One-pot catalytic conversion of cellulose and of woody biomass solids to liquid fuels

Article abstract of DOI:10.1021/ja205436c

Efficient methodologies for converting biomass solids to liquid fuels have the potential to reduce dependence on imported petroleum while easing the atmospheric carbon dioxide burden. Here, we report quantitative catalytic conversions of wood and cellulosic solids to liquid and gaseous products in a single stage reactor operating at 300-320 °C and 160-220 bar. Little or no char is formed during this process. The reaction medium is supercritical methanol (sc-MeOH) and the catalyst, a copper-doped porous metal oxide, is composed of earth-abundant materials. The major liquid product is a mixture of C₂-C₆ aliphatic alcohols and methylated derivatives thereof that are, in principle, suitable for applications as liquid fuels.

Full text of DOI:10.1021/ja205436c

ARTICLE
pubs.acs.org/JACS
One-Pot Catalytic Conversion of Cellulose and of Woody Biomass
Solids to Liquid Fuels
†
†,‡
†,§
,†
Theodore D. Matson, Katalin Barta, Alexei V. Iretskii, and Peter C. Ford*
†
Department of Chemistry and Biochemistry, University of California, Santa Barbara, Santa Barbara, California 93110-9510,
United States
‡
Center for Green Chemistry and Engineering, Yale University, New Haven, Connecticut 06520, United States
§
Department of Chemistry and Environmental Sciences, Lake Superior State University, Sault Sainte Marie, Michigan 49783,
United States
S Supporting Information
b
ABSTRACT: Efficient methodologies for converting biomass
solids to liquid fuels have the potential to reduce dependence on
imported petroleum while easing the atmospheric carbon
dioxide burden. Here, we report quantitative catalytic conver-
sions of wood and cellulosic solids to liquid and gaseous
products in a single stage reactor operating at 300ꢀ320 °C
and 160ꢀ220 bar. Little or no char is formed during this
process. The reaction medium is supercritical methanol
(
sc-MeOH) and the catalyst, a copper-doped porous metal oxide, is composed of earth-abundant materials. The major liquid
product is a mixture of C ꢀC aliphatic alcohols and methylated derivatives thereof that are, in principle, suitable for applications as
2
6
liquid fuels.
’
INTRODUCTION
should be compatible with operation scales that match the
transportation limitations of biomass feedstocks. Furthermore,
the catalyst is based on earth-abundant elements.
With the goal of reducing anthropogenic CO , we have
2
developed a clean, single-stage, high-yield conversion of biomass
solids to combustible liquids. Woody biomass, primarily a matrix
of the biopolymers lignin, cellulose, and hemicellulose, is a
renewable feedstock for sustainable energy production that does
not compromise the world food supply. Efficient methodologies
for its utilization in the production of liquid fuels or chemicals can
reduce societal dependence on fossil-carbon based energy. Two
types of approaches have historically received consideration for
the transformation of wood to liquid fuels: (a) acid pretreatment
and separation followed by fermentation or liquid phase proces-
sing, and (b) high temperature conversion such as gasification or
pyrolysis to bio-oils. Each has specific problems relevant to
efficient conversion of lignocellulose carbon to value-added
products. For (a), the lignin fraction, comprising up to 30% of
the carbon content, is often simply burned, although there has
been recent effort into feedstock pretreatment
utilization of aqueous sugars and lignin byproducts. For (b),
a significant fraction of the feedstock is transformed into
Simple chemical conversion of lignocellulosic biomass is
challenged by the complexity of this material; indeed, many of
the studies of new biomass conversion methodologies have
focused on simpler molecular models, especially for lignin.
Effective utilization requires depolymerizing cellulose and lignin,
the latter being nearly refractory toward hydrolysis of its phenolic
ether linkages, and substantially reducing overall oxygen content
without producing chars. We have previously addressed the issue
of lignin depolymerization by demonstrating that the simple
12
model dihydrobenzofuran and the more complex organosolv
13
1
lignin undergo catalytic hydrogenolysis of phenolic ether
linkages in supercritical methanol in the presence of a copper-
doped porous metal oxide (Cu20-PMO).
2
The Cu20-PMO also catalyzes methanol reforming (eq 1)
3ꢀ5
and better
CH3OH h 2H2 þ CO
and the water gas shift (eq 2).
CO þ H2O h H2 þ CO2
ð1Þ
6,7
8,9
1
0
intractable carbonaceous waste (i.e., “char”), and refinery-scale
infrastructure is needed for processing. Moreover, the geogra-
phically dispersed nature of biomass feedstocks both makes
forest-to-refinery transport a key economic consideration and
therefore limits the size of processing units. We offer a unique
alternative, the methanol-mediated, catalytic conversion of
woody biomass and of cellulose solids to liquids and gases
ð2Þ
The resulting H provides the reducing equivalents necessary
2
both to cleave the aromatic ethers and for further reduction and
deoxygenation. We now show that this single-stage process also
11
Received: June 12, 2011
Published: August 01, 2011
(Scheme 1). This process is quantitative (little or no char) and
r 2011 American Chemical Society
14090
dx.doi.org/10.1021/ja205436c
|J. Am. Chem. Soc. 2011, 133, 14090–14097

Journal of the American Chemical Society
ARTICLE
Scheme 1. Conversion of Lignocellulose Solids to Liquid Fuels As Mediated by sc-MeOH and Copper-Doped Hydrotalcite Based
Porous Metal Oxides
depolymerizes and reduces cellulose and converts much more
complex lignocellulose composites, various finely divided wood
particles, to a mixture of alcohols potentially usable as liquid fuels.
Liquids Analysis. At UCSB, liquid phase samples were obtained
with a microliter syringe and injected into a programmable Agilent
model 6890 GC with a 30 m DB-5 column equipped with a 10 m guard
column and a flame ionization detector (FID). The injector was
maintained at 290 °C. The carrier gas was helium at a flow of
’
EXPERIMENTAL SECTION
ꢀ1
2
0 mL min . The column was maintained at 30 °C for 5 min, raised
Materials. Sawdust samples were obtained by sanding oven-dried
ꢀ1
ꢀ1
at 10 °C min to 90 °C, then raised at 25 °C min to 250 °C and
maintained for 20 min. Identification of the alcohols was achieved by
retention time comparison to standards. In some cases, samples were
mixed with authentic standards and analyzed with two different tem-
perature programs to obtain positive identification. Mass fraction was
determined by integration.
woods followed by sifting through a screen of a predetermined mesh or
were used as provided from commercial sources. Coniferous wood
ponderosa pine flour (140 mesh maximum) was purchased from
American Wood Fibers. Powdered torrefied wood was used as provided
by Energy Institute of The Netherlands. Cellulose fibers were purchased
from Aldrich (#C6288). The Cu20-PMO catalyst typically used has a
At Yale, the GCꢀFID analyses used a Shimadzu GC equipped with
an FID detector and a SHRXI-5MS column of 30 m ꢁ 0.25 mm i.d. The
carrier gas was helium at a total flow of 19.6 mL/min. Split injection
mode was used with a split ratio of 15.2. The oven program maintained T
at 40 °C for 10 min, heated at 10 °C/min to 90 °C where it paused for
5 min, then heated at 15 °C/min to 260 °C where it paused for 5 min,
then finally heated at 25 °C/min to 290 °C.
1
5/60/25 Cu/Mg/Al molar ratio and was prepared by calcining a
12,13
copper-doped hydrotalcite as previously reported.
Small-Scale Batch Reactions. In a typical run, a finely divided
substrate (wood or cellulose), catalyst (usually Cu20-PMO), and
methanol (3.0 mL) were added to a 10 mL stainless steel, Swagelok
13
union minireactor. The reactor was sealed, shaken, and placed in a
preheated aluminum block at the desired temperature (T) and allowed
to react at the specified T for a prescribed period. For analysis, the
reaction was quenched by rapid cooling, after which the reactor was
weighed to ensure that no leakage had occurred. (Any sample deviating
Solids Analysis. Solids were washed extensively with methanol and
water, and then slurried in ∼20-fold excess by weight H
2
O. Concen-
trated nitric acid was added dropwise with stirring and a flow of air was
maintained to remove evolved gases. At a nominal concentration
>
10 mg from the prereaction weight was discarded.) The gaseous
3
between 10 and 12 M HNO , complete oxidation and dissolution of
products were collected and the volume measured by displacement of
water in a specially designed volumetric displacement apparatus
equipped with sampling port. (WARNING: These minireactors develop
high internal pressures during a catalysis run and must be handled and
opened with care.) A sample of 500 μL of gas was analyzed by GCꢀTCD
according to the methods described below. The liquid products were
separated by filtration or by centrifugation from the remaining solids, which
were dried and weighed after washing with methanol and ether. Approxi-
mately 1.0 μL sample of this liquid was analyzed by GCꢀFID as described
below. In several cases, the solids were further treated with ∼10 mL of
MeOH and the filtrate was analyzed by GCꢀFID to verify that the
concentrate obtained initially was representative of the bulk liquid.
Gas Analysis. Gas phase samples were obtained with a gas tight
syringe, then injected into a programmable Agilent model 6890 GC with
a 10 m Carbosieve packed column and a thermal conductivity detection
copper was obtained. The contents of the vial (10 mL for a 100 mg
sample of solid) were transferred to a flask and further treated with
3
40 mL of 5 M HNO . The contents of the flask were then gravity
filtered on a pretared medium porosity frit and washed with 10 ꢁ 10 mL
of H O, 3 ꢁ 10 mL of MeOH, and 3 ꢁ 10 mL of Et O. The frit was then
2
2
dried in vacuo and the extent of conversion determined by mass
difference.
Elemental analyses of dissolved catalysts, ash, and other residues were
accomplished using a Thermo iCAP Model 6300 Inductively Coupled
Plasma Atomic Emission Spectrometer (ICP).
Larger-Scale Batch Reactions. Runs were also carried out in a
Parr series 5500 compact reactor equipped with a stirrer and a 100 mL
base (total volume 113 mL) or a 25 mL base (38 mL). Inorganic catalyst
and organic material were added to the reactor base with a weighed
quantity of methanol. The reactor was then sealed and raised to the
designated temperature over 20ꢀ30 min. Stirring was maintained at
1000 rpm. After the time period designated for the reaction, the reactor
was cooled. Volatile species were collected in a 5 L Tedlar gas bag
equipped with a septum sampling port. The reactor was opened, and
liquid and solid contents were discharged onto a medium porosity frit.
The filtrate was analyzed by GCꢀFID, and the contents of the gas bag
were analyzed by GCꢀTCD and GCꢀFID (VOC’s).
(
TCD). The injector was maintained at 290 °C. The carrier gas was
ꢀ
1
helium at a flow of 7.3 mL min . The column was maintained at 50 °C
ꢀ1
for 5 min, raised at 40 °C min to 250 °C and held for 10 min, then
ꢀ1
finally raised at 10 °C min to 290 °C and held for 45 min. The gases
H , CO, CH , CO , C H , and C H are well resolved with this
2
4
2
2
6
3 8
configuration and detected by a thermal conductivity detector. Quanti-
fication was achieved by comparison to calibration curves determined
for each gas.
1
4091
dx.doi.org/10.1021/ja205436c |J. Am. Chem. Soc. 2011, 133, 14090–14097

Journal of the American Chemical Society
ARTICLE
a
Table 1. Summary of Minireactor Experiments
substrate (amount)
conditions: T (°C), time (h)
conversion (%)
gases (mmol) H
2
, CO, CO
2
, CH
4
HAE (wt %)
CAE (wt %)
maple sawdust (100 mg)
maple sawdust (100 mg)
maple sawdust (100 mg)
pine flour (207 mg)
320, 8
99
99
22, 6.1, 4.7, 0.9
20, 5.8, 6.0, 1.1
not determined
not determined
21, 4.5, 4.8, 0.8
23, 3.7, 6.2, 0.9
27, 4.0, 5.0, 1.5
5.9
5.7
4.6
9.1
5.8
8.6
0.3
2.0
2.0
1.8
4.0
1.4
0.4
0
320, 8 (second run)
300, 24
∼100
96
b
320, 8
torrefied wood (100 mg)
cellulose fibers (100 mg)
no substrate
320, 8
89
c
320, 8
98
320, 8
a
Runs carried out in 10 mL Swagelok union reactors with methanol (2.4 g) and 100 mg of solid catalyst. The gas and liquid compositions were determined
b
by GCꢀTCD and GCꢀFID techniques, respectively, while conversion was based on the amount of solid residues not soluble in aq HNO solution. The
3
c
9
mg residue after dissolution in HNO
3
was visually assessed to be unreacted wood. The 2 mg residue was a white powder presumed to be cellulose.
’
RESULTS AND DISCUSSION
sawdust. Analogous products were observed, and acid digestion
of the remaining solids showed that overall conversion was
Small-Scale Reactions with Sawdust. Typically, substrate
>
98%. Thus, the copper-doped PMO catalyst can be used in
(
100ꢀ600 mg), catalyst (100 mg), and MeOH (2.4 g) were
multiple runs (see below).
Similar observations were made with pine sawdust and
powdered torrefied wood, although only 89% conversion was
sealed in 10 mL minireactor, heated to specified T for a
prescribed period, then quenched by cooling. Gaseous products
were collected and analyzed by GCꢀTCD, and liquid products
were separated from the remaining solids and analyzed by
GCꢀFID. The solids were dried and weighed, then dissolved
found for the latter substrate (Table 1). In each case, a H , CO,
2
CO , CH gas mixture was generated while the liquid fraction
2
4
contained HAE and CAE in a ratio consistent with cellulose/
hemicellulose and lignin origins, respectively.
The reaction time required for solubilization of the biomass
solid was probed in a series of experiments where mixtures of
pine sawdust (100 mg), Cu20-PMO (100 mg), and MeOH
in aq HNO , and the resulting mixture was filtered to measure
3
quantitatively any insoluble organic residues.
We applied this procedure to several wood sawdusts, to
powdered torrefied wood, and to cellulose fibers and have
identified and quantified the major components in the gas and
liquid phases in each case. The liquid products fall largely into
two groups of monomeric alcohols and ethers: HAE (higher
alcohols and ethers), which are C ꢀC species, and CAE
(
3.0 mL) were heated at 320 °C in minireactors for the various
time intervals: 0.17, 0.33, 0.5, 1.0, 2, 4, 6, 8, 10, and 12 h. The
extent of solubilization was determined by weighing the solid
residues after the reaction, and these data (summarized in
Figure 2 and Supporting Information Table S-1) show this to
be ∼75% within 30 min and essentially complete within an hour.
GCꢀFID analysis of the liquid products from an 8 h experiment
with pine sawdust is reported in SI Figure S-1. After 20 min, the
liquid fraction, once separated from the solids, was a bright
orange, which suggested the presence of organosolv lignin.
However, after only 1 h reaction, the color changed to a light
yellow then after 2 h it became completely colorless (Figure 3).
The solids were additionally subjected to the nitric acid test to
establish the extent of conversion. The inorganic fraction of the
solid residues quickly dissolved, while the remaining unreacted
biomass did not, and no char was observed in samples where the
remaining solid weighed 100 ( 5 mg. GCꢀFID results of these
samples are further reported in the Supporting Information.
SI Table S-1 also summarizes control reactions where a pine
sawdust sample (100 mg) was subjected to analogous conditions
for 0.5, 2, and 24 h without catalyst. The behavior was markedly
different from those with added Cu20-PMO. For example, only
about 40% of the substrate was solubilized after 30 min and the
60 mg solid organic residue was visually assessed to be a mixture of
unreacted biomass and char (SI Figure S-2). Furthermore, sig-
nificant amounts of char remained after longer reaction times.
Qualitatively, the liquid phases obtained were strongly colored in
every case, and when analyzed by GCꢀFID, only a small fraction
proved to be volatile. When a sawdust sample was treated similarly
for 24 h with a Mg/Al hydrotalcite derived PMO (100 mg) not
containing copper, the system behaved similarly to the control
reaction without any catalyst (SI Table S-1). Additionally, the
initially white PMO was black after filtration, washing and drying,
showing char formation when Cu was not present.
2
6
(
substituted cyclohexyl alcohols and ethers), which are largely
9
14
C ꢀC components. No organic char residues were observed
12
in these experiments.
In one such experiment, a minireactor was charged with maple
sawdust (100 mg, <63 μm), Cu20-PMO (100 mg), and MeOH
(
2.4 g), purged with argon, sealed, weighed, and heated at 300 °C
for 24 h. Upon opening the reactor, H , CO, CO , CH and other
2
2
4
volatiles were expelled. Analysis by GCꢀFID showed the clear,
colorless liquid product (1900 mg) to contain unreacted metha-
nol and dimethyl ether and a mixture of HAE and CAE
corresponding to 4.6 and 1.8 wt % of this solution (∼87 and
∼
34 mg) in a ratio roughly consistent with the origins of these
components being cellulose and lignin, respectively. We note
that these amounts are somewhat more than expected purely
from lignocelluloses; increases in mass can be attributed to the
incorporation of some methanol-derived carbon into the pro-
ducts (see below).
The solids were collected quantitatively by filtration and
washed with MeOH, diethyl ether, and pentane, and then dried
to give a blackish-red material weighing 96 mg. After treat-
ing this with aq HNO to dissolve the copper and inorganic
oxides, the remainder was a colorless, flocculent solid weighing
3
<
1 mg. Thus, the substrate was converted to MeOH-soluble
or volatile products without formation of intractable organic
waste (char).
Nearly identical results were obtained for an analogous run
conducted at 320 °C for 8 h (Table 1). The liquid products were
a mixture of HAE and CAE in methanol according to GCꢀFID
analysis (Figure 1). The solids recovered were washed with
methanol and directly recycled into a second run with maple
1
4092
dx.doi.org/10.1021/ja205436c |J. Am. Chem. Soc. 2011, 133, 14090–14097

Journal of the American Chemical Society
ARTICLE
Figure 1. GCꢀFID trace of the liquids from the reaction of maple sawdust and Cu20-PMO in a 10 mL minireactor at 320 °C for 8 h, noting some key
chemical entities, identified by comparison to standard samples. (Recorded on a Agilent model 6890 GC with a 30 m DB-5 column.) (DME is dimethyl
ether. Note: The vertical axis is FID signal intensity; the horizontal axis is retention time in minutes.).
Figure 3. Photograph of liquid phases from experiments probing the
temporal solubilization of pine sawdust in the presence of Cu20-PMO
catalyst in experiments detailed in SI Table S-1. The strongly colored
sample at the left is after 10 min reaction at 320 °C in a 10 mL
minireactor.
4
0 Torr, virtually no higher boiling components (<2 mg)
24
remained. In contrast, analogous treatment of the liquid
product from a pine sawdust (100 mg) reaction gave ∼20 mg
of higher boiling residue. GCꢀFID analysis of this residue
corresponded to CAE, consistent with it being of lignin origin.
GCꢀFID traces for analogous runs with pine sawdust and with
cellulose are shown in SI Figure S-1, illustrating the much greater
quantity of the CAE component in the former.
Figure 2. Temporal solubilization of pine sawdust for mixtures of
substrate (100 mg), Cu20-PMO (100 mg), and MeOH (3.0 mL) heated
at 320 °C in minireactors.
Reactions with cellulose (200 mg) and Cu20-PMO (100 mg)
were also carried out at different reaction times (0.5ꢀ8 h). At
Small-Scale Reactions with Cellulose. The reactivity of
cellulose alone was verified by heating cellulose fibers (100 mg),
Cu20-PMO (100 mg), and MeOH (2.4 g) in a minireactor for
0
.5 h, 60% solubilization of cellulose was achieved, and this was
complete within the first 2 h (SI Table S-2). Interestingly, the
product stream, as evaluated by GCꢀFID, continued to evolve
during this time sequence with the lighter components becoming
more prominent and components having the longest retention
times becoming less so (see Figure S-3) presumably as the latter
undergo further processing via retro-Aldo and dehydration
processes (see below). Full solubilization of cellulose was also
obtained after 8 h reaction at a lower temperature (280 °C).
Additionally, the effect of different cellulose loadings (100, 200,
400, and 600 mg) at the same catalyst loading was probed for an
8 h reaction time at 320 °C (SI Table S-2). All experiments resulted
in approximately 100% conversion of cellulose to methanol soluble
8
h at 320 °C. Conversion was ∼98%, and the standard gas mixture
of H , CO, CO , and CH was observed. Most significantly, the
2
2
4
methanolic liquid collected was 8.6 wt % HAE with only a trace
of CAE (0.4%) (Figure 4). We attribute formation of small
quantities of C + alcohols and ethers from cellulose to con-
7
densation and/or hydroformylation of reactive intermediates
generated by dehydration/hydrogenation under the processing
conditions as well as possibly some polymethyl glycosides or
15
their partially reduced/deoxygenated forms.
When the liquid phase product from an analogous cellulose
conversion (100 mg) was subjected to rotary evaporation at
1
4093
dx.doi.org/10.1021/ja205436c |J. Am. Chem. Soc. 2011, 133, 14090–14097

Journal of the American Chemical Society
ARTICLE
Figure 4. GCꢀFID trace of the liquids from the reaction of cellulose fibers and Cu20-PMO and methanol in a minireactor at 320 °C for 8 h. (Recorded
on a Agilent model 6890 GC with a 30 m DB-5 column). Note: The vertical axis is FID signal intensity; the horizontal axis is retention time in minutes.
species, and the GCꢀFID chromatograms of the liquid phase
products were nearly the same (SI Figure S-4).
as seen with the minireactors but with a measurable quantity of
CAE (0.08 g). Conversion to liquid products was >98%.
We also probed the effect of added syngas on reaction
patterns. The Parr reactor was charged with pine flour (1.0 g),
Reaction with Glucose. A minireactor experiment with
glucose (200 mg), MeOH (2.4 g), and Cu20-PMO (100 mg)
at 320 °C for 8 h was carried out. GCꢀFID analysis showed the
liquid products to be similar to those obtained by analogous
treatment of cellulose (SI Figure S-5), suggesting that cellulose
may be first converted to monosaccharide units under these
conditions.
Cu20-PMO (0.5 g), MeOH (7.8 g), CO (133 mmol), and H
2
(40 mmol) to an initial pressure of 52 bar. The mixture was
heated for 2.0 h at 320 °C and then for 4.0 h at 300 °C, during
which a maximum pressure of 194 bar was observed. Analysis
found H , CO, CO , and CH (139, 116, 33, and 2.3 mmol,
2 2 4
In a control reaction, methanol and Cu20-PMO were heated
in a minireactor at 320 °C for 8 h in the absence of substrate. The
liquid products were HAE (0.3 wt %) and dimethyl ether (DME,
respectively) in the gases released and 10.0 wt % HAE and 5.8 wt %
CAE in the liquid methanol phase. Conversion was ∼85%. Thus,
it appears that CO/H mixtures may, at least partially, supplant
2
0
.3 wt %) in MeOH (SI Figure S-6). The quantity of HAE thus
reducing equivalents from CH OH.
3
formed is minor compared to that found when a biomass-derived
substrate is present.
Larger-Scale Batch Reactions. These experiments demon-
strated that the liquid products can be readily separated into
lighter and heavier fractions. Excellent mass balances were
observed. For example, the reaction of eucalyptus wood chips
Catalyst Stability Studies. Minireactor runs were carried out
in which the catalyst was recycled and mixed with fresh MeOH
and solid substrate. Each run was 3.0 h at 320 °C. Initially, MeOH
(2.4 g) and Cu20-PMO (220 mg) were heated without substrate,
the liquids and gases collected and the solids washed with MeOH
(∼10 mL). The solids were dried, weighed, and returned to the
same reactor with fresh MeOH (2.4 g) plus cellulose fibers (400
mg), and the process was repeated four times with new substrate.
The mass of solid (catalyst and unreacted substrate) recycled
to the reactor was ∼255ꢀ275 mg, and it was estimated that ∼3ꢀ
5 mg of solids were lost each time due to handling. The product
solution for each cycle with substrate was analyzed by GCꢀFID,
and the run-to-run product distributions were essentially indis-
tinguishable with <1% deviation in relative intensities of indivi-
dual peaks (Figure 5). Following the 5 cycles (4 with substrate), a
50 mg portion of the remaining solid (265 mg total) was titrated
(
6.6 g, 1.4ꢀ2.0 mm size range), Cu20-PMO (1.0 g), and MeOH
(15.8 g) for 4.5 h at 300 °C in the larger reactor led to an
autogenous pressure of 180 bar. Product workup gave: (a) the
gases H , CO, CO , and CH (134, 4.6, 67, and 15.1 mmol,
2
2
4
respectively, ∼3.6 g total) plus volatile organics (0.5 g), which
were collected by room temperature, atmospheric distillation
into a gas collection bag; (b) a pale yellow liquid (15.6 g)
collected by room temperature vacuum distillation (0.1 Torr);
(
(
c) methanol-soluble organics (0.8 g) recovered by washing; and
d) residual solids (2.5 g). GCꢀFID analysis of the liquid
distillate found methanol, HAE (2.8 g), and CAE (0.8 g) (SI
Figure S-7 and Table S-3). On the basis of the mass of the residual
solids, the above reaction gave 77% conversion in 4.5 h. When
these solids were subjected to further processing with methanol
for 8 h at 320 °C, half the unreacted biomass was solubilized;
however, acid digestion of the inorganic components suggested
that the remainder of the organic residue contained some char.
A Parr reactor (38 mL) run with cellulose (0.79 g) gave a
mixture of liquid products (SI Figure S-8) mostly HAE (0.69 g)
with aq HNO ; the insoluble fraction was 5 mg of a white,
3
flocculent solid. Thus, we estimated that ∼27 mg of cellulose was
carried over between runs. The remaining solid (215 mg) plus
MeOH (2.4 g) was returned to the reactor at 320 °C for 3 h and
the residue (190 mg) was completely soluble in aqueous HNO
indicating that overall substrate conversion was >99%.
,
3
This process was duplicated with eucalyptus chips (<250 μm)
with an ash content of 0.9 wt % determined by published
16
procedures. Cu20-PMO (320 mg) was first treated with pure
1
4094
dx.doi.org/10.1021/ja205436c |J. Am. Chem. Soc. 2011, 133, 14090–14097

Journal of the American Chemical Society
ARTICLE
MeOH (2.4 g) for 3.0 h at 320 °C, while the subsequent five
cycles contained eucalyptus chips (400 mg) and MeOH (2.4 g).
The mass of solids recycled to the reactor was ∼425ꢀ440 mg
(
after washing with MeOH and drying) with 3ꢀ5 mg lost
between runs due to handling. Compared to the cellulose
reactions, a greater run-to-run variability was observed in the
GCꢀFID analyses of product liquids from the eucalyptus chips
(SI Figure S-9) with greater quantities of heavier products being
detected in the later runs. After 6 cycles (5 with substrate), a
7
5 mg sample of the solid (430 mg total) was titrated with nitric
acid and found to contain ∼10 mg of a dark brown, flocculent
solid. We estimated that organic residue carried over between
runs was ∼57 mg. When the remaining solid was heated in
MeOH (2.4 g) at 320 °C for 3 h, the organic material was fully
converted to soluble products; thus, total conversion was >99%.
Therefore although apparent changes occur with continued conver-
sion of lignocellulose by the UCSB Process (in contrast with pure
cellulose), the conversion of solid biomass to liquids remains highly
efficient.
Figure 5. GCꢀFID trace showing liquid products of a single cellulose
conversion experiment (blue) compared to the pool of liquid products
from the four cellulose conversion experiments (red). (Recorded on a
Agilent model 6890 GC with a 30 m DB-5 column). Note: The vertical
axis is relative FID signal intensity; the horizontal axis is retention time in
minutes.
The washed and dried solids (260 mg) from the multiple
eucalyptus chips runs were calcined at 450 °C for 5 h to obtain a
very dark green material (240 mg). A 100 mg sample was
provides conditions suitable for solubilizing constituent polymers
to give “active cellulose” and “active lignin”. Depolymerization of
17
dissolved in aq HNO and subjected to ICP elemental analysis.
the solubilized cellulosic components would be followed by
3
2
2
Notably, the Cu/Mg/Al weight ratio was approximately the same
as that of the initial catalyst. However, the composition of the
recovered solid reflected collection of some wood ash as evi-
denced by increased sodium, potassium, and calcium (SI Table
S-3). ICP analysis of product solutions, combined washes, and so
forth, showed extremely small quantities of Na, K, and Al, and
essentially no Cu, Mg, or Ca; hence, the catalyst is resistant to
leaching by the medium.
Mechanistic Interpretation. Conversion of lignin and cellu-
lose to liquid fuels requires reduction of these highly function-
alized biopolymers. The process described here uses catalytic
methanol reforming to provide the necessary reducing equiva-
lents R, although the introduction of syngas may also serve this
purpose. If we define the 1-electron reducing equivalents pro-
sequential dehydrations and retro-aldol condensations coupled
to catalytic hydrogenation of unsaturated intermediates.
The “active lignin” component is depolymerized by aryl ether
1
2,13
hydrogenolysis followed by aromatic ring hydrogenation.
Successive dehydration and catalytic hydrogenations would lead
to propylcyclohexanol derivatives, the latter step being faster
than competitive polymerizations of reactive intermediates to
chars. The fact that glucose and cellulose display qualitatively
similar product suites suggests that rapid solvolysis or hydrogenolysis
of cellulose to monosaccharide units is the first step. More specific
proposals regarding transformations of cellulosic components to
HAE are offered in Scheme S-1 of the Supporting Information.
The treatment of lignin models by the process described here
also led to considerable ring methylation of the eventual
12,13
vided by the system as n , then n = 6n + 2nH2 + 2nCO. For
products.
Analogous methylation of unsaturated intermedi-
R
R
MeOH
example, in the larger-scale conversion of cellulose described
ates is a likely source of branched alcohols in the HAE products
from the treatment of cellulose.
above, n equaled 720 mmol at the beginning of the reaction and
R
6
10 mmol at completion. The difference (Δn = 110 mmol) is
Copper/metal oxide are also proficient catalysts for carbonyl
R
20
consistent with the reduction of the cellulose (4.9 mequiv of
hydrogenation, aryl ether hydrogenolysis, and aromatic ring
12
C (H O) ) to a mixture of C ꢀC alcohols such as in the
hydrogenations, but are not FischerꢀTropsch catalysts. Thus,
the combination of in situ reduced Cu20-PMO and the sc-MeOH
medium appear to be ideally suited for facilitating the complex set
of physical processes and chemical reactions required to convert
lignocellulose composites to liquid fuels.
6
2
5
2
6
simplified model illustrated in eq 3.
1
=yfC ðH OÞ g þ 24R f 2C H OH
6
2
5
y
2
5
þ C4H9OH ð þ 2H2O, assuming 24R ¼ 12H2Þ
ð3Þ
Potential Implementation. Transportation costs limit the
range over which biomass solids can be collected and moved to a
17
Cellulose has been shown to form char under vacuum and in
1
8
11
sc-MeOH at the temperatures utilized in the present studies.
Similarly, organosolv lignin forms chars in sc-MeOH in the
processing facility. We envision that biomass conversion via the
process described here would occur at moderate-sized facilities
located where lignocellulose wastes or crops are generated.
Three variations of this methodology are described in Scheme 3.
The “methanol option” involves utilizing MeOH both as the
reaction medium and as a “liquid syngas” for reducing biomass
components. This option should be the mosteasilyscaled fromthe
13
absence of catalyst. Since this undesired byproduct is absent
in nearly every case when cellulose or woody materials are treated
in the process described here, chemical pathways under these
conditions must intercept the unsaturated intermediates that
would otherwise polymerize to such wastes. Scheme 2 proposes a
model for this transformation of an insoluble lignocellulose
matrix to liquids.
current technology. The “methane option” uses CH -to-syngas
4
reforming to generate the necessary reducing equivalents. In the
United States, an advantage of the latter option is that the methane
distribution network is already extensive, including areas in the
Key roles for the copper/metal oxide catalyst include metha-
1
9
nol reforming to give the necessary reducing equivalents,
hydrogenation of unsaturated intermediates generated by
dehydration, and the water gas shift. The sc-MeOH medium
2
0
23
Midwest that would be major suppliers of biomass feedstocks.
21
The disadvantage would be the addition of another process to the
1
4095
dx.doi.org/10.1021/ja205436c |J. Am. Chem. Soc. 2011, 133, 14090–14097

Journal of the American Chemical Society
ARTICLE
a
Scheme 2. Model for Converting Lignocellulose Solids to Fluid Alcohols by the UCSB Process
a
(a) “Cellulose” refers to cellulose, hemicellulose, and the cellulosic component of lignocellulose; (b) “active cellulose” consists of soluble oligomers and
methylated monomers; (c) “lignin” refers to the lignified component; (d) “active lignin” refers to solubilized lignified material or lignin oligomers; (e)
depolymerization and demethoxylation by hydrogenolysis; (f) aromatic hydrogenation; (g) dehydration-hydrogenation.
Scheme 3. Options for Biomass Conversion via the UCSB
Process (Material Inputs and Outputs Indicated by Ovals)
biomass solids via this process would have the potential of reducing
the CO burden from liquid fuel utilization without compromising
2
food supply. Ongoing studies are focused on characterizing and
optimizing the earth-abundant catalysts used in this process, on
elucidating key mechanistic issues, and on evaluating how the
reaction variables control yields and selectivity in product streams.
For example, one approach may be to develop simple deoxygena-
tion procedures to generate clean mixtures of alkanes.
’
ASSOCIATED CONTENT
S
Supporting Information. Tables and figures describing
b
the temporal conversion of substrates, displaying GCꢀFID data
for additional experiments and summarizing analyses of catalysts
before and after reactions. Scheme of hypothetical reaction
network leading to products from cellulose. This material is
available free of charge via the Internet at http://pubs.acs.org.
’
AUTHOR INFORMATION
Corresponding Author
ford@chem.ucsb.edu
front end of this process, although methane reforming is well-
established technology. The “biomass-only option” uses a bio-
mass-to-syngas conversion at the front end of the biomass-to-fuels
process. The advantage would be an even lower CO footprint
2
25
’ ACKNOWLEDGMENT
combining an existing technology (biomass-to-syngas) with a
new one (the UCSB process), and in the long run would clearly be
a desirable goal. With the latter two options, the methanol utilized
as the reaction medium could be generated internally.
In summary, we have described the quantitative conversion of
lignocellulose solids to combustible liquids via a unique single stage,
methanol mediated process. Little or no char is formed, and this
result plus the observation that the copper-based metal oxide
catalyst can be reused suggests that this process is compatible with
high yield, continuous operation. Successful utilization of renewable
We thank Professor Robert Rinker of UCSB for discussion
and insight regarding analytical procedures. This research re-
ceived partial support from the Center for Enabling New
Technology through Catalysis (National Science Foundation
grant CHE-0650456), from the UCSB Partnership for Interna-
tional Research and Education in Electron Chemistry and
Catalysis at Interfaces (National Science Foundation grant
OISE-05302680) and from the Petroleum Research Fund ad-
ministered by the American Chemical Society (46828-AC3).
1
4096
dx.doi.org/10.1021/ja205436c |J. Am. Chem. Soc. 2011, 133, 14090–14097

Journal of the American Chemical Society
REFERENCES
ARTICLE
’
(20) Twigg, M. V.; Spencer, M. S. Appl. Catal., A 2001, 212, 161–174.
(
21) Gokhale, A. A.; Dumesic, J. A.; Mavrikakis, M. J. Am. Chem. Soc.
008, 130, 1402–1414.
22) Sasaki, M.; Goto, K.; Tajima, K.; Adschiri, T.; Arai, K. Green
Chem. 2002, 4, 285–287.
23) About U.S. Natural Gas Pipelines; http://www.eia.doe.gov/
pub/oil_gas/natural_gas/analysis_publications/ngpipeline/index.html.
24) A referee has suggested that, under hydrogenating conditions,
(1) Regalbuto, J. R. Science 2009, 325, 822–824.
2
(2) Huber, G. W.; Iborra, S.; Corma, A. Chem. Rev. 2006, 106, 4044–
(
4098.
(
3) (a) Suganuma, S.; Nakajim, K.; Kitano, M; Yamaguchi, D.; Kato,
(
H.; Hayashi, S.; Hara, M. J. Am. Chem. Soc. 2008, 130, 12787–12793. (b)
Binder, J. B.; Raines, R. T. J. Am. Chem. Soc. 2009, 131, 1979–1985. (c)
Toor, S. S.; Rosendahl, L.; Rudolf, A. Energy 2011, 36, 2328–2342.
(
monosaccharides could be converted to sugar alcohols that would not be
seen by gas chromatographic techniques. The absence of any evapora-
tion residue from the liquid product fractions when glucose or cellulose
components are used as substrates suggests that the unimolecular
dehydration and retro-aldol reactions in sc-MeOH are rapid relative to
the hydrogenation of glucose and related precursors. On the other hand,
hydrogenation of the resulting unsaturated species, precursors to
intractable chars, must, in turn, be rapid relative to rate-limiting
solubilization of biomass solids given that nonvolatile organics are at
most a very small fraction of the products.
(
4) (a) Ji, N.; Zhang, T.; Zheng, M. Y.; Wang, A.; Wang, H.; Wang,
X.; Chen, J. G. Angew. Chem., Int. Ed. 2008, 47, 8510–8513. (b) Pang, J.
F.; Zheng, M. Y.; Wang, A. Q.; Zhang, T. Ind. Eng. Chem. Res. 2011,
5
0, 6601–6608.
5) vom Stein, T.; Grande, P.; Sibilla, F.; Commandeur, U.; Fischer,
R.; Leitner, W.; de Maria, D. Green Chem. 2010, 12, 1844–1849.
6) Huber, G. W.; Shabaker, J. W.; Dumesic, J. A. Science 2003,
(
(
3
2
00, 2075–2077. (b) Rinaldi, R.; Schueth, F. ChemSusChem 2009,
, 1096–1107. (c) Geilen, F. M. A.; Engendahl, B.; Harwardt, A.;
Marquardt, W.; Klankermayer, J.; Leitner, W. Angew. Chem., Int. Ed.
010, 49, 5510–5514.
7) (a) Mehdi, H.; Fabos, V.; Tuba, R.; Bodor, A.; Mika, L.; Horvath,
(25) Klass, D. L. Biomass for Renewable Energy, Fuels, and Chemicals;
2
Academic Press; San Diego, CA, 1998; pp 271ꢀ326.
(
I. T. Top. Catal. 2008, 48, 49–54. (b) Hanson, S. K.; Baker, R. T.;
Gordon, J. C.; Scott, B. L; Sutton, A. D.; Thorn, D. L. J. Am. Chem. Soc.
2
009, 130, 428–429. (c) Palkovits, R. Angew. Chem., Int. Ed. 2010,
49, 4336–4338.
(8) (a) Yan, N.; Zhao, C.; Dyson, P. J.; Wang, C.; Liu, L.; Kou, Y.
ChemSusChem 2008, 1, 626–629. (b) Hanson, S. K.; Baker, R. T.;
Gordon, J. C.; Scott, B. L; Thorn, D. L. Inorg. Chem. 2010,
49, 5611–5618. (c) Zakzeski, J.; Bruijnincx, P. C. A.; Jongerius, A. L.;
Weckhuysen, B. M. Chem. Rev. 2010, 110, 3552–3599. (d) Nichols,
J. M.; Bishop, L. M.; Bergman, R. G.; Ellman, J. A. J. Am. Chem. Soc.
2
010, 132, 12554–12555.
9) (a) Zhao, C.; Kou, Y.; Lemonidou, A. A.; Li, X.; Lercher, J. A.
(
Angew. Chem., Int. Ed. 2009, 48, 3987–3990. (b) Calvo-Flores, F. G.;
Dobado, J. A. ChemSusChem 2010, 3, 1227–1235.
0
(
10) Milne, T. A.; Evans, R. J. Abatzoglou, N.Biomass Gasifier Tars:
Their Nature, Formation and Conversion; Report No. NREL/TP-570-
5357; National Renewable Energy Laboratory: Golden, CO, 1998;
http://www.osti.gov/bridge.
2
(
11) Willems, P. A. Science 2009, 325, 707–708.
(12) Macala, G. M.; Matson, T. D.; Johnson, C. L.; Lewis, R. S.;
Iretskii, A. V.; Ford, P. C. ChemSusChem 2009, 2, 215–217.
13) Barta, K.; Matson, T. D.; Fettig, M. L.; Scott, S. L.; Iretskii, A. V;
Ford, P. C. Green Chem. 2010, 12, 1640–1647.
14) Operationally, the HAE are defined as those eluting during
(
(
the first 10.3 min including 1-hexanol (10.26 min) of the GCꢀFID
program using the Agilent Gas Chromatograph described in the
Experimental Section. Major peaks were confirmed by comparison
to authentic samples. Similarly, CAE are defined as those eluting
after 10.3 min and correspond to materials of the type identified by
GCꢀMS experiments as being formed from lignin precursors (ref 13).
Analyses using the GCꢀFID program of Shimadzu GC (see Experi-
mental Section) gave very similar product patterns but with longer
retention times.
(
15) (a) Comparable results were obtained with cellulose when the
Cu20-PMO was replaced by a Cu/Zn/Al metal oxide catalyst (100 mg)
with 27/53/20 molar ratio (Cu/ZnO/Al ), prepared according to
2 3
O
literature procedures (ref 15b) and known to be active for methanol
reforming and the water gas shift. (b) Herman, R. G.; Klier, K.;
Simmons, G. W.; Finn, B. P.; Bulko, J.l B.; Kobylinski, T. P. J. Catal.
1
1
1
979, 56, 407–429.
16) Misra, M. K.; Ragland, K. W.; Baker, A. J. Biomass Bioenergy
993, 4, 103–116.
17) Bradbury, A. G. W.; Sakai, Y.; Shafizadeh, F. J. Appl. Polym. Sci.
979, 23, 3271–3280.
(
(
(18) Ishikawa, Y.; Saka, S. Cellulose 2001, 8, 189–195.
(19) Sunggyu, L. Methanol Synthesis Technology; CRC Press: Boca
Raton, FL, 1990.
1
4097
dx.doi.org/10.1021/ja205436c |J. Am. Chem. Soc. 2011, 133, 14090–14097