2nd manuscript

CatalysisScience &Technology

PAPER

Cite this: DOI: 10.1039/c5cy01307k

Received 9th August 2015,Accepted 27th September 2015

DOI: 10.1039/c5cy01307k

www.rsc.org/catalysis

Hydrodeoxygenation of microalgae oil to greendiesel over Pt, Rh and presulfided NiMo catalysts

Lin Zhou* and Adeniyi Lawal

The catalytic characteristics, activity and selectivity of 1% Pt/Al2O3, 0.5% Rh/Al2O3 and presulfided NiMo/

Al2O3 catalysts have been investigated in the hydrodeoxygenation of microalgae (Nannochloropsis salina)

oil to produce green diesel in a microreactor. Coke accumulation decreased in the order NiMo > Pt > Rh.

The amount of formed coke over NiMo increased with reaction duration, while no on-stream time depen-

dence was found over Pt and Rh. Rhodium was found to be very active for CH4 production via hydrocrack-

ing at its freshly reduced state. The activity and selectivity of all three investigated catalysts were positively

affected by increased reaction pressure, temperature, H2/oil ratio and residence time. The selectivity of

NiMo for the hydrodehydration (DHYD) route was changed to a hydrodecarbonylation/hydro-

decarboxylation (DCO/DCO2) route at a reduced H2/oil ratio and residence time, while the selectivity of Pt

and Rh for the DCO/DCO2 route was not affected by the reaction conditions. The highest hydrocarbon

yield, 76.5%, was obtained over 1% Pt (310 °C, 500 psig, 1000 SmL mL−1 gas/oil ratio, 1.5 s residence time),

which is 13.8% higher than that over NiMo (360 °C, 500 psig, 1000 SmL mL−1 gas/oil ratio, 1 s residence

time). The 50 °C decrease in the reaction temperature for Pt indicates possible energy saving via heat

supply.

1. Introduction

Over the next few decades, as economies expand and evolve,the global energy demand for transportation is expected tocontinue to grow significantly. It is estimated that the globalenergy demand for commercial transportation will rise by 70percent from 2010 to 2040. Due to its affordability, availabil-ity, portability and high energy density, liquid fuel, includinggasoline, diesel, jet fuel and fuel oil, will remain the energy ofchoice for most types of transportation. Seventy-five percentof the heavy-duty transportation energy requirements are metby diesel fuel, due to its widespread availability and therobust ability of diesel engine technology to handle heavyloads. With an estimate of about 80 percent growth in heavy-duty transportation fuel requirements, the demand for dieselis expected to grow sharply by about 75 percent from 2010 to2040.1 Petroleum diesel is refined from crude oil, and approx-imately 12 gallons of diesel fuel can be produced from each42 gallon barrel of crude oil.2 The CO2 emission from dieselcan be calculated based on its carbon content, 2778 g per gal-lon of diesel, which is 10 084 g of CO2 per gallon of diesel.3

Therefore, the combination of continued crude oil depletion,

the ever-growing need for renewable energy resources and theCO2 emission problem is a motivation for research and devel-opment into advanced biofuels.

It has been reported that oleaginous feedstock, the maincomponents of which are triglycerides and free fatty acids,has higher energy content than lignocellulose.4 Microalgae,as a source of oleaginous feedstock for green diesel produc-tion, is currently attracting the attention of researchers.Green diesel, produced from hydrotreatment of triglycerides,is chemically similar to petroleum diesel with excellent stabil-ity and cold-flow properties, and can be blended with petro-leum diesel in any proportion. It can be transported inexisting infrastructure and used in diesel engines withoutany modification. Microalgae, which provide the route forcollection, conversion and storage of solar energy in the formof oxygenated hydrocarbon compounds, have been consid-ered as potential candidates for fuel production.5 The utiliza-tion of microalgae for green diesel production will not com-promise the food and oil production from land crops. Energyis captured by microalgae through CO2 fixation via photosyn-thesis, which will help to maintain and improve the air qual-ity. Moreover, because the nutrients in wastewater (especiallynitrogen and phosphorus) could be exploited by algaegrowth, algae cultivation can also be coupled with wastewatertreatment.5

The sulfur, nitrogen, and phosphorus contents in micro-algae oil are relatively low, and the main heteroatom needed

Catal. Sci. Technol.This journal is © The Royal Society of Chemistry 2015

New Jersey Center for Microchemical Systems, Dept. of Chemical Engineering and

Materials Science, Stevens Institute of Technology, 1 Castle Point on Hudson,

Hoboken, NJ 07030, USA. E-mail: [email protected]; Fax: +1 201 216 8306;

Tel: +1 201 216 8314

Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article OnlineView Journal

http://crossmark.crossref.org/dialog/?doi=10.1039/c5cy01307k&domain=pdf&date_stamp=2015-10-12

http://dx.doi.org/10.1039/c5cy01307k

http://pubs.rsc.org/en/journals/journal/CY

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2015

to be removed to produce green diesel is oxygen. Triglycer-ides constitute the major component (>80%) of algae oil,and due to the nature of oxygen bonding in triglycerides, oxy-gen is proposed to be removed by adding hydrogen throughhydrodeoxygenation (HDO), which occurs via two routes: (1)hydrodehydration (DHYD), in which oxygen is removed inthe form of H2O, and (2) hydrodecarboxylation (DCO2) orhydrodecarbonylation (DCO), in which oxygen is removed inthe form of CO2 and CO, respectively.6

Catalyst selection is crucial in the hydrotreatment step,due to its role in controlling the yield and selectivity to differ-ent products, which in turn impact both capital and operat-ing costs. Because the oxygen content in crude oil is at ratherlow levels, on the order of 0.5%, not much concern is givento HDO in petroleum refining. Therefore, no hydrotreatingcatalysts have been specially formulated for HDO of oleagi-nous feedstock with much higher oxygen content than crudeoil. Thus, one of the critical challenges posed by HDO ofalgae oils is the development of specific catalysts to make theprocess economical.

Compared to non-selective acidic catalysts, which give abroad distribution of hydrocarbons, supported metal cata-lysts in their reduced or sulfided state are more commonlyused in industry because they produce a narrow range ofhydrocarbons.7 In conventional petroleum refining, the sul-fides of transition metals, Mo or W, promoted with Co or Nimetal sulfides have been utilized in hydrotreating units toremove sulfur (hydrodesulfurization, HDS), nitrogen (hydro-denitrogenation, HDN), metals (hydrodemetalation, HDM)and oxygen (hydrodeoxygenation, HDO) from the heavy gasoil feedstock.4 These conventional hydrotreating catalystshave been studied in the HDO process.8–14 The sulfided NiMocatalyst is found to be more active than the sulfided CoMocatalyst in hydrotreatment of oxygen-containing compounds,such as fatty acid methyl esters (FAME).15,16 However, it hasbeen reported that the leaching of sulfur from the sulfidedcatalyst surface will cause the deactivation of the catalyst andcontaminate the product.7,17–19 Water produced from DHYDhas been found to accelerate sulfur leaching and decreasethe activity, thus affecting the selectivity of both sulfidedNiMo and CoMo catalysts.16 Moreover, the conventionalhydrotreating catalyst was found to prefer the DHYD route,which requires higher H2 consumption to producehydrocarbons.

Supported noble metal-based catalysts have been investi-gated in the reduced state, with palladium (Pd) being themost studied.20 Pd-based catalysts were found to prefer theDCO2 route, which does not require a stoichiometric level ofH2 consumption to produce renewable diesel. The best cata-lytic performance was obtained from a commercial Pd/C cata-lyst in the HDO of stearic acid in a continuous reactor at 360°C. The conversion of steric acid after a time-on-stream of 92h (5500 min) was about 40%, although it was 90% at thebeginning of the experiment.21 The rapid deactivation of thePd/C catalyst was also reported in the HDO of lauric acid at255–300 °C.21 The deactivation of the Pd/C catalyst was

caused by a combination of coking and poisoning by the pro-duced gases, mainly CO and CO2.

21 Other supported preciousmetal catalysts, including Pt, Rh, Ru and Ir, have also beenstudied mostly for the HDO of model compounds, such asstearic acid, ethyl stearate and tristearin, in batch or semi-batch reactors.22–24 The lack of experimental data on the per-formance of replacement catalysts for sulfided NiMo, espe-cially the precious metal-based catalysts, in the HDO ofactual algae oil feedstock in a continuous flow reactor is oneof the key technical challenges that needs to be addressed.

In the present study, the hydrodeoxygenation of micro-algae oil extracted from the Nannochloropsis salina strain, theworkhorse of the algae industry, was conducted in a micro-reactor with excellent mass transfer characteristics. There is adearth of experimental data in the literature on the hydro-deoxygenation of microalgae oil, and our overall goal was tocontribute reliable data that can be used in pilot studies in aprelude to industrial scale-up. In relation to that, micro-reactors provide ease of scale-up through the numbering-upapproach which has been demonstrated elsewhere.14,25–30

Three catalysts, 1% Pt/Al2O3, 0.5% Rh/Al2O3 and presulfidedNiMo/Al2O3, were investigated in terms of their catalytic char-acteristics, activity and selectivity by analyzing the gas andliquid products. Comparison of their performance will guidefurther catalyst formulation and process scale-up andoptimization.

2. Experimental section2.1 Materials

Reactants. Hydrogen (5.0 grade, ultra-high purity), nitro-gen (5.0 grade, ultra-high purity), and air (5.0 grade, ultra-high purity) were all purchased from Praxair.

In order to make the microalgae oil-to-green diesel processcommercially feasible, the investment required by all aspectsof the product chain needs to be taken into consideration.Due to its high lipid content (up to 70% of dry weight) andrelatively high biomass yield, the microalgae oil extractedfrom Nannochloropsis salina and provided to us by ValicorRenewables, LLC Inc. was used in this present work. Unlikevegetable oils which are essentially 100% acylglycerides, only47.45% of the Nannochloropsis salina oil was identified to beacylglycerides and free fatty acids, with the remainder catego-rized as an unidentified portion. The quantification resultscan be found elsewhere,14 and the corresponding fatty acid(FA) profile as determined by Valicor is given in Table 1. Thealgae oil is in the form of a semisolid at room temperature;therefore, a solvent is needed to process the oil in a continu-ous reactor. One gram of algae oil was dissolved in 100 mL ofn-dodecane (99+%, Alfa Aesar) to make a 1.3 wt% solution,and the solution was vacuum-filtered through a grade 410 fil-ter paper (VWR, 1 μm retention) to remove any residual parti-cles (>10 wt% of the oil) before feeding into the reactor.Although the potential of blockage to the reactor system islow for an industrial reactor, the residual particles that accu-mulate over a period of time and are adsorbed on the catalyst

Catalysis Science & TechnologyPaper

Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



surface can block the active sites or the pores of the catalyst,which will reduce the apparent catalyst activity and result inlow system productivity. It should also be mentioned that,based on our experience with microalgae oil extracted fromother algae species, such as Chlorella which has negligibleresidual solid content and flows freely at room temperature,the solvent and vacuum-filtration pretreatment process maynot be needed. However, for industrial implementation ofour current process, dodecane can be replaced with middledistillates from refinery, such as the light atmospheric gas oil(LAGO), and this will not affect the process economics signif-icantly. The compatibility of all process equipment and pipe-lines with the properties of algae oil, especially the corrosiveproperties, needs to be considered when processing purealgae oil, as constraints imposed by material selection canincrease both the capital and operating investment. The costof vacuum filtration is negligible in comparison to the cost ofalgae oil production, which comprises mainly algae cultiva-tion and oil extraction. This is the cost that determines theeconomic feasibility of the whole biofuel production process.Therefore, the co-processing of Nannochloropsis salina algaeoil with a solvent after vacuum filtration is proposed as ourcurrent process for green diesel production.

Catalyst. The NiMo/γ-Al2O3 catalyst was provided byAlbemarle, Houston, TX (presulfided and supplied byEurecat, USA). A typical composition of this commercialhydrotreating catalyst is 4.03% NiO and 13.20% MoO3 on adry basis.31

Commercial catalysts, 1% Pt and 0.5% Rh, were purchasedfrom Alfa Aesar, a Johnson Matthey company. The two cata-lysts were all supported on 2.7–3.3 mm γ-Al2O3 pellets, andwere ground and sieved to a 75–150 μm particle size fractionbefore packing into the reactor. Previous studies25–27,29,30

have revealed that internal diffusion has negligible influenceon the reaction rate for this catalyst particle size range. Thetwo catalysts have the same bulk density, and the molar massof Rh (102.90 g mol−1) is nearly half that of Pt (195.09 gmol−1). Therefore, the number of Pt atoms is the same as thatof Rh for the same volume of catalyst.

2.2. Experimental setup and procedure

As shown in Scheme 1, the hydrodeoxygenation of microalgaeoil was conducted in a continuous flow microreactor, theinner diameter of which was 0.762 mm. All three investigatedcatalysts were ground and sieved to 75–150 μm before being

Table 1 Fatty acid profile of microalgae oila

Fatty acid profile (wt%)

C14:0b C14:1 C16:0 C16:1n7 C18:0 C18:1-cis C18:1-trans C18:2 C18:3 C20:0

2.19 0.04 12.21 15.91 0.41 3.05 1.10 0.96 0.04 0.06C20:1 C20:2n6 C20:4n6 C20:3n3 C20:5n3 C22:0 C21:5n3 C24 C24:1 Other0.03 0.38 1.43 3.52 7.66 0.07 0.05 0.10 0.05 0.04

a Microalgae oil is extracted from Nannochloropsis salina and provided by Valicor Renewables, LLC Inc. b The nomenclature shows the numberof carbon atoms, the number of CC double bonds, and the position of the CC double bond.

Scheme 1 Schematic of microalgae oil hydrodeoxygenation setup.

Catalysis Science & Technology Paper

Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



packed into the reactor. At the entrance and exit of the reac-tor, a 1/8 inch outer diameter tube was connected to the reac-tor and was packed with glass beads of 90 μm. To retain thepacked particles in each tube section, a Hastelloy micron fil-ter cloth (200 × 1150 meshes, Unique Wire Weaving Co., Hill-side, NJ) was placed inside each union between the 1/8 inchtube and the reactor. The reactor was placed in an electricfurnace with digital temperature control and readout, andthe temperature of the furnace was also measured by a hand-held temperature meter for confirmation. The pressure at theentrance and exit of the reactor was measured by two pres-sure transmitters. The system pressure was adjusted by aback pressure regulator (BPR) installed downstream. Gas flowwas regulated by a mass flow controller (sccm, standardcubic centimeter per minute), which was calibrated with anAgilent ADM 2000 universal gas flowmeter routinely.According to previous studies, introduction of gas and liquid

flows head-to-head generates a short slug length whichenhances mixing in two-phase flows;28 therefore, the liquidfeed pumped by an HPLC pump (mL min−1) first made con-tact with the gas flow at the T-mixer before entering the reac-tor. Due to the low product flow rate, a sample loop wasconnected between the reactor exit and the BPR, and the liq-uid product was collected every hour under ambient condi-

tions for off-line analysis. The remaining product stream con-tinued to the gas–liquid separator, where the liquidcondensed and the gas proceeded through an inline filter tothe GC-TCD for analysis.

In situ reduction of fresh catalysts was carried out withpure hydrogen flow at 310 °C and 500 psig for 2 hours beforestarting any reaction run. In a microreactor, it was difficult torecover the used catalyst and clean it with solvent. In order toremove any liquid or solid residues from the catalyst that mayaffect the coke formation result, the system was purged withcompressed nitrogen. Therefore, the “coke” mentioned insection 3.3 was referred to as “all carbonaceous residues thatcannot be removed through nitrogen purging”. For gas prod-uct analysis (sections 3.4 and 3.5), nitrogen which was fedinto the system as a mixture with the hydrogen reactant wasused as an internal standard. The composition of the nitro-gen/hydrogen mixture was also confirmed by GC-TCD beforestarting each reaction.

2.3. Product analysis

Liquid product analysis. As shown in Table 1, C14, C16,C18, and C20 fatty acids are the main components of the

microalgae (Nannochloropsis salina) oil fatty acid profile.According to the HDO mechanism mentioned above, thediesel-range alkanes, C13 to C20 hydrocarbons, were thefocus of this study. To identify the components of the liquidproduct and quantify them, the liquid product was directlyinjected by a CP-8400 autosampler into a Varian 450 gaschromatograph (GC) equipped with a flame ionization detec-tor (FID) under the conditions stated below:

Column: ZB-1HT nonpolar capillary column (30 m × 0.25mm × 0.25 μm)

Oven: 50 °C (1 min), 15 °C min−1 to 240 °C (5 min)Injector: 250 °CDetector: FID, 280 °CCarrier gas: heliumInjection: 1 μ, 1 : 20 splitBefore analyzing each batch of samples, GC external cali-

bration was performed with 4 standards containing C13–C20

hydrocarbons of different concentrations to obtain theresponse factor and retention time of each hydrocarbon. TheC13–C20 hydrocarbon yield was used to evaluate the catalystactivity, and was calculated based on the quantified FA(47.45%) in the liquid feed.

The even-numbered carbon hydrocarbon to odd-numbered carbon hydrocarbon ratio (HCĲ2n)/HCĲ2n − 1)) is a

reflection of catalyst selectivity, and is defined as:The space-time yield (STY) indicates the rate of hydrocar-

bon formation, and is defined as:

(3)

Gas product analysis. To study the coke formation (section3.3), methane formation (section 3.4) and hydrogen con-sumption (section 3.5), quantification of H2, N2, CH4, andCO2 was of vital importance. The gas sample was injectedinto a Shimadzu GC-14B through a 10-port sampling valveand analyzed using a thermal conductivity detector (TCD).The sample eluted by argon (Ar) first flowed through the HP-PLOT Q column and then the HP-Molsieve column. The tem-perature program in the column oven was initially held at 60°C for 13.5 min, then ramped at 25 °C min−1 to 160 °C andheld for 6.5 min. After each experiment, the GC was heatedat 220 °C for 4 hours to get rid of any liquid residue from theflow system. In the coke formation determination, pure airwas fed into the system, and the nitrogen contained thereinwas used as the internal standard to quantify CO2

(1)

(2)


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



production. In the methane formation and hydrogen con-sumption study, the added nitrogen (1 sccm, 20 mol%) wasused as the internal standard to account for the gas flow ratechange during the reaction.

3. Results and discussion3.1. Analysis of external mass transfer limitation

In the present study, all the catalyst evaluation experimentswere performed in the Taylor flow regime, consisting of alter-nating liquid slugs and gas bubbles. For this heterogeneousreaction of microalgae oil with hydrogen on the γ-aluminasupported catalyst, the main external mass transfer resis-tances occur at the gas–liquid and liquid–solid interfaces,with the former being predominant. The mass transfer ratesthrough interfaces depend greatly on flow velocity; therefore,the overall external mass transfer limitation can be investi-gated experimentally by studying the change in hydrocarbonyield rate at different superficial flow velocities. The otherprocess conditions, including the reactor ID, gas-to-oil ratio,residence time, and reaction temperature and pressure, needto be kept the same for all experiments, with the constantresidence time achieved by varying the reactor length, inother words the amount of the catalyst. Fig. 1 shows that theSTY of C13 to C20 hydrocarbons is independent of the super-ficial flow velocity within the studied range. This indicatesthe absence of external mass transfer limitation beyond avelocity of 0.117 m s−1, which was chosen as the baselinevelocity for all subsequent performance studies. The samenon-external mass transfer-controlled flow velocity range hasalso been reported with respect to the presulfided NiMo cata-lyst.14 Since the external mass transfer is not catalyst speciesdependent, the flow velocity range without external masstransfer limitation is expected to be the same for the Rh cata-lyst, thus the performance studies over the Rh catalyst werealso conducted in the same range.

3.2. Analysis of internal mass transfer limitation

Internal mass transfer refers to the diffusion of reactantsfrom the catalyst pellet pore entrance into the pore. Reac-tions are limited by internal mass transfer when there is aconcentration gradient within the catalyst, with the concen-tration inside the pores being much lower than that at theentrance. In order to account for variations in concentrationthroughout the catalyst pellet, the Weisz–Prater parameter(Cwp) is introduced.

32

(4)

where− r′AĲobs): observed reaction rate, 9.22 × 10−5 ghydrocarbon gcat

−1 s−1,ρp: density of catalyst pellets, 0.29 gcat cm

−3,R: characteristic diameter of the catalyst pellets, 112.5 ×

10−4 cm,CAS: hydrogen concentration at the external surface of the

catalyst pellets, which is the hydrogen solubility in dodecane,approximately 360 g m−3, and

De: effective diffusivity of hydrogen in the catalyst pellets,De = DABϕPσC/τ, where DAB: binary diffusivity of hydrogen indodecane, 1.12 × 10−7 m2 s−1,33 ϕP: porosity of the catalyst pel-lets, 0.4–0.6, σC: catalyst constriction factor, 0.7–0.8, and τ:catalyst tortuosity, 2–8.32 Then, the De is calculated to bebetween 2.69 × 10−8 and 3.92 × 10−9 m2 s−1.

Based on the ranges of the catalyst porosity, constrictionfactor, and tortuosity, the calculated Weisz–Prater parameterwas between 0.240 and 0.035. According to the Weisz–Pratercriterion, CWP ≪ 1 means that there is no concentration gra-dient within the pellet, and consequently no diffusion limita-tions existed in our experiments.

3.3 Coke formation

In the HDO of microalgae oil, the carbon lost to the gasphase, including CO, CO2, CH4 and light hydrocarbons, andsolid phase, mainly coke, will affect the carbon recovery inthe liquid product. Moreover, coke formation during the cata-lytic upgrading of carbon-rich feedstock is one of the majorcauses of catalyst deactivation. Therefore, the extent of cokeformation in the HDO of microalgae oil over Pt, Rh and NiMowas first studied. The definition of what constitutes coke inliquid-phase reactions, especially for liquids with a high boil-ing point, depends on the coke quantification method, and isquite different from that in gas-phase reactions.34 In thepresent study, coke is referred to as all carbonaceous residuesthat cannot be removed through purging with gas (N2).

In order to investigate coke accumulation with time-on-stream, the HDO of microalgae oil was conducted for 10 h,20 h and 30 h using Pt, Rh and NiMo under the same processconditions, starting each experimental run with reducedfresh catalysts. Upon completion of the HDO reaction, thereactor system was purged with nitrogen to get rid of anyremaining oil from the catalyst surface. Because it is difficultto directly quantify the formed coke accurately in a micro-reactor, alternatively, the amount of coke can be determined

Fig. 1 Effect of superficial flow velocity on C13 to C20 hydrocarbonspace-time yield (1% Pt, 1.32 wt% Nannochloropsis salina oil indodecane, temperature = 310 °C, H2 pressure = 500 psig, residencetime = 1 s, H2/oil = 1000 SmL mL−1).


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



through coke burn-off (oxidization). The concentration(mol%) of produced CO2 in the combustion gas product is anindication of the extent of coke formation during the HDOreaction. The trend of coke formation during different reac-tion periods over the 3 catalysts is depicted in Fig. 2, with theslope of each trend line representing the coke formation rate.Under the experimental conditions, the coke formationdecreased in the following order for all studied HDO reactiontimes-on-stream: NiMo > 1% Pt > 0.5% Rh. The amount ofcoke accumulated on the presulfided NiMo catalyst increasedwith HDO time-on-stream, while no time dependence of cokeaccumulation was observed on the Pt- and Rh-based catalysts.This result for Pt is in agreement with previous studies onthe hydrogenation of vegetable oils conducted in a batchreactor using a different coke quantification method, whichreported that the formation of coke precursors on Pt was veryslow.34 The amount of coke formed and the rate of formationof coke for NiMo are always much higher than the values forthe two precious metal catalysts studied; therefore, theamount of carbon lost is higher for NiMo than those for Ptand Rh. However, because dilute algae oil solution (1.32wt%) was used in this study, the actual amount of cokedeposit was small, and no obvious reduction in C13 to C20hydrocarbon yield was observed over 30 hours of reaction. Itshould be noted that, in industrial practice, although a sig-nificantly larger amount of catalyst is used to process higheralgae oil concentration feedstock, the accumulation of cokedeposit with reaction time on the NiMo catalyst may lead togradual catalyst deactivation. This will result in the decreasein oxygen removal efficiency as well as carbon recovery. Inaddition, a process involving more frequent regeneration willbe required by NiMo to restore the catalyst activity and selec-tivity, in comparison to the precious metal-based catalysts.

3.4 Methane formation

Methane is known to be produced through methanation (CO+ 3H2 → CH4 + H2O) and hydrocracking in the HDO process.

In addition to the reduction of carbon recovery in the liquidphase, the formation of gaseous products will also negativelyaffect the hydrogen partial pressure, which has been reportedto potentially reduce the catalyst activity.35 Moreover, the for-mation of light hydrocarbons, especially CH4, of which theH/C ratio is the highest for all hydrocarbons, will increasethe hydrogen consumption, and consequently render thewhole upgrading process less economically effective. It isnoteworthy that the reduction in the formation of methane,which is one of the greenhouse gases (GHG) with a globalwarming potential (GWP) 23 times higher than that of CO2,will significantly provide environmental benefits.36

The yield of CH4 per gram of algae oil processed over the3 catalysts is depicted in Fig. 3. NiMo showed the lowestactivity for CH4 formation, and no CH4 was detected in thegas product. The precious metal-based catalyst showed highactivity in CH4 production initially, which quickly decreasedwith reaction time-on-stream. A similar result had beenreported in a study of Pt.37 For the first 7 h, Rh was muchmore active than Pt in producing CH4, but less active after 7h. After 16 h, there was no CH4 detectable in the Rh gas prod-uct. It should be also noted that, during the first 5 hours ofreaction over Rh, the CH4 yield was much higher than themaximum theoretical CH4 production through methanationfrom CO, which indicated that CH4 was mainly formedthrough hydrocracking. Analysis of the liquid productshowed that more cracking products, i.e., hydrocarbons thatcontain less than 12 carbons, were found in the Rh product,but decreased with time, which validated the hydrocrackingactivity of the freshly reduced Rh catalyst. By examining thetrend in the change of CH4 formation with reaction time, itis expected that the amount of produced methane will stabi-lize below the detection limit of GC-TCD over all three cata-lysts, but longer time is needed to reach the stability whenusing Pt. Therefore, in industrial practice, where the catalystlife is expected to be much longer than 30 hours (typically>24 months), methane formation will not be much of a con-cern in the HDO of algae oil for all three studied catalysts.

Fig. 2 CO2 concentration in combustion product gas (reaction: 80 mgcatalyst, 1.32 wt% Nannochloropsis salina oil in dodecane, 0.05 mlmin−1, temperature = 360 °C, H2 pressure = 500 psig, H2/oil = 100 Smlml−1; combustion: 5 sccm air, 650 °C).

Fig. 3 CH4 formation per gram of microalgae oil processed (80 mgcatalyst, 1.32 wt% Nannochloropsis salina oil in dodecane, 0.05 mlmin−1, temperature = 360 °C, reaction pressure = 500 psig, H2 flowrate = 4 sccm, N2 flow rate = 1 sccm).


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



3.5. Hydrogen consumption

Hydrogen consumption based on per gram of algae oilprocessed over the NiMo, Pt and Rh catalysts has been inves-tigated experimentally, and the results are shown in Fig. 4.Hydrogen consumption follows the same trend as CH4 forma-tion over Rh up to 7 hours (Fig. 3). For 1 mole of CH4 pro-duced, 3 moles of H2 will be consumed through methanationof CO, while only 1 mole of H2 is demanded for hydrocrack-ing. However, the H2 consumption over Rh was not shown tobe 3 times that of the formed CH4. Therefore, it again con-firms that Rh was initially very active for hydrocracking toproduce CH4. After 16 h, the H2 consumption over Rh stabi-lized around 24 × 10−3 mol of H2 per g of AO. For the NiMoand Pt catalysts, H2 consumption increased with time ini-tially and stabilized around 40 × 10−3 mol of H2 per g of AOand 30 × 10−3 mol of H2 per g of AO, respectively. It shouldbe noted that CH4 formation over Pt was mainly throughhydrocracking initially, which consumed less hydrogen thanthe HDO reaction. The stabilized H2 consumption over Pt isslightly higher than that over Rh due to its slightly higherproduction of CH4 mainly through methanation. NiMo con-sumed much more H2 than Pt and Rh, because the DHYDroute, which is favored by NiMo, demands more H2 than theDCO route, the path favored by Pt and Rh. Therefore, due tothe difference in selectivity, the precious metal catalysts con-sume less hydrogen in removing the same amount of oxygen,though 1 mole of carbon will be sacrificed in the form of CO,CO2 and CH4 lost into the gas phase for every 2 moles ofhydrogen saved.

As previously reported,14 the content of fatty acids includ-ing monoacylglycerides (MAG), diacylglycerides (DAG), triacyl-glycerides (TAG) and free fatty acids (FFA) in Nannochloropsissalina is 47.45 wt%, and the corresponding fatty acid profileis given in Table 1. The mechanism of the HDO of FA hasbeen fully illustrated and the detailed reaction pathway hasbeen provided elsewhere.6,38 Based on the reaction pathways,the hydrogen demand for each step of the reaction is listedin Table 2. Oxygen removal through the hydrodehydration

route involves reactions 1–5, while that through the hydro-decarbonylation route involves reactions 1–3 and 6. In con-junction with the fatty acid profile in Table 1, the theoreticalamounts of H2 required to remove oxygen from only the acyl-glyceride and free fatty acid portions of algae oil through theDHYD and DCO routes were calculated to be 9.5 × 10−3 molof H2 per g of AO and 4.3 × 10−3 mol of H2 per g of AO,respectively. However, the measured amounts of stabilizedhydrogen consumption over the NiMo and precious metal-based catalysts were higher than twice the theoretically calcu-lated values. Therefore, oxygen removal from the unidentifiedportion of algae oil consumed much more hydrogen thanthat from the identified acylglyceride and free fatty acid por-tions. However, the oxygen content and degree ofunsaturation are only slightly higher in the latter.14 It indi-cates that the HDO mechanism of the unidentified portion ofNannochloropsis salina oil is different from that of theacylglycerides.

3.6. Effect of reaction pressure

In order to study the effect of reaction pressure on catalystactivity and selectivity, experiments were conducted in thepressure range of 300 to 500 psig. Other process parametersincluding the reaction temperature, H2/oil ratio, fluid flowrates and weight hourly space velocity (WHSV) were kept thesame. Since the actual gas volumetric flow rate was at least62 times that of the liquid volumetric flow rate, the residencetime was mainly a function of gas flow rate. Therefore, due tothe pressure effect on the actual volumetric flow rate of gas,the residence time increased with the increase in reactionpressure. In addition, the concentration of dissolved H2 wasalso positively affected by the pressure increase.

As shown in Fig. 5(a) and (b), the hydrocarbon yieldincreases with an increase in reaction pressure during the 7hours of reaction, which shows the positive effect ofdissolved H2 concentration and residence time on the activ-ity of the precious metal-based catalysts. The initial hydro-carbon yield at 500 psig was around 60% for Pt, Rh andNiMo;14 however, unlike the NiMo catalyst whose initial cat-alytic activity was not affected by the reaction pressure,14 theinitial activity of Pt and Rh both increased with reactionpressure. It can be observed that the hydrocarbon yield overPt and Rh kept decreasing during the first 4 hours of run,and then stabilized; however, no stabilized hydrocarbonyield was observed in the results for NiMo, and the decreas-ing rate of NiMo activity kept increasing with reaction time-on-stream, which indicated the occurrence of deactivationpossibly due to the accumulation of by-products, such asoxygenated intermediates and coke.14 It is seen that theincrease in reaction pressure has a negative effect on thestability of activity of the precious metal-based catalysts. Itcould be due to the difficulty in desorption of product mole-cules from the catalyst surface at high reaction pressure.Moreover, as mentioned above, due to Rh's initial activity inhydrocracking, which produced a substantial amount of

Fig. 4 Hydrogen consumption (80 mg catalyst, 1.32 wt%Nannochloropsis salina oil in dodecane, 0.05 ml min−1, temperature =360 °C, reaction pressure = 500 psig, H2 flow rate = 4 sccm, N2 flowrate = 1 sccm).


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



CH4, the hydrogen partial pressure was greatly reduced. As aresult, the hydrocarbon yield over Rh decreased much fasterduring the first 4 hours than that over both Pt and NiMo. At300 psig, the initial hydrocarbon yield over Rh was higherthan that over Pt, and then both stabilized at around 10%after 4 hours. For the reaction pressure of 400 psig andabove, the stabilized activity of Pt is higher than that of Rh,and both are higher than the extrapolated stable activity ofNiMo.

The ratios of even-numbered carbon hydrocarbons toodd-numbered carbon hydrocarbons at different reactionpressures were all less than 1 for 7 h on stream(Fig. 5(c) and (d)), which confirmed that both Pt and Rh pre-fer the DCO/DCO2 route over the DHYD route. It can also beobserved that the HCĲ2n)/HCĲ2n − 1) ratio decreases withreaction pressure, which indicates that the reaction pressureaffects both catalyst activity and selectivity to the DCO/DCO2reaction in the same direction. Furthermore, the HCĲ2n)/HC-Ĳ2n − 1) ratio increases with on-stream time, which confirmsthe correlation of catalyst activity with its selectivity.

From the analysis above, it can be concluded that, withinthe experimental pressure range, a higher reaction pressureis needed to obtain a higher hydrocarbon yield with a lowerHCĲ2n)/HCĲ2n − 1) ratio for both the Pt and Rh catalysts. Asreported previously, not only is hydrogen a reactant but italso serves to protect the catalyst by preventing the adsorp-tion of oxygenated intermediates and coke.14,39 However, thisbeneficial effect was found not to be monotonically increas-ing in the study of model compounds, that is, a decrease inhydrocarbon yield was observed when the hydrogen partialpressure was above a certain range, which is in agreementwith the previously reported results for NiMo.14,39 This indi-cates the existence of an optimum hydrogen-to-oil ratio onthe catalyst surface, which could be due to the competitiveadsorption of H2 and algae oil molecules on the catalyst sur-face.40 A detailed kinetic study is needed to verify thishypothesis.

3.7. Effect of reaction temperature

Experiments were carried out in the temperature range of280 °C to 360 °C to investigate the temperature effect on cata-lyst activity and selectivity. The reaction pressure, H2/oil ratio,fluid flow rates and weight hourly space velocity (WHSV) werekept the same. It should be noted that despite the tempera-ture effect on the actual gas volumetric flow rate, the

residence time varied between the narrow range of 0.92 s to1.05 s, and was not considered to be a factor in the analysisof product yield and product distribution results as shown inFig. 6.

Fig. 6(a) and (b) show that the hydrocarbon yield increaseswith an increase in reaction temperature over both Pt and Rhcatalysts, and a similar temperature effect was also observedfor the NiMo catalyst.14 Also, at elevated temperatures, thethree catalysts all lost their activity with time at a moderaterate. As was reported previously, this beneficial effect fromthe increase in reaction temperature could be explained bythe Arrhenius theory of the dependence of reaction rates ontemperature. Moreover, the increase in temperature willeffectively lower the viscosity of the liquid feed and enhancethe gas–liquid mass transfer, which will result in anincreased concentration of dissolved H2.

41 At 360 °C, the C13to C20 hydrocarbon yield over all three catalysts was shownto be high, and increased following the order: Pt >

NiMo14 > Rh. There was no sign of reduction in NiMo activ-ity for at least 7 hours; however, a slight decrease in Pt andRh activity was observed, which indicates that the mecha-nism of deactivation of Pt and Rh is different from that ofNiMo. The rate of decrease in Rh activity was much lower atan elevated temperature, which indicates that reduction inactivity caused by H2 partial pressure reduction could be mit-igated by elevating the reaction temperature. It should bementioned that the cracking of biofuel molecules over thedifferent catalysts may occur at different temperatures, withRh having the lowest cracking temperature among the threestudied catalysts. The light hydrocarbons (less than 13 car-bons) that were produced from hydrocracking over Rh werenot accounted for in this comparison. The yield of hydro-cracking products was found to also increase with reactiontemperature, which decreased the C13 to C20 hydrocarbonyield over Rh. Moreover, it would also decrease the hydrocar-bon yield increase per degree of reaction temperatureincrease, which weakened the apparent temperature effect onthe improvement of Rh activity. At 280 °C, Rh showed thehighest initial activity among the three investigated catalysts,and after 4 hours, the lowest stabilized hydrocarbon yieldwas obtained over NiMo. At a reaction temperature between360 °C and 280 °C, the hydrocarbon yield over the preciousmetal catalysts decreased during the first 4 hours of run, andthen stabilized; however, again no stability could be reachedover NiMo within 7 hours of run. By examining the trend ofthe hydrocarbon yield with time, it is expected that the

Table 2 Hydrogen demand in hydrodeoxygenation of acylglycerides and free fatty acids

Reaction Hydrogen consumption per mole of FA chain (mol of H2 per mol of FA)

1. Saturation of CC double bond 1 mol of H2 demanded per mol of double bond2. Hydrogenolysis to produce FFA 13. FFA → R–CHO (fatty aldehyde) 14. R–CHO → R–CH2OH (fatty alcohol) 15. R–CH2OH → HCĲ2n) 16. R–CHO → HC(2n − 1) 0


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



Fig. 5 Effect of reaction pressure on C13 to C20 hydrocarbon yield:(a) 1% Pt, (b) 0.5% Rh; ratio of even-numbered carbon hydrocarbon toodd-numbered carbon hydrocarbon: (c) 1% Pt, (d) 0.5% Rh (60 mg cat-alyst, 1.32 wt% Nannochloropsis salina oil in dodecane, temperature =310 °C, WHSV = 0.65 goil gcat

−1 h−1).

Fig. 6 Effect of reaction temperature on C13 to C20 hydrocarbonyield: (a) 1% Pt, (b) 0.5% Rh; ratio of even-numbered carbon hydrocar-bon to odd-numbered carbon hydrocarbon: (c) 1% Pt, (d) 0.5% Rh (60mg catalyst, 1.32 wt% Nannochloropsis salina oil in dodecane, H2 pres-sure = 500 psig, WHSV = 0.65 goil gcat

−1 h−1).


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



stabilized C13 to C20 hydrocarbon yield should decrease fol-lowing the order: NiMo < Rh < Pt.

As shown in Fig. 6(c) and (d), the HCĲ2n)/HCĲ2n − 1) ratioover Rh was slightly smaller than that over Pt at the sametemperature. This ratio decreases with the increase in reac-tion temperature but at different rates, which reflect theincrease of the Pt and Rh selectivity to the DCO/DCO2 reac-tion with reaction temperature. However, as reported previ-ously, the selectivity of NiMo was less sensitive to tempera-tures of 310 °C and above.14 It can also be concluded that theHCĲ2n)/HCĲ2n − 1) ratio tends to be more stable at elevatedtemperatures, which again confirms the correlation betweenthe activity and selectivity of the precious metal-based cata-lysts when the operating temperature is varied.

3.8. Effect of hydrogen-to-oil ratio

The hydrogen-to-oil ratio (H2/oil) is as an important processparameter in the upgrading of algae oil via HDO. It repre-sents the ratio of the hydrogen feed (sccm, standard cubiccentimeter per minute) to the total liquid feed (mL min−1,solvent included). In the present work, five H2/oil ratios rang-ing between 385 SmL mL−1 and 1000 SmL mL−1 were studiedby varying the liquid flow rate at the same gas flow rate.Therefore, the residence time was kept constant at differentH2/oil ratios. Since the catalyst loading remained the samefor all runs, WHSV was reduced when the H2/oil ratio wasincreased.

As depicted in Fig. 7(a) and (b), the hydrocarbon yield overall three studied catalysts was found to increase with theincrease in H2/oil ratio. This could be explained by theenhancement of the gas–liquid convective mass transfer,which is a strong function of the liquid slug length.42 Inaddition, it has been reported that the increase in H2/oil ratiocan effectively decrease the formation of oligomerized (>C18)and cracked products, which can lead to the increase in theyield of C13–C18 hydrocarbons.43 The initial activity of NiMois more sensitive to the H2/oil ratio than that of the studiedprecious metal-based catalysts.14 Moreover, for H2/oil ratiosof 715 SmL mL−1 and below, the initial activity of Pt and Rhwas hindered by the insufficient mass transfer, and this inhi-bition on NiMo was found to occur at an even lower H2/oilratio. This could be due to the high active site density of theNiMo catalyst. With time, the activity of Rh was found todecrease much faster than that of Pt during the first 4 hours,which could be caused by the reduction of H2 partial pres-sure due to the production of CH4. After reaching stability,the activity was positively affected by the H2/oil ratio; more-over, the stabilized hydrocarbon yield was about the same forNiMo, Pt and Rh. At a H2/oil ratio of 1000 SmL mL−1, the sta-bilized hydrocarbon yield over Pt is higher than that over Rh;however, no stability was reached over NiMo within 7 hoursof reaction and its decreasing rate increased with reactiontime-on-stream.14

The H2/oil ratio had the greatest effect on the initialselectivity of NiMo, and the least on Pt. The ratio of HCĲ2n)/

Fig. 7 Effect of H2/oil ratio on C13 to C20 hydrocarbon yield: (a) 1%Pt, (b) 0.5% Rh; ratio of even-numbered carbon hydrocarbon to odd-numbered carbon hydrocarbon: (c) 1% Pt, (d) 0.5% Rh (60 mg catalyst,1.32 wt% Nannochloropsis salina oil in dodecane, temperature = 310°C, H2 pressure = 500 psig, residence time = 1 s, GHSV = 55000 h−1).


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



HCĲ2n − 1) (Fig. 7(c) and (d)) over Pt and Rh is less than 1,although it changes with time during the 7 hour reaction,which reflects primarily a selectivity to the DCO/DCO2 route.However, the HCĲ2n)/HCĲ2n − 1) ratio over NiMo was foundto be less than 1 for the H2/oil ratio of less than 1000 SmLmL−1 after 4 hours of reaction, which shows that the selectiv-ity of NiMo was affected by the H2/oil ratio and time-on-stream.14 The stability of the Pt and Rh activity seemed to benegatively affected by the H2/oil ratio. For the H2/oil ratio lessthan 715 SmL mL−1, the selectivity of both Pt and Rh was notaffected by the H2/oil ratio and was stable for at least 7hours, with Pt giving a slightly lower HĲ2n)/HCĲ2n − 1) ratio;however, for the ratio above 715 SmL mL−1, the H2/oil effectson the selectivity of Pt and Rh were opposite: the higher theH2/oil ratio, the lower the HĲ2n)/HCĲ2n − 1) ratio for Rh, butthe higher for Pt.

3.9. Effect of residence time

Residence time is an indication of the average time that thereactants spend inside the reactor. It can be calculated bydividing the volume of the reactor with the total superficialvolumetric flow rates of the reactants based on the emptybed assumption. In this study, the residence time wasadjusted by varying the reactor length while keeping the gasand liquid flow rates constant. Since the catalyst loadingincreased with the increase in reactor length, the WHSVdecreased with the increase in residence time. The reactor IDwas kept the same for all runs; therefore, the flow velocitywas held constant and the convective mass transfer did notaffect the reactor performance results in this set ofexperiments.

The results summarized in Fig. 8(a) and (b) show that thecatalyst activity and its stability both increase with theincrease in residence time. The same phenomenon wasobserved when studying the NiMo catalyst.14 At the same resi-dence time, the catalyst stability increases in the followingorder: Pt > Rh > NiMo. For the residence time of 1 s andabove, the hydrocarbon yield over Pt and Rh stabilized, andthe stabilized hydrocarbon yield over Pt was higher than thatover Rh. However, no stability was reached over NiMo within7 hours. The prolongation of residence time over Rh and Pteffectively helped to maintain the catalysts' activity. At theresidence time of 1.5 s, there was no sign of reduction in thehydrocarbon yield over Pt for at least 7 hours. For the resi-dence time of less than 1 s, the hydrocarbon yield stabilizedaround the same value over the three catalysts. The residencetime effect on the HCĲ2n)/HCĲ2n − 1) ratio over Pt and Rh isdepicted in Fig. 8(c) and (d). At the residence time of lessthan 1 s, the HCĲ2n)/HCĲ2n − 1) ratio was reported to changeover NiMo with residence time, but the selectivity of Pt andRh was not affected. For the residence time below 1.5 s,unlike the results for NiMo,14 the HCĲ2n)/HCĲ2n − 1) ratioover Pt and Rh was not affected by the increase in residencetime, but all stabilized around the same value. The lowestratio was obtained at the longest residence time, whereas for

NiMo, the highest ratio was attained as reported in ref. 14.The improved performance obtained at longer residence time

Fig. 8 Effect of residence time on C13 to C20 hydrocarbon yield: (a)1% Pt, (b) 0.5% Rh; ratio of even-numbered carbon hydrocarbon toodd-numbered carbon hydrocarbon: (c) 1% Pt, (d) 0.5% Rh (1.32 wt%Nannochloropsis salina oil in dodecane, temperature = 310 °C, H2

pressure = 500 psig, superficial flow velocity = 0.117 m s−1).


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



could be due to the reduction of WHSV, which indicates thatper unit mass, the catalyst is exposed to a lower amount offeedstock. Although no further increase in residence time isneeded in the present reactor system for the hydrotreatmentof microalgae oil, it is still necessary to point out that anincrease in residence time can possibly lead to the decreasein the formation of desired diesel range products due to theincrease in secondary reactions of hydrocarbons, such ascracking and oligomerization.44

4. Conclusions

Green diesel production from microalgae (Nannochloropsissalina) oil via hydrodeoxygenation has been investigated over1% Pt/Al2O3, 0.5% Rh/Al2O3 and presulfided NiMo/Al2O3 cata-lysts in a microreactor. Coke accumulation decreased in theorder NiMo > Pt > Rh, and the amount of coke formed overNiMo increased with reaction duration, while no on-streamtime dependence was found over Pt and Rh. Platinum andrhodium were both active for CH4 formation at their freshlyreduced state, with Rh giving the highest CH4 yield and Pthaving the longest (>30 hours) CH4 formation activity; how-ever, no detectable amount of CH4 was formed over NiModuring the reaction. As on-stream time increases, the forma-tion of CH4 over Pt and Rh both decreased, and no methanewas produced after the reaction reached stability. Methanewas mainly produced through hydrocracking initially, andthen via methanation, which was confirmed by the liquidproduct analysis and the hydrogen consumption results. Thestabilized hydrogen consumption over the three investigatedcatalysts decreased following the order NiMo > Pt > Rh,which confirms that the DHYD route demands more hydro-gen than the DCO/DCO2 route in removing the same amountof oxygen. By comparing the measured H2 consumption withthe theoretical H2 demand based on the identified FA in thefeedstock, it can be concluded that the HDO mechanism ofthe unidentified portion of the algae oil is different from thatof FA.

After investigating the susceptibility to coking, dispositiontowards methane formation, and hydrogen consumption ofeach catalyst, on the basis of external and internal masstransfer analysis, a performance study was conducted to eval-uate the activity and selectivity of these three catalysts bystudying the changes in the yield and distribution of diesel-ranged alkanes as a function of operating parameters,namely reaction temperature, pressure, hydrogen-to-oil ratioand residence time. The results reveal that the Pt and Rh cat-alysts favor the DCO/DCO2 route and their selectivity was notaffected by the experimental conditions; however, the selec-tivity of NiMo was changed from DHYD to DCO/DCO2 atreduced H2/oil ratio and residence time. The HCĲ2n)/HCĲ2n − 1)ratio over Pt and Rh was more sensitive to temperaturethan that over NiMo. The activity and selectivity of all threeinvestigated catalysts were positively affected by the elevatingreaction pressure, temperature, H2/oil ratio and residencetime. Under these conditions, the hydrocarbon yield over Pt

and Rh was found to be stabilized after 4 hours of reaction,but no stability was reached over NiMo within 7 hours. It isreasonable to extrapolate that the activity of Pt and Rh cata-lysts will be maintained for a longer on-stream time underelevated operating conditions. Moreover, the NiMo activitydecreases with time-on-stream, indicating the occurrence ofdeactivation, possibly due to the accumulation of by-prod-ucts. At the highest reaction temperature (360 °C), althoughthe activity of Pt and Rh was shown to be high, a slightdecrease was observed with time-on-stream, which indicatesthat the mechanism of deactivation of Pt and Rh was differ-ent from that of NiMo.

The formation of CH4 reduced the H2 partial pressure andcaused the fast decrease in Rh activity initially. However, itcan be mitigated by increasing the reaction temperature andthe catalyst loading, which will increase the residence timeand decrease the WHSV. It should also be noted that thecracking product over Rh was not accounted for in the hydro-carbon yield results, but its amount increases with increasingreaction temperature, which indicates that the oxygenremoval efficiency was not fully reflected by only looking atthe C13 to C20 hydrocarbon yield. The highest hydrocarbonyield, 76.5%, was obtained over 1% Pt at 310 °C, 500 psig, aH2/oil ratio of 1000 SmL mL−1 and a residence time of 1.5 s.Under these conditions, the HCĲ2n)/HCĲ2n − 1) ratio wasbelow 0.3 and both the catalyst activity and selectivity couldbe maintained for at least 7 hours. Compared with the bestperformance result of 62.7% obtained over the NiMo catalystat 360 °C and a residence time of 1 s, the reaction tempera-ture was decreased by 50 °C with an increase of 0.5 s in resi-dence time from 1 second, which could potentially result insubstantial energy saving.

Based on our experience in the catalytic hydrotreatment ofmicroalgae oil, we found three major challenges that need tobe addressed to render the algae-to-biofuel process economi-cally feasible. The growth and cultivation of microalgae, asmeasured by the algae area productivity and oil content, rep-resent by far the biggest cost factor. Genetic engineeringappears to be a promising approach to enhancing bothparameters which will provide a path to the production oflow-cost feedstock. The development of more cost-effectivecatalysts that can be commercially used in hydrotreatment ofhigh concentration algae oils for an extended period of on-stream time without significant deactivation will also contrib-ute to cost reduction. The results of the present study provideguidance on future work in this direction. Loss of catalystactivity will eventually occur, especially with high algae oilconcentration feedstock; therefore, an effective catalyst regen-eration process needs to be developed to recover the activitylost due to coke formation, impurities in the feed solution oradsorption of oxygenates on active catalyst sites.

Acknowledgements

Financial support from the DOE-Bioenergy Technologies Officethrough grant DE-EE0006063 is gratefully acknowledged. The


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online



authors appreciate the support from our principal industrialpartner Valicor Renewables LLC, and we are especially gratefulto Dr. Brian Goodall for his technical contributions to thestudy. Special thanks are given to Dr. James Manganaro forhis inspiring guidance, and the painstaking review of theexperimental results as well as the manuscript. The authors,of course, bear the sole responsibility for any errors.

References

1 E. N. Parekh, The outlook for energy: a view to 2040,ExxonMobil, 2014.

2 U.S.E.I. Administration. http://www.eia.gov/energyexplained/index.cfm?page=diesel_home, Last accessed May 5, 2015.

3 E. Facts, US EPA. http://www.chargepoint.com/files/420f05001.pdf, Last accessed May 5, 2015.

4 J. A. Melero, J. Iglesias and A. Garcia, Energy Environ. Sci.,2012, 5, 7393–7420.

5 L. Brennan and P. Owende, Renewable Sustainable EnergyRev., 2010, 14, 557–577.

6 B. Peng, Y. Yao, C. Zhao and J. A. Lercher, Angew. Chem.,2012, 124, 2114–2117.

7 D. Kubička and J. Horáček, Appl. Catal., A, 2011, 394, 9–17.8 B. Donnis, R. G. Egeberg, P. Blom and K. G. Knudsen, Top.

Catal., 2009, 52, 229–240.9 J. Holmgren, C. Gosling, K. Couch, T. Kalnes, T. Marker, M.

McCall and R. Marinangeli, Petroleum Technology Quarterly,2007, 12, 119.

10 J. Holmgren, C. Gosling, R. Marinangeli, T. Marker, G.Faraci and C. Perego, Hydrocarbon Process., 2007, 86, 67–72.

11 E. Laurent and B. Delmon, Appl. Catal., A, 1994, 109, 77–96.12 E. Laurent and B. Delmon, Appl. Catal., A, 1994, 109, 97–115.13 P. Šimáček, D. Kubička, G. Šebor and M. Pospíšil, Fuel,

2009, 88, 456–460.14 L. Zhou and A. Lawal, Energy Fuels, 2015, 29, 262–272.15 O. Şenol, T.-R. Viljava and A. Krause, Catal. Today,

2005, 100, 331–335.16 O. I. Senol, T. Viljava and A. Krause, Catal. Today, 2005, 106,

186–189.17 A. Bridgwater, Appl. Catal., A, 1994, 116, 5–47.18 O. Şenol, E.-M. Ryymin, T.-R. Viljava and A. Krause, J. Mol.

Catal. A: Chem., 2007, 277, 107–112.19 O. Şenol, T.-R. Viljava and A. Krause, Appl. Catal., A,

2007, 326, 236–244.20 K. O. Albrecht and R. T. Hallen, A Brief Literature Overview of

Various Routes to Biorenewable Fuels from Lipids for the NationalAlliance for Advanced Biofuels and Bio-products (NAABB) Consor-tium, Pacific Northwest National Laboratory Richland, WA, 2011.

21 P. Mäki-Arvela, M. Snåre, K. Eränen, J. Myllyoja and D. Y.Murzin, Fuel, 2008, 87, 3543–3549.

22 V. Dundich and V. Yakovlev, Chemistry for SubstainableDevelopment, 2009, vol. 17, pp. 521–526.

23 I. Kubičková, M. Snåre, K. Eränen, P. Mäki-Arvela and D. Y.Murzin, Catal. Today, 2005, 106, 197–200.

24 A. T. Madsen, C. H. Christensen, R. Fehrmann and A.Riisager, Fuel, 2011, 90, 3433–3438.

25 N. Joshi and A. Lawal, Chem. Eng. Sci., 2012, 74, 1–8.26 N. Joshi and A. Lawal, Chem. Eng. Sci., 2012, 84, 761–771.27 N. Joshi and A. Lawal, Ind. Eng. Chem. Res., 2013, 52, 4049–4058.28 D. Qian and A. Lawal, Chem. Eng. Sci., 2006, 61, 7609–7625.29 S. Tadepalli, R. Halder and A. Lawal, Chem. Eng. Sci.,

2007, 62, 2663–2678.30 S. Tadepalli, D. Qian and A. Lawal, Catal. Today, 2007, 125,

64–73.31 M. A. Halabi, A. Stanislaus, A. Qamra and S. Chopra, Stud.

Surf. Sci. Catal., 1996, 100, 243–251.32 H. S. Fogler, Elements of Chemical Reaction Engineering,

fourth ed. Prentice Hall Professional Technical References,2005, pp. 839–841.

33 M. A. Matthews, J. B. Rodden and A. Akgerman, J. Chem.Eng. Data, 1987, 32, 319–322.

34 J. Edvardsson, P. Rautanen, A. Littorin and M. Larsson,J. Am. Oil Chem. Soc., 2001, 78, 319–327.

35 R. Egeberg, N. Michaelsen and L. Skyum, Novel hydrotreatingtechnology for production of green diesel, Lyngby, Denmark:Haldor Topsoe A/S, 2012, p. 3.

36 J. T. Houghton, Y. D. J. G. Ding, D. J. Griggs, M. Noguer, P. J.van der Linden, X. Dai, K. Maskell and C. A. Johnson,Climate change 2001: the scientific basis, 2001.

37 P. T. Do, M. Chiappero, L. L. Lobban and D. E. Resasco,Catal. Lett., 2009, 130, 9–18.

38 D. Kubička and L. Kaluža, Appl. Catal., A, 2010, 372, 199–208.39 P. Mäki-Arvela, et al. Catalytic deoxygenation of fatty acids

and their derivatives, Energy Fuels, 2007, 21.1, 30–41.40 L. Boda, et al. Catalytic hydroconversion of tricaprylin and

caprylic acid as model reaction for biofuel production fromtriglycerides, Appl. Catal., A, 2010, 374(1), 158–169.

41 O. B. Ayodele, H. F. Abbas and W. M. A. W. Daud,Hydrodeoxygenation of stearic acid into normal and iso-octadecane biofuel with zeolite supported palladium-oxalatecatalyst, Energy Fuels, 2014, 28.9, 5872–5881.

42 M. T. Kreutzer, P. Du, J. J. Heiszwolf, F. Kapteijn and J. A.Moulijn, Chem. Eng. Sci., 2001, 56, 6015–6023.

43 M. Anand and A. K. Sinha, Temperature-dependent reactionpathways for the anomalous hydrocracking of triglyceridesin the presence of sulfided Co–Mo-catalyst, Bioresour.Technol., 2012, 126, 148–155.

44 Y. Yang, et al. Hydrotreating of C18 fatty acids tohydrocarbons on sulphided NiW/SiO2–Al2O3, Fuel Process.Technol., 2013, 116, 165–174.


Publ

ishe

d on

29

Sept

embe

r 20

15. D

ownl

oade

d by

RSC

Int

erna

l on

30/1

0/20

15 0

9:19

:29.

View Article Online

http://www.eia.gov/energyexplained/index.cfm?page=diesel_home

http://www.eia.gov/energyexplained/index.cfm?page=diesel_home

http://www.chargepoint.com/files/420f05001.pdf

http://www.chargepoint.com/files/420f05001.pdf


2nd manuscript

Documents