Enzyme and Microbial Technology 38 (2006) 209214

Optimization of fermentation conditions for the production of ethanol

from sago starch by co-immobilized amyloglucosidase and cells of
Zymomonas mobilis using response surface methodology
Veera Venkata Ratnam Bandaru a, , Subba Rao Somalanka b,1 , Damodara Rao Mendu c,2 ,
Narasimha Rao Madicherla b,3 , Ayyanna Chityala b,4
a

Department of Neurology, Johns Hopkins University School of Medicine, 600 North Wolfe Street, Pathology 233/Meyer 222, Baltimore, MD 21287, USA
b Center for Biotechnology, Department of Chemical Engineering, College of Engineering, Andhra University, Visakhapatnam 530003, India
c Department of Pediatrics, Division of Infectious Diseases, Johns Hopkins University School of Medicine, 720 Rutland Avenue,
Ross 1135B, Baltimore, MD 21205, USA
Received 15 May 2005; received in revised form 6 June 2005; accepted 8 June 2005

Abstract
Statistical experimental design was used to optimize the conditions of simultaneous saccharification and fermentation (SSF), viz. temperature, pH and time of fermentation of ethanol from sago starch with co-immobilized amyloglucosidase (AMG) and Zymomonas mobilis
MTCC 92 by submerged fermentation. Maximum ethanol concentration of 55.3 g/l was obtained using a starch concentration of 150 g/l. The
optimum conditions were found to be a temperature of 32.4 C, pH of 4.93 and time of fermentation of 17.24 h. Thus, by using SSF process
with co-immobilized AMG and Z. mobilis cells MTCC 92, the central composite design (CCD) was found to be the most favourable strategy
investigated with respect to ethanol production and enzyme recovery.
2005 Elsevier Inc. All rights reserved.
Keywords: Co-immobilization; Amyloglucosidase; Sago starch; Ethanol; Response surface methodology

1. Introduction
The new developments in biotechnology will play an
important role in resolving part of the energy and food problems that lie ahead. One important development, which has
stimulated worldwide interest is the utilization of renewable carbohydrate sources for the production of ethanol as
a liquid fuel [13]. Sago starch is an agricultural material abundantly produced in India and other tropical coun

Corresponding author. Tel.: +1 443 287 3717; fax: +1 410 502 6736.
E-mail addresses: vbandar2@jhmi.edu, bvvratnams@yahoo.co.in
(V.V.R. Bandaru), dmendu1@jhmi.edu (D.R. Mendu),
mnrmd au@yahoo.com (N.R. Madicherla), ayyanna123@rediffmail.com
(A. Chityala).
1 Tel.: +91 891 284 4890/2534032.
2 Tel: +1 410 614 0058.
3 Tel.: +91 891 284 4883/2537641, mobile: +91 9440403268.
4 Tel.: +91 893 3222959/891 2552405.
0141-0229/$ see front matter 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.enzmictec.2005.06.002

tries [4] and is an alternative source of energy. The starch

is a product extracted from the seeds or tubers, stems of
palms and cycads such as Metroxylon sagu and is a mixture of 27% (w/w) amylose and 73% (w/w) amylopectin
[5]. There are a variety of products that can be obtained
from starch biomass via hydrolysis. Alcohol is one of
the largest volume of products that can be produced from
biomass. Recently, there has been active research aimed
at attaining an increase in ethanol yield by immobilized
techniques.
Zymomonas mobilis cells and AMG were co-immobilized
in the form of alginate beads and SSF was carried using the
co-immobilized cells and enzyme. The SSF process combines enzymatic hydrolysis of starch to glucose and ethanol
fermentation into a single operation. Consequently, this process offers a great potential of increased rate of hydrolysis,
reduction of fermentation time, decreased capital cost [6] and
removing end point inhibition as well as eliminating the need

210

V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209214

for separate reactors. SSF process has attracted many investigators [7,8].
The traditional one-factor at a time technique used for
optimizing a multivariable system is not only time consuming but also often easily misses the alternative effects between
components. Also, this method requires to carry out a number
of experiments to determine the optimum levels, which are
untrue. These drawbacks of single factor optimization process can be eliminated by optimizing all the affecting parameters collectively by CCD using response surface methodology
(RSM). Recently, many statistical experimental design methods have been employed in bioprocess optimization. Among
them, RSM is the one suitable for identifying the effect of
individual variables and for seeking the optimum conditions
for a multivariable system efficiently. This method has been
successfully applied to optimize alcoholic fermentation and
other fermentation media [916]. A detailed account of this
technique has been outlined [17]. Basically, this optimization
process involves three major steps: performing the statistically designed experiments, estimating the coefficients in a
mathematical model and predicting the response and checking the adequacy of the model. Hence, the authors report the
application of the RSM using the BoxWilson design [18] of
experiments to develop a mathematical correlation between
the temperature, pH and time of fermentation and concentration of ethanol.
In the present study, the optimal conditions of temperature,
pH and time of fermentation for maximum ethanol yield have
been quantified from the BoxWilson CCD.

2. Materials and methods

2.5. Co-immobilization
The enzyme was immobilized on powdered chitin using
the procedures of Stanley et al. [19]. Exponentially growing
Z. mobilis cells (8 g dry cell weight) were centrifuged and
resuspended with the immobilized glucoamylase on chitin
in 50 ml physiological saline. The suspension was carefully
mixed with 50 ml 4% sodium alginate solution. The slurry
was then added drop wise to a 0.05 M CaCl2 solution with
continuous stirring using a 5 ml disposable pipette tip. Beads
of 34 mm diameter were formed in this solution. The total
volume of beads was approximately 68 ml. The concentrations of the chitin immobilized glucoamylase and Z. mobilis
cells in the beads were 77.2 and 93.2 g dry weight/l beads,
respectively.
2.6. Production media and fermentation
Starch liquefaction was carried out by adding 0.2% (v/w)
-amylase to the slurry at pH 6.5 and heating at 95 C for
1 h. No problem was faced during the solubilization of starch
because of the reduction in viscosity of fermentation mashes
by enzymes. Fermentation media were composed of 150 g
sago starch, 10 g yeast extract and co-immobilized enzyme
and cell beads (120 ml bead volume) in 1 l water. Fermentation was carried out in a Biostat M fermentor supplied by B.
Braun Co., Germany, with all necessary controls. The reactor
was of 2 l capacity and the working volume was 1 l. The operating conditions were maintained at a temperature of 30 C
and pH 5.0. The reactor was maintained under anaerobic
conditions.

2.1. Substrate

2.7. Analytical methods

Sago starch was collected from cultivators, East Godavari

District, Andhra Pradesh, India.

Ethanol was estimated by GLC in which a flame ionization

detector and stainless steel column (2.0 m length, 3.0 mm
i.d.) packed with Porapak-Q (5080 mesh, manufactured by
Nucon Engineers, India) were used. The column oven was
operated isothermally at 150 C and the detector and injection
ports were kept at 170 C. Nitrogen was used as carrier gas
at a flow rate of 30 cm3 /min and the combustion gas was a
mixture of hydrogen and air [13]. Sugars were determined
using Millers method [20].

2.2. Organism
Z. mobilis MTCC 92 obtained from IMTECH, Chandigarh, India, was used throughout this study.
2.3. Enzymes
-Amylase from Bacillus licheniformis and AMG from
Aspergillus niger were obtained from Sigma Chemical Co.
(St Louis, MO). The activities of the two enzymes were
60 KNU/g and 200 AGU/ml, respectively.
2.4. Growth conditions
The Z. mobilis MTCC 92 was maintained on agar slants
having composition (g/l): glucose, 100; yeast extract, 10;
KH2 PO4 , 1; (NH4 )2 SO4 , 1; MgSO4 7H2 O, 0.5 and the cells
were grown at a temperature of 35 C and pH of 5.5.

2.8. Experimental design and optimization

CCD [18] was used in the optimization of ethanol production. Temperature (X1 , C), pH (X2 ) and time of fermentation
(X3 , h) were chosen for the independent variables shown in
Tables 1 and 2. Ethanol concentration (Yi , g/l) was used as
the dependent output variable. For statistical calculations the
variables Xi were coded as xi according to Eq. (1):
xi =

Xi x i
,

xj

i = 1, 2, 3, . . . , k

(1)

V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209214

211

Table 1
Independent variables in the experimental plan

Table 3
Experimental and the predicted yields for ethanol

Variables

Run no.

Temperature, X1
pH, X2
Time, X3 (h)

Coded levels
( C)

1.682

1.682

13.18
3.318
6.59

20
4
10

30
5
15

40
6
20

46.82
6.682
23.41

Table 2
The central composite design matrix employed for three independent variables (actual values given in Table 1)
Run no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

X1

X2

X3

1
1
1
1
0
0
1
1
1
1
0
0
1.682
1.682
0
0
0
0
0
0

1
1
1
1
0
0
1
1
1
1
0
0
0
0
1.682
1.682
0
0
0
0

1
1
1
1
0
0
1
1
1
1
0
0
0
0
0
0
1.682
1.682
0
0

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

X1

20
40
20
40
30
30
20
40
20
40
30
30
13.18
46.82
30
30
30
30
30
30

X2

4
4
6
6
5
5
4
4
6
6
5
5
5
5
3.318
6.682
5
5
5
5

X3

10
20
20
10
15
15
20
10
10
20
15
15
15
15
15
15
6.59
23.41
15
15

Ethanol yield (g/l)

Experimental

Predicted

25.92
43.54
32.57
19.48
52.10
51.92
26.43
30.11
24.16
40.64
51.01
52.40
11.45
22.62
37.42
34.60
32.57
39.37
52.00
51.89

26.2817
41.2643
30.3534
20.2888
51.8215
51.8215
24.0027
30.7082
24.8173
38.6599
51.8215
51.8215
12.8248
23.5334
38.865
35.4431
30.34735
43.8805
51.8215
51.8215

where xi is dimensionless value of an independent variable,

Xi the real value of an independent variable, x i the real value
of the independent variable at the center point and
xj is the
step change.
A 23 -factorial central composite experimental design,
with six axial points ( = 3)and six replications at the center points (n0 = 6) leading to a total number of 20 experiments
was employed (Table 2) for the optimization of the parameters. The second degree polynomials (Eq. (2)) were calculated
with the statistical package (Stat-Ease Inc., Minneapolis,
MN, USA) to estimate the response of the dependent variable:

rial CCD and regression analysis. Also this method evaluating

the effective factors and building models to study interaction
and select optimum conditions of variables for a desirable
response. The full CCD, based on three basic principles
of an ideal experimental design, primarily consists of (1)
a complete 2n factorial design, where n is the number of
test variables, (2) n0 center points (n0 1) and (3) two axial
points on the axis of each design variable at a distance of
1.682 (2n/4 = 1.682 for n = 3) from the design center. Hence,
the total number of design points (20) is N = 2n + 2n + n0 .
The suitable temperature, initial pH and time of fermentation were also determined using statistical CCD. The experimental design matrix is given in Tables 1 and 2. Twenty
experiments were performed using different combinations of
the variables as per the CCD. Using the results of the experiments the following second order polynomial equation giving
the ethanol as a function of temperature (X1 , C), pH (X2 ) and
time of fermentation (X3 , h) was obtained:

Yi = b0 + b1 X1 + b2 X2 + b3 X3 + b11 X12 + b22 X22

Yi = 223.529 + 7.609949X1 + 51.6821X2 + 3.16389X3

+ b33 X32 + b12 X1 X2 + b23 X2 X3 + b13 X1 X3

(2)

where Yi is predicted response, X1 , X2 , X3 the independent

variables, b0 the offset term, b1 , b2 , b3 the linear effects, b11 ,
b22 , b33 the squared effects and b12 , b23 , b13 are interaction
terms.

3. Results and discussion

The most important physical factors, which affect the fermentative production of ethanol are the temperature, initial
pH and time of fermentation. The RSM includes full facto-

0.11891X12 5.18443X22 0.20794X32

0.22387X1 X2 + 0.39075X2 X3 + 0.064175X1 X3
(3)
The predicted levels of ethanol using the above equation
were given along with experimental data in Table 3. The
coefficients of the regression model (Eq. (3)) calculated are
listed in Table 4, in which they contain three linear, three
quadratic and three interaction terms and one block term.
The effects of all three parameters, i.e. temperature, pH and
time of fermentation and their interactions with each other
on percent ethanol concentration were found to be significant

V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209214

212

Table 4
Coefficients, t-statistics and significance probability of the model
Term

Coefficient

Value

Standard error of coefficient

t-Value

p-Value

Constant
Temperature
pH
Time
Temperature temperature
pH pH
Time time
Temperature pH
pH time
Temperature time

b0
b1
b2
b3
b11
b22
b33
b12
b23
b13

223.529
7.609949
51.6821
3.16389
0.11891
5.18443
0.20794
0.22387
0.39075
0.064175

26.01494
0.603539
7.032349
1.207079
0.006087
0.608655
0.024346
0.081707
0.163413
0.016341

8.59232
12.60887
7.349195
2.621113
19.5373
8.51784
8.54107
2.73998
2.391174
3.927156

6.27E06
1.83E07
2.46E05
0.025549
2.7E09
6.77E06
6.61E06
0.020833
0.037883
0.002833

Fig. 1. Parity plot showing the distribution of experimental vs. predicted

values of ethanol yield.

(p 0.05). The parity plot showed a satisfactory correlation

between the values of experimental values and predictive values (Fig. 1), wherein, the points cluster around the diagonal
line which indicates the good fit of the model, since the deviation between the experimental and predictive values was
less. And also the goodness of the model could be checked
by different criteria. The coefficient of determination, R2 is
0.9824 which implies that 98.24% of the sample variation
in the ester yield is attributed to the independent variables.
The R2 value also indicates that only 1% of the variation is
not explained by the model. The value of R is 0.99. The corresponding analysis of variance (ANOVA) was presented in
Table 5. Chi-square test was also carried out to check the best
2 < 2 , since
fit of the model. The model was a good fit. cal
tab
2
2
cal is 1.6052 and tab is 30.14. The predicted optimum levels
of temperature pH and time of fermentation were obtained by
applying the regression analysis to the Eq. (3). The predicted

Fig. 2. Response surface and contour plot of temperatures vs. pH on ethanol

production (time was kept constant at 15 h).

and experimental ethanol concentration at the optimum levels of fermentation conditions were also determined by using
Eq. (3).
Figs. 24 represent the isoresponse contour and surface
plots for the optimization of fermentation conditions of
ethanol. The effects of the pH and temperature on the ethanol
production showed in Fig. 2. An increase in the temperature with pH up to the optimum point increased the ethanol
production to a maximum level and a further increase in
the temperature with pH the trend is reversed. The interaction effect of the time and pH on the ethanol production in
Fig. 3 clearly indicates a proper combination for production
of ethanol. An increase in the pH with time increased the
ethanol production gradually but at a higher pH and time the
trend is reversed. The optimum for maximum ethanol production lies near the centre point of the pH and time. A similar

Table 5
ANOVA for the entire quadratic model
Source of variation

Sum of squares (SS)

Degrees of freedom (d.f.)

Mean squares (MS)

F-value

Probe > F

Regression
Residual

2994.509
53.40789

9
10

332.7233
5.340789

62.29852

Total

3047.917

19

V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209214

Fig. 3. Response surface and contour plot of pH vs. time on ethanol production (temperature was kept constant at 30 C).

213

at fixed levels is laborious and time consuming. This method

requires a complete series of experiments for every factor of
interest. Moreover, such a method does not provide means
of observing possible factors interactions. In contrast, CCD
offer a number of important advantages. For instance, the
researchers could easily determine factor effects with considerably less experimental effort, identify factors, find optima,
offer greater precision and facilitate system modeling.
Thus, the present study using the RSM with CCD enables
to find the importance of factors at different levels. A high
similarity was observed between the predicted and experimental results, which reflected the accuracy and applicability
of RSM to optimize the process for ethanol production. The
results of this study have clearly indicated RSM is an effective method for maximum production of ethanol using SSF
with AMG and Z. mobilis MTCC 92.

References

Fig. 4. Response surface and contour plot of temperature vs. time on ethanol
production (pH was kept constant at 15).

effect on the response was observed for the temperature at

any level of the time an increase in the temperature with time
up to the optimum point increased the ethanol production
to maximum level and a further increase in the temperature
with time decreased the ethanol production is shown in Fig. 4.
Therefore, an optimum was observed near the central value of
pH, temperature and time. The optimum conditions for maximum ethanol concentrations were obtained at a temperature
of 32.4 C, pH of 4.93 and time of fermentation of 17.24 h.
A maximum ethanol concentration of 55.3 g/l was obtained
at these optimum parameters.

4. Conclusion
The one factor at a time is the most frequently used operation in optimization process. This technique is based on
changing one parameter at a time, while keeping the others

[1] Scheller WA. The clean fuel from biomass and wastes. In: Proceedings of the institute of technology. 1977.
[2] Prescott SC, Dunn CG. Industrial microbiology. 3rd ed. New York:
McGraw-Hill; 1959.
[3] Lindeman LR, Rocchiccioli C. Ethanol in Brazil: brief summary
of the state of industry in 1977. Biotechnol Bioeng 1979;31:
110719.
[4] Tan K. Production of sago starch. In: Stanton WR, Flach M, editors.
Proceedings of the second international sago symposium. The Hague:
Martinus Nijho; 1980. p. 56. ISBN 90-2472470-8.
[5] Swinkels JJM. In: Van Beynum GMA, Roels JA, editors. Starch
conversion technology. New York: Marcel Dekker; 1985. p. 23. ISBN
0-82477194-X.
[6] Kosaric N, Wieczorek A, Cosentino G, Magee R. In: Dellweg H,
editor. Biotechnology, vol. 3. Weinheim: Verlag Chemie; 1983. p.
315. ISBN 3-52725765-9.
[7] Deshpande V, Raman HS, Rao M. Simultaneous saccharification
and fermentation of cellulose to ethanol using penicillium funiculosum cellulose and free or immobilized Saccharomyces uvarum cells.
Biotechnol Bioeng 1983;25:167984.
[8] Spangler DJ, Emert GH. Simultaneous saccharification and fermentation with Z. mobilis. Biotechnol Bioeng 1986;28:1158.
[9] Chen SL. Optimization of batch alcohol fermentation of glucose
syrup substrate. Biotechnol Bioeng 1981;23:182736.
[10] Bowman L, Geiger E. Optimization of fermentation conditions for
alcohol production. Biotechnol Bioeng 1984;26:14927.
[11] Zertuche L, Zall RR. Optimizing alcohol production from whey
using computer technology. Biotechnol Bioeng 1985;27:54754.
[12] Ratnam BVV. Studies on physico-chemical and nutritional parameters for the production of ethanol from palmyra jaggery by submerged fermentation using Saccharomyces cerevisiae. Ph.D. Thesis,
Andhra University, Visakhapatnam, AP, India; 2001.
[13] Ratnam BVV, Narasimha Rao M, Damodara Rao M, Subba Rao S,
Ayyanna C. Optimization of fermentation conditions for the production of ethanol from sago starch using response methodology. World
J Microbiol Biotechnol 2003;19:5236.
[14] Sunitha I, Subba Rao MV, Ayyanna C. Optimization of medium
constituents and fermentation conditions for the production of
L-glutamic acid by the coimmobilized whole cells of Micrococus glutanicus and Pseudomonas reptilivora. Bioprocess Eng
1998;18:3539.
[15] Ambati P, Ayyanna C. Optimizing medium constituents and fermentation conditions for citric acid production from palmyra jag-

214

V.V.R. Bandaru et al. / Enzyme and Microbial Technology 38 (2006) 209214

gery using response surface method. World J Microbiol Biotechnol

2001;17:3315.
[16] Maddox IS, Reichert SH. Use of response surface methodology for
the rapid optimization of microbiological media. J Appl Bacteriol
1977;43:197204.
[17] Cochran NG, Cox GM. Experimental designs. John Wiley and Sons
Inc.; 1968. p. 611.

[18] Box GEP, Wilson KB. On the experimental attainment of optimum

conditions. J R Stat Soc (Ser B) 1951;13:145.
[19] Stanley WL, Watters GG, Kelly SH, Olson AC. Glucoamylase
immobilized on chitin with Glutaraldehyde. Biotechnol Bioeng
1978;20:13540.
[20] Miller GL. Use of DNS reagent for determination of reducing sugars.
Anal Chem 1959;31:4268.