Visit https://ebookfinal.

com to download the full version and

explore more ebooks

Annual Reports on NMR Spectroscopy 64 1st

Edition Graham A. Webb (Eds.)

_____ Click the link below to download _____

https://ebookfinal.com/download/annual-reports-on-nmr-
spectroscopy-64-1st-edition-graham-a-webb-eds/

Explore and download more ebooks at ebookfinal.com

Here are some suggested products you might be interested in.
Click the link to download

Annual Reports on NMR Spectroscopy 66 1st Edition Graham

A. Webb (Eds.)

https://ebookfinal.com/download/annual-reports-on-nmr-
spectroscopy-66-1st-edition-graham-a-webb-eds/

Annual Reports on NMR Spectroscopy Volume 83 1st Edition

Webb

https://ebookfinal.com/download/annual-reports-on-nmr-spectroscopy-
volume-83-1st-edition-webb/

Annual Reports on NMR Spectroscopy Volume 87 First Edition

Webb

https://ebookfinal.com/download/annual-reports-on-nmr-spectroscopy-
volume-87-first-edition-webb/

Annual Reports on NMR Spectroscopy 62 1st Edition G.A.

Webb (Eds.)

https://ebookfinal.com/download/annual-reports-on-nmr-
spectroscopy-62-1st-edition-g-a-webb-eds/
Understanding NMR Spectroscopy 1st Edition James Keeler

https://ebookfinal.com/download/understanding-nmr-spectroscopy-1st-
edition-james-keeler/

In Vivo NMR Spectroscopy 2nd Edition Robin A. De

Graaf(Auth.)

https://ebookfinal.com/download/in-vivo-nmr-spectroscopy-2nd-edition-
robin-a-de-graafauth/

Annual reports in computational chemistry 11 1st Edition

Dixon

https://ebookfinal.com/download/annual-reports-in-computational-
chemistry-11-1st-edition-dixon/

Protein NMR Spectroscopy Principal Techniques and

Applications 1st Edition Gordon Roberts

https://ebookfinal.com/download/protein-nmr-spectroscopy-principal-
techniques-and-applications-1st-edition-gordon-roberts/

Solid State NMR Spectroscopy Principles and Applications

1st Edition Melinda J. Duer

https://ebookfinal.com/download/solid-state-nmr-spectroscopy-
principles-and-applications-1st-edition-melinda-j-duer/
Annual Reports on NMR Spectroscopy 64 1st Edition
Graham A. Webb (Eds.) Digital Instant Download
Author(s): Graham A. Webb (Eds.)
ISBN(s): 9780123743374, 0123743370
Edition: 1
File Details: PDF, 2.45 MB
Year: 2008
Language: english
 Annual Reports on
NMR SPECTROSCOPY
VOLUME 64
This page intentionally left blank
 Annual Reports on
NMR SPECTROSCOPY
VOLUME 64

Edited by
GRAHAM A. WEBB
Royal Society of Chemistry
Burlington House
Piccadilly, London, UK

Amsterdam  Boston  Heidelberg  London  New York  Oxford

Paris  San Diego  San Francisco  Singapore  Sydney  Tokyo
Academic Press is an imprint of Elsevier
Academic Press is an imprint of Elsevier
84 Theobald’s Road, London WC1X 8RR, UK
Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands
Linacre House, Jordan Hill, Oxford OX2 8DP, UK
30 Corporate Drive, Suite 400, Burlington, MA 01803, USA
525 B Street, Suite 1900, San Diego, CA 92101-4495, USA
First edition 2008
Copyright r 2008 Elsevier Ltd. All rights reserved
No part of this publication may be reproduced, stored in a retrieval system or
transmitted in any form or by any means electronic, mechanical, photocopy-
ing, recording or otherwise without the prior written permission of the
publisher
Permissions may be sought directly from Elsevier’s Science & Technology
Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44)
(0) 1865 853333; email: [email protected]. Alternatively you can
submit your request online by visiting the Elsevier web site at http://
www.elsevier.com/locate/permissions, and selecting Obtaining permission to
use Elsevier material
Notice
No responsibility is assumed by the publisher for any injury and/or damage
to persons or property as a matter of products liability, negligence or other-
wise, or from any use or operation of any methods, products, instructions or
ideas contained in the material herein. Because of rapid advances in the
medical sciences, in particular, independent verification of diagnoses and drug
dosages should be made

ISBN: 978-0-12-374337-4
ISSN: 0066-4103

For information on all Academic Press publications

visit our website at books.elsevier.com

Printed and bound in Great Britain

08 09 10 11 12 10 9 8 7 6 5 4 3 2 1
 CONTENT S

Contributors vii
Preface ix
19
1. Theoretical and Experimental Studies on F NMR Shieldings in Mineral
Glasses, Zeolites and Related Silsesquioxanes 1
J.A. Tossell
1. Introduction 2
2. Recommended Computational Methods 3
3. Fluorine in Aluminosilicate Glasses 5
4. Fluorine in Zeolites (Low-Density, Large-Pore Aluminosilicates) and in
Silsesquioxanes 11
5. Al in Hydroxyfluorides: Rosenbergite 16
6. Conclusion 18
Acknowledgment 18
References 18

2. NOE Studies of Solvent–Solute Interactions 21

J.T. Gerig
1. Introduction 22
2. Methods Used to Examine Solvent–Solute Interactions 24
3. Scope of this Chapter 26
4. Theoretical Background 26
5. Experimental Considerations 36
6. Applications 38
Acknowledgments 69
References 69

3. DFT Computations of Transition-Metal Chemical Shifts 77

Michael Bühl
1. Introduction 78
2. Methodological Aspects 80
3. 3d-Metals 85
4. 4d-Metals 99
5. 5d-Metals 111
6. Actinides 118
7. Concluding Remarks 118
Abbreviations 119
Acknowledgements 121
References 121

v
vi Contents

4. Solid-State NMR Studies of Collagen Structure and Dynamics in Isolated

Fibrils and in Biological Tissues 127
Daniel Huster
1. Introduction 128
2. Experimental Methods to Study the Molecular Dynamics in Collagen 132
3. Selected Examples 139
4. Conclusions 155
Acknowledgements 156
References 156

5. NMR Study of Beverages 161

J. Kidrič
1. Introduction 161
2. Novel Methods for the Analysis of Beverages 162
3. Analysis and Authentication 163
References 169

6. Contribution of NMR Spectroscopy to Flavour Release and Perception 173

L. Tavel, E. Guichard and C. Moreau
1. Introduction 174
2. NMR Diffusion of Small Solutes 175
3. Aroma–Macromolecule Interactions Using NMR Spectroscopy 183
4. Conclusion 186
References 186

7. Polymer Blend Miscibility 189

Jeffery L. White and Marcin Wachowicz
1. Introduction 190
2. Recent Developments in Spin-Diffusion and Polymer Blends 191
3. Blends of Polymers that Contain Heteroatoms 196
4. Binary Blends of Polymers Containing Only sp3 Carbons: Polyolefins 200
5. Blends Containing Vinyl or Diene Polymer Components 203
6. Biopolymer Blends 204
7. Conclusions 206
References 206

Subject Index 211

 CONTRIBUTORS

Michael Bühl
Max-Planck Institut für Kohlenforschung, Kaiser-Wilhelm-Platz 1, D-45470
Mülheim an der Ruhr, Germany; Present address: School of Chemistry, North
Haugh, University of St. Andrews, St. Andrews, Fife, KY16 9ST, UK
J.T. Gerig
Department of Chemistry and Biochemistry, University of California, Santa
Barbara, CA 93106, USA
E. Guichard
UMR1129 FLAVIC, ENESAD, INRA, Université de Bourgogne, 17 rue Sully,
B.P. 86510, 21065 Dijon, France
Daniel Huster
Institute of Medical Physics and Biophysics, Härtelstr. 16–18, D-04107 Leipzig,
Germany
J. Kidrič
National Institute of Chemistry, Hajdrihova 19, 1000 Ljubljana, Slovenia
C. Moreau
UMR1129 FLAVIC, ENESAD, INRA, Université de Bourgogne, 17 rue Sully,
B.P. 86510, 21065 Dijon, France
L. Tavel
UMR1129 FLAVIC, ENESAD, INRA, Université de Bourgogne, 17 rue Sully,
B.P. 86510, 21065 Dijon, France
J.A. Tossell
Department of Chemistry and Biochemistry, University of Maryland, College
Park, MD 20742, USA
Marcin Wachowicz
Deparment of Chemistry, Oklahoma State University, Stillwater, Oklahoma 74078
Jeffery L. White
Department of Chemistry, Oklahoma State University, Stillwater, Oklahoma
74078

vii
This page intentionally left blank
 PREFACE

It is my great pleasure to introduce Volume 64 of Annual Reports on NMR. As is

usual with this series of cutting edge reports the great importance of NMR in
many areas of scientific research is highlighted.
The volume commences with a chapter on Theoretical and Experimental
Studies on 19F NMR Shieldings in Mineral Glasses, Zeolites and Related
Silsequioxanes by J.A. Tossell. Chapter 2 is on NOE Studies of Solvent–Solute
Interactions by G.T. Gerig. Chapter 3 is a state-of-the-art account of DFT
Computations of Transition Metal Chemical Shifts by M. Bühl; this is followed by
a contribution from D. Huster on Solid-State NMR Studies of Collagen Structure
and Dynamics in Isolated Fibrils and in Biological Tissues; J. Kidrič reports on
NMR Studies of Beverages; the Contribution of NMR Spectroscopy to Flavour
Release and Perception is covered by L. Tavel, E. Guichard and C. Moreau;
the final contribution by J.L. White is on Polymer Blend Miscibility.
My gratitude for their very interesting reports is due to all of these
contributors. My thanks also go to the production staff at Elsevier for their help
in the timely appearance of volumes of Annual Reports on NMR Spectroscopy.

G.A. Webb
Royal Society of Chemistry
Burlington House
Piccadilly
London, UK

ix
This page intentionally left blank
 CHAPT ER 1
Theoretical and Experimental
Studies on 19F NMR Shieldings in
Mineral Glasses, Zeolites and
Related Silsesquioxanes
J.A. Tossell

Contents 1. Introduction 2
2. Recommended Computational Methods 3
3. Fluorine in Aluminosilicate Glasses 5
4. Fluorine in Zeolites (Low-Density, Large-Pore Aluminosilicates)
and in Silsesquioxanes 11
5. Al in Hydroxyfluorides: Rosenbergite 16
6. Conclusion 18
Acknowledgment 18
References 18

Abstract While much attention has been devoted to the measurement and
calculation of O NMR shieldings in solid oxides and silicates, F has been
seriously neglected. This is because, except for the metal fluorides, F enters
most solids as a low-concentration impurity or defect. Yet, the presence of
only small amounts of F can exercise enormous influence upon physical
properties, such as phase relations and viscosity, and upon chemical
reactivity and mechanism of formation. Since the 1990s a number of studies
have shown that F NMR shieldings in solids can be calculated quite
accurately using quantum methods, which use large flexible basis sets and
partially incorporate electron correlation, so long as the cluster model for
the solid is adequate. Studies directed toward the assignment of particular
unexpected peaks in specific glasses or disordered solids, as well as more
general studies of trends in F NMR shifts have both been performed. We
now have a good general understanding of the effect of local and mid-range
structure about F on its NMR shift. However, for any given site trends often

Department of Chemistry and Biochemistry, University of Maryland, College Park, MD 20742, USA
Annual Reports on NMR Spectroscopy, Volume 64 r 2008 Elsevier Ltd.
ISSN 0066-4103, DOI 10.1016/S0066-4103(08)00001-X All rights reserved.

1
2 J.A. Tossell

compete. For example, F sites can differ both in the number of other ligands
in the coordination sphere and in the degree of oligomerization, so that
considerably different F sites could give the same shift. Nonetheless, for
many different classes of solids conclusions about both structures and
energetics can now be drawn from the F NMR shieldings observed.
Keywords: quantum calculations; 19F NMR; amorphous solids; disordered
solids; silsesquioxanes; zeolites

1. INTRODUCTION

There are relatively few minerals (naturally occurring inorganic crystalline solids)
that have F-rich compositions. One unusual one, rosenbergite, will be discussed at
the end of this paper. However, F can occur at low concentration in a number of
oxide, silicate and aluminosilicate mineral solid solutions and glasses. Since the
valence of F is one (rather than the two of O, more commonly found in minerals) it
usually forms either a strong single bond or a moderate number of weak bonds to
high-coordination-number cations. The decrease in overall bond number and
bond strength caused by F substitution for O within an oxide, silicate or
aluminosilicate mineral weakens the strength of the network, decreasing its
melting point. Dissolved ‘‘volatile components’’ such as F in melts of mineral
composition, or magmas, can strongly influence their physical properties and
eruptive behavior. F is one of the most important such volatile components,
significantly influencing phase relations,1 lowering viscosity2 and increasing
diffusion coefficients of cations. F is also an important additive to industrial
silicate glasses, in which it is primarily used to lower the refractive index.3 These
effects are very important in ceramic sciences and geosciences. 19F NMR has also
been used to determine the positions and to help understand the energetics and
mechanisms for incorporation of F into open aluminosilicate cage molecules, such
as zeolites4 and silsesquioxanes,5 important in catalysis and materials science.
In fact the main focus of 19F NMR studies is to obtain structural information
which will allow us to understand the effects of substitution of small amounts of
F on physical properties such as viscosities of magmas, refractive indexes of
doped aluminosilicate glass or the stabilities of low-defect, low-density, large-
pore structures for mesoporous or microporous aluminosilicates.
Identification of the species produced can sometimes be made by comparison
of observed NMR shifts with those in well-characterized crystalline solids.
However, an important characteristic of Al oxides, silicates, aluminosilicates and
hydroxyfluorides is their tendency to form amorphous solids and/or oligomeric
solution complexes. For characterization of amorphous materials and solution
complexes multinuclear NMR can be a very powerful approach, particularly if
coupled with cross-polarization techniques to determine connectivity of different
atoms. For example, we might anticipate that for a F-bearing aluminosilicate glass
a combination of highly resolved F, Na, Si and Al NMR (with perhaps a bit of
anion–cation cross-polarization information) would allow us to construct a
 Theoretical and Experimental Studies on 19F NMR Shieldings 3

detailed picture of the medium-range structure of the glass from information about
the local structure around each atom. This procedure is sometimes effective but
suffers from the problem that the local geometries involved may not exist in any
well-studied crystalline material so that a fingerprinting approach is not adequate.
The NMR shielding range of F in solids is substantial and is strongly
influenced by nearest neighbor environment. In many cases, details of the
environment such as bond distance are important. Since F often occupies
impurity or partially occupied sites characterization of its environment with
X-rays is fraught with difficulty. Since NMR is inherently a local probe it can
sometimes give a more accurate and detailed information on the F environment
rather than the averaged picture sometimes given by X-ray diffraction. Since
many of the F environments are not well characterized by X-ray, establishing a
relationship between structure and spectra is difficult.
In this respect calculation of the NMR shielding of cations or anions for
models with different local structures can be very valuable.6 Such calculations
can be done from different perspectives, namely concentrating on accurate
reproduction of experimental data for a particular F structural site or focusing on
uncovering general trends in F NMR shielding with changes in structural
parameters. While the identity of the cations bonded to the F will always exercise
an important influence on its shielding, the number and identity of other ligands
within the coordination sphere may also be important. In many cases both the
atoms of a bond may show shifts in the same direction due to changes in the local
energy spectrum, with a lowering of occupied–unoccupied orbital energy
differences or decrease in bond gap increasing the magnitude of the deshielding,
paramagnetic contributions for both atoms.

2. RECOMMENDED COMPUTATIONAL METHODS

NMR calculations for solids can be done using both local cluster methods and
periodic methods.7 All calculations on fluorides to date have utilized a cluster
approach, partly because much of the experimental interest has been in
amorphous materials, more difficult to treat within a periodic approach. In
many cases certain atoms distant from the F have been replaced by point
charges.8 This is consistent with the expectation that many fluorine compounds
are quite ionic in nature. Ionic crystals, in general, will have larger numbers of
weak interactions and relatively high F coordination numbers. Thus, the number
of atoms or ions even in the first coordination shell can be fairly large and
calculations incorporating several shells can be quite demanding. For the
viewpoint of interpreting the computational results, large clusters may also make
qualitative interpretations somewhat dubious. The basis set and quantum
method requirements for accurate shielding calculations are much like those
for the more studied oxygen.9 Difficulties in obtaining accurate local geometries
using cluster methods must now be considered the most important factor
limiting the accuracy of calculated F NMR shieldings for most cases of
4 J.A. Tossell

disordered materials or those which are incompletely characterized on a local

scale. We have found that to accurately describe the full range of 19F shifts in
solids requires large bases (e.g., 6311+G(3df,2p) or the equivalent) and methods
which treat at least some aspects of correlation (e.g., at least at the hybrid
Hartree–Fock–density functional techniques (HF–DFT) B3LYP level or the MP2
level). We typically average the HF and B3LYP as described originally in Liu and
Nekvasil.10 This procedure gives improved agreement with the experimental
shifts for both molecules and solids. A similar scaling procedure has been
developed by Chesnut.11
Experimental findings indicate that broad trends in F shieldings can be
understood using rather modest cluster models of solids, even without
neutralizing counterions or point charges. Body et al.12 have used minimal
cluster models some of which are shown in Figure 1 for a range of metal fluorides
and find quite reasonable overall trends of calculated shielding vs. experimental

Figure 1 Example of clusters used in calculation. (a) [FAl2F8]3cluster for AlF3, (b) [FMg3F11]3
cluster for MgF2, (c) [FAlF4]2 cluster for the unshared fluorine atom of KAlF4,
(d) [FAlF4FBaFBa2F2] retained cluster for the unshared fluorine atom F7 in Ba3Al2F12 and
(e) [FBa4]7+ cluster for BaF2. Gray and black spheres represent fluorine and metal atoms,
respectively. (Adapted from ref. 12.)
 Theoretical and Experimental Studies on 19F NMR Shieldings 5

Figure 2 Calculated isotropic chemical shift values vs. experimental values for some fluoride
compounds. The solid line corresponds to diso,cal = diso,exp. The dashed line corresponds to the
linear regression. (Adapted from ref. 12.)

as shown in Figure 2. Given adequate computer resources one can construct

appropriate clusters and apply reliable quantum methodologies to them.

3. FLUORINE IN ALUMINOSILICATE GLASSES

It is generally acknowledged that in F-bearing aluminosilicate glasses the F bonds

mainly with Al, and that much of the Al is in 4-coordination and is highly
polymerized, just as in the F-free case. However, much interest attaches to the
identification of that small number of defect or unusual sites, F replacing O at
4- or 5-coordinate Al sites, which may influence the properties of the glass greatly
out of proportion to their number, i.e., the sites most effective in determining the
physical or chemical properties of the glass.
It is always sobering to look back over one’s previous work and assess
problems of concern then which have now been solved and assumptions,
implicit or explicit, since shown to be false. In ref. 13 the overall goal was
actually to assign both major and minor peaks in the 27Al NMR spectra of
some F-containing aluminosilicate glasses. This study was prompted by the
27
Al and 19F .27Al cross-polarization study of Kohn et al.14 and the 19F NMR
6 J.A. Tossell

study of Schaller et al.15 on F-bearing aluminosilicate glasses. Schaller et al.15 was

the first to apply 19F NMR to determination of the local structure about F in
F-doped aluminosilicate glasses. The overall intent of these NMR studies (and
related studies on F-bearing aluminosilicate glasses using Raman and X-ray
absorption spectroscopies) was to understand the enormous lowering of
viscosity caused by even small amount of F. In Tossell,13 a single species
AlF3(O–)2 was identified which had 27Al and 19F NMR spectra (including the
quadrupole coupling constant observed for 27Al) as well as IR spectra and
energetics which seemed most consistent with experiment. Methods were used
for the NMR calculations which are now obsolete (core-corrected common
origin HF and RPA-LORG,16 both with polarized double-zeta basis sets). In
addition, the quantum methods employed could not even really give accurate
geometries for gas-phase molecules of this size and no corrections could be
made for the effects of the external environment on structures or shieldings.
Nonetheless, such methods could yield important insights into trends in
shielding. For example, it was found that 19F NMR trends paralleled trends in
central atom (Al or Si) shieldings – increases in coordination number of the
central polyhedra increased the shielding for both Al (or Si) and F. Some results
are shown in Table 1.
It is observed that the shielding for F atoms bonded to Al increases with the
Al coordination number, and therefore with the Al NMR shielding. Substitution
of –OH for F shields the remaining F’s and the F shielding is smaller for F bound
to Si, an atom which shows a significantly larger paramagnetic deshielding than
does Al. These qualitative results have stood the test of time and of other better
calculations. Although it has never been proven, there appears to be a general
correlation between the shieldings of the two atoms of a bond, causing them to
generally shift together, as expected from naive arguments based on the
eigenvalue spectrum.
On the other hand, the precise identification of species may be illusory (even
when several properties are considered) since several species may exist which fit

Table 1 Calculated F isotropic NMR shieldings using a 3–21G* basis and 3–21G* optimized
geometries

Species, symmetry CHF Core-corrected CHF

AlF4 533 476

AlF2
5 , C4v
540(eq), 549(ap) 487 av
AlF2
5 , D3n
543(eq), 541(ap) 488 av
AlF3
6
577 505
AlFðOHÞ 3 527 471
AlF2 ðOHÞ 2 522 465
AlF3 ðOHÞ22
532, 530 477, 476
SiF2
6
528 472
Source: Adapted from ref. 13.
 Theoretical and Experimental Studies on 19F NMR Shieldings 7

the data equally well, particularly if structural effects arising from thermal
motion or medium effects are taken into account. Also, an inability to treat an
effect (such as that of the surrounding medium) unfortunately does not in itself
ensure that such effects are small. However, this study did establish that the fairly
simple methods which at that time were yielding good results for the NMR
shieldings of cations could be readily extended to anions and that there were
interesting, simple trends in F shielding associated with the local environment.
At this time only trends associated with a single-coordination polyhedron could
be explored, because of method and computer limitations. As noted at the
beginning of this manuscript, F may be weakly bonded to more than one cation,
i.e., it may be involved in various types of oligomerization or polymerization in
which it bridges coordination polyhedra at their corners or their edges. Study of
the effects of such environments on the shielding had to wait for bigger and faster
computers and better methodologies for calculating shieldings.
When Liu and Nekvasil10 returned to this problem a decade later, such
improvements had been made and they were able to study much larger systems,
using methods that yielded much more accurate geometries as well as more
accurate NMR shieldings. In particular, they used geometries calculated using
hybrid HF–DFT and they used a scaling procedure for the NMR shifts based on
both HF and hybrid HF–DFT calculations which has been previously mentioned,
along with much larger basis sets. Even using these better, yet more
computationally demanding methods, advances in software and computer
environments allowed them to study larger systems in which F was in a bridging
environment between polyhedra as well as to examine the effects of inclusion of
counterions, like Na+, and variation of structural parameters, such as Si–O–Si
angles. They examined a range of Si- and Al-centered polyhedra, attempting to
assign a 19F NMR resonance at 147 ppm, which had been tentatively assigned to
a Si–F species.
Later studies of F in amorphous SiO217 yielded a major peak at 146 ppm that
was consistent with their calculations for SiF(OSiH3)3 (their model for SiO3/2F)
groups at 146 ppm. However, there is yet no explanation for a minor peak at
135 ppm, since calculations on models for the SiF(O–)4 group suggested by
Youngman and Sen17 give F shifts around 40 ppm deshielded from those for
SiO3/2F (Tossell, unpublished results).
Both the Tossell13 and the Liu and Nekvasil10 papers followed an approach
quite common in the application of quantum techniques to problems in material
and mineral sciences: focus on new experimental data, which has been
interpreted in terms of unexpected and broadly interesting species and attempt
to test the assignment. This assures interest within the experimental community
in a way that a general theoretical paper, focusing on overall trends in properties,
does not. However, it may tempt us to do more with the calculations that we
safely can.
At this time there were also no comprehensive experimental studies of F as
impurity in either SiO2, Al2O3 and its hydrated forms or aluminosilicate
minerals. Although 19F is rather a good NMR nuclide, since it is 100% abundant
and has a spin of 1/2, the lack of understanding of details of the local structure in
8 J.A. Tossell

Figure 3 Typical 19F chemical shifts of octahedral aluminum environments with oxygen and
fluorine in the first coordination sphere, AlO6xFx, vs. x. (Adapted from ref. 18.)

materials inherently poorly crystalline or amorphous had prevented the

development of a general framework for interpreting F NMR shifts, as explained
by Chupas et al.18 However, in that work the literature was scrutinized
sufficiently to produce a very useful correlation of F NMR shift with number
of F and O within the local octahedral environment for F bonded to 6-coordinate
Al, as seen in Figure 3.18 Again, increased shielding for F was correlated with
increased numbers of F within the local octahedral, which in turn correlated with
increased shielding for the Al.
Partly to provide more systematic data Liu and Tosell,19 used the same hybrid
HF–DFT methods with large basis sets as in Liu and Nekvasil,10 to examine
much larger Al-centered polyhedral oligomeric clusters. Some of these polyhedra
are shown in Figure 4.19 The cluster method and quantum methodology
were tested by calculating the shielding for some well-characterized crystalline
fluorides, obtaining the excellent agreement with experiment shown in
Table 2 from Liu and Tossell19. Perhaps inevitably, as more species were
considered the number of distinct species that could be fitted to any
given 19F NMR peak generally increased, the opposite of the desired result.
However, at the same time interesting patterns began to emerge. Figure 519
shows such patterns, with 19F NMR shifts becoming less negative as central
polyhedron coordination numbers increase and also as the linkage of the
polyhedral elements increases, e.g., from corner to edge-shared linkages.
Substitution of Si for Al invariably leads to a deshielding of the F adjacent
to the substitution site. Thus, there were a number of factors determining
the F NMR shift: (1) the identity of the cation to which it was bound, (2) the
coordination number of its local polyhedron, (3) the ligand composition
of that polyhedron and (4) the degree and type of polymeric linkage of
the polyhedra.
Thus, it became apparent that some general principles could be formulated to
predict the 19F NMR shift from its the local and mid-range geometry about F, but
that competition between trends could lead to a considerable spread of shielding
 Theoretical and Experimental Studies on 19F NMR Shieldings 9

Figure 4 Cluster models used for corner-shared bridging F atoms in 4-coordinated Al species,
and their chemical-shift values. (Adapted from ref. 19.)

for each of the different local structures. For example, a corner shared F bonded
to a 4-coordinate Al could have the same shift as a terminal, nonbridging F
bonded to a 5-coorinate Al. There was some suggestion in the calculations of Liu
and Tossell19 that such F species could also be distinguished on the basis of their
chemical-shift anisotropy (CSA) values. Invariably the CSA is much greater for
10 J.A. Tossell

Table 2 Calculated and experimental 19F NMR results for crystalline phases
19
Phase F NMR result (ppm)

HF B3LYP Scaled Experimental

(chemical shift)

Na5Al3F14 (chiolite, 151.21) 141.6 179.2 160.4 165

Na2SiF6 129.6 177.0 153.3 152
K2SiF6 117.6 156.1 136.9 136
LiF 178.0 220.0 199.0 204
NaF 193.3 250.9 222.1 225
KF 126.1 142.8 134.5 132.4
Source: Adapted from ref. 19.

Figure 5 Diagram of the distribution of calculated 19F chemical shifts for terminal, corner-
shared, and edge-shared F atoms in various aluminum-containing species. Numbers in the
figure specify 4-, 5- or 6-coordinated Al species. (Adapted from ref. 19.)

Si–F than for Al–F bonds and it seems to decrease somewhat with increase in Al
coordination number. However, for comparison of terminal vs. bridging F’s
changes in CSA are subtle and show no obvious uniformity and the presence and
position of counterions like Na+ have surprisingly large effects.
The study of Liu and Tossell was used to assist the assignment of the 19F NMR
spectra of Na2O–Al2O3–SiO2 aluminosilicate glasses by Mysen et al.20 They
 Theoretical and Experimental Studies on 19F NMR Shieldings 11

assigned peaks to four different fluoride complexes: (1) Na–F complexes (NF);
(2) Na–Al–F complexes with Al in fourfold coordination (NAF); (3) Na–Al–F
complexes with Al in sixfold coordination with F(CF); and (4) Al–F complexes
with Al in sixfold, and possibly also fourfold coordination (TF, for tourmaline-
like F). The latter three complexes could be linked to the aluminosilicate network
via Al–O–Si bridges. Raman spectra along the mineral series were also examined
and equilibrium constants were evaluated for a number of different reactions
affecting the silicate network polymerization. One of the main conclusions was
that the dissolution of F in aluminosilicate melts was highly composition
dependent and changed greatly with the proportion of bridging O (or degree of
polymerization) of the melt.

4. FLUORINE IN ZEOLITES (LOW-DENSITY, LARGE-PORE

ALUMINOSILICATES) AND IN SILSESQUIOXANES
Zeolites are used as catalysts for a number of organic reactions, both as formed,
and with additional catalytic species added, usually to enhance their acidity.
Although they have been wildly successful as heterogeneous catalysts there is
still room for improvement, i.e., there are many useful acid-catalyzed reactions
for which there are no zeolite catalysts or for which their activity tends to be short
and unstable. Attention is constantly focused upon producing zeolite-type
materials which retain stability and have even larger pore spaces, increased
acidity and increased ease and yield in their synthesis. It was observed early that
the structures of zeolites could be modified by introducing ‘‘structure directing
agents’’ (SDA) and ‘‘mineralizers’’ (principally OH and F) into the reaction
mixture . The structure directing agents change the ring structures of the zeolites,
or more generally the class of ‘‘microporous and mesoporous materials’’ now
being employed (with P, Ge and other elements in addition to the basic O, Al and
Si). It may well be that the SDA and the mineralizers actually work together to
produce the final product structure.22,23 There is enormous activity in this field,
both experimental and theoretical.
In general, F could have at least three potentially different types of location in
zeolites, as discussed by Attfield et al.24 It could be close to the SDA cation,
essentially an ion-pair, it could be at or near the middle of an aluminosilicate cage
or it could bond to a Si of an aluminosilicate cage to produce a 5-coordinate Si.
These possibilities are shown in Figure 6.24 The particular cage chosen for display
in the figure is what is variously called a double 4-ring (D4R) or a 46, a cage
bounded by six square sides, i.e., a cube. Attfield et al. found for two all-silica
zeolite structures that the 5-coordinate form was energetically preferred and they
were able to calculate an accurate S–F bond length around 1.76 Å. They also
explained why the X-ray crystallographic studies on these compounds obtained a
much longer erroneous bond length due to structural averaging. Another
possibility (but one never observed in crystalline zeolites) is that the F might
12 J.A. Tossell

Figure 6 A schematic representation of the three types of Fion environments found in

siliceous zeolites: (i) as part of an ion pair, (ii) in the center of a small cage far from any Si
atoms and (iii) coordinated to a Si atom to form part of a pentacoordinated SiO4F unit.
(Adapted from ref. 24.)

Figure 7 Structure of tetrabutylammonium octaphenyloctasilsesquioxane fluoride (3).

(Adapted from ref. 5.)

replace one of the O bonded to Si, retaining the 4-coordinate Si and reducing the
degree of polymerization.
Such structural possibilities can be more readily explored by focusing on a
slightly less complicated group of materials often used to model zeolites, the
silsesquioxanes, which can be readily synthesized from trisilanolchloride
reactants in aqueous solution. Some silsesquioxane molecules, with general
formula (RSiO1.5)n and D4R cage structures, have been found to encapsulate
F ions at their centers.5 One such structure is shown in Figure 7 above, taken
from ref. 5. In zeolites and mesoporous silicas F are found either at the centers of
D4R cages as in octadecasil25 or near one of the apical Si atoms in larger cages.26
For slightly different reaction systems, encapsulated F has not been
 Theoretical and Experimental Studies on 19F NMR Shieldings 13

characterized but the yield of D4R cages in the synthesized zeolite or mesoporous
silica is greatly increased by the presence of fluorides in the reaction mixture.4
The presence of such D4R is correlated with low density, large void space,
and consequently high internal surface area of the mesoporous solid, increasing
their capabilities as catalysts. In order to design such low-density mesoporous
solids it is important to understand which types of silsesquioxane cages can
incorporate F, how the properties of the F are affected and how the internal
and external properties of the silsesquioxane cages are influenced by the
entrapped F.
A particularly interesting study of 19F shieldings for F in D4R-mixed
Si/Ge zeolites was performed by Pulido et al.27 They initially optimized
geometries using empirical force field techniques but also refined the local
geometries using DFT methods. They calculated the shieldings at the DFT level
using several different potentials and basis sets and for models of the D4R
of two different sizes. Their optimized geometries showed that F invariably
bonded to Ge when it was present to form a 5-coordinate species, while
for the pure Si end member the equilibrium F position was at the center
of the D4R. The existence of 5-coordinate Ge had been indicated by a preliminary
study,28 the first computational study of 19F NMR shifts in zeolites. Their
calculated F NMR shifts were in quite good agreement with experiment
for the 8Si/0Ge, 7Si/1Ge and 0Si/Ge8 members, the only ones studied
experimentally (see Table 3 of ref. 27 for details). The 8Si/0Ge species, with
the F in the center of the cage, had the most negative F shift (31 to 36
depending upon the details of model and method) compared to 38 obtained
experimentally.25
In our own studies29 we used similar DFT methodology to obtain equilibrium
geometries and the combined HF–DFT methodology, discussed earlier, to
calculate F NMR shieldings for a series of DnR species, with n = 36, as
well as for the 5-coordinate monomer and a series of single-ring compounds,
nRF, with both –H and –OH terminating groups on the Si. Some very clear
trends emerged. The shift of F in Si(OH)4F was about –115 ppm, but it
became more positive for the nRF– rings (in which the optimized position of
F was over the ring) with the effect largest for the smallest rings. As the rings
became larger the F shift approached that in free monomeric Si(OH)4F. For
F inside the DnR, when the F was bonded to a single F (n = 56) the shifts
were like those in the single-ring species, but when the equilibrium position
for F was at the center of the ring (n = 34) the shifts were much more
positive, even reaching +40 ppm (vs. the CCl3F reference) for n = 3. However,
when we constrained the F to be at the center of the large DnR, n = 56, the
F shifts became very negative, with the value of 170.8 ppm for D6R(H)
central F, beginning to approach that we calculated for a free F anion
(276.4 ppm). The simplest qualitative explanation is that the F shift is
determined at least partly by the bond strength received at F, with the
free ion value around 280 ppm for no bonds and values around 115 ppm
for a single bond from Si. Of course, some of these structures we considered
14 J.A. Tossell

Table 3 Calculated 19F NMR shieldings, isotropic shifts and spans (in ppm, average of HF and
B3LYP GIAO calculations using 6311+G(2d,p) bases evaluated at 631G* HF geometries) for
reference species and for ring and double ring silsesquioxane species with –H or –OH
termination

Type Formula 1/2(sHF+sB3LYP) dF shift OF span

Reference CCl3F 203.8 0 90.1

Free F F 480.2 276.4 0
M(OH) Si(OH)4F 318.6 114.8 109.5
3R(OH) Si3O9H6F 271.6 67.5 81.9
4R(OH) Si4O12H8F 291.1 87.3 92.4
5R(OH) Si5O15H10F 291.4 87.7 110.2
6R(OH) Si6O18H12F 294.0 90.2 111.3
D3R(H) Si6O9H6F 163.8 +40.0 55.8
D4R(H) Si8O12H8F 242.2 38.4 6.8
exp = 26
R = Ph
D5R(H) Si10O15H10F 258.0 54.2 87.9
D5R(H) central F Si10O15H10F 317.0 112.6 22.6
D6R(H) Si12O18H12F 272.5 68.7 104.7
D6R(H) central F Si12O18H12F 374.1 170.8 27.8
D3R(OH) Si6O15H6F 173.7 +30.0 45.1
D4R(OH) Si8O20H8F 263.1 59.3 0.7
exp = 38
silicalite
D5R(OH) Si10O25H10F 265.2 61.4 79.7
Opened Si10O27H14F 101.5
D5R(OH) opt. R(Si– 276.5 72.7
F) = 1.749
X-ray R(Si– 252.8 49.0
F) = 1.916 exp = 64
silicalite = 1
Source: Adapted from ref. 29.

may be hypothetical, with such low bond strength to F– that they cannot exist.
For example, the central F position in D6R(H) (Si12O18H12F–) is calculated
to be about 26 kcal/mol less stable than the apex position bonding to a single
Si at a distance around 1.75 Å. Results of these calculations are shown in
Table 3.29
A particularly interesting case is the Si10O27H14F cluster, which can be
derived from a D5R(OH) cluster by hydrolyzing two Si–O–Si linkages,
thus converting five 4-rings into one 4-ring and two 6-rings. We call this an
‘‘opened’’ D5R(OH). This cluster occurs in silicalite 1 and we calculate for it
an F shift of 72.7 ppm, which agrees well with the experimental value
of 64 ppm.30
 Theoretical and Experimental Studies on 19F NMR Shieldings 15

−
Si10O27H14F (Adapted from ref. 29)

It should also be anticipated that similar changes in coordination and F

shielding will be found for DnRs involving other elements like Ge and element
combinations like Al and P. For example, we have found that for the D3R(H)
species Ge6O9H6F the equilibrium F position is in the center of the cage, as shown
below (rather than singly bonded to Ge as in the D4R) and the F shift, evaluated in

Ge6O9H6F −
16 J.A. Tossell

the same way as in ref. 29 is 34.0 ppm, i.e., substantially deshielded from that
seen in the all Ge species of D4R type, in which F is apex bonded to a single Ge.
Thus, there are large and systematic changes in both the 19F NMR shift and
span, depending upon the size of the cage. Therefore, with a sufficient database
of experimental 19F NMR and crystal structures and/or calculated 19F NMR and
structures, it would be possible to determine the location of the F from its NMR
alone in silsesquioxanes and related mesoporous compounds. When the F is at
the center of a smaller DnR it can be strongly deshielded and its NMR span is
substantially reduced.
In ref. 29 the formation of the D4RF species Si8O12H8F is correctly predicted
to be thermodynamically favorable in toluene, but not in aqueous solution. The
favorability of encapsulation is greatest for the D4R, although the encapsulation
of F in the double 5- and 6-rings is only 5–6 kcal less favorable. Energetics for
encapsulation of F and OH are very similar for the D4R so that F and OH
will compete for this location. The mechanism for the formation of low-density
mesoporous silicas using F as mineralizer has been discussed by Corma and
Davis21 and Zones et al.31 We hope that the energetics calculated in this paper
will help to refine the proposed mechanisms.
The favorability of the encapsulation process is primarily a result of a
stabilizing electrostatic interaction between the (strained) Si8O12H8 cage and F.
Qualitatively, interaction of the F with the cage is responsible for the
deshielding of the 19F NMR.

5. AL IN HYDROXYFLUORIDES: ROSENBERGITE

In their study of aluminum hydroxyfluorides Chupas et al.18 determined 19F and

27
Al NMR spectra for the mineral rosenbergite, AlF[F0.5(H2O)0.5]4. H2O (one
polymorph of AlF3  3H2O), unusual because it has a large and fixed percentage
of F. The crystal structure of this mineral consists of infinite chains of F corner
sharing octahedral units. It thus gave us a chance to test more stringently the
prediction in Liu and Tossell19 that corner sharing fluorines were systematically
deshielded compared to terminal fluorines. The experimental F NMR showed a
major peak at about 145 ppm and additional weak peaks up to about –160 ppm.
Chupas et al. attributed this range to cis/trans ordering of fluorines with diverse
H-bonding arrangements, but our results suggested some breakdown of order
with the creation of terminal, as well as bridging F’s. In ref. 32 we used two
different experimental X-ray geometries to construct a Al2 F3 ðOH2 Þþ3 8 geometry
cluster and calculated bridging F shifts of 147 and nonbridging or terminal F
shifts of 155 to 158 ppm. We also considered a larger cluster of Al3F4 type,
obtaining essentially the same value for the F shifts (but slightly different values
for central and terminal Al shifts). The various clusters considered are shown in
Figure 8.
This study thus confirmed some of the systematics we have previously
predicted but it also showed some limitations to our approach. Optimizing the
geometries of the Al2 F3 ðOH2 Þþ3
8 , whether as a free ion or within a polarizable
 Theoretical and Experimental Studies on 19F NMR Shieldings 17

Figure 8 Ball-and-stick drawings of some of the molecules considered. (Adapted from ref. 32.)

continuum, gave Al–F bond distances and shifts for bridging and terminal F’s
that were in much poorer agreement with experiment. The observed value of the
main F NMR peak in rosenbergite (146 ppm) is also very close to the values
around 147 ppm observed for a minor peak in F-bearing anhydrous Na
aluminosilicate glasses, as discussed in ref. 19. Thus, such a corner-shared
bridging F bonded to 6-coordinate Al is a possibility for assigning this feature in
the glasses (in which the OH2 grounds would presumably be replaced by OSi
groups). We also speculated that species like that in rosenbergite could be present
in Al hydroxyfluoride aqueous solutions giving F NMR peaks around 154
to 157 ppm. When the geometry of the cluster Al2 F3 ðOH2 Þþ3 8 is optimized in
a PCM appropriate to water the Al–terminal F distance is reduced and the
Al–bridging F distance increases, so that the bridging F becomes more shielded,
with a calculated shift around 153. This thus provides a speculative means for
adapting polymeric structures observed in solids to smaller oligomeric species
18 J.A. Tossell

anticipated in solution, which will have slightly different local geometries and
therefore slightly different NMR shifts. We also established that the monomeric
unit like Al(OH2)5F2+ had a F shift around 174 (since its F is terminal) and
therefore could not explain deshielded peaks around 153 to 157 observed in
solution.

6. CONCLUSION
F is an uncommon element in aluminosilicate minerals, mineral glasses and
inorganic materials such as zeolites and silsesquioxanes, but it can strongly
influence their physical and chemical properties and reactivities. Since many of
these materials are amorphous, disordered or have complex crystal structures,
the local environment of the F is hard to determine with X-ray crystallography. In
such a case, multinuclear NMR might seem an ideal technique to determine local
and some aspects of mid-range structure. However, recovering structure from F
NMR has been difficult since the shielding systematics were poorly understood.
In recent-year theoretical studies, directed either at the identification of particular
unassigned or puzzling peaks in F NMR spectra or toward elucidation of general
trends, have given us enhanced ability to recover structure from the F NMR
spectra. For small molecules or well-characterized solids, a methodology going
somewhat beyond HF theory, incorporating electron correlation at the MP2 or
B3LYP level, along with the use of large (at least local) basis sets, seems to yield
accurate agreement with experiment. The main difficulty for less well-
characterized materials seems to be obtaining accurate geometries. The general
trends we see are (1) less negative shifts when F is bonded to Si rather than its
more common partner Al, (2) less negative shifts as the coordination number of
the central atom decreases, e.g., from 6 to 5 to 4 for Al, (3) less negative shifts as
other F’s in the coordination environment are replaced by –OH or –O-Si, (4) less
negative shifts as F’s became part of shared polyhedral elements, i.e., as they
form shared corners or edges and (5) a range of shifts from highly negative to
positive when F occupies the centers of DnR cages in silsesquioxanes or zeolites.

ACKNOWLEDGMENT
This work was supported by NSF Grant EAR-0001031 and 0539109 and DOE Grant DE-FG02-
94ER14467.

REFERENCES
1. D. A. C. Manning, Contrib. Mineral. Petrol, 1981, 76, 206–2165.
2. D. B. Dingwell, Am. Mineral., 1989, 74, 333–338.
3. T. M. Duncan, D. C. Douglas, R. Csencsits and K. L. Walker, J. Appl. Phys., 1986, 60, 130–136.
4. P. Caullet, J.-L. Paillaurd, A. Simon-Masseron, M. Soulard and J. Patarin, Comp. Rendu. Chimie.,
2005, 8, 245–266.
5. A. R. Bassindale, D. J. Parker, M. Pourny, P. G. Taylor, P. N. Horton and M. B. Hursthouse,
Organometallics, 2004, 23, 4400–4405.
 Theoretical and Experimental Studies on 19F NMR Shieldings 19

6. J. A. Tossell, Using quantum mechanical calculations of NMR properties to elucidate local and
min-range structures in amorphous oxides, nitrides and aluminosilicates, in: Computational
Materials Chemistry: Methods and Applications, L. A. Curtiss and M. S. Gordon, eds., Kluwer, Boston,
MA, 2004, pp. 1–34.
7. M. Profeta, F. Mauri and C. J. Pickard, J. Am. Chem. Soc., 2003, 125, 541–548.
8. D. Solis, J. Mol. Struct., 2002, 602–603, 159–164.
9. J. R. Cheeseman, G. W. Trucks, T. A. Keith and M. J. Frisch, J. Chem. Phys., 1996, 104, 5497–5509.
10. Y. Liu and H. Nekvasil, Am. Mineral., 2002, 87, 339–346.
11. D. B. Chesnut, Chem. Phys. Lett., 2003, 380, 251–257.
12. M. Body, G. Silly, C. Legein and J.-Y. Buzare, J. Phys. Chem., 2005, 109, 10270–10278.
13. J. A. Tossell, Am. Mineral, 1993, 78, 16–22.
14. S. C. Kohn, R. Dupree, M. G. Mortuza and C. M. B. Henderson, Am. Mineral., 1991, 76, 309–312.
15. T. Schaller, D. B. Dingwell, H. Keppler, W. Knoller, L. Merwin and A. Sebald, Geochim. Cosmochim.
Acta, 1991, 56, 701–707.
16. A. E. Hansen and T. D. Bouman, J. Chem. Phys., 1985, 82, 5033–5047.
17. R. E. Youngman and S. Sen, J. Non-Cryst. Solids, 2004, 337, 182–186.
18. P. J. Chupas, D. R. Corbin, V. N. M. Rao, J. C. Hanson and C. P. Grey, J. Phys. Chem. B, 2003, 107,
8327–8336.
19. Y. Liu and J. Tossell, J. Phys. Chem. B, 2003, 107, 11280–11289.
20. B. O. Mysen, G. D. Cody and A. Smith, Geochim. Cosmochim. Acta, 2004, 68, 2745–2769.
21. A. Corma and M. E. Davis, Chem. Phys. Chem., 2004, 5, 304–313.
22. A. Pulido, A. Corma and G. Sastre, J. Phys. Chem. B, 2006, 110, 23951–23961.
23. G. Sastre, Phys. Chem. Chem. Phys., 2007, 9, 1052–1058.
24. M. P. Attfield, C. R. A. Catlow and A. A. Sokol, Chem. Mater., 2001, 13, 4708–4713.
25. P. Caullet, J. L. Guth, J. Hazm, J. M. Lamblin and H. Gies, Eur. J. Solid State Inorg. Chem., 1991, 28,
345–361.
26. H. Koller, A. Wolker, L. A. Vallaescusa, M. J. Diaz-Cabanas, S. Valencia and M. A. Camblor, J. Am.
Chem. Soc., 1999, 121, 3368–3376.
27. A. Pulido, G. Sastre and A. Corma, Chem. Phys. Chem., 2006, 7, 1092–1099.
28. G. Sastre, A. Pulido and A. Corma, Chem. Comm., 2005, 2357–2359.
29. J. A. Tossell, J. Phys. Chem. C, 2007, 111, 3584–3590.
30. L. Delmotte, F. Soulard, A. Guth, A. Seive, A. Lopez and J. L. Guth, Zeolites, 1990, 10, 778–783.
31. S. I. Zones, R. J. Darton and S. J. Hwang, J. Phys. Chem. B, 2005, 109, 652–661.
32. J. A. Tossell and Y. Liu, Magn. Res. Chem., 2004, 42, S34–S40.
This page intentionally left blank
 CHAPT ER 2
NOE Studies of Solvent–Solute
Interactions
J.T. Gerig

Contents 1. Introduction 22
2. Methods Used to Examine Solvent–Solute Interactions 24
3. Scope of this Chapter 26
4. Theoretical Background 26
4.1 Intramolecular dipole–dipole cross relaxation 28
4.2 Intermolecular dipole–dipole cross relaxation 29
4.3 Intermolecular dipole–dipole cross relaxation in the presence
of long-lived association 31
4.4 Molecular dynamics simulations 32
4.5 Some calculations 32
5. Experimental Considerations 36
5.1 Pulse sequences 37
5.2 Other details 38
6. Applications 38
6.1 Interactions between solvent molecules 38
6.2 Dissolved xenon 39
6.3 Ion–ion and ion–solvent interactions 40
6.4 Ionic liquids 41
6.5 Cyclodextrin complexes 43
6.6 Solvation of small organic molecules 44
6.7 Selective solvation of small organic molecules 48
6.8 Selective solvation of peptides 50
6.9 Chiral interactions 58
6.10 Micelles and aggregated systems 59
6.11 Protein–small molecule interactions 62
6.12 Models of biological membranes 65
6.13 Probing molecular surfaces 67
6.14 Toward the future 68
Acknowledgments 69
References 69

Department of Chemistry & Biochemistry, University of California, Santa Barbara, CA 93106, USA
Annual Reports on NMR Spectroscopy, Volume 64 r 2008 Elsevier Ltd.
ISSN 0066-4103, DOI 10.1016/S0066-4103(08)00002-1 All rights reserved.

21
22 J.T. Gerig

Abstract Intermolecular Overhauser effects resulting from interactions of solvent

molecules with a solute are readily detected with modern instrumentation.
After a discussion of how these effects arise, applications of intermolecular
Overhauser effects in studies of solute–solvent and other interactions in
solution are reviewed. Observed effects may reflect (a) a local concentration
of solvent molecules near the solute that is different from the concentration
of the bulk, (b) kinematic behavior of solvent molecules near the solute that
is different from that characteristic of the bulk solvent, or (c) formation of
solute–solvent complexes that are long-lived relative to the time associated
with diffusive encounters of solute and solvent. In mixed solvents, a solute–
solvent Overhauser effect may indicate preferential association of one
solvent component with the solute; such interactions may be regiospecific.
Determination of solvent–solute intermolecular Overhauser effects have
been done in systems that range from solutions of monoatomic gases to
proteins. It is shown that additional experimental developments are needed
to enable unambiguous interpretation of observed effects.
Keywords: overhauser effect; solvent interactions; NOE; intermolecular
NOE; solvent effects; cross-relaxation; intermolecular dipole-dipole
interaction; selective solvation; complex formation

1. INTRODUCTION
Beginning chemistry students are taught that a solution is a homogenous mixture
of two or more materials. The component in highest concentration is called the
solvent while the lesser component is the solute. The student’s instructor tries to
make clear that a solution will not form unless there is a net favorable free energy
change resulting from interactions between the solvent and solute molecules
within the mixture, as compared to interactions present in the separated solvent
and solute. These simple concepts, of course, hide a world of complexity, for
understanding the interactions between and within molecules lies at the heart of
all chemistry and biochemistry. Further, the intrinsic disorder characteristic of the
liquid state presents a huge experimental challenge to defining the structure, and
structural fluctuations, present in a solution.
Sufficiently strong non-covalent interactions are central to the existence of
condensed phases of pure materials and solutions.1 These interactions can be the
result of hydrogen-bond formation, electrostatic effects arising between dipoles and
ions, dispersion effects resulting from electron behavior of atoms and molecules at
short range, charge-transfer interactions and entropy-driven (solvophobic, hydro-
phobic) effects.1–3 All interactions would preferably be described in quantum
mechanical terms,2 but empirical or classical approaches can often be adequate for
the purpose at hand.
A change of solvent can alter the dominant conformation of a dissolved
molecule,4–6 change the rate of a reaction,7,8 shift the position of an equilibrium,8 or
modify the interaction of a species with electromagnetic radiation. The influence of
solvent on reaction rates can be dramatic. For example, the rate constant for the SN2
 NOE Studies of Solvent–Solute Interactions 23

nucleophilic displacement reaction of bromide from methyl bromide by hydroxide

ion is reduced by about 20 orders of magnitude when the reaction takes place in
aqueous solution compared to the gas phase. A significant portion of this effect is
probably related to the hydration of hydroxide ion8,9 and the gas phase reaction is
likely more complex mechanistically than the usual picture of an SN2 reaction.10,11
A number of empirical ‘‘solvent scales’’ have been developed to correlate
observed solvent effects on reactivity with some physical or chemical property of
the solvent or solvent mixture. Such correlates include the polarity or ‘‘ionizing
power’’ of the solvent and the solvatochromic effect of solvent on a spectroscopic
probe molecule.8,12–15
Successful completion of a chemical reaction in solution may require a
mixture of solvents. An extensive literature indicates that it is usually possible to
construct mixtures of liquids that have desired physical and chemical properties,
such as density or viscosity, in addition to possessing the ability to dissolve the
materials of interest.16 Use of a solvent mixture raises questions about which
solvent component or components interact with a particular solute. The notion of
selective interactions of solutes with solvent components in mixed systems has
been extant in the chemical community for at least 50 years, and a variety of
experimental methods have been developed to explore the composition and
dynamics of the critical layer(s) of solvent molecules adjacent to a solute in a
mixed solvent. These have included (a) spectroscopic studies of dye molecules
whose transition energies are dependent on solvent composition and solvent
molecule dynamics,17–20 (b) examination of chemical reactivity,21,22 (c) mass
spectrometry,23,24 (d) molecular dynamics (MD) simulations,25 and (e) various
theoretical/thermodynamic approaches.26,27
A challenge to theoretical chemistry is a reliable prediction of the effect of a
given solvent or solvent mixture on the rate of a reaction in solution. Early
theoretical studies of solvent effects on chemical processes viewed the solute as
occupying a cavity in a continuum of solvent molecules, typically characterized
by a set of bulk solvent molecule properties. However, theory based on
continuum models probably cannot reveal subtle effects such as the difference
between the effects of ethanol and acetonitrile, two liquids which have similar
molecular weights and bulk dielectric constants. Continuum theories for mixed
solvents are rarely successful due to selective solvation and microscopic dielectric
constants that differ from the dielectric constant of the bulk medium.
Theoretical descriptions of solvent–solute interactions that recognize to some
extent the discrete nature of solvent molecules include various quantum chemical
approaches and MD calculations.2,28–30 Molecular dynamics simulations of
solvent–solute systems rely on descriptions of the interactions of a limited
number of solute and solvent molecules constrained by some procedure to a space
of defined shape and volume. The interactions between entities within that
volume are described by a potential energy function that may contain some
combination of empirical or semi-empirical terms, usually based in some manner
on experimental information. The potential energy function describing the
interaction of solvent molecules with each other is important to the success of
these approaches. It is noteworthy that a potential energy function (force field) for
24 J.T. Gerig

the interaction of water molecules obtained using only quantum mechanical

methods has been recently described.31 When used in MD simulations, this
function predicts spectroscopic and thermodynamic properties of liquid water that
are in agreement with experiment. Computational (and experimental) methods
have developed to the extent that it is possible to achieve detailed understandings,
for example, of reactivity of conformations differing only by rotation around a
single carbon–carbon bond32 or of multi-reagent processes such as water catalysis
of the reaction between hydroxyl radicals and acetaldehyde in the gas phase.33
It is natural that understanding a subject as broad and important as the roles
of solvent molecules in defining chemical reactivity has produced a large
literature. It would be impossible and inappropriate to survey that literature for
this article. However, to provide some context for the discussions which follow
the following works should be mentioned.
An early monograph by Amis summarizes treatments of solvent effects on rates
of reactions involving an ion and a dipolar reactant, two dipolar reactants, or
electron exchange.34 A multi-author series of volumes presents the status of
theoretical and experimental approaches to examination of ion–solvent interactions
through 1984.35 The volume ‘‘Molecular Theory of Solvation’’, edited by Hirata36,
provides an indication of the current status of statistical mechanical theories of
the liquid state based on reference interaction site models. The indispensable
‘‘Handbook of Solvents’’ is a comprehensive collection of short chapters on a wide
variety of topics, including theoretical and empirical descriptions of solvents and
solute–solvent interactions, solvent effects on reactions, practical information on
the industrial uses of solvents, and biological and environmental considerations in
the use of solvents.37 The third edition of Reichardt’s ‘‘Solvents and Solvent Effects
in Organic Chemistry’’ offers extensive coverage of the literature through 2001 on
topics related to solute–solvent interactions, solvent effects of chemical equilibria
and chemical reactivity, and solvent effects on various spectroscopies of organic
molecules.8 Egelstaff38 has provided an introduction to descriptions of the liquid
state; a contemporaneous volume by Balucani and Zoppi39 focuses on dynamics of
the liquid state. Older introductory texts provide useful introductions to theories of
the liquid state and experimental methods for study of this state.40,41

2. METHODS USED TO EXAMINE SOLVENT–SOLUTE INTERACTIONS

Infrared, Raman, ultraviolet-visible, NMR, ESR, microwave, and ultra-sound

spectroscopic experiments have been applied to the study of solute–solvent
interactions. A survey of spectroscopic methods that have contributed to
understanding the nature of electrolyte solutions is provided by a volume edited
by Dogonadze et al.35 Muller-Dethlefs and Hobza1 describe the application of
these and other spectroscopic methods to study of non-covalent interactions in
condensed media. Reichardt8 provides a chapter on solvent effects on ultraviolet-
visible, fluorescence, CD, infrared, ESR, and NMR spectra.
Various scattering experiments, particularly neutron scattering, provide
valuable information about the microstructure of solutions.39,42
 NOE Studies of Solvent–Solute Interactions 25

Electrospray and other ionization techniques enable mass spectrometric exami-

nation of solute–solvent clusters in the gas phase. These clusters presumably are
reflective of species present in the solution from which they are formed and their
study provides clues about the stoichiometry and energetics of solvent–solute
interaction. Self-association in binary solvent mixtures23,24,43–45 and selective
solvation of solutes23,44,45 have been examined recently by these techniques.
Kirkwood–Buff integrals are related to pair correlation functions for solution
components and can be obtained from such data as isothermal compressibilities,
molar volumes, and vapor composition.46,47 They can provide a direct,
quantitative measure of the preferential solvation of solution components.
Kirkwood–Buff integrals for a number of completely miscible aqueous binary
mixtures have been provided by Marcus.48,49
NMR techniques have proven to be especially revealing of solute–solvent
interactions.50 Indeed, the first chapter of the first volume of Annual Reviews of
NMR Spectroscopy in 1968 included discussion of hydrogen bonding and other
solvent effects on NMR spectra51 and this topic has been a staple of such reviews
ever since. The solvent used for liquid state NMR influences chemical shielding
parameters, scalar and dipolar coupling constants, and relaxation behavior.
Bagno et al.52 provide a recent survey of results that indicate how these quantities
respond to a solvent change. Solvent effects on chemical shifts and scalar
coupling constants are also discussed by Reichardt.8 An early review considered
NMR relaxation studies of solute–solvent interactions.53 NMR methods
readily provide information about translational and rotational diffusion in
solutions;54 these properties are tightly linked to molecular shape and volume
and can give indications of solute-solvent interactions, aggregation behavior, and
the binding of ligands to large molecules.
Nuclear Overhauser effects (NOEs) and the related rotating frame nuclear
Overhauser effects (ROEs) result from magnetic interactions between spin
dipoles that produce nuclear spin-lattice and transverse relaxation.55–57 Intramo-
lecular NOEs and ROEs are familiar to most chemists and biochemists – these are
effects that arise from interactions between nuclear spins that are present within
the same molecule. Intramolecular spin dipolar interactions are strongly
dependent on internuclear distance, to the extent that intramolecular NOEs and
ROEs are not detectable if the distance between the interacting spins is greater
than 5–6 Å. It is the strong internuclear distance dependence that makes
determinations of intramolecular NOEs and ROEs such powerful tools for
deducing three-dimensional (3D) structures.
Spin dipoles on different molecules may also interact, produce spin relaxation
and, therefore, lead to intermolecular Overhauser effects. Intermolecular NOEs
were first reported by Kaiser58 who used c.w. techniques to observe an
approximately 30% increase in the intensity of the chloroform proton signal
from a 2 : 9 mixture of chloroform and cyclohexane when the cyclohexane proton
resonance was saturated. In other early work, Homer and Cedeno59 determined
intermolecular NOEs in mixtures of organic solvents and demonstrated that
intermolecular interactions with protons may contribute to carbon-13 relaxation
in these systems. Using c.w. experiments, Pitner et al.60,61 provided evidence that
26 J.T. Gerig

intermolecular NOEs could be used to explore the interaction of water and

organic solvents with peptides.
Detection of the chloroform–cyclohexane NOE in Kaiser’s system was
realized by a two-dimensional (2D) NMR experiment by Macura and Ernst62,63
in 1980. Subsequent improvements in instrumentation, particularly the introduc-
tion of stable, self-shielded pulsed field gradient technologies, have facilitated
detection of solvent–solute and other intermolecular NOEs and ROEs.52,54,64
In the process, it has become clear that determination of intermolecular NOEs
can be a useful experiment for the study of selective solvation and the time-scales
for solvent–solute interactions.

3. SCOPE OF THIS CHAPTER

An effort has been made to include citations of all papers from the 10 years prior
to June 2007 that describe results of intermolecular NOE experiments directed
toward understanding solute–solvent and other intermolecular interactions.
Sufficient citations of earlier reports are included so that the reader can find other
relevant papers using contemporary searching techniques. The author readily
acknowledges that the chapter is probably not exhaustive and apologizes in
advance to those whose work should have been mentioned but was not. The goal
has been to provide a digestible and didactic summary of emerging NMR
techniques that provide information about the strength and duration of
interactions of dissolved molecules with solvent molecules or other species in
solution.

4. THEORETICAL BACKGROUND

Overhauser effects are the result of those processes in a solution that produce
spin–lattice (T1) and spin–spin (T2) nuclear spin relaxation. The relaxation times
characterizing these processes are measurable (bulk) properties of the sample.
The populations of allowed nuclear spin states in a macroscopic collection of
spins (the 1H spins in a 0.01 M solution of chloroform in CCl4, for example) are
specified by the Boltzmann distribution law. After any perturbation of these
populations, the system returns to the Boltzmann populations by exchanging
energy with its surroundings. The rate at which equilibrium is re-established is
quantified by a spin–lattice relaxation time T1.
Relaxation requires the presence of magnetic fields within or near the
molecules of interest that change at frequencies near the frequencies that
correspond to the energy of transitions between the allowed spin energy states.
Thus, relaxation of the protons in the chloroform sample mentioned will be
driven by magnetic fields within or near individual chloroform molecules that
change at a frequency of B500 MHz if the sample is in a magnetic field of 14.74 T.
The spacing between allowed nuclear spin energy levels depends on the
laboratory magnetic field in which the sample resides. As a result, relaxation
 NOE Studies of Solvent–Solute Interactions 27

processes and measured relaxation times may be dependent on the magnetic

field of the spectrometer used to measure T1 or T2.
In what follows, we limit the discussion of relaxation and NOEs to collections
of spin 1/2 nuclei in an isotropic medium. Nuclei that have spin W1/2, or are in
situations where molecular motions are anisotropic due to orientational or
positional ordering, need somewhat more complex treatments.65–68
Consider a spin system containing two spin 1/2 nuclei, A and X. The two spins
could be of the same spin type, say both 1H nuclei, but characterized by different
chemical shifts, or could be different species, perhaps A  1H and X  19F. The
spins A and X could be in the same molecule or in different molecules. When in
the same molecule, spins A and X may be spin coupled to each other, or not.
A system consisting of two spin 1/2 nuclei has four allowed nuclear spin
energy states.69–71 Taking into account all of the possible transitions between the
allowed states as avenues for relaxation leads to the Solomon equations for spin–
lattice relaxation (Equations (1) and (2))
dAZ ðtÞ
¼ rA ðAZ ðtÞ  A1 1
Z Þ  sAX ðXZ ðtÞ  XZ Þ (1)
dt

dXZ ðtÞ 1
¼ rX ðXZ ðtÞ  X1
Z Þ  sXA ðAZ ðtÞ  AZ Þ (2)
dt
Here AZ(t) and XZ(t) are components at time t of the sample magnetizations
associated with spin A and spin X along the direction of the laboratory magnetic
field (taken to be the z-axis in a Cartesian reference frame), and A1 1
Z and XZ are
these magnetizations at equilibrium (i.e., t = N). The coefficients rA and rX are
sums of transition probabilities between the allowed states, such that rA = 1/T1A,
the spin–lattice relaxation rate for spin A that would be observed in an
experiment in which the populations of energy levels associated with spin A are
inverted, with a similar meaning for rX. The term sAX(  sXA) is commonly called
the cross relaxation rate; it has the same units as 1/T1A(s1). The cross relaxation
rate depends on the difference between the probabilities of transition between the
lowest and highest allowed energy states of the AX system and the probabilities
of transition between the two intermediate allowed energy states of this two-spin
system.
The mutual interaction of the nuclear spin dipole of spin X with the nucleus of
spin A, and vice versa, leads to relaxation of both spins and potentially
contributes to a non-zero value of sAX. There are other mechanisms or processes
by which spin A and spin X can be relaxed but most of these do not contribute to
the cross relaxation term sAX.
Two ingredients determine the contribution of a nuclear spin dipole–nuclear
spin dipole interaction to relaxation. One is the strength of the interaction or,
more correctly, the Hamiltonian that describes the interaction. The dipolar
Hamiltonian for the A–X interaction is discussed in many texts.69,72–74 The second
component is a description of the time dependence of the A–X interaction. That
is, how rapidly and in what manner the magnetic field created by a dipole–dipole
interaction changes. This second aspect will be fundamentally different when
28 J.T. Gerig

both spin A and spin X are covalently bound to the same molecule, producing an
intramolecular dipolar interaction between the spins, compared to the situation
where A and X are parts of different molecules, leading to an intermolecular A–X
relaxation interaction.

4.1 Intramolecular dipole–dipole cross relaxation

Spin coupling between spins A and X has been neglected in the discussion that
follows. Neuhaus and Williamson,56 Kowalewski and Maier,75 and Ghalebani
et al.,76 among others, have considered the complications introduced by such
coupling.
Assuming that the distance between them (rAX) is invariant, the dipolar
interaction between two interacting spins A and X present in the same molecule
is modulated by the rotational motions of the molecule. As an approximation, it
is often assumed that the spins are fixed in a spherical molecule that has radius
rAX. The sphere reorients randomly and isotropically due to collisions with
solvent molecules. In this simplified but often useful description, a rotational
correlation time (tR) is used to describe the random tumbling of the AX-
containing molecule, where the time tR corresponds roughly to the average time
it takes the molecule to rotate through 1 rad in any direction.
A molecule of volume Vð¼ 4pr3AX =3Þ within a solvent of bulk viscosity Z at
temperature T has a rotational correlation time given by the Stokes–Einstein–
Debye equation (Equation (3)), where kB is Boltmann’s constant and fR is a
microviscosity factor defined by Gierer and Wirtz.77
VZf R
tR ¼ (3)
kB T
Equation (3) agrees with one’s intuition, showing that the rotational correlation
time increases as the size of the molecule holding A and X increases, and
decreases as the sample temperature is increased. An increase in solution
viscosity is expected to slow the rate of molecular tumbling, leading to an
increase in tR. The viscosity of the solution likely decreases as a sample is
warmed, thereby also contributing to the reduction of tR when temperature is
increased.
The NOE cross relaxation term sNOE
AX in Equation (1) is given by
!
NOE KAX 1 6tR tR
sAX ¼  (4)
10 r6AX 1 þ ðoA þ oX Þ2 t2R 1 þ ðoA  oX Þ2 t2R
where
 m 2
0
KAX ¼ ðgA gX _Þ2 (5)
4p
gA and gX are the gyromagnetic ratios of the interacting spins, and oA and oX are
the Larmor frequencies of these spins. The factor KAX is referred to as the dipolar
coupling constant and has the value 5.66  1011 Å6s2 when the interspin distance
is expressed in Angstrom units and both spins are 1H nuclei.
 NOE Studies of Solvent–Solute Interactions 29

In the rotating frame Overhauser experiment the cross relaxation term sROE
AX is
given by
!
K AX 1 3t R 2t R
sROE
AX ¼ þ (6)
10 r6AX 1 þ ðoA Þ2 t2R 1 þ ðoA  oX Þ2 t2R

The cross relaxation term sROE NOE

AX is always positive, while sAX can be positive,
zero or negative depending on the values of the terms within the parentheses
of Equation (4). It should be mentioned that sROE AX can only be experimentally
determined if spin A and spin X are of the same nuclear type. Typically both are
1
H nuclei. In practice, the sNOE ROE
AX and sAX terms are only detectable if the interspin
distance rAX is less than B5.5 Å.
Equations (4) and (6) assume that reorientation of the A–X spin pair in solution
is isotropic. This is clearly a simplification. Real molecules have asymmetric
shapes and undergo rotational diffusion more readily in some directions than
others. Further, the spin pair of interest may be able to undergo local motions that
are not reflected in the overall correlation time. Treatments of intramolecular
relaxation in which reorientation is anisotropic, or in which the A–X spin pair has
mobility relative to the molecule in which the spin pair resides, are available.55,57,78

4.2 Intermolecular dipole–dipole cross relaxation

When spin A and spin X are on different molecules, the time dependence of their
mutual interaction is defined by the diffusion of the molecules containing them
relative to each other. Various models have been proposed to describe diffusion
and interaction of molecules in solution.79–82 One approach posits that A and X
are at the centers of separate hard (impenetrable) spheres of radius rA and rX,
respectively. The spheres diffuse due to Brownian motion with characteristic
translational self-diffusion coefficients DA and DX; the two spins cannot come
closer than rA+rX, the ‘‘distance of closest approach’’. A correlation time (tD) for
the interaction of A and X is usually taken as the time required to diffuse the
distance rA+rX, with tD given by80,83,84
ðrA þ rX Þ2
tD ¼ (7)
ðDA þ DX Þ
The cross relaxation rate sNOE
AX for the intermolecular interaction of the spins
A and X when the signal for spin A is observed is84
6K0AX pN X
sNOE
AX ¼ ð6J ðoA þ oX Þ  J 2 ðoA  oX ÞÞ (8)
5ðDA þ DX ÞðrA þ rX Þ 2

while
6K0AX pN X
sROE
AX ¼ ð3J ðoA Þ þ 2J 2 ðoA  oX ÞÞ (9)
5ðDA þ DX ÞðrA þ rX Þ 2
Exploring the Variety of Random
Documents with Different Content
covering force on the heights of San Marcial[662], which was at the
same time menaced by Castañon with a brigade of the 4th Galician
Division, who had come up the high road from Renteria.
The French brigadier evacuated the position at once. Reille had
told him that he was only placed there to protect the retreat of Foy.
He crossed the Bidassoa, leaving only a detachment of sixty men in
the tête de pont which covered the south or Spanish end of the
bridge of Behobie; it consisted of a stone blockhouse surrounded
with palisades. The further or French end of the long bridge was
blocked by the fortified village of Hendaye.
Next morning Graham resolved that the tête de pont must be
taken or destroyed, so that the enemy should have no open road by
which he might return to Spain. While some Spanish infantry
exchanged a vague fusillade across the river with the French in
Hendaye, ten guns[663] were turned on the blockhouse. The garrison
tried to blow it up and to retreat, but their mine failed, their
commanding-officer was killed, and they were about to surrender
when Foy sent two fresh companies over the bridge to bring them
off. This was done, at the cost of four officers and 64 men, killed and
wounded while crossing the much-exposed bridge. Foy then ordered
the structure to be set on fire; and as its floor was composed of
wood this was easily done: the four arches nearest France were
burnt out by the following morning (July 1).
This evacuation, without any attempt at defence, of the
bridgehead on the Bidassoa angered the Minister of War at Paris,
and still more Napoleon, when the news reached him at Wittenberg
nine days later. Foy and Reille had men enough to have held the
heights of San Marcial, against anything short of an attack by the
whole of Graham’s and Giron’s infantry. And it is certain that Graham
would not have delivered any such assault, but would have halted in
front of them, and waited for orders from Wellington. The Emperor,
who—as we shall presently see—was set on a counter-offensive from
the moment when the news of Vittoria reached him, was wild with
wrath at the abandonment of the foothold in Spain. ‘It was insane,’
he wrote to Clarke, ‘to recross the Bidassoa: they all show
themselves as timid as women[664].’
 But whether it would have been possible for the French to hold
the Irun-Behobie-San Marcial tête de pont, when Wellington had
come up in person from the south, is another matter. Probably he
would have decided that the enemy must be thrust back across the
Bidassoa, before he dared to sit down to beleaguer San Sebastian.
And undoubtedly he had men enough to carry out that operation.
But this was no excuse for Reille’s evacuation of the position, one of
the highest strategical value, before he was compelled to do so by
force. The fact was that Reille, like most of the other French
generals, was demoralized at the moment by the recent disaster of
Vittoria, and had lost confidence both in himself and in his troops.
 SECTION XXXVII: CHAPTER III
THE EAST COAST. MURRAY AT TARRAGONA
[N.B.—For Map of Catalonia andPlan of Tarragona see Vol. IV, pp. 538
and 524]

There are certain episodes of the Peninsular War which the British
historian has to narrate with a feeling of some humiliation, but which
have to be set forth in full detail, if only for the purpose of
illustrating the manifold difficulties with which Wellington had to
cope. Of these by far the most distressing is the story of General Sir
John Murray at Tarragona.
It will be remembered that a diversion on the East Coast formed
an essential part of Wellington’s great scheme for the expulsion of
the French from Spain, and that he had devoted much care to
instructing Murray in the manner in which it was to be carried out. If
sufficient shipping to embark 10,000 men could be procured at
Alicante, the bulk of the Anglo-Sicilian army was to be transported to
Catalonia, and to strike at Tarragona, getting what aid it could from
the local Spanish forces under the Captain-General Copons. If, as
was to be expected, Suchet should fly northward from Valencia with
all his available field-army, to rescue Tarragona, the two Spanish
units in the kingdom of Murcia, Elio’s and Del Parque’s armies, were
to take the offensive against the detachments which the French
Marshal would have to leave behind him to hold down his southern
conquests. Murray might fail in Catalonia, if Suchet were rapid and
lucky in his combinations; but in that case Elio and Del Parque ought
to get possession of the city of Valencia and all its fertile plainland.
Or, on the other hand, Suchet might be loth to abandon his
advanced position, might hold it in force, and might order Decaen
and the Army of Catalonia to make head against the Anglo-Sicilian
expeditionary force. If this should happen, the Spanish generals
might be held in check, but Murray would have a free hand at
Tarragona. With the aid of Copons he ought to be able to take the
place, and to throw all the French occupation of Catalonia into
disorder. In the end, Suchet would have to evacuate Valencia, in
order to save Decaen and the Catalan garrisons. At the worst the
expedition of the Anglo-Sicilian army ought at least to have the
effect of giving Suchet so much to think about, that he would have
no attention to spare for the perils of King Joseph and the fate of
Castile.
This last minimum result was all that was achieved. Suchet, it is
true, had an anxious time during the critical days of Wellington’s
march to Vittoria, and sent no help to the King. But neither was
Tarragona taken by the Anglo-Sicilians, nor Valencia by the armies of
Elio and Del Parque. Both of those forces endured humiliating
checks, from an enemy over whom they had every strategical
advantage. And the story of Murray’s operations about Tarragona is
not the story of an honest and excusable failure, but one which
provokes bitter irritation over the doings of a British general who
showed himself not only timid and incompetent, but shifty,
mendacious, and treacherous to his allies. There is nothing in the
whole history of the Peninsular War which produces such an
unpleasant impression as the facts revealed by the minutes of
Murray’s court-martial, supplemented by certain documents which
ought to have been forthcoming at that trial, but unfortunately were
not.
But to proceed to the details of this unhappy campaign. In
obedience to Wellington’s orders, Murray began to draw his army in
to Alicante between the 25th and 27th of May, the forward positions
which the Anglo-Sicilians had held being handed over to Elio’s
troops, while those of Del Parque, who had at last been brought up
from the borders of Andalusia, took post on Elio’s left, about Yecla
and Chinchilla. Both these armies were to move forward, as soon as
Suchet should be detected in the act of detaching divisions
northward to deal with Murray’s oncoming invasion. The appearance
of the Spaniards on ground hitherto held by British outposts gave
the Marshal warning that some new plan was developing. A raid by
sea was an obvious possibility, but he could not tell whether it might
not be directed on a point as far south as Valencia or as far north as
Rosas. Till Murray showed his hand only precautionary movements
could be made. Suchet was at this moment stronger than he had
been at the time of the battle of Castalla. Warned of his danger by
the results of that fight, he had strengthened his troops on the
littoral at the expense of the garrisons inland. Severoli’s Italian
division had been ordered down from Saragossa to Valencia[665]. This
heavy draft on the northern section of his army was rendered
possible by the fact that Clausel had come far forward into Aragon in
pursuit of Mina, so that Saragossa and its region could be held with
smaller numbers than usual. The Spanish irregular forces in this
direction had full occupation found for them, by the raid of the Army
of Portugal into their sphere of operations. And this suited well with
Wellington’s general plan—the more that French troops were drawn
down to the Mediterranean, the less would there be of them
available for service in Castile, when his own blow came to be
delivered.
Murray had at his disposal in the harbour of Alicante transports
sufficient to carry much more than the force of 10,000 men, which
Wellington had named as the minimum with which a raid on
Catalonia might be attempted. He was able to embark the whole of
his own army, with the exception of the regiment of Sicilian cavalry
(he was short of horse-transports), and in addition nearly the whole
of Whittingham’s Spaniards—all indeed save one battalion[666] and
the attached squadrons. This made up a force of 14,000 infantry,
800 cavalry, and 800 artillery, with 24 field-guns and the battering
train which had been sent round from Portugal. The British
contingent was a little stronger than at the time of the battle of
Castalla, for if one battalion of the King’s German Legion had been
sent back to Sicily since that fight, Wellington had permitted Murray
to draw in the 2/67th, long in garrison at Cartagena, and had sent
him a Portuguese and a British company of artillery to man his
battering train[667]. Moreover, two squadrons of Brunswick Hussars
had arrived direct from England.
The whole force, having been swiftly embarked at Alicante, sailed
on May 31st; and being favoured with a strong south-west wind
came in sight of the high-lying Tarragona on June 2nd. The fleet of
transports ran into the bay sheltered by Cape Salou, eight miles
south of its goal. There would have been no object in risking a more
difficult disembarkation on the long open beaches at the mouth of
the Francoli river, closer to Tarragona. Before landing his main body
Murray shipped off two battalions (2/67th and De Roll-Dillon) under
Colonel Prevost, to seize the defile of the Col de Balaguer, the point
twenty miles to the south of Tarragona where the coast-road from
Tortosa curves round a steep headland between a precipice and the
sea. There was a small French fort, San Felipe de Balaguer, blocking
the Col, and Prevost was ordered to take it if he could. But its fall
was not an absolutely essential condition to the success of the siege,
for the road could be cut, blasted away, or blocked with
entrenchments north of the fort, at several points where a thousand
men could stop a whole army corps. It was desirable to take this
precaution, because the Col de Balaguer road was the only route by
which succours coming from Valencia could reach the plain of
Tarragona, without taking an immense détour inland, by paths
impracticable for artillery.
On hearing of the arrival of the British fleet off Cape Salou,
General Copons, Captain-General of Catalonia, rode down from his
head-quarters at Reus, ten miles away, to report to Murray that he
had received Wellington’s instructions, and had done his best to
carry them out. The Spanish Army of Catalonia consisted of no more
than 15,000 men, even after it had received the two battalions
which Wellington had sent to it by sea during the winter[668]. Over
5,500 of them were locked up in garrisons in the interior; many of
these were untrained recruits, and none were available for the field.
Of the remainder, Copons had brought down twelve battalions to the
neighbourhood of Tarragona, leaving only two under his second-in-
command, Eroles, to watch the French garrisons in the north[669]. He
had also with him his handful of cavalry—370 sabres; field-guns the
army had none. Altogether there were 7,000 men ready to join
Murray at once: 1,500 more might be brought in, if the French of the
northern garrisons should move down to join General Decaen at
Barcelona. Copons had certainly done all that was in his power to aid
Wellington’s scheme. Murray asked him to lend two battalions to join
the brigade that was to strike at Fort San Felipe and to block the Col
de Balaguer, and to arrange the rest so as to cover at a distance the
disembarkation of the Anglo-Sicilian army. Copons consented, and
on the next morning the whole force came ashore, Prevost’s brigade
in a creek near the Col de Balaguer, where it found the two Spanish
battalions already arrived, the rest of the army at Salou Bay. The
expeditionary force was little cumbered with transport, and had but
a small allowance of horses and mules: the infantry and some of the
field-guns with the greater part of the cavalry were ashore by the
early afternoon, and marched that same night on Tarragona, which
was invested from sea to sea, Mackenzie and Adam taking up their
position by the mouth of the Francoli, Clinton occupying the Olivo
heights, and Whittingham extending down to the shore east of the
city. The French garrison kept quiet—being of no strength sufficient
to justify the showing of a man outside the walls.
General Bertoletti had with him two battalions, one French and
one Italian[670], a company of Juramentados, two companies of
artillery, and the armed crews of three small vessels which were
blockaded in the port—they were turned on to act as auxiliary
gunners. The whole did not exceed 1,600 men. This was an entirely
inadequate force, and the defences were in an unsatisfactory
condition. After Suchet had captured the city in 1811, he had no
intention of leaving locked up behind him a garrison of the size
required for such a large fortress. The outer enceinte had been left
in the condition of ruin consequent on the siege[671], and only the
Upper City on its high cliff was occupied. Its western front, where
the breaches had been, was repaired; but the Lower City and its
fortifications remained practically untouched. All that had been done
was to patch up two isolated strong-points, the so-called Fort Royal
and Bastion of St. Carlos. These had been cut off from the mass of
the ruins, and closed in at the rear: each was armed with one gun.
The object of this was merely to prevent British ships from entering
Tarragona roadstead and mooring there. These two outlying posts,
dangerously remote from the city above, were held by no more than
a company each. Bertoletti thought for a moment of abandoning
them, since he dared not detach reinforcements from his inadequate
garrison, and his communication with the forts was across half a
mile of exposed ground: nothing was more likely than that the
enemy would slip detachments among the ruined houses and walls
of the Lower City, and dig himself in between the Upper City and the
weak outlying posts, which must inevitably fall. But reflecting on the
advantage of keeping the harbour unusable, he resolved to hold on
to them till the last minute. And his policy turned out to be justified,
for had they been evacuated, even the torpid and timorous Murray
could hardly have avoided the temptation of closing in on the Upper
City, which was in no condition to hold out for the space of nine days
during which the Anglo-Sicilian army lay in front of it.
From the first moment of his landing Murray seems to have been
obsessed with the idea that every disposable French soldier in
Catalonia and Valencia would be on his back within a very few days.
As his evidence during his court martial shows, he had a notion that
Suchet would practically evacuate Valencia, and march against him
with three-quarters of his available men, while at the same time
Decaen would abandon all Catalonia save the largest towns and
bring an even greater force to Tarragona from the north. He had
made elaborate, and in part correct, calculations as to the gross
force of the two French armies, by which he made out that the
enemy might conceivably concentrate 25,000 men against him. For
the fact that such a force would have to be scraped together from
very remote points, between which communication must be very
difficult, he made insufficient allowance. Still less did he calculate out
the handicap on the enemy caused by the fact that Suchet and
Decaen were out of touch with each other, and would obviously look
upon the problem presented to them from different points of view,—
all the former’s action being influenced by his wish to hold on to
Valencia, all the latter’s by his anxiety not to have his communication
with France cut off. Murray assumed that all roads marked on his
map would be practicable to the enemy, that Suchet’s information
would always be correct, that his troops would march every day the
possible maximum, and that they would have no difficulties
concerned with food, water, or weather. Every conceivable hazard of
war was to fall luckily for the enemy, unluckily for himself. At his trial
in 1814 he explained that he was never sanguine of success, and
that he did not expect when he sailed that he could take
Tarragona[672]. He chose to regard himself as the blind and unwilling
instrument of Wellington’s orders, which he would carry out, so far
as he could, with an expectation that they would lead to failure. And
he observed that the dominating motive which influenced all his
doings was that Wellington had written that ‘he would forgive
everything excepting that the corps should be beaten or
dispersed’[673]. Deducing from this phrase the general policy that he
must pursue, Murray came to the conclusion ‘the first principle is the
army’s safety’. He started intending to subordinate all chances of
success to the remotest risk of defeat. His mind obsessed with this
miserable prepossession, he was, in fact, defeated before he had
ever set sail. Yet he hid his resolve from his generals, even from the
senior officer, Clinton, who would have succeeded to the command if
he had fallen sick or received a chance bullet. They all complained
that he never gave them any hint of what were his intentions, or
showed them Wellington’s orders which it was his duty to carry out.
‘We were totally uninformed,’ said Clinton at the court martial, ‘of the
instructions which the Commander of the forces might have for his
guidance[674]. The first that we knew of these instructions of the
Commander of the forces, and then partially only, was when he
produced them at a Council of War on June 17th,’ after the siege of
Tarragona was over. Clinton, Mackenzie, and Adam, the three
commanders of units, were all very confident in their men: ‘they
were in the highest state of discipline and equipment[675],’ said
Mackenzie, and spoiling for a fight. Hence their entire amazement as
they discovered that Murray was intending to avoid all offensive
operations: it even led to an infraction of discipline, when Mackenzie,
Adam, and General Donkin[676] called together on their commander
to urge on him a more active policy[677], and were chased out of the
room with the words ‘it will not do’—a decision which, as Mackenzie
remarked, ‘was unanswerable’. For any further urging of the point
would have amounted to military disobedience. From the moment
when the army landed at Cape Salou on June 3rd, down to the day
of its ignominious flight on June 12th, Murray was thinking of
nothing but horrible possibilities—he was what the French call a
catastrophard.
Probably his most disastrous resolve of all was that which he
came to when first he surveyed Tarragona, on the evening of his
disembarkation. Though he noted the half-ruinous condition of the
two outworks, on which the enemy was working to the last moment,
their isolated position, and the fact that they were surrounded by all
sorts of cover easily to be seized, he resolved to lay formal siege to
them, as if they were the solid front of a regular line of defence. As
Napier remarks with perfect good sense, they should have been
dealt with as Wellington dealt with the Redoute Renaud at Rodrigo,
or Hill with the forts of Almaraz, which were far more formidable
works than the Fuerte Real or San Carlos[678]. They should have been
taken by force, escaladed, on the night of the formation of the
blockade. And being incomplete, ill-flanked, and without palisades or
ditch, and under-manned, they undoubtedly could have been
rushed. But Murray, instead of trying to gain time for the prompt
attack on the main fortress and its badly-stopped breaches,
proceeded to lay out approaches and commence batteries on the
low ground by the mouth of the Francoli river, with the object of
reducing the two outworks by regular operations. On June 4th one
battery was commenced near the sea, 600 yards from San Carlos,
another farther inland, 900 yards from the Fuerte Real. Their
construction was covered by a naval bombardment: Admiral
Hallowell moved into the roadstead a brig, three bomb-vessels, and
two gunboats, which shelled the Upper City freely, in order to
distract the attention of the garrison from the work on the batteries
by the Francoli. The fire was kept up from dusk till dawn on the 4th-
5th, and repeated on the 6th-7th during the same hours—
throughout the day the workers in the trenches kept low. This
bombardment had the desired result of permitting the batteries to
be finished without molestation, but inflicted no great damage on
the city, though it set fire to some houses, and caused casualties
both among the garrison and the inhabitants. But unaimed night-fire
had of course no effect on the walls of the enceinte.
By dawn on the 6th the two batteries were ready and opened on
the outworks with six guns: they kept up the fire all day, and with
some effect, suffering themselves very little from the distant
counter-fire of the Upper City. At dark, according to the French
narrative of the defence, parties of skirmishers came out of the
trenches, took cover in the ruins of the lower city and kept up a
persistent tiraillade against the outworks[679]. It was expected that
they would try to rush them at some chosen hour—and this would
undoubtedly have been the right policy. But nothing of the kind
happened; the British parties withdrew at dawn, and Bertoletti
began to ask himself whether the whole of these feeble operations
were not a mere demonstration, intended to draw the French armies
of Valencia and Catalonia toward Tarragona, while the real blow was
being delivered in some other quarter.
During the second night of naval bombardment, that of the 6th-
7th, Murray had ordered a third breaching battery to be built, near
the bridge of the Francoli, 300 yards closer to the Fuerte Real than
the original battery, No. 2. On the morning of the 7th all three
batteries were hard at work, and with good effect; the gorge of the
Fuerte Real was blown to bits by flank fire from the new battery, its
one gun silenced, its parapet levelled for a long space: the garrison
had to keep under cover. At dusk Major Thackeray, the senior
engineer of the army, reported that the work could be stormed at
any moment. According to Murray’s narrative Thackeray made at the
same moment the curious comment, that if the fort were escaladed
and occupied, the ground gained would be of no immediate use for
the attack on the Upper City, whose most accessible front—the
bastions of San Juan and San Pablo—might be much more easily
battered from the slopes of the Olivo hill, farther inland, than from
the low-lying site of the Fuerte Real. To storm the outworks would
cost men—to build new advanced batteries on or near them would
cost many more, since they were completely commanded by the
Upper City. ‘As the state of the fort was now such that it could be
taken whenever convenient,’ wrote Murray, ‘I consented to defer the
attack, and directed that the fire upon it should continue only
sufficiently to prevent its being re-established[680].’
The decision seems of more than doubtful wisdom. It was from
ground near the Fuerte Real that Suchet had pushed forward his
approaches in the siege of 1811, and his batteries in the lower town
had proved effective. To resolve, after four days spent on battering
the outworks, that it was better to attack from a new front, was
equivalent to sacrificing the whole of the exertions of those four
days, and starting the siege anew. But Murray accepted Thackeray’s
scheme, though the engineer warned him that he should require
fourteen days more of open trenches to reduce the Upper City. This
would relegate the crisis of the final assault to July 21st, and
meanwhile Suchet and Decaen would have had three weeks to
concentrate, instead of one. The time-problem looked very
unsatisfactory.
However the new plan was taken up—more artillery and
engineer-stores and more guns were landed on the beach west of
the mouth of the Francoli, as were also the remainder of the horses
of the cavalry and the field-guns. The disembarkation was not
always easy, as the surf on the beach grew dangerous whenever the
wind was high. On several days communication with the shore had
to be given up for many hours on end. Murray complained that the
foreign troops worked slowly and unwillingly, and that he had to
replace them with British parties, in order to get up the ammunition
from the boats to the batteries[681].
However, progress was made, and in the four days between the
7th and the 10th of June two heavy batteries were thrown up on the
high-lying slopes of the Olivo hill, in positions from which they could
bring an enfilading as well as a frontal fire to bear upon the three
corner bastions of the Upper City, including the roughly repaired
breaches in the curtain, by which Suchet had made his entry in
1811. On the 9th and again on the 10th Admiral Hallowell sent his
available vessels in-shore to resume the bombardment; and late on
the latter day the batteries on the Olivo began their fire, which they
continued on the following morning: it was very effective, all the
attacked bastions and the curtain between them being much
damaged. Meanwhile the old batteries by the Francoli overwhelmed
the Fuerte Real and San Carlos forts with renewed fire, destroying
such repairs as the garrisons had been able to carry out. The
Governor, Bertoletti, made up his mind that there would be a general
assault both on the forts and on the Upper City on the night of the
11th, and made such preparations as he could to receive it. But the
prospect was gloomy—the garrison was worn out, the walls were
crumbling, and there seemed no hope of succour from without.
Meanwhile, harassed as the Governor might be, Murray was in an
even greater state of depression. He had never believed, as he
acknowledged at his court martial, that he could succeed. The only
gleam of hope which ever entered his mind was when on the
morning of the 8th he received news that the fort of Balaguer had
been taken, so that the road by which Suchet could most easily
arrive was completely blocked. At any rate there would be a day or
two gained, since the army of Valencia would have to take difficult
and circuitous roads, instead of a short and direct one, when it came
up.
The siege of the Fort of San Felipe de Balaguer had lasted four
days. The place was small, only sixty yards square; it mounted
twelve guns, but was held by only a single company. It was in a
rocky and inaccessible position, and when Colonel Prevost had
landed with his brigade, and had reconnoitred the place on the 3rd,
he found that it must be battered by heavy guns. Shelling it with
field pieces proved unavailing. Accordingly aid was sought from the
fleet, and with great difficulty the sailors of the Invincible got two
12-pounders and a howitzer ashore, to a spot 700 yards from the
fort. But the ground was so steep and rocky that it was difficult to
construct a battery, and when the fire was opened it proved not very
effective. It was only when more guns had been hoisted up the
rocks, to a position only 300 yards from the fort, that anything was
accomplished. On the evening of the 7th a lucky bomb from a
mortar exploded one of the French magazines, and the commandant
surrendered[682]. It was thought that he might have held out longer,
as his main magazine was intact—but nearly a third of his garrison
of 150 men had been hurt, and their morale was low. With the
capture of Fort San Felipe the coast road became absolutely blocked
to any troops coming from Suchet’s direction, and Murray, as he
confessed, ‘entertained a ray of hope, not so much of the capture of
Tarragona itself, but that the expedition might prove, as Lord
Wellington wished it to prove, an effective diversion in favour of the
allied army in Valencia[683].’ It was this success, as he explained,
which encouraged him to remain two days longer in front of
Tarragona than he would otherwise have done, since the loss of the
coast-road would add two marches to the distance which Suchet had
to cover in order to join Decaen.
For already this downhearted general was obsessed with panic
fears that the enemy might be upon him at any moment. He was
gleaning in every rumour of the near approach of Suchet and
Decaen, however incredible. On the 4th he had received an express
from Prevost, with a message that the Marshal had reached Tortosa
and might be at the Col de Balaguer on the 5th. On the 7th there
was a report that a heavy French column was at Amposta, near the
mouth of the Ebro, marching north. And on the other side Decaen
was said to be in movement—‘Could I ever have expected,’ said
Murray, a year later, ‘that his army would not be united, that his
movable column would have remained divided at Gerona, Figueras,
and Barcelona[684]?’ Accordingly he wrote that night to Wellington, ‘I
am much afraid we have undertaken more than we are able to
perform. But to execute your Lordship’s orders I shall persevere as
long as prudence will permit. I have as yet no certain information of
Suchet’s movements, nor of Decaen to the eastward. But there are
reports of both, and if they prove true, in five or six days I may be
attacked by a force infinitely superior, without the hope of a retreat
in case of misfortune. I calculate that Suchet can bring into the field
24,000 or 25,000 men without difficulty[685].’ There is not a thought
in Murray’s brain of the chance that one of the enemy’s columns
might be late, and that it might be possible from his central position
to fall upon the other, with his own forces and those of Copons
united.
 As a matter of fact things were working out most favourably for
him. Suchet had seen the great transport-fleet pass the coast of
Valencia on the 31st May; but he was wholly in doubt whether the
expedition might be intending to strike at Tortosa or Tarragona, at
Barcelona or Rosas. The Marshal had to make up his mind to act,
before he knew his enemy’s objective or his exact numbers. After
many searchings of heart he resolved to keep the bulk of his forces
in the kingdom of Valencia, which he was most unwilling to give up,
and to march with a column of moderate strength to reinforce the
Army of Catalonia. He left Harispe in command in the South, with his
own division, that of Habert, Severoli’s Italians, and the bulk of his
cavalry, and resolved to move on Tortosa with Musnier’s division, his
hussar regiment, three batteries of artillery, and an improvised
brigade under General Pannetier, composed of four battalions
borrowed some time back from the Army of Catalonia, one battalion
of his own, and an odd squadron of Westphalian light horse
belonging to the Army of the Centre. The whole made up about
8,000 men[686], a force so weak that it was clear that he dared not
attack Murray till he should be joined by the troops of Decaen.
Pannetier’s brigade had a long start, as it was about Castellon and
Segorbe when the order to march arrived: it reached Tortosa on
June 8th, and Perello on the Balaguer road on the 10th. Then the
news came in that the fort of San Felipe had capitulated two days
before. Pannetier halted, and sent back the information to Suchet,
who had reached Tortosa on the 9th, escorted by a squadron of
dragoons. But Musnier’s division had taken some time to assemble,
only left Valencia on the 7th, and was far behind. Suchet was in no
small perplexity this day. The coast-road was blocked: it was the
only one by which guns and transport could move directly on
Tarragona. No news whatever had been received of Decaen and the
Army of Catalonia. Musnier could not reach Tortosa till the 11th.
Should he recall Pannetier, wait for the arrival of the rest of his
column, and then march with his whole force by the circuitous inland
road along the Ebro, by Ginestar, Tivisa, and Momblanch, so as to
reach the plain of Tarragona from the north? This would lead to
insufferable delays—the country was desolate and waterless, and
when the column reached its goal the Army of Copons, with help
perchance from Murray, would be in the way. Ten days might easily
be wasted, and meanwhile Tarragona would probably have fallen.
Information must at all costs be got, and the Marshal finally ordered
Pannetier to drop all his impedimenta, and push with his infantry
alone by mountain paths from Perello to Monroig, on the edge of the
hills overlooking the plain of Tarragona. This was to take a
dangerous risk—a brigade of 2,500 bayonets might easily be
surrounded by the Catalans and cut off. But the attention of Copons
was at this moment distracted to the other direction. Pannetier
reached the slopes above Monroig on the night of the 11th-12th,
and could get no information there—the people had fled up into the
mountains. The only fact that came to his notice was that no
cannonading could be heard next morning from the direction of
Tarragona: the most natural deduction was that the place had fallen.
But in case it might still be resisting, Pannetier ordered a row of
bonfires to be lighted along the hillsides, which he thought would be
visible to the beleaguered garrison, and would show that succour
was at hand. He then drew off again by the same rough paths by
which he had come, and returned to Valdellos, half-way back to
Perello. From thence he could only send a report of a negative kind
to Suchet, who was left none the wiser for this risky reconnaissance.
Meanwhile Musnier’s division had at last come up, but there had also
arrived the news that Del Parque and Elio were on the move in
Valencia, and were pushing back Harispe’s advanced troops. On the
12th-13th-14th the Marshal remained stationary, waiting vainly for
news, and fearing the worst. Murray, if he had but known it, had
nothing to fear from this quarter.
On the other flank also things were working out in the best
possible fashion for the besieger of Tarragona. There was a long
delay before the Army of Catalonia could prepare a field force which
could dare to face the Anglo-Sicilians. Decaen himself was at
Gerona, far away to the North, and got the news of Murray’s landing
on June 5th, by a dispatch sent him by Maurice Mathieu, Governor
of Barcelona. He had no troops under his hand save the four
battalions of Beurmann’s brigade of Lamarque’s division; he could
only collect more men by cutting down the garrisons of Figueras and
Puycerda, and calling in two brigades (those of Petit and Espert)
which were acting as flying columns at the moment, and were out of
touch, in the sub-Pyrenean foot-hills. The news that a very large
disembarkation had taken place near Tarragona struck him at first
with such dismay that he replied to Maurice Mathieu that they could
not hope to resist such a force, and that it might even be necessary
to evacuate Barcelona[687]. However, resolving to do what he could,
he ordered Beurmann’s brigade to march on Barcelona (June 8), and
sent orders around for a general shifting of the northern garrisons,
so that he hoped to collect another 4,000 men at Gerona in the
course of a week, if the flying columns could be discovered and
brought in. With this reserve he intended to come down to join the
rest of the field force about the 14th or 15th. Meanwhile he ordered
Maurice Mathieu to demonstrate against the enemy, without risking
anything, or quitting the valley of the Llobregat and the vicinity of
Barcelona.
Maurice Mathieu was the only one of the generals in Catalonia,
French or allied, who deserves any credit for his conduct in this
campaign of blunders. He resolved that the one thing necessary was
to take the offensive, and to threaten the besiegers of Tarragona,
even if he dared not venture to attack them. Beurmann’s brigade
having arrived on June 10th, he marched next day with it and four
battalions of his own to Villafranca, half-way between Barcelona and
Tarragona, and drove in Copons’ outposts—leaving his base-fortress
occupied by a very inadequate garrison. He had only 6,000 infantry
and 300 horse with him, so that he was wholly incapable of facing
Murray’s expeditionary force if it should show fight. Meanwhile he
sent letters to Decaen, telling him that the honour of the Army of
Catalonia was at stake, and that it was necessary for him to come
down from Gerona without delay, and with every available man. But
the Commander-in-Chief did not appear—he was detained by a naval
demonstration in the Bay of Rosas. For Sir Edward Pellew, then in
command of the Mediterranean Fleet, had run down from his usual
cruising ground opposite Toulon, and concentrated a numerous
squadron off the coast of the Ampurdam. He came close in-shore,
made a great display of boats, and even landed a few hundred
marines on June 8th. The news of this disembarkation filled Decaen
with the idea that the Tarragona expedition was only a snare,
intended to make him draw off all his forces southward, and that the
true blow would be struck at Rosas. He concentrated his scattered
troops with the object of parrying it, and was so long in detecting his
mistake that it was only on June 15th that he set out from Gerona,
with four battalions and one squadron, to join Maurice Mathieu,
having at last discovered that Pellew could do no real mischief.
The Governor of Barcelona, therefore, had never more than
6,000 men at his disposition during the critical days of the campaign
—June 11th-12th-13th-14th. He had, as has been mentioned above,
reached Villafranca on the first-named day. There he received what
had been longed for in vain up till now, a detailed dispatch from
Suchet, setting forth his intentions. But it was no less than twelve
days old, having been written at Valencia on May 31st, before the
Marshal had any knowledge of Murray’s strength, objective, or
intentions. It stated that he was intending to march via Tortosa, and
that he hoped that Decaen with all his disposable field force would
come down to meet him at Reus, unless indeed the Anglo-Sicilians
were aiming at some more northern point, in which case he would
have to follow them to Upper Catalonia. But he added that it was
possible that the whole naval expedition might be a mere feint,
intended to lure him northward beyond the Ebro; the fleet might
turn back again and re-land the whole of Murray’s force near
Valencia[688].
The dispatch was hopelessly out of date, and Maurice Mathieu
had no means of knowing whether Suchet had carried out his
original intentions. But his first impression was that he must seek for
the Marshal at Reus, according to the directions given. Accordingly
on June 12 he pushed his vanguard as far as Arbos, six miles in
advance of Villafranca, and some 24 miles from Tarragona. But
nothing could be heard of the Marshal’s approach, and Copons’
troops were gathering in from all sides to block the road, while
Murray was only one march away. Seized with sudden misgivings at
finding himself close to 20,000 enemies with such a trilling force,
Mathieu made up his mind that it would be madness to push on. At
10 o’clock on the night of the 12th-13th[689] he evacuated both Arbos
and Villafranca, and retreated in haste to Barcelona, to await the
arrival of Decaen and the reserve.
Murray’s miserable timidity now intervened, to save Tarragona,
just when both the French forces had found themselves foiled, and
had given up the relief of the fortress as impossible. The story reads
like the plot of a stupid theatrical farce, where every character does
the wrong thing, in order to produce absurd complications in the
situation. On the 12th Pannetier, on the hills above Monroig, heard
no bombardment, because the bombardment had ceased. And
Maurice Mathieu on the night of that same day was running away
from an enemy who had already absconded that very afternoon. So,
after coming within 35 miles of each other, the two French generals
had turned back in despair and given up the game. But Murray had
given it up also.
The bombardment had gone on very successfully throughout the
11th June, and the engineers reported at noon that the Fuerte Real
could be stormed at any moment, and that the works of the Upper
City were crumbling in many places. Orders were issued that
Mackenzie’s division should storm the Fuerte Real at 10 p.m., while
Clinton’s was to make a demonstration against the Upper City. All
arrangements were made, and after dark the troops designated for
the assault filed into the advanced trenches: the signal was to be by
a flight of rockets[690]. But the rockets never went up.
For some days Murray had been in a state of agonized indecision.
On the 9th he had received information from a trusted secret agent
in Valencia that Suchet had marched for Tortosa on the 7th with
9,000 men—this was absolutely certain and showed that all the
previous rumours from the South had been false. On the same day a
dispatch came in from Eroles at Vich to say that a French column
(Beurmann’s brigade) had left Gerona on the 8th, marching for
Barcelona, and that he intended to follow it with his own
detachment, and would join Copons before the enemy could
concentrate. The French then were on the move on both flanks—but
still far off. On the night of the 10th General Manso, commanding on
Copons’ right flank, reported the approaching departure of Maurice
Mathieu from Barcelona, but overrated his force at 10,000 men. He
undertook to detain them in the defiles beyond Villafranca for at
least one day. This news was corroborated next evening by an
officer of Whittingham’s staff, who had seen the column, estimated it
at 7,000 to 8,000 infantry, and reported that it had entered
Villafranca at 4 p.m. on the 11th[691]. But it was the movements of
Suchet which gave Murray the greater alarm: by ingenious
miscalculations[692] he had arrived at the statistical conclusion that
the Marshal had got 12,000 or 13,000 men of all arms with him,
instead of the real 8,000[693]. And when the arrival of Pannetier’s
brigade at Perello on the 10th was reported to him on the following
day, he proceeded to assume that the French column was all closed
up: ‘the Marshal with 13,000 men was within two long marches of
Tarragona[694].’ Many generals would have asked for no better
opportunity than that of being placed in a central position with some
23,000 men—the Anglo-Sicilian troops and those of Copons
exceeded that figure—while two hostile columns, one of 13,000 men
and the other of 10,000, 35 miles apart, were trying to join each
other across a difficult country by bad roads. (As a matter of fact, of
course, the French force was much smaller—no more than 8,000
men on the West, and 6,000 on the East, though Murray must not
be too much blamed for the over-estimate.) But the two governing
ideas that ruled in the brain of the unfortunate general were firstly
the memory of Wellington’s warning that ‘the one thing that could
not be forgiven would be that the corps should be beaten or
dispersed’, and secondly the fact that his army was composed of
heterogeneous material, some of which might be found wanting at a
crisis. But he concealed his downheartedness till the last moment,
both from his own lieutenants and from the Spaniards. On the 10th
he rode out to meet General Copons at Torre dem Barra, to which
place the Catalan head-quarters had just been moved, and agreed
with him to defend the line of the Gaya river against the column
coming from Barcelona. He promised to send up all his cavalry,
8,000 infantry, and two field batteries to join the Spanish army,
which was now concentrated across the two roads by which
Tarragona could be approached from Villafranca. Five battalions
were on the northern road, by the Col de Santa Cristina, four and
the three squadrons of horse on the southern route, nearer the sea.
Warnings were issued to Clinton’s and Whittingham’s divisions, and
also to Lord Frederick Bentinck, the cavalry brigadier, to be ready to
march to the line of the Gaya[695]. This looked like business, and the
spirits of the Anglo-Sicilians were high that night.
On the morning of the 11th, while the bombardment of
Tarragona was going on in a very satisfactory way, Murray rode out
again, met Copons at Vendrils, behind the Gaya, and spent much
time in inspecting the chosen positions. He disliked them; the river
was fordable in many places, and a very long front would have to be
guarded: it was considered that the French might break through by
one of the roads before the troops guarding the other could arrive.
However, adhering to his promise of the 10th, Murray ordered up
Bentinck with two squadrons of hussars and two guns to the mouth
of the Gaya, where they took over the outposts on the coast-road.
The infantry was not brought up—as the assault on Tarragona was
to take place that evening. The only news from the front received in
the afternoon were rumours that Maurice Mathieu had actually
reached Villafranca[696].
At seven o’clock Murray started home, and reached his head-
quarters before Tarragona two hours later. There he found a batch of
reports awaiting him, which finally broke down his resolution to
fight, and drove him to the ignominious flight which he had already
contemplated on more than one day during the preceding week.
Two Spanish officers had come in from the Col de Santa Cristina, to
report that the Barcelona column had certainly occupied Villafranca,
and was apparently pushing on beyond it: this was discouraging. But
the document which Murray regarded as all-important was a note
from his adjutant-general, Donkin, to the effect that peasants, who
had just come in from Perello, reported that Suchet’s column had
continued its advance: ‘This corps of infantry may be at Reus to-
morrow, if they think proper to march by Perello without artillery.
And Decaen (i. e. Mathieu), if he marched this day to Villafranca, can
also reach Reus to-morrow. This possibility may, perhaps, make
some change in your arrangements[697].’ As a matter of fact, we have
already seen that it was only Pannetier’s brigade which had gone
across the hills from Perello towards Monroig, and that this was a
mere reconnaissance.
But Donkin’s picture of Suchet and the Barcelona column joining
at Reus, in his rear, on the 12th, with forces which Murray’s
imagination raised to 25,000 men, was so much in consonance with
the fears which had been obsessing the brains of the Commander-
in-Chief throughout the last ten days, that he felt that his nightmare
was coming true. There could be no doubt that he was on the brink
of a disaster like the Ostend catastrophe of 1798[698]. Of course
Suchet’s active brain had planned his complete destruction. He tells
us in his defence[699] that he asked himself whether it was probable
that an officer of the Marshal’s activity and reputation would have
left a man more than he could help in Valencia. ‘Was it the character
of a French general to act with inadequate means when ample
means were within his reach? Was it probable that he would have
brought a small force only from the Xucar by the fatiguing march to
Tortosa? Would he have left a man idle in the south when in danger
of losing his communication with France? Every disposable corps,
many more than what might be calculated to be fairly disposable,
must be with the Marshal.’ He might have not 13,000 men but many
more. And the Barcelona column might be not 10,000 men but
13,000.
Rather than take the risk of waiting one day longer in his present
position, Murray resolved to abscond by sea, while the enemy was
still twenty miles away from Tarragona. By 9.30 p.m. he had sent his
staff officers with messages to his divisional generals to stop the
projected assault, to order all guns, horses, and stores to be got on
shipboard, and the infantry after them. He calculated, in his panic,
that he had only eighteen hours in hand: the re-embarkation must
be over by dusk on the 12th. But the most disgraceful part of his
scheme was that he had resolved to leave Copons and the Catalans
to their fate, after having brought them to the Gaya by definite
promises of assistance, without which they would never have taken
up their fighting position, and this though he had renewed his
pledges that same morning. He sent no warning of his real
intentions to his colleague, but only told him that recent information
had made it necessary for him to re-embark his battering train; but
six Anglo-Sicilian battalions should be sent out next morning, to
strengthen the force behind the Gaya.
There followed a night and a morning of confusion. No one in the
expeditionary force had hitherto suspected Murray’s wavering
confidence, except Admiral Hallowell, to whom he had on June 9th
made the remark that he imagined that they ought to be thinking
about getting away in safety rather than about prolonging the siege,
and the Quartermaster-General, Donkin, to whom on the same day
Murray had said that he suspected that they would have ere long to
depart, whereupon that officer had drawn up a secret scheme for
the details of re-embarkation[700]. But since nothing more had been
heard about such a move on the 10th or the 11th, Hallowell and
Donkin had supposed that the idea was abandoned. Clinton and the
other senior officers had been kept entirely in the dark, till the
sudden orders of 9.30 p.m. were delivered to them. At midnight
Admiral Hallowell came in to Murray’s quarters to protest against the
hasty departure, which would cause all manner of confusion and
ensure the loss of much valuable material: they parted after an
angry altercation. Colonel Williamson (commanding the artillery) also
appeared, to say that in the time given him he could get off the
guns in the batteries near the shore, but not those on the distant
Olivo. He understood the general to reply that he might be granted
some extra hours, and that the Olivo guns might be brought down
after dark on the 12th[701].
But in the morning Murray’s apprehension grew progressively
worse. He had at first intended to do something to cover Copons’
inevitable retreat, and ordered Clinton to throw out six battalions
towards the Gaya. But he soon cancelled this order, and directed
Bentinck to bring back his cavalry and guns from Altafulla without
delay. At 9 a.m. a message was sent to Williamson, to say that the
guns on the Olivo must be spiked or destroyed, as it would be
perilous to wait till night[702]. Half an hour later Murray’s notions of
retreat flickered round to a new scheme—the troops on the shore
should embark there; but those on the northern heights—the
divisions of Clinton and Whittingham—should march to the Col de
Balaguer via Constanti, and take ship in the much better harbourage
behind Cape Salou. Half an hour later he abandoned this scheme,
and ordered them down to the beach by the Francoli, there to
embark without delay.
This dispatch reduced Clinton to a state of cold rage: at the court
martial in 1814 he produced seven separate orders which he had
received between dawn and 1.30 that day: they were all
contradictory, and deserve record as showing Murray’s state of mind
during the critical hours. (1) The first, received early, directed him to
take six battalions towards the Gaya, to cover the retreat of Copons
from Altafulla. (2) The second, sent off at 9 o’clock, told him not to
execute this march, but to wait till the Spaniards had cleared off, and
then to move, not with six battalions but with the whole of his own
and Whittingham’s divisions, to Constanti[703]. (3) The third was to
the effect that the baggage should be sent to the Col de Balaguer, to
which the whole army would now proceed[704]. (4) Half an hour later
Clinton was told to cancel the last two dispatches, and to come
down to head-quarters on the nearest beach. (5) A supplementary
verbal order directed that the guns in the Olivo batteries should be
spiked. (6) Twenty minutes later Clinton was told that the guns
might still be saved: Whittingham’s Spaniards should remain on the
Olivo and guard them till dusk, when the artillery would try to get
them down to the shore. (7) Lastly, at 1.30, the final order cancelled
the sixth, it directed (once more) that Whittingham’s troops must
follow Clinton’s for instant embarkation, and that the guns should be
spiked without delay. This was done, to the intense disgust of the
gunners, who had been getting everything ready for an orderly
retreat after dusk. Seventeen heavy pieces in good condition, and
one more which had been disabled, were spiked and left in the Olivo
batteries, while the infantry hurried down to the shore.
The momentary wavering in Murray’s orders during the early
morning, when he seemed inclined to risk a longer stay, and to
march Clinton’s division and the baggage by land to the anchorage
by the Col de Balaguer, was apparently caused partly by new
remonstrances from Hallowell, partly by an interview with some of
his subordinates, somewhere between 8 and 9 a.m., Mackenzie,
Adam, Donkin, and Thackeray, the chief engineer, entered the house
of the Commander-in-Chief together, and Adam, as their spokesman,
urged him with great heat not to embark, but to advance, join
Copons, and attack the French on the Gaya with every available
man. Murray replied that if he did so, the French would refuse to
fight and give back toward Barcelona; while Suchet, coming from
the other direction, would cut him off from the fleet and relieve
Tarragona. But if, on the other hand, he were to march against
Suchet, and stop him, then the Barcelona column would relieve
Tarragona and get the expeditionary force separated from its
transports. Either course would be equally ruinous. He also said that
Wellington had told him ‘not to commit the army’. The generals
withdrew—further insistence would have amounted to military
insubordination. The final hurry-scurry order, to destroy the guns
and embark pell-mell, was caused by news from Copons that the
French at Villafranca had advanced as far as Arbos, only ten miles
from the Gaya, and seemed still to be coming on.
During the night and the morning that followed all the guns from
the lower batteries, much valuable material, some of the cavalry,
and the infantry of Adam, were got aboard. The shipment of artillery
stores was still proceeding at 10 o’clock, when Murray ordered that
nothing but men should now be embarked—all else must be
abandoned. Admiral Hallowell, who was superintending the work on
the beach in person, was much incensed with this resolve, and took
it upon himself to direct the sailors in the boats to refuse infantry,
and to go on lading stores. Then came a deadlock—Hallowell said
that there was plenty of time to take off everything: Donkin, coming
down from Murray’s quarters, maintained that the French were only
two and a half hours’ march away, and that the infantry must be got
off at all costs: horses and mules might have to be shot, and food
and ammunition abandoned[705]. There was much disputing, but the
naval men, obeying the admiral, continued to embark horses and
artillery material till midday, when Mackenzie’s infantry began to
pour down to the beaches, followed by Clinton’s and Whittingham’s
battalions[706]. All through the afternoon the shipping off of troops
continued without any interruption, till only a rearguard of 500 of
Clinton’s men was left on shore. No news of the French coming to
hand, there was a relaxation of the wild hurry which had prevailed
between 9 a.m. and 2 p.m. Murray permitted Bentinck to take that
part of the cavalry which had not yet embarked round to the
anchorage by the Col de Balaguer by land[707], and sent on behind
them twelve field-guns escorted by a half battalion of the 2/27th.
Meanwhile, the troops having all got off, the boats began to load
again with transport mules, entrenching tools, spare shot, platform
timbers, sand bags, and biscuit. The rear guard was not taken off till
late at night; and, even after it was gone and dawn had come, the
sailors continued to find and take off various valuable leavings[708].
No molestation whatever was suffered from the garrison, who
regarded the whole movement as inexplicable, and only crept out
with caution after the last British troops had disappeared, to find 18
spiked guns and some artillery stores in the Olivo batteries, and a
certain amount of flour and beef barrels left on the shore. The most
distressing part of the chronicle of this wretched day’s work is
Murray’s dealings with Copons. He had induced the Captain-General
to bring his army to the line of the Gaya, by the promise of
supporting him with 8,000 men, renewed on the morning of the
11th. At dawn on the 12th the message was sent that six battalions
only could be spared. An hour or two later Copons received the
crushing news that the whole expeditionary army was about to re-
embark, abandoning him to his own resources. If Murray’s view of
the situation had been correct, the Spaniards must have been
caught between Suchet advancing on Reus and Mathieu converging
on the same point from Villafranca. His moral guilt, therefore, was
very great. But, as a matter of fact, nothing disastrous happened.
On hearing that he had been left in the lurch, Copons withdrew the
troops which he had at Altafulla on the coast-road, to join those at
the Col de Santa Cristina, leaving the way to Tarragona open to
Maurice Mathieu, and preparing to retreat into the mountains if
necessary. He kept a close look-out upon the French column, and
was astonished on the morning of the 13th to learn that, after
sending a vanguard to Arbos on the previous day, six miles down the
southern road, it had turned back at ten o’clock at night, and was
now in full retreat on Barcelona. The crisis was over for the moment.
Copons returned to his old head-quarters at Reus, threw out a slight
screen of troops to the line of Gaya, and reoccupied Villafranca with
a few cavalry. He was quite ready to retreat into the mountains once
more if any untoward developments should supervene.
 On June 13th Murray had all his army on shipboard,[709] save
Prevost’s two battalions at the Col de Balaguer, and the guns and
cavalry with Bentinck, which had marched by land to that same
pass. Wellington’s orders told him that if he failed at Tarragona he
was to return to Valencia, join Elio and Del Parque, and fall upon the
diminished French force on the Xucar before Suchet could get back.
It was his obvious duty to pick up Prevost’s and Bentinck’s
detachments, and to depart at once. He did nothing of the kind, but
proceeded to employ himself for several days in minor operations, in
the vain hope of redeeming the disgrace of June 12th.
On the morning of the 13th Murray received news from Colonel
Prevost at San Felipe to the effect that the French column in front of
him (Pannetier’s brigade), about 3,000 strong, had marched from
Perello by mountain paths eastward, but had not been followed by
Suchet’s main body. Prevost thought that it might be intending to cut
him off from the rest of the army by dropping down into the coast-
road in his rear. Pannetier’s movement was also reported by
Bentinck, whose flank patrols had run against similar parties of the
French while on their way to the Col de Balaguer. It was necessary
to bring off Prevost, the cavalry, and the guns by sea, if the French
were trying to get round them. So Murray asked Admiral Hallowell to
send ships in-shore to the Col, and with them part of Mackenzie’s
division[710], to act as a covering force for the re-embarkation of
Prevost’s detachment. These troops—apparently three battalions—
were landed near the fort late on the evening of the 13th.
Now during that day Pannetier, conscious that he was too far
away from his chief, and quite ‘in the air,’ had drawn back from near
Monroig to Valdellos, ten miles nearer to Perello. The same idea had
occurred to Suchet, who simultaneously brought up his main body
from Tortosa to Perello, though he knew that the coast-road was
blocked by the British force holding Fort San Felipe. But on the
morning of the 14th the Marshal, exploring toward the Col, sighted
not only Mackenzie’s infantry on shore, but the whole transport fleet
lying off the coast from opposite the fort as far as Cape Salou—180
ships small and great, as he counted. A frigate and two brigs took a
number of long shots at the Marshal and his large escort, which had
to retire in haste. Suchet saw the whole expeditionary force before
him, and recognized that he was too weak to tackle it. Wherefore he
sent orders to Pannetier to fall back and rejoin him—and came to a
halt. Early on the morning of the 15th he wrote to Decaen:
‘The loss of the Col de Balaguer has foiled all my plans. The
English fleet fired more than a thousand shot at us. Menaced by
forces of four times my own strength, I wish to know what you
propose to do. The enemy [Del Parque] has attacked my lines on the
Xucar[711].’
This last fact, at the moment, was worrying Suchet even more
than the unknown fate of Tarragona, for Harispe had just reported
that the two Spanish armies in Valencia had advanced against him in
force, and on a long front. Now, while the Marshal halted in
indecision, Murray was seized with a spasmodic fit of energy—he
knew that Maurice Mathieu was for the moment out of the game; he
saw that Suchet was blocked, and that Pannetier was in a
dangerously advanced position. He conceived the idea that he might
land more troops at the Col, and strike in between the French main
body and the brigade at Valdellos. To the astonishment of his
subordinates he ordered the rest of Mackenzie’s troops and all
Clinton’s to be landed, as also the remainder of the cavalry, and
instructed Mackenzie to make a forced march on Valdellos and beat
up Pannetier’s camp. The General did so, but arrived only just in
time to see the French rearguard absconding on the 15th[712]. This
demonstration of energy having failed, as it was pretty certain to do
(for events like the surprise of Arroyo dos Molinos are not common),
Murray had another inspiration. On the evening of the 15th[713] he
suggested to Admiral Hallowell that the garrison of Tarragona must
now be quite off their guard, that they had not been strengthened,
and that it might be possible to re-embark 5,000 men secretly at
Fort Balaguer, land them at Cambrils, and make a sudden dash at
Tarragona by escalade or surprise. The Admiral answered that as
Murray had not felt himself equal to storming Tarragona when the
whole army was with him, and the batteries playing on the place, he
(Hallowell) did not think him equal to it now with 5,000 men[714]. The
idea was dropped: it was not without its merits, but though Picton or
Craufurd might have succeeded in such a desperate stroke, Murray
would certainly have failed from lack of nerve at the decisive
moment, as the Admiral meant to insinuate.
This wretched campaign was now nearing its end—but still
reserved two extraordinary surprises for its last days. On the 16th
Suchet, much impressed with the strength which Murray had shown,
judging from the stroke at Valdellos that he was thinking of taking
the offensive, and harassed by fresh news of Del Parque’s advance
on Valencia, retired from Perello, and took his main body back across
the Ebro to Amposta, which he reached on the 17th. He left only
Pannetier’s brigade and the hussar regiment to watch Murray. Thus
the Anglo-Sicilians had opposite them on the south only 3,000 men,
from whom no possible danger could be apprehended. But their
General had no idea as yet that only a rearguard lay before him, and
imagined that the Marshal and his main body were close in
Pannetier’s rear.
The second surprise took place on the other front. Maurice
Mathieu had got back to Barcelona by a forced march on the 13th.
On the next day spies brought him the news that Murray had raised
the siege of Tarragona, and gone away with his whole army by sea.
There was nothing to prevent him from relieving Bertoletti’s garrison,
save Copons’ troops holding the line of the Gaya. Therefore,
although Decaen had not yet arrived from the North, and indeed had
written that he could not start from Gerona until the 17th, Mathieu
resolved to make a second attempt to reach Tarragona. He could still
count on no greater force than the 6,000 men with which he had
made his first fruitless expedition, but he thought this sufficient to
deal with the Catalans. He moved once more upon Villafranca early
on the 15th, and reached that place the same night. Next day a
forced march of twenty-four miles brought him to Tarragona
unopposed; for Copons would not fight when unsupported by
Murray, and had withdrawn his troops to Valls. At Tarragona Mathieu
learned that the Anglo-Sicilians had not disappeared entirely from
Catalonia, as he had hitherto believed, but were concentrated at and
about the Col de Balaguer. Though no news had been received from
Suchet, it might be conjectured that Murray had landed at the Col in
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebookfinal.com