Comparative investigation of fine bubble and macrobubble aeration on gas utility and biotransformation productivity

The sufficient provision of oxygen is mandatory for enzymatic oxidations in aqueous solution, however, in process optimization this still is a bottleneck that cannot be overcome with the established methods of macrobubble aeration. Providing higher mass transfer performance through microbubble aerators, inefficient aeration can be overcome or improved. Investigating the mass transport performance in a model protein solution, the microbubble aeration results in higher k L a values related to the applied airstream in comparison with macrobubble aeration. Comparing the aerators at identical k L a of 160 and 60 1/h, the microbubble aeration is resulting in 25 and 44 times enhanced gas utility compared with aeration with macrobubbles. To prove the feasibility of microbubbles in biocatalysis, the productivity of a glucose oxidase catalyzed biotransformation is compared with macrobubble aeration as well as the gas ‐ saving potential. In contrast to the expectation that the same productivities are achieved at identically applied k L a , microbubble aeration increased the gluconic acid productivity by 32% and resulted in 41.6 times higher oxygen utilization. The observed advantages of microbubble aeration are based on the large volume ‐ specific interfacial area combined with a prolonged residence time, which results in a high mass transfer performance, less enzyme deactivation by foam formation, and reduced gas consumption. This makes microbubble aerators favor-able for application in biocatalysis.


| INTRODUCTION
To compare a novel fine bubble with classic macrobubble aeration techniques and prove the feasibility for application in biocatalysis, this contribution is focusing on oxygen as a gaseous substrate.
Oxygen is by mass the third most abundant element in the universe and essential for a majority of life forms on earth. The air we are breathing is containing 0.21 vol.% oxygen (Weiss, 1970), which is a comparatively cheap and simply accessible natural source of oxygen for application in oxidation reactions (Toftgaard Pedersen et al., 2017). In industry, oxygen is employed as an oxidant in several applications from enzymatic oxidation reactions, biological aerobic wastewater treatment to whole-cell aerobic fermentation (Chapman, Cosgrove, Turner, Kapur, & Blacker, 2018;Hone & Kappe, 2018;Terasaka, Hirabayashi, Ai, Nishino, Fujioka, & Kobayashi, 2011). To make use of oxygen in enzymatic oxidation reactions, it has to be dissolved in a liquid phase to become accessible (Gemoets, Hessel, & Noël, 2016;Woodley, 2019). These reactions are often accompanied by mass transport limitations caused by the low oxygen solubility in aqueous media (Chapman et al., 2018;Garcia-Ochoa & Gomez, 2009). Therefore, the oxygen mass transfer performance of established aeration techniques are usually the bottleneck in process optimizations (Chapman et al., 2018;Romero, Gómez Castellanos, Gadda, Fraaije, & Mattevi, 2018). Considering this, there is a need for highly efficient aeration techniques in the industry. Requirements for these aeration systems are a high mass transfer performance, efficient gas utilization, low-pressure drop, low shear stress, and reduced foaming tendency (Terasaka, Hirabayashi, Nishino, Fujioka, & Kobayashi, 2011). One approach to achieve these goals is the costintensive application of pressure to enhance oxygen solubility in the system (Bolivar, Mannsberger, Thomsen, Tekautz, & Nidetzky, 2019;Xing et al., 2014). Another option is the bubble-free aeration through permeable or porous materials, which is challenged by the diffusion rate of oxygen through the membrane material (Kaufhold et al., 2012;Rissom, Schwarz-Linek, Vogel, Tishkov, & Kragl, 1997). A third possibility is aeration with fine bubbles, which are defined in the ISO 20480-1 to be a gaseous dispersed phase with a volume equivalent diameter of less than 100 µm (Tsuge, 2014). They are further differentiated in ultra-fine and microbubbles, which are less than 1 µm and between 1 µm and 100 µm of size, respectively. The generation of air containing fine bubbles through porous spargers, which are capable of generating microbubbles, are in the focus of this contribution. The mass transfer of fine bubbles as well as macrobubble aeration could further be enhanced by aeration with pure oxygen or enriched air (Lee, 2020;Lindeque & Woodley, 2020). Fine bubbles are successfully applied for the defouling of various surfaces, cleaning of sanitary facilities and in flotation processes (Tao, 2005;Wang, Wang, Yan, Wang, & Cao, 2017), and wastewater treatment (Wang et al., 2019;Khuntia, Majumder, & Ghosh, 2012;Temesgen, Bui, Han, Kim, & Park, 2017). The reported charged surface of small bubbles results in a lower coalescence, bursting tendency, and increased foam formation (Collins, Motarjemi, & Jameson, 1978;Huo et al., 2019;Oliveira & Rubio, 2011). In consequence, the introduced shear forces compared to macrobubbles are reduced as illustrated in Figure 1 (Duval, Adichtchev, Sirotkin, & Mermet, 2012;Ushikubo et al., 2010). Nevertheless, the application of microbubble instead of macrobubble aeration is known to reduce the foaming tendency as a result of high mass transfer and lower gassing rates (Tsuge, 2014). It seems that microbubbles completely dissolve in to the surrounding medium, which avoids or reduces the need for antifoaming agents (Guy, Carreau, & Paris, 1986;Jia, Xiao, & Kang, 2019).
Microbubble aeration offers several advantages due to its simple generation and promising high dissolution rates in aqueous media (Iwakiri et al., 2017;Jia et al., 2019). This enhanced mass transfer performance results first from the high volume-specific interfacial area and second from the prolonged residence time by low rising velocity as shown in Figure 1 (Duval et al., 2012;Struthwolf & Blanchard, 1984;Terasaka et al., 2011). Especially, the high interfacial area leads to the decrease in bubble size (shrinking) by quick dissolution, which is resulting in an exponential increase of the Young-Laplace pressure (inner bubble pressure), when the bubble size is decreased towards the microbubble region (Iwakiri et al., 2017). The bubble inner pressure is linked to the concentration gradient by the law of Henry (Atkins, de Paula, & Michael, 2013) in the boundary layer of the bubble. According to Henry's law, the partial pressure difference is the driving force for mass transferred from the bubble into the bulk phase which is the fundamental reason of fast shrinking and quick dissolution of microbubbles (Epstein & Plesset, 1950;Iwakiri et al., 2017;Ludwig & Macdonald, 2005). To make use out of these promising properties of microbubbles, the aeration has to be implemented in established reactor types, such as bubble column and stirred tank reactor (STR; B. Liu et al., 2019). The behavior of microbubbles is different compared with classic macrobubble aeration under stirrer induced shear forces, because of their lower volume, immobile bubble surface, and higher residence time (Liu, Zhang, Zhang, Yang, & Zhang, 2014;Struthwolf & Blanchard, 1984). For macrobubbles, the mass transfer performance is depending on the dispersion of the gaseous phase by agitation induced  shear forces (Gaddis, 2013;B. Liu et al., 2019). This is often coupled to high energy input, inefficient gas consumption, and waste of large amounts of gas (B. Liu et al., 2019;Terasaka et al., 2011). In consequence of the high gassing rates and energy input, macrobubble systems are simultaneously mixed extensively. In contrast, microbubbles have a low terminal rising velocity and provide, therefore, less efficient mixing induced by the applied gas stream (Parkinson, Sedev, Fornasiero, & Ralston, 2008). To ensure proper mixing conditions in microbubble aerated reactions, some agitation is needed. Thus, there is a potential to reduce the necessary energy input to a minimum.
For the investigation of microbubble application in biotransformation, an enzymatic model reaction system is studied in an aerated STR set-up. This study is focusing on the comparison of micro-and macrobubbles generated in an aerated STR (Figure 2).
To investigate the advantages of microbubble aeration techniques, the oxygen-limited biotransformation of β-D-glucose with oxygen towards gluconic acid and hydrogen peroxide catalyzed by glucose oxidase (GOx) from Aspergillus niger is chosen. The productivity (after 3.5 h), gas utilization, and mass transport performance of macro-and microbubble aeration are compared. To get a deeper understanding and interpretation of the effects influencing the productivity after 3.5 h, a detailed characterization of the model enzyme GOx Type VII from A.
niger is carried out.

| Chemicals, enzymes and proteins
GOx from A. niger Type VII, peroxidase (POD) Type II from horseradish with 450 U/mg, catalase from the bovine liver with 5500 U/mg, and o-dianisidine more than or equal to 98% POD substrate were purchased from Sigma-Aldrich/Merck. The bovine serum albumin (BSA) was purchased from Carl Roth. The reaction media were composed of demineralized water, D(+)-glucose p.a. (Carl Roth), sodium acetate more than or equal to 99% (Sigma-Aldrich/Merck), acetic acid ROTIPURAN® 100% (Carl Roth), and sodium hydroxide more than or equal to 98% (Carl Roth).

| Devices
A UV/VIS photometer UvikonXL including temperature control was used in the activity assay. Incubation was carried out in a 500 ml glass reactor including a heating jacket connected to a Lauda 003 heating

| Photometric activity assay
The determination of the initial enzyme activity was carried out based on the GOx/POD assay by Bateman and Evans. A substrate F I G U R E 2 Concept of enzymatic sodium gluconate synthesis in a two-phase system. The boundary layer is indicated with dashed lines solution containing 0.21 mM o-dianisidine and 9.44 mM D-glucose in 10 mM sodium acetate buffer pH 5.3 was prepared. The samples contained 936 µl substrate solution mixed with 32 µl POD solution (60 U/ml) and 32 µl GOx solution (40 U/ml). As reference measurement, a composition of 936 µl substrate solution, 32 µl POD solution (60 purpurogallin U/ml), which was filled up to 1 ml with 10 mM sodium acetate buffer pH 5.3 was measured. The assay was initiated by the addition of the GOx solution. According to Bateman and Evans, the unit definition is: one unit is oxidizing 1.0 μmole of β-Dglucose to D-gluconolactone and H 2 O 2 per minute at pH 5.3 and 30°C, which is equivalent to an O 2 uptake of 22.4 μmol per min (Bateman & Evans, 1995).

| Determination of mass transfer performance (k L a)
The k L a measurements were performed in 1 L (liquid) medium at 25°C in a 2-L STR. To ensure turbulent flow regime Re ≥ 10 4 , a pitched blade turbine with a power input of 0.12 W/L was installed. The following aerators were tested: 1 µm pore size SPG, 2 µm pore size transfer performance was determined according to the dynamic method described by (Garcia-Ochoa & Gomez, 2009;Tribe, Briens, & Margaritis, 1995). This method uses nitrogen for degassing to deplete the medium of oxygen and air to reach the saturation concentration c*. During aeration, the change in oxygen concentration c n is measured. Applying the following linearization in Equation (1) using the time difference t n − t 0 between the oxygen measurements, the dynamic k L a was determined.

| Experiments in an STR
The reaction setup was composed of a 500 ml jacketed glass reactor including three baffles and a pitch blade stirrer. To reduce the use of expensive enzymes in the gluconic acid synthesis, a 500-ml geometric equal reactor to the 2 l reactor filled with 300 ml medium was used. In this reactor, a down pumping pitch blade stirrer (45°blade, d = 46 mm) was applied. The gassing rate was adjusted by a flow meter. The tested aerators were the open pipe, SPG, and sintered frit as shown above. The substrate solution was containing 0.22 M D-glucose in 10 mM sodium acetate buffer. A temperature of 35°C and 450 rpm (0.12 W/L) was adjusted. The reaction was initiated by the addition of 11,865 U of GOx and approximately 12,000 U of catalase to the reaction mixture. To ensure a constant pH and monitoring of the productivity of gluconic acid, an auto titration unit was included in the setup. The reaction progress was measured by pH-stat titration of gluconic acid towards sodium gluconate. For the biotransformation, a constant pH of 5.3 was ensured with the continuous titration of 1 M sodium hydroxide solution.

| Bubble size and surface tension measurement
An endoscopic optical probe with a detection range of 9-1200 µm was used for the determination of the BSD. Measurements were carried out in an aerated STR, containing 83.3 mg/L GOx in 10 mM sodium acetate buffer pH 5.3, at a constant power input of 0.12 W/L (pitch blade stirrer) and a temperature of 25°C. In addition, the surface tension of different ratios of glucose and gluconic acid was measured with a DVT50 drop volume tensiometer from Krüss at 35°C.

| RESULTS AND DISCUSSION
For the comparison of aeration systems in biocatalysis regarding their gas utility, productivity, and oxygen atom economy, a suitable model enzyme is required. Therefore, the β-D-GOx Type VII from A.
niger with an oxygen consumption rate of 450 µmol/min/mg Enzyme (specific activity of v max = 450 U/mg Enzyme ) is chosen. According to Equation (1), a k L a value below 417 1/h per mg of GOx (c* = 0.215 mmol/L, pure water, 35°C, 0.3 L) has to be adjusted to overcome the mass transfer limitation of the biotransformation. Due to this, an aerator dependent investigation of macro-and microbubble aeration (fine bubble), by studying the potential to enhance or overcome mass transfer limitation on the example of a model reaction system in biocatalysis, is carried out. Simultaneously, optimal reaction conditions for the biotransformation have to be adjusted to avoid the interfering effects from stability, pH, type of buffer, and so forth on the aeration related effects. In the literature, a pH optimum of 5-6 in sodium acetate buffer for GOx from A. niger (Bankar, Bule, Singhal, & Ananthanarayan, 2009) is reported, which is in good agreement to the determined optimal pH range in this study. The temperaturedependent activity optimum is measured to be in the range of 30-60°C, which confirms the literature data (Bankar et al., 2009;Wilson & Turner, 1992 (Mafra, Furlan, Badino, & Tardioli, 2015).

| Determination of aerator mass transfer performance in the BSA model protein solution
Besides improving the oxygen saturation concentration through increased system pressure (Bolivar et al., 2019;Woodley, 2019) or through additives such as silicone oil (Leung, Poncelet, & Neufeld, 1997;Zhai et al., 2020), the system-specific aerator characterization in view of the oxygen mass transfer rate is the key to increase reaction rates. In respect to the oxygen solubility of 0.213 mM (35°C) and the above named K m,O2 of 0.18 mM of GOx from A. niger (Nakamura et al., 1976), the gluconic acid synthesis is challenged by the low solubility of oxygen in water (Woodley, 2019). To circumvent this limitation, the optimization of the oxygen transfer rate (OTR) is necessary. In Equation 2, the cases for an oxygen (O 2 ) consuming reaction given mass transfer limitation is expressed. Therefore, the relation of OTR (µmol/min) to maximum reaction rate (v max in µmol/ min/mg) is used: To overcome the oxygen mass transfer limitation for the above described GOx reaction, the ratio of oxygen supply (OTR) to oxygen demand (v max ) needs to be equal or higher than 1. In consequence, the enhancement of OTR in gluconic acid synthesis is essential to ensure high reaction rates. In respect to this and Equation (1), the OTR is directly linked to the volumetric mass transfer coefficient (k L a), which is a measure for mass transfer performance for the comparison of different aerator systems.  (Mersmann, 1977;Räbiger et al., 2013). Especially the combination of high Young-Laplace pressure, immobile bubble surface (We/Fr < 0.3) and high volume-specific surface area of microbubbles is compared with classic macrobubble aeration. In contrast to microbubbles, macrobubbles show low self-compression (Young-Laplace pressure), mobile bubble surface (instabile primary bubbles), and low volume to surface ratio to investigate the effect of bubble size in enzymatic catalyzed model reactions (Mersmann, 1977;Räbiger et al., 2013 rates and k L a values. This is proving that aeration in the jet gassing range is present for both macrobubble aerators, when using them to achieve a k L a of 160 and 60 1/h, respectively. In addition, the critical criteria for We 2 /Fr >> 675, which is the critical ratio of secondary bubble formation, is fulfilled. Therefore, the correlation from Mersmann (1977) for both macrobubble aerators is applicable. According to this, large Weber numbers indicate that the impulse of gassing is high. In this case, instable primary bubbles formed at the aerator. These instable primary bubbles disintegrate shortly after generation and result in a secondary bubble formation. Out of the force balance, maximum stable bubble size is calculated, which is under the described applied gas stream, and Weber number a function of the liquid properties (Mersmann, 1977 (Durst & Beer, 1969;Mersmann, 1977;Räbiger et al., 2013).  Huo, & Stenstrom, 2006;Wagner & Johannes Pöpel, 1996). In consequence, the formation of the resulting bubble size is expected to be highly influenced due to the process of secondary bubble formation and the disintegration of unstable bubbles (Durst & Beer, 1969). In contrast to macrobubbles, stirrer introduced shear forces affect the gas dispersion of microbubbles less, due to their small size (Matthes et al., 2020). In the experi- (4) In respect to this, the macrobubble aeration results in 25 times higher gas consumption compared to the microbubble aeration at a k L a of 160 1/h. Comparing the gas utility at k L a of 60 1/h, even lower aeration rates are necessary and the potential for the gas utility of microbubble aeration is even higher (44 times). As mentioned in For example, a bubble with a diameter d of 100 µm results in a volume-specific area of 6 × 10 4 m −1 , which is 10 folds higher than a macroscopic bubble of d = 1000 µm with 6 × 10 3 m −1 . Most of the bubble generators can produce a defined bubble size distribution that can be adjusted by gassing rate and pressure differences (Terasaka et al., 2011), which applies to the examined microbubble generators. With the assumption of ideal spherical bubbles (φ = 1) and the Sauter mean diameter d 32 (µm), a specific surface area SV (µm −1 ) is calculated, described by the Equation (5) (Stieß, 2005).  (Table 1). The second mass transfer influencing parameter is the residence time of bubbles in the medium. As a result of the decrease in bubble diameter, the rising velocity is decreased, which results in prolonged residence time. Therefore, the bubble interact longer with the surrounding medium. The Stokes law is the classic approach to calculate the theoretical terminal rising velocity of a gas bubble (Parkinson et al., 2008). The assumption of an immobile surface at low Re-numbers Re <<1 is given by the equation of Stokes (Stokes, 1851). The other interpretation (Parkinson et al., 2008) of Naiver-Stokes was done in 1911 by Hadamard and Rybczynsk, which takes the boundary condition, the internal viscosity η (N·s/m 2 ), and the dispersive phase into account.
In respect to the Stokes equation, a microbubble in water, for example, a 10-µm bubble results in a low rising velocity of 0.05 mm/s. Taking the height h (m) of the reactor into account, the resulting residence time τ (s) can be calculated with known gravity constant g (m 3 /kg/s 2 ) as well as the density of gas and liquid ρ (g/m 3 ) for ideal conditions without stirring.
The third influencing parameter is the Young-Laplace pressure within the bubble, which can be calculated with Equation 7. 2 (mol/L) in the bubble thin layer (Atkins et al., 2013). Consequently, the shrinking behavior as well as the bubble formation is depending on the surface tension. In a previous study, we reported a decrease of 15.5% in surface tensions of a buffer solution containing 67.8 mg/L BSA or GOx, compared with pure water (Matthes et al., 2020). This change applies to the underlying system. In contrast to this, the variation of the gluconic acid and glucose ratio is resulting in no significant change in surface tension compared with pure water with a measured value of 72.4 mN/m (25°C). Furthermore, the bubble diameter is directly affecting the pressure difference between the inner bubble and the system pressure (Tanaka, Kastens, Fujioka, Schlüter, & Terasaka, 2020). According to Equation 7, the pressure difference is calculated to be 17 times higher, when comparing microbubbles with Sauter mean diameter of 320 µm to macrobubbles with 5 mm (Table 1). The increase in Young-Laplace inner bubble pressure is one of the major reasons for the quick shrinking behavior of fine bubbles. This concentration gradient is the driving force for the diffusion of gas from the bubble into the bulk medium.
As reported by Tsuge (2014), microbubbles show quick dissolution amplified by self-compression during the bubble shrinking (Iwakiri et al., 2017). The effect of bubble shrinking is getting more prominent with decreasing bubble size towards a critical diameter of less than 30 µm (Iwakiri et al., 2017;Tanaka et al., 2020).

| Comparative study of aeration systems on the example of a GOx catalyzed gluconic acid synthesis
To assess the feasibility and gas-saving potential of microbubble aeration in biocatalysis, the productivity of a GOx catalyzed gluconic acid synthesis is investigated. In consequence of the demonstrated high difference in mass transfer performance ( Figure 3  Note: Conditions for calculation of Δp and: T = 25°C, η = 0,81 mPas, g = 9.81 m/s 2 ρ gas = 1.18 kg/m 3 , ρ L = 1000 kg/m 3 ; Kell, 1975;Mersmann, 1977;Straub, Rosner, & Grigull, 1980. a   Overall, a significant improvement in productivity by application of microbubble in comparison to macrobubble aeration is achieved. Furthermore, microbubble aeration improves the gas utilization, which is compared by the atom balance of oxygen applied to the reaction medium related to the bound oxygen during the biotransformation. Therefore, in Equation 9, the oxygen atoms bound in the product gluconic acid (n /2 gluconic acid ) per reaction time are normalized to the molar stream of oxygen, which was introduced by the airflow rate into the system (• t n O ,total 2 ). This is offering the possibility to reduce the production costs by saving gas and reducing off-gas streams in industrial biocatalysis and chemical applications.