Jiggle bed reactor for testing catalytic activity of olivine

0 downloads 0 Views 966KB Size Report
Dec 2, 2016 - tional fossil fuels such as petroleum oil, natural gas and coal to produce heat and ... gasification. Moreover, reactor design in which a sufficient gas-solid ...... [9] M. Tripathi, J.N. Sahu, P. Ganesan, Effect of process parameters on production of bio- .... softwood bark as a liquid fuel for gas turbines. Part II: ...
Powder Technology 316 (2017) 400–409

Contents lists available at ScienceDirect

Powder Technology journal homepage: www.elsevier.com/locate/powtec

Jiggle bed reactor for testing catalytic activity of olivine in bio-oil gasification Mohammad Latifi a,b,⁎, Franco Berruti a, Cedric Briens a a b

Institute for Chemicals and Fuels from Alternative Resources (ICFAR), Department of Chemical Engineering, Faculty of Engineering, The University of Western Ontario, London, Ontario, Canada Process Engineering Advanced Research Lab (PEARL), Department of Chemical Engineering, Ecole Polytechnique de Montreal, Montreal, Quebec, Canada

a r t i c l e

i n f o

Article history: Received 7 June 2016 Received in revised form 21 November 2016 Accepted 29 November 2016 Available online 2 December 2016 Keywords: Jiggle bed reactor Gasification Olivine Bio-oil Syngas Hydrogen

a b s t r a c t The Jiggle bed reactor (JBR) is a new batch-wise micro fluidized bed reactor that was designed for screening the catalysts at various gasification operating conditions. Natural olivine mineral, (Mg,Fe)2SiO4, was selected as a suitable catalyst for bio-oil gasification in fluidized bed reactors due to its iron content and its attrition-resistance property. However, because iron can be present in the form of different phases in the olivine structure, e.g. iron oxide or metallic iron, a clear understanding of the link between the iron phase and catalytic activity of olivine is essential. Therefore, three pre-treated olivine samples were prepared as the bed material for catalytic tests in the JBR: calcined with air at 850 °C and 1000 °C and reduced with hydrogen at 800 °C. Silica sand was also utilized as an inert bed material for non-catalytic tests. A thermodynamic model was developed to predict the products mole fraction at equilibrium as a basis to evaluate behavior of the bed materials. Results proved that olivine was an active catalyst for bio-oil steam gasification when it was reduced with hydrogen, due to the presence of iron metal, on the contrary to calcined olivines that released oxygen, due to the presence of iron oxide, which reacted with the combustible products. According to thermodynamic analysis, the product gases reached predicted equilibrium state in the case of the reduced olivine despite the event of the calcined olivines. On the other hand, the activity of the calcined olivine samples revealed that such materials might have promising potentials to be applied in chemical looping gasification processes. © 2016 Elsevier B.V. All rights reserved.

1. Introduction There are some drawbacks attributed to utilization of the conventional fossil fuels such as petroleum oil, natural gas and coal to produce heat and valorized chemicals. For instance, they are limited energy resources, and their conversion processing generates harmful emissions, e.g. CO2, SOx and NOx, leading to a rise in human health, environmental and global warming problems. Development of viable and profitable processes to substitute fossil fuels with renewable resources has been severely practiced in the last decades, particularly, in countries that have to import fossil fuels. Processing the renewable energy resources provides advantages such as reduction of greenhouse gas emissions to stay with environmental rules as well as fuel supply security and reduction of fossil fuel imports. Biomass is an attractive renewable source of fuel and energy. Thermochemical processes can convert biomass either to a liquid bio-oil, bio-char and gases through flash pyrolysis [1–10], for instance, or to a ⁎ Corresponding author at: Process Engineering Advanced Research Lab (PEARL), Department of Chemical Engineering, Ecole Polytechnique de Montreal, Montreal, Quebec, Canada E-mail address: mohammad.latifi@polymtl.ca (M. Latifi).

http://dx.doi.org/10.1016/j.powtec.2016.11.057 0032-5910/© 2016 Elsevier B.V. All rights reserved.

syngas [11–22]. Syngas, which is a mixture of hydrogen and carbon monoxide [23] can also be produced from gasification of bio-oil [24– 34]. In fact, there are some advantages by conversion of biomass to bio-oil and then conversion of bio-oil to syngas: bio-oil, for instance, can be consumed directly as a fuel or upgraded to chemicals and fuels. As an economic advantage, in contrast with both raw and bulky biomass and gas, the transportation cost will be considerably saved if the dense bio-oil is produced in small-distributed units, stored and then transported to a syngas production plant at a central bio-refinery. It is also noteworthy that all primary, secondary and tertiary tar compounds that will have detrimental effects on the downstream equipment are produced through direct gasification of biomass to syngas whereas tertiary tars are not produced through gasification of bio-oil to syngas [35]. Syngas is an essential building block chemical mixture. Not only can platform chemicals and clean fuels be produced from syngas, but hydrogen is itself an alternative fuel. A high hydrogen production is usually desired: for example, methanol production requires a syngas with a molar H2/CO ratio of 2. Therefore, maximum hydrogen production has been a primary objective in steam reforming/gasification of bio-oil. Despite hydrocarbons that typically contain carbon and hydrogen and to some extend sulfur, a bio-oil is a carbon-hydrogen-oxygen bearing fuel. This elemental composition is due to its oxygenated contents

M. Latifi et al. / Powder Technology 316 (2017) 400–409

such as organic acids, aldehydes, alcohols, ketones, furans as well as phenolic and cyclic oxygenates along with 20–40 wt% water [36–38]. The composition of the mentioned oxygenates varies from a bio-oil to another depending on the upstream biomass pyrolysis conditions and type of the biomass. Accordingly, the required operating conditions to produce syngas in a bio-oil gasification process depend on its feedstock, i.e. bio-oil composition. Despite the presence of oxygenates in a bio-oil, as discussed in Section 3, some steam is required in favor of maximum production of hydrogen and carbon monoxide. In other words, if the water content of bio-oil is not enough, excess steam should be consumed in the gasifier to reach the desired H2/CO ratio. In addition to steam, an appropriate catalyst is required in bio-oil gasification. Moreover, reactor design in which a sufficient gas-solid contact should be maintained contributes significantly to the efficiency of the process [39]. For instance, due to poor gas-solid contact and nonuniform temperature profile in a fixed bed reactor, severe coke formation occurs on the surface of catalysts leading to fast catalyst deactivation [28,40]. Therefore, fluidized bed reactors are preferred to fixed bed reactors because they provide enhanced heat and mass transfer conditions. On the other hand, since much smaller particles of a catalyst are employed, particles attrition and entrainment are problematic in a fluidized bed reactor that may result in catalyst loss [41]. So, it is very crucial to find and employ an appropriate catalyst with an optimum formulation and physicochemical properties, which provides the required activity for maximum conversion of bio-oil and yield of syngas, while it has high stability against deactivation and attrition in fluidized bed reactors. Therefore, sometimes an array of different catalysts must be investigated for a given bio-oil. On the other hand, testing catalysts with new formulations and physicochemical properties under different gasification operating conditions such as temperature, reaction time and steam to carbon ratio in a pilot-scale fluidized bed reactor is indeed costly and time and labor demanding. These problems are, particularly, because there are limitations in bio-oil supply to a large scale and long term stability of bio-oil in the storage [38,42–49]. Thus, utilization of a micro reactor to screen the array of the catalysts is advantageous concerning cost, time and labor. There are some drawbacks with the available micro reactors, particularly, to test endothermic gasification reactions, and as a result, operating conditions in the large scale fluidized bed reactors cannot be mimicked in such reactors [39,50]. For instance, since they employ an external furnace to supply the heat, i.e. heat is transferred from the reactor wall to the bed, there is a high temperature gradient in the bed, heat transfer coefficient is slight and parasitic thermal cracking reactions occur on the wall. Besides, because agitators are employed to circulate the gas in the reactor, the hot gas may leak out. Therefore, the jiggle bed reactor (JBR) that is a batch-wise fluidized bed micro reactor was designed and developed at the Institute for Chemicals and Fuels from Alternative Resources (ICFAR) to overcome the above mentioned challenges with the pilot plant setups and the micro reactors. The synthesized nickel based catalysts supported mostly on alumina have been investigated for conversion of bio-oil or its model compounds [29]. There are also some natural minerals that have potentials to be employed as a catalyst. As such, olivine is a mineral that has been investigated by several researchers to crack the tars, produced during biomass gasification, in the product gases [51–59]. It has been shown that olivine, with general formula of (Mg,Fe)2SiO4, has enough strength against attrition in a fluidized bed reactor, and its iron content makes it a suitable catalyst for gasification processes [60–62]. Since different phases of iron, e.g. metallic iron, FeO, Fe2O3, Fe3O4 and MgFe2O4, can be present in the structural matrix of olivine, it is important to understand that how iron behaves as a catalyst in the catalytic gasification tests. In other words, proper pre-treatment of olivine is required to provide the catalytically active phase of iron in the reactor. However, this process is not very well discussed in the literature.

401

It has been proposed based on some temperature programmed reduction (TPR) tests that if olivine were pre-calcined with air at an optimum temperature, a maximum amount of iron oxide would come off the bulk of olivine. So when it is introduced into the gasification reactor, gases such as hydrogen and carbon monoxide, which are produced in the reactor, reduce the iron oxide to metal iron (Fe0) that is an active catalyst for gasification reactions [55]. However, we have seen in our catalytic bio-oil gasification tests in a pilot fluidized bed reactor that when the calcined olivine material was loaded into the reactor, production of hydrogen and carbon monoxide was low while production of carbon dioxide was considerably high. Therefore, a new motivation was emerged to investigate the relation between pre-treatment of olivine and its catalytic activity in bio-oil gasification. We employed the jiggle bed reactor (JBR) to perform catalytic tests on different pre-treated olivine samples, two samples of calcined olivine and one sample of reduced olivine, to monitor their activity in the gasification environment. Also, a thermodynamic model based on equilibrium constant of the independent reactions between the gasification products was developed to solidify the experimental observations. 2. Materials and methods 2.1. Bio-oil characteristics Dynamotive Energy Systems Corporation in Canada supplied the bio-oil that was produced by fast pyrolysis of hardwood. Since we had stored the bio-oil in the lab for three years, we rechecked whether its elemental composition was as of the received bio-oil. Table 1 illustrates that the elemental composition changed over the time; this could be due to either moisture absorption, loss of volatile organics or reaction with oxygen of the lab ambient [40]. Although it is far from the scope of this research, aging and change of the bio-oil composition (elemental and molecular) should be investigated to find out an optimized storage condition. Nonetheless, we utilized the old bio-oil for gasification tests of this research. 2.2. Preparation of the bed material We employed silica sand and olivine as the bed materials with a size distribution of 106–212 μm. We prepared three pre-treated olivine particles: two samples of olivine calcined with air at 850 °C and 1000 °C each for 24 h and one sample of olivine reduced with hydrogen at 800 °C for 24 h. The olivine calcination took place inside a tubular furnace. To avoid its re-oxidation happening, the olivine reduction occurred in the jiggle bed reactor in situ by a 100 ml/min flow of hydrogen; afterward, we kept the bed under argon until before doing a reaction test. Pre-treatment temperatures were chosen to be higher than the reaction temperature in order to obtain particles with a stable structure in the reactor. Silica sand was also utilized to carry out thermal cracking tests under similar operating conditions to compare the yield data with the yield data from the catalytic tests. The bed mass was 10 g in all gasification trials. Table 1 Elemental analysis of the Dynamotive bio-oil.

H, wt% C, wt% N, wt% O, wt% (balance) Water content, wt% General formula C/H, mol/mol C/O, mol/mol

Fresh bio-oil

Old bio-oil

7.20 41.67 0.27 51.13 24.34 CH2.071O0.920N0.0053 0.48 1.09

8.04 36.30 0.30 55.36 27.52 CH2.624O1.121N0.0065 0.38 0.87

402

M. Latifi et al. / Powder Technology 316 (2017) 400–409

Table 2 Surface and pore analysis of olivine.

Catalyst

BET surface area, m2/g

Average pore sizea, nm

Total pore volumeb, cm3/g

Original olivine Reduced olivine

2.734 2.144

3.751 3.878

0.0026 0.0021

a

Adsorption average pore diameter (4 V/A° by BET). Single point adsorption total pore volume of pores b81.06 nm diameter at P/P0 = 0.976. b

2.3. Catalyst characterization Surface area analyses of the original and the reduced olivine were carried out by the instrument ASAP 2010 through adsorption of N2 at 77.35 K. According to Table 2, reduction of olivine with hydrogen did not significantly change the surface properties of the original olivine. Besides, their BET surface area was negligible with a very tiny volume of the pores. In other words, olivine looked like a non-porous material. Fusion X-Ray fluorescence (XRF) analysis of the original olivine was carried out to determine the composition of the olivine. According to Table 3, silicate, magnesium, and iron were the main components of the olivine although there were traces of other metals. Attempts were made to measure the H2 consumption of the olivine by TPR and H2 chemisorption to estimate the mass of the active metals on the olivine surface; however, it was not successful since hydrogen consumption did not stop. Probably because the dominant phase was a silicate, this phase was not reduced quickly nor reduced at all at the tested conditions (below 800 °C). Another reason could be due to mass transfer limitations with respect to molecular structure of olivine where iron with different phases is distributed across the olivine. Fig. 1 presents X-ray powder diffraction (XRD) patterns of the original and the pre-treated olivine samples. Information from the literature made interpretation of the diffraction lines possible [51,53]. The main phases of olivine including (Mg,Fe)2SiO4, MgSiO3 and Mg3Si2O5(OH)4 present on the original olivine was also present on the pre-treated olivines. The applied thermal pre-treatment though changed the intensity of these silicates for the pre-treated samples of olivine. While the original olivine and the olivine calcined at 850 °C had no iron oxide phase, the olivine calcined 1000 °C had a phase of iron oxide (α-Fe2O3) at 2θ of about 33° that means this olivine gained oxygen in its structure during the calcination. Although not detected on the olivine calcined at 850 °C, iron oxide should also have been formed in this sample with a concentration below XRD detection limit. In opposite to the original and calcined olivines, XRD graph depicts that the reduced olivine contained a phase of metallic iron indicating that either some iron oxide on the surface of this olivine converted to metal iron, or some iron managed to leave its bulk and stayed on the surface ready for the catalytic reactions.

Table 3 XRF analysis data of fresh olivine. Component oxide

Composition, wt%

SiO2 TiO2 Al2O3 Fe2O3 MnO MgO CaO K2O Na2O P2O5 Cr2O3 LOI

42.15 b0.01 0.15 7.43 0.1 49.74 0.02 0.01 0.1 b0.01 0.47 0.45

Fig. 1. XRD patterns of the (a) original Olivine (b) Olivine calcined at 850 °C (c) Olivine calcined at 1000 °C and (d) Olivine reduced with hydrogen. ● Presents phases of (Mg,Fe)2SiO4, + presents phase of MgSiO3, Δ presents phase of Mg,Si2O5, ♦ presents phase of α-Fe2O3 and ■ presents phases of metallic iron (Fe0).

2.4. Equilibrium model for bio-oil gasification Given the fact that a catalyst helps product gases reach equilibrium quickly, we developed a thermodynamic model to have a correct understanding of the catalytic activity of the pre-treated olivines based on the equilibrium constant of the independent reactions between the gasification products. The model applies to equilibrium analysis of gasification of any feedstock with a general formula of CHmOn, where m and n ≥ 0. We assumed that the products at equilibrium conditions were hydrogen, carbon monoxide, carbon dioxide, steam, coke, methane, ethane, ethylene, propane, and propylene (Eq. (1)). CHm On þ b1 H2 O þ b2 O2 →a1 H2 þ a2 CO þ a3 CO2 þ a4 CH4 þ a5 H2 OðgÞ þ a6 C þ a7 C2 H6 þ a8 C2 H4 þ a9 C3 H10 þ a10 C3 H8

ð1Þ

Eqs. (2) through (4) present the elemental mass balance equations: atom C : 1 ¼ a2 þ a3 þ a4 þ a6 þ 2a7 þ 2a8 þ 3a9 þ 3a10

ð2Þ

atom H : m þ 2b1 ¼ 2a1 þ 4a4 þ 2a5 þ 6a7 þ 4a8 þ 10a9 þ 8a10

ð3Þ

atom O : n þ b1 þ 2b2 ¼ a2 þ 2a3 þ a5

ð4Þ

Eqs. (5) through (11) present the independent reactions between product gases at equilibrium: CH4 þ H2 OðgÞ ↔CO þ 3H2

ð5Þ

C2 H6 þ 2H2 OðgÞ ↔2CO þ 5H2

ð6Þ

C2 H4 þ 2H2 OðgÞ ↔2CO þ 4H2

ð7Þ

C3 H8 þ 3H2 OðgÞ ↔3CO þ 7H2

ð8Þ

C3 H6 þ 3H2 OðgÞ ↔3CO þ 6H2

ð9Þ

CO þ H2 OðgÞ ↔CO2 þ H2

ð10Þ

C þ H2 OðgÞ ↔CO þ H2

ð11Þ

And, Eq. (12) presents the relation between equilibrium constant, concentration of reactants and products and the standard Gibbs energy of the independent reaction j:  Kj ¼

P





  ∏yi

vi

¼ j

G° exp − RT

!! ð12Þ j

M. Latifi et al. / Powder Technology 316 (2017) 400–409

403

By solving a set of non-linear equations generated from the elemental mass balance equations and equilibrium constant of the independent reactions the unknown variables of a1 through a10 were estimated. 2.5. Jiggle bed reactor

Fig. 2. Diagram of the JBR: 1. on/off feed valves; 2. Inlet of carrier gas; 3. Thermocouple; 4. Inlet of feed and carrier gas; 5. Ceramic crucible with insulation; 6. Insulation disk; 7. Insulation disk; 8. Linear pneumatic actuator; 9. Outlet gas valve; 10. Stainless steel support rods; 11. Copper coil; 12. Copper disk; 13. Aluminum disk mounted on the actuator; 14. Stainless steel scalloped disk.

where the standard Gibbs energy of reaction j is estimated by Eq. (13): ΔG ° ΔGr° −ΔH r° ΔHr° 1 T 1T dT ∫ ΔC p° dT− ∫ ΔC p° ¼ þ þ RT T r R Tr T RT RT r RT

! ð13Þ j

The jiggle bed reactor (JBR) is a batch-wise bubble free fluidized bed reactor (Fig. 2). A non-porous ceramic crucible made of 99.8% alumina with an internal diameter of 2.54 cm and a height of 7.3 cm forms the reaction compartment. A fast alternating vertical motion induced by a pneumatic linear actuator fluidizes the catalyst particles in the ceramic crucible. The optimum frequency and amplitude of the actuator are 3 Hz and 0.1 m, respectively. The bed of particles expands and contracts alternatively during an operation. This sort of motion displaces the gas in the entire length of the crucible accordingly inducing intense axial and radial mixing of the gas and solid phases. It also promotes heat and mass transfer between the gas and the catalyst particles. Fig. 3 illustrates some sequences of the particles mixing in a transparent crucible. Fluidization dynamics of the JBR can be found elsewhere [39]. The JBR benefits from an induction heating mechanism to supply the heat required for endothermic reactions. The induction heating induces eddy and hysteresis currents on the surface of eight Inconel rods, symmetrically arranged inside the ceramic crucible, leading to a generation of heat on the rods' surface. This internally generated heat is transferred to the bed through the convection heating. An investigation has shown there was a slight temperature difference between the hot surface of the rods and the catalyst bed, which minimized the parasitic thermal reactions that would occur in other micro reactors. Also, heat transfer coefficient in the JBR was similar to that in the large-scale fluidized bed reactors [39]. 2.6. Experimental procedure Fig. 4 presents a schematic of the whole experimental setup developed to perform gasification tests with the JBR. A detailed explanation of the experimental procedure including bio-oil injection, reaction operation, gas collection and analysis is available elsewhere [28]. Since

Fig. 3. Sequences of mixing of catalyst particles in the jiggled bed reactor: (a) bed expansion during downward actuator retraction (b) bed contraction during upward actuator extension.

404

M. Latifi et al. / Powder Technology 316 (2017) 400–409

the JBR operation is vibrational, flexible tubing interface between the reactor and stationary parts, i.e. gas cylinders, flow meters, condenser and micro GC. We employed the Drummond capillary tubes with 4 μl volume to inject a precise amount of the bio-oil into the batch reactor. This small mass was to keep the atmospheric pressure in the reactor. We flushed out the reaction products by a flow of argon; as a result, a very dilute gas used to be collected in the sampling back making gas analysis with the micro GC a hurdle. Therefore, a Varian CP4900 micro GC was tuned to operate with the sampling time of 500 ms, and with the extremely high resolution of the thermal conductivity detectors. We conducted the tests at 800 °C a temperature at which maximum yield of syngas is achievable thermodynamically. In a previous work, the bio-oil almost entirely converted to product gases at this temperature with a nickel based catalyst [40]. Comparison of the previous data with those from catalytic activity of olivine at 800 °C would be technically interesting. The reaction time varied between 10 s and 240 s to monitor the evolution of the product gas concentrations in comparison with the predicted equilibrium data. The reaction time was the time between the bio-oil injection and the flushing with argon of the product gases into the gas sampling bag. Each experiment was repeated for three times to make sure experimental data were reproducible. The data reported in this paper are in fact average values of the obtained data from the similar tests. Maximum spread between the replicate data was ±4%.

production of hydrogen and the less production of carbon monoxide. H2 O þ CO↔H2 þ CO2

Combining the former equations, maximum achievable hydrogen is (2 + m/2 − k) mole from 1 mol of bio-oil and (2 − k) mole of steam, i.e. Eq. (16). It is noteworthy though that gasification is favoured at temperatures above 700 °C, so the rate of the water-gas shift reaction should be smaller than the rate of steam cracking reaction.   CHm Ok þ ð2−kÞH2 O→CO2 þ 2 þ m =2 −k H2

CHm Ok →CHy Oz þ gases ðH2 ; H2 O; CO; CO2 ; CH4 ; …Þ þ coke

ð17Þ

2CO↔C þ CO2

ð18Þ

CH4 þ H2 O→CO þ 3H2

ð19Þ

As discussed elsewhere [40], our idea was to count on the water content of the bio-oil as a source of steam, so we did not apply additional steam to the process. Therefore, we calculated the yield of the product gases by Eqs. (20)–(22):   g mass of hydrogen as H2 in product gas ¼ g mass of H in the raw bio−oil

Y H2

Eq. (14) explains an ideal case of conversion of the bio-oil entirely to hydrogen and carbon monoxide. This is when (1 − k) mole of steam is supplied for 1 mol of a bio-oil with a general formula of CHmOk:

Y CO;CO2 ;CH4

  CHm Ok þ ð1−kÞH2 O→CO þ 1 þ m =2 −k H2

Y C2 H4 ;C2 H6

However, upon production of carbon monoxide, some steam reacts for the water-gas shift reaction, i.e. Eq. (15), leading to the further

ð16Þ

Thermal decomposition of bio-oil, i.e. Eq. (17), Boudouard reaction, i.e. Eq. (18), and steam reforming of hydrocarbons, e.g. Eq. (19) also take place during gasification.

3. Results and discussion

ð14Þ

ð15Þ

ð20Þ

  mol moles of the product gas ¼ mol moles of the bio−oil

ð21Þ

  mol moles of the product gas ¼ mol 0:5  moles of the bio−oil

ð22Þ

We calculated the carbon conversion by summing up the yields of carbon containing product gases.

Fig. 4. Schematic of the JBR experimental setup for the catalytic gasification tests.

M. Latifi et al. / Powder Technology 316 (2017) 400–409

3.1. Effect of bed material on composition of product gases We applied the thermodynamic model developed in this research to compare the equilibrium and experimental mole fraction of the product gases obtained from different bed materials. Fig. 5 illustrates the evolution of the experimental mole fraction of the product gases for various bed materials. Also, a horizontal line indicates equilibrium mole fractions estimated by the thermodynamic model. It must be pointed out that the estimated equilibrium mole fractions of the product gases are expressed on a dry basis to be able to compare them with the experimental data. The jiggle bed reactor let conduct the tests from short to long reaction times. Except for the silica sand for which experiments were also carried out at a reaction time of

405

600 s, we performed the experiments at reaction times between 10 s and 240 s for all the tested bed materials. With the reduced olivine, it is clear that the product gases reached mole fractions near the equilibrium values at the shortest reaction times. This observation indicates the reduced olivine acted as an active and efficient catalyst that exceedingly speeded up the steam gasification reaction, i.e. Eq. (14). Since the iron oxide on the olivine surface had been reduced to metallic iron during the reduction pre-treatment with hydrogen, this result proves that metallic iron was responsible for the olivine catalytic activity. We observed a similar catalytic activity between the reduced olivine and the nickel-based catalysts tested previously [40], i.e. negligible production of hydrocarbons CH4, C2H4 and C2H6 and maximum production

Fig. 5. Measured mole fraction of product gases compared to their predicted equilibrium mole fraction, on a dry basis. Bed materials: olivine and sand.

406

M. Latifi et al. / Powder Technology 316 (2017) 400–409

of hydrogen. Whereas their production is tiny even at very short reaction times, results in Fig. 5 show that hydrocarbons could be fully cracked provided long enough reaction time was available. The product gases reached the equilibrium state at much longer reaction times during the tests with a bed of the sand. In other words, a reaction time of 600 s was required to reach equilibrium in opposite to the reaction time of 20 s necessary for the bed of the reduced olivine. However, it does not mean that sand had any catalytic activity; as explained in the later section, bio-oil conversion with sand was little. The yield data associated with the bed of the calcined olivines reveal that mole fractions of the product gases, especially, hydrogen, CO and CO2 were far from the equilibrium values. In other words, while mole fractions of hydrogen, CO and hydrocarbons decreased at longer reaction times and dropped below the equilibrium mole fractions, the mole fraction of CO2 increased at longer reaction times and was always above the equilibrium mole fraction. Since the equilibrium mole fractions were calculated according to the elemental composition of the raw bio-oil, this observation suggests that the bed of a calcined olivine acted as an extra source of oxygen in the reactor, which could lead to a significant production of CO2 and steam. The XRD plots of Fig. 1 confirm this hypothesis: α-Fe2O3 was present on the olivine calcined at 1000 °C. Such an oxide phase should have also been formed on the olivine calcined at 800 °C although it was not detected by XRD. Elemental analysis applied to the original and the calcined olivines by a CHNOS elemental analyzer showed that oxygen content of the calcined olivines had increased relatively compared to the original olivine so that the largest oxygen content was present in the olivine calcined at 1000 °C. In other words, the calcined olivines worked like an oxygen sponge so that the released oxygen reacted with the combustible product gases. This observation is also realized from the comparison of the activity of the two calcined olivines depicted in Fig. 5. The olivine calcined at 1000 °C caused less H2 and hydrocarbons and more CO2 production indicating that there was a larger mass of iron oxide on the surface of this olivine, and therefore, more mass of hydrogen and hydrocarbons was burned off. As it can be seen from the evolution of the product gases associated with the calcined olivines, it seems that mole fractions tended to reach a new equilibrium at longer reaction times. The developed thermodynamic model was applied to estimate how much oxygen should have been released from the calcined olivines to reach the new equilibrium state: the stoichiometric factor b2 in Eq. (1) was varied to achieve mole fractions which were close to the experimental data. For instance, Table 4 compares experimental mole fractions of the product gases associated with the olivine calcined at 800 °C, obtained after a reaction time of 240 s, with the estimated equilibrium mole fractions where b2 in Eq. (1) was assumed to be 1.1. Under this condition, an extra oxygen mass 1.96 times as large as oxygen mass in the injected bio-oil should have been consumed in the reactor to reach the observed experimental mole fractions. It is proposed in the literature that olivine should have been pre-calcined [55] in continuous processes whereas the calcined olivines were inappropriate catalysts in our tests that might be because of the small bio-oil to catalyst mass ratio in the batch jiggle bed reactor. However, our observation of bio-oil gasification with a calcined olivine in a pilot plant fluidized bed reactor was similar to that in the JBR. This contradiction might be due to several reasons such as the overwhelming excess of oxygen in the calcined olivine compared to flow rate and composition of

bio-oil, complex phases of iron and its dynamic displacement in the matrix of olivine, type of olivine and kinetics of oxygen pumping from the olivine surface. Results obtained with the recued olivine at different reaction times, nonetheless, indicated enough stability of this active catalyst. Therefore, we suggest that olivine should be pre-reduced before being loaded into a fluidized bed gasifier. The molar H2/CO ratio is a critical parameter for the gasification downstream processes. The equilibrium model was a useful tool to predict an available amount of this ratio by elemental analysis of the feedstock. Fig. 6 illustrates the predicted molar H2/CO ratio at equilibrium that was expected from gasification of the bio-oil. This value was reached within 30 s with the reduced olivine and within 60 s with the inert sand. However, a reaction time of 60 s would not be suitable with sand because full carbon conversion of the bio-oil might not be obtained even at much longer reaction times (Fig. 7). The predicted molar H2/CO ratio was, however, 1.65 without the addition of extra steam. Therefore, the addition of steam would be necessary if a high value of the molar H2/CO ratio, such as 2, was required for the downstream process. According to the developed thermodynamic model, it was estimated that a molar H2/CO ratio of 2 could be obtained at 800 °C when supplementary steam was injected into the reactor with steam/bio-oil mass ratio of 0.344 g/g. 3.2. Effect of bed material on hydrogen yield and bio-oil conversion Fig. 7 depicts hydrogen yield versus reaction time for the different bed materials tested in the jiggle bed reactor. It also presents carbon conversion of the bio-oil. Silica sand is known to be inert in term of catalytic activity, so this material was a good basis to compare catalytic activity of the pre-treated samples of olivine. The yield of hydrogen was increasing versus reaction time with the reduced olivine and silica sand while it had a decreasing trend with the calcined olivines. It is also noteworthy that production of hydrogen was quite stable at longer reaction times for all the bed materials. In the case of the reduced olivine, the produced hydrogen comprised 83 wt% and 93 wt% of the hydrogen content of the bio-oil, respectively, after reaction times of 10 s and 60 s. A similar trend of hydrogen yield versus reaction time is observed in the case of silica sand where the hydrogen yield increased from 0.35 at a reaction time of 10 s to a maximum hydrogen yield of 0.55 at a reaction time of 60 s. This data indicates that provided enough reaction time was available, high hydrogen yields could be achieved through the thermal cracking reactions although still far from the optimum yield.

Table 4 Comparison between estimated equilibrium and experimental mole fractions (dry basis) from olivine calcined at 1000 °C. Temperature: 800 °C, Reaction time: 240 s.

Equilibrium mole fraction Experimental mole fraction

H2

CO

CO2

CH4

C2H4 + C2H6

0.091 0.032

0.071 0.056

0.837 0.823

1.13 × 10−7 0.078

0.000 6.57 × 10−3

Fig. 6. Experimental molar H2/CO ratio versus predicted molar H2/CO ratio at equilibrium; Bed materials: olivine and sand.

M. Latifi et al. / Powder Technology 316 (2017) 400–409

407

Fig. 7. Hydrogen yield and bio-oil conversion versus reaction time; Bed materials: olivine and sand.

By comparing the data associated with the reduced olivine and silica sand, it is again proved that the reduced olivine was an active catalyst that dramatically dropped the reaction time required to reach a desired high hydrogen yield. Carbon conversion of bio-oil with the reduced olivine was 96% at the lowest reaction time and complete conversion was achieved at longer reaction times. Compared to carbon conversion with silica sand where maximum conversion of 66% was gained, it is understood that the reduced olivine was also very effective to crack large molecules inside bio-oil and enhanced reaction between oxygenated molecules and steam according to Reaction 14. Bio-oil conversion and hydrogen yield data obtained from the calcined olivines confirm that the released oxygen from the olivine surface burned off the combustible gases so that more carbon conversion was obtained with the olivine calcined at 1000 °C. Although calcination of olivine was found ineffective to increase the syngas yield in this research, its oxygen storage capability could be advantageous in cracking the complex molecules. In such cases, an optimum mass of the calcined olivine might be employed to partially burn off the large molecules to facilitate the formation of the lighter molecules conversion of which to syngas is easier than the original feedstock. 3.3. Chemical looping gasification with olivine Although we suggest pre-reduction of olivine to get as an active catalyst for the bio-oil gasification, behavior of the calcined olivines suggest that there might be some potentials in chemical looping gasification processes. In other words, research should be carried out to figure out if olivine can circulate, with an optimum flow rate, between an oxidizing reactor where enough iron oxide is generated in its matrix and a gasification reactor where the iron oxide is reduced. In such a scenario, the oxidized olivine should, early enough before leaving the gasification reactor, get reduced by the reducing gases generated through initial thermal cracking of bio-oil. Afterward, the metallic iron on the olivine surface would act as an active catalyst to convert bio-oil toward the maximum production of syngas. Assuming iron oxide Fe2O3 was the primary oxygen pumper on olivine, and hydrogen was the responsible reducing gas to reduce iron oxide, as a hypothesis, metallic iron would be generated on the olivine according to Eq. (25), i.e. combination of Eqs. (23) and (24): Fe2 O3 þ H2 →2FeO þ H2 O

ð23Þ

FeO þ H2 → Fe0 þ H2 O

ð24Þ

Fe2 O3 þ 2H2 →FeO þ Fe0 þ 2H2 O

ð25Þ

As seen in Fig. 5, 55 wt% of the hydrogen content of the bio-oil could be converted to H2 through thermal cracking. Also, if olivine were calcined at an appropriate temperature, most of the iron would be in the form of Fe2O3 [55]. Having said so, according to the general formula of the bio-oil in Table 1 and mass fraction of Fe2O3 in Table 3, we estimated that an initial bio-oil mass flow rate of 4.38 kg/h must be thermally cracked to reduce a 100 kg/h mass flow rate of a circulating olivine. However, as mentioned earlier, a thorough research is required to solidify this hypothesis. 4. Conclusions The jiggle bed reactor (JBR) was a beneficial batch-wise micro reactor to run a series of screening tests over different reaction times to evaluate catalytic activity of the bed material when accurate and reproducible experimental data, supported by the developed thermodynamic model, were obtained. The calcined olivines were inappropriate for gasification processes because they pumped out oxygen, and it burned the combustible gases. On the other hand, upon reduction pretreatment, free iron metals would be on the olivine surface that would make it a very active catalyst for complete conversion of bio-oil and a high hydrogen yield at relatively short reaction times. However, the behavior of the calcined olivines suggested that olivine might have promising potentials to be employed in a chemical looping process. From the syngas yield point of view, estimated by the thermodynamic model, a supplementary 0.344 g of steam per 1 g of the Dynamotive bio-oil would lead to reaching a molar H 2 /CO ratio of 2. Nomenclature Indices

i j Δ

Index of gasification species Index of independent equilibrium reactions Change of a parameter or a property

Symbols

ai b1

Number of moles of species i at equilibrium Number of moles of consumed steam per one mole of atom carbon

408

b2 Cp,i ΔG°j (ΔG°r)j (ΔH°r)j Kj m n P P° R T Tr Yi yi νi

M. Latifi et al. / Powder Technology 316 (2017) 400–409

Number of moles of consumed oxygen per one mole of atom carbon Special heat capacity of species i Standard Gibbs energy of reaction j at equilibrium Standard Gibbs energy change of reaction j at reference temperature Standard enthalpy change of reaction j at reference temperature Equilibrium constant of independent reaction j Number of moles of atom hydrogen per one mole of carbon atom on the C\\H\\O feed Number of moles of atom oxygen per one mole of carbon atom on the C\\H\\O feed Pressure, bar Standard pressure, bar Univeral constant of gases, 8.314 mol:J °K Temperature, °K Reference temperature, 298.15°K Yield of species i Mole fraction of species i Stoichiometric coefficient of species i

Acknowledgments The authors gratefully acknowledge the financial support of the Natural Sciences and Engineering Research Council of Canada (NSERC), of the Minister of Research and Innovation of Ontario and Agriculture and AgriFood Canada. References [1] C. Briens, J. Piskorz, F. Berruti, Biomass valorization for fuel and chemicals production – a review, Int. J. Chem. React. Eng. 6 (2008). [2] M.I. Jahirul, M.G. Rasul, A.A. Chowdhury, N. Ashwath, Biofuels production through biomass pyrolysis – a technological review, Energies 5 (2012) 4952–5001. [3] F. Abnisa, W.M.A. Wan Daud, A review on co-pyrolysis of biomass: an optional technique to obtain a high-grade pyrolysis oil, Energy Convers. Manag. 87 (2014) 71–85. [4] M. Asadieraghi, W.M. Ashri Wan Daud, H.F. Abbas, Heterogeneous catalysts for advanced bio-fuel production through catalytic biomass pyrolysis vapor upgrading: a review, RSC Adv. 5 (2015) 22234–22255. [5] S. Huiping, W. Zhendong, W. Junhui, C. Jie, L. Kaiyan, Review on pyrolysis carbonization technology of biomass, Adv. Mater. Res. 1092–1093 (2015) 72–78. [6] S. Papari, K. Hawboldt, A review on the pyrolysis of woody biomass to bio-oil: focus on kinetic models, Renew. Sustain. Energy Rev. 52 (2015) 1580–1595. [7] A. Sharma, V. Pareek, Z. Dongke, Biomass pyrolysis-a review of modelling, process parameters and catalytic studies, Renew. Sustain. Energy Rev. 50 (2015) 1081–1096. [8] Tao K, Strezov V, Evans TJ. Lignocellulosic biomass pyrolysis: a review of product properties and effects of pyrolysis parameters. Renew. Sustain. Energy Rev. 2016; 57:1126–40. [9] M. Tripathi, J.N. Sahu, P. Ganesan, Effect of process parameters on production of biochar from biomass waste through pyrolysis: a review, Renew. Sustain. Energy Rev. 55 (2016) 467–481. [10] M.N. Uddin, W.M.A.W. Daud, H.F. Abbas, Effects of pyrolysis parameters on hydrogen formations from biomass: a review, RSC Adv. 4 (2014) 10467–10490. [11] N. Abdoulmoumine, S. Adhikari, A. Kulkarni, S. Chattanathan, A review on biomass gasification syngas cleanup, Appl. Energy 155 (2015) 294–307. [12] A. Abuadala, I. Dincer, A review on biomass-based hydrogen production and potential applications, Int. J. Energy Res. 36 (2012) 415–455. [13] T.Y. Ahmed, M.M. Ahmad, S. Yusup, A. Inayat, Z. Khan, Mathematical and computational approaches for design of biomass gasification for hydrogen production: a review, Renew. Sust. Energy Rev. 16 (2012) 2304–2315. [14] G.-Y. Chen, W. Kong, Y. Xu, W.-G. Li, L.-L. Ma, B.-B. Yan, et al., Review of hydrogen production from biomass by chemical conversion process, J. Zhejiang Univ. Eng. Sci. 48 (2014) 1318–1328. [15] K. Goransson, U. Soderlind, H. Jie, Z. Wennan, Review of syngas production via biomass DFBGs, Renew. Sust. Energy Rev. 15 (2011) 482–492. [16] L. Guo, J. Chen, Reviews on hydrogen production from thermochemical gasification of biomass with supercritical water heated by concentrated solar energy, Dianli Xitong Zidonghua/Autom. Electric Power Syst. 37 (2013) 38–46. [17] S.E. Hosseini, M. Abdul Wahid, M.M. Jamil, A.A.M. Azli, M.F. Misbah, A review on biomass-based hydrogen production for renewable energy supply, Int. J. Energy Res. 39 (2015) 1597–1615. [18] P. Parthasarathy, K.S. Narayanan, Hydrogen production from steam gasification of biomass: influence of process parameters on hydrogen yield – a review, Renew. Energy 66 (2014) 570–579.

[19] J. Udomsirichakorn, P.A. Salam, Review of hydrogen-enriched gas production from steam gasification of biomass: the prospect of CaO-based chemical looping gasification, Renew. Sust. Energy Rev. 30 (2014) 565–579. [20] K. Urbaniec, R.R. Bakker, Biomass residues as raw material for dark hydrogen fermentation – a review, Int. J. Hydrogen Energy 40 (2015) 3648–3658. [21] L. Weijuan, C. Guanyi, W. Chao, Z. Xinli, Syngas production from biomass: a review, Adv. Mater. Res. 608–609 (2013) 402–405. [22] P.J. Woolcock, R.C. Brown, A review of cleaning technologies for biomass-derived syngas, Biomass Bioenergy 52 (2013) 54–84. [23] P. Spath, D. Dayton, Assessment of Synthesis Gas to Fuels and Chemicals with Emphasis on the Potential for Biomass-derived Syngas, National Renewable Energy Laboratory, Colorado, 2003. [24] D. Rennard, R. French, S. Czernik, T. Josephson, L. Schmidt, Production of synthesis gas by partial oxidation and steam reforming of biomass pyrolysis oils, Int. J. Hydrogen Energy 35 (2010) 4048–4059. [25] J.R. Marda, J. DiBenedetto, S. McKibben, R.J. Evans, S. Czernik, R.J. French, et al., Noncatalytic partial oxidation of bio-oil to synthesis gas for distributed hydrogen production, Int. J. Hydrogen Energy 34 (2009) 8519–8534. [26] M. Balat, Hydrogen-rich gas production from biomass via pyrolysis and gasification processes and effects of catalyst on hydrogen yield, Energy Source Part A: Recov. Utilization Environ. Effects 30 (2008) 552–564. [27] S. Czernik, R. French, C. Feik, E. Chornet, Hydrogen by catalytic steam reforming of liquid byproducts from biomass thermoconversion processes, Ind. Eng. Chem. Res. 41 (2002) 4209–4215. [28] M. Latifi, Gasification of bio-oils to syngas in fluidized bed reactors(Ph.D. Thesis) Western University, London, Ontario, 2012. [29] S. Ayalur Chattanathan, S. Adhikari, N. Abdoulmoumine, A review on current status of hydrogen production from bio-oil, Renew. Sust. Energy Rev. 16 (2012) 2366–2372. [30] F. Seyedzadeh Khanshan, R.H. West, Developing detailed kinetic models of syngas production from bio-oil gasification using Reaction Mechanism Generator (RMG), Fuel 163 (2016) 25–33. [31] S. Czernik, R. French, Distributed production of hydrogen by auto-thermal reforming of fast pyrolysis bio-oil, Int. J. Hydrogen Energy 39 (2014) 744–750. [32] Y.M. Li, P. Fu, W.M. Yi, X.Y. Bai, Hydrogen production by steam reforming of acetic acid and bio-oil using Ni/−Al2O3 catalysts, Int. J. Agric. Biol. Eng. 8 (2015) 69–76. [33] X. Huaqing, Y. Qingbo, Z. Zongliang, H. Zhicheng, Y. Xin, Q. Qin, Hydrogen production via sorption-enhanced catalytic steam reforming of bio-oil, Int. J. Hydrogen Energy 41 (2016) 2345–2353. [34] J.R. Marda, J. DiBenedetto, S. McKibben, R.J. Evans, S. Czernik, R.J. French, et al., Noncatalytic partial oxidation of bio-oil to synthesis gas for distributed hydrogen production, Int. J. Hydrogen Energy 34 (2009) 8519–8534. [35] G. Van Rossum, S.R.A. Kersten, W.P.M. Van Swaaij, Catalytic and noncatalytic gasification of pyrolysis oil, Ind. Eng. Chem. Res. 46 (2007) 3959–3967. [36] D. Wang, S. Czernik, D. Montane, M. Mann, E. Chornet, Biomass to hydrogen via fast pyrolysis and catalytic steam reforming of the pyrolysis oil or its fractions, Ind. Eng. Chem. Res. 36 (1997) 1507–1518. [37] M. Marquevich, S. Czernik, E. Chornet, D. Montane, Hydrogen from biomass: steam reforming of model compounds of fast-pyrolysis oil, Energy Fuels 13 (1999) 1160–1166. [38] A. Oasmaa, S. Czernik, Fuel oil quality of biomass pyrolysis oils – state of the art for the end users, Energy Fuels 13 (1999) 914–921. [39] M. Latifi, F. Berruti, C. Briens, A novel fluidized and induction heated microreactor for catalyst testing, AICHE J. 60 (2014) 3107–3122. [40] M. Latifi, F. Berruti, C. Briens, Thermal and catalytic gasification of bio-oils in the jiggle bed reactor for syngas production, Int. J. Hydrogen Energy 40 (2015) 5856–5868. [41] K.A. Magrini-Bair, S. Czernik, R. French, Y.O. Parent, E. Chornet, D.C. Dayton, et al., Fluidizable reforming catalyst development for conditioning biomass-derived syngas, Appl. Catal. A Gen. 318 (2007) 199–206. [42] J.D. Adjaye, R.K. Sharma, N.N. Bakhshi, Characterization and stability analysis of wood-derived bio-oil, Fuel Process. Technol. 31 (1992) 241–256. [43] S.P.R. Katikaneni, N.N. Bakhshi, D. Huffman, The new bio-crude: a renewable substitute for petroleum? IECEC 96 Proceedings of the 31st Intersociety Energy Conversion Engineering Conference, 11–16 Aug 1996, IEEE, New York, NY, USA 1996, pp. 1735–1740. [44] M.E. Boucher, A. Chaala, H. Pakdel, C. Roy, Bio-oils obtained by vacuum pyrolysis of softwood bark as a liquid fuel for gas turbines. Part II: stability and ageing of bio-oil and its blends with methanol and a pyrolytic aqueous phase, Biomass Bioenergy 19 (2000) 351–361. [45] A. Chaala, T. Ba, M. Garcia-Perez, C. Roy, Colloidal properties of bio-oils obtained by vacuum pyrolysis of softwood bark: aging and thermal stability, Energy Fuels 18 (2004) 1535–1542. [46] M. Garcia-Perez, A. Chaala, H. Pakdel, D. Kretschmer, D. Rodrigue, C. Roy, Evaluation of the influence of stainless steel and copper on the aging process of bio-oil, Energy Fuels 20 (2006) 786–795. [47] H. Nguyen Lyckeskog, C. Mattsson, L.-E. Amand, L. Olausson, S.-I. Andersson, L. Vamling, et al., Storage stability of bio-oils derived from the catalytic conversion of softwood Kraft lignin in subcritical water, Energy Fuels 30 (2016) 3097–3106. [48] B. Pidtasang, S. Sukkasi, A. Pattiya, Effect of in-situ addition of alcohol on yields and properties of bio-oil derived from fast pyrolysis of eucalyptus bark, J. Anal. Appl. Pyrolysis 120 (2016) 82–93. [49] R. Yin, L. Zhang, R. Liu, Y. Mei, W. Yu, Optimization of composite additives for improving stability of bio-oils, Fuel 170 (2016) 1–8. [50] M. Latifi, J. Chaouki, A novel induction heating fluidized bed reactor: its design and applications in high temperature screening tests with solid feedstocks and prediction of defluidization state, AICHE J. 61 (2015) 1507–1523.

M. Latifi et al. / Powder Technology 316 (2017) 400–409 [51] C. Courson, E. Makaga, C. Petit, A. Kiennemann, Development of Ni catalysts for gas production from biomass gasification. Reactivity in steam- and dry-reforming, Catal. Today 63 (2000) 427–437. [52] C. Courson, L. Udron, D. Swierczynski, C. Petit, A. Kiennemann, Hydrogen production from biomass gasification on nickel catalysts: tests for dry reforming of methane, Catal. Today 76 (2002) 75–86. [53] R. Rauch, K. Rosch, H. Hofbauer, D. Swierczynski, C. Courson, A. Kiennemann, Comparison of different olivines for biomass steam gasification, in: A.V. Bridgwater, D.G.B Boocock (Eds.), Science in Thermal and Chemical Biomass Conversion Conference, CPL Press, Victoria BC, 2004. [54] J. Corella, J.M. Toledo, R. Padilla, Olivine or dolomite as in-bed additive in biomass gasification with air in a fluidized bed: which is better? Energy Fuels 18 (2004) 713–720. [55] D. Swierczynski, C. Courson, L. Bedel, A. Kiennemann, S. Vilminot, Oxidation reduction behavior of iron-bearing olivines (FexMg1 − x)2SiO4 used as catalysts for biomass gasification, Chem. Mater. 18 (2006) 897–905. [56] D. Swierczynski, C. Courson, L. Bedel, A. Kiennemann, J. Guille, Characterization of Ni-Fe/MgO/olivine catalyst for fluidized bed steam gasification of biomass, Chem. Mater. 18 (2006) 4025–4032.

409

[57] G. Hu, S. Xu, S. Li, C. Xiao, S. Liu, Steam gasification of apricot stones with olivine and dolomite as downstream catalysts, Fuel Process. Technol. 87 (2006) 375–382. [58] D. Swierczynski, S. Libs, C. Courson, A. Kiennemann, Steam reforming of tar from a biomass gasification process over Ni/olivine catalyst using toluene as a model compound, Appl. Catal. B Environ. 74 (2007) 211–222. [59] M.L. Mastellone, U. Arena, Olivine as a tar removal catalyst during fluidized bed gasification of plastic waste, AICHE J. 54 (2008) 1656–1667. [60] C. Quan, S. Xu, C. Zhou, Steam reforming of bio-oil from coconut shell pyrolysis over Fe/olivine catalyst, Energy Convers. Manag. (2016) (in press). [61] J.R. Kastner, S. Mani, A. Juneja, Catalytic decomposition of tar using iron supported biochar, Fuel Process. Technol. 130 (2015) 31–37. [62] P. Ja Young, K. Mi Ae, L. Seung Jae, J. Jongjin, J. Hye Mi, P.P. Upare, et al., Preparation and characterization of carbon-encapsulated iron nanoparticles and their catalytic activity in the hydrogenation of levulinic acid, J. Mater. Sci. 50 (2015) 334–343.