Bioresource Technology 203 (2016) 295–302

Contents lists available at ScienceDirect

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

Open fermentative production of fuel ethanol from food waste

by an acid-tolerant mutant strain of Zymomonas mobilis
Kedong Ma a,1, Zhiyong Ruan b,1, Zongxia Shui c, Yanwei Wang c, Guoquan Hu c, Mingxiong He c,⇑
a
College of Environmental and Chemical Engineering, Dalian University, Dalian 116622, PR China
b
Key Laboratory of Microbial Resources (Ministry of Agriculture, China), Institute of Agricultural Resources and Regional Planning, CAAS, Beijing 100081, PR China
c
Key Laboratory of Development and Application of Rural Renewable Energy, Ministry of Agriculture, Biomass Energy Technology Research Centre, Biogas Institute of Ministry
of Agriculture, Chengdu 610041, PR China

h i g h l i g h t s

 An economical bioprocess for ethanol production from food waste was developed.
 Open fermentation by employing acid-tolerant Z. mobilis ZMA7-2 was achieved.
 Bench scale ethanol fermentation and cell reusability was feasible.
 Z. mobilis ZMA7-2 is an ideal organism for food waste based ethanol production.

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

Article history: The aim of present study was to develop a process for open ethanol fermentation from food waste using
Received 19 October 2015 an acid-tolerant mutant of Zymomonas mobilis (ZMA7-2). The mutant showed strong tolerance to acid
Received in revised form 17 December 2015 condition of food waste hydrolysate and high ethanol production performance. By optimizing fermenta-
Accepted 19 December 2015
tion parameters, ethanol fermentation with initial glucose concentration of 200 g/L, pH value around 4.0,
Available online 23 December 2015
inoculum size of 10% and without nutrient addition was considered as best conditions. Moreover, the
potential of bench scales fermentation and cell reusability was also examined. The fermentation in bench
Keywords:
scales (44 h) was faster than flask scale (48 h), and the maximum ethanol concentration and ethanol yield
Food waste
Fuel ethanol
(99.78 g/L, 0.50 g/g) higher than that of flask scale (98.31 g/L, 0.49 g/g). In addition, the stable cell growth
Zymomonas mobilis and ethanol production profile in five cycles successive fermentation was observed, indicating the mutant
Open fermentation was suitable for industrial ethanol production.
Acid-tolerant mutant Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction based ethanol production led to the problems of crop competition

In the past decades, renewable biomass-based energy produc-
tion has gained great interest due to the growing global energy
demand and environmental pollution associated with the fossil
fuels (Ma et al., 2014a; Ma and Ruan, 2015). Among them, bioetha-
nol as one of the most promising alternative to those fossil fuels
has been intensively studied and its production is increasing over
the years, reaching the level of 100 billion liters during the year
2013 (Ho et al., 2014). Traditionally, bioethanol is produced mainly
from the first generation feedstocks i.e. sugar (sugar cane and sugar
beet) or starch energy crops (corn, wheat and rice). But the crop-
⇑ Corresponding author. Tel./fax: +86 28 85242281.
E-mail address: hemingxiong@caas.cn (M. He).
1
Two authors equally contributed to this work and should be considered co-first
author.
and landfill use conflict in addition to the high production cost hin-
dered the global application of bioethanol (Ho et al., 2014; Choi
et al., 2015; Horisawa et al., 2015). To solve these problems, the
research emphasis has then moved forward to the ethanol produc-
tion using the second generations feedstocks comprise of non-food
lignocellulosic biomass such as agricultural byproducts, forestry
residues or municipal waste (Choi et al., 2013; Ho et al., 2014;
Saha et al., 2015). Even though lignocellulosic biomass is the most
abundant and renewable resource suitable for bioethanol produc-
tion, their variable composition and high lignin content making
the pretreatment process very difficult and costly, thus limit its
industrial scale production. (da Costa et al., 2015; Ho et al., 2013).
Food waste is organic solid wasted discharged from sources
including food processing plants, household kitchens and restau-
rants (Kiran and Liu, 2015). Over a billion tons of food waste is gen-
erated per year, and only in Japan, the annual generation of food

http://dx.doi.org/10.1016/j.biortech.2015.12.054
0960-8524/Ó 2015 Elsevier Ltd. All rights reserved.
296 K. Ma et al. / Bioresource Technology 203 (2016) 295–302
wastes is about 20 million tons (Han et al., 2015; Jin et al., 2015;
Pleissner et al., 2013; Tang et al., 2008; Tashiro et al., 2013). The
increasing amount of food waste has put tremendous pressure on
the municipal solid waste management system. Moreover, dioxin-
related compounds released in the incineration treatment poten-
tially causes air pollution, and disposal of residues is also becoming
increasingly expensive because of land limitations and transport
and labor costs (Kiran et al., 2014). Therefore, from both environ-
mental and economical viewpoint, an appropriate treatment and
recycling of food waste is strongly needed (Kiran et al., 2015). Food
waste mainly consists of starch, protein, sugar and cellulose, which
can be converted to fermentable sugars easily by simple environ-
mental friendly pretreatment method for bioenergy production
(Kiran and Liu, 2015). Furthermore, the natural low pH of food
waste is also an attractive characteristic compare to other waste
biomass. The acid condition of medium not only can prevent the
contamination by the microorganism but also make the open
(non-sterilized) fermentation applicable (Ma et al., 2009a). It is
reported that ethanol production under non-sterilized condition
can save 30–40% energy consumption in cooking starch and steril-
ization during ethanol production (Ma et al., 2014b; Tao et al.,
2005). Hence, along with the fact that food waste is in abundant
supply, suggest that it may be a potential source for ethanol
production.
Zymomonas mobilis, a kind of Gram-negative ethanol-producing
bacteria have been attracting increasing attention for fuel ethanol
production in recent years for some favorite industrial characteris-
tics over yeasts like Saccharomyces cerevisiae, including (i) anaero-
bic growth ability, (ii) high sugar and ethanol tolerance and (iii)
metabolize sugar via the Entner–Doudoroff (ED) pathway (He
et al., 2014). Using Entner–Doudoroff (ED) pathway, less ATP and
less biomass is produced and more carbon sources are thus chan-
neled to ethanol, resulting in high ethanol yield and a higher
glucose metabolic flux, normally three- to fivefold that of S. cere-
visiae (Bai et al., 2008; Wirawan et al., 2012). Although Z. mobilis
is better than yeast in many aspects, the drawback of narrow sub-
strate utilization (only glucose, fructose and sucrose) and sensitive
to acetic acid and other compound produced in pretreatment pro-
cess prevented its commercial use (Shui et al., 2015). Therefore,
many efforts have been made to develop Z. mobilis strains with
desirable features in terms of enhanced inhibitor-tolerant and
wide substrate utilization capability. In contrast to conventional
mutagenesis and traditional genetic engineering, adaptive labora-
tory evolution (ALE) has the advantage of obtaining beneficial
mutations in a rather simple stress selective way. And thus, as a
powerful metabolic engineering tool for strain development and
optimization has been successfully applied in Escherichia coli
(Hua et al., 2007; Lee and Palsson, 2010) and S. cerevisiae (Çakar
et al., 2012). These engineered microorganisms increase their resis-
tance of acetic acid stress by successive exposure to high concen-
tration of sodium acetate and the acid tolerant mechanism are
investigated (Arnold et al., 2001). Currently, ALE strategy was also
employed for Z. mobilis strain improvement. Agrawal et al. (2011)
reported highly efficient xylose-fermenting Z. mobilis strain, and
Ma et al. (2009b) documented acid-tolerant strain by adaptive
mutation, respectively. It is worthy to investigate the possibility
of using those engineered strains to produce ethanol from a wide
range of waste biomass for promoting the application of Z. mobilis
strains in ethanol production.
In this study, we employed a mutant acetic acid tolerant Z.
mobilis strain for ethanol production using food waste as substrate.
This acid-tolerant strain was developed by ALE approach for cellu-
losic ethanol production, and the profiling of cell growth, glucose
utilization and ethanol fermentation under normal or stress condi-
tions was preliminarily examined by our research group (Shui
et al., 2015). For the purpose of examining the fermentation
performance of mutant strain in food waste based ethanol produc-
tion, the research was carried out from the following aspects: (1)
ethanol fermentation performance of mutant strain in the presence
of acetic acid at different conditions, (2) the optimum fermentative
parameters, (3) the fermentation performance under different
ethanol production scale, (4) the feasibility of biomass recycling
and (5) the evaluation of over mass balance of our process.
2. Methods
2.1. Experimental design
In this study, an experiment comprises of three parts were
designed and the overall experimental flowchart was summarized
in Fig. 1 in Supporting information. In Part I, the feasibility of etha-
nol production from food waste and experimental parameters on
ethanol production were investigated, respectively. In Part II and
III, the potential of biodiesel and biogas production using recovered
waste oil and residue solids are being studied in our lab.
2.2. Microorganism and medium
Z. mobilis ZM4 (ATCC 31821) was purchased from the American
Type Culture Collection (Rockville, MD). Z. mobilis ZMA7-2 as new
bacterial strain that could tolerate high levels of acetic acid was
mutated from the wild-type strain (Z. mobilis ZM4) by adaptive
laboratory evolution (ALE) experiments (Shui et al., 2015). The
strains were maintained at 4 °C on agar plates containing 20 g glu-
cose, 10 g yeast extract, and 20 g agar per liter of deionized water,
with fortnightly sub-culturing until usage. The cells were the
transformation recipient and statically grown overnight in a rich
medium (RM) (Goodman et al., 1982) at 30 °C. For long-term stor-
age, stock cultures were maintained in 40% glycerol at 80 °C.
2.3. Materials and chemicals
The food waste used in this study was collected from the local
supermarket (Jusco, Wakamatsu, Kitakyushu and Fukuoka, Japan).
The non-recyclable materials (plastic, wrapper, metal, and glass
pieces) were separated if present in the waste, and recyclable
organic materials were stored in plastic baskets at 4 °C until used
in a day or two. The food waste consisted of the carbohydrate
group (i.e. rice, bread, pasta, noodles, etc.), the protein group (i.e.
fish, beef, pork, chicken, etc.) and the vegetable/fruit group. Repre-
sentative characteristics of the food waste are shown in Table 1.
All the chemicals were analytical grade and purchased from
Wako pure Chemical Industries, Ltd. of Japan and used as received
without further purification.
2.4. Saccharification of food waste
The food waste (50 kg) was chopped into small pieces using a
food-processing chopper (Model MKB C-22, Masuko Sangyo Co.
Ltd. Japan), transferred to a 90 L bioreactor and 25.0 kg tap water
by a ratio 2:1 (w/w) was added. Next, glucoamylase (Glucoteme,
20,000 U/g; Nagase sangyo, Osaka, Japan) was added to a final con-
centration of 180 mg protein kg 1 wet waste and saccharification
was performed at 50 °C, 200 rpm for 6 h. After saccharification,
the saccharified broth was separated by decanter centrifuge
(Model HS-25 TP, IHI Co. Ltd. Japan) at 3500g. The recovered
saccharified liquid was then concentrated by frozen and thawed
process and adjusted the reducing sugar to the concentration of
approximately 200 g/L. A clear saccharified food waste solution
was obtained and used as the substrate for ethanol fermentation
in this work. The remaining saccharification residue and the
 K. Ma et al. / Bioresource Technology 203 (2016) 295–302 297

Table 1
Characteristics and composition of food waste before and after saccharification.

Parameters Food waste Residue solids (% w/w)a Recycle oil (% w/w)a

(% w/w)a
Moisture 72.5 ± 0.65 76.3 ± 0.35 0.0
Total solids (TS) 27.5 ± 0.71 23.7 ± 0.35 0.0
Combustible component 27.4 ± 0.16 18.1 ± 0.15 100.0 ± 0.00
Ash 2.8 ± 0.02 1.8 ± 0.02 0.0
Total sugars 11.1 ± 0.75 3.8 ± 0.20 0.0
Protein 4.5 ± 0.28 ND 0.0
Lipid 8.3 ± 0.23 ND 0.0
Holocellulose 0.9 ± 0.02 ND 0.0
C (based on dry weight) 47.2 ± 0.56 44.6 ± 0.50 76.6 ± 0.10
H (based on dry weight) 7.0 ± 0.08 6.7 ± 0.10 11.7 ± 0.05
O (based on dry weight) 34.2 ± 0.45 38.1 ± 0.25 11.7 ± 0.15
N (based on dry weight) 2.3 ± 0.52 3.7 ± 0.12 0.0
C/Nb 20.5 12.1

ND = no data.
a
The proportion of ingredient of food waste listed were average data calculated from sources of hospital and school cafeteria, convenient store, supermarket, restaurant,
food processing plant and kitchen garbage. Mean values and standard deviations of three determinations.
b
C/N means the ratio of total carbon and total nitrogen.

recovered oil can be used as potential cheap feedstocks for other
bioenergy production.
2.5. Fermentation
2.5.1. Preparation of inoculum
Z. mobilis cells grown on 2% glucose agar plates were inoculated
into 100 mL of inoculum medium (Inoculum A) or saccharified liq-
uid medium (SLM) of food waste with glucose 30–50 g/L at natural
pH (Inoculum B). Inoculum A medium consisted of 20 g glucose,
10 g yeast extract, 1 g KH2PO4, 1 g MgCl2 and 1 g (NH4)2SO4 per
liter of deionized. The phosphate was autoclaved separately from
the other ingredients and the final pH (approximately 5.6) was
not adjusted. In order to stimulate the growth of the Z. mobilis
strain, 2 g/L of yeast extract and 1 g/L of peptone were also added
in Inoculum B medium. Pre-cultivation was performed anaerobi-
cally at 30 °C for overnight without shaking, and then the resulting
pre-cultivation broth was used as inoculum for ethanol fermenta-
tion experiments.
2.5.2. Preparation of ethanol fermentation media
The base-line fermentation medium contained the following
components (g/l): glucose, 100; yeast extract, 10; KH2PO4, 1;
MgCl2, 1 g; (NH4)2SO4, 1 g. For food waste studies, saccharified liq-
uid medium (SLM) was used without sterilization. The initial pH
value of SLM medium was around 4.0, and its glucose concentra-
tion was around 200 g/L by a concentrating process.
2.5.3. Ethanol fermentation under different acetic acid stress
To analyze the acetic acid tolerance of wild-type strain and
mutant strain, 10% (v/v) Inoculum A of two strains were inoculated
into 100 mL base-line fermentation medium (mentioned above)
containing 0%, 0.2%, 0.4%, 0.6% (v/v) of acetic acid in Erlenmeyer
flasks and performed the fermentation, respectively. The initial cell
concentrations were both set at OD600 of 0.35 after inoculation.
2.5.4. Open fermentation test at laboratory scale
Laboratory fermentation assays were conducted in 250-mL
Erlenmeyer flasks, with a working volume of 100-mL. The inocu-
lum was added at a ratio of 10% (v/v) to the fermentation mixture
under aseptic conditions. The fermentation process was conducted
openly (without sterilization) and without shaking to lasted for
48 h under anaerobic conditions. The flasks were sealed with rub-
ber stoppers equipped with needles for CO2 venting and nitrogen
of 0.5 vvm was flowed into the flask for 10 min to remove dis-
solved oxygen in fermentation broth.
To evaluate the experimental parameters in terms of initial
glucose concentration, pH, nitrogen sources and inoculum size on
ethanol production, the initial glucose concentration in the SLM
was adjusted to approximately 180, 200, 220, 240, 260 g/L. The ini-
tial pH of SLM was set at 3.5, 4.0, 5.0 and 6.0 with 3 N HCl or 5 N
NaOH; Four types of nitrogenous nutrients of yeast extract, pep-
tone, Urea and (NH4)2SO4 (abbreviated to N-1, N-2, N-3, N-4) were
added to the SLM; Five inocula sizes (4%, 6%, 8%, 10% and 12%) were
used to determine the effect of inoculum size on ethanol produc-
tion, respectively. Pure SLM under natural pH, non-sterilized con-
ditions and without any supplementation of other nitrogen was
used as control in this study.
2.5.5. Open fermentation test at bench scale
The bench scales batch fermentation were carried out anaerobi-
cally in 3 L (Biostat B) and 20 L (Biostat D) fermentors (B. Brown),
10% (v/v) Inoculum B broth was transferred aseptically, with final
working volumes of 2.0 L and 14 L, respectively. The experiments
were conducted openly (without sterilization) at 30 °C with agita-
tion rate of 100 rpm under anaerobic conditions. The pH of the cul-
ture was recorded constantly but not regulated. The CO2 evolution
rate was used as an indicator of the progress and completion of fer-
mentation. The fermentation broth was sparged with N2 (men-
tioned above) to achieve the anaerobic condition for fermentative
ethanol production. All bench scales batch fermentations were car-
ried out at the optimized ethanol production conditions. Samples
were taken at regular intervals to follow the cells growth, glucose
and ethanol concentrations.
2.6. Open ethanol production using the recycled Z. mobilis ZMA7-2
cells
The effect of using recycled acid-tolerant mutant strain of
Z. mobilis cells on ethanol production was examined in 3 L bench
fermentor. 10% (v/v) inoculum was used as starter culture for ini-
tiating fermentation. After incubation at 30 °C for 44 h, the fer-
mented broth was transferred to the sterilized centrifuge tubes
which were spun at 5000g at 4 °C for 5 min in a refrigerated cen-
trifuge. The supernatant was analyzed for glucose and ethanol con-
centrations, while the biomass was vortexed and adjusted to 3.5 of
OD600 for inoculation of the subsequent batch. All the batches were
prepared in the same way, and the experiments were conducted
for five successive cycles.
298 K. Ma et al. / Bioresource Technology 203 (2016) 295–302
2.7. Analysis methods
2.7.1. Compositional analysis of food waste samples
Total solids (TS) were analyzed in accordance to standard meth-
ods (Tang et al., 2008). The moisture content in the food waste was
determined by the decrease in weight following drying overnight
at 105 °C constant mass in an oven. The ash content was deter-
mined at 600 °C by combustion for 2 h. The protein content of
the food waste was calculated as 6.25  nitrogen content. The lipid
content was determined by hexane/isopropanol (3:2) method
(Hara and Radin, 1978). The content of carbon, nitrogen, hydrogen
and oxygen of dry samples was determined using a CHN coder
MT-3 (Yanako, Tokyo, Japan) (Tang et al., 2008). Holocellulose
was analyzed following NREL Laboratory Analytical Procedure
(LAP) (Sluiter et al., 2008) using two-step acid hydrolysis.
2.7.2. Sugar, ethanol, cells and organic acid determination
The total reducing sugars in the hydrolyzate were measured by
the DNS (3,5-dinitrosalicylic acid) method, using glucose as stan-
dard. The glucose concentration was enzymatically analyzed by
the glucose oxydase–peroxydase enzyme (Wako pure Chemical
Industries, Ltd. Osaka, Japan). Fructose, glucose and sucrose were
analyzed employing an HPLC apparatus (Shimadzu LC-20AD)
equipped with a refractive index (RI) detector (Shimadzu RID
10A) and an Aminex HPX-87P (Bio-Rad, 300  7.8 mm) chro-
matography column. Mobile phase was degassed HPLC-water at
a flow rate of 0.6 mL/min and column temperature of 70 °C. For
ethanol concentration, an HPLC apparatus equipped with a refrac-
tive index (RI) detector. An Aminex HPX-87H (Bio-Rad,
300  7.8 mm) chromatography column was used. Mobile phase
was 5 mM H2SO4 in degassed HPLC-water at a flow rate of
0.6 mL/min and column temperature was 40 °C. The optical cell
density was determined using a spectrophotometer detector
(Shimadzu SPD-20A, Japan) at wavelength 600 nm after proper
dilution with deionized water to give an absorbance range of
0.05–0.90. Un-inoculated fermentation medium was used as a
blank. Volatile fatty acids (VFAs), including lactic acid and acetic
acid, were analyzed as described previously (Ma et al., 2014a).
2.8. Statistical analysis
All the experiments values presented in graphs and tables are
the mean ± SD of three replications calculated using MS Excel.
Multiple comparison tests were performed with Student’s t test
(significance levels = 0.05) (Ma and Ruan, 2015).
3. Results and discussion
3.1. Ethanol fermentation under different acetic acid stress
In this study, an acid-tolerant mutant Z. mobilis ZMA7-2 was
adopted for food waste based ethanol production. To examine the
ethanol fermentation performance of mutant at pH range around
4.0, three types of medium with acetic acid concentration of 0.2%
(pH 4.22), 0.4% (pH 3.97) and 0.6% (pH 3.85) were prepared. Base-
line medium without acetic acid addition and wild type strain of
Z. mobilis ZM4 (ATCC 31821) was used as control. When fermenta-
tion was conducted in base-line medium in the absence of acetic
acid, almost no differences in cell density and ethanol yield was
observed between ZMA7-2 and ZM4. The glucose was completely
consumed to produce 50.02 g/L and 49.88 g/L of ethanol within
the 24 h by two strains (Table 2). However, with the increasing of
acetic acid concentration in medium, the cell growth and ethanol
production capability of wild type was obviously inhibited. The cell
density (OD600) gradually decreased from 3.95 to 2.81, 1.23 and
finally reached 0.46 under 0.2%, 0.4% and 0.6% acetic acid,
corresponding to ethanol yield of 0.44 g/g, 0.22 g/g and 0.06 g/g,
respectively. In contrast to wild type, minor fluctuation of cell
growth and ethanol production was observed in mutant strain. As
shown in Table 2, the cell density (OD600) was 3.83, 3.86 and 3.76
under 0.2%, 0.4% and 0.6% acetic acid, and ethanol concentration
was 49.05 g/L, 48.69 g/L and 47.53 g/L, corresponding to ethanol
yield of 0.49 g/g, 0.49 g/g and 0.48 g/g, respectively. The results
obtained were very close to that of control medium, implying the
feasibility of ethanol fermentation at pH range around 4.0 by
ZMA7-2, and hence the mutant was subjected for further
investigation.
3.2. Optimization fermentation parameters at laboratory scale
3.2.1. Effect of initial glucose concentration on ethanol production
The ethanol fermentation performance of ZMA7-2 under various
initial glucose concentrations was evaluated, and the result was
summarized in Fig. 1A. When the initial sugar concentration was
adjusted at 180 g/L and 200 g/L, the glucose could be completely
exhausted by ZMA7-2 within 48 h. The cell density (OD600), ethanol
concentration and ethanol efficiency was 3.86, 88.76 g/L and 96.5%
under 180 g/L glucose, and 3.66, 99.32 g/L and 96.2% under 200 g/L
glucose. However, ZMA7-2 showed a decreasing sugar consump-
tion tendency as well as cell growth and ethanol production perfor-
mance with the increasing initial glucose concentration. The cell
density (OD600), ethanol and ethanol efficiency dropped to 1.26,
35.6 g/L and 34.32% when glucose concentration in medium was
260 g/L, indicating that glucose concentration higher than 200 g/L
gave negative influence on cell growth and ethanol fermentation
performance. This negative influence might have several causes,
including some level of inhibitory compound to microorganism
by excessive sugar concentration. In addition, high osmotic pres-
sure exerted by the concentrated sugar medium also caused a sub-
strate inhibitory effect to microorganism.
3.2.2. Effect of initial pH on ethanol production
The effect of the initial pH value of food waste hydrolysate on
ethanol production was examined at a pH range of 3.5–6.0
(Fig. 1B). All experiments were conducted openly (without steril-
ization) condition. The cell density (OD600), ethanol and fermenta-
tion efficiency obtained at natural pH condition was 3.93, 99.8 g/L
and 97.36%, respectively. A slight increase in ethanol concentration
was observed when the initial pH raised from 3.5 to 4.0. The cell
density (OD600), ethanol and fermentation efficiency increased
from 3.86, 96.2 g/L and 93.84% at pH 3.5 to 3.9, 99.6 g/L and
97.16% at pH 4.0. In opposite, the fermentation kinetics gradually
decreased when pH increasing to 5.0 and 6.0, finally reduced to
3.52, 76.5 g/L and 74.6%. This phenomenon might contribute to
the contamination caused by other microorganism such as lactic
acid bacteria, which could be proved by high concentration of
acetic acid and lactic acid in medium, consequently inhibited the
cell growth and ethanol fermentation performance of ZMA7-2. In
our previous study, we found that the initial pH of saccharified
food waste liquid was around 4.0 due to lactic acid and acetic acid
produced by microorganism during the transport and storage pro-
cess (Ma et al., 2009a). According to analysis data, 200.0 g/L food
waste hydrolysate usually comprised 7.0–10.0 g/L lactic acid and
3.0–4.0 g/L acetic acid. It had been reported that lactic acid and
acetic acid present together in a medium exhibited highly syner-
gistic inhibitory effect to yeast (Narendranath et al., 2001). There-
fore, although the mutant strain exhibited excellent growth and
ethanol production profile in the presence of acetic acid in base-
line medium, it still remained uncertain whether the mutant
developed for acetic acid tolerance could cross-protect the cells
from lactic acid in saccharified food waste liquid. The maximum
ethanol and fermentation efficiency was generated at pH value
 K. Ma et al. / Bioresource Technology 203 (2016) 295–302 299

Table 2
Growth and ethanol production by Z. mobilis ZM4 and ZMA7-2 in base-line fermentation medium under different acetic acid stress conditions.

Z. mobilis Z. mobilis ZM4 Z. mobilis ZMA7-2

strains
Parameters Cell density Ethanol Ethanol yield Ethanol efficiency Cell density Ethanol Ethanol yield Ethanol efficiency
(OD600)a (g/L)a (g/g glucose)a (%)a (OD600)a (g/L)a (g/g glucose)a (%)a
Acetic acid 3.95 ± 0.20 50.02 ± 0.08 0.50 ± 0.00 97.81 ± 0.18 3.91 ± 0.15 49.88 ± 0.05 0.50 ± 0.00 97.54 ± 0.35
(v/v) 0
0.2% 2.81 ± 0.30 43.36 ± 0.60 0.44 ± 0.02 85.18 ± 0.23 3.83 ± 0.20 49.05 ± 0.45 0.49 ± 0.09 95.91 ± 0.26
0.4% 1.23 ± 0.10 22.35 ± 1.00 0.22 ± 0.03 43.70 ± 0.10 3.86 ± 0.15 48.69 ± 0.50 0.49 ± 0.05 95.21 ± 0.15
0.6% 0.46 ± 0.13 5.81 ± 1.21 0.06 ± 0.05 11.36 ± 0.23 3.76 ± 0.20 47.53 ± 0.67 0.48 ± 0.06 92.94 ± 0.35
a
Mean values and standard deviations of three determinations.

120 10.00 120 10.00

A B
100 100

Fermentation efficiency (%)

Fermentation efficiency (%)

Ethanol concentration (g/l)

8.00
Ethanol concentration (g/l)

8.00

Volatile fatty acids (g/l)

Cell density (OD600)
Cell density (OD600)
80 80
6.00 6.00
60 60
4.00 4.00
40 40

2.00 2.00
20 20

0 0.00 0 0.00
180 200 220 240 260 Control 3.5 4.0 5.0 6.0
Glucose concentration (g/l) pH
Ethanol Fermentration efficiency OD Ethanol Fermentation efficiency OD Lactic acid Acetic acid

120 10.00 120 10.00

C D
D
100 100
Fermentation efficiency (%)

Fermentation efficiency (%)

Ethanol concentration (g/l)

8.00
Ethanol concentration (g/l)

8.00
Cell density (OD600)

Cell density (OD600)

80 80
6.00 6.00
60 60
4.00 4.00
40 40

2.00 2.00
20 20

0 0.00 0 0.00
Control N-1 N-2 N-3 N-4 4 6 8 10 12
Nitrogen sources Inoculum size (%, v/v)
Ethanol Fermentation efficiency OD Ethanol Fermentation efficiency OD

Fig. 1. Effect of fermentation parameters on ethanol production from food waste by Z. mobilis ZMA7-2. (A) Initial glucose concentration; (B) initial pH; (C) nitrogen sources;
(D) inoculum size.

around 4.0, in which the medium containing 7.32 g lactic acid and addition was used as control (Fig. 1C). Almost no difference in cell
3.34 g acetic acid. And the result obtained in food waste medium density (OD600) (3.95, 4.01), ethanol (99.43 g/L, 99.25 g/L) and fer-
was relative higher than that of in base-line medium only contain- mentation efficiency (97.1%, 96.83%) was observed in the fermen-
ing acetic acid, implying acetic acid tolerance led to tolerance to tation broth with yeast extract supplementation and without any
biomass derived other inhibitors such as lactic acid. This cross pro- added ingredient. The results obtained from other cases were sim-
tection made our acetic acid tolerant strain more attractive for use iliar, indicating the richness of Japanese food waste saccharified
in ethanol production from food waste under non-sterilized liquor could provide adequate nutrients for cell growth and addi-
condition. tional supplementation was not necessary. This result was also
coincidence with our previous study when using S. cerevisiae as
fermentation performer (Ma et al., 2009a).
3.2.3. Effect of various nitrogen sources
It is important to note that in ethanol production, nitrogen 3.2.4. Effect of initial inoculum size
sources were a critical element in the overall cost of production Ethanol fermentation was carried out with 4%, 6%, 8%, 10% and
because this parameter was a determinant of fermentation capac- 12% inoculum to examine the effect of inoculum size on ethanol
ity and thus contributes to capital costs. Therefore, the effect of fermentation. The maximum cell density (OD600) (3.95), ethanol
nitrogen sources on ethanol production by ZMA7-2 was examined (98.32 g/L) and fermentation efficiency (96.06%) was obtained with
in this study. Yeast extract, peptone, Urea and (NH4)2SO4 were an initial inoculums of 10%. There was a significant difference
applied as nitrogen sources, and medium without any nitrogen between an inocular size of 10% and other inoculums sizes
300 K. Ma et al. / Bioresource Technology 203 (2016) 295–302

(Fig. 1D). The higher or lower initial inoculum size generated low 220 120
value of cell density and ethanol concentration. These phenomena A
might be explained as low initial inoculum resulting in less growth 200
of cell and low amount of ethanol. On the contrary, overgrowth of 100
180
cells by high initial inoculums in fermentation broth led to their
competitions on substrate, consequently reducing the conversion 160

Glucose concentration (g/L)

Ethanol concentration (g/L)

rate of glucose to ethanol. The result above illustrated that the
80
inoculum size had a significant effect for ethanol production with 140
Z. mobilis strain and, 10% was the optimum inoculums size. Trial I
120
Trial II 60
3.3. Scale-up open fermentation
100 Trial III

To study the possibility of industrial scale ethanol production 80 Trial I

from food waste by acid tolerant strain of ZMA7-2, open fermenta- 40
Trial II
tion was conducted on laboratory scale (250 ml) and bench scales 60
Trial III
(3 L and 20 L), respectively. The initial glucose concentration was
40 20
200.0 g/L and the inoculums size was 10% in all experiments.
Fig. 2B shows the growth curve of ZMA7-2 as a function of time. 20
In laboratory scale (250 ml) fermentation, stationary phase was
achieved at the time point of around 42 h. However, bench scales 0 0
(3 L and 20 L) fermentation was faster, with the stationary phase 0 6 12 18 24 30 36 42 48 54 60
of cell growth being reached at the 36 h and a higher cell concen-
Time (h)
tration was obtained. The difference of cell growth rate observed
between laboratory scale and bench scales fermentation might 5.0
due to the better homogenization provided by the reactor, and B
thus, enhancing mass transfer processes between the media and 4.5
the cells. Fig. 2A represented the kinetics of glucose consumption
4.0
and ethanol concentration during fermentation. In the case of lab-
oratory scale fermentation, 200.0 g/L of glucose was completely 3.5
exhausted around the 48 h and maximum ethanol concentration
Cell density (OD600)

was 98.31 g/L, achieved at the same time. In the bench scales (3 L 3.0
and 20 L) cases, the fermentation was both finished at the 44 h
with final ethanol concentration of 98.65 g/L and 99.78 g/L, respec- 2.5
tively. As expected, the process was faster than that of laboratory Trial I
scale due to higher cell concentrations. The results were summa- 2.0
rized in Table 3. Ethanol yield and volumetric ethanol productivity Trial II
1.5
after 48 h fermentation in laboratory scale was 0.49 g/g and
Trial III
2.03 g/L/h. In the case of bench scales (3 L and 20 L) fermentation, 1.0
the ethanol yield and volumetric ethanol productivity after 44 h
fermentation calculated as 0.49 g/g and 2.23 g/L/h, 0.50 g/g and 0.5
2.35 g/L/h, respectively. The glucose uptake rate and ethanol fer-
mentation performance of ZMA7-2 in reactor was relative higher 0.0
than in laboratory scale fermentation, which might contribute to 0 6 12 18 24 30 36 42 48 54 60
the agitation power and the different geometry of the fermentor.
Time (h)
Taken together, the results above indicated that industrial scale
ethanol production using food waste by ZMA7-2 was possible. Fig. 2. Ethanol production (A) and growth curves (B) by ZMA7-2 in different
fermentation scales. (A) Closed key, glucose; open key, ethanol. Trial I: laboratory
3.4. Ethanol production using recycled ZMA7-2 cells scale (250 mL); Trial II: bench scale (3 L); Trial III: bench scale (20 L).

productivity calculated was 0.49 g/g and 2.23 g/L/h in the first
The utilization of recycled cells for successive fermentation was
cycle, and 0.48 g/g and 2.30 g/L/h in the final cycle, respectively,
one of the strategies for cost effective ethanol production in indus-
suggesting that ZMA7-2 cell could be successfully used for five suc-
trial scale, since it reduced the labor cost and energy used for
cessive fermentation cycles. The results obtained demonstrated
inoculums preparation for each fresh batch. However, high ethanol
high tolerance to stress condition and stable fermentation perfor-
concentrations, osmotic stress due to sugar and acidity were con-
mance of recycled cells, and thus ZMA7-2 might be suitable for
sidered as stress conditions faced by microorganism, particularly
applying in continuous ethanol production. Furthermore, the cell
the recycled cells during the industrial process. Hence, strong tol-
density of fifth cycle declined 8% compared to that of first cycle
erance to stress conditions and stable ethanol production capabil-
which might due to cell loss occurred during separation and fer-
ity of recycled cells was necessary in successive fermentation
mentation process, suggesting that immobilization of microbial
(Dhaliwal et al., 2011). To examine the potential of using recycled
cells should be employed to future continuous fermentation to pre-
ZMA7-2 in successive fermentation, we performed a five cycles of
vent the cell loss from downstream process.
batch fermentation. The ZMA7-2 cells were collected after the glu-
cose was completely consumed, and then used as inoculum for the
next cycle. The ethanol concentrations obtained in each cycle were 3.5. Mass balance in the process
comparable (Table 4). Ethanol concentration slightly declined by
about 2% after the fifth cycle compared to ethanol obtained in The mass and energy balance of ethanol production by food
the first cycle. Moreover, the ethanol yield and volumetric ethanol waste obtained based on the bench scale experimental results
 K. Ma et al. / Bioresource Technology 203 (2016) 295–302 301

Table 3
Kinetics parameters of bioethanol production for different fermentation scales.

Parameters Fermentation time (h) Cell density Ethanol Volumetric ethanol productivity Ethanol yield Ethanol efficiency (%)a
Scales (OD600)a (g/L)a (g/L/h)a (g/g glucose)a
Laboratory scale (250 mL) 48 3.81 ± 0.15 98.31 ± 0.65 2.03 ± 0.02 0.49 ± 0.01 95.90 ± 0.15
Bench scale (3 L) 44 3.93 ± 0.20 98.65 ± 0.50 2.23 ± 0.15 0.49 ± 0.01 96.25 ± 0.22
Bench scale (20 L) 44 3.96 ± 0.15 99.78 ± 0.45 2.35 ± 0.03 0.50 ± 0.00 97.35 ± 0.10

The initial glucose concentration was 200.36 ± 0.45 g/l.

a
Mean values and standard deviations of three determinations.

Table 4
Evaluation of kinetic parameters for ethanol production using recycled Z. mobilis ZMA7-2 cells during successive fermentation cycles.

Parameters Fermentation time (h) Cell density (OD600)a Ethanol Volumetric ethanol productivity Ethanol yield (g/g glucose)a Ethanol efficiency (%)a
Scales (g/L)a (g/L/h)a
1st 44 3.86 ± 0.10 98.17 ± 0.50 2.23 ± 0.02 0.49 ± 0.01 96.00 ± 0.15
2nd 32 3.81 ± 0.20 98.34 ± 0.85 3.07 ± 0.03 0.49 ± 0.01 95.48 ± 0.20
3rd 32 3.75 ± 0.15 97.32 ± 0.67 3.04 ± 0.06 0.48 ± 0.00 95.15 ± 0.35
4th 38 3.66 ± 0.15 97.30 ± 0.90 2.56 ± 0.05 0.48 ± 0.02 95.08 ± 0.20
5th 42 3.55 ± 0.20 96.65 ± 1.02 2.30 ± 0.10 0.48 ± 0.02 94.41 ± 0.35
a
Mean values and standard deviations of three determinations.

Impurities Tab water Residual Solids Recycle oil

(15 kg) (50 kg) (52 kg) (5.5 kg)

Food waste Recyclable

(Recyclable and non-recyclable Sorting and Saccharification Solid and liquid Saccharified-liquid
organic materials
materials) (115 kg fresh basis) disrupting pulp (150 kg) separation (92 kg)
(100 kg)

Zymomonas mobilis

Concentrated food Concentration

Anhydrous ethanol Distillation Fermentation broth Ethanol
waste hydrolysates Freezing and thawing
(3.8 kg/4.8 L) Dehydration (37.5 kg) fermentation
(40.5 kg)
Glucose concentration: 200 g/L
Waste liquid CO2 H2O
(33.7 kg) (3 kg) (51.5 kg)

Fig. 3. Overall mass balance of fuel ethanol fermentation of food waste.

Table 5
and described in Fig. 3. 100 kg food waste could produce 3.8 kg Comparison of ethanol yields, storage, and potential ethanol production for raw
ethanol, 52 kg residual solid and 5.5 kg recycle oil in our process. biomasses and food waste in Japan.
In saccharified broth, 10% reducing sugar was composed of 9.1%
Raw Ethanol yield Storage Ethanol potential
glucose and 0.9% fructose. Fermentation using ZMA7-2 could material (L/kg) (104 t/year) (104 kL/year)
simultaneously convert all glucose and partial fructose, thus etha-
Wheata 0.43 9.6 4.10
nol yield calculated close to theoretical value of 0.50 g/g. Moreover, Molassesa 0.20 2.4 0.48
the component of residue solids and recycle oil was analyzed and Waste 0.21 735.0 39.00
summarized in Table 1. After saccharification, 3.8% sugars wooda
remained in residue solids, which could be converted into biogas Cull (ed) 0.15 370.0 28.00
wooda
(Fig. 1 in Supporting information: Part III). In addition, 100% of
Food wasteb 0.14 2000.0 80.00
combustible component in recycle oil suggested that it might be
a
potential feedstock for biodiesel production (Fig. 1 in Supporting Reference to Ministry of the Environment Government of Japan.
b
Our results.
information: Part II). It was reported that the amount of ethanol
produced from food waste was lower than that of starchy biomass,
i.e. wheat, corn, barley molasses, but comparable to those of ligno-
ZMA7-2, open (without sterilization) fermentation with high etha-
celluloses biomass, i.e. culled wood (Table 5). Because the cropland
nol yield achieved, suggesting our process was an economic pro-
limitation of Japan, crop based biomass storage was considerable
cess for low-cost industrial ethanol production.
lower than food waste, and thus could not support the large-
scale industrial ethanol production. On the other hand, the conver-
sion process of lignocelluloses biomasses to ethanol still too costly 4. Conclusions
since the complicated pretreatment process involved more energy
consumption. In this regard, using food waste for ethanol produc- In this study, a process of open ethanol fermentation from food
tion was more advantageous than starchy or lignocelluloses waste using an acid-tolerant mutant of Z. mobilis (ZMA7-2) was
biomass for its abundant storage and relatively simple pretreat- established. The mutant exhibited high tolerance to acid condition
ment process. Furthermore, by employing acid-tolerant strain of and ethanol production with ethanol yield of 0.50 g/g close to
302 K. Ma et al. / Bioresource Technology 203 (2016) 295–302
theoretical ethanol yield. Moreover, shorter fermentation time and
higher fermentation kinetics was achieved in bench scales ethanol
production, combing with the reusability of mutant cells, which
verified by a five cycle successive fermentation, we concluded that
ZMA7-2 was an excellent fermentation performer for food waste
based ethanol fermentation and the process developed was valu-
able for low-cost industrial ethanol production.
Acknowledgements
This research was financially supported by Open Funds of the
Key Laboratory of Development and Application of Rural Renew-
able Energy, Ministry of Agriculture, China (2015009), the National
Natural Science Foundation of China (Grant No. 31570055).
Partially supported by Youth Science and Technology Foundation
of Sichuan Province (2015JQO047).
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.biortech.2015.12.
054.
References
Agrawal, M., Mao, Z., Chen, R.R., 2011. Adaptation yields a highly efficient xylose-
fermenting Zymomonas mobilis strain. Biotechnol. Bioeng. 108, 777–785.
Arnold, C.N., McElhanon, J., Lee, A., Leonhart, R., Siegele, D.A., 2001. Global analysis
of Escherichia coli gene expression during the acetate-induced acid tolerance
response. J. Bacteriol. 183, 2178–2186.
Bai, F.W., Anderson, W.A., Moo-Young, M., 2008. Ethanol fermentation technologies
from sugar and starch feedstocks. Biotechnol. Adv. 26, 89–105.
Çakar, Z.P., Turanlı-Yıldız, B., Alkım, C., Yılmaz, Ü., 2012. Evolutionary engineering of
Saccharomyces cerevisiae for improved industrially important properties. FEMS
Yeast Res. 12, 171–182.
Choi, I.S., Kim, J.H., Wi, S.G., Kim, K.H., Bae, H.J., 2013. Bioethanol production from
mandarin (Citrus unshiu) peel waste using popping pretreatment. Appl. Energy
102, 204–210.
Choi, I.S., Lee, Y.G., Khanal, S.K., Park, B.J., Bae, H.J., 2015. A low-energy, cost-
effective approach to fruit and citrus peel waste processing for bioethanol
production. Appl. Energy 140, 65–74.
da Costa, J.A., Marques Jr., J.E., Gonçalves, L.R.B., Rocha, M.V.P., 2015. Enhanced
enzymatic hydrolysis and ethanol production from cashew apple bagasse
pretreated with alkaline hydrogen peroxide. Bioresour. Technol. 179, 249–259.
Dhaliwal, S.S., Oberoi, H.S., Sandhu, S.K., Nanda, D., Kumar, D., Uppal, S.K., 2011.
Enhanced ethanol production from sugarcane juice by galactose adaptation of a
newly isolated thermotolerant strain of Pichiakudriavzevii. Bioresour. Technol.
102, 5968–5975.
Goodman, A.E., Rogers, P.L., Skotnicki, M.L., 1982. Minimal medium for isolation of
auxotrophic Zymomonas mutants. Appl. Environ. Microbiol. 44, 496.
Han, W., Ye, M., Zhu, A.J., Zhao, H.T., Li, Y.F., 2015. Batch dark fermentation from
enzymatic hydrolyzed food waste for hydrogen production. Bioresour. Technol.
191, 24–29.
Hara, A., Radin, N.S., 1978. Lipid extraction of tissues with a low toxicity solvent.
Anal. Biochem. 90, 420–426.
He, M.X., Wu, B., Qin, H., Ruan, Z.Y., Tan, F.R., Wang, J.L., Hu, Q.C., 2014. Zymomonas
mobilis: a novel platform for future biorefineries. Biotechnol. Biofuels 7, 101.
Ho, S.H., Huang, S.W., Chen, C.Y., Hasunuma, T., Kondo, A., Chang, J.S., 2013.
Bioethanol production using carbohydrate-rich microalgae biomass as
feedstock. Bioresour. Technol. 135, 191–198.
Ho, D.P., Ngo, H.H., Guo, W., 2014. A mini review on renewable sources for biofuel.
Bioresour. Technol. 169, 742–749.
Horisawa, S., Ando, H., Ariga, O., Sakuma, Y., 2015. Direct ethanol production from
cellulosic materials by consolidated biological processing using the wood rot
fungus Schizophyllum commune. Bioresour. Technol. 197, 37–41.
Hua, Q., Joyce, A.R., Palsson, B.Ø., Fong, S.S., 2007. Metabolic characterization of
Escherichia coli strains adapted to growth on lactate. Appl. Environ. Microbiol.
73, 4639–4647.
Jin, Y., Chen, T., Chen, X., Yu, Z.X., 2015. Life-cycle assessment of energy
consumption and environmental impact of an integrated food waste-based
biogas plant. Appl. Energy 151, 227–236.
Kiran, E.U., Liu, Y., 2015. Bioethanol production from mixed food waste by an
effective enzymatic pretreatment. Fuel 159, 463–469.
Kiran, E.U., Trzcinski, A.P., Ng, W.J., Liu, Y., 2014. Bioconversion of food waste to
energy: a review. Fuel 134, 389–399.
Kiran, E.U., Trzcinski, A.P., Liu, Y., 2015. Enhancing the hydrolysis and methane
production potential of mixed food waste by an effective enzymatic
pretreatment. Bioresour. Technol. 183, 47–52.
Lee, D.H., Palsson, B.Ø., 2010. Adaptive evolution of Escherichia coli K-12 MG1655
during growth on a nonnative carbon source, L-1, 2-propanediol. Appl. Environ.
Microbiol. 76, 4158–4168.
Ma, K.D., Ruan, Z.Y., 2015. Production of a lignocellulolytic enzyme system for
simultaneous bio-delignification and saccharification of corn stover employing
co-culture of fungi. Bioresour. Technol. 175, 586–593.
Ma, K.D., Wakisaka, M., Sakai, K., Shirai, Y., 2009a. Flocculation characteristics of an
isolated mutant flocculent Saccharomyces cerevisiae strain and its application
for fuel ethanol production from kitchen refuse. Bioresour. Technol. 100, 2289–
2292.
Ma, H., Wang, Q., Qian, D., Gong, L., Zhang, W., 2009b. The utilization of acid-
tolerant bacteria on ethanol production from kitchen garbage. Renewable
Energy 34, 1466–1470.
Ma, K.D., Maeda, T., You, H.Y., Shirai, Y., 2014a. Open fermentative production of L-
lactic acid with high optical purity by thermophilic Bacillus coagulans using
excess sludge as nutrient. Bioresour. Technol. 151, 28–35.
Ma, H., Xing, Y., Yu, M., Wang, Q., 2014b. Feasibility of converting lactic acid to
ethanol in food waste fermentation by immobilized lactate oxidase. Appl.
Energy 99, 89–93.
Narendranath, N.V., Thomas, K.C., Ingledew, W.M., 2001. Effects of acetic acid and
lactic acid on the growth of Saccharomyces cerevisiae in a minimal medium. J.
Ind. Microbiol. Biotechnol. 26, 171–177.
Pleissner, D., Lam, W.C., Sun, Z., Lin, C.S.K., 2013. Food waste as nutrient source in
heterotrophic microalgae cultivation. Bioresour. Technol. 137, 139–146.
Saha, B.C., Nichols, N.N., Qureshi, N., Kennedy, G.J., Iten, L.B., Cotta, M.A., 2015. Pilot
scale conversion of wheat straw to ethanol via simultaneous saccharification
and fermentation. Bioresour. Technol. 175, 17–22.
Shui, Z.X., Qin, H., Wu, B., Ruan, Z.Y., Wang, L.S., Tan, F.R., He, M.X., 2015. Adaptive
laboratory evolution of ethanologenic Zymomonas mobilis strain tolerant to
furfural and acetic acid inhibitors. Appl. Microbiol. Biotechnol. 99, 5739–5748.
Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., Crocker, D., 2008.
Laboratory Analytical Procedure (LAP): Determination of Structural
Carbohydrates and Lignin in Biomass. Technical Report: NREL/TP-510-42618.
National Renewable Energy Laboratory, Golden Co, USA.
Tang, Y.Q., Koike, Y., Liu, K., An, M.Z., Morimura, S., Wu, X.L., Kida, K., 2008. Ethanol
production from kitchen waste using the flocculating yeast Saccharomyces
cerevisiae strain KF-7. Biomass Bioenergy 32, 1037–1045.
Tao, F., Miao, J.Y., Shi, G.Y., Zhang, K.C., 2005. Ethanol fermentation by an acid-
tolerant Zymomonas mobilis under non-sterilized condition. Process Biochem.
40, 183–187.
Tashiro, Y., Matsumoto, H., Miyamoto, H., Okugawa, Y., Pramod, P., Miyamoto, H.,
Sakai, K., 2013. A novel production process for optically pure L-lactic acid from
kitchen refuse using a bacterial consortium at high temperatures. Bioresour.
Technol. 146, 672–681.
Wirawan, F., Cheng, C.L., Kao, W.C., Lee, D.J., Chang, J.S., 2012. Cellulosic ethanol
production performance with SSF and SHF processes using immobilized
Zymomonas mobilis. Appl. Energy 100, 19–26.