercial, and agronomic applications, and have been found in the
140
natural environment (Brumley et al., 1991, Pereira et al., 1983, Stuermer et al., 1982). We also
141
included two nonpolar aromatic compounds (benzene and naphthalene) for comparison; their
142
adsorption data were taken from a previous study (Lattao et al., 2014). The reference adsorbent
143
was graphite, which has an unhindered, unfunctionalized polyaromatic (graphene) surface.
144
Graphite has been used previously as a reference for the polyaromatic surfaces of PCMs (Lattao
145
et al., 2014, Pignatello, 2013, Xiao and Pignatello, 2015, Zhu et al., 2005).
146
M AN U
SC
RI PT
138
2. Methods and materials
148
Biochars were made from maple wood shavings heated for 2 h under a flow of N2 at six final
149
heat treatment temperatures (HTTs) from 300 to 700 ºC as previously described (Cao et al.,
150
2012). The chars are hereafter referred to as B300 through B700. Graphite (Sigma-Aldrich;
151
99.99% C; particle size < 20 µm) had 100.1% C, 99.9%; pKa = 5.23), 1,3-
159
diazine (≥ 98.0%; pKa = 1.23), quinoline (98%; pKa = 4.95)]. ‘Nanopure water’ is tap water
160
passed through a commercial purifier (Milli-Q) to achieve a resistivity of > 18.2 MΩ cm. 8
ACCEPTED MANUSCRIPT
Adsorption was conducted in glass vials at 20 ºC by a method described previously (Xiao
162
and Pignatello, 2014). The liquid phase was synthetic agricultural runoff prepared with nanopure
163
water containing both monovalent and divalent ions (1.0 × 10−3 mol/L NaHCO3, 0.5 × 10−3 mol/L
164
CaCl2, 0.2 × 10−3 mol/L MgSO4, 0.1 × 10−3 mol/L KCl, 1.0 × 10−3 mol/L NaCl) and also
165
containing 3.1 × 10−3 mol/L NaN3 to inhibit microbial growth. The calculated ionic strength is
166
7.5 × 10−3 mol/L, not including salts from the biochar that may have dissolved. The adsorbent
167
(biochar or graphite) was pre-wetted in the liquid phase for 48 h. The vials were then spiked with
168
adsorbates from stock solutions in methanol (atrazine, simazine and terbutryn; final methanol
169
concentration in liquid phase, < 0.2%) or water (ametryn, prometon, pyridine, 1,3-diazine and
170
quinoline), and allowed to mix end-over-end at 40 revolutions per minute (rpm) at 20 ºC in the
171
dark. The solid-to-liquid ratio was adjusted to achieve 25−75% adsorption. The solid-to-liquid
172
ratio did not influence adsorption on graphite (Fig. S1 in the Supplementary data). The
173
adsorption equilibrium time was chosen on the basis of separate kinetic tests to be sufficient to
174
reach apparent equilibrium. Equilibration times of 50 d (for pyridine and quinoline) and 60 d (for
175
triazine herbicides) were chosen for biochar, and 48 h was chosen for graphite. An adsorption
176
time of 7 d for 1,3-diazine on biochar was selected on the basis of the uptake curve for pyridine.
177
Controls without adsorbents showed no significant abiotic loss during the uptake period. After
178
adsorption, the aqueous phase was sampled and microfiltered (0.45 µm, nylon). Adsorption to
179
filters was negligible. Aqueous concentrations were determined within two days by high-
180
performance liquid chromatography (HPLC) with UV detection at 228 nm for herbicides and
181
quinoline and 256 nm for 1,3-diazine and pyridine, using external standards. The analytical
182
HPLC methods can be found in the Supplementary data.
183
AC C
EP
TE D
M AN U
SC
RI PT
161
Adsorption isotherms were fit to the Freundlich model:
9
ACCEPTED MANUSCRIPT
184
Cs = K F Cwn
(1)
where Cs (mg/kg) and Cw (mg/L) are the adsorbed and aqueous-phase concentrations,
186
respectively; n is the Freundlich exponent providing an indication of isotherm nonlinearity; and
187
KF (mg1–n Ln kg–1) is the Freundlich adsorption coefficient. They were obtained by nonlinear
188
least-squares regression of experimental data weighted by the dependent variable. The observed
189
(concentration-dependent) distribution ratio K (L/kg) (Kb for biochar and Kg for graphite) is
190
defined as the adsorbed-to-solution concentration ratio,
193 194 195
The K is related to KF by K = K FCwn−1 .
SC
Cs Cw
(2)
M AN U
192
K=
The uptake data were fitted to an analytical solution of the Fickian diffusion equation for a sphere in a well-mixed batch system (Crank, 1975):
Cs,t Cs,e
6α (α + 1) exp(− Dbapp qn2t / a 2 ) 9 + 9α + qn2α 2 n =1 ∞
= 1− ∑
TE D
191
RI PT
185
(3)
where Cs,t and Cs,e are the concentrations of solute adsorbed after time t and equilibrium,
197
respectively; a (cm) is the characteristic diffusion length scale within the solid; Dbapp (cm2/sec) is
198
the apparent diffusion coefficient over a; α (unitless) is a parameter incorporating the final
199
fractional uptake of solute; and qn is the non-zero root of
AC C
200
EP
196
tan q n =
3q n 3 + α q n2
(4)
201
The value of Dbapp a 2 was obtained by minimizing the sum of the square of the differences
202
between measured and fitted values, with Dbapp a 2 as the only adjustable parameter. The
203
regression was performed for values of n up to 4 because
app 6α (α + 1) exp(− Dbiochar qn2t / a 2 ) 9 + 9α + qn2α 2
10
ACCEPTED MANUSCRIPT
204
approached zero for n > 4. The adjusted (for degrees of freedom) coefficient of determination
205
Radj2 was used to evaluate fitting performance (Xiao et al., 2013).
206
3. Results and discussion
208
3.1. Microporosity and mesoporosity and other properties of biochars as a function of HTT
209
Increasing HTT caused significant and nonlinear increases in C and decreases in O and H
210
contents of biochar (Fig. S2, Supplementary data). The loss of oxygen-containing substance and
211
volatile organic compounds during pyrolysis results in the formation of pores (Lehmann and
212
Joseph, 2015 (in press)). The total porosity of biochar tripled from HTT of 300 ºC to 500−700 ºC.
213
The microporosity of biochar (d < 2 nm) increased by 2.5 times as the HTT increased from 300
214
ºC to 700 ºC (Fig. 1). Considerable mesoporosity (2 nm < d < 50 nm) had developed by the HTT
215
of 500 ºC. The mesoporosity decreased two-fold when the HTT further increased from 500 oC to
216
600 ºC and was unchanged from 600 ºC to 700 ºC. B500 had the highest mesoporosity, whereas
217
B700 had the highest microporosity (Fig. 1). Lua et al. also found that the N2 BET surface area
218
of pistachio-nut shell-derived carbon increases with the pyrolysis temperature from 250 to 500
219
o
220
Keiluweit et al. found that a general increase in N2 BET surface area of wood- and grass-derived
221
chars as the pyrolysis temperature increasing from 100 to 700 oC (Keiluweit et al., 2010). The
222
effects of pyrolysis temperature on the evolution of biochar pore geometry are not well
223
understood, and they are potentially intertwined with other pyrolysis conditions and feedstock
224
type. In addition, the isoelectric point determined by zeta potential measurement is pHiep < ~2 for
225
all the biochars (Fig. S3, Supplementary data). As shown in Fig. S3, at neutral pH, the low-
226
temperature biochars (e.g,. B300) tends to be more negatively charged than high-temperature
TE D
M AN U
SC
RI PT
207
AC C
EP
C but then decreases (moderately) from 500 to 800 oC (Lua et al., 2004). On the other hand,
11
ACCEPTED MANUSCRIPT
biochars (e.g., B700), which is likely due to the loss of surface oxygenic functional groups with
228
temperature (see Fig. S2 of Supplementary data). The mean minimum aromatic fused ring cluster
229
size estimated by nuclear magnetic resonance spectroscopy increased approximately linearly
230
from 20 to 76 carbons with HTT from 300 to 700 oC (Cao et al., 2012).
RI PT
227
231 232
Fig. 1 is near here.
SC
233
3.2. Adsorption as a function of biochar physical-chemical properties
235
We initially examined the relationships between adsorption and biochar physical-chemical
236
properties using atrazine and the reference, planar compound quinoline as probes. Adsorption
237
isotherms of atrazine on B300, B400, B500, B600 and B700 are shown in Fig. S4
238
(Supplementary data), and those of quinoline in Fig. S5. Many studies of organic compound
239
adsorption by PCMs (biochar and activated carbon) (Ahmaruzzaman, 2008, Graber et al., 2012,
240
Hao et al., 2013, Kearns et al., 2014, Wang et al., 2010, Yang et al., 2010, Zhang et al., 2011)
241
report a trend of increasing adsorption with increasing SA or microporosity. In addition, atrazine
242
adsorption was found to decrease with increasing carbonization (as H/C ratio) or increasing
243
polarity [as (O + H)/C ratio] (Hao et al., 2013). However, our results are not in agreement with
244
these reported trends. Biochar B700 has the highest SA, largest microporosity, lowest H/C,
245
lowest O/C, and highest mean minimum fused ring cluster size, indicating it has the largest and
246
most hydrophobic surface of all the chars; thus, B700 was expected to have the greatest
247
adsorption affinity for the test compounds according to the previously-reported trends. However,
248
no straightforward relationship was found between atrazine or quinoline adsorption and each of
249
the following biochar properties: micropore SA (CO2 accessible), N2 BET SA, CO2-accessible
AC C
EP
TE D
M AN U
234
12
ACCEPTED MANUSCRIPT
microporosity, mesoporosity, H/C, O/C, and fused ring cluster size (Figs. S6 and S7,
251
Supplementary data). As shown in Figs. S4 and S6 in Supplementary data, adsorption of atrazine
252
peaked at B500, which has the highest mesoporosity. The same is true for quinoline. The results
253
are consistent with a recent study (Lattao et al., 2014) that found no straightforward relationship
254
between these same biochar properties and adsorption affinity for benzene, naphthalene, or 1,4-
255
dinitrobenzene.
RI PT
250
Lattao et al. (Lattao et al., 2014) found that the “reduced” carbon-normalized distribution
257
coefficient (Kb-C,r) for benzene, naphthalene, or 1,4-dinitrobenzene could be predicted by the
258
weighted sum of microporosity and mesoporosity:
M AN U
SC
256
K b-C,r = a ⋅ microporosity + b ⋅ mesoporosity
259
(5)
where a and b are the coefficients (g/cm3) corresponding to micropore adsorption and mesopore
261
adsorption, respectively. Kb-C,r is calculated from Cs-C (carbon-normalized Cs) using the
262
Freundlich equation and the reduced aqueous concentration. The reduced concentration is given
263
by
264
Cr = Cw Cwsat ( L)
TE D
260
(6)
where Cwsat ( L) is saturated water solubility of the liquid or subcooled liquid (atrazine, 1068 mg/L
266
(Capel et al., 2008); quinoline, 6110 mg/L (SRC, 2014)). In this study, Eq. 5 was extended to
267
include two heteroaromatic amines (atrazine and quinoline), and the values of Kb-C,r were
268
calculated at Cr of 0.001 and 0.01.
AC C
269
EP
265
A visual presentation of model performance of Eq. 5 is given in Fig. 2, and the regression
270
results appear in Table S1 (Supplementary data). The coefficient of determination (R2) falls in
271
the range 0.88−0.99 (Table S1), suggesting a strong correlation. However, in each case there are
272
only five data points (five biochars) against two independent variables, which may lead to over-
13
ACCEPTED MANUSCRIPT
fitting (Xiao et al., 2013). The adjusted coefficient of determination (Radj2), which takes into
274
account the degree of freedom, falls in the range 0.51−0.66, indicating a moderately strong
275
correlation. The prediction of Kb-C,r is good to within a factor of 3 (0.48 log units) for all but
276
three data points (Fig. 2).
Fig. 2 is near here.
3.3. Adsorption rates
SC
277 278 279 280 281
RI PT
273
Uptake rates of prometon, pyridine, and quinoline by biochars were determined at pH ~7, where
283
they are predominantly neutral in solution, and at pH ~2 where they are predominantly cationic.
284
The data were fitted to the diffusion model in Eq. 3. The uptake curves along with model fits are
285
shown in Fig. S8 (Supplementary data). The fits by Eq. 3 are generally better for the neutral than
286
the cationic species. The fitted values of Dbapp a 2 are given in Table S2 (Supplementary data) and
287
displayed in the bar graphs of Fig. 3.
288 Fig. 3 is near here.
EP
289 290
TE D
M AN U
282
Steric hindrance to diffusion becomes important as the ratio of pore aperture to the
292
minimum critical molecular diameter drops below ~10, and becomes severe as that ratio
293
approaches 1 (Kärger and Ruthven, 1992). The computed molecular volumes and shortest (width)
294
and longest (length) molecular dimensions are given in Table 1. The reference compounds,
295
benzene, naphthalene, pyridine, 1,3-diazine and quinoline are characterized by co-planarity of all
296
atoms in the molecule. The single-ring compounds vary 2.74−2.81 Å in length and the double-
297
ring compounds vary 5.02−5.07 Å in length (Table 1). The triazine herbicides are larger in both
298
dimensions than the reference compounds due to ring substituents. In addition, the triazines have
AC C
291
14
ACCEPTED MANUSCRIPT
greater bulk in the direction (z) perpendicular to the plane of the ring due to the inability of all
300
substituent atoms to simultaneously assume co-planarity with the ring atoms. This can further
301
restrict diffusion within slit-like pores, which become more prevalent as the biochar evolves to a
302
more polyaromatic structure with increasing pyrolysis temperature.
303 304 305
Table 1 is near here.
RI PT
299
For a given biochar, Dbapp a 2 follows the order, pyridine > quinoline > prometon, which
307
is the reverse order in their molecular volume. The value of Dbapp a 2 follows the same order for
308
the corresponding cationic form. An inverse correlation between Dbapp a 2 and molecular volume
309
is expected if steric effects are important. Quinoline is 79% longer and 56% more voluminous
310
than pyridine, and prometon is the largest among the three compounds in all dimensions (Table
311
1). The value of Dbapp a 2 reaches a maximum at B500 with all compounds, both in neutral and
312
cationic form. B500 has the largest mesoporosity and the highest ratio of mesoporosity to total
313
porosity (Fig. 1). This suggests that mesopores provide pathways for molecules to access deeper
314
and smaller pores. The slowest uptake for all three compounds occurred for B700, which has the
315
highest microporosity. Restricted diffusion will be manifested more in micropores than
316
mesopores, and the steric hindrance in B700 micropores may lead to its low adsorption affinity
317
towards solutes in comparison with B500 that has lower microporosity but greater mesoporosity.
318
It can also be seen in Fig. 3 that Dbapp a 2 is several times smaller for the cation than for
AC C
EP
TE D
M AN U
SC
306
319
the corresponding neutral molecule of pyridine, quinoline and prometon. This observation is
320
novel. The slower uptake of the cation is likely due to one or more of the following effects that
321
would retard its diffusion in pores compared to the neutral molecule: i) back diffusion of native
15
ACCEPTED MANUSCRIPT
cations (e.g., Na+) to maintain electroneutrality; ii) co-diffusion of a counterion (e.g., Cl−) to
323
maintain electroneutrality; or iii) a larger hydration sphere of strongly-bound water molecules
324
surrounding the cation and counterion. The relative contribution of these effects requires further
325
study.
RI PT
322
326
3.4. Probing electronic and steric effects using graphite as a reference adsorbent
328
The adsorption isotherms of the triazines and the reference compounds, pyridine and quinoline,
329
were determined on B400 and graphite for comparison. Graphite is nonporous and sparsely-
330
functionalized and should therefore interact primarily on the infinite polyaromatic sheet surface,
331
with some adsorption along edges and defects. The weight-based adsorption of the (nonplanar)
332
triazine herbicides was generally lower on B400 than on graphite (Fig. S9a), despite the higher
333
SA of B400 than graphite (N2-accessible SA: 9.3 m2/g vs 4.5 m2/g; CO2-accessible SA: 388 m2/g
334
vs 15.8 m2/g (Lattao et al., 2014)). On the other hand, on a weight basis the planar heterocyclic
335
amines (pyridine, quinoline, and 1,3-diazine) prefer B400 over graphite.
TE D
M AN U
SC
327
The N2 SA-normalized adsorption isotherms on B400 and graphite are shown in Fig. 4.
337
These isotherms are mirrored by the corresponding CO2 SA-normalized isotherms shown in Fig.
338
S9b (Supplementary data). Based on N2 SA, adsorption of the triazine herbicides is 101- to 102-
339
times greater on graphite than on B400. The origin of this preference for graphite may be steric
340
or electronic, or both. The steric bulk of the triazines may limit their access to interior pore SA of
341
B400 that is otherwise available to N2 at 77 K. In this regard, the preference for graphite is not
342
exhibited for the smaller, flatter molecules pyridine, quinoline and 1,3-diazine. An electronic
343
origin of the preference would be remarkable because graphite lacks polar (at least, dipolar)
344
functionality.
AC C
EP
336
16
ACCEPTED MANUSCRIPT
345 346 347
Fig. 4 is near here.
To better understand the relative importance of steric and electronic effects on adsorption,
349
the B400−water and graphite−water isotherms were used to construct the SA normalized
350
B400−graphite distribution isotherms for eight heteroaromatic amines (pyridine, 1,3-diazine,
351
quinoline, and five triazines) studied in the present work, in comparison with those for two
352
aromatic compounds (benzene and naphthalene) obtained from a previous study (Lattao et al.,
353
2014). The isotherms of a reference compound, 1,3,5-triazine could not be obtained because the
354
compound is unstable in water. The B400−graphite isotherms were constructed using the
355
experimental adsorption data points for B400 (Cw, Cs,b) and calculated values of Cs,g for graphite
356
at the same Cw using the Freundlich parameters for graphite converted to unit SA basis (Lattao et
357
al., 2014). The biochar−graphite distribution ratio is given by: K b -g =
358
TE D
M AN U
SC
RI PT
348
Cs,b C w Cs,b = Cs,g C w Cs,g
(7)
Figure S10a shows B400−graphite adsorption isotherms normalized by the N2 BET SA,
360
( SA bN and SA gN 2 , respectively). The same isotherms normalized by CO2 SA in Fig. S10b have
361
identical
362
2 2 SA bN 2 SA CO ⋅ ( SA gN 2 SA CO ) . The isotherms indicate increasing preference for graphite over b g
363
B400 in the order, benzene < pyridine ~ naphthalene < 1,3-diazine < quinoline < triazine
364
herbicides.
365 366
2
relationships,
AC C
spatial
EP
359
but
are
just
shifted
by
a
constant
equal
to
−1
The free energy of B400−graphite partitioning of a compound i, relative to i = benzene, can be expressed by (Lattao et al., 2014),
17
ACCEPTED MANUSCRIPT
367
∆i −BEN ∆ b-gG = ∆i − BEN ∆ b-g G ster + ∆i −BEN ∆ b-g G polar + ∆i −BEN ∆ b-gG hydr
(8)
where the terms on the right correspond to the contributions to overall free energy by steric,
369
polar, and hydrophobic effects. The polar term lumps dipole−dipole, dipole-induced dipole,
370
hydrogen bonding, and quadrupole−quadrupole forces. The hydrophobic effect term lumps
371
dispersion forces at the surface and interactions in the liquid phase that result in disruption of the
372
cohesive forces of bulk water. The hydrophobic free energy may be taken as proportional to the
373
free energy of solvent−water partitioning, such that
SC
∆i−BEN ∆b-gGhydr = ( ∆i −BEN ∆b-g a ) ∆i -BEN ∆S-wG + ( ∆i −BEN ∆b-gb )
(9)
M AN U
374
RI PT
368
where S is a suitable reference solvent and a and b are proportionality constants. For a closely-
376
related series of compounds, the hydrophobic effects driving adsorption to B400 and graphite are
377
expected to be roughly equal, which is equivalent to saying that the values of ∆i−BEN∆b-g a and
378
∆i−BEN∆b-gb are close to zero. Therefore, the hydrophobic term in Eq. 8 can be ignored.
379
Substituting the appropriate B400−graphite adsorption equilibrium constants for ∆BEN-i ∆b-gG , Eq.
380
8 becomes,
383
EP
382
∆ i −BEN ∆ b-g G ster + ∆ i −BEN ∆ b-g G polar = − RT ln
K b-g,i
K b-g,BEN
+ const
(10)
where const is a constant that depends on the reference states chosen for adsorption.
AC C
381
TE D
375
Figure 5 plots ∆ i − BEN ∆ b-g G ster + ∆ i − BEN ∆ b-g G polar versus the adsorbed concentration on
384
graphite for all the eight heteroaromatic amines studied in the present work and two aromatic
385
compounds (benzene and naphthalene) from a previous study (Lattao et al., 2014). One can see
386
that the curve for quinoline is positively displaced relative to the curve for pyridine. Since
387
quinoline and pyridine are expected to undergo similar polar interactions with B400, this
388
suggests that steric effects suppress quinoline adsorption relative to pyridine adsorption on B400. 18
ACCEPTED MANUSCRIPT
As such, the curve for naphthalene is positively displaced (less favorable adsorption by B400)
390
relative to the curve for benzene. Since naphthalene does not undergo significant polar
391
interactions with either biochar or graphite, the positive displacement of the curve for
392
naphthalene has been attributed to steric effects (Lattao et al., 2014).
RI PT
389
393 Fig. 5 is near here.
SC
394 395
The free energy curves for pyridine and 1,3-diazine are also positively displaced relative
397
to the curve for benzene. Since pyridine and 1,3-diazine are isosteric with benzene, this indicates
398
that pyridine and 1,3-diazine have greater affinity for graphite than for B400 for electronic
399
reasons. Pyridine and 1,3-diazine can undergo dipole−dipole, H bonding and π−π EDA
400
interactions with B400. However, since graphite has no significant functionality, of these forces,
401
π−π EDA interactions with graphite appear to dominate the ∆ i − BEN ∆ b-g G polar term for pyridine and
402
1,3-diazine.
TE D
M AN U
396
The π−π EDA force may occur when electron-poor and electron-rich π systems interact
404
through their quadrupoles. The polyaromatic surfaces of PCMs are π-electron rich, while the
405
solute molecule is π-electron poor by virtue of the electronegative N atoms incorporated into the
406
ring. The polyaromatic sheet of graphite is infinite, whereas the mean minimum fused ring size
407
of B400 is about 18 C atoms (Cao et al., 2012). The polyaromatic component of the total SA of
408
B400 is expected to be less than that of graphite due to the presence of incompletely carbonized
409
structure (Keiluweit et al., 2010) in biochar. In addition, B400 contains almost 20% O—some of
410
it in functional groups like hydroxyl, carboxyl, ether and carbonyl attached to rings—that likely
411
reduce the π-donor character of the ring systems.
AC C
EP
403
19
ACCEPTED MANUSCRIPT
Electrostatic potential (Ep, kJ/mol) has been used as an indicator of a compound’s π-
413
donor or π-acceptor character, and thus its potential to engage in π−π EDA interactions (Cockroft
414
et al., 2007, Mecozzi et al., 1996, Xiao and Pignatello, 2015). The molecular Ep surface of
415
solutes was determined by density functional theory (B3LYP, 6-31G* basis set; Spartan 10,
416
Wavefunction Inc., Irvine, CA) (Xiao and Pignatello, 2015). The value of Ep at the midpoint of
417
the ring (Ep,mid) was taken as a measure of the π-donor or π-acceptor character of the ring as a
418
whole, according to past practice (Xiao and Pignatello, 2015). The value of Ep,mid (kJ/mol)
419
increases with increasing N incorporation in the order, benzene (−96.0) < pyridine (−44.2)
S0043-1354(15)00277-8
DOI:
10.1016/j.watres.2015.04.040
Reference:
WR 11269
To appear in:
Water Research
Received Date: 9 January 2015 Revised Date:
1 April 2015
Accepted Date: 13 April 2015
Please cite this article as: Xiao, F., Pignatello, J.J., Interactions of triazine herbicides with biochar: Steric and electronic effects, Water Research (2015), doi: 10.1016/j.watres.2015.04.040. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
Interactions of triazine herbicides with biochar: Steric and electronic
2
effects
4
Feng Xiao and Joseph J. Pignatello*
5
SC
6
RI PT
3
Department of Environmental Sciences, The Connecticut Agricultural Experimental Station, 123
8
Huntington St., P.O. Box 1106, New Haven, Connecticut 06504-1106, United States
M AN U
7
9
11 12
15 16
corresponding author contact information: Phone: +1-203-974-8518; fax: +1-203-974-8502; e-mail
EP
14
*
address: [email protected]
AC C
13
TE D
10
1
ACCEPTED MANUSCRIPT
ABSTRACT: We studied the adsorption of triazine herbicides and several reference
18
heteroaromatic amines from water onto a temperature series of hardwood biochars (300 – 700 oC,
19
labelled B300 – B700). Adsorption on biochars correlated poorly with pyrolysis temperature,
20
H/C, O/C, mean minimum fused ring size, surface area (N2 or CO2), microporosity, and
21
mesoporosity, but correlated well with a weighted sum of microporosity and mesoporosity.
22
Steric effects were evident by the negative influence of solute molecular volume on adsorption
23
rate. Adsorption rate maximized for the biochar with the greatest mesoporosity-to-total-porosity
24
ratio, suggesting that mesopores are important for facilitating diffusion into pore networks. The
25
cationic forms of amines adsorb more slowly than the neutral forms. To further probe steric and
26
electronic effects, adsorption on a biochar (B400) was compared to adsorption on graphite—a
27
nonporous reference material with an unhindered, unfunctionalized graphene surface—and in
28
comparison with reference compounds (benzene, naphthalene, pyridine, quinoline and 1,3-
29
triazine). Relative to benzene, the surface area-normalized adsorption of the triazine herbicides
30
was disfavored on B400 (favored on graphite) by 11−19 kJ/mole, depending on concentration. It
31
is estimated that steric suppression of B400 adsorption comprises 6.2 kJ/mol of this difference,
32
the remainder being the difference in polar electronic effects. Based on the behavior of the
33
reference amines, the difference in polar effects is dominated by π−π electron donor–acceptor
34
(EDA) interactions with sites on polyaromatic surfaces, which are more electropositive and/or
35
more abundant on graphite. Overall, our results show that mesoporosity is critical, that
36
adsorption rate is a function of solute molecular size and charge, that steric effects in the solute
37
largely suppress equilibrium adsorption, and that π−π EDA forces play a role in triazine polar
38
interactions with the biochar.
AC C
EP
TE D
M AN U
SC
RI PT
17
39
2
ACCEPTED MANUSCRIPT
40
Keywords: Herbicides; Soil contamination; Leaching; Groundwater; Solid waste conversion;
41
Biochar
42
RI PT
43 44 45
SC
46 47
M AN U
48 49 50 51 52
TE D
53 Abstract word count: 279
55
Text word count: 5104
56
Figures (6): 5 × 300 (Figs. 1−4 and 6) + 1 × 600 (Fig. 5) = 2100
57
Table (1): 1 × 300 = 300
58
Total word count: 7783 = Abstract (279) + Text (5104) + Figures (2100) + Table (300)
AC C
59 60
EP
54
3
ACCEPTED MANUSCRIPT
Graphical Abstract
SC
RI PT
61 62 63
M AN U
64 65 66
AC C
EP
TE D
67 68
4
ACCEPTED MANUSCRIPT
1. Introduction
70
Triazine herbicides are used world-wide in agriculture to control broadleaf and grassy weeds.
71
The triazines have low to moderate soil sorption coefficient, moderate solubility in water, and
72
low volatility, making them vulnerable to leaching. In the U.S., triazine herbicides, especially
73
atrazine, simazine and prometon, are frequently detected in ground and surface waters at
74
concentrations up to several hundred micrograms per liter (Gilliom, 2007, Hoffman et al., 2000,
75
Kolpin et al., 1998). Exposure to some triazine herbicides can lead to adverse human health and
76
ecosystem effects, including hormone disruption (atrazine) (Chapin et al., 1996) and damage to
77
testes, kidneys, and thyroid (simazine) (USEPA, 2002; 2005). Effective approaches for
78
mitigating the leaching of triazine herbicides from agricultural fields are clearly needed. Biochar,
79
which is made by anaerobic pyrolysis of biomass waste materials, has attracted attention as soil
80
amendments for reducing the physical and biological availabilities of soil-borne chemicals,
81
among other functions (Lehmann and Joseph, 2015 (in press)). Biochar addition to soils
82
increases adsorption of certain herbicide compounds (Cabrera et al., 2014, Graber and Kookana,
83
2015, Loganathan et al., 2009, Spokas et al., 2009), which can reduce leaching (Delwiche et al.,
84
2014, Graber and Kookana, 2015, Jones et al., 2011, Lu et al., 2012) but at the same time
85
increases the application rate needed to achieve the same level of weed control (Graber and
86
Kookana, 2015). Chars also present intrinsically in soil as the charred products of biomass
87
burning.
SC
M AN U
TE D
EP
AC C
88
RI PT
69
The objective of the present study was to better understand the interaction of triazines
89
with biochar at a fundamental level. Published studies with this objective in mind are few and
90
were seldom carried out by varying solutes and sorbents in a systematic manner. Hao et al. found
91
that the adsorption of atrazine on corncob biochars increases with decreasing H/C and (O + N)/C
5
ACCEPTED MANUSCRIPT
92
ratios, stating that “aromatic C and hydrophobic structures” are the major adsorption sites (Hao
93
et al., 2013). Cao et al. proposed that the adsorption of atrazine by low-temperature (200 and 350
94
◦
95
2009). Zhang et al. found that decreasing the solution pH from ~9.7 to ~3.7 led to a 1.5-fold
96
increase in the adsorption of atrazine and simazine on biochar made from a mixture of hardwood
97
chips and barks, and believed that the increase was due to the coulombic attraction between the
98
triazine cations and the negatively charged surfaces (Zheng et al., 2010). In spite of the insights
99
gained from these studies, our mechanistic understanding of the adsorption of triazines and other
RI PT
SC
agrochemicals to biochar is still rudimentary.
M AN U
100
C) biochars is due to partitioning into an uncarbonized biochar organic matter phase (Cao et al.,
First, the effect of biochar pore geometry on the adsorption is not clear. Biochar and
102
many pyrogenic carbonaceous materials (PCMs) are generally nanoporous, with pores having
103
apertures within the range of solute molecular diameters. Several rate models based on pore-
104
diffusion and pore-filling mechanisms have been developed for the adsorption on porous PCMs
105
(activated carbon) (Ahn et al., 2005, Crittenden et al., 1991, Kleineidam et al., 2002). However,
106
little attention has been paid so far to steric effects that could result in molecular exclusion from
107
pores or hindered diffusion within pore networks. Some studies of simple neutral aromatic
108
compounds have shown that adsorption on biochar becomes hindered with increasing degree of
109
substitution on the ring and increasing fused ring size (Lattao et al., 2014, Pignatello, 2013, Zhu
110
et al., 2005). However, the role of biochar geometry, especially the relative importance of
111
microporosity and mesoporosity of biochar, in relation to the size of solute molecules has not
112
been well characterized.
AC C
EP
TE D
101
113
Second, the influence of the electronic properties of heteroaromatic amines on adsorption
114
is also incompletely understood. Previous studies have shown that π-electron rich polyaromatic
6
ACCEPTED MANUSCRIPT
surfaces of biochar and graphite can interact with certain cyano- or nitroaromatic compounds
116
that are π-electron poor to form π–π electron donor-acceptor (EDA) complexes (Lattao et al.,
117
2014, Zhu and Pignatello, 2005). The π–π EDA force results primarily from attraction between
118
opposing quadrupoles. We recently showed that cationic triazine herbicides (ametryn and
119
prometon) and polyaromatic surfaces undergo π+–π EDA interactions, which are a combination a
120
π−π EDA force and a cation−π force (Xiao and Pignatello, 2015). Zhang et al. suggested that
121
uncharged atrazine undergoes π−π EDA interactions on manure-derived biochars (Zhang et al.,
122
2013). Since the neutral form of the triazines predominates in the soil under most conditions, it is
123
of interest to establish beyond speculation whether neutral triazine herbicides can associate
124
through π–π EDA interactions with the polyaromatic surfaces of PCMs including biochar.
M AN U
SC
RI PT
115
The effect of biochar amendment on the leaching of herbicides also strongly depends on
126
the adsorption rate; however, reports of adsorption rates of organic compounds including
127
herbicides by biochar are rare (Chen et al., 2012). Chen et al. state that the diffusion of organic
128
compounds in an uncarbonized ‘partition’ phase is slower than diffusion in the carbonized
129
phases of biochar; hence, naphthalene adsorption was slower in low-temperature biochars (200
130
and 300 oC) than in high-temperature biochars (> 400 oC), which had a higher degree of
131
carbonization (Chen et al., 2012).
EP
The objective of this study was to systematically probe key physical-chemical properties
AC C
132
TE D
125
133
of the biochar and key electronic and steric properties of the triazine herbicides that govern the
134
overall adsorption rate and extent. The test set included five triazine herbicides―ametryn,
135
atrazine, prometon, simazine, and terbutryn. The study was aided by the use of reference
136
compounds and sorbents. The reference compounds included three heteroaromatic amines
137
(pyridine, quinoline and 1,3-diazine). Quinoline is a steric homolog of pyridine; 1,3-diazine is
7
ACCEPTED MANUSCRIPT
isosteric with pyridine and an electronic homolog of pyridine. Pyridine and quinoline are used in
139
a variety of industrial, commercial, and agronomic applications, and have been found in the
140
natural environment (Brumley et al., 1991, Pereira et al., 1983, Stuermer et al., 1982). We also
141
included two nonpolar aromatic compounds (benzene and naphthalene) for comparison; their
142
adsorption data were taken from a previous study (Lattao et al., 2014). The reference adsorbent
143
was graphite, which has an unhindered, unfunctionalized polyaromatic (graphene) surface.
144
Graphite has been used previously as a reference for the polyaromatic surfaces of PCMs (Lattao
145
et al., 2014, Pignatello, 2013, Xiao and Pignatello, 2015, Zhu et al., 2005).
146
M AN U
SC
RI PT
138
2. Methods and materials
148
Biochars were made from maple wood shavings heated for 2 h under a flow of N2 at six final
149
heat treatment temperatures (HTTs) from 300 to 700 ºC as previously described (Cao et al.,
150
2012). The chars are hereafter referred to as B300 through B700. Graphite (Sigma-Aldrich;
151
99.99% C; particle size < 20 µm) had 100.1% C, 99.9%; pKa = 5.23), 1,3-
159
diazine (≥ 98.0%; pKa = 1.23), quinoline (98%; pKa = 4.95)]. ‘Nanopure water’ is tap water
160
passed through a commercial purifier (Milli-Q) to achieve a resistivity of > 18.2 MΩ cm. 8
ACCEPTED MANUSCRIPT
Adsorption was conducted in glass vials at 20 ºC by a method described previously (Xiao
162
and Pignatello, 2014). The liquid phase was synthetic agricultural runoff prepared with nanopure
163
water containing both monovalent and divalent ions (1.0 × 10−3 mol/L NaHCO3, 0.5 × 10−3 mol/L
164
CaCl2, 0.2 × 10−3 mol/L MgSO4, 0.1 × 10−3 mol/L KCl, 1.0 × 10−3 mol/L NaCl) and also
165
containing 3.1 × 10−3 mol/L NaN3 to inhibit microbial growth. The calculated ionic strength is
166
7.5 × 10−3 mol/L, not including salts from the biochar that may have dissolved. The adsorbent
167
(biochar or graphite) was pre-wetted in the liquid phase for 48 h. The vials were then spiked with
168
adsorbates from stock solutions in methanol (atrazine, simazine and terbutryn; final methanol
169
concentration in liquid phase, < 0.2%) or water (ametryn, prometon, pyridine, 1,3-diazine and
170
quinoline), and allowed to mix end-over-end at 40 revolutions per minute (rpm) at 20 ºC in the
171
dark. The solid-to-liquid ratio was adjusted to achieve 25−75% adsorption. The solid-to-liquid
172
ratio did not influence adsorption on graphite (Fig. S1 in the Supplementary data). The
173
adsorption equilibrium time was chosen on the basis of separate kinetic tests to be sufficient to
174
reach apparent equilibrium. Equilibration times of 50 d (for pyridine and quinoline) and 60 d (for
175
triazine herbicides) were chosen for biochar, and 48 h was chosen for graphite. An adsorption
176
time of 7 d for 1,3-diazine on biochar was selected on the basis of the uptake curve for pyridine.
177
Controls without adsorbents showed no significant abiotic loss during the uptake period. After
178
adsorption, the aqueous phase was sampled and microfiltered (0.45 µm, nylon). Adsorption to
179
filters was negligible. Aqueous concentrations were determined within two days by high-
180
performance liquid chromatography (HPLC) with UV detection at 228 nm for herbicides and
181
quinoline and 256 nm for 1,3-diazine and pyridine, using external standards. The analytical
182
HPLC methods can be found in the Supplementary data.
183
AC C
EP
TE D
M AN U
SC
RI PT
161
Adsorption isotherms were fit to the Freundlich model:
9
ACCEPTED MANUSCRIPT
184
Cs = K F Cwn
(1)
where Cs (mg/kg) and Cw (mg/L) are the adsorbed and aqueous-phase concentrations,
186
respectively; n is the Freundlich exponent providing an indication of isotherm nonlinearity; and
187
KF (mg1–n Ln kg–1) is the Freundlich adsorption coefficient. They were obtained by nonlinear
188
least-squares regression of experimental data weighted by the dependent variable. The observed
189
(concentration-dependent) distribution ratio K (L/kg) (Kb for biochar and Kg for graphite) is
190
defined as the adsorbed-to-solution concentration ratio,
193 194 195
The K is related to KF by K = K FCwn−1 .
SC
Cs Cw
(2)
M AN U
192
K=
The uptake data were fitted to an analytical solution of the Fickian diffusion equation for a sphere in a well-mixed batch system (Crank, 1975):
Cs,t Cs,e
6α (α + 1) exp(− Dbapp qn2t / a 2 ) 9 + 9α + qn2α 2 n =1 ∞
= 1− ∑
TE D
191
RI PT
185
(3)
where Cs,t and Cs,e are the concentrations of solute adsorbed after time t and equilibrium,
197
respectively; a (cm) is the characteristic diffusion length scale within the solid; Dbapp (cm2/sec) is
198
the apparent diffusion coefficient over a; α (unitless) is a parameter incorporating the final
199
fractional uptake of solute; and qn is the non-zero root of
AC C
200
EP
196
tan q n =
3q n 3 + α q n2
(4)
201
The value of Dbapp a 2 was obtained by minimizing the sum of the square of the differences
202
between measured and fitted values, with Dbapp a 2 as the only adjustable parameter. The
203
regression was performed for values of n up to 4 because
app 6α (α + 1) exp(− Dbiochar qn2t / a 2 ) 9 + 9α + qn2α 2
10
ACCEPTED MANUSCRIPT
204
approached zero for n > 4. The adjusted (for degrees of freedom) coefficient of determination
205
Radj2 was used to evaluate fitting performance (Xiao et al., 2013).
206
3. Results and discussion
208
3.1. Microporosity and mesoporosity and other properties of biochars as a function of HTT
209
Increasing HTT caused significant and nonlinear increases in C and decreases in O and H
210
contents of biochar (Fig. S2, Supplementary data). The loss of oxygen-containing substance and
211
volatile organic compounds during pyrolysis results in the formation of pores (Lehmann and
212
Joseph, 2015 (in press)). The total porosity of biochar tripled from HTT of 300 ºC to 500−700 ºC.
213
The microporosity of biochar (d < 2 nm) increased by 2.5 times as the HTT increased from 300
214
ºC to 700 ºC (Fig. 1). Considerable mesoporosity (2 nm < d < 50 nm) had developed by the HTT
215
of 500 ºC. The mesoporosity decreased two-fold when the HTT further increased from 500 oC to
216
600 ºC and was unchanged from 600 ºC to 700 ºC. B500 had the highest mesoporosity, whereas
217
B700 had the highest microporosity (Fig. 1). Lua et al. also found that the N2 BET surface area
218
of pistachio-nut shell-derived carbon increases with the pyrolysis temperature from 250 to 500
219
o
220
Keiluweit et al. found that a general increase in N2 BET surface area of wood- and grass-derived
221
chars as the pyrolysis temperature increasing from 100 to 700 oC (Keiluweit et al., 2010). The
222
effects of pyrolysis temperature on the evolution of biochar pore geometry are not well
223
understood, and they are potentially intertwined with other pyrolysis conditions and feedstock
224
type. In addition, the isoelectric point determined by zeta potential measurement is pHiep < ~2 for
225
all the biochars (Fig. S3, Supplementary data). As shown in Fig. S3, at neutral pH, the low-
226
temperature biochars (e.g,. B300) tends to be more negatively charged than high-temperature
TE D
M AN U
SC
RI PT
207
AC C
EP
C but then decreases (moderately) from 500 to 800 oC (Lua et al., 2004). On the other hand,
11
ACCEPTED MANUSCRIPT
biochars (e.g., B700), which is likely due to the loss of surface oxygenic functional groups with
228
temperature (see Fig. S2 of Supplementary data). The mean minimum aromatic fused ring cluster
229
size estimated by nuclear magnetic resonance spectroscopy increased approximately linearly
230
from 20 to 76 carbons with HTT from 300 to 700 oC (Cao et al., 2012).
RI PT
227
231 232
Fig. 1 is near here.
SC
233
3.2. Adsorption as a function of biochar physical-chemical properties
235
We initially examined the relationships between adsorption and biochar physical-chemical
236
properties using atrazine and the reference, planar compound quinoline as probes. Adsorption
237
isotherms of atrazine on B300, B400, B500, B600 and B700 are shown in Fig. S4
238
(Supplementary data), and those of quinoline in Fig. S5. Many studies of organic compound
239
adsorption by PCMs (biochar and activated carbon) (Ahmaruzzaman, 2008, Graber et al., 2012,
240
Hao et al., 2013, Kearns et al., 2014, Wang et al., 2010, Yang et al., 2010, Zhang et al., 2011)
241
report a trend of increasing adsorption with increasing SA or microporosity. In addition, atrazine
242
adsorption was found to decrease with increasing carbonization (as H/C ratio) or increasing
243
polarity [as (O + H)/C ratio] (Hao et al., 2013). However, our results are not in agreement with
244
these reported trends. Biochar B700 has the highest SA, largest microporosity, lowest H/C,
245
lowest O/C, and highest mean minimum fused ring cluster size, indicating it has the largest and
246
most hydrophobic surface of all the chars; thus, B700 was expected to have the greatest
247
adsorption affinity for the test compounds according to the previously-reported trends. However,
248
no straightforward relationship was found between atrazine or quinoline adsorption and each of
249
the following biochar properties: micropore SA (CO2 accessible), N2 BET SA, CO2-accessible
AC C
EP
TE D
M AN U
234
12
ACCEPTED MANUSCRIPT
microporosity, mesoporosity, H/C, O/C, and fused ring cluster size (Figs. S6 and S7,
251
Supplementary data). As shown in Figs. S4 and S6 in Supplementary data, adsorption of atrazine
252
peaked at B500, which has the highest mesoporosity. The same is true for quinoline. The results
253
are consistent with a recent study (Lattao et al., 2014) that found no straightforward relationship
254
between these same biochar properties and adsorption affinity for benzene, naphthalene, or 1,4-
255
dinitrobenzene.
RI PT
250
Lattao et al. (Lattao et al., 2014) found that the “reduced” carbon-normalized distribution
257
coefficient (Kb-C,r) for benzene, naphthalene, or 1,4-dinitrobenzene could be predicted by the
258
weighted sum of microporosity and mesoporosity:
M AN U
SC
256
K b-C,r = a ⋅ microporosity + b ⋅ mesoporosity
259
(5)
where a and b are the coefficients (g/cm3) corresponding to micropore adsorption and mesopore
261
adsorption, respectively. Kb-C,r is calculated from Cs-C (carbon-normalized Cs) using the
262
Freundlich equation and the reduced aqueous concentration. The reduced concentration is given
263
by
264
Cr = Cw Cwsat ( L)
TE D
260
(6)
where Cwsat ( L) is saturated water solubility of the liquid or subcooled liquid (atrazine, 1068 mg/L
266
(Capel et al., 2008); quinoline, 6110 mg/L (SRC, 2014)). In this study, Eq. 5 was extended to
267
include two heteroaromatic amines (atrazine and quinoline), and the values of Kb-C,r were
268
calculated at Cr of 0.001 and 0.01.
AC C
269
EP
265
A visual presentation of model performance of Eq. 5 is given in Fig. 2, and the regression
270
results appear in Table S1 (Supplementary data). The coefficient of determination (R2) falls in
271
the range 0.88−0.99 (Table S1), suggesting a strong correlation. However, in each case there are
272
only five data points (five biochars) against two independent variables, which may lead to over-
13
ACCEPTED MANUSCRIPT
fitting (Xiao et al., 2013). The adjusted coefficient of determination (Radj2), which takes into
274
account the degree of freedom, falls in the range 0.51−0.66, indicating a moderately strong
275
correlation. The prediction of Kb-C,r is good to within a factor of 3 (0.48 log units) for all but
276
three data points (Fig. 2).
Fig. 2 is near here.
3.3. Adsorption rates
SC
277 278 279 280 281
RI PT
273
Uptake rates of prometon, pyridine, and quinoline by biochars were determined at pH ~7, where
283
they are predominantly neutral in solution, and at pH ~2 where they are predominantly cationic.
284
The data were fitted to the diffusion model in Eq. 3. The uptake curves along with model fits are
285
shown in Fig. S8 (Supplementary data). The fits by Eq. 3 are generally better for the neutral than
286
the cationic species. The fitted values of Dbapp a 2 are given in Table S2 (Supplementary data) and
287
displayed in the bar graphs of Fig. 3.
288 Fig. 3 is near here.
EP
289 290
TE D
M AN U
282
Steric hindrance to diffusion becomes important as the ratio of pore aperture to the
292
minimum critical molecular diameter drops below ~10, and becomes severe as that ratio
293
approaches 1 (Kärger and Ruthven, 1992). The computed molecular volumes and shortest (width)
294
and longest (length) molecular dimensions are given in Table 1. The reference compounds,
295
benzene, naphthalene, pyridine, 1,3-diazine and quinoline are characterized by co-planarity of all
296
atoms in the molecule. The single-ring compounds vary 2.74−2.81 Å in length and the double-
297
ring compounds vary 5.02−5.07 Å in length (Table 1). The triazine herbicides are larger in both
298
dimensions than the reference compounds due to ring substituents. In addition, the triazines have
AC C
291
14
ACCEPTED MANUSCRIPT
greater bulk in the direction (z) perpendicular to the plane of the ring due to the inability of all
300
substituent atoms to simultaneously assume co-planarity with the ring atoms. This can further
301
restrict diffusion within slit-like pores, which become more prevalent as the biochar evolves to a
302
more polyaromatic structure with increasing pyrolysis temperature.
303 304 305
Table 1 is near here.
RI PT
299
For a given biochar, Dbapp a 2 follows the order, pyridine > quinoline > prometon, which
307
is the reverse order in their molecular volume. The value of Dbapp a 2 follows the same order for
308
the corresponding cationic form. An inverse correlation between Dbapp a 2 and molecular volume
309
is expected if steric effects are important. Quinoline is 79% longer and 56% more voluminous
310
than pyridine, and prometon is the largest among the three compounds in all dimensions (Table
311
1). The value of Dbapp a 2 reaches a maximum at B500 with all compounds, both in neutral and
312
cationic form. B500 has the largest mesoporosity and the highest ratio of mesoporosity to total
313
porosity (Fig. 1). This suggests that mesopores provide pathways for molecules to access deeper
314
and smaller pores. The slowest uptake for all three compounds occurred for B700, which has the
315
highest microporosity. Restricted diffusion will be manifested more in micropores than
316
mesopores, and the steric hindrance in B700 micropores may lead to its low adsorption affinity
317
towards solutes in comparison with B500 that has lower microporosity but greater mesoporosity.
318
It can also be seen in Fig. 3 that Dbapp a 2 is several times smaller for the cation than for
AC C
EP
TE D
M AN U
SC
306
319
the corresponding neutral molecule of pyridine, quinoline and prometon. This observation is
320
novel. The slower uptake of the cation is likely due to one or more of the following effects that
321
would retard its diffusion in pores compared to the neutral molecule: i) back diffusion of native
15
ACCEPTED MANUSCRIPT
cations (e.g., Na+) to maintain electroneutrality; ii) co-diffusion of a counterion (e.g., Cl−) to
323
maintain electroneutrality; or iii) a larger hydration sphere of strongly-bound water molecules
324
surrounding the cation and counterion. The relative contribution of these effects requires further
325
study.
RI PT
322
326
3.4. Probing electronic and steric effects using graphite as a reference adsorbent
328
The adsorption isotherms of the triazines and the reference compounds, pyridine and quinoline,
329
were determined on B400 and graphite for comparison. Graphite is nonporous and sparsely-
330
functionalized and should therefore interact primarily on the infinite polyaromatic sheet surface,
331
with some adsorption along edges and defects. The weight-based adsorption of the (nonplanar)
332
triazine herbicides was generally lower on B400 than on graphite (Fig. S9a), despite the higher
333
SA of B400 than graphite (N2-accessible SA: 9.3 m2/g vs 4.5 m2/g; CO2-accessible SA: 388 m2/g
334
vs 15.8 m2/g (Lattao et al., 2014)). On the other hand, on a weight basis the planar heterocyclic
335
amines (pyridine, quinoline, and 1,3-diazine) prefer B400 over graphite.
TE D
M AN U
SC
327
The N2 SA-normalized adsorption isotherms on B400 and graphite are shown in Fig. 4.
337
These isotherms are mirrored by the corresponding CO2 SA-normalized isotherms shown in Fig.
338
S9b (Supplementary data). Based on N2 SA, adsorption of the triazine herbicides is 101- to 102-
339
times greater on graphite than on B400. The origin of this preference for graphite may be steric
340
or electronic, or both. The steric bulk of the triazines may limit their access to interior pore SA of
341
B400 that is otherwise available to N2 at 77 K. In this regard, the preference for graphite is not
342
exhibited for the smaller, flatter molecules pyridine, quinoline and 1,3-diazine. An electronic
343
origin of the preference would be remarkable because graphite lacks polar (at least, dipolar)
344
functionality.
AC C
EP
336
16
ACCEPTED MANUSCRIPT
345 346 347
Fig. 4 is near here.
To better understand the relative importance of steric and electronic effects on adsorption,
349
the B400−water and graphite−water isotherms were used to construct the SA normalized
350
B400−graphite distribution isotherms for eight heteroaromatic amines (pyridine, 1,3-diazine,
351
quinoline, and five triazines) studied in the present work, in comparison with those for two
352
aromatic compounds (benzene and naphthalene) obtained from a previous study (Lattao et al.,
353
2014). The isotherms of a reference compound, 1,3,5-triazine could not be obtained because the
354
compound is unstable in water. The B400−graphite isotherms were constructed using the
355
experimental adsorption data points for B400 (Cw, Cs,b) and calculated values of Cs,g for graphite
356
at the same Cw using the Freundlich parameters for graphite converted to unit SA basis (Lattao et
357
al., 2014). The biochar−graphite distribution ratio is given by: K b -g =
358
TE D
M AN U
SC
RI PT
348
Cs,b C w Cs,b = Cs,g C w Cs,g
(7)
Figure S10a shows B400−graphite adsorption isotherms normalized by the N2 BET SA,
360
( SA bN and SA gN 2 , respectively). The same isotherms normalized by CO2 SA in Fig. S10b have
361
identical
362
2 2 SA bN 2 SA CO ⋅ ( SA gN 2 SA CO ) . The isotherms indicate increasing preference for graphite over b g
363
B400 in the order, benzene < pyridine ~ naphthalene < 1,3-diazine < quinoline < triazine
364
herbicides.
365 366
2
relationships,
AC C
spatial
EP
359
but
are
just
shifted
by
a
constant
equal
to
−1
The free energy of B400−graphite partitioning of a compound i, relative to i = benzene, can be expressed by (Lattao et al., 2014),
17
ACCEPTED MANUSCRIPT
367
∆i −BEN ∆ b-gG = ∆i − BEN ∆ b-g G ster + ∆i −BEN ∆ b-g G polar + ∆i −BEN ∆ b-gG hydr
(8)
where the terms on the right correspond to the contributions to overall free energy by steric,
369
polar, and hydrophobic effects. The polar term lumps dipole−dipole, dipole-induced dipole,
370
hydrogen bonding, and quadrupole−quadrupole forces. The hydrophobic effect term lumps
371
dispersion forces at the surface and interactions in the liquid phase that result in disruption of the
372
cohesive forces of bulk water. The hydrophobic free energy may be taken as proportional to the
373
free energy of solvent−water partitioning, such that
SC
∆i−BEN ∆b-gGhydr = ( ∆i −BEN ∆b-g a ) ∆i -BEN ∆S-wG + ( ∆i −BEN ∆b-gb )
(9)
M AN U
374
RI PT
368
where S is a suitable reference solvent and a and b are proportionality constants. For a closely-
376
related series of compounds, the hydrophobic effects driving adsorption to B400 and graphite are
377
expected to be roughly equal, which is equivalent to saying that the values of ∆i−BEN∆b-g a and
378
∆i−BEN∆b-gb are close to zero. Therefore, the hydrophobic term in Eq. 8 can be ignored.
379
Substituting the appropriate B400−graphite adsorption equilibrium constants for ∆BEN-i ∆b-gG , Eq.
380
8 becomes,
383
EP
382
∆ i −BEN ∆ b-g G ster + ∆ i −BEN ∆ b-g G polar = − RT ln
K b-g,i
K b-g,BEN
+ const
(10)
where const is a constant that depends on the reference states chosen for adsorption.
AC C
381
TE D
375
Figure 5 plots ∆ i − BEN ∆ b-g G ster + ∆ i − BEN ∆ b-g G polar versus the adsorbed concentration on
384
graphite for all the eight heteroaromatic amines studied in the present work and two aromatic
385
compounds (benzene and naphthalene) from a previous study (Lattao et al., 2014). One can see
386
that the curve for quinoline is positively displaced relative to the curve for pyridine. Since
387
quinoline and pyridine are expected to undergo similar polar interactions with B400, this
388
suggests that steric effects suppress quinoline adsorption relative to pyridine adsorption on B400. 18
ACCEPTED MANUSCRIPT
As such, the curve for naphthalene is positively displaced (less favorable adsorption by B400)
390
relative to the curve for benzene. Since naphthalene does not undergo significant polar
391
interactions with either biochar or graphite, the positive displacement of the curve for
392
naphthalene has been attributed to steric effects (Lattao et al., 2014).
RI PT
389
393 Fig. 5 is near here.
SC
394 395
The free energy curves for pyridine and 1,3-diazine are also positively displaced relative
397
to the curve for benzene. Since pyridine and 1,3-diazine are isosteric with benzene, this indicates
398
that pyridine and 1,3-diazine have greater affinity for graphite than for B400 for electronic
399
reasons. Pyridine and 1,3-diazine can undergo dipole−dipole, H bonding and π−π EDA
400
interactions with B400. However, since graphite has no significant functionality, of these forces,
401
π−π EDA interactions with graphite appear to dominate the ∆ i − BEN ∆ b-g G polar term for pyridine and
402
1,3-diazine.
TE D
M AN U
396
The π−π EDA force may occur when electron-poor and electron-rich π systems interact
404
through their quadrupoles. The polyaromatic surfaces of PCMs are π-electron rich, while the
405
solute molecule is π-electron poor by virtue of the electronegative N atoms incorporated into the
406
ring. The polyaromatic sheet of graphite is infinite, whereas the mean minimum fused ring size
407
of B400 is about 18 C atoms (Cao et al., 2012). The polyaromatic component of the total SA of
408
B400 is expected to be less than that of graphite due to the presence of incompletely carbonized
409
structure (Keiluweit et al., 2010) in biochar. In addition, B400 contains almost 20% O—some of
410
it in functional groups like hydroxyl, carboxyl, ether and carbonyl attached to rings—that likely
411
reduce the π-donor character of the ring systems.
AC C
EP
403
19
ACCEPTED MANUSCRIPT
Electrostatic potential (Ep, kJ/mol) has been used as an indicator of a compound’s π-
413
donor or π-acceptor character, and thus its potential to engage in π−π EDA interactions (Cockroft
414
et al., 2007, Mecozzi et al., 1996, Xiao and Pignatello, 2015). The molecular Ep surface of
415
solutes was determined by density functional theory (B3LYP, 6-31G* basis set; Spartan 10,
416
Wavefunction Inc., Irvine, CA) (Xiao and Pignatello, 2015). The value of Ep at the midpoint of
417
the ring (Ep,mid) was taken as a measure of the π-donor or π-acceptor character of the ring as a
418
whole, according to past practice (Xiao and Pignatello, 2015). The value of Ep,mid (kJ/mol)
419
increases with increasing N incorporation in the order, benzene (−96.0) < pyridine (−44.2)
