Access the full text.
Sign up today, get DeepDyve free for 14 days.
Limited applicability and scarce availability of analytical equipment for micro- and millifluidic applications, which are of high interest in research and development, complicate process development, control, and monitoring. The low-cost sensor presented in this work is a modular, fast, non-invasive, multi-purpose, and easy to apply solution for detecting phase changes and concentrations of optically absorbing substances in single and multi-phase capillary flow. It aims at generating deeper insight into existing processes in fields of (bio-)chemical and reaction engineering. The scope of this work includes the appli- cation of the sensor to residence time measurements in a heat exchanger, a tubular reactor for concentration measurements, a tubular crystallizer for suspension detection, and a pipetting robot for flow automation purposes. In all presented applications either the level of automation has been increased or more information on the investigated system has been gained. Further applications are explained to be realized in the near future. Article highlights • An affordable multipurpose sensor for phase differentiation, concentration measurements, and process automation has been developed and characterized • The sensor is easily modified and can be applied to various tubular reaction/process units for analytical and automation purposes • Simple integration into existing process control systems is possible Keywords Multiphase flow · N on-invasive sensor · Online concentration measurement · Process automation · Light dependent resistor · Residence time distribution especially popular due to their convenience and availability Introduction of various configurations of tubes. Introducing bends and curves to tubular reactors effects the centrifugal force and In research and development of chemical and process engi- induces secondary flow, which in turn improves mixing [5 ]. neering, continuous-flow processes have become more and Together with the growing interest in continuous pro- more popular . Although the flexibility of continuous cesses, miniaturization and modularization of chemical reac- processes lacks some features compared to batch processes tors are on the rise . Their large specific surface enables , attributes such as product quality , throughput, effi - high mass and heat transfer rates , and they play a signifi- ciency and safety  are typical and unique characteristics cant role in process intensification [8 , 9]. Particularly effects of continuous processes. The most basic continuous reac- that cause improved internal mixing accompanied with nar- tor is a pipe or tubular reactor. Reactants are introduced at row residence time distributions (RTDs) are systematically the inlet and after a desired hydrodynamic residence time made use of to achieve high product uniformity [10, 11]. the products can be collected at the outlet. These setups are A specific example of a capillary reactor, which is also frequently used for miniaturized chemical reactors is the * Stefan Höving capillary Coiled Flow Inverter (CFI). A basic schematic is Stefan.Hoeving@tu-dortmund.de given in Fig. 1. CFIs are tubular or capillary reactors that Department of Biochemical and Chemical Engineering, have been characterized in different configurations and Laboratory of Equipment Design, TU Dortmund University, applications [7, 10, 12, 13]. The coil direction of the tube Emil-Figge-Straße 68, 44227 Dortmund, Germany Vol.:(0123456789) 1 3 186 Journal of Flow Chemistry (2022) 12:185–196 Fig. 1 a Schematic of a Coiled Flow Inverter (CFI). Grey is the sup- tion of three different positions in the CFI. The blue color gradient is porting structure (frame) for the tube (blue) d : diameter of the frame, the velocity profile of the fluid inside the tube. Dashed lines describe d : diameter of coil windings, L : Length of one coil, p: pitch dis- the circulation of Dean Vortices. c Summed up the circulation of ct c tance between two turns, d : inner diameter of the tube. b Cross-sec- Dean Vortices along the tube coordinate is changed by 90° after a certain amount of turns. The coils priced sensors capable of phase detection and quantitative cause centrifugal forces on the fluid elements, resulting in analysis in micro- and millifluidic applications are often secondary flow (Dean flow) and therefore in increased mix- limited to unwieldy camera [20, 25] and ultrasonic setups ing and a narrow residence time distribution in single phase , that do not always fit into small-scale and laboratory flow . When it comes to two-phase flow within the slug setups. flow regime, secondary Taylor vortices are also induced Minimal invasive conductive sensors for detecting phases due to the interfacial boundary of the phases. This provides are described in , measuring changes in electrical resist- for internal convection and enhanced interfacial diffusion ance of the fluids of interest. A capacitive, and therefore between contacting fluid compartments (slugs) [15– 17]. non-invasive, solution is presented in  and used to dif- More precise CFI design guidelines can be found in . ferentiate between oil and water in a tubular flow. By meas- Considering a two-phase liquid/liquid system in a tubular uring the impedance instead of the fluid's resistance, it has reactor, two slugs of the discontinuous phase can be seen as been shown that non-invasive measurement of conductivity individual reaction compartments that do not directly inter- is possible to distinguish between different phases, too . act with each other due to spatial separation. For microflu- In a recent contribution, von Vietinghoff et al. presented a idic applications this has already been used in droplet-based multi-transistor sensor used to determine slug flow parame- high-throughput screening systems for chemical analysis ters of a triphasic slug flow of 1-hexanol, water, and nitrogen such as described in . . The determination of multi-phase flow parameters has The analysis of the phase of interest is often done with been demonstrated for microfluidic tubular reactor setups. offline methods for which a phase separation is necessary. Glotz and Kappe constructed an open-source photometer and However, when the flow regime should not be disturbed, demonstrated applications in flow chemistry . The work - phase separators and invasive analytical devices are not ing principle is similar to the sensor concept presented in eligible. This also holds for most commercially available this work and used for reactor characterization and the deter- UV–Vis spectrometers that disturb the flow regime due to mination of reaction kinetics. It uses optical fibers to connect changes in the cross sectional area of the flow cell. Nonin- emitter and detector of the sensor to the region of interest vasive and online analysis of the individual reaction com- within the tubular experimental setup. This enables for good partments is often done with camera setups observing the separation of the reactor from the sensors electric compo- phases in optically accessible tubing . Non-invasive nents, which is of interest for applications in explosion-risk methods include, next to the mentioned optical ones, also areas. Additionally, heat dissipation from the LED to the acoustical ones. Contributions reporting the quantification tube containing the liquid medium is prevented. The setup of phase proportions, velocities and contents can be found presented here omits the optical fibers, electrical amplifiers [21–24]. However, flexible, non-invasive and reasonably and filters making the whole setup less complex and easier to 1 3 Journal of Flow Chemistry (2022) 12:185–196 187 understand and apply for the non-specialist. The 3D-printed with 5 mm diameter as emitter and light depending resis- housing makes it possible to clamp the sensor to a desired tor (LDR) (NSL-19M51; Luna Innovations Inc., Virginia, position of the used tube instead of having to uninstall the United States) as detector with a detection optimum at tubing and threading the tube through the sensor. 550 nm. For the presented use cases a red LED showed Additionally, the sensor concept with the printed hous- the highest sensitivity, but can be easily substituted by ing allows for rapid modifications in the design to fit spe- other colors (wavelengths) depending on the application. cific needs that goes beyond a straight capillary used for the Here the optical properties of the tubing material has to determination of reactor kinetics and reaction kinetics. be kept in mind. A comparison of different LEDs can be The sensor presented in this work can be applied to dif- found in SI10. LDRs contain a photosensitive semiconduc- ferent tubular setups due to its non-invasive and optical tor element with a CdS resistor that changes its electrical characteristics. The working principle is known for many resistance depending on the light reaching the element. years  and is based on the optical absorption to infer Thus, the intensity of the light reaching the detector can from the attenuation of light to the composition of a sample. be correlated to the electrical resistance of the LDR. To be The sensor discussed here represents a flexible application able to gain insights on the contents of a tube or capillary, technique of the mentioned principle. The use cases in this LED and LDR are positioned opposite of each other with contribution describe tubular/capillary applications of an the tube between them. To shield the sensor from light optical non-invasive sensor, demonstrating the flexibility and from the outside, two different sensor housing versions versatility in four die ff rent applications. The working princi - have been designed and implemented. The larger version ple as well as details of the developed sensor are described fits on tubes with an outer diameter of 3.2 mm, hereafter for each application. Together with the conclusion of this called Version A. The smaller version fits to tubes with work an outlook on further application is given with relevant an outer diameter of 1.6 mm, hereafter called Version B development opportunities. (cf. Fig. 2). Besides differences in the tube diameter, the versions differ in shape, which resemble two different approaches to prevent stray light from the outside. Both approaches are the result of iterative preliminary experi- Materials and methods ments investigating the influence of outside light on the sensor signal. A systematic investigation concerning the This chapter starts with a recap of the Lambert–Beer Law design approaches was therefore not carried out. Version and continues with the construction and operation of the A has elongated ends that help preventing light from the developed sensor. outside reaching the detector of the sensor. Version B has two 90° bends forming an S-shape to impede the path of Lambert–Beer law the ambient light. Both versions consist of two 3D-printed housing parts (Black PLA, BASF SE, Germany, printed The Lambert–Beer Law (Eq. 1) links the radiation intensity on Ultimaker S5 (Ultimaker, Utrecht, Netherlands) with of an emitter Io passing through a sample (I) with the con- 0.1 mm layer height and 40 % infill) that can be manually centration c, the extinction coefficient ε , and the thickness e clamped together. Resource files of technical drawings and to the absorbance A. CAD-files for both versions can be found in SI2-SI7 of the Supporting Information. Modifying these parts is easy and A ∼ E = log = ce (1) can be done iteratively with appropriate software to fulfill specific needs. LDR and LED are fixed in the designated The application of Lambert–Beer Law comes with sev- holes inside the housing parts using molten polylactic acid eral limitations that have to be met to fulfill validity. This (PLA). This way, electrical connectors are still accessible includes that the medium of interest must not scatter, e.g. from the outside of the housing, while both compartments must not be turbid. Furthermore, the concentrations of are protected from light. After fixing the sensor around a solved substances have to stay moderately low because the tube, LDR and LED face each other with the tube right coefficient of absorption becomes dependent on the concen- between them (cf. Fig. 2). This way, the non-invasive tration for high concentrations and a linearity is not given working principle can be ensured. anymore. [28, 29] Due to their optical properties tubes made from fluori- nated ethylene propylene (FEP) (Bohlender GmbH, Ger- Sensor details many), have been selected. FEP is optically transparent and has nearly the same refractive index as water (1.333; The central piece of the sensor is a light-emitting diode depending on the temperature) , allowing for an opti- (LED) (peak at 620–640 nm, Tru Components, Germany) cal analysis of the contents. 1 3 188 Journal of Flow Chemistry (2022) 12:185–196 Fig. 2 a Schematic setup of the two sensor versions and cross-section. b Photograph of the s-shaped sensor clamped to a FEP tube (1.6 mm o.d., 1 mm i.d.). The tube is filled with diluted dye. c Photograph of the straight sensor clamped to a FEP tube (3.2 mm o.d., 1.6 mm i.d.) to ground. This way, the LDRs resistance can be measured Software and electric circuit as an analog signal by the microcontroller. The resistance of the LDR correlates with the incoming light. The Arduino As it has been mentioned before, the electrical compo- contains a 10-bit analog to digital converter. This means that nents of the sensor are LED and LDR. To make processing the analog resistance (or signal) is mapped to integer values and easy interfacing possible, a microcontroller has been from 0 to 1023. This resolution did not pose challenges in selected with the Arduino UNO (Arduino S.r.l., Monza, the conducted experiments. If a higher resolution was neces- Italy). As it can be seen in Fig. 3b), the LED, providing sary, different platforms or expansion boards could be used. the light, is powered with 5 V from a digital pin (2) of the The program (cf. Fig. 3a)) installed on the Arduino is Arduino. A variable resistor or potentiometer connected in written in C + + running on the Arduino IDE (integrated series with the LED can be used to adjust the intensity of development environment). A minimum working example the emitted light. An LED with a forward voltage of 2.1 V (MWE) is provided in Fig. 3a) and in SI8. A more complex in series with resistors of more than 220 Ω would con- code for two sensors is provided in SI9. For the operation sume < 30 mW. A need for active cooling was not observed. only 8 lines of code are necessary. In the setup() the LED The LDR is connected from the 5 V pin to an analog input is turned on and a serial connection to a serial host with a pin of the Arduino. A 10 kΩ pulldown-resistor is connected Fig. 3 a Screenshot of the Arduino IDE with a minimum working example code used to run the sensor. b Simplified electronic wiring diagram of the Arduino and sensor. This setup can be operated with the program in a) 1 3 Journal of Flow Chemistry (2022) 12:185–196 189 baud rate of 9600 is established. Inside the loop(), the analog installed invasively. Therefore, the developed sensor is used value obtained from the LDR is saved to the SensorValue to track the concentration of an optically visible tracer solu- variable, which is repeatedly printed via the serial connec- tion to have insights on the RTD of milli-fluidic elements in tion. A delay of 20 ms is inserted within the loop(). This a non-invasive manner by simply clamping the sensor to the time scale is a compromise between the performance of the existing FEP tubing of the fluidic element. Arduino and measurement resolution as well as amount of From the recorded raw data the RTD function of the generated data. With this code, a measuring frequency of heat exchanger at the outlet E(t) and the cumulative func- 50 Hz is achieved which did not pose challenges regarding tion F(τ) have been calculated according to . Apply- data lag due to the faster serial connection. The use of the ing the axial dispersion model and a curve fitting method millis() function would make the timing more efficient, while characteristic values such as the axial dispersion coefficient increasing the program's complexity. For reasons of simplic- (D ) and consequently the Bodenstein number (Bo) can ax ity, the delay() function is used here. be determined . The fluidic element considered in this The data is provided via the serial port of the Arduino, use case is a heat exchanger passage with one inlet and one from where it can be accessed by a serial host. Depending outlet for heat exchanger medium flow. Inlet and outlet of on the platform, the data can be directly integrated into an the heat exchanger are connected to an FEP tube (3.2 mm application, or a simple program can be written that allows o.d., 1.6 mm i.d.). Further specifications and details of the for access to the data. The serial interface ensures maximum heat exchanger are beyond the scope of this work. The heat flexibility. In our use case, it was implemented into different exchanger will therefore be treated as a black-box. Two sen- systems such as a custom written process automation system sors (Version A) are clamped to the tubes, one directly at the installed on a computer  and the process automation outlet, and one directly at the inlet (edge to edge). Deion- −1 software LabVision® (HiTec Zang GmbH, Germany). ized water was pumped (320.3 mL min ) through the heat exchanger with a gear pump (Ismatec®ISM446B; Cole- Parmer GmbH, Wertheim, Germany) as process medium. Results from use cases Through a septum, water soluble tracer solution (0.15 mL black ink (Metzger & Mendle GmbH, Fischach, Germany)) In this section four different use cases that have been inves- was quickly (< < 1 s) injected by hand into the process tigated are presented. Qualitative analysis for the sake of medium stream right before the first sensor using a syringe automation but also quantitative analysis has been performed (schematic in Fig. 4). This does influence the flowrate during in different tubular experimental setups. The capability to the injection. The effect of this on the RTD however should distinguish between phases in multi-phase setups, deter- be negligible. Both sensor signals were recorded over time mine the concentration of certain substances in a solution (Fig. 4). and associated opportunities for process automation will be From Fig. 4 a narrow peak is visible that resembles the presented. signal measured at the inlet of the heat exchanger (blue). The tracer travels through the heat exchanger and exits at the Use case 1: Residence time distribution in micro‑ outlet, where sensor 2 recorded the output signal (brown). and milli‑fluidic elements As one can see from the signals the tracer enters the fluidic element with a narrow pseudo-Dirac delta function (unit The residence time of a process fluid in an equipment of impulse), experiences axial mixing, and exits with a sub- interest is an important design parameter. As mentioned stantially broader RTD. before, narrow residence time distributions (RTD) can posi- tively impact the process, such as favoring the main product Use case 2: Enzymatic gas–liquid reactions of a chemical reaction. Considering a tubular reactor with laminar flow accompanied by secondary flow, high radial Biocatalytic reactions are of high interest in industry and mixing -from secondary flow- and low axial dispersion lead research. This is mainly due to the enantio-selective prop- to narrow residence time distributions. Commonly residence erties of enzymes and the mild reaction conditions in an time distributions are determined with sensors that measure aqueous environment [20, 25]. A significant role in the the electrical resistance of the fluid to track a highly con- respecting reactor systems plays the mass transport of the ductive tracer solution as in . Furthermore, tracking the substrate to the enzyme's active site. Therefore, investi- absorption of a tracer with a commercially available ana- gations of different mixing strategies and their influence lytical optical device such as an UV-/Vis is possible . on mass transport are performed with various approaches However, the installation of an UV-/Vis-spectrometer in a . One specific approach makes use of two-phase flow process is not always possible for reasons of available space. induced vortices in capillary flow . The setup used for The same holds for conductivity sensors that even have to be the presented data is already introduced in . Here, the 1 3 190 Journal of Flow Chemistry (2022) 12:185–196 Fig. 4 a Schematic of heat exchanger and sensor setup. b Analog and presented sensor by tracking the intensity of injected dye. Input curve unprocessed sensor signal from the Inlet and Outlet Sensor. c Resi- is not completely shown dence time distribution (RTD) of a heat exchanger measured with the kinetics of the enzymatic oxidation of 2,2′-azino-bis(3- from triple measurements. The absorbance was calculated ethylbenzothiazoline-6-sulfonicacid) (ABTS) (Sigma from the sensors analog signal using Lambert-Beers Law Aldrich, Missouri, USA) is investigated. The catalyzing (Eq. 1). enzyme used is laccase (Sigma Aldrich, Missouri, USA). The experimental data show a linear dependency of the This reaction system is frequently used as a model reac- concentration and the absorption, as it is required for Lam- tion because of the availability of all reactants but even bert–Beer to be applied . By changing the resistance R more due to the striking change in color from colorless it is possible to change the sensor's sensitivity, too, which (ABTS) to blue/green (ABTS ). The chemical analysis is resembled by the slope of the calibration lines in Fig. 4. ox here is done in an optical and non-invasive system with The blue curve shows deviation from a straight line (dot- a dedicated photography setup . Offline techniques ted). A linear dependency for concentrations larger than −1 would require significant interventions to the system with 0.1 mg mL cannot be observed for R = 100 kΩ. However, −1 quenching procedure to stop enzyme activity would also for concentrations below 0.1 mg mL the sensitivity of the be necessary in many cases [20, 25]. As an alternative ana- calibration curve is better compared to the samples evaluated lytic procedure, the optical sensor (version A) discussed with smaller values for R , which show a linear dependency here was integrated into the tubular reaction system. The across the whole measuring range. sensor gives insights into the concentration of ABT S It was possible to show that the sensor can be calibrated ox inside the two-phase reaction system. Figure 5 shows to concentrations of ABTS , which opens up a new method ox the results of a single-phase calibration procedure done for mass transport investigations in tubular reaction systems with different concentrations of ABTS and different light such as in [20, 25]. Furthermore, it was demonstrated that intensities that are varied using different resistors for R the sensitivity of the sensor is strongly dependent on the (cf. Fig. 3b)) to later investigate the two-phase system. intensity of the emitter which was varied by systematically Therefore, the different concentrations of ABTS in deion-changing R . A suitable tradeoff between sensitivity and lin- ized water have been prepared and oxidized using syn- ear dependency can be found in order to identify an operat- thetic air until full conversion can be assumed. The pre- ing window. −1 pared solution was continuously pumped (9.7 mL min ) This can be of particular interest when using new sub- through an FEP tube (3.2 mm o.d., and 1.6 mm i.d.) to stance systems because instead of having to decrease the which the sensor was clamped. To pump the fluid a peri- concentration of a substance of interest by dilution, one can staltic pump (ISM597, Ismatec®, Cole Palmer GmbH, simply decrease the intensity of the emitter to reach the lin- Wertheim, Germany) was used. Data points are achieved ear range of the calibration curve. This could even easily 1 3 Journal of Flow Chemistry (2022) 12:185–196 191 Fig. 5 Calibration curves for different concentrations of ABTS and different values for ox R . The absorbance has been calculated with Lambert- Beers law be integrated into an automation system that checks if the pumped (LabDos easy-load, HiTec Zang GmbH, Germany) collected data agrees with the conditions to apply the Lam- through an FEP tube (3.2 mm o.d., 1.6 mm i.d.). In a tube- bert–Beer Law. in-tube setup cooling medium is pumped (CC304, Huber Kältemaschinenbau AG, Germany) through a bigger tube Use case 3: Suspension detection for continuous in a counter current fashion to be able to control the tem- crystallization perature along the axial coordinate of the crystallizer. As a model system used for characterization experiments the Research interest in continuous (tubular) crystallizers is substance system substance system l -alanine/water has been increasing, especially in the chemical and pharmaceutical used. Further information on the crystallizer can be found field due to product quality consistency and overall effi- in  and . ciency. Examples for these crystallizers are [33, 34], while To prevent the washing liquid from being collected they can also have a CFI design [35, 36]. As reported, the together with the product suspension and possibly affect - generation of solids inside a continuously operated tube is ing the product properties, the described sensor was inte- prone to blocking [33, 36, 37]. Clogging can occur due to grated. It was installed at the end of the CFI, 0.5 m from agglomeration of the solids and growth of crystals on walls, the outlet, before the valve that switches between filling the invasive sensors, and valves that act as mechanical hurdles process medium into the product container and filling it into or thermal bridges to the outside. Therefore, flushing and the waste container. Due to the open-source electronics, it cleaning of the tubular reactors is necessary to maintain the was possible to integrate the Arduino, operating the sensor functionality of the device. Since the demand for continuous version A, into the process control system (PCS) LabMan- automated processes is high, an automated cleaning cycle ager® (HiTec Zang GmbH, Germany), where a logic cou- needs to be integrated into the experimental setup. However, pling between the sensors signal and the valve position was rinsing the reactor with solution or even solvent influences implemented. the product exiting the crystallizer. A possible dilution in the Figure 6b shows the sensor signal plotted over time for a product vessel or subsequent process steps might even lead crystallization process with two cleaning cycles (10–13 min, to the dissolution of product crystals. 18–22 min). The images taken with microscope (Bresser In this use case a continuous tubular crystallizer with a Science ADL 601P, Bresser GmbH, Rhede, Germany) and twin stack CFI design is used (cf. Fig. 6). Process medium is an attached camera (Z6, Nikon GmbH, Tokyo, Japan) show 1 3 192 Journal of Flow Chemistry (2022) 12:185–196 Fig. 6 a Schematic setup of the CFI crystallizer with a camera setup depicts the static threshold used for the evaluation of crystal presence. at the suspension outlet. The sensor (version A) installed close to the The images are exemplary for both states microscope delivers the analog signal shown in b). Here the grey box both states, crystal suspension and solvent. As one can see, injection unit (cf. Fig. 7a) bottom left) into the CFI with flow −1 −1 the sensor signal changes significantly depending on the pre- rates in the range of 40 µL min to 3 mL min by a syringe vailing state enabling the usage of the signal as a trigger for pump (LAMBDA VIT-FIT, LAMBDA Instruments GmbH, the valve position. Baar, Switzerland). For validation experiments a continu- −1 Utilizing this information and the subsequent signal pro- ous phase of colored (0.25 mg mL Sudan Blue II (Sigma cessing of the PCS as well as a suitable valve positioning Aldrich, St. Louis, Missouri, USA)) polydimethylsiloxane strategy, it was possible to separate the solvent for purging (PDMS) (ELBESIL-Öle, B, L. Böwing GmbH, Hofheim, −3 from the product suspension stream in order to increase the Germany, 1 cSt, 945 kg·m ) and a disperse phase of deion- overall product quality and process performance. ized water were used. Using the injection unit, the ADoS is capable of injecting water slugs of desired volume into Use case 4: Automation of continuous liquid–liquid the continuous phase. Figure 7d) shows the signal of water reaction systems slugs of different volume that have been recorded by sensor 2 (Version B) which is positioned at the outlet of the CFI Lab automation has become more and more prominent in (cf. Fig. 7a)). The flowrate during the measurement was set −1 recent years [39, 40]. It opens up potentials concerning to 63.12 µL min . Figure 7d) shows photographs of the decreasing manual and tedious work while maintaining and same slugs also taken after they had passed the reactor. increasing reproducibility and significance of experimental It was not only possible to distinguish the continuous results . Especially for batch screenings pipetting robots from the disperse phase, from the signal it may also be pos- have shown to be a reliable although financially expensive sible to use the sensor in applications where determining the way to handle liquid dosing operations [39, 41]. According length of a slug is of interest. Additionally, aqueous slugs to this trend, an automated dosage system (ADoS) has been with different concentrations of ABT S have been injected ox developed in our research group to increase the throughput into the system following the same procedure as described of chemical batch reactions . before (cf. Fig. 7b)). Here the PDMS has not been colored As a step toward automated continuous processes, a tubu- blue. It can be seen that the concentration of ABTS inside ox lar reactor in form of a CFI (cf. Fig. 7) bottom right) has the slugs causes different signals of the sensor, which makes been coupled to the pipetting robot (cf. Fig. 7a)). The reac- a correlation between both possible. tor consists of 4 coils with 15 turns of FEP tubing (1.6 mm Low flow rates generate high residence times, however, o.d., 1 mm i.d.). Each coil consists of 1 m tube. More spe- good mixing within the slugs can only be achieved with cific information on the reactor configuration can be found high flow rates causing short residence times in the CFI. For in . A continuous phase is pumped through a droplet reactions that should be screened for a longer time than the 1 3 Journal of Flow Chemistry (2022) 12:185–196 193 Fig. 7 a Continuous CFI setup inside a pipetting robot with auto- signal for aqueous ABT S slugs of different concentrations with ox mated dosing system ADoS. Sensor positions for concentration meas- PDMS as continuous phase. c Signal for aqueous slugs of different urements and changing the pump direction are highlighted. b Sensor volume in colored PDMS. d Images of the slugs shown in c actual residence time the presented CFI reactor can provide output. In all presented applications the sensor setup was for, sensor 1 before and sensor 2 after the reactor come into operated by a microcontroller that delivers the output data play. The sensors detect the slugs entering and leaving the stream via a standard serial protocol. This enables to simply reactor and the process automation system automatically read the raw data from the controller, while an integration changes the pump direction of the syringe pump that is feed- into common automation systems is possible for further con- ing the continuous phase in terms of a feedback. By pumping trol purposes. the slugs back and forth between the two sensors a desired The sensor might face limitations when it comes to high residence time can be achieved. concentrations of dissolved analytes due to the nonlinearity of concentration and optical properties in such cases. This is a general issue in optical analysis. The sensor´s working Discussion principle depends highly on the properties of the materi- als used. Not only the tubing material needs to be optically The presented sensor and its use cases, which resemble an accessible but also the fluid of interest needs to be some- extract of possible applications, has shown to be a viable what accessible for (visible) light to be evaluated by the alternative over traditional analytical equipment. Neither sensor. Here, the sensor was used with FEP-tubes with an high amounts of data is generated nor complicated process- outer diameter of 3.2 mm (1.6 mm i.d.) and 1.6 mm (1 mm ing of such is necessary to operate the sensor or evaluate the i.d.), however the sensors geometry is easily modifiable and 1 3 194 Journal of Flow Chemistry (2022) 12:185–196 should work for different tube sizes or view glass arrange - makes it possible to integrate a large number of inline ments as well. Tube material and configurations such as wall sensors to get more data out of a process. To get relevant thickness and opacity can influence the sensor´s sensitivity. information from the data is another step, which has to be In multi-phase systems, where the phases have the same tackled with proper data management and analysis tools or similar optical properties such as absorption maxima and . refractive indices the sensor might reach its limits. The sen- sor signal may only change in a non-distinctive way and not overcome a certain threshold necessary for distinction. Here it was possible to color one of the phases with an inert dye Conclusion & outlook which helped to overcome this challenge. For fully devel- oped slug flows a discrimination between the two phases was The presented optical sensor and its working principle for possible this way. However, a qualitative information about the analysis of capillary single and multi-phase flow phe- complex systems with different phases or concentrations is nomena are an interesting choice compared to common often valuable for better process understanding. analytical methods such as spectrometer and conductivity In this contribution the sensor has been connected to a meter as well as camera setups that generate information flexible Arduino UNO microcontroller reading the data with from images. Although the working principle of using opti- a frequency of up to 50 Hz. For applications that require a cal absorption to infer from the attenuation of light to the faster measurement frequency than 50 Hz it would be possi- composition of the sample is known for a long time , ble to exchange the Arduino Uno with a development board this contribution presents new aspects regarding laboratory that has a faster chipset. Stepping away from the Arduino applications. By detecting differences in light absorption of platform could also have a beneficial impact. Arduinos are samples or process streams the presented sensor provides developed to be user friendly and functional without focus- reliable online process information with a simple electrical ing performance. Standard functions in the Arduino environ- setup. Its capabilities with regards to solid phase detection, ment cause an overhead that can be avoided by removing the process automation, process monitoring for biochemical Arduino bootloader and using the programming language reactions, and determination of RTD characteristics have preferred by the chipset manufacturer. been demonstrated in a quantitative or qualitative way. The overall ease of use, flexibility, cost efficiency, and The installed program making the data accessible is non-invasive character can speed up process develop- easy to understand for non-professionals and adaptable for ment time in different areas. The modular and adaptable specific needs. Further applications would allow for paral- design of the multipurpose sensor enables transferring the lelized systems with more than two sensors connected to working principle to applications and problems that are one microcontroller recoding the intensity of several wave- beyond those presented in this contribution. The “clamp- lengths at once, achieving more specific results of a sample. on” feature makes mounting and dismounting to setups Also, skipping the microcontroller, while connecting the that already contain optically accessible tubular compart- sensor directly to an existing process control system is pos- ments especially easy. Emitter and detector of the same sible. There is a wide variety of possible applications within form factor are available in different configurations (e.g. the field of flow chemistry. Especially the differentiation of emission/ detection spectra and optima) which allows for phases, capabilities in automation, and the determination custom tailored solutions with regard to the optical proper- of RTDs of fluidic elements are of high interest. Straight- ties of emitter, detector, and the substance of interest. This forward software and hardware design enable for the inte- way it would be possible to track concentration changes gration of both into already existing processes constituting of a reactant along a tube with several sensors installed in an attractive possibility to increase process knowledge and series along the length of the tube. It would also be feasi- automation. ble to design the housing of the sensor to fit curved tubes Supplementary Information The online version contains supplemen- so that measurements could also be performed between the tary material available at https://doi. or g/10. 1007/ s41981- 021- 00214-w . individual coils of a CFI at the 90° bends. Applications are conceivable for electromagnetic waves outside the visible Acknowledgements This research was partly supported by the AiF spectrum, too. Another field of application could be the foundation (grant no. ZF4595601LL8 for UC1), the German Research Foundation (DFG, grant no. KO2349/13‐1 for UC2), and German Min- use of the sensor as a tool to classify the product of tubular istry of Economic Affairs and Energy BMWi (ENPRO2.0-TeiA, grant crystallizers. Here the change of absorbance for different no. 03ET1528A). We would like to thank research associates and stu- particles can be utilized.With regards to cost, the sensor dents that let us implement the sensor into their setups or even imple- itself is very affordable with material costs less than 1.5 mented them themselves. These include Mira Schmalenberg, Stephanie Kreis, Julia Grühn, Hüseyin Talha Eroglu, Murat Oruç, Daniel Becker, € per piece (see SI1). The microcontroller is not included and Jan-Hendrik Seifert. because it is optional and not limited to one sensor. This 1 3 Journal of Flow Chemistry (2022) 12:185–196 195 Funding Open Access funding enabled and organized by Projekt 14. Saxena AK, Nigam KDP (1984) Coiled configuration for flow DEAL. inversion and its effect on residence time distribution. AIChE J 30:363–368. https:// doi. org/ 10. 1002/ aic. 69030 0303 15. Sharma L, Nigam K, Roy S (2017) Single phase mixing in coiled Declarations tubes and coiled flow inverters in different flow regimes. Chem Eng Sci 160:227–235. https:// doi. org/ 10. 1016/j. ces. 2016. 11. 034 Conflict of interest The authors have declared no conflict of interest. 16. Kumar V, Aggarwal M, Nigam K (2006) Mixing in curved tubes. Chem Eng Sci 61:5742–5753. https://d oi.o rg/1 0.1 016/j.c es.2 006. Open Access This article is licensed under a Creative Commons Attri-04. 040 bution 4.0 International License, which permits use, sharing, adapta- 17. Castelain C, Berger D, Legentilhomme P et al (2000) Experimen- tion, distribution and reproduction in any medium or format, as long tal and numerical characterisation of mixing in a steady spatially as you give appropriate credit to the original author(s) and the source, chaotic flow by means of residence time distribution measure- provide a link to the Creative Commons licence, and indicate if changes ments. Int J Heat Mass Transfer 43:3687–3700. https:// doi. org/ were made. The images or other third party material in this article are 10. 1016/ S0017- 9310(99) 00363-4 included in the article's Creative Commons licence, unless indicated 18. Kurt SK, Akhtar M, Nigam KDP et al (2017) Continuous reac- otherwise in a credit line to the material. If material is not included in tive precipitation in a coiled flow inverter: inert particle track - the article's Creative Commons licence and your intended use is not ing, modular design, and production of uniform CaCO 3 par- permitted by statutory regulation or exceeds the permitted use, you will ticles. Ind Eng Chem Res 56:11320–11335. https:// doi. org/ 10. need to obtain permission directly from the copyright holder. To view a 1021/ acs. iecr. 7b022 40 copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . 19. Engl W, Tachibana M, Colin A et al (2008) A droplet-based high-throughput tubular platform to extract rate constants of slow chemical reactions. Chem Eng Sci 63:1692–1695. https:// doi. org/ 10. 1016/j. ces. 2007. 11. 006 20. Grühn J, Vogel M, Kockmann N (2021) Digital image process- References ing of gas-liquid reactions in coiled capillaries. Chem Ing Tech 93:825–829. https:// doi. org/ 10. 1002/ cite. 20200 0240 1. Baumann M, Moody TS, Smyth M et al (2020) A perspective on 21. Vietinghoff N von, Lungrin W, Schulzke R et al. (2020) Photo- continuous flow chemistry in the pharmaceutical industry. Org electric sensor for fast and low-priced determination of Bi- and Process Res Dev 24:1802–1813. https://doi. or g/10. 1021/ acs. opr d. triphasic segmented slug flow parameters. Sensors 20.https:// 9b005 24 doi. org/ 10. 3390/ s2023 6948 2. Nagy ZK, El Hagrasy A, Litster J (2020) Continuous pharmaceuti- 22. Demori M, Ferrari V, Strazza D et al (2010) A capacitive sensor cal processing. Springer, Cham system for the analysis of two-phase flows of oil and conductive 3. Plumb K (2005) Continuous processing in the pharmaceutical water. Sens Actuators A 163:172–179. https://doi. or g/10. 1016/j. industry. Chem Eng Res Des 83:730–738. https://doi. or g/10. 1205/ sna. 2010. 08. 018 cherd. 04359 23. Wang YX, Ji HF, Huang ZY et al (2017) Online measurement of 4. Grundemann L, Schoenitz M, Scholl S (2012) Shorter time-to- conductivity/permittivity of fluid by a new contactless imped- market with micro-conti processes. Chem Ing Tech 84:685–693. ance sensor. Rev Sci Instrum 88:55111. h t t p s : / / d o i . o r g / 1 0 . https:// doi. org/ 10. 1002/ cite. 20110 0238 1063/1. 49832 08 5. Schönfeld F, Hardt S (2004) Simulation of helical flows in micro- 24. Fan S, Yan T (2014) Two-phase air-water slug flow measure - channels. AIChE J 50:771–778. https://d oi.o rg/1 0.1 002/a ic.1 0071 ment in horizontal pipe using conductance probes and neural 6. Hessel V, Löwe H, Schönfeld F (2005) Micromixers—a review network. Trans Instrum Meas 63:456–466. https:// doi. org/ 10. on passive and active mixing principles. Chem Eng Sci 60:2479– 1109/ TIM. 2013. 22804 85 2501. https:// doi. org/ 10. 1016/j. ces. 2004. 11. 033 25. Bobers J, Grühn J, Höving S et al (2020) Two-phase flow in a 7. Mridha M, Nigam K (2008) Coiled flow inverter as an inline coiled flow inverter: process development from batch to con- mixer. Chem Eng Sci 63:1724–1732. https://do i. org/ 10. 1016/j. tinuous flow. Org Process Res Dev 24:2094–2104. https:// doi. ces. 2007. 10. 028 org/ 10. 1021/ acs. oprd. 0c001 52 8. Jähnisch K, Hessel V, Löwe H et al (2004) Chemistry in micro- 26. Glotz G, Kappe CO (2018) Design and construction of an open structured reactors. Angew Chem Int Ed Engl 43:406–446. https:// source-based photometer and its applications in flow chemis- doi. org/ 10. 1002/ anie. 20030 0577 try. React Chem Eng 3:478–486. https://doi. or g/10. 1039/ C8RE0 9. Hessel V, Löwe H (2003) Microchemical engineering: compo- 0070K nents, plant concepts, user acceptance – part III. Chem Eng Tech- 27. Perkampus H-H (1986) UV-VIS-Spektroskopie und ihre nol 26:531–544. https:// doi. org/ 10. 1002/ ceat. 20039 0079 Anwendungen. Springer, Berlin 10. Kumar V, Mridha M, Gupta AK et al (2007) Coiled flow inverter 28. Mayerhöfer TG, Popp J (2019) Beer’s law - why absorbance as a heat exchanger. Chem Eng Sci 62:2386–2396. https://d oi.o rg/ depends (almost) linearly on concentration. ChemPhysChem 10. 1016/j. ces. 2007. 01. 032 20:511–515. https:// doi. org/ 10. 1002/ cphc. 20180 1073 11. Schmalenberg M, Krieger W, Kockmann N (2019) Modular coiled 29 Skoog DA, West DM, Holler FJ (2013) Fundamentals of analyti- flow inverter with narrow residence time distribution for process cal chemistry, 9th edn. Cengage, Belmont development and production. Chem Ing Tech 91:567–575. https:// 30. Soboll S, Bittorf L, Kockmann N (2018) Axial backmixing and doi. org/ 10. 1002/ cite. 20180 0172 residence time distribution in a miniaturized, stirred-pulsed 12. Mandal MM, Aggarwal P, Nigam KDP (2011) Liquid-liquid mix- extraction column. Chem Eng Technol 41:134–142. https://doi. ing in coiled flow inverter. Ind Eng Chem Res 50:13230–13235. org/ 10. 1002/ ceat. 20170 0152 https:// doi. org/ 10. 1021/ ie200 2473 31. Schmalenberg M, Weick LK, Kockmann N (2021) Nuclea- 13. Mansour M, Thévenin D, Zähringer K (2020) Numerical study of tion in continuous flow cooling sonocrystallization for coiled flow mixing and heat transfer in helical pipes, coiled flow invert- capillary crystallizers. J Flow Chem. https:// doi. org/ 10. 1007/ ers and a novel coiled configuration. Chem Eng Sci 221:115690. s41981- 020- 00138-x https:// doi. org/ 10. 1016/j. ces. 2020. 115690 1 3 196 Journal of Flow Chemistry (2022) 12:185–196 32. Levenspiel O (1999) Chemical reaction engineering, 3rd edn. on Nanochannels, Microchannels and Minichannels, June 10–13, Wiley, New York 2018, Dubrovnik, Croatia. The American Society of Mechanical 33. Eder RJP, Radl S, Schmitt E et al (2010) Continuously seeded, Engineers, New York continuously operated tubular crystallizer for the production of 39. Prabhu GRD, Urban PL (2017) The dawn of unmanned analyti- active pharmaceutical ingredients. Cryst Growth Des 10:2247– cal laboratories. Trends Anal Chem 88:41–52. https://doi. or g/10. 2257. https:// doi. org/ 10. 1021/ cg901 57881016/j. trac. 2016. 12. 011 34. Besenhard MO, Neugebauer P, Ho C-D et al (2015) Crystal size 40. Urban PL (2018) Prototyping instruments for the chemical labora- control in a continuous tubular crystallizer. Cryst Growth Des tory using inexpensive electronic modules. Angew Chem Int Ed 15:1683–1691. https:// doi. org/ 10. 1021/ cg501 637m Engl 57:11074–11077. https:// doi. org/ 10. 1002/ anie. 20180 3878 35. Hohmann L, Greinert T, Mierka O et al (2018) Analysis of 41. Kong F, Yuan L, Zheng YF et al (2012) Automatic liquid handling crystal size dispersion effects in a continuous coiled tubular for life science: a critical review of the current state of the art. crystallizer: experiments and modeling. Cryst Growth Des J Lab Autom 17:169–185. https:// doi. org/ 10. 1177/ 22110682 11 18:1459–1473. https:// doi. org/ 10. 1021/ acs. cgd. 7b013 83 435302 36. Hohmann L, Schmalenberg M, Prasanna M et al (2019) Suspen- 42. Bobers J, Škopić MK, Dinter R et al (2020) Design of an auto- sion flow behavior and particle residence time distribution in mated reagent-dispensing system for reaction screening and vali- helical tube devices. Chem Eng J 360:1371–1389. https:// doi. dation with DNA-tagged substrates. ACS Comb Sci 22:101–108. org/ 10. 1016/j. cej. 2018. 10. 166https:// doi. org/ 10. 1021/ acsco mbsci. 9b002 07 37. Wang Y, Su M, Bai Y (2020) Mechanism of glycine crystal adhe- 43. Wulf C, Beller M, Boenisch T et al (2021) A unified research data sion and clogging in a continuous tubular crystallizer. Ind Eng infrastructure for catalysis research – challenges and concepts. Chem Res 59:25–33. https:// doi. org/ 10. 1021/ acs. iecr. 9b059 77 ChemCatChem 13:3223–3236. https://doi. or g/10. 1002/ cctc. 20200 38. Schmalenberg M, Hohmann L, Kockmann N (2018) Miniaturized 1974 Tubular Cooling Crystallizer With Solid-Liquid Flow for Process Development. In: Proceedings of the ASME 16th International Publisher’s note Springer Nature remains neutral with regard to Conference on Nanochannels, Microchannels and Minichannels jurisdictional claims in published maps and institutional affiliations. 2018: Presented at ASME 2018 16th International Conference 1 3
Journal of Flow Chemistry – Springer Journals
Published: Jun 1, 2022
Keywords: Multiphase flow; Non-invasive sensor; Online concentration measurement; Process automation; Light dependent resistor; Residence time distribution
Access the full text.
Sign up today, get DeepDyve free for 14 days.