BIOTECHNOLOGY AND BIOENGINEERING

VOL. X, PAGES 845-864 (1968)

Kinetics of Product Inhibition

in Alcohol Fermentation
S. AIBA, 31. SHODA, and 11. SAGATAXI, Institute of Applied
Microbiology, Cniversify of Tokyo, Tokyo, J a p a n

Summary
The inhibitory effect of ethanol concentration p in a medium on the specific
rates of growth p and ethanol production v of a specific strain of bakers yeast was
studied in a chemostat, where except for ethanol as the product, only the concentration of glucose S was controlled t o limit the metabolic activity of the yeast.
This was designed t o supplement the previous findings from the batch experiment,
in which ethanol was added artificially and no substrate components were limiting
the metabolism of the same yeast, that p = p ~ e - ~ i pand v = v ~ e - ~ z pwhere
,
kl
and l i z are empirical constants and subscript the 0 denotes respective values a t
p = 0. The effects of p on the values of p and v were confirmed by the Lineweaver-Burk plot to belong t o noncompetitive inhibition. The formulas here for
p and v as affected by p , if extrapolated t o the case of no limiting substrates,
were in good agreement in respective forms with those derived previously from
the batch experiment, though the values of corresponding coefficients in these
formulas were different. The differential equations for p and v as functions of
both p and S and, in addition for the rate of glucose consumption as correlated
by the yield factors either with the cell growth rate or the rate of ethanol production, were solved properly with a digital computer. A kinetic pattern calculated
so far was discussed with reference to the data obtained in the batch experiment
and those relevant to actual sake brewing.

INTRODUCTION

It is well known that in an alcohol fermentation the accumulation

of ethanol continues incessantly even after a cessation of cell growth
at a level of about lo8 cells/ml. I t is not unusual in this stationary
phase to keep the temperature of a fermenting broth lower-e.g.,
around l5OC in the brewing of sake to prevent an inactivation of
the yeast cells due to the ethanol produced.
The fact that ethanol is inhibitory to the metabolic activity of the
845

846

S. AIBA, M. SHODA, M. NAGATANI

yeast cells is clear, if only a qualitative description is of interest.

The inhibitory effect of ethanol on the yeast cell growth is apparent
a t a relatively lower concentration, i.e., several per cent of ethanol,
while the fermentation activity of the cells seems to have tolerance
until ethanol approaches about 20% for example.
Despite a seemingly discrete effect of ethanol on yeast metabolism,
its inhibitory effect remains to be discussed in quantitative terms
from an aspect of kinetics. This implies that the effect should be in
continuous rather than in discrete patterns, provided the kinetics on
growth and ethanol production activities of yeast cells can be assessed
properly. It is difficult, indeed, to find any reference in which the
kinetic pattern of an alcohol fermentation is discussed extensively.
In this context, Hinshelwood derived from his experiments with
Bact. lactis aerogenes the following equation to describe the effect of
alcohol added into a growing culture.'
P = Pm(1 -

ap)

(1)

provided p,,, = maximum value of specific growth rate; p = concentration of alcohol; a = an empirical constant, depending presumably
on the species of alcohol added.
Recently, Holzberg et al. studied the effect of ethanol produced by
yeast cells on its rate of product (ethanol) formation in a fermentation
of grape juice.2 The region they studied was in a stationary phase of
the cell growth. The empirical equation presented by them is as
f 0llO~Vr-S
:
dp/dt

B(pm - p )

(2)

where B = an empirical coefficient; p m = the maximum value of

ethanol ever attainable.
The fact that neither the term of cell number nor that of cell
growth is found in the above equation originates from the fact that
alcohol in this instance is produced in a period lasting much longer
than that experienced from an inoculation until a cessation of the cell
growth. It is difficult from this approach to correlate the rate of
ethanol production with that of cell growth.
Now it is evident that one cannot establish a kinetic pattern of
product inhibition in alcohol fermentation unless the rates of cell
growth and ethanol production are studied regarding the case where
only one component of the original substrates becomes limited.
BIOTECHNOLOGY AND BIOENGINEERING, VOL. X, ISSVE 6

PRODUCT INHIBITION

847

A continuous fermentation in single vessels is used here to facilitate

the experiment in which glucose and only glucose in an original
culture medium limits the metabolic activity of the yeast cells. The
purpose of this work is to assess the inhibitory effect of ethanol on
the rates of cell growth and ethanol production such that one can gain
an access to the product inhibition kinetics both in continuous and in
batch runs of alcohol fermentation.

MATERIALS AND METHOD

Strain
The strain used was a respiration-deficient mutant of a bakers
yeast, H-1 (Japan Sugar Refinery Ltd., Yokohama), a protoplasmic
mutant of a tetraploid organism.
A principal reason for having adopted the specific strain was for
convenience of experimentation, because no strictly anaerobic conditions in the experiment were required oli-ing to the use of this strain.

Culture Media
The composition of the medium used is shown in Table I. The
concentration of glucose in the fresh medium was controlled a t all
times to l-2yo to realize the condition of glucose-limiting in a
chemostat in this work. This criterion of 1-2y0in the glucose concentration was predetermined from the preliminary batch experiments
that the values of specific growth rate p of the cells decreased appreciably when the concentration of the remaining glucose became lower
than this figure, provided other components, especially the yeast
extract in the original medium, were the same prescription as that in
Table I.3 Ethanol was added into the fresh medium to different
TABLE I
Composition of Culture Medium Used
Glucose
KHzPOa
(NH,)zSO4
M,SOI. 7H20
Yeast extract
Tap water

10 or 20 g
j g
2g
0.4 g
2g

1000 ml
pH 4.0

848

S. AIBA, M. SHODA, hf. NAGATANI

concentrations in each case to study the inhibitory effect of ethanol,

because ethanol due to be produced from the original medium
(1-2y0 in glucose concentration) was too dilute to produce the
adverse effect .
The medium composition used for precultivating the cells a t 30C
for about 1.5 hr in shaken flasks (rotary type; each working volume =
2 liters) was the same as in Table I except that glucose was increased
from the original @yo)
to .5yo.

Analytical Method
Cell Mass. The optical density was measured a t a wavelength of
610 mF with a photometer (Node1 FPW-4, Hitachi Works, Tokyo);
the value was converted to that of dry cell mass per unit volume of
the medium (grams per liter) by using a linear calibration.
Glucose. The Somogyi method or the Somogyi-Kelson method
was used for the analysis of glucose. The latter method was adopted
when the glucose concentration was less than about 0.2 g/l. A
spectrophotometer used for the analysis a t a wavelength of 320 mfi
was model 101 from Hitachi Works, Tokyo.
Ethanol. The oxidation method by dichromate was used for the
determination of ethan01.~ Clearly, the ethanol to be analyzed from
time to time by samplings (see below) was a sum total of that which
was preadded into the medium and that which was produced in the
continuous fermentation.
Experimental Procedure
The chemostat used is shown schematically in Figure 1. The
body of a fermentor (nominal volume = 30 liters; working volume =
5 or 10 liters) is a cylindrical glass; the lid and the bottom, both of
which were made of stainless steel were bolted together in assembling.
The fermentor was equipped with four baffle plates and the broth in
the vessel was rotated by a standard flat-blade turbine a t 200 rpm
throughout each run. The vessel n-as fixed in a water bath, the temperature of which xvas adjusted to 30C by a controller as shown in
Figure 1, while the pH value of the broth was controlled a t 4.0.
The medium reservoir was made of stainless steel (nominal volume
= 32 liters; working volume = 30 liters of fresh and sterile medium
impregnated appropriately by ethanol, if required).
BIOTECHNOLOGY AND BIOENGINEERING, VOL. X, ISSUE 6

849

PRODUCT INHIBITION

14

P
I

2
3

I 10 I PIPE FOR ALKALI

1 FERMENTOR
I FLAT-BLADE TURBINE I I I I THERMOMETER

ADD.

~~

SAMPLING PIPE
BAFFLE PLATE

12 HEATER
13 TEMP. CONTROLLER

5
6

WATER BATH
DH ELECTRODE

14
15

7
8

I ALKALI

pH CONTROLLER

I ELECTRIC

I 16 I

VALVE

RESERVOIR

17

1
1

PLUNGER PUMP
MEDIUM RESERVOIR
PIPE FOR MEDIUM ADD.
SAMPLING FLASK

Fig. 1. Schematic diagram of experimental apparatus in continuous system.

A principal part of the plunger pump was made of glass (Type

T-63K, Sanyo Scientific Inst. Co., Ltd., Tokyo); the feed rate was
rather small (0.<5-1.2 l/hr).
Due attention was paid to guarantee a constant feed rate by a
direct measurement from time to time in each run. The maximum
fluctuation of the feed rate ever experienced was about f 1 0 7 G . If
the feed rate exceeded the allowable limit, its stroke was adjusted.
The vessel was sterilized by blowing live steam (100C for 1 hr)

850

S. AIBA, M. SHODA, M. NAGATANI

before initiating each run of the continuous fermentation. A vinyl

pipe for transporting the medium and the pH and reference electrodes
in Figure 1 were also sterilized with either live steam or with an
aqueous solution of ethanol (SOTo v/v).
In addition, the fresh medium (30 1) in the reservoir was placed
en bloc in an autoclave to effect sterilization a t 120C for 10 min; after
cooling the reservoir by exposing it to a running tap water, pure
ethanol was added into the medium, if needed.
A preculture (2 or 5 liters) was charged into the vessel, followed by
an addition of the sterile medium from the reservoir. The liquid
volume in the fermentor was adjusted to either 5 or 10 liters, depending on the dilution rate employed in each series of experiments.
Then a continuous run was started by rotating the impeller and by
actuating the plunger pump. The liquid was discharged constantly
through an overflow pipe as is clear from Figure 1.
After 20-24 hr a t a constant feed rate in each run, the sampling
was begun every 1-2 hr to confirm whether or not the steady state
could be attained. If the values of residual glucose and the yeast
cells determined by several samplings neither increased nor decreased
beyond each experimental error of measurement (less than 5Yo), the
steady state was assumed. Ethanol concentration in the medium
was then determined by the oxidation method.
The procedure of samplings used here must be mentioned briefly.
A long time was required to accumulate the flowing medium out of
the fermentor (see Fig. 1) in the case of smaller dilution rates (see
Table 11). For this particular case, a sampling flask in Figure 1 was
replaced uith a measuring cylinder, into which an aqueous solution
(20 ml) of 0.1% mercuric chloride had been charged. The medium
from the overflow pipe was discharged into the cylinder till the
liquid volume amounted to 45 ml. This was followed immediately
by neutralization with an aqueous solution of sodium hydroxide
(40%) first and, second, by supplemental distilled water to a final
volume of liquid exactly equal to 50 ml. The cells could be inactivated during this sampling; after separating the cells by centrifugation the supernatant liquid u-as subjected to a spectrophotometric
determination of the residual glucose. An existence of the fungicide
was confirmed from preliminary experiments to have no adverse
effect on the measurement.
BIOTECHNOLOGY AND BIOENGINEERING, VOL. X. ISSUE 6

PRODUCT INHIBITION

851

TABLE I1
Experimental Data of Steady State in Continuous Alcohol Fermentation
pa, g/l

X, g/!

7.97
13.7
21.3
28.4
42.7
57.0

0
7.86
15.1
21.9
36.9
53.3

2.0
2.22
2.05
2.08
2.12
1.87

0.334
0.221
0.254
0.263
0.230
0.166

10.9
9.99
9.58
10.7

4.7
19.3
33.4
51.5

0
15.5
30.4
48.3

1.20
1.37
1.40
1.22

0.392
0.278
0.214
0.262

0.138
0.194
0.326
1.63

21.2
21.1
21.6
20.4

8.57
22.7
40.0
55.4

0
15.3
32.6
48.6

2.40
2.43
2.30
1.68

0.571
0.487
0.516
0.648

0.186
0.723
1.82

20.7
20.7
20.2

8.44
24.1
37.2

0
16.3
29.8

2.33
2.14
2.03

0.717
0.729
0.721

0.226
1.09

10.8
10.4

4.51
18.7

0
15.0

1.25
0.919

0.875
0.974

D=~,hr-'

S,gA

So, g/l

p, g/l

0.084

0.054
0.096
0.122
0.127
0.118
0.212

21.5
19.5
19.9
20.5
20.9
20.5

0.079
0.091
0.114
0.115

"

<'
I'

"

'
0. loo
"

'
0.160
"

'1

0.198
"

0.242

'

9,

hr-I

As for a larger value of the dilution rate, there was no necessity of

this particular procedure; each sampling \vas made directly into a
flask by 10 ml.
After the data required were taken in each run, another run of eontinuous fermentation was started without changing the feed rate.
A series of runs of the continuous experiments was repeated without
interruption a t a specific feed rate with different concentrations of
ethanol in the fresh medium flowing into the vessel.
The fermentor with its auxiliary equipment was disassembled for
cleaning and sterilizing before another series of experiments with
different values of the feed rate could be initiated.

852

S. AIBA, hZ. SHODA, 11. NAGATANI

RESULTS AND DISCUSSION

The experimental data are summarized in Table 11. The values
of specific rate of ethanol production u in Table I1 were calculated
from the data ( D , X , p , and po) in each run as follows:
u =

D(p

p,)ix

(31

The value of specific growth rate p of the cells is equal in this study
to each dilution rate. Most of the figures and discussion which will
appear later on originate from the data in Table 11.
The values of S are plotted against p in Figure 2 , the parameter
being the dilution rate, D (=p). Broken curves through the data
points deviate markedly from the curves, presumably due to experimental errors inherent especially in the smaller values of dilution
rate. It is difficult to discern directly from the figure to \\hat
extent ethanol did affect adversely the growth rate of the cells. A
symptom to this effect recognizable from Figure 2 is that the values of
S in a larger range of p more than 20 g/1 for example, depended
appreciably on the values of D; the appearance is similar to that
observable when n-ashout occurs. Otherwise, the curves should
have been closer together as can be seen near the origin of Figure 2 .
The Lineneaver-Burk plot n a s made in Figure 3 to rectify an
equivocal situation of the inhibition in Figure 2 . It must be remarked here t h a t the values of S pertaining to each value of p selected
for convenience a t round numbers were estimated from each broken
curve in Figure 2. It is evident from the plot in Figure 3 that the
inhibition is noncompetitive, because each data point at constant
value of p is on a straight line and, in addition, these straight lines
converge eventually a t a point 011 the abscissa of Figure 3.j It was
not difficult to draw these straight lines starting from the same point
on the abscissa, even if one pays particular attention t o specific points
which deviate markedly from respective broken curves constructed
in Figure 2 .
The value of a saturation constant, K , in this correlation can be
estimated from Figure 3 as K , = 0.22 gtl. The effect of ethanol
on the cell gron-th is then manifested in the maximum reaction
(growth) rate in each concentration of ethanol. Each maximum
value of the specific growth rate could be estimated by the intercepts
of the straight lines from the ordinate of Figure 3. The values of p
BIOTECHNOLOGI .IND BIOENGINEERING, VOL. X, ISSUE 6

853

PRODUCT INHIBITION
2.c

1
I
)=0.242

0=0.160

D=y.l9

I"

I
I
I

4I

1:.

I
I
I
I

I
I

I
I

5
-m

v)

1.c

I
I

+
I

0.5

I
I
I

I
/
/
/
I
/

D= 0.100//:

",/b
/'.
'0
<DL

0.084

I
3

Fig. 2 . S vs. p in the steady state (single vessels).

estimated from Figure 3 are plotted semilogarithmically against the

values of p as shown in Figure 4.
The experimental data (p versus p ) which 1%-ere
obtained separately
in the shaken-flask tests3 are cited in Figure 4. The values of p in
the shaken-flask tests were those pertaining to the logarithmic
growth phase of the cells which were not "limited" by any substrates.
It is seen from the figure that the data points in both cases are
correlated by straight lines on semilogarithmic paper.

854

S. AIBA, M. SHODA, M. NAGATANI

Fig. 3. Lineweaver-Burk plot;

l/p

vs. l/S in the steady state (single vessels).

Accordingly, the inhibitory effect of ethanol on the cell growth could

be formulated irrespective of the difference of operation (batch and
continuous) by t,he following equation:

provided: k, = an empirical coefficient; p o = specific growth rate

a t p = 0.
Equation (4)is assumed to represent the inhibitory effect of ethanol
in a batch fermentation when glucose becomes limiting, though no
experiment along this line was made. Also, the value of K , in eq. (4)
is assumed to be independent of operations, i.e., batch or continuous.
Discussion on these assumptions will be made later on in this work.
BIOTECHNOLOGY AND BIOENGINEERING, VOL. X , ISSUE 6

PRODUCT INHIBITION

855

Fig. 4. Specific growth rate p as affected by ethanol concentration p when

no substrates are limiting; for the continuous run the values of p were taken
from the intercepts of the Lineweaver-Burk plot in the ordinate of Fig. 3.

However, by using these assumptions, the point to differentiate the

continuous from the batch run in eq. (4)is in the value of kl. The
values of k, determined from Figure 4 are 0.016 and 0.028 g/l for the
continuous and shaken-flask tests, respectively.
The difference in the values of 10 in eq. (4)depending on the type of
operation (see Fig. 4) is attributed presumably to that of equipment
used: i.e., one, the shaken-flask, the other, the jar fermentor. The
above discussion regarding eq. (4) [cf. eq. (5) below] will be of use
from a kinetic approach to the scaleup.
The values of v cited from Table I1 are plotted against ethanol
concentration p in Figure 5 ; another parameter is the dilution rate D .
The values of v are expected from the figure to increase with an increase of dilution rate. However, it is difficult to assess the form of a
function to define the dependence of v values on the values of p in each
dilution rate. Therefore, the broken curves were prepared temporarily as shown in Figure 3 to estimate the value of v which could
correspond to any value of p other than that actually measured.
Unacceptable as this procedure may seem, it will become apparent
that this step is not accompanied by any grave error which totally
invalidates a conclusion here.

856

S. AIBA, M. SHODA, M. NAGATANI

Fig. 5 .

m. p in the steady state (single vessels).

The experimental data in Figure 5 were rearranged into the Lineweaver-Burk plot (l/v versus 1/S)in Figure 6. This plot was made
possible by reading the values of v from Figure 5 a t round numbers of
p arbitrarily selected with the use of respective broken curves in the
figure first and, second, by taking the corresponding values of S from
Figure 2 with further reference to the broken curves already established in the same figure.
BIOTECHNOLOGY A N D BIOENGINEERING, VOL. X, ISSUE 6

PRODUCT INHIBITION

857

Straight lines could be drawn easily through the data points as

shown in Figure 6 such that the lines converge to a point on the
abscissa. Then it is concluded that ethanol inhibition to the fermentation activity of the cells is also noncompetitive. The value
of Ks, the saturation constant, was estimated as 0.44 1/g from the
figure.
The intercepts of the straight lines from the ordinate in Figure 6
correspond to the values of v which are not limited by any substrates.
Consequently, the values of v estimated thus from Figure 6 \\-ere
plotted against ethanol concentration p in Figure 7. The values of
v could be correlated fairly well with those of p by an exponential
function, irrespective of the difference of experimental procedure.
The only difference is that of the initial value of v = v o a t p = 0 and
that of the slope of these lines in Figure 7.
The discussion mentioned earlier in Figures 3 and 4 for the specific
gron th rate p of the cells may also hold exactly for the values of v in
Figures 6 and 7. Then eq. (5) was derived to formulate the inhibitory effect of ethanol on the fermentation activity of the yeast
cells.

Fig. 6. Linen-eaver-Burk plot; 1,v vs. 1/S in the steady state (single vessels).

858

S. AIBA, M. SHODA, M. NAGATANI

p (94)

Fig. 7. Specific rate of ethanol production Y a8 d e c t e d by ethanol concentration p when no substrates are limiting; for the continuous run the values of Y
were taken from the intercepts of the Lineweaver-Burk plot in the ordinate of
Fig. 6.

provided: k , = an empirical coefficient; V O = specific rate of ethanol

production a t p = 0. The values of k z estimated from Figure 7 are
0.029 and 0.015 l/g for the continuous run and the Warburg respirometer, respectively.
One could calculate the values of yield factors Y P , sand Y x , ~from
the data shown in Table 11. The yield factors were plotted against
p in Figures 8 and 9. It is noted from these figures that both values
of yield factors remained nearly unchanged in the range of ethanol
concentration studied here. iilthough the data points scattered
considerably in Figures 8 and 9, the average values of YP,sand Y X , S
were taken as Y p I s= 0.35 and Y x l S = 0.10. The fact that the cell
growth was appreciable in this study could account for the lower
value of Yp,s compared with the theoretical value ( = 0.51).
The following equations which define either the yield factor, Yp,s
or another yield factor, YX/S must be used with eqs. (4)and (5) for
the calculation of a kinetic pattern of the yeast cells.

dS

--

dt

-1 d p YP,s dt

-1 d X
Y X I Sdt

A calculation of a batch alcohol fermentat.ion with a digital computer (HITAC 5020T, Computer center, University of Tokyo) by
BIOTECHNOLOGY AND BIOENGINEERING, VOL. X, ISSUE 6

PRODUCT INHIBITION

859

0.5

04

0.3

SYMBOL'

In

2
>
0.2

DILUTION RATE, D

0 1 9 8

0.I

10

20

I
30

40

I
50

C 3

p (g/Pl

Fig. 8. Y,,s vs. p in the steady state (single vessels).

10

20

30

40

50

60

p (Sd)

Fig. 9. Yx,s vs. p in the steady state (single vessels).

using eqs. (4), ( 5 ) , and (6) is shown in Figure 10; the ordinate represents either the values of S, p , or X , while the abscissa is fermentation time t. Data points described in the figure are bhe reproduction
of the batch fermentation with the jar ferrnent~r.~

S. AIBA, M. SHODA, 31. NAGATANI

860

20

10

(hr)

Fig. 10. Batch alcohol fermentation, observed and calculated. Xumerical

values substituted into respective coefficiennts of the kinetic equations (see the
text) are: k , = 0.028, k p = 0.015, p o = 0.408, y o = l.0,3K , = 0.22, K, = 0.44,
Y,,s = 0.35 ( I x , ~
= 0.10).

Solid curves in Figure 10 are the results of the computation. It

must be pointed out that the kinetic patterns could only be calculated starting from the initiation of the logarithmic phase to the period
when the substrate became nearly exhausted. The numerical values
substituted into the terms k 1 and k 2 in eqs. (4)and ( 3 ) for Figure 10
were obtained e~perirnentally;~
k l = 0.028, k2 = 0.015. The values
of po and v o used for the computation (see above the caption of Fig.
10) have no significant implication, n hereas the values of K,, Ks,
and Y, used \\-ere determined in this work.
-4pparently the calculation justifies that eqs. (4), ( 5 ) , and (6)
represent the kinetic pattern either in batch or in continuous operation, provided the values of k l and k2 in cqs. (4) arid ( 3 ) are duly
adopted, in particular.
BIOTECHNOLOGY AND B I O E S G I N E E R I N G , VOL. X, ISSUE 6

PRODUCT INHIBITION

861

This suggests that the kinetic pattern can be used to predict the
rate of ethanol production even after a practical cessation of the cell
growth, provided glucose is still available as in the brewing of sake,
for instance.
An example of the specific brewing of sake is cited in Figure 11.
The experimental data on which the figure was prepared came from
the Research Institute of Brewing, Tokyo. The ordinate of Figure 11
represents either the value of ethanol production rate d p l d t in a
logarithmic scale or the broth temperature, while the abscissa is
ethanol concentration, p (v/v 70)
of the fermenting broth. Various
symbols in the figure demonstrate the difference of runs, in which
various species of yeast including a specific yeast characterized by the
least amount of froth generation in the fermentation are employed.
The values of d p / d t were assessed graphically from the plot of p
versus time (days) ; incidentally, the period of fermentation for the
2,

l
0
L

w
c

Fig. 11. Example of sake brewing, accumulation of ethanol in the period after
a cessation of cell growth.

862

S. AIBA, M. SHODA, M. NAGATANI

data in Figure 11 was about 10 days. It must be mentioned here

that the values of p were taken from the original record of each
fermentation after the froth of the fermenting broth calmed down
onto the broth surface. Otherwise, an appropriate evaluation of
each ethanol production activity of the yeast cells becomes difficult,
because most of the yeast cells which tend to be adsorbed a t the
froth surface are due to loose contact with the broth during an
especially vigorous frothing period in each run, underestimating
thus far the value of ethanol production activity dp/dt of the cells in
question.
The cell growth in the fermenting period (Fig. 11) could be assumed
to be practically zero. An average pH value of the broth in the
figure ranged from 4.2 to 4.5, while the temperature was lowered with
the progress of each fermentation. Glucose, on the other hand, was
abundant in each run; in fact, the residual glucose a t the termination
of each run was about 1%. With reference to the value of K, = 0.44
g/l determined in this work, glucose was assumed not to have limited
the fermentation (Fig. 11).
It is interesting to find from the upper part of Figure 1 1 that the
solid curves prepared by hand through each set of the data points
were rather parallel. In particular, the curves could be assumed to
be straight lines in the semilogarithmic plot pertaining to a considerable range of p which started from p = 10% in the figure.
By using the assumption that glucose did not limit the fermentation, eq. ( 5 ) could be modified as follows:
dp/dt

voX e - k 2 p

(5)

Then,
log dp/dt = log VOX- 0.43kzp

(5j

Equation (5) suggests that a plot of dp/dt against p on semilogarithmic paper is shown by a straight line, the intersection of which
with the ordinate corresponds to the value of O X .
Apparently, each set of the data points in Figure 11 can be represented by eq. (5j for a wide range of p, starting from the origin of
the figure. Take, for example, the data points with an elliptic
symbol in the figure. Assuming that these data points are on a
straight line starting from p = 10 to p = 16y0in Figure 11, the
value of k2 could be estimated from its slope as 0.14s yo- = 0.019
BIOTECHNOLOGY AND BIOENGINEERING, VOL. X, ISSUE 6

PRODUCT INHIBITION

863

l/g [see eq. ( 5 ) ] . The point that the value of kz estimated

regarding the actual brewing of sake was near the value ever determined with the Warburg manometer3 (kq = 0.015 l/g; see
Fig. 7) seems interesting.
Taking the example further, the value of vOX could be calculated
from eq. (5) as 7.8yO/day = 61.5 g/l/day. If the value of Y O is
taken as 1.0 hr- (at 30C; see Fig. 7),3the value of X was estimated
as 2.56 g/l
7.7 x lo7 cells/ml. The value of v o for this example
of calculation may be much smaller than 1.O hr-I primarily due to the
lower temperature of operation (around 15C); however, another
point that the value of cell concentration X estimated above was in
the alleged order of magnitude in the alcohol fermentation seems
also interesting.
This illustration suggests that one can estimate from eq. (5) the
value of X in the actual sake brewing, if the value of d p / d t is
measured properly, and if the values of Y O and liq can be assumed most
appropriately. This approach may be of help from a standpoint of
control in the brewery.
Also, it is noted from Figure 11 that the data points deviate considerably from each straight line (not shown) assumed in the semilogarithmic plot. Although the factors responsible for this deviation
cannot be discussed definitely, the following factors might have been
contributing to this phenomenon.
One is the fact that the broth temperature was generally lowered
with the progress of each fermentation. Equation (5) was derived
by assuming implicitly that the operating temperature was constant.
The other is concerned with a possibility that the yeast cells might
have lost their viability with time in an environment of high ethanol
content. The latter possible factor responsible for the peculiar
degradation of fermentation activity of the cells around the end of
each run in Figure 11 is beyond the scope of eq. (5). With these
particular situations in mind, the kinetic equations discussed in this
study are considered promising from aspects of both basic comprehension and application of kinetics in an alcohol fermentation.

Nomenclature
a
B

empirical constant
empirical coefficient
dilution rate ( = F / V ) , hr-I

864

F
K,
K,'

k&
P

Po
P*
S

SO
t

I.
X
Y,,s
YX,S
@
YO
V
yo

S. AIBA, hl. SHODA, M. NAGATANI

feed rate, l/hr
saturation constant, g i l
saturation constant, g/l
empirical exponents, l/g
product (ethanol) concentration, concentration of various sorts of
alcohol, g/l, v / v %
ethanol concentration in fresh medium, g/l
maximum value of ethanol ever attained, g/l
substrate (glucose) concentration, g/l
substrate (glucose) concentration in fresh medium, g/l
time, hr, day
liquid volume, 1
dry cell mass concentration, g 11
yield factor ( = - A p / A S )
yield factor ( = - A X / A S )
specific growth rate, hr-1
specific growth rate a t p = 0, hr-'
specific rate of ethanol production, hr-I
specific rate of ethanol production a t p = 0, hr-'

References
1. C. N. Hinshelwood, The Chemical Kinetics of the Bacterial Cell, Clarendon,
Oxford, 1952, p. 105.
2 . I. Holzberg, R. K. Finn, and K. H. Steinkraus, Biotechnol. Bioeng., 9, 413
(1967).
3. M. Nagatani, M. Shoda, and S. Aiba, J . Ferm. Technol., in press.
4. J. F. Guymon and E. A. Crowell, J . rissoc. O$c. Agri. Chem. Baltimore,
42, 393 (1959).
5 . E. 9.Dawes, Quantitative Problems in Bzochemastry E. S. Livingstone,
London, 1962, pp. 99-154.

Received March 22, 196s

BIOTECHNOLOGT AND BIOENGINEERING, VOL. X, ISSUE 6