january-march 2014 vol.20, number 1, 1-153

ISSN 1451 - 9372(Print)ISSN 2217 - 7434(Online)JANUARY-MARCH 2014Vol.20, Number 1, 1-153

www.ache.org.rs/ciceq

Journal of the Association of Chemical Engineers of Serbia, Belgrade, Serbia

EDITOR-In-Chief

Vlada B. Veljković Faculty of Technology, University of Niš, Leskovac, Serbia

E-mail: [email protected]

ASSOCIATE EDITORS Branko Bugarski

Department of Chemical Engineering, Faculty of Technology and Metallurgy,

University of Belgrade, Belgrade, Serbia

Jonjaua RanogajecFaculty of Technology, University of Novi

Sad, Novi Sad, Serbia

Srđan Pejanović Department of Chemical Engineering, Faculty of Technology and Metallurgy,

University of Belgrade, Belgrade, Serbia Milan Jakšić

ICEHT/FORTH, University of Patras, Patras, Greece

EDITORIAL BOARD (Serbia) Đorđe Janaćković, Sanja Podunavac-Kuzmanović, Viktor Nedović, Sandra Konstantinović, Ivanka Popović

Siniša Dodić, Zoran Todorović, Olivera Stamenković, Marija Tasić, Jelena Avramović

ADVISORY BOARD (International)

Dragomir Bukur

Texas A&M University, College Station, TX, USA

Milorad Dudukovic

Washington University, St. Luis, MO, USA

Jiri Hanika

Institute of Chemical Process Fundamentals, Academy of Sciences of the Czech Republic, Prague, Czech Republic

Maria Jose Cocero

University of Valladolid, Valladolid, Spain

Tajalli Keshavarz

University of Westminster, London, UK

Zeljko Knez

University of Maribor, Maribor, Slovenia

Igor Lacik

Polymer Intitute of the Slovak Academy of Sciences, Bratislava, Slovakia

Denis Poncelet

ENITIAA, Nantes, France

Ljubisa Radovic

Pen State University, PA, USA

Peter Raspor

University of Ljubljana, Ljubljana, Slovenia

Constantinos Vayenas

University of Patras, Patras, Greece

Xenophon Verykios

University of Patras, Patras, Greece

Ronnie Willaert

Vrije Universiteit, Brussel, Belgium

Gordana Vunjak Novakovic

Columbia University, New York, USA

Dimitrios P. Tassios

National Technical University of Athens, Athens, Greece

Hui Liu

China University of Geosciences, Wuhan, China

FORMER EDITOR (2005-2007) Professor Dejan Skala

University of Belgrade, Faculty of Technology and Metallurgy, Belgrade, Serbia

Journal of the Association of Chemical Engineers of Serbia, Belgrade, Serbia

Vol. 20 Belgrade, January-March 2014 No. 1

Chemical Industry & Chemical EngineeringQuarterly (ISSN 1451-9372) is published

quarterly by the Association of ChemicalEngineers of Serbia, Kneza Miloša 9/I,

11000 Belgrade, Serbia

Editor:Vlada B. Veljković

[email protected]

Editorial Office:Kneza Miloša 9/I, 11000 Belgrade, Serbia

Phone/Fax: +381 (0)11 3240 018E-mail: [email protected]

www.ache.org.rs

All the manuscripts are not to be returned

For publisher:Tatijana Duduković

Secretary of the Editorial Office:Slavica Desnica

Marketing and advertising:AChE Marketing Office

Kneza Miloša 9/I, 11000 Belgrade, SerbiaPhone/Fax: +381 (0)11 3240 018

Publication of this Journal is supported by theMinistry of Education and Science of the

Republic of Serbia

Subscription and advertisements make payableto the account of the Association of Chemical

Engineers of Serbia, Belgrade, No. 205-2172-71, Komercijalna banka a.d., Beograd

Computer typeface and paging:Vladimir Panić

Printed by:Faculty of Technology and Metallurgy,

Research and Development Centre of PrintingTechnology, Karnegijeva 4, P. O. Box 3503,

11120 Belgrade, Serbia

Abstracting/Indexing:Articles published in this Journal are indexed inThompson Reuters products: Science Citation

Index - ExpandedTM - access via Web of Science®, part of ISI Web of KnowledgeSM

CONTENTS

Maja Lj. Bulatović, Marica B. Rakin, Ljiljana V. Mojović, Svetlana B. Nikolić, Maja S. Vukašinović Sekulić, Aleksandra P. Đukić Vuković, Improvement of pro-duction performance of functional fermented whey- -based beverage ..................................................................... 1

Ramchandra Pandey, Pravin O. Patil, Sanjay B. Bari, Dinesh M. Dhumal, Simultaneous estimation of etodolac and thiocolchicoside in bulk and in tablet formulation by UV-spectrophotometry ........................................................... 9

B. Jayalakshmi, K.A. Raveesha, K.N. Amruthesh, Evaluation of antibacterial and antioxidant potential of Euphorbia cotinifolia Linn. leaf extracts ................................................. 19

Jian Ding, Huihui Wang, Keke Dai, Yuhua Zi, Zhongping Shi, Prediction of porcine interferon α antiviral activity in fermentation by Pichia pastoris based on multivariable regression and artificial neural network ................................ 29

Shweta S. Havele, Sunil R. Dhaneshwar, Determination of glibenclamide, metformin hydrochloride and rosi-glitazone maleate by reversed phase liquid chro-matographic technique in tablet dosage form ...................... 39

Maria Valderez Ponte Rocha, Jocélia Sousa Mendes, Maria Estela Aparecida Giro, Vânia Maria M. Melo, Luciana Rocha Barros Gonçalves, Biosurfactant production by Pseudomonas aeruginosa MSIC02 in cashew apple juice using a 24 full factorial experimental design ................ 49

Miodrag N. Tekić, Ivana M. Šijački, Milenko S. Tokić, Predrag S. Kojić, Dragan Lj. Petrović, Nataša Lj. Lukić, Svet-lana S. Popović, Hydrodynamics of a self-agitated draft tube airlift reactor .................................................................. 59

Vesna Nađalin, Žika Lepojević, Mihailo Ristić, Jelena Vladić, Branislava Nikolovski, Dušan Adamović, Investigation of cultivated lavender (Lavandula officinalis L.) extraction and its extracts ..................................................... 71

Majid Mazhar, Majid Abdouss, Zahra Shariatinia, Mojdeh Zargaran, Graft copolymerization of methacrylic acid monomers onto polypropylene fibers .................................... 87

Mohammad A. Behnajady, Shahrzad Yavari, Nasser Modir-shahla, Investigation on adsorption capacity of TiO2- -P25 nanoparticles in the removal of a mono-azo dye from aqueous solution: A comprehensive isotherm analysis ................................................................................. 97

Kulandaivelu Karunakaran, Gurusamy Navaneethan, Elango

Kuppanagounder Pitchaimuthu, A validated stability-

Contents continued indicating RP-HPLC method for paracetamol and lornoxicam: application to pharmaceutical dosage forms .... 109

A. Pobudkowska, U. Domańska, Study of pH-dependent drugs solubility in water ...................................................... 115

Abhishek Kumar Singh, Mausumi Mukhopadhyay, Response surface methodology for optimizing the glycerolysis reaction of olive oil by Candida rugosa lipase ...................... 127

Nataša Nedeljković, Marijana Sakač, Anamarija Mandić, Đorđe Psodorov, Dubravka Jambrec, Mladenka Pes-torić, Ivana Sedej, Tamara Dapčević Hadnađev, Rheo-logical properties and mineral content of buckwheat enriched wholegrain wheat pasta ........................................ 135

S.M. Peyghambarzadeh, A. Hatami, A. Ebrahimi, S.A. Alavi Fazel, Photographic study of bubble departure diameter in saturated pool boiling to electrolyte solutions ............................................................................. 143

Chemical Industry & Chemical Engineering Quarterly

Available on line at

Association of the Chemical Engineers of Serbia AChE www.ache.org.rs/CICEQ

Chem. Ind. Chem. Eng. Q. 20 (1) 1−8 (2014) CI&CEQ

1

MAJA LJ. BULATOVIĆ MARICA B. RAKIN

LJILJANA V. MOJOVIĆ SVETLANA B. NIKOLIĆ

MAJA S. VUKAŠINOVIĆ SEKULIĆ

ALEKSANDRA P. ĐUKIĆ VUKOVIĆ

Faculty of Technology and Metallurgy, University of Belgrade,

Belgrade, Serbia

SCIENTIFIC PAPER

UDC 637.344:663.1:579.864

DOI 10.2298/CICEQ120715096B

IMPROVEMENT OF PRODUCTION PERFORMANCE OF FUNCTIONAL FERMENTED WHEY-BASED BEVERAGE

Article Highlights • Effect of culture mixing and different fermentation temperature on fermentation time,

titratable acidity, total cell count, aroma and storage stability of functional whey-based beverage were investigated

• Culture mixing contributes to the aroma improvement • Temperature has significant influence on the fermentation dynamic and cell viability

during the storage of produced beverage • A beverage produced by mixed culture of Lactobacillus helveticus ATCC 15009 and

Streptococcus thermophilus S3 at 42 °C achieved high storage stability with a shelf life of 22 days

Abstract

The aim of this study was improvement of the performances for the production of whey-based beverages with highly productive strains of Lactobacillus. Individual or mixed cultures containing Lactobacillus helveticus ATCC 15009, Lactobacillus delbrueckii ssp. lactis NRRL B-4525 and Streptococcus ther-mophilus S3 were studied. The scientific hypothesis was that production per-formances, especially aroma and viable cell count, are positively affected by the strains combination and temperature. Based on the results, beverages obtained by mixed cultures Lb. helveticus ATCC 15009 - S. thermophilus S3 and Lb. delbrueckii ssp. lactis - S. thermophilus S3 had higher aroma values than beverages obtained by individual strains. The symbiosis of tested strains had a positive impact on the aroma of produced beverage. In addition, the temperature had significant effects on cell viability during storage and fermen-tation dynamics. The beverages produced by mixed cultures Lb. helveticus ATCC 15009 - S. thermophilus S3 and Lb. delbrueckii ssp. lactis - S. thermo-philus S3 at 42 °C achieved higher storage stability (19 to 22 days) than beverages produced at 37 and 45 °C (13 to 19 days). Subsequently, at 42 °C fermentation time for both mixed cultures was 1.5 h shorter, compared to the time achieved at 37 °C.

Keywords: whey, functional beverages, probiotics, Lactobacillus, fermentation, stability.

Whey is a major by-product of the cheese industry often disposed as waste, causing high envi-ronmental contamination because of high COD (57- -75 g/L) [1] and BOD5 (35-40 g/L) [2] values, which is completely at odds with the nutritional potential that this material possesses. Considerable efforts have

Correspondence: M. Rakin, Faculty of Technology and Metal-lurgy, University of Belgrade, Karnegijeva 4, 11000 Belgrade, Serbia. E-mail: [email protected] Paper received: 15 July, 2012 Paper revised: 18 October, 2012 Paper accepted: 22 October, 2012

been made over the past years to find new outlets of whey utilization in terms to reduce environmental pollution [3-5]. Whey itself has the ability to act as an antioxidant, antihypertensive, antitumor, hypolipide-mic, antiviral and chelating agent [6]. Thus, in recent years the bioconversion of whey has become an inte-resting process from the viewpoint of human nutrition, especially for therapeutic purposes, in regard to eco-nomy, and with advantage for reducing pollution [7].

As main nutritive components, whey contains: 0.50–0.55% of beta-lactoglobulin (source of essential and branched chain amino acids), 0.20–0.25% of

M.Lj. BULATOVIĆ et al.: IMPROVEMENT OF PRODUCTION PERFORMANCE… Chem. Ind. Chem. Eng. Q. 20 (1) 1−8 (2014)

2

alpha-lactoalbumin (primary protein found in human breast milk also of essential and branched chain amino acids), 0.10-0.15% of immunoglobulins (pri-mary protein found in colostrum with immune modul-ating benefits), 0.01-0.02% of lactoferrin (antioxidant, antibacterial, antiviral and antifungal agent which pro-motes growth of beneficial bacteria), 0.005% of lacto-peroxidase (inhibits growth of bacteria), 0.05-0.10% of bovine serum albumin (large protein which is source of essential amino acids) and 0.10-0.15% of glycomacropeptide (source of branched amino acids that lacks aromatic amino acids such as phenyl-alanine, tryptophan and tyrosine) [6].

The fermentation of whey by lactic acid bacteria allows the production of beverages with significantly improved characteristics. Fermented whey contains: a) lactic acid and possibly antimicrobial compounds important for maintaining of intestinal microflora; b) flavor compounds (e.g., acetaldehyde in yoghurt and cheese) and other metabolites (e.g., extracellular polysaccharides) that will provide a product with the organoleptic properties desired by the consumer; c) free amino acids and vitamins which improve the nutritional value of whey; d) substances that provide a special therapeutic or prophylactic effect against can-cer and control of serum cholesterol levels [8-12].

Numerous strains of Lactobacillus genera are already known as highly productive in lactic acid fermentation [13]. In addition to their role in fermen-tation processes, some of these lactic acid bacteria have been studied as dietary sources of substances destined to promote a positive impact in the host by improving the health benefits. The strains Lacto-bacillus helveticus and Lactobacillus delbrueckii ssp. lactis, beside high fermentation productivity, have recently been considered as important bacteria for human health. Therefore, the main effect attributed to strain L. helveticus includes the production of whey hydrolysates that contains potent angiotensin I-con-verting enzyme (ACE) inhibitory peptides, with a high inhibition rate [14,15]. The strain Streptococcus ther-mophilus S3 is also marked as a good producer of exopolysaccharides [16], which could be considered as prebiotics because of their positive impact on human gut microflora. Also, these exopolysaccha-rides can enhance the viability of probiotic bacteria in cases when they are present in beverages. High pro-ductivity of these strains is very important in terms of profitability of the beverage production process. The use of highly productive strains shortens the fermen-tation time, which can significantly decrease the beve-rage production costs and valorize whey from cheese production what is important in term of production generated income.

In addition, from the consumer’s point of view, the manufacture of whey-based beverages through lactic fermentation must provide desirable sensory profiles of product [17,18]. This is not always the case, especially when the fermentation is performed with highly productive strains that produce high level of lactic acid and substances with unacceptable odour and taste. Combining different species or strains can lead to a significant improvement of production perfor-mance due to symbiotic interaction between micro-organisms [19]. There are many combinations of mic-roorganisms that can provide production of beverages with satisfactory sensory characteristics, and the necessary beverage production criteria such as low cost of production, functionality and storage stability. However, this area has not been fully explored.

The aim of this study was improvement of the performances for the production of functional whey- -based beverages with highly productive strains of Lactobacillus genera. The scientific hypothesis was that whey-based beverage performances, especially aroma and viable cell count are positively affected by the strain combination and appropriate temperature. Influence of fermentation temperature on the stability of viable cell count during the storage of whey-based beverage was also investigated.

MATERIALS AND METHODS

Microorganisms and media

The strains L. helveticus (ATCC 15009), L. del-brueckii ssp. lactis (NRRL B-4525) and S. thermo-philus (S3) used in this work were obtained from the Culture Collection of Department of Biochemical Engi-neering and Biotechnology, Faculty of Technology and Metallurgy, Belgrade, Serbia. Stock cultures were stored at -20 °C in 3 mL vials containing MRS broth and 50% (v/v) glycerol as a cryoprotective agent. For the preparation of laboratory cultures, a drop of stock culture were transferred in 3 mL of the MRS broth and incubated for 18 h under anaerobic conditions at the optimal growth temperature (37 °C). All working cul-tures were pre-cultured twice in an MRS broth prior to experimental use.

Whey fermentation

Sweet whey powder (Lenic Laboratories, Bel-grade, Serbia), with following composition: proteins 12.11%, fat 1.0%, and carbohydrates 69.62%, was reconstituted to contain 8% of dry matter. A volume of 300 mL of the reconstituted whey with pH 6.2 was poured into sterile glass bottles of 500 mL. Samples were pasteurized at 60 °C for 60 min, cooled at fer-


3

mentation temperature (37, 42 and 45 °C) and imme-diately inoculated by adding 2% (v/v) of individual strains or mixed cultures. For the preparation of mixed cultures highly productive strains L. helveticus and L. delbrueckii ssp. lactis were mixed with strain S. thermophilus in ratio 1:1. The fermentations were car-ried out until pH 4.6 was attained. During the incu-bation time samples were withdrawn every 1 h for determination of pH value. When pH 4.6 was reached, fermentations were stopped by quick cooling. The resulting beverages were distributed in sterile plastic bottles in triplicates and stored at 4 °C for 35 days. Viable cell count (log CFU/mL) was determined every 5 days of the storage.

Chemical and microbiological analysis

Acidity of whey samples was analyzed as pH and titratable acidity. The pH was measured by a pH meter (WTW pH 720), and titratable acidity (°SH) by the Soxhlet-Henkel method [20].

The fermented samples were analyzed for viable cell count by pour plate technique on MRS agar [21].

Sensory evaluation

Aroma as the most pronounced sensory charac-teristic was analyzed by a panel group of 5 sensory analysts and evaluated with grades from 1 to 5. For aroma evaluation, the following scale was used: 1 – on sauerkraut, 2 – on sourdough, 3 – on whey, 4 – on mild yogurt, 5 – on yogurt [22].

The experiments were done in triplicate, and the results are shown as average values.

RESULTS AND DISCUSSION

Effect of culture mixing on beverage production performances

In order to investigate the possibilities of impro-vement of beverage production performances the fermentation with individual and mixed cultures was performed. Fermentations were carried out at 37 °C, statically. Production performances were evaluated by determining the fermentation time (h), titratable acidity (°SH), viable cell count (log CFU/mL) and aroma value.

As shown in Figure 1, both assayed strains showed relatively short fermentation time between 5.0 and 6.5 h. Short fermentation time can substantially decrease the beverage production costs and valorize whey from cheese production. The fermentation with L. helveticus was 1.5 h shorter than the fermentation with L. delbrueckii ssp. lactis. Based on the obtained results, it could be said that both strains are highly productive, which is very important in terms of pro-duction costs. Beside shorter fermentation time, strain L. helveticus showed higher titratable acidity (11.4 °SH) and viable cell count (8.0 log CFU/mL) than the strain L. delbrueckii ssp. lactis (10.4 °SH and 6.6 log CFU/mL). It is interesting to note that the strain L. delbrueckii ssp. lactis showed very low cell growth in addition to high amount of produced lactic acid. This

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18 Fermentation time, h Titratable acidity, °SH Viable cell count, log(CFU/mL) Aroma value

Lb. helveticus ATCC 15009 Lb. delbrueckii ssp. lactis NRRL B-4525

Figure 1. Fermentation time, titratable acidity, total cell count and aroma value of whey fermented by individual strains Lactobacillus helveticus ATCC 15009 and Lactobacillus delbrueckii ssp. lactis NRRL B-4525.


4

could be due to differences in lactic acid production abilities and growth characteristics, which, in the case of this strain, are independent of each other. Both strains produced beverages with unpleasant aroma and exhibited low aroma values. Strain L. helveticus exhibited value 1 for aroma which was on sauerkraut, while strain L. delbrueckii ssp. lactis exhibited value 2 for aroma which was on sourdough. This could be because the tested strains do not produce diacetyl, which has long been known as a major contributor to aroma in fermented dairy products [23]. In addition, both strains produced beverages without presence any type of precipitate and sour-salty taste because of high amount of present lactic acid.

To improve production performances, individual strains were mixed with strain S. thermophilus. Since both mixed cultures were prepared by adding the same amount of S. thermophilus, a proportional change in fermentation time was expected. But, as shown in Figure 2, the fermentation time with mixed culture L. delbrueckii ssp. lactis-S. thermophilus did not change, while the fermentation with mixed culture L. helveticus-S. thermophilus was longer for 1.5 h than the fermentation with individual strain. Different impact of the culture mixing on the fermentation time is probably a consequence of different proteolytic acti-vity of used Lactobacillus strains. According to Pes-cuma et al. [24], species L. delbrueckii have high

proteolytic activity, which provides amino acids that are a growth factor of many microorganisms including S. thermophilus. On the other hand, L. helveticus requires the presence of certain amino acids for growth and fermentation activity [25] that cannot be provided by S. thermophilus. It could be assumed that L. delbrueckii ssp. lactis has higher proteolytic activity than L. helveticus and probably was able to provide the amino acids corresponding to strain S. thermo-philus. Therefore, the symbiosis of strains Lb. del-brueckii ssp. lactis and S. thermophilus proved to be better. The reduced amount of L. delbrueckii ssp. lactis in the inoculum can be compensated by the addition of S. thermophilus so that the fermentation time remains the same. On the other hand, the sym-biosis of strains L. helveticus and S. thermophilus gives mixed culture with lower fermentation activity which leads to a prolongation of the fermentation. In contribution to this assumption, a higher titratable activity (13 °SH) was achieved in the sample fer-mented with mixed culture L. delbrueckii ssp. lactis-S. thermophilus. This also could be due to presence of amino acids that promote the metabolism of S. ther-mophilus strain.

The sensory characteristics are a very important factor for product placement. As shown in Figure 2, the aroma value was higher (value 5 – on yogurt) in both mixed cultures compared with the values shown

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

Lb. helveticus ATCC 15009

- S. thermophilus S3

Lb. delbrueckii ssp. lactis NRRL B-4525 - S. thermophilus S3

Fermentation time, h

Titratable acidity, oSH Viable cell count, log(CFU/mL) Aroma value

Figure 2. Fermentation time, titratable acidity, viable cell count and aroma value of whey fermented by mixed cultures Lactobacillus helveticus ATCC 15009-Streptococcus thermophilus S3 and Lactobacillus delbrueckii ssp. lactis NRRL B-4525-Streptococcus

thermophilus S3.


5

in Figure 1 obtained by individual strains. This is probably due to presence of S. thermophilus S3 that is marked as a good producer of exopolysaccharides and probably diacetyl. Also, produced exopolysac-charides possibly induce L. helveticus and L. del-brueckii ssp. lactis metabolism to produce diacetyl, a major contributor of flavor and aroma. In this regards, it could be said that the culture mixing in both cases contributes significantly to improvement of sensory characteristics.

It must be considered that the national con-sumer preferences could also play an important role in sensory analysis. There are not too many whey-based drink products in the Serbian market and there is no tradition of drinking whey and whey-based beve-rages in Serbia. These products are more widely available and accepted by people in countries such as Germany, Austria, and Switzerland where there is a longer tradition of whey consumption [26]. In this respect, this study favorizes the beverages with sen-sory characteristics that are similar to the fermented products present on Serbian market to which the con-sumers are already accustomed.

Effect of incubation temperature on production performances and beverage storage stability

In order to improve the beverage production performances, the influence of various incubation temperatures on fermentation parameters and sto-rage stability was investigated. Fermentations were performed at three different temperatures by using mixed cultures, statically. The data on the effect of temperature of 37, 42 and 45 °C on fermentation time (h), titratable acidity (°SH), viable cell count (log CFU/mL) and aroma value of the beverage are pre-sented in Tables 1 and 2. Table 1 compares fermen-tation time, titratable acidity, viable cell count and aroma value of whey fermented by mixed culture L. helveticus-S. thermophilus at different temperatures. Table 2 compares fermentation time, titratable acidity, viable cell count and aroma value of whey fermented by mixed culture L. delbrueckii ssp. lactis-S. thermo-philus at different temperatures.

As shown in Tables 1 and 2, for both mixed cultures the fermentation time at 42 and 45 °C was decreased for 1.5 h, compared to the time achieved at 37 °C. Therefore, the temperature increase has a significant influence on the fermentation dynamics. Reduced fermentation time reduces costs of the pro-duction and valorizes obtained product. Subse-quently, the temperature increase leads to decrease of titratable acidity and increase of viable cell count in both samples. Temperature of 40–45 °C is optimal for

S. thermophilus which is probably one of the reasons for faster whey fermentation. Both samples had excel-lent aroma values at all temperatures. With respect to viable cell count as a main fermentation parameter, the best results were observed at a temperature of 45 °C for both mixed cultures.

Table 1. Effect of incubation temperature on fermentation time, titratable acidity, viable cell count and aroma value of beverage fermented by mixed culture Lactobacillus helveticus ATCC 15009 – Streptococcus thermophilus S3

Parameter Incubation temperature, °C

37 42 45

Fermentation time, h 6.5 5.0 5.0

Titratable acidity, °SH 11.4 11.2 11.0

Viable cell count, log(CFU / mL) 8.30 8.38 8.60

Aroma value 5 5 5

Table 2. Effect of incubation temperature on fermentation time, titratable acidity, viable cell count and aroma value of beverage fermented by mixed culture Lactobacillus delbrueckii ssp. lactis NRRL B-4525–Streptococcus thermophilus S3

Parameter Incubation temperature, °C

37 42 45

Fermentation time, h 6.5 5.0 5.0

Titratable acidity, °SH 13.0 11.4 11.2

Viable cell count, log(CFU / mL) 8.00 8.25 8.36

Aroma value 5 5 5

To examine the effect of incubation temperature on storage stability of the fermented product, the beverages were kept in the refrigerator at 4 °C for 35 days. Stability was evaluated after 0, 5, 10, 15, 20, 25, 30 and 35 days of the storage by determining the pH value, titratable acidity, viable cell count and aroma values. Data for pH, titratable acidity and aroma values are not presented because the most interesting aspect of our study was to determine bac-erial survival during the storage and the influence of fermentation temperature on total bacterial count during the storage. Figure 3 compares the effect of incubation temperature on viable cell count during the storage of whey fermented by a mixed culture L. hel-veticus-S. thermophilus. Figure 4 compares the effect of incubation temperature on viable cell count during the cool storage (4 °C) of whey fermented by a mixed culture L. delbrueckii ssp. lactis-S. thermophilus.

As shown in Figure 3, the viable cell count ≥ 6 log units has been held until the 23rd day of the storage in the sample fermented by L. helveticus-S. thermophilus at 42 °C. In samples fermented at 37 and 45 °C, the viable cell count was less than 6 log units already after 15 and 20 days of storage, respect-


6

ively. From the comparison of the viable cell count on the 23rd day of storage (Figure 3) in samples fer-mented at different temperatures, it can be concluded that the temperature of 42 °C had a positive impact on the stability of viable cell count during the storage period.

As shown in Figure 4, temperature 42 °C showed positive effects on the stability of the viable cell count of mixed culture L. delbrueckii ssp. lactis-S. thermo-philus during the storage period. The viable cell count ≥ 6 log units was held until about 20 days of storage.

In samples fermented at 37 and 45 °C, the viable cell count was less than 6 log units already after 13 and 15 days of storage, respectively.

As shown in Figures 3 and 4, after 20 days of storage viable cell count in sample fermented by L. helveticus-S. thermophilus was 6.5 log CFU/mL while in sample fermented by L. delbrueckii ssp. lactis-S. thermophilus was 5.36 log CFU/mL. Based on these results, it can be concluded that the mixed culture L. helveticus-S. thermophilus was more stable then

0 5 10 15 20 25 30 35 400

1

2

3

4

5

6

7

8

9

10

Via

ble

cell

cou

nt,

log(

CF

U/

mL

)

Storage time, day

T=37oC

T=42oC

T=45oC

Figure 3. Effect of incubation temperature on viable cell count during the cool storage (4 °C) of beverage fermented by mixed culture Lactobacillus helveticus ATCC 15009-Streptococcus thermophilus S3.

0 5 10 15 20 25 30 35 400

1

2

3

4

5

6

7

8

9

10

Via

ble

cell

coun

t, lo

g(C

FU

/ m

L)

Storage time, day

37oC

42oC

45oC

Figure 4. Effect of incubation temperature on viable cell count during the cool storage (4 °C) of beverage fermented by mixed culture Lactobacillus delbrueckii ssp. lactis NRRL B-4525-Streptococcus thermophilus S3.


7

mixed culture L. delbrueckii ssp. lactis-S. thermo-philus after the fermentation at 42 °C.

Regardless of the temperature, both mixed cul-tures had excellent aroma values at the end of the storage period (value 5). Titratable acidity increased in all samples for 20 days (about 0.5–1.0 °SH) of storage, regardless of temperature. After 20 days, the titratable acidity was stable to the end of the storage period. A significant pH decrease (around 0.5 pH units) was noticed in both samples fermented at 42 and 45 °C after 15 days of storage, while samples fermented at 37 °C showed pH decrease (around 0.4 pH units) after 5 days of storage. The pH value was stable until the end of the storage period.

CONCLUSIONS

Whey fermentation by mixed cultures L. hel-veticus-S. thermophilus and L. delbrueckii ssp. lactis- -S. thermophilus showed very similar characteristics. Symbiosis of testing cultures contributes to the increase of the aroma values, with negligible pro-longation of fermentation time, which can be avoided by increasing the fermentation temperature.

Based on the presented results, for both mixed cultures maximal viable cell count (about 8.5 log CFU/mL) was achieved at 45 °C, but during the storage period this count was decreased significantly faster than in samples fermented at 42 °C. Sub-sequently, the temperature of 42 °C is estimated as optimal for the fermentation of whey by both mixed cultures. Samples fermented at this temperature showed longer stability of viable cell count during the storage period than samples fermented at 45 °C. Viable cell count ≥ 6 log units in the sample fer-mented at 42 °C by L. helveticus ATCC 15009-S. thermophilus S3 has been held until 22nd day, while in the sample fermented by L. delbrueckii ssp. lactis-S. thermophilus until 19th day of storage. Subsequently, the mixed culture L. helveticus-S. thermophilus was more stable during the storage and had a longer shelf life.

Аcknowledgements

This work was funded by the Serbian Ministry of Education, Science and Technological development (TR 31017).

REFERENCES

[1] LJ. Tratnik, Mljekarstvo 53 (2003) 325-352

[2] N. Verma, V. Kumar, M. C. Bansal, J. Ind. Res. Tech. 1 (2011) 88-91

[3] C. González-Martínez, M. Becerra, M. Cháfer, A. Albros, J.M. Carot, A. Chiralt, Trends Food Sci. Technol. 13 (2002) 334-340

[4] C. Douaud, Whey proteins sees demand from functional drinks, http://www.nutraingredients-usa.com (accessed 12 December 2007)

[5] R. Jeličić, R. Božanić, Lj. Tratnik, Mljekarstvo 58 (2008) 257-274

[6] K. Marshall, Altern. Med. Rev. 9 (2004) 136-156

[7] T. Kar, A.K. Misra, Rev. Microbiol. 30 (1999) 163-169

[8] G.V. Reddy, K.M. Shahani, M.R. Banerjee, J. Natl. Cancer Inst. 50 (1973) 815–817

[9] C.F. Fernandes, K.M. Shahani, M.A. Amer, FEMS Microb. Rev. 46 (1987) 343–356

[10] S.E. Gilliland, FEMS Microb. Rev. 87 (1990) 175–188

[11] M.G. O’Sullivan, G. Thornton, G.C. O’Sullivan, J.K. Col-lins, Trends Food Sci. Techn. 3 (1992) 309–314

[12] S.Y. Lin, J.W. Ayres, W. Winkler, W.E. Sandine, J. Dairy Res. 72 (1989) 2885–2889

[13] J. Vijayakumar, R. Aravindan, T. Viruthagiric, Chem. Biochem. Eng. Q. 22 (2008) 245-264

[14] T. Stefanova, Z. Urshev, Z. Dimitrov, N. Fatchikova, S. Minkova, Biotechnol. Biotec. Eq. 23 (2009) 1368-1371

[15] A. Jae-Eun, M. Sc. Thesis, Dept. of Food Science and Agricultural Chemistry, Macdonald College of McGill Uni-versity, Montreal, Quebec, 2001, p. 44

[16] E.J. Faber, M.J. van den Haak, J.P. Kamerling, J.F.G. Vliegenthart, Carbohydr. Res. 331 (2001) 173–182

[17] S. Salminen, S. Gorbach, K. Salminen, Food Technol. 45 (1991) 112

[18] L. Skudra, A. Blija, E. Sturmovica, E. Dukalska, A. Abol-tins, D. Karklina, Acta Biotechnol. 18 (1998) 277-288

[19] V.M. Marshall, A.Y. Tamime, Int. J. Dairy Technol. 50 (1997) 35-41

[20] L. Varga, Int. J Food Microbiol. 108 (2006) 272-275

[21] Lj. Vrbaški, S. Markov, Praktikum iz mikrobiologije, Prometej, Novi Sad, 1993, pp. 102-105 (in Serbian)

[22] I. Dragalić, Lj. Tratnik, R. Božanić, Lait 85 (2005) 171-179

[23] M.Y. Pack, W.E. Sandine, P. R. Elliker, E.A. Day, R.C. Lindsay, J. Dairy Sci. 47 (1964) 981-986

[24] M. Pescuma, E.M. Hébert, F. Mozzi, G. Font de Valdez, Food Microbiol. 25 (2008) 442-451

[25] E.M. Hebert, R.R. Raya, G.S. de Giori, Appl. Environ. Microb. 66 (2000) 5316-5321

[26] V. Legarová, L. Kouřimská, Mljekarstvo 60 (2010) 280- –287.


8

MAJA LJ. BULATOVIĆ

MARICA B. RAKIN

LJILJANA V. MOJOVIĆ

SVETLANA B. NIKOLIĆ

MAJA S. VUKAŠINOVIĆ SEKULIĆ

ALEKSANDRA P. ĐUKIĆ

VUKOVIĆ

Tehnološko-metalurški fakultet, Univerzitet u Beogradu, Karnegijeva 4,

11000 Beograd, Srbija

NAUČNI RAD

UNAPREĐENJE PERFORMANSI PROIZVODNJE FUNKCIONALNOG FERMENTISANOG NAPITKA NA BAZI SURUTKE

Postoji veliki broj sojeva iz roda Lactobaccillus koji su već poznati kao visoko produktivni u

procesu mlečno-kiselinske fermentacije. Primena ovih visoko produktivnih sojeva skraćuje

vreme trajanja fermentacije, smanjuje troškove proizvodnje napitaka na bazi surutke i valo-

rizuje surutku nastalu tokom procesa proizvodnje sira. Cilj ovog istraživanja je bio unapre-

đenje performansi procesa proizvodnje napitaka na bazi surutke primenom visoko produk-

tivnih sojeva iz roda Lactobacillus. Proučavane su pojedinačne ili mešane kulture koje

sadrže sojeve Lactobacillus helveticus ATCC 15009, Lactobacillus delbrueckii ssp. lactis

NRRL B-4525 i Streptococcus thermophilus S3. Polazna naučna hipoteza je bila da na

performanse procesa proizvodnje napitaka, posebno na aromu i ukupni broj ćelija, pozi-

tivno utiče kombinovanje sojeva i primenjena temperatura fermentacije. Takođe je ispitivan

uticaj temperature fermentacije na preživljavanje primenjenih sojeva tokom procesa skla-

dištenja napitaka. Na osnovu dobijenih rezultata, simbioza ispitivanih sojeva doprinosi

unapređenju arome. Napici dobijeni primenom mešanih kultura imaju vrlo prihvatljivu

aromu što je veoma važno za njihovo uključivanje u ishranu ljudi. Rezultati pokazuju da

temperatura ima veoma značajan uticaj na dinamiku procesa fermentacije kao i na pre-

življavanje primenjenih sojeva tokom procesa skladištenja. Napitak proizveden pomoću

mešane kulture sastavljene od sojeva Lactobacillus helveticus ATCC 15009 i Strepto-

coccus thermophilus S3, fermentacijom na temperaturi 42 °C ispoljava visoku stabilnost

tokom skladištenja sa rokom trajanja od 22 dana.

Ključne reči: surutka, funkcionalni napici, probiotici, Lactobacillus, fermentacija, stabilnost.





9

RAMCHANDRA PANDEY1

PRAVIN O. PATIL1

SANJAY B. BARI1 DINESH M. DHUMAL2

1Department of Quality Assurance, H.R. Patel Institute of

Pharmaceutical Education and Research, Shirpur, Dist: Dhule

(M.S.), India 2Department of Pharmaceutical

Chemistry, R.C. Patel Institute of Pharmaceutical Education and Research, Shirpur, Dist: Dhule

(M.S.), India

SCIENTIFIC PAPER

UDC

DOI 10.2298/CICEQ120114098P

SIMULTANEOUS ESTIMATION OF ETODOLAC AND THIOCOLCHICOSIDE IN BULK AND IN TABLET FORMULATION BY UV-SPECTROPHOTOMETRY

Abstract

Two simple, rapid and reproducible simultaneous equation and Q-analysis UV--spectrophotometric methods have been developed for simultaneous esti-mation of etodolac (ETO) and thiocolchicoside (THC) in combined tablet dosage form. The methods involved solving simultaneous equations andQ-value analysis based on measurement of absorbance at wavelengths, 223 (λmax of ETO), 259.4 (λmax of THC) and 236 nm (iso-absorptive point). Linearity was found in the concentration range of 1–6 µg/mL and 4−24 µg/mL for ETO and THC, respectively, with correlation coefficients 0.9998 and 0.9992. The amounts of drugs estimated by the proposed methods are in excellent agreement with the label claimed. Furthermore, the methods were applied for the determination of ETO and THC in spiked human urine. The degradation behavior of ETO and THC was investigated under acid hydrolysis, alkali hyd-rolysis, photo- and oxidative degradation. The subsequently generated samp-les were used for degradation studies using the developed method. THC was found to degrade extensively under alkali hydrolysis and unaffected by other stress conditions, while ETO was found to be stable in all stress conditions. The methods were validated according to ICH guidelines. The method, suit-able for routine quality control, has been successfully applied to the deter-mination of both drugs in commercial brands of tablets.

Keywords: etodolac, thiocolchicoside, simultaneous equation method, Q- absorbance ratio method, spiked human urine, sress degradation study.

Etodolac is a nonsteroidal anti-inflammatory drug (NSAID) and used as anti-inflammatory and analgesic. It is chemically 1,8-diethyl-1,3,4,9-tetrahyd-ropyrano[3,4-b]indole-1-acetic acid [1-2]. ETO inhibits cyclo-oxygenase enzyme and subsequently inhibits prostaglandin synthesis, and is hence used as an analgesic [3]. Thiocolchicoside is also an anti-inflam-matory analgesic with muscle relaxant action [4], it is chemically N-[(7S)-3-(beta-D-glucopyranosyloxy)-1,2- -dimethoxy-10-(methylsulfanyl)-9-oxo-5,6,7,9-tetra-hydrobenzo[a]heptalen-7-yl] acetamide [1-2]. THC has selective affinity for GABA receptors and activate

Correspondence: P.O. Patil, H.R. Patel Institute of Pharmaceu-tical Education and Research, Shirpur, Dist: Dhule (M.S.), 425405 India. E-mail: [email protected] Paper received: 14 January, 2012 Paper revised: 22 September, 2012 Paper accepted: 25 October, 2012

GABA inhibitory pathway thereby acting as a potent muscle relaxant [5].

Etodolac is official in Indian Pharmacopoeia [2], United State Pharmacopoeia [6] and British Pharma-copeia [7]. Literature survey reveals one LC-MS [8] method was found for estimation of ETO in biological fluids. Few RP-HPLC [9-10], UV-spectrophotometric [11-14] and HPTLC [15] method have been reported for estimation of ETO in combination with other drugs in bulk and in pharmaceutical dosage forms. Thiocol-chicoside is official in Indian Pharmacopoeia [2]. Several RP–HPLC [16-21], UV-spectrophotometric [22-24] and HPTLC [25-26] methods have been stu-died for determination of THC in bulk and in pharma-ceutical formulations. Literature survey revealed simultaneous estimation of ETO and THC using UV spectrophotometric [27] and RP-HPLC [28] methods.

To the best of our knowledge, UV-spectro-photometric methods have not been yet reported for

R. PANDEY et al.: SIMULTANEOUS ESTIMATION OF ETODOLAC… Chem. Ind. Chem. Eng. Q. 20 (1) 9−17 (2014)

10

simultaneous estimation of ETO and THC in com-bined dosage form. In the present work, a successful endeavor has been made to estimate both these drugs simultaneously in tablet dosage form by two simple UV-spectrophotometric methods (simultane-ous equation method and Q-absorbance ratio method) [29]. However, the reported methods have poor sensitivity and are applicable only for pharma-ceuticals and not for biological fluids. The methods were successfully applied to the determination ETO and THC in spiked human urine. These methods were validated according to the ICH guidelines [30-32]. The chemical structures of both drugs are as shown in Figures 1 and 2.

Figure 1. Chemical structure of etodolac.

Figure 2. Chemical structure of thiocolchicoside.

MATERIAL AND METHODS

Chemicals and Reagents

Thiocolchicoside and etodolac bulk drugs were obtained from Vital Lab. Pvt. ltd, Mumbai, and Inchem Lab. Pvt. Ltd, Hyderabad, India, respectively, as gift samples. Methanol (HPLC grade) was purchased from Merck (India) Ltd., Worli, and Mumbai, India. Tablet (proxym-MR) was purchased from Indian mar-ket, containing 200 mg of ETO and 4 mg of THC.

Drug free human urine was obtained from a healthy male aged about 24 years.

Instrumentation

A UV-visible spectrophotometer (Shimadzu-1700, UV Probe 2.21 software) with spectral band-width 1 nm was employed for all spectroscopic mea-surements, using a pair of 1.0 cm matched quartz cells.

Selection of common solvent

Methanol was selected as common solvent for studying spectral characteristics of drugs.

Preparation of stock standard solutions

Stock standard solutions of ETO and THC were separately prepared by dissolving 10 mg in 100 ml volumetric flask containing 50 mL methanol and the volume was made up to the mark with water to obtain final concentrations of 100 µg/mL for each sample.

Simultaneous equation method (Method-I)

From the stock solution of 100 µg/mL, working standard solutions of drugs were prepared by appro-priate dilution and scanned in the UV-region, i.e., 400−200 nm. Linearity was found in the concentration range of 1-6 µg/mL and 4−24 µg/mL for ETO and THC, respectively (Table 1). From the overlaid spec-tra (Figures 3 and 4) two wavelengths, 223 (λmax of ETO) and 259.4 nm (λmax of THC) were selected for the construction of simultaneous equation. Standard solutions were prepared at concentrations 1–6 µg/mL for ETO and 4–24 µg/mL for THC. The absorbances of these standard solutions were measured at 223 and 259.4 nm and calibration curves were plotted. Two simultaneous equations (in two variables C1 and C2) were formed using E (1%, 1 cm) values (Table 2).

=1 ETO THC1383.17 +509.67A C C (1)

+2 ETO THC=262.66 433.16A C C (2)

where CETO and CTHC are the concentrations in g/100 mL in sample solution; A1 and A2 are absorbance of

Table 1. Linearity studies; SEE - standard error of estimate; SD - standard deviation

Parameter THC ETO

Linearity, µg mL-1 4-24 1–6

Linearity equation Y = 0.0335X + 0.0342 Y = 0.1276X + 0.0393

Slope ± SD 0.0335±0.0001 0.1276±0.0003

Intercept ± SD 0.0342±0.0005 0.0393±0.0008

Correlation Coefficient ± SD

SEE

Chi-square

Residual SD

0.9992±0.0001

0.00790

0.00069

0.0037

0.9998±0.0003

0.00343

0.0001

0.00307


11

mixture at selected wavelength 223 and 259.4 nm, respectively.

By applying Cramer’s rule [26] to Eqs. (1) and (2), the concentrations CETO and CTHC can be obtained as follows:

( ) ( )= × − × −ETO 2 1433.16 509.67 / 591451C A A (3)

( ) ( )= × − × −THC 1 2262.66 1383.17 / 591451C A A (4)

Q-Absorbance ratio method (Method-II)

From the overlain spectrum of ETO and THC, two wavelengths were selected one at 259.4 nm, λmax of THC and other at 236 nm, which was the isoabsorptive point for both drugs. The E (1%, 1 cm) values for both the drugs at selected wavelengths are shown in Table 2.

The method employed Q-values, and the con-centrations of drugs in sample solutions were deter-mined using the following equations:

1THC

1

m y

x y

Q Q AC

Q Q ax

−= ×

− (5)

2ETO

1

m x

y x

Q Q AC

Q Q ay−= ×−

(6)

where A1 and A2 are the absorbances of mixture at 236 and 259.4 nm, Qm = A2/A1, Qy = ay2/ay1 and Qx = = ax2/ax1, ax1 (397.66), ax2 (262.66), ay1 (397.83) and ay2 (433.16) are absorptivities (1%, 1 cm) of ETO and THC at 236 and 259.4 nm.

Assay of tablet formulation by Method-I and Method-II

Twenty tablets containing ETO and THC were weighed and the mean weight was calculated. These tablets were crushed and accurately weighed tablet powder equivalent to 10 mg of ETO that contains 0.2 mg THC was transferred into a 100 mL volumetric flask containing 50 mL of methanol, then 9.8 mg of THC working standard was added and the volume was made up to the mark with water, filtered through 0.45 μm Whatmann filter paper. An appropriate

Figure 3. Overlain spectra of ETO and THC.


12

volume of solution was further diluted with water to obtain concentrations 4 μg/mL of ETO and 4 μg/mL of THC. Absorbance of sample solution was recorded at 223, 236 and 259.6 nm and the concentrations of two drugs in the sample were determined using Eqs. (3) and (4) (Method-I) and (5) and (6) (Method-II). The analysis procedure was repeated for six times with tablet formulation. The results of analysis are men-tioned in Table 3.

Percentage recovery studies

The accuracy of the proposed analytical method was determined by recovery experiments. The reco-very studies were carried out at three different con-centration levels. A known amount of drug was added to pre-analysed tablet formulation at 80, 100 and 120%, and percentage recoveries were calculated. The results of recovery studies were satisfactory and are presented in Table 4.

Figure 4. Overlain spectra of ETO and THC and their mixture.

Table 2. E (1%, 1 cm) value of ETO and THC at 223, 259.4 and 236 nm

Parameter (E 1%, 1cm) at 223 nm (E 1%, 1cm) at 259.4 nm (E 1%, 1cm) at 236 nm

ETO THC ETO THC ETO THC

Mean 1383.17 509.67 262.67 433.16 397.66 397.83

SD 1.83 1.51 1.84 1.94 1.58 1.60

RSD / % 0.13 0.29 0.69 0.44 0.39 0.40

Table 3. Results of analysis of tablets by the proposed methods and statistical comparison of the results with the reference method

Method Tablet content Label claim, mg Found, label claim±SD, % (n = 6)

Reference method [33] Proposed method

I ETO 200 99.60±0.0032 99.25±0.29, F Test: 0.0011, t Test: 0.179

THC 4 104.50±0.0018 98.45±0.55, F Test: 1.542, t Test: 0.002

II ETO 200 99.10±0.0021 101.71±0.92, F Test: 1.496, t Test: 0.038

THC 4 103.25±0.0031 99.00±0.56, F Test: 4.865, t Test: 0.0056


13

Precision

Precision of an analytical procedure expresses the closeness of agreement between a series of measurements obtained from multiple samplings of the same homogenous samples under the prescribed conditions. The precision of the method was verified by intra-day, inter-day and repeatability studies. Intra-day precision was determined by analyzing three different concentrations (4, 5 and 6 µg mL-1) of ETO and THC, for three times in the same day. Day-to-day variability was evaluated using the three concen-trations listed above and analyzed on three different days. Repeatability of the sample solution was mea-sured by taking absorbances of a homogenous sample of 4 µg mL-1 of ETO and 4 µg mL-1 of THC six times.

Sensitivity

Sensitivity of the methods for drugs was indi-vidually determined by calculating the LOD, LOQ and Sandell’s sensitivity (µg/cm2), which can be defined as the smallest weight of substance that can be detected in the column of solution of unit cross section.

Analysis of ETO and TCH in human urine

From stock solutions of ETO and THC, 0.4 mL of sample was spiked in 5 ml of human urine and centrifuged at 3000 rpm, 4 °C for 10 min, then the supernatant was collected and the volume was made up with methanol to 10 ml, followed by sonication for 5 minutes to obtain concentrations of 4 μg/mL of ETO and 4 μg/mL of THC. The absorbance was measured directly at 223, 236 and 259.6 nm and the concentra-tions of two drugs in the spiked urine sample [33-35] were determined using Eqs. (3) and (4) (Method-I) and (5) and (6) (Method-II).

Stress degradation studies

The ICH guidelines require stability testing of new drug substances and products that to reveal the inherent stability characteristics of the active sub-stance. The aim of the present study was to perform

the stress degradation studies on ETO and THC using the developed method.

Conduct of stress studies

Acid and base-induced degradation was attempted by adding 10 mg of ETO and THC in 10 mL each of 0.1 M HCl and 0.1 M NaOH solutions. The solutions were kept for 8 h at room temperature in the dark in order to exclude the possible degradative effect of light. The solutions were neutralized and diluted with methanol. The absorbances were mea-sured directly at 223, 236 and 259.6 nm. For oxidative degradation, 10 mg of ETO and THC was added to 10 mL of 10% (v/v) hydrogen peroxide solution. The mix-tures were kept for 8 h. The solutions were diluted with methanol and treated as described for acid and base-induced degradation. Photodegradation was studied by exposing 1 mg/mL solution in methanol to sunlight for 72 h. The resulting solutions were diluted with methanol and analyzed using the methods des-cribed above.


Etodolac and thiocolchicoside followed linearity in the concentration ranges of 1–6 μg/mL and 4–24 μg/mL, respectively, and results are shown in Table 1 and Figures 5 and 6. The marketed brand of tablet was analyzed and recovery for ETO and THC deter-mined by proposed methods I and II was found to be 99.25 and 101.71% for ETO and 98.45 and 99.00% for THC, respectively. The recoveries range from 99.54 to 101.12% for ETO and THC in Method-I and -II, respectively. The results of the proposed method were statistically compared with those obtained by the reference method [27]. Statistical analysis of the results, using Student’s t-test and F-test revealed no significant difference between the performance of the proposed and reference method at 95% confidence level (Table 3). The advantage of the present method over the reference method [27] is that the author had used 0.1 M NaOH which may be responsible for the

Table 4. Results of recovery studies

Level of recovery, % Amount of drug added, μg/mL Drug Recovery, %

Method-I Method-II

80 3.2 ETO 100.55 100.42

3.2 THC 99.86 99.72

100 4 ETO 99.87 101.12

4 THC 99.75 99.87

120 4.8 ETO 99.77 100.68

4.8 THC 99.54 100.34


14

degradation of thiocolchicoside. Linearity study was carried out with good precise range in accordance with the dose ration of both drugs in used formulation, which were not reported in the reference method. The proposed methods are more sensitive because of low values of LOD and LOQ and additionally Sandell’s sensitivity, as compared to the reference method. Percentage label claim of both drugs by the proposed methods were found to be equivalent and more pre-cise over previously published spectrophotometric methods. The proposed methods can be applied to the determination of ETO and THC in spiked human urine. Precision was calculated as inter and intraday variations (RSD is less than 2%) for both drugs and as repeatability (RSD is less than 2%) and are pre-sented in Tables 5 and 6. LOD, LOQ and Sandell’s Sensitivity of ETO and TCH were found to be suffi-ciently low (Table 7), showing that much lower amounts of both drugs can be effectively detected by these methods. Ruggedness for these drugs was carried

out using two different laboratories and different anal-ysts; no significant difference was obtained between the results in the present study. The specificity and selectivity of the method was investigated by observ-ing any interference encountered from the common tablet excipients such as talc, lactose, starch and magnesium stearate. These excipients did not inter-fere with the proposed method.

Additionally, the developed methods were applied to determine ETO and THC in spiked urine sample. The recoveries of ETO and THC from spiked human urine sample were found to be satisfactory. As shown in Table 8, the percentage recovery values in the range 90.26 to 94.72 for Method-I as well as Method- -II with relative standard deviation values of less than 2 proved that the accuracy and reproducibility of the proposed method for the determination of both the drugs simultaneously in spiked human urine. The UV spectrum obtained from action of acid, alkali, hyd-rogen peroxide, and light showed that THC was

Figure 5. Calibration curve of THC.

Figure 6. Calibration curve of ETO.


15

Table 5. Results of precision studies

Precision (RSD / %) Method-I Method-II

ETO THC ETO THC

Intra–day (n = 3) 0.14 0.15 0.37 0.25

Inter–day (n = 3) 0.25 0.14 0.14 0.29

Table 6. Results of repeatability study

Method Drug (4 µg/mL) Amount found in µg/mL (n = 6) RSD / %

I ETO 3.98 0.25

THC 3.96 0.15

II ETO 3.99 0.14

THC 3.97 0.21

Table 7. Sensitivity

Parameter ETO TCH

LOD / µg mL–1 0.088 0.129

LOQ / µg mL–1 0.291 0.949

Sandell’s sensitivity, µg/cm2 0.00723 0.0230

Table 8. Analysis in human urine; amount spiked in urine: 4 µg/mL

Method Drug Amount found, % (n = 5) RSD / %

I ETO 92.88 2.86

THC 94.72 1.98

II ETO 90.26 2.04

THC 93.24 2.42

Figure 7. A) ETO in methanol, 0.1 M HCl and 0.1 M NaOH; B) THC in methanol, 0.1 M HCl and 0.1 M NaOH; C) ETO and THC mix in methanol, 0.1 M HCl and 0.1 M NaOH.


16

Table 9. Stress degradation studies of ETO and THC

Sample-exposure condition λmax / nm Recovery, %

ETO THC ETO THC

0.1 M HCl, 8 h 223 259.6 98.52 99.08

0.1 M NaOH, 8 h 223 241.2a 98.69 42.84

10% H2O2, 8 h 223 259.6 100.21 99.58

Sunlight, 8 h 223 259.6 100.54 100.20 aChange in λmax of THC indicates that THC is susceptible to alkali condition and it does some transformation in THC structure

degraded only by alkali hydrolysis (Figure 7) because of change in λmax of THC and there was no effect of other stress conditions on THC as well as ETO. The amount remaining (%) and recovery (%) were cal-culated for ETO and THC in stability samples by using Eqs. (3) and (4) (Method-I) and (5) and (6) (Method-II), Table 9.

CONCLUSION

The developed methods were found to be simple, accurate and rapid for the routine estimation of ETO and THC in tablet formulation as well as in human urine. Degradation products resulting from stress studies did not interfere with the simultaneous estimation of ETO and THC except alkali conditions. The methods were validated according to ICH guide-lines. The method, suitable for routine quality control, has been successfully applied to the determination of both drugs in commercial brands of tablets.

Acknowledgements

The authors are thankful to H.R. Patel Institute of Pharmaceutical Education and Research, Shirpur (M.S.), India, for providing facilities to carry out this research work.

REFERENCES

[1] The Merck Index – An Encyclopedia of Chemicals, Drugs and Biologicals, 14th ed., Merck Research Laboratories, Whitehouse Station, NJ, 2006, pp. 685, 1662

[2] Indian Pharmacopoeia, Govt. of India, Ministry of Health and Family Welfare, published by Indian Pharmacopoeial convention, Ghaziabad, Vol. II, 2010, pp. 1318-1322, 2213-2215

[3] T. Ishizaki, K. Katsumata, A. Tsuchida, T. Wada, Y. Mori, M. Hisada, H. Kawakita , T. Aoki, Int. J. Mol. Med.17(2) (2006) 357-362

[4] D. F.Soonawalla, N. Joshi, J. Indian Med. Assoc. 106(5) (2008) 331–335

[5] A.R. Umarkar, N.S. Rewatkar, M.S. Charde, R.M. Charde, A.V. Kasture, J. Pharm. Res. 4(5) (2011) 1307- -1308

[6] United States Pharmacopoeia-32, National Formulary-27, The United States Pharmacpoeial Convention, Rockville, Vol. II, 2009, pp. 2334-2336

[7] British Pharmacopoeia, The Stationary Office Medicinal and Pharmaceutical Substances (A–I), Vol. III, London, 2009, pp. 2567-2569

[8] H.S. Lee, I.M. Kang, H.W. Lee, J. H. Ryu, M. J. Lee, S.Y. Jeong, K.T. Lee, J. Chromatogr., B 863(1) (2008) 158- -162

[9] V.G. Dongre, S.B. Shah, G.S. Bayes, M. Phadke, V.K. Jadhav, Chromatographia 69(9) (2010) 1019-1023

[10] O.A. Saleh, A.A. Eazzouny, A.M. Badawey, M.S. Rashed, J. Liq. Chromatogr. Relat. Technol. 32(17) (2009) 2584- -2599

[11] A.A. Gouda, W.S. Hassan, Chem. Cent. J. 2(7) (2008) 1-8

[12] N.M. El Kousy, J. Pharm. Biomed. Anal. 20(2) (1999) 185-194

[13] S.T. Ulu, J. Food Drug Anal. 19(1) (2011) 94-101

[14] R. Patidar, U.S. Baghel, S. Paetla, M. Singhal, N. Patidara, G. Englaa, N. Dongrea, J. Glo, Pharma Tech. 1(1) (2009) 62-66

[15] P.H Chaube, S.V Gandhi, P.B. Deshpande, V.G. Kulkarni, J. Pharm. Biomed. Sci. 7(13) (2010) 1-6

[16] S.S. Chitlange, P.S. Shinde, G.R. Pawbake, S.B. Wankhede, Der Pharmcia Lettre 2(2) (2010) 86-93

[17] S.B. Wankhede, S.S. Zambare, N.R. Dixit, S.S. Chitlange, Der Pharmcia Lettre 2(3) (2010) 315-320

[18] M. Sahoo, P. Syal, S. Ingale, K. Ingale, S. Sindhe, V.P. Choudhari, B.S. Kuchekar, Int. J. Res. Pharm. Sci. 2(1) (2011) 1-7

[19] Y.J. Chen, S.M. Huangd, C.Y. Liu, P.H. Yeh, T.H. Tsai, Int. J. Pharm. 350 (2008) 230–239

[20] S.R. Dhaneshwar, K.O. Raut, V.K. Bhusari, Res. J. Pharm. Biol. Chem. Sci. 2(2) (2011) 435-445

[21] N. Goyal, A. Bhandari, S. Jain, R. Patel, Int. J. Pharm. Stud. Res. 2(1) (2011) 106-109

[22] S.K. Acharjya, P. Mallick, P. Panda, M.M. Annapurna, J. Pharm. Educ. Res. 1(1) (2010) 51-57

[23] V.P. Choudhari, A.R. Chabukswar, S.N. Savakhande, M.U. Tryambake, V.M. Suryawanshi, P.K. Sayal, Int. J. Curr. Res. Rev. 2(12) (2010) 3-10

[24] S.A. Hapse, R.R Thorave, P.T. Kadaskar, A.S. Shirsath, M.D. Dokhe, J. Pharm. Res. 4(11) (2011) 3928-3929

[25] V.S. Rajmane, S.V. Gandhi, U.P. Patil, M.R. Sengar, J. Pharm. Res. 4(5) (2011) 201-212


17

[26] S.T. Patil, V.K. Bhusari, S.R. Dhaneshwar, Int. J. Pharma Bio Sci. 2(2) (2011) 482-490

[27] R. Tiwari, S. Pillai, Res. J. Pharm. Technol. 4(12) (2011) 1891-1895

[28] R. Pandey, P.O. Patil, S.B. Bari, Asian J. Bio. Pharm. Res. 1(2) (2012) 381-390

[29] A.H. Beckett, J.B. Stenlake, Practical Pharmaceutical Chemistry, 4th ed., Part II, CBS Publishers and Distri-butors, New Delhi, 1997, pp. 275-277

[30] ICH, Q2A: Text on Validation of Analytical Procedures, International Conference on Harmonization, October, 1994

[31] ICH, Q3B: Validation of Analytical Procedures: Metho-dology, International Conference on Harmonization, Nov-ember, 1996

[32] ICH Harmonized Tripartite Guideline, Q2 (R1): Validation of Analytical Procedures: Text and Methodology, November, 2005

[33] N. Rajendraprasad, K. Basavaiah, K.B. Vinay, Chem. Ind. Chem. Eng. Q. 17(3) (2011) 259-267

[34] J. Parojcic, K. Karljikovic Rajic, Z. Duric, M. Jovanovic, S. Ibric, Biopharm. Drug Dispos. 24 (2003) 309–314

[35] B.N. Barsoum, M.S. Kamel, M.M.A. Diab, Res. J. Agric. Biol. Sci. 4(5) (2008) 471-484.

RAMCHANDRA PANDEY1

PRAVIN O. PATIL1

SANJAY B. BARI1

DINESH M. DHUMAL2

1Department of Quality Assurance, H.R. Patel Institute of Pharmaceutical

Education and Research, Shirpur, Dist: Dhule (M.S.), India

2Department of Pharmaceutical Chemistry, R.C. Patel Institute of

Pharmaceutical Education and Research, Shirpur, Dist: Dhule (M.S.),

India

NAUČNI RAD

SIMULTANO ODREĐIVANJE ETODOLAKA I TIOKOLČIKOZIDA U FARMACEUTSKOJ SUPSTANCI I TABLETAMA

U radu su razvijene dve jednostavne, brze i reproduktivne spektrofotometrijske metode za

simultano određivanje etodolaka (ETO) i tiokolčikozida (THC) u tabletama koje sadrže obe

aktivne supstance. Metode uključuju rešavanje sistema jednačina uz korišćenje rezultata

merenja apsorbanci na 223 nm (λmax za ETO), 259,4 (λmax za THC) i 236 nm (izo-apsorp-

tivna tačka). Nađeno je da pod optimalnim uslovima za reakcije važi Berov zakon u

opsegu koncentracija 1-6 µg/mL, odnosno 4-24 µg/mL za THC, odnosno ETO, a odgova-

rajući koeficijenti korelacije su 0,9998 i 0,9992. Količine određenih aktivnih supstanci su u

odličnom slaganju sa količinama koje su označene na etiketi leka. Takođe, metode su

uspešno primenjene za određivanje ETO i THC u ljudskom urinu. Ispitana je degradacija

ETO i THC u uslovima kisele i bazne hidrolize i foto i oksidacione degradacije. Nastali

degrdacioni produkti su analizirani razvijenom metodom. Nađeno je da se tiokolčikozid

intenzivno razgrađuje u uslovima bazne hidrolize, a da je otporan na druge ispitivane

uslove, dok je etodolak stabilan na sve ispitivane uslove. Metode su validirane u skladu sa

ICH smernicama. Metoda je uspešno primenjena za određivanje obe aktivne supstance u

tabletama i pogodna je za rutinsku kontrolu.

Ključne reči: etodolak, tiokolčikozid, simultana metoda, Q-apsorpciona metoda, ljudski urin, forsirane degradacije leka.





19

B. JAYALAKSHMI1 K.A. RAVEESHA2

K.N. AMRUTHESH3 1Department of Botany, Maharani’s Science College for Women, J.L.B.

Road, Mysore, Karnataka, India 2Herbal Drug Technology

Laboratory, Department of Studies in Botany, University of Mysore,

Manasagangotri Mysore, Karnataka, India

3Applied Plant Pathology Laboratory, Department of Studies

in Botany, University of Mysore, Manasagangotri Mysore,

Karnataka, India

SCIENTIFIC PAPER

UDC 615.281:582.682.1:66:54

DOI 10.2298/CICEQ120610099J

EVALUATION OF ANTIBACTERIAL AND ANTIOXIDANT POTENTIAL OF Euphorbia cotinifolia LINN. LEAF EXTRACTS

Abstract

Antibacterial activity of aqueous and solvent extracts of Euphorbia cotinifolia leaves was tested against some human pathogenic bacteria, viz. Escherichia coli, Klebsiella pneumoniae, Bacillus subtilis, Bacillus cereus, Salmonella typhi, Enterobacter aerogenes and Staphylococcus aureus by agar cup diffu-sion and broth microdilution methods. Antioxidant properties were evaluated for different solvent extracts by di(phenyl)-(2,4,6-trinitrophenyl)iminoazanium (DPPH), nitric oxide (NO) and hydrogen peroxide methods and IC50 values were calculated and compared with the standard ascorbic acid and butylated hydroxyanisole. Among the aqueous and organic solvent extracts, methanol and ethyl acetate showed significant activity against B. subtilis and E. aero-genes, which recorded a maximum inhibition zone of 17.25 mm. Minimum inhibitory concentrations of methanol and ethyl acetate extracts for different bacteria ranged from 0.3–1.25 mg/mL. In DPPH method, IC50 values of chloro-form, petroleum ether, ethyl acetate and methanol were found to be 15, 17, 18 and 19 mg/mL, respectively, lesser than the standard, ascorbic acid (25 mg/mL). Phytochemical analysis of aqueous, ethyl acetate and methanol extract showed the presence of flavonoids, terpenoids, tannins and steroids. Further work is in progress to isolate the active compound(s).

Keywords: Euphorbia cotinifolia, antibacterial activity, antioxidant activity, flavonoids, steroids..

Infectious diseases are major causes of mor-bidity and mortality in the developing world and account for about 50% of all deaths. Besides inci-dents of epidemics, death due to drug resistant micro-organisms poses enormous public health concerns [1]. Conventional antibiotics are useful medicines that are highly potent and save lives, but they cause more harmful effects than useful ones [2]. Bacterial resis-tance to antibiotics increases mortality, likelihood of hospitalization and length of stay in the hospital [3]. In developing countries, due to poor sanitation, igno-rance and bad hygienic practices, a large number of people are exposed to infectious agents. Pathogenic bacteria like Escherichia coli, Salmonella typhi, Pro-

Correspondence: K.N. Amruthesh, Applied Plant Pathology Laboratory, Department of Studies in Botany, University of Mysore, Manasagangotri Mysore-570 006, Karnataka, India. E-mail: [email protected];

[email protected] Paper received: 10 June, 2012 Paper revised: 25 October, 2012 Paper accepted: 25 October, 2012

teus spp., Shigella spp., Pseudomonas aeruginosa and Staphylococcus aureus, which are usually pre-sent as commensals, have several virulent factors and colonize in a biofilm fashion causing a variety of intestinal and extra intestinal diseases [4]. Thus, there is a need to search for some newer, safer, effective and above all inexpensive antimicrobial agents to tackle these problems.

Higher plants have the capacity to produce a large number of phytochemicals with complex struc-tural diversity known as secondary metabolites, which are produced for self-defense [5]. Over the last 30 years, a large number of secondary metabolites from various plant species have been evaluated for anti-microbial activity. Since these secondary metabolites from natural resources have been elaborated within living systems, they are often perceived as showing more drug likeness and biological friendliness than totally synthetic molecules, making them good can-didates for further drug development. The success story of modern medicine lies in the continuous

B. JAYALAKSHMI et al.: EVALUATION OF ANTIBACTERIAL… Chem. Ind. Chem. Eng. Q. 20 (1) 19−28 (2014)

20

search for new drugs to counter the challenges posed by resistant strains of bacteria. There are several reports in the scientific literature describing the anti-microbial properties of crude extracts prepared from plants [6-8].

Lipid oxidations are one of the important pro-cesses that produce free radicals in food, chemicals and also in living systems. These are involved in food spoilage and chemical degradation, which contribute to several diseases in humans [9]. Antioxidants play an important role in scavenging free radicals and/or chain breaking of the oxidation reactions. The inhi-bition of oxidative reactions in food, pharmaceutical and prevention of oxidative stress related diseases in humans are some of the potential functions of anti-oxidants [10]. Chemically synthesized compounds such as butylated hydroxyanisole (BHA) and butyl-ated hydroxytoluene (BHT) are used as antioxidants, but the use of BHA and BHT has proven to be car-cinogenic [11]. Hence, naturally produced antioxi-dants are preferred as they are capable of inhibiting free radical reactions, retarding oxidative rancidity of lipids, protecting the human body from diseases and preserving food from spoilage [12].

Euphorbia cotinifolia L. is a plant belonging to the family Euphorbiaceae. The family comprises of 300 genera and 5000 species distributed in tropical countries [13]. Several Euphorbia species have been subjected to various biological activities by several researchers [14-18]. E. cotinifolia is a perennial shrub with woody stem. It has delicate oval shaped leaves, with milky sap and small insignificant flowers. Older leaves are dark burgundy, while new foliage has a brighter red tone. Loose flower clusters have small white to pale yellow bracts during spring season [19].

Antibacterial activity of E. cotinifolia has been evaluated by Rojas and co-workers [20]. In their report, dried leaves of E. cotinifolia were extracted with isopropyl alcohol and fractionated by column chromatography. None of the fractions eluted from the column separation of E. cotinifolia extract showed antibacterial activity against S. aureus, Enterococcus faecalis, E. coli, Klebsiella pneumoniae and P. aeruginosa. The present investigation reports soxhlet based extraction successively from low polar to high polar solvents. These extracts have been subjected to antibacterial activity against E.coli, K. pneumoniae, B. subtilis, B. cereus, S. typhi, Enterobacter aerogenes and S. aureus. Antioxidant activity of the extracts has also been evaluated.


Plant material

Healthy leaves of Euphorbia cotinifolia were col-lected in 2010 from Mysore, Karnataka, India, and used for the preparation of aqueous and different solvent extracts. A voucher specimen of the plant has been deposited in the Herbarium, Department of Studies in Botany, University of Mysore, Mysore, Karnataka State, India (voucher/specimen number is MGBH01). A group of Taxonomists in the Department of Studies in Botany, University of Mysore, India, have identified the plant.

Test pathogens

Authentic cultures of human pathogenic bacteria viz., Escherichia coli (MTCC 7410), Klebsiella pneumoniae (MTCC 7407), Bacillus subtilis (MTCC 121), Bacillus cereus (MTCC 1272), Salmonella typhi (MTCC 733), Enterobacter aerogenes (MTCC 7325) and Staphylococcus aureus (MTCC 7443) which served as test bacteria were obtained from Microbial Type Culture Collection, Chandigarh, India.

Preparation of extract

Aqueous extract

Samples (50 g) of thoroughly washed fresh leaves of E. cotinifolia were macerated with 50 mL sterile distilled water in a blender (Warning Inter-national, New Hartford, CT, USA) for 10 min. The macerate was first filtered through double layered cloth and then centrifuged at 4000 rpm for 30 min. The supernatant was filtered through Whatman No. 1 filter paper and sterilized at 120 °C for 10 min. The extract was cooled to room temperature and pH was recorded just before subjecting it (1:1) to antibacterial activity assay.

Solvent extract

Thoroughly washed mature leaves were shade dried and then powdered in a blender. 50 g of the powder was filled in the thimble and extracted suc-cessively with petroleum ether, chloroform, ethyl ace-tate and methanol using a soxhlet extractor for 48 h. All the extracts were concentrated using rotary flash evaporator and preserved at 5 °C in an airtight bottle until further use. All the extracts were tested for their antibacterial potential.

Phytochemical analysis

Phytochemical analysis of petroleum ether, chloroform, ethyl acetate, methanol and water extract was carried out for the detection of active secondary metabolite or different constituents such as tannins,


21

alkaloids, flavonoids, terpenoids, steroids, carbohyd-rates, proteins and saponins. The dried extracts were reconstituted in methanol and 1 mL of each extract was subjected to standard phytochemical analysis according to the procedure described [21]. The phyto-chemical tests are given briefly in the following para-graphs.

Test for alkaloids

Mayers test. The test extract was treated with Mayers reagent (potassium mercuric iodide solution). Formation of cream precipitate indicates the presence of alkaloids.

Dragendroff’s test. The extract was treated with Dragendroff reagent (potassium bismuth iodide solu-tion); formation of orange brown precipitate indicates the presence of alkaloids

Test for phenols

Ferric chloride test. The extract was treated with ferric chloride; formation of green precipitate indicates the presence of phenols.

Test for flavonoids

Shinoda test. The extract was warmed with magnesium turnings with three drops of conc. hydro-chloric acid, an orange pink coloration indicate the presence of flavonoids.

Test for terpenoids

Dinitrophenyl hydrazine test. The extract was treated with 0.5% (2,4-dinitrophenyl)hydrazine solu-tion to give yellow-orange coloration indicating the presence of terpenoids.

Test for tannins

Ferric chloride test. The extract treated with ferric chloride, formation of bluish black colored pre-cipitate indicates the presence of tannins.

Gelatin test. The extract was treated with 1% gelatin in sodium chloride solution; formation of white precipitate indicates the presence of tannin.

Test for steroids

Liebermann-Burchard test. The extract was treated with 0.5 mL each of the acetic anhydride and chloroform and cooled well in ice. A volume of 1 mL conc. sulfuric acid was added to the sides of the test tube carefully, which produces reddish violet color-ation at the junction indicates the presence of ste-roids.

Salkowski test. The extract was treated with chloroform, shaken well and kept aside for three minutes. Few drops of conc. sulfuric acid were added carefully through the sides of the test tube. Red color at the junction indicates the presence of steroids.

Test for glycosides

Legal’s test. the extract was treated with sodium nitroprusside in pyridine and methanolic alkali. The formation of pink color indicates the presence of glycosides.

Test for carbohydrates

Molisch’s test. The extract was treated with Molisch reagent (α-naphthol in 95% ethanol) and few drops of conc. sulfuric acid were added at the sides of the test tube; violet ring at the junction indicates the presence of carbohydrates.

Fehling’s test. The extract was treated with Feh-ling’s reagent and the mixture was heated to give the brick-red color; precipitate indicates the presence of carbohydrates.

Test for proteins

Millons test. Extract was treated with Millons reagent, which yields red coloration.

Biuret test. Extract was treated with sodium hyd-roxide and copper sulfate solution added drop wise, which yields violet coloration.

Test for saponins

Froth test. The extract was dissolved in 5 mL of distilled water in a test tube and shaken vigorously and then allowed to stand for 15 min. The appearance of frothing persists indicates the presence of sapo-nins.

Antibacterial activity assay

Antibacterial activities of aqueous and solvent extracts were determined by agar cup diffusion method [22]. A 7 mm cork borer was used for making wells on nutrient agar medium and inocula containing 106 CFU/mL of bacteria were spread on the solid media with a sterile swab moistened with the bacterial suspension. The dried solvent extracts were recon-stituted in methanol to a concentration of 100 mg/mL.

Aqueous and solvent extracts of 100 µL were placed in the well made in the inoculated plates. Sterilized distilled water and methanol of same quan-tity were placed in the cups separately, which served as the control. Streptomycin (streptomycin sulfate IP; 1 mg/mL) served as the standard drug. The plates were incubated for 24 h at 37 °C and zones of inhi-bition were measured. For each treatment 3 repli-cates were maintained and all assays were repeated.

Determination of minimum inhibitory concentration (MIC)

MIC was determined in 96 well sterile flat bottom micro-titer plates based on a micro dilution assay, which is an automated turbidometric and colorimetric method as described by [23,24]. Inoculum of the test


22

bacteria was prepared from 24 h culture of the bac-teria and a suspension were made in sterile/saline water and adjusted to 0.5 McFarland standard solu-tion turbidity.

The crude extracts of methanol and ethyl ace-tate were diluted to the concentration of 100 mg/mL, which served as the stock solution. The 96 well plates were prepared by dispensing 200 µL of broth and 100 µL of the extract to the first well. A two-fold serial dilution was made and the final concentrations were 5–0.019 mg/mL. The inoculum suspension of 10 µL of each bacterial strain was added to each well.

The wells containing nutrient broth with ino-culum and solvent served as negative control. The plates were incubated at 37 °C for 24 h and turbidity was measured at 620 nm using a microplate reader (LT4000, LABTECH Instruments, UK). The lowest concentration that inhibited visible growth was recorded as the MIC based on the readings.

The MIC was also detected by adding 10 µL/well of TTC (2,3,5-triphenyl-2H-tetrazolium chloride, Sigma) dissolved in water (TTC, 2 mg/mL) and incubated under appropriate cultivation conditions for 30 min in the dark [25]. Viable organism reduced the dye to pink color compound. The lowest concentration at which the color change occurred was taken as the MIC value. All MIC tests were repeated in triplicate.

Statistical analysis

Statistical calculations were carried out using one way analysis of variance (ANOVA) and the sig-nificances of the differences between means were calculated using Tukey’s multiple range test at a sig-nificance level of P < 0.05.

Antioxidant assay

Different radical scavenging methods have been frequently used to estimate the antioxidant capacity of several plants extracts [26]. The present investigation reports the antioxidant assay of E. cotinifolia by di(phenyl)-(2,4,6-trinitrophenyl)iminoazanium (DPPH), nitric oxide (NO) and hydrogen peroxide methods.

DPPH Radical scavenging assay

The free radical-scavenging assay of different extracts was measured in terms of hydrogen donating or radical-scavenging ability using the stable radical DPPH described by Blois method [27]. Stock solu-tions of the extracts (0.001 g mL-1) were prepared by dissolving in dimethyl sulfoxide (DMSO). Different concentrations (20, 40, 60, 80 and 100 μg) of test solutions were prepared from stock and made up to 2 mL with methanol. Solution of DPPH (0.1 mmoL) in methanol was prepared and 1 ml of this solution was

added to each of the above test solutions. The mix-ture was shaken vigorously and incubated for 30 min, and the absorbance of each test solution was mea-sured at 517 nm. All the tests were run by triplicate and expressed as the mean ± standard deviation (SD). Ascorbic acid (AA) was used as a standard or positive control and DMSO was used as a negative control. The capability to scavenge the DPPH radical was calculated using the following equation:

DPPH• scavenging effect (%) = − ×c b

c

100A A

A (1)

where Ac is the absorbance of the negative control, i.e., without sample, and Ab is the absorbance with the extract.

Hydroxyl radical scavenging assay

The ability of the compounds to effectively sca-venge hydrogen peroxide was determined according to the reported method [28], where it was compared with that of butylated hydroxyanisole (BHA) as the standard. The hydroxyl radicals (OH•) in aqueous media was generated through the Fenton system. The test solutions of the different extracts were pre-pared with DMSO (0.001g mL-1). A volume of 5 ml assay mixture contained the following reagents: saf-ranin (11.4 µmoL), EDTA–Fe(II) (40 µmoL), H2O2 (1.76 µmoL), the test solution (4, 8, 12, 16 and 20 µL (5 mg/mL)) and a phosphate buffer (0.067 moL, pH 7.4). The assay mixtures were incubated at 37 °C for 30 min in a water bath and the absorbance was mea-sured at 520 nm. BHA which suppressed hydroxyl radical was used as positive control. All the tests were assayed in triplicate and expressed as the mean ± standard deviation (SD). The suppression ratio for OH• was calculated by the following equation:

Suppression ratio (%) = − ×i

0

0 100A A

A (2)

where A0 is the the absorbance of the control and Ai is the absorbance of the test solution.

Nitric oxide scavenging assay

Nitric oxide radical scavenging assay was per-formed according to [29]. The assay was based on generation of nitric oxide (NO) from sodium nitro-prusside (SNP) and it was measured by the Griess reagent. At physiological pH, aqueous solution of sodium nitroprusside spontaneously generates nitric oxide, which interacts with oxygen to produce nitrite ions. The generated nitrite ion was quantified by the Griess reagent. A volume of 1.5 mL (10 mM) sodium nitroprusside in phosphate buffer saline with pH 7.4 was mixed with various concentrations (20, 40, 60, 80


23

and 100 μg) of 1 mL extract and the mixture was incubated at 25 °C for 150 min. After incubation, the formed nitric oxide reacts with 1.5 mL Griess reagent (1% sulfanilamide, 2% phosphoric acid and 0.1% naphthylethylenediamine dihydrochloride). The reac-tion mixture was allowed 30 min at room temperature. The test solution (4 mL) was made up with the phos-phate buffer (pH 7.4). The absorbance of this solution was measured at 546 nm against the appropriate blank. All the samples and controls were performed in triplicate. IC50 values were determined using the fol-lowing equation:

Nitric oxide· scavenging effect (%) = − ×c

c

t 100A A

A (3)

where Ac is the absorbance of control and At is the absorbance in presence of the sample mixture with extract.

RESULTS

Phytochemical analysis

The results of phytochemical analysis of E. coti-nifolia are given in Table 1. The flavonoids, terpe-noids, tannins and steroids were present in aqueous, methanol and ethyl acetate extracts, whereas alka-loids, carbohydrates, saponins and proteins were absent in aqueous, methanol and ethyl acetate extracts. All the constituents namely alkaloids, flavor-noids, terpenoids, tannins, proteins and saponins were absent in petroleum ether and chloroform extracts. Steroids and carbohydrates were present in petro-leum ether and chloroform extracts.

Antibacterial activity assay

The results of antibacterial activity of aqueous and different solvent extracts of E. cotinifolia against the test bacteria are presented in Table 2. Aqueous extract did not show activity against any of the test bacteria. Among the solvent extracts, ethyl acetate

and methanol showed significant activity, while neg-ligible activity was found with chloroform extract. The inhibition range was recorded between 12–17 and 10– –14 mm for methanol and ethyl acetate extracts, res-pectively. K. pneumoniae, B. subtilis and E. aero-genes were highly susceptible to methanol extract with the maximum inhibition zone of 17 mm. B. cereus was the less susceptible to methanol extract. Ethyl acetate extract showed a uniform inhibition zone in the range of 10–14 mm against all the tested bac-teria. Petroleum ether extract showed minimum acti-vity against B. cereus and S. typhi while it has neg-ligible activity against other bacteria. The inhibition zone of crude methanol extract was slightly lesser than that of streptomycin, a standard antibiotic. The MIC for different test bacteria ranged from 0.312–1.25 mg/mL for methanol and, ethyl acetate extracts. The minimum MIC concentration was 0.312 mg/mL recorded for B. subtilis for both extracts (Table 3).

Radical scavenging activity

The free radical scavenging ability of tested extracts was evaluated by the change in absorbance by reduced DPPH. The ability of extracts to neutralize hydroxyl radical was expressed as 50% inhibitory concentration, IC50, in μg/mL. The results of DPPH test showed that the chloroform extract of E. cotini-folia was the most active with an IC50 value of 15 μg/mL, followed by petroleum ether, ethyl acetate and methanol with IC50 values of 17, 18, and 19 μg/mL, respectively (Figure 1). These extracts showed higher radical scavenging activity compared to standard ascorbic acid with IC50 value of 25 μg/mL. All the extracts recorded H-donor activity that was concen-tration dependent. The lower the IC50 values, the better was the scavenging ability of the extract.

Hydroxyl radical scavenging

The scavenging effect of the tested extracts on HO• is concentration related and the suppression ratio increases with the increasing concentration of sample

Table 1. Phytochemical analysis of aqueous and solvent extracts of Euphorbia cotinifolia (leaf extract); + = present; - = absent

Active principle Aqueous Petroleum ether Chloroform Ethyl acetate Methanol

Alkaloids – - - - -

Flavanoids + - - + +

Terpenoids + - - + +

Tannins + - - + +

Steroids + + + + +

Glycosides + + + + +

Carbohydrates - + + - -

Proteins - - - - -

Saponins - - - - -

B. JAYALAKSHMI, K.A. RAVEESHA, K.N. AMRUTHESH: EVALUATION OF ANTIBACTERIAL… CI&CEQ 20 (0) 000−000 (2014)

24

Table 2. Antibacterial activity of different extracts of E. cotinifolia against human pathogenic bacteria (in mm); values are means of four independent replicates. Figures followed by different letters in columns differ significantly when subjected to Tukey (P < 0.05); – means no activity

Bacteria Aqueous Petroleum ether Chloroform Ethyl acetate Methanol Streptomycin

Bacillus cereus 0.00 - 8.75±0.47b 11.25±0.47ab 12.25±0.50c 21.0±0.40bc

Bacillus subtilis 0.00 8.25±0.2a 8.50±0.28b 12.75±0.62a 17.25±0.62bc 20.75±0.47bc

Escherichia.coli 0.00 - 8.50±0.28b 10.75±0.25b 13.5±0.64a 23.75±0.47a

Enterobacter aerogenes 0.00 - 10.25±0.2a 13.0±0.57a 17.25±0.62a 21.25±0.62bc

Klebsiella pneumoniae 0.00 8.0±0.40a 8.75±0.25b 12.75±0.25a 17.0±0.40a 19.75±0.62c

Salmonella typhi 0.00 - 8.25±0.25b 11.5±0.28ab 16.0±0.40a 22.5±0.28ab

Staphylococcus aureus 0.00 - 0.00±0 10.75±0.25b 15.5±0.28ab 19.25±0.47c

Table 3. MIC of methanol and ethyl acetate extract (mg/mL) of E. cotinifolia against some human pathogenic bacteria

Bacteria Bacillus cereus

Bacillus subtilis

Escherichiacoli

Enterobacter aerogenes

Klebsiella pneumoniae

Salmonella typhi

Staphylococcus aureus

MIC of methanol 1.25 0.312 1.25 0.625 0.625 0.312 0.312

MIC of ethyl acetate 1.25 0.312 0.625 0.312 0.625 0.625 0.625

Figure 1. DPPH Radical scavenging capacity of different extracts at different concentrations. Each value was the average of triplicates, representing±SD.

(Figure 2). The IC50 values varied in the following order: chloroform < petroleum ether < methanol < < ethyl acetate < BHA, indicating moderate activity of the extracts with regard to the synthetic antioxidant BHA.

Nitric oxide radical scavenging

The toxicity of NO increases greatly when it reacts with superoxide radical, forming the highly reactive peroxynitrite anion (ONOO–) [30]. The gene-rated nitric oxide from sodium nitroprusside reacts with oxygen to form nitrite. The extract inhibits nitrite formation by directly competing with oxygen in the reaction with nitric oxide. In the present study, extracts have less potent nitric oxide scavenging acti-

vity compared to that of the standard ascorbic acid (Figure 3). E.cotinifolia extracts caused a moderate dose-dependent inhibition of nitric oxide with an IC50 (Table 4) in the order of chloroform < petroleum ether < methanol < ethyl acetate < AA. The IC50 value of the extract was slightly less than that of the standard AA.

DISCUSSION

Plants are sources of potent biochemicals. These are obtained from various parts of the plant [31]. Herbal remedies in traditional folk medicine provide a still largely unexplored field for the develop-ment of potentially new drugs for chemotherapy, which helps to overcome the growing problem of drug


25

Figure 2. Hydroxyl scavenging activity of different extracts at different concentrations.

Each value is the average of triplicates, representing±SD.

Figure 3. Nitric oxide radical scavenging of different extracts of E. cotinifolia at different concentrations.

Each value is the average of triplicates, representing±SD.

Table 4. Minimum percentage inhibitory concentration of DPPH radical, hydrogen peroxide and nitric oxide radical scavenging by diffe-rent extracts of E. cotinifolia; SD = standard deviation (average of three replicates); AA = ascorbic acid; BH = butylated hydroxy anisole

Compound (IC50±SD) / µg mL-1

DPPH H2O2 Nitroprusside

Petroleum ether 17±0.23 30±0.03 44±0.64

Chloroform 15±0.35 33±0.14 47±0.56

Ethyl acetate 18±0.30 25±0.04 33±0.75

Methanol 19±0.62 27±0.08 36±0.01

AA 25±0.34 – 23±0.21

BHA – 22±0.08 –


26

resistance and avoid the toxicity of the currently available antibiotics. Therefore, the present research is focused on the above considerations and com-prises the antibacterial activity and antioxidant assay of different solvent extracts of E. cotinifolia.

Several researchers have shown the antibac-terial and antioxidant activity of different Euphorbia sp. and the reports support the use of these plants in traditional medicine for the treatment of various diseases [16-18,32]. E. cotinifolia, the test plant, has showed molluscidal activity [33], moderate antiviral and cytotoxic activity [34]. Antibacterial activity of E. cotinifolia has been evaluated by Rojas and co-workers [20]. In their report, dried leaves of E. coti-nifolia were extracted with isopropyl alcohol and frac-tionated by column chromatography. None of the frac-tions eluted from the column separation of E. coti-nifolia extract showed antibacterial activity against S. aureus, Enterococcus faecalis, E. coli, K. pneumoniae and P. aeruginosa. In the present investigation, the authors have reported soxhlet based extraction suc-cessively from low polar to high polar solvents. These extracts have been subjected to antibacterial activity against E. coli, K. pneumoniae, B. subtilis, B. cereus, S. typhi, E. aerogenes and S. aureus. Antioxidant activity of the extracts has also been evaluated.

The evaluated methanol and ethyl acetate extracts of leaves of E. cotinifolia revealed significant antibacterial activity. The results of antibacterial assay obtained by well diffusion method are in agree-ment with the values of MIC by turbidometric and colorimetric micro titer methods (Tables 2 and 3). The results of the present work indicate that solvent extracts exhibited greater antimicrobial activity because the antimicrobial principle was either polar or non-polar and they were extracted through success-ive solvent extraction, which was supported by many investigators [35,36]. A maximum inhibition zone of 17 mm was recorded in methanol extract against K. pneumoniae, B. subtilis and E. aerogenes, which was slightly less than the standard antibiotic.

The test bacteria subjected to antibacterial acti-vity are associated with different types of infections such as urinary tract infections, wound infections, gastroenteritis and pneumonia [37]. Among all the bacteria tested the Gram-positive bacteria were slightly more susceptible to the extracts than Gram-negative bacteria. The high resistance among the Gram-negative bacteria could be due to the diffe-rences in the cell membrane of these bacteria. The external membrane of Gram-negative bacteria ren-ders their surface highly hydrophilic, whereas lipophi-lic ends of the cell membrane of Gram-positive bac-

teria may facilitate penetration by hydrophobic com-pounds.

Many researchers have shown that methanol is the most suitable solvent for extracting the secondary metabolites from plant materials [38,39]. The signi-ficant inhibition of these bacteria by methanol extract of E. cotinifolia is an indication that there is a possi-bility of sourcing alternative antibiotic substance from this plant for development of new antimicrobial agents.

Phytochemical analysis of methanol and ethyl acetate extracts revealed the presence of flavonoids, terpenoids, tannins and steroids, which could be the active principle. The plants of Euphorbiaceae contain diterpenoids, which have tigliane, ingenane and daphnane skeletons. The toxicity of tannins on micro-organisms operates either by direct action on micro-bial membrane or by metal ion depletion [40]. Flavo-noids and phenolic compounds of plants exhibited antimicrobial activity inhibiting bacterial growth by reacting with DNA and disrupting DNA replication [41]. Herbs belonging to Euphorbiaceae were reported to have antioxidant principles like phenols, flavonoids and terpenoids showing organ protective properties [42,43].

Hydrogen donating ability is an index of primary antioxidants. DPPH Radical was used as a substrate to evaluate the free radical scavenging activity of E. cotinifolia extracts. DPPH Radical contains an odd electron, which is responsible for the blue color. When DPPH accepts an electron from the extracts, it bleaches the blue color; the decrease in absorbance was measured spectrophotometrically at 517 nm. The different extracts of E. cotinifolia having good sca-venging activity with IC50 value at lower concentration of 17, 15, 18 and 19 µg/mL of petroleum ether, chloro-form, ethyl acetate and methanol extracts, respec-tively which is lesser than the IC50 value of standard whose IC50 value is 25 µg/mL. The results indicate that the different solvent extracts are more potent scavenging agents than the standard ascorbic acid.

Hydrogen peroxide is a weak oxidizing agent, but sometimes can be toxic to the cell membrane, generating hydroxyl radical while entering into the cell, which is probably due to reactions with iron(II) and copper(II) ions, which is toxic to the cell. It can inactivate a few enzymes directly, usually by oxidation of thiol (–SH) groups. E. cotinifolia extracts have shown H2O2 decomposition activity at lower concen-tration, i.e., 30, 33, 25 and 27 µg/mL of petroleum ether, chloroform, ethyl acetate and methanol extracts, respectively which is nearly equal to the IC50 value of standard BHA.


27

Nitric oxide (NO) is a potent pleiotropic mediator of physiological processes such as smooth muscle relaxant, neuronal signaling, inhibition of platelet aggregation and regulation of cell mediated toxicity. In pathological conditions, NO reacts with superoxide anion and form potentially cytotoxic molecules. All the extracts showed moderate activity by decreasing the amount of nitrite which is toxic to the cell. The anti-oxidant activity was slightly lower than the standards used, indicating that the extracts have moderate activity.

The results of the present study suggest that tested plant extracts have moderate to potent free radical scavenging activity due to the presence of phenols and flavonoids in the extracts. The defensive effects of natural antioxidants are mainly due to the presence of these major groups, vitamins, phenols, flavonoids and carotenoids [44,45]. The radical sca-venging activity of the E. cotinifolia extracts against DPPH was very effective while slightly less effective against hydroxyl and nitric oxide than the commer-cially available synthetics like BHA and AA.

In the present research, the antioxidant potential and antibacterial activity of E. cotinifolia against human pathogenic bacteria has been evaluated. Methanol extracts of E. cotinifolia posses a broad spectrum of activity against several pathogenic bac-teria. Further investigation is necessary for identi-fication of active compounds of methanol extract, which could lead to the discovery of new antimicrobial drugs from the test plant.

Acknowledgements

The first author, B. Jayalakshmi, is thankful to the University Grants Commission (UGC), New Delhi, India, for awarding FIP - Teacher Fellowship, and also Department of Collegiate Education, Govt. of Karnataka, Bangalore, India, for granting permission to undertake PhD Programme. The authors are grate-ful to UGC, New Delhi, India, for providing the infra-structure facilities developed under “100-Crore Spe-cial Grants of MHRD – UGC – University of Mysore – Centre for Excellence” on the campus and DST-FIST Programme facilities, at the Department of Studies in Botany, University of Mysore, India. We also thank our Department Taxonomists for identifying the plant species used in the study.

REFERENCES

[1] A. D. Khosravi, A. Behzadi, Pak. J. Med. Sci. 22 (2006) 429-432

[2] H. C. Neu, Sci. 257 (1992) 1050-1055

[3] T. G. Winstanley, D. L. Limb, R. Eggenton, F. Hancock, J. Antimicrob. Chemother. 40 (1997) 591-594

[4] D. Leno, Afr. Health Sci. 6 (2006) 31-35

[5] F. Evans, S. Taylor, Pro-inflammatory, tumor promoting and antitumor diterpene of the plant families Eupho-rbiaceae and Thymelaeaceae, in: W. Herz, H. Grisebach, G. Kirby (Eds.), Progress in the chemistry of organic Natural products, Vol. 44, Springer-Verlag, New York, 1983, pp. 1-99

[6] J.H. Doughari, S. Manzara, Afr. J. Microbiol. Res. 2 (2008) 67-72

[7] O. Boussaada, J. Chriaa, S. Ammar, D. Saidana, M.A. Mahjoub, I. Chraeif, A.N. Helal, Z. Mighri, Microbiol. Res. 163 (2008) 87-95

[8] A. Banso, J. Med. Plants. Res. 3 (2009) 82-85

[9] H. Ayub Rashid, M. Zahid Qureshi, A.R. Syed, J. William, M. Arshad, An. Univ. Bucuresti Chim. 19 (2010) 23-30

[10] A. Moure, J. M. Cruz, D. Franco, M. Dominguez, J. Sine-iro, H. Dominguez, Food Chem. 72 (2001) 145-171

[11] H. Ebrahimabadi, E.H. Brahimabadi, Z. Djafari-Bidgoli, J. K. Fereshtec, M. Asma, B. Hossein, Food Chem. 119 (2010) 452-458

[12] J. Terao, M.K. Piskula, Flavonoids as inhibitors of lipid peroxidation in membranes, in: C.A. Rice-Evans, L. Packer (Eds.), Flavonoids in health and disease, Marcel Dekker, New York, 1997, pp. 277-295

[13] G. Webster, Annal. Miss. Bot. Gard. 81 (1994) 133-144

[14] T. Pullaiah, Medicinal plants in India, Regency Pub-lications, New Delhi, 2002, p 241

[15] Natarajan, S. John Britto, K. Srinivasan, N. Nagamu-rugan, C. Mohanasundari, G. Perumal, J. Ethnopharma-col. 102 (2005) 123-126

[16] J. Hohmann, D. Re-dei, P. Forgo, J. Molnar, G. Dombi, T. Zorig, J. Nat. Prod. 66 (2003) 976-979

[17] H. Haba, C. Lavaud, H. Harkat, A. Alabdul, L. Marcourt, M. Benkhaled, Phytochem. 68 (2007) 1255-1260

[18] S. Hore, V. Ahuja, G. Mehta, P. Kumar, S. Pandey, A. Ahmad, Fitoterapia 77 (2006) 35-38

[19] L.H. Bailey, Manual of Cultivated Plants, Revised edition, McMillan, New York, Vol. 1, 1949, p. 88

[20] J. Rojas, J. Velasco, A. Morales, T. Díaz, G. Meccia, BLACPMA 7 (2008) 199-202.

[21] J.B. Harborne, Phytochemical Methods. Chapman and Hall Publications, London, 1998, pp. 7-8

[22] A. Karthikeyan, V. Shanthi, A. Nagasathaya, Int. J. Green. Pharm. 3 (2009) 78-79

[23] J.N. Eloff, Planta Med. 64 (1998) 711-713

[24] J.H. Jorgensen, J.D. Turnidge, J.A. Washington, ASM Press, Washington D.C., 1999, pp. 1526-1562

[25] S. Qaiyami, Macro- and Microdilution methods of anti-microbial susceptibility testing, in: R. Schwalbe, L. Steele- -Moore, A.C. Goodwin (Eds.), Antimicrobial Suscebtibility Testing Protocols, Taylor & Francis Group, New York, 2007, pp. 75-81


28

[26] R. Gupta, M. Sharma, L. Ramakrishnan, D. Prabhakaran, K. Srinath Reddy, Indian J. Biochem. Biophys. 46 (2009) 126-129

[27] M.S. Blois, Nature 181 (1958) 1199-1200

[28] R.J. Ruch, S.J. Cheng, J.E. Klaunig, Carcinogenesis 10 (1989) 1003-1008

[29] L.C. Green, D.A. Wagner, J. Glogowski, P.L. Skipper, J.K. Wishnok, S.R. Tannenbaum, Anal. Biochem. 126 (1982) 131–136

[30] R.E. Huie, S. Padmaja, Free Radical Res. Commun. 18 (1993) 195-199

[31] M.M. Cowan, Clin. Microbiol. Rev. 12 (1999) 564-582

[32] T.M. Jyothi, M.M. Shankaraiah, K. Prabhu, S. Lakshmi-narasu, G.M. Srinivasa, S. Ramachandra Setty, Iranian J. Pharmacol. Ther. 7 (2008) 25-30

[33] J.P. Pereira, C.P. de Souza, N.M. Mendes, Rev. Bras. Pesqui. Med. Biol. 11 (1978) 345-351

[34] L.A. Galvis Betancur, G.E. Morales, J.E. Forero, J. Roldan, Mem. Inst. Oswaldo Cruz Rio de Janeiro 97 (2002) 541-546

[35] Mohanasundari, D. Natarajan, K. Srinivasan, S.A. Umamaheswari, A. Ramachandran, Afr. J. Biotechnol. 6 (2007) 2650-2653

[36] M.P. Raghavendra, S. Sathisha, K.A. Raveesha, Arch. Phytopathol. Plant Prot. 40 (2008) 1-9

[37] S. Gupte, The short text book of medical microbiology. 9th ed., Jaype Brothers Medical Publishers, New Delhi, 2006, pp. 43-56

[38] H.J. De Boer, A. Kool, A. Broberg, W.R. Mziray, I. Fiedberg, J.J. Levenflors, J. Ethnopharmacol. 96 (2005) 461-469

[39] R. Ahameethunisa, W. Hooper, BMC Complement Altern. Med. 10 (2010) 1-6

[40] M. McDonald, I. Mila, Scalbert, J. Agric. Food Chem. 44 (1996) 599-606

[41] S. Maleki, S.M. Seyyednejad, N. Mirzane Damabi, H. Motamedi, Pak. J. Bio. Sci. 11 (2008) 1286-1289

[42] V.S. Pracheta, R. Paliwal, S. Sadhana, Int. J. Pharm. Tech Res. 3 (2011) 124-132

[43] S. Kumar, M. Rashmi, K. Dinesh, Int. Res. J. Pharm. 1 (2010) 150-156

[44] T.P.T. Cushine, A.J. Lamb, Int. J. Antimicrob. Agents 26 (2005) 343-356

[45] K. Thaipong, U. Boonprakob, K. Crosby, L. Cisnero-Zevallos, D.H. Byrne, J. Food Comp. Anal. 19 (2006) 669-675.

B. JAYALAKSHMI1

K.A. RAVEESHA2

K.N. AMRUTHESH3

1Department of Botany, Maharani’s Science College for Women, J.L.B.

Road, Mysore, Karnataka, India 2Herbal Drug Technology Laboratory,

Department of Studies in Botany, University of Mysore, Manasagangotri

Mysore, Karnataka, India 3Applied Plant Pathology Laboratory,

Department of Studies in Botany, University of Mysore, Manasagangotri

Mysore, Karnataka, India

NAUČNI RAD

EVALUACIJA ANTIBAKTERIJSKOG I ANTIOKSIDATIVNOG POTENCIJALA EKSTRAKTA LIŠĆA Euphoria cotinifolia LINN.

Antibakterijska aktivnost različitih ekstrakata dobijenih iz lišća Euphoria cotinifolia je

testirana u odnosu na humane patogene bakterije Escherichia coli, Klebsiella pneumonia,

Bacillus subtilis, Bacillus cereus, Salmonella typhi, Enterobacter aerogenes i

Staphilococcus aureus pomoću disk-difuzione i mikro-dilucione metode. Antioksidativna

svojstva ekstrakata su određene pomoću 1,1-difenil-2-pikril-hidrazila (DPPH), azot(II)-

-oksida (NO) i vodonik-peroksida. Dobijene IC50 vrednosti su izračunate i poređene sa

askorbinskom kiselinom i butilovanim hidroksianizolom kao standardima. Ekstrakti dobijeni

pomoću metanola i etil-acetata kao rastvarača, pokazuju značajnu aktivnost u odnosu na

B. subtilis i E. aerogenes sa zonama inhibicije 17,25 mm. Minimalna inhibitorna kon-

centracija pomenutih ekstrakata za različite bakterije kreće se u intervalu 0,3-1,25 mg/ml.

DPPH metodom je utvrđeno da su IC50 vrednosti za ekstrakte sa hloroformom, petrol-

-etrom, etil-acetatom i metanolom (15, 17, 18 i 19 mg/ml, redom), manje od vrednosti

dobijene za askorbinsku kiselinu kao standard (25 mg/ml). Fitohemijska analiza vodenog,

etil-acetatnog i metanolskog ekstrakta je pokazala prisustvo flavonoida, terpenoida, tanina

i steroida.

Ključne reči: Euphoria cotinifolia, antibakterijska aktivnost, antioksidativna aktivnost, flavonoidi, steroidi.





29

JIAN DING

HUIHUI WANG KEKE DAI YUHUA ZI

ZHONGPING SHI

Key Laboratory of Industrial Biotechnology, Ministry of

Education, School of Biotechnology, Jiangnan

University, Wuxi, China

SCIENTIFIC PAPER

UDC 579.62:616-097:004:519.87

DOI 10.2298/CICEQ120704097D

PREDICTION OF PORCINE INTERFERON α ANTIVIRAL ACTIVITY IN FERMENTATION BY Pichia pastoris BASED ON MULTIVARIABLE REGRESSION AND ARTIFICIAL NEURAL NETWORK

Article Highlights • In this work, porcine interferon α was produced by Pichia pastoris in fermentor • An ANN model was built to describe the relationship between antiviral activity and

process variables • The model was optimized by back-propagation and genetic algorithm respectively • The model optimized by genetic algorithm performs well in predicting antiviral activity Abstract

One of the most important methods for production of porcine interferon α is microbial fermentation. In the present study, recombinant Pichia pastoris was used. Broth’s antiviral activity is the key index of the expression level of porcine interferon α. Measurement of antiviral activity is a time-consuming and difficult task, which makes the research and production work inconvenient and uncer-tain. To solve this problem, multivariable regression and artificial neural net-work were applied to predict the antiviral activity based on five on-line vari-ables (induction time, temperature, dissolved oxygen, O2 uptake rate and CO2 evolution rate) and two off-line variables (methanol consumption rate and total protein concentration). Parameters of the multivariable quadratic polynomial regression equation were estimated using least square methods. Optimization of artificial neural network (ANN) was achieved by back-propagation and gene-tic algorithm. Verified by test set, the ANN optimized by genetic algorithm had the best predictive performance and generalization. The sensitivity analysis showed that CO2 evolution rate, O2 uptake rate and methanol consumption rate were the most relevant factors for the model’s output, except for the anti-viral activity’s own previous value.

Keywords: porcine interferon α, antiviral activity, multivariable regression, artificial neural network, back-propagation algorithm, genetic algorithm.

Porcine interferon α (pIFN-α), a broad-spectrum antiviral veterinary drug, has good effects on prevention and treatment of many frequent viral infectious diseases [1,2]. pIFN-α is also widely utilized due to its features of lower cost and fewer side-effects. Microbial fermentation has recently become one of the major measures to increase pIFN-α yield.

Correspondence: J. Ding, School of Biotechnology, Jiangnan University, 1800 Lihu Road, Wuxi 214122, Jiangsu Province, China. E-mail: [email protected] Paper received: 4 July, 2012 Paper revised: 25 October, 2012 Paper accepted: 26 October, 2012

In a shake flask, pIFN-α with antiviral activity has been successfully expressed in E.coli and P. pastoris systems [3,4]. For future work, it is urgent to realize pIFN-α’s high level expression in a fermentor of labo-ratory scale or even of industrial scale. As the main index of pIFN-α’s expression level in fermentation, antiviral activity of broth guides all the work focused on optimizing fermentation technology and building fermentation system. However, measuring antiviral activity is really complex and time-consuming work involving animal cell culture which is hard to operate in a microbial fermentation laboratory or factory, thus no measurement results can be obtained in time, making research work and production process uncer-

J. DING et al.: PREDICTION OF PORCINE INTERFERON α ANTIVIRAL ACTIVITY… Chem. Ind. Chem. Eng. Q. 20 (1) 29−38 (2014)

30

tain. The antiviral activity values are unknown until the end of fermentation and much manpower, time, and finances is commonly wasted for a failed batch. Therefore, it is crucial to avoid uncertainty in research and production when fermentation is still in progress. The key to solving the problem is to predict antiviral activities by building an accurate soft sensor model based on fermentation process variables. Thus, multi-variable regression (MR) and artificial neural network (ANN) were employed.

MR is a classical approach to seeking relation-ship between two or more independent variables and dependent variables. In MR, a proper regression equa-tion, such as multivariable linear function or multi-variable polynomial, should be firstly selected or built. The parameters of the equation need to be estimated by using least square method. For simple processes or those for which precise mathematical equations are known, MR is an ideal tool to model them [5,6].

ANN, a mathematical model built by imitating the structure of the human brain, consists of a group of artificial neurons. Each pair of two interconnected neurons has its own weight; each single neuron, except for the ones in input layer, has its own bias; each layer, except for the input layer, has its own activation function. Information from inputs transfers from layer to layer following the rules determined by weights, biases and activation functions. The func-tional relationship between inputs and outputs is des-cribed by its transfer function. Learning ability is ANN’s most significant feature. The training process of an ANN is actually the process of optimizing weights and biases. Due to its complex structure and optional nonlinear activation functions, an ANN model has the ability to describe complex nonlinear func-tional relationships accurately. With the obvious advantage, ANN has been applied in many fields to model processes and predict target variables [7–10]. However, its application in the fermentation producing target protein by recombinant P. pastoris, especially pIFN-α, has seldom been reported.

To make ANN model accurately describe the specific functional relationship, training or parameters optimization is essential. Back-propagation (BP) algo-rithm is the most commonly and widely used algo-rithm for optimizing ANN model’s parameters. It is a supervised learning method requiring desired output for any input. Error signal is calculated with desired output and actual output at first. Then model’s para-meters are corrected according to the propagation of error signal from output layer to input layer [11]. Although BP algorithm is popular and has solved a series of thorny problems, it performs worse than

genetic algorithm (GA) in searching global best solu-tions. GA belongs to evolution algorithms, and it is an effective optimization algorithm inspired by the evo-lution action of population existing in nature world. Solutions are coded to binary strings, which are con-sidered as chromosomes. The chromosome repre-senting best solution can be generated and retained by genetic operations including selection, crossover, and mutation. Because of its excellent global search-ing ability and few restriction of objective function, such as continuity and differentiability, GA has been applied in numbers of optimization issues [12–15].

Thus, MR and ANN were employed to describe the functional relationship between fermentation process variables and broth’s antiviral activity. Nine process variables affecting antiviral activity were selected as a matter of experience. Determined by the number of the variables, a multivariable quadratic polynomial with nine independent variables and a three-layer ANN model with nine neurons in input layer were built. The activation functions of hidden and output layers were sigmoid function and linear function respectively. Antiviral activity was the only dependent variable (output) of the models. Ferment-ation data from fourteen batches in different ferment-ation status (failed or successful) was divided into training set and test set, and they were normalized into a range from 0 to 1. The number of neurons in ANN’s hidden layer being set from 3 to 15, the model was trained for the same times respectively. The minimal mean square error (MSE) was obtained when 10 neurons were included in hidden layer, so the number of neurons in hidden layer was decided. The quadratic polynomial’s and ANN model’s parameters were optimized with the same training set. Both BP and GA were applied in ANN model’s training, and least square method was used in the parameters’ estimation of quadratic polynomial. Three optimized models were obtained. The models’ predictive perfor-mances and generalizations were evaluated by com-paring MSE and correlation coefficients of test set, R2. Results showed that the ANN model optimized by GA is superior to the others. Finally, sensitivity analysis was carried out to reveal nine process variables’ correlations with model’s output.


Strain

Expression plasmid pPICZ-αIFN was con-structed by ligation of pINF-α gene into pPICZα (Invi-trogen, Carlsbad, CA, USA) at downstream of the pro-moter AOX1. pPICZ-αIFN was linearly integrated into


31

the chromosome DNA of the host P. pastoris KM71 (Muts his-, PAOXII, Invitrogen, Carlsbad, CA, USA) before transformation. The construction of the recom-binant P. pastoris KM71 (IFNα-pPICZαA) was imple-mented at the Animal Husbandry and Veterinary Research Institute, Shanghai Academy of Agricultural Science, China.

Fermentation medium

Seed medium (in g/L, unless otherwise spe-cified): glucose 20, peptone 20, yeast extract 10.

PTM1 (in g/L, unless otherwise specified): CuSO4·5H2O 6, NaI 0.08, MnSO4·H2O 3, Na2MoO4·2H2O 0.2, H3BO3 0.02, CoCl2 0.5, ZnCl2 20, FeSO4·7H2O 65, biotin 0.3, H2SO4 5 mL.

Bath medium for jar fermentor (in g/L, unless otherwise specified): glycerol 20, MgSO4 1, K2SO4 1, (NH4)2SO4 5, CaSO4 0.1, H3PO4 2%, v/v, PTM1 10 mL/L, pH 6.0.

Feeding medium for growth (in g/L, unless otherwise specified): glycerol 500, (NH4)2SO4 0.5, KH2PO4 0.5, MgSO4 0.03, PTM1 10 mL/L, pH 6.0.

Feeding medium for induction (in g/L, unless otherwise specified): methanol 500, (NH4)2SO4 0.5, KH2PO4 0.5, MgSO4 0.03, PTM1 10 mL/L, pH 5.5.

Mixed feeding medium for transition phase: growth medium/induction medium volumetric ratio of 25:1.

pIFN-α expression by P. pastoris fed-batch cultivation in 5 L bioreactor

Fed-batch cultures were carried out in a 5 L bench-scaled bioreactor (BIOTECH-5BG, Baoxing Co., China), with the initial batch medium of 1.5 L. The entire operation details for growth, transition and induction phases were exactly the same as those described in the previous report [16]. The previously proposed ANNPR-Ctrl approach [17] was used for feeding glycerol and glycerol/methanol mixture during growth phase (26 h) and the short transition phase (3–4 h), allowing cells to reach high density (about 120– -150 g DCW/L) [18]. The induction phase was started by feeding methanol-based medium at about 30 h after glycerol was completely used out. During induc-tion phase, cell concentration basically stayed at a high density level (120–150 g DCW/L) constantly with-out significant variation. Based on the off-line measu-rement, methanol concentration was control at any required level by adjusting rotate speed of peristaltic pump (BT11-50M, Langer Co., China). An electronic balance (JA1102, Haikang Electronic instrument Co., China) connected to the PC via RS232 communi-cation cable was used to on-line monitor and to cal-culate the methanol consumption rate by weighing the weight loss of methanol feeding reservoir. The O2 and

CO2 partial pressures in the exhaust gas, as well as the corresponding O2 uptake rate (OUR) and CO2 evolution rate (CER), were measured on-line and cal-culated by a gas analyzer (LKM2000A, Lokas Co., Korea). These data were also collected into the PC via RS232 communication cable for storage at 1 min intervals. If O2 partial pressure exceeded the measu-rement range (0–30%) of the analyzer when sparging oxygen-enriched air, the exhaust gas measurements were stopped. Temperature was controlled within the range of 20–30 °C according to requirements through-out the induction phase, by using either tap water or water from a temperature-controllable circulating bath (MP-10, Shanghai Permanent Science and Techno-logy, Co., China). After shifting into induction phase, the aeration and agitation rates were fixed without further manual adjustment.

Analytical methods

Measurements of methanol concentration. Metha-nol was detected by a gas chromatograph (GC112A, FID detector, Shanghai Precision & Scientific Instru-ment Co., China) with an Alpha-Col AC20 capillary column (SGE Int’l Pty. Ltd., Australia).

Measurements of pIFN-α antiviral activity. The samples were centrifuged at 11000 rpm for 10 min and the supernatant was used for the determination of pIFN-α antiviral activity. The pIFN-α antiviral acti-vity was determined according to Chinese pharmaco-poeia [19], using human amniotic cell WISH and vesi-cular stomatitis virus (VSV) (Wanxing Bio-Pharma-ceutical Co., China). The commercially available human interferon was used as the standard sample. Before measurement, the standard and fermentation supernatant samples were properly diluted according to specific requirements. The pIFN-α antiviral activity was defined as the reciprocal of the maximal dilution rate, and under this rate, 50% cytopathic inhibition or 50% virus plaque formation could be reached. IU refers to the abbreviation of “International Unit”.

Measurements of total protein concentration. Total protein concentrations in broth were measured using Coomassie Brilliant Blue G-250 [20].

Calculations

Training and test set designing

Selecting the model’s inputs (independent vari-ables) was mainly based on the experiences accu-mulated in a long term of fermentation research. In induction phase of the fermentation mentioned above, broth’s antiviral activity (AA) increased accompanied by the accumulation of pIFN-α. Increasing rate of antiviral activity was reflected or determined by some


32

process variables, which were temperature (T), dis-solved oxygen (DO), O2 uptake rate (OUR), CO2 evolution rate (CER), methanol consumption rate (MCR) and total protein concentration in broth (TPC). TPC and AA increased gradually with pIFN-α’s syn-thesis, so the TPC and AA values both in current and previous sampling points must be considered as inputs when modeling. Besides this, microbes’ meta-bolic activity and pIFN-α’s synthesis rate had their own variations even if all the operation conditions remained constant, so the induction time (t) was also selected as one of the model’s inputs.

On-line variables including t, T, DO, OUR, CER were all saved by PC automatically at 1 min intervals. Methanol concentration (MC, g/L) was measured off-line by gas chromatography at 120 min intervals and the methanol feeding amount (MFA, g) between two sampling points could be calculated based on the weight loss of methanol feeding reservoir. MCR (g L-1 h–1) could be calculated by Eq. (1):

previous current

2

MC V MC V MFAMCR

V

× − × += (1)

where MCcurrent and MCprevious represent methanol con-centrations in current and previous sampling points respectively, MFA is the methanol feeding amount between the two sampling points and V (L) is the volume of broth. TPC and AA were measured off-line without constant interval, but they were regressed by two RBF-ANN models [21]. A series of TPC and AA values at N min intervals used by training and testing could be estimated by the models. So the nine inputs, also namely independent variables, were chosen and expressed by Eqs. (2) to (10):

1x t= (2)

( )2

1 t

i t Nx T i

N = −= (3)

( )3

1 t

i t Nx DO i

N = −= (4)

( )4

1 t

i t Nx OUR i

N = −= (5)

( )5

1x CER

= −=

t

i t Ni

N (6)

( )6x MCR t= (7)

( )7x TPC t= (8)

( )8x TPC t N−= (9)

( )( )9 logx AA t N−= (10)

The output (dependent variable) was:

y =log (AA(t))

Data of fourteen batches of fermentation with different status (failed or successful) was collected in experiments. Both of the antiviral activity and the process variables affecting it distributed in large ranges. Their lower and upper bounds were shown in Table 1. The data diversity could ensure the accuracy of model. These data need to be divided into training set and testing set. At first, batches were sorted by ascending their maximum antiviral activities. The data coming from the batches located in third, sixth, ninth, and twelfth of the arrangement was selected as test-ing set and other was training set. The method divid-ing data was reasonable because ANN models are empirical, and they should be used for interpolation and not extrapolation. After divided, training and test set were normalized to the range from 0 to 1.

Table 1. Lower and upper bounds of antiviral activity and pro-cess variables affecting it

Process variable Lower bound Upper bound

t / h 10 91.8

T / °C 19.3 30.7

DO / % 0 72.9

OUR / mmol L–1 h–1 0 377.9

CER / mmol L–1 h–1 0 182.8

MCR / g L–1 h–1 0 57.6

TPC / g L–1 0.22 2.2

log AA 0 7.3

Models building

It is widely known that the fermentation process is a highly nonlinear system. So the commonly used multivariable linear function is unsuitable to model for the process. Considering the fermentation system’s nonlinearity, the multivariable quadratic polynomial was introduced as MR’s regression equation. Three types of formulas are included in multivariable quad-ratic polynomial: 1) pure quadratic type with constant, linear and squared terms; 2) interaction type with constant, linear and interaction terms; 3) full quadratic type with constant, linear, interaction and squared terms all. For the aforementioned fermentation pro-cess, the first one was superior to the others in fitting (data not shown), so Eq. (11) was used as the regres-sion equation for MR:

20

1 9 1 9i i jj j

iy x xβ β β

≤ ≤ ≤ ≤= + +

j (11)

An ANN model with three layers (Figure 1) was built. Hypothesizing number of neurons in hidden


33

layer was m, then W1 was an m×9 matrix including all of the weights between input layer and hidden layer, element in row i and column j was the weight value between jth neuron in input layer and ith neuron in hidden layer. b1 was an m×1 matrix, elements in which representing bias values of hidden layer. m ele-ments n 1×m matrix W2 represented m weights between m neurons in hidden layer and the sole neu-ron in output layer respectively. b2 was a scalar repre-senting the bias value of the neuron in output layer. Sigmoid function and linear function were selected as activation functions of hidden layer and output layer respectively. If inputs were represented by X = = (x1,x2,x3,x4,x5,x6,x7,x8,x9)

T, the output of the ANN model could be calculated by the following equation:

( )( )2 2 1 1 1 2y f W f W X b b= + + (12)

Parameters estimation and optimization

The regression equation for MR was decided. Then its parameters need to be estimated with train-ing set including K pairs of independent variables and dependent variable. In MR, parameters estimation was actually optimizing parameters to minimize resi-dual sum of squares, which is defined by the following equation:

( )2

1

ˆK

i

Q y y=

= − (13)

where y is the observed value and y is the estimated value. Minimizing Q, optimal βi (1≤i≤9) and βjj (1≤jj≤9) were computed. For comparing with the ANN model, MSE in MR was also calculated, and found to be 0.0023. The regression equation with estimated para-meters was named as MR-MQP.

Training (optimization) was essential for the ANN model to describe specific relationship between inputs (fermentation process variables) and output (pIFN-α antiviral activity). The same training set was used. Desired outputs and real outputs were repre-sented by s(i) and p(W1,W2,b1,b2,i), respectively. MSE could be calculated by the following equation:

( )

( )

1 2 1 2

2

1 2 1 2

1

, ,

1( ) , , , ,

K

MSE W W b b

s i p W W b b iK =

=

= − i

,

(14)

The optimization target was searching the mini-mal value of MSE, as well as the parameters (W1,W2,b1,b2) values when MSE reached the mini-

f1

f1

f1

f1

f1

f1

f1

f1

f1

f1

f2

W1

W2

x1

x2

x3

x4

x5

x6

x7

x8

x9

y

b1 b2

Input layer Hidden layer Output layer

f1

f1

f1

f1

f1

f1

f1

f1

f1

f1

f2

W1

W2

x1

x2

x3

x4

x5

x6

x7

x8

x9

y

b1 b2

Input layer Hidden layer Output layer

Figure 1. Topology of the ANN model: nine neurons were included in input layer; ten neurons were included in hidden layer; one neuron was included in output layer. Activation functions, f1 and f2, were sigmoid function and linear function, respectively.


34

mum. So the parameter optimization problem could be described as:

( )1 2 1 2

1 2 1 2, ,min , ,

W W b bMSE W W b b

,, (15)

The number of neurons in hidden layer should be decided firstly before normal training. Fixing target MSE and training times, the model was trained with different neuron numbers in hidden layer, from 3 to 15. Minimal MSE was obtained when the number is 10. So in next work, m was set as 10. Then the optimization problem described by Eq. (15) was solved by BP at first with 0.005 as target MSE value. Through 2810 times of iterations, MSE converged to the target value. The variation of MSE in training is shown in Figure 2A.

In GA optimization, fitness proportionate selec-tion, which was also called roulette-wheel selection, would be adopted. The selection method demanded that fitness values must be positive and optimization target must be maximizing fitness function. To meet the demand, the optimization problem in Eq. (15) was converted to its equivalent form expressed by Eq. (16):

( )1 2 1 2

1 2 1 2, ,max , ,

W W b bMSE W W b b

,, (16)

where F = 1/(1+MSE), and was directly considered as fitness function. Initial population including 20 chro-mosomes in binary form was generated randomly, representing 20 solutions. Exploitation of new solu-tions was achieved by genetic operations including crossover and mutation. Crossover probability and mutation probability were set as 0.05 and 0.005, res-pectively. Twenty chromosomes were selected using the roulette-wheel method, constituting the population of first generation. The whole process including cross-

over, mutation and selection was considered as one time of iteration. After 200 times of iterations, maximal fitness value in the whole population converged to 0.9830, which meant the MSE converged to 0.0173. The variation of MSE is shown in Figure 2B. The ANN models optimized by BP algorithm and GA were saved and named as BP-ANN and GA-ANN, respecti-vely. Predictive performances and generalizations of BP-ANN and GA-ANN were evaluated by the test set designed above. Finally, sensitivity analysis was car-ried out. Removing each input from training set res-pectively, nine new training sets with eight inputs only was generated. Then an ANN model was trained for the same times (300 times) by BP algorithm with the nine new training sets respectively. The nine inputs’ importance was studied by comparing the training accuracies in the absence of themselves. Lower train-ing accuracy (higher MSE) meant stronger sensitivity of the absent input. Calculation involving estimating parameters, optimizing parameters and evaluating models’ performances was implemented in Matlab soft-ware (version R2010b, The MathWorks Inc., USA).


Predictive abilities of MR-MQP, BP-ANN and GA-ANN were verified by the test set. The prediction period N being set as 240 min, test data was inputted to models conforming to Eqs. (2)–(10). It should be emphasized that the ninth input x9 here was the pre-dictive value of previous sampling point, whereas in optimizing x9 was the measured value of previous sampling point. Outputs of models were a series of pIFN-α antiviral activities corresponding to induction time x1. Comparison between the measured values and predictive values calculated by the three models are shown in Figure 3. It could be observed that the

0 500 1000 1500 2000 2500 3000

0.00

0.05

0.10

0.15

0.20A

Iterative times

MSE

0 50 100 150 2000

1

2

3

4

5B

Iterative times

MSE

Figure 2. Variations of MSE values when implementing BP algorithm and GA. A: BP algorithm; B: GA.


35

predictive results of GA-ANN were closer to the mea-sured values than those of MR-MQP and BP-ANN. MR-MQP’s predictive values were far away from the measured ones. MSE and correlation coefficients R2 for each group of test data could be used to accu-rately evaluate predictive performances of the models. MSE of each batch were calculated by the Eq. (14) based on the predictive results of the three models (Figure 4A). MSE value closer to 0 means better predictive performance. GA-ANN’s MSE were the smallest except for batch #3. Although MR-MQP had the smallest MSE for batch #3, its MSE for the other batches were too high to be accepted. R2 of test set for models were computed and compared in Figure 4B. In contrast to MSE, R2 value closer to 1 means better predictive performance. MR-MQP had the highest R2 for batch #3 but for the other batches its R2 values were too low. R2 values of GA-ANN were the highest except that for batch #3, which was still at an acceptable level. So the three models’ MSE and R2 for all the batches being comprehensively anal-yzed, GA-ANN had the best predictive performance and generalization, and MR-MQP behaved worst in prediction.

The prediction result of MR-MQP was unsatis-factory, which might be because the fermentation process producing pIFN-α by P. pastoris is too com-plicated for a multivariable quadratic polynomial to describe. A terrible result was inevitable when a com-plex system is described with a simple model. In future work, it is necessary to discuss whether other improved elementary functions could bring better results when they are used as regression equations in MR. Compared with elementary functions, ANN had strong ability to model for complicated and nonlinear system. BP algorithm and GA have been widely used in ANN model’s optimization. In the optimizing pro-cess implemented by the BP algorithm, smaller MSE between output and expected output could be obtained compared with GA. For example, in the aforemen-tioned work, MSE reached 0.005 in BP algorithm training, whereas in GA optimizing MSE only con-verged to 0.0173. However, the BP algorithm has some disadvantages, the most common of which are overfitting and local optima. Overfitting generally occurs when an ANN model is excessively complex, such as having too many neurons in hidden layer, or having too many hidden layers. An ANN model which has been overfit will generally have poor predictive

0 10 20 30 40 50 60 70100

102

104

106

108

batch#1 Ant

ivira

l act

ivity

(UI/m

l)

Ant

ivira

l act

ivity

(UI/m

l)

Induction time (h)0 10 20 30 40 50 60 70

100

102

104

106

108

Induction time (h)

0 10 20 30 40 50 60 70100

102

104

106

108

Ant

ivira

l act

ivity

(UI/m

l)

Induction time (h)0 10 20 30 40 50 60 70

100

102

104

106

108

batch#4batch#3

batch#2

Ant

ivira

l act

ivity

(UI/m

l)

Induction time (h)

Figure 3. Comparison between measured and predictive antiviral activities: open circle: measured values; dash line: values predicted by GA-ANN; dot line: values predicted by BP-ANN; dash dot line: values predicted by MR-MQP.


36

performance, as it can exaggerate minor fluctuations in the data. Local optima of an optimization problem are solutions those are optimal within neighboring sets of solutions. Optimal solutions searched by BP easily are trapped in local optima for BP’s character of local search, and GA is expected to avoid local optima due to its excellent global searching ability [22]. In the present work BP algorithm performed worse than GA. The probable reason may be overfit-ting or local optima. Changing ANN model’s topology or repeated training may further improve the BP algo-rithm’s performance.

The correlations of independent variables with the model’s output were finally studied. If the absence

of one input leads to a severe deterioration of training accuracy, which is marked by a much higher MSE, it is highly relevant to model’s output. Figure 5 shows the MSE values in training when each input was absent. The antiviral activity in previous sampling point (x9) had the most influence on the model’s accu-racy. CER (x5), OUR (x4) and MCR (x6) were the factors mostly affecting antiviral activity except for its own previous value. The other inputs by descending importance were DO (x3), previous TPC (x8), current TPC (x7), t (x1) and T (x2).

The fermentation process producing pIFN-α by P. pastoris was divided into three phases: growth phase, induction phase and a short transition phase

2.48

1.56

2.55

4.76

0.890.61

1.27

0.59

8.267.91

0.35

5.79

0.65

0.77

0.680.7

0.79

0.85

0.77

0.86

0.35

0.47

0.87

0.56

batch #1 batch #2 batch #3 batch #40

3

6

9

B

MSE

BP-ANN GA-ANN MR-MQP

A

batch #1 batch #2 batch #3 batch #40.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

R2

BP-ANN GA-ANN MR-MQP

Figure 4. Comparison of MSE and R2 values for test set in prediction: A: MSE; B: R2.


37

between them. Cells grew quickly in glycerol-based medium in the growth phase and cell concentration reached 120-150 g DCW/L. During the induction phase, cell concentration basically stayed at the high density level constantly without significant variation. In research of fermentation, measuring pIFN-α anti-viral activity of broth is a time-consuming and difficult task. A measuring procedure does not generally start with for a few samples. The antiviral activities are always measured after the end of a number of batches. From animal cell cultivation to the measurement, around 7 days are needed. So real-time monitoring of the expression of pIFN-α is impossible, which brings inconveniences for fermentation research at the lab scale and industrial scale production. To avoid direct measurement of antiviral activity, a soft sensor method was proposed to realize real-time monitoring based on fermentation variables that can be mea-sured online or easily and quickly measured offline. With the ANN optimized by GA, pIFN-α antiviral acti-vity, which represents the expression level of pIFN-α, could be predicted in the fermentation process, providing significant basis for research.

CONCLUSION

In pIFN-α fermentation by P. pastoris, multiva-riable regression and ANN were introduced to model for the functional relationship between antiviral acti-vity and process variables including induction time, temperature, dissolved O2, O2 uptake rate, CO2 evo-lution rate, methanol consumption rate and total pro-tein concentration. Parameters of regression equation

and ANN model were optimized and three models were obtained to predict antiviral activity. Verified by test set, the ANN model optimized by GA had the best predictive ability and generalization. In sensitivity analysis, CER, OUR and MCR were the factors having the most correlations with antiviral activity except the previous value of its own.

REFERENCES

[1] L.T. Jordan, J.B. Derbyshire, Vet. Microbiol. 38 (1994) 263-276

[2] J. Chinsangaram, M. Koster, M.J. Grubman, J. Virol. 75 (2001) 5498-5503

[3] F. Lefevre, C. La Bonnardiere, J. Interferon. Res. 6 (1986) 349-360

[4] R.B. Cao, X.Q. Xu, B. Zhou, D.S. Chen, P.Y. Chen, Chin. J. Biotechnol. 20 (2004) 291-294

[5] A. Aranda, G. Ferreira, M.D. Mainar-Toledo, S. Scarpellini, E. Llera Sastresa, Energy Buildings 49 (2012) 380-387

[6] R. Mac Nally, Biodivers. Conser. 11 (2002) 1397-1401

[7] U. Atici, Expert Syst. Appl. 38 (2011) 9609-9618

[8] S.C. Chelgani, J.C. Hower, E. Jorjani, S. Mesroghli, A.H. Bagherieh, Fuel Process. Technol. 89 (2008) 13-20

[9] A. Celekli, S.S. Birecikligil, F. Geyik, H. Bozkurt, Bio-resour. Technol. 103 (2012) 64-70

[10] T.J. Shankar, S. Bandyopadhyay, Food Bioprod. Proc. 85 (2007) 29-33

[11] S. Agatonovic-Kustrin, R. Beresford, J. Pharm. Biomed. Anal. 22 (2000) 717-727

[12] A.S. Sahin, B. Kilic, U. Kilic, Heat Mass Transfer 47 (2011) 1553-1560

X1 X2 X3 X4 X5 X6 X7 X8 X90.005

0.010

0.015

0.020

MSE

Figure 5. Result of sensitivity analysis: MSE values in training in the absence of each input.


38

[13] S.-H. Chen, P.-C. Chang, T.C.E. Cheng, Q. Zhang, Comput. Oper. Res. 39 (2012) 1450-1457

[14] H. Nazif, L.S. Lee, Appl. Math. Model. 36 (2012) 2110- –2117

[15] S.K. Lahiri, N.M. Khalfe, Chem. Ind. Chem. Eng. Q. 15 (2009) 175-187

[16] R. Yu, S. Dong, Y. Zhu, H. Jin, M. Gao, Z. Duan, Z. Zheng, Z. Shi, Z. Li, Bioprocess Biosyst. Eng. 33 (2010) 473-483

[17] H. Jin, Z. Zheng, M. Gao, Z. Duan, Z. Shi, Z. Wang, J. Jin, Biochem. Eng. J. 37 (2007) 26-33

[18] H. Jin, G. Liu, X. Ye, Z. Duan, Z. Li, Z. Shi, Biochem. Eng. J. 52 (2010) 91-98

[19] Chinese pharmacopoeia, People' s Health Press, Beijing, 2005, p. 343

[20] J.J. Sedmak, S.E. Grossberg, Anal. Biochem. 79 (1977) 544-552

[21] S. Seshagiri, H.K. Khalil, IEEE Trans. Neural Networks 11 (2000) 69-79

[22] A. Ghaffari, H. Abdollahi, M.R. Khoshayand, I.S. Bozcha-looi, A. Dadgar, M. Rafiee-Tehrani, Int. J. Pharm. 327 (2006) 126-138.

JIAN DING

HUIHUI WANG

KEKE DAI

YUHUA ZI

ZHONGPING SHI

Key Laboratory of Industrial Biotechnology, Ministry of Education,

Jiangnan University, Wuxi, China

NAUČNI RAD

PREDVIĐANJE ANTIVIRUSNE AKTIVNOSTI SVINJSKOG INTERFERONA-α TOKOM FERMENTACIJE SA Pichia pastoris POMOĆU MULTIVARIJANTNE REGRESIJE I VEŠTAČKE NEURONSKE MREŽE

Jedna od najznačajnijih metoda za dobijanje porcin interferon-α je mikrobiološka

fermentacija. U ovom radu je korišćen rekombinantni soj Pichia pastoris. Anivirusna ak-

tivnost fermentacionog medijuma je ključni indeks nivo ekspresije svinjskog interferona-α.

Merenje antivirusne aktivnosti je dugotrajan i mukotrpan zadatak, koji čini istraživački i

proizvodni rad nepogodnim i neizvesnim. Da bi se rešio ovaj problem, primenjene su

multivarijantna regresija i veštačka neuronska mreža za predviđanje antivirusne aktivnosti

na osnovu pet on-line promenljivih (vreme indukcije, temperature, rastvoreni kiseonik,

brzina potrošnje kiseonika i brzina izdvajanja CO2 i dve off-line promenljive (brzina trošenja

metanola i koncentracija ukupnih proteina). Parametri kvadratne polinomne regresione

jednačine su izračunati metodom najmanjih kvadrata. Optimizacija veštačke neuronske

mreže je postignuta algoritmom sa povratnom propagacijom i genetskim algoritmom.

Veštačka neuronska mreža, verifikovana i optimizovana genetskim algoritmom, imala je

najbolju moć predviđanja i generalizacije. Analiza osetljivosti je pokazala da su brzina

evolucije CO2, brzina potrošnje kiseonika i brzina potrošnje metanola najrelevantniji faktori

za rezultat modela, izuzev za prethodne vrednosti antivirusne aktivnosti.

Ključne reči: svinjski interferon-α, antivirusna aktivnost, multivarijantna regresija, veštačka neuronska mreža, algoritam sa povratnom propagacijom, genetski algoritam.





39

SHWETA S. HAVELE1

SUNIL R. DHANESHWAR2 1Research and Development

Centre in Pharmaceutical Sciences and Applied Chemistry, Poona College of Pharmacy, Bharati

Vidyapeeth University, Erandwane, Pune, India

2Department of Pharmaceutical Chemistry, RAK Medical & Health

Sciences University College of Pharmaceutical Sciences Ras Al

Khaimah, U.A.E.

SCIENTIFIC PAPER

UDC 543.544.5:54:615.252:61

DOI 10.2298/CICEQ120607101H

DETERMINATION OF GLIBENCLAMIDE, METFORMIN HYDROCHLORIDE AND ROSIGLITAZONE MALEATE BY REVERSED PHASE LIQUID CHROMATOGRAPHIC TECHNIQUE IN TABLET DOSAGE FORM

Abstract

A simple, precise and accurate high performance liquid chromatography (HPLC) method was developed for the simultaneous estimation of metformin hydrochloride, rosiglitazone maleate, glibenclamide present in multicomponent dosage forms. Chromatography was performed on a 25 cm×4.6 mm i.d., 5-μm particle, C18 column with 78:22 (v/v) methanol:20 mM potassium dihydrogen phosphate buffer as mobile phase at a flow rate of 1.0 ml/min and UV detec-tion at 238 nm for metformin hydrochloride, rosiglitazone maleate and gliben-clamide. The total elution time was shorter than 9 min. This method was found to be precise and reproducible. The proposed method was successfully applied for the analysis of metformin hydrochloride, rosiglitazone maleate, gli-benclamide as a bulk drug and in pharmaceutical formulation without any inter-ference from the excipients.

Keywords: reverse phase high performance liquid chromatography, rosiglitazone maleate, metformin hydrochloride, glibenclamide.

Currently, the most commonly prescribed medi-cations for the treatment of non-insulin dependent type 2 diabetes mellitus are drugs such as biguani-des. For example, metformin hydrochloride, 1,1-di-methylbiguanide hydrochloride (Figure 1a), is an anti-hyperglycemic agent [1].

It improves glucose tolerance in patients with type 2 diabetes and reduces both basal and post-prandial plasma glucose [2]. Sulfonylurea gliben-clamide (Gly), 1-{4-[2-(5-chloro-2-methoxybenzami-do)ethyl]benzensulfonyl}-3-cyclohexylurea (Figure 1c), is a second generation hypoglycemic agent [3] that appears to lower blood glucose by stimulating the release of insulin from the pancreas [4-5]. Thiazoli-dinedione (TZD) derivatives such as rosiglitazone ma-leate (Rosi), chemically (±)-5-{4-[2-(N-methyl-N(2-py-ridyl)amino)ethoxy]benzyl}-2,4-dionethiozolidine (Fi-

Correspondence: S.R. Dhaneshwar, Department of Pharma-ceutical Chemistry, RAK Medical & Health Sciences University College of Pharmaceutical Sciences Ras Al Khaimah, P.O. Box 11172, U.A.E. E-mail: [email protected] Paper received: 7 June, 2012 Paper revised: 31 October, 2012 Paper accepted: 1 November, 2012

gure 1b) [6], are potent new oral antihyperglicemic agents that reduce insulin resistance in patients with type 2 diabetes by binding to peroxisome proliferator-activated receptors gamma (PPAR-γ) [7-9]. For many patients with type 2 diabetes, monotherapy with an oral anti-diabetic agent is not sufficient to reach target glycaemic goals and multiple drugs may be neces-sary to achieve adequate control [10]. The use of combination of biguanides, sulfonylureas and TZDs is commonly observed in clinical practice. This com-bination can be achieved by taking each of the drugs separately or alternatively fixed formulations have been developed. Combinations of Met, Rosi and Gly are available commercially as single tablets. Although many methods have been reported in literature for the estimation of Met [11-28], Rosi [29-37] and Gly [38- -43] individually, only a few methods are available for the simultaneous estimation of Met and Rosi [44-46], Met and Gly [47-50] and Rosi with Gly [51]. However, no analytical method has been published for the simultaneous analysis of three drugs combinations whether in pure forms or in the pharmaceutical pre-paration, which became the aim of this work. The method described is rapid, economical, precise, and

S.S. HAVELE, S.R. DHANESHWAR: DETERMINATION OF GLIBENCLAMIDE… Chem. Ind. Chem. Eng. Q. 20 (1) 39−47 (2014)

40

accurate and can be used for routine analysis of tablets. It was validated as per ICH guidelines [52].

(a)

(b)

(c)

Figure 1. Chemical structures of a) metformin hydrochloride, b) glibenclamide and c) rosiglitazone.

EXPERIMENTAL

Materials and methods

Pharmaceutical grade working standards met-formin HCl [Met] (batch no. 1997418), rosiglitazone maleate [Rosi] (batch no. 758001) and glibenclamide [Gly] (batch no. 2022198) were obtained from Ran-baxy Laboratories, Dewas, India.

The commercial tablet (brand name: Diabetrol 3D, Piramal Health Care, batch No. 20605016), label claim: 500 mg of Met, 2 mg Rosi and 5 mg of Gly per tablet) was purchased in March 2010 from a local pharmacy in Pune.

All chemicals and reagents were of HPLC grade and were purchased from Merck Chemicals, Mumbai, India.

Instrumentation

The LC system consisted of a pump (Jasco PU- -1580 intelligent LC pump) with auto injecting facility (AS-1555 sampler) programmed at 20 µl capacity per injection. The detector consisted of a UV–Vis (Jasco UV 1575) model operated at a wavelength of 238 nm. The software used was Jasco Borwin version 1.5, LC- -Net II/ADC system. The column used was HiQ Sil C18HS 250 mm×4.6 mm, 5.0 µm (Kya Technologies Corporation). Different mobile phases were tested in order to find the best conditions for separation of Met, Rosi and Gly. The mobile phase contained 78:22 (v/v) methanol:20 mM potassium dihydrogen phosphate buffer and the flow rate was maintained at 1.0 ml/min UV detection was carried out at 238 nm (Figure 2). The mobile phase and samples was filtered using a 0.45 µm membrane filter. Mobile phase was degas-sed by ultrasonic vibrations prior to use. All deter-minations were performed at ambient temperature.

Standard solutions and calibration graphs for chromatographic measurement

Met, Rosi and Gly were weighed accurately and separately transferred to 10 ml volumetric flasks. All the drugs were dissolved in HPLC-grade methanol to prepare 1000 µg/ml standard stock solutions. Calib-ration standards at five levels were prepared by appropriately weighed and mixed standard solutions in the concentration range of 50–250 µg/ml for Met and 0.4–2.0 µg/ml for Rosi and 0.6–3.0 µg/ml for Gly. Samples were made in triplicate for each concen-tration and peak areas were plotted against the cor-responding concentrations to obtain the calibration graphs.

Sample preparation

For the analysis of tablets, 20 tablets were weighed and finely ground in a mortar. The portion equivalent to 500 mg of Met, 2 mg of Rosi and 2.5 mg of Gly, was transferred in a 25 ml volumetric flask separately, 20 ml of methanol was then added, and sonication was done for 45 min with swirling. After sonication, the volume was made up to the mark with the diluent, and mixed well. The solution was filtered through Whatman filter paper (#41) then injected into the chromatographic system, and analyzed quanti-tatively. The analysis was repeated six times. The possibility of excipients interference with the analysis was examined.


41

Optimization of HPLC method

The HPLC procedure was optimized with a view to develop a simultaneous assay method for Met, Rosi and Gly. The mixed standard stock solution (200 µg/ml of Met, 0.8 µg/ml of Rosi and 1.0 µg/ml of Gly) injected in HPLC. Different ratios of methanol and potassium dihydrogen phosphate buffer at different pH and molarities were tested.

Method validation

The method was validated according to the ICH guidelines. The following validation characteristics were addressed: linearity, accuracy, precision, and specificity, limits of detection and quantitation and robustness.

Linearity and range

Calibration standards at five levels were pre-pared by appropriately weighed and mixed standard solutions. From the mixed standard stock solution (50–250 µg/ml for Met and 0.4–2.0 µg/ml for Rosi and 0.6–3.0 µg/ml for Gly). Each concentration was injected six times into the LC system keeping the injection volume constant. The peak areas were plotted against the corresponding concentrations to obtain the calibration graphs.

Precision

Method repeatability was obtained from RSD (relative standard deviation, %) by repeating the analysis six times for three concentrations in the same day for intra-assay precision. The repeatability of sample injection and measurement of peak area for active compound were expressed in terms of RSD. Intermediate precision was assessed by repeating the analysis on three different days. The repeatability and intermediate precision variation was carried out at

three different concentration levels (50, 150 and 250 µg/ml for Met, 0.4, 1.2 and 2 µg/ml for Rosi and 0.6, 1.8 and 3 µg/ml for Gly).

Limit of detection and quantification

In order to estimate the limit of detection (LOD) and limit of quantification (LOQ), blank methanol was injected 6 times following the method as described in the instrumentation Section. The signal-to-noise (S/N) ratio was specified as 3:1 for LOD and 10:1 for LOQ. The LOD and LOQ were experimentally verified by diluting known concentrations of standard solutions of Met, Rosi and Gly until the average responses were approximately 3 or 10 times the standard deviation (SD) of the responses for 6 replicate determinations.

Robustness

The robustness was studied by evaluating the effect of small but deliberate variations in the chro-matographic conditions. The conditions studied were flow rate (altered by ±0.1 ml/min), mobile phase com-position (methanol ±2 ml). These chromatographic variations were evaluated for resolution between Met, Rosi and Gly.

Solution stability

To assess the solution stability, three different concentrations of (2, 4 and 6 µg/ml) were prepared from sample solutions and kept at room temperature for 8 days. These solutions were compared with freshly prepared standard solutions.

System suitability

The system suitability parameters with respect to theoretical plates, tailing factor, repeatability and resolution between Met, Rosi and Gly peaks were defined.

Figure 2. Overlaid UV spectra of metformin HCl rosiglitazone maleate and glibenclamide measured from 200 to 400 nm.


42

Specificity

Extracts of commonly used placebos were injected to demonstrate the absence of interference with the elution of the Met, Rosi, and Gly. For deter-mining selectivity of the method, a powder blend of typical tablet excipients containing lactose mono-hydrate, mannitol, maize starch, povidone K30, citric acid anhydrous granular, sodium citrate, natural lemon and lime flavor and magnesium stearate was prepared and analyzed. All chromatograms were examined to determine if the compounds of interest co-eluted with each other or with any additional excipients peaks.

Accuracy

Accuracy of the method was carried out by applying the method to drug sample to which known amounts of Met, Rosi and Gly standard powder cor-responding to 80, 100 and 120% of label claim had been added (standard addition method), mixed, and the powder was extracted and analyzed by running chromatograms in optimized mobile phase. These mixtures were analyzed by the proposed method. The experiment was performed in triplicate and recovery (%) and RSD were calculated.

Analysis of marketed formulation

The marketed formulation was assayed as des-cribed above. The peak areas were measured at 238 nm and concentrations in the samples were deter-mined using multilevel calibration developed on the same LC system under the same conditions, and analyzed using linear regression as described earlier.


Method development and optimization

The HPLC procedure was optimized with the aim of developing a suitable LC method for the anal-ysis of Met, Rosi and Gly in fixed dose combined dosage form. Initially, methanol and water in different ratios were tried. However, a broad peak shape was obtained for Met, so water was replaced by potassium dihydrogen phosphate buffer (20 mM), and mixture of methanol and potassium dihydrogen phosphate buffer in different ratios were tried. It was found that methanol:potassium dihydrogen phosphate buffer (20 mM) at a ratio of 78:22, v/v, resulted in acceptable retention time (tR 2.51 min for Met, 3.90 min for Rosi and 8.12 min for Gly), plates, and good resolution for Met, Rosi and Gly at a flow rate of 1.0 ml/min (Figure 3).

Figure 3. Chromatogram of metformin hydrochloride (200 μg/ml), tR: 2.35 min; rosiglitazone maleate (0.8 μg/ml), tR: 3.90 min; glibenclamide (1 μg/ml), tR: 8.12 min; measured at 238 nm, mobile phase: methanol/potassium dihydrogen phosphate buffer

(20 mM) (78/22, v/v).


43

Validation

Linearity

Linearity was evaluated by analysis of working standard solutions of Met, Rosi and Gly of five diffe-rent concentrations. The range of linearity was from

50–250 µg/ml for Met, 0.4–2.0 µg/ml for Rosi and 0.6– -3.0 µg/ml for Gly. The regression data obtained are represented by calibration curves (Figures 4–6). The results showed that within the concentration range mentioned above, there was an excellent correlation between peak area and concentration of each drug

Figure 4. Calibration curve of Met.

Figure 5. Calibration curve of Rosi.

Figure 6. Calibration curve of Gly.

S.S. HAVELE, S.R. DHANESHWAR: DETERMINATION OF GLIBENCLAMIDE… CI&CEQ 20 (0) 000−000 (2014)

44

Precision

The results of the intra-day and inter-day pre-cision experiments are given in Table 1. The deve-loped method was found to be precise, with RSD values for intra-day and inter-day precision < 2%, as recommended by ICH guidelines. Separation of the drugs was found to be similar when analysis was performed on different chromatographic systems on different days, as shown in Table 1.

LOD and LOQ

The LOD and LOQ values were found to be 0.02 and 0.06 µg/ml for Met, 0.01 and 0.032 µg/ml for Rosi and 0.003 and 0.01 µg/ml for Gly (Table 2).

Table 2. Limit of detection and quantitation

Compound LOD LOQ

RSD / %

Met 0.009 0.72

Rosi 0.55 0.28

Gly 0.12 0.43

Specificity

Extracts of commonly used placebos were injected to demonstrate the absence of interference with the elution of the drugs. The results demon-strated that there was no interference from other materials in the tablet formulation, thereby confirming the specificity of the method (Fig. 7).

System suitability

System suitability parameters such as the num-ber of theoretical plates, HETP and peak tailing were determined. The obtained results are shown in Table 3.

Robustness of the method

To ensure the insensitivity of the developed HPLC method to minor changes in the experimental conditions, it is important to demonstrate its robust-

ness. None of the alterations caused a significant change in resolution between Met, Rosi and Gly, peak area, RSD, %, tailing factor and theoretical plates (Table 4).

Table 3. Statistical analysis of parameters required for system suitability testing of the proposed HPLC method

Parameter Met Rosi Gly

Theoretical plates 35547.65 3563 3213.85

Resolution - 3.01 11.99

Peak asymmetry 1.31 1.35 1.25

RSD / % 0.32 0.09 0.04

Solution stability studies

Three different concentrations 2, 4 and 6 µg/ml were prepared from sample solution and stored at room temperature for 8 days. They were then injected into the HPLC system and no additional peaks were found in the chromatogram, indicating the stability of Met, Rosi and Gly in the solution (Table 5).

Table 1. Repeatability and intermediate precision; n = 6, average of 50, 150 and 250 µg/ml for Met; 0.4, 1.2 and 2 µg/ml for Rosi and 0.6, 1.8 and 3 µg/ml for Gly

Compound Intraday Interday

Mean % of assay RSD / % Mean % of assay RSD / %

Met 100.89 1.14 99.89 1.09

Rosi 100.71 1.03 99.54 0.37

Gly 100.03 0.51 100.53 0.29

Figure 7. Representative chromatogram obtained for the commonly used excipients.


45

Table 5. Stability of drugs in sample solutions; n = 6, average of (50, 150 and 250 µg/ml for Met, 0.4, 1.2 and 2 µg/ml for Rosi; 50, 150 and 250 µg/ml for Met and 0.6, 1.8 and 3 µg/ml for Gly)


RSD / % 0.74 0.04 1.10

Recovery studies

Good recoveries of Met, Rosi and Gly were obtained at various added concentrations for the tab-lets (Table 6).

Analysis of a commercial formulation

Experimental results of the amount of Met, Rosi and Gly in tablets, expressed as a percentage of label claims were in good agreement with the label claims, thereby suggesting that there is no interference from any of the excipients that are normally present in tablets. Fixed dose combination tablets were analyzed using the proposed procedures (Table 7).

The summary of validation parameters is listed in Table 8.

Table 8. Summary of validation parameters


Linearity range, µg/ml 50-100 0.4-2 0.6-3.0

Correlation coefficient 0.9997 0.9997 0.9999

Limit of detection, µg/ml 0.02 0.01 0.003

Limit of quantitation, µg/ml 0.06 0.03 0.01

Recovery (n = 6) 100.19 99.50 99.08

Precision (RSD / %)

Intraday

Interday

1.14

1.09

1. 03

0.37

0.51

0.29

Robustness Robust Robust Robust

CONCLUSION

The new HPLC method described in this paper provides a simple, convenient and reproducible approach for the simultaneous identification and quantification that can be used to determine metformin hydrochloride, rosiglitazone maleate, glyburide in routine quality control.

Table 6. Recovery studies (n = 6)

Label claim Amount of drug added, % Total amount, mg Amount recovered, mg Recovery, %

Met

500 mg

80 900 899.28 99.92

100 1000 1008.3 100.83

120 1100 1098.02 99.82

Rosi

2 mg

80 3.6 3.601 100.05

100 4 3.97 99.38

120 4.4 4.35 99.08

Gly

2.5 mg

80 4.5 4.45 98.91

100 5 4.96 99.32

120 5.5 5.44 99.03

Table 7. Applicability of the HPLC method for the analysis of the pharmaceutical formulations

Sample Label claim, mg Drug content, % RSD / %

Met 500 100.39 0.08

Rosi 2 99.38 0.28

Gly 2.5 99.88 0.14

Table 4. Robustness testing (n = 6)

Chromatographic factor Level Retention time, tR / min Resolution (Rs) Asymmetry (As)

– Met Rosi Gly Met Rosi Gly Met Rosi Gly

Flow rate, ml/min 0.9 2.30 3.78 7.98 0 6.01 2.39 1.13 1.46 1.38

1.0 2.35 3.90 8.12 0 3.01 11.99 1.31 1.35 1.25

1.1 2.39 3.70 4.00 0 6.49 2.48 1.01 1.30 1.25

Methanol content, % 76 2.40 3.99 8.50 0 3.70 12.06 1.70 1.66 1.40

78 2.35 3.90 8.12 0 3.01 11.99 1.31 1.35 1.25

80 2.30 3.10 4.08 0 6.09 2.01 1.30 1.34 1.24


46

Nomenclature

Met: Metformin hydrochloride Rosi: Rosiglitazone maleate Gly: Glibenclamide ICH: International Conference on Harmonisation HPLC: High Performance Liquid chromatography

REFERENCES

[1] British Pharmacopoeia, Here Majesty Queen office, London, 2002, p. 1123

[2] M. Garber, Endocrine Practice 3 (1997) 359-370

[3] Martindale – The Complete Drug Reference, Pharmaceu-tical Press, London, 2002

[4] The Merck Index, 13th ed., Merck, Whitehouse Station, NJ, 2001, p. 1335

[5] R. Padwal, S. R. Majumdar, J. A. Johnson, J. Varney, F. A. McAlister, Diabetes Care 28 (2005) 736–744

[6] C.J. Tack, P. Smits, Neth. J. Med. 55 (1999) 209–211

[7] The Martindale Extra Pharmacopoeia 32nd ed., Royal Pharmaceutical Society, London, 1999, p. 331

[8] E.A. Gale, Lancet 357 (2001) 1870–1875

[9] J.M. Malinowski, S. Bolesta, Clin. Ther. 22 (2000) 1151– -1168

[10] J.A. Balfour, G.L. Plosker, Drugs 57 (1999) 921–930

[11] H.N. Mistri, A.G. Jangid, P.S. Shrivastav, J. Pharm. Biomed. Anal. 45 (2007) 97–106

[12] M.A.S. Marques, A. de S. Soares, O.W. Pinto, P.T.W. Barroso, D.P. Pinto, M. Ferreira-Filho, E. Werneck-Barroso, J. Chromatogr., B 852 (2007) 308–316

[13] C.G. Ding, Z. Zhou, Q.H. Ge, X.J. Zhi, L.L. Ma, Biomed. Chromatogr. 21 (2007) 132–138

[14] N. Koseki, H. Kawashita, M. Niina, Y. Nagae, N. Masuda, J. Pharm. Biomed. Anal. 36 (2005) 1063–1072

[15] Y. Wang, Y. Tang, J. Gu, J.P. Fawcett, X. Bai, J. Chromatogr., B 808 (2004) 215–219

[16] K. Heinig, F. Bucheli, J Pharm. Biomed. Anal. 34 (2004) 1005–1011

[17] R.T. Sane, V.J. Banavalikar, V.R. Bhate, V.G. Nayak, Indian Drugs 26 (1989) 47-648

[18] M.A. Hegazy, M.R. El-Ghobashy, A.M. Yehia, and A.A. Mostafa, Drug Test Anal. 1 (2009) 339–349

[19] S.Z. El-Khateeb, H.N. Assaad, M.G. El-Bardicy, A.S. Ahmad, Analytica Chimica Acta 208 (1988) 321–324

[20] M. Kar, P.K. Choudhury, Indian J. Pharm. Sci. 71 (2009) 318–320

[21] S. Havele, S. Dhaneshwar, WebmedCentral Pharmaceutical Sci. 1(10) (2010) WMC001078

[22] S. Havele, S. Dhaneshwar, J. Nanomedic. Nanotechnol. 1 (2010) 202

[23] S.S. Havele, S.R. Dhaneshwar, J. Liq. Chromatogr. Rel. Technol. 34 (2011) 966–980

[24] S.S.M. Hassan, W.H. Mahmoud, M.A.F. Elmosallamy, A.H.M. Othman, Anal. Chim. Acta 378 (1999) 299–311

[25] M.S. Arayne, N. Sultana, M.H. Zuberi, Pak. J. Pharm. Sci. 19 (2006)231–235

[26] I. Ali, H.Y. Aboul-Enein, V.K. Gupta, Comb. Chem. High Throughput Screen. 10 (2007) 611–615

[27] I. Hamdan, A.K.B. Jaber, A.M. Abushoffa, J. Pharm. Biomed. Anal. 53 (2010)1254–1257

[28] H.I. Habib, M.S. Kamel, Talanta 60 (2003)185–190

[29] M.W. Hruska, R.F. Frye, J. Chromatogr., B 803 (2004) 317–320

[30] R.S. Pedersen, K. Brøsen, F. Nielsen, Chromatographia 62 (2005)197–201

[31] P. Gomes, J. Sippel, A. Jablonski, M. Steppe, J. Pharm. Biomed. Anal. 36 (2004) 909–913

[32] P. Gomes, M. Steppe, J. AOAC Int. 89 (2006) 1296–1299.

[33] S. Gayathri, A. Shantha, V. Vaidyalingam, Indian J. Pharm. Sci. 65 (2003) 663–665

[34] R.T. Sane, M. Francis, A. Moghe, S. Khedkar, A. Anerao, J. Planar Chromatogr 15 (2002) 192–195

[35] R.N.V.S. Mamidi, B. Benjamin, M. Ramesh, N.R. Srinivas, Biomed. Chromatogr. 17 (2003) 417–420

[36] J. He, Y.F. Hu, L.F. Duan, Z.R. Tan, L.S. Wang, D. Wang, W. Zhang, Z. Li, J. Liu, J.H. Tu, Y.M. Yao, H.-H. Zhou, J. Pharm. Biomed. Anal. 43 (2007) 580–585

[37] T. Radhakrishna, J. Satyanarayana, A. Satyanarayana, J. Pharm. Biomed. Anal. 29 (2002) 873–880

[38] W.J. Adams, D.S. Krueger, J. Pharm. Sci. 68 (1979) 1138–1140

[39] H. Emilsson, S. Sjöberg, M. Svedner, I. Christenson, J. Chromatogr. 383 (1986) 93–102

[40] S. Othman, O. Shaheen, I. Jalal, A. Awidi, W. Al-Turk, J. Assoc. Off. Anal. Chem. 71 (1988) 942–944

[41] S.Havele, S. Dhaneshwar, Phar. Lettre 2 (2010) 440-446.

[42] T. Rydberg, E. Wåhlin-Boll, A. Melander, J. Chromatogr. 564 (1991) 223–233

[43] S. Hsieh, K. Selinger, J. Chromatogr., B 772 (2002) 347– –356

[44] L. Zhang, Y. Tian, Z. Zhang, Y. Chen, J. Chromatogr., B 854 (2007) 91–98

[45] M. Vasudevan, J. Ravi, S. Ravisankar, B. Suresh, J. Pharm. Biomed. Anal. 25 (2001) 77–84

[46] A.R. Ali, I.I. Duraidi, M.M. Saket, E.S.M. Abu-Nameh, J. AOAC Int. 92 (2009) 119–124

[47] D.H. Khanolkar, V.M. Shinde, Indian Drugs 36 (1999) 739–742

[48] H.N. Mistri, A.G. Jangid, P.S. Shrivastav, J. Pharm. Biomed. Anal. 45 (2007) 97–106

[49] P.K. Chaturvedi, R. Sharma, Acta Chromatogr. 20 (2008) 451–461

[50] S. Aburuz, J. Millership, J. McElnay, J. Chromatogr., B 817 (2005) 277–286

[51] P. Venkatesh, T. Harisudhan, H. Choudhury, R. Mullangi, N. R. Srinivas, Biomed. Chromatogr. 20 (2006) 1043– –1048

[52] International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, Validation of Analytical Procedures Q2 (R1) (1994) Incorporated (2007).


47

SHWETA S. HAVELE1

SUNIL R. DHANESHWAR2

1Research and Development Centre in Pharmaceutical Sciences and Applied

Chemistry, Poona College of Pharmacy, Bharati Vidyapeeth

University, Erandwane, Pune, India 2Department of Pharmaceutical

Chemistry, RAK Medical & Health Sciences University College of

Pharmaceutical Sciences Ras Al Khaimah, U.A.E.

NAUČNI RAD

ODREĐIVANJE GLIBENKLAMIDA, METFORIN- -HIDROHLORIDA I ROSIGLITAZON-MALEATA U TABLETAMA REVERSNO-FAZNOM HROMATOGRAFIJOM POD VISOKIM PRITISKOM

Razvijena je jednostavna, precizna i tačna HPLC metoda za simultano određivanje

metformin-hidrohlorida, rosiglitazone-maleata i glibenklamida u multikomponentnim lekov-

itim preparatima. Određivanje je izvedeno na C18 koloni dimenzija 25 cm×4.6 mm i.d., i

veličina čestica 5 μm. Kao mobilna faza korišćen je rastvarač metanol:kalijum-dihidro-

genfosfat (20 mM) 78:22 (v/v) sa protokom 1,0 ml/min. Metformin-hidrohlorid, rosiglitazon-

-maleat I glibenklamid su detektovani na 238 nm. Ukupno vreme analize je kraće od 9 min.

Metoda je precizna i reproduktivna i uspešno je primenjena za određivanje metformin

hidrohlorida, osiglitazone osigli I glibenklamida u aktivnim supstancama I farmacetskim

formulacijama bez interferencije sa dodatim ekscipijensima.

Ključne reči: RP-HPLC, rosiglitazon-maleat, metformin-hidrohlorid, glibenklamid.





49

MARIA VALDEREZ PONTE ROCHA1

JOCÉLIA SOUSA MENDES1

MARIA ESTELA APARECIDA GIRO1

VÂNIA MARIA M. MELO2

LUCIANA ROCHA BARROS GONÇALVES1

1Universidade Federal do Ceará – Depto. de Engenharia Química -

Bloco 709, Campus do Pici – CEP. 60.455-760, Fortaleza – CE, Brasil 2Universidade Federal do Ceará –

Depto. de Biologia - Laboratório de Ecologia Microbiana e

Biotecnologia (LemBiotech) - Bloco 909, Campus do Pici – CEP.

60.455-760, Fortaleza – CE, Brasil

SCIENTIFIC PAPER

UDC 579.841.1:60:66

DOI 10.2298/CICEQ120518100R

BIOSURFACTANT PRODUCTION BY Pseudomonas aeruginosa MSIC02 IN CASHEW APPLE JUICE USING A 24 FULL FACTORIAL EXPERIMENTAL DESIGN

Abstract

In this work, the production of biosurfactants from cashew apple juice by Pseudomonas aeruginosa MSIC02 was investigated by carrying out a 24 full factorial experimental design, using temperature, glucose concentration from cashew apple juice, phosphorous concentration and cultivation time as vari-ables. The response variable was the percentage of reduction in surface ten-sion in the cell-free culture medium, since it indicates the surface-active agent production. Maximum biosurfactant production, equivalent to a 58% reduction in surface tension, was obtained at 37 °C, with glucose concentration of 5.0 g/L and no phosphorous supplementation. Surface tension reduction was signi-ficant, since low values were observed in the cell-free medium (27.50 dyn/cm), indicating that biosurfactant was produced. The biosurfactant emulsified diffe-rent hydrophobic sources and showed stability in the face of salinity, exposure to high temperatures and extreme pH conditions. These physiochemical pro-perties demonstrate the potential for using biosurfactants produced by P. aeru-ginosa MSIC02 in various applications.

Keywords: biosurfactant, surface tension, Pseudomonas aeruginosa, full factorial experimental design, cashew apple juice, enhanced oil recovery.

Surfactants constitute a very important class of chemical compounds widely used in a variety of industrial sectors, because they act as dispersants and/or solubilizing agents of organic compounds. Most of the commercially used surfactants are syn-thesized from petroleum derivates [1]. Biosurfactants are complex biological molecules, which display pro-perties similar to those of the well-known synthetic surfactants, and have been reported as being pro-duced both on the microbial cell surface and excreted extracellularly [2]. They include microbial compounds that exhibit surfactant properties, e.g., polysaccha-ride-protein complexes, lipopeptides, fatty acids, gly-colipids, phospholipids and neutral lipids [3,4].

The most important advantage of biosurfactants over chemical surfactants is probably their ecological

Correspondence: L.R.B. Gonçalves, Departamento de Engen-haria Química, Universidade Federal do Ceará, Campus do Pici, Bloco 709, Fortaleza, CE, Brasil, Zip code: 60455-760. E-mail: [email protected], [email protected] Paper received: 18 May, 2012 Paper revised: 31 October, 2012 Paper accepted: 1 November, 2012

sustainability. Biosurfactants are biodegradable and thus, problems of toxicity and accumulation in natural ecosystems are avoided. They have received more attention lately due to their low toxicity, biodegra-dability and effectiveness in improving the solubility and biodegradation of hydrophobic compounds [2]. Some potential applications include bioremediation of water-insoluble pollutants, enhanced oil recovery and use in health care and crude oil drilling industries [5,6]. Other potential applications can be found in agriculture, cosmetics, pharmaceuticals, detergents, food processing industries, among others [7,8]. Gly-colipids, consisting of hydrophilic carbohydrate and long chain aliphatic acids or hydroxyl-aliphatic acids, are the most common class of the microbially-pro-duced surface active compounds [9]. Of this class, rhamnolipids possess surfactant, antibacterial and antiviral properties [10], as well as spreading abilities, particularly in alkanes.

Rhamnolipids, a glycolipid surfactant containing one or two molecules of rhamnose and 3-hydroxy fatty acids, are produced by bacteria of the genus Pseudomonas. Their properties depend on the bacte-

M.V.P. ROCHA et al.: BIOSURFACTANT PRODUCTION BY P. aeruginosa… Chem. Ind. Chem. Eng. Q. 20 (1) 49−58 (2014)

50

rial strain, culture conditions and composition of the culture medium [11]. Furthermore, rhamnolipids can be produced using hydrophilic and hydrophobic sub-strates [12].

Relatively high production costs prevent biosurf-actants from being widely employed in the industry. The use of alternative substrates, such as agro-industrial wastes, may be a viable strategy for redu-cing costs [13]. However, the formulation of culture medium, i.e. determining which waste has the proper nutrients for cell growth and accumulation of the desired product, is a challenge [14]. According to the literature [15], biosurfactant production is controlled by different operating parameters, which should be maintained within a certain range of operational con-ditions in order to achieve maximum production of biosurfactant. To achieve this goal, the authors sug-gest the use of a 24 full factorial experimental design.

On the north coast of Brazil, cashew apple juice (CAJ) occurs as a residue of the cashew agro-industry, since less than 20% of the total cashew apple (90% of the fruit) is processed by the local industry (beverages and deserts) [16–18]. Further-more, most of the cashew apple production is left to rot in the soil. These facts, together with its rich com-position, make cashew apple juice an interesting and inexpensive (R$ 1.00/kg) culture medium. However, CAJ characterization indicated that supplementing the juice with essential nutrients is required [19]. Therefore, in this work, nutritional and cultural factors affecting biosurfactant production, in flask-scale, by a new strain of Pseudomonas aeruginosa were inves-tigated by using a 24 full factorial experimental design. Furthermore, the stability of the biosurfactant relative to some environmental stress conditions (pH, tempe-rature and salinity) was also evaluated.

EXPERIMENTAL

Microorganism

The Pseudomonas aeruginosa MSIC02 strain used in this study was isolated from an oil spill off the coast of Ceará. Its rRNA 16S sequence is deposited in the Genbank with the following access number: FJ876297. The culture was maintained on nutrient agar (Biolife) slants at 4 °C and sub-cultured at regu-lar time.

Preparation and characterization of substrate

The physiochemical composition of cashew apple varies widely, depending on the variety, matu-ration, size, duration of the post-harvesting period, and regional environmental variations [20]. Thus, in

this work, all of the cashew apple juice used was standardized and characterized. In order to obtain consistent results, the same batch of juice was used in all experiments.

Cashew apple juice was extracted by compres-sing the cashew apple (Anarcardium occidentale L.). Since fruits of various origins and maturations were used, after compressing, the juice was mixed and centrifuged at 3500 rpm for 20 min (BIO ENG, BE-6000), filtered and stored at –18 °C. The pH of the cashew apple juice was determined using a Tecnal potentiometer (model Tec-3MP) at approximately 27 °C.

Inoculum preparation and biosurfactant production

The bacterial strains were streaked on a nutrient agar (5.0 g/L of peptone, 3.0 g/L of yeast extract and 15 g/L of agar) slant and incubated for 24 h at 30 °C. Three loops of culture were inoculated in 50 mL of nutrient broth (5.0 g/L of peptone and 3.0 g/L of yeast extract) in a 250 mL Erlenmeyer flask and incubated in a rotary shaker (Tecnal – TE240, BR) at 30 °C and 150 rpm for 18–24 h. Afterwards, the optical density (600 nm) of bacterial suspension was adjusted to 0.5 and an aliquot of 1 mL of inoculum (2%, v/v) was transferred to a 250 mL Erlenmeyer flask, containing 50 mL of medium (CAJ), and incubated at 30 °C, 150 rpm in a rotary shaker (Tecnal – TE240, BR). The initial pH of CAJ medium was adjusted to 7.0 with NaOH 1 mol/L. Samples were collected at time-defined intervals and submitted to analysis. During the assays, NaNO3 was used as an inorganic nitrogen source, in a concentration of 5 g/L. Nutrient broth was sterilized at 121 °C for 15 min. The media containing cashew apple juice (CAJ) were sterilized by filtration using a sterile membrane (Millipore cellulose ester) with a pore diameter of 0.45 µm.

Effect of variables on surface tension

The production of biosurfactants by P. aerugi-nosa MSIC02 in cashew apple juice was carried out by 24 full factorial experimental design, using tem-perature (X1), glucose concentration from cashew apple juice (X2), phosphorous concentration through KH2PO4 (X3) supplementation and cultivation time (X4) as variables. Cashew apple juice was diluted with water in order to achieve the desired glucose concen-tration. The percentage of reduction in surface ten-sion was adopted as the response variable, since it indicates the surface-active agent production. Accord-ing to the literature [21–23], the measurement of sur-face tension may be used to detect biosurfactant production and most of the other methods that measure the surface properties of biosurfactant use surface tension reduction as the standard. A 24 full


51

factorial experimental design was used, with three central points to determine error, made up of 19 expe-riments (Table 1). The factors studied were select based on the literature [4,24–28]. The minimum and maximum ranges of the variables investigated and the full experimental plan with respect to their values are listed in Table 1. The data were statistically analyzed using Statistica 6.0 (Statsoft Inc., USA) via multiple regression analysis using the least squares method, taking into account the isolated terms and interaction of the studied variables.

The results of the experiments were analyzed in order to determine the Pareto chart, the significant variables, the effect, and the intensity of the stationary point, i.e., with the aim of evaluating the existence of a maximum or minimum point.

Determination of the effect of environmental factors on biosurfactant activity

Stability studies were conducted using cell-free broth obtained after 48 h of cultivation [25,29,30]. The pH stability was performed by adjusting the broth to different pH values (2.0–11.0), using 1.0 mol/L NaOH or HCl depending on the desired value. To study the effect of salinity on the stability of biosurfactant, diffe-rent concentrations of NaCl (0.0–20%, w/v) were added to broth samples and mixed until completely dissolved. Broth samples were heated in a boiling water bath at constant temperature (4, 15, 30, 50 and

100 °C) and in autoclave (121 °C) for 15 min and cooled at room temperature. Samples were also heated in a boiling water bath (100 °C) for different time intervals (0, 15, 30, 60 and 75 min) and cooled at room temperature. After each treatment, the surface tension values were determined.

Analytical methods

Biomass content. Cell growth was determined by measuring the optical density of samples, using a UV-visible spectrophotometer (20 Genesis, BR) at 600 nm. Cell concentration was determined by dry weight by filtering through a 0.45 μm previously weighted Millipore membrane [31].

Carbohydrate concentration. Substrate concen-tration (glucose and fructose), present in CAJ, was measured by HPLC using a Waters high-perfor-mance-liquid chromatograph equipped with a refrac-tive index detector and a Shodex Sugar SC1011 column (8.0 mm×300 mm). Milli-Q water was used as a solvent with a flow rate of 0.6 mL/min at 80 °C. The samples were identified by comparing the retention times with those of carbohydrate standards [19].

Surface tension determination. Surface tension was determined with a Tensiometer (Krüss) at 30 °C, according to the De Nöuy ring method. The deter-mination of surface tension was replicated and it was performed using cell-free supernatants obtained after centrifugation.

Table 1. Results of experimental using four independent variables and three center points showing observed values for biosurfactantproduction by P. aeruginosa MSIC02: X1 - temperature, X2 – glucose concentration, X3 – phosphorous concentration, X4 – cultivation time, R1 – surface tension and R2 – surface tension reduction

Assay X1 / °C X2 / g L–1 X3 / g L–1 X4 / h R1 / dyn cm–1 R2 / %

1 30 5 0 24 31.00±0.0 40.10

2 37 5 0 24 28.00±0.0 47.00

3 30 10 0 24 34.00±0.71 24.44

4 37 10 0 24 28.50±0.0 56.55

5 30 5 1 24 29.50±0.0 51.73

6 37 5 1 24 28.50±0.0 51.07

7 30 10 1 24 38.25±0.49 15.47

8 37 10 1 24 31.10±0.41 30.58

9 30 5 0 72 27.50±0.0 46.86

10 37 5 0 72 27.50±0.0 47.95

11 30 10 0 72 31.00±0.0 41.87

12 37 10 0 72 27.50±0.0 58.08

13 30 5 1 72 30.00±0.0 50.92

14 37 5 1 72 31.00±0.0 46.78

15 30 10 1 72 37.00±0.0 18.23

16 37 10 1 72 28.50±0.0 36.38

17 (C) 33.5 7.5 0.5 48 28.63±0.25 17.21

18 (C) 33.5 7.5 0.5 48 29.00±0.0 18.40

19 (C) 33.5 7.5 0.5 48 29.00±0.0 16.98


52

Emulsifying activity. Emulsifying activity against gasoline, diesel, hexane, kerosene and soy oil was determined according to Rocha et al. [19] and reported as (E24).


All the experiments were carried out with three independent replicates. In order to verify significant differences, results were evaluated statistically, at 95% confidence level (p < 0.05), using the Analysis of variance (ANOVA) and Tukey multiple comparison tests, available in Microcal Origin 8.1 software (Mic-rocal Software Inc., Northampton, MA, USA).


Biosurfactant production by a 24 full factorial experimental design

In a previous work [32], the ability of biosurfac-tant production by Pseudomonas aeruginosa ATCC 10145 in batch cultivation using cashew apple juice (CAJ) and mineral media was evaluated. This initial study indicated that traditional carbon sources for biosurfactant production could be replaced by CAJ, supplemented with peptone. Since the production of biosurfactant on CAJ was demonstrated, further stu-dies were conducted in order to enhance biosurfac-tant production. For that purpose, and due to the complex nature of biological processes [15], a 24 full factorial experimental design was utilized to investi-gate the reduction in surface tension (biosurfactant production) and to determine the significance of pro-cess parameters and their interactions. Table 1 shows the results obtained in the 24 full factorial design from the studied variables: temperature (X1), glucose concentration (X2), phosphorous concen-tration (X3) and cultivation time (X4), using the isolated

P. aeruginosa MSIC02 strain. Table 1 also shows the surface tension values for the media after fermen-tation for each experiment.

The results shown in Table 1 indicate that maxi-mum biosurfactant synthesis was obtained in the ninth, tenth and twelfth experiments. A significant reduction in surface tension was observed, once tension of the cell-free medium reached 27.50 dyn/cm. Based on the results, one can conclude that CAJ is also a suitable substrate for the production of biosurfactants by P. aeruginosa MSIC02, since the surface tension of the cell-free broth was reduced to below 30 dyn/cm [11,33].

Table 2 shows a regression analysis of the esti-mates and hypothesis tests for the coefficients of regression. The determination of the significant para-meters was performed through a hypothesis test (Stu-dent’s t-test) with a 5% level of significance.

The results of this analysis were also shown using the Pareto chart (Figure 1) and they very clearly present the most significant effects. According to Garrido-Lopez and Tena [34], the length of the bars is proportional to the absolute value of the estimated effects. In this work, the dashed line represents 95% of confidence interval. Effects that cross this line are significant values with respect to the response. In the Pareto chart for surface tension reduction, the signi-ficant effect is due to glucose concentration, followed by temperature. Enhancing glucose concentration negatively affects biosurfactant production, while enhancing temperature positively affects biosurfac-tant production. The negative influence of glucose concentration on biosurfactant production may be explained by the effect caused by inhibitors, such as tannin, which are present in cashew apple juice. The interactions between glucose concentration and phos-phorous concentration (X2X3), and glucose concen-

Table 2. Results of regression analysis using 24 full factorial design

Parameter Effect Pure error t(2) p Coefficient Standard error Level of confiance (95%)

Inferior limit Superior limit

Mean/interc. 37.7158 0.1748 215.6996 0.00002 37.71579 0.1748 215.6996 0.00002

(X1) Temperature 10.5963 0.3811 27.8056 0.00129 5.29813 0.1905 27.8056 0.00129

(X2) Glucose -12.6013 0.3811 -33.0669 0.00091 -6.30063 0.1905 -33.0669 0.00091

(X3) Phosphorus -7.7113 0.3811 -20.2350 0.00243 -3.85563 0.1905 -20.2350 0.00243

(X4) Time 3.7663 0.3811 9.8830 0.01008 1.88312 0.1905 9.8830 0.01008

X1X2 9.7988 0.3811 25.7128 0.00151 4.89938 0.1905 25.7128 0.00151

X1X3 -3.4813 0.3811 -9.1351 0.01177 -1.74063 0.1905 -9.1351 0.01177

X1X4 -2.7688 0.3811 -7.2655 0.01842 -1.38438 0.1905 -7.2655 0.01842

X2X3 -12.3588 0.3811 -32.4305 0.00095 -6.17938 0.1905 -32.4305 0.00095

X2X4 3.1138 0.3811 8.1708 0.01465 1.55687 0.1905 8.1708 0.01465

X3X4 -2.9013 0.3811 -7.6132 0.01682 -1.45063 0.1905 -7.6132 0.01682


53

tration and temperature (X1X2), negatively affected biosurfactant production.

Maximum production of biosurfactants was observed, with a 58% reduction in surface tension, at the minimum level (–1) of glucose concentration (5.0 g/L) and the maximum level (+1) of temperature (37 °C) when phosphorous concentration and cultivation time are fixed at their respective center point values of 0.5 g/L and 48 h.

Reduction in surface tension increases with temperature in the range studied, while increasing the glucose concentration lead to a decrease in the values of “reduced surface tension”. Therefore, within the experimental data, it can be observed that when there is an increase in glucose concentration from cashew apple juice, a decrease in biosurfactant pro-duction occurs. The opposite was observed with tem-perature, in which a greater decrease in surface ten-sion occurred when the temperature rose from 30 to 37 °C. However, there is not enough significant evi-dence of the interaction of these two variables with the other two variables in the study (phosphorous concentration and cultivation time). Wei et al. [26] also observed that rhamnolipid production by an indigenous P. aeruginosa J4 was affected by tempe-rature and agitation rate, being 30 °C and 200 rpm the best conditions for rhamnolipid production. Other authors [35] determined the optimum values of tem-perature (35 °C) for a rhamnolipid biosurfactant-producing strain (Pseudomonas aeruginosa F-2), which is in the range studied in this work for P. aeruginosa MSIC02.

According to data in the literature [28], phos-phate metabolism can influence biosurfactant pro-

duction, and inorganic phosphate is important for the capacity buffer. Other authors [24,36] mentioned that the carbon/phosphorous (C/P) ratio is important in the production of rhamnolipids by P. aeruginosa. Mulligan et al. [28] studied the influence of phosphate meta-bolism on biosurfactant production by P. aeruginosa ATCC 9027 using glucose as carbon source and they concluded that a shift in phosphate metabolism coin-cided with biosurfactant production. However, in the present work, it was observed that the used phos-phorous concentration did not significantly influence biosurfactant production.

It was also observed that cultivation time had no expressive influence on biosurfactant production. Low values of surface tension of the cell-free broth were achieved after 24 h of fermentation, independent of the experiment. Therefore, in that time period, the concentration of biosurfactant in the medium was already above critical micelle concentration (CMC) and an increase in the concentration of biosurfactant that was produced would not reduce the surface ten-sion any further. This fact can explain the lack of influence of cultivation time on biosurfactant pro-duction. Some authors [37] report the same behavior and explain that even in the presence of a small con-centration of biosurfactants, CMC may be achieved, from which no variation in the surface tension can be observed.

Kinetics of biosurfactant production

The kinetics of growth and the production of biosurfactants by P. aeruginosa MSIC02 at 37 °C and 150 rpm in an optimized cashew apple juice medium were studied. Several process parameters were

Figure 1. Biosurfactant production by P. aeruginosa MSIC02 using CAJ: Pareto chart of standardized effects.


54

monitored during the fermentation period and the results are presented in Figure 2. It can be observed that the variation in biomass concentration (ln(X/X0)) is a typical microbial growth curve. A comparison between cell growth, surface tension and substrate uptake allows one to observe that biosurfactant pro-duction coincides with the extinguishing of the sub-strate and the beginning of the stationary phase (Figure 2). According to the literature [38], several biosurfactants were recognized as secondary meta-bolites, while others were considered growth asso-ciated. In this study, the observed behavior is typical of secondary metabolites.

The consumption pattern of glucose and fruc-tose during the fermentation was confirmed by moni-toring these sugars (Figure 2). It can be seen that fructose concentration stayed almost constant along fermentation, meaning that only glucose was con-sumed by the microorganism. The same behavior was observed before when P. aeruginosa ATCC 10145 was cultivated in cashew apple juice [32].

The surface tension of the culture broth decreased from 47.7 to 28.0 dyn/cm, and remained constant for up to 72 h (Figure 2). The lowest surface tension was achieved at the stationary phase (48 h). P. aeruginosa is known to produce rhamnolipids, which are capable of reducing the surface tension of the media by approximately 30–60% [39]. In this work, when CAJ was used, the surfactant produced by the bacteria was found to reduce the surface tension of

the medium by 58.08% (Table 1), which is com-parable with the earlier reports.

Emulsifying activity of the biosurfactant against non-aqueous phase liquids

According to the literature [30], the emulsi-fication properties of any surfactant depend on the tested solvent. Therefore, different non-aqueous phase liquids were tested as a substrate for emul-sifying activity by the biosurfactant and the results are shown in Figure 3.

As shown in Figure 3, all of the hydrocarbons and vegetable oil tested served as substrate for emul-sification by the biosurfactant, but it showed appre-ciable emulsification indices (more than 40%) with gasoline, diesel, kerosene and soy oil. Kerosene (83.5 % emulsified) was the best substrate, while hexane (10.7% emulsified) was the poorest. Most microbial surfactants are substrate specific, solubilizing or emulsifying different hydrocarbons at different rates [40,25]. The results of emulsifying activity (high values) indicate potential use, for instance, for enhanced oil recovery. The water-oil emulsions showed to be compact and remained stable for more than 4 months at room temperature, suggesting that the addition of such biosurfactant into a remediation process may enhance the availability of the recalcitrant hydrocarbon. Furthermore, the ability to emulsify vegetable oils suggests a potential application in the pharmaceutical and cosmetic indus-tries.

Figure 2. Kinetics of growth and production of biosurfactants by P. aeruginosa MSIC02 at 37 °C and 150 rpm in an optimized cashew apple juice medium, contained glucose concentration of 5 g/L and no phosphorous supplementation. (●) Optical density; (■) surface

tension; (○) glucose concentration (g/L); (□) fructose concentration (g/L).


55

Biosurfactant stability analysis

The stability of biosurfactants under extreme conditions is a pre-requisite for their potential appli-cations in enhanced oil recovery, and in environmen-tal and industrial applications [2,3,41,42]. The con-ditions that affect the performance of a biosurfactant are usually salinity, pH and temperature [30].

The effect of NaCl concentration on the surface activity of the biosurfactant produced is shown in Figure 4A, which was evaluated in order to investigate its applicability in the bioremediation of saline environments. Statistical tests (ANOVA and Tukey multiple comparison) showed that the surface activity was not significantly influenced by NaCl concentrations ranging from 0 to 10% and 20%, w/v, as shown in Figure 4B. The surface tension presented a significant decreased when 15%, w/v, NaCl was added, which may be explained by the presence of a net negative charge in the solution/air interface. Helvaci et al. [43] stated that electrolytes directly affect the carboxylate groups of the rhamnolipids. At pH 6.8, carboxylic acid groups are ionized and strong repulsive electrostatic forces between rhamnolipid molecules is promoted. As a result, a decrease in surface tension values is observed, probably because of the formation of a close-packed monolayer, caused by the fact that the negative charge is shielded by Na+ in the electrical double layer in the presence of NaCl [25,43].

Although surface tension was significant affected by NaCl concentration, it is important to notice that the surface tension remained in range of 29–30

dyn/cm, meaning that the biosurfactant retained its surface activity. Even higher salt tolerance (up to 35%) has been observed for biosurfactants obtained from a marine P. aeruginosa [44]. Considering that 3% is the highest salinity of sea in the world [25], these results indicate that the biosurfactant from P. aeruginosa MSIC02 could be applied in contaminated marines.

The effect of pH on biosurfactant activity is shown in Figure 4B. The surface activity was retained over a pH range of 6–11 with minimal deviation in surface tension, but the maximum surface activity was reached at pH 7–9, showing no significant diffe-rence in the surface tension. The rhamnolipids have their optimum aqueous solubility at neutral to alkaline pHs, which is attributed to their acidic nature, once the reported pKa is 5.6 [25]. The stability loss at low pH scale (<5) is probably due to the occurrence of precipitation, caused by the consequent insolubility of the biosurfactant produced by P. aeruginosa at these pH values [21,41,45]. It has been reported that as the pH increases from 5 to 8, the negative charge of the polar head increases, promoting an increase in aque-ous solubility [46].

Studies on the effect of heat treatment demon-strated that no appreciable change in surfactant surface activity had occurred (Figure 4C). Although surface tension means are significantly different, the surface tension values (28–30 dyn/cm) remained stable after exposure to high temperatures (100 °C), even after 75 min. The stability of the biosurfactant

Figure 3. Emulsifying activity (E24) of biosurfactant (produced by P. aeruginosa MSIC02 using cashew apple juice) obtained different hydrocarbons and vegetable oil. Error bars represent standard deviations.

M.V.P. ROCHA et al.: BIOSURFACTANT PRODUCTION BY Pseudomonas aeruginosa… CI&CEQ 20 (0) 000−000 (2014)

56

(A) (B)

(C) (D)

Figure 4. A) Effect of NaCl on activity of biosurfactant, B) effect of pH on activity of biosurfactant, C) effect of time of exposure at 100 °C and D) effect of temperature on activity of biosurfactant produced by P. aeruginosa MSIC02 at 37 °C and 150 rpm in an optimized

cashew apple juice medium. Values with different letters present statistically significant differences (p < 0.05).

was also tested over a wide temperature range. The biosurfactant produced showed to be stable during incubation for 15 min at temperatures ranging from 4 to 100 °C (Figure 4D), with no significant difference on surface tension values. When submitted to auto-clave sterilization (121 °C/15 min), the surface activity was also maintained (29.1±0.1 dyn/cm). These results indicate the usefulness of the produced bio-surfactant in industries where heating to achieve ste-rility is of paramount importance.

Likewise, thermo-tolerant and thermo-stable bio-surfactants from P. aeruginosa [25,46], Bacillus sub-tilis [3,30] and from Bacillus licheniformis [47] have been reported. The biosurfactant that was produced in this study showed stability when exposed to high temperatures, indicating that it could be used under

extreme temperature conditions, such as in micro-bially enhanced oil recovery – MEOR [46]. Emulsifying biosurfactants that are stable in environments with high pH and salinity would find applications in the bioremediation of spills at sea. Furthermore, the bio-surfactant may also be useful for bioremediation in hot and slightly alkaline environments.

CONCLUSION

Through the 24 full factorial design, it was observed that of the 4 variables studied, glucose con-centration and temperature were the most significant for biosurfactant production by Pseudomonas aeru-ginosa MSIC02. Increasing the cultivation tempera-ture and reducing the glucose concentration, present


57

in the cashew juice, caused a greater reduction in surface tension, indicating greater biosurfactant pro-duction. In addition, the properties (minimum surface tension and emulsifying activity) of the produced biosurfactant, as well as its high stabilities at high salinities, elevated temperatures, and over a wide pH range, makes these biosurfactants potential can-didates to be used in bioremediation of contaminated sites and in the petroleum industry (MEOR) where drastic conditions are very common.

Nomenclature

C/P Carbon/phosphorous ratio CAJ Cashew apple juice CMC Critical micellar concentration E24 Emulsifying activity, % X1 Temperature X2 Glucose concentration from cashew apple juice X3 Phosphorous concentration (KH2PO4) X4 Cultivation time

Acknowledgements

The authors would like to thank ANP, CAPES, CNPq and FINEP (from Brazil) for the financial sup-port that made this work possible.

REFERENCES

[1] C.J.B. Lima, F.P. França, E.F.C. Sérvulo, M.M. Resende, V.L. Cardoso, Appl. Biochem. Biotecnol. 137-140 (2007) 463-470

[2] C.N. Mulligan, Environ. Pollut. 133 (2005) 183-198

[3] M. Nitschke, G.M. Pastore, Bioresource Technol. 97 (2006) 336-341

[4] H. Yin, J. Qiang, Y. Jia, J. Ye, H. Peng, H. Qin, N. Zhang, B. He, Process Biochem. 44 (2009) 302–308

[5] I.M. Banat, R.S. Makkar, S.S. Cameotra, Appl. Microbiol. Biotechnol. 53 (2000) 495-508

[6] Y.H. Wei, I.M. Chu, Enzyme Microb. Tech. 22 (1998) 724-728

[7] R.S. Makkar, S.S. Cameotra, Appl. Microbiol. Biot. 58 (2002) 428-434

[8] M. Nitschke, S.G.V.A.O. Costa, J. Contiero, Process Biochem. 46 (2011) 621-630

[9] L. Sim, O.P. Ward, Z-Y. Li, J. Ind. Microbiol. Biot. 19 (1997) 232-238

[10] S. Itoh, T. Suzuki, Agric. Biol. Chem. 36 (1972) 2233- –2235

[11] M. Nitschke, S.G.V.A.O. Costa, R. Haddad, L.A. Gon-çalves, M.N. Eberlin, J. Contiero, Biotechnol. Progr. 21 (2005) 1562-1566

[12] M. Benincasa, J. Contiero, M.A. Manresa, I.O. Moraes, J. Food Eng. 54 (2002) 283-288

[13] M.E. Mercade, M.A. Manresa, J. Am. Oil Chem. Soc. 71 (1994) 61-64

[14] R.S. Makkar, S.S. Cameotra, J. Surfactants Deterg. 2 (1999) 237-241

[15] A.R. Najafi, M.R. Rahimpour, A.H. Jahanmiri, R. Roos-taazad, D. Arabian, Z. Ghobadi, Chem. Eng. J. 163 (2010) 188-194

[16] J.F. Morton, Fruits of warm climates, Echo Point Books & Media Brattleboro, VT, 1997, p. 239

[17] D.C.P. Campos, A.S. Santos, D.B. Wolkoff, V.M. Matta, L.M.C. Cabral, S. Couri, Desalination 148 (2002) 61-65

[18] R.B. Assunção, A.Z. Mercadante, J. Food Composition Anal. 16 (2003) 647-657

[19] M.V.P. Rocha, R.V.G. Barreto, V.M.M. Melo, L.R.B. Gonçalves, Appl. Biochem. Biotecnol. 155 (2009) 63-75

[20] J.A. Damasceno Júnior, F.C. Bezerra, Rev. Bras. Frutic. 24 (2002) 258-262

[21] R.S. Makkar, S.S. Cameotra, J. Ind. Microbiol. Bio-technol. 18 (1997) 37-42

[22] R.S. Makkar, S.S. Cameotra, J. Ind. Microbiol. Biotech-nol. 20 (1998) 48-52

[23] N.H. Youssef, K.E. Duncan, D.P. Nagle, K.N. Savage, R.M. Knapp, M.J. McInerney, J. Microbiol. Meth. 56 (2004) 339

[24] A. Bazire, A. Dheilly, F. Diab, D. Morin, M. Jebbar, D. Haras, A. Dufour, FEMS Microbiol. Lett. 253 (2005) 125– –131

[25] S.N.R.L. Silva, C.B.B. Farias, R.D. Rufino, J.M. Luna, L.A. Sarubbo, Colloids Surfaces, B 79 (2010) 174-183

[26] Y.-H. Wei, C.-L. Chou, J.-S. Chang, Biochem. Eng. J. 27 (2005) 146–154

[27] S. Pansiripat, O. Pornsunthorntawee, R. Rujiravanit, B. Kitiyanan, P. Somboonthanate, S. Chavadej, Biochem. Eng. J. 49 (2010) 185–191

[28] C.N. Mulligan, B.F. Gibbs, Appl. Environ. Microb. 55 (1989) 3016-3019

[29] H. Ghojavand, F. Vahabzadeh, E. Roayaei, A.K. Shah-raki, J. Colloid Interf. Sci. 324 (2008) 172

[30] A. Daverey, K. Pakshirajan, Colloids Surfaces, B 79 (2010) 246-253

[31] F.A.S.L. Reis, E.F.C. Sérvulo, F.P. De França, Appl. Biochem. Biotecnol. 113–116 (2004) 899-912

[32] M.V.P. Rocha, M.C.M. Souza, S.C.L. Benedito, M.S. Bezerra, G.R. Macedo, G.A.S. Pinto, L.R.B. Gonçalves, Appl. Biochem. Biotecnol. 136 (2007) 185-194

[33] L.M. Santa Anna, G.V. Sebastian, N. Pereira Jr., T.L.M. Alves, E.P. Menezes D.M.G. Freire, Appl. Biochem. Biotecnol. 91-93 (2001) 459-467

[34] A. Garrido-Lopez, M.T. Tena, J. Chromatogr., A. 1099 (2005) 75-83

[35] P. Yan, M. Lu, Q. Yang, H.-L. Zhang, Z.-Z. Zhang, R. Chen, Bioresource Technol. 116 (2012) 24–28

[36] L.H. Guerra-Santos, O. Kappeli, A. Fiechter, Appl. Mic-robiol. Biotechnol. 24 (1986) 443-448

[37] C. Ferraz, A.A. De Araujo and G.M. Pastore, Appl. Biochem. Biotecnol. 98 (2002) 841-847


58

[38] J.D. Desai, I.M. Banat Microbiol. Mol. Biol. Rev. 61 (1997) 47-64

[39] S. Lang, D. Wullbrandt, Appl. Microbiol. Biotechnol. 51 (1999) 22-32

[40] M.O. Ilori, C.J. Amobi, A.C. Odocha, Chemosphere 61 (2005) 985-992

[41] K.S.M. Rahman, T.J. Rhaman, S. McClean, R. Marchant, I.M. Banat, Biotechnol. Prog. 18 (2002) 1277 -1281

[42] R. Sen, Prog. Energ. Combustion 34 (2008) 714-724

[43] S.S. Helvacı, S. Peker, G. Özdemir, Colloids Surfaces, B 35 (2004) 225–233

[44] R. Thavasi, S. Jayalakshmi, T. Balasubramanian, I.M. Banat, Lett. Appl. Microbiol. 45 (2007) 686-691

[45] S. Joshi, C. Bharucha, A.J. Desai, Bioresource Technol. 99 (2008) 4603-4608

[46] A.M. Abdel-Mawgoud, M.M. Aboulwafa, N.A.H. Has-souna, Appl. Biochem. Biotechnol. 157 (2009) 329-345

[47] P. Ramnani, S.S. Kumar, R. Gupta, Process Biochem. 40 (2005) 3352 -3359.

MARIA VALDEREZ

PONTE ROCHA1

JOCÉLIA SOUSA MENDES1

MARIA ESTELA

APARECIDA GIRO1

VÂNIA MARIA M. MELO2

LUCIANA ROCHA

BARROS GONÇALVES1

1Universidade Federal do Ceará – Depto. de Engenharia Química - Bloco

709, Campus do Pici – CEP. 60.455-760, Fortaleza – CE, Brasil

2Universidade Federal do Ceará – Depto. de Biologia - Laboratório de

Ecologia Microbiana e Biotecnologia (LemBiotech) - Bloco 909, Campus do

Pici – CEP. 60.455-760, Fortaleza – CE, Brasil

NAUČNI RAD

PRODUKCIJA BIOSURFAKTANTA POMOĆU Pseudomonas aeruginosa MSIC02 NA PODLOZI SA SOKOM JABUKE KAŽUJE POMOĆU PUNOG FAKTORIJELNOG EKSPERIMENTALNOG PLANA 24

Produkcija biosurfaktanta pomoću Pseudomonas aeruginosa MSIC02 u podlozi sa sokom

jabuke kažuje je ispitivana pomoću punog faktorijelnog eksperimentalnog plana 24 u kome

su nezavisne promenljive: temperature, koncentracija glukoze (iz soka jabuke kažuje),

koncentracija fosfora i vreme kultivacije. Zavisna promenljiva je bilo procentno smanjenje

površinskog napona fermentacione tečnosti bez suspendovanih ćelija, kao indikatora

produkcije površinski aktivnog agensa. Maksimalna produkcija biosurfaktanta, ekviva-

lentna smanjenju površinskog napona od 58%, dobijena je na 37 °C pri koncentraciji

glukoze 5,0 g/L i bez dodatka fosfora. Smanjenje površinskog napona je bilo značajno

budući da je mala vrednost izmerena u hranljivoj podlozi (27,50 dyn/cm), što je ukazalo na

produkciju biosurfaktanta. Ovaj biosurfaktant emulguje različite hidrofobne supstance i

pokazuje stabilnost prema salinitetu, visokim temperaturama i ekstremnim pH vred-

nostima. Ove fizičkohemijske osobine pokazuju potencijal za različite primene biosurf-

aktanta proizvedenog pomoću P. aeruginosa MSIC02.

Ključne reči: biosurfaktant, površinski napon, Pseudomonas aeruginosa, puni faktorijelni eksperimentalni plan, sok jabuke kažuje, uvećano izdvajanje ulja.





59

MIODRAG N. TEKIĆ

IVANA M. ŠIJAČKI

MILENKO S. TOKIĆ

PREDRAG S. KOJIĆ

DRAGAN LJ. PETROVIĆ

NATAŠA LJ. LUKIĆ

SVETLANA S. POPOVIĆ

Department of Chemical Engineering, Faculty of

Technology, University of Novi Sad, Novi Sad, Serbia

SCIENTIFIC PAPER

UDC 532.5:66.069.82:66.021.2

DOI 10.2298/CICEQ120627102T

HYDRODYNAMICS OF A SELF-AGITATED DRAFT TUBE AIRLIFT REACTOR

Article Highlights • A novel-constructed draft tube airlift reactor, self-agitated by ten impellers, was

investigated • The insertion of impellers caused bubble breakup and reduction of mean bubble size • The riser gas holdup increases while the downcomer almost diminishes, bringing to

lower overall gas holdup • Circulation time is prolonged, downcomer liquid velocity decreases and mixing time

increases • Constructed reactor shows better performance compared to previous studies with

other internals Abstract

The main hydrodynamic characteristics of a novel-constructed, self-agitated draft tube airlift reactor (DT-ALR) were investigated. Ten impellers, driven only by means of gas throughput and induced liquid circulation, were inserted in the draft tube. The insertion of impellers caused bubble breakup and reduction of both mean bubble size and coalescence, even under the conditions of high gas throughputs. Although the impellers induced energy losses, the resistance to the flow was relatively lower due to their rotation, unlike the internals used in other research reported in the literature. In comparison to the conventional configuration of a DT-ALR, it was found that the presence of impellers led to significant changes in hydrodynamics: riser gas holdup and mixing time increased, while overall gas holdup and liquid velocity in the downcomer decreased.

Keywords: draft-tube airlift reactor, internals, self-agitated impellers, hyd-rodynamics.

Airlift reactors (ALRs) have important appli-cations in chemical and biochemical processes, due to numerous advantages they offer as efficient and economical devices for enabling an intimate inter-facial contact in different gas-liquid and gas-liquid-solid operations. The main aim to enhance the bubble breakup, but also to intensify the mass transfer in these reactors, led to intensive research of different reactor designs. Some of the solutions were to intro-duce a kind of an internal “obstacle”, which would alter the hydrodynamics and mass transfer in the investigated contactors. Table 1 summarizes the research to date in this matter.

Correspondence: I.M. Šijački, Department of Chemical Engineering, Faculty of Technology, University of Novi Sad, 403063 Novi Sad, Serbia. E-mail: [email protected] Paper received: 27 June, 2012 Paper revised: 30 October, 2012 Paper accepted: 1 November, 2012

Lin et al. [1] studied the performance of an external loop airlift reactor (EL-ALR) with slanted baffles placed on the riser wall, facing downwards at an angle of 30°. Although the mixing and the cir-culation time were drastically prolonged with the incorporated baffles, the mass transfer was improved, leading to increased product yields in such fermentors.

Different packing materials can be used as catalysts or biomass carriers in ALRs. Placed in the riser section, packing induces change in the flow patterns in comparison to the unpacked reactor, leading to greater flow tortuosity, longer path lengths for bubbles to travel and therefore, intensified inter-action between the bubbles and the packing material [2,3]. Gopal and Sharma [2] investigated the influence of Pall rings in the riser of a draft tube airlift reactor (DT-ALR). Although such packing increased the resis-tance to the liquid flow, decreasing the liquid circul-ation velocity and disabling the bubble penetration in

M.N. TEKIĆ et al.: HYDRODYNAMICS OF A SELF-AGITATED DRAFT TUBE… Chem. Ind. Chem. Eng. Q. 20 (1) 59−69 (2014)

60

the downcomer, the increase in the interfacial area (a) by a factor of 3-5 was observed. The research of Meng et al. [3], conducted in an external loop airlift bioreactor (EL-ALR) packed with woven nylon in the riser section, showed that optimal hydrodynamic con-ditions occurred at high packing porosity with the full packing height. Such conditions permitted high holdup of immobilized biomass attached to the pack-ing, highest gas holdup to improve the mass transfer and large void space to reduce plugging and liquid frictional losses.

Breakup of the bubbles in the ALRs can be achieved with static mixers. Stejskal and Potůček [4] reported that Kenics static mixers in the draft tube enhanced the origination of very fine gas bubbles, which recirculated through the column. The bubbles were held in the draft tube longer, but the downcomer gas holdup was lower in the presence of the motion-less mixer, because of lower liquid velocity. Gaspillo and Goto [5,6] investigated the influence of a static mixer in a riser of slurry DT-ALR and EL-ALR. In their research, static mixers were effective in dispersion of large bubbles created by a single nozzle. On the other hand, when a different distributor (which created very fine bubbles) was used, the presence of static mixer induced coalescence [5,6]. The research of Gavrilescu et al. [7], conducted in an EL-ALR with Sulzer static mixers in the riser, showed that in the non-Newtonian solutions the riser gas holdup was significantly increased. However, the liquid circulation

superficial velocity in the riser section was dimi-nished, due to a decrease in downcomer to riser cross-section ratio in the presence of the motionless mixer. The presence of static mixer led to an increase in mass transfer coefficient in both water-like [5,6,8] and non-Newtonian media [4,8].

Insertion of perforated plates could also modify the gas phase dispersion. Zhao et al. [9] studied the influence of baffles in several design modes of bubble columns (BCs) and DT-ALRs filled with Newtonian and non-Newtonian liquids. Two types of baffles were used: a baffle cap, formed by mounting a perforated plate on top of a short tube, and a perforated plate. Placed in the draft tube of a DT-ALR, the baffles caused an accumulation of bubbles underneath. When the pressure drop across the orifice of the per-forated plate was overcome, the gas was redis-tributed. Therefore, the gas was bubbled periodically through the baffles and the liquid circulation velocity was decreased. Mass transfer was enhanced with introduction of baffles in all reactor types with viscous Newtonian liquids. Nevertheless, the effect of baffles was less pronounced in case of non-Newtonian liquids. Complete investigations of hydrodynamics and mass transfer in a DT-ALR with three perforated plates in the draft tube were performed by Vora-pongsathorn et al. [10]. The research showed that presence of baffles reduced liquid circulation velocity, as they obstructed flow and increased the resistance to the liquid flow. Also, the baffles reduced the bubble

Table 1. Review of studies in airlift reactors with internals

Reference Reactor type and charateristics Internal Investigated parameters

Lin et al. (1976) EL-ALR; DR = 0.15 m, DD = 0.05 m, H = 3 m Slanted baffles kLa, tm , tc

Gopal and Sharma (1982) DT-ALR; D = 0.2 m, DDT = 0.11 m, HDT = 0.96 m

16 mm Stainless steell Pall rings

WLD, WLR, εG, εGR, a, kLa, kL, Δp

Meng et al. (2002) EL-ALR; DR = 0.089 m, DD = 0.047 m, HR = 1.81 m

Woven nylon packing εG, WL, axial dispersion, bubble size distribution

Stejskal and Potůček (1985) DT-ALR; D = 0.11 m, DDT = 0.058 m Kenics-180° static mixer kLa, NG, εG, εGD

Gaspillo and Goto (1991) slurry DT-ALR; D = 0.097 m, DDT = 0.027 m, H = 0.37 m

Static mixer Δp, UG,min, kLa

Goto and Gaspillo (1992) Slurry EL-ALR; DR = 0.027 m, DD = 0.061 m, HR = 0.585 m, HD = 0.39 m

Static mixer Δp, ULR, UG,min, kLa

Chisti et al. (1990) EL-ALR; DR = 0.050 m, DD = 0.075 m

SMV-12 Static mixer elements

kLa

Gavrilescu et al. (1997) EL-ALR; AD/AR = 0.1225 Sulzer type static mixer WL, εGR

Zhao et al. (1994) DT-ALR; D = 0.14 m, DDT = 0.09 m, H = 1.7 m Baffle caps, sieve plates kLa, εG, WL

Vorapongsathorn et al. (2001) DT-ALR; D = 0.137 m, DDT = 0.093 m, HDT = 1 m

3 Perforated baffle plates εG, εGD, εGR, kLa, WL

Krichnavaruk and Pavasant (2002)

DT-ALR; D = 0.098 m, DDT = 0.068 m, H = 2.4 m, HDT = 2.07 m

Perforated plates WLD, kLa, εGR

Chisti and Jauregui-Haza (2002)

DT-ALR annulus sparged; D = 0.755 m, DDT = 0.50 m, H = 3.21 m, HDT = 2.06 m

2 Impellers mechanically agitated

kLa, εG


61

rise velocity, thus leading to a slight enhancement of the riser holdup in comparison to the non-baffled configuration. Downcomer gas holdup was decreased by lowering the liquid velocity. The result was a negligible effect of baffles on the overall gas holdup, and related volumetric mass transfer coefficient. The presence of baffles led to a development of stagnant regions underneath the plates, which streamlined the flow path. This phenomenon resulted in the rise of liquid velocity with the increase of the power through-put. Exhaustive analysis of the perforated plate effect on bubble forming or breakage in a DT-ALR was presented by Krichnavaruk and Pavasant [11]. By employing a photographical technique, a very stu-dious analysis on this phenomenon was performed in several reactor designs with various plate number and plate configurations. The role of the perforated plate was to break large bubbles into smaller ones. However, it was observed that they coalesced again after they left perforated plate and gained almost equal size as before they reached the plate. Plates with too large free area were less effective in bubble breakage. The presence of plates reduced riser liquid velocity and circulation velocity, but augmented riser gas holdup and interfacial mass transfer area (a). Although the mass transfer coefficient (kL) was decreased, the overall volumetric mass transfer coef-ficient was as much as twice the value obtained from the conventional DT-ALR.

Improved agitation in the DT-ALR was evaluated through research of Chisti and Jauregui-Haza [12]. Two low-power hydrofoil impellers, mechanically driven by a motor, were used to enhance fluid cir-culation in the reactor. The gas was sparged into the annular zone to avoid the impeller flooding. It was found that the gas holdup rose with increasing aera-tion and agitation rates. The effect of mechanical agitation on mixing time was pronounced at relatively low aeration rates. At higher aeration velocities, rising bubbles were the dominant cause of the mixing. The oxygen transfer capability was improved, but the oxygen transfer efficiency was reduced by use of mechanical agitation. Generally, the authors con-cluded that the use of mentioned impellers in the downcomer of an ALR could substantially enhance the rate of liquid circulation, mixing and gas-liquid mass transfer relative to operation without the agi-tator. However, the performance improvement was made through disproportionate increase in the power consumption [12].

Investigations of the DT-ALR with different type of internals in the draft tube have shown that these contactors are mainly superior to the conventional

configuration. Insertion of internals, such as baffles, static mixers or packing increases the gas holdup, breaks up the bubbles, and therefore, enhances the mass transfer by increasing the specific interfacial area (a). However, these internals increased the resistance to the liquid flow. Thus, the liquid velocity is lower and the pressure drop is significantly higher. In such conditions the volumetric mass transfer coef-ficient (kLa) might be decreased.

It is known that addition of a separator can also alter the hydrodynamics in a DT-ALR [13]. This kind of modification leads to lower downcomer gas holdup, in comparison to configuration without the separator. In this case, the investment costs are probably significantly lower than in reactors with different baf-fles, static mixers or other internals. However, the presence of the separator affects the separation of the gas phase at the top, therefore on the driving force for the liquid circulation, but has no additional influence on the bubble breakup. Although additional parts such as baffles, packing, static mixers, etc. (see Table 1) bring further costs and lead to a more complex construction of reactors, generally known as simple-constructed, they have their specific role as biomass carriers, turbulence promoters, etc.

As it can be seen from Table 1, previous research of modified reactors included only the con-figurations with packing, static baffles, perforated plates or static mixers. Until now, reactors with incur-porated impellers were mechanically agitated. The objective of this experimental work was to introduce agitation in the DT-ALR using the already present energy of the gas throughput. So, the hydrodynamics of a self-agitated DT-ALR, as a novel-constructed contactor, with inserted impellers in the riser section was investigated and compared to the conventional DT-ALR. In this initial stage of our research only a single orifice (known as the least effective sparger) was used as the gas distributor, in order to point out the strong impact of the impellers on bubble breakup and additional mixing. The impellers were rotated by the gas throughput and induced liquid circulation; therefore, the resistance should be lower in com-parison to motionless internals.

EXPERIMENTAL SETUP

The experiments were performed at 20±1 °C and atmospheric pressure in a glass DT-ALR, with geometrical details presented in Figure 1. Two con-figurations of the same reactor were used: con-ventional DT-ALR and a configuration with impellers installed in the riser. A shaft with 10 impellers was


62

mounted in the centre of the draft tube. In order to reduce the friction between the impellers and the shaft, Teflon rings were placed in the plastic impel-ler’s shell. The impellers were driven only by the gas throughput and induced liquid circulation. The dis-tance between the impellers and their geometrical details are depicted in Figure 1. The number and disposition of the impellers were chosen based on a preliminary research, in order to achieve the best performance: breakup of the bubbles with the mini-mum resistance to the flow. It was noticed that all the impellers started rotating simultaneously at the lowest gas flow used in this experimental work (QG = 200 l/h). This minimum gas flow was chosen having in mind that the obtained gas holdup in this case over-comes the error of the employed experimental method.

The air, sparged through a single orifice into the draft tube, was used as the gas phase. The gas flow rates were controlled and measured by a rotameter. Tap water was used as the liquid phase.

The overall gas holdup was determined by the volume expansion technique with an error less than 10%. The gas holdup values along the downcomer were obtained by measuring the pressures at five points using piezometric tubes. In order to reduce the liquid surface fluctuations in the piezometric tubes, the capillaries (50 mm in length and 0.7 mm i.d.) were inserted at the entrance of the tubes. Therefore, the relative average error of these measurements was diminished to 2%. Values of the gas holdup in the separator were also determined by piezometric tube placed at the entrance of the separator. The riser gas holdup was calculated upon the balance equation:

Figure 1. Experimental setup.

ε ε εε + − − − += ( ) ( )( )GL D R G DT D GD GL DT D R GSGR

DT R

H A A H A H H A AH A

(1)


63

Downcomer liquid velocity, liquid circulation time and the mixing time were measured by a pulse tracer technique. Two self-constructed conductivity sensors were placed in the downcomer section (as it is shown in Figure 1) at L = 0.75 m distance from each other. Solution of 4 mol/l NaCl, used as a tracer, was injected in the downcomer (Figure 1). The signal, caused by the tracer pass between the sensors, was measured using an A/D converter and recorded on a PC as a conductivity-time function with characteristic peaks. Measurements were performed with a samp-ling frequency of 0.2 s until no change was observed in the liquid phase conductivity. The moment when the tracer was registered by the sensor (t1 for the first and t2 for the second sensor), later used as an output value, was obtained as half peak width at base. The time delay between the signals of both sensors was used for calculation of liquid velocity based on the following equation:

=−2 1( )LDL

Wt t

(2)

For each value of a gas flow rate, at least five measurements were performed and the average value of downcomer interstitial liquid velocity was calcul-ated. The relative average error of this method was ±5%.

Liquid circulation time (tc) was determined with the relative average error of ±5% from the same measu-rements, as the time difference between the two adja-cent peaks on the output signals for both sensors.

Mixing time (tm) was calculated from the output signal for the first sensor, as the time required for the decrease of output signal to 5% of the maximum signal value. The relative average error of the measu-rements was ±10%.

The impeller rotation was recorded on a Sam-sung VP-D353i camera. One of the blades on the impeller was marked with vivid yellow color. There-fore, the impeller speed was later calculated by track-ing the marked blade through several rotations by analyzing the video sequences.


Hydrodynamics – main observations and the influence of impellers. Visually, it was observed that all the impellers in the modified DT-ALR started rota-tion under the minimal gas throughput employed in this experimental work. The rotation speed was almost uniform regardless to the impeller position along the riser. As it can be seen from Figure 2a, the impellers speed increased with the increase of the superficial gas velocity. At low gas throughputs this

increase is very high, according to the slope of the function impellers speed vs. superficial gas velocity. However, as the observed transition between the regimes in the riser occurred, the increase in the impellers speed became less steep (see Figure 2a). In the turbulent regime, the backward liquid flow pro-bably hindered the impellers rotation. Therefore, above the superficial gas velocities of 0.05 m/s, the impellers rotation speed tended to be constant.

Pressure of the gas entering the column was measured in order to quantify the resistance of impel-lers. As it is obvious from Figure 2b, no major increase in the pressure value was noticed, in com-parison to the reactor without the impellers, until UG ≈ ≈ 0.050 m/s, which corresponds to the slug flow in the riser. On the contrary, it seems that the impeller’s rotation streamlined the flow, thus lowering the expected resistance. From that point (UG ≈ 0.050 m/s) forward, the resistance to the gas throughput was larger in the presence of impellers because of the intensified circulation of bubbles and detention of the gas in the riser. Also, the energy necessary for the impeller rotation was increased, as the riser was filled with gas plugs. All of this resulted in about 17% higher pressure of gas at the inlet in the self-agitated reactor.

Figure 2. Average impellers speed and pressure of gas entering the column as a function of superficial gas velocity.


64

Insertion of impellers in the draft tube of the DT- -ALR led to significant changes in the hydrodynamics, in comparison to the conventional configuration. Rota-tion of the impellers caused breakup of the large bub-bles originated in the draft tube and contributed to a uniform radial distribution of the bubbles, for lower superficial gas velocities. The bubble coalescence in the draft tube was reduced, in comparison to the con-ventional DT-ALR. Although the impellers could be observed as obstacles to the gas upflow in the riser, neither the accumulation of bubbles nor the formation of air pockets under the impellers were noticed. Inten-sive bubble breakup enabled origination of many, very small, bubbles (< 1 mm in diameter) that were dragged into the downcomer. At low gas throughput, these bubbles accumulated at the entrance of the downcomer, forming a kind of resistive layer. With an increase of the gas superficial velocity, this resistance to the liquid circulation was overcome, and the bub-bles started recirculating. However, in the presence of the impellers, even under high gas throughputs, the downcomer was filled with tiny bubbles of 1–2 mm in diameter and less, because of which the downcomer bulk was milky. The rise velocity of such small bub-bles was very low; therefore, they are detained in the column for a few minutes even if the gas flow was stopped. Also, the impellers enhanced the lagging of bubbles. In the modified configuration of the ALR, even under the conditions of the slug flow in the riser (UG ≈ 0.050 m/s), the bubbles in the downcomer were smaller in diameter and much more stable to the ten-dency of coalescence, in comparison to the conven-tional reactor. The coalescence occurred only at the entrance in the annulus, but at much higher gas throughputs than in the conventional DT-ALR. Although the bubbles of larger diameters (4–6 mm) were present in the downcomer at superficial gas velocities of about 0.053 m/s, they formed a stagnant layer. Through their accumulation a certain zone was created, whose boundary with the underneath region of small bubbles was sharp. As the superficial gas velocity rose, the zone of larger bubbles was moving downward the annulus, until they started recirculating at superficial gas velocities of about 0.062 m/s. From this point forward, the number of larger, stable bub-bles in the downcomer increased noticeably.

Gas holdup. Figure 3 illustrates the comparison of the overall, riser and downcomer gas holdup between the two airlift configurations – with and without the impellers. Presence of the impellers induced significant increase in the riser gas holdup, which became even higher with an increase of the gas throughput. By the following: breaking up the

bubbles, reducing their coalescence and prolonging detention of bubbles in the riser, the gas holdup in the draft tube became 80–200% (mainly about 100%) larger, in comparison to the conventional reactor. Also, the appearance of the very large air bubbles, which fill out almost the whole cross section of the riser, was prolonged and attenuated. On the other hand, the presence of very small bubbles in the down-comer resulted in a very low downcomer gas holdup, which, in a self-agitated ALR, did not exceed the value of 2% unless a significant increase of the gas throughput took place. The reduction of the down-comer gas holdup in the reactor with incorporated impellers was about 88%, in comparison to the reac-tor without impellers. As the result of simultaneous enlargement of the riser holdup and extreme reduc-

Figure 3. Overall, riser and downcomer gas holdup as a functionof superficial gas velocity and reactor configuration.


65

tion of the downcomer gas holdup, the overall gas holdup in the novel DT-ALR was either equal to the gas holdup in the conventional DT-ALR or about 24% lower. Obviously, for this reactor geometry, it is expected that the changes in the downcomer gas holdup have dominant influence on the overall gas holdup.

Based on the research of Blažej et al. [14], higher downcomer, riser and overall gas holdups and higher liquid circulation velocity could be obtained in reactors of larger scale. In larger reactors, different specific friction against liquid circulation exists in com-parison to smaller reactors. Also, the gas phase in larger reactors recirculated even in the case of the lowest values of the superficial gas velocity. Having in mind that reactor scale strongly affects the hydro-dynamics [14], research with added impellers should also be conducted in larger reactors, in order to exa-mine the results from the small-scale reactor used in this work.

Liquid phase velocity and circulation time. The difference between the riser and the downcomer holdup, as the driving force for the liquid circulation through the column, is presented in Figure 4, as a function of the superficial gas velocity for both reactor with and reactor without the impellers. The dispro-portional increase in the riser holdup and decrease in the downcomer gas holdup led to significant enhance-ment of the driving force of about 370% in the modified reactor configuration. It is to be expected that this enlarged driving force will increase the liquid circulation velocity. However, the circulation time was about 1.9 times higher (see Figure 4b), thus the liquid circulation velocity was lower, if the impellers were present. The reason could be found in energy losses due to maintaining the impellers rotation, but also in extended flow path and flow tortuosity. Also, the decrease in the circulation time with an increase in superficial gas velocity was more pronounced in the reactor with added impellers. The dampening of the circulation velocity led to an entrainment of only tiny bubbles in the downcomer. As the bubble breakup by the impellers enhanced the origination of very small bubbles (< 1 mm), they hoarded at the downcomer entrance and formed a resistive layer for liquid circul-ation. Therefore, the downcomer liquid velocity was about 2.2 times lower in the DT-ALR with impellers (Figure 4c). By entrainment of larger bubbles into the downcomer and after their accumulation at the upper part of this region, for higher superficial gas velocities, the hydraulic resistance to the liquid flow increased causing the descent in liquid velocity. It can be noticed that the downcomer liquid velocity tended to

settle at almost constant value, despite the increase of gas throughput.

Figure 4. Driving force, circulation time and downcomer liquid velocity as a function of superficial gas velocity and reactor

configuration.

The transition between the following regimes in the downcomer: 1) presence of small bubbles, 2) formation of stagnant swarm of bubbles and 3) circul-ation of bubbles through the column, could be observed in Figures 3 and 4, for both reactors, as the changes in the slopes of the curves representing the overall gas holdup, driving force and downcomer


66

liquid velocity. The transition between the named regimes shifts to higher gas throughputs in the pre-sence of impellers, as indicated by the dashed lines in the mentioned figures.

Mixing time. The influence of superficial gas velocity in both reactor with and reactor without impellers on mixing time is depicted in Figure 5. For both reactor configurations, the mixing time decreased with increased gas throughput. Insertion of impellers in the draft tube deteriorated the mixing performan-ces. In the self-agitated reactor, three characteristic peaks (see Figure 5), showing local minima of the mixing time function, could be observed for superficial gas velocities of about 0.032, 0.050 and 0.063 m/s. The mentioned velocities correspond to the visually observed transition moments when the increased liquid velocity was sufficient enough to enable the recirculation of the accrued larger bubbles in the downcomer. Growing presence of bubbles and their circulation was an important factor for mixing impro-vement. Similar observations, meaning that flow regimes and the presence of solids or internals could alter the hydrodynamics and, therefore, have strong influence on the mixing time, are found in other papers [15,16]. Pandit and Joshi [15] reported that mixing phenomenon in gas-liquid contactors depends upon the gross liquid circulation and the microscale turbulence, while the molecular diffusion is usually negligible. They also concluded that introduction of the draft tube in the BC reduced the turbulence in

annulus and enhanced the clear circulation pattern, thus increasing the mixing time in comparison to the BC. Petrović et al. [16] investigated the mixing in the gas-liquid-solid DT-ALR, but the similar observations in the mixing time were reported as in this paper. The flow regime of bubbles had great influence on the mixing and circulation time. Bubble entrainment in the annulus led to less intensive mixing as the stationary cloud of bubbles presented the hydraulic resistance to the flow. As the stationary cloud of bubbles was formed, the mixing time versus superficial gas velocity curve shows the local minima. When this resistance was overcome and the circulation of bubbles through the column started, the mixing time decreased and the local maxima on the mixing time curve appeared. Chisti and Jauregui-Haza [12] also confirmed that the movement of bubbles through the contactor is the dominant cause of mixing.

Blažej et al. [14] concluded that the increase in reactor scale provided higher circulation velocities and shorter mixing times, but also a better distribution of the gas phase. Therefore, it is to be expected that the use of impellers in reactor of larger scale could lead to better performance in sense of mixing.

Comparison of results obtained in this work with the results of previous experiments is presented in Figure 6. Figure 6a depicts the overall gas holdup in reactors with different types of internals [2,4,8,10] along with the values obtained in this experimental work. The differences between the mentioned values,

Figure 5. Mixing time as a function of superficial gas velocity and reactor configuration. Legend: ○ reactor without impellers, ● reactor with 10 impellers in the riser.


67

besides the insert type, could be associated mainly with different reactor geometry, as well as type of used gas sparger. Stejskal and Potůček [4] conducted experiments in a DT-ALR with riser to column diame-ter ratio (DR/D) equal to the ratio in our work. Before reaching UG ≈ 0.05 m/s, the values of the overall εG, in both cases, were similar. Later, for higher super-ficial gas velocities our experimental εG values were 25–48% higher, which led to the conclusion that the impellers provided better gas dispersion than static mixers, even with less effective gas distributor. The values for overall gas holdup, reported by Gopal and Sharma [2], are about 50% higher than ours, for all UG values less than 0.06 m/s. After that, we obtained

about 25% higher values, owing to the advantages of impellers in comparison to the packing. Our results are about 100% lower than the ones of Vorapong-sathorn et al. [10], probably because they employed a shorter draft tube and a reactor with higher DR/D ratio. However, in comparison to Chisti and Jauregui-Haza [12], who introduced agitation by mechanically driven impellers, our results are about 200% higher. Having in mind that the column used in their experimental work was of much larger scale, the results are not completely quantitatively comparable with ours.

In the case of the riser gas holdup, our results show enhancement of about 100% after insertion of impellers. Figure 6b presents the results of other authors [10,11] in comparison to the results of this experimental work. In both cases [10,11], the enhan-cement in riser gas holdup is lower. Vorapongsathorn et al. [10] found that, in comparison to the conven-tional reactor, the inserted perforated baffle plates increased the riser gas holdup for about 15%, but Krichnavaruk and Pavasant [11] reported the increase of about 73% with the same internals. Nevertheless, comparison of riser gas holdup values from Figure 6b shows that our results are about 25–45% higher than the results of others [10,11] because of our modi-fication.

CONCLUSIONS

The results of the presented work show that the insertion of impellers in the riser section of a DT-ALR strongly alters its hydrodynamics. In comparison to the conventional DT-ALR, the following effects are noticed:

• The gas holdup in the riser noticeably inc-reases, but the downcomer gas holdup almost dimi-nishes, resulting in slightly lower overall gas holdup.

• Circulation time is prolonged and the down-comer liquid velocity decreases.

• Mixing time increases, thus, the overall mixing is not enhanced due to presence of impellers.

• The results show that the obtained riser gas holdup values in this experimental work, especially under the conditions of higher gas throughputs, are much higher in comparison to previous experimental studies with other internals. Also, the appearance of the slug flow in the riser was prolonged, thus leading to more stable reactor operation.

Therefore, better understanding of the impellers effect on the hydrodynamics and mass transfer of self-agitated DT-ALR requires future research that would allow application of various liquids as well as

Figure 6. Comparison of previous research with this study. a) Influence of internals on the overall gas holdup; Legend: ■

Gopal and Sharma [2] (reactor with Pall rings), □ Stejskal and Potůček [4] (reactor with Kencis static mixers), ● this study

(reactor with impellers), ○ Vorapongsathorn et al. [10] (reactor with perforated baffle plates), ▲ Chisti and Jauregui-Haza [12]

(reactor without mechanical agitation by impellers), Δ Chisti and Jauregui-Haza [12] (reactor with mechanical agitation,

impeller speed 170 rpm)). b) Riser gas holdup comparison of previous research with this study; Legend: □ Vorapongsathorn

et al. [10] (reactor without baffles), ■ Vorapongsathorn et al. [10] (reactor with baffles), Δ Krichnavaruk and Pavasant [11]

(reactor without perforated plates), ▲ Krichnavaruk and Pavasant [11] (reactor with perforated plates), ○ this study

(reactor without impellers), ● this study (reactor with impellers).


68

measurements of mass transfer coefficients on the novel-constructed and other similar reactors.

Acknowledgement

This research was supported by the Ministry of Education, Science and Technological Development of the Republic of Serbia (Project No. 172025).

Nomenclature

A cross sectional area (m2) D diameter of column (m) DR inner diameter of riser (m) DD inner diameter of downcomer (m) kLa volumetric mass transfer coefficient (1/s) L distance between the conductivity sensors (m) H height (m) NG gas power output (W) Δp pressure drop (bar) tc liquid circulation time (s) tm mixing time (s) t time (s) UG superficial gas velocity, column based (m/s) WLD downcomer interstitial liquid velocity (m/s)

Abbreviations

ALR airlift reactor BC bubble column DT-ALR draft tube airlift reactor EL-ALR external loop airlift reactor rps rotations per second

Greek letters

εG gas holdup

Subscripts

C circulation D downcomer DT draft tube G gas phase

L liquid phase R riser S separation zone

REFERENCES

[1] C.H. Lin, B.S. Fang, C.S. Wu, H.Y. Fang, T.F. Kuo, C.Y. Hu, Biotechnol. Bioeng. 18 (1976) 1557-1572

[2] J.S. Gopal, M.M. Sharma, Can. J. Chem. Eng. 60 (1982) 353-362

[3] A.X. Meng, G.A. Hill, A.K. Dalai, Ind. Eng. Chem. Res. 41 (2002) 2124-2128

[4] J. Stejskal, F. Potucek, Biotechnol. Bioeng. 27 (1985) 503-508

[5] P.-A.D. Gaspillo, S. Goto, J. Chem. Eng. Jpn. 24 (1991) 680-682

[6] S. Goto, P.D. Gaspillo, Chem. Eng. Sci. 47 (1992) 3533- –3539

[7] M. Gavrilescu, R.V. Roman, R.Z. Tudose, Bioprocess. Biosyst. Eng. 16 (1997) 93-99

[8] Y. Chisti, M. Kasper, M. Moo-Young, Can. J. Chem. Eng. 68 (1990) 45-50

[9] M. Zhao, K. Niranjan, J.F. Davidson, Chem. Eng. Sci. 49 (1994) 2359-2369

[10] T. Vorapongsathorn, P. Wongsuchoto, P. Pavasant, Chem. Eng. J. 84 (2001) 551-556

[11] S. Krichnavaruk, P. Pavasant, Chem. Eng. J. 89 (2002) 203-211

[12] Y. Chisti, U.J. Jauregui-Haza, Biochem. Eng. J. 10 (2002) 143-153

[13] J. Klein, Š. Godo, O. Dolgoš, J. Markoš, J. Chem. Technol. Biotechnol. 76 (2001) 516-524

[14] M. Blazej, M. Kisa, J. Markos, Chem. Eng. Process. 43 (2004) 1519-1527

[15] A.B. Pandit, J.B. Joshi, Chem. Eng. Sci. 38 (1983) 1189- –1215

[16] M.M. Milivojević, A.P. Duduković, B. Obradović, Hem. Ind. 58 (2004) 10-18.


69

MIODRAG N. TEKIĆ

IVANA M. ŠIJAČKI

MILENKO S. TOKIĆ

PREDRAG S. KOJIĆ

DRAGAN LJ. PETROVIĆ

NATAŠA LJ. LUKIĆ

SVETLANA S. POPOVIĆ

Department of Chemical Engineering, Faculty of Technology, University of

Novi Sad, Novi Sad, Serbia

NAUČNI RAD

HIDRODINAMIKA SAMO-MEŠAJUĆE BARBOTAŽNE KOLONE SA KONCENTRIČNOM CEVI

U ovom radu su istražene glavne hidrodinamičke karakteristike novog tipa barbotažne

kolone sa koncentričnom cevi, sa samo-pokretajućim impelerima. U centralnoj cevi kolone

bilo je postavljeno deset impelera koji su bili pokretani isključivo pomoću uvođenja gasa u

kolonu i usled toga, indukovane cirkulacije tečnosti. Prisustvo impelera je uzrokovalo

razbijanje mehurova, smanjenje njihove veličine kao i redukciju koalescencije, čak i pri

velikim brzinama gasa. Mada su impeleri uzrokovali dodatne gubitke energije, otpor

proticanju tečnosti je bio relativno mali, zbog njihovog okretanja, u poređenju sa podacima

drugih autora dostupnih u literaturi, koji su koristili druge umetke U poređenju sa

uobičajenom konfiguracijom barbotažne kolone sa koncentričnom cevi, primećeno je da je

prisustvo impelera uzrokovalo značajne hidrodinamičke promene: sadržaj gasa u

centralnoj cevi i vreme mešanja su povećani, dok su ukupan sadržaj gasa i brzina tečnosti

u anulusu smanjeni.

Ključne reči: barbotažna kolona sa koncentričnom cevi, umetci, samo-pokretajući impeleri, hidrodinamika





71

VESNA NAĐALIN1

ŽIKA LEPOJEVIĆ2

MIHAILO RISTIĆ3

JELENA VLADIĆ2

BRANISLAVA NIKOLOVSKI2

DUŠAN ADAMOVIĆ4

1Technical College of AppliedSciences in Zrenjanin, Zrenjanin,

Serbia2University of Novi Sad, Faculty of

Technology, Novi Sad, Serbia3Institute for Medicinal PlantResearch “Dr Josif Pančić”,

Belgrade, Serbia4Institute of Field and Vegetable

Crops, Novi Sad, Serbia

SCIENTIFIC PAPER

UDC 582.929.4:66.061:543.544.3

DOI 10.2298/CICEQ120715103N

INVESTIGATION OF CULTIVATED LAVENDER (Lavandula officinalis L.) EXTRACTION AND ITS EXTRACTS

Article Highlights • Physicochemical properties of lavender essential oil were determined • Lavender flower was extracted with supercritical CO2 under isothermal and isobaric

conditions • Modeling the extraction system lavender flower-supercritical CO2 was performed • Essential oil and CO2 extracts of lavender flower analysis was done by GC/MS and

GC/FID Abstract

In this study essential oil content was determined in lavender flowers andleaves by hydrodistillation. Physical and chemical characteristics of the isolated oils were determined. By using CO2 in supercritical state, the extrac-tion of lavender flowers was performed with a selected solvent flow underisothermal and isobaric conditions. Qualitative and quantitative analysis of the obtained essential oil and supercritical extracts (SFE) was carried out usinggas chromatography in combination with mass spectrometry (GC/MS) and gaschromatography with flame ionisation detector (GC/FID). Also, the analysis ofindividual SFE extracts obtained during different extraction times was per-formed. The main components of the analysed samples were linalool, linaloolacetate, lavandulol, caryophyllene oxide, lavandulyl acetate, terpinen-4-ol and others. Two proposed models were used for modelling the extraction system lavender flower - supercritical CO2 on the basis of experimental results obtained by examining the extraction kinetics of this system. The appliedmodels fitted well with the experimental results.

Keywords: lavender, extraction, extracts, modelling, supercritical CO2.

Lavender (Lavandula officinalis L.) is a member of the Lamiaceae family and is a plant species that predominantly contains lipophilic components (essen-tial oil) that are responsible for the anti-inflammatory, antiseptic, sedative and spasmolytic activity [1,2]. In addition, lavender contains hydrophilic components (phenolic compounds, flavonoids – mainly flavone gly-cosides, anthocyanins, tannins, etc.).

The complex chemical composition of this plant species represents the basis for its medicinal pro-perties. Essential oil content in lavender flowers, obtained by steam distillation, ranges from 1–3%, and oil quality significantly depends on the total amount of Correspondence: V. Nađalin, Technical College of Applied Sci-ences in Zrenjanin, Đorđa Stratimirovića 23, Zrenjanin, Serbia. E-mail: [email protected] Paper received: 15 July, 2012 Paper revised: 31 October, 2012 Paper accepted: 1 November, 2012

esters, which is around 35–55% recalculated on linalyl acetate [3]. In the essential oil of lavender more than 100 components have been isolated, of which the fol-lowing were identified esters: linalyl acetate (17.6– –53%), lavandulyl acetate (15.95%) and genaryl ace-tate (5.0%); alcohols: linalool (26–49%), α-terpineol (6.7%) and terpinen-4-ol (0.03–6.4%); sesquiterpenes: β-caryophyllene (2.6–7.6%); monoterpenes: cis-β-oci-mene (1.3–10.9%); oxides: 1.8-cineole (0.5–2.5%). It was confirmed that lavender possesses 12% of tan-nins, while the content of camphor, ketones is less than 1%, which is much lower than other types of Lavandulaceae [4]. For this reason lavender found its application in the cosmetics and perfume industry, in contrast to Lavandulaceae with high camphor con-tent, which are used as insecticides, rubefacients and for other purposes [5].

V. NAĐALIN et al.: INVESTIGATION OF CULTIVATED LAVENDER… Chem. Ind. Chem. Eng. Q. 20 (1) 71−86 (2014)

72

It was established that the pharmacological acti-vity of lavender derives from linalool [6]. Due to pre-sence of linalool, linalyl acetate, and the mentioned esters and alcohols, the essential oil of lavender exerts anti-inflammatory, antifungal, antiseptic and sedative effects [7]. Higher concentrations of alco-hols, aldehydes, esters, ketones and sesquiterpenes induce spasmolytic effects on smooth musculature [8,9].

It is particularly important to note that, with con-firmed antibacterial and antifungal activity, lavender oil operates in-vitro on methicillin resistant strain of Staphylococcus aureus and vancomycin resistant strain Enterococcus faecium in concentrations less than 1% (v/v) [2,10].

The antioxidant property of lavender is asso-ciated with the presence of polyphenolic compounds. A qualitative analysis of these polyphenolic com-pounds in lavender extracts showed that rosmarinic acid is the most abundant phenolic constituent, which is known for its antiviral, antibacterial, antioxidant, anti-inflammatory and immune stimulating effect [11].

In studies carried out with ethanol extracts of lavender it was found that polyphenol content varies, not only in different parts of the plant (stem, flower and leaf), but also among Lavandula species. In the leaf extracts the concentrations of phenolic acids, flavonoids, procyanidins, total tannins and polyphe-nols are higher compared to extract of flowers and stems. Phenolic acids are the most abundant in the extract of lavender ranging from 2.41 to 5.32% in the leaves, followed by the flowers and then the stems. The content of total polyphenols in the extracts of leaves is 9.20%, flowers 8.46% and stems 4.54%. In the extracts of flowers and leaves the flavonoid con-tent varies from 0.09–0.26%. The total tannin content in the leaves is 3.18%, flowers 2.77% and stems 1.38% [12].

Compared to conventional extraction proce-dures, the extraction of plant materials by gases under pressure, most notably carbon dioxide, has an important role for a number of reasons [13-15]. A supercritical fluid has a wide range of possibilities for selective extraction, fractionation and purification of natural materials [16]. Changing the pressure and/or temperature above the critical value for CO2 (Tc = = 31.1 °C; pc = 7.38 MPa; dc = 0.469 g/ml) leads to changes in density and dielectric constant of CO2, which makes it possible to control the yield and composition of the obtained extracts [14]. The extrac-tion procedure is performed at moderate tempera-tures and relatively low pressures, so it can be used for isolating thermolabile compounds [16,17]. In pro-

cedures of supercritical fluid extraction, carbon diox-ide represents the most suitable solvent, due to its non-toxicity, chemical inertness, physiological inacti-vity, nonflammability and spontaneous release from the extract at atmospheric conditions [18,19].

Mathematical modeling

Supercritical carbon dioxide extraction kinetic data of lavender flowers were modelled using two mathematical models that assume the extraction rate to be controlled by internal diffusion. The first model was a modified model [20] of Reverchon and Sesti Osseo [21], which had been successfully applied to supercritical fluid extraction of lavender flowers [22]. This model had two adjustable parameters. The second model was a simplified model suggested by Sovová [23] with three adjustable parameters. This model contains, in addition, an equation for the initial extraction from the surface of the plant. Sovová [23] used this model for mathematical modelling of some previously published experimental results of super-critical fluid extraction of lavender oil [24]; however, for a narrower range of the extraction pressures com-pared to this work. For modelling lavender oil extrac-tion the shrinking-core model could also be used [24], as well as the model developed by Žižović et al. [25].

Modified model [20] of Reverchon and Sesti Osseo model [21]

Reverchon and Sesti Osseo proposed a model for supercritical extraction of basil oil by carbon dioxide [21]. This model has been expressed by the following equation:

i100 1ttY e

− = −

(1)

where Y is the normalized yield of extract:

( )e max 100Y y y=

t is the extraction time, ti is the internal diffusion time, ye is the experimental value and ymax is the yield maximum value. In order to avoid the use of ti in Eq. (1), it was assumed that for a certain extraction system ti could be considered approximately as cons-tant, so the following expression has been derived:

Z at b= + (2)

where Z is the modified Y defined as:

( )ln 1 100Z Y= − , i1a t= −

and b is a correction term.


73

Equation (2) has been applied on our data of lavender flower extraction by supercritical carbon dioxide. The modification of Eq. (1) is then [20]:

( )( )m 100 1 at bY e += − (3)

where Ym normalized yield obtained from modelling:

mm

max

100y

Yy

=

Sovová model [23]

Sovová [23] proposed a series of simplified models that can be successfully used for preliminary modelling of the extraction of various plant materials, which are based on combining the characteristic time for individual periods/phases that can be observed during the entire extraction process, such as external mass transfer, internal mass transfer, the hypothetical equilibrium extraction excluding the resistance of mass transfer, i.e., combining the characteristic time of mass transfer in the fluid phase, tf, characteristic time of internal mass transfer, ti, characteristic time of extraction equilibrium, teq, and the mean residence time of the solvent in the extractor, tr [23].

Taking into account that lavender belongs to the Lamiaceae family, Sovová suggested the model for a preliminary simulation of the extraction of lavender oil, which includes a plug flow of the solvent through the extractor [23] as:

f

e u 1 11

for

1m

t Gy Gx t t

tK q e Θ

−= ≤ =

−

(4)

( )1

comb,ie u u 11 for t

t tty x G x e t

− − = − − >

Three adjustable parameters of the model are Km, G and ti.

The mass-related partition coefficient, Km, is the ratio of the equilibrium mass concentrations on the particle surface: the initial concentrations of solute in the fluid, y0, and solid phase, x0, and it is of the utmost importance when there is a solute-matrix interaction, as is the case with the extraction of lavender oil [23].

The value of G is closely associated with the degree of fragmentation of particles. Namely, assum-ing that the content of extractible substances in the plant at the time when the material is fed into the extractor is xu, the initial concentration of easily accessible oil in the broken (open) cells of the mate-rial is (xu–xk), and the initial concentration of oil within the whole (intact) particles is xk, and the share of oil in broken cells marked with G (G < 1), the fraction of oil

that is trapped in the intact cells, 1–G, is extracted much slower than those in open cells.

Characteristic time of internal mass transfer, ti, can be used for calculation of the intraparticle diffu-sivity (effective diffusivity), De:

( )2/32

e

1

15 i

R GD

t

−= (5)

The dimensionless mass transfer resistance in the fluid phase, Θf, represents the ratio of the charac-teristic time of mass transfer in the fluid phase, tf, and the fluid residence time in the extractor, tr [23]:

ff

r

tt

Θ = (6)

The characteristic time of mass transfer in the fluid phase, tf, was calculated based on the coefficient of mass transfer in the phase of supercritical CO2:

ff 0

tk a

ε= (7)

The coefficient of mass transfer in the super-critical fluid, kf, depends on the fluid speed and it is usually estimated based on the correlations between dimensionless numbers (Sherwood number, Sh, Rey-nolds number, Re, Schmidt number, Sc and Grashof number, Gr), such as proposed by King et al. (1997) [26]:

( ) 1/2

31 1.5 1 0.2548

ShRe

Scε + − = (8)

where the values of the Sherwood number, Sh, Reynolds number, Re, and Schmidt number, Sc, are given by the following equations:

f

12

k dSh

D= (9)

udRe

ρμ

= (10)

12

ScDμ

ρ= (11)

Due to the good transportation properties of supercritical carbon dioxide the characteristic time of mass transfer in the fluid phase, tf, is much lower than the characteristic time of internal mass transfer, ti. Still for small values of distribution coefficient Km, the characteristic time of mass transfer during extraction depends on tf and ti; it is obtained as a combination of these two times, i.e., the combined characteristic time


74

which replaces the time ti, can be calculated from the following equation:

fcomb,i i

r

1

m

tt t

t qK= + (12)

For sufficiently low values of Θf, tcomb,i = ti. Also, this simplification could be assumed if the dissolution of extract does not affect the internal diffusion rate [23]. Thus, the expression in the denominator of t1, in Eq. (4), becomes:

f

1

1 1e Θ−

− = (13)

Taking into account that:

( )e u 100Y y x=

for a sufficiently low Θf, the extraction yield according to the second model is:

11 m

100 for t G

Y G t tt K q

= ≤ = (14)

( )1

1100 1 1 for i

t ttY G e t t

−−

= − − >

To evaluate the model adjustable parameters (Km, G and ti), the model equation was fitted to vari-ous sets of experimental data by minimizing the dis-crepancies through the minimum sum of squares cri-terion.

EXPERIMENTAL

Plant material

Lavender was collected in full flowering in July 2011 from cultivated plants at the Institute of Field and Vegetable Crops Novi Sad, Department of Organic Production and Biodiversity-Bački Petrovac. Plant material was air-dried and the leaves, flowers and stems were separated and then milled. The milled and separated leaves, flowers and stems of lavender were used in this work.

Chemicals

Commercial carbon dioxide (Messer, Novi Sad, Serbia) was employed as the extracting agent. All other chemicals were of analytical reagent grade.

Determination of essential oil content

Essential oil content in the drug (flower and leaf) was determined by hydrodistillation according to procedure described in the literature [27], using a dis-

tillation apparatus defined in the German Pharmaco-poeia DAB-8 [28].

Determination of basic physicochemical properties of lavender essential oil

Relative density ( 2020d ) was determined by a

pycnometer (5.0 ml), as suggested by the European Pharmacopoeia [27], while for determination of ref-ractive indices ( 20

Dn ), an Atago RX-1000 digital refrac-tometer was used. Determination of optical rotation of essential oil samples was done on a Polamat A device. The specific rotation was calculated using the following equation:

( )D

read value

1.17543

t Vml

αα = (15)

where α is the read value of the measured optical rotation, V is the volume of the sample (cm3), l is the length of the cuvette (dm) and m is the mass of the sample (g).

Drug extraction by supercritical carbon dioxide

The supercritical fluid extraction (SFE) with carbon dioxide was performed using a laboratory-scale high-pressure extraction plant – HPEP (Nova-Swiss, Effretikon, Switzerland). The main part and characteristics (manufacturer specification) of the plant were as follows: diaphragm-type compressor (up to 1000 bar), extractor with an internal volume of 200 ml (pmax = 700 bar), separator with internal volume of 200 ml (pmax = 250 bar) and maximum CO2 mass flow rate of approximately 5.7 kg/h.

Lavender flower (30 g) with an average particle diameter, ds (ds = 0.58 mm), was extracted. The cumulative extraction yield was determined by mea-suring the mass of isolated extract after 0.5, 1.0, 1.5, 2.0, 2.5 and 3.0 h. Extraction was carried out under isothermal conditions at T = 313 K and at pressures of 10, 15, 20, 25 and 30 MPa, as well as at tempera-tures, T, of 323 and 333 K and pressures of 10, 20 and 30 MPa, while the flow of extractant was 0.194 kg/h in all cases.

According to the previously described procedure the extraction of lavender flower was performed at a pressure p = 10 MPa and temperature T = 313 K, with the exception that during the time intervals of the extraction process (0.5, 1.0, 1.5, 2.0, 2.5 and 3.0 h) appropriate CO2 extracts were taken (SFE-1, SFE-2, SFE-3, SFE-4, SFE-5 and SFE-6). These CO2 extracts were subjected to quantitative and qualitative analysis.

The dependence of lavender flower extraction yield from the on the solvent flow rate, w (w1 = 0.095


75

kg/h; w2 = 0.194 kg/h; w3 = 0.277 kg/h) at a pressure of 10 MPa and temperature 313 K was also exa-mined. Isolation of CO2 extract, in all cases, was per-formed at a pressure of 1.5 MPa and temperature 298 K.

Analytical gas chromatography (GC/FID)

Using GC/MS and GC/FID, the qualitative and quantitative composition of essential oil obtained by hydrodistillation of lavender flower and leaves was determined, and the samples were labelled as La-ec and La-el. In addition, the analysis of CO2 extracts obtained by extracting lavender flower at pressures of 10, 20 and 30 MPa and temperature 313 K was per-formed, while the samples were labelled as SFE-I, SFE-II and SFE-III. Samples of CO2 extracts obtained at 10 MPa and 323 and 333 K were labelled as SFE- -IV and SFE-V.

Sample preparation

Samples of essential oil and CO2 extracts were prepared by dissolving them in methylene chloride-methanol mixture (9:1, v/v), with an approximate con-centration of 20 mg/ml.

GC/FID analysis of the samples was carried out on an Agilent Technologies 7890A gas chromato-graph equipped with split-splitless injector and auto-matic liquid sampler (ALS), attached to HP-5 column (30 m×0.32 mm, 0.25 µm film thickness) and fitted to a flame ionisation detector (FID). Carrier gas flow rate (H2) was 1 ml/min, injector temperature was 250 °C, detector temperature 300 °C, while column tempera-ture was linearly programmed from 40–260 °C (at a rate of 4 °C/min), and held isothermally at 260 °C for the following 20 min. Solutions of lavender isolates were consecutively injected by ALS (1 µl, splitless mode). Area percent reports, obtained as result of standard processing of chromatograms, were used as the base for the quantification purposes.

Gas chromatography/mass spectrometry (GC/MS)

The same analytical conditions as those men-tioned for GC/FID were employed for GC/MS anal-ysis, along with column HP-5MS (30 m×0.25 mm, 0.25 µm film thickness), using an HP G1800C Series II GCD system (Hewlett-Packard, Palo Alto, CA, USA). Instead of hydrogen, helium was used as carrier gas. The transfer line was heated at 260 °C. Mass spectra were acquired in EI mode (70 eV), in m/z range 40– –450. Sample solutions were injected by ALS (1 µl, splitless mode).

The constituents were identified by comparison of their mass spectra to those from Wiley275 and NIST/NBS libraries, using different search engines. The experimental values for retention indices were

determined by the use of calibrated Automated Mass Spectral Deconvolution and Identification System software [29], compared to those from available lite-rature [30], and used as an additional tool to confirm MS findings.


The essential oil content was determined by hydrodistillation of lavender flowers and leaves, and found to be 2.39±0.015 and 0.52±0.011 ml/100 g drug, respectively. From the results, it can be seen that the essential oil content in the lavender flower is nearly 4.6 times higher than in the leaf. The essential oil of the flower has been isolated and its refractive index turned out to be 20

Dn = 1.462±0.0042, its rela-tive density 20

20d = 0.890±0.0044 and specific rotation 20D[ ]α = –7.0±0.065 (the values of physicochemical

parameters represent an average value of three measurements ± standard deviation).

Supecritical CO2 extraction of lavender flower

Effect of solvent flow on the extraction yield

Further testing was carried out by using an extraction procedure with supercritical CO2 as an extractant in order to obtain lavender flower extracts and investigate the extraction kinetics. Firstly, the influence of supercritical CO2 flow on extraction yield was examined. The following extractant flows were used: w1 = 0.095 kg/h, w2 = 0.195 kg/h and w3 = 0.277 kg/h, while the other extraction conditions were: p = = 10 MPa; T = 313 K and t = 3 h. The results are shown in Figure 1.

From the extraction kinetics results, it can be seen that the flow of the solvent has a notable inf-luence on the extraction yield for the specific con-sumption of CO2, with q values up to 10 kg CO2/kg plant material. The highest yield for an extraction time of 3 h was achieved at the flow w3 (5.56 g/100 g drug), and the lowest at the flow w1 (4.62 g/100 g drug). However, for that time, the solvent’s specific consumption values at flows w3 and w1 were about 28 and 10 kg CO2/kg plant material, respectively. At the solvent flow w2 the extraction yield for the extraction time of 3 h was 5.16 g/100 g drug, and q value was about 20 kg CO2/kg plant material, so this flow of supercritical CO2 was used in further tests.

Effect of pressure on the extraction yield

The extraction of lavender flowers was per-formed by supercritical CO2 to investigate the extrac-tion yield at different working pressures but under constant temperature (isothermal conditions, T = 313 K) during a 3 h extraction. Experimentally determined


76

values of the lavender extract yield versus extraction time, obtained at pressures 10, 15, 20, 25 and 30 MPa are presented in Figure 2.

With the increase of pressure from 10 to 30 MPa, at constant temperature (313 K), the extraction yield increased from 5.16 to 7.08 g/100 g drug, which is understandable, because with the increase of pres-sure the density of the extractant also increases and thus the polarity of the solvent or the dielectric cons-tant, i.e., the dissolving power of the extractant.

Effect of temperature on extraction yield

The influence of temperature (313, 323 and 333 K) at pressures of 10, 20 and 30 MPa was inves-tigated. The effect of temperature on the extraction yield of lavender extract at a flow rate of 0.194 kg/h of CO2 and at pressures of 10, 20 and 30 MPa versus extraction time is shown in Figures 3a–c, respectively.

From the results shown in Figure 3 it can be seen that at the pressure of 10 MPa the extraction yield decreased with the increase of temperature, due

Figure 2. The influence of pressure on extraction yield with supercrticial CO2 at 313 K. Symbols – experimental results (ye); lines – modelled values (ym) by the modified model [20].

Figure 1. Results of modelling the lavender flower extraction yield in dependence of the solvent flow rate (p = 10 MPa; T = 313 K). Symbols – experimental results (ye); lines – modelled values (ym) by the modified model [20].


77

(a)

(b)

(c)

Figure 3. The influence of temperature on extraction yield at different pressures: a) 10, b) 20 and c) 30 MPa. Symbols – experimental results; lines – modelled values (ym) by the modified model [20].


78

to the fact that mostly essential oils are being extracted at the examined pressure, so the vapour pressures of the diluted components have an insig-nificant effect on the extraction yield. However, at higher pressures (Figure 3) and especially the results shown in Figure 3c indicated that with the increase of pressure the solvent density also increased. This led to the extraction of other components (wax and resin), besides the essential oil, whose vapour pressures significantly affect the extraction yield based on the applied temperature. This temperature effect is parti-cularly expressed at the results shown in Figure 3c. At the pressure of 30 MPa and temperatures of 313, 323 and 333 K, it can be seen that the extraction yield was the highest at 333 K, and the lowest at 313 K because with the temperature increase the vapour pressure of diluted components increased, so the solubility depended on the equilibrium between the solvent density and changes of vapour pressure of the diluted components.

Kinetic modelling of the extraction system lavender flower-supercritical CO2

Based on the results of examining the kinetics of lavender flower extraction by supercritical CO2 at 10, 15 20, 25 and 30 MPa and at 313 K, the extraction system lavender flower - supercritical CO2 was modelled by applying the modified model [20]. The modelling results are shown in Figure 2.

Since the modified model equation includes the value Z (Z = ln (1–Y/100)), i.e., Z = at + b, Figure 4 shows the dependence between Z and t, from which parameters a and b were calculated (Table 1).

Table 1. Parameters in Eq. (2) and correlation coefficient, r

Pressure, MPa Parameters in Eq. (2)

a b |r|

10 -1.327 0.2061 0.988

15 -1.322 -0.107 0.996

20 -1.636 0.3162 0.966

25 -1.611 0.1665 0.991

30 -1.302 0.0649 0.992

High values of the correlation coefficient, r (from 0.966 to 0.992) show that there is a strong correlation between Z and t.

The effect of temperature on the extraction yield and the modelling results is shown in Figure 3. By applying the same procedure as in the previous case the necessary parameters for obtaining the value of ym were calculated. Based on the values of the influence of pressure on extraction yield under iso-thermal conditions, as well as the influence of tempe-rature on extraction yield, the values of mean relative deviation (MRD) were between 0.97 and 8.42 %, so it can be concluded that the applied modified model fits the experimental results of the extraction system lavender flower - supercritical CO2 relatively well.

The results of investigating the kinetics of laven-der flower - supercritical CO2 extraction system were also modelled by applying the model proposed by Sovová [23] to the extraction experimental results, i.e., the extraction yield as a function of time at pressures of 10, 15, 20, 25 and 30 MPa, temperature of 313 K and solvent flow rate 0.194 kg/h.

The physical parameters of supercritical CO2 under experimental conditions are given in Table 2.

Figure 4. Dependence of Z (Z = ln (1–Y/100)) on time.


79

The residence time of the solvent in the extrac-tor was calculated based on the value of the specific flow rate, porosity of the layer and density of both phases. The density of the solid phase was taken from literature data [24]. For the values of supercri-tical fluid density and coefficient of dynamic viscosity corresponding values for CO2 were taken. The den-sity of CO2 was estimated based on the Dohrn-Praus-nitz equation, using PE2000 software [31], and the coefficient of dynamic viscosity on the basis of Jossi, Stiel and Thodos method [32]. The dimensionless resistance to mass transfer in the supercritical fluid, Θf, was calculated on the basis of the coefficient of mass transfer in the phase of supercritical CO2. Binary diffusion coefficient of lavender oil in the car-bon dioxide, D12, was estimated as an average value based on the binary diffusion coefficient values for the most abundant components of lavender oil (camphor, fenchone, eucalyptol, campholenal, fenchol, cam-phene, thymol, myrtenol, furfuryl alcohol, linalyl ace-tate) and for the calculation the method proposed by Catchpole and King [33] was used. The self-diffusion coefficient of carbon dioxide is 4.944×10–8 m2 s-1 [32], while the critical molar volume of the solute was determined by a method proposed by Joback [32]. The values of mass transfer coefficient in the phase of supercritical CO2 were estimated based on the Sher-wood number values calculated from a correlation proposed by King et al. [26]. The characteristic dimension is the particle diameter.

During supercritical fluid extraction of any essential oil from its plant material, the essential oils and cuticular waxes are co-extracted at all experi-mental conditions; but if the extraction pressure is increased the contribution of the waxes in the extract will be more relevant, meaning that solubility of waxes will be increased and consequently its’ extraction kinetics will be changed compared to some other ope-

rating conditions. Nonetheless, to selectively extract the essential oil, for example, in order to investigate its extraction kinetics, extraction step must be fol-lowed by fractional separation. The essential oil con-tent in the plant material at the time when the material is fed into the extractor, xu, can be easily determined by hydrodistillation. Moreover, xu equals the essential oil yield maximum value, ymax. Therefore, the essen-tial oil extraction curves have the same asymptote for t approaching infinity regardless of different operating conditions.

In this work, the obtained lavender extracts were not additionally separated and contained, besides the lavender essential oil, a larger or smaller amount of co-extracted substances, depending on extraction pressure, temperature and extraction time. Following the previously adopted procedure for assessment of the essential oil yield maximum value, the yield maxi-mum value, ymax, could be determined as the asymp-tote of extraction curves for t approaching infinity. Thus, ymax becomes in fact an additional adjustable parameter of the model for extraction that depends on the extraction pressure and temperature; the value determined by this way is more reliable for expe-riments that last longer.

Due to consistency, the same way of calculation of the asymptotic yield was also applied in the Sovová model [23], in order to obtain the content of extrac-table substances in the plant, xu, to be pressure and temperature dependent parameter. The model adjust-able parameters (Km, G and ti) and xu were deter-mined by fitting the model in Eq. (4) to various sets of experimental data by minimizing the discrepancies through the minimum sum of squares criterion.

The results of modelling of the extraction of lavender flowers at 313 K and pressures 10, 15, 20, 25 and 30 MPa by the Sovová model [23] are pre-sented in Table 3. The comparison between the expe-

Table 2. Physical parameters of supercritical CO2 at various experimental conditions. Mass flow rate of CO2: 0.194 kg h–1

T / K p / MPa u×104 / m·s-1 ρ / kg·m-3 ν×108 / m2·s-1 Re D12×108 / m2·s-1 kf×106 / m·s-1

313 10 0.68 630 7.66 0.51 2.3 5.6

15 0.55 781 8.81 0.36 1.6 3.8

20 0.51 841 9.47 0.31 1.4 3.4

25 0.49 880 9.99 0.28 1.3 3.1

30 0.47 911 10.4 0.26 1.2 2.9

323 10 1.11 385 7.41 0.87 4.4 11.1

20 0.54 785 8.87 0.36 1.6 3.8

30 0.49 871 9.87 0.29 1.3 3.1

333 10 1.48 290 8.34 1.03 6.0 15.4

20 0.59 724 8.36 0.41 2.0 4.7

30 0.52 830 9.38 0.32 1.5 3.5


80

rimental extraction curves of the lavender extract at 313 K and pressures 10, 15, 20, 25 and 30 MPa and those obtained by the Sovová model [23] is shown in Figure 5. The average value of the effective internal diffusivity was calculated as De = 4.4×10–11 m2 s–1 (standard deviation of 1.57×10–11 m2 s–1) by using Eq. (5), and based on the characteristic time of internal mass transfer, ti, given in Table 3.

In literature [23], it could be found that the characteristic time of internal diffusion, the diffusion of the lavender oil through the wall of intact trichomes, was pressure and temperature dependent and varied between 123 min at (10 MPa, 308 K) and 500 min at (8 MPa, 323 K). In our work, this effect was not noti-ceable, mainly because the fact that the extract com-prised not only the oil but co-extractable substances too, as well as due to the way of xu determination. The

characteristic time of mass transfer during extraction depends only on the characteristic time of internal mass transfer (tcomb,i ≈ ti) and therefore instead of Eq. (4), Eq. (14) can be used for mathematical modelling.

The results of modelling of the extraction of lavender flowers at 323 and 333 K and pressures 10, 15, 20, 25 and 30 MPa by the Sovová model [23] are presented in Table 4. Comparison between the expe-rimental extraction curves of the lavender extract at 313 K and pressures 10, 15, 20, 25 and 30 MPa and those obtained by the Sovová model [23] is shown in Figure 5.

Values of the partition coefficient Km (Tables 3 and 4) varied between 0.08 at 20 MPa, 333 K and 0.18 at 15 MPa, 313 K. The literature values of the partition coefficient Km varied between 0.085 at 8 MPa, 323 K and 0.29 at 14 MPa, 323 K [25]. Accord-

Figure 5. Comparison between the experimental extraction curves of the lavender extract at 313 K and pressures 10, 15, 20, 25 and

30 MPa and those obtained by the Sovová model [23]. Symbols – experimental results; lines – calculated values by Eq. (4).

Table 3. The results of modelling the extraction of lavender at 313 K, 0.194 kg h–1 CO2 flow rate and different pressures. Model: Sovová [23]

Parameter p / MPa

10 15 20 25 30

xu×103 / kg ext. (kg matter)–1 57.8 57.5 59.3 64.0 84.6

Θf 0.013 0.015 0.016 0.017 0.017

G 0.40 0.50 0.50 0.54 0.62

Km 0.12 0.18 0.14 0.15 0.09

t1 / min 28 23 30 30 57

tcomb.i / min 77 62 101 62 145

ti / min 76 62 100 61 144

AARDa / % 1.42 1.10 6.30 3.93 1.49

MRDb / % 0.47 0.37 2.10 1.31 0.50

SDc / % 0.05 0.06 0.28 0.21 0.09

a , ,

,

100 e j m j

j e j

y yAARD

N NP y

−=

− ; b , ,

,

100 e j m j

j e j

y yMRD

N y

−= ;

c

2, ,( )

1

e j m jj

y y

SDN NP

−=

− −


81

ing to Sovová [23], as the glandular trichomes rup-tured not only during the pre-treatment of the plant but also when exposed to dense CO2, where the effect depends on the extraction conditions, the adjusted parameter G varied between 0.35 at 10 MPa, 308 K and 0.69 at 14 MPa, 323 K. In the present work, G varies from 0.39 (10 MPa, 333K) to 0.62 (30 MPa, 313 K). Therefore, the results of this preliminary eva-luation with the simplified model indicate that the model of essential oil extraction on micro-scale des-cribing the rupture of glandular trichomes in detail [25] would be appropriate to simulate these data.

Based on the AARD values (from 1.10 to 6.30%) it can be concluded that the Sovová model [23] fits the experimental results of the extraction system lavender flower-supercritical CO2 fairly well.

Isolation of CO2 extracts (SFE)

In order to isolate CO2 extracts during different time intervals of extraction, tests were carried out with lavender flower at 10 MPa and 313 K. SFE extracts were obtained at different time intervals during extrac-tion (0.5, 1.0, 1.5, 2.0, 2.5 and 3.0 h).

The highest yield of SFE extract was achieved in the first 0.5 h (2.39%). The yield of other SFE extracts is much lower and ranges between 0.09 and 0.57%. The first two extracts (SFE-1 and SFE-2) make up more than 75% of the total extract obtained during the 3 h extraction.

Qualitative and quantitative analysis of essential oil and SFE extracts by gas chromatography

Qualitative and quantitative analysis of essential oil and SFE extract samples were carried out using gas chromatography. The results are given in Table 5.

Table 5. GC analysis of essential oils and CO2 extracts of lavender flower; KIE = Kovats (retention) index experimentally determined [29] - uncorrected value; KIL = Kovats (retention) index - literature data [30]; n.i. = not identified; n/a = not available; tr. = traces (< 0.01%); La-el = leaf essential oil; La-ec = flower essential oil; SFE-1 (p = 10 MPa; T = 313 K; τ: 0.0-0.5 h); SFE-I (p = 10 MPa; T = 313 K; τ = 3 h); SFE-II (p = 20 MPa; T = 313 K; τ = 3 h); SFE-III (p = 30 MPa; T = 313 K; τ = 3 h); SFE-IV (p = 10 MPa; T = 323 K; τ = 3 h); SFE-V (p = 10 MPa; T = 333 K; τ = 3 h)

Component

Component content, mass%

KIE KIL Essential oila Supercritical fluid extracts (SFE)b

Sample label

La-el La-ec SFE-1 SFE-I SFE-II SFE-III SFE-IV SFE-V α-Pinene 0.08 0.06 tr. 0.04 0.07 0.07 0.04 0.04 924.0 932

Camphene 0.12 0.05 0.04 0.06 0.08 0.10 0.06 0.05 937.8 946 β-Pinene 0.41 0.43 0.31 0.24 0.26 0.31 0.24 0.24 966.0 974

3-Octanone 0.08 0.16 0.12 0.11 0.10 0.12 0.08 0.08 981.6 979

Myrcene 0.38 0.58 0.29 0.12 0.17 0.29 0.14 0.13 984.0 988

Butyl butanoate 0.05 0.09 0.03 0.08 0.05 0.04 0.06 0.06 991.4 993

Dehydroxy-cis-linalool oxide 0.08 0.24 0.07 0.16 0.09 0.07 0.11 0.11 1001.4 1006

Table 4. The results of modelling the extraction of lavender at 323 and 333 K, 0.194 kg h–1 CO2 flow rate and pressures 10, 20 and 30 MPa. Model: Sovová [23]

Parameter p / MPa

10 20 30

T / K 323 333 323 333 323 333

xu×103 / kg ext. (kg matter)–1 53.7 38.7 74.7 45.5 79.9 84.2

Θf 0.011 0.010 0.015 0.013 0.017 0.016

G 0.40 0.39 0.49 0.49 0.50 0.50

Km 0.11 0.14 0.11 0.08 0.10 0.12

t1 / min 30 23 40 52 42 35

tcomb.i / min 65 27 61 46 84 78

ti / min 64 27 60 45 82 77

AARD / % 3.03 1.24 3.64 5.44 4.48 2.27

MRD / % 1.01 0.62 1.21 0.80 1.49 0.76

SD / % 0.14 0.05 0.19 0.13 0.25 0.16


82

Table 5. Continued

Component


KIE KIL Essential Oila Supercritical fluid extracts (SFE)b

Sample label

La-el La-ec SFE-1 SFE-I SFE-II SFE-III SFE-IV SFE-V

Hexyl acetate 0.48 0.04 0.02 tr. tr. 0.04 0.01 0.01 1009.6 1007

p-Cymene 1.11 0.19 0.08 0.06 0.07 0.07 0.05 0.04 1016.7 1020

Limonene 0.69 0.35 0.17 0.13 0.10 0.14 0.11 0.10 1019.8 1024

1.8-Cineole 1.04 0.36 0.17 0.15 0.14 0.15 0.13 0.12 1021.7 1026

cis-β-Ocimene 0.25 0.55 0.82 0.65 0.36 0.54 0.48 0.55 1031.3 1032

trans-β-Ocimene 0.24 0.43 0.20 0.10 0.08 0.14 0.11 0.11 1041.4 1044

γ-Terpinene 0.06 0.06 0.02 0.02 tr. 0.03 0.03 0.01 1050.4 1054

cis-Sabinene hydrate 0.08 0.05 0.17 0.12 0.05 0.06 0.09 0.15 1061.0 1065

cis-Linalool oxide (furanoid) 0.66 1.58 1.22 1.11 1.15 1.24 1.15 1.10 1065.3 1067

Terpinolene 0.15 0.02 tr. 0.02 tr. tr. 0.02 0.01 1079.6 1086

trans-Linalool oxide (furanoid) 0.60 1.23 1.09 1.00 1.10 1.24 1.02 1.00 1081.3 1084

Linalool 25.75 48.44 31.48 28.25 25.24 25.64 29.55 29.76 1099.1 1095

1-Octen-3-yl acetate 0.48 0.54 0.44 0.47 0.34 0.34 0.48 0.43 1107.7 1110

cis-p-Menth-2-en-1-ol 0.35 0.10 0.07 0.09 0.18 0.16 0.10 0.09 1116.5 1118

Octan-3-ol acetate 0.05 0.03 0.05 0.05 0.06 0.08 0.04 0.04 1119.5 1120

trans-Pinocarveol 0.26 0.12 0.05 0.03 tr. 0.03 tr. 0.03 1132.4 1135

Camphor 2.35 0.43 0.25 0.25 0.22 0.24 0.29 0.24 1135.4 1141

Borneol 0.08 0.05 0.09 0.14 0.15 0.16 0.16 0.16 1160.0 1165

Lavandulol 11.11 4.85 4.97 2.91 3.13 3.76 3.30 3.35 1163.8 1165

Terpinen-4-ol 2.31 3.39 2.70 2.45 2.26 2.36 2.64 2.53 1170.5 1174

Cryptone 1.21 0.11 0.07 0.08 tr. 0.13 0.09 0.04 1179.7 1183

α-Terpineol 3.27 1.40 0.66 0.99 tr. 0.70 0.97 0.82 1186.0 1186

(3Z)-Hexenyl butanoate 1.66 2.89 1.08 1.00 2.70 2.02 1.21 1.10 1189.6 1187

Verbenone 1.07 0.29 0.14 tr. 0.13 0.16 0.14 0.14 1202.5 1204

cis-Sabinene hydrate acetate 0.05 0.14 0.07 0.06 0.19 0.18 0.07 0.06 1218.8 1219

Nerol 0.91 0.74 0.09 0.17 0.15 0.17 0.16 0.09 1225.3 1227

Hexyl 2-methyl butanoate 0.23 0.10 0.02 0.04 0.08 0.08 0.01 0.01 1230.1 1233

Cumin aldehyde 1.23 0.35 0.15 0.15 0.16 0.13 0.09 0.09 1235.7 1238

Carvone 0.54 0.13 0.09 0.08 0.11 0.06 tr. tr. 1239.4 1239

Linalool acetate 12.59 11.85 27.57 25.60 22.95 22.83 26.94 27.11 1252.3 1254

Dihydrolinalool acetate tr. tr. 0.08 0.04 0.06 0.10 0.04 0.05 1268.4 1272

Bornyl acetate 0.62 0.14 0.35 0.32 0.67 0.80 0.41 0.36 1277.2 1287

Lavandulyl acetate 4.49 5.34 2.53 4.68 3.89 3.20 4.91 4.75 1285.0 1288

p-Cymen-7-ol 0.41 0.14 0.11 0.15 0.22 0.21 0.15 0.17 1291.9 1289

Carvacrol 0.07 tr. 0.09 0.10 0.18 0.17 0.13 0.03 1306.0 1298

p-Vinylguaiacol 0.12 tr. 0.34 0.41 1.36 1.07 0.46 0.46 1311.3 1309

Hexyl tiglate 0.04 0.06 0.12 0.09 0.07 0.07 0.10 0.10 1324.9 1330

3-Oxo-p-menth-1-en-7-al 0.31 0.09 1.32 0.99 1.26 1.41 1.25 1.17 1330.1 1330

n.i. tr. 0.03 0.08 0.13 0.22 0.21 0.15 0.12 1341.3

n.i. 0.07 0.13 0.59 0.40 0.19 0.26 0.27 0.64 1345.9

n.i. 0.04 0.07 0.56 0.33 tr. 0.17 0.18 0.60 1349.1

Neryl acetate 0.03 tr. 0.09 0.24 0.27 0.12 0.20 0.23 1358.1 1359

α-Copaene 0.49 0.76 0.07 0.10 0.20 0.19 0.12 0.12 1364.9 1374

Geranyl acetate 1.11 1.45 0.35 0.24 0.24 0.42 0.40 0.44 1377.4 1379

n.i. 0.16 0.03 0.02 0.03 2.79 2.77 0.14 0.06 1393.1

n.i. 0.05 tr. 0.13 0.15 0.23 0.26 0.18 0.16


83

Table 5. Continued

Component


KIE KIL Essential Oila Supercritical fluid extracts (SFE)b

Sample label

La-el La-ec SFE-1 SFE-I SFE-II SFE-III SFE-IV SFE-V

n.i. 0.03 tr. 0.16 0.14 0.18 0.20 0.16 0.15

n.i. 0.04 0.02 0.12 0.11 0.13 0.13 0.13 0.13

trans-Caryophyllene 4.33 2.20 5.86 5.72 5.08 5.06 6.22 6.22 1408.5 1417

trans-α-Bergamotene 0.15 0.02 0.12 0.11 0.17 0.17 0.17 0.10 1425.3 1432

cis-β-Farnesene 0.13 tr. 0.21 0.19 0.28 0.40 0.33 0.21 1434.8 1440

α-Humulene 0.57 0.30 1.37 1.31 1.20 1.22 1.46 1.45 1441.9 1452

trans-β-Farnesene 0.14 0.10 0.29 0.23 0.33 0.40 0.26 0.27 1448.3 1454

Linalool isovalerate tr. 0.01 0.14 0.11 0.37 0.43 0.11 0.12 1457.9 1466

γ-Muurolene 0.34 0.16 0.53 0.70 0.99 0.94 0.75 0.73 1469.7 1478

trans.trans-α-Farneseneb tr. tr. 0.27 0.57 0.95 0.67 0.66 0.47 1498.1 1498

γ-Cadinene 0.18 0.02 0.29 0.56 0.93 0.71 0.61 0.46 1502.7 1513

Photosantalol 1.27 0.09 0.13 0.05 0.12 0.14 0.05 0.06 1510.9 1511

Elemol 0.19 0.01 0.06 0.09 0.31 tr. 0.12 0.08 1535.1 1548

n.i. 0.04 tr. 0.09 0.18 0.39 0.38 0.21 0.14

n.i. 0.53 0.26 0.24 0.24 0.37 0.35 0.28 0.23 1541.3

n.i. 0.03 tr. 0.13 0.17 0.20 0.19 0.19 0.18

Caryophyllene oxide 8.56 4.04 4.15 3.74 3.66 3.95 4.27 4.16 1572.7 1582

Humulene epoxide II 0.34 0.08 0.13 0.12 0.15 0.14 0.18 0.08 1597.7 1608

epi-α-Cadinol (τ-cadinol) 0.72 0.11 0.06 0.06 0.06 0.05 0.13 0.07 1631.7 1638

14-Hydroxy-(Z)-caryophyllene 0.58 0.17 0.07 0.03 0.05 0.04 0.13 0.03 1649.7 1666

Cadalene 0.82 0.41 0.32 0.31 0.33 0.38 0.38 0.35 1664.2 1675

Mustakone 0.21 0.03 0.03 tr. tr. 0.04 0.03 0.01 1669.1 1676

n.i. 0.15 0.10 0.33 0.28 0.33 0.38 0.36 0.31 1672.4

n.i. 0.25 0.03 0.05 0.13 0.16 0.11 0.14 0.13

n.i. 0.02 0.01 0.06 0.10 0.11 0.07 0.10 0.10

n.i. 0.04 0.01 0.13 0.10 0.05 0.10 0.09 0.14 1754.6 aEssenital oil obtained by hydrodistillation;

bextracts obtained by supercritical CO2;

ctentative identification

From the results shown in Table 5 it can be seen that essential oil and CO2 extracts obtained from the extraction of lavender flower at different pressures and temperatures have complex chemical compo-sitions. Components that are present in the highest percentage are: linalool (from 25.24 to 48.4%); lina-lool acetate (from 11.85 to 27.11%); lavandulol (from 2.91 to 11.11%); caryophyllene oxide (from 3.66 to 8.56%); lavandulyl acetate (from 3.20 to 4.91%); terpinen-4-ol (from 2.26 to 3.39%).

The essential oil of the flower contains a much higher percentage of linalool (48.44%) than the essential oil of leaf (25.75%), while its content in the extracts ranges between 25.24 and 29.76%. The lina-lool acetate content in essential oils is nearly the same (about 12%) and significantly higher in the extracts (from 22.83 to 27.11%). Esters are probably hydrolysed in the process of obtaining essential oil

from the drug by hydrodistillation, which has been reported in the literature [15, 24].

Furthermore, it can be seen that CO2 extracts contain a higher number of compounds than essential oils, especially the components with higher retention times, which are mainly present in waxes.

The analysis of SFE extracts (from SFE-1 to SFE-6) was performed. The content of dominant com-ponents in the extract SFE-1 is shown in Table 5 while the complete chemical composition other iso-lated extracts (from SFE-2 to SFE-6) is not shown in this work. On the other hand, from the change of dominant component content in the isolated extracts (Figure 6), it can be seen that their chemical com-position is similar to the extracts obtained at different pressured and temperatures.

The results shown in Figure 6 represent the extraction of lavender flower at different times of extraction (0.0–0.5; 0.5–1.0; 1.0–1.5; 1.5–2.0; 2.0–2.5;


84

2.5–3.0 h), where their qualitative and quantitative analysis was carried out by GC/MS and GC/FID. For example, for the extraction period 0.0–0.5 h the extract labelled SFE-1 was obtained, for the period 0.5–1.0 h extract SFE-2 and so on. Extract SFE-1 represents 60% of the total mass of all obtained extracts and extract SFE-2 about 14%, which means that over 70% of the extraction is completed in the first hour of the extraction process.

From the results shown in Figure 6 it can be seen that the content of linalool and linalool acetate in the first two CO2 extracts is relatively high, in contrast to their content in other extracts. Since the first two extracts represent, as noted earlier, about 75% of the total extract yield, it can be concluded that for an extraction time of one hour a high yield of CO2 extract is obtained with a relatively high content of dominant components (linalool and linalool acetate).

The content of other components in the extracts was practically the same.

Based on the results of extracts analysis by gas chromatography, it can be particularly concluded that the contents of components such as hydrocarbons (heptacosane, 2-methyloctacosane, nonacosane, untri-acontane) and hexadecanoic acid in the extracts constantly increase from SFE-1 to SFE-6.

CONCLUSION

Using the hydrodistillation procedure, the con-tent of essential oil in lavender flower was determined to be 2.39 ml/100 g drug, in contrast to the essential oil content in the leaf (0.52 ml/100 g drug).

By using GC/MS and GC/FID it was found that the examined samples (essential oils and SFE

extracts) contain high quantities of linalool (25.24– –48.44%), linalool acetate (11.85–27.57%), lavandulol (2.91–11.11%), caryophyllene oxide (3.66–8.56%) and others as main components.

Under isothermal conditions (T = 313 K), the extraction yield increases with the increase of pres-sure, which is in accordance with the increase of the dissolution power of the extractant. The highest yield was obtained with CO2 at a pressure of 30 MPa (7.08 mass%). At higher pressures, especially at 30 MPa, the extraction yield was the highest at 333 K and the lowest at 313 K, because vapour pressure inceased with the increase of temperature; thus, the solubility was determined by the equilibrium between the sol-vent density and the changes of vapour pressure of diluted components.

Based on the results of the extraction of laven-der flower and obtained individual extracts for diffe-rent periods of extraction, it was concluded that after 1 h of extraction, an extract was obtained with a high yield (about 75%) and with almost the same quali-tative and quantitative composition compared to the yield after 3 h of extraction. These results are of inte-rest in practice.

Based on the MRD (%) values (from 0.37 to 2.10 and from 0.97 to 8.42) for the Sovová model and the modified model respectively, it can be concluded that the applied models fit the experimental results of the extraction system lavender flower-supercritical CO2 well.

Acknowledgement

This work was supported by the Integrated and Interdisciplinary Research Project No. 173021.

Figure 6. The change of dominant component content in CO2 extracts obtained by extracting lavender flower with supercritical carbon dioxide (p =10 MPa; T = 313 K).


85

Nomenclature

a0 specific surface area, m2 m–3 a the adjustable parameter of Eq. (2), s–1 b the adjustable parameter of Eq. (2) dp average particle diameter, m ds average particle diameter, mm D12 binary diffusion coefficient of lavender oil in the

carbon dioxide, m2 s–1 De effective intraparticle diffusion coefficient, m2 s–1 G initial fraction of extract in open cells, the

adjustable parameter of Eq. (4), kf mass transfer coefficient in fluid phase (external

coefficient), m2 s–1

Km mass-related partition coefficient, the adjust-able parameter of Eq. (4), kg plant (kg solvent)–1

MRD mean relative deviation N number of experimental points at one extrac-

tion curve NP number of adjustable parameters in a model q specific flow rate, kg·(kg plant)–1·s–1 R spherical particle radius, m r correlation coefficient t extraction time, s t1 time of the end of the first extraction period, s tcomb,i combined characteristic time of mass transfer –

substitute for ti, s teq characteristic time of equilibrium extraction, s tf characteristic time of the fluid phase mass

transfer, s ti characteristic time of the solid phase mass

transfer (internal diffusion), s tr residence time, s u superficial velocity, m s–1

w solvent mass flow rate, kg s–1

x0 initial solid phase concentration, kg·(kg plant)–1 xk the initial concentration of solute within the

whole (intact) particles, kg·(kg plant)−1 xt transition concentration, kg·(kg plant)–1 xu concentration of extract in the plant before

extraction, kg·(kg plant)–1 Y (=ye/ymax) 100 normalized yield of extract Ym normalized yield obtained from modelling, Eq.

(3) ye extraction yield, kg·(kg plant)–1

ymax the maximum yield, kg·(kg plant)–1

ym (= ymaxYm/100) extraction yield obtained from the Ym

y0 initial fluid phase concentration, kg·(kg sol-vent)–1

ysat solubility, kg·(kg solvent)–1

Z (= ln(1-Y/100)) variable (transformed norma-lized yield of extract)

ε free void

Θf (= tf/tr) dimensionless external mass transfer resistance

ρ solvent density, kg m–3 μ solvent viscosity, Pa s A cross sectional area (m2) D diameter of column (m) DR inner diameter of riser (m) DD inner diameter of downcomer (m) kLa volumetric mass transfer coefficient (1/s) L distance between the conductivity sensors (m) H height (m) NG gas power output (W) Δp pressure drop (bar) tc liquid circulation time (s) tm mixing time (s) t time (s) UG superficial gas velocity, column based (m/s) WLD downcomer interstitial liquid velocity (m/s)

Abbreviations

ALR airlift reactor BC bubble column DT-ALR draft tube airlift reactor EL-ALR external loop airlift reactor rps rotations per second

Greek letters

εG gas holdup

Subscripts

C circulation D downcomer DT draft tube G gas phase L liquid phase R riser S separation zone

REFERENCES

[1] H. Afshari, A. G. Ebad, Am. J. Sci. Res. 18 (2011) 13-17

[2] R.R.S. Nelson, J. Antimicrob. Chemother. 40 (1997) 305– –306

[3] S. Kulevanova, G. Stetkov, M. Ristic, Bull. Chem. Tech-nol. Maced. 19 (2000) 165-169

[4] A.R. Fakhari, P. Salehi, R. Heydari, S. N. Ebrahimi, P.R. Haddad, J. Chromatogr., A 1098 (2005) 14–18

[5] H.M.A. Cavanagh, J.M. Wilkinson, Phytother. Res. 16 (2002) 301–308

[6] A. Prasher, I.C. Locke, C.S. Evans, Cell Proliferat. 37 (2004) 221–299

[7] S.S. Denner, Holist. Nurs. Pract. 23 (2009) 57–64

[8] T. Atsumi, K. Tonosaki, Psychiatry Res. 150 (2007) 89–96

[9] M. Lis-Balchin, S. Hart, Phytother. Res. 13 (1999) 540– –542


86

[10] M. Lis-Balchin, S.G. Deans, E. Eaglesham, Flavour Frag. J. 13 (1998) 98–104

[11] D. Komes, A. Belščak-Cvitanović, D. Horžić, G. Rusak, S. Likić, M. Berendika, Phytochem. Anal. 22 (2011) 172–180

[12] B. Blažeković, S. Vladimir-Knežević, A. Brantner, M. Bival Štefan, Molecules 15 (2010) 5971-5987

[13] T. Lua, F. Gaspar, R. Marriott b, S. Mellorc, C. Watkin-sond, B. Al-Duria, J. Sevillea, R. Santos, J. Supercrit. Fluids 41 (2007) 68–73

[14] Z. Zeković, Ž. Lepojević, A. Tolić, Sci. Tech. 36 (2001) 3459-3472

[15] E. Reverchon, G. Della Porta, F. Senatore, J. Agr. Food Chem. 43 (1995) 1654-1658

[16] K. Ghafoor, F. Al-Juhaimi, Y. Hee Choi, Pak. J. Bot. 43 (2011) 1581-1586

[17] D. Skala, I. Žižović, S. Gavrančić, Hem. Ind. 56 (2002) 179-190

[18] B.M. Damjanović, D. Skala, D. Petrović-Đakov, J. Baras, J. Essent. Oil Res. 15 (2003) 91-93

[19] E. Reverchon, I. de Marco, J. Supercrit. Fluids 38 (2006) 144-166

[20] B. Pekić, A. Tolić, Ž. Lepojević, Z. Zeković, Application of our modification of the Reverchon-Sesti Osseo equation to the supercritical extraction of chamomile flower by carbon dioxide, in Proceedings of the 1st International Conference of the Chemical Societies of the South-East European Countries, Chemical Sciences and Industry, Halkidiki, Greece, 1998, p. 654

[21] E. Reverchon, L. Sesti Osseo, Modeling the Supercritical Extraction of Basil Oil, in Proceedings of the 3rd Inter-national Symposium on Supercritical Fluids, Perrut and Brunner, Strasbourg, Austria, 1994, p.189

[22] E. Reverchon, G. Della Porta, F. Senatore, J. Agr. Food Chem. 43 (1995) 1654

[23] H. Sovová, J. Supercrit. Fluids 66 (2012) 73-79

[24] M. Akgün, N. Akgün, S. Dinçer, Ind. Eng. Chem. Res. 39 (2000) 473–477

[25] I. Žižović, M. Stamenić, A. Orlović,. D. Skala, Chem. Eng. Sci. 60 (2005) 6747–6756

[26] J.W. King, M. Cygnarowicz-Provst, F. Favati, Ital. J. Food Sci. 9 (1997) 193-204

[27] European Pharacopoeia (Ph. Eur. 5.0.), Council of Europe, France, Strasbourg, 2005

[28] Deutsches Arzneibuch 8 (DAB 8), Ausgabe und Nacht-räge, Deutscher Apotheker Verlag. Stuttgart, Govi Verlag, GmbH, Frankfurt/Main, 1987

[29] W.G. Mallard, J. Reed, Automated Mass Spectral Decon-volution and Identification System software (AMDIS ver.2.1.), National Institute of Standards and Technology (NIST), Standard Reference Data Program, MD 20899, Gaithersburg, USA, 1997

[30] R.P. Adams, Identification of Essential Oil Components by Gas Chromatography / Mass Spectrometry, 4th ed., Allured Publishing Corporation, Carol Stream, Illinois, USA, 2007

[31] O. Phohl, S. Petkov, G. Brunner, PE2000-A powerful tool to correlate phase equilibria, Herbert Utz Verlog GmbH, München, 2000

[32] R.C. Ried, J.M. Prausnitz, B.E. Poling, The properties of gases & liquids, 4th ed., McGraw-Hill Book Company, New York, 1987, p. 577

[33] O.J. Catchpole, M.B. King, Ind. Eng. Chem. Res. 33 (1994) 1828-1837.

VESNA NAĐALIN1

ŽIKA LEPOJEVIĆ2

MIHAILO RISTIĆ3

JELENA VLADIĆ2

BRANISLAVA NIKOLOVSKI2

DUŠAN ADAMOVIĆ4

1Visoka tehnička škola strukovnih studija u Zrenjaninu, Zrenjanin, Srbija

2Univerzitet u Novom Sadu, Tehnološki Fakultet, Novi Sad, Srbija

3Institut za proučavanje lekovitog bilja “Dr Josif Pančić”, Beograd, Srbija

4Institut za ratarstvo i povrtarstvo, Novi Sad, Srbija

NAUČNI RAD

ISPITIVANJE EKSTRAKCIJE I EKSTRAKATA GAJENE LAVANDE (Lavandula officinalis L.)

U radu je postupkom hidrodestilacije izvršeno određivanje sadržaja etarskog ulja u cvetu i

listu lavande. Izolovanom etarskom ulju određene su fizičko-hemijske karakteristike.

Korišćenjem CO2 u superkritičnom stanju izvršena je ekstrakcija cveta lavande odabranim

protokom rastvarača, primenom izotermnog i izobarnog postupka. Primenom gasne hro-

matografije sa masenom spektrometrijom (GC/MS) i gasne hromatografije sa plameno

jonizujućim detektorom (GC/FID), izvršena je kvantitativna i kvalitativna analiza etarskog

ulja i superkritičnih ekstrakata (SFE). Takođe je izvršena i analiza pojedinačnih SFE

ekstrakata dobijenih tokom različitih perioda ekstrakcije. Nađeno je da su glavne kom-

ponente analiziranih uzoraka: linalool, linalool-acetat, lavandulol, kariofilen-oksid, lavan-

dulil-acetat, terpinen-4-ol i dr. Korišćenjem odabranih modela izvršeno je modelovanje

ekstrakcije sistema cvet lavande – superkritični CO2 na osnovu eksperimentalnih rezultata

ispitivanja kinetike ekstrakcije. Odabrani modeli dobro fituju eksperimentalne rezultate.

Ključne reči: lavanda, ekstrakcija, ekstrakti, modelovanje, superkritični CO2.





87

MAJID MAZHAR1

MAJID ABDOUSS1

ZAHRA SHARIATINIA1

MOJDEH ZARGARAN2 1Department of Chemistry,

Amirkabir University of Technology (Polytechnic), Tehran, Iran

2Building and Housing Research Center (BHRC), Tehran, Iran

SCIENTIFIC PAPER

UDC 677.494.742.017:66.095.26-922.3

DOI 10.2298/CICEQ120428104M

GRAFT COPOLYMERIZATION OF METHACRYLIC ACID MONOMERS ONTO POLYPROPYLENE FIBERS

Abstract

In recent years, graft copolymerization has been widely used to insert various functional groups onto polymers. In our study, methacrylic acid monomers are grafted to polypropylene (PP) fibers to make them hydrophilic while main-taining their mechanical properties. Experiments are designed based on Taguchi method and influence of temperature, monomer concentration; cross-linker concentration and time of reaction are investigated. Grafting of methac-rylic acid and divinyl benzene is investigated by FTIR spectra and confirmed by SEM micrographs. Tensile strength and toughness of specimens are mea-sured and compared with raw fibers. The effects of grafting on the tensile strength and toughness of fibers are measured and the critical grafting degree to maintain tensile strength and toughness of fibers is defined.

Keywords: mechanical properties, PP fiber, Taguchi method, graft copolymerization, SEM..

Polypropylene (PP) fibers have been used for valuable purposes in different areas like textile, con-crete and some chemical industries [1]. Despite their noticeable properties such as chemical and thermal stability, stiffness, low density, impact resistance and low material cost, PP fibers are restricted in their applications because of the lack of chemical function-alities and nonpolarity [2]. For example, they cannot be used under conditions that require high tempera-tures. These drawbacks can be removed by effecting functionalization of the backbone polymer through grafting [3]. When the interaction with other materials is important, a modification of these polyolefins should be carried out [4]. Modification of different types of polyolefins with polar groups by various methods which includes graft-copolymerization method has been studied almost comprehensively [3,5-8]. The chemistry associated with specific functional groups may change the properties of the graft-modified poly-mer and potentially improve its usability in applica-tions requiring reactivity, paint-ability, adhesion, stab-

Correspondence: M. Abdouss and Z. Shariatinia, Department of Chemistry, Amirkabir University of Technology (Polytechnic), P.O. Box 159163-4311, Tehran, Iran. E-mail: [email protected] (M. Abdouss), [email protected] (Z. Shariatinia) Paper received: 28 April, 2012 Paper revised: 12 August, 2012 Paper accepted: 2 November, 2012

ility or impact strength [9]. Graft copolymerization can be performed by various methods such as grafting initiated by chemical means [10,11], grafting initiated by photoradiation [12,13], γ-irradiation [14], thermal [15] and enzymatic grafting [16,17]. Free radical graft-ing of monomers is one of the most attractive ways for the chemical modification of polymers. It involves the reaction between a polymer and a vinyl-containing monomer, which is able to form grafts onto the poly-mer backbone in the presence of free radical gene-rating chemicals, such as peroxides [18,19]. As some literature confirms, grafting of the fibers with high grafting degree can weaken their strength in tension [20,21]. But there is not a comprehensive study on the influence of chemical grafting on PP fibers mecha-nical properties. It is notable that grafting of methac-rylic acid onto PP fibers has been performed by the radiation-induced method [22]. In this research, influ-ences of temperature, monomer concentration, cross-linker concentration and reaction time on grafting degree of PP fibers are studied based on the Taguchi method, while effect of grafting on tensile strength and toughness of PP fibers are also investigated. The Taguchi method is a statistical method for experiment designing [23].

M. MAZHAR et al.: GRAFT COPOLYMERIZATION OF METHACRYLIC ACID… Chem. Ind. Chem. Eng. Q. 20 (1) 87−96 (2014)

88


Materials

The PP fiber used was prepared commercially and characterized as follow: tensile strength 350 MPa, 1111 dtex, cross-sectional area of 0.12 mm2 and specific gravity of 0.91 g/cm3. Methacrylic acid (MAA) and benzoyl peroxide (BPO) were both pur-chased from Merck. BPO was recrystallized and then used. MAA was distilled in vacuum at 88 ºC before use. Divinyl benzene (DVB) was utilized as a cross-linker and toluene was used as a solvent (both pur-chased from Merck).

Scanning electron microscopy (SEM) micro-graphs were obtained using a Philips XL30 instru-ment, under vacuum, accelerated at 30 kV. Infrared (FT-IR) spectra were recorded on a Shimadzu spec-trometer. Elemental analysis was performed using a Heraeus CHN-O-RAPID apparatus. The tensile strengths were measured using an Amsler Universal Tension Machine (Germany).

Design of experiments

The experiments are designed by the Taguchi method [23], which is an accepted statistic method for heavily reducing the number of required experiments to investigate the influence of some parameters on the result of a specific reaction. It was selected in this paper to overcome the multiplicity of the required experiments. Without the Taguchi method, it is neces-sary to perform 44 (256) experiments to fully inves-tigate the influence of 4 factors at 4 levels on the

grafting reaction, while just 16 experiments are required to study it in an acceptable way. An L16–array was selected as the appropriate array for this research. This array was modified to four factors at four levels by Qualitek 4 software [24]. To investigate the influence of four factors (temperature, monomer concentration, cross-linker concentration and time) at four levels (Table 1) on the grafting degree, the expe-riments were designed as shown in Table 2.

Table 1. Levels of factors

No. Factor Level

1 2 3 4

1 Temperature, °C 25 30 35 40

2 MAA conc., mass%a 25 40 55 70

3 DVB conc., mass%a 0 5 10 20

4 Time, h 1 2 3 4 aRefers to weight percentage which is calculated as (the component

weight) / (total weight of the solution)

Procedure of graft-copolymerization

To remove any chemicals adsorbed on the fiber surface, the fibers were washed with acetone, two times for 3 min and dried in an oven at 30 °C for 5 h before initial mass was determined. The initiator BPO was dissolved in a small amount of toluene and this solution was mixed with a solution of monomer MAA (0.5 mol%) in 50 ml toluene. The mixture was purged by nitrogen gas for 5 min and then dropped in fibers glass flask under nitrogen atmosphere. Then spe-cified amounts of DVB were added to the mixture.

Table 2.Experimental conditions

Trial no. Temperature, °C MAA conc., mass%a DVB conc., mass%a Time, h Degree of grafting, %

1 25 25 0 1 2.22

2 25 40 5 2 5.33

3 25 55 10 2 11.13

4 25 70 20 4 12.67

5 30 25 5 3 7.13

6 30 40 0 4 8.78

7 30 55 20 1 8.33

8 30 70 10 2 9.67

9 35 25 10 4 9.67

10 35 40 20 3 11.13

11 35 55 0 2 8.67

12 35 70 5 1 8.33

13 40 25 20 2 10.13

14 40 40 10 1 9.78

15 40 55 5 4 15.33

16 40 70 0 3 14.67aRefers to weight percentage which is calculated as (the component weight) / (total weight of the solution)


89

The flask was placed in an oil bath at various tem-peratures and was shaken for specified times with a shaking speed of 150 rpm. MAA-grafted PP fibers were removed from the reaction flask and washed with a solution of 5% NaOH in water/ethanol (50:50, v/v) and extracted using methanol as the solvent with a soxhlet extractor for 24 h to remove any residual monomer and homo polymers. The fibers were then washed with large amounts of water, dried and weighed. Degrees of grafting (Dg) were determined using the equation Dg (%) = [(W1 – W0)/W0]×100, in which W1 is the weight of grafted fibers and W0 is the initial weight.


Grafting results

Experiments were performed according to the above procedure and conditions were controlled as mentioned in Table 2. The results (Table 2) were interpreted by Qualitek 4 software. Influences of four different factors on the grafting reaction are presented in the following sections.

Temperature effect

Interpreting the experimental results by Qualitek 4 software revealed the effects of factors on the grafting reaction. For temperature efficacy (Figure 1), it can be concluded that high temperatures lead to an increment in grafting degrees. Also, the slope of the curve is increased by raising temperature. This behavior indicates that to achieve high grafting deg-

rees, it is essential to use higher temperatures. Rais-ing temperature increases the number of free radicals generated by the thermo-decomposition of the initia-tor, which results in more active sites on the fiber sur-face. Another effect is that high temperatures enhance the diffusion of the monomer to the active site for polymerization. This procedure certainly has a limitation and increasing temperature partly can increase the grafting degree, because higher tem-peratures cause the initiator to decompose rapidly and increase the number of generated radicals, lead-ing to the formation of homopolymers, which surely will decrease expected amount of copolymer and grafting degree. The maximum effect of temperature on grafting degree in the studied interval was achieved at 40 °C (level 4 for temperature).

Monomer concentration effect

Interpretation of grafting results by the software revealed the monomer concentration effect on the graft copolymerization as shown in Figure 2. Raising the monomer concentration increases the grafting degree but the slope of curve logically decreases in high monomer concentrations. High monomer con-centration results in high viscosity of solution because the gel effect might be generated in homogenous reaction systems. Thus, the diffusion of monomers from bulk solution to fiber surface is prevented by high viscosity, which results in a decrease of the curve slope and could even decrease the grafting yield. The optimum effect of monomer concentration is achieved with 70% w/w monomer concentration.

Figure 1. Effect of temperature levels on grafting degrees.


90

Cross-linker concentration effect

As mentioned, DVB was used as a cross-linker in this work. Interpretation of the grafting results by the software revealed the DVB concentration effect on the graft copolymerization as shown in Figure 3. DVB has a slight influence on grafting yield, but raising its concentration relatively increases the grafting degree. The presence of DVB in the reaction environment branches side-chains, which results in more grafting of monomers. DVB has two vinyl groups and can participate in the reaction with both of

these groups. When doing so, it makes the side-chains to generate and branch. If produced branches reach to other side-chains, DVB acts as cross-linker and if not, it acts to branch the side chains. In the latter case, because of multiplicity of growing branches, more monomers can attach to the growing chain, as Figure 4 demonstrates.

Reaction time effect

Using the grafting results, Qualitek 4 software interpreted the effect of reaction time on grafting degree (Figure 5). The reaction time has a remark-

Figure 2. Effect of monomer concentration levels on grafting degrees.

Figure 3. Effect of cross-linker concentration levels on grafting degrees.


91

able effect on grafting yield. With raising the reaction time, amounts of monomers that are grafted to the fibers are increased. The slope of the curve tends to decrease by raising the time of reaction. Graft copoly-merization is a relatively slow reaction. Whenever required amounts of the substances are added to the reaction environment, they need time to react with each other. Therefore, having more time for the reac-tion results in more attachment of monomers to the growing side and more grafting yield. Obviously, after elapsing appropriate time and consuming reactants, the rate of grafting decreases and the grafting reac-tion tends to stop and be completed. The appropriate time to achieve higher grafting degrees in the studied interval was determined as 4 h.

Optimum condition

The software determined the main effects of levels of each factors on grafting degrees (Table 3). Main effect of each level represents contribution of that level to the grafting degree. Greater main effects

C

CH3

O

O n

CH3 CH3

H

H

PP-chain

side-chain grafted to PP-chain

main side-chain

DVB branching mechanism

Figure 4. Structure of synthesized copolymer including DVB branching mechanism and vibrational modes of some important bonds in FTIR spectroscopy.

Figure 5. Effect of the reaction time levels on grafting degrees.

Table 3. Main effects (contributions) of levels of the factors to the grafting degree

No. Factor Level

1 2 3 4

1 Temperature, °C 7.837 8.477 9.449 12.477

2 MAA conc., mass%a 7.287 8.755 10.864 11.335

3 DVB conc., mass%a 8.585 9.03 10.062 10.565

4 Time, h 7.164 8.449 11.014 11.612aRefers to weight percentage which is calculated as (the component

weight) / (total weight of the solution)


92

result in greater grafting degrees. From this table, it can be inferred that the level four of each factor has more effect in increasing the grafting yield. Further-more, if the optimum condition refers to the highest amount of grafting, these levels could be recom-mended. The expected grafting degree at optimum condition was estimated by the software as 17.305%. It should be considered that the reaction was opti-mized with 16 experiments based on the Taguchi method instead of 256 required experiments by vary-ing each of the parameters. The reported maximum degree of grafting is for the conditions studied in this research. In fact, achieving high grafting yield was not our purpose in this work, because high grafting yields result in decrement of mechanical properties of the fibers. Accordingly, moderate conditions were sel-ected here. Obviously, if greater amounts were sel-ected for the factors, grafting degrees became higher than those achieved.

Verification of graft copolymerization

To confirm that grafting is performed, FT-IR spectra of three specimens were investigated. Figu-

res 6–8 are related to the raw PP fiber, specimen (16) grafted by methacrylic acid monomers without cross-linker (DVB) and specimen (13) grafted by methac-rylic acid with contribution of DVB as cross-linker, respectively. The peak at 1716 cm–1 in Figure 7 refers to the stretching of the C=O group in MAA grafted to PP fiber. Due to the stretching of the O–H bond, a distinct peak is seen at about 3400 cm–1 in Figure 7. There is a peak at 1255 cm–1 and two peaks around 950–990 cm–1 that can be attributed to the stretching of the C–O bond and out of plane vibration of the O–H bond in MAA grafted to the PP fiber, respectively. These peaks confirm the grafting of MAA to the PP fiber. In Figure 8, there is a group of peaks around 1500–1650 cm–1 that refers to the C=C bond and another group around 3100–3200 cm–1 that is attri-buted to the =C–H group, whose both bonds are in DVB monomers that are grafted to PP fibers. These peaks verify DVB contribution in the graft copolymeri-zation. Some important vibrational modes are demon-strated in Figures 7 and 8.

Comparison of SEM micrographs of raw and modified fibers confirms that graft copolymerization

Figure 6. FTIR Spectrum of raw PP fiber.


93

Figure 7. FTIR Spectrum of MAA-g-PP.

Figure 8. FTIR Spectrum of MAA and DVB grafted to PP fiber.


94

has been occurred. The SEM micrographs of the modified fibers exhibit varying degrees of grafting (Figure 9). White solids on the fibers are increasing by raising grafting degrees which can verify occur-rence of grafting reaction.

Mechanical properties of fibers

As mentioned before, grafting affects the mechanical properties of fibers. To investigate this effect, tensile strength and toughness of raw and modified fibers were examined. Tensile strength tests were repeated five times for each specimen and the average of maximum points of load-displacement curves per fibers cross-sectional area were consi-dered as their average ultimate tensile strength. To realize grafting effect on the tensile strengths of fibers, average ultimate tensile strengths of the fibers were plotted against grafting degrees in Figure 10. Tensile strength values of the fibers grafted under 5.33% are very close to the value of raw fibers. It

shows that grafting below 5.33% does not affect sig-nificantly the fibers mechanical properties. Conse-quently, 5.33% in a strict conclusion can be defined as critical point to graft PP fibers with methacrylic acid monomers in the mentioned conditions with maintain-ing their tensile strength. On the other hand, in the interval of 5–9%, the decrease in tensile strength is negligible too. Thus, to achieve high grafted fibers, little decrement in their mechanical properties in some utilizations can be acceptable. With this attitude, about 9% grafting can be suggested as the critical point. This behavior demonstrates that grafting of PP fibers more than 9% results in significant loss in ten-sile strength of them and makes them inappropriate for applications such as FRC’s which need fibers with high tensile strengths.

The toughness of fibers, like their tensile strength, can be affected by grafting. Toughness is the ability of a material to absorb energy and plastically deform without fracturing. The area under the stress-strain

Figure 9. SEM Micrographs: a) raw PP fiber ,b) specimen 7 with grafting degree of 8.33% and c) specimen 15 with grafting degree of 15.33%.

Figure 10. Effect of grafting reaction on tensile strength of the fibers.


95

curve of fibers represents their toughness. To realize the effect of grafting on toughness of fibers, their average toughness is plotted versus grafting degrees (Figure 11). Toughness of the fibers obviously decreases by increasing the grafting degrees. This

reduction is more significant after 9% grafting. This procedure reveals that to protect fibers from remark-able decrement in toughness, it is essential to keep degree of grafting below 9%.

Considering the tensile strength and toughness of the fibers demonstrates that 9% grafting of fibers can be a critical point. In other words, in this research, 9% grafting by hydrophilic monomers can be defined as a critical point to improve PP fibers hydrophilicity with protecting their tensile strength and toughness, approximately. Justifications to decrement in mecha-nical properties of modified fibers needs more researches to be decisive but possibly, it can occur because of attacks of radicals generated from pero-xides to PP chains, which results in rupturing of the chains and decreasing tensile strengths and tough-ness of fibers. In addition, another possibility comes from electrostatic and steric repulsions between side chains, which results in creation of tensile residual stresses on PP chains and therefore, decreases the tensile strength of the fiber. Moreover, a combination of both supposed pathways could occur and affect the mechanical properties of the fibers. The mechanical properties data can be almost a confirmation for the latter assumption because the most loss in mecha-nical properties occurs in grafting degrees greater than 9%, which lead to high concentration of radicals and repulsion between side chains due to high graft-

ing degree that causes tensile residual stresses to be significant in these fibers. However, as mentioned, further investigation is needed to comprehend this behavior.

CONCLUSIONS

In this work, the effects of temperature, mono-mer concentration, cross-linker concentration and reaction time on graft copolymerization of MAA in moderate conditions were studied and the optimum conditions to achieve maximum grafting degree were determined. The effects of grafting on the mechanical properties of PP fibers were also investigated and it was concluded that grafting decreases tensile strength and toughness of fibers. This decrement is remark-ably higher after 9% of grafting and this point can be approximately defined as a critical point to make PP fibers hydrophilic with maintaining their mechanical properties.

Acknowledgements

The financial support of this work by the Research Councils of Amirkabir University of Tech-nology (Polytechnic) and Building and Housing Research Center (BHRC) is gratefully acknowledged.

REFERENCES

[1] A.M. Alhozaimy, P. Soroushian, F. Mirza, Cem. Concr. Compos. 8 (1996) 85-92

[2] K. Cao, Z.Shen, Z.Yao, B. Qu, X. Pang, Z. Lu, Y. Li, Z. Chen, Chem. Eng. Sci. 65 (2010) 1621–1626

Figure 11. Effect of grafting reaction on toughness of the fibers.


96

[3] I. Kaur, B.N. Misra, S. Kumar, J. Appl. Polym. Sci. 69 (1998) 143-152

[4] K. Yamamoto, H. Tanaka, M. Sakaguchi, S. Shimada, Polymer 44 (2003) 7661-7669

[5] E. Passaglia, P. Sicialiano, F. Ciardelli, G. Maschio, Poly-mer Int. 49 (2000) 949-952

[6] J.G. Fatou, B. Rojas Gáscue, Polym., B 44 (2000) 285- –292

[7] C. Plessier, B. Gupta, A. Chapiro, J. Appl. Polym. Sci. 69 (1998) 1343-1348

[8] G.-S. Tong, T. Liu, G.-H. Hu, L. Zhao, W.-K. Yuan, J. Supercrit. Fluid. 43 (2007) 64-73

[9] V.M. Hoo, R.A. Whitney, W.E. Baker, Polymer 41 (2000) 4367-4371

[10] C. Yu, L. Yun-Fei, T. Hui-Min, J. Jian-Xin, Carbohyd. Polym. 75 (2009) 287-292

[11] J. Yuan, L. Chen, X. Jiang, J. Shena, S. Lin, Colloids Surfaces, B 39 (2004) 87-94

[12] J. Balart, V. Fombuena, R. Balart, J. M. España, R. Navarro, O. Fenollar, J. Appl. Polym. Sci. 116 (2010) 3256-3264

[13] R. Guan, J. Appl. Polym. Sci. 77 (2000) 96–103

[14] A.K. Mukherjee, P.K. Sharma, J. Appl. Polym. Sci. 30 (1985) 1707-1715

[15] J. Zhao, G. Geuskens, Eur. Polym. J. 35 (1999) 2115- –2123

[16] A. Fillat, O. Gallardo, T. Vidal, F.I.J. Pastor, P. Díaz, M.B. Roncero, Carbohyd. Polym. 87 (2012) 146-152

[17] T. Kudanga, E.N. Prasetyo, J. Sipilä, G.S. Nyanhongo, G.M. Guebitz, Enzym. Microb. Technol. 46 (2010) 272- –280

[18] Y.-F. Duann, Y.-C. Chen, J.-T. Shen, Y.-H. Lin, Polymer 45 (2004) 6839-6843

[19] A.V. Machado, J.A. Covas, M. van Duin, Polym. 42 (2001) 3649-3655

[20] A. Vahdat, H. Bahrami, N. Ansari, F. Ziaie, Rad. Phys. Chem. 76 (2007) 787-793

[21] M. Pei, D. Wang, Y. Zhao, X. Hu, Y. Xu, J. Wu, D. Xu, J. Appl. Polym. Sci. 92 (2004) 2637-2641

[22] A.K. Mukherjee, B.D. Gupta, J. Appl. Polym. Sci. 30 (1985) 3479-3491

[23] R.K. Roy, Design of Experiments Using the Taguchi Approach: 16 Steps to Product and Process Improve-ment, Wiley-Interscience, New York, 2001, p. 8

[24] Qualitek 4 software demo, www.nutek-us.com/wp-q4w (access date: June 2012).

MAJID MAZHAR1

MAJID ABDOUSS1

ZAHRA SHARIATINIA1

MOJDEH ZARGARAN2

1Department of Chemistry, Amirkabir University of Technology (Polytechnic),

Tehran, Iran 2Building and Housing Research

Center (BHRC), Tehran, Iran

NAUČNI RAD

GRAFT KOPOLIMERIZACIJA MONOMERA METAKRILNE KISELINE NA POLIPROPILENSKA VLAKNA

U poslednjih nekoliko godina, graft kopolimerizacija se uveliko koristi za umetanje različitih

funkcionalnih grupa na polimere. U ovom radu, monomeri metakrilne kiseline su kalemljeni

na polipropilenska (PP) vlakna, čineći ih hidrofilnim i zadržavajući njihove mehaničke

osobine. Eksperimenti su izvođeni na osnovu Taguchi metode, pri čemu je proučavan

uticaj temperature, koncentracije monomera, koncentracije umreživača i vremena reakcije.

Kalemljenje metakrilne kiseline i divinilbenzena je praćeno pomoću FTIR spektroskopije i

potvrđeno SEM mikrografijom. Zatezna čvrstoća i žilavost uzoraka su mereni i upoređivani

sa sirovim vlaknima. Uticaj kalemljenja na zateznu čvrstoću i žilavost vlakana su takođe

izmereni, a određen je i kritični stepen kalemljenja pri kome su zatezna čvrstoća i žilavost

vlakana definisani.

Ključne reči: mehaničke osobine, PP vlakna, Taguchi metoda, graft kopoli-merizacija, SEM.





97

MOHAMMAD A. BEHNAJADY SHAHRZAD YAVARI

NASSER MODIRSHAHLA

Department of Chemistry, College of Science, Tabriz Branch, Islamic

Azad University, Tabriz, Iran

SCIENTIFIC PAPER

UDC 66.081.3:677:54

DOI 10.2298/CICEQ120610105B

INVESTIGATION ON ADSORPTION CAPACITY OF TIO2-P25 NANOPARTICLES IN THE REMOVAL OF A MONO-AZO DYE FROM AQUEOUS SOLUTION: A COMPREHENSIVE ISOTHERM ANALYSIS

Article Highlights • TiO2-P25 nanoparticles show considerable adsorption capacity • The desired pH for adsorption of AR27 onto TiO2-P25 nanoparticles was 3 • We report a comprehensive isotherm analysis for adsorption of AR27 onto TiO2-P25

nanoparticles • The kinetics of the adsorption process is found to follow the pseudo-second-order

kinetic model Abstract

In this work TiO2-P25 nanoparticles with high surface area have been used as adsorbent for the removal of C.I. Acid Red 27 (AR27), as an organic contam-inant from aqueous solution. Characteristics of phases and crystallite size of TiO2-P25 nanoparticles were achieved from XRD and the surface area and pore size distribution were obtained from BET and BJH techniques. TiO2-P25 nanoparticles with almost 80% anatase and 20% rutile phases, the average crystallite size of 18 nm, have specific surface area of 56.82 m2 g–1. The effect of various parameters like initial AR27 concentration, pH, contact time and adsorbent dosage was carried out in order to find desired adsorption con-ditions. The desired pH for adsorption of AR27 onto TiO2-P25 nanoparticles was 3. The equilibrium data were analyzed with various 2-, 3- and 4-parameter isotherm models. Equilibrium data fitted very well by the 4-parameter Fritz-–Schluender model. The results of the adsorption kinetics study indicated that the pseudo-second order kinetics provided the best fit with correlation coeffi-cients close to unity.

Keywords: adsorption; TiO2-P25 nanoparticles; isotherm analysis; nonli-near regression analysis; kinetics.

Wastewaters produced from industries contain different types of organic and inorganic pollutants. The textile industry uses large volumes of water and generates considerable quantities of wastewater con-taining large amount of dissolved dye compounds [1– -6]. Amongst synthetic dyes, azo dyes constitute the largest and the most important class of commercial dyes with wide range applications [7]. Some dyes are carcinogenic, and the biodegradation of many dyes Correspondence: M.A. Behnajady, Department of Chemistry, Faculty of Science, Tabriz Branch, Islamic Azad University, Tabriz, I.R. Iran. E-mail: [email protected]; [email protected] Paper received: 10 June, 2012 Paper revised: 3 November, 2012 Paper accepted: 4 November, 2012

yields toxic aromatic amines [8,9]. C.I Acid Red 27 (AR27) was used as a food dye, textile dye for wool and silk as well as in photography [10]. However, in 1970, Russian studies [11] showed that AR27 was carcinogenic. The disposal of wastewaters containing dye compounds is a threat to the environment. The removal of dyes from effluents is a major problem for textile industries. The various chemical and physical treatment methods may be efficient for the removal of dyes from aqueous solutions including precipitation, coagulation-flocculation, reverse osmosis and oxida-tion with ozone or hydrogen peroxide [12–14]. Adsorp-tion has proven to be one of the most effective methods for removal of organic and inorganic pollu-tants from aqueous solutions. This process has low

M.A. BEHNAJADY et al: INVESTIGATION ON ADSORPTION CAPACITY… Chem. Ind. Chem. Eng. Q. 20 (1) 97−107 (2014)

98

capital investment cost, simplicity of design and ope-ration. Adsorption onto oxides such as NiO, MgO and TiO2 has been found to be important for removal of organic and inorganic pollutants. TiO2 is widely used

as a pigment in paints and as a semiconductor in pho-tocatalytic removal of pollutants [15]. Adsorption is a vital stage in photocatalytic degradation of organic pollutants on TiO2 surface [16].

The aim of this work was to study the character-istics of TiO2-P25 nanoparticles using some tech-niques and investigation of their adsorption capacity for AR27, as an organic pollutant from textile industry. Non-linear trial and error method was used for anal-ysis of equilibrium data with various 2-, 3- and 4-para-meter isotherm models.

EXPERIMENTAL

Materials

The commercial azo dye AR27 as a model pol-lutant from textile industry was purchased from Merck. The AR27 is a monoazo anionic dye with C.I. number 16185 and λmax 522 nm. The chemical structure of AR27 is given in Figure 1. The TiO2 sample was sup-plied by Degussa, Germany (TiO2-P25). Also, HCl and NaOH were used for adjusting pH of solutions.

Figure 1. Chemical structure of C.I. Acid Red 27.

Characterization of TiO2 nanoparticles

XRD analysis (Siemens D5000 diffractometer) using CuKα radiation at 2θ angle from 5 to 70° was used in order to determine the crystallite size and phase composition of the TiO2-P25 nanoparticles. The average crystallite size of the particles was cal-culated according to Scherrer’s equation [17]:

cos

kD

λβ θ

= (1)

where D is the average crystallite size (nm), k is a constant equal to 0.89, λ is the X–ray wavelength equal to 0.154056 nm, β is the full width at half maximum intensity (FWHM) and θ is the diffraction angle.

If a sample contains only anatase and rutile, the mass fraction of rutile can be calculated from the fol-lowing equation [18]:

A

R

100Rutile phase%

1 0.8II

=

+

(2)

where IA and IR are integrated intensities of the ana-tase and rutile peaks, respectively.

Surface area and porosity were defined by Bru-nauer–Emmett–Teller (BET) and Barrett-Joyner-Hal-enda (BJH) measurements using Belsorp mini II ins-trument based on N2 adsorption-desorption.

Adsorption studies

The adsorption experiments were performed via the batch technique at the temperature of 25 °C. Adsorption of AR27 onto TiO2-P25 nanoparticles was done under various conditions such as initial pH (3, 5 and 9), adsorbent dosage (0.2, 0.5 and 1 g L–1) and initial AR27 concentration (10, 20, 40, 60 and 80 mg L–1). A stock solution of AR27 was prepared in 200 mg L–1 concentration and the diluted to the appropriate concentration. The initial pH was adjusted to the required values with small amount of NaOH or HCl solutions before mixing with the TiO2-P25 nanopar-ticles. Samples were centrifuged at 4000 rpm for 5 min and measurement of AR27 remaining concen-tration in the solution was done using a UV-Visible spectrophotometer (Pharmacia Biotech Ultrospec 2000). The adsorption yield (%) was calculated from the following equation:

0

0

Adsorption% 100tC CC−= × (3)

where C0: initial AR27 concentration (mg L–1); Ct: AR27 concentration at any time (mg L–1).

The amount of AR27 adsorbed onto TiO2-P25 nanoparticles at equilibrium, qe (mg g–1), was cal-culated by the following relationship:

0 ee

( )C C Vq

W− ×= (4)

where Ce is the equilibrium concentration of AR27 (mg L–1), V is the volume of the solution (L) and W is the weight of the TiO2-P25 nanoparticles as an adsor-bent (g). In the all experiments the volume of the solu-tion was 100 mL.


99


Characterization of TiO2 nanoparticles

The XRD pattern of TiO2-P25 nanoparticles (Figure 2) for 2θ diffraction angles between 5 and 70° shows five primary peaks at 25.2, 38, 48.2, 55 and 62.5°, which can be attributed to different diffraction planes of anatase TiO2 and four different peaks at 27.5, 36, 54 and 69°, which can be attributed to dif-ferent diffraction planes of rutile phase of TiO2. According to the XRD results TiO2-P25 nanoparticles have almost 80% anatase and 20% rutile phases. The average particle size of TiO2-P25 nanoparticles

was determined from the XRD pattern according to Scherrer’s equation as 18 nm.Figure 3a shows nitrogen adsorption-desorption isotherms onto TiO2-P25 nanoparticles, which is in accordance with classical type III isotherm of IUPAC classifications suggesting the presence of mesoporous structure [19]. BET surface area and total pore volume of the TiO2-P25 nanoparticles which determined by N2 physisorption experiments were 56.82 m2 g–1 and 0.3879 cm3 g–1, respectively. Figure 3b shows the pore size distribution of the TiO2-P25 nanoparticles as estimated according to the BJH method from the adsorption branch. As it is evident, the sample

Figure 2. XRD Pattern of TiO2-P25 nanoparticles.

(a) (b)

Figure 3. Nitrogen adsorption and desorption isotherms and pore size distribution curve calculated from the adsorption branch for TiO2-P25 nanoparticles.


100

presents a quite extended pore size distribution range from 2 to over 100 nm. TiO2-P25 nanoparticles produced through hydrolysis of TiCl4 in a hydrogen flame do not contain pores in each TiO2 crystallites. Therefore, the formation of the pore structures in the samples could be attributed to the aggregations of TiO2 crystallites [20]. TiO2-P25 nanoparticles with high surface area can be used for efficient adsorption of environmental pollutants with optimization of operational parameters.

Adsorption behavior

The effect of initial pH

The adsorption of AR27 was studied at different initial pHs and adsorbent dosages. Figure 4 indicates the effect of initial pH on the adsorption percent of AR27 onto TiO2 at 25 °C after 30 min equilibration for 60 mg L–1 initial AR27 concentration. The results show that the pH of the system plays an important role in this process and higher adsorption can be obtained at lower pH values; thus, highest adsorption occurred at pH 3.

0

10

20

30

40

50

60

0 2 4 6 8 10

pH

Ads

orpt

ion%

Adsorbent = 1 g/L

Adsorbent = 0.2 g/L

Figure 4. Effect of initial pH on the adsorption of AR27 onto TiO2-P25 nanoparticles. [AR27]0 = 60 mg L–1.

The interaction of solute with metal oxide sur-face is determined by surface charge of metal oxide and ionization state of solute. Nanoparticles of TiO2 in water are of amphoteric nature and are known to have the following acid-base equilibrium [21]:

1

2TiOH TiOH HapK+ +⎯⎯⎯→> +←⎯⎯⎯ (5)

1TiOH TiO HapK − +⎯⎯⎯→> +←⎯⎯⎯ (6)

The pH of zero point of charge (pHzpc) rep-resents surface charge of metal oxide at different initial pHs. At low pH under conditions in which pH < < pHzpc the positively charged surface sites on the adsorbent increase, while at high pH under conditions in which pH > pHzpc the negatively charged surface sites increase. The pHzpc for TiO2-P25 nanoparticles is 6.25 [22]. Interactions between AR27 molecules with anionic character and adsorbent sites will be favored at low pH (pH < pHzpc). In the other words, formation of 2TiOH+> at low pH values and electro-static attraction is mainly responsible for AR27 adsorption onto TiO2 surface [8,24,25].

The effect of initial dye concentration, contact time and adsorbent dosage

Figure 5 shows the effect of contact time and initial AR27 concentration in the adsorption percent onto TiO2-P25 nanoparticles. Results in this figure indicate that adsorption percent decreases signi-ficantly from 97 to 39% with increasing of initial con-centration of AR27 from 10 to 80 mg L–1. Lower adsorption percent at higher initial concentrations of AR27 can be related to the saturation of adsorption sites onto TiO2-P25 nanoparticles. According to the results in this figure majority of adsorption occurred in the first 5 min and then the adsorption % increases

0

20

40

60

80

100

0 5 10 15 20 25 30 35

Time (min)

Ads

orpt

ion%

[AR27] = 10 mg/L[AR27] = 20 mg/L[AR27] = 40 mg/L[AR27] = 60 mg/L[AR27] = 80 mg/L

Figure 5. The effect of contact time and initial AR27 concentration on the adsorption of AR27 onto TiO2-P25 nanoparticles. TiO2-P25 = 1 g L–1, pH 3.


101

with a very low slope. The rapid adsorption at the initial 5 min indicates surface bound adsorption to available positively charged surface of TiO2-P25 nanoparticles as a result of electrostatic attraction and the slow second phase is due to the slow diffusion of solute onto the pores of TiO2-P25 nanoparticles [23]. The results in this figure show that the required con-tact time to reach the equilibrium of AR27 onto TiO2- -P25 nanoparticles is 20 min. The results in Table 1 show the effect of initial AR27 concentration in the adsorbed amount at equilibrium time (qe). As can be seen from these results, qe increases with increasing of initial AR27 concentration. This is probably due to a high driving force for mass transfer. In the other words, higher initial concentrations provide an impor-tant driving force for adsorption of AR27 onto TiO2- -P25 nanoparticles. The results in Table 1 indicate that adsorption percent increases and qe decreases with increasing of adsorbent dosage. Increasing of adsorption % is due to the enhancing of adsorbent surface area which results from increasing of adsor-bent dosage. The qe decrease in the result of adsor-bent dosage increase can be related to the unsatu-ration of some adsorption sites in the high adsorbent dosages.

Adsorption isotherm analysis

The adsorption isotherms can be used to quan-tify the amount of adsorbed substrate on an adsor-bent as a function of concentration at a given tempe-rature. Various 2, 3 and 4-parameter isotherm models were proposed in literature for describing the amount of adsorbate at equilibrium conditions. The different parameters in adsorption models provide some infor-mation about sorption mechanism, surface properties and affinity of sorbent [26].

Two-parameter (2-p) isotherm models

A list of 2-p isotherm models is given in Table 2. All the model parameters were evaluated by non- -linear trial and error method using Polymath 6.0 soft-ware, because non-linear method avoids the variation in error distribution due to linearization and is more appropriate method to determine optimum sorption

isotherm [26]. The calculated adsorption models para-meters at 95% confidence interval; regression coeffi-cient (R2) and adjusted regression coefficient (R2

adj) are reported in Table 3. In this work R2 and R2

adj were used to gauge the goodness-of-fit. For the regression model to be stable and statistically valid, the confi-dence intervals must be much smaller than the res-pective parameter values. Figure 6 show non-linear regression of 2-p isotherm models to experimental data.

Table 2. 2-Parameter isotherm models

Model Equation Parameters

2-p Langmuir m L ee

L e1

q K Cq

K C=

+ qm, KL

2-p Freundlich 1

e F enq K C= KF, n

2-p Temkin e T eln( )

RTq K C

Q=

Δ KT, ΔQ

Langmuir isotherm model

The Langmuir adsorption model is the most common isotherm model, which describes the mono-layer coverage of adsorbate molecules on adsorbent sites without any interactions between adsorbate molecules on adjacent sites. In this isotherm all sites are equivalent and the surface is homogeneous [27]. The results of non-linear regression analysis of Lang-muir model for adsorption of AR27 onto TiO2-P25 indicate a relatively high R2 between 2-p isotherm models and also good accordant between R2 and R2

adj. The maximum mono-layer capacity (qm) obtained from Langmuir model was 38.88 mg g–1. Table 4 lists maximum capacity of the mono-layer adsorption of TiO2-P25 and other metal oxide nanoparticles reported in the literature. In general tested TiO2-P25 in this work exhibited comparable adsorption capacity in comparison with other metal oxide nanoparticles.

KL in the Langmuir model is a constant attributed to the affinity between the adsorbate and adsorbent [29]. A dimensionless constant called the separation factor (KR) can be used for indicating the type of adsorption using the Langmuir constant (KL) as fol-lows:

Table 1. The AR27 adsorption and qe in different adsorbent dosages and initial adsorbate concentrations

TiO2 dosage g L–1

AR27 concentration, mg L–1

10 20 40 60 80

Adsorption %

qe

mg g-1 Adsorption

% qe

mg g-1 Adsorption

% qe

mg g-1 Adsorption

% qe

mg g-1 Adsorption

% qe

mg g-1

0.2 36.065 18.905 30.087 34.401 19.230 44.938 13.785 56.921 9.508 57.05 0.5 70.954 14.669 64.13 26.818 31.744 35 27.509 37.479 24.826 44.587

1 91.250 9.360 78.174 16.178 69.231 26.657 55.044 34.814 38.824 35


102

RL 0

1

1K

K C=

+ (7)

KR value indicate the type of adsorption as: KR = 0 irreversible adsorption, 0<KR<1 favorable adsorption, KR = 1 linear adsorption, KR>1 unfavorable adsorption.

KR values obtained for adsorption of AR27 onto TiO2-P25 at different initial concentrations of adsor-bate are between zero and one, indicating the adsorption is favorable.

Freundlich isotherm model

The Freundlich isotherm is the most important adsorption isotherm for rough and heterogeneous surfaces with interaction between adsorbed mole-cules. In this isotherm 1/n is the heterogeneity factor and indicates deviation from linear adsorption. If the

value of n = 1, the adsorption is linear; if n < 1, the adsorption is chemical; and if n > 1, the adsorption is a favorable physical process [30]. The R2 value obtained for Freundlich isotherm is slightly lower than other 2-p isotherm models. According to the experi-mental data in high concentrations of adsorbate a plateau is reached (Figure 6). The Freundlich iso-therm cannot predict this plateau [29]. The value of 3.458 for n indicates that adsorption of AR27 onto TiO2-P25 is a favorable physical process.

Temkin isotherm model

Temkin studied [31] the heat of adsorption and the adsorbate-adsorbate interactions. This isotherm gave a satisfactory fit to experimental data. In this isotherm ΔQ is related to the heat of adsorption, T is

Table 4. Comparison of the maximum adsorption capacity of TiO2-P25 with other metal oxide nanoparticles

Dye Adsorbent qm / mg g-1 Reference

Reactive Brilliant Red X-3B NiO nanosheets 30.40 [28]

Congo Red 35.15 [28]

Fuchsin Acid 22 [28]

Reactive Yellow 145 TiO2/SiO2 ≈ 9.47 [25]

Reactive Black 5 ≈ 6.82 [25]

Acid Red 27 TiO2-P25 38.88 This work

Table 3. 2-Parameter isotherm model parameter values (±95% confidence level)

Model R2 R2adj Parameters Parameters values

2-p Langmuir 0.975 0.969 qm

KL

38.884 (±0.145)

0.242 (±0.004)

2-p Freundlich 0.963 0.954 KF

n

12.517 (±5.7×10-4)

3.458 (±1.7×10-4)

2-p Temkin 0.975 0.969 RT/ΔQ

KT

6.629 (±2.5×10-4)

5.262 (±7.9×10-4)

0

10

20

30

40

0 10 20 30 40 50

Ceq (mg L-1)

q eq

(mg

g-1)

ExperimentalLangmuirFreundlich Temkin

Figure 6. 2-parameter isotherm models for adsorption of AR27 onto TiO2-P25 nanoparticles. TiO2-P25 = 1 g L–1, pH 3.


103

the absolute temperature (K), R is the universal gas constant (8.314 J mol–1 K–1) and KT is the Temkin model constant (L mg–1) [31]. The adsorption energy obtained for adsorption of AR27 onto TiO2-P25 was 373.75 J mol–1, which indicates an endothermic adsorp-tion process.

Three-parameter (3-p) and four-parameter (4-p) isotherm models

A list of 3-p and 4-p isotherm models is given in Table 5. All the model parameters for these isotherms evaluated by non-linear method and R2 and R2

adj

values are reported in Table 6. Figure 7 also shows non-linear regression of 3-p and 4-p isotherm models to the experimental data.

Table 5. 3- and 4-Parameter isotherm models

Model Equation Parameters

3-p Redlich-Peterson

ee g

e1

ACq

B C=

+ A, B, g

3-p Sips m a e

ea e

( )

1 ( )

q K Cq

K C

α

α=+

qm, Ka, α

3-p Toth m TH ee 1

TH e[1 ( ) ]

q K Cq

K C β β=

+ qm, KTh, β

4-p Fritz-Schluender

ee

e1

K

M

J Cq

LC=

+ J, K, L, M

Redlich-Peterson isotherm model

The Redlich-Peterson isotherm is an empirical equation with three parameters, which has features of the both Langmuir and Freundlich isotherm equa-tions. In this equation A, B and g are model constants, which the value of g lies between 0 and 1 [32]. When g = 1, this isotherm transforms to Langmuir isotherm and when g = 0 transforms to linear form of Freund-lich isotherm (Henry’s law) [26]. The value of 0.9 for g

in Redlich-Peterson isotherm indicates that the iso-therm is approaching the Langmuir but not Freundlich isotherm, which proves previous results of better fit-ting of experimental data by the Langmuir isotherm model.

Sips isotherm model

An isotherm was proposed by Sips in 1948 [33] that has a similar form to the Freundlich isotherm and at low adsorbate concentrations reduces to this iso-therm. The Sips isotherm at high adsorbed concentra-tions predicts a mono-layer adsorption capacity simi-lar to the Langmuir isotherm. The parameter of α in this isotherm is related to the system’s heterogeneity [34]. The Sips isotherm shows a good prediction of AR27 adsorption onto TiO2-P25 nanoparticles, with relatively high value of R2. The value of α is closer to 1 and indicates that the adsorption of AR27 onto TiO2-P25 nanoparticles is approaching Langmuir form more than Freundlich form.

Toth isotherm model

Toth isotherm was derived from potential theory for adsorption in heterogeneous systems. This iso-therm assumes most adsorption sites have sorption energy less than the mean value [35]. Unlike to Freundlich and Sips isotherm models, this isotherm reduces into the correct Henry law in the low concen-trations and for β = 1 this isotherm transforms to the Langmuir isotherm. The value of β in this study is closer to 1, meaning that AR27 adsorption onto TiO2- -P25 nanoparticles is more of a Langmuir form.

Fritz-Schluender isotherm model

This isotherm contains more parameters than any other isotherms and is an empirical isotherm [36]. In the absence of a theoretical model that could account for the chemical heterogeneity of the surface

Table 6. 3- and 4-Parameter isotherm model parameter values (±95% confidence level)

Model R2 R2adj Parameters Parameters values

3-p Redlich-Peterson 0.977 0.962 A

B

g

16.314 (±0.145)

0.7 (±0.9)

0.9 (±0.2)

3-p Sips 0.983 0.972 qm

Ka

α

45.749 (±1.8×10-3)

0.156 (±2.7×10-5)

0.691 (±9.3×10-5)

3-p Toth 0.977 0.962 qm

KTh

β

28.526 (±40.1813)

0.654 (±2.6475)

0.873 (±0.4045)

4-p Fritz-Schluender 0.995 0.988 J

K

L

M

10.069 (±3.5×10-4)

0.426 (±1.1×10-5)

0.0002 (±2.8×10-8)

2.071 (±4.5×10-5)


104

and simultaneous prevalence of different adsorption mechanisms; perhaps a model with more parameters could be able to predict equilibrium data better [29]. A comparison between R2 values for fitting experimental data by various isotherm models indicating that the 4-p Fritz-Schluender model fits experimental data better than other isotherms. In comparison with other curves obtained for various isotherms in Figures 6 and 7, better coincidence was obtained between experimental and results obtained from the 4-p Fritz-Schluender model.

Adsorption kinetic studies

Kinetic modeling

Adsorption kinetics is expressed as the adsor-bent removal rate which is important in estimating the required residence time of adsorbent in the system. Various kinetic models have been used in adsorption process. The pseudo-first-order and pseudo-second-order are the widely used kinetic models in this pro-cess [34].

The Lagergren pseudo-first-order kinetic model is expressed as follows [37]:

1e elog( ) log

2.303tk t

q q q− = − (8)

where qe is the adsorption capacity at equilibrium (mg g–1), qt is the adsorption capacity at time t (mg g–1) and k1 is the pseudo-first-order rate constant (min–1). The plot of log(qe – qt) against t gives a slope of k1 and intercept of log qe. The values of R2 for the pseudo-first-order kinetic model in the adsorption of different concentrations of AR27 onto TiO2-P25 are lower than 0.92 (Table 7), which indicates that the pseudo-first-order kinetic model can not fit perfectly the experi-mental data.

The Ho pseudo-second-order kinetic model is expressed as follows [38]:

2e2 e

1

t

t tq qk q

= + (9)

The constants of this equation were determined experimentally by plotting t/qt versus t (Figure 8). The values of k2, qe and R2 for pseudo-second-order kine-tic model are given in Table 7. Results in Table 7 show that AR27 adsorption onto TiO2-P25 nanopar-ticles has very good compliance with pseudo-second-

0

10

20

30

40

0 10 20 30 40 50

Ceq (mg L-1)

q eq

(mg

g-1)

ExperimentalRedlich-Peterson Sips Toth Fritz-Schluender

Figure 7. 3- and 4-parameter isotherm models for adsorption of AR27 onto TiO2-P25 nanoparticles. TiO2-P25 = 1 g L–1, pH 3.

Table 7. The pseudo-first-order, pseudo-second-order and intraparticle diffusion kinetic model constants for the adsorption of the AR27onto TiO2-P25 for different AR27 initial concentrations. TiO2-P25 = 1 g L-1, pH 3

[AR27]0

mg L-1 qe(exp.)

mg g-1

Pseudo-first-order kinetic model Pseudo-second-order kinetic model Intraparticle diffusion kinetic model

k1 min-1

qe(cal.)

mg g-1

R2 k2 g mg-1 min-1

qe(cal.) mg g-1

R2 kp,2

mg g-1 min-0.5 Ci

mg g-1 R2

10 9.36 0.25 4.93 0.854 0.425 9.42 0.999 0.188 8.42 0.914

20 16.18 0.32 5.14 0.744 0.791 16.23 1 0.087 15.76 0.959

40 28.66 0.24 18.28 0.917 0.068 29.41 0.999 1.055 23.72 0.954

60 34.81 0.20 22.23 0.875 0.046 35.46 0.999 1.498 27.33 0.932

80 33.12 0.08 15.42 0.782 0.047 33.44 0.998 1.300 26.23 0.962


105

order kinetic model. The values of R2 are greater than 0.99 for all experiments. Also, the calculated qe values are in good agreement with the experimental data. According to this model two reactions occur, the first one is fast and reaches equilibrium quickly and the second is slow that can continue for a long time [39].

0

0.5

1

1.5

2

2.5

3

3.5

0 10 20 30 40

Time (min)

t/qt


Figure 8. The pseudo-second-order kinetic plot for adsorption of AR27 onto TiO2-P25 nanoparticles. TiO2-P25 = 1 g L–1, pH 3.

Adsorption mechanism

In order to identify adsorption mechanism Weber and Morris [40] intraparticle diffusion model was tested. This model is expressed as follows:

0.5pt i iq k t C= + (10)

where kpi is the rate parameter of stage i (mg g–1 min–0.5) and Ci is the intercept of stage i that gives an idea about the thickness of boundary layer. The plot of qt

against t0.5 (Figure 9) indicates that the adsorption process is comprised of two phases, involving adsorption on the external surface and intraparticle or pore diffusion. In the other words, the initial linear portion of the plot indicates boundary layer effect, while the second linear portion is due to intraparticle diffusion. Therefore the slope of second linear portion of the plot has been considered as the kp (Table 7) [39,41].

CONCLUSION

In this work TiO2-P25 nanoparticles with 18 nm crystallite size and 58.6 m2 g–1 surface area containing 80% anatase and 20% rutile were used as an efficient adsorbent for removal of AR27 as a monoazo anionic dye from aqueous solutions. Experimental data indi-cate that pH is a very important factor in this process and optimal pH for highest adsorption of AR27 onto TiO2-P25 nanoparticles is 3. Adsorption percent is sensitive to AR27 initial concentration and decreases from 97 to 39% with increasing initial concentration of AR27 from 10 to 80 mg L–1. The maximum mono-layer capacity of TiO2-P25 in the adsorption of AR27 reached 38.88 mg g–1, which is comparable with other metal oxide nanoparticles. A comprehensive isotherm analysis for adsorption of AR27 onto TiO2-P25 nano-particles indicates that the best fit to the equilibrium data was provided by the 4-p Fritz-Schluender model. The kinetics of the adsorption process was found to follow the pseudo-second-order kinetic model. Gene-rally, the TiO2-P25 nanoparticles were effectively cap-able to remove the wide range concentrations of AR27 from aqueous solutions in acidic pHs.

0

5

10

15

20

25

30

35

40

0 1 2 3 4 5 6

Time 0.5 (min 0.5 )

q t (m

g g-1

)


Figure 9. The intraparticle diffusion plot for adsorption of AR27 onto TiO2-P25 nanoparticles. TiO2-P25 = 1 g L-1, pH 3.


106

Acknowledgements

The authors would like to thank the financial support of Tabriz branch, Islamic Azad University, and the Iranian Nanotechnology Initiative Council, Iran.

REFERENCES

[1] S.A. Figueiredo, J.M. Louriero, A.R. Boaventura, Water Res. 39 (2005) 4142-4152

[2] C. Sourja, D. Sirshendu, D. Sunando, K.B. Jayanta, Chemosphere 58 (2005) 1079-1086

[3] C. Namasivayam, S. Sumithra, J. Environ. Manage. 74 (2005) 207-215

[4] O. Abdelwahab, A. El Nemr, A. El-Sikaily, A. Khaled, Chem. Ecol. 22 (2006) 253-266

[5] A. El-Sikaily, A. Khaled, A. El Nemr, O. Abdelwahab, Chem. Ecol. 22 (2006) 149-157

[6] B. Noroozi, G. A. Sorial, H. Bahrami, M. Arami, J. Hazard. Mater., B 139 (2007) 167-174

[7] M. Neamtu, I. Siminiceanu, A. Yediler, A. Kettrup, Dyes Pigments 53 (2002) 93-99

[8] M.F. Boeniger, in Carcinogenity of azo dyes derived from benzidine, Pub. No. 8-119, Cincinnati, OH, 1980

[9] T. Zheng, T.R. Holford, S.T. Mayne, P.H. Owens, P. Boyle, B. Zhang, Eur. J. Cancer 38 (2002) 1647-1652

[10] M. Karkmaz, E. Puzenat, C. Guillard, J.M. Herrmann, Appl. Catal., B 51 (2004) 183-194

[11] M. Perez-Urquiza, J. L. Beltran, J. Chromatogr., A 898 (2000) 271-275

[12] T. Robinson, G. McMullan, R. Marchant, P. Nigam, Bio-resour. Technol. 77 (2001) 247-255

[13] P. Cooper, J. Soc. Dyers Colour. 109 (1993) 97-100

[14] Y.M. Slokar, A. Majcen Le Marechal, Dyes Pigments 37 (1998) 335-356

[15] I.K. Konstantinou, T.A. Albanis, Appl. Catal., B 49 (2004) 1-14

[16] V. Belessi, D. Lambropoulou, I. Konstantinou, R. Zboril, J. Tucek, D. Jancik, T. Albanis, D. Petridis, Appl. Catal., B 87 (2009) 181-189

[17] A.L. Patterson, Phys. Rev. 56 (1939) 978-982

[18] R.A. Spurr, H. Myers, Anal. Chem. 29 (1957) 760-762

[19] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T. Siemieniewska, Pure Appl. Chem. 57 (1985) 603-619

[20] J. Yu, H. Yu, B. Cheng, M. Zhou, X. Zhao, J. Mol. Catal., A 253 (2006) 112-118

[21] K. Bourikas, M. Stylidi, D.I. Kondarides, X. Verykios, Langmuir 21 (2005) 9222-9230

[22] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann, Chem. Rev. 95 (1995) 69-96

[23] M. Ahmaruzzaman, S.L. Gayatri, J. Chem. Eng. Data 55 (2010) 4614–4623

[24] P.K. Dutta, A.K. Ray, V.K. Sharma, F.J. Millero, J. Colloid Interf. Sci. 278 (2004) 270-275

[25] A. Aguedach, S. Brosillon, J. Morvan, E.K. Lhadi, Appl. Catal., B 57 (2005) 55-62

[26] K. Vasanth Kumar, S. Sivanesan, Dyes Pigments 72 (2007) 130-133

[27] I. Langmuir, J. Am. Chem. Soc. 38 (2000) 2221-2295

[28] Z. Song, L. Chen, J. Hu, R. Richards, Nanotechnol. 20 (2009) 275707

[29] J.U.K. Oubagaranadin, Z.V.P. Murthy, AICHE J. 56 (2010) 2312-2322

[30] H.M.F. Freundlich, Z. Phys. Chem. 57A (1906) 385-470

[31] M.J. Temkin, V. Pyzhev, Acta Physiochim. URSS 12 (1940) 217-225

[32] O. Redlich, D.L. Peterson, J. Phys. Chem. 63 (1959) 1024-1024

[33] R. Sips, J. Chem. Phys. 16 (1948) 490-495

[34] J. Febrianto, A. Natasia Kosasih, J. Sunarso, Y. H. Ju, N. Indraswati, S. Ismadji, J. Hazard. Mater. 162 (2009) 616- –645

[35] J. Toth, J. Colloid Interf. Sci. 225 (2000) 378-383

[36] W. Fritz, E. U. Schluender, Ceram. Eng. Sci. Proc. 29 (1974) 1279-1282

[37] S. Lagergren, Handlingar 24 (1898) 1-39.

[38] Y.S. Ho, G. McKay, Process Biochem. 34 (1999) 451-465

[39] Y. Khambhaty, K. Mody, S. Basha, B. Jha, Chem. Eng. J. 145 (2009) 489-495

[40] W.J. Weber, J.C. Morris, J. Sanit. Eng. Div. Am. Soc. Civ. Eng. 89 (1963) 31-60

[41] M. Dogan, H. Abak, M. Alkan, J. Hazard. Mater. 164 (2009) 172-181.


107

MOHAMMAD A. BEHNAJADY

SHAHRZAD YAVARI

NASSER MODIRSHAHLA

Department of Chemistry, Faculty of Science, Tabriz Branch, Islamic Azad

University, Tabriz, Iran

NAUČNI RAD

ISPITIVANJE ADSORPCIONOG KAPACITETA TIO2-P25 NANOČESTICA ZA UKLANJANJE MONO-AZO BOJE IZ VODENOG RASTVORA: SVEOBUHVATNA ANALIZA IZOTERMI

U ovom radu, TiO2-P25 nanočestice velike specifične površine su korišćene kao adsorbent

za uklanjanje azo boje Acid Red 27 (AR27) iz vodenog rastvora. Karakteristike faza i

veličina kristala TiO2-P25 nanočestica su određene XRD analizom, dok su specifična

površina i raspodela veličine pora dobijene BET i BJH tehnikom. TiO2-P25 nanočestice sa

gotovo 80% anataz faze i 20% rutil faze, srednje veličine kristala 18 nm, imaju specifičnu

površinu 56,82 m2 g-1. Da bi se odredili željeni uslovi adsorpcije, ispitivan je uticaj različitih

parametara (početna koncentracija AR27, pH, kontaktno vreme i doza adsorbenta).

Poželjna pH vrednost za adsorpciju AR27 na TiO2-P25 nanočesticama je bio 3. Ravno-

težni podaci su analizirani dvo-, tro- i četvoro-parametarskim modelima adsorpcione izo-

terme. Ravnotežni podaci dobro fituju četvoro-parametarski model Fritz-Schluendera. Re-

zultati proučavanja kinetike adsorpcije su pokazali da model sa kinetikom pseudo drugog

reda daje najbolj slaganje, sa koeficijentom korelacije blizu jedan.

Ključne reči: TiO2-P25 nanočestice, analiza izotermi, nelinearna regresiona ana-liza, kinetika.





109

KULANDAIVELU KARUNAKARAN1

GURUSAMY NAVANEETHAN1

ELANGO KUPPANAGOUNDER

PITCHAIMUTHU2 1Department of Chemistry, Sona

College of Technology (Anna University), Salem, India

2Department of Chemistry, Gandhigram Rural Institute

(Deemed University), Gandhigram, India

SCIENTIFIC PAPER

UDC 543.544.5.068.7:615

DOI 10.2298/CICEQ120502106K

A VALIDATED STABILITY-INDICATING RP-HPLC METHOD FOR PARACETAMOL AND LORNOXICAM: APPLICATION TO PHARMACEUTICAL DOSAGE FORMS

Article Highlight • The developed method is user-friendly for routine release analysis in pharmaceutical

industry and it achieved very low level of LOQ value, hence it can be used for the cleaning verification of LR and PR

Abstract

A new method for the simultaneous determination of paracetamol (PR) and lornoxicam (LR) has been developed by reversed phase HPLC from the com-bination drug product. The separation achieved on a C18 column using aceto-nitrile and 0.02 M potassium dihydrogen phosphate in the ratio of 35:65 (v/v) as the mobile phase at a flow rate of 1.0 mL/min. Both the components were monitored at a single wavelength at 260 nm and the column temperature was maintained at 30 °C throughout the analysis. A linear response was found in the concentration range of 125–375 µg/mL for PR and 2–6 µg/mL for LR, with the correlation coefficient of more than 0.999. Although the tablet contained a high dose of PR (500 mg) and a low dose of LR (8 mg), a single HPLC method was developed and the intra- as well as inter-day precision was obtained at less than 2% of RSD. The accuracy results obtained were between 98% and 102%. The drug was intentionally degraded under acidic, basic, peroxide, ther-mal and photolytic conditions. The major degradation observed for both PR and LR under peroxide condition indicated that the drug product is susceptible to oxidation. The degraded peaks were properly resolved from PR and LR. Hence, the method is stability indicating.

Keywords: RP-HPLC, paracetamol, lornoxicam, stability indicating, forced degradation, validation.

Lornoxicam (LR), an oxicam derivative, is an NSAID. It is used for the treatment of musculoskeletal and joint disorders such as osteoarthritis and rheuma-toid arthritis. It is also used for the treatment of other painful conditions including postoperative pain. For the treatment of osteoarthritis and rheumatoid arthri-tis, LR is administered in an oral dosage form at a daily dose of 12 mg, in two or three divided doses. Paracetamol (PR) is a widely used over-the-counter as an analgesic and an antipyretic. It is also used for

Correspondence: K. Karunakaran, Department of Chemistry, Sona College of Technology (Anna University), Salem 636 005, India. E-mail: [email protected] Paper received: 2 May, 2012 Paper revised: 8 November, 2012 Paper accepted: 8 November, 2012

fever, headaches and other minor aches and pains. It is a major ingredient in numerous cold and flu medi-cines. PR is also used in the management of very severe pain such as pain caused by cancer or post-operative pain [1]. Chemically, PR is called N-(4-hyd-roxyphenyl)ethanamide and LR, 6-chloro-4-hydroxy- -2-methyl-N-2-pyridinyl-2H-thieno-[2,3-e]-1,2-thiazine- -3-carboxamide 1,1-dioxide (Figure 1).

A literature survey revealed that RP-HPLC [2–5] and UV-Vis [6] methods have been reported for the estimation of PR in many combination drugs. The estimation of PR individually has been carried out using grapheme-based electrochemical sensor [7], HPTLC [8] and GC-MS [9] methods. Similarly, for the estimation of LR in plasma LC-MS [10] and pharma-ceutical formulation, derivative spectroscopy [11] and polarography [12] methods have been reported to be

K. KARUNAKARAN et al.: A VALIDATED STABILITY-INDICATING RP-HPLC… Chem. Ind. Chem. Eng. Q. 20 (1) 109−114 (2014)

110

effective. Stability-indicating methods have also been reported to be effective for the estimation of LR indi-vidually [13] and in combination with other oxicams [14,15] and thiocolchicoside [16].

OH

NH

CH3

O

(a)

NNH

O

NS

S

OH

Cl

OO

CH3

(b)

Figure 1. The chemical structures of a) paracetamol and b) lornoxicam.

Other than the above methods, there is no offi-cial method for the estimation of LR in any of the pharmacopoeias. Recently, a UV-Vis spectrophoto-metric method [17] and TLC [18] method were pub-lished for the simultaneous determination of PR and LR. Some of the non-stability indicating HPLC methods were reported for the combination drug product [19,20]. Dinesh Kumar et al. published stability indi-cating HPLC and HPTLC methods [21]. However, this method does not discuss the degradation product in the HPLC method. Srinivasu et al. reported the non-stability indicating UPLC method [22]. Therefore, the aim of the present endeavor is to develop a simple, rapid, accurate and stability-indicating isocratic RP- -HPLC method for the simultaneous estimation of both drugs in a combined tablet dosage form in pre-sence of their degradation products.


Reagents

The working standards of PR and LR have been received as gift samples from M/S GS laboratory, Chennai, India, and M/S Pharma laboratory, Baddi, India. LORNISTAR-P, a commercially available drug, containing 500 mg of PR and 8 mg of LR was used. HPLC grade acetonitrile and AR grade potassium dihydrogen phosphate were purchased from Merck, India, and were used as received. The water used in the present study was provided by a Milli-Q water

purification system. All other reagents employed were of high purity analytical grade.

Instrumentations and chromatographic conditions

The chromatographic system (Waters Alliance), equipped with quaternary solvent delivery system with 2487 UV and 2996 PDA detectors was employed in the present study. A Kromasil C18 column, 5 µm, 250 mm×4.6 mm, was used as a stationary phase. The separation was achieved in the isocratic mode, with mobile phase consisting of a mixture of 0.02 M potas-sium dihydrogen phosphate and acetonitrile in the ratio of 65:35 (v/v), with a flow rate of 1.0 mL/min. The eluants were monitored at 260 nm. The analysis was carried out at 30 °C at an injection volume of 50 µL. All mobile phases were filtered through a 0.45 µm Millipore filter.

Preparation of standard solutions

A standard solution containing 250 µg/mL of PR and 4 µg/mL of LR was prepared by dissolving PR and LR working standards in the diluent (70:30 (v/v) acetonitrile and water).

Preparation of sample solutions

Five tablets were weighed and finely powdered. The total powder sample was transferred to a 500 mL standard volumetric flask. Approximately 300 mL of the diluent was added and sonicated for 30 min with intermittent shaking. The volume was made up to the mark with the diluent. The solution was filtered and further diluted with the diluent to have a final working concentration of 250 µg/mL of PR and 4 µg/mL of LR.


Method development

The RP-HPLC method was optimized with a goal of developing a stability indicating assay method to quantify PR and LR simultaneously from the phar-maceutical dosage form. The degraded samples were used for the optimization of the method. Both pure drugs and degraded samples were injected in diffe-rent mobile phases and columns. The initial results observed in the mobile phase containing phosphate buffer and acetonitrile with C18 column were positive. The ratio of acetonitrile was adjusted in such a way that all co-eluting degradation products were well separated from the PR and LR peaks. The primary focus during development was to achieve good peak shape of LR. Various ion-pairing agents such as triethylamine and trifluoroacetic acid were tested to improve peak properties such as peak height and peak tailing, but with no positive effect. A different


111

commercial C18 column was tried finally good peak shape and theoretical plates found in Kromasil C18 column without using any ion-pair agent. The wave-length selected was 260 nm at which both the drugs got good response. Satisfactory results were obtained with the above-mentioned chromatographic condi-tions with respect to tailing, theoretical plates and resolution.

Optimization of sample preparation

Twenty tablets were weighed and crushed. The powder equivalent of 40 mg of LR was taken in a 500 mL volumetric flask; 300 mL of the diluent was added, sonicated for 30 min, and then made up to the mark with the diluents, 5 mL of the solution was further diluted to 100 mL with the diluent. The samples were analyzed using the proposed analytical method. The recovery of LR was found between 82 and 111%. The RSD, %, was found to be more than 10 in six replicate sample preparations. In order to improve the recovery and consistency of LR, five tablets were crushed with the help of mortar and pestle. The total powder sample was taken in a 500 mL volumetric flask; 300 mL of the diluent was added and sonicated for 30 min and then made up to the mark with the diluents, 5.0 mL of the solution was further diluted to 100 mL with the diluent. The recovery found was between 98% and 102% for LR and PR. The %RSD was found to be less than 2.0, indicating precision of the method.

Method validation

The optimized HPLC method was validated with respect to specificity, selectivity, precision, rugged-ness, linearity, robustness and accuracy [23,24].

System suitability

To check the system and column performance, the standard solution was injected five times. The following parameters were monitored, i.e., tailing factor (not more than 2.0), theoretical plates (not less than 5000 for PR and not less than 10000 for LR), RSD of PR and LR (not more the 2%) and the resolution between PR and LR (not less than 2.0). The system suitability results are shown in Table 1.

Specificity and selectivity

The intentional degradation study was attempted to prove the stability indicating power of the optimized LC method and to find the degradation pathway of the

drug product. All forced degradation samples were prepared at a test concentration of 250 µg/mL of PR and 4 µg/mL of LR. For acid (0.1 M HCl), base (0.1 M NaOH) hydrolysis and peroxide oxidation (10% H2O2) the solution were refluxed at 80 °C for 8 h. Deg-radation was also carried out in solid sample by exposing the drug product at 105 °C temperature and light at 254 nm for 48 h. To prove the selectivity of the method, individual components of PR and LR were injected. All the stressed samples were analysed and all the peak purity tests were carried out using the PDA detector. The percentage degradation was cal-culated against the standard solution of PR and LR.

The results are shown in Table 2, indicating that the drug product was stable under photolytic and thermal stress conditions, while minor degradation occurred in acid and base degraded samples. Sig-nificant degradation of the drug product under oxide-tive condition led to the formation of unknown deg-radation product peaks at 2.82, 3.97, 4.82, 5.36, 7.21 and 9.46 min (Figure 2). The results of the peak purity test using PDA confirmed that the PR and LR peaks obtained from all the stressed samples analyzed were homogeneous and pure. The close eluting degrada-tion products were well resolved from the pure drugs. The purity angle was observed to be within the purity threshold limit in all the stressed samples, demon-strating the stability indicating power of the method. No blank interference was found at the RT of PR and LR (Figure 3).

Precision

The precision of the developed method was evaluated by injecting six individual preparations of the sample at the test concentration. For intermediate precision, six individual samples were analysed by different analysts on different days using different HPLC systems. The percentage assay and RSD were calculated for PR and LR. The RSD of PR and LR was found to be less than 2.0 proving the precision of the method.

Linearity

The linearity curves were plotted in the range of 125–375 µg/mL for PR and 2–6 µg/mL for LR against their peak areas. Excellent linearity was obtained, which is represented by the following linear regres-sion equations:

Table 1. System suitability results

Compound RT / min Resolution USP Tailing factor USP Theoretical plates

PR

LR

3.29

10.49

-

26.51

1.2

1.2

8045

11519


112

Table 2. Results of forced degradation study; acceptance criteria: purity angle is less than purity threshold

Conditions PR LR

Assay, % Purity angle Purity threshold Assay, % Purity angle Purity threshold

Unstressed sample

Acid hydrolysis (0.1 M HCl for 8h)

Base hydrolysis (0.1 M NaOH for 8 h)

Oxidation (10% H2O2 for 8 h)

Thermal (105 °C for 48 h)

Photolytic (254 nm for 48 h)

99.5

97.5

96.2

82.3

98.9

99.1

0.390

0.412

0.451

0.515

0.512

0.362

1.022

1.015

0.962

0.981

1.011

1.032

99.1

97.0

95.5

75.2

98.9

99.1

0.588

0.513

0.423

0.315

0.526

0.423

0.820

0.932

0.852

0.810

0.921

0.791

Figure 2. Typical LC chromatogram obtained from a sample after oxidative stress.

Figure 3. Typical HPLC chromatograms of blank (A) and unstressed sample (B).

( )2PR 14745 48000 0.999Y x r= + =

( )2LR 10192 926 0.999Y x r= + =

Accuracy

Accuracy of the developed method was con-firmed by the recovery study as per ICH guidelines at three different concentration levels, viz., 80, 100 and 120%. The study was performed in triplicate to check the recovery of the drugs at different levels in the formulations. The result of the accuracy study is reported in Table 3. The percentage recovery ranged from 99.32 to 100.05% for PR and from 99.25 to

100.10% for LR. According to the recovery results, it is clear that the proposed method is very accurate for the estimation of PR and LR in the combined dosage form.

Robustness

The effects of introducing small changes in the chromatographic parameters on the results were examined. The effect of flow rate of the mobile phase (±0.1 units), column temperature (25–35 °C), solvent concentration (±10%) and wavelength change (±2 nm) were investigated. In all these deliberate changes, the assay variability of PR and LR was within ±2%.


113

The low values of %RSD obtained after introducing small changes in the chromatographic parameters indicated the robustness of the method.

Limit of detection (LOD) and limit of quantification (LOQ)

In order to estimate the LOD and LOQ, the serial dilution of PR and LR solutions was performed and the signal-to-noise ratio was determined. The signal-to-noise ratios of 3 and 10 were considered as LOD and LOQ, respectively. The LOD and LOQ concen-trations for PR were found to be 0.238 µg/mL and 0.665 µg/mL, respectively. The LOD and LOQ con-centrations for LR were found to be 0.192 and 0.480 µg/mL, respectively. The LOQ precision was studied at six different preparations. The RSD for PR and LR was found to be less than 5.

Market sample analysis

The proposed method was verified by means of six replicated estimations of the pharmaceutical pre-parations of the commercially available tablet LORNISTAR-P, containing 500 mg of PR and 8 mg of LR. The average assay for PR and LR was found to be 99.51 and 100.15%, respectively. The RSD values for the six assay values were 1.10% for PR and 1.21% for LR.

CONCLUSION

The proposed isocratic RP-HPLC method deve-loped for the simultaneous determination of PR and LR was found to be specific, linear, precise, accurate and robust. All validation parameter results were within the acceptance range. As the LOQ and LOD values of PR and LR were achieved at a very low level, this method can be suitable for cleaning vali-dation in the pharmaceutical industry. The developed method is stability indicating and can be conveniently used by the quality control department to determine the assay of PR and LR simultaneously in commercial and stability samples.

Acknowledgments

The author wishes to thank M/S GS lab (Chen-nai, India) and M/S Pharma Lab (Baddi, India) for providing gift samples of PR and LR standards. REFERENCES

[1] S.C. Sweetman, Martindale: The Complete Drug Refe-rence, 36th ed., The Pharmaceutical Press, London, 2009, pp. 77, 108

[2] J.T. Franeta, D. Agbaba, S. Eric, S. Pakvok, M. Aleksic, S. Vladimirov, Farmaco. 57 (2002) 709-713

[3] K.A. Shaikh, A.B. Devkhile, J. Chromatogr. Sci. 46 (2008) 649-652

[4] H. Cenyuva, T. Ozden, J. Chromatogr. Sci. 40 (2002) 97- –100

[5] P. Reddy, Int. J. Pharm. Tech. Res. 1 (2009) 514-516

[6] S.J. Wadher, P.R. Pathankar, P. Manisha, R.O. Ganj-wale, P.G. Teole, Ind. J. Pharm. Sci. 70 (2008) 393-395

[7] X. Kang, J. Wang, H. Wu, J. Liu, A.A. Ilhan, Y. Lin, Talanta 81 (2010) 754-759

[8] D.K. Laxman, Y. Asmita, Kamble, V.M. Mahadeo, R.D. Sunil, J. AOAC Int. 93 (2010) 765-770

[9] T. Belal, T. Award, C. Randall Clark, J. AOAC Int. 92 (2009) 1622-1630

[10] Y.H. Kim, H.Y. Ji, E.S. Park, S.W. Chae, H.S. Lee, Arch. Pharm. Res. 30 (2007) 905-910

[11] E. Nemutlu, S. Demicran, S. Kir, Parmazie 60 (2005) 421-425

[12] C. Ibrahim, K. Nisa, A. Sule, C. B. U. J. Sci. 5 (2009) 11- –18

[13] R.P. Kiran, P.R. Vipul, N.S. Jaiprakash, B.S. Devanand, Chromatographia 69 (2009) 1001-1005

[14] E.A. Taha, N.N. Salama, L. El-Said Abdel Fattah, J. AOAC. Int. 87 (2004) 366-373

[15] J. Joseph-Charles, M. Bertucat, J. Liq. Chromatogr. Related Technol. 22 (1999) 2009-2021

[16] M.B. Shekhar, M.P. Dasharath, P.K. Amit, C.N. Patel, J. Chem. Pharm. Res. 2 (2010) 563-572

[17] S. Lakshmi, K.S. Lakshmi, T. Tintu, Int. J. Pharm. Sci. 2 (2010) 166-168

[18] P.J. Dhara, P.P. Vivek, Int. J. Chem Tech. Res. 2 (2010) 1929-1932

[19] M.N. Santosh, N.S. Surekha, L.R. Swaroop, S.N. Jaiprakash, Pharmacia Sinica 2 (2011) 138-144

Table 3. Accuracy results

Drug Amount added

Recovery, % (n = 3) RSD / % (n=3) % µg/ml

PR 80

100

120

200

250

300

100.05

99.51

99.32

1.23

0.89

1.05

LR 80

100

120

3.2

4.0

4.8

99.51

100.10

99.25

1.10

1.23

1.15


114

[20] M. Attimarad, Pharmaceutical Methods 2 (2011) 61-66

[21] J. Dinesh Kumar, D. Nitin, S. Yash, Int. J. Pharm. Biomed. Sci. 2 (2011) 55-60

[22] T. Srinivasu, B.N. Rao, M.M. Annapurna, T.G. Chandra-shekhar, Int. J. Pharm. Sci. Res. 3 (2012) 1149-1154

[23] ICH Validation of Analytical Procedures (2005). Text and Methodology Q2 (R1), International Conference on Harmonization, IFPMA, Geneva, 2005

[24] United States Pharmacopoeia (2008), 31st ed., United States Pharmacopoeia Convention Inc., Rockville, MD, pp, 683-687.

KULANDAIVELU

KARUNAKARAN1

GURUSAMY NAVANEETHAN1

ELANGO KUPPANAGOUNDER

PITCHAIMUTHU2

1Department of Chemistry, Sona College of Technology (Anna

University), Salem, India 2Department of Chemistry, Gandhigram

Rural Institute (Deemed University), Gandhigram, India

NAUČNI RAD

VALIDACIJA RP-HPLC METODE ZA PRAĆENJE STABILNOSTI PARACETAMOLA I LORNOKSIKAMA U FARMACEUSTKIM PREPARATIMA

Razvijena je nova, jednostavna metoda za simultano određivanje paracetamola (PR) i

lornoksikama (LR) u kombinovanim preparatima RP-HPLC hromatografijom. Razdvajanje

je izvedeno na C18 koloni, koristeći kao mobilnu fazu rastvarač acetonitrile-kalijum-dihid-

rogen-fosfatni puffer (0,02 M) u odnosu 35:65, v/v. Protok mobilne faze bio je 1,0 ml/min.

Temperatura kolone je održavana na 30 °C , a obe komponente su detektovane na 260

nm. Metoda je linerana za određivanje paracetamola u opsegu 125-375 µg/mL, a lorno-

ksikama u opsegu 2-6 µg/mL sa koeficijentom korelacije većim od 0,999. Iako tablete

sadrže visoku koncentraciju paracetamola (500 mg) i malu koncentraciju lornoksikama (8

mg), moguće je određivanje obe komponente jednom HPLC analizom. Varijabilnost

metode u toku jednog i više dana je <2% RSD. Tačnost je između 98 i 102%. Aktivne sup-

stance su podrvgnute kiseloj, baznoj, peroksidnoj, termalnoj i fotolitičkoj degradaciji.

Nađeni degradacioni produkti za PR i LR pri peroksidnoj degradciji su pokazali da se ove

lekovite supstance lako oksidišu. Degradacioni produkti se uspešno razdvajaju od PR i LR

razvijenom metodom, pa je ona pogodna za praćenje stabilnosti ispitivanih preparata.

Ključne reči: RP-HPLC, paracetamol, lornoksikam, praćenje stabilnosti leka, forsirana degradacija, validacija.





115

A. POBUDKOWSKA

U. DOMAŃSKA

Department of Physical Chemistry,

Faculty of Chemistry, Warsaw University of Technology, Warsaw,

Poland

SCIENTIFIC PAPER

UDC 615.07:54

DOI 10.2298/CICEQ120531116P

STUDY OF PH-DEPENDENT DRUGS SOLUBILITY IN WATER*

Abstract

The solubilities of five sparingly soluble drug-compounds in water were mea-sured at constant temperatures (298.2 and 310.2 K) by the classical saturation shake-flask method. All substances presented in this work are derivatives of anthranilic acid: flufenamic acid (FLU), mefenamic acid (MEF) niflumic acid (NIF), diclofenac sodium (DIC) and meclofenamic sodium (MEC). All of them have anti-inflammatory action. Since the aqueous solubility of the ionized drug is significantly higher than the unionized, the experimental conditions that affect equilibrium solubility values such as composition of aqueous buffer were examined. The Henderson-Hasselbalch (HH) relationship was used to predict the pH-dependent solubility profiles of chosen drugs at two temperatures. For this purpose the pKa values of the investigated drugs were determined using the Bates-Schwarzenbach spectrophotometric method at a temperature of 310.2 K. At temperature of 298.2 K these values were reported previously. Similar values of pKa were obtained from the solubility measurements.

Keywords: derivatives of anthranilic acid, pH-solubility profile, pKa; shake flask method, Henderson-Hasselbalch approach.

The solubility of a drug is defined as the maxi-mum quantity of a drug dissolved in a given volume of a solvent at chosen temperature, pressure and pH. For ionizable drugs, the solubility can be affected by the pH of the solution, and the intrinsic solubility (S0) is defined as the concentration of a saturated solution of the neutral form of the drug, in equilibrium with its solid at constant temperature and pressure.

Nowadays, drug design approaches based on a combination of chemistry and quantitative structure- –activity relationship led to new active substances that are less water soluble and more lipophilic. Not very lipophilic drugs reveal lower solubility in water and have trouble crossing membranes. The acidic group of a drug molecule becomes negatively charged by losing a hydrogen ion at pH < 7. Research in phar-maceutical chemistry has devoted little attention to

Correspondence: A. Pobudkowska, Department of Physical Chemistry, Faculty of Chemistry, Warsaw University of Tech-nology, Noakowskiego 3, 00-664 Warsaw, Poland. E-mail: [email protected] *This paper was presented at the Second World Conference on Physico–Chemical Methods in Drug Discovery and Develop-ment, September 18-22, 2011, Zadar, Croatia. Paper received: 31 May, 2012 Paper revised: 19 October, 2012 Paper accepted: 26 November, 2012

the physicochemical properties of the chemical leads and has focused mainly on optimization of the in vitro activity [1–3]. The rate at which a drug goes into the solution when it is dissolved in an acidic or a basic medium is proportional to the solubility of the drug. Many drugs have different solubilities at different pHs. These pH-dependent solubility differences lead to pH- -dependent dissolution profiles. The solubility-pH pro-file of drugs or amines has already been reported by many authors [4–13].

The Henderson-Hasselbalch (HH) equation [14] has been used many times for the mathematical des-cription of the solubility-pH profile of drugs or amines existing in the solution as a monomer [4,6,7].

Aqueous solubility has an essential role in the bioavailability of oral drug formulations. There is an established classification, namely, the biopharma-ceutical classification system (BCS), which divides drugs into four classes in terms of their solubility and permeability [15]. The BCS classification correlates the in vitro solubility and permeability to the in vivo bioavailability.

In recent years, the problem of drug solubility in water has become more acute and more common as pharmaceutical companies have improved drugs for certain therapeutic areas [16]. The accuracy of many

A. POBUDKOWSKA, U. DOMAŃSKA: STUDY OF PH-DEPENDENT DRUGS… Chem. Ind. Chem. Eng. Q. 20 (1) 115−126 (2014)

116

predictive methods can be discussed when it is pos-sible to compare calculated and measured solubility [17,18]. A number of useful experimental methods are reviewed, including the miniaturized shake-flask mic-rotitre plate, the micro-solubility self-calibrating direct UV, the potentiometric and micro-dissolution methods [5] as well as the “new shake-flask method” [19].

The aim of the present study was to measure the solubility of five ionisable drugs: flufenamic acid (FLU), mefenamic acid (MEF), niflumic acid (NIF), diclofenac sodium (DIC) and meclofenamic sodium (MEC) at two temperatures, 298.2 and 310.2 K using the traditional shake-flask solubility method. We have already measured the thermodynamic solubility of these drugs as a function of temperature at natural pH 7 in water, ethanol and 1-octanol [20–22]. These results will be compared to the buffer solutions at the same temperature and pH. The solubility-pH profile of mefenamic acid as well as the pKa values at two tem-peratures, 298.2 and 310.2 K have also been pre-sented earlier [5] and will be compared to our new values.

The novelty of the present work is to show the effect of pH on the solubility in buffer solutions at two temperatures and ambient pressure. The intrinsic solubility, S0 at two temperatures will be developed from the solubility-pH profile measurements. The pKa values obtained using the Bates-Schwarzenbach spectrophotometric method will be compared with the pKa values, coming from the solubility-pH profile mea-surements. The sigmoidal relationship of solubilities of weak acid drugs and two salts will be predicted using the HH equation.


Chemicals and reagents

The following investigated drugs were obtained from Sigma Aldrich: flufenamic acid (CAS registry No. 530-78-9, ≥0.99), mefenamic acid (CAS registry No. 61-68-7, ≥0.99), niflumic acid (CAS registry No. 4394-00-7; ≥0.99), diclofenac sodium salt (CAS registry No. 15307-79-6; ≥0.99), meclofenamic sodium (CAS registry No. 6385-02-0; ≥0.99). The drugs were used

without purification and were used as powder or small crystals. Other chemicals were as follows: methanol for HPLC – super gradient (CAS registry No. 67-56-1, POCH, 99.9%), hydrochloric acid (CAS registry No. 7647-01-0, POCH), sodium hydroxide (CAS registry No. 1310-73-2, POCH), dipotassium hydrogen phosphate (CAS registry No. 7758-11-4, POCH, ≥0.99), disodium tetraborate (CAS registry No. 1303-96-4, POCH, ≥0.999), sodium chloride (CAS registry No. 231-598-3, POCH, ≥0.999),

All solvents were filtrated twice with the Schott funnel with 4 μm pores. They were stored under freshly activated molecular sieves of type 4 Å. Water used as a solvent was twice distilled, degassed and filtered with Milipore Elix 3. The names, abbreviations, systematic (IUPAC) names, molecular formulas and molar mass of the drugs are given in Table 1.

The pKa measurements

The pKa measurements were performed with the Bates-Schwarzenbach (BS) method using a UV-Vis spectrophotometer (Perkin Elmer Life and Analytical Sciences, Shelton, USA). The method was described in our previous paper [21]. The UV-Vis spectra for acidity constant measurements at temperature 310.2 K are presented at three conditions: buffer, 0.2 M HCl, and 0.12 M NaOH in Figures 1–5. The pKa values are also determined from the solubility-pH profiles and are compared to those obtained with BS method.

pH-Dependent solubility studies

The solubility experiment was performed with a new small-scale shake flask method [23] at constant temperatures of 298.2 and 310.2 K. The shake-flask method proposed by Higuchi and Connors [24] is the most reliable and widely used solubility measurement method. This method determines thermodynamic solubility and could be carried out in several steps. Each drug was added in excess to 10 ml of dipo-tassium hydrogen phosphate (0.15 M) in a test tube. The test tubes were placed on a plate shaker. Using a pH-meter up to a stable pH (solubility equilibrium), the pH of each drug suspension was measured and adjusted if necessary with either diluted 0.2 M HCl, or

Table 1. Basic properties of drugs used in the investigations

Name of compound/ abbreviation

Systematic (IUPAC) name Molecular formula Molar mass M / g mol–1

Flufenamic acid/ FLU 2-{[3-(Trifluoromethyl)phenyl]amino}benzoic acid C14H10F3NO2 281.23

Mefenamic acid/MEF 2-(2,3-Dimethylphenyl)aminobenzoic acid C15H15NO2 241.30

Niflumic acid/ NIF 2-{[3-(Trifluoromethyl)phenyl]amino}nicotinic acid C13H9F3N2O2 282.22

Diclofenac sodium/ DIC 2-(2-(2,6-Dichlorophenylamino)phenyl)acetic sodium C14H10Cl2NO2Na 318.13

Meclofenamic sodium/MEC 2-[(2,6-Dichloro-3-methylphenyl)amino]benzoic acid, sodium salt C14H10Cl2NO2Na 318.13


117

0

0.1

0.2

0.3

270 280 290 300 310 320A

λ/nm

Figure 1. UV-Vis spectra for acidity constant measurement at temperature 310.2 K for flufenamic acid + solvent: () buffer, () 0.2 M HCl, () 0.2 M NaOH.

0

0.05

0.1

0.15

260 270 280 290 300 310 320

A

λ/nm

Figure 2. UV-Vis spectra for acidity constant measurement at temperature 310.2 K for mefenamic acid + solvent: () buffer, () 0.2 M HCl, () 0.2 M NaOH.

0

0.5

1

1.5

270 280 290 300 310

A

λ/nm

Figure 3. UV-Vis spectra for acidity constant measurement at temperature 310.2 K for niflumic acid + solvent: () buffer, () 0.2 M HCl, () 0.2 M NaOH.

0

1

2

3

4

250 270 290 310

A

λ/nm

Figure 4. UV-Vis spectra for acidity constant measurement at temperature 310.2 K for diclofenac sodium salt + solvent: () buffer, () 0.2 M HCl, () 0.2 M NaOH.


118

0

0.3

0.6

210 220 230 240A

λ/nm

Figure 5. UV-Vis spectra for acidity constant measurement at temperature 310.2 K for meclofenamic sodium salt + solvent: () buffer, () 0.2 M HCl, () 0.2 M NaOH.

0.2 M NaOH to a selected pH value. The experiment was completed when at least three pH measurements performed at an early, an intermediate and a late time-point of the 24 h period resulted in the same pH value. The equilibrium solubility was attained within 24 h. The pH values were also measured for the supernatants obtained after centrifugation, in order to certify that the drug solutions analyzed had the same pH value as the suspensions. Measurements were made using pH-meter (CPC-401 Elmetron) with an associated uncertainty of 0.01. The test-tubes were thermostated by a temperature control thermostat (Lauda A3, Germany) through the jacket of the vessel with uncertainty 0.1 K. Samples were withdrawn after 24 h, since a previous study had shown that a majo-rity of drugs reach their solubility equilibrium within this time-scale [7,23]. Furthermore, this time-scale often is used in drug development settings. Excess solid was equilibrated using a rotating-bottle appa-ratus (Hettich Zenrifugen, EBA 20) at 300000 rpm for 30 min. Following centrifugation, the supernatant was collected and used for solubility and pH determi-nations. The concentration of the drug in the super-natant solution was determined by an HPLC proce-dure with single-wavelength UV detection. All com-pounds were analyzed in quadruplicates at each investigated pH value, as it was shown in the pre-vious works [23,25,26].

The prerequisites in the selection of the buffer system were that the buffer should display an osmotic

pressure comparable to the osmotic pressure of the intestinal fluid (278 mOsm/kg) and that it should have an acceptable buffer capacity. Thus, 150 mM K2HPO4 (340 mOsm/kg) was chosen as buffer, and this buffer was mixed with pure drug to obtain the desired pH values. For pH below the pKa value the HCl was added and for the values above the pKa values, the NaOH was added [7].

HPLC Analysis

Each sample contained excess of drug and buf-fer solutions, sodium hydroxide or hydrochloric acid solution. Drug concentration in each sample was measured using an HPLC-UV-Vis apparatus delivered by Agilent Technologies, consisting of: 1200 Series Quat pump, 1200 vacuum degasser, 1200 DAD/ /MWD. A C18 analytical column (4.6 mm×150 mm) with a mean particle size of 5 μm was used. One buf-fer was prepared for mobile phase: disodium tetrabo-rate (0.01038 M Na2B4O7 and 0.01925 M NaCl). As a mobile phase two solutions were used: methanol (A) and borate buffer, pH 9 (B). Injection volumes of 5 μl were used during the analysis. The chromatographic conditions for all the drugs are shown in Table 2.

Data analysis

The modified HH equation may be used for the pH-dependent solubility prediction in two forms, for monoprotic acids:

Table 2. Condition during chromatography separation

Compound Mobile phase

Flow rate, ml min–1 λ / nm Methanol (A), % Borate buffer (B), %

FLU 75 25 1.5 290

MEF 75 25 1.5 210

NIF 65 35 1.5 290

DIC 80 20 1.0 280

MEC 70 30 1.0 210


119

pH p a0log log log(1 10 )KS S −= + + (1a)

and for monoprotic bases:

p a pH0log log log(1 10 )KS S −= + + (1b)

where S0 is the intrinsic solubility, S is the predicted solubility at a given pH, and pKa is a pH at which the concentration of unionized and ionized forms of a monoprotic drug in the solution are equal. The HH equation describes solubility as a function of pH. This equation always predicts an increasing solubility of all weak acids as pH increases, and descending solu-bility of bases when pH increases. In this paper all solubility values, S (mol dm–3) are expressed as log S and are calculated from the HH equation for mono-protic acids (Eq. (1a)). The intrinsic solubility, S0, was developed from the solubility-pH profile. Using the HH equation, the calculated curves were drawn for all experimentally studied drugs.


The pKa of an ionisable compound is an impor-tant property, describing the charge state of the drug at a certain pH. However, according to Avdeef [6], it is not recommended to determine pKa values from the solubility-pH measurements (pKaS). The results listed in Table 3 show a reasonably good agreement between the results obtained by the two different methods at two temperatures: from the solubility-pH measurements (pKaS), and with the precise spectro-photometric Bates-Schwarzenbach method (pKaB-S). The pKaS values were obtained from the crossing of two lines interpolated from the experimental points for the unionized and ionized form of the chosen drug.

The parameters and correlation coefficients are listed in Table 3. All the literature data of pKa values, obtained with different methods (including ours at 298.2 K for comparison) are presented in Table 4. Generally, our values are higher than those from lite-rature with the exception of MEF. In our opinion the Bates-Schwarzenbach method is more precise because it does not use the co-solvent and the extra-polated values to pure aqueous solutions.

The classical shake-flask method was applied to measure the equilibrium solubility at two temperatures for five drugs over a wide pH range from 2 to 8 or 9. The time of stirring was chosen as 24 h at the same temperature and pH to get the repeatable results. After stirring, the two phases (solution and solid mate-rial) of the saturated solution were separated, and after 6 h (the necessary time needed for the sepa-ration of two phases) the supernatant was taken out for the concentration of drug measurements by HPLC-UV-Vis spectrometry. The theoretical HH solu-bility, S, was predicted based on the pKa

B-S values and the intrinsic solubilities, S0 determined by the shake-flask method at two temperatures. Data are listed in Table 4 together with the literature data of S0. The results are discussed for each drug separately.

Flufenamic acid

During the last few years there were different values of pKa and S0 discussed in the literature at temperature 298.2 K for FLU. Our value of pKa = 4.62

[20] is close to the presented by Muñoz et al. [27], pKa = 4.17 and higher than all remaining data. In our opinion, the reason may also be insufficient control of the equilibrium temperature. The same parameters are responsible also for the values of the intrinsic solubility, S0. The value obtained in this work at T =

Table 3. The linear regression parameters, a1, b1 for the unionized form and a2, b2 for the ionized form, the corresponding correlation coefficients, R1 and R2, the pKa

S developed from the solubility-pH profile and the pKaB-S obtained from the Bates-Schwarzenbach method

Compound Parameter

pKaS pKa

B-S

a1 b1 R12 a2 b2 R2

2

298.1 K

FLU 0.159 -5.835 0.989 1.124 -10.290 0.987 4.62 4.62 [20]

MEF 0.134 -6.698 0.939 1.455 -11.818 0.998 3.87 3.88 [21]

NIF 0.100 -4.465 0.959 0.677 -7.035 0.991 4.45 4.42 [20]

DIC 0.061 -4.322 0.968 1.022 -9.767 0.997 5.66 5.70 [20]

MEC 0.079 -7.138 0.901 1.073 11.543 0.994 4.43 4.39 [22]

310.2 K

FLU 0.150 -4.996 0.874 1.143 -10.220 0.997 5.26 5.23

MEF 0.184 -6.159 0.976 1.690 -12.680 0.998 4.33 4.33

NIF 0.142 -4.466 0.939 0.743 -7.234 0.998 4.60 4.60

DIC 0.096 -4.387 0.972 0.768 -7.423 0.998 4.52 4.51

MEC -0.010 -6.631 0.912 1.036 -10.810 0.999 4.00 3.99


120

= 298.2 K, S0 = 7.40×10–6 mol dm–3 is close to that presented by Avdeef [5], S0 = 8.32×10–6 mol dm–3, or by Box et al. [28], S0 = 4.47×10–6 mol dm–3. The two other values presented in Table 4 are higher. New values, presented in this work at T = 310.2 K are: pKa = 5.23 and S0 = 5.20×10–5 mol dm–3. As for most of chemical compounds, with an increase of tempe-rature the solubility increases. The pKa also increases with an increase of temperature. Figure 6 shows the pH-equilibrium solubility profile for FLU at 298.2 (Figure 6a) and 310.2 K (Figure 6b). As the pKa values inform below pH 4.6 at 298.1 K and pH 5.2 at 310.2 K, a constant value of the solubility of this unionized form of weak acid is observed. Above these values of pH the compound transformed to the anionic form. At around pH 8 the horizontal plateau due to sodium-salt solubility was observed. The theo-retical HH calculated curve from Eq. (1a) closely follows the experimental points.

Mefenamic acid

The mefenamic acid is the only compound for which the solubility-pH profile at two temperatures (298.2 and 310.2 K) was presented in the open lite-rature [6]. The pKa = 4.33 value at 310.2 K is close to the published earlier pKa = 4.64 [6]. The other lite-rature values of pKa and intrinsic solubility are close to ours [28,29] or much higher i.e., S0 = 1.7×10–4 mol dm–3 [30] in comparison with our value S0 = 5.75×10–7

mol dm–3 at T = 298.2 K. Also the S0 = 3.31×10–4 mol dm–3 at T = 310.2 K [31] is far from our data S0 = = 3.55×10–6 mol dm–3. All data are presented in Table 4. The solubility-pH profile for MEF is shown in Figures 7a and 7b for 298.2 and 310.2 K, res-pectively. Below pH 3.8 and 4.3 at 298.2 and 310.2 K there is a constant value of S of the unionized acid form. Above these two pH values MEF is present in the solution as ionized form. At around pH 6.5 the horizontal plateau due to sodium-salt solubility was observed. A significant deviation from the HH eq. can be observed for the ionized form of MEF (see Figure 7a and 7b). The slope of the linear part of the pH- –solubility curve is far from the theoretical HH eq. defined as –1 for monoprotic weak acids. This can be explained as the salting out effect of phosphate counter-ions [4,7]. The pKa and solubility increase with an increase of temperature.

Niflumic acid

NIF is a popular drug for which the different values of pKa and S0 are presented in the literature at temperature 298.2 K. Our value of pKa = 4.42 [20] is close to the presented by Box et al. [28, 32] pKa = = 4.44 and higher than presented by Takács-Novák et al. [33], pKa = 2.28. The reason may be the different experimental method and the buffer used. The value of the intrinsic solubility at temperature 298.2 K, S0 = = 7.61×10–5 mol dm–3 is close to those presented by

Table 4. The literature values of pKalit, intrinsic solubility, S0lit, and experimental intrinsic solubility, S0

exp, developed from solubility-pH profile at two temperatures, 298.2 and 310.2 K

Compound pKalit S0lit/ mol dm–3 S0

exp/ mol dm–3

FLU 4.62B-S,a

3.97b,c

4.17d

3.84e

3.90f

4.47×10-6 b

8.32×10-6 g

3.24×10-5 h

2.38×10-5 i

7.40×10-6

5.20×10-5 (310.2 K)

MEF 3.88B-S,j

4.22b,c

4.5 g

4.64 (310.2 K)g

4.57×10-7b,c

8.70×10-8 g,k

1.70×10-4 i; 1.66×10-6 l

3.31×10-4 (310.2 K)l; 2.45×10-7 (310.2 K)g

5.75×10-7

3.55×10-6 (310.2 K)

(310.2 K)

NIF 4.42B-S,a

4.44b,m

4.86n

2.28n

3.39×10-5b,m

1.05×10-4 m

7.61×10-5

1.35×10-4 (310.2 K)

DIC 5.70B-S,a

3.99o,p

4.00o

3.90r

2.58×10-6 o, p 1.01×10-4

1.16×10-4 (310.2 K)

MEC 4.39B-S,s

4.10b

1.38×10-7b 1.58×10-7

2.11×10-7 (310.2 K) aRef. [20]; bRef. [28]; cRef. [29]; dRef. [27]; eRef. [35]; fRef. [36]; gRef. [6]; hRef. [37]; iRef. [30]; jRef. [21]; kRef. [5]; lRef. [31]; mRef. [32]; nRef. [33]; oRef. [34];pRef. [9]; rRef. [38]; sRef. [22]


121

-8

-6

-4

-2

2 4 6 8 10

log

S

pH

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (a) (b)

Figure 6. pH-dependent solubility profile of flufenamic acid (points are log S values measured by shake-flask method): a) experimental data at 298.2 K; b) experimental data at 310.2 K; the solid line was calculated with the Henderson-Hasselbalch equation; the dashed line

was linear interpolation of the experimental points.

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (a)

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (b)

Figure 7. pH-dependent solubility profile of mefenamic acid (points are log S values measured by shake-flask method): a) experimental data at 298.2 K; b) experimental data at 310.2 K; the solid line was calculated with the Henderson-Hasselbalch equation; the dashed line

was linear interpolation of the experimental points.

Box et al. [28,32] (see Table 4). The new values, presented in this work at T = 310.2 K are: pKa = 4.60 and S0 = 1.35×10–4 mol dm–3. The pKa and S increase when the temperature increases. Figure 8 shows the pH-equilibrium solubility profile for NIF at 298.2 (Figure 8a) and 310.2 K (Figure 8b). Below pH 4.4 at

298.2 K and pH 4.6 at 310.2 K a constant value of the solubility of the unionized form of NIF is observed. Above these values of pH the compound transformed to the anionic form. At around pH 7.5, the horizontal plateau due to sodium-salt solubility was observed. The theoretical HH calculated curve (Eq. (1a)) closely


122

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (a)

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (b)

Figure 8. pH-dependent solubility profile of niflumic acid (points are log S values measured by shake-flask method): a) experimental data at 298.2 K; b) experimental data at 310.2 K. The solid line is calculated with the Henderson-Hasselbalch equation; the dashed line is a

linear interpolation of the experimental points.

follows the experimental points for the unionized form and slightly deviates for ionized form. The pKa and solubility increase with an increase of temperature.

Diclofenac sodium

The sodium salt of diclofenac, DIC was used for the solubility measurements. The pKa values are 5.70 [20] at T = 298.2 K and 4.51 at T = 310.2 K. All lite-rature values at 298.2 K are much lower, which is shown in Table 4. Its intrinsic solubility is S0 = = 1.01×10–4 mol dm–3 at T = 298.2 K, which is much higher than that in literature S0 = 2.58×10–6 mol dm–3 [9,34]. The intrinsic solubility obtained by the shake-flak method in this work at T = 310.2 K is S0 = = 1.16×10–4 mol dm–3. Solubility increases as the tem-perature increases but the pKa decreases. In the later the inverse property is observed in comparison with weak acids. Figure 9 show the log S-pH profile of DIC. In solution at low pH DIC as an acid predo-minates; at high pH the ionized form of salt and asso-ciates of sodium cation predominate. The HH equa-tion excellent describes the solubility, especially at T = 298.2 K.

Meclofenamic sodium

The meclofenamic sodium salt, MEC was used for the solubility-pH profile measurements. The pKa values are 4.39 [22] and 3.99 at T = 298.2 K and T = = 310.2 K respectively. The literature value at T = = 298.2 K is close to that, pKa = 4.10 [28] (see Table 4). The intrinsic solubility is S0 = 1.58×10–7 mol dm–3 at T = 298.2 K, which is similar to that in literature S0 = = 1.38×10–7 mol dm–3 [28]. The intrinsic solubility obtained by the shake-flak method in this work at T = = 310.2 K is S0 = 2.11×10-7 mol dm–3. Solubility increases as the temperature increases but the pKa decreases as it was observed for DIC. Figure 10 shows the log S-pH profile of MEC. Below the pKa values a constant value of the solubility of the unionized form of MEC is observed. In solution at low pH MEC as an acid predominates; at high pH the ionized form of salt and associates of sodium cation predominate. The HH equation fits extremely well to measured solubility values at both temperatures.


123

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (a)

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (b)

Figure 9. pH-dependent solubility profile of diclofenac sodium (points are log S values measured by shake-flask method): a) ♦, experimental data at 298.2 K; b) , experimental data at 310.2 K. The solid line is calculated with the Henderson-Hasselbalch equation;

the dashed line is a linear interpolation of the experimental points at 298.2 K.

Some practical comparisons

The values of pKa increase as the temperature rises for weak acids and decrease for sodium salts. The dissociation reaction and the corresponding con-stants for a monoprotic weak acids and other sub-stances were defined in an excellent review paper [6]. The dissociation, dimerization, or aggregation cons-tant of different species in the solution is a tempera-ture dependent constant. The van’t Hoff equation shows that for a reaction that is exothermic under standard conditions (ΔrH < 0) the equilibrium constant K decreases as the temperature rises, which is observed for the measured sodium salts. The oppo-site occurs in the case of endothermic reactions, as for acids measured in this work.

The shake-flask method, used by pharmaceut-ical laboratories for the measurement of solubility of drugs in an aqueous buffer solution, assumes the thermodynamic equilibrium between the solid and liquid phase at constant temperature and pressure in saturated solution [4]. The dynamic solubility mea-surements at natural pH, or shake–flask method at natural pH used in physico-chemical, thermodynamic

models/laboratories also assumes the thermodyna-mic equilibrium between the solid and liquid phase at constant temperature and pressure in saturated solu-tion [22]. The only important difference between these two methods, as we can see, is an aqueous buffer solution in comparison with pure water. The solubility of drug in buffer solution reaches equilibrium at cons-tant pH after few to 24 h. During the first few hours the pH of solution changes due to ionization, salt for-mation, association of cation, or anion of compound, the salting out effects coming from the buffer used, the common ion effect, the equilibrium formation of different associates, self-association by forming mixed-charge micelles or micelle-like structures and possible aggregation of different species. The influ-ence of the kind of buffer used was recently discus-sed for the promethazine pH-equilibrium solubility [4]. Table 5 shows the results obtained at the same laboratory for the same drugs with two different methods in pure water solvent, ST and in an aqueous buffer solution at the same temperature, S, an ambient pressure and the same pH. In general the solubilities of weak acids is higher in an aqueous buffer solution


124

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (a)

-8

-6

-4

-2

2 4 6 8 10

log

S

pH (b)

Figure 10. pH-dependent solubility profile of meclofenamic sodium (points are log S values measured by shake-flask method): a) experimental data at 298.2 K; b) experimental data at 310.2 K. The solid line is calculated with the Henderson-Hasselbalch equation;

the dashed line is a linear interpolation of the experimental points at 298.2 K.

Table 5. Solubility of drugs derivatives of anthranilic acid in water obtained by dynamic method in natural pH 7, ST, and by shake-flask method in buffer solutions, S, at pH 7

Compound ST / mol dm–3 S / mol dm–3

298.2 K

FLU 3.38×10-5a 3.81×10-3

MEF <5.55·10-6b 4.87×10-3

NIF 9.11×10-5a 5.11×10-3

DIC 2.89×10-2a 2.47×10-3

MEC 0.903c 9.31×10-5

310.2 K

FLU 1.49×10-4a 6.00×10-3

MEF <5.55×10-6b 1.62×10-2

NIF 1.39×10-4a 7.30×10-3

DIC 0.202a 9.19×10-3

MEC 2.09c 2.75×10-4

aRef. [20]; bRef. [21]; cRef. [22]

but that of the sodium salts are lower. However, MEF can exists at pH 7 as a sodium salt, the solubility is three ranges of order higher than that coming from the dynamic method. It is evident that weak acids drugs tend to be more soluble in an aqueous buffer

solution than would be measured in pure water, or predicted by the thermodynamic models. The number of polar groups and hydrogen bond donors and acceptors always tend to have influence on the solu-bility [32].


125

CONCLUSIONS

This paper presents a systematic study of the pKa (at 310.2 K) and equilibrium solubility-pH profile measurements by the saturation shake-flask method of five drugs at 298.2 and 310.2 K. The solubility results depend on temperature and pH. Each of the investigated compounds reveals a specific pH-depen-dent solubility profile. It was shown that for some drugs as mefenamic acid or niflumic acid the HH equation has a limited applicability in phosphate buf-fer. The solubility of all drugs increases with an increase of pH and temperature. For all solutions above certain pH the characteristic plateau was observed. The pKa increases with an increase of temperature for weak acids and decreases for sodium salts.

Acknowledgements

Authors thank the Warsaw University of Tech-nology for funding and Dr. Svava Ósk Jónsdóttir for two-years helpful discussion and for funding of the meclofenamic sodium.

REFERENCES

[1] J. Devane, Pharm. Technol. 22 (1998) 68-80

[2] A.J. Humberstone, W.N Charman, Adv. Drug Deliv. Rev. 25 (1997) 103-128

[3] F. De Jaeghere, E. Allémann, R. Cerny, B. Galli, A.F. Steulet, I. Müller, H. Schütz, E. Doelker, R. Gurny, Pharm. Technol. 3 (2001) 1-8

[4] G. Völgyi, E. Baka, K.J. Box, J.E.A. Comer, K. Takács-Novák, Anal. Chim. Acta 673 (2010) 40-46

[5] A. Avdeef, S. Bendels, O. Tsinman, K. Tsinman, M. Kansy, Pharm. Res. 24 (2007) 530-545

[6] A. Avdeef, Adv. Drug Deliv. Rev. 59 (2007) 568-590

[7] C.A.S. Bergström, K. Luthman, P. Artursson, Eur. J. Pharm. Sci. 22 (2004) 387-398

[8] A. Avdeef, B. Testa, CMLS Cell. Molec. Life Sci. 59 (2002) 1681-1689

[9] A. Avdeef, C.M. Berger, Ch. Brownell, Pharm. Res. 17(1) (2000) 85-89

[10] L-Y. Lim, M-L. Go, Eur. J. Pharm. Sci. 10 (2000) 17-28

[11] G. Granero, M.M. de Bertorello, M.C. Briñón, Inter. J. Pharm. 190 (1999) 41-47

[12] X. Yu, G.L. Zipp, G.W.R. Davidson III, Pharm. Res. 11 (1994) 522-527

[13] S.D. Roy, G.L. Flynn, Pharm. Res. 6 (1989) 147-151

[14] K.A. Hasselbalch, Z. Die Biochem. 78 (1916) 112-144

[15] G.L. Amidon, H. Lennernas, V.P. Shah, J.R. Crison, Pharm. Res. 12 (1995) 413-420

[16] J. Dressman, Ch. Repass, Adv. Drug Deliv. Rev. 59 (2007) 603-616

[17] B. Faller, P. Erlt, Adv. Drug Delivery Rev. 59 (2007) 533- –545

[18] L. Du-Cuny, J. Huwyler, M. Wiese, M. Kansy, Eur. J. Med. Chem. 43 (2008) 501-512

[19] E. Baka, J.E.A. Comer, K. Takács-Novák, J. Pharm. Biomed. Anal. 46 (2008) 335-341

[20] U. Domańska, A. Pobudkowska, A. Pelczarska, J. Phys. Chem., B 115 (2011) 2547-2554

[21] U. Domańska, A. Pobudkowska, A. Pelczarska, M. Winiarska-Tusznio, P. Gierycz, J. Chem. Thermodyn. 42 (2010) 1465-1472

[22] U. Domańska, A. Pobudkowska, A. Pelczarska, P. Gierycz, J. Phys. Chem., B 113 (2009) 8941-8947

[23] C.A.S. Bergström, U. Norinder, K. Luthman, P. Artursson, Pharm. Res. 19 (2002) 182-188

[24] T. Higuchi, K.A. Connors, Adv. Anal. Chem. Instrum. 4 (1965) 117-212

[25] M. Mizutani, Z. Phys. Chem. 119 (1925) 318-326

[26] A. Li, S.H. Yalkowsky, J. Pharm. Sci. 83 (1994) 1735- –1740

[27] A. Muñoz de la Peña, N. Mora Díez, D. Bohoyo Gol, A.C. Olivieri, G.M. Escander, Anal. Chim. Acta 569 (2006) 250-259

[28] K.J. Box, G. Völgyi, E. Baka, M. Stuart, K. Takács-Novák, J. Pharm. Sci. 95(6) (2006) 1298-1307

[29] G. Ottaviani, D.J. Gosling, C. Patissier, S. Rodde, L. Zhou, B. Faller, Eur. J. Pharm. Sci. 41 (2010) 452-457

[30] E. Rytting, K.A. Lentz, X.-Q. Chen, F. Qian, S. Ven-katesh, AAPS J. 7 (2005) E78-E105

[31] Ch.N. TenHoor, V. Bakatselou, J. Dressman, Pharm. Res. 8 (1991) 1203-1205

[32] K.J. Box, J.E.A. Comer, Curr. Drug Metab. 9 (2008) 869- –878

[33] K. Takács-Novák, K.Y. Tam, J. Pharm. Biomed. Anlys. 21 (2000) 1171-1182

[34] N.T. Hansen, I. Kouskoumvekaki, F.S. Jorgensen, S. Brunak, S.Ó. Jónsdóttir, J. Chem. Inf. Model. 46 (2006) 2601-2609

[35] S.M. Sabry, Anal. Chim. Acta 367 (1998) 41-53

[36] M. El-Kemary, H. El-Shamy, M.M. Mosaad, Mater. Chem. Phys. 118 (2009) 81-85

[37] P.R. Duchowicz, A. Talevi, L.E. Bruno-Blanch, E.A. Castro, Bioorgan. Med. Chem. 16 (2008) 7944-7955

[38] U.A. Boelsterli, Toxicol. Appl. Pharmacol. 192 (2003) 307-332

[39] A. Avdeef, Pharm. Pharmacol. Commun. 4 (1998) 165- –178

[40] R.G. Bates, R. Gary, J. Res. NBS 65A (1961) 495-505

[41] D.M. Freeman, S.M. Hattangadi, T.F. Weiss, Auditory Neuroscience 3 (1997) 363-375

[42] W.L. Jorgensen, E.M. Duffy, Bioorg. Med. Chem. Lett. 10 (2000) 1155-1158

[43] T.T. Weiss, D.M. Freeman, Hearing Res. 111 (1997) 55-64.


126

A. POBUDKOWSKA

U. DOMAŃSKA

Department of Physical Chemistry,

Faculty of Chemistry, Warsaw University of Technology, Warsaw,

Poland

NAUČNI RAD

ISPITIVANJE RASTVORLJIVOSTI LEKOVA U VODI NA RAZLIČITIM pH VREDNOSTIMA

Rastvorljivost pet slabo rastvornih lekova u vodi je merena pri konstantnim temperaturama

(298,2 i 310,2 K) klasičnom metodom zasićenja u erlenmajeru. Sve testirane supstance su

derivati antranilne kiseline: flufenaminska kiselina (FLU), meklofenaminska kiselina (MEF),

nifluminska kiselina (NIF), diklofenak-natrijum (DIC) i meklofenamik-natrijum (MEC), i

poseduju anti-inflamatorno dejstvo. Pošto je rastvorljivost u vode jonizovanog leka znatno

veća nego nejonizovanog, ispitani su eksperimentalni uslovi koji utiču na vrednosti ravno-

težne rastvorljivosti, kao što je sastav vodenog pufera. Henderson-Hasselbalch jednačina

(HH) je korišćena za predviđanje zavisnosti rastvorljivosti izabranih lekova od pH na

pomenutim temperaturama. Za ovu svrhu su pKa vrednosti ispitivanih lekova utvrđene

Bates-Schvarzenbach spektrofotometrijskom metodom na temperaturi 310,2 K, s obzirom

na to da su vrednosti dobijene pri temperaturi od 298,2 K ranije publikovane. pKa dobijene

ovim metodama i merenjem rastvorljivosti imaju slične vrednosti.

Ključne reči: derivati antranilne kiseline, rastvorljivost, pKa, metoda erlenmajera, Henderson-Hasselbalchova jednačina.





127

ABHISHEK KUMAR SINGH MAUSUMI MUKHOPADHYAY

Department of Chemical Engineering, Sardar Vallabhbhai National Institute of Technology,

Surat, Gujarat, India

SCIENTIFIC PAPER

UDC 547.426.1:665.327.3:66

DOI 10.2298/CICEQ120626117S

RESPONSE SURFACE METHODOLOGY FOR OPTIMIZING THE GLYCEROLYSIS REACTION OF OLIVE OIL BY Candida rugosa LIPASE

Article Highlights • Glycerolysis of olive oil with an immobilized Candida rugosa lipase is reported • The response surface method employed for production of mono- and diglyceride • Effect of process parameters studied • The time, temperature, and lipase amount were observed to be most significant • Yield of mono- and diglyceride above 38 wt.% were obtained Abstract

In the present work, solvent free olive oil glycerolysis for the production of monoglycerides (MG) and diglycerides (DG) with an immobilized Candida rugosa lipase was studied. MG and DG production was optimized using experiment design techniques and response surface methodology (RSM). RSM based on five-level, a five-variable central composite design (CCD) was used to optimize MG and DG production: reaction time, temperature, molar ratio of glycerol to oil, amount of lipase, and water content in glycerol. The reaction time, temperature, and amount of lipase were observed to be the most significant factors on the process response. The immobilized C.rugosa lipase revealed optimum yield of MG and DG as 38.71 and 40.45 wt.%, respectively, following a 5 h reaction time with 0.025 g of lipase and 5% water content in glycerol at 40 °C temperature. The yield of MG and DG production can be enhanced 1.5 fold by RSM.

Keywords: monoglycerides, diglycerides, immobilized Candida rugosa lipase, response surface methodology, modeling.

Monoglycerides (MG) and diglycerides (DG) are anionic surfactants and emulsifiers with both hyd-rophilic and hydrophobic parts [1]. MG and DG are widely used in cosmetic, food, pharmaceutical, lubri-cant and chemical industry [2–6]. Commercial manu-facture of MG and DG are either by direct esterifica-tion of fats or oils and glycerol with or without inor-ganic alkaline catalysts (NaOH or Ca(OH)2) under a nitrogen atmosphere [7] or by chemical glycerolysis of fats and oils at higher temperatures (220–250 °C) [8]. The major drawbacks of this process are low yield, dark-colored and burnt-taste by-products. This pro-cess also requires product post-purification by mole-

Correspondence: M. Mukhopadhyay, Department of Chemical Engineering, Sardar Vallabhbhai National Institute of Tech-nology, Surat 395007, Gujarat, India. E-mail: [email protected];

[email protected] Paper received: 26 June, 2012 Paper revised: 27 November, 2012 Paper accepted: 28 November, 2012

cular distillation [9]. In recent year, synthesis of MG and DG using lipase as catalysts has been widely investigated. The use of lipase is advantageous for its stereo and positional specificities, and high catalytic efficiency [10]. The enzymatic glycerolysis synthesis is a possible alternative to chemical glycerolysis syn-thesis because of mild reaction conditions, improved product quality and lower energy consumption [9]. Olive oil is a source of long-chain monounsaturated fatty acids, such as, oleic acid (C18:1, 78%), linoleic acid (C18:2, 16%), palmitic acid (C16, 12%) and stea-ric acid (C18, 5%). Olive oil is produces a consider-able amount of MG and DG in glycerolysis reaction compared to the conventional oils [11].

Several lipase-catalyzed glycerolysis systems have been investigated without organic medium [1,12], with free [3,13,14] or immobilized lipase [1,12], in ionic liquids [15] or using compressed fluids as reaction media [16]. Many works reported on the gly-cerolysis of fats or oils with glycerol using lipase in the

A.K. SINGH, M. MUKHOPADHYAY: RESPONSE SURFACE METHODOLOGY… Chem. Ind. Chem. Eng. Q. 20 (1) 127−134 (2014)

128

presence of solvents [3,9,13,14]. In this system, the separation of organic solvents at the end of the reaction is difficult. In the present work, solvent-free olive oil glycerolysis for MG and DG production is reported. As no solvent separation step is involved, solvent-free glycerolysis lowers the final product cost. Efficient recovery of MG and DG without further puri-fication steps [17,18] is observed.

Response surface methodology (RSM) is a col-lection of statistical and mathematical techniques for designing experiments. RSM is useful for the model-ing and finding optimum conditions of factors for desirable response [19]. RSM also quantifies the rela-tionship between the controllable input parameters and the obtained response surfaces [1]. RSM has been recently used on modeling and optimization of lipase catalyzed reaction surfaces [1,20,21].

The optimization of solvent free glycerolysis of olive oil for MG and DG production with immobilized Candida rugosa lipase has not yet been described in full detail. The aim of this study is to identify relation-ship between the five input variables (reaction time, reaction temperature, amount of lipase, molar ratio of glycerol to oil and water content in glycerol) and MG, DG yield for the olive oil glycerolysis reaction and glycerol using immobilized C. rugosa lipase. The experimental design techniques and response sur-face methodology are used to optimize the glycerol-ysis reaction conditions for MG and DG production. The aim is also to develop predictive models that can be used in the design of new applications.


Materials

The immobilized (immobilized on macroporous acrylic beads) C. rugosa lipase, glycerol (99.9%) and olive oil were purchased from Sigma-Aldrich Che-micals Pvt. Ltd, Mumbai, India. Distilled water (Milli-pore, India) was mixed with glycerol. Standard for high-performance liquid chromatography (HPLC) analysis such as tripalmitate, dipalmitin, glyceryl tri-oleate 1, DL-α-palmitin, 2-diolein, 1,3-diolein and 1-oleoyl-rac-glycerol were purchased from Sigma-Ald-rich Chemicals Pvt. Ltd, Mumbai, India. The solvent used for analysis, e.g., acetone, acetonitrile, n-hex-ane, isopropanol and acetic acid of HPLC grade pur-chased from Merck, Mumbai, India.

Experimental procedure

The enzymatic glycerolysis reaction was perfor-med in a batch system. The immobilized C. rugosa lipase was placed in screw-capped flasks containing

glycerol and olive oil (gly:oil 1:1, 1.5:1, 2:1, 2.5:1 and 3:1). The reaction time, temperature, molar ratio of glycerol and oil (gly/oil), amount of lipase and water content in glycerol (3.5, 5.0, 6.5, 8.0 and 9.5%) were varied following experimental design. The reaction mixture was stirred in a water bath using a magnetic stirrer (IKA RCT Basic S22, Bangalore, India) at 600 rpm. Aliquot fractions (0.10 ml) of the reaction mixture were taken periodically. The lipase was separated by centrifugation (2900g) for 15 min. and the oil phase collected from reaction mixture. The excess of gly-cerol in oil phase was removed by washing with hot distilled water in equal amount of oil phase and re-centrifuged. The compositions of the glycerolysis pro-duct were determined by high-performance liquid chromatography. The free fatty acid (FFA) composi-tion of product obtained from the olive oil glycerolysis reaction was not investigated. Five random experi-mental conditions were selected and glycerolysis reaction was performed further in batch system to validate the developed model.

Analytical methods

Quantitative analyses of the product were car-ried out by HPLC (Agilent, USA Series 1100), with a Zorbax C18 column (4.6 m×250 mm, 5 µm) and a refractive index detector. The conditions were: detec-tor temperature 45 °C; column temperature 35 °C; flow rate 1.0 mL/min, mobile phase n-hexane and iso-propyl alcohol (4:5, v/v). The n-hexane and isopropyl alcohol were used as a sample dissolving solvent with injection volume of 40 µL. The quantification of MG and DG was carried out by comparing with MG and DG standard. Calibration charts were plotted, and the results were expressed as weight percentage of MG and DG.

Experimental design

The experimental design techniques for maxi-mum production of MG and DG are studied with cen-tral composite design (CCD). This method is sufficient to describe the majority of steady state process res-ponse and optimize the effective factor with a mini-mum number of experiments [22]. Reaction time (X1, h), temperature (X2, °C), amount of lipase (X3, g), glycerol to oil molar ratio (X4), and water content in glycerol (X5, wt.%) are selected as independent vari-ables and yield of MG and DG (wt.%) as dependent variables (Table 1).

The five independent variables are studied at five levels (−2, −1, 0, +1, +2) (Table 1). According to the full factorial five-level design, 32 experiments are required for optimization of the glycerolysis reaction (Table 1). It consists of 24 factorial CCD, with ten axial


129

points (α = 2) and six replication at the center points (number = 6). When response data are obtained from the test work, a regression analysis with Statistica (StatSoft, Inc., USA) is carried out with the second order polynomials equation to determine the coeffi-cients of the response model (A0, A1,…, An), their stan-dard errors and significance. For five variables under consideration, the response model is:

5 5 4 52

0

1 1 1 1

i i ii i ij i ji i i j i

Y b b X b X b X X= = = = +

= + + + (1)

where Y is the predicted response, Xi and Xj are the independent variables, b0 is the offset term, bi are

linear coefficients, bii are quadratic coefficients and bij are cross-product coefficients.

RESULT AND DISCUSSIONS

Statistical analysis and model fitting

Response surface optimization is more useful compared to the traditional single parameter optimi-zation as it saves reaction time, space and raw ma-terial [23]. The thirty two designed experimental points in terms of coded and un-coded independent variables are presented in Table 1. The factors that affect the production of MG and DG are time, tem-perature, amount of lipase, molar ratio of glycerol to

Table 1. The CCD matrix used for five independent variables and the comparison between experimental and predicted responses for MG and DG yields

Run Time (h), X1

Temperature (°C), X2

Amount of lipase (g), X3

Ratio of glycerol to oil X4

Water content (%), X5

MG content, wt.% DG content, wt.%

Experimental Predicted Experimental Predicted

1 -1(3) -1(20) -1(0.015) -1(1.5:1) 1(8) 14.77 13.49 23.03 22.35

2 -1(3) -1(20) -1(0.015) 1(2.5:1) -1(5) 16.67 16.97 28.81 28.59

3 -1(3) -1(20) 1(0.025) -1(1.5:1) -1(5) 20.58 20.49 26.43 27.16

4 -1(3) -1(20) 1(0.025) 1(2.5:1) 1(8) 14.56 13.60 24.02 22.38

5 -1(3) 1(40) -1(0.015) -1(1.5:1) -1(5) 21.03 20.00 25.17 24.71

6 -1(3) 1(40) -1(0.015) 1(2.5:1) 1(8) 18.27 16.38 23.80 20.97

7 -1(3) 1(40) 1(0.025) -1(1.5:1) 1(8) 32.45 30.17 35.93 34.05

8 -1(3) 1(40) 1(0.025) 1(2.5:1) -1(5) 27.81 27.11 31.63 30.21

9 1(5) -1(20) -1(0.015) -1(1.5:1) -1(5) 18.04 19.18 23.28 25.56

10 1(5) -1(20) -1(0.015) 1(2.5:1) 1(8) 13.11 13.38 20.38 20.28

11 1(5) -1(20) 1(0.025) -1(1.5:1) 1(8) 19.28 19.17 25.45 26.31

12 1(5) -1(20) 1(0.025) 1(2.5:1) -1(5) 16.99 18.46 22.78 24.09

13 1(5) 1(40) -1(0.015) -1(1.5:1) 1(8) 22.62 21.57 27.93 27.59

14 1(5) 1(40) -1(0.015) 1(2.5:1) -1(5) 20.45 20.98 25.21 25.33

15 1(5) 1(40) 1(0.025) -1(1.5:1) -1(5) 38.71 38.85 40.45 41.52

16 1(5) 1(40) 1(0.025) 1(2.5:1) 1(8) 25.12 24.40 33.20 31.90

17 -2(2) 0(30) 0(0.02) 0(2:1) 0(6.5) 18.16 21.33 24.41 28.03

18 2(6) 0(30) 0(0.02) 0(2:1) 0(6.5) 27.40 25.77 33.60 31.07

19 0(4) -2(10) 0(0.02) 0(2:1) 0(6.5) 10.27 9.11 20.82 18.97

20 0(4) 2(50) 0(0.02) 0(2:1) 0(6.5) 22.59 25.29 25.92 28.86

21 0(4) 0(30) -2(0.01) 0(2:1) 0(6.5) 18.21 18.92 23.54 24.08

22 0(4) 0(30) 2(0.03) 0(2:1) 0(6.5) 30.67 31.50 34.09 34.64

23 0(4) 0(30) 0(0.02) -2(1:1) 0(6.5) 26.27 27.75 34.47 33.10

24 0(4) 0(30) 0(0.02) 2(3:1) 0(6.5) 19.78 19.84 24.26 26.72

25 0(4) 0(30) 0(0.02) 0(2:1) -2(3.5) 31.95 30.27 35.49 33.21

26 0(4) 0(30) 0(0.02) 0(2:1) 2(9.5) 19.58 22.80 24.50 27.87

27 0(4) 0(30) 0(0.02) 0(2:1) 0(6.5) 32.79 32.73 35.21 35.02

28 0(4) 0(30) 0(0.02) 0(2:1) 0(6.5) 32.98 32.73 35.21 35.02

29 0(4) 0(30) 0(0.02) 0(2:1) 0(6.5) 32.88 32.73 35.21 35.02

30 0(4) 0(30) 0(0.02) 0(2:1) 0(6.5) 33.76 32.73 35.21 35.02

31 0(4) 0(30) 0(0.02) 0(2:1) 0(6.5) 32.98 32.73 35.21 35.02

32 0(4) 0(30) 0(0.02) 0(2:1) 0(6.5) 32.58 32.73 35.21 35.02


130

oil and water content in glycerol. The experimental and predicted value of MG and DG yield are also pre-sented in Table 1.

The corresponding analysis of variance (ANOVA), which indicates the significance of independent vari-ables, is done using square terms of the independent variables and first order interaction terms for each paired combination of independent variables for maxi-mum yield of MG and DG. The p-values are used to determine the effects (main or interaction) in the model which is statistically significant [24]. For MG production the linear coefficients X1–X5, quadratic term coefficients X1

2–X52 and the interaction coefficient

(X2X3) are found significant (p < 0.05). The first order interaction terms like X1X2, X1X3, X1X4, X1X5, X2X4, X2X5, X3X4, X3X5 and X4X5 are statistically insignificant terms (p > 0.1) for MG production, whereas for DG production the linear coefficients X1–X5, quadratic term coefficients X1

2–X52 and the interaction coefficient

(X2X3) are found significant (p < 0.05). The inde-pendent variable X1 and the first order interaction terms like X1X2, X1X3, X1X4, X1X5, X2X4, X2X5, X3X4, X3X5 and X4X5 are statistically insignificant terms (p > > 0.1) for DG production.

From the experimental design and results (Table 1), ANOVA and Eq. (1), the second order response functions representing the relationship between MG and DG yield and operating parameters, time, tempe-rature, amount of lipase, molar ratio of glycerol to oil and water content in glycerol are obtained as in Eqs. (2) and (3).

For MG yield model equation:

1 1 2 3

2 24 5 1 2

2 2 23 2 3 4 5

32.73 1.11 4.04 3.14

1.97 1.86 2.29 3.88

2.05 1.88 2.23 1.54

Y X X X

X X X X

X X X X X

= + + + −

− − − − +

+ − − −

(2)

For DG yield model equation:

2 1 2 3

2 24 5 1 2

2 2 23 2 3 4 5

35.02 0.76 2.47 2.64

1.59 1.33 1.36 2.77

2.24 1.41 1.27 1.12

Y X X X

X X X X

X X X X X

= + + + −

− − − − +

+ − − −

(3)

The predicted optimal levels of time, tempera-ture, amount of lipase, molar ratio of glycerol to oil and water content in glycerol of the glycerolysis reac-tion are calculated. The predicted yield of MG and DG along with experimental value is shown in Figure 1. The predicted values match the experimental values reasonably well with an R2 value of 0.96 and 0.97 for MG and DG yield. The quadratic regression model Eqs. (2) and (3) indicates that the models are ade-quate for prediction within the range of experimental variables.

Optimization of MG and DG yield

The effect of the process parameters and interaction parameter on the yield of MG and DG are presented in Figures 2–5. The three-dimensional surface plots (Figure 2) show the yields of MG and DG as a function of amount of lipase and time at fixed molar ratio of glycerol to oil (1.5), reaction tempera-ture (40 °C) and water content (5 wt.%). It can be seen from the plot that enhancing the lipase amount from 0.015 to 0.025 g could increase the yield of MG and DG, but further increase in lipase amount leads to a decline of MG and DG yield. Due to protein aggregation many of the active sites of the lipase molecules present are not exposed to the reactant at high lipase concentration, which decreases the MG and DG yield. The present results are in agreement with the previous study [12] where a similar trend of decreasing MG and DG production on increasing the

Figure 1. Correlation between experimental and predicted yield of A) MG and B) DG.


131

lipase amount observed. The effect of reaction time on the production of MG and DG is significant. The yield increases as the reaction time increase from 2 to 5 h, however, the yield decreases by further increas-ing reaction time. The yield relatively decreases due to reduction of enzyme activity after 5 h. Yang et al. [25] reported maximum MG production (17 wt.%) in lipase-glycerolysis of sunflower oil in a solvent-free system at 5 h reaction time.

Figure 3 shows the three-dimensional surface plots at different reaction temperatures and lipase amounts at fixed reaction time (5 h), molar ratio of glycerol to oil (1.5), and water content (5 wt.%). It can be seen that the yield of MG and DG increases with

the increase of lipase amount from 0.01 to 0.025 g, drops slightly from 0.025 to 0.03 g, increases rapidly with the increase in temperature from 10 to 40 °C and decreases rapidly from 40 to 50 °C. Temperature has a very important role in the enzymatic reaction sys-tem. Glycerol has a low miscibility with fats and oils. With increase in temperature, the mixture viscosity is reduced, and the mutual solubility or substrate diffu-sion process increases, thus reducing mass transfer limitations and favoring interaction between enzyme and oil. However, if the temperature is set too high, lipase denaturation can occur. Therefore, an optimal working temperature value should be selected. The result shows a decrease in the yield of MG and DG,

Figure 2. Response surface plot of production of A) MG and B) DG during glycerolysis of olive oil and glycerol by lipase, as a function of reaction time, lipase concentration, at X2 = 40 °C, X4 = 1.5:1 and X5 = 5 wt.%.

Figure 3. Response surface plot of production of A) MG and B) DG during glycerolysis of olive oil and glycerol by lipase, as a function of reaction temperature, lipase concentration, at X1 = 5 h , X4 = 1.5:1 and X5 = 5 wt.%.


132

with increase in the reaction temperature. Such a drop may be due to lipase denaturation occurring at elevated temperatures (above 40 °C). The reaction temperature of 40 °C is suitable for the glycerolysis of sunflower oil using Novozym 435 [25] in a solvent-free system.

Figure 4 shows the three-dimensional surface plots at varying reaction time and molar ratio at fixed reaction temperature (40 °C), amount of lipase (0.025 g), and water content (5 wt.%). It can be seen that the yield of MG and DG increases with increasing molar ratio of glycerol and oil from 1:1 to 1.5:1, but a further increase in molar ratio of glycerol and oil leads to a decrease in MG and DG yield. The decrease in reac-tion yields is presumably due to increased viscosity as the glycerol content increases. Glycerolysis repre-sents a complex reaction system that involves two phases, where the reaction can take place in one or both phases, or at its interface. The moderate agita-tion rate in glycerolysis reaction is necessary to eli-minate the mass transfer limitation and reduce the viscosity of the reaction mixture. Mass transfer between oil and glycerol phase directly relates to the interfacial area, which is dependent on the shear rate and the two phase volume ratio [15]. High contents of MG and DG (about 22 and 28 wt.%) at glycerol to oil molar ratio of 1:2 in the glycerolysis of butter oil at 45 °C in a solvent-free system [26] are reported. Tuter and Aksoy [27] found that 2:1 molar ratio of glycerol to palm kernel oil by Humicola lanuginose lipase pro-duces maximum MG (31 wt.%) and DG (42 wt.%). The yields of MG and DG increase rapidly with the increase of time from 1 to 5 h, and drop slightly when the reaction time is increased to 6 h.

The three-dimensional surfaces plots based on independent variable reaction time and temperature are shown in Figure 5, while the other three indepen-dent variables, amount of lipase, molar ratio of gly-cerol to oil, and water content are kept at 0.025 g, 1.5, and 5 wt.%, respectively. As can be seen, enhancing reaction temperature from 10 to 40 °C drastically enhances the yield of MG and DG to 38.7 and 40.45 wt.%. Here reaction time is the significant factor and the yield of MG and DG increases with increase with time until 5 h. Pawongrat et al. [14] reported the effect of temperature on the MG and DG production from tuna oil with IM-AK lipase enzyme in tert-butyl methyl ether. The MG and DG production are 25 and 42 wt.%, after 24 h at 45 °C.

Regression model verification

Five random experimental conditions (different from CCD) and within the range investigated were evaluated for validation of model. Table 2 presents experimental and predicted yield values for MG and DG. In all cases, the model prediction was in good agreement with the experimental data (considering the experimental error). The experimental and pre-dicted MG and DG yields at optimum levels of glyce-rolysis reaction condition were also evaluated. The CCD (from Table 2) enables finding the accurate values of the glycerolysis reaction parameters for production of MG and DG. The immobilized C. rugosa lipase has revealed the optimum yields of MG and DG as 38.71 and 40.45 wt.%, respectively, following a 5 h reaction time with 0.025 g of lipase and 5 wt.% water content in glycerol at 40 °C. This indicates that the regression model can efficiently predict the yield

Figure 4. Response surface plot of production of A) MG and B) DG during glycerolysis of olive oil and glycerol by lipase,

as a function of reaction time, molar ratio of glycerol to oil, at X2 = 40 °C , X3 = 0.025 g and X5 = 5 wt.%.


133

of MG and DG production. The yield of MG and DG can be enhanced from 23.9 and 30.3 wt.% in single variable optimization to 38.71 and 40.45 wt.% by RSM optimization giving approximately a 1.5 fold increase in MG and DG yield.

CONCLUSIONS

Immobilized Candida rugosa lipase was used as a catalyst in production of MG and DG from olive oil and glycerol in solvent-free system. RSM was used to determine the effect of process variables on the gly-cerolysis reaction and a second order response model was evaluated. Reaction time, temperature, and amount of lipase were the main factors affecting the glycerolysis reaction rate. The immobilized C. rugosa lipase revealed the optimum yield of MG and DG as 38.71 and 40.45 wt.%, respectively, following a 5 h reaction time with 0.025 g of lipase and 5 wt.% water content in glycerol at a temperature of 40 °C. The yield of MG and DG production can be enhanced approximately 1.5 fold by RSM optimization.

Acknowledgments

The authors wish to thank Dr. Sanjay M. Maha-jani, Chemical Engineering Department, Indian Insti-tute of Technology, Mumbai, India, for their HPLC analysis support of this research.

REFERENCES

[1] H. Ghamgui, N. Miled, A. Rehai, M. Karra-Chaabouni, Y. Gargouri, Enzyme Microb. Technol. 39 (2006) 717–723

[2] D.S. Negi, F. Sobotka, T. Kimmel, G. Wozny, R. Scho-macker, Ind. Eng. Chem. Res. 45 (2006) 3693–3696

[3] M. Kaewthong, S. Sirisansaneeyakul, P. Prasertsan, A. H-Kittikun, Process Biochem. 40 (2005)1525–1530

[4] M.A. Langone, M.E. de Abreu, M.J. Rezende, G.L. Jr Sant-Anna, Appl. Biochem. Biotechnol. 98 (2002) 987– –996

[5] Z. Guo, D. Kahveci, B. Ozcelik, X. Xu, New Biotechnol. 26 (2009) 37–43

[6] A.K. Singh, M. Mukhopadhyay, Appl. Biochem. Biotech-nol. 166 (2012) 486–520

[7] G.P. Mcneill, S. Shimizu, T. Yamane, J. Am. Oil Chem. Soc. 68 (1991) 1–5

Figure 6. Response surface plot of production of A) MG and B) DG during glycerolysis of olive oil and glycerol by lipase, as a function of reaction time, reaction temperature, at X3 = 0.025 g, X4 = 1.5:1 and X5 = 5 wt.%.

Table 2. Data for the validation of the theoretical predictions

Run Factor Yield of MG, wt.% Yield of DG, wt%

X1 X2 X3 X4 X5 Exp. Predicted Exp. Predicted

1 3 30 0.02 2:1 6.5 21.99 22.89 29.43 28.98

2 4 20 0.02 2.5:1 6.5 17.89 16.64 24.43 23.79

3 4 30 0.025 1.5:1 6.5 24.31 24.56 32.57 33.21

4 4 40 0.015 2:1 6.5 26.83 27.46 33.84 34.68

5 5 20 0.025 2:1 6.5 19.74 18.98 25.62 24.08


134

[8] J.B. Kristensen, X. Xu, H. Mu, J. Agric. Food Chem. 53 (2005) 7059−7066

[9] S. Ferreira-Dias, A.C. Correia, F.O. Baptista, M.M.R. Fonseca, J. Mol. Catal. B: Enzym. 11 (2001) 699–711

[10] H.L. Fan, Y. Chu, G.X. Yang, W. Zhang, J.L. Liu, Z.S. Wu, S.G. Cao, D.L. You, Ann. NY Acad. Sci. 864 (1998) 267–272

[11] D. Firestone, Bailey’s Industrial Oil and Fat Products, United States Food and Drug Administration, John Wiley & Sons, Inc. New York, 2005

[12] A.K. Singh, M. Mukhopadhyay, Grasas Aceites 63 (2012) 202–208

[13] M.L. Damstrup, J. Abildskov, S.Z. Kiil, A.D. Jensen, F.V. Sparso, X. Xu, J. Agric. Food Chem. 54 (2006) 7113– –7119

[14] R. Pawongrat, X. Xu, A. H-Kittikun, Food Chem. 104 (2007) 251–258

[15] Z. Guo, X. Xu, Green Chem. 8 (2006) 54–62

[16] M.A. Jackson, J.W. King, J. Am. Oil Chem. Soc. 74 (1997) 103–106

[17] A.R.M. Yahya, W.A. Anderson, M. Moo-Young, Enzyme Microb. Technol. 23 (1998) 438–450

[18] A. Valério, R.L. Kruger, J. Ninow, F.C. Corazza, D.D. Oliveira, J.V. Oliveira, M.L. Corazza, J. Agric. Food Chem. 57 (2009) 8350–8356

[19] H. Shafaghat, G.D. Najafpour, S. Rezaei, M. Sharifzadeh, Chem. Ind. Chem. Eng. Q. 16 (2010) 199–206

[20] L.V. Fregolente, C.B. Batistella, R.M. Filho, M.R.W. Maciel, J. Am. Oil Chem. Soc. 82 (2005) 673–678

[21] Y. Zheng, X.M. Wu, C. Branford-White, J. Quan, L.M. Zhu, Bioresour. Technol. 100 (2009) 2896–2901

[22] R. Azargohar, A.K. Dalai, Micropor. Mesopor. Mater. 85 (2005) 219–225

[23] S. Pinzi, J.M. Mata-Granados, F.J. Lopez-Gimenez, M.D. Luque de Castro, M.P. Dorado, Bioresour. Technol. 102 (2011) 1059–1065

[24] V.I. Vitanov, N. Javaid, D.J. Stephenson, Surf. Coat. Technol. 204 (2010) 3501–3508

[25] T. Yang, M. Rebsdorf, U. Engelrud, X. Xu, J. Food Lipid. 12 (2005) 299–312

[26] H.S. Garcia, B. Yang, K.L. Parkin. Food Res. Int. 28 (1996) 605–609

[27] M. Tüter, H.A. Aksoy.. Biotechnol. Lett. 22 (2000) 31–34.

ABHISHEK KUMAR SINGH

MAUSUMI MUKHOPADHYAY

Department of Chemical Engineering, Sardar Vallabhbhai National Institute of

Technology, Surat, Gujarat, India

NAUČNI RAD

OPTIMIZACIJA REAKCIJE GLYCEROLIZE MASLINOVOG ULJA POMOĆU LIPAZE Candida rugosa POMOĆU METODOLOGIJE POVRŠINE ODZIVA

U ovom radu je proučavana gliceroliza maslinovog ulja pomoću imobilisane lipaze

Candida rugosa u odsustvu rastvarača radi dobijanja mono- (MAG) i diacilglicerola (DAG).

Dobijanje MAG i DAG je optimizovano pomoću eksperimentalnog plana i metodologije

površine odziva (MPO). RSM, bazirana centralnom kompozitnom planu sa pet promenljivih

na pet nivoa, je korišćena za optimizaciju dobijanja MAG i DAG u odnosu na reakciono

vreme, temperaturu, molski odnos glicerol/ulje, količinu lipase i sadržaj vode u glicerolu.

Utvrđeno je da su reakciono vreme, temperatura i količina lipaze najznačajniji faktori.

Optimalni prinosi MAG i DAG pomoću imobilisane lipaze C. rugosa od 38,71 and 40,45%,

respektivno dobijeni su na 40 °C za 5 h korišćenjem 0,025 g lipaze i 5% vode u glicerolu.

Prinosi MAG i DAG se mogu povećati 1,5 puta pomoću RSM.

Ključne reči: monoacilgliceroli, diacilgliceroli, imobilisana lipaza Candida rugosa, metodologija površine odziva, modelovanje.





135

NATAŠA NEDELJKOVIĆ MARIJANA SAKAČ

ANAMARIJA MANDIĆ ĐORĐE PSODOROV

DUBRAVKA JAMBREC MLADENKA PESTORIĆ

IVANA SEDEJ TAMARA

DAPČEVIĆ HADNAĐEV

Institute of Food Technology in Novi Sad, University of Novi Sad,

Novi Sad, Serbia

SCIENTIFIC PAPER

UDC 664.69:66:664.641.2

DOI 10.2298/CICEQ120801125N

RHEOLOGICAL PROPERTIES AND MINERAL CONTENT OF BUCKWHEAT ENRICHED WHOLEGRAIN WHEAT PASTA

Article Highlights • Wheat blend containing 20% LBF expressed the most similar rheological behavior to

WWF as a control sample • The BWWP pasta possessed higher contents of P, Mg, K and Zn compared to WWP • The reduction in mineral content of BWWP during cooking was significantly higher

compared to WWP • The supplementation improved the mineral content of dry pasta, but not of cooked

pasta Abstract

Light buckwheat flour (LBF) was used to substitute 20% of whole wheat flour (WWF) in the formulation of wholegrain wheat pasta. Wholegrain wheat pasta (WWP) and buckwheat enriched wholegrain wheat pasta (BWWP) were produced on an industrial scale. The substitution level of buckwheat flour (20%) was based on previously conducted rheological tests on LBF/WWF blends, which were performed using 10, 20 and 30% of LBF. The obtained Mixolab profiles have indicated that wheat blend containing 20% LBF expres-sed the most similar rheological parameters to WWF. Proximate composition, cooking quality and mineral content of BWWP were analysed and compared with those of WWP. The substitution of WWF with LBF in the pasta formulation resulted in significantly increased (P < 0.05) contents of P, Mg, K and Zn com-pared to WWP in dry pasta. The reduction in mineral content of BWWP during cooking was significantly higher (P < 0.05) compared to WWP. The content of P, Mg and K were at same level in both type of pasta after cooking. The obtained results suggest that enrichment of WWP with LBF at the level of 20% did not improve the mineral content of cooked pasta, although an increase in minerals was observed in dry pasta.

Keywords: pasta, light buckwheat flour, dough rheology, Mixolab, mine-ral content.

Pasta products are consumed all over the world, and they are frequently manufactured from wheat flour, which is known to be the best raw material suitable for its production. The best properties of wheat flour for pasta production result from the pro-tein structure of wheat that enables to form a matrix with encapsulated starch granules. On the other hand, wheat flour is characterized by relatively low

Correspondence: M. Sakač, University of Novi Sad, Institute of Food Technology, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia. E-mail: [email protected] Paper received: 1 August, 2012 Paper revised: 30 November, 2012 Paper accepted: 6 December, 2012

lysine, methionine and threonine content, as well as some minerals and vitamins [1]. Fortification of wheat pasta in order to improve its nutritional quality and to produce functional pasta product has been described in the literature [2,3].

Buckwheat (Fagopyrum esculentum Moench) is an important pseudocereal known as a dietary source of protein containing high levels of essential amino acids [4], starch and dietary fibres [5], essential mine-rals [6] and trace elements [7]. This crop also con-tains antioxidant compounds, primarily rutin [8], which is responsible for beneficial health effects. There is some evidence that regular consumption of buck-wheat containing products may reduce the risk of high blood pressure, prevents oedema and hemorrhagic

N. NEDELJKOVIĆ et al.: RHEOLOGICAL PROPERTIES AND MINERAL… Chem. Ind. Chem. Eng. Q. 20 (1) 135−142 (2014)

136

diseases, prevents diabetes and reduces the risk of arteriosclerosis [9–11]. Also, buckwheat minerals have beneficial effects on health: magnesium may con-tribute to maintenance of normal muscle and nerve function, healthy immune function, and bone health; potassium may reduce the risk of high blood pressure and stroke, in combination with a low sodium diet; zinc is the component of many enzymes and its deficiencies lead to retarded development of children, skin affections, acne and weakening of taste; phos-phorus is an essential component of bones and teeth [12,13].

Buckwheat flour has been used for supplement-ation of wheat pasta in order to produce functional pasta with nutritional benefits [14–17]. Furthermore, buckwheat bran has been incorporated into spaghetti formulation to improve its quality [18,19].

The majority of research of buckwheat enriched pasta has focused on its mechanical strength and cooking quality related to decreased gluten matrix [20] and sensory properties [17,20].

The production and cooking of pasta influence its nutritional quality. Therefore, the supplementation of pasta with functional ingredient(s) has to be fol-lowed by availability of added compounds that con-tribute to desirable nutritional benefits. The loss of amino acids, i.e., lysine, during processing [21], as well as some minerals [19] and polyphenols [22] during cooking of pasta has been reported.

Knowing that production and cooking influence the mineral content of pasta, the aim of this work was to investigate if there is a possibility to produce buck-wheat enriched wholegrain wheat pasta with accept-able properties from the rheological point of view, and to determine the effect of processing and cooking on mineral content of pasta. The obtained information would be useful for optimization of process para-meters that would provide the stability of nutritional components at the time of pasta consumption.

MATERIAL AND METHODS

Raw materials

Light buckwheat flour (LBF) was obtained from Hemija Commerce, Novi Sad, Serbia, and whole wheat flour (WWF) was purchased from Žitko, Bačka Topola, Serbia.

Pasta dough formulation

Wholegrain wheat pasta (WWP) was produced using WWF and buckwheat enriched wholegrain wheat pasta (BWWP) was obtained by substitution of WWF with LBF at the level of 20%. WWF or WWF-

-LBF mixture was hydrated with deionised water to 320 g kg–1 absorption in order to achieve proper dough consistency [18, 23].

Rheological characteristics of pasta dough

The rheological behaviour of whole wheat flour, as well as WWF/LBF blends containing 10, 20 and 30% LBF were examined using Mixolab (Chopin, Tripette et Renaud, Paris, France). All measurements were performed using the modified Mixolab “Chopin +” protocol and applied parameters were: initial equi-librium at 30 °C for 8 min, heating to 90 °C for 15 min (at a rate of 4 °C/min), holding at 90 °C for 7 min, cooling to 50 °C for 10 min (at a rate of 4 °C/min) and holding at 50 °C for 5 min. The mixing speed was kept constant at 80 rpm [24]. Modification of the “Chopin +” protocol is due to dough weight increase from 75 g to 90 g because of the specific nature of the buckwheat flour [25].

Industrial pasta production

Two types of pasta (WWP and BWWP) were produced on an industrial scale by using Ital past Mac 60 (Parma, Italy). WWF or WWF-LBF mixture was hydrated and mixed in pre-mixer for 12 min. After that, the entire quantity was transferred to a mixer and mixed for 6 min. The obtained dough was extruded at the extrusion speed of 42 rpm as tagliatella for 42 min. The extruded tagliatella was dried in a dryer (Ital past D200, Parma, Italy) for 13.5 h at the temperature of 41.3 °C. The humidity was controlled since the automatic mode was used, and the final relative humi-dity was between 75 and 77%.

Proximate composition

The proximate composition of WWF, LBF, WWP and BWWP was analyzed using AOAC methods [26] for determining the moisture (14.004), crude protein (14.142), ash (14.006), crude cellulose (7.065), crude fat (14.019) and starch content (14.031).

Mineral composition

The mineral composition (P, Mg, K, Zn, Fe and Mn) of WWF, LBF, WWP and BWWP (uncooked and cooked) was determined using a Varian Spectra AA 10 (Varian Techtron Pty Limited, Mulgrave Victoria, Australia) atomic absorption spectrophotometer equip-ped with a background correction (D2-lamp). The sample preparation consisted of a dry ashing pro-cedure at 450 °C as described by Pavlović et al. [27].

Pasta cooking quality

Optimal cooking time. Pasta sample (100 g) was cooked in 1000 mL boiling and salted (5 g NaCl)


137

deionised water. Every 30 s, one noodle was taken out and pressed between two pieces of glass [28]. The cooking time was reached at the time when a white core could no longer be seen. This time was noted as an optimal cooking time and used for fol-lowing evaluations.

Cooking loss. During cooking, some parts of pasta dissolve in water. This cooking loss was eva-luated gravimetrically (90 min at 105 °C) by weighing the residues after evaporating a defined portion of cooking water [28]. Cooking loss was expressed as a percentage of the starting material.

Volume increase. The coefficient that represents the increase of pasta volume during cooking was calculated by dividing the volume of cooked sample with the uncooked sample. The volume measurement was performed by placing a certain amount of sample into a volumetric flask containing 1000 mL water and recording the height of water which represents the volume of the weighted sample [28].


All analyses were performed in triplicate, and the mean values with the standard deviations (S.D.) are reported. Analysis of variance and Duncan's multiple range test were used. Statistical data anal-ysis software system Statistica (StatSoft, Inc. (2011), version 10.0) was used for analysis. P values < 0.05 were regarded as significant.


Proximate composition of flours

The proximate compositions of the commercially available WWF and LBF are presented in Table 1.

The protein content was significantly higher (P < < 0.05) for LBF compared to WWF, followed by higher ash and fat content. The results obtained in the pre-sent study are comparable with the published data [5,7,29].

The pseudocereals are reported to contain higher levels of minerals in comparison to wheat [29]. Therefore, the contents of P, Mg and K in LBF were found to be much higher compared to WWF (Table 1). Bilgiçli [15] has detected that buckwheat flour was rich in K, Mg and P contents. It was reported that the levels of Mg, Zn, K, P and Cu in buckwheat flour were higher when compared with other cereals [30]. The obtained mineral contents of WWF and LBF are within the ranges reported by Steadman et al. [6] and Bonafaccia et al. [7].

Rheological characteristics of pasta dough

Physical and chemical changes take place at microstructural levels in dough when some part of wheat flour in the formulation is replaced with another type of flour [31], for example buckwheat flour. It has already been concluded that rheological tests on dough can predict material’s performance during pro-cessing [32]. Therefore, the rheological properties of wholegrain wheat dough and dough enriched with LBF were investigated using the Mixolab, which mea-sures dough behaviour during mixing and heating in a

Table 1. Proximate composition (%, dry basis) and minerals (mg/100 g) of whole wheat flour (WWF) and light buckwheat flour (LBF).Nitrogen-to-protein conversion factors are: 5.7 for WWF and 6.25 for LBF in case of crude protein (Nxfactor). Values are means of three determinations ± standard deviation. Values of the same row with the same superscript are not statistically different (P < 0.05)

Proximate composition WWF LBF

Moisture 11.7±0.02a 12.1±0.02b

Crude protein (Nxfactor) 13.6±0.12a 15.8±0.19b

Ash 1.46±0.01a 2.00±0.0b

Crude cellulose 1.55±0.03b 0.87±0.01a

Crude fat 1.52±0.02a 2.49±0.11b

Starch 66.6±0.18b 58.5±0.01a

Others 3.57±0.05 8.24±0.06

Minerals

P 253±4.26a 447±18.15b

Mg 81.9±1.38a 177±1.77b

K 272±5.38a 535±4.68b

Zn 2.09±0.11a 2.38±0.02b

Fe 3.51±0.09a 4.20±0.08b

Mn 2.79±0.11b 1.72±0.02a


138

single test simulating the mixing and baking pro-cesses. Hence, by using Mixolab it is possible to monitor both the protein and starch behaviour during processing, and it is well known that pasta dough rheology and cooking quality greatly depend on pro-tein content and nature, as well as amylose content and ratio of small starch granules [33–35].

Therefore, the feasibility of buckwheat flour as pasta ingredient was investigated by substituting 10, 20 and 30% of whole wheat flour for buckwheat flour.

Higher levels of substitution were not performed since the results obtained by Maeda [36] revealed that addi-tion of light buckwheat flour above 30% distinctly decreased the dough strength and sensory pro-perties.

Mixolab curves of WWF enriched with 10, 20 and 30% of LBF are presented in Figure 1, while the parameters are summarized in Table 2.

The first part of a Mixolab curve describes the protein characteristics of dough expressed as dough

Figure 1. Mixolab profiles of dough systems made by using whole wheat flour (WWF) and mixtures of whole wheat flour and light buckwheat flour at the level of 10, 20 and 30% (90WWF-10LBF, 80WWF-20LBF and 70WWF-30LBF).

Table 2. Mixolab properties of dough systems made by using whole wheat flour (WWF) and mixtures of whole wheat flour and light buckwheat flour at the level of 10, 20 and 30% (90WWF-10LBF, 80WWF-20LBF and 70WWF-30LBF). C2 – minimum consistency; C3 –peak torque; C3-C4 – breakdown torque; C5-C4 – setback torque. Values are means of three determinations ± standard deviation. Values of the same column with the same superscript are not statistically different (P < 0.05)

Dough type

Water ab-sorption, %

Development

time, min

Stability time min

C2 torqueN m

Pasting tem-perature, °C

C3 torqueN m

Peak tem-perature, °C

C3-C4 torque N m

C5-C4 torqueN m

100WWF 64.77±0.23d 5.20±0.55b 7.68±0.09b 0.32±0.01b 55.9±0.46a 1.91±0.04b 76.33±1.15a 0.54±0.03c 0.64±0.03a

90WWF-10LBF

64.07±0.12c 4.54±0.68ab 6.96±0.23a 0.23±0.02a 55.83±0.55a 1.78±0.04a 75.37±0.91a 0.44±0.05b 0.64±0.01a

80WWF-20LBF

63.07±0.12b 4.15±0.05a 7.03±0.21a 0.30±0.02b 57.2±0.9b 1.92±0.04b 78.97±0.15b 0.34±0.02a 0.68±0.03a

70WWF-30LBF

62.50±0.17a 3.72±0.03a 6.82±0.21a 0.23±0.02a 56.7±0.1ab 1.73±0.03a 78.07±1.05b 0.31±0.03a 0.64±0.05a


139

development time, water absorption, stability time and C2 value. Dough development time represents the time required to achieve appropriate consistency, which was defined as a torque of 1.1 N m. Stability (time until the loss of consistency is lower than 11% of the maximum consistency reached during mixing) indicates the dough resistance to kneading, while the C2 value represents the ability of the protein network to withstand both mechanical (kneading) and thermal (heating) influences. As it can be seen in Table 2, addition of buckwheat flour led to decrease in water absorption and dough development time. The high water absorption of the whole wheat flour is mainly related to the presence of water absorbing arabino-xylans [37]. Moreover, it was estimated that whole wheat flour was characterized with higher level of crude cellulose than buckwheat flour (Table 1). It was also found that the addition of LBF resulted in two peaks in the Mixolab curve at 30% substitution level. The same phenomenon was noticed by Maeda [36]. However, while in the research performed by Maeda [36] peaks have become prominent at 50% addition level, in this study the peaks were noticeable at lower substitution level (30%), which was probably inf-luenced by the difference in the used wheat flours (white wheat flour versus whole wheat flour). In gene-ral, Farinograms of strong wheat flours show a second peak after dough development time, since flour with weak gluten in which gliadin fraction prevail will express early dough development, while glutenin as a predominant fraction of strong flour will lead to dough stiffening [38]. Since buckwheat flour is cha-racterized by absence of gluten, the second peak observed in this study could not be ascribed to increase in glutenin fraction relative to gliadin, but probably to the nature of buckwheat hydrocolloids, which express slower hydrating process [39]. The addition of buckwheat flour to WWF resulted in lower stability, i.e., ability to resist deformation for longer time. However, although buckwheat flour possesses lower quality proteins from the technological point of view [40], the substitution of WWF with LBF at the level of 20% did not significantly affect the protein weakening (C2 parameter). In general, noncontinuous variation of the rheological parameters can be inter-preted as the result of interactions between system components [41]. Torbica et al. [25] have indicated the possibility of interactions between buckwheat and rice flour ingredients. In pasta processing, proteins of good mechanical properties that could resist extrusion process are required. In general, during extrusion, extrusion auger kneads the dough and forces it through the die, which leads to friction and heat

generation. Therefore, the C2 value could be the indi-cator of dough behaviour during extrusion, since it involves a combination of mechanical and thermal stresses.

The second part of the curve follows the changes in dough structure caused by temperature increase and mechanical forces of mixing. Following parameters were recorded: pasting and peak tempe-ratures, maximum torque at C3 point, C3–C4 value and C5–C4 value. Concerning the pasting properties, the C3 value, which represents the peak torque, was higher for wheat flour formulation compared to formul-ation containing 10 and 30% buckwheat flour, while the pasting temperature was lower in comparison to 20% LBF, probably due to lower lipid content of wheat flour (Table 1). According to Jane [42], lipids are known to form stable complexes with starch chains which restrict granule swelling. Difference in the behaviour of flour containing 20% buckwheat flour in comparison to 10 and 30% substitution level could be ascribed to possible interactions between composite flour components. The C3–C4 value (breakdown torque) is related to hot paste stability and enzymatic activity. It can be seen that wheat flour possessed a signi-ficantly higher (P < 0.05) C3–C4 value in comparison to buckwheat flour mixtures due to higher α-amylase activity in wheat flour. Ikeda et al. [43] proved that buckwheat seed contains an α-amylase inhibitor and therefore the addition of buckwheat flour resulted in lower C3–C4 values. The C5-C4 (setback torque) value represents the degree of starch retrogradation. The addition of buckwheat flour in dough formulations resulted in no significant changes in C5–C4 value, since pure buckwheat flour has only slightly lower setback torque in comparison to WWF [44].

Based on the results presented in Table 2, as well as the excellent matching of control and 20% LBF curves (Figure 1), it can be concluded that the rheological behaviour of these two investigated sys-tems was the most similar. Therefore, further investi-gations on pasta nutritional and technological quality were performed with the sample containing 20% of light buckwheat flour. Moreover, it was shown that the production of buckwheat enriched wholegrain wheat pasta (BWWP) in industrial conditions without modi-fying equipment and process parameters of whole wheat pasta production could be enabled.

Proximate composition of pasta

The proximate composition of wholegrain wheat pasta (WWP) and buckwheat enriched wholegrain wheat pasta (BWWP) are presented in Table 3. The pasta containing LBF is superior in protein, ash and


140

fat content than WWP due to higher content of these nutrients in LBF compared to WWF (Table 1). The obtained results were expected as buckwheat flour was reported to contain higher protein and mineral content than wheat flour.

Pasta cooking quality

Optimal cooking time was shorter for BWWP when compared to WWP (Table 3). This may be due to the physical disruption of gluten matrix which provided water absorption into the buckwheat enriched wholegrain wheat pasta. Similar observations were reported by Manthey et al. [18] and Chillo et al. [17]. The weakening of the gluten matrix in BWWP caused the significant increase in cooking loss of enriched pasta (Table 3).

Alamprese et al. [45] and Bilgiçli [14] also found that matter loss in cooking water was higher in pasta containing buckwheat. The obtained results could be considered as an acceptable cooking loss level and are in agreement with those reported by Dick and Youngs [46].

Moreover, Alamprese et al. [45] detected that buckwheat-containing pasta had a significantly higher weight increase during cooking than white wheat pasta. These authors explained this observation underlining the high non-starch polysaccharide con-tent of buckwheat with high water absorption capacity and the structure of buckwheat starch granules with irregular structure containing more amorphous areas than those of white wheat. However, the results of this study (Table 3) show that the volume increase for WWP and BWWP is not significantly different. This

can be explained by the fact of low substitution level in this pasta formulation.

Mineral content of pasta

Mineral contents of wholegrain wheat pasta (WWP) and buckwheat enriched wholegrain wheat pasta (BWWP) are presented in Table 4.

The substitution of WWF with LBF in the pasta formulation at the level of 20% resulted in significantly increased (P < 0.05) contents of P, Mg, K and Zn, which were expected considering the mineral content of WWF and LBF (Table 2). Previous studies showed that high K, Mg and P contents of buckwheat flour increased the mineral content of tarhana [14], gluten-free tarhana [47] and eriste [15], which were produced following the modified formulations that included buckwheat flour instead of a part of wheat flour. The increasing amount of buckwheat flour in gluten-free breads was resulted in increased macroelements content (Ca, Mg, P and K) [48].

Processing of dough into both types of pasta (WWP and BWWP) had little or no effect on mineral content (Tables 1 and 4). This finding was in accord-ance with observation of Manthey and Hall [19], who reported that pasta processing did not influence the mineral composition of pasta.

During cooking some amount of material is released into the cooking water. The loss of materials for nontraditional pasta is greater than for wheat pasta due to the weak gluten matrix of nontraditional pasta that permits a greater leaching of nutrients into the cooking water [20]. The reduction in mineral con-tent of BWWP during cooking was significantly higher

Table 3. Proximate composition (% dry basis) and cooking quality of wholegrain wheat pasta (WWP) and buckwheat enriched (20%) wholegrain wheat pasta (BWWP). Values are means of three determinations ± standard deviation. Values of the same column with thesame superscript are not statistically different (P < 0.05)

Pasta sample

Moisture Crude protein Ash Crude cellulose Crude fat Starch Optimal cook-ing time, min

Cooking loss, %

Volume increase

WWP 11.3±0.03b 13.1±0.11a 1.30±0.03a 1.46±0.01b 1.30±0.01a 65.9±0.18b 9.10±0.32b 7.83±0.65a 2.86±0.24a

BWWP 11.1±0.01a 14.1±0.08b 1.59 ±0.02b 1.36±0.02a 1.58±0.08b 64.1±0.01a 8.05±0.12a 9.33±0.40b 3.07±0.46a

Table 4. Effect of cooking on mineral content (mg/100 g dry basis) of wholegrain wheat pasta (WWP) and buckwheat enriched (20%) wholegrain wheat pasta (BWWP). Values are means of three determinations ± standard deviation. Values of the same column with thesame superscript are not statistically different (P < 0.05)

Pasta sample P Mg K Zn Fe Mn

Uncooked

WWP 268±0.83a 94.5±2.81a 277±0.32c 2.03±0.01a 3.92±0.12a 3.53±0.14a

BWWP 345±16.5b 125±5.20b 402±8.48d 2.32±0.01b 4.09±0.28a 2.82±0.07b

Cooked

WWP 253±12.1a 96.8±2.62a 47.7±0.02a 2.06±0.10a 3.92±0.10a 3.58±0.08a

BWWP 268±7.39a 96.0±1.09a 68.6±1.87b 2.43±0.16b 4.08±0.07a 2.87±0.09b


141

(P < 0.05) compared to WWP (Table 4). The reduc-tion in P, Mg and K in BWWP was 22, 23 and 88%, respectively. On the contrary, there were no signi-ficant reduction in P and Mg for WWP, but the reduc-tion in K was 83%. These results are in agreement with average reduction in minerals of about 28% in spaghetti containing 25% buckwheat bran flour pub-lished by Manthey and Hall [19]. These authors reported that the major reduction was detected for K (62%). Other literature data suggest that the decline of 10-30% in minerals during cooking of traditional pasta was typical, with the exception of K that leached out by 67% [49].

In contrast, the contents of Zn and Fe in both types of pasta were unaffected by cooking (Table 4). In general, the increased contents of P, Mg and K in enriched dry pasta that had been achieved by substi-tution of WWF with LBF were decreased and reached the same levels as in WWP after cooking. This implies that enrichment of WWP with LBF at the level of 20% did not improve the mineral content of cooked pasta, although increase in minerals was observed in dry pasta.

CONCLUSION

The substitution of WWF with LBF in the pasta formulation at the level of 20% resulted in significantly increased (P < 0.05) contents of P, Mg, K and Zn compared to the control pasta (WWP). The increased contents of these minerals in buckwheat enriched wholegrain wheat pasta (BWWP) declined and reached the same levels as in WWP after cooking. The obtained results suggest that enrichment of WWP with LBF did not result in significant impro-vement in mineral content of cooked pasta, although an increase in minerals was observed in dry pasta.

Acknowledgement

This paper is a result of the research within the project TR31029 financed by the Ministry of Edu-cation, Science and Technological Development, Republic of Serbia.

REFERENCES

[1] J. Heger, Z. Frydrich, Nutr. Rep. Int. 36 (1987) 1105-1112

[2] P. Rayas-Duarte, C.M. Mock, L.D. Satterlee, Cereal Chem. 73 (1996) 381-387

[3] R.L. Shogren, G.A. Hareland, Y.V. Wu, J. Food Sci. 71 (2006) 428-432

[4] G. Bonafaccia, M. Marocchini, I. Kreft, Food Chem. 80 (2003) 9-15

[5] V. Skrabanja, I. Kreft, T. Golob, M. Modic, S. Ikeda, K. Ikeda, S. Kreft, G. Bonafaccia, M Knapp, K. Kosmelj, Cereal Chem. 81 (2004) 172-176

[6] K.J. Steadman, M.S. Burgoon, B.A. Lewis, S.E. Edward-son, R.L. Obendorf, J. Sci. Food Agric. 81 (2001) 1094- -1100

[7] G. Bonafaccia, L. Gambelli, N. Fabjan, I. Kreft, Food Chem. 83 (2003) 1-5

[8] M. Holasava, V. Fiedlerova, H. Smrcinova, M. Orsak, L. Lachman, S. Vavreinova, Food Res. Int. 35 (2002) 207- –211

[9] S.L. Kim, S.K. Kim, C.H. Park, Food Res. Int. 37 (2004) 319-327

[10] G. Prestano, A. Pedrazuela, E. Penas, M.A. Lasuncion, G. Arroyo, Nutr. Res. 23 (2003) 803-814

[11] B. Filipčev, O. Šimurina, M. Bodroža-Solarov, M. Vuja-ković, Chem. Ind. Chem. Eng. Q. 17 (2011) 291-298

[12] IFIC – International Food Information Council Foundation. (2009). Background on Functional Foods: Food Insight. http://www.foodinsight.org/Content/6/functionalfoodsbackgrounder.pdf

[13] H. Gambuś, F. Gambuś, D. Pastuszka, P. Wrona, R. Ziobro, R. Sabat, B. Mickowska, A. Nowotna, M. Sikora, Int. J. Food Sci. Nutr. 60 (2009) 31-50

[14] N. Bilgiçli, LWT- Food Sci. Technol. 42 (2009) 514-518

[15] N. Bilgiçli, Int. J. Food Sci. Nutr. 60 (2009) 70-80

[16] G. Bonafaccia, I. Kreft, Fagopyrum 14 (1994) 35-42

[17] S. Chillo, J. Laverse, P. Falcone, A. Protopapa, M. Del Nobile, J. Cereal Sci. 47 (2008) 144-152

[18] F.A. Manthey, S.R. Yalla, T.J. Dick, M. Badaruddin, Cereal Chem. 81 (2004) 232-236

[19] F.A. Manthey, C.A. Hall, J. Sci. Food Agric. 87 (2007) 2026-2033

[20] R.P. Duarte, C.M. Mock, L.D. Satterlee, Cereal Chem. 73 (1996), 381-387

[21] J.E. Dexter, R. Tkachuk, R.R. Matsuo, J. Food Sci. 49 (1984) 225-228

[22] V. Verardo, D. Arráez-Román, A. Segura-Carretero, E. Marconi, A. Fernnádez-Gutiérrez, F. M. Caboni, J. Agric. Food Chem. 59 (2011) 7700-7707

[23] R. Schoenlechner, J. Drausinger, V. Ottenschlaeger, K. Jurackova, E.Berghofer, Plant Foods Human Nutr. 65 (2010) 339-349

[24] ICC, Approved Standard 173. International Association for Cereal Science and Technology, Vienna, 2006

[25] A. Torbica, M. Hadnađev, T. Dapčević, Food Hydro-colloids 24 (2010) 626-632

[26] AOAC, Official Methods of Analysis (13th ed.), Asso-ciation of Official Analytical Chemists, Washington, DC, USA, 1980

[27] A. Pavlović, Ž. Kevrešan, Đ. Kelemen-Mašić, A. Mandić, in Proceedings of the 8th Symposium on Analytical and Environmental Problems, Szeged, Hungary, 2001, pp. 117-122


142

[28] AACC, Approved Methods of the American Association of Cereal Chemists (9th ed.), American Association of Cereal Chemists, St. Paul, MN, USA, 1995

[29] L. Alvarez-Jubete, E. K. Arendt, E. Gallagher, Int. J. Food Sci. Nutr. 60 (2009) 240-257

[30] H.H. Wijngaard, E.K. Arendt, Cereal Chem. 83 (2006) 391-401

[31] N. Nikolić, J. Dodić, M. Mitrović, M. Lazić, Chem. Ind. Chem. Eng. Q. 17 (2011) 349-357

[32] B.J. Dobraszczyk, M.P. Morgenstern, J. Cereal Sci. 38 (2003) 229–245

[33] J.E. Dexter, R.R. Matsuo, Cereal Chem. 55 (1978) 44-57

[34] J.E. Dexter, R.R. Matsuo, Cereal Chem. 56 (1979) 190- –195

[35] M. Sissons, Food 2 (2008) 75-90

[36] T. Maeda, K. Miyake, M. Tahara, N. Morita, Fagopyrum 21 (2004) 99-103

[37] D.R. Lineback, V.F. Rasper, in Wheat: Chemistry and Technology, Y. Pomeranz, Ed., AACC, St. Paul, 1988, pp. 277–372

[38] D. Weipert, in: Future of Flour a Compendium of Flour Improvement, L. Popper, W. Schafer, W. Freund, Eds., Agrimedia, Bergen, Dunne, 2006, pp. 117-146

[39] T. Bojňanská, D. Urminská, M. Bennár, R. Gažar, Inter-national conference on food inovation, http://

//www.foodinnova.com/foodInnova/docu2/32.pdf, Valencia, Spain, 2010

[40] M. Mariotti, M. Lucisano, M. A. Pagani, S. Iameti, J. Agric. Food Chem. 56 (2008) 4258-4267

[41] S.M. Fitzsimons, D.M. Mulvihill, E.R. Morris, Food Hydrocolloids 22 (2008) 576–586

[42] J. Jane, in: Starch: Chemistry and Technology, 3rd ed., J. BeMiller, R. Whistler, Eds., Academic Press, Amsterdam, 2009, pp. 193-237

[43] M. Ikeda, K. Shida, M. Kisheda, Fagopyrum 14 (1994) 3-6

[44] T. Dapčević Hadnađev, A. Torbica, M. Hadnađev, Pro-cedia Food Science 1, 11th International Congress on Engineering and Food (ICEF11), Athens, Greece, 2011, pp. 328–334

[45] C. Alamprese, E. Casiraghi, M.A. Pagani, Eur. Food Res. Technol. 225 (2007) 205-213

[46] J.W. Dick, V.L. Youngs, in: Durum Wheat Chemistry and Technology, G. Fabrian, C. Lintas Eds., AACC, St. Paul, 1988, pp. 237-248

[47] N. Bilgiçli, Int. J. Food Sci. Nutr. 60 (2009) 1-8

[48] M. Wronkowska, D. Zielińska, D. Szawara-Nowak, A. Troszyńska, M. Soral-Śmietana, Int. J. Food Sci. Technol. 45 (2010) 1993-2000

[49] A.A.E. Yaseen, Nahrung 37 (1993) 449-455.

NATAŠA NEDELJKOVIĆ

MARIJANA SAKAČ

ANAMARIJA MANDIĆ

ĐORĐE PSODOROV

DUBRAVKA JAMBREC

MLADENKA PESTORIĆ

IVANA SEDEJ

TAMARA

DAPČEVIĆ HADNAĐEV

Naučni institut za prehrambene tehnologije Novi Sad, Univerzitet u

Novom Sadu, Novi Sad, Srbija

NAUČNI RAD

REOLOŠKA SVOJSTVA I SADRŽAJ MINERALNIH MATERIJA U INTEGRALNOJ PŠENIČNOJ TESTENINI OBOGAĆENOJ HELJDINIM BRAŠNOM

Belo heljdino brašno (LBF) korišćeno je za supstituisanje integralnog pšeničnog brašna

(WWF) u formulaciji za integralnu pšeničnu testeninu na nivou od 20%. Integralna

pšenična testenina (WWP) i integralna pšenična testenina obogaćena heljdinim brašnom

(BWWP) proizvedene su u industrijskim uslovima. Primenjeni nivo supstitucije (20%)

zasnovan je na prethodno sprovedenim reološkim ispitivanjima testa smeša LBF/WWF,

koje su sadržale 10, 20 i 30% LBF. Dobijeni Mixolab parametri ukazali su da je testo sa

20% LBF najsličnije testu od WWF po reološkim parametrima. Osnovni hemijski sastav,

svojstva pri kuvanju i sadržaj mineralnih materija određeni su u ispitivanim testeninama

(WWP i BWWP). Supstitucija WWF korišćenjem LBF u formulaciji za testeninu rezultirala

je signifikantnim porastom (P < 0,05) sadržaja P, Mg, K i Zn u nekuvanoj WWP testenini.

Redukcija sadržaja mineralnih materija u BWWP tokom kuvanja signifikantno je viša (P <

< 0,05) u poređenju sa WWP. Sadržaji P, Mg i K bili su na istom nivou u obe ispitivane

testenine nakon kuvanja. Dobijeni rezultati ukazuju da obogaćivanje WWP korišćenjem

LBF na nivou od 20% ne povećava sadržaj mineralnih materija kuvane testenine, mada je

porast sadržaja minerala zabeležen u nekuvanoj testenini.

Ključne reči: testenina, belo heljdino brašno, reologija testa, Mixolab, sadržaj mineralnih materija.





143

S.M. PEYGHAMBARZADEH A. HATAMI

A. EBRAHIMI S.A. ALAVI FAZEL

Department of Chemical Engineering, Mahshahr branch,

Islamic Azad University, Mahshahr, Iran

SCIENTIFIC PAPER

UDC 544.6:66:54

DOI 10.2298/CICEQ120707120P

PHOTOGRAPHIC STUDY OF BUBBLE DEPARTURE DIAMETER IN SATURATED POOL BOILING TO ELECTROLYTE SOLUTIONS

Article Highlights • Larger bubbles are generated in electrolyte solutions compared with water at similar

conditions • Increasing the salt concentrations makes the bubbles slightly larger • Bubble diameter decreases with increasing heat flux in all three electrolyte solutions • Larger bubbles appear during boiling of NaCl solutions since it has higher surface

tension Abstract

Bubble departure diameters during saturated pool boiling to pure water and three different electrolyte solutions including NaCl, KNO3 and KCl aqueous solutions are experimentally measured. Variable heat fluxes up to 90 kW/m2 and different salt concentrations from 10.6 to 69.6 kg/m3 are applied in order to investigate their effects on the bubble size during pool boiling around the hori-zontal rod heater. Visual observations demonstrate that larger vapor bubbles generate on the heat transfer surface at higher salt concentrations and lower heat fluxes in all of the solutions tested while in distilled water bubbles become slightly larger with increasing heat flux. Furthermore, the effects of different important physical properties like surface tension, viscosity, and density of the solutions on the bubble departure diameter are also discussed. NaCl solutions have surface tension higher than the other electrolyte solutions. Furthermore, the addition of NaCl to distilled water slightly increases the viscosity of the solution whereas other salts have no measurable effect on the viscosity. Therefore, larger bubbles are expected to appear on the heat transfer surface during the boiling of NaCl solutions, which is in agreement with the experi-mental results.

Keywords: bubble departure diameter, NaCl, KNO3, KCl, pool boiling, heat transfer.

In numerous applications in chemical pro-cesses, air–separation, refrigeration, and many other industries, heat is transferred or substances are sepa-rated by nucleate boiling. As a result, many experi-mental and theoretical studies on nucleate boiling heat transfer have been conducted. The vapor forms bubbles that begin growing at the heater surface and rise through the liquid after reaching a certain size.

Correspondence: S.M. Peyghambarzadeh, Department of Che-mical Engineering, Mahshahr branch, Islamic Azad University, Mahshahr, Iran. E-mail: [email protected] Paper received: 7 July, 2012 Paper revised: 8 December, 2012 Paper accepted: 8 December, 2012

Since the flow patterns in the boiling depend on the bubble formation and growth, the fluid motion and heat transfer processes are coupled. It is believed that heat transfer mechanisms responsible for the large boiling heat transfer coefficients are directly linked to the bubble activity on the heated surface. Therefore, almost all of the correlations developed to predict heat transfer coefficients in the nucleate pool boiling have a term related to bubble dynamics spe-cially bubble departure diameter (see e.g. correlations proposed by McNelly [1], Stephan and Abdelsalam [2], Benjamin and Balakrishnan [3], Jung et al. [4,5], Kedzierski [6], Vinayak Rao and Balakrishnan [7]). The bubble departure diameter (db) which is used in

S.M. PEYGHAMBARZADEH et al.: PHOTOGRAPHIC STUDY OF BUBBLE… Chem. Ind. Chem. Eng. Q. 20 (1) 143−153 (2014)

144

the above correlations is usually recommended to be calculated using the correlation of Fritz [8] as follows:

bl v

20.0146

( )d

gσθ

ρ ρ=

− (1)

Fritz [8] only considered the simple force balance between the surface tension force and the buoyancy force and obtained the above result. This correlation, which was used by several authors, showed acceptable prediction ability.

The bubble departure diameter also strongly influences heat exchanger fouling. It was confirmed by several authors that the major contribution towards the deposition of dissolved salts in water is due to the evaporation at the base of growing bubbles [9,10] and under most conditions, fouling is more severe during boiling heat transfer because of the mechanisms of bubble formation and detachment during boiling [11]. For example, it was shown that at higher surface temperature, greater fouling resistance occurs [10]. The variation in surface temperature varies the bub-ble size and also the frequency of bubble generation on the heat transfer surface. This phenomenon can consequently cause higher supersaturation beneath the vapor bubble and greater fouling resistance. Therefore, it is necessary to measure and understand bubble size during boiling conditions and large num-ber of investigations published until now confirmed this idea.

Kiper [12] described a method for determining the bubble departure diameter in saturated nucleate pool boiling. He showed that the minimum departure size varies only with the Jakob number. Yang et al. [13] established a correlation between bubble depar-ture diameter and bubble growth time and suggested a prediction formula for bubble departure diameter by considering the analogue between nucleate boiling and forced convection. Chen et al. [14] conducted visualization experiments for nucleate pool boiling of propane and isobutane on a horizontal smooth tube and two kinds of structured enhanced tubes. The effects of physical properties on the bubble dynamics were shown using different liquids. By comparing with propane, the departure diameter for iso–butane was larger due to the higher surface tension, and the nucleation site density was also higher, which is mainly caused by the much lower vapor density.

In one of the highly related paper, Jamialahmadi et al. [15] investigated pool boiling heat transfer to NaCl, Na2SO4, and KCl solutions. The effects of the dissolved salts on the nucleation site density, bubble departure diameter and bubble frequency were also investigated. The authors concluded that the pre-

sence of a small amount of electrolyte in water is sufficient to increase the bubble size in the low heat flux regime significantly and any further increase in electrolyte concentration has only a minor effect. The effect of heat flux on the mean bubble departure diameter found in that investigation on electrolyte solutions could be correlated by the following equa-tion:

b

1 0.0142596.75

ln

qd q

= + (2)

Kim and Kim [16] performed a quantitative anal-ysis of bubble departures during nucleate pool boiling using a characteristic bubble radius and time scales. Recently, Sarafraz et al. [17] studied subcooled flow boiling heat transfer characteristics of water and ethanol binary solutions in an annulus. The authors indicated that the bubbles were suppressed by increasing fluid flow rate and inlet subcooling.

Meanwhile, some investigations were performed on the measurement of bubble departure diameter in microgravity. For example, Straub [18] summarized the results of a series of microgravity experiments on pool boiling. A great amount of data of bubble depar-ture diameters have been compared with several correlations used in the literature. Zhao et al. [19] reported bubble behaviors observed in microgravity and presented a new semi-theoretical model of bub-ble growth which takes into account more factors acted on the growing bubble during pool boiling in different gravity conditions. Satisfying agreement between the observation and the predictions was obtained. Xiao and Yu [20] and Xiao et al. [21] pro-posed a fractal model for subcooled flow boiling heat transfer. The proposed fractal model was a function of wall superheat, liquid subcooling, bulk velocity of fluid, fractal dimension, the minimum and maximum active cavity size, the contact angle and physical properties of fluid.

In this paper, bubble departure diameters during pool boiling heat transfer to pure water and three different electrolyte solutions include NaCl, KNO3, and KCl aqueous solutions are measured and com-pared. Furthermore, the effects of heat flux and salt concentration and also the influences of the physical properties of the solutions on the bubble size are investigated.

EXPERIMENTAL

Apparatus and procedure

Figure 1 shows the complete pool boiling appa-ratus, consisting of a cylindrical stainless steel tank


145

(diameter of 0.25 m and height of 0.3 m) containing 38 liters of test liquid. Further, a vertical condenser is used to condense the evaporated liquid with tap water and recycle it to the pool. A large volume of liquid is considered to ensure that its temperature is not changed by the recycling flow. The test section is mounted horizontally within the tank and can be observed and photographed through observation glasses at both sides of the tank. The tank and con-denser are heavily insulated to reduce heat losses to the ambient air. The temperature in the tank is regul-ated by an electronic temperature controller and a variable electrical transformer in conjunction with a band heater covering the complete cylindrical outside surface. The pressure in the apparatus is monitored continuously and a pressure relief valve is installed for safety reasons. All the experiments were per-formed at atmospheric pressure. Boiling occurs at the outside of a cylindrical stainless steel test heater with a diameter of 10.67 mm, and a heated length of 99.1 mm. The test heater consists of an internally heated stainless steel sheathed rod, which is covered by a thick copper layer to generate uniform heat flux. The power supplied to the test heater could be calculated from the measured current and voltage drop.

Initially, the test section and the tank were cleaned and the system connected to a vacuum pump. Once the pressure of the system reached approximately 10 kPa, the test solution was intro-duced. Following this, the tank band heater was switched on and the temperature of the system allowed to rise. Once the system was de-aerated, it was left at the desired pressure and the corres-

ponding saturation temperature for about three hours to obtain a homogenous condition throughout. Then, the power was supplied to the test heater and kept at a predetermined value. Some runs were repeated later to check the reproducibility of the experiments, which proved to be excellent.

The sizes of bubbles are measured using image analysis software and the mean value of at least 10 bubbles was reported as the bubble departure dia-meter in that condition. In almost all conditions, the heating surface includes some oversized/undersized bubbles, which are related to bubble coalescence and breakage phenomenon and are excluded from mea-surements. Image analysis is performed in the area located at the height of 5 cm above the heating element. Bubble diameter was calculated using Sigma Scan software. This software enables mea-suring the projected areas of all the selected bubbles in the images. The average value of these areas equates with the area of equivalent circle to obtain bubble departure diameter. This measurement has uncertainty less than 5%.

The test liquids included different aqueous elec-trolyte solutions, i.e., NaCl, KNO3 and KCl solutions, which were used at different concentrations. Bubble departure diameters in NaCl, KNO3 and KCl solutions were measured at six various heat fluxes (below 90 kW/m2) and seven different salt concentrations (below 70 kg/m3). The boiling points of these solutions are shown are directly related to the salt concentration (Figure 2). All the physical properties of the solutions and their boiling point data were taken from Inter-national Critical Tables [22] except surface tension,

Figure 1. Simplified drawing of experimental setup.


146

which was measured experimentally. Surface tension was measured using the commonly utilized maximum bubble pressure method which was used by several authors (see, e.g., [23]).

RESULTS AND DISCUSSIONS

Bubble departure diameter in distilled water

Distilled water was used to check the accuracy of the experimental results and the calibration of the system for two reasons:

a) The physical properties of distilled water are well known with high accuracy.

b) Pool boiling heat transfer of distilled water and the related bubble dynamics has been studied by several investigators over a wide range of heat fluxes. Therefore, it can be used to check the accuracy of the experimental set up.

Figure 3 shows the bubble departure diameter of water as a function of heat flux. It can be seen that bubble departure diameter increases slightly with heat flux due to the rise in the wall superheat. Further-more, prediction of the correlation proposed by Fritz [8] and Ruckenstein [24] were also demonstrated in this figure. As can be seen in Figure 3, Ruckenstein [24] correlation shows better prediction especially for

Figure 2. Boiling points of different electrolyte solutions used in this study.

Figure 3. Bubble departure diameter as a function of heat flux for distilled water.


147

the effect of heat flux with absolute average error of about 16%. The experimental results are not in good agreement with the prediction of Fritz [8] correlation which is inherently independent of heat flux. This correlation overpredicts the experimental data with an absolute average error of 22%.

Figure 4 shows the appearances of heat transfer surface during pool boiling of water at three different heat fluxes. It is seen that the departing bubbles have irregular shapes at all values of heat fluxes. It is also shown that the number of generating bubbles increases with increasing heat flux. Furthermore, very large bubbles appear far from the heat transfer sur-face due to sever bubble coalescence in distilled water especially at higher heat fluxes.

Bubble departure diameter in electrolyte solutions

The results obtained for the bubble departure diameter of NaCl, KNO3 and KCl solutions are shown in Figure 5A-C, respectively. As can be seen, increas-ing heat flux causes the bubble diameter to be sharply reduced. This is in contradiction to the behavior of bubbles in distilled water boiling. Figure 3 shows that the bubble departure diameter increases with increasing heat flux in distilled water. Increasing heat flux causes more nucleation sites to be active and slightly larger bubbles generated in distilled water while in the electrolyte solutions the situations become reverse. The precence and accumulation of dissolved salts in the microlayer inhibits the activity of nucleation sites and less sites are active in the same heat flux. Therefore, larger bubbles are generated at same heat flux compared with distilled water.

When the heat flux increases in the electrolyte solutions, more nucleation sites become active and generate smaller bubbles. When a vapor bubble geneates in the electrolyte solution, it contains enti-rely water vapor. Therefore, the salt concentration in the vicinity of the heater becomes greater compared with the liquid bulk. This concentration difference pro-

duces diffusion of salt molecules from the bubble interface to the liquid bulk and reversely diffusion of water molecules from liquid bulk to the bubble inter-face. In these conditions, mass transfer of water from the bulk of liquid to the bubble interface becomes the limiting process. This phenomenon causes heat transfer reduction in the electrolyte solutions compare with distilled water as previously demonstrated by Jamialahmadi et al. [15]. This mass transfer inter-ference effect was also observed in the boiling of binary mixtures [25,26]. Alavi Fazel and Shafaee [27] performed an experimental study on pool boiling of electrolyte solutions. Their results show that the bub-ble detachment diameter increases with increasing either boiling heat flux or electrolyte concentration. Their results regarding the effect of heat flux on bub-ble diameter are in direct contradiction with the pre-sent investigation and the experimental study of Jamialahmadi et al. [15]. Although it is not highlited by Alavi Fazel and Shafaee [27], this conclusion is due to the subcooled pool boiling conditions existed in their experiments. In the subcooled pool boiling of electrolyte solutions, bubble departure diameter increases with increasing heat flux while in the satu-rated pool boiling bubble departure diameter decreases with increasing heat flux.

The effect of salt concentration on the bubble departure diameter is shown in Figure 6 at a constant heat flux. It is clear from Figure 6 that the bubbles get larger when a small amount of salt is added to distilled water. The addition of dissolved salts in water causes their accumulation in the microlayer beneath the vapor bubbles. It inhibits the activity of nucleation sites and fewer sites are active in the same heat flux. Therefore, to repel the heat from the heat transfer surface, larger bubbles are generated at the same heat flux compared to distilled water. Also, it is obvious that the bubble departure diameter in the NaCl solution is slighlty larger than two other elec-trolyte solutions in all the concentration investigated.

Figure 4. Photographs from the teat transfer surface during bubble generation in distilled water boiling at different heat fluxes.


148

(A)

(B)

(C)

Figure 5. Bubble departure diameter at different concentrations and heat fluxes in A) NaCl, B) KNO3 and C) KCl solution.


149

Figure 6. Effect of salt concentration on bubble dynamics in different electrolyte solutions.

Visual observations

Photographic studies of bubble dynamics are customary in the literature. The main objective of this study is to measure visually the bubble departure diameter of three kinds of electrolyte solutions in different concentrations and heat fluxes. Some of the selected photos taken from the heat transfer surface are presented in Figure 7. It is interesting to have different bubble appearances in these electrolyte solutions at same conditions. It is obvious that bub-bles in NaCl solutions are almost spherical, while in KCl and KNO3 solutions they have oblate, hemisphe-rical, and irregular shape. Also, the number of dis-persed bubbles in NaCl solutions seems to be greater than that in KCl and KNO3 solutions, while their sizes in NaCl solutions seem to be smaller. These obser-vations are in contradiction with the results presented in Figure 6, which indicate larger bubbles in NaCl solutions. It should be mentioned that larger bubbles in the photos taken from KCl and KNO3 solutions are generated as a result of bubble coalescence (merging of two or more small bubbles in the larger one). One can find large coalesced bubbles far from the heated surface while we measured the bubble diameter just at the departing time. Bubbles in the vicinity of the heat transfer surface in KCl and KNO3 solutions are smaller than those in NaCl solutions. Since larger bubbles appeared far from heated surface in KCl and KNO3 solutions compared with NaCl solutions, it may be concluded that more bubble coalescence may occur in the KCl and KNO3 solutions comparing with NaCl solutions. It was previously shown that coales-cence occurs easily in pure water and increasing con-centrations of added electrolytes there is a transition

to coalescence inhibition [28–30]. This point can be inferred from the results of the present study com-paring Figures 4 and 7. Bubble coalescence is related to the surface tension force and is discussed later in this paper.

Effect of different physical properties

The process of bubble formation is governed by many operating parameters and physicochemical properties such as liquid viscosity, liquid density, and nature of liquid (i.e., polar or nonpolar, etc.), which decide the mode of bubble formation and subse-quently reflects on bubble size. The main forces act-ing on a moving bubble are gravity, buoyancy, drag, viscous forces, added mass force, and the lift force. Fluid physical properties have the main responsibility for the magnitude of these forces. In this chapter, their influences on the bubble departure diameter have been discussed according to literature and the pre-sent experimental data.

Effect of surface tension

Apparently, the addition of a small amount of salt to water has no strong influence on the surface tension of the solution since these salts are not surface active agent. Effects of the addition of NaCl, KNO3 and KCl to distilled water on the surface tension of the solutions are presented in Figure 8A. As can be seen in Figure 8A, the addition of these salts slightly improves the surface tension of the solutions with respect to water (58.8 mN/m at 100 °C [22]) while further increase in salts concentrations has no mea-surable effect on the surface tension. Generally, NaCl solutions have higher surface tension than the other electrolyte solutions. Therefore, larger bubbles must


150

Figure 7. Heat transfer surface appearance during bubble generation in the boiling of electrolyte solutions at different concentrations and heat fluxes.

appear on the heat transfer surface during the boiling of NaCl solutions. The bubble diameter for the three electrolyte solutions previously shown in Figure 6 confirms this idea. Inoue et al. [23], in the study of pool boiling to water/ethanol binary mixtures, observed that the bubble departure diameter becomes small

and the number of nucleation site increases with an increase in the surfactant concentration (or reduction of surface tension). Wasekar and Manglik [31] also observed the same phenomena in pool boiling of aque-ous solutions with sodium dodecyl or lauryl sulfate.


151

(A)

(B)

(C)

Figure 8. Physical properties vs. concentration for different electrolyte solutions at the boiling point temperature: A) surface tension; B) viscosity; C) liquid density.


152

Effect of viscosity

The effect of viscosity on the various phenol-mena related to the boiling such as bubble formation, bubble growth and detachment, rising velocity from the heat transfer surface, and consequently the heat transfer coefficient that is related to these pheno-mena, has been little investigated. It is obvious from Figure 8B that the addition of NaCl to distilled water slightly increases the viscosity of the solution whereas KCl has no measurable effect on the viscosity. It should be mentioned that almost all of the corre-lations proposed for the prediction of bubble depart-ure diameter exclude the effect of viscosity [8,24,32,33]. One of the correlations that include the effect of viscosity of the liquid belongs to Kutateladze and Gogonin [34].

Effect of density

In all the electrolyte solutions used in this study, the vapor bubbles produced generally contain pure water vapor and therefore the vapor phase densities are the same. For the comparison, Figure 8C shows the liquid densities of these solutions as a function of the salt concentration. It can be seen that all of these electrolyte solutions have similar liquid densities and therefore, it can be concluded that this physical pro-perty has no measurable contribution in the bubble size and appearance.

CONCLUSION

In pure water, bubble diameter increases with increasing heat flux, whereas in all three electrolyte solutions bubble diameter decreases with increasing heat flux. This difference is due to the mass transfer resistance created in the bubble vicinity for pool boil-ing of electrolyte solutions.

• In electrolyte solutions, larger bubbles are generated on the heat transfer surface compared with distilled water at same heat flux. The presence and accumulation of dissolved salts in the microlayer inhi-bits the activity of nucleation sites and therefore, less sites are active at the same heat flux.

• Increasing the salt concentrations makes the bubbles slightly larger.

• NaCl solutions have higher surface tension than the other electrolyte solutions. Therefore, larger bubbles appeared on the heat transfer surface during the boiling of NaCl solutions.

• Bubbles in NaCl solutions are almost spheri-cal, while in KCl and KNO3 solutions they have oblate, hemispherical, and irregular shape.

Number of dispersed bubbles in NaCl solutions seems to be greater than that in KCl and KNO3 solutions, while their sizes in NaCl solutions seem to be smaller. It is due to the fact that increasing con-centrations of added electrolytes inhibits the coales-cence of bubbles.

REFERENCES

[1] M.J. McNelly, J. Imperial College Chem. Eng. Soc. 7 (1953) 18–34

[2] K. Stephan, M. Abdelsalam, Int. J. Heat Mass Transfer 23 (1980) 73–87

[3] R.J. Benjamin, A.R. Balakrishnan, Int. J. Heat Mass Transfer 39 (1996) 2495–2504

[4] D. Jung, Y. Kim, Y. Ko, K. Song, Int. J. Refrig. 26 (2003) 240–248

[5] D. Jung, K. Song, K. Ahn, J. Kim, Int. J. Refrig. 26 (2003) 764–771

[6] M.A. Kedzierski, Int. J. Refrig. 26 (2003) 337–348

[7] G. Vinayak Rao, A.R. Balakrishnan, Exp. Therm. Fluid Sci. 29 (2004) 87–103

[8] W. Fritz, Physik Zeitschr 26 (1935) 379–384

[9] S.M. Peyghambarzadeh, A. Vatani, M. Jamialahmadi, App. Therm. Eng. 50 (2013) 848-856

[10] S.M. Peyghambarzadeh, A. Vatani, M. Jamialahmadi, App. Therm. Eng. 39 (2012) 105–113

[11] M. Jamialahmadi, R. Blochi, H. Müller–Steinhagen, Chem. Eng. Process. 26 (1989) 15–26

[12] A.M. Kiper, Int. J. Heat Mass Transfer 14 (1971) 931–937

[13] C. Yang, Y. Wu, X. Yuan, C. Ma, Int. J. Heat Mass Transfer 43 (2000) 203–208

[14] Y. Chen, M. Groll, R. Mertz, R. Kulenovic, Exp. Therm. Fluid Sci. 28 (2004) 171–178

[15] M. Jamialahmadi, A. Helalizadeh, H. Müller–Steinhagen, Int. J. Heat Mass Transfer 47 (2004) 729–742

[16] J. Kim, M. H. Kim, Int. J. Multiphase Flow 32 (2006) 1269– –1286

[17] M.M. Sarafraz, S.M. Peyghambarzadeh, N. Vaeli, Chem. Ind. Chem. Eng. Q. 18(2) (2012) 315−327

[18] J. Straub, Adv. Heat Transfer 35 (2001) 157-172

[19] J.F. Zhao, G. Liu, S.X. Wan, N. Yan, Microgravity Sci. Technol. 20(2) (2008) 81-89

[20] B. Xiao, B. Yu, Int. J. Multiphase Flow 33 (2007) 1126- –1139

[21] B. Xiao, B. Yu, Z. Wang, L. Chen, Physics Letters, A 373 (2009) 4178–4181

[22] E.W. Washburn, International Critical Tables of Nume-rical Data, Physics, Chemistry and Technology, Knoven, 2003, p. 181

[23] T. Inoue, Y. Teruya, M. Monde, Int. J. Heat Mass Transfer 47 (2004) 5555–5563

[24] E. Ruckenstein, Bul. Institutului Politech. Bucaresti. 33 (1963) 79–88


153

[25] S.M. Peyghambarzadeh, M. Jamialahmadi, S.A. Alavi Fazel, S. Azizi, Braz. J. Chem. Eng. 26(1) (2009) 33–43

[26] S.M. Peyghambarzadeh, M. Jamialahmadi, S.A. Alavi Fazel, S. Azizi, Exp. Therm. Fluid Sci. 33 (2009) 903–911

[27] S.A. Alavi Fazel, S.B. Shafaee, J. Heat Transfer, Trans-actions of the ASME 132 (2010) 081502

[28] J. Zahradnik, M. Fialova, V. Linek, Chem. Eng. Sci. 54(21) (1999) 4757–4766

[29] Y.H. Tsang, Y.–H. Koh, D.L. Koch, J. Colloid Interface Sci. 275(1) (2004) 290–297

[30] C.P. Ribeiro Jr., D. Mewes, Chem. Eng. Sci. 62(17) (2007) 4501–4509

[31] V.M. Wasekar, R.M. Manglik, J. Heat Transfer 122 (2000) 708–715

[32] N. Zuber, Appl. Mech. Rev. 17 (1964) 663–672

[33] V.M. Borinshansky, F.S. Fokin, Trudy TsKTI 62 (1963) 1-9

[34] S.S. Kutateladze, I.I. Gogonin, High Temp. 17 (1979) 667–671.

S.M. PEYGHAMBARZADEH

A. HATAMI

A. EBRAHIMI

S.A. ALAVI FAZEL

Department of Chemical Engineering, Mahshahr branch, Islamic Azad

University, Mahshahr, Iran

NAUČNI RAD

FOTOGRAFSKO PROUČAVANJE PREČNIKA MEHURA PRI KLJUČANJU ZASIĆENIH RASTVORA

Prečnik mehura pri ključanju zasićene vode iz tri vodena rastvora elektrolita (NaCl, KNO3 i

KCl) je eksperimentalno meren. Toplotni fluks je menjan do 90 kW/m2, a koncentracije

rastvora od 10,6 do 69,6 kg/m3, radi ispitivanja njihovog uticaja na prečnik mehura tokom

ključanja zasićene tečnosti oko horizontalnog grejača oblika šipke. Vizuelno je uočeno da

se veliki mehuri pare generišu na površini grejača pri velikim koncentracijama rastvora soli

i manjim fluksevima toplote, dok u slučaju destilovane vode mehurovi postaju malo veći sa

povećanjem fluksa toplote. Takođe, diskutovani su uticaji različitih značajnih fizičkih oso-

bina (površinski napon, viskozitet i gustina rastvora) na prečnik mehura. Rastvori NaCl

imaju veći površinski napon od rastvora druga dva elektrolita. Dodatak NaCl destilovanoj

vodi blago povećava viskozitet rastvora, dok druge soli imaju zanemarljiv uticaj na visko-

zitet rastvora. Zbog toga se očekuje pojavljivanje većih mehura na površini grejača tokom

ključanja rastvora NaCl, što se slaže sa eksperimentom.

Ključne reči: prečnik mehura, NaCl, KNO3, KCl, ključanje zasićene tečnosti, prenos toplote.

january-march 2014 vol.20, number 1, 1-153

Documents