Bioresource Technology 124 (2012) 190–198

Contents lists available at SciVerse ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Direct liquefaction of Dunaliella tertiolecta for bio-oil in sub/supercriticalethanol–water

Chen Yu a, Wu Yulong a,⇑, Zhang Peiling a,b, Hua Derun a, Yang Mingde a,⇑, Li Chun c, Chen Zhen a, Liu Ji a

a Institute of Nuclear and New Energy Technology, Tsinghua University, Beijing 100084, PR Chinab School of Chemistry and Chemical Engineering, Shihezi University, Shihezi 832000, Xinjiang, PR Chinac School of Life Science, Beijing Institute of Technology, Beijing 100081, PR China

h i g h l i g h t s

" Bio-oil was produced by liquefactionof Dunaliella tertiolecta in sub/supercritical ethanol–water.

" The ethanol and water showedsynergistic effects on the directliquefaction of D. tertiolecta.

" XPS and SEM were used to verify theliquefaction behavior of D. tertiolectaand its solid residue.

" A plausible reaction mechanism ofthe main chemical component in D.tertiolecta is proposed.

" The optimal D. tertiolecta conversionwas 98.24%, with a maximum bio-oilyield of 64.68%.

0960-8524/$ - see front matter Crown Copyright � 2http://dx.doi.org/10.1016/j.biortech.2012.08.013

⇑ Corresponding authors. Tel.: +86 108 979 60 88;E-mail addresses: wylong@tsinghua.edu.cn (Y. W

(M. Yang).

g r a p h i c a l a b s t r a c t

The bio-oil preparation by direct liquefaction of microalgae (Dunaliella tertiolecta) was carried out withsub/supercritical ethanol–water mixture as the medium in a batch autoclave with high temperatureand high pressure. The results indicated that ethanol and water showed synergistic effects on the directliquefaction of D. tertiolecta. The optimal D. tertiolecta conversion was 98.24%, with a maximum bio-oilyield of 64.68% in the sub/supercritical ethanol–water mixture at a reaction temperature of 593 K, witha holding time of 30 min, a ratio of the material to reaction medium of 1:10, and an ethanol volume frac-tion of 40% (v/v).

a r t i c l e i n f o

Article history:Received 7 May 2012Received in revised form 1 August 2012Accepted 3 August 2012Available online 17 August 2012

Keywords:Sub/supercritical ethanol–waterBio-oilDirect liquefactionDunaliella tertiolectaX-ray photoelectron spectroscopy (XPS)

a b s t r a c t

This paper presents bio-oil preparation by direct liquefaction of Dunaliella tertiolecta (D. tertiolecta) withsub/supercritical ethanol–water as the medium in a batch autoclave with high temperature and highpressure. The results indicated that ethanol and water showed synergistic effects on direct liquefactionof D. tertiolecta. The maximum bio-oil yield was 64.68%, with an optimal D. tertiolecta conversion of98.24% in sub/supercritical ethanol–water. The detailed chemical compositional analysis of the bio-oilwas performed using an EA, FT-IR, and GC–MS. The empirical formulas of the bio-oil obtained usingthe ethanol–water co-solvent (40%, v/v) and sole water as the reaction medium were CH1.52O0.14N0.06

and CH1.43O0.23N0.09, with calorific values of 34.96 and 29.80 MJ kg�1, respectively. XPS and SEM resultsshowed that ethanol–water is a very effective reaction medium in the liquefaction. A plausible reactionmechanism of the main chemical component in D. tertiolecta is proposed based on our results and theliteratures.

Crown Copyright � 2012 Published by Elsevier Ltd. All rights reserved.

012 Published by Elsevier Ltd. All r

fax: +86 106 977 14 64.u), yangmd@tsinghua.edu.cn

ights reserved.

Y. Chen et al. / Bioresource Technology 124 (2012) 190–198 191

1. Introduction

Recently, energy from biomass resources has received remark-able attention due to the increasing worldwide energy require-ments, and the demand for biofuels has rapidly increased in therecent decades. Biofuels can be broadly defined as solid, liquid, orgaseous material consisting of or derived from renewable biomassresources (Zou et al., 2010a; Duan and Phillip, 2011). Among thebiomass resources, microalgae are especially promising feedstockbecause of their higher photosynthetic efficiency, growth rate,and area-specific yield (Patil et al., 2008; Clarens et al., 2010; Zouet al., 2010b). Furthermore, microalgae can be cultivated on a largescale in either freshwater or saline environments, and do not com-pete with the use of agriculturally productive or environmentallysensitive land (Sarmidi, 2009; Yu et al., 2011). Microalgae had beenconsidering as very good candidate for the third-generation biofu-els and chemicals (Pereira et al., 2004).

So far, many efforts have been made to convert microalgae toliquid fuels and chemicals by thermochemical conversion tech-niques, including pyrolysis, hydrothermal liquefaction, gasification,etc. (Zou et al., 2010a). Among these techniques, hydrothermal liq-uefaction was considering as an important thermochemical lique-faction method with a quick and efficient way due to uniqueproperties of sub/supercritical water: low viscosity and controllablesolvent power (Savage, 1999; Sakaki et al., 2002). Our previous workshowed sub/supercritical water has good performance on Dunaliellatertiolecta liquefaction (Zou et al., 2010c). However, the higher crit-ical temperature and critical pressure of water (647.3 K, 22.1 MPa)meant challenging operation conditions (George et al., 2006). Be-sides, using water as the medium for biomass liquefaction resultedin lower yield and lower calorific value for bio-oil (Matsumura et al.,1999).

Organic solvents, such as ethanol, methanol, acetone, etc., havebeen utilized as the reaction medium instead of water to enhancethe bio-oil yield with lower oxygen content (Yamazaki et al., 2006;Yuan et al., 2007; Liu and Zhang, 2008; Xu and Etcheverry, 2008).Cheng et al. proved that alcohol (methanol or ethanol) and watershow synergistic effects on direct liquefaction of biomass (Chenget al., 2010). Methanol, ethanol and butanol all can be derived fromrenewable resources, however, methanol is a poisonous chemical,and ethanol is cheap and easily available in the process ofindustrial production comparing with butanol. Thereby, theethanol–water co-solvent was selected as the medium has broadapplication prospects.

The introduction of ethanol has several advantages (Jade andPattarapan, 2010): First, the critical temperature and critical pres-sure of ethanol (516.2 K, 6.38 MPa) is far below that of water, somuch milder reaction conditions can be obtained. Second, ethanolcan provide active hydrogen as a hydrogen-donor in the liquefac-tion process. Third, ethanol can react with acidic components inthe bio-oil by esterification reaction to obtain fatty acid ethyl es-ters similar to biodiesel. Finally, due to relatively lower dielectricconstant, ethanol can readily dissolve relatively high-molecular-weight products derived from microalgae. There are many papersthat have reported on the liquefaction of biomass (chosen as ligninor cellulose) with sub/supercritical ethanol–water as the medium.However, the biochemical components of microalgae differ fromthose of terrestrial biomass remarkably, which have high proteincontent. Published works involving the liquefaction of algae withsub/supercritical ethanol–water as the medium are not available.

In this study, sub/supercritical ethanol, water, and their co-sol-vents were tested for direct liquefaction of Dunaliella tertiolecta (D.tertiolecta) at a temperature range of 533–613 K. The effects of theethanol–water co-solvent composition and the reaction tempera-ture were examined with a fixed reaction time. Features of the

obtained bio-oils were characterized by Fourier transform infraredspectroscopy (FT-IR), gas chromatography–mass spectrometry(GC–MS), and elemental analysis (EA). X-ray photoelectron spec-troscopy (XPS) and scanning electron microscopy (SEM) were usedto verify the liquefaction behavior on the basic of D. tertiolecta andsolid residues. A plausible reaction mechanism of main chemicalcomponent in D. tertiolecta is proposed based on our results andthe literatures.

2. Methods

2.1. Materials

D. tertiolecta was obtained from Tianjin Microalgae BioTechnol-ogies Co., Ltd. (Tianjin, China). The samples were prepared throughpulverization in a mortar to <75 lm and dried at 378 K for 12 h be-fore use. The proximate and ultimate analysis results of the sampleand its structural compositions are given in Table 1. The Analysismethods were reported in our previous work (Zou et al., 2010c).

2.2. Thermochemical liquefaction

The reaction was performed in a stainless autoclave with 50 mLcapacity. The autoclave was heated with an external electrical fur-nace, and the temperature was measured with a thermocouple andcontrolled to ±1 K. In a typical run, D. tertiolecta (approximately4 g), ethanol [the co-solvent with the content of ethanol rangingfrom 0% to 100% (v/v)], and water were fed into the reactor, thensealed and heated up to the desired reaction temperature (from533 to 613 K), and the temperature was kept 30 min. When thereaction ended, the electric furnace was removed, and the auto-clave was cooled down to room temperature.

The separation of liquefaction products is depicted in Fig. 1. Theproducts included the gaseous, organic phase, and water-solublephases, as well as the solid residue. The gaseous product wasvented after the autoclave had cooled down. The liquid phase frac-tion and the autoclave wall were washed with chloroform thrice,and the contents were separated through dispersion. The chloro-form solvent was removed in a rotary evaporator at 313 K underreduced pressure, and the liquid fraction remained was called‘‘bio-oil’’. The water-insoluble fraction remaining on the filter pa-per was dried at 378 K for more than 24 h, then weighed, and des-ignated as the ‘‘solid residue’’ (SR). The solid residues wereclassified as Residue-CS when the ethanol–water co-solvent usedas the medium and as Residue-SS when sole water as the medium.The testing of the gaseous and water-soluble reaction products isbeyond the scope of this paper, but it could be accomplished inthe future by analysis of the gaseous molecules and organic sub-stances in the aqueous. The experimental errors for the liquefac-tion yields and D. tertiolecta conversion are lower than 5% bythree runs repeatedly at the same conditions.

2.3. Analysis of bio-oil

The ash, moisture, and density of bio-oil were determinedaccording to American Society for Testing and Materials (ASTM)standard. Every bio-oil sample was analyzed by three runs repeat-edly at the same conditions, and taken the average as the final re-sults. The elemental composition was performed on a CE440elemental analyzer. The calorific value was calculated accordingto the Dulong formula (Huang et al., 2011):

Calorific valueðMJ kg�1Þ ¼ 0:3383Cþ 1:442ðH—O=8Þ ð1Þ

where C, H, and O are the weight percentage of carbon, hydrogen,and oxygen in the bio-oil, respectively.

Table 1Analysis results of D. tertiolecta.

Industrial analysis/% Chemical composition analysis/% Elemental analysis/%

Moisture War 25.48 Protein 50.30 C 39.00Ash Aar 9.76 Fat 17.80 H 5.37Volatiles Var 53.13 Carbohydrate 21.70 Oa 43.88Fixed carbon CFar 11.63 Cellulose 5.24 N 1.99

Othersa 4.96

a Calculated by difference.

Fig. 1. Procedure for the separation of reaction products.

192 Y. Chen et al. / Bioresource Technology 124 (2012) 190–198

FT-IR analysis was performed on a Spectrum GX series FT-IRspectrometer to determine the functional groups. GC–MS analysiswas carried out on a Trace DSQ GC–MS system with an AB-5MScapillary column (30 m � 0.25 mm id, 0.25 lm film thickness), He-lium was used as carrier gas, with a flow rate of 1 mL min�1. Thecolumn temperature was programmed from 343 to 573 K at a rateof 10 K min�1 after an initial two-minute isothermal period, then itwas kept at the final temperature for 10 min. The inlet temperaturewas set to 573 K, and the split ratio was 1:50. The mass spectrom-eter was set to an ionizing voltage of 70 eV with a mass range from35 to 650 amu. SEM images were obtained using a JSM-6490LV(JEOL Ltd.). XPS measurements were carried out using a PHI Quan-tera microprobe (ULVAC-PHI, Inc.) equipped with an aluminum an-ode as the monochromatized X-ray source (1486.7 eV run at 10 kVand 15 mA in fixed analyzer transmission mode). The peak fittingprocedure was performed with the XPS Peak (version 4.1) program.The C–C peak was set to 284.8 eV. The oxidized to unoxidizedcarbon ratio (Cox/unox) was calculated as follows (Matuana andKamdem, 2002):

Cox=unox ¼ Coxidized=Cunoxidaized ¼ ðC2 þ C3 þ C4Þ=C1 ð2Þ

2.4. Calculation

The bio-oil and SR yield, as well as the D. tertiolecta conversion,were expressed in% w/w, and calculated based on the dry organicmatter as follows:

Bio-oil yield ¼ x1=x0 � 100% ð3Þ

SR yield ¼ x2=x0 � 100% ð4Þ

Others yield ¼ 100%� Bio-oil yield� SR yield ð5Þ

D:tertiolecta conversion ¼ 100%� SR ð6Þ

where x0 (g), x1 (g), and x2 (g) were defined as the weights of D.tertiolecta, the bio-oil, and the SR, respectively. For the approxima-tion, the total yield of the discarded products (gas + water-soluble),defined as ‘‘others yield’’, was obtained from the weight difference.

3. Results and discussion

3.1. Physical and chemical properties of bio-oil

The physical and chemical properties of bio-oil, which directlyaffect its application and use efficiency, are very important toestimate the treatment technology and the selection of the processequipment. In this study, the chemical and elemental composi-tions, as well as the calorific value of bio-oil obtained fromD. tertiolecta, were analyzed. The results are listed in Table 2. Theethanol–water co-solvent represents that the ethanol content is40% (v/v) in this study unless otherwise specific noted.

Low ash content was obtained in direct liquefaction of D. tertio-lecta using either ethanol–water co-solvent or sole water as themedium, with ash content of 0.19% and 0.28%, respectively. Thisfinding indicates that the bio-oil obtained with the ethanol–waterco-solvent as the medium has more obvious potential as a cleanerfuel oil compared with bio-oil obtained in sole water medium.

0 20 40 60 80 1000

20

40

60

80

100(98.24%) (95.73%) (91.81%)(98.39%)(98.45%)(95.39%)

Yie

ld (

%)

Ethanol content (v/v%)

SR Others Bio-oil

Fig. 2. Effect of ethanol content on the bio-oil, others and SR yields for directliquefaction of D. tertiolecta at 593 K in ethanol–water co-solvent. The value of D.tertiolecta conversion (%) is shown in parentheses.

Y. Chen et al. / Bioresource Technology 124 (2012) 190–198 193

The organic acid was generated from the oxidation of a low-molecular-weight fraction liquefaction of D. tertiolecta, so the liq-uefaction product was acidic, as denoted by the pH value. Thebio-oil pH (5 to 6) approached neutrality using ethanol–waterco-solvent as the medium. This result might be caused by the reac-tion between ethanol and fatty acids produced in the liquefactionprocess to form fatty acid ethyl esters.

Table 2 also presents the EA results. The carbon and hydrogencontents in the bio-oil produced using ethanol–water co-solventas the medium were both higher than those of the raw materialsand of the bio-oil with sole water as the medium. This finding illus-trates that the bio-oil obtained using the co-solvent as the mediumhas a higher energy density, with a calorific value of 34.96 MJ kg�1

on the average. The empirical formulas of the bio-oils obtainedusing ethanol–water co-solvent and sole water as the reactionmedium were CH1.52O0.14N0.06 and CH1.43O0.23N0.09, with calorificvalues of 34.96 and 29.80 MJ kg�1, respectively.

3.2. Effects of ethanol–water composition

D. tertiolecta was liquefied in sub/supercritical ethanol–water atdifferent ethanol contents, varying from 0% to 100% (v/v) at 593 Kfor 30 min to determine the optimal ethanol content. The bio-oil,SR, and others yields as well as the conversion, were plottedagainst the ethanol content as shown in Fig. 2.

Fig 2 reveals that using either ethanol or water as the medium isless effective for conversion or bio-oil yield compared with theethanol–water co-solvent. Ethanol and water showed synergisticeffects on direct liquefaction of D. tertiolecta. The conversion andbio-oil yield peaked with the 40% (v/v) ethanol solution, but inter-estingly, both the conversion and bio-oil yield were greatly re-duced as the ethanol content in the co-solvent increased to 80%(v/v). This phenomenon can be explained as follows, when the eth-anol content is 40% (v/v), the critical temperature of the co-solventis about 599 K (Yuan et al., 2007), and the actual liquefaction tem-perature (593 K) is very close to the critical point. With increasingof ethanol content, the critical pressure and critical temperature ofthe co-solvent decrease, therefore, the liquefaction process wascarried out under sub/supercritical conditions in this study. Sub/supercritical water can provide ionic, polar non-ionic and free rad-ical in biomass liquefaction for bio-oil (Cheng et al., 2010). Sinceethanol is slightly weaker acids than water, the increase ethanolcontent resulted in the decrease of free-radical derived from ioni-zation water, which is not conducive to D. tertiolecta liquefaction,and brought on the bio-oil yield reduce. The best ethanol contentwas chosen as 40% (v/v), where the optimal conversion and bio-oil yield were 98.24% and 64.68%, respectively. This bio-oil yieldis significantly higher than that of other published results (Clarens

Table 2Physical properties of bio-oil using different reaction medium.

Properties Bio-oil

Water solvent Ethanol–water co-solvent

Ash/wt.% 0.28 0.19H2O/wt.% 0.73 0.15pH 3–5 5–6C/wt.% 65.43 71.78H/wt.% 7.82 9.10O/wt.%a 19.98 13.58N/wt.% 6.49 5.35H/C 1.43 1.52O/C 0.23 0.14Density/g cm�3 1.31 1.04Empirical formula CH1.43O0.23N0.09 CH1.52O0.14N0.06

Calorific value/MJ kg�1 29.80 34.96

a Calculated by difference.

et al., 2010; Zou et al., 2010c), however, the amount of ethanol con-verted into the bio-oil was lower than 5%, in accordance to the es-ter concentration in the bio-oil according as the GC–MS results(Table S1 in Supplementary data). The molecular weight (MW) ofheptadecanoic acid ethyl ether and hexadecanoic acid ethyl etheris 298 and 284, respectively. According to the reaction equationsbetween acid and ethanol, each generate an ester molecule; theethanol contribution to MW of ester is 28, lower than 10%. Coupledwith the fact that the amount of this two ethers appearing in thebio-oil can reach 50% based on GC–MS results, the amount of eth-anol converted into the bio-oil is lower than 5% finally.

3.3. Effects of the reaction temperature

In the present study, slightly higher temperatures from 533 to613 K were tested for direct liquefaction of D. tertiolecta. The ef-fects of the reaction temperature on the bio-oil, SR, and othersyields, as well as the conversion in ethanol–water co-solvent areshown in Fig. 3. The results clearly showed that the bio-oil yield in-creased continuously from 47.11% to 64.68% as the reaction tem-perature increased from 533 to 593 K, whereas conversion raiseslightly from 92.84% (at 533 K) to 98.24% (at 593 K). At the begin-ning of liquefaction, the dominant reaction was thermal cracking;thus, the elevated temperature was conducive to liquefaction.However, the bio-oil yield obviously declined as the temperatureexceeded 593 K, probably due to the formation of gas and/or smallwater-soluble molecules through further thermal cracking of thebio-oil. Similar observations were reported previously in direct liq-uefaction of algae in water (Clarens et al., 2010). Thus, 593 K is theoptimal temperature for the liquefaction of D. tertiolecta to producebio-oil.

The following competitive reactions are discussed to further ex-plain this phenomenon. The liquefaction process can be described astwo reactions. First, the formation of solids through the cyclization,condensation, and repolymerization of the liquid products, and theformation of gases from the decomposition of the liquid products,can decrease the bio-oil yield. On the other hand, the formation ofliquid products via the decomposition of solids and the aggrega-tion/condensation of the gases result in the increase of the bio-oilyield (Liu and Zhang, 2008). It could conclude that the effect of en-hanced temperature on the second reaction outweighed the firstone based on our research results. During the liquefaction, the maincomponents of algae were decomposed and depolymerized into

525 540 555 570 585 600 6150

20

40

60

80

100(99.35%)(98.24%)(97.02%)(95.17%)(92.84%)

Yie

ld (

%)

SR Others Bio-oil

Temperature (K)

Fig. 3. Effect of temperature on the bio-oil, others and SR yields for directliquefaction of D. tertiolecta with ethanol–water co-solvent (40%, v/v). The value ofD. tertiolecta conversion (%) is shown in parentheses.

194 Y. Chen et al. / Bioresource Technology 124 (2012) 190–198

lighter molecules or fragments, and these unstable fragments wererearranged through condensation, cyclization, and polymerizationto form new compounds (Qu et al., 2003; Yuan et al., 2007; Liet al., 2009). In addition, ethanol can react with the decompositionintermediates, especially at higher temperatures, to promote bio-oil formation and relatively prevent SR formation from the decom-position intermediates. Li et al. had earlier proposed that ethanolcould react with the free radicals produced from the decompositionof sludge (Li et al., 2010).

3.4. Effects of the addition methods of ethanol

A two-step method was carried out in this study to furtherstudy the effect of ethanol on direct liquefaction of D. tertiolecta.This method involves D. tertiolecta liquefaction with only wateras the medium at 593 K for 30 min, with a solvent: material ratioof 10:1. Then, 40% (v/v) ethanol was added to the liquefactionproducts at temperature same with liquefaction process for30 min. The obtained results are shown in Table S1 in Supplemen-tary data. The bio-oil components produced in the two-stepmethod were different from those produced via direct liquefaction.Much of the main products, such as heptadecanoic acid ethyl ester,hexadecanoic acid ethyl ester, 4,4,7a-trimethyl-5,6,7,7a-tetrahy-dro-2(4H)-benzofuranone, N-ethyl hexadecane amide, hexadeca-noic amides, and so on, could be observed in both process, butthere were many products in the bio-oil only obtained by thetwo-step method, such as a-eicosene, 1,1,3-trimethyl-2-cyclohexanone, 1-(hexahydro-pyrrolizin-3-ylidene)-propan-2-one,(9E)-N,N-dimethyl-9-octadecenamide, 2,6,10,14-tetramethyl-2-hexadecene, N-butyloctadecanamide, and so on. Interestingly,when sole ethanol as the medium, the bio-oil composition changesremarkably, and only hexadecanoic acid ethyl ester, 4,4,7a-tri-methyl-5,6,7,7a-tetrahydro-2(4H)-benzofuranone, and (Z)-9-octa-decenoic acid ethyl ester can be observed.

The research results indicated that the adding of ethanol notonly just reacted with the amides and/or acids to form ethyl esters,but also used as hydrogen-donor in the liquefaction process.Previously studies had mentioned the role of ethanol as hydro-gen-donors in algae liquefaction processes for bio-oil (Yuan et al.,2007; Cheng et al., 2010; Chumpoo and Prasassarakich, 2010; Zhouet al., 2012). However, the liquefaction of algae with sub/supercrit-ical ethanol–water as the medium is a very complex processes, andlots of reactions could take place, the hydrogen-donor role of

ethanol cannot be regarded as a simple dehydrogenation of etha-nol, direct evidence of the hydrogen-donor effect could be obtainedby hydrogen isotopic tracer method, and it is beyond the scope ofthis paper, but it could be accomplished in my future work.

3.5. Characterization of liquefaction products

3.5.1. FT-IR analysisThe FT-IR spectra of D. tertiolecta, bio-oil (Ethanol–water co-sol-

vent as the medium), and Residue-CS are shown in Fig. S1 inSupplementary data. The IR adsorption profiles of D. tertiolectapresented a broad and strong absorption peak from 3100 to3600 cm�1, which indicates the presence of higher amino-contain-ing and hydroxyl-containing compounds, likely to be protein, cel-lulose, and/or polysaccharide compounds. The adsorption at1700 cm�1 may be ascribed to C=O stretching, which identifiesthe carbonyl compounds, and are usually derived from lipidsand/or proteins. Compared with the IR adsorption profiles ofD. tertiolecta, the absorption weakened obviously in the bio-oil,was between 3100 and 3600 cm�1. This finding suggests that thecracking reactions of the protein, cellulose, and/or polysaccharidecompounds converted them into other molecules; thus, the con-centrations of the amino-containing and/or hydroxyl-containingcompounds also decrease. The typical hydroxyl group absorptionat 3400 cm�1 caused by the combination and overlap of aliphaticand aromatic O–H stretching from the phenolic compounds as wellas from the moisture, was inevitably observed in the samples. Theabsorption between 2800 and 2960 cm�1 could be attributed to thesymmetrical and asymmetrical C–H stretching vibrations of themethyl and methylene groups, respectively. The presence of bothO–H and C=O stretching (absorption around 1700 cm�1) vibrationsmay also indicate the presence of carboxylic acids and their deriv-atives. The band at 1379 cm�1 was attributed to C–H bending.Meanwhile, the bands between 700 and 900 cm�1 indicated thepresence of single, polycyclic, and substituted aromatic groups.However, from the IR adsorption profiles of Residue-CS, the bandsof the ether groups (1000 to 1400 cm�1) are relatively strongerthan those in bio-oil, which suggests that the ether groups be-tween fragments and ethanol can be improved after dehydrationand condensation reactions. There were no characteristic absorp-tion peaks at 1400 to 4000 cm�1; thus, most of the proteins, carbo-hydrates, fats, and/or cellulose had formed other chemicalproducts in the D. tertiolecta liquefaction process. The resultverified that D. tertiolecta could be effectively liquefied using etha-nol–water co-solvent as the medium.

3.5.2. GC–MS analysisThe organic compounds obtained from bio-oil were identified

by GC–MS. The identification of GC–MS peaks was based, in mostcases, on a comparison with the spectra of the NIST 98 spectrumlibrary. Over 100 peaks were displayed in the GC–MS chromato-grams. Fig. S2 in Supplementary data presents the total ion current(TIC) of the chemical compounds in the bio-oils from the D. tertio-lecta liquefaction in the ethanol–water co-solvent [with the etha-nol content varying from 0% to 100% (v/v)]. The relative percentarea for each compound identified (defined by the percentage ofthe chromatographic area of the compound out of the total area)and the total percent area for all of the identified compounds areshown in the Table S1 in Supplementary data when the ethanolcontent was 0, 40, and 100% (v/v), respectively. It should be notedthat the present percent area values in this study only illustrate therelative concentration of each compound in the fraction of the bio-oil that can be vaporized and passed through the GC column.

From Fig. S2 in Supplementary data, the TIC characteristicspeaks of 5,6,7,7a-tetrahydro-4,4,7a-trimethyl-2(4H)-benzofura-none and 2,6,10,14-tetramethyl-2-hexadecene were shown to

0 20 40 60 80 100

0

20

40

60

80Hexadecanoic AcidHexadecanoic Acid Ethyl EsterHexadecanoic amides

Rel

ativ

e co

nten

t (%

)

Ethanol content (%)

Fig. 4. Relative content changes of hexadecanoic amides, hexadecanoic acid, andhexadecanoic acid ethyl ester in the bio-oil from D. tertiolecta liquefaction at 593 Kin ethanol–water co-solvent vs. ethanol content.

Y. Chen et al. / Bioresource Technology 124 (2012) 190–198 195

gradually weaken along with the increase of the ethanol content,whose retention times were 21.54 and 28.32 min, respectively,when the ethanol content varied from 0% to 100% (v/v). Mean-while, the hexadecanoic acid (retention time as 30.08 min) contentdecreased and the hexadecanoic acid ethyl ester (31.26 min) con-tent increased as the ethanol content increased, suggesting thatethanol can react with hexadecanoic acid to form hexadecanoicacid ethyl ester. This phenomenon results in the significant de-crease of the amide and acid contents in the bio-oil. There is a largenumber of heptadecanoic acid ethyl ester in the bio-oil with theethanol–water as the medium, compared with the few heptadeca-noic acid or heptadecanoic amides produced when sole water usedas the medium. This phenomenon may be related to the hydrogen-donor effect of ethanol.

The bio-oil obtained with sole water as the medium contained alarger amount of amides and acids. Likewise, there were more es-ters compared with products when the ethanol–water used as themedium. This finding suggests that the sub/supercritical ethanol–water could facilitate ester formation. There are two main reasonsfor this phenomenon. First, ethanol can react with amides andacids to form ethyl esters, which results in decreased amide andacid content in the bio-oil, whereas the ester components contentsare increased with ethanol–water as the medium. On the otherhand, ethanol is an important hydrogen-donor solvent; thus,hydrogen transfer can take part in the liquefaction reaction.

Hexadecanoic acid derivatives are one of the most importantcomponents of the bio-oil obtained in the direct liquefaction pro-cess. The content of hexadecanoic acid, hexadecanoic acid ethyl es-ter, and hexadecanoic amides in the bio-oils against the ethanolcontent are shown in Fig. 4. The content of hexadecanoic acid ethylester and hexadecanoic amides was increasing as the ethanol con-tent varies from 0% to 40% (v/v) compared with the decrease ofhexadecanoic acid content. However, when the ethanol contentwas above 40% (v/v), the hexadecanoic acid content was constant,the hexadecanoic acid ethyl ester content still increased, and thehexadecanoic amide content begun to decline. Therefore, there isa competitive reaction with carboxylic acid between ammoniaand ethanol to form amide and ester during the liquefactionprocess. Thus, the amide amount increases compared with thereduced amount of the carboxylic acid. When the ethanol contentis above 40% (v/v), the amides begin to react with ethanol. Accord-ingly, the amide and acid contents are reduced. This findingsuggests that the hexadecanoic acid ethyl esters are not only ob-tained from the esterification reaction of hexadecanoic acid andethanol, but also from alcoholysis of hexadecanoic amides.

3.6. The liquefaction behavior of D. tertiolecta

3.6.1. SEM analysisThe morphological changes of D. tertiolecta before and after

liquefaction were observed by SEM. Fig. 3 in Supplementary datais showed the comparison among the SEM images of D. tertiolecta,Residue-SS, and Residue-CS. In Fig. S3a in Supplementary data,D. tertiolecta indicated abundant natural spherical structure com-pared with Residue-SS (Fig. S3b in Supplementary data) and Resi-due-CS (Fig. S3c in Supplementary data). However, there was nocell structures existed the Residue-CS replaced by some crystalsubstances. This result may be due to the formation of ether inthe solid residue, and is in accord with the results of the FT-IRresults. This phenomenon may be explained by the relatively lowerdielectric constant of ethanol, it can readily dissolve the relativelyhigh-molecular-weight products derived from D. tertiolecta.However, there are few spherical structures in Residue-SS. Fromthese results, it can conclude that ethanol–water as the mediumis superior to water for D. tertiolecta liquefaction.

3.6.2. XPS analysisXPS analysis has been reported as an interesting technique to

characterize the biomass surface chemical composition in relationwith interfacial phenomena (Sernek et al., 2004; Nzokou and Kam-dem, 2005). D. tertiolecta consist of carbon, hydrogen, nitrogen, andoxygen. In this study, the main elements detected using XPS wereoxygen and carbon. XPS was also conducted on the C1s and O1s re-gion for D. tertiolecta, Residue-CS, and Residue-SS to determine thetypes and amounts of carbon–oxygen bonds present. The C1s peaksof the XPS spectra were fitted using various carbon components(Fig. 5). Table 3 shows that the spectra of the C1s carbon can bedeconvoluted into four components (C1, C2, C3, and C4). The C1 peakcorresponds to carbon bound to carbon or hydrogen (C–C or C–H),which is ascribed to the carbon present in fats, carbohydrates, andcellulose. The C2 peak is attributed to carbon singly bonded to oxy-gen (C–O), present in the hydroxyl or ether groups of carbohy-drates and cellulose. The C3 peak represents carbon with doublebonds to oxygen (C=O), which is mainly due to the carbonyl groupsin the protein. The C4 peak identifies the carbon bonded to threeoxygen atoms (O–C=O or HO–C=O), which can correspond to thecarbonyl of carboxylic acids or to the esters present in proteinsor fats.

An O/C atomic ratio is calculated as the initial indication ofmaterial liquefaction (Table 3). The O/C ratio in the Residue-CS is0.51, which is remarkably higher than that of D. tertiolecta and Res-idue-SS, primarily as a result of a large number of hydrogen andcarbon converted into bio-oil and a relative increase in the concen-tration of oxygen atoms after liquefaction. So the O/C ratio in-creased after liquefaction for both Residue-CS and Residue-SS,indicating that the hydrogen and carbon of the surface materialhad been effectively transformed. Differences in the oxidized-to-unoxidized carbon ratio for D. tertiolecta before and after liquefac-tion process using different solvents were apparent (Table 3). Theliquefaction process decreased the concentration of carbon–oxy-gen bonds (C2 + C3 + C4). This result was expected because the ma-jor components of D. tertiolecta have a larger percentage ofoxidized carbon in their chemical structure. The carbonyl and hy-droxyl groups are the main carbon–oxygen functional groups inD. tertiolecta, which became evident as the larger C2 and C3 values.As expected, liquefaction significantly decreased the Cox/unox ratio,proving that the carbonyl and hydroxyl groups of the D. tertiolectawere converted into other forms during the liquefaction process. Inaddition, the Cox/unox ratio in Residue-CS is significantly lower thanthat of Residue-SS. This finding may be due to the formation of

292 290 288 286 284 282 280 2780

400

800

1200

1600

2000

C4

C3

C1

C2

MicroalgaeC1s

Binding Energy (ev)

Inte

nsit

y (a

.u)

540 538 536 534 532 530 528 526 524400

800

1200

1600

2000

2400

MicroalgaeO1s

O1s2

Inte

nsit

y (a

. u)

Binding Energy (ev)

O1s1

292 290 288 286 284 282 280 2780

400

800

1200

1600

2000

C4

C3

C2

Residue-SSC1s

Inte

nsit

y (a

.u)

Binding Energy (ev)

C1

540 538 536 534 532 530 528 526 5240

400

800

1200

1600

2000

2400

O1s2

O1s1

Inte

nsit

y (a

.u)

Binding Energy (ev)

Residue-SSO1s

292 290 288 286 284 282 2800

400

800

1200

1600

2000

Inte

nsit

y (a

.u)

Binding Energy (ev)

Residue-CSC1s

C1

C2

C3

C4

540 538 536 534 532 530 528 526 524400

800

1200

1600

2000

2400

Residue-CSO1s

Inte

nsit

y (a

. u)

Binding Energy (ev)

O1s2O1s1

Fig. 5. C1s and O1s photoelectron spectra of D. tertiolecta before and after liquefaction.

196 Y. Chen et al. / Bioresource Technology 124 (2012) 190–198

ether in the liquefaction process when D. tertiolecta reacts withethanol, thereby increasing the C1 value and leading to thedecreased Cox/unox ratio of Residue-CS.

The C1 peak does not have a carbon–oxygen bond, whereas C2,C3, and C4 all possess a carbon–oxygen bond. Table 3 and Fig. 5summarize the oxygen and carbon concentrations and the abun-dance of each deconvoluted carbon signal. The relative intensitiesof the different peaks are in agreement with the chemical

composition. In different solvent conditions, the relative concen-trations of C1, C2, C3, and C4 change greatly. The peak intensitiesof the four peaks showed a decreasing trend in either Residue-CSor Residue-SS, which indicates that the carbon compounds in D.tertiolecta have been effectively transformed during liquefactionprocess. A drop in the C3 peak was clearly seen, which suggestedthat the protein in D. tertiolecta is effectively converted. Thesignificant change in C4 indicates the effective conversion of fat

Table 3C1s and O1s test data from D. tertiolecta and residue by XPS.

Species Binding energy (eV) Difference value of binding energy (eV) Relative content (%)

D. tertiolecta Residue-SS Residue-CS D. tertiolecta Residue-SS Residue-CS D. tertiolecta Residue-SS Residue-CS

C1 282.92 283.92 283.94 0.00 0.00 0.00 9.41 23.40 45.48C2 284.00 285.00 285.00 1.08 1.08 1.06 24.23 48.08 36.23C3 284.98 285.98 285.96 2.06 2.06 2.02 47.79 22.33 18.29C4 286.31 287.31 287.31 3.39 3.39 3.37 18.57 6.19 0.00O1s1 530.65 531.42 531.43 0.00 0.00 0.00 32.71 59.36 60.31O1s2 531.82 532.67 532.68 1.17 1.25 1.25 67.29 40.64 39.69C1s 75.77 63.99 57.91O1s 22.08 31.29 40.03O/C 0.22 0.37 0.51Cox/unox 9.63 3.27 1.20

Y. Chen et al. / Bioresource Technology 124 (2012) 190–198 197

in D. tertiolecta. It is interesting to note that in the liquefaction of D.tertiolecta, the C4 peak was not found in Residue-CS, which indi-cates that all fats in the raw material were transformed in the li-quid phase. That is, the ethanol–water co-solvent is the mosteffective reaction medium. The absolute value of C2 changes a littleand the relative value of C2 increases during liquefaction processusing either co-solvent of ethanol–water or sole water as the med-ium. This finding can be explained by the fact that the proteins andfats are more susceptible to liquefaction than carbohydrates andcellulose. However, the absolute value of C2 in Residue-CS isslightly lower than that of Residue-SS, which shows that the sub/supercritical ethanol–water are conducive to the liquefaction ofcarbohydrates and cellulose. Simultaneously, the XPS spectra ofthe O1s oxygen can be fitted into two main parts by the Gauss–Lorentzian line shapes: O1s1 (C–OH. . .O) and O1s2 (C–OH). Thebinding energy of C–OH. . .O is a little lower than that of C–O dueto the absorption action of the oxygen anions. The O1s1 peak cor-responds to the relative content of the crystalline regions, and theO1s2 peak corresponds to the relative content of the amorphousregion in D. tertiolecta. From Fig. 5, the relative content of O1s2 de-creases as that of O1s1 increases in Residue-SS after the liquefac-tion process compared with D. tertiolecta. This result indicatesthat the amorphous component is relatively easy to liquefy. Therelative content of O1s1 and O1s2 increased in Residue-CS com-pared with D. tertiolecta, and its absolute value was also higherthan that of D. tertiolecta. This finding may be due to the effectiveconversion of the amorphous components in D. tertiolecta in sub/supercritical ethanol–water, whereas ethanol can react with resi-due to generate ethers and other substances. These results con-firmed the results of infrared analysis and SEM.

3.7. Reaction scheme of the possible reaction pathways

Based on our results, as well as those obtained on the biomassliquefaction process in the literature, a possible reaction pathwayof D. tertiolecta liquefaction is proposed in Fig. S4 in Supplementarydata. The cellulose decomposition pathways in supercritical waterhave been previously reported by George et al. (George et al.,2006). As listed in Table 1, the main chemical components of D. ter-tiolecta are proteins, carbohydrates, fats, and cellulose. The conver-sion of D. tertiolecta into bio-oil is actually the reaction of thesemain components. A large number of different reactions can occur,and an approximate liquefaction mechanism of the main chemicalcomponents in D. tertiolecta in the sub/supercritical ethanol–wateris described.

Sub/supercritical ethanol–water are weak acid, the D. tertiolectaliquefaction is considering as an acid-catalyzed process. Underacidic conditions, proteins first form a long peptide chain, whichis then hydrolyzed to form amino acids. The amino acids undergocracking, condensation, decarboxylation, and deamination, etc. to

form a liquefied product. Cellulose and other carbohydrates under-go dehydration to form monosaccharides, a part of which may thenreact with ethanol to form the ether that exists in the solid resi-dues. Most of the monosaccharides may further react to generatecarboxylic acid or other organic compounds that undergo theacid-catalyzed process (Andrew et al., 2008). The carboxylic acidcan then react with the solvent (ethanol) via alcoholysis to formcarboxylic acid esters and to undergo ammonolysis with ammoniamolecules from the decomposition of protein to obtain amides. Theamides can also undergo alcoholysis to generate carboxylic acid es-ters when the concentration of ethanol is high. The esterificationreaction can occur for fat in the supercritical ethanol–water. Inaddition, during the acid-catalyzed decomposition of proteins,aside from carboxylic acid and its derivatives, ammonia moleculesmay also be produced, which may be used as an ammonolysisreagent in the direct liquefaction process.

Overall, the reduction of organic acids and amides, as well asthe increase in fatty acid esters, could significantly improve thebio-oil quality. Generally, the low organic acids reduce acid valueand corrosivity of the bio-oil. Similar to biodiesel, the bio-oil ob-tained in liquefaction contains numerous organic acid esters,which improve the bio-oil quality. In brief, the addition of ethanolnot only makes reaction condition mild and elevates the bio-oilyield, but also improves the quality of the bio-oil. The study pre-sents the novel approach to produce bio-oil by direct liquefactionof D. tertiolecta. Testing of the gas and water-soluble reaction prod-ucts is being carried out, and results will be reported in future.

4. Conclusions

In this study, D. tertiolecta was effectively liquefied in sub/supercritical ethanol–water. Ethanol and water showed synergisticeffects on direct liquefaction process. The optimal D. tertiolectaconversion was 98.24%; with a maximum bio-oil yield of 64.68%at a reaction temperature of 593 K, with a holding time of30 min, and an ethanol content of 40% (v/v) in sub/supercriticalethanol–water. The empirical formulas of the bio-oil obtainedusing the ethanol–water co-solvent as the reaction medium wereCH1.52O0.14N0.06 with calorific values of 34.96 MJ kg�1. Finally,XPS and SEM results showed that ethanol–water is a very effectivereaction medium in the liquefaction.

Acknowledgements

The authors are grateful for the financial supports from NationalNatural Science Foundation of China (No. 21176142), Science andPetroChina Innovation Foundation (No. 2010D50060406), Inde-pendent Research Programs of Tsinghua University (No.2011Z08141), and National Key Technology R&D Program (No.2011BAD14B01).

198 Y. Chen et al. / Bioresource Technology 124 (2012) 190–198

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.biortech.2012.08.013.

References

Andrew, A.P., Frédéric, V., Russell, P.L., Morgan, F., Michael, J.A.J., Jefferson, W.T.,2008. Thermochemical biofuel production in hydrothermal media: a review ofsub- and supercritical water technologies. Energy Environ. Sci. 1, 32–65.

Cheng, S.N., Dcruz, I., Wang, M.C., 2010. Highly efficient liquefaction of woodybiomass in hot-compressed alcohol–water co-solvents. Energy Fuels 24, 4659–4667.

Chumpoo, J., Prasassarakich, P., 2010. Bio-oil from hydro-liquefaction of bagasse insupercritical ethanol. Energy Fuels 24, 2071–2077.

Clarens, A.F., Resurreccion, E.P., White, M.A., Colosi, L.M., 2010. Environmental lifecycle comparison of algae to other bioenergy feed stocks. Crit. Rev. Environ. Sci.Technol. 44, 1813–1819.

Duan, P.G., Phillip, E.S., 2011. Hydrothermal liquefaction of a microalga withheterogeneous catalysts. Ind. Eng. Chem. Res. 50, 52–61.

George, W.H., Sara, I., Avelino, C., 2006. Synthesis of transportation fuels frombiomass: chemistry, catalysts, and engineering. Chem. Rev. 106, 4044–4098.

Huang, H.J., Yuan, X.Z., Zeng, G.M., Wang, J.Y., Li, H., Zhou, C.F., Pei, X.K., You, Q.,Chen, L., 2011. Thermochemical liquefaction characteristics of microalgae insub- and supercritical ethanol. Fuel Process. Technol. 92, 147–153.

Jade, C., Pattarapan, P., 2010. Bio-oil from hydro-liquefaction of bagasse insupercritical ethanol. Energy Fuels 24, 2071–2077.

Li, H., Yuan, X.Z., Zeng, G.M., Tong, J.Y., Yan, Y., Cao, H.T., Wang, L.H., Cheng, M.Y.,Zhang, J.C., Yang, D., 2009. Liquefaction of rice straw in sub- and supercritical1,4-dioxane–water mixture. Fuel Process. Technol. 90, 657–663.

Li, H., Yuan, X.Z., Zeng, G.M., Huang, D.L., Huang, H.J., Tong, J.Y., You, Q., Zhang, J.C.,Zhou, M., 2010. The formation of bio-oil from sludge by deoxy-liquefaction insupercritical ethanol. Bioresour. Technol. 101, 2860–2866.

Liu, Z., Zhang, F.S., 2008. Effect of various solvents on the liquefaction of biomass toproduce fuels and chemical feedstocks. Energy Convers. Manage. 49, 3498–3504.

Matsumura, Y., Nonaka, H., Yokura, H., Tsutsumi, A., Yoshida, K., 1999. Co-liquefaction of coal and cellulose in supercritical water. Energy Fuels 78,1049–1056.

Matuana, L.M., Kamdem, D.P., 2002. Accelerated ultraviolet weathering of PVC/wood-flour composites. Polym. Eng. Sci. 42, 1657–1666.

Nzokou, P., Kamdem, D.P., 2005. You have free access to this content X-rayphotoelectron spectroscopy study of red oak- (Quercus rubra), black cherry-(Prunus serotina) and red pine- (Pinus resinosa) extracted wood surfaces. Surf.Interface Anal. 37, 689–694.

Patil, V., Tran, K.Q., Giselrod, H.R., 2008. Toward sustainable production of biofuelsfrom microalgae. Int. J. Mol. Sci. 9, 1188–1195.

Pereira, S.L., Leonard, A.E., Huang, Y.S., Chuang, L.T., Mukerji, P., 2004. Identificationof two novel microalgal enzymes involved in the conversion of the omega 3-fatty acid, eicosapentaenoic acid, into docosahexaenoic acid. Biochem. J. 384,357–366.

Qu, Y.X., Wei, X.M., Zhong, C.L., 2003. Experimental study on the direct liquefactionof Cunninghamia lanceolata in water. Energy 28, 597–606.

Sakaki, T., Shibata, M., Sumi, T., Yasuda, S., 2002. Saccharification of cellulose using ahot-compressed water-flow reactor. Ind. Eng. Chem. Res. 41, 661–665.

Sarmidi, A., 2009. Review on biofuel oil and gas production processes frommicroalgae. Energy Convers. Manage. 50, 1834–1840.

Savage, P.E., 1999. Organic chemical reactions in supercritical water. Chem. Rev. 99,603–621.

Sernek, M., Kamke, F.A., Glasser, W.G., 2004. Comparative analysis of inactivatedwood surfaces. Holzforschung 58, 22–31.

Xu, C., Etcheverry, T., 2008. Hydro-liquefaction of woody biomass in sub andsupercritical ethanol with iron-based catalysts. Energy Fuels 87, 335–345.

Yamazaki, J., Minami, E., Saka, S., 2006. Liquefaction of beech wood in varioussupercritical alcohols. J. Wood Sci. 52, 527–532.

Yu, G., Zhang, Y.H., Schideman, L., Funk, T., Wang, Z.C., 2011. Distributions of carbonand nitrogen in the products from hydrothermal liquefaction of low-lipidmicroalgae. Energy Environ. Sci. 4, 4587–4595.

Yuan, X.Z., Li, H., Zeng, G.M., Tong, J.Y., Xie, W., 2007. Sub- and supercriticalliquefaction of rice straw in the presence of ethanol–water and 2-propanol–water mixture. Energy 32, 2081–2088.

Zhou, D., Zhang, S.C., Fu, H.B., Chen, J.M., 2012. Liquefaction of macroalgaeenteromorpha prolifera in sub-/supercritical alcohols: direct production ofester compounds. Energy Fuels 26, 2342–2351.

Zou, S.P., Wu, Y.Y., Yang, M.D., Imdad, K., Li, C., Tong, J.M., 2010a. Production andcharacterization of bio-oil from hydrothermal liquefaction of microalgaeDunaliella tertiolecta cake. Energy 35, 5406–5411.

Zou, S.P., Wu, Y.L., Yang, M.D., Li, C., Tong, J.M., 2010b. Pyrolysis characteristics andkinetics of the marine microalgae Dunaliella tertiolecta using thermogravimetricanalyzer. Bioresour. Technol. 101, 359–365.

Zou, S.P., Wu, Y.L., Yang, M.D., Li, C., Tong, J.M., 2010c. Bio-oils production from sub-and supercritical water liquefaction of microalgae Dunaliella tertiolecta andrelated properties. Energy Environ. Sci. 3, 1073–1078.