Current Analytical Chemistry, 2005, 1, 3-27 3

Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy for

the Investigation of Supra- and Biomolecular Systems
Ingo Schnell*

Max-Planck-Institut für Polymerforschung, Postfach 3148, D-55021 Mainz, Germany

Abstract: After the fields of liquid- and solid-state NMR spectroscopy have developed largely independently
of each other over the past 50 years, materials of current interest call for new approaches which combine and
merge spectroscopic techniques from either side. A particular possibility to promote NMR spectroscopy into
this direction is provided by magic-angle sample spinning (MAS), which can be performed either at high
speeds in solid-state NMR or at moderate speeds in combination with high-resolution NMR. In this way, new
NMR methods and experiments have recently been developed, as will be discussed in this review.
In solid-state NMR, new and versatile experiments are based on coherent spin-states and coherent transfer
mechanisms, provided that fast MAS ensures efficient averaging of anisotropic interactions, and recoupling
schemes selectively re-introduce the interactions of interest. In high-resolution NMR, MAS can be borrowed
from solid-state NMR in order to ensure sufficient spectral resolution for samples with residual anisotropies or
heterogeneities (so-called HRMAS). In such HRMAS studies, recoupling techniques become applicable and
provide fully quantitative information on molecular structure and dynamics.
Keywords: NMR spectroscopy, Fast magic-angle spinning (MAS), High-resolution magic-angle spinning (HRMAS), Dipolar
recoupling, Double-quantum coherences, Polarisation transfer.

1. INTRODUCTION NMR spectra of such samples consist of very sharp

The past 60 years have seen the tremendous development
of nuclear magnetic resonance (NMR) from its discovery by
P URCELL, P OUND [1] and B LOCH [2] in 1945/46 to one of
the most powerful characterisation methods with widely
spread applications in all areas of science. In the course of
this development, three major areas of NMR research
activities emerged: liquid-state (or high-resolution) NMR
spectroscopy, solid-state NMR spectroscopy, and magnetic-
resonance imaging (MRI). When NMR methods and
techniques became more and more sophisticated and
specialised, the separation of these “principal” NMR areas
intensified such that they are largely considered as
independent research fields with little overlap.
Having this in mind, it is interesting to recall that
P URCELL and BLOCH (Nobel prize for Physics in 1952)
performed their first experiments independently of each other
on a solid and a liquid sample, respectively, namely paraffin
and water using proton (1H) resonances [1, 2]. In a sense,
NMR was discovered in parallel on solid and liquid
samples, and it was therefore clear from the very beginning
that NMR would be applicable to both, liquid and solid
phases of condensed matter. Despite this common start,
liquid-state and solid-state branches bifurcated in the
following, which was somewhat initiated by the discovery
of chemical shifts and spin multiplets between 1949 and
1951 by PROCTOR and WU [3], and PACKARD and
ARNOLD [4-6]. When molecules can tumble rapidly and
isotropically in a dilute solution, the interactions that
determine NMR spectra are reduced to their isotropic parts,
i.e., isotropic chemical shifts and so-called J-couplings.
*Address correspondence to this author at the Max-Planck-Institut für
Polymerforschung, Postfach 3148, D-55021 Mainz, Germany; Tel: +49
6131 379-240; Fax: +49 6131 379-100; E-mail: schnelli@mpip-
mainz.mpg.de
resonance lines which provide detailed information on the
solubilised molecules. The chemical shift allows nuclei of
the same isotope to be sensitively distinguished according to
their individual chemical environment, while the J-couplings
reflect the network of chemical bonds. These effects form the
basis for high-resolution NMR spectroscopy, whose
development into a uniquely powerful characterisation
method was particularly driven by the ongoing quest for
homogeneous, superconducting high-field magnets, the
introduction of pulsed Fourier-transform spectroscopy by
ERNST in 1971 [7, 8] and the extension to multidimens-
ional spectra proposed by JEENER[9] and pioneered by
ERNST in the 1970s [10, 11].
Over the past 30 years, high-resolution NMR at high
magnetic fields has not only become a routine method for
molecular structure determination in Chemistry and
Pharmaceutics, but has also gained outstanding importance
for the characterisation of complicated biological systems, in
particular proteins. Rumour has it that BLOCH liked to say
it was him who he did the first biological NMR experiment
in 1946 when he put his finger into the probe coil of his
nuclear induction apparatus and received a strong 1H NMR
signal from it. However, the first serious high-resolution
NMR studies of living systems began with the phosphorus
spectra of intact red blood cells by M OON and R ICHARDS in
1973 [12]. Subsequently, the new field of “biological NMR”
was particularly promoted by the work of WüTHRICH on
protein structure determination [13], which has become an
important subject of world-wide research efforts in numerous
high-field NMR centres. The versatility and significance of
high-resolution NMR is also reflected by two Nobel prizes
in Chemistry awarded to ERNST [14] and WüTHRICH [15].
in 1991 and 2002, respectively.
While liquid-state NMR has developed into a routine
characterisation method with numerous applications in

1573-4110/05 $50.00+.00 © 2005 Bentham Science Publishers Ltd.

4 Current Analytical Chemistry, 2005, Vol. 1, No. 1
diverse fields of science and medicine, one might be tempted
to say that solid-state NMR spectroscopy has remained a
delicate, though powerful, tool for specialists. Technically,
this is predominantly due to the strong anisotropic
interactions present in solid materials, which make higher
demands on NMR techniques and give rise to broad and
complex lines in solid-state NMR spectra. A groundbreaking
idea how to deal with these anisotropic interactions came
from ANDREW [16] and LOWE [17] who independently
invented magic-angle spinning (MAS) already in the late
1950s. Making use of the orientation dependence of
anisotropic interactions relative to the magnetic field (B0),
the sample is spun rapidly about an axis which is inclined at
the so-called magic angle relative to B0. In this way,
anisotropic parts of interactions are averaged, provided that
the spinning frequency exceeds the strength of the
interaction. Consequently, the line narrowing performance is
in many cases limited by the technically achievable MAS
frequency, which is why it was not until the 1990s that
sufficient spinning frequencies [18, 19] paved the way for
widely spread applications of solid-state MAS NMR
experiments including protons [20, 21].
In 1998, equipment supporting MAS frequencies in
excess of 30 kHz became commercially available. In
combination with the technical progress of radio-frequency
circuits and pulse generators, solid-state NMR equipment
has meanwhile reached a status which allows for an efficient
handling and averaging of anisotropic interactions (apart
from strong quadrupolar couplings, though). It should
therefore be possible to design solid-state NMR experiments
by incorporating techniques and concepts from liquid-state
NMR, which were so far supposed to be dedicated to the
liquid state only. Thus, it is time for solid-state NMR
spectroscopists to remind themselves of the wealth of liquid-
state NMR techniques.
Interestingly, the same also applies in reverse. In the last
years, high-resolution NMR has taken substantial advantage
of slightly aligning the molecules under inspection, for
example by dissolving them in a dilute liquid-crystalline
medium. In this way, small residual anisotropic interactions
are re-introduced and become available for structural
investigations of (large) molecules, in particular proteins.
This strategy, pioneered by B AX [22, 23] and PRESTEGARD
[24, 25] has attracted considerable interest in the biological
NMR community and undoubtedly bears further potential
for high-resolution NMR applications in general. Another
field of current interest is focused on materials in which the
molecules of interest can move rapidly, but not fully
isotropically, because they are attached to an immobile
species. In addition to the motional restriction, such
inhomogeneous mixtures of mobile and immobile
components usually exhibit large magnetic susceptibility
gradients which broaden the resonance lines and destroy the
spectral resolution. In order to remove these unwanted
susceptibility gradients, together with other residual
anisotropies, and to restore spectral resolution, MAS has
recently been successfully introduced into high-resolution
NMR spectroscopy, and the approach has been termed
“HRMAS” spectroscopy [26-28]. Considering HRMAS and
the strategy of partial alignment outlined above, it is
palpable that liquid-state NMR is about to take up concepts
from solid-state NMR.
Ingo Schnell
In view of this background, the work described in this
review aims at a thorough methodological combination of
liquid-state and solid-state NMR by merging concepts from
either side. Based on this strategy, new MAS NMR
methods and experiments have been developed for organic
supramolecular and biomolecular systems. These classes of
materials not only gain more and more importance in current
research, but also clearly show that future NMR methods
will need to be applicable to samples which are neither rigid
solids nor isotropic liquids. Thus, it is about time that
NMR spectroscopy overcomes the “traditional” separation
between liquid state and solid state and brings its common
roots back to its mind.
2. INTRODUCING LIQUID-STATE NMR CONCEPTS
INTO SOLID-STATE MAS NMR
In NMR, interactions become orientation dependent due
to the presence of a strong external magnetic field, B0. This
field gives rise to the famous ZEEMAN interaction, which
usually exceeds all internal interactions by several orders of
magnitude. Consequently, the latter are treated as a first-
order perturbation of the ZEEMAN interaction in the so-called
“rotating frame” [29], and the B0 field effectively represents
the quantisation axis of the spin system. Based on this, two
classes of interactions can be distinguished, namely
anisotropic and isotropic interactions, which exhibit and do
not exhibit, respectively, an orientation dependence relative
to B0.
The fundamental difference between solid- and liquid-
state NMR can also be traced back to this distinction, as it
arises from the absence and presence, respectively, of rapid
isotropic motions of the molecules under investigation. In
this context, “rapid” refers to the timescale of the
interactions which are averaged and, thus, reduced to their
isotropic properties by the motion of the molecules.
Typically, NMR interactions give rise to energy level
differences and transition frequencies of the order of ∆E/h =
100 ... 2·104Hz [29], when large quadrupole couplings are
ignored. Correspondingly, molecular motions need to occur
isotropically with correlation times of τ C < 10µs, in order to
ensure complete averaging of anisotropic properties. This is
usually easily fulfilled in diluted solutions of relatively
small molecules. Therefore, this class of samples constitutes
the classical field of application of liquid-state (or high-
resolution) NMR spectroscopy.
In solids, obviously, molecules do not undergo such
rapid and isotropic motions, and NMR has to deal with
pronounced effects arising from anisotropic interactions
[30, 31], such as anisotropies of the chemical shielding
(CSA) as well as dipole-dipole and quadrupole couplings,
just to mention the most prominent ones. These effects
dominate typical solid-state NMR spectra and make them
appear fundamentally different from high-resolution spectra
obtained from solutions. Thus, the transfer of concepts from
liquid-state NMR into its solid-state counterpart relies on
techniques that allow for an efficient averaging or, at least,
handling of anisotropic properties. In the past five years,
significant advances into this direction have become possible
by the technical progress in the field of fast magic-angle
spinning (MAS), which will be discussed in the first part of
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy
this section. Consecutively, ideas will be introduced which
enable the selective restoration, i.e. “recoupling”, of certain
anisotropic properties. In this way, coherent spin states can
be generated, and numerous experimental schemes can be
designed, in which the strength of the anisotropic interaction
of interest becomes measurable through the coherent
evolution of spin states. Measurement strategies of this kind
form the basis of a whole wealth of experiments in high-
resolution NMR spectroscopy, and the scope of this work is
to introduce these concepts into solid-state NMR
experiments.
2.1. The Quest for Spectral Resolution
The impact of high-resolution NMR spectroscopy in
modern science can largely be attributed to its ability to
sensitively distinguish different species of a nucleus, such as
1H, 13C, 15N etc., due to their chemical shifts. Being an
isotropic property, these shifts become very well resolved
when the molecules tumble rapidly and isotropically in
solution. In solids, resonance lines are severely broadened by
anisotropic interactions, which pose the first and most
important obstacle to be overcome before considering
analogies between liquid- and solid-state NMR.
Isotropic and Anisotropic Interactions
The concepts of liquid-state NMR spectroscopy are
largely built on the two isotropic interactions, namely the
isotropic chemical shifts, which provides spectral resolution,
and the indirect spin-spin (or J-) couplings, which reflect the
network of chemical bonds [29]. In principle, these two
interactions do not hamper spectral resolution, as they lead
to sharp resonance lines, although multiplet splittings
arising from J-couplings might confuse and complicate the
interpretation of the spectra. To accomplish spectral
resolution in solid-state NMR spectra, line broadening
effects of anisotropic interactions need to be effectively
removed, while retaining the isotropic chemical-shift
information.
Basically, this work is restricted to spin-1/2 nuclei, in
particular to 1H, 13C and 15N due to their fundamental role
in biomolecules. From a methodological point of view, this
reduces the relevant anisotropic interactions to homo- and
heteronuclear dipole-dipole couplings and chemical shifts,
while quadrupole couplings can be ignored. With respect to
the size of the interactions, this means a considerable
simplification, because quadrupole couplings can be very
strong and reach in some cases the same order as the
ZEEMAN interaction. Then, they can no longer be considered
within the so-called secular approximation as a “weak” first-
order perturbation of the dominating ZEEMAN interaction,
and it becomes technically very demanding (if not
impossible) to reduce them efficiently [32, 33]. Therefore, it
is rather difficult to propose general concepts of merging
liquid- and solid-state NMR including nuclei with strong
quadrupole couplings. An important exception is the
deuteron, 2H, because its typical quadrupole couplings
hardly exceed 200 kHz and are thus only by a factor of < 10
stronger than the strongest dipole-dipole couplings.
Therefore, they can be treated fully analogously to dipole-
dipole and chemical-shift interactions. Moreover, deuterons
Current Analytical Chemistry, 2005, Vol. 1, No. 1 5
are of practical importance in NMR research, because they
can be used to selectively replace protons at sites of interest.
In the secular approximation, the Hamiltonians of the
relevant anisotropic NMR interactions can be expressed in
terms of operators, which are components of rank two and
order zero of irreducible spherical tensor operators [34] where
λ denotes the type of interaction, while i refers to the spin i
in case more than one is involved. For CSAs, hetero- (DIS )
and homonuclear (Dij) dipole-dipole and quadrupole
couplings (Q) one finds[30, 31] for the space parts (given in
the principal axes system of the interaction, PAS), and for
the spin parts. The space parts consist of the respective
coupling constants plus scaling factors, which deviate from
the usual conventions [30] because normalisation factors of
the spin parts have been incorporated for reasons of clarity.
The spin parts differ considerably for the interactions, but
analogies are obvious between CSA and heteronuclear
dipole-dipole couplings as well as between homonuclear
dipole-dipole and quadrupole couplings.
Hλ = Σ A λ2,0, i T 2,0
λ, i
, (1)
i
= δ , A DIS ,P AS = 2DIS , A D I J , PA S = 3D, ,ij
CSA .PA S
A 2,0 2,0 2,0
Q ,P A S
= 3
δQ (2a)
and A 2,0 2
(i) (j)
= 2IZ S Z , T ij = 3IZ I Z I (i) I (j) ,
CSA D D
T 2,0 = IZ , T IS
2,0 2,0
Q
and T 2,0= 3I Z I Z I I (2b)
Averaging Techniques
To remove an anisotropic interaction, i.e., to average the
respective Hamiltonian (see Eq. 1), it is sufficient to cancel
either the spin part or the space part. Formally, all averaging
procedures involve a transformation of reference frames into
the laboratory system (LAB) with B0 || z. In other words, the
orientation dependences of the anisotropic interactions result
from the transformation of the reference frames into the LAB
frame. Generally, concerning the space part, this
transformation occurs from the PAS to the LAB frame and
can be expressed as a rotation using the reduced
elements d (2)
m,0 (β) of WIGNER rotation matrices [30,31],
where α and β are the polar and azimuthal angles specifying
the orientation of the interaction tensor. Eq. 3 can be applied
analogously to the spin operators. In spin space, rotations
are effectively accomplished by pulsed radio-frequency (RF)
fields that are applied at the resonance (i.e. LARMOR)
frequency of the respective nucleus. Then, the angles α and
β represent the phase and the flip angle, respectively, of the
RF pulses [30].
λ, P AS
A λ,2,0LA B = Σ A 2,m
-ima (2)
e d m.0 (β) (3)
m
All averaging schemes, acting either in real space or in
spin space, are formally based on Eq. 3. To achieve
averaging, α is effectively made time dependent by
introducing ω t. This results in harmonic oscillations,
exp(+-imω t), which average to zero over full oscillation
periods. In addition, β is chosen such that the static non-
oscillating component (m = 0) disappears as well, which is
6 Current Analytical Chemistry, 2005, Vol. 1, No. 1
the case for the so-called “magic angle” βm = 54.7°.
Therefore, the per-formance of such averaging techniques
crucially depends on the precision of the magic-angle
adjustment and on the frequency ω . With respect to the
latter, it is important to recall the distinction between
homogeneous and inhomogeneous line broadening
phenomena, as introduced by MARICQ and WAUGH [35].
Theoretically, inhomoge-neous interactions, such as CSAs,
heteronuclear dipole-dipole or quadrupole couplings, are
fully averaged over one oscillation period irrespective of the
frequency, while the homogeneous character of homonuclear
dipole-dipole couplings requires the period of the averaging
cycle to be short enough to prevent the formation of a
dipolar-coupled multi-spin network. In practice, pure
inhomogeneous line broadening is rarely found, because
homogeneous effects are encountered as soon as homonuclear
dipole-dipole couplings are present in the spin system. This
is usually the case, as most systems of interest contain a
significant amount of protons with multiple strong dipole-
dipole couplings among them. Therefore, the averaging
frequency ω , in general, needs to be high enough to
efficiently suppress proton-proton dipolar interactions, i.e.,
ω /2π ≥ 25 kHz as a rule of thumb.
As mentioned above, the averaging procedures can
formally be applied in real space as well as in spin space. In
real space, magic-angle spinning (MAS) is the most
commonly used technique which involves a rapid mechanic
rotation of the sample about an axis that is inclined at the
magic angle relative to the B0 field [16, 17]. In spin space,
the same type of rotation is accomplished by the application
of suitable RF fields. Presently, maximum performance is
provided by LEE-GOLDBURG pulse schemes [36], their
frequency-switched [37] and phase modulated [38] variants
Fig. (1). Spectral resolution achievable at various MAS frequencies in the 1 H solid-state MAS spectrum of a crystalline benzoxazine
dimer (the chemical structure of which is depicted in the inset).
Ingo Schnell
as well as by numerically optimised pulse schemes [39].
Such continuous RF schemes are advantageous insofar as
high averaging frequencies of the order of 100 … 200 kHz
can be realised experimentally. However, the observed
isotropic chemical shifts are effectively scaled by factors
< 1 3 [30], which significantly reduces the separation of
the resonance lines in the spectra. Moreover, in order to
detect signal or to apply other RF pulse sequences at the
same time, the RF averaging schemes need to be interrupted,
which usually reduces their efficiency and severely
complicates the RF pulse programming. MAS, in contrast,
only allows for averaging frequencies of the order of 30 kHz
(at the present stage), but it leaves the RF side of a NMR
spectrometer completely free and available for pulse
sequences of NMR experiments. In a sense, MAS mimics
the rapid isotropic motions the molecules undergo in dilute
solutions, and paves the way for the introduction of liquid-
state techniques into solid-state NMR. Due to this
conceptual analogy, the work described here concentrates on
MAS and its ability to average anisotropic interactions.
Resolution Enhancement by Magic-angle Spinning
For most (bio)organic spin-1/2 systems, the spectral
resolution achievable in solid-state NMR is determined by
the ability to suppress dipole-dipole couplings to and among
protons, i.e., to “decouple” protons. Therefore, simple 1H
NMR spectra recorded at different MAS frequencies can
serve to illustrate the decoupling performance of MAS,
which is shown in (Fig. 1) using a crystalline benzoxazine
dimer as an example [20].
It is obvious from (Fig. 1) that, for 1H spectra of rigid
solids, MAS needs to be applied at frequencies ≥ 25 kHz. At
higher spinning frequencies, the resolution still increases
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy
gradually, but the maximum resolution is not only
determined by the technically available MAS frequencies,
but also by the material itself. When the crystallinity, i.e.,
the structural order, of a material is not perfect, the local
distributions of electrons around the nuclei are affected by
local disorder or variations of the packing structure. These
phenomena give rise to dispersions of the chemical shifts
and, thus, to an inherent line width which cannot possibly
be reduced by averaging techniques. Therefore, in amorphous
samples, maximum 1H resolution is already achieved at
MAS frequencies of about 30 … 35 kHz, and faster spinning
would not further increase the spectral resolution.
Despite this limitation, the quest for higher MAS
frequencies [40] is still perfectly legitimate, because the
decoupling of the dipolar-coupled proton network is not
only required to enhance resolution in 1H spectra, but also to
reduce transverse relaxation phenomena of other nuclei X,
such as 13C or 15N. Usually, such heteronuclear decoupling
between X and 1H is accomplished by powerful RF pulse
schemes [41-44] which are applied on the 1H channel while
observing the X signal. This simple and efficient approach,
however, provides full decoupling between 1H and X nuclei,
which is no longer feasible when correlated spin states of the
form X-H are to be observed. Then, a decoupling scheme is
required which does not destroy the correlated X-H spin state
(as brute-force RF decoupling would do), but efficiently
prevents transverse relaxation. This issue gains importance
for so-called inverse detection schemes, which are discussed
in the following section.
2.2. The Creation of Coherent Spin States
Aiming at the removal of anisotropic interactions, the
previous considerations were mainly concerned with
restoring the information on isotropic chemical shifts. There
is, however, another isotropic interaction, namely the
indirect spin-spin (or J-) coupling, which is of outstanding
importance in high-resolution NMR spectroscopy, because it
enables the correlation of spin states and the creation of
multi-spin coherences [10, 29]. In this way, J-couplings
form the basis for virtually all multidimensional NMR
experiments in the liquid state.
In the presence of a J-coupling, represented by a
Hamiltonian of the form HJ = πJ . Iz S z , a two-spin system
(I,S) oscillates between a correlated (IYS Z) and an
uncorrelated, transverse spin state (IX), which can be
expressed as [10] where J is the J-coupling constant. When
observing the I-spin signal, the cosine-modulation gives rise
to the well-known multiplet splittings in the spectra [10,29].
In view of multidimensional NMR, however, the correlated
spin state is the more interesting one, as it provides a means
to coherently transfer states from one spin to another. In the
case of J-couplings, such transfer processes occur between
chemically bonded spins, which means that spin states are
propagated through the molecule via the network of chemical
bonds. Ideally, due to the coherent nature of the processes
and the (formal) absence of relaxation, no information is lost
in the course of the spin system’s evolution. This direct link
between spectroscopic properties and molecular structure
represents one of the fundamental reasons for the success and
versatility of high-resolution NMR spectroscopy.
Current Analytical Chemistry, 2005, Vol. 1, No. 1 7
πJ IZ SZ
IX IX cos(πJt) + 1Y SZ sin (πJt), (4)
In view of this successful concept, the question arises
whether and to what extent the idea of generating and
transferring coherent two- or multi-spin states would also be
applicable under solid-state conditions. Obviously, in
solids, anisotropic interactions and the associated line-
broadening and relaxation effects represent the main obstacle.
However, if averaging techniques can reduce these
anisotropies efficiently, and unravel the isotropic
information, similar experiments become possible in solid-
state NMR as well. For this purpose, the recovery of the
isotropic chemical shifts is no doubt essential, but with
regard to the J-couplings, there is an alternative way to
couple nuclei in solid samples, namely via dipole-dipole
couplings. Having said this, three major differences between
dipole-dipole and J-couplings are obvious:
(i) J-couplings are electron-mediated indirect spin-spin
couplings which are, as a consequence of this,
restricted to nuclei linked by chemical bonds. Dipole-
dipole couplings, in contrast, act directly through
space and only depend on the distance, r, between the
nuclei according to 1/r3. Therefore, they are not
directly related to the network of chemical bonds,
which on the one hand hampers the investigation of
the chemical structure of molecules, but on the other
hand enables the detection of intermolecular contacts.
(ii) For the nuclei discussed here, J-couplings are
basically fully isotropic interactions, while dipole-
dipole couplings are fully anisotropic and do not
possess an isotropic component at all.
(iii) The two types of couplings also differ considerably
with respect to coupling sizes: Heteronuclear 15N-1H
and 13C-1H J-couplings are of the order of
80 … 150 Hz, and typical homonuclear 1H-1H J-
couplings do not exceed 15 Hz. The dipolar
counterparts are of the order of 10 … 20 kHz, and
thus larger by factors of about 102 … 10 3. This
comparison, however, only holds in the absence of
(molecular) motions, because dipole-dipole couplings
are averaged and reduced by molecular dynamics
and/or mechanical sample rotation (MAS) due to their
anisotropic nature.
At first sight, the significant differences between dipole-
dipole and J-couplings seem to suggest that a transfer of
experimental schemes from liquid- to solid-state NMR relies
on techniques which provide access to J-couplings in solids.
As J-couplings are very weak compared to typical
anisotropies encountered in solids, powerful averaging
techniques are required, which make quite high demands on
the technical equipment in terms of available RF power and
RF stability. Such approaches exist, pioneered by LESAGE
and EMSLEY [45-49], but at the present stage, only RF-
based decoupling schemes in combination with MAS are
capable of providing sufficient averaging of anisotropic
interactions. At the current limit of spinning frequencies,
MAS alone does not suffice to unravel J-couplings in solids.
Nevertheless, the solid-state NMR methods presented
herein do not consider J-couplings, but are solely based on
dipole-dipole couplings because of two reasons: Firstly,
8 Current Analytical Chemistry, 2005, Vol. 1, No. 1
dipole-dipole couplings are much stronger, and it is therefore
much easier to use them in the experiments rather than
battling against them with sophisticated pulse sequences.
Secondly, dipole-dipole couplings are efficiently averaged by
fast MAS, but can conveniently be re-introduced by so-
called recoupling pulse sequences, which are discussed
below. In this way, the experiments gain an important
degree of flexibility, because dipole-dipole couplings can be
switched on and off during the experiments basically at will.
During recoupling periods, coherent dipolar two- and multi-
spin states can be generated, and concepts of high-resolution
NMR experiments can be adapted to solid-state MAS
conditions, including coherence transfers. Even more, the
peculiarities of MAS and recoupling procedures allow for an
extension of such concepts towards sensitive measurements
of couplings. As a starting point, however, the effect of
MAS on anisotropic interactions needs to be considered first
in more detail.
Rotor-Modulated Interactions
Under MAS, the effective Hamiltonian of an anisotropic
interaction represented by a second-rank tensor can be
approximated using the zeroth-order average Hamiltonian
[30, 50]
λ λ
t1 λ LA B
A ,
1
t0 - t1 t∫ 2,0
HM AS (t0 ; t1) = T 2,0 (ωRt)dt
0
λ
= T2,0 t - t Φ λ (t0 ; t1)
1
(5)
0 1
λ λ
Α L, M and Τ L,M represent the spatial part and the spin
part (see Eq. 2a and 2b) of the interaction λ, and are given as
components of irreducible spherical tensor operators of order
M and rank L [30,34]. The first term with L = M = 0 is
neglected, as it represents the isotropic part of the
interaction, which is zero for dipole-dipole and quadrupole
interactions: Α0.0 = ΑQ λ
0.0 = 0 The integrated phase Φ (t 0;t 1)
D
contains the spatial part of the interaction, which is
modulated by the sample rotation [30] where Α λ.P 2.0
AS
is the
relevant tensor component in the PAS of the interaction λ
(as given explicitly in Eq. 2a), and ω R denotes the MAS
frequency. β and γ are the azimuthal and polar angles that
describe the orientation of the interaction tensor in the rotor
frame (see Fig. 2a). γ can also be viewed as the initial rotor
phase, as the time-dependent rotor phase, ω Rt, is always
added to this angle. In Eq. 6, it is assumed that the
interaction is symmetric, which greatly simplifies the
description in the case of dipole-dipole couplings and many
2H quadrupole couplings. For CSAs, asymmetries need to
be taken into account, which requires the tensor components
ΑCSA
2,+2
. PAS
to be included. Using the above expressions for
an averaged Hamiltonian under MAS, an initial state of
transverse magnetisation, ρ0 = Ix (ρ 0 = I x(i ) + Ix(j) ) for a spin
pair ij with a homonuclear dipole-dipole coupling), evolves
under an anisotropic interaction according to where W
denotes the spin state into which the system evolves under
the interaction λ, i.e. W = IY, W = 2IYS Z,
W = 2(IY(i) I(j) I(i) I(j) ) and W = 2(IYIZ + IZIY) for
Z Z Y
CSA, hetero- and homonuclear dipole-dipole and quadrupole
coupling, respectively. Eq. 7 represents a general form of
Eq. 4, which has been adapted to rotor-modulated
anisotropic interactions, as encountered in solids under MAS
conditions.
Ingo Schnell
λ.PAS
A 2,0
Φλ (t0;t1) = 1 sin2 β(sin (2ω t +2γ) -sin(2ω t + 2γ)
ωR 2
R1 R0
2 sin 2β (sin (ωRt1 + γ) - sin (ωRt0 + γ) (6)
λ
HMAS
ρ0 ρ0 cosΦ λ + W sin Φ λ (7)
Recoupling Schemes
The MAS NMR experiments considered in this part of
the work employ sequences of strong RF pulses separated by
delays, during which the spin system evolves under the
internal interactions. These pulse sequences are often referred
to as “laboratory frame” sequences, because the quantisation
axis is given by the static magnetic field B0 along the
laboratory z-axis. “Rotating frame” pulse sequences, in
contrast, are characterised by a continuous application of RF
fields, so that the spin system evolves no longer freely but
in the presence of transverse RF fields (see for example the
helical pulse sequences discussed in Chapter 3).
The principle of laboratory-frame recoupling pulse
sequences is well established [51, 52] and can easily be
understood by considering the rotor modulations which are
introduced by the sample rotation (see Eq. 6). The spatial
parts of the interaction Hamiltonians are modulated by the
MAS frequency ω R and twice the MAS frequency 2ω R. For
illustration, the respective oscillations are depicted in (Fig.
2b). By averaging – in other words: integrating – over a full
rotor period, the modulated interactions vanish because
positive and negative parts of the oscillations cancel each
other. This cancellation effect can also be seen from Eq. 4,
when t1 – t0 = n⋅τ R and n is an integer number. An obvious
way to avoid this cancellation is to invert the sign of the
spin part by a pulse sequence simultaneously to the sign
changes imposed on the spatial part by MAS. Effectively,
recoupling is accomplished by inducing a “counter-rotation”
in spin space relative to MAS rotation in real space. As there
are two modulation frequencies, ω R and 2ω R, the sign
change in the spin part can be performed in intervals of
either τ R/2 or τ R/4 [93].
The type of RF pulses required for this sign inversion
λ
depends on the nature of the spin part T2,0 of the respective
interaction λ (see Eq. 2b). For interactions depending
linearly on a spin (i.e., CSA and heteronuclear dipole-dipole
couplings), the sign inversion is straightforwardly achieved
by π pulses. The approach of utilizing π-pulse trains with
pulse spacings of τ R/2 was first realised in the REDOR
(rotational echo, double resonance) experiment, which was
initially introduced by SCHAEFER and GULLION [53-56] in
order to recouple the heteronuclear dipole-dipole coupling
between two kinds of low-γ nuclei. Such REDOR-type π-
pulse trains are also commonly used for CSA recoupling
[57, 58] or for recoupling 1H-X heteronuclear dipole-dipole
couplings, 51, 52, 55, 56] where X denotes a low-γ spin-1/2
nucleus. Interactions with a bilinear spin dependence (i.e.,
homonuclear dipole-dipole and quadrupole couplings) can be
recoupled in an analogous fashion by the so-called Back-to-
Back (BABA) pulse sequence [21, 59]. In this sequence,
each π-pulse of REDOR-type recoupling is replaced by a
pair of π/2-pulses with orthogonal phases. Effectively, these
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy Current Analytical Chemistry, 2005, Vol. 1, No. 1 9

Fig. (2). (a) Geometric relations under MAS conditions, including the definition of the angles β and γ. They relate the main axis of a
tensor representing a symmetric anisotropic interaction to the rotor. (b) Modulations of an anisotropic interaction under MAS (top)
and pulse sequences for recoupling interactions which depend on the nuclear spin in a bilinear (middle) or linear fashion (bottom).
The pairs of 90° RF pulses (middle) recouple homonuclear dipole-dipole couplings and (weak) quadrupole couplings, while the 180°
pulses (bottom) recouple heteronuclear dipole-dipole couplings or CSAs.

λ
pairs of π/2-pulses invert the sign of the spin part T 2.0 of
the homonuclear dipole-dipole and quadrupole coupling.
In Fig. 2b, the principles of BABA and REDOR
recoupling are illustrated [93]. Clearly, pulse trains with
pulse spacings of τ R/2 recouple only the ω R-modulated
terms, while pulse trains with pulse spacings of τ R/4
recouple only the 2ω R-modulated terms. Henceforth, these
two types of recoupling will be referred to as “ω R-
recoupling” and “2ω R-recoupling”. [93]. These experiments
give rise to the following integrated phases for axially
symmetric interactions [93],
2 Φ λ (t;t + τR /2) = Φ ωλ (t)
R defines the starting time of the recoupling period in the
Aλ,2,0PA S
= ω 2 sin 2β sin (ωRt + γ) (8a)
R
λ
4 Φ λ (t;t + τ R /4) = Φ2ω (t) =
R
λ, PA S
A 2,0
= ω sin β sin (2ωR t + 2γ)
2 (8b)
R spin state, when the (recoupled) interaction is not a spin-spin
The factors 2 and 4 have been introduced to make the
total timing of both expressions equal (i.e. τ R) and to allow
for a comparison of their absolute scaling. Then, it becomes
obvious that the two expressions differ not only in their
orientation dependencies (sin 2β and sin 2 β), but also in their
amplitudes. Under ω R-recoupling, the effective Hamiltonian
is larger by a factor of 2 as compared to 2ω R-
recoupling, which means that, theoretically, 2ω R-recoupling
needs to be applied longer by a factor of, 2 to achieve
the same signal amplitude as ω R-recoupling. Consequently,
ω R-recoupling is in general favourable with respect to the
recoupling efficiency, and 2ω R-recoupling is disadvanta-
geous unless sample orientation becomes an issue, which
will be discussed below.
Two-Spin Correlations and Double-Quantum Coherences
The recoupling procedures outlined above allow for the
generation of states W according to Eq. 7, where the
integrated phases Φ λ are to be replaced by the recoupled
λ
phases NR Φ nω R (t) (with n =1 or 2) given in Eqs. 8a and
8b. N R denotes the number of rotor periods, during which
recoupling is applied (i.e., the recoupling time), while t
experiment, which reflects the so-called rotor encoding of the
recoupled interaction [21, 60] and will become crucial for the
generation of sideband patterns (see section 2.3). Before
turning to these, the state W deserves some discussion.
As mentioned above, W can take different forms.
Evolving out of an initial state ρ0 = IX, W remains a single
coupling, but instead represents the anisotropy of the
chemical shielding (CSA) or the quadrupole coupling
between the nuclear quadrupole moment and the electric field
gradient at the nucleus. This can directly be seen from the
Hamiltonians, which contain operators of a single spin only.
In spin-1/2 cases (CSA), the spin simply stays in a
transverse state, the phase of which is modulated by the
recoupled CSA. In the terminology of coherences, such
transverse states correspond to single-quantum coherences
(SQCs), because the spin state can be understood as a
10 Current Analytical Chemistry, 2005, Vol. 1, No. 1
coherent superposition of the states with magnetic spin
quantum numbers mI = +1/2 and mI = –1/2. This involves a
transition of |∆mI | = 1, which is a SQ transition.
In the quadrupolar case (i.e., nuclei with spin ≥ 1),in
contrast, the spins can evolve into coherences of higher order
under the quadrupole coupling. As this work is only
concerned with 2H as a quadrupolar nucleus, the discussion
can be restricted to the spin-1 case with relatively weak
quadrupole couplings. The only higher-order coherence
available to deuterons is the double-quantum coherence
(DQC) which involves a transition of |∆mI | = 2. At first
sight, it might appear a bit surprising that in NMR theory
such DQCs are commonly considered although quantum
mechanics says that the DQ transition is forbidden. In
NMR, however, this rule does not apply because the spin
states under consideration are non-equilibrium states that
have been created by a strong perturbation of the spin system
(i.e., the application of a high power RF pulse) [29].
Consequently, in the case of deuterons W represents a DQC
that is generated through the recoupled quadrupole coupling.
When spin-spin couplings, such as J- or recoupled
dipole-dipole couplings, act on the spin system, W
represents two-spin states, which again directly reflects the
properties of the Hamiltonians. Compared to the single-spin
scenarios described above, the two-spin interactions provide
access to two- and multi-spin correlations and, subsequently,
to coherence transfer processes. A homonuclear two spin-1/2
system with a dipole-dipole coupling bears analogy to a
spin-1 quadrupolar nucleus insofar as the coupled two-spin
system can form a DQC. In contrast to the 2H case,
however, the dipolar DQC is, in general, not the highest
possible coherence order, because further spins can couple to
the DQC and, in this way, produce multiple-quantum
coherences of various orders [61, 62]. When the number of
spins and couplings increases, the coherent nature of the spin
state becomes more and more concealed and is finally lost.
In other words, such multi-spin effects bridge the gap
between coherent spin states and incoherent relaxation
phenomena.
In a heteronuclear two spin-1/2 system, the situation is
simplified by the large difference in LARMOR frequencies
between the two spins involved. As a result, the spin part of
the Hamiltonian is truncated (see Eq. 2b) [30], and the spins
can separately be manipulated by RF pulses. Therefore, in
the experiments, an initial state ρ0 = IX evolves first into a
two-spin correlation, or “antiphase coherence”, of the form
W = 2IYS Z, which does not evolve into a DQC unless an
additional RF pulse is applied on the S-spins. More
generally speaking, in heteronuclear experiments, RF pulses
can be used to switch between transverse (IX,Y, SX,Y) and
longitudinal spin states (IZ, SZ) independently on the two
spins involved. In this way, spin states, such as transverse
magnetisation, can straightforwardly be transferred from the
I- to the S-spin through a transfer sequence of the form
DIS
H MAS (π2) I (π2) S
IX 2IY SΖ sin Φ λ 2IΖ SY sin Φ λ
DIS
H MAS
SX sin Φ λ sin Φ λ (9)
Ingo Schnell
Procedures of this kind are well established as building
blocks of liquid-state NMR experiments [10], and they can
also be realised in solid-state experiments by combining the
averaging performance of fast MAS with suitable recoupling
pulse sequences. As the transfer occurs coherently, the
amplitude of the final state (see Eq. 9) contains the full
information on the heteronuclear dipole-dipole coupling.
Thus, coherent transfers not only occur in a well defined
manner, they also enable the measurement of the involved
coupling, as will be discussed below. In these respects, they
are superior to incoherent transfer processes, such as the
well-established technique of “cross polarisation” in solid-
state NMR [30].
2.3. The Measurement of Couplings
After describing how and which coherent spin states can
be generated in solid-state MAS NMR experiments by
recoupling anisotropic interactions, it is now time to turn to
the actual purposes of all these efforts, namely to
experiments which allow anisotropic interactions to be
measured in a selective and sensitive fashion. These
measurements aim to provide detailed insight into the
structure of molecules, their packing and their dynamics. In
this respect, high-resolution NMR spectroscopy has already
proven its enormous versatility for characterising solubilised
systems, and the goal of this work is the design of
experiments which enable solid-state NMR to unearth as
much information as possible about solid materials, which
is so far buried in the spectra by solid-state effects.
All solid-state NMR experiments discussed in this
review, in principle, follow the same scheme, which is
depicted in (Fig. 3). While fast MAS is continuously
applied, anisotropic interactions, primarily dipole-dipole
couplings, are recoupled by appropriate pulse sequences
during two periods. These are separated by an indirect
spectral dimension t1, which is inserted to shift the phases
of the two recoupling blocks relative to each other and/or to
record a spectrum with chemical-shift resolution. In some
cases, the latter can alternatively be achieved by placing a
spectral dimension t1 before the first recoupling block.
Optionally, the experiment starts with a preparation period
during which the initial state of the spin system required for
the experiment is prepared; for example, transverse 13C
magnetisation can initially be prepared by a conventional
cross polarisation of 1H magnetisation. At the end of all
pulse sequences, the signal is detected during the direct
spectral dimension t2.
Rotor Encoding and Sideband Patterns [93]
An efficient and sensitive way to measure anisotropic
interactions is the generation of spinning sideband patterns
via rotor encoding (RE) of the recoupled interactions [21,
60]. Such rotor encoding is accomplished by combining two
recoupling blocks with different initial rotor phases,
meaning that the second block is shifted with respect to the
first by a time t1 which is incremented in steps ∆t1 < τ R
(see Fig. 3). The amplitude of the recoupled signal is then
modulated (for ω R-recoupling) according to [93] depending
on the coherence pathway selected and observed in the
recoupling experiment. For 2ω R-recoupling, ΦωλR would
have to be used instead of ΦωλR (see Eqs. 8a and 8b). The
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy Current Analytical Chemistry, 2005, Vol. 1, No. 1 11

Fig. (3). Schematic representation of the two-dimensional recoupling experiments described in this work.

time t1 represents the delay between the two recoupling
blocks, and N exc and N rec denote the numbers of rotor
periods during which the first and second recoupling
sequence is applied (referring to “excitation” and
“reconversion”, respectively). Following Eq. 7, the
observation of the initial state ρ0 would give rise to a signal
of the form Scos (Eq. 10a), while the evolution into a state
W and back into observable magnetisation would yield a
modulation of the form Ssin (Eq. 10b, as also found for the
transfer described by Eq. 9). In both cases, the spin states ρ0
and W need to be selected between the two recoupling
blocks, and other contributions need to be removed by an
appropriate phase cycle of the RF pulse sequence. Without
such a selection procedure, a superposition of sin- and cos-
type evolutions leads to a signal of the form Ssin,cos
according to Eq. 10c. It should be noted that this
consideration is simplified insofar as any evolution of the
spin system (in a non-longitudinal state) occurring during
the short t1 intervals is neglected. “Pure” ω R- or 2ω R-
recoupling without contributions from the other modulation
is only ensured when the spin state present during t1 is not
subject to evolution. Otherwise, the evolution during t1
gives rise to additional ω R- and 2ω R-modulations that enter
into the overall average Hamiltonian [93].
While incrementing t1, the observed signal forms echoes
when Φ λωR (tI ) = Φ λωR (0) or Φ λ2ωR (tI ) = Φ λ2ωR (0). It is
obvious from Eqs. 6 and 8 that, for ω R- and 2ω R-
recoupling, full echoes occur in the t1 signal every full and
half rotor period, respectively. Therefore, after Fourier
transformation along t1, spinning sidebands are, in principle,
found at all multiples of ω R in the ω R-recoupled spectra,
while 2ω R-recoupling leads to sidebands only at multiples
of 2ω R. More specifically, a signal of the form Ssin,cos gives
rise to sidebands at all multiples of ω R and 2ω R (for ω R-
and 2ω R-recoupling, respectively), while Ssin and Scos lead
to sidebands only at odd and even multiples, respectively, of
ω R and 2ω R [93]. An example for sidebands arising from
ω R- and 2ω R-recoupling of a 1H-13C dipole-dipole coupling
is given in Fig. 4. The signals and sideband patterns shown
there correspond to the Ssin-type (Eq. 10b).
S cos (t1) ~ cos Ne xc Φωλ (0) cos Nre c Φωλ (t1 ) (10a)
R R
Ssin (t1 ) ~ sin N ex c ΦωλR (0) sin Nre c Φωλ (t1 ) , or (10b)
R
S sin c os (t1 ) ~ sin Ne xc ΦωλR (0) + cos Ne xc ΦωλR (0)
sin N rec Φωλ (t1) + cos N rec ΦωλR (t1) (10c)
R

Fig. (4). Modulations of the t1 -signal (on the left), arising from rotor-encoding in a 1 H-13 C dipolar recoupling experiment performed
under ωR - and 2ωR -recoupling conditions. The corresponding sideband patterns are shown in the middle. The experimental data was
recorded on the CH resonance of powdered L-alanine at 25 kHz MAS. Under ωR - and 2ωR -recoupling conditions, the 1 H-13 C dipole-
dipole couplings were recoupled for 2τR and 4τR , respectively (see experimental scheme in Fig. 3). The dotted sideband patterns are
calculations using r CH = 113 pm. On the right, the dependence of the sideband patterns on the product of the dipole-dipole coupling
and the recoupling time is shown for a series of calculated patterns.
12 Current Analytical Chemistry, 2005, Vol. 1, No. 1 Ingo Schnell

Directed Recoupling and Inverse Detection [63-67]
In experimental pulse schemes, the recoupling sequences
described above can be combined with spectral dimensions
in various ways. The insertion of a t1 dimension between the
two recoupling blocks has already been discussed with
respect to the rotor encoding and the sideband patterns
resulting from this. Of course, in many cases chemical shift
resolution is also desired, which is provided by the spectrum
recorded in the direct dimension (t2, see Fig. 3) and can
additionally be provided by an indirect dimension. In the
experiments discussed here, the correlation of two such
spectral dimensions occurs via recoupled dipole-dipole
couplings, because they are capable of transferring spin states
from one spin to another. In homonuclear cases (e.g., 1H-
1H), two principal experimental schemes can be designed
(Figs. 5a and 5b), which differ in the spin state detected in
the indirect dimension: either a transverse single spin state,
i.e., a SQC, is detected before the dipolar transfer step (Fig.
5a), or a two-spin correlation, i.e., a DQC, is observed in
the middle of the (split) transfer sequence (Fig. 5b). The two
experiments therefore represent two-dimensional SQ-SQ or
DQ-SQ correlations [68], respectively.
the spin system can lead to a reversal of the role of I and S
spins. In the various types of heteronuclear correlation
experiments, different pathways of transverse coherences can
be chosen [65]. The potential of this option lies in the
fundamental difference of the coupling topologies when one
of the two spin species considered is isotopically dilute (or
spatially isolated). A transverse S coherence evolves as [65]
while transverse I coherences evolve as [65] where Φ(i) =
ΦD IS.(i) or Φ D IS.(i) , depending on the type of recoupling.
ωR 2ωR
Comparing Eqs. 11 and 12, it becomes obvious that the
structure of the modulation experienced by evolving I-spins
is much simpler, because it consists of a sum of pair
couplings. When I-spins are transverse for the whole course
of the experiment, i.e., an II experiment is performed (see
Fig. 6c), the final signal is a superposition of pair
signals [65].

Fig. (6). Coherence transfer pathways in hetero-nuclear dipolar

2D correlation experiments, referred to as (a) IS, (b) SS, (c) II and
(d) SI recoupling. (c) and (d) involve “inverse” detection.
(i)
Σ iΦ(i) IΖ SΖ
SX S γ Π i cos Φ (i)
(i)
Σi IΖ S γ sin Φ(i) Π j i cos Φ (i ) ... , (11)
=
Fig. (5). Homo- and heteronuclear dipolar 2D correlation while transverse Icoherences evolve as [65]
experiments resulting from different combinations of dipolar (i)
Σ iΦ (i) IΖ S Ζ
Σ i IX cos Φ
recoupling blocks and spectral dimensions. (i) (i) (j)
Σ i IX
For heteronuclear experiments (e.g., 1H-13C
or 1H-15N),
Σ i IY S Ζ sin Φ
(i) (i)
the presence of two distinguishable species of nuclei allows ... , (12)
for one more degree of freedom in the design of pulse
sequences, because during the recoupling periods each of the ~ Σ sin Ne xc Φ (i) (0)
SII sin Nre c Φ (i) (t1 ) , (13)
two spin species can separately be chosen to occupy a i
transverse state, as schematically depicted in Fig. 6. To
discuss this possibility and the consequences arising from it In the corresponding two-dimensional spectrum, the
in more detail, a simple spin system consisting of abundant detected I-spins can be resolved from each other by their
I spins and rare S spins will be considered in the following, chemical shift, while in the indirect dimension, in general,
as for example represented by a system with 1H and 13C in two options are possible: The two-spin correlation can be set
natural abundance under the application of a pulse sequence to a full longitudinal state of heteronuclear dipolar order
probing a heteronuclear two-spin state. It should be noted (HDOR, i.e., IZS Z) [65, 69, 70], which does not evolve at
that isotopic labelling schemes and a specific topology of all. As a consequence of this, pure rotor encoding leads to
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy
sideband patterns, which may be a sum of two-spin patterns,
because different I spins couple to different S-spins.
However, if S-spin chemical-shift resolution is required, a
single-quantum antiphase state with a transverse S compo-
nent (i.e., IZS X,Y), can provide the necessary chemical-shift
separation also in the indirect dimension [65, 70].
Direct S detection (SS experiment, see Fig. 6b), in
contrast, leads to signals of the type [65].
SSS ~ Σ sin Nexc Φ(i) (0) sin Nrec Φ(i) (t1)
i shortcomings as the IS experiment. The need for an initial
(i) (j)
Π cos Nexc Φ (0) cos Nrec Φ (t1)
j=i (14)
The additional cosine terms lead to problems in
REDOR-based correlation experiments, as they introduce an
angular dependence into the experimental data and
complicate the interpretation [71, 72]. Corresponding
interference effects are also known to arise from multiple J-
couplings in liquid-state NMR [10]. There is even an
angular dependence which does, however, not result from an
orientation dependence of the J-couplings (as there is none),
but from the relative angles of chemical bonds [10].
An analysis of the spinning sidebands is also possible
for an SS experiment when only the dominating couplings
are to be determined [70]. Then, the cosine terms in Eq. 14
representing weak-coupling perturbations merely lead to a
slight overestimation of the measured coupling values. A
second coupling which reaches a similar magnitude as the
primary one renders the analysis of the pattern in terms of a
single spin-pair critical. On including the second coupling,
however, the angle between the two involved coupling
vectors becomes accessible [65]. In fact, the sideband
patterns are surprisingly sensitive to the angle when
“asymmetric” recoupling sequences, i.e., different excitation
and reconversion times, are applied [65]. In the case of I2S
spin systems, for example, the large potential of directed
recoupling becomes appreciable. The individual coupling
constants can be determined precisely by performing an II
experiment (which might not necessarily be an inverse-
detected experiment, for example when the S spin is a
proton), while the SS experiment can be used to determine
the angle, using the known coupling constants as fixed
parameters.
We are now left with the IS and SI recoupling schemes
(shown in Figs. 6a and 6d), which involve a polarisation
transfer (termed “recoupled polarisation transfer”, REPT)
[69]. The IS experiment (Fig. 6a, with S being identified
with 13C in natural abundance) is conceptually simple
because the pulse sequence does not contain a cross
polarization, and has already proven to be useful in a variety
of molecular dynamics applications [73, 76], where the
reductions of CH dipole-dipole couplings due to fast
anisotropic dynamics were measured quantitatively. Its
signal modulation reads [65].
SSS ~ Σ sin Nexc Φ (i) (0) sin Nre c Φ (i) (t1 )
i called “z-gradients”). Effectively, such PFGs introduce a
Πj cos N rec Φ (j) (t1 ) (15)
=i
Current Analytical Chemistry, 2005, Vol. 1, No. 1 13
This leads to sideband patterns which, in comparison to
the SS experiment, are even less influenced by the remote
couplings represented by the cosine terms. Only the first-
order sidebands suffer from remote couplings to an
appreciable degree, and can be excluded from the analysis. It
should be noted that the one-dimensional version of REPT
is also known as TEDOR (transferred-echo, double-
resonance, see also Fig. 5c) [77].
The SI recoupling scheme (inverse-detected REPT, Fig.
5d) features essentially the same advantages and
cross-polarisation step to transfer the initial magnetization to
the rare S nuclei and the subsequent TEDOR transfer back to
the abundant I-spins may at a first glance represent a serious
disadvantage as compared to the other three variants.
However, it turns out that the SI experiment is the inverse
variant which is so far best applicable to samples naturally
abundant in 13C or 15N [64, 65].
Using II- and SI-type recoupling schemes, the concept of
inverse detection, which is well established and of great help
in high-resolution NMR [11, 108], can be adapted to solid-
state conditions. The inverse, i.e., 1H, detection schemes not
only provide the flexibility discussed above for determining
multiple couplings, but also and more importantly allow the
achievable signal-to-noise ratios to be significantly enhanced
by factors of 10 and more [63,64,66]. In this way, natural-
abundance 15N spectroscopy becomes feasible in solid-state
NMR [64, 65].
Signal Selection and Suppression Using PFGs [78]
An important technical difficulty associated with inverse
(1H) detection schemes in 1H-X correlation spectroscopy is
the handling of excess 1H magnetisation. As the
heteronucleus X is usually much less abundant in the
materials than 1H (for comparison: 1H 100%, 13C 1.1% and
15N 0.35% natural isotope abundance), the vast majority of
1H nuclei does not come into touch with the heteronuclei of
interest. Their intense signal contribution must therefore be
removed from the detected signal or, preferably, be
suppressed before the signal is detected. In conventional
phase cycling schemes of the RF pulses, unwanted signals
are cancelled by subtracting the signal of consecutive
transients of the pulse sequence. This procedure, however,
becomes the more disadvantageous the stronger the
unwanted signal contributions are. This is because the
digital receiving units of the spectrometer are at some point
saturated with the unwanted signal, while the weak wanted
contributions fall below the digitisation threshold.
In high-resolution NMR, this problem has been solved
by introducing pulsed field gradients (PFGs), [79, 81]. For
a short time in the range of a few milliseconds (“pulse”),
additional magnetic fields are applied such that a small
gradient is imposed on the resulting magnetic field.
Depending on the geometry of the coils used for these
PFGs, the gradients can adopt basically any geometry. The
most common ones in high-resolution NMR are linear
gradients along the direction of the external field B0 (so-
inhomogeneity into the B0 field and give rise to a well-
defined dephasing of all transverse spin states, which can be
reversed by application of an inverse PFG. Based on these
14 Current Analytical Chemistry, 2005, Vol. 1, No. 1
dephasing and rephasing effects, PFGs are well established
tools for the selection of coherence transfer pathways in
high-resolution NMR experiments [80, 81]. Their particular
advantage over pulse phase cycling schemes is that they
develop their effect within a single transient of the pulse
sequence. Consequently, signal subtraction procedures over a
number of consecutive transients can be avoided, which
helps to reduce experiment time requirements and, moreover,
allows the dynamic range of the signal receiving unit to be
optimally adjusted to the incoming signal of interest.
In solid-state NMR, however, PFGs are relatively rarely
used [82-85], because their usual application times are in the
order of a few milliseconds and, thus, relatively long as
compared to transverse relaxation times of solids. Transverse
spin states, which are required for PFGs to take effect, tend
to be subject to strong spin-spin relaxation so that the signal
largely disappears before a dephasing/rephasing cycle is
completed. Hence, PFGs need to be combined with
decoupling techniques that efficiently average the
interactions responsible for the relaxation mechanisms, i.e.
predominantly dipole-dipole couplings in spin-1/2 systems.
In heteronuclear 1H-X systems, conventional RF 1H
decoupling pulses would serve this purpose, but due to the
action of the field gradient the RF bandwidth has to be
correspondingly broad, which is usually beyond the
capabilities of available RF decoupling pulse schemes. In
addition, RF pulses and field gradient pulses would have to
be applied at the same time, which also makes high
demands on the NMR probe electronics; in particular,
coupling phenomena between the individual circuits need to
be prevented. Magic-angle spinning (MAS), in contrast,
does not give rise to such interference problems with PFGs,
and therefore appears as the method of choice.
Recently, PFGs have been made available to solid-state
NMR by technically combining them with fast MAS
equipment and experiments [78]. The required field gradients
are provided by a simple pair of coils which have been built
into a standard fast-MAS probe. The selective dephasing and
rephasing capabilities of PFGs are used to design
experimental techniques for 1H signal selection and
suppression in solids, which have originally been developed
for liquid-state NMR. In particular, WATERGATE [86-89].
and DANTE [90, 91]. pulse sequences can be used under
MAS to selectively excite or suppress peaks in 1H solid-
state spectra. These tools can readily be incorporated into
solid-state MAS NMR experiments, as has, for example,
been demonstrated for 1H-1H double-quantum NMR spectra
of supramolecular systems, where the 1H signals of interest
Fig. (7). (a) Pulse sequence for an inverse (1 H) detected 1 H-15 N correlation experiment in the solid state under fast MAS. (M2, M3) The
sequence consists of a TEDOR-type polarisation transfer, in the middle of which a t1 ’ dimension introduces rotor encoding, while a
spectral t1 dimension provides a 15 N spectrum with chemical-shift resolution (split-t1 approach). PFGs help with suppressing
unwanted 1 H signal. (b) Pair of field-gradient coils attached to the top and the bottom of a Bruker 2.5 mm fast-MAS stator in a
commercial 1 H-X double-resonance probehead with a standard bore diameter (54 mm).
Ingo Schnell
are relatively weak and need to be observed despite the
presence of the strong 1H signal of long alkyl side chains
[92]. PFG-assisted suppression of this strong perturbing
signal is shown to be particularly useful for obtaining
unambiguous results. Moreover, PFGs are also used in
inverse (1H) detected 1H-15N correlation experiments to
remove strong unwanted 1H signal contributions [64, 65].
Probing Molecular Orientation [93]
The recoupling schemes and PFGs approaches described
in the previous sections were concerned with providing
means to experimentally determine dipole-dipole couplings
in a sensitive and selective fashion. Such couplings contain
information on the structure and dynamics of molecules,
because they depend on the distance of the two nuclei
involved as well as on the orientation of the coupling vector
relative to B0. In liquid-state NMR, these dependences are
also exploited in various ways, for example to derive
distance constraints from proton-proton contacts or to infer
molecular conformations from residual dipole-dipole
couplings after partially aligning the molecules in a liquid-
crystalline medium which adopts a preferred orientation
along the B0 axis [22-24]. The latter approach has recently
gained considerable importance for biological NMR
applications, and emphasises the relevance of orientation
measurements by NMR.
In solid-state NMR, such measurements are very well
known, because the anisotropies of the interactions, in
particular CSAs and 2H quadrupole couplings, allow the
orientation of molecular segments to be straightforwardly
detected by simple solid-state NMR spectra recorded on
static samples [31]. The restriction to static samples means
that the sites of interest need to be selectively labelled with
NMR-active isotopes, such as 13C or 2H, because an
analysis of the data is only possible when a single or very
few resonances contribute to the spectrum. To measure
molecular orientation at multiple sites, the orientation
information needs to be correlated with a spectral dimension
which provides chemical-shift resolution, and this is exactly
how such experiments are performed in high-resolution
NMR. An analogous strategy can be followed in solid-state
NMR when orientation measurements are performed under
(fast) MAS conditions, which can efficiently be
accomplished by using the orientation dependence of
recoupling pulse sequences [93].
From Eqs. 6, 8a and 8b, it can be seen that the
recoupling efficiency depends on the orientation of the PAS
in the rotor frame, namely on the angle β which represents
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy
the tilt between the rotor axis and the main axis of the
anisotropic interaction. In detail, the ω R- and 2ω R-
modulated terms exhibit different orientation dependencies,
namely sin 2β and sin2 β, respectively. Obviously, from a
sin 2β dependence, the angle β can only be extracted
unambiguously over the interval β = [0°,45°], while the
sin 2 β function is invertible over the interval β = [0°,90°].
Hence, 2ω R-recoupling widens the range of distinguishable
molecular orientations by a factor of two. Moreover, in β-
ranges where the angle sensitivity of sin 2β is relatively low,
sin 2 β depends strongly on β and vice versa. In this way, the
ranges with high β-sensitivity are complementary in the ω R-
and 2ω R-modulated terms [93].
In ref. [93] the effects of sample orientation have been
investigated using highly oriented polyethylene fibres as a
model system, and the feasibility of extracting orientational
order parameters from sideband patterns of rotor-encoded
MAS experiments has been demonstrated. In the solid-state,
sample orientation phenomena are of particular interest for
biological systems or for supramolecular structures, such as
discotic liquid crystals, since the performance of such
materials in optoelectronic devices strongly depends on the
degree of orientational order within the material [94]. In
particular, an extruded fibre of a columnar discotic mesogen
[73]. has been studied by novel ω R- and 2ω R-recoupling
methods, and the MAS NMR results have been compared to
X-ray diffraction studies [93].
Scaling of Couplings by OMAS [95]
Instead of using the different orientation dependences of
ω R- and 2ω R-recoupling methods, orientation measurements
can also be based directly on the evolution of the spin
system under an orientation dependent (i.e., anisotropic)
interaction, provided that the anisotropies are not
(completely) removed by an averaging technique. In the case
of MAS, anisotropic interactions can straightforwardly be
allowed to partially “survive” the averaging process by
slightly misadjusting the magic angle. Under these off-MAS
(OMAS) conditions [31], anisotropic interactions are scaled
by 12 (3cos2 θ − 1) , which would be zero for θm = 54.7°.
Technically, slight misadjustments in the range of a few
degrees can easily be realised using standard MAS
equipment. Since the effective scaling factor is then still
relatively small, this OMAS approach is basically restricted
to strong anisotropic interactions. At this point, however, a
contradiction is encountered, as residual anisotropic
interactions would of course lead again to significant line
broadening and destroy the chemical-shift resolution. The
solution to this problem is to separate the measurement of
the (residual) anisotropic interaction from the chemical-shift
information in two spectral dimensions.
A particularly intriguing way to realise this separation is
possible in the case of 2H NMR spectroscopy and the 2H
quadrupole coupling because of a fortunate combination of
two properties: Firstly, the 2H quadrupole coupling in solids
is strong enough to allow all relevant information to be
extracted from a 2H NMR lineshape even if its spectral
width is considerably reduced under OMAS conditions.
Secondly, the first-order quadrupole coupling affects all
transitions and, thus, NMR frequencies of the spins except
for the maximum order transition. Therefore, the 2H double-
quantum (DQ) transitions between the spin states of
Current Analytical Chemistry, 2005, Vol. 1, No. 1 15
magnetic quantum numbers ±1 are not affected by the 2H
quadrupole coupling. Consequently, 2H DQ resonances do
not exhibit a quadrupolar lineshape. The same phenomenon
constitutes the fundamental idea behind multiple-quantum
(MQ) MAS NMR spectroscopy of half-integer quadrupolar
nuclei, which was initially proposed by F RYDMAN and
HARWOOD [96], and has emerged as a powerful branch of
solid-state NMR spectroscopy with widely spread
applications [97-103].
By combining the ideas of MQMAS and OMAS, in a
sense, a two-dimensional 2H NMR experiment can be
designed, which correlates isotropic 2H chemical shifts in
the indirect dimension with scaled pseudostatic 2H
lineshapes in the direct dimension. In this way, it is
possible to separate multiple 2H resonance lines according to
the DQ chemical shifts of the involved 2H nuclei. In ref.
[95] the theory of such two-dimensional 2H DQ OMAS
experiments is described, and the principal features and
applicability is demonstrated using a sample of L-histidine
hydrochloride monohydrate, which was deuterated at four
positions by simple recrystallisation from D2O.
For 2H NMR spectroscopy, this DQ OMAS experiment
is particularly helpful because 2H virtually always needs to
be isotopically enriched in the samples. This in turn
provides a practically perfect site selectivity of 2H NMR
investigations, as long as chemical synthesis allows the site
of interest to be deuterated in a fully selective fashion.
Although, in chemical synthesis, the “art” of site-selective
deuteration has gained a remarkable degree of proficiency
[104, 105], the required synthetic efforts may still form an
undesirable obstacle that calls for simplification. In this
context, a particular challenge is encountered for materials
where simple deuteration procedures are no longer selective,
but yield samples with multiple sites carrying 2H. This is,
for example, always the case for acidic protons which can
conveniently be replaced by deuterons by exposing the
material to a deuterated solvent capable of exchanging its
deuterons with the solubilised material. Using deuterated
water, D2O, for example, such a simple exchange procedure
places 2H at all sites of sufficient acidity, i.e.,
predominantly NH n and OH positions in organic materials,
while CHn sites are usually unaffected. 2H NMR
investigations of such multiply deuterated samples are
considerably aggravated by the fact that the spectra consist of
multiple superimposed lines which are, in general, difficult
to separate by deconvolution procedures without perturbing
the shapes of the individual resonance lines.
3. INTRODUCING SOLID-STATE NMR TECHNI-
QUES INTO HIGH-RESOLUTION MAS NMR
When high-resolution NMR spectroscopy is performed
on systems in dilute solutions, the spectral resolution of the
resonance lines is primarily an issue with respect to the
strength and the homogeneity of the static magnetic field.
Aiming at maximum separation of sharp resonance lines,
ongoing technical efforts are devoted to the increase of the
magnetic field (presently reaching about 1 GHz 1H
frequency) while possibly retaining its homogeneity (at a
level of < 0.1 Hz). As long as the systems under
investigation can be isotropically dissolved in suitable
16 Current Analytical Chemistry, 2005, Vol. 1, No. 1 Ingo Schnell

solvents, residual anisotropies of internal interactions do not The extreme situation of anisotropic interactions causing
play a significant role with respect to spectral resolution. severe line broadening and obscuring chemical shift
information is familiar from solid-state NMR and has been
In fact, liquid-crystalline media are increasingly used to
discussed in the previous Chapter. In partially mobile
introduce a small and directed anisotropy of the samples by
systems, the situation is certainly not as extreme as in
effectively aligning the molecules to some degree along the
solids, but when standard solution-state NMR experiments
external magnetic field [22-25]. Even under these conditions,
fail due to line broadening effects of residual anisotropic
the spectral resolution is practically not hampered. The
interactions, techniques from solid-state NMR appear to be
success of this relatively new approach and the significance
well suited to help overcome these problems. Following this
it gained over only a few years clearly indicates that systems
idea, MAS has successfully been combined with standard
with residual anisotropies, be they deliberately introduced or
solution-state NMR techniques, such as COSY [10], HMQC
be they inherently present, are about to become more and
[106, 107], HSQC [108] or NOESY [109], and the approach
more relevant in NMR research. Fields of applications
has been termed high-resolution (HR) MAS [26-28]. In the
include, among others, solid-phase synthesis and on-bead
majority of current HRMAS applications, however, MAS is
characterisation, tissue samples, or membrane proteins.
merely used to make the system behave more liquid-like and
Therefore, NMR methodology needs to make experiments
to allow standard solution-state experiments to be performed
and techniques available which are capable of delivering the
[110-112]. Considering the character of ”borderline” samples
desired pieces of information on such systems.

O
O
H6' O Si
PivO H2
H6
H4 O
PivO NH2
H5 OPiv
H3 H1

Fig. (8). 1 H NMR spectra of a small glycoside attached to swollen beads of polystyrene resin, as commonly used in solid-phase
synthesis. The spectral resolution is obscured mainly due to susceptibility effects which can be averaged by MAS.
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy
that are neither isotropic solutions nor true solids, further
progress can thus be expected when solid-state and solution-
state NMR methods are combined more extensively.
In this work, the concept of recoupling, which is well
established in solid-state NMR, is introduced to HRMAS.
In the HRMAS experiments performed so far, MAS was
used to efficiently remove anisotropic interactions
throughout the experiment. The line-narrowing effect of
MAS, however, is only required during the spectral
dimensions, but not during the periods where the (residual)
interactions are essential as a driving force for informative
processes, such as exchange of magnetisation. By
introducing suitable recoupling techniques, this source of
information becomes accessible and can be exploited for
characterising the systems under investigation more
thoroughly.
In the dipolar recoupling experiments presented below,
the anisotropic interactions are still efficiently averaged by
MAS during the t1 and t2 dimensions, thereby ensuring
spectral resolution, but the dipole-dipole couplings are active
during the exchange periods at (nearly) their “non-MAS”
strength. In addition, dipolar recoupling pulse sequences can
be designed such that only the dipolar interaction is re-
introduced under MAS conditions, while unwanted
anisotropic interactions are suppressed. Moreover, the
possibility of tailoring average interaction Hamiltonians by
RF pulse sequences also allows dipolar decoupling to be
accomplished and, in this way, an isotropic mixing
Hamiltonian to be generated, which gives rise to exchange
processes purely driven by J-couplings. In the following, it
will be demonstrated how dipolar and J-driven exchange can
be distinguished and independently measured in two-
dimensional NMR correlation experiments, corresponding to
the distinction of “through-space” and “through-bond”
contacts, by a combination of dipolar recoupling and
decoupling schemes under HRMAS conditions.
3.1. The Quest for Spectral Resolution
Considering systems with small residual anisotropies,
the fundamental motivation for applying MAS in NMR
experiments is no doubt the lack of spectral resolution under
static conditions. On the top of Fig. 8, a static 1H spectrum
is shown which is typically observed when samples from
routine solid-phase synthesis are investigated. In such cases,
the systems of interest, mostly peptides or glycopeptides,
are attached to swollen polymer resins. This greatly
simplifies the synthetic procedures, but causes severe
problems in the 1H NMR spectra, because they are
dominated by the broad resonances of the resin, while the
signals of the attached systems are totally obscured. As
illustrated in Fig. 8, this problem can largely be overcome
by applying MAS. Moderate spinning frequencies of about
4 kHz usually suffice to remove essentially all line
broadening effects.
Residual Anisotropies and Line Broadening
Considering the MAS spectra at low spinning
frequencies more closely, it becomes obvious that the broad
lines more or less instantaneously split into a bunch of sharp
sidebands as soon as the sample is rotated. This behaviour is
Current Analytical Chemistry, 2005, Vol. 1, No. 1 17
typical for “inhomogeneous” line broadening, following the
terminology introduced by MARICQ and WAUGH [35]. At
first sight, this appears a bit surprising from a solid-state
NMR point of view, because proton-proton dipole-dipole
couplings are usually expected to play the dominating role
in line broadening. Their effect, however, would be of
homogeneous nature, which means that the lines should
gradually narrow on increasing the MAS frequency. As this
does not fit to the observations, the broadening must be of
different origin. In this context, one needs to recall typical
effects encountered in high-resolution NMR and, in
particular, in magnetic resonance imaging, where the
heterogeneity of the sample is known to affect the achievable
resolution. In this respect, technical precautions are routinely
taken in high-resolution NMR by using an optimised
sample geometry (long glass tubes) and, moreover, by
slowly spinning the samples.
In HRMAS studies, the heterogeneous character of many
systems (e.g., mixtures of solvent and resin beads) gives rise
to significant susceptibility gradients. The magnetic
susceptibility, χ, is a phenomenological factor relating the
magnetic field vector H0 with the magnetisation M acquired
by the (nonferromagnetic) material: M = χH0. The magnetic
induction B0 (which, in NMR literature, is commonly,
though in fact incorrectly, referred to as the “magnetic field”)
is proportional to the sum of magnetic field and
magnetisation where µ0 is the vacuum permeability. As long
as χ is a global parameter that is valid throughout the
sample, it merely leads to a minor overall shift of B0.
However, when the quantity χ is considered as a scalar
whose value is a local property χ(dV) of fractional sample
volumes dV, the resulting magnetisation is also a local
property. Consequently, each nuclear spin experiences local
dipolar fields arising from the local magnetisation of
surrounding fractional volumes. In analogy to the dipolar
spin-spin coupling, the interaction can be viewed as a
dipole-dipole coupling between a spin and the local
magnetisation M(i) of fractional sample volumes dVi in the
proximity ( of the spin. The corresponding Hamiltonian
Hx α ΣiM Zi) IZ is analogous to that of a heteronuclear
dipole-dipole coupling where the second spin is replaced by
the effective magnetisation ΣiM Zi as experienced by the
()
spin I. [28].
B0 = µ0 (H0 + M) = µ 0 (I + X ) H 0 , (16)
Magic-Angle Spinning
Based on this analogy, the effect of MAS on a sample
exhibiting susceptibility gradients can straightforwardly be
understood: While in the static case the resonance lines are
“inhomogeneously” broadened by the additional couplings
Hχ arising from local susceptibility differences, MAS
averages these interactions to zero and narrows the lines. In
principle, the situation is complicated by the fact that the
susceptibility χ is not a scalar quantity, but rather a tensor
due to its anisotropic properties. The problem arising from
the anisotropy is that H0 may no longer parallel B0 such that
the setting of the magic angle relative to the external field
H0 is no longer correct with respect to B0. Thus, for
appreciably anisotropic χ, MAS is effectively carried out
under slight “off-angle” conditions and cannot develop its
full averaging performance. This is, however, not the
18 Current Analytical Chemistry, 2005, Vol. 1, No. 1
dominating effect, as MAS is usually capable of averaging
susceptibility gradients and, thus, provides satisfactory
resolution.
Thus, with respect to line broadening, the residual
anisotropies of internal interactions play often only a minor
role in heterogeneous systems. The molecular dynamics,
though slightly restricted, is usually pronounced enough to
ensure efficient pre-averaging and to leave only small
anisotropies, which can then be removed easily and
completely by MAS at low spinning frequencies.
Nevertheless, there are residual anisotropies in systems of
this kind, and the work described in the following aims to
exploit their information content for determining molecular
structure and dynamics.
3.2. The Creation of Coherent Spin States
When MAS NMR experiments are performed on partially
immobilised systems, anisotropic interactions are averaged
both by molecular dynamics and by MAS. The weak
residual interactions, which are left over by these two
averaging processes, merely give rise to relaxation effects.
These can be understood and interpreted in a
phenomenological manner, but the direct information on
individual couplings is lost. In order to restore this
information, the relaxation phenomena need to be replaced
by actual quantum-mechanical evolution processes which
directly and individually reflect the action of the couplings
on certain spins. This requires the interactions of interest to
be amplified such that they emerge out of the “background
noise” of fluctuating interactions of various kinds.
With respect to the experiments and systems discussed
here, interactions can only be amplified in terms of
preventing them from being reduced. While it is naturally
not possible to compensate for the pre-averaging effect of
molecular motions by means of NMR techniques, it is
possible to prevent the additional averaging effect of MAS
by recoupling pulse sequences. In this way, the spin system
experiences no longer weak residual interactions, but instead
recoupled interactions at their “original” strengths which
they would develop in the static case, i.e., in the absence of
MAS. This “non-MAS” strength can suffice to make the
spin systems evolve in a quantum-mechanical manner such
that individual interactions become detectable and
measurable. In the following section, this strategy will be
discussed for the case of homonuclear dipole-dipole
couplings among protons. The same principles apply
analogously to CSAs as well as to heteronuclear dipole-
dipole and quadrupole couplings.
Recoupling Schemes [113]
In systems with pronounced molecular mobilities,
anisotropic interactions are in general relatively weak,
typically on the order of a few Hertz. To ensure efficient
recoupling under these conditions, the pulse sequences have
to be applied for durations of 100 ms and more, which is
very long compared to typical timescales of solid-state
NMR. Thus, the pulse sequences need to be equipped with
internal compensation mechanisms for the effects of other
interactions, experimental misadjustments and pulse
imperfections. In practice, helical pulse sequences, developed
Ingo Schnell
by LEVITT and coworkers [114, 115], have proven to
perform well under HRMAS conditions, and were therefore
used in the experiments to recouple homonuclear 1H-1H
dipole-dipole couplings.
Helical pulse sequences are based on the simultaneous
rotation of space and spin part through MAS and RF pulses,
respectively, and aim at creating an average Hamiltonian
consisting of selected tensor components Al,mT λ,µ by means
of symmetry principles. The selection rules and the design
of appropriate sequences are described in detail in ref. [115].
To accomplish dipolar recoupling for 1H nuclei, a pulse
sequence is required which recouples the homonuclear
dipole-dipole coupling, while providing optimum
suppression of all other, potentially perturbing, interactions,
in particular chemical shift or, equivalently, off-resonance
effects. By tailoring the tensor component T 2,0 in the spin
part, the properties of the “original” dipole-dipole coupling
can be retained (see Eq. 1). T 2,0 gives rise to zero-quantum
coherences (ZQCs), involving “flip-flop” type transitions
 ↓ ↔ ↓
↓ ↓
in pairs of coupled spins. It might appear
natural to recouple this component, but it bears the
disadvantage that unwanted interactions are difficult to
suppress, because components T 2,0 exist for all interactions.
A unique selection of the dipolar interaction is possible
through its components T 2,±2. By recoupling these
components, dipolar double-quantum coherences (DQCs) are
generated between pairs of coupled spins, involving
transitions of the ”flop-flop” type  ↔ ↓↓ .
↓↓
In principle, both ZQCs and DQCs can be used for
dipolar exchange spectroscopy. Although for the DQ
process, in contrast to ZQCs, the energy of the spin system
is not conserved, the DQ approach is favourable in terms of
suppressing unwanted interactions. In addition, as will be
discussed later, it imposes a quantum-chemical evolution on
states of longitudinal magnetisation in NOESY-type
experiments. The DQ-type of dipolar recoupling is realised
in the C7 21 pulse-sequence [114], which selects the terms
A2,1T 2,2 and A2,-1 T 2,-2 of the dipolar Hamiltonian. For an ij
spin pair the resulting average Hamiltonian read [114, 115]
where the coefficient ω D is given by.
Dij is the dipolar coupling constant of the ij pair, and
ω Rt0 is the rotor phase at the start of the pulse sequence. β
and γ denote the azimuthal and polar angle of the
internuclear vector in the rotor frame. Apart from the
orientation dependence and the phase factors, the average
Hamiltonian of C721 contains an overall scaling factor of
~ 0.232 for the recoupled dipole-dipole coupling in
comparison to the static case [114].
HC7 = ωD T2,2 + ωD T2,-2 + 2πJ ij I(i) I(j) , (17)
dipole-dipole J-coupling
343(i + e iπ/14)
(18)
ωD = Dij sin(2β) e i(ωR t0-γ)
520π 2
“Through-Bond” and “Through-Space” Information
[113]
It is important to note that, in addition to the dipolar
part, the average Hamiltonian in Eq. 17 also contains the
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy Current Analytical Chemistry, 2005, Vol. 1, No. 1 19

isotropic J-coupling between the spins i and j. The two can be restored and investigated at effectively ~ 20% of their
types of couplings differ in the information they convey, as “non-MAS” strengths. Apart from this “amplification”
J-couplings propagate through chemical bonds, while dipole- achievable by recoupling, there are important differences
dipole interactions act directly through space. In fact, J- between the Hamiltonians of C7 21-recoupled and (weak)
couplings are part of the average Hamiltonian not only for residual dipole-dipole couplings, which are obvious from
C7 21, but for all possible variants of helical pulse sequences. their spin parts:
This invariability of J-couplings opens up the possibility of
creating an average Hamiltonian which solely consists of J- HDC7 ~ T2,2 + T2,-2 ~ I (i) I(j) + 1 (i) I (j) (21a)
couplings, while all other interactions are suppressed due to offset (i) (j)
HD ~ 3I(i) (j) (i) (j)
Z IZ I I 2I Z IZ (21b)
their symmetry and transformation properties. Using helical
pulse sequences, the simplest scheme providing a pure J- The recoupled and residual interactions exhibit DQ and
coupling Hamiltonian is C931 [116, 117]. The average ZQ character, respectively. For residual couplings, in
Hamiltonian is simply given by which represents an addition, second averaging effects need to be taken into
isotropic mixing Hamiltonian. The coupling between the account, since the chemical-shift difference between the spins
spins i and j is completely independent of the molecular i and j often exceeds the residual coupling and truncates the
orientation and is also unaffected by resonance offsets, which dipolar Hamiltonian by suppressing the “flip-flop” exchange
would cause second averaging effects as soon as the term (see Eq. 21b). Dipolar recoupling, in contrast, prevents
frequency difference of the two resonances exceeds the J- such second-averaging effects, because the pulse sequence
coupling. In the context of isotropic mixing, a number of removes chemical-shift and resonance-offset contributions to
pulse sequences have been proposed in liquid-state NMR the average Hamiltonian.
[118, 119, some of which have recently been re-examined
and optimised with respect to their performance under Thus, considering an( )arbitrary state of longitudinal
HRMAS conditions [120, 121]. Here, the C931 sequence magnetisation Ρ 0 α a IZi b I(i) Z , and its quantum-
was chosen in order to keep the approaches for dipolar mechanical evolution under the different types of dipolar
recoupling (C721) and isotropic mixing (C931) as similar as interactions, three cases need to be distinguished: Firstly,
possible for combination purposes: The difference the “fully truncated” Hamiltonian HD α 2I(i) I(j)
Z Z , does not
Hamiltonian HC7 - HC9 = ω D . T 2.2 + ωD* T2.-2 is a pure lead to any evolution of the state ρ0, as is reflected by the
“dipolar” Hamiltonian, from which J-couplings have been commutation relation
removed. It cannot be produced as an average Hamiltonian of (i ) (i) (j)
a pulse sequence, but its effect on the spin system can be a I Z + b IZ(j) , 2IZ I Z = 0 (22)
approximated by the difference of the signals observed in a
C7 21 and a C931 recoupling experiment where τ rcpl denotes Secondly, under the “original” homonuclear dipolar
the recoupling time in the experiment. In the above Hamiltonian, HD α 3I(i) I(j) (i) (j)
Z Z - 1 I a state of longitudinal
approximation, mixed terms arising from the non-vanishing magnetisation evolves in terms of a ZQ exchange process,
commutator [HC7, HC9] 0 are neglected. Such terms provided that it is a non-equilibrium state with a = b ,
would only contribute to second and higher orders and have because the respective commutator reads
therefore no effect for two-spin systems. Even when multiple
spins are coupled, there is practically no effect for HC7 .τrepl (i ) (j) (i ) (j )
a IZ + b IZ , 3IZ I Z I(i)I(j ) =
<1 and HC9 .τrepl <1. Thus, by combining a C721 and a (j )
C9 31 recoupling experiment, it is possible to record NMR 2 (a b) (I (i) I I(i) I(j) ) (23)
spectra which contain information on either dipole-dipole Thirdly, the recoupled dipolar DQ Hamiltonian HC7 D
couplings or J-couplings [113]. In this way, “through-bond” generates DQ coherences out of longitudinal magnetisation,
contacts can be distinguished from “through-space” contacts as long as a = -b . These differences between the three cases
between spins. have particular consequences for NOESY-type experiments
H C9 = 2πJ i j .I (i) I (j) (19) (see below), where during a mixing time exchange processes
occur between dipolar-coupled spins that occupy states of
exp i(H C7 HC 9 )τ rc pl ρ 0 exp i(H C7 H C9 )τ
rc pl
longitudinal magnetisation. For residual dipole-dipole
couplings represented by a “fully truncated” Hamiltonian,
≈ ρ 0 + exp iH τrcpl ρ 0 exp τrc pl
C7 C7
iH there is no quantum-mechanical evolution of the spin
system. Consequently, exchange processes are solely based
on diffusion of magnetisation, which occurs on much longer
C712 -experiment time scales than quantum-chemical evolution. Such diffusive
processes are described in a phenomenological manner [10,
exp iHC9 τrcpl ρ 0 exp iHC9 τ rcpl 122] and are therefore difficult to quantify in terms of
individual two-spin couplings. In particular, the knowledge
of motional correlation times is required in order to
C912 -experiment determine dipole-dipole couplings from conventional
(20) NOESY data (see below).
Recoupled and Residual Interactions
By applying the C721 pulse sequence, MAS experiments a I (i)
Z
+ b I (j)
Z
,T 2.+2 = 2(a + b) T2,+2 (24)
are no longer restricted to the observation of very weak
residual dipolar interactions, but the dipole-dipole couplings
20 Current Analytical Chemistry, 2005, Vol. 1, No. 1
If the truncation of the dipolar Hamiltonian is avoided by
appropriate recoupling, a quantum-mechanical evolution of
longitudinal spin states is re-introduced. For ZQ- and DQ-
type Hamiltonians, this evolution is proportional to the
degree of non-equilibrium (a – b in Eq. 23) and equilibrium
(a + b in Eq. 24), respectively. Conversely, the sum
magnetisation Σ = I(i)Z
I (j)
Z and the difference magnetisation
∆ = I Z I Z are conserved under ZQ- and DQ-type
(i) (j)
evolution, respectively.
The latter property also opens up a possibility to
distinguish J-couplings from dipole-dipole couplings. The
two types of couplings provide complementary pieces of
information about the arrangement of two spins, namely
their molecular connectivity and their spatial proximity. In
general, a connection via chemical bonds which causes a
detectable J-coupling comes along with a certain degree of
spatial proximity, but the precise measurement of the spin-
spin distance allows, for example, the molecular
conformation to be elucidated. Furthermore, the orientation
dependence of the dipole-dipole coupling can, in principle,
be used for determining the orientation as well as, in the
case of mobile systems, the local mobility and dynamics of
molecular segments. In standard solid- and liquid-state
NMR, dipole-dipole couplings and J-couplings do not
interfere and there is no need to unravel them, because one of
the two interactions clearly predominates. In partially mobile
systems, however, the motional averaging is incomplete,
such that the two types of couplings can effectively be of
comparable strength, and means of distinction become
valuable.
In conclusion, the dipolar recoupling procedures outlined
above take advantage of
• the amplification, i.e. recovery, of dipole-dipole
couplings under MAS,
• the absence of second averaging,
• the presence of quantum-mechanical evolutions
instead of relaxation processes, and
• the conservation of ∆ instead of Σ, when DQ-type
recoupling is performed, which provides a means to
distinguish homonuclear dipole-dipole from J-
couplings.
These features make the C721 pulse sequence an attractive
candidate for replacing the diffusive exchange process in
MAS NOESY experiments. The above considerations
focused on homonuclear dipole-dipole couplings between
protons, but they apply analogously to all interactions
recoupled by appropriate pulse sequences, in particular to
quadrupole and heteronuclear dipole-dipole couplings.
3.3 The Measurement of Couplings
After discussing the motivation for and the principal
concepts of recoupling approaches under HRMAS
conditions, it is time to turn to measurements of couplings
which have been performed by recoupling experiments. In
the following, it shall be demonstrated how homonuclear
1H-1H and heteronuclear 1H-13C dipole-dipole couplings as
Ingo Schnell
well as 2H quadrupole couplings can be determined
selectively and accurately by different recoupling techniques.
Recoupled Exchange [113]
An experiment frequently used in high-resolution NMR
is two-dimensional 1H-1H NOESY (“Nuclear Overhauser
Enhancement SpectroscopY”) [109], the pulse sequence of
which is depicted in Fig. 9a. In the spectra, distinguishable
species of protons are correlated by signals (“cross peaks”)
which reflect a transfer of magnetisation from one species to
the other. This transfer is an incoherent cross-relaxation
effect which is driven by fluctuating residual dipole-dipole
couplings between the two spins involved. Using a
phenomenological model, the signal intensity can be
described by an exponential relation of the form [10, 122]
where the rate constants RC and RL stand for the cross-
relaxation and the “leakage” rate, respectively, and τ mix is
the mixing time during which the transfer processes occur.
While the former contains the coupling between the two
spins ij of interest, RC = 1/ 5 D2ij τ c ,the latter represents the
interaction of the pair with further spins. However, the cross-
correlation rate, RC, depends not only on the coupling Dij,
but also on the motional correlation time τ C, which is
commonly used to distinguish between the limits of fast and
slow motion with respect to the time scale of the Larmor
frequency, ω 0. For the samples investigated here, the
validity of the slow motion limit (also called “spin diffusion
limit”), i.e., τ c >> 2π . ω 0-1 = 10 ns, can be safely
assumed, but the precise value of τ C is not known, such that
the evaluation of the NOESY data alone yields relative
dipole-dipole coupling constants only.
e-RC τ mix ) e-RC τ mix ,
(25)
S(t) ~ (1
Instead of relying on such cross-relaxation phenomena,
the homonuclear 1H-1H dipole-dipole coupling can be
recoupled and, effectively, be amplified under MAS
conditions, as discussed above. For this purpose, a dipolar
recoupling sequence needs to be applied during the mixing
time (see (Fig. 9b). In this way, the incoherent cross-
relaxation processes are replaced by coherent transfer
processes via dipolar DQCs, and the signal intensities can be
expressed as where the “dipolar frequency” ω D is given by
Eq. 18 when the C721 sequence is used. In this way,
residual dipole-dipole couplings Dij can be determined from
experimental signal intensities S(τ mix ) observed for
individual spin pairs ij. The 1H HRMAS spectra resulting
from the simple NOESY experiment and the “C7-recoupled”
experiment are shown in Figs. 10a and 10b, respectively,
and two major differences are evident: Firstly, the mixing
times required for efficient transfer processes are shorter by a
factor of ~ 25 under dipolar recoupling conditions, because
the couplings are amplified by preventing the averaging
effect of MAS. Secondly, the “C7-recoupled” exchange
signals are (partially) negative in the spectra. This proves
that the transfer occurs in a coherent fashion via DQCs,
because during this process the difference (and not the sum)
magnetisation of the respective spin pair ij is conserved (see
above), which renders the exchange signals negative.
S (τ mix ) ~ sin2 ( 1 2 ωD τ mi x) β, γ (26)
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy Current Analytical Chemistry, 2005, Vol. 1, No. 1 21

In the exchange experiments discussed so far, DQCs of
spin pairs ij serve as transition states for i j magnetisation
transfer processes. Instead of merely using DQCs as
transition states, they can also be spectroscopically detected
in a (indirect) DQ dimension that is inserted into the mixing
time and replaces the t1 dimension of the exchange
experiment (see schemes in Figs. 9b and 9c). The resulting
dipolar DQ-SQ correlation experiment is usually referred to
as dipolar DQ spectroscopy, which has been established in
solid-state MAS NMR [21, 68]. It bears analogy to the
INADEQUATE [123, 124] experiment known from liquid-
state NMR, which in contrast makes use of J-couplings for
the generation of DQCs. From a two-dimensional dipolar
DQ spectrum, pairs of coupled spins can be directly
identified. Pairs of unlike spins, ij, give rise to two signals
placed at (ω i + ω j, ω i) and (ω i + ω j, ω j), while pairs of like
spins, ii, are observed on the DQ-SQ “diagonal” at (2ω i,
ω i). In an exchange experiment, in contrast, an ij correlation
gives rise to four signals, placed at (ω i, ωi), (ω i, ωj), (ω j,
ω i) and (ω j, ω j), or a single signal at (ω i, ωi), if the
chemical shifts of i and j are identical. As a consequence of
this, in exchange spectra, signals tend to overlap on the
diagonal, and exchange between like spins is difficult to
observe, while the respective signals are separately observed
in DQ spectra. In fact, the appearance of DQ signals arising
from pairs of like spins is a noteworthy feature of dipolar
DQ spectra. In the case of J-mediated correlation spectra (i.e.
INADEQUATE), such signals are not observable because
they rely on the anisotropy of the interaction.
The DQ signal intensity depends on the dipole-dipole
coupling Dij according to Eq. 26, but experimentally, the
oscillations are usually obscured by an exponential decay.
This effect is known from solid-state 1H-1H DQ experiments
[21]. In the limit of short recoupling times, i.e.
Dijτ exc/rec/2π<1, the sine dependence in Eq. 26 may be
approximated by the leading linear term, which yields
expressions of the form SD Q ~ ω D2 β.γ τ ex
2
c/rec and SDQ ~ rij
-

Fig. (9). Experimental schemes of conventional NOESY-type (a) and recoupled dipolar (b) exchange experiments. Instead of merely
serving as transition states, the dipolar DQCs can also be detected spectroscopically (c).

Fig. (10). Two-dimensional 1 H-1 H HRMAS spectra resulting from the experiments and pulse sequences depicted in Fig. 9: (a)
Conventional NOESY-type exchange, (b) “C7-recoupled” dipolar exchange, (c) DQ spectroscopy. Dotted contour lines indicate
negative signals. The sample under investigation, together with the assignment of the 1 H resonance lines, is given in Fig. 8.
22 Current Analytical Chemistry, 2005, Vol. 1, No. 1
6 (fora given recoupling time). Thus, the signals of dipolar
DQ spectra provide direct insight into the (average) distances
between distinct nuclei. Experimentally, distance ranges of
up to rij ≤ 4 … 5 Å have found to be accessible [21].
In the recoupled exchange experiments described so far,
effects potentially arising from J-couplings have been
neglected. In DQ spectroscopy, there are indeed no such
effects, because only dipole-dipole couplings can give rise to
DQCs, while J-couplings cannot due to the ZQ-character of
their Hamiltonian (see Eq. 19). In exchange experiments,
however, J-couplings may give rise to signals, which can
also been seen from the spectrum in Fig. 10b. The cross
peaks are not purely negative, as they would be for recoupled
dipolar exchange, but exhibit positive contributions from J-
couplings between the respective nuclei. In this way, a
distinction of the effects is possible in a qualitative manner,
but there is need for a quantitative separation, which can be
accomplished by the combination of C721 recoupling and
C9 31 decoupling, as discussed in the previous section. (Fig.
11) shows the two-dimensional 1H-1H spectra observed for
C7 21-recoupling and C931-decoupling, as well as the “C7-
C9” difference spectrum. The latter contains pure dipolar
(“through-space”) information and, correspondingly, only
negative cross peaks are observed.
Rotational Resonance [125]
The C721 pulse sequence used above provides broadband
recoupling of dipole-dipole couplings between all types of
protons in a system, while the selection of spin pairs ij
relies fully on the resolution of the resonance lines in the
two-dimensional spectra. However, if spectral resolution is
the only means to separate information, it is not possible to
distinguish whether a transfer i ←
← ← → j occurs really directly
or via a third spin i ← k ← j . The latter effect is well
known as “relayed transfer”, and usually prevents the
determination of weak couplings Dij in the presence of
stronger ones, Dik and Dkj. To overcome this problem,
recoupling schemes are needed which selectively restore a
single type of coupling, while leaving all others unaffected.
A particularly efficient and robust method for selective
dipolar recoupling is rotational resonance (RR), which was
Fig. (11). Two-dimensional 1 H-1 H HRMAS exchange spectra resulting from (a) C72 1 -recoupling, (b) C93 1 -decoupling and (c) the
difference of the two. The C93 1 spectrum contains pure “through-bond” information, because the exchange occurs only via J-
couplings. The “C7-C9” spectrum, in contrast, contains pure dipolar (“through-space”) information, because practically all
contributions from J-couplings are removed by the subtraction.
Ingo Schnell
initially developed by LEVITT and G RIFFIN, [126, 127] and
belongs to the routine experiments in solid-state MAS
NMR. The recoupling effect of RR is restricted to dipole-
dipole couplings between pairs of spins whose chemical-
shift difference exactly matches the applied MAS frequency
(or a small integer multiple of it). The bandwidth of RR
depends on the ZQ-relaxation properties of the system [127-
129], and was found to be around 10 Hz for proton-proton
couplings under HRMAS conditions [125].
At the start of a RR exchange experiment, a state of
maximum difference magnetisation, ρ0 α I(i) Z I(j)
Z , is
prepared by inverting the state of one of the two spins.
Consecutively, the spins are allowed to exchange
magnetisation during a mixing time under RR conditions,
which ideally leads to a magnetisation difference of zero.
The difference is recorded as a function of the mixing time,
and analysed by fitting calculated curves (see Fig. 12a).
From the initial slope of the curve, the effective dipole-
dipole coupling Dij between the spins can then be
determined selectively. The RR experiment has been
introduced to HRMAS conditions, and 1H-1H dipole-dipole
couplings as low as 0.3 Hz could be measured between two
methyl groups [125]. The selectivity of the RR approach
originally relies on the resolution of the 1H resonance lines
in the spectrum, but it can be significantly enhanced by
selective isotope enrichment if necessary. For example, 13C-
labelled methyl groups can be used, whose 1H signals can be
selected by means of a conventional high-resolution 1H-13C
HSQC/HMQC step [10, 29] before the RR experiment. In
this way, liquid-state (HSQC/HMQC) and solid-state (RR)
NMR methods are directly combined in a single HRMAS
experiment [125].
Rotor Encoding and Sideband Patterns [125]
Under HRMAS conditions, the 1H-1H dipole-dipole
coupling determined by the RR experiment usually contains
two pieces of information. Firstly, the couplings depend on
the internuclear distances, rij; secondly, the couplings are
reduced by molecular motions that occur on timescales
below microseconds. In order to separate the structural and
dynamic aspects by means of dipolar recoupling
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy Current Analytical Chemistry, 2005, Vol. 1, No. 1 23

experiments, measurements need to be performed where one
of the two parameters (i.e., either the distance or the motion)
is known. Obviously, this criterion is met for dipole-dipole
couplings of chemically-bonded 1H-13C entities, as their
(bond) distance is known. Consequently, such 1H-13C
dipole-dipole couplings are only affected by motions of the
molecular segment they belong to.
Experimentally, various pulse sequences are available
from solid-state NMR, which recouple heteronuclear dipole-
dipole couplings in the presence of MAS [51,52, 56]. For
HRMAS applications, the rotor-encoded REDOR
(REREDOR) scheme has proven to be particularly useful, as
it combines technical robustness with a sensitive means of
detection, namely rotor-encoded sideband patterns (see
Chapter 2.3). Initially, REREDOR experiments have been
developed in this work as solid-state applications for fast
MAS [65], but they can equally be applied under HRMAS
conditions. Fig. 12b shows three HRMAS REREDOR
sideband patterns from which 1H-13C dipole-dipole
couplings as low as 10 Hz can be determined independently
for the three methyl groups [125]. The ratio of these
measured couplings to the (hypothetical) value of a rigid
group can be interpreted as a reduction factor S, which
represents the motional pre-averaging in terms of the residual
anisotropy of the respective molecular segment. In the
tetrapeptide studied in ref. [125] these factors were of the
order of S = 2. 10-3 , which shows that the systems under
investigation move rapidly and almost isotropically, and
only a small residual anisotropy is left. Nevertheless, dipolar
recoupling approaches are still feasible.
Using the reduction factors S, the motional pre-averaging
of the homonuclear 1H-1H dipole-dipole couplings
determined by RR experiments (see above) can be taken into
account and removed from the coupling constants.
Depending on the molecular structure, it might be necessary
to combine the reduction factors of different molecular
segments in order to appropriately model the dynamics of
the molecule [125]. The “corrected” coupling constants,
Dij/S, so obtained depend basically only on the effective
distance between the protons i and j. In this way,
information on structure and dynamics can be separated from
each other, and effective (i.e., averaged) distances of up to
8 Å could be measured [125].
Under HRMAS conditions, the local or segmental
dynamics of a molecule can be determined not only by
recoupling heteronuclear 1H-13C dipole-dipole couplings,
but also by recoupling 2H quadrupole couplings. For this
purpose, a homonuclear recoupling scheme has to be
applied, which means that the REDOR-based pulse
sequences for 1H-13C recoupling need to be replaced by, for
example, the back-to-back pulse sequence [21, 59]. The latter
recouples homonuclear dipole-dipole couplings as well as
(weak) quadrupole couplings, because a homonuclear spin
system consisting of two dipolar-coupled spin-1/2 nuclei is
formally identical to a quadrupolar spin-1 system. For such
2H NMR studies, the sample needs to be deuterated at the

Fig. (12). (a) Rotational-resonance decay curves observed for the two 1 H-1 H methyl-methyl couplings indicated above in the
tetrapeptide structure. (b) Sideband patterns of the three methyl groups, observed in a 1 H-13 C rotor-encoded REDOR experiment. The
respective dipole-dipole couplings obtained from the experiments are given in the figure. For the 1 H-13 C couplings, reduction factors
S are also given, which reflect the motion of the respective methyl group. To derive these, a 1 H-13 C coupling of D = 6.9 kHz was
assumed for a rigid methyl group (that only rotates rapidly about its threefold symmetry axis).
24 Current Analytical Chemistry, 2005, Vol. 1, No. 1 Ingo Schnell

sites of interest. Using the tripeptide depicted in (Fig. 13) as
an example [125], deuterated methyl and methylene groups
have been introduced by esterifying the acid functionalities
with fully deuterated methanol and ethanol. The resulting
tripeptide sample is depicted in Fig. 13, together with
spinning sideband patterns which are generated by rotor-
encoding the recoupled quadrupole couplings. The 2H
chemical-shift resolution (see also ref. [95] suffices to
separate all three 2H resonances and to analyse the sideband
patterns individually Fig. 13b. By comparing the
expeimental patterns with calculated ones (Fig. 13a), the
quadrupole couplings are determined, which can then be
converted into motional reduction factors S by relating the
measured couplings to the 2H coupling of a rigid methyl
group (δQ/2π = 40 kHz, where “rigid” means that each
group only rotates about its threefold symmetry axis
[31, 129].
When the segmental dynamics has been determined by
means of such parameters S at several sites of the same
molecule, it is further possible to characterise the internal
dynamics, or residual order, within the molecule. In the tri-
and tetrapeptides (Figs. 12 and 13), for example, the residual
order from one aspartic-acid sidechain end to the other is
reflected by the ratio of the respective parameters, i.e.,
SMe /SEt ≈ 0.8. This means that the peptide as a whole is
highly mobile (with S = 2 . 10-3 ) due to the flexibility of the
linker by which it is attached to the polymer resin.
Internally, however, the peptide adopts a relatively immobile
and stable structure (with S ≈ 0.8). It is therefore reasonable
to refer to a structure of the peptide molecule with noticeable
internal dynamics and a pronounced overall motion [125].
4. CONCLUSIONS AND OUTLOOK
After the fields of liquid- and solid-state NMR
spectroscopy have developed largely independently of each
other over the past 50 years, materials of current interest,
particularly in biological NMR, call for new approaches
which combine and merge spectroscopic techniques from
either side. In this respect, recent technical progress opens up
a number of possibilities to promote NMR methodology
into this direction, for example by employing high-speed
MAS in solid-state NMR [21] and by performing high-
resolution NMR under moderate MAS conditions [28].

CD3 CD2
O
O

O O Si
H
N O
O N N
H H H
O O N
O
O O
O
CD3

Fig. (13). Molecular dynamics of the deuterated methyl and methylene groups in the given tripeptide structure, as determined by
sideband patterns resulting from the rotor-encoding of recoupled 2 H quadrupole couplings under HRMAS conditions. (a) Calculated
and (b) experimental patterns with different coupling constants (CQ = δ Q/2 π).
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy
Based on this background, this review intended to provide
an overview of new NMR methods and experiments which
have been developed by merging concepts from liquid-state
and solid-state NMR.
In solid-state NMR, a wealth of new and versatile
experiments can be designed by using coherent spin-states
and coherent transfer mechanisms. These concepts form the
basis for virtually all multidimensional liquid-state NMR
experiments [11, 13, 29], and the principal ideas can be
adapted to solid-state conditions when (i) fast MAS ensures
efficient averaging of anisotropic interactions, and (ii)
recoupling schemes selectively re-introduce the interactions
of interest, i.e., predominantly dipole-dipole couplings
which basically serve as a substitute for the much weaker J-
couplings. In this way, a wealth of experimental schemes
can be transferred from liquid-state to solid-state NMR
[63-65, 78, 95], including two important and well-
established techniques, namely inverse (i.e., 1H) detection
schemes for sensitivity enhancement [63-65] and pulsed
field-gradients (PFGs) for signal selection and suppression
[78].
Due to the anisotropy inherent to solid-state interactions,
recoupling schemes can be performed under MAS in such a
way that the recoupled interaction becomes “rotor-encoded”,
which gives rise to characteristic sideband patterns in the
spectra [65]. These patterns provide a very sensitive means
to measure the underlying coupling, and can, in a sense, be
viewed as the solid-state MAS analogues to the typical line
splittings observed in high-resolution NMR spectra.
Correspondingly, the sideband patterns can readily serve as a
measure for molecular structure [65, 66, 130, 131-133],
dynamics [134, 135], and orientation phenomena [93].
Furthermore, it is possible to re-introduce and to detect weak
residual interactions in solid-state NMR by performing
MAS under “off-angle” conditions (so-called OMAS). This
concept is particularly suited for 2H quadrupole interactions,
which can conveniently be determined by 2H double-
quantum OMAS spectroscopy in a site-selective fashion
[95].
In high-resolution NMR, samples with residual
anisotropies or heterogeneities are increasingly being studied
[22-28]. Either are weak anisotropic properties deliberately
introduced, for example by aligning molecules in oriented
media, or the sample itself is partially insoluble and of
inhomogeneous nature, such as proteins embedded in lipid
membranes or molecules attached to solid supports. In the
latter case, MAS has recently been borrowed from solid-state
NMR and combined with conventional high-resolution
NMR experiments in order to ensure sufficient spectral
resolution (so-called HRMAS) [28]. In addition to this
rather superficial combination, however, techniques from
liquid- and solid-state NMR can be merged more
extensively. In this work, it was demonstrated that
recoupling techniques are generally applicable under
HRMAS conditions [113, 125]. In this way, information on
molecular structure and dynamics no longer needs to be
derived in a rather phenomenological and semi-quantitative
manner from incoherent relaxation processes, but can be
inferred from interactions, in particular from dipole-dipole
couplings, which are measured accurately and selectively by
means of coherent processes.
Current Analytical Chemistry, 2005, Vol. 1, No. 1 25
Under HRMAS conditions, diffusive processes of 1H-1H
magnetisation transfer can, for example, be replaced by
specific 1H-1H correlations via dipolar double-quantum
coherences, when a suitable recoupling pulse sequence is
applied during the mixing period of 1H-1H NOESY
experiments. [113] Highly selective dipolar recoupling is
provided by rotational resonance techniques, [127] which
enabled the selective determination of 1H-1H dipole-dipole
couplings as weak as 0.3 Hz between two methyl groups of
an oligopeptide attached to a polymer resin [125].
Heteronuclear dipolar recoupling, e.g. 1H-13C, allows for the
measurement of residual dipole-dipole couplings of
molecular segments, which directly reflects the dynamics of
these segments. The same information is, in principle, also
accessible from residual 2H quadrupole couplings which can
also be measured by recoupling approaches under HRMAS
conditions. Notably, all these recoupling experiments
provide the coupling information in combination with
chemical-shift resolution, i.e., in a site-selective fashion.
Overall, MAS NMR experiments are attractive tools for
studying molecular structure and dynamics in solids as well
as in “borderline” materials which are neither rigid solids nor
isotropic solutions or liquids. The information on structure
and dynamics is contained in anisotropic interactions, and
can therefore be obtained in a highly selective and accurate
manner from recoupling experiments which involve the
generation of coherent spin states. These strategies apply
equally well to solid-state and high-resolution NMR, and
the differences between these two traditional fields of NMR
spectroscopy come down to a few technical details
concerning the fine adjustment of the experiments. As a
consequence of this, there is hardly a reason to keep the
separation of solid- and liquid-state NMR. In fact, there is
need to promote their “re-unification” in order to be able to
investigate the vast field of materials which combine
properties of solids and liquids, but do not solely belong to
either side.
REFERENCES
[1] Purcell, E. M.; Torrey, H. C.; Pound, R. V. Phys. Rev., 1946, 69,
37.
[2] Bloch, F.; Hansen, W. W.; Packard, M. E. Phys. Rev., 1946, 70,
474.
[3] Proctor, W. G.; Yu, F. C. Phys. Rev., 1950, 77, 717.
[4] Packard, M. E.; Arnold, J. T. Phys. Rev., 1951, 83, 210.
[5] Arnold, J. T.; Dharmatti, S. S.; Packard, M. E. J. Phys. Chem.,
1951, 19, 507.
[6] Packard, M. E.; Arnold, J. T. J. Phys. Chem., 1951, 19, 1608.
[7] Ernst, R. R. J. Magn. Reson., 1971, 4, 280.
[8] Ernst, R. R. J. Magn. Reson., 1971, 5, 398.
[9] Jeener, J. Ampere NMR Int. Summer School II, Basko Polje,
Yugoslavia, 1971.
[10] Aue, W. P.; Bartholdi, E.; Ernst, R. R. J. Chem. Phys., 1976, 64,
2229.
[11] Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of Nuclear
Magnetic Resonance in One and Two Dimensions, Clarendon,
Oxford, 1987.
[12] Moon, R. B.; Richards, J. H. J. Biol. Chem., 1973, 248, 7276.
[13] Wüthrich, K. NMR of Proteins and Nucleic Acids, Wiley, New
York, 1986.
[14] Ernst, R. R. Angew. Chem. Int. Ed. Engl., 1992, 31, 805.
[15] Wüthrich, K. Angew. Chem. Int. Ed. Engl., 2003, 42, 3340.
[16] Andrew, E. R.; Bradbury, A.; Eades, R. G. Nature, 1958, 182,
1659.
[17] Lowe, I. Phys. Rev. Lett., 1959, 2, 285.
26 Current Analytical Chemistry, 2005, Vol. 1, No. 1 Ingo Schnell

[18] Jakobson, H. J. In Encyclopedia of Nuclear Magnetic Resonance, [62] Munowitz, M. Coherence and NMR, Wiley, New York, 1988.
D. M. Grant, R. K. Harris, Ed.; Wiley, Chichester, 1996; Vol. 1, p. [63] Schnell, I.; Langer, B.; Söntjens, S. H. M.; van Genderen, M. H.
398. P.; Sijbesma, R. P.; Spiess, H. W. J. Magn. Reson., 2001, 150, 57 -
[19] Samoson, A.; Tuherm, T.; Gan, Z. Solid State Nucl. Magn. Reson., 70 (2001).
2001, 20, 130. [64] Schnell, I.; Saalwächter, K. J. Am. Chem. Soc., 2002, 124, 10938.
[20] Schnell, I.; Brown, S. P.; Low, H. Y.; Ishida, H.; Spiess, H. W. J. [65] Saalwächter, K.; Schnell, I. Solid State Nucl. Magn. Reson., 2002,
Am. Chem. Soc., 1998, 120, 11784. 22, 157.
[21] Schnell, I.; Spiess, H. W. Adv. Magn. Reson. - J. Magn. Reson., [66] Goward, G. R.; Schnell, I.; Brown, S. P.; Spiess, H. W.; Kim, H.-
2001, 151, 153. D.; Ishida, H. Magn. Reson. Chem., 2001, 39, S5.
[22] Tjandra, N.; Bax, A. Science, 1997, 278, 1111. [67] Goward, G. R.; Sebastiani, D.; Schnell, I.; Spiess, H. W.; Kim, H. -
[23] Wu, Z.; Bax, A. J. Am. Chem. Soc., 2002, 124, 9672. D.; Ishida, H. J. Am. Chem. Soc., 2003, 125, 5792.
[24] Tolman, J. R.; Flanagan, J. M.; Kennedy, M. A.; Prestegard, J. H. [68] Geen, H.; Gottwald, J.; Graf, R.; Schnell, I.; Spiess, H. W.;
Proc. Natl. Acad. Sci. USA, 1995, 92, 9279. Titman, J. J. J. Magn. Reson., 1997, 125, 224.
[25] Prestegard, J. H. Nature Struct. Biol., 1998, 5, 517. [69] Saalwächter, K.; Graf, R.; Spiess, H. W. J. Magn. Reson., 1999,
[26] Keifer, P. A. Drug Discovery Today, 1997, 2, 468. 140, 471.
[27] Shapiro, M. J.; Wareing, J. R. Curr. Opin. Chem. Biol., 1998, 2, [70] Saalwächter, K.; Spiess, H. W. J. Chem. Phys., 2001, 114, 5707.
372. [71] Bertmer, M.; Eckert, H. Solid State Nucl. Magn. Reson., 1999, 15,
[28] Lippens, G.; Bourdonneau, M.; Dhalluin, C.; Warrass, R.; Richert, 139.
T.; Seetharaman, C.; Boutillon, C.; Piotto, M. Curr. Org. Chem., [72] Fyfe, C. A.; Lewis, A. R. J. Chem. Phys. B., 2000, 104, 48.
1999, 3, 147. [73] Fechtenkötter, A.; Saalwächter, K.; Harbison, M. A.; Müllen, K.;
[29] Levitt, M. H. Spin Dynamics, Wiley, New York, 2001. Spiess, H. W. Angew. Chem. Int. Ed. Engl., 1999, 38, 3039.
[30] Mehring, M. Principles of High Resolution NMR in Solids, [74] Wind, M.; Saalwächter, K.; Wiesler, U.-M.; Müllen, K.; Spiess,
Springer, Berlin, 1983. H. W. Macromolecules, 2002, 35, 10071.
[31] Schmidt-Rohr, K.; Spiess, H. W. Multidimensional Solid State [75] Saalwächter, K. Macromol. Rapid Commun., 2002, 23, 286.
NMR and Polymers , Academic Press, New York, 1994. [76] Fischbach, I.; Saalwächter, K.; Pakula, T.; Minkin, P.;
[32] Mueller, K. T.; Sun, B. Q.; Chingas, G. C.; Zwanziger, J. W.; Fechtenkötter, A.; Müllen, K.; Spiess, H. W. J. Phys. Chem. B.,
Terao, T.; Pines, A. J. Magn. Reson.,1990, 86, 470. 2002, 106, 6408.
[33] Samoson, A.; Lippmaa, E.; Pines, A. Mol. Phys., 1998, 65, 1013. [77] Hing, A. W.; Vega, S.; Schaefer, J. J. Magn. Reson., 1992, 96,
[34] Spiess, H. W. In NMR Basic Principles and Progress, P. Diehl, E. 205.
Fluck, R. Kosfeld, Ed.; Springer, Berlin, 1978; Vol. 15, p. 55. [78] Fischbach, I.; Thieme, K.; Hoffmann, A.; Hehn, M.; Schnell, I. J.
[35] Maricq, M. M.; Waugh, J. S. J. Chem. Phys., 1979, 70, 3300. Magn. Reson., 2003, 65, 102.
[36] Lee, M.; Goldburg, W. I. Phys. Rev. A., 1965, 140, 1261. [79] Bernstein, M. A.; Trimble, L. A. Magn. Reson. Chem., 1994, 32,
[37] Bielecki, A.; Kolbert, A. C.; de Groot, H. J. M.; Griffin, R. G.; 107.
Levitt, M. H. Adv. Magn. Reson., 1990, 14, 111. [80] Berger, S. Progr. Nucl. Magn. Reson. Spectr., 1997, 30, 137.
[38] Vinogradov, E.; Madhu, P. K.; Vega, S. Chem. Phys. Lett., 1999, [81] Sattler, M.; Schleucher, J.; Griesinger, C. Progr. Nucl. Magn.
314, 443. Reson. Spectr., 1999, 34, 93.
[39] Sakellariou, D.; Lesage, A.; Hodgkinson, P.; Emsley, L. Chem. [82] Cory, D. G.; v. Os, J. W. M.; Veeman, W. S. J. Magn. Reson.,
Phys. Lett., 2000, 319, 253. 1989, 76, 543.
rd [83] Fyfe, C. A.; Skibsted, J.; Grondey, H.; Meyer, H. Chem. Phys.
[40] Samoson, A. 3 Alpine Conference on Solid-State NMR,
Lett., 1998, 281, 44.
Chamonix, France, 2003.
[84] Fritzhanns, T.; Hafner, S.; Demco, D. E.; Spiess, H. W.; Laukien,
[41] Bennett, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V.;
F. H. J. Magn. Reson., 1998, 134, 355.
Griffin, R. G. J. Chem. Phys., 1995, 103, 6951.
[85] Maas, W. E.; Bielecki, A.; Ziliox, M.; Laukien, F. H.; Cory, D. G.
[42] Detken, A.; Hardy, E. H.; Ernst, M.; Meier, B. H. Chem. Phys.
J. Magn. Reson., 1999, 141, 29.
Lett., 2002, 356, 298.
[86] Piotto, M.; Saudek, V.; Sklenar, V. J. Biomol. NMR, 1992, 2, 661.
[43] Lesage, A.; Sakellariou, D.; Hediger, S.; Eléna, B.; Charmont, P.;
[87] Sklenar, V.; Piotto, M.; Leppik, R.; Saudek, V. J. Magn. Reson.
Steuernagel, S.; Emsley, L. J. Magn. Reson., 2003, 163, 105.
Ser. A., 1993, 102, 241.
[44] De Paepe, G.; Lesage, A.; Emsley, L. J. Chem. Phys., 2003, 119,
[88] Turner, D. J. Magn. Reson., 1983, 54, 146.
4833.
[89] Kupce, E.; Freeman, R. J. Magn. Reson., 1992, 99, 644.
[45] Lesage, A.; Sakellariou, D.; Steuernagel, S.; Emsley, L. J. Am.
[90] Bodenhausen, G.; Freeman, R.; Morris, G. A. J. Magn. Reson.,
Chem. Soc., 1998, 120, 13194.
1976, 23, 171.
[46] Lesage, A.; Bardet, M.; Emsley, L. J. Am. Chem. Soc., 1999, 121,
[91] Freeman, R. In A Handbook of Nuclear Magnetic Resonance,
10987.
Longman, Harlow, 1987; p. 207.
[47] Lesage, A.; Emsley, L. J. Magn. Reson., 2001, 148, 449.
[92] Fischbach, I.; Thieme, K.; Hoffmann, A.; Hehn, M.; Schnell, I. J.
[48] Sakellariou, D.; Lesage, A.; Emsley, L. J. Am. Chem. Soc., 2001,
Magn. Reson., 2003, 65, 102.
123, 5604.
[93] Fischbach, I.; Ebert, F.; Spiess, H. W.; Schnell, I. Chem. Phys.
[49] Brown, S. P.; Perez-Torralba, M.; Sanz, D.; Claramunt, R. M.;
Chem., 2004, 5, 895.
Emsley, L. J. Am. Chem. Soc., 2002, 124, 1152.
[94] Tracz, A.; Jeszka, J. K.; Watson, M. D.; Pisula, W.; Müllen, K.;
[50] Haeberlen, U. Adv. Magn. Reson. Suppl. 1, Academic Press, San
Pakula, T. J. Am. Chem. Soc., 2003, 125, 1682.
Diego, 1976.
[95] Hoffmann, A.; Schnell, I. Chem. Phys. Chem., 2004, 5, 966.
[51] Bennett, A. E.; Griffin, R. G.; Vega, S. In NMR Basic Principles
[96] Frydman, L.; Harwood, J. S. J. Am. Chem. Soc., 1995, 117, 5367.
and Progress, P. Diehl, E. Fluck, R. Kosfeld, Ed.; Springer, Berlin,
[97] Fernandez, C.; Amoureux, J. P. Chem. Phys. Lett., 1995, 242, 449.
1994; Vol. 33, p. 1.
[98] Fernandez, C.; Amoureux, J. P. Solid State Nucl. Magn. Reson.,
[52] Dusold, S.; Sebald, A. Ann. Rep. NMR Spectrosc., 2000, 41, 185.
1996, 5, 315.
[53] Gullion, T.; Schaefer, J. J. Magn. Reson, 1989, 81, 196.
[99] Massiot, D.; Touzo, B.; Trumeau, D.; Coutures, J. P.; Virlet, J.;
[54] Gullion, T.; Schaefer, J. Adv. Magn. Reson., 1989, 13, 57.
Florian, P.; Gradinetti, P. J. Solid State Nucl. Magn. Reson., 1996,
[55] Gullion, T. Magn. Res. Rev., 1997, 17, 83.
6, 73.
[56] Gullion, T.; Concepts Magn. Res. 1998, 10, 277.
[100] Brown, S. P.; Wimperis, S. J. Magn. Reson., 1997, 128, 42.
[57] Hong, M. J. Am. Chem. Soc., 2000, 122, 3762.
[101] Medek, A.; Frydman, L. J. Braz. Chem. Soc., 1999, 10, 263.
[58] deAzevedo, E. R.; Hu, W.-G.; Bonagamba, T. J.; Schmidt-Rohr,
[102] Ashbrook, E.; Wimperis, S. J. Magn. Reson., 147, 238.
K. J. Chem. Phys., 2000, 112, 8988.
[103] Goldbourt, A.; Madhu, P. K. Monatsh. Chem., 2002, 133, 1497.
[59] Feike, M.; Demco, D. E.; Graf, R.; Gottwald, J.; Hafner, S.;
[104] Thomas, A. F. Deuterium Labelling in Organic Chemistry,
Spiess, H. W. J. Magn. Reson. A., 1996, 122, 214.
Appleton-Century-Crafts, New York, 1971.
[60] Friedrich, U.; Schnell, I.; Brown, S. P.; Lupulescu, A.; Demco, D.
[105] The outstanding work of H. Zimmermann needs to be
E.; Spiess, H. W. Mol. Phys., 1998, 95, 1209.
acknowledged here; see for example: Meinel, C.; Zimmermann,
[61] Baum, J.; Munowitz, M.; Garroway, A. N.; Pines, A. J. Chem.
H.; Schmitt, H.; Haeberlen, U. Appl. Magn. Reson., 2003, 24, 25.
Phys., 1985, 83, 2015.
[106] Müller, L. J. Am. Chem. Soc., 1979, 101, 4481.
Merging Concepts from Liquid-State and Solid-State NMR Spectroscopy Current Analytical Chemistry, 2005, Vol. 1, No. 1 27

[107] Bax, A.; Griffey, R. H.; Hawkins, B. L. J. Magn. Reson., 1983, [123] Bax, A.; Freeman, R.; Kempsell, S. P. J. Am. Chem. Soc., 1980,
55, 301. 102, 4849.
[108] Bodenhausen, G.; Ruben, D. J. Chem. Phys. Lett., 1980, 69, 185. [124] Bax, A.; Freeman, R.; Frenkiel, T. A. J. Am. Chem. Soc., 1981,
[109] Jeener, J.; Meier, B. H.; Bachmann, P.; Ernst, R. R. J. Chem. 103, 2102.
Phys., 1979, 71, 4546. [125] Thieme, K.; Schnell, I. J. Am. Chem. Soc., 2003, 125, 12100.
[110] Dhalluin, C.; Boutillon, C.; Tartar, A.; Lippens, G. J. Am. Chem. [126] Raleigh, D. P.; Levitt, M. H.; Griffin, R. G. Chem. Phys. Lett.,
Soc., 1997, 119, 10494. 1988, 146, 71.
[111] Warrass, R.; Wieruszeski, J. M.; Boutillon, C.; Lippens, G. J. Am. [127] Levitt, M. H.; Raleigh, D. P.; Creuzet, F.; Griffin, R. G. J. Chem.
Chem. Soc., 2000, 122, 1789. Phys., 1990, 92, 6347.
[112] Furrer, J.; Piotto, M.; Bourdonneau, M.; Limal, D.; Guichard, G.; [128] Costa, P. R.; Sun, B.; Griffin, R. G. J. Magn. Reson., 2003, 164,
Elbayed, K.; Raya, J.; Briand, J. P.; Bianco, A. J. Am. Chem. Soc., 92.
2001, 123, 4130. [129] 0 Macho, V.; Brombacher, L.; Spiess, H. W. Appl. Magn. Reson.,
[113] Thieme, K.; Zech, G.; Kunz, H.; Spiess, H. W.; Schnell, I. Org. 2001, 20, 405.
Lett., 2002, 4, 1559. [130] Schnell, I.; Langer, B.; Söntjens, S. H. M.; van Genderen, M. H.
[114] Lee, Y. K.; Kurur, N. D.; Helmle, M.; Johannessen, O. G.; P.; Sijbesma, R. P.; Spiess, H. W. Phys. Chem. Chem. Phys., 2002,
Nielsen, N. C.; Levitt, M. H. Chem. Phys. Lett., 1995, 242, 304. 4, 3750.
[115] Brinkmann, A.; Edén, M.; Levitt, M. H. J. Chem. Phys., 2000, [131] Percec, V.; Glodde, M.; Bera, T. K.; Miura, Y.; Shiyanovskaya,
112, 8539. I.; Singer, K. D.; Balagurusamy, V. S. K.; Heiney, P. A.; Schnell,
[116] Hardy, E. H.; Verél, R.; Meier, B. H. J. Magn. Reson., 2001, 148, I.; Rapp, A.; Spiess, H. W.; Hudson, S. D.; Duan, H. Nature, 2002,
459. 419, 384.
[117] Heindrichs, A. S. D.; Geen, H.; Giordani, C.; Titman, J. J. Chem. [132] Díez-Peña, E.; Quijada-Garrido, I.; Barrales-Rienda, J. M.;
Phys. Lett., 2001, 335, 89. Schnell, I.; Spiess, H. W. Macromol. Chem. Phys., 2004, 205, 430.
[118] Braunschweiler, L.; Ernst, R. R. J. Magn. Reson., 1983, 53, 521. [133] Díez-Peña, E.; Quijada-Garrido, I.; Barrales-Rienda, J. M.;
[119] Bax, A.; Davies, D. G.; Sarkar, S. K. J. Magn. Reson., 1985, 63, Schnell, I.; Spiess, H. W. Macromol. Chem. Phys., 2004, 205, 438.
230. [134] Rapp, A.; Schnell, I.; Sebastiani, D.; Brown, S. P.; Percec, V.;
[120] Kupce, E.; Keifer, P. A.; Delepierre, M. J. Magn. Reson., 2001, Spiess, H. W. J. Am. Chem. Soc., 2003, 125, 13284.
148, 115. [135] Fischer, M.; Lieser, G.; Rapp, A.; Schnell, I.; Mamdouh, W.; De
[121] Kupce, E.; Hiller, W. Magn. Reson. Chem., 2001, 39, 231. Feyter, S.; De Schryver, F.; Höger, S. J. Am. Chem. Soc., 2004,
[122] Macura, S.; Ernst, R. R. Mol. Phys., 1980, 41, 95. 126, 214.

Received:26 August, 2004 Accepted: 15 September, 2004