Paramagnetic NMR of Phenolic Oxime Copper Complexes: A Joint Experimental and Density Functional Study

Article abstract of DOI:10.1002/chem.201602567

¹H and¹³C pNMR properties of bis(salicylaldoximato)copper(II) were studied in the solid state using magic-angle-spinning NMR spectroscopy and, for the isolated complex and selected oligomers, using density-functional theory at the PBE0- 1/3//PBE0-D3 level. Large paramagnetic shifts are observed, up to δ(¹H)=272 ppm and δ(¹³C)=1006 ppm (at 298 K), which are rationalised through spin delocalisation from the metal onto the organic ligand and the resulting contact shifts arising from hyperfine coupling. The observed shift ranges are best reproduced computationally using exchange-correlation functionals with a high fraction of exact exchange (such as PBE0- 1/3). Through a combination of experimental techniques and first-principles computation, a near-complete assignment of the observed signals is possible. Intermolecular effects on the pNMR shifts, modelled computationally in the dimers and trimers through effective decoupling between the local spins via A-tensor and total spin rescaling in the pNMR expression, are indicated to be small. Addition of electron-donating substituents and benzannelation of the organic ligand is predicted computationally to induce notable changes in the NMR signal pattern, which suggests that pNMR spectroscopy can be a sensitive probe for the spin distribution in paramagnetic phenolic oxime copper complexes.

Full text of DOI:10.1002/chem.201602567

DOI: 10.1002/chem.201602567
Full Paper
&
Materials Science
Paramagnetic NMR of Phenolic Oxime Copper Complexes: A Joint
Experimental and Density Functional Study
[
a]
[a]
[a]
[a]
Michael Bꢀhl,* Sharon E. Ashbrook,* Daniel M. Dawson, Rachel A. Doyle,
[b]
[b]
[a]
Peter Hrobꢁrik, Martin Kaupp, and Iain A. Smellie
1
13
Abstract: H and C pNMR properties of bis(salicylaldoxima-
to)copper(II) were studied in the solid state using magic-
angle-spinning NMR spectroscopy and, for the isolated com-
perimental techniques and first-principles computation,
a near-complete assignment of the observed signals is possi-
ble. Intermolecular effects on the pNMR shifts, modelled
computationally in the dimers and trimers through effective
decoupling between the local spins via A-tensor and total
spin rescaling in the pNMR expression, are indicated to be
small. Addition of electron-donating substituents and
benzannelation of the organic ligand is predicted computa-
tionally to induce notable changes in the NMR signal pat-
tern, which suggests that pNMR spectroscopy can be a sensi-
tive probe for the spin distribution in paramagnetic phenolic
oxime copper complexes.
plex and selected oligomers, using density-functional theory
1
at the PBE0- = //PBE0-D3 level. Large paramagnetic shifts are
3
1
13
observed, up to d( H)=272 ppm and d( C)=1006 ppm (at
98 K), which are rationalised through spin delocalisation
2
from the metal onto the organic ligand and the resulting
contact shifts arising from hyperfine coupling. The observed
shift ranges are best reproduced computationally using ex-
change-correlation functionals with a high fraction of exact
1
exchange (such as PBE0- = ). Through a combination of ex-
3
II
Introduction
tions. Cu complexes are paramagnetic, which gives rise to the
question of whether communication between the spins on dif-
ferent molecules contributes to the interaction. The metal cen-
tres in the solids are usually too far apart for direct or indirect
exchange coupling, but it is unclear to what extent spin deloc-
alisation onto the organic ligands occurs, which might in turn
modulate any intermolecular interaction. Significant spin densi-
ty on H and N atoms has been detected through EPR spectros-
[
1]
Hydrometallurgy, the extraction of metals from ores using
aqueous chemical processes, is an important alternative to
[
2]
energy-intensive techniques involving smelting. Bespoke ex-
tractants are used on a large scale to separate and concentrate
a variety of metals. In this respect, phenolic oximes are notable
for their high affinity and selectivity toward copper, as they
II
[5]
form very stable Cu complexes that can be subject to liquid–
copy in solution and in the solid state, but in the absence of
1
3
liquid extraction, separation, and purification. Their ease of
preparation and use also makes such oximes excellent targets
C-enriched samples, the precise spin distributions over the
organic backbone are unknown. We now present an experi-
[
3]
1
13
for undergraduate laboratory courses.
The complexes usually crystallise readily, and numerous
mental and computational H and C solid-state NMR study of
phenolic oxime copper(II) complexes at natural abundance in
order to fill that gap.
[4]
structures have been characterised by X-ray crystallography.
Clearly there are attractive intermolecular interactions between
the complexes that can be tuned through appropriate sub-
stituent patterns for tailoring the solvation properties, with p-
stacking between the aromatic moieties, arguably, the most
important factor modulating or even governing these interac-
NMR is an ideal technique to probe the local structure and
short-range order or disorder in solid materials. While applica-
tions to diamagnetic solids are becoming more and more rou-
tine (at least for “benign” nuclei with high sensitivity and/or
low or vanishing quadrupole moment), acquisition of NMR
spectra for paramagnetic materials is still a challenge. Advan-
ces in instrumentation (such as very fast magic-angle spinning,
MAS) are beginning to make more and more classes of para-
magnetic compounds amenable to study by solid-state NMR
[a] Prof. M. Bꢀhl, Prof. S. E. Ashbrook, Dr. D. M. Dawson, R. A. Doyle,
Dr. I. A. Smellie
University of St. Andrews
School of Chemistry and Centre of Magnetic Resonance
North Haugh, St Andrews, Fife, KY16 9ST, Scotland (UK)
E-mail: buehl@st-andrews.ac.uk
[6]
[7]
spectroscopy, including metalloproteins, smaller complexes,
[8]
infinitely-connected metal–organic frameworks, and dense
sema@st-andrews.ac.uk
[9]
oxides. However, the assignment of individual resonances to
[b] Dr. P. Hrobꢁrik, Prof. M. Kaupp
specific chemical or crystallographic sites can still be a formida-
ble task and may even require laborious syntheses for site-spe-
cific isotope labelling.
Technische Universitꢂt Berlin, Institut fꢀr Chemie
Strasse des 17. Juni 135, D-10623 Berlin (Germany)
Supporting information for this article can be found under http://
dx.doi.org/10.1002/chem.201602567.
Chem. Eur. J. 2016, 22, 1 – 13
1
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
It is highly desirable to confirm, and eventually to make,
such assignments based on reliable quantum-chemical compu-
tation of the salient NMR parameters, in particular isotropic
shifts. Such calculations have proven very useful during assign-
ment and interpretation of the multinuclear NMR spectra of di-
NMR experiments
1
The H MAS NMR spectrum of 1 (at 298 K) is shown in Fig-
ure 2a, and contains resonances at d=272, 26.0, 22.7, 5.5 and
ꢀ5.4 ppm. Variable-temperature experiments over the range
278–338 K (shown in Figure 2b) show that, as might be ex-
pected, the more unusual isotropic shifts are the most temper-
[
10,11]
amagnetic materials.
However, application of such compu-
tations to paramagnetic species is a considerable challenge.
The first quantum-chemical methods to calculate paramagnetic
NMR (pNMR) properties that went beyond the pure contact
1
ature dependent. Figure 2c shows plots of H diso against
1000/T, with corresponding lines of best fit and regression co-
[
12,13]
2
shifts were only devised in the early 2000s
and, even
efficients (R ). The linear relationship expected for the “high-
though refinements and further developments have been
temperature” paramagnetic regime is observed in all cases,
except the resonance at d=5.5 ppm, the position of which is
essentially independent of temperature (to within d=
ꢁ0.1 ppm over the entire temperature range).
[
14–17]
made since then,
pNMR calculations remain far from rou-
[
18]
tine. Extension of the present computational techniques to
[10,11]
pNMR properties of infinite periodic solids is in its infancy
1
and is probably one of the final frontiers of computational
NMR spectroscopy. As a further step toward that goal, we pres-
ent a computational pNMR study alongside experimental
measurements on phenolic oxime copper complexes, calling
special attention to intermolecular effects on the pNMR chemi-
cal shifts. Such effects are shown to be relatively small, but the
calculations prove instrumental for guiding the assignments
and for prediction and future study of substituent effects.
In addition to the unusual shifts, three of the observed H
resonances have reasonably large shift anisotropies (defined
by the span (W=d ꢀd )). By far the largest is for the reso-
11
33
nance at d=22.6 ppm, which has W ꢁ326 ppm, leading to in-
tense spinning sidebands even with the rapid MAS rate
(37.5 kHz) used in the experiment. The other resonances have
anisotropies that can be estimated in a “slow” MAS experiment
(shown in Figure S1 in the electronic Supporting Information).
The slowest MAS rate that could practically be used was
1
6 kHz, to avoid overlap of spinning sidebands. In addition, at
1
slower MAS rates, the line broadening from the H homonu-
clear dipolar interaction begins to affect the spectral resolu-
tion. Table 1 summarises the NMR parameters for the H species
observed.
Results and Discussion
The main target of this study is the bis(salicylaldoximato)cop-
per(II) complex (1, Figure 1). The first X-ray crystallographic
[
20]
characterisation was published in 1964, but a number of al-
ternate refinements or polymorphs have been reported
1
Table 1. H NMR parameters for 1 at 298 K, and the temperature depend-
[
21]
1
13
since. We will first describe the H and C NMR-spectroscop-
ic characterisation of 1, followed by a discussion of the com-
puted pNMR shifts. We will then comment on the assignments
and, finally, discuss substituent effects in selected derivatives.
1
ence of diso. The y intercepts of the plots in Figure 2c are given by diso
The magnitude of the shift anisotropy is described by the span (W).
.
iso1
diso [ppm]
ddiso/d(1/T) [ppmK]
d
[ppm]
W [ppm]
4
2
2
2
5
72(2)
6.0(1)
2.7(3)
.5(1)
8.87(8)ꢃ10
ꢀ26(3)
4.8(8)
8.4(3)
4.95(9)
4.76(5)
69(5)
89(5)
326(20)
75(5)
70(5)
3
6.34(25)ꢃ10
3
4.24(9)ꢃ10
2
1.62(28)ꢃ10
ꢀ3.03(2)ꢃ10
3
ꢀ
5.4(1)
13
The C MAS NMR spectrum of 1 at 298 K is shown in Fig-
ure 3a, and contains resonances at d=1006, 963, 244, 148,
124, 118 and 79 ppm. The resonances at d=1006 and
963 ppm are very broad and overlapping, and were originally
thought to be a single resonance centred at d=ꢁ980 ppm
until the resonance was re-examined in light of the results of
the DFT calculations (discussed below). Minor resonances (at-
tributed to an unidentified impurity) are also observed at d=
85 and 143 ppm. Variable-temperature experiments over the
range of 278–338 K (shown in Figure 3b) show that the posi-
tions of all resonances are temperature dependent, with the
more unusual isotropic shifts varying the most. Figure 3c
13
shows plots of C diso against 1000/T and their corresponding
2
1
lines of best fit and R . As for H, a linear relationship is ob-
served in all cases, although the plots for the two resonances
at higher shift have lower regression coefficients owing to
Figure 1. a) Structure of complex 1 with atom numbering; b) packing in the
solid (from ref. [21]; stereoview).
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
2
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
1
Figure 2. a) H (9.4 T, 298 K, 37.5 kHz MAS) NMR spectrum of 1 (relative to TMS), with insets showing expansions of the isotropic resonances and spinning
1
sidebands indicated with asterisks (*). b) Variable-temperature H (9.4 T, 37.5 kHz MAS) NMR spectra of 1, in which only the isotropic regions are shown (* indi-
1
cates spinning sidebands of the resonances at d=22.6 and 5.5 ppm). c) Plots of the variation in isotropic H peak position as a function of 1000/T. The lines
2
of best fit and their equations and regression coefficients (R ) are given.
1
3
Figure 3. a) C (9.4 T, 298 K, 37.5 kHz MAS) NMR spectrum of 1 (relative to TMS), with insets showing expansions of the isotropic resonances and spinning
1
3
sidebands indicated with asterisks (*). b) Variable-temperature C (9.4 T, 37.5 kHz MAS) NMR spectra of 1, in which only the isotropic regions are shown.
1
3
2
c) Plots of the variation in isotropic C peak position as a function of 1000/T. The lines of best fit and their equations and regression coefficients (R ) are
given.
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
3
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
nating (see also below). As expected, the percentage of Har-
tree–Fock (HF) exchange is, thus, crucial. At the PBE level,
much larger shift ranges are computed (ca. d=450 and
1
3
Table 2. C shifts for 1 at 298 K, and the temperature dependence of
1
d
iso. The y intercepts of the plots in Figure 3c are given by diso
.
iso1
d
iso [ppm]
ddiso/d(1/T) [ppmK]
d
[ppm]
1
13
1750 ppm for H and C, respectively) than observed (ca. d=
5
1
9
2
1
1
006 (5)
63 (5)
44(1)
48(1)
24(1)
3.36(16)ꢃ10
ꢀ115(51)
90(84)
300 and 930 ppm, respectively). This finding is consistent with
the tendency of GGA functionals to overestimate spin delocali-
sation from the metal to the ligand (an effect of the inherent
self-interaction error in DFT). Inclusion of HF exchange reduces
the extent of spin delocalisation and, hence, the computed
5
2.62(26)ꢃ10
4
3.89(4)ꢃ10
114.0(14)
137.4(4)
126.9(2)
122.3(1)
126.1(1)
3
2.97(13)ꢃ10
2
ꢀ8.6(6)ꢃ10
3
118(1)
ꢀ1.33(5)ꢃ10
3
1
79.0(5)
ꢀ1.403(3)ꢃ10
shift ranges notably (to ca. d=340 and 1200 ppm for H and
1
3
C, respectively, with PBE0, and to about d=300 and
1
1
100 ppm, respectively, with PBE0- = ). This spin delocalisation
3
their overlap at higher temperature. Table 2 summarises the
is illustrated by the computed total spin density (PBE/II level,
Figure 4b), which essentially arises from the singly occupied
13
NMR parameters for the C species observed.
DFT computation
To aid the spectral assignments, we performed DFT calculations
of the pNMR shifts of 1 by using a well-established methodolo-
[
13–15]
gy.
terised by X-ray crystallography, a monomeric unit, a dimer
“slipped sandwich”, cf. Figure 1b), in which two such units are
Starting from the structure of one polymorph charac-
[
21]
(
shown), as well as a trimer were optimised. From the agree-
ment between computed and observed distances (see Table S1
in the Supporting Information), the PBE0-D3 level of DFT was
[
22]
chosen for all further geometry optimisations.
Initial pNMR shift calculations were performed using a variety
of functionals, basis sets and other approximations. Results for
9
a monomeric 1, which is a 3d spin doublet, are given in
Table 3. Inclusion of the spin-orbit terms for the A tensor of
the lighter elements (Aorb) has little effect, up to approximately
13
d=5 ppm for C (compare PBE/II and PBE/II(orb) entries in
Table 3). As they increase the CPU time significantly (by
a factor of 4 for this system) these spin-orbit terms were ne-
glected in all subsequent calculations.
In contrast, the results are very sensitive to the exchange-
correlation functional. It is clear from the observed (Tables 1
and 2) and computed (Table 3) chemical shift ranges that the
contact term arising from spin density on the nucleus is domi-
Figure 4. a) a-HOMO and b) spin density of 1 (PBE/II//PBE0/AE1(*) level, iso-
density surfaces at 0.02 a.u. for MO and 0.0004 a.u. for density); see Fig-
ure 1a for atom numbers.
1
13
Table 3. Computed H and C pNMR chemical shifts (in ppm relative to TMS at
[
a]
298 K) of pristine 1.
HOMO (Figure 4a). This orbital has significant anti-
bonding contributions between the dx2-y2 orbital on
[
b]
1
1
Nucleus
PBE/II
PBE/II (orb)
PBE0/II
PBE0/III
PBE0- =
3
/ II
3
PBE0- = / III
Cu and s -bonding orbitals, which are delocalised
p
H7
H5
H(br)
H3
H4
H6
C6
C7
C5
C4
C2
C3
C1
426.5
57.9
57.1
14.2
10.6
ꢀ11.4
1732.5
1696.7
344.4
115.7
233.9
212.7
ꢀ15.3
426.5
57.7
55.2
14.6
10.6
ꢀ11.3
1737.9
1700.4
343.2
115.2
235.8
212.4
ꢀ19.5
338.1
38.6
34.4
12.7
7.8
ꢀ7.9
1303.6
1113.7
262.2
135.1
133.0
130.8
52.1
330.0
38.6
37.1
12.1
8.0
ꢀ8.1
1271.8
1104.3
264.6
137.6
137.9
135.3
75.8
306.6
35.0
31.7
13.4
6.5
ꢀ6.8
1166.6
931.4
235.3
144.8
132.4
108.8
79.2
297.7
35.2
34.6
12.9
6.8
ꢀ6.8
1183.2
922.9
237.6
147.1
137.8
113.5
99.3
over large parts of the ligands. The contributions
from the in-plane p-orbitals are clearly seen from the
nodal planes passing through most of the lighter
nuclei. It is the hybridisation through admixture of s-
orbitals that gives rise to large hyperfine couplings at
some of these nuclei, and, hence, to their large
pNMR shifts. Among the H atoms it is the H atom at
the imido moiety (H7 in Figure 1a) that carries the
highest spin density, in excellent agreement with
[5b]
recent low-temperature EPR results.
As a conse-
quence, this H atom is the one with the most pro-
[
a] See Figure 1 for labelling scheme. [b] Including Aorb contributions in the A tensor.
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
4
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
nounced downfield shift and can safely be assigned to the res-
onance observed at d=272 ppm (Table 1).
mers. Two test calculations for the “Cu–Zn dimer” at the PBE0-
1
=3/IGLO-II level using both choices for the gauge origin afford-
Among the C atoms, significant s-spin density is accumulat-
ed on the imido C atom and the ortho-C atom opposite (C6
and C7 in Figure 1a). Both have the highest computed shifts,
approaching or exceeding 1000 ppm and appear very close to
each other (within ca. 200 ppm at all levels of theory investi-
gated). Initially the very broad signal in this region in the ex-
perimental spectrum had been assumed to be a single reso-
nance (centred at dꢁ980 ppm). However, closer inspection in
the light of these calculated shifts revealed that the resonance
can be decomposed into two signals separated by dꢁ40 ppm,
in qualitative agreement with the computation.
ed virtually identical results (at most d=0.1 ppm difference for
13
some C nuclei, data not shown) so that, subsequently, the
centre of charge was used throughout.
In the solid the metal atoms reside on inversion centres so
that both salicylaldoximate ligands are equivalent by symme-
try. In a pristine dimer they are not, so that different shifts are
computed for nuclei that are formally equivalent in experi-
ment. In Table 4, we report the arithmetic average for such
equivalent pairs, along with the individual differences for each
of the resonances (ꢂ values in parentheses).
Compared to the impact of the amount of HF exchange in
the functional, the effect of increasing the basis set beyond
IGLO-II for the ligand atoms is less pronounced. On going from
IGLO-II to IGLO-III basis, the computed pNMR shifts change by
1
13
Table 4. H and C pNMR isotropic shifts (in ppm relative to TMS at
98 K) of 1 and its adducts with one M(C NO moiety (dimer) or sand-
/ IGLO-II level, doublet states.
2
7
H
6
2 2
)
1
wiched by two of these (trimer), PBE0- =
3
1
13
up to approximately d=8 and 30 ppm for H and C, respec-
tively (Table 3). These maximum changes are observed for the
nuclei experiencing the largest paramagnetic downfield shifts
Nucleus Pristine Dimer
(M=Zn)
Dimer
(M=Cu)
Trimer
(M=Zn)
Trimer
(M=Cu)
[
a]
[a,b]
[b]
1
H7,7’
H5,5’
H(br,br’)
H3,3’
H4,4’
H6,6’
C6,6’
C7,7’
C5,5’
C4,4’
C2,2’
C3,3’
C1,1’
306.6 303.5(ꢂ1.5) 303.7(ꢂ2.9). 302.2
301.9
34.4
32.6
14.2
6.0
(H7, C6 and C7); the results for other nuclei in the more
35.0 33.7(ꢂ0.6)
31.7 32.9(ꢂ1.6)
13.4 13.6(ꢂ0.2)
6.5 5.6(ꢂ0.2)
34.7(ꢂ0.3)
30.9(ꢂ2.16)
14.1(ꢂ0.2)
5.4(ꢂ0.4)
32.6
32.9
13.4
5.9
“
normal” (diamagnetic) chemical shift ranges are much less
basis-set dependent.
As a first step toward modelling intermolecular effects,
which may be important for the observed shifts in the solid,
we performed computations for a “slipped-sandwich” dimer of
complex 1 [in Figure 1b) two such dimers can be discerned,
one in the background at the top left, one in the foreground
ꢀ6.8 ꢀ7.1(ꢂ0.7) ꢀ7.4(ꢂ0.5)
ꢀ6.6
-6.9
1166.6 1142.1(ꢂ7.9) 1134.4(ꢂ13.7) 1127.6
931.4 901.3(ꢂ8.6) 903.5(ꢂ12.7) 894.6
1109.7
903.2
221.1
148.4
142.6
106.6
58.4
235.3 228.0(ꢂ4.6) 221.9(ꢂ1.6)
144.8 147.6(ꢂ2.1) 151.3(ꢂ0.5)
132.4 138.7(ꢂ6.2) 138.8(ꢂ8.6)
108.8 106.0(ꢂ2.2) 102.9(ꢂ1.6)
79.2 68.2(ꢂ21.7) 56.3(ꢂ10.1)
232.8
144.0
145.5
108.8
80.1
[
23]
at the bottom right]. The slippage is such that the aromatic
ring of one molecular unit is close to the axial site of the Cu
centre of the other (with Cu···C distances of approximately
[a] Mean value for the pair of nuclei indicated in the first column (in pa-
rentheses difference of each individual nucleus within the pair from this
average). [b] sorb, g, and A tensors taken from the high-spin calculation,
A-tensors and pNMR shift expression rescaled to that of a doublet (see
the text for details).
3.6 ꢄ indicating van der Waals contacts). Starting from the X-
ray coordinates, this dimer structure was essentially maintained
during optimisation (at the PBE0-D3 level, see Table S1 in the
Supporting Information). Finally, a third molecule was added
to form a trimer in C symmetry (Figure S2 in the Supporting
i
Information). As slight structural differences from the periodic
structure began to appear for this trimer (e.g., a slightly too
short Cu···Cu distance of 4.64 ꢄ as compared to 4.91 ꢄ in the
solid), the trimer was re-optimised with the Cu···Cu distance
For the “Cu–Zn dimer”, these individual differences are small
1
13
to modest (up to ca. d=1.5 and 22 ppm for H and C, respec-
tively), and the changes of the mean values from those in the
pristine monomer are small for most nuclei (a few ppm),
[
24]
13
fixed to the value in the solid.
except for the most deshielded C resonances around d=
The dimer and trimer were optimised by assuming ferro-
magnetic coupling between the Cu centres (triplet and quartet
states, respectively). The precise nature of this coupling in the
1000 ppm, in which changes of up to d=30 ppm are comput-
ed (compare entries for “pristine 1” and “dimer (M=Zn)” in
Table 4). Similarly, small to modest changes are obtained for
the trimer, the central copper complex is sandwiched by two
Zn moieties (again simply replacing the corresponding Cu
atoms in the optimised structure, see “trimer (M=Zn)” entry in
Table 4).
[
25]
full solid is unknown (see below). Based on the results for
monomeric 1 discussed above, we will discuss primarily PBE0-
1
=
3/IGLO-II pNMR results, although selected PBE data are pro-
vided in the Supporting Information.
To probe for the spatial proximity of a second molecule
without possible complications from the coupling between
spin centres, we performed initial pNMR calculations for
a dimer in which one of the Cu centres was replaced with Zn
Effects from dia- or paramagnetic ring currents circulating
around the metal centre or the aromatic rings of one mono-
mer on the nuclei of others nearby can be probed by evaluat-
ing the orbital contributions to the shielding constant (siso(orb)
)
(without reoptimisation), affording a simple doublet. By de-
at those points in space around pristine 1 in which the nuclei
fault, the gauge origin for the g-tensor evaluation is taken as
the centre of electronic charge, which is identical with the Cu
position in monomeric 1, but at the centre of inversion in the
dimer, that is, off-centre with respect to each of the mono-
of the second molecule would be placed. Such anisotropy ef-
1
fects can be significant, for example, for H chemical shifts of
hydride ligands directly bonded to heavier d- or f-block ele-
[26]
ments, but are small in this case as the molecules are rather
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
5
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
far apart (Cu···Cu distance in the solid: 4.9 ꢄ). sorb values at ap-
propriate points in the space around 1 (at the “virtual” C and
H positions of the nearest neighbour) are all less than d=
latter encoding the spin distribution), can adjust to the pres-
ence of the neighbouring monomers, while the rescaling of
the total spin and the A tensors implies a spin-decoupled cal-
culation of the pNMR shifts. In the limit of weakly coupled
monomers, as in our case (see below for a justification), we be-
lieve this is a reasonable approximation.
[
27]
1
ppm. By and large, the spin distribution within a molecular
moiety appears to be perturbed little by the presence of
neighbouring closed-shell, diamagnetic complexes.
In contrast, much larger effects are predicted for the dimer
with two ferromagnetically-coupled Cu centres. In particular,
the more deshielded pNMR shifts are increased greatly by the
presence of the second spin on the nearest neighbour: by ap-
proximately d=100 and 300 ppm for H7 and C6, respectively
The resulting data for the dimer and trimer are included in
Table 4 (M=Cu entries). Compared to the zincated doublet
models (M=Zn entries), the presence of the open-shell neigh-
bours brings about some minor modifications of the pNMR
shifts, but typically less than observed on going from the pris-
1
(compare entries for 1 and M=Cu in Tables S2 and S3, Sup-
tine monomer to the zincated models. While the computed H
porting Information). Adding a third molecule of 1 to form
a centrosymmetric trimer in its high-spin state brings about
similarly large shifts (compare entries for M=Cu and trimer in
Tables S2 and S3, Supporting Information). Compared to the
pristine monomer, such ferromagnetically-coupled oligomers,
therefore, show strongly enhanced pNMR shifts in poor agree-
ment with experiment (Tables 1 and 2).
shifts change only little with the environment, a somewhat
13
larger sensitivity is apparent for the C shifts. On going from
the monomer to the trimer (M=Cu), the largest changes (be-
tween ca. d=20 and 60 ppm) are obtained for C1, C6, C2 and
C7, which have the highest spin density (Figure 4b) and are
closest to the neighbouring Cu centre. Importantly, no qualita-
tive changes are brought about by these cluster models for
the solid, that is, the relative sequence of the shifts is main-
tained.
Zero-field splitting (ZFS) in these oligomers with total spin
1
larger than = is expected to be small because of the expected
2
weak exchange coupling between the Cu centres. For strongly
coupled spins located on the same metal centre, ZFS can be
included in the pNMR expressions by modifications arising
The pNMR shifts for the trimer in a doublet state were also
computed, converging a suitable broken-symmetry (BS) wave-
function, in which the spin on one of the terminal
Cu(C H NO ) moieties has been flipped (denoted “BS fl››” in
[
14]
from the D-tensor. It is debatable if this methodology devel-
oped for high-spin metal centres can be applied to more dis-
tant ferromagnetically coupled spins. Tables S2 and S3, Sup-
porting Information, show that if it is applied, only very small
effects on the final pNMR shifts are calculated, with maximum
7
6
2 2
[31]
1
Table S3, Supporting Information).
At the PBE0- = /IGLO-II
3
level, this BS state is slightly more stable than the high-spin
quartet, but the preference for this antiferromagnetic coupling
is very small, with a predicted Heisenberg exchange coupling
1
13
ꢀ1 [32]
changes of approximately d=3–4 ppm for d( H) or d( C)
compare “M=Cu no ZFS” and “M=Cu with ZFS” entries in Ta-
constant J of around ꢀ0.3 cm . This result fully justifies the
(
“spin-decoupled” approach discussed above. The same result
is found for a BS open-shell singlet state for the dimer. There
are no paramagnetic NMR shifts for such a BS singlet, but its
[
28]
bles S2 and S3, Supporting Information). Similarly small ef-
fects from the D-tensor contributions are obtained for the
high-spin trimer (results not shown) and are, therefore, ne-
glected henceforth.
orbital contributions to the isotropic shielding constant, siso(orb)
,
are virtually identical to those of the high-spin triplet state (to
within fractions of a ppm in all cases). Therefore, in all pNMR
calculations for the trimer, the orbital contributions were taken
from the high-spin quartet state. The pNMR shifts of the cen-
tral moiety in the BSfl›› trimer are of the same order of mag-
nitude as those of the monomer and the other oligomer
models (Table 4). However, because the direct use of a BS-
wavefunction with its huge spin contamination in property cal-
culations without spin-projection procedures is ill-justified we
will not discuss these results further.
From the results for the high-spin cluster models it is appar-
ent that the current theory of pNMR shifts cannot be applied
to an aggregate of weakly spin-coupled subunits when this ag-
gregate is to be treated as a high-spin supermolecule. The
reason is quite simple: the expression for the A tensor ele-
ꢀ
1 [29]
ˆ
ments contains a factor hS i , that for the pNMR contribu-
z
[12,14]
tions to the magnetic shielding contains a factor S(S+1).
Thus, the pNMR shifts diverge as more and more monomers
are added with concomitant increase of the total spin. This is
a consequence of the fact that the solid does not correspond
to a ferromagnetically coupled state but likely exhibits a statisti-
cal distribution of local spins. As a pragmatic workaround, we
calculated the pNMR shifts of these oligomers by treating
them as simple doublets through manually setting the total
spin to 0.5 in the pNMR expression and by rescaling the A ten-
sors from the high-spin calculation by multiplying all tensor el-
For the nuclei with the most pronounced paramagnetic
shifts, H7, C1 and C6, the shielding constants are broken down
into the constituent contributions according to reference [14]
(Table 5). As expected, it is the contact shift (g ·A ) that is the
e
FC
dominant contribution. The data in Table 5 illustrate how, in
particular for C6, this contribution is modulated through the
small changes of the hyperfine coupling constant (A ) upon
iso
ˆ
ˆ
ements by 2hS i (in which S is the expectation value in the
interaction with the neighbouring molecules. Deviations of the
isotropic g-value from the corresponding free-electron value
z
z
high-spin case, that is, 1 and 1.5 for the triplet and quartet, re-
[
30]
spectively). In this way it is possible to effectively spin-de-
couple one monomer from the spin of its neighbours, forming
an isolated doublet. The wavefunction and its associated pri-
mary response properties, namely sorb, g, and A tensors (the
modify the contact shifts further (Dg ·A contribution).
iso FC
For comparison with the observed tensor properties and
temperature dependencies (Tables 1 and 2) of the experimen-
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
6
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
31.7 ppm, and none calculated to be at d=26.0 or 22.7 ppm.
Table 5. EPR parameters and resulting contributions to the pNMR shield-
Therefore, the calculations were used initially as a guide to
help assign the spectra, but with experimental confirmation
sought where practically possible. As preparing isotopically en-
riched material is too expensive and time consuming (as well
as requiring the development of new synthetic procedures),
ing
M(C
tensor
NO ·Cu(C
for
NO
selected
·M(C NO
nuclei
in
1
and
its
1
[
7
H
6
2
)
2
7
H
6
2
)
2
7
H
6
2
)
2
] trimer models, PBE0- =
3
/ IGLO-II
level (g is dimensionless, A is in MHz and pNMR contributions are in
ppm).
[
a]
Nucleus Property
Monomer Trimer (M=Zn) Trimer (M=Cu)
2
D NMR experiments were carried out to aid assignment. How-
g
iso
2.089635
2.094185
2.092923
ever, owing to the very fast nuclear relaxation in paramagnetic
systems, the number and sophistication of the experiments
available (and the range of information they can provide) is
limited.
H7
C6
C1
s
iso(orb)
iso
23.7
24.1
24.1
A
10.86
ꢀ286.8
ꢀ12.5
0.8
10.66
ꢀ282.4
ꢀ13.0
0.8
10.65
ꢀ282.3
ꢀ12.8
0.8
e
g ·AFC
In this case, the most useful tool for experimental assign-
Dgiso·AFC
[
[
[
b]
b]
b]
1
13
Dganiso·Adip
ment proved to be H– C heteronuclear correlation (HETCOR)
1
spectra, using cross polarisation (CP) from H as the means of
s
iso(orb)
iso
62.7
9.49
ꢀ998.2
ꢀ43.5
0.0
63.8
9.11
ꢀ959.8
ꢀ44.0
1.5
61.6
8.91
ꢀ939.3
ꢀ42.5
ꢀ0.5
magnetisation transfer. Although this experiment transfers
magnetisation by the through-space dipolar interaction, by
using a short (100 ms) CP “contact time”, the transfer can be
limited to short (essentially one-bond) distances, allowing only
A
g
e
·AFC
Dgiso·AFC
Dganiso·Adip
1
13
the observation of directly bonded CꢀH pairs. The H– C
HETCOR spectrum of 1 is shown in Figure 5 (note that the
spectrum had to be recorded with stepwise acquisition in both
dimensions to enable efficient spin locking of all resonances,
and the figure presents a composite of four separate experi-
ments). The cross-peak at d=(963, 272) ppm is a reasonable
match for C7–H7, predicted to be at d=(931.4, 306.6) ppm;
the resonance at d=(1021, ꢀ5.4) ppm is a reasonable match
for C6–H6, predicted to be at d=(1166.6, ꢀ6.8) ppm, and the
resonance at d=(244, 26.0) ppm is a reasonable match for C5–
H5 predicted to be at d=(235.3, 35.0) ppm. The only other in-
tense resonances observed are at d=(124, 5.5) and (118,
s
iso(orb)
iso
20.5
ꢀ0.79
81.1
3.5
20.7
ꢀ0.76
80.1
3.7
19.6
ꢀ0.98
103.5
4.7
A
e
g ·AFC
Dgiso·AFC
Dganiso·Adip
4.8
4.5
2.9
[
a] A-tensor and pNMR contributions scaled to S=0.5 (see text). [b] One
third of the trace.
tal measurements the corresponding quantities have been
computed for monomeric 1 and are given in Table 6.
5.5) ppm, and must correspond to C3–H3 and C4–H4, which
are predicted to be at d=(108.8, 13.4) and (144.8, 6.5) ppm, re-
spectively, despite the poor agreement between experiment
and calculation for H3 and C4. However, it was shown in Fig-
Spectral assignment
Assignment of the spectra would be impossible (or at least
very challenging and time consuming) without input from the
DFT calculations. However, several discrepancies can immedi-
ately be seen between the calculated (Table 6) and experimen-
1
ure 2c that the position of the H resonance at d=5.5 ppm is
essentially independent of temperature, indicating that the
DFT calculations may have placed a small amount of spurious
spin density on H3 and H4 when, in reality, these species
appear to have essentially no hyperfine coupling. Indeed, as
1
tal parameters (Tables 1 and 2), particularly for H, in which
there are no resonances observed at d=13.4, 35.0 or
13
1
seen in Figure 3c, the C resonances correlated to the H reso-
nance at d=5.5 ppm also display the smallest temperature de-
13
1
13
1
pendences observed for the C spectrum, indicating smaller
hyperfine couplings than calculated. The only other resonance
observed in the HETCOR spectrum is at d=(79, 5.5) ppm, sug-
gesting a quaternary C species spatially close to either H3 or
H4. Figure S3, Supporting Information, shows all of the
through-space H–C contacts under 3.5 ꢄ, and Table S5, Sup-
porting Information, lists all CꢀH contacts under 3.5 ꢄ (includ-
ing intermolecular and intramolecular distances). The two qua-
ternary C species are C1 and C2, and it can be seen that the
shortest C1-H3/H4 contact is 3.40 ꢄ to H3, whereas C2 is
2.17 ꢄ from H3 and also 3.43 ꢄ from H4. This would seem to
suggest that the cross-peak from C2–H3 should be the most
intense of these three possible correlations when the contact
time is short, thus assigning the resonances at d=79 and
Table 6. Calculated H and C NMR parameters for 1 (PBE0- =
3
/ IGLO-II
level).
[
a]
iso1
[a]
Site
d
iso [ppm]
ddiso/d(1/T) [ppmK]
d
[ppm]
W [ppm]
4
H7
H5
H(br)
H3
H4
H6
C6
C7
C5
C4
C2
C3
C1
306.6
35.0
8.91ꢃ10
8.21ꢃ10
6.30ꢃ10
1.84ꢃ10
7.76
7.43
10.6
102
24
349
19
40
95
283
922
394
238
295
255
748
3
3
31.7
13.4
6.5
3
7.22
6.85
7.03
2
ꢀ1.01ꢃ10
3
ꢀ6.8
1166.6
931.4
235.3
144.8
132.4
108.8
79.2
ꢀ4.13ꢃ10
5
3.11ꢃ10
114.0
149.6
139.4
120.6
119.1
140.1
169.6
5
2.31ꢃ10
4
2.83ꢃ10
3
7.14ꢃ10
3
3.89ꢃ10
3
ꢀ9.25ꢃ10
4
ꢀ2.67ꢃ10
148 ppm to C2 and C1, respectively. However, this contradicts
[a] Calculated for 298 K.
the calculations, which predict exactly the opposite sequence
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
7
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
1
13
Figure 5. H– C (9.4 T, 298 K, 37.5 kHz CP MAS) HETCOR spectrum of 1 with a contact time of 100 ms (the spectrum was recorded in four sections to allow ef-
1
13
1
13
ficient spin locking at all H and C offset frequencies). The H and C MAS NMR spectra (recorded at 298 K) are shown along the respective axes. Dotted
grey lines indicate isotropic peak positions. No resonances were observed in the spectrum shown on the lower-right panel. The contour levels in each panel
are arbitrary and not to the same scale.
(final estimates ca. d=148 and 75 ppm for C2 and C1, respec-
tively, see below).
Table 7. Comparison of experimental and calculated isotropic shifts (in
ppm) at 298 K. For estimated errors in the experimental values, see
Tables 1 and 2.
A second HETCOR spectrum was recorded with a longer
contact time (750 ms) and is shown in Figure S4, Supporting In-
formation (note that only the region shown in the top right
panel of Figure 5 was recorded with the longer contact time).
This spectrum shows additional resonances corresponding to
C5-H3/4 (most likely H4, owing to its closer proximity to C5),
C5–H6 (the same distance as C5–H4), C1–H6, and C2–H6. How-
ever, most importantly, there is no correlation observed at d=
Species
Exptl. diso
DFT diso
(final estimate)
Difference DFT
[
a]
d
isoꢀexptl diso
H3
H4
H5
H6
H7
Hbr
C1
C2
C3/C4
C5
5.5
5.5
13.7
6.3
34.6
+8.2
+0.8
+8.6
ꢀ1.5
26.0
ꢀ5.4
272
22.7
79
148
124, 118
244
ꢀ6.9
293.0
35.5
+21
(148, 5.5) ppm, suggesting that the corresponding C species is
+12.8
ꢀ4
spatially distant from H3 and H4, and should, therefore, indeed
be C1, rather than C2. On the other hand, there is also no
cross-peak observed at either d=(118, 26.0) or (124,
74.9
148.0
111.3, 150.7
223.4
1126.3
894.7
ꢂ0
ꢀ13 to +33
ꢀ21
26.0) ppm, (expected for C4–H5,due to a very short intramolec-
C6
C7
1006
963
+120
ꢀ68
ular CꢀH distance of 2.09 ꢄ, Table S5, Supporting Information),
and the peaks at d=(148, ꢀ5.4) and (79, ꢀ5.4) ppm involving
H6 are of similar intensity, even though one must be a coupling
between nuclei separated by two bonds, and the other
through three. The absence of a cross-peak may, thus, not
always be unequivocal for the assignments, and the purely ex-
1
[
a] Obtained by adding the computed packing effects at the PBE0- =
3
/II
1
3
level to the PBE0- = /III data (see text).
1
=3/III values in Table 3. At this stage, the only resonances that
cannot be conclusively assigned by experiment or calculation
are C3 and C4 at d=118 and 124 ppm. It can be seen that, in
general, the difference between the DFT and experimental
shifts is, at most, on the order of 10% of the shift range con-
13
perimental assignment of the C resonances for C1 and C2 re-
mains tentative. However, in view of the resulting excellent
agreement between theory and experiment, it is preferable to
13
assign the C resonances at d=79 and 148 ppm to C2 and C1,
respectively. This is not only in agreement with the isotropic
shifts, but also with their temperature dependence (compare
1
sidered. It should be noted that, while the H shift anisotropy
values were not necessary for the final assignment, the experi-
mental and calculated values quoted in Tables 1 and 3 are gen-
erally in very good agreement (particularly given the large
errors on the experimental values and the neglect of longer-
range crystal packing effects in the calculations).
[
33]
the dd /d(1/T) entries in Tables 2 and 6).
iso
Table 7 gives the final assignment of the resonances, com-
bining information from the solid-state NMR experiments and
DFT results. The final estimates were obtained adding the com-
puted packing effect upon trimer formation (the difference be-
tween the first and last data entries in Table 4) to the PBE0-
Despite its spatial proximity to the Cu centre, the bridging H
atom does not show the largest isotropic paramagnetic
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
8
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
[
34]
shift; it does show, however, the largest anisotropy by far
Tables 1 and 6), indicating that the shielding and deshielding
ing to a very similar shift range in all cases. It can be seen from
the predicted shifts for 3 that the presence of the unpaired
spin is not expected to significantly affect the second ring of
the naphthalene moiety, as perhaps expected given that the
hyperfine coupling relies on the s character of the s bonds,
rather than being transferred through the p orbitals.
(
effects in the principal components cancel to a large extent in
the isotropic shift. The ability to calculate accurately the shift
anisotropy may prove useful in assigning the NMR spectra of
other paramagnetic solids, for which resonances may have
similar isotropic shifts but different anisotropies (e.g., H5 and
H(br) in 1 have an isotropic shift difference of just d=3 ppm,
but a difference in W of over d=200 ppm).
1
13
In comparison to the H resonances, some of the C isotrop-
ic shifts are much more sensitive to the substituents. Most no-
tably, the d values for C1 and C5, the substituent of which
remain unchanged, show the largest variations across the
series, up to dꢁ100 ppm. In some cases, a complete change
of the sequence of the signals can ensue (compare, e.g., data
for 1 and 4 in Table 8). Among the most deshielded resonances
computed for C6 and C7, the shift of C6 is the most sensitive,
varying by up to d=150 ppm (compare values for 3 and 4 in
Table 8). Qualitatively, however, this region of the spectrum
should remain much the same across the whole series (al-
though the two resonances may overlap, depending on line-
width). For 2, the calculation that the methyl carbon atom
should be strongly deshielded (d=458.9 ppm) is confirmation
that the spin density is very high on the imido moiety.
Substituent effects
A variety of derivatives of complex 1 can be synthesised and
those shown in Scheme 1 were chosen for initial theoretical in-
vestigation of the effects of the substituents on the NMR and
EPR parameters. The substituent effects were probed computa-
1
tionally for the pristine monomeric complexes (PBE0- = /IGLO-II
3
level) and the results are summarised in Table 8.
Conclusions
1
In summary, we have recorded and analysed the solid-state H
13
and C NMR spectra of the bis(salicylaldoximato)copper (II)
complex 1. First-principles computation of the salient NMR pa-
rameters, mainly isotropic shifts and their temperature de-
pendence, were instrumental for the assignment of the signals.
Near-complete assignment was possible with additional two-
dimensional heteronuclear correlation spectra.
Scheme 1. Ligand substitution pattern studied in this work (only one of two
equivalent ligand shown).
Computationally, we have studied intermolecular effects on
the pNMR shifts in small cluster models as a first step toward
modelling a molecular crystal consisting of weakly interacting
paramagnetic molecules. This is achieved through effective de-
coupling between the local spins, justified by the essentially
vanishing exchange coupling between them. Judging from
these results, such intermolecular effects on the pNMR shifts
are relatively small. For the Cu complexes in this study, these
effects are dwarfed by the intricacies of the electronic structure
A more detailed NMR spectroscopic investigation of these
[
35]
derivatives will be published elsewhere, but, herein, we offer
the following general assessments of the computed values.
1
The isotropic H shifts are not very sensitive to the substituent
patterns explored here: the strongly deshielded H7 remains
close to d=300 ppm and H6 is always slightly shielded, lead-
II
1
Table 8. Substituent effects on monomeric salicylaldoxime–Cu complexes, PBE0- =
3
/ IGLO-II level (labelling as in Scheme 1).
[
a]
complex
H3
H4
H5
H6
H7
H(br)
H(R)
H9
H10
7.0
H11
9.4
1
2
3
4
5
13.4
15.1
–
12.3
13.2
6.5
6.7
–
5.0
–
35.0
35.2
43.0
40.2
35.0
ꢀ6.8
ꢀ3.7
ꢀ8.2
–
306.6
–
303.7
308.6
306.5
31.7
28.2
33.5
32.0
31.5
–
ꢀ29.0
8.0
4.0
10.5
ꢀ7.7
1.1
[
a]
C1
79.2
84.6
138.2
175.9
70.5
C2
C3
C4
C5
C6
C7
C(R)
–
C9
C10
C11
1
2
3
4
5
132.4
120.2
122.6
131.2
130.0
108.8
77.4
86.1
99.9
102.4
144.8
136.7
156.1
149.3
168.0
235.3
270.4
266.8
163.8
229.6
1167
1146
1209
1059
1174
931.4
951.4
932.4
919.2
946.9
458.9
142.5
47.4
120.0
142.5
127.5
[
b]
35.5
[a] H8,C8; mean values for H shifts of Me groups. [b] Quaternary C atom; mean value for Me groups: 33.0.
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
9
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
of the monomeric units, which are reflected, for example, in
the large dependencies of the computed pNMR shifts on the
exchange-correlation functional. Commensurate with previous
(C1), 153.2 (C7), 131.5 (C5), 130.9 (C3), 119.9 (C4), 116.9 (C6) and
116.4 ppm (C2).
[
36]
Bis(salicylaldoximato)copper(II): Salicylaldoxime (0.14 g, 1 mmol)
experience, a high fraction of Hartree–Fock exchange (33%
in the present case) is beneficial for this purpose. The compu-
tational investigation was then extended to predict the NMR
parameters of a series of complexes with substituted oximate
ligands.
3
in absolute ethanol (12 cm ) was added to a hot solution of 0.02m
3
aqueous copper(II) sulfate (25 cm ). The mixture was stirred for
1
5 min and filtered under suction to afford a pale-brown precipi-
tate, which was washed with deionised water (80 mg, 47% yield).
Analysis of the computed shielding tensors confirms the ex-
pectation that the isotropic pNMR shifts are governed by the
contact shifts, essentially probing the isotropic (Fermi-contact)
part of the hyperfine coupling at the nucleus in question. For
NMR spectroscopy
Solution-state NMR spectra were recorded using a Bruker Avance II
1
spectrometer, equipped with a 9.4 T superconducting magnet ( H
13
13
and C Larmor frequencies of 400.13 and 100.66 MHz, respective-
C nuclei in particular, pNMR spectroscopy can thus be an im-
ly). Chemical shifts are quoted relative to (CH ) Si, using the residu-
portant complement to EPR spectroscopy, where such cou-
plings can be difficult to assign at natural abundance. In con-
junction with theory, the pNMR spectra afford exquisitely de-
tailed insights into the spin distribution within these metal
complexes and into the way this spin distribution can be
modulated through substituents attached to the organic li-
gands. More work will be needed to establish relationships be-
tween these spin distributions, molecular and crystal struc-
tures, and properties of potential interest for technological ap-
plications such as metal–ligand binding affinities or solubilities
of the complexes.
3 4
al CHCl and CDCl solvent peaks (d=7.26 and 77.16 ppm, respec-
tively).
3
3
Solid-state NMR spectra were recorded using a Bruker Avance III
spectrometer, equipped with a 9.4 T wide-bore superconducting
1
13
magnet ( H and C Larmor frequencies of 400.13 and 100.66 MHz,
respectively). Experiments were carried out using a 1.9 mm MAS
1
13
probe, with MAS rates between 16 and 37.5 kHz. H and C NMR
chemical shifts are quoted relative to (CH ) Si, using the NH and
3
4
3
CH resonances of l-alanine (d=8.5 and 20.5 ppm, respectively) as
3
secondary references. MAS spectra were recorded using a rotor-
synchronised spin-echo pulse sequence with an echo delay of one
1
rotor period. Signal averaging was carried out for 512 ( H) or
Neither pNMR measurements nor computational modelling
of solid materials are routine tasks at present. Challenges for
further work abound, namely to find the appropriate condi-
tions to record the spectra (e.g., spinning speed, temperature
control) and to find the appropriate quantum-chemical tools
to assign and interpret them. On the theoretical side, the chal-
lenges comprise the accurate modelling of effects from the
bulk solid (most efficiently through inclusion of periodic boun-
dary conditions), and probably thermal and zero-point correc-
tions; all compounded by the huge sensitivity of the results to
the theoretical level (i.e., the exchange-correlation functional if
DFT is used). Few of these challenges appear insurmountable
already, so that further work in this area may be rewarding.
13
8
1920–448512 transients ( C) with a recycle interval of 100 ms in
all cases. The magnitude of the chemical shift anisotropy is defined
by the span (W=d ꢀd ), in which the three principal components
11
33
of the diagonalised shift tensor are ordered with d ꢃd ꢃd and
11
22
33
d
=(d +d +d )/3. The HETCOR spectra were recorded using
11 22 33
iso
1
1
CP from H with a contact pulse (ramped for H) of 100 or 750 ms,
and are the result of averaging between 440 and 32768 transients
for each of between 14 and 80 t increments of 12.5 to 26.67 ms,
1
with a recycle interval of 100 ms. The sample temperature was
controlled using a Bruker BCU-II chiller and Bruker BVT/BVTB-3000
temperature controller and heater booster. The sample tempera-
ture (including frictional heating effects arising from sample spin-
87
[37]
ning) was calibrated using the isotropic Rb shift of solid RbCl.
Computational details
[
38]
Experimental Section
Structures were optimised with the Gaussian 09 program at RI-
[39]
[40,41]
[42,43]
BP86,
PBE0,
and PBE0-D3
levels of density functional
theory, employing the AE1(*) basis, that is, a Wachters basis aug-
Synthesis
[44]
mented with two diffuse p and one diffuse d sets for Cu (8s7p4d,
Salicylaldoxime: Salicylaldehyde (1.83 g 15 mmol) in 80% aqueous
ethanol (40 cm ) was added to a solution of hydroxylamine hydro-
[
45]
3
full contraction scheme 62111111/3311111/3111), 6-31G** for the
H(br) atoms and 6-31G* for all other atoms. The s , g, and A ten-
chloride (2.08 g, 30 mmol) and sodium acetate (3.04 g, 37 mmol) in
orb
[39]
1
[46]
3
sors were computed at the PBE,
PBE0, and PBE0- =
levels
deionised water (10 cm ) and the solution was stirred at room tem-
3
using 9s7p4d basis sets on 3d metals that were constructed specif-
ically for hyperfine coupling constant calculations (full contraction
perature for 3 h. The solvent was removed in vacuo and deionised
3
water (20 cm ) was added to the crude residue. The product was
[47]
3
scheme 621111111/3311111/3111),
and the IGLO-II or IGLO-III
extracted using ethyl acetate (3ꢃ20 cm ) and the three organic
[48]
3
basis on the ligands (sometimes just denoted II and III, respec-
tively). The sorb calculations employed gauge-including atomic or-
bitals and fine integration grids as implemented in Gaussian 09; g,
fractions were combined and washed with brine (3ꢃ15 cm ). The
organic layers were dried over anhydrous sodium sulfate and the
solvent was removed in vacuo. The crude product was recrystal-
lized from boiling 60–80 petroleum ether (20 mL) to afford white
[49]
and A tensors were computed with the ORCA program (tight
SCF convergence and fine integration grid, Grid5 option). Unless
otherwise noted, all NMR and EPR properties were computed
using the same functional/basis-set combinations.
crystals of salicylaldoxime (1.06 g, 7.73 mmol (51.5% yield). M.p.
1
(
56.2–58.38C); H NMR (400.13 MHz, CDCl , Me Si): d=9.79 (1H, s,
3
4
C-OH), 8.24 (1H, t, J=0.4 Hz, H7), 7.30 (1H, s, N-OH), 7.29 (1H, ddd,
J=8.3, 7.3, 1.7 Hz, H5), 7.19 (1H, ddt, J=7.7, 1.7, 0.4 Hz, H3), 6.99
Magnetic shielding tensors s were computed using the formalism
according to reference [14]. Herein we only consider the isotropic
average:
(
1
1H, ddt, J=8.3, 1.2, 0.4 Hz, H6), 6.93 ppm (1H, ddd, J=7.5, 7.4,
1
3
1
.2 Hz, H4); C { H} NMR: (100.66 MHz, CDCl , Me Si): d=157.3
3
4
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
10
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
siso ¼ sisoðorbÞꢀSðS þ 1Þb =ð3kTg b Þ½g ꢄ A þ
McKinlay, I. A. Smellie, R. Cadou, N. S. Keddie, R. E. Morris, S. E. Ash-
brook, Phys. Chem. Chem. Phys. 2013, 15, 919–929.
9] J. Kim, D. S. Middlemiss, N. A. Chernova, B. Y. X. Zhu, C. Masquelier, C. P.
Grey, J. Am. Chem. Soc. 2010, 132, 16825–16840.
e
N
N
e
FC
1
ge ꢄ APC þ Dgiso ꢄ AFC þ = TrðDg
ꢄ AdipÞꢅ
3
aniso
[
in which siso(orb) is the isotropic orbital shielding, S is the effective
spin, b and b are Bohr magneton and nuclear magneton, respec-
[10] See, for example: Calculation of NMR and EPR Parameters: Theory and
Applications (Eds.: M. Kaupp, M. Bꢀhl, V. G. Malkin), Wiley-VCH, Wein-
heim, 2004.
e
N
tively, T is the temperature (set to 298.15 K), g and g are the free-
e
N
[
11] For some overviews in the context of solid-state NMR, see for example:
a) C. Bonhomme, C. Gervais, F. Babonneau, C. Coelho, F. Pourpoint, T.
Aza ¨ı s, S. E. Ashbrook, J. M. Griffin, J. R. Yates, F. Mauri, C. J. Pickard,
Chem. Rev. 2012, 112, 5733–5779; b) S. E. Ashbrook, D. McKay, Chem.
Commun. 2016, 52, 7186–7204.
electron and nuclear g-values, respectively, AFC and Adip are the
usual isotropic Fermi-contact and anisotropic traceless spin-dipolar
contribution to the A-tensor, respectively, APC the isotropic pseudo-
contact (PC) term arising from spin-orbit corrections to the A-
tensor, and Dgiso and Dganiso are the isotropic and anisotropic parts
of the g-tensor, respectively (in the usual representation of the g-
tensor in the form g=g +Dg ·1+Dg ). Chemical shifts d are re-
[12] Z. Rinkevicius, J. Vaara, L. Telyatnyk, O. Vahtras, J. Chem. Phys. 2003, 118,
2550–2561.
[13] S. Moon, S. Patchkovskii, in Calculation of NMR and EPR Parameters:
Theory and Applications (Eds.: M. Kaupp, M. Bꢀhl, V. G. Malkin), Wiley-
VCH, Weinheim 2004, pp. 325–340.
e
iso
iso
ported relative to TMS according to:
[
14] P. Hrobꢁrik, R. Reviakine, A. V. Arbuznikov, O. L. Malkina, V. G. Malkin, F.
Koehler, M. Kaupp, J. Chem. Phys. 2007, 126, 024107.
d ¼ sisoðorbÞðTMSÞꢀsisoðorbÞ
[
15] a) T. O. Pennanen, J. Vaara, J. Chem. Phys. 2005, 123, 174102; b) T. O.
Pennanen, J. Vaara, Phys. Rev. Lett. 2008, 100, 133002; c) S. A. Rouf, J.
in which the isotropic orbital shieldings of TMS have been comput-
ed at the same level (see Table S4, Supporting Information, for indi-
vidual values).
ˇ
Mares, J. Vaara, J. Chem. Theory Comput. 2015, 11, 1683–1691.
[16] B. Martin, J. Autschbach, J. Chem. Phys. 2015, 142, 054108.
[17] a) W. Van den Heuvel, A. Soncini, Phys. Rev. Lett. 2012, 109, 073001;
b) W. Van den Heuvel, A. Soncini, J. Chem. Phys. 2013, 138, 054113.
[
[
[
18] Although some such calculations have been proposed for undergradu-
ate computational chemistry “experiments”: B. P. Pritchard, S. Simpson,
E. Zurek, J. Autschbach, J. Chem. Educ. 2014, 91, 1058–1063.
19] For example: a) D. Carlier, M. Menetrier, C. P. Grey, C. Delmas, G. Ceder,
Phys. Rev. B 2003, 67, 174103; b) D. S. Middlemiss, A. J. Ilott, R. J. Clꢅm-
ent, F. C. Strobridge, C. P. Grey, Chem. Mater. 2013, 25, 1723–1734.
20] CSD refcode SALCOP: M. A. Jarski, E. C. Lingafelter, Acta. Crystallogr.
Acknowledgements
We thank A. Mondal for helpful discussions. This work was sup-
ported by EaStCHEM and the School of Chemistry. Computa-
tions were performed on local computer clusters maintained
by Dr. H. Frꢀchtl. We would like to thank the EPSRC for compu-
tational support through the Collaborative Computational Proj-
ect on NMR Crystallography (CCP-NC), via EP/M022501/1 and
EP/J501542/1. SEA would like to thank the Royal Society and
Wolfson Foundation for a merit award. Work at the TU Berlin
has been carried out within the EU Marie-Curie Initial Training
Network “pNMR”. For access to research data and metadata
1
964, 19, 1109–1112.
[21] For example, CSD refcode SALCOP03: K. N. Lazarou, A. K. Boudalis, V.
Psycharis, C. P. Raptopoulou, Inorg. Chim. Acta 2011, 370, 50–58.
[
22] The PBE0 level performs very well for metal–ligand bond lengths in
transition-metal complexes (M. Bꢀhl, C. Reimann, D. A. Pantazis, T.
Bredow, F. Neese, J. Chem. Theory Comput. 2008, 4, 1449–1459), and
the dispersion correction is instrumental for describing intermolecular
interactions, as in the oligomers that have been considered here.
23] The shortest distance between the perpendicularly oriented stacks is
between H6 in one stack and the phenolic O atom in the other (C·O
distance 3.40 ꢄ), implying a weak CH···O hydrogen bond.
[
see
DOI:
http://dx.doi.org/10.17630/4e9b2566-aa84-4494-
b95b-f1a9ee2840bd.
[24] This procedure was adopted (rather than extracting sections from the
experimental crystal structure directly), because optimised structures
tend to perform better in NMR chemical shift calculations than raw X-
ray derived coordinates (see e.g.: S. E. Ashbrook, M. E. Cutajar, C. J. Pick-
ard, R. I. Walton, S. Wimperis, Phys. Chem. Chem. Phys. 2008, 10, 5754–
Keywords: chelates · density functional calculations · materials
science · NMR spectroscopy · quantum chemistry
5764).
[
25] Depending on the substituent pattern, both ferromagnetic and antifer-
romagnetic behaviour has been reported for phenolic oxime Cu com-
plexes, namely, ref. [5a].
[
[
1] F. Habashi, Hydrometallurgy 2005, 79, 15–22.
2] M. A. Wilson, P. J. Bailey, P. A. Tasker, J. R. Turkington, R. A. Grant, J. B.
Love, Chem. Soc. Rev. 2014, 43, 123–134.
[26] See, for example: a) Y. Ruiz-Morales, G. Schreckenbach, T. Ziegler, Orga-
nometallics 1996, 15, 3920–3923; sometimes spin-orbit-effects can be
overriding: b) P. Hrobꢁrik, V. Hrobarikova, F. Meier, M. Repisky, S. Komor-
ovsky, M. Kaupp, J. Phys. Chem. A 2011, 115, 5654–5659; c) P. Hrobꢁrik,
V. Hrobarikova, A. H. Greif, M. Kaupp, Angew. Chem. Int. Ed. 2012, 51,
[
3] I. A. Smellie, R. S. Forgan, C. Brodie, J. S. Gavine, L. Harris, D. Houston,
A. D. Hoyland, R. P. McCaughan, A. J. Miller, L. Wilson, F. M. Woodhall, J.
Chem. Educ. 2016, 93, 362–367.
[4] About half of the more than 120 transition-metal phenolic bis(oxime)
complexes deposited in the Cambridge Structure Database (Version
10884–10888; Angew. Chem. 2012, 124, 11042–11046.
II
1
5.37, November 2015) contain Cu as metal centre, see for example:
[27] These sorb values (PBE0- =
3
/II, level, data not shown) are essentially the
A. G. Smith, P. A. Tasker, D. J. White, Coord. Chem. Rev. 2003, 241, 61–85.
5] a) A. M. Whyte, B. Roach, D. K. Henderson, P. A. Tasker, M. M. Matsushita,
K. Awaga, F. J. White, P. Richardson, N. Robertson, Inorg. Chem. 2011, 50,
same for paramagnetic 1 and a diamagnetic variant in which Cu has
been replaced with Zn, indicating that anisotropy effects from the para-
[
magnetic metal centre are negligible in our case.
[28] The D-tensor calculations did not converge at the PBE0- =
1
12867–12876; frozen solution: b) M. R. Healy, E. Carter, I. A. Fallis, R. S.
3
level; for the
Forgan, R. J. Gordon, E. Kamenetzky, J. B. Love, C. A. Morrison, D. M.
Murph, P. A. Tasker, Inorg. Chem. 2015, 54, 8465–8473.
6] I. Bertini, L. Emsley, M. Lelli, C. Luchinat, J. Mao, G. Pintacuda, J. Am.
Chem. Soc. 2010, 132, 5558–5558.
7] N. P. Wickramasinghe, Y. Ishii, J. Magn. Reson. 2006, 181, 233–243.
8] a) G. de Combarieu, M. Morcrette, F. Millange, N. Guillou, J. Cabana, C. P.
Grey, I. Margiolaki, G. Fꢅrey, J. M. Tarascon, Chem. Mater. 2009, 21,
dimer the results for the D-tensor obtained at the PBE/IGLO-II level (in
ꢀ
1
which the computed D-value is 17.4 cm ) were used along with the re-
1
[
3
maining quantities obtained at the PBE0- = /IGLO-II level.
[29] See, for example: M. Munzarova, in: Calculation of NMR and EPR Param-
eters. Theory and Applications (Eds.: M. Kaupp, M. Bꢀhl, V. G. Malkin),
Wiley-VCH, Weinheim, 2004, pp. 463–482.
[30] When applied to truly uncoupled systems with single unpaired elec-
trons, as for instance two methyl radicals 1000 ꢄ away of each other,
[
[
1602–1611; b) D. M. Dawson, L. E. Jamieson, M. I. H. Mohideen, A. C.
Chem. Eur. J. 2016, 22, 1 – 13 www.chemeurj.org
11
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
which the full system treated as a triplet, this procedure recovers the
results for a monomeric doublet.
31] Very similar results are obtained for the other possible doublet state,
for which the spin on the other terminal metal fragment has been flip-
ped in addition, that is, “fl›fl” (the resulting pNMR shifts are within d=
A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara,
K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O.
Kitao, H. Nakai, T. Vreven, J. J. A. Montgomery, J. E. Peralta, F. Ogliaro, M.
Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, T. Keith, R.
Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S.
Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Strat-
mann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L.
Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dan-
nenberg, S. Dapprich, A. D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz,
J. Cioslowski, D. J. Fox, Gaussian, Inc., Wallingford CT, 2010.
[39] a) A. D. Becke, Phys. Rev. A 1988, 38, 3098–3100; b) J. P. Perdew, Phys.
Rev. B 1986, 33, 8822–8824; c) J. P. Perdew, Phys. Rev. B 1986, 34, 7406.
[40] a) J. P. Perdew, in Electronic Structure of Solids (Eds.: P. Ziesche, H. Eis-
chrig, Akademie, Berlin, 1991; b) J. P. Perdew, Y. Wang, Phys. Rev. B
1992, 45, 13244–13249.
[
0.2 ppm of those for the “BS fl››” state in Table S3, Supporting Infor-
mation, data not shown).
[
32] This value is irrespective of the particular approximation for the spin-
Hamiltonian analysis, which uses the energy difference and information
ˆ
2
from the total spin and/or the expectation values of the S operator:
a) A. P. Ginsberg, J. Am. Chem. Soc. 1980, 102, 111; b) L. Noodleman,
E. R. Davidson, Chem. Phys. 1985, 100, 131; c) T. Soda, Y. Kitagawa, T.
Onishi, Y. Takano, Y. Shigeta, H. Nagao, Y. Yoshioka, K. Yamaguchi, Chem.
Phys. Lett. 2000, 319, 223–230.
[
33] The computed infinite-temperature limits are also consistent with this
1
assignment: For C1 and C2, diso =169.6 and 119.1 ppm, respectively,
are obtained (Table 6), in very good agreement with the observed reso-
[41] a) J. P. Perdew, M. Ernzerhof, K. Burke, J. Phys. Chem. 1996, 105, 9982–
9985; b) C. Adamo, V. Barone, J. Chem. Phys. 1999, 110, 6158–6170.
[42] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 2010, 132,
154104.
[43] Including Becke–Johnson damping: a) A. D. Becke, E. R. Johnson, J.
Chem. Phys. 2005, 122, 154104; b) E. R. Johnson, A. D. Becke, J. Chem.
Phys. 2006, 124, 174104.
2
+
nances in the diamagnetic Ni analogue (d=158.9 and 116.4 ppm, re-
spectively, D. M. Dawson, R. A. Doyle, S. E. Ashbrook, unpublished
work). The corresponding experimental values in Table 2 are not con-
clusive in that respect, because they contain the thermal effects on dorb
on top and are associated with some uncertainty from the long-range
extrapolation to 1/T!0).
[
34] The chemical shift predicted from the siso(orb) part of monomeric 1, d=
[44] a) A. J. H. Wachters, J. Chem. Phys. 1970, 52, 1033–1036; b) P. J. Hay, J.
Chem. Phys. 1977, 66, 4377–4384.
11.2 ppm, is similar to that in other diamagnetic complexes, in which
the resonances of these bridging protons have been observed, for ex-
ample, at d=10.9 ppm for the Pt complex in CCl solution (S. F.
4
Kaplan, V. Y. Kukushkin, S. Shova, K. Suwinska, G. Wagner, A. J. L. Pom-
beiro, Eur. J. Inorg. Chem. 2001, 1031–1038).
[45] a) W. J. Hehre, R. Ditchfield, J. A. Pople, J. Chem. Phys. 1972, 56, 2257–
2261; b) P. C. Hariharan, J. A. Pople, Theor. Chim. Acta 1973, 28, 213–
222.
[46] C. A. Guido, E. Bremond, C. Adamo, P. Cortona, J. Chem. Phys. 2001, 115,
201104.
[47] M. Munzarovꢁ, M. Kaupp, J. Phys. Chem. A 1999, 103, 9966.
[48] W. Kutzelnigg, U. Fleischer, M. Schindler, in NMR Basic Principles and
Progress, Vol. 23, Springer, Berlin, 1990, p.165.
[49] Orca, Version 3.0.3: a) F. Neese, WIREs Comput. Mol. Sci. 2012, 2, 73–78;
b) https://orcaforum.cec.mpg.de (accessed 20 May 2016).
II
[
[
35] R. A. Doyle, D. M. Dawson, I. A. Smellie, M. Bꢀhl, D. B. Cordes, A. M. Z.
Slawin, S. E. Ashbrook, unpublished results.
36] For example, for organic mixed-valence radicals: a) M. Renz, K. Theilack-
er, C. Lambert, M. Kaupp, J. Am. Chem. Soc. 2009, 131, 16292–16302;
b) M. Renz, M. Kess, M. Diedenhofen, A. Klamt, M. Kaupp, J. Chem.
Theory Comput. 2012, 8, 4189–4203.
[
[
37] J. Skibsted, H. J. Jakobsen, J. Phys. Chem. A 1999, 103, 7958–7971.
38] Gaussian 09, Revision B.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E.
Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Men-
nucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian,
Received: May 30, 2016
Published online on && &&, 0000
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
12
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
FULL PAPER
Quantum chemistry: Armed with good-
quality DFT calculations, pNMR spectros-
copy can give detailed insights into the
spin distribution of phenolic oxime cop-
per(II) complexes in the solid, and into
the way this can be modulated by inter-
molecular and substituent effects.
&
Materials Science
M. Bꢀhl,* S. E. Ashbrook,* D. M. Dawson,
R. A. Doyle, P. Hrobꢁrik, M. Kaupp,
I. A. Smellie
&
& – &&
Paramagnetic NMR of Phenolic Oxime
Copper Complexes: A Joint
Experimental and Density Functional
Study
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
13
ꢂ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ