A scalable carboxylation route to furan-2,5-dicarboxylic acid

Article abstract of DOI:10.1039/c7gc01059a

Furan-2,5-dicarboxylic acid (FDCA) is a biomass-derived diacid that can be used to make polymers including polyethylene furandicarboxylate (PEF), a highly attractive substitute for petroleum-derived polyethylene terephthalate (PET). Current FDCA syntheses require edible fructose as the feedstock, entail a difficult oxidation step that generates undesirable aldehyde impurities, and have moderate yields. As an alternative, carbonate-promoted C-H carboxylation enables the synthesis of FDCA from 2-furoic acid and CO₂. This route is potentially advantageous because 2-furoic acid is made from furfural, a feedstock produced commercially from inedible lignocellulosic biomass, and it obviates late-stage oxidation. In the carboxylation reaction, salt mixtures composed of alkali furan-2-carboxylate (furoate) and alkali carbonate (M₂CO₃) are heated under CO₂ in the absence of solvent or catalysts to form furan-2,5-dicarboxylate (FDCA^2-), which is subsequently protonated to produce FDCA. Previously, high yields were achieved on small-scale reactions using caesium furoate and Cs₂CO₃. In this work, we investigate the carboxylation reaction using alkali furoate/M₂CO₃ salts containing cation blends and describe reaction conditions that provide high yields on a preparative scale. We show that the carboxylation proceeds efficiently with K⁺/Cs⁺ blends that have a high K⁺ content (up to 4:1 K⁺:Cs⁺). Removing H₂O, which is a by-product of the reaction, is important for suppressing decomposition pathways. The accumulation of the FDCA^2- product inhibits the reaction. Integrating these lessons, we demonstrate the carboxylation of furoate on a 1 mol scale using a fixed-bed flow reactor with 89% isolated yield of pure FDCA upon protonation.

Full text of DOI:10.1039/c7gc01059a

Green Chemistry
View Article Online
View Journal
PAPER
A scalable carboxylation route to furan-2,5-
dicarboxylic acid†
Cite this: DOI: 10.1039/c7gc01059a
Graham R. Dick, Amy D. Frankhouser, Aanindeeta Banerjee and
Matthew W. Kanan
*
Furan-2,5-dicarboxylic acid (FDCA) is a biomass-derived diacid that can be used to make polymers
including polyethylene furandicarboxylate (PEF), a highly attractive substitute for petroleum-derived poly-
ethylene terephthalate (PET). Current FDCA syntheses require edible fructose as the feedstock, entail a
difficult oxidation step that generates undesirable aldehyde impurities, and have moderate yields. As an
alternative, carbonate-promoted C–H carboxylation enables the synthesis of FDCA from 2-furoic acid
2
and CO . This route is potentially advantageous because 2-furoic acid is made from furfural, a feedstock
produced commercially from inedible lignocellulosic biomass, and it obviates late-stage oxidation. In the
carboxylation reaction, salt mixtures composed of alkali furan-2-carboxylate (furoate) and alkali carbonate
(
M
2
CO
3
) are heated under CO
2
in the absence of solvent or catalysts to form furan-2,5-dicarboxylate
FDCA −), which is subsequently protonated to produce FDCA. Previously, high yields were achieved on
CO . In this work, we investigate the carboxylation
2
(
small-scale reactions using caesium furoate and Cs
2
3
reaction using alkali furoate/M
2
CO
3
salts containing cation blends and describe reaction conditions that
+
provide high yields on a preparative scale. We show that the carboxylation proceeds efficiently with K /
+
+
+
+
Received 8th April 2017,
Accepted 27th April 2017
Cs blends that have a high K content (up to 4 : 1 K : Cs ). Removing H O, which is a by-product of the
2
reaction, is important for suppressing decomposition pathways. The accumulation of the FDCA2− product
DOI: 10.1039/c7gc01059a
inhibits the reaction. Integrating these lessons, we demonstrate the carboxylation of furoate on a 1 mol
rsc.li/greenchem
scale using a fixed-bed flow reactor with 89% isolated yield of pure FDCA upon protonation.
synthetic routes to FDCA and its derivative dimethyl-
furandicarboxylate (DMFD) have been advanced to pilot-scale
Introduction
Polyethylene terephthalate (PET) and polyethylene furandicar- production and plans have been announced for the first com-
1
2
boxylate (PEF) were both originally patented nearly 70 years mercial scale plant. All of these routes use a C6 sugar
1
,2
7
ago. PET has since grown to become a >10 ton per year (hexose) as the feedstock and follow a similar synthetic strat-
polymer, with applications in rigid packaging, fibres, and egy. The hexose is first dehydrated to form hydroxymethyl fur-
films. This growth was made possible by the development of fural (HMF), which is then oxidized to FDCA (Scheme 1).
scalable syntheses of purified terephthalic acid (PTA), the DMFD is made either by esterifying FDCA or etherifying HMF
diacid monomer, which is made by oxidation of petroleum- to form methoxymethyl furfural (MMF) and then oxidizing it
3
13
derived para-xylenes. PEF is an attractive replacement for PET in methanol. This basic approach—hexose dehydration fol-
because its diacid monomer, furan-2,5-dicarboxylic acid lowed by oxidation—has several drawbacks. In their current
4
(
FDCA), is derived from biomass, which provides an opportu- form, these syntheses require edible fructose as the hexose
5
nity to produce a polymer with a much lower greenhouse gas feedstock. Edible fructose has a large GHG footprint and its
GHG) footprint. In addition, PEF has superior gas barrier use for commodity-scale plastic production would compete
5
(
and mechanical properties compared to PET for certain packa- with food production. Using hexoses derived from ligno-
6
–8
ging applications such as bottled beverages.
cellulosic biomass would avoid these issues, but thus far it has
PEF and other FDCA-based polymers have attracted proven very challenging to obtain FDCA economically from
9
–11
14
increased interest in the past several years.
A number of lignocellulose-derived hexoses. Another major drawback is
the difficulty of oxidizing HMF to FDCA (or MMF to DMFD).
This reaction is typically performed under highly corrosive
Department of Chemistry, Stanford University, 337 Campus Drive, Stanford,
California 94305, USA. E-mail: mkanan@stanford.edu
15–17
conditions in acetic acid solvent,
similar to the conditions
3
used to oxidize para-xylenes to terephthalic acid. Moreover,
†
Electronic supplementary information (ESI) available: Experimental pro-
cedures, additional data, NMR spectra. See DOI: 10.1039/c7gc01059a
even under optimized conditions, the reaction also forms alde-
This journal is © The Royal Society of Chemistry 2017
Green Chem.

View Article Online
Paper
Green Chemistry
Scheme 1 (Top) Route for producing FDCA from edible biomass. A common impurity of the oxidation of hydroxymethylfurfural (HMF) is the incom-
plete oxidation product, 5-formyl-2-furoic acid. (Bottom) Diagram depicting the overall scheme for producing FDCA from lignocellulose through
2−
furfural using CO
3
-promoted C–H carboxylation. Recovery of the alkali cations is achieved by protonation of the FDCA salt and bipolar membrane
electrodialysis of the resulting alkali chloride salt.
2
3
hyde impurities that cause chain terminations during poly- duction. Second, oxidizing furfural to 2-furoic acid is easier
ester synthesis and discolour the polymer. Removing these than oxidizing HMF to FDCA. Although 2-furoic acid is not
impurities requires a separate chemical step—either hydrogen- currently synthesized on a commodity scale, multiple reports
1
8,19
ation or a second oxidation.
While much progress has describe high-yielding oxidations of furfural or furfuryl alcohol
under relatively mild conditions (e.g.
Lastly, CO32−-promoted C–H carboxylation
requires no solvent or transition metal catalysts and installs
We recently described a C–H carboxylation reaction that the final carbon in the appropriate oxidation state.
makes it possible to synthesize FDCA from 2-furoic acid and Utilizing this chemistry to supply FDCA for polymer pro-
CO in high yield without energy-intensive reagents or tran- duction requires finding carboxylation conditions that are
been made in advancing FDCA and DMFD synthesis toward to 2-furoic acid using O
commercial production, alternative routes that avoid these 50 °C, 1 bar).
drawbacks are highly desirable.
2
2
5–28
2
2
0
sition metal catalysts. This reaction is performed by heating amenable to very large-scale reactions with minimal by-pro-
a salt mixture composed of alkali furoate (the deprotonated ducts and facile purification. In our previous work, carboxyla-
2
−
form of 2-furoic acid) and alkali CO
3
2
under a CO atmo- tion reactions were performed by heating mixtures of caesium
2
0
sphere. With the appropriate salt composition (see below), furoate and Cs
CO3 promotes the C–H carboxylation to form furan-2,5-dicar- up to 89% for FDCA were obtained on a 1–10 mmol scale,
boxylate (FDCA ). Protonation with strong acid (e.g. HCl) in but a diminished yield of 71% was obtained upon scaling to
water yields precipitated FDCA product and an aqueous salt 100 mmol. Isotope exchange experiments suggested that the
solution (e.g. MCl) (Scheme 1). The stoichiometric base and reaction proceeds by deprotonation of the C–H bond at the 5
2
CO
3 2
under static CO pressure. High yields of
2
−
2−
2
−
2
−
acid consumed in this synthesis can be regenerated by subject- position of 2-furoate by CO
ing the salt solution to bipolar membrane electrodialysis.
3
to form a carbanion that under-
2
1,22
2−
goes a C–C bond-forming reaction with CO
2
. CO
3
is ordina-
2
9
This carboxylation route to FDCA has a number of potential rily incapable of deprotonating this C–H bond (pK 35), but
a
advantages over conventional hexose-based routes. First, in a solvent-free caesium furoate/Cs
2
2
CO
-furoic acid is made from furfural, a compound that has been high concentration of Cs cations that can stabilize the anionic
produced from lignocellulosic biomass industrially for almost species. At the temperatures required for carboxylation, the
3
mixture there is a
+
2
3,24
a century.
Furfural is made by depolymerizing and dehy- mixture is semi-molten, providing sufficient ion mobility for
drating the pentose-rich hemicellulose component of ligno- the reaction to occur. We attributed the diminished yield upon
cellulose, which is much easier to process than the hexose-rich scaling to 100 mmol to a reduced surface area-to-volume ratio,
cellulose component. Multiple biomass waste streams have which impedes transport of CO
been used for furfural production, including corncobs, O. In the absence of other additives, alkali furoate/alkali
bagasse, and the hemicellulose waste of wood pulp pro- CO3 mixtures with a single cation other than Cs do not
2
into the salt and removal of
H
2
2
−
+
Green Chem.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Green Chemistry
Paper
undergo C–H carboxylation, suggesting that these salts remain
in an unreactive crystalline form at elevated temperatures.
However, we demonstrated that furoate carboxylation could be
+
achieved in small-scale (1 mmol) reactions with all K salts by
combining potassium furoate/K CO with a potassium carbox-
2
3
ylate co-salt (e.g. potassium isobutyrate) to increase ion mobi-
2
0
lity at elevated temperatures.
In this work, we investigate furoate carboxylations with
salts containing cation mixtures and develop conditions for
mole-scale carboxylation in high yield using fixed-bed flow
reactors. We show that carboxylation proceeds efficiently with
+
+
2 3
alkali furoate/M CO salts if K /Cs cation mixtures are used,
+
+
even with high K : Cs ratios. This advance enables carboxyla-
+
tion with a reduced Cs loading without requiring a co-salt.
Although the cations would be recycled in a practical appli-
+
cation, reducing the Cs loading is desirable because the
+
30
amount of Cs salts mined annually is relatively low. More
+
+
2 3
importantly, alkali furoate/M CO salts with K /Cs mixtures
do not undergo a bulk phase transition to a molten phase,
which facilitates scale-up. Using a fixed-bed reactor under
+
+
2
flowing CO , we demonstrate furoate carboxylation with K /Cs
salts on a 1 mol scale with 89% isolated yield for FDCA upon
acidic work up.
Fig. 1 The C–H carboxylation of the 5 position of furoate promoted by
2
−
+
+
CO
“
3
at various temperatures and ratios of K : Cs . Reactions denoted
gas streams that were pre-saturated with water
+ water” had CO
2
Results
Small-scale reactions
before introduction to the reactions. All yields were determined by com-
parison to an internal standard by NMR. Reactions were performed in
triplicate and the error bars represent the standard deviation.
2
−
The synthesis of FDCA via C–H carboxylation is performed
by heating furoate and CO
2
−
3
2
under CO . The yield and reac-
tion rate depend on the precise composition and morphology
of the salt mixture, the temperature, and CO mass transport. products. The decomposition products consist mostly of in-
2
+
To assess whether pure Cs salts could be substituted with soluble char and smaller amounts of acetate and malonate.
cation blends, we initially surveyed the reaction on an analyti-
The cation composition affects the morphology of the salt
cal scale (0.5 mmol substrate) using different K : Cs ratios. To under the reaction conditions. In the all Cs reaction, the salt
+
+
+
prepare the solid reactants, 2-furoic acid was added to an mixture is partially molten at the temperatures where good
+
+
aqueous K
2
CO
3 2 3
/Cs CO
solution (1.25 total molar equivalents yields are observed. With a K /Cs blend, however, the salts do
2
−
of CO3 ) in a glass vial and dried to a solid, which is com- not appear to undergo a bulk phase transition. The ability to
2
−
2−
posed of 2-furoate and 0.75 equivalents of residual CO
3
. This synthesize FDCA without forming a bulk molten solution
material was then heated in a vial block with a constant purge provides flexibility in reactor design and facilitates scale-up
of CO through the headspace for 17 h. The product mixture (see below).
2
was analysed directly by NMR using an internal standard. The
To gain insight into the origin of the decomposition pro-
yields are far from optimal in this vial block reactor (see ducts, we first assessed the effect of the cation composition on
below), but the setup permits rapid initial assessment of the the thermal stability of the substrate and product in the
2
−
effects of varying the salt composition.
absence of CO
2 3
. Heating furoate and 0.75 equivalents of CO
2
−
Fig. 1 shows the FDCA yield as a function of temperature under N for 6 h at 265 °C to 295 °C caused decomposition to
for several K : Cs ratios. As a reference point for these vial similar products seen in the carboxylation reaction (Fig. 2).
block reactions, the yield with only Cs (caesium furoate + The amount of decomposition increased with temperature and
) reached a maximum of 71% at 260 °C. Reactions with with the Cs content of the salt mixture. Thus, the higher the
loading, the greater the intrinsic thermal stability of
do not form any FDCA regardless of the temperature (data furoate. Regardless of the cation composition, the amount of
2
+
+
+
+
Cs
2
CO
3
+
+
only K in the absence of co-salts (potassium furoate + K CO )
K
2
3
2
−
+
+
not shown). However, K /Cs blends containing a high pro- decomposition observed under N
portion of K (1 : 1 to 4 : 1 K : Cs ) gave 60–70% yield of greater than the amount of FDCA produced under CO
2
was comparable to or
+
+
+
2−
at the
2
2
−
+
+
+
+
FDCA . The optimal temperature for the K /Cs reactions same temperature. For example, at 285 °C, the 2 : 1 K : Cs
+
2−
increased with the K content. The mass balance of the reac- blend produced 63% FDCA under CO
2
after 17 h, whereas
tions was composed of unreacted furoate and decomposition >90% of the furoate decomposed after only 6 h under N . This
2
This journal is © The Royal Society of Chemistry 2017
Green Chem.

View Article Online
Paper
Green Chemistry
+
2
balance for the all Cs reaction indicated that H O mostly
2
−
slowed down FDCA synthesis without substantially accelerat-
ing decomposition (i.e. both yield and furoate conversion were
2
reduced). In contrast, H O greatly increased the decomposition
+
+
products for the 2 : 1 K : Cs reaction. This result suggests that
removing the H O by-product is important for suppressing
2
decomposition during carboxylation when cation blends are
used, which is further supported by experiments in the fixed-
bed reactor (see below).
We next probed the kinetics of carboxylation in the vial
block reactor by performing a time course experiment. A set of
+
+
identical reaction mixtures with 2 : 1 K : Cs was heated under
CO at 285 °C and individual reactions were stopped at
2
2
−
different reaction times for analysis (Fig. 3). The FDCA yield
rose steadily to 50% in the first 5 h, but thereafter the reaction
slowed dramatically, requiring an additional 12 h to reach 63%
yield. Quantification of the substrate consumption indicated
that 14% decomposed in the first 5 h and an additional 10%
decomposed over the next 12 h. The greatly attenuated sub-
strate decomposition relative to experiments under N (Fig. 2)
2
2
confirms that CO inhibits the decomposition pathways.
Additional time course experiments were performed with
2
−
2−
different amounts of CO3 . Increasing the CO3
loading
beyond 0.75 equivalents increased the initial reaction rate, but
did not eliminate the sharp drop in the reaction rate after
Fig. 2 The thermal decomposition of the furoate (dashed line) and
2
−
2−
2 3
FDCA (solid line) salts under N atmospheres in the presence of CO
+
+
at various temperatures and ratios of K : Cs . The percent decompo-
sition was determined by comparison to an internal standard by NMR.
Reactions were performed in triplicate and the error bars represent the
standard deviation.
2
result suggests that CO greatly curtails the rate of furoate
decomposition, in addition to making the carboxylation reac-
2
−
tion possible. In contrast to furoate, FDCA showed good
thermal stability under N with all cation compositions up to
85 °C (Fig. 2). Some decomposition was observed at 295 °C,
2
2
with a similar cation dependence as seen for the substrate.
The decomposition products observed during carboxylation
therefore arise from furoate decomposition rather than the
2
−
FDCA product.
Water is a by-product of CO
3
2−-promoted C–H carboxyla-
tion. We hypothesized that water could be deleterious to
furoate carboxylation by providing a source of hydroxide ions
that cause decomposition via nucleophilic attack on furoate.
+
+
+
To test this hypothesis, the all Cs and the 2 : 1 K : Cs carbox-
ylations were repeated at their optimal temperatures using a
CO
2
stream passed through a water bubbler to hydrate the gas
Fig. 3 The kinetics of the C–H carboxylation of the 5 position of
before flowing into the glass vials (Fig. 1, “+ water” con-
ditions). Under these conditions, the FDCA
reduced from 71% to 41% in the all Cs reaction and from
2
−
furoate promoted by CO
3
with or without the addition of 0.5 equiva-
2
−
yield was
2−
lents of FDCA . All yields were determined by comparison to an internal
standard by NMR. Reactions were performed in triplicate and the error
+
+
+
7
0% to 31% in the 2 : 1 K : Cs reaction. Analysis of the mass bars represent the standard deviation.
Green Chem.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Green Chemistry
Paper
50% FDCA − formation (Fig. S8†). To see if FDCA2− itself seen under flowing conditions can be attributed to water
2
∼
slows the furoate carboxylation, a time course was performed removal. Additionally, flow forces CO2 through the solid
2
−
using a salt mixture spiked with 0.5 equivalents of FDCA
.
material, which may expose salt particles to CO
2
more efficien-
2
−
Under these conditions, the synthesis rate and yield of FDCA
were reduced, and furoate decomposition increased (Fig. 3). ditions gave a 56% yield of FDCA for a reaction with a
tly than under static conditions. Remarkably, flowing con-
2
−
2
−
+
+
This experiment shows that the build-up of FDCA slows 20 : 1 K : Cs ratio (Table 1), providing proof-of-concept for the
down the carboxylation of remaining furoate, either by imped- ability to perform carboxylation with cation blends that have
+
ing CO transport or slowing the rate of deprotonation.
very low Cs content.
2
Preparative-scale reactions
Fixed-bed flow reactor
Fixed-bed flow conditions were used to evaluate the carboxyla-
tion on preparative scales. Experiments were performed with
The vial block reactor provides relatively poor gas transport
because CO can only be flowed through the headspace above
2
+
+
2
: 1 K : Cs blends using the reactor design described above
the solid reactants. To determine whether increased contact
between the salt mixture and CO2 improves the yield, we
assembled a fixed-bed flow reactor. In brief, the reactor con-
sists of a temperature-controlled stainless steel tube (the reac-
tion vessel) connected to a CO2 pre-heating apparatus (see
with larger reaction vessels to accommodate the increased
reactant volumes (see ESI†). At the conclusion of the reactions,
the solid products were dissolved in H O, filtered to remove in-
2
soluble by-products, and acidified with HCl to precipitate
FDCA. Isolated yields are reported for dry, purified FDCA
ESI† for details). The reactor accommodates 8 bar CO
2
at elev-
(
Table 2). The reaction was first attempted on a 100 mmol
ated temperature and flow rates up to several hundred ml
−
1
−
1
2
scale at 285 °C with 140 ml min CO at 8 bar. After a 24 h
min . For fixed-bed flow experiments, we used industrial
grade CO (99.5%). We first evaluated the effects of static vs.
flowing CO on a 10 mmol scale for different cation compo-
2
2
Table 2 Preparative-scale CO
3
2−-promoted C–H carboxylation of
furoate
sitions (Table 1). The solid reactants were prepared as
described above and transferred to the reaction vessel as dry
powders. The reactions were performed at 265 °C to 300 °C
under 8 bar CO for 14 h with either no flow or a flow rate of
2
1
+
2−
1
40 ml min− . With all Cs , the FDCA yield was 91% in the
absence of flow at 265 °C. This result is consistent with the
high carboxylation yield observed on the 10 mmol scale with
+
20
all Cs in a Parr reactor.
4
For reactants with 1 : 1 to
Ramp
rate
Soak
time
Starting
material
+
+
2−
a
a
b
: 1 K : Cs , however, the FDCA yield was only 22%–56%
Scale
T
FDCA
under static conditions and large amounts of decomposition
products were formed. Upon switching from static to flowing
1
1
00 mmol 285 °C
00 mmol 260–285 °C 2 °C h
260–285 °C 2 °C h
N/A
24 h
24 h
24 h
66%
86%
89%
0%
3%
2%
−1
−1
2
−
CO
2
, the FDCA yields increased to 71%–79% and decompo- 1 mol
sition products were suppressed. Under flowing conditions,
water formed during the reaction is stripped away. Since water
a
The temperature was ramped from 260 °C to 285 °C at the specified
b
ramp rate and then held (soaked) at 285 °C for 24 h. Isolated yield of
is deleterious to the carboxylation reaction, the boost in yield the diacid.
Table 1 Effect of flowing CO
2
on CO
3
2−-promoted C–H carboxylation of furoate
a
Static
Flowing
FDCA2−
+
+
FDCA2−
K : Cs
Scale
T
Starting material
Starting material
+
Cs only
1 mmol
265 °C
265 °C
275 °C
290 °C
290 °C
290 °C
300 °C
91%
91%
56%
22%
39%
33%
10%
4%
0%
12%
0%
0%
0%
—
—
—
—
+
Cs only
10 mmol
10 mmol
10 mmol
10 mmol
10 mmol
10 mmol
1
2
3
4
2
: 1
: 1
: 1
: 1
0 : 1
79%
78%
74%
71%
56%
10%
0%
15%
18%
26%
69%
a
2
All yields were determined by comparison to an internal standard by NMR. The flow rate of CO .
was 140 ml min−1
This journal is © The Royal Society of Chemistry 2017
Green Chem.

View Article Online
Paper
Green Chemistry
reaction, FDCA was isolated in 66% yield, which is consider- tions, the salt mixture slowly compacts as the reaction pro-
2
−
ably lower than the 78% FDCA
yield observed in the ceeds, which greatly slows the reaction and promotes
1
0 mmol scale reaction (Table 1). Large amounts of decompo- decomposition. These problems are intrinsic to the use of a
sition products were formed. These results reflect the chal- fixed bed reactor. Strategies that improve the mixing between
lenge of maintaining good solid–gas contact as the scale is CO2 and the solid reactants, such as fluidization or other
3
2
increased in the fixed-bed reactor. In particular, the solid reac- modes of agitation, are expected to greatly improve the rates
tants start as a fine powder, but compact into a large mass and diminish decomposition products upon further scale up.
over the course of the reaction. This process reduces solid–gas Such approaches have been used for the carboxylation of solid
33
contact, which may impede the removal of water and slow car- sodium phenolate in the industrial synthesis of aspirin.
boxylation, thereby increasing decomposition. The use of CO for the synthesis of high-volume chemicals
To reduce decomposition and improve the yield, we investi- and fuels has been a longstanding goal in chemistry.
gated the effects of using a temperature gradient. A 100 mmol Carbonate-promoted C–H carboxylation is distinguished from
2
3
4,35
2 2
scale reaction was performed with the same CO flow rate other chemical approaches to CO utilization because it forms
using a 2 °C h− ramp from 265 °C to 285 °C, followed by 24 h C–C bonds without resorting to expensive, energy intensive
at 285 °C. This protocol resulted in nearly complete conversion reagents. While FDCA has yet to be proven as a viable com-
of furoate and an 86% isolated yield of FDCA. Gratifyingly, modity, it has thus far shown great potential to replace ter-
1
7
when the same temperature ramp was used on a 1 mol scale, ephthalic acid in high volume (>10 ton per year) applications.
FDCA was isolated in 89% yield (139 g). The gradient may The carboxylation route described here provides a unique
improve the reaction by slowing or subtly altering the compac- opportunity to use CO and waste biomass for a high-volume
2
tion of the solids, thereby enabling better CO access to the chemical synthesis.
2
reactants in the centre of the reactor and more effective water
removal.
Conclusions
We have shown that CO32−-promoted C–H carboxylation of
furoate using K /Cs mixtures is an efficient method for
Discussion
+
+
synthesizing FDCA on a preparative scale. Salt mixtures with
high K : Cs ratios retain their bulk solid morphology under
the reaction conditions, which facilitates reactions under
Our results demonstrate that furoate carboxylation can be
used to synthesize FDCA in high yield on a preparative scale
+
+
(>100 g) and provide insights for further scale-up. The simpli-
flowing CO . Carboxylation competes with furoate decompo-
2
city of the carboxylation reaction is potentially advantageous
for process development. No solvent or catalyst is used, which
obviates the burden of recovering and/or replenishing these
components. Furthermore, the carboxylation performs well
2
sition. Removal of the H O by-product suppresses decompo-
sition and improves the yield of FDCA. Using a simple fixed
bed reactor, we have demonstrated mole-scale furoate carboxy-
+
+
lation using a 2 : 1 K : Cs blend with 89% isolated yield of
FDCA. Further improvement of mass transport is expected to
enable similar performance on a larger scale.
with relatively low-purity CO
energy-intensive purification steps required for high-purity
gases. The CO could easily be re-circulated in a large-scale
process to maximize CO conversion. One possible concern for
2
(99.5%), which avoids costly and
2
2
scale-up is the use of caesium. In the absence of a co-salt,
+
Acknowledgements
some Cs is required for efficient carboxylation, and the
+
+
results here indicate that a K : Cs ratio of 2 : 1 to 4 : 1 is desir-
We thank Stanford University and the Global Climate and
Energy Project. The Camille and Henry Dreyfus Foundation
also supported this work through a Teacher-Scholar award to
MWK. GRD gratefully acknowledges the Center for Molecular
Analysis and Design for a graduate fellowship.
3
1
able. Although caesium is more abundant than tin, caesium
salts have historically been mined on a small scale because of
3
0
their low demand. It is important to emphasize that the use
of carboxylation for large-scale FDCA synthesis would not
consume caesium salts because the cations would be recycled
after protonation by using bipolar membrane electrodialysis
+
(
Fig. 1). Because of its very high water solubility, Cs can be
Notes and references
2
0
quantitatively recovered upon FDCA precipitation. Assuming
+
+
+
Cs has similar transport properties to Na and K in cation
exchange membranes, bipolar membrane electrodialysis
would require a modest energy demand of a few kWh per kg
FDCA.
1 W. J. Rex and D. J. Tennant, US Pat 2465319, 1949.
2 J. Lincoln and N. D. J. Gordon, US Pat 2551731, 1951.
3 R. J. Sheehan, in Ullmann’s Encyclopedia of Industrial
Chemistry, Wiley-VCH Verlag GmbH & Co. KgaA, Weinheim,
7th edn, 2012, vol. 36, pp. 17–28.
4 T. Werpy, G. Petersen, A. Aden, J. Bozell, J. Holladay,
J. White, A. Manheim, D. Eliot, L. Lasure and S. Jones, Top
value added chemicals from biomass. Volume 1-Results of
The reaction kinetics must be substantially improved in
order for carboxylation to be used on much larger scales. Our
results suggest that this challenge can be addressed by improv-
ing mass transport. For the 100 mmol and 1 mol scale reac-
Green Chem.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Green Chemistry
Paper
screening for potential candidates from sugars and synthesis 21 K. Mani, J. Membr. Sci., 1991, 58, 117–138.
gas, DTIC Document, 2004.
A. J. J. E. Eerhart, A. P. C. Faaij and M. K. Patel, Energy
Environ. Sci., 2012, 5, 6407–6422.
S. K. Burgess, J. E. Leisen, B. E. Kraftschik, C. R. Mubarak,
R. M. Kriegel and W. J. Koros, Macromolecules, 2014, 47,
22 J. R. Davis, Y. Chen, J. C. Baygents and J. Farrell, ACS
Sustainable Chem. Eng., 2015, 3, 2337–2342.
23 H. Hoydonckx, W. Van Rhijn, W. Van Rhijn, D. De Vos and
P. Jacobs, in Ullmann’s Encyclopedia of Industrial Chemistry,
5
6
Wiley-VCH Verlag GmbH
2007.
&
Co. KgaA, Weinheim,
1
383–1391.
7 S. K. Burgess, O. Karvan, J. Johnson, R. M. Kriegel and 24 J. P. Lange, E. van der Heide, J. van Buijtenen and R. Price,
W. J. Koros, Polymer, 2014, 55, 4748–4756. ChemSusChem, 2012, 5, 150–166.
8 S. K. Burgess, R. M. Kriegel and W. J. Koros, 25 E. Taarning, I. S. Nielsen, K. Egeblad, R. Madsen and
Macromolecules, 2015, 48, 2184–2193. C. H. Christensen, ChemSusChem, 2008, 1, 75–78.
9 E. de Jong, M. Dam, L. Sipos and G. Gruter, Biobased mono- 26 J. van Haveren, S. Thiyagarajan and T. Morita, WIPO
mers, polymers, and materials, 2012, vol. 1105, pp. 1–13. International Application WO2013/096998, 2013.
0 A. Pellis, E. H. Acero, L. Gardossi, V. Ferrario and 27 K. Gupta, R. K. Rai, A. D. Dwivedi and S. K. Singh,
G. M. Guebitz, Polym. Int., 2016, 65, 861–871. ChemCatChem, 2016, DOI: 10.1002/cctc.201600942.
1 A. Sousa, A. Fonseca, A. Serra, C. Freire, A. Silvestre and 28 S. S. Stahl, A. B. Powell, T. W. Root, D. S. Mannel
1
1
1
J. Coelho, Polym. Chem., 2016, 7, 1049–1058.
2 A. H. Tullo, Chemical & Engineering News, 2016, vol. 94,
issue 41, p. 13.
and M. S. Ahmed, US Pat. Application 2015/0134272,
2015.
29 R. R. Fraser, T. S. Mansour and S. Savard, Can. J. Chem.,
1985, 63, 3505–3509.
1
1
3 G. J. M. Gruter and F. Dautzenberg, US Pat 8133289, 2012.
4 J. Zhang, J. Li, Y. Tang, L. Lin and M. Long, Carbohydr. 30 W. C. Butterman, W. E. Brooks and R. G. Reese Jr., Mineral
Polym., 2015, 130, 420–428. commodity profiles: cesium, Report 2331-1258, 2005.
5 C. M. de Diego, W. P. Schammel, M. A. Dam and 31 K. H. Wedepohl, Geochim. Cosmochim. Acta, 1995, 59, 1217–
G. J. M. Gruter, US Pat 8519167, 2013. 1232.
6 C. M. de Diego, M. A. Dam and G. J. M. Gruter, US Pat 32 J. Werther, Fluidized-Bed Reactors, Wiley-VCH Verlag GmbH
865921, 2014. & Co. KgaA, Weinheim, 2007.
7 M. E. Janka, D. Lange, M. C. Morrow, B. R. Bowers, 33 G. Stopp, H. Karkossa and V. Trescher, US Pat 4508920,
1
1
1
8
K. R. Parker, A. S. Shaikh, L. R. Partin and C. Sumner, US
Pat. Application 2015/0051412, 2015.
8 M. E. Janka, K. R. Parker, P. Moody, A. S. Shaikh,
L. R. Partin, B. R. Bowers and M. C. Morrow, US Pat
1985.
34 A. M. Appel, J. E. Bercaw, A. B. Bocarsly, H. Dobbek,
D. L. DuBois, M. Dupuis, J. G. Ferry, E. Fujita, R. Hille,
P. J. A. Kenis, C. A. Kerfeld, R. H. Morris, C. H. F. Peden,
A. R. Portis, S. W. Ragsdale, T. B. Rauchfuss, J. N. H. Reek,
L. C. Seefeldt, R. K. Thauer and G. L. Waldrop, Chem. Rev.,
2013, 113, 6621–6658.
1
8
809556, 2014.
1
2
9 M. E. Janka, C. E. Sumner Jr., S. K. Kirk, A. S. Shaikh and
K. R. Parker, US Pat 8969404, 2015.
0 A. Banerjee, G. R. Dick, T. Yoshino and M. W. Kanan, 35 M. Aresta, A. Dibenedetto and A. Angelini, Chem. Rev.,
Nature, 2016, 531, 215–219. 2014, 114, 1709–1742.
This journal is © The Royal Society of Chemistry 2017
Green Chem.