Accepted Manuscript Title: Impact of biomolecule solute size on the transport and performance characteristics of analytical porous polymer monoliths Author: Ivo Nischang PII: DOI: Reference:
S0021-9673(14)00819-X http://dx.doi.org/doi:10.1016/j.chroma.2014.05.053 CHROMA 355445
To appear in:
Journal of Chromatography A
Received date: Revised date: Accepted date:
3-3-2014 25-4-2014 20-5-2014
Please cite this article as: I. Nischang, Impact of biomolecule solute size on the transport and performance characteristics of analytical porous polymer monoliths, Journal of Chromatography A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.053 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
3
Ivo Nischang*
4
Institute of Polymer Chemistry, Johannes Kepler University Linz, Welser Strasse 42, A-4060
5
Leonding, Austria
6
*Corresponding author:
7
E-mail:
[email protected], phone: +43 732 671547-66, fax: +43 732 671547-62
8
---------------------------------------------------------------------------------------------------------------------------
9
Abstract
us
cr
ip t
2
Impact of biomolecule solute size on the transport and performance characteristics of analytical porous polymer monoliths
Porous monolithic poly(styrene-co-divinylbenzene) stationary phases in 4.6 mm ID
11
analytical format have been investigated with respect to their transport properties probed by
12
solutes of biological origin varying vastly in size. Elucidation of several properties of these
13
benchmark and robust materials gave complementary insight. These are: (i) the porous
14
polymers’ apparent dry-state microscopic appearance, (ii) the columns porosity probed by
15
the biomolecules and modulated by mobile phase solvent composition, (iii) the impact of
16
probe solute size on apparent retention at varying mobile phase solvent compositions, and
17
(iv) the elution performance under both nonretained and retained elution conditions. By
18
varying the volume percentage of acetonitrile in the mobile phase, it is demonstrated that the
19
monolithic scaffold shows a variable porosity experienced in particular by the larger sized
20
solutes, while the smaller solutes are gradually less affected. The nanoscale swelling and
21
solvation of porous monolithic adsorbents resulting in gel porosity varied with mobile phase
22
solvent composition was, therefore, indicated. The plate height curves for the solutes under
23
nonretained conditions show a moderate increase at increased flow velocity while
24
approaching plateau values. These plateau values were in conjunction with a trend of a
25
decreased performance at an increased molecular weight of the solute. The systematic
26
shape of the plate height curves at increased flow velocity indicates pre-asymptotic
27
dispersion. This is because the column bed aspect ratio of length-to-diameter is equal or
28
smaller than 10. Imposing retention on the solutes at a constant flow velocity deteriorates
29
isocratic elution performance, more pronouncedly for the larger sized solutes at even weak
30
retention. This is explained with slow pore fluid-gel interface diffusion. Additionally, the
31
apparent retention factor for elution of the probe solutes becomes a function of flow rate,
32
consequently a function of imposed pressure experienced by the scaffold.
Ac ce p
te
d
M
an
10
33 1
Page 1 of 32
34
Keywords: adsorption; efficiency; gel porosity; mass transfer; partition; pore-fluid gel
35
interface; retention; size exclusion
36 37 38
1. Introduction
39
micrometer-sized flow-through pores have found significant attention in recent years.
40
Typically, the scaffolds are prepared via free radical cross-linking (co)polymerization of small
41
organic precursors in porogenic diluents. Such entities enable excellent separation of large
42
molecules in gradient elution mode [1-8]. This application for large molecules has been
43
extended to that of smaller proteins, peptides, and amino acids with varying success in the
44
monoliths’ application [9-12]. Another recent focus of the monolithic column technology can
45
be found in the exploitation of a variety of chromatographic modes such as hydrophobic
46
interaction
47
chromatography
[18-20],
48
chromatography
[22]. Recent efforts were also directed toward scaling capabilities,
49
microfluidic applications, and novel types of functionalization to explore a diversity of
50
implementations [4, 6, 23, 24].
chromatography
[13-15],
ion-exchange
chromatography
chromatography
[21],
and
[16,17],
affinity
hydrodynamic
M
an
size-exclusion
us
cr
ip t
Porous polymeric monoliths based on a cross-linked material intertwined by
More detailed studies on the chromatographic properties of porous polymeric
52
monoliths, however, indicated that the properties measured in the dry-state of the materials
53
allow only limited interpretation regarding their nanoscale dynamics and associated transport
54
performance [25-28]. These materials, no matter how heavily cross-linked, develop gel
55
porosity on a nanometer-scale, a porosity that is absent in the dry-state of the cross-linked
56
polymeric materials and which stems from nanoscale solvation and swelling of the polymer
57
scaffold features [27]. Such phenomena have not yet been investigated for large molecule
58
transport, since gradient elution conditions leave limited room for interpretational efforts of
59
mass transfer processes [28, 29]. In general, the poor performance of polymeric monoliths in
60
isocratic (equilibrium) elution mode is frequently reported. The poor performance under
61
isocratic elution conditions could partly be associated with the flow-through pore
62
heterogeneity [28, 29, 30-33] resulting in convective flow dispersion, an aspect that has
63
received significant attention in recent years. Experimental elucidations in this direction via
64
reconstruction [31-33], and modelling attempts [31, 32], have already been reported. The
65
efficiency analysis in the elution of small (retained) solutes also reveals the strong impact of
66
nanoscale gel porosity [25-29]. This aspect is detached, but interplays, with the
67
morphological aspects of polymer monoliths [28]. In fact, polymer monoliths show
68
heterogeneities on the nano-, micro-, and potentially confine-associated length scales. These
69
heterogeneities have their origin in the very early beginnings of the uncontrolled, though
Ac ce p
te
d
51
2
Page 2 of 32
70
adjustable, free-radical polymerization processes and the inherently associated phenomena
71
of scaffold formation [25, 27]. The separation of small molecules, peptides, and (small) proteins is well-served by
73
silica-based monoliths [34, 35], packed beds of porous adsorbent particles, and the recent
74
revival of core-shell particles [36, 37] with a noticeable extension to the analysis of large
75
molecules [38]. Theoretical description of transport and performance characteristics through
76
such types of porous media has advanced considerably in recent years and this has allowed
77
correlation to actual elution experiments with small solutes [39, 40].
ip t
72
Using such small solutes with typical polymer monoliths enters a regime of
79
chromatography that is ruled by convective flow dispersion, as well as diffusion and
80
partitioning into the solvated polymer gel matrix. This is irrespective of what happens to the
81
solutes once they are adsorbed, or whether mesopores are present in the dry-state or not
82
[28]. Such small solutes have sizes that correspond to the molecular framework openings of
83
the heterogeneously cross-linked and solvated polymeric material found in porous monoliths
84
[25-29]. From a fundamental point of view, this marks the major difference to hard matter
85
silica-based (meso)porous materials [29]. In turn, polymer monoliths are analogous to their
86
earlier generation bead-based polymeric counterparts [29, 41-46]. A practical and easy
87
indication for the discussed issues on column performance can be found by estimating the
88
resistance to mass transfer contribution (classically C-Term contribution) to the plate height
89
of small molecules. Once systematically probed, isocratic performance is impacted by the
90
solute properties such as size, functional group content, as well as retention [47]. The C-
91
Term contribution to band broadening can dominate overall achievable performance of these
92
materials even at very low flow velocities [27].
us
an
M
d
te
Ac ce p
93
cr
78
We have introduced and explored the concept of gel porosity as an important tool for
94
indirectly studying the soft matter structure of macroscopically rigid polymer monoliths [25]. It
95
has helped in explaining the performance of porous polymeric monoliths for small molecules
96
including the impact of mobile phase composition. The mobile phase composition modulates
97
both polymer nanoscale gel structure and associated retention-dependent performance in
98
the elution of small retained solutes for both methacrylate and styrene/divinylbenzene based
99
chemistries [25, 26, 47]. In both of these cases, a specific mobile phase solvent composition
100
translates to a variably solvated polymer scaffold. Direct demonstration of this aspect has
101
been possible by confocal Raman spectroscopy [48]. While water was seen almost
102
exclusively in the large, micrometer-sized flow-through pores of a hydrophobic porous
103
polymeric poly(styrene-co-divinyl benzene) material, acetonitrile was present even within
104
individual globular features. The resultant distribution equilibrium of mobile phase
105
components (associated with a dynamic gel porosity) likely then also determines variable 3
Page 3 of 32
106
permeation of the cross-linked polymer gel by small-sized solutes based on partition and
107
adsorption. Varying polymer monoliths feature size, and cross-link density then are aspects
108
that have to be considered for a straightforward performance analysis. Small molecules
109
permeate the cross-linked polymer to a varying degree under operation in liquid
110
chromatography [28, 47]. Porous polymer monoliths show their optimum performance for gradient elution of
112
large molecules. The impact of gel porosity on the metric description of transport
113
performance for larger solutes such as proteins should yet require significant attention [28,
114
29]. Related studies in this area mostly utilize the nonretained elution regime in the absence
115
of adsorption at the pore-fluid gel interfaces and indicate a rapid loss of efficiency at even low
116
retention [49, 50]. The recently introduced modeling approaches for flow [31] and dispersion
117
[32] in commercial monolithic disc materials further found disagreement of both shape of
118
plate height curves, and magnitude of large molecule transport performance under
119
nonretained conditions. Theoretically predicted values of the (reduced) minimum plate height
120
for a limited model domain of a commercial CIMTM disc material were smaller than 10 µm and
121
were increasing with flow velocities. This shape of the derived plate height curve was in
122
contrast to experimental results [32]. A knowledge of the theoretically achievable
123
performance should spur efforts and further interest in improving the polymer monoliths’
124
chromatographic performance.
d
M
an
us
cr
ip t
111
It is purpose of the current experimental study to have a closer look at the
126
characteristics of porous polymeric monoliths for elution of biomolecular solutes under
127
isocratic conditions. The experiments involve a set of small peptides and proteins ranging in
128
molecular weight from 181.19 to 44287 Da. This study is complemented by the use of a
129
suitable low molecular weight mobile phase velocity tracer as a reference.
Ac ce p
130
te
125
131
2. Experimental
132
2.1. Chemicals and materials
133
Solutes uracil (112.09 g/mol), tyrosine (181.19 g/mol), val-tyr-val (379.45 g/mol), [D-
134
Ala2]-leucine enkephalin (569.65 g/mol), [Asn1Va5]-angiotensin II (1031.17 g/mol), insulin
135
(from bovine pancreas) (5733 g/mol), cytochrome c (from equine heart) (12384 g/mol),
136
myoglobin (from equine skeletal muscle) (17699 g/mol), albumin (from chicken egg white)
137
(44287 g/mol), and the mobile phase additive formic acid were acquired from Sigma–Aldrich
138
(Vienna, Austria). LC-MS grade acetonitrile was received from VWR (Vienna, Austria). Water
139
was purified on a Milli-Q Reference water purification system from Millipore (Vienna, Austria).
140
Sample solutions were prepared in running mobile phases containing various volume
141
percentages of acetonitrile/water (v/v) and 0.1% (v/v) formic acid. 4
Page 4 of 32
142 143
2.2. Equipment, column, and chromatographic measurements A 1290 Infinity UPLC system (Agilent Technologies, Vienna, Austria) was used for all
145
chromatographic experiments. The overall measured dead volume from injection to detection
146
with the column replaced by a zero dead-volume connection was 25.7 µL. The data
147
acquisition rate was set to 160 Hz to allow sufficient data sampling for the very short
148
residence times and small peak widths. This is particularly important with rapid elution times
149
that can have serious effects on estimated performance [51].
ip t
144
The commercially available ThermoScientific (TM) ProSwift (TM) RP-1S and RP-3U
151
monoliths with a poly (styrene-co-divinylbenzene) cross-linked macroporous monolithic
152
polymer were a kind gift from Thermo Fisher Scientific (Sunnyvale, California) as research
153
samples. The total extra-column volume was calculated to be ≤ 4 % of the overall column
154
volume according to the provided bed dimension of 4.6 x 46 mm (RP-1S) or 4.6 x 40 mm
155
(RP-3U), respectively. The (macroscopic) bed aspect ratios of length-to-diameter were 10
156
(RP-1S) or 8.7 (RP-3U), respectively. The measured extra-column volume has been
157
considered for the calculation of the apparent column porosities from elution studies with the
158
set of solutes. The porosity was calculated as the ratio of corrected elution volume and
159
empty tube volume. Injections of 1 µL sample of all solutes with a concentration of a typically
160
100 µg/mL dissolved in the mobile phase were performed. Chromatographic experiments
161
were carried out at a controlled column oven temperature of 25 °C.
d
M
an
us
cr
150
After conditioning the columns with acetonitrile with at least thirty column volumes,
163
the column was equilibrated with the desired binary mobile phase solvent composition. The
164
binary mobile phase solvent composition comprised varying percentages of acetonitrile and
165
water (v/v), both containing 0.1 % (v/v) formic acid as a typical additive.
167 168
Ac ce p
166
te
162
Scanning electron micrographs were obtained using a Crossbeam 1540 XB electron
microscope (Carl Zeiss SMT AG, Oberkochen, Germany).
169
3. Results
170
3.1. Microscopic structure and solvated-state porosity
171
Figure 1 shows example scanning electron microscopy (SEM) images of the
172
microscopic dry-state properties of the monoliths used in this work. It demonstrates the
173
typical structure of porous polymer monoliths. While the RP-3U column (Figure 1a) shows
174
large globule agglomerates of irregular size and large intertwining pores, the RP1S column
175
(Figure 1b) shows a much finer porous structure. Indeed, these results show the much
176
smaller domain size of the RP-1S monolith. 5
Page 5 of 32
Knowing about the difference in structural properties between the dry- and solvated-
178
state [28], the apparent elution porosity probed by the solutes used in the present study was
179
investigated at different volume percentages of acetonitrile in the mobile phase. The
180
experiments were performed at a superficial flow velocity of 0.5 mm/s, i.e. a constant flow
181
rate of 0.5 mL/min for both columns. It can be seen in Figure 2 that the apparent elution
182
porosity for the RP-1S column varies widely. At a mobile phase composition of 50/50 (v/v)
183
acetonitrile/water, uracil experiences a porosity of 58 % and ovalbumin experiences only 51
184
%. This can be explained by the existence of an apparent micro- and mesoporosity under
185
such mobile phase solvent composition for the RP-1S column. The data show that the large
186
ovalbumin has limited accessibility to the porous structure of the column in the presence of
187
50/50 (v/v) acetonitrile/water. More interestingly, increasing volume percentage of acetonitrile
188
in the mobile phase leads to a more pronounced difference in elution porosity for the probe
189
solutes, in particular observed for the larger sized solutes (Figure 2). This may partly be
190
explained also by reduction of any residual interactions with the scaffold.
an
us
cr
ip t
177
These results, however, may confirm the apparent swelling propensity imposed by
192
the use of porous polymer monoliths with an increased acetonitrile concentration. It indicates
193
closing of the existing pores at a volume fraction of greater than 50% (v/v) acetonitrile in the
194
mobile phase. The permeation volume of the smaller sized solutes is only influenced less
195
significantly. Recent studies on other types of polymer monoliths with atomic force
196
microscopy in the dry- and solvated-state came to the conclusion that initially existing
197
mesopores were not observed in the solvated-state [52]. This situation is not surprising since
198
nanoscale polymer solvation and swelling enhances the free volume of the polymer, i.e.
199
generates gel porosity [29]. This decreases the average pore size of the column, a quantity
200
measured by elution experiments such as in Figure 2. In comparison to the RP-1S column,
201
the RP-3U column (filled grey squares in Figure 2) shows similar, but systematically smaller,
202
apparent porosities for all solutes at a mobile phase composition of 50/50 (v/v)
203
acetonitrile/water.
205 206
d
te
Ac ce p
204
M
191
3.2. Elution performance under nonretained conditions To highlight the impact of solute molecular weight (consequently size) on elution
207
performance, Figure 3 shows plate height values at two selected flow velocities against
208
molecular weight of the probe solutes for the RP-1S and the RP-3U column. It can be
209
concluded that at superficial flow velocities of 1 and 2 mm/s, elution performance
210
experiences a decrease for the probe solutes larger than 1000 to 6000 Da. This range of
211
molecular weights indicates only partial sampling of the porous structure (Figure 2) and, at a
212
first approximation the solutes experience the hydrodynamic flow dispersion in the 6
Page 6 of 32
heterogeneously structured flow-through pore space. The relatively low diffusion coefficients
214
of solutes cannot compensate for velocity biases across the column cross-section or, more
215
precisely, the amount of cross-section that becomes effectively sampled by lateral diffusion
216
(vide infra). This is most pronounced for the solutes having largest molecular weight and
217
lowest diffusivity, apparently increasing the height equivalent to a theoretical plate.
218
Comparison to the RP-3U column reveals its significantly lower performance than that of the
219
RP-1S column over the same superficial flow-velocity range for all eluted solutes,
220
irrespective of their molecular weight. For both columns, the curves slightly shift upwards at
221
larger superficial flow velocities, more so for the larger domain size RP-3U column.
cr
ip t
213
Figure 4 shows plate height curves for two selected solutes, i.e. the uracil (Figure 4a)
223
and myoglobin (Figure 4b), with a mobile phase solvent composition of 50/50 (v/v)
224
acetonitrile and water. These experiments demonstrate the substantially better efficiency of
225
the RP-1S column in the elution of both solutes for the whole range of superficial flow
226
velocities. The much finer porous structure of the RP-1S column (Figure 1b), as compared to
227
the RP-3U column (Figure 1a) must be accompanied with some structural homogeneity in
228
the solvated state. Therefore, it provides a better efficiency than that of the apparently more
229
heterogeneous porous structure of the RP-3U column. Additionally, both plate height curves
230
show a relatively weak dependence on superficial flow velocity and appear to taper off at
231
increased velocities, more so for the RP-1S column. This confirms recent conclusions that
232
polymer monoliths do not, or only insignificantly lose performance at increased flow velocities
233
and is discussed later in this study (vide infra). Indeed, we may argue that there is only a
234
small resistance to mass transfer even for the RP-1S column (that shows significant
235
population of small pores) under nonretained elution conditions (Figure 2).
an
M
d
te
Ac ce p
236
us
222
For closer observation, Figure 5 shows plate height curves of all probe solutes for
237
superficial flow velocities of up to 2.5 mm/s for the RP-1S column. The lower molecular
238
weight solutes are shown in Figure 5a and the larger solutes are shown in Figures 5b and 5c.
239
While some variations in elution performance are observed for all solutes, common features
240
of the plate height curves are apparent. In general, the performance at relatively low
241
superficial flow velocity ranges from 30 to 50 µm, which is in good agreement of what we
242
observed recently under nonretained elution conditions using this column format [28, 47]. It
243
can be explained by the flow through-pore heterogeneity of the porous polymeric monoliths
244
as one of the major sources of band broadening [30].
245
While the current results highlight the well-known structural flow-through pore
246
heterogeneity of porous polymer monoliths providing large “eddy” dispersion, its “true value”
247
does not become accessible with the current column format. It is stunning, that all solutes
248
seem to approach plate heights that upon further increase in superficial flow velocity taper off 7
Page 7 of 32
(Figures 4 and 5). For the larger flow velocity range, the plate height data indicate
250
independence of efficiency with respect to increased flow speed. In the quest, to understand
251
the shape of the plate height curves, we noticed a ratio of column length to its diameter of ≤
252
10, that is at least orders of magnitude smaller than that for monoliths typically found in
253
capillary sized conduits [6, 11, 25, 26, 53]. Here, isocratic experiments over comparable flow
254
velocity ranges (even under nonretained or only slightly retained conditions) indicate a loss of
255
column performance with increased flow-velocity for both polymer-based [11, 25], and silica-
256
based monoliths [53]. This observation further hints toward the pre-asymptotic (macroscopic)
257
sampling of efficiency with the current columns. We observed short elution times of only tens
258
of seconds (e.g. ~ 13 s for uracil at 2 mLmin-1 flow rate) that imply liquid chromatography
259
performed under pre-asymptotic dispersion conditions [54-56]. Here, the solutes entering at
260
the column head do not achieve radial dispersion equilibrium that is related to the observed
261
and accessible (macroscopic) efficiency. This situation becomes more climactic at increased
262
flow rates since the lateral (statistical) displacement of solutes significantly decreases [54],
263
which makes the plate height curves taper off. Therefore, we suggest observation of
264
apparently better efficiencies than expected for a longer column bed of identical architecture.
265
As a consequence, the unusual shape of the plate height curves does not allow a
266
straightforward analysis with common models of chromatographic dispersion processes.
267
Experimentally, the exact column inlet and outlet design including flow distribution in the frit
268
then imposes further difficulties in rationalizing performance [54-56].
d
M
an
us
cr
ip t
249
Unfortunately, the nonretained elution conditions are not appropriately reflecting the
270
practical domain of the monoliths’ application. Therefore we move toward the more
271
interesting retained elution regime. This condition considers the superposition of the
272
adsorption/desorption kinetics and potential impact of (slow) diffusion at the pore-fluid gel
273
interfaces.
Ac ce p
274
te
269
275
3.3. Retention dynamics of the scaffold and transition of performance under
276
nonretained elution conditions to that under retained conditions.
277 278
3.3.1 Retention dynamics and performance at a constant superficial flow
velocity
279
Figure 6a shows the apparent retention (with uracil as elution reference) of all solutes
280
against volume percentage of acetonitrile in the mobile phase at a constant superficial flow
281
velocity of 1.5 mm/s. We have seen earlier that the effect of molecular weight and
282
consequently size needs to be accounted for in the effective elution volume of larger
283
biomolecules. Practically, this means that the large solutes experience slight retention, but
284
still elute before or close to that of the mobile phase velocity tracer uracil. Uracil permeates 8
Page 8 of 32
an overall larger pore space due to its smaller size (Figure 2). The relatively steep slope of
286
apparent retention for the proteins is evident. Their elution volume is changing marginally,
287
shortly before a steep rise at the respective volume percentages of acetonitrile. Here,
288
adsorption at the pore-fluid gel interfaces starts to have a dominating effect upon elution
289
volume. As an example, the retention factor varies more than two orders of magnitude by just
290
changing 1.6 volume percent of acetonitrile in the mobile phase for ovalbumin (open
291
hexagons in Figure 6a), the largest solute used in the current study. This change is smaller
292
for the smaller sized solutes, though a generalization is not possible. For example,
293
enkephalin (open circles in Figure 6a) requires a change of almost 20% acetonitrile in the
294
mobile phase to accomplish a similar change in apparent retention as compared to that of
295
ovalbumin. The dependency of retention on mobile phase composition is strongly affected by
296
the molecular surface (and volume) of a specific solute undergoing interaction but also the
297
constitution of the pore fluid-gel interface (and volume) available for interactions at a specific
298
mobile phase composition.
an
us
cr
ip t
285
Monitoring retention (Figure 6a) and associated elution performance (Figure 6b)
300
shows the pronounced loss in performance for the larger sized solutes at weak retention.
301
Ovalbumin and myoglobin seem barely retained (eluting close to the small uracil, Figure 2)
302
and their elution performance rapidly decreases with slightly decreasing volume percentage
303
of acetonitrile in the mobile phase (open hexagons and filled hexagons in Figure 6). This
304
must have its origin in the adsorption/desorption kinetics and slow diffusion at the pore-fluid
305
gel interfaces, i.e. once adsorbed, the protein molecule barely moves. The steep slope of
306
retention against mobile phase solvent composition for the larger sized solutes (Figure 6a)
307
makes isocratic separations practically impossible. This indicates the well-known necessity to
308
use gradient elution conditions in separation practice. Steep gradient elution conditions
309
accomplish a nonretained elution performance with a zone sharpening history. Thus, for
310
larger solutes the column is effectively operated under nonretained elution conditions with a
311
gradient determined peak “release” after an effective focusing on the column head. It then
312
also demonstrates the option to increase flow speed and gradient steepness realizing
313
narrowly eluting peak bandwidth at short timescales. At increased flow speed, only a
314
relatively moderate decrease in performance under nonretained conditions is observed
315
(Figures 4 and 5).
Ac ce p
te
d
M
299
316 317
3.3.2 Retention dynamics and performance at varying flow velocity
318
The RP-1S column has a superficial velocity-based hydrodynamic permeability of 1.1 -14
m2, determined by measurements with 50/50 (v/v) acetonitrile/water and 0.5 mL/min
319
x 10
320
flow rate, an estimate well in good agreement to a previously reported value [28]. The large 9
Page 9 of 32
321
pre-asymptotically probed value of eddy dispersion under nonretained conditions indicates
322
substantial heterogeneity in the flow-through pore space. Additionally, it was observed that retention depends on superficial flow velocity that
324
logically increased the systems’ pressure. Figures 8 and 9 show plots of apparent retention
325
and associated performance for a selected set of solutes with variation in superficial flow
326
velocity. These solutes comprised the small val-tyr-val tripeptide with 379.45 Da (Figure 7)
327
and myoglobin with 17699 Da molecular weight (Figure 8). To allow a proper reference for
328
comparison, Figure 8b also shows efficiency data for the uracil at the respective mobile
329
phase compositions and flow velocities as tested for myoglobin.
cr
ip t
323
As seen in both figures, the apparent retention factor at constant mobile phase
331
compositions varies significantly with flow velocity, and asymptotically approaches a plateau
332
at superficial flow velocities larger than 1 mm/s (Figures 7a and 8a). Under these conditions
333
val-tyr-val shows retention even at high flow velocities. Retention increases by an order of
334
magnitude while decreasing the flow rates (Figure 7a). In some cases, such as for
335
myoglobin, elution is realized before the uracil at flow velocities greater than 1 mm/s, while
336
retention becomes significant at flow velocities smaller than 1 mm/s (Figure 8a).
M
an
us
330
This aspect is most surprising since high pressures in liquid chromatography on silica-
338
based materials such as studied by Gritti and Guiochon are known to increase retention for
339
insulin (from 75 bar upwards) [57]. This behavior was recently confirmed also for other
340
biomolecular solutes and columns operated at several hundred bars of pressure [58, 59].
341
The physical explanation of the increase in retention with high pressure is still under
342
extensive experimental investigation. In contrast to this aforementioned work with derivatized
343
silica, retention at increased flow rates and pressures (of only up to 150 bar system pressure
344
at 2 mL/min) appears to decrease significantly with the current column and experimental
345
conditions. We hypothesize that an increased pressure, translating to an increased pore
346
pressure inside the monolithic scaffolds, modulates the adsorption-desorption dynamics of
347
the bed. A change in pressure not only imposes conformational changes and varying molar
348
volumes on large molecules reported for much larger pressure [58, 59], it may rather change
349
the nanoscale interfacial structure of the pore-fluid gel interfaces found in polymer monoliths.
350
Likely, the free volume of the solvated polymer gel participating in retention may be
351
modulated, an indication of which could be found in the variation of the physicochemical
352
properties of individual structural (globular) elements found in such materials [48]. Varying
353
pressure may potentially impact the free volume of cross-linked solvated polymer material
354
under increased flow shear. In another study, Trilisky and Lenhoff argue that an increased
355
fluid drag may reduce retention of large biomolecules [50]. Eventually, under more extreme
356
conditions, macroscopic compression of the polymer as recently discussed by Podgornik et
Ac ce p
te
d
337
10
Page 10 of 32
357
al. [59] may be the ultimate result. Yet, varying pore pressure in the scaffold under flow may
358
impact the internal solvated polymer properties before such (macroscopic) deformation.
359
Additionally, we have seen that the solutes undergo pre-asymptotic dispersion under non-
360
retained conditions because of varying lateral displacement, an issue further complicating
361
substantiation of this observation. Taking these arguments and inspecting plate height curves for the two solutes then
363
shows the observation of an apparent “B-Term”. This apparent “B-Term” is totally absent
364
under nonretained elution conditions (Figures 4 and 5) and realized with the uracil at the
365
same mobile phase composition as for myoglobin (shown by the grey symbols in Figure 8b).
366
The plate height curves for the uracil collapse to one single master curve. Due to its more
367
than one order of magnitude lower molecular weight as compared to myoglobin, uracil should
368
show much more pronounced axial diffusion effects at low flow velocities (Figure 8b). This
369
discrepancy in the observed plate height curves is believed to be introduced by an increased
370
(apparent) retention of myoglobin at low flow rates, consequently reduced system pressure.
371
This situation reflects the decrease of performance at increased retention shown for the
372
larger superficial (but constant) flow velocity (Figure 6). Further seen is an apparent absence
373
of an increase in plate height at increased mobile phase velocities and an achievement of
374
performance that again approaches a plateau value within the tested velocity range (e.g.
375
Figures 7b and 8b). This is congruent to an approaching plateau value in the retention
376
dynamics at increased pressure (Figures 7a and 8a). Very slight changes in the volume
377
percentage of acetonitrile strongly impact the large biomolecule myoglobin at weak retention
378
only, while such changes are observable for the small peptide in a different range of retention
379
and mobile phase solvent compositions (Figures 8a vs. 7a).
381 382
cr
us
an
M
d
te
Ac ce p
380
ip t
362
4. Discussion
An experimental study of the porous and hydrodynamic properties of analytical
383
poly(styrene-co-divinylbenzene) monolithic scaffolds, using a set of biomolecular solutes
384
ranging in molecular weight from 181.19 Da to 44287 Da, is reported. The current study of
385
these robust columns focuses on different properties that are: (i) the columns apparent dry-
386
state microscopic appearance, (ii) the columns apparent porosity probed by the biomolecules
387
differing vastly in size and potentially modulated by mobile phase solvent composition, (iii)
388
the impact of probe solute size on apparent retention, and (iv) the elution performance under
389
both nonretained and retained elution conditions. Considering these complementary “model”
390
experiments accessible by the most simple and straightforward chromatographic
391
measurements of an identical material, the following findings can be set in context. 11
Page 11 of 32
At a most fundamental level, the polymer monoliths show a dynamic scaffold structure
393
(Figure 2). Clearly, porosity experienced by solutes varies with mobile phase solvent
394
composition. Principally, the total (physical) porosity does not change since the amount of
395
polymer material in the confine does not change, just its physicochemical state (i.e. free
396
volume) under varying experimental conditions. The apparent variation of porosity probed by
397
solutes of varying molecular weight does not seem detrimental under nonretained elution
398
conditions (Figures 3-5). However, it does, once only weak retention is imposed for the larger
399
biomolecular solutes and somewhat more strongly retained conditions are imposed for the
400
smaller sized solutes. This is irrespective of the actual amount of organic modifier in the
401
mobile phase (Figures 6-8). It is therefore most likely supported by the current experiments
402
that while changing the mobile phase solvent composition, stationary phase nanoscale gel
403
porosity becomes modulated [28]. Specific, well-chosen gradient elution conditions
404
effectively access a nonretained elution performance after zone sharpening on the column
405
head. Yet, the obtained results blur the performance picture in view of the still decent
406
performance obtained under purely nonretained elution conditions, i.e. several tens of
407
micrometer plate height (Figures 3-5). The performance situation becomes climactic if
408
retention at the pore-fluid gel interfaces becomes imposed that is affected both by molecular
409
surface (and volume) of a specific solute undergoing interaction but also the constitution of
410
the pore fluid-gel interface (and volume) available for interactions (Figure 6). We will not see
411
this for the materials typical application in a gradient elution scenario of proteins employing
412
relatively steep gradients. Here, the peaks elute at a timescale larger than implied under
413
nonretained isocratic conditions but with a peak bandwidth that is at least as narrow to tune
414
as after a shorter isocratic nonretained elution time. Apparently, this benefit becomes more
415
difficult for the smaller sized solutes that have a less steep dependence of retention on
416
mobile phase compositions (Figure 6). While this issue is well-known and utilized for the
417
practical application of porous polymer monoliths for large molecules, it is an area where the
418
detailed design of the microscale stationary phase morphology and homogeneity can
419
contribute to a significant advancement. This tailored microscale structure should further be
420
realized on length scales of the confine by minimizing wall effects, that may originate from
421
preparatory conditions in small and large dimensions [6, 61]. The described experiments
422
therefore provide more simple insight on the improvement of morphology and associated
423
performance of polymer monoliths. We should simply perform isocratic elution of small [28,
424
47] and large molecules in the absence and presence of partition and adsorption that gives
425
us a complete picture of the stationary phase performance realized in these different regimes
426
of elution. It is further suggested that this approach helps identifying if new preparation
Ac ce p
te
d
M
an
us
cr
ip t
392
12
Page 12 of 32
427
methods, aiming at new morphologies and chemistries, are readily useful and worthwhile to
428
implement.
429 430
5. Conclusions Results presented in this study clearly demonstrate that the efficiency properties of
432
porous polymer monoliths probed by solutes of biological origin varying vastly in size are a
433
relative sensitive measure for the flow-through pore heterogeneity in monolithic scaffolds and
434
related performance characteristics under nonretained conditions. Though the microscale
435
heterogeneity of these structures clearly has a significant impact on performance, it is
436
sampled in the pre-asymptotic regime. More detailed experiments show that the nanoscale
437
scaffold structure clearly becomes modulated in the presence of varying mobile phase
438
solvent composition. Existing micro- and mesopores, indicated for the material at a specific
439
solvated state, become modulated. We have seen an increase of free volume of polymer at
440
the cost of existent pore space accessible to large molecules. Under conditions of
441
nonretained elution this does not translate to an extremely detrimental effect on
442
hydrodynamic band dispersion for solutes that do not undergo partition in the actual gel
443
structure on a significant length scale due to their size and of course related adsorption, a
444
situation in line with previous conclusions [25, 28].
M
an
us
cr
ip t
431
When moving from the nonretained elution regime to a regime of retention at existent
446
pore-fluid gel interfaces, a distinct contribution to band broadening can be observed. This
447
contribution is associated with the adsorption/desorption kinetics and varying pore fluid-gel
448
interface diffusion. Here, the proteins show lowest mass transfer efficiency, in particular
449
thermodynamically relevant, as they generally move slowly by any diffusion process. This is
450
irrespective if diffusion occurs in solution or at pore-fluid gel interfaces.
te
Ac ce p
451
d
445
A factor quantitatively difficult to address in the current context and also for future
452
studies is the conformational change of already denatured large protein in the aqueous-
453
organic mobile phase and upon adsorption, a situation well-accepted under these
454
chromatographic elution conditions using hydrophobic monolithic backbones. Most
455
importantly and in contrast to small molecules that undergo both partition and adsorption
456
[29], the absence of an apparent C-Term does not completely disguise effects associated
457
with flow-through pore heterogeneity. This situation helped to identify a pre-asymptotic
458
elution regime for the current columns. This pre-asymptotic elution regime may yet mimic a
459
better chromatographic performance than expected from identical column morphology and
460
chemistry with larger column length-to-diameter aspect ratio, in particular at high flow
461
speeds. 13
Page 13 of 32
With respect to the presented effects we can conclude that the impact of molecular
463
size, retention, and mobile phase composition on the performance of porous polymer
464
monoliths is a useful tool to identify inherent column characteristics. This makes a
465
straightforward performance analysis possible to perform, but we have to accept that
466
physicochemical parameters of the polymer monolithic beds underlie dynamic changes. In
467
the current study, the existence of structural flow-through pore heterogeneity, though
468
practically not quantifiable in the studied pre-asymptotic regime of chromatography, is
469
identified as the major contribution to band broadening probed under nonretained conditions
470
and the described monolithic structures.
cr
ip t
462
Lastly, it becomes re-affirmed that shrinking structural elements of polymer monoliths
472
(translating to a reduced domain size) at the cost of a lower permeability, again significantly
473
improves performance for both small [28] and also large molecules. This realized reduced
474
domain size still has significant heterogeneity limiting further improvements in performance.
us
471
477 478
6. Acknowledgement
Dr. Kelly Flook from Thermo Fisher Scientific (Sunnyvale, California) is acknowledged
M
476
an
475
for the generous gift of the columns, without which this study would not have been possible.
480
d
479
7. References
482
[1] S. Hjerten, J.L. Liao, R. Zhang, J. Chromatogr. 473 (1989) 273.
483
[2] T.B. Tennikova, M. Bleha, F. Svec, T.V. Almazova, B.G. Belenkii, J. Chromatogr. 555,
484
(1991) 97.
485
[3] Q.C. Wang, F. Svec, J.M.J. Fréchet, J. Chromatogr. A 669 (1994) 230.
486
[4] I. Nischang, F. Svec, J.M.J. Fréchet, J. Chromatogr. A 1216 (2009) 2355.
487
[5] A. Jungbauer, R. Hahn, J. Chromatogr. A 1184 (2008) 62.
488
[6] I. Nischang, F. Svec, J.M.J. Fréchet, Anal. Chem. 81 (2009) 7390.
489
[7] I. Nischang, O. Brueggemann, F. Svec, Anal. Bioanal. Chem. 397(2010) 953.
490
[8] A. Premstaller, H. Oberacher, C.G. Huber, Anal. Chem. 72 (2000) 4386.
491
[9] H. Oberacher, C.G. Huber, TrAC-Trends Anal. Chem. 21 (2002) 166.
492
[10] P. Pruim, M. Öhman, Y. Huo, P.J. Schoenmakers, W.T. Kok, J. Chromatogr. A 1208
493
(2008) 109.
494
[11] J.H. Mohr, R. Swart, C.G. Huber, Anal. Bioanal. Chem. 400 (2011) 2391.
495
[12] M.H.M. van de Meent, S. Eeltink, G.J. de Jong, Anal. Bioanal. Chem. 399 (2011) 1845.
496
[13] Y. Li, H.D. Tolley, M.L. Lee, Anal. Chem. 81 (2009) 4406.
Ac ce p
te
481
14
Page 14 of 32
[14] T.D. Christopher, R.D. Arrua, M. Talebi, N.A. Lacher, E.F. Hilder, J. Sep. Sci. 36 (2013)
498
2782.
499
[15] P. Hemström, A. Nordborg, K. Irgum, F. Svec, J.M.J. Fréchet, J. Sep. Sci. 29 (2006) 25.
500
[16] Y.Y. Li, H.D. Tolley, M.L. Lee, Anal. Chem. 81 (2009) 9416.
501
[17] J. Krenkova, A. Gargano, N.A. Lacher, J.M. Schneiderheinze, F. Svec,
502
J Chromatogr. A 1216 (2009) 6824.
503
[18] M. Lonnberg, Y. Dehnes, M. Drevin, M. Garle, S. Lamon, N. Leuenberger, T. Quach, J.
504
Carlsson, J. Chromatogr. A 1217 (2010) 7031.
505
[19] E. L. Pfaunmiller, M. Hartmann, C. M. Dupper, C.M. S. Soman, D.S. Hage, J.
506
Chromatogr. A 1269 (2012) 198.
507
[20] E.L. Pfaunmiller, M.L. Paulemond, C.M. Dupper, M. Courtney, D.S. Hage, Anal. Bioanal.
508
Chem. 405 (2013) 2133.
509
[21] Y. Li, H.D. Tolley, M.L. Lee, Anal. Chem. 81(2009) 4406.
510
[22] R. Edam, S. Eeltink, D.J.D. Vanhoutte, W.Th. Kok, P.J. Schoenmakers, J. Chromatogr.
511
A 1218 (2011) 8638.
512
[23] J. Krenkova, F. Foret, F. Svec, J. Sep. Sci. 35 (2012) 1266.
513
[24] D. Connolly, S. Currivan, B. Paull, Proteomics 12 (2012) 2904.
514
[25] I. Nischang, O. Brüggemann, J. Chromatogr. A 1217 (2010) 5389.
515
[26] I. Nischang, I. Teasdale, O. Brüggemann, J. Chromatogr. A 1217 (2010) 7514.
516
[27] I. Nischang, I. Teasdale, O. Brüggemann, Anal. Bioanal. Chem. 400 (2011) 2289.
517
[28] I. Nischang, J. Chromatogr. A 1236 (2012) 152.
518
[29] I. Nischang, J. Chromatogr. A 1287 (2013) 39.
519
[30] H. Aoki, N. Tanaka, T. Kubo, K. Hosoya, J. Sep. Sci. 32 (2009) 341.
520
[31] H. Koku, R.S. Maier, K.J. Czymmek, M.R. Schure, A.M. Lenhoff, J. Chromatogr. A, 1218
521
(2011) 3466.
522
[32] H. Koku, R. S. Maier, M.R. Schure, A.M. Lenhoff, J. Chromatogr. A, 1237 (2012) 55.
523
[33] T. Müllner, A. Zankel, C. Mayrhofer, H. Reingruber, A. Höltzel, Y. Lv, F. Svec, U.
524
Tallarek, Langmuir 28 (2012) 16733.
525
[34] A.M. Siouffi, J. Chromatogr. A 1000 (2003) 801.
526
[35] J. Rozenbrand, W.P. van Bennekom, J. Sep. Sci. 34 (2011) 1934.
527
[36] S. Fekete, K. Ganzler, J. Fekete, J. Pharm. Biomed. Anal. 54 (2011) 482.
528
[37] S. Fekete, R. Berky, J. Fekete, J.-L. Veuthey, D. Guillarme, J. Chromatogr. A 1236
529
(2012) 177.
530
[38] B.M. Wagner, S.A. Schuster, B.E. Boyes, J.J. Kirkland, J. Chromatogr. A, 1264 (2012)
531
22.
532
[39] S. Bruns, T. Hara, B.M. Smarsly, U. Tallarek, J. Chromatogr. A,1218 (2011) 5187.
Ac ce p
te
d
M
an
us
cr
ip t
497
15
Page 15 of 32
[40] S. Bruns, D. Stoeckel, B.M. Smarsly, U. Tallarek, J. Chromatogr. A,1268 (2012) 53.
534
[41] K. Jerabek, Anal. Chem. 57 (1985) 1598.
535
[42] F. Nevejans, M. Verzele, Chromatographia 20 (1985) 172.
536
[43] J.V. Dawkins, L.L. Lloyd, F.P. Warner, J. Chromatogr. 352 (1986) 157.
537
[44] R.M. Smith, D.R. Garside, J. Chromatogr. 407 (1987) 19.
538
[45] F. Nevejans, M. Verzele, J. Chromatogr. 406 (1987) 325.
539
[46] B. Ells, Y. Wang, F.F. Cantwell, J. Chromatogr. A 835 (1999) 3.
540
[47] T.J. Causon, E.F. Hilder, I. Nischang, J. Chromatogr. A 1263 (2012) 108.
541
[48] M. Laher, T. Causon, W. Buchberger, S. Hild, I. Nischang, Anal. Chem. 85 (2013) 5645.
542
[49] E.I. Trilisky, H. Koku, K.J. Czymmek, A.M. Lenhoff, J. Chromatogr. A 1216 (2009) 6365.
543
[50] E.I. Trilisky, A.M. Lenhoff, J. Chromatogr. A 1217 (2010) 7372.
544
[51] R. Hahn, A. Jungbauer, Anal. Chem. 72 (2000) 4853.
545
[52] J.-L. Cabral, D. Bandilla, C.D. Skinner, J. Chromatogr. A 1108 (2006) 83.
546
[53] D. Hlushkou, S. Bruns, A. Höltzel, U. Tallarek, Anal. Chem. 82 (2010) 7150.
547
[54] S. Khirevich, A. Holtzel, A. Seidel-Morgenstern, U. Tallarek, Anal. Chem. 81 (2009)
548
7057.
549
[55] F. Gritti, G. Guiochon, J. Chromatogr. A 1262 (2012) 107.
550
[56] F. Gritti, G. Guiochon, Anal. Chem. 85 (2013) 3017.
551
[57] F. Gritti, G. Guiochon, Anal. Chem. 81 (2009) 2723.
552
[58] S. Fekete, J.-L. Veuthey, D.V. McCalley, D. Guillarme, J. Chromatogr. A 1270 (2012)
553
127.
554
[59] S. Fekete, K. Horvath, D. Guillarme, J. Chromatogr. A 1311 (2013) 65.
555
[60] A. Podgornik, A. Savnik, J. Jancar, N.L. Krajnc, J. Chromatogr. A 1333 (2014) 9.
556
[61] A. Podgornik, J. Jancar, I. Mihelic, M. Barut, A. Strancar, Acta Chim. Slov. 57 (2010) 1.
558 559
cr
us
an
M
d
te
Ac ce p
557
ip t
533
Figure Captions
560
Fig. 1. Scanning electron micrographs of the cross-sectional structure of (a) the RP-3U, and
561
(b) the RP-1S column at different magnifications as indicated.
562 563
Fig. 2. Plot of molecular weight of probe solutes against measured apparent scaffold porosity
564
at a superficial flow velocity of 0.5 mm/s. In this study the volume fraction of acetonitrile in the
565
mobile phase was varied for the RP-1S column: (filled squares) 50% acetonitrile, (open
566
circles) 60% acetonitrile, (filled triangles) 65% acetonitrile, (open diamonds) 70% acetonitrile,
567
(filled hexagons) 80 % acetonitrile, (open pentagons) 85 % acetonitrile. The data for the RP-
568
3U column (filled grey squares) are included for comparison. 16
Page 16 of 32
569 570
Fig. 3. Height equivalent to a theoretical plate against molecular weight of the probe solutes
571
with the RP-1S (circles) and RP-3U column (squares) and two selected superficial flow
572
velocities of 1 mm/s (filled symbols) and 2 mm/s (open symbols). The mobile phase solvent
573
composition was 50/50 acetonitrile/water (v/v).
ip t
574
Fig. 4. Plate height curves measured with the RP-3U and RP-1S monolithic column at a
576
mobile phase composition of 50/50 acetonitrile/water (v/v). (a) Uracil RP-1S (filled squares),
577
uracil RP-3U (open squares); (b) myoglobin RP-1S (filled circles), myoglobin RP-3U (open
578
circles).
cr
575
us
579
Fig. 5. Plate height curves with a mobile phase composition of 50/50 acetonitrile/water (v/v)
581
for (a) uracil (filled squares), tyrosine (half-filled squares), val-tyr-val (filled circles),
582
enkephaline (half-filled circles); (b) angiotensine (filled hexagons), insulin (half-filled
583
hexagons), cytochrome c (filled diamonds); (c) myoglobin (half-filled diamonds), ovalbumine
584
(filled pentagons).
M
an
580
585
Fig. 6. Apparent retention factors and elution performance for the probe solutes plotted
587
against volume percentage of acetonitrile in the mobile phase at a constant superficial flow
588
velocity of 1.5 mm/s. (a) Apparent retention factor with uracil as elution reference for
589
calculation of the retention factor, and (b) height equivalent to a theoretical against volume
590
percentage of acetonitrile in the mobile phase. Symbols: tyrosine (filled squares), val-tyr-val
591
(open squares), angiotensine (filled circles), enkephaline (open circles), insulin (filled
592
triangles), cytochrome c (open triangles), myoglobin (filled hexagons), and ovalbumine (open
593
hexagons).
te
Ac ce p
594
d
586
595
Fig. 7. Apparent retention factors (with uracil as elution reference) and elution performance
596
for val-tyr-val as the probe solute, plotted against superficial flow velocity. (a) Apparent
597
retention factor of val-tyr-val with uracil as elution reference, and (b) height equivalent to a
598
theoretical plate of val-tyr-val. Symbols indicating volume percentage of acetonitrile: 5%
599
(filled squares), 6% (half-filled squares), 7% (filled circles), 10% (half-filled circles).
600 601
Fig. 8. Apparent retention factors (with uracil as elution reference) and elution performance
602
for myoglobin as the probe solute plotted against superficial flow velocity. (a) Apparent
603
retention factor of myoglobin, and (b) height equivalent to a theoretical plate for myoglobin
604
against superficial flow velocity. Uracil is shown with grey symbols at the same volume 17
Page 17 of 32
605
percentages of acetonitrile in the mobile phase. Symbols indicating volume percentage of
606
acetonitrile in the mobile phase: 34% (filled squares), 35% (half-filled squares), 36% (filled
607
circles), 37% (half-filled circles).
Ac ce p
te
d
M
an
us
cr
ip t
608
18
Page 18 of 32
608 609 610 611
Highlights Elution performance of porous polymer monoliths for small and large biomolecules
614 615
Modulated gel porosityby elution experiments at varying mobile phase compositions
616 617
Performance atnonretainedconditions and biomolecule molecular weight is addressed
618
Plate height curves taper off at high flow speedandpre-asymptotic dispersion
619 620
Retention and efficiency depend strongly on mobile phase composition and flow rate
an
us
cr
ip t
612 613
Ac ce p
te
d
M
621
19
Page 19 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 1
Page 20 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 2
Page 21 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 3
Page 22 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4a
Page 23 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4b
Page 24 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 5a
Page 25 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 5b
Page 26 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 5c
Page 27 of 32
Ac ce p
te
d
M
an
us
cr
ip t
Figure 6ab
Page 28 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 7a
Page 29 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 7b
Page 30 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 8a
Page 31 of 32
Ac
ce
pt
ed
M
an
us
cr
i
Figure 8b
Page 32 of 32