Liquefaction of Cellulose in Hot Compressed Water under Variable

Experiments are carried out using a batch reactor that equips a temperature controller. The heating profile is controlled to make a proportional relat...
70 downloads 4 Views 441KB Size
Download PDF
4944

Ind. Eng. Chem. Res. 2006, 45, 4944-4953

Liquefaction of Cellulose in Hot Compressed Water under Variable Temperatures Eiji Kamio,† Susumu Takahashi,† Hidehiko Noda,‡ Chouji Fukuhara,† and Takanari Okamura*,† Department of Chemical Engineering on Biological EnVironment and Department of Mechanical Engineering, Hachinohe Institute of Technology, 88-1 Ohiraki, Myo, Hachinohe, 031-8501, Japan

The liquefaction mechanism of solid cellulose particles in hot compressed water is estimated experimentally and theoretically. Experiments are carried out using a batch reactor that equips a temperature controller. The heating profile is controlled to make a proportional relationship with time. Cellulose is drastically decomposed and liquefied when the maximum temperature is set over ∼513 K. It decomposed to oligosaccharides, monosaccharides, and pyrolysis products such as glyceraldehydes and 5-hydroxymethyl-2-furaldehyde. The concentration of oligosaccharides and monosaccharides shows maximal values for the maximum temperature set. The theoretical treatment of cellulose decomposition is investigated for stepwise processes of decomposition from (1) cellulose to oligosaccharides, (2) oligosaccharides to monosaccharides and pyrolysis products, and (3) monosaccharides to pyrolysis products. The theoretical equation derived for the proposed mechanism inserts the temperature variation with Arrhenius relationship. The concentration profiles on maximum temperature are calculated numerically. The calculated results show a reasonable correlation to the experimental data. Introduction The development of an effective utilization process to deal with the energy problem from a lack of fossil fuels is desired austerely. Biomass is an important resource that can be converted to energy. Particularly, wood biomass is one of the most useful resources. In the not too distant past, humans were getting energy by burning wood biomass. However, the energy conversion efficiency by burning wood biomass is low due to its high moisture content; high moisture biomass is not efficiently gasified by conventional thermal systems. In 1978, Modell et al. suggested a new process to solve this problem. They used supercritical water as a reaction medium and studied gasification of wood biomass.1 Irrespective of the moisture amount, the supercritical water treatment produced hydrogen and methane from wood biomass as energy fuel. Thereafter, many studies about wood biomass re-forming and various organic waste treatments with supercritical water, subcritical water, and hot compressed water were carried out.2-22 The early studies, which were mainly carried out by Elliott and his colleagues, are summarized in the review presented by Sealock et al.2 Antal and his colleagues also carried out extensive work on supercritical, subcritical, and hot compressed water conversion of biomass feedstock and biomass-related organics.3-7 They investigated the solvolysis of lignin and hemicellulose, which are the major components of wood biomass, as is cellulose. In their research, they reported that lignin and hemicellulose undergo solvolysis when the temperature is above 463 K. They also investigated the effect of the reactor’s wall material, catalyst, and maximum temperature on gasification of organic feedstock. They reported that large yields (>2 × 10-3 m3/g) of gas with a high content of hydrogen (57 mol %) were realized using an appropriate catalyst and operating at high temperature. Minowa and co-workers investigated the liquefaction and gasification of cellulose with hot compressed water and supercritical water.8-14 They investigated the effect of temperature on cellulose decomposition and reported that cellulose decom* Corresponding author. Tel.: +81-178-25-8089. Fax: +81-17825-8015. E-mail: [email protected]. † Department of Chemical Engineering on Biological Environment. ‡ Department of Mechanical Engineering.

position was started below 473 K, became fast around 513-543 K, and achieved complete decomposition at 553 K. Arai and co-workers investigated cellulose re-forming with subcritical and supercritical water.15-20 They also investigated the gasification of lignin with supercritical water treatment.19,21,22 Not for fundamental purposes, the practical use of the process was examined. At current, two main methods are studied for practical uses. One of the methods is a process with a largescale reactor. Elliott et al. studied a variety of feedstock and catalysts using a batch type and continuous flow type reactor process called TEES (Thermochemical Environmental Energy System).23,24 Useful gases such as methane and hydrogen were obtained by using appropriate catalysts. The scaled-up TEES had the performance of 0.5 ton/day liquid feed with a design flow rate of 22 dm3/h. Another method is a more compact process, which achieved by a so-called microreactor. The microreactor can achieve quick heating and rapid quenching of the reaction.17,20,25-29 By using the microreactor, the aimed reaction can be rapidly and selectively promoted by setting the desired temperature and preferable residence time. In fact, a remarkable rapid decomposition of cellulose, of several seconds or less without using catalysis, was reported by Sasaki et al.17 A numbering-up of the device allows the rapid manufacturing in large quantities and the direct commercialization of laboratory experiments. In recent years, we started to investigate the gasification of wood biomass based on the concept of the microreactor device. The final object of a series of studies was to achieve a rapid manufacturing process of hydrogen from Japanese cedar. Before examining the gasification of wood biomass, we faced a serious problem (i.e., the liquefaction of a solid sample of wood biomass). A solid sample having large particle diameter cannot be fed into the microreactor, which has a narrow channel. If we use the microreactor for the treatment of solid material, it is necessary to liquefy it beforehand. Considering a practical operation, a batch type reactor is preferable to treat solid wood biomass. In this study, we examined the liquefaction of a solid sample by using a batch reactor with its volume of 9.7 × 10-5 m3 as the first of a series in our study. In the actual process of liquefying a solid biomass by batch reactor, quick heating and

10.1021/ie060136r CCC: $33.50 © 2006 American Chemical Society Published on Web 06/03/2006

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006 4945

Figure 1. Reaction pathway of cellulose decomposition for the mathematical derivation of the kinetics model.

rapid quenching are difficult to achieve. It is important to know the reaction profile in the batch reactor that accompanies the heating and quenching processes. In fact, Sinagˇ et al. reported that the heating rate influenced the hydrothermal decomposition of glucose in the supercritical water.30 The aim of this study was to estimate the reaction profile of solid cellulose as a model sample of wood biomass under a variable temperature condition by using a batch reactor with a temperature controller. The experiments were carried out at relatively low temperatures to prevent char products from generating. The liquefaction profile of solid cellulose particles in hot compressed water was experimentally measured and theoretically predicted. Theoretical Section In the derivation for the kinetic model in this study, we assumed a simple reaction pathway as shown in Figure 1. The cellulose particle decomposes according to the following stepwise processes: (1) solid cellulose particle to oligosaccharides, (2) oligosaccharides to monosaccharides and pyrolysis products, and (3) monosaccharides to pyrolysis products. In the derivation of the theoretical model, we assumed the stoichiometric relationship as follows:

cellulose(s) + 75.7H2O(l) f 76.7oligomer(l); r1 (1) oligomer(l) + 2H2O(l) f 3monomer(l); r2

(2)

monomer(l) + H2O(l) f pyrolysis(l); r3

(3)

oligomer(l) + 2H2O(l) f 3pyrolysis(l); r4

(4)

In these equations, (s) and (l) mean the state of the materials: (s) is solid state and (l) is liquid state. Cellulose means a crystalline cellulose particle, which consists of an average of 230 molecules of glucose, because the crystalline cellulose used in this study has 230 of a viscosity averaged degree of polymerization. It exists in an aqueous phase as a solid state. Oligomer is the general term of oligosaccharides such as cellobiose, cellotriose, cellotetraose, etc. We assumed that an oligomer consists of an average of 3 units of glucose. Monomer is the general term of monosaccharide, which consists of glucose and fructose. Pyrolysis means the general term of pyrolysis products such as glyceraldehyde, glycolaldehyde, erythrose, and 5-hydroxymethyl-2-furaldehyde (5-HMF). We assumed that the average molecular weight of pyrolysis is same as that of monosaccharides. r1-r4 mean the reaction rates of each of the elementary processes. For the chemical reactions of each process, a first-order rate equation can be applied.25,26,31 As reported by Sasaki et al.,17 a crystalline cellulose particle gradually shrinks below 523 K. That is to say, the decomposition reaction of cellulose to oligosaccharides occurs at the particle surface of microcrystalline cellulose. It is a heterogeneous reaction. Therefore, the surface reaction model is progressed to

Figure 2. Representation of the reacting cellulose particle proposed in this study.

describe the first process. The theoretical kinetic equations are derived for each chemical in the following sections. Decomposition of Solid Cellulose Particle to Oligosaccharides. The representation of the reacting cellulose particle proposed in this study is shown in Figure 2. In the mathematical derivation of the kinetic model, it is assumed that a cellulose particle is a spherical shape with r0 as its initial particle diameter. The chemical reaction of cellulose decomposition to oligosaccharides occurs at the surface of the cellulose particle.17,20 The decomposition reaction of cellulose is hydrolysis by the water monomer. The water monomer is formed in both the aqueous bulk and the aqueous film near the cellulose particle, but it is consumed for the hydrolysis reaction at the surface of a cellulose particle. Therefore, the concentration gradient of the water monomer is formed near the surface of a cellulose particle. A crystalline cellulose particle has no pores, so the water monomer cannot diffuse into the particle, and its concentration becomes 0 mol/m3 at the surface. The following two steps are occurring in succession during decomposition of a cellulose particle. (A) Diffusion of water monomer through the aqueous film surrounding a cellulose particle to the surface of the particle. (B) Hydrolysis reaction of water monomer with cellulose molecule at the surface of a cellulose particle. The first mass-transfer step of the water monomer develops as follows. In this study, we assumed a linear driving force approximation for the transportation of a water monomer from the bulk to the surface of a cellulose particle. The mass-transfer rate of the water monomer through the aqueous film is as follows:

WA1 ) 4πr2kc(Cb - Cs)

(5)

where WA1 is the mass transfer rate of water monomer; Cb and Cs are the concentration of water monomer in the bulk and that at the surface of a cellulose particle, respectively; kc is the mass transfer coefficient of water monomer through the aqueous film surrounding a cellulose particle; and r is the radial position of cellulose surface after the reaction started. The radius of a cellulose particle decreases as the hydrolysis reaction progresses. Subsequently, we developed a mathematical expression of the second hydrolysis reaction step. We assumed that the hydrolysis reaction rate could be expressed as a first-order reaction for water monomer. The hydrolysis reaction rate is expressed as follows:

WA2 ) 4πr2ksCs

(6)

where WA2 is the hydrolysis reaction rate at the surface of a cellulose particle, and ks is the first-order rate constant. By assuming a quasi-steady-state for the overall process of this cellulose decomposition, it is considered that eqs 5 and 6

4946

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006

are equal. In addition, they are also considered to be equal to the apparent reaction rate of water monomer for one cellulose particle. Therefore, the following expression is available:

-rA ) 4πr kc(Cb - Cs) ) 4πr ksCs 2

2

(7)

where rA is the apparent reaction rate of water monomer for one cellulose particle. Equation 7 can subsequently be rewritten to eq 8:

-rA )

4πr2(Cb - Cs) 4πr2Cs ) 1/kc 1/ks

4πr2Cb 1/kc + 1/ks

(9)

Meanwhile, we assess the decomposition rate from an aspect of the mass balance of a cellulose particle. At a certain period during the reaction, the radial position of the cellulose surface becomes r, and the molar amount of cellulose particle remaining is 4πr3F/3, where F is the molar density of the cellulose particle. At that time, the mass balance of the unit cellulose particle is written as follows:

d 4 3 πr F ) -rB dt 3

(

)

(10)

where rB means the apparent reaction rate of the cellulose molecule for a unit cellulose particle; t is time. From the stoichiometric relationship of eq 1, the following relationship between rA and rB is available:

-rA ) -75.7rB

(11)

By substituting eqs 9 and 11 into eq 10, the following differential equation can be obtained:

4πr2Cb dr ) dt 75.7(1/kc + 1/ks)

-4πr2F

(12)

This eq 12 is the differential equation that expresses the variation of the radial position of the surface of a cellulose particle over time. We can transform eq 12 to the differential equation that expresses the variation of the decomposition ratio of cellulose particle over time. When the radial position of the surface becomes r, the decomposition ratio of cellulose particle is written as follows:

4 3 πr F 3 r ) 1-x) 4 3 r0 πr F 3 0

()

3

(13)

where x means the decomposition ratio of a unit cellulose particle. By differentiating eq 13 with t, we obtain the following relationship:

-

dx 3 dr ) r2 dt r 3 dt

4πr03 dx 4πr02(1 - x)2/3Cb F ) 3 dt 75.7(1/kc + 1/ks)

(14)

0

By substituting eqs 13 and 14 into eq 12, we get the following

(15)

The decomposition ratio of cellulose particle can be expressed by the concentration of cellulose molecule according to the following relationship:

x)

(8)

By adding the denominator of the second term to that of the third term and also adding the numerator of the second term to that of the third term, eq 8 can be rewritten as follows:

-rA )

differential equation which expresses the variation of the decomposition ratio of cellulose particle over time:

CCel,0 - CCel CCel,0

(16)

where CCel,0 and CCel are the concentration of cellulose molecules at t ) 0 and t, respectively. Substituting eq 16 into eq 15 and after simplification, the decomposition rate of a cellulose particle is as follows:

rCel ) -r1 )

dCCel ) -k1CCel2/3 dt

(17)

where rCel is the decomposition rate of a cellulose particle and k1 is apparent rate constant of cellulose decomposition described as follows:

k1 )

3CCel,01/3Cb 75.7Fr0(1/kc + 1/ks)

(18)

It should be noted that the temperature of the system is continuously changed under the experimental conditions of this study. Therefore, the rate constants also change with a change in temperature. The temperature dependence of hydrolysis rate constant, ks, can be represented by the Arrhenius’ law:

ki ) k0i exp(-Eai/RT)

(19)

where k0i and Eai are the general expression of the frequency factor and the activation energy of the reaction. Equation 19 can be applied not only to k0s and Eas but also to k02-k04 and Ea2-Ea4, which correspond to eqs 2-4. R is the gas constant, and T is temperature, which is expressed as a function of time:

heating process: T ) VTt + Tinit quenching process: T)

6000(Tmax - 273.15) (Tmax - Tinit) t+ 6000 VT

(20)

+ 273.15 (21)

where VT is heating rate. Tinit and Tmax are the initial temperature and the maximum temperature, which are set as an experimental condition, respectively. For eq 21, there is no actual physical inference, since this equation is applied to describe the quenching profile by natural cooling. Figure 3 shows the experimentally obtained temperature profile and calculated result from eqs 20 and 21. In Figure 3, Tmax is set at 523 K. In the calculation, eq 20 is immediately alternated to eq 21 when the temperature reached Tmax. As shown in Figure 3, the calculated temperature profile describes the experimental data well. By substituting eqs 18-21 into eq 17, we obtain a theoretical equation for cellulose decomposition. The variation of cellulose concentration over time can be calculated by solving the obtained equation. However it is impossible to solve the obtained equation analytically. Therefore, in this study, we solved the obtained

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006 4947

cellulose particle. For the decomposition of oligosaccharides to monosaccharides or pyrolysis products, the following kinetic equation is available from the stoichiometric relationship shown in eqs 2 and 4:

rOligo,d ) rOligo,d2 + rOligo,d4 ) -r2-r4

(23)

where rOligo,d2 and rOligo,d4 are the decomposition rates of oligosaccharides to monosaccharides and pyrolysis products, respectively; rOligo,d is the total decomposition rate of oligosaccharides. The reaction rates of eqs 2 and 4 can be expressed by the first-order kinetic equation for oligosaccharides concentration as described in eqs 24 and 25, respectively:

Figure 3. Temperature profile of the aqueous phase inside the cell. Comparison of experimental and calculated data. The heating rate is 0.167 K/s. The quenching is done by natural cooling. Table 1. Parameters Used in Calculating the Concentration of Each Chemical parameters

set value (increment unit)

Eas Ea2 Ea3 Ea4 k0s k02 k03 k04 kc Tinit Tmax CCel0 r0 F m0 MWCel Vaq Cb

141 102 130 141 7.3 × 106 6.0 × 107 1.0 × 1010 3.0 × 1011 0.001 303.15 423.15-573.15 (1 K) 0.243 8.0 × 10-5 555 0.7 41436.8 6.93 × 10-5 5.0 × 103-1.0 × 104 (4.2 × 102 mol/m3) 8.314 1

R h

unit kJ/mol kJ/mol kJ/mol kJ/mol

m3 mol/m3 J/(K‚mol) s

equation numerically. In the numerical calculation, we adopted the fourth-order Runge-Kutta method. The parameters set in the calculation are shown in Table 1. The initial conditions of the differential equation are CCel,0 ) (m0/MWCel)/Vaq ) 0.243 mol/m3 at t ) 0 and COligo,0, CMono,0, and CPyro,0 ) 0 at t ) 0. Where m0 is the loaded weight of cellulose in the reactor, MWCel is averaged molecular weight of cellulose, and Vaq is the volume of ultrapure water loaded in the reactor. COligo,0, CMono,0, and CPyro,0 mean the initial concentrations of oligomer, monomer, and pyrolysis, respectively. These initial conditions are also adopted for the calculation of COligo,0, CMono,0, and CPyro,0 described later. It should be also noted that the calculated concentration of cellulose is set at 0 mol/m3 after it was obtained as a negative value over time. Formation and Decomposition of Oligosaccharides. The stoichiometric equations of the formation and the decomposition reactions of oligosaccharides are described by eqs 1, 2, and 4, respectively. For the formation of oligosaccharides from a cellulose particle, the following kinetic equation is available because of the stoichiometric relationship shown in eq 1:

rOligo,f ) r1 ) -rCel 76.7

(22)

where rOligo,f is the formation rate of oligosaccharides from a

(24)

r4 ) k4COligo

(25)

where COligo means the concentration of the oligomer, and k2 and k4 are the rate constants of eqs 2 and 4, respectively. By substituting eqs 24 and 25 into eq 23, we obtain the total decomposition rate of oligosaccharides. The overall reaction rate of the oligosaccharide, rOligo, is obtained by adding rOligo,f and rOligo,d:

rOligo )

m/s K K mol/m3 m mol/m3 g

r2 ) k2COligo

dCOligo ) rOligo,f + rOligo,d ) dt 76.7k1CCel2/3 - (k2 + k4)COligo (26)

In the calculation of the variation of the oligomer concentration over time, the rate constants also vary with changes in temperature. Equations 19, 20, and 21 are also adopted for eq 26, and the calculation is also carried out by adopting the fourthorder Runge-Kutta method. The calculated CCel value for each step of the calculation from eq 17 is substituted for the calculation of eq 26. It should be noted that eq 27 is used to calculate the variation in oligosaccharides concentration after the cellulose concentration becomes 0 mol/m3, because oligosaccharides are not formed when the cellulose concentration is 0 mol/m3:

rOligo )

dCOligo ) rOligo,d ) -(k2 + k4)COligo dt

(27)

Formation and Decomposition of Monosaccharides. Similar to the reaction of oligosaccharides, one can derive the rate equation for monosaccharides. According to the stoichiometric relationship described in eqs 2 and 3, the formation and decomposition rate equations of monosaccharides are described as follows:

rMono,f ) r2 ) -rOligo,d2 3

(28)

rMono,d ) -r3

(29)

where rMono,f and rMono,d are the formation and decomposition rates of monosaccharides; r3 is expressed by the first-order kinetic equation for monosaccharides concentration as described in eq 30:

r3 ) k3CMono

(30)

where CMono is the concentration of the monomer. The overall reaction rate of monomer, rMono, is obtained by adding rMono,f and rMono,d as follows:

4948

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006

rMono )

dCMono ) rMono,f + rMono,d ) 3k2COligo + k3CMono dt (31)

The Arrhenius temperature dependency, the fourth-order RungeKutta method, and the calculated COrigo value for each step of calculation from eqs 26 or 27 are also adapted for the calculation of the variation of the monosaccharides concentration with eq 31. Formation of Pyrolysis Products. According to the stoichiometric relationship described in eqs 3 and 4, the formation rate equation of pyrolysis products are described as follows:

rPyro )

dCPyro ) rPyro,f ) rPyro,f3 + rPyro,f4 ) r3 + 3r4 ) dt k3CMono + 3k4COligo (32) Figure 4. Schematic of batch reactor system.

where rPyro, rPyro,f, and rPyro,d are the overall formation rate of pyrolysis products, the formation rate of pyrolysis products from oligosaccharides, and from monosaccharides, respectively. CPyro is the concentration of the pyrolysis products. The Arrhenius temperature dependency, the fourth-order Runge-Kutta method, and the calculated COrigo and CMono values for each step of calculation are also adapted for the variation calculation of the concentration of pyrolysis products with eq 32. To check the calculated results for each chemical, CPyro was also calculated from the mass balance and compared the obtained CPyro values from both eq 32 and mass balance. The obtained CPyro values show reasonable agreement within a range of 1 × 10-4 mol/m3. Experimental Section Reagents. Cellulose used was de-ashed microcrystalline cellulose (Avicel No. 2331; average particle diameter 20-100 µm) purchased from Merck. The reagents used for High Performance Liquid Chromatography (HPLC) analysis are as follows: cellobiose (>99%), cellotriose (>98%), cellotetraose (>95%), cellopentaose (>95%), erythrose (50%), pyrvaldehyde (40%), glyceraldehyde (>95%), and 5-HMF (>99%) (SigmaAldrich Japan K. K. Co. Ltd. (Tokyo)); glucose (>99%), fructose (>99%) (Wako Pure Chemicals Industries Ltd. (Osaka)). Cellulose Liquefaction Experiments and Product Analysis. The experiments were performed on a batch basis in a standard 9.7 × 10-5 m3 stirred reactor fabricated from HastelloyC-22 with four cartridge heaters. The schematic of the batch reactor system is presented in Figure 4. The reactor equipped two type K thermocouples. One was set to measure the temperature of the inner aqueous phase of the reactor, while the other was to measure the temperature of the reactor wall in which the cartridge heaters were stuck. The thermocouples and the cartridge heaters were connected to the system controller. The temperature of the inner aqueous phase during the heating process was controlled by PDI control. The temperature was varied according to the temperature profile which was set by a computer program. The monitored temperatures were recorded in notebook PC. The reactor also equips a pressure indicator and N2 line to purge the air. The system was also configured with pressure and temperature alarms, which turned off the furnace if the pressure or temperature exceeded preset limit. The batch system was also equipped with a gas sampling part. After the reaction, the produced gas sample was collected through a diaphragm pump. The obtained gas samples were analyzed with a gas chromatograph (GC) system (Shimadzu, GC-2014). Carbon dioxide and oxygen gas were detected when

the maximum temperature was set over 533 K, but their quantities were extremely low (they were below 7 × 10-4 mol even when maximum temperature was set at 553 K). Therefore, in this study, the produced gas sample was neglected for data treatment. The liquefaction experiment was carried out as follows. A 0.7 g cellulose sample and 69.3 g of ultrapure water were introduced into the reactor cell. The reactor cell was attached to the system and sealed. The reactor was purged and pressurized with N2 gas to 1 MPa. This high loading N2 gas prevented evaporation of water. The reactor was checked for leaks. The stirrer was subsequently started, and the reactor and its contents were heated according to the pre-set temperature program. In this study, the heating rate of the aqueous medium was set at 0.1667 K/s. The maximum temperatures were set at 443, 473, 503, 513, 523, 533, and 553 K. As soon as the medium temperature reached the maximum temperature, heating was ceased immediately. The reactor was subsequently cooled by air-cooling at room temperature. The pressure in the system was continuously changed according to the gas-liquid phase equilibrium during the experiment. An example of pressure variation during experiment is shown in Figure 3. After the reactor contents were fully cooled, the reactor cell was separated from the system. The entire contents in the reactor cell were collected. The collection ratios were over 99% for all experiments. The collected sample was then filtrated through a PTFE membrane filter with a pore size of 0.5 × 10-6 m. The solid residue was dried at room temperature in a vacuum desiccator and weighed. The decomposition ratio of cellulose particle was calculated from the mass balance of the solid sample before and after reaction. The composition of the liquid filtrate was analyzed by HPLC (Shimadzu co. Ltd.). The HPLC equipment consisted of an isocratic pump (LC-10ADvp), an autosampler (SIL-10ADvp), and a system controller (SCL-10Avp). The column used for the analysis was a Shim-pack SPR-Ca. The HPLC was operated at an oven temperature of 358 K, with a flow of the water solvent of 1.0 × 10-8 m3/s. The detector used was a refractive index detector (RID-10A). Peak identification was established by standard solution of the pure compound. Calibration of the peaks was performed using standard solutions of varying concentrations to develop a linear relationship between the peak areas and the corresponding concentrations. Results and Discussion The HPLC chromatograms of the products at temperatures of 443, 473, 493, 513, and 533 K are shown in Figure 5a. That

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006 4949

Figure 7. Example of a calculated result. Tmax ) 523 K, VT ) 0.167 K/s, Cb ) 7000 mol/m3.

Figure 5. HPLC chromatograms of recovered liquid samples. (a) Results for several Tmax. (b) Result at Tmax ) 523 K and retention times of pure chemicals.

Figure 6. Effect of Tmax on cellulose and oligosaccharides concentrations.

of 523 K is also shown in Figure 5b with the retention times of the pure chemicals, which are indicated by vertical broken lines. As shown in Figure 5a, the concentration of glucose, fructose, glyceraldehyde, and 5-HMF increase by increasing the maximum temperature. Although it is hard to read from Figure 5a, the concentration of oligosaccharides also depends on the maximum temperature. Figure 6 shows both concentration profiles for oligosaccharides and cellulose. As shown in Figure 6, at 473 K as maximum temperature, the concentration of cellotetraose increased. At that maximum temperature, cellulose decomposition had started slightly. At 493 K as maximum temperature, cellotriose and cellobiose were also detected. The cellotetraose concentration shows a maximal value at Tmax ) 503 K. The maximal concentration of cellotriose is at 513 K and that of cellobiose is at 533 K. The concentration of cellulose monotonically decreased above ∼473 K. These results show

that cellulose is decomposed to oligosaccharides, having a high degree of polymerization (DP), and the higher DP-oligosaccharides are decomposed to lower DP-oligosaccharides by degrees. Considering the results of Figures 5 and 6 together, it is further suggested that lower DP-oligosaccharides are decomposed to monosaccharides and pyrolysis products. The suggested pathway of decomposition reaction supports the pathway as previously reported Sasaki et al.17,20 and also supports the assumption of the theoretical model that is proposed in this study. Subsequently, we investigated the theoretical calculation. Figure 7 shows an example of the calculated result. For the calculations shown in Figure 7, Tmax was set at 523 K. As shown in Figure 7, the concentration of oligosaccharides shows a plateau above ∼3000 s. At 3000 s, the temperature of the system had decreased to about 470 K. This calculated result shows that the oligosaccharides are not decomposed below ∼470 K. This temperature of 470 K is equal to the temperature at which oligosaccharides were detected in Figure 6. These experimental and calculated results therefore show a reasonable agreement. That is to say, the proposed theoretical model qualitatively illustrates the decomposition reaction of cellulose, and we are going to compare the experimental with the calculated results. Since we could not obtain the reactant concentrations from our batch system in the middle of the operation, we investigated the effect of Tmax on the concentration of chemicals after the reaction had stopped. In the calculation, we defined the concentration of chemicals when T had decreased to 303 K as the concentration of chemicals after the reaction, Ce. For various Tmax, the concentration profiles, which are like those shown in Figure 7, were calculated, and Ce values were determined for each Tmax. They are plotted in Figure 8 together with the experimental results. Figure 8a shows the relationship between the Ce of chemicals and Tmax. Figure 8b shows a magnified vertical axis of Figure 8a, which elucidates the concentration profiles of cellulose and oligomer. As shown in Figure 8, a reasonable agreement of theoretical to experimental data is obtained. The correlation factors (R2) are 0.986 (cellulose), 0.845 (oligomer), 0.949 (monomer), and 0.948 (pyrolysis). They were the Pearson product-moment correlation coefficient calculated according to following equation: n

R2 )

{(Cexp,i - Cexp)(Ccal,i - Ccal)} ∑ i)1

x∑ n

i)1

x∑

(33)

n

(Cexp,i - Cexp)2‚

(Ccal,i - Ccal)2

i)1

where Cexp and Ccal mean the experimentally obtained Ce and

4950

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006

Figure 9. Arrhenius plot of rate constants for each reaction.

Figure 8. (a) Effect of Tmax on Ce. VT ) 0.167 K/s, Cb ) 7000 mol/m3. (b) Magnified part of the vertical axis in panel a. Table 2. Activation Energies and Frequency Factors as Obtained from a Literature study

reaction

puressure (MPa)

literature

25

136a Sasaki31 7.28 × 106 a 91b Sasaki et al.25 2.83 × 107 b 96 Kabyemela et al.26 2.57 × 1011 c 88 Amin et al.32

25

121

Bobleter and Pape33

25

121 1.33 × 1010 122.6 1.26 × 1010

Matsumura et al.34

cellulose decomposition

30

cellobiose hydrolysis

25

glucose decomposition

25-30

cellobiose pyrolysis

Ea (kJ/mol) frequency factor (-)

25

Sasaki et al.25

a Calculated from given data in Sasaki.31 b Calculated from given data in Sasaki et al.25 c Calculated from given data in Kabyemela et al.26

calculated Ce, respectively. n is the number of data. In this study, n equals 8, because Cexp was obtained for eight different Tmax. Cexp and Ccal are the mean of Cexp and Ccal, respectively. In the next section, we will evaluate the validity of the parameters such as the activation energy, the frequency factor, and the concentration of the water monomer, which are listed in Table 1. In Table 2, the activation energy, which were reported in or determined from the previously reported papers, is shown. In addition, the rate constants calculated from the activation energy and frequency factors determined in this study are plotted in Figure 9 with the Arrhenius plots obtained from previously reported papers.25,26,31-33 The activation energies used

in this study, which are listed in Table 1, show a reasonable agreement with those determined in the literature, except for Ea3. The activation energy for the glucose decomposition presented by Kabyemela et al.26 is smaller than the Ea3 used in the current study. However, when all values determined by Kabyemela et al.,26 Amin et al.,32 and Bobleter and Pape,33 which are shown in the figure presented by Kabyemela et al.,26 are used to determine the activation energy, we get 133.6 kJ/mol as activation energy for glucose decomposition. On one hand, the activation energy of glucose decomposition in the temperature ranging from 448 to 673 K reported by Matsumura et al.34 is 121 kJ/mol. These values are considerably closer to Ea3 used in this work. As shown in Figure 9, the lines calculated by using the Eai and k0i values listed in Table 1 show a good agreement with the experimental data shown in the literature. This result demonstrates the validity of the Eai and k0i values used in this study. In the meantime, there are many investigations on thermal decomposition of cellulose. Mamleev et al.35 investigated the decomposition of cotton in air in modulated thermogravimetry. They evaluated activation energies for the degradation of cotton cellulose with the model-free analysis. As a result, they concluded that the main step of degradation of cotton is transglycosylation of cotton cellulose and its related polymers formed by degradation of cotton. The activation energy of transglycosylation is evaluated as Ea ) 195.9 kJ/mol. This large activation energy shows that cellulose is easily decomposed by the hydrolysis reaction with hot compressed water rather than thermal degradation. The concentration of the water monomer is another unknown parameter. We set the concentration of the water monomer at 7000 mol/m3 in the calculation. Walrafen et al. measured the Raman OH stretching peak frequencies from water vapor below 647 K.36 They concluded that the molar ratio of water monomer and dimmer at the critical point were ∼0.3 and ∼0.7, respectively. If the monomer ratio is 0.3 and there are only monomers and dimmers in the system, the concentration of the water monomer is estimated about 9800 mol/m3. Under the conditions investigated in this study, the temperature is below the critical point. Therefore, it is suggested that the monomer concentration of water is lower than 9800 mol/m3. Ohtomo et al. proposed a tetrahedrally coordinated pentamer-monomer model to estimate the liquid structure of water.37 They concluded that the molar ratio of the water monomer and pentamer at 368 K were 0.45 and 0.55, respectively. If the monomer ratio is 0.45 and there are only monomers and pentamers in the system, the concentration of the water monomer is estimated to be about 7000 mol/m3. Under the conditions investigated in this study, the temperature is higher than 368 K. Therefore, it is suggested that the monomer concentration of water is higher than 7000 mol/m3. From these estimations, it can be assumed that 7000 mol/m3 is adequate for the monomer concentration of water. To compare the effects of monomer concentration of water and

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006 4951

The mathematical model that involves the effect of temperature variation during the reaction was derived according to the predicted decomposition mechanism. The calculated results showed a reasonable agreement to the experimentally obtained data. The used parameters for the calculation also gave a good correlation to those reported in the literature. In conclusion, it was indicated that the proposed decomposition mechanism of cellulose and the derived mathematical model were appropriate. By using the proposed mathematical model, one can easily predict the time to achieve the complete decomposition. One can also estimate the appropriate heating rate, the most suitable maximum temperature, and the required holding time at the maximum temperature for complete liquefaction. One can estimate the reaction profile of cellulose for various conditions such as heating rate and maximum operation temperature. It is necessary to consider the effect of the heating rate when we use the batch reactor, which requires slow heating and quenching processes. In the practical operation, slow heating and quenching processes will be necessary to treat a huge amount of wood biomass. They will significantly influence a design of the process. The mathematical model proposed in this study should be applicable to design a practical process of liquefaction of wood biomass. Acknowledgment This study was supported in part by a Grant-Aid from The Ministry of Education, Culture, Sports, Science and Technology of Japan. Nomenclature

Figure 10. Relationship among Tmax, Ce, and Cb. VT ) 0.167 K/s.

to confirm the validity of calculated result shown in Figure 8, we calculated the concentration profiles of cellulose, oligomer, monomer, and pyrolysis for various concentrations of water monomer. Figure 10 shows the calculated results. The monomer concentration of water was changed from 5000 to 10000 mol/m3 in increment units of 416 mol/m3. As shown in Figure 10, decomposition of cellulose becomes faster with increasing monomer concentration in water, but the concentration profile varies just a little. From these suggestions, the suitability of the used parameters is shown. This also indicates that the proposed decomposition mechanism of cellulose and the derived mathematical model are appropriate. The mathematical model should help to set the operating conditions and predict the composition of products in a closed batch reactor. It should also be applied to a practical use. Conclusion In this study, we investigated the cellulose liquefaction in hot compressed water by using a batch type reactor, which is preferable for the treatment of solid sample, but requires slow heating and quenching processes. The stepwise reaction pathway of cellulose decomposition was predicted from the experimentally obtained data profile. From the result, the mechanism of cellulose decomposition was estimated as follows. A microcrystalline cellulose decomposes to oligosaccharides with a high degree of polymerization. The produced oligosaccharides decompose down up to monosaccharides and pyrolysis products.

Cb ) concentration of water monomer in the bulk (mol/m3) Ccal ) calculated Ce (mol/m3) Ccal ) mean of Ccal (mol/m3) Ce ) concentration of chemicals after the reaction (mol/m3) Cexp ) experimentally obtained Ce (mol/m3) Cexp ) mean of Cexp (mol/m3) Ci ) concentration of chemicals (mol/m3) Cs ) concentration of water monomer at the surface of cellulose particle (mol/m3) DP ) degree of polymerization Eai ) activation energy (kJ/mol) h ) time interval set for the calculation of the fourth-order Runge-Kutta method (s) k0i ) frequency factor (-) k1 ) apparent rate constant of cellulose decomposition (mol1/3/ (m‚s)) k2 ) rate constant of eq 2 (1/s) k3 ) rate constant of eq 3 (1/s) k4 ) rate constant of eq 4 (1/s) kc ) mass transfer coefficient of water monomer through the aqueous film surrounding a cellulose particle (m/s) ks ) first-order rate constant of hydrolysis reaction for cellulose decomposition (m/s) m0 ) loaded weight of cellulose particle (g) MWCel ) Averaged molecular weight of cellulose () [molecular weight of glucose] × DP) (-) n ) number of data R ) gas constant (J/(mol‚K)) R2 ) correlation factor (-) r ) radial position of cellulose surface (m) r0 ) radius of unreacted cellulose particle (m) r1 ) reaction rate of elementary process shown as eq 1 (mol/ (m3‚s))

4952

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006

r2 ) reaction rate of elementary process shown as eq 2 (mol/ (m3‚s)) r3 ) reaction rate of elementary process shown as eq 3 (mol/ (m3‚s)) r4 ) reaction rate of elementary process shown as eq 4 (mol/ (m3‚s)) rA ) apparent reaction rate of water monomer for unit cellulose particle (mol/s) rB ) apparent reaction rate of cellulose molecule for unit cellulose particle (mol/s) rCel ) decomposition rate of cellulose particle (mol/(m3‚s)) ri ) overall reaction rate of chemicals (mol/(m3s)) ri,d ) total decomposition rate of chemicals (mol/(m3‚s)) ri,f ) total formation rate of chemicals (mol/(m3‚s)) rOligo,d2 ) decomposition rate of oligosaccharides to monosaccharides (mol/(m3‚s)) rOligo,d4 ) decomposition rate of oligosaccharides to pyrolysis products (mol/(m3‚s)) rPyro,f3 ) formation rate of pyrolysis products from monosaccharides (mol/(m3‚s)) rPyro,f4 ) formation rate of pyrolysis products from oligosaccharides (mol/(m3‚s)) T ) temperature (K) t ) time (s) Tinit ) initial temperature (K) Tmax ) maximum temperature set as an experimental condition (K) Vaq ) volume of an ultrapure water loaded in the reactor (m3) VT ) heating rate (K/s) WA1 ) mass transfer rate of water monomer (mol/s) WA2 ) hydrolysis reaction rate at the surface of a cellulose particle (mol/s) x ) decomposition ratio of unit cellulose particle (-) Greek Letters r ) molar density of the cellulose particle (mol/m3) Subscripts 1∼4 ) reaction described in eqs 1∼4 Cel ) cellulose molecule d ) decomposition f ) formation Mono ) monosaccharides Oligo ) oligosaccharides Pyro ) pyrolysis products Literature Cited (1) Modell, M.; Reid, R. C.; Amin, S. Gasification process. U.S. Patent 4,113,446, September 12, 1978. (2) Sealock, L. J., Jr.; Elliott, D. C.; Baker E. G.; Butner, R. S. Chemical processing in high-pressure aqueous environments. 1. Historical perspective and continuing developments. Ind. Eng. Chem. Res. 1993, 32, 1535. (3) Mok, W. S. L.; Antal, M. J., Jr. Uncatalyzed solvolysis of whole biomass hemicellulose by hot compressed liquid water. Ind. Eng. Chem. Res. 1992, 31, 1157. (4) Yu, D.; Aihara M.; Antal, M. J., Jr. Hydrogen production by steam reforming glucose in supercritical water. Energy Fuels 1993, 7, 574. (5) Xu, X.; Matsumura, Y.; Stenberg, J.; Antal, M. J., Jr. Carboncatalyzed gasification of organic feedstocks in supercritical water. Ind. Eng. Chem. Res. 1996, 35, 2522. (6) Antal, M. J., Jr.; Va´rhegyi, G.; Jakab, E. Cellulose pyrolysis kinetics: revisited. Ind. Eng. Chem. Res. 1998, 37, 1267. (7) Antal, M. J., Jr.; Allen, S. G.; Schulman, D.; Xu, X. Biomass gasification in supercritical water. Ind. Eng. Chem. Res. 2000, 39, 4040. (8) Minowa, T.; Ogi T.; Yokoyama, S. Effect of pressure on lowtemperature gasification of wet cellulose into methane using reduced nickel catalyst and sodium carbonate. Chem. Lett. 1995, 24, 285.

(9) Minowa, T.; Ogi T.; Yokoyama, S. Hydrogen production from wet cellulose by low-temperature gasification using reduced nickel catalyst. Chem. Lett. 1995, 24, 937. (10) Minowa, T.; Zhen, F.; Ogi, T.; Va´rhegyi, G. Liquefaction of cellulose in hot compressed water using sodium carbonate: products distribution at different reaction temperatures. J. Chem. Eng. Jpn. 1997, 30, 186. (11) Minowa, T.; Zhen, F.; Ogi, T.; Va´rhegyi, G. Decomposition of cellulose and glucose in hot-compressed water under catalyst-free conditions. J. Chem. Eng. Jpn. 1998, 31, 131. (12) Minowa, T.; Zhen, F. Hydrogen production from cellulose in hot compressed water using reduced nickel catalyst: product distribution at different reaction temperatures. J. Chem. Eng. Jpn. 1998, 31, 488. (13) Szabo´, P.; Minowa, T.; Ogi, T. Liquefaction of cellulose in hot compressed water using sodium carbonate. Part 2: Thermogravimetricmass spectrometric study of the solid residues. J. Chem. Eng. Jpn. 1998, 31, 571. (14) Inoue, S.; Hanaoka, T.; Minowa, T. Hot compressed water treatment for production of charcoal from wood. J. Chem. Eng. Jpn. 2002, 35, 1020. (15) Adschiri, T.; Hirose, S.; Malaluan, R.; Arai, K. Noncatalytic conversion of cellulose in supercritical and subcritical water. J. Chem. Eng. Jpn. 1993, 26, 676. (16) Sasaki, M.; Kabyemela, B.; Malaluan, R.; Hirose, S.; Takeda, N.; Adschiri, T.; Arai, K. Cellulose hydrolysis in subcritical and supercritical water. J. Supercrit. Fluids 1998, 13, 261. (17) Sasaki, M.; Fang, Z.; Fukushima, Y.; Adschiri, T.; Arai, K. Dissolution and hydrolysis of cellulose in subcritical and supercritical water. Ind. Eng. Chem. Res. 2000, 39. 2883. (18) Watanabe, M.; Inomata, H.; Arai, K. Catalytic hydrogen generation from biomass (glucose and cellulose) with ZrO2 in supercritical water. Biomass Bioenergy 2002, 22, 405. (19) Osada, M.; Sato, T.; Watanabe, M.; Adschiri, T.; Arai, K. Lowtemperature catalytic gasification of lignin and cellulose with a ruthenium catalyst in supercritical water. Energy Fuels 2004, 18, 327. (20) Sasaki, M.; Adschiri, T.; Arai, K. Kinetics of cellulose conversion at 25 MPa in sub- and supercritical water. AIChE J. 2004, 50, 192. (21) Watanabe, M.; Inomata, H.; Osada, M.; Sato, T.; Adschiri, T.; Arai, K. Catalytic effects of NaOH and ZrO2 for partial oxidative gasification of n-hexadecane and lignin in supercritical water. Fuel 2003, 82, 545. (22) Saisu, M.; Sato, T.; Watanabe, M.; Adschiri, T.; Antal, M. J., Jr. Conversion of lignin with supercritical water-phenol mixtures. Energy Fuels 2003, 17, 922. (23) Elliott, D. C.; Sealock, L. J., Jr.; Baker, E. G. Chemical processing in high-pressure aqueous environments. 3. Batch reactor process development experiments for organics destruction. Ind. Eng. Chem. Res. 1994, 33, 558. (24) Elliott, D. C.; Phelps, M. R.; Sealock, L. J., Jr.; Baker, E. G. Chemical processing in high-pressure aqueous environments. 4. Continuousflow reactor process development experiments for organics destruction. Ind. Eng. Chem. Res. 1994, 33, 566. (25) Sasaki, M.; Furukawa, M.; Minami, K.; Adschiri, T.; Arai, K. Kinetics and mechanism of cellobiose hydrolysis and retro-aldol condensation in subcritical and supercritical water. Ind. Eng. Chem. Res. 2002, 41. 6642. (26) Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Kinetics of glucose epimerization and decomposition in subcritical and supercritical water. Ind. Eng. Chem. Res. 1997, 36, 6. 1552. (27) Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Degradation kinetics of dihydroxyacetone and glyceraldehyde in subcritical and supercritical water. Ind. Eng. Chem. Res. 1997, 36, 6. 2025. (28) Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K.; Ohzeki, H. Rapid and selective conversion of glucose to erythrose in supercritical water. Ind. Eng. Chem. Res. 1997, 36, 5063. (29) Kabyemela, B. M.; Takigawa, M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Mechanism and kinetics of cellobiose decomposition in sub- and supercritical water. Ind. Eng. Chem. Res. 1998, 37, 357. (30) Sinagˇ, A.; Kruse, A.; Rathert, J. Influence of the heating rate and the type of catalyst on the formation of key intermediates and on the generation of gases during hydropyrolysis of glucose in supercritical water in a batch reactor. Ind. Eng. Chem. Res. 2004, 43, 502. (31) Sasaki, M. Ph.D. Thesis, Tohoku University, Sendai, Japan, 2000. (32) Amin, S.; Reid, R. C.; Modell, M. Reforming and decomposition of glucose in an aqueous phase. In Intersociety Conference on EnVironmental Systems, San Francisco, CA, 1975; The American Society of Mechanical Engineers (ASME): New York, 1975; ASME Paper 75-ENAs-21. (33) Bobleter, O.; Pape, G. Hydrothermal decomposition of glucose. Monatsh. Chem. 1968, 99, 1560.

Ind. Eng. Chem. Res., Vol. 45, No. 14, 2006 4953 (34) Matsumura, Y.; Yanachi, S.; Yoshida, T. Glucose decomposition kinetics in water at 25 MPa in the temperature range of 448-673 K. Ind. Eng. Chem. Res. 2006, 45, 1875. (35) Mamleev, V.; Bourbigot, S.; Le Bras, M.; Yvon, J.; Lefebvre, J. Model-free method for evaluation of activation energies in modulated thermogravimetry and analysis of cellulose decomposition. Chem. Eng. Sci. 2006, 61, 1276. (36) Walrafen, G. E.; Yang, W.-H.; Chu, Y. C. Raman spectra from saturated water vapor to the supercritical fluid. J. Phys. Chem. B 1999, 103, 1332.

(37) Ohtomo, N.; Tokiwano, K.; Arakawa, K. The structure of liquid water by neutron scattering. II. Temperature dependence of the liquid structure. Bull. Chem. Soc. Jpn. 1982, 55, 2788.

ReceiVed for reView February 2, 2006 ReVised manuscript receiVed May 3, 2006 Accepted May 4, 2006 IE060136R