The Chapman rearrangement in a continuous-flow microreactor

Article abstract of DOI:10.1039/c9ra01347d

The Chapman rearrangement is of practical significance in pharmaceutical and fine chemical industries. It is a high temperature reaction with an exothermic nature in numerous cases. The conventional batch-wise synthesis is limited by its operational compl

Full text of DOI:10.1039/c9ra01347d

RSC Advances
View Article Online
View Journal | View Issue
PAPER
The Chapman rearrangement in a continuous-flow
microreactor†
Cite this: RSC Adv., 2019, 9, 9270
bc
Jingjie Fang,a Miaolin Ke,bc Guanxin Huang,bc Yuan Tao,bc Dang Cheng
*
bc
*
and Fen-Er Chen
The Chapman rearrangement is of practical significance in pharmaceutical and fine chemical industries. It is
a high temperature reaction with an exothermic nature in numerous cases. The conventional batch-wise
synthesis is limited by its operational complexities, temperature control difficulties and scale-up hurdles. In
this work, a microreactor-based continuous-flow approach was developed to perform the rearrangement
in a highly controlled and safer manner. High conversions were obtained within short residence times (#20
minutes). The detailed kinetics of this reaction, using 2,6-dichloro-phenyl N-phenyl benzimidate and 2-
carbomethoxy-phenyl N-phenyl benzimidate as the representative reactants, was explored at varying
temperatures to understand the intensified reaction behavior, and was modelled based on the obtained
experimental data. The continuous process was scaled up to a 16-fold larger reactor volume by increasing
the diameter of the microreactor while maintaining the residence time without further optimization. A very
slight variation was observed in the conversion for the larger-sized flow system. Upscaling the batch
reaction to a 10 times larger volume, by contrast, resulted in a dramatic decrease in the conversion. The
simplicity of scaling up continuous-flow system was clearly demonstrated. A CFD model coupled with the
obtained rearrangement kinetics was developed and well validated against the experimental data, which
provided a robust platform for guiding the relevant process design and optimization of the continuous-
flow processes. The results presented shed new light on the developments and applications of continuous-
flow method for the classical Chapman rearrangement that require harsh high temperatures.
Received 22nd February 2019
Accepted 4th March 2019
DOI: 10.1039/c9ra01347d
rsc.li/rsc-advances
amides with aryl halides,15 and the Ullmann-type cross-coupling
reactions between aryl halides and anilines typically represents
the alternative route to afford diphenylamines.16 However, the
1. Introduction
The Chapman rearrangement of aryl-N-arylbenzimidates (1) to
N,N-diarylbenzamides (2) (Scheme 1) is of high synthetic signi-
cance in pharmaceutical and ne chemical industries, since it
provides a useful approach for preparing N,N-diarylbenzamides (2)
and/or diphenylamines (3) (via facile subsequent hydrolysis of (2))
that can otherwise be difficult to access.1–6 N,N-diarylbenzamides
and diphenylamines are found in a variety of medicinally active
compounds7,8 and among the most important building blocks for
a host of chemotherapeutic drugs such as quinacrine,9,10 fenben-
dazole,11 chlorpromazine12 and diclofenac sodium,13,14 etc.
cross-coupling methods oen require metal-catalysts,
Other methods for the preparation of N,N-diarylbenzamides
generally involve C–N cross-coupling of acyclic secondary
aCollaborative Innovation Center of Yangtze River Delta Region Green
Pharmaceuticals, School of Pharmaceutical Sciences, Zhejiang University of
Technology, 18 Chao Wang Road, 310014, Hangzhou, PR China
bEngineering Center of Catalysis and Synthesis for Chiral Molecules, Department of
Chemistry, Fudan University, 220 Handan Road, Shanghai 200433, PR China.
E-mail: dcheng@fudan.edu.cn; rfchen@fudan.edu.cn
cShanghai Engineering Research Center of Industrial Asymmetrical Catalysis for Chiral
Drugs, 220 Handan Road, Shanghai 200433, PR China
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c9ra01347d
Scheme 1
9270 | RSC Adv., 2019, 9, 9270–9280
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
RSC Advances
sophisticated ligands, high temperatures and long reaction painfully long, case specic and expensive procedures.13 When
times.15,16 These factors are undesirable from economic, envi- upscaling a batch reactor to larger sizes, the volume increases
ronmental and operational safety viewpoints considering the much faster than the external surface area, so if a small reactor is
cost of noble metal catalysts (e.g., palladium) and sophisticated operated under an isothermal condition, increasing its volume
ligands as well as the strict regulations to remove trace amounts would move the reactor towards adiabatic operational regime.28
of metal species in the synthesis of medicinally active mole- In this case, safety concerns would be raised as a result. It is thus
cules. What is more, only a limited substrate scope can be highly desired that the Chapman rearrangement could be alter-
effected via cross-coupling reactions.17 Therefore, the Chapman natively carried out in a more convenient, efficient, controllable
rearrangement is the method of choice in numerous synthetic and safer approach without scale-up hurdles.
routes of practical relevance.
Fortunately, the rapid development of micro-reaction tech-
It is an intramolecular reaction via four-membered transition nology (MRT) has provided new alternatives for synthetic organic
state, as conrmed by previous experimental studies,18,19 in which chemistry and the chemical industry.25 Chemical reactions could
a 1,3 shi of an aryl group from oxygen to nitrogen occurs. The be conveniently performed in continuous-ow microreactors.
appealing usefulness of the rearrangement specically lies in Due to the high surface-to-volume ratio, exceptional heat and
several aspects. Firstly, it is a unimolecular reaction, only the mass transfer efficiencies are enabled for microreactors, in which
desired product is resulted as the end product. Hence, the the temperatures can be uniformly and accurately controlled in
operation and work-up procedures involved are simple. Secondly, a timely manner.29 Besides, the closed system combined with
it has been carried out successfully with a wide substrate scope, small hold-up volume provide an inherently safe environment to
and the yields have generally been high.20 Thirdly, the reactant (1) implement high temperature, exothermic, hazardous and even
can be readily prepared by the utilization of simple as well as explosive reactions.25 Moreover, scale up can be straightforward
widely accessible starting materials.21
for microreactors via the numbering-up strategy or simply
Whereas, the main concern is that a high temperature (e.g., adopting longer channels while maintaining the same dimen-
typically 473.15–573.15 K) is required to facilitate the aromatic sionless transport characteristics or running for an extended
migrating process.22 While if the temperature is too high, which period of time to increase throughput.30,31 Accordingly, the
is far above the temperature needed to provide the activation continuous-ow microreactors have attracted a great deal of
energy, it would decompose the reactant or product in a large attention in organic chemistry community.
number of cases.23,24 Furthermore, the situation becomes
The process design and optimization of continuous-ow
complex for the cases that are markedly exothermic. For processes can be effectively accelerated with the help of
example, the temperature was observed to have risen as much detailed knowledge of the uid ow dynamics, heat and mass
as 30 K for the reactant R1]o–COOMe, R2]H (in Scheme 1) transfer characteristics including chemical kinetics. Conven-
when operating the rearrangement at 543.15 K on a very small tionally, the employment of exhaustive experiments to assess
laboratory scale.24 More importantly, the temperature increase the importance of various parameters in large scale applications
depends partly on the scale of the operation, which is especially is a tedious, time-intensive as well as costly task, especially for
true for batch-wise processes.24 Consequently, the Chapman those cases in which the reactants/reagents/catalysts are not
rearrangement ideally needs to be run at an optimal tempera- commercially available on the market, but needs to be in-house
ture in a well-controlled manner.
prepared. For this reason, computational uid dynamics (CFD)
It is worth noting that the Chapman rearrangement reactions technique can be employed as a powerful tool to acquire an in-
are still overwhelmingly carried out in batch-wise operations. depth understanding on the processes as well as to identify the
From an operational viewpoint, it is difficult to operate and crucial parameters for the highly efficient design and optimi-
manage high temperature reactions in batch reactors, in partic- zation of large-scale applications.
ular for large scale applications. In industrial practices, reaction
Many synthetic reactions requiring high temperatures have
temperatures over 473.15 K are generally outside of normal been successfully implemented in continuous-ow micro-
operating ranges for most standard batch reactors in pharma- reactors.32,33 However, to our best knowledge, the Chapman
ceutical production facilities.25 Large-volume high-temperature/ rearrangement has never been attempted in continuous-ow
high-pressure customized reactors, on the other hand, are very microreactors in the open literature, despite of all the potential
bulky as well as costly.25 In spite of the continuous innovations in benets. Ergo, the Chapman rearrangement was carried out in
batch reactors, precise thermal control to ensure a spatially continuous-wise microreactors in this work in order to explore its
uniform temperature environment remains a big challenging performance and behavior in this new reaction fashion. Although
problem for conventional batch reactors,26 particularly when rearrangement of this type has been heavily investigated in batch
high temperature reactions of exothermic nature are considered processes since its discovery, its reaction performance and
because of the strong nonlinear dynamics and the intrinsically behavior in microreactors were never discussed and analyzed
unsteady behavior of batch reactors.27 Since the spatial temper- before. As is oen the case, continuous-wise reaction behaviors
ature gradients are unlikely to be completely eliminated, the are prone to deviate substantially from the batch-wise counter-
long-time exposure of reaction mixture to inhomogeneous heat- parts. With this in mind, it becomes necessary to study the
ing is problematic, and may cause severe decease in the yields. detailed continuous-wise reaction behaviors in order to provide
Worse still, the scale up of batch processes from laboratory size to new knowledge and insights into the rearrangement performed
commercial scale is non-trivial, which comprises complicated, in continuous manner, and thus improve upon the conventional
This journal is © The Royal Society of Chemistry 2019
RSC Adv., 2019, 9, 9270–9280 | 9271

View Article Online
RSC Advances
Paper
batch mode as well as dene the optimal operating regimes in The detailed preparation procedures are elaborated in the ESI
the continuous-ow processes. In addition, chemical kinetics (Pages S2–S8).†
could be coupled with the CFD model, thus allowing the devel-
opment of a robust predictive platform for further reliable point apparatus. HRMS was measured by the Bruker micrOTOF
process design and optimization. spectrometer. HPLC analysis was performed on an Agilent 1200
Melting points were measured on a WRS-1B digital melting-
In this work, an initial understanding on the Chapman HPLC system with UV detector and PHENOMENEX OOG-4375-
rearrangement was rst obtained by examining the effects of E0 column (250 mm ꢀ 4.6 mm). Eluents were composed of
temperature and reaction time. Then, the performance of this H2O (A) and CH3CN (B) with VA/VB ¼ 40 : 60–10 : 90. Peaks were
1
rearrangement in a continuous-ow microreactor was investi- found at l 220 nm with a ow rate of 1.0 ml minꢁ1. H (400
gated for a number of reactants. Subsequently, the reaction MHz) and 13C (100 MHz) NMR were analyzed on a Bruker
behavior was detailedly explored and the corresponding kinetic Avance 400 spectrometer in d6-DMSO using tetramethyl silane
model was formulated to describe the reaction process in (TMS) as the internal standard. The chemical shis were given
a quantitative way. Following that, scale up of both the in d (ppm) units relative to TMS. Coupling constant (J) values
continuous-ow and batch-wise approaches was discussed in were given in Hz. Multiplicities were designated by the
order to showcase the easy scale-up of ow devices. A CFD following abbreviations: s, singlet; d, doublet; t, triplet; q,
model coupled with the obtained chemical kinetics was devel- quartet; br, broad; m, multiplet. Products were puried by ash
oped to aid the analysis of transport characteristics of the column chromatography on silica gel purchased from Qingdao
continuous-ow processes nally. The results reported would Haiyang Chemical Co., Ltd.
shed new lights on the developments and applications of
continuous-ow method for the classical Chapman rearrange-
2.2 Experimental setup and method
ment that requiring harsh high temperatures.
The graphic representation of the ow system is shown in Fig. 1.
Unless otherwise specied, the adopted microreactor consisted
of a 250 ml piece of stainless steel coil with an internal diameter
(d) of 0.588 mm and a length (L) of 920 mm. In the experiments,
the reactant in tetraethylene glycol dimethyl ether (tetraglyme)
2. Experimental
2.1 Materials and analytics
Since the reactants used in this work were not commercially was introduced into the microreactor by a syringe pump (Fusion
available, they were in-house prepared according to Scheme 2. 400, Chemyx). A check value was installed between the pump
Scheme 2 General procedure for preparing the Chapman rearrangement reactants.
Fig. 1 Schematic overview of the experimental system for the Chapman rearrangement.
9272 | RSC Adv., 2019, 9, 9270–9280
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
RSC Advances
and the microreactor in order to prevent the reverse ow. The The continuous-ow process allows this rearrangement to
microreactor was immersed in an oil bath to ensure the operate in a safer and more controllable manner. The effects of
required reaction temperature. A stainless termination loop temperature (T) and reaction time (tR) were rst examined to
with an internal diameter of 0.588 mm at room temperature was have an initial understanding on this reaction based on the
connected to the microreactor. Because the Chapman rear- reactant 2,6-dichloro-phenyl N-phenyl benzimidate.
rangement required a certain high temperature, so it stopped
Fig. 2(a) reveals the temperature dependence of this rear-
proceeding once the reaction mixture owing into the termi- rangement operated in a microreactor. It is of interest to note
nation loop. The whole system was pressurized by a back that a temperature increase from 493.15 to 533.15 K resulted in
pressure valve to avoid boiling point limitations. The use of about 4.5 times higher conversion (X), but the increase in
high-boiling solvent tetraglyme (bp 548.15 K) could make the conversion slowed down with further increase in the tempera-
operation safer by considerably decreasing the pressure needed ture. It is worth pointing out that over 92% conversion was
to overcome the boiling point limitations.
achieved within only 5 min at 533.15 K, and over 99% conver-
The reaction time inside the microreactor was accurately sion was realized in just 2.5 min at 553.15 K.
controlled via varying the ow rate. The quenched effluent was
Fig. 2(b) shows the reaction time dependence of the rear-
collected aer a period of the corresponding residence time in rangement in both the microreactor and the batch reactor. As
the ow system. The collected samples were subject to further can be seen, more than 94.42% reactant was quickly converted
purication and quantitative analysis.
into product within 20 min in continuous-ow approach, but
Additionally, unless otherwise stated the relevant batch-wise the batch-wise reaction was much slower. Approximately, it took
syntheses were carried out in a 25 ml standard round-bottomed about 3 times more time for the batch-wise reaction to reach the
ask with reaction volume of 10 ml in order to compare the two similar nal conversion.
methods. The analytical and spectral data for all the obtained
Following the initial examination of the reaction conditions
reactants and products are given in the ESI (Pages S2–S8 and (Fig. 2), a group of reactants was tested as shown in Table 1 to
S13–S18 for the reactants; Pages S9–S11 and S19–S24 for the evaluate the performance of the continuous-ow method.
products).†
It appears that high conversions were achieved in only
20 min in the continuous-ow microreactors compared to
that many hours were consumed in batch-wise
approaches.2,19,21,34 In addition, it is important to emphasize
that no by-products were found in these tested reactions. The
results shown in Table 1 illustrate that the Chapman rear-
rangement could be efficiently performed in continuous-ow
microreactors.
3. Results and discussion
3.1 The Chapman rearrangement
The continuous-ow microreactor provided a convenient means
to conduct high temperature reactions. In this approach, the
reactant was continuously delivered into the microreactor with
a pump, then it owed through the reaction channel set at the
desired temperature with chemical transformation occurred
3.2 Chemical kinetics
during the course. Aer a period of the residence time, the Although chemical kinetics is useful to understand the reac-
reaction mixture was collected when it exited the system. No tion behavior as well as plays a signicant role in process
external intervention was needed aer the system started up. design and optimization, the detailed kinetics of this
Fig. 2 Effects of temperature and reaction time on the conversion for 2,6-dichloro-phenyl N-phenyl benzimidate. (a) Effect of temperature in
a microreactor for tR ¼ 5 min. (b) Effect of reaction time in a microreactor and a batch reactor.
This journal is © The Royal Society of Chemistry 2019
RSC Adv., 2019, 9, 9270–9280 | 9273

View Article Online
RSC Advances
Paper
Table 1 The Chapman rearrangement for varying reactants in a continuous-flow microreactor (Cr ¼ 0.3 mol Lꢁ1
)
Entry
Product
tR (min)
T (K)
Conversiona
Isolated yield
1
20
513.15
94.42%
81%
2
3
20
20
513.15
563.15
>99%
92%
82.8%
73.8%
4
5
20
20
563.15
563.15
78.5%
72.4%
67.4%
65.9%
6
20
563.15
83.3%
53.8%
a
Conversion was quantied by offline HPLC analysis of aliquots of reaction mixture with internal standard of diphenyl ether (bp 531.7 K).
Cr ¼ Cr,0eꢁk t
(2)
r
continuous-ow Chapman rearrangement has never been re-
ported. Efforts were thus undertaken in the present study to
In order to determine the value of kr from the experimental
data, taking the natural log of eqn (2) leaded to
explore the intensied kinetic behavior. Since it is a unim-
olecular reaction,18,19 the rate of transformation of the reactant
was derived as
ln(Cr) ¼ ln(Cr,0) ꢁ krt
(3)
vCr
ꢁ
¼ krCr
(1)
vt
As is known, the Arrhenius equation quantitatively correlates
the reaction rate constant with the corresponding energy of
activation (Ea, kJ molꢁ1) and pre-exponential factor (A, 1/s) as
where Cr is reactant concentration (mol Lꢁ1), kr the reaction rate
constant (1/s) and t time (s).
Integrating and applying the condition that at t ¼ 0 s, Cr¼
Ea
ln krðTÞ ¼ ꢁ
þ lnðAÞ
(4)
Cr,0 (initial reactant concentration), we arrived at
RT
9274 | RSC Adv., 2019, 9, 9270–9280
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
The kinetic parameters were calculated from the equations
RSC Advances
The typical reactants 2,6-dichloro-phenyl N-phenyl benzimi-
date and 2-carbomethoxy-phenyl N-phenyl benzimidate were
employed to study the detailed kinetics because they are of
practical importance in the syntheses of diclofenac sodium13 and
quinacrine,9,10 respectively. A series of kinetic experiments were
conducted in the microreactor at 493.15, 513.15, 533.15, and
553.15 K with the reactant concentration of 0.3 mol Lꢁ1. Reactant
concentrations over time were closely followed in order to obtain
the corresponding reaction rate constants. Fig. 3 shows the
variations of reactant concentrations over time. Obviously, the
rearrangement is highly temperature-dependent, and increasing
temperature would dramatically facilitate the transformation.
The conversions with time are depicted in Fig. 4. Specically,
2,6-dichloro-phenyl N-phenyl benzimidate reached only about
50.73% conversion at 493.15 K within 20 min of reaction time,
35,36
below
DH† ¼ Ea ꢁ RT
kBT
(5)
(6)
DG†
RT
kr ¼
eꢁ
h
DG† ꢁ DH†
DS† ¼
(7)
T
where R is the universal gas constant (8.314 J molꢁ1 Kꢁ1), T the
absolute temperature (K), DG† the Gibbs energy of activation (kJ
molꢁ1), DH† the enthalpy of activation (kJ molꢁ1), DS† the
entropy of activation (J molꢁ1 Kꢁ1), h Planck's constant and kB
the Boltzmann's constant.
Fig. 3 Measured reactant concentrations vs. residence time in a continuous-flow microreactor at varying temperatures for (a) 2,6-dichloro-
phenyl N-phenyl benzimidate, (b) 2-carbomethoxy-phenyl N-phenyl benzimidate.
Fig. 4 Comparison of measured and simulated conversions vs. reaction time in a continuous-flow microreactor at varying temperatures for (a)
2,6-dichloro-phenyl N-phenyl benzimidate, (b) 2-carbomethoxy-phenyl N-phenyl benzimidate (scattered dots: experimental data; solid lines:
CFD simulated data).
This journal is © The Royal Society of Chemistry 2019
RSC Adv., 2019, 9, 9270–9280 | 9275

View Article Online
RSC Advances
Paper
but over 94.42% reactant was successfully converted within the
same period of reaction time (20 min) when increasing the
temperature to 513.15 K. Further increasing it to 553.15 K, the
conversion was complete in 2.5 min. Note that over 95%
conversion was realized in 20 min at 493.15 K for 2-
carbomethoxy-phenyl N-phenyl benzimidate, and almost
complete conversion was reached in 1.5 min at 553.15 K.
The data shown in Fig. 4 offers valuable information for
aiding the choices of optimum operating variables including
the residence times and reaction temperatures. Because in
practical process design, it is oen challenging to dene the
optimal operating conditions. Higher temperature means more
energy consumption. It might not be the best practice to simply
select the highest possible temperature for large scale applica-
tions, as conversion need to be trade-off with energy
consumption and the overall operating cost.
The reaction rate constants at varying temperatures were
obtained via tting eqn (3) to the experimental data. The rate
constants of reaction as well as the corresponding kinetic
parameters (DG†, DH† and DS†) are tabulated in Tables 2 and 3.
Arrhenius correlations between reaction rate constants and
inverse temperatures are plotted in Fig. 5. The energy of acti-
vation Ea and pre-exponential factor A were obtained from the
slope and intercept, respectively (Tables 2 and 3).
Fig. 5 Arrhenius correlations between reaction rate constants and
inverse temperatures for continuous-flow processes.
3.3 Scale-up
The throughput of a single channel on sub-millimeter scale is
oen too low to supply large amounts of materials to meet
commercial needs. The straightforward approaches to
increase the throughput involve numbering up the ow
devices or adopting longer channels while maintaining the
same dimensionless transport characteristics or simply
running the reaction for an extended period of time. However,
numbering-up of sub-milliliter volume microreactor is not
industrially preferred taking into account the complexity of
simultaneously controlling ows and temperatures in
numerous microreactors in parallel. On the other hand, the
exceedingly long microchannel results in a very high pressure
drop. Alternatively, it was suggested to rst scale the micro-
channel to a larger size while maintaining transport charac-
teristics and then number up a smaller number of ow
devices.37
Fig. 5 quantitatively indicates that the higher the tempera-
ture, the larger the reaction rate constant for the Chapman
rearrangement. Note that the reaction rate constants of 2-
carbomethoxy-phenyl N-phenyl benzimidate are larger than the
counterparts of 2,6-dichloro-phenyl N-phenyl benzimidate. This
fact justies the better conversions of 2-carbomethoxy-phenyl
N-phenyl benzimidate compared to those of 2,6-dichloro-
phenyl N-phenyl benzimidate at the same operating condi-
tions as shown in Fig. 4.
Table 2 Kinetic rate constants and kinetic parameters for continuous-
flow transformation of 2,6-dichloro-phenyl N-phenyl benzimidatea
T (K)
kr (1/s)
DG†
DH†
DS†
493.15
513.15
533.15
553.15
0.0337
0.1389
0.4893
1.51
136.739
136.412
136.317
136.417
139.657
139.491
139.325
139.159
ꢁ5.918
ꢁ6.000
ꢁ5.642
ꢁ4.956
Ea ¼ 143.757 kJ molꢁ1, A ¼ 5.826 ꢀ 1013 (1/s).
a
Table 3 Kinetic rate constants and kinetic parameters for continuous-
flow
transformation
of
2-carbomethoxy-phenyl
N-phenyl
benzimidatea
T (K)
kr (1/s)
DG†
DH†
DS†
493.15
513.15
533.15
553.15
0.1471
0.5515
1.2172
3.7164
130.697
130.529
132.277
132.275
114.823
114.657
114.491
114.325
32.188
30.931
33.360
32.451
Fig. 6 Measured conversions vs. reaction time for the two batch
reactors (T ¼ 513.15 K and Cr ¼ 0.3 mol Lꢁ1).
Ea ¼ 118.923 kJ molꢁ1, A ¼ 6.126 ꢀ 1011 (1/s).
a
9276 | RSC Adv., 2019, 9, 9270–9280
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
RSC Advances
Fig. 7 Simulated reactant concentration distributions inside microreactors: (a–c) for MR1, (d–f) for MR2 (2,6-dichloro-phenyl N-phenyl ben-
zimidate, T ¼ 513.15 K, tR ¼ 20 min and Cr ¼ 0.3 mol Lꢁ1).
Based on these considerations, the sub-millimeter micro- rate of MR1 was 62 mg hꢁ1, while it reached 0.949 g hꢁ1 for
reactor (MR1) was scaled up to a 16-fold larger volume by MR2. The production rate of MR2 represents a 15.31-fold
employing
comparison was made such that the diameter and length were product per day. The scale-up experiments demonstrated the
changed while the residence time was retained constant. simplicity of scaling up continuous-ow Chapman
a larger-sized stainless steel coil (MR2). The increase in scale, which could be translated to 22.776 g
Meanwhile, the batch-wise reaction volume was scaled from rearrangement.
10 (BR1) to 100 ml (BR2) as well. Notably, the conversion
dramatically decreased by more than 33.3% when upscaling the
batch reaction to a 10 times larger volume.
3.4 CFD modelling
It is difficult to experimentally measure the transport charac-
Fig. 6 displays the conversions with time for the two batch
reactors. It is observed that the reaction in larger volume was
much slower than in the small one, suggesting a strong scale-up
effect for the conventional batch-wise synthesis.
However, in stark contrast, only a very slight decrease was
found in the conversion when increasing 16-fold the reaction
volume for the continuous-ow method. Clearly, it exemplied
the ow system could be easily scaled to a larger dimension
without palpably affecting the reaction results. More impor-
tantly, no further optimization was needed. The production
teristics (e.g., species concentration distributions, temperature
distributions, etc.) inside the microreactor used in this work
because it is made of opaque stainless steel and has a very small
dimension. CFD modelling can thus be used to obtain an in-
depth understanding on the transport characteristics inside
the microreactors, as well as a powerful tool in the relevant
process design and optimization. Since the microreactors used
are circular tubes, a 2D axisymmetric computational domain
was reasonably assumed to decrease the computing burden.
The uid ow inside was laminar with Reynolds number in the
This journal is © The Royal Society of Chemistry 2019
RSC Adv., 2019, 9, 9270–9280 | 9277

View Article Online
RSC Advances
Paper
Fig. 8 Temperature distributions in MR1 and MR2 (2,6-dichloro-phenyl N-phenyl benzimidate, T ¼ 513.15 K, tR ¼ 20 min and Cr ¼ 0.3 mol Lꢁ1).
Table 4 Comparison of scale-up properties between continuous-flow and batch-wise syntheses using 2,6-dichloro-phenyl N-phenyl benzi-
midate (T ¼ 513.15 K and Cr ¼ 0.3 mol Lꢁ1
)
Reactor
Geometric parameters
Reaction volume
Reaction time
Conversionc
Isolated yield
MR1
MR2
BR1
BR2
d ¼ 0.588 mma, L ¼ 920 mma
d ¼ 1.755 mma, L ¼ 1650 mma
VR ¼ 25 ml askb
0.25 ml
4.0 ml
10 ml
20 min
20 min
20 min
20 min
94.42%
92.7%
70.92%
47.27%
81%
78%
d
ꢁ
d
VR ¼ 250 ml askb
100 ml
ꢁ
a
b
c
Internal diameter (d) and length (L) of the microreactor. Standard round-bottomed ask, VR is the total reactor volume. Conversion was
d
quantied by offline HPLC analysis of aliquots of reaction mixture with internal standard of diphenyl ether. Not isolated.
range of 0.13–0.74. The uid was assumed to be constant eqn (11). The numerical details are given in the ESI (Page
density considering that it was pressurized in the system.
The continuity and momentum equations were given by the
Navier–Stokes equations
S12).†
The CFD simulated values of conversion are compared to the
corresponding experimental data as shown in Fig. 4. As can be
V$(ru) ¼ 0
(8)
(9)
Â
À
ÁÃ
vu
vt
T
r
þ rðu$VÞu ¼ V$ ꢁpI þ m Vu þ ðVuÞ
where u is the ow eld, p the pressure and r the viscosity.
The energy conservation equation in the microreactor was
modelled as
ꢀ
ꢁ
vT
vt
rCp
þ ðu$VÞT þ V$ðꢁkhVTÞ ¼ krCrDH
(10)
where T is the temperature (K), Cp the specic heat, kh the
thermal conductivity and DH the enthalpy of reaction.
The mass conservation of the reactant in the microreactor
was formulated as
vCr
þ u$VðCrÞ ꢁ V$ðDMVCrÞ ¼ ꢁkrCr
(11)
vt
where DM is the molecular diffusivity. In the CFD modelling, the
obtained kinetics eqn (4) was included into the source term of
Fig. 9 Comparison of experimental and CFD simulated conversions.
9278 | RSC Adv., 2019, 9, 9270–9280
This journal is © The Royal Society of Chemistry 2019

View Article Online
Paper
RSC Advances
seen, the simulated conversions (solid lines) matched well with process design and optimization of the continuous-ow
the experimental data (scattered dots).
Chapman rearrangement.
The simulated reactant concentration distributions inside
MR1 and MR2 are shown in Fig. 7. Be noted that the geome-
tries were scaled by a factor of 200 for both MR1 and MR2 as
they had very large aspect ratios. Interestingly, it is noted that
the plug ow pattern was very closely presented in MR1, while
the ow prole obviously deviated from the ideal plug ow in
MR2.
Conflicts of interest
There are no conicts to declare.
Acknowledgements
Fig. 8 exhibits the temperature distributions in MR1 and
MR2. The CFD simulations revealed that the reaction uid was
heated up to 513.15 K in 4.03 seconds for MR1 and in 5.88
seconds for MR2 respectively aer it was pumped into the ow
system from room temperature. The similarly distributed
temperatures in MR1 and MR2 ensured that the reaction results
were close to each other. Nevertheless, the deviation from plug
ow pattern in MR2 might be the cause for the slight variation
in the conversion as compared to that of MR1 (Table 4).
The simulated values of conversion and experimental data
are compared in Fig. 9 with the mean absolute error of 1.06%.
Almost all the simulated data fell within 3% absolute error
with the experimental data. Therefore, good agreement was
obtained between the simulated values and the measured
data, which conrmed that the CFD model was robust and
could be further employed as a powerful predictive platform
in the process design and optimization when translating the
developed continuous-ow approach into industrial
practices.
This work was nancially sponsored by Shanghai Pujiang
Program (No. 18PJ1401100) and State Key Laboratory of
Chemical Engineering (No. SKL-ChE-18A03).
References
1 A. W. Chapman, J. Chem. Soc., 1929, 569–572.
2 F. Benington, E. V. Shoop and R. Poirier, J. Org. Chem., 1953,
18, 1506.
3 G. Bock, Chem. Ber./Recl., 1967, 100, 2870–2884.
4 V. Burdukovskii, D. Mognonov, S. Allayarov, S. Botoeva and
Z. P. Mazurevskaya, Russ. Chem. Bull., 2004, 53, 1773.
5 A. Kaczor, L. M. Proniewicz, R. Almeida, A. Gomez-Zavaglia,
M. L. S. Cristiano, A. M. M. Beja, M. R. Silva and R. Fausto, J.
Mol. Struct., 2008, 892, 343–352.
6 A. Imberdis, G. Lefevre, P. Thuery and T. Cantat, Angew.
Chem., Int. Ed., 2018, 57, 3084–3088.
7 A. Kumar and A. K. Mishra, J. Pharm. BioAllied Sci., 2015, 7,
81–85.
8 J. Wagner, M. L. Wagner and W. A. Hening, Ann.
Pharmacother., 1998, 32, 680–691.
4. Conclusion
In this work, a microreactor-based continuous ow method
9 M. Kimura, J. Chem. Soc., Perkin Trans. 2, 1987, 205–209.
was developed for the high temperature Chapman rearrange- 10 W. G. Dauben and R. L. Hodgson, J. Am. Chem. Soc., 1950, 72,
ment. The process was found to operate in a highly controlled 3479–3480.
and safer manner as compared to the conventional batch 11 L. Peterson, A. Douglas and R. Tolman, J. Heterocycl. Chem.,
mode. High conversions were reached for a range of reactants 1981, 18, 659–662.
within short residence times (#20 minutes). The continuous- 12 F. Lopez-Munoz, C. Alamo, G. Rubio and E. Cuenca, Prog.
ow rearrangement was several times faster than the batch- Neuropsychopharmacol. Biol. Psychiatry, 2004, 28, 205–208.
wise counterpart. The detailed kinetics of this reaction, 13 I. Grafe, H. Schickaneder and K. H. Ahrens, US Pat.,
using 2,6-dichloro-phenyl N-phenyl benzimidate and 2- US4978773, 1990.
carbomethoxy-phenyl N-phenyl benzimidate as the represen- 14 F. Nohara, UK Patent, GB1532087, 1978.
tative reactants, was explored at varying temperatures to 15 J. D. Hicks, A. M. Hyde, A. M. Cuezva and S. L. Buchwald, J.
understand the intensied reaction behavior, and was
Am. Chem. Soc., 2009, 131, 16720–16734.
modelled based on the obtained experimental data. The 16 J. F. Hartwig, Angew. Chem., Int. Ed., 1998, 37, 2046–2067.
determined kinetic parameters indicated that the activation 17 A. R. Katritzky and A. J. Cozens, J. Chem. Soc., Perkin Trans. 1,
energy of 2,6-dichloro-phenyl N-phenyl benzimidate is higher
than that of 2-carbomethoxy-phenyl N-phenyl benzimidate. 18 K. B. Wiberg and B. I. Rowland, J. Am. Chem. Soc., 1955, 77,
The continuous-ow synthesis was scaled up 16 times in 2205–2209.
volume by increasing the diameter of the microreactor while 19 O. H. Wheeler, F. Roman and O. Rosado, J. Org. Chem., 1969,
maintaining the residence time. Only a very slight decrease 34, 966–968.
1983, 11, 2611–2615.
was observed in the conversion for the larger-sized system. By 20 J. Schulenberg and S. Archer, Org. React., 2004, 14, 1–51.
contrast, upscaling the batch-wise reaction to a 10 times larger 21 O. H. Wheeler, F. Roman, M. V. Santiago and F. Quiles, Can.
volume resulted in a dramatic decrease in the conversion. The
J. Chem., 1969, 47, 503–504.
simplicity of scaling up continuous-ow system was well 22 H. M. Relles, J. Org. Chem., 1968, 33, 2245–2249.
demonstrated. A CFD model coupling with the rearrangement 23 A. W. Chapman, J. Chem. Soc., 1927, 1743–1751.
kinetics was developed and well validated against the experi- 24 M. M. Jamison and E. Turner, J. Chem. Soc., 1937, 0, 1954–
mental data, which provided a robust platform for guiding the
1959.
This journal is © The Royal Society of Chemistry 2019
RSC Adv., 2019, 9, 9270–9280 | 9279

View Article Online
RSC Advances
Paper
25 B. Gutmann, D. Cantillo and C. O. Kappe, Angew. Chem., Int. 30 K. S. Elvira, X. C. I. Solvas, R. C. R. Wootton and A. J. deMello,
Ed., 2015, 54, 6688–6728. Nat. Chem., 2013, 5, 905–915.
26 M. Huzmezan, B. Gough and S. Kovac, Proceedings of the 31 S. G. Newman and K. F. Jensen, Green Chem., 2013, 15, 1456–
2002 American Control Conference (IEEE Cat. No. CH37301),
2002, vol. 2, pp. 1156–1161.
27 F. Caccavale, M. Iamarino, F. Pierri and V. Tufano, Control and
1472.
32 T. Razzaq, T. N. Glasnov and C. O. Kappe, Eur. J. Org. Chem.,
2009, 9, 1321–1325.
Monitoring of Chemical Batch Reactors, Springer, London, 2011. 33 T. Razzaq and C. O. Kappe, Chem.–Asian J., 2010, 5, 1274–
28 J. Worstell, Batch and Semi-Batch Reactors-Practical Guides in 1289.
Chemical Engineering, Butterworth-Heinemann, Waltham, 34 R. Barclay, Can. J. Chem., 1965, 43, 2125–2131.
2015. 35 J. Boche and O. Runquist, J. Org. Chem., 1968, 33, 4285–4286.
29 N. Kockmann and D. M. Roberge, Chem. Eng. Technol., 2009, 36 K. J. Laidler and M. C. King, J. Phys. Chem., 1983, 87, 2657–2664.
32, 1682–1694.
37 Y. J. Zhang, S. C. Born and K. F. Jensen, Org. Process Res. Dev.,
2014, 18, 1476–1481.
9280 | RSC Adv., 2019, 9, 9270–9280
This journal is © The Royal Society of Chemistry 2019