A simple flowcell for reactionmonitoring by NMR

Article abstract of DOI:10.1002/mrc.2610

A simple, cheap and flexible flowcell based on a standard 5mmNMR tube, designed for the monitoring of reactions but of wide applicability, is described. No modification of the NMR instrument is needed, allowing the system to be employed with any conventional NMR probe and magnet. The system is robust and economical in use of reagents, and can be used for studying both homogeneous and heterogeneous reactions.

Full text of DOI:10.1002/mrc.2610

Research Article
Received: 16 February 2010
Revised: 19 March 2010
Accepted: 25 March 2010
Published online in Wiley Interscience: 29 April 2010
(
www.interscience.com) DOI 10.1002/mrc.2610
A simple flowcell for reaction monitoring
by NMR
a
b
a∗
M. Khajeh, M. A. Bernstein and G. A. Morris
A simple, cheap and flexible flowcell based on a standard 5 mm NMR tube, designed for the monitoring of reactions but of wide
applicability, is described. No modification of the NMR instrument is needed, allowing the system to be employed with any
conventional NMR probe and magnet. The system is robust and economical in use of reagents, and can be used for studying
both homogeneous and heterogeneous reactions. Copyright ꢀc 2010 John Wiley & Sons, Ltd.
1
Keywords: NMR; H; flowcell; kinetics; reaction monitoring
Introduction
PEEK HPLC tubing (8) runs concentrically to the base of the tube.
The PEEK tubing is held concentric by grooved PTFE spacers at the
bottom of the tube and above the active volume, and by four PTFE
grooved sleeves (18) that also serve to reduce the dead volume of
the cell. The NMR tube is held in the headpiece by an M8 threaded
bush (4) that compresses an O-ring (5). The inlet HPLC tubing runs
continuously through the headpiece, and the outlet tube (9) is
mated to the headpiece, both held by O-rings (3) compressed by
M3 threaded bushes (2). The flowcell is held in the appropriate
turbine (17), at the appropriate depth, for the probe used. As a
precaution against probe damage from reactant leaks, a liquid
sensor (7; Sensortechnics, type OLP01B0F3) is placed below the
headpiece. If the turbine design permits (as with Varian turbines),
the active element of the sensor can be placed directly on the
turbine, and a glass tube (16) can be glued to the turbine with
epoxy resin to contain any leaks; if not (as with Bruker turbines),
the sensor can be placed on a PEEK disc through which the NMR
tube passes and which is held within the glass tube, in both
cases sealed with O-rings. The liquid sensor is connected to an
electronic interlock (21) that switches off the pump in the event
of a leak being detected. (To date, the only occasion on which
the safety system was activated was when a loose union above
the headpiece leaked as a result of a reactant precipitating in the
system.)
The application of NMR spectroscopy to kinetic and mechanistic
studies of chemical reactions is not limited to static methods,
in which a solution is placed in an NMR tube and inserted into
[
1]
a magnet for measurements, but also includes stopped or
[
2]
continuous flow methods. Flow NMR techniques have been
[
3]
used since 1951 in a variety of areas, such as studies of protein
folding,[ food science[6] and HPLC-NMR ; recent examples of
the use of flow NMR for reaction monitoring include Refs [8–10].
Most applications of flow NMR use dedicated flow probes.
These allow the balance between sample volume, sensitivity and
resolution to be optimised, but are expensive and are relatively
inflexible, typically being optimised for a single nucleus or group
of nuclei; they also generally have a limited temperature range
and are very vulnerable to blockage. This paper describes a simple,
cheap and flexible flowcell based on a standard 5 mm NMR tube.
Several designs for rapid mixing cells based on 5 mm NMR tubes
4,5]
[7]
[
9,11]
have recently been published
; in contrast, the cell described
here is designed for continuous or stopped-flow use (but not
for reactions on sub-minute timescales). No modification of the
NMR instrument is needed, allowing the system to be employed
with any conventional 5 mm or greater NMR probe and magnet.
The system is robust and economical in use of reagents, and
can be used for studying both homogeneous and heterogeneous
reactions. The only materials in contact with the reaction mixture
are glass, poly(tetrafluoroethylene) (PTFE) and poly(ether ether
ketone) (PEEK), accommodating a wide range of solvents and
reactants. The reaction vessel is placed outside of the magnet;
therefore the composition of the sample can be varied during the
experiment if any reagents need to be added. It is also relatively
straightforward to perform reactions at high or low temperature,
or under an inert atmosphere. The spectral quality obtainable is
comparable to that available from conventional high-resolution
probes.
A number of features of the design are noteworthy. First, the use
of a concentric inlet pipe minimises disturbance of the magnetic
field homogeneity by maintaining cylindrical symmetry, while the
use of narrow bore tubing ensures that the frequency-shifted
signal from within the tubing is very small. Any deviation from
concentricity will lead to significant degradation in line shape
and difficulty in shimming. Second, the cell geometry is intended
−
1
for relatively low flow rates (<2 ml min ), since pre-polarisation
takes place largely in the portion of the NMR tube between the
∗
Correspondence to: G. A. Morris, School of Chemistry, The University of
Experimental
Manchester, Oxford Road, Manchester M13 9PL, UK.
E-mail: Gareth.Morris@manchester.ac.uk
Flowcell
a School of Chemistry, The University of Manchester, Manchester M13 9PL, UK
b AstraZeneca R&D Charnwood, Loughborough, Leics LE11 5RH, UK
The flowcell (Figs 1 and 2) consists of a standard 5 mm NMR tube
(
6) fitted with a PEEK headpiece (1), through which a length of
Magn. Reson. Chem. 2010, 48, 516–522
Copyright ꢀc 2010 John Wiley & Sons, Ltd.

A simple flowcell for reaction monitoring
(
(
2)
3)
(
1)
Top
Side
Base
(
4)
(
5)
0
mm
20
Figure 1. Scale drawing (left) and photographs (right) of the flowcell headpiece assembly. (1) headpiece; (2) inlet and outlet pipe bushes; (3) and
5) O-rings; (4) NMR tube retaining bush. The left part of the composite photograph shows the inlet and outlet pipes and their respective bushes and
(
O-rings, together with the liquid sensor cable, the latter and the inlet pipe running continuously through the headpiece. The top right shows the top
view, and the bottom right the bottom view, of the headpiece.
(9)
(
21)
Safety cut-off
system
pump
(8)
(15) (13)
(13) (12)
(11)
(14)
(8) (9)
(10)
(
16)
(
(
2)
3)
(
8)
(
17)
(1)
(
18)
(
(
5)
4)
(
6)
(7)
(19)
(20)
(6)
Figure 2. Structure of the flow system used. (1) flowcell headpiece; (2) inlet and outlet pipe bushes; (3) and (5) O-rings; (4) NMR tube retaining bush;
6) NMR tube; (7) optical liquid sensor; (8) inlet pipe; (9) outlet pipe; (10) reaction vessel; (11) reaction vessel outlet pipe; (12) and (15) unions; (13) PTFE
peristaltic tubing; (14) peristaltic pump; (16) glass tube, glued to turbine (17); (18) long grooved PTFE spacers; (19) short grooved PTFE spacers;
20) cross-section of spacers; (21) safety cutoff.
(
(
base and the active volume. If higher flow rates are required, for
which the residence time in the pre-polarisation volume is less
than the spin–lattice relaxation time, an extra reservoir could be
used above the flowcell headpiece, or the direction of flow could
be reversed. Third, the system is designed for low-pressure, fail-
safe operation; if the pressure rises too much (significantly above
should yield well before the pressure limit of the NMR tube itself,
leaking into the space above the turbine and triggering the liquid
detector. The system may be used with HPLC pumps, but in this
case a low pressure-limit should be used. Fourth, the PEEK HPLC
tubing usedmayeasilybereplacedwiththosewithdifferentinside
diameters, to accommodate solvents of different viscosities within
the pressure constraints of the system.
−
2
1
MN m ), the O-ring holding the NMR tube to the headpiece
Magn. Reson. Chem. 2010, 48, 516–522
Copyright ꢀc 2010 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/mrc

M. Khajeh, M. A. Bernstein and G. A. Morris
Flow system
A sample homogeneous reaction
The flowcell is designed to be used with an external reaction
vessel, outside of the magnet, with reactant mixture circulated by
a low-pressure pump. The system used here, illustrated in Fig. 2,
comprises a reaction vessel (10), HPLC PEEK and PTFE piping,
unions, peristaltic pump (14), flowcell headpiece (1) and NMR tube
Inthereactionof2-methoxyphenylacetatewithphenylethylamine
in aqueous solution, transesterification (Scheme 1(a)) and hydrol-
ysis (Scheme 1(b)) proceed in parallel. The hydrolysis is catalysed
by the amine, which is gradually consumed, complicating kinetic
analysis.
(
6). The first inlet pipe (11; PTFE tubing, length 30 cm, i.d. 1.6 mm,
The two second-order parallel reactions in Scheme 1 can be
summarised as follows:
o.d. 3.2 mm) leads from the reaction vessel (10) to the peristaltic
pump (14; Cole-Parmer, type KH-77912-07), connected by a union
k1
(
(
12). The pump uses 4-mm o.d., 2 mm i.d. PTFE peristaltic tubing
13; Cole-Parmer, type WW 77390-50). The output of the pump is
A + B −−−→ C + D
A + B −−−→ C + AH + E
k2
+
(1)
connectedbyaunion(15)totheflowcellinlettube(8) (PEEK,length
m, i.d. 0.5 mm, o.d. 1.6 mm) that leads directly to the bottom
3
+
where A denotes phenylethylamine; AH , the phenylethylammo-
nium ion; B, 2-methoxyphenyl acetate; C, 2-methoxyphenol; D,
phenylethylacetamide and E, acetate ion. The rates of change of
species A–E are then
of the NMR tube (6). The outlet pipe (9; PEEK tubing, length 3 m,
i.d. 0.5 mm, o.d. 1.6 mm leads from the headpiece (1) back to the
reaction vessel. The total volume of reaction mixture required
to fill the pipes, pump and flowcell is approximately 3.5 ml; this
could be reduced somewhat by the use of lower bore and shorter
peristaltic tubing.
The flowcell was placed in the turbine (17) and lowered by its
input and output tubing into the standard 5 mm diameter indirect
detection probe of a 300 MHz Varian INOVA spectrometer. All
d[A]
=
=
=
=
=
−k1([A] − [E])[B]
dt
d[B]
−k1([A] − [E])[B] − k2([A] − [E])[B]
k1([A] − [E])[B] + k2([A] − [E])[B]
k1([A] − [E])[B]
dt
d[C]
◦
experiments were conducted at room temperature (20 ± 2 C).
dt
d[D]
Results
dt
d[E]
Spectra
k2([A] − [E])[B]
(2)
dt
Spectra obtained with the flowcell typically showed good
resolution and acceptable line shape, with a full width at half-
height of around 1 Hz both under static conditions and at low (ca
To follow this reaction, the flow system was initially filled by
pumping a solution of phenylethylamine and trimethylsilylpropi-
onic acid, sodium salt (TSP) (3.5 ml of 0.03 M phenethylamine in
D2O, and 300 µl of 0.24 M TSP in D2O) into the system. After filling,
flow was stopped to allow locking and shimming of the magnetic
field and setting up of pre-saturation of the residual HDO signal. A
mixtureof2 mlof0.03 M2-methoxyphenylacetate(Sigma)inD2O,
−
1
1
ml min ) flow rate. Figure 3 shows the spectrum obtained for a
line shape test solution of chloroform in acetone-d6 containing a
small amount of ethanol stabiliser. The full width at half-height of
theTMSsignalwas0.75 Hz, onlyslightlygreaterthanthatroutinely
obtained with conventional non-spinning samples.
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
ppm
8
7
6
5
4
3
2
1
0 ppm
1
Figure 3. H spectrum obtained with the flow cell under static conditions for a solution containing protiochloroform, ethanol and tetramethylsilane in
deuteroacetone.
www.interscience.wiley.com/journal/mrc
Copyright ꢀc 2010 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2010, 48, 516–522

A simple flowcell for reaction monitoring
h
OMe
(
a)
O
C
a
g
c
CH3
e
NH2
OH
O
NHCOCH3
D O
2
b
+
f
+
OMe
d
h
OMe
O
C
(
b)
c
CH3
OH
+
a
NH3+
O
D O
i
2
-
b
CH COO
3
OMe
d
NH2
Scheme 1. Reactions of 2-methoxyphenyl acetate with phenylethylamine in D2O. Lower-case letters denote assignments for methylene and methyl
protons in the spectra shown in Fig. 4.
Figure 4. Time course array for the solution of phenylethylamine and 2-methoxyphenyl acetate in D2O, with every 35th spectrum from the first (bottom)
to the last (top) shown, from 0 to 4 h reaction time. The lower-case letters from ‘a’ to ‘i’ correspond to the methyl and methylene signals highlighted in
Scheme 1. During the course of the reaction, the chemical shifts of signals ‘a’, ‘b’ and ‘h’ change as a result of the change in pH as hydrolysis proceeds.
2
0
ml of 0.03 M phenylethylamine (Alfa Aesar) in D2O and 300 µl of
.24 M TSP [sodium 3-(trimethylsilyl)-propionate-2,2,3,3,-d4; Goss
functions, with width, amplitude, frequency, phase and zero- and
first-order baseline corrections as variable parameters. In the case
of the triplet signals a, b, e and f, the relative intensities of the
inner and outer triplet components were constrained to equal
that of the central component, and the triplet splittings were held
constant;inbothcases, thisisagoodapproximationforthedegree
of strong coupling present. For the first spectrum in the series,
the initial parameter values for the singlet Lorentzian functions
were estimated by inspection. From then on, each successive
spectrum was fitted using starting parameters extrapolated from
the preceding fits. This minimises the number of iterations
required, reduces problems with signals that drift in frequency
and allows signal assignments to be retained even when the
chemical shift ordering changes during a reaction. For the four
triplet signals, fitting was performed in reverse order, starting with
the last spectrum, because of the overlap between triplets b and
f at the start of the reaction. In the very early spectra, where f is
completely buried under b, the intensities of the f signals were
constrained to equal those of e, allowing b and f to be separated
in the fitting.
Scientific]asreferencewasthenprepared,andimmediatelyplaced
in the reaction vessel. The input pipe to the peristaltic pump was
placed in the reaction vessel, pumping initiated at 1 ml/min and
1
acquisition of a time course of H spectra initiated. For the first
3.5 min of the time course, the output from the flowcell was di-
verted to waste, to discard the initial contents of the system; the
output was then returned to the reaction vessel to recycle the
reacting mixture. The total duration of the experiment was 5 h,
with a recycle delay of 7 s, 4 transients per acquisition, a saturation
delay of 4 s and a 90 pulse of duration 9 µs, for each H spectrum.
The spectral time course array for the reaction is illustrated in
Fig. 4.
All the data obtained were Fourier transformed, reference
deconvoluted using a 2.4 Hz Lorentzian target line shape,
normalised to correct systematic drifts in receiver gain and
baseline-corrected in Vnmr 6.1C. Data were then exported to
◦
1
[
12]
Mathematica version 6.0
and fitting. The peaks marked in Fig. 4 were fitted to Lorentzian
as a text file for statistical analysis
Magn. Reson. Chem. 2010, 48, 516–522
Copyright ꢀc 2010 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/mrc

M. Khajeh, M. A. Bernstein and G. A. Morris
(
a) 0.035
.03
.025
.02
.015
.01
.005
(b) 0.035
0.03
(c) 0.035
0
0.03
0.025
0.02
0
0
0
0.025
0.02
0
0.015
0.01
0.015
0.01
0
0.005
0.005
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
Time/h
Time/h
Time/h
Figure 5. Experimental (dots) and calculated (solid lines) kinetic plots for reactants and products in the reactions of Scheme 1. (a) Kinetic plot for
phenethylamine (reactant) and N-phenylethylacetamide (product), signals ‘b’ and ‘f’ in Fig. 4. (b) Kinetic plot of 2-methoxyphenyl acetate (reactant) and
−
3
2
5
1
-methoxy phenol (product), signals ‘d’ and ‘h’ in Fig. 4. (c) Kinetic plot of acetate, signal ‘i’ in Fig. 4. Optimised values of k1 and k2 were 4.0 × 10 and
−3 −1 −1
.9 × 10 l mol
s
, respectively. The effects of incomplete mixing are seen in the oscillations in the reactant curves (a) and (b) for about the first
0 min of the reaction; these data were omitted from the fitting.
O
c
C
H
H
a
b
d
NH2
THF/acetic acid
N
C
Me
+
H2O
+
Me
p-tolualdehyde
cyanoborohydride
cyanoborohydride
Me
i
e
f
CH2
g
N
h
NH
CH2
Me
CH2
main product
byproduct
i
Me
Scheme 2. Reductive amination of benzylamine with p-tolualdehyde, involving p-dimethyl benzyl benzylimine as an intermediate. The imine is reduced
to p-dimethyl benzyl benzylamine as the main product; the by-product (signals h, i) was tentatively identified as 2-(p-dimethyl benzyl) benzylamine but
this was not confirmed. The protons for which signals are assigned in the spectra below are marked with lower-case letters.
The integrals obtained were converted to reactant concentra-
tions using the known initial total concentration. Signals a and b
reflect total amine concentration, as a result of rapid exchange
A sample heterogeneous reaction
Heterogeneous reactions are difficult to study in static conditions
in an NMR tube. As the magnetic susceptibilities of solids and
liquids are normally different, the homogeneity of the magnetic
field is severely perturbed, and it is difficult to maintain the solid
component in suspension. One of the advantages of the flow
system described is the ability to study heterogeneous reactions
while avoiding having solid particles in the detection region of the
probe. As a test heterogeneous reaction, the reductive amination
of an aldehyde (Scheme 2) was studied. As shown in Scheme 3,
+
between free amine (A) and ammonium (AH ) species, so the free
amine concentration was obtained by subtraction of the acetate
concentrationknownfromsignali.Theconcentrationtimecourses
werefittedtonumericalsolutionsofEqn (2),withtheresultsshown
in Fig. 5. The slight oscillations in reactant concentration seen in
Fig. 5(a) and (b) are caused by imperfect mixing in the pipework,
pump and flowcell at the start of the experiment; not all of the
initial contents of the system were flushed out by the reactant
mixture, with the result that small damped oscillations in reactant
concentration occur with a period equal to the transit time for
the system. This problem can be reduced (at the expense of an
increase in the total volume of reactants required) by increasing
the time for which the flowcell output is discarded, or (at the risk
of trapping bubbles) by starting the reaction with the flow system
empty.
k1
A
A
B
C
k2
Scheme 3. Parallel first-order kinetics for the conversion of imine (A) to
product (B) and by-product (C).
www.interscience.wiley.com/journal/mrc
Copyright ꢀc 2010 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2010, 48, 516–522

A simple flowcell for reaction monitoring
Figure 6. Time course array for a solution of benzylamine and p-tolualdehyde in THF; every 10th spectrum is shown the first (bottom) to the last (top).
The lower-case letters from ‘a’ to ‘i’ correspond to proton signals highlighted in Scheme 2. The starred signals are ×1: carboxylic proton of acetic acid, in
exchange with residual water; ×2 overlapped aromatic signals of reactants and products; ×3 THF signal, partially saturated; ×4 impurity; ×5 methyl of
acetic acid; ×6 THF signal, not saturated. The signals from the reactant aldehyde are not observed, as the initial imine formation is very fast. One of the
methylene signals of the putative by-product is not assigned, and is assumed to be overlapped by another signal. Multiple signals of minor species are
seen. The effect of small temperature changes during initial equilibration can be seen in the moving COOH signal ×1.
dialkylated amine,[13] but these could not be completely avoided.
The reaction mixture consisted of 2.43 ml of 0.26 M benzylamine in
THF, 2.43 ml of 0.13 M p-tolualdehyde in THF, 0.34 ml glacial acetic
g
acid, 1.82 ml THF, 434 mg polymer-supported cyanoborohydride
d
and 250 µl TMS (Scheme 3).
Spectra were run unlocked, with the THF resonance at 3.6 ppm
1
3
pre-saturated; broadband C decoupling was used during
1
3
acquisition to avoid C satellites of THF signals overlapping
with signals of interest. The total duration of the experiment was
i
6
h 19 min, with a recycle delay of 8 s, 16 transients per spectrum,
◦
a saturation delay of 3 s, saturation power 5 dB and a 30 flip
angle (3 µs pulse width). Immediately after mixing, the reaction
solution (7 ml) was pumped into the dry, empty flow system at a
flow rate of 1 ml/min. Data acquisition was started when the first
drop emerged from the output pipe; representative spectra from
the time course are shown in Fig. 6.
0.0
0.5
1.0
1.5
Time/h
Figure 7. Kinetic plot for the signals ‘d’, ‘i’, ‘g’ in Fig. 6. The dots represent
the experimental data, which were fitted to first-order parallel kinetics
Initial data processing, export and fitting to Lorentzian
lineshapes were carried out as described for the homogeneous
reactions above. The integrals of the methyl peaks d, g and i were
converted to concentrations and their time courses fitted to the
analytical solutions for first-order parallel kinetics (Scheme 3) with
the results shown in Fig. 7.
−4
−1 −1
s and k2 of
(
6
solid lines), yielding a k1 of 1.86 ± 0.08 × 10 l mol
.08 ± 0.011 × 10−4 l mol
−1 −1
s .
the reaction chosen was between benzylamine (Alfa Aesar) and p-
tolualdehyde(Sigma)in(protio-)tetrahydrofuran(THF)/aceticacid
with cyanoborohydride on a polymer support (Sigma, product no.
1
7337). The reaction cell containing the solid phase and reactant
mixture was continuously stirred to keep the polymer-supported Conclusions
cyanoborohydride in suspension, while a stainless steel frit on
the inlet of the pipe leading to the pump was used to retain
the solid in the reaction vessel and keep it from blocking the
pipework or being pumped into the probe. An excess of primary
amine was used in an attempt to avoid side products such as
The flowcell described offers a simple and economical route to the
acquisition of spectra under stopped or continuous flow. While
primarily designed for reaction monitoring, it is adaptable to a
wide range of uses; one example would be to allow hydrogen
Magn. Reson. Chem. 2010, 48, 516–522
Copyright ꢀc 2010 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/mrc

M. Khajeh, M. A. Bernstein and G. A. Morris
gas to be bubbled through a sample in para-hydrogen-induced
polarisationexperiments. Theresolution, lineshapeandsensitivity
obtainable are only slightly poorer than for a standard 5 mm NMR
tube, and the system is compatible with a wide range of probes
and nuclei. It is compatible with standard NMR instrumentation
and requires only small (millilitre) volumes of solution.
[3] G. Suryan, Proc. Indian Acad. Sci. 1951, 33, 107.
[
4] M. Hamang, A. Sanson, L. Liagre, V. Forge, P. Berthault, Rev. Sci.
Instrum. 2000, 71, 2180.
[
5] N. A. J. van Nuland, V. Forge, J. Balbach, C. M. Dobson, Acc. Chem.
Res. 1998, 31, 773.
[6] C. Tellier, F. Mariette, P. S. B. G. A. Webb, M. J. Mccarthy, Ann. Rep.
NMR Spectrosc. 1995, 31, 105.
[
[
7] N. Watanabe, E. Niki, Proc. Jpn. Acad. Ser. B 1978, 54, 194.
8] M. A. Bernstein, M. Scarontefinovi, C. J. Sleigh, Magn. Reson. Chem.
Acknowledgements
2007, 45, 564.
[
9] A. Mix, P. Jutzi, B. Rummel, K. Hagedorn, Organometallics 2010, 29,
Support from the Engineering and Physical Sciences Research
Council (grant references EP/D05592X and EP/E057888/1) and
from AstraZeneca (studentship to MK) is gratefully acknowledged.
442.
[10] M. Maiwald, H. H. Fischer, Y.-K. Kim, K. Albert, H. Hasse, J. Magn.
Reson 2004, 166, 135.
11] P. V. Yushmanov, I. Fur o´ , J. Magn. Reson. 2005, 175, 264.
[
[
12] Wolfram Research, Inc., Mathematica, Version 6.0, Champaign, IL,
2
007.
References
[13] A. F. Abdel-Magid, K. G. Carson, B. D. Harris, C. A. Maryanoff,
R. D. Shah, J. Org. Chem. 1996, 61, 3849.
[
[
1] D. B. Green, J. Lane, R. M. Wing, Appl. Spectrosc. 1987, 41, 847.
2] E. Bayer, K. Albert, J. Chromatogr. A 1984, 312, 91.
www.interscience.wiley.com/journal/mrc
Copyright ꢀc 2010 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2010, 48, 516–522