Historical developments in hydroprocessing bio oils

This paper provides only an overview of the types of technology for liquid fuels production from biomass, such as hydrothermal processing or fast pyrolysis.. Hydrothermal Liquefaction Pr

1792 Energy & Fuels 2007, 21, 1792-1815

Historical Developments in Hydroprocessing Bio-oils

Douglas C Elliott*

Pacific Northwest National Laboratory, P.O Box 999, 902 Battelle BouleVard,

Richland, Washington 99352 ReceiVed January 25, 2007 ReVised Manuscript ReceiVed March 16, 2007

This paper is a review of the developments in the field of catalytic hydroprocessing of biomass-derived

liquefaction conversion products (bio-oil) over the past 25 years Work has been underway, primarily in the

U.S and Europe, in catalytic hydrotreating and hydrocracking of bio-oil in both batch-fed and

continuous-flow bench-scale reactor systems A range of heterogeneous catalyst materials have been tested, including

conventional sulfided catalysts developed for petroleum hydroprocessing and precious metal catalysts The

important processing differences have been identified, which required adjustments to conventional

hydropro-cessing as applied to petroleum feedstocks This application of hydroprohydropro-cessing is seen as an extension of

petroleum processing and system requirements are not far outside the range of conventional hydroprocessing

The technology is still under development but can play a significant role in supplementing increasingly

expensive petroleum

Introduction

The objective of this review was to gather the relevant

information from the literature (which is in specific cases more

or less accessible from earlier times) to describe the

development of hydrotreating upgrading technology to produce

transportation liquid fuels from thermochemically derived

biomass liquids This paper provides only an overview of the

types of technology for liquid fuels production from biomass,

such as hydrothermal processing or fast pyrolysis The

important consideration is that these liquid products from

biomass are not useful as fuels other than direct boiler firing

and possibly for some types of turbine and large diesel

applications after significant modifications In order for the

biomass liquids to be useful as transportation fuels, they require

chemical transformation to increase volatility and thermal

stability and reduce viscosity through oxygen removal and

molecular weight reduction The reader may refer to

Bridgwater et al for a discussion of how these types of

technologies relate to each other.1 Maggi and Delmon provide

a detailed comparison of fast and slow pyrolysis.2 The field of

hydrotreating of hydrothermal (high-pressure) liquefaction oils

from biomass provides the initial basis for this review This

study then focuses on the hydroprocessing of bio-oil from fast

pyrolysis to liquid fuels As such, it does not provide details of

the liquefaction processes themselves but will focus on the

catalytic processes used to convert the highly oxygenated

bio-oil to a liquid more similar to petroleum-derived fuels

This review included collection of data from the literature in

the field of study highlighting developments from the 1980s to

the present A useful summary of the early work has been

published.3 This review is a collection and summary of the

actual

* E-mail address: dougc.elliott@pnl.gov.

(1) Bridgwater, A V.; Meier, D.; Radlein, D An Overview of Fast

Pyrolysis of Biomass Org Geochem 1999, 30, 1479-1493.

(2) Maggi, R.; Delmon, B Comparison between “Slow” and “Flash”

Pyrolysis Oils from Biomass Fuel 1994, 73 (5), 671-677.

(3) Elliott, D C.; Beckman, D.; Bridgwater, A V.; Diebold, J P.; Gevert,

S B.; Solantausta, Y Developments in Direct Thermochemical Liquefaction

of Biomass: 1983-1990 Energy Fuels 1991, 5 (3), 399-410.

data from the experimentation of handling and upgrading bio- oils by catalytic hydrotreating

Biomass Liquefaction Processes

By way of introduction, the thermochemical conversionmethods for producing oil from biomass are reviewed Thereare basically two methodsshigh-pressure liquefaction (typicallyhydrothermal for biomass) and atmospheric-pressure fast py-rolysis

Hydrothermal Processes As early as the 1920s,

experi-mental data from Berl supported the concept of oil productionfrom biomass in hot water using alkali as catalyst.4 Heinemanncontinued this work into the 1950s.5 Following the Arab oilembargo of 1974, initial efforts in biomass liquefaction in theU.S focused on the high-pressure (hydrothermal) liquefactionprocess based on this concept developed at the PittsburghEnergy Research Center (PERC) and demonstrated at theAlbany Biomass Liquefaction Experimental Facility at Albany,Oregon,6 at a scale of 100 kg/h In this process, wood powderwas slurried in recycle oil product and mixed with water and asodium carbonate catalyst The slurry was pumped to apressurized reactor where carbon monoxide was added at 20.8MPa The slurry was held at a temperature of around 350 Cfor 20 min to 1 h after which it was depressurized, and oilproduct was separated from water This heavy oil product had amelting point near room temperature with an oxygen content of

12 to 14% and dissolved water of 3 to 5%

A later derivative of this process was the HTU process fromShell.7 In this process, similar conditions were used but the

(7) Goudriaan, F.; van de Beld, B.; Boerefijn, F R.; Bos, G M.; Naber,

J E.; van der Wal, S.; Zeevalkink, J A Thermal Efficiency of the HTU

Process for Biomass Liquefaction In Progress in Thermochemical Biomass ConVersion; Bridgwater, A.V., Ed.; Blackwell Science, LTD.: Oxford,

England, 2001; pp 1312-1325.

10.1021/ef070044u CCC: $37.00 © 2007 American Chemical Society

Published on Web 05/02/2007

DeVelopments in Hydroprocessing Bio-oils Energy & Fuels, Vol 21, No 3, 2007 1

reducing agent alkali/CO was left out; the resulting product had

an oxygen content nearer to 18% with a melting point of about

80 C.8 This process has been scaled up to a 10 kg/h dry

biomass feed pilot plant The pilot plant was operated for a 500

h design run in 2004

Fast Pyrolysis Processes Most of the work up to the year

2000 is covered in detail in a review article by Bridgwater and

Peacocke,9 but some of the highlights are described below

Modern developments in fast pyrolysis can be traced to the

Occidental/Garrett process development work in the 1970s That

effort focused on waste processing, but a detailed report exists

which describes results with biomass materials.10 The University

of Waterloo first published results in a fluidized bed for fast

pyrolysis of biomass in a government report in 1981 After

extensive research, the technology was scaled up in Spain

(Union Fenosa) and is now a patented technology held by

Dynamotive Energy Systems Corporation A company which

grew out of the University of Waterloo, Resource Transforms

International Ltd., is still developing refinements and chemical

products from fast pyrolysis

A second version of fast pyrolysis which utilizes circulating

fluidized beds was developed out of the University of Western

Ontario and is now commercialized by Ensyn Technologies

(RTP, rapid thermal processing) Their technology is used in

food flavoring production plants in the U.S It was scaled up

for further development in Italy in the 1990s but was not

operated extensively

Other versions of fast pyrolysis include the ablative reactor

first demonstrated in its vortex form by the National Renewable

Energy Laboratory (formerly the Solar Energy Research

Insti-tute) and now being developed at Aston University and Twente

University with rotating plate and cone reactors The Twente

version has been scaled up in Malaysia The simple entrained

flow reactor was attempted at the Georgia Tech Research

Institute in the 1980s as an extrapolation from the Tech-Air

upflow pyrolysis/gasification technology The operations could

never achieve as high oil yields as those found in the other

reactors, apparently because of the limited heat transfer

Vacuum pyrolysis was developed to demonstration scale at

Laval University but was not found to be economical with

biomass

Bio-oil Hydroprocessing

Upgrading biomass-derived oils to hydrocarbon fuels requires

oxygen removal and molecular weight reduction As a result,

there is typically a formation of an oil phase product and a

separate aqueous phase product by hydroprocessing To

mini-mize hydrogen consumption in hydroprocessing,

hydrodeoxy-genation (HDO) must be emphasized, without saturation of the

aromatic rings Hydroprocessing of biomass-derived oils differs

from processing petroleum or coal liquids because of the

importance of deoxygenation, as opposed to nitrogen or sulfur

removal At the time that research began to evaluate the HDO

of biomass-derived oils, HDO had received only limited

attention in the literature.11 Over the past 20 plus years, there

(8) Goudriaan, F.; Zeevalkink, J A.; Naber, J E HTU Process Design

and Development: Innovation Involves Many Disciplines In Science in

Thermal and Chemical Biomass ConVersion; Bridgwater, A V., Boocock,

D G B., Eds.; CPL Press: Newbury Berks, U.K., 2006; pp 1069-1081.

(9) Bridgwater, A V.; Peacocke, G V C Fast Pyrolysis Processes for

Biomass Renewable Sustainable Energy ReV 2000, 4, 1-73.

(10) Boucher, F B.; Knell, E W.; Preston, G T.; Mallan, G M Pyrolysis

of Industrial Wastes for Oil and ActiVated Carbon RecoVery; report no.

EPA-600/2-77-091, project no S-801202, U.S EPA: Washington, D.C.,

May 1977.

(11) Furimsky, E Catal ReV.sSci Eng 1983, 25 (3), 42.

have been wide-ranging efforts reported in the literature, as arecent review describes.12 A large portion of the body of workaddresses the catalytic chemistry by hydrotreating modelcompounds containing oxygen Many of these models arerelevant to bio-oil hydroprocessing, such as phenolics andaromatic ethers This report provides a detailed review of thoseresearch efforts that focused on processing actual bio-oilproducts, and model compound studies are addressed only whenperformed as a part of the bio-oil upgrading research effort

Hydrothermal Liquefaction Product Upgrading

As the first wood liquefaction oil produced at pilot scalebecame available in 1979, oil upgrading process research began.The catalytic hydrotreating of the heavy oil product followedthe methods used in conventional petroleum processing technol-ogy

Pacific Northwest National Laboratory (PNL/PNNL)

Bio-oil upgrading work at PNNL has focused on heterogeneouscatalytic hydroprocessing Initial work involved batch reactortests of model phenolic compounds13 with various catalysts.Commercial samples of catalysts were used representingCoMo, NiMo, NiW, Ni, Co, Pd, and CuCrO to hydrogenatephenol at

300 or 400 C (1 h at temperature) P-Cresol, ethyl-phenol,

dimethyl-phenol, trimethyl-phenol, naphthol, and guaiacol oxy-phenol) were also tested with a CoMo catalyst at 400 C

(meth-Of the catalysts tested, the sulfided form of CoMo was mostactive, producing a product containing 33.8% benzene and 3.6%cyclohexane at 400 C The Ni catalyst was also active,producing a product with 16.9% benzene and 7.6%cyclohexane The sulfided Ni catalyst still produced 8.0%cyclohexane, but its yield of benzene dropped to near zero(0.4%) A Pd catalyst produced a 7.8% benzene product with2.7% cyclohexane but

5.5% cyclohexanone At lower temperature (300 C), hexanone was the primary product at 8.1% and benzene andcyclohexane were nearly equal at 2.0 and 2.5%, respectively.The original work with hydrotreating biomass-derived liquidswas the effort to make gasoline from the high-pressure liquefac-tion oil produced from wood at the Albany Biomass Liquefac-tion Pilot Plant The oil from high-pressure liquefaction (ahydrothermal, alkali-catalyzed process) was a more deoxygen-ated product than fast pyrolysis bio-oil As a result, it was morethermally stable and required less hydrogenation to produce agasoline product Processing technology was adapted fromconventional petroleum hydrotreating using nickel and sulfidedcobalt-molybdenum catalysts14 in a continuous-flow, fixedcatalyst bed reactor system operated in an upflow configuration.The catalysts used in these tests are identified and described

cyclo-in Table 1

Preliminary results with a light oil fraction showed that thesulfided form of the CoMo catalyst was much more active thanthe oxide form A copper-chromite catalyst was also testedand found to be even less active The results are presented inTable 2 for extinction hydrotreating/hydrocracking of wholewood oil and a distillate product The whole oil could behydrotreated much like the distillate oil but required a lowerspace velocity and operating pressure and resulted in higherhydrogen consumption The nickel catalyst exhibited activity

(12) Furimsky, E Catalytic Hydrodeoxygenation Appl Catal., A: Gen.

2000, 199, 147-190.

(13) Elliott, D C Hydrodeoxygenation of Phenolic Components of

Wood-Derived Oil Prepr Pap.sAm Chem Soc., DiV Pet Chem 1983,

28 (3), 667-674.

(14) Elliott, D C.; Baker, E G Upgrading Biomass Liquefaction

Products through Hydrodeoxygenation Biotechnol Bioeng Symp 1984,

2 Energy & Fuels, Vol 21, No 3, 2007 Elliott

14, 159-174.

C 1.65 1.5ox

yg 0.0 0.8de

nsi 0.84 0.91 H/C

atom 1.72 1. 1.8

C -

> 3 7

Table 1 Catalysts Used in Hydrotreating/Hydrocracking Tests

Table 2 Hydrotreating Hydrothermal Products

Experimental Operating Conditions

catalyst CoMo 0402/S HT 400/S Ni-1404

feedstock distillate whole oil whole oil

(LHSV), vol oil/(vol cat h) 0.38 0.24 0.30 total oil product, L/L feedaqueous phase, L/L feed 0.990.20 0.92 0.20

a The distillate oil test was of limited duration, and the carbon on catalyst

percentage was proportionally higher at 19%; but, the actual carbon weight

percentage was similar to the other tests at 12%, suggesting that the carbon

laydown occurs early in the run and reaches a steady state.

while the TR12 contained more multiring phenolics Undersimilar processing conditions as shown in Table 3, a lighterhydrocarbon product was produced from the TR7 oil A needfor additional hydrocracking function in the catalyst wasindicated for use with TR12 The relationships of space velocity

to deoxygenation and gasoline yield when using the HT400-Scatalyst are shown in Figures 1 and 2 for the two high-pressure

16

similar to the sulfided CoMo catalyst except for having a much

higher gas yield and requiring hydrogen consumption The Ni

catalyst seemed to lose activity over the several hours of

operation as shown by the two product analyses in Table 2

Distillations of these products showed that they were mostly

gasoline range material (50-225 C BP) GC-MS analysis

showed that the cyclic ketones and phenolics in the liquefaction

product were converted to cyclic alkanes and aromatics

In further studies,15 comparisons were made between the

whole oil presented above (TR7 water slurry, single pass) and

an alternative form (TR12 produced with oil recycle) By

GC-MS analysis of the two feed oils, it was found that the TR7 oil

contained primarily cyclic ketones and single-ring phenolics,

liquefaction bio-oils

Catalyst stability was also evaluated and neither sulfur lossnor carbon buildup was judged to be the cause The alkalicontent of the TR12 oil and its deposition on the catalyst wasconcluded to be the cause of catalyst deactivation by a poreplugging mechanism over a 48 h test The resulting loss ofactivity in the mixed Haldor Topsoe catalyst system is shown

in Figure 3

Extensive studies with several catalysts provided a useful set

of results for comparing residual oxygen content and hydrogen

to carbon ratio as well as gasoline yield as a function of spacevelocity16 (see Figures 4 and 5)

It was concluded that high yields of high-quality gasolinecan be produced from biomass-derived oils; however, low space(15) Baker, E G.; Elliott, D C Catalytic Upgrading of Biomass

Pyrolysis Oils In Research in Thermochemical Biomass ConVersion;

Bridgwater, A V., Kuester, J L., Eds.; Elsevier Science Publishers, LTD.:

Barking, England, 1988; pp 883-895.

(16) Baker, E G.; Elliott, D C Catalytic Hydrotreating of

Biomass-Derived Oils In Pyrolysis Oils from Biomass: Producing, Analyzing and Upgrading; ACS Symposium Series 376; Soltes, E J., Milne, T A., Eds.;

American Chemical Society: Washington, D.C., 1988; pp 228-240.

Figure 1 Oxygen content of hydrotreated oils (400 C, 13.8 MPa).

velocities are required Cracking and hydrogenation of the

higher molecular weight components are the rate-limiting steps

in upgrading biomass-derived oils Future catalyst

development should be directed at these reactions

An advanced process concept was also tested wherein the

initial hydrotreating for oxygen reduction was followed by

separation of the aqueous, gasoline, and gas phases and a

sec-ond processing step of hydrocracking the heavy components.17

By performing the multistep process, it was possible to

maximize the yield of aromatic gasoline by removing a tion from the system before the rings became saturated.Furthermore, hydrogen consumption was minimized by using

frac-it only to hydrodeoxygenate the oxygen-containing nents and it was not wasted on the saturation of aromaticcompounds

compo-(17) Baker, E G.; Elliott, D C Method of Upgrading Oils Containing Hydroxyaromatic Hydrocarbon Compounds to Highly Aromatic Gasoline U.S Patent Number 5,180,868, January 19, 1993.

Figure 3 Catalyst deactivation with Haldor Topsoe catalyst and TR12 Oil (400 C, 13.8 MPa).

Figure 4 Effect of catalysts on the quality of hydrotreated oil from TR7 (400 C, 13.8 MPa).

In batchwise recycle tests, the results of the recycle of heavy

components to a hydrotreating bed (recycle concept) or to a

hydrocracking environment (multistep concept) were compared

Both concepts showed good gasoline yield, but the

hydrocrack-ing catalysts were more active toward molecular weight

reduc-tion and produced more gasoline These batchwise recycle tests

were even more impressive when compared to the single-pass

test, as shown in Table 4 The two-stage hydrotreatment (withintermediate separation) results in a 13% reduction in hydrogenconsumption for equivalent gasoline yield The space velocity

is increased by a factor of 4 With the use of a hydrocrackingstep, there was a similar reduction in hydrogen consumptionand a similar fourfold increase in the space velocity resulting

in a gasoline yield increase of up to 30%

Figure 5 Effect of catalysts on the quality of hydrotreated oil from TR12 (400 C, 13.8 MPa).

two

-sin gle CoM

a TR12 bio-oil feed was used in the HT step, distilled to remove volatiles <160 C before the 2nd step.

In comparison with the biomass-derived oils, the hydrotreated

products are significantly upgraded The oxygen content is

greatly reduced and, coincidentally, so is the density of the

products The density difference has a significant impact

because, although the mass yield of the hydrotreated products

is in the range of 80%, the volume yield in many cases exceeds

100% A primary concern throughout the research at PNNL

was the maintenance of the aromatic character of the biomass

oil in order to minimize hydrogen consumption and to produce

a higher octane gasoline blending stock, as recorded in one

of their patents.18 The extent of saturation as shown by the

H/C ratio is a useful indicator of the aromatic character of the

product Saturation of the aromatic components has a strongly

deleterious effect on the octane of the product A review of the

literature shows that cyclic hydrocarbons have poor octanes

similar to straight-chained hydrocarbons Although the crude

hydrotreated products did contain minor amounts of oxygen,

water solubility in the products was low In addition, although

sulfided catalysts were used in the hydrotreating, little

incor-poration of sulfur into the nearly sulfur-free biomass oils

For most of the samples listed in Table 5, the D-1319 datacompared quite favorably with the C-13 NMR results Thearomatic and aliphatic portions were nearly identical TheD-1319 method consistently showed a small olefin fraction inthe oil, while the NMR analysis detected essentially no olefiniccarbon atoms Further analysis of the fractions from the D-1319separation was performed by gas chromatography with a massselective detector (HP 5970) Individual components in eachfraction were identified and semiquantitatively determined by(18) Baker, E G.; Elliott, D C Method of Upgrading Oils Containing

Hydroxyaromatic Hydrocarbon Compounds to Highly Aromatic Gasoline.

U.S Patent Number 5,180,868, January 19, 1993.

(19) Elliott, D.C.; Schiefelbein, G F Liquid Hydrocarbon Fuels from

Biomass Prepr Pap.sAm Chem Soc., DiV Fuel Chem 1989, 34 (4),

1160-6.

Table 5 Distillate Products from Hydrotreatment

23 -

a “Actual aromatic” is the sum of the aromatic carbon and the nonaromatic substituents on the aromatic rings.

Table 6 Components of D-1319 Chromatography Fractions (within Table 7 Hydrotreating Hydrothermal Products

Each Fraction, from Highest Total Ion Count)

desalted isooctane xylene s

t

o

c

Feed Oil Composit

as Fed p

0.79 1 d

r

al

c

Product Analysi H/C atomic

s 1.38 1 e

methylpropylbenzene ethyl phenol

the intensity of the total ion current for each peak Components

in each of the fractions were listed in Table 6 With this

analysis, the NMR results were confirmed, as the primary

components of the olefin fraction were found to be bicyclic

components Some difficulty was encountered with this

analysis because of the small fraction size and the

contamination by the aliphatic fraction However, no mass

spectra of olefin components were confirmed, and the primary

components in the fraction could be determined by comparison

with the aliphatic fraction analysis Chalmers Institute of

Technology Chalmers also performed catalytic

hydroprocessing research using the TR12 product oil from the

U.S high-pressure (hydrothermal) biomass liquefaction effort at

Albany, Oregon The effort focused on preprocessing of the oil

to facilitate subsequent catalytic processing Solvent extraction

was evaluated as a means to separate a light oil from the

heavier components including the mineral material (residual

sodium salts derived from the liquefaction catalyst) Extraction

yields varied from 74% with acetone to 56% with xylene, to

30% with octane or decalin, and only 3% with pentane.20

The decalin extract of the TR12 oil was used as a feedstock

to batch hydroprocessing tests using the Akzo Ketjen 742

catalyst (sulfided CoMo on alumina).21 Conditions evaluated

(20) Gevert, B S.; Otterstedt, J.-E Upgrading of Liquefied Biomass to

Transportation Fuels by Extraction Energy from Biomass & Wastes X;

IGT: Chicago, 1986; pp 845-854.

(21) Gevert, B S.; Otterstedt, J.-E Upgrading of Directly Liquefied

Biomass to Transportation FuelssHydroprocessing Biomass 1987, 13,

105-115.

to 5 h Typical yields at 350 C and 100 bar were 1% gas, 16%naphtha, 41% atmospheric gas oil, 34% vacuum gas oil, and8% vacuum residue from a feedstock which contained 2.5%naphtha, 38.2% atmospheric gas oil, 39.5% vacuum gas oil, and19.8% vacuum residue (based on gas chromatography simulateddistillation) Coke laydown on the catalyst was evaluated andfound to be at a minimum between 350 and 375 C The cokedeposition amounted to 2-4 wt % on the catalyst, appeared tooccur during the first 2.5 h of the test, and did not increasewith longer time at the same temperature The operatingtemperature was identified as the most important factoraffecting yields with the pressure effect being much less Athigher temperature, more of the heavier components in thefeed were converted to lighter components By extrapolation, itwas found that, if the reaction could be done at 410 C, thenthe residue would be eliminated

In subsequent tests, the wood oil was desalted (water washedwhile in isooctane solvent) before hydroprocessing in a down-flow, continuous-feed, fixed-bed reactor using the same cata-lyst.22 The three feedstocks listed in Table 7 are all desalted;the last two were sequentially extracted with the indicatedsolvent from the desalted oil In tests performed at 350 C and

10 MPa of pressure, 100 ppm carbon disulfide was added tomaintain activity of the Ketjen 742 catalyst The nonextractedbut desalted oil was more readily hydrotreated than theextracted oils, and the authors concluded that desalting wassufficient with no need for extraction before hydroprocessing.(22) Gevert, B S.; Andersson, P.; Jaras, S.; Sandqvist, S Hydropro-

cessing of Desalted Directly Liquefied Biomass Prepr Pap.sAm Chem.

Soc., DiV Fuel Chem 1988, 33 (4), 913-919.

In subsequent testing, the use of wide-pore hydroprocessing

catalysts showed no improvement for hydrodeoxygenation,

hydrogenation, or hydrocracking of the desalted oil as shown

by product oxygen content, H/C ratio of the product, and the

amount of distillation residue in the product, respectively.23

However, by pore volume analysis, it was determined that the

more active small-pore catalyst more readily lost pore volume

during use and would likely be less active over time, even

though the carbon deposition measured by weight percent was

Fast Pyrolysis Bio-oil Upgrading Slow Pyrolysis Oil Upgrading

Some work in the 1970s envisioned the use of slow pyrolysis

reactor systems for wood tar production as a biomass-derived

crude oil Compared to fast pyrolysis, the slow pyrolysis oil

was recovered in lower mass yields but with lower oxygen and

water content and better thermal stability Initial research on

upgrading this oil had early success as a result of the more

easily processed oil

Texas A&M University Pyrolytic oil produced in a

Tech-Air reactor system from southern pine was the feedstock for

this research along with other tars produced in an updraft

gasifier Batch and continuous-flow reactors were used, and 20

catalyst formulations were tested.24 Decahydronaphthalene was

used as a solvent carrier at a nominal 2-to-1 ratio with the

pyrolysis oil in all these tests On the basis of batch reactor test

results at 400 C for 1 h, the Pd on alumina catalyst was

determined to be most useful with the highest liquid yield Pt

or Re on alumina and Raney nickel were nearly as useful but

produced higher gas yields Ru and Rh were found to be the

most active for gas formation The use of a carbon support in

place of the alumina support caused more gas production

Sulfided CoMo, NiMo, and NiW catalysts were found to have

much lower activity compared to the precious metal catalysts

Subsequent work focused on processing the Tech-Air pine

pyrolysis oil using the 5% Pt on alumina catalyst and

com-mercial CoMo, NiMo, and NiW catalysts in the

continuous-flow reactor.25 The reactor system was a trickle-bed type with

a 19 mm i.d and 813 mm long, of which only the top 508 mm

were packed with catalyst with inert 3 mm Pyrex beads in the

bottom The temperature of the reactor bed ranged from 190

C up to the reaction temperature, which varied from 350 to

400 C Hydrogen was fed at 107 vol/vol of oil, and the reactor

pressure was maintained at from 5.3 to 10.4 MPa The oil was

processed at a weight hourly space velocity (WHSV) from 0.5

to 3

The Pt catalyst (Strem 78-166 mixed with 20% silica and

extruded) was found to be the most active for oxygen removal

Oxygen removal ranged from 27 to 45% over the temperature

range, while that of NiMo and CoMo ranged from 15 to 39%

At the highest operating pressures, the oxygen removal was

increased to 51% Product alkanes and aromatics increased to

almost 20% using the Pt catalyst at the highest temperature at

a WHSV of 2 and 8.7 MPa, while they never exceeded 13%

using the CoMo catalyst and this was only 8% with NiMo The

Upgrading of fast pyrolysis bio-oil is a difficult propositiondue to the reactivity of the condensed bio-oil recovered fromfast pyrolysis The hydrodeoxygenation tactic used with thehigh-pressure liquefaction (hydrothermal) bio-oil is still anoverriding concern in order to make a more useful product fromthe bio-oil However, direct application of hydrotreating tech-nology from petroleum processing is not possible with fastpyrolysis bio-oil

Pacific Northwest National Laboratory (PNL/PNNL)

Bio-oil upgrading work at PNNL has focused on catalytic processing The initial work as described above focused on theproduct from high-pressure liquefaction, a hydrothermal, alkali-catalyzed process which produced a more deoxygenated productthan fast pyrolysis bio-oil As a result, the hydrothermalproduct was more thermally stable and required lesshydrogenation to produce a gasoline product An initial testwith fast pyrolysis bio-oil clearly demonstrated that thehydrotreating approach for hydrothermal product wasinappropriate for bio-oil26 as it resulted in heavy product tarplugging the reactor system and the catalyst bed encased in acokelike product The bio-oil from a poplar wood pyrolyszed in

hydro-a fluid bed rehydro-actor whydro-as hydrotrehydro-ated over hydro-a sulfided CoMocatalyst at 355 C and 13.8 MPa with a liquid hourly spacevelocity (LHSV) of 0.35 and a hydrogen flow at 318 std L/L

of bio-oil Hydrogen consumption was calculated at 127 L/L.There was only a 23% mass yield of a liquid product whichhad 3.6 and 5.9% oxygen with H/C atomic ratios of 1.55-1.45

Two-Stage Hydrotreatment In order to address the instability

of the bio-oil under hydrotreating conditions, a two-stageprocess was developed Catalytic hydrotreatment attemperatures below

300 C with either Ni or sulfided CoMo catalyst were effective

in producing a stabilized oil product This concept was patented

by Battelle at the time.27 In the attempt at a higher temperature(310 C), the reactor bed plugged with solid cokelike material

As reported,28 catalyzed hydrotreatment produced a viscous,black oil with a density of about 1, containing about 25%oxygen and an H/C atomic ratio of about 1.5 The liquid yieldwas about

0.42 L/L from a hardwood bio-oil produced in an flow reactor The presence of a hydrogenation catalyst wascritical as shown by the result in the test where only an aluminaspacer was present wherein the bed plugged with cokelikematerial Processing data are given in Table 8

entrained-Further testing of the low-temperature (stage one) cessing demonstrated that the space velocity and hydrogenflows

hydropro-NiW was reported to not be effective in oxygen removal

(26) Elliott, D C.; Baker, E G Biomass Liquefaction Product Analysis (23) Gevert, S B.; Andersson, P B W.; Sandqvist, S P

Hydroprocessing of Directly Liquefied Biomass with Large-Pore

Catalysts Energy Fuels

1990, 4, 78-81.

(24) Soltes, E J.; Lin, S.-C K.; Sheu, Y.-H E Catalyst Specificities in

High Pressure Hydroprocessing of Pyrolysis and Gasification Tars Prepr.

Pap.sAm Chem Soc., DiV Fuel Chem 1987, 32 (2), 229-239.

(25) Sheu, Y.-H E.; Anthony, R G.; Soltes, E J Kinetic Studies of

Upgrading Pine Pyrolysis Oil by Hydrotreatment Fuel Process Technol.

1988, 19, 31-50.

and Upgrading Comptes Rendus de l’Atelier de TraVail sur la Lique

´faction de la Biomasse; report 23130, NRCC: Sherbrooke, Quebec,

Table 8 Stage One Hydroprocessing of Bio-oil

Experimental Operating Conditions catalyst Ni3266 Ni3266 Ni3266 HT-400 R-Al 2 O 3

to the aqueous phase is lower Processing the aqueous phasethrough the high-temperature hydrotreating process apparentlyconverts the aqueous phase carbon to gas with an additionalconsumption of hydrogen gas

On the basis of material and elemental balance calculations,these results were produced only at near-steady state Acalculation shows that steady-state operation would be expected

to produce 0.50-0.55 L of product oil/L of feed The productoil product yield,

L/L feed 0.28 0.42 reactor plugged

Wet Product Analysis

0.42 reactor plugged oil would be expected to contain 2-3 wt % oxygen and would

have an H/C ratio of 1.5 and a density of 0.92 g/mL About50-60% of the oil would be in the gasoline boiling range; ofthe remainder, about 30% would be distillable gas oil (diesel)and the rest would be resid Gasoline yield and residual oxygencontent in the product can both be directly related to theboth played a minor role in the processing results As seen in

Table 9, space velocity variation in the range of 0.44-0.62

(three leftmost columns) has an effect on the extent of oxygen

removal and hydrogen consumption, while at higher space

velocities up to 1.6 the results remain about level Reduction of

the hydrogen flow to zero at moderate space velocity (three

rightmost columns) reduced the hydrogen incorporation

noticeably without affecting oxygen removal The product oil

quality was reduced as measured by density and viscosity

The thermal stability of the low-temperature, hydrotreated

product oil (it could be distilled batchwise without coke

formation in the pot) as well as its elemental composition

(70.7% C, 8.1% H, 0.1% N, 20.9% O, calculated to a dry

basis) indicated that it was significantly upgraded from the

original bio-oil Chemical composition analysis by GC-MS

verified that the carbonyl components were destroyed as well

as the saturation of the olefinic side chains The relative

amount of phenolic components increased at the expense of

the aromatic ethers Saturated cyclic alcohols were also present

indicating some hydrogenation of the aromatic rings

On the basis of these results, a two-stage process for

hydrotreating fast pyrolysis bio-oil to liquid fuels was

demon-strated at bench scale Gasoline-range hydrocarbons were

produced from low-temperature hydrotreated bio-oil, using a

sulfided cobalt molybdenum catalyst, at conditions of

ap-proximately 350 C and 13.8 MPa Carbon conversion to

gasoline was in excess of 80 wt % with a liquid product yield

of 0.77 L/L Hydrogen consumption was measured at 728 L/L

of oil feed based on gas-phase measurements in and out of the

reactor The space velocity in the upflow reactor was relatively

low at 0.07 vol of oil/(vol of catalyst h).29 The additional results

of the two-stage processing are presented in Table 10, as well

as calculation of the combination of the two steps.30

Nonisothermal Hydrotreatment To simplify the process and

maximize liquid product yield, the two stages were combined

in a single nonisothermal catalyst bed This concept was tested

at the bench-scale, in an upflow configuration, with several

wood-derived bio-oil feedstocks and also a peat fast pyrolysis

(29) Elliott, D C.; Baker, E G Hydrotreating Biomass Liquids to

Produce Hydrocarbon Fuels Energy from Biomass & Wastes X; IGT:

Chicago, 1986; pp 765-784.

(30) Baker, E G.; Elliott, D C Catalytic Upgrading of Biomass

Pyrolysis Oils Research in Thermochemical Biomass ConVersion;

Bridgwater, A V., Kuester, J L., Eds.; Elsevier Science Publishers, LTD.:

Barking, England, 1988; pp 883-895.

processing space velocity.15

With the peat oil,31 the results in Table 11 represent state operation The quality of the product oil from peat issomewhat better than the hydrotreated wood-derived oils Thepeat product contained more paraffinic material (includingalkylcyclohexanes), which accounts for the higher H/C ratio.However, it also contained 1.1 wt % nitrogen, a level that isseldom seen with wood-derived products

steady-More extensive testing was done with the bio-oil produced

by vacuum pyrolysis at the University of Laval in Canada Theresults32 from nonisothermal catalytic hydrotreating of primarycondenser oil are presented in Table 12 The first test wasperformed at relatively low temperatures and a high bio-oil feedrate As a result, the product quality is low, as shown by highoxygen content and density The second test was performed atconditions optimized to improve product qualityshigher tem-perature and lower feed rate Hydrogen consumption was higher

as was gas formation Because of the reduced density of theproduct, the volumetric yield increased to over 0.4 L/L based

on bio-oil feed This data set probably does not representsteady- state operation The last two data sets are derived fromlonger- term tests (18-22 h) They are believed to betterrepresent steady-state operation after break-in of the catalyst.These process tests can be further compared by plotting theproduct quality versus processing rate.33 In Figure 6, oxygencontent in the upgraded product oil is plotted as a function ofspace velocity for nonisothermal catalytic hydrotreating of bio-oils In this plot, the data for all the atmospheric fast pyrolysisprocesses (flash pyrol.) appear to fall on the same line Weconclude that the oil source components in all the feedstocksare essentially the same, so that the same type of product isproduced from all the tests However, the vacuum pyrolysis oil(vac pyrol.) appears to be somewhat more difficult to deoxy-genate to low levels In Figure 7, gasoline yield is plottedversus space velocity In this comparison, the atmospheric fastpyrolysis upgraded products again fall on the same line,suggesting an

(31) Elliott, D C.; Baker, E G.; Piskorz, J.; Scott, D S.; Solantausta,

Y Production of Liquid Hydrocarbon Fuels from Peat Energy Fuels 1988,

2, 234-235.

(32) Elliott, D C Perform Hydrotreatment of Biomass Liquefaction Products Including Vacuum Pyrolysis Oils; contract no 2312012867, letter

report to Zeton, Incorporated: Burlington, Ontario, October 20, 1988.

(33) Elliott, D C Perform Hydrotreatment of Biomass Liquefaction Products RTP (Rapid Thermal Processing) LiquidsEnsyn Engineering Associates, Inc.; contract no 2312012867, letter report to Zeton, Incorpo-

rated: Burlington, Ontario, February 8, 1989.

on

to

oil/a 55 80/1 87/ 83/5 83/7 87/11 83/9 85

to

gas oil product yield, L/L feed7 5 4 4 0.4210 10 0.56 0.69 0.66 0.65 0.70 0.61 0.61

Wet Product Analysis

Table 10 Two-Stage Hydroprocessing of Bio-oil

Experimental Operating Conditions feed oil bio-oil stage 1 product combined

equivalent yield, while the vacuum pyrolysis bio-oil gives a

somewhat higher yield of the gasoline-range product by

hydrotreating

Low-SeVerity Hydrotreatment Using the above results, a

prediction of results under less-severe processing conditions was

determined.34 As shown in Figure 8, an extrapolation of data

to higher space velocity processing can be used to predict the

oxygen content of hydrotreated biomass fast pyrolysis oils

Using this data, it was predicted that a space velocity of 0.4

would produce a product with an oxygen content of

ap-proximately 15% Such an oil should require less hydrogen

consumption and would be produced in higher yield because

there would be less gas formation This prediction was based

on continued use of the high-temperature regime (up to 400

C) and 13.8 MPa of pressure in the reactor

Initial low-severity hydrotreating tests were made using an

upflow configuration.35 It is difficult to operate in the upflow

mode because of the tendency of the product to fractionate and

plugging in the reactor by high-viscosity products The

products

(34) Elliott, D C Biomass Fast Pyrolysis OilsLow-SeVerity

Hydrotreat-ing; contract no AIR-CT-92-0216, final report to Arusia: Perugia, Italy,

December 1996.

(35) Elliott, D C.; Hart, T R.; Neuenschwander, G G.; McKinney, M.

D.; Norton, M V.; Abrams, C W EnVironmental Impacts of

Thermo-chemical Biomass ConVersion; NREL/TP-433-7867, National Renewable

Energy Laboratory: Golden, Colorado, 1995 (also PNL-10413, Pacific

Northwest National Laboratory: Richland, Washington).

with oxygen contents less than 15% were low viscosity, asgood or better than the feed oil The products with 15 to 25%oxygen were several orders of magnitude higher in viscosityand difficult to recover from the reactor system In the upflowmode, there is a tendency for the low-viscosity (and alsohigh-volatility) products to fractionate from the moreviscous, low-volatility products and leave the reactor with theexcess hydrogen In this manner, the catalyst bed slowly fillswith high-viscosity oil which exits the reactor indisproportionately high yields later in the experiment

The products from the near-steady-state operations appearmuch heavier than expected as a result of the fractionationoccurring during the initial period of the experiments Asignificant period of non-steady-state operation occurs at thebeginning of the tests in which product recovery is incompletewhile volatile products are swept from the reactor and heavier(liquid-phase) products slowly build up in the catalyst bed.Eventually (after several hours online), the catalyst bed fillswith liquid phase, the “whole” product oil begins to exit thereactor, and a near-steady-state operation is achieved.However, there remains the accounting of the light productsrecovered in the early stages of the test These must be addedback to the heavy whole products to account for a true massbalance These initial- period products are the type of lightfuel oil which is being sought after for the desired turbine-fuel product

In these tests wherein the vapor-phase product and the phase product are so different, it is difficult to bring short-termtests to steady-state operation The fixed-bed reactor in anupflow mode of operation naturally allows productfractionation In addition, the difference in product propertiesmakes control of the process equipment and recovery of theproduct more difficult than in gasoline-production tests with atotally vapor- phase product

liquid-Catalyst analysis after these initial low-severity tests indicatedsome changes had occurred There was clear evidence in some

of the catalyst pellets of the reaction of the alumina support tothe hydroxide (boehmite) This reaction in aqueous processingunder similar conditions has been reported.36,37 The high-surface-area aluminas (gamma and delta) are not stable in the presence

of water at 350 C The stability of the sulfide phases was not(36) Elliott, D C.; Sealock, L J., Jr.; Baker, E G Chemical Processing

in High-Pressure Aqueous Environments 2 Development of Catalysts for

Gasification Ind Eng Chem Res 1993, 32, 1542-1548.

(37) Laurent, E Etude et controˆ le des re´actions d’hydrode´soxyge

´nation lors de l’hydroraffinanage des huiles de pyrolyze de la biomasse Ph.D Thesis Universite´ Catholique de Louvain, Louvain-la-Neuve,

Belgium, 1993.

0.8

6

Table 11 Nonisothermal Hydroprocessing of Bio-oil

Experimental Operating Conditions feed

a strong hydrotreating catalyst effect in the reactor, exothermicpyrolysis condensation reactions caused major temperatureLHSV,

vol oil/(vol cat h) 0.28 0.13 0.11 0.13 excursions and led to coke buildup on the catalyst Buildup of

heavy tar products in the reactor and effluent lines causedExperimental Results and Product Analyses plugging and premature experiment terminations Product

of the biocrude still as a heavy tar product Total carbonbalances were difficult to measure, but very low spacevelocities would

Product Inspections H/C atomic ratio 1.47 1.73

C 5 -225 C, vol % NA 76

quantified, but there was no nickel or molybdenum oxide

evident by X-ray diffraction while the sulfides were both

identified

The results presented below were produced in the continuous

feed catalytic hydrotreater operated in a downflow

configura-tion.38 The tests were made with a NiMo on alumina (Haldor

Topsoe TK751, 1 mm extrudates) conventional hydrotreating

catalyst, presulfided processing an ablative reactor bio-oil

gener-ated from poplar hardwood by NREL A second catalyst was

also tested (BASF K8-11) This CoMo catalyst is unlike the

NiMo in that the support is a spinel not the conventional

high-surface-area γ-alumina commonly used in petroleum

hydrotreat-ing catalysts Better chemical stability of the support was

envisioned; but, the catalyst, which is not normally used for

petroleum hydrotreating, exhibited much lower activity overall

The results of these tests are summarized in Table 13 below

Higher levels of conversion were seen in the downflow

(38) Elliott, D C.; Neuenschwander, G G Liquid Fuels by Low-Severity

Hydrotreating of Biocrude DeVelopments in Thermochemical Biomass

ConVersion; Bridgwater, A V., Boocock, D G B., Eds.; Blackie Academic

& Professional: London, 1996; Vol 1, pp 611-621.

be required to effectively hydrotreat the biocrude over this catalyst.

The viscosity of the product oil is also a function of theoxygen content, as seen in Figure 9 The range shown goes allthe way from the low viscosity required for turbine fuels ofless than 5 cps to the heavy tar products with high oxygencontents and viscosities >100 000 cps whose pour points would

be around room temperature These results suggest that onlythe highly upgraded oils with oxygen contents of 5% or lesshave the potential for direct use as turbine fuels because ofviscosity limitations Figure 9 also includes the data for thethree bio-oils The raw bio-oils show decreasing viscosity withincreasing oxygen content because of increasing water content

in the raw biocrude

Figure 10 shows the effect of temperature on the viscosity

of several of the heavy oil products These results suggest thatthe highly oxygenated products could not be used as turbinefuels because of their high viscosity even with preheating The10% oxygen oil might be used as turbine fuel with preheating

to 50 C or higher

Organic contamination of the water byproduct from drotreating is a concern relative to the overall wastewatertreatment requirements for the plant Figure 11 clearly showsthat the contamination of the byproduct water as represented

hy-by the carbon content increases with the oxygen content of the

Figure 6 Effect of space velocity on product oxygen content.

Figure 7 Effect of space velocity on the yield of gasoline range hydrocarbon.

lyptus oil from fluid-bed pyrolysis (Union Fenosa) Table 14provides some of the results Although the eucalyptus-oil testwas performed at higher temperature and lower space velocity,the product-oil quality was significantly lower, with higheroxygen content and density Further substantiating the conclu-sion of lower reactivity is the lower gas yield and also the lowerhydrogen consumption The combination of high viscosity andpoor pumping performance with the lower reactivity make theeucalyptus oil more difficult to hydrotreat

A number of comparisons were made between the PNNLwork and that published by Veba Oel.39 The experimentalreactor results are evaluated in terms of deoxygenation which

Figure 8 Processing rate effect on product oxygen content.

product oil up to a range of 5-10 wt % when the product oil

oxygen content is >10% At that level, the oil-water separation

is difficult and oil phase contamination of the water samples

leads to great variation in the results Samples representing part

and all of the oil product for two-stage upflow tests are given

The effect of biocrude feed properties on the reactivity can

be significant Whereas the work at PNNL over the years had

used a number of bio-oils including several hardwood oils from

different fluid-bed pyrolysis reactors, pine oil from ablative

pyrolysis, hardwood oil from vacuum pyrolysis, and peat oil

from fluid-bed pyrolysis without identifying any major

differ-ences in reactivity; more recent work showed large differdiffer-ences

between poplar oil from ablative pyrolysis (NREL) and

euca-incorporates the effects of temperature, pressure, and catalyst.The deoxygenation is defined as the percent of chemicallycombined oxygen in the bio-oil removed when compared tothe oil product Deoxygenation as a function of space velocityshows a dramatic effect In Figure 12, space velocity ispresented in terms of volume (liquid hourly space velocity) Bythis measure, the new downflow results in this paper stand outclearly compared to all of the earlier-published results Thus,our earlier agreement with Veba that the effect of downflowoperation compared to upflow was small considering theoverall reaction and little difference between our earlier resultsand those

(39) Baldauf, W.; Balfanz, U Upgrading of Pyrolysis oils from Biomass in Existing Refinery Structures; final report JOUB-0015, Veba

Oel AG: Gelsenkirchen, 1992.

Figure 9 Relation of oil viscosity to oxygen content.

Figure 10 Effect of temperature on oil viscosity.

Table 13 Low-Severity Hydrotreating Results in a Downflow Operation

3

3

4

1

1 pr

of Veba However, Figure 13 provides a data comparison on a

weight basis (weight hourly space velocity) wherein the effect

of the diluted CoMo catalyst bed used by Veba has a dramatic

effect By using the diluted CoMo catalyst bed, Veba achieved

much higher processing rates based on the weight of catalyst

The one unexplained result is the poor showing of the NiMo

catalyst in the downflow experiments at Veba Again, our

downflow results are much improved compared to our earlierupflow results In all of these tests, the catalysts are supported

on an alumina base The differences in the use of CoMo versusNiMo are relatively small compared to the difference betweenupflow and downflow

The product-oil density is clearly a function of product-oil oxygen content as seen in Figure 14 The molecular weight may