Microwave-assisted heating in a novel thin film-liquid spinning coaxial reactor

1.   Introduction

The development of green chemistry technologies requires new processes and hardware, allowing increased conversion rates, improved product selectivity and yields, enhanced safety conditions, decreased energy usage, and low product cost ^[1,2,3,4]. These technologies often include systems controlled through wireless networks in human-free conditions. For instance, there is significant interest in launching chemical laboratories into space for enhanced synthesis of biochemical products ^[5,6,7].

Accelerating chemical processes can be reached by different means, guided by a modified Arrhenius equation ^[8]:

where $K$ is the rate constant, i.e., frequency of collisions resulting in a reaction, and $\rho$ is the steric factor depending on the shape of interacting molecules $A$ and $B$ and their orientations at a collision. ${Z_{{\text{AB}}}}$ is the number of binary collisions of molecules $A$ and $B$, ${E_a}$ is the chemical reaction activation energy, $R$ is the Avogadro constant, and $T$ is the absolute temperature. It is clear that the main influencing factor of the reaction rate is temperature.

Another critical factor is the number of collisions ${Z_{AB}}\left(T \right)$, dependent on the delivery of molecules to the interaction points. This is achieved by preliminary mixing of reagents in a solvent using laminar, non-laminar, diffusion, or boiling flows in batch chemistry.

Rotational mixing is routinely used in liquid chemistry ^[9,10]. Among these devices are tube-in-tube reactors, distinguished by increased conversion rates. They are built on an inner rotating cylinder inserted into another stator tube. The heated mixture is pumped between the two mentioned cylinders. If the mixture's viscosity is relatively high and the gap is large enough, Taylor-Couette vortices are generated along the reactor, facilitating better mixing and increasing the conversion rate ^{[9,10,11,12,13]}.

Different designs have been proposed for thin-film liquid tube-in-tube reactors ^[9,12,13,14]. These reactors pump fluids in a narrow sub-millimeter gap, allowing the generation of micrometric vortices for fine mixing of externally heated reagents.

On the contrary, microwave (MW)-assisted chemistry involves heating polar liquids directly in reactors ^[10,15,16]. Unfortunately, the conventional MW ovens based on hollow cavities are unsuitable for tailored chemistry because of their employed frequency of 2.45 GHz; they also require a large amount of energy, even for synthesis in small vials. Miniature and energy-efficient MW reactors can be built on transversal electromagnetic (TEM) waveguides and semiconductor generators. Some designs use coaxial waveguides ^{[17,18,19,20,21,22,23,24,25,26,27,28,29]}, which allow scalability ^[18,20]. For high-loss liquids, the residence time in reactors is increased with dielectric-layered quasi-TEM coaxial waveguides ^[26,28].

MW coaxial designs are well-suited for rotating thin-film liquid reactors. The first device of this kind was proposed in ^[29], featuring a central conductor inserted into a hollow dielectric rotor. The liquid flows in the narrow gap between this rotor and the outer shield under MW heating. Enhanced conversion rates are expected due to improved mixing by exciting the sub-millimeter liquid vortices in the narrow gap ^[9] and specific MW effects in microchannels ^[30].

This paper aims to confirm the workability of the design proposed in ^[29] and investigate key effects and challenges for optimizing the thin-film MW-assisted spinning coaxial reactors.

2.   Microwave spinning coaxial reactor

One of the developed devices, Kinel-2.0, is shown in Figures 1 and 2. It consists of a spool-like coaxial reactor with installed alumina (Appendix, Table 1A, row 1) or quartz (Table 1A, row 2) tube, closed at one end and rotating around the central conductor (shown in black in Figure 2) using a flexible coupler by an electromotor (Table 1A, row 3). This wire conductor extends inside the tube to a length ${H_c}$ and serves as an open-end applicator, heating the liquid as it moves from the inlet to the outlet through the circular slot between the spinning rotor and coaxial shield (Figure 2). The slot size (<1 mm) and rotation speed $\left({ \approx 3{\text{ }}000 - 4{\text{ }}000{\text{ rpm}}} \right)$ were chosen according to literature data ^[9,12,13,14] on non-microwave spinning reactors.

This reactor is fed by MWs from below the board using a coaxial connector (Table 1A, row 4). Rubber rings (shown in blue in Figure 2) seal this reactor from both ends, and the caps support a certain pressure to ensure hermeticity. Wetting of the upper ring is achieved by maintaining the required liquid pressure internally and/or with liquid provided through the circular cap slot between the rotor and the cap wall.

Microwaves are initially excited in the reactor through the mentioned sealed coaxial connector in the cavity below the rotor end (Figure 2) as a transverse electromagnetic (TEM) wave. The wave then travels to the partially filled coaxial waveguide. Its cross-section is mainly occupied by the dielectric rotor. This waveguide's primary mode is a quasi-TEM type. This wave is reflected from the end of the central rod, and power partially comes to the layered lossy circular waveguide, whose size dampens the propagation of its main mode. The cross-sections of the inlet/outlet channels are too small for wave propagation.

3.   Measurement setup

Figures 3 (block diagram) and 4 (physical implementation) illustrate the setup for measuring the temperature of liquids in MW spinning reactors. All commercially available component data are given in Appendix, Table 1A.

The setup consists of an MW generator (Table 1A, row 5) whose power can be adjusted from 0 to 250 W in 10 W increments, controlled by a computer. Approximate output data and reflected power values are available directly from this generator. More accurate power measurement results are available in the reflectometer (Table 1A, row 6), which samples small portions of the generator's and reflected powers, relayed through coaxial cables (Table A1, row 7) to power meters PM-I and PM-R (Table 1A, rows 8 and 9). Unfortunately, this generator lacks fine-tuning capabilities for precise power adjustment using its digital interface. Besides, like most high-power generators and amplifiers, it may show some thermal noise and power level drift, especially in multi-hour experiments. Nevertheless, this instrument, intended for MW heating, provided acceptable performance in our research.

The attenuators ${A_1}$ and ${A_2}$ (Table 1A, rows 10 and 11) adjust the incoming levels for accurate measurements. A coaxial cable connects the reflectometer to the reactor (Table 1A, row 12). A polar liquid from vessel 1 reaches the reactor using a pump (Table 1A, row 13) and tubing (Table 1A, row 14). At the outlet, the temperature is measured by an optic fiber thermometer (O-TM; Table 1A, row 15), whose sensor is installed directly in the reactor cavity near the rotating dielectric cylinder. This sensor is not influenced by MW irradiation.

The heated liquid is evacuated to a glass vessel 2 using Tygon tubing. A DC source (24 V) of a regulated and controlled current (0–15 A) feeds the reactor's electromotor. The speed of dielectric rotors is measured using a non-contact tachometer (Table 1A, row 16).

The level of possible parasitic MW irradiation is measured by a power indicator (Table 1A, row 17). Observed levels were significantly below the maximum allowed power density of 5 mW/cm², even close to the reactor and connectors during heating experiments.

The frequency properties of the reactors filled with ethanol (Table 1A, row 18) and methanol (Table 1A, row 19) at room temperature were analyzed preliminary using a network analyzer (Table 1A, row 20). The last three instruments are not shown in Figures 3 and 4.

All experiments were performed at the Department's Chemical Laboratory, equipped with air-hood chambers and safety hardware.

4.   Materials and components

Details and characteristics of all used materials and components are given in Tables 1, 2, and Appendix 1A.

4.1.   Reference liquids

As reference liquids, ethanol $\left({{{\text{C}}_{\text{2}}}{{\text{H}}_{\text{5}}}{\text{OH}}} \right)$ and methanol $\left({{\text{C}}{{\text{H}}_{\text{3}}}{\text{OH}}} \right)$ were used ^[31,32]. The dielectric parameters of these liquids depend on frequency and temperature. At 2.5 GHz and 20 ℃, the complex relative permittivity of ethanol is $\tilde \varepsilon _{{\text{ET}}}^{\left(r \right)} \approx 6.59 - j6.30$; for methanol, this parameter is $\tilde \varepsilon _{\left({{\text{MT}}} \right)}^{\left(r \right)} \approx 21.30 - j13.88$ ^[31]. The first one is more lossy due to comparable real and imaginary permittivity components.

The parameters mentioned above depend on the inner structure of the alcohol molecules and their interactions with each other. For instance, ethanol molecules' polarity is weaker than methanol due to longer carbon chains, resulting in a lower real permittivity. Ethanol has six hydrogen atoms, which can be coupled to neighboring molecules. The boiling point of ethanol (78.23 ℃) is higher than that of methanol (64.7 ℃), which has only four hydrogen atoms. These interactions define other parameters of these liquids, as shown in Table 1.

4.2.   Reactor materials

In this paper, the developed reactor was made of brass, whose specific electric conductivity is 28% that of copper (5.96 × 10⁷ Sim/m). The central conductor was stainless steel ($1.45 \times {10^7}{\text{ }}{{{\text{Sim}}} \mathord{\left/ {\vphantom {{{\text{Sim}}} {\text{m}}}} \right. } {\text{m}}}$), chosen for its mechanical strength. The rotors are made from alumina ceramic or quartz. Their electric parameters are shown in Table 2.

5.   Results

5.1.   Practical matching of reactors

The design is based on an open-end central conductor of a coaxial waveguide (Figure 2). Untuned reactors have increased reflection ^[32,33]. Some additional mismatching occurs due to the parasitics from connectors, reflectometers, and cables, as well as due to the temperature drift of dielectric parameters of heated liquids ^[32]. The narrow bandwidth matching leads to transients near generator activation. Due to the inertia of the heating process, slight variations of incident power during the heating process do not influence the temperature of a liquid at the outlet.

Reactors can be matched using the outer mechanicalor electromechanical tuners or by varying the geometry of heaters ^[32]. Particularly, in ^[33], a technique was used to change the rotor's immersed length inside the spool reactor (Figure 2) when filled with a liquid at room temperature.

In this experiment, two different rotors were used. One is made of alumina (Table 1A, row 1), and another of quartz (Table 1A, row 2). A calibrated network analyzer is connected directly to the reactor filled with room-temperature liquid. Measurements show that the untuned geometry $h = 0{\text{ mm}}$ (see Figure 2) leads to strong reflection.

Similarly to ^[33], the tuning of the reactor is achieved by changing the height $h$ of the mini-cavity below the dielectric rotor (Figure 2). Particularly, for an alumina tube, $h = 14{\text{ mm}}$ was used; for quartz, $h = 10{\text{ mm}}$. In this case, the reflection coefficient calculated according to the measured power data (${P_{\text{G}}}\left({t, {t^ \circ }} \right)$ and ${P_{{\text{ref}}}}\left({t, {t^ \circ }} \right)$) at the reflectometer ports (Figures 3 and 4) is within the 25%–35% limit (Figures 5–11), which is acceptable for the research of these reactors.

This simple technique has its advantages and disadvantages. For instance, the liquid-filled cavity below the rotor's end may hinder its geometrical scalability due to the complicated physics of multiphase-heated liquids. Outer matching may be preferable for finalized designs with a fixed minimal length of cavities below the end of the rotors.

5.2.   Results of heating experiments

Temperature measurements of heated ethanol and methanol were performed at the reactor liquid outlet (Figures 2, 3). The rate of a ten-roller pump was supported at 1.5 mL/min for all measurements. The reactor was connected to this pump by a transparent Tygon tube of 1 m length to minimize the rest of the liquid pulsation originating from the peristaltic pumping. Each measurement starts with a preliminary cooled reactor at room temperature.

5.2.1.   Heating experiments with ethanol and ceramic rotor reactor

Consider some results on ethanol heating in the reactor with a ceramic rotor. Initially, this device was studied in the static rotor regime, applying different generator power levels. For instance, the heating experiment for the generator power around 70 W is shown in Figure 5.

Transient behavior is found in the applied and reflected power trends near the switching time of the generator (Figure 5(a)). An associated temperature jump is seen in Figure 5(b) with a slight delay, meaning that the liquid is heated by microwaves, even at the end of the reactor, close to the outlet tube.

Some quasiperiodic few-degree temperature oscillations (Figure 5b) occurred despite relatively steady power trends (Figure 5a). A similar behavior of the decreased magnitude was found for the lower applied power levels; this is shown in Figure 6, where the generator's power is around 60 W.

The nature of these oscillations is not definitely known. One reason is intensive bubbling in the cavity below the rotor end, where the power density of the applied MW field is very high. Also, these bubbles are seen along the used outlet Tygon transparent tube, although the liquid temperature is reduced due to heat exchange with the reactor body. Similar behavior was found in ^[32], where the heating of liquids in plastic tubes was studied. Another possible cause is the overheating of liquids in the mentioned cavity and the slugged motion of overheated drops to the outlet.

These quasi-periodical effects diminished in the rotating regime, shown in Figure 7. Here, the reflected power is much noisier due to the non-ideal centering of the design.

Despite the intensive generator's thermal noise arising after several hours of experiments, this temperature trend does not show the mentioned quasi-periodical oscillations explained by improved mixing ethanol in the cavity below the rotor and in the narrow cylindrical gap (0.5 mm) between the dielectric rotor and reactor wall (Figure 7). More research is required to precisely establish the nature of this quasi-periodical effect in temperature trends.

5.2.2.   Heating experiments with methanol and ceramic rotor reactor

Similar experiments were performed with methanol. Figure 8 shows the power and temperature curves for the methanol-filled ceramic rotor reactor.

Applying power around 80–72 W reaches the methanol boiling temperature measured at the outlet channel. In this case, the influence of bubbling on temperature is less visible. Still, the generator power is noisier than in Figure 5, and its level cannot be tuned precisely due to the generator design.

Figure 9 shows the methanol heating experiments with a spinning ceramic rotor. Methanol's trend pulsations are less influenced by the rotation than in Figure 6, mostly because this liquid has twice less viscosity than ethanol (Table 1). Excitation of micro-vortices is less effective in this case.

In MW heating conditions, the temperature of liquids can surpass boiling temperature even at normal atmospheric pressure ^[32]. Figure 9 shows a similar effect that can be additionally strengthened by increasing liquid pressure inside the reactor and liquid degassing.

5.2.3.   Heating experiments with ethanol and quartz rotor reactor

Some studies were performed with static (Figure 10) and rotating (Figure 11) quartz tubes and ethanol. Figure 10 shows the power (a) and temperature (b) trends for the static state of the quartz rotor. The behavior (occlusion effect) of liquid motion is similar to what was seen in Figure 5.

This occlusion effect is diminished in the rotating regime (Figure 11). By analyzing Figures 5 and 10, we need to consider the surface quality of alumina and quartz rotors; the first one has increased roughness (estimated visually), which may cause more effective excitation of liquid micro-vortices.

6.   Discussion

In this paper, a novel reactor was studied for liquid-heating experiments using a high-power MW generator, power meters, and an optic fiber thermometer resistant to MW irradiation. This study aimed to confirm the proposed reactor's workability in the dielectric rotor's static and spinning regime immersed in a sealed coaxial cavity ^[29,33]. This research was performed in realistic conditions, in which the high-power generator had limitations in fine-tuning power levels, some thermal noise, and power drift.

The manufactured reactor design also presented its own limitations in its MW and mechanical tuning. For instance, the used MW matching allowed only the regulation of the reactor reflection with the joined cables and connectors in a limited manner; ideally, it requires the outer tuners to adapt toward the input impedance temperature drift. In this case, the relatively large cavities with complicated two-phase liquid dynamics could be deleted from the design to improve the predictability of thermal processes.

Non-ideal centering may lead to vibration of the central rod of the coaxial cavity and a noise-like response of the reflected power (see Figures 7 and 9). This vibration can be diminished by better centering using PTFE bearings. Using massive metallic parts of the reactor body exposed to convective air leads to slow warming of the flowing liquids in the thin channel between the rotor and the coaxial outer shield. Glass-metal technology can alleviate this problem ^[27,28].

Nevertheless, the studied reactor showed stable heating of ethanol and methanol in the static and spinning regimes of ceramic and quartz rotors. Spinning rotors damped the slugging effects and destroyed large bubbles of liquid vapor in the mentioned channel. The proposed design is open to moderate scalability if an outer tuning is applied without keeping large cavities below the rotor's end.

7.   Conclusions

In this work, a novel microwave thin-film liquid spinning coaxial reactor was developed, manufactured, and studied in heating experiments. It consists of a coaxial vessel sealed from both ends, where the central conducting rod is installed into a hollow dielectric rotor with a regulated immersion length. This coaxial vessel is excited from the one end by microwaves using a sealed connector. The liquids are pumped to the reactor and following further to the sub-millimeter cylindrical gap between the stator (outer conductor of a coaxial waveguide) and dielectric rotor and heated. While the dielectric cylinder is rotating at high speed (2000–4000 rpm), fine turbulent flow arises due to the interaction of liquids with the rotating dielectric surface, which improves the mixing of reagents as shown in known non-microwave reactors of this type ^[9].

The experiments with MW-assisted heating ethanol and methanol confirmed the workability of the proposed device in the static and spinning regimes of the alumina and quartz rotors. For instance, rotating dampens the occlusion effect when the liquid flow is slugged and liquid drops are moving fast under the vapor pressure. Further design and developments will be done to enhance these reactors, with numerical simulations and experimental heating studies in different regimes, including analyzing bubbling and overheating effects. The scalability of this hardware should also be investigated.

The thin-film liquid spinning coaxial reactors are promising for rapid chemical synthesis where acceleration is achieved due to the fine mixing of reagents ^[9] and specific MW effects in sub-millimeter channels ^[30]. Assessing these reactors in increased safety environment applications, including space applications ^[5,6], is also fascinating.

Acknowledgments

The Author thanks the Small-Scale Project Fund (Information Technology and Electrical Engineering Faculty, NTNU) for financial support and Dr. S. Pettersen for his help in Chemical Laboratory.

He thanks the unknown reviewers for their critiques and advice.

Conflict of interest

The author declares that there is no conflict of interest in this paper.

Appendix