A modular, low footprint and scalable flow platform for the expedient α-aminohydroxylation of enolizable ketones

Article abstract of DOI:10.1039/d0gc04395h

The unique reactivity profile of α-chloronitroso derivatives is expressed to its fullest potential through the development of an integrated, modular and scalable continuous flow process for the electrophilic α-aminohydroxylation of various enolizable ketones. Flow conditions contribute to mitigating the high reactivity and toxicity of α-chloronitroso derivatives and provide an efficient, versatile and safe protocol for the α-aminohydroxylation of ketones with a minimal footprint. Fundamental aspects of the α-aminohydroxylation process were computed by DFT and further supported the experimental observations, hence leading to the unprecedented α-chloronitroso-based α-aminohydroxylation of primary, secondary and tertiary substrates. Recycling of the carbon backbone of the α-chloronitroso derivatives provides a high atom economy for the preparation of value-added molecules. This work showcases α-chloronitroso derivatives as economic and efficient vehicles for transferring electrophilic synthons of hydroxylamine toward nucleophilic enolates. A representative range of precursors and analogs of pharmaceutical active ingredients, including WHO essentials and drugs in shortage (such as epinephrine and ketamine), are prepared within minutes according to a fully concatenated process. The process features sequential modules with distinct unit operations including chemical transformations and multiple in-line extractions. The process relies on an upstream chemical Generator that manages the preparation of α-chloronitroso derivatives and that feeds downstream a series of α-aminohydroxylation modules. The setup is amenable to the addition of libraries of compounds for feeding upstream the process of discovery in medicinal chemistry and is transposable to pilot scale. Several layers of in-line analytical procedures are featured to improve process control and safety.

Full text of DOI:10.1039/d0gc04395h

Volume 23
Number 6
21 March 2021
Pages 2187-2542
Green
Chemistry
Cutting-edge research for a greener sustainable future
rsc.li/greenchem
ISSN 1463-9262
PAPER
Jean-Christophe M. Monbaliu et al.
A modular, low footprint and scalable flow platform for the
expedient α-aminohydroxylation of enolizable ketones

Green Chemistry
View Article Online
View Journal | View Issue
PAPER
A modular, low footprint and scalable flow
platform for the expedient α-aminohydroxylation
of enolizable ketones†
Cite this: Green Chem., 2021, 23,
2336
Victor-Emmanuel H. Kassin, a Romain Morodo,a Thomas Toupy,a
Isaline Jacquemin,a Kristof Van Hecke, b Raphaël Robiette c and
a
Jean-Christophe M. Monbaliu
*
The unique reactivity profile of α-chloronitroso derivatives is expressed to its fullest potential through the
development of an integrated, modular and scalable continuous flow process for the electrophilic
α-aminohydroxylation of various enolizable ketones. Flow conditions contribute to mitigating the high reactiv-
ity and toxicity of α-chloronitroso derivatives and provide an efficient, versatile and safe protocol for the
α-aminohydroxylation of ketones with a minimal footprint. Fundamental aspects of the α-aminohydroxylation
process were computed by DFT and further supported the experimental observations, hence leading to the
unprecedented α-chloronitroso-based α-aminohydroxylation of primary, secondary and tertiary substrates.
Recycling of the carbon backbone of the α-chloronitroso derivatives provides a high atom economy for the
preparation of value-added molecules. This work showcases α-chloronitroso derivatives as economic and
efficient vehicles for transferring electrophilic synthons of hydroxylamine toward nucleophilic enolates. A
representative range of precursors and analogs of pharmaceutical active ingredients, including WHO essentials
and drugs in shortage (such as epinephrine and ketamine), are prepared within minutes according to a fully
concatenated process. The process features sequential modules with distinct unit operations including chemi-
cal transformations and multiple in-line extractions. The process relies on an upstream chemical Generator
that manages the preparation of α-chloronitroso derivatives and that feeds downstream a series of
α-aminohydroxylation modules. The setup is amenable to the addition of libraries of compounds for feeding
upstream the process of discovery in medicinal chemistry and is transposable to pilot scale. Several layers of
in-line analytical procedures are featured to improve process control and safety.
Received 28th December 2020,
Accepted 1st February 2021
DOI: 10.1039/d0gc04395h
rsc.li/greenchem
Introduction
Nitrogen, which is the fourth element in abundance in bio-
genic material, is ubiquitous in biomolecules from peptides to
neurotransmitters. Nitrogen-containing organics are therefore
privileged scaffolds for active pharmaceutical ingredients
(APIs), among which β-amino alcohols and α-aminoketones
are abundantly represented (Fig. 1). The development of amin-
aCenter for Integrated Technology and Organic Synthesis, MolSys Research Unit,
University of Liège, B-4000 Liège (Sart Tilman), Belgium.
Fig. 1 Representative examples of β-amino alcohol and α-aminoketone
APIs and biologically active compounds. Pseudoephedrine and ephedrine
are listed as WHO essential drugs,20 and are mainly used as sympathomi-
metic sinus decongestant and stimulant, respectively. Epinephrine, also
known as adrenalin, is a neurotransmitter and a WHO essential drug cur-
rently in shortage.20–22 Salbutamol is a β-amino alcohol active pharma-
ceutical listed as essential by WHO used mainly for the treatment of
E-mail: jc.monbaliu@uliege.be; http://www.citos.uliege.be
bXStruct, Department of Chemistry, Ghent University, Krijgslaan 281-S3, B-9000
Ghent, Belgium
cInstitute of Condensed Matter and Nanosciences, Université catholique de Louvain,
Place Louis Pasteur 1, 1348 Louvain-la-Neuve, Belgium
†Electronic supplementary information (ESI) available: Fluidic components,
experimental procedures, details of the fluidic setup and procedures (micro- and
mesoscale), off-line analysis with structural assignments (NMR and X-ray diffrac-
tion) and computations. CCDC 2046130–2046136. For ESI and crystallographic
data in CIF or other electronic format see DOI: 10.1039/d0gc04395h
asthma.20 Propranolol, another WHO essential β-amino alcohol, is
a
common β-blocker medication for treating cardiovascular conditions.20
Ketamine and bupropion are two representative α-aminoketones prescribed
mainly for treating severe depression. Ketamine is listed as WHO essential.20
2336 | Green Chem., 2021, 23, 2336–2351
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
tially versatile intermediates for rapidly accessing various
β-amino alcohol and α-aminoketone scaffolds, which are com-
monly encountered in pharmaceuticals and other biologically
active molecules (Fig. 3a). In this context, Oppolzer’s seminal
work for electrophilic aminations of enolizable carbonyls with
α-chloronitroso compounds (Fig. 3b) is still a landmark.57–62
Oppolzer’s amination was also amenable to stereoselective pro-
tocols, either using chiral auxiliaries on the substrate or chiral
nitroso compounds.
Fig. 2 Illustration and detailed mechanism for the electrophilic amin-
ation of enolizable ketone substrates (also known as the nitroso aldol
reaction) with α-chloronitroso derivatives 1a,b showing the formation of
an intermediate nitrone nit-2 and its subsequent solvolysis toward a free
α-aminohydroxylated ketone 3.
ation procedures starting from common, widely available sub-
strates therefore stimulates lots of research efforts.1–4 Typical
amination procedures exploit the inherent nucleophilic nature
of nitrogen for its insertion on activated electrophilic
substrates.5–11 Alternative methods relying on the umpolung of
nitrogen, known as electrophilic aminations, are also abun-
dantly documented in the literature.12–19
A variety of electrophilic amination reagents have been
reported, among which azodicarboxylates,23–29 diazoesters,30
haloamines,30–32 hydroxylamines and derivatives,30,32–35
oxaziridines,11,36 as well as nitroso derivatives are documen-
ted (Fig. 2). The latter class counts several subclasses that
share a common electrophilic NvO moiety that is modulated
through specific structural features. Nitroso derivatives are
versatile chemicals that have been used for formal amination
reactions starting from various substrates mainly through
hetero Diels–Alder,37–44 ene40,45–48 and O- or N-nitroso aldol
reactions.48–51
Among nitroso derivatives, α-chloronitroso (also known as
1-chloro-1-nitroso or gem-chloronitroso) derivatives such as
1-chloro-1-nitrosocyclopentane 1a and 1-chloro-1-nitrosocyclo-
hexane 1b (Fig. 2) are underrepresented in the literature,
despite their inherent appealing reactivity profile. Alike other
nitroso derivatives, they can be involved in hetero Diels–
Alder52–56 and nitroso aldol reactions,57–64 yet unlike other
nitroso species, the carbon backbone is readily released
through solvolysis, hence avoiding tedious or wasteful de-
protection protocols to release the aminohydroxyl function
(Fig. 2). The mechanism, which is illustrated in Fig. 2 for a
Fig. 3 (a) α-Aminohydroxylated ketones obtained from the reaction of
enolizable ketones and α-chloronitroso 1a,b are versatile intermediates
for rapidly accessing various β-amino alcohol and α-aminoketone
nitroso-aldol reaction (i.e. the electrophilic amination of an scaffolds. (b) Oppolzer’s α-aminohydroxylation on chiral amides (N* =
(2R)- or (2S)-bornane-10,2-sultam) with 1b.57–62 (c) Large scale prepa-
ration of anti-inflammatory drug candidate BMS-561392 involving a key
electrophilic α-aminohydroxylation using 1a.65 (d) This work significantly
extends the scope of the α-aminohydroxylation with α-chloronitroso
derivatives such as 1a to primary, secondary and tertiary enolizable sub-
enolizable substrate), involves the formation of an intermedi-
ate nitrone nit-2, which solvolyzes in lower alcohols or in
acidic water. The carbon backbone is released as a ketal (solvo-
lysis in lower alcohols) or as a ketone (acidic hydrolysis). The
α-aminohydroxylated ketones obtained accordingly are poten- strates under scalable flow conditions.
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 2336–2351 | 2337

View Article Online
Paper
Green Chemistry
There is only one example in the recent literature that col was first validated on a series of model primary and sec-
illustrates a large-scale batch application relying on com- ondary enolizable ketone derivatives, and then broadened to
pound 1a (Fig. 3c). Compound 1a was used to install a qua- other enolizable ketones, including tertiary substrates. The
ternary center on a key intermediate toward anti-inflamma- protocol is versatile and also features options for controlling
tory compound BMS-561392 (Fig. 3c).65 This protocol was the α-aminohydroxylation of unsymmetrical enolizable
developed at Bristol-Myers Squibb and required labor- and ketones under thermodynamic and kinetic conditions. Last,
energy-intensive conditions. Besides, the authors indicated but not least, this protocol also enables the unprecedented
that “data substantiated that the aminating agent 1-chloro-1- α-aminohydroxylation of primary enolizable ketones, which is
nitrosocyclopentane was not safe to use”65 on such a prepara- very challenging to achieve under conventional batch con-
tive scale. The unstable and explosive nature of 1a imposed ditions. A DFT study indicated that the α-aminohydroxylation
−78 °C for the effective amination of a leucine-derived enoliz- step is diffusion-controlled and supported the experimental
able amide.65
observations. Besides the wide functional diversity obtained
With increasingly restrictive legislations both at the R&D through this method, the by-product of the reaction, i.e. the
and commercial scales, the use of reactive and potentially carbon backbone of the α-chloronitroso derivative (i.e. cyclo-
hazardous chemicals has become limiting if not impossible pentanone for 1-chloro-1-nitrosocyclopentane 1a) is recycl-
with conventional batch reactors. The concept of chemical able and readily transformed with hydroxylamine hydro-
generators under continuous flow conditions has been exten- chloride into the corresponding oxime for feeding Generator
sively documented by Kappe,66–72 hence providing a robust II, hence showcasing α-chloronitroso derivatives as economic
answer for exploiting reactive intermediates yet benefiting and efficient vehicles for transferring electrophilic synthons
from all the safety and process advantages of continuous flow of hydroxylamine toward nucleophilic enolates.
operation. A continuous flow chemical generator actually relies
on stable and widely available precursors and on the concate-
nation of upstream operations aiming at the preparation of
unstable, toxic or reactive chemicals, which are next immedi-
ately consumed within downstream operations (Fig. 3d). Such
Experimental section
General information
a scheme alleviates the safety concerns of handling and stock- Conversion and yield were determined by HPLC/DAD or by
1
piling reactive chemicals. Building upon our expertise in devel- high field H NMR spectroscopy. Structural identity was con-
oping continuous flow processes73–77 and the handling of firmed by 1H and 13C NMR spectroscopy (400 MHz Bruker
nitroso species,37,38,78,79 a solution was sought for exploiting, Avance spectrometer) in CDCl3, CD3OD or D2O and by X-ray
under safe conditions, the unique reactivity of α-chloronitroso diffraction on single crystals of compounds rac-3g·HCl, 3q,
derivatives at its fullest.
3r, 3t·HCl, 3u, 5, 6 and 7 (CCDC 2046130–2046136†).
We describe herein the development of a low footprint, Compounds 6 and 7 are described in the ESI only, see sec-
modular, safe and scalable continuous flow process for the tions 2.3, 2.5 and 2.6†. The chemical shifts are reported in
electrophilic α-aminohydroxylation of various enolizable ppm relative to TMS as an internal standard or to the solvent
ketones (Fig. 3d). Building upon the highly reactive nature of residual peak. Solvents (diethyl ether, absolute ethanol,
α-chloronitroso derivatives, this flow process revisits methanol, hexanes, petroleum ether (40–60) and ethyl
Oppolzer’s classical procedure for the α-aminohydroxylation acetate) were used as received, unless otherwise stated.
of enolizable ketones and is amenable to either R&D or pro- Tetrahydrofuran (THF) and methyl tert-butyl ether (MTBE)
duction scales. Flow conditions contribute to mitigate the were distilled over Na/benzophenone. Sodium hypochlorite
high reactivity and toxicity of α-chloronitroso derivatives. The pentahydrate, tert-butanol, glacial acetic acid, acetoxime,
flow process features two chemical generators connected in 2-butanone oxime, cyclopentanone oxime, cyclohexanone
series: (a) Generator I for the preparation of strong organic oxime, cycloheptanone oxime, (−)-menthone, (+)-camphor,
oxidizer tert-butyl hypochlorite (tBuOCl) from common and acetophenone, propiophenone, 4′-methylpropiophenone,
stable precursors; (b) and Generator II for the generation of 3′-chloro-propiophenone, 4′-chloro-propiophenone, 4′-fluoro-
α-chloronitroso derivatives from stable and widely available propiophenone, 4′-(trifluoromethyl)-propiophenone, 2′-(tri-
oximes and tBuOCl (from Generator I). Generator II then fluoromethyl)-propiophenone,
feeds series of fluidic modules for the phenone, 3′-nitropropiophenone, 4′-methoxy-propiophenone,
α-aminohydroxylation of enolates with α-chloronitroso deriva- valerophenone, octanophenone, 1-(2-thienyl)-1-propanone,
tives, producing a variety of racemic α-aminohydroxylated 1,3-diphenylpropan-1-one, 1-(benzo[d][1,3]-dioxol-5-yl)
3′-(trifluoromethyl)-propio-
a
ketone hydrochlorides in water. Each fluidic element also propan-1-one, isobutyrophenone, 1,2-diphenylethan-1-one,
includes additional operations such as extraction, quench, cyclobutyl phenyl ketone, cyclopentyl phenyl ketone, 2-chlor-
hydrolysis and in-line analysis, hence providing a complex, ophenyl cyclopentyl ketone, cyclohexyl phenyl ketone, 2-phe-
fully integrated process. Generators I and II alleviate the nylcyclohexanone, lithium bis(trimethylsilyl)amide (ca. 26%
storage and handling of two toxic and unstable chemicals, in THF), hydroxylamine hydrochloride, sodium acetate tri-
i.e. tBuOCl and α-chloronitroso derivatives, hence drastically hydrate, pyridine, zinc chloride (ca. 25% in THF), sodium
improving the safety features of the process. The flow proto- bicarbonate, potassium tert-butoxide, potassium hydroxide,
2338 | Green Chem., 2021, 23, 2336–2351
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
sodium borohydride and zinc dust, calcium chloride, potassium (BPR-1000). Downstream membrane separation relied on a
iodide, sodium thiosulfate, potassium bichromate were membrane separator from Zaiput Flow Technologies (SEP-200,
obtained from commercial sources and used as received.
equipped with a 1 µm pore hydrophobic membrane). In-line
analysis was optionally performed with an IR spectrometer
(FlowIR™ from Mettler-Toledo equipped with a DTGS detector
Experimental setup
Microfluidic setups. Microfluidic reactors consisted of using HappGenzel apodization, a Silicon probe connected via
modular continuous flow assemblies constructed from PFA a FlowIR™ sensor and a high pressure heated 50 µL cell) or a
tubing (1.58 mm outer diameter, 750 µm internal diameter) benchtop NMR (43 MHz Spinsolve™ 1H NMR spectrometer
equipped with PEEK/ETFE connectors and ferrules (IDEX/ from Magritek® equipped with a flow-through cell).
Upchurch Scientific). Feed and collection lines consisted of
Computational study
PFA or PEEK tubing (1.58 mm outer diameter, 750 μm internal
diameter) equipped with PEEK/ETFE connectors and ferrules Calculations were carried out using the Jaguar 8.5 program
(IDEX/Upchurch Scientific). Liquid feeds were handled with package.80 Geometry optimizations were performed at the
high force Chemyx Fusion 6000 syringe pumps equipped with B3LYP-D3/6-31+G* level of theory. Solvent effects were mod-
SS syringes and DuPont Kalrez O-rings or with HPLC pumps elled by using the polarizable continuum-Poisson method as
(ThalesNano micro HPLC). The coil reactors were thermoregu- incorporated in Jaguar, using the parameters for MTBE, i.e., a
lated with a Heidolph MR Hei-Tec equipped with a Pt-1000 dielectric constant of 2.6, and a solvent probe radius of 2.42 Å.
temperature sensor. Downstream pressure was regulated with Full calculation of vibrational frequencies was performed at
back pressure regulators from Zaiput Flow Technologies the B3LYP-D3/6-31+G* level using ultrafine grids to character-
(BPR-10) or from IDEX/Upchurch Scientific. Liquid–liquid ize the stationary points (local minima no imaginary frequen-
extraction was carried out with a Zaiput Flow Technologies cies and transition states one imaginary frequency) and to
membrane separator (SEP-10, equipped with a 1 µm pore obtain Gibbs free energy correction. The calculated reaction
hydrophobic membrane).
profile was verified for each transition state by following the
Mesofluidic setup – lab scale. Lab scale mesofluidic experi- intrinsic reaction coordinate (IRC). In Jaguar, the translational
ments were carried out in a Corning® Advanced-Flow™ Low partition function is computed for ideal gas standard con-
Flow Reactor (0.5 mL internal volume glass fluidic modules). ditions, corresponding to a pressure of 1 atm at 298.15 K. For
Feed and collection lines consisted of PFA tubing (1/8″ o.d.) solution reactions, the standard condition is instead 1 mol
equipped with PFA or SS Swagelok connectors and ferrules. L−1. Accordingly, the free energy value computed in Jaguar was
Liquid feeds were handled with FLOM or ThalesNano micro corrected by a concentration term, equal to RT ln(V_mol_gas_1
HPLC pumps. The reactor was maintained at reaction temp- atm/V_mol_1 M), i.e., 1.89 kcal mol−1 at 298.15 K. Electronic
erature with a LAUDA Integral XT 280 thermostat and a energies were computed at the B3LYP-D3/6-311+G** level
LAUDA Proline RP845. Downstream pressure was regulated whereas solvation energies were obtained at the B3LYP-D3/6-
with a back pressure regulator from Zaiput Flow Technologies 31+G*(MTBE). A systematic attempt to locate all possible local
(BPR-10 or BPR-1000). Downstream membrane separation minima (at the B3LYP-D3/6-31+G* level) was made, with the
relied on
a
membrane separator from Zaiput Flow data presented referring to the lowest energy form.
Technologies (SEP-10, equipped with a 1 µm pore hydro-
phobic membrane). In-line analysis was performed on an in-
Typical runs
line IR (FlowIR™ from Mettler-Toledo equipped with a DTGS CAUTION: Reactor setups involving tBuOCl, 1-chloro-1-nitroso-
detector using HappGenzel apodization, a Silicon probe con- cyclopentane (1a) and 1-chloro-1-nitrosocyclohexane (1b) were
nected via a FlowIR™ sensor and a high pressure heated covered in thin foil to prevent light exposure. tBuOCl, 1a and
1
10 µL cell) or a benchtop NMR (43 MHz Spinsolve™ H NMR 1b are reactive and unstable species with a pungent smell that
spectrometer from Magritek® equipped with a flow-through decompose under heat or UV irradiation. They must be
cell).
handled with great care under a fume hood in the absence of
Mesofluidic setup – pilot scale. Mesofluidic experiments direct light.81,82 If necessary, storage over CaCl2 at −8 °C in a
were carried out in a Corning® Advanced-Flow™ G1 Reactor brown glass bottle is recommended. Careful neutralization of
(8 mL internal volume glass fluidic modules for the prepa- tBuOCl over thioanisole83 and of 1a,b over triphenyl-
ration of tBuOCl and 1a,b) or in a Corning® Advanced-Flow™ phosphine84 is strongly advised before disposal.
G1 SiC Reactor (8 mL internal volume silicon carbide fluidic
Concatenated continuous flow preparation of 1-chloro-1-
modules for the preparation of rac-3b and rac-3k). Feed and nitrosocyclopentane 1a (lab scale). The feed solution of
collection lines consisted of PFA tubing (1/8″ or 1/4″ o.d.) sodium hypochlorite (1.5 M in water) and an aqueous solution
equipped with PFA or SS Swagelok connectors and ferrules. of acetic acid and tert-butanol (1 M, 1 : 1) were both injected at
Liquid feeds were handled with Corning® dosing lines (FUJI 0.1 mL min−1 and reacted in a PFA capillary coil for 5 min at
Technologies™ and HNP Mikrosysteme pumps). The reactor 25 °C. MTBE was injected downstream (0.2 mL min−1) and
was maintained at reaction temperature with a LAUDA Integral tBuOCl was extracted through a short column loaded with
XT 280 thermostat. Downstream pressure was regulated with a glass beads (ϕ = 0.1 mm). After liquid–liquid separation
back pressure regulator from Zaiput Flow Technologies through a hydrophobic membrane separator, the organic
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 2336–2351 | 2339

View Article Online
Paper
Green Chemistry
permeate was mixed with a solution of cyclopentanone oxime 1 : 1 MTBE/MeOH mixture were added dropwise at room temp-
4a (1 M in MTBE, 0.2 mL min−1) through a PEEK T-mixer and erature. The mixture was left under intense stirring for 1 h to
reacted in a PFA capillary coil (2 mL internal volume, 5 min complete neutralization (the neutralization rate was measured
residence time) at 25 °C. The reactor effluent was passed through HPLC/DAD by monitoring the formation of triphenyl-
through an in-line IR spectrometer for reaction monitoring phosphine oxide, see ESI, section 2.2.15†). Upon neutraliz-
and then quenched with a stream of aqueous sodium carbon- ation, the characteristic blue color of 1a slowly faded away.
ate (10 wt%, 0.2 mL min−1) under 5 bar of counterpressure.
Concatenated continuous flow α-aminohydroxylation of pro-
Downstream liquid–liquid separation through a hydrophobic piophenone 2b. A feed solution of sodium hypochlorite
membrane separator yielded 1-chloro-1-nitrosocyclopentane (1.5 M) and an aqueous mixture of acetic acid and tert-butanol
(1a) in MTBE (0.5 M). The aqueous retentate was redirected to (1 M, 1 : 1) were both injected at 0.1 mL min−1 though a PEEK
a waste container.
T-mixer and reacted in a PFA capillary coil for 5 min at 25 °C.
Concatenated continuous flow preparation of 1-chloro-1- MTBE was injected downstream the reactor (0.2 mL min−1) to
nitrosocyclopentane 1a (mesofluidic scale). An aqueous solu- extract tBuOCl through a short column loaded with glass
tion containing tert-butanol (1 M) and acetic acid (1 M) was beads (ϕ = 0.1 mm). The biphasic mixture was next processed
reacted with another aqueous solution of sodium hypochlorite through a membrane separator. The aqueous retentate was
(2.25 M) in a Corning® Advanced-Flow™ Low Flow Reactor redirected to a waste tank. The organic permeate (tBuOCl, 1 M
(4 fluidic modules connected in series, 2.0 mL total internal in MTBE) was redirected toward a PEEK T-mixer where it was
volume, operated at 25 °C). MTBE was subsequently injected mixed with a stream of cyclopentanone oxime (4a, 1 M in
into the system and two additional fluidic modules were used MTBE, 0.2 mL min−1). The resulting mixture was reacted in a
for the extraction (1 mL total internal volume, operated at PFA capillary coil (2 mL internal volume, 5 min residence
25 °C). The resulting biphasic stream was processed through a time) at 25 °C. The reactor effluent was connected to an in-line
hydrophobic membrane separator. The organic permeate IR spectrometer for reaction monitoring and then quenched
(tBuOCl, 1 M in MTBE) was next mixed with a solution con- with a stream of aqueous sodium carbonate (10 wt%, 0.2 mL
taining cyclopentanone oxime (4a, 1 M in MTBE) and reacted min−1). Downstream liquid–liquid separation through a hydro-
into a second reactor consisting of 3 fluidic modules con- phobic membrane separator yielded 1-chloro-1-nitrosocyclo-
nected in series (1.5 mL total internal volume, operated at pentane (1a) in MTBE (0.5 M). The organic permeate was
10 °C). The effluent was next passed through an in-line IR passed through a short column loaded with molecular sieve
spectrometer for real-time monitoring under 5 bar of counter- (MS
3 Å) prior to further downstream processing. The
pressure. All pumps were set to 2 mL min−1, affording an esti- upstream enolization reaction involved two pumps used to
mated residence time of 30 s for the preparation of tBuOCl deliver a solution of propiophenone (2b, 0.5 M in dry THF)
and of 22.5 s for the preparation of 1a.
and LiHMDS (0.5 M in dry THF). Both pumps were set to
Concatenated continuous flow preparation of 1-chloro-1- 0.4 mL min−1 and the resulting mixture was reacted in a PFA
nitrosocyclopentane 1a (pilot81 scale). An aqueous solution capillary coil (0.8 mL internal volume, 1 min residence time)
containing tert-butanol (1 M) and acetic acid (1 M) was mixed at 0 °C. The reactor effluent from the enolization step was
with an aqueous solution of sodium hypochlorite (2.25 M) in a mixed with the stream of 1a through a PEEK T-mixer and
Corning® Advanced-Flow™ G1 Reactor (2 fluidic modules con- reacted in a PFA capillary coil (1.2 mL internal volume, 1 min
nected in series, 16 mL total internal volume, operated at residence time) at 0 °C. The reactor effluent was next mixed
25 °C). MTBE was subsequently injected into the system and through a PEEK T-mixer with an aqueous solution of HCl
the extraction was carried out in an additional fluidic module (6 M) injected with a pump set at 0.4 mL min−1. The mixture
(8 mL internal volume, operated at 25 °C). The resulting bipha- was reacted in a PFA capillary coil (1.6 mL internal volume,
sic stream was processed into a membrane separator. The 1 min residence time) at 60 °C. The entire setup was operated
organic permeate (tBuOCl, 1 M in MTBE) was next mixed with under 5 bar of counterpressure. A final liquid–liquid mem-
a solution containing cyclopentanone oxime (4a, 1 M in brane separator was inserted downstream and the aqueous
MTBE) and reacted in an additional fluidic module (8 mL retentate was collected at steady state, diluted with MeOH, and
internal volume, operated at 10 °C). The effluent was next analyzed by HPLC/DAD (97% conv., 90% yield).
passed through an in-line IR spectrometer for real-time moni-
General
continuous
flow
protocol
for
the
toring under 5 bar of counterpressure. The reactor effluent was α-aminohydroxylation of enolizable ketones 2a-u. The pumps
neutralized in a collection tank containing a solution of tri- used to deliver ketones 2a-u (0.5 M in dry THF) and LiHMDS
phenylphosphine (1 M in a 1 : 1 MTBE/MeOH mixture). All (0.5 M in dry THF) were both set to 0.1 mL min−1. Both
pumps were set to 15 mL min−1, affording an estimated resi- streams were mixed through a PEEK T-mixer. The mixture was
dence time of 36 s for the preparation of tBuOCl and of 18 s reacted in a PFA capillary coil (0.2 mL internal volume, esti-
for the preparation of 1a.
Neutralization of
mated 1 min residence time) at 0 °C. The reactor effluent was
1a. next mixed through a PEEK T-mixer with 1-chloro-1-nitroso-
1-chloro-1-nitrosocyclopentane
CAUTION: the neutralization of 1a is exothermic. To 5 mL of a cyclopentane (1a, 0.5 M in MTBE, 0.11 mL min−1) and redir-
stirred solution of 1a (1 M, 5 mmol, 1 equiv.), 20 mL of a solu- ected toward a PFA capillary coil (0.31 mL internal volume,
tion of triphenylphosphine (0.25 M, 5 mmol, 1 equiv.) in a 1 min residence time) operated at 0 °C. The reactor effluent
2340 | Green Chem., 2021, 23, 2336–2351
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
was next mixed through a PEEK T-mixer with an aqueous solu- propiophenone 2b and 1-(3-chlorophenyl)propan-1-one 2k
tion of HCl (6 M) injected with a pump set to 0.3 mL min−1
(0.5 M in dry THF) and 1a (0.5 M in MTBE) were both set to
.
The mixture was reacted in PFA capillary coil (0.61 mL internal 30 mL min−1 and injected in a Corning® Advanced-Flow™ G1
volume, 1 min residence time) at 60 °C. The entire setup was SiC Reactor (6 fluidic modules connected in series, 48 mL
operated under 5 bar of counterpressure. A final liquid–liquid internal volume, 48 s residence time at 0 °C). The reactor
membrane separator was inserted downstream and the effluent was directly collected in aqueous HCl (1 M) under vig-
aqueous retentate was collected at steady state, diluted with orous stirring to hydrolyze the intermediate nitrone (nit-
MeOH, and analyzed by HPLC/DAD. Conversion and isolated 2b,2k). Conversion and isolated yield: starting from propiophe-
yield: starting from acetophenone 2a (3a, 70% conv., 62% none 2b (3b, 98% conv., 82% isolated yield) or 1-(3-chlorophe-
yield), propiophenone 2b (3b, 97% conv., 90% yield), 1-(p- nyl)propan-1-one 2k (3k, 88% conv., 73% isolated yield).
tolyl)propan-1-one 2c (3c, 96% conv. 88% yield), 1-(4-methoxy-
Semi batch preparation of norephedrine. The aqueous
phenyl)propan-1-one 2d (3d, 97% conv., 78% yield), 1-(benzo reactor effluent from the previous pilot scale experiment with
[d][1,3]dioxol-5-yl)propan-1-one 2e (3e, 80% conv., 74% yield), 2b was poured in a separatory funnel in the presence of
1-(4-chlorophenyl)propan-1-one 2f (3f, 97% conv., 95% yield), 250 mL of HCl (1 M) and extracted with Et2O (3 × 250 mL). The
1-(4-(trifluoromethyl)phenyl)propan-1-one 2g (3g, 79% conv., aqueous layer was concentrated under vacuum and the crude
73% yield), 1-(4-fluorophenyl)propan-1-one 2h (3h, 91% conv., residue was dissolved in 700 mL of MeOH. Subsequently,
86% yield), 1-(3-(trifluoromethyl)phenyl)propan-1-one 2i (3i, 40.7 g (1.1 mol, 2.2 equiv.) of NaBH4 were added portion wise
75% conv., 64% yield), 1-(3-nitrophenyl)propan-1-one 2j (3j, over 60 min and the mixture was stirred for 2 h at 0 °C. Then,
75% conv., 65% yield), 1-(3-chlorophenyl)propan-1-one 2k (3k, 3 M HCl (400 mL) were added to adjust the pH to 1. The
88% conv., 80% yield), 1-(2-(trifluoromethyl)phenyl)propan-1- solvent (MeOH) was then removed under reduced pressure
one 2l (3l, 77% conv., 49% yield), 1-(thiophen-2-yl)propan-1- and 700 mL of a 2 : 1 HCl/AcOH mixture were added to the
one 2m (3m, 93% conv., 60% yield), 1-phenylpentan-1-one 2n resulting solution. Afterwards, 162.5 g (2.5 mol, 5 equiv.) of
(3n, 83% conv., 79% yield), 1-phenyloctan-1-one 2o (3o, 76% zinc dust were added at 0 °C and the resulting suspension was
conv., 72% yield), 1,3-diphenylpropan-1-one 2p (3p, 89% vigorously stirred for 3 days at room temperature. The resulting
conv., 74% yield), 2-methyl-1-phenylpropan-1-one 2q (3q, 85% solid residue was removed by filtration and the pH of the fil-
conv., 82% yield), cyclobutyl(phenyl)methanone 2r (3r, 87% trate was adjusted to 14 using a 50 wt% sodium hydroxide
conv., 84% yield), cyclohexyl(phenyl)methanone 2s (3s, 78% aqueous solution. Brine (250 mL) was added and the crude
conv., 75% yield), cyclopentyl(phenyl)methanone 2t (3t, 82% reaction mixture was extracted with Et2O (5 × 500 mL) to give a
conv., 72% yield), (2-chlorophenyl)(cyclopentyl)methanone 2u colorless ethereal solution. The ethereal solution was dried
(3u, 98% conv., 80% yield).
over MgSO4, filtered and concentrated under reduced pressure
Continuous flow protocol for the α-aminohydroxylation of to give crude norephedrine as a white solid. The solid was
ketone 2v under thermodynamic or kinetic conditions. The then recrystallized in Et2O/hexane (1 : 2) to give 43 g (59%
pumps used to deliver 2-phenylcyclohexan-1-one (2v, 0.5 M in overall isolated yield) of pure norephedrine.
dry THF) and either LiHMDS (0.5 M in dry THF, kinetic con-
ditions) or potassium tert-butoxide (0.5 M in dry THF, thermo-
dynamic conditions) were both set to 0.1 mL min−1. Both
streams were mixed through a PEEK T-mixer and the resulting
Results and discussion
Continuous flow preparation of α-chloronitroso species
mixture was reacted in a PFA capillary coil (0.2 mL internal
volume, 1 min residence time) at either 0 °C (kinetic con- α-Chloronitroso species are typically prepared from a variety of
ditions) or 25 °C (thermodynamic conditions). The reactor substrates,85–89 although the corresponding oximes are typi-
effluent was next mixed through a PEEK T-mixer with 1-chloro- cally privileged substrates. The most represented protocols use
1-nitrosocyclopentane (1a, 0.5 M in MTBE, 0.11 mL min−1). chlorine-based oxidizers. Procedures based on chlorine gas
The resulting mixture was reacted in a PFA capillary coil were amongst the earliest reported.65,84,90–96 Given the hazards
(0.31 mL internal volume, 1 min residence time) at 0 °C. The related with the handling of Cl2 gas on a preparative scale, a
reactor effluent was next mixed through a PEEK T-mixer with procedure relying on its in situ generation from aqueous H2O2/
an aqueous solution of HCl (6 M) injected with a pump set at HCl was reported by Terent’ev for the preparation of various
0.3 mL min−1. The mixture was reacted in a PFA capillary coil α-chloronitroso compounds.97 Alternative methods based on
(0.61 mL internal volume, 1 min residence time) at 60 °C. The N-chloroderivatives such as N-tert-butyl-chlorocyanamide,98
entire setup was operated under 5 bar of counterpressure. A N,N′-dichloro bis(2,4,6-trichlorophenyl)urea,99 1-chloro-1H-
final liquid–liquid membrane separator was inserted down- benzo[d][1,2,3]triazole,79 or N-chlorosuccinimide,82 provided a
stream and the aqueous retentate was collected at steady state, low atom economy100 and
a high environmental factor
diluted with MeOH, and analyzed by HPLC/DAD (kinetic con- (E-factor).101 Hypochlorites have attracted attention for con-
ditions: 3v-1, 77% conv., 66% isolated yield; thermodynamic verting oximes into α-chloronitroso compounds since the early
conditions: 3v-2, 81% conv., 71% isolated yield).
1960s,102 and either inorganic103 or organic38,102 hypochlorites
Pilot scale α-aminohydroxylation of enolizable ketones 2b can be utilized. Among organic hypochlorites, tBuOCl (Fig. 4)
and 2k. The pumps used to deliver the enolates derived from emerged as a convenient and readily available reagent. The
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 2336–2351 | 2341

View Article Online
Paper
Green Chemistry
extraction with an organic solvent. Chlorinated solvents such
as dichloromethane and chloroform gave high extraction
efficiency and facilitated the in-line NMR monitoring of the
reaction, yet their environmental impact led us to consider
MTBE as a greener alternative. Compared to other ethereal sol-
vents, MTBE is considered as a safer option since it does not
form potentially explosive peroxides and does not react with
tBuOCl.65,104 A column packed with glass beads (φ = 0.1 mm)
was inserted before the membrane separator to enhance the
extraction efficiency, giving tBuOCl in 98% yield (ESI, Fig. S1
and Table S3†). The aqueous effluent from the membrane
separator was collected in a waste. The ability to produce
tBuOCl in a first generator and to directly use it downstream
for reaction with oximes enhances process safety, since it
avoids its isolation and storage. Adjustment of the flow of
MTBE enabled the preparation of a 1 M solution. During the
preliminary development stages of the flow process toward
tBuOCl, off-line redox titration was carried out to determine
its concentration after extraction (ESI, section 2.2.2†). After
optimization, off-line titration was conveniently replaced with
Fig. 4 (a) The method of Mintz and Walling81 for the preparation of
tBuOCl in batch. (b) Simplified microfluidic flowchart (Generator I) for
the preparation of tert-butyl hypochlorite (tBuOCl) showcasing the
extraction module. The column packed with glass beads upstream the
membrane separator is omitted for clarity (ESI for details, section 1.5.1†).
An in-line, low field NMR can be inserted downstream to monitor the
conversion of tBuOH into tBuOCl (see ESI section 2.2.3†).
1
in-line low field H NMR monitoring (ESI, section 2.2.3†). The
first procedures were reported using trichlorofluoromethane organic effluent can be collected over CaCl2 and stored at low
as solvent, a banned chemical responsible for the ozone layer temperatures, preferably in a brown glass bottle. Samples
depletion.38
could be maintained accordingly for several months, yet a
A typical procedure for the preparation of tBuOCl follows direct concatenation to downstream operations is highly desir-
an adaptation of the method of Mintz and Walling81 and con- able for mitigating chemical hazards.
sists of reacting sodium hypochlorite with tert-butanol in the
Since the cyclopentyl and cyclohexyl nitroso derivatives 1a,b
presence of acetic acid at 0 °C in the dark (Fig. 4a). are the most described α-chloronitroso reagents for electrophi-
Commercial bleach can advantageously replace aqueous solu- lic amination reactions in the literature,57–62,65 they were
tions of reagent grade sodium hypochlorite. The formation of selected as model compounds for the optimization of
tBuOCl is almost instantaneous, and it separates from the Generator II (Fig. 5b). The corresponding cyclopentanone and
aqueous reaction mixture. The absence of precipitates makes cyclohexanone oximes 4a,b, respectively, are readily available
the transposition under continuous flow conditions straight-
forward, yet it was never reported.
tBuOCl is an extremely potent electrophilic chlorinating
agent that offers one of the most convenient and atom-econ-
omic procedures for accessing α-chloronitroso species from
oximes. Unlike chlorine, its handling is facilitated by its liquid
nature, although its toxicity, photo- and thermal instability
raise significant safety concerns even for lab scale operation.
Handling of tert-butyl hypochlorite as a secondary oxidizer
thus justifies a flow protocol to maintain discrete amounts at
any time upon operation. The formation of tBuOCl is con-
veniently achieved from widely accessible commercial com-
pounds, including aqueous NaOCl, AcOH and tBuOH. Under
continuous flow, the formation of tBuOCl is completed (>99%
conversion) within 5 min of residence time with an optimal
NaOCl/tBuOH/AcOH ratio of 1.5 : 1 : 1 (Fig. 4b).
Generator I was constructed from PFA coils and HPLC-type
connectors. In a typical procedure, a feed solution of NaOCl in
Fig. 5 (a) Use of tBuOCl for the conversion of oximes 4a,b into the
water (1.5 M) and a feed solution of tBuOH (1 M) and AcOH corresponding α-chloronitroso species 1a,b. The advancement of the
(1 M) in water were injected at 0.1 mL min−1 through a static
T-mixer and reacted for 5 min at 25 °C in a PFA coil; a stream
reaction can be qualitatively monitored with the appearance of the
typical blue color of nitroso derivatives 1a,b. Typical reaction conditions
imply the slow addition of tBuOCl at room temperature over 30 min. (b)
Simplified microfluidic flowchart for Generator II for the production of
of MTBE was injected downstream to extract tBuOCl. Although
tBuOCl is poorly soluble in aqueous medium, its recovery
model α-chloronitroso species 1a,b showcasing the in-line IR and the
through membrane separation was drastically enhanced upon extraction module.
2342 | Green Chem., 2021, 23, 2336–2351
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
and can be ordered on the multikilogram scale (Fig. 5a). The each) embedded with a heat exchanger and connected in
next step thus consisted in the downstream use of tBuOCl for series to mimic the microfluidic configuration of Generators
the conversion of oximes 4a,b into 1a,b (Fig. 5b).
I–II (Fig. 4b and 5b). Upon minimal readjustment of the
One of the main challenges for optimizing the second gen- process conditions (ESI, section 2.2.13†), the process was then
erator relied on the appropriate selection of an off- or in-line transposed to pilot scale with a modular Corning® Advanced-
analytic protocol to monitor reaction optimization. The Flow™ G1 reactor (Fig. 6).
unstable nature of α-chloronitroso species precluded direct iso-
The transposition of Generator I to the pilot scale involved
lation and characterization via typical off-line analytic pro- 2 G1 fluidic modules (FMs) connected in series (8 mL internal
cedures such as NMR, GC and HPLC. We opted for non- volume each) and operated at 25 °C for the reaction of an
destructive in-line IR spectroscopy (ESI, section 2.2.18†), pro- aqueous solution containing tert-butanol (1 M) and acetic acid
viding a very convenient protocol to optimize reaction para- (1 M) with aqueous sodium hypochlorite (2.25 M) within 36 s
meters by monitoring the relative variations of the specific of residence time. The effluent from Generator I was mixed
vibration bands of the starting oxime (˜υOH = 3550–3600 cm−1
˜υCN = 1665–1720 cm−1) and of the corresponding nitroso formed in an additional FM (8 mL internal volume) operated
derivatives (˜υNO = 1539–1621 cm−1).
at 25 °C. The resulting biphasic stream was next processed
,
with a stream of MTBE and the extraction of tBuOCl was per-
The organic effluent coming from the upstream Generator through a pilot scale membrane separator (Zaiput Flow
I, consisting of a 1 M solution of tBuOCl in MTBE, was then Technologies liquid–liquid separator SEP-200) equipped with a
redirected to the upstream section of the Generator II (Fig. 5b), PTFE hydrophobic membrane (1 μm pore size). Qualitative
1
where it was mixed with a solution of oxime 4a (1 M in MTBE, monitoring was carried out through in-line low field H NMR
0.1 mL min−1) and reacted at 10 °C for 5 min in a PFA coil. (ESI, section 2.2.3†). The MTBE permeate (tBuOCl, 1 M) was
The reactor effluent was continuously extracted with an next redirected to Generator II, where it quantitatively reacted
aqueous 10 wt% Na2CO3 solution and separated through with oximes 4a,b into another FM (8 mL internal volume)
hydrophobic membrane technology. The final extraction was operated at 10 °C with a residence time of 18 s (Fig. 6).
paramount to remove traces of acetic acid, tBuOH and other
The outlet of Generator II was connected through a
by-products that would otherwise interfere with downstream sampling loop to an in-line IR spectrometer for real-time
operations. The aqueous stream was redirected to a waste and monitoring. The reactor effluent was collected in a tank con-
the organic permeate was collected. Residual water in the taining a solution of triphenylphosphine (1 M) for chemical
organic stream (1a in MTBE) was determined by Karl-Fisher destruction.84,105,106 Once the generation of nitroso was vali-
titration (ESI, section 2.2.19†). The titration results empha- dated on the pilot scale, the α-aminohydroxylation step was
sized that the membrane performed very well, with a residual considered.
water content of 19 ppm in the organic MTBE permeate;
further drying with MS 3 Å decreased the moisture content to
3
ppm. Downstream telescoping with the actual
α-aminohydroxylation would therefore require the addition of
a drying column filled with MS 3 Å on the permeate prior to
injection over enolates (see section “α-Aminohydroxylation on
model ketones 2a,b” below).
The conversion toward compound 1a was quantitative, and
the process was extended for the production of 1b and per-
formed likewise. The organic effluent can be collected over
CaCl2 and stored at −8 °C, preferably in a brown glass bottle.
Samples could be maintained accordingly up to 2 weeks, yet a
direct concatenation to downstream operations is highly desir-
able. In view of the preparation of rac-α-aminohydroxylated
ketones, the development of a reliable Generator II of nitroso
1a,b was considered as sufficient milestone, although
Generator II was also amenable for the preparation of a small
library of other α-chloronitroso derivatives (see ESI section
2.2.17† for details) upon minor temperature adjustments.
Fig. 6 Close-up of the pilot scale Generator II relying on an 8 mL
internal volume Corning® Advanced-Flow™ G1 reactor fluidic module
with integrated HEART™-shape static mixers (courtesy of Corning®).
The total residence time is 18 s, yet the characteristic blue color accom-
panying the formation of α-chloronitroso 1a,b appears nearly instan-
taneously after the injection of tBuOCl and oximes 4a,b. FM1: fluidic
module; H1,2: in/outlets for thermofluid; I1: inlet for tBuOCl (1 M MTBE,
feeding from backside); I2: inlet for oximes 4a,b (1 M MTBE, feeding
Assessment of scalability for Generators I–II
With the optimization of Generators I–II on the microfluidic
scale, we next envisioned its progressive scalability from lab to
pilot scales. The first step of the transposition relied on a
modular Corning® Advanced-Flow™ Low Flow reactor. The
mesofluidic setup consisted of glass fluidic modules (0.5 mL from backside); O1: outlet for α-chloronitroso 1a,b in MTBE.
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 2336–2351 | 2343

View Article Online
Paper
Green Chemistry
α-Aminohydroxylation on model ketones 2a,b
total reaction time). LC monitoring however indicated that the
reaction with cyclopentyl α-chloronitroso 1a was faster and
more selective than with 1b. This observation was further con-
firmed under microfluidic conditions (ESI, section 2.2.16†),
hence further trials were only conducted with 1a.
The ability to produce a reliable and steady stream of 1a,b
opens up new possibilities for downstream electrophilic amin-
ation setup. We shifted gears for revisiting Oppolzer’s57–62
electrophilic amination under continuous flow conditions.
Two simple enolizable ketones were selected as model sub-
strates: acetophenone 2a and propiophenone 2b. While the
electrophilic amination with ketone 2a was never reported in
the literature, 2b served as a privileged target in Oppolzer’s
work, since it is a fast track entry toward (pseudo)ephedrine
and related APIs. 2b would hence set boundaries and would
help us to calibrate our protocol.
The typical procedure in Oppolzer’s original papers relied
on the formation of a sodium enolate at low temperature, fol-
lowed by the electrophilic amination with 2b in the presence
of a Lewis acid, such as ZnCl2 or Zn(OTf)2. A direct transposi-
tion under microfluidic conditions systematically led to clog-
ging. However, computations on model compounds indicated
that the reaction in the presence and in the absence of ZnCl2
provides similar reaction profiles, hence indicating that Zn2+
catalysis is in fact not mandatory for this reaction (ESI, sec-
tions 2.2.7 and 4.1†). Accordingly, going back to batch prelimi-
nary optimizations (Fig. 7), lithium bis(trimethylsilyl)-amide
(LiHMDS) was identified as the best performing base for the
rapid and complete enolization of 2b in the presence of
nitroso compounds 1a,b (ESI, section 2.2.14†) and in the
absence of any Lewis acid at −78 °C in THF/MTBE (30 min
Complete conversion to nitrone nit-2b was reached after
90 min in batch. Nitrone nit-2b was next hydrolyzed with
aqueous HCl (1 M) for 12 h at room temperature, leading to
the desired α-aminohydroxylated ketone rac-3b (in 90% iso-
lated yield; Fig. 7a). From this preliminary optimization, it
came out that the slow addition of 1a at low temperature was
critical to avoid vigorous exothermic decomposition. With this
in hand, the α-aminohydroxylation was attempted on aceto-
phenone 2a. The conversion was complete, yet the product
obtained was not the expected α-aminohydroxylated ketone 3a,
but amidoxime 5 (Fig. 7a), the structure of which was con-
firmed by X-ray diffraction.107 We reasoned that the formation
of 5 would occur from two consecutive α-aminohydroxylations
on 2a. Further optimizations did not yield the desired
α-aminohydroxylated ketone compound (3a).
In order to identify the main parameters governing the reac-
tivity and the selectivity of the α-aminohydroxylation, compu-
tations (see ESI, section 4.2†) were undertaken starting from
the lithium enolates en-2a and en-2b (Fig. 7b). Experimentally,
a second amination is only observed in the case of 2a, leading
after hydrolysis to amidoxime 5. Our calculations predicted
that for both en-2a and en-2b, the electrophilic amination with
1a proceeds with no enthalpy barrier, meaning that it is extre-
mely fast and diffusion-controlled. The main hypothesis was
that the second electrophilic amination proceeds through
deprotonation of nit-2a,b by enolate en-2a,b to form en-nit-2a
and en-nit-2b, respectively. The latter would then add onto
another equivalent of 1a to yield bisnitrone adducts (bis-nit-2a
and bis-nit-2b) upon the elimination of chloride. These latter
would next undergo acidic hydrolysis to yield the corres-
ponding amidoximes, yet experimentally this only happens for
2a. Our computational study revealed that the acid–base equili-
brium toward en-nit-2a and en-nit-2b involves a low free
energy barrier (ΔG‡ = 3.7 and 9.4 kcal mol−1 for nit-2a and nit-
2b, respectively) and is highly displaced toward the formation
of en-nit-2a and en-nit-2b (ΔG = −22.8 and −19.7 kcal mol−1
for en-nit-2a and en-nit-2b, respectively). Subsequent addition
onto cyclopentyl α-chloronitroso 1a occurs with a free energy
barrier of 11.4 and 19.5 kcal mol−1 for en-nit-2a and en-nit-2b,
respectively. These results indicate that the selectivity for the
first amination vs. the second amination must be determined
by the competition between the addition onto 1a and the
deprotonation of nit-2a and nit-2b, respectively. In both
cases, for en-2a,b, the addition onto 1a (first amination) is
diffusion-controlled (vide supra). Concerning the deprotona-
Fig. 7 (a) Electrophilic amination of acetophenone 2a and propiophe- tion, for nit-2a, it involves a free energy barrier of 3.7 kcal
mol−1, meaning that the rate of this step (k ∼ 6.109 M−1 s−1) is
none 2b with cyclopentyl α-chloronitroso 1a under optimized con-
ditions. (b) Computations at the B3LYP-D3/6-311+G**(MTBE)//
B3LYP-D3/6-31+G*(MTBE) level of theory for the electrophilic amination
of the enolates of acetophenone (en-2a) and of propiophenone (en-2b),
respectively. Free energy barriers are indicated in orange (kcal mol−1).
Cp = cyclopentyl. See ESI (section 4.2†) for details.
also at, or very close to, the diffusion limit. In the case of nit-
2a, the free energy barrier is higher (9.4 kcal mol−1),
predicting a rate significantly lower than the diffusion limit
(k ∼ 2.105 M−1 s−1).
2344 | Green Chem., 2021, 23, 2336–2351
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
From our computational study, it became clear that in the
case of ketone 2a, the second amination is competitive, hence
leading to the bis-amination product 5 in the presence of a
local excess of enolate en-2a. Any inhomogeneity in the reac-
tion mixture (typically related to inefficient mixing in batch)
leads to poor control of the local stoichiometry, hence trigger-
ing side reactions. It is now well accepted that reactions sensi-
tive to local stoichiometry can be advantageously developed
under continuous flow conditions, often with a significant
reduction in side reactions.74,108
With this mechanistic understanding, we envisioned the
transposition of the electrophilic amination of model ketones
2a,b under microfluidic conditions. Three steps had to be
implemented according to a concatenated microfluidic setup
relying on three modules (Fig. 8a). The first module was dedi-
cated to the enolization of ketones 2a,b with LiHMDS. The
improved heat exchange efficiency under continuous flow con-
ditions is typically exploited to perform highly exothermic reac-
tions at higher temperatures than batch would normally
allow.108,109 Upon optimization, the enolization of ketone 2a in
module I required 1 min of residence at 0 °C with 1.1 equiv. of
LiHMDS in THF (0.5 M), rather than typically −78 °C in batch.
The effluent of module I was then directly mixed with a stream
of 1a (0.5 M in MTBE), in provenance from Generator II that
was operated continuously to feed the electrophilic amination.
Screening of the electrophilic amination conditions on
model ketones 2a,b emphasized that the electrophilic amin-
ation reaction could advantageously be performed at a much
higher temperature in flow (0 °C) than in batch (typically
−78 °C), while avoiding decomposition and reaching almost
quantitative conversion within 1 min of residence time
(Fig. 8b). Higher temperatures than 0 °C for shortening the
residence time even further are not advised in flow since two
runs at room temperature led to the explosive decomposition
of the reaction mixture.
Fig. 8 (a) Fully concatenated setup showing Modules I (enolization), II
(electrophilic amination, connected upstream to Generator II) and III
(hydrolysis) for the electrophilic amination of enolizable ketones 2a,b
with α-chloronitroso 1a. The organic permeate was further processed to
recover cyclopentanone, which could be converted back to cyclopenta-
none oxime 4a with 83% yield and could be used to feed Generator II.
(b) Screening of the reaction conditions for the electrophilic amination
of model substrate 2b with nitroso 1a under continuous flow conditions.
The number in the isovalue plot indicates the HPLC conversion of 2b. At
room temperature, exothermic decomposition of nitroso 1a occurred
leading to either clogging or rupture of module II.
The final hydrolysis of nitrone nit-2a,b was performed in
module III. The effluent from module II was immediately
mixed with aqueous HCl (6 M) at 0.3 mL min−1 and reacted in
a PFA coil. Increasing the temperature from room temperature
to 60 °C in module III led to a significant reduction in reaction
time, conversions up to 97% being attained within 1 min. The
conversion of ketones 2a,b was of 70 and 96%, respectively,
with a quantitative selectivity within 3 min of total residence
time toward the desired α-aminohydroxylated ketones 3a,b (as
their hydrochlorides), with no traces of the bisamination
product 5. The fast mixing and fine control on the local stoi- carefully engineered in order to avoid the contamination of
chiometry, as well as the effective concatenation helped sup- active pharmaceutical ingredient intermediates with poten-
pressing the occurrence of an amidoxime by-product 5. tially toxic nitroso compounds.
Besides, the electrophilic amination on primary enolizable
ketones was never reported with nitroso derivatives.
The insertion of a hydrophobic membrane separator, com-
bined with an upstream hydrolysis with hydrochloric acid, pro-
A final in-line membrane separation step was integrated to vided a simple yet elegant and robust solution to segregate the
module III to effect continuous separation between the nitroso-containing
organic permeate from the
aqueous stream containing essentially pure 3a,b as their α-aminohydroxylated ketone hydrochlorides in water. The
hydrochlorides and the organic effluent containing traces of organic permeate was treated with PPh3 (1 M in THF) to neutral-
unreacted nitroso 1a and any other organic by-products (ESI, ize any leftover of 1a. The aqueous effluent was collected, and
section 2.2.20†). The design of the downstream extraction was processed to recover 3a,b as free bases (ESI, section 2.2.7†). The
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 2336–2351 | 2345

View Article Online
Paper
Green Chemistry
aspects of recovering and reusing the nitroso carbon backbone
from the organic permeate were also considered. The permeate
was further processed to recover cyclopentanone, that could be
converted back to cyclopentanone oxime 4a with 83% isolated
yield (ESI, section 2.2.9†). The recovery of the carbon backbone
of nitroso 1a significantly reduces the overall footprint of the
process, since the cyclopentyl backbone is only borrowed as a
vehicle for transferring a hydroxylamine electrophilic synthon
to the enolizable ketone substrate (Fig. 8a).
Scope of the α-aminohydroxylation with nitroso 1a
With a fully functional continuous flow system including a
chemical generator for nitroso species 1a and a modular setup
for the electrophilic amination of model enolizable ketones,
we aimed at pushing further the horizon of the microfluidic
setup. An ideal scenario would be to keep the same setup
while only changing the feed solution of enolizable ketone
substrates and assess its potential for the edition of discrete
libraries of α-aminohydroxylated ketones. Following this idea,
the setup was left unchanged and the reactor was fed with
various enolizable ketones (Fig. 9 and 10), preferably bearing
an aromatic ring to facilitate off-line LC monitoring.
We began exploring the scope of this protocol with deriva-
tives of propiophenone 2b-l, from which several biologically
active molecules can be accessed through this procedure,
including norephedrine (from 3b), cathinone (from 3b),
mephedrone (from 3c) and bupropion (from 3k).
Feed solutions of ketones 2b-l were prepared in dry THF
(0.5 M) and successively subjected to electrophilic amination
with nitroso 1a, giving the corresponding hydrochlorides of
α-aminohydroxylated ketones rac-3b-l. Neutralization, followed
by extraction in MTBE and evaporation gave rac-3b-l free bases
in 49–95% isolated yield (for some examples, the hydro-
chloride was isolated and analyzed, see ESI, sections 2.3, 2.5
and 2.6†). Analytical samples were obtained through either
recrystallization or after column chromatography on silica gel.
The lowest conversion and isolated yield concerned rac-3l that
features an ortho-CF3 group, which might sterically hamper
the reaction. For all other propiophenone-related substrates,
EWG-substituted derivatives came with a slightly lower conver-
sion, which most likely relates to a lower nucleophilicity of the
corresponding enolates. The difference with EDG-substituted
propiophenone derivatives is, however, difficult to exploit
according to a Hammett relationship.
The conditions are also amenable to heteroaromatic ketone
2m (93% conv.) and sterically more demanding 2n (83% conv.)
and 2o (76% conv.). Besides secondary ketones (2b-p), the elec-
trophilic amination under flow conditions was also success-
fully applied to a variety of tertiary enolizable ketones, hence
providing quaternary α-aminohydroxylated ketones 3q-u in
good to high isolated yields (72–84%). Unexpected oxidation
products were obtained for two additional substrates (diethyl
malonate and 2-phenylacetophenone) that are only described
in the ESI (sections 2.3, 2.5 and 2.6†).
Fig. 9 Scope of the electrophilic amination on primary, secondary and
tertiary ketones. Conversion and isolated yield (in parentheses) are indi-
cated. All compounds were obtained according to the setup described
in Fig. 8a. See also Fig. 10.
Our interest toward ketamine and analogs75 prompted us to
consider (rac)-2-phenylcyclohexanone (2v) as a substrate. In
2346 | Green Chem., 2021, 23, 2336–2351
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
was simplified to spare resources, and we focused the efforts on
the most critical steps (module 2). The hydrolysis of the nitrone
was carried out in a collection tank. Two pharmaceutically rele-
vant intermediates were selected as demonstrators, namely,
2-(hydroxyamino)-1-phenylpropan-1-one (rac-3b) and 1-(3-chloro-
phenyl)-2-(hydroxyamino)propan-1-one (rac-3k), which are
advanced intermediates for the preparation of norephedrine and
bupropion, respectively.
The process was then transposed to pilot scale with a
Corning® Advanced-Flow™ G1 SiC Reactor (ESI sections 1.5.8
and 2.2.21†). Silicon carbide was selected here rather than
glass to avoid any corrosion issue with LiHMDS. The setup for
the pilot scale α-aminohydroxylation involved several G1 SiC
FMs connected in series (about 48 mL total internal volume)
and operated at 0 °C. To simplify the setup, preformed solu-
tions of the enolates (0.5 M) derived from 2b (en-2b) or 2k (en-
2k) were prepared by mixing LiHMDS with the enolizable sub-
strate and maintaining the feed solutions at 0 °C. The reactor
effluent from pilot scale Generator II was dried with MS 3 Å
and was reacted with enolates en-2b,k. With a total flow rate of
60 mL min−1, a shorter residence time than for the microflui-
dic scale was obtained (48 s). The reactor effluent was collected
directly in aqueous HCl (1 M) to perform the hydrolysis of the
intermediate nitrones nit-2b,k toward the corresponding
α-aminohydroxylated ketones rac-3b and rac-3k in the collec-
tion flask. Samples were collected at steady state and hydro-
lyzed prior to LC analysis. Daily productivities of 3.49 kg (98%
conv., 82% isolated yield) and of 3.19 kg (88% conv., 73% iso-
lated yield) were achieved for rac-3b and rac-3k, respectively.
Both compounds were further reduced in batch for the prepa-
ration of norephedrine and analogs.
Fig. 10 (a) Adaptation of the microfluidic setup to 2-phenylcyclohexa-
none (2v) for the electrophilic amination under kinetic or thermo-
dynamic conditions. (b) Details of the area of interest for the 13C APT
spectra of the kinetic and thermodynamic α-aminohydroxylated ketones
3v-1,2 compared to the starting ketone 2v.
contrast to the other substrates of the study, 2v bears two
potentially enolizable positions, hence providing either a
direct precursor of norketamine and ketamine (3v-2) or an
analog (3v-1, Fig. 10). The selectivity between the two enoliz-
able sites is a classical control of kinetic vs. thermodynamic
conditions for the generation of the corresponding enolates
kin-en-2v and therm-en-2v. Module I of the microfluidic plat-
form described in Fig. 8a was adapted accordingly (Fig. 10):
the formation of kin-en-2v was obtained as previously, within
1 min of residence time at 0 °C with LiHMDS (0.5 M in THF),
while the formation of therm-en-2v was performed at room
temperature, within 1 min of residence time with tBuOK
(0.5 M in THF). The rest of the setup was left unchanged.
Under kinetic conditions for the formation of the enolate,
the electrophilic amination occurred selectively at the least
substituted position, thus leading to the formation of the
corresponding secondary α-aminohydroxylated ketone 3v-1 in
66% isolated yield (77% conv.). With module I operated at
room temperature for the enolization of 2v with tBuOK, the
electrophilic amination occurred selectively at the thermo-
dynamic position, yielding α-aminohydroxylated ketone 3v-2 in
71% isolated yield (81% conv.).
Discussion on the overall footprint of the electrophilic amination
Besides the unquestionable safety assets of such an integrated
platform that concatenates two upstream continuous flow gen-
erators of highly reactive species (tBuOCl and nitroso 1a), its
versatility was also demonstrated through the generation of
libraries of structurally diverse α-aminohydroxylated ketones.
On the microfluidic scale, the electrophilic amination platform
could be deployed for feeding upstream the process of discovery
of new hits/leads in medicinal chemistry, or at least accelerating
the preparation of α-aminoketone and β-amino alcohol
scaffolds. Besides operation at the lab scale, we have also
demonstrated the scalability of each step, opening up perspec-
tives on the production scale with low footprint reactors. The
fully concatenated system, including Generators I–II and the
3 modules for the electrophilic amination step was successfully
transposed to pilot scale in a commercial mesofluidic reactor
with a minimal footprint (L × W × H 94 × 42 × 40 cm), excluding
the auxiliaries, i.e. pumps, thermostats and in/off-line analytical
devices. The assets of flow chemistry were clearly demonstrated
Assessment of scalability for the electrophilic amination
With a wide chemical compatibility and the potential for rapidly in terms of high heat exchange efficiency (with reactions per-
accessing libraries of structurally diverse α-aminohydroxylated formed at much higher temperatures than conventional batch
ketones, this method has proven its added-value for R&D appli- processes) and in terms of mixing efficiency and control over
cations. In the next stage of the project, we wanted to assess the the local stoichiometry (with the suppression of a bisamination
scalability of the critical electrophilic amination step. The setup by-product for a model primary ketone).
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 2336–2351 | 2347

View Article Online
Paper
Green Chemistry
Besides process performance, the chemistry developed is (<0.1 for oil refining; <1–5 for bulk chemicals; 5–50 for fine
also advantageous according to its calculated environmental chemicals and 25–100 for pharmaceuticals). Since the context
factor (E-factor) (ESI, section 3.1†). The E-factor was calculated of this research is at the interface between fine chemicals and
according to the equation [E-Factor = (mass of total inputs) − pharmaceuticals, an E-factor comprised between 5–100 would
(mass of valuable products)/mass of valuable products] and be acceptable.
following the recommendations of Andraos.110 The two main
From the analysis of the E-factors presented in Table 1, it
elements of the process, namely the preparation of nitroso 1a can be seen that our electrophilic amination procedure is very
and the electrophilic amination on acetophenone 2b as a favorable (entry 1). It outclasses previous protocols for the
model substrate, were compared with similar strategies preparation of nitroso 1a with an E-factor = 6. It is also worth
reported in the literature, and their respective E-factors were mentioning that the carbon backbone of 1a is recyclable (see
calculated and compared (see Table 1). As a brief reminder, above). For the electrophilic amination, one can see from
E-factor values inherently relate to the nature of the process Table 1 that a calculated E-factor of 11 was obtained for this
protocol (entry 5), which is very competitive given the complex-
ity of the sequence, despite the fact that the protocol produces
racemic α-aminohydroxylated ketones (enantioselective
Table 1 Calculated environmental factors (E-factors) for the prepa-
version: see entry 8, E-factor = 26).
ration of 1a and the subsequent electrophilic amination step
A more refined analysis was next performed using the semi-
quantitative EcoScale from EcoSynth.114,115 The EcoScale con-
siders various parameters with a direct impact on the output
of a reaction (i.e. yield, costs of chemicals, safety, technical
setup, energy requirements, as well as work up & purification).
Generator I is associated with an honourable score of 81% on
the EcoScale, while Generator II is associated with a score of
76% (ESI, section 3.2†). The electrophilic amination on 2b as
described in this manuscript came with a score of 60% (the
calculation does not consider the recycling of the carbon back-
bone of nitroso 1a that ultimately makes the process even
more appealing), which constitutes a significant improvement
compared to the score of 43% obtained for the adaptation of
Oppolzer’s original amination in batch.
Starting
Entry material
Reagents and
solvents
Product
E-factor
Preparation of α-chloronitroso 1a
1
tBuOCl MTBE
6a
2
34% aq. HCl
37% aq. H2O2
CH2Cl2
32 (ref. 97)
3
4
11 (ref. 79)
12 (ref. 99)
Electrophilic amination on propiophenone 2b and derivatives
11a
Conclusion
5
This work discloses a unique, scalable and efficient continu-
ous flow strategy for the preparation of libraries of
α-aminohydroxylated ketones in good to high isolated yields
from the corresponding enolizable ketones and reactive
α-chloronitroso 1a. A library of 22 α-aminohydroxylated
ketones was prepared starting from primary, secondary and
tertiary ketones, hence opening the way to several intermedi-
ates for pharmaceutical production (including WHO essen-
tials and drugs on shortage). The development of a fully con-
catenated setup, featuring two upstream generators, enables
to exploit the unique reactivity profile of α-chloronitroso
derivatives to its fullest potential while mitigating chemical
risk. The main assets of continuous flow processing, includ-
ing high heat exchange and mixing efficiencies, the pro-
duction of discrete amounts of toxic and potentially explosive
compounds as well as the integration of in-line analytics were
very supportive for the design of a modular, safe and efficient
integrated flow protocol. The careful selection of reaction
conditions and chemicals, as well as the design of simple yet
robust downstream purification and recycling protocols con-
tributed to a significant reduction in overall footprint.
Beyond the development of robust conditions in conjunction
6
7
8
18 (ref. 111)
10 (ref. 112)
26 (ref. 59)
9
34 (ref. 113)
a This work.
2348 | Green Chem., 2021, 23, 2336–2351
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
with favorable metrics, this work also discloses a versatile
platform amenable to either discovery on lab scale or to pro-
duction at commercial scale. While maintaining a compact
and highly integrated reactor setup design (<2 m2 including
the auxiliaries) for the final demonstrator, a productivity
of up to 3.49 kg per day of high value-added
α-aminohydroxylated ketone intermediates was achieved.
Work is now undergoing for the development of recyclable
chiral α-chloronitroso derivatives, which would not only serve
as vehicles for NH2OH, but also for transferring stereochemi-
cal information to the ketone substrates.
10 R. Dorel, C. P. Grugel and A. M. Haydl, Angew. Chem., Int.
Ed., 2019, 58, 17118–17129.
11 Y. Park, Y. Kim and S. Chang, Chem. Rev., 2017, 117,
9247–9301.
12 C. Greck, B. Drouillat and C. Thomassigny, Eur. J. Org.
Chem., 2004, 1377–1385.
13 A. M. R. Smith and K. K. Hii, Chem. Rev., 2011, 111, 1637–
1656.
14 X. Yan, X. Yang and C. Xi, Catal. Sci. Technol., 2014, 4,
4169–4177.
15 K. S. Williamson, D. J. Michaelis and T. P. Yoon, Chem.
Rev., 2014, 114, 8016–8036.
16 M. Corpet and C. Gosmini, Synthesis, 2014, 46, 2258–2271.
17 T. J. Barker and E. R. Jarvo, Synthesis, 2011, 24, 3954–3964.
18 X. Dong, Q. Liu, Y. Dong and H. Liu, Chem. – Eur. J., 2017,
23, 2481–2511.
Conflicts of interest
There are no conflicts to declare.
19 Z. Zhou and L. Kürti, Synlett, 2019, 30, 1525–1535.
20 WHO Model Lists of Essential Medicines, https://www.
who.int/medicines/publications/essentialmedicines/en/,
(Accessed on December 28, 2020).
21 Drug shortages, https://www.fda.gov/drugs/drug-safety-
and-availability/drug-shortages, (Accessed on December
28, 2020).
Acknowledgements
This work was supported by the University of Liège (Welcome
Grant WG-13/03, JCMM) and by the F.R.S.-FNRS (Incentive
grant for scientific research MIS F453020F, JCMM).
Computational resources were provided by the “Consortium
des Équipements de Calcul Intensif” (CÉCI), funded by the
“Fonds de la Recherche Scientifique de Belgique” (F.R.S.-FNRS)
under Grant No. 2.5020.11 and by the Walloon Region. VEK
and TT are FRIA PhD fellows. KVH thanks the Research
Foundation – Flanders (FWO) (projects AUGE/11/029 and
G099319N) for funding. RR is a Maître de recherche of the F.R.
S.-FNRS. The authors thank Florence Houdouin and Geofrey
Vandendeyck for their experimental support. Dr Guillaume
Gauron and Alessandra Vizza (Corning SAS) are acknowledged
for the loan of the AFR reactors and the logistic, technical and
scientific support for the scalability trials.
22 Shortages
catalogue,
https://www.ema.europa.eu/en/
human-regulatory/post-authorisation/availability-medicines/
shortages-catalogue, (Accessed on December 21, 2020).
23 X. Yang and F. D. Toste, J. Am. Chem. Soc., 2015, 137,
3205–3208.
24 G. A. Shevchenko, G. Pupo and B. List, Synlett, 2015, 26,
1413–1416.
25 H. M. Nelson, J. S. Patel, H. P. Shunatona and F. D. Toste,
Chem. Sci., 2015, 6, 170–173.
26 X. Fu, H. Y. Bai, G. D. Zhu, Y. Huang and S. Y. Zhang, Org.
Lett., 2018, 20, 3469–3472.
27 M. K. Reddy, I. Ramakrishna and M. Baidya, Org. Lett.,
2018, 20, 4610–4613.
28 A. Yanagisawa, Y. Yamashita, C. Uchiyama, R. Nakano,
M. Horiguchi and K. Ida, Synlett, 2019, 30, 738–742.
29 B. M. Trost, J. S. Tracy and E. Y. Lin, ACS Catal., 2019, 9,
11082–11087.
Notes and references
1 S. Mirabella, F. Cardona and A. Goti, Org. Biomol. Chem.,
2016, 14, 5186–5204.
2 M. Grom, G. Stavber, P. Drnovšek and B. Likozar, Chem.
Eng. J., 2016, 283, 703–716.
30 T. Baumann, M. Bachle and S. Brase, Org. Lett., 2006, 8,
3797–3800.
3 B. Robnik, B. Likozar, B. Wang, T. S. Ljubin and Z. Časar,
Pharmaceutics, 2019, 11, 1–21.
31 T. Miura, M. Morimoto and M. Murakami, Org. Lett.,
2012, 14, 5214–5217.
4 M. Trampuž, G. Stavber and B. Likozar, Chem. Eng. J.,
2019, 374, 924–936.
32 W. Yang, J. Sun, X. Xu, Q. Zhang and Q. Liu, Chem.
Commun., 2014, 50, 4420–4422.
5 T. Nobuta, G. Xiao, D. Ghislieri, K. Gilmore and
P. H. Seeberger, Chem. Commun., 2015, 51, 15133–15136.
6 Y. Park, Y. Kim and S. Chang, Chem. Rev., 2017, 117,
9247–9301.
33 Y. Shen and G. K. Friestad, J. Org. Chem., 2002, 67, 6236–
6239.
34 B. Zhou, J. Du, Y. Yang, H. Feng and Y. Li, Org. Lett., 2014,
16, 592–595.
7 R. Morodo, R. Gérardy, G. Petit and J.-C. M. Monbaliu,
Green Chem., 2019, 21, 4422–4433.
35 X. Yan, X. Yang and C. Xi, Catal. Sci. Technol., 2014, 4,
4169–4177.
8 A. de la Torre, V. Tona and N. Maulide, Angew. Chem., Int.
Ed., 2017, 56, 12416–12423.
9 O. I. Afanasyev, E. Kuchuk, D. L. Usanov and D. Chusov,
Chem. Rev., 2019, 119, 11857–11911.
36 C. W. Kang, M. P. Sarnowski, Y. M. Elbatrawi and J. R. Del
Valle, J. Org. Chem., 2017, 82, 1833–1841.
37 J. C. Monbaliu and J. Marchand-Brynaert, Tetrahedron
Lett., 2008, 49, 1839–1842.
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 2336–2351 | 2349

View Article Online
Paper
Green Chemistry
38 J. C. Monbaliu, B. Tinant and J. Marchand-Brynaert,
J. Org. Chem., 2010, 75, 5478–5486.
39 B. S. Bodnar and M. J. Miller, Angew. Chem., Int. Ed., 2011,
50, 5630–5647.
40 T. A. Wencewicz, B. Yang, J. R. Rudloff, A. G. Oliver and
M. J. Miller, J. Med. Chem., 2011, 54, 6843–6858.
41 S. Carosso and M. J. Miller, Org. Biomol. Chem., 2014, 12,
7445–7468.
42 B. Maji and H. Yamamoto, J. Am. Chem. Soc., 2015, 137,
15957–15963.
43 L. Brulíková, A. Harrison, M. J. Miller and J. Hlaváč,
Beilstein J. Org. Chem., 2016, 12, 1949–1980.
44 M. G. Memeo and P. Quadrelli, Chem. Rev., 2017, 117,
2108–2200.
45 W. Adam and O. Krebs, Chem. Rev., 2003, 103, 4131–4146.
46 M. Baidya and H. Yamamoto, Synthesis, 2013, 45, 1931–
1938.
47 L. Zhai, X. Tian, C. Wang, Q. Cui, W. Li, S. H. Huang,
Z. X. Yu and R. Hong, Angew. Chem., Int. Ed., 2017, 56,
11599–11603.
J. P. Adams, I. T. Forbes, N. J. Press and M. J. Thompson,
J. Chem. Soc., Perkin Trans. 1, 2002, 2, 1494–1514.
65 N. A. Magnus, S. Campagna, P. N. Confalone, S. Savage,
D. J. Meloni, R. E. Waltermire, R. G. Wethman and
M. Yates, Org. Process Res. Dev., 2010, 14, 159–167.
66 B. Gutmann, D. Cantillo and C. O. Kappe, Angew. Chem.,
Int. Ed., 2015, 54, 6688–6728.
67 F. J. Strauss, D. Cantillo, J. Guerra and C. O. Kappe, React.
Chem. Eng., 2016, 1, 472–476.
68 B. Gutmann and C. O. Kappe, J. Flow Chem., 2017, 7, 65–71.
69 M. Köckinger, C. A. Hone and C. O. Kappe, Org. Lett.,
2019, 21, 5326–5330.
70 A. Steiner, J. D. Williams, O. De Frutos, J. A. Rincón,
C. Mateos and C. O. Kappe, Green Chem., 2020, 22, 448–
454.
71 A. Steiner, P. M. C. Roth, F. J. Strauss, G. Gauron,
G. Tekautz, M. Winter, J. D. Williams and C. O. Kappe,
Org. Process Res. Dev., 2020, 24, 2208–2216.
72 D. Dallinger, B. Gutmann and C. O. Kappe, Acc. Chem.
Res., 2020, 53, 1330–1341.
48 M. Baidya, K. A. Griffin and H. Yamamoto, J. Am. Chem.
Soc., 2012, 134, 18566–18569.
73 R. Gérardy, M. Winter, A. Vizza and J.-C. M. Monbaliu,
React. Chem. Eng., 2017, 2, 149–158.
49 K. Ohmatsu, Y. Ando, T. Nakashima and T. Ooi, Chem,
2016, 1, 802–810.
50 P. Merino, T. Tejero, I. Delso and R. Matute, Synthesis,
2016, 48, 653–676.
51 I. Ramakrishna, V. Bhajammanavar, S. Mallik and
M. Baidya, Org. Lett., 2017, 19, 516–519.
52 Y. A. Arbuzov and A. Markovskaya, Bull. Acad. Sci. USSR,
Div. Chem. Sci., 1952, 1, 355–358.
53 N. J. Leonard, A. J. Playtis, F. Skoog and R. Y. Schmitz,
J. Am. Chem. Soc., 1971, 93, 3056–3058.
54 O. Werbitzky, K. Klier and H. Felber, Liebigs Ann. Chem.,
1990, 267–270.
55 S. Ranganathan and K. S. George, Tetrahedron, 1997, 53,
3347–3362.
56 A. Hall, P. D. Bailey, D. C. Rees, G. M. Rosair and
R. H. Wightman, J. Chem. Soc., Perkin Trans. 1, 2000, 329–
343.
74 R. Gérardy, N. Emmanuel, T. Toupy, V. Kassin,
N. N. Tshibalonza, M. Schmitz and J.-C. M. Monbaliu,
Eur. J. Org. Chem., 2018, 2301–2351.
75 V. E. H. Kassin, R. Gérardy, T. Toupy, D. Collin,
E. Salvadeo, F. Toussaint, K. Van Hecke and J.-C.
M. Monbaliu, Green Chem., 2019, 21, 2952–2966.
76 N. Emmanuel, P. Bianchi, J. Legros and J.-C.
M. Monbaliu, Green Chem., 2020, 22, 4105–4115.
77 V. E. H. Kassin, T. Toupy, G. Petit, P. Bianchi, E. Salvadeo
and J.-C. M. Monbaliu, J. Flow Chem., 2020, 10, 167–
179.
78 J.-C. M. Monbaliu, A. Cukalovic, J. Marchand-Brynaert
and C. V. Stevens, Tetrahedron Lett., 2010, 51, 5830–5833.
79 J.-C. M. Monbaliu, L. K. Beagle, J. Kovacs, M. Zeller,
C. V. Stevens and A. R. Katritzky, RSC Adv., 2012, 2, 8941–
8945.
80 Jaguar, version 8.5, Schrodinger, Inc., New York, 2014.
81 M. J. Mintz and C. Walling, Org. Synth. Coll., 1973, 5, 183–
187.
57 W. Oppolzer and O. Tamura, Tetrahedron Lett., 1990, 31,
991–994.
58 W. Oppolzer, O. Tamura and J. Deerberg, Helv. Chim. Acta,
1992, 75, 1965–1978.
82 V. Dhayalan and P. Knochel, Synthesis, 2015, 47, 3246–
3256.
59 W. Oppolzer, O. Tamura, G. Sundarababu and M. Signer,
J. Am. Chem. Soc., 1992, 114, 5900–5902.
83 C. R. Johnson and J. J. Rigan, J. Am. Chem. Soc., 1969, 91,
5398–5399.
60 W. Oppolzer and E. Merifield, Helv. Chim. Acta, 1993, 76,
957–962.
84 I. Sakai, N. Kawabe and M. Ohno, Bull. Chem. Soc. Jpn.,
1979, 52, 3381–3383.
61 W. Oppolzer, P. Cintas-Moreno, O. Tamura and
F. Cardinaux, Helv. Chim. Acta, 1993, 76, 187–196.
62 W. Oppolzer, C. G. Jochet and E. Merifield, Tetrahedron
Lett., 1994, 35, 7015–7018.
63 E. M. Stocking, R. A. Martinez, L. A. Silks, J. F. Sanz-
Cervera and R. M. Williams, J. Am. Chem. Soc., 2001, 123,
3391–3392.
85 M. Pape, Fortschr. Chem. Forsch., 1967, 7, 559.
86 E. G. Bozzi, C.-Y. Shiue and L. B. Clapp, J. Org. Chem.,
1973, 38, 56–59.
87 M. Kugelman, K. Mallams and F. Vernay, J. Chem. Soc.,
Perkin Trans. 1, 1976, 1113–1126.
88 R. Bou-Moreno, S. Luengo-Arratta, V. Pons and
W. B. Motherwell, Can. J. Chem., 2013, 91, 6–12.
89 R. Bou-Moreno, S. Luengo-Arratta and W. B. Motherwell,
Tetrahedron Lett., 2011, 52, 2097–2099.
64 E. C. Davison, M. E. Fox, A. B. Holmes, S. D. Roughley,
C. J. Smith, G. M. Williams, J. E. Davies, P. R. Raithby,
2350 | Green Chem., 2021, 23, 2336–2351
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
90 E. Muller and H. Metzger, Chem. Ber., 1954, 87, 1282– 104 D. Prat, J. Hayler and A. Wells, Green Chem., 2014, 16,
1293. 4546–4551.
91 E. Muller, H. Metzger and D. Fries, Chem. Ber., 1954, 87, 105 M. Ohno and I. Sakai, Tetrahedron Lett., 1965, 50, 4541–4544.
1449–1460.
106 M. Ohno and N. Kawabe, Tetrahedron, 1966, 33, 3935–
3938.
107 S. Buscemi, A. Pace, A. P. Piccionello, G. Macaluso,
N. Vivona, D. Spinelli and G. Giorgi, J. Org. Chem., 2005,
70, 3288–3291.
92 C. Schenk and T. J. de Boer, Recl.: J. R. Neth. Chem. Soc.,
1979, 98, 18–21.
93 D. Ranganathan, S. Ranganathan, C. B. Bao and
K. Raman, Tetrahedron, 1981, 37, 629–635.
94 B. C. Oxenrider and M. M. Rogic, J. Org. Chem., 1982, 47, 108 M. B. Plutschack, B. Pieber, K. Gilmore and
2629–2633. P. H. Seeberger, Chem. Rev., 2017, 117, 11796–11893.
95 M. Tordeux, K. Boumizane and C. Wakselman, J. Fluor. 109 B. Picard, K. Pérez, T. Lebleu, D. Vuluga, F. Burel,
Chem., 1995, 70, 207–214.
96 A. K. Gupta, D. K. Dubey and M. P. Kaushik, Org. Prep.
Proced. Int., 2005, 37, 294–298.
97 A. O. Terent’ev, I. B. Krylov, Y. N. Ogibin and
G. I. Nikishin, Synthesis, 2006, 3819–3824.
98 V. Kumar and M. P. Kaushik, Tetrahedron Lett., 2005, 46,
8121–8123.
D. C. Harrowven, I. Chataigner, J. Maddaluno and
J. Legros, J. Flow Chem., 2020, 10, 139–143.
110 J. Andraos, Reaction green metrics: problems, exercises, and
solutions, Taylor & Francis Group, LLC, 1st edn, 2018.
111 A. Armstrong, M. A. Atkin and S. Swallow, Tetrahedron:
Asymmetry, 2001, 12, 535–538.
112 D. M. Scarpino Schietroma, M. R. Monaco, V. Bucalossi,
P. E. Walter, P. Gentili and M. Bella, Org. Biomol. Chem.,
2012, 10, 4692–4695.
99 A. K. Gupta, J. Acharya, D. Pardasani and D. K. Dubey,
Tetrahedron Lett., 2007, 48, 767–770.
100 R. A. Sheldon, ACS Sustainable Chem. Eng., 2018, 6, 32–48.
101 R. A. Sheldon, Green Chem., 2017, 19, 18–43.
113 Y. Yamashita, H. Ishitani and S. Kobayashi, Can. J. Chem.,
2000, 78, 666–672.
102 H. Diekmann and W. Lüttke, Angew. Chem., Int. Ed. Engl., 114 http://ecoscale.cheminfo.org/calculator
1968, 7, 387–388. January 28, 2021).
103 A. H. A. Mohammed and G. Nagendrappa, J. Chem. Sci., 115 K. Van Aken, L. Strekowski and L. Patiny, Beilstein J. Org.
(Accessed
on
2011, 123, 433–441.
Chem., 2006, 2, 1–7.
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 2336–2351 | 2351