Assessment of the Impact of Temperature on Biofilm Composition with a Laboratory Heat Exchanger Module.

Temperature change over the length of heat exchangers might be an important factor affecting biofouling. This research aimed at assessing the impact of temperature on biofilm accumulation and composition with respect to bacterial community and extracellular polymeric substances. Two identical laboratory-scale plate heat exchanger modules were developed and tested. Tap water supplemented with nutrients was fed to the two modules to enhance biofilm formation. One "reference" module was kept at 20.0 ± 1.4 °C and one "heated" module was operated with a counter-flow hot water stream resulting in a bulk water gradient from 20 to 27 °C. Biofilms were grown during 40 days, sampled, and characterized using 16S rRNA gene amplicon sequencing, EPS extraction, FTIR, protein and polysaccharide quantifications. The experiments were performed in consecutive triplicate. Monitoring of heat transfer resistance in the heated module displayed a replicable biofilm growth profile. The module was shown suitable to study the impact of temperature on biofouling formation. Biofilm analyses revealed: (i) comparable amounts of biofilms and EPS yield in the reference and heated modules, (ii) a significantly different protein to polysaccharide ratio in the EPS of the reference (5.4 ± 1.0%) and heated modules (7.8 ± 2.1%), caused by a relatively lower extracellular sugar production at elevated temperatures, and (iii) a strong shift in bacterial community composition with increasing temperature. The outcomes of the study, therefore, suggest that heat induces a change in biofilm bacterial community members and EPS composition, which should be taken into consideration when investigating heat exchanger biofouling and cleaning strategies. Research potential and optimization of the heat exchanger modules are discussed.

1. Introduction

Heat exchangers constitute a crucial part of process equipment such as cooling towers, where it is used to transfer the heat from a primary fluid to cold n class="Chemical">water [1]. Biofilm formatioclass="Chemical">n iclass="Chemical">n heat exchaclass="Chemical">ngers caclass="Chemical">n have stroclass="Chemical">ng class="Chemical">negative impacts oclass="Chemical">n process efficieclass="Chemical">ncy aclass="Chemical">nd is, therefore, coclass="Chemical">nsidered a problematic challeclass="Chemical">nge faced iclass="Chemical">n class="Chemical">numerous iclass="Chemical">ndustrial sites. The depeclass="Chemical">ndeclass="Chemical">nce of iclass="Chemical">ndustrial processes to heat exchaclass="Chemical">ngers makes the iclass="Chemical">nterruptioclass="Chemical">n of operatioclass="Chemical">n very difficult aclass="Chemical">nd, iclass="Chemical">n maclass="Chemical">ny cases, requires coclass="Chemical">nticlass="Chemical">nuous operatioclass="Chemical">n for several years before cleaclass="Chemical">niclass="Chemical">ng or replacemeclass="Chemical">nt of parts caclass="Chemical">n be performed [2]. Despite the use of biocides aclass="Chemical">nd dispersaclass="Chemical">nts, biofilms still develop over loclass="Chemical">ng periods of time aclass="Chemical">nd coclass="Chemical">nsiderably reduce heat traclass="Chemical">nsfer efficieclass="Chemical">ncy. Additioclass="Chemical">nally, they caclass="Chemical">n serve as a protective habitat for pathogeclass="Chemical">nic bacteria or bacteria iclass="Chemical">nvolved iclass="Chemical">n microbially iclass="Chemical">nduced corrosioclass="Chemical">n, causiclass="Chemical">ng irreversible damage to the equipmeclass="Chemical">nt [3,4]. These detrimeclass="Chemical">ntal coclass="Chemical">nsequeclass="Chemical">nces stroclass="Chemical">ngly affect the capital aclass="Chemical">nd operatioclass="Chemical">nal costs of the heat exchaclass="Chemical">nge iclass="Chemical">nstallatioclass="Chemical">ns [5,6].

Early investigations on biofilm resistance to heat transfer revealed a thermal conductivity of biofilm of 0.6 W/m·K [7], comparable to class="Chemical">water, which is class="Chemical">not surprisiclass="Chemical">ng siclass="Chemical">nce class="Chemical">n class="Chemical">water forms the major mass fraction of a biofilm [8]. Many studies have used this key parameter for monitoring of biofilm growth and assessment of control methods in heat exchanger experiments [9,10,11]. Tian et al. used this approach to evaluate the impact of SiO2 particles inclusion on biofilm structures and heat transfer resistance [12]. It was also used by Chang et al. to investigate the efficiency of thermal shock on biofilm growth inhibition [9].

A potential important aspect for biofilm growth is the presence of a temperature gradient over the length of the heat exchanger affecting the bulk class="Chemical">water temperature. Bacterial commuclass="Chemical">nities composiclass="Chemical">ng the biofilm aclass="Chemical">nd produciclass="Chemical">ng extracellular class="Chemical">n class="Chemical">polymeric substances (EPS) necessary for the biofilm integrity are subjected to this temperature change and may adapt. There is, however, limited literature available on its effect on biofilm composition since most research studies were performed under isothermal conditions [13,14,15]. Among the few studies considering the impact of thermal gradient, Yang et al. have shown that co-current or counter-current configurations of heat exchangers cause different fouling behaviors [16]. Undoubtedly, change in temperature exerts a strong influence on the extent and physical characteristics of biofilms. Due to these observed changes, it is reasonable to assume that variations in the bacterial communities and extracellular compounds also occur along heat exchangers. Characterizing the variations in biofilm composition is of importance to identify or predict which parts of heat exchangers are more detrimentally affected by the biofilm, and to optimize cleaning methods accordingly.

Laboratory scale heat exchanger set-uclass="Chemical">ps have beeclass="Chemical">n desigclass="Chemical">ned iclass="Chemical">n the past [17,18,19] to predict the impact of operatioclass="Chemical">nal variables oclass="Chemical">n biofouliclass="Chemical">ng rates for full-scale applicatioclass="Chemical">ns. These laboratory elemeclass="Chemical">nts are, however, class="Chemical">not suitable for biofilm collectioclass="Chemical">n aclass="Chemical">nd compositioclass="Chemical">n aclass="Chemical">nalyses. A compact plate heat exchaclass="Chemical">nger module, iclass="Chemical">nspired from a well-established membraclass="Chemical">ne fouliclass="Chemical">ng simulator [20], was, therefore, desigclass="Chemical">ned aclass="Chemical">nd built to eclass="Chemical">nable the growth of biofilms of sigclass="Chemical">nificaclass="Chemical">nt thickclass="Chemical">ness aclass="Chemical">nd characterizatioclass="Chemical">n of its compoclass="Chemical">neclass="Chemical">nts. Iclass="Chemical">n this chapter, we describe (i) the developmeclass="Chemical">nt of this class="Chemical">new siclass="Chemical">ngle-plate heat exchaclass="Chemical">nger module for biofilm laboratory iclass="Chemical">nvestigatioclass="Chemical">ns, (ii) its implemeclass="Chemical">ntatioclass="Chemical">n to study the effect of temperature oclass="Chemical">n the bacterial commuclass="Chemical">nity aclass="Chemical">nd class="Chemical">n class="Chemical">EPS composition of biofilms, and (iii) the potential applications and limitations of the module in biofilm characterization studies.

2. Materials and Methods

2.1. Laboratory Heat Exchanger Module

The laboratory heat exchanger modules were designed as single plate heat exchangers. This design was selected due to its convenience for laboratory experiments, allowing easy access to cold and hot channels, and visual monitoring. The flat plate surface facilitates collection of biofilms. Modules were made of two identical PVC elements, fitting each other, and separated by a single class="Chemical">metal plate (Figure 1). PVC coclass="Chemical">nclass="Chemical">nectors built iclass="Chemical">n each side of the chaclass="Chemical">nclass="Chemical">nels allowed coclass="Chemical">nclass="Chemical">nectioclass="Chemical">ns to tubiclass="Chemical">ng. Diffusers were iclass="Chemical">ncluded iclass="Chemical">n the geometry of the chaclass="Chemical">nclass="Chemical">nels to provide a homogeclass="Chemical">neously distributed flow. O-riclass="Chemical">ngs were placed oclass="Chemical">n each side of the class="Chemical">n class="Chemical">metal plate to tightly maintain the plate and separate the water from each channel. Corrosion-resistant metal, Hastelloy C-22, was used for the metal plate. L × W × H dimensions were as follows: 189 × 34 × 8 mm3 for the channels and 189 × 34 × 1 mm3 for the plate. Two identical modules were produced by STT Products BV and operated simultaneously during our experiments. The technical drawing of the modules is provided in the Supplementary Materials (Figure S1).

Figure 1

Heat exchanger module observed from the top side (A), from the left side (B), and view of the inside of the module with a rubber O-ring and metal plate (C).

2.2. Experimental Design

In all experiments, one module was operated in single-pass flow without heating, denoted as “reference module”, and a second module was operated in counter-flow with continuous heating, denoted as “heated module”. The class="Disease">cold channel of each module was fed with class="Chemical">n class="Gene">tap water at a temperature of 20.0 ± 1.4 °C and a flow rate of 0.18 L/min. A nutrient solution composed of sodium acetate trihydrate (CH3COONa·3H2O; for a ‘C’ source), sodium nitrate (NaNO3; for a ‘N’ source) and sodium phosphate monobasic monohydrate (NaH2PO4·H2O; for a ‘P’ source) was dosed to enhance biofilm growth. The concentrations in elements C, N and P were maintained at 500, 100 and 50 µg/L, respectively, in the feed water, corresponding to a mass ratio C:N:P of 100:20:10 [21]. The second channel of the heated module only was operated with recirculating water at 50 °C at a flow rate of 1.8 L/min, without nutrient dosage. The full laboratory set-up is illustrated in Figure 2. Biofilms were grown over a period of 40 days. The experiments were performed in triplicate, denoted as “Experiments 1, 2 and 3”.

Figure 2

Schematic diagram of the heat exchanger set-up in operation.

2.3. Temperature Monitoring

n class="Chemical">Platinum resistaclass="Chemical">nce thermometers of high-precisioclass="Chemical">n (CTP5000 WIKA, Kliclass="Chemical">ngeclass="Chemical">nberg am Maiclass="Chemical">n, Germaclass="Chemical">ny) were located iclass="Chemical">nside the iclass="Chemical">nlet aclass="Chemical">nd outlet tubiclass="Chemical">ng of the chaclass="Chemical">nclass="Chemical">nels, with a 10-cm deep immersioclass="Chemical">n. Temperature probes were calibrated aclass="Chemical">nd showed 0.04 °C accuracy. All probes were coclass="Chemical">nclass="Chemical">nected to a display uclass="Chemical">nit (CTR2000, WIKA, Kliclass="Chemical">ngeclass="Chemical">nberg am Maiclass="Chemical">n, Germaclass="Chemical">ny) aclass="Chemical">nd data were recorded oclass="Chemical">nliclass="Chemical">ne via a data loggiclass="Chemical">ng system oclass="Chemical">n a local computer.

Heat transferred (Q) to the n class="Disease">cold channel iclass="Chemical">n module 2 was calculated with Equatioclass="Chemical">n (1):

m: mass flow rate of n class="Chemical">water (L/s)

Cp: heat capacity of n class="Chemical">water—4184 J/(kg·K)

With n class="Chemical">dT = T

The overall heat transfer coefficient (U) was approximated with Equation (2):

A: area of the plate

LMTD: counter-current logarithmic mean temperature difference

With

Uref was calculated as average U over a period of 1 day, without dosages of nutrients. Uexp was monitored during the experiments to approximated the resistance (R) to heat transfer caused by the biofilm:

The approximated thickness (t) of the biofilm was then obtained with Equation (4): with λ: thermal conductivity of biofilms (0.6 W/m·K)

2.4. Biofilm Collection

At the end of each experiment, both modules were carefully drained and disassembled to recover the biofilms from the surface of the plates. 1 × 2 cm2 areas were sampled from the inlet and the outlet and preserved at −20 °C before DNA extraction and 16S rRNA gene amplicon sequencing. The remaining biofilms were scraped from the plates, frozen at −80 °C and lyophilized before n class="Chemical">EPS extractioclass="Chemical">n.

2.5. DNA Extraction and 16S rRNA Gene Amplicon Sequencing

The genomic DNA was extracted using the DNeasy UltraClean Microbial Kit (Qiagen, Hilden, Germany). Extraction was done following company’s standard protocol with addition of an alternative lysis step. This included a combination of 5 min of heat (65 °C) followed by 5 min of bead-beating on the filters for cell disruption on a Mini-Beadbeater-24 (Biospec, Bartlesville, OK, USA). Samples were sent to Novogene Ltd. (Hongkong, China) for amplicon sequencing of the V3-4 region of the 16S-rRNA gene (position 341–806) on an Illumina paired-end platform. The raw data were processed with the software Mothur v.1.40.5 (Ann Arbor, MI, USA) [22]. Raw sequences were quality filtered, aligned, checked for chimera and operational taxonomic units (OTUs) were generated based on 97% similarities after removal of singletons. The alignment and taxonomic classifications were performed using the SILVA database [23]. The representative OTU sequence of n class="Species">Cupriavidus was compared to the RefSeq NCBI database usiclass="Chemical">ng the Basic Local Aligclass="Chemical">nmeclass="Chemical">nt Search Tool (BLAST) for species ideclass="Chemical">ntificatioclass="Chemical">n. Beta diversity measuremeclass="Chemical">nt was assessed with priclass="Chemical">ncipal coordiclass="Chemical">nate aclass="Chemical">nalysis (PCoA) iclass="Chemical">n Mothur v.1.40.5 usiclass="Chemical">ng the thetaYC distaclass="Chemical">nce matrix. The spatial separatioclass="Chemical">ns visualized iclass="Chemical">n the PCoA are used to compare similarities aclass="Chemical">nd dissimilarities betweeclass="Chemical">n samples.

2.6. EPS Extraction

The extraction of class="Chemical">EPS was performed oclass="Chemical">n the freeze-dried raw deposits. class="Chemical">n class="Chemical">EPS was extracted at 80 °C in alkaline conditions, following a method previously described [24,25], and lyophilized. The freeze-dried EPS samples were kept in a dry environment before further analyses.

2.7. Fourier Transform Infrared Spectroscopy (FTIR)

FTIR spectra of the freeze-dried extracted class="Chemical">EPS samples were carried out oclass="Chemical">n a FTIR Spectrophotometer (Perkiclass="Chemical">nElmer, Waltham, MA, USA) at room temperature, with a waveclass="Chemical">number raclass="Chemical">nge from 700 to 4000 cm−1. Resolutioclass="Chemical">n of 4 cm−1 aclass="Chemical">nd accumulatioclass="Chemical">n of 8 scaclass="Chemical">ns were applied oclass="Chemical">n each sample. FTIR spectra were baseliclass="Chemical">ne-corrected aclass="Chemical">nd a miclass="Chemical">n–max class="Chemical">normalizatioclass="Chemical">n was applied with respect to the class="Chemical">n class="Chemical">amide I peaks (between 1700 and 1600 cm−1) with Matlab R2018b software (Mathworks, Natick, MA, USA).

2.8. Polysaccharide Quantification

class="Chemical">Polysaccharides were quaclass="Chemical">ntified followiclass="Chemical">ng a method developed by [26]. Iclass="Chemical">n short, the freeze-dried class="Chemical">n class="Chemical">EPS extracts were diluted in a sodium hydroxide solution (0.01 M in ultrapure water) to a final concentration of 1000 mg/L. For the standards, a sugar mixture of 1000 mg/L was prepared with equal amounts of fucose, rhamnose, galactose, glucose, xylose, mannose, and ribose. Sugars were selected due to their previous detection in bacterial EPS [27]. The sugar mixture was diluted to concentrations of 0, 10, 25, 50, 75, 100, and 200 mg/L in order to establish the calibration curve. Then, 200 µL of each sample or standards were pipetted in a glass reaction tube, followed by 200 µL of 5% w/v phenol solution and 1000 µL of 95% sulfuric acid. Tubes were vortexed, left at room temperature for 20 min and vortexed again before analyses. The absorbance of the samples and standards were measured at 482 nm with a spectrophotometer (DR3900 Hach, Loveland, CO, USA).

2.9. Protein Quantification

The freeze-dried class="Chemical">EPS extracts were diluted iclass="Chemical">n a class="Chemical">n class="Chemical">sodium hydroxide solution (0.01 M in ultrapure water) to a final concentration of 250 mg/L. Total protein concentrations were determined with the BCA assay using the Protein Quantification Kit (Interchim, Montluçon, France). The standards were prepared by dilution of a bovine serum albumin (BSA) solution to concentrations of 0, 5, 10, 50, 100, 200, 350, and 500 mg/L. Reagents were added to the standards and samples according to manufacturer’s instruction in a 96-well plate. The plate was shaken for 30 s and incubated in the dark at room temperature for 2 h before analysis. Absorbance was measured at 562 nm by a plate reader (Infinite M200 PRO Tecan, Männedorf, Switzerland).

3. Results

3.1. Operating Parameters

The reference module was operated with the class="Disease">cold channel oclass="Chemical">nly, without heaticlass="Chemical">ng. Iclass="Chemical">nlet aclass="Chemical">nd outlet measuremeclass="Chemical">nts iclass="Chemical">ndicated class="Chemical">negligible variatioclass="Chemical">n of the bulk class="Chemical">n class="Chemical">water temperature over the channel length (Table 1). The cold channel of the heated module was operated with the same flow rate, feed source and nutrient dosage as the reference module throughout the experiments.

Table 1

Operational parameters monitored during the operation of the modules, with the interval (±) representing the range of variations. Indicated values combine the data of all experiments.

Module	Temperature Hot Channel (°C)		Temperature Cold Channel (°C)		Flow RateHot Channel(L/h)	Flow RateCold Channel(L/h)	Heat Loss(%)
	Inlet	Outlet	Inlet	Outlet
Reference	r.t. *	r.t. *	20.0 ± 1.4	20.1 ± 1.2	0	11.4	approx. 0
Heated	50.4 ± 0.2	49.3 ± 0.2	20.1 ± 1.3	27.3 ± 1.3	108	11.4	27 ± 3

* r.t.: room temperature.

class="Chemical">Water heated at 50 °C was recirculaticlass="Chemical">ng iclass="Chemical">n couclass="Chemical">nter-flow through a secoclass="Chemical">nd chaclass="Chemical">nclass="Chemical">nel (Figure 1), aloclass="Chemical">ng which a decrease of 1 °C was measured. The heated flow caused a heat traclass="Chemical">nsfer through the class="Chemical">n class="Chemical">metal plate, increasing the bulk water temperature of the cold channel from 20.1 ± 1.3 °C to 27.3 ± 1.3 °C between the inlet and outlet. When no biofilm was present in the heated module, the calculated overall heat transfer coefficient reached 660 ± 40 W/m2 °C, which was used as the reference value for the later calculation of the biofilm heat resistance. Biofilm growth caused a reduction in the temperature difference between inlet and outlet by 1.6 ± 0.2 °C in the cold channel of the heated module. Collection of temperature data along the heated system allowed the monitoring of the biofilm thickness forming on the surface of the metal plate.

3.2. Monitoring of Biofilm Growth in the Heated Module

Approximated biofilm thicknesses were calculated with Equation (4) from online temperature measurements of inlets and outlets of each heated module. Biofilm growth curves of the heated modules are shown in Figure 3. Experiment 1 revealed disturbances due to intermittent air bubbles entering the system causing fluctuations in temperature. This n class="Disease">anomaly was fixed iclass="Chemical">n Experimeclass="Chemical">nt 2 aclass="Chemical">nd Experimeclass="Chemical">nt 3. Despite slight differeclass="Chemical">nces iclass="Chemical">n profile, all experimeclass="Chemical">nts show a similar growth curve with a ficlass="Chemical">nal thickclass="Chemical">ness betweeclass="Chemical">n 200 aclass="Chemical">nd 250 µm after 40 days of operatioclass="Chemical">n. Reproducibility of the study was achieved iclass="Chemical">n terms of biofilm developmeclass="Chemical">nt iclass="Chemical">n the heated heat exchaclass="Chemical">nger module.

Figure 3

Approximated thickness of the biofilms monitored in the heated module over the duration of Experiment 1 (A), Experiment 2 (B), and Experiment 3 (C).

3.3. Biofilm Characterization

3.3.1. Bacterial Community Structure along the Heat Exchanger

Accumulated biofilms were collected from the inlet and outlet of each module to evaluate the changes in bacterial communities along the heated and reference systems, and assess the effect of a temperature gradient.

Dissimilarities between the structure of bacterial communities are illustrated in Figure 4. Although a deviation is visible for a set of data points at coordinates (0.22, −0.64), corresponding to Experiment 3, inlet and outlet of the reference module showed high similarity in each experiment. Main identified grouclass="Chemical">ps iclass="Chemical">n Experimeclass="Chemical">nts 1 aclass="Chemical">nd 2 refereclass="Chemical">nces were Variovorax aclass="Chemical">nd Pisciclass="Chemical">nibacter geclass="Chemical">nera aclass="Chemical">nd Microscillaceae family. Iclass="Chemical">n Experimeclass="Chemical">nt 3, the geclass="Chemical">nus Aquabacterium represeclass="Chemical">nted a large fractioclass="Chemical">n of the total relative abuclass="Chemical">ndaclass="Chemical">nce (Figure S2). Oclass="Chemical">n the coclass="Chemical">ntrary, iclass="Chemical">nlet aclass="Chemical">nd outlet of the heated module experieclass="Chemical">nced a stroclass="Chemical">ng chaclass="Chemical">nge iclass="Chemical">n bacterial commuclass="Chemical">nity structure iclass="Chemical">n each experimeclass="Chemical">nt, represeclass="Chemical">nted iclass="Chemical">n Figure 4 by dashed arrows. Although Aquabacterium, Curvibacter, Microscillaceae, aclass="Chemical">nd Variovorax grouclass="Chemical">n class="Chemical">ps were commonly identified in the inlets, Cupriavidus was dominating the outlets of all three heated modules. Therefore, changes observed along the metal plate in absence of heating were negligible compared to the changes that occurred under heated condition. The temperature gradient along the metal plate thus resulted in a change in biofilm composition in terms of bacterial community members and abundances in the heated module.

Figure 4

Principal coordinate analysis (PCoA) of biofilm bacterial community structures from the heated and reference modules in the three experiments. The bigger the distance between data points, the stronger the dissimilarity in community structures, based on the presence and abundance of operational taxonomic units. Dashed arrows connect inlet and outlet of each module.

Interestingly, outlet biofilm samples of the heated module from the three experiments converge to a similar structure. Taxonomic identification indicated a significant increase in n class="Species">Cupriavidus respiraculi abuclass="Chemical">ndaclass="Chemical">nce, reachiclass="Chemical">ng 52%, 79%, aclass="Chemical">nd 58% iclass="Chemical">n the heated outlet biofilms of Experimeclass="Chemical">nts 1, 2, aclass="Chemical">nd 3, respectively (Figure S2). Results suggest the loss iclass="Chemical">n diversity of maiclass="Chemical">n commuclass="Chemical">nity members at elevated temperatures, with the selectioclass="Chemical">n of class="Chemical">n class="Species">Cupriavidus respiraculi.

3.3.2. EPS Amounts

The total biofilm from each class="Chemical">metal plate was collected to obtaiclass="Chemical">n sufficieclass="Chemical">nt class="Chemical">n class="Chemical">EPS for further characterization. This allowed the comparison of the biofilm composition of the reference module to the biofilm composition of the heated module but not the evaluation of changes along the plates. The relative low amount of EPS per plate did not allow for sampling different areas of the plate separately. The average freeze-dried deposit collected in the reference module reached 10.8 ± 3.0 mg with an extracted EPS yield of 52 ± 2%. These values were not significantly different (p-values > 0.05) to the heated module, with 11.2 ± 2.6 mg and 57 ± 8% of EPS yield. In conclusion, similar amounts of accumulated biofilm and EPS were found in all systems.

3.3.3. EPS Characterization

The FTIR spectra of the extracted class="Chemical">EPS samples display the proteiclass="Chemical">n baclass="Chemical">nd at 1500–1700 cm−1 with the peaks correspoclass="Chemical">ndiclass="Chemical">ng to C=O aclass="Chemical">nd C-N grouclass="Chemical">n class="Chemical">ps of amide I and N-H, C-N, and C-C groups of amide II [28]. The characteristic carbohydrate band is visible at 940–1200 cm−1, corresponding to stretching vibrations of C-O in C-OH and C-C group [29,30]. The comparison of spectra indicates high similarity in molecular vibrations with a large overlapping fraction of the curves of the extracted EPS from the heated and reference modules. Only the carbohydrate peak does not overlap (Figure 5), with a striking divergence between both conditions. A lower absorbance intensity is observed in the EPS grown under heated conditions, suggesting that a relatively lower sugar content of the EPS was produced at elevated temperature.

Figure 5

FTIR of the extracted EPS of the biofilm samples collected on the heated and reference modules. The band associated to carbohydrates at 940–1200 cm−1 shows a strong divergence between the heated and reference conditions [29].

The quantitative analyses of protein and class="Chemical">polysaccharide coclass="Chemical">nteclass="Chemical">nts iclass="Chemical">n the extracted class="Chemical">n class="Chemical">EPS samples led to the results shown in Figure 6A. The percentage of proteins remained quite stable with variations between 48% and 54% of the total EPS, while polysaccharides constituted between 5% and 13% of the total EPS. As observed with the FTIR analyses, the reference EPS seemed to contain a higher amount of polysaccharide than the EPS from the heated module, especially in Experiment 2 (p-value < 0.05) and 3 (p-value < 0.05). The ratio of proteins to polysaccharides was therefore calculated for each module and the values from both tested conditions were compared. Despite a partial overlap in standard deviations, the statistical analysis indicates a significant difference (p-value < 0.05) between the conditions, with a higher PN/PS ratio in the EPS extracted from the heated biofilm, 7.8 ± 2.1% compared to 5.4 ± 1.0% in the EPS extracted from the reference biofilm (Figure 6B).

Figure 6

Composition of EPS in weight percentages of polysaccharides and proteins in the heated and reference modules (A) and protein-to-polysaccharide (PN/PS) ratio under both conditions, averaged over the triplicate experiments (B). The error bars indicate the standard deviations.

In summary, despite similarity in amounts of accumulated class="Chemical">EPS iclass="Chemical">n the refereclass="Chemical">nce aclass="Chemical">nd heated modules, class="Chemical">n class="Chemical">EPS characterization reveals a small but significant change of protein to polysaccharide ratio caused by a relatively lower extracellular sugar production at elevated temperatures.

4. Discussion

4.1. Impact of Temperature on Biofilm Composition

The triplicate experiments performed with the newly designed modules led to reproducible results regarding biofilm growth profile (Figure 3) and class="Chemical">EPS characterizatioclass="Chemical">n (Figure 5 aclass="Chemical">nd Figure 6). Oclass="Chemical">nly the bacterial commuclass="Chemical">nity structures showed a deviatioclass="Chemical">n iclass="Chemical">n Experimeclass="Chemical">nt 3 compared to Experimeclass="Chemical">nts 1 aclass="Chemical">nd 2 (Figure 4), likely caused by the effect of exterclass="Chemical">nal factors affecticlass="Chemical">ng the class="Chemical">n class="Gene">tap water planktonic community. Variation in external temperatures and overall water use in the building could have affected the bacterial community structure [31,32,33] since the experiments were performed sequentially. Despite this deviation, conclusions could be drawn from the results. They are further discussed below.

Experimental analyses of the biofilms grown in the reference and heated modules revealed (i) variations in biofilm community structures caused by the temperature increase along the plate heat exchanger, (ii) comparable amounts of accumulated n class="Chemical">EPS, aclass="Chemical">nd (iii) a relatively lower productioclass="Chemical">n of extracellular class="Chemical">n class="Chemical">polysaccharides at elevated temperature by the biofilm members.

Abundant members of our study include bacterial grouclass="Chemical">ps frequeclass="Chemical">ntly fouclass="Chemical">nd iclass="Chemical">n driclass="Chemical">nkiclass="Chemical">ng class="Chemical">n class="Chemical">water (Figure S2), such as the Piscinibacter genus [34], the Aquabacterium genus populating biofilm with optimal growth at 20 °C [35] or the Variovorax genus [36]. Their high abundance was particularly marked in the reference biofilms, without heating. Selection of the species Cupriavidus respiraculi occurred in the biofilm grown in the heated module (Figure S2). Cupriavidus respiraculi is a gram-negative, obligate aerobe bacterium which has been observed to grow at 28, 32, and 37 °C [37,38]. The species was isolated from patients with cystic fibrosis suggesting that they are well adapted for growth at physiological temperatures. Dominance of Cupriavidus over the drinking water biofilm groups with increasing temperature is an indication that the temperature change faced in heat exchangers leads to biofilm bacterial community compositions diverging from the feed water source communities. It is also important to note that the processes of detachment and dispersion or reattachment of biofilm fragments [39], observed in mature biofilms, did not seem to have affected the community structure of the heated biofilm in the 40-day studies.

These results reveal that the temperature gradient in heat exchangers causes the selection of different biofilm bacterial groun class="Chemical">ps aloclass="Chemical">ng the system, aclass="Chemical">nd caclass="Chemical">n, therefore, cause damages of varyiclass="Chemical">ng class="Chemical">nature aclass="Chemical">nd magclass="Chemical">nitude aloclass="Chemical">ng the process, e.g., regardiclass="Chemical">ng the exteclass="Chemical">nt of biofouliclass="Chemical">ng or microbial corrosioclass="Chemical">n [40]. The effect of heat gradieclass="Chemical">nt should, therefore, be takeclass="Chemical">n iclass="Chemical">nto accouclass="Chemical">nt aclass="Chemical">nd samples should be collected aloclass="Chemical">ng the system wheclass="Chemical">n characteriziclass="Chemical">ng biofilms from heat exchaclass="Chemical">ngers.

Analyses showed that the amounts of biofilms and accumulated class="Chemical">EPS were class="Chemical">not sigclass="Chemical">nificaclass="Chemical">ntly differeclass="Chemical">nt betweeclass="Chemical">n the refereclass="Chemical">nce aclass="Chemical">nd the heated biofilms. It is surprisiclass="Chemical">ng, siclass="Chemical">nce more severe biofouliclass="Chemical">ng is usually observed at higher temperatures [41,42]. Felz et al. has showclass="Chemical">n that class="Chemical">n class="Chemical">EPS yield can be strongly affected by the method of EPS extraction [24]. However, the comparable protein content from both extracted EPS (Figure 6) does not suggest disparities in cell lysis by the alkaline extraction. Biofilm growth and EPS production in the heated biofilm might be limited by the carbon supply or by the wide temperature range along the plate and through the thickness of the biofilm. For a better assessment, the exact temperatures at the surface of the plate and through the biofilm should be monitored in future studies.

The quantitative measurements of proteins and class="Chemical">sugars revealed a lower class="Chemical">n class="Chemical">sugar content of extracted EPS from the heated biofilms (Figure 6), which is in line with the lower band intensity in the region 940–1200 cm−1 observed in the FTIR analyses (Figure 5). The change in temperature therefore induced variations in the produced extracellular compounds. These changes in EPS composition along heated surfaces can affect the efficiency of biofilm removal chemicals such as enzymatic treatments. Studies at larger scale and after long periods of heat exchanger operations are needed for comparison with our laboratory results. Sampling and EPS analyses from a fouled full-scale heat exchanger surface with wide temperature gradient would be of high interest to evaluate the extend of such variations and the implications from an industrial point of view.

4.2. Advantages and Limitations of the Heat Exchanger Laboratory Module

The purpose of the developed lab-scale heat exchanger module is to reproducibly investigate the impact of a continuous temperature gradient on biofilm formation, in terms of active biomass and class="Chemical">EPS amouclass="Chemical">nt aclass="Chemical">nd compositioclass="Chemical">n. It was, therefore, desigclass="Chemical">ned accordiclass="Chemical">ng to characteristics required for carryiclass="Chemical">ng out the research study. Its laboratory-suited size makes it easy to assemble aclass="Chemical">nd disassemble aclass="Chemical">nd allows a flexible use of the device iclass="Chemical">n various locatioclass="Chemical">ns. The height of the chaclass="Chemical">nclass="Chemical">nel was choseclass="Chemical">n so that it caclass="Chemical">n sustaiclass="Chemical">n the growth of a thick biofilm without causiclass="Chemical">ng a build-up iclass="Chemical">n pressure or sigclass="Chemical">nificaclass="Chemical">nt variatioclass="Chemical">n iclass="Chemical">n fluid velocity which could eveclass="Chemical">ntually affect the biofilm developmeclass="Chemical">nt, e.g., through aclass="Chemical">n iclass="Chemical">ncrease iclass="Chemical">n shear rate [43,44]. Stable coclass="Chemical">nditioclass="Chemical">ns could, thus, be maiclass="Chemical">ntaiclass="Chemical">ned duriclass="Chemical">ng the operaticlass="Chemical">ng period. The material used for the class="Chemical">n class="Chemical">metal plate (hastelloy C-22) was selected for its anti-corrosion properties and to allow an optimal heat transfer from the hot to cold channel of the module—hastelloy has a thermal conductivity of 11.1 W/m·K—while the external case was made of PVC to limit heat loss. In addition, the plate can easily be replaced to test different materials, which can broaden the spectrum of potential studies to (microbiologically influenced) corrosion investigations.

There are, however, some limitations to the heat exchanger set-up. A further developed version of the module should include online measurements of plate surface temperatures and investigation of the temperature profile over the height of the biofilm. The temperature at the surface of the plate is generally higher than the bulk temperature and is affected by the boundary layer, which could be approximated by modeling of the system. class="Chemical">Oxygen profile probes would also provide iclass="Chemical">nformatioclass="Chemical">n oclass="Chemical">n the class="Chemical">n class="Chemical">oxygen diffusion through the layer. Design could be enhanced by using material able to handle higher temperatures than PVC can resist, while maintaining an acceptable heat loss. For temperatures higher than 50 °C, it is recommended to change material to thermoplastics with high heat deflection temperature such as polyethersulfones (PES). Lastly, some loss of biofilm was observed during the drainage of the cold channels leading to a small loss in biofilm materials. The handling of this step could be improved in order to reach a full recovery of the grown biofilm.

4.3. Potential Applications in Future Research

When operated over a sufficiently long period of time, the system can be used to perform studies on biofilm properties and compositions that would not be possible with biofilms of a few µm-thickness. The testing of a broader range of temperatures and the use of additional chemical characterization methods of n class="Chemical">EPS would build oclass="Chemical">n the results compiled iclass="Chemical">n this study. Iclass="Chemical">n additioclass="Chemical">n to the prelimiclass="Chemical">nary results described iclass="Chemical">n this research, we also propose some study liclass="Chemical">nes of iclass="Chemical">nterest, which could be tested with the laboratory plate heat exchaclass="Chemical">nger module.

Regarding the characterization of the microbiome, variations can be assessed along the temperature gradient, as introduced in this study. Investigation of the bacterial community changes could also be performed across the thickness of the biofilm [45], by collecting layers of the biofilm formed between the class="Chemical">metal plate aclass="Chemical">nd the bulk class="Chemical">n class="Chemical">water. The differences in main bacterial group, their physiology and interactions would provide a deeper understanding of how the biofilm is able to cope with the significant changes in conditions, not only linked to the diffusion of nutrient and oxygen [46] but also the effect of thermal dissipation.

The applied monitoring provides an approximated thickness, assuming a flat and homogeneous structure of the biofilm over the plate surface. Flow profile, temperature, and nutrient load can, however, affect the growth and the distribution of the biofilm [47]. In addition, it has been seen in pure-culture biofilm that temperature can cause a regulation of n class="Chemical">EPS productioclass="Chemical">n iclass="Chemical">nduciclass="Chemical">ng morphological chaclass="Chemical">nges of the biofilm [15]. To evaluate this assumptioclass="Chemical">n accurately, the temperature moclass="Chemical">nitoriclass="Chemical">ng should be combiclass="Chemical">ned with measuremeclass="Chemical">nt of the thickclass="Chemical">ness. For more iclass="Chemical">n-depth iclass="Chemical">nvestigatioclass="Chemical">n of the physical structure, iclass="Chemical">n-situ imagiclass="Chemical">ng of the biofilm would provide valuable iclass="Chemical">nformatioclass="Chemical">n oclass="Chemical">n the biofilm developmeclass="Chemical">nt aclass="Chemical">nd morphology (e.g., roughclass="Chemical">ness) wheclass="Chemical">n subjected to a thermal gradieclass="Chemical">nt, iclass="Chemical">n a similar way as it was performed for differeclass="Chemical">nt tyclass="Chemical">n class="Chemical">pes of flow [48].

Due to the adjustability of the module, the use of class="Chemical">metal plate proclass="Chemical">ne to corrosioclass="Chemical">n such as class="Chemical">n class="Chemical">copper or carbon steel can be explored. Microbiologically influenced corrosion causes serious damages in full-scale plants [49], and is very challenging to control due to the combination of processes involved in the corrosion mechanisms [50]. The module can for example be used for the testing of metals or innovative coatings, and assessment of their ability to withstand microbial corrosion under different temperatures.

In short, the heat exchanger can be used for a multitude of research purposes. The fact that thick biofilms can be grown in the module and collected in substantial amounts allow the use of multiples biofilm analyses for in-depth characterization of its composition.

5. Conclusions

A laboratory-scale plate heat exchanger module for the assessment of the impact of temperature gradient on biofilm composition was built, tested and applied to a preliminary biofilm investigation.

The online monitoring and analyses showed that:

the developed plate heat exchanger was able to monitor online biofilm growth by measuring its resistance to heat transfer;

the laboratory module is suitable for a large range of applications related to the effect of a thermal field, such as bacterial community identification through the height of the biofilm, composition and morphology of n class="Chemical">EPS at various temperatures, or microbial corrosioclass="Chemical">n iclass="Chemical">nvestigatioclass="Chemical">ns.

The uniqueness of the module lies in its suitability to sustain and monitor extensive biofilm formation with negligible effect on flow properties, and to allow collection of sufficient biofilm material to perform destructive analyses of composition. Some suggestions based on the exploratory study were proposed for further improvement of the module.

Exploratory study on the effect of temperature gradient revealed:

comparable amounts of biofilm and accumulated n class="Chemical">EPS formed iclass="Chemical">n the class="Chemical">noclass="Chemical">n-heated aclass="Chemical">nd heated systems over the 40-day experimeclass="Chemical">nts;

differences in proteins-to-class="Chemical">polysaccharides ratio iclass="Chemical">n extracellular class="Chemical">n class="Chemical">polymeric substances caused by the thermal field, with a lower production of polysaccharides at elevated temperature;

differences in biofilm bacterial groun class="Chemical">ps resulticlass="Chemical">ng from the temperature chaclass="Chemical">nge at the surface of the plate.