A method and apparatus for RNA synthesis

Abstract

A method of RNA synthesis comprising; introducing into a first microfluidic reactor module 10 via a plurality of inlet ports 14, a plurality of reactants including, a nucleoside triphosphate (NTP), DN

Description

Field of invention The present invention relates to a method of RNA synthesis using a microfluidic device and method. Background art Flow reactors, alternatively termed continuous reactors, provide for a continuous flow of materials or reactance and enable different experimental configurations with regard to mixing properties, temperature control, monitoring, residence times etc. Example prior art flow reactors are typically formed as an assembly of individual modules that are connected face-to-face to form a unitary block through which a fluid is directed. Example flow reactors are described in WO 2013 / 050764. Micro-manufacturing offers more sustainable, flexible and effective reaction systems configurable as micro-factories or downsized production systems that decrease the use factory floor space, reduce energy consumption and improve resource utilisation. Accordingly, it is possible to achieve a sustainable growth with better environmental impact, cost effectiveness and flexibility than current manufacturing systems. In the bio-chemical sector it is increasingly common to perform processes on Lab-on-chip (LOC) devices. Such LOC devices use micro-manufacturing techniques to miniaturize and integrate laboratory analysis and functionalities into a small chip. These small devices as well as using less space, consuming fewer samples and producing less waste bring faster response times and easier operation. Many functionalities can be integrated into a LOC device such as microchannels, filters, cell sorters, mixers, DNA or RNA synthesis. It is important that LOC devices are formed from reaction-inert materials and are preferably transparent, in order to enable visual inspection of the reaction in real-time in the chip. Also, a device that is transparent, enables the possibility for spectroscopic analysis and investigations. The most common material used for such applications is PDMS (Polydimethylsiloxane). In 2019, the World Health Organization identified ten worldwide threats to the global health. Eight of these ten threats are dealing with pandemics and accessibility to vaccines especially in LMIC (low- and medium-income countries) where pandemics such as Ebola or Dengue have / are impacting population. Therefore, there is an urgent need for apparatus and methods suitable for use in the manufacturing of vaccines and vaccine pre-cursors that can meet the demands of large quantities, low cost and rapid bulk synthesis. In particular there is a need for a chemical compound manufacturing platform that can be used to meet urgent vaccine demand i.e., within a couple of days, is easy to store and deploy around the world, and is highly productive and cost effective to run. Summary of the Invention Accordingly it is an objective of the present invention to provide a modular manufacturing platform capable of synthesis and / or filtering of chemical compounds and biological and non-biological molecules including in particular RNA and subsequently vaccines. It is a specific objective to provide apparatus and a system configurable as a micro factory for biological molecule synthesis to enable downstream vaccine manufacturing. It is a specific objective to provide apparatus and method forming a microfluidic integrated system comprising component modular parts including microfluidic bioreactor modules and microfluidic filtration modules interconnected via respective conduits / channels to create a fluid flow pathway that may be operated continuously and may be automated using electronic and / or software control utilities involving the use of membranes, component actuators such as pumps, valves, gates, delivery ports, outlet ports, syringe pumps, sensors etc. as will be appreciated. It is a further specific objective to provide a microfluidic system for the reaction synthesis of selected chemical components such as biological and non-biological molecules via a process that may be continuous or non-continuous and that may be automated or semiautomated. It is a further specific objective to provide a modular system in which individual modular units may be interconnected depending upon the synthetic pathway. Such units may comprise ports, valves and suitable connections to enable the fluid communication and interconnection between respective modules. It is a further specific objective to provide a reaction system for a fluid configurable to measure and be responsive to characteristics of the system such as pressure, temperature, pH, a volume or ratio of one or more selected chemical components of the fluid, a flow rate of the fluid and a reaction status of the fluid. It is a further specific objective to provide a microfluidic filtration apparatus and method in which chemical components of a fluid may be separated within an automated or semiautomated fluidic flow system. It is an objective to provide a filtration system that may be operated continuously via suitable electronic control and fluid delivery, recirculation and / or driving via fluid pressure. It is a further specific objective to provide a microfluidic system having one or a plurality of reactor modules and filtration modules interconnected to form a network. It is a further objective to provide a method and apparatus for the synthesis of RNA. It is a further specific objective to provide apparatus and method for the synthesis of RNA from DNA. It is a yet further objective to provide a form of synthetic RNA that may be used for the preparation of a vaccine. According to a first aspect of the present invention there is provided a method of RNA synthesis via a continuous flow process comprising: introducing into a first microfluidic reactor module via at least one inlet a plurality of reactants comprising: a nucleoside triphosphate (NTP), DNA, a salt solution and RNA polymerase; driving a flow of a fluid comprising the reactants through an elongate reaction channel of the first microfluidic reactor module; allowing at least some of the reactants to react within the elongate reaction channel; filtering reaction products of the reactants using a first filtration module; and recirculating at least some of the reactants output from an outlet of the filtration module to a region of the at least one inlet of the first reactor module via at least one recirculation conduit. Optionally, the method of RNA synthesis comprises at least some of the reactants comprises recirculating the RNA polymerase from an outlet of the first filtration module to an inlet region of the first reactor module. Optionally, the steps of filtering reaction products and recirculating at least some of the reactants comprises using a gate to filter DNA output from an outlet end of the elongate reaction channel and recirculating the DNA to an inlet region of the first reactor module. Optionally, the method of RNA synthesis comprises delivering at least some of the filtered reaction products from the first filtration module into a second microfluidic reactor module in combination with inputting a capping enzyme into the second reactor module. Optionally, the method of RNA synthesis comprises delivering a fluid output from the second reactor module to a second microfluidic filtration module. Optionally, the method of RNA synthesis comprises recirculating any unreacted NTP to an inlet region of the first reactor module and recirculating any unreacted capping enzyme to the inlet region of the second reactor module. Optionally, the salt solution comprises MgCh. Optionally, the RNA polymerase comprises T7 polymerase. According to a further aspect of the present invention there is provided RNA prepared by the method as claimed herein. According to a further aspect of the present invention there is provided use of RNA prepared by the method of any preceding claim in the preparation of a vaccine. According to a other aspects microfluidic apparatus is described herein for processing a fluid comprising: a first reactor module having a reaction flow channel or well, at least one inlet and at least one outlet; a first filtration module having a fluidic filtration region, at least one inlet and at least one outlet, said inlet provided in fluid communication with the outlet of the first reactor module. Optionally, the apparatus comprises a second reactor module having a reaction flow channel or well, at least one inlet and an outlet, said inlet provided in fluid communication with the outlet of the first filtration module. Optionally, the apparatus comprises a second filtration module having a fluidic filtration region, at least one inlet and an outlet, said inlet provided in fluid communication with the outlet of the second reactor module. Optionally, the apparatus comprises a fluid injection port provided in fluid communication with an inlet region of the first filtration module. Optionally, the apparatus comprises a fluid injection port provided in fluid communication with an inlet region of the second reactor module. Optionally, the apparatus comprises a fluid injection port provided in fluid communication with an inlet region of the second filtration module. Optionally, the apparatus comprises a first recirculation conduit extending between a region of the outlet of the first reactor module and a region of the at least one inlet of the first reactor module. Optionally, the apparatus comprises a second recirculation conduit extending between a region of the outlet of the first filtration module and a region of the at least one inlet of the first reactor module. Optionally, the apparatus comprises a third recirculation conduit extending between a region of the outlet of the second filtration module and a region of the at least one inlet of the first reaction module. Optionally, the apparatus comprises a plurality of inlet ports to allow the input of fluidic chemical component into the reaction channel or well. Optionally, the apparatus comprises at least one pump coupled to the at least one inlet of the first reactor module to drive fluid flow through the reaction channel or within the well. Optionally, the apparatus comprises at least one, valve, fluid flow gate, fluid flow port, heating element, fluid storage or retention reservoir provided in fluid communication with the fluid within the apparatus. Optionally, the pump comprises a syringe pump. Optionally, the apparatus has a plate-like structure such that the first reactor module and the first filtration module are formed at least in part, as channels or recessed grooves provided on or internally at the plate-like structure. Optionally, the apparatus comprises a plurality of sensors positioned at different fluid flow regions of the apparatus. Optionally, the sensors comprise any one or a combination of at least one temperature sensor, at least one flow rate sensor, at least one pressure sensor, at least one pH sensor, at least one flow volume sensor, at least one spectroscopic sensor, at least one optical sensor, at least one optical fibres or spectroscopic optic fibre. Optionally, the apparatus comprises a control unit coupled to the at least one pump, valve, fluid flow gate, fluid flow port, heating element, fluid storage or retention reservoir and the sensors to control a characteristic of the fluid flow within the apparatus in response to a status of a physical, chemical or mechanical characteristic of the fluid determined by the sensors. Optionally, the control unit comprises a CPU, a processor, a PCB, a PLC, a handheld electronic device. Optionally, the control unit comprises a control module comprising any one or a combination of the following: software; electronic components; a data storage utility; wired or wireless communication modules and / or ports; a visual display output; a user interface; at least one actuator to actuate any one or a combination of the at least one pump, valve, fluid flow gate, fluid flow port, heating element, fluid storage or retention reservoir and the sensors. According to other aspects a microfluidic system is described herein for processing a fluid comprising: a plurality of the apparatus as described herein, a final fluid flow outlet of each of the apparatus being coupled to a collection unit to combine the fluid output from each of the apparatus. Optionally, the microfluidic system comprises control unit coupled to the at least one pump, valve, fluid flow gate, fluid flow port, heating element, fluid storage or retention reservoir and the sensors of each or a selection of the apparatus to control a characteristic of the fluid flow within the apparatus in response to a status of a physical, chemical or mechanical characteristic of the fluid determined by the sensors. Optionally, the control unit comprises a CPU, a processor, a PCB, a PLC, a handheld electronic device. Optionally, the control unit comprises a control module comprising any one or a combination of the following: software; electronic components; a data storage utility; wired or wireless communication modules and / or ports; a visual display output; a user interface; at least one actuator to actuate any one or a combination of the at least one pump, valve, fluid flow gate, fluid flow port, heating element, fluid storage or retention reservoir and the sensors. According to other aspects a method of processing a fluid using microfluidic apparatus is described herein comprising: introducing at least one fluid into a first reactor module via at least one inlet and allowing the fluid to flow through or within a reaction flow channel or well and outputting the fluid at an outlet of the first reactor module; driving at least a portion of the fluid from said outlet through a filtration module having a fluid filtration region via at least one inlet and outputting the fluid from the first filtration module at an outlet; wherein a reaction occurs in the first reactor module between at least two chemical components of the fluid and a filtration of the fluid occurs in the first filtration module. Optionally, the method comprises driving fluid flow through the apparatus using at least one fluidic pump. Optionally, the method comprises recirculating at least a portion of the fluid from the outlet of the reactor module and / or the filtration module to a region of the at least one inlet of the reactor module via at least one recirculation conduit. Optionally, the method comprises monitoring a status of a chemical reaction occurring within the fluid between the chemical components using at least one sensor. Optionally, the method comprises controlling fluid flow through the apparatus in response to the status of the chemical reaction identified by the sensor. Optionally, the method comprises monitoring a physical, chemical and / or mechanical characteristic of the fluid within the apparatus using at least one sensor. Optionally, the method comprises controlling fluid flow within the apparatus in response to the identified physical, chemical and / or mechanical characteristic of the fluid by the sensor. Optionally, the step of controlling the fluid flow comprises controlling or deactivating at least one actuator, pump, valve, gate or port provided in fluid communication with the fluid within the apparatus. Optionally, the controlling the fluid flow comprises actuating or deactivating the at least one actuator, pump, valve, gate or port using a CPU, processor, PCB, PLC or handheld electronic device. Brief description of drawings A specific implementation of the present invention will now be described, by way of example only, and with reference to the accompanying drawings in which: Figure lisa schematic illustration of a continuous reaction and filtration apparatus suitable for RNA production; Figure 2 is a perspective view of a reactor and filtration module suitable for use in the apparatus of figure 1; Figure 3 is a plan view of a well or batch bioreactor module suitable for use in the apparatus of figure 1; Figure 4 is a plan view of a reaction channel-shaped bioreactor module suitable for use in the apparatus of figure 1; Figure 5 A is a plan view of a filtration module suitable for use in the apparatus of figure 1; Figure 5B is an image of a membrane within the filtration module of figure 5A; Figure 6A is an illustration of a part of a filtration module having a region configured for centrifugal force separation of chemical components suitable for use in the apparatus of figure 1; Figure 6B is an image of a part of a filtration module having regions for the centrifugal force separation of chemical components; Figure 7 is a vertical tangential flow filtration (VTFF) module for use in the present microfluidic apparatus; Figure 8 is a magnified view of a region of the filtration module of figure 7; Figure 9 is a plan view of a spiral tangential flow filtration module; Figure 10A is a component of a filtration module according to specific implementation; Figure 1 OB is a schematic view of further parts of the filtration module of figure 10A; Figure 10C is a further illustration of parts of the filtration module of figures 10A and 10B; Figure 11A is a plan view of a further filtration module formed as a two-layer microchannel TFF; Figure 1 IB is a schematic illustration of a filtration mechanism and filtration modules according to specific implementations; Figure 12 is a schematic view of a channel-flow bioreactor module; Figure 13 is a graph used in the calculation of the design of a microfluidic bioreactor and filtration system; Figure 14 is a schematic illustration of a reactor module coupled to a filtration module to form a microfluidic device; Figure 15 is perspective illustrations of a TFF module; Figure 16 is an image of microscopic analysis of the channel dimension of the TFF module; Figure 17 is a schematic illustration of a summary of a prototype building process; Figure 18A is a schematic illustration of a microfluidic system comprising a reactor module, a series of sensors, a reaction status monitoring arrangement and a control unit according to a specific implementation; Figure 18B is a schematic illustration of components of a reaction status monitoring arrangement suitable as a part of the microfluidic system of figure 18 A; Figure 18C is a schematic illustration of a series of sensors, a reaction status monitoring arrangement and a control unit forming part of the microfluidic system of figure 18A according to a specific implementation; Figure 19 is a schematic view of an architecture of a part of the microfluidic system of figure 18 including a part of a spectrometer arrangement; Figure 20 is a schematic illustration of a serial communication between an Arduino and a Python IDE control utility: Figure 21 is a graph of a relationship between an RNA yield and proportion of various reactants including NaCl, MgCh, NTPs etc; Figure 22 is a graph of the data analysis (RNA yield VS magnesium ion) and specific reaction time selection according to aspects of the subject invention; Figure 23 A is a schematic illustration of a velocity profile analysis result simulation; Figure 23B is a graph of a velocity analysis graph; Figure 24 is a schematic illustration of the velocity in a tangential flow filtration channel within a filtration module; Figure 25A is a graph of the pressure drop in the bioreactor channel; Figure 25B is a graph of the pressure drops as multistep down; Figure 26A is a photograph of a mask for the manufacture of a microfluidic reactor and filtration apparatus; Figure 26B is a photograph of a continuous microfluidic flow reactor mask; Figure 27 is an image of continuous flow reactor sealed with Kapton tape; Figure 28 is an image of the complete set up using two Acrylic plates; Figure 29 is an image of the 3D printed mould after use; Figure 30A is an image of cracked and broken PDMS mask Figure 30B is an image of PDMS sample cured in a foil paper recipient; Figure 31 is a photograph of an experimental microfluidic chemical compound synthesis platform; Figure 32 is a live plot spectrometer data (Intensities against wavelength) in Python; Figure 33 is an image of samples for use with a microfluidic reactor and filtration system having different concentrations (A: 25mM / L, B: 50mM / L, C:75mM / L, D:100Mm / L); Figure 34 is a graph of a UV-Vis spectral analysis; Figure 35 is an image of a live plot PH data in Python; Figure 36 are graphs of Predicted Plot (On the left is the reaction with ImM dNTP and on the right is the reaction with 4mM dNTP); Figure 37 illustrates graphs of a prediction Profiler for 1st and 2nd experiments (Upon is the reaction with ImM dNTP and below is the reaction with 4mM dNTP); Figure 38 is a summary of various data of the microfluidic bioreactor and filtration (left is the reaction with ImM dNTP and right is the reaction with 4mM dNTP LogWorth =-loglO(p-value)); Figure 39 is a schematic illustration of a microfluidic system for the processing of a fluid and synthesis of batch chemical compounds and optionally biomolecules comprising multiple bioreactor modules / units and filtration modules / units interconnected as a unitary modular system according to a specific implementation; Figure 40A is an illustration of a source code of absorbance data and real-time plot (Absorbance against Wavelength) in Python; Figure 40B is an illustration of a source code of live plot PH data in Python. Detailed description of preferred embodiment of the invention Hydrodynamics Principles in Micro Porous Channels Advances in microfabrication make it possible to build microchannels with micrometre dimensions. Since microchannels are usually integrated into these microsystems, it is important to determine the characteristics of the fluid flow in microchannels for better design of various microflow devices. The understanding of fluid behaviour in porous media, especially in microporous media that can be used for Lab on a chip design for synthetic RNA vaccine manufacturing is very limited due to the restriction of techniques. The porous medium is taken to be saturated with the fluid of interest in the sense that fluid^fluid interfaces do not form, and a single fluid prevails in the pore space. Let dp be the particle size and U, a velocity scale. (1) has been found to be applicable for Reynolds number of up to unity, namely (1) The permeability of a porous medium is a property that depends on the pore size and the pore structure. Dimensional analysis shows permeability to be a function of porosity e and particle diameter dp; each representing the pore geometry and pore size, respectively. The Carman-Kozeny relationship connects these quantities empirically but with dimensional correctness as ------- (2) 180( 1 - r j * With the particle diameter expressed in meters, permeability has dimensions of m2. In applications, one can expect to have a length scale for the device itself, say L,and the ratio defined as Da - 4 (3) which is called the Darcy number. In many applications, the particle diameter is of the order of a fraction of a millimetre while the device length scale is around a meter or beyond. Eq. 2 can be properly stated as being applicable in the limit of Da«l; Re<l. Bahrami et al. (M.Bahrami, 2006) developed a general model for prediction of pressure drop in microchannels of arbitrary cross section. Selection of the characteristic length is an arbitrary choice and will not affect the final solution. According to the model of Bahrami et al., pressure drop of laminar fully developed flow in arbitrary cross-section microchannels can be obtained from: Ar - (4) w 1 + c = where Ip* = Ip / A is the specific polar momentum of inertia of the microchannel cross section, r = 4(W+H) is the microchannel cross-section perimeter, L is the fully developed length, Q is the volumetric flow rate, e = Aspect ratio (Width / Height) of the rectangular microchannels. Reynolds number was calculated knowing the volumetric flow rate, Q, and the cross-sectional area, A, from Tt (5) Minor Losses, APmin: Other pressure losses associated with the measured pressure drop are inlet, exit, and bend losses. These losses are usually obtained from the traditional relationships used in macroscale. Phillips (ref) showed that the minor pressure losses can be obtained from: Where A and Atare the channel and connecting tube cross-sectional areas, respectively. Kb is the loss coefficient for the bend, and Kc and Ke represent the contraction and expansion loss coefficients due to area changes. Phillips recommended Kb to be approximately 1.2 for a 90 deg bend. Assuming equal cross-sectional areas for the channel and connecting tubes and maximum possible values for Kc and Ke, relative minor losses with respect to the measured pressure drop are negligible compared with the measured pressure drop. Hooman and Merrikh (M.Bahrami, 2010) have developed analytical solutions for flow and pressure drop inside porous channels as: and h, L and W are the depth, length, and width of the porous channel respectively. The cross-sectional aspect ratio, e in the samples tested in the present study is 0.5. Therefore, instead of considering the whole rectangular cross-section, the sample can be modelled as a porous medium sandwiched between two parallel plates. Choosing an appropriate cassette depends on the total sample volume, required process time, and desired final sample volume. Use the following equation to calculate the membrane area required for processing a sample in a specified time: V A= _____ JXT (8) Where, A=Membrane Area (m2) V= Volume of Filtrate generated (Litres) J= Filtrate Flux Rate (Litres / m2 / hour, LMH) T= Process time (hours) Microfluidics Simulation and Study The channel height and type of liquid affect on the performance of the microchannel. The computational fluid dynamic (CFD) FLUENT software was used for the simulation of microchannel. From the CFD, the velocity profile has been analysed for under and fully develop the region to study the fluid flow behaviour. The property of liquid influences the fluid flow inside the microchannel. We require the smallest kinematic viscosity and low surface tension that depend on select correct liquid. The microfluidic plays an important role in leading small-scale devices to integrate with many applications because of their potential in chemical and biochemical engineering. The mechanism and fundamental of single-phase fluid flow have been review using experimental data for different types of liquid and surface roughness. Therefore, in a micro scale, the impact of liquid properties such as surface tension and viscosity is dominated. [Nawi, M. N. M., Manaf, A. A., Arshad, M. R., &Sidek, O. (2013). Numerical simulation of the microchannel for the microfluidic based flow sensor. Proceedings - 2012 IEEE International Conference on Control System, Computing and Engineering, ICCSCE 2012, 345-348.} The Reynolds number is the ratio of inertial and viscous forces on fluid flowing in a channel (i.e., a ratio of the momentum of the fluid to the friction force imparted by the channel walls). A low Reynolds-number flow is a laminar or layered flow in which fluid streams flow in parallel to each other and mix only through convective and molecular diffusion. A high Reynolds-number flow is turbulent flow in which various-sized “parcels” of fluid exhibit motions that are simultaneously random in both space and time causing rapid mixing throughout the channel. The transition between the laminar and turbulent flow typically occurs at Re=2000 in internal flows [Schulte, T. H., Bardell, R. L., &Weigl, B. H. (2002). Microfluidic-technologies-in-clinical-diagnostics_2002_Clinica-Chimica-Acta. 321, 1-10] Although a turbulent flow field is more complex than a laminar one, there are significant second-order influences. One and two-dimensional models depicting the time-dependent evolution of analysis distribution under such conditions have been improved. Confocal fluorescent microscopy experiments and three-dimensional numerical modelling can help to confirm the quantitative description of reaction-diffusion processes near the walls [Schulte, T. H., Bardell, R. L., &Weigl, B. H. (2002). Microfluidic-technologies-in-clinical-diagnostics_2002_Clinica-Chimica-Acta. 321, 1-10] Sensor Integrated Platform In the study, a sensor integrated platform should be established to obtain and monitor data throughout the RNA transcription process. Gruber, P. [Gruber, P., Marques, M., Szita, N. and Mayr, T. (2017). Integration and application of optical chemical sensors in microbioreactors. Lab on a Chip, 17(16), pp.2693-2712.] proposed a flow diagram describing the path towards robust sensor integration in a microfluidic device, this diagram could be a guidance for the integration study. Various integrated platform systems have been designed and developed for real-time reaction monitoring [Wang, T., Kim, S. and An, J. (2017). A novel CMOS image sensor system for quantitative loop-mediated isothermal amplification assays to detect food-borne pathogens. Journal of Microbiological Methods, 133, pp.1-7.]. In Wang’s methodology, the complementary-metal-oxide semiconductor (CMOS) technology has shown great potential for integration and detection. CMOS image sensor works as an effective reaction and detection platform, enabling it to monitor real time photon changes as per the amplification process. The CMOS image sensor observed the photons and converted into digital units. In addition, UV-spectral studies, optical colour intensity detection and pH analysis were carried out to prove the efficiency of the CMOS [Wang, T„ Devadhasan, J., Lee, D. and Kim, S. (2016). Real-time DNA Amplification and Detection System Based on a CMOS Image Sensor. Analytical Sciences, 32(6), pp.653-658.]. Lopez-Huerta [Lopez-Huerta, F., Woo-Garcia, R., Lara-Castro, M., Estrada-Lopez, J. and Herrera-May, A. (2013). An Integrated ISFET pH Microsensor on a CMOS Standard Process. Journal of Sensor Technology, 03(03), pp.57-62.] has come up with an integrated ISFET pH microsensor on a CMOS standard process. The complete system was integrated in a 1.12 mm2 silicon area; it presented a 59 mV / pH linearity, within a concentration range of 2 to 12 of pH level, making it a good alternative for biological or medical applications. CMOS is a relevant choice, but considering the cost and time, it is difficult to achieve. More realistically, the integration scheme of each sensor has been reviewed. For pH sensor, a study of real-time feedback control of pH within microfluidics using integrated sensing and actuation [Welch, D. and Christen, J. (2014). Real-time feedback control of pH within microfluidics using integrated sensing and actuation. Lab on a Chip, 14(6), p.l 191.] has been found. The system utilized an extended-gate ion-sensitive field-effect transistor (ISFET) along with an integrated pseudo-reference electrode to monitor pH values within a microfluidic reaction chamber. For temperature senor, a serpentine-shaped temperature sensor (linewidth: 50 mm) and heater (linewidth:400 mm) were integrated beneath the centre of the reaction chamber [Sun, H., Olsen, T., Zhu, J., Tao, J., Ponnaiya, B., Amundson, S., Brenner, D. and Lin, Q. (2015). A bead-based microfluidic approach to integrated single-cell gene expression analysis by quantitative RT-PCR. RSC Advances, 5(7), pp.4886-4893.]. Choi, K., Mudrik, J. and Wheeler, A. (2015). [A guiding light: spectroscopy on digital microfluidic devices using in-plane optical fibre waveguides. Analytical and Bioanalytical Chemistry, 407(24), pp. 7467-7475.] have presented a novel method for in-plane digital microfluidic spectroscopy. In this technique, a custom manifold aligned optical fibres with a digital microfluidic device, allowing optical measurements to be made in the plane of the device, which has provided an idea for optical detection. Methodology Continuous Flow Reaction A schematic representation of the continuous RNA production process is shown in figure 1. As illustrated in figure 1, the microfluidic apparatus comprises a first bioreactor module 10, a first tangential flow filtration (TFF) module 11, a second bioreactor module 12 and a second tangential flow filtration (TFF) module 13. Each of the modules 10 to 13 are interconnected via respective channels or conduits. A plurality of fluid delivery input ports 14 are provided at the first reactor module 10 for the introduction of respective reactants into the first bioreactor 10. Each port 14 comprises a respective syringe pump 15. The first TFF comprises a respective input / inlet port 16, the second bioreactor module comprises a respective inlet / input port 17 and the second TFF module comprises a corresponding input / inlet port 18. Each of the port 16 to 18 may be provided with respective associated pumps or syringe ports for the introduction of fluid to drive fluid flow under pressure within elongate reaction channels 24 of each reactor and elongate filtration channels 25 within each respective module. A plurality of recirculating channels 19, 20, 21 interconnect various fluid flow regions (at or towards the respective inlets / outlet end regions) of the modules 10, 11, 12, 13 to provide fluid communication and return flow of reactants / reagent. The apparatus of figure 1 also comprises sensors, a reaction status monitoring device, a control unit and other associated electronic and actuator components (not shown) as described elsewhere herein so as to provide a fully automated and continuous microfluidic synthesis platform for the manufacture of chemical compounds including biological and non-biological species. The first bioreactor 10 at an outlet end of the elongate reaction channel 24 comprises a gate, 23, Gate 23 is configured to filter large molecular weight chemical components such as DNA from the fluid based on molecular size and / or molecular weight. The desired filtered fluid is then recirculated via channel 19 to an inlet region of reactor 10. This way, selected reagents and reactants can be recycled to provide a continuous or semi-continuous process. Referring to both figure 1 and figure 2, figure 2 illustrates a variation of the fluidic system of figure 1 in which reactor 10, in contrast to the elongate reaction channel 24 of figure 1, comprises a reaction well 50. As described with reference to figure 1, a plurality of injection ports 14 are coupled in fluid communication with well 50 for the delivery of the desired chemical compounds, reagents, solvents etc for the reaction,. Filtration modules 11, 12 may comprises a series of extraction ports 51, 22 enabling the extraction of selected species, reagents, solvents from the filtration module and also collection of an end product for example at port 22 such as the fully capped RNA. As will be appreciated, the specific details of the channel profiles, the use of ports, gates, inlets, outlets, pumps and fluid flow control actuators may be implemented with the arrangements as described herein. Referring to figure 39, a plurality of reactant and filtration systems as described referring to figures 1 and 2 (and indicated generally as plate-like units 10, 11, 12, 13) may be connected together to form a fully integrated modular microfluidic reaction and filtration system. In particular, suitable connection conduits provide fluid communication between the individual plate-like units 10, 11, 12, 13 such that the output of each unit (at output 22) may be combined to provide a total yield. The collective system of figure 39 may comprise a single control unit, multiple control units and respective sensors and reaction status monitoring utilities and units as described herein referring to figures 18A to 20. Using RNA synthesis as an example, during the process, a segment of DNA is copied into RNA by the enzyme RNA polymerase. The transcription mixture is composed of NTP (nucleoside triphosphate), DNA, MgCh (reacts with P2O74 to generate precipitate, affecting productivity of RNA), T7 Polymerase (catalysis), which are injected into the first bioreactor channel by syringe pumps respectively. Hydroxy naphthol blue (HNB) is used as a colorimetric indicator for the titration of Mg2 ions. Initially, the HNB and Mg2+ ions are combined in HNB-Mg2+ complexes. The decrease of Mg2+ ions results in colour change from violet to sky blue. The mixture stays in the 1st reactor 10 for approximately 6 hours, during which the RNA transcription reaction is going on all the time and RNA begins to be produced, accompanied with magnesium pyrophosphate precipitate (Mg2P2O?). All the above components except DNA flow through the first tangential flow filtration (TFF) module 11, which has a large molecular weight (MW) cut-off membrane. The DNA is filtered back to the original inlet port 14(2) via recirculation channel 19 and kept in the 1st reactor, while others are filtered to next stage. After the first TFF stage, the above components are permeated downstream with the T7 Polymerase recirculated into inlet port 14(4). Then the m7G methyltransferase (5’ cap) is injected into a port 17 and flow into the second reactor 12 and channel 24, where the contents stay and react for 2 hours. The permeate enters the second TFF unit 13, which has a medium MW cut-off membrane. In this TFF stage, unreacted NTPs and 5’ caps are retained and recirculated (via 21 and 20) into the bioreactor 12 and / or 10. Finally, purified RNA is extruded at the outlet 22. Simultaneously, other components like Mg2PPb water, salts, are collected in another vessel or port 51 and optionally discarded or onward-processed. The output (capped RNA) may then be processed via further processing procedures (not described herein but familiar to those skilled in the art) to yield a desired vaccine. An objective is to achieve non-stop reaction and higher throughput of RNA. It is necessary to add the raw materials into the reactor via parts 14 to compensate for the consumed molecules in order to maintain a continuous reaction of the process, which could be realised by regulating the actuator syringe pumps, and the feedback loop of material flow. Substantial improvements in RNA yield outputs might be possible by optimizing the reaction conditions (such as magnesium concentration, pH, temperature, reaction time). SBCE Methodology Lean product development methodology has been chosen as the main framework to develop the design. Lean product development methodology is Set-Based Concurrent Engineering (SBCE). SBCE is a process where the product that wants to be developed is divided into different subsystems, allowing to develop a set of possible solutions for each of the subsystems in parallel. As the design evolves, the set of solutions for each subsystem gets narrowed down by knowledge-based decisions, using tools such as simulations, prototyping, testing or other gained knowledge. Following the SBCE methodology, the first essential step is to divide the product in different subsystems that can be developed individually. Considering the conceptual design of the systems that needs to be developed, there are four main identified processes that could be grouped in just two functionalities that will be developed: The bioreactor and the filtration process. Simultaneously, the system needs to be integrated with a control system, capable of controlling the pressure from the pumps as well as having sensors in strategic parts of the product in order to measure key parameters. Having mapped a conceptual design of the system, it has been possible to define different subsystems that can be developed in parallel. Table 1 shows the different subsystems, with a brief description and the corresponding level of innovation that has been defined for each subsystem. Table 1 The subsystems of the conceptual design Subsystem Innovation Level Description Bioreactor 1 (10-fig 1) Medium The considered volume for the bioreactor is 1 mL. There is one feeding channel for each of the raw materials, that goes into the reactor. The reaction needs to follow a continuous flow. Filter 1 (11 - fig 1) High Several filters must be designed and developed in order to increase the RNA yield and to reuse some of the raw materials. This filter needs to separate and recirculate T7 Polymerase to the rest of the fluid. Bioreactor 2 (12-fig 1) Medium The reaction that goes on in the second reactor, concerns RNA and M7G Methyltransferase. It is essential that this reactor keeps the continuous flow of the process. Filter 2 (13-fig 1) High The second filter needs to achieve an RNA purity up to 99%. The M7G Methyltransferase needs to be cut off and taken and reused in the second reactor. Control / Sensing system High A control system needs to be integrated with the whole system in order to support the different sensor / actuator devices. Subsystem Comparison and Selection Process Bioreactor 5 Regarding the bioreactor, two different designs have been developed. The first design is inspired by the conventional batch process used in chemistry for RNA synthesis. The design is compound by four inputs 14 for each of the raw materials, that are connected to a main 1 mL vessel where the reaction happens after the residence time has gone by. Once the reaction is done, the product goes out of the main vessel into the output channel where 10 it can be tested with the sensors or sent to the next step of the process. Figures 2 and 3 show a design for a bioreactor. The complexity that this design adds to the system is that it doesn’t work in a fully continuous way. This design needs a complex and well-coordinated control system with some type of gate that keeps all the raw materials in the main vessel while the reaction is happening. After the residence time is finished, the result product needs to be fed into the rest of the system. The second design for the bioreactor is inspired by a paper from Lawrence, [J., O’Sullivan, B , Lye, G.J., Wohlgemuth, R , Szita, N. (2013) Microfluidic multi-input reactor for biocatalytic synthesis using transketolase. Journal of Molecular Catalysis: Enzymatic. Published by Elsevier, B. V. ], which shows a reactor in the shape of a channel. This design tackles the issues that the previous design has, related to the continuous flow. In this design there are still four inputs for the raw materials, but they are fed into a channel. The channel needs to be designed in a way that with the appropriate pressure at the inputs, the raw materials get mixed and react in the channel while they are flowing through. This way, the product reaches the end of the channel-shaped bioreactor after the residence time, goes into a small vessel designed for sensing different parameter, and continues flowing through to the next steps of the process. Figure 4 shows a design of the channel-shaped bioreactor. In conclusion, the batch-type reactor needs a more complex integrated control system whereas the channel-shaped reactor needs more calculations and is harder to design but it would blend in better to the continuous flow process. Filtration There are several types of filtration, but generally, when we talk about filtration, there are two main types: Direct flow filtration (DFF) and tangential flow filtration (TFF). DFF consists in having the main flow perpendicular to the filter whereas TFF has a flow in a parallel direction to the filter. DFF has a much higher risk of clogging and it is demonstrated that while increasing the filtered volume, the filtrate flux rate is higher for the TFF that for the DFF. Moreover, typical applications for TFF are concentration, diafiltration and fractionation of biomolecules, as well as clarification and removal of cells; which are similar applications to the one present in this project. In other words, for this project’s application, TFF is the most effective option and consequently, the chosen type of filtration. For the TFF process, several design solutions have been developed and evaluated using mainly literature review, although other aspects such as manufacturability or possibility of integration with the whole system have been taken into consideration for the decision making. The first TFF design solution consists in a serpentine-shaped path with two channels at the same height, separated by the filtration membrane. This idea is inspired by X. Chen et al. (2007) [Microfluidic Chip for Blood Cell Separation and Collection Based on Crossflow Filtration. Sensors and Actuator B 130 (2008) (p.216-p.221). China.], who explain that this TFF design is a feasible separation method that has rapid separating time, high reproducibility and is low cost compared to other solutions. Figures 5A and 5B show a design solution and an image of the membranes in the microchannel. The second filtration design solution does not need any filter membrane. It works with the principals of centrifugal force, which make some particles go faster than others if the fluid has a curved trajectory with a certain speed. So, having one bend microchannel that separates into two different channels, it is possible to have different sized particles go to different outlet channels [Blatter, C., Jurischka, R., Tahhan, I., Schoth, A., Kerth, P., Menz, W. (2005). Microfluidic Blood / Plasma Separation Unit Based on Microchannel Bend Structures. Proceedings of the 3rd Annual International IEEE EMBS. Hawaii.]. It is important to mention that this filtration method only works under appropriate conditions such as the fluid speed being over 1 [m / s]. With the right conditions, this separation technique can get efficiencies up to 90% and can simplify greatly the filtration process due to the lack of membrane. Figures 6A and 6B show the CAD design of this design solution. The next design solution is very similar to the first TFF design solution mentioned previously. The main difference is that the channel is shaped as a spiral. As Z. Geng et al. (2012) [Continuous Blood Separation Utilizing Spiral Filtration Microchannel with Gradually Varied Width and Micro-Pillar Array. Sensors and Actuators B 180 (2013) (p. 122-p. 129). China.] explain, this design adds the centrifugal effect to the conventional serpentine-shaped TFF (that has been presented previously). This centrifugal effect increases the separation efficiency and reduces the risk of clogging. Figures 9 to 10C show the design of this solution. The fourth and final design solution for the TFF process is inspired by the TFF design shown in a paper authored by X. Li et al. (2014) [Continuous-Flow Microfluidic Blood Cell Sorting for Unprocessed Whole Blood Using Surface-Micromachined Microfiltration Membranes. Royal Society of Chemistry.]. In this case, the microchannel has two layers; the top one and the bottom one, which are separated by the filtration membrane. The fluid comes from the reactor and goes into the TFF process through the top layer channel. Due to a pressure difference, some of the molecules in the fluid go through the membrane and in to the bottom layer channel, whereas the molecules that are too big to go through the membrane stay on the top layer channel. This design solution offers good efficiency rates and most importantly, it offers a new perspective in terms of manufacturability. Figures 11A and 1 IB show a design of this solution. Final System Selection Regarding the final decision for the filtration process, the second design solution (microchannel bend structure) has been discarded because the fluid may not have enough speed to be effectively separated with this methodology. The serpentine-shaped TFF and the spiral TFF are quite similar, although the spiral TFF has higher effectiveness and lower clogging risk, so the final decision is between the spiral TFF and the multi-layer TFF. The key element for this decision has been manufacturability. The entire manufacturing process of the apparatus is explained further herein, but to make the decision for the filtration process, it is essential to know that part of the manufacturing process is done with 3D printing, more specifically, with a Stereo Lithography Apparatus (SLA). For the spiral TFF, the membrane needs to be to be manufactured with 3D printing whereas with the multi-layer TFF, the membrane can be bought from a supplier and assembled on to the microfluidic chip. The best resolution for the available SLA machines is around 10 micrometres, but since the molecules that need to be filtered are smaller than that, it is no possible to take this option. The result of all considerations explained herein is that the best design solution for the filtration process, and consequently the chosen one, is the multi-layer TFF. Referring to figures 7 and 8, tangential flow filtration module preferably comprises a first 5 plate-like layer 30, a second plate-like layer 31 positioned with their main faces opposed to one another and sandwiching therebetween a membrane 36 to provide a lamella structure. Each of the plates 30, 31 comprises serpentine channels having straight sections 34 and bent or curved sections 35 to create respective fluid flow channels 37, 38 in direct contact with and at least, in part, defined by intermediate membrane 36. A collection, input or 10 buffer reservoir 33 is provided at or towards respective lengthwise ends of any one of the first or second channels 37, 38. Accordingly, a fluid is capable of flowing through channel 37 of the upper plate with selected chemical components of the fluid having a smaller molecular size capable of defusing through the membrane 36 and into the adjacent fluid flow channel 38 of the second plate 31. Accordingly, TFF module of figures 7 and 8 is 15 configured for the separation of a permeate and a retentate differentiated by particle size (particle diameter) and / or molecular weight as will be appreciated. Microfluidics Studies and Simulation As illustrated in Table 2, there are two different sizes of bioreactor channel that are 20 required. Computational fluid dynamics (CFD) software has been used to analyse the assumption data in the microchannel. The microchannel requires a steady state flow rate. Table 2 The calculation results of different sizes of bioreactor channel Inlet Flow Rate (pl / min) Width (pm) Height (pm) Length (mm) Viscosity (cP) Density (g / cm3) Pressure Drop (Pa) Velocity (m / s) Reynold’s Number Flow Time (min) 3 800 400 2500 3.5 1.06 149.9 0.00016 0.0252 260.42 3 1000 500 2500 3.5 1.06 81.9 0.00013 0.0269 320.51 25 FLUENT: Fluent is computational fluid dynamics (CFD) software which helps to address fluid flow problems. It uses the finite volume method to solve the governing equations for fluid and provides many different physical models such as laminar or turbulent, viscous or not viscous, compressible or incompressible. Geometry and grid generation is doing GAMBIT which is the pre-processor bundled with FLUENT. A solution can be obtained by following these steps: Geometry; Mesh; Setup; Solution; Results. The channel is represented in a 2D CAD design. After, material properties and boundary conditions are set. Finally, the domain must be meshed. FLUENT converges the problem until the convergence limit is met or the number of iterations as specified is achieved. a) Geometry The geometry consists of walls, inlets and outlet boundaries, which is shown, in figure 12. b) Mesh Coarse fine mesh types are available. Mesh density varies based upon the Refinement Factor, as illustrated in table 3. Table 3 Coarse fine mesh types Mesh Density Refinement Factor Fine 1 Coarse 3 Physical models for FLUENT: Regarding the Reynolds number (Re = 0.0269) Laminar Flow has been selected. Air and water-liquid need to be selected as materials. The following material properties can be specified: Density and Viscosity. The following boundary conditions have been assigned in FLUENT as shown in table 4. Table 4 Boundary conditions assigned in FLUENT Boundary conditions Inlet Velocity inlet Outlet Outflow Wall Wall c) Solution The mesh is exported to FLUENT along with the physical properties and the initial conditions specified. When the solution is converged or the specified number of iterations is met, FLUENT exports the data. As table 5 shows, the total flow time is 14400 seconds, which is 4 hours so that every 60 seconds CFD software record data which needs to repeat 240 times, Then the CFD software analyses the velocity of X and F directions, energy and continuity. Finally, after 38 iteration the results have converged as shown in figure 13. Table 5 Input and Output Data FLUENT Input and Output Data Time Step Size(s) 60 Number of Time Steps 240 Flow time(s) 14400 Converged 38 Total iteration 276 Prototype Building To build the prototype different steps have been followed. Moulds have been created first and the chips have been casted from them. The different sub systems have been built separately in order to test each design. Experiments will be done on each sub system to validate the design. Different versions of the moulds have been edited. The design evolution is represented in figure 14. After experiments made on each mould some improvements have been made between the two versions. • The two moulds have been regrouped into one making the all preparation chip building process easier. • The cross-section area of the channels has been increased from 0.4mmx0.8mm to 0.5mmx 1mm avoiding the walls of the channels to collapse when testing the chip • Extra space between the channels (0.5mm to 1mm) and between the mould’s border and the features have been added to improve the quality of the chip and avoid any accidents when removing the PDMS mask from the mould. • Finally, some walls have been added to the borders of the moulds to improve quality of the whole PDMS chip when casting. Regarding the Vertical TFF device the top and the bottom layer have been edited separately to create two different layers which will be sealed together with the filtration membrane in the middle. Figure 15 illustrates the design of two layers. Table 6 explains the different features of the two designs. Table 3 Feature explanation of the two designs Features Explanation Main vessel 40 The main vessel has a cross section area of 0.5mmxlmm. The two moulds have identical channel. The two channels are reverse to aloud to have same channel when sealing the two layers one on top of each other. Male and female plugs 41 On both masks a female and a male plug have been added to help the positioning between the two layers during the assembly. Walls 42 For the same reason as the reactor walls have been added around the moulds to improve the quality of the chip. Soft lithography & SLA 3D printed moulds Microfluidics chip requires precision, accurate method and technique in order to make reactions or filtration work as expected. The most common method that is used for microfluidics chip is Soft Lithography. In this case Soft Lithography consists in creating a mould where PDMS will be casted from [Kamei, K. ichiro, Mashimo, Y., Koyama, Y., Fockenberg, C., Nakashima, M., Nakajima, M., ... Chen, Y (2015). 3D printing of soft lithography mold for rapid production of poly dimethyl siloxane-based microfluidic devices for cell stimulation with concentration gradients. Biomedical Microdevices,}. The mould is 3D printed from the CAD design of the wanted part. Although different techniques can be used for 3D printing, in the project case Stereo Lithography has been selected. With the different moulds designed, the parts were sent to be 3D printed using SLA. SLA printing allows to build complex shapes with a precision of 10pm in approximately 24 hours. Regarding the utilisation and the different shapes and features that the moulds required, the best resolution was set on the SLA machines to optimize the resolution and precision of the printing. The first version of the mask was analysed through a microscope to compare the size between the CAD file and the 3D printed part. Figure 16 illustrates two microscopic analysis of the channel dimension. Table 7 Microscopic analysis value of channel dimension Measurement Value CAD Dimension LI: Distance between channels 437 pm 400pm L2: Channel width at the 90° bend 902pm 800pm L3: Channel width in regular section 902pm 800pm L4: Channel width in regular section 910pm 800pm L5: Distance between channels 440pm 400pm Finally, the moulds have been printed in two different materials. The two are photopolymer resin with one being a high temperature resin. The main difference of these two materials are the maximum temperature before heat deflection. The first one can reach 50°C when the second one can go up to 250°C. Filtration membrane selection The selection of specific filtration membrane has been managed in two steps. The first steps consisted in finding the molecular cut-off size of the two different membranes for the two filtration stages. In fact, each filtration stages are dealing with different molecules with different molecular weight. As shown herein with the conceptual process some of the reagent’s molecules need to be filtered and recycled back to their respective bioreactors due to their high cost. Table 8 details the different molecules and their size or molecular weight. Table 4 Molecular weight information Molecules Molecular weight and spherical dimension Filtration stages DNA 10 Ml 9a 30 nm End of Bioreactor 1 RNA 5MD a 23 nm Final purified product T7 polymerase 99 kE >a -> 6.3 nm Vertical TFF 1 M7g + Methyltransferase 300 k Da 9 nm Vertical TFF 2 Regarding the two vertical TFF stages, the first one is dealing with T7 polymerase whereas the second vertical TFF is dealing with M7g+Methyltransferase. Considering the different molecules sizes, the selected filtration membrane needs to have a molecular cut-off size three times smaller than the molecules that needs to be kept, in this scenario it is the RNA molecules [General Electric, 2014). In addition, the filter needs to be large enough to let the recycling molecules go through it. Given all this information the molecular cut-off size of each membranes has been allocated as followed: • Vertical TFF 1 -> Molecular weight cut-off membranes of 300kDa • Vertical TFF 1 -> Molecular weight cut-off membranes of 500kDa The second step of the process was to select an appropriate membrane supplier. Therefore, because of time restriction, Imxlm flat sheet membrane from the company Synder filtration have been selected. Three types of membranes have been ordered with different pore size: 300kDa, 400kDa and 500kDa allowing to make different test and experiments and analyse the filtration effectiveness of each membrane. PDMS mask and casting With the mould printed and ready, Polydi methyl siloxane (PDMS) microfluidics can be casted from it. The product used for creating PDMS chip is Sylgard 184 silicone elastomer. This product is delivered with two elements, one silicone elastomer part and a curing agent responsible of the elastomers’ molecules crosslinking, which will solidify the elastomer while keeping casting properties. However, some preparations are required before applying PDMS over the mould. Table 9 shows the curing temperature with their curing time. • Silicone elastomer is mixed with curing agent with a mass ratio of 10:1 (Silicone / curing agent) for 10 minutes in a beaker. • After the mixing, the air bubbles in the mixing need to be removed with the help of a desiccator. Repeat this operation until all the bubbles have disappeared. • If small bubbles are remaining let the beaker rest. The bubbles will pop by themselves. • When the mix is ready, applied it on the 3D printed mould and let it rest for 10 to 20 minutes Put the mask with the PDMS on top in the oven or let it rest at ambient temperature. Table 9 Curing temperature and time Curing temperature Curing time 22°C 48 H 60°C 5H 100°C 35 min 125°C 20 min 150°C 10 min Different temperatures have been tested to compare PDMS curing properties and aspect. However, the first mould used in this process were not able to support high temperature. It is why the ambient temperature curing have been tested over a week end as well. Finally, with the different features casted on the PDMS mask, this one needs to be sealed on the open side. Two different methods have been used to seal the mask to allow different result from tests (figure 17 is a summary of the different steps of a prototype building process to create a continuous flow bioreactor system): • The first one consisted in sealing the PDMS masks with another flat layer of PDMS. This method allowed to add extra thickness of the whole mask and aided the integration of tubing and fittings to it. • The second method consisted in sealing the PDMS masks with Kapton tape on the open side. This method allows a perfect sealing between the features of the PDMS mask and the tape. However, the integration of tubing and fitting will be more complicated as the thickness of the mask is reduced. Sensing Methodology The main sensing system is designed combining the feature of the continuous flow microfluidics system and the reaction. Firstly, the sensing system of the microfluidics chip is limited by the chip size and the characteristics of the factors that need to be detected. Furthermore, to select suitable sensors for obtaining accurate and real-time detection results of the chip is another key point. The sensing system of the micro-factory comprises spectrometer, pH sensor, and pressure controller and sensors. The theories behind the spectrometer are discussed by L. Zhu, (2005) [, Nuno Miguel Matos Pires, (2014) and Kihwan Choi, (2015). The theory behind the pressure controller and sensors have also been explored by Yung-Shin Sun, (2016). Material selection a) Spectrometer The spectrometer (USB2000+ Miniature Fiber Optic Spectrometer, Ocean Optics inc, UK) can store 1000 full spectrum per second, and its detection range is from 200 to 1100 nm wavelength. In addition, to use the official software, programming with open source packages in Python according to the characteristics of the sensing system is the main. This spectrometer is suitable for monitoring the continuous reactions. b) pH meter The measuring range of pH meter (HI-1093B pH Electrode, Hanna Instruments Ltd, Bedfordshire UK) is 0 to 13. In order to integrate the pH meter with the microfluidies chip, the size of pH meter is limited due to the design specifications. The diameter of measuring cell on the microfluidics chip is 5 mm2, and the body of the pH meter is 3mm diameter. c) Pressure controller and sensors The CFD analysis of the microfluidics chips suggests that the pressure of the bioreactors and TFF sections is 0.27 Bar. The pressure sensor (MPS Microfluidic High Precision Pressure Sensor, ELVEFLOW, Paris France) can provide 5 ranges of measurement, from 70 mBar to 7 Bar. It also allows to detect an ultra-small internal volume to 7.5 pL, which can apply to the 0.5 mL volume of the microfluidics chip. The flow rate regulation of this sensor is sensitive and responsive, so it is appropriate for the real time monitoring of the minor changes in flow rate. The mating sensor reader (Sensor Reader, ELVEFLOW, Paris France) can provide a high-speed capability and is easy to integrate into the chips, which makes the measurement simple and feasible. d) Pressure supplier Syringe pumps (C3657 C-Series Syringe Pumps, Tricontinent, CA USA) are used as pressure supplier to feed the reactant materials into the microfluidic chip. The stroke speed of the pumps are 1.2 seconds to 100 minutes per stroke, and pump resolutions are 3,000 steps of standard mode and 24,000 steps of high-resolution mode. The reaction time of the system is 4 to 6 hours. Also, these syringe pumps can provide stable flow rate in the microfluidic systems. The pressure of 5 inlets for injecting the responsive components can control the proportions of reactants to achieve an economical and efficient yield of RNA vaccine. Method The key parameters of the reactions need to be monitored are RNA and Mg2+ concentration, pH value of the reactions and the temperature. At the same time, inlet pressure needs to be controlled, in order to achieve the appropriate proportion of RNA yield, and to adjust the flow rate of the mixture solution in the continuous flow channel for a proper reaction time. There are 3 detection points on the microfluidics chip, two points are used for quantitation of RNA and Mg2+ and the other for pH value measurement. Also, the pressure sensors are fabricated outside the chip. The concentration level change of Mg2+ and RNA can be used to detect the rate of reactions. The method of converting absorbance to concentration is a common way to accomplish the quantitation of RNA and Mg2+. UV absorbance measurements at 260 nm is a gold standard of RNA quantitation, and the Mg2+ quantitation can be obtained with absorbance measurements at 680 nm. The absorbance measurements of RNA and Mg2+ can be converted to concentration using the Beer-Lambert law. In the design, PDMS is used to make the bioreactors, which is sufficiently transparent. This makes the effective RNA detection viable. In the reaction, the corresponding thermostable DNA polymerase is the main source of pH value. The pH of 8.3 to 9.0 can give optimal results in the system. Also, increasing the pH value in the reaction can help to stabilize the DNA template and enhance the transcription results. The recombinant enzyme activity is significantly influenced by the pH of the assay. For integration of devices, Arduino micro controllers (ARDUINO UNO REV3, Arduino, UK) have been used at the core for data collection. Integrated Experimental Platform A schematic illustration of the present microfluidic reaction system is illustrated in figures 18A to 18C. The system comprises a plurality of sensors adapted for the measuring of various characteristics of the fluid flowing within the microfluidic system including pressure, temperature, pH, flow rate, flow volume etc. Also incorporated within the system is a plurality of valves, injection ports, pumps in particular syringe pumps, gates etc. to control the flow pathway of components of the fluid within the apparatus. For example, components of the fluid may be recirculated from an outlet port at the lengthwise end of a bioreactor and / or filtration module to return to an inlet port of an upstream bioreactor and / or filtration module as desired. The system further comprises a control unit that typically comprises a CPU, microprocessor or other suitable electronic processor device. According to a specific embodiment, the processor is integrated into a PCB or PLC. The control unit is provided with a user interface, sensing data storage utility, software, visual display device, communication ports and / or modules, at least one analogue to digital converter and / or a data source library. Such components may be utilised by the control unit in combination with the sensors, injection ports, pumps, valves etc. to control continuously and automatically fluid flow within the microfluidic system. Accordingly, a continuous reactor system is provided for the continuous input of reagents to enable the continuous output of desired reactant products. The present system is advantageous to recirculate and reuse reagents, solvents etc. to minimise waste and to maximise efficiency. Once the facilities are ready, the layout of them will be designed and the experiment environment will be set up. Necessarily, the PDMS chip ought to be fixed and clamped by a fixture. Then the locations of other instruments are determined accordingly. At the beginning, 4 syringe pumps were used to put pressure into the 4 input vessels respectively. Tubing and fittings are generally used to transport small volumes of samples to the PDMS chip. Another pump will be connected to a second vessel, which is located at the start point of the second bioreactor (end of 1st TFF), making it possible to inject the m7G methyltransferase and regulate pressure in the chip, the use of a pressure sensor with a feedback loop will increase the responsiveness of the flow control significantly. The pressure sensors are considered to measure at inlet and outlet of the channels. The target is to keep the syringe pumps working while still being able to ensure a constant pressure in the device. To implement the concentration detection, UV / Vis light should go through the detection chamber, where is at the end of the first bioreactor (serpentine channel), are transmitted through optical fibres. Two miniature optical fibres which match the specific wavelength needs are installed vertically to each other: a source fibre that is connected to the light source, and a collection fibre connected to the mini spectrometer. During the continuous bio-reaction, a pH sensor is planned to put into a buffer to measure the PH value, while temperature sensors can be placed on the board with a heater on the other side to adjust the temperature. The pH sensors and temperature sensors are equipped with BNC connector, which can be connected with Arduino. Thus, the data could be obtained from those sensors. Data Acquisition and Monitoring Spectrometer Data The Lambert-Beer law describes the linear relationship between the analyte concentration and light absorbance of specific wavelength: / 0 (9) Absorbance = log10( — ) = e ■ L ■ C 'i Where: / 0 is the initial intensity of the light source, 4 is the light intensity after it passes through the sample, e represents the wavelength-dependent molar absorptivity coefficient with units of M'1 cm'1, L is the path length, and C is the analyte concentration. Therefore, the expected spectrometer data is exactly the absorbance spectrum. The Ocean Optics spectrometer has provided an easy way to access the data from Python. That is the Python-Seabreeze package, which wraps the Seabreeze library to communicate with the spectrometer. It provides a fully working and tested reference implementation of the Ocean Optics USB interface, which means the spectrometer data can be read and monitored using python. The connection between python and spectrometer is shown in figure 19. Once the pH sensor has been connected with Arduino through BNC connector, the sensor data can be obtained, and the real time data can be checked on serial monitor. The code for this is shown in figure 19. The same mechanism is used to obtain the temperature sensor data. Five full factorial experiments were conducted focussing on the influence of the different proportion of reagents as well as reaction time and the proportion of Magnesium icon. The experiments were operated based on the fed-batch system with a volume of 500pL. Background experiment data is shown in below. Several conclusions can be drawn from the data. Figure 21 shows the maximum output coming from the mixture of lOmM NTPs, lOrnM NaCl, 75mM MgCh with Acetate ion, which is around 900 mM RNA yield, which could be regarded as the optimum combination. The optimum reaction time for RNA yield is shown in Figure 22. The RNA yield can reach a highest level when the proportion of Mg(OAc)2 is approximately 80 mmol. After four hours reaction, the RNA yield achieves stability at around 1800mM. The optimum combination of the reaction condition depends on the four main factors: the proportion of MgCF, NaCl and NTPs as well as the reaction time. For the further experiment, the proportion of MgCh, NaCl and NTPs need to be monitored to keep it at a certain level for getting the maximum yield of RNA. The most remarkable difference of the experiments between Imperial College and Cranfield University is the reaction system design. Due to the manufacturing consideration, continues follow system is designed for the bioreactor process instead of the fed-batch reaction system. To test the continuous flow system, the experiment is designed based on the previous experiments conducted by Shattock Group at Imperial College. Three levels are chosen for four factors (table 10) to test the optimum proportion for RNA yield in a continuous flow system. The selected factors refer to the result of fed-batch system experiments. Additionally, the values of DNA and T7 Polymerase are the same for the recycle using reason. With these factors, the new design of experiment is listed as table 11. The experiment aims to find the relationship between RNA yield and four factors and to identify the trend of reaction. Table 10 Factors and Levels selection for continuous follow system experiment No Factors Levels Response 1 MgC12 70 75 80 RNA Yield 2 NTP 2 4 6 3 DNA 8 10 12 4 T7 Polymerase 8 10 12 Table 11 DoE table for continuous follow system (mM) Run MgC12 NTP DNA T7 RNA Yield 1 1 0.25 0.05 0.05 2 1 0.5 0.1 0.01 3 1 0.1 0.05 0.01 4 1 0.5 0.1 0.1 5 1 0.25 0.05 0.1 6 2 0.5 0.01 0.1 7 2 0.5 0.05 0.05 8 2 0.1 0.1 0.01 9 2 0.1 0.1 0.1 10 2 0.25 0.1 0.05 11 2 0.1 0.01 0.01 12 2 0.1 0.01 0.1 13 2 0.25 0.05 0.05 14 3 0.5 0.01 0.1 15 3 0.5 0.1 0.01 16 3 0.5 0.01 0.01 17 3 0.25 0.01 0.05 18 3 0.1 0.1 0.1 5 RESULTS AND DISCUSSION Hydrodynamics and CFD Analysis Suitable governing equations are applied for hydrodynamics study in the bioreactor and TFF parts to calculate the filtrate volume with constant fluid velocity and suitable pressure gradient under steady state flow conditions. Equation (6) has been utilised to get the prediction of pressure drop in the microchannels. Selection of the characteristic length is an arbitrary choice and will not affect the final solution as illustrated in table 12 and table 13 5 Table 12 Fluid properties under different channel length of 1000 pm X 500 pm Inlet Flow Rate (pl / min) Width (pm) Height (pm) Total Length (mm) Viscosity (cPa) Density (g / cm3) Pressure Drop (Pa) Velocity (m / s) Reynold’s No Flow Time (min) 3 1000 500 1000 3.5 1.06 24.57 0.0001 0.0202 166.67 3 1200 29.48 0.0001 0.0202 200.00 3 1500 36.85 0.0001 0.0202 250.00 3 2000 49.14 0.0001 0.0202 333.33 3 2200 54.05 0.0001 0.0202 366.67 3 2500 61.43 0.0001 0.0202 416.67 Table 5 Fluid properties under different flow rate of 1000 pm X 500 pm Inlet Flow Rate (pl / min) Width (pm) Height (pm) Total Length (mm) Viscosity (cPa) Density (g / cm3) Pressure Drop (Pa) Velocity (m / s) Reynold’s No Flow Time (min) 4 1000 500 1000 3.5 1.06 32.76 0.00013 0.0269 128.21 4 1200 39.31 0.00013 0.0269 153.85 4 1500 49.14 0.00013 0.0269 192.31 4 2000 58.96 0.00013 0.0269 230.77 4 2200 65.52 0.00013 0.0269 256.41 4 2500 81.90 0.00013 0.0269 320.51 Other pressure losses associated with the measured pressure drop are inlet and exit bend 10 losses. Bend pressure losses can be obtained within different number of bends in table 14. Table 14 Bend pressure loss percentage calculations with different number of bends No. of Bends Flow Rate (pl / min) Density (g / cm3) Bend Pressure Loss (Pa) % of Pressure Loss 12 3 1.06 1.48 0.99% 16 3 1.06 1.97 1.32% 20 3 1.06 2.47 1.65% 24 3 1.06 2.96 1.97% 28 3 1.06 3.45 2.30% 32 3 1.06 3.95 2.63% 36 3 1.06 4.44 2.96% 40 3 1.06 4.93 3.29% Hydrodynamics Study in TFF part Cerman-Kozeny relationship is used to obtain the permeability constant to solve the fluid 5 parameters through microporous channels under suitable boundary conditions. Here different particle diameters and pore sizes are applied because of the variations in input particle sizes in table 15. Table 15 Permeability constant for different particle diameters and pore sizes Particle Diameter, dp (pm) Pore Size, e(pm) Permeability, K ( / pm3) Darcy Number, Da 0.020 0.060 5.4323E-10 2.122E-22 0.021 0.063 6.9776E-10 2.726E-22 0.022 0.066 8.8616E-10 3.462E-22 0.023 0.069 L1139E-09 4.351E-22 0.024 0.072 1.3869E-09 5.418E-22 0.025 0.075 L7120E-09 6.688E-22 The pressure drop for the geometry can be predicted from equation 10 The limitation of the equation is that it gives mathematical error if the units of measurements in microns. Pressure drop over length in Micro-porous Channel, as shown in table 16. (10) Where, p= Dynamic Viscosity Q= Flow Rate h= height / depth of channel K= Permeability Table 16 Pressure loss over length in microporous channel for different flow rates Flow Rate, Q (pl / mi n) pQ.sinh.(h / A) Kh 2< / h(-l+cosh(hW k)) sinhfh / C k) i , i 11 ■ ; IM jw j-j=j AP / L (KPa ) 3 1.943 0.000020 48 3.1514508 3 0.185 6.07529E-05 31.9 7 4 2.590 0.000020 48 3.1514508 3 0.185 6.07529E-05 42.6 3 5 3.238 0.000020 48 3.1514508 3 0.185 6.07529E-05 53.2 8 8 5.180 0.000020 48 3.1514508 3 0.185 6.07529E-05 85.2 6 10 6.475 0.000020 48 3.1514508 3 0.185 6.07529E-05 106. 58 The result that are obtained in this study is just a prediction which depends on the geometry, fluid conditions and boundary conditions. According to the membrane selection 5 parameters in the Pall Corporation article, the filtrate volumes can be estimated which is given in table 17. Table 17 Filtrate volume prediction for different filter length area Length (m) Membrane Area, A (m2) Filtrate Flux Rate, J (LMH) Process Time, T (Hours) Filtrate Volume V, (ml) 0.4 0.00032 10 0.50 1.6 0.5 0.0004 10 2.00 8.0 0.6 0.00048 10 1.00 4.8 0.7 0.00056 10 1.50 8.4 0.8 0.00064 10 2.00 12.8 1.0 0.0008 10 2.00 16.0 Computational Fluid Dynamics Analysis Velocity profile in Bioreactor channel As figures 23A and 23B demonstrates, the X direction velocity goes through the whole bioreactor channel. There are two different colours show the same velocity, which go between 0.0001 msA-l and 0.0002 msA-l, the reason being that the yellow colour indexes positive direction and the green colour indexes negative direction. At the beginning, figure 23B shows that the velocity increased because the initial pump gives higher pressure and then declined to the constant velocity of about 0.00033msA-l which is around two times higher than the predicted result. Tangential Flow Filtration channel Figure 24 shows the velocity in tangential flow filtration channel. The colours demonstrate that different velocity in microchannel are between 0.0001327 msA-l and 0.0001858msA-l which is close to the inlet velocity. Moreover, the data from table 18 indicates that the outlet velocity is slightly declining from 0.000145 msA-l to 0.000138 msA-l, as expected. Table 18 Velocity magnitude Velocity Magnitude (m / s) Inlet 0.000145 Outlet 0.000138 Pressure contours in bioreactors Figure 25 illustrates the pressure drops in the bioreactor channel. The colours represent the different pressure values. Figure 25B demonstrates that the pressure drops down as a consequence of having many bends in the bioreactor channel. Those different results have been considered in the design evolution. Experimental Results PDMS Masks Tests Bioreactors As different versions of moulds have been realised through the present work, the evolution of the result has been observed and will be explained. a) First Mould Generation First, both first version of the two bioreactors have been tested. The different sealing options, Kapton tape and second PDMS layer have been experimented, and results have been extracted. Both reactors have been curing over two days at 35°C. The two reactors have been firstly sealed with another flat PDMS’ layer. Green water has been introduced in one of the inlets of each masks thanks to a syringe. Figure 26A and figure 26B show the first result about the fluid behaviour for each mask. The two pictures show that on the batch bioreactor, when the green water is introduced in the mask, the water is contouring the middle of the reactor. This has been explained by the fact that the middle of the reactor has collapsed during the curing as it is possible to see on blue section of figure 26A. Furthermore, figure 26B shows that green water sample has gone through the T junction of the masks and followed the main channel pathway. However, the sample has avoided some areas. This can be explained by the size of the channel’s walls, that couldn’t support the pressure and collapse during the curing process. Secondly, the continuous flow reactor sealed with Kapton tape has been tested. Figure 27 shows the fluid behaviour inside the new set up. The fluid (dark green) has followed the pathway correctly. The sealing has worked better than the previous one. However, due to the size of the channel the curing did not work properly again, and some parts of the mask did not peel properly from the mould creating some defects zone in the mask. From all the observations made on the different tests, the design has been changed. In addition, it has been decided to create a platform where the PDMS mask can be placed between two acrylic plates, increasing the pressure between the two PDMS layers and, therefore, avoiding the channel to collapse when testing the masks. These two acrylic plates would also allow to connect fittings and tubing to the chip. Figure 28 shows that Kapton sealed masks have been placed between the two plates, and has been connected with tubing, fittings and some sensing instrument to allow first PH and absorbance testing and measurements. b) Second mould generation Using the high temperature resistant material to print the second mould generation the masks have been casted with a curing temperature of 70°C for four hours. As figure 29 shows, the mould after use, the side in direct contact with the mould has bonded with it. Small parts of the PDMS mask could not be peeled off and removed from the mask. Specially between the channels, thin layers of PDMS were sticking on the mask. Although the curing has worked properly, the peeling of the mask from the mould has failed. After trying to clean the mould as much as possible with isopropanol, another tentative was given with a curing temperature of 100°C for 35 minutes. The same problem occurred unfortunately. An explanation of such problem can be given by the fact the high temperature resistant material from the 3D printers has bonded with the PDMS layer directly in contact with it. The walls may have, as well, an impact on destabilizing the curing homogeneity of the whole mask. Vertical Tangential Flow Filtration Only one version of the vertical TFF mould have been printed using the high temperature resistant material. As for the bioreactor masks the same process have been followed and a PDMS sample, poured in a foil paper recipient, has been placed in the oven with the two masks in order to compare the different aspect and result of the process. Figure 30A and figure 3 OB show the aspect of the masks as well as the sample cured in the oven. As for the bioreactor mask, the PDMS in contact with the mould has bonded with it. When peeling it off from the mould, the mask has cracked and broken as shown in figure 30A. However, the sample cured in a foil paper has cured without any defaults and nothing happen when peeling it off from the foil. On top of that, the sample has casted perfectly the shapes of the foil as shown in figure 30B. From these different observations some analysis can be made, and result found: • The chemistry and the protocol cannot be challenged as the sample in the foil worked. • The same defaults have occurred several times with the different masks. Therefore, the high temperature material has a negative impact on the PDMS mask casting and should not be used for such experiments • The first material used for the first experience should be kept and the curing parameters should be set at ambient temperature with 48 hours of curing, to let the PDMS mask cast properly and avoid any heat deflection from the mould. Despite of these results, a new platform has been created, as shown in figure 31. Based on the integrated experiment design, all the facilities have been collected in the lab, and the experimental platform has been set up to verify the effectiveness of the system. The built experimental platform in the lab is shown in figure 31, the PDMS chip is fixed and clamped by a fixture and connected the pumps and chip by tubing and fittings. Two optical fibres are installed vertically to each other: a source fibre that is connected to the light source, and a collection fibre connected to the mini spectrometer (Ocean Optics USB2000+). Absorbance Measurement and Digital Output Precise absorbance measurement For precise absorbance measurements, the baseline needs to be set up. Firstly, record the spectrum of the background source of radiation without the sample. Secondly, repeat the experiment by adding the sample in the absorption path. Thirdly, subtract the first spectrum from the second one and get a clean absorption spectrum of that sample. To cancel the noise effect, the background spectrum should be subtracted from both sample and reference spectra. So, equation (9) should become: ., , , ^sample ^B (11) Absorbance = — log10 —------— ‘Ref —‘B Where: Isampie is the light intensity after passing through the sample, 1B is the background light intensity, which is recorded by the spectrometer with the light source switched-off or blocked and without sample. IRef is the reference light intensity. UV-Vis Data analysis Through Python-Seabreeze API, it’s easy to get the real-time data from the UV-Vis spectrometer and plot the figure of intensities against wavelength. As figure 32 shows, the light intensity reaches a peak value when the wavelength is around 580nm. To verify the relationship between absorbance and wavelength at different concentrations, various concentrations of Vitamin B12 solutions have been used, which are A (25mM / L), B (50mM / L), C (75mM / L), D (lOOmM / L) respectively, see figure 33. The UV-Vis spectral result is shown in figure 34, the absorbance increases with the increase of sample concentration, and the absorbance reaches the maximum at a wavelength of 550 nm. PH Sensor Calibration and Digital Output PH sensor calibration To ensure accuracy, the PH probe used for the first time needs to be calibrated, two standard buffer solutions are used to calibrate the PH sensor, which are of 4.0 and 7.0. Following the calibration step provided by DFROBOT, the calibration has been completed. PH value plot in real time The next step consisted in testing the effectiveness of the PH sensor with acid and alkaline solutions. The result, in figure 35, shows that real-time figure dynamic changes follow PH changes, which proves its validity. Discussion Critical Result Evaluation The work and results obtained are effective attempts of implementing lab-on-a-chip technology to achieve continuous flow reaction with maximum RNA output. Applying the SBCE methodology has allowed to build an entire design framework that has been followed from the beginning to the end of the project. Dividing the device in different subsystems has allowed to build different innovative solutions by thinking outside the box. Considering that the work started with little information on chip design, it has been possible to develop novel design solutions that entirely correlate to the project requirements. As a result of the present work, the reactor and filtration processes have been defined as the key elements of the device. • Two different design solutions have been developed for the reactor which allow to change the RNA manufacturing from a manual batch process (conventional RNA synthesis) to a continuous flow process. In fact, the developed continuous flow reactor is an innovative idea that has been proved to work (J. Lawrence et al., 2013) but hasn’t been developed to this extent, and it tackles one of the most challenging requirements. • The filtration process has been deeply studied and analysed in order to find a system that could meet the requirement while being easy to manufacture with the available resources. Therefore, starting from the most general and commonly used tangential flow filtration, which has proved to be effective in RNA purification processes (A. Eon-Duval et al., 2002), a device has been developed with the target of having a straightforward and cost-effective manufacturability. Indeed, aspects of the vertical TFF is the separation of the device into two different independent layers sealing together with a filtration membrane in between. This way the vertical TFF device is easy to manufacture while keeping all the complexity that a filtration process requires. In addition, the channels allow to keep both retentate and permeate in the continuous flow process as well as recycling and preserving the costly reagents such as T7 Polymerase or M7G methyltransferase. Although the filtration device has not been tested, there are good expectations that it will provide efficient results. Regarding the whole system, a new integrated platform has been developed in order to run tests and experiments. The platform is compound by an acrylic structure that integrates pumps, sensors and fittings for inlet and outlet. It allows to test a continuous flow reaction or filtration while getting instantaneous feedback from the different elements and stages of each process. Furthermore, the system has been manufactured in a modular way that allows an easy to use framework for running the experiments. 3D printing has been the chosen tool for manufacturing the moulds due to its high flexibility, low cost and short lead times. Moreover, these technologies have been used previously for microfluidic chip applications, and more specifically, for soft lithography casting [Kamei, K ichiro, Mashimo, Y., Koyama, Y., Fockenberg, C., Nakashima, M., Nakajima, M., ... Chen, Y. (2015). 3D printing of soft lithography mold for rapid production of polydimethylsiloxane-based microfluidic devices for cell stimulation with concentration gradients. Biomedical Microdevices}. Unfortunately, in this case, 3D printing hasn’t provided the optimum results. The materials used by the SLA printers was not ideal for this application because it has not allowed to have an effective casting process of the PDMS mask. Removing the mask from the mould has been one of the most problematic operations during the manufacturing process even while changing some of the parameters such as the curing temperature and time, the curing agent ration, or the support used while curing. The UV-Vis data shows that an appropriate way has been found to clarify the material content and proportion in the reactor through real-time monitoring the concentrations of the samples against wavelengths, which can provide the information about to what extent the RNA transcription has proceeded, and give a feedback to the perfusion system (pumps, tubing, fittings etc.) for regulation of input raw materials to compensate for the consumed molecules in order to maintain a continuous reaction of the process. The successful pH sensor data acquisition makes it easy to know exactly the pH value in the reactor at any point, helping to optimize the reaction conditions. The integrated experimental platform having been built partly completed what have been planned to test. Because of the failure of PDMS masks testing, there was reduced opportunity to test the real value in the RNA transcription reaction, just the effectiveness of spectrometer and pH sensor has been demonstrated. The real testing in the chip can be executed only after the raw material samples are injected into the PDMS chip with a smooth flow without any leakage or intermission. The results obtained in the serpentine channels of the Bioreactor parts are reliable and the equations that we applied are suitable for the present work. But it is very difficult to get equations to solve the pressure and velocity in the microporous channels due to different design parameters and geometry of the channels. In other research areas the pressure difference was readily available to measure the flow rates and velocity profile for continuous flow. In the work, it is needed to calculate the pressure gradient by applying appropriate equations to get correct results to build the chips. Although the integrated platform to run the tests, and the different designs for the device are ready, there is some further investigation required on microfluidic PDMS chip manufacturing process. New pathways for manufacturing the moulds have been identified and these include: • Applying a coating on the mould to facilitate the casting process, specially the peeling operation. • Using different materials when printing with the SLA machines or even try using FDM printers, as long as the resolution meets the requirements. • Manufacturing the mould on a metallic material by implementing micro machining in the process. These new suggestions should be studied in depth but still following the existing framework. The new ideas should maintain the ability to be integrated with the setup that has been developed. • Based on the successfully working PDMS prototype, operating the experiment to test the parameters (pH, temperature, concentration, pressure, etc.) For the future work, the first step is to continue the experiment. When the PDMS mask is verified working, the experiment can be conducted with the current facility and the experimental results can be compared with the calculation ones, which will be the fundamental of the control system development. • Outsourcing the sensors design and manufacturing process. Based on the prototype design, redesign the sensors into proper sizes that could be integrated into the chip. • Developing the automatic control system. Create a feedback loop to achieve a dynamic control of pressure in the channels, automatically change the proportion of input raw materials to maintain stable flux. Hydrodynamics study in microfluidics systems is new technology in its area of 5 applications and therefore, the predicted results need to be verified in the experiments to confirm the design parameters. We may work more on the cross-sections of the channels to get the optimum flow under steady state flow conditions. Optionally, the reactor modules and filtration modules may be manufactured by printing eg 10 lithography, machining (scoring or etching of a solid plate or disc), moulding (such as injection or blow moulding) or casting. As will be appreciated, the modular fluidic processing units may be manufactured by a variety of techniques with such units being formed from a polymer material, glass or a metallic material. 15

Claims

1. A method of RNA synthesis via a continuous flow process comprising:introducing into a first microfluidic reactor module via at least one inlet a plurality of reactants comprising: a nucleoside triphosphate (NTP), DNA, a salt solution and RNA polymerase;driving a flow of a fluid comprising the reactants through an elongate reaction channel of the first microfluidic reactor module;allowing at least some of the reactants to react within the elongate reaction channel;filtering reaction products of the reactants using a first filtration module; and recirculating at least some of the reactants output from an outlet of the the filtration module to a region of the at least one inlet of the first reactor module via at least one recirculation conduit.

2. The method as claimed in claim 1 wherein the step of recirculating at least some of the reactants comprises recirculating the RNA polymerase from an outlet of the first filtration module to an inlet region of the first reactor module.

3. The method as claimed in claim 1 or 2 wherein the steps of filtering reaction products and recirculating at least some of the reactants comprises using a gate to filter DNA output from an outlet end of the elongate reaction channel and recirculating the DNA to an inlet region of the first reactor module.

4. The method as claimed in any one of the preceding claims comprising delivering at least some of the filtered reaction products from the first filtration module into a second microfluidic reactor module in combination with inputting a capping enzyme into the second reactor module.

5. The method as claimed in claim 4 comprising delivering a fluid output from the second reactor module to a second microfluidic filtration module.

6. The method as claimed in claim 5 wherein the step of recirculating comprises recirculating any unreacted NTP to an inlet region of the first reactor module and recirculating any unreacted capping enzyme to the inlet region of the second reactor module.

57. The method as claimed in any preceding claim wherein the salt solution comprises MgCl2.

8. The method as claimed in any preceding claim wherein the RNA polymerase10 comprises T7 polymerase.

Images