Sustainable production of 5-hydroxymethyl furfural from glucose for process integration with high fructose corn syrup infrastructure

Article abstract of DOI:10.1039/d1gc00311a

5-Hydroxymethyl furfural (HMF) is a platform chemical, which can be derived from lignocellulosic biomass, and used for production of liquid fuels and polymers. We demonstrate a process for production of HMF using sequential enzymatic and catalytic reactions of glucose to synthesize HMF, and simulated-moving-bed (SMB) separation to purify HMF. The adsorption thermodynamic parameters of glucose, fructose, and HMF on a commercial chromatography resin are experimentally determined for modeling the SMB-based HMF production process. The experimental data are used to develop a rigorous process model and then estimate the cost of production. Chromatographic separation of HMF has 16% lower operating costs compared to an extraction-based process and has a minimum selling price of approximately $1478 per ton. We demonstrate that the HMF process can be integrated with the high fructose corn syrup (HFCS) process, and we performed analyses considering two systems including construction of a new integrated facility and retrofitting an existing HFCS facility to produce HMF. Our analyses suggest that the latter approach is a promising short-term low-risk strategy to advance the HMF production technology to commercial scale.

Full text of DOI:10.1039/d1gc00311a

Volume 23
Number 9
7
May 2021
Pages 3143-3488
Green
Chemistry
Cutting-edge research for a greener sustainable future
rsc.li/greenchem
ISSN 1463-9262
PAPER
James A. Dumesic et al.
Sustainable production of 5-hydroxymethyl furfural from
glucose for process integration with high fructose corn
syrup infrastructure

Green Chemistry
View Article Online
View Journal | View Issue
PAPER
Sustainable production of 5-hydroxymethyl
furfural from glucose for process integration with
high fructose corn syrup infrastructure†
Cite this: Green Chem., 2021, 23,
277
3
a
a
a
Hochan Chang, ‡ Ishan Bajaj,‡ Ali Hussain Motagamwala,
a
a
b,c
Arun Somasundaram,
James A. Dumesic
George W. Huber,
Christos T. Maravelias
and
a,b
*
5-Hydroxymethyl furfural (HMF) is a platform chemical, which can be derived from lignocellulosic
biomass, and used for production of liquid fuels and polymers. We demonstrate a process for production
of HMF using sequential enzymatic and catalytic reactions of glucose to synthesize HMF, and simulated-
moving-bed (SMB) separation to purify HMF. The adsorption thermodynamic parameters of glucose, fruc-
tose, and HMF on a commercial chromatography resin are experimentally determined for modeling the
SMB-based HMF production process. The experimental data are used to develop a rigorous process
model and then estimate the cost of production. Chromatographic separation of HMF has 16% lower
operating costs compared to an extraction-based process and has a minimum selling price of approxi-
mately $1478 per ton. We demonstrate that the HMF process can be integrated with the high fructose
corn syrup (HFCS) process, and we performed analyses considering two systems including construction
of a new integrated facility and retrofitting an existing HFCS facility to produce HMF. Our analyses suggest
that the latter approach is a promising short-term low-risk strategy to advance the HMF production
technology to commercial scale.
Received 27th January 2021,
Accepted 22nd March 2021
DOI: 10.1039/d1gc00311a
rsc.li/greenchem
7
(
ethylene furanoate) and poly(butylene furanoate). The
Introduction
polarity of furan ring in these monomers has been shown to
5
-Hydroxymethyl furfural (HMF) is a widely studied biomass- be replacements for PET that have improved barrier pro-
8
based platform chemical that has been used to produce liquid perties. Aldol condensation of HMF and acetone has been
fuels and bio-based polymers. Hydrodeoxygenation of HMF used for synthesizing a new platform chemical that can be
over CuRu catalyst can be employed to produce a mixture of used as an organic dye or monomers.
9
branched hydrocarbons and aromatic compounds, which can
be used as gasoline additive. Diesel and jet fuel can be pro- chemical and technological challenges for production of HMF
duced by aldol condensation of HMF to increase the carbon from biomass resources. For instance, glucose and fruc-
A large body of research has been performed to address the
1
1
0
11
2
–4
12
chain, followed by hydrodeoxygenation.
converted into 1,6-hexanediol and tetrahydrofuran dimethanol yield, respectively, in ionic liquid over an acid catalyst.
THFMD) which are both α–ω diols that can be used to Fructose was converted to HMF with 90% selectivity at 95%
HMF can also be tose can be dehydrated to produce HMF at 60% and 90%
(
5
13,14
produce polyesters and other polymers. Oxidation of HMF fructose conversion in organic solvent mixtures.
However,
can be carried out to produce an α–ω diacid, 2,5-furan dicar- the use of expensive and toxic solvents (e.g., ionic liquids,
6
boxylic acid (FDCA), a monomer used in the synthesis of poly organic solvent) negatively affected the process design.
Glucose has been utilized as a feedstock due to its inexpensive
market price and wide availability. Glucose is an widely avail-
able feed, but the unreacted glucose feed must be separated
a
Department of Chemical and Biological Engineering, University of Wisconsin–
Madison, Madison, WI, USA. E-mail: jdumesic@wisc.edu
b
from HMF to produce high purity HMF for applications
because the experimental yields of HMF from glucose are
DOE Great Lakes Bioenergy Research Center, University of Wisconsin-Madison,
Madison, WI, USA
c
15
Department of Chemical and Biological Engineering, Princeton University,
limited to be around 50–60%. Therefore, glucose is firstly
Princeton, NJ, USA
converted to fructose, either by isomerization over an immobi-
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/
d1gc00311a
These authors contributed equally.
1
6
17
lized enzyme or over heterogeneous catalysts, such as Sn-β,
1
8
followed by dehydration for production of HMF.
‡
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 3277–3288 | 3277

View Article Online
Paper
Green Chemistry
AVA Biochem developed a commercial scale (20 tons per lower operating cost compared to a MIBK extraction-based
1
8
year) process for production of HMF by using hydrothermal process and the process becomes more cost competitive with
carbonization and purifying HMF by crystallization. However, increase in plant scale. The minimum selling price (MSP) of
crystallization of HMF in aqueous solution, which is per- HMF was computed to be $1478 per ton for a facility proces-
formed to produce high purity HMF, requires a large energy sing 2000 kg h− glucose and producing 8.74 kton of HMF
input to operate refrigeration, and degradation of HMF during annually. We demonstrate that the proposed process can be
the aqueous dehydration reduces the production rate of HMF. integrated with a high fructose corn syrup (HFCS) process to
Accordingly, development of catalytic reactions with an allow sugar producers more flexibility in their product
effective strategy for separating HMF from unreacted sugars streams. We also develop a process model and perform TEA
and solvents has been a key challenge for production of HMF for the integrated process. Analyses are performed for two
from sugar resources since HMF degrades in presence of acid systems that include building a new HFCS/HMF facility (green-
1
1
9
19
catalyst via condensation and is also thermally unstable.
field) and retrofitting an existing HFCS facility with an HMF
An organic/aqueous bi-phasic solvent system was used to sim- production section (brownfield). The proposed process can be
2
0
ultaneously produce and separate HMF. Excess amounts of combined with bio-refineries that produce glucose from
2
4,25
26
salt, such as NaCl, were added to the aqueous phase to maxi- biomass,
mize HMF partitioning to the organic phase, but the use of
salt leads to equipment corrosion and high separation costs.
HMF was produced in high yield (63%) from glucose in an
and/or embed HMF upgrading process.
acetone/water mono-phasic solvent system, and the separation Experimental method
was achieved by HMF extraction in methyl isobutyl ketone
Materials
1
8
(MIBK) solvent. While extraction is an effective separation
The following materials were used as received: D-glucose
technique to isolate pure HMF from the solvent and unreacted
sugars, vacuum evaporation and refrigeration were required to
separate MIBK from HMF. Moreover, MIBK is an expensive
solvent which further increases the operating costs of the
process. Chromatographic separation can be an alternative
technique that minimizes the operating costs of the process by
reducing the refrigeration cost and eliminating the use of
additional solvents, like MIBK. Moving-bed chromatographic
separation is widely used in industry to purify high fructose
(
>99.5%, Sigma-Aldrich), D-fructose (>99%, Fisher chemical),
D-cellobiose (>98%, Sigma-Aldrich), D-mannose (>99%, Sigma-
Aldrich), 1,6-anhydro-β-D-glucose (levoglucosan, >99%, Sigma-
Aldrich), 5-hydroxymethyl furfural (HMF, 98% AK Scientific),
acetone (HPLC grade, Fisher Scientific), DOWEX monosphere
99 Ca/320 separation resin (DOW), Amberlyst-15 (hydrogen
form, Sigma-Aldrich), glucose isomerase (≥350 unit per g,
from Streptomyces murinus, Sigma-Aldrich), quartz wool (Ohio
Valley Specialty), Silicon dioxide (99.9%, 4–20 mesh, Sigma-
Aldrich), Mill-Q water (∼18 MΩ cm), 10 ml glass reactor
2
1
22
corn syrup and p-xylene from xylenes mixtures.
In this paper, we develop a process based on sequential
enzymatic isomerization and catalytic dehydration to produce
HMF from glucose in a green co-solvent (acetone/water
mixture), followed by chromatographic separation to purify
HMF. The solubilities of glucose and fructose in an acetone/
water solvent were examined to utilize a maximum concen-
tration of glucose feed at the upper solubility limit. The first
step in the process is isomerization of glucose into fructose by
using an immobilized glucose isomerase (from Streptomyces
murinus), in an acetone/water solvent system. The isomerized
solution, containing fructose and glucose, is then dehydrated
to produce HMF over a solid acid-catalyst, Amberlyst-15.
Reaction conditions for dehydration were determined to maxi-
mize the HMF yield. Single column chromatography was used
to separate HMF from the mixture of the unreacted glucose
and fructose. DOWEX 99 Ca/320 resin was used as a stationary
phase in chromatography since it is commercially used to sep-
arate glucose and fructose for the production of high fructose
(
(
Chemglass), and straight bore burette with PTFE stopcock
100 mL, Kimble Kimax, Class A).
Product analysis (HPLC)
The concentrations of the components of the isomerized and
dehydrated solution were quantified by High Performance
Liquid Chromatography (HPLC) analysis using a Bio-Rad
Aminex HPX-87H column on a Waters 2695 system equipped
with RI-2414 and PDA-2998 detectors in series. Glucose, fruc-
tose, HMF, sugar by-products (cellobiose, levoglucosan,
mannose), and levulinic acid were calibrated using standard
solutions. An aliquot of the product mixture was diluted 10
times with Milli-Q water and was filtered through a 0.2 μm
PTFE filter before analysis. The temperature of the HPLC
column was maintained at 50 °C, and the flow rate of the
mobile phase (pH 2 water, acidified by sulfuric acid) was kept
constant at 0.6 mL min− . Glucose, fructose, sugar by-pro-
ducts, and levulinic acid were analyzed by a RI detector, while
the HMF concentration was measured with a Waters 2998 PDA
detector at 320 nm. The conversions and yields were calculated
by the following equations.
1
2
3
corn syrup at industrial scale. Experiments were performed
to determine the adsorption parameters for the adsorption of
glucose, fructose, and HMF on the resin. The adsorption para-
meters were used to optimize simulated-moving-bed (SMB)
separation. Using the experimental data, a process model was
developed, and a techno-economic analysis (TEA) was per-
formed. TEA suggested that the proposed approach resulted in
Moles of final HMF
Moles of initial fructose
HMF yield from fructose ¼
ꢀ 100ð%Þ
3278 | Green Chem., 2021, 23, 3277–3288
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
the resin was filtered. These washing cycles were repeated 5
times to prepare the wet resin for chromatographic separation.
The chromatography column was prepared by filling a 100 mL
straight bore buret with quartz wool, silicon dioxide, and
Fructose conversion ¼
Moles of initial fructose ꢁ Moles of final fructose
ꢀ
100ð%Þ
Moles of initial fructose
4
2 mL (apparent volume) of wet resin (void fraction = 0.3). The
Glucose conversion ¼
desorbent flow was driven by gravity, resulting in monotonic
decrease of the desorbent velocity, because the level of desor-
bent in column decreased with time proceeding the separ-
ation. The linear velocity profiles of the loaded feed and the
desorbent during the separation are shown in Fig. S3.† Slow
flow rate of the loaded feed facilitated HMF separation from
sugars by maximizing the contact time of the chemicals and
the resin, whereas a faster flow of the desorbent was used to
recollect glucose and fructose after HMF collection. 42 mL of
the simulated solution was loaded on the column. Then, the
desorbent (acetone/water (80/20, v/v) solvent) was slowly added
Moles of initial glucose ꢁ Moles of final glucose
ꢀ
100 ð% Þ
Moles of initial glucose
Enzymatic isomerization of glucose in acetone/water (80/20,
v/v) solvent
0
.38 g of glucose was dissolved in 1.6 mL of Milli-Q water to
prepare an aqueous glucose solution (19 wt% glucose). 5 g of
acetone was added to the aqueous glucose solution for prepar-
ing acetone/water (80/20, v/v) solvent. 0.25 g of immobilized
glucose isomerase (from Streptomyces murinus) was ground
and added to the glucose feed. The glass vial reactor was
placed in an oil bath at 65 °C and the feed solution was stirred
at 560 rpm. Reaction time was varied from 10 to 180 min. After
the reaction, the product solution was quenched in water bath
to stop the reaction. 21 mol% of sugars were precipitated with
the enzyme as the temperature decreased from 65 to 20 °C.
The quenched solution was centrifuged at 3000 rpm for
(
Fig. S3(a)†) to initiate chromatography of the loaded sample
to separate HMF. After 85 mL of eluent collection for HMF iso-
lation, 2.5 L of desorbent was passed through the column with
fast low rate (Fig. S3(b)†) to recollect glucose and fructose. The
eluent samples from the chromatography were injected into
the HPLC after filtration by 0.2 μm PTFE filter without dilution
for analysis.
1
0
0 min to separate the precipitate from the liquid solution.
.5 mL of separated liquid solution was diluted 10 times with
Adsorption isotherms of glucose, fructose, and HMF on resin
The wet resin was dried by vacuum treatment on filter paper.
Different amounts of glucose (0.02, 0.04, 0.06, 0.08, 0.1 g) were
dissolved in 5 mL of acetone/water (80/20, v/v) solvent, and
Milli-Q water and filtered through a 0.2 μm PTFE filter before
analysis. 8 mL Milli-Q water was added to the precipitated
(mixture of glucose, fructose, enzyme) to dissolve the sugars in
0.1 g of dried resin was added to each glucose sample. Then,
aqueous solution. The aqueous sugar solution was analyzed
without dilution.
the set of five glucose solutions, containing the dried resin,
was placed in an oil bath at different temperatures (23, 40,
55 °C). Five fructose samples (0.02, 0.04, 0.06, 0.08, 0.1 g of
fructose with 0.1 g of dried resin in 5 mL of acetone/water
solvent) were prepared for fructose adsorption experiments.
Then, the set of five fructose samples was placed in an oil bath
at different temperatures (24, 40, 60 °C). A set of HMF adsorp-
tion experiments was prepared by mixing 0.02, 0.04, 0.06, 0.08,
Dehydration of simulated isomerization solution over
Amberlyst-15
A simulated isomerization solution was prepared as the feed
for dehydration reaction. 0.14 g of glucose and 0.23 g of fruc-
tose were dissolved in 1.6 mL of Milli-Q water. 5.0 g of acetone
was added into the aqueous sugar solution. 0.2 g of Amberlyst-
0.1 g of HMF with 0.1 g of dried resin in 5 mL of acetone/water
1
5 (40 μm to 150 μm) was mixed with the feed in 10 mL thick-
walled glass reactors (Chemglass) and placed in an oil bath at
20 °C. A triangular stir bar was used to stir the reaction
mixture at 750 rpm. Reaction time was varied from 40 to
00 min. After dehydration, the product solution was
solvent. Then, the set of HMF samples with resin was placed
in an oil bath at different temperatures (24 °C). All the
samples were stirred at 750 rpm for overnight (13–15 h).
1
0.2 mL of samples from HMF, glucose and fructose adsorption
1
experiments were diluted 3 times in Milli-Q water and filtered
by 0.2 μm PTFE filter for HPLC analysis. The adsorbed
amounts of chemicals were calculated by comparing the
chemical concentrations before and after the adsorption. The
adsorption equilibrium constants and adsorbed amounts
quenched in a water. The quenched solution was centrifuged
at 3000 rpm for 10 min to separate the solid catalyst. 0.5 mL of
the dehydrated solution was diluted 10 times with Milli-Q
water and filtered through a 0.2 μm PTFE filter for HPLC
analysis.
(
measured by HPLC) were calculated by the following
Column chromatography separation of glucose, fructose, and
HMF
equations.
KadsC
Langmuir isotherm adsorption : q ¼ qmax
0
.69 g of glucose, 0.09 g of fructose, and 0.70 g of HMF were
1 þ KadsC
dissolved in a mixture of Milli-Q water (9.57 g) and acetone
Linear isotherm adsorption : q ¼ KadsC
apparent volume) of dry resin (DOWEX 99 Ca/320) was stirred where, q: adsorbed amount of chemical (g chemical per g
(
(
30.11 g) to prepare the simulated solution in Fig. 2. 100 mL
in 400 mL of acetone/water (80/20, v/v) solvent for 15 min and resin), qmax: maximum adsorbed amount of chemical (g
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 3277–3288 | 3279

View Article Online
Paper
Green Chemistry
chemical per g resin), C: chemical concentration in solution levulinic acid, and humins. The selectivity of HMF, sugar by-
mol L− ), Kads: adsorption constant (L mol ).
1
−1
products, and levulinic acid were 78.6%, 15.2%, and 3.2%,
(
Mass of dissolved chemical before adsorption ꢁ Mass of dissolved chemical after adsorption
Adsorbed chemical amount ¼
ðg=gÞ
Mass of resin
respectively. 3.0% of glucose and fructose were converted to
water-soluble humins by acid-catalyzed self-condensation. The
humins can be selectively adsorbed on activated carbon and
Results and discussion
Enzymatic glucose isomerization and acid-catalyzed fructose
dehydration
18
purged by filtering the activated carbon. The three major
Immobilized glucose isomerases have been used for industrial chemicals in the dehydrated solution are HMF, glucose, and
scale processes of HFCS production due to their high activity fructose (Fig. S2†).
2
7
and stability. Over 500 tons of the immobilized enzymes are
synthesized and enable the production of approximately
Chromatographic separation of glucose, fructose, and HMF
2
8
1
0 million tons of HFCS per year. We used the commercial A simulated solution, containing the major three compounds
immobilized enzyme (≥350 unit per g) from Streptomyces (glucose, fructose, HMF) in the dehydrated solution (Fig. S2†),
murinus, to catalyze glucose isomerization to produce fructose
in an acetone/water (80/20, v/v) solvent at 65 °C. Note that the
use of acetone solvent enhanced the activity and stability of
2
9
the enzyme for isomerization. The concentration (5.4 wt%)
of glucose in an acetone/water (80/20, v/v) solvent was the
upper solubility limit at the reaction temperature (65 °C). As
shown in Fig. S1,† a 5.7 wt% glucose solution was not soluble
in acetone/water at 65 °C, whereas 5.4 wt% glucose solution
was soluble. Fructose has higher solubility (Fig. S1(c)†) than
glucose in an acetone/water solvent and shows mono-phasic
product solution at 65 °C after isomerization. Thermodynamic
equilibrium was obtained after 40 min of reaction, resulting in
6
1.3% glucose conversion and >98% fructose selectivity in
Fig. 1(a). Enzymatic isomerization reaches thermodynamic
equilibrium (Keq = 1.59) after 40 min of reaction at 65 °C, com-
pared to Sn-β catalyst (with the reaction quotient, Q = 0.75,
1
8
from previous work) at 80 °C after 180 min of reaction.
Fructose dehydration in an acetone/water (80/20, v/v) solvent
was investigated over Amberlyst-15 catalyst at 120 °C. The
highest HMF yield (78.7%) and 91.3% conversion of fructose
was obtained at 90 min as shown in Fig. 1(b). A reaction time
of 100 min resulted in decreased HMF yield due to the pro-
duction of levulinic acid by the acid-catalyzed rehydration of
3
0
HMF. The HMF selectivity varied between 80–86% during
fructose dehydration.
In subsequent experiments, HMF was produced from
glucose (5.4 wt%, 270 mM) by sequential reactions of enzy-
matic isomerization followed by acid-catalyzed fructose de-
hydration as shown in Fig. 2(a). Glucose was converted to fruc-
tose in an acetone/water (80/20, v/v) solvent by isomerization
without significant by-product (mannose, cellobiose, anhydro-
sugar) formation (<2% selectivity). A simulated solution
(
glucose and fructose mixture in an acetone/water solvent) was
used to investigate the dehydration since 25 mol% of glucose
and 18 mol% of fructose were precipitated with the enzyme Fig. 1 (a) Enzymatic isomerization of glucose (5.4 wt%, 270 mM) in
during the temperature quenching (from 65 to 20 °C) after iso- acetone/water (80/20, v/v) solvent at 65 °C (black dot: glucose conver-
sion, blue dot: fructose yield, pink diamond: total carbon balance); (b)
acid-catalyzed fructose (7.4 wt%, 388 mM) dehydration over Amberlyst-
merization. Fig. 2(b) shows the overall moles of consumed
sugars and the produced chemicals by the dehydration over
15 catalyst in acetone/water (80/20, v/v) solvent at 120 °C (blue dot:
Amberlyst-15. 88.9% fructose and 14.1% glucose were con-
verted to HMF and by-products, including sugar by-products, selectivity).
fructose conversion, red dot: HMF yield, purple triangle: HMF
3280 | Green Chem., 2021, 23, 3277–3288
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
tose required large volume of desorbent to be removed from
the resin. For example, when 92% of loaded fructose was recov-
ered, 58 times higher volume of desorbent was required than
the volume of the loaded feed (desorbent/feed (v/v) = 58)
(
Fig. 3(d)).
Adsorption model for glucose, fructose, and HMF on the
chromatography resin
Equilibrium adsorption isotherms for glucose, fructose, and
HMF on the chromatography resin (DOWEX 99 Ca/320) were
obtained at 23–24 °C (Fig. 4). Fructose was the most strongly
adsorbed component and followed a Langmuir adsorption iso-
therm at 24 °C. The adsorption equilibrium constant and the
maximum adsorbed amount of fructose were estimated to be
.13 ± 2.09 L mol− and 0.44 ± 0.08 g fructose per g resin (95%
1
7
confidence interval), respectively. Glucose was weakly
adsorbed and followed a linear isotherm adsorption model at
3 °C with 1.28 L mol− of adsorption equilibrium constant.
1
2
HMF showed no adsorption at 24 °C. The adsorption models
for glucose, fructose, and HMF explain the separation order of
the loaded chemicals in the chromatography separation.
Specifically, adsorption of fructose on resin contributes to the
longest elution time, whereas HMF passes through the column
fastest since it is not adsorbed on the resin.
Adsorption thermodynamic parameters of glucose and
fructose
The adsorption equilibrium constants of fructose and glucose
were measured at different temperatures to determine the
adsorption thermodynamic parameters (ΔHads, ΔSads). The
adsorption equilibrium constant of fructose was calculated to
be 7.13, 5.40, and 4.49 L mol− at 24, 40, and 60 °C, respect-
Fig. 2 (a) Concentrations of glucose, fructose and HMF in the reacted
solution; (b) moles of reacted and produced chemicals by the dehydra-
tion reaction in 8 mL dehydrated solution (reaction conditions: glucose
isomerization was operated at 65 °C for 40 min over enzyme; fructose
1
dehydration operated at 120 °C for 90 min over Amberlyst-15; acetone/ ively. The maximum adsorption capacity of fructose was deter-
water (80/20, v/v) solvent was used in both the reactions.).
mined to be 0.42 g fructose per g resin based on the Langmuir
isotherm adsorption model in Fig. 5(a). The glucose adsorp-
tion constant was 1.28, 1.22, and 1.17 L mol− at 23, 40, and
1
was prepared to demonstrate the chromatographic separation 55 °C, respectively (Fig. 5(b)). The adsorption enthalpy (ΔHads
)
at room temperature (23–25 °C). The three compounds (89% and entropy (ΔSads) of the fructose and glucose on the resin
mole fraction) were quantified by HPLC analysis, while each were calculated by Fig. 5(c) and (d). ΔHads and ΔSads for fruc-
−
1
by-product was not detected by HPLC because of its low con- tose adsorption were estimated to be −10582 ± 910 J mol
,
centration (2–4%) after dilution by the chromatographic desor- and −19.3 ± 2.9 J mol−
1
−1
K , respectively (90% confidence
bent. The concentration of the major chemicals (80 mM interval). For glucose adsorption, ΔH and ΔS were −2321
ads
−1 −1
ads
glucose, 10.5 mM fructose, 116 mM HMF) in the simulated ± 198 J mol− , and −5.8 ± 0.6 J mol
1
K
(90% confidence
solution was the same as the experimentally obtained products interval), respectively. The adsorption thermodynamic para-
Fig. 2(a)). The acetone/water (80/20, v/v) solvent was used as meters were used to optimize the simulated-moving-bed (SMB)
(
the desorbent during chromatography, and DOWEX 99 Ca/320 separation in techno-economic analysis model. The retention
as the chromatography resin. HMF was firstly eluted by the times of the by-products (e.g., cellobiose, mannose, anhydro-
chromatographic separation (Fig. 3(a)). Therefore, the initially sugar, levulinic acid) are similar (7–17 min) to glucose and
collected sample contained pure HMF (>99% purity by HPLC) fructose (9–10 min) in high-performance liquid chromato-
in the acetone/water desorbent. Then, glucose and fructose fol- graphy (HPLC) analysis (Fig. S4†), whereas retention time of
lowed the HMF collection. Glucose had a smaller retention HMF is relatively high (retention time: 32 min). This result
volume than fructose (inset of Fig. 3(a)). All the loaded HMF indicates that the by-products can be collected along with fruc-
was collected at desorbent/feed (v/v) ratio = 0.6 (Fig. 3(b)). tose by chromatographic separation. Thus, the adsorption
Similarly, 100% of loaded glucose and 61% of loaded fructose thermodynamic parameter of fructose was used to simulate
were collected at desorbent/feed (v/v) = 22 (Fig. 3(c)). The fruc- the separation of by-products for the SMB model in TEA.
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 3277–3288 | 3281

View Article Online
Paper
Green Chemistry
Fig. 3 (a) Concentration profiles of glucose (black), fructose (blue), and HMF (red) during the experimental chromatography separation at room
temperature (23–25 °C). Acetone/water (80/20) solvent was used as desorbent to separate HMF, glucose and fructose. Inset figure represents
magnified concentration profiles of glucose and fructose (solid lines are the visual guidance); comparison of chemical amounts ‘before (black bar)’
and ‘after (grey bar)’ the chromatography at different desorbent/feed (v/v) ratios, (b) desorbent/feed (v/v) = 0.6, (c) desorbent/feed (v/v) = 22, (d) des-
olvent/feed (v/v) = 58 (loaded feed conditions: 80 mM glucose, 10.5 mM fructose, 116 mM HMF in acetone/water (80/20, v/v) solvent).
System-level studies of HMF production process
We developed a process model for production of HMF from
glucose (shown in Fig. 6(a)), using the experimental data in
this paper and Aspen Plus Process Simulator (V10.0 Aspen
Technology). A complete material balance is shown in
Table S1.† These data were then used to perform a techno-
economic analysis. Experimental results were used to model
the reactors and adsorption column for separating humins,
whereas Aspen was used to model the evaporators. The
glucose feed is combined with an acetone/water stream and a
2
glucose, fructose, H O, and acetone recycle stream and fed
into the isomerization reactor (R-1). The products from R-1 are
then dehydrated in R-2. The products are then passed over an
activated carbon bed that removes the humins. It is assumed
that 1% of sugars and by-products are also lost with humins.
The products are then sent to a simulated moving bed (SMB)
separator (S-2). S-2 separates the HMF form the glucose and
fructose. We modeled a SMB for separation which is more
industrially practical than batch chromatography as used in
Fig. 4 Isothermal adsorption of fructose (blue triangle), glucose (black
circle), and HMF (red cross) on DOWEX 99 Ca/320 at 23–24 °C.
Adsorption model: Langmuir isotherm adsorption (blue line), linear iso-
therm adsorption (black line); adsorption temperature was at 23–24 °C.
3282 | Green Chem., 2021, 23, 3277–3288
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
Fig. 5 Adsorption isotherms of glucose and fructose at different temperatures (a) Langmuir isotherm adsorption of fructose at different tempera-
tures (circle: 24 °C, triangle: 40 °C, cross: 60 °C), (b) linear isotherm adsorption of glucose at different temperatures (circle: 23 °C, triangle: 40 °C,
cross: 55 °C); Van’t Hoff plot of (c) fructose adsorption and (d) glucose adsorption.
our experiments. The batch chromatography uses a higher
Based on the optimization results of SMB, the volumetric
volume of desorbent than SMB and is not a continuous flow rate of desorbent (stream 14) needs to be 1.12 times the
3
1
process.
SMB was modeled and simulated in Aspen volumetric flow rate of feed (stream 6). A part of the desorbent
Chromatography (V10.0 Aspen Technology) using the isotherm requirement is fulfilled by stream 12 and the rest is satisfied
parameters obtained in Fig. 4. A molecular sieve column (S-4) by separating the solvent from the extract. First, stream 9 is
was used to remove excess water formed in the dehydration split in two streams. One of them (stream 15) is passed
reactor. We optimized the design and operating conditions of through a flash column operating at 90 °C to separate sugars
SMB, and obtained HMF recovery and purity of 96.5% and from the solvent system. The solvent stream (stream 17) is
9
6.8%, respectively in the raffinate stream using a desorbent/ combined with stream 12 to satisfy the requirement on desor-
feed ratio (v/v) of 1.12 (Fig. S5†). bent amount. To prevent accumulation of by-products, a frac-
00% of glucose and 85.3% of fructose are recovered in the tion of stream 18 is discharged and sent to the waste treatment
1
extract stream (stream 9), whereas 96.5% of HMF is recovered facility. The remaining fraction (stream 19) is combined with
in the raffinate stream (stream 8). It is assumed that the sugar stream 16 and recycled to the isomerization reactor. A base
by-products and levulinic acid are recovered in the extract design case is simulated by considering the glucose feed flow
stream. The volumetric flow rate of the raffinate is 0.97 times rate of 2000 kg h− resulting in annual HMF production of
the SMB feed (stream 6). Stream 8 is then passed through a 8.74 kton. The material balance information for the main
series of flash columns operating at low temperatures and streams is given in Table S1.†
1
pressures to separate HMF from the solvent. Stream 10 con-
Next, we estimated the total energy requirements of the
taining acetone/water solvent is passed through a molecular process. It is assumed that steam, cold water, refrigerant, and
sieve adsorber to remove excess water formed in the dehydra- electricity are available for purchase to meet the heating,
tion reactor. It is assumed that 1% of acetone is also lost with cooling, and power requirements, which are 15.4 MW, 19.3
water.
MW, and 24.2 kW, respectively. 24% of the cooling require-
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 3277–3288 | 3283

View Article Online
Paper
Green Chemistry
Fig. 6 (a) Overall process flow diagram for production of HMF; (b) a summary of costs and revenue of the proposed process; (c) comparison of
MSP of extraction and SMB-based approaches with feedstock flowrate.
ments are satisfied by the refrigerant and the remaining by
cooling water.
The operating costs of the process comprise of fixed and
variable components. The fixed operating costs consist of
After modeling the process in Aspen Plus, capital and oper- labor and other overhead items. The variable operating costs
ating costs of the process were estimated. To provide a fair are estimated based on the utility, catalyst, and raw material
1
8
comparison with another HMF production process, the (feedstock and makeup chemicals) consumption. The total
equipment and material costs are adjusted to a common basis operating costs are estimated to be $8.9 million per year
year of 2017. The unit installed cost, C, for capacity S in the (summary in Table S4†).
cost-year is computed using,
Based on the estimated capital and operating costs, and the
economic parameters listed in Table S5,† the minimum selling
price (MSP) of HMF was computed using discounted cash flow
analysis. The MSP of HMF was calculated to be $1478 per ton
ꢀ
ꢁ ꢀ
ꢁ
β
S
Icost
Ibase
C ¼ C0
ð1Þ
S0
(
Fig. 6(b)) for the base design.
where C
0
is the installed cost for capacity S
0
in the base year,
We also compared the economics of the proposed approach
and Icost and Ibase are the plant cost indices in the cost and with an extraction-based process, where methyl isobutyl
base years, respectively. The parameters in eqn (1) of the main ketone (MIBK) was used as a solvent to extract HMF from
1
8
equipment are given in Table S2.† Once the total equipment sugars. After extraction, HMF was separated from MIBK by
cost has been determined, the total capital investment (TCI) is vacuum evaporation. Thus, vacuum evaporation was required
estimated by considering other direct and indirect costs to separate (i) MIBK from HMF and (ii) HMF from the solvent
(
summary in Table S3†). For the base design of the HMF plant system. The SMB approach does not require MIBK. In the SMB
producing 8.74 kton per year, TCI is estimated to be approach, vacuum evaporation is required only once to separ-
$
31.8 million.
ate HMF from the solvent. A summary of the differences
3284 | Green Chem., 2021, 23, 3277–3288
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
Table 1 Economic analysis of producing 8.74 kton HMF annually using extraction-based and SMB-based approaches
Glucose concentration
(mM)
Yield (g HMF per
g glucose)
Capital costs
($ per ton)
Variable operating
costs ($ per ton)
Fixed operating
costs ($ per ton)
MSP
($ per ton)
Process
Extraction-based
SMB-based
222
270
0.63
0.5
357
459
961
804
165
215
1483
1478
between the two approaches is given in Table 1. The extrac- MSP of HMF increases to $1800. In general, capital costs are
tion-based approach has a lower capital, higher yield, and subject to significant uncertainty, especially when pioneer
3
2
higher variable operating costs than the SMB approach. The plants are constructed. A ± 50% variation in total capital
higher capital cost for the SMB approach is because of high investment changes the MSP by 22.7%. For the sake of simpli-
SMB cost. The higher yield of the extraction-based approach is city, fixed operating costs (FOC) are computed based on the
because of higher selectivity in dehydration reaction. The MSP capital investment for the base design case. However, for a
of the extraction-based approach is slightly less than the pro- small-scale plant, this method most likely underestimates the
posed SMB-based approach for the base case.
labor costs component of FOC. We estimate the labor costs
The equipment costs and fixed operating costs exhibit econ- based on the number of operators and see that the FOC
omies of scale. However, the variable operating costs, which increases by $1 million, resulting in an increase in MSP by
account for a significant share of the total operating costs for 7.6%.
the two processes, vary linearly with scale. A comparison of the
We study the impact of SMB performance on MSP of HMF.
variable operating costs of the two approaches is shown in The results obtained in the current study are based on conven-
Fig. S6.† The SMB approach leads to 16% reduction in the vari- tional SMB comprising of four sections with five columns in
able operating costs compared to the extraction-based each section. The development of alternative operation modes
approach mainly due to the elimination of MIBK use and (e.g., dynamic configuration variations and flow rate modu-
3
3
reduced refrigerant consumption. Thus, the proposed lations) and optimization of column configurations have
approach becomes more competitive with increase in the shown to enhance product purity, recovery, and productivity,
process scale as illustrated in Fig. 6(c). The analysis suggests and reduce the required number of columns. We study two
that for a facility producing more than 8.39 kton per year of parameters related to SMB performance – desorbent/feed ratio
HMF, it is more economical to use the SMB-based approach.
and HMF recovery. While increasing the former increases the
We performed sensitivity analyses with respect to technical capital costs and energy consumption of the evaporator S-5
(
e.g. HMF recovery using SMB and desorbent/feed ratio for and increases the MSP, increasing the latter improves the
SMB) and economic parameters (e.g. refrigerant prices, ROI overall HMF yield and reduces the MSP.
discount rate, total capital investment, and fixed operating
costs) to identify the major cost drivers. Each variable was Integration with HFCS production facility
changed to a minimum and maximum value with all other
High fructose corn syrup (HFCS) is a sweetener made from
variables held constant. The results are shown in Fig. 7.
Although refrigerant satisfies only 24% of the cooling require-
ments, it contributes moderately (8%) to the MSP because of
its high price. A 20% reduction in the refrigerant price leads to
milled corn and used predominantly in food and beverage
industries. It became popular in the late 1970s when the price
of regular sugar was high. However, due to the ongoing debate
regarding its correlation to obesity and heart disease, the
industry has experienced decreased demand over the past
several years. We now show that by integrating HMF pro-
duction process in a corn biorefinery producing HFCS, the
economics of the biorefinery can be improved.
1
.6% reduction in the MSP. Internal rate of return (IRR) also
greatly impacts the MSP. If the IRR is increased to 20%, the
An integrated process for the production of HFCS and HMF
is shown in Fig. 8. First, corn is wet milled to recover and
purify starch and other coproducts. The first step in wet
milling involves softening the texture of corn kernel by
soaking it in water and sulfur dioxide. The grain is then
ground and the constituents are separated by screening,
washing, hydrocloning, and centrifuging. More details on the
3
4
wet milling process can be found in Blanchard. The end pro-
ducts of wet milling are germ, corn gluten feed, corn gluten
meal, and starch. The germ is used to produce corn oil, which
has applications in food and cosmetic industries. Corn gluten
feed is high fiber, low protein feed for beef cattle. Corn gluten
meal contains high protein and low fiber, used to feed poultry
Fig. 7 Sensitivity analysis results.
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 3277–3288 | 3285

View Article Online
Paper
Green Chemistry
Fig. 8 An overview of integrated HFCS/HMF process.
and swine. While starch can be used as a food additive or
adhesive with minimal further processing, its conversion to
sweeteners is economically more favorable. Thus, starch is
hydrolyzed to yield a syrup with over 95% glucose (dry solids
basis) and the remaining sugars are maltose, maltotriose with
traces of higher oligomerized sugars. A fraction of the glucose
stream is sent to the syrup production section, where glucose
is converted to fructose such that concentration of fructose is
4
2% on a dry weight basis. The converted HFCS-42 syrup is
further purified and concentrated. Generally, the remaining
glucose stream is sold as a coproduct, but we propose to use it
to produce HMF. We perform technoeconomic analyses for the
integrated HFCS/HMF process considering two systems for
different fractions of glucose going to either produce HMF or
HFCS. First, we analyze the economics of building a new facil-
ity for the integrated process. Second, we examine retrofitting
an existing HFCS production facility to produce HMF.
3
5
We use SuperPro Designer model by Intelligen to perform
mass and energy balances, and estimate capital and operating
costs of the HFCS process. A base design is considered with a
−
1
corn feed flow rate of 50 000 kg h . TCI and operating costs
are estimated to be $189.4 million and $123.7 million per year,
respectively for the greenfield facility. The relevant parameters
for the analysis are given in Table S6.†
We analyze the process and economic conditions that make
the integrated HFCS/HMF process economically more favor-
able than the HFCS process. The results of the analyses are
shown in Fig. 9. One of the glucose streams from the starch
hydrolysis section is used to produce HFCS-42, whereas the
other stream is either sold as a by-product in HFCS process or
utilized to produce HMF in HFCS/HMF process. The combined
HFCS/HMF process requires more capital investment than the Fig. 9 Economic analyses and comparison of greenfield HFCS/HMF
and HFCS systems. (a) MSP of HFCS-42 and IRR as a function of fraction
of glucose sent to syrup production section ( fg→HFCS) for HFCS and
HFCS/HMF processes. (b) Difference in the net present values (NPV) of
HFCS/HMF and HFCS processes as a function of difference in glucose
HFCS process alone. The amount of HMF produced is less
compared to glucose produced by HFCS process. Therefore,
the selling price of HMF needs to be higher for the integrated
HFCS/HMF process to be more economically favorable. For the
analysis shown in Fig. 9(a), we have assumed HMF price to be
and HMF selling price and fg→HFCS
.
$
1500 per ton. To understand the effect of changing the pro-
duction rate of HFCS-42 on the economics of the two pro- than 185 kton per year (fraction of glucose sent to syrup pro-
cesses, we estimate the MSP of HFCS-42 and IRR. MSP is esti- duction, fg→HFCS = 0.93). Thus, a corn refinery using integrated
mated based on 15% IRR. The results are shown in Fig. 9(a). HFCS/HMF process can afford to offer HFCS at a lower price
While the MSP of HFCS-42 increases with decrease in its pro- and be more competitive. We also observe that IRR decreases
duction rate for the two processes, we see that MSP is lower for with decrease in production rate, however, it decreases more
HFCS/HMF process if the production rate of HFCS-42 is lower rapidly for HFCS process. The analysis suggests that in com-
3286 | Green Chem., 2021, 23, 3277–3288
This journal is © The Royal Society of Chemistry 2021

View Article Online
Green Chemistry
Paper
parison to glucose, HMF compensates for the loss in revenue
due to lower HFCS production to a higher degree and there-
fore, it is more economical to convert the glucose stream to
produce HMF.
Clearly, for HFCS/HMF process to be more profitable, the
selling price of HMF needs to be higher than glucose and it
needs to be produced at sufficiently large scale to take advan-
tages of economies of scale. Fig. 9(b) shows that with increase
in the price difference between HMF and glucose, and
decrease in HFCS-42 production, HFCS/HMF process becomes
more favorable.
Now, we consider the brownfield system wherein an existing
HFCS facility that is fully depreciated is retrofitted to integrate
an HMF production section. Compared to a stand-alone HMF
production system, this system is expected to reduce costs and
investment risks because of existing energy recovery facilities,
experienced workforce and a reliable source of feedstock. In a
Fig. 10 Economic analysis for retrofitting an existing HFCS process to
produce HMF. Marginal payback period and marginal return on invest-
ment are shown as a function of HMF selling price for three different
3
6
study by de Jong et al., the benefits of retrofitting were quan-
tified and the authors estimated a 12% reduction in total values of fg→HFCS
capital investment compared to a stand-alone plant because of
reduced capital expenditures for buildings, service facilities,
.
land and yard works. A comparison of TCI for a stand-alone
HMF, greenfield and brownfield HFCS/HMF systems for
different production capacities of HMF is given in Table 2. TCI
for a brownfield system is an order of magnitude lower than a
greenfield system and 12% lower than a stand-alone HMF
system. Moreover, with declining market of HFCS, it is less
risky to retrofit an existing HFCS plant than constructing a
new HFCS/HMF facility.
chosen to be 0.9, 0.95, and 0.98 corresponding to an annual
HMF production of 17.5, 9.2, and 3.5 kton, respectively.
Increasing the production rate of HMF allows us to take
advantages of economies of scale, resulting in lower MPBP
and higher MROI.
Conclusions
To determine whether retrofitting an existing HFCS plant
to integrate an HMF production section is advantageous, we We have studied a process for HMF production from glucose
use marginal payback period (MPBP) and marginal return on followed by chromatographic separation to purify HMF.
investment (MROI) as the performance metrics. MPBP is the Glucose feed (5.4 wt%) was dissolved in an acetone/water
amount of time required to recover the capital investment for cosolvent at the isomerization temperature (65 °C). Enzymatic
building the HMF production section through additional isomerization of glucose produced 61.3% yield of fructose with
profit obtained by selling HMF. It is defined as the ratio of >98% selectivity. The isomerized solution was dehydrated to a
TCI and sum of the difference in profits of HFCS/HMF and solution containing HMF, glucose, and fructose as major com-
HFCS processes and depreciation. MROI is the additional ponents (49, 32, and 7 mol%, respectively). We demonstrated
return possible because of investing in HMF production that chromatography effectively separated HMF from the
section. It is defined as the ratio of the difference in the unreacted glucose and fructose because HMF was not
profits of the two processes and TCI. The fraction of glucose adsorbed on the chromatography resin at 24 °C. In contrast to
from starch hydrolysis section sent to syrup production HMF, fructose and glucose were adsorbed on the resin and fol-
section ( fg→HFCS) and HMF price significantly affect the lowed Langmuir and linear isotherm adsorption models,
process economics. The effect of the two factors on MPBP respectively. The thermodynamic parameters of the sugar
and MROI are shown in Fig. 10. The values of fg→HFCS are adsorptions were determined experimentally and used in
simulations to evaluate the economic potential of the process
by optimizing the performance of simulated-moving-bed
(
SMB) separation. Techno-economic analysis indicates that the
Table 2 TCI for a stand-alone HMF system, greenfield and brownfield
HFCS/HMF systems
process can produce HMF at an MSP of $1478 per ton of HMF
for the base design (2000 kg h− of glucose feed flow rate, 8.74
kton per year of HMF production). A comparison of the econ-
omics of the SMB-based process to an extraction-based
1
TCI for
TCI for
TCI for
brownfield
HFCS/HMF
HMF
production
stand-alone
HMF system
(MM$)
greenfield
HFCS/HMF
system (MM$)
1
8
(kton per year)
system (MM$) process illustrates that the SMB-based approach is economi-
cally more favorable when the plant produces more than 8.39
kton HMF per year. We demonstrate that the proposed process
28.9
3
9
7.5
.5
.2
18.4
32.8
48.3
208.4
222.5
237.4
16.2
1
42.5
can be integrated to the existing infrastructure, such as high
This journal is © The Royal Society of Chemistry 2021
Green Chem., 2021, 23, 3277–3288 | 3287

View Article Online
Paper
Green Chemistry
fructose corn syrup (HFCS) production process and performed 11 S. Hu, Z. Zhang, J. Song, Y. Zhou and B. Han, Green Chem.,
TEA to determine the process and economic conditions that 2009, 11, 1746–1749.
make the integration economically favorable. We considered 12 J. B. Binder and R. T. Raines, J. Am. Chem. Soc., 2009, 131,
two systems that include constructing a new HFCS/HMF 1979–1985.
process and retrofitting an existing HFCS process to produce 13 J. N. Chheda, Y. Román-Leshkov and J. A. Dumesic, Green
HMF. While the former system has high TCI, the latter is a par- Chem., 2007, 9, 342–350.
ticularly useful short-term strategy with low risk and low 14 J. N. Chheda and J. A. Dumesic, Catal. Today, 2007, 123,
investment costs to enable the transition of emerging HMF
production technology to commercial scale.
59–70.
15 T. Wang, M. W. Nolte and B. H. Shanks, Green Chem., 2014,
16, 548–572.
1
1
1
1
2
2
2
2
6 A. Converti and M. Del Borghi, Bioprocess Eng., 1997, 18,
27–33.
Conflicts of interest
7 Y. Román-Leshkov, M. Moliner, J. A. Labinger and
M. E. Davis, Angew. Chem., Int. Ed., 2010, 49, 8954–8957.
8 A. H. Motagamwala, K. Huang, C. T. Maravelias and
J. A. Dumesic, Energy Environ. Sci., 2019, 12, 2212–2222.
9 P. Y. Nikolov and V. A. Yaylayan, J. Agric. Food Chem., 2011,
Authors declare that there is no conflict of interest.
Acknowledgements
59, 10104–10113.
This material is based upon work supported in part by the
Great Lakes Bioenergy Research Center, U.S. Department of
Energy, Office of Science, Office of Biological and
Environmental Research under Award Number DE-SC0018409
and in part by U.S. Department of Energy under Award
Number DE-EE0008353. We thank Dr Kefeng Huang for pro-
viding the model of extraction-based process. We thank
Dr Demetri Petrides at Intelligen for his help with the HFCS
model.
0 F. K. Kazi, A. D. Patel, J. C. Serrano-Ruiz, J. A. Dumesic and
R. P. Anex, Chem. Eng. J., 2011, 169, 329–338.
1 A. Toumi and S. Engell, Chem. Eng. Sci., 2004, 59, 3777–
3
792.
2 M. Minceva and A. E. Rodrigues, Ind. Eng. Chem. Res., 2002,
1, 3454–3461.
3 D. J. Gisch, C. H. Martin and C. R. Eicher, U.S. Patent,
0180001228, 2018.
4
2
2
2
4 K. L. Smiley, Biotechnol. Bioeng., 1971, 13, 309–317.
5 J. Michels, in 4.3.2 Starch Biorefinery, ed. J. Pietzsch,
Springer Verlag, 2017, pp. 91–94.
Notes and references
26 H. Chang, A. H. Motagamwala, G. W. Huber and
1
2
3
4
5
Y. Román-Leshkov, C. J. Barrett, Z. Y. Liu and
J. A. Dumesic, Nature, 2007, 447, 982–985.
R. M. West, Z. Y. Liu, M. Peter, C. A. Gärtner and
J. A. Dumesic, J. Mol. Catal. A: Chem., 2008, 296, 18–27.
R. M. West, Z. Y. Liu, M. Peter and J. A. Dumesic,
ChemSusChem, 2008, 1, 417–424.
G. W. Huber, J. N. Chheda, C. J. Barrett and J. A. Dumesic,
Science, 2005, 308, 1446–1450.
J. He, K. Huang, K. J. Barnett, S. H. Krishna, D. M. Alonso,
J. A. Dumesic, Green Chem., 2019, 21, 5532–5540.
27 V. J. Jensen and S. Rugh, Methods Enzymol., 1987, 136, 356–
370.
28 R. Di Cosimo, J. Mc Auliffe, A. J. Poulose and G. Bohlmann,
Chem. Soc. Rev., 2013, 42, 6437–6474.
29 K. M. Vilonen, A. Vuolanto, J. Jokela, M. S. A. Leisola and
A. O. I. Krause, Biotechnol. Prog., 2004, 20, 1555–1560.
30 D. M. Alonso, J. M. R. Gallo, M. A. Mellmer, S. G. Wettstein
and J. A. Dumesic, Catal. Sci. Technol., 2013, 3, 927–931.
Z. J. Brentzel, S. P. Burt, T. Walker, W. F. Banholzer, 31 G. Dünnebier and K. U. Klatt, Chem. Eng. Sci., 2000, 55,
C. T. Maravelias, I. Hermans, J. A. Dumesic and
G. W. Huber, Faraday Discuss., 2017, 202, 247–267.
A. H. Motagamwala, W. Won, C. Sener, D. M. Alonso,
C. T. Maravelias and J. A. Dumesic, Sci. Adv., 2018, 4,
eaap9722.
J. C. Morales-Huerta, A. M. De Ilarduya and S. Muñoz-
Guerra, Polymer, 2016, 87, 148–158.
L. Sun, J. Wang, S. Mahmud, Y. Jiang, J. Zhu and X. Liu,
Eur. Polym. J., 2019, 118, 642–650.
373–380.
32 E. W. Merrow, K. E. Phillips and C. W. Myers,
Understanding Cost Growth and Performance Shortfalls in
Pioneer Process Plants, 1981.
33 A. Rajendran, G. Paredes and M. Mazzotti, J. Chromatogr. A,
2009, 1216, 709–738.
34 P. Blanchard, Technology of corn wet milling and associated
processes, Elsevier, 1992.
35 D. Petrides, Corn Refinery - Modeling and Evaluation with
SuperPro Designer, 2016.
6
7
8
9
H. Chang, I. Bajaj, G. W. Huber, C. T. Maravelias and
J. A. Dumesic, Green Chem., 2020, 22, 5285–5295.
0 T. W. Walker, A. H. Motagamwala, J. A. Dumesic and
36 S. de Jong, R. Hoefnagels, A. Faaij, R. Slade, R. Mawhood
and M. Junginger, Biofuels, Bioprod. Biorefin., 2015, 9, 778–
800.
1
G. W. Huber, J. Catal., 2019, 369, 518–525.
3288 | Green Chem., 2021, 23, 3277–3288
This journal is © The Royal Society of Chemistry 2021