Predicting Organic Cation Sorption Coefficients - ACS Publications

Subscriber access provided by University of Colorado Boulder

Article

Predicting organic cation sorption coefficients: Accounting for competition from sorbed inorganic cations using a simple probe molecule William C. Jolin, Reaha Goyetche, Katherine Carter, John Medina, Dharni Vasudevan, and Allison A MacKay Environ. Sci. Technol., Just Accepted Manuscript • Publication Date (Web): 01 May 2017 Downloaded from http://pubs.acs.org on May 1, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28

Environmental Science & Technology

1

Predicting organic cation sorption coefficients:

2

Accounting for competition from sorbed inorganic

3

cations using a simple probe molecule

4

William C. Jolin†, Reaha Goyetcheҗ, Katherine Carter җ, John Medina җ, Dharni Vasudevanҗ*,

5

and Allison A. MacKay‡*

6

†

Department of Civil and Environmental Engineering, University of Connecticut, Storrs, CT

7

06269 җ

8 9

‡

Department of Civil, Environmental and Geodetic Engineering, The Ohio State University,

10

11

12

Department of Chemistry, Bowdoin College, Brunswick, ME 04011

Columbus, OH 43210

ABSTRACT

With the increasing number of emerging contaminants that are cationic at environmentally

13

relevant pH values, there is a need for robust predictive models of organic cation sorption

14

coefficients (Kd). Current predictive models fail to account for the differences in the identity,

15

abundance, and affinity of surface-associated inorganic exchange ions naturally present at

16

negatively charged receptor sites on environmental solids. To better understand how organic

17

cation sorption is influenced by surface-associated inorganic exchange ions, sorption coefficients

18

of ten organic cations (including eight pharmaceuticals and two simple probe organic amines)

ACS Paragon Plus Environment

1


Page 2 of 28

19

were determined for six homoionic forms of the aluminosilicate mineral, montmorillonite.

20

Organic cation sorption coefficients exhibited consistent trends for all compounds across the

21

various homoionic clays with sorption coefficients (Kd) decreasing as follows: KdNa+ > KdNH4+ ≥

22

KdK+ > KdCa2+ ≥ KdMg2+ > KdAl3+. This trend for competition between organic cations and

23

exchangeable inorganic cations is consistent with the inorganic cation selectivity sequence,

24

determined for exchange between inorganic ions. Such consistent trends in competition between

25

organic and inorganic cations suggested that a simple probe cation, such as

26

phenyltrimethylammonium or benzylamine, could capture soil-to-soil variations in native

27

inorganic cation identity and abundance for the prediction of organic cation sorption to soils and

28

soil minerals. Indeed, sorption of two pharmaceutical compounds to 30 soils was better described

29

by phenyltrimethylammonium sorption than by measures of benzylamine sorption, effective

30

cation exchange capacity alone, or a model from the literature (Droge & Goss, Environ. Sci.

31

Technol., 2013, 47, 14224). A hybrid approach integrating structural scaling factors derived from

32

this literature model of organic cation sorption, along with phenyltrimethylammonium Kd values,

33

allowed for estimation of Kd values for more structurally complex organic cations to homoionic

34

montmorillonites and to heteroionic soils (mean absolute error = 0.27 log units). Accordingly,

35

the use of phenyltrimethylammonium as a probe compound was concluded to be a promising

36

means to account for the identity, affinity and abundance of natural exchange ions in the

37

prediction of organic cation sorption coefficients for environmental solids.

38 39

INTRODUCTION

40

A growing number of environmental contaminants of interest, including pharmaceuticals,

41

surfactants, and ionic liquids,1 are positively charged under environmentally-relevant pH values,


2

Page 3 of 28


42

creating the need for accurate sorption models for organic cations.2-5 Traditional sorption

43

prediction approaches (e.g. EPISUITE6) accommodate only compounds that are neutral, with no

44

ionic moieties. In the case of organic cation sorption, positive charge on the amine moiety

45

interacts with negative charge sites on aluminosilicate clay mineral and organic matter

46

components of environmental solids.3-5,7-12 Local charge neutrality is maintained because the

47

sorbing organic cations displace a naturally-occurring inorganic cation from the charged site;

48

however, additional sorption energy contributions from the non-charged portions of the organic

49

cation (hydrophobic exclusion from the aqueous phase, polar interactions with the solid phase13-

50

15

51

counterparts for sorption sites at orders of magnitude lower concentration or abundance.17 As a

52

result, accurate predictive models for organic cation sorption must integrate both characteristics

53

of the environmental solids (site availability) and structural features of the compound itself.

- for more exhaustive listing see ref [16]) allows these compounds to outcompete their inorganic

54

To date, the greatest progress toward a quantitative model for organic cation sorption to

55

environmental solids has come through the detailed work of Droge and Goss.14,15,18 Their model

56

for organic cation sorption to soils integrates environmental solids characteristics through use of

57

the solid’s total cation exchange capacity and an attributed cation exchange contribution for the

58

organic matter content of the solids. Structural features of the sorbing cations are captured

59

through the compound molar volume, amine rank (primary, secondary or tertiary) and corrective

60

factors for additional structural moieties, such as multiple rings, -Cl groups, etc. In this model,

61

the extent of exchange site to soil aluminosilicate mineral and soil organic matter was separated

62

because sorption studies with pure phase minerals and organic matter showed distinct effects of

63

compound structure (sign reversal in empirical term) for the amine rank contributions to

64

sorption, as well as some differences in correction factors, between the two sorbent types.14,15


3


Page 4 of 28

65

Overall, this approach of separating cation exchange to aluminosilicate minerals and organic

66

matter and representing structural feature contributions with relationships developed from pure

67

phase sorbents (illite clay and Pahokee peat organic matter) was quite successful in predicting

68

organic cation sorption to two representative soils. This success advances prior modeling efforts

69

for organic cations that did not include fundamental site density measures5 or appropriate

70

structural feature representations.14

71

One limitation of the Droge and Goss model is the inability to account for inorganic

72

cation identity and abundance at negative charge sites. Earlier, we identified exchangeable

73

inorganic cation identity and abundance as a potentially important factor in organic cation

74

sorption.16 Droge and Goss’ model development with pure phase sorbents and validation efforts

75

with soils were undertaken in the presence of 5 mM CaCl2 that likely yielded Ca2+ to be the

76

charge balancing cations in their sorbent systems. Although calcium is an important soil cation,

77

the weathering and anthropogenic histories of many soils can yield solids with a mixture of

78

inorganic cations. Prior studies of inorganic cation sorption to soils19-21 and organic polymers22

79

show clear trends in ‘selectivity’, or relative affinity for the surface, between cations –

80

Li+ ≈ Na+ < NH4+ ≈ H+ ≈ K+ < Mg2+ < Ca2+ < Al3+

81

that suggest ion identity should play a role in organic cation exchange. From Li+ through to Al3+,

82

inorganic cations with increased affinity to soils should impart increased competition with

83

sorbing organic cations, thereby yielding reduced sorption coefficients for sorbing organic

84

cations. Indeed, this has been demonstrated in the case of Na+ versus Ca2+ as the saturating

85

cation: sorption coefficients for Na+-clays and soils are about one order of magnitude greater

86

than for Ca2+-clays and soils, a trend that is observed consistently for organic cations,

87

independent of sorbate structure.15,22-24 Extension of such comparisons to a broader suite of

(Sequence I)


4

Page 5 of 28


88

inorganic cations has been undertaken only for a small number of cationic pyridine, purine,

89

nucleotide and biguanide compounds.22,24 Measured sorption coefficients in these cases

90

decreased for homoionic clays with inorganic cations of greater relative affinity for the surface

91

according to Sequence I.

92

montmorillonite than Ca2+-montmorillonite, indicating that other properties, such as hydration of

93

the inorganic cation25-27, may also affect to organic cation sorption. It is unclear from such a

94

small dataset whether other inorganic cations impart a consistent influence on organic cation

95

sorption for all compound structures. The potential identification of such a consistent

96

competition sequence for inorganic cation displacement by organic cations opens the door to

97

quantitative integration of inorganic cation identity into organic cation sorption models. To this

98

end, we offer that complex variations in inorganic cation identity, affinity and abundance can be

99

better accounted for through the use of a cationic probe compound16,28 rather than the

100

cumbersome quantitative assessment of concentrations of individual exchange ions on

101

environmental solids.

The one exception was lower sorption observed for Mg2+-

102

Previously, we have conceptualized that experimentally measured sorption coefficients

103

for a cationic probe compound with a simple structure, such as benzylamine or

104

phenyltrimethylammonium (PTMA) (structures, Figure S1), could provide a combined measure

105

of cation exchange site density and the baseline driving force for sorption to an environmental

106

solid while implicitly accounting for competition from naturally-occurring exchangeable

107

inorganic cations .16 If a consistent sequence for inorganic cation displacement is noted for a

108

wide range of organic cation structures, we propose that the probe compound sorption coefficient

109

for a given sorbent should be correlated with sorption coefficients for other organic cations with

110

varied structures.

Thus, sorption coefficients for any organic cation, i, (Kid,pred) could be


5


Page 6 of 28

111

predicted from the experimentally-determined sorption coefficient of the probe compound to that

112

particular soil (Kd,exptprobe) and a structural scaling factors (Sisoil) that captures the free energy

113

differences in sorption arising from structural differences between the organic cation of interest

114

and the probe compound: , = ,

×

(1)

115

Structural differences, manifest in both hydrophobic and electrostatic contributions to sorption

116

interactions, could be determined experimentally or from empirical models.14,15,18 While this

117

proposed probe approach may imply the assumption of one negative site ‘type’ in Eq. 1, it does

118

still facilitate a hybrid approach in which the influence and distribution site characteristic

119

including, sorbed inorganic cation identities and abundance could be captured through Kd,exptprobe,

120

and the distinct influence of organic cation structural characteristics on sorption to

121

aluminosilicate clay minerals and organic matter (Sisoil) could be integrated explicitly through a

122

modified form of the Droge and Goss soil sorption model, for example. Without knowledge of

123

the exact framework for Sisoil, a simple structure for the probe compound was conceptualized

124

because we anticipated that it would be easier to add adjustments for other moieties (-Cl, -OH,

125

etc) to build up to more complex structures than to use a structurally-complex probe that could

126

be more representative of all of the potential sorption interactions yet require a more complicated

127

adjustment for the structural correction. With the limited application to date of probe compounds

128

for predicting sorption of other ionic organic compounds,28 there is a need to evaluate whether

129

probe compounds could be used to quantitatively account for the identity and abundance of a

130

heterogeneous group of exchangeable inorganic cations, using an extensive soil set. Note that

131

our emphasis herein are conditions of low sorbate surface coverage (< 2% sites) where linear

132

isotherms are applicable.29


6

Page 7 of 28


133

This study is aimed at improving the predictive capability of organic cation sorption to

134

soils. To this end, we (i) identify the influence of exchangeable inorganic cation identity on the

135

extent of organic cation sorption and quantify this competitive effect for homoionic

136

montmorillonite systems; (ii) establish measures of probe cation sorption (Kd,exptprobe) for

137

heteroionic systems (soils) that accesses overall organic cation exchange affinity of a soil or a

138

soil mineral, while capturing competition effects from naturally-occurring exchangeable

139

inorganic cations, and (iii) develop improved predictive models for organic cation sorption to

140

aluminosilicate minerals and soils (Eq. 1) that integrate measures of probe sorption (Kd,exptprobe)

141

with scaling factors (Si) delineated by leveraging the Droge and Goss structure-based models for

142

sorption to aluminosilicate minerals and organic matter.

143

METHODS

144

Sorbents and chemicals

145

Texas Ca-montmorillonite (STx-1, CEC = 0.844 meq kg-1) was obtained from the Clay

146

Minerals Society. Silicon carbide (SiC) was from Alfa Aesar. Thirty soil samples were collected

147

from 28 sites across the eastern United States as part of an earlier study by Jones et al.30 Soils

148

were previously characterized for native soil pH, effective cation exchange capacity (ECEC), and

149

concentrations of exchangeable Na+, K+, Ca2+, Mg2+, and Al3+ (Table S1). Sorbate compounds

150

phenyltrimethylammonium (PTMA), benzylamine, serotonin, atenolol, metoprolol, tramadol,

151

trimethoprim, propranolol, desipramine, and diltiazem (Figure S1) were from Sigma Aldrich. All

152

other chemicals and reagents were ACS grade. Solutions were made with high purity 18.2 MΩ

153

water (DI water) from a MilliQ system (Waters).

154

Sorption to Homoionic Montmorillonite: Kd Values Obtained by Column Chromatography


7


Page 8 of 28

155

Pulse input column chromatography was used to obtain sorption coefficients for 10 organic

156

cations (Figure S1) each with six homoionic (Na+, K+, NH4+, Ca2+, Mg2+, and Al3+)

157

montmorillonite systems. Montmorillonite was chosen as a representative aluminosilicate

158

mineral because of (i) the pH-independent nature of the cation exchange sites,21(ii) its use in

159

many prior studies,4,9,12,31,32 and (iii) similarities in sorbent behavior to other common clay

160

minerals.15 Columns were carefully packed with mixtures of montmorillonite (MMT) and inert

161

SiC. Following methods detailed and validated previously,33 montmorillonite-to-water ratios of

162

19, 5, and 0.75 grams of montmorillonite per liter of column pore water were used in individual

163

columns to balance compound retention against peak spreading. A comparative control column

164

was packed entirely with SiC to verify that test compounds had no sorptive interactions with this

165

inert solid. Column porosity was similar for both the ‘SiC-only’ and ‘SiC-montmorillonite’

166

columns because of the small amounts of montmorillonite used.33 Packed columns were loaded

167

into a standard HPLC system (Jasco PU-980 pump, AS-950 auto sampler, 40 µL injection loop,

168

and MD-1510 multiwavelength detector). Montmorillonite was converted to homoionic Na+,

169

NH4+, K+, Ca2+, and Mg2+ forms by flushing the respective column with air-equilibrated (pH 6 ±

170

0.3), 15 mM (NaCl, NH4Cl, KCl) or 5 mM (CaCl2, MgCl2) solutions for 24 hrs (2,800 pore

171

volumes) before compound injection. Al3+-montmorillonite was created by flushing with 2.5 mM

172

AlCl3 adjusted to pH 3 (HCl) to avoid precipitation of Al(OH)3. Flushing solutions were

173

designed to have ionic strengths that matched conditions of the Droge and Goss dataset15 and

174

they also served as the experimental background eluent solutions for experiments with the

175

respective clays. An operating flowrate of 100 µL min-1, previously determined to allow sorptive

176

equilibrium to montmorillonite,33 was used for all experiments. All experiments were performed

177

at room temperature (25°C).


8

Page 9 of 28


178

The concentrations of test compounds injected into the column were varied from 4.3 × 10-6 to

179

2.6 × 10-4 M in a solution that matched the background eluent. Sorption isotherms consisted of

180

at least three concentration points that were each measured in triplicate on both the ‘SiC-only’

181

and the ‘montmorillonite-SiC’ columns. Absorbance vs. time data collected following each

182

injection were exported directly to a MATLAB routine to calculate Kd values using compound

183

and tracer retention times as detailed in the Supplemental Information (SI, 3.1). All Kd values

184

were assumed to be in the linear range of the isotherm if surface coverage by organic cations was

185

less than 2%, two or more initial concentrations had similar Kd values (within 1 standard

186

deviation of each other) and skewness values of runs on ‘montmorillonite-SiC’ columns matched

187

(within 4%) those on the ‘SiC-only’ column.33 Sorption coefficients for benzylamine, PTMA,

188

and desipramine on Ca2+-montmorillonite were re-measured at the end of the study to ensure that

189

the columns remained unchanged through the duration of this work.

190

confirmed to provide identical retention times and sorption coefficients for these compounds

191

indicating that no loss of mass, or irreversible sorption, occurred.

192

Sorption to Soils: Kd from Batch Sorption Experiments

All columns were

193

Sorption of four organic cations (benzylamine, PTMA, tramadol, and desipramine) to 30

194

soils was measured using standard batch experiments. Column chromatography was not utilized

195

to evaluate sorption to soils because eluting columns with DI water or a background electrolyte

196

(NaCl or CaCl2) would modify the distribution of naturally occurring exchangeable inorganic

197

cations on the soil. A predetermined mass of soil (see SI, 3.2, for details) was placed in a 15-mL

198

centrifuge tube and 10 mL of 5 × 10-5 M solution (pH 6-7) of the test compound prepared in DI

199

water was added. Soil-free reactors, prepared in the same manner, provided a measure of initial

200

concentration and allowed us to confirm the absence of other loss processes. Soil-free and soil-


9


Page 10 of 28

201

containing reactors were rotated end-over-end for 18-24 hr (based on preliminary kinetic

202

experiments) in the dark to attain sorptive equilibrium. Then, solutions from the soil-free reactors

203

and supernatant from the soil-containing reactors were filtered with 0.45 µm nylon filers,

204

dispensed into clean HPLC vials and centrifuge tubes for analysis of aqueous organic cation

205

concentrations (Cw, mol L-1) and pH, respectively. Organic cation aqueous concentrations were

206

measured by high performance liquid chromatography with diode array detector (HPLC-DAD)

207

using an Agilent 1100 Series system (see SI for HPLC methods). Corresponding sorbed

208

compound concentrations (Cs, mol kg-1) were calculated by difference, and Kd values were then

209

obtained as Cs/Cw. Based on our earlier studies29, we note that our experimental conditions were

210

designed to ensure that Kd values obtained were in the linear range of sorption isotherm. Batch

211

experiments with soils were conducted using organic cation dissolved in DI water to preserve the

212

native heteroionic composition of the soil; we expected that the use of a buffer or a background

213

electrolyte would change native soil pH and the composition of exchangeable inorganic cations

214

associated with the soil. All experiments were performed in triplicate at native soil pH.

215

Desorption experiments using BaCl2 were used to confirm that loss from solution was due

216

primarily to sorption phenomena.

217 218

RESULTS AND DISCUSSION

219

Effect of Cation Identity on Organic Cation Sorption to Homoionic Montmorillonite

220

We evaluated whether exchangeable inorganic cations showed a consistent sequence of

221

competition across several organic cations by measuring the sorption coefficients of ten organic

222

compounds (Figure S1), including eight pharmaceuticals, with six homoionic montmorillonite

223

clays. Test organic cations were chosen to have a range of factors that could influence cation


10

Page 11 of 28


224

exchange, including hydrophobic exclusion from the bulk aqueous phase, amine rank and the

225

presence of polar groups. These structural differences contributed to a wide range in Kd values

226

across a given montmorillonite system (e.g., 83 to 29600 L/kg for Na-MMT, Table S3) even

227

though all test organic cations had the same charge of +1. Nevertheless, some important trends

228

were evident across the test organic cations. For all the organic cations examined, Kd values for

229

sorption to homoionic montmorillonite saturated with monovalent cations (Na-MMT, NH4-

230

MMT, K-MMT) were greater than Kd values for divalent homoionic montmorillonite (Ca-MMT

231

and Mg-MMT) which in turn were greater than values for montmorillonite saturated with

232

trivalent cations (Al-MMT) (Figure 1). The resulting Kd sequence for the test organic cations was

233

found to be:

234

KdNa-MMT > KdNH4-MMT ≥ KdK-MMT > KdCa-MMT ≥ KdMg-MMT > KdAl-MMT (Sequence II)

235

Because variations between KdX-MMT values are an implicit measure of the extent to which the

236

exchangeable inorganic cation, Xn+, suppressed organic cation sorption, the trend of cation order

237

from left to right in Sequence II represents increased competition for cation exchange sites from

238

inorganic cations. This inorganic cation ordering showed a similar pattern as Sequence I for

239

exchange between inorganic cations. The one exception to ordering between Sequence II and

240

Sequence I occurred for the divalent cations: Mg2+ showed stronger, or similar, competition with

241

organic cations than Ca2+ (Sequence II) compared to inorganic cation affinities to cation

242

exchange sites (Sequence I). This trend of KdCa-MMT ≥ KdMg-MMT for organic cations has been

243

observed previously by others.22,24


11


Page 12 of 28

244 245

Figure 1. Sorption coefficients (Log Kd) of organic cations to Na+ (grey), NH4+ (black), K+ (red),

246

Ca2+ (green), Mg2+ (yellow), and Al3+ (blue) exchanged montmorillonite followed a general

247

pattern of decreased sorption with increased inorganic cation charge. Compound numbers refer

248

to structures found in Figure S1.

249 250

The experimentally-determined Kd values from our set of test organic compounds were

251

used to quantify relative ion affinity constants for each inorganic cation. The ion affinity

252

constant, Exion-X, was defined as:

253

= log log

(2)

254

where KdX-MMT (L/kg) is the experimentally determined Kd value for montmorillonite homoionic

255

in cation X. Ca-montmorillonite was chosen as a reference homoionic clay because a detailed

256

structural model for organic cation sorption to Ca2+-saturated clay was developed by Droge and

257

Goss;15 by definition, Exion-Ca = 0. For the other five cations, Exion-X values were remarkably


12

Page 13 of 28


258

similar across the ten test organic cations. The average values were determined to be Exion-Na =

259

0.85 ± 0.06; Exion-NH4 = 0.65 ± 0.12; Exion-K = 0.58 ± 0.13; Exion-Mg = -0.15 ± 0.09 and Exion-Al = -

260

0.90 ± 0.13. First, we note that our Exion-Na for Na+ matches well with the +1 log unit offset in

261

cation exchange capacity-normalized Kd values recommended by Droge and Goss to translate

262

their structure-based organic cation sorption model from Ca2+-saturated to Na+-saturated clays.

263

Second, the coefficients of variation for Exion-X ranged from 7% (Na+) to 22% (K+), quantitatively

264

supporting our prior hypothesis that sorption of each organic cation was influenced similarly by a

265

particular inorganic cation. The only exception was Exion-Mg that varied by 60% across the

266

compounds because of the differential effect of Mg2+ on Kd values between compounds 1-7

267

(Exion-Mg = -0.17 ± 0.06) and compounds 8-10 (Exion-Mg = -0.017 ± 0.006). Our experimentally-

268

derived Exion-X values allow for simple adjustments to the Droge and Goss model to predict

269

organic cation sorption to other homoionic clays by adding the value of Exion-X to log Kd values

270

calculated from the Ca2+-clay model. (Note that Droge and Goss define sorption coefficients as

271

normalized to solid cation exchange capacity instead of mass as is our definition here; however,

272

application of a constant CEC value as a divisor to the two Kd terms in Eq. 2 yields the same

273

values for the two approaches.) The close similarity in Exion-X values across our test organic

274

cations shows that each inorganic cation affects the sorption of organic cations in a consistent

275

and predictable manner, regardless of organic cation structure. Importantly, this observation

276

supports the promise of using a probe cation to account for inorganic exchange ion identity,

277

affinity and abundance in the prediction of organic cation sorption coefficients.

278

Delineation of Scaling Factors (Si) and Incorporation of Kd,exptprobe into a Predictive Model for

279

Cation Sorption in Homoionic Systems


13


Page 14 of 28

280

Using our data for organic cation sorption to homoionic montmorillonites (Table S3),we

281

first validated the concept of using probe compound sorption (Kd,exptprobe), along with a scaling

282

factor (Si), to predict Kd values for other compounds. PTMA was chosen to be the probe

283

compound because structural corrections (Si, Eq. 1) would largely be additive to extrapolate to

284

other compounds in our test set. Although benzylamine also meets this criteria, the low Kd

285

values for this compound resulted in greater relative error (e.g., 30% for Al3+-MMT, Table S3),

286

thus favoring PTMA. Here, we implemented a hybrid approach by using Droge and Goss’

287

detailed structural model of organic cation sorption15 to calculate structural scaling factors by

288

utilizing differences between molar volume and amine rank for PTMA and organic cation i: log ( ! ) = 1.22(& &'( ) 0.22*+, ± ./ !

(3)

289

where Vxi (L mol-1) is the molar volume, NAii is amine rank or number of hydrogen atoms

290

bonded to the cationic amine group, and CFclay represents corrective factors for compound

291

structural moieties (Table S5). By definition NAi = 0 for PTMA. Sorption coefficients of the

292

other nine test compounds (all except PTMA) for the six homoionic clays were predicted with

293

Eq. 1 using Kd,exptPTMA for each clay type, along with scaling factors, Siclay, obtained using Eq. 3.

294

The resultant Kd,predX-MMT values were within 0.3 log units of the experimental values (Figure 2).

295

We note that predictions for each individual homoionic clay (Xn+-MMT) are uniformly

296

distributed around the 1:1 line, even for Mg2+-MMT that had the most variation in Exion-Mg values

297

across the compound set. These observations suggest that there was no bias in the application of

298

the probe concept (Eq. 1 with Eq. 3) for a particular homoionic clay.


14

Page 15 of 28


299 300

Figure 2. Application of PTMA as a probe compound to predict the sorption of other organic

301

cations onto MMT showed a strong correlation between predicted and experimental Kd values

302

(R2 = 0.97, p < 0.01), indicating that Kd,exptPTMA and Siclay were able to account for competition

303

from different inorganic exchange ions in the cases of homoionic Na+ (grey circles), NH4+ (black

304

circles), K+ (red circles), Ca2+ (green circles), Mg2+ (yellow circles) and Al3+ (blue circles)

305

montmorillonite. Solid line is 1:1 and dashed lines represent ± 0.3 log units.

306

A potential source of bias in application of PTMA as a probe to predict organic

307

compound sorption to the homoionic montmorillonite surface is the choice of CFclay factors in

308

the calculation of Si (Eq. 3). Our test compounds possessed six structural moieties for which

309

CFclay as defined by Droge and Goss (Table S5) were required. Poorly ascribed correction

310

factors for a given compound would shift predicted Kd,predX-MMT values systematically above or

311

below the 1:1 line (Figure 2). Of all the CFclay factors used, we observed that only the CFclay for

312

ether moieties (-COC-) were ineffective: Predicted Kd values were 2 to 10 times lower than

313

measured values, possibly resulting from overcorrection by assuming additivity when more than

314

one ether moiety was present. Therefore, values in Figure 2 were obtained by setting CFclay = 0


15


Page 16 of 28

315

for -COC- (Table S5). All other CFclay values assigned by Droge and Goss were effective in

316

arriving at our scaling factors. Together, these results indicated that Kd,exptPTMA could be used to

317

account for exchangeable inorganic cation identity in pure phase aluminosilicates without

318

explicit measures of exchangeable cation identities or surface abundances.

319 320

Establishing a Hybrid Approach for Using a Probe Compound to Predict Sorption to

321

Heteroionic and Heterogeneous Soil Sites

322

Soils are heteroionic systems with a variety of inorganic cations balancing negative

323

charge at surface sites. Quantitative analysis of inorganic cations displaced in the measurement

324

of CEC does reveal the distinct composition of inorganic cations associated with the soil (e.g.,

325

f(Exi), Table S5); however, it is unclear how empirical Exion-X values might be coupled with such

326

analyses to apportion the relative competition effects of inorganic soil cations on sorbing organic

327

cations. Therefore, we evaluated the potential for using sorption coefficients from a probe

328

compound to establish an aggregate measure of inorganic cation competition on aluminosilicate

329

mineral-containing soils.

330

Evidence that inorganic cation identity and abundance influences organic cation sorption

331

to soils was obtained through several regression analyses. First, we performed linear regressions

332

of experimental Kd,expt values with soil cation exchange capacity (Fig. 3A, S2A). Experimental

333

log Kd,expt values for PTMA, a compound with a simple structure, and tramadol, a

334

pharmaceutical, showed correlations with log ECECsoil that were statistically significant (p

Predicting Organic Cation Sorption Coefficients - ACS Publications

Recommend Documents