- Email: [email protected]
Statistical optimization of parameters affecting lipid productivity of microalga Chlorella protothecoides cultivated in photobioreactor under nitrogen starvation
south african journal of chemical engineering 23 (2017) 26e37
Contents lists available at ScienceDirect
South African Journal of Chemical Engineering journal homepage: http://www.journals.elsevier.com/ south-african-journal-of-chemical-engineering
Statistical optimization of parameters affecting lipid productivity of microalga Chlorella protothecoides cultivated in photobioreactor under nitrogen starvation Prakash Binnal*, P. Nirguna Babu Department of Chemical Engineering, Siddaganga Institute of Technology, Tumkur, 572103, Karnataka, India
article info
abstract
Article history:
In the present work, a lab scale photobioreactor was used to evaluate lipid productivity and
Received 11 October 2016
carbon dioxide fixation rate of microalgae Chlorella protothecoides under nitrogen deplete
Received in revised form
conditions. Effect of environmental conditions such as pH, temperature, light intensity,
1 January 2017
photoperiod (light to dark cycle ratio), CO2 concentration in air and aeration rate on lipid
Accepted 12 January 2017
productivity and carbon fixation rate of microalga was studied. Response surface methodology was adopted to optimize these conditions. All the parameters were found to be
Keywords:
statistically significant. Best operating conditions were evaluated to be: pH-6.51, Temper-
Chlorella protothecoides
ature-28.63 C, light intensity-5.31 klux, Photoperiod-15.36 h:8.64 h, CO2 concentration in
Photobioreactor
air-6.26% (v/v), Aeration rate 2.92 lpm. Lipid productivity under these conditions was
Nitrogen starvation
found to be 274.15 mg/(L day) which was 3.94 times higher than the value obtained in Nþ
Lipid productivity
experiment (69.46 mg/(L day)). Carbon fixation rates under Nþ and N conditions were
Carbon dioxide fixation
286.12 and 273.66 mg/(L day) respectively. © 2017 Published by Elsevier B.V. on behalf of Institution of Chemical Engineers. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/ by-nc-nd/4.0/).
1.
Introduction
Lipid productivity (LP) is a critical variable for evaluating algal species for biodiesel production. It is calculated as the product of biomass productivity (BP) (g dry weight/L day) and lipid content (% dw) to give an indicator of oil produced on a basis of both volume and time (Griffiths and Harrison, 2009). Key environmental variables affecting LP of microalgae under phototrophic conditions include: light intensity (LI), temperature, pH, photoperiod (ratio of light to dark cycle), CO2 concentration in air and aeration rate. Several reports are available in literature which illustrate the effect of these variables on LP under nitrogen replete (Nþ) conditions (Lee and Lee, 2001; Pal et al., 2011; Xin et al., 2011; Cabello et al., 2015;
Razzak et al., 2015). But very few authors have reported the influence under nitrogen deplete (N) conditions (Fernandez et al., 2012; Liu et al., 2012; Bruer et al., 2013; Toledo et al., 2013; Fakhry and Maghraby, 2015). Moreover, the number of variables in these studies was restricted to a maximum of 3. However, LP of microalgae is significantly affected by all aforementioned environmental variables. Especially, under nitrogen deprivation, microalgal cells are more sensitive to environmental conditions. They modify their biosynthetic pathways to accumulate lipids during this phase. Thus, knowledge of the influence of environmental conditions on LP under N conditions is very useful. It leads to development of useful kinetic expressions that could be applied for designing and modelling
* Corresponding author. Fax: þ91 816 2214070. E-mail addresses: [email protected] (P. Binnal), [email protected] (P.N. Babu). http://dx.doi.org/10.1016/j.sajce.2017.01.001 1026-9185/© 2017 Published by Elsevier B.V. on behalf of Institution of Chemical Engineers. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
south african journal of chemical engineering 23 (2017) 26e37
photobioreactors to predict LP and optimize the operating conditions (Da Silva et al., 2006). In this regard, the present study aims at the statistical optimization of aforementioned environmental parameters influencing the LP and RC of microalga Chlorella protothecoides in a lab scale externally illuminated photobioreactor (PBR) under N conditions.
1.1. Design of experiments by response surface methodology Response surface methodology (RSM) is a statistical technique which allows the simultaneous consideration of many variables at different levels and the interactions between those variables, using a smaller number of observations than conventional procedures. It is very useful for developing, improving, and optimizing processes in which a response of interest is influenced by several variables and the objective is to optimize this response. RSM has important application in the design, development and formulation of new products, as well as in the improvement of existing product design. It defines the effect of the independent variables, alone or in combination, on the processes. In addition to analysing the effects of the independent variables, this experimental methodology generates a mathematical model which describes the chemical or biochemical processes (Anjum et al., 1997). Several authors have described the use of RSM as an optimization tool (Singh et al., 2015; Amosa, 2016; Amosa and Majozi, 2016; Chiranjeevi and Mohan, 2016; Jami et al., 2016). Among RSM techniques, Box-Behnken design is considered as an efficient option and an ideal alternative to central composite designs. It combines a fractional factorial with incomplete block designs to avoid the extreme vertices and presents an approximately rotatable design with only three levels per factor. The number of experiments (N) required for the development of BBD is defined as N ¼ 2k(k1) þC0, (where k is number of factors and C0 is the number of central points). For, k ¼ 6, Co ¼ 6, N is 56 (Deming and Morgan, 1993). Considering all the linear terms, square terms and linear linear interaction items, the quadratic response model can be described as: Y ¼ bo þ
i¼k X i¼1
bi xi þ
k X i¼1
bii x2i þ
k X k X ii > j
bij xi xj þ ε
j
where Y is the predicted response variable, xi is the independent variable, bo , bi , bii , bij are the regression coefficients
27
and ε is the random error. Other advantages of BBD are: i) It permits estimation of the parameters of the quadratic model (ii) building of sequential designs (iii) detection of lack of fit of the model and (iv) use of blocks (Ferreira et al., 2007).
2.
Materials and methods
2.1.
Construction of photobioreactor
The pictorial, schematic and PID diagrams of Photobioreactor (PBR) used in the present study are shown in Figs. 1e3 respectively. The set up consists of a 5 L borosilicate vessel connected to a panel of straight glass tubes through a diaphragm pump. The panel of straight glass tubes consists of 6 glass tubes (1 inch diameter, 80 cm long) connected to each other through bends, flexible couplings and O-rings. The media from 5 L vessel is circulated through straight glass tubes and returned to the vessel by diaphragm pump. The vessel is equipped with pH, temperature and light intensity controllers. pH electrode (Mettler Tolerdo) and Pt-100 sensors have been used to measure pH and temperature of media in reactor. LED panels are arranged around 5 L vessel and straight glass tubes. The light intensity (0e10 klux) was adjusted from control panel. Two rotameters are provided for controlling flow rate of Carbon dioxide and air (Range of rotameter for air: 1 LPMe5 LPM, range of rotameter for CO2: 1 mLPM to 50 mLPM). The mixed CO2 enriched air was bubbled through media in 5 L vessel using ring sparger.
2.2.
Algae strain collection and culture condition
The original strain of C. protothecoides (SAG 211-10C) was obtained from Sammulung von Algenkulturen (SAG), Germany, and maintained on agar slants containing BG11 medium consisting of (g/L): NaNO3-1.5 g, K2HPO4$3H2O-0.04, KH2PO4$3H2O0.2, EDTA-0.0005, Fe ammonium citrate-0.005, citric acid-0.005, Na2CO3-0.02 and 1 ml of trace metal solution. The trace metal solution contains (g/L): H3BO3-2.85, MnCl2$4H2O-1.8, ZnSO4$7H2O-0.02, CuSO4$5H2O-0.08, CoCl2$6H2O-0.08 and Na2MoO4$2H2O-0.05. The pH of the medium was adjusted to 6.8. Stock cultures were inoculated into 100 ml of sterilized medium in 250 ml Erlenmayer flasks. The flasks were then incubated at 24 C in a rotary shaker and agitated at 120 rpm. After seven days, the algal biomass was recovered by centrifugation (REMI c-24 bl) at 10,000 rpm for 10 min and used as inoculam to PBR containing 3 L BG11 medium with nitrogen (for Nþ experiments) or BG11 medium without nitrogen (for N experiments)and the
Fig. 1 e Pictorial view of photobioreactor.
28
south african journal of chemical engineering 23 (2017) 26e37
Fig. 2 e Schematic view of photobioreactor.
initial biomass concentration was adjusted to 0.2 g/L. For Nþ experiment, operating conditions were set as: Temperature27 C, Light intensity-6 klux, photoperiod-10 h:12 h, pH-7.2, aeration rate-3 lpm. These conditions were optimized in our previous study (Babu and Binnal, 2015). For Nþ experiment, 1% CO2 air was continuously bubbled through medium.
2.3. Design of nitrogen depletion experiments in photobioreactor The ranges of parameters selected in the present study are depicted in Table 1. After inoculation, operating conditions of PBR were adjusted at specified values in a given experiment (Table 2). Response Surface Methodology (RSM) was used to design these experiments. A three level Box-Behnken design was used to study the effect of six variables (LI, temperature, pH of cultivation medium, photoperiod, CO2 concentration in air and aeration rate) influencing LP and RC of C. protothecoides under N conditions. A total of 54 experiments were conducted. During each run, samples were collected from PBR at
regular intervals of 24 h and analysed for biomass concentration, lipid content and carbon content. BP, LP and RC were evaluated as explained in section 2.4. Highest values of LP and RC obtained in each run were noted. Design expert software (Version 8, Stat Ease, USA) was used to analyse the data and generate response surfaces and contour plots. The accuracy of quadratic model obtained through regression analysis was tested by the coefficient of determination (R2). The response surface plot analysis was made by keeping four independent variables constant (at zero levels), while changing the other two independent variables.
2.4.
Analytical methods
2.4.1.
Estimation of biomass
The conical flask containing culture was shaken thoroughly. 1.5 ml of culture was pipetted out and transferred to previously weighed 2 ml microcentrifuge tubes (Make: TARSONS CILLOK). This was done in duplicate. The tubes were centrifuged in mini centrifuge (Model: GeNei™, SLM-CFT-10K) for
Fig. 3 e Process instrumentation diagram of photobioreactor.
south african journal of chemical engineering 23 (2017) 26e37
Table 1 e Experimental range and levels of the variables. Variable
Symbol
Temperature ( C) Light intensity (klux) pH Photoperiod (h:h) CO2 concentration in air (%, v/v) Aeration rate (lpm)
X1 X2 X3 X4 X5 X6
Coded factors and levels 1
0
1
20 1 3 8:24 1 1
30 5.5 6.5 16:8 8 3
40 10 10 24:0 15 5
29
10 min at 10,000 rpm. The supernatant in tubes was discarded and the tubes were dried in an oven for 8 h. The weight of dried tubes was noted down. W1 ¼ Weight of microcentrifuge tubes (g) W2 ¼ Weight of microcentrifuge tubes þ dried biomass (g) W ¼ W2W1 ¼ Weight of dry biomass per 1.5 ml culture medium (g)
Concentration of biomass; Xðg=LÞ ¼ W 103 1:5
(1)
Table 2 e Design of experiments along with responses.
2.4.2.
Run
X1
X2
X3
X4
X5
X6
LP
RC
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
20 40 20 40 20 40 20 40 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 20 40 20 40 20 40 20 40 30 30 30 30 30 30 30 30 20 40 20 40 20 40 20 40 30 30 30 30 30 30
1 1 10 10 1 1 10 10 1 10 1 10 1 10 1 10 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 1 10 1 10 1 10 1 10 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5
6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 3 3 10 10 3 3 10 10 3 10 3 10 3 10 3 10 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 3 3 10 10 3 3 10 10 6.5 6.5 6.5 6.5 6.5 6.5
8 8 8 8 24 24 24 24 16 16 16 16 16 16 16 16 8 8 24 24 8 8 24 24 8 8 24 24 8 8 24 24 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16
8 8 8 8 8 8 8 8 1 1 1 1 15 15 15 15 8 8 8 8 8 8 8 8 1 1 1 1 15 15 15 15 1 1 15 15 1 1 15 15 8 8 8 8 8 8 8 8 8 8 8 8 8 8
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 1 1 1 1 5 5 5 5 3 3 3 3 3 3 3 3 1 1 1 1 5 5 5 5 1 1 1 1 5 5 5 5 3 3 3 3 3 3
30.17 4.42 66.11 21.15 54.19 7.96 118.97 38.07 40.82 78.15 51.79 92.44 24.00 45.97 30.44 53.59 57.28 97.82 83.77 104.45 103.08 176.08 150.78 188.16 100 41.16 126.04 66.26 55.74 22.87 70.14 30.96 90.49 117.31 17.45 30.79 162.87 163.8 24.43 43.1 51.12 23.45 66.79 41.15 74.47 46.79 95.49 61.15 266.92 266.92 266.92 266.92 266.92 266.92
107.84 20.14 153.93 61.79 148.42 27.81 196.50 61.79 18.19 52.22 7.49 54.08 20.85 50.79 7.16 18.71 51.14 10.77 85.44 27.08 95.16 26.11 140.38 61.82 74.77 21.74 109.17 26.48 14.12 12.21 16.28 17.74 84.59 107.15 17.82 28.82 104.48 121.88 21.12 29.75 30.43 11.15 21.78 5.11 52.18 31.08 51.48 33.15 268.49 268.49 268.49 268.49 268.49 268.49
Estimation of total lipids was done following the protocol of Bligh and Dyer (1959). To a 15 ml vial containing algal biomass (300 mg), 2 ml methanol, and 4 ml chloroform were added. The mixture was agitated in a cyclomixer (Model: CM 101, REMI) for 2 min. 1 ml of chloroform was again added and the mixture was shaken vigorously for 1 min. 1.8 ml of distilled water was added and the mixture was mixed in a vortex again for 2 min. The two layers were separated by centrifugation for 15 min at 10,000 rpm. The lower layer was filtered through Whatman No. 1 filter paper into a previously weighed clean vial (W1). The process was repeated three times. All the chloroform phases were collected together, and evaporated in a Rotary vacuum evaporator (Superfit™, Rotavap, Model:PBV-7D). The final weight of vial with lipids was measured (W2). The lipids obtained were calculated by subtracting W1 from W2, and expressed as % dcw.
2.4.3.
Estimation of total lipids
Estimation of lipid productivity
Biomass productivity and lipid productivity were calculated as, Biomass productivity½mg dry biomass=ðL dayÞ ¼ ðXn XoÞ=n (2) LP mg lipid L1 day1 ¼ BP Lipid contentðwt%Þ 100
(3)
where Xo and Xn are biomass concentrations (in g/L) on zeroth and nth day respectively.
2.4.4.
Estimation of carbon dioxide fixation
The carbon content of algae biomass was determined using Perkin Elmer 2400 CHNS/O elemental analyzer. The carbon fixation rate was calculated as (Tang et al., 2011), RC ¼ CC BP MCO2 =MC
(4)
where CC is weight % of carbon in algae biomass, MCO2 and MC are molecular weights of carbon dioxide and carbon respectively.
3.
Results and discussion
3.1.
Statistical analysis of parameters
Table 2 depicts the highest values LP and RC obtained in each run. It was observed that, the time duration to attain these values varied for each run. This is obvious as the cultivation conditions were different for each run (except for Runs 49e54). The experimental data was analysed by multiple regressions using Design Expert software. The final estimative models for LP and RC in terms of uncoded factors were as follows.
30
south african journal of chemical engineering 23 (2017) 26e37
3.2. Influence of environmental factors on LP under nitrogen deplete conditions
RC ¼ 1251:49 þ 46:11X1 þ 25:64X2 þ 98:85X3 þ 25:73X4 þ 30:13X5 þ 97:35X6 0:813X1 2 2:03X2 2 7:89X3 2 0:76X4 2 2:07X5 2 15:22X6 2
(5)
LP ¼ 1330:65 þ 58:8X1 þ 48:63X2 þ 76:01X3 þ 20:81X4 þ 16:31X5 þ 69:63X6 1:015X1 2 4:14X2 2 5:59X3 2 0:6X4 2 1:27X5 2 9:84X6 2
(6)
where X1, to X6 are the actual values of independent variables. Tables 3 and 4 describe the ANOVA results for the second order response surface models. Fisher's variance ratio, which is defined as the ratio of mean lack of fit to mean square of pure experimental error, were high for both models (F ¼ 64.91 and 44.1 for LP and RC respectively) with a low probability (p < 0.0001), which demonstrates the high significance of the models. The significance of each term was determined by their respective p-values. The smaller the p-value, the more significant is the corresponding coefficient. In both models, all the linear terms and quadratic terms were statistically significant. All mutual interactions terms were found to be insignificant in both models (p > 0.05). Adequate precision measures the signal to noise ratio. A ratio greater than 4 is desirable. The values obtained in present work (28.51 mg/(L day) and 24.17 mg/(L day) for LP and RC respectively) indicated adequate signals. R2 (Coefficient of regression) is the proportion of variation in the depended variable explained by regression model. Adjusted R2 is coefficient of determination adjusted for the number of independent variables in the regression model (Ferreira et al., 2007). High values of R2 obtained in the present work for both models (0.97 and 0.94) indicated that the they would accurately predict the relationships between the parameters. Further, the “Predicted R2” values were in reasonable agreement with the “Adjusted R2” which confirmed that models could be used to navigate the design space.
3.2.1.
Mutual interactions of temperature and light intensity
Basically, to achieve highest LP, both lipid content and biomass concentration of microalgae should be increased. However, cultivation conditions that increase lipid content (high LI, high temperature) can adversely reduce biomass. In the same way, conditions that favour high biomass concentration will negatively affect lipid content. Hence, an optimum LI and temperature should be selected to maximize LP. The effect of irradiance and temperature on biomass and lipid content has been explored by various researchers. There has been a consensus with reference to effect of LI on biomass that, at low LI, photosynthesis rates are low. Increasing intensity improves BP and RC till saturation light intensity. Above saturation LI, biomass growth rate is hampered due to photoinhibition. However, there has been no clear consensus about effect of LI on lipid content. Gordillo et al. (1998) reported that lipid content of Dunaliella viridis kept decreasing with increase in LI from 35 to 700 mmol/(m2 s), while Liu et al. (2012) reported that lipid content of Scenedesmus sp. 11-1 increased rapidly with increasing irradiance and a highest lipid content of 41.12% was obtained at 250 mmol/(m2 s). Basically, when exposed to high light (HL), lipids serve as a sink of excessive light energy absorbed by photosynthetic apparatus in microalgae. Triacylglycerol (TAG), the main component in neutral lipids synthesis requires large amounts of ATP and NAD(P)H (which are produced by the photosynthesis). Therefore, lipid accumulation is helpful in the dissipation of excess light energy and prevention of the photochemical damage of algal cells. The excess carbon flux generated from the photosynthesis is channelled to the lipid accumulation on a unit biomass basis when individual cell is exposed to a large quantity of light energy (Courchesne et al., 2009). Therefore, the HL could promote the final lipid content (especially neutral lipid content). However, under ‘HL coupled with N conditions’, chloroplast lamellae are disrupted more
Table 3 e Analysis of variance (ANOVA) for response surface quadratic model (LP). Source Model X1 X2 X3 X4 X5 X6 X21 X22 X23 X24 X25 X26 Residual Lack of fit Pure error Total CV ¼ 21.08% R2 ¼ 0.97 Adjusted R2 ¼ 0.93 Predicted R2 ¼ 0.911 *Significant at p 0.05.
Sum of squares
df
Mean square
F value
p-value
2.902Eþ005 10573.7 4550.29 3259.44 2902.67 19359.9 10767.32 1.06Eþ005 72402.57 48252.84 15552.88 40019.4 15938.25 15272.52 15272.52 0.000 3.054Eþ005 Mean ¼ 91.51 Adequate precision ¼ 28.51
12 1 1 1 1 1 1 1 1 1 1 1 1 41 36 5 53
24180.56 10573.7 4550.29 3259.44 2902.67 19359.9 10767.32 1.06Eþ005 72402.57 48252.84 15552.88 40019.4 15938.25 372.5 424.24 0.00
64.91 28.39 12.22 8.75 7.79 51.98 28.91 284.5 194.37 129.54 41.75 107.43 42.79
<0.001 <0.001 0.0012* 0.0051* 0.0079* <0.0001 <0.0001 <0.0001 <0.001 <0.0001 <0.0001 <0.0001 <0.0001
Significant
31
south african journal of chemical engineering 23 (2017) 26e37
Table 4 e Analysis of variance (ANOVA) for response surface quadratic model (RC). Source Model X1 X2 X3 X4 X5 X6 X21 X22 X23 X24 X25 X26 Residual Lack of fit Pure error Total CV ¼ 31.95% R2 ¼ 0.94 Adj.R2 ¼ 0.90 Predicted R2 ¼ 0.8722
Sum of squares
df
Mean square
F value
p-value
3.247E005 17426.05 5148.01 4115.41 3091.23 11566.19 3439.1 68067.33 17463.93 96137.67 24351.36 1.068Eþ005 38164.1 24851.4 24851.4 0.000 3.456 þ 005 Mean ¼ 77.06 Adequate precision ¼ 24.173
12 1 1 1 1 1 1 1 1 1 1 1 1 41 36 5 53
26731.62 17426.05 5148.01 4115.41 3091.23 11566.19 3439.1 68067.33 17463.93 96137.67 24351.36 1.068Eþ005 38164.1 606.13 690.32 0.000
44.1 28.75 8.49 6.79 4.98 19.08 5.67 112.3 28.81 158.61 40.18 176.27 62.96
<0.001 <0.001 0.0058 0.0127 0.0312 <0.0001 0.0219 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
Significant
*Significant at p 0.05.
severely than Nþ conditions which reduce photosynthesis efficiency and carbon fixation efficiency. Thus, it can be concluded that effect of LI on lipid content of microalgae depends upon the ability of the individual species to capture carbon at high light. Fig. 4A and B depict the combined effect of LI and temperature on LP and RC respectively. At low temperature and low LI (20 C and 1 klux), LP and RC were 30.1 mg/(L day) and 107.8367 mg/(L day) respectively (Run 1). When LI was increased above 1 klux, both LP and RC increased at all temperatures till an optimum LI of 5.5 klux. Above 5.5 klux, both responses declined attaining a minimum value at 10 klux. However, the responses at 10 klux were higher than those obtained at 1 klux (Run 3: LP ¼ 66.10667 mg/(L day), RC ¼ 153.9267 mg/(L day)). At 20 C and 1 klux, BP and lipid content were 0.19 g/(L day) and 15.72%, while at 20 C and 10 klux, the corresponding values were 0.22 g/(L day) and 30.04%. Thus, lipid content increased with increasing LI. Since no significant changes in BP were observed at 1 klux and 10 klux, it was concluded that higher values of LP and RC obtained at 10 klux were due to high lipid content (Equation (1)). Also, lipid contents in Runs 2, 5, 6 were greater than for runs 4,7 and 8, compared in the same order (Run 2 v/s 4: 4.7%/17.45%, Run 5 v/s 7: 22.49%/39.39%, Run 6 v/s 8: 6.8%/25.1%), while BPs were (Run 2 v/s 4: 0.094/0.123, Run 5 v/s 7: 0.24/0.301, Run 6 v/s 8: 0.117/0.152). Thus, HL increased the lipid content of cells due to light stress. Therefore, LP and RC at 10 klux (Run 4, 7, 8) were greater than the corresponding values at 1 klux (Run 2,5,6). Similar to irradiance, maintaining an optimal temperature in PBR was required to promote high LP. If we analyse the effect of temperature on biomass, it has been suggested in literature that changes in cytoplasmic viscosity under suboptimal temperature conditions is responsible for less efficient carbon and nitrogen utilization. Hence, increasing temperature beyond the optimum reduces protein synthesis and consequently results in decreased growth rates. On the other hand, the effect of temperature on lipid content of microalgae is species specific and relies upon concentration of ROS (reactive oxygen species). ROS, if present in adequate
concentrations (which can occur either at low or high temperature) create stress on microalgal cells thus improving their lipid content (Chokshi et al., 2015). In present work, analysis of Runs (1e8, 25e32) indicated that both lipid content and BP at 20 C were higher than at 40 C. This can be attributed to two factors. At lower temperature, lipid % was higher due to generation of ROS which positively affected LP. The reduced BP at 40 C could be due to the fact that, at high temperature, CO2 solubility in water decreases, which further reduces RC. Also, at high temperature, photosystem (II) gets severely damaged which reduces rate of photosynthesis.
3.2.2. Mutual interactions of temperature and CO2 concentration Theoretically, increasing CO2 concentration in air should result in increased biomass concentration and lipid content of microalgae. However, extremely high concentration of CO2 will cause substrate inhibition and lead to decreased RC and LP. Hence, air containing an optimum concentration of CO2 must be supplied to algae culture to obtain its best performance. In the present work, low LP and RC were recorded at 1% CO2 (Fig. 4C and D). As depicted in these two figures, both LP and RC increased with increasing CO2 concentration until an optimum CO2 concentration of 8% at all temperatures. Above this concentration, both the responses decreased due to substrate inhibition. This is due to the fact that, increasing CO2 concentration in air results in increased rate of photosynthesis, which results in increased concentration of oxygen concentration in medium. Oxygen competes with CO2 on carboxylating CO2 enzyme (RuBISCO) thus reducing rate of carbon fixation. This leads to reduced BP. However, under nitrogen deprivation, the absorbed carbon, instead of being incorporated into proteins and assist cells to grow, is diverted to lipid synthesis path way that would be taken as a carbon sink. Hence, above 8% CO2, although lipid content improved, a decline in LP was observed due to decreased growth rate. The negative effect of increasing CO2 concentration above optimum was more severe at higher temperature than low temperatures (Run 29e32).
32
south african journal of chemical engineering 23 (2017) 26e37
Fig. 4 e Response surface plots: A, B- Interactions of LI and temperature; C, D- Interactions of temperature and CO2 concentration in air; E, F- Interactions of CO2 concentration in air and pH.
3.2.3.
Mutual interactions of pH and CO2 concentration
Fig. 4E and F depict the effect of pH on LP and RC respectively. pH of the culture medium plays an important role in influencing the lipid accumulation in the algae. Activation of
ACCase (key enzyme in lipid biosynthesis) is pH dependent. In present study, a comparison of LP and RC values for runs: 9 v/s 11, 10 v/s 12,14 v/s 16,17 v/s 18,19 v/s 20 and 21 v/s 22 (Runs 9, 10, 13, 14, 17, 19 were at pH 3 and Runs 11, 12, 15,16, 18, 20 were
south african journal of chemical engineering 23 (2017) 26e37
at pH 10) clearly indicated that, in all these cases, LP and RC were higher at pH 10 than at pH 3. At low pH of 3, CO2 inhibits the activity of enzyme carbonic anhydrase (CA), which plays an important role in carbon fixation. Thus, both BP and lipid content were low leading to lower values of LP. With increasing pH, activity of RUBISCO increases thus improving carbon fixation which further improves LP. Thus, LP and RC increased till an optimum pH of 6.9. However, at alkaline pH, although a rise in lipid content was noticed, BP reduced owing to non-optimum pH conditions thus reducing LP. For example, lipid content for Run 21 and 22 were 32.1% and 71.2, while BP were 0.32 and 0.25 g/(L day) respectively. It has been known that, environmental stress along with nitrogen deprivation leads to higher lipid content. Thus, there was not much difference in BP values for pH 3 and 10. Hence, the higher value of LP at pH 10 was due to higher lipid content. It is well known that, CO2 in medium is present in culture medium in three forms (CO2 3 , HCO3 and dissolved CO2). The concentrations of these ions are pH dependent. At alkaline pH, HCO 3 dominates. Thus, C. protothecoides cell was able maintain its growth at alkaline conditions by consuming HCO 3.
3.2.4.
Mutual interactions of photoperiod and light intensity
It can be observed from Table 1 that, at 20 C and LI of 1 klux, when photoperiod was increased from 8:12 to 24:0, LP increased from 30.1 mg/(L day) to 54.19 mg/(L day) and RC increased from 107.83 mg/(L day) to 148.41 mg/(L day). Similarly, an analysis of responses from Run 17 to 32 clearly indicated that increasing light period from 8 h to 24 h improved LP and RC. In the present work, at low LI and low light periods, both BP and lipid content were low due to photolimtation condition and insufficient time duration of light period. Increasing photoperiod at low LI improved LP and RC. However, at higher LI, an increase in photoperiod initially improves LP till an optimum. Further increase in light period leads to decrease in LP (Fig. 5A). High light intensities cause damage to photosynthesis apparatus which can be counterbalanced by exposing the cells to darkness (Pulz, 2001). During light period, light energy absorbed by microalgal cells is converted to chemical energy providing NADPH2 and ATP. During dark period, the excess light energy absorbed in light period is used for endergonic reactions. Maximum photosynthetic efficiency are achieved when light/dark cycle period approaches photosynthetic turnover time (Wahidin et al., 2013). Fig. 5AeB indicates that 16:8 is the optimum photoperiod. This result indicates that cell division was not under circadian control but light responsive (diurnal). The diurnal pattern of cell growth in the light and cell division in the dark is thought to allow for maximum collection of light energy.
3.2.5. Mutual interactions of photoperiod and CO2 concentration Fig. 5CeD depict the interactions among CO2 concentration and photoperiod. Increasing light period for low to optimum CO2 concentration (8% v/v) improves carbon fixation. This is due to the fact that microalgae can absorb CO2 only in the presence of light and an increase in duration of light naturally leads to increased RC. However, above optimum CO2 concentration, due to inhibition of CA enzyme, RC decreases.
3.2.6. Mutual interactions of aeration rate and CO2 concentration Fig. 5EeF illustrate the effect of aeration rate and CO2 concentration on LP and RC. It is evident that, irrespective of CO2
33
concentration, LP and RC increased with the aeration rate. Rate of aeration and CO2 concentration are very important parameters influencing the performance of PBR as they together deicide the amount of available carbon in PBR. The rate of carbon supply should match with rate of its consumption by cells. In general, at low aeration rate (1 lpm) and low CO2 concentration (1%), photosynthetic rates were low due to inadequate mixing of broth and insufficient amount of inorganic carbon. With increase in flow rate from 1 lpm to 4 lpm, both LP and RC improved for all CO2 concentrations. This can be attributed to better mixing at high flow rate which facilitated removal of dissolved oxygen from medium, thus improving rate of photosynthesis. However, an increase of aeration rate from 4 lpm to 5 lpm slightly reduced LP and RC. This may be a result of high shear on microalgal cells that might have reduced biomass concentration. Also, higher flow rate decreases residence time of CO2 in the reactor. Further, it was observed that, at all aeration rates, above 8% CO2, a reduction in LP and RC was observed. As mentioned earlier, high photosynthetic efficiency is achieved when rate of carbon supply is equal to consumption rate by microalgal cells. Thus, aeration rate of 4 lpm with 8% CO2 was found to be adequate.
3.3. Optimization of environmental parameters influencing LP and RC under N conditions using desirability function Desirability function is a popular and established technique for the simultaneous determination of optimum settings of input variables that can determine optimum performance levels for one or more responses. It ranges from zero outside the limits, to one at the goal. The goal fields for response have five options: none, maximum, minimum, target and within range. In the present work, ‘Numerical optimization’ of the software has been chosen in order to find the specific point that maximizes the desirability function. The program seeks to maximize this function. The goal seeking begins at a random starting point and proceeds up the steepest slope to a maximum. There may be two or more maxima because of curvatures in the response surfaces and their combination in the desirability function. Starting the search of optimum from several points in the design space improves the chances of finding the best local maximum (Amini et al., 2008). The individual desirabilities are then the combined to give the overall desirability (D). The main objective of preset study was maximize LP and RC while following a cost driven approach. Hence, the goal ‘maximize’ was selected for LP and RC with highest importance. In order to reduce cost of process, the goal ‘minimize’ was selected for LI, photoperiod, aeration rate and Carbon dioxide supply. For temperature and pH, the option ‘in the range’ was selected. The number of cycles was set as 50 and the duplicate solution filter selected was epsilon (minimum difference) for eliminating duplicate solutions. The results of optimization are shown in Table 5. Seven solutions were suggested by software. Highest desirability (0.76) was observed for solution No.1. However, LP and RC for this solution were least for this solution. Highest LP and RC were predicted for solutions 5 and 6 respectively. However, desirability functions for these solutions were lower than those reported for solutions 1e4. A comparison of Solution 2 with 5 and 6 revealed that, LP can be increased by 2.53% and 2.2% by: i) increasing LI by 12.42% and
34
south african journal of chemical engineering 23 (2017) 26e37
Fig. 5 e Response surface plots: A, B- Interactions of photoperiod of LI; C, D- Interactions of photoperiod and CO2 concentration in air; E, F- Interactions of CO2 concentration in air and aeration rate.
35
south african journal of chemical engineering 23 (2017) 26e37
Table 5 e Suggested solutions by ‘Numerical optimization’ technique. Solution No. 1 2# 3 4 5 6 7
Temperature ( C)
LI (klux)
pH
Photoperiod (h:h)
Volume % of CO2 in air
Aeration rate (lpm)
LP (mg/(L day))
RC (mg/(L day))
Desirability
28.79 28.63 28.68 28.62 28.87 28.67 28.61
4.95 5.31 5.65 5.47 5.77 5.97 5.84
6.58 6.51 6.48 6.48 6.67 6.47 6.42
14.32 15.36 16.11 15.69 16.61 16.89 16.5
5.59 6.26 6.59 6.41 6.44 6.9 6.85
2.73 2.92 3.12 3.03 3.33 3.29 3.18
260 268.37 272.3 270.78 275.17 274.26 273.27
256 267.28 274 269.94 271.11 273.9 273.75
0.76 0.722 0.675 0.63 0.61 0.54 0.428
# Selected optimum solution.
8.66% and ii) by increasing aeration rate by 14% and 12.67% respectively. Since the main cost components of microalga cultivation are LI and aeration rates, solutions 5 and 6 were rejected as they are not economically attractive. Solution 2 was accepted as global optimum as it suggested reasonably low LI (5.31 klux), low photoperiod (15.36:8.64), low aeration rate (2.92 lpm) and lower concentration of CO2 in air (6.26%) while predicting relatively high LP and RC (268.37 mg/(L day) and 267.28 mg/(L day)). Fig. 6 shows the ramp desirability of solution 2 that was generated via numerical optimization. The individual desirability functions (di) for each of the responses, and the calculated geometric mean as maximum over all desirability (D ¼ 0.722) is represented in Fig. 7. For validation of this Solution 2, a confirmatory experiment was conducted at optimum results. The values of LP and RC obtained were found to be 274.15 mg/(L day) and 273.66 mg/ (L day) respectively (% error from model predictions were 2.15% and 2.4% for LP and RC respectively).
3.4. Comparison of LP and RC under Nþ and N conditions Comparison of LP and RC obtained under Nþ and N conditions when cultivated under their respective optimum conditions are depicted in Table 6. It is evident that, LP under N conditions was 3.94 times greater than the value obtained under Nþ conditions. Despite nitrogen starvation, C. protothecoides was able to maintain the growth rate till 7th day, after which a stationary phase begun. However, biomass concentration in N was slightly lower than in N cultures (2.85 g/L v/s 2.81 g/L). The reason for sustaining growth rate under N conditions could be due to the fact that, an intracellular nitrogen pool such as that in chlorophyll could be used as nitrogen source to maintain the growth rate. Similar results were obtained by Ho et al. (2014), who reported that, biomass concentration of Scenedesmus obliquus CNW-N continued to rise after depletion of
Fig. 6 e Desirability ramp of optimum solution.
36
south african journal of chemical engineering 23 (2017) 26e37
Fig. 7 e Bar graph representing individual desirability of all responses (di) in correspondence with desirability (D).
Table 6 e Comparison of LP and RC obtained under Nþ conditions with N¡ conditions. Time (Day)
Nþ conditions
N conditions
X (g/L) Lipid content LP (mg/(L day)) RC (mg/(L day)) X (g/L) Lipid content (%) LP mg/(L day) RC mg/(L day) 0 1 2 3 4 5 6 7 8 9
0.2 0.32 0.61 1.42 1.82 2.17 2.49 2.85 2.12 2.05
5.89 7.55 10.54 12.55 14.46 16.22 18.12 22.48 21.48 22.29
e 9.1 21.61 50.95 58.52 63.9 69.46* 67.45 51.55 35.83
e 128.85 161.12 186.45 222.23 261.17 266.12 271.15 277.15 286.12*
0.2 0.33 0.54 1.31 1.74 2.12 2.33 2.81 2.44 1.95
5.89 9.2 17.44 27.53 40.18 49.62 69.47 73.48 70.98 71.1
e 11.96 29.64 31.2 154.93 190.54 250.87 274.15* 198.74 138.25
e 124.41 157.52 184.17 209.81 254.32 261.16 273.66* 261.94 254.57
* Highest values obtained.
nitrogen in the medium. However, lipid content increased much rapidly in N cultures than Nþ cultures (5.89%e73.8% in 7 days for N cultures as compared to 5.89% to 22.48% in 6 days for Nþ cultures). These factors resulted in higher LP in N cultures than in Nþ cultures.
4.
Conclusions
A lab scale externally photobioreactor was used in the present work to evaluate lipid productivity of microalga C. protothecoides under nitrogen deprivation.54 experiments were conducted in accordance with Box-Behnken design in order to study and optimize the effect of six environmental factors on lipid productivity and rate of carbon dioxide fixation. Among 54 trails, 6 trails were replicates (at centre points). Under these conditions, local optimum values were observed (LP ¼ 266.92 mg/(L day) and RC ¼ 268.4867 mg/(L day)). However, to obtain global optimum, numerical optimization technique of Design Expert software was used. A cost driven approach which minimizes costs of light intensity, aeration rate, photoperiod, rate of CO2 supply and maximizes LP was adopted. Seven solution were suggested by software. The solution which provides a cost effective process while producing high LP was selected. The global optimum values were: pH6.51, Temperature-28.63 C, light intensity-5.31 klux, Photoperiod-15.36 h:8.64 h, CO2 concentration in air-6.26% (v/v) and Aeration rate 2.92 lpm. Under these conditions, highest LP of 274.15 mg/(L day) and RC of 273.66 mg/(L day) were obtained.
Thus, C. protothecoides can be a potential feedstock for biodiesel production. The regression models for both responses had high R2 (0.97 and 0.94 for LP and RC respectively), high Fischer variance ratio (64.91 and 44.1 for LP and RC respectively), and adequate precision (28.51 and 24.173 for LP and RC respectively). Further, Adjusted R2 for both regression models were close to predicted R2 (Tables 3 and 4). Thus, models could be used to predict the experimental results with high accuracy. In order to obtain high volumetric productivity of microalgal system in large photobioreactors, precise optimization of the environmental variables is crucial. In the present study, optimization of variables was done in small lab scale reactor of 5 L capacity. The approach can be extended to large scale photobioreactors to obtain their highest performance.
Acknowledgement The authors thank financial support from Visveswaraya Technological University, Belagavi, Karnataka, India, under VTU research Grants Scheme-2010-11 (VTU/Aca./2011-12/A-9/ 6380).
Nomenclature LP (mg/L day) Lipid Productivity RC (mg/L day) Carbon fixation rate BP (g/(L day) Biomass productivity
south african journal of chemical engineering 23 (2017) 26e37
LI (klux) HL PBR X (g/L) CC (wt %) MC MCO2
Light intensity High light Photobioreactor Biomass concentration Carbon content of biomass Molecular mass of Carbon Molecular weight of carbon dioxide
References Amini, M., Younesi, H., Bahramifar, N., Lorestani, A.A.Z., Ghorbani, F., Danesh, A., Sharifzadeh, M., 2008. Application of response surface methodology for optimization of lead biosorption in an aqueous solution by Aspergillus niger. J. Hazard. Mater 154, 694e702. Amosa, M.K., 2016. Sorption of water alkalinity and hardness from high-strength wastewater on bifunctional activated carbon: process optimization, kinetics and equilibrium studies. Environ. Technol. 37 (16), 2016e2039. Amosa, M.K., Majozi, T., 2016. GAMS supported optimization and predictability study of a multi-objective adsorption process with conflicting regions of optimal operating conditions. Comput. Chem. Eng. 94, 354e361. Anjum, M.F., Tasadduq, I., Al-Sultan, K., 1997. Response surface methodology: a neural network approach. Eur. J. Oper. Res. 101 (1), 65e73. Babu, P.N., Binnal, P., 2015. Lipid productivity of microalgae Chlorella vulgaris and Nannochloropsis oculata in externally illuminated lab scale photobioreactor. Int. J. ChemTech Res. 7 (5), 2217e2221. Bligh, E.G., Dyer, W.J., 1959. A rapid method of total lipid extraction and purification. Can. J. Biochem. Phys. 37 (8), 911e917. Breuer, G., Lamers, P.P., Martens, D.E., Draaisma, R.B., Wijffels, R.H., 2013. Effect of light intensity, pH, and temperature on triacylglycerol (TAG) accumulation induced by nitrogen starvation in Scenedesmus obliquus. Bioresour. Technol. 143, 1e9. nchez, L., Revah, S., Cabello, J., Toledo-Cervantes, A., Sa Morales, M., 2015. Effect of the temperature, pH and irradiance on the photosynthetic activity by Scenedesmus obtusiusculus under nitrogen replete and deplete conditions. Bioresour. Technol. 181, 128e135. Chiranjeevi, P., Mohan, S.V., 2016. Critical parametric influence on microalgae cultivation towards maximizing biomass growth with simultaneous lipid productivity. Renew. Energy 98, 64e71. Chokshi, K., Pancha, I., Trivedi, K., George, B., Maurya, R., Ghosh, A., Mishra, S., 2015. Biofuel potential of the newly isolated microalgae Acutodesmus dimorphus under temperature induced oxidative stress conditions. Bioresour. Technol. 180, 162e171. Courchesne, N.M.D., Parisien, A., Wang, B., Lan, C.Q., 2009. Enhancement of lipid production using biochemical, genetic and transcription factor engineering approaches. J. Biotechnol. 141 (1), 31e41. Da Silva, N.D.L., Maciel, M.R.W., Batistella, C.B., Maciel Filho, R., 2006. Optimization of biodiesel production from castor oil. Appl. Biochem. Biotechnol. 130 (1e3), 405e414. Deming, S.N., Morgan, S.L., 1993. Experimental Design: a Chemometric Approach. Elsevier.
37
Fakhry, E.M., El Maghraby, D.M., 2015. Lipid accumulation in response to nitrogen limitation and variation of temperature in Nannochloropsis salina. Bot. Stud. 56 (1), 1e8. ndez, I., Acie n, F., Ferna ndez, J., Guzma n, J., Maga n, J., Ferna Berenguel, M., 2012. Dynamic model of microalgal production in tubular photobioreactors. Bioresour. Technol. 126, 172e181. Ferreira, S.C., Bruns, R., Ferreira, H., Matos, G., David, J., Brandao, G., da Silva, E., Portugal, L., Dos Reis, P., Souza, A., 2007. Box-Behnken design: an alternative for the optimization of analytical methods. Anal. Chim. Acta 597 (2), 179e186. Gordillo, F.J., Goutx, M., Figueroa, F.L., Niell, F.X., 1998. Effects of light intensity, CO2 and nitrogen supply on lipid class composition of Dunaliella viridis. J. Appl. Phycol. 10 (2), 135e144. Griffiths, M.J., Harrison, S.T., 2009. Lipid productivity as a key characteristic for choosing algal species for biodiesel production. J. Appl. Phycol. 21 (5), 493e507. Ho, S.H., Chang, J.S., Lai, Y.Y., Chen, C.N.N., 2014. Achieving high lipid productivity of a thermotolerant microalga Desmodesmus sp. F2 by optimizing environmental factors and nutrient conditions. Bioresour. Technol. 156, 108e116. Jami, M.S., Rosli, N.S., Amosa, M.K., 2016. Optimization of manganese reduction in biotreated POME onto 3A molecular sieve and clinoptilolite zeolites. Water Environ. Res. 88 (6), 566e576. Lee, K., Lee, C.G., 2001. Effect of light/dark cycles on wastewater treatments by microalgae. Biotechnol. Bioprocess Eng. 6 (3), 194e199. Liu, J., Yuan, C., Hu, G., Li, F., 2012. Effects of light intensity on the growth and lipid accumulation of microalga Scenedesmus sp. 11-1 under nitrogen limitation. Appl. Biochem. Biotechnol. 166 (8), 2127e2137. Pal, D., Khozin-Goldberg, I., Cohen, Z., Boussiba, S., 2011. The effect of light, salinity, and nitrogen availability on lipid production by Nannochloropsis sp. Appl. Microbiol. Biotechnol. 90 (4), 1429e1441. Pulz, O., 2001. Photobioreactors: production systems for phototrophic microorganisms. Appl. Microbiol. Biotechnol. 57 (3), 287e293. Razzak, S.A., Ilyas, M., Ali, S.A.M., Hossain, M.M., 2015. Effects of CO2 concentration and pH on mixotrophic growth of Nannochloropsis oculata. Appl. Biochem. Biotechnol. 176 (5), 1290e1302. Singh, P., Guldhe, A., Kumari, S., Rawat, I., Bux, F., 2015. Investigation of combined effect of nitrogen, phosphorus and iron on lipid productivity of microalgae Ankistrodesmus falcatus KJ671624 using response surface methodology. Biochem. Eng. J. 94, 22e29. Tang, D., Han, W., Li, P., Miao, X., Zhong, J., 2011. CO2 biofixation and fatty acid composition of Scenedesmus obliquus and Chlorella pyrenoidosa in response to different CO2 levels. Bioresour. Technol. 102 (3), 3071e3076. Toledo-Cervantes, A., Morales, M., Novelo, E., Revah, S., 2013. Carbon dioxide fixation and lipid storage by Scenedesmus obtusiusculus. Bioresour. Technol. 130, 652e658. Wahidin, S., Idris, A., Shaleh, S.R.M., 2013. The influence of light intensity and photoperiod on the growth and lipid content of microalgae Nannochloropsis sp. Bioresour. Technol. 129, 7e11. Xin, L., Hong-Ying, H., Yu-Ping, Z., 2011. Growth and lipid accumulation properties of a freshwater microalga Scenedesmus sp. under different cultivation temperature. Bioresour. Technol. 102 (3), 3098e3102.
Contents lists available at ScienceDirect
South African Journal of Chemical Engineering journal homepage: http://www.journals.elsevier.com/ south-african-journal-of-chemical-engineering
Statistical optimization of parameters affecting lipid productivity of microalga Chlorella protothecoides cultivated in photobioreactor under nitrogen starvation Prakash Binnal*, P. Nirguna Babu Department of Chemical Engineering, Siddaganga Institute of Technology, Tumkur, 572103, Karnataka, India
article info
abstract
Article history:
In the present work, a lab scale photobioreactor was used to evaluate lipid productivity and
Received 11 October 2016
carbon dioxide fixation rate of microalgae Chlorella protothecoides under nitrogen deplete
Received in revised form
conditions. Effect of environmental conditions such as pH, temperature, light intensity,
1 January 2017
photoperiod (light to dark cycle ratio), CO2 concentration in air and aeration rate on lipid
Accepted 12 January 2017
productivity and carbon fixation rate of microalga was studied. Response surface methodology was adopted to optimize these conditions. All the parameters were found to be
Keywords:
statistically significant. Best operating conditions were evaluated to be: pH-6.51, Temper-
Chlorella protothecoides
ature-28.63 C, light intensity-5.31 klux, Photoperiod-15.36 h:8.64 h, CO2 concentration in
Photobioreactor
air-6.26% (v/v), Aeration rate 2.92 lpm. Lipid productivity under these conditions was
Nitrogen starvation
found to be 274.15 mg/(L day) which was 3.94 times higher than the value obtained in Nþ
Lipid productivity
experiment (69.46 mg/(L day)). Carbon fixation rates under Nþ and N conditions were
Carbon dioxide fixation
286.12 and 273.66 mg/(L day) respectively. © 2017 Published by Elsevier B.V. on behalf of Institution of Chemical Engineers. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/ by-nc-nd/4.0/).
1.
Introduction
Lipid productivity (LP) is a critical variable for evaluating algal species for biodiesel production. It is calculated as the product of biomass productivity (BP) (g dry weight/L day) and lipid content (% dw) to give an indicator of oil produced on a basis of both volume and time (Griffiths and Harrison, 2009). Key environmental variables affecting LP of microalgae under phototrophic conditions include: light intensity (LI), temperature, pH, photoperiod (ratio of light to dark cycle), CO2 concentration in air and aeration rate. Several reports are available in literature which illustrate the effect of these variables on LP under nitrogen replete (Nþ) conditions (Lee and Lee, 2001; Pal et al., 2011; Xin et al., 2011; Cabello et al., 2015;
Razzak et al., 2015). But very few authors have reported the influence under nitrogen deplete (N) conditions (Fernandez et al., 2012; Liu et al., 2012; Bruer et al., 2013; Toledo et al., 2013; Fakhry and Maghraby, 2015). Moreover, the number of variables in these studies was restricted to a maximum of 3. However, LP of microalgae is significantly affected by all aforementioned environmental variables. Especially, under nitrogen deprivation, microalgal cells are more sensitive to environmental conditions. They modify their biosynthetic pathways to accumulate lipids during this phase. Thus, knowledge of the influence of environmental conditions on LP under N conditions is very useful. It leads to development of useful kinetic expressions that could be applied for designing and modelling
* Corresponding author. Fax: þ91 816 2214070. E-mail addresses: [email protected] (P. Binnal), [email protected] (P.N. Babu). http://dx.doi.org/10.1016/j.sajce.2017.01.001 1026-9185/© 2017 Published by Elsevier B.V. on behalf of Institution of Chemical Engineers. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
south african journal of chemical engineering 23 (2017) 26e37
photobioreactors to predict LP and optimize the operating conditions (Da Silva et al., 2006). In this regard, the present study aims at the statistical optimization of aforementioned environmental parameters influencing the LP and RC of microalga Chlorella protothecoides in a lab scale externally illuminated photobioreactor (PBR) under N conditions.
1.1. Design of experiments by response surface methodology Response surface methodology (RSM) is a statistical technique which allows the simultaneous consideration of many variables at different levels and the interactions between those variables, using a smaller number of observations than conventional procedures. It is very useful for developing, improving, and optimizing processes in which a response of interest is influenced by several variables and the objective is to optimize this response. RSM has important application in the design, development and formulation of new products, as well as in the improvement of existing product design. It defines the effect of the independent variables, alone or in combination, on the processes. In addition to analysing the effects of the independent variables, this experimental methodology generates a mathematical model which describes the chemical or biochemical processes (Anjum et al., 1997). Several authors have described the use of RSM as an optimization tool (Singh et al., 2015; Amosa, 2016; Amosa and Majozi, 2016; Chiranjeevi and Mohan, 2016; Jami et al., 2016). Among RSM techniques, Box-Behnken design is considered as an efficient option and an ideal alternative to central composite designs. It combines a fractional factorial with incomplete block designs to avoid the extreme vertices and presents an approximately rotatable design with only three levels per factor. The number of experiments (N) required for the development of BBD is defined as N ¼ 2k(k1) þC0, (where k is number of factors and C0 is the number of central points). For, k ¼ 6, Co ¼ 6, N is 56 (Deming and Morgan, 1993). Considering all the linear terms, square terms and linear linear interaction items, the quadratic response model can be described as: Y ¼ bo þ
i¼k X i¼1
bi xi þ
k X i¼1
bii x2i þ
k X k X ii > j
bij xi xj þ ε
j
where Y is the predicted response variable, xi is the independent variable, bo , bi , bii , bij are the regression coefficients
27
and ε is the random error. Other advantages of BBD are: i) It permits estimation of the parameters of the quadratic model (ii) building of sequential designs (iii) detection of lack of fit of the model and (iv) use of blocks (Ferreira et al., 2007).
2.
Materials and methods
2.1.
Construction of photobioreactor
The pictorial, schematic and PID diagrams of Photobioreactor (PBR) used in the present study are shown in Figs. 1e3 respectively. The set up consists of a 5 L borosilicate vessel connected to a panel of straight glass tubes through a diaphragm pump. The panel of straight glass tubes consists of 6 glass tubes (1 inch diameter, 80 cm long) connected to each other through bends, flexible couplings and O-rings. The media from 5 L vessel is circulated through straight glass tubes and returned to the vessel by diaphragm pump. The vessel is equipped with pH, temperature and light intensity controllers. pH electrode (Mettler Tolerdo) and Pt-100 sensors have been used to measure pH and temperature of media in reactor. LED panels are arranged around 5 L vessel and straight glass tubes. The light intensity (0e10 klux) was adjusted from control panel. Two rotameters are provided for controlling flow rate of Carbon dioxide and air (Range of rotameter for air: 1 LPMe5 LPM, range of rotameter for CO2: 1 mLPM to 50 mLPM). The mixed CO2 enriched air was bubbled through media in 5 L vessel using ring sparger.
2.2.
Algae strain collection and culture condition
The original strain of C. protothecoides (SAG 211-10C) was obtained from Sammulung von Algenkulturen (SAG), Germany, and maintained on agar slants containing BG11 medium consisting of (g/L): NaNO3-1.5 g, K2HPO4$3H2O-0.04, KH2PO4$3H2O0.2, EDTA-0.0005, Fe ammonium citrate-0.005, citric acid-0.005, Na2CO3-0.02 and 1 ml of trace metal solution. The trace metal solution contains (g/L): H3BO3-2.85, MnCl2$4H2O-1.8, ZnSO4$7H2O-0.02, CuSO4$5H2O-0.08, CoCl2$6H2O-0.08 and Na2MoO4$2H2O-0.05. The pH of the medium was adjusted to 6.8. Stock cultures were inoculated into 100 ml of sterilized medium in 250 ml Erlenmayer flasks. The flasks were then incubated at 24 C in a rotary shaker and agitated at 120 rpm. After seven days, the algal biomass was recovered by centrifugation (REMI c-24 bl) at 10,000 rpm for 10 min and used as inoculam to PBR containing 3 L BG11 medium with nitrogen (for Nþ experiments) or BG11 medium without nitrogen (for N experiments)and the
Fig. 1 e Pictorial view of photobioreactor.
28
south african journal of chemical engineering 23 (2017) 26e37
Fig. 2 e Schematic view of photobioreactor.
initial biomass concentration was adjusted to 0.2 g/L. For Nþ experiment, operating conditions were set as: Temperature27 C, Light intensity-6 klux, photoperiod-10 h:12 h, pH-7.2, aeration rate-3 lpm. These conditions were optimized in our previous study (Babu and Binnal, 2015). For Nþ experiment, 1% CO2 air was continuously bubbled through medium.
2.3. Design of nitrogen depletion experiments in photobioreactor The ranges of parameters selected in the present study are depicted in Table 1. After inoculation, operating conditions of PBR were adjusted at specified values in a given experiment (Table 2). Response Surface Methodology (RSM) was used to design these experiments. A three level Box-Behnken design was used to study the effect of six variables (LI, temperature, pH of cultivation medium, photoperiod, CO2 concentration in air and aeration rate) influencing LP and RC of C. protothecoides under N conditions. A total of 54 experiments were conducted. During each run, samples were collected from PBR at
regular intervals of 24 h and analysed for biomass concentration, lipid content and carbon content. BP, LP and RC were evaluated as explained in section 2.4. Highest values of LP and RC obtained in each run were noted. Design expert software (Version 8, Stat Ease, USA) was used to analyse the data and generate response surfaces and contour plots. The accuracy of quadratic model obtained through regression analysis was tested by the coefficient of determination (R2). The response surface plot analysis was made by keeping four independent variables constant (at zero levels), while changing the other two independent variables.
2.4.
Analytical methods
2.4.1.
Estimation of biomass
The conical flask containing culture was shaken thoroughly. 1.5 ml of culture was pipetted out and transferred to previously weighed 2 ml microcentrifuge tubes (Make: TARSONS CILLOK). This was done in duplicate. The tubes were centrifuged in mini centrifuge (Model: GeNei™, SLM-CFT-10K) for
Fig. 3 e Process instrumentation diagram of photobioreactor.
south african journal of chemical engineering 23 (2017) 26e37
Table 1 e Experimental range and levels of the variables. Variable
Symbol
Temperature ( C) Light intensity (klux) pH Photoperiod (h:h) CO2 concentration in air (%, v/v) Aeration rate (lpm)
X1 X2 X3 X4 X5 X6
Coded factors and levels 1
0
1
20 1 3 8:24 1 1
30 5.5 6.5 16:8 8 3
40 10 10 24:0 15 5
29
10 min at 10,000 rpm. The supernatant in tubes was discarded and the tubes were dried in an oven for 8 h. The weight of dried tubes was noted down. W1 ¼ Weight of microcentrifuge tubes (g) W2 ¼ Weight of microcentrifuge tubes þ dried biomass (g) W ¼ W2W1 ¼ Weight of dry biomass per 1.5 ml culture medium (g)
Concentration of biomass; Xðg=LÞ ¼ W 103 1:5
(1)
Table 2 e Design of experiments along with responses.
2.4.2.
Run
X1
X2
X3
X4
X5
X6
LP
RC
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
20 40 20 40 20 40 20 40 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 20 40 20 40 20 40 20 40 30 30 30 30 30 30 30 30 20 40 20 40 20 40 20 40 30 30 30 30 30 30
1 1 10 10 1 1 10 10 1 10 1 10 1 10 1 10 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 1 10 1 10 1 10 1 10 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5
6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 3 3 10 10 3 3 10 10 3 10 3 10 3 10 3 10 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 3 3 10 10 3 3 10 10 6.5 6.5 6.5 6.5 6.5 6.5
8 8 8 8 24 24 24 24 16 16 16 16 16 16 16 16 8 8 24 24 8 8 24 24 8 8 24 24 8 8 24 24 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16
8 8 8 8 8 8 8 8 1 1 1 1 15 15 15 15 8 8 8 8 8 8 8 8 1 1 1 1 15 15 15 15 1 1 15 15 1 1 15 15 8 8 8 8 8 8 8 8 8 8 8 8 8 8
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 1 1 1 1 5 5 5 5 3 3 3 3 3 3 3 3 1 1 1 1 5 5 5 5 1 1 1 1 5 5 5 5 3 3 3 3 3 3
30.17 4.42 66.11 21.15 54.19 7.96 118.97 38.07 40.82 78.15 51.79 92.44 24.00 45.97 30.44 53.59 57.28 97.82 83.77 104.45 103.08 176.08 150.78 188.16 100 41.16 126.04 66.26 55.74 22.87 70.14 30.96 90.49 117.31 17.45 30.79 162.87 163.8 24.43 43.1 51.12 23.45 66.79 41.15 74.47 46.79 95.49 61.15 266.92 266.92 266.92 266.92 266.92 266.92
107.84 20.14 153.93 61.79 148.42 27.81 196.50 61.79 18.19 52.22 7.49 54.08 20.85 50.79 7.16 18.71 51.14 10.77 85.44 27.08 95.16 26.11 140.38 61.82 74.77 21.74 109.17 26.48 14.12 12.21 16.28 17.74 84.59 107.15 17.82 28.82 104.48 121.88 21.12 29.75 30.43 11.15 21.78 5.11 52.18 31.08 51.48 33.15 268.49 268.49 268.49 268.49 268.49 268.49
Estimation of total lipids was done following the protocol of Bligh and Dyer (1959). To a 15 ml vial containing algal biomass (300 mg), 2 ml methanol, and 4 ml chloroform were added. The mixture was agitated in a cyclomixer (Model: CM 101, REMI) for 2 min. 1 ml of chloroform was again added and the mixture was shaken vigorously for 1 min. 1.8 ml of distilled water was added and the mixture was mixed in a vortex again for 2 min. The two layers were separated by centrifugation for 15 min at 10,000 rpm. The lower layer was filtered through Whatman No. 1 filter paper into a previously weighed clean vial (W1). The process was repeated three times. All the chloroform phases were collected together, and evaporated in a Rotary vacuum evaporator (Superfit™, Rotavap, Model:PBV-7D). The final weight of vial with lipids was measured (W2). The lipids obtained were calculated by subtracting W1 from W2, and expressed as % dcw.
2.4.3.
Estimation of total lipids
Estimation of lipid productivity
Biomass productivity and lipid productivity were calculated as, Biomass productivity½mg dry biomass=ðL dayÞ ¼ ðXn XoÞ=n (2) LP mg lipid L1 day1 ¼ BP Lipid contentðwt%Þ 100
(3)
where Xo and Xn are biomass concentrations (in g/L) on zeroth and nth day respectively.
2.4.4.
Estimation of carbon dioxide fixation
The carbon content of algae biomass was determined using Perkin Elmer 2400 CHNS/O elemental analyzer. The carbon fixation rate was calculated as (Tang et al., 2011), RC ¼ CC BP MCO2 =MC
(4)
where CC is weight % of carbon in algae biomass, MCO2 and MC are molecular weights of carbon dioxide and carbon respectively.
3.
Results and discussion
3.1.
Statistical analysis of parameters
Table 2 depicts the highest values LP and RC obtained in each run. It was observed that, the time duration to attain these values varied for each run. This is obvious as the cultivation conditions were different for each run (except for Runs 49e54). The experimental data was analysed by multiple regressions using Design Expert software. The final estimative models for LP and RC in terms of uncoded factors were as follows.
30
south african journal of chemical engineering 23 (2017) 26e37
3.2. Influence of environmental factors on LP under nitrogen deplete conditions
RC ¼ 1251:49 þ 46:11X1 þ 25:64X2 þ 98:85X3 þ 25:73X4 þ 30:13X5 þ 97:35X6 0:813X1 2 2:03X2 2 7:89X3 2 0:76X4 2 2:07X5 2 15:22X6 2
(5)
LP ¼ 1330:65 þ 58:8X1 þ 48:63X2 þ 76:01X3 þ 20:81X4 þ 16:31X5 þ 69:63X6 1:015X1 2 4:14X2 2 5:59X3 2 0:6X4 2 1:27X5 2 9:84X6 2
(6)
where X1, to X6 are the actual values of independent variables. Tables 3 and 4 describe the ANOVA results for the second order response surface models. Fisher's variance ratio, which is defined as the ratio of mean lack of fit to mean square of pure experimental error, were high for both models (F ¼ 64.91 and 44.1 for LP and RC respectively) with a low probability (p < 0.0001), which demonstrates the high significance of the models. The significance of each term was determined by their respective p-values. The smaller the p-value, the more significant is the corresponding coefficient. In both models, all the linear terms and quadratic terms were statistically significant. All mutual interactions terms were found to be insignificant in both models (p > 0.05). Adequate precision measures the signal to noise ratio. A ratio greater than 4 is desirable. The values obtained in present work (28.51 mg/(L day) and 24.17 mg/(L day) for LP and RC respectively) indicated adequate signals. R2 (Coefficient of regression) is the proportion of variation in the depended variable explained by regression model. Adjusted R2 is coefficient of determination adjusted for the number of independent variables in the regression model (Ferreira et al., 2007). High values of R2 obtained in the present work for both models (0.97 and 0.94) indicated that the they would accurately predict the relationships between the parameters. Further, the “Predicted R2” values were in reasonable agreement with the “Adjusted R2” which confirmed that models could be used to navigate the design space.
3.2.1.
Mutual interactions of temperature and light intensity
Basically, to achieve highest LP, both lipid content and biomass concentration of microalgae should be increased. However, cultivation conditions that increase lipid content (high LI, high temperature) can adversely reduce biomass. In the same way, conditions that favour high biomass concentration will negatively affect lipid content. Hence, an optimum LI and temperature should be selected to maximize LP. The effect of irradiance and temperature on biomass and lipid content has been explored by various researchers. There has been a consensus with reference to effect of LI on biomass that, at low LI, photosynthesis rates are low. Increasing intensity improves BP and RC till saturation light intensity. Above saturation LI, biomass growth rate is hampered due to photoinhibition. However, there has been no clear consensus about effect of LI on lipid content. Gordillo et al. (1998) reported that lipid content of Dunaliella viridis kept decreasing with increase in LI from 35 to 700 mmol/(m2 s), while Liu et al. (2012) reported that lipid content of Scenedesmus sp. 11-1 increased rapidly with increasing irradiance and a highest lipid content of 41.12% was obtained at 250 mmol/(m2 s). Basically, when exposed to high light (HL), lipids serve as a sink of excessive light energy absorbed by photosynthetic apparatus in microalgae. Triacylglycerol (TAG), the main component in neutral lipids synthesis requires large amounts of ATP and NAD(P)H (which are produced by the photosynthesis). Therefore, lipid accumulation is helpful in the dissipation of excess light energy and prevention of the photochemical damage of algal cells. The excess carbon flux generated from the photosynthesis is channelled to the lipid accumulation on a unit biomass basis when individual cell is exposed to a large quantity of light energy (Courchesne et al., 2009). Therefore, the HL could promote the final lipid content (especially neutral lipid content). However, under ‘HL coupled with N conditions’, chloroplast lamellae are disrupted more
Table 3 e Analysis of variance (ANOVA) for response surface quadratic model (LP). Source Model X1 X2 X3 X4 X5 X6 X21 X22 X23 X24 X25 X26 Residual Lack of fit Pure error Total CV ¼ 21.08% R2 ¼ 0.97 Adjusted R2 ¼ 0.93 Predicted R2 ¼ 0.911 *Significant at p 0.05.
Sum of squares
df
Mean square
F value
p-value
2.902Eþ005 10573.7 4550.29 3259.44 2902.67 19359.9 10767.32 1.06Eþ005 72402.57 48252.84 15552.88 40019.4 15938.25 15272.52 15272.52 0.000 3.054Eþ005 Mean ¼ 91.51 Adequate precision ¼ 28.51
12 1 1 1 1 1 1 1 1 1 1 1 1 41 36 5 53
24180.56 10573.7 4550.29 3259.44 2902.67 19359.9 10767.32 1.06Eþ005 72402.57 48252.84 15552.88 40019.4 15938.25 372.5 424.24 0.00
64.91 28.39 12.22 8.75 7.79 51.98 28.91 284.5 194.37 129.54 41.75 107.43 42.79
<0.001 <0.001 0.0012* 0.0051* 0.0079* <0.0001 <0.0001 <0.0001 <0.001 <0.0001 <0.0001 <0.0001 <0.0001
Significant
31
south african journal of chemical engineering 23 (2017) 26e37
Table 4 e Analysis of variance (ANOVA) for response surface quadratic model (RC). Source Model X1 X2 X3 X4 X5 X6 X21 X22 X23 X24 X25 X26 Residual Lack of fit Pure error Total CV ¼ 31.95% R2 ¼ 0.94 Adj.R2 ¼ 0.90 Predicted R2 ¼ 0.8722
Sum of squares
df
Mean square
F value
p-value
3.247E005 17426.05 5148.01 4115.41 3091.23 11566.19 3439.1 68067.33 17463.93 96137.67 24351.36 1.068Eþ005 38164.1 24851.4 24851.4 0.000 3.456 þ 005 Mean ¼ 77.06 Adequate precision ¼ 24.173
12 1 1 1 1 1 1 1 1 1 1 1 1 41 36 5 53
26731.62 17426.05 5148.01 4115.41 3091.23 11566.19 3439.1 68067.33 17463.93 96137.67 24351.36 1.068Eþ005 38164.1 606.13 690.32 0.000
44.1 28.75 8.49 6.79 4.98 19.08 5.67 112.3 28.81 158.61 40.18 176.27 62.96
<0.001 <0.001 0.0058 0.0127 0.0312 <0.0001 0.0219 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
Significant
*Significant at p 0.05.
severely than Nþ conditions which reduce photosynthesis efficiency and carbon fixation efficiency. Thus, it can be concluded that effect of LI on lipid content of microalgae depends upon the ability of the individual species to capture carbon at high light. Fig. 4A and B depict the combined effect of LI and temperature on LP and RC respectively. At low temperature and low LI (20 C and 1 klux), LP and RC were 30.1 mg/(L day) and 107.8367 mg/(L day) respectively (Run 1). When LI was increased above 1 klux, both LP and RC increased at all temperatures till an optimum LI of 5.5 klux. Above 5.5 klux, both responses declined attaining a minimum value at 10 klux. However, the responses at 10 klux were higher than those obtained at 1 klux (Run 3: LP ¼ 66.10667 mg/(L day), RC ¼ 153.9267 mg/(L day)). At 20 C and 1 klux, BP and lipid content were 0.19 g/(L day) and 15.72%, while at 20 C and 10 klux, the corresponding values were 0.22 g/(L day) and 30.04%. Thus, lipid content increased with increasing LI. Since no significant changes in BP were observed at 1 klux and 10 klux, it was concluded that higher values of LP and RC obtained at 10 klux were due to high lipid content (Equation (1)). Also, lipid contents in Runs 2, 5, 6 were greater than for runs 4,7 and 8, compared in the same order (Run 2 v/s 4: 4.7%/17.45%, Run 5 v/s 7: 22.49%/39.39%, Run 6 v/s 8: 6.8%/25.1%), while BPs were (Run 2 v/s 4: 0.094/0.123, Run 5 v/s 7: 0.24/0.301, Run 6 v/s 8: 0.117/0.152). Thus, HL increased the lipid content of cells due to light stress. Therefore, LP and RC at 10 klux (Run 4, 7, 8) were greater than the corresponding values at 1 klux (Run 2,5,6). Similar to irradiance, maintaining an optimal temperature in PBR was required to promote high LP. If we analyse the effect of temperature on biomass, it has been suggested in literature that changes in cytoplasmic viscosity under suboptimal temperature conditions is responsible for less efficient carbon and nitrogen utilization. Hence, increasing temperature beyond the optimum reduces protein synthesis and consequently results in decreased growth rates. On the other hand, the effect of temperature on lipid content of microalgae is species specific and relies upon concentration of ROS (reactive oxygen species). ROS, if present in adequate
concentrations (which can occur either at low or high temperature) create stress on microalgal cells thus improving their lipid content (Chokshi et al., 2015). In present work, analysis of Runs (1e8, 25e32) indicated that both lipid content and BP at 20 C were higher than at 40 C. This can be attributed to two factors. At lower temperature, lipid % was higher due to generation of ROS which positively affected LP. The reduced BP at 40 C could be due to the fact that, at high temperature, CO2 solubility in water decreases, which further reduces RC. Also, at high temperature, photosystem (II) gets severely damaged which reduces rate of photosynthesis.
3.2.2. Mutual interactions of temperature and CO2 concentration Theoretically, increasing CO2 concentration in air should result in increased biomass concentration and lipid content of microalgae. However, extremely high concentration of CO2 will cause substrate inhibition and lead to decreased RC and LP. Hence, air containing an optimum concentration of CO2 must be supplied to algae culture to obtain its best performance. In the present work, low LP and RC were recorded at 1% CO2 (Fig. 4C and D). As depicted in these two figures, both LP and RC increased with increasing CO2 concentration until an optimum CO2 concentration of 8% at all temperatures. Above this concentration, both the responses decreased due to substrate inhibition. This is due to the fact that, increasing CO2 concentration in air results in increased rate of photosynthesis, which results in increased concentration of oxygen concentration in medium. Oxygen competes with CO2 on carboxylating CO2 enzyme (RuBISCO) thus reducing rate of carbon fixation. This leads to reduced BP. However, under nitrogen deprivation, the absorbed carbon, instead of being incorporated into proteins and assist cells to grow, is diverted to lipid synthesis path way that would be taken as a carbon sink. Hence, above 8% CO2, although lipid content improved, a decline in LP was observed due to decreased growth rate. The negative effect of increasing CO2 concentration above optimum was more severe at higher temperature than low temperatures (Run 29e32).
32
south african journal of chemical engineering 23 (2017) 26e37
Fig. 4 e Response surface plots: A, B- Interactions of LI and temperature; C, D- Interactions of temperature and CO2 concentration in air; E, F- Interactions of CO2 concentration in air and pH.
3.2.3.
Mutual interactions of pH and CO2 concentration
Fig. 4E and F depict the effect of pH on LP and RC respectively. pH of the culture medium plays an important role in influencing the lipid accumulation in the algae. Activation of
ACCase (key enzyme in lipid biosynthesis) is pH dependent. In present study, a comparison of LP and RC values for runs: 9 v/s 11, 10 v/s 12,14 v/s 16,17 v/s 18,19 v/s 20 and 21 v/s 22 (Runs 9, 10, 13, 14, 17, 19 were at pH 3 and Runs 11, 12, 15,16, 18, 20 were
south african journal of chemical engineering 23 (2017) 26e37
at pH 10) clearly indicated that, in all these cases, LP and RC were higher at pH 10 than at pH 3. At low pH of 3, CO2 inhibits the activity of enzyme carbonic anhydrase (CA), which plays an important role in carbon fixation. Thus, both BP and lipid content were low leading to lower values of LP. With increasing pH, activity of RUBISCO increases thus improving carbon fixation which further improves LP. Thus, LP and RC increased till an optimum pH of 6.9. However, at alkaline pH, although a rise in lipid content was noticed, BP reduced owing to non-optimum pH conditions thus reducing LP. For example, lipid content for Run 21 and 22 were 32.1% and 71.2, while BP were 0.32 and 0.25 g/(L day) respectively. It has been known that, environmental stress along with nitrogen deprivation leads to higher lipid content. Thus, there was not much difference in BP values for pH 3 and 10. Hence, the higher value of LP at pH 10 was due to higher lipid content. It is well known that, CO2 in medium is present in culture medium in three forms (CO2 3 , HCO3 and dissolved CO2). The concentrations of these ions are pH dependent. At alkaline pH, HCO 3 dominates. Thus, C. protothecoides cell was able maintain its growth at alkaline conditions by consuming HCO 3.
3.2.4.
Mutual interactions of photoperiod and light intensity
It can be observed from Table 1 that, at 20 C and LI of 1 klux, when photoperiod was increased from 8:12 to 24:0, LP increased from 30.1 mg/(L day) to 54.19 mg/(L day) and RC increased from 107.83 mg/(L day) to 148.41 mg/(L day). Similarly, an analysis of responses from Run 17 to 32 clearly indicated that increasing light period from 8 h to 24 h improved LP and RC. In the present work, at low LI and low light periods, both BP and lipid content were low due to photolimtation condition and insufficient time duration of light period. Increasing photoperiod at low LI improved LP and RC. However, at higher LI, an increase in photoperiod initially improves LP till an optimum. Further increase in light period leads to decrease in LP (Fig. 5A). High light intensities cause damage to photosynthesis apparatus which can be counterbalanced by exposing the cells to darkness (Pulz, 2001). During light period, light energy absorbed by microalgal cells is converted to chemical energy providing NADPH2 and ATP. During dark period, the excess light energy absorbed in light period is used for endergonic reactions. Maximum photosynthetic efficiency are achieved when light/dark cycle period approaches photosynthetic turnover time (Wahidin et al., 2013). Fig. 5AeB indicates that 16:8 is the optimum photoperiod. This result indicates that cell division was not under circadian control but light responsive (diurnal). The diurnal pattern of cell growth in the light and cell division in the dark is thought to allow for maximum collection of light energy.
3.2.5. Mutual interactions of photoperiod and CO2 concentration Fig. 5CeD depict the interactions among CO2 concentration and photoperiod. Increasing light period for low to optimum CO2 concentration (8% v/v) improves carbon fixation. This is due to the fact that microalgae can absorb CO2 only in the presence of light and an increase in duration of light naturally leads to increased RC. However, above optimum CO2 concentration, due to inhibition of CA enzyme, RC decreases.
3.2.6. Mutual interactions of aeration rate and CO2 concentration Fig. 5EeF illustrate the effect of aeration rate and CO2 concentration on LP and RC. It is evident that, irrespective of CO2
33
concentration, LP and RC increased with the aeration rate. Rate of aeration and CO2 concentration are very important parameters influencing the performance of PBR as they together deicide the amount of available carbon in PBR. The rate of carbon supply should match with rate of its consumption by cells. In general, at low aeration rate (1 lpm) and low CO2 concentration (1%), photosynthetic rates were low due to inadequate mixing of broth and insufficient amount of inorganic carbon. With increase in flow rate from 1 lpm to 4 lpm, both LP and RC improved for all CO2 concentrations. This can be attributed to better mixing at high flow rate which facilitated removal of dissolved oxygen from medium, thus improving rate of photosynthesis. However, an increase of aeration rate from 4 lpm to 5 lpm slightly reduced LP and RC. This may be a result of high shear on microalgal cells that might have reduced biomass concentration. Also, higher flow rate decreases residence time of CO2 in the reactor. Further, it was observed that, at all aeration rates, above 8% CO2, a reduction in LP and RC was observed. As mentioned earlier, high photosynthetic efficiency is achieved when rate of carbon supply is equal to consumption rate by microalgal cells. Thus, aeration rate of 4 lpm with 8% CO2 was found to be adequate.
3.3. Optimization of environmental parameters influencing LP and RC under N conditions using desirability function Desirability function is a popular and established technique for the simultaneous determination of optimum settings of input variables that can determine optimum performance levels for one or more responses. It ranges from zero outside the limits, to one at the goal. The goal fields for response have five options: none, maximum, minimum, target and within range. In the present work, ‘Numerical optimization’ of the software has been chosen in order to find the specific point that maximizes the desirability function. The program seeks to maximize this function. The goal seeking begins at a random starting point and proceeds up the steepest slope to a maximum. There may be two or more maxima because of curvatures in the response surfaces and their combination in the desirability function. Starting the search of optimum from several points in the design space improves the chances of finding the best local maximum (Amini et al., 2008). The individual desirabilities are then the combined to give the overall desirability (D). The main objective of preset study was maximize LP and RC while following a cost driven approach. Hence, the goal ‘maximize’ was selected for LP and RC with highest importance. In order to reduce cost of process, the goal ‘minimize’ was selected for LI, photoperiod, aeration rate and Carbon dioxide supply. For temperature and pH, the option ‘in the range’ was selected. The number of cycles was set as 50 and the duplicate solution filter selected was epsilon (minimum difference) for eliminating duplicate solutions. The results of optimization are shown in Table 5. Seven solutions were suggested by software. Highest desirability (0.76) was observed for solution No.1. However, LP and RC for this solution were least for this solution. Highest LP and RC were predicted for solutions 5 and 6 respectively. However, desirability functions for these solutions were lower than those reported for solutions 1e4. A comparison of Solution 2 with 5 and 6 revealed that, LP can be increased by 2.53% and 2.2% by: i) increasing LI by 12.42% and
34
south african journal of chemical engineering 23 (2017) 26e37
Fig. 5 e Response surface plots: A, B- Interactions of photoperiod of LI; C, D- Interactions of photoperiod and CO2 concentration in air; E, F- Interactions of CO2 concentration in air and aeration rate.
35
south african journal of chemical engineering 23 (2017) 26e37
Table 5 e Suggested solutions by ‘Numerical optimization’ technique. Solution No. 1 2# 3 4 5 6 7
Temperature ( C)
LI (klux)
pH
Photoperiod (h:h)
Volume % of CO2 in air
Aeration rate (lpm)
LP (mg/(L day))
RC (mg/(L day))
Desirability
28.79 28.63 28.68 28.62 28.87 28.67 28.61
4.95 5.31 5.65 5.47 5.77 5.97 5.84
6.58 6.51 6.48 6.48 6.67 6.47 6.42
14.32 15.36 16.11 15.69 16.61 16.89 16.5
5.59 6.26 6.59 6.41 6.44 6.9 6.85
2.73 2.92 3.12 3.03 3.33 3.29 3.18
260 268.37 272.3 270.78 275.17 274.26 273.27
256 267.28 274 269.94 271.11 273.9 273.75
0.76 0.722 0.675 0.63 0.61 0.54 0.428
# Selected optimum solution.
8.66% and ii) by increasing aeration rate by 14% and 12.67% respectively. Since the main cost components of microalga cultivation are LI and aeration rates, solutions 5 and 6 were rejected as they are not economically attractive. Solution 2 was accepted as global optimum as it suggested reasonably low LI (5.31 klux), low photoperiod (15.36:8.64), low aeration rate (2.92 lpm) and lower concentration of CO2 in air (6.26%) while predicting relatively high LP and RC (268.37 mg/(L day) and 267.28 mg/(L day)). Fig. 6 shows the ramp desirability of solution 2 that was generated via numerical optimization. The individual desirability functions (di) for each of the responses, and the calculated geometric mean as maximum over all desirability (D ¼ 0.722) is represented in Fig. 7. For validation of this Solution 2, a confirmatory experiment was conducted at optimum results. The values of LP and RC obtained were found to be 274.15 mg/(L day) and 273.66 mg/ (L day) respectively (% error from model predictions were 2.15% and 2.4% for LP and RC respectively).
3.4. Comparison of LP and RC under Nþ and N conditions Comparison of LP and RC obtained under Nþ and N conditions when cultivated under their respective optimum conditions are depicted in Table 6. It is evident that, LP under N conditions was 3.94 times greater than the value obtained under Nþ conditions. Despite nitrogen starvation, C. protothecoides was able to maintain the growth rate till 7th day, after which a stationary phase begun. However, biomass concentration in N was slightly lower than in N cultures (2.85 g/L v/s 2.81 g/L). The reason for sustaining growth rate under N conditions could be due to the fact that, an intracellular nitrogen pool such as that in chlorophyll could be used as nitrogen source to maintain the growth rate. Similar results were obtained by Ho et al. (2014), who reported that, biomass concentration of Scenedesmus obliquus CNW-N continued to rise after depletion of
Fig. 6 e Desirability ramp of optimum solution.
36
south african journal of chemical engineering 23 (2017) 26e37
Fig. 7 e Bar graph representing individual desirability of all responses (di) in correspondence with desirability (D).
Table 6 e Comparison of LP and RC obtained under Nþ conditions with N¡ conditions. Time (Day)
Nþ conditions
N conditions
X (g/L) Lipid content LP (mg/(L day)) RC (mg/(L day)) X (g/L) Lipid content (%) LP mg/(L day) RC mg/(L day) 0 1 2 3 4 5 6 7 8 9
0.2 0.32 0.61 1.42 1.82 2.17 2.49 2.85 2.12 2.05
5.89 7.55 10.54 12.55 14.46 16.22 18.12 22.48 21.48 22.29
e 9.1 21.61 50.95 58.52 63.9 69.46* 67.45 51.55 35.83
e 128.85 161.12 186.45 222.23 261.17 266.12 271.15 277.15 286.12*
0.2 0.33 0.54 1.31 1.74 2.12 2.33 2.81 2.44 1.95
5.89 9.2 17.44 27.53 40.18 49.62 69.47 73.48 70.98 71.1
e 11.96 29.64 31.2 154.93 190.54 250.87 274.15* 198.74 138.25
e 124.41 157.52 184.17 209.81 254.32 261.16 273.66* 261.94 254.57
* Highest values obtained.
nitrogen in the medium. However, lipid content increased much rapidly in N cultures than Nþ cultures (5.89%e73.8% in 7 days for N cultures as compared to 5.89% to 22.48% in 6 days for Nþ cultures). These factors resulted in higher LP in N cultures than in Nþ cultures.
4.
Conclusions
A lab scale externally photobioreactor was used in the present work to evaluate lipid productivity of microalga C. protothecoides under nitrogen deprivation.54 experiments were conducted in accordance with Box-Behnken design in order to study and optimize the effect of six environmental factors on lipid productivity and rate of carbon dioxide fixation. Among 54 trails, 6 trails were replicates (at centre points). Under these conditions, local optimum values were observed (LP ¼ 266.92 mg/(L day) and RC ¼ 268.4867 mg/(L day)). However, to obtain global optimum, numerical optimization technique of Design Expert software was used. A cost driven approach which minimizes costs of light intensity, aeration rate, photoperiod, rate of CO2 supply and maximizes LP was adopted. Seven solution were suggested by software. The solution which provides a cost effective process while producing high LP was selected. The global optimum values were: pH6.51, Temperature-28.63 C, light intensity-5.31 klux, Photoperiod-15.36 h:8.64 h, CO2 concentration in air-6.26% (v/v) and Aeration rate 2.92 lpm. Under these conditions, highest LP of 274.15 mg/(L day) and RC of 273.66 mg/(L day) were obtained.
Thus, C. protothecoides can be a potential feedstock for biodiesel production. The regression models for both responses had high R2 (0.97 and 0.94 for LP and RC respectively), high Fischer variance ratio (64.91 and 44.1 for LP and RC respectively), and adequate precision (28.51 and 24.173 for LP and RC respectively). Further, Adjusted R2 for both regression models were close to predicted R2 (Tables 3 and 4). Thus, models could be used to predict the experimental results with high accuracy. In order to obtain high volumetric productivity of microalgal system in large photobioreactors, precise optimization of the environmental variables is crucial. In the present study, optimization of variables was done in small lab scale reactor of 5 L capacity. The approach can be extended to large scale photobioreactors to obtain their highest performance.
Acknowledgement The authors thank financial support from Visveswaraya Technological University, Belagavi, Karnataka, India, under VTU research Grants Scheme-2010-11 (VTU/Aca./2011-12/A-9/ 6380).
Nomenclature LP (mg/L day) Lipid Productivity RC (mg/L day) Carbon fixation rate BP (g/(L day) Biomass productivity
south african journal of chemical engineering 23 (2017) 26e37
LI (klux) HL PBR X (g/L) CC (wt %) MC MCO2
Light intensity High light Photobioreactor Biomass concentration Carbon content of biomass Molecular mass of Carbon Molecular weight of carbon dioxide
References Amini, M., Younesi, H., Bahramifar, N., Lorestani, A.A.Z., Ghorbani, F., Danesh, A., Sharifzadeh, M., 2008. Application of response surface methodology for optimization of lead biosorption in an aqueous solution by Aspergillus niger. J. Hazard. Mater 154, 694e702. Amosa, M.K., 2016. Sorption of water alkalinity and hardness from high-strength wastewater on bifunctional activated carbon: process optimization, kinetics and equilibrium studies. Environ. Technol. 37 (16), 2016e2039. Amosa, M.K., Majozi, T., 2016. GAMS supported optimization and predictability study of a multi-objective adsorption process with conflicting regions of optimal operating conditions. Comput. Chem. Eng. 94, 354e361. Anjum, M.F., Tasadduq, I., Al-Sultan, K., 1997. Response surface methodology: a neural network approach. Eur. J. Oper. Res. 101 (1), 65e73. Babu, P.N., Binnal, P., 2015. Lipid productivity of microalgae Chlorella vulgaris and Nannochloropsis oculata in externally illuminated lab scale photobioreactor. Int. J. ChemTech Res. 7 (5), 2217e2221. Bligh, E.G., Dyer, W.J., 1959. A rapid method of total lipid extraction and purification. Can. J. Biochem. Phys. 37 (8), 911e917. Breuer, G., Lamers, P.P., Martens, D.E., Draaisma, R.B., Wijffels, R.H., 2013. Effect of light intensity, pH, and temperature on triacylglycerol (TAG) accumulation induced by nitrogen starvation in Scenedesmus obliquus. Bioresour. Technol. 143, 1e9. nchez, L., Revah, S., Cabello, J., Toledo-Cervantes, A., Sa Morales, M., 2015. Effect of the temperature, pH and irradiance on the photosynthetic activity by Scenedesmus obtusiusculus under nitrogen replete and deplete conditions. Bioresour. Technol. 181, 128e135. Chiranjeevi, P., Mohan, S.V., 2016. Critical parametric influence on microalgae cultivation towards maximizing biomass growth with simultaneous lipid productivity. Renew. Energy 98, 64e71. Chokshi, K., Pancha, I., Trivedi, K., George, B., Maurya, R., Ghosh, A., Mishra, S., 2015. Biofuel potential of the newly isolated microalgae Acutodesmus dimorphus under temperature induced oxidative stress conditions. Bioresour. Technol. 180, 162e171. Courchesne, N.M.D., Parisien, A., Wang, B., Lan, C.Q., 2009. Enhancement of lipid production using biochemical, genetic and transcription factor engineering approaches. J. Biotechnol. 141 (1), 31e41. Da Silva, N.D.L., Maciel, M.R.W., Batistella, C.B., Maciel Filho, R., 2006. Optimization of biodiesel production from castor oil. Appl. Biochem. Biotechnol. 130 (1e3), 405e414. Deming, S.N., Morgan, S.L., 1993. Experimental Design: a Chemometric Approach. Elsevier.
37
Fakhry, E.M., El Maghraby, D.M., 2015. Lipid accumulation in response to nitrogen limitation and variation of temperature in Nannochloropsis salina. Bot. Stud. 56 (1), 1e8. ndez, I., Acie n, F., Ferna ndez, J., Guzma n, J., Maga n, J., Ferna Berenguel, M., 2012. Dynamic model of microalgal production in tubular photobioreactors. Bioresour. Technol. 126, 172e181. Ferreira, S.C., Bruns, R., Ferreira, H., Matos, G., David, J., Brandao, G., da Silva, E., Portugal, L., Dos Reis, P., Souza, A., 2007. Box-Behnken design: an alternative for the optimization of analytical methods. Anal. Chim. Acta 597 (2), 179e186. Gordillo, F.J., Goutx, M., Figueroa, F.L., Niell, F.X., 1998. Effects of light intensity, CO2 and nitrogen supply on lipid class composition of Dunaliella viridis. J. Appl. Phycol. 10 (2), 135e144. Griffiths, M.J., Harrison, S.T., 2009. Lipid productivity as a key characteristic for choosing algal species for biodiesel production. J. Appl. Phycol. 21 (5), 493e507. Ho, S.H., Chang, J.S., Lai, Y.Y., Chen, C.N.N., 2014. Achieving high lipid productivity of a thermotolerant microalga Desmodesmus sp. F2 by optimizing environmental factors and nutrient conditions. Bioresour. Technol. 156, 108e116. Jami, M.S., Rosli, N.S., Amosa, M.K., 2016. Optimization of manganese reduction in biotreated POME onto 3A molecular sieve and clinoptilolite zeolites. Water Environ. Res. 88 (6), 566e576. Lee, K., Lee, C.G., 2001. Effect of light/dark cycles on wastewater treatments by microalgae. Biotechnol. Bioprocess Eng. 6 (3), 194e199. Liu, J., Yuan, C., Hu, G., Li, F., 2012. Effects of light intensity on the growth and lipid accumulation of microalga Scenedesmus sp. 11-1 under nitrogen limitation. Appl. Biochem. Biotechnol. 166 (8), 2127e2137. Pal, D., Khozin-Goldberg, I., Cohen, Z., Boussiba, S., 2011. The effect of light, salinity, and nitrogen availability on lipid production by Nannochloropsis sp. Appl. Microbiol. Biotechnol. 90 (4), 1429e1441. Pulz, O., 2001. Photobioreactors: production systems for phototrophic microorganisms. Appl. Microbiol. Biotechnol. 57 (3), 287e293. Razzak, S.A., Ilyas, M., Ali, S.A.M., Hossain, M.M., 2015. Effects of CO2 concentration and pH on mixotrophic growth of Nannochloropsis oculata. Appl. Biochem. Biotechnol. 176 (5), 1290e1302. Singh, P., Guldhe, A., Kumari, S., Rawat, I., Bux, F., 2015. Investigation of combined effect of nitrogen, phosphorus and iron on lipid productivity of microalgae Ankistrodesmus falcatus KJ671624 using response surface methodology. Biochem. Eng. J. 94, 22e29. Tang, D., Han, W., Li, P., Miao, X., Zhong, J., 2011. CO2 biofixation and fatty acid composition of Scenedesmus obliquus and Chlorella pyrenoidosa in response to different CO2 levels. Bioresour. Technol. 102 (3), 3071e3076. Toledo-Cervantes, A., Morales, M., Novelo, E., Revah, S., 2013. Carbon dioxide fixation and lipid storage by Scenedesmus obtusiusculus. Bioresour. Technol. 130, 652e658. Wahidin, S., Idris, A., Shaleh, S.R.M., 2013. The influence of light intensity and photoperiod on the growth and lipid content of microalgae Nannochloropsis sp. Bioresour. Technol. 129, 7e11. Xin, L., Hong-Ying, H., Yu-Ping, Z., 2011. Growth and lipid accumulation properties of a freshwater microalga Scenedesmus sp. under different cultivation temperature. Bioresour. Technol. 102 (3), 3098e3102.