Stopped-Flow 19F NMR Spectroscopic Analysis of a Protodeboronation Proceeding at the Sub-Second Time-Scale

Article abstract of DOI:10.1002/ejoc.202100290

In-situ NMR spectroscopic analysis of homogeneous reactions is an essential tool for mechanistic analysis in organic and organometallic chemistry. However, rapid non-equilibrium reactions, that are initiated by mixing, require specialized approaches. We report herein on a study of the factors that ensure quantitative results in a recently-developed technique for stopped-flow NMR spectroscopy. The influence of some of the key parameters on quantitation is studied by ¹⁹F NMR spectroscopic analysis of the kinetics and activation parameters for the base-catalyzed protodeboronation of highly-reactive polyfluorinated arylboronic acids, with half-lives as low as 0.1 seconds. The effects of spin relaxation, pre-magnetization, heat-transfer versus reaction enthalpy, and mixing-efficiency are analyzed in detail. We also compare and contrast choice of pulse angle, interscan delay, and use of pseudo real-time by interleaving, as means to achieve an optimal balance between temporal resolution and sensitivity.

Full text of DOI:10.1002/ejoc.202100290

Full Papers
doi.org/10.1002/ejoc.202100290
Stopped-Flow 19F NMR Spectroscopic Analysis of a
Protodeboronation Proceeding at the Sub-Second
Time-Scale
Ran Wei,[a] Andrew M. R. Hall,[a] Richard Behrens,[b] Mark S. Pritchard,[c] Edward J. King,[c] and
Guy C. Lloyd-Jones*[a]
In-situ NMR spectroscopic analysis of homogeneous reactions is
an essential tool for mechanistic analysis in organic and
organometallic chemistry. However, rapid non-equilibrium re-
actions, that are initiated by mixing, require specialized
approaches. We report herein on a study of the factors that
ensure quantitative results in a recently-developed technique
for stopped-flow NMR spectroscopy. The influence of some of
the key parameters on quantitation is studied by 19F NMR
spectroscopic analysis of the kinetics and activation parameters
for the base-catalyzed protodeboronation of highly-reactive
polyfluorinated arylboronic acids, with half-lives as low as
0.1 seconds. The effects of spin relaxation, pre-magnetization,
heat-transfer versus reaction enthalpy, and mixing-efficiency are
analyzed in detail. We also compare and contrast choice of
pulse angle, interscan delay, and use of pseudo real-time by
interleaving, as means to achieve an optimal balance between
temporal resolution and sensitivity.
Introduction
eters ubiquitous in chemistry departments, NMR spectroscopy
offers such a wealth of information that it remains one of the
most widely utilized techniques for kinetic analysis of organic
reactions.
The elucidation of chemical reaction mechanisms is essential for
the informed optimization of known processes, and a powerful
impetus in the design of new ones.[1] Tracing the components
of a reaction, and the chronology of their evolution,[2] offers
invaluable insights, including identification of intermediates,
side-products, and catalysts, and quantification of selectivity.
Techniques for determining reaction kinetics should ideally a)
provide structural information for identification of chemical
species evolving throughout the reaction; b) allow straightfor-
ward quantitation of concentrations; and c) be applicable under
the conditions under which the reaction of interest is usually
employed (concentration, solvent, temperature, etc.). In general,
in-situ NMR spectroscopy fulfils all of the requirements outlined
above and is applicable to a broad range of nuclei common in
organic and organometallic reactions. Indeed, with spectrom-
Reactions that proceed ’slowly’, by which we mean half-
lives, t1/2, on the order of 30 seconds or more, can be studied by
routine NMR techniques.[3] However, those that are ’rapid’
require a more specialized approach. Rapid reactions can be
divided into two classes (Scheme 1): systems at equilibrium and
systems that evolve irreversibly after initiation. The analysis of
rapid equilibria by NMR spectroscopy using line-shape
analysis,[4–6] magnetization or saturation transfer experiments,[7,8]
and spin-echo[9–11] or CPMG methods,[12,13] is well-established.
For example, SHARPER[14] allows activation parameters to be
derived from temperature-dependent line-broadening and
distortion effects, and the CPMG spin-echo pulse sequence
reveals protein motions at microsecond to millisecond
timescale.[15,16] However, monitoring rapid irreversible reactions
by NMR spectroscopy poses a considerably different challenge,
[a] R. Wei, Dr. A. M. R. Hall, Prof. Dr. G. C. Lloyd-Jones
School of Chemistry
The University of Edinburgh
Joseph Black Building, David Brewster Road, Edinburgh, EH9 3FJ, UK
E-mail: guy.lloyd-jones@ed.ac.uk
http://www.lloyd-jones.chem.ed.ac.uk
[b] Dr. R. Behrens
Laboratory of Engineering Thermodynamics (LTD)
Technische Universität Kaiserslautern
Erwin-Schrödinger-Straße 44, Kaiserslautern, 67663, Germany
[c] M. S. Pritchard, E. J. King
TgK Scientific Ltd.
Bradford on Avon, Wiltshire, BA15 1DH, UK
http://www.hi-techsci.com
Supporting information for this article is available on the WWW under
https://doi.org/10.1002/ejoc.202100290
© 2021 The Authors. European Journal of Organic Chemistry published by
Wiley-VCH GmbH. This is an open access article under the terms of the
Creative Commons Attribution License, which permits use, distribution and
reproduction in any medium, provided the original work is properly cited.
Scheme 1. Generic NMR analysis techniques for the kinetics of reversible,
and irreversible, rapid reactions A+B=C. Herein ’fast’ refers to reactions
with (pseudo) first-order half-lives, t1/2, on the order of seconds or lower.
Eur. J. Org. Chem. 2021, 2331–2342
2331
© 2021 The Authors. European Journal of Organic Chemistry
published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
especially when such processes are initiated by mixing, rather
than by a readily controlled external stimulus such as light.[17]
Over the last decade we have investigated a wide range of
anion-mediated processes involving organoboron[18–21] and
organosilicon[22–26] reagents, including hydrolytic activation and
degradation of boronic acids,[18,21] and their derivatives.[19,20] In
1
the majority of examples we were able to employ standard H,
Scheme 2. Base-catalyzed protodeboronation of polyfluorophenyl boronic
acids, with half-lives ranging from 7 months (3-fluorophenylboronate) to 2.6
milliseconds (2,3,4,5,6-pentafluorophenylboronate).[21]
11B, 19F, and 29Si NMR spectroscopic techniques to analyze the
reaction kinetics. However, some reactions, for example the
protodeboronation of polyfluorophenylboronic acids, 1,
Scheme 2,[21] required fast in-situ spectroscopic techniques, such
as SF-FT-IR, or very rapid ex-situ quench methods.[27–29]
can. These aspects led us to explore application of the two
general techniques to monitor relatively fast irreversible reac-
tions by NMR: rapid-injection,[30–35] and stopped-flow,[36–43] Fig-
ure 1. The primary difference between the two techniques lies
in the way they mix the analytes. For rapid-injection methods, a
solution containing the final reagent(s) needed to initiate the
reaction is injected at high speed into an NMR tube containing
the remainder of the reagents, using the tube itself as the
Results and Discussion
Whilst the IR/quench techniques allowed us to determine a
wide range of protodeboronation kinetics, n=0 to 5, Scheme 2,
they did not provide the key insights that NMR spectroscopy
Ran Wei conducted her undergraduate studies
at the University of Science and Technology
Beijing, in 2010. She obtained her Master’s
degree in Chemistry from the University of
Manchester, during which period she under-
took her MSc project focusing on pure shift
NMR and developed her interest in this area.
In 2019 she joined the Lloyd-Jones group as a
PhD student, her work primarily focuses on
new NMR methods and their application in
physical-organic mechanistic investigations.
Mark Pritchard is an electronics engineer with
a wide range of experiences in the mechanical
and electronic design of scientific instruments
and specialist products for chemistry and the
bio-physical sciences. He has specialized in
design techniques and project management
and has made major contributions to the
development of IR and NMR projects in the
Lloyd-Jones group Edinburgh.
Edward (’Ted’) King is a chartered engineer
with a wide range of experiences and knowl-
edge; especially in the design of scientific
instruments and specialist products for
chemistry and the bio-physical sciences. With
Andrew Hall graduated from Imperial College,
London with an MSci in Chemistry in 2014. He
then moved to the University of Bath to carry
out an integrated PhD at the Centre for
Sustainable Chemical Technologies, under the
supervision of Dr Antoine Buchard and Dr
Ulrich Hintermair. After completing his PhD in
2018 he carried out postdoctoral studies in
the group of Dr Giuseppe Pileio at the
University of Southampton, before moving to
the University of Edinburgh in January 2020.
His research in the Lloyd-Jones group focuses
on the development of new stopped-flow
instrumentation for monitoring fast reactions
by NMR spectroscopy.
a
strong background in UV/Vis, IR, and
fluorescence spectroscopy and related phys-
ical methods applied to fast reactions in
solution, his interests have more recently
expanded to NMR, through a rich and produc-
tive collaboration with the Lloyd-Jones group
at the University of Edinburgh.
Guy Lloyd-Jones studied Chemical Technology
at Huddersfield Polytechnic, obtained his
doctorate at Oxford with John Brown FRS, did
postdoctoral research with Andreas Pfaltz at
Basel, and began his independent career in
1996 at Bristol. In 2013 he moved to the
Forbes Chair of Organic Chemistry at the
University of Edinburgh. Over the last 25 years
his research has focused on the mechanism
and kinetics of synthetic processes and reac-
tions, and more recently on the development
of new methods for UV, IR and NMR analysis,
in a series of highly productive collaborations
with Ted King and Mark Pritchard.
Richard Behrens obtained his Bachelor’s de-
gree from FH Kaiserslautern and his Master’s
degree in mechanical engineering with
a
specialization in fluid process engineering
from TU München in 2014. He then worked at
the Laboratory of Engineering Thermodynam-
ics at TU Kaiserslautern under the supervision
of Prof. Erik von Harbou and Prof. Hans Hasse.
In 2018, he visited the Lloyd-Jones group as
guest researcher to compare and contrast
stopped-flow NMR designs at Kaiserslautern
and Edinburgh. Richard received his PhD in
2019 and is currently working at BASF SE in
Ludwigshafen in the process research depart-
ment for gas scrubbing.
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2332
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
In 2018 we reported on a detailed study of the mechanism
[23]
of anion-initiated CF3 transfer from TMSCF3
in which we
employed a stopped-flow insert (Figure 1c) that we custom-
built for the investigation. In our subsequent analyses of CF2
generation,[24] and arene CÀ H silylation by TMSCF3,[46] we have
optimized our design and use of the stopped-flow system.
Herein we report in detail on the physical and NMR parameters
that help enable quantitative and accurate kinetics when
implementing the technique, using the rapid protodeborona-
tion (Scheme 2) of 2,3,4,6-tetrafluorophenyl boronic acid 1a,
and 2,3,4,5,6-pentafluorophenyl boronic acid 1b, as challenging
examples for analysis.[21,47]
General Design Features. There are a number of distinct
features between the approach discussed herein (Figure 2b)
and the conventional design for stopped-flow NMR spectro-
scopy (Figure 2a). The major difference is the use of three
independent drive syringes, thereby allowing systematic varia-
tion in reagent concentrations, using a single set of stock
solutions (A, B, C). The use of computer-controlled drives allows
the flow to be stopped by the resistance present in the open-
ended circuit (see ’waste’ in Figure 2b) after arresting the
movement of the drives. The trigger signal is sent to the NMR
console from the syringe drive unit, rather than from a
microswitch on a stop syringe. This configuration reduces the
acute pressure and flow oscillations induced by the ’hard stop’
of a classic stop-syringe (Figure 2a). In both designs (Figure 2a,
Figure 2b) reactant nuclei are polarized in pre-magnetization
reservoirs (see later) placed immediately before the mixer. The
latter employs
a sufficiently high flow-rate to generate
Figure 1. Schematic representations of a) rapid injection apparatus designed
to insert into a standard NMR probe, b) a customized stopped-flow probe,
and c) a stopped-flow apparatus designed to insert into a standard NMR
probe.
turbulence and enable the efficient mixing required to initiate
the reaction with a clearly defined start point. The flow is also
used to purge the previous contents from the sample cell and
replace it with the nascent reaction mixture primed for NMR
spectroscopic measurement. Herein we have employed flow
rates ranging from 0.2 mL/s to 2 mL/s, resulting in dead-times
ranging from 0.75 to 0.075 seconds, respectively, to prime the
3 mm outer-diameter 180 μL capacity glass flow-cell. The
capacity of the flow-cell in the detection region of the NMR
receiver coil is approximately 93 μL and this results in about 3-
fold lower S/N compared to a normal 5 mm NMR tube.
However, in contrast to custom-built stopped-flow NMR probes
(Figure 1b), the system described herein is inserted directly into
a standard NMR probe head (Figure 1c) allowing use of highly-
sensitive He or N2 cryoprobes.
Establishing the Boltzmann Distribution. In conventional
NMR spectroscopy, a Boltzmann distribution of spin populations
is developed in the analyte solution in the probe-head at the
center of the magnetic field (B0) of the spectrometer. Radio-
frequency (RF) pulses applied at a direction perpendicular to B0
disturbs this distribution, and it is the relaxation of the excited
spins back to equilibrium that gives rise to the NMR signal.
Whilst it takes a significant time for the spins to develop their
initial Boltzmann distribution (99% equilibration requires 5×T1),
in routine NMR spectroscopy this ’pre-magnetization’ occurs
during the pre-acquisition processes of locking, shimming and
tuning, after the sample tube has been inserted into the
spectrometer.
mixing chamber, Figure 1a. In the stopped-flow technique the
reactants are passed through a mixing device and the nascent
reaction mixture delivered to an NMR flow cell, Figure 1b,
Figure 1c.
Stopped-Flow NMR. Since the first reports of stopped-flow
NMR spectroscopy techniques in the 1970s,[36–38] there has been
a focus on minimizing the phenomenological dead-time: the
time duration between the initiation of the reaction and the
start of data acquisition. There are three main contributors to
the dead-time in a stopped-flow NMR experiment: i) mixing, ii)
transport from the mixer to the NMR flow cell, and iii) cessation
of flow and pressure oscillations. An early publication from
Ernst et al.[38] highlighted the potential of stopped-flow NMR in
detecting fast irreversible reactions, despite the limitations of
instrumentation at the time, and line-shape distortions caused
by oscillations induced by the rapid (’hard’) stopping motion.
More recent developments have included rapid mixing cells
using pneumatic drives and solenoid valves,[40] customized
probes for greater reagent polarization and temperature
control,[42–44] and high pressure stopped-flow devices.[39] Landis
and Christianson have also reported simulations of the changes
in intensity, line-width, and phase, in NMR spectra arising from
very rapid reactions.[45]
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2333
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
Figure 2. Schematic representations of a) classical dual-input ’hard-stop’ stopped-flow apparatus, and b) variable-ratio triple input ’soft-stop’ stopped-flow
apparatus; see text for full discussion. c) A simulated first-order reaction profile (reactant!product) showing the difference between the real concentrations
(solid lines) versus those determined from the intensity of NMR signals arising from non-pre-magnetized reactants (dashed lines), when the reaction rate (k) is
the same as the longitudinal relaxation rate (1/T1). d) An analogous simulation where k@1/T1. Inset: magnification of the first 1.5 seconds; see text for full
discussion.
The solid lines in Figure 2c show the temporal concentration
profiles for a first-order process (’reactant’!’product’) where
kT1 =1. The dashed lines show the apparent concentrations that
would be determined based on the NMR signal intensities
when the reaction is initiated without prior establishment of the
Boltzmann distribution of spins (for details see the SI), indicative
of a major underestimation of the reactant concentration. When
the reaction rate substantially exceeds the relaxation rates of
the studied spins (e.g. Figure 2d, kT1 =12) the NMR signal for
the reactant can barely be detected, as it is consumed more
rapidly than it is magnetized. Moreover, the NMR signal for the
product predominantly reflects the magnetization process, not
the change in concentrations due to the chemical reaction of
interest.
Pre-magnetization. In the system shown in Figure 2b polar-
ization of the reactants (A, B, C) is achieved by a set of three
pre-magnetization reservoirs located just above the probe
head, see Figure 3. This has the potential to afford high (but in
contrast to bespoke probe heads, Figure 2a, not complete)
polarization of the nuclei in the reactants before they are mixed
and delivered to the flow-cell. The extent of the polarization
(Figure 3a) differs mainly based on the delivered volumes and
the concentric arrangement of the coil-shaped reservoirs (Fig-
ure 3b). Since a total delivered volume (A+B+C) of �600 μL is
required to fully purge the previous sample and refill the flow-
cell with nascent reaction mixture, the pre-magnetization
profile allows definition of the limits in individual delivery
volumes (A, B, C) to afford >90% pre-magnetization.
Thus, for stopped-flow NMR analysis of reactions that
proceed faster than the Boltzmann distributions of the various
nuclei involved can be achieved, attaining signal intensities that
directly reflect concentrations requires that the reaction
components are exposed to the magnetic field (B0) for >5×T1
before the reaction is initiated by mixing. Unlike the NMR
spectrometers employed in the early stages of development of
stopped-flow techniques,[36–38] modern superconducting mag-
nets have very little stray field. Thus, the reactants must be pre-
magnetized in reservoirs located as close as possible to the
field-center and the NMR probe head.
Mixing efficiency. Rapid mixing is essential in all stopped-
flow experiments to ensure a well-defined reaction start time.
The system employed herein passes the reagent streams (A, B,
C) through a passive micromixer and down to the reaction
flow-cell through a narrow bore glass capillary, ideally as a
homogeneous reaction mixture. To assess this aspect, the
mixing time was estimated by spectrophotometry using the
Villermaux-Dushmann protocol.[48,49] This method uses compet-
ing iodate/iodine and acid/base reactions, where the concen-
tration set is chosen such that in a well-mixed sample the acid
is entirely neutralized by a fast reaction with the borate buffer
before any iodine is formed. In a poorly mixed sample, pockets
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2334
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
Figure 4. Mixing time characterization as a function of flow rate through the
mixer (Figure 3b), estimated using the Villermaux-Dushmann method.[48,49]
Transitions between laminar and turbulent flow regimes in the mixer were
estimated using the Reynolds number calculated for water using the
nominal dimensions of the mixer. The specific power dissipation of the mixer
was estimated from the mixing time as described by Falk.[50]
residence time, meaning that the sample is fully mixed by the
time it exits the mixer. Using three inputs (A, B, C) resulted in
slightly shorter mixing times than using only two (A, B), likely
due to parallel lamination of the reagent streams.[51]
Reaction Enthalpy. One of the advantages of custom-built
stopped-flow NMR probes (Figure 1b) is that a single, highly-
effective, fluid-based thermostatting system can be incorpo-
rated in the probe head where the reaction is conducted.[42,43] In
contrast, a system based on insertion of a stopped-flow cell into
a standard NMR probe (Figure 1c) requires two thermostatic
systems. An auxiliary system provides thermostatting to the
pre-magnetization, mixing, and sample transport stages,[52] with
control transitioning to the spectrometer during measurement.
For the work herein we used an aqueous ethylene glycol heat
transfer medium, and a Pt100 temperature probe located close
to the mixer to pre-calibrate the auxiliary system to be within
�0.5 K of the spectrometer thermostat. After delivery of the
nascent reaction to the NMR flow-cell and cessation of the flow,
the reaction temperature must remain sufficiently stable to
provide reliable kinetic data. Standard NMR spectrometers use
gas flow (dry air or N2) to stabilize the temperature of analyte
solutions in 5 mm NMR tubes, a system that is not designed to
regulate fast process with large enthalpies of reaction.
Figure 3. a) Pre-magnetization profiles of input solutions A, B, C. Volumes
ranging from 50–900 μL in increments of 50 μL were studied, the dashed
lines through the datapoints are for visual guidance only. b) vertical cross-
section of the pre-magnetization reservoirs. The reagents from solution
syringes flow through the pre-magnetization coils in an upward manner
until they reach the mixer. The homogeneously mixed volume is then
delivered back down to the flow-cell in the probe head at the magnetic field
center. The relative vertical position of the coils to NMR probe and the
corresponding magnetic field strength experienced are illustrated in the left-
hand section of the figure.
of high acid concentration are only partially neutralized by the
buffer, allowing the slower iodate/iodine reaction to occur. The
concentration of iodine was measured by UV-Vis spectropho-
tometry and the mixing time was estimated using a model
derived from known reaction rates and ionic strengths.[48,50]
H4BO4À þ Hþ ! H3BO3 þ H2O
IO3À þ 5IÀ þ 6Hþ ! 3I2 þ 3H2O
To evaluate the limits of the spectrometer thermostat, we
used the technique of von Harbou et al. to estimate the error in
the apparent rate constant for a reaction, as a function of its
half-life, enthalpy change, and concentration.[43] We first
determined the heat transfer coefficient from the flow-cell to
the surrounding gas-flow in the spectrometer. Methanol was
The mixing time for the system employed herein (Figure 2b
and Figure 3) was determined to be between 0.9 and 14.1 milli-
seconds (Figure 4); a range comparable to other similarly sized
micromixers.[50] As expected, the mixing time is inversely
proportional to flow rate, with higher flow rates resulting in
greater power dissipation and more-efficient mixing. The
mixing time is between 4 and 9 times shorter than the
°
heated to 50 C using the auxiliary system, and then delivered
via a standard stopped-flow ’shot’ to the 3 mm glass NMR tube
°
flow-cell that had been pre-heated to 30 C in the spectrometer
gas flow (Figure 5a). The cooling rate (Figure 5b) was estimated
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2335
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
constant. The energy and component balances result in three
coupled partial differential equations of the concentration and
the temperature of the reactants, and of the temperature of the
glass of the NMR flow cell. The partial differential equations
were solved numerically (see SI) and used to construct a non-
isothermal model, at four different initial concentrations of
substrate, Figure 6a–d.
The results show that unless the reaction is very exothermic,
the standard spectrometer thermostat system (Figure 5a)
provides sufficient control for reactions with half-lives as low as
100 msec, when the initial concentration is 0.05 M (Figure 6b)
or lower. Higher concentrations can be used for weak to
moderately exothermic reactions, affording enhanced sensitivity
without compromising the accuracy of the kinetics.
Longitudinal Relaxation. The relaxation time constant T1 is
an essential but sometimes overlooked parameter for consid-
eration in NMR kinetic studies, especially for rapid reactions
when data acquisition needs to be repeated in a relatively short
time window. For quantitative NMR, T1 constants govern the
repetition rate between successive acquisitions in the basic
experiment for reaction monitoring (Figure 7a). In general, an
interscan delay, tR, of 5×T1 is necessary to recover >99% of the
°
magnetization between the application of 90 pulses. Using an
interscan delay that is too short will result in incomplete
relaxation of the nuclei between scans and saturation of some
or all signals, which compromises quantification.
One approach is to use smaller pulse angles, trading
°
reduced sensitivity for enhanced data density. For example, 30
pulses effectively shorten the required interscan delay between
two transients to 2.9×T1 for >99% relaxation recovery, at the
°
penalty of a 50% sensitivity loss compared to using 90 pulses.
°
Use of 10 pulses further enhances the temporal resolution by
shortening the interscan delay to just 0.77 *T1, but at the cost of
a 6-fold reduction in sensitivity. Therefore, a knowledge of the
T1 constant(s), including that of the internal standard(s), aids
selection of acquisition parameters to achieve optimal balance
between temporal resolution and sensitivity. Moreover, incom-
plete pre-magnetization will result in spins continuing to be
polarized inside the reaction-cell, in a T1-dependent manner
(Figure 7c). Thus, if the reactants or internal standards have
substantially different T1 constants, unequal magnetization rates
will compromise the accuracy of quantification of the data
acquired within the reaction period, as discussed below.
Figure 5. a) Experimental determination of the heat transfer coefficient using
the methanol NMR thermometer technique.[53] The co-axial thermostat circuit
is analogous to that of a known continuous-flow NMR system.[54] b)
Temperature evolution of pre-heated methanol upon injection into the flow-
cell including fitted parameters; the air flow-rate was 535 L/h.
1
using the H NMR chemical shift difference between CH3/OH
Pulse angle, θ, and Sensitivity. Accurate quantitation in
NMR requires a high signal-to-noise ratio (S/N), and thus
acquisition time, receiver gain, and signal-averaging (number of
FIDs) are important parameters. However, in rapid irreversible
reactions the analytes evolve too fast to allow signal averaging,
constraining the number of FIDs to one per time point. In
addition, the interscan delay, tR, between consecutive pulses
determines the data density achievable in a fixed window of
time. In our earlier studies,[23,24,46] we routinely used a small
protons.[53] With the assumption that heat transfer from the
methanol occurs by radial conduction, the heat transfer
coefficient, (α, 110 WmÀ 2 KÀ 1) was then determined from the
flow-cell dimensions, and standard properties of the methanol
(see SI).
A chemical reaction that proceeds with a substantial
enthalpy change will establish a radial temperature gradient in
the reaction medium, thus leading to a radial gradient of the
rate of reaction, and in turn to a radial concentration gradient.
The impact of this effect on the measured versus real rates of
reaction for a generic first-order process was evaluated using
the Arrhenius equation. All properties (heat conductivity,
enthalpy of reaction, heat capacity, etc.) were considered to be
°
radiofrequency pulse angle, typically θ=10 , to minimize tR and
achieve higher data density, equation (1).[55] Herein, we revised
our approach and used θ values optimized for spectral
sensitivity (S/N) versus data density, especially in the context of
interleaving (see later).
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2336
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
Figure 6. Non-isothermal modelling of a first order reaction (A!P) proceeding in the stopped-flow NMR cell, at four initial concentrations of A. The graphs
show the deviation (% - contour scale) of the measured net reaction rate from the real reaction rate that would occur if the sample were thermally
homogeneous and at the temperature of gas passed over the external surface of the tube by the spectrometer thermostat system. The x-axis and y-axis show
how the deviation varies as a function of reaction half-life and enthalpy. The calculations are based on the empirical heat transfer coefficient Figure 5. See SI
for full details.
!
tR=
q ¼ arccos eÀ
data density through spectral ’interleaving’. The principle for
this is straightforward; the reaction is performed multiple times,
incrementing the delay (Figure 7a) between the trigger and the
first excitation pulse for each experiment. Data points obtained
from separate experiments can then be combined to produce
an interleaved dataset.
The process is illustrated schematically in Figure 9 for a
reaction that is complete in about 4 seconds. To simulate the
spectra, the T1 constant of the measured species was set to 2 s,
and the interscan delay tR to 1.5 s, with the optimal pulse angle
T1
(1)
The half-life of 2,3,4,6-tetrafluorophenyl boronate (2a)
generated in-situ from boronic acid 1a,[21] is 10.4 seconds at
300 K, Figure 8. The calculated enthalpy-change for the process
2a!3a is À 0.5×102 kJmolÀ 1 and thus near-isothermal kinetics
(Figure 6b) are expected when the initial concentration is
0.05 M or lower.[56] The kinetics (2a!3a, Figure 8) were
analyzed by stopped-flow 19F NMR using a variety of pulse
angles, θ, with larger angles affording beneficial increases in
S/N. However, because of the different longitudinal relaxation
rates (1/T1) in the 19F nuclei monitored in the substrate (2a), the
internal standard (TFA), and the product (3a), differential signal
saturation develops when the pulse angle is too large for the
°
calculated as 60 . Three separate series (blue, red, yellow) were
simulated with incremented trigger delays, and each series
used to generate one data subset. These three data subsets
were combined to generate the final dataset which contains
three-fold higher data density. The trigger delays, interscan
delay tR, and pulse angle can be adjusted based on the
estimated reaction rate, known T1 constants, and the required
sensitivity and data density. The general stopped-flow method
allows reactions to be conveniently and reproducibly initiated
as a series of identical ’shots’ from the same set of stock
solutions. The interleaving experiment solely requires accurate
timing of the trigger signal sent to the NMR spectrometer, so
that the delay period can be reliably incremented.
°
interscan delay, tR. In the example in Figure 8c, where θ=90 ,
this phenomenon leads to a noticeably-distorted reaction
profile (compare with Figure 8b), and a significant error in the
measured rate constant (see SI).
Sub-second Protodeboronation Kinetics via Interleaving.
As is evident from Equation 1, when larger pulse angles, 0<θ<
°
90 , are employed to gain S/N, the interscan delay, tR, must be
extended to avoid saturation. This in turn leads to lower data
density and thus impacts on the quantitation of the kinetics of
reactions that proceed faster than the longitudinal relaxation
rate, 1/T1. We have addressed this by manually enhancing the
We have previously measured the rate of the base-catalyzed
protodeboronation of highly-reactive[47] pentafluorophenyl bor-
onic acid 1b using the rapid quenched flow technique.[21] The
boronate (2b) undergoes the fastest protodeboronation of any
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2337
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
Figure 7. a) pseudo-2D pulse sequence used for reaction monitoring. The
pulse angle, θ, and interscan delay, tR, are adjustable acquisition parameters.
b) an example of a first-order reaction profile acquired using the pulse
sequence shown in a). c) continued polarization of analytes in the flow tube
(circles) after being ‘shot’ from the stopped-flow solution syringes, and
simulated polarization of analytes without pre-magnetization (lines).
Figure 8. Stopped-flow 19F NMR analysis of the protodeboronation of
°
°
°
boronate 2a (0.04 M) using 10 , 45 , and 90 pulses. In (a) and (b), non-linear
regressions of the first-order decay of 2a were identical (t1/2 =10.4 s). In (c)
differential saturation (see text and SI) results in deviation from the true rate
(see dashed lines). 19F NMR longitudinal relaxation constants, T1, blue-
labelled F in 2a, 1.347�0.009 s, red-labelled F in 3a, 2.810�0.005 s; TFA
2.209�0.002 s.
of the wide series of (hetero)arylboronates that we have tested
to date,[18,21] with a half-life of <0.003 s at 343 K. As shown
below (Figure 10), even at 300 K the process is very rapid, with
a half-life of just 0.12 seconds.
Using the interleaving technique, we analyzed the kinetics
of the protodeboronation of 2b at 300 K using stopped-flow 19
F
NMR spectroscopy, Figure 10. With the reaction >97% compete
in 0.6 seconds, only a single data point can be obtained from Conclusions
°
each stopped-flow ’shot’. Consequently, θ was set to 90 to
maximize the sensitivity, and a series of 10 shots interleaved.
The phenomenological deadtime (0.22 s at a flow rate of 1 mL/
s) resulted in the kinetics being analyzed for the last 20% of
conversion. By increasing the flow rate to 2 mL/s, to achieve a
phenomenological deadtime of 0.15 s, the activation parame-
ters were able to be determined between 284 K and 323 K. We
previously reported the rate of protodeboronation of 2b at
343 K using the rapid quench flow technique.[21] At this temper-
ature the rate of reaction is considerably faster than the
phenomenological dead time of the stopped-flow NMR appara-
tus and the protodeboronation complete (>99% 2b!3b, in
�0.02 s) before the first NMR pulse can be applied (0.15 s).
Nonetheless, the data determined by stopped-flow NMR
between 284 K and 323 K (red data points in Figure 11)
correlate reasonably well with the rate determined by rapid
quench flow at 343 K (blue datapoint).
We have demonstrated how using an insert that is compatible
with a standard NMR spectrometer fitted with a He or N2
cryoprobe, the stopped-flow NMR technique can be used to
study the kinetics of rapid irreversible reactions such as the
protodeboronation of 2,3,4,5,6-pentafluorophenyl boronate
(2b, t1/2 =0.12 s at 300 K). Pre-magnetization coils (Figure 3)
located close to the magnet core of the spectrometer induce
about 90% polarization of the reagents before they are rapidly
mixed (Figure 4) and enter the detection region. The design
uses computer-controlled stepper-motor syringe drives to allow
variable mixing ratios of reagents. The mixing time, pre-
magnetization volumes, and phenomenological deadtime have
also been assessed. The limiting reactant concentration for
processes with large enthalpies of reaction, with sufficient
thermostatting by the spectrometer to maintain approximately
isothermal reaction kinetics has been evaluated by NMR cooling
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2338
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
Figure 9. Schematic overview of the process of interleaving a series of spectra using an incremented trigger delay. The process effects manually enhanced
temporal resolution when studying reactions that are too fast to acquire enough datapoints within a single ’shot’ of the stopped-flow system. The three
different colored subsets represent data obtained from identically initiated experiments; they are then combined to afford three-fold higher data density than
the individual components.
Figure 10. Interleaved 19F stopped-flow spectra of the protodeboronation of 2,3,4,5,6-pentafluorophenyl boronic acid (1b) mediated by KOH in 1:1 dioxane/
H2O at 300 K. Inset: first-order decay of the boronate intermediate (2b), each differently-colored data point is generated from an interleaved set of two
spectra, with an incremented trigger delay (only the first ten spectra are shown), grey line: non-linear regression to obtain the first-order rate constant,
k
obs =5.9 sÀ 1; t1/2 =0.12 s.
experiments, and mathematical modelling (Figure 5 and Fig-
ure 6). We have also conducted an evaluation of the key NMR
parameters for quantitative analysis of the kinetics of relatively
rapid irreversible reactions (equation 1, and Figure 7 and Fig-
ure 8). These parameters ensure sufficient nuclei relaxation
between pulses, which is particularly important for reactions
where the relaxation rate (1/T1) is similar to, or smaller than, the
reaction rate (k). Optimization of the pulse angle, θ, and
interscan delay, tR, allows the optimal balance between S/N and
data density to be obtained. The data-density can also be
manually enhanced by interleaving spectra (Figure 9 and
Figure 10) from multiple experiments with different time offsets.
We are currently developing an alternative design for stopped-
flow NMR spectroscopy that affords significant reductions in the
phenomenological dead-time, and are also measuring the rates
of protodeboronation of polyfluorinated boronate esters,[47]
relative to the corresponding boronic acids.[21,47] We will report
in full on both of these investigations in due course.
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2339
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
°
tube. Each experiment was acquired with 3 loops of 90 pulses, and
the interscan delay tR was set to 15.76 s to ensure full relaxation
recovery.
T1-dependent polarization measurement
Pre-magnetization effects with respect to different T1 constants
were examined by interleaved 31P stopped-flow experiments
performed on a mixture of 0.5014 M PPh3 (T1 =15.997�0.001 s)
and 0.5006 M O=PPh3 (T1 =3.601�0.001 s) in 9:1 THF/ H2O. The
pulse sequence shown in Figure 7a was used in 26 individual
experiments with incremented trigger delays from 0.039 s to
°
44.04 s. Each experiment was acquired with 3 loops of 90 pulses,
and the interscan delay tR was set to 70.11 s to ensure full relaxation
recovery.
Mixing time measurement
Mixing time measurements were performed using the Villermaux-
Dushmann protocol.[48] Stock solutions of A) potassium iodide
(0.032 M) and potassium iodate (0.006 M) in a sodium hydroxide/
boric acid buffer (0.09 M) and B) sulfuric acid (0.06 M) were
prepared. The stopped-flow instrument was prepared and primed
with reagent line A containing the iodide/iodate stock solution, line
B the sulfuric acid solution and line C containing distilled water.
The waste line from the stopped-flow instrument was connected to
a UV-Vis spectrometer flow cell with a 5 mm path length. For each
mixing time measurement, syringes A and B were driven at a
constant flow rate for the full volume (2.5 mL per syringe). For
measurements using three syringes, line C was filled with the
sulfuric acid solution and the three syringes were driven in a 2:1:1
ratio (2.5:1.25:1.25 mL). The absorbance at 353 nm was monitored
until a stable signal was observed, with the final absorbance
measured after the flow was stopped. The measurements were
repeated three times at each flow rate and averaged. Mixing times
and power dissipation were estimated from the measured
absorbance values using the master curve reported by Falk et. al.:[48]
Figure 11. a) Base-catalyzed protodeboronation of pentafluorophenyl bor-
onic acid (1b) via the corresponding boronate 2b (See Figure 10) at 313 K
(t1/2 =0.05 s). (b) Eyring analysis of rate data obtained by stopped-flow 19
F
NMR analysis at 284 K, 300 K, 305 K, 313 K, and 323 K; the datapoint colored
blue is a value previously determined by rapid quench flow, then 19F NMR
analysis (343 K; t1/2 =0.003 s).[21]
Experimental Section
NMR spectra were recorded on
a 400 MHz Bruker Ascend
spectrometer equipped with a 5 mm BBFO nitrogen cryoprobe.
Stopped-flow experiments were performed with the custom built
stopped-flow system, that we have previously outlined.[23] Temper-
ature control in the pre-magnetization coils and mixer was
achieved using a Huber Petite Fleur heater/chiller circulating a
tm ¼ 0:33 � ðabsorbanceÞ�
½Hþ�À 4:55½KI�À 1:5½KIO3� ½NaOH�À 2½H3BO3�À 2
5:8
tm ¼ 0:15 eÀ 0:45
d
�
50/50 ethylene glycol water mixture, with
a Pt100 resistor
tm ðtheoreticalÞ
¼
arcsinhð0:76 PeÞ
8 u
terminated just below the mixer block. Temperature control of the
reaction volume in the flow cell was achieved by use of the
standard NMR spectrometer system in which dry air (or N2) is
passed up and around the sample. All NMR data were processed by
Bruker TopSpin® 3.6.2 and MestReNova (version 14.0.0).
Where tm is the mixing time (s), ɛ is the dissipated power (W.kgÀ 1),
d is the characteristic diameter of the mixer (m), ū is the mean
velocity of the fluid (m.sÀ 1) and Pe is the Peclet number (propor-
tional to Reynolds number).
Pre-magnetization profiles measurement
Protodeboronation of 2,3,4,6-tetrafluorophenyl boronic acid
1a
Pre-magnetization profiles for three solution syringes were gener-
ated from 19F stopped-flow experiments on samples of 0.1007 M 2-
fluorophenyl boronic acid (T1 =1.649�0.020 s) and 0.0999 M 4-
fluorophenyl boronic acid (T1 =2.062�0.021 s) in 1:1 dioxane/H2O.
Stock solutions were prepared separately to allow flexible control
of variable input volumes. The pre-magnetization profile of each
syringe was studied individually, during which time the studied
syringe was filled with the stock solution and all other syringes with
solvent. Input volumes from the syringe containing the stock
solution were incremented from 50 μL to 900 μL in equal steps of
50 μL. For volumes <600 μL, solvent was added from one of the
other syringes to maintain a sufficient volume to flush the NMR
A stock solution of 0.083 M 2,3,4,6-tetrafluorophenyl boronic acid
(T1 =1.347�0.009 s for signal δ=À 109 ppm) and 0.0197 M tri-
fluoroacetic acid (IS) (T1 =2.123�0.004 s before reaction and T1 =
2.209�0.002 s after reaction, δ=À 76 ppm, CF3CO2K) was freshly
prepared in 1:1 dioxane/H2O. A solution of 0.3022 M (3 equivalent)
of KOH in 1:1 dioxane/H2O was prepared separately. T1 constants of
19F nuclei in the starting materials and product were measured
individually in the flow-cell of the stopped flow system using
standard ’inversion recovery’ experiments (see SI for full details).
The reaction was initiated by simultaneously inputting 300 μL of
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2340
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
[1] B. A. Abel, C. A. L. Lidston, G. W. Coates, J. Am. Chem. Soc. 2019, 141,
boronic acid and 300 μL of base solution from syringe A and
syringe B, at a flow rate of 1 mL/s. 19F NMR spectra were then
acquired with a minimum trigger delay of 0.039 s. The interscan
delay was 1.36 s (relaxation delay of 0.5 s and acquisition time of
12760–12769.
[2] D. G. Blackmond, J. Am. Chem. Soc. 2015, 137, 10852–10866.
[3] F. Susanne, D. S. Smith, A. Codina, Org. Process Res. Dev. 2012, 16, 61–
64.
[4] H. S. Gutowsky, A. Saika, J. Chem. Phys. 1953, 21, 1688–1694.
[5] H. M. McConnell, J. Chem. Phys. 1958, 28, 430–431.
[6] J. Kaplan, P. P. Yang, G. Fraenkel, J. Chem. Phys. 1974, 60, 4840–4845.
[7] S. Forsén, R. A. Huffman, J. Chem. Phys. 1963, 39, 2892–2901.
[8] J. Jeener, B. H. Meier, P. Bachmann, R. R. Ernst, J. Chem. Phys. 1979, 71,
4546–4553.
°
0.8 s respectively) for 10 pulses, and 1.86 s (relaxation delay of 1 s
°
°
and acquisition time of 0.8 s respectively) for 45 and 90 pulses.
Protodeboronation of 2,3,4,5,6-pentafluorophenyl boronic
acid 1b
[9] Z. Luz, S. Meiboom, J. Chem. Phys. 1963, 39, 366–370.
[10] A. Allerhand, H. S. Gutowsky, J. Chem. Phys. 1964, 41, 2115–2126.
[11] A. Allerhand, H. S. Gutowsky, J. Chem. Phys. 1965, 42, 1587–1599.
[12] H. Y. Carr, E. M. Purcell, Phys. Rev. 1954, 94, 630–638.
[13] S. Meiboom, D. Gill, Rev. Sci. Instrum. 1958, 29, 688–691.
[14] A. B. Jones, G. C. Lloyd-Jones, D. Uhrín, Anal. Chem. 2017, 89, 10013–
10021.
[15] D. D. Boehr, H. J. Dyson, P. E. Wright, Chem. Rev. 2006, 106, 3055–3079.
[16] A. Mittermaier, L. E. Kay, Sci. J. 2006, 312, 224–228.
[17] Y. Ji, D. A. DiRocco, C. M. Hong, M. K. Wismer, M. Reibarkh, Org. Lett.
2018, 20, 2156–2159.
A stock solution of 0.0993 M 2,3,4,5,6-pentafluorophenyl boronic
acid (T1 =0.636�0.005 s for signal δ=À 134 ppm) and 0.0261 M
trifluoroacetic acid (IS) was freshly prepared in 1:1 dioxane/H2O. A
solution of KOH 0.3212 M (3 equivalent) in 1:1 dioxane/H2O was
prepared separately. The thermostats on the spectrometer and
auxiliary system were both set to 300 K. The reaction was initiated
by computer-controlled simultaneous delivery of 300 μL of the
boronic acid solution, and 300 μL of base solution, at a flow rate of
1 mL/s. Interleaved 19F NMR spectra were acquired with incre-
mented trigger delays of 0.039, 0.08, 0.12, 0.16, 0.20, 0.25. 0.30,
0.35, 0.4, and 0.45 s. The interscan delay was 0.36 s, and the pulse
[18] P. A. Cox, A. G. Leach, A. D. Campbell, G. C. Lloyd-Jones, J. Am. Chem.
Soc. 2016, 138, 9145–9157.
[19] A. J. J. Lennox, G. C. Lloyd-Jones, J. Am. Chem. Soc. 2012, 134, 7431–
7441.
°
angle was 60 .
A solution of 0.1031 M 2,3,4,5,6-pentafluorophenyl boronic acid and
0.0246 M trifluoroacetic acid (IS) was freshly prepared in 1:1
dioxane/H2O. A solution of KOH 0.2994 M (3 equivalent) in 1:1
dioxane/H2O was prepared separately. The thermostats on the
spectrometer and auxiliary system were both set to 313 K. The
reaction was initiated by computer-controlled simultaneous deliv-
ery of 300 μL of the boronic acid solution, and 300 μL of base
solution, at a flow rate of 2 mL/s. Interleaved stopped-flow spectra
were acquired with incremented trigger delays of 0.039, 0.06, 0.08,
0.10, 0.12, 0.14. 0.16, 0.18, 0.20, and 0.22 s. The interscan delay was
[20] J. A. Gonzalez, O. M. Ogba, G. F. Morehouse, N. Rosson, K. N. Houk, A. G.
Leach, P. H.-Y. Cheong, M. D. Burke, G. C. Lloyd-Jones, Nat. Chem. 2016,
8, 1067–1075.
[21] P. A. Cox, M. Reid, A. G. Leach, A. D. Campbell, E. J. King, G. C. Lloyd-
Jones, J. Am. Chem. Soc. 2017, 139, 13156–13165.
[22] E. Kühnel, D. D. P. Laffan, G. C. Lloyd-Jones, T. Martínez Del Campo, I. R.
Shepperson, J. L. Slaughter, Angew. Chem. Int. Ed. 2007, 46, 7075–7078;
Angew. Chem. 2007, 119, 7205–7208.
[23] C. P. Johnston, T. H. West, R. E. Dooley, M. Reid, A. B. Jones, E. J. King,
A. G. Leach, G. C. Lloyd-Jones, J. Am. Chem. Soc. 2018, 140, 11112–
11124.
[24] A. García-Domínguez, T. H. West, J. J. Primozic, K. M. Grant, C. P.
Johnston, G. G. Cumming, A. G. Leach, G. C. Lloyd-Jones, J. Am. Chem.
Soc. 2020, 142, 14649–14663.
°
0.36 s, and the pulse angle was 90 .
[25] L. T. Ball, T. J. A. Corrie, A. J. Cresswell, G. C. Lloyd-Jones, ACS Catal.
2020, 10, 10420–10426.
[26] H. J. A. Dale, C. Nottingham, C. Poree, G. C. Lloyd-Jones, J. Am. Chem.
Soc. 2021, 143, 2097–2107.
Acknowledgements
[27] F. Song, R. D. Cannon, M. Bochmann, J. Am. Chem. Soc. 2003, 125, 7641–
Professor Erik von Harbou (Kaiserslautern) facilitated the visit of
R.B. to Edinburgh and provided assistance in the calculations of
heat-transfer coefficients and reaction thermodynamics. Dr.
Andrew Leach (Manchester) conducted the DFT calculations on
the reaction enthalpies for protodeboronation. Professor John
Blacker (Leeds) made very useful suggestions regarding estimates
of mixing efficiency and the use of the Villermaux-Dushmann
protocol. Dr. Ruth Dooley, Dr. Ariana Jones and Professor Dušan
Uhrín (Edinburgh) made numerous valuable contributions and
suggestions in the early stages of this project; Hannah Hayes
(Edinburgh) provided detailed advice for the studies of the kinetics
of the protodeboronation reactions.
7653.
[28] T. E. Barman, S. R. W. Bellamy, H. Gutfreund, S. E. Halford, C. Lionne, Cell.
Mol. Life Sci. C. 2006, 63, 2571–2583.
[29] J. F. Eccleston, S. R. Martin, M. J. Schilstra, in Biophys. Tools Biol. Vol. One
Vitr. Tech., Academic Press, 2008, pp. 445–477.
[30] J. F. McGarrity, C. A. Ogle, Z. Brich, H. R. Loosli, J. Am. Chem. Soc. 1985,
107, 1810–1815.
[31] K. H. Mok, T. Nagashima, I. J. Day, J. A. Jones, C. J. V. Jones, C. M. Dobson,
P. J. Hore, J. Am. Chem. Soc. 2003, 125, 12484–12492.
[32] S. E. Denmark, B. M. Eklov, P. J. Yao, M. D. Eastgate, J. Am. Chem. Soc.
2009, 131, 11770–11787.
[33] R. Franco, A. Favier, P. Schanda, B. Brutscher, J. Magn. Reson. 2017, 281,
125–129.
[34] S. E. Denmark, B. J. Williams, B. M. Eklov, S. M. Pham, G. L. Beutner, J.
Org. Chem. 2010, 75, 5558–5572.
[35] S. Bowen, C. Hilty, Phys. Chem. Chem. Phys. 2010, 12, 5766–5770.
[36] J. Grimaldi, J. Baldo, C. McMurray, B. D. Sykes, J. Am. Chem. Soc. 1972,
94, 7641–7645.
[37] J. J. Grimaldi, B. D. Sykes, Rev. Sci. Instrum. 1975, 46, 1201–1205.
[38] R. O. Kühne, T. Schaffhauser, A. Wokaun, R. R. Ernst, J. Magn. Reson.
1979, 35, 39–67.
Conflict of Interest
The authors declare no conflict of interest.
[39] S. Funahashi, K. Ishihara, S. I. Aizawa, T. Sugata, M. Ishii, Y. Inada, M.
Tanaka, Rev. Sci. Instrum. 1993, 64, 130–134.
[40] P. V. Yushmanov, I. Furó, J. Magn. Reson. 2005, 175, 264–270.
[41] M. D. Christianson, E. H. P. Tan, C. R. Landis, J. Am. Chem. Soc. 2010, 132,
11461–11463.
Keywords: Kinetics · NMR spectroscopy · Protodeboronation ·
Rapid reactions · Stopped-flow
[42] A. L. Dunn, C. R. Landis, Magn. Reson. Chem. 2017, 55, 329–336.
[43] E. von Harbou, R. Behrens, J. Berje, A. Brächer, H. Hasse, Chem. Ing. Tech.
2017, 89, 369–378.
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2341
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202100290
[44] A. Scheithauer, A. Brächer, T. Grützner, D. Zollinger, W. R. Thiel, E.
[53] C. Ammann, P. Meier, A. Merbach, J. Magn. Reson. 1982, 46, 319–321.
[54] D. A. Foley, E. Bez, A. Codina, K. L. Colson, M. Fey, R. Krull, D. Piroli, M. T.
Zell, B. L. Marquez, Anal. Chem. 2014, 86, 12008–12013.
[55] R. R. Ernst, W. A. Anderson, Rev. Sci. Instrum. 1966, 37, 93–102.
[56] Estimated enthalpies have previously been reported, using DFT
methods.[21] The enthalpy change for the initial conversion to the
boronate (1a!2a) is calculated to be larger, about À 1.6×102 kJmolÀ 1.
However this equilibrium process proceeds extremely rapidly[14] and is
complete within the timescale of mixing, under control of the auxiliary
thermostat.[52]
von Harbou, H. Hasse, Ind. Eng. Chem. Res. 2014, 53, 17589–17596.
[45] M. D. Christianson, C. R. Landis, Concepts Magn. Reson. Part A 2007, 30 A,
165–183.
[46] A. García-Domínguez, P. H. Helou de Oliveira, G. T. Thomas, A. R.
Sugranyes, G. C. Lloyd-Jones, ACS Catal. 2021, 11, 3017–3025.
[47] Y. P. Budiman, S. A. Westcott, U. Radius, T. B. Marder, Adv. Synth. Catal.
2021, 363, DOI 10.1002/adsc.202001291.
[48] J.-M. Commenge, L. Falk, Chem. Eng. Process. Process Intensif. 2011, 50,
979–990.
[49] L. Falk, J.-M. Commenge, in Micro Process Eng., Wiley-VCH Verlag GmbH
& Co. KGaA, Weinheim, Germany, 2013, pp. 145–173.
[50] L. Falk, J.-M. Commenge, Chem. Eng. Sci. 2010, 65, 405–411.
[51] C.-Y. Lee, W.-T. Wang, C.-C. Liu, L.-M. Fu, Chem. Eng. J. 2016, 288, 146–
160.
[52] On exiting the mixer, the nascent reaction mixture is transported to the
flow-cell via a narrow-bore glass capillary that passes through the liquid
heat-transfer medium thus dissipating any significant enthalpy of
mixing.
Manuscript received: March 8, 2021
Revised manuscript received: March 25, 2021
Accepted manuscript online: March 26, 2021
Eur. J. Org. Chem. 2021, 2331–2342
www.eurjoc.org
2342
© 2021 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH