Accepted Manuscript Title: pider diagram: a universal and versatile approach for system comparison and classification. Application to solvent properties Author: E. Lesellier PII: DOI: Reference:
S0021-9673(15)00242-3 http://dx.doi.org/doi:10.1016/j.chroma.2015.02.017 CHROMA 356275
To appear in:
Journal of Chromatography A
Received date: Revised date: Accepted date:
1-12-2014 4-2-2015 5-2-2015
Please cite this article as: E. Lesellier, Sigmapider diagram: a universal and versatile approach for system comparison and classification. Application to solvent properties, Journal of Chromatography A (2015), http://dx.doi.org/10.1016/j.chroma.2015.02.017 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.
1
Σpider diagram : a universal and versatile approach for system comparison
2
and classification. Application to solvent properties
3
E. Lesellier
5
Université d’Orléans, Institut de Chimie Organique et Analytique (ICOA), CNRS UMR 7311,
6
B.P. 6759, rue de Chartres, 45067 Orléans cedex 2, France.
7
[email protected] ip t
4
8
10
ABSTRACT
cr
9
Classification methods based on physico-chemical properties are very useful in analytical chemistry, both for extraction and separation processes. Depending on the number
12
of parameters, several classification approaches can be used: by plotting two- or three-
13
dimensional maps (triangles, cubes, spheres); by calculating comparison values for one
14
system with reference to another one, i.e. the ranking factor F, or the Neue selectivity
15
difference s2; or with chemometric methods, (principal component analysis - PCA or
16
hierarchical cluster analysis - HCA). All these methods display advantages and drawbacks:
17
some of them are limited by the number of studied parameters (e.g. three for triangle or
18
sphere plots); others require a new calculation when changing the reference point (F; s2),
19
while for chemometric methods (PCA, HCA), the relationships between the clusters and the
20
physico-chemical properties are not always easily understandable.
an
M
ed
pt
21
us
11
From previous studies performed in supercritical fluid chromatography for stationary phase classification on the basis of linear solvation energy relationships (LSER) including
23
five parameters, we developed a classification map called the Σpider diagram. This diagram
24
allows plotting in a two-dimensional map the location of varied systems, having as many
25
parameters as the ones required to get a satisfactory classification. It can be three, five, eight,
26
or any number.
27
Ac ce
22
In the present paper, we apply this diagram, and the calculation mode to obtain this
28
diagram, to different solvent classifications: Snyder triangle, solvatochomic solvent
29
selectivity, Hansen parameters, and also to LSER Abraham descriptors and COSMO-RS
30
parameters. The new figure based on Snyder data does not change the global view of groups,
31
except by the use of corrected data from literature, and allows to add the polarity value onto
32
the map. For the solvatochromic solvent selectivity, it leads to achieve a better view of
33
solvents having no acidic character.
Page 1 of 52
34
For Hansen parameters, the “flattening” of the spherical view down to a single plane
35
could be found easier to use. For COSMO-RS and with Abraham descriptor, a more subtle
36
classification is achieved, mainly due to the use of five parameters instead of three. A strong
37
reversed correlation is established between the Rohrschneider polarity P’ and the normalized
38
V (molecular volume) parameter.
39
The study of the location of solvents used for reversed-phase liquid chromatography and the Arizona system for counter-current chromatography is discussed, as well as the
41
replacement of unsafe solvents by greener ones, or the use of these classifications for the
42
study of compound solubility.
cr
44
Besides, this paper also shows the ability to the spider diagram to plot on a single plane three axes from principal component analyses.
us
43
ip t
40
45
1. Introduction
an
46 47 48
Solvents are extensively used in the field of analytical chemistry, for separation or extraction. They are mainly chosen on the basis of the interactions they are able to develop
50
with analytes, which can be related to analyte solubility, or on solvent miscibility or non-
51
miscibility, in the case of counter-current chromatography (CCC) or liquid-liquid extraction.
52
Of course, it is well known that reversed-phase liquid chromatography (RPLC) mainly uses
53
hydro-organic mixtures with methanol or acetonitrile, normal-phase liquid chromatography
54
(NPLC) and thin-layer chromatography on silica plates both use mixtures of organic solvents,
55
while hydrophilic interaction chromatography (HILIC) essentially relies on acetonitrile-water
56
mixtures. What is more, beyond the necessary solubility, selectivity of the solvent, that is to
57
say the ability to separate classes of compounds, is rather based on subtle differences in
58
interactions between the liquid phase and the compounds. Besides, the change in the
59
stationary phase properties due to mobile phase adsorption should not be neglected in
60
chromatographic processes.
61
Ac ce
pt
ed
M
49
Varied classification systems were developed in the past to rank the solvents based on
62
their estimated interaction capabilities. Two types of scales were essentially used:
63
solvatochromic, based solely on the solvent properties, and eluotropic, measuring the solvent
64
properties in the presence of a stationary phase. For the first type, one can cite the Reichardt’s
65
scale (ET(30), Nile red), the solvent selectivity triangle (Snyder) (SST), the solvatochromic
Page 2 of 52
66
solvent selectivity (SSS, Kamlet and Taft), Hildebrand and Hansen solubility parameters
67
(HSP), Abraham solvation descriptors (LSER) and the COSMO-RS approach.
68
Some of these scales are based on spectroscopic measurements (SSS, ET(30)), whereas
69
others are based on energy measurements (SST, Hildebrand, Hansen, Abraham descriptors),
70
or by using theoretical descriptors on the base of the -potential (Cosmo-RS). Beyond the data, their use for a classification depends on the number of parameters
72
extracted. For two parameters, a plane is sufficient, and its understanding easy. For three data,
73
3D spaces (Hansen sphere) are required, or triangle with normalized data (Teas or Snyder
74
diagrams). For more descriptors, no direct representation is generally used but hierarchical
75
cluster analysis (HCA), or principal component analysis (PCA) are often used to reduce data
76
dimensionality. Both are obtained with a mathematical treatment applied to the parameters,
77
and plotted on one or several 2D graphs by combination of the principal components PCi
78
describing the higher variance percentage (PC1/PC2; PC1/PC3). However, this approach does
79
not evidence a simple relation between the classification and relevant descriptors.
an
us
cr
ip t
71
80
From the Rohrschneider’s gas-liquid partition data [1], Snyder defined the overall
82
measure of solvent strength by the P’ polarity scale, which varies from 0.1 for hexane to 10.2
83
for water [2,3]. The measurement of this gas-liquid partition coefficient is based on the ability
84
of liquids to dissolve six selected probes. These partition coefficients were measured by
85
analyzing the gas phase of a sealed flask maintained at 25°C during several hours, and
86
containing 2 ml of the solvent and 5 µl of the selected probes. They represent the ratio of the
87
concentration of the probes in the dilution solvent and of the concentration of the probes in
88
the gaseous phase above the liquid. Data were thus provided for 81 liquids.
ed
pt
Ac ce
89
M
81
Based on a reduced data set, by using three (ethanol, p-dioxane and nitromethane) of the
90
six initial reference probes (plus n-octane, toluene, 2-butanone), Snyder suggested three
91
parameters, xe, xd and xn respectively describing hydrogen bond basicity (ethanol), acidity (p-
92
dioxane) and the solvent dipolarity (nitromethane). To get these parameters, the values of the
93
gas/liquid partition coefficients (K) are adjusted with regard to the solvent molar volume
94
K’=KVs
95
(1)
96 97
then corrected by subtracting the distribution constant of an alkane (Kv) of the same molar
98
volume
Page 3 of 52
K”= K’/Kv
99
(2)
100 101
Consequently, this scale is a polarity scale, which does not include the capability for dispersive interactions. Other works had earlier shown that dispersive interactions represented
103
between 52 % (acetonitrile) and 70 % (acetone) of the total interactions (cohesion) in pure
104
solvents [4], possibly explaining the more or less accurate relation between the
105
solvatochromic scales which include or not the dispersive interactions. The sum of the
106
logarithm of these three corrected partition coefficients K” provides the P’ polarity value:
107 108
P’ = log(K”)ethanol + log(K”)dioxane + log(K”)nitromethane
110
(3)
us
109
cr
ip t
102
Then the ratios of logarithm of these adjusted partition coefficients and P’ :
an
111 112
M
113
(4)
114
determine xe (the part of basicity), xd (the part of acidity), xn (the part of dipole) for each
116
solvent. The sum of the three last parameters (xi) is equal to 1. For instance, for ethyl acetate,
117
the values are: xe = 0.36; xd = 0.22 and xn = 0.42, meaning that ethyl acetate mainly interacts
118
through dipole-dipole interactions (xn), and as an acceptor in hydrogen bonding (xe).
119
Obviously this statement depends on the solute properties, and could be valid for non-polar
120
compounds.
pt
Ac ce
121
ed
115
Then, each solvent was plotted in a selectivity triangle depending on these three relative
122
values. In this figure, solvents having the same properties were located in a close area. Eight
123
groups were defined. The drawbacks of this scale were discussed previously [5,6]. In
124
particular, this scale leads to classify alcohols solvents as basic, and aromatic solvents are
125
classified as acidic. These inconsistencies were also reported to justify the introduction of the
126
solvatochromic solvent selectivity approach [7].
127
In this approach, the three solvatochromic parameters (, and *; hydrogen bond
128
acidity, basicity and polarity-polarizability) were obtained by spectroscopic UV-visible
129
measurements [8]. The solvatochromic parameters were normalized. For instance, * = 0 for
130
cyclohexane and 1 for dimethylsulfoxide (DMSO) [9]. The parameter is calculated based on
Page 4 of 52
131
a reference (hexamethylphosphamide) equal to 1 [10], whereas the parameter is calculated
132
based on another reference (methanol) equal to 1 [11]. The values were also corrected for
133
dispersion interactions by referencing to an alkane of similar size, and then normalized.
134
However, the sum of the three solvatochromic parameters can be any values.
135
The correlations between the Snyder and the solvatochromic parameters were studied [7] and were rather disappointing. Another triangle was used to plot the data allowing the
137
solvent classification. However, due to the absence of acidity for aromatic solvents or amides,
138
amines, esters, a lot of solvents were located on a line joining the basic and dipolar summits
139
of the triangle, leading to a poor discrimination of a large number of solvents.
cr
140
ip t
136
The Hansen parameters [12,13] were derived from the Hildebrand ones [14].
Hildebrand described the use of solubility parameters (), which is related to cohesive energy
142
density (CED) through the energy of vaporization ΔE per unit volume Vm:
an
us
141
M
143
(5)
ed
144
pt
145
It is also related to four partial solubility parameters covering dispersive (d), dipole (o),
147
proton acceptor (b) and proton donor (a), based on free energy of vaporization of the pure
148
solvents per molar volume unit [15].
149
Ac ce
146
By extension, Hansen suggested that the total solubility parameter was the sum of three
150
contributions, dispersive (d), polar (p) (Keesom and Debye) and hydrogen bonding (h)
151
(donor and acceptor) [16]:
152 153
(6)
154 155
This simplification of the Hildebrand theory which described five partial solubility
156
parameters, also took into account the dispersive interactions, which were not included in the
157
Snyder and solvatochromic triangles. They were based on the energy of vaporization per unit
Page 5 of 52
volume, which is related to the cohesive energy of the liquid solvent. This energy is composed
159
by dispersive (Ed), polar (Ep) and hydrogen bonding (Eh) interactions. Thus compared to the
160
previous scales discussed, that scale added dispersive interactions, and joined the hydrogen
161
bond acidity and basicity. The representation of these parameters can be achieved by using a
162
three-dimensional space, which is rather uncommon. By calculating fractional parameters,
163
again normalized units, a triangle plot called Teas diagram [17,18] was used similarly to the
164
SST and the SSS parameters. A general discussion about the use of selectivity triangles
165
emphasized the varied uses of such plots [19].
ip t
158
The solvation parameter model describes five interactions by using five descriptors
167
related to the compound properties (Abraham descriptors) [20-21]. E is the excess molar
168
refraction, related to the presence of n- and -electrons resulting in charge transfer, -
169
interactions or dipole – induced dipole interactions; S stands for the presence of dipoles and
170
polarizability; A and B describe hydrogen bond acidity and basicity, and V is McGowan’s
171
volume, related to dispersive interaction and cavity energy formation.
us
an
172
cr
166
The V descriptor is calculated based on the compound structure (atoms and bonds). Instead of the V descriptor suited for transfer between condensed phases, the L descriptor can
174
be used for transfer from a gas to a condensed phase. L is related to the gas-liquid partition
175
coefficient with n-hexadecane as solvent at 25°C. This descriptor is well suited for the
176
investigation of gas chromatography systems. E is calculated from the refractive index, A and
177
B are determined by liquid-liquid partition, and S from liquid-liquid partition and
178
chromatographic measurements [22].
ed
pt
179
M
173
These descriptors are known for thousands of compounds. Linear solvation energy relationships (LSERs) calculated with these descriptors were extensively used to describe
181
liquid-phase and gas-phase chromatographic systems [23]:
182 183 184 185
Ac ce
180
X = c + vV + eE + sS + aA + bB
(7)
X being the studied property. The coefficients obtained from such models (e, s, a, b, v
186
or l) are related to the descriptors allowing the comparison of numerous stationary and mobile
187
phases [24] or solvents [25], or biological partition systems. However, due to the large
188
number of descriptors (five), no classification map can be easily drawn and these system
189
comparisons were generally achieved by using cluster analysis or ranking methods [24,25].
190
PCA approach seems ineffective because none of the resulting score plots provide a useful
Page 6 of 52
191
solvent classification, despite the high variance percentage (93%) described by the four
192
principal components [25].
193
The works of Klamt [26-32] reported that COSMO-RS calculates the dielectric
195
screening charges and energies of a van der Waals-like molecular surface. This calculation
196
considers the solute embedded in a virtual conductor, and uses a gas-phase reference energy.
197
From the plot of the -potentia, i.e. the plot of the chemical potential s() vs. the polarization
198
charge density of the surface fragment (), hydrophobicity and hydrogen bonding (donor,
199
acceptor) properties of a solvent can be calculated. The partition coefficients of a solute into a
200
solvent can be represented as a linear combination of -moments. The set of some relevant
201
moments allows to compare any solvents. Recently, 61 descriptors were extracted and treated
202
by PCA analyses (due to the great number of descriptors) [33,34]. These works classified
203
solvents into 10 clusters having various properties by using three principal components (PC1,
204
PC2, PC3), i.e. by a 3D space, which represented 85% of the data variance. PC1 and PC2
205
were closely related to hydrogen bond donor and acceptor, whereas PC3 could be correlated
206
both to lipophobicity and dipolarity. Moreover, some loss of information (15%) is noticed
207
when reducing the data set from 61 to 3 parameters [33], showing that the use of a reduced set
208
of data could achieve a performing classification.
cr
us
an
M
ed
209
ip t
194
The comparison of two solvent classifications [35] was also studied, by comparing the results provided by the use of the five LSER descriptors (E, S, A, B, V) and the five
211
COSMOments of Klamt’s [30-32]: Sig2 (overall electrostatic polarity) and Sig3 (asymmetry
212
of the profile) for polarity/polarizability, Hb don3 for hydrogen bond acidity, HB acc3 for
213
hydrogen bond basicity, and CSA for the surface area. It showed a large overlap of the
214
information content, despite a different distribution of the properties [35].
Ac ce
215
pt
210
Correlation between these five COSMOments and the Hansen parameters was also
216
checked [36], showing ability of the COSMOments to describe the Hansen parameter. We
217
must point out that both the LSER and Hansen approaches are heuristic ones, i.e. based on
218
numerous experiments, whereas the COSMO-RS one is rather a theoretical one, based on
219
time-consuming calculation. Such as previous conclusions, these studies [35,36] also indicate
220
that the solvent properties space can be well described by a five-dimensional space, and that
221
the use of the five basic moments avoids the over-parametrization when training the artificial
222
neutral network [36]
223
Page 7 of 52
224
The goal of this paper is to provide a new visual approach to the classification of
225
solvents, based on published data describing their chemical properties. Previously, we have
226
extensively used this classification called spider (or star) diagram for the classification of
227
supercritical fluid chromatography systems, which showed its ability for comparing numerous
228
and diverse data such as those describing very different chromatographic systems.
230
We applied this approach for numerous solvent classification systems, to provide a simple way of comparing the solvents used in varied separation and extraction systems.
ip t
229
231
233
cr
232
2. Material and methods
235
us
234
All data were collected in the literature and are presented in Tables 1 to 6. The spider diagram construction was first presented elsewhere [37]. Basically, a star comprising as many
237
branches as necessary (depending on the number of parameters, from three to five in this
238
paper) must first be plotted, with equal angle spacing between the branches. For models using
239
five parameters, they should not be placed at random but setting side by side those that are the
240
most positively correlated (for instance E and S, Sig2 and Sig3, or A and B, Hb Acc3 and Hb
241
Don3), in opposition those that are negatively correlated, while the least correlated ones
242
should be placed in an orthogonal fashion if possible. For LSER descriptors, other
243
arrangements of axes were studied, for instance E, A, B, S, V or B, S, A, E, V, but none of
244
these other arrangements provided a better separation of the groups of solvents in regards to
245
the one presented. The centre of the star serves as centre of the diagram, and each point is
246
plotted according to its parameters (Figure 1). The parameters can be viewed as the
247
coordinates of a solvation vector
248
point of this solvation vector. The way of reading this figure is not obvious at first sight as
249
anyone familiar with principal component analysis score plots would be tempted to interpret
250
the proximity of a point and an axis as an indication of the dominant factor. However, it is not
251
the case. The star is only represented to indicate the origin of the reference space and the
252
directions that allowed placing the points, but only the distances between the points are
253
significant. On figure 1, the continuous blue lines support the five values of the E, S, A, B, V
254
descriptor of formamide, making a geometrical figure, and the blue dotted lines indicate the
255
point displacement from the centre to the extremity of the solvation vector. For hexane, in
Ac ce
pt
ed
M
an
236
Ei, Si, Ai, Bi, Vi) [38], and each point is placed at the end-
Page 8 of 52
256
brown of figure 1, the displacement only follows the V axis because all the values for other
257
descriptors are equal to zero.
258
In the case of LSERs, we had shown that it was preferable to plot normalized
259
parameters. Normalization can be achieved by dividing the coordinates by the length of the
261
solvation vector, which is calculated as the square root of the sum of squares of all
262
coordinates. This way, the comparisons were related to the chromatographic system’s
263
selectivity rather than system’s retention. For other classification systems, other normalization
264
modes can be used.
cr
ip t
260
Moreover, to indicate the overall strength of interactions, the points are preferably
266
presented as bubbles, with the size of the bubbles being related to the length of the solvation
267
vector (in the case of LSERs) or to polarity parameters as further detailed below. Thus when
268
two bubbles are close in the spider diagram, it indicates that the solvents have close
269
selectivity, while the largest bubble indicates the strongest interactions.
an
us
265
The spider diagram for the Snyder and the solvatochromic data sets display the same
270
axes location as for the triangles in referenced literature. Examples of use of these
272
classification maps will be presented and discussed.
M
271
276 277 278
Results and discussions
3.1. Snyder’s solvent selectivity triangle (SST)
pt
275
3.
Ac ce
274
ed
273
Figure 2 shows the spider diagram obtained from the SST data (table 1). Each point
279
represents one solvent, and the bubble size is related to the polarity parameter P’. The small
280
black point into each group is the barycentre indicating the average value of all the solvents
281
included in one group. The group number, as well as the circles surrounding these groups are
282
the same as the ones used by Snyder. This number will also be used for figures 3 and 4,
283
plotting respectively the solvatochromic and the Hansen parameters.
284
Obviously, the relative position of each group to the other is the same as on the Snyder
285
triangle. One exception is for group 5 which is located close to the dipolar corner on the
286
Snyder triangle whereas it is located between groups VII and VIII on the spider diagram.
287
However, this change is not related to the data treatment, as it was announced by a re-
Page 9 of 52
288
evaluation of Snyder data [7], after the shift of chloroform and methylene chloride based on
289
new data values.
290
Besides, the addition of polarity values through the bubble size clearly indicates the high polarity of water in regards of the one of methanol and acetonitrile, the two organic
292
solvents classically used in RPLC. When going from methanol or acetonitrile towards water,
293
the polarity of the mobile phase increases, explaining the decrease in the eluotropic strength
294
observed in RPLC.
295
ip t
291
In this case, the introduction of the polarity P’ in the spider diagram can compensate for the lack of dispersive interactions, because there is a reversed relationship between polarity
297
(in the sense of the P’ calculation) and the dispersive interactions.
To conclude, for the Snyder’s data, the use of the spider diagram does not change the
us
298
cr
296
visual use of this classification, except by taking into account the solvent polarity. The
300
drawbacks of the measurements are identical, i.e. due to the choice of the three solvent probes
301
selected to assess the distribution coefficients.
302
3.2. Solvatochromic solvent selectivity
M
303 304
Figure 3 shows the spider diagram obtained from the solvatochromic parameters (table
ed
305
an
299
2). The spider diagram representation displays a clear advantage in regards to the triangle
307
graph: the aromatic solvents (group VII) and the amide solvents (group III) are not stacked
308
along an axis, but scattered in the selectivity space depending on their different basic
309
character. The alcohols (group II) clearly display their acidic and basic properties; acetic acid
310
is located with water due to its acidic character, whereas ethylene glycol and benzyl alcohol
311
are far from group VIII. The chlorinated solvents (group V) are always located between
312
groups VII and VIII, but some overlapping is observed between groups VIb and VII on the
313
one hand, or groups VIa and III on the other hand. To simplify diagrams from solvatochromic
314
data, the plot of the polarity π* vs. basicity βwas suggested in the goal to replace common
315
solvents by greener ones [40]. This simplification works only because the data set is first
316
divided into two groups: protic and non-protic solvents, otherwise the lack of the acidic value
317
α would lead to an overlapping of alcohols with a lot of non-protic solvents.
318
Ac ce
pt
306
The position of hexane, limonene and p-cymene, all close to the center of the spider
319
diagram because of their low values for each of the three plotted parameters, evidences the
320
lack of a parameter accounting for dispersive interactions in the SSS model.
321
Page 10 of 52
322
3.3. Hansen parameters
323 324
Figure 4 shows the spider diagram using the Hansen parameters (table 3). Because of the fusion of the two components of hydrogen bonds (acidic and basic), the value of h for
326
water is very high, as well as that of alcohols and acids, although to a lesser extent. Due to the
327
presence of a parameter accounting for dispersive interactions, the other organic solvents are
328
located in a narrower area than for the two previous diagrams (Fig.2 and Fig. 3). The addition
329
of dispersive interactions to the data set (with regards to the two previous scales discussed
330
above), and the mixture between acidic and basic hydrogen bonds both induce some
331
overlapping of several groups: II and IV, which contain alcohols and acetic acid, I (ester) and
332
VI, V (Chlorinated), VIb (nitrile) and VII (aromatic). The resulting clusters are thus not very
333
satisfying on the base of the groups defined by Snyder. However, this figure shows other
334
clusters including solvents having close behavior: methyl-t-butyl ether (MTBE) /
335
tetrahydrofuran (THF) / ethyl acetate / di-ethyl ether, or di-propyl ether / chloroform /
336
limonene, or benzene / toluene / carbon tetrachloride.
an
us
cr
ip t
325
As an example, the solubility of fullerene C60 is plotted on the same figure: it is located
338
very close to hexane, benzene, toluene, or carbon tetrachloride, probably indicating acceptable
339
solubility of C60 in these solvents. This is an example that the spider diagram can be used in
340
the goal to select suitable solvents for purification steps, or to study its compatibility in
341
polymers to provide new materials, due to its unusual optical and redox properties [41].
342
Besides, DMSO, which is often used as a dilution solvent before chromatographic analysis to
343
improve analyte solubility, is located at the centre, showing the ability to this solvent to
344
develop varied interactions with solutes.
Ac ce
pt
ed
M
337
345
The spider diagram can also be used when looking for environmentally-friendly
346
solvents for extraction processes in agreement with sustainable development (table 4).
347
Beyond solvent toxicity (dioxane, formaldehyde and acetonitrile), the question of “green
348
solvents” can be related to varied points: energy required to manufacture and number of
349
chemical steps in the synthesis (for instance, both of them are high for THF), energy to
350
distillate, ozone layer depletion, production from renewable resources etc [40, 42].
351
For natural compounds extracted from rapeseeds (fatty acids, triglycerides, tocotrienols,
352
tocopherols, sterols) [43], Figure 5 shows, as could have been expected, that alcohols are less
353
adapted than hexane for the extraction of lipophilic compounds. As was reported by the
354
authors, our diagram shows that hexane can be replaced by limonene, p-cymene or α-pinene
355
which all display close properties. The location of carbon dioxide [47] in the same area of the
Page 11 of 52
356
diagram shows its possible successful use to ensure high recovery of the studied compounds,
357
avoiding the concentration step of the extraction liquid when using terpenes as extractive
358
solvents. The Hansen parameters were recently used for the selection of green solvents to be used
360
in industrial processes [46]: triterpenes (α-terpineol), propylene carbonate, solketal, isosorbide
361
and glycol derivatives. For instance, the replacement of methylene chloride, largely used in
362
normal-phase and non-aqueous reversed-phase HPLC is of a prime interest [48].
ip t
359
Figure 5 shows that solketal or a mixture of carbon dioxide and methanol (90:10 v/v)
363
could be used instead of pure methanol. Unfortunately, the viscosity of solketal is about
365
eleven times the viscosity of water, meaning that no HPLC pumping system could provide the
366
necessary pressure to deal with such high viscosity [49]. On the opposite, the low viscosities
367
of CO2-organic solvent mixtures permit the use of both long column lengths and high flow
368
rates to ensure short and efficient analyses [50-54]. In fact, most green solvents are currently
369
applied to extraction processes or selective solvation, while their employment in
370
chromatography remains rather limited [43-47].
an
us
cr
364
However, in a recent publication, the use of propylene carbonate instead of acetonitrile
M
371
for HILIC was reported [55]. Because propylene carbonate is not fully miscible to water,
373
ternary mixtures with ethanol were prepared. Despite some drawbacks, such as the increase in
374
the pressure drop (the viscosity of propylene carbonate is equal to 2.4 cp at 25°C), some
375
reduction of both the chromatographic efficiency and the mass spectrometer sensitivity, the
376
use of propylene carbonate instead of acetonitrile for HILIC chromatography is affordable for
377
polar compounds. Moreover, it is suitable for large volume injection in bioanalytical analyses
378
[56].
pt
Ac ce
379
ed
372
The Hansen parameters are generally used to assess the solubility of compounds in a
380
solvent. For instance, Figure 6 shows the spider diagram for paracetamol solubility.
381
Compared to Hansen sphere, it is rather easier to find suitable solvents that show close
382
proximity to the paracetamol point (they are circled on Fig. 6). Nevertheless, as suggested
383
previously, it does not ensure the relevance of the calculated parameters with regards to the
384
real solubility data [57,58]. These data, as well as those further described in the present paper,
385
do not describe the effect of different polymorphs (crystalline forms) of a drug, which can
386
modify their respective solubility [59,60], and do not take account of the varied models used
387
to describe solvent properties.
388 389
Page 12 of 52
390
3.4. Abraham descriptors
391 392
Generally, the studies carried out with the LSER model use the model coefficients e, s, a, b, and v or l to describe the properties of the chromatographic or solubility systems. The
394
classification of solvents commonly used for separation processes, from values (e, s, a, b, l)
395
obtained for gas-to-solvent transfer of, was made by hierarchical cluster analysis [25]. For the
396
classification of these thirty-six solvents, seven clusters were defined, together with four
397
independent solvents (trifluoroethanol, water, dimethylsufoxide and dimethylformamide)[25].
398
Figure 7 shows these values (e, s, a, b, l) plotted for 30 solvents on the spider diagram. The
399
same seven clusters and the independent solvents are indicated on Figure 7. The location of
400
the clusters from the spider diagram is well in accordance with the one gained from the HCA
401
[25], but provides an additional information as it shows position of the solvents relative to
402
each type of interactions. For instance, acetonitrile and propylene carbonate are strongly
403
dipolar, displaying higher s values than alcohols or ketones, esters and ethers (KEE) (acetone,
404
dioxane…), whereas they are less acidic than alcohols and KEE. However, due to the
405
projection of five coefficients, which obviously causes some loss of information, acetonitrile
406
and propylene carbonate appear closer to KEE, whereas the HCA [25] indicated that they are
407
equally far from alcohols and KEE.
410
cr
us
an
M
ed
409
Nevertheless, this comparison shows the ability of the spider diagram for classifying the solvents on the basis of their chemical properties. Rather than plotting gas-to-liquid partition data (e, s, a, b, v), the molecular descriptors
pt
408
ip t
393
of the solvent molecules (E, S, A, B, V)(table 5), indicating their capabilities for defined
412
interactions, can also be compared (Figure 8). It is worth noting that these descriptors were
413
developed to be used as solute scales, not solvent scales, which can change the value of the
414
parameters (e.g., the basicity of bulk solvent hydrogen bond donors versus their behavior at
415
infinite dilution).
416 417
Ac ce
411
The bubble size here is defined by the ratio V/U. U is the length of the solvation vector associated to the chromatographic descriptor, and calculated from equation (8):
418 419
Ui
E i2 S i2 Ai2 B i2 V i 2
(8)
420 421 422
This vector length is a valuable tool to compare the strength of the interaction capabilities for each solvent.
Page 13 of 52
423
Figure 9 shows the relationship between P’, the Rohrschneider polarity, and either V or V/U. Obviously, the slope is negative because polarity, i.e. polar interactions, and the
425
McGowan volume V, i.e. the molecular volume indicating capabilities for dispersive
426
interactions are opposite. However, the regression coefficient is significantly improved when
427
using the normalized V/U rather than the V value, explaining our choice of V/U for the
428
diagram in Figure 8. With this parameter, the point size is inversely related to polarity,
429
explaining that the bubble representing water is the smallest, whereas the size for alkanes is
430
the largest and equal to 1. Additionally, the use of V/U is useful to compare the polarity of
431
solvents for which no P’ values are available.
cr
ip t
424
In a general way, several groups of solvents are rather well distinguished and their
433
relative location makes sense (Fig. 8). Moreover, the relative position of the solvents in Figure
434
8 is similar to the one observed Figure 7, indicating that the molecular descriptor values (E, S,
435
A, B, V) are appropriate to compare solvent properties.
an
436 437
us
432
As expected, water is located at the bottom right, showing its high acidity (A is large) and weak hydrophobicity (V is low). In the vicinity of water are located alcohols, then acetic
439
acid and formamide. Nitriles (like acetonitrile) display higher dipole interactions and are
440
located at the right-hand-side of the plot, above alcohols. Alkanes, with high hydrophobicity,
441
are naturally at the opposite of the figure, on the left, close to the V axis. Aromatic solvents
442
are at the top of the diagram, around the E axis. THF, 1,4-dioxane, acetone and ethyl acetate,
443
which were generally located in the same cluster or in close clusters in other diagrams (Fig. 2
444
and Fig. 3), are also located in the same group here, at the center of the spider diagram. As
445
expected, Figure 10a shows that these four solvents display close values for all the five
446
descriptors. Compared to previous figures, an improvement in the classification can be
447
noticed for instance for nitrobenzene and nitromethane. These two solvents were classified
448
with toluene and benzene in the Snyder diagram (Fig. 2), whereas nitrobenzene is now closer
449
to benzonitrile, and nitromethane to acetonitrile. With regards to the Hansen plot (Fig. 4), the
450
use of five descriptors provides a better discrimination of MTBE and ethyl-acetate, two
451
solvents which are often used for liquid extraction of plant matrices [65,66], ethyl-acetate
452
having a higher dipole (S) value (Fig. 10a).
Ac ce
pt
ed
M
438
453
However, some discrepancies are noticed. First, the location of ethers that are placed
454
close to alcohols. Figure 10b shows the descriptor values for 1-butanol, dimethyl-, diethyl-
455
and dipropyl-ether. Ethers have zero values for E and A descriptors, whereas 1-butanol
456
displays comparable values for these two descriptors. Due to the spatial arrangement selected
Page 14 of 52
457
for the five descriptors in the spider diagram, E and A axes are almost opposite. As a result,
458
the two combinations, close values for E and A (1-butanol) or zero values for E and A (ethers)
459
yield a close location for these solvents, because they also display almost identical values for
460
S, B and V. A similar issue can be reported by comparing the location of pyridine and chlorinated
461
solvents (methylene chloride, chloroform and carbon tetrachloride). Figure 10c shows that
463
pyridine has greater B, E and S values than chloroform, but the combination of the three
464
differences, arranged in opposite directions (B vs. E and S), leads to a close location of the
465
two solvents, whereas the solvents properties are not exactly identical. However a close
466
location of chlorinated solvents and pyridine was also reported on the Hansen diagram (Fig.
467
3).
cr
us
468
ip t
462
Finally, Figure 10d shows the comparison of descriptors for acetic acid and ethylene glycol, which are mainly differing be the E and B values, but once again, these small
470
differences compensate on the diagram leading to a very close location.
471
an
469
The relationship between the miscibility and the location of the solvents onto the spider diagram can be discussed. In the group including THF, 1,4-dioxane, acetone and ethyl acetate,
473
the first three are miscible with water, whereas the solubility of ethyl acetate in water is only
474
partial (around 10% at ambient temperature). MTBE is not miscible to water neither. When
475
looking at the descriptor values (Fig.10a), one can see that whatever the values of E, S, A and
476
B, the two non-miscible solvents (MTBE and ethyl acetate) display V values higher than 0.7,
477
whereas the three others (THF, 1,4-dioxane and acetone) have lower V values. This would
478
indicate that solvent miscibility to water is first related to dispersive interactions, i.e. is not
479
related to the global location onto the diagram but to specific solvent properties. There are
480
two exceptions to this hypothesis, for chlorinated solvents: chloroform and methylene
481
chloride (V is respectively equal to 0.617 and to 0.494) (Table 5). However, among the
482
solvents having a V value lower than 0.7, chloroform and methylene chloride do not display
483
basic properties, probably explaining, in that case, the non miscibility between water and
484
these two solvents. However, we may point out that the location of chloroform and
485
dichloromethane in the diagram is far away of water. We can thus conclude that good
486
miscibility to water is ensured with small molecular volume, together with hydrogen-bond
487
acceptor capabilities.
Ac ce
pt
ed
M
472
488 489 490
This diagram (Figure 11) also allows comparing the location of the solvents used, either for RPLC (water, methanol, acetonitrile, THF) or for the Arizona system of solvents used in
Page 15 of 52
491
counter-current chromatography (heptane - ethyl acetate – methanol - water) [65,68-69], or
492
for other CCC solvent systems (chloroform – methanol - water or ethyl acetate- 2-butanol -
493
water)[70] and liquid-liquid extraction solvents.
494
As regards RPLC, we have already pointed out that the high acidic character of water together with a low V value leads to a location at the bottom right of the diagram (Fig. 8). The
496
V value for the three organic solvents usually combined to water in RPLC (methanol,
497
acetonitrile or THF) is lower than 0.7, ensuring their good miscibility. Besides, Figure 12a
498
shows that the E and B values of the three RPLC organic solvents are not strongly different,
499
explaining their relative proximity in the spider diagram. It also displays that the three organic
500
solvents are complementary, methanol being the most acidic, and acetonitrile displaying the
501
greatest dipole value. The use of THF, having the greatest V value, favors the higher
502
solubility of most of the organic compounds through dispersive interactions, explaining its
503
high eluting strength in RPLC.
cr
us
an
504
ip t
495
Figure 12b shows the changes in the descriptor values for the four solvents used in the Arizona system for counter-current chromatography (hexane, ethyl acetate, methanol, water).
506
Good miscibility between hexane and ethyl acetate is related to the high V descriptor, which
507
shows that these two solvents are able to interact by dispersive interactions. Basic character
508
and dipole interactions favor mixing between ethyl acetate and methanol, whereas hydrogen
509
bond acidity, basicity, and dipole interactions allow good miscibility between methanol and
510
water.
ed
Some attempts were done to replace heptane by limonene, in the goal to reduce the
pt
511
M
505
toxicity of the solvents used in CCC [71]. Limonene is a cyclic monoterpene coming from
513
orange and citrus stripper oil, easily available and renewable, which is considered as a greener
514
solvent than alkanes. This study showed that the use of limonene instead of heptane in CCC
515
slightly favored the solubility of benzene, toluene and diethylphthalate. When comparing the
516
location of the two solvents (limonene and heptane) to the ones of the extracted compounds
517
(benzene, toluene and diethylphthalate) (Fig. 11), it appears that limonene is as close to these
518
three compounds as is heptane. This example evidences the usefulness of the spider diagram
519
in the search of sustainable substitute solvents for various applications.
Ac ce
512
520 521
3.5. COSMO-RS descriptors
522 523 524
Figure 13 displays the classification obtained from the five COSMOments CSA, Sig2, Sig3, Hb acc3 and Hb don3 reported in other papers [30-32,35](table 6). Such as for Abraham
Page 16 of 52
descriptors, these values are calculated for isolated compounds, but were previously used for
526
solvent classification by using PCA [33-34]. The spatial arrangement of the five axes is close
527
to the one used for the LSER classification, as CSA and V are related to the surface or volume
528
of the molecule; Sig2, Sig3, E and S are related to polarizability and dipole interactions; Hb
529
acc3, Hb don3, A and B are related to hydrogen bond donors and acceptors. A comparison
530
between these two data sets showed the large overlap of the two descriptor sets, although the
531
chemical information was differently distributed among the descriptors [35]. Indeed, the E
532
term is not well correlated to the COSMOments, whereas if A and B are mainly depending on
533
Hb don3 and Hb acc3, Sig2 and Sig3 are also related to A and B, meaning that the chemical
534
content of Sig2 and Sig3 probably includes a part of A and B information.
cr
ip t
525
Figure 13 can be compared to Figure 8 in terms of relevance of clusters. Despite the
536
similar ordering of the five parameters (V, E, S, A, B) and (CSA, Sig2, Sig3, Hb don3, Hb
537
acc3), the location of the clusters is not identical. Some of these clusters are well separated
538
from the others (as alkanes, chloro-alkanes and aromatics), but others overlap (as alcohols and
539
non-protic dipolar solvents ethylacetate, THF, acetonitrile and acetone). Several papers report
540
the good relevance of this classification on the base of hydrogen bonds, but also underline its
541
weakness for taking into account the dipolar character and the dispersive interactions [33,
542
72,73].
an
M
ed
543
us
535
With the data of the PCA used for this classification based on the COSMO-RS data[33], we can also apply the spider diagram to plot the coordinates of the first three principal
545
components (F1, F2 and F3) on the same figure (figure 14). It shows that this type of spider
546
diagram can also be applied to present results from PCA with only one figure including the
547
three main components F1, F2 and F3, whereas this type of result is presented with bi-
548
dimensional plots, F1 vs F2 or/and F1 vs F3. Figure 15 shows the spider diagram for classical
549
solvents from this PCA coordinates. This classification, obtained on the basis of 61 COSMO-
550
RS descriptors, displays some surprising ranking, for instance, nitromethane is located in the
551
same group as water; alkanes and aromatics are together in a group of non-polar solvents;
552
acetone, acetonitrile, DMSO and 1,4-dioxane are located in the same group; N-
553
methylformamide is clustered with methanol. On another hand, methanol is not included in
554
the same group as ethanol.
Ac ce
pt
544
555 556 557
4. Conclusion
Page 17 of 52
558 559 560 561
The spider diagram approach has been applied for classification of solvents on the basis of their physico-chemical properties. This diagram provides a universal way to compare the varied and well known classification scales, and overcomes some difficulties in the representation of properties
563
encountered with other visual presentations (triangles, spheres, cubes), for instance in the case
564
of zero values for one parameter (solvatochromic parameters), or for a parameter number
565
higher than three (Abraham descriptors, COSMOments). Obviously, this new presentation
566
does not modify the raw data, i.e. their relevance or their failure. At least, it allows to add to
567
the studied interactions one more parameter, i.e. the polarity P’ or the total solubility
568
parameter for the Snyder, the solvatochromic or the Hansen parameters, which is represented
569
through bubble size on the map.
us
cr
ip t
562
Applied to Abraham descriptors, and with combination of the V/U parameters, it leads
571
to a simple view of the solvent groups having similar or different properties. By comparison
572
with the diagrams using three parameters, this classification, using five parameters, is slightly
573
different, and the shift of some solvents with regards of other classification scales makes
574
sense, from a chemist’s point of view. Besides, the plot of COSMOments could appear less
575
relevant than the plot based on Abraham descriptors, because of the overlapping of some
576
clusters.
M
ed
577
an
570
Finally, the spider diagram permits to compare a lot of scales of solvent properties by using an identical treatment of data , and is very useful to select suitable solvents with
579
regards of the desired analytical method, for extraction, separation or purification approaches,
580
and for solubility studies, or as an aid for choosing greener solvents.
Ac ce
581
pt
578
Despite the simplicity of the spider diagram classification, and whatever the scales used,
582
the future limitations of its use will be related to the calculation of parameters or descriptors
583
for new solvents or for mixtures of solvents, for instance for modifier/CO2 mixtures.
584 585 586
References
587 588
[1] L. Rohrschneider, Solvent characterization by gas-liquid partition coefficient of selected
589
solutes, Anal. Chem. 45 (1973) 1241-1247.
590
[2] L.R. Snyder, Classification of the solvent properties of common liquids, J. Chromatogr.
591
Sci. 16 (1978) 223-234.
Page 18 of 52
[3] L.R. Snyder, Classification of the solvent properties of common liquids, J. Chromatogr. 92
593
(1974) 223-230.
594
[4] E.F. Meyer, T.A. Renner, K.S. Stec, Cohesive energies in polar organic liquids. II. The n-
595
alkyl nitriles and the 1-chloroalkanes, J. Phys. Chem. 75 (1971) 642-648.
596
[5] L. Szepesy, Possibilities and pitfalls in defining selectivity in HPLC, Chromatographia, 51
597
(2000) S98-S107.
598
[6] V.J. Barwick, Strategies for solvent selection - A literature review, Trends Anal. Chem. 16
599
(1997) 293-309.
600
[7] S.C. Rutan, P.W. Carr, W.J. Cheong, J. H. Park, L.R. Snyder, Re-evaluation of the solvent
601
triangle and comparison to solvatochromic based scales of solvent strength and selectivity, J.
602
Chromatogr. 462 (1989) 21-37.
603
[8] M.J. Kamlet, R.W. Taft, P.W. Carr, M.H. Abraham, J. Chem. Soc. Faraday Trans. I,
604
Linear solvation energy relationships. Part 9-Correlation of gas/liquid partition coefficients
605
with the solvatochromic parameters, π*, α and β, 78 (1982) 1689-1704.
606
[9] M.J. Kamlet, J. L. Abboud, R.W. Taft, The solvatochromic comparison method. 6. The π*
607
scale of solvent polarities, J. Am. Chem. Soc. 99 (1977) 6027-6038.
608
[10] M.J. Kamlet, R.W. Taft, The solvatochromic comparison method. I. The β scale of
609
solvent hydrogen-bond acceptor (HBA) basicities, J. Am. Chem. Soc. 98 (1976) 377-383.
610
[11] M.J. Kamlet, R.W. Taft, The solvatochromic comparison method. II. The α scale of
611
solvent hydrogen-bond acceptor (HBD) acidities, J. Am. Chem. Soc. 98 (1976) 2886-2894.
612
[12] C.M. Hansen, The three dimentional solubility parameters - Key to paint component
613
affinities. I. Solvents, plasticizers, polymers and resins, J. Paint Technol. 39 (1967) 104-117.
614
[13] C.M. Hansen, Solvent for coating, Chem. Tech. 2 (1972) 547-553.
615
[14] J.H. Hildebrand, J.M. Prausnitz, R.L. Scott, Regular and Related Solutions, Van
616
Nostrand-Reinhold, New York, 10970.
617
[15] J.H. Hildebrandt, R.L. Scott, The solubility of non-electrolytes, 3rd ed. Dover,
618
Publications, New York, 1964.
619
[16] C.M. Hansen, Hansen solubility parameters: A user’s handbook, CRC press, Inc., Boca
620
Raton FL, 2007.
621
[17] J.P. Teas, Graphic analysis of resin solubility, J. Paint Technol. 40 (1968) 19-25.
622
[18] E. Stefanis, C. Panayiou, Prediction of Hansen solubility parameters with a new group-
623
contribution method, Int. J. Thermphys. 29 (2008) 568-585
624
[19] A.R. Johnson, M.F. Vitha, Chromatographic selelctivity triangles, J. Chromatogr. A 1218
625
(2011) 556-586.
Ac ce
pt
ed
M
an
us
cr
ip t
592
Page 19 of 52
[20] M.H. Abraham, Scales of solutes hydrogen-bonding: their construction and application to
627
physicochemical and biochemical processess, Chem. Soc. Rev. 22 (1993) 73-83.
628
[21] M.H. Abraham, A. Ibrahim, A. M. Zissimos, Determination of solute descriptors from
629
chromatographic measurement, J. Chromatogr. A 1037 (2004) 29-47.
630
[22] C. Poole, S.N. Atapattu, S.K. Poole, A.K. Bell, Determination of solute descriptors by
631
chromatographic method, Anal. Chim. Acta 652 (2009) 32-53.
632
[23] M. Vitha, P.W. Carr, The chemical interpretation and practice of linear solvation energy
633
relationships in chromatography, J. Chromatogr. A 1112 (2006) 143-194.
634
[24] C.F. Poole, S.K. Poole, Column selectivity from the perspective of the solvation
635
parameter model, J. Chromatogr. A 965 (2002) 263-299.
636
[25] C. Poole, T. Karunasekara, Solvent classification for chromatography and extraction, J.
637
Planar Chromatogr. 25 (2012) 190-199.
638
[26] A. Klamt, Conductor-like screening model for real solvents; a new approach to the
639
quantitative calculation of solvation phenomena, J. Phys. Chem. 99 (1995) 2224-2235.
640
[27] A. Klamt, V. Jonas, T. Burger, J.C.W. Lohrenz, Refinement and parametrization of
641
COSMO-RS, J. Phys. Chem. A 102 (1998) 5074-5085.
642
[28] A. Klamt, F. Eckert, Cosmo-RS: a novel and efficient method for the a priori prediction
643
of thermophysical data of liquids, Fluid Phase Equi. 172 (2000) 43-72.
644
[29] F. Eckert, A. Klamt, Fast solvent screening via Quantum chemistry: Cosmo-RS
645
approach, AIChe 48 (2002) 369-385.
646
[30] C. Mehler, A. Klamt, W. Peukert, Use of COSMO-RS for the prediction of adsorption
647
equilibria, AIChe 48 (2002) 1093-1099.
648
[31] A. Klamt, F. Eckert, M. Hornig, Cosmo-RS: a novel view to physiological solvation and
649
partition questions, J. Comput. Aid. Mol. Des. 15 (2001) 355-365.
650
[32] A. Klamt, F. Eckert, M. Diedenhofen, Prediction of soil sorption coefficient with a
651
conductor-like screening model for real solvents, Env. Tox. Chem. 21 (2002) 2562-2566.
652
[33] M. Durand, V. Molinier, W. Kunz, J.M. Aubry, Classification of organic solvents
653
revisited by using the COSMO-RS approach, Chem. Eur. J. 17 (2011) 5155-5164.
654
[34] L. Moity, M. Duran, A. Benazzouz, C. Pierlot, V. Molinier, J.M. Aubry, Panorama of
655
sustainable solvents using the COSMO-RS approach, Green Chem. 14 (2012) 1132-1145.
656
[35] A. M. Zissimos, M.H. Abraham, A. Klamt, F. Eckert, J. Wood, A comparison between
657
the two general stes of linear free energy descriptors of Abraham and Klamt, J. Chem. Inf.
658
Comput. Sci. 42 (2002) 1320-1331.
Ac ce
pt
ed
M
an
us
cr
ip t
626
Page 20 of 52
[36] G. Jarvas, C. Quellet, A. Dallos, Estimation of Hansen solubility parameters using
660
multivariate non linear QSPR modeling with COSMO Screening charge density moment,
661
Fluid Phase Equi. 309 (2011) 8-14.
662
[37] C. West, E. Lesellier, Characterisation of stationary phases in subcritical fluid
663
chromatography by the solvation parameter model II. Comparison tools, J. Chromatogr. A
664
1110 (2006) 191-199.
665
[38] Y. Ishihama, N. Asakawa, Characterization of lipophilicity scales using vectors from
666
solvation energy descriptors, J. Pharm. Sci. 88 (1999) 1305-1312.
667
[39] J.H. Clarck, D. J. Macquarrie, J. Sherwood, A quantitative comparison between
668
conventional and bio-derived solvents from citrus waste in esterification and amidation
669
kinetic studies, Green Chem. 14 (2012) 90-93.
670
[40] P. G. Jessop, Searching for green solvents, Green Chem. 13(2011)1391-1398.
671
[41] C. M. Hansen, A. L. Smith, Using Hansen solubility parameters to correlate solubility,
672
Carbon 42 (2004) 1591-1597.
673
[42] C. Capello, U. Fisher, K. Hungerbühler, What is a green solvent? A comprehensive
674
framework for the environmental assessments of solvents, Green Chem. 9 (2007) 927-934.
675
[43] Y. Li, F. Fine, A-S. Fabiano-Tixer, M. Albert-Vian, P. Carre, X. Pages, F. Chemat,
676
Evaluation of alternative solvents for improvement of oil extraction from rapeseeds, C. R.
677
Chimie, 17 (2014) 212-217.
678
[44] T. Fardi, E. Stefanis, C. Panayiotou, S. Abbott, S. van Loon, Artwork conservation
679
material and Hansen solubility parameters: a novel methodology towards critical solvent
680
selection, J. Cul. Heritage, 2013, in press
681
[45] web address: https://pirika.com
682
[46] A. Benazzouz, L. Moity, C. Pierlot, M. Sergent, V. Molinier, J. M. Aubry, Selection of a
683
greener set of solvent evenly spread in the Hansen space by space-filling design, Ind. Eng.
684
Chem. Res., 52 (3013) 16585-16597.
685
[47] M.W. Rowe, J. Phomakay, J.O. Lay, O. Guevara, K. Srinivas, W.K. Hollis, K.L.
686
Steelman, T. Guilderson, T.W. Strafford Jr., S.L. Chapman, J.W. King, Application of
687
supercritical carbon dioxide solvent mixtures for removal of organic material from
688
archeological artifacts for radiocarbon dating, J. Supercri. Fluids, 79 (2013) 314-323.
689
[48] J. P. Taygerly, L.M. Miller, A. Yee, E.A. Peterson, A convenient guide to help select
690
replacement solvents for dichloromethane in chromatography, Green Chem. 14 (2012) 3020-
691
3025.
Ac ce
pt
ed
M
an
us
cr
ip t
659
Page 21 of 52
[49] J. I. Garcia, H. Garcia-Martin, E. Pires, Glycerol based solvents: synthesis, properties and
693
applications, Green Chem. 16 (2014) 1007-1033.
694
[50] F. S. Deschamps, E. Lesellier, J. Bleton, A. Baillet, A. Tchapla, P. Chaminade,
695
Glycolipid class profiling by packed column subcritical fluid chromatography, J. Chromatogr.
696
A 1040 (2004) 115-121.
697
[51] K. Gaudin, E. Lesellier, P. Chaminade, D. Ferrier, A. Baillet, A. Tchapla, Retention
698
behaviour of ceramides in sub-critical fluid chromatography in comparison with non-aqueous
699
reversed-phase liquid chromatography, J. Chromatogr. A 883 (2000) 211-222.
700
[52] E. Lesellier, A. Latos, A. Lopes de Oliveira, Ultra high efficiency/low pressure
701
supercritical fluid chromatography with superficially porous particles for triglyceride
702
separation, J. Chromatography A 1327 (2014) 141-148.
703
[53] D.A. Roston S. Ahmed, D. Williams, T. Catalano, Comparison of drug substance
704
impurity profiles generated with extended length columns during packed-column SFC, J.
705
Pharm. Bioned. Ana. 26 (2011) 339-355.
706
[54] V. Abrahamsson, I. Rodriguez-Meizoso, C. Turner, Determination of carotenoids in
707
microalgae using supercritical fluid extraction and chromatography, J. Chromatogr. A 1250
708
(2012) 63-68.
709
[55] F. Tache, S. Udrescu , F. Albu, F. Micale, A. Medvedovici, Greening pharmaceutical
710
applications of liquid chromatography through using propylene carbonate-ethanol mixtures
711
instead of acetonitrile as organic modifier in the mobile phases, J. Pharm. Biomed. Anal., 75
712
(2013) 230-238.
713
[56] M. Cheregi, F. Albu, S. Udrescu, N. Raducanu, A. Medvedovici, Grenner bioanalytical
714
approach for LC/MS-MS assay of enalapril and enalaprilat in human plasma with total
715
replacement of acetonitrile throughout all analytical stages, J. Chromatogr. B 927 (2013) 124-
716
132.
717
[57] R.A. Granberg, A. C. Rasmussen, Solubility of paracetamol in pure solvents, J. Chem.
718
Eng. Data 44 (1999) 1391-1395.
719
[58] F.L. Mota, A.P. Carneiro, A.J. Queimada, S. P. Pinho, E. A. Macedo, Temperature and
720
solvent effects in the solubility of some pharmaceutical compounds: measurements and
721
modeling, Eur. J. Pharm. Sci. 37 (2009) 499-507.
722
[59] V. Jouyban, M. Khoubnasabjafari, F. Martinez, A. Pena, A. Jouyban, Solubility of drugs
723
in ethyl acetate-ethanol mixtures at various temperatures, J. Drug Del. Sci. Tech. 22 (2012)
724
545-547.
Ac ce
pt
ed
M
an
us
cr
ip t
692
Page 22 of 52
[60] J. Barra, F. Lescure, E. Doelker, P. Bustamante, The expanded Hansen approach to
726
solubility parameters. Paracetamol and cotric acid in individual solvents, J. Pharm.
727
Pharmacol. 49 (1997) 644-651.
728
[61] J. Jover, R. Bosque, J. Sales, Determination of abraham solute parameters from
729
molecular stucture, J. Chem. Inf. Comput. Sci. 44(2004)1098-1106
730
[62] T. Karunasekara, C. F. Poole, Determination of descriptors for flagrance compounds and
731
liquid-liquid partition, J. Chromatogr. A 1235(2012)159-165
732
[63] M. H. Abraham, R. E. Smith, R. Luchtefeld, A.J. Boorem, R. Luo, W.E.Acree Jr.,
733
Prediction of solubility of solubility of drugs and other compounds in organic solvents, J.
734
pharmacy. Sci. 99(2010) 1500-1515
735
[64] M.H. Abraham; R. Kumarsingh; J. E. Cometto-Miniz, W. S. Cain, /. Rosés, E. Bosch, M.
736
L. Diaz, The determination of solvation descriptors for terpenes and the prediction of nasal
737
purgency thresholds, J. Chem. Soc. Perkin Trans.2, (1998)2405-2411
738
[65] N. Zga, Y. Papastamoulis, A. Toribio, T. Richard, J.C. Delaunay, P. Jeandet, J.H.
739
Renault, J.P. Monti, J.M. Mérillon, P. Waffo-Téguo, J. Chromatogr. B 877 (2009) 1000-1004.
740
[66] T. Michel, E. Destandau, G. Le Floch, M.E. Lucchesi, C. Elfakir, Antimicrobial,
741
antioxydant and phytochemical investigations of sea buckthorn (Hippophaë rhamnoides l.)
742
leaf, stem, root and seed, Food Chem. 131 (2012) 754-760.
743
[67] A. Berthod, M. Hassom, M.J. Ruiz-Angel, Alkane effect in the Arizona liquid systems
744
used in counter-current chromatography, Anal. Bioanal. Chem. 383 (2005) 327-340.
745
[68] A. Marston, K. Hostettmann, Developments in the application of counter-current
746
chromatography to plant analysis, J. Chromatogr. A 1112 (2006) 181-194.
747
[69] A. Berthod, T. Maryutina, B. Spivanov, O. Shpigun, I.A. Sutherland, Countercurrent
748
chromatography in analytical chemistry, Pure Appl. Chem, 81 (2009) 355-387.
749
[70] Y. Lu, A. Berthod, R. Hu, W. Ma, Y. Pan, Screening of complex natural extracts by
750
countercurrent chromatography using a parallel protocol, Anal. Chem. 81 (2009) 4048-4059.
751
[71] K. Faure, E. Bouju, P. Suchet, A. Berthod, Use of limonene in countercurrent
752
chromatography: A green alkane substitute, Anal. Chem. 85 (2013) 4644-4650.
753
[72] H. Grensemann, J. Gmelhing, Performance of a conductor-like screening model for real
754
solvents model in comparison to classical group contribution methods, Ind. Eng. Chem. Res.
755
44 (2005) 1610-1624.
756
[73] T. Mu, J. Rarey, J. Gmelhing, Performance of COSMO-RS with sigma profiles from
757
different model chemistries, Ind. Eng. Chem. Res. 46 (2007) 6612-6629.
Ac ce
pt
ed
M
an
us
cr
ip t
725
758
Page 23 of 52
762 763
Figure captions
764 765
Figure 1. Example showing how to place a point in the spider diagram with five axes.
766
Figure 2. Spider diagram based on Snyder values Xn, Xd, Xe (data from table 1). The point
768
size is the Rohrschneider polarity P’.
ip t
767
769
Figure 3. Spider diagram based on solvatochromic parameters (data from table 2).
771
The point size is the Rohrschneider polarity P’.
cr
770
us
772
Figure 4. Spider diagram based on Hansen parameters d, p, h (data from table 3). The
774
point size is the total parameter t.
an
773
775 776
Figure 5. Spider diagram based on Hansen parameters for green solvents (data from table 4).
M
777
Figure 6. Spider diagram based on Hansen parameter showing the solubility (s) of
779
paracetamol in varied solvents at 30°C (indicated in parenthesis). The first solubility value
780
comes from ref. [57], and the second from ref. [58]. The solvents in the blue circle indicate
781
good solubility.
pt
782
ed
778
Figure 7. Spider diagram based on the LSER system coefficients (e,s,a,b,v) measured for gas-
784
to-liquid transfer [data from ref. 25].
785
Ac ce
783
786
Figure 8. Spider diagram based on Abraham descriptors E,S,A,B,V (data from table 5). The
787
point size is the V/U ratio.
788 789
Figure 9. Relationship between P’ (Rohrschneider polarity) and V (McGowan molecular
790
volume) (blue squares and interrupted line) or V/U (U defined in equation (8)) (red diamonds
791
and line) for numerous classical solvents.
792 793
Figure 10. Comparison of Abraham descriptor values for various solvents.
794
a/ yellow =acetone; green = THF; red = Dioxane; Blue = ethyl-acetete; light blue =MTBE
Page 24 of 52
795
b/ yellow = chloroform; green = pyridine
796
c/ green = 1-butanol; yellow = di-methyl-ether; orange = di-ethyl-ether; red = di-propyl-ether
797
d/ yellow = acetic acid; green = ethylene-glycol
798
Figure 11. Spider diagram for the Abraham descriptors and various analytical systems: water
800
– methanol – THF - acetonitrile: reversed-phase liquid chromatography solvents (RPLC);
801
heptane – ethyl acetate – methanol - water: Arizona counter current chromatography (CCC)
802
solvents; water - 2-butanol – ethyl acetate and water – methanol - chloroform: other CCC
803
systems.
cr
ip t
799
804
Figure 12. Abraham descriptor values for solvents used in RPLC and CCC.
806
(a) RPLC; green = water; pink = MeOH; blue = acetonitrile; light blue =
807
(b) Arizona Counter Current Chromatography; yellow = limonene; green = hexane; pink =
808
ethyl-acetate; blue = MeOH; light blue = water
809
Figure 13. Spider diagram based on COSMOments (data from table 6).
M
810
an
us
805
811
Figure 14. Spider diagram for 153 solvents based on principal components coordinates
813
F1,F2,F3 extracted from COSMOments (based on values from ref. [33]).
814
ed
812
Figure 15. Spider diagram for classical solvents based on principal components coordinates
816
F1,F2,F3 extracted from COSMOments (based on values from ref. [33]).
818
Ac ce
817
pt
815
Page 25 of 52
*Highlights (for review)
Highlights
The pider diagram unify the presentation of solvents properties This simple presentation is applied for scales having both 3 or five parameter
This classification is useful for separation, solubility and extraction methods
Ac
ce pt
ed
M
an
us
cr
This classification can be use for replacing toxic solvents by green ones
ip t
Snyder, Kamlet/Taft, Hansen, COSMO-RS and LSER scales are compared
Page 26 of 52
Ac
ce pt
ed
M an
us
cr
i
Figure
Figure 1
Page 27 of 52
i cr us
Xe Basic
M an
Ethanol Ether I
II
IV
ce pt
Trifluoro Water ethanol
Methanol
ed
Acetic acid Ethylene glycol Benzylic alcohol
Nitrobenzene
Ac
VIII
III
Formamide
VIa VIb
Chloroform
Xd Acidic Figure 2
CCl4 V CH2Cl2
VII
THF Di-methyl formamide pyridine Ethyl acetate Dioxane Acetone Acetonitrile Benzonitrile Chloroethane Toluene
Nitromethane
Xn Dipolar Page 28 of 52
i cr
b
Ethylene Dimethyl acetamide basic glycol THF
Ethanol
us
M an
I
Ether
II
Dioxane
Ac VIII
Figure 3
ed
Acetonitrile
Nitrobenzene
Glycerol
Nitromethane V
p* Dipolar
Water Chloroforme
Trifluoro ethanol
VIb VII
Glycerol J. Acetic acid
Acidic
Acetone
VIa
Limonene
ce pt
a
DMSO
III
p-Cymene
Formamide
Dimethyl formamide
IV
Hexane
Methanol
Ethyl acetate
CH2Cl2
Page 29 of 52
i Water
M an
us
hydrogen bond
cr
dh
Methanol
Glycerol Ethylene glycol
Formamide
ce pt
ed
VIII
MTBE
IV
Ethyl acetate VIa
THF
Dioxane
Ac
Dimethyl formamide
II
Propylene glycol Ethanol Acetic acid Acetone Benzylic alcohol
CO Chloroforme Pyridine I III 2 VII CH2Cl2 DMSO Acetonitrile Benzene V dp VIb Toluene CCl4 Polar C60 dd Benzonitrile Nitrobenzene Hexane Nitromethane Dispersive
Figure 4
Page 30 of 52
i cr
dh
M an
us
hydrogen bond
CO2/MeOH 90/10
ed
Ethyl lactate
Ac
ce pt
glycerol trimethylether Polym. Linseed oil
Ethanol Solketal iso-propanol n-Butanol
dp
Dimethyl Glycerol isosorbide carbonate CO2
Polar Propylene carbonate
Figure 5
Glycerol a-terpineol triacetate Cholesterol TG Tocotrienol
FA
Hexane
Glycerol limonene triet/trilutyl p-cymene ether CH2Cl2 a-pinene
Pine resin Tocopherols Sterols
dd Dispersive Page 31 of 52
cr
i Water(17/32)
us
dh
M an
hydrogen bond
Methanol(371)
Acetic acid (83)
ce pt
ed
Ethylene glycol
Dimethyl formamide(1012)
Ac
Paracétamol
dp
DMSO(1132) Acetonitrile(32)
Polar
Figure 6
Ethanol(232/242) Butanol(132)/propanol(93) Acetone(111/140) THF(155)
Ethyl acetate(10/32) Dioxane(17) CH2Cl2(0,3) Toluene(0,3)
CCl4(0,9/22) d d Dispersive
Page 32 of 52
i cr us
e hexane heptane CCl4 -1,500
cyclohexane
benzene
toluene
M an
s
l
ed
CH2Cl2
b
1-Octanol
1-butanol
acetonitrile 1,4 dioxane acetone
THF
di-ethylether
Ac
Propylene carbonate
ethyl acetate MTBE
ce pt
chloroform
2-propanol ethanol methanol
2-butanol 1-propanol
diméthyl formamide
DMSO formamide
ethylene glycol
trifluoroethanol
water
a
Figure 7
-6,000
Page 33 of 52
i cr
E
us
nitrobenzene
tert-butylbenzene
a terpineol
benzene
toluene CCl4 cyclohexane
ed
limonene
M an
p-cymene a-pinene
heptane hexane
Pyridine chloroform CH2Cl2
Phenyl ethanol
THF
ce pt
V
Propylene carbonate
1-octanol
ethyl acetate
nitromethane acetonitrile
S
diméthyl formamide
propionitrile acetone
1-butanol
-1,500
di-propylether MTBE
Paracetamol
1,4 dioxane
di-methylether
Ac
benzonitrile
anisole p-xylene
ethanol
1-propanol 2-butanol 2-propanol methanol
di-ethylether
Glycerol
formamide
ethylene glycol acetic acid trifluoroethanol
1,3 propane diol
B
water
Figure 8 -1,000
A
Page 34 of 52
i cr us M an
P’
12
y = -9,7611x + 10,112 R² = 0,836
10
ed
8
4 2
Ac
0
ce pt
6
-2
Figure 9
y = -7,8912x + 9,5509 R² = 0,5794
0,0
0,2
0,4
V/U
0,6
0,8
1,0
1,2
1,4
Page 35 of 52
i cr c
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
S
A
b
E
Figure 10
S
A
B
E
V
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
Ac
1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
B
ce pt
E
ed
M an
us
a
1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0
V
S
A
B
V
d
E
S
A
B
V Page 36 of 52
i cr
E
us
Di-ethylphthalate
nitrobenzene
p-cymene
toluene p-xylene
a-pinene
limonene
M an
tert-butylbenzene
benzonitrile
anisole
benzene
CCl4
Pyridine
chloroform CH2Cl2
ed
cyclohexane
V
Alcool Phenyl ethanol benzylique hexane 1,4 dioxane
a terpineol
ce pt
THF
heptane
acetone
diméthyl formamide
S
nitromethane
Acetonitrile propionitrile
di-methylether
ethyl acetate
formamide
di-ethylether
Ac
-1,500
di-propylether
MTBE 2-butanol 2-propanol
ethanol
methanol
water
B Figure 11
acetic acid
-1,000
A
Page 37 of 52
cr
i S
A
B
Ac
ce pt
E
ed
M an
1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0
us
a
V
1,4 1,2 1,0 0,8 0,6 0,4 0,2 0,0
b
E
Figure 12
S
A
B
V Page 38 of 52
i cr
Sig2
Glycerol
us
DMSO
CSA 1-Octanol
Phenyl ethanol
a-terpineol
M an
Propylene carbonate
Benzylic alcohol benzonitrile
Sig3
ethylene glycol
1,4 dioxane
diméthyl formamide
ed
formamide 1-butanol nitrobenzene acetic acid ethyl acetate 1-propanol Pyridine anisole 2-propanol 2-but-ol acetone di-ethylether ethanol MTBE p-xylene acetonitrile
ce pt
-1,000
Ac
toluene di-propylether THF nitromethane benzene
heptane hexane
Hb acc3
CH 2Cl2 chloroform
water
methanol
di-methylether
Hb don3
CCl4 cyclohexane
Figure 13
-0,700 Page 39 of 52
i cr
IX. Organic acidic
M an
us
F3
X. Polar structured
ce pt
ed
I. Strong electron-pair donor bases
Ac
VI. Asymetric halogenated hydrocarbon
VII. Amphiprotic VIII. Polar protic
II. Weak electron-pair donor bases
F2
Electron pair donor V. non polar
Figure 14
III. aprotic dipolar IV. aprotic highly dipolar
F1 Page 40 of 52
Hydrogen donor
i
cr
F3
Methanol
Acetic acid
Ethylene glycol
1-Octanol 1-propanol 1-Butanol
ce pt
Propylene glycol
Ethanol
ed
Water
M an
us
Trifluoro ethanol
2-propanol 2-butanol Glycerol Formamide N-methyl CH2Cl2 formamide Cyclohexane Hexane Benzonitrile Pyridine CCl4 DiEt Dipr Heptane MTBE ether Nitromethane Ether THF benzene Propionitrile Etac Acetonitrile Toluene Nitrobenzene 1,4-Dioxane Acetone Propylene DMSO carbonate F2
Ac
Chloroform
Figure 15
F1 Page 41 of 52
ed
octanol n-butanol Isopropanol n-propanol ethanol Methanol
Xd 0,13 0,14 0,1 0 0,17 0,18 0,17 0,19 0,19 0,22
Xe 0,53 0,48 0,51 0 0,58 0,54 0,57 0,54 0,52 0,48
us
Xn 0,34 0,38 0,39 0,2 0,25 0,28 0,26 0,27 0,29 0,31
M an
Solvents diethyl ether dibuthylether di-isopropyl ether
cr
i
Tables
0,41 0,36 0,4 0,41
0,19 0,22 0,21 0,2
0,4 0,42 0,39 0,38
benzyl alcohol ethylene glycol acetic acid
0,31 0,28 0,3
0,29 0,29 0,3
0,4 0,43 0,41
CH2Cl2
0,4
0,33
0,27
CCl4 chloroform dichloethane
0,34 0,34 0,49
0,4 0,35 0,21
0,26 0,31 0,3
ethyl acetate dioxane acetone
0,42 0,4 0,42
0,22 0,23 0,23
0,36 0,37 0,35
Ac
ce pt
THF pyridine dimethylformamide N,N dimetacetamide
Page 42 of 52
i 0,43 0,45 0,45 0,43 0,42 0,4
ed
ce pt
formamide trifluoroethanol water
cr
nitrobenzene toluene benzene anisole benzylic ether nitromethane
0,3 0,27 0,25
Xd 0,26 0,25
Xe 0,32 0,33
0,29 0,27 0,28 0,29 0,28 0,31
0,29 0,28 0,27 0,27 0,3 0,28
0,33 0,33 0,37
0,36 0,4 0,37
us
Xn 0,41 0,42
M an
Solvents benzonitrile acetonitrile
Ac
Table 1. Data of solvent strenght selectivity (Snyder). Xn : dipole; Xd : acidity; Xe : basicity; P’ : Rochschneider polarity
Page 43 of 52
0,47 0,48
0,79 0,76
0,88 0,95
0,54 0,6
0,83 0,93
0,77 0,62
pyridine dimethylformamide THF DMSO
0,87 0,88 0,58 1
0 0 0 0
0,64 0,69 0,55 0,74
CH2Cl2 chloroethane chloroforme CCl4
0,82 0,81 0,58 0,28
0,2 0 0,44 0,00
0 0 0 0,00
ethyl acetate dioxane acetone
0,55 0,55 0,71
0 0 0,08
0,45 0,37 0,48
N,N dimetacetamide ethylene glycol
0,49 0,92
0 0,52
0,57 0,9
benzonitrile acetonitrile nitrobenzene
0,9 0,75 1,01
0 0,19 0
0,41 0,31 0,39
ed
Ac
a 0 0 0
i
cr
n-butanol isopropanol n-propanol ethanol methanol
M an
p* 0,27 0,24 0,27
us
b 0,47 0,46 0,49
ce pt
Solvents diethyl ether dibuthylether di-isopropyl ether
Page 44 of 52
i
b 0,11 0,1 0,22 0,41 0,25 0 0,13
1,51 1,17 1,12 0,71 0,62 0,94
0 0,47 0,45 0,44 0,51 0,51
us
cr
0,73 1,09 0,64 0,97 1,21 1
a 0 0 0 0 0,12 0 0
ce pt
ed
trifluoroethanol Water acetic acid formamide glycerol glycerol Jessop[40]
p* 0,55 0,59 0,73 0,8 0,85 0,16 0,39
M an
Solvents toluene benzene anisole benzylic ether nitromethane limonene[39]) p cymene[39]
Ac
Table 2. Data of solvatochromic solvent selectivity. p* : polarity-polarisability ;a : hydrogen-bond acidity; ;b: hydrogen-bond basicity. The Rochschneider polarity values are in table 1. For Limonene, p-cymene, hexane, and glycerol this value was arbitrary set at 5 for plotting the point.
Page 45 of 52
i dd
dp
14,5 17,4
2,9 3,7
n-butanol isobutanol n-propanol ethanol methanol MTBE THF pyridine dimethylformamide
16 15,1 16 15,8 15,1 14,8 16,8 19 17,4
cr
Solvents diethyl ether dibenzylether
dh
5,7 5,7 6,8 8,8 12,3 4,3 5,7 8,8 13,7
15,8 16 17,4 19,4 22,3 5 8 5,9 11,3
23,20 22,73 24,60 26,52 29,61 16,20 19,46 21,75 24,86
18,4 17 14,5 16,8
6,3 11 8 9,4
13,7 26 13,5 23,3
23,79 32,95 21,37 30,22
dichlorométhane CCl4 chloroforme dichloethane
18,2 17,8 17,8 16,6
6,3 0 3,1 8,2
6,1 0,6 5,7 0,4
20,20 17,81 18,95 18,52
ethyl acetate dioxane acetone
15,8 19 15,5
5,3 1,8 5,3
7,2 7,4 11,7
18,15 20,47 20,13
benzonitrile ACN
17,4 15,3
9 18
3,3 6,1
19,87 24,40
Ac
M an
ed
ce pt
benzylic alcohool ethylene glycol acetic acid propylene glycol
us
5,1 7,4
dt 15,64 19,27
Page 46 of 52
dd
dp
cr
i Solvents nitrobenzene toluene benzene anisole nitromethane hexane
20 18 18,4 17,8 15,8 15
8,6 1,4 0 4,1 18,8 0
4,1 2 2 6,8 5,1 0
dt 22,15 18,16 18,51 19,49 25,08 15,00
formamide water fullerene C60[41]
17,2 15,6 19,7
26,2 16 2,9
19 42,3 2,7
36,65 47,84 20,09
us
M an
Ac
ce pt
ed
dh
Table 3. Data of Hansen parameters for common solvents. dd: dispersion parameter; dp :polar parameter; dh:hydrogen bond parameter; dt: total parameter
Page 47 of 52
10,6
cr
i ce pt
Ac
CO2/MeOH 90/10 [47]
dh
1,8 1,3 2,3 4,7 0,9 1,8 0 3,1 1,6 6 16,4 6,3 3,8 18 7,6 25,5 7,1 11,3 7,2 5,7 4,5 4,8 3,8
4,3 2,2 2,4 2,2 2,8 3,2 3,5 5,7 4,5 7 10,2 5,7 10 4,1 12,5 17,4 7,5 27,2 19,3 7,5 9,1 3,6 3,2
5,3
12,4
M an
17,2 16,4 17,3 15,8 16,6 17,2 17,5 16,5 18,7 16 18,4 15,7 17 20 16 17,9 17,6 17,4 16 15,5 16,5 15,5 15,7
dp
us
dd
ed
Solvents limonene a-pinene p-cymene TG[43] tocopherol[43] sterols[43] a-tocotrienol[43] fatty Acid (FA)[43] pine resin [44] polym. Lindseed oil[44] DMSO [45] carbon dioxyde [45] a-terpineol[46] propylene carbonate[46] ethyl lactate[46] glycerol carbonate[46] dimethyl isosorbide[46] glycerol[45] solketal[46] glyceroyl trimethylether[46] glycerol triacetate[46] glycerol triethylether[46] glycerol tributether[46]
a
dt 17,82 16,60 17,62 16,63 16,86 17,59 17,85 17,73 19,30 18,47 26,68 17,85 20,09 27,22 21,68 35,69 20,41 34,21 26,08 18,14 19,37 16,62 16,47 11,85
Table 4. Data of Hansen parameters for « Green » solvents. a : calculated from data in ref. [47]
Page 48 of 52
benzene toluene tert-buthyl-benzene p-xylene
0,61 0,601 0,619 0,613
0,52 0,52 0,49 0,52
0 0 0 0
methanol ethanol 1-propanol 1-butanol 2-butanol 2-propanol
0,278 0,246 0,236 0,224 0,217 0,212
0,44 0,42 0,42 0,42 0,36 0,36
ed
ce pt
Ethylene glycol
B 0 0 0
V 0,954 1,095 0,845
U 0,954 1,095 0,904
0,14 0,14 0,18 0,16
0,716 0,857 1,28 0,998
1,084 1,177 1,515 1,291
0,43 0,37 0,37 0,37 0,33 0,33
0,47 0,48 0,48 0,48 0,56 0,56
0,308 0,449 0,59 0,731 0,731 0,59
0,878 0,898 0,973 1,062 1,065 0,972
i
A 0 0 0
cr
S 0 0 0,1
us
E 0 0 0,305
M an
Solvents hexane heptane cyclohexane
0,46
0,76
0,6
0,69
0,507
1,372
0,237 0,313 0,162 0,313 0,319 1,12
0,9 0,95 0,9 1,31 1,037 1,66
0,07 0,06 0,02 0 0 0,91
0,32 0,31 0,36 0,74 0,6 0,93
0,404 0,424 0,545 0,647 0,697 1,17
1,066 1,131 1,124 1,667 1,422 2,659
acetone ethyl acetate THF 1,4-dioxane
0,179 0,106 0,289 0,329
0,7 0,62 0,52 0,75
0,04 0 0 0
0,49 0,45 0,48 0,64
0,547 0,747 0,622 0,681
1,031 1,075 0,985 1,243
CH2Cl2
0,387
0,57
0,1
0,05
0,494
0,855
CHCl3
0,425
0,49
0,15
0,02
0,617
0,908
CCl4
0,458
0,38
0
0
0,739
0,949
Ac
ACN nitromethane propionitrile dimethyl-formamide propylene-carbonate paracetamol
Page 49 of 52
i 0,784
0,83
benzo-nitrile nitrobenzene
0,742 0,871
1,11 1,11
anisole pyridine
0,708 0,631
0,75 0,84
water acetic acid formamide
0 0,265 0,468
cr
phenyl-ethanol
B 0,45 0,58 0,41 0,59
V 0,731 1,01 0,45 0,872
U 0,895 1,176 0,666 1,074
0,3
0,66
1,057
1,717
0 0
0,33 0,28
0,871 0,891
1,628 1,692
0 0
0,29 0,52
0,916 0,675
1,41 1,353
0,45 0,65 1,3
0,82 0,61 0,62
0,35 0,44 0,6
0,167 0,465 0,265
1,013 1,129 1,65
0,488 0,446 0,553 0,607
0,28 0,14 0,61 0,49
0 0 0,2 0
0,45 0,12 0,7 0,19
1,323 1,257 1,42 1,28
1,506 1,346 1,796 1,511
glycerol 1,3 propane diol trifluoroethanol
0,512 0,397 0,015
0,76 0,89 0,6
0,47 0,77 0,57
1,43 0,87 0,25
0,707 0,649 0,502
1,899 1,649 1
diethyl-phthalate
0,729
1,4
0
0,88
1,711
2,489
ce pt
Ac
limonene a pinene a terpineol p-cymene
A 0 0 0 0
us
S 0,25 0,16 0,27 0,21
M an
E 0,041 0,006 0 0,024
ed
Solvents Diethyl-ether Propyl-ether Dimethylether MTBE
Table 5. Data for Abraham descriptors (from ref. 20,21, 61 -64) Page 50 of 52
cr
i Ac
us
Hb don3 0 0 0 0 0 2,14 1,99 1,69 1,98 1,411 1,4 0 0 1,211 0 0 0 4,482 2,014 0 0 0 0 1,75 0 3,83 0 3,85 0 3,62 0
M an
Sig3 0,43 0,69 0,399 -0,53 1,51 20,28 23,48 21,95 22,83 19,726 25,68 17,76 36,052 28,24 -15,81 30,64 38,25 51,318 17,66 6,164 -21,19 8,58 29,65 10,34 11,075 -3,29 -2,95 13,14 58,94 16,09 2,13
ed
sig2 7,92 8,92 5,71 28,26 28,58 53,58 53,53 52,02 57,13 48,4627 51,42 50,057 47,91 45,04 32,85 31,43 35,789 59,114 73,8 52,471 26,79 44,13 46,833 66,66 51,248 71,99 9,91 75,299 61,74 88,38 55,794
ce pt
Solvents hexane heptane cyclohexane benzene toluene methanol ethanol 1-propanol 1-butanol 2-butanol 2-propanol ACN acetone ethyl acetate CH2CL2 diethyl ether THF dioxane phenyl ethanol nitrobenzene chloroforme anisole pyridine benzylic alcohol benzo nitrile acetic acid CCl4 water dimethyl formamide formamide nitromethane
HB acc3 0 0 0 0 0 4,22 4,47 3,92 4,27 3,5425 4,178 1,346 2,793 0 0 2,63 3,509 0 3,58 0,241 0 0,425 3,39 2,84 0,858 1,97 0 5,75 5,99 5,82 0,219
CSA 156,9 167,8 126,0 118,7 135,8 66,4 85,9 104,2 128,4 125,8 106,4 82,6 102,7 129,7 98,5 130,3 111,8 120,5 161,5 149,4 117,5 150,4 117,0 150,0 145,7 92,3 134,2 43,1 115,9 76,7 88,4
Page 51 of 52
i cr us Hb don3 2,837 0 0 1,752 9,74 0 1,09 0 2,591 1,987
M an
Sig3 28,07 31,52 30,25 10,341 83,528 33,35 20,599 26,87 38,481 22,83
ed
sig2 82,61 33,48 31,48 66,667 85,92 34,24 50,55 72,86 100,289 57,13
HB acc3 5,514 3 2,64 2,839 0,03 3,188 2,61 1,707 6,402 4,2717
CSA 97,6 163,7 89,9 150,0 111,7 139,6 187,4 124,2 121,6 197,0
Ac
ce pt
Solvents ethylene glycol propyl ether dimethylether benzy alcohol DMSO MTBE a terpineol propylene carbonate glycerol 1-octanol
Table 6. Data for COSMOment s(from ref. 33-35) Page 52 of 52