Expedient preparation of active pharmaceutical ingredient ketamine under sustainable continuous flow conditions

Article abstract of DOI:10.1039/c9gc00336c

A robust three-step continuous flow procedure is presented for the efficient and sustainable preparation of active pharmaceutical ingredient ketamine. The procedure relies on the main assets of continuous flow processing, starts from commercially available chemicals, utilizes low toxicity reagents and a FDA class 3 solvent under intensified conditions. The procedure features a unique hydroxylation step with molecular oxygen, a fast imination relying on triisopropyl borate and a thermolysis employing Montmorillonite K10 as a heterogeneous catalyst, all three steps being performed in ethanol. The three individual steps can be run independently or can be concatenated, thus providing a compact yet efficient setup for the production of ketamine. The scalability of the critical hydroxylation step was assessed in a commercial pilot continuous flow reactor. The process can also be adapted for the preparation of ketamine analogs. A thorough computational study on the backbone rearrangement of the cyclopentylphenylketone scaffold under thermal stress rationalizes the experimental selectivity and the various experimental observations reported herein.

Full text of DOI:10.1039/c9gc00336c

View Article Online
View Journal
Green
Chemistry
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: V. H. Kassin, R.
Gérardy, T. Toupy, D. Collin, E. Salvadeo, F. Toussaint, K. Van Hecke and J. M. Monbaliu, Green Chem.,
2019, DOI: 10.1039/C9GC00336C.
This is an Accepted Manuscript, which has been through the
Volume 18 Number
7 7 April 2016 Pages 1821–2242
Royal Society of Chemistry peer review process and has been
accepted for publication.
Green
Chemistry
Accepted Manuscripts are published online shortly after
Cutting-edge research for a greener sustainable future
www.rsc.org/greenchem
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
ISSN 1463-9262
CRITICAL REVIEW
G. Chatel et al.
Heterogeneous catalytic oxidation for lignin valorization into valuable
chemicals: what results? What limitations? What trends?
rsc.li/green-chem

Please do not adjust margins
Page 1 of 15
Green Chemistry
View Article Online
DOI: 10.1039/C9GC00336C
Green Chemistry
ARTICLE
Expedient Preparation of Active Pharmaceutical Ingredient
Ketamine under Sustainable Continuous Flow Conditions
<Received 00th January 20xx,
Accepted 00th January 20xx
Victor-Emmanuel Kassin,a,† Romaric Gérardy,a,† Thomas Toupy,a Diégo Collin,a Elena Salvadeo,a,b
François Toussaint,a Kristof Van Heckec and Jean-Christophe M. Monbaliua,*
DOI: 10.1039/x0xx00000x
A robust three-step continuous flow procedure is presented for the efficient and sustainable preparation of active
pharmaceutical ingredient ketamine. The procedure relies on the main assets of continuous flow processing, starts from
commercially available chemicals, utilizes low toxicity reagents and a FDA class 3 solvent under intensified conditions. The
procedure features a unique hydroxylation step with molecular oxygen, a fast imination relying on triisopropyl borate and
a thermolysis employing Montmorillonite K10 as a heterogeneous catalyst, all three steps being performed in ethanol. The
three individual steps can be run independantly or can be concatenated, thus providing a compact yet efficient setup for
the production of ketamine. The scalability of the critical hydroxylation step was assessed in a commercial pilot continuous
flow reactor. The process can also be adapted for the preparation of ketamine analogs. A thorough computational study
on the backbone rearrangement of the cyclopentylphenylketone scaffold under thermal stress rationnalizes the
experimental selectivity and the various experimental observations reported herein.
www.rsc.org/
Ketamine was first accepted as an ingredient for anesthetic
cocktails in the late 1960s on humans and for veterinary use.1
Introduction
()-2-(2-Chlorophenyl)-2-(methylamino)-cyclohexanone (()-
1a), also known as ketamine, belongs to a class of active
pharmaceutical ingredients (APIs) derived from the parent
phencyclidine scaffold (arylcyclohexylamines, Figure 1). The
two enantiomers of ketamine are usually referred to as
esketamine ((S)-1a) and arketamine ((R)-1a).
As a consequence of strong dissociative side effects, it was
progressively withdrawn from the pharmacopeia, although it is
still listed as a general anesthetic on the World Health
Organization (WHO) list of essential medicines.2 Since 1999, it
is also listed as a Schedule III non-narcotic substance under the
Controlled Substances Act.3,4
The last few years have witnessed a renewed interest for
ketamine, since it is now considered
a breakthrough
medication for treating major depressive disorders with
imminent risk for suicide.1,5 Depression has become a major
public health challenge, both at the social and economic levels
according to WHO.6 Despite the availability of various
antidepressants, several weeks, if not months, are necessary
to be effective on depressive patients. Ketamine, on the
contrary, is effective after the first intake.1 The Janssen
Pharmaceutical Companies of Johnson & Johnson released
earlier this year the results from two Phase III clinical studies
emphasizing the significant and rapid reduction of depressive
symptoms while using (S)-1a as a nasal spray formulation.7 The
US Food and Drug Administration (FDA) has just accepted
(March 5, 2019) this formulation as medication for treatment-
resistant depression. Ketamine acts on the central nervous
Fig.
1 General structure of the arylcyclohexylamine class of active
pharmaceutical ingredients, the parent phenylcyclidine and ketamine (1a).
a. Center for Integrated Technology and Organic Synthesis, Research Unit MolSys,
University of Liège, B-4000 Liège (Sart Tilman), Belgium
jc.monbaliu@uliege.be www.citos.uliege.be
b. Alysophil SAS, 2000 Route des Lucioles Starter Business Center – CS 70337 Les
Algorithmes - Bât. Thalès, 06410 Biot - Sophia Antipolis, France
c. XStruct, Department of Chemistry, Ghent University, Krijgslaan 281-S3, B-9000
Ghent, Belgium
system as
a non-competitive glutamatergic N-methyl-D-
aspartate (NMDA) receptor antagonist. Enantiomer (S)-1a is
more potent, although the pharmacology profile of ketamine
is complex and not completely resolved yet.1,8 The most
common pharmaceutical formulation of ketamine still consists
†
These authors have contributed equally to this manuscript
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
diffraction) and computations. See DOI: 10.1039/x0xx00000x
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 1
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 2 of 15
ARTICLE
Journal Name
of an aqueous solution of the racemate as a hydrochloride salt
(Ketalar, Calypsol, Vetalar).
View Article Online
DOI: 10.1039/C9GC00336C
Fig. 2 Retrosynthetic strategies for the preparation of racemic ketamine, ()-1a and its hydrochloride ()-1aHCl.
The pioneering work that eventually led to the preparation °C and relied on a complex chiral organophosphoric acid
of racemic ketamine (()-1a) was reported in the early 1960s catalyst, the preparation of which required several steps. The
by Stevens in a series of papers describing the complex procedure was successfully applied to an advanced ketone
behavior of -hydroxy imines and isomeric -amino ketones intermediate for the preparation of (S)-1a, which was obtained
under thermal stress (Figure 2a).9–15 Stevens’ procedure in 30% yield (99% ee) over three steps. As mentioned earlier,
involved a bromination-imination-thermolysis sequence and ()-1aHCl is a very common pharmaceutical formulation; (S)-
was adopted in the recent literature for the preparation of 1a can nevertheless be obtained by resolution of the racemate
libraries of ketamine analogs. 1,16–18
with (L)-(+)-tartaric acid22 or (L)-pyroglutamic acid.17
The bromination step on an advanced ketone intermediate
It is now well established that the assets of continuous flow
(2) toward
bromosuccinimide/benzoyl peroxide, bromine or copper chemical performance and to drastically reduce the overall
bromide in class I or II solvents such as carbon tetrachloride or footprint of
given process.23,24 Although transposing
3 is typically wasteful and involves N- micro/mesofluidic reactors can be exploited to enhance global
a
dichloromethane. The next step usually involves the reaction chemical reactions under flow conditions is not a sufficient
of brominated intermediate 3 with pure methylamine in large condition to ensure sustainability, the combination of a
excess, leading to hydroxyimine 4a. During the last step, carefully (re)designed process and the unique assets of
hydroxyimine 4a is thermolyzed at ~ 180 °C in refluxing decalin continuous flow reactors can contribute to reduce the global
or in refluxing dichlorobenzene for 20-120 min, preferably in environmental impact while boosting productivity.
the presence of hydrochloric acid. More recently, a copper-
We describe herein the development of a continuous flow
assisted method utilizing ceric ammonium nitrate process for the production of racemic ketamine. Building up on
((NH4)2Ce(NO3)6) in excess for the direct nitration of cyclic the unique, yet capricious, reactivity profile of the
ketones was reported by Zhang et al.,19 and the method was cyclopentylphenylketone scaffold, we revisit Stevens’ classical
amenable for the preparation of ()-1a, with an overall procedure for the preparation of ketamine, and propose an
isolated yield of 25% (3 steps, Figure 2b). The key nitration was economically and environmentally favorable alternative that
carried out on ketone 5 in dichloroethane at 80 °C for 12 h, bypasses problematic steps or conditions (Figure 2). The
and the corresponding nitroketone 6 was next subjected to process utilizes exclusively low toxicity reagents and a FDA
further cosmetic steps including nitro reduction and reductive class 3 solvent under intensified conditions that minimize
amination.
waste products. This procedure features 3 robust steps
The 3-4 fold higher potency of (S)-1a stimulated the starting from an advanced, readily available ketone precursor
development of enantioselective strategies. For instance, 2, and enables the atom-economic preparation of active
Kiyooka et al. reported in 2009 an asymmetric synthesis of (S)- pharmaceutical ingredient ()-1a and analogs, such as
1 according to 10 steps in 21% overall yield (99% ee) and low norketamine ()-1b and N-benzyl norketamine ()-1c. Each
atom economy.20 More recently, Toste et al. reported a step can be run individually, or can be concatenated within a
procedure for the direct asymmetric electrophilic amination of single uninterrupted reactor network upon some adjustments
α-substituted ketones with di-tert-butyl azodicarboxylates.21 in feed concentration and temperature. This work also reports
The electrophilic amination step itself required 40-60 h at 45 for the first time a computational study on the backbone
2 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 3 of 15
Green Chemistry
Please do not adjust margins
Journal Name
ARTICLE
rearrangement of the cyclopentylphenylketone scaffold under syringe pumps equipped with SS syringes and Dupont Kalrez O-
View Article Online
DOI: 10.1039/C9GC00336C
thermal stress, and the results rationalize the selectivity and rings or with HPLC pumps (ThalesNano micro HPLC). Gas feeds
the different experimental observations reported herein.
(O2 or air) were handled with Bronkhorst EL FLOW Prestige
mass flow controllers. The coil reactors were thermoregulated
with a Heidolph MR Hei-Tec® equipped with a Pt-1000
temperature sensor, unless otherwise stated (T > 180 °C).
Downstream pressure was regulated with back pressure
regulators from Zaiput Flow Technologies or from
IDEX/Upchurch Scientific.
Experimental section
General information
Conversion and yield were determined by high performance
liquid chromatography coupled to Diode-Array Detection
1
(HPLC/DAD) or by high field H NMR spectroscopy. Selectivity
Mesofluidic setup. Mesofluidic experiments (hydroxylation
step) were carried out in a Corning® Advanced-Flow™ SiC G1
reactor (6 fluidic modules connected in series, 60 mL total
internal volume). In some experiments, a residence time unit
(PFA loop, 1/8” o.d., 40 mL internal volume) was added
downstream the reactor. Feed and collection lines consisted of
PFA tubing (1/4” o.d.) equipped with PFA or SS Swagelok
connectors and ferrules. Liquid feeds were handled with
Corning dosing lines (FUJI TechnologiesTM pumps), and the gas
feed (oxygen) was handled with a Bronkhorst EL FLOW
Prestige or EL FLOW Select mass flow controllers. The reactor
was maintained at reaction temperature with a LAUDA Integral
XT 280 thermostat. Downstream pressure was regulated with
a back pressure regulator from Zaiput Flow Technologies.
is defined as the ratio between the yield and the conversion.
Structural identity was confirmed by 1H and 13C NMR
spectroscopy (400 MHz Bruker Avance spectrometer) in CDCl3
(Supporting Information) and by X-ray diffraction on single
crystals of compounds iso-7, 4a and ()-1a. The chemical shifts
are reported in ppm relative to TMS as internal standard or to
solvent residual peak. Solvents were used as received, unless
otherwise stated. Magnesium, cyclopentylbromide, 2-
chlorobenzonitrile, 2-chlorophenyl cyclopentyl ketone,
triethylphosphite, 1,4,7,10,13,16-hexaoxacyclooctadecane (18-
C-6),
1,4,7,10,13-pentaoxacyclopentadecane
(15-C-5),
1,4,7,10-tetraoxacyclododecane (12-C-4), poly(ethylene glycol)
(average Mw 400, PEG-400), ethylene glycol, glycerol,
triethylamine, 1,8-diazabicyclo(5.4.0)undec-7-ene (DBU), 2-
tert-butyl-1,1,3,3-tetramethylguanidine (Barton’s base), 2,8,9-
trimethyl-2,5,8,9-tetraaza-1-phosphabicyclo[3.3.3]undecane
Computational study
Computations were performed at the B3LYP/6-31+G** level of
theory with the Gaussian 09 package of programs (Revision
D.01).25 Computations were run with implicit solvent (PCM, ε =
24.852 for ethanol). Stationary points were optimized with
gradient techniques (tight optimization convergence).
Transition states were localized using the Newton-Raphson
algorithm, and the nature of the stationary points was
determined by analysis of the Hessian matrix. Activation and
reaction energies were obtained from the thermochemical
calculations (298.15 K, 1 atm). Intrinsic reaction coordinate
(IRC) calculations were performed on each transition state.
(Verkade's
base),
2-tert-butylimino-2-diethylamino-1,3-
(BEMP), tert-
dimethylperhydro-1,3,2-diazaphosphorine
butylimino-tris(pyrrolidino)-phosphorane (BTPP), tetramethyl-
(tris(dimethylamino)phosphoranylidene)-phosphorictriamid-
ethyl-imine (P2-Et), polystyrene-supported phosphazene base
P2-t-Bu, Amberlyst A26, Dowex Marathon A, Amberlite IRN 78,
Ambersep 900, lithium ethoxide, sodium ethoxide, potassium
ethoxide, potassium tert-butoxide, tetramethylammonium
hydroxide (TMAOH), cesium carbonate, cesium hydroxide,
sodium hydroxide, potassium hydroxide, methylamine (33 wt%
in ethanol), ammonia (7 M in methanol), benzylamine,
trimethyl borate, triethyl borate, triisopropylborate, titanium
(IV) isopropoxide, Montmorillonite K 10, hydrochloric acid (5-6
M in isopropanol or 3 M in butanol) were obtained from
commercial sources and used as received. Oxygen (Alphagaz 1)
and air (Alphagaz 1) were obtained from commercial sources.
Typical runs
Continuous flow hydroxylation toward (2-chlorophenyl)(1-
hydroxycyclopentyl)methanone 7 under homogeneous conditions
(microfluidic setup). The pump used to deliver the solution of
(2-chlorophenyl)cyclopentyl ketone (2, 1 M), KOH (0.5 M),
PEG-400 (1 M) and P(OEt)3 (1.1 M) in EtOH was set to 0.2 mL
min-1. The mass flow controller used to deliver oxygen was set
to 10 mLn min-1, and both streams were mixed through a PEEK
T-mixer. The biphasic mixture was reacted in a PFA capillary
coil (6.5 mL internal volume, estimated 5 min residence time)
at 25 °C under 11 bar of counterpressure (Figure 3a). The
reactor effluent was collected at steady state, diluted with
water and EtOH, and analyzed by HPLC/DAD (99% conversion,
99% selectivity, see Table 1).
Experimental setup
Microfluidic setups. Microfluidic reactors consisted of modular
continuous flow assemblies constructed with plug-and-play
thermoregulated SS packed-bed column (12.5 cm × 7 mm o.d.
× 4 mm i.d. or 10 cm × 13 mm o.d. × 9 mm i.d.), PFA (1.58 mm
outer diameter, 800 µm internal diameter) or stainless steel
(SS) coils (1.58 mm outer diameter, 500 μm internal diameter)
equipped with SS connectors and ferrules (Valco) or PEEK/ETFE
connectors and ferrules (IDEX/Upchurch Scientific). Feed and
collection lines consisted of PFA or PEEK tubing (1.58 mm
outer diameter, 750 μm internal diameter) equipped with
PEEK/ETFE connectors and ferrules (IDEX/Upchurch Scientific).
Liquid feeds were handled with high force Chemyx Fusion 6000
Continuous flow hydroxylation toward (2-chlorophenyl)(1-
hydroxycyclopentyl)methanone
7
under
heterogeneous
conditions (microfluidic setup). The pump used to deliver the
solution of (2-chlorophenyl)cyclopentyl ketone (2, 0.5 M) and
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 3
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 4 of 15
ARTICLE
Journal Name
P(OEt)3 (1 M) in EtOH was set to 0.15 mL min-1. The mass flow material: 3.7 g of Montmorillonite K10 in granules) at 180 °C
View Article Online
controller used to deliver oxygen was set to 4.2 mLn min-1, and under 35 bar of counter-pressure (Figure 9.b.). After 60 min of
DOI: 10.1039/C9GC00336C
both streams were mixed through a PEEK T-mixer. The operation, the reactor effluent was collected, diluted with
biphasic reaction mixture was reacted in a packed-bed column EtOH and analyzed by HPLC/DAD (69% conversion, 89%
(12.5 cm × 7 mm o.d. × 4 mm i.d.; loading material: Ambersep selectivity, see Table 6). Concentration of the reactor effluent
900 hydroxide, 1.6 cm³) operated at 0 °C under 8.5 bar of
counter-pressure (Figure 4b). The reactor effluent was
collected after 60 min of equilibration, diluted with EtOH, and
analyzed by HPLC/DAD (88% conversion, >99% selectivity, see
Table 2).
and cooling to -18 °C afforded crystals of ()-1a free base.
(36% isolated yield, 99% purity). Alternatively, addition of 2
volumes of HCl (1 M in diethyl ether) gave selective
precipitation of 1a·HCl (62% isolated yield, 99% purity).
Concatenated process toward ()-ketamine (()-1a), ()-
norketamine (()-1b) and ()-N-benzylnorketamine (()-1c). The
feed solution of (2-chlorophenyl) cyclopentyl ketone (2, 1 M),
KOH (0.5 M), PEG-400 (0.5 M) and P(OEt)3 (1.1 M) in EtOH was
injected at a flow rate of 0.25 mL min-1 and reacted with
oxygen (10 mLn min-1) at 25 °C for 5 min of residence time (11
bar). The reactor effluent was next mixed with a neat solution
of triisopropyl borate (0.13 mL min-1, 2 equiv.) and with Feed
solution 2 (option a: MeNH2, 33 wt. % in EtOH, 0.05 mL min-1;
option b: NH3 7 M in MeOH, 0.29 mL min-1; option c: neat
benzylamine, 0.04 mL min-1). The three streams were mixed
through a PEEK cross junction, and the homogeneous mixture
was reacted in a PFA coil at 25 °C for 2 min of residence time (7
bar). The reactor effluent containing either 4a (option a), 4b
(option b) or 4c (option c) was diluted with EtOH and analyzed
by HPLC/DAD (option a: conv. (2) = 98%, selec. = 95%; option
b: conv. (2) = 55%, selec. = 98%; option c: conv. (2) = 99%,
selec. = 99%). The hydroxylation-imination reactor effluent
was collected in a surge, diluted with ethanol (0.02 M in 4a-c)
and reinjected at a flow rate of 1.6 mL min-1. The solution was
preheated in a SS loop (0.5 mL internal volume), and reacted in
a SS packed-bed column (10 cm × 13 mm o.d. × 9 mm i.d.,
packing material: 3.7 g of Montmorillonite K10 in granules) at
180 °C under 35 bar of counter-pressure (see Figure 11). After
20 min of operation, the effluent of the reactor was collected,
diluted with EtOH, and analyzed by HPLC/DAD (()-1a: conv.
(4a) = 70%, selec. 93%; ()-1b: conv. (4b) = 99%, selec. 99% or
()-1c: conv. (4b) = 60%, selec. 81%).
Continuous flow hydroxylation toward (2-chlorophenyl)(1-
hydroxycyclopentyl)methanone 7 under homogeneous conditions
(mesofluidic setup). The solution of (2-chlorophenyl)
cyclopentyl ketone (2, 1 M), KOH (0.5 M), PEG-400 (1 M) and
P(OEt)3 (1.1 M) in EtOH was injected at a flow rate of 5 mL min-
1 and oxygen was delivered at a flow rate of 0.125 Ln min-1. The
two streams were mixed and reacted in a Corning® Advanced-
Flow™ G1 SiC Reactor (60 mL internal volume) operated at 40
°C under 11 bar of counterpressure. The reactor effluent was
sampled at steady state and analysed by HPLC/DAD (85%
conversion, 95% selectivity.
Continuous
flow
imination
toward
1-((2-
(4a)
chlorophenyl)(methylimino)methyl)cyclopentanol
(microfluidic setup). The syringe pump used to deliver the neat
solution of (2-chlorophenyl)(1-hydroxycyclopentyl)methanone
(7) was set to 0.38 mL min-1 (1 equiv.), the syringe pump used
to deliver the neat solution of triisopropyl borate was set to
0.79 mL min-1 (2 equiv.) and the syringe pump used to deliver
the solution of methylamine (33 wt. % in EtOH) was set to 0.32
mL min-1 (1.5 equiv.). The three streams were mixed through a
PEEK cross junction, and the homogeneous mixture was
reacted in a PFA capillary coil (1.5 mL internal volume, 1 min of
residence time) at 60 °C under 5.2 bar of counterpressure
(Figure 7b). The reactor effluent was collected at steady state,
diluted with EtOH, and analyzed by HPLC/DAD (>99%
conversion and selectivity, see Table 3).
Continuous flow thermolysis toward ()-ketamine (()-1a) under
homogeneous conditions (microfluidic setup). The HPLC pump
Results and Discussion
used
to
deliver
the
solution
of
1-((2-
chlorophenyl)(methylimino)methyl)cyclopentanol (4a, 0.42 M)
in ethanol was set to 0.8 mL min−1. The solution was
thermolyzed in a SS coil (4 mL internal volume, 5 min of
residence time) at 220 °C under 35 bar of counter-pressure
(Figure 9a). The reactor effluent was was collected at steady
state, diluted with EtOH and analyzed by HPLC/DAD (71%
conversion, 78% selectivity, see Table 4).
Hydroxylation of ketone 2 under microfluidic conditions
While considering alternatives for designing an effective flow
process for the preparation of ()-1a, hydroxyketone 7 rapidly
emerged as a robust alternative to bromoketone 3, which is a
pivotal intermediate in Stevens’ original procedure.
Bromoketone 3 is readily available from ketone 2, but its
preparation is wasteful. Since ketone 2 is commercially
available, but can also be synthesized from cheap and widely
available chemicals (Supporting Information, section 2.2.1), we
decided to keep 2 as starting material.
A wide variety of methods for the preparation of tertiary -
hydroxycarbonyl compounds directly from ketones or analogs
has been reported so far, although many procedures require
either the use of transition metal catalysts or environmentally
harmful chemicals. Among these procedures, several
Continuous flow thermolysis toward ()-ketamine (()-1a) under
heterogeneous conditions (microfluidic setup). The HPLC pump
used
to
deliver
the
solution
of
1-((2-
chlorophenyl)(methylimino)methyl)cyclopentanol (4a, 0.42 M)
in ethanol was set to 0.8 mL min−1. The solution was preheated
in a SS loop (0.5 mL internal volume), and reacted in a SS
packed-bed column (10 cm × 13 mm o.d. × 9 mm i.d., packing
4 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 5 of 15
Green Chemistry
Please do not adjust margins
Journal Name
ARTICLE
considered the direct use of molecular oxygen as a primary toward isomeric iso-7 under basic conditions. Isomeric
View Article Online
oxidizer, which is very appealing given its wide availability and hydroxyketone iso-7 was isolated anDdOIc:h10a.r1a0c3t9e/Cri9zGeCd0, 03w3i6tCh
low toxicity profile.26 For instance, Jiao reported
a
unambiguous structure determination by X-ray diffraction on a
hydroxylation procedure using molecular oxygen in the single crystal (Figure 3b).
presence of a basic catalyst (20 mol% Cs2CO3) and a reductant
(P(OEt)3) in DMSO.27 An enantioselective approach involving
molecular oxygen, aqueous NaOH, a chiral phase transfer
catalyst and a reductant (P(OEt)3 or dppe) in benzene was
reported by Zhao.28 A few years later, Gnanaprakasam
reported an alternative hydroxylation procedure in the
presence of air and a stoichiometric base (tBuOK) in DMSO or
toluene, in the absence of an external reductant.29 Jiao’s
procedure was applied in batch to ketone 2, providing
compound 7 in high yield and selectivity after 3 h on a 1 g scale
in DMSO. On a 40 g scale, about 10 h were required to reach
completion. The procedure was difficult to scale-up in batch,
mostly due to a high exotherm and the poor mass transfer
between oxygen and the liquid phase, giving inconsistent
results. Further optimization, including the use of ethanol as a
solvent, led to longer reaction times (up to 24 h) to reach a
Fig. 3. a. Metal-free hydroxylation of ketone 2 with molecular oxygen under
homogeneous conditions. b. Molecular structure by single crystal X-ray
diffraction analysis of iso-7.
similar conversion. As is, the conditions were not directly
transposable to flow, and a solution was sought.
DMSO is a FDA class 3 solvent, but it is listed as
problematic for large scale applications by CHEM21, GSK and
Sanofi solvent guides.30–33 According to the same solvent
selection guides, ethanol is considered as a green solvent, and
is therefore recommended for lab and pilot scale applications.
In addition, ethanol is also a FDA class 3 solvent and is
compatible with the entire reaction sequence, thus potentially
enabling direct telescoping without intermediate solvent swap
or other operations that would negatively affect the overall
process footprint. A preliminary set of conditions was screened
for the hydroxylation of 2 in ethanol.
Ketone 2 has a good solubility in ethanol (up to 1 M), the
limiting factor becoming the base additive. Oxygen was
selected as the primary oxidizer, and homogeneous bases
were screened. Unlike the conditions reported by
Gnanaprakasam,29 the presence of an external reductant was
mandatory since the reaction proceeds through the formation
of an intermediate hydroperoxide per-7 that needs to be
reduced . There are several reductants that can be used to
convert the intermediate hydroperoxide per-7 to
hydroxyketone 7 such as phosphines,27,28 thiourea,34 or
inorganic reductants such as sodium thiosulfate,35 although
only an ethanol-soluble reductant was eligible for this
continuous flow application. Triethylphosphite is a low cost,
widely available liquid reductant with a high solubility in
ethanol, and is likely the least toxic reductant suitable for this
application.36 In addition, its oxidation product, i.e. triethyl
phosphate, is also highly soluble in ethanol and has a low
toxicity profile. In addition to the aforementioned,
triethylphosphite and the corresponding phosphate were
found unreactive under the conditions of steps 2-3, and could
stay in the reactor effluent, hence avoiding wasteful
intermediate purifications.
With a 1 M feedstock solution of 2 in ethanol, the
optimization commenced with a thorough screening of the
base additive under continuous flow conditions (Figure 3 and
Table 1). The reaction was very exothermic, and the
temperature of the T-mixer, where mixing of the liquid and
gaseous feeds occurred, rapidly increased (actual temperature
not measured). Decision was made to thermostatize the T-
mixer as well as the reaction unit. A variety of commercially
available organic and inorganic bases soluble in ethanol were
considered (see Supporting Information, Tables S3-8 for the
detailed optimization). The nature and the concentration of
the base as well as the reaction temperature had a profound
impact on the reaction outcome (Table 1). Cesium carbonate
(0 and 20 mol%) was tested first to mimic Jiao’s procedure
under flow conditions (entries 1,2); however, in ethanol, only
12% conversion were observed at 25 °C within 5 min of
residence time.
Non-ionic bases such as amines, amidines, guanidines and
phosphazenes were selected for preliminary trials. No
conversion was obtained with triethylamine (1 equiv.) or DBU,
regardless of the residence time and the temperature (entries
3,4). With phosphazenes such as BTPP and P2-Et, the
conversion of 2 reached 40% and 95% (99% and 98%
selectivity), respectively, at 25 °C with 50 mol% of the base
(entries 5-8) within 5 min of residence time. Lower loadings of
the base decreased the conversion. Increasing the
temperature to 40 °C led to a slight increase for the conversion
with BTPP up to 58% (99% selectivity, entry 6), while with P2-
Et, the conversion of 2 reached 99%, but the selectivity
declined to 94% (entry 8). Despite the very high efficiency of
phosphazene P2-Et for the selective preparation of 7, its cost
and its impact on the global footprint of the transformation
make it quite questionable for designing a sustainable
approach toward ketamine ()-1a.
The optimization was complex due to the capricious
nature of 7 that is prone to a fast -ketol rearrangement
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 5
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 6 of 15
ARTICLE
Journal Name
conversion (97%) and selectivity (98%) were achieved at 25 °C
View Article Online
in the presence of 18-C-6 (entry 17). DeDcrOeI:a1s0in.1g03t9h/Ce9rGeCs0id0e3n36cCe
time to 2.5 min slightly decreased the conversion to 93%, but
left the selectivity unaffected.
Table 1. Process optimization for the homogeneous hydroxylation of ketone 2
Base
(mol%)
Additive
(mol%)
T
(°C)
Conv.
(%)b
Selec.
(%)b
Entrya
To further improve the atom economy of the process,
simple alkaline hydroxides were also assessed (entries 19-26).
Sodium, potassium and cesium hydroxides were soluble in a 1
M feed solution of ketone 2 in ethanol, while lithium
hydroxide was not soluble in the feed solution thus precluding
reaction under flow conditions. The trend was similar to the
one observed for the ethoxides. The temperature had a
positive impact on the conversion, but also a deleterious effect
on the selectivity, with significant amounts of isomeric iso-7
observed at 40 °C. In the worst cases, isomeric iso-7 crashed
out from solution and clogged the reactor. Larger, softer
countercations increased both the conversion and selectivity.
To further confirm this observation, tetramethylammonium
hydroxide (TMAOH) was also tested as a basic additive (entries
27-28), and up to 95% conversion (95% selectivity) was
observed at 25 °C, although a larger amount of base was
required (1:1 2/TMAOH ratio).
Similarly to alkaline alkoxides, the presence of the
appropriate crown ethers improved both conversion and
selectivity, but since crown ethers are both expensive and very
toxic,37 alternative additives were sought. The best results
were obtained for potassium hydroxide in the presence of
PEG-400 (entries 24-25). The hydroxylation reaction on ketone
2 reached 94% within 5 min of residence time at 25 °C in the
presence of 50 mol% of KOH and PEG-400. With a (1:1 2/PEG-
400 ratio), both conversion and selectivity toward 7 were total.
The hydroxylation procedure was not directly transposable to
other substrates than ketone 2 (see Supporting Information,
Tables S11-12). In addition to its low toxicity profile and wide
availability, PEG-400 was also found unreactive under the
conditions of steps 2-3, and could stay in the reactor effluent
without affecting the process. This again was seen as a major
asset since it avoided wasteful intermediate purifications and
enabled direct reaction telescoping. Intermediate purification
of hydroxyketone 7 was however possible through column
chromatography for getting analytical samples (see Supporting
Information, section 2.2.2.9).
/
95
/
1
/
/
25
25
25
140
25
40
25
40
25
40
25
40
25
40
25
25
25
40
25
25
40
25
40
25
25
25
25
0
12
0
2
3
Cs2CO3 (20)
Et3N (100)
DBU (100)
BTPP (50)
BTPP (50)
P2-Et (50)
P2-Et (50)
LiOEt (50)
LiOEt (50)
NaOEt (50)
NaOEt (50)
KOEt (50)
KOEt (50)
KOtBu (100)
KOtBu (50)
KOtBu (50)
KOtBu (50)
NaOH (50)
KOH (50)
KOH (50)
KOH (50)
KOH (50)
KOH (50)
KOH (50)
CsOH (50)
TMAOH (50)
TMAOH
/
/
/
4
/
/
0
>99
>99
98
94
10
5
5
40
58
95
>99
32
40
99
99
>99
>99
>99
77
97
99
88
85
91
96
>99
94
>99
>99
55
6
/
7
/
8
/
9
12-C-4 (50)
12-C-4 (50)
15-C-5 (50)
15-C-5 (50)
18-C-6 (50)
18-C-6 (50)
18-C-6 (50)
/
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
72
54
95
82
25
70
98
82
69
71
57
97
92
>99
>99
96
97
18-C-6 (50)
18-C-6 (50)
15-C-5 (50)
/
/
18-C-6 (50)
18-C-6 (50)
PEG-400 (50)
PEG-400 (100)
18-C-6 (100)
/
28
/
25
95
95
(100)
a Typical conditions: [2] = 1 M; [P(OEt)3] = 1.1 M; residence time = 5 min; P =
11 bar; liquid flow rate = 0.20 mL min-1; O2 flow rate = 10 mLn min-1.
Conversion and yield were determined at steady state by HPLC/DAD
processed at 220 nm.
b
More affordable ionic organic bases with a good solubility
in ethanol were next considered. Similarly to strong non-ionic
bases, the alkaline ethoxides had to be used in
substoichiometric amount to maximize the selectivity toward
7. Starting with alkaline ethoxides, both conversion and
selectivity were strongly affected by the nature of the alkaline
countercation (entries 9-18), and increasing conversion and
selectivity were observed while increasing the softness of the
countercation (Li+ < Na+ < K+). A complex HPLC reaction profile
was obtained with LiOEt. Significant improvements were
obtained while adding crown ethers, and the combination of
KOEt (50 mol%) and 18-crown-6 (18-C-6, 50 mol%) gave the
best results with complete conversion and 95% selectivity
toward 7 at 25 °C (entries 13-14). Again, the reaction was very
sensitive to the temperature, and significant amounts of
isomeric hydroxyketone iso-7 were obtained at 40 °C (entries
10-18). Shifting to the more hindered KOtBu, excellent
Next, the hydroxylation of 2 (1 M in ethanol) was further
assessed with the optimized parameters (5 min of residence
time, 25 °C, 50 mol% of KOH, 1:1 2/PEG-400 ratio) while
changing the gas feed from industrial oxygen to air. As
expected, the conversion dropped (14%) while maintaining an
excellent selectivity toward
7
(98%). Increasing the
temperature to 40 °C boosted the conversion to 20%, but had
a deleterious impact on the selectivity (85%).
Heterogeneous bases were also considered to further
increase cost-efficiency of the process and to simplify
downstream processing (Table 2 and Figure 4). For the trials
involving heterogeneous bases, less drastic conditions were
utilized with a 0.5 M feedstock solution of 2 in ethanol and 2
equiv. of triethylphosphite. No chelating additive was required
for these trials. A variety of commercially heterogenized base
6 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 7 of 15
Green Chemistry
Please do not adjust margins
Journal Name
ARTICLE
reagents were considered (see Supporting Information, Table these conditions, the conversion decreased from V9ie5w%ArtaicflteeOrnl3in0e
S13 for the detailed optimization).
min to 78% after 2 h of continuous oDpOerI:a1t0io.10n39(e/Cn9tGryC0503a3n6Cd
Heterogenized phosphazene such as polystyrene- Supporting Information, Figure S13). Despite the added value
supported P2-t-Bu (PS-P2-t-Bu) gave at best a conversion of 7% of a supported base, the decrease in performance over
(Supporting Information, Table S13, Entry 3). The most operation time drastically reduced the practical interest.
promising results were obtained with quaternary ammonium-
Based upon these observations, a likely mechanism implies
type anion exchange resins such as Amberlyst A26, Dowex the formation of enolate en-2 through the reaction of ketone 2
Marathon A, Amberlite IRN 78 and Ambersep 900 (Table 2). with base MB, and its reaction with molecular oxygen to form
The heterogeneous process was also strongly exothermic, and hydroperoxide anion per-7. Intermediate per-7 can
the packed-bed filled with the heterogeneous base was deprotonate in turn ketone 2 to form more of en-2, although
thermostatized. Similarly to the homogeneous process, higher ethanol is likely involved (not shown). The next step involves
temperatures had a deleterious impact on the selectivity, with the reduction of hydroperoxide per-7 with triethylphosphite to
increasing amounts of isomeric iso-7 being formed.
form hydroxyketone 7 (Scheme 1). Hydroperoxide per-7 is
unstable, and all attempts for isolation and characterization
failed.
Fig. 4. a. Representative examples of heterogenized bases. b. Metal-free
hydroxylation of ketone 2 with molecular oxygen under heterogeneous
conditions
At 25 °C, Amberlyst A26 and Amberlite IRN 78 deactivated
quite rapidly, while Ambersep 900 and Dowex Marathon A
gave steady conversions for at least 90 min of operation
(Supporting Information, Figure S11). The optimized conditions
implied a packed-bed column filled with Ambersep 900
operated at 0 °C to maintain the selectivity toward compound
7 at 99%, and using a 0.15 mL min-1 flow rate for the liquid
feed and an excess of oxygen (2.5 equiv.; 4.2 mLn min-1). Under
Scheme 1. Mechanism for the hydroxylation of ketone 2b toward
intermediate 7, showing the competitive formation of isomeric
hydroxyketone iso-7. MB is a hydroxide or alkoxide base (M is a
countercation).
The results from both the homogeneous and the
heterogeneous studies emphasized that the formation of
isomeric hydroxyketone iso-7 is enhanced by: (a) an excess of
base, (b) the presence of hard cations and (c) heat. DFT
computations on the -ketol rearrangement of 7 to isomeric
iso-7 provided important insights in agreement with our
observations and the mechanism depicted in Scheme 1. The -
ketol rearrangement proceeds through a unimolecular pseudo
3-membered ring transition state (Figure 5), and results in a
profound backbone modification. Although quite common for
-hydroxyketones, the -hydroxylcyclopentylketone is
particularly sensitive to such structural rearrangement. The -
Table 2. Process optimization for the heterogeneous hydroxylation of ketone 2.
T
(°C)
Conv. Selec. STY (mol L-1 day-
Entrya
Base
1
b
(%)b
(%)b
)
1
2
Amberlyst A-26
Dowex Marathon
A
25
25
56
51
94
95
47.4
43.6
3
4
5c
Amberlite IRN 78
Ambersep 900
Ambersep 900
25
25
0
46
64
88
97
93
40.2
53.6
59.4
ketol rearrangement on 7 has a high activation barrier (G≠
neut
= 39.1 kcal mol-1) when proceeding on neutral structures.
These values are in good agreement with experimental
data collected in the 1970s by Stevens and others.38 Under
basic conditions, the activation barrier decreases significantly
>99
a Typical conditions: [2] = 0.5 M; [P(OEt)3] = 1 M; P = 8.5 bar; liquid flow rate = 0.2
mL min-1; O2 flow rate = 3 mLn min-1, volume of the bed = 1.6 cm3 (neat catalyst). b
Conversion and selectivity were determined after 60 min of equilibration, by
(G≠ = 11.2 kcal mol-1), and thus the formation of isomeric
an
HPLC/DAD processed at 220 nm. STY
= space-time yield after 60 min of
iso-7 becomes significant. In the presence of a hard cation
(here modelled as a proton), the activation barrier for the -
equilibration. c Liquid flow rate = 0.15 mL min-1, O2 flow rate = 4.2 mLn min-1.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 7
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 8 of 15
ARTICLE
Journal Name
Table 3. Process optimization for the imination of hydroxyketone 7 with
ketol rearrangement is also lower than for the neutral pathway
View Article Online
methylamine in ethanol (33 wt%).
(G≠ = 14.6 kcal mol-1). These computational results are in
DOI: 10.1039/C9GC00336C
cat
agreement with the mechanism in Scheme 1 and the
Yield
[7]
(M)
MeNH2
(equiv.)
Additive
Residence
time(min)
T
(°C)
Entrya
(4a)
(%)b
94
>99c
86
>99d
87
experimental observations:
a
higher temperature will
accelerate the -ketol rearrangement, in particular when
combined with an excess of base or a hard countercation.
Excess or even stoichiometric basic reagents would negatively
affect the selectivity of the reaction, and trigger the formation
of isomeric hydroxyketone iso-7. The addition of a strong
chelating agent such as a crown ether or PEG is therefore
helpful to counteract the activation effect of hard cations.
Further computations on the parent cyclopentylphenylketone
1
2
3
4
5
0.5
0.5
neat
neat
neat
1.5
1.5
1.5
1.5
1.5
B(OiPr)3
B(OiPr)3
B(OiPr)3
B(OiPr)3
B(OiPr)3
1
1
1
1
0.5
40
60
40
60
60
a
b
Typical conditions: 2 equivalents of additive; P = 5.2 bar. Yield was
determined at steady state by HPLC/DAD processed at 216 nm.
Productivity: 65 g day-1. d Productivity: 585 g day-1.
c
emphasized a similar behavior (G≠
= 38.5 kcal mol-1, G≠
neut
an
= 11.6 kcal mol-1, G≠ = 21.8 kcal mol-1, see Supporting
cat
assessed in a commercial mesofluidic silicon carbide reactor
(Corning® Advanced-Flow™ G1 SiC reactor) equipped with 6
fluidic modules (Figure 6). The selection of silicon carbide
rather than glass was dictated by the combination of basic
conditions (KOH) and high exothermicity. The outlet of the
reactor was optionally connected to a residence time unit.
The operating specifications for such reactor impose rather
high flow rates for maintaining the highest mixing
performance (min. 30 mL min-1 up to max. 200 mL total flow
rate), thus leading to short residence times. For these
experiments, the concentration in 2 was set at 0.5 M in
ethanol with KOH (50 mol%), PEG-400 (50 mol%) and P(OEt)3
(1.1 equiv.) to reduce the viscosity of the feed solution. With a
20 g min-1 flow rate for the liquid feed and 1 Ln min-1 for
oxygen, an estimated residence time of 30 s was obtained. We
reasoned that higher temperature in conjunction with short
residence time might still afford good conversion (up to 63% at
120 °C). However, as observed under microfluidic conditions,
increasing the temperature also degraded the selectivity
toward 7 (27%, at 120 °C) with increasing amounts of iso-7
being formed. A selectivity of 99 % was obtained with 30 s of
residence time at 40 °C (11 bar), with a conversion of 15%
(daily productivity of 0.24 kg for 7).
Information, section 3.1), thus excluding a specific effect from
the ortho-chloro substituent on the phenyl group (other than
decreasing the pKa of the enolizable position in 2). Such
rearrangement was however not reported by Jiao upon the
hydroxylation of cylcopentylphenyl ketone.27
It rapidly appeared that a longer residence time was
necessary to reach higher conversions, and decision was made
to degrade the mixing efficiency and operate the reactor out
of specifications (5 mL min-1 for the liquid feed and 125 mLn
min-1 for the oxygen feed). Lower flow rates thus came with a
longer residence time and also authorized higher feed
concentration (1 M for 2 in ethanol). At 40 °C (11 bar), a
conversion of 85% with a selectivity of 95% was maintained,
thus corresponding to a productivity of 1.3 kg per day.
Fig. 5. Structures and Gibbs free energy (G≠) of reaction corresponding to
the transition states associated with the -ketol rearrangement of
hydroxyketone 7 under basic (green, TS7_an), cationic (red, TS7_cat) and
neutral (blue, TS7_neut) conditions. For details and cartesian coordinates,
see the Supporting Information (section 3.1).
Assessment of scalability for the hydroxylation process toward
hydroxyketone 7
The handling of oxygen combined with flammable organics is
typically seen as a significant safety concern, especially for
large scale applications. Continuous flow technology offers a
safe alternative for such processes, as demonstrated by Kappe
and coworkers:26 the actual amount of reacting oxygen in the
continuous flow setup per unit of time remains low, and is
consumed during the course of the reaction. The excess
oxygen is vented upon collection in the surge (see the
description of the telescoped process) and diluted with a flow
of nitrogen for direct release in the fume hood.
With the optimized microfluidic conditions for the
hydroxylation of ketone 2 in hands, further scalability was
Fig. 6. Photograph of the Corning® Advanced-Flow™ G1 SiC reactor setup for
the hydroxylation of ketone 2
8 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 9 of 15
Green Chemistry
Please do not adjust margins
Journal Name
ARTICLE
Imination of hydroxyketone 7 under microfluidic conditions
Alkyl borates have received attention the lastVfieewwAryticeleaOrsnlianes
low cost and low toxicity Lewis acids DfOorI: 1v0a.1r0io39u/sC9rGeCa0c0ti3o3n6sC,
including amidation39 and imination reactions. For instance,
Reeves et al. reported a general imination procedure with
tris(2,2,2-trifluoroethyl)borate (B(OCH2-CF3)3).40 To keep the
process economically viable and with a low environmental
impact, simple liquid trialkyl borates such as trimethyl, triethyl
and triisopropyl borate were considered. Preliminary results
were collected by performing reactions in HPLC vials,
indicating that total conversion could be attained with only 1.5
equiv. of methylamine in the presence of 2 equivalents of
B(OMe)3, B(OEt)3 or B(OiPr)3 within 1 min of reaction time at
60 °C. These results were next validated under continuous flow
conditions (Figure 7b). Three separate high pressure pumps
delivered hydroxyketone 7, MeNH2 in ethanol (33 wt%) and
neat triisopropyl borate, respectively. Quantitative conversion
and selectivity toward the desired hydroxyimine 4a was
obtained within 1 min of residence time at 60 °C in the
presence of 1.5 equiv. of methylamine and 2 equiv. of
triisopropyl borate. No noticeable differences were observed
between the alkyl borates. Most importantly, the conditions
were amenable to crude 7 (no purification of the reactor
effluent from the hydroxylation step) without impacting the
conversion and the selectivity (see Supporting Information,
section 2.2.6). The excess triisopropyl borate could be easily
removed for getting pure hydroxyimine 4a (see Supporting
Information, section 2.2.5). However, intermediate purification
was alleviated since the thermolysis step could accommodate
isopropyl borate with no decrease in performance (see Table
5), hence avoiding wasteful intermediate purifications.
The imination of 7 toward the formation of hydroxyimine 4a
was next studied. Preliminary data were collected from batch
experiments. The imination of 7 in ethanol with a large excess
of methylamine (5 equiv.) proceeds slowly at room
temperature and typically requires 3 days to reach completion
(97% isolated yield). The direct transposition of these
conditions under continuous flow (SS coil) was somewhat
cumbersome (Figure 7a). With
a solution of 0.5 M
hydroxyketone 7 in ethanol, the reaction profile was clean up
to 100 °C (Supporting Information, Figure S14). The conversion
of hydroxyketone 7 toward 4a reached a mere 38% (10 min of
residence time). From 120 °C, ()-1a could be detected in the
crude reactor effluent (2%) along with unreacted 7 (43%), 4a
(53%) and other impurities. Pushing the temperature up to 180
°C did not improve further the reaction, since increasing
amounts of ()-1a (18%) and other side products were
observed
.
Based upon these preliminary observations, we looked at
alternatives for increasing the conversion while maintaining
the reaction time below 5 min at lower temperature. Both the
results collected upon optimization of the hydroxylation of 2
and the preliminary imination trials have emphasized the
emergence of competitive side-reactions upon heating the
reaction mixture: above 40 °C, the amount of isomeric ketone
iso-7 becomes significant and above 160 °C the formation of
()-1a already becomes noticeable.
Imine formation is typically enhanced by the use of
molecular sieves, dehydration reagents and/or Lewis acids.
Similar process conditions were amenable to the
preparation of hydroxyimines 4b,c, as precursors of
norketamine ()-1b and N-benzylnorketamine ()-1c (see
Figure 11a), respectively. Iminols 4b,c were obtained in good to
excellent conversion and selectivity (see below and Supporting
Information, section 2.2.6).
Thermolysis of iminol 4a under microfluidic conditions
The rearrangement of -iminols, alike -ketols, was
extensively studied by Stevens9–15,38 and described as a general
phenomenon (Figure 8a) that can be triggered by thermal
stress, an acid or a base.41 Aiming at the preparation of
aminoketones of type 1, Stevens studied the rearrangement of
aminoketones of type iso-4 as well as iminols of type 4 (Figure
8b). It emphasized that the cyclopentyl fragment on 4 and
derivatives was a key feature, leading upon rearrangement to
a relaxed cyclohexyl backbone in high yield and selectivity;
while with other acyclic substituents on the starting
hydroxyimine, fragmentation or other side reactions were
reported.9–15,38 The series of papers concluded that the
thermolysis of iminols of type 4 was more efficient than the
thermolysis of aminoketones of type iso-4 for producing
aminoketones of type 1. Such thermolysis was described to
occur at temperatures > 180 °C without a catalyst (reaction
time 2-3 h). At temperatures higher than 230 °C, the yield
dropped and extensive decomposition was reported. Typical
yields at 180 °C in dichlorobenzene or decane are of 70 and
Fig. 7. a. High temperature imination of hydroxyketone 7 with methylamine.
b. Trialkyl borate-promoted imination of hydroxyketone 7. c. Molecular
structure by single crystal X-ray diffraction analysis of 4a.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 9
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 10 of 15
ARTICLE
Journal Name
75%, respectively. Several acids were also tested for the setup required specific adaptations for high temperature and
View Article Online
thermal rearrangement toward 1; in all cases, the iminol salts pressure operation in the presenceDOoI:f10.1039/C9GC00336C
a strong acid.
afforded higher yields in shorter reaction times and at lower Preliminary trials under microfluidic conditions using PFA
temperature than with the free bases of type 4. For instance, fluidic assemblies rapidly showed their limitations and were
up to 95% yield was obtained upon thermolysis of 4 in the not compatible with process conditions involving
presence of HCl in dichlorobenzene. The authors also reported temperatures > 180 °C and pressures > 15 bar. We thus
that the thermal rearrangement was also sensitive to solvent changed the PFA microfluidic setups to SS microfluidic setups
polarity, and swapping to more polar solvents increased the to improve the mechanical resistance under heat and pressure
conversion. It was suggested that the presence of an acid stresses, although this definitively excluded HCl as a coreagent.
would trigger the formation of an ammonium salt of 1, thus
allowing 1 to escape the series of interconnected equilibria.
Preliminary trials indicated that the thermal rearrangement
of hydroxyimine 4a requires at least 180 °C and 25 min of
The conclusions of Stevens’ series of papers can be easily residence time to reach a conversion of 83%. ()-1a could be
transposed for the preparation of ()-1a or analogs ()-1b,c, isolated in 30% yield after recrystallization (Figure 9a).
and a potential complex reaction profile is expected as a Prolonged reaction time at 180 °C or higher temperatures led
consequence of four interconnected isomeric species (Figure to significant decomposition.
8b). Several studies utilized the thermal rearrangement of
By contrast, exposure to high temperature at shorter
iminol 4a in the presence of HCl for the preparation of residence times provided much cleaner reaction profiles with
ketamine hydrochloride, as well as for the preparation of conversion of up to 80% (Table 4, entry 16) and up to 58%
ketamine analogs and metabolites.16,17,42,43
yield (77% selectivity, Table 4, entry 14) in ()-1a (see
Supporting Information, Table S16).
Fig. 8 a. General thermal rearrangement for -ketols, -iminols, -
aminoketones and -aminoimines. b. Successive -iminol/-aminoketone
rearrangements connecting iminol 4 to aminoketone iso-4. All these
structures are isomers. Blue dots refer the same carbon among isomeric
structures.
Fig. 9. a. Homogeneous thermolysis of iminol 4a under microfluidic
conditions under neutral or acid-catalyzed conditions (the cooling loop is
omitted for clarity). b. Thermolysis of 4a in the presence of Montmorillonite
K10 as heterogeneous catalyst under flow conditions (the pre-heating and
cooling loops are omitted for clarity). c. Molecular structure by single crystal
X-ray diffraction analysis of free base ()-1a. Only one molecule of the
asymmetric unit is shown (See Supporting Information 2.8.2.3).
We first reproduced the original conditions reported by
Stevens: the thermolysis of 4a in the presence of 1 equiv. of
HCl in dichlorobenzene at 180 °C gave ()-1aHCl in 74% yield
after 30 min. When HCl was omitted, ketamine could not be
extracted from the reaction mixture. Further screening of
conditions for the thermolysis of 4a and preliminary
optimization were carried out in batch trials under microwave
irradiation (see Supporting Information, Table S14).17,44 For the
sake of process simplification, ethanol was selected as solvent
for the thermolysis of hydroxyimine 4a. Although the addition
of HCl was recommended by Stevens, its utilization led to two
main issues: (a) a significant amount of ketone 7 was formed
through the hydrolysis of hydroxyimine 4a, and (b) the flow
10 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 11 of 15
Green Chemistry
Please do not adjust margins
Journal Name
ARTICLE
View Article Online
DOI: 10.1039/C9GC00336C
Table 5. Process optimization for the thermal rearrangement of -iminol 4a into
Table 4. Process optimization for the thermal rearrangement of -iminol 4a into
(±)-1a free base under homogeneous conditions without HCl or additives.
(±)-1a (free base) under homogeneous conditions in the presence of
2
equivalents of Lewis acids.
Selec.
for 1a
(%)b
80
84
61
60
26
27
75
78
38
56
29
28
81
77
n.d.d
45
Selec.
for 8
(%)b
12
10
4
39
73
45
18
15
2
38
70
45
15
12
n.d.d
50
T
(°C)
Conv.
(%)b
Entrya
solvent
Selec.
for 1a
(%)b
78
74
48
81
24
58
Selec.
for 8
(%)
12
8
12
14
15
39
T
(°C)
Conv.
(%)b
Entrya
additive
1
2c
3
4
5
6
7
8c
9
10
11
12
13
14c
15
16
17
18
EtOH
EtOH
[EMIM][ES]
DMSO
MeTHF
toluene
EtOH
EtOH
[EMIM][ES]
DMSO
MeTHF
toluene
EtOH
200
200
200
200
200
200
220
220
220
220
220
220
240
240
240
240
240
240
66
63
33
10
7
1
2
3
4
5
6
Ti(OiPr)4
B(OiPr)3
Ti(OiPr)4
B(OiPr)3
Ti(OiPr)4
B(OiPr)3
200
200
220
220
240
240
16
10
33
61
33
76
15
62
71
78
59
17
31
71
75
n.d.d
80
40
47
a
Conditions: solvent = EtOH; [4a] = 0.1 M; [additive] = 0.2 M; residence time = 5
min; P = 35 bar. Conversion and selectivity were determined by HPLC/DAD
processed at 216 nm.
b
results are thus very encouraging for the perspectives of
process concatenation (see below), although no activation
effect could be emphasized with such homogeneous Lewis
acids.
EtOH
[EMIM][ES]
DMSO
MeTHF
toluene
29
34
62
28
Clays and silica can also be used to promote the
rearrangement of -ketols and -iminols.41,47 In particular,
Montmorillonite K10 is a well-studied solid heterogeneous
acid catalyst.48 Preliminary trials in vials under microwave
irradiation (see Supporting Information, Table S15) gave very
promising results, with short reaction times, high conversion
and selectivity in the presence of Montmorillonite K10. These
results were next transposed under continuous flow
conditions. The microfluidic setup consisted of a SS packed-
bed reactor filled with 3.7 g of Montmorillonite K10 connected
upstream to a HPLC pump for delivering the liquid feed at 0.8-
1.6 mL min-1 flow rates. The packed-bed reactor was inserted
in a thermoregulated oven, and backpressure regulation was
included downstream (Figure 9b). Samples were taken
periodically from the reactor effluents, and no deactivation of
the catalyst could be evidenced upon processing 50 mL of
feedstock solution. Several solvents were assessed as well, and
ethanol gave the best results (Table 6 and Supporting
Information, Table S17). The reaction is dramatically enhanced
in the presence of Montmorillonite K10, even using solvents
that were not favorable under homogeneous conditions, such
as 2-methyltetrahydrofuran and toluene. High conversion
(75%) and selectivity (95%) were obtained at 180 °C, under 35
bar of counter-pressure in ethanol, with a 0.1 M feedstock
solution of 4a that was delivered at 1.6 mL min-1 (entry 1). The
HPLC trace of the reactor effluent was very clean, showing
essentially (±)-1a and unreacted 4a (see Supporting
Information, Figure S27). The influence of 2 equiv. of B(OiPr)3
on the reaction was also evaluated in the 180-220 °C range in
ethanol. Identical results as for the additive-free experiments
were obtained (see Supporting Information, Table S17).
[EMIM][ES] = ethylmethylimidazolium ethylsulfate. a Typical conditions: [4a] = 0.1
b
M; residence time = 5 min; P = 35 bar. Conversion and selectivity were
c
d
determined by HPLC/DAD processed at 216 nm. [4a] = 0.42 M. Complex HPLC
chromatogram.
A small library of solvents with either low toxicity or
considered as sustainable alternatives, as well as more
conventional reaction media30,31,33,45 (listed by increasing
polarity:
ethylmethylimidazolium
toluene,
2-methyltetrahydrofuran,
ethylsulfate
ethanol,
and
[EMIM][ES]
dimethylsulfoxide) was screened at temperatures ranging
between 200 and 240 °C (35 bar, 5 min of residence time).
Above 180 °C, a fragmentation process leading to N-(2-
chlorobenzylidene) methanamine (8) and cyclopentanone
became significant (see Supporting Information, Table
S16).13,14 The fragmentation process is favored in apolar
solvents (toluene and MeTHF) and is inhibited in [EMIM][ES]
and ethanol. The formation of hydroxyketone 7 in up to 10%,
as well as other unidentified side-products was however
observed in [EMIM][ES]. The cleanest reaction profiles with
the highest conversion were obtained in ethanol in the 200-
240 °C range (5 min of residence time, 35 bar), regardless of
the concentration of 4a (0.1 or 0.42 M, Table 4 and Supporting
Information, Tables S16 and S17).
Since -ketol and -iminol rearrangements can also be
promoted by Lewis acids,41,46 homogeneous Lewis acids
compatible with the upstream operations were screened
(Table 5). In particular, (Ti(OiPr)4) or B(OiPr)3 were were
investigated as homogeneous co-additives (2 equiv.) in the
feed solution of 4a (0.1 M in ethanol). While Ti(OiPr)4
prevented the rearrangement of 4a to occur and yielded
complex reaction profiles in the 200-240 °C range, B(OiPr)3
somehow inhibited the reaction at 200 °C, but gave similar
results at 220 °C as entry 7 in Table 4 (no additives). These
Eventually, experiments were carried out at 180 °C with a
more concentrated feedstock solution of 4a (0.42 M in EtOH).
Higher concentrations led to a decrease in conversion from 75
(0.1 M) to 57% (0.42 M) with a flow rate of 1.6 mL min-1
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 11
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 12 of 15
ARTICLE
Journal Name
23.3 kcal mol-1), followed with the neutral pathway (G≠
31.6 kcal mol ) and the cationic pathway (i.e. a thermal
rearrangement in the presence of a Brønsted acid, G≠
=
neut
View Article Online
Table 6. Process optimization for the thermal rearrangement of -iminol 4a into
(±)-1a (free base) under heterogeneous conditions.
-1
DOI: 10.1039/C9GC00336C
=
cat
31.9 kcal mol-1). The experimental data collected upon
preliminary optimization showed that the addition of a
Brønsted base or a Brønsted acid led to complex reaction
profiles, despite the formation of ()-1a. The addition of Lewis
acids of increasing strength did not lead to a decrease of the
activation barrier, in agreement with our experimental results:
in the presence of boron trifluoride or boron trimethoxide, the
computed activation barriers were of 36.4 and 36.0 kcal mol-1,
respectively.
The entire reaction profile connecting the four isomers of
ketamine is depicted in Figure 10 (see Supporting Information,
sections 3.1-3.5 for details). Computations emphasized that
these rearrangements proceed through unimolecular
processes with rather high activation barriers. Once formed,
ketamine ()-1a can further isomerize through a concerted
asynchronous migration of the phenyl group leading to its
cyclohexyliminol isomer iso-1a. The thermal isomerization of
()-1a to iso-1a has an activation barrier of 27.9 kcal mol-1,
which is ca. 3.7 kcal mol-1 below the thermal isomerization of
4a to ()-1a. Compound iso-1a can next undergo a ring
Selec. Selec.
T
(°C)
Conv.
(%)b
Entrya
solvent
for 1a
(%)b
95
91
89
for 8
(%)b
<1
<1
<1
<1
9
1
2c
3c,d
4
5
6
7
8
9
EtOH
EtOH
EtOH
180
180
180
75
57
69
26
16
73
75
75
40
30
73
78
[EMIM][ES] 180
94
69
DMSO
MeTHF
toluene
EtOH
180
180
180
200
89
92
91
85
69
81
68
<1
<1
<1
<1
12
<1
5
[EMIM][ES] 200
10
11
12
DMSO
MeTHF
toluene
200
200
200
[EMIM][ES] = ethylmethylimidazolium ethylsulfate. a Typical conditions: [4a] = 0.1
M; flow rate = 1.6 mL min-1; P = 35 bar; 3.7 g of Montmorillonite K10 in the
packed-bed. Conversion and selectivity were determined after injection of 50
mL of feedstock solution through the packed-bed, by HPLC/DAD processed at 216
b
nm. c [4a] = 0.42 M. d flow rate = 0.8 mL min-1.
(entries 1-2). This behavior was quite expected, as the
availability of the active sites on the catalyst decreases with
the concentration of the feedstock solution. Decreasing the
flow rate to 0.8 mL min-1 helped to restore the conversion
(69%) while maintaining a high selectivity (89%) for (±)-1a
(entry 3). Upon concentration and cooling of the reactor
effluent, (±)-1a crystallized and could be recovered by simple
filtration (36% isolated yield, 99% purity). The concentrated
mother liquor containing unreacted α-iminol 4a could be
contraction through
a similar mechanism as for the
isomerisation of 4a to ()-1a, leading to aminoketone iso-4a
yet with a smaller activation barrier (22.9 kcal mol-1). These
computations thus corroborate Stevens’ conclusions in the
sense that the most straightforward chemical route to prepare
ketamine starts from -iminol 4a rather than from
aminoketone iso-4a. Both kinetic and thermodynamic data
from these thermochemistry computations also emphasize the
complexity of the process, and the difficulty to achieve high
conversion and selectivity for the formation of ()-1a. The
unfavorable position of the equilibrium between 4a and ()-
1a, as well as the high activation barriers and low G≠
between the competitive reactions are only a few elements
underneath such complexity. Similar trends were obtained for
other structurally related compounds, such as the precursors
of norketamine and N-benzylnorketamine (±)-1b,c,
respectively (see Supporting Information, section 3.2).
recycled for further thermolysis, thus enabling
a daily
productivity of 71 g for (±)-1a (free base). Alternatively, (±)-
1aHCl could be collected directly by slow addition of a
solution of hydrochloric acid in diethyl ether to the crude
reactor effluent (62% isolated yield, 99% conversion).
The complexity of the thermal isomerization of
hydroxyimine 4a arises mainly from the emergence of
interconnected equilibria that relate the 4 structural isomers
of ()-1a, in addition to other competitive processes including
hydrolysis, fragmentation and other decomposition reactions.
DFT computations were undertaken on the unimolecular
rearrangement of -iminol 4a toward ()-1a considering basic
(anionic), neutral and acidic (cationic) conditions (Figure 10)
for gathering mechanistic information. Thermochemistry
computations indicated that the ring expansion from 4a to ()-
1a is an endergonic reaction (see Supporting Information,
sections 3.2 and 3.3), and it critically depends on the
conditions: G°an = 13.8 kcal mol-1, G°neut = 4.4 kcal mol-1 and
G°cat = 26.5 kcal mol-1. The corresponding transition states are
structurally related to those identified for the -ketol
rearrangement (see Figure 5). The activation barrier is less
sensitive to the presence of a base than for the -ketol
rearrangement. The anionic pathway, thus involving the
conjugated base of 4a (i.e. a thermal rearrangement in the
Fig. 10. Transition state structures and relative Gibbs free energy (G) for
the thermal isomerization of 4a under neutral conditions.
presence of a base), gave the lowest activation barrier (G≠
=
an
12 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 13 of 15
Green Chemistry
Please do not adjust margins
Journal Name
ARTICLE
Concatenation of the individual steps under microfluidic
conditions and preparation of ketamine and analogs
The direct concatenation toward the thermolysis of iminols
4a-c was attempted, and the reactor effluent from the
View Article Online
DOI: 10.1039/C9GC00336C
The concatenation of steps 1-2 was rather straightforward imination step was reacted in a packed-bed reactor filled with
upon minor adjustments for the hydroxylation and the Montmorillonite K10 operated at 180 °C (35 bar). Good
imination steps (Figure 11). For the hydroxylation step, conversion and selectivity (()-1a: conv. (4a) = 70%, selec.
additive PEG-400 was set at 50 mol% in the feedstock solution 93%; ()-1b: conv. (4b) = 99%, selec. 99% or ()-1c: conv. (4b):
to reduce the viscosity. Upon optimization of the imination 60%, selec. 81%) were maintained for 20 min, but then rapidly
step from hydroxyketone 7, the best conditions involved a decreased. The deactivation of the catalyst is likely due to
slight excess of methylamine (1.5 equiv.) in the presence of residual potassium hydroxide from the hydroxylation step or
B(OiPr)3 (2 equiv.) at 60 °C (1 min of residence time, Table 3). amines from the imination step, which would eventually
However, upon direct concatenation with the hydroxylation poison acidic sites of K10.
step, significant amounts of isomeric hydroxyketone iso-7
Upon collection of the final reactor effluent containing
(26%) were detected along with hydroxyimine 4a. The ketamine, triethyl phosphate, PEG-400 and the excess
formation of isomeric iso-7 can be easily rationalized: the triisopropyl borate remained in the mother liquor regardless of
combination of a higher temperature for the imination step the isolation procedure (cooling crystallization to get ketamine
(60 °C) and the presence of residual KOH from the free base or reactive crystallization with HCl in diethyl ether to
hydroxylation step likely favors the competitive -ketol get ketamine hydrochloride). No attempts for recycling triethyl
isomerization of 7 to iso-7 (see above). An obvious solution phosphate, PEG-400 or triisopropyl borate were made. The
was therefore to decrease the temperature for the imination mother liquor was then handled as a non-halogenated organic
step (25 °C), combined with an increase of the residence time waste, and disposed through the EH&S Department.
(2 min). Under such conditions, compound 4a was obtained in
98% conversion and 95% selectivity.
Conclusion
By changing Feed 2 from methylamine to ammonia or
benzylamine, iminols 4b,c were obtained as precursors of
norketamine ()-1b and N-benzylnorketamine ()-1c. The
imination of 7 with ammonia gave iminol 4b with a lower
conversion (55%) but with excellent selectivity (98%)
regardless of the excess of ammonia, while the imination with
benzylamine proceeded extremely well with 99% conversion
and selectivity. The reactor effluent from the concatenated
hydroxylation-imination steps was collected in a surge, and
diluted with ethanol prior thermolysis.
This work illustrates an economically and environmentally
favorable alternative for the preparation of active
pharmaceutical
ingredient
ketamine
that
bypasses
problematic steps or conditions. It provides concrete, atom
economic and robust conditions for the continuous flow
preparation of ketamine. The procedure relies on the main
assets of continuous processing and utilizes exclusively low
toxicity reagents and a FDA class 3 solvent (ethanol) under
intensified conditions. The procedure is amenable to the
production of ketamine analogs, being of paramount
importance within the current pharmaceutical perspectives for
new potent antidepressant medications.
Starting from a commercially available ketone (that can
also be synthesized from cheap, widely and available
chemicals), this flow process features three steps including (a)
a unique continuous flow hydroxylation procedure step relying
on molecular oxygen in the presence of a substoichiometric
amount of potassium hydroxide and PEG-400, (b) a fast
imination with methylamine
(1.5 equiv.) relying on
triisopropyl borate (2 equiv.) and (c) a thermolysis relying on
Montmorillonite K10 as a heterogeneous catalyst. Each step
can be run independently, or can be concatenated within one
single uninterrupted reactor network, while keeping high
productivity with a minimal footprint. Ketamine free base can
be directly crystallized from ethanol upon concentration of the
thermolysis reactor effluent, affording 36% isolated yield (99%
purity). Alternatively, ()-1a·HCl can be precipitated in 62%
isolated yield (99% purity) upon addition of HCl in diethyl ether
to the effluent of the reactor. The most critical optimizations
were supported by a computational study that provided
insights on mechanisms and selectivity, and hence concrete
solutions to avoid the formation of undesirable side products.
To further complement the study, the scalability of the critical
Fig. 11. Concatenation toward racemic ketamine and analogs ()-1a-c.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 13
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 14 of 15
ARTICLE
Journal Name
hydroxylation step was assessed in
mesofluidic reactor.
a
pilot continuous
1965, 30, 2962–2966.
C. L. Stevens, I. L. Klundt, M. E. Munk and M. D. Pillai, J.
Org. Chem., 1965, 30, 2967–2972.
C. L. Stevens, H. T. Hanson and K. G. Taylor, J. Am. Chem.
Soc., 1966, 88, 2769–2774.
View Article Online
DOI: 10.1039/C9GC00336C
11
12
13
Conflicts of interest
There are no conflicts to declare.
C. L. Stevens, A. B. Ash, A. Thuillier, J. H. Amin, A. Balys, W.
E. Dennis, J. P. Dickerson, R. P. Glinski, H. T. Hanson, M. D.
Pillai and J. W. Stoddard, J. Org. Chem., 1966, 31, 2593–
2601.
C. L. Stevens, A. Thuillier, K. G. Taylor, F. A. Daniher, J. P.
Dickerson, H. T. Hanson, N. A. Nielsen, N. A. Tikotkar and R.
M. Weier, J. Org. Chem., 1966, 31, 2601–2607.
C. L. Stevens, F. E. Glenn and P. M. Pillai, J. Am. Chem. Soc.,
1973, 95, 6301–6308.
L. Chen, Y. Gong and R. Salter, J. Label. Compd.
Radiopharm., 2018, 61, 864–868.
P. J. Morris, R. Moaddel, P. Zanos, C. E. Moore, T. Gould, C.
A. Zarate Jr. and C. J. Thomas, Org. Lett., 2017, 19, 4572–
4575.
S. A. Chambers, J. M. DeSousa, E. D. Huseman and S. D.
Townsend, ACS Chem. Neurosci., 2018, 9, 2307–2330.
Z.-Q. Zhang, T. Chen and F.-M. Zhang, Org. Lett., 2017, 19,
1124–1127.
R. Yokoyama, S. Matsumoto, S. Nomura, T. Higaki, T.
Yokoyama and S.-I. Kiyooka, Tetrahedron, 2009, 65, 5181–
5191.
X. Yang and F. D. Toste, J. Am. Chem. Soc., 2015, 137,
3205–3208.
K. Steiner, S. Gangkofner, J.-M. Grunenwald, US Pat.
6040479, 2000.
Acknowledgments
14
This work was supported by the University of Liège (Welcome
Grant WG-13/03, JCMM). Computational resources have been
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. Michaël Schmitz and Julien Delaisse are
acknowledged for the large-scale preparation of ketone 2. KVH
thanks the Hercules Foundation (project AUGE/11/029 “3D-
SPACE: 3D Structural Platform Aiming for Chemical
Excellence”) and the Special Research Fund (BOF) – UGent
(project 01N03217) for funding.
15
16
17
18
19
20
Notes and references
1
M. W. Tyler, H. B. Yourish, D. F. Ionescu and S. J. Haggarty,
ACS Chem. Neurosci., 2017, 8, 1122–1134.
2
World Health Organization, WHO Model List of Essential
21
22
23
Medicines
-
20th
List,
2017,
http://www.who.int/medicines/publications/essentialmedi
cines/en/ (Accessed January 19, 2019).
3
4
U. S. Drug Enforcement Administration, The Controlled
R. Gérardy, N. Emmanuel, T. Toupy, V.-E. Kassin, N. N.
Tshibalonza, M. Schmitz and J.-C. M. Monbaliu, Eur. J. Org.
Chem., 2018, 2301–2351.
Substances
Act,
https://www.dea.gov/controlled-
substances-act (Accessed January 19, 2019).
The Center for Integrated Technology and Organic
Synthesis (CiTOS) is licensed for the manufacturing of
ketamine by the Federal Agency of Medicine and Health
Products (License #620029).
24
R. Gérardy and J.-C. M. Monbaliu, Multistep Continuous-
Flow Processes for the Preparation of Heterocyclic Active
Pharmaceutical Ingredients in Topics in Heterocyclic
Chemistry: Flow Chemistry for the Synthesis of
Heterocycles, eds. U. K. Sharma and E. V. Van der Eycken,
Springer International Publishing, Cham, 2018, pp. 1–102.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M.
A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B.
Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li,
H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L.
Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J.
Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H.
Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta, F.
Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V.
N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari,
A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N.
Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V.
Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J.
W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski,
G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D.
Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski
and D. J. Fox, Gaussian 09 (Revision D.01), Gaussian Inc.,
Wallingford CT, 2009.
5
6
Y. Yang, Y. Cui, K. Sang, Y. Dong, Z. Ni, S. Ma and H. Hu,
Nature, 2018, 554, 317–322.
World
Health
Organization,
Mental
Health,
25
https://www.who.int/mental_health/management/depres
sion/en/ (Accessed January 19, 2019).
7
U. S. National Library of Medicine, A Study to Evaluate the
Efficacy, Pharmacokinetics, Safety and Tolerability of
Flexible Doses of Intranasal Esketamine Plus an Oral
Antidepressant in Adult Participants With Treatment-
resistant
Depression,
https://clinicaltrials.gov/ct2/show/NCT03434041
(Accessed January 19, 2019).
8
P. Zanos, R. Moaddel, P. J. Morris, P. Georgiou, J. Fischell,
G. I. Elmer, M. Alkondon, P. Yuan, H. J. Pribut, N. S. Singh,
K. S. S. Dossou, Y. Fang, X.-P. Huang, C. L. Mayo, I. W.
Wainer, E. X. Albuquerque, S. M. Thompson, C. J. Thomas,
C. A. Zarate Jr and T. D. Gould, Nature, 2016, 533, 481–486.
C. L. Stevens, R. D. Elliott and B. L. Winch, J. Am. Chem.
Soc., 1963, 85, 1464–1470.
9
10
C. L. Stevens, A. Thuillier and F. A. Daniher, J. Org. Chem.,
14 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 15 of 15
Green Chemistry
Please do not adjust margins
Journal Name
ARTICLE
26
27
C. A. Hone and C. O. Kappe, Top. Curr. Chem., 2019, 377, 2.
Y.-F. Liang and N. Jiao, Angew. Chem. Int. Ed., 2014, 53,
548–552.
S.-B. D. Sim, M. Wang and Y. Zhao, ACS Catal., 2015, 5,
3609–3612.
M. B. Chaudhari, Y. Sutar, S. Malpathak, A. Hazra and B.
Gnanaprakasam, Org. Lett., 2017, 19, 3628–3631.
R. K. Henderson, C. Jiménez-González, D. J. C. Constable, S.
R. Alston, G. G. A. Inglis, G. Fisher, J. Sherwood, S. P. Binks
and A. D. Curzons, Green Chem., 2011, 13, 854–862.
D. Prat, O. Pardigon, H.-W. Flemming, S. Letestu, V.
Ducandas, P. Isnard, E. Guntrum, T. Senac, S. Ruisseau, P.
Cruciani and P. Hosek, Org. Process Res. Dev., 2013, 17,
1517–1525.
47
48
X. Zhang, Y. Dai and W. D. Wulff, Synlett, 2018, 29, 2015–
2018.
M. Hechelski, A. Ghinet, B. Louvel, P. Dufrénoy, B. Rigo, A.
Daïch and C. Waterlot, ChemSusChem, 2018, 11, 1249–
1277.
View Article Online
DOI: 10.1039/C9GC00336C
28
29
30
31
32
33
34
F. P. Byrne, S. Jin, G. Paggiola, T. H. M. Petchey, J. H. Clark,
T. J. Farmer, A. J. Hunt, C. R. McElroy and J. Sherwood,
Sustain. Chem. Process., 2016, 4, 7.
D. Prat, A. Wells, J. Hayler, H. Sneddon, C. R. McElroy, S.
Abou-Shehada and P. J. Dunn, Green Chem., 2016, 18, 288–
296.
C. Mendoza, N. Emmanuel, C. A. Páez, L. Dreesen, J.-C. M.
Monbaliu and B. Heinrichs, ChemPhotoChem, 2018, 2, 890–
897.
35
36
S. Isayama, Bull. Chem. Soc. Jpn., 1990, 63, 1305–1310.
The H412 hazard statement for triethylphosphite (MSDS)
indicates potential harm to aquatic life with a lasting
effect. However, the procedure developed in this
manuscript requires a very slight excess of P(OEt)3 (1.1
equiv.) that is oxidized into triethyl phosphate under the
reaction conditions. Triethyl phosphate is a non-toxic
substance according to its MSDS.
37
E. Weber, Crown Ethers in Ullmann’s Encyclopedia of
Industrial Chemistry, Wiley-VCH Verlag GmbH & Co. KGaA,
Weinheim,
Germany,
2007,
DOI:
10.1002/14356007.a08_091.pub2.
38
39
40
C. L. Stevens, T. A. Treat and P. M. Pillai, J. Org. Chem.,
1972, 37, 2091–2097.
M. T. Sabatini, L. T. Boulton and T. D. Sheppard, Sci. Adv.,
2017, 3, e1701028.
J. T. Reeves, M. D. Visco, M. A. Marsini, N. Grinberg, C. A.
Busacca, A. E. Mattson and C. H. Senanayake, Org. Lett.,
2015, 17, 2442–2445.
41
42
L. A. Paquette and J. E. Hofferberth, The α-Hydroxy Ketone
(α-Ketol) and Related Rearrangements in Organic
Reactions, John Wiley & Sons, Inc., Hoboken, NJ, USA,
2004, DOI: 10.1002/0471264180.or062.03.
J. Jose, S. A. Gamage, M. G. Harvey, L. J. Voss, J. W. Sleigh
and W. A. Denny, Bioorg. Med. Chem., 2013, 21, 5098–
5106.
43
44
45
46
H. Elhawi, H. Eini, A. Douvdevani and G. Byk, Molecules,
2012, 17, 6784–6807.
T. N. Glasnov and C. O. Kappe, Chem. Eur. J., 2011, 17,
11956–11968.
J. P. Hallett and T. Welton, Chem. Rev., 2011, 111, 3508–
3576.
X. Zhang, R. J. Staples, A. L. Rheingold and W. D. Wulff, J.
Am. Chem. Soc., 2014, 136, 13971–13974.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 15
Please do not adjust margins