Bioresource Technology 120 (2012) 140–148

Contents lists available at SciVerse ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Effects of fermentation substrate conditions on corn–soy co-fermentation

for fuel ethanol production
Linxing Yao, Show-Ling Lee, Tong Wang ⇑, Juliana M.L.N.de Moura, Lawrence A. Johnson
Department of Food Science and Human Nutrition, Center for Crops Utilization Research, Iowa State University, Ames, IA 50011, United States

h i g h l i g h t s

" A successful integration of soy and corn biorefineries is shown.

" Soy skim from soybean aqueous extraction processing was shown to have rate promotion effect on corn-ethanol fermentation without affecting the
ethanol yield.
" The soy skim provided all nutrients needed for the yeast.
" Such corn–soy co-fermentation benefits soybean aqueous extraction processing and ethanol fermentation.
" Soy-enhanced whole stillage had high protein and lysine contents and can provide feed with improved quality.

a r t i c l e i n f o a b s t r a c t

Article history: Soy skim, a protein-rich liquid co-product from the aqueous extraction of soybeans, was co-fermented
Received 1 December 2011 with corn to produce ethanol. Effects of soy skim addition level, type of skim, corn particle size, water-
Received in revised form 17 April 2012 to-solids ratio, and urea on co-fermentation were determined. The addition of 20–100% skim increased
Accepted 20 April 2012
the fermentation rate by 18–27% and shortened the fermentation time by 5–7 h without affecting ethanol
Available online 8 May 2012
yield. Finely ground corn or high water-to-solids ratio (P3.0) in the mash gave higher fermentation rates,
but did not increase the ethanol yield. When the water was completely replaced with soy skim, the addi-
Keywords:
tion of urea became unnecessary. Soy skim retentate that was concentrated by nanofiltration increased
Corn–soy co-fermentation
Distiller’s dry grains
fermentation rate by 25%. The highest level of skim addition resulted in a finished beer with 16% solids,
Dry-grind corn ethanol fermentation 47% protein (dwb) containing 3.6% lysine, and an ethanol yield of 39 g/100 g dry corn.
Enzyme-assisted aqueous extraction Ó 2012 Elsevier Ltd. All rights reserved.
process
Soy skim

1. Introduction has achieved notable enhancement on fermentation performance.

The continuous growing demand on fuel ethanol encourages
the exploration of new and improved technologies to increase
the efficiency of traditional alcohol fermentation and co-product
utilization. Fermentation rate and final ethanol concentration are
usually limited by nutrient deficiencies and ethanol intolerance
of yeast cells. Supplementing fermentation media with nutrients
such as soy flour (Ju et al., 1983; Damiano and Wang, 1985; Viegas
et al., 1985; Maia and Nelson, 1994), oryzenin (a rice protein),
albumin, and koji mold mycelia (Hayashida et al., 1974), and finger
millet flour (Reddy and Reddy, 2006) in fuel ethanol production
⇑ Corresponding author. Address: 2312 Food Science Building, Department of
Food Science and Human Nutrition, Center for Crops Utilization Research, Iowa
State University, Ames, IA 50011, United States. Tel.: +1 515 294 5448; fax: +1 515
294 8181.
E-mail address: tongwang@iastate.edu (T. Wang).
Although the mechanism of the promoting effect from these pro-
teins or protein–lipid complexes is not fully understood, the major
contribution may be the remediation of nutrient deficiencies in the
media (Viegas et al., 1985); however, inexpensive sources of sup-
plements are needed to maintain or reduce the total production
cost of fuel ethanol.
An integrated biorefinery concept with corn–soy co-fermenta-
tion was proposed by Yao et al. (2011) to utilize the co-product
of the enzyme-assisted aqueous extraction of soybeans (EAEP),
soy skim, in dry-grind corn fermentation. The authors found that
soy skim increased the fermentation rate and thus shortened
fermentation time. The co-fermentation, which uses the liquid
EAEP co-product rich in hydrolyzed soy protein, not only benefits
soybean EAEP by utilizing and adding value to its dilute protein
stream (skim), but also introduces an inexpensive protein source
to supplement alcohol fermentation. The study conducted by Yao
et al. (2011) compared the performance of co-fermentation with

0960-8524/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.biortech.2012.04.071
 L. Yao et al. / Bioresource Technology 120 (2012) 140–148
different levels of skim, but the ratio of water-to-solids was not
controlled. The authors suggested that the fermentation rate could
be further increased by adjusting the water-to-solids ratio of the
initial corn–soy slurry because such a ratio was known to affect
the ethanol production rate (Wang et al., 1999). Other parameters,
such as corn particle size, yeast and enzyme doses also affect either
ethanol yield or final ethanol concentration (Naidu et al., 2007;
Rathore et al., 2009).
The stillage obtained from corn–soy co-fermentation should
have an increased lysine content attributable to the high lysine
content in soybean. Together with the high solid and protein con-
tents, the stillage from the co-fermentation may have significantly
improved feed quality compared to traditional distillers’ dried
grains with soluble (DDGS) (Yao et al., 2011). Thus a complete
compositional analysis of the resulting stillage is needed for future
animal feeding trials.
The objectives of the present study were to determine the effect
of soy skim addition with a fixed ratio of water-to-solids on fer-
mentation performance in order to provide conclusive evidence
for the beneficial effect of soy protein, and the effects of water-
to-solids ratio, corn particle size, addition of urea, and type of skim
on corn–soy co-fermentation performance, and to present the
composition of the soy-enhanced whole stillage.
2. Methods
2.1. Corn
Yellow dent corn produced in 2008 was acquired from Heart of
Iowa Cooperative (Nevada, IA). The corn was ground using a Fitz
Mill (Model DAS 06, Fitzpatrick Co., Elmhurst, IL) at 3000 rpm with
a screen 1531–0125 to give coarsely ground corn. Finely ground
corn was obtained by first using the same mill at 5000 rpm with
screen 1531–0125, and then feeding back to the mill for a second
grinding at 7000 rpm with screen 1532–0040. The coarsely ground
corn, which was similar to that used in commercial ethanol plants
in the Midwest, was used unless otherwise specified. It contained
11.4% moisture, 3.8% (dwb) oil, 1.2% (dwb) ash, and 73% (dwb)
starch. The finely ground corn had 10.1% moisture. Corn flour pur-
chased from a local grocery store had 11.7% moisture, 67% (dwb)
starch, and 2.9% (dwb) oil. The finely ground corn and corn flour
were only used in the experiment testing the effect of corn particle
size on the fermentation rate as discussed in Sections 2.5 and 3.1.
Fig. 1. Particle size distribution of corn materials.
141
The particle size distributions of the three corn materials was
determined by passing through U.S. Standard Sieves with mesh #
20, 30, 40, 50, and 70 (with 0.84, 0.60, 0.42, 0.30, and 0.21 mm
opening) (Fig. 1).
2.2. Soy skim and skim retentate after nanofiltration
Soy skim was produced in the pilot plant of the Center for
Crops Utilization Research, Iowa State University (Ames, IA) via
two-stage counter-current EAEP (de Moura et al., 2011) and kept
frozen in plastic bottles at 14 °C until used. The detailed proce-
dure to produce soy skim can be found in the study of de Moura
et al. (2011). The proximate composition of soy skim was 11% sol-
ids, which contained 56% (dwb) protein, 9% oil (dwb), and 13%
(dwb) ash. The soy skim was further concentrated by using a
cross-flow membrane filtration system at pilot-plant scale
(NCSRT’s OptiSep 800, Raleigh, NC). The system was composed
of a membrane module containing three membrane plates (two
membrane sheets per plate) with 0.5-mm channel heights, a
30-L jacketed glass reactor, a flow-meter, a heat exchanger, and
a centrifugal pump (CR4- model D- grundfos, 3 HP, 3450 rpm,
Fresno, CA). A flat polyethersulfone nanofiltration membrane
(PES 10% NaCl rejection) (Mycrodyn-Nadir, Raleigh, NC) with an
effective area of 0.096 m2 was used. Skim filtration was per-
formed at 50 °C, transmembrane pressure of 4.82 bar, and a recir-
culation flow rate of 21 L/min. About 30 L of skim were
nanofiltered to a concentration factor (volume of feed/volume of
retentate) of 4.5. The resulting skim retentate contained 27% sol-
ids, which contained 67% (dwb) protein, 8.7% (dwb) oil, and 11%
(dwb) ash.
2.3. Materials used in fermentations
Liquid R-amylase SPEZYME Xtra (13,642 R-amylase units/g) and
a saccharifying enzyme G-ZYME 480 Ethanol (401 gluco-amylase
units/g), both from Genencor Inc. (Cedar Rapids, IA), were used
for liquefaction and saccharification, respectively. Lactrol (462 g
virginiamycin/lb), an antibiotic extract, was from PhibroChem
(Ridgefield Park, NJ). Dry yeast (Saccharomyces cerevisiae) Ethanol
Red was acquired from Fermentis, a division of Lesaffre Yeast Corp.
(Headland, AL). Urea was supplied by Keytrade USA Inc. (Kordova,
TN). All these materials were industrial grade and are used in dry-
grind ethanol plants.
142 L. Yao et al. / Bioresource Technology 120 (2012) 140–148
2.4. General fermentation procedure
The liquefaction and simultaneous saccharification and fermen-
tation were done using the Alcohol Fermentation Monitor (or AFM,
Metrohm Applikon, Netherlands) that could handle six samples at
a time. The total amount of corn slurry was maintained at 400 to
410 g. Ground corn was mixed with cold water or soy skim in a
500-mL AFM glass bottle, and then amylase (0.5 g) was added.
The bottle was connected to the AFM and the mixture was heated
to 76 °C (maximum temperature for the instrument) with continu-
ous stirring at 300 rpm. After 2 h, the bottle was cooled to 30 °C
with tap water. The bottle was removed from the AFM to adjust
the pH to 4.7 with 6 mol/L sulfuric acid. Dry yeast (0.5 g), glucoam-
ylase (0.5 g), lactrol (10 mg), and urea (250 mg) were added. Urea
was added to the fermentation media in most studies except for
those described in Sections 2.7 and 3.3, in which the effect of
nitrogenous nutrient adequacy was examined. The bottle was
tightly connected to the AFM, and the AFM started to monitor
the fermentation by measuring the CO2 flow rate. The CO2 produc-
tion rate (mL/min) and total CO2 production (mL) were obtained
from the automated AFM as shown in Fig. 2.
2.5. Effects of skim replacement level and corn particle size on co-
fermentation performance
The ratio of water-to-solids (solids = dry corn + skim solids) was
kept constant at 2.4 (the lowest water-to-solids ratio used in this
study). The water in the control, which had no soy skim added,
was proportionally replaced by skim in five treatments, namely
20%, 40%, 60%, 80%, and 100% replacements. Three corn materials
were used, coarsely ground corn, finely ground corn, and corn
flour. Using coarsely ground corn, three replications were carried
out for each of the six skim replacements, and two replications
were carried out for the finely ground corn and corn flour. A
two-way factorial analysis was performed to test the significance
of the differences among corn particle sizes and levels of skim
Fig. 2. Alcohol Fermentation Monitor graphs of CO2 production rate (A) and total
CO2 production (B).
replacements as well as the interaction between corn particle size
and skim replacement level.
2.6. Effect of water-to-solids ratio on the co-fermentation performance
The ratios of water-to-solids were set at 2.4, 3.0, 3.5 and 4.0 to
examine the effect of the initial solid concentration on the co-
fermentation performance. The water-to-solids ratio of 2.4 was
the same as that used previously (Yao et al., 2011) so that a com-
parison could be made between the two studies. A lower ratio than
2.4 caused problems in the AFM as the corn particles blocked the
gas outlet. Two ratios of dry corn to skim solids were used, 4 and
19, which were equivalent to the 100% and 20% skim replacements,
respectively. The control had no soy skim added. Only coarsely
ground corn was used in this experiment, since it represented
the size used in commercial production. Three replications were
carried out on the control at each water-to-solids ratio, two repli-
cations were carried out on all treatments using 100% and 20%
skim replacement levels. The statistical design and data analysis
were the same as for the study described in Section 2.5.
2.7. Effect of urea nitrogen on the co-fermentation performance
The effect of urea on the fermentation rate was examined.
Treatments with 20% and 100% skim replacements with a 3.5
water-to-solids ratio were used for the addition of 0.06% urea. Only
coarsely ground corn was used in this experiment. Two replica-
tions were carried out for each treatment. The statistical design
and data analysis were the same as described in Section 2.5.
2.8. Corn–soy co-fermentation with nanofiltered soy skim
The skim retentate obtained from nanofiltration, which con-
tained 27% solids, was diluted with distilled water to obtain two
additional solid concentrations, 15% and 20%. Together with the
starting skim retentate, the three skim retentate samples were des-
ignated as ‘‘15% conc.’’, ‘‘20% conc.’’, and ‘‘27% conc.’’, and were
used to replace of the water in the fermentations. Fermentation
rates and ethanol yields were compared with those of the control
and 100% skim (regular skim with 11% solids and used for 100%
replacement level). The water-to-solids ratio was kept at 2.4, the
same as for those in the experiment described in Section 2.5. Three
replications were carried out on the control, 100% skim, and the
27% conc., and two replications were carried out on the 15% and
20% conc. treatments. The replications for 0% and 100% skim were
from separate runs, independent from those in other experiments
in order to perform randomized block statistical analysis.
2.9. Proximate analysis of whole stillage
After 48 h of fermentation, the bottle was capped and heated in a
water bath at 65 °C for 20 min to deactivate the yeast. To obtain rep-
resentative samples from the high-solid beer, special sampling as
described by Yao et al. (2011) was followed. First the cooled whole
stillage was transferred to a 1-L Erlenmeyer flask. While the flask
was vigorously shaken, about 40 mL of slurry was poured into a
400-mL beaker, and such a sample transferring was repeated four
times until a combined small portion of sample was collected. The
combined sample was transferred to appropriate weighing dishes
while being stirred with a spatula for compositional analyses. The
solids content was determined by weighing the dry solids after
oven-drying at 110 °C overnight. Oil content was determined by
an acid hydrolysis method (AOAC official method 922.06), and pro-
tein content by using the Dumas nitrogen combustion method with
an ElementarVario MAXCN analyzer (Elementar Analysensysteme
GmbH, Hanau, Germany) and a nitrogen-to-protein conversion
 L. Yao et al. / Bioresource Technology 120 (2012) 140–148
factor of 6.25. Total residual sugar was determined by using a Mega-
zyme total starch kit (Megazyme International Ltd., Wicklow,
Ireland) following the procedure of Srichuwong and Jane (2011).
Ash content was determined by heating the dry stillage at 550 °C
for 5 h (AOAC 923.03). Total carbohydrate content was obtained
by subtracting the oil, protein, and ash contents from 100%. Total fi-
ber content was obtained by subtracting the total residual sugar
content from total carbohydrate content. Amino acid content was
determined by the University of Missouri Agricultural Experiment
Station Chemical Laboratories (Columbia, MO).
2.10. Total ethanol determination
The ethanol production was determined by three methods
(weight loss, AFM, and HPLC). For the weight loss method, the
AFM bottle weights before and after fermentation were used to cal-
culate the mass loss. Water evaporation and material loss on the
heating rod and stir bar, 0.42 g on average, was subtracted from
the mass loss. Thus, the ethanol yield was calculated as (Wang
et al., 2009):
Ethanol yield ðg per 100g dry cornÞ
¼ 100  ½46  ðg of mass loss  0:42Þ=44=ðg of dry cornÞ
To calculate ethanol yield from AFM data, the total CO2 (mL)
production at 48 h was obtained (Fig. 2). Then the ethanol yield
was calculated by using the following equation:
Ethanol yield ðg per100g dry cornÞ
¼ 100  ð46  mL of total CO2 produced
 0:001842=44Þ=ðg of dry cornÞ
where the constant 0.001842 was the density (g/mL) of CO2 at
20 °C and 1 atm. The HPLC method gave ethanol concentrations
(g/mL) of the liquid portion of the finished beer by injecting 5 lL
of microfiltered sample into a Varian HPLC equipped with an
Aminex HPX-87H column (300  7.8 mm, Bio-Rad, Hercules, CA),
and a refractive index detector at 30 °C. The column was eluted
with 5 mM H2SO4 at a 0.6 mL/min flow rate at 65 °C. The ethanol
concentration of the finished beer was converted to ethanol yield
by the following equation:
Ethanol yield ðg per100g dry cornÞ
¼ 100  ½g of finished beer  ð1-solid%Þ=0:9527Þ
 ðg=mL ethanol concentration from HPLCÞ=ðg of dry cornÞ
where the constant 0.9527 was the average density of the liquid
portion of the finished beer. The liquid density was estimated by
weighing a known volume of filtered beer.
Paired t tests were performed to test the difference between the
AFM method and the other two methods. Statistical analyses of
Table 1
Effects of skim replacement level and corn particle size on maximum production rate (mean ± SD) of CO2i.
Corn % Skim replacement
0 20
Coarsely ground corn 17.76 ± 1.74 20.96 ± 1.26
Finely ground corn 21.84 ± 0.49 23.96 ± 1.65
Corn flour 18.03 ± 0.35 21.96 ± 0.34
LSMEANiii 19.21b 22.30a
i
LSMEAN is least square mean. LSMEAN followed by different letters are significantly different at p < 0.05.
ii
LSMEAN for main effect of corn materials.
iii
LSMEAN for main effect of skim replacement level.
iv
The p value for the comparison between LSMEAN for coarsely ground corn and corn flour is 0.0853.
143
ethanol yield were performed on data obtained with all three
methods.
2.11. Statistical analysis
Data were analyzed with SAS (Version 9.1, SAS Institute Inc.
Cary, NC, USA) to test significant difference at p = 0.05. The effects
of fermentation parameters, such as water-to-solids ratio, corn
particle size, skim replacement level, and urea addition on ethanol
yield.
3. Results and discussion
3.1. Effects of skim replacement and corn particle size on the co-
fermentation performance
Fig. 2 presents an example of AFM graphs for CO2 production
rate (mL/min) and total CO2 production (mL) during fermentation.
Measuring weight loss caused by CO2 release is a commonly used
method to monitor ethanol production during yeast fermentation
studies (Wang et al., 2009). The changes in weight of the reaction
mixture can be stoichiometrically converted to the mass of ethanol
ð1Þ produced. Thus, the maximum CO2 production rate, lmax;CO2 , which
is the peak value labeled in Fig. 2(A), corresponds to the maximum
ethanol production rate. A greater (maximum) ethanol production
rate enables shorter fermentation time. The example given in Fig. 2
illustrates that the ‘‘20% Skim’’ had higher lmax;CO2 than the ‘‘Con-
trol’’, and, thus, used short fermentation time to produce the same
volume of ethanol as the ‘‘Control’’. Therefore, the maximum CO2
production rate, lmax;CO2 , is used in the following discussion as
ð2Þ one of the indicators of fermentation performance.
The level of skim replacement and corn particle size signifi-
cantly affected lmax;CO2 . Regardless of the level of skim replace-
ment, finely ground corn gave the highest lmax;CO2 , followed by
corn flour, and then coarsely ground corn (Table 1). Similar obser-
vations were reported by Naidu et al. (2007) who observed that
corn with a smaller particle size tended to give a higher ethanol
yield. The corn flour had a slightly lower oil and starch content
(see Section 2) than the coarsely ground corn, which indicated that
part of the germ and bran were removed. The processing of com-
mercial corn flour might have affected its fermentation perfor-
mance, and resulted in lower fermentation rates than the finely
ground corn that had particle size similar to that of corn flour
(Fig. 1).
Regardless of the corn material used, the control gave signifi-
cantly lower lmax;CO2 than any other skim treatments (Table 1).
ð3Þ
The interaction between skim replacement level and corn particle
size was insignificant (p = 0.88). When examining the fermentation
rates from each corn material, for coarsely ground corn, the
averaged lmax;CO2 of all the skim replacement treatments was
21 mL/min, which was 10% higher than the rate of the control;
for corn flour, the averaged lmax;CO2 of all the skim replacement
40 60 80 100 LSMEANii
20.93 ± 1.85 22.05 ± 1.28 22.06 ± 1.71 22.62 ± 1.75 21.07y
23.64 ± 1.70 24.29 ± 0.66 24.51 ± 0.97 25.17 ± 0.34 23.90x
22.35 ± 0.17 22.35 ± 0.79 24.27 ± 0.21 24.02 ± 0.68 22.16y,iv
22.31a 22.90a 23.62a 23.94a
144 L. Yao et al. / Bioresource Technology 120 (2012) 140–148
treatments was 23 mL/min, about 28% higher than that of the con-
trol; and for finely ground corn, the averaged lmax;CO2 of all the
skim replacement treatments was 24 mL/min, 10% higher than that
of the control. Compared with the control the greatest rate increase
was from corn flour. Since these experiments were designed with a
fixed water-to-solids ratio of 2.4, the effects of solids concentration
of initial slurry on liquefaction and the following saccharification
and fermentation were minimal when the same corn material
was used. The particle size of skim solids was greatly different from
ground corn, thus, the difference might affect the mash viscosity
and consequently influence the performance of enzymatic hydro-
lysis and yeast fermentation when different proportions of skim
were added to the fermentation media. The results showing that
the fermentation rate-promoting effect of soy skim were consis-
tent for corn materials having various particle sizes removed such
uncertainty even though the ethanol industry will not likely use
such finely ground corn in the actual production due to increased
energy input. The increase of CO2 production rate as affected by
the skim addition resulted in fermentations completed in 5 to
7 h shorter time than for the control.
This experiment confirmed the results by Yao et al. (2011) that
soy skim increases ethanol production rate and shortens the total
fermentation time. However, the final ethanol yield in this experi-
ment was unaffected by the skim addition. Studies by Ju et al.
(1983), Damiano and Wang (1985), and Viegas et al. (1985)
showed that both ethanol production rate and ethanol concentra-
tion increased when soy flour was added to corn fermentation.
Vidal et al. (2009) pretreated raw dry-milled starch with protease
and found that proteolysis increased early ethanol production rate
but had no effect on final ethanol yield. The discrepancy between
the results from the above studies may be caused by the different
fermentation and procedures used. Viegas et al. (1985) suggested
that the mechanism of soy flour enhancement of the fermentation
performance was nutrient satisfaction. Since all the treatments
presented in Table 1 were supplied with adequate urea as a nitro-
gen source, the rate-promoting effect of skim cannot be solely
accredited to excess nitrogen in the media. The easily utilizable
nitrogenous matters (i.e., the nature of nitrogen sources) for yeast
and possibly the minor nutrients, such as minerals and vitamins
present in soy skim, are likely the main reasons for the rapid fer-
mentation. Soy skim contained considerable amounts of free ami-
no acids and small peptides, some of which were the amino acids
favorably utilized by yeast (Yao et al., 2011). These amino nitrogen
Fig. 3. Effects of water-to-solids ratio on the maximum CO2 production rate of the control, 20% and 100% skim replacements. The means within each skim replacement group
followed by different letters are significantly different with p < 0.05 (n = 3 for control, n = 2 for 20% and 100% skim). Error bars represent standard deviations.
sources may be utilized faster by yeast (Kitagawa et al., 2008), en-
able the yeast to consume the carbon skeleton from amino acids
(Thomas and Ingledew, 1990), and increase protein biosynthesis
in yeast cells (Jimenez-Marti and del Olmo, 2008).
3.2. Effect of water-to-total solids ratio on the fermentation
performance
The water-to-solids ratio affects fermentation efficiency in typ-
ical corn fermentations (Wang et al., 1999). If the amount of dry
corn was kept the same in corn–soy co-fermentation and more
water was replaced with the skim, the water-to-solids ratio would
decrease. Thus, optimizing the water-to-solids ratio in corn–soy
co-fermentation is needed. Since it is not necessary to study the ef-
fect of water-to-solids ratio on each of the five skim replacement
levels, we chose 20% and 100% skim replacements. With 20% skim
replacement, more corn can be used compared to 100% skim and,
thus, more total ethanol can be produced per batch given each
replacement has a fixed water-to-solids ratio. At the 100% skim
replacement, more skim from soybean EAEP can be utilized and
higher protein content in DDGS can be obtained but with the sac-
rifice of ethanol yield per batch.
The statistical results from the 2-way factorial design experi-
ment showed that the fermentation rate obtained at 2.4 water-
to-solids ratio was significantly lower than those at higher
water-to-solids ratios when 20 and 100% skim replacement were
used (Fig. 3). The water-to-solids ratio did not considerably affect
lmax;CO2 when skim was not used. No significant evidence was
found for rate differences among 3.0, 3.5 and 4.0 water-to-solids
ratios. Higher than 3.0 water-to-solids ratio gave satisfactory fer-
mentation rates when soy skim was added. Nevertheless, one can-
not conclude that 3.0, 3.5 or 4.0 are the optimum water-to-solids
ratios to use in corn–soy co-fermentation system simply because
it is not cost-effective. As less corn is used, less ethanol will be pro-
duced because soy skim has a limited amount of fermentable sug-
ars. Therefore, economics studies will have to be done to determine
the optimum water-to-solids ratio. Regardless of the water-to-
solids ratio used, 20% and 100% skim replacements gave similar
fermentation rates, however, both of which were much higher than
that of the control. This result further confirmed the advantages of
soy skim on the fermentation rate improvement during corn–soy
co-fermentation. The total ethanol yields were not significantly af-
fected by the water-to-solids ratios.
 L. Yao et al. / Bioresource Technology 120 (2012) 140–148
3.3. Effect of urea nitrogen on the co-fermentation performance
Yao et al. (2011) reported that soy skim with 11% solids produced
from EAEP provided adequate nitrogenous nutrients for yeast
growth. We hypothesized that soy skim not only could replace the
water used in corn fermentation but also could provide the neces-
sary nitrogen when the 100% replacement was used. Fig. 4 shows
that fermentations with or without urea gave the same rates when
the 100% skim replacement was used; however, the 20% skim
replacement without urea gave significantly lower rate than the cor-
responding treatment with urea, which indicated that the 20% skim
replacement had inadequate nitrogenous nutrients for yeast growth.
This agrees with our earlier observation that the yeast growth rate
decreased and less cell biomass was produced when soy skim was di-
luted because the diluted skim did not contain enough utilizable
amino nitrogen to support normal yeast growth (Yao et al., 2011).
The control without soy skim did not give measurable CO2 when urea
was not added. No significant difference in total ethanol yield was
found between the treatments with or without urea.
3.4. Corn–soy co-fermentation with skim retentate from nanofiltration
The purpose of this experiment was to use a concentrated EAEP
soy skim in the co-fermentation and thereby maximize EAEP skim
utilization and further increase the protein content of the DDGS.
Experiments using nanofiltered skim showed that the 15% conc.
treatment gave the highest fermentation rate (Fig. 5). Fermentation
rates decreased when the skim retentates had solids contents
>15%. The 27% conc. sample had about the same fermentation rate
as the 100% skim. The nanofiltered (diluted and undiluted) skims
gave significantly higher rates than the control. Seemingly, there
is an optimum percentage solids content of skim that leads to
the fastest fermentation. With media having glucose as the sole
carbon source, soy skim with about 13% solids gave the highest
yeast growth rate (Yao et al., 2011), which also suggested an
optimum solids content of skim. The presence of an optimum skim
solids content may be attributed to several factors including min-
eral or salt content of the sample. No difference in final ethanol
yield was found among these treatments even though some treat-
ments had much faster rates of fermentation than others.
Fig. 4. Effects of urea addition to 20% and 100% skim replacements on the maximum CO2 production rate. The means in each skim replacement followed by different letters
are significantly different with p < 0.05 (n = 2). Error bars represent standard deviations.
145
3.5. Comparison of three methods of determining ethanol yield
Since there were no significant differences in ethanol yield
(by all three quantification methods) among various treatments,
ethanol yields from all the fermentation runs were treated as rep-
lications to examine whether there was any discrepancy among
the three ethanol determination methods. The AFM method
where ethanol yield was based on released CO2 gave significantly
lower values than the other two methods. The mean difference
between AFM and HPLC methods was 3.45 g per 100 g of dry corn
with a standard error of 0.16, the difference between AFM and
weight loss methods was 4.23 g per 100 g of dry corn with a stan-
dard error of 0.13. The weight loss method gave slightly higher
values than the HPLC method where the difference was 0.78 g
per 100 g of dry corn with a standard error of 0.17. The correla-
tions between the AFM method and the other two methods were
significant (p < 0.0001). The linear regression model of HPLC (y)
and AFM (x) methods can be describe in an equation of
y = 13.76 + 0.69 x, with a mean square error (MSE) of 1.38,
whereas the relationship between weight loss (y) and AFM (x)
methods can be described in an equation of y = 14.13 + 0.703 x,
with a MSE of 1.07. Since the ethanol data were distributed in a
relatively narrow range of 29 to 40 g per 100 g of dry corn from
a statistical point of view, the R square, which has often been
used to determine how well the regression model fits the data,
tended to be low with this type of data and were replaced by
MSE as reported here. In addition, the AFM method gave the least
number of outliers as shown in Fig. 6 and the standard deviation
of 102 ethanol data points from AFM was the lowest among the
three methods. Considering the main purpose of using the AFM
instrument is to facilitate parameter screening for the ethanol fer-
mentation experiments, AFM provided reasonably consistent and
informative data for such studies although ethanol yield was
underestimated. Since AFM monitored CO2 production by mea-
suring the flow rate of the output gas, it is possible that the lower
values were caused by inappropriate flow rate calibration of the
new instrument.
Fermentation efficiency (the ratio between ethanol yield and
theoretical ethanol yield) of all the treatments performed in the
present study was not significantly affected by the fermentation
conditions (p > 0.05), and ranged from 85% to 95%.
146 L. Yao et al. / Bioresource Technology 120 (2012) 140–148

Fig. 5. Maximum CO2 production rate of corn–soy co-fermentations with nanofiltered skim. The means followed by different letters are significantly different with p < 0.05
(n = 2 for treatments 15% conc. and 20% conc., n = 3 for other treatments). The data for the control and 100% skim were not the same as those in Fig. 3 because new runs were
performed to enable randomized block design. Error bars represent standard deviations.

Fig. 6. Ethanol yields calculated by using AFM, HPLC, and weight loss methods.

3.6. Chemical composition of soy-enhanced whole stillage solid
The proximate analyses of the whole stillages (actual measure-
ments were made on the beers and the total solids) showed that
the values from the control were not substantially different from
those of typical DDGS except that the total residual sugar content
was out of the range (Liu, 2011). The low values of residual sugar
contents probably resulted from the experimental scale performed.
The values were similar to those reported by Srichuwong and Jane
(2011) who also used a laboratory scale so that the fermentation
was well controlled and more complete, whereas the higher values
reviewed by Liu (2011) were of the DDGS from commercial ethanol
plants. Compared to the control, the soy-enhanced whole stillage
(SEWS) had increased solids, protein, and ash contents with the
addition of soy skim, whereas its oil and fiber contents decreased
(Table 2). The composition at each skim replacement level was dif-
ferent from those reported by Yao et al. (2011) because the ratios
of water-to-solids used in the two studies were not the same; how-
ever, their general trends were similar. Compared with the control,
solids and protein contents of SEWS from the 100% skim replace-
ment increased by 32% and 62%, respectively. Its ash content was
almost 2-fold higher than that of the control and fiber content
was reduced by 54%. The decrease in fiber content was mainly be-
cause less corn was used to maintain proper water-to-solids ratio,
and consequently, the reduction of fiber in this study was accom-
panied by less total ethanol produced per batch. If one wants to
increase total ethanol output, more corn should be added and
the fiber content of the resulting whole stillage will increase.
Further decreases in fiber content may be achieved by improved
saccharification techniques and better selection of yeast strains.
 L. Yao et al. / Bioresource Technology 120 (2012) 140–148 147

Table 2
Proximate analysis of the finished beera.

Component % Skim replacement

0b 20 40 60 80 100
Solids (%) 12.48 ± 0.63c 12.98 ± 0.72c 14.30 ± 0.10b 14.65 ± 0.81b 15.76 ± 0.23a 16.47 ± 0.24a
Protein (% dwb) 28.69 ± 2.83d 34.89 ± 3.33d 38.65 ± 1.43c 41.34±1.07bc 43.02 ± 1.22b 46.56 ± 3.89a
Oil (% dwb) 12.32 ± 0.86a 12.32 ± 1.23a 11.99 ± 0.96a 11.28 ± 1.15ab 11.03 ± 0.54b 10.32 ± 0.98b
Total carbohydrates (% dwb) 54.87 ± 3.51a 47.27 ± 4.42b 43.23 ± 1.85c 40.24 ± 2.61cd 38.42 ± 1.68d 35.14 ± 3.96d
Total residual sugar (% dwb) 1.92 ± 0.14a 1.40 ± 0.18b 0.89 ± 0.06c 0.76 ± 0.03c 0.65 ± 0.06c 0.63 ± 0.08c
Total fiber (% dwb) 52.95 ± 4.30a 45.87 ± 4.63b 42.34 ± 2.43bc 39.48 ± 2.54cd 37.77 ± 1.99cd 34.51 ± 4.51d
Ash (% dwb) 4.12 ± 0.09e 5.52 ± 0.21d 6.28 ± 0.22c 7.15 ± 0.17b 7.62 ± 0.25ab 8.04 ± 0.18a
a
The finished beer were from the treatments that had a fixed water-to-solids ratio of 2.4 and used coarsely ground corn as described in Section 2.5. Mean ± S.D. followed by
different letters within each row are significantly different with p < 0.05 (n = 3).
b
Control.

SEWS from the 100% skim replacements has almost the same pro- ers (Lumpkins et al., 2004), 30% in nursery and grow-finish pigs
tein and oil contents as soybean meal, which currently is the most diets (Stein and Shurson, 2009), 30% in dairy cattle (Kalscheur,
suitable supplemental protein for swine and poultry feed and typ- 2005), and 20% in beef cattle (Huls et al., 2008), had no detrimental
ically contains 44–48% protein and 1.5% of oil (Shurson, 2006). effect on growth performance. Using SEWS may lead to an increase
Soy protein is rich in lysine, the main limiting amino acid in cer- of the proportion of ethanol fermentation co-product that can be
eal grains as well as in traditional DDGS. Incorporation of soy skim included in swine and poultry diet formulations as well as rumi-
to the corn ethanol fermentation resulted in SEWS with higher ly- nant feed.
sine content (Table 3). SEWS from the 100% skim replacement con-
tained lysine at 3.6% of the total amino acids, which was a 126% 4. Conclusion
increase from the control although it was still lower than that of
soybean meal (6.4%). Other amino acids with elevated ratios in The promoting effect of soy skim on the fermentation rate of
SEWS included aspartic acid, glutamic acid, arginine, isoleucine, corn–soy co-fermentation was confirmed. Such co-fermentations
and tryptophan. Tryptophan, the second limiting amino acid in provide a viable solution to the downstream treatment of soy skim
swine diets, increased by 107%. fraction from EAEP, and will alleviate concerns of water use for
The application of regular DDGS is usually limited to ruminant both ethanol production and EAEP. The evaluation of the effect of
diets because of the variation in the nutrient composition, low en- various fermentation conditions including skim addition level,
ergy and high fiber content, and poor protein quality such as lysine water-to-solids ratio, and the requirement for nitrogen source on
deficiency (Batal and Bregendahl, 2012). Having balanced amino the co-fermentation provides directions for further scale-up and
acids and high protein digestibility is important to swine and poul- process optimization. The compositional data of soy-enhanced
try diets (Shurson, 2006), which makes SEWS attractive for uses in whole stillage can be used as the basis for the feed formulation
both ruminant and non-ruminant feeds. It is also worthy to men- for animal nutrition studies.
tion that the protein content of SEWS from the 100% skim replace-
ment was more than twice of that of high protein DDGS, which is Acknowledgements
now available on the feed market (Shurson, 2006). Animal feeding
studies have suggested that inclusion of DDGS up to 5–8% in broil- This research was funded by a USDA NIFA Grant (Contract No.
2010-34432-20955). We also thank the consulting group in the
Department of Statistics at Iowa State University and Dr. Hui Wang
Table 3 for acquiring and grinding the corn in the pilot plant of the Center
The amino acid profiles (wt.%) of soy-enhanced whole stillage solidsa.
for Crops Utilization Research.
Amino acid % Skim replacement
0b 20 40 60 80 100 References
Aspartic acid 7.07 8.04 8.91 9.40 9.62 9.97
Batal, A.B., Bregendahl, K., 2012. Feeding ethanol coproducts to poultry. In: Liu, K.,
Threoninea 4.39 4.44 4.36 4.27 4.26 4.21
Rosentrater, K.A. (Eds.), Distillers Grains: Production, Properties, and Utilization.
Serine 4.84 4.82 4.55 4.40 4.46 4.37
CRC Press, Boca Raton, FL, pp. 317–338.
Glutamic acid 14.49 15.36 16.04 16.28 16.54 16.92 Damiano, D., Wang, S.S., 1985. Improvements in ethanol concentration and
Proline 8.02 7.65 7.35 7.10 7.03 6.39 fermentor ethanol productivity in yeast fermentations using whole soy flour
Glycine 4.61 4.67 4.78 4.80 4.72 4.69 in batch and continuous recycle systems. Biotechnol. Lett. 71, 135–140.
Alanine 7.73 7.03 6.51 6.19 5.99 5.84 Hayashida, S., Feng, D.D., Hongo, M., 1974. Function of the high concentration
Cysteine 2.08 1.94 1.89 1.81 1.78 1.73 alcohol-production factor. Agric. Biol. Chem. 38, 2001–2006.
Valinec 5.95 5.72 5.80 5.88 5.74 5.71 Huls, T.J., Luebbe, M.K., Erickson, G.E., Klopfenstein, T.J., 2008. Effect of inclusion
Methioninec 2.28 2.05 1.89 1.78 1.72 1.67 level of modified distillers grains plus solubles in finishing steers. Nebraska Beef
Isoleucinec 4.62 4.66 4.87 5.03 4.98 5.03 Cattle Rep. 36–38.
Leucinec 13.14 11.87 10.86 10.14 9.85 9.53 Jimenez-Marti, E., del Olmo, M., 2008. Addition of ammonia or amino acids to a
Tyrosine 4.69 4.54 4.35 4.20 4.24 4.23 nitrogen-depleted medium affects gene expression patterns in yeast cells
Phenylalaninec 6.20 5.98 5.77 5.68 5.61 5.52 during alcoholic fermentation. FEMS Yeast Res. 8, 245–256.
Ju, N., Damiano, D., Shin, C., Kim, N., Wang, S.S., 1983. Continuous ethanol
Lysinec 1.60 2.03 2.39 2.99 3.18 3.61
fermentation of Zymomonas mobilis using soy flour as a protective agent.
Histidinec 2.83 2.71 2.71 2.61 2.67 2.66
Biotechnol. Lett. 5, 837–842.
Arginine 4.93 5.65 6.12 6.46 6.61 6.81
Kalscheur, K., 2005. Impact of feeding distillers grains on milk fat, protein, and yield.
Tryptophanc 0.54 0.82 0.84 1.00 1.01 1.12 Proceeding of the Distillers Grains Technology Council. 10th Annual
a Symposium, Louisville, KY.
The stillage were from the treatments that had a fixed water-to-solids ratio of
Kitagawa, S., Mukai, N., Furukawa, Y., Adachi, K., Mizuno, A., Iefuji, H., 2008. Effect of
2.4 and used coarsely ground corn as described in Section 2.5.
b
soy peptide on brewing beer. J. Biosci. Bioeng. 105, 360–366.
Control. Lumpkins, B.S., Batal, A.B., Dale, N.M., 2004. Evaluation of distillers dried grains with
c
Essential amino acids. solubles as a feed ingredient for broilers. Poult. Sci. 83, 1891–1896.
148 L. Yao et al. / Bioresource Technology 120 (2012) 140–148
Liu, K., 2011. Chemical composition of distillers grains, a review. J. Agric. Food
Chem. 59, 1508–1526.
Maia, A.B.R.A., Nelson, D.L., 1994. A comparative study of effects of soy and corn
flours on the evolution of alcohol fermentation in successive batches. J. Chem.
Technol. Biotechnol. 59, 171–179.
de Moura, J.M.L.N., Maurer, D., Jung, S., Johnson, L.A., 2011. Pilot-plant proof-of-
concept for countercurrent two-stage enzyme-assisted aqueous extraction
processing of soybeans. J. Am. Oil Chem. Soc. 88, 1649–1658.
Naidu, K., Singh, V., Johnston, D.B., Rausch, K.D., Tumbleson, M.E., 2007. Effects of
ground corn particle size on ethanol yield and thin stillage soluble solids. Cereal
Chem. 84, 6–9.
Reddy, L.V.A., Reddy, O.V.S., 2006. Rapid and enhanced production of ethanol in very
high gravity (VHG) sugar fermentation by Saccharomyces cerevisiae: role of
finger millet (Eleusine coracana L.) flour. Process. Biochem. 41, 726–729.
Rathore, S.S.S., Paulsen, M.R., Sharma, V., Singh, V., 2009. Optimization of yeast and
enzyme dose for dry-grind corn fermentation process for ethanol production.
Trans. ASABE 52, 867–875.
Srichuwong, S., Jane, J., 2011. Methods for characterization of residual starch in
distiller’s dried grains with soluble (DDGS). Cereal Chem. 88, 278–282.
Shurson, J., 2006. The value of high-protein distillers coproducts in swine feeds.
Distillers Grains Quarterly First Quarter 2006, 22–25.
Stein, H.H., Shurson, G.C., 2009. Board-invited review: the use and application of
distillers dried grains with solubles in swine diets. J. Anim. Sci. 87, 1292–
1303.
Thomas, K.C., Ingledew, W.M., 1990. Fuel alcohol production: effect of free amino
nitrogen on fermentation of very-high-gravity wheat mashes. Appl. Environ.
Microbiol. 56, 2046–2050.
Vidal Jr., B.C., Rausch, K.D., Tumbleson, M.E., Singh, V., 2009. Protease treatment to
improve ethanol fermentation in modified dry grind corn processes. Cereal
Chem. 86, 323–328.
Viegas, C.A., Sá-Correia, I., Novais, J.M., 1985. Nutrient-enhanced production of
remarkably high concentrations of ethanol by Saccharomyces bayanus through
soy flour supplementation. Appl. Environ. Micorbiol. 50, 1333–1335.
Wang, S., Ingledew, W.M., Thomas, K.C., Sosulski, K., Sosulski, F.W., 1999.
Optimization of fermentation temperature and mash specific gravity for fuel
alcohol production. Cereal Chem. 76, 82–86.
Wang, H., Wang, T., Johnson, L.A., 2009. Effect of low-shear extrusion on corn
fermentation and oil partition. J. Agric. Food Chem. 57, 2302–
2307.
Yao, L., Wang, T., Wang, H., 2011. Effect of soy skim from soybean aqueous
extraction processing on the performance of corn ethanol fermentation.
Bioresour. Technol. 102, 9199–9205.