The vital role of citrate buffer in acetone–butanol–ethanol (ABE) fermentation using corn stover and high-efficient product recovery by vapor stripping–vapor permeation (VSVP) process

The effect of citrate buffer on enzymatic hydrolysis and ABE fermentation

The composition of raw corn stover used in this work contained 41.2 % cellulose, 21.6 %
hemicellulose, 32.7 % lignin and ash. After 2 % NaOH pretreatment, the solid residues
varied to contain 62.2 % cellulose, 24.6 % hemicellulose, 11.3 % lignin and ash, with
weight loss of ~40 %. The weight loss of corn stover in alkali pretreatment was mainly
attributed to the solubilization of components in corn stover such as lignin, hemicellulose
and other soluble extractants. Thus, ~90 % of cellulose was recovered in alkali pretreatment,
but ~80 % of lignin was removed from corn stover due to the solubilization of lignin
in NaOH solution. Compared with dilute H
₂
SO
₄
, lime and NH
₃
/HCl pretreatment, pretreatment of corn stover with 2 % NaOH was proved to substantially
increase the lignin removal and improve the accessibility and digestibility of cellulose
6]. Furthermore, the enzymatic hydrolysate from NaOH-pretreated corn stover contained
higher content of fermentable sugars and less inhibitors. Compared with dilute acid
and steam explosion pretreatment, alkali pretreatment only produced acetic acid in
the liquid stream, but dilute acid and steam explosion produced inhibitors (furfural,
HMF, acetic acid and formic acid) soluble in the liquid stream 13]. The furfural and HMF existed in fermentation broth were considered as inhibitors
in microorganism fermentation. Therefore, the hydrolysate from alkali pretreatment
may be the most favorable carbon source for microorganism.

The sodium citrate buffers with different concentrations were used to investigate
the effect on enzymatic hydrolysis and ABE fermentation using corn stover. When 10 g
of corn stover solid residues from alkali pretreatment was added into 100 mL sodium
citrate buffers with the concentration ranges of 20–100 mM, respectively, 81.8 ± 2.3 g/L
total sugars (54.0 ± 1.5 g/L glucose, 18.8 ± 0.5 g/L xylose, 7.5 ± 0.7 g/L cellobiose,
1.7 ± 0.1 g/L arabinose) were released in the enzymatic hydrolysis (see Table 1). There were no prominent differences in sugars concentrations released from corn
stover when using different strengths of the citrate buffer. The one-way ANOVA analysis
indicated that citrate buffers in the test concentrations had no significant effect
on fermentable sugars released from corn stover due to P value of 0.05. After inoculation of the seed and addition of other P2 medium components,
the corn stover hydrolysates with various citrate strengths were used for ABE fermentations,
with initial total sugars of 72.2 ± 2.8 g/L (glucose 46.1 ± 1.0 g/L, xylose 17.8 ± 1.0 g/L,
cellobiose 6.1 ± 0.5 g/L, arabinose 1.5 ± 0.1 g/L, respectively). When sodium citrate
concentrations in the hydrolysate increased from 20 to 30 mM, butanol and ABE titer
increased from 9.4 and 15.8 to 11.2 and 19.7 g/L, respectively, but then gradually
decreased to 6.4 and 11.0 g/L when sodium citrate in the hydrolysate increased to
100 mM. The maximum butanol concentration, yield and productivity were obtained with
11.2 g/L, 0.28 g/g, 0.19 g/L/h, respectively, when 30 mM citrate buffer was used for
enzymatic hydrolysis. Compared with other buffer strengths, more glucose were consumed
in ABE fermentation under the scenario with 30 mM citrate buffer. There was no decrease
of cellobiose and arabinose concentrations in fermentation broth as Clostridium beijerinckii CC101 could not utilize them as carbon sources. The one-way ANOVA analysis indicated
that citrate buffers had very significant effect on butanol production as P value was less than 0.001.

Table 1. The performance of enzymatic hydrolysis and ABE fermentation under various citrate
buffer strengths using corn stover

Kinetics of cell growth in ABE fermentation in various strengths of citrate buffer
is shown in Fig. 1. The maximum cell growth was obtained in the corn stover hydrolysate medium with
30 mM citrate buffer. The growth of C. beijerinckii CC101 was deterred by the hydrolysate medium with more than 50 mM citrate buffer.
Higher citrate strengths inhibited cell growth by reducing the cells internal pH and
proton motive force, and changing cell membrane permeability 14]. Higher strength of citrate buffer will lead to higher concentration of undissociated
citric acid and higher medium osmolality, which can directly affect cell growth. In
addition, citrate buffers may chelate trace elements, which may influence the optimum
cell growth in the medium 8], 15]. Since P value was between 0.05 and 0.001, citrate buffers had significant effect on cell
growth in corn stover hydrolysate. To further verify the inhibitory effect of citrate
buffer on cell growth and butanol production, the corn stover mediums with 30 and
100 mM citrate buffer were diluted twice and then used for ABE fermentation. As shown
in Fig. 1 and Table 1, the cell growths were enhanced in these two diluted mediums. Compared with corn
stover medium with 100 mM citrate buffer, the dilute medium was more effective for
cell growth, with the maximum OD and butanol productivity increased by 90 and 27.3 %,
respectively.

Fig. 1. Kinetics of cell growth in ABE fermentation in various strengths of citrate buffer.
The strengths of citrate buffer are in the range of 10–100 mM. a Corn stover hydrolysate as carbon source in citrate buffer with different concentrations
(20, 30, 40, 60, 80, 100 mM); b glucose as carbon source in citrate buffer with different concentrations (10, 20,
30, 40, 60, 80, 100 mM)

To investigate the effect of citrate buffer on ABE fermentation, the sodium citrate
buffers with the concentrations of 10–100 mM were added to P2 medium using initial
glucose of 66.0 ± 2.0 g/L as sole carbon source, respectively. As shown in Table 2, when sodium citrate concentration increased from 10 to 60 mM, butanol and ABE concentration
decreased gradually from 9.1 and 14.2 to 4.6 and 7.3 g/L in the time course of 48 h,
respectively. When the sodium citrate concentrations were at 80 and 100 mM, there
were no glucose consumption and ABE production in fermentation broth. As P value was less than 0.001, the strength of citrate buffer had very significant effect
on butanol production in glucose P2 medium. As shown in Fig. 1b, the cell growth gradually decreased with the increase of citrate buffer strength,
and there were no cell growth at 80 and 100 mM of citrate buffer. The one-way ANOVA
analysis indicated that citrate buffer had very significant effect on cell growth
in glucose P2 medium as P value was much less than 0.001. Furthermore, estimated by P value, the effect of citrate buffer on cell growth in glucose P2 medium was more
significant than that in corn stover hydrolysate P2 medium. When using glucose as
sole carbon source in the P2 medium, sodium citrate had strong toxic effect on cell
growth and ABE production. But, when using corn stover hydrolysate, the extractants
from corn stover as well as cellulose cocktail enzymes may interact with citrate buffer
and alleviate the effect of citrate buffer on cell growth.

Table 2. The performance of ABE fermentation with various citrate buffer strengths using glucose

Product recovery from fermentation broth by vapor stripping–vapor permeation process

The vapor stripping–vapor permeation (VSVP) process with the pure PDMS membrane was
carried out to investigate the performance of product recovery from active fermentation
broth. The VSVP process is the membrane-based technology in which the solvent mixture
vaporizes by gas stripping and then contacts with one side of the membrane. The vapor
mixture is diffused into the membrane, then desorbed to the permeate side as vapor
under vacuum. Finally, the vapor is condensed at a low temperature. The fermentation
broth containing 11.2 g/L butanol, 7.5 g/L acetone and 1.1 g/L ethanol in 75 mL was
used in vapor stripping–vapor permeation process, which was derived from corn stover
hydrolysate treated with 30 mM citrate buffer in the enzymatic pretreatment (see Table 1). When vapor stripping–vapor permeation experiment was carried out in 4 h, ~90 %
of ABE solvent could be recovered from fermentation broth. The butanol, acetone and
ethanol concentration in fermentation broth decreased from 11.2, 7.5 and 1.1 to 0.7,
0.6 and 0.8 g/L, respectively. The recovery rate of butanol and ABE were 93.7 and
89.4 %, respectively. In the first hour of product recovery process, the condensate
containing 147.5 g/L butanol, 70.0 g/L acetone and 8.8 g/L ethanol was achieved, with
totally 236.2 g/L ABE solvents (see Fig. 2). Then, the butanol, acetone and ethanol concentration in the condensate gradually
decreased to 41.4, 26.4 and 6.4 g/L at 4 h, respectively. The total flux was relatively
stable in the range of 217.2–243.1 g/m
²
/h, while butanol flux decreased with time due to the decreased butanol concentration
in fermentation broth. The separation factors of butanol and acetone increased with
time, indicating that the selectivities of butanol and acetone over water in VSVP
process were higher in the feed with low butanol and acetone concentrations. There
was no variation of ethanol separation factor as the ethanol concentration in feed
solution maintained stable at a low level. The average butanol and ABE concentrations
were 100.4 and 153.5 g/L, respectively. The average separation factors of butanol,
acetone and ethanol were 34.2, 13.9 and 8.1, respectively. The demonstrating results
showed that the VSVP process was very effective for ABE solvents recovery from fermentation
broth.

Fig. 2. The performance of ABE recovery from fermentation broth using vapor stripping–vapor
permeation process. a ABE concentrations in condensate and total flux; b separation factors of butanol, acetone and ethanol

Simulation of in situ product recovery during ABE fermentation

To mimic in situ product recovery during ABE fermentation, the vapor stripping–vapor
permeation process was conducted using fermentation broth with butanol, acetone and
ethanol concentrations of 9.7, 4.9 and 2.3 g/L at 37 °C. According to our previous
studies, in situ product recovery during ABE fermentation are usually conducted at
~10 g/L butanol in fermentation broth, which not only could alleviate butanol toxicity
to cells but also obtain a high butanol concentration recovered in condensate 16]. Therefore, the fermentation broth containing ~10 g/L butanol in 500 mL was used
to investigate the performance of VSVP process to continuously recover ABE solvents.

Since the volume of feed solution was much greater than the recovered volume per hour,
the performance of the VSVP process was very stable due to the ABE concentrations
in feed solution maintained at stable level. When butanol, acetone and ethanol concentrations
in feed solution were in the range of 9.4–9.7, 4.5–4.7 and 2.1–2.3 g/L, respectively,
the VSVP process produced the recovered condensate containing 212.0–232.0 g/L butanol,
86.3–115.5 g/L acetone and 8.3–8.6 g/L ethanol, with total flux of 117.2–124.1 g/m
²
/h (see Table 3). The average separation factors of butanol, acetone and ethanol were 29.8, 24.3
and 3.9, respectively.

Table 3. The comparison of ABE recovery by VSVP process, pervaporation and gas stripping

However, using the same PDMS membrane and feed solution, pervaporation only produced
71.5–77.4 g/L butanol, 35.0–39.8 g/L acetone and 6.3–6.7 g/L ethanol in the condensate,
with total flux of 48.8–54.3 g/m
²
/h (see Table 3). The average separation factors of butanol, acetone and ethanol were 8.2, 8.3 and
3.2, respectively. According to our previous studies, the separation factors of homogeneous
PDMS membranes with different thicknesses were 7.5–8.0 for butanol recovery at 37 °C,
which was enhanced to ~20 when the temperature increased to 80 °C 17]. It was also reported that the PDMS membrane by pervaporation coupling with ABE fermentation
was used to recover butanol from fermentation broth, with butanol separation factor
of 7.0–10.3 18]. Therefore, from butanol purity point of view, it should be noted that the VSVP process
for butanol recovery is more effective than pervaporation. In addition, hydrophobic
fillers or layer composited with the PDMS membrane could improve the separation factor
of butanol and mass flux in pervaporation 19], 20]. Therefore, the performance of the VSVP process could be dramatically enhanced if
using the composite PDMS membrane.

Gas stripping is an alternative technique that allows the selective removal of volatile
solvents from fermentation broth, with solvents recovery from the vapor phase by condensation
in a cold trap or via a molecular sieve. The main difference of VSVP process with
gas stripping is that the stripped volatile solvents from the vapor phase permeate
through a membrane before condensation in a cold trap. When the same fermentation
broth with butanol, acetone and ethanol concentrations of 9.7, 4.9 and 2.3 g/L at
37 °C was used as feed, gas stripping produced the condensate containing 99.8–106.5 g/L
butanol, 41.1–46.3 g/L acetone and 7.4–7.7 g/L ethanol (see Table 3). The separation factors of butanol, acetone and ethanol were 10.9, 10.0 and 3.8,
respectively. In comparison with the VSVP and gas stripping, the stripping rates from
fermentation broth in two processes should be theoretically equal if all of the stripped
solvents could be recovered and condensed. But the apparent stripping rate of VSVP
process could be limited by the membrane performance. If the stripped solvents could
not be completely recovered by the membrane, some of stripped solvents and water would
circulate back to fermentation broth. Thus, the VSVP process has a lower apparent
stripping rate than gas stripping. In present study, the flow rate of stripping gas
in the VSVP process is twofold higher than that in gas stripping, the stripping rate
of the VSVP process is supposed to be twofold higher than gas stripping. But due to
limitation of membrane flux, some of stripped solvents circulated back to fermentation
broth, and finally the apparent stripping rate of the VSVP process is 2-fold higher
than that of gas stripping. The apparent stripping rate could be enhanced by increasing
the membrane area or performance. In addition, the butanol selectivities over acetone
(SF
_B
/SF
_A
) in the three processes are also shown in Table 3. It should be noted that the VSVP process has the highest butanol selectivity over
acetone. Therefore, the VSVP process has the best performance for ABE solvents recovery
among these three processes. The separation factors of butanol and acetone in VSVP
process are at least two times higher than those in pervaporation and gas stripping.

Fermentative production of butanol is limited to low concentration of typically less
than 2 % (w/v) butanol, which leads to high separation energy demand by conventional
distillation approaches. The process simulation of hybrid vapor stripping–vapor permeation
(membrane-assisted vapor stripping system, MAVS) indicated that significant reductions
in energy demand are possible for MAVS systems compared with conventional distillation
systems to separate ABE solvents from butanol/water binary solutions and ABE/water
solutions. The MAVS system is estimated to require 6.2 MJ-fuel/kg-butanol to produce
99.5 % (w/v) butanol from a 1 % (w/v) butanol feed solution, with energy saving of
63 % relative to a benchmark distillation/decanter system 21]. Furthermore, the MAVS pilot unit shows an excellent demonstration that the energy
usage of 10.4 MJ-fuel/kg-butanol is required to achieve 85 % butanol recovery from
a 1.3 % (w/v) solution 22]. The energy usage could be further reduced by more heat-integrated design. Many scholars
invested their attention on studies of membrane-assisted pervaporation for butanol
recovery. The butanol vapor stripped from the feed is more concentrated during the
stripping process, and the concentrated vapor contacting with one side of the membrane
in the VSVP process could dramatically improve the separation factor of butanol. Furthermore,
when butanol fermentation was integrated with pervaporation, the membranes tend to
be contaminated when in contact with fermentation broth for pervaporation. The cells
and macromolecules in the fermentation broth tended to adsorb on or infiltrate into
the membrane, which induced the membrane fouling. Membrane fouling significantly increases
the downtime and running cost, even though the membrane could be easily recovered
by water rinsing. In the VSVP process, the volatile vapor containing ABE solvents
and water contacts with both sides of the membrane, which could not induce the membrane
contamination. The cells and macromolecules would be detained in the fermentation
broth and have no chance to contact with the membrane because they are nonvolatile.
Therefore, the VSVP process coupling with ABE fermentation has potential application
in the industrial production of biobutanol for long duration.