Abstract

Thermal pre-flocculation to enable dispersed air flotation is an economical and ecofriendly technology for harvesting microalgae from water. However, the underlying mechanism and optimal conditions for this method remain unclear. In this study, Chlorella vulgaris (C. vulgaris) and Scenedesmus obliquus (S. obliquus) were harvested using a thermal flotation process. The surface structure and characteristics (morphology, electricity, and hydrophobicity) of the microalgae were analyzed using FT-IR (Fourier transform infrared spectroscopy), SEM (scanning electron microscopy), zeta potential, and a hydrophobic test. Further, response surface methodology (RSM) was used to optimize the flotation process. The hydrophobicity of S. obliquus exceeded that of C. vulgaris; as such, under the thermal pre-flocculation, S. obliquus (88.16%) was harvested more efficiently than C. vulgaris (47.16%). Thermal pre-flocculation denatured the lipids, carbohydrate, and proteins of microalgal cell surfaces. This resulted in a decrease in the electrostatic repulsion between the cells and air bubbles. The highest harvesting efficiency was 91.96% at 70 °C, 1,412 rpm, and 13.36 min. The results of this study demonstrate the potential for economic and ecofriendly harvesting of microalgae for biofuels and other bioproducts industries.

INTRODUCTION

Microalgae are unicellular organisms, which usually grow in wastewater, providing several advantages for producing biofuels due to their higher productivity, high lipid content, short growth cycle, high biomass yield, and lack of need for agricultural land compared to conventional oil crops (such as corn and soybean) (Zhou et al. 2011; Uggetti et al. 2018; Xu et al. 2019). However, harvesting is considered the most energy-intensive process (nearly 30% of the total production cost) (Xu et al. 2018b). This is due to the low concentration of cultivated microalgae (the mass concentration is less than 2 g L−1 in photobioreactors, and is far less in open ponds) and the small size of the microalgae, with mostly negative surface charges in wide pH ranges (Rao et al. 2013). Therefore, harvesting optimization is a fundamental need.

Several separation methods are available to harvest microalgae from suspension, including centrifugation (Schenk et al. 2008), filtration (Zhang et al. 2010), sedimentation (Molina Grima et al. 2003) and flocculation (Brennan & Owende 2010). However, due to their shortcomings such as high energy input requirement, being time consuming, and having high cost or low efficiency, these methods are unfavorable for industrial application at large scale. Besides these methods, flotation is widely used to remove solid suspensions from water with the help of chemical surfactants, due to its high efficiency and short harvesting time (Rubio et al. 2002). Flotation can be generally divided into dissolved air flotation (DAF), dispersed air flotation (DiAF), and electro flotation (EF) (Ndikubwimana et al. 2016). Foam flotation, a type of DiAF, shows great potential as a microalgal biomass harvesting and enrichment method (Alkarawi et al. 2018). Through the addition of chemical collectors, the hydrophobicity of particles can be increased and their floatability can also be improved (Zhang & Zhang 2019). However, the use of chemical collectors may harm the aquatic environment and downstream products due to their high toxicity and non-biodegradability.

Thermal flocculation for DiAF (thermal flotation) is considered to be a promising microalgal harvesting method, because it is more environmentally friendly and practical, particularly at a large scale, and has a low cost. Laamanen et al. (2016) found the heating process caused quartets of Scenedesmus sp. to form into small clumps. This clumping emulated the addition of a flocculating agent in traditional bubble flotation systems, without contamination from metal salts or other additives. This process benefits subsequent flotation and improves harvesting efficiency. However, additional research is needed to identify a mechanism to improve the harvesting efficiency of this method and to identify flotation conditions that will maximize harvesting efficiency.

In this study, Chlorella vulgaris (C. vulgaris) and Scenedesmus obliquus (S. obliquus) were used in flotation experiments. A series of tests explored the mechanism of thermal pre-flocculation. In addition, several factors can affect the harvesting efficiency of microalgae biomass, including agitation speed, microalgae concentration, flotation time, pH value of the microalgal suspension, and heating temperature (Ahmed & Hameson 1989; Chen et al. 1998; Wen et al. 2017; Laamanen & Scott 2017; Zou et al. 2018). Therefore, response surface methodology (RSM) was applied as a useful tool to screen the significant variables in the flotation process. A factorial design of 12 experiments were conducted using a Plackett-Burman design (PBD), which is an efficient and effective approach for systematically investigating and evaluating the effects of medium components (Wen et al. 2019). The independent variables were altered to assess optimum harvesting efficiency. The variables included temperature (25–90 °C), agitation speed (800–2,400 rpm), pH value (3–9), microalgae concentration (0.039–0.156 g/L) and flotation time (5–15 min). In addition, a Box-Behnken design (BBD) was used to optimize the thermal flotation process. This method is a class of rotatable or nearly rotatable second-order designs based on three-level incomplete factorial designs (Zou et al. 2019).

MATERIALS AND METHODS

Microalgae cultivation

The microalgae strains (C. vulgaris, FACHB-8; S. obliquus, FACHB-12) used in this study were obtained from the Freshwater Algae Culture Collection at the Institute of Hydrobiology (FACHB-collection, Wuhan, China). A Pyrex photobioreactor (PBR, Shanghai Guangyu Biological Technology Co., Ltd, China), with a base diameter of 25 cm and a height of 100 cm, was used to cultivate the algae. C. vulgaris was grown at 25 ± 3 °C at a light intensity between 3,000 and 3,500 Lux (12 h/d); air was fed at 15 L h−1. The pH was measured and automatically maintained at 7–7.5 by supplying 1 mol/L NaOH and HCl with a peristaltic pump. Microalgae in the stationary growth phase (less than 5% increases in cell numbers per day) were washed twice with distilled water and then used to prepare microalgal suspension samples for the subsequent experiments and measurements. To simulate cultivation in an open pond, concentrations of both of S. obliquus and C. vulgaris biomass were diluted to 0.26 g/L.

THERMAL PRE-FLOCCULATION PROCESS

In this study, thermal flocculation was used as a pretreatment method for microalgae harvesting. Before flotation, the temperature of the microalgae samples was increased using a water bath. A 50–90 °C temperature range was selected for the harvesting test, with a microalgal heating time of 1 hour. After heating, the microalgae samples were placed at room temperature (232 °C) for flotation experiments. To obtain the surface free energy of the C. vulgaris and S. obliquus cells after the heating pre-treatment, the contact angles were measured using three liquids (water, glycol, and glycerol) with known surface free energy components. The surface free energy of the prepared microalgae cells was determined by contact angle measurements using a contact angle goniometer (Kenuo contact angle instrument SL200KS, USA) based on the sessile drop technique. Microalgae cells were pre-concentrated by centrifugation at 2,500 g for 4 min. A flat layer of cells was deposited on 1% (wt/vol) agar and 10% (vol/vol) glycerol for 30 min to stabilize the microalgae cell moisture content. Contact angles were measured as a function of drying time of the cellular lawns in air. The contact angles increased with drying time until a plateau was reached (after about 3 h) (Dengis et al. 1995; Xu et al. 2018a). The degree of hydrophilicity and hydrophobicity of the algae surfaces were determined based on their free energy of cohesion (ΔGcoh). A negative ΔGcoh indicates hydrophobicity where surface-surface interactions are stronger than surface-water interactions; a positive value indicates hydrophilicity (Hiemenz & Rajagopalan 1997). The calculations are shown in Equations (1) and (2): 
formula
(1)
 
formula
(2)

In these equations, θ is the measured contact angle; and subscripts of s and l refer to the solid surface and probe liquid, respectively. γLW, γ, and γ+ are the Lifshitz-van der Waals component, the Lewis acid component, and a Lewis base component, respectively. The subscripts s and l denote the solid surface (microalgae) and the liquid. The reference data for surface tension components of three probe liquids (including in water, glycerol and glycol) employed in the present contact angle measurements in this study were present (Table S1, available with the online version of this paper) (Oss 1995). Because the medium of the flotation system is water, water was chosen to calculate the value of .

The microalgae cells were characterized before and after thermal pre-flocculation using a Fourier transform infrared method (FT-IR, PerkinElmer Spectrum Two, American) and zeta meter (DelsaTM Nano Beckman Coulter, USA). The hydrophobicity of microalgae was quantified using a modified adherence-to-hydrocarbon method (Rosenberg et al. 1980). The test assesses essentially the distribution ratio of cells between water and an organic phase. The microalgae sample (4 mL) was placed in a test tube to which 1 mL of 98% pure n-hexane was added and shaken vigorously for 1 min; the resulting suspension was allowed to settle for 2 min. Afterwards, 2 mL was carefully drawn from the aqueous layer at the bottom of the test tube, placed in a UV cuvette, and the absorbance was read at 620 nm using a spectrophotometer (Shimadzu, Japan UV-2,450); the proportion of microalgae cells that had moved to the water-hexane interface can be determined (Garg et al. 2012). The extractability (H) of the hexane layer on organic substances in the algal suspension was calculated using the following expression: 
formula
(3)
In this expression, A0 is the initial absorbance of the microalgae suspension and Aw is the absorbance of the aqueous phase after settling for 2 min.

Flotation experiments

Flotation experiments were conducted using a 1.0 L Denver Flotation Cell (ShunZe, XFD-1, China). The pH of the flotation pulp was adjusted using HCl (1 mol/L) or NaOH (1 mol/L). Initially, the microalgal suspension was conditioned using mixing at 800 rpm for 5 min, followed by mixing at 600 rpm for 10 min for the flotation test. All flotation harvests were conducted under an air flow rate of 180 litres/h. For the first set of experiments, the pH value of the flotation experiments was the initial pH of the microalgae medium (7–7.5). When mixing was stopped, the ‘microalgae-bubble’ aggregates that floated to the top of the flotation column after 10 minutes were harvested. The concentration of both S. obliquus and C. vulgaris biomass was 0.26 g/L. Only the temperature of these six experiments was different (25, 50, 60, 70, 80 and 90 °C). All results are presented as the average of three measurements.

Microalgal harvesting efficiency and concentration factor were determined using Equations (4) and (5). 
formula
(4)
 
formula
(5)
In these expressions, is the initial absorbance at 540 nm (C. vulgaris) or 680 nm (S. obliquus); (at the same absorbance as OD1) is the final absorbance of the microalgae that floated on the water surface; is the final concentration of microalgae of the subnatant after flotation.

Plackett-Burman design

The experimental process to optimize the harvesting efficiency of S. obliquus was designed using RSM. The optimal RSM design applied five factors at one response level. Design-Expert 10 software was used for RSM design. The PBD was used for screening, to determine the key process variables and the best combination for thermal flotation (Naveena et al. 2005). Based on previous studies on air flotation and thermal pre-flocculation (Laamanen et al. 2016; Chen et al. 2018), temperature (A, 25–90 °C), agitation speed (B, 800–2,400 rpm), pH value (C, 3–9), microalgae concentration (D, 0.039–0.156 g/L), and flotation time (E, 5–15 min) were tested for their significance in S. obliquus harvesting. Each variable was set at a high level (+1) and a low level (−1), as well as a middle level of 0. A total of 12 groups of experiments were designed. Table S2 (available online) shows the design in the coded units and the experimental results for the harvesting efficiency.

Box-Behnken design

The BBD was used to optimize the significant factors in the PBD screening, including temperature, agitation speed, and flotation time. Table S4 (available online) shows the design of variables for BBD. Each independent variable was coded with three different levels (−1, 0, +1). A total of 17 runs were used to optimize harvesting efficiency. Table S4 shows the design in coded units and the harvesting efficiency experimental results. The harvesting efficiency (%) of S. obliquus was used as the response variable. The relationship between the independent variables was estimated as: 
formula
(6)
In this expression, Y is the predicted response; Xi and Xj are independent factors; is the intercept; is the linear coefficient; is the quadratic coefficient; and is the interaction coefficient.

RESULT AND DISCUSSION

Harvesting efficiency of microalgae with thermal pre-flocculant

Figure 1 shows the results of the first set of experiments, which assessed harvesting efficiency of C. vulgaris and S. obliquus under different pre-flocculation temperatures. When the temperature rose from 25 °C to 70 °C, the harvesting efficiency and concentration factor of C. vulgaris increased by 19.37% and 2.46%, respectively. The harvesting efficiency increased at 47.16%, when the temperature reached 80 °C. However, S. obliquus harvesting efficiency significantly increased by 59.28% when the temperature rose from 60 °C to 70 °C. When the temperature exceeded 70 °C, the harvesting efficiency and concentration factor remained stable at about 90.00% and 10.00, respectively. The highest harvesting efficiency of C. vulgaris and S. obliquus was 47.16% and 88.16%, respectively. The highest concentration factor of these two strains of microalgae was 6.28 and 11.82, respectively.

Figure 1

The harvesting efficiency of (a) C. vulgaris and (b) S. obliquus under different pre-flocculation temperatures. SEM images showing that microalgae cells began to gather and form flocs at 80 °C. Error bars represent the standard error of the mean (n = 3).

Figure 1

The harvesting efficiency of (a) C. vulgaris and (b) S. obliquus under different pre-flocculation temperatures. SEM images showing that microalgae cells began to gather and form flocs at 80 °C. Error bars represent the standard error of the mean (n = 3).

Zhao et al. (2015) found that, for a stable ‘bubble-microalgae’ attachment, a liquid film formed between the bubble and microalgae surface needs to be ruptured, and then a liquid/gas/solid three-phase contact line must be formed during the flotation process. With the increase of pretreatment temperature, the absolute value of zeta potential of microalgae cells decreased and the microalgae cells turned from hydrophilicity to hydrophobicity. Therefore, the electrostatic repulsion between cells and bubbles decreased and the hydrophobic attraction force between air bubbles and particles was enhanced. These processes can promote the attachment of microalgae cells and bubbles, as well as the rupture of the liquid film. Thus, the harvesting efficiency and concentration factor of the microalgae increased.

The scanning electron microscope (SEM) images in Figure 1 show that some microalgae cells began to gather and form flocs as the temperature increased. Compared with C. vulgaris, there were more flocs in the SEM images of S. obliquus. Therefore, the heating process can emulate the addition of flocculants, making microalgae flocs form more easily. Chen & Hu (1985) identified many factors influencing the rising velocity of ‘bubble-microalgae’ aggregates in a flotation system. These factors affect the harvesting efficiency of microalgae, and include the density difference between water and the ‘bubble-microalgae’ aggregates, aggregate diameters, and the dynamic state of the flotation process. The microalgae forming the flocs adhered more easily to more bubbles, leading to a smaller floc density. Zou et al. (2018) found it necessary to flocculate microalgae cells, to form flocs through the pre-flocculant process. Using a thermal pre-flocculant causes the microalgae cells to form flocs and avoids chemical flocculant pollution. This makes it a useful and ecofriendly pre-flocculant process.

The effects of heating on the surface properties of microalgae

Figure 2(b) shows the FT-IR spectroscopy of S. obliquus and C. vulgaris. This analysis was applied to obtain information about the biomass structure and chemical changes that occurred during pretreatment (Salehian et al. 2013). For the C. vulgaris without thermal pretreatment, the band around 3,420 cm−1 corresponded to O-H stretching. The broad adsorption band around 2,928 cm−1 was attributed to the stretching of C-H (Lipid-carbohydrate). The band around 1,659 cm−1 was the amide I group (mainly C = O stretching), and the band at 1,547 cm−1 was the band of amide II (mainly N-H bending and C-N stretching) (Ng et al. 2012). The C-O-C band of polysaccharides and the nucleic acid P = O stretching of phosphodiesters were assigned to the wavenumber ranges 1,155 cm−1 and 1,060 cm−1. The peak around 1,383 cm−1 was ascribed to the protein (C-H) bending of the methyl and carboxylic acid C-O of the COO groups of carboxylates. The surface substances of C. vulgaris should include lipids, carbohydrates, and proteins. Compared with the spectra of C. vulgaris at 90 °C, there seems to have been no obvious change in the FT-IR spectra of C. vulgaris at 25 °C. Figure 2(a) shows that the position of the absorption band in the FT-IR spectra for S. obliquus was the same as for C. vulgaris; however, the absorption intensity differed. In Figure 2(a), the untreated sample showed a small band around 1,060 cm−1 (C-O-C of polysaccharides), 1,660 cm−1 (protein amide I band and C = O stretching), and 2,932 cm−1 (Lipid-carbohydrate C-H). As the temperature rose, the thermal pretreatment affected the carbohydrate, lipid, and protein structures. Therefore, it can be concluded that thermal pretreatment had few effects on the cell surface composition of C. vulgaris. However, this process may significantly change the surface characteristics of S. obliquus.

Figure 2

The FT-IR absorption spectra of (a) S. obliquus and (b) C. vulgaris. Yellow, red, and blue lines indicate the microalgae thermal pre-treatment at 25 °C, 70 °C, and 90 °C, respectively. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/wst.2019.287.

Figure 2

The FT-IR absorption spectra of (a) S. obliquus and (b) C. vulgaris. Yellow, red, and blue lines indicate the microalgae thermal pre-treatment at 25 °C, 70 °C, and 90 °C, respectively. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/wst.2019.287.

Surface energy

Table 1 summarizes the surface characteristics of C. vulgaris and S. obliquus before and after the heating treatment. At a normal temperature (25 °C), the values of C. vulgaris and S. obliquus were 6.074 and 5.996, respectively. As the temperature increased, the measured value of gradually improved in all three tested liquids. At 70 °C, the values (−15.322) for the microalgae reversed. According to Shen et al. (2018), the ΔGcoh value is related to the hydrophobicity of the particle surface. A negative ΔGcoh value indicates hydrophobicity, while a positive value indicates hydrophilicity (Ozkan & Berberoglu 2013a, 2013b). Meanwhile, the surface hydrophobicity of the microalgae was a key factor affecting microalgae flotation (Garg et al. 2014). The hydrophobic attraction force between air bubbles and microalgae cells can be enhanced by increasing particle hydrophobicity at room temperature (Zhang & Zhang 2019). When the temperature reached 90 °C, the value of of S. obliquus was two times that of C. vulgaris. Meanwhile, the harvesting efficiency and concentration factor values for S. obliquus were 88.16% and 11.82, respectively. The highest harvesting efficiency and concentration factor values for C. vulgaris were 47.16% and 6.24, respectively. Compared with C. vulgaris, S. obliquus was more suitable for thermal pre-flocculation, due to different compositions of lipids, carbohydrate, and proteins between the C. vulgaris and S. obliquus cell surfaces; the hydrophobicity can be significantly influenced by thermal pretreatment.

Table 1

The surface characteristics of C. vulgaris (S. obliquus) before and after heating treatment

 C. vulgaris contact angles/(°)
 
 
   
25 °C 62.99 ± 1.82 55.35 ± 1.79 68.59 ± 2.13 6.074 
50 °C 64.49 ± 0.63 55.24 ± 1.12 68.63 ± 2.05 2.016 
70 °C 70.63 ± 0.67 44.03 ± 1.35 61.98 ± 0.54 −15.322 
90 °C 91.34 ± 1.03 45.26 ± 1.17 65.27 ± 0.92 −28.284 
 S. obliquus contact angles/(°)
 
 
     
25 °C 50.80 ± 1.07 49.39 ± 1.14 62.57 ± 1.74 5.996 
50 °C 54.88 ± 1.53 46.25 ± 1.32 60.50 ± 1.25 1.526 
70 °C 80.74 ± 1.32 44.65 ± 1.56 63.78 ± 0.73 −25.332 
90 °C 95.22 ± 0.75 47.48 ± 0.84 68.24 ± 1.16 −61.129 
 C. vulgaris contact angles/(°)
 
 
   
25 °C 62.99 ± 1.82 55.35 ± 1.79 68.59 ± 2.13 6.074 
50 °C 64.49 ± 0.63 55.24 ± 1.12 68.63 ± 2.05 2.016 
70 °C 70.63 ± 0.67 44.03 ± 1.35 61.98 ± 0.54 −15.322 
90 °C 91.34 ± 1.03 45.26 ± 1.17 65.27 ± 0.92 −28.284 
 S. obliquus contact angles/(°)
 
 
     
25 °C 50.80 ± 1.07 49.39 ± 1.14 62.57 ± 1.74 5.996 
50 °C 54.88 ± 1.53 46.25 ± 1.32 60.50 ± 1.25 1.526 
70 °C 80.74 ± 1.32 44.65 ± 1.56 63.78 ± 0.73 −25.332 
90 °C 95.22 ± 0.75 47.48 ± 0.84 68.24 ± 1.16 −61.129 

Data given as average values with standard error of the mean (n = 3).

The relationship between hydrophobicity and zeta potential

A previous study showed that the electrostatic properties of the bubbles and particles are the most important parameters controlling the harvesting efficiency of flotation experiments (Han 2002). Without the addition of chemical coagulants, air bubbles have an IEP (isoelectric point, pH of net charge) at pH < 3 and negative zeta potentials of about −25 mV over the pH range of 6–8 (Dockko & Han 2004; Edzwald 2010). Besides, the zeta potentials of these two kinds of microalgae species were negative. Therefore, there was an electrostatic repulsion between microalgae cells and air bubbles, hindering the contact between cells and bubbles. Figure 3(a) shows that the zeta potential of the two microalgae strains first increased and then remained stable. When temperature rose from 25 to 90 °C, the zeta potential of C. vulgaris and S. obliquus increased by 6.77 and 9.18 mV, respectively. Hence, the electrostatic repulsion can be decreased and the harvesting efficiency can be improved. Compared with C. vulgaris, the absolute value of the zeta potential of S. obliquus significantly decreased because of the thermal pretreatment. This process made S. obliquus easier to form the ‘S. obliquus-bubble’ aggregates. Thus, the harvesting efficiency of S. obliquus can be greatly improved.

Figure 3

(a) The microalgae hydrophobicity and zeta potential of C. vulgaris and S. obliquus under different temperatures; (b) the linear fit of surface hydrophobicity and zeta potential for C. vulgaris and S. obliquus under different heating temperature. Error bars represent the standard error around the mean (n = 3).

Figure 3

(a) The microalgae hydrophobicity and zeta potential of C. vulgaris and S. obliquus under different temperatures; (b) the linear fit of surface hydrophobicity and zeta potential for C. vulgaris and S. obliquus under different heating temperature. Error bars represent the standard error around the mean (n = 3).

Figure 3(a) shows the hydrophobicity and zeta potential of C. vulgaris and S. obliquus at different temperatures. At room temperature, the hydrophobicity values of C. vulgaris and S. obliquus were 15.62% and 20.37%, respectively. When the temperature increased from 25 °C to 90 °C, the maximum hydrophobicity values of C. vulgaris and S. obliquus were 31.90% and 64.68%, respectively. Zhang & Zhang (2019) reported that with the increase of hydrophobicity of particle, the floatability of particles can be also improved. The hydrophobicity of S. obliquus was higher than C. vulgaris; as such, the harvesting efficiency of S. obliquus was also better than C. vulgaris.

Figure 3(b) shows a linear fit between the surface hydrophobicity and zeta potential for C. vulgaris and S. obliquus at different temperatures. The Pearson correlation coefficients (r) for C. vulgaris and S. obliquus were −0.9482 and −0.9929, respectively. The surface hydrophobicity of these two strains microalgae were strongly negatively correlated with the absolute value of zeta potential. As the temperature increased, the homocharge of microalgae cells decreased, leading to a decline in the electrostatic repulsion between cells. The aggregation of microalgae resulted in the partial compression of protein molecules at the interface (Wong et al. 2011). This may be why the hydrophobicity of microalgae cells increased as the absolute value of zeta potential decreased.

Screening of the significant factors using Plackett-Burman design

Jarvis et al. (2009) indicated that the differences in removal for different algae reflected the differences in structure between species. As aforementioned, S. obliquus was more suitable for thermal flotation, because the surface characteristic (zeta potential and hydrophobicity) of S. obliquus can be significantly adjusted by thermal pretreatment. In contrast, few effects on the cell surface composition of C. vulgaris were found. Therefore, S. obliquus was selected for further optimization experiments. In the second set of experiments, RSM was used to improve the harvesting efficiency of S. obliquus, using thermal pre-flocculation. The PBD test was used to identify the most significant variables from those that were less important. The possible influencing factors included temperature (°C), agitation speed (rpm), flotation time, pH value, and microalgae concentration.

Table S3 (available with the online version of this paper) shows the assigned concentration of variables at different levels and the effect estimates using PBD. Table S3 shows that the model was significant, as evaluated by the p-value. The correlation coefficient (R2) was an appropriate indicator for the regression of the model. In this study, the R2 (93.31%) showed that only 6.69% of the total variability could not be explained by the model. The higher the R2 was, the closer the data fit the model. An R2 value alpha level of 0.10 was established to highlight all possible significant variables and variable interactions (Coward et al. 2013). According to the p-value displayed in Table S3, temperature, agitation speed, and flotation time were the significant factors (<0.05) driving flotation in the thermal pre-flocculation.

Temperature was the factor with the largest effect (p = 0.0002); the coefficient estimate showed a positive effect between the temperature and harvesting efficiency. Ducker et al. (1994) found that the hydrophobicity of particles was directly correlated with the attractive force between particles and bubbles. In thermal flotation, with the increase of temperature, the characteristics of the microalgae cells can be changed. Since the zeta potential absolute value of microalgae cells decreased, the electrostatic repulsion between microalgae cells decreased. According to Wong et al. (2011), the aggregation of microalgae resulted in the partial compression of protein molecules at the interface, increasing the hydrophobicity of microalgae cells. Based on the present microalgae hydrophobicity tests, the hydrophobicity of microalgae cells was improved. The hydrophobic microalgae cells are easier to attach to bubbles. Therefore, the harvesting efficiency can be improved by increasing the hydrophobic attraction and decreasing electrostatic repulsion. For example, when temperature increased from 25 °C to 90 °C, the zeta potential absolute value decreased from 15.26 to 6.08 mV, the hydrophobicity increased from 20.37% to 64.68% and the harvesting efficiency increased to over 90.00%.

Flotation time had the second largest effect on the concentration factor (p = 0.0212). The coefficient estimate showed that the flotation time positively affected harvesting efficiency. Therefore, a longer flotation time can improve harvesting efficiency. With the increase of flotation time, the chance of bubble-cell collision can be increased (Chen et al. 1998). The longer the flotation time was, the more aggregates were collected.

The agitation speed was also significant (p = 0.0212), with a coefficient estimate of −10.23. This indicated a negative effect between agitation speed and harvesting efficiency. In this thermal flotation process, air is introduced into the negative pressure region formed by the high-speed stirring of an impeller, and subsequently, air is sheared into microbubbles under intense agitation. A slow agitation speed may decrease the number of air bubbles, negatively affecting the formation of ‘microalgae-bubble’ aggregates. Therefore, there may also be a decrease in the number of microalgae cells floating on the suspension surface. Under high agitation speed, more small bubbles with thin liquid films were generated (Ahmed & Hameson 1989). Xing et al. (2017) noted that the film drainage speed increased with decreasing bubble size due to the increased driving pressure. Coward et al. (2015) reported that small bubbles could enhance the collision and attachment efficiencies so as to improve the flotation performance of microalgal harvesting. Therefore, an increased number of bubbles enhanced the collision frequency between air bubbles and S. obliquus cells, and thus improved the harvesting efficiency of S. obliquus. In addition, with the increase of small bubbles, the amount of liquid trapped between the foam lamellae increased and more liquid was removed from suspension (Coward et al. 2013). In conclusion, the increase of agitation speed can produce more small bubbles, which was generally beneficial in promoting collision between bubbles and particles and thus improved the harvesting efficiency. However, when the agitation speed was too fast, it destroyed the microalgae flocs formed in the thermal pre-flocculation process, decreasing the harvesting efficiency of S. obliquus.

Optimization of significant variables: Box-Behnken design

Based on the analysis of variance (ANOVA) of PBD test (Table S3), temperature (x1), agitation speed (x2) and flotation time (x3) were selected as the significant factors. The BBD was used to optimize the thermal flotation process, with each factor tested at three levels (−1, 0, +1). Table S5 (available online) shows the ANOVA of the estimated effects and the coefficients obtained for S. obliquus harvesting efficiency. The value associated with a ‘lack of fit’ was 1.88 (>0.05), which showed agreement between the experimental data and any predicated response value. This indicated that the model diagnostic is appropriate. The model F-values exceeded 75.60 and very low P-values (<0.05) indicated that all models were significant (Zheng et al. 2012). The R2 value was 98.98%, indicating that only 1.02% of the total variability cannot be explained by the model. The BBD tests indicate that the predicted response, harvesting efficiency (Y), can be obtained and presented using (Equation (7)). Additional experiments were conducted using the optimized conditions. 
formula
(7)

To verify model accuracy in predicting optimum conditions (agitation speed of 1,142 rpm, microalgae suspension temperature of 90 °C and flotation time of 13.36 min), three sets of additional experiments were conducted in optimized conditions. These additional experiments (91.96%) showed the model accurately predicts harvesting efficiency.

Figure 4 shows the 3D response surface plots of the model equations fitted to the data. It shows the relationship between the factors and helps determine the optimum level of each factor for maximum response. In Figure 4(a), an increased temperature significantly improved the harvesting efficiency of S. obliquus. Harvesting efficiency first increased rapidly at a low agitation speed and then decreased gradually to a stable value as the agitation speed increased. Compared with temperature, agitation speed didn't have a significant influence on harvesting efficiency. Harvesting efficiency reached its maximum level (91.96%) when the agitation speed was 1,412 rpm and the temperature was 90 °C. As aforementioned, harvesting efficiency slightly increased by 1.56% when the temperature increased from 70 °C to 90 °C. To save energy of this harvesting method, experiments were not needed at a higher temperature. Thus, the optimized temperature was set at 70 °C.

Figure 4

The 3D response surface curve for harvesting efficiency (%) of S. obliquus as a function of (a) agitation speed (rpm) and temperature (°C), (b) flotation time (min) and temperature (°C), and (c) flotation time (min) and agitation speed (rpm).

Figure 4

The 3D response surface curve for harvesting efficiency (%) of S. obliquus as a function of (a) agitation speed (rpm) and temperature (°C), (b) flotation time (min) and temperature (°C), and (c) flotation time (min) and agitation speed (rpm).

Figure 4(b) shows the relationship between temperature, flotation time, and harvesting efficiency. Similar to Figure 4(a), temperature had a great effect on the harvesting efficiency compared with flotation time. However, the harvesting efficiency gradually increased with increasing flotation time; the harvesting efficiency was highest (91.02%) with a medium flotation time (13.36 min). However, Figure 4(c) shows that the interaction of agitation speed and flotation time was not significant and the highest point was 91.93%. The model shows that the optimal conditions for S. obliquus harvesting efficiency with thermal pre-flocculation were a temperature of 70 °C, an agitation speed of 1,412 rpm, and a flotation time of 15 min.

CONCLUSION

This study identified the mechanism involved in the thermal flotation process, by investigating the flotation behaviors of S. obliquus and C. vulgaris. Experimental results showed that the harvesting efficiencies of S. obliquus and C. vulgaris were 47.16% and 88.16%, respectively. Measurement results using FT-IR spectroscopy, zeta potential, and hydrophobic testing revealed that thermal pre-treatment had a more significant effect (through enhanced hydrophobicity) on S. obliquus compared to C. vulgaris. Thus, the harvesting efficiency of S. obliquus exceeded that of C. vulgaris.

In addition, a PBD test screened the most significant factors in the thermal flotation process. The BBD test revealed that harvesting efficiency was approximately 91.96% at a temperature of 70 °C, an agitation speed of 1,412 rpm, and a flotation time of 13.36 min. Moreover, waste heat from thermal plants can support the heating process and improve the harvesting efficiency. Furthermore, microalgae can be used to purify industrial exhaust gas (NOx, CO2, SOx) emissions. Therefore, thermal flotation has a beneficial cost-benefit relationship compared to other microalgae harvesting methods. Future research should focus on applying this process and developing an industrial-scale approach.

ACKNOWLEDGEMENTS

The authors would like to thank the National Natural Science Foundation of China (51478045), Key Laboratory of Degraded and Unused Land Consolidation Engineering of the Ministry of Land and Resources of China (SXDJ2017-6), supported by the Fund Project of Shaanxi Key Laboratory of Land Consolidation (2018-ZD04), and the Special Fund for Basic Scientific Research of Central Colleges, Chang'an University (300102299703 and 300102299708) for funding this project.

CONFLICT OF INTEREST STATEMENT

There are no potential financial or other interests that could be perceived to influence the outcomes of the research.

REFERENCES

REFERENCES
Ahmed
N.
Hameson
G. J.
1989
Flotation kinetics
.
Process. Extr. Metall. Rev.
5
(
1–4
),
77
99
.
Alkarawi
M.
Caldwell
G.
Lee
J.
2018
Continuous harvesting of microalgae biomass using foam flotation
.
Algal Res.
36
,
125
138
.
Chen
Y.
Hu
B.
1985
Research and Application of Air Floatation Water Purification Technology
.
Shanghai Science and Technology Press
,
Shanghai, China
.
Chen
Y. M.
Liu
J. C.
Ju
Y. H.
1998
Flotation removal of algae from water
.
Colloids Surf., B
12
(
1
),
49
55
.
Coward
T.
Lee
J. G. M.
Caldwell
G. S.
2013
Development of a foam flotation system for harvesting microalgae biomass
.
Algal Res.
2
(
2
),
135
144
.
Dengis
P. B.
Nélissen
L. R.
Rouxhet
P. G.
1995
Mechanisms of yeast flocculation: comparison of top- and bottom-fermenting strains
.
Appl. Environ. Microbiol.
61
(
2
),
718
728
.
Dockko
S.
Han
M.
2004
Fundamental characteristics of bubbles and ramifications for the flotation process
.
Water Sci. Technol. Water Supply
50
(
12
),
207
214
.
Edzwald
J. K.
2010
Dissolved air flotation and me
.
Water Res.
44
(
7
),
2077
2106
.
Garg
S.
Li
Y.
Wang
L.
Schenk
P. M.
2012
Flotation of marine microalgae: effect of algal hydrophobicity
.
Bioresour. Technol.
121
(
10
),
471
474
.
Hiemenz
P. C.
Rajagopalan
R.
1997
Principles of Colloid and Surface Chemistry
.
CRC
,
New York, NY, USA
.
Jarvis
P.
Buckingham
P.
Holden
B.
Jefferson
B.
2009
Low energy ballasted flotation
.
Water Res.
43
(
14
),
3427
3434
.
Laamanen
C. A.
Senhorinho
G. N. A.
Ross
G. M.
Scott
J. A.
2016
Heat-aided flocculation for flotation harvesting of microalgae
.
Algal Res.
20
,
213
217
.
Molina Grima
E.
Belarbi
E. H.
Acién Fernández
F. G.
Robles Medina
A.
Chisti
Y.
2003
Recovery of microalgal biomass and metabolites: process options and economics
.
Biotechnol. Adv.
20
,
491
515
.
Ndikubwimana
T.
Chang
J.
Xiao
Z.
Shao
W.
Zeng
X.
Ng
I.
Lu
Y.
2016
Flotation: a promising microalgae harvesting and dewatering technology for biofuels production
.
Biotechnol. J.
11
(
3
),
315
326
.
Ng
M.
Liana
A. E.
Liu
S.
Lim
M.
Chow
C. W.
Wang
D.
Drikas
M.
Amal
R.
2012
Preparation and characterisation of new-polyaluminum chloride-chitosan composite coagulant
.
Water Res.
46
(
15
),
4614
4620
.
Ozkan
A.
Berberoglu
H.
2013a
Cell to substratum and cell to cell interactions of microalgae
.
Colloids Surf., B
112
,
302
309
.
Ozkan
A.
Berberoglu
H.
2013b
Physico-chemical surface properties of microalgae
.
Colloids Surf., B
112
,
287
293
.
Rubio
J.
Souza
M. L.
Smith
R. W.
2002
Overview of flotation as a wastewater treatment technique
.
Miner. Eng.
15
(
3
),
139
155
.
Salehian
P.
Karimi
K.
Zilouei
H.
Jeihanipour
A.
2013
Improvement of biogas production from pine wood by alkali pretreatment
.
Fuel
106
(
2
),
484
489
.
Schenk
P. M.
Thomas-Hall
S. R.
Stephens
E.
Marx
U. C.
Mussgnug
J. H.
Posten
C.
Kruse
O.
Hankamer
B.
2008
Second generation biofuels: high-efficiency microalgae for biodiesel production
.
Bioenergy Res.
1
(
1
),
20
43
.
Uggetti
E.
García
J.
Álvarez
J. A.
García-Galán
M. J.
2018
Start-up of a microalgae-based treatment system within the biorefinery concept: from wastewater to bioproducts
.
Water Sci. Technol.
78
(
1
),
114
124
.
Wen
H.
Li
Y.
Shen
Z.
Ren
X.
Zhang
W.
Liu
J.
2017
Surface characteristics of microalgae and their effects on harvesting performance by air flotation
.
Int. J. Agr. Biol. Eng.
10
,
125
133
.
Zhang
X.
Hu
Q.
Sommerfeld
M.
Puruhito
E.
Chen
Y.
2010
Harvesting algal biomass for biofuels using ultrafiltration membranes
.
Bioresour. Technol.
101
(
14
),
5297
5304
.
Zheng
H.
Gao
Z.
Yin
J.
Tang
X.
Ji
X.
Huang
H.
2012
Harvesting of microalgae by flocculation with poly (γ-glutamic acid)
.
Bioresour. Technol.
112
,
212
220
.
Zou
X.
Xu
K.
Wen
H.
Xue
Y.
Zhao
S.
Xie
W.
Li
Y.
2019
A novel method to recover microalgae by compound buoyant-bead flotation
.
Sep. Purif. Technol.
211
,
658
666
.

Supplementary data