www.elsevier.com/locate/ultsonch

Ultrasonics Sonochemistry 15 (2008) 95–100

Optimization of ultrasound extraction of phenolic compoundsfrom coconut (Cocos nucifera) shell powder

by response surface methodology

Sueli Rodrigues a,*, Gustavo A.S. Pinto b, Fabiano A.N. Fernandes c

a Departamento de Tecnologia de Alimentos, Universidade Federal do Ceara, Campus do Pici, Bloco 858, Caixa Postal 12168,

CEP 60021-970, Fortaleza-CE, Brazilb Embrapa Agroindustria Tropical, Rua Sara Mesquita 2270, Pici, 60511-110, Fortaleza-CE, Brazil

c Departamento de Engenharia Quımica, Universidade Federal do Ceara, Campus do Pici Bloco 709, CEP 60455-760, Fortaleza-CE, Brazil

Received 24 November 2006; received in revised form 23 January 2007; accepted 27 January 2007Available online 3 February 2007

Abstract

Coconut is a tropical fruit largely consumed in many countries. In some areas of the Brazilian coast, coconut shell represents morethan 60% of the domestic waste volume. The coconut shell is composed mainly of lignin and cellulose, having a chemical compositionvery similar to wood and suitable for phenolic extraction. In this work, the use of ultrasound to extract phenolic compounds from coco-nut shell was evaluated. The effect of temperature, solution to solid ratio, pH and extraction time were evaluated through a 24 experi-mental planning. The extraction process was also optimized using surface response methodology. At the optimum operating condition(30 �C, solution to solid ratio of 50, 15 min of extraction and pH 6.5) the process yielded 22.44 mg of phenolic compounds per gram ofcoconut shell.� 2007 Elsevier B.V. All rights reserved.

Keywords: Coconut; Phenolic extracts; Response surface methodology; Ultrasound extraction; Agriculture waste reduction

1. Introduction

Plant food processing may generate by-products rich inbioactive compounds, such as phenolic compounds [1],which may exhibit a wide range of physiological properties,such as anti-allergenic, anti-artherogenic, anti-inflamma-tory, anti-microbial, antioxidant, anti-thrombotic, cardio-protective and vasodilatory effects [2–5]. The beneficialeffects derived from phenolic compounds in human lifehave been attributed to their antioxidant activity. Theavailability of phenolic compounds from agricultural andindustrial residues have been reviewed by Moure et al.[6], and phenolic compounds with antioxidant activity havebeen identified in several agricultural by-products, such asrice hulls, buckwheat hulls and almond hulls [7–11].

1350-4177/$ - see front matter � 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.ultsonch.2007.01.006

* Corresponding author. Tel.: +55 85 3366 9656; fax: +55 85 3458 3407.E-mail address: sueli@ufc.br (S. Rodrigues).

From a process point of view the extraction of activeingredients from plants is essential if their active com-pounds are to be of prophylactic or therapeutic value tohuman subjects. Factors such as solvent composition,extraction time, extraction temperature, solvent to solidratio and extraction method influence the extraction effi-cacy [12–14]. Optimization of the process can be achievedby empirical or statistical methods and is essential for thecommercial application of the process.

Large amounts of coconut (Cocos nucifera) water andcoconut products are processed and consumed in Braziland many countries. This market for coconut products isincreasing and the volume of coconut shell waste corre-sponds to more than 60% of the domestic waste volumein some areas of the Brazilian coast. According to officialdata, the annual production of coconut shell is approxi-mately 6.7 million tons. Coconut shell is composed mainlyof lignin and cellulose being suitable as a phenolic source.

Nomenclature

R solution to solid ratioT temperature (�C)t time (min)

Table 1First experimental planning (full face centered central composite 24)

Run Temperature (�C) pH Ratio Time (min) Phenolicsa (mg/g)

1 30 4.5 20 20 5.72 ± 0.042 30 4.5 20 60 5.22 ± 0.473 30 4.5 50 20 13.13 ± 0.224 30 4.5 50 60 15.00 ± 2.785 30 6.5 20 20 12.09 ± 0.626 30 6.5 20 60 8.05 ± 0.217 30 6.5 50 20 21.37 ± 0.638 30 6.5 50 60 17.15 ± 0.159 60 4.5 20 20 13.70 ± 0.2710 60 4.5 20 60 17.99 ± 0.7311 60 4.5 50 20 12.24 ± .0012 60 4.5 50 60 18.05 ± 0.213 60 6.5 20 20 18.00 ± 0.9914 60 6.5 20 60 17.14 ± 0.2015 60 6.5 50 20 14.88 ± 0.8816 60 6.5 50 60 17.75 ± 0.2517 30 5.5 35 40 11.51 ± 0.5218 60 5.5 35 40 15.53 ± 0.3419 45 4.5 35 40 11.32 ± 1.0820 45 6.5 35 40 13.32 ± 0.2221 45 5.5 20 40 8.10 ± 0.1022 45 5.5 50 40 13.04 ± 0.9623 45 5.5 35 20 14.15 ± 0.4124 45 5.5 35 60 14.55 ± 0.4525 45 5.5 35 40 10.95 ± 0.6926 45 5.5 35 40 10.42 ± 0.5227 45 5.5 35 40 10.76 ± 1.13

a Mean values ± SD (n = 6).

96 S. Rodrigues et al. / Ultrasonics Sonochemistry 15 (2008) 95–100

Phenolics from coconut shell are similar to the woods usedto store and age alcoholic beverages. During the ageingprocess phenolic compounds are extracted by lignin ethan-olysis from the wooden casks. Phenolics are also used inthe food industry as anti-microbial agent, antioxidantagent and food stabilizer.

Nowadays, the use of coconut shell waste was onlyapplied in agriculture and in some industries which usescoconut fiber. The use of coconut shell as a source of chem-ical compounds, mainly to produce phenolic compounds, isnew. The use of ultrasound extraction instead of traditionalSohxlet method is increasing and its use has been investi-gated in the pharmaceutical, chemical and food industries.Ultrasound extraction became a good alternative extrac-tion method when compared to classical extraction meth-ods because of its high efficiency, low energy requirementand low water consumption (no reflux or refrigerationare needed). Besides, ultrasound assisted extraction is awell-established method in the processing of plant material,particularly to extract low molecular weight substances[15], and in the extraction of bioactive substances fromplants. The enhancement on the extraction process causedby ultrasound is attributed to the disruption of the cellwalls, reduction of the particle size and the enhancementon the mass transfer of the cell content to the solventcaused by the collapse of the bubbles produced by cavita-tions [16,17].

2. Materials and methods

2.1. Raw material

Coconut shells were washed chopped into smaller piecesand sun dried for 4 days. Then the material was disinte-grated in an industrial equipment and milled in a knife mill.The milled material was sun dried for more 4 days reachinga final moisture content of 20%. The fibers were separatedfrom the powder using a 1.5 mm nylon sieve. The powderwas classified by particle size distribution and stored atroom temperature.

2.2. Thermal treatment

Thermal treatment was carried out to enhance phenolicextraction from the coconut shell powder. Samples con-taining 20 g of coconut shell powder were submitted tothermal treatment (toasting) in a conventional drying ovenwith forced air circulation (Marconi model MA-085). Thepowder was toasted at 100 �C for 60 min.

2.3. Phenolic extraction

A full face centered central composite experimentalplanning (24) was carried out to study the effect of temper-ature, solvent, pH, solvent to solid ratio (g/g) and time ontotal phenolic content of the extracts (Table 1). Consider-ing the best condition found in the experimental planninga second experimental planning (full face centered centralcomposite 22) was carried out to optimize the extractionprocess. The software Statistica (Statsoft version 5.0) wasused to generate the experimental planning.

Phenolic compounds from the toasted powder wereextracted using a 50% (v/v) ethanol/water solution withpH adjusted with HCl to improve ethanolysis of woodcompounds. An open rectangular ultrasonic cleaning bath(Marconi Model USC, 25 kHz, 150 W) with of 2.7 L (inter-nal dimensions: 14 · 24 · 9 cm) was used to carry out theextractions. The equipment operated at an ultrasoundintensity of 4870 W/m2. The temperature was controlledand maintained at the desired value circulating externalwater from a thermostated water bath. Extractions were

Table 2Estimated effect of independent variables on total phenolic content (firstexperimental planning)

Factor Effect Sdt. Err t(12) p

Mean* 11.5804 0.2929 39.5384 4.42E�14T* 4.0127 0.3747 10.7100 1.70E�07T2* 3.0090 0.9913 3.0354 1.04E�02pH* 3.0485 0.3747 8.1367 3.16E�06pH2 0.6081 0.9913 0.6134 5.51E�01R* 4.0587 0.3747 10.8329 1.50E�07R2* �2.8960 0.9913 �2.9215 1.28E�02t 0.6176 0.3747 1.6483 1.25E�01t2 4.6685 0.9913 4.7096 5.06E�04T · pH* �1.7187 0.3974 �4.3249 9.88E�04T · R* �4.9426 0.3974 �12.4375 3.24E�08T · t* 2.3682 0.3974 5.9594 6.62E�05pH · R 0.0023 0.3974 0.0060 9.95E�01pH · t* �2.2189 0.3974 �5.5836 1.19E�04R · t* 0.9364 0.3974 2.3563 3.63E�02

* Significant factors at 95% of confidence level.

Table 3Analysis of variance of the regression model (Eq. (1))

Source ofvariation

Sum ofsquares

Degrees offreedom

Meansquare

F-value

Regression 394.14 14 28.15 44.68Residual 7.58 12 0.63Total 401.72 26Correlation

coefficient0.9811

F listed value(95%)

F14,12 = 2.63

Fig. 1. Fitted surface for total phenolic content as a function of pH(extraction time = 20 min, solvent to solid ratio = 50).

S. Rodrigues et al. / Ultrasonics Sonochemistry 15 (2008) 95–100 97

carried out using 1.5 g of the toasted powder and theproper ethanol solution volume. The extraction time ran-ged from 20 to 60 min. Experiments were done in replicate.

2.4. Determination of the total phenolic content in theextracts

The total phenolic content in the extracts was deter-mined by the Folin–Ciocalteau method [18] with somemodifications. Tannic acid was used as standard. All chem-icals were analytical grade and were purchased from Vetec(Vetec Quımica, Rio de Janeiro, Brazil).

In a 100 mL volumetric flask a fresh working solution wasprepared mixing 1.0 mL of a 2% (w/v) potassium sodiumtartarate (NaC4H4O6 Æ 4H2O) solution with 1.0 mL of a 1%(w/v) CuSO4 solution. The volume was completed to100 mL with a 0.1 M NaOH solution containing 2% (w/v)of CaCO3. The Folin–Ciocalteau reagent was diluted (1:3)with distilled water. One milliliter of the working solutionwas added to 250 lL of the sample. The assay tube was keptfor 10 min in the dark and then 100 lL of the diluted Folin–Ciocalteau reagent was added to the tube, which was kept formore 30 min in the dark. The absorbance was read at660 nm. The assay was performed at room temperature(28 �C). Samples were assayed in triplicate and the data weregiven as mean value ± SD.

2.5. Statistical analysis

Response surface methodology was used to analyze theresults. The software Statistica (Statsoft version 5.0) wasused to handle the results. Calculations were done at 95%of confidence level.

3. Results and discussion

About 56% of the powder presented mean diameterbetween 0.300 and 0.600 mm, followed by a fraction of36% mean diameter between 0.600 and 1.000 mm. Less than10% of the material presented mean diameter below 0.300 mm.

The first experimental planning is presented in Table 1.The phenolics extracted from the coconut shell powder ran-ged from 5.22 ± 0.47 to 21.37 ± 0.63 mg/g. Table 2 presentsthe estimated effects of each variable, as well as their inter-actions on extracted phenolic. The regression model fortotal phenolic content in the extracts is presented in Eq.(1). Table 3 depicts the ANOVA table of the fitted model:

Phenolic ðmg=gÞ¼ �18:36þ 7:36� 10�2T þ 6:70� 10�2T 2

þ 2:97pHþ 3:04� 10�1pH2 þ 1:02R� 6:41� 10�3R2

� 3:79� 10�1tþ 5:80� 10�3t2 � 5:73� 10�2T � pH

� 1:10� 10�2T � Rþ 3:90� 10�3T � tþ 1:10

� 10�4pH� R� 5:55� 10�2pH� t þ 1:60� 10�3R� t

ð1Þ

Most effects were significant at 95% of confidence level (Ta-ble 2). According to the ANOVA table, the regressionmodel was significant at the considered confidence levelsince a satisfactory correlation coefficient was obtained

98 S. Rodrigues et al. / Ultrasonics Sonochemistry 15 (2008) 95–100

and the F-value was more than three times the listed F-value [19]. Surface response graphs, obtained using the fit-ted model presented in Eq. (1), are presented in Figs. 1–6.

According to the surface graphs presented in Figs. 1–3,the phenolic extraction was maximized when high pH val-ues, high solvent to solid ratio and low temperature wereemployed. According to Figs. 4–6, increasing extractiontime did not increase the amount of phenolic that wereextracted. Thus high solvent to solid ratio (R = 50), highpH values (pH = 6.5) at low temperature (T = 30 �C) and20 min of ultrasound treatment maximized the amount ofphenolic that were extracted from the coconut shell powderaccording to the results obtained in the first experimentalplanning (Table 1). This condition corresponds to experi-mental run #7. To optimize the extraction process a second

Fig. 2. Fitted surface for total phenolic content as a function of solvent tosolid ratio and temperature (extraction time = 20 min, pH = 6.5).

Fig. 3. Fitted surface for total phenolic content as a function of solvent tosolid ratio and pH (extraction time = 20 min, temperature = 30 �C).

Fig. 5. Fitted surface for total phenolic content as a function of solvent tosolid ratio and extraction time (pH = 6.5, temperature = 30 �C).

Fig. 4. Fitted surface for total phenolic content as a function of extractiontime and pH (solvent to solid ratio = 50, temperature = 30 �C).

experimental planning was carried out with extraction timeranging from 5 to 20 min and solvent to solid ratio rangingfrom 40 to 60, respectively. The experiments were per-formed at 30 �C and pH 6.5. The experimental planningand the results are presented in Table 4. Table 5 presentsthe estimated effects of each variable, as well as their inter-actions on the total phenolic extracted. The regressionmodel for total phenolic content in the extracts is presentedin Eq. (2). Table 6 depicts the ANOVA table of the fittedmodel. Fig. 7 presents the surface response graph obtainedusing the fitted model (Eq. (2)):

Phenolic ðmg=gÞ¼ �57:49þ 2:47R� 2:10� 10�2R2 þ 2:06t

� 1:59� 10�2t2 � 2:5� 10�2R� t ð2Þ

Fig. 6. Fitted surface for total phenolic content as a function of extractiontime and temperature (pH = 6.5, solvent to solid ratio = 50).

Table 4Second experimental planning (face centered central composite 22)

Run Ratio Time (min) Phenolicsa (mg/g)

1 40 5.0 12.74 ± 0.102 40 20.0 22.25 ± 0.153 60 5.0 17.35 ± 0.314 60 20.0 19.13 ± 1.405 40 12.5 17.53 ± 0.706 60 12.5 18.80 ± 1.357 50 5.0 16.39 ± 0.898 50 20.5 22.44 ± 1.429 50 12.5 20.39 ± 0.39

10 50 12.5 21.80 ± 0.47

a Mean values ± SD (n = 6).

Table 5Estimated effect of independent variables on total phenolic content(second experimental planning)

Factor Effect Sdt. Err t(4) p

Mean* 20.6969 0.4481 46.1840 1.31E�6R 0.9512 0.6123 1.5535 1.95E�1R2* �4.2533 0.9818 �4.3321 1.23E�2t* 5.8118 0.6122 9.4921 6.87E�4t2 �1.7557 0.9818 �1.7882 1.48E�1R · t* �3.8147 0.7499 �5.0871 7.04E�3

* Significant factors at 95% of confidence level.

Table 6Analysis of variance of the regression model (Eq. (2))

Source ofvariation

Sum ofsquares

Degrees offreedom

Meansquare

F-value

Regression 80.16 5 16.03 27.64Residual 2.30 4 0.58Total 82.46 9Correlation

coefficient0.9721

Listed F-value(95%)

F4,5 = 6.26

Fig. 7. Fitted surface for total phenolic content as a function of extractiontime and solvent to solid ratio (pH = 6.5, temperature = 30 �C).

S. Rodrigues et al. / Ultrasonics Sonochemistry 15 (2008) 95–100 99

Only the linear effect of solvent to solid ratio and the qua-dratic effect of the extraction time were not significant atthe considered confidence level (95%). According to theANOVA analysis the fitted model was significant becausethe F-value was higher than three times the listed F-value[19]. According to Fig. 7 the extraction time presentedthe highest effect on phenolic extraction when the processwas carried out at 30 �C and pH 6.5. The experimental con-dition that maximized phenolic extraction was found atextraction times between 15 and 20 min.

Mass transfer controls solvent extraction of any compo-nent from a plant matrix. When the solvent saturates withthe extracted compound the concentration gradientbecomes null and the phenomena stops. In ultrasoundextraction of phenolics from coconut shell powder masstransfer stops after 20 min and the process can be inter-rupted. Solvent to solid ratio for values above 50 slightlyincreased the amount of phenolic that was extracted.According to Balanchandran et al. [20] sonication leadsto an increase in the effective diffusivity of the mass transferprocess and this effect is maximum at short sonicationtimes as was verified for phenolics extraction for coconutshell powder.

An amount of 7.20 ± 0.28 mg/g of phenolics wasobtained in a control experiment carried out at the opti-mized operating condition (solvent to solid ratio of 50,pH of 6.5 and extraction time of 20 min) without sonica-tion. This control experiment showed that the use of soni-cation enhanced the extraction yield by three folds.

4. Conclusion

Coconut shell powder is a suitable phenolic source. Theoptimum operating condition that maximizes the extraction

100 S. Rodrigues et al. / Ultrasonics Sonochemistry 15 (2008) 95–100

of phenolic compounds by ultrasound was found using a50% (v/v) ethanol:water solution, pH adjusted to 6.5, sol-vent to solid ratio of 50 and the process carried out at30 �C. The process required 15 min of sonication at ultra-sound intensity of 4870 W/m2. At this condition22.44 mg/g of phenolics were extracted from the coconutshell.

References

[1] A. Schieber, P. Hilt, P. Streker, H.U. Endreß, C. Rentschler, R. Carle,Inn. Food Sci. Emerg. Technol. 4 (2003) 99.

[2] O. Benavente-Garcia, J. Castillo, J. Lorente, A. Ortunio, J.A. DelRio, Food Chem. 68 (2000) 457.

[3] C. Manach, A. Mazur, A. Scalbert, Curr. Opin. Lipidol. 16 (2005) 77.[4] E. Middleton, C. Kandaswami, T.C. Theoharides, Pharmacol. Rev.

52 (2000) 673.[5] K.E. Heim, A.R. Tagliaferro, D.J. Bobilya, J. Nutr. Biochem. 13

(2002) 572.[6] A. Moure, J.M. Cruz, D. Franco, J.M. Dominguez, J. Sineiro, H.

Dominguez, M.J. Nunez, J.C. Parajo, Food Chem. 72 (2001) 145.

[7] N. Ramarathnam, T. Osawa, M. Namiki, S. Kawakishi, J. Agric.Food Chem. 37 (1989) 316.

[8] M. Watanabe, Y. Ohshita, T. Tsushida, J. Agric. Food Chem. 45(1997) 1039.

[9] G.R. Takeoka, L.T. Dao, J. Agric. Food Chem. 51 (2002) 496.[10] S. Bryngelsson, B. Mannerstedt-Fogelfors, A. Kamal-Eldin, R.

Andersson, L.H. Dimberg, J. Sci. Food Agric. 82 (2002) 606.[11] N. Balasundram, K. Sundram, S. Samman, Food Chem. 99 (2006)

191.[12] M. Wettasinghe, F. Shahidi, J. Agric. Food Chem. 47 (1999) 1801.[13] J.E. Cacace, G. Mazza, J. Agric. Food Chem. 50 (2002) 5939.[14] J.E. Cacace, G. Mazza, J. Food Sci. 68 (2003) 240.[15] Z. Hromadkova, A. Ebringerova, P. Valachovic, Ultrason. Sono-

chem. 9 (2002) 37.[16] L. Paniwnyk, E. Beaufoy, J.P. Lorimer, T.J. Mason, Ultrason.

Sonochem. 8 (2001) 299.[17] M.H. Entezani, S.H. Nazary, M.H.H. Khodaparast, Ultrason.

Sonochem. 11 (2004) 379.[18] E. Ragazzi, G. Veronese, J. Chromatogr. 77 (1973) 369.[19] S. Rodrigues, L.M.F. Lona, T.T. Franco, J. Food Eng. 75 (2006) 433.[20] S. Balanchandran, S.E. Kentish, R. Mawson, M. Ashokkumar,

Ultrason. Sonochem. 13 (2006) 471.