Production of hydrogen, alkanes and polyols by aqueous phase processing of wood-derived pyrolysis oils

View Article Online / Journal Homepage / Table of Contents for this issue

PAPER

www.rsc.org/greenchem | Green Chemistry

Production of hydrogen, alkanes and polyols by aqueous phase processing

of wood-derived pyrolysis oils

Tushar P. Vispute and George W. Huber*

Received 8th April 2009, Accepted 4th June 2009

First published as an Advance Article on the web 9th July 2009

DOI: 10.1039/b912522c

Pyrolysis oils are the cheapest liquid fuel derived from lignocellulosic biomass. However, pyrolysis

oils are a very poor quality liquid fuel that cannot be used in conventional diesel and internal

combustion engines. In this paper we show that hydrogen, alkanes (ranging from C

1

to C ) and

6

polyols (ethylene glycol, 1,2-propanediol, 1,4-butanediol) can be produced from the aqueous

fraction of wood-derived pyrolysis oils (bio-oils). The pyrolysis oil was ﬁrst phase separated into

aqueous and non-aqueous fraction by addition of water. The aqueous phase of bio-oil contained

sugars; anhydrosugars; acetic acid; hydroxyacetone; furfural and small amounts of guaiacols. The

aqueous fraction was subjected to a low temperature hydrogenation with Ru/C catalyst at

◦

1

25–175 C and 68.9 bar. The hydrogenation step converts the various functionalities in the

bio-oil (including aldehydes; acids; sugars) to corresponding alcohols. Undesired methane and

light gases are also produced in this low-temperature hydrogenation step. Diols (ranging from C2

to C4) and sorbitol are obtained as major products in this step. After the low temperature

hydrogenation step either hydrogen or alkanes can be produced by aqueous-phase reforming

(

APR) or aqueous-phase dehydration/hydrogenation (APD/H) respectively. APR was done with

◦

a 1 wt% Pt/Al

2

O

3

catalyst at 265 C and 55.1 bar. Hydrogen selectivities of up to 60% were

observed. The hydrogen selectivity was a function of space velocity. A 4 wt% Pt/SiO

catalyst at 260 C and 51.7 bar was used for alkane production by APD/H. The carbon

2

-Al

2

O

3

◦

conversion to gas phase products of 35% with alkane selectivity of 45% was obtained for a WHSV

of 0.96 h^-when hydrogen is produced in situ from bio-oil. Alkane selectivity can be improved by

supplying hydrogen externally. Alkane selectivities as high as 97% can be obtained when HCl is

added to the aqueous-phase of the bio-oil and hydrogen is supplied externally. Model compounds

for further bio-oil conversion studies are suggested.

1

.0 Introduction

The bio-oil can be produced in weight yields as high as

2

7

5 wt% of the original dry biomass weight. The bio-oil product

Diminishing petroleum resources combined with environmental

concerns related to fossil fuels are making it increasingly

important to ﬁnd renewable and sustainable liquid fuel sources.

The only sustainable source of carbon that can be used to

make liquid fuels is carbon ﬁxed by photosynthesis in plant

biomass. Furthermore, bio-fuels, fuels derived from biomass,

3,4

typically contains 60–75% of the initial energy of the biomass.

Furthermore, fast pyrolysis has signiﬁcantly lower capital and

operating costs than competing technologies such as gasiﬁcation

5

andfermentation/hydrolysis. A wide variety of feedstock can be

used to produce bio-oil including wood, corn stover, agricultural

waste and forest waste.

1

are the only sustainable liquid fuels. Lignocellulosic biomass

Bio-oil is the cheapest liquid fuel that can be made from

(

which includes trees, grasses, energy crops and agricultural

5

lignocellulosic biomass. However, the bio-oil is a very poor

wastes) is currently the cheapest and most abundant form of

plant biomass. However, there are many challenges associated

with developing cost-effective processes for biomass conversion

to liquid fuels. One very promising conversion technology

is fast pyrolysis of biomass. Biomass fast pyrolysis involves

rapidly heating the biomass without any external oxygen to

intermediate temperatures producing a liquid product called

bio-oil (or pyrolysis oil).

quality fuel that cannot be used in gasoline and diesel fuel

engines for several reasons. Bio-oils have half the energy density

compared to petroleum derived fuels due to their large oxygen

content. Bio-oil is nearly insoluble with petroleum fuels. Bio-oils

are acidic, having a pH of around 2.5. Bio-oils also degrade with

time. Bio-oils are a complex mixture of more than 300 different

3,6

compounds. This mixture arises from the depolymerization

and fragmentation of cellulose, hemicellulose and lignin, the

three building blocks of the lignocellulosic biomass. Bio-oil,

depending on its source, is found to contain various acids

(

formic, acetic, propanoic), aldehydes (acetaldehyde, hydroxy-

Department of Chemical Engineering, University of Massachusetts,

Amherst, 686 North Pleasant Street, Amherst, MA 01003, USA.

E-mail: huber@ecs.umass.edu; Fax: +1 413-545-1647; Tel: +1

acetaldehyde), ketones (hydroxyacetone, hydroxybutanone, ace-

tone), sugars (glucose, xylose, fructose), guaiacols, syringols,

furans, furfurals and water.

3,7

4

13-545-0276

This journal is © The Royal Society of Chemistry 2009

Green Chem., 2009, 11, 1433–1445 | 1433

View Article Online

In spite of the difﬁculties of bio-oils they are very promising

feedstocks for the production of conventional transportation

fuels. The majority of effort of bio-oil conversion to date has

focused on hydrotreating of the bio-oil and catalytic cracking

2.0 Experimental and materials

The bio-oil used in these studies was obtained from Renewable

Oil International (ROI) LLC and was made from oak wood

16,17

6,8–11

using ROI’s fast pyrolysis process.

ROI uses a proprietary

of the bio-oil.

In this paper, we report a new approach for

Auger reactor to make the bio-oils in high yield. The bio-oil was

stored in the refrigerator to minimize ageing.

conversion of the bio-oil by aqueous phase processing (APP).

Aqueous phase processing, developed by Dumesic and co-

workers, involves the selective conversion of sugars and polyols

12,13

to targeted hydrogen and alkanes.

The major advantage of

2

.1 Elemental analysis, viscometry and catalyst

this approach is that targeted products can be selectively pro-

duced by carefully controlling the chemistry that occurs in the

aqueous phase. Other advantages of aqueous phase processing

include: (1) liquid fuels can be made that can be used with

gasoline and diesel engines, (2) high thermal efﬁciencies because

the process occurs in the liquid phase, (3) no energetically-intense

distillation steps, and (4) high rates of production per reactor

volume. Liquid-phase catalytic processing of biomass-derived

compounds offers unique opportunities for achieving high yields

of speciﬁc, and well-deﬁned, liquid fuels from biomass.

characterization

Elemental analysis (C, H and O) of the bio-oil and its various

fractions was done at Schwarzkopf Microanalytical Laboratory,

Woodside, NY, USA. Ash content of the bio-oil was found

by heating about 1 g of bio-oil in a furnace in presence of

◦

air at 600 C for 6 hours. The amount of ash remaining was

measured at the end of the run. The viscosity of the bio-oil and

its various fractions was measured in a capillary glass viscometer

(

from Cannon Instrument Company). Viscosity measurement

of the water insoluble bio-oil (WIBO) was carried out in a TA

instrument, AR2000 using a concentric cylinders geometry. All

the viscosity measurements were done at 25 C. Catalysts were

characterized by hydrogen chemisorption in a Quantachrome

Autosorb 1C. About 0.1 g catalyst was loaded in a chemisorption

cell and reduced. Reduction temperature regime was: room

Hydrogen is an important reactant in a majority of reﬁnery

14

processes to make liquid fuels and biofuels. Hydrogen can be

produced in our process by aqueous-phase reforming (APR)

which is a part of APP. APR involves C–C bond cleavage (to

produce CO) followed by the water gas shift reaction over

◦

13

supported metal catalysts. The overall reaction for APR of

sorbitol is shown in eqn (1). Side reactions such as methanation

and dehydration/hydrogenation take place to produce C1 to C6

alkanes as by-products.

◦

-1

◦

temperature to 260 C at 30 C h and hold at 260 C for

◦

2

8

h. The isotherms were then taken at 30 C and 267, 533,

00, 1067 and 1333 N m . The chemisorption cell was then

-

2

◦

evacuated for 30 minutes. Isotherms were again taken at 30 C

-

2

and 267, 533, 800, 1067 and 1333 N m to calculate the

extent of physisorption. The amount of chemisorbed hydrogen

was calculated from the difference between the two isotherms.

The liquid samples were analyzed for the carbon content by a

Shimadzu 5000A Total Organic Carbon (TOC) analyzer. The

aqueous samples were further diluted by distilled water to a

concentration below 1000 ppm carbon for the TOC analysis. The

TOC samples were fortiﬁed with 0.5 N HCl (1.2 gm for 38.8 gm

C

6

H

14

O

6

+ 6H

2

O → 6CO

2

+ 13H

2

(1)

Alkanes ranging from C1 to C6 can be produced more selec-

tively by aqueous phase dehydration/hydrogenation (APD/H),

also a subset of APP. This involves passing the aqueous

solution of oxygenated hydrocarbon over a bi-functional catalyst

containing metal and acidic sites such as Pt supported on silica-

alumina. Hydrogen can be produced either in situ or ex situ on

metal sites according to eqn (1). The alkanes are produced by

dehydration reactions on solid acid sites to produce dehydrated

products, and hydrogenation reactions on metal sites, as shown

in eqn (2). If the alkane and hydrogen production reactions are

balanced, then alkanes can be directly produced from sorbitol

diluted sample) to remove dissolved CO

standardized by sorbitol solutions of known concentrations.

2

. The TOC analyzer was

2.2 Bio-oil extraction and pre-treatment

with CO

2

and H

2

O as the byproducts, as shown in eqn (3).

The bio-oil was mixed with distilled water to separate into two

phases: an aqueous phase (WSBO: water soluble fraction of

bio-oil) and an organic phase (WIBO: water insoluble bio-oil

fraction). The mixture was then centrifuged in a Marathon 2100

centrifuge (Fisher Scientiﬁc) at 10 000 rpm for 30 minutes to

ensure the phase separation. The two phases, aqueous (top)

and non-aqueous (bottom), were then separated by decanting.

The weight of the aqueous faction was measured to determine

the amount of bio-oil that dissolved in water. For experimental

purposes, 28 g of bio-oil was added to 112 g of water and mixed

well. The aqueous and non-aqueous phases were separated by

centrifugation followed by decanting. The resulting aqueous

solution is about 13.4 wt% water soluble bio-oil in water, which

is about 5 wt% carbon in water. This aqueous solution was used

in the batch hydrogenation experiments. The product of batch

hydrogenation was further diluted to about 2 wt% carbon in

water. This solution was then used as the feed for further APP

Smaller alkanes are obtained as by-products due to the C–C

bond cleaving reactions.

C

6

H

14

O

6

+ 6H

2

→ C

6

H

14 + 6H

2

O

(2)

(3)

(19/13)C

6

H

14

O

6

→ (36/13)CO

2

+ C

6

H

12 + (42/13)H O

2

The majority of the previous APP work was done with pure

compounds.

12,13,15

In this paper we show that APP chemistry can

be extended for conversion of the aqueous fraction of bio-oil

that is derived from lignocellulosic biomass. This involves the

conversion of a large amount of functionalities to alkanes and

hydrogen. This paper thus proposes a new route for production

of renewable liquid fuels from biomass by combining fast

pyrolysis with aqueous phase processing.

1

434 | Green Chem., 2009, 11, 1433–1445

This journal is © The Royal Society of Chemistry 2009

View Article Online

experiments. In the remaining part of the paper this will be called

WSBO.

Product No. 78–2010) in distilled water was added drop wise

to the silica-alumina powder (SiO to Al

2

O

3

ratio = 4, Davison

SIAL 3125) with continuous mixing. The wet catalyst was then

◦

2.3 Hydrogenation of aqueous fraction of bio-oil

dried in an oven at 80 C for 7–8 h. The catalyst was then

-

1

calcined in air ﬂowing at 300 ml min . The temperature regime

Hydrogenation of the aqueous fraction of the bio-oil was carried

out in a 170 ml Parr batch reactor. About 80 ml of the aqueous

fraction of the bio-oil (about 5 wt% carbon) was loaded in the

reactor along with 3–4 g (wet basis, 50 wt% moisture content) of

◦

for calcining was: room temperature to 260 C in 3 h, then hold at

◦

2

60 C for 2 h. Both the catalysts (Pt/Al

2

O

3

and Pt/SiO

2

-Al

2

3

O )

were reduced in the ﬂow reactor with hydrogen ﬂowing from

-

1

the bottom at about 200 ml min . The temperature regime used

5

4

wt% Ru/activated C catalyst (Strem Chemicals, Product No.

4–4059). The reactor was then purged at least 4–5 times with

for reduction of Pt/SiO

to 450 C at 50 C h , then hold at 450 C for 2 h. The

2

-1

-Al

2

O

3

catalyst was: room temperature

◦

helium gas to get rid of the air present in the reaction vessel.

The reactor was then purged with hydrogen at least 4–5 times to

replace all the helium with hydrogen. The reactor pressure was

set to 48.3 bar by adding hydrogen and the heating and stirring

were started. Once the temperature reached the desired value,

the reactor pressure was increased to 68.9 bar total by adding

more hydrogen. Additional hydrogen was added to the reactor

during the course of reaction to compensate for the hydrogen

consumption. The total pressure was maintained at 68.9 bar.

The amount of hydrogen consumed during the reaction was

calculated from the decrease in pressure. Liquid samples were

withdrawn during the run from the liquid sampling tube. The

liquid samples were ﬁltered before analysis to remove the catalyst

particles. The liquid samples as well as the reactant and the ﬁnal

product of the batch hydrogenation were analyzed for carbon

content by a TOC analyzer. The product and feed compositions

were measured with a Shimadzu gas chromatograph (GC)

temperature regime used for reducing Pt/Al

room temperature to 260 C at 30 C h , then hold at 260 C

for 2 h.

2

O

3

catalyst was:

◦

-1

◦

The reactor tube was heated by a Lindberg (type 54032)

furnace. The liquid feed was fed to the reactor from the bottom

with the help of a JASCO PU980 HPLC pump at the rate of

-

1

0

.04 ml min . A gas–liquid separator was employed after the

reactor tube. Helium was supplied from the top as the carrier

-

1

-1

gas at ﬂow rates from 30 ml min to 60 ml min . The gaseous

products from the reactor (and helium carrier gas) ﬂow through

a back pressure regulator, used to maintain the pressure of

the reaction system. External hydrogen, required for catalyst

-

1

-1

reduction (~200 ml min ) or for the reaction (was ~100 ml min )

supplied from the bottom of the reactor. No carrier gas was used

when hydrogen was supplied externally. The gaseous products

were further analyzed by two online gas chromatographs (HP

5

H

890 series II). Permanent gases in the gaseous product (CO

) were analyzed by a thermal conductivity detector (TCD).

An Alltech HAYESEP DB 100/120 packed column (Part No.

2

and

(

model 2010) and a high performance liquid chromatograph

HPLC). A ﬂame ionization detector (FID) was used on the GC

2

to quantify all the reactants and products except sugars, sugar

alcohol and levoglucosan. The reactants and products were also

veriﬁed by GC-MS. A Restek Rtx-VMS (Catalog No. 19915)

column was used with a constant column linear velocity of

2

7

836PC) was used with the oven temperature held constant at

◦

5

C. The TCD and the injection port were held at 160

C

◦

-1

and 120 C, respectively. The column ﬂow rate was 1 ml min

with helium as carrier gas. Alkanes in the gaseous product were

analyzed on a FID with an Alltech AT-Q capillary column (Part

No. 13950). Helium was used as the carrier gas with a column

ﬂow rate of 1 ml min . The injection port and the detector

were both held at 200 C. The following GC oven temperature

-

1

3

1.3 cm s . Ultra high purity helium was used as the carrier gas;

◦

injector and detector were both held at 240 C. The GC oven was

programmed with following temperature regime: hold at 35 C

for 5 min, ramp to 240 C at 10 C min and hold at 240 C for

min. On HPLC, an RI detector (held at 30 C) was used to

◦

-

1

◦

-1

◦

5

◦

regime was used: hold at 40 C for 6 min, ramp to 180 C at

5

quantify sugars, sugar alcohol and levoglucosan in the feed and

product of batch hydrogenation. Bio-Rad’s Aminex HPX-87H

column (Catalog No. 125–0140) was used with distilled water as

◦

-1

◦

C min and hold at 180 C for 25 min. Carbon selectivity

to a particular alkane was calculated by dividing the carbon

moles in a particular alkane by total carbon moles in all of the

alkanes. For a particular catalyst loaded in the reactor, liquid

feed was started at time t = 0. Steady state was usually reached

within 8 hours. At least 3 gas samples were analyzed to ensure

the steady state. Liquid product accumulated in the gas–liquid

separator was drained then. The liquid product was analyzed

for the carbon content. Reaction parameters were then changed

for further studies. Time on stream for a particular catalyst from

the start of reaction (t = 0) was noted for each sample and was

denoted by time on stream for that catalyst.

-

1

the mobile phase with the ﬂow rate of 1 ml min . The column

◦

oven temperature was held constant at 30 C.

2

.4 Aqueous phase processing

The hydrogenated aqueous fraction of the bio-oil was diluted by

the addition of distilled water to about 2 wt% carbon in water.

This diluted product was used as the feed for the further liquid

phase processing. The hydrogenation of the aqueous fraction of

◦

bio-oil was carried out at 175 C and 68.9 bar total pressure

1

for 3 hours. A ₄¢¢ or ₂¢¢ stainless steel tube was packed with

the reforming or dehydration/hydrogenation catalyst with glass

wool plugs on both the sides. The catalyst used for reforming

3.0 Results

was 1 wt% Pt/Al

center (Product No. 4761–137). The dehydration/hydrogenation

catalyst was 4 wt% Pt/SiO -Al and was prepared by the

incipient wetness method. The appropriate amount of the

solution of tetraammineplatinum (II) nitrate (Strem Chemicals,

2

O

3

, and was obtained from the UOP research

3

.1 Characterization of bio-oil

2

O

3

Various characterization techniques were used to characterize

the bio-oil. Physical characterization was done by viscosity

and solubility measurements. GC-MS was used to identify the

This journal is © The Royal Society of Chemistry 2009

Green Chem., 2009, 11, 1433–1445 | 1435

View Article Online

components. GC-MS, GC-FID and HPLC were used for the

quantiﬁcation of the components.

Table 1 Identiﬁcation of major components of aqueous fraction of

bio-oil. The aqueous fraction of bio-oil was made by mixing 80 g of

water with 9 g of bio-oil

3

.1.1 Elemental analysis. Elemental analysis of the oak

Quantiﬁcation

method

Concentration

(mmole carbon L ) carbon

% of total

wood bio-oil was found to be 47.0 wt% carbon, 8.2 wt%

hydrogen and the rest oxygen. Nitrogen was not detected. The

oxygen content of 44.8 wt% is higher as compared to the typical

bio-oil compositions. Typical bio-oil oxygen content is in the

-1

Species

GC-FID

Hydroxyacetone

Hydroxyacetaldehyde

Guaiacols and

derivatives

Sugars

Levoglucosan

Acetic acid

135.5

28.1

30.8

6.5

1.4

1.5

18

range of 35 to 40 wt%. In addition to carbon, oxygen and

hydrogen, nitrogen (0 to 0.2 wt%) can be present in the bio-

HPLC

GC-FID

GC-MS

377.4

390.6

182.2

100.0

18.2

18.8

8.8

6

oil. Mineral components of the bio-mass (including potassium,

sodium, calcium and magnesium) also end up in the bio-oil in

trace quantities. Typical ash content of bio-oils is 0–0.2 wt%.

The ash content of our bio-oil was 0.3 wt%.

Furfural and

4.8

3

6

2

-furanone

Total carbon content 1244.6

identiﬁed by GC &

HPLC

60.0

3

.1.2 Solubility studies. Upgrading of the bio-oil without

using any solvent is preferred if possible. However, this route has

many problems, such as high viscosity and thermal instability of

bio-oil, clogging of reactors, considerable coking, and catalyst

deactivation due to coking. Bio-oil can be dissolved in a suitable

solvent and then treated for upgrading to help with these

problems. The oak wood bio-oil used in this study was found

to be almost completely miscible in methanol, iso-propanol and

acetone with some residue left that can be ﬁltered. About 14 wt%

of the bio-oil was found to be soluble in toluene, which is

a gasoline like solvent. Solubility of the bio-oil in diesel fuel

was about 4 wt%. The solubilities were measured by mixing

bio-oil and solvent in 1:1 proportion at room temperature

and the two layers formed were separated by decanting after

centrifugation. Toluene and diesel fuel fractions were found to

consist of various guaiacols. Low solubility of bio-oil in toluene

and diesel fuel signiﬁes its incompatibility with the conventional

liquid transportation fuels. An ideal solvent for hydroprocessing

of the bio-oil should dissolve a considerable fraction of bio-oil

in it and should be inert to hydrogenation. Alcohols such as

methanol and iso-propanol are not inert to the hydrogenation.

Whereas gasoline, diesel fuel or any saturated liquid alkane are

inert to hydrogenation but are incompatible with the bio-oil.

Two phases form when our bio-oil is mixed with water at a

water to bio-oil weight ratio of greater than 1 to 4. For the bio-

oil in this study, about 60 to 65 wt% of the bio-oil is soluble

in water irrespective of the amount of water added. Water is

inert for hydrogenation. Furthermore, a signiﬁcant fraction of

the bio-oil is water soluble, making water an ideal solvent for

hydroprocessing of bio-oil. The elemental analysis for mixing

bio-oil with water is shown in Fig. 1 for a water to bio-oil weight

mixture of 1 : 1. As can be seen in this ﬁgure the WSBO has

a higher oxygen content than the WIBO. About 62 wt% of the

original bio-oil is in the aqueous phase. This also includes water

that was present in the bio-oil. Our ﬁndings are similar to those

of Sipila et al., where 60 wt% of the bio-oil (made from mixed

Total carbon content 2075.9

measured by TOC

100

Fig. 1 Liquid–liquid extraction of the bio-oil in water.

carbon in the aqueous fraction of the bio-oil with our methods

used in this study. This did, however, help us in understanding

some of the bio-oil conversion reactions. The rest of the carbon

was probably present in the form of compounds like formic

acid, various furans, and higher molecular weight sugars (i.e.

cellobioses and trioses). The column used in HPLC can separate

sugar, sugar alcohols and anhydro sugars. But it cannot separate

various sugars from each other. A broad peak was observed

for sugars in HPLC. Various sugars may be present, including

glucsoe, xylose, fructose, mannose and galactose. Piskorz et al.

have identiﬁed hydroxyacetaldehyde up to 10 wt% in the bio-

oil. We found that only 1.5% of the carbon in WSBO was

from hydroxyacetaldehyde. Formic acid could not be observed

in our solution due to the large amounts of water we used and

hence is not quantiﬁed here. High molecular weight degradation

products of pentoses, hexoses are also present in the bio-oil and

were not detected by our analytical methods. Cellobiosan is

20

3

19,21

known to be present in bio-oil in a signiﬁcant amount.

et al. have identiﬁed phenolic compounds (phenol and its alkyl

derivatives) up to 20 wt% in a bio-oil made from P. indicus.

Phenol and its alkyl derivatives were not detected in signiﬁcant

amounts in the bio-oil we used for this study.

Luo

19

maple and oak wood) was found to be water soluble.

The water insoluble bio-oil (WIBO) was mixed with methanol

and analyzed by GC-MS. It mainly consists of guaiacol and

its substituted forms which are known to come from the

lignin part of the wood. The WSBO was analyzed with

GC-MS, GC-FID, HPLC and TOC as shown in Table 1. The

major components of the aqueous fraction are acetic acid,

hydroxyacetone, hydroxyacetaldehyde, levoglucosan, furfural,

22

3.1.3 Viscosity measurements. The various bio-oil frac-

tions have a wide range of viscosities, which are probably due

to both intramolecular and intermolecular interactions. Bio-

oils are also known to undergo viscosity changes during ageing

2

-furanone and sugars. We were only able to identify 60% of the

1

436 | Green Chem., 2009, 11, 1433–1445

This journal is © The Royal Society of Chemistry 2009

View Article Online

Table 2 Viscosity of the bio-oil, bio-oil fractions and bio-oil mixtures

converted to corresponding alcohols such as ethylene glycol,

propylene glycol and tetrahydrofurfuryl alcohol, respectively.

The challenge with the low temperature aqueous phase hy-

drogenation step is to selectively hydrogenate targeted C–O

bonds and not break C–C or C–O bonds. Cleavage of C–C and

C–O bonds results in formation of lighter products including

undesired methane. The aqueous phase hydrogenation reactions

in this study were all performed in a batch reactor.

a

Viscosity /cP

Bio-oil

153

8

5

2

0 wt% bio-oil in methanol

17.1

2.93

1.01

50 000

95.1

5.1

b

Water insoluble bio-oil

8

5

2

0 wt% water insoluble bio-oil in methanol

Batch hydrogenation of the aqueous fraction of bio-oil was

1.04

1.65

◦

done in the temperature range of 125–175 C with a 5 wt%

c

Water soluble bio-oil solution

Ru/Carbon catalyst. The hydrogen uptake of this catalyst was

a

-1

All the viscosity measurements are done in a capillary glass viscometer

except for the water insoluble bio-oil) at 25 C. TA instrument AR2000

used with concentric cylinders geometry. Solution was 50 wt% water

soluble bio-oil in water.

33.2 mmol H g dry catalyst, which corresponds to a surface

◦

c

b

(

to bulk Ru ratio of 6.7%. Table 3 depicts the data gathered for

batch hydrogenation where the temperature is increased in steps

◦

from 125 C to 150 C to 175 C. Ethylene glycol, propylene

glycol, butanediols, tetrahydrofurfuryl alcohol, g-butyrolactone

and 1,2-cyclohexanediol all reach their respective maximum

10,23

and bio-oil upgrading.

Table 2 shows the viscosity of bio-

oil, bio-oil mixtures and bio-oil fractions prior to upgrading.

The oak wood bio-oil used in our studies was found to have

the viscosity of 153 cP at 25 C. Typical viscosity values of the

bio-oil reported in the literature depicts an order of magnitude

variation. According to Bridgwater et al., the viscosity of bio-

oils ranges from 30 to 200 cP at 40 C. The viscosity can be as

high as 1000 cP at 40 C depending on the feedstock and process

conditions. The viscosity of the bio-oil decreases exponentially

upon addition of small amount of solvents. As shown in Table 2,

addition of 20 wt% of methanol to bio-oil results in an order of

magnitude decrease in the viscosity.

The water insoluble bio-oil is a highly viscous sticky brown

liquid with a viscosity of more than 50 000 cP. Its viscosity

also decreases exponentially upon addition of methanol. The

phenolic compounds originating from lignin are present in

◦

concentrations within 2.5 h at 125 C. All the hydroxyacetone

is also consumed during this same period. Acetic acid does not

react at the reaction conditions used.

◦

Sugars and levoglucosan do not undergo complete conversion

◦

at 125 C. Levoglucosan has a very slow rate of hydrogena-

◦

24

◦

tion at 125 C. Initially, levoglucosan concentration decreases

◦

-

1

-1

rapidly from 390.6 mmol-C L to about 275 mmol-C L at

1

6

◦

25 C. However, it then stays the same for 3 h. The levoglucosan

concentration does decrease when the temperature is increased

further. This implies that the high reaction temperature is

required for the hydrogenation of levoglucosan. Levoglucosan

is converted to sorbitol in two steps, hydrolysis of levoglucosan

21

to glucose, followed by hydrogenation of glucose to sorbitol.

The ﬁrst reaction is an acid catalyzed reaction, whereas the

second reaction is catalyzed by a hydrogenation catalyst, in this

case Ru/C. In the absence of any externally added acid in the

reaction mixture, it is possible that the ﬁrst reaction is catalyzed

by acids that are present in the aqueous fraction of bio-oil.

Disappearance of the sugars follow a similar but less obvious

3

the bio-oil as both monomeric and oligomeric compounds.

These phenolic compounds are present in signiﬁcantly higher

concentrations in the WIBO than in the water soluble bio-oil.

These high molecular weight oligomers (molecular weight up

to 5000) can form networks due to intermolecular interactions,

resulting in high viscosity of the bio-oil. A rheological study

of bio-oil obtained from the soft wood bark residue indicates

the existence of self-aggregating intermolecular interactions

forming long-range network structures. These interactions will

be stronger in the water insoluble bio-oil due to its concentrated

nature, explaining the very high viscosity observed for it.

Addition of methanol to bio-oil or WIBO results in disrupting

of the intermolecular network due to solvation, signiﬁcantly

reducing the viscosity.

◦

-1

trend as levoglucosan. At 125 C, sugars (377 mmol-C L

-

1

to 175 mmol-C L ) disappear twice as fast as levoglucosan

390 mmol-C L to 275 mmol-C L ) does in 3.5 hours. This

-

1

-1

(

implies that the ﬁrst step (levoglucosan to glucose) is the

slower one and hence the rate limiting step in the conversion

of levoglucosan to sorbitol. Increasing the acidity of the feed

could therefore help expedite the conversion of levoglucosan to

25

21

sorbitol.

A low hydrogenation temperature is preferred to minimize

carbon loss from liquid in the form of methane. About 25% of

◦

carbon is converted to methane at 175 C, whereas at 125 C,

only about 10% carbon is converted to methane. However, sugars

and levoglucosan are not completely converted to corresponding

3

.2 Low temperature hydrogenation of the aqueous fraction of

bio-oil

◦

The aqueous fraction of bio-oil contains thermally unstable

compounds (e.g. glucose, levoglucosan) that decompose when

heated to high temperature. These compounds must be con-

alcohols at 125 C for shorter reaction times (<3.5 h). Thus,

there requires some optimization of reaction temperature and

time. At high temperature, shorter reaction times (just sufﬁcient

to convert all the reactants to desired products) should be used

to keep carbon loss to methane low. At low temperature, longer

reaction time may be required to convert the reactants to respec-

tive alcohols. Hydrogen is consumed during the hydrogenation

step (see Table 3).

1

2,13

verted into thermally stable compounds prior to APP.

If

they are not converted to thermally stable compounds then they

can cause deactivation of the catalyst by coke formation. The

thermally unstable compounds can be converted to thermally

stable compounds by a low temperature aqueous phase hydro-

genation. In the low temperature hydrogenation, compounds

such as hydroxyacetaldehyde, hydroxyacetone, and furfural are

Hydrogen consumption is about 0.034 g/g water soluble bio-

◦

oil at 125 C and doubles for every 25 C rise in the temperature.

This journal is © The Royal Society of Chemistry 2009

Green Chem., 2009, 11, 1433–1445 | 1437

View Article Online

Table 3 Hydrogenation of the aqueous fraction of bio-oil with initial carbon concentration of 24 900 mg L^-(by TOC), catalyst: 3 gm of 5 wt%

1

Ru/C (wet basis), total P: 68.9 bar

◦

Temperature/ C

25

125

150

175

Minutes

0

30

60

150

210

270

320

360

400

455

545

Concentration/mmol-C L^-

1

Method

GC-FID Hydroxyacetone

GC-FID Hydroxyacetaldehyde

GC-FID Ethylene glycol

GC-FID Propylene glycol

135.5 55.7

28.1 0.0

0.0 90.6

0.0 25.4

0.0 13.3

0.0 9.9

0.0 24.7

30.8 28.8

377.4 242.5

0.0 0.0

390.6 274.8

182.2 126.1

24.5

0.0

172.3 212.2

113.1 159.0

36.6

12.5

31.7

0.0

232.6 N.A.

0.0 N.A.

0.0

216.9

164.1

55.5

14.9

29.4

0.0

174.7

56.6

276.7

167.1

0.0

198.9

154.9

51.8

13.5

30.5

0.0

140.4

89.8

235.8

186.0

0.0

191.6

163.0

59.1

11.4

30.3

0.0

204.6

172.5

63.0

13.3

30.7

0.0

222.6

186.6

64.3

12.0

30.8

0.0

80.4

186.7

150.7

206.8

0.0

210.9

182.7

61.1

15.0

29.8

0.0

67.3

204.7

132.0

210.2

0.0

197.3

171.0

57.9

18.0

26.9

0.0

47.5

213.8

123.6

202.6

1058.6

63.0

a

GC-FID Butanediols

52.7

15.7

30.4

0.0

b

GC-FID THFA

GC-FID g-Butyrolactone

c

GC-FID Guaiacols

0.0

HPLC

Sugars

Sorbitol

Levoglucosan

125.9

117.1

216.0

196.3

117.7

141.2

212.9

201.1

271.4 N.A.

192.6 170.9

GC-FID Acetic acid

Total mmol-C L^-

1

1144.5 891.8 1087.2 N. A.

55.2 50.5

100.0 85.1

1155.9 1101.7 1110.8 1157

63.8

87.3

1141.0 1114

64.6

85.1

%

H

C identiﬁed in liquid

C in liquid by TOC

59.8 N. A.

87.6 86.1

0.016 0.024

59.6

89.2

0.038

61.8

86.7 N.A.

0.048 0.061

61.7

63.8

84.2

0.091

81.0

0.120

2

Consumption (g/g WSBO)

0

0.007

0.034

0.070

a

b

c

1

,2 and 1,4-butanediol. Tetrahydrofurfuryl alcohol. Contains guaiacol and methyl guaiacol, N.A. = not analyzed.

Hydrogen consumption at the end of the run was found to be

.12 g/g water soluble bio-oil. Hydrogen consumption should

0

be minimized by using as low temperature as possible. At low

temperature unwanted reactions, such as saturation of aromatic

rings, which consume a considerable amount of hydrogen, are

suppressed. The hydrogen consumption at low temperature

◦

(

125 C) is comparable to that reported by Baker and Elliott

for the two stage hydroprocessing process, where 0.034 gm H

consumed per gm of bio-oil.

2

is

8

3

.3 Aqueous-phase reforming of the aqueous fraction of Bio-oil

We produced hydrogen from the aqueous fraction of the bio-oil

by passing it over a 1 wt% Pt/Al catalyst as shown in Table 4

and Fig. 2. The Pt/Al catalyst had a hydrogen uptake of

0.6 mmol H g catalyst which corresponds to a surface to bulk

2

O

3

2

O

3

Fig. 2 Alkane distribution for aqueous-phase reforming of WSBO and

-

1

◦

4

5

wt% sorbitol solution at 265 C and 55.1 bar with 1 wt% Pt/Al

2

O

3

metal ratio of 79.2%. The concentration of the WSBO in water

was 4–5 wt% or 1.5–2 wt% carbon in water.

The concentration varied depending on the pretreatment of

the bio-oil. Prior to each run the exact carbon content of the

catalyst. Feed and reaction key (see Table 4) (ꢀ) WSBO-A, ( ) WSBO-

B, ( ), (ꢁ) Sorbitol, ( ) WSBO-C.

◦

Table 4 Aqueous-phase reforming of WSBO and 5 wt% sorbitol solution with a 1 wt% Pt/Al

2

O

3

catalyst. Reaction conditions: 265 C and 55.1 psi.

◦

The WSBO contains 1.5–2 wt% carbon and is prepared by hydrogenation of the aqueous fraction of bio-oil at 175 C and 68.9 bar with a 5 wt%

Ru/C catalyst for 3 h

Hours catalyst

Run code Feed pre-treatment WHSV/h^-¹^aon stream

Alkane

% Carbon in gas % Carbon in liquid

d

b

c

H

2

selectivity (%)

selectivity (%) phase efﬂuent

phase efﬂuent

WSBO-A Hydrogenation

WSBO-B Hydrogenation

WSBO-C None

0.73

0.13

0.73

75

20

10

34

60

50

36

64

21

35

30

16

21

41

32

44

73

43

22

51

e

f

Sorbitol

None

a

-1

b

WHSV = ﬂow rate of the aqueous fraction of bio-oil (g h ) divided by grams of catalyst in the reactor.

H

2

selectivity = (molecules H

2

produced/C

atoms in gas phase) ¥ (1/reforming ratio) ¥ 100, reforming ratio (RR) is the ratio of molecules of H

that all the water soluble bio-oil goes to H

(

H

2

produced to C atoms in the gas phase assuming

c

2

and CO

2

only, for hydrogenated water soluble bio-oil RR is 3, RR for sorbitol is 13/6. Alkane selectivity =

d

total moles of carbon atoms in alkane products)/(total moles of carbon atoms in the feed) ¥ 100. Gas phase contains C1 to C6 alkanes, CO

2

and

e

f

2

. Decreases to 23% with the catalyst on stream for 22 h. Decreases to 38% with the catalyst on stream for 48 h.

1

438 | Green Chem., 2009, 11, 1433–1445

This journal is © The Royal Society of Chemistry 2009

View Article Online

◦

Table 5 Production of alkanes from WSBO and sorbitol with a 4 wt% Pt/SiO

2

–Al

2

O

3

catalyst. Reaction conditions: 260 C and 51.7 bar. The

◦

WSBO feeds contain 1.5–2 wt% carbon and are prepared by hydrogenation of the aqueous fraction of bio-oil at 175 C and 68.9 bar with a 5 wt%

Ru/C catalyst for 3 h.

Hours catalyst

on stream

Alkane

selectivity (%)

% Carbon in gas

phase efﬂuent

% Carbon in liquid

phase efﬂuent

a

-1

b

c

Run code

Feed

WHSV (h )

d

WSBO-D

WSBO-E

WSBO-D-H

WSBO-E-H

WSBO-HCl

WSBO-HCl-H

Sorbitol

WSBO

WSBO + H

WSBO + HCl

WSBO + HCl + H

Sorbitol

0.96

0.20

0.96

0.20

0.96

48

25

80

45

42

77

85

55–60

97

42

35

38

43

59

37

47

32

19

40–50

18

2

55

56

200

215

24

40–45

55–60

72

2

a

-1

b

WHSV = ﬂow rate of the aqueous fraction of bio-oil (g h ) divided by grams of catalyst in the reactor. Alkane selectivity = (total moles of carbon

c

d

atoms in alkane products)/(total moles of carbon atoms in the feed) ¥ 100. Gas phase contains C1 to C6 alkanes and CO

the catalyst on stream for 60 h.

2

.

Decreases to 30% with

feed was measured with TOC. The liquid phase carbon content

was measured by TOC analysis of the feed and liquid products.

non-hydrogenated water soluble bio-oil, demonstrating the

necessity of the hydrogenation step.

The gas phase contained C1 to C6 alkanes, CO

products tested.

2

and H for all

2

3.4 Liquid phase dehydration/hydrogenation of aqueous

fraction of bio-oil

The carbon conversion to gas phase products was just above

0% for the hydrogenated aqueous fraction of bio-oil (WSBO-

2

Alkanes can be produced from the aqueous fraction of the

bio-oil by aqueous-phase dehydration/hydrogenation with a

-

1

A) at WHSV of 0.73 h . The H

2

and alkane selectivities were

6

0% and 21% respectively. This indicates that at low conversion,

12

bi-functional catalyst. Platinum supported on silica-alumina

is the catalyst used in these studies. The reactions that occur

on Pt include hydrogenation; reforming and water–gas shift.

The acidic silica-alumina catalyzes dehydration reactions. The

high hydrogen selectivity can be obtained from the aqueous

fraction of bio-oil. At similar conditions, sorbitol had a carbon

conversion of 44% and H

respectively. Decreasing the WHSV to 0.13 h for the WSBO

WSBO-B) increases the gas phase conversion to 41%. At this

2

and alkane selectivity of 64% and 16%

-

1

data gathered with 4 wt% Pt/Al

2

O

3

-SiO catalyst is depicted

2

(

in Table 5 and in Fig. 3 and 4. This data was collected on the

hydrogenated WSBO with hydrogen as a co-feed and without

hydrogen as a co-feed. The concentration of WSBO in water

was 4–5 wt%, which is 1.5–2 wt% carbon in water.

higher conversion, the hydrogen selectivity decreases to 50%.

Thus, the hydrogen and alkane selectivities are a function of

the gas phase conversion. When WSBO-B data is compared to

sorbirtol (both have similar conversion), it can be concluded

that the aqueous fraction of the bio-oil is less active than

sorbitol. The alkane selectivity for the WSBO and sorbitol are

signiﬁcantly different, as shown in Fig. 2. The WSBO more

selectively produces the heavier alkanes as compared to the

sorbitol.

Our results are similar to that of Cortright et al. who produced

◦

hydrogen from 1 wt% sorbitol solution at 265 C and 55.1 bar

13

with a 3 wt% Pt/Al

2

O catalyst. Carbon conversion to gas

3

phase, CO selectivity and hydrogen selectivity of 90%, 68%

2

and 46% respectively, were reported. The difference between

our runs and that of Cortright et al. is that our catalyst had

-

1

signiﬁcantly fewer surface sites (40.6 mmol H g catalyst of

-

1

hydrogen uptake vs. 105 mmol g catalyst of CO uptake). Also

the sorbitol space velocity used by Cortright et al. is 0.008 g

sorbitol per gram catalyst as compared to our 0.0365 g sorbitol

per gram catalyst. To compare our catalyst to that of Cortright

et al., we fed our reactor with 5 wt% sorbitol in water with same

Fig.

3

Alkane distribution for the liquid phase dehydra-

◦

tion/hydrogenation of WSBO and 5 wt% sorbitol solution at 260

and 51.5 bar with 4 wt% Pt/SiO -Al

C

operating conditions. Conversion to gas phase, CO selectivity

2

O

3

catalyst. Feed and reaction

and hydrogen selectivity for sorbitol were 44%, 84% and 64%,

respectively. We see low activity for sorbitol due the low catalyst

concentration and less active catalyst we used.

key (see Table 5): (ꢀ) WSBO-D, ( ) WSBO-E, ( ) WSBO-HCl, (ꢁ)

sorbitol.

Reforming of a non-hydrogenated aqueous fraction of the

bio-oil (WSBO-C) was also studied at WHSV of 0.73 h . The

gas phase conversion decreased from 32% at 10 h to 23% at

The WSBO was tested at two different space velocities

WSBO-D and WSBO-E) without adding hydrogen as a feed.

-

1

(

The carbon conversion to gas phase products increased from

2

2 h. A hydrogen selectivity of only 36% was observed with

-1

3

5% to 40–50% when the WHSV decreased from 0.96 h to

This journal is © The Royal Society of Chemistry 2009

Green Chem., 2009, 11, 1433–1445 | 1439

View Article Online

of 1 wt%. The product was then subjected to the aqueous phase

dehydration/hydrogenation. The gas phase carbon conversion

did not increase when HCl was added. However, the alkanes

carbon selectivity increased with and without hydrogen being

co-fed to the reactor (see WSBO-HCl-H and WSBO-HCl).

2

Low methane selectivity and a fairly even distribution of C2

to C6 alkanes is observed in case of the acidiﬁed feed (see Fig. 3

and 4).

4

.0 Discussion

4

.1 Proposed reaction system

From the elemental analysis data, the non-hydrogenated WSBO

can be represented by a molecular formula C

2

5.4

H O2.05 with an ar-

bitrary molecular weight of about 62. The amount of hydrogen in

water soluble bio-oil increases during the hydrogenation step and

depends on the hydrogen consumption in this step. About 25% of

Fig.

4

Alkane distribution for the liquid phase dehydration/

◦

hydrogenation of WSBO at 260 C and 51.7 bar with 4 wt% Pt/SiO

2

-

,

Al

2

O

3

catalyst. Feed and reaction key (see Table 5): ( ) WSBO-D-H

2

the carbon is converted to CH in the hydrogenation step under

4

(

ꢀ) WSBO-E-H , (ꢁ) WSBO-HCl-H

2

.

current conditions. Clearly, reducing the amount of methane

formed in the hydrogenation step is of critical importance in

obtaining commercially relevant yields. Methane can form by

the hydrogenation of CO produced upon decarbonylation of

bio-oil compounds. A mole of water is also formed when a mole

-

1

0.2 h . The alkane carbon selectivity was 45% for WSBO-D

with hexane and pentane being the most abundant alkanes

see Fig. 3). A similar (42%) alkane selectivity is observed

(

at both the space velocities. However, the alkane distribution

shifts towards lower alkanes at higher conversions (see Fig. 3).

At similar operating conditions to the WSBO, sorbitol had a

carbon conversion of 72% with the alkane selectivity of 42%.

The sorbitol shows a similar distribution of C1-C6 alkanes

as the WSBO. For sorbitol (72% conversion) and WSBO-E

of CH is obtained from hydrogenation of CO. The hydrogen

4

consumption in this step is about 0.06 g/g WSBO for a reaction

◦

temperature and time of 175 C and 3 h, respectively. For 0.06 g

H consumed per gram of WSBO, the hydrogenated WSBO

2

can be represented by a molecular formula C1.5

molecular weight of 48.9 (see reaction 4).

6.1

H O1.55 with a

(

up to 50% converison), similar alkane selectivity (42%) is

observed. Bio-oil is clearly less active than pure sorbitol for

alkane production, but comparable alkane selectivites can be

achieved. Huber et al. reported 95% conversion with 60% alkane

selectivity for sorbitol with the same temperature, pressure,

catalyst concentration and a WHSV of 1.3 h . Our catalyst is

less active for dehydration/hydrogenation as compared to that

of Huber et al. As the time of on stream increases from 48 h to

C H

O

+1.87H → C H O +0.5CH +0.5H O

5.4 2.05 2 1.5 6.1 1.55 4 2

(4)

Non-hydrogenated WSBO WSBO (or hydrogenated WSBO)

2

Hydrogen is produced by reaction 5 from the bio-oil and

water by aqueous-phase reforming. The hydrogen can react

with the bio-oil to produce alkanes (pentane is shown as an

example), as shown in reaction 6. (No carbon-carbon bond

forming reactions are taking place in Reaction 6.) Addition of

reaction 5 and 6 so that the hydrogen produced in reaction 5 is

consumed in reaction 6, gives reaction 7. From reaction 5 it can

be seen that the theoretical highest yield of hydrogen is 0.184 g

hydrogen per g of hydrogenated WSBO or 0.145 g hydrogen

per g of non-hydrogenated WSBO. The net hydrogen production

accounting for hydrogen consumption in hydrogenation step is

-

1 12

6

0 h the conversion for WSBO-D decreases from 35% to 30%,

indicating catalyst deactivation.

Hydrogen was supplied externally to improve the carbon

12

selectivity towards alkanes (WSBO-D-H

The alkane selectivity increases when hydrogen is added from

2–45% to 77–85%. This is due to the suppression of hydrogen

2

and WSBO-E-H

2

).

4

producing reforming reactions in the presence of external

hydrogen. Catalyst deactivation can be the reason for the low

0

.085 g hydrogen per g non-hydrogenated WSBO. The selectivity

of alkanes produced from the water soluble bio-oil depends

upon its composition. However, if we assume that pentane is

the major product, the theoretical highest yield of pentane that

can be produced is 0.326 g per g of non-hydrogenated WSBO

conversion observed as the WSBO-D-H

2

data was gathered on

the same catalyst after the collection of WSBO-D data. At low

-

1

WHSV (0.2 h ), carbon conversion was 56% with 85% alkane

selectivity when the hydrogen was supplied externally (WSBO-

(

reaction 7).

E-H ). When hydrogen is supplied externally, very high hexane

2

and pentane selectivities are observed with almost complete

suppression of methane formation (Fig. 4). However the hexane

and pentane carbon selectivity does decrease with increasing

conversion (Fig. 4).

The alkane selectivity is a function of relative rates of

dehydration and hydrogenation reactions. Hydrochloric acid

was added to the reaction mixture to increase the rate of

dehydration reactions. HCl was added to the aqueous fraction

of bio-oil before the hydrogenation step to a HCl concentration

C

1.5

H

6.1

O

1.55 + 1.45 H

2

O → 1.5 CO

2

+ 4.5 H

2

(5)

(6)

(7)

C

1.5

H

6.1

O

1.55 + 0.3 H

2

→ 1.55 H O + (1.5/5) C

2

5

H

12

1

6 C1.5

H

6.1

O1.55 → 1.5 CO

2

+ 21.8 H

2

O + 4.5 C

5

H

12

In our experiments we were only able to obtain 0.031 g H

per g of non-hydrogenated WSBO. This is only about 21%

of the maximum possible yield (0.145 g). This yield could be

2

1

440 | Green Chem., 2009, 11, 1433–1445

View Article Online

increased by recycling the un-reacted product and decreasing

the amount of methane we produce in the low-temperature

hydrogenation step. Currently, our hydrogen production in the

reforming step is less than our hydrogen consumption in the

hydrogenation step. Better reforming catalysts are required to

increase the hydrogenproduction. If higher hydrogen production

is achieved, the excess hydrogen produced can be used for the

low temperature hydrogenation reaction and to hydrogenate

the WIBO. The water insoluble fraction of the bio-oil can be

hydrotreated with the hydrogen produced from the aqueous

fraction of the bio-oil to produce aromatic fuel. WIBO also has

without the addition of external water. The extent of separation

of bio-oil in the two phases can also depend on the method of

19

addition of water to it. The oak wood bio-oil used in our studies

phase separates upon addition of small amount of water. The

minimum water to bio-oil ratio required for phase separation

was 1 : 4 (weight ratio). The amount of bio-oil extracted in

water was independent of the amount of water added, as long

as the water to bio-oil ratio was above 1 : 4. This ensures that

concentrated aqueous solutions of water soluble bio-oil can be

used in our process. However, dilute WSBO solutions were used

in our studies primarily because of the lack of large amounts of

bio-oil.

6

the potential to replace phenol in phenol–formaldehyde resins.

4

.2 Proposed model bio-oil compounds for the aqueous phase of

4.4 Low temperature hydrogenation of the aqueous fraction of

bio-oil

Bio-oil is a complex mixture of several hundred compounds.

Hence a complete chemical analysis of bio-oil is very difﬁcult.

In our case we were able to identify only 60% of the carbon

in the aqueous fraction of the bio-oil using GC-MS and

HPLC. For catalyst and process development, it is important

to understand the chemistry involved in the hydrogenation and

aqueous phase processing of bio-oil. Various model compound

studies should be very helpful in this context. Based on our

Low temperature hydrogenation makes the aqueous fraction of

bio-oil amenable to further processing. Ruthenium on activated

carbon was selected, as this catalyst is known to exhibit high

activity and stability for similar hydrogenation reactions in the

27–29

aqueous phase.

The temperature and reaction time both

inﬂuence the products that are formed. Optimization of the

reaction temperature and time is required, as discussed in

Section 3.2. Strong acid (e.g. H

2

SO , HCl) can be added to the

4

1,3,4,7,20,24,26

own bio-oil analysis and the available literature,

the

reaction mixture to increase the rate of acid catalyzed hydrolysis

of levoglucosan to glucose. Helle et al. showed that increasing

the concentration of sulfuric acid increases the rate of hydrolysis

following model compounds are suggested for the aqueous

phase of bio-oils: levoglucosan, glucose, acetic acid, formic acid,

hydroxyacetaldehyde, hydroxyacetone, 2-furanone, furfural and

guaiacol. Table 6 shows the recommended composition of

a model aqueous-phase of bio-oil. Model compound studies

should help us understand how each compound behaves during

hydrogenation, reforming and APH/D steps. It should also help

to understand the interaction between different components in

the bio-oil.

◦

21

of levoglucosan in water at 110 C.

The hydrogen consumption in this step is up to 0.12 g/g

◦

WSBO at 175 C (Table 3). At high hydrogenation temperature,

large amounts of undesired methane is produced. This results in

a signiﬁcant increase in hydrogen consumption during this step.

Methane formation also results in reduced hydrogen production

in APR.

Based on GC-MS and HPLC analysis of the feed and product

of the hydrogenation reaction, the major reactions occurring

during this step are identiﬁed and are depicted in Table 7. The

last reaction in Table 7 is the hydrogenolysis reaction, wherein

a,w-diols are produced from the sugar alcohols such as sorbitol

in the presence of hydrogen. Ruthenium metal can catalyze the

4

.3 Extraction of bio-oil in water

Bio-oil phase separates readily into an aqueous and non-

aqueous phases upon addition of water. The aqueous fraction

of bio-oil is easy to handle and process. Water is an ideal solvent

for subsequent processing steps. It is inert for hydrogenation and

takes part in the liquid phase reactions producing hydrogen and

alkanes. The composition of the bio-oil, the storage conditions

and storage time are the factors affecting the bio-oil phase

separation. Some bio-oils can phase separate during storage

30–32

hydrogenolysis reactions.

Ruthenium can also catalyze the

C–C bond breaking reactions. The relative rates of hydrolysis

of C–O bonds and cleavage of C–C bonds decide the ﬁnal

product distribution. Apart from the reaction mentioned in

Table 7, decarbonylation and methanation reactions also happen

during this step. Another process option is to extract the polyols

from the hydrogenated aqueous-fraction of the bio-oil. The

three major products after hydrogenation are ethylene glycol,

propylene glycol and 1,4-butanediol. We produced 0.102 g

ethylene glycol, 0.064 g propylene glycol and 0.021 g 1,4-

butanediol per g of non-hydrogenated WSBO. The polyols can

be separated from the aqueous fraction either by distillation

or nanoﬁltration. The concentration of the aqueous phase will

decide the distillation efﬁciency. Nanoﬁltration is an attractive

option where molecules can be separated based on size and

functionality. The smaller polyols (C2 to C4) can be separated

from the hydrogenated aqueous fraction, and be used for the

production of hydrogen. Whereas the larger polyols (sugar

alcohols) can be used to produce alkanes.

Table 6 Recommended model bio-oil aqueous phase composition

Component

wt% range

Formic acid

Acetic acid

Hydroxyacetaldehyde

Hydroxyacetone

Furfural

4–8

8–12

5–10

8–12

2–5

2

-Furanone

2–5

Glucose

15–20

4–8

Levoglucosan

Guaiacol

Water

15–20

Green Chem., 2009, 11, 1433–1445 | 1441

View Article Online

Table 7 Reactions in hydrogenation of water soluble fraction of bio-oil

Reactant Product

Type of reaction

Hydrogenation

Hydroxyacetaldehyde

Ethylene glycol

Hydroxyacetone

Propylene glycol

Hydrogenation

Furfural

Tetrahydrofurfuryl alcohol

Hydrogenation

2

-Furanone

g-Butyrolactone

Guaiacol

1,2-Cyclohexanediol

Glucose

Hydrogenation

Hydrolysis

Levoglucosan

Glucose

Sorbitol

Hydrogenation

Hydrogenolysis

Sorbitol

Ethylene glycol, propylene glycol and 1,4-butanediol

4.5 Reforming of the aqueous fraction of bio-oil to produce

hydrogen

In this study it is shown that the aqueous-fraction of bio-oil can

be converted to hydrogen and alkanes. A process ﬂow diagram

that shows how bio-oil could be converted into liquid fuels and

chemicals is shown in Fig. 5. Fig. 5 shows mass balances for the

different reactions assuming 100% of the theoretical yield and

2

5% of methane is produced.

In the ﬁrst step in this process, the bio-oil is mixed with water

(

1 : 1 weight mixture) and separated into a water soluble fraction

WSBO) and a water insoluble bio-oil (WIBO). Approximately

6

2 wt% of the initial bio-oil goes into the WSBO and 38 wt%

of the initial bio-oil goes into the WIBO. The WSBO is

then sent to a low temperature hydrogenation unit where the

hydrogen consumption is up to 0.12 g/g bio-oil. In Fig. 5,

hydrogen consumption of 0.06 g/g non-hydrogenated WSBO

is shown. The purpose of this unit is to stabilize the bio-oils. The

hydrogenated WSBO is then sent to an aqueous-phase reformer

where hydrogen is produced. The maximum amount of hydrogen

Fig. 5 Process ﬂow diagram for aqueous phase reforming of bio-oil.

-

1

step. The net hydrogen production for this process is 5.3 kg h .

After the hydrogenation step, the polyols can be separated out

if desired. The unreacted bio-oil is re-concentrated and recycled

back to the reactor.

-

1

-1

that can be produced from 100 kg h of bio-oil is 9 kg h . In the

process, 66 kg h^-CO

1

and 8 kg h of methane will also form.

-1

2

-

1

The hydrogen consumption is 3.7 kg h in the hydrogenation

1

442 | Green Chem., 2009, 11, 1433–1445

View Article Online

In our experiments, hydrogen is produced from the aqueous

fraction of the bio-oil with high selectivity (50–60%). The major

challenge is that the bio-oils have low activity. This could be

due to poisoning of the catalyst surface by acids or other

impurities. Lower hydrogen selectivity is observed for bio-oil

(

50%) as compared to sorbitol (64%) at similar conversion.

Bio-oil exhibits low reforming activity, though with moderate

hydrogen selectivity. The moderate hydrogen selectivity can be

attributed to the presence of smaller oxygenated hydrocarbons

in the aqueous fraction of bio-oil. Smaller oxygenates present

in bio-oil are converted to smaller polyols such as ethylene

glycol and propylene glycol upon hydrogenation. These polyols

can produce hydrogen in high yield as compared to larger

polyols, such as sorbitol. These polyols also have higher rates

Fig.

6 Process ﬂow diagram for the aqueous phase dehydra-

13,15

tion/hydrogenation of bio-oil.

of reaction than sorbitol.

Even though it contains smaller

polyols, the aqueous fraction shows lower hydrogen selectivity as

compared to sorbitol. This can be due the presence of unreacted

sugars and other unidentiﬁed compounds in the aqueous

fraction.

Conversion for bio-oil is almost half of that for sorbitol. From

Fig. 6, it can be seen that 0.289 to 0.328 g alkanes can be

produced per gram of non-hydrogenated WSBO. We were able

to produce up to 0.139 g alkanes per gram of non-hydrogenated

WSBO, which is 42–48% of the theoretical yield. The alkane

selectivity and the heavier alkane distributions are better for the

bio-oil than the sorbitol. The alkane selectivity can be altered

by supplying hydrogen externally or by adding a strong acid to

the feed. External hydrogen suppresses the C–C bond breaking

Theoretically, about 0.085 g of hydrogen can be produced

from a gram of non-hydrogenated WSBO when about 25% of

the carbon is converted to methane in the hydrogenation step.

We were able to produce 0.031 g of hydrogen from a gram of

non-hydrogenated WSBO, while consuming 0.06 g hydrogen in

the hydrogenation step. The low hydrogen production is due

to two reasons (a) high carbon loss and hence the hydrogen

loss to gas phase during hydrogenation and (b) low activity

of bio-oil in reforming. About 0.032 g hydrogen per g non-

hydrogenated WSBO is lost in the form of methane during

reactions that produce CO and CH as the by-products. The near

4

absence of methane when hydrogen is supplied externally implies

that the methane is a result of C–C bond breaking reactions

and not of dehydration/hydrogenation reactions of bio-oil

components containing a single carbon atom (e.g. methanol).

The strong acid can act as a dehydration catalyst and hence can

expedite the dehydration reactions resulting in higher alkanes

selectivity.

◦

hydrogenation at 175 C. To reduce the carbon loss, optimization

of reaction temperature and time is necessary, as discussed

earlier. A catalyst with low decarbonylation activity is desired

for the hydrogenation step. For reforming, a catalyst with high

activity for bio-oil is desired. The low reforming activity seen

The theoretical selectivities for the individual alkanes from

WSBO and hydrogenated WSBO can be calculated from the

composition shown in Table 3. These theoretical selectivities

assume that no carbon–carbon bond cleavage occurs in aqueous-

phase dehydration hydrogenation. This provides an upper limit

for alkane production by APD/H. In Fig. 7, the theoretical

selectivities are compared with the actual selectivities observed

for WSBO-E-H2. Theoretically, no methane should be produced

from the hydrogenated WSBO but in actuality up to 16%

methane selectivity is observed for hydrogenated WSBO (see

Fig. 3). This implies the presence of C–C bond cleaving

reactions during APD/H. The actual ethane selectivity is

lower than the theoretical value, implying that C2 compounds

are involved in the production of methane. Propane, butane

and pentane selectivities are higher than what is expected

theoretically, implying that other than the respective C3, C4 and

C5 compounds, these alkanes are also produced from the C6

compounds.

for Pt/Al

2

3

O catalyst can also be due to various impurities in

the bio-oil. Effects of the bio-oil impurities (e.g. alkali salts of

organic acids) on the reforming process needs to be studied. Bi-

metallic catalysts are known to be highly active for the reforming

15

of ethylene glycol and can potentially be used for reforming of

aqueous fraction of bio-oil.

4.6 Liquid phase dehydration/hydrogenation of aqueous

fraction of bio-oil

The process ﬂow diagram for the production of alkanes from the

aqueous fraction of the bio-oil is shown in Fig. 6. The amount

of alkanes produced is shown with 100% carbon conversion to

alkanes. The process is essentially the same as that for APR,

except that the APR reactor is replaced by the APD/H reactor.

Theoretically, when hydrogen is not supplied externally, about

-

1

8.0 to 20.4 kg h of C1 to C6 alkanes mixture can be produced

-

1

-1

from 100 kg h of bio-oil (62.2 kg h non-hydrogenated WSBO).

At the same time, 3.5 to 16.0 kg h of CO

In this study, only smaller alkanes (C1 to C6) are produced

from the aqueous fraction of the bio-oil. These alkanes are

less valuable as fuel. The aqueous fraction of the bio-oil can

essentially be used to produce larger gasoline range alkanes

by carrying out some C–C bond formation reactions (e.g.

-

1

-1

2

and 8 kg h of

CH will be produced. The composition of the alkane mixture

4

produced depends on the composition of WSBO. Hydrogen can

be supplied externally to the APD/H reactor to increase the

carbon selectivity towards alkanes.

33,34

aldol condensation) within it, followed by APD/H.

Presence

In our experiments, alkanes are produced at modest yields

from the aqueous fraction of bio-oil over a bi-functional catalyst.

of various aldehydes and ketones makes the bio-oil aqueous

fraction an excellent candidate for such reactions.

Green Chem., 2009, 11, 1433–1445 | 1443

View Article Online

Ca and Mg salts in the ash can end up in the aqueous fraction

and can de-activate the APP catalyst.

5

.0 Conclusion

The general conclusion from this study is that hydrogen and

alkanes can be produced from the aqueous fraction of bio-oil

by aqueous phase processing. Thus, previous work on aqueous

12,13,33

phase processing of model biomass compounds

can be

applied to processing of feedstocks derived from lignocellulosic

biomass. This thus offers a new concept for the conversion of

lignocellulosic biomass into fuels and chemicals.

The ﬁrst step in this process is to add water to the bio-oil

and separate it into aqueous and organic phases. The aqueous

phase is then sent to a low temperature hydrogenation unit where

thermally unstable functionalities are hydrogenated to thermally

stable compounds. In this hydrogenation step aldehydes are

converted to alcohols, sugars to sugar alcohols, and aromatics

are hydrogenated. Undesired methane is also formed from this

reaction. A key need in this low temperature step is to reduce

the amount of hydrogen that is consumed.

Fig. 7 Theoretical and actual alkane selectivities for the production of

alkanes from the aqueous fraction of the bio-oil. Legend (refer Table 5):

(

ꢀ) non-hydrogenated WSBO – theoretical, ( ) hydrogenated WSBO –

theoretical, (ꢁ) WSBO-E-H

2

.

Hydrogen is produced at a hydrogen selectivity of 60% from

the water soluble part of bio-oil. This selectivity is comparable

to that observed for pure sorbitol at similar conversions.

This makes bio-oil a feasible feedstock for the production of

hydrogen. We have demonstrated here the feasibility to produce

hydrogen from the water soluble bio-oil, but future research is

needed to determine how the carbon conversion to gas phase can

be increased maintaining the high selectivity. Concerns about

catalyst stability must also be answered to efﬁciently use aqueous

phase processing.

Alkanes are produced from the water soluble bio-oil by

aqueous-phase dehydration/hydrogenation with a bifunctional

catalyst. An alkane selectivity of 77% is obtained when hydrogen

is co-fed with the bio-oil. Alternatively, an alkane selectivity of

45% is obtained when hydrogen is produced in situ from the

4

.7 Comparison of aqueous-phase processing of bio-oil to

hydrotreating of bio-oil

To date, all the efforts to upgrade bio-oil have revolved around

hydrotreating and zeolite upgrading. In hydrotreating, oxygen

is removed in the form of water in the presence of a catalyst

at high temperature and high hydrogen pressure. Baker and

Elliott developed a two step process in which bio-oil is subjected

9,10

◦

to mild hydrotreating at 300 C and 137.8 bar (total P) to

make a stabilized product, followed by further hydrocracking

◦

8

at 350 C and 137.8 bar. A relatively low LHSV of 0.07

volume of oil per volume of catalyst per h is used in the

second step. Sulﬁded Co-Mo based catalyst was used in both

the steps. About 75% carbon is converted to the oil phase

containing 2.3 wt% oxygen. Hydrogen consumption is about

0

.034 g/g of bio-oil. The advantage of this bio-oil upgrading

bio-oil itself. The C

5

plus C carbon selectivity ranges from 35

6

approach is that all the reactions appear in one single step. The

advantage of our approach is that we process the aqueous phase

of the bio-oil differently than the organic phase. The aqueous

phase is signiﬁcantly different in composition than the organic

phase. This allows us to design catalysts that are well suited for

conversion of both the aqueous and organic phases. We believe

that in the future this will allow us to achieve higher overall

yields for conversion of bio-oils into liquid fuels and chemicals.

It would be advantageous to separate smaller molecules from the

bio-oil prior to hydrotreating, especially since these molecules

form lighter gases during hydrotreating. These smaller molecules

could then be sold as chemicals or reformed to hydrogen. Bio-oil

composition changes depending on the feedstock. We envision

that our process can handle any type of bio-oil. All the small

aldehyde and ketonic functionalities in the aqueous fraction will

be converted to corresponding polyols. These small polyols are

excellent feedstock for the production of hydrogen and alkanes.

The challenge would be converting all types of sugars and

anhydrosugars to corresponding alcohols. Strong acid may need

to be added in the hydrogenation step to facilitate hydrolysis

of anhydrosugars to corresponding sugars. Another challenge

would be processing bio-oils with high ash content. The Na, K,

to 76% depending on the process conditions.

For these reactions to become an industrial process the overall

yield must be increased. Catalyst deactivation problems must

be overcome and a more active catalyst must be developed.

However, in spite of these current difﬁculties it is highly likely

that more efﬁcient catalytic processes will be developed for

the conversion of pyrolysis oils into fuels and chemicals as we

continue to understand the chemistry of this process and develop

new generations of catalysts.

Acknowledgements

The authors gratefully acknowledge Phil Badger of Renewable

Oil International LLC for providing the bio-oil samples. We also

thank Hakan Olcay for help with chemisorption measurements.

This work was supported by DOE-STTR and NSF-MRI

(

CBET-0722802) grants.

References

1

D. L. Klass, Biomass for renewable energy, fuels and chemicals,

Academic Press, San Diego, 1998.

2 A. V. Bridgwater, Chem. Eng. J., 2003, 91, 87–102.

1

444 | Green Chem., 2009, 11, 1433–1445

View Article Online

3

4

5

D. Mohan, C. U. Pittman and P. H. Steele, Energy Fuels, 2006, 20,

48–889.

G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006, 106, 4044–

098.

M. M. Wright and R. C. Brown, Biofuels, Bioproducts and Bioreﬁning,

007, 1, 49–56.

S. Czernik and A. V. Bridgwater, Energy Fuels, 2004, 18, 590–598.

T. A. Milne, F. Agblevor, M. Davis, S. Deutch, and D. Johnson,

in Thermal Biomass Conversion, ed. A. V. Bridgwater, and

D. G. B. Boocock, Blackie Academic and Professional, London,

UK, 1997, pp. 409–424.

18 A. Oasmaa and S. Czernik, Energy Fuels, 1999, 13, 914–921.

19 K. Sipil a¨ , E. Kuoppala, L. Fagernas and A. Oasmaa, Biomass

Bioenergy, 1998, 14, 103–113.

20 J. Piskorz, D. S. Scott, and D. Radlien, in Pyrolysis Oils from

Biomass: Producing Analyzing and Upgrading, ed. E. J. Soltes, and

T. A. Milne, American Chemical Society, Washington DC, 1988, pp.

167-178.

21 S. Helle, N. M. Bennett, K. Lau, J. H. Matsui and S. J. B. Duff,

Carbohydr. Res., 2007, 342, 2365–2370.

8

4

2

6

7

22 Z. Y. Luo, S. R. Wang, Y. F. Liao, J. S. Zhou, Y. L. Gu and K. F. Cen,

Biomass Bioenergy, 2004, 26, 455–462.

8

9

E. G. Baker, and D. Elliott, in Thermochemical Biomass Conversion,

ed. A. V. Bridgwater, and J. L. Kuester, Elsevier Science Publishers

Ltd, Barking, England, 1988, pp. 883–895.

R. Maggi, and B. Delmon, in Hydrotreatment and Hydroprocessing of

Oil Fractions, ed. G. F. Froment, B. Delmon and P. Grange, Elsevier

Science B. V., 1997.

23 J. P. Diebold, A Review of the Chemical and Physical Mechanisms of

the Storage Stability of Fast Pyrolysis Bio-Oils, National Renewable

Energy Laboratory Report NREL/SR-570–27613, Golden, CO,

2000.

24 A. V. Bridgwater, D. Meier and D. Radlein, Org. Geochem., 1999, 30,

1479–1493.

1

0 D. C. Elliott, Energy Fuels, 2007, 21, 1792–1815.

25 M. Garc `ı a-P e´ rez, A. Chaala, H. Pakdel, D. Kretschmer, D. Rodrigue

and C. Roy, Energy Fuels, 2006, 20, 364–375.

26 R. J. Evans and T. A. Milne, Energy Fuels, 1987, 1, 123–138.

27 J. Wisnlak and R. Simon, Ind. Eng. Chem. Prod. Res. Dev., 1979, 18,

50–57.

28 P. Gallezot, N. Nicolaus, G. Fl e` che, P. Fuertes and A. Perrard,

J. Catal., 1998, 180, 51–55.

29 Z. Zhang, J. E. Jackson and D. J. Miller, Appl. Catal., A, 2001, 219,

89–98.

1 R. K. Sharma and N. N. Bakhshi, Energy Fuels, 1993, 7, 306–314.

2 G. W. Huber, R. D. Cortright and J. A. Dumesic, Angew. Chem., Int.

Ed., 2004, 43, 1549–1551.

1

3 R. D. Cortright, R. R. Davda and J. A. Dumesic, Nature, 2002, 418,

9

64–967.

4 T. A. Milne, C. C. Elam, and R. J. Evans, Hydrogen from Biomass:

State of the Art and Research Challenges, International Energy

Agency Report IEA/H2/TR-02/001, 2002.

1

5 G. W. Huber, J. W. Shabaker, S. T. Evans and J. A. Dumesic, Appl.

30 I. T. Clark, Ind. Eng. Chem., 1958, 50, 1125–1126.

31 D. K. Sohounloue, C. Montassier and J. Barbier, React. Kinet. Catal.

Lett., 1983, 22, 391–397.

32 M. Schlaf, Dalton Trans., 2006, 4645–4653.

33 G. W. Huber, J. N. Chheda, C. J. Barrett and J. A. Dumesic, Science,

2005, 308, 1446–1450.

Catal., B, 2006, 62, 226–235.

6 P. Badger, Method or System for Accomplishing Flash or Fast

Pyrolysis with Carbonaceous Materials, US Pat., Application No.

1

1/480 914.

1

7 P. Badger, and P. Fransham, Improved Methods of Accomplishing

Flash or Fast Pyrolysis with Carbonaceous Materials, US Pat.,

Application No. 11/480 915.

34 E. L. Kunkes, D. A. Simonetti, R. M. West, J. C. Serrano-Ruiz, C. A.

G a¨ rtner and J. A. Dumesic, Science, 2008, 322, 417–421.

Green Chem., 2009, 11, 1433–1445 | 1445

Article Doi

Production of hydrogen, alkanes and polyols by aqueous phase processing of wood-derived pyrolysis oils

DOI: 10.1039/b912522c

Source and publish data:

Authors:

Article abstract of DOI:10.1039/b912522c

Full text of DOI:10.1039/b912522c

Products guided by the article

R&D Labs maybe for 488-44-8

Relevant to this article

Hot Product