Author Manuscript: Published in Journal of Energy Chemistry, 23, 4, July 2014, 535-541

DOI: doi:10.1016/S2095-4956(14)60182-0

Production of Octyl Levulinate Biolubricant over Modified

H-ZSM-5: Optimization by Response Surface Methodology

Kakasaheb Y. Nandiwalea, Sunil K. Yadavab, and Vijay V. Bokadea*

aCatalysis and Inorganic Chemistry Division, CSIR - National Chemical Laboratory, Pune-

411008, India.

bUniversity Institute of Chemical Technology, North Maharashtra University, Jalgaon-

425001, Maharashtra, India.

*Correspondence to:

Dr. Vijay V. Bokade,

Catalysis and Inorganic Chemistry Division,

CSIR - National Chemical Laboratory,

Pune-411008, India.

Ph; +91-20-25902458, Fax; +91-20-25902634

E-mail: vv.bokade@ncl.res.in

Published in: Journal of Energy Chemistry 23 535-541 2014.

DOI: http://www.sciencedirect.com/science/article/pii/S2095495614601820

1

2

Abstract

The present study highlighted the use of modified H-ZSM-5 (Meso-HZ-5) as

heterogeneous catalyst for the synthesis of octyl levulinate biolubricant by catalytic

esterification of biomass derived renewable levulinic acid (LA) with n-octanol. The process

variables such as catalyst loading (X1), n-octanol to LA molar ratio (X2) and reaction

temperature (X3) were optimized through response surface methodology (RSM), using the

Box–Behnken model. Analysis of variance was performed to determine the adequacy and

significance of the quadratic model. The yield of octyl levulinate was obtained to be 99% at

optimum process parameters.

The quadratic model developed was found to be adequate and statistically accurate

with correlation value (R2) of 0.9971 to predict the yield of octyl levulinate biolubricant. The

study also extended on validation of theoretical and experimental data, including catalyst

reusability.

Keywords: Biolubricant, esterification, H-ZSM-5, Levulinic acid, octyl levulinate, response

surface methodology.

3

1 Introduction

Extensive use of non-renewable mineral oil based lubricant is creating numerous

environmental concerns such as groundwater contamination, surface water, air pollution, soil

contamination, agricultural product and food contamination [1]. Emphasis on the

development of renewable, biodegradable and eco-friendly lubricants has resulted in the

extensive employ of natural oils, fats for non-edible purposes [2]. The application of animal

fats and plant oils for industrial purposes, particularly as lubricants, has been in practice for

several years [3].

Lubricant is synthesized by hydrolysis of vegetable oil to fatty acids followed by

catalytic conversion of fatty acids to their corresponding esters with higher alcohols (C8 to

C14) [1, 4]. The use of different lower and higher alcohols, namely methanol, ethanol, n-

propanol and n-octanol, for the synthesis of methyl, ethyl, propyl and octyl esters by

transesterification of vegetable oil has been reported in literature [5-6]. Fatty acid alkyl esters

possessing 22 to 26 carbon atoms can act as biolubricant components. The numerous

vegetable oils such as soyabean [7], jatropha [8], castor [9], waste cooking oil [10] have been

extensively investigated for biolubricant production. Both catalytic as well as enzymatic

methods have been employed for biolubricant synthesis [5,6,9-11].

The octyl ester biolubricant can be obtained by catalytic esterification of levulinic

acid (LA) with octanol. LA is one of the top biomass derived platform molecule that can be

made from C6 sugar carbohydrates derived from renewable ligno-cellulose [12]. Hence

production octyl ester biolubricant from renewable biomass feedstock may set new important

area of research for academic as well industrial interest.

The renewable bio-based lubricants represent promising substitutes to other synthetic

and mineral-oil based lubricants due to their specific functional features such as high flash

point, high viscosity index, high lubricity, very low volatility, bio-degradability [13]. Here,

4

esterification of LA with n-octanol over modified H-ZSM-5 aiming to octyl levulinate

biolubricant is studied.

Response surface methodology (RSM) is an excellent tool for optimizing a catalytic

process and has been widely adopted to improve process conditions [1, 16-18]. It is a

collection of statistical and mathematical techniques which are first used to generate optimal

data and then analysed by means of regression methods. A set of multivariable data can be

obtained by appropriately designed experiments. Graphical representation were obtained by

RSM, which serve as a visual aid to better understanding of the reaction process [1,8].

Optimization of yield of octyl levulinate by utilizing RSM analysis may allow a more

comprehensive analysis on the interactions between experimental variables than a single-

factor experimental design. Consequently, this could lead to a better understanding and

knowledge influencing process parameters and subsequently maximizes the yield of octyl

levulinate biolubricant. Apart from that, it also reduces the number of experimental runs

required to generate statistically-validated results [18].

The novelty of the present study lies in the use of biomass derived LA as feedstock for

biolubricant production by esterification with n-octanol over modified H-ZSM-5 catalyst.

Octanol has been selected as the working alcohol since it is the cheapest among the higher

chain alcohols (C8 to C14). This study also focuses on, use of design expert software to

optimize the process parameters for esterification reaction in view to maximize yield of octyl

luvulinate by using response surface methodology (RSM). The three crucial variables such as

catalyst loading (X1), octanol to LA molar ratio (X2) and reaction temperature (X3) were

investigated for production octyl levulinte biolubricat using RSM. The effects of these

variables on the yield of octyl ester (response) were studied with Box–Behnken model and

subsequently an empirical mathematical model correlating the response to the variables was

5

developed and presented as well. The study also extended on validation of theoretical and

experimental data along with reusability of modified H-ZSM-5 catalyst at optimal conditions.

2 Experimental

2.1 Materials

Levulinic acid (99%) and n-octanol (99%) were obtained from M/s E. Merck,

Mumbai (India). NaOH and ammonium nitrate were obtained from M/s Loba chemie

Mumbai (India). All the reagents were of analytical grade and used without further

purification.

2.2 Catalyst synthesis and characterization

H-ZSM-5 with Si/Al ratio 37 was synthesized as per the reported procedure [14]. The

modified H-ZSM-5 (Meso-HZ-5) was obtained by following procedure. 300 mL of 0.2 M aq.

NaOH was mixed with 10 g of H-ZSM-5 in a flask and kept at 338 K for 30 min. The zeolite

sample was transformed into ammonium forms by threefold ion exchange with aq. 0.1M

ammonium nitrate (in the proportion 10 mL g-1 of product for 5 h) without calcinations in

between ion-exchange procedures. Finally, samples were transformed into the hydrogen

forms by calcinations in air at 823 K for 5 h. The sample obtained at the final stage was

designated as Meso-HZ-5.

The detailed characterization of synthesized catalysts samples can be found in our

previous publication [15]. The phase identification, degree of crystallization and purity were

determined by powder X-ray diffraction (XRD). It can be seen as XRD pattern in Figure 1 (a)

the crystallinity of Meso-HZ-5 was 97% and it also confirmed the phase purity of H-ZSM-5

and Meso-HZ-5. Figure 1 (b) shows N2 adsorption-desorption isotherm of H-ZSM-5 and

Meso-HZ-5. H-ZSM-5 showed type I isotherm, indicating that H-ZSM-5 is microporous

material and that of Meso-HZ-5 represent both type I and type IV isotherms which suggests

the presence of both micro and mesoporosity. The specific surface area of synthesized

6

catalysts was calculated using Brunaer-Emmett-Teller (BET) method (Table 1). The overall

acidity of H-ZSM-5 and Meso-HZ-5 (Table 1) were measured by Temperature Programmed

Ammonia Desorption (TPAD) using a Micromeritics AutoChem (2910, USA) equipped with

thermal conductivity detector. Prior to the measurements, sample was dehydrated at 773 K in

He (30 cm3 min−1) for 1 h. The temperature was then decreased to 323 K and then NH 3 was

allowed to adsorb by exposing sample to a gas stream containing 10% NH3 in He for 1 h. It

was then flushed with He for another 1 h. The NH3 desorption was carried out in He flow (30

cm3 min−1) by increasing the temperature up to 723 K with a heating rate of 10 K min−1.

2.3 Catalytic evaluation

The experiments of synthesis of octyl levulinate (biolubricant) were carried out in

a100 mL cylindrical stainless steel batch reactor, under autogeneous pressure (20-25 psi). LA,

n-octaol and a given amount of catalyst totalling to 30 ml reaction volume was mixed in the

reactor. The reactor was heated by an electric heater with PID controller. The temperature

was maintained within an accuracy of ±0.5 K by PID controller. The experiments were

conducted at a temperature range of 373-393 K, catalyst loading of 10-30 (wt. % of LA), n-

octanol to LA molar ratio of 4-10 and reaction time of 5 h, respectively. After running the

reaction for a desired duration, the reactor was quenched by quickly immersing in a cool

water bath to terminate the reaction. After quenching, the sample was filtered and analysed.

All experiments were performed in triplicate and average values were reported.

2.4 Analysis of reaction feed and product

The liquid reaction feed and product were analysed by using gas chromatography (GC)

Chemito GC-1000, capillary column, BP-5 (50m length and 0.3mm width) with nitrogen as a

carrier gas and Flame Ignition Detector (FID) in programmable temperature range of 313 to

473 K. The reaction products were also confirmed by GC-MS (Shimadzu-QP 5000).

2.5 Response surface methodology

7

An experimental design for the series of parameters used for octyl levulinate

biolubricant synthesis by esterification of LA with n-octanol over Meso-HZ-5 was built by

RSM with the Design-Expert® Version 8.0.7.1 (Stat- Ease, Inc., Minneapolis, USA) [20-21].

A Box–Behnken factorial model was employed in this study requiring 17 experimental runs

[8,19]. Different formulations of the design consisted of all possible combinations of the

independent factors at all levels and were conducted in a fully randomized order.

The independent variables chosen were catalyst loading (X1), octanol to LA molar

ratio (X2) and reaction temperature (X3). Percentage yield of octyl levulinate biolubricant (Y)

was chosen to be the target or response parameter as a dependent variable. Each factor in the

experiment was established and coded into levels -1, 0 and +1 as shown in Table 2. The Box-

Behnken design (BBD) matrix of the three variables in coded units, natural units and

corresponding response values are given in Table 3. The significance of the model was

determined by the statistical parameter. Model graphs were plotted resulting from the

equation, using the same software.

3 Results and discussion

3.1 Performance of catalysts in octyl levulinate synthesis

The synthesised H-ZSM-5 and Meso-HZ-5 catalysts were evaluated for octyl

levulinate synthesis. The reaction parameters used: catalyst loading of 10% of LA, molar

ratio of octanol to LA of 6, reaction temperature of 373 K and reaction time of 6 h. In general

esterification is autocatalytic reaction; hence esterification of LA with octanol able occurs

even in absence of catalyst. For this reason, thermal reaction (blank) was performed at 373 K.

The time courses for the yields of octyl levulinate over blank, H-ZSM-5 and Meso-HZ-5

catalysts are represented as Figure 2. The maximum yield of octyl levulinate obtained after 4

h over blank, H-ZSM-5 and Meso-HZ-5 were 7.6%, 29% and 56% respectively. The higher

yield of octyl levulinate obtained over Meso-HZ-5 than H-ZSM-5 may be attributed to the

8

higher acidity, surface area and mesoporosity generation in Meso-HZ-5 due to selective

extraction of Si with NaOH treatment, which overcome diffusional resistances (Table 1) [15].

The selectivity towards octyl levulinate in all the experiments was 100%. It has been

observed that yield of octyl levulinate reached maximum at reaction time of 4 h, there after it

was stable (Figure 2). Hence the optimum reaction time of 4 h was used for all further

experiments. In present case, Meso-HZ-5 was found to be potential catalyst for production of

octyl levulinate biolubricant. Hence influence of various parameters for octyl levulinate

production over Meso-HZ-5 was further investigated with RSM design. The optimization of

process parameters in view to maximize the yield of octyl levulinate biolubricat over Meso-

HZ-5 catalyst is presented later. The reusability Meso-HZ-5 catalyst was also tested at

optimized process parameters.

3.2 Model analysis

The experiments of esterification reactions were carried out as per Table 3 and

obtained results were indicated. The analysis of results was performed with design expert

software Version 8.0.7.1. This software offers a response surface model and ANOVA

(analysis of variance).

The quadratic model was generated by RSM design; this was used to evaluate the

responses. Experimental and predicted values of the LA conversion are shown in Table 3. The

reported yield of octyl levulinate was obtained as an average of triplicate determinations after

fixed reaction time of 5 h. The selectivity towards octyl levilinate was 100% in all

experiments. As shown in the table, the yield of octyl levulinate increased from 56% to 98%,

depending on the reaction conditions. Based on the RSM analysis, the second order quadratic

model of coded units for yield of octyl levulinate (Y) is presented as Eq. (1):

Y = 93 + 11.63X1 + 2.88X2 + 9.75X3 + 0.5X1X2 + 4.75X1X3 + 3.25X2X3 - 12.25X12 - 3.25X2

2

- 9X32 (1)

9

where Y represents the yield of octyl levulinate biolubricant and X1, X2 and X3 are the

coded variables in esterification reaction. Figure 3 demonstrated the good linear correleation

between the actual values and predicted yield of octyl levulinate (calculated from Eq. (1)).

Positive sign of the coefficients in linear terms reveals that with an increasing the variables

the yield increases linearly (synergistic effect), while negative sign indicates antagonistic

effect. From the Eq. (1), yield of octyl levulinate has linear and quadratic effects by the three

process variables. Eq. (1) indicates that the catalyst loading (X1) has the strongest effect on

the response since the coefficient of X1 (11.63) is the largest compared to the other

investigated factors. Next most significant effects on the response is reaction temperature

(X3), followed by (X2), slightly weaker interaction effects between parameters (X1X3 and

X2X3) and much lower interaction effect between parameters (code X1X2).

Statistical analysis includes the interaction effects and the main effects of the

variables on the yield of octyl levulinate. The ANOVA tests the statistical significance of each

effect by comparing with the mean square against the estimated experimental error within the

range of experimental conditions. Statistical analysis of variance (ANOVA) of the main

effects and the interactions for the chosen response, together with the test of statistical

significance for the response surface quadratic model are shown in Table 4.

The sum of squares is used to estimate the F-values (F), which are defined as the ratio

of the respective mean square effect and the mean square error [8,18]. The Model F-value of

264.78 implies the model is significant. There is only a 0.01% chance that a “Model F-Value”

this large could occur due to noise. Statistical model fit summary consisting of a sequential

model sum of squares and lack of fit tests suggested a quadratic model as the best fit model.

Values of “Prob > F” less than 0.05 indicate model terms are significant in this case X 1, X2,

X3, X1X3, X2X3, X12, X2

2 and X32 are significant model terms.

10

Correlation value (R2) of 0.9971 indicated that the model could explain 99.71% of the

variability in esterification process. The R2-predicted of 0.9531 is in reasonable agreement

with the R2-adjusted of 0.9933. Adequate precision measures the signal to noise ratio. A ratio

greater than 4 is desirable. In present study, adequate precision ratio of 48.488 indicates an

adequate signal. Hence this model can be used to navigate the design space. The response (Y)

should be checked for the maximum and minimum ratios. Generally, a ratio greater than 10

indicates a higher probability that the transformation of the response may improve the model

[20]. In present case, the ratio of maximum to minimum response (Y) is 1.75 indicating that

the transformation is not required. On these bases, it can be concluded that the selected model

was fairly adequate for predicting the yield of octyl levulinate and the predicted results were

satisfying.

3.3 Effect of process variables on yield of octyl levulinate

The results in Table 4, shows that interactions between variables have significant

effect on the yield of octyl levulinate. Therefore, instead of studying single variable (as in

conventional method) the interactions will be investigated which is significance for a

comprehensive optimization study. As mentioned earlier, model Eq. (1) can be presented in

the form of three dimensional response surface plots and two dimensional interaction plots

for the yield of octyl levulinate for various values of catalyst loading and molar ratio (Figure

4), catalyst loading and reaction temperature (Figure 5) and molar ratio and reaction

temperature (Figure 6). In all of these cases other two variables are maintained constant.

The effect of interaction between catalyst loading and molar ratio of octanol to LA at

constant reaction time of 4 h and reaction temperature of 383 K is presented in Figure 4. The

yield of octyl levulinate was higher than 56% when the catalyst loading was between 10 to

30% and the molar ratio of octanol to LA was from 4 to 8. Therefore it is evident that

esterification reaction is very much dependent on the amount of catalyst loading. More

11

catalyst reveals more active sites which participate in the reaction and catalyse the production

of lubricant [10]. At 10–30% catalyst loading, the yield of octyl levulinate was slightly

affected by the molar ratio. It seemed that the increase in octanol to LA molar ratio had less

effect on the yield of octyl levulinate at different catalyst amounts. This supports the result

that the octanol to LA molar ratio (p-value = 0.0002) was less significant parameter in

comparison with other parameters (value of p< 0.0001) (Table 3). In general, esterification

reaction of LA with alcohol is an endothermic reaction [15]. Reaction temperature plays a

crucial role in determining the reaction rate in esterification reaction which influences the

yield of octyl levulinate. For instance, higher temperature induces faster reaction rate

compared to lower temperature [21].

The effect of catalyst loading and reaction temperature on yield of octyl levulinate at a

constant molar ratio of 6 and reaction time of 4 h is shown in Figure 5. The yield of octyl

levulinate was influenced significantly by the catalyst loading and reaction temperature. At

low catalyst loading, the yield of octyl levulinate slightly increases with an increase in

reaction temperature. The yield of octyl levulinate increased as reaction temperature

increased at moderate levels of catalyst. It signifies that catalyst loading has bigger positive

effect on yield of octyl levulinate and is supported by high coefficient for linear term in

model equation (Eq. (1)). Figure 6 shows the effects of different of molar ratio of octanol to

LA and reaction temperature on the yield of octyl levulinate in three dimensional surface

response and two-dimensional interaction plots at constant catalyst loading of 20% and

reaction time of 4 h. From the figures, it is obvious that at any designated quantity of molar

ratio from 4 to 8, the yield of octyl levulinate increase proportionally with reaction

temperature. The observed phenomenon occurred as increasing the reaction temperature

enhanced the reaction rate of esterification reaction and eventually the yield of octyl

levulinate [22].

12

3.4 Optimization of process parameters for esterification

In order to generate optimal conditions for synthesis of octyl levulinate biolubricant,

numerical feature of the Design-Expert® Version 8.0.7.1 software was applied. The

independent parameters used in numerical optimization includes catalyst loading, molar ratio

of octanol to LA and reaction temperature were set within the range between low (-1) and

high (+1) while the yield of octyl levulinate was set to maximum value [23]. Table 5

summarizes the constraints used for the optimization of process parameters. Subsequently, 43

solutions for the optimum conditions were generated by the software and the solution with

the highest desirability and yield of octyl levulinate was selected to be verified by

experiments. The optimum conditions including the predicted and experimental yield of octyl

levulinate are shown in Table 6. The experimental value of yield of octyl levulinate

represented in table is the average of three independent experiments. The obtained average

optimum yield of octyl levulinate of 99% is well in agreement with the predicted value, with

a relatively insignificant error of 3.23%. As the experimental error is less than ±5%, it can be

concluded that the proposed statistical model was adequate for predicting the yield of octyl

levulinate.

3.5 Catalyst reusability

The reusability of Meso-HZ-5 catalyst was tested for octyl levulinate biolubricant

production at optimized process parameters obtained by RSM design (Table 5), with catalyst

loading of 25.4%, molar ratio (octanol to LA) of 7.56, reaction temperature of 393 K and

reaction time of 4 h (Figure 7). The Meso-HZ-5 catalyst activity to produce biolubricant was

observed to be stable for six cycles (fresh and five reuses) (Figure 7). After sixth cycle the

marginal decrease in yield of octyl levulinate 99 to 95% was observed. This concludes that,

catalyst is highly active, stable and reusable. The present method for production of octyl

levulinate biolubricant over Meso-HZ-5 catalyst offers greener methodology with potential

13

advantages with respect to higher yield of octyl levulinate of 99% and catalyst reusability for

six cycles without considerable loss in activity.

4 Conclusions

The synthesis of octyl levulinate biolubricant by esterification of renewable levulinic

acid (LA) over heterogeneous Meso-HZ-5 zeolite is presented probably for the first time.

Response surface methodology (RSM) design of experiments was carried out in order to

optimize process parameters of the esterification reaction in view to maximize the yield of

octyl levulinate. The RSM indicated that the catalyst loading and reaction temperature are

most significant parameters in esterification reaction.

The yield of octyl levulinate over Meso-HZ-5 at optimized process parameters was

found to be 99%. The Meso-HZ-5 catalyst was reusable for six cycles without considerable

loss in activity. This study confirms that Meso-HZ-5 is a highly active, stable and reusable

promissory solid acid catalyst for synthesis of octyl levulinate biolubricant. The present

approach represents a viable means of producing lubricant from biomass derived LA which is

renewable in nature and can be alternative to non-renewable mineral oil feedstocks.

Acknowledgements

The financial assistance from CSIR is acknowledged for CSIR-XII FYP Networking

Project BLB.

References

[1] Wai L, Li X, Yanxun L, Tingliang L, Guoji L. Chem Eng Technol, 2013, 36 (4): 559.[2] Salimon J, Salih N, Yousif E. Eur J Lipid Sci Technol, 2010, 112: 519.[3] Kumar A, Sharma S. Ind Crops Prod, 2008, 28: 1.[4] Rudnick L R. Synthetics, Mineral Oils and Bio-Based Lubricants Chemistry and

Technology. CRC Press, Taylor & Francis Group, Boca Raton, 2006. 362.[5] Bokade V V, Yadav G D. Proc Safety Environ Protect, 2007, 85: 372.[6] Akerman C O, Gaber Y, Ghani N A, Lamsa M, Hatti-Kaul R. J Mol Cat B: Enzymatic,

2011, 72: 263.[7] Ting C C, Chen C C. Measurement, 2011, 44: 1337.[8] Liao C C, Chung T W. Chem Eng Res Des, 2013, 1: 2457–2464[9] Singh A K. Ind Crops Prod, 2011, 33: 287.[10] Avisha C, Debarati M, Dipa B. J Chem Technol Biotechnol, 2013, 88: 139.

14

[11] Akerman C O, Hagstrom A E V, Mollaahmad M A, Karlsson S, Hatti-Kaul R. Process Biochem, 2011, 46: 2225.

[12] Yan K, Wu G, Wen J, Chen A. Catal Commun, 2013, 34: 58.[13] Rudnick L R. Synthetics, Mineral Oils, and Bio-Based Lubricants: Chemistry and

Technology, Taylor & Francis, CRC Press, Boca Raton, USA, 2006.[14] Shiralkar V P, Joshi P N, Eapen M J, Rao B S. Zeolites, 1991, 11: 511.[15] Nandiwale K Y, Sonar S K, Niphadkar P S, Joshi P N, Deshpande S S, Patil V S,

Bokade V V. Appl Catal A: Gen, 2013, 460– 461: 90.[16] Boey P L, Ganesan S, Maniam G P, Khairuddean M, Efendi J. Energy Convers Manage,

2013, 65: 392.[17] Gao Y Y, Chen W W, Lei H, Liu Y, Lin X, Ruan R. Biomass Bioenerg 2009, 33: 277.[18] Montgomery D C. Design and analysis of experiments, fifth ed., John Wiley and Sons,

2001.[19] Kishore D, Kayastha A M. Food Chem, 2012, 134: 1650.[20] Stat-Ease, Multifactor RSM tutorial (part 2-optimization). Design-expert software

version 7.1.5, user’s guide, 2008.[21] Saka S, Isayama Y. Fuel, 2009, 88: 1307.[22] Tan K T, Lee K T, Mohamed A R. Biomass Bioenerg, 2009, 33: 1096.[23] Tan K T, Lee K T, Mohamed A R, Bioresource Technol, 2010, 101: 965.

15

Figure captions:Figure 1 Characterization of H-ZSM-5 and Meso-HZ-5: (a) powder X-ray diffraction

(XRD) patterns and (b) N2 adsorption-desorption isotherms.Figure 2 Catalytic performance of H-ZSM-5 and Meso-HZ-5 catalyst for octyl levulinate

synthesis.Figure 3 Predicted versus experimental values of LA conversion (%) over DH-ZSM-5

catalyst.Figure 4 Response surface plot for octyl levulinate synthesis over Meso-HZ-5 catalyst as

a function of catalyst loading and molar ratio of octanol to LA at constantreaction time of 4 h and reaction temperature of 383 K.

Figure 5 Response surface plot for octyl levulinate synthesis over Meso-HZ-5 catalyst asa function of catalyst loading and reaction temperature at constant molar ratio of6 and reaction time of 4 h.

Figure 6 Response surface plot octyl levulinate synthesis over Meso-HZ-5 catalyst as afunction of molar ratio of octanol to LA and reaction temperature at constantcatalyst loading of 20% and reaction time of 4 h.

Figure 7 Reusability of Meso-HZ-5 catalyst for octyl levulinate synthesis at optimizedprocess parameters of catalyst loading of 25.4%, molar ratio of 7.56, reactiontemperature of 393 K and reaction time of 4 h.

10 20 30 40 50

Inte

nsity

2(degree)

(a)

H-ZSM-5

Meso-HZ-5

16

0.0 0.2 0.4 0.6 0.8 1.040

60

80

100

120

140

160

180

200

220

Ad

so

rbe

d A

mo

un

t (m

l/g)

Relative Pressure (p/p0)

H-ZSM-5 Meso-HZ-5

(b)

Figure 1

Figure 2

17

Figure 3

18

Fig. 4

19

Fig. 5

20

Figure 6

21

1 2 3 4 5 6 70

10

20

30

40

50

60

70

80

90

100Y

ield

of

Oc

tyl

Le

vu

lin

ate

(%

)

Number of Cycles

Figure 7

22

Table 1 Characterization of catalystsCatalyst Total

acidity(mmol g-1

)

BET surfacearea (m2 g-1

)

Volume (cm3/g) Porediameter

(Å)

Relativecrystallinity

(%)Meso Micro Total

H-ZSM-5 0.51 300.8 - - - 5.5 100Meso-HZ-5 0.73 427.6 0.128 0.190 0.318 29.78 97

23

Table 2 Selected variables and coded levels used in the Box-Behnken design.

Variables Symbol Coded levels

-1 0 +1

Catalyst Loading (wt. %) X1 10 20 30

Molar Ratio (octanol to LA) X2 4 6 8

Reaction Temperature (K) X3 373 383 393

24

Table 3 The Box-Behnken design matrix of the four variables in coded units and the

response values.

RunExperimental variables in

coded unitsExperimental variables in

natural unitsYield of Octyl

Levulinate, Y (%)X1 X2 X3 X1 X2 X3 Experimental Predicted

1 0 0 0 20 6 383 93 932 0 + + 20 8 393 98 96.633 0 0 0 20 6 383 93 934 - + 0 10 8 383 67 68.255 0 - - 20 4 373 70 71.356 0 0 0 20 6 383 93 937 0 - + 20 4 393 84 84.378 - - 0 10 4 383 64 63.59 + 0 - 30 6 373 69 68.8810 0 + - 20 8 373 71 70.6311 0 0 0 20 6 383 93 9312 - 0 + 10 6 393 65 65.1213 0 0 0 20 6 383 93 9314 - 0 - 10 6 373 56 55.1215 + 0 + 30 6 393 97 97.8816 + - 0 30 4 383 87 85.7517 + + 0 30 8 383 92 92.5

25

Table 4 ANOVA for response surface quadratic model.Source Sum of

squares Df Mean

squareF-value p-value

Prob > FModel 3148.99 9 349.89 264.78 < 0.0001 significantX1 1081.12 1 1081.12 818.15 < 0.0001X2 66.13 1 66.13 50.04 0.0002X3 760.5 1 760.5 575.51 < 0.0001X1X2 1 1 1 0.76 0.4132X1X3 90.25 1 90.25 68.3 < 0.0001X2X3 42.25 1 42.25 31.97 0.0008X1

2 631.84 1 631.84 478.15 < 0.0001X2

2 44.47 1 44.71 33.66 0.0007X3

2 341.05 1 341.05 258.09 < 0.0001Residual 9.25 7 1.32Lack of Fit 9.25 3 3.08Pure Error 0 4 0Cor Total 3158.24 16R2 = 0.9971; R2 (adjusted) = 0.9933; Df = Degree of freedom.

26

Table 5 The pre-set criteria for optimization of Meso-HZ-5 catalysed esterificationFactor/response Goal Lower limit Upper limitCatalyst loading (wt. %), X1 Is in range 10 30Molar ratio (octanol to LA), X2 Is in range 4 8Reaction temperature (K), X3 Is in range 373 393Yield of Octyl levulinate (%), Y Maximize 56 98

27

Table 6 Optimum process parameters for synthesis of octyl levulinate biolubricant and

validation model adequacyParameters Catalyst loading

(wt. %)Molar ratio

(octanol to LA)Reaction

temperature (K)Yield of octyllevulinate (%)

Predicted 25.36 7.56 393.14 102.23Experimental 25.4 7.56 393 99

28

Description portion:

Box-Behnken design with quadratic model was found to be adequate and statistically

accurate to optimize the yield of octyl levulinate biolubricant to 99% over Meso-HZ-5

catalyst.

29