Available via license: CC BY 4.0
Content may be subject to copyright.
Citation: Haro Mares, N.B.; Döller,
S.C.; Wissel, T.; Hoffmann, M.; Vogel,
M.; Buntkowsky, G. Structures and
Dynamics of Complex Guest
Molecules in Confinement, Revealed
by Solid-State NMR, Molecular
Dynamics, and Calorimetry. Molecules
2024,29, 1669. https://doi.org/
10.3390/molecules29071669
Academic Editors: Michele R.
Chierotti and Simona Goliˇc
Grdadolnik
Received: 29 February 2024
Revised: 29 March 2024
Accepted: 5 April 2024
Published: 8 April 2024
Copyright: © 2024 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
molecules
Review
Structures and Dynamics of Complex Guest Molecules in
Confinement, Revealed by Solid-State NMR, Molecular
Dynamics, and Calorimetry
Nadia B. Haro Mares 1, Sonja C. Döller 1, Till Wissel 1, Markus Hoffmann 2,* , Michael Vogel 3,* and
Gerd Buntkowsky 1, *
1Eduard-Zintl-Institut für Anorganische und Physikalische Chemie, Technische Universität Darmstadt,
Peter-Grünberg-Str. 8, D-64287 Darmstadt, Germany; haromares@chemie.tu-darmstadt.de (N.B.H.M.);
sonja.doeller@tu-darmstadt.de (S.C.D.); wissel@chemie.tu-darmstadt.de (T.W.)
2Department of Chemistry and Biochemistry, State University of New York at Brockport, Brockport,
NY 14420, USA
3Institute for Condensed Matter Physics, Technische Universität Darmstadt, Hochschulstr. 6,
D-64289 Darmstadt, Germany
*Correspondence: mhoffman@brockport.edu (M.H.); michael.vogel@physik.tu-darmstadt.de (M.V.);
gerd.buntkowsky@chemie.tu-darmstadt.de (G.B.)
Abstract: This review gives an overview of current trends in the investigation of confined molecules
such as water, small and higher alcohols, carbonic acids, ethylene glycol, and non-ionic surfactants,
such as polyethylene glycol or Triton-X, as guest molecules in neat and functionalized mesoporous
silica materials employing solid-state NMR spectroscopy, supported by calorimetry and molecular
dynamics simulations. The combination of steric interactions, hydrogen bonds, and hydrophobic
and hydrophilic interactions results in a fascinating phase behavior in the confinement. Combining
solid-state NMR and relaxometry, DNP hyperpolarization, molecular dynamics simulations, and
general physicochemical techniques, it is possible to monitor these confined molecules and gain deep
insights into this phase behavior and the underlying molecular arrangements. In many cases, the
competition between hydrogen bonding and electrostatic interactions between polar and non-polar
moieties of the guests and the host leads to the formation of ordered structures, despite the cramped
surroundings inside the pores.
Keywords: confinement; NMR; molecular dynamics; mesoporous silica
1. Introduction
Ordered periodical mesoporous silica (PMS) [
1
,
2
], like MCM-41 (Mobil Composition of
Matter No. 41) [
3
] and SBA-15 (Santa Barbara Amorphous) [
4
,
5
] and their many derivates,
exhibit characteristic narrow pore-diameter distributions and large specific surface areas.
Their high chemical stability makes them easy to handle under ambient conditions. Their
reactive surface silanol groups (Si-OH) provide an easy pathway to chemical functionaliza-
tion and tailored surface design, e.g., by post-synthetic grafting of functional groups such as
amino, amide, carboxyl, phosphate [
6
,
7
], or by co-condensation with molecules containing
such groups [
8
,
9
]. They have a high application potential in many technical processes, such
as heterogeneous catalysis, separation technology, encapsulation of molecules, drug deliv-
ery, or selective adsorption [
10
–
12
]. Moreover, they provide an ideal model environment
investigating the physicochemical properties of fluid guest molecules confined in a porous
environment, showing strong competition with solid–liquid and liquid–liquid interactions
or in biomineralization [13].
Revealing these properties necessitates the combination of several analytical and
computational techniques, such as X-ray (XRD) and neutron diffraction techniques for the
investigation of crystallinity and long-range order [
14
–
17
], small angle scattering (SAXS and
Molecules 2024,29, 1669. https://doi.org/10.3390/molecules29071669 https://www.mdpi.com/journal/molecules
Molecules 2024,29, 1669 2 of 26
SANS) for the characterization of the pore geometry and pore ordering, differential scanning
calorimetry (DSC) [18] for the study of phase or glass transition processes in confinement,
gas adsorption (BET, BJH) for the characterization of the pore ordering, specific surface areas
and pore diameters [
19
,
20
], solid-state NMR (SSNMR), and NMR diffusometry
[21–25],
possibly supported by Dynamic Nuclear Polarization (DNP) [
26
–
33
], to boost the NMR
sensitivity or chemical shift calculations to help in the interpretation [
34
,
35
], or the study
of the local ordering and dynamics on the molecular level and molecular dynamics (MD)
simulations for the modelling of the dynamics and structures of the confined guests in the
host material.
Hydrogen-bonded liquids in nanoscale confinements are a highly topical field of
research [
36
]. In view of their enormous relevance to science and application, particularly
intensive research efforts ascertained the properties of neat water and aqueous solutions
under such circumstances [
37
,
38
]. Also, the properties of confined alcohol molecules
in dependence on the size and chemistry of the confining framework were often a re-
search focus [
36
,
39
]. In such investigations, phase behaviors, structures, and dynamics
of hydrogen-bonded liquids were addressed, applying a wide range of experimental and
computational methods [
39
]. For studies of dynamical aspects, broadband dielectric spec-
troscopy [
40
], quasielastic neutron scattering [
41
], NMR spectroscopy and diffusometry [
42
],
and MD simulations [
43
,
44
] were very suitable methods. Various comprehensive review
articles summarized these research efforts [36–41,43,44].
Here, we review NMR approaches to the structural and dynamical properties of
hydrogen-bonded liquids in porous frameworks. In doing so, we build upon several
review articles covering this topic [
45
–
49
]. Using isotope selective approaches in NMR
spectroscopy, both rotational motion and translational diffusion in confinement were suc-
cessfully determined. To characterize the reorientation of confined molecules,
2
H NMR
studies of deuterated compounds proved to be a powerful tool [
50
–
52
]. In particular, when
combining
2
H spin-lattice relaxation (SLR) analysis, including
2
H field-cycling relaxome-
try [
53
,
54
], with
2
H stimulated-echo experiments (STE), it was possible to follow MD over a
broad range of correlations times [
55
–
57
],
τ≈
10
−11
–10
0
s. Furthermore, NMR experiments
in magnetic field gradients enabled measurements of self-diffusion coefficients [
23
]. In
these approaches, it was advantageous to use a static field gradient (SFG) rather than a
pulsed one [
58
,
59
]. The SFG method enabled an application of stronger gradients and, in
this way, an observation of diffusion on smaller-length scales down to roughly 100 nm.
By exploiting these capabilities, it was possible to ensure that diffusion inside a particular
framework is probed, e.g., inside a given pore, whereas distorting effects from an escape
of the confinement can be neglected, e.g., fast displacements in empty space between
mesoporous silica particles.
While these NMR approaches provide a quantitative evaluation of the molecular
motions present, their qualitative interpretation on the molecular level can be challenging.
To overcome this challenge, MD simulations provide a powerful theoretical approach
to gain such molecular level insights. In MD simulations, an ensemble of molecules is
allowed to evolve in time to obtain essentially a movie that reveals the present dynamic
processes, intermolecular interactions, and the resulting structural patterns. The potentials
of all bonding and non-bonding interactions between all present atoms must be defined at
each time increment in order to obtain a new set of velocities at which each atom moves
during the next simulation step. In ab-initio MD (AIMD) simulations, the potentials are re-
calculated ab-initio at each simulation time step [
60
]. The AIMD method is computationally
very demanding, limiting its use to smaller-sized systems. For that reason, more commonly
used are MD simulations that employ classical potential functions during the simulation.
Bonding interactions typically consist of harmonic oscillator functions to describe chemical
bonds and bond-angle vibrations and sinusoidal functions to describe the energy barriers
for dihedral rotations. Non-bonding interactions are typically comprised of the Coulomb
potential between (partial) charges and the Lennard Jones potential for describing the
London dispersion forces. Numerous sets of parameters classically describing all of these
Molecules 2024,29, 1669 3 of 26
interactions have been developed over time, and are referred to as force fields. Some of
the most popular force fields include AMBER [
61
,
62
], OPLS/AA [
63
], CHARMM [
64
],
and GROMOS [
65
]. To reduce computation times, some forcefields lump groups of atoms
together, as is the case for the GROMOS force field where CH
2
groups for example are
described as one constituent. Even more coarse-grained force fields have been developed,
as well such as the MARTINI [
66
] force field that was specifically optimized for simulating
polymers. The quality of the classical force field is assessed by comparison between
simulated and experimental data. Aside reproducing experimental data, the most common
analysis tasks of the MD simulations include the evaluation of radial distribution functions,
which provide direct insights into the structural organization of the studied systems, and
thus the present interactions, as well as the inspection of various correlations functions,
from which time constants of present dynamics can be extracted and compared with
experimental values. In the case of systems that engage in hydrogen bonding, these
can be directly assessed from MD simulations, which experimentally is very difficult to
achieve [
67
]. The interested reader is referred to several references for more details about
MD simulations, such as the application of periodic boundary conditions and the pressure
and temperature equilibration procedures [68,69].
In a recent review [
47
] some of us gave an extensive overview about the state of the
art of confinement studies of small molecules, such as confined water, small aromatic
molecules, alcohols, or carbonic acids [
50
,
70
–
77
] hosted in these materials, and discussed
how the confinement affects thermophysical properties such as freezing and melting
points of the guest molecules. While confinement effects on small guest molecules with
simple physicochemical properties in mesoporous environments are well investigated,
until recently not much was known about the local structures of more complex molecules,
such as surfactants, in mesoporous confinement. In continuation with the previous review,
the present paper gives an overview of a series of newer studies, where more complex
molecules are confined inside these materials. The larger number of functional molecular
sites permits a larger number of possible interactions, which enables these molecules
to form more complex or richer structures than simple small molecules like benzene or
pyridine. Of particular interest here is the competition between surface–guest hydrogen
bonds and intermolecular (and possibly also intramolecular) guest–guest hydrogen bonds.
This review only summarizes the findings since 2020. For a very extensive overview of
older work, the reader is referred to the previous review [47].
The rest of this review is organized as follows: Section 2gives an introduction into the
preparation and surface modification of the mesoporous host materials and the investigated
surfactants; Section 3discusses the physicochemical properties of these guest molecules
in their bulk phases and their behavior inside the confinement; and the review concludes
with a Summary and Outlook into possible future developments in the field.
2. Materials and Methods
2.1. Host-Materials
Mesoporous silica materials like MCM-41 or SBA-15 combine large and adjustable (via
the preparation) pore sizes, specific volumes, and specific surface areas with high thermal
stability, low specific weights, and narrow pore-diameter distributions [
78
–
80
]. Both types
of materials are relatively easy to prepare and to functionalize, following, e.g., the synthesis
protocol by Grünberg et al. [
81
,
82
] or Grün et al. [
83
] (for details, see refs.
[36,80,84])
. Their
quality, pore dimensions, and surface parameter can be easily determined by the combina-
tion of nitrogen adsorption (BET and BJH),
29
Si SSNMR spectroscopy, and SAXS. Important
to note is that freshly prepared samples contain a substantial amount of surface-bound wa-
ter molecules [
49
,
85
,
86
], which in general have to be removed for confinement [
22
] studies
employing special drying protocols for the preparation of “water-free” silica samples [
87
].
Molecules 2024,29, 1669 4 of 26
2.2. Probe Molecules
The probe molecules considered in this review are water, octanol, ethylene glycol, and
the surfactants E
5
, polyethylene glycol, C
10
E
6
, and Triton-X (see Figure 1). Each of these
chemical structures contains hydroxy as well as ether moieties, which both can engage
in hydrogen-bonding interactions. While octanol, C
10
E
6
, and Triton-X have only a single
hydroxyl group, which can interact with the silica surfaces, water, ethylene glycol (EG),
and its polymers (PEGs) can interact via two terminal hydroxyl groups. Moreover, in the
case of E
5
, C
10
E
6
, and PEG, there is a length-dependent number of ether-oxygens, which
can serve as a hydrogen-bond acceptor in competition to the hydroxyl groups.
Molecules 2024, 29, x FOR PEER REVIEW 4 of 27
be removed for confinement [22] studies employing special drying protocols for the prep-
aration of “water-free” silica samples [87].
2.2. Probe Molecules
The probe molecules considered in this review are water, octanol, ethylene glycol,
and the surfactants E5, polyethylene glycol, C10E6, and Triton-X (see Figure 1). Each of
these chemical structures contains hydroxy as well as ether moieties, which both can en-
gage in hydrogen-bonding interactions. While octanol, C10E6, and Triton-X have only a
single hydroxyl group, which can interact with the silica surfaces, water, ethylene glycol
(EG), and its polymers (PEGs) can interact via two terminal hydroxyl groups. Moreover,
in the case of E5, C10E6, and PEG, there is a length-dependent number of ether-oxygens,
which can serve as a hydrogen-bond acceptor in competition to the hydroxyl groups.
Figure 1. (A) Structures of 1-Octanol, EG, and the surfactants studied in this work. Except for E5, the
surfactants are polydisperse mixtures (exact compositions are given in ref. [88]). (B) Sketch of neat
and functionalized mesoporous silica material.
2.2.1. 1-Octanol
1-octanol, an unbranched saturated fay alcohol with the molecular formula
CH3(CH2)7OH, is commonly employed in the synthesis of esters. Owing to the hydrophilic
hydroxy-group and the lipophilic alkyl chain, it is an ideal small model surfactant. It is
often employed for evaluating the lipophilicity of pharmaceutical products. A quantita-
tive measure for this is the water octanol partition coefficient or p-value Kow [89]. It can be
employed, e.g., for estimations of the partitioning of dissolved drug molecules between
the cytosol and lipid membranes of living systems in pharmacology, or the behavior of
water/oil mixtures in geology or environmental science [90]. Water–octanol mixtures are
Figure 1. (A) Structures of 1-Octanol, EG, and the surfactants studied in this work. Except for E
5
, the
surfactants are polydisperse mixtures (exact compositions are given in ref. [
88
]). (B) Sketch of neat
and functionalized mesoporous silica material.
2.2.1. 1-Octanol
1-octanol, an unbranched saturated fatty alcohol with the molecular formula
CH
3
(CH
2
)
7
OH, is commonly employed in the synthesis of esters. Owing to the hydrophilic
hydroxy-group and the lipophilic alkyl chain, it is an ideal small model surfactant. It is
often employed for evaluating the lipophilicity of pharmaceutical products. A quantitative
measure for this is the water octanol partition coefficient or p-value K
ow
[
89
]. It can be
employed, e.g., for estimations of the partitioning of dissolved drug molecules between
the cytosol and lipid membranes of living systems in pharmacology, or the behavior of wa-
ter/oil mixtures in geology or environmental science [
90
]. Water–octanol mixtures are ideal
model systems of the phase behavior of immiscible liquids in bulk and confined phases
by combinations of solid-state NMR techniques such as 1D
1
H-MAS NMR,
29
Si-CP-MAS
(Cross-Polarization Magic Angle Spinning) NMR, and
1
H/
29
Si-HETCOR-(FSLG)-NMR
(Heteronuclear Correlation by Frequency Switched Lee–Goldburg Decoupling) combined
Molecules 2024,29, 1669 5 of 26
with MD simulations are commonly employed in such studies. The combination of these
techniques reveals important information such as the distributions of the two liquids inside
a confinement, as was shown recently by Kumari et al. in a series of papers [91–93].
2.2.2. Ethylene Glycol, Pentaethylene Glycol, and Polyethylene Glycol (PEG)
Ethylene glycol (EG) is the smallest vicinal diol and the simplest example of a poly-
hydric alcohol. EG finds wide application in the production of polyester fibers and for
antifreeze formulations. Owing to the presence of the two hydroxyl groups, it can be em-
ployed as a simple model for confined liquids that can interact via several hydrogen bonds
with host surface groups. Typical examples of these studies include NMR studies of small
confined molecules inside zeolites [
22
,
94
–
96
] or mesoporous silica materials [
45
,
97
–
99
]
and their functionalized derivates [
100
,
101
]. The low molecular weight representatives of
polyethylene glycols (PEG, H-[O-CH
2
-CH
2
]
n
-OH) possess environmentally benign proper-
ties including no toxicity, low vapor pressure, reducing exposure through inhalation and
biodegradability. PEG is widely and relatively inexpensively available with an industrial
annual production of about 500,000 tons per year [
102
]. Commercial PEG is manufactured
as polydisperse mixtures, where the average molar weight is part of the product name. For
example, PEG200 has an average molar weight of approximately 200 g mol
−1
. PEG is a
very good solvent for a wide variety of chemicals including some mineral salts [
103
], which
facilitates its use, e.g., in transition metal catalyzed reactions, including heterogeneously
catalyzed reactions where the transition metal catalyst is immobilized on a solid material of
large surface area [
104
]. Accordingly, PEG has been increasingly used as a very attractive
alternative solvent for Green Chemistry [
105
]. Its properties as an alternative solvent for
chemical synthesis were reviewed in several recent articles [
106
–
108
]. To further aid these
efforts in heterogenous catalysis using PEG as a solvent, they need to be studied under
confinement to understand how their physical and chemical properties change under these
conditions as a function of the degree of polymerization. Such studies should combine
experimental studies employing NMR and thermodynamic measurements with theoretical
methodologies such as MD simulation in order to be able to correctly interpret experimental
results at the molecular level. In this review, confinement studies of the monomer EG, two
different EG polymers, namely E
5
, a monodisperse polymer with chain length of five and
a polydisperse polymer with a distribution of chain lengths (PEG200), are reported and
compared to the commercial surfactants C10E6and Triton X-100 (see Figure 1).
2.3. Simulation Methods
As there are a number of excellent reviews on MD simulations [
109
–
111
], here, only
the salient features relevant for this review are shortly summarized. The simulations were
carried out using the GROMACS platform [
112
,
113
]. For liquids, the typical simulation
protocol consisted of the steps of randomly inserting usually 1000 molecules into a virtual
box that is chosen in size to be too large, removing close contacts between atoms that may
have arisen during the random molecule insertion, letting the density equilibrate at the
desired constant simulation temperature and pressure, and finally, after establishing the av-
erage density, simulating long enough at constant temperature and volume corresponding
to the average density so that the system reaches the diffusive regime, which then allows
extraction of the self-diffusion coefficient. For PEG200, simulation times of 300 ns were
typical at a temperature of 328 K saving at least 10,000 frames for analysis. The simulations
results summarized in this review were mostly obtained with the OPLS/AA force field.
As is common, periodic boundary conditions were employed along with a Verlet cutoff
scheme [
114
], and treatment of long-rang electrostatic interactions were employed with a
smooth Particle-Mesh Ewald (PME) grid-wise cubic interpolation [
115
]. Temperature con-
trol was established with the Bussi–Donadio–Parinello velocity-rescaling thermostat [
116
],
while pressure was controlled with the Parrinello-Rahman barostat [
117
,
118
]. Analysis
of the obtained simulations was carried out mainly using modules available with the
Molecules 2024,29, 1669 6 of 26
GROMACS platform augmented by some self-developed script files that can be found
along with a very detailed description of the simulation details in Hoffmann et al. [119].
2.4. Differential Scanning Calorymetry (DSC)
The DSC analysis were performed using a Heat Flux DSC. In this type of DSC, the
sample and the reference are heated through the same heating source. Nitrogen was used
as the purge and protective gas during the experiments.
The samples mentioned in this review were prepared and packed under inert condi-
tions and tested within a temperature range of 100 to 300
◦
C under various heating/cooling
rates, depending on the sample. For more detailed information about a specific sample, the
reader is referred to the original papers [48,120].
3. Exemplary Studies
When studying the temperature-dependent dynamics of confined molecules that
readily crystallize, such as water, it is important to consider that confinement usually affects
the freezing and melting behaviors. This necessitates the determination whether specific
findings relate to the fully liquid state above the melting temperature (T
m
) or the partially
frozen state below this temperature. In partially frozen states, crystalline regions near the
pore center coexist at equilibrium with a liquid layer at the pore wall [
45
,
52
,
121
–
123
]. As
specific examples of NMR reorientation and diffusion studies on fully liquid or partially
frozen states in nanoscale confinements, we will discuss the dynamical properties of water
(H
2
O and D
2
O), ethylene glycol, and LiCl aqueous solutions in native and functionalized
mesoporous silica. This approach will provide detailed insights into the dependence of the
rotational and diffusive motions of hydrogen-bonded liquids on the size, i.e., the diameter
d, of the pores and the chemistry of their walls.
3.1. Water and Ion Dynamics
In partially frozen states, the molecules of the liquid layer interact with the porous
framework in their immediate neighborhood very closely, enabling detailed insights into
the influence of the confinement chemistry on liquid dynamics. Exploiting this possibility,
2
H NMR was utilized to investigate D
2
O reorientation near different biomimetic interfaces,
specifically near silica walls functionalized with various amino acids [
58
,
124
]. It was found
that the rotational correlation times
τ
obtained from
2
H SLR analysis were longer in the
functionalized than the pristine silica pores, but with values strongly dependent on the
type of the amino acid; see Figure 2. The longest correlation times were observed for
lysine (LYS), followed by alanine (ALA) and, finally, glutamic acid (GLU) functionalization.
Based on these results, it was concluded that the flexibility of the surface groups is not the
decisive parameter for the mobility of neighboring water molecules [
58
]. Instead, it was
proposed that water reorientation is slower near amino acids with basic residues than near
those with acidic ones. In addition, it is evident from Figure 2that, independent of the
functionalization, water dynamics exhibited a high temperature dependence in the narrow
interfacial layers between the pore walls and the ice cores, described by an activation of
about 1 eV, corresponding to nearly 100 kJ/mol.
Other studies employed very narrow confinements with diameters of d
≈
2 nm to fully
suppress crystallization of water [
53
,
56
,
57
]. In these cases, a combination of
2
H SLR and
STE studies provided access to the slowdown of water reorientation when approaching a
glass transition. An important question of such research was to what degree the dynamics
of confined water resembles that of bulk water in the supercooled regime, which is difficult
to access due to rapid crystallization. In particular, it was vigorously debated whether
or not dynamical crossovers observed for confined water [
40
,
125
,
126
] may be taken as
evidence for the existence of a second critical point associated with a liquid–liquid phase
transition of bulk water, which was proposed to be at the origin of water’s anomalies [
127
].
Molecules 2024,29, 1669 7 of 26
Molecules 2024, 29, x FOR PEER REVIEW 7 of 27
Figure 2. Rotational correlation times τ of D2O in pristine (d = 5.4 nm) [57] and functionalized [124]
mesoporous silica. For the functionalization, SBA-15 silica was functionalized via co-condensation
with 3-(aminopropyl)triethoxysilane (APTES), yielding SBA-APT (d = 6.8 nm). Afterwards, the
amino acids lysine (LYS, d = 5.8 nm), alanine (ALA, d = 5.9 nm), or glutamic acid (GLU, d = 5.6 nm)
were coupled to this SBA-APT batch. The surface densities of the linked amino acids were 0.4–0.5
nm−2. The dashed lines are Arrhenius fits with activation energies of ca. 1 eV.
Other studies employed very narrow confinements with diameters of d ≈ 2 nm to
fully suppress crystallization of water [53,56,57]. In these cases, a combination of 2H SLR
and STE studies provided access to the slowdown of water reorientation when approach-
ing a glass transition. An important question of such research was to what degree the dy-
namics of confined water resembles that of bulk water in the supercooled regime, which
is difficult to access due to rapid crystallization. In particular, it was vigorously debated
whether or not dynamical crossovers observed for confined water [40,125,126] may be
taken as evidence for the existence of a second critical point associated with a liquid–liquid
phase transition of bulk water, which was proposed to be at the origin of water’s anoma-
lies [127].
Electrolyte solutions in nanoscale confinements are of enormous relevance across
various fields. Their applications in heterogeneous catalysis and energy conversion are
fundamental to modern society, and ion channels are crucial for the biological functions
of living cells. However, many properties of electrolyte solutions in interfaces remain in-
sufficiently understood to this day. The isotope selectivity of NMR was exploited to sep-
arately analyze water and ion dynamics for LiCl solutions in the bulk [128] and various
confinements [129,130]. Figure 3 shows diffusion coefficients of LiCl–7H2O solution ob-
tained from 1H and 7Li SFG studies, which reflect the mobility of the water molecules and
lithium ions, respectively. This composition, which is close to the eutectic one, was chosen
to suppress crystallization and enable investigations in a broad temperature range. For
the bulk solution, it was found that the 1H diffusivity was slightly larger than the 7Li dif-
fusivity, and both exhibited a prominent non-Arrhenius temperature dependence typical
of many glass-forming liquids. When confining the LiCl–7H2O solution to a pristine silica
pore with a diameter of d = 3.0 nm, the difference of the 1H and7Li diffusivities increased
to nearly an order of magnitude, and the temperature dependence again changed to an
Arrhenius behavior with an activation energy of Ea = 0.26 eV [129]. It was argued that these
prominent confinement effects resulted from Stern layer formation at the usually nega-
tively charged silica walls. In a silica material with dye molecules grafted to the inner sur-
faces, even smaller 7Li diffusion coefficients were reported [130]. To rationalize this
Figure 2. Rotational correlation times
τ
of D
2
O in pristine (d= 5.4 nm) [
57
] and functionalized [
124
]
mesoporous silica. For the functionalization, SBA-15 silica was functionalized via co-condensation
with 3-(aminopropyl)triethoxysilane (APTES), yielding SBA-APT (d= 6.8 nm). Afterwards, the amino
acids lysine (LYS, d= 5.8 nm), alanine (ALA, d= 5.9 nm), or glutamic acid (GLU, d= 5.6 nm) were
coupled to this SBA-APT batch. The surface densities of the linked amino acids were 0.4–0.5 nm
−2
.
The dashed lines are Arrhenius fits with activation energies of ca. 1 eV.
Electrolyte solutions in nanoscale confinements are of enormous relevance across
various fields. Their applications in heterogeneous catalysis and energy conversion are
fundamental to modern society, and ion channels are crucial for the biological functions
of living cells. However, many properties of electrolyte solutions in interfaces remain
insufficiently understood to this day. The isotope selectivity of NMR was exploited to
separately analyze water and ion dynamics for LiCl solutions in the bulk [
128
] and vari-
ous confinements [
129
,
130
]. Figure 3shows diffusion coefficients of LiCl–7H
2
O solution
obtained from
1
H and
7
Li SFG studies, which reflect the mobility of the water molecules
and lithium ions, respectively. This composition, which is close to the eutectic one, was
chosen to suppress crystallization and enable investigations in a broad temperature range.
For the bulk solution, it was found that the
1
H diffusivity was slightly larger than the
7
Li
diffusivity, and both exhibited a prominent non-Arrhenius temperature dependence typical
of many glass-forming liquids. When confining the LiCl–7H
2
O solution to a pristine silica
pore with a diameter of d= 3.0 nm, the difference of the
1
H and
7
Li diffusivities increased
to nearly an order of magnitude, and the temperature dependence again changed to an
Arrhenius behavior with an activation energy of E
a
= 0.26 eV [
129
]. It was argued that
these prominent confinement effects resulted from Stern layer formation at the usually
negatively charged silica walls. In a silica material with dye molecules grafted to the inner
surfaces, even smaller
7
Li diffusion coefficients were reported [
130
]. To rationalize this
finding, it was conjectured that these bulky functional groups protrude into the interior
of the pores and, in this way, form obstacles for the long-range transport of the highly
hydrated lithium ions.
Molecules 2024,29, 1669 8 of 26
Molecules 2024, 29, x FOR PEER REVIEW 8 of 27
finding, it was conjectured that these bulky functional groups protrude into the interior
of the pores and, in this way, form obstacles for the long-range transport of the highly
hydrated lithium ions.
Figure 3. 1H and 7Li diffusion coefficients of a LiCl-7·H2O solution in the bulk [128], in pristine silica
pores with a diameter of d = 3.0 nm [129], and in functionalized silica pores (APT + DYE, d = 5.8 nm)
[130]. For the functionalization, SBA-15 was functionalized with APTES via co-condensation, and
then further modified by adding 4-(5-methoxormatioridin-4-yl)thiazol-4-yl)benzoic acid dye mole-
cules, yielding a grafting density of ~1 dye molecule per nm2. The dashed line is a VFT interpolation
of the 1H diffusion coefficients of the bulk solution. The solid lines are Arrhenius fits of the 1H and
7Li diffusion coefficients in the pristine pores, yielding the same activation energy of Ea = 0.26 eV.
Despite their high practical relevance, a comprehensive understanding of the par-
tially frozen states of confined liquids is still lacking. For example, it is unclear whether
these two-phase states are ergodic, i.e., whether the crystalline and liquid phases exchange
molecules over time. 2H NMR spectroscopy yielded important information about the ice–
water equilibrium in silica nanopores below the melting temperature Tm [131]. The
method exploited the fact that the 1D 2H NMR line shape enabled a discrimination be-
tween molecules in the different phases; see Figure 4. Specifically, molecules of the less
mobile ice phase contributed a broad Pake paern (ν ≠ 0, in general), while those of the
more mobile water phase added a narrow Lorenian line (ν ≈ 0). Under such circum-
stances, 2D 2H NMR spectra provided access to an exchange between both fractions. Spe-
cifically, molecules that belonged to water prior to the mixing time (tm) exhibited ν1 ≈ 0
and became ice during this period of the 2D experiment so that they showed ν2 ≠ 0, pro-
duced a Pake spectrum along the frequency axis ν1 = 0. Vice versa, molecules that were a
part of the ice phase before the mixing time tm and of the water phase afterwards contrib-
uted such line shape along ν2 = 0. Together, a cross-like 2D spectral intensity along the
frequency axes indicated ice–water exchange during the mixing time tm. In a 2D 2H NMR
spectrum of D2O in SBA-15 pores measured at 220 K for a mixing time of tm = 5 ms, a cross-
like intensity along the frequency axes was clearly observed; [131] see Figure 4. Unlike in
a 1H NMR approach [132], where spin diffusion was faster and at the underlying process
of an exchange between the magnetizations of ice and water phases, the 2H NMR findings
could be traced back to an exchange of molecules between both phases. Explicitly, a de-
tailed analysis of the mixing-time dependence of the cross-like and other 2D spectral in-
tensities revealed that the residence time of a molecule in either phase was characterized
by an exchange time of 5.7 ms at 220 K. Thus, the ice–water equilibrium was highly dy-
namic, or, in other words, ergodicity restoration occurred relatively fast for the two-phase
state of water in nanoscale confinement. In this context, it should be mentioned that the
crystal structure of confined ice was discussed for years, and the existence of stack-
Figure 3.
1
H and
7
Li diffusion coefficients of a LiCl-7
·
H
2
O solution in the bulk [
128
], in pristine
silica pores with a diameter of d= 3.0 nm [
129
], and in functionalized silica pores
(APT + DYE,
d= 5.8 nm
) [
130
]. For the functionalization, SBA-15 was functionalized with APTES via co-
condensation, and then further modified by adding 4-(5-methoxormatioridin-4-yl)thiazol-4-yl)benzoic
acid dye molecules, yielding a grafting density of ~1 dye molecule per nm
2
. The dashed line is a VFT
interpolation of the
1
H diffusion coefficients of the bulk solution. The solid lines are Arrhenius fits
of the
1
H and
7
Li diffusion coefficients in the pristine pores, yielding the same activation energy of
Ea= 0.26 eV.
Despite their high practical relevance, a comprehensive understanding of the partially
frozen states of confined liquids is still lacking. For example, it is unclear whether these
two-phase states are ergodic, i.e., whether the crystalline and liquid phases exchange
molecules over time.
2
H NMR spectroscopy yielded important information about the
ice–water equilibrium in silica nanopores below the melting temperature T
m
[
131
]. The
method exploited the fact that the 1D
2
H NMR line shape enabled a discrimination between
molecules in the different phases; see Figure 4. Specifically, molecules of the less mobile
ice phase contributed a broad Pake pattern (
ν=
0, in general), while those of the more
mobile water phase added a narrow Lorentzian line (
ν≈
0). Under such circumstances,
2D
2
H NMR spectra provided access to an exchange between both fractions. Specifically,
molecules that belonged to water prior to the mixing time (t
m
) exhibited
ν1≈
0 and became
ice during this period of the 2D experiment so that they showed
ν2=
0, produced a Pake
spectrum along the frequency axis
ν1
= 0. Vice versa, molecules that were a part of the
ice phase before the mixing time t
m
and of the water phase afterwards contributed such
line shape along
ν2
= 0. Together, a cross-like 2D spectral intensity along the frequency
axes indicated ice–water exchange during the mixing time t
m
. In a 2D
2
H NMR spectrum
of D
2
O in SBA-15 pores measured at 220 K for a mixing time of t
m
= 5 ms, a cross-like
intensity along the frequency axes was clearly observed; [
131
] see Figure 4. Unlike in a
1
H
NMR approach [
132
], where spin diffusion was faster and at the underlying process of an
exchange between the magnetizations of ice and water phases, the
2
H NMR findings could
be traced back to an exchange of molecules between both phases. Explicitly, a detailed
analysis of the mixing-time dependence of the cross-like and other 2D spectral intensities
revealed that the residence time of a molecule in either phase was characterized by an
exchange time of 5.7 ms at 220 K. Thus, the ice–water equilibrium was highly dynamic,
or, in other words, ergodicity restoration occurred relatively fast for the two-phase state
of water in nanoscale confinement. In this context, it should be mentioned that the crystal
structure of confined ice was discussed for years, and the existence of stack-disordered ice
comprising interlaced layers of cubic ice (I
c
) and hexagonal (I
h
) ice was proposed in recent
studies [133,134].
Molecules 2024,29, 1669 9 of 26
Molecules 2024, 29, x FOR PEER REVIEW 9 of 27
disordered ice comprising interlaced layers of cubic ice (I
c
) and hexagonal (I
h
) ice was pro-
posed in recent studies [133,134].
Figure 4. 1D and 2D
2
H NMR spectrum of D
2
O in SBA-15 silica pores with a diameter of d = 5.4 nm
at 220 K [131]. A mixing time of t
m
= 5 ms was used to record the 2D spectrum. The dashed lines
indicate the frequency axes ν
1
= 0 and ν
2
= 0 of the 2D spectrum.
3.2. Octanol
In the following, it is described how solid-state NMR techniques combined with MD
simulations and thermodynamic techniques can be employed to investigate the structural
arrangement and dynamics of confined molecules, employing monohydric alcohol 1-oc-
tanol [92,135] as a model compound for surfactants. The techniques employed for these
investigations were originally developed for the study of confined isobutyric acid
[136,137].
The structural arrangement of the octanol molecules [92] can be investigated with the
application of
1
H/
13
C CP-MAS FSLG HETCOR experiments (see Figure 5, left panel). These
HETCOR experiments are sensitive to the magnetic dipolar interaction, thus providing
information about the distance between the carbon nuclei of the octanol and the carbon
and silica protons of the octanol and the silica host, respectively. By variation of the contact
time in these experiments, it is possible to sense how close various moieties of the confined
solvent molecules are to the pore surface. The analysis of the resulting spectra allows to
deduce neighborship relations, which can be interpreted in terms of orientations and ar-
rangements of confined molecules relative to the pore surfaces of the silica host (Figure 5,
right panel).
Figure 5. Left: Low temperature
1
H/
29
Si CP-MAS FSLG HETCOR obtained with 9 ms contact time
of an 80:20 mol% 1-octanol:water mixture confined in SBA-15. Referencing of the
1
H-dimension was
performed by employing the technique described in ref. [93]. Right: Graphical visualization of a
Figure 4. 1D and 2D
2
H NMR spectrum of D
2
O in SBA-15 silica pores with a diameter of d= 5.4 nm
at 220 K [
131
]. A mixing time of t
m
= 5 ms was used to record the 2D spectrum. The dashed lines
indicate the frequency axes ν1= 0 and ν2= 0 of the 2D spectrum.
3.2. Octanol
In the following, it is described how solid-state NMR techniques combined with MD
simulations and thermodynamic techniques can be employed to investigate the structural
arrangement and dynamics of confined molecules, employing monohydric alcohol 1-
octanol [
92
,
135
] as a model compound for surfactants. The techniques employed for these
investigations were originally developed for the study of confined isobutyric acid [
136
,
137
].
The structural arrangement of the octanol molecules [
92
] can be investigated with the
application of
1
H/
13
C CP-MAS FSLG HETCOR experiments (see Figure 5, left panel). These
HETCOR experiments are sensitive to the magnetic dipolar interaction, thus providing
information about the distance between the carbon nuclei of the octanol and the carbon
and silica protons of the octanol and the silica host, respectively. By variation of the contact
time in these experiments, it is possible to sense how close various moieties of the confined
solvent molecules are to the pore surface. The analysis of the resulting spectra allows
to deduce neighborship relations, which can be interpreted in terms of orientations and
arrangements of confined molecules relative to the pore surfaces of the silica host (Figure 5,
right panel).
Molecules 2024, 29, x FOR PEER REVIEW 9 of 27
disordered ice comprising interlaced layers of cubic ice (I
c
) and hexagonal (I
h
) ice was pro-
posed in recent studies [133,134].
Figure 4. 1D and 2D
2
H NMR spectrum of D
2
O in SBA-15 silica pores with a diameter of d = 5.4 nm
at 220 K [131]. A mixing time of t
m
= 5 ms was used to record the 2D spectrum. The dashed lines
indicate the frequency axes ν
1
= 0 and ν
2
= 0 of the 2D spectrum.
3.2. Octanol
In the following, it is described how solid-state NMR techniques combined with MD
simulations and thermodynamic techniques can be employed to investigate the structural
arrangement and dynamics of confined molecules, employing monohydric alcohol 1-oc-
tanol [92,135] as a model compound for surfactants. The techniques employed for these
investigations were originally developed for the study of confined isobutyric acid
[136,137].
The structural arrangement of the octanol molecules [92] can be investigated with the
application of
1
H/
13
C CP-MAS FSLG HETCOR experiments (see Figure 5, left panel). These
HETCOR experiments are sensitive to the magnetic dipolar interaction, thus providing
information about the distance between the carbon nuclei of the octanol and the carbon
and silica protons of the octanol and the silica host, respectively. By variation of the contact
time in these experiments, it is possible to sense how close various moieties of the confined
solvent molecules are to the pore surface. The analysis of the resulting spectra allows to
deduce neighborship relations, which can be interpreted in terms of orientations and ar-
rangements of confined molecules relative to the pore surfaces of the silica host (Figure 5,
right panel).
Figure 5. Left: Low temperature
1
H/
29
Si CP-MAS FSLG HETCOR obtained with 9 ms contact time
of an 80:20 mol% 1-octanol:water mixture confined in SBA-15. Referencing of the
1
H-dimension was
performed by employing the technique described in ref. [93]. Right: Graphical visualization of a
Figure 5. Left: Low temperature 1H/29Si CP-MAS FSLG HETCOR obtained with 9 ms contact time
of an 80:20 mol% 1-octanol:water mixture confined in SBA-15. Referencing of the
1
H-dimension was
performed by employing the technique described in ref. [
93
]. Right: Graphical visualization of a
feasible bilayer formation of 1-octanol (blue) inside the pore. Water molecules (blue) are concentrated
near the pore wall, as well as in the pore center (adapted from Kumari et al. [92]).
Molecules 2024,29, 1669 10 of 26
The dynamics of 1-octanol-d
17
in its bulk phase and confined in mesoporous silica
SBA-15 were also investigated by a combination of DSC experiments and
2
H-variable
temperature solid-state NMR (solid-echo and MAS NMR experiments) in the region of the
solid–liquid phase transition [
135
]. Compared to previous studies of smaller molecules
such as benzene [
50
,
138
], naphtalene [
139
], or bipyridine [
140
], where relatively broad
activation energy distributions were observed, the DSC results could be modeled by the
Kissinger model, employing a single activation energy of (313.6
±
2.1) kJ mol
−1
for the
bulk and a single activation energy of (172
±
17) kJ mol
−1
for the confined octanol [
141
].
The smaller activation energy of the confined octanol is reflected in its lower melting
point, which is approximately 38 K below the bulk value (e.g., 219.7 K versus 257.3 K at a
heating rate of 5 K min
−1
). The larger uncertainty in the value of the confined molecules is
already an indication of larger structural disorder and the coexistence of different species
with different melting points [
142
]. In order to gain further insights into the effects of the
confinement on the octanol, the
2
H-solid state NMR spectra of bulk and confined octanol are
compared in Figure 6A,B. Generally, the melting points observed in the NMR experiments
agree with those determined using the DSC measurements. Their line-shape analysis
(Figure 6C) reveals a superposition of different spectral components. The static spectra
below the melting point display a superposition of two Pake patterns with different width
and intensity. The broader Pake pattern shows C
Q≈
170 kHz just until the melting point, a
value typical for an immobile deuteron of a -CD bond [
143
]. The quadrupolar coupling
constant of C
Q≈
55 kHz of the narrower Pake pattern is characteristic for a CD
3
-group
moving around its C
3
-axis in a three-fold jump [
144
]. Accordingly, the two Pake patterns are
assigned to the methyl and methylene deuterons of the alkyl chain. An additional narrow
Lorentzian signal appears close to the melting point. This type of signal is characteristic
for the onset of melting and the presence of mobile molecules. Below temperatures of
195 K for bulk octanol-d
17
and 170 K for octanol-d
17
in SBA-15, an additional broad and
unstructured component is present in the static spectra. The latter is attributed to deuterons
whose motions are falling into the intermediate exchange regime and have relatively short
effective T
2
values [
145
,
146
]. This interpretation is corroborated with the
2
H MAS NMR
spectra, where only the large and the small Pake pattern are visible, plus the narrow
Lorentzian signal close to the melting point of the respective compound, but not the broad
unstructured component, due to its short T
2
. The distribution of activation energies for the
melting process is calculated from the mole-fractions of the spectral components employing
the Roessler model [147] (see ref. [135] for details).
The resulting distributions of activation energies of melting corroborated the results
of the Kissinger analysis. For the bulk octanol-d
17
, a narrow distribution of a well-ordered
crystalline solid was observed [
135
], and for the confined octanol a broad distribution
denoting a melting process involving species in a distribution of different environments
and activation energies, and possibly a distribution of rigidly and less-rigidly ordered
molecules [
71
,
142
], was observed. Moreover, the melting curves for the confined octanol-
d
17
exhibit clear deviations between the experimental and calculated curves towards lower
temperature, which are indicative for a non-Gaussian distribution of activation energies.
Additionally, this non-Gaussian distribution is illustrated by the numerical derivative of
the experimental data (Figure 6, right), especially for the confined octanol-d
17
under MAS
conditions. For this sample, a shoulder on the low-temperature flank of the main curve
is visible, indicating lower melting points for the molecules involved in the pre-melting
process compared to the full melting.
Molecules 2024,29, 1669 11 of 26
Molecules 2024, 29, x FOR PEER REVIEW 10 of 27
feasible bilayer formation of 1-octanol (blue) inside the pore. Water molecules (blue) are concen-
trated near the pore wall, as well as in the pore center (adapted from Kumari et al. [92]).
The dynamics of 1-octanol-d17 in its bulk phase and confined in mesoporous silica
SBA-15 were also investigated by a combination of DSC experiments and 2H-variable tem-
perature solid-state NMR (solid-echo and MAS NMR experiments) in the region of the
solid–liquid phase transition [135]. Compared to previous studies of smaller molecules
such as benzene [50,138], naphtalene [139], or bipyridine [140], where relatively broad ac-
tivation energy distributions were observed, the DSC results could be modeled by the
Kissinger model, employing a single activation energy of (313.6 ± 2.1) kJ mol−1 for the bulk
and a single activation energy of (172 ± 17) kJ mol−1 for the confined octanol [141]. The
smaller activation energy of the confined octanol is reflected in its lower melting point,
which is approximately 38 K below the bulk value (e.g., 219.7 K versus 257.3 K at a heating
rate of 5 K min−1). The larger uncertainty in the value of the confined molecules is already
an indication of larger structural disorder and the coexistence of different species with
different melting points [142]. In order to gain further insights into the effects of the con-
finement on the octanol, the 2H-solid state NMR spectra of bulk and confined octanol are
compared in Figure 6A,B. Generally, the melting points observed in the NMR experiments
agree with those determined using the DSC measurements. Their line-shape analysis (Fig-
ure 6C) reveals a superposition of different spectral components. The static spectra below
the melting point display a superposition of two Pake paerns with different width and
intensity. The broader Pake paern shows CQ ≈ 170 kHz just until the melting point, a
value typical for an immobile deuteron of a -CD bond [143]. The quadrupolar coupling
constant of CQ ≈ 55 kHz of the narrower Pake paern is characteristic for a CD3-group
moving around its C3-axis in a three-fold jump [144]. Accordingly, the two Pake paerns
are assigned to the methyl and methylene deuterons of the alkyl chain. An additional nar-
row Lorenian signal appears close to the melting point. This type of signal is character-
istic for the onset of melting and the presence of mobile molecules. Below temperatures of
195 K for bulk octanol-d17 and 170 K for octanol-d17 in SBA-15, an additional broad and
unstructured component is present in the static spectra. The laer is aributed to deuter-
ons whose motions are falling into the intermediate exchange regime and have relatively
short effective T2 values [145,146]. This interpretation is corroborated with the 2H MAS
NMR spectra, where only the large and the small Pake paern are visible, plus the narrow
Lorenian signal close to the melting point of the respective compound, but not the broad
unstructured component, due to its short T2. The distribution of activation energies for the
melting process is calculated from the mole-fractions of the spectral components employ-
ing the Roessler model [147] (see ref. [135] for details).
Molecules 2024, 29, x FOR PEER REVIEW 11 of 27
Figure 6. Comparison of 2H MAS NMR spectra and solid echo 2H spectra of (A) bulk octanol-d17 and
(B) octanol-d17 confined in mesoporous SBA-15 as a function of temperature. Exemplary line-shape
analysis of a static spectrum and calculated distributions of activation energies are shown in the
boom panels: (C) line-shape analysis of the solid echo 2H NMR spectrum of bulk octanol-d17 at 150
K (black) revealing (blue) narrow Pake paern of methyl deuterons; (orange) broad Pake paern of
methylene deuterons and unstructured component (green) of deuterons with short relaxation times.
(D,E): Distribution of activation energies for the melting process, calculated with the Roessler model
(blue: octanol-d17 in SBA-15, red: bulk octanol-d17, solid lines: MAS conditions, dashed lines: static
conditions). (Adapted from Döller et al. [135]).
The resulting distributions of activation energies of melting corroborated the results
of the Kissinger analysis. For the bulk octanol-d17, a narrow distribution of a well-ordered
crystalline solid was observed [135], and for the confined octanol a broad distribution de-
noting a melting process involving species in a distribution of different environments and
activation energies, and possibly a distribution of rigidly and less-rigidly ordered mole-
cules [71,142], was observed. Moreover, the melting curves for the confined octanol-d17
exhibit clear deviations between the experimental and calculated curves towards lower
temperature, which are indicative for a non-Gaussian distribution of activation energies.
Additionally, this non-Gaussian distribution is illustrated by the numerical derivative of
the experimental data (Figure 6, right), especially for the confined octanol-d17 under MAS
conditions. For this sample, a shoulder on the low-temperature flank of the main curve is
visible, indicating lower melting points for the molecules involved in the pre-melting pro-
cess compared to the full melting.
In studies of guests that are not fully isotope labelled, it is often necessary to employ
DNP enhancement for the characterization of the confined guest molecules. The efficacy
of these DNP enhancements in hyperpolarized NMR crucially depends on a good com-
patibility between the polarizing agent (PA), which is typically a dissolved organic mono-
or bi-radical, and the employed frozen solvents, often called the DNP matrix. A homoge-
neous distribution of the radicals, both in space and orientation, is necessary to achieve
good enhancements. In order to investigate this, Döller et al. [148] studied the behavior of
four different commercially available DNP polarizing agents confined in the non-ionic
model surfactant 1-octanol as analyte and established a novel relative quantification
method for the comparison of the proportion of the direct and indirect polarization trans-
fer pathway efficacies, which is able to take concentration effects into account. This study
revealed that the hydrophilicity of the PA is the key factor in the way polarization is trans-
ferred from the polarizing agent to the analyte.
3.3. Ethylene Glycol
As the starting point of the investigation of larger confined guest molecules who are
capable to perform several hydrogen bonding interactions, a study involving partially
deuterated ethylene glycol monomer (EG-d4) in its bulk phase, confined in SBA-15, as well
as APTES-modified SBA-15 was conducted. A combination of thermodynamic
Figure 6. Comparison of
2
H MAS NMR spectra and solid echo
2
H spectra of (A) bulk octanol-d
17
and
(B) octanol-d
17
confined in mesoporous SBA-15 as a function of temperature. Exemplary line-shape
analysis of a static spectrum and calculated distributions of activation energies are shown in the
bottom panels: (C) line-shape analysis of the solid echo
2
H NMR spectrum of bulk octanol-d
17
at
150 K (black) revealing (blue) narrow Pake pattern of methyl deuterons; (orange) broad Pake pattern
of methylene deuterons and unstructured component (green) of deuterons with short relaxation
times. (D,E): Distribution of activation energies for the melting process, calculated with the Roessler
model (blue: octanol-d
17
in SBA-15, red: bulk octanol-d
17
, solid lines: MAS conditions, dashed lines:
static conditions). (Adapted from Döller et al. [135]).
In studies of guests that are not fully isotope labelled, it is often necessary to employ
DNP enhancement for the characterization of the confined guest molecules. The efficacy of
these DNP enhancements in hyperpolarized NMR crucially depends on a good compati-
bility between the polarizing agent (PA), which is typically a dissolved organic mono- or
bi-radical, and the employed frozen solvents, often called the DNP matrix. A homogeneous
distribution of the radicals, both in space and orientation, is necessary to achieve good
enhancements. In order to investigate this, Döller et al. [
148
] studied the behavior of four
different commercially available DNP polarizing agents confined in the non-ionic model
surfactant 1-octanol as analyte and established a novel relative quantification method for
the comparison of the proportion of the direct and indirect polarization transfer pathway
efficacies, which is able to take concentration effects into account. This study revealed that
the hydrophilicity of the PA is the key factor in the way polarization is transferred from the
polarizing agent to the analyte.
3.3. Ethylene Glycol
As the starting point of the investigation of larger confined guest molecules who are
capable to perform several hydrogen bonding interactions, a study involving partially
deuterated ethylene glycol monomer (EG-d
4
) in its bulk phase, confined in SBA-15, as well
Molecules 2024,29, 1669 12 of 26
as APTES-modified SBA-15 was conducted. A combination of thermodynamic measure-
ments, solid-state NMR, and MD simulations were employed [
120
]. The phase behavior
(i.e., melting, crystallization, glass formation, etc.) of EG-d
4
in these three systems was stud-
ied using DSC (Figure 7). Through line-shape analysis of the
2
H ssNMR spectra recorded at
different temperatures, two signal patterns were identified for each of the three investigated
systems: a Lorentzian pattern, indicative of a liquid-like state, and a Pake pattern character-
istic of a solid-like state. Using a two-phase model, the distribution of activation energies
for the dynamic processes in each system was calculated. The spectra reveal an interesting
behavior of the confined EG. On the one hand, similar to the previously studied confined
1-octanol [
135
], the
2
H NMR spectra indicated the formation of a crystalline solid inside the
pores at reduced phase transition temperatures compared to unconfined EG. On the other
hand, DSC scans of the same samples, shown exemplarily in Figure 7, indicated the forma-
tion of an amorphous glass under rapid cooling. Interestingly, during the heating phase
of the DSC temperature cycle, the formed amorphous glass relaxes to form a crystalline
solid that later melts at further elevated temperatures. Moreover, the behavior of the EG
depends strongly on the surface modification of the SBA-15 host. On non-functionalized
surfaces, strong hydrogen-bonding interactions between EG and surface silanol groups
were revealed by causing a slowing down of EG dynamics [149].
In contrast, in the case of APTES-functionalized surfaces, where the polar surface-
silanol groups are, to a large extent, removed by the binding of the APTES, the surface is far
less polar and has a much lower capacity to form hydrogen bonds with EG-molecules [
150
].
This results in weaker interactions between pore-surface and EG molecules, placing more
importance to the EG–EG interactions and a higher tendency to form crystalline EG phases.
As a result, a substantially larger portion of EG in the pore remains solid after the first
melting event. These effects are schematically shown in the lower panel of Figure 7(see ref.
Haro et al. [120] for more details).
The rotational motion and translational diffusion of EG in mesoporous silica is studied
in detail in refs. [
120
,
149
]. To determine the pore-size dependence of these dynamics
unaffected by crystallization, a
1
H and
2
H NMR study focused on the fully liquid state
above T
m
[
149
]. Correlation times (
τ
) from
2
H SLR measurements indicated that the
molecular reorientation mildly slows down as the pore size decreases to d= 2.4 nm;
see Figure 8. A slightly more pronounced pore-size dependence was observed for the
diffusion coefficients (D) from
1
H SFG experiments. However, both reorientation and
diffusion exhibited similar temperature dependencies. In both cases, the non-Arrhenius
temperature behavior of the bulk liquid turned into Arrhenius behavior in sufficiently
severe confinements. In more detail, the Stokes–Einstein–Debye (SED) relation,
Dτ=2
9R2
H
was obeyed not only in the bulk liquid, where the experimental values of
τ
and Dindicated
a hydrodynamic radius of RH= 1.15 Å, but also in the silica pores. However, the different
pore-size dependencies of the rotational and translation motions manifested themselves
in reduced hydrodynamic radii, e.g., R
H
= 0.8 Å for the narrowest confinement with
d= 2.4 nm.
Correlation times
τ
of EG in lysozyme and elastin matrices obtained from
2
H
NMR SLR and STE studies resembled those in silica pores [
151
]. In particular, an Arrhenius
temperature dependence with an activation energy of E
a≈
0.6 eV was also observed. Thus,
the dynamics of EG does not depend on the exact chemistry of a confining framework, at
least as long as the latter allows for a formation of hydrogen bonds.
Molecules 2024,29, 1669 13 of 26
Molecules 2024, 29, x FOR PEER REVIEW 13 of 27
Figure 7. (A–C): DSC scans of EG confined in SBA-15 decorated with APTES at scan rates of 5 K
min
−1
(A), 10 K min
−1
(B), and 15 K min
−1
(C), respectively. At 5 and 10 K min
−1
, only partial freezing
is observed during cooling, as evidenced by the broad negative peak near 205 K. As the sample
relaxes during heating, the release of heat near 200 K indicates the formation of a crystalline solid
that then melts again in two steps near 200 K and 260 K, where the laer is presumed to indicate the
presence of EG not confined in the pores. At 15 K min
−1
(and higher rates, not shown) the small step
between 150–160 K during cooling indicates formation of a glass. (D–F):
2
H ssNMR spectra obtained
in the temperature range between 200 and 230 K for EG-d4 in SBA-15 (sample 2). (D)
2
H static NMR
experimental data, (E) fied
2
H static NMR spectra, and (F)
2
H MAS experimental data. (G,H):
Figure 7. (A–C): DSC scans of EG confined in SBA-15 decorated with APTES at scan rates of
5 K min−1(A),
10 K min
−1
(B), and 15 K min
−1
(C), respectively. At 5 and 10 K min
−1
, only partial
freezing is observed during cooling, as evidenced by the broad negative peak near 205 K. As the
sample relaxes during heating, the release of heat near 200 K indicates the formation of a crystalline
solid that then melts again in two steps near 200 K and 260 K, where the latter is presumed to indicate
the presence of EG not confined in the pores. At 15 K min
−1
(and higher rates, not shown) the small
step between 150–160 K during cooling indicates formation of a glass. (D–F):
2
H ssNMR spectra
obtained in the temperature range between 200 and 230 K for EG-d4 in SBA-15 (sample 2). (D)
2
H
static NMR experimental data, (E) fitted
2
H static NMR spectra, and (F)
2
H