An open-source approach to automation in organic synthesis: The flow chemical formation of benzamides using an inline liquid-liquid extraction system and a homemade 3-axis autosampling/product-collection device

Article abstract of DOI:10.1016/j.tet.2018.02.043

Several open-source hardware and software technologies (RAMPS, Python, PySerial, OpenCV) were used to control an automated flow chemical synthesis system. The system was used to effect the synthesis of a series of benzamides. An inexpensive Raspberry Pi single board computer provided an electronic interface between the control computer and the RAMPS motor driver boards.

Full text of DOI:10.1016/j.tet.2018.02.043

Accepted Manuscript
An open-source approach to automation in organic synthesis: The flow chemical
formation of benzamides using an inline liquid-liquid extraction system and a
homemade 3-axis autosampling/product-collection device
Matthew O'Brien, April Hall, John Schrauwen, Joyce van der Made
PII:
S0040-4020(18)30188-1
10.1016/j.tet.2018.02.043
TET 29314
DOI:
Reference:
To appear in: Tetrahedron
Received Date: 17 October 2017
Revised Date: 15 February 2018
Accepted Date: 19 February 2018
Please cite this article as: O'Brien M, Hall A, Schrauwen J, van der Made J, An open-source approach to
automation in organic synthesis: The flow chemical formation of benzamides using an inline liquid-liquid
extraction system and a homemade 3-axis autosampling/product-collection device, Tetrahedron (2018),
doi: 10.1016/j.tet.2018.02.043.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
An open-source approach to automation in organic synthesis: the flow
chemical formation of benzamides using an inline liquid-liquid extraction
system and a homemade 3-axis autosampling/product-collection device.
Matthew O'Briena*, April Halla, John Schrauwenab and Joyce van der Madeab
aSchool of Chemical and Physical Sciences, Lennard-Jones laboratories, Keele University, Borough of
Newcastle-under-Lyme, Staffordshire, ST5 5BG, United Kingdom of Great Britain and Northern
Ireland.
bAvans University of Applied Sciences, Lovensdijkstraat 61 – 63, 4818 AJ Breda, North Brabant, the
Netherlands.
Graphical Abstract:
Abstract
Several open-source hardware and software technologies (RAMPS, Python, PySerial, OpenCV) were
used to control an automated flow chemical synthesis system. The system was used to effect the
synthesis of a series of benzamides. An inexpensive Raspberry Pi single board computer provided an
electronic interface between the control computer and the RAMPS motor driver boards.
Introduction
Automation, particularly since the industrial revolution, has had a profound impact on the
technological and economic evolution of human society. There are now very few areas of human
activity which have not seen the introduction of significant automation in one form or another.1 It is,
therefore, somewhat surprising that chemical synthesis (at least in the majority of academic
research laboratories) is still performed using largely the same apparatus and labour intensive
techniques developed when the discipline was still young. Whilst a number of very well engineered
and robust automated chemical synthesis platforms are available commercially, and have obtained
growing popularity in industrial research laboratories, their relatively high price often puts them
beyond the reach of many academic research groups.
1

ACCEPTED MANUSCRIPT
In a growing context of chemist-led technological innovation,2 we have been interested in using
open-source hardware and software technologies to develop alternative low-cost automation
platforms for chemical synthesis.
In the last two decades or so, a number of significant developments have taken place which have
made enabling technologies much more accessible to 'non experts'. In terms of coding, the advent of
high-level scripting languages, like Python,3 have made it easier to write highly functional programs
without the need to understand underlying low-level processes. A major attraction of Python is the
availability of a vast array of libraries that provide specific functionality across a diverse range of
application areas. These often act as wrappers around code written in computationally more
efficient languages such as C/C++ or Fortran. In terms of electronic hardware, the growing
availability of easy to program microcontrollers (such as Arduino,4 PICAXE5 and
PyBoard/MicroPython6) as well as General-Purpose-Input-Output (GPIO) enabled single-board
computers (such as the Raspberry Pi7 and BeagleBoard8) has enabled the ready interfacing of control
systems with electronic machinery and components.9 In addition to the technologies themselves,
the community-driven models of open innovation and applications that have evolved alongside
them have also played a major role in their success and popularity.10
In our previous work,11 we carried out a series of acid catalysed silyl group deprotections using a
homemade 3-axis autosampler and a series of motor actuated valves to select and load starting
materials into a holding loop and to control the liquid flow through the system. In that case, whilst
the selection of the starting materials was automated, the collection of reaction products into
separate containers was not and switching between the collection vessels had to be performed
manually.
In this manuscript we describe an improved automation system which selects and loads starting
materials and also collects products into separate collection flasks.
2

ACCEPTED MANUSCRIPT
Figure 1. Simplified schematic of our previous (top) and current (bottom) reaction systems.
Shown in Figure 1 (top) is a simplified schematic representation of our previous system. Starting
materials were taken from the autosampler and loaded into the holding loop by means of Syringe
Pump 1 (the holding loop was not completely filled, thereby avoiding contamination of Syringe
Pump 1). The common reagent was loaded into Syringe Pump 2 and, after switching of valves, the
Piston Pump and Syringe Pump 2 would push the two solutions through the reaction/purification
zone and product flow streams would emerge at the outlet to be collected in suitable vessels. As the
autosampler was not also being used to collect products, the necessary flushing of the residual
starting material (in the line between the autosampler and 3-Way Valve 2) into a waste channel
could be left until after the reaction had completed (by moving Syringe Pump 1 back to its original
starting position).
3

ACCEPTED MANUSCRIPT
For our updated autosampling/product-collection system, the flushing of the line to the autosampler
could not be left until after the reaction had completed as the residual starting material would then
end up in the product collection vessel. In order to incorporate this step in a way which minimised
the amount of additional hardware required, we added an extra 3-Way Valve and a secondary
Holding Loop (Figure 1, bottom). This would allow Syringe Pump 1 to load the required starting
material into Holding Loop 1 and then, after switching of 3-Way Valves 2 and 4 and moving of the
autosampler needle to the waste position, to flush out the residual starting material (between the
autosampler and 3-Way Valve 4) by moving back to its original position. After moving the
autosampler needle to the correct collection vessel, the Piston Pump could then be used to push the
starting material flow stream through the reaction system. The volume of Holding Loop 2 was
chosen to be slightly higher than that of Holding Loop 1 in order to prevent unwanted premature
mixing of the starting material with the common reagent stream (from Syringe Pump 2).
In order to demonstrate the operation of the system, we sought to carry out the flow chemical12
formation of a series of benzamides, 4, using benzoyl chloride, 1, as the common limiting reagent
and a series of primary amines, 2, as the liquid-liquid extractable13 starting materials used in excess
(Scheme 1). In order to facilitate rapid reaction in the flow system at room temperature, we also
planned to use 2 equivalents of 4-dimethylaminopyridine (DMAP, 3) as the liquid-liquid extractable
stoichiometric base due to its ability to also act as a nucleophilic catalyst.14 One of the key limitations
of flow chemistry is that it is generally incompatible with reactions that form precipitates. As the
most likely precipitate from the reactions in our solvent of choice (dichloromethane) would be the
HCl salt of DMAP, we carried out preliminary solubility studies of this material (which was made by
the addition of aqueous HCl to DMAP in THF solution).
Scheme 1
Unsurprisingly, the material was insoluble in neat dichloromethane at realistic concentrations.
Acetonitrile was attemped as a cosolvent but this did not provide significant additional solubility.
Methanol did prove to be a useful cosolvent and a 3 : 1 mixture of dichloromethane : methanol was
found capable of dissolving DMAP-HCl at a concentration of at least 0.17 M. When reactions
between dichloromethane solutions of benzoyl chloride (1) and methanolic solutions of excess
benzylamine (2a) and DMAP (3) were attempted, although the reactions proceeded rapidly at room
temperature, in the absence of precipitation and with all of the excess DMAP and amine being
efficiently removed by extraction with aqueous HCl, the amide product 4a was accompanied by
small amounts of methyl benzoate, indicating that methanol was competing as a nucleophile with
the amine. In order to reduce the nucleophilicity whilst still providing the requisite polarity,
isopropanol was investigated as an alternative, more sterically hindered cosolvent. Again, significant
concentrations of DMAP hydrochloride could be obtained using isopropanol as a cosolvent with
dichloromethane (e.g. greater than 0.13 M in 1 : 1 dichloromethane : isopropanol). Pleasingly, when
we attempted the same formation of benzylbenzamide using this solvent mixture, the reaction
4

ACCEPTED MANUSCRIPT
proceeded rapidly at room temperature, again without any precipitation, and with no apparent
formation of isopropyl benzoate as a side product.
With suitable reaction conditions in hand, our attention turned to the utilisation of the automation
system to carry out a series of flow chemical benzamide formations. A full schematic of the flow
system is shown in Figure 2.
-
H2PO4 N+H3R
PhCOCl
0.13 M
DCM
H3PO4
-
H2PO4 HDMAP+
(0.5 M aq.)
Piston
Pump 2
aqueous
3-way
valve 3
in
2.0 mL/min
2-way
Reaction
M
valve
loop
M
aqueous
out
Syringe Pump 2
mixer
T-1
Holding
Loop 2
10 mL
(t = 10 min)
T-2
Holding
6.0 mL
Loop 1
Piston
webcam
3-way
valve 2
Pump 1
M
M
4.7 mL
3-way
valve 1
separation
vessel
DCM
check
valve
3-way
valve 4
M
O
Python
OpenCV
R-NH2 2a-j
+ DMAP 3
R
N
Syringe Pump 1
H
0.26 M IPA
....
....
....
4a-j
R-Pi
3-axis
autosampler/collector
MX MY MZ
Figure 2. Schematic of the flow system for the automated formation of benzamides. --- Motor
control from Raspberry Pi. --- Control computer link to webcam and pump control. ― Flow line.
The main Python control script for the system ran on a laptop computer. This computer also
communicated directly with Syringe Pumps 1 and 2 as well as Piston Pump 1 through USB-RS232
connections. In total, 8 stepper motors were required for mechanical actuation (4 x 3-Way-Valves, 1
x 2-Way Valve, 3 axes of the autosampler/product-collector). These were controlled indirectly
through an intermediate Raspberry Pi single board computer. Connection between the laptop and
the Raspberry pi was via a UART TTL-232R-3V3 cable. In our previous system, the Pololu A4988
stepper driver boards15 used for motor control were mounted on single breadboards. With an
increased number of motors, it became more important to minimise the number of wires and
contacts. For this reason, we used 2 RAMPS boards16 (each of which can accommodate up to 5
Pololu A4988 boards) which were connected directly to the Raspberry Pi. Although RAMPS boards
are primarily designed as shields for Arduino compatible input-output interface layouts, this did
5

ACCEPTED MANUSCRIPT
provide significant improvements to the organisation of the wiring (see Supporting Information for
further details).
Solutions of 10 amine starting materials 2a-j and DMAP 3 in isopropanol were placed in vials whose
openings were covered with two layers of parafilm (which would allow the take-up needle to easily
pass through but would prevent any premature evaporation of solvent). These were placed in a test
tube rack which was positioned on the bed of the autosampler. As mentioned above, the
autosampler was used to select the appropriate amine/DMAP solution and load it into Holding Loop
1 using Syringe Pump 1. A total of 5 mL of solvent was taken in by Syringe Pump 1. The total volume
of the tubing line between the amine solution and 3-Way Valve 2 was 1 mL, so that only 4 mL would
enter Holding Loop 1 and thus be available for reaction (the residual 1 mL would be flushed to
waste). Holding Loop 1, which was 4.7 mL in volume, was therefore large enough to accommodate
this quantity of amine/DMAP solution without the risk of contamination upstream of 3-Way Valve 1.
After switching of 3-Way Valves 2 and 4, and moving the autosampler needle to the waste position,
Syringe Pump 1 moved back to its original position, thereby pushing the starting material solution
from Holding Loop 1 to Holding Loop 2 whilst at the same time flushing out the residual starting
material downstream of 3-Way Valve 4. Once Syringe Pump 2 (which was preprimed with 5 mL
benzoyl chloride solution) had taken in the appropriate quantity of benzoyl chloride solution (0.13 M
in DCM, 3 mL) 3-Way Valve 3 was switched to allow the two streams to meet and react. As the acid
chloride was the limiting and non-extractable reagent and the acid-extractable amine was being
used in excess, it was important to make sure that the acid chloride solution was always
accompanied by amine. As Holding Loop 2 was 6 mL in volume (to fully accommodate all of the
amine solution), there was an approximately 1 mL volume of neat dichloromethane at the 'front' of
Holding Loop 2 and 1 mL of neat dichloromethane at the 'tail' (as 5 mL of solvent were injected by
Syringe Pump 1 to move the starting material between the two Holding Loops). Syringe Pump 2
began pumping the benzoyl chloride solution (for 6 minutes at a flow rate of 0.5 mL min-1) 3 minutes
after Piston Pump 1 started pumping into the starting material stream (corresponding to 1.5 mL at
0.5 mL min-1, the flow rate of Piston Pump 1). This meant that there would be an approximately 0.5
mL 'overlap' of amine solution entering t-piece T-1 prior to the acid chloride solution and an
approximately 0.5 mL 'overlap' after the acid chloride solution had ceased entering T-1. During the
batch studies of cosolvent screening, we established that a 10 minute reaction time was at least
sufficient for complete reaction conversion at room temperature. A 10 mL reaction loop was
therefore placed downstream of T-1 to provide the appropriate residence time. To maintain a
constant 1 mL min-1 flow rate through the reaction loop, the flow rate of Piston Pump 1 was
increased to 1 mL min-1 (from 0.5 mL min-1) once Syringe Pump 2 had ceased pumping. The reaction
was quenched downstream of the reaction loop, at t-piece T-2, by the addition of an aqueous
stream of 0.5 M phosphoric acid solution at 2.0 mL min-1. Phosphoric acid was chosen, rather than
an alternative such as hydrochloric acid, as it is generally compatible with the types of stainless steel
commonly used in HPLC pumps. The combined biphasic aqueous/organic stream was thoroughly
mixed by passing through an active mixer that consisted of several small PTFE coated stirrer bars
inside a glass omnifit column placed on a magnetic stirrer plate.17 Upon exiting the mixer, the flow
stream entered the liquid-liquid separation vessel. The position of the liquid-liquid interface was
monitored using a webcam in conjunction with an OpenCV hue filtering algorithm in the main
Python control script.18 This was indirectly achieved by using a green plastic 'float' that had a density
in between that of the aqueous and organic phases. When the position of the interface moved
6

ACCEPTED MANUSCRIPT
below a lower set bound, the 2-Way Valve would be opened so that the aqueous phase could leave
the separation vessel through the upper outlet (the path of least resistance) causing the interface to
rise. When the interface reached a set upper bound, the 2-Way Valve would close again, forcing the
dense organic phase to leave through the lower outlet. Experimentation with plastic mixtures
identified an approximately 2.9 : 1 mixture of polyethyelene (from the plunger of a disposable
syringe) and polyacetal (from a green Keck clip) as appropriate for a range of biphasic water-
dichloromethane-isopropanol combinations. Despite the fact that isopropanol solutions were, for
short bursts, injected into the flow stream in the absence of dichloromethane (which forms the
dense organic phase), no adverse effects on the operation of the liquid-liquid extraction process
were detected. The organic layer remained the most dense and the float remained at the liquid-
liquid interface throughout the entire run. It is likely, due to dispersion effects, that the organic flow
stream entering the liquid-liquid extraction stage retained
dichloromethane at all times.
a significant proportion of
From experimental runs using coloured dyes, it was determined that a flow time of 30 minutes for
Piston Pump 1, after the acid chloride had been injected, would be sufficient to ensure that all
product/reagent had cleared the system. During this time, product solutions were collected into 65
mL beakers, with approximately 38 mL of outlet solution being collected for each run. At the end of
each reaction, a second flushing procedure was necessary to clear the remaining residual starting
material from the tubing between 3-Way Valves 2 and 4. It should be pointed out that, during this
final flushing step, and also during the filling of Holding Loop 1, the 2-Way Valve controlling the
aqueous outlet was held in an open position as the lower organic outlet was closed off (at 3-Way
Valve 4) during these stages. The total time for each run, including time taken to change the
autosampler position, switch valves and flush at the end of the run, was about 60 minutes. A step-
by-step graphical description of the flow sequence for the reaction runs is provided in the
Supporting Information, which also includes the Python source code for the scripts running on the
control computer and the Raspberry Pi.
At the end of the series of reactions, the benzamide products were isolated by removing the solvent
under reduced pressure (using a rotary evaporator followed by further drying with a 2-stage rotary
vane pump). As the agitation/mixing and separation stages of the liquid-liquid extraction process
occur in different vessels, the lower organic layer did not contain any visible droplets of the aqueous
phase. For this reason, it was not necessary to use a dessicant (e.g. magnesium sulfate), which would
typically be required when using a separating funnel. Any dissolved water that may have been
present was removed along with the solvent under vacuum. The results of the reactions are shown
in Scheme 1. The products were obtained in high yield and with high levels of purity (as indicated by
1H and 13C NMR, spectra are provided in the Supporting Information section).
7

ACCEPTED MANUSCRIPT
Scheme 2. Results of automated flow chemical benzamide formations.
In conclusion, several open-source hardware and software technologies were used in the
development of a low-cost automated synthesis platform. Whereas our previous system automated
only the selection and loading of starting materials, the inclusion of an additional motor-actuated
valve and holding loop also facilitated the automation of product collection. This was then used to
effect the flow chemical synthesis of a series of benzamides. An inline computer-vision controlled
liquid-liquid extraction stage, using aqueous phosphoric acid, effectively removed excess amine and
DMAP from the reaction, affording the amide products in high yields and purities. We are currently
working on expanding the functionality of the system and will report our findings in due course.
Experimental Section
NMR spectra were recorded in CDCl3 solutions at room temperature with a Bruker Ascend/Avance III
1
400 MHz spectrometer (400 MHz for H, 101 MHz for 13C). CDCl3 was purchased from Cambridge
1
Isotopes. The H and 13C chemical shifts are reported relative to tetramethylsilane (0.0 ppm).
Coupling constants (J-values) are reported in Hz. Protons coupling to two other non-identical protons
but with identical coupling constants are generally referred to as doublets of doublets, rather than
triplets (or apparent triplets). TLC analysis was carried out using silica gel impregnated with
8

ACCEPTED MANUSCRIPT
fluorescent indicator on aluminium backed plates. Plates were visualized either by UV fluorescence
(254 nm) or by staining with acidic vanillin solution or alkaline potassium permanganate solution.
Dichloromethane and isopropanol were purchased from VWR and used as supplied. Benzoyl
chloride, DMAP and the amine starting materials 2a-j were purchased from Sigma-Aldrich, Alfa-
Aesar, Acros, Fluorochem or Maybridge. A Knauer Azura P 4.1S pump was used to pump
dichloromethane (piston pump 1). This was connected to the control computer using a Newlink USB-
0039DBL USB-to-RS232 adapter (purchased from CPC) (via a crossover /‘null modem’ serial cable). A
Jasco PU 980 HPLC pump was used to pump the aqueous phosphoric acid solution. The syringe
pumps used were NE-1000 (New Era) and were purchased from World Precision Instruments. These
were connected to the control computer through Newlink USB-0039DBL USB-to-RS232 adapters
using custom made RS232 cables (female DB9 to RJ11-6p4c). The webcam used was a Microsoft
Lifecam Cinema. For the fluidic system FEP or PFA flow tubing was used (1.0 mm o.d 0.6 mm i.d.) and
connected using Omnifit/Diba adapters and interconnects (1/4-28-UNF thread). All dynamic
functions of the webcam (auto-white-balance, auto-focus, auto-exposure) were disabled and the
same settings were used for each run. The main Python control script ran on a Dell Studio 1558
laptop from an IDLE shell (Python 2.7.13, Anaconda distritubution 4.4.0, 64 bit). The Raspberry Pi
used was an RPi 3 model B (Raspbian Jessie Linux OS) and the Python script was run from a
command line.
Experimental procedure for the flow synthesis of amides 4a-j.
Prior to initiating the run, the position of the needle on the autosampler was manually adjusted
(with the power to the stepper motor drivers off) so that it was just above the waste position vial.
The system (see Figure 2 for a schematic) was primed with dichloromethane and aqueous
phosphoric acid by pumping through for around 30 minutes. Valves were switched to ensure
complete filling of the system. The syringe in Syringe Pump 2 was primed by manually taking in the
benzoyl chloride solution. The control script was activated with the 2-Way Valve (aqueous out) in the
closed position. Once the system appeared to be operating normally, the reaction sequence was
initiated by pressing the appropriate key on the keyboard. When all products had been collected,
(approximately 10 hours), the products were isolated by removing the solvent under reduced
pressure. The main control script and the script running on the Raspberry Pi are provided in the
Supporting Information section.
N-Benzylbenzamide (4a)
N-Benzylbenzamide (4a) was synthesised using the general procedure outlined above, using benzoyl
chloride 1 and N-benyzlamine 2a. Yield = 95% (78 mg from 3 mL of 0.13 M benzoyl chloride). 1H NMR
(400 MHz, CDCl3) δ 7.78 (d, J = 8.5, 2H), 7.45 (m, 1H), 7.41 – 7.33 (m, 2H), 7.33 – 7.21 (m, 5H), 6.80 (s,
1H), 4.58 (d, J = 5.7 Hz, 2H). 13C NMR (101 MHz, CDCl3) δ 167.38, 138.22, 134.32, 131.41, 128.64,
128.45, 127.76, 127.43, 126.95, 43.96. The data are consistent with those reported previously in the
literature.19
N-Amylbenzamide (4b)
N-Amylbenzamide (4b) was synthesised using the general procedure outlined above, using benzoyl
chloride 1 and N-benyzlamine 2b. Yield = 94% (70 mg from 3 mL of 0.13 M benzoyl chloride). 1H NMR
(400 MHz, CDCl3) δ 7.78 (d, J = 7.2 Hz, 2H), 7.46 (m, 1H), 7.38 (t, J = 7.2 Hz, 2H), 6.65 (s, 1H), 3.41 (dt,
9

ACCEPTED MANUSCRIPT
J = 7.2, 6.0 Hz, 2H), 1.67 – 1.50 (m, 2H), 1.40 – 1.22 (m, 4H), 0.89 (t, J = 7.0 Hz, 3H). 13C NMR (101
MHz, CDCl3) δ 167.51, 134.79, 131.09, 128.33, 126.81, 40.00, 29.23, 29.05, 22.27, 13.85. The data
are consistent with those reported previously in the literature.20
N-(2-Furfuryl)benzamide (4c)
N-(2-Furfuryl)benzamide (4c) was synthesised using the general procedure outlined above, using
benzoyl chloride 1 and furfurylamine 2c. Yield = 93% (73 mg from 3 mL of 0.13 M benzoyl chloride).
1H NMR (400 MHz, CDCl3) δ 7.78 (d, J = 7.3 Hz, 2H), 7.48 (m, 1H), 7.43 – 7.31 (m, 3H), 6.79 (s, 1H),
6.31 (m, 1H), 6.26 (d, J = 2.9 Hz, 1H), 4.60 (d, J = 5.5, 2H). 13C NMR (101 MHz, CDCl3) δ 167.24,
151.24, 142.14, 134.12, 131.47, 128.44, 126.98, 110.42, 107.52, 36.91. The data are consistent with
those reported previously in the literature.21
tert-Butyl-2-benzoylhydrazinecarboxylate (4d)
tert-Butyl-2-benzoylhydrazinecarboxylate (4d) was synthesised using the general procedure outlined
above, using benzoyl chloride 1 and tert-butyl-2-hydrazinecarboxylate 2d. Yield = 89% (82 mg from 3
mL of 0.13 M benzoyl chloride). 1H NMR (400 MHz, CDCl3) δ 9.12 (s, 1H), 7.81 (d, J = 7.4 Hz, 2H), 7.44
(t, J = 7.4 Hz, 1H), 7.33 (t, J = 7.4 Hz, 2H), 7.17 (s, 1H), 1.44 (s, 9H). 13C NMR (101 MHz, CDCl3) δ
166.85, 155.71, 132.23, 131.81, 128.65, 127.25, 82.03, 28.13.
N-(3-Phenyl)propylbenzamide (4e)
N-(3-Phenyl)propylbenzamide (4e) was synthesised using the general procedure outlined above,
using benzoyl chloride 1 and (3-phenyl)propylamine 2e. Yield = 93% (87 mg from 3 mL of 0.13 M
benzoyl chloride). 1H NMR (400 MHz, CDCl3) δ 7.70 (dd, J = 8.3, 1.3 Hz, 2H), 7.42 (m, 1H), 7.32 (t, J =
7.5 Hz, 2H), 7.24 (t, J = 7.2 Hz, 2H), 7.15 (m, 3H), 6.80 (s, 1H), 3.42 (dt, J = 7.0, 7.0 Hz, 2H), 2.71 – 2.52
(m, 2H), 1.99 – 1.75 (m, 2H). 13C NMR (101 MHz, CDCl3) δ 167.43, 141.45, 134.66, 131.25, 128.50,
128.43, 128.35, 126.78, 125.99, 39.78, 33.48, 31.07. The data are consistent with those reported
previously in the literature.22
N-(2-Thiophenemethyl)benzamide (4f)
N-(2-Thiophenemethyl)benzamide (4f) was synthesised using the general procedure outlined above,
using benzoyl chloride 1 and (2-thiophenemethyl)amine. Yield = 92% (78 mg from 3 mL of 0.13 M
1
benzoyl chloride. H NMR (400 MHz, CDCl3) δ 7.67 (dd, J = 8.0, 1.2 Hz, 2H), 7.33 (m, 1H), 7.23 (t, J =
8.0 Hz, 2H), 7.18 (s, 1H), 7.07 (dd, J = 5.1, 1.0 Hz, 1H), 6.85 (dd, J = 3.5, 1.0 Hz, 1H), 6.79 (dd, J = 5.1,
3.5 Hz, 1H), 4.61 (d, J = 5.7 Hz, 2H). 13C NMR (101 MHz, CDCl3) δ 167.15, 140.81, 134.14, 131.56,
128.53, 126.98, 126.91, 126.14, 125.24, 38.78. The data are consistent with those reported
previously in the literature.23
10

ACCEPTED MANUSCRIPT
N-Butylbenzamide (4g)
N-Butylbenzamide (4g) was synthesised using the general procedure outlined above, using benzoyl
chloride 1 and N-butylamine 2g. Yield = 91% (63 mg from 3 mL of 0.13 M benzoyl chloride). 1H NMR
(400 MHz, CDCl3) δ 7.78 – 7.73 (m, 2H), 7.48 (m, 1H), 7.45 – 7.37 (m, 2H), 6.23 (s, 1H), 3.65 – 3.24 (m,
2H), 1.66 – 1.54 (m, 2H), 1.48 – 1.34 (m, 2H), 1.0 – 0.8 (m, 3H). 13C NMR (101 MHz, CDCl3) δ 167.50,
134.91, 131.24, 128.50, 126.80, 39.78, 31.74, 20.13, 13.74. The data are consistent with those
reported previously in the literature.24
N-(4-Phenyl)butylbenzamide (4h)
N-(4-Phenyl)butylbenzamide (4h was synthesised using the general procedure outlined above, using
benzoyl chloride 1 and (4-phenyl)butylamine 2h. Yield = 96% (95 mg from 3 mL of 0.13 M benzoyl
1
chloride). H NMR (400 MHz, CDCl3) δ 7.73 (m, 2H), 7.45 (m, 1H), 7.38 (m, 2H), 7.29 – 7.22 (m, 2H),
7.20 – 7.12 (m, 3H), 6.40 (s, 1H), 3.43 (dt, J = 6.5, 6.5 Hz, 2H), 2.63 (t, J = 7.3 Hz, 2H), 1.73 – 1.56 (m,
4H). 13C NMR (101 MHz, CDCl3) δ 167.50, 142.00, 134.71, 131.20, 128.40, 128.32, 128.26, 126.80,
125.73, 39.82, 35.42, 29.16, 28.62. The data are consistent with those reported previously in the
literature.25
N-(4-Methyl)benzylbenzamide (4i)
N-(4-Methyl)benzylbenzamide (4i) was synthesised using the general procedure outlined above,
using benzoyl chloride 1 and (4-methyl)benzylamine 2i. Yield = 97% (85 mg from 3 mL of 0.13 M
benzoyl chloride). 1H NMR (400 MHz, CDCl3) δ 7.76 (d, J = 7.3 Hz, 2H), 7.45 (m, 1H), 7.35 (t, J = 7.3 Hz,
2H), 7.19 (d, J = 8.0 Hz, 2H), 7.11 (d, J = 8.0 Hz, 2H), 6.80 (s, 1H), 4.53 (d, J = 5.7 Hz, 2H), 2.31 (s, 3H).
13C NMR (101 MHz, CDCl3) δ 167.28, 137.22, 135.16, 134.45, 131.39, 129.37, 128.48, 127.86, 126.92,
43.83, 21.03. The data are consistent with those reported previously in the literature.23
N-sec-Butylbenzamide (4j)
N-sec-Butylbenzamide (4j) was synthesised using the general procedure outlined above, using
benzoyl chloride 1 and sec-butylamine 2j. Yield = 90% (62 mg from 3 mL of 0.13 M benzoyl chloride).
1H NMR (400 MHz, CDCl3) δ 7.69 (d, J = 7.0 Hz, 2H), 7.36 (m, 1H), 7.30 – 7.23 (m, 2H), 6.33 (d, J = 7.1
Hz, 1H), 4.00 (m, 1H), 1.56 – 1.35 (m, 2H), 1.11 (d, J = 6.6 Hz, 3H), 0.84 (t, J = 7.5 Hz, 3H). 13C NMR
(101 MHz, CDCl3) δ 166.87, 135.05, 131.10, 128.38, 126.77, 47.03, 29.68, 20.38, 10.34. The data are
consistent with those reported previously in the literature.26
Acknowledgements
We thank the European Union Erasmus+ scheme (JS and JVDM).
11

ACCEPTED MANUSCRIPT
References
1
2
(a) Noble, D. Forces of Production: A Social History of Industrial Automation; Transaction
Publishers, Piscataway, 2011; (b) Autor, D. H. Journal of Economic Perspectives 2015, 29, 3.
For reviews and selected examples, see: (a) Ley, S. V.; Fitzpatrick, D. E.; Ingham, R. J.; Myers,
R. M. Angew. Chem. Int. Ed. 2015, 54, 3449; (b) Capel, A. J.; Wright, A.; Harding, M. J.;
Weaver, G. W.; Li, Y. Q.; Harris, R. A.; Edmondson, S.; Goodridge, R. D.; Christie, S. D. R.
Beilstein J. Org. Chem. 2017, 13, 111; (c) Monaghan, T.; Harding, M. J.; Harris, R. A.; Friel, R.
J.; Christie, S. D. R. Lab Chip 2016, 16, 3362; (d) Fitzpatrick, D. E.; Ley, S. V. Reac. Chem. Eng.
2016, 1, 629; (e) Ley, S. V.; Fitzpatrick, D. E.; Myers, R. M.; Battilocchio, C.; Ingham, R. J.
Angew. Chem. Int. Ed. 2015, 54, 10122; (f) Kitson, P. J.; Glatzel, S.; Chen, W.; Lin, C. G.; Song,
Y. F.; Cronin, L. Nature Protocols 2016, 11, 920; (g) Kitson, P. J.; Symes, M. D.; Dragone, V.;
Cronin, L. Chem. Sci. 2013, 4, 3099; (h) Ley, S. V.; Ingham, R. J.; O'Brien, M.; Browne, D. L.
Beilstein J. Org. Chem. 2013, 9, 1051; (i) Machida, K.; Hirose, Y.; Fuse, S.; Sugawara, T.;
Takahashi, T. Chem. Pharm. Bull. 2010, 58, 87; (j) Masui, H.; Yosugi, S.; Fuse, S.; Takahashi, T.
Beilstein J. Org. Chem. 2017, 13, 106; (k) Ingham, R. J.; Battilocchio, C.; Hawkins, J. M.; Ley, S.
V. Beilstein J. Org. Chem. 2014, 10, 641.
3
4
(a) G. van Rossum, Python tutorial, Technical Report CS-R9526, Centrum voor Wiskunde en
Informatica (CWI), Amsterdam, May 1995. https://www.python.org/ (accessed February
2018); (b) Oliphant, T. E. Comp. Sci. Eng. 2007, 9, 10.
(a) Cressey, D. Nature 2017, 544, 125; (b) https://www.arduino.cc/ (accessed February
2018).
5
6
7
http://www.picaxe.com/ (accessed February 2018).
https://micropython.org/ (accessed February 2018).
(a) https://www.raspberrypi.org/ (accessed February 2018); (b) Severance, C. Computer
2013, 46, 14.
8
9
https://beagleboard.org/ (accessed February 2018).
For a well documented example of the use and development of open-source software to
control and interface to a range of flow chemical apparatus, see: Ingham, R. J. Octopus,
version 0.1, 2014. https://github.com/richardingham/octopus/ (accessed February 2018).
(a) http://reprap.org/ (accessed February 2018); (b) https://hackaday.io/ (accessed February
2018); (c) https://www.hackster.io/ (accessed February 2018).
10
11
12
O'Brien, M.; Konings, L.; Martin, M.; Heap, J. Tetrahedron Lett. 2017, 58, 2409.
(a) Wegner, J.; Ceylan, S.; Kirschning, A. Adv. Synth. Catal. 2012, 354, 17; (b) Webb, D.;
Jamison, T. F. Chem. Sci. 2010, 1, 675; (c) Wiles, C.; Watts, P. Chem. Commun. 2011, 47,
6512; (d) Knowles, J. P.; Elliott, L. D.; Booker-Milburn, K. I. Beilstein J. Org. Chem. 2012, 8,
2025; (e) Hessel, V.; Kralisch, D.; Kockmann, N.; Noel, T.; Wang, Q. ChemSusChem 2013, 6,
746; (f) Gutmann, B.; Cantillo, D.; Kappe, C. O. Angew. Chem. Int. Ed. 2015, 54, 6688; (g)
McQuade, D. T.; Seeberger, P. H. J. Org. Chem. 2013, 78, 6384; (h) Brzozowski, M.; O'Brien,
M.; Ley, S. V.; Polyzos, A. Acc. Chem. Res. 2015, 48, 349; (i) Pastre, J. C.; Browne, D. L.; Ley, S.
V. Chem. Soc. Rev. 2013, 42, 8849; (j) Ley, S. V. Chemical Record 2012, 12, 378; (k) Polyzos,
A.; O'Brien, M.; Petersen, T. P.; Baxendale, I. R.; Ley, S. V. Angew. Chem. Int. Ed. 2011, 50,
1190; (l) O'Brien, M.; Baxendale, I. R.; Ley, S. V. Org. Lett. 2010, 12, 1596; (m) O'Brien, M.;
Taylor, N.; Polyzos, A.; Baxendale, I. R.; Ley, S. V. Chem. Sci. 2011, 2, 1250.
Day, C.; Saldarriaga, A.; Tilley, M.; Hunter, H.; Organ, M. G.; Wilson, D. J. Org. Proc. Res. Dev.
2016, 20, 1738.
13
14
(a) Ragnarsson, U.; Grehn, L. Acc. Chem. Res. 1998, 31, 494; (b) Neises, B.; Steglich, W.
Angew. Chem. Int. Ed. 1978, 17, 522.
15
16
https://www.pololu.com/product/1182 (accessed February 2018).
http://reprap.org/wiki/RAMPS_1.4 (accessed February 2018).
12

ACCEPTED MANUSCRIPT
17
18
Gugger, R. E.; Mozersky, S. M., US 4054270, 1977; Vol. 4054270.
(a) O'Brien, M.; Koos, P.; Browne, D. L.; Ley, S. V. Org. Biomol. Chem. 2012, 10, 7031; (b)
O'Brien, M.; Cooper, D. A.; Dolan, J. Tetrahedron Lett. 2017, 58, 829; (c) O'Brien, M.; Cooper,
D. Synlett 2016, 27, 164; (d) O'Brien, M.; Cooper, D. A.; Mhembere, P. Tetrahedron Lett.
2016, 57, 5188; (e) Bourne, S. L.; O'Brien, M.; Kasinathan, S.; Koos, P.; Tolstoy, P.; Hu, D. X.;
Bates, R. W.; Martin, B.; Schenkel, B.; Ley, S. V. ChemCatChem 2013, 5, 159.
Lim, D. S. W.; Lew, T. T. S.; Zhang, Y. Org. Lett. 2015, 17, 6054.
Gaspa, S.; Porcheddu, A.; De Luca, L. Org. Biomol. Chem. 2013, 11, 3803.
Gockel, S. N.; Hull, K. L. Org. Lett. 2015, 17, 3236.
Nack, W. A.; He, G.; Zhang, S.-Y.; Lu, C.; Chen, G. Org. Lett. 2013, 15, 3440.
Li, F.; Ma, J.; Lu, L.; Bao, X.; Tang, W. Cat. Sci. Technol. 2015, 5, 1953.
Kaicharla, T.; Thangaraj, M.; Biju, A. T. Org. Lett. 2014, 16, 1728.
19
20
21
22
23
24
25
26
Häring, A. P.; Biallas, P.; Kirsch, S. F. Eur. J. Org. Chem. 2017, 2017, 1526.
Kraiem, J.; Ollevier, T. Green Chem. 2017, 19, 1263.
13