Streamer formation dynamics with mixed bacterial species: Effects of cultivation conditions, hydrodynamics, and species

Bacterial biofilms are assemblies of micro-organisms in which cells are encased in a self-excreted matrix of extracellular polymeric substances (EPSs) sticking to each other on living or non-living surfaces. The formation of bacterial biofilm begins with the adhesion of individual bacteria to a surface, and the biofilm colonizes the surface and grows through cell division and enclosure by polysaccharide matrices.¹ Recently, the study of biofilms has attracted significant interdisciplinary attention, mainly motivated by the fact that this form of bacterial collection can play both positive and negative roles in natural and engineered systems. For instance, biofilms can help in bioremediation (in the removal of organics) in waste-water treatment, in CO₂ sequestration,^2,3 and in enhancing biogas and methane yield in an anaerobic digester.⁴ On the other hand, biofilms can play negative roles such as in shipping industries causing biofouling,⁵ in membrane water purification systems for drinking water production forming active biofilm on the porous surface,⁶ and in clinical environments and medical devices (such as in heart stents and catheters) posing clogging and persistent infection risks aggravated by their significant resistance to antibiotics and antimicrobial agents when in biofilm form.^7–10 

The mechanisms by which bacteria form biofilm on surfaces and the conditions promoting this process are not yet fully understood. However, some studies have demonstrated that the biofilm formation process on a porous surface is affected by a number of factors such as the bacterial cell type, the surface properties of the bacteria, their physiology, the pH and ionic strength of the bacterial suspension, and the flow conditions.^11–21 The complex interplay between these biological and physico-chemical factors controls the biofilm formation process. Depending on these conditions, bacteria form biofilms with different architectures and morphologies on filters. Stoodley and co-workers²² have reported that, when under flow through filters, bacteria form filamentous structures called streamers, porous-matrix structures formed mainly by externally imposed shear flow moving through a background flow. They are dominantly positioned in the bulk flow, not at the walls or stagnation points.^1,23 Recently, experimental studies using microfluidic devices have demonstrated that bacterial streamers can form in laminar flow conditions in zig-zag microchannels with different curvature angles²⁴ and in arrays of microchannels with different geometries and channel sizes.²³ Rusconi and co-workers²⁵ showed that hydrodynamics, especially the presence of secondary flow in the system, plays an important role in the development of streamers. On the other hand, Hassanpourfard et al.²⁶ investigated streamer distribution in a microfluidic device with cylindrical micro-pillars along the z axis using confocal laser scanning microscopy and reported that no localization of streamers was found; they found that streamers were distributed at different heights. Battin et al.²⁷ confirmed the presence of secondary flow around the grains of the fluidic devices using numerical simulations (COMSOL Multiphysics). Such contradictory findings suggest that the mechanism of streamer formation in relation to channel geometry or media topography remains poorly understood. Furthermore, the surface chemistry of the media significantly affects streamer inception and evolution;^28–30 Ghosh et al.³¹ reported that smooth surfaces are more attractive to biofilms and to the resulting streamer formation by shearing due to flow-induced deformation affecting the initial stages of biofouling and resulting in streamer detachment on rough surfaces.

Bacteria of all taxa are known to synthesize diverse intracellular and extracellular biopolymers. Few intracellular biopolymers are studied with limited use, while there has been a vast range of studies on extracellular biopolymers with valuable structural and functional properties having different applications and promising prospects. Extracellular biopolymers are collectively referred to as extracellular polymeric substances, or exopolysaccharides (EPS). EPS play crucial roles in the development and stability of biofilms at solid–liquid interfaces such as adhesion, bacterial cell aggregation, water retention, biofilm cohesion, nutrient source and energy sink, protective barrier to various antibiotics, binding of enzymes, and sorption of organic compounds and inorganic ions.^32–34 The extent of EPS synthesis varies with bacterial species and the physico-chemical incubation conditions. The physico-chemical factors playing a central role in the yield of EPS include pH, temperature, incubation time, and medium composition (carbon, nitrogen, and cation sources).³² Carbon availability, with the limitation of other nutrients such as nitrogen, oxygen, or phosphorus, controls EPS synthesis for most bacterial species.^35,36 A high carbon-to-nitrogen ratio is usually required to achieve a high yield of EPS synthesis.^37,38

On the other hand, the initiation phenomena for streamer formation and quantitative evaluation of the formation and evolution dynamics remain puzzles until now. Some researchers argued that streamer formation occurs only in turbulent flows,^22,39–42 while others confirmed later, with the advent of microfluidic devices, that it also occurs in creeping flow regimes.^{1,5,26,33,34,43–47} These observations confirm that bacterial streamers can form under a variety of flow conditions, ranging from creeping to turbulent flow regimes.³¹ Furthermore, the investigations on streamers until now have been limited to pure culture bacterial species, overlooking the aggregative nature of the economically relevant biofouling systems, which may most often involve mixed bacterial species,^27,47–51 which brings significant biological complexity to the analysis. The studies so far mostly did not account for inter-species interactions, disregarding cell-to-cell signaling,⁵² which in turn may influence the material composition of the EPS. The nature or material composition of the EPS significantly determines the structure and hydrodynamics of the streamers.³¹ Hence, investigations on streamer formation dynamics considering mixed bacterial species are still lacking and are challenging, demanding interdisciplinary developments of theoretical and experimental techniques for their qualitative analyses and quantitative evaluations at different spatio-temporal scales of scrutiny.

In this study, we present the effects of bacteria cultivation conditions, hydrodynamics, and bacterial species type (pure vs mixed species) on the bacterial streamer formation during filtration in microfluidic filters to get time- and space-resolved dynamic insights. The microfluidic systems mimic important engineering systems such as biomedical devices^53–55 and filtration units.^1,56 The cultivation conditions influence EPS synthesis, which in turn has a dominant role in streamer development. Enterobacter A47 species is used for the cultivation conditions analysis. The bacterium Enterobacter A47 has been studied for its synthesis of fucose-rich EPS (FucoPol) using glycerol as a carbon source.^57–60 The main control parameter in this study is the carbon-to-nitrogen ratio in the nutrient. Glycerol and ammonium phosphate are used as carbon and nitrogen sources, respectively. We also analyzed the extent of streamer formation by using mixed bacterial culture (a binary mixture) and compared it with the formation by pure culture. In the binary mixture, Enterobacter A47 is mixed with Cupriavidus necator, a non-EPS producing species with a larger size compared to Enterobacter A47. Cupriavidus necator produces intracellular polyhydroxyalkanoates (PHA) when excess sugar is available. The experiments are performed at relatively high flow velocities (0.14 to 0.83 m/s) using different geometries of microchannels in order to investigate the effect of the hydrodynamics on the streamer formation.

The microfluidic devices used for this work were designed using CleWin 4 software and fabricated in the MESA Plus (MESA+) nanolab at the University of Twente, The Netherlands. The dry reactive-ion etching process is used to fabricate the chips, and the devices are fabricated from silicon and glass wafers. Dry Reactive-Ion Etching (DRIE) of Silicon is used to etch the channels where the analysis takes place, the upstream and downstream of the channels, and the inlets and outlets. Anodic bonding is used for the bonding of the silicon wafer on which the microchannels are engraved with the Borofloat glass wafer after manually aligning and prebonding. The bonding takes place at 400 °C in a vacuum at a high voltage (1000 V). This temperature is low enough to prevent plastic deformation of the glass wafer and, thereby, the possible closing of the channels. After the anodic bonding process, dicing is performed using a Disco-DAD dicing saw to get individual chips from the wafers. For use in filtration experiments, the chips are assembled in a chip holder that has inlets and outlets to which nano-ports and tubes are connected (Fig. 1).

The microchannels used for this experimental work have widths ranging from 2 to 5 μm with different geometries: straight microchannels (DS), microchannels with staggered square pillars (DA), and microchannels with staggered polygon pillars (DP). The microfluidic device assembly, the different designs of the microchannels, and the design parameters for a 5 μm straight channel chip are shown in Fig. 1. All the main channels and the microchannels have a depth of 20 μm (i.e., a single layer system).

Enterobacter A47 (DSM 23139), a Gram-negative motile bacterium containing flagella, is grown on glycerol by-products from the biodiesel industry to study the effects of cultivation conditions (which influence the production of EPS) on streamer formation during filtration with microfluidic devices. The bacterium Enterobacter A47 (DSM 23139) is rod-shaped with dimensions of 0.7–0.8 μm × 1.2–2.5 μm.⁶¹ 

Enterobacter A47 produces a fucose-containing EPS (FucoPol) by cultivation in mineral medium supplemented with glycerol.⁶⁰ EPS synthesis by Enterobacter A47 has been shown to be influenced by both the initial glycerol and nitrogen concentrations and by the nutrients’ feeding rate.⁶⁰ Therefore, two types of Enterobacter A47 cultures are cultivated by using different substrate ratios to control the EPS production.

The first culture (EPS-producing culture) is grown on a medium supplemented with glycerol by-product (40 gl⁻¹), while nitrogen concentration is kept low (<0.1 gl⁻¹). The second culture (a non-EPS producing culture) is cultivated with the same medium and cultivation conditions but with an increased nitrogen concentration (1.27 g/l) to suppress the EPS production.

Mixed bacterial culture (Enterobacter A47 and Cupriavidus necator in a 1:1 ratio) is also cultivated aerobically in the same medium as the first case to study streamer formation. The cultivation is performed at a controlled temperature of 30 °C and a pH of 7. Both bacterial species have a cell concentration of 6 g per dry weight after the growth phase. Enterobacter A47 produces EPS, while Cupriavidus necator produces intracellular polyhydroxyalkanoates (PHA) when there is excess sugar. Cupriavidus necator is a non-EPS producing bacterium and is larger in size compared to Enterobacter A47.

Filtration experiments are performed in dead-end mode at constant flow velocity by using a syringe pump (Harvard Apparatus PHD 2000). The filtration process is recorded using a light sensitive camera (Olympus). The dynamics of streamer formation during filtration is recorded by taking images every 30 s. The experimental setup for the filtration to study streamer formation phenomena is shown in Fig. 2. During filtration, flow conditions are kept in a laminar regime, and the maximum Reynolds number is 6.64, calculated at the maximum flow velocity (0.83 m/s) in the feed channel. The flow velocity is varied from 0.14 to 0.83 m/s (corresponding to flow rates from 0.05 to 0.30 ml/min).

Physical effects (i.e., flow conditions and geometry of the micro-separator) and biological factors (i.e., bacterial type and cultivation conditions) influence streamer formation dynamics during the filtration of bacterial suspensions through micron-sized separators. In this experimental work, we used microfluidic devices to investigate how flow, channel geometry, and bacterial cultivation conditions, in particular the substrate composition, affect streamer development in both pure and mixed bacterial cultures. The experiments are performed at a relatively high flow regime (Re ≤ 6.64). Constant flow velocity filtration experiments are performed using three different geometries of the separation section with both pure and mixed bacterial cultures. These geometries are used to study the effect of the connectivity of the microchannels and the sharp edges of the pillars on the extent of streamer formation (Sec. III A 1). These geometry effects are mainly related to the secondary flow that occurs at the edges of the pillars. The microchannel size is also varied (2 and 5 μm) to investigate the effect of the constriction on the streamer formation dynamics. The effect of the cultivation conditions is presented in Sec. III A 2. The flow velocity is varied from relatively low to high values (i.e., 0.14, 0.28, and 0.55 m/s) in experiments with pure culture suspensions (Sec. III A 3) and abruptly increased to 0.83 m/s after forming a dense streamer at 0.55 m/s to analyze the effect of a sudden change in hydrodynamic conditions. Filtration experiments with mixed bacterial suspensions are performed at intermediate flow velocity (i.e., 0.28 m/s), at which high streamer formation is observed with pure bacterial suspensions (Sec. III B). Two microchannel geometries are used for filtration experiments with mixed culture: microchannels with staggered square pillars and polygon pillars. The following sections present the effects of all these parameters on the streamer formation dynamics.

Experiments are performed with microchannels having different tortuosity and channel sizes at a constant flow velocity of 0.55 m/s. A very low accumulation of bacteria and streamer formation is observed for non-EPS producing Enterobacter A47 pure cultures cultivated at low C:N ratios. Figure 3(a) shows a 2D micrograph of bacterial accumulation for two channel sizes (2 and 5 μm) and two channel configurations (straight channels and staggered square pillars). The accumulation of materials on the upstream side of the filtering zone is observed frequently, and the possible reasons for this could be the glycerol remaining in the suspension, protein agglomeration, the presence of inorganic residues, contamination from the chips and connections, or a combination of these factors. Relatively dense streamers were formed when filtering the same bacterial species with tortuous channels (channels with alternatively aligned pillars). Figure 3(b) demonstrates the temporal increment of the density of streamers with time in staggered square pillar channels.

Figure 4 shows that the average streamer length increased more rapidly with the tortuous microchannels. The average streamer length is calculated by determining the area covered by the streamers and dividing the value by the width of the channel. The area covered by the streamers is calculated by ImageJ software. The tortuous nature of the microchannels has a more significant effect on streamer formation than the decrease in channel size, which is mainly due to the presence of secondary flow.²⁵ The inset image shows the attachment and gliding phenomena of bacteria during the inception of the streamers. The average streamer length for the tortuous channels is about 34 μm, while for the 2 μm channels, it is 6 μm after 100 min of filtration time. Recent studies have demonstrated that the presence of complex surface geometries enhances the formation of flexible streamers, which causes clogging in flow systems.^53,62,63 This could enhance streamer growth through the filtration of bacteria and other components in the system by interception from the bulk fluid.²⁷ 

In these systems, the hydrodynamics control the formation and the three-dimensional structure of the bacterial streamers, while bacterial motility is insignificant since the experiments are performed at relatively high flow rates. Therefore, the flow field around the pillars of the different channel geometries plays a major role in shaping the streamers in different structures. The tortuous nature of the flow (Fig. 3) actively enables the development of dense streamers close to the outlets of the channels. In contrast, long and highly porous filaments are formed by the flow in straight channels. This shows that flow determines the three-dimensional structure of bacterial streamers. The temporal oscillation in average streamer length for tortuous channels in Fig. 4 is due to the detachment of streamers from the channel outlets by the flow caused by the difference in pressure gradient in the background flow and in the streamer jets.⁶⁴ The streamer breakup is thus purely a flow instability phenomenon in the viscous filamentous jets.

The presence of EPS in pure cultures has been confirmed and observed using various electron microscopy techniques. The EPS are characterized by their adsorption ability, biodegradability, and hydrophilicity/hydrophobicity. The main components in EPS, though it is a dynamically evolving entity, are carbohydrates, proteins, humic substances, and nucleic acids.^65,66 The combination of the above special characteristics and the contents of the main components of EPS crucially affect the physicochemical properties of microbial aggregates, including the structure, surface characteristics, mass transfer, adsorption ability, and stability. Substrate type and nutrient levels have substantial effects on the microbial communities and their metabolism, influencing the production and composition of EPS.^67–71 Furthermore, EPS composition varies with species origin, growth temperature, and hydrodynamic shear.^{65,66,72–74} In this study, we analyze the effect of the presence of EPS on streamer formation. The C:N ratio in the substrate (which is kept at 50:1 for the EPS-producing culture and at 4:1 for the non-EPS producing culture) is the main controlling parameter for the EPS synthesis by the bacteria. The synthesis of EPS by Enterobacter A47 bacterium in nitrogen limited cultivation conditions may be related to the absence of nitrogen, restricting other metabolic pathways and leaving place for EPS synthesis.⁶⁰  Fig. 5 shows very rapid streamer formation in the downstream zone of the separation section during filtration with the EPS-producing culture with straight microchannels, while there is no streamer formation by the non-EPS producing culture. The time scale for the streamer formation is in the order of a few minutes in the former case; the average streamer length reaches about 50 μm in just 5 min of filtration time. The streamers are very long but highly porous in structure. They are also very dynamic undulating in the background flow.

The inset images in Fig. 5 show the extent of the difference in streamer formation between the EPS-producing culture and the non-EPS producing culture after an hour of filtration time. Therefore, it can certainly be noted that EPS secretion by bacteria is the main cause, if not the only one, of streamer formation during filtration through constrictions.

Filtration experiments using EPS-producing pure culture, Enterobacter A47, are performed at three different flow rates of 0.05, 0.10, and 0.20 ml/min (corresponding velocities of 0.14, 0.28, and 0.55 m/s) to analyze the hydrodynamics effect on the bacterial streamer formation. The experiments are performed with 5 μm wide straight channels in dead-end mode. Figure 6 shows the average streamer length with time, and it can be noted that streamer formation is favored at intermediate flow velocity (0.28 m/s), with the average streamer length reaching up to 125 μm. The average streamer length at the highest flow velocity (i.e., 0.55 m/s) becomes almost constant (about 65 μm) after 20 min of filtration time, except for small oscillations. These oscillations are due to the balance between the formation and the subsequent detachment of streamers by the flow. The average streamer length increases throughout the filtration time at the lowest flow velocity (0.14 m/s), where no detachment of streamers by the flow is observed.

In order to analyze further the effect of the flow on the streamer stability and long-term fate, the flow velocity was suddenly increased from 0.55 to 0.83 m/s (corresponding flow rate of 0.30 ml/min in Fig. 7), after the formation of dense streamers, and the average streamer length was evaluated with time (Fig. 7). It can be noted from Fig. 7 that the average streamer length drops substantially and then starts to increase, revealing that the background fluid flow plays a significant role in the stability of bacterial streamers.^44,63,75 Valiei et al.⁶³ hypothesized instability of the streamer jets for failure or detachment, while Biswas et al. proposed instability due to necking as a mechanism for localized failure⁷⁵ or due to void growth at the attachment surfaces.⁷⁶ Streamer breakage or failure could also occur due to clogging of the channels after the formation of dense streamers, thereby resulting in localized failure by the stick-slip mechanism.⁴⁴ The instabilities may arise from localized pressure cavities, tearing, and bacteria losses. The interaction between the background flow and the structure of the streamers may change with time as the streamers get denser, which may create localized flow channels for the fluid, causing instability and failure. The biological life phenomena of the system in the streamers further complicate the analysis. Hence, a detailed investigation of the nucleation and growth of such instabilities and final failure mechanisms at the nanoscale is still a scientific challenge. The inset images in Fig. 7 demonstrate that the streamers are dense with less detachment at 0.55 m/s, while there is significant streamer detachment at 0.83 m/s and the structure is more porous.

Almost all of the studies reported so far on the production of EPS and streamer formation are mainly conducted by employing pure bacterial cultures.^1,26,63,77 However, aggregative forms of bacteria, such as flocs or biofilms, are often supposed to comprise mixed species. The important interactions between microbes, i.e., inter-species and intra-species interactions, have been overlooked. Few researchers have reported that using mixed cultures over single cultures promotes EPS synthesis.^78–81 The cooperative growth of bacterial cells that occurs in mixed cultures of EPS-producing and non-EPS producing bacterial strains can influence the concentration and characteristics of the EPS.⁸² The presence of certain non-EPS producing bacterial species along with EPS producing species in a mixed culture can stimulate the production of EPS. Figure 8 shows the comparison between the streamer formation dynamics during filtration of EPS producing pure culture (Enterobacter A47) and mixed bacterial cultures of Enterobacter A47 and Cupriavidus necator at intermediate velocity (0.28 m/s) using microchannels with staggered square and polygon pillars.

Very long streamers (about 330 μm) are formed by the mixed culture at a filtration time of 80 min with the staggered square pillar geometry. The corresponding average streamer length is about 240 μm for the pure culture, which shows about a 37% increment in average streamer length with the mixed culture. The initial stages of streamer formation show similar trends for both cultures. One possible reason for this could be that the initial stages of the bacterial adhesion process may be dominated by the Enterobacter A47 species during filtration with the mixed culture. On the other hand, the initial stages of streamer formation may be dominated by the geometry of the micro-separator rather than the bacterial species during the filtration of both pure and mixed cultures. After a certain filtration time (about 20 min), the streamer formation is more pronounced for the mixed culture. At later stages, higher concentrations of biological particles (bacteria) in mixed culture may enhance adhesion; cell-to-cell signaling or “co-evolutionary” relationships may also impact streamer formation due to an increase in carrying capacity through niche construction.²⁷ Scheidweiler et al.²⁶ cultured Pseudomonas flourescens bacterium in LB and M9 media and reported that significantly smaller flocs were formed in the latter case and streamer formation was not observed; this shows that growth conditions principally control streamer formation.

It has been reported that when two or more microbial strains form a biofilm, the presence of EPS may not only assist in establishing the biofilm but also promote greater biofilm growth than the comparable single species biofilm.⁸³ The EPS promotes not only the adhesion and growth of the cells that synthesize it but also those of other microbial species. Symbiotic relationships between the microbial strains may also be beneficial for EPS production in the mixed culture. These relationships between EPS and non-EPS producing bacterial strains could be useful for producing EPS with higher flocculation efficiency, which increases the adhesion probability of the bacterial cells or aggregates. Mixed culture of EPS-producing and non-EPS producing bacterial strains may also reveal higher viscosity and higher molecular weight EPS production than that of pure culture,⁸³ which in turn enhances the streamer formation process.

In many systems, for instance, membrane bioreactors, the presence of EPS is a serious problem due to fouling, which occurs by pore clogging, floc adhesion, and cake layer and/or streamer formation.⁸⁴ The important factors determining the extent and severity of the fouling in membrane systems handling materials containing EPS are the concentration and the chemical characteristics of the EPS.^85,86 Different strategies have been proposed to mitigate this membrane fouling problem such as control of EPS production, modification of EPS characteristics, and removal of EPS from the system.⁸⁶ The implementation of these strategies requires a detailed analysis of the different components and the chemical and functional properties of the EPS to understand its role in fouling. EPS production control is challenging because EPS secretion is a general attribute of micro-organisms in both natural systems (e.g., wastewater sludge) and in laboratories with controlled conditions occurring in both eukaryotes and prokaryotes.

Because of its special characteristics, the presence of EPS in microbial suspensions highly influences the properties of the suspension during flow. Provided that EPS is available with the desired properties, it can play fundamental roles in the matrix, including adherence to surfaces, aggregation of bacterial cells, formation of flocs and biofilms, cell-cell recognition, resistance against external forces such as shear by acting as structural elements, protective barrier for cells, water retention to avoid desiccation of cells, sorption of exogenous organic compounds and inorganic ions, etc. The biopolymers in EPS play an important role in the flocculation of particles by bridging mechanisms that encourage aggregation. They can also reduce the particle surface charge by using oppositely charged cations and/or bioflocculants, reducing the distance between the particles and making attractive forces more effective than repulsive forces between the particles. Therefore, EPS takes part in the aggregation of particles through the charge neutralization mechanism. The higher the molecular weight of the biopolymers (which is the case for Enterobacter A47), the more effective the bridging mechanism is. Flocculation with high molecular weight EPS involves more adsorption sites.

The production of EPS by the Enterobacter A47 bacterium and its composition are largely influenced by a number of factors that govern bacterial metabolism, including the cell growth phase, the carbon and nitrogen sources and their ratios, the pH of the system, the temperature of the system, and aeration (aerobic or anaerobic conditions).⁶⁰ Increasing the initial nitrogen concentration led to higher Enterobacter A47 growth, but it was detrimental to EPS synthesis. Higher nitrogen concentrations not only significantly reduced EPS production but also resulted in the synthesis of polymers with a very different composition, namely, lower fucose content and higher glucose content;⁶⁰ this may be linked with the production of an enzyme that degrades biopolymers.

A profound emphasis is given to the carbon-to-nitrogen ratio in relation to EPS production because it has a significant effect on microbial metabolism and, hence, on biopolymer production. The carbon-to-nitrogen ratio has been varied over a wide range in order to analyze its effect on EPS synthesis and streamer formation during flow through microchannels. Increasing the carbon-to-nitrogen ratio from 4:1 to 50:1 largely increased EPS production. Filtration of bacterial suspensions of the two types (non-EPS producing and EPS producing) through microchannels shows that the presence of EPS in the system significantly controls streamer formation. Ye et al. (2011b)⁸⁷ reported that decreasing the carbon-to-nitrogen ratio from 100 to 20 (favorable for EPS production) resulted in decreased carbohydrate content and increased protein content in the EPS. Durmaz and Sanin (2001)⁶⁷ also reported that a carbon-to-nitrogen ratio of 5 in activated sludge resulted in EPS rich in proteins but low in carbohydrates; increasing the ratio to 40 led to a sharp decrease in the amount of protein.

Our analysis of the effect of EPS presence in bacterial suspension filtration shows that EPS plays a crucial role in streamer formation, as demonstrated in Fig. 5. All the characteristics of the EPS described earlier may have important roles in this dynamics. The EPS can have a significant role in the formation of bacterial aggregates, or flocs, which increases the adhesion probability and hence the streamer formation. The biopolymers can also play a bridging role between different bacterial aggregates, and the entanglements among themselves and with other molecules lead to streamer formation. The formation of tinny filamentous structures connecting to form nets growing into streamer jets initiates the process, and our experiments confirm this hypothesis, leading to the conclusion that the EPS produced by bacteria a the main cause of streamer formation. The production of EPS is in turn controlled by the carbon-to-nitrogen ratio in the substrate. Therefore, the cultivation conditions of bacteria control the streamer formation phenomenon during flow through constrictions.

Our study also shows that the presence of non-EPS producing bacterial species (Cupriavidus necator) along with EPS-producing species (Enterobacter A47) in a mixed culture enhances streamer formation. This could be related to the cooperative growth that occurs in the mixed culture of the bacterial species, which influences the concentration and characteristics of the EPS.⁸² The different characteristics of EPS may increase the adhesive properties of surfaces. Symbiotic relationships between the two bacterial strains might also enhance EPS production by the EPS-producing species. Furthermore, EPS synthesis may increase due to an increase in substrate utilization efficiency, which results from the secretion of certain enzymes by the microbes helping them fulfill their nutrient requirements. The flow properties of the mixed cultures (such as the viscosity) could be largely influenced by these phenomena, and this could enhance the extent of adhesion and streamer formation.

Analysis of the rheological characteristics of the cultivation broth of Enterobacter A47 bacterium showed that there is a considerable viscosity buildup, and the cultivation broth developed a non-Newtonian behavior essentially due to EPS concentration, molecular weight, chemical composition, the formation of new interactions between individual EPS molecules, and the other components of the broth. These new interactions and the resulting viscosity buildup can have a prominent impact on the formation and morphology of the bacterial streamers.

Concomitant with EPS synthesis and its paramount effect on bacterial streamer formation, hydrodynamic conditions play important roles in streamer inception dynamics. In this regard, we have analyzed the parametric effects of the channel size, the architecture of the micro-separator, and the flow velocity. The dynamics of the streamer formation and the resulting morphology is highly influenced by the size and geometry of the filter. The temporal growth of the bacterial streamers is quantified by evaluating their average length. The dynamics of the streamers in the flow system is analyzed in terms of the area differences between two consecutive images captured during the filtration experiment. The morphological differences between streamers formed by different filter geometries are analyzed by evaluating the “particles” size distribution.

In general, small channel sizes and increased tortuosity of channels resulted in rapid streamer formation and growth (Fig. 4). This could be related to an increase in the capture efficiency of the bacteria as the channel size decreased and approached the size of the bacteria and the role played by the secondary flow in the entanglement between the polymer chains as the channels became tortuous with sharp edge corners. The streamers formed with straight channels by the EPS-producing bacterial suspension of Enterobacter A47 are long and highly porous. In contrast, the streamers formed during filtration using tortuous microchannels are relatively short and dense in structure. This shows that the filter geometry largely influences the morphology of the filamentous structures formed by the bacterial aggregates. This morphological difference between the streamers formed by the two geometries is analyzed by calculating the “particle” size distribution by binarizing the images captured after an hour of filtration. The “particle” size distribution evaluation gives an estimation of how segregated or densely populated the streamer jets are. Figures 9(a) and 9(b) present the results of the “particle” size distribution for streamers formed with straight microchannels (non-tortuous) and tortuous geometry, respectively.

The results show that the “particles” are largely distributed for the streamers formed with both geometries and form a non-Gaussian distribution. The “particle” sizes for streamers formed using tortuous microchannels range from 0.5 to 350 μm, while the corresponding values for streamers formed with straight microchannels range from 0.5 to 163 μm in diameter. A large number of small particles and a small number of relatively large “particles” are obtained from streamers formed with straight microchannels. Compared to this, the structure formed by the tortuous geometry results in a small number of small “particles” and a relatively large number of large “particles.” The mean diameter of “particles” for streamers formed with straight microchannels is 1.217 μm, while the corresponding value for streamers formed with tortuous microchannels is 3.701 μm. The sphericities of “particles” formed with straight and tortuous microchannels are 0.64 and 0.65, respectively. These data indicate that very long and highly porous streamers are formed during filtration with straight microchannels [Fig. 9(a)]. On the contrary, the tortuosity of microchannels results in relatively large “particle” sizes, implying that the streamers are dense in structure. This could be related to the role of the sharp corners of the pillars (or the secondary flow) in the aggregation of the bacterial cells wrapped by the large molecular weight biopolymers.

High flow velocity enhanced the rapid formation and growth of streamers during the initial periods of the filtration experiments (Fig. 6). However, there is no further growth of the streamers, and it seems to reach a limit with small fluctuations in the average length due to the detachment of some streamer jets by the background flow. Relatively low velocity filtration of the bacterial suspension resulted in a slow but steady increase in the average streamer length with no fluctuations. The longest streamers (about 130 μm using straight microchannels) are formed while filtering at intermediate flow velocity (0.28 m/s), with some fluctuations at long filtration times. Therefore, flow velocity plays an important role in the kinetics of streamer growth and the resulting average streamer length. A sudden increase in velocity (from 0.55 to 0.83 m/s) after the formation of dense streamers resulted in the depletion of the streamers due to detachment by the background fluid flow. This could be related to the instability created in the streamer jets by the flow due to the pressure gradient difference between the fluid flow and the streamer jets.

As the streamers thicken, they appear to become more “rigid.” It has been observed during the filtration experiments that they cease to undulate in the flow. Frequent detachment of streamer jets is also observed as the filtration experiment proceeds. Figure 10 shows the area differences between the consecutive images, indicating the streamer formation dynamics. There are two possible sources for the area difference: one is due to the formation of new streamers, and the other possibility is the mobility of the streamers by the background flow.

Higher fluctuations in area coverage by streamers in the initial stages of the filtration experiment can be noted in Fig. 10 for both straight and tortuous channels. This is due to the high mobility or undulation of the streamers by the fluid flow and the rapid growth of streamers during this period, both of which were clearly observed during the filtration experiments. This fluctuation is more pronounced and lasts longer (more than an hour) for straight channels, while for tortuous channels, the area difference shows a steady increase after about 20 min. This shows that denser and lower mobility streamers are formed by the tortuous microchannels. At the end of the filtration (after 70 min), a limit is reached, and the area differences are leveled-off for both cases. This is because the streamer formation dynamics reach equilibrium due to the balance between the formation and the detachment of streamers in the fluid flow.

In this work, we present the effects of bacterial cultivation conditions and hydrodynamics on streamer formation during the flow of both pure (EPS-producing) and mixed (EPS producing and non-EPS producing) bacterial suspensions through constrictions. The carbon-to-nitrogen ratio is used as a control parameter to analyze the effects of cultivation conditions on the bacterial species, which influence EPS synthesis and streamer formation. Complex microchannel geometries are used, and the flow velocities are varied in order to analyze the effect of the hydrodynamics on the streamer formation.

The results show that the substrate type and nutrient levels supplied to bacterial species influence the synthesis and composition of EPS, which in turn controls streamer formation. Nitrogen limited cultivation conditions favor EPS synthesis, which is the main factor in streamer formation. Rapid and pronounced streamers are developed during filtration with EPS producing culture. The tortuosity of microchannels plays a significant role in streamer formation and the morphology of the streamer due to the occurrence of secondary flow, which enhances the clogging in flow systems. Tortuous microchannels augment the rapid formation of dense streamers. Furthermore, moderate hydrodynamic conditions (i.e., intermediate flow velocities) favor streamer growth and stability, which are mainly related to the background fluid flow. Studies on streamer formation have so far overlooked the important interactions between microbes. The streamer formation dynamics comparison between pure and mixed bacterial cultures shows that streamer formation is more pronounced in the mixed culture. In summary, the results from our experimental study demonstrated that bacterial streamer formation is influenced by both biological (such as the type of culture and the cultivation conditions) and hydrodynamic (such as the flow conditions and the micro-separator geometry) conditions.

The authors acknowledge the financial support of the Education, Audio-visual and Culture Executive Agency (EACEA) division of the European Union under the program of the EUDIME (European Doctorate in Membrane Engineering, http://www.eudime.unical.it) grant in the Erasmus Mundus framework.

The authors have no conflicts to disclose.

Zenamarkos B. Sendekie: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Patrice Bacchin: Conceptualization (equal); Formal analysis (equal); Funding acquisition (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Writing – review & editing (equal). Rob G. H. Lammertink: Conceptualization (equal); Formal analysis (equal); Funding acquisition (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – review & editing (equal). João G. Crespo: Conceptualization (equal); Formal analysis (equal); Funding acquisition (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article.