7434(Online) APRIL-JUNE 2018 Vol.24, Number 2, 93-200

ISSN 1451 - 9372(Print)ISSN 2217 - 7434(Online)APRIL-JUNE 2018Vol.24, Number 2, 93-200

www.ache.org.rs/ciceq

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

Vol. 24 Belgrade, April-June 2018 No. 2

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

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, Science and

Technological Development 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 Printing Technology, 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

Le Pham Tan Quoc, Nguyen Van Muoi, Physicochemical properties of Polygonum multiflorum Thunb. root powder produced with different carrier agents ..................... 93

E. Abdul Jaleel, K. Aparna, Identification of a heat-integrated distillation column using hybrid support vector regression and particle swarm optimization ....................... 101

Nebojša N. Ristić, Ivica R. Dodić, Ivanka P. Ristić, The influ-ence of surfactant structure on the dyeing of polyamide knitting with acid dyes ........................................................ 117

Bojana Ž. Bajić, Damjan G. Vučurović, Siniša N. Dodić, Zorana Z. Rončević, Jovana A. Grahovac, Jelena M. Dodić, The biotechnological production of xanthan on vegetable oil industry wastewaters. Part II: Kinetic modelling and process simulation ...................................... 127

Tutuk Djoko Kusworo, Hadiyanto Hadiyanto, Deariska Dear-iska, Lutfi Nugraha, Enhancement of separation performance of asymmetric cellulose acetate mem-brane for produced water treatment using response surface methodology .......................................................... 139

Dragan Psodorov, Vera Lazić, Marijana Ačanski, Đorđe Pso-dorov, Senka Popović, Dragana Plavšić, Kristian Pas-tor, Danijela Šuput, Zvonko Nježić, Fatty acid profile changes in ricotta-filled pastry during storage investigated by a GC/MS-ANOVA ...................................... 149

Faezeh Sharifi, Mansour Jahangiri, Investigation of the stability of vitamin D in emulsion-based delivery systems .............................................................................. 157

Shilin Huang, Juan Li, Chang-Feng Yan, Zhida Wang, Changqing Guo, Yan Shi, Synthesis and charac-terization of Cu-X/γ-Al2O3 catalyst by intermittent microwave irradiation for hydrogen generation from dimethyl ether steam reforming .......................................... 169

Natnirin Phanthumchinda, Tanapawarin Rampai, Budsa-bathip Prasirtsak, Sitanan Thitiprasert, Somboon Tanasupawat, Suttichai Assabumrungrat, Nuttha Thongchul, Alternative reverse osmosis to purify lactic acid from a fermentation broth ........................................... 179

Svetolik Maksimovic, Vanja Tadic, Jasna Ivanovic, Tanja Radmanovic, Stoja Milovanovic, Milica Stankovic, Irena Zizovic, Utilization of the integrated process of supercritical extraction and impregnation for incor-poration of Helichrysum italicum extract into corn starch xerogel ................................................................................ 191

Activities of the Association of Chemical Engineers of Serbia are supported by: - Ministry of Education, Science and Technological Development, Republic of Serbia - Hemofarm Koncern AD, Vršac, Serbia - Faculty of Technology and Metallurgy, University of Belgrade, Belgrade, Serbia - Faculty of Technology, University of Novi Sad, Novi Sad, Serbia - Faculty of Technology, University of Niš, Leskovac, Serbia - Institute of Chemistry, Technology and Metallurgy, University of Belgrade, Belgrade, Serbia

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 Jonjaua Ranogajec

Faculty of Technology, University of Novi Sad, Novi Sad, Serbia

Srđan PejanovićDepartment 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

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. 24 (2) 93−100 (2018) CI&CEQ

93

LE PHAM TAN QUOC1,2

NGUYEN VAN MUOI1 1Departement of Food Technology, College of Agriculture and Applied Biology, Can Tho University, Ninh

Kieu District, Can Tho City, Vietnam

2Institute of Biotechnology and Food Technology, Industrial

University of Ho Chi Minh City, Go Vap District, Ho Chi Minh City,

Vietnam

SCIENTIFIC PAPER

UDC 66.04:544:582.665.11:547.56.06

PHYSICOCHEMICAL PROPERTIES OF Polygonum multiflorum THUNB. ROOT POWDER PRODUCED WITH DIFFERENT CARRIER AGENTS

Article Highlights • Spray dried powder remains, TPC and AC of Polygonum multiflorum Thunb. root extract• TPC and AC values of GA after spray drying process were higher than that of MD • Physicochemical properties of spray dried powder were absolutely different with the

initial material Abstract

Polyphenol is a valuable compound found in plants. Unfortunately, it is quite sen-sitive to heat, light and oxygen in the air. This is a disadvantage making the sto-rage of these compounds for longer periods of time difficult. However, this problem can be overcome by encapsulation with carrier agents as maltodextrin, gum arabic, modified starch, etc. The efficiency of maltodextrin (MD, DE16-19) and gum arabic (GA) on spray drying of Polygonum multiflorum Thunb. root extract was investigated. The incorporation of gum arabic to the extract had the total polyphenol content (TPC) and antioxidant capacity (AC) higher than malto-dextrin. The obtained powders from gum arabic and maltodextrin were analyzed for encapsulation yield, moisture content, color parameters, total phenolic con-tent, antioxidant capacity, bulk density, wettability, hygroscopicity, water solubility index, particle size and microstructure. The results showed the types of carrier agents which significantly affected the physicochemical properties of powders produced by spray drying.

Keywords: carrier agent, gum arabic, maltodextrin, Polygonum multi-florum Thunb., spray drying.

It is well known that polyphenol from plant based foods and medicines has an important role in human health. Polyphenols in plants are antioxidant compounds which can combat many syndromes such as cancer, cardiovascular diseases, diabetes, intel-lectual disability, neurodegenerative disorders, etc. [1]. In addition, phenolic compounds are reducing agents and have high antioxidant capacity. There are many studies which have found these bioactive compounds safe and easy to use in medicine and food to replace

Correspondence: L.P.T. Quoc, Departement of Food Technology, College of Agriculture and Applied Biology, Can Tho University, Campus II 3/2 street, Ninh Kieu District, Can Tho City, Vietnam. E-mail: [email protected] Paper received: 29 March, 2017 Paper revised: 10 May, 2017 Paper accepted: 6 June, 2017

https://doi.org/10.2298/CICEQ170329021Q

synthetic antioxidants, especially natural antioxidants, in prevention of lipid oxidation in seafood [2].

Hà thủ ô đỏ (in Vietnamese), Polygonum multi-florum Thunb., is widely planted in Vietnam, China and Korea, and is one of the special herbal plants containing high levels of bioactive compounds in the root like polyphenol, gallic acid, resveratrol, catechin, physcion, rhein, emodin and more than 100 other compounds. For thousands of years, local people have used this wild plant as a traditional medicine for its anti-aging effects, hepatoprotective activities, effects against cancer, etc. [3]. In Vietnam, harvesting the root from this plant is quite difficult because P. multiflorum Thunb. is a wild herbal plant and is largely distributed in the mountainous region in the north of the country. Besides, it takes about 4 to 5 years for the plant to grow to be ready for harvesting [4].

L.P.T. QUOC, N.V. MUOI: PHYSICOCHEMICAL PROPERTIES OF P. multiflorum… Chem. Ind. Chem. Eng. Q. 24 (2) 93−100 (2018)

94

Nowadays, there are many methods to extract phenolic compounds from P. multiflorum Thunb., but the microwave-assisted extraction with acetone as solvent was used in this study because it can achieve high TPC and TEAC values and make the spray dried product good. Spray drying of the root extract may be a good method for these compounds to keep their quality and maintain low water activity and would make them easier to store and transport [5]. This method has been widely used for commercial pro-duction of dried vegetables, fruits and medicines. Moreover, spray drying is a highly suitable method for the oxygen- and light-sensitive components, espe-cially polyphenol. This process has been successfully applied for polyphenol stability in plant foods, for instance Morinda Citrifolia L. [6], guava leaves [7] and Emblica officinalis [8].

The changes of properties of the spray dried product depend on many factors of the spray drying process such as inlet/outlet temperature, feed flow rate, air flow speed, type of carrier agent, atomizer pressure, etc. However, we are interested in the phys-icochemical properties of the type of carrier agents in this study. There are many carrier agents, such as gum arabic, gelatin, modified starch, maltodextrin and whey protein isolate, which may serve as a drying aid to core encapsulation. Among them, gum arabic and maltodextrin are widely used for spray drying because they increase the stability of polyphenol and are com-mercially available and reasonably priced. Until now, there has been few published studies about the phar-maceutical values of P. multiflorum Thunb. root, no research studies about the spray drying process of its extract nor comparisons of physicochemical pro-perties of the encapsulating agents. Hence, the main aim of this research was to investigate the different physicochemical properties of maltodextrin and gum arabic before and after spray drying and to inves-tigate, under the same conditions of the spray drying process, the encapsulation yield, moisture content, colour parameters, total phenolic content, antioxidant capacity, bulk density, wettability, hygroscopicity, water solubility index, particle size and microstructure.

EXPERIMENTAL

Sample preparation

P. multiflorum Thunb. roots were harvested from the Cao Bang province, Vietnam. The roots were then washed with tap water, sliced and dried at 60 °C until the moisture level was less than 12%. The slices were then ground into fine powder (diameter less than 0.5 mm) and vacuum-packed.

Chemicals and reagents

Maltodextrin (MD, 16-19DE) was obtained from GPC, USA, and and gum arabic (GA) was supplied by Tianjin Dengfeng, China. Folin-Ciocalteu and DPPH (2,2-diphenyl-1-picrylhydrazyl, purity: ≥ 90%) rea-gents were purchased from Merck. A trolox (6-hyd-roxy-2,5,7,8-tetramethylchroman-2-carboxylic acid, purity: 97%) reagent was purchased from Sigma-Ald-rich, USA and all other chemicals and organic sol-vents were of analytical reagent grade.

Microwave-assisted extraction (MAE)

Polyphenols from dried powder of P. multiflorum Thunb. roots were extracted by a microwave system with acetone concentration of 57.35%, solid/solvent ratio of 1/39.98, extraction time of 289 s and micro-wave power of 127 W. The crude extract was filtered by Whatman paper. The filtered extract was evapor-ated at 45 °C to increase the extract concentration until a level of 4% of soluble solid was reached. The extract was then stored in a closed container at 4 °C [9].

Spray drying of Polygonum multiflorum Thunb. root extracts

Initial extracts after concentration by evaporation were at a level of 4 % of soluble solid. After that, gum arabic and maltodextrin were added into the extract with levels of soluble solid reaching 15 and 25%, res-pectively. The solution was well mixed and then fed into a Lab Plant SD-06 spray dryer. The feed flow rate, inlet/outlet temperature and air flow speed were approximately set at 500 mL/h, 160/70 °C and 5 m/s. After the spray drying process was completed, the dried powders were collected and vacuum-packed.

Encapsulation yield (EY)

EY was measured as the ratio of the dried mass of obtained powder to the dried mass of the initial substances, including the dried added carrier agent and the dried substances in the extract.

Determination of total polyphenol content (TPC)

The TPC in the extract was slightly modified and determined by the Folin-Ciocalteu colorimetric method [10]. The results were based on a standard curve obtained with gallic acid. TPC was expressed as mg of gallic acid equivalent per gram of dry weight (mg GAE/g DW).

Determination of antioxidant capacity (AC)

The AC of the extract was determined by DPPH assay which was adapted from Soto et al. (2014) [11] and slightly modified. Trolox was used as the stan-


95

dard. AC was expressed in TEAC (Trolox equivalent antioxidant capacity) determined as μmol of Trolox per gram of dry weight (μmol TE/g DW).

Bulk density

An amount of 2 g powder was added into an empty 10 mL graduated cylinder and shaked by a vortex vibrator for 1 min. The bulk density value was determined by the ratio of mass of the powder and the volume occupied in the cylinder [12].

Hygroscopicity

For hygroscopicity, 1.5 g of the powder was placed in a closed container containing the saturated solution of sodium carbonate. After 1 week, the pow-der sample was weighed and hygroscopicity was expressed as gram of adsorbed moisture per 100 g of powder [13].

Water solubility index (WSI)

Spray dried powder (2.0 g) and deionized water (25 mL) were vigorously mixed, incubated at 37 °C in a water bath for 30 min. Then the mixture was cen-trifuged for 20 min at 10000 rpm. The supernatant was separated and dried at 103±2 °C in an oven. The WSI (%) was expressed as the percentage of dried supernatant to the amount of the original powder [14].

Flowability

Flowability was determined by using measure-ment of angle of repose (AOR). The powder was poured slowly into the funnel which was held at a fixed height above the flat base. AOR was calculated by the height and radius of the powder on the flat base [15].

Wettability

The method to determine wettability was adapted from Freudig et al. (1999), with slight changes [16]. A volume of 100 mL water was first placed in a 250 mL beaker at room temperature. Then, 1 g of powder was gently tipped from a dish. The time for the whole amount of powder to visibly sink beneath the water surface was recorded as an indicator of wettability.

Color parameter

The color parameter consists of L* (lightness), a* (redness and greenness) and b* (yellowness and blueness) values, which were determined using a Chroma Meter CR-400 (Minolta, Japan).

Particle size

A laser scattering particle size distribution ana-lyzer (Horiba LA-960, Japan) was used to determine the particle sizes of the spray dried powder.

Scanning electron microscopy (SEM)

The morphology of the spray dried powder was examined by a Jeol JSM-7401F scanning electron microscope system. Samples were observed at 2000× magnification.

Data analysis

The experimental data was analyzed by the one-way analysis of variance (ANOVA) method and significant differences among the means from tripli-cate analysis at p < 0.05 were determined by Fisher’s least significant difference (LSD) procedure using Statgraphics software (Centurion XV). The values obtained were expressed in the form of a mean±stan-dard deviation (SD).

RESULTS AND DISCUSSION

Moisture, TPC, TEAC and EY of spray dried powder

Table 1 shows that the moisture of the two samples after spray drying significantly decreased in comparison with their corresponding initial materials (p < 0.05). The decrease of moisture depends on the properties of the carrier agent. Concentration of car-rier agent increases with the decrease of the moisture of the spray dried product. This result is in agreement with Kha et al. who spray dried Gac powder and increased concentration of MD from 10 to 20%. It resulted in a decrease of moisture content of the samples from 4.87 to 4.06% [17]. In addition, the moisture of the product also depends on the drying temperature where increasing drying temperature results in a significant drop in moisture content and

Table 1. TPC, TEAC, EY and moisture of spray dried powder; different superscript letters in the same column denote significant differ-ences (p < 0.05); MD-E: maltodextrin and extract after spray-drying; GA-E: gum arabic and extract after spray-drying

Sample Moisture, % TPC / mg GAE g-1 DW TEAC / µmol TE g-1 DW EY / %

Extract - 47.53±0.79c 334.07±3.04c -

MD 6.01±0.16c - - -

GA 10.97±0.03d - - -

MD-E 1.26 ±0.15a 20.54±1.12a 127.06±3.76a 60.4±2.01a

GA-E 4.01±1.15b 39.35±0.23b 146.97±2.13b 65.17±1.34b


96

the rate of heat transfer to the particle is greater. Besides, changes of moisture content are quite com-plicated when there are combinations of many carrier agents [18]. The moisture content of MD-E and GA-E in this study were 1.26 and 4.01%, respectively. This moisture is quite safe for storage because of low water activities.

High inlet temperature is a disadvantage for the spray drying process. It strongly affects TPC and TEAC of the received products. Compared with the initial extract, TPC and TEAC of MD-E retained approximately 43 and 38%; TPC and TEAC of GA-E retained 83 and 44%, respectively.

Although the sensitive bioactive compounds easily decomposed at high temperature, both of GA-E and MD-E retained TPC and TEAC. However, the received result shows that using GA as the carrier agent was more effective than MD because GA-E has the highest TPC and TEAC values. The reason for this is that the mean diameter of GA-E (19.17 µm) was higher than that of MD-E (17.52 µm); high dia-meter particles can contain a large amount of phe-nolic compounds. Tables 2 and 3 show the low WSI value and high wettability value of GA-E are advan-tages in protecting bioactive compounds inside the core. In other words, GA-E is more stable than MD-E. Therefore, the appearance of GA in the wall matrix can improve some properties of the product. Other research has yielded similar results, for instance TPC value of the wall matrix with 10% GA and 20% MD was higher than that of the wall matrix with 30% MD for spray drying process of guava leaf extract [7]. According to Tonon et al., powders produced with GA had the greatest polyphenol retention after the spray drying process of Euterpe oleraceae Mart. extract,

followed by the sample produced with MD 10 DE and 20 DE, while there was no significant difference in antioxidant capacity [19].

ANOVA results showed that there was a sig-nificant difference between the EY of spray dried pro-ducts (p < 0.05). Particularly, EY of GA-E was higher than MD-E. EY depends on many factors, for instance inlet/outlet temperature, type of carrier agent, feed flow rate, air flow speed, etc. This may be due to the volatilization of the active component [17]. EY of the spray drying process does not reach the maximum level because of the wet powder stuck to the upper part of the chamber wall. EY of product in this study was lower than encapsulation of passion fruit extract (75-78%) [20], higher than encapsulation of Morinda citrifolia L. leaf extract (39%) [6] and similar with encapsulation of Lippia sidoides leaf extract (66-69%) [21].

Bulk density, hygroscopicity, water solubility index and flowability of spray dried powder

There was a significant difference in the bulk density of the samples (p < 0.05). Bulk density of powder after spray drying process was lower than that of the initial powder. High inlet temperature results in the rapid formation of a dried layer at the droplet surface and case-hardening of the droplets occurs. Vapor bubbles and vapor-impermeable films appear on the drop surface. This leads to droplet expansion and decreases bulk density [22]. Besides, bulk den-sity also depends on total solid feed, feed flow rate, air flow rate, atomizer pressure and especially on the type of carrier agents [23]. Table 2 shows that bulk density of GA-E was lower than MD-E. However, MD- -E in this study was higher than the finding of Mishra et al. who used MD for spray drying Emblica offi-

Table 2. Bulk density, hygroscopicity, water solubility index and flowability of spray dried powder; different superscript letters in the same column denote significant differences (p < 0.05); MD-E: maltodextrin and extract after spray-drying; GA-E: gum arabic and extract after spray-drying

Sample Bulk density, g/mL Hygroscopicity, g/(100 g) Water solubility index (WSI / %) Flowability (repose of angle, °)

MD 0.71±0.01c 26.34±0.47a 93.50±0.71b 31.66±0.82b

GA 0.69±0.00b 26.82±0.47a 84.50±0.71a 31.13±0.60b

MD-E 0.69±0.00b 29±1.41b 94.50±0.71b 26.13±0.78a

GA-E 0.53±0.01a 37.16±0.1c 92.00±2.83b 37.66±2.75c

Table 3. Wettability and color parameter of spray dried powder; different superscript letters in the same column denote significant differences (p < 0.05); MD-E: maltodextrin and extract after spray-drying; GA-E: gum arabic and extract after spray-drying

Sample Wettability, s L* a* b*

MD 57±1a 99.44±0.43d -0.34±0.06a 1.28±0.02a

GA 365±12c 84.72±0.25b 2.84±0.11c 18.34±0.05d

MD-E 155±2b 93.84±0.95c 3.14±0.14d 3.00±0.06b

GA-E 461±11d 81.83±0.44a 2.26±0.03b 17.32±0.03c


97

cinalis fruit extract and obtained a bulk density of 0.52 g/mL [8]. In addition, dextrose equivalent value and moisture content increase with the increase of high bulk density due to its stickiness [5].

Hygroscopicity of spray dried powder increased strongly, especially GA-MS. The links between the hydrogen present in water molecules and the hyd-roxyl groups available in the amorphous and crys-talline regions of subtract also affect hygroscopicity, especially MD-E. MD-E and GA-E consist of a high number of hydrophilic groups and thus easily bind water molecules from the surrounding air [5]. Hygro-scopicity of MD-E and GA-E augmented approx-imately 10 and 39%, respectively. In addition, particle size significantly affects hygroscopicity; the smaller the particle size, the bigger the exposed surface and this leads to greater water adsorption from the sur-rounding air. The results were in agreement with Tonon et al. who spray dried Euterpe oleraceae Mart. fruit extract [19]. Besides, inlet air temperature inc-reased with the decrease of the moisture of products. There is a great moisture gradient between the dried product and the surrounding air so it is quite easy for the product to absorb the moisture from the surround-ing air. The high hygroscopicity is a disadvantage for storage but the result of this study was lower than that of other researchers such as Ersus and Yurdagel who spray dried carrot extract by MD and hygroscopicity achieved ranged from 72.83 to 83.33 g/(100 g) [24], while Mishra et al. used MD to encapsulate Emblica officinalis fruit extract and achieved hygroscopicity ranging from 46.03 to 53.01 g/(100 g) [8].

Table 2 shows that WSI of MD and MD-E were not significantly different (p<0.05). Conversely, WSI of GA-E changed dramatically, increasing from 84.5 to 92%. WSI of GA was lower than MD because GA has high viscosity, which makes it difficult to dissolve in water. Bigger particle size resulted in the smaller exp-osed surface, reduced contact with the continuous phase and low value WSI. WSI increases with inc-reasing concentration of the carrier agent and inlet temperature [25] or decreasing dry air flow rate [5]. Besides, the type of the carrier agent can significantly affect WSI. In general, WSI of dried products was quite high, from 92 to 94.5%; these values were higher than WSI of tomato powder (17.65–26.73%) [26], WSI of Gac powder (36.91–38.25%) [17] and similar to Ginger powder (93.82%) [25].

The measurement of the angle of repose (AOR) can determine the flowability of particles. AOR of MD dropped from 31.66 to 26.13° while AOR of GA inc-reased from 31.13 to 37.66° (Table 2). AOR inc-reases with a decrease in flowability and flowability of

GA-E was lower than MD-E. According to Carr (1965, 1970), flowability of MD-E in this study was better (AOR < 30°) and GA-E was cohesive (AOR: 30–40°) [27,28]. The flowability of particles depends on many factors such as storage temperature, moisture of par-ticle and relative humidity. The dried powder abs-orbed moisture from the surrounding air on particle surface. It tends to dissolve soluble components and form liquid bridges between particles making them more cohesive [29]. The shape and particle size also affect flowability; the particle size decreases with an increase in surface area per unit mass and leads to reduction in flowability. Frictional forces resist the flow and more surface area is available for cohesive forces [30].

Wettability and color parameter of spray dried powder

Table 3 shows that there was a significant dif-ference among the wettability of the powder samples (p < 0.05). Carrier agents also had different wet-tability, but wettability of GA was quite higher than MD, reaching 365 s. After spray drying, wettability values of MD-E and GA-E increased sharply and achieved 155 and 461 s. Wettability depends on shape, particle size and type of carrier agent. Besides, the moisture of the powder also strongly affects wettability. Normally, the moisture of initial materials is higher than in spray drying produce and there is the aggregation of dispersed material to mat-erial units of larger size. Then, the water penetrates easily into the pores of the powder and shortens the wettability time [31]. This result was considerably different from Fernandes et al., who used GA and modified starch to spray dry rosemary essential oil and obtained wettability ranging between 301 and 254 s [32]. This study was also different from Wu et al. who spray dried skim milk powder at various feed solid contents and obtained wettability ranged between 45 and 79 s [33].

The results of color analysis are described in Table 3. Color parameters (L*, a*, b*) of carrier agents were the considerably different. MD was lighter than the rest of the samples; MD-E and GA-E were darker than initial MD and GA. The a* and b* value of samples also changed considerably. Changes were due to the color of the extract which was mixed with carrier agents. In addition, color parameters were affected by inlet temperature, feed flow rate, air flow rate and soluble solid content [34]. In addition, many phenolic compounds degrade at high inlet tempera-ture but some phenolic compounds remain, such as tannins, which can react slowly with iron in the abs-ence of oxygen and form dark-colored complexes [8].


98

Smaller particle size and the increased surface area cause rapid pigment oxidation [35]. This leads to changes in the color of the samples.

Microstructure and particle size of spray dried powder

MD-E and GA-E were the encapsulating agents enabling the formation of homogeneous particles. Dried powder particles have spherical shapes with many wrinkles on their surfaces, especially GA-E. The rest of GA-E particles have many indentions and wrinkles, while MD-E has smooth surfaces and less wrinkles (Figure 1). The formation of these inden-tations is usually attributed to particle shrinkage due to the drastic loss of moisture during the spray drying process which is followed by cooling in a cyclone [18,19]. In addition, concave wrinkled surfaces have high surface and increase frictional forces. They change physical properties of particles such as AOR, bulk density, hygroscopicity, water solubility index, etc. The microstructure of particles depends on the inlet temperature and the type of carrier agent. According to Phisut, when the inlet air temperature is high, the particles have a smooth surface, causing faster water evaporation which leads to the formation of a smooth and hard crust; decreasing drying tem-perature results in a larger number of particles with a

shriveled surface [5]. Microstructure of the encapsul-ating agents in this study was found to be in agree-ment with Pham et al., who used MD and the com-bined MD and GA to spray dry guava leaf extract [7], and Pang et al. who spray dried Orthosiphon stami-neus extract by MD and whey protein isolate [36].

Figures 2 and 3 show that particle size of the initial carrier agent fluctuated strongly from 1.5 to 600 µm for MD (45 sizes) and from 7 to 500 µm for GA (32 sizes). Mean diameters of MD and GA were 99.66 and 123.71 µm, respectively. All sizes of MD and GA were under 6.2%.

Particle size of MD-E and GA-E changed signi-ficantly and were smaller than that of MD and GA. Diameters of MD-E ranged between 0.15 µm and 260 µm (56 sizes), mean diameter of 17.52 µm and each size was under 9%, while diameters of GA were sep-arated into 2 areas; one area had particles size from 0.115 to 0.6 µm (13 sizes), cumulative percent of 26 and the other area has particle size from 2 to 260 µm (37 sizes), cumulative percent of 74%. Mean diameter of GA-E was 19.17 µm. The results show that the dia-meter of particle decreases after the spray drying pro-cess. Particles size depends on the type of carrier agent, inlet temperature, feed flow rate [22], atomizer pressure, feed total solids, air flow rate [23] and vis-

Figure 1. SEM microphotographs of MD-E (a) and GA-E (b) at 2000× magnification.

Figure 2. Particle size distribution curve of MD (a) and MD-E (b).


99

cosity of the encapsulated material [13,37]. Changes of particles size lead to changes in the physical properties of the powder such as bulk density, shape, flowability and dispensability.

CONCLUSION

All P. multiflorum Thunb. powder samples were produced by the spray drying method with different carrier agents which had low moisture content (1.26- –4.01%). This was an advantage for storage. MD-E had the highest bulk density and WSI, while GA-E had the highest encapsulation yield, hygroscopicity, flowability, wettability and mean diameter. The color of dried powder changed completely and the micro-structures of the powder were spherical with many shriveled surfaces, especially GA-E. The TPC and TEAC retention of GA were higher than those of MD after the spray drying process. These results show that the type of carrier agent is a very important factor that strongly affects physicochemical properties of the dried product. Therefore, using GA as the carrier agent was the best choice in this study because there are three main factors that have the highest values: TPC, TEAC and EY.

Acknowledgments

The authors wish to thank Nguyen Thi Hai, Dinh Thi Tuyet Hang, Ta Thi Kim Hue and Le Thi Thu Thao for their support.

REFERENCES

[1] R.A. Vacca, D. Valenti, S. Caccamese, M. Daglia, N. Braidy, S.M. Nabavi, Neurosci. Biobehav. Rev. 71 (2016) 865–877

[2] S. Maqsood, S. Benjakul, A. Abushelaibi, A. Alam, Compr. Rev. Food Sci. Food Saf. 13 (2014) 1125–1140

[3] L. Lin, B. Ni, H. Lin, M. Zhang, X. Li, X. Yin, C. Qu, J. Ni, J. Ethnopharmacol. 159 (2015) 158–183

[4] T.L. Đỗ, Những cây thuốc và vị thuốc Việt Nam, Nhà xuất bản Y Học, Hà Nội (2004), p. 1274. (in Vietnamese)

[5] N. Phisut, Int. Food Res. J. 19 (2012) 1297–1306

[6] D. Krishnaiah, R. Sarbatly, R. Nithyanandam, Chem. Eng. Res. Des. (2012) 622–632

[7] M.T. Pham, T.V.A. Hoang, T.H.C. Le, N.Q.C. Vu, T.B.O. Dang, B.K. Dang, D.T.M. Pham, H.T. Mai, T.T.S. Nguyen, S.M. Dam, T.N. Tran, Ann.: Food Sci. Technol. 17 (2016) 34–40

[8] P. Mishra, S. Mishra, C.L. Mahanta, Food Bioprod. Pro-cess. 92 (2014) 252–258.

[9] L.P.T. Quoc, N.V. Muoi, J. Sci. Tech. 53 (2015) 1–11

[10] A. Siddiqua, K.B. Premakumari, R. Sultana, S. Vithya, Int. J. ChemTech Res. 2 (2010) 205-208

[11] C. Soto, E. Caballero, E. Pérez, M.E. Zúñiga, Food Bio-prod. Process. 92 (2014) 328–333

[12] A.M. Goula, K.G. Adamopoulous, N.A. Kazakis, Dry. Technol. 22 (2004) 1129–1151

[13] Y.Z. Cai, H. Corke, J. Food Sci. 65 (2000) 1248–1252

[14] R.A. Anderson, H.F. Conway, V.F. Pfiefer, J.R. Griffin, Cereal Sci. Today 14 (1969) 4–11

[15] D. Geldart, E.C. Abdullah, A. Hassanpour, L.C. Nwoke, I. Wouters, China particuol. 4 (2006) 104-107

[16] B. Freudig, S. Hogekamp, H. Schubert, Chem. Eng. Process. Process Inten. 38 (1999) 525–532

[17] T.C. Kha, M.H. Nguyen, P.D. Roach, J. Food Eng. 98 (2010) 385–392

[18] P.I. Silva, P.C. Stringheta, R.F. Teófilo, I.R.N. Oliveira, J. Food Eng. 117 (2013) 538–544

[19] V.R. Tonon, C. Brabet, D. Pallet, P. Brat, D.M. Hubinger, Int. J. Food Sci. Tech. 44 (2009) 1950-1958

[20] T.N.M. Nguyet, D.V. Hiep, J. Sci. Dev. 9 (2006) 69-75

[21] L.P. Fernandes, R.C. Candido, W.P. Oliveira, Food Bio-prod. Process. 90 (2012) 425–432

[22] G.R. Chegini, B. Ghobadian, Dry. Technol. 23 (2005) 657–668

[23] R.Y. Jumah, B. Tashtoush, R.R. Shaker, A.F. Zraiy, Dry. Technol. 18 (2000) 967–984

[24] S. Ersus, U. Yurdagel, J. Food Eng. 80 (2007) 805–812

[25] S. Phoungchandang, A. Sertwasana, Sci. Asia 36 (2010) 40–45

[26] A.S. Sousa, S.V. Borges, N.F. Magalhães, H.V. Ricardo, A.D. Azevedo, Braz. Arch. Biol. Technol. 51 (2008) 807– –814

Figure 3. Particle size distribution curve of GA (a) and GA-E (b).


100

[27] R.L. Carr, Chem. Eng. 1 (1965) 69–72

[28] R.L. Carr, Br. Chem. Eng. 15 (1970) 1541–1549

[29] E. Teunou, J.J. Fitzpatrick, J. Food Eng. 42 (1999) 109– –116

[30] J.J. Fitzpatrick, T. Iqbal, C. Delaney, T. Twomey, M.K. Keogh, J. Food Eng. 64 (2004) 435–444

[31] R.A. Buffo, K. Probst, G. Zehentbauer, Z. Luo, G.A. Rein-eccius, Flavour Fragrance J. 17 (2002) 292–299

[32] R.V.B. Fernandes, S.V. Borges, D.A. Botrel, Carbohydr. Polym. 101 (2014) 524-532

[33] W.D. Wu, W. Liu, T. Gengenbach, M.W. Woo, C. Selo-mulya, X.D. Chen, M. Weeks, J. Food Eng. 123 (2014) 130–135

[34] D. Saydi, M.S. Hatamipour, Iran. J. Chem. Eng. 9 (2012) 55–62

[35] S.A. Desobry, F.M. Netto, T.P. Labuza, J. Food Sci. 62 (1997) 1158–1162

[36] S.F. Pang, M.M. Yusoff, J. Gimbun, Food Hydrocolloids 37 (2014) 159–165

[37] S.M. Jafari, E. Assadpoor, Y. He, B. Bhandari, Dry. Technol. 26 (2008) 816–835.

LE PHAM TAN QUOC1,2

NGUYEN VAN MUOI1

1Departement of Food Technology, College of Agriculture and Applied

Biology, Can Tho University, Ninh Kieu District, Can Tho City, Vietnam

2Institute of Biotechnology and Food Technology, Industrial University of Ho Chi Minh City, Go Vap District, Ho Chi

Minh City, Vietnam

NAUČNI RAD

FIZIČKO–HEMIJSKE OSOBINE PRAHA KORENA Polygonum multiflorum THUNB. PROIZVEDENOG SA RAZLIČITIM NOSAČIMA

Polifenoli su vrlo važna jedinjenja za biljke. Nažalost, to su jedinjenja koja su prilično osetljiva na toplotu, svetlost i dejstvo kiseonika iz vazduha. Ovo je nedostatak o kome se mora voditi računa ako se ova jedinjenja skladište na duži vremenski period. Među-tim, ovaj problem se može prevazići inkapsulacijom sa nosačima, kao što su: malto-dekstrin, gumi arabika, modifikovani skrob, itd. U ovom radu analiziran je uticaj sprej su-šenja na sastav ekstrakta korena Poligonum multiflorum Thunb. posle inkapsuliranja maltodekstrinom (MD, DE16-19) i gumi arabikom (GA). Inkorporacijom gumi arabike u ekstrakt dobija se veći ukupan sadržaj polifenola (TPC) i antioksidativnog kapaciteta (AC) nego kada se inkorporira maltodekstrin. Kod prahova koji su dobijeni sa gumi ara-bikom i maltodekstrinom analizirani su sledeći parametri: prinos inkapsulacije, sadržaj vlage, parametre boje, ukupan sadržaj fenola, antioksidativni kapacitet, ukupna gustina, omekšivost, higroskopnost, indeks rastvorljivosti vode, veličina čestica i mikrostruktura. Rezultati pokazuju koji nosači imaju značajan uticaj na fizičkohemijske osobine prahova proizvedenih sprej sušenjem.

Ključne reči: nosač, gumi arabika, maltodekstrin, Poligonum multiflorum Thunb., sprej sušenje.




101

E. ABDUL JALEEL

K. APARNA

Dept. of Chemical Engineering, National Institute of Technology,

Calicut, Kerala, India

SCIENTIFIC PAPER

UDC 66.048.6:66

IDENTIFICATION OF A HEAT-INTEGRATED DISTILLATION COLUMN USING HYBRID SUPPORT VECTOR REGRESSION AND PARTICLE SWARM OPTIMIZATION

Article Highlights • Support vector regression (SVR) was used for identification of HIDC • Particle swarm optimization (PSO) performed the optimization of SVR • HYSYS software was used for generating data needed for identification • RMSE, R and MAE regression plots used for validating the performance • Hybrid PSO-SVR outperformed GA-SVR and BF-SVR Abstract

Distillation is the most commonly used method for separating fluid mixtures in oil and gas industries. It is a process that requires high energy usage. One of the efficient ways to save energy in a distillation column is by heat integration. One such type of distillation column is called a heat-integrated distillation column (HIDC). In HIDC, the prediction of mole fractions of the component in the product can be made using proper identification, or modeling, of the HIDC. However, nonlinear modeling of HIDC is a highly challenging task. Methods based on first principles are not sufficient for a highly nonlinear HIDC. Hence, a novel method for identification of HIDC using a non-parametric “support vector regression (SVR)” method for predicting benzene composition in benzene-toluene HIDC is proposed in this work. The data used for identification is generated using pro-cess simulation software HYSYS. 100 samples of data were used for training and 50 samples of data were employed for validating the model. Particle swarm optimization (PSO) was also incorporated with SVR for obtaining optimized parameters of SVR. The proposed model is compared with other SVR models optimized with optimization methods other than PSO. The proposed model showed better performance over others.

Keywords: SVR, HIDC, identification, PSO.

Distillation is a widely employed unit operation in petrochemical industries where 95% of liquid sep-aration is carried out using this process [1]. It causes significant energy consumption in refineries and pet-rochemical plants [2]. It accounts for 3% of world energy consumption [3]. Much research has been done to find methods for energy conservation since 1970 [4]. Process integration is an effective way to

Correspondence: E. Abdul Jaleel, Dept. of Chemical Engineering, National Institute of Technology, Calicut, Kerala 63601, India. E-mail: [email protected] Paper received: 18 November, 2016 Paper revised: 10 May, 2017 Paper accepted: 13 June, 2017

https://doi.org/10.2298/CICEQ161118023J

increase energy efficiency [5]. One of the methods developed is heat integration of two distillation columns [6]. Hence, the thermodynamic efficiency of the system can also be increased using this approach [7]. M. Nakaiwa et al. [4], Nakanishi et al. [8], Ponce [9] and Li et al. [10] proposed a heat integrated dis-tillation column (HIDC). Energy conservation was observed in all of the above proposed models when compared with conventional distillation columns.

The required purified products in HIDC, like in a conventional distillation column, can be obtained by controlling the product compositions [11,12]. The exact prediction of product compositions is necessary to preserve the products with essential purity and thereby, optimal control of compositions is also feasible [13]

E.A. JALEEL, K. APARNA: IDENTIFICATION OF A HEAT-INTEGRATED… Chem. Ind. Chem. Eng. Q. 24 (2) 101−115 (2018)

102

and realistic modeling of HIDC is needed. Two approaches are practiced for the modeling of these types of complexed systems. One method employed is to produce a model based on the first principle where model parameters are determined from the process descriptions [14]. An alternative approach to the first principle is developing the system model from the process input-output data. This type of modeling is called system identification or simply, identification.

Several researchers practiced using first prin-ciple-based modeling of the heat-integrated distil-lation column in research works [8,15-18]. Numerous assumptions are used for this kind of modeling, and it results in deviation from the actual process for the performance of the system [14]. Further, complex dif-ferential equations also appear when this kind of modeling is employed [14] and finding the solutions of these equations takes longer time [20]. These kinds of derived models based on the first principle are inadequate to represent nonlinear systems accurately [21,22]. The distillation column shows strong nonlin-ear dynamic characteristics during high purity oper-ations [21,23]. An HIDC process is a process with complex dynamics and is highly interactive [24]. Operating a HIDC is also more complicated than operating a conventional column [5]. Since it is a non-linear process, first principle-based modeling is inade-quate for system modeling.

For modeling of these type of complex and non-linear systems, different types of identification non-parametric methods can be used, such as artificial neural network, fuzzy logic, etc. The neural network has the robust capability of approximating any func-tion, the ability for parallel processing, the ability to learn from data sets and the capacity to learn the system from input-output data. Due to the advantages mentioned above, it has been used for identification of different systems, including distillation. Although identification of many systems is carried out using the artificial neural network, it has some shortcomings like empirical and structural risks [25]. Moreover, the limited number of training samples causes over-fitting and leads to poor generalization in the case of the artificial neural network [26]. Due to all of these short-comings of the neural network, it is difficult to design a good model of the neural network and its variants, especially to those who have little experience and little prior knowledge. These limitations, when the traditional neural network is used, can be solved by support vector regression [27,28]. Support vector reg-ression is one of the applications of a support vector machine. A support vector machine is a statistical learning algorithm. It can be used for classification [29], fault diagnosis [25,30] and regression. A support

vector machine for data regression (SVR) is used as a powerful tool for learning in many applications [31- –35]. The greatest advantage of SVR is structural risk minimization, rather than empirical risk used in the neural network. Structural risk minimization minimizes the upper bound of generalization errors, rather than of empirical errors used in the neural network [32]. Due to structural risk minimization, the optimal struc-ture is achieved by SVR.

To obtain better accuracy for support vector reg-ression, the model parameters of SVR have to be chosen selectively. By trial and error, it takes a long time to achieve the best parameters. Hence, the para-meters of a SVR model can be optimized using an evolutionary algorithm like genetic algorithm approach [36], bacterial foraging [37], particle swarm optimization [38], etc. Yian et al. [25], Chen et al. [39] and Ustun et al. [40] used genetic algorithm (GA) for finding optimal parameters of SVR. Wu et al. emp-loyed bacterial foraging (BF) for optimization of a sup-port vector machine (SVM) [41]. Yang et al. practiced BF for choosing optimal parameters of SVR [42]. Yian et al. [25], Kong et al. [31] and Lou et al. [43] developed models using SVR optimized with PSO. These optimized SVR models are called PSO-SVR models. Among these algorithms, particle swarm opti-mization is more attractive because of its simple imp-lementation, real convergence to the right solution with less time meeting the requirement of the object-ive function [44]. It has good stochastic global opti-mization as well. PSO also has excellent computati-onal efficiency, requiring less memory space and less CPU speed and a lesser number of parameters to tune [45]. PSO can be programmed easily using fund-amental mathematical and logical operations. Many nonlinear and optimization problems can be efficiently solved using PSO [46]. PSO can achieve the best solution, even though it does not need any gradient information about the objective function. PSO also has good convergence property to satisfactory sol-utions [46]. Hence, PSO is used for optimizing para-meters of SVR in this work.

Any of the non-parametric methods above have not been applied in existing literature for identification of a heat-integrated distillation column. The objective of this work is to develop a support vector regression model for heat-integrated distillation.

MATERIALS AND METHODS

Heat-integrated distillation column

A distillation column consists of a feed section, reboiler and condenser. The separating agent is heat and the reboiler provides this heat. Supplied heat is


103

lost at the condenser. Reboiler temperature is higher than the condenser temperature. Hence, heat added at a higher temperature is lost at the lower tempe-rature in the distillation column. So, thermal energy is lost in the condenser. Thermal energy is not reused in the conventional distillation column. In this case, energy is degraded from reboiler to condenser.

To overcome this energy degradation and to improve thermal efficiency, two methods are applied to the heat integrated distillation column: 1) Inter-coolers and inter-heat exchangers are used. 2) The distillation column is divided into two sections: a high pressure (HP) column and a low pressure (LP) column. To establish heat transfer between two parts,

operating pressure and temperature at the HP column should be higher than one in the LP column. There-fore, lower operating pressure and temperature is used in the LP section, whereas higher operating pressure and temperature is employed in the HP section. Reflux flow is carried out in the HP section, and vapor flow is held in the LP section. Therefore, the condenser in the HP section and the reboiler in the LP section can be avoided. The conceptual flow sheet of the heat-integrated distillation column used in this work is shown in Figure 1. The HYSYS flow sheet of the heat-integrated distillation column is illustrated in Figure 2.

Figure 1. Conceptual diagram of a heat integrated distillation column (HIDC).

Figure 2. HYSYS simulation diagram of a heat integrated distillation column (HIDC).


104

In this work, a Peng-Robinson fluid package was used for simulation. The stream feed is fed to HP and LP columns. The streams D and B are the dis-tillate stream from the LP column and the bottom pro-duct from the HP column, respectively. Feed condi-tions and compositions are given in Figure 1. The feed consists of benzene (light component) and tolu-ene (heavy component). This HIDC was used for the separation of benzene and toluene. Feed stream pressures are at a higher pressure than therequired feed stage pressure. This pressure is reduced to the required levels using control valves V1 and V2.

Feed locations were chosen to satisfy minimum reboiler duty for the HP column and minimum duty for the auxiliary heater for the LP column. Heat duty versus feed location for the HP and the LP column is shown in Figure 3.

The degree of freedom of one was observed for both HP and LP column. Hence, a 0.001 mol fraction of benzene at the reboiler stage was specified for the HP column, and a 0.99 mol fraction of benzene was chosen at the condenser stage for the LP column.

HP and LP column parameter specifications

Cooling water at 25 °C is inexpensive and it can be used in the condenser for cooling purposes. A temperature difference of 20 °C is essential for heat transfer in the condenser [47]. Therefore, the reflux drum temperature in the LP column can be selected as 45 °C, for heat transfer to occur. To accomplish 45 °C at the reflux drum, a pressure of 34.47 kPa (0.34 atm) has to be chosen in the condenser of the LP column, as shown in Figure 1. To establish a pres-sure difference between top and bottom sections of column, 44.47 kPa (10 kPa greater than in the con-denser) is used in the bottom of the LP column. This pressure results in a temperature of 82.95 °C at the bottom of the LP column. The pressure difference of

10 kPa between the top and the bottom is equally distributed in each tray of the LP column. Since 20 trays were used in this work, a pressure drop of nearly 0.5 kPa occurs in each tray.

To achieve heat transfer between the bottom section of the LP column and the top section of the HP column, a sufficient differential temperature between these two sections is essential [47]. In this work, heat has to be transferred from the top section of the HP column to the bottom section of the LP column. Hence, the temperature in the top section of the HP column should be higher in value compared to the temperature in the bottom section of the LP col-umn. To attain higher temperature in the top of the HP column, a higher pressure value of 344.47 kPa or 3.4 atm (300 kPa greater than the bottom pressure of the LP column) is chosen at the top of the HP column. The temperature at the top of the HP column then reaches 128.5 °C. This differential temperature of 45.55 °C (128.5-82.9 °C) between the top of the HP column and the bottom of the LP column is enough for heat transfer. To accomplish a pressure difference between the top and bottom sections of the column, 3544.47 kPa (10 kPa greater than at the top of the column) is applied in the reboiler of the HP column. The liquid composition and temperature profile for LP and HP columns in a steady state are shown in Fig-ures 4 and 5.

Heat integration

Heat integration occurs between the top section of the HP column (344.47 kPa, 128.5 °C) and the bottom section of the LP column (44.47 kPa, 82.95 °C), as shown in Figure 1. Heat integration between the two columns is carried out through the heat exchanger. The top section of the HP column works as a condenser and the bottom section of the LP column works as a reboiler, as described in literature

Figure 3. Heat duty versus feed location: a) HP column; b) LP column.


105

Figure 4. Composition and temperature profile of LP column: a) composition; b) temperature.

Figure 5. Composition and temperature profile of HP column: a) composition; b) temperature.

[16,18,19,47]. The operations in these two sections are as follows.

The process fluid at a temperature of 128.5 °C coming from the HP column enters the inlet part of the tube side of the heat exchanger. This process fluid leaves from the outlet part of the tube side of the

exchanger with a temperature of 124.6 °C. The tem-perature of this stream is again reduced using the auxiliary cooler. This cooler is used with lower duty compared to the condenser. The temperature of the process fluid is coming from the outlet of the cooler at 122.6 °C. This process fluid is refluxed to the top of


106

the HP column through a drum. Hence, a decrease in temperature in the refluxed stream is obtained through heat integration, i.e., the condenser action is carried out through heat integration.

The process fluid at a temperature of 82.95 °C coming from the bottom of the LP column enters the inlet part of the shell side of the heat exchanger. This process fluid leaves from the outlet part of the shell side of the exchanger with a temperature of 128.3 °C. The temperature of this stream is again increased using the auxiliary heater. This heater is also used with lower duty compared to reboiler duty. The tempe-rature of process fluid coming from the outlet of this heater is 132.2 °C. This process fluid is fed to the bottom of the LP column to provide heat. As a separ-ate auxiliary cooler and heater are used, cooling required at the top of the HP column and heat needed at the bottom of the LP column can be individually controlled. Therefore, heat is supplied to the lower part of the LP column through heat integration, i.e. reboiler action is carried out at the bottom of the LP column through heat integration.

This type of heat integrated distillation column rivals the conventional column in terms of energy saving and total annual cost saving. Economic ana-lysis comparison of a conventional distillation column and a heat integrated distillation column is shown in Table 1. The total energy cost and TAC is lesser in HIDC compared to conventional distillation columns.

Support vector regression (SVR)

For a regression-based support vector, the goal is to find a function f that maps actual input to actual output in such a way that the predicted output has, at most, a deviation of ε from the actual output y with maximum flatness as possible. Errors which are less than ε are not taken into considerations, but at the same time, errors greater than є are not accepted. Let us suppose the observation samples consist of samples of ( 1_LPr , 1_D LPx ), ( 2 _LPr , 2 _D LPx ), ( 3 _LPr , 3 _D LPx ),…,( _n LPr , _Dn LPx ) and _i LPr ϵ Rn and

_Di LPx ϵ R. _i LPr represent the reflux rate samples used in the LP column and 1_D LPx represent the mole fraction of the benzene component in the dis-tillate in the LP column, corresponding to reflux rate samples _i LPr :

= +LPf kr d (1)

For nonlinear support vector regression, the input data points are transformed to higher dimen-sional feature space, i.e., Φ: χ → ψ, so that ( 1_LPr , 2 _LPr ,…, _n LPr ) in the input space is trans-formed to high dimensional feature space as (ψ( 1_LPr ), ψ( 2 _LPr ),…,ψ( _n LPr )). The function which relates inputs (ψ( 1_LPr ), ψ( 2 _LPr ),…,ψ( _n LPr )) can be expressed by SVR, as given in Eq. (2):

( )ψ= +LPf k r d (2)

Table 1. Economic results of HIDC

Parameter Conventional distillation

column HP section of HIDC LP section of HIDC

HIDC (combined HP and LP)

No of trays 20 20 20 40 Feed Tray 11 11 10 - Feed rate, kmol/h 90.72 45.36 45.36 90.72 Temperature of condenser 45 - 45 45 Temperature of reboiler 82 165 - 165 Condenser duty, kW 1573.75 - 506.187 506.187 Reboiler duty, kW 1684.22 - 956.24 956.24 Area of condenser, m2 92.36 - 29.71 29.71 Area of reboiler, m2 68.96 19.81 - 19.81 Diameter of column, m 1.5 0.749 0.789 - Length of column, m 13.176 13.176 13.176 - Column cost, 104$ 21.4952 10.2479 10.8338 21.0817 Condenser cost, 104$ 13.8241 - 6.6135 6.6135 Reboiler Cost, 104$ 11.4330 5.0816 - 5.0816 Total capital cost, 104$ 46.7523 15.3295 17.4473 32.7768 Cooling water cost, 104$ 1.0528 - 0.3386 0.3386 LP stream vapor cost, 104$ 10.0985 - - - HP stream vapor cost, 104$ - 7.5441 - 7.5441 Total energy cost, 104$/y 11.1513 7.5441 0.3386 7.8827 Total annual cost (TAC), 104$/y 26.7354 12.6539 6.1544 18.8083


107

For achieving maximum flatness, optimization can be expressed by Eq. (3):

minimize 212

k

subject to ( )

( )ξ

ξ

− Ψ − ≤

Ψ + − ≤

_ _

_ _

Di LP i LP

i LP Di LP

x k r d

k r d x, i = 1,2,…, n (3)

If some amount of training errors outside the ξ sensitive zone are allowed, slack variables δi and ∗δi are introduced, as described by eEq. (4):

minimize ( )∗

=+ δ + δ2

1

12

n

i ii

k C

subject to

( )( )

ξ

ξ−

∗−

∗

− Ψ − ≤ + δ Ψ + − ≤ + δ

δ ,δ ≥

_

_

0

Di LP i LP i

i LP Di LP i

i i

x k r d

k r d x (4)

i = 1,2,…,n

> 0C indicates the trade-off between the flat-ness of f and the amount up to which deviations larger than ξ can be tolerated. Optimization problem sol-ution by Lagrangian method is given by Eq. (5):

( )

( ) ( )

( )( )

( )( )

μ μ

μ μ

ξ ψ

ξ ψ

∗ ∗ ∗

∗ ∗ ∗

= =

=

∗ ∗

=

δ,δ ,β,β = +

+ β δ + δ − δ + δ −

− β + δ − + +

− β + δ − − −

2

1 1

_ _1

_ _1

1, , , ,

2n n

i i i i i i ii i

n

i i Di LP i LPin

i i Di LP i LPi

La k C k

C

x k r b

x k r d

(5)

La is Lagrangian and μ μ∗ ∗β ,β ,, ,i i i i are Lagrange multi-pliers and μ μ∗ ∗β ,β ,, ,i i i i > 0 .

By minimizing with respect to primal variables ( )∗δ , δ, , i ik d , Eqs. (6)–(9) are obtained:

( ) ( )

( ) ( )

ψ

ψ

∗

=

∗

=

∂ = − β − β = ∂

= β − β

_1

_1

0 0n

i i i LPi

n

i i i LPi

La k rk

k r

(6)

( )∗

=

∂ = β − β =∂

1

0 0n

i ii

Lad

(7)

∂ = − β − δ =∂δ

0 0i ii

La C (8)

∗ ∗∗

∂ = − β − δ =∂δ

0 0i ii

La C (9)

By substituting, optimization can be written as:

maximize

( )

( ) ( ) ( ) ( )( )

ξ

ψ ψ

∗

∗ ∗

= =

∗ ∗

=

β,β =

− (β + β + (β − β = − β − β β − β

_1 1

_ _, 1

) )

12

n n

i i Di LP i ii i

n

i i j j i LP j LPi j

K

x

r r

subject to ( )∗

=

β − β =

β ,β

1

0

Є 0,

n

i ii

i j C

, i = 1,2, …,n (10)

Regression function can be written as shown below:

( ) ( ) ( ) ( )( )ψ ψ∗

=

= + _1

β -β ,n

LP i i i LP LPi

f r r r d (11)

Predicted output (mole fraction of benzene com-position in the distillate in the LP column) concerning support vectors is given by Eq. (12):

( ) ( ) ( )( )ψ ψ∗

=

= β − β +_ _ˆ ,D LP i i i LP LPi SV

x r r d (12)

In Eq. (12), ( ( ) ( )ψ ψ_ ,i LP LPr r ) is represented by the kernel function ( )_ ,i LP LPK r r . Predicted output using support vectors and the kernel function can be represented as per Eq. (13):

( ) ( )∗

=

= β − β +_ _ˆ ,D LP i i i LP LPi SV

x K r r d (13)

( ) ( )

( ) ( )

ξ

ξ

∗

<β <

∗

<β <

∗

= −

− β − β − +

+ −

− β − β −

_

_

_0

_ _r Є

_0

_ _r Є

1{ [

, ]

[

, ]}

i

j LP

j

j LP

Di LPC

j j j LP i LPSV

Di LPC

j j j LP i LPSV

d xs

K r r

x

K r r

(14)

where s is the number of support vectors in Eq. (14). The general structure of SVR for input samples of x(1), x(2),…,x(n) and support vectors x1,x2,…,xs is shown in Figure 6.

Denoting ∗β = β ,β[ ]T T Tn n , where βT

n = = β ,β , ...,β1 2[ ]n and ∗β T

n = ∗ ∗ ∗β ,β ,...,β1 2[ ]n and kernel matrix ×n nK where ( )=, _ _,i j i LP j LPK K r r , then the optimization problem (10) can be described by con-vex quadratic programming, as per Eq. (15):


108

minimize ξξ

− − β β + β +−

_

_

12

Tn D LPT

n D LP

XK KXK K

(15)

Different type of kernel functions K such as poly-nomial, sigmoid kernel, and radial kernel can be used. In this work, the radial kernel is used, which is given by Eq. (16):

( ) ( )= − − σ2 2

_ _, exp /LP D LP LP D LPK R X R X (16)

Particle swarm optimization (PSO)

Particle swarm optimization is originally based on the social behavior of bird flocks and fish schools, to search for candidate positions. When the birds search for food, the current position of the bird which is nearest to the food is searched. Every individual bird or particle is associated with a particular position and velocity. In the searching space, information is shared between individuals and thereby it helps to get the optimal solution of the particle. Each particle updates their moving trajectory, i.e., current position and velocity based on its experience or the compa-nion’s experience. As this process repeats, ultimately the particles achieve the optimal positions.

The initial position and velocity of the particle are randomly initialized in particle swarm optimiz-ation. The performance of the particle can be eva-luated by finding the fitness value of the particle. It helps to get the information of the current position and velocity whether they are good or bad. The fitness value of each particle is calculated at every iteration. The best particle is the particle with minimum fitness value if the problem is a minimization problem,

whereas if the problem is a type of a maximization problem, the best particle is the particle with maxi-mum fitness value. In this work, the problem is of the minimization category. Hence, the particle with mini-mum fitness value is chosen as the best particle. There-fore, the optimal particle solution is the position of each particle with minimum fitness value from the first iteration to the current iteration, and the global optimal solution is the best global position of all the particles from the first iteration to the current iteration [44].

Suppose there are P particles in the swarm. Let =t

ppbest 1 2, ,..., ,...,t t t tp p pd pDpbest pbest pbest pbest be

the best previous position of a particle p at iteration t and =tgbest 1 2, ,..., ,...,t t t t

d Dgbest gbest gbest gbest be the best global position of all the particles at iteration t. Let ( )t

p pf X and ( )tgf G be the fitness function

value of the thp particle and global optimal fitness value, respectively at the tht iteration. For the thp particle, if the value of fitness function at the current iteration is smaller than the value of fitness function at previous −1t iteration, then the best position of the particle pbest will be substituted by particle location at the current iteration step. Otherwise, pbest value remains constant. The same procedure is used for calculating gbest global optimum position of the all the particles, as illustrated from the following Eqs. (17) and (18):

−

=

1

tp

tp

Xpbest

pbest ( ) ( )−< 1

otherwise

t tp p p pf X f pbest

(17)

−

=

1

t

t

Ggbest

gbest ( ) ( )−< 1

otherwise

t tp pf G f gbest

(18)

Figure 6. Output predicting using SVR.


109

At iteration −1t , the position and velocity of the particle are updated as per Eqs. (19) and (20):

( ) ( )+ = + − + −11 1 2 2

k t t tp p p pV kV f r pbest X f r gbest X (19)

+ += +1 1t t tp p pX X V

Where tpV , +1t

pV , tpX , +1t

pX are velocity of the particle, updated velocity of the particle, position of the particle, updated position of the particle and iner-tia weight respectively. 1f and 2f are acceleration constants and 1f is called social constant and 2f is called cognitive constant. 1r and 2r are random values between 0 and 1. k is called inertia weight and its value decreases from a maximum value to a minimum value for improving the performance as the iteration is progressing. pX and pV are being kept within range [ minX maxX ] and [ minV maxV ], respect-ively, for each particle. The flow diagram of PSO is given in Figure 7.

Proposed model

Data needed for identification of a heat-inte-grated distillation column is created from HYSYS soft-ware. This software is widely used in chemical and oil and gas refineries. Here, the output considered is the benzene composition. The manipulating variable is used to change the compositions. The reflux rate is used as the manipulating variable. Hence, this mani-pulating variable is used as the input variable for iden-tification. A selective excitation input signal has to be used for identification [48]. As the system is non-lin-ear, random excitation signals are adopted [49]. 150 samples of data are collected from the HYSYS. Out of 150 samples collected, 100 samples are employed for training, and 50 samples are selected for validation of the proposed model.

The model used for predicting output variable mole fractions of benzene compositions is the SVR model. SVR is trained using the training input data to develop an SVR model. This SVR model trained with

Figure 7. Flow diagram of PSO.


110

data creates a function which relates the input vari-able reflux rate with the output variable mole fractions of the benzene composition. This developed model is also tested with validation input data to ensure the accuracy of the model. Usually, the accuracy of SVR mainly depends on the parameters of SVR, C, σ and ξ . Hence, these parameters are also optimized using a simple global optimization PSO algorithm in this work. Parameters of PSO utilized in this work are listed in Table 2.

Table 2. Parameters of PSO

Name Value

Number of particles 20

Maximum iteration 20

Maximum inertia weight 1.9

Minimum inertia weight 1.4

Social constant 2

Cognitive constant 2

After dividing the data, the training data is applied to the PSO algorithm to choose the best para-meters of SVR. The position variables of PSO are the solutions of parameters. During the training of PSO, each particle’s positions (solutions of the SVR) are applied to Eq. (15). Values of β and β ∗ are obtained by solving the Eq. (15). These values are used to cal-culate the predicted output, according to Eq. (12). Fit-ness values of the corresponding particle (solutions of parameters of SVR) are calculated based on the pre-dicted outputs values. Fitness value is calculated based on the root mean square error (RMSE).

From the calculations of fitness values of par-ticles, the pbest value of each particle and the gbest values of particles are calculated. Calculation of pbest and gbest are repeated until a maximum

number of iterations is achieved. Finally, the optimal position values of PSO (optimal solutions of paramet-ers of SVR) are obtained.


The input-output data used for identification is illustrated in Figure 8. 150 samples of input (reflux rate) and 150 samples of output variables (mole fractions of benzene composition) are illustrated in Figure 8. Figure 8a represents the input data samples and Figure 8b illustrates the output data samples.

In this study, five solutions of PSO parameters were used for identification of a heat-integrated distillation column. These optimal values of parameters of SVR were obtained using PSO. These five solutions of parameters are given in Tables 3 and 4. Usually, minimum and maximum values of position variables (solutions of SVR parameters) are initialized during PSO training. In this case, zero was used as the minimum value of all variables. But different maxi-mum values of the parameters σ and C were used during training of SVR using PSO. In these cases, dif-ferent values of optimal parameters were obtained. The number of support vectors was also different for various optimal values of the parameters, as shown in Table 2. The minimum value of support vectors (5)

Figure 8. Input-output data used for identification; a) input data; b) output data.

Table 3. Number of support vectors for various optimized para-meters of SVR

Parameters of SVR No. of support vectors C σ ξn

48.1634 0.7672 0.0155 48

100 1.1602 0.0138 55

188.3618 0.8135 0.0222 6

320.1290 0.6008 0.0201 12

877.8386 1.5671 0.0226 5


111

was observed in the parameter selection: 877.8386 (C), 1.5671 (σ ) and 0.0226 ( ξn ). The maximum number of support vectors (55) showed for the para-meter selection group: 100 (C), 1.1602( σ ) and 0.0138 ( ξn ).

Response of the proposed model

The response of the proposed model for training data and validation data is shown in Figures 9 and 10. In both cases, actual values and predicted output values are very close, as shown in Figures 9a and 10a. Errors between predicted output values and actual experimental values in validation data are illus-trated in Figure 10b. These errors were slight (between -0.0003 and 0.0003).

Figure 9 illustrates the predicting capability of SVR from the provided samples. Figure 10 shows the capacity of SVR to predict outputs close to the real value from unknown samples (samples which are not used for training).

Performance analysis

Performance analysis of the system is carried out through four degrees of measure: 1) regression plot; 2) root mean square error (RMSE); 3) correlation coefficient (R); 4) mean average error (MAE). The

equations for RMSE, MAE and R are given by Eqs. (21)-(23). In Eqs. (21)-(23), ( )y k , ( )y k , ( )y k and

( )y k are actual or experimental output value, pre-dicted output value, average value of actual or expe-rimental output values, and average value of pre-dicted output values, respectively:

( ) ( )( )=

= −2

1

1 ˆN

iRMSE y k y k

N (21)

( ) ( )=

= −1

1 ˆN

iMAE y k y k

N (22)

( ) ( )( ) ( ) ( )( )( ) ( )( ) ( ) ( )( )

=

= =

− −=

− −

1

22

1 1

ˆ ˆ

ˆ ˆ

N

iN N

i i

y k y k y k y kR

y k y k y k y k (23)

Regression plot shows a linear relation between the predicted output value and the actual value. The predicted output values are also displayed in the regression plot. Actual values are positioned along the x-axis and predicted output values along the y-axis. The equation relating predicted output value and the real value is given by Eq. (24):

Predicted output value = Target value + Constant (24)

Table 4. Table of comparison of statistical criteria for different values of optimized parameters

Parameters of SVR Training data Validation data

C σ ξn RMSE MAE R RMSE MAE R

48.1637 0.7672 0.0155 0.0013 0.0011 0.9998 0.0010 0.00082 0.9999

100 1.1602 0.0138 0.0013 0.0010 0.9999 0.00085 0.00065 0.9995

188.3618 0.8135 0.0222 0.0018 0.0015 0.9998 0.0016 0.0013 0.9998

320.1290 0.6008 0.0201 0.0017 0.0014 0.9981 0.0016 0.0013 0.9998

877.8386 1.5671 0.0226 0.0018 0.0015 0.9997 0.0016 0.0013 0.9998

Figure 9. Training data response: (a) Actual versus predicted; (b) Regression plot.


112

If the predicted output values are closer to the regression line, the model will be more accurate. Pre-dicted data points of PSO-SVR in Figure 9b and also in Figure 10c are lying very close to the regression line. It also shows the accuracy of the proposed SVR model.

RMSE is also used for the performance analysis of the model. RMSE value is between zero and one. RMSE value of zero indicates perfect matching of the predicted model and the actual model of the system. RMSE value of one represents greater mismatching between the predicted model and the real one.

RMSE for both training and validation data are small in all SVR models with the mentioned SVR parameter group in Table 4. The lowest value of RMSE (0.0013 for training and 0.00085 for validation) for both training (0.0013) and validation (0.00085) observed in the parameters group 100 (C), 1.1602 (σ ) and 0.0138( ξn ). MAE can also be used for per-formance analysis. MAE (0.0010 for training and 0.0065 for validation) has a lower value in the SVR model with the parameter group 100 (C), 1.1602 (σ ) and 0.0138 ( ξn ).

Another performance parameter is R-value. The R-value is between 0 and 1. R-value of one repre-sents the perfect matching of the predicted model and the actual model. R-value of zero represents no rel-ation between the predicted model and the actual

model of the system. Hence, as R-value gets closer to one, the predicted model will be closer to the real model of the system. R values are close to one (0.9998, 0.9999, 0.9998, 0.9981 and 0.9997 for train-ing data, 0.9999, 0.9995, 0.9998 and 0.9998 for valid-ation data) in all the SVR models with optimized para-meters, shown in Table 3.

Comparison of PSO-SVR with artificial neural network

To compare the performance of PSO-SVR with the artificial neural network (ANN), model responses, regression plots of both PSO-SVR and the neural net-work for training data are plotted in Figure 9. The Levenberg-Marquardt algorithm is used for the train-ing neural network. The model response of PSO-SVR model is very close to the real sample values com-pared with the neural network in Figure 9a. The data points of the SVR model are also nearer to the reg-ression line compared with ANN output values in Figure 9b. Validation data comparison of both ANN and PSO-SVR is shown in Figure 10. The model res-ponse and regression response of validation data of PSO-SVR is better compared with ANN, as shown in Figure 10b and c. Figure 10b represents the error between real values and model output values which is less in PSO-SVR compared to ANN.

RMSE and R values, for both training and valid-ation data in the case of PSO-SVR and ANN, are

Figure 10. Validation data response: a) actual versus predicted; b) error in the predicted outputs of validation data; c) regression plot.


113

shown in Table 5. RMSE values of the proposed model (0.0013 for training data and 0.0016) are com-paratively lesser than ANN (0.0157 for training data and 0.0137 for validation data). High value of R (0.99979 for training data and 0.99981 for validation data) is observed for PSO-SVR compared with ANN (0.98375 for training data and 0.98574 for validation data). The discussions above reveal that the perform-ance of PSO-SVR is better compared to ANN.

Table 5. Table of comparison of PSO-SVR and ANN

Algorithm Training data Validation data

RMSE R RMSE R

PSO-SVR 0.0018 0.99979 0.0016 0.99981

ANN 0.0157 0.98375 0.0137 0.98574

The ability to predict any model from a small number of samples with desired accuracy is an imp-ortant characteristic of SVR. From Tables 3 and 4, It is understood that SVR can predict the whole system using five support vectors or five numbers of samples (last row in Tables 3 and 4). But prediction of the entire system using such a small number of samples (5) is not possible with the neural network. To illus-trate the generalization property of SVR, RMSE of validation data for models trained with 50 and 100 numbers of sample data are shown in Table 6. RMSE in both cases (50 numbers of samples for training and 100 numbers of samples for training) are similar. But RMSE values differ more in ANN models.

Table 6. RMSE of validation data for model trained with differ-ent number of training data samples

Algorithm No. of the samples of training data

50 100

SVR 0.0023 0.0016

ANN 0.1111 0.0137

Comparison of PSO-SVR model with GA-SVR, BF SVR

To compare the accuracy of optimized para-meters of SVR, PSO-SVR is compared with a genetic algorithm-based SVR (GA-SVR) and a bacterial for-aging (BF)-based SVR (BF-SVR). In GA-SVR, the optimization of SVR parameters is carried out by the

genetic algorithm. Optimization of SVR parameters is employed using a bacterial foraging global optimiz-ation algorithm in BF-SVR. Statistical criteria RMSE and R are used for comparison. In this case, the mini-mum and maximum values of the variable used and the maximum number of iterations (20) are the same for the three optimization techniques. Optimized para-meter values and RMSE and R-values for both training and validation are shown in Table 7. RMSE values show little values for PSO-SVR for both train-ing (0.0018) and validation (0.0016) data compared with GA-PSO and BF-SVR. RMSE for GA is 0.0042 for training and 0.0043 for validation data. Similarly, RMSE values of BF-SVR are higher for training (0.0040) and validation (0.0040) compared with PSO- -SVR. The R-value has the higher value for both train-ing (0.99979) and validation (0.99981) data for PSO- -SVR compared to other models. From the statistical criteria above of SVR models, the better capability of the PSO-SVR model to predict the outputs from the input data is understood.

CONCLUSIONS

In this study, identification of a heat-integrated distillation column was carried out using optimized sup-port vector regression. Parameters of SVR were opti-mized using particle swarm optimization. The results showed high accuracy of the PSO-SVR model com-pared with GA-SVR and BF-SVR models. The pro-posed model was also compared with a neural net-work model and showed better performance over the artificial neural network (ANN). Five groups of opti-mized parameters were obtained, keeping the differ-ent maximum value of optimized parameters during the optimization stage. For various parameter groups, different number of support vectors was observed. The model accuracy was found to be nearly the same for these different numbers of support vectors. Statistical criteria was used for both training and validation data. The result showed the SVR capability to predict mole fractions of benzene compositions (output) from the manipulated variable reflux rate (input) with high accur-acy. This work can be extended to provide more accur-acy when using modified or improved PSO for opti-mization.

Table 7. Table of comparison of different algorithms for performance analysis

Algorithm Parameters of SVR Training data Validation data

C σ ξn RMSE R RMSE R

PSO-SVR 877.8386 1.5671 0.0226 0.0018 0.99979 0.0016 0.99981

BF-SVR 825.5206 1.1045 0.0452 0.0040 0.99935 0.0040 0.99925

GA-SVR 857.3357 1.1089 0.0456 0.0042 0.99919 0.0043 0.99894


114

REFERENCES

[1] E. Jaleel, K. Aparna, Procedia Technol. 25 (2016) 92-99

[2] R.S.H. Mah, J.J. Nicholas, R.B. Wodnik, AIChE J. 23 (1977) 651-658

[3] A.K. Jana, Appl. Energy. 87 (2010) 1477-1494

[4] M. Nakaiwa, M. Owa, T. Akiya, T. Nakane, M. Sato, Energy (1997) 621-625

[5] K. Naito, M. Nakaiwa, K. Huang, A. Endo, K. Aso, T. Nakanishi, T. Nakamura, H. Noda, T.Takamatsu, Comput. Chem. Eng. 24 (2000) 495-499

[6] M. Gadalla, Z. Olujic, L. Sun, A. De Rijke, P.J. Jansens, Chem. Eng. Res. Des. 83 (2005) 987-993

[7] M. Gadalla, Chem. Eng. Res. Des. 87 (2009) 1658-1666

[8] T. Nakanishi, T. Takamatsu, M. Nakaiwa, K. Aso, H. Noda, N. Kuratani, Comput. Chem. Eng. 23 (1999) S855- -S858

[9] G.H.S.F. Ponce, M. Alves, J.C.C. Miranda, R. Maciel Filho, M.R. Wolf Maciel, Chem. Eng. Res. Des. 95 (2015) 55-63

[10] R. Li, Q. Ye, X. Suo, X. Dai, H. Yu, Chem. Eng. Res. Des. 105 (2016) 1–15

[11] E. Abdul Jaleel, K. Aparna, ICACC, in Proceedings of Fifth International Conference on Advances in Computing and Communications, Kochi, India, 2015, pp. 358-361

[12] E. Abdul Jaleel, K. Aparna, INDICON, New Delhi, 2015, doi: 10.1109/INDICON.2015.7469599

[13] N. Mohamed Ramli, M.A. Hussain, B. Mohamed Jan, B. Abdullah, Neurocomputing 131 (2014) 59-76

[14] J.C. Macmurray, D.M. Himmelblau, Comput. Chem. Eng. 19 (1995) 1077-1088

[15] L. Chang, X. Liu, Sep. Purif. Technol. 134 (2014) 73-81

[16] J. Ma, M. Li, H. Chen, K. Huang, N. Wei, C. Xia, Chem. Eng. Res. Des. 91 (2013) 2299-2308

[17] H. Shahandeh, J. Ivakpour, N. Kasiri, Energy 64 (2014) 875-886

[18] X. Zhang, K. Huang, H. Chen, S. Wang, Comput. Chem. Eng. 35 (2011) 2017-2033

[19] Y. Wang, K. Huang, S. Wang, Ind. Eng. Chem. Res. 49 (2010) 3349-3364

[20] Y. Wang, K. Huang, S. Wang, Ind. Eng. Chem. Res. 49 (2010) 3349-3364

[21] A.A. Bachnas, R. Tóth, J.H.A. Ludlage, A. Mesbah, J. Process Control. 24 (2014) 272-285

[22] A. Nooraii, J.A. Romagnoli, J. Figueroa, J. Process Control. 9 (2006) 247-264

[23] S. Ramchandran, R.R. Rhinehart, J. Process Control. 5 (1995) 115-128

[24] M. Nakaiwa, K. Huang, M. Owa, T. Akiya, T. Nakane, T. Takamatsu, Comput. Chem. Eng. 1354 (1997) 389-393

[25] S. Yian, M. Fu, F. Wu, Neurocomputing 151 (2015) 296- –306

[26] N. Cristianini, J. Shawe-Taylor, An Introduction to Support Vector Machines and Other Kernel-based Learning Methods, Cambridge University Press, Cam-bridge, 2000

[27] V. Vapnik , Statistical Learning Theory, Willey, New York, 1998

[28] V. Vapnik, Statistical Learning Theory, Springer-Verlag, New York, 1999

[29] M. Zhong, X. Nie, A. Yan, Chem. Res. Toxicol. 26 (2013) 741-749

[30] S.Yin, X. Zhu, C. Jing, Neurocomputing 145 (2014) 263- –268

[31] X. Kong, X. Liu, R. Shi, Neurocomputing 169 (2015) 449- –456

[32] H. Zhao, S. Yin, Appl. Math. Model. 33 (2009) 3997-4012

[33] W. Yan, H. Shao, X. Wang, Comput. Chem. Eng. 28 (2004) 1489-1498

[34] S. Zaidi, Chem. Eng. Res. Des. 98 (2015) 44-58

[35] P. Jain, I. Rahman, B.D. Kulkarni, Chem. Eng. Res. Des. 85 (2007) 283-287

[36] D.E. Goldberg, Genetic Algorithm in Search, Optimization and Machine Learning, Adhison-wesely, New York, 1989

[37] K.M. Passino, IEEE Control Syst. Mag. 22 (2002) 52-67

[38] J. Kennedy, R.C. Eberhart, in Proceedings of Inter-national Conference on Neural Networks, 1995 IEEE, Vol. 4,1995, pp.1942-1948

[39] K. Chen, C. Wang, Tourism Manage. 28 (2007) 215-226

[40] B. Üstün, W.J. Melssen, M. Oudenhuijzen, L.M.C. Buydens, Anal. Chim. Acta 544 (2005) 292-305

[41] Q. Wu,X. Chen, L. Ding, C. Wei, H. Ren, R. Law, H. Dong, X. Li, IEEE Trans. Autom. Sci. Eng. (2016) 1-16

[42] D.L. Yang, X.J. Li, K. Wang, L.L. Jiang, Adv. Mater. Res. 216 (2011) 153-157

[43] I. Lou, Z. Xie, W. Ung, K. Mok, Appl. Math. Model. 39 (2015) 5907-5916

[44] J. Lu, H. Hu, Y. Bai, Neurocomputing 152 (2015) 305-315

[45] L. Lin, F. Guo, X. Xie, Neurocomputing 149 (2015) 1003- –1013

[46] D. Tungadio, B. Numbi, M. Siti, Neurocomputing 148 (2015) 175-180

[47] W.L. Luyben, Distillation Design and Control using Aspen Simulation, Wiley, New York, 2006

[48] L. Ljung, System Identification Theory for the User, Prentice Hall, Englewood Cliffs, NJ, 1987

[49] G. Ravi Sriniwas, Y. Arkun, I.-L. Chien, B.A. Ogunnaike, J. Process Control. 5 (1995) 149–162.


115

E. ABDUL JALEEL K. APARNA

Dept. of Chemical Engineering, National Institute of Technology,

Calicut, Kerala, India

NAUČNI RAD

IDENTIFIKACIJA TOPLOTNO INTEGRISANE DESTILACIONE KOLONE KORIŠĆENJEM HIBRIDNE VEKTORSKI PODRŽANE REGRESIJE I OPTIMIZACIJE ROJA ČESTICA

Destilacija je najčešće korišćena metoda za razdvajanje tečnosti u industriji nafte i gasa. To je proces koji koristi mnogo energije. Jedan od efikasnih načina za uštedu energije u destilacionoj koloni je toplotna integracija. Ovakva vrsta destilacione kolone se naziva toplotno integrisana destilaciona kolona (TIDK). U slušaju TIDK, predviđanje sastava proizvoda može se postići korišćenjem odgovarajuće identifikacije ili modela TIDK. Međutim, nelinearno modelovanje TIDK-a je izuzetno težak zadatak. Metode zasnovane na prvim principima nisu dovoljne za jako nelinearnu TIDK. Stoga se u ovom radu pred-laže novi metod za identifikaciju TIDK korišćenjem neparametričke metode “vektorski podržane regresije (VPR)” za predviđanje koncentracije benzena u koloni za razdva-janje benzena od toluena. Podaci koji su korišćeni za identifikaciju generisani su pomoću softvera za simulaciju procesa HYSYS. Za obuku je korišćen je set od 100 podataka, dok je 50 podataka korišćeno za validaciju modela. Optimizacija roja čestica (ORČ) je, takođe, kombinovana sa VPR radi dobijanja optimizovanih parametara VPR. Predloženi model su upoređeni sa VPR modelima optimizovanim drugim metodama optimizacije (različte od ORČ). Predloženi model pokazao je bolje performanse u odnosu na druge.

Ključne reči: vektorski podržana regresija, toplotno integrisana destilaciona kolona, Identifikacija, optimizacija roja čestica.




117

NEBOJŠA N. RISTIĆ

IVICA R. DODIĆ IVANKA P. RISTIĆ

Vocational High School for Technology and Art, Leskovac,

Serbia

SCIENTIFIC PAPER

UDC 677.494.675.027:544.4:66

THE INFLUENCE OF SURFACTANT STRUCTURE ON THE DYEING OF POLYAMIDE KNITTING WITH ACID DYES

Article Highlights • Surfactants as dyeing auxiliaries play an important role in achieving fast and leveled

dyeing • Surfactant mechanism of action as a leveling agent depends on its chemical com-

position • Nonionic surfactants reduce the chemical potential and substantivity of dyes in the

solution • Anionic surfactants act as retardation agents bonding to the sites of fiber onto which

dye ions bond • Surfactant concentration is optimized so that the dying process is efficient and eco-

nomical Abstract

The influence of nonionic and anionic surfactants on the dyeing kinetics of polyamide 6 knitting was studied in this work. The influence of surfactants on the dyeing process is presented by determining the kinetic and thermodynamic parameters of dyeing. Nonionic surfactants create unstable polydisperse asso-ciates which reduce the concentration of individual ionic forms of dyes in the solution, slowing down the dyeing process, with the dyeing having higher leveling. Interactions were confirmed by measuring the cloud point of nonionic surfactants and they are stronger with more hydrophobic dye and are related to the results of the studied dyeing kinetics. The anionic surfactant as a retarding agent, which behaves like a colorless dye in the studied dyeing system, makes a significant contribution even at concentrations of 0.5 and 1 g/dm3, indicating the conclusion that the surfactant concentration in a dyeing solution should be optimized by previous trials, so that the process would be efficient with high utilization of dye.

Keywords: polyamide, acid dye, surfactant, kinetics of dyeing, leveling effect, dye substantivity.

Polyamide (PA) fibers are thermoplastic fibers which by their chemical, physical and mechanical pro-perties represent a bridge between hydrophilic and hydrophobic fibers and therefore have a variety of applications. The dyeing of these fibers is possible with a wide range of ionizing and non-ionizing dyes. The most commonly used are acid dyes, often tail-ored to the specific dyeing characteristics of poly-

Correspondence: N.N. Ristić, Vocational High School for Technology and Art, Vilema Pušmana, 17, Leskovac, Serbia. E-mail: [email protected] Paper received: 2 November, 2016 Paper revised and accepted: 15 July, 2017

https://doi.org/10.2298/CICEQ161102025R

amide material. The dyeing rate of polyamide varies and depends on the number of amino groups and sub-molecular structure of the fiber and that is why the so-called kinetic classification of PA fibers is done. Uneven distribution of amino groups in the fiber, rapid adsorption of dye due to low glass tran-sition temperature and reduced fiber crystallinity under the dyeing conditions, as well as varying deg-rees of dye sulfonation in the dye solution, often lead to uneven dyeing of the polyamide, described as the stripe mark effect [1,2]. The causes of the stripe mark effect can be of chemical and physical nature. The chemical cause of stripe mark effect is a different number of end amino groups, due to different lengths

N.N. RISTIĆ et al.: THE INFLUENCE OF SURFACTANT STRUCTURE… Chem. Ind. Chem. Eng. Q. 24 (2) 117−125 (2018)

118

of polyamide polymer chains. If the chains are shorter there are more amino groups and vice versa. There are several physical causes and one of the most significant is a different crystallinity degree in fiber areas due to the non-uniform stretching of fibers.

Leveled and reproductive dyeing of polyamide products can be obtained by slowing down the dye adsorption on fiber, which can be achieved by adjust-ing the pH and temperature and by adding leveling agents. Also, the thermostabilization has an impact on the dyeing rate and color intensity of the fiber. Acid dyes for polyamide fibers are fixed by ionic bonds and van der Waals interactions where the pH of the sol-ution plays an important role, which is why the sel-ection of the acid donor for pH control appeared in a number of substantial papers [3-7]. Instead of the mostly used ammonium sulfate, the use of hydro-lyzing organic esters is suggested for higher util-ization of acid dyes.

Surfactants in a dyeing bath have various pur-poses: they facilitate the wetting of textile material, act as dispersants and leveling agents. Investigation of dye-surfactant interactions in an aqueous solution at surfactant concentrations below and above critical micelle concentration (CMC), indicated the formation of complexes with stoichiometric ratios of 1 to 1 below surfactant CMC [8,9] and above this, complexes with higher stoichiometric ratios are formed. The dye-surf-actant complexes may arise between dyes and surf-actants, having opposite charge, and between micel-les of nonionic surfactants and anionic dyes. The most important factors that influence complex form-ation are the type and the structure of the surfactant, especially the length of the hydrophobic chain and dye molecular mass. Depending on the molar ratio of dye/surfactant it can be water soluble or insoluble. The higher molar ratio of the dye/surfactant complex,

the more water solubility is observed but dye util-ization is lower, due to the increased affinity of the dye to the aqueous phase [10].

Thermostabilization treatment can reduce or inc-rease the adsorption of anionic dyes on polyamide by changing the supramolecular structure of the fiber and by chemical changes [11,12]. If the thermostabil-ization is done with hot air, the volume fraction of amorphous areas and end amino group concentration is reduced, causing reduced adsorption of anionic dyes on the polyamide. If the thermostabilization is done with water vapor or hot water, the number of amino groups is not reduced and molecular mobility, due to humidity, increases the accessibility of amor-phous areas, causing increased adsorption of anionic dyes on polyamide.

The objective of this work is the study of dyeing of polyamide knitting in the presence of surfactants of various chemical compositions, in order to determine the influence of surfactant chemical structure and concentration on the dyeing rate of polyamide.

EXPERIMENTAL

Materials and methods

In the experiment, the dyeing of the samples of polyamide knitting (PA6), having a surface mass of 172 g/m2, was performed. The samples had dimen-sions of 15 cm×15 cm. Before dyeing the knittings were washed in a solution of nonionic agent (0.5 g/dm3) at 45 °C for 30 min with 1:30 bath liquor ratio. After washing, the samples were rinsed several times with distilled water and dried at room temperature.

For the dyeing of polyamide the following acid dyes were used: C.I. Acid Red 114 (Mr = 830.81) and C.I. Acid Blue 324 (Mr = 473.43, Figure 1). The used dyes were of technical quality. The dyeing of samples

Figure 1. Structures of acid dyes.


119

was performed in an Ahiba apparatus with vertical movement of material at 70 °C, liquor ratio 150:1 and dye concentration of 1.5% on the mass of material. The acidity of the solution was adjusted by adding acetic acid until reaching pH 5, measured with a digi-tal pH meter Hanna - model HI 2209. The following surfactants were used in the experiments: FN-10, non-ionic surfactant (nonyl phenyl ethylene oxide) with 10 ethoxy groups (Henkel, Serbia) CMC = 0.75×10-4 mol/dm3, Slovaton CR, non-ionic oxy-ethylene product of high molecular weight fatty alco-hols (Chemapol, Czech Republic) and Alviron P96, anionic leveling agent for wool and polyamide dyeing (Textilcolor, Switzerland).

Surfactant concentrations in the dyeing bath were: 0.5, 1 and 3 g/dm3. Table 1 shows the design-ations of the samples.

Analysis

Reflection curves of the dyed samples were obtained with a reflection spectrophotometer Spectra-flash SF600X (Datacolor, USA). The color differ-ences, including the total color difference ΔE*ab according to CIELab system, were determined. Based on the reflection value (R) at the wavelength of max-imum color absorption, the color intensity for each sample (K/S) was determined using the Kubelka- -Munk equation:

2(1 )2

K RS R

−= (1)

where K is the absorption coefficient, S is the scat-tering coefficient and R is the reflectance factor (0 ≤ ≤ R ≤ 1).

The total color difference is calculated from Eq. (2):

2 2 2ab* ( *) ( *) ( *)E L a bΔ = Δ + Δ + Δ (2)

where ΔL*, Δa* and Δb* are differences between cor-responding parameters of samples (p) and standards (s). The standards are specimens dyed with the acid dye solution without the addition of surfactant and the

samples are specimens dyed with the acid dye sol-ution with the addition of surfactants.

The wavelength of maximum absorption of C.I. Acid Red 114 is 530 nm and of C.I. Acid Blue 324 is 620 nm.

Relative dye exhaustion (I) was determined by measuring the absorbance of the dyeing solution before dyeing (Ao) and after a certain period of time (At), using Eq. (3):

o

o

100 tA AIA−= (3)

Based on the relative exhaustion and maximum amount of dye on the fiber (15 mg/g), the quantity of dye after a certain period of time was determined.

The dyeing solution absorbance was measured on an MA 9507 colorimeter (Iskra, Slovenia) after 5, 10, 15, 20, 30, 45 and 60 min, using the green filter for red color and the yellow filter for blue color. The samples were dyed for 240 min to reach the equilib-rium state.

For each dyeing system dye substantivity (K) was determined, which is a measure of the ability of the dye to move from the solution to the fiber. Sub-stantivity was determined using Eq. (4) [13]:

%100 %

E LKE

×=−

(4)

where %E is dye exhaustion in % at equilibrium and L the dyeing bath liquor ratio.

The leveling of the dyeing was calculated by measuring K/S values on 20 random points at the maximum absorption wavelength, λ, and using Eqs. (5) and (6) [14,15]:

2,

1(( ) ( ) )

( )1

n

ii

K KS S

n

λ λ

σ λ =−

=−

(5)

,1

1( ) ( )

n

ii

K KS n Sλ λ

== (6)

where σ(λ) is the standard deviation of each random

Table 1. Designations of the samples

Dye Surfactant Concentration of surfactants, g/dm3

C.I. Acid Red 114 – 0 0.5 1 3

FN 10 C0 C1 C2 C3

Slovaton CR C4 C5 C6

Alviron P96 C7 C8 C9

C.I. Acid Blue 324 FN 10 P0 P1 P2 P3

Slovaton CR P4 P5 P6

Alviron P96 P7 P8 P9


120

point K/S value from (K/S)λ, λ is the wavelength of dye maximum absorption, n is the number of tested points, and (K/S)i,λ is the K/S value of each random point. The leveling of the dyeing is increased when the σ(λ) value is declining.

Color fastness to washing at 40 °C of dyed samples was determined, according to ISO 105 C10: :2006 standard.

RESULTS AND DISCUSION

Color differences

Tables 2 and 3 show color differences between the sample dyed without surfactant addition and the samples dyed with the addition of surfactants.

All test samples dyed with C.I. Acid Red 114 with the addition of FN-10 (Table 2) have higher lightness (+ΔL*) and lower chromaticity (-ΔC*ab) values. The color differences in all color coordinates increase proportionally with increased surfactant con-centration in the dyeing bath.

When using the nonionic agent Slovaton CR, a significant difference in color of the dyed test samples is obvious, compared to the standard sample. The observed color differences have a similar trend as with FN-10, i.e., the samples have higher lightness (+ΔL*) and lower chromaticity (-ΔC*ab) compared to

the standard, when the concentration of Slovaton CR in the dyeing bath is higher. The total color difference (ΔE*ab) is higher with FN-10, meaning that this non- -ionic surfactant has greater influence on the dyeing of PA knittings with C.I. Acid Red 114 compared to Slovaton CR. The greatest effect on the dyeing of polyamide with the acid dye C.I. Acid Red 114 has the anionic agent Alviron P69. The differences in light-ness (ΔL*), chromaticity (ΔC*ab) and the total color dif-ference (ΔE*ab) are several times higher than with non-ionic agents.

On the samples dyed with C.I. Acid Blue 324 (Table 3), significant differences were observed in the attributes of test samples compared to the standard. Color differences regarding lightness (ΔL*), chroma-ticity (ΔC*ab) and total color difference (ΔE*ab) have the same trend as with C.I. Acid Red 114, i.e. test samples are brighter (+ΔL*), greener (-Δa*) and bluer (-Δb*). The total color difference (ΔE*ab) increases with the concentration of the leveling agent in the dye-ing bath. The highest total difference was observed on the samples dyed with the addition of anionic agent Alviron P96, but the differences were signific-antly lower in comparison with C.I. Acid Red 114.

Kinetics of dyeing

Figure 2 shows the adsorption rates of C.I. Acid Red 114 on polyamide knitting in the first 60 min of

Table 2. Color differences of C.I. Acid Red 114 with illuminant D65/10, standard C0

Sample mark ΔE*ab ΔL* Δa* Δb* ΔC*ab ΔH*ab

C1 5.44 3.32 -1.16 -4.15 -2.72 -3.34

C2 10.14 6.71 -2.39 -7.23 -4.96 -5.78

C3 12.91 8.15 -4.64 -8.87 -7.62 -6.48

C4 4.92 3.39 -0.70 -3.50 -2.04 -2.92

C5 5.83 3.85 -1.19 -4.22 -2.77 -3.40

C6 8.30 5.50 -1.76 -5.96 -3.93 -4.81

C7 13.37 8.78 -4.32 -9.11 -7.40 -6.85

C8 24.25 15.85 -12.45 -13.49 -16.47 -8.10

C9 51.45 32.56 -37.37 -13.79 -39.50 5.21

Table 3. Color differences of C.I. Acid Blue 324 with illuminant D65/10, standard P0

Sample mark ΔE*ab ΔL* Δa* Δb* ΔC*ab ΔH*ab

P1 0.01 0.54 -0.43 -0.43 0.42 -0.44

P2 1.37 0.97 -0.64 -0.72 0.71 -0.65

P3 4.36 3.62 -2.43 -0.05 0.09 -2.43

P4 1.26 1.00 -0.69 -0.31 0.31 -0.69

P5 1.80 1.43 -0.76 -0.78 0.78 -0.76

P6 2.88 2.51 -1.25 -0.69 0.69 -1.25

P7 2.58 2.10 -1.11 -1.00 1.00 -1.11

P8 5.21 4.41 -2.35 -1.46 1.50 -2.33

P9 19.25 17.41 -7.74 2.80 -2.10 -7.96


121

dyeing. The highest dye concentration on the knitting was achieved after 60 min of dyeing at 70 °C on the sample dyed without the presence of the surfactant in the technological dyeing solution (9.9 mg/g). In the presence of the nonionic surfactant FN-10, dye con-centration was reduced proportionally to the surfact-ant quantity in the dyeing solution. After 60 min of dyeing the dye concentration on the knitting was sev-eral times lower compared with the standard sample.

Figure 2. Kinetics of dyeing of polyamide knittings with C.I. Acid

Red 114 in the presence of FN-10: (a), Slovaton CR (b) and Alviron P96 (c).

The non-ionic surfactant Slovaton CR shows a significant retarding effect on acid dye adsorption, already at minimum concentration. By further inc-rease of surfactant concentration, the concentration of dye present on the knitting changed minimally. It can be seen that FN-10 has a stronger effect on adsorp-tion of C.I. Acid Red 114 on the knitted samples than Slovaton CR. In systems with the anionic surfactant Alviron P96 the dyeing process was dramatically slowed down in such a way that at surfactant concen-tration of 3 g/dm3, only a minimum amount of dye was present on the knitting after 60 min of dyeing.

Also, the dyeing with C.I. Acid Blue 324 showed the highest adsorption degree with the dyeing sys-tems without surfactants (9.7 mg/g, Figure 3). The used surfactants affect the retardation of polyamide dyeing kinetics, but this effect is lesser than with C.I. Acid Red 114.

Figure 4 shows the color strength of C.I. Acid Red 114 on PA knitting after 60 min of dyeing. The highest strength has the standard sample due to the highest dye exhaustion. The samples dyed in the sol-ution with the addition of surfactants had lower color strength and it decreased progressively with the inc-rease of surfactant concentration. Extremely low values of color strength of C.I. Acid Red 114 were recorded with the addition of the anionic agent Alviron P96.

Also, C.I. Acid Blue 324 color intensities on the samples dyed in baths containing surfactants were regularly lower than on the standard. The reduction of intensity was higher at higher concentrations of the agent and the highest was with the anionic agent (Figure 5). By analyzing dyeing kinetics of used acid dyes, it can be seen that all surfactants used in the experiment have a stronger influence on the retarding effect in polyamide dyeing with the dye of greater molecular mass - C.I. Acid Red 114.

Reduced adsorption of acid dyes on polyamide knitting in the presence of nonionic surfactants can be explained by the changed state of acid dyes in the solution. Nonionic surfactants in an aqueous solution interact with anionic dyes, making associates based on hydrophilic and hydrophobic interactions, which was confirmed by different methods of UV-Vis spec-tral photometry, surface tension measurement and cloud point determination [16-19]. Interactions between non-ionic surfactants and ionic dyes occur at surfact-ant concentrations higher than CMC. Interactions between dye ions and single molecules of nonionic surfactants are negligible [20]. Non-ionic surfactants affect the textile dyeing mechanism by reducing the dye chemical potential in the solution. These agents effectively reduce the concentration of the mono-mole-


122

Figure 3. Kinetics of dyeing of polyamide knittings with C.I. Acid

Blue 324 in the presence of FN-10: (a), Slovaton CR (b) and Alviron P96 (c).

cular/soluble form of the dye in the solution, which can only penetrate the fiber structure. For the form-ation of the complex between the anionic dye and the micelle, nonionic surfactant hydrophobic interactions are responsible to a greater extent and interaction strength is increased with increased surfactant hydro-philicity and dye hydrophobicity [21]. In this work, the int-

Figure 4. Color intensity of Acid Red 114 on the samples of

polyamide knittings dyed in the presence of FN-10: (a), Slovaton CR (b) and Alviron P96 (c).

eractions of acid dyes with micelles of non-ionic surf-actant were studied by measuring the cloud point of FN-10 at surfactant concentration (1.5×10-3 mol/dm3) which is much above CMC (0.75×10-4 mol/dm3). The higher the cloud point, the stronger is the anionic dye nonionic/micelle of surfactant interaction [22]. The FN-10 cloud point regularly increases with increasing concentration of acid dyes (Figure 6). As it can be seen from the graph, the dye with the greater mole-cular mass, C.I. Acid Red 114, has more intense associations with the micellar nonionic surfactant, based on hydrophobic attraction. Anionic dye/micelle of non-ionic surfactant associates created in the dye-


123

Figure 5. Color intensity of Acid Blue 324 on the samples of

polyamide knittings dyed in the presence of FN-10: (a), Slovaton CR (b) and Alviron P96 (c).

ing process behave as dye depots, since the popul-ation of individual dye ions is adsorbed on the fiber first and afterwards the associates decompose, enab-ling further increase of dye utilization. This slower way of dye adsorption from the solution to the fiber allows for better dye distribution and produces dyeing with higher leveling, confirmed by leveling results shown in Table 4, where lower values of standard deviation σ(λ) were obtained for the samples dyed in the presence of surfactants than for the samples dyed without surfact-ants. The same table shows color fastness to wash-ing, which has a value of 4-5, and it can be consi-

dered that the dyeing in the presence of surfactants does not affect the color fastness of acid dyes on polyamide knitting.

0.0 0.1 0.2 0.3 0.4

50

60

70

80

90

Clo

ud p

oint

, o C

Dye concentration, g/dm3

C.I. Acid Blue 324 C.I. Acid Red 114

Figure 6. Cloud point of FN-10 depending on the concentration

of acid dyes.

Table 4. Leveling of dyeing of polyamide knittings, color fastness to washing and dye substantivity

Dye Sample σ(λ) Color fastness to

washing %E K

C.I. Acid Red 114

C0 0.111 4-5 79.2 571.1

C1 0.086 4-5 54.6 180.4

C2 0.084 4-5 37.2 88.8

C3 0.083 4-5 30.1 64.6

C4 0.088 4-5 54.0 176.1

C5 0.085 4-5 49.2 145.3

C6 0.082 4-5 44.4 119.8

C7 0.082 4-5 30.0 64.3

C8 0.077 4-5 14.4 25.2

C9 0.076 4-5 2.47 3.8

C.I. Acid Blue 324

P0 0.122 4-5 77.3 510.8

P1 0.115 4-5 68.4 324.7

P2 0.110 4-5 65.7 287.3

P3 0.108 4-5 37.5 90.0

P4 0.110 4-5 62.4 248.9

P5 0.108 4-5 59.3 218.5

P6 0.105 4-5 51.6 159.9

P7 0.098 4-5 59.5 220.4

P8 0.086 4-5 43.9 117.4

P9 0.080 4-5 27.1 55.8

Anionic surfactant Alviron P96 showed the greatest effect in reducing acid dye bonding rate to polyamide, which is consistent with literature results [23]. Anionic leveling agents behave like colorless dyes, i.e., they have substantivity to fiber and occupy cationic groups on the fiber in competition with dyes. As the anionic surfactant bonds faster to the substrate


124

site for which the dye has affinity, too, the differences in color intensities due to the stripe mark effect on polyamide are reduced, which is in accordance with the results for leveling (Table 4). Anionic surfactants have a leveling effect only when they bond to the sub-strate before the dyes. In that case, they slow down dye exhaustion, increase dye migration and reduce the final exhaustion. In the dyeing process with tem-perature increase, adsorption of ionic surfactants is reduced and ions of anionic surfactant are pushed out from the active sites of polyamide fibers and replaced with dye anions, which contributes to leveled and reproductive dyeing of polyamide knitting.

Table 4 shows the values of dye substantivity without and in the presence of surfactants. Substan-tivity indicates how much a dye would prefer to transit from the dyeing bath onto the substrate. The higher substantivity, the dye would faster transit to the fiber. Dyeing systems without surfactants have the highest substantivity, because the highest values of equilib-rium exhaustion are achieved. The addition of any surfactant reduces acid dye substantivity, because with nonionic surfactants, the number of dye ions act-ively participating in dyeing is reduced, due to dye- -surfactant association, i.e., the chemical potential of the dye in the solution is reduced, but anionic surf-actant systems block fiber active sites, slowing down dye adsorption and diffusion to the fiber.

CONCLUSIONS

The addition of nonionic surfactants with con-centration higher than CMC to the aqueous solution of acid dyes affects the state of acid dyes in such a manner that, due to hydrophobic attraction between dye ions and micelles of nonionic surfactant, acid dye-micelle of non-ionic surfactant hydrophilic com-plexes, with limited stability, are created. Solvated associates behave as dye depots which slowly rel-ease individual dye ions, so that adsorption on the surface of polyamide knitting takes place more slowly and more leveled dyeing is achieved proportionally to the surfactant concentration in the dyeing bath. According to experimental results, the dye of greater molecular mass, C.I. Acid Red 114, has more intense associations with micelles of non-ionic surfactants which reduces chemical potential and substantivity to a greater extent compared to C.I. Acid Blue 324. It can be concluded that the selected concentration region of nonionic surfactant can be successfully

applied in practical dyeing to achieve leveled dyeing of polyamide knitting.

The anionic surfactant acts as a retarding agent, which temporarily occupies, by ionic bond, protonated amino groups of polyamides to which ions of acid dyes also bond. During dyeing, the dye anions push out surfactant anions and slowly become adsorbed to the polyamide knitting, producing higher leveling.

REFERENCES

[1] D.M. Lewis, Rev. Prog. Coloration 28 (1998) 12-17

[2] M. Espinosa-Jimenez, R. Perea-Carpio, R. Padilla-Weig-and, A. Ontiveros, J. Colloid Interf. Sci. 238 (2001) 33-36

[3] K.A. Bello, O. Ajayi, J. Soc. Dyers Colour. 113 (1997) 131-135

[4] J. Koh, G. Sim, J. Kim, Color. Technol. 117 (2001) 156- –160

[5] J. Koh, G. Sim, J. Kim, Color. Technol. 117 (2001) 337- –341

[6] J. Koh, S.H. Park, G.H. Shim, D.H. Cho, J.P. Kim, Fib. Polym. 5 (2004) 110-116

[7] K.S. Lee, B.S. Lee, Y.H. Park, Y.C. Park, Y.M. Kim, S.H. Jeong, S.D. Kim, Fib. Polym. 6 (2005) 35-41

[8] S. Gokturk, M. Tuncay, J. Surfact. Deterg. 6 (2003) 325- –330

[9] M. Bielska, A. Sobczynska, K. Prochaska, Dyes Pigments 80 (2009) 201-2015

[10] A. Navarro, J. Cegarra, J. Valldeperas, D. Jocic, Dyes Pigments 69 (2006) 40-44

[11] W.B. Achwal, M.R. Nagar, Indian J. Text. Res. 2 (1977) 82-84

[12] R. Atav, A. Cay, A.E. Korlu, K. Duran, Color. Technol. 122 (2006) 277-281

[13] C.W. Kan, C.W. M. Yuen, Plasma Process Polym. 3 (2006) 627-635

[14] F. Zhang, Y. Chen, H. Lin, Y. Lu, Color. Technol. 123 (2007) 351-357

[15] N. Ristić, I. Ristić, J. Eng. Fiber. Fabr. 7 (2012) 113-121

[16] J. Oakes, P. Gratton, Color. Technol. 119 (2003) 91-99

[17] M. Sarkar, S. Poddar, Spectrochim. Acta, A. 55 (1999) 1737-1742

[18] N. Ristić, M. Šmelcerović, I. Ristić, Tekstil 62 (2013) 1-7

[19] K. Blus, J. Bemska, Autex Res. J. 10 (2010) 64-68

[20] D. Pramanick, D. Mukherjee, J. Colloid Interf. Sci. 157 (1993) 131-134

[21] M. Kert, B. Simončić, Tekstilec 50 (2007) 187-207

[22] D. Jocić, R. Trajković, Tekstil. Ind. 37 (1991) 30-36

[23] K. Zdovc, M. Gorjanc, M. Kert, Tekstilec 56 (2013) 312- –322.


125

NEBOJŠA N. RISTIĆ

IVICA R. DODIĆ IVANKA P. RISTIĆ

Visoka tehnološko umetnička strukovna škola, Leskovac, Srbija

NAUČNI RAD

UTICAJ STRUKTURE TENZIDA NA BOJENJE POLIAMIDNE PLETENINE KISELIM BOJAMA

U ovom radu je proučavan uticaj nejonogenih i anjonskog tenzida na kinetiku bojenja poliamid 6 pletenine. Uticaj tenzida na proces bojenja prikazan je određivanjem kine-tičkih i termodinamičkih parametara bojenja. Nejonogeni tenzidi stvaraju labilne poli-disperzne asocijate koji smanjuju koncentarciju monojonskog oblika boje u rastvoru, zbog čega je proces bojenja usporen a obojenje ima veću ravnomernost. Interakcije su potvrđene merenjem tačke zamućenja nejonogenog tenzida, jače su kod hidrofobnije boje i dovedene su u vezu sa rezultatima kinetike bojenja. Anjonski tenzid kao retarda-ciono sredstvo koje ispoljava ponašanje bezbojne boje u proučavanom sistemu bojenja ima značajan učinak već pri koncentracijama od 0,5 i 1 g/dm3, što upućuje na zaključak da predhodnim probama treba optimizovati koncentraciju tenzida u kupatilu za bojenje tako da proces bude efikasan a iskorišćenost boje velika.

Ključne reči: poliamid, kisele boje, tenzid, kinetika bojenja, egalizacioni efekat, supatantivnost boje.




127

BOJANA Ž. BAJIĆ

DAMJAN G. VUČUROVIĆ SINIŠA N. DODIĆ

ZORANA Z. RONČEVIĆ JOVANA A. GRAHOVAC

JELENA M. DODIĆ

University of Novi Sad, Faculty of Technology, Department of

Biotechnology and Pharmaceutical Engineering, Novi Sad, Serbia

SCIENTIFIC PAPER

UDC 628.3:547.815:60

THE BIOTECHNOLOGICAL PRODUCTION OF XANTHAN ON VEGETABLE OIL INDUSTRY WASTEWATERS. PART II: KINETIC MODELLING AND PROCESS SIMULATION

Article Highlights • Simulation software was used for the xanthan production process and cost model

development • Simulation model was developed based on defined kinetic models • Results represent a basis for defining a general design for the suggested bioprocess Abstract

Xanthan is a microbial biopolymer with a wide range of industrial applications and it is expected that the demand for this product will significantly increase in the coming decade and for this reason it is important to constantly work on improving all aspects of this biotechnological process. The aim of this research was to examine the kinetics of batch cultivation of Xanthomonas campestris ATCC 13951 using vegetable oil industry wastewaters as a basis for the culti-vation medium, in order to produce the biopolymer xanthan. Kinetic modelling is very important for process control, reducing process costs and increasing product quality. By performing xanthan production on a medium with optimized content, the experimental values of content of biomass, carbon source and the desired product were obtained and used to determine the kinetics of biosyn-thesis. In order to describe biomass multiplication, product formation and carbon source consumption, the logistics, the Luedeking-Piret and modified Luedeking--Piret equation, respectively, were successfully used. Additionally, using process simulation software (SuperPro Designer®), a process and cost model for a xan-than production facility was developed. The developed model represents the basis for a 21,294.29 and 23,107.97 kg/year xanthan production facility, which uses a vegetable oil industry wastewater-based medium and a semi-synthetic medium. The simulation model of the suggested xanthan production process, developed and based on defined kinetic models, represents an excellent basis for its further improvement and for increasing its efficiency.

Keywords: batch cultivation, kinetic modelling, vegetable oil industry wastewaters, xanthan.

Using modern process modelling and simulation techniques facilitates the development of new bioprocesses. Process modelling and simulation inc-

Correspondence: B.Ž. Bajić, Department of Biotechnology and Pharmaceutical Engineering, Faculty of Technology, University of Novi Sad, Bulevar cara Lazara 1, 21000Novi Sad, Serbia. E-mail: [email protected] Paper received: 10 March, 2017 Paper revised: 17 July, 2017 Paper accepted: 20 July, 2017

https://doi.org/10.2298/CICEQ170310004B

reases our insight and understanding of the process and helps us to identify key points where it can be improved, as well as any potential issues. Bioprocess modelling and simulation improve insight into the understanding of the bioprocess itself and help in identifying its key points. During the development of the model it is necessary that the suggested appli-cation is clear and that the model is developed based on it [1,2]. Introducing these models into various sim-ulation software packages simplifies the identification of potential improvements and/or problems which

B.Ž. BAJIĆ et al.: THE BIOTECHNOLOGICAL PRODUCTION OF XANTHAN… Chem. Ind. Chem. Eng. Q. 24 (2) 127−137 (2018)

128

may occur during the production process. Bioprocess simulation software is a series of computer algorithms capable of mathematical modelling of the perform-ance of individual unit operations, whose main adv-antage is the fact that all aspects of process manage-ment are incorporated into one format and can be simultaneously tested [1,3]. Although different soft-ware packages were used to simulate various biopro-cesses, the available scientific literature contains no references to using SuperPro Designer® to obtain a model of xanthan production using vegetable oil ind-ustry wastewater.

Xanthan is an industrial polysaccharide with the highest commercial production and is obtained by cul-tivation with Xanthomonas campestris, an obligatory aerobic microorganism, able to utilize both complex and completely defined media [4]. The substrate for the commercial production of xanthan is glucose, typically favoring batch over continuous production due to its efficiency [5,6]. As the carbon source cost is one of the major factors contributing to the cost of xanthan production, research on its production from low-cost substrates is very important. Previous res-earch showed that various low-cost substrates, most importantly agricultural wastes and food industry wastewaters, can be used as raw materials for xan-than production [7-13]. Increased industrial production of xanthan is becoming increasingly promising due to its extensive applications and wide use in the food, toiletry, oil recovery, cosmetic, water-based paint and other industries. The major application of xanthan is as a suspending and thickening agent in the food ind-ustry [4,14].

Several authors provided detailed explanations of the characteristics of kinetic models used to des-cribe xanthan production kinetics in their papers. The available literature shows that the testing of unstruc-tured kinetic models during batch xanthan production using the production microorganism Xanthomonas campestris was performed by many research groups [5,15-21]. The used models based on kinetic equa-tions which describe biomass growth and the pro-duction of the desired product can be classified into two groups: models where growth and production are dependent on medium nutrients and models where growth and production are only a function of temporal changes of biomass. The only difference between them is what type of kinetic equations are used for the description of the growth and production [16].

The aim of this research was to examine the kin-etics of xanthan production in laboratory scale batch cultivation by Xanthomonas campestris ATCC13951 using wastewaters of vegetable oil industry as a basis

for the cultivation medium. The additional aim of this research was to generate a simulation process model which primarily predicts process indicators, but also economic indicators of the analysed biotechnological process. Considering that previous research [22] proved that xanthan production from vegetable oil ind-ustry wastewater is viable and optimised the medium that is based on this wastewater, regarding the most significant nutrients (carbon, nitrogen and phosphorus content), current research represents a step towards scaling up this process to the industrial level using modelling and simulation of this biotechnological pro-cess.

EXPERIMENTAL

Production microorganism

The reference culture Xanthomonas campestris ATCC 13951, was used as the production micro-organism for the experiments. The inoculum was pre-pared in two steps: first, by refreshing the culture by incubation for 24 h, at 26 °C, on yeast maltose (YM®, Difco) agar slants, and second, by double passaging of the microorganism on the synthetic YM® (Difco) media for 36 h, at 26 °C. Samples were spontane-ously aerated and externally stirred (laboratory shaker, 150 rpm).

Cultivation media

Wastewater obtained from the vegetable oil ind-ustry was used as a basis for the cultivation medium. Initial total nitrogen content was 0.018 g/L, total phos-phorus was 0.0037 g/L, COD value was 7240 mg/L, BOD value was 3200 mg/L. The vegetable oil industry wastewater did not contain any digestible sugars.

The optimized cultivation medium composition (average value from the optimized range: carbon source content 15.5 g/L, nitrogen content 0.065 g/L and phosphorus content 0.0145 g/L) was used for this experiment, based on previous research results [22]. Utilized wastewaters were enriched by adding glu-cose so that the initial concentration of the carbon source was 15.5 g/L. Yeast extract and (NH4)2SO4 (in 2:1 ratio) were added as a nitrogen source so that the total nitrogen content is 0.065 g/L. K2HPO4 was added as a phosphorus source so that phosphorus content is 0.0145 g/L. Semi-synthetic glucose-based medium was prepared to contain the same amounts of the carbon source, nitrogen and phosporus. Addi-tionally, the pH value of the cultivation media was set to 7.0 and sterilized in an autoclave at 121 °C and overpressure of 1.1 bar during 20 min.


129

Cultivation

The biotechnological xanthan production was carried out in a 3 L laboratory bioreactor (Biostat® A plus, Sartorius AG, Germany) with 2 L of the culti-vation medium. The biosynthesis was carried out in batch mode under aerobic conditions at the defined temperature (internal temperature regulation) with stirring, using two parallel Rushton turbines for 96 h. The initial bioprocess temperature of 26 °C increased to 30 °C after 36 h of cultivation. Additionally, the air flow rate was increased from the initial value of 1 to 2 vvm and agitation speed was increased during bio-synthesis from its starting value of 150 to 300 rpm, after 36 h of cultivation, due to the increased viscosity of the cultivation broth. All experiments were per-formed in triplicate.

Analytical methods

The content of biomass in samples of cultivation media was determined by centrifuging the sample diluted with the same volume of saline solution at 10,000 rpm (Hettich Rotina 380 R, Germany) for 30 min. The obtained sediment is rinsed with saline sol-ution and dried to constant mass at 105 °C.

The samples of the cultivation broth were cen-trifuged at 10,000 rpm for 30 min (Hettich Rotina 380R, Germany) and the obtained supernatants were used to determine residual carbon content. The supernatants were filtered through a 0.45 μm nylon membrane (Agilent Technologies, Germany) and sub-sequently analyzed using HPLC (Thermo Scientific Dionex UltiMate 3000series). The HPLC instrument was equipped with a HPG-3200SD/RS pump, WPS- -3000(T) SL autosampler (10 μL injection loop), ZORBAX NH2 column (250 mm×4.6 mm, 5 μm) and a RefractoMax520 detector. A 75 vol.% acetonitrile sol-ution was used as the eluent at a flow rate of 1.2 mL/min and an elution time of 20 min at column tem-perature of 25 °C.

Prior to centrifugation (10,000 rpm, 30 min) the obtained cultivation broths were diluted with 4 volumes of distilled water in order to determine resi-dual nitrogen and phosphorus content. The residual nitrogen content was determined from the obtained supernatants using the Kjeldahl method [23], and a standard method was used to determine the residual phosphorus content [24].

Rheological properties of the cultivation broth samples were determined using a rotational visco-meter (REOTEST 2 VEB MLV Prüfgeräte-Verk, Men-dingen, SitzFreitel), with a double gap coaxial cylinder sensor system, spindle N. The rheological parameters were calculated using the Ostwald de Vaele equation.

Product separation

Xanthan was recovered by precipitation with 96 vol.% ethanol in the presence of KCl. Ethanol was gradually added to the supernatant at 15 °C until the alcohol content in the mixture was 60 vol.%, at cons-tant stirring. A saturated solution of KCl was added when half of the necessary ethanol amount was poured into the supernatant in a quantity to obtain a final content of 1 vol.%. After precipitation, the mixture of xanthan was kept on 4 °C, for 24 h and then cen-trifuged (4,000 rpm, 15 min). The precipitate was dried to a constant mass at 60 °C and this data was used to calculate the xanthan yield.

Kinetic models and calculation of kinetic parameters

Kinetic model for biomass formation A logistic equation was applied to describe mic-

roorganism growth. The logistic equation represents a model which does not include the effect of substrate consumption and can be applied to describe the exp-onential and stationary growth phases but not the decline phase [25]:

( )( ) ( )( )

0

0

exp( )

1 / 1 expm

m m

X tX t

X X tμ

μ=

− − (1)

where X is the biomass content (g/L), X0 initial bio-mass content (g/L), Xm maximum biomass content (g/L) and μm maximum specific growth rate (h-1). The logistic model checked the values of X0, Xm and µm.

Kinetic model for product biosynthesis Xanthan production kinetics was described

using the Luedeking-Piret equation which is partially connected to microorganism growth:

d dd dP X Xt t

α β= + (2)

where P is xanthan content (g/L) and the parameters α (gP/gX) and β (gP/gX·h) represent growth asso-ciated and non-growth associated constants. The aforementioned parameters are empirical constants which can vary with cultivation conditions such as temperature, pH value and agitation speed [9].

Integrating Eq. (2) under the starting parameters of t = 0, P = P0 and substituting Eq. (1) into Eq. (2) produces the following equation:

( )( )

( )( )0 0

0

( )

ln 1 1 expmm

m m

P t P X t X

XX tX

α

β μμ

= + − +

+ − −

(3)

which can be rearranged into:


130

( )( )

( )( )

00

0

( ) ln 1 1 expmm

m m

XXP t P tX

X t X

β μμ

α

− − − − =

= −

(4)

Since it can be considered that at the beginning of biosynthesis there is no produced xanthan in the cultivation medium (P0 = 0), the value of α can be cal-culated as the slope of the curve obtained from the correlation of the left side of Eq. (4) and (X(t)-X0). Since β is a parameter independent of growth and its value isn't connected to the exponential phase, it can be determined based on data from the stationary phase where dX/dt = 0 using the following equation:

( )stat.phased / d

m

P t

Xβ = (5)

Kinetic model for carbon source consumption The kinetics of substrate (carbon source) con-

sumption is described using the modified Luedeking- -Piret equation:

d dd dS X Xt t

γ δ− = + (6)

Where S is substrate content (g/L) and the para-meters γ (gS/gX) and δ (gS/gX·h) represent growth associated and non-growth associated constants.

Integrating Eq. (6) under the starting parameters of t = 0, S = S0 and substituting Eq. (1) into Eq. (6) produces the following equation:

( )( )

( )( )0 0

0

( )

ln 1 1 expmm

m m

S S t X t X

XX tX

γ

δ μμ

− = − +

+ − −

(7)

which can be rearranged into:

( )( )

( )( )

00

0

( ) ln 1 1 expmm

m m

XXS S t tX

X t X

δ μμ

γ

− − − − =

= −

(8)

The value of γ can be calculated as the slope of the curve obtained from the correlation of the left side of Eq. (8) and (X(t)-X0).

Since δ is a parameter independent of growth and its value isn't connected to the exponential phase, it can be determined based on data from the stationary phase where dX/dt = 0 using the following equation:

( )stat.phased / d

m

S t

Xδ = (9)

Using the commercially available SigmaPlot®11 software (Systat Software Inc, USA), the analyzed models were fitted into experimental data using non- -linear regression analysis which provided a statistical indicator (R2) for regression goodness of fit as well as all other significant parameters of the used models.

Simulation of xanthan biosynthesis

The process and cost models were developed using SuperPro Designer® software (Intelligen Inc., Scotch Plains, NJ). The defined process flowsheet of the proposed design of xanthan production is shown in Figure 1. Input data and data on operating con-ditions of the examined process are obtained from the experiments and literature, while equipment and pro-cess data are obtained directly from the SuperPro Designer® software.

As the cultivation medium in this process, veget-able oil industry wastewater is enriched with neces-sary nutrients in a blending storage tank (P-07, 6.2 m height and 2.1 m diameter), from where they are sent to sterilization (P-08) with an operating throughput of 56,553 L/h, at a temperature of 121 °C, for 20 min. In order to estimate the efficiency of the xanthan pro-duction process using wastewaters as the cultivation medium, the same process is evaluated using a semi-synthetic glucose-based medium. The cultivation media requirement for the facility is 20,000 L/batch.

The sterilized cultivation media is then transfer-red to a bioreactor (P-09) and inoculated with the appropriate amount of Xanthomonas campestris ino-culum. The inoculum is prepared in several steps, using one test tube (P-01), two shake flasks (P-02 and P-03) and two reactors (P-04 and P-05). The first step are test tubes containing yeast maltose agar (YMA), the second step are two shake flasks contain-ing 200 mL and 4 L of yeast maltose agar broth (YMB), and the third and final step are the 80 L (1.1 m height and 0.4 m diameter) and 1600 L (2.8 m height and 0.9 m diameter) reactors containing YMB and wastewater or semi-synthetic-based medium, res-pectively. Each step in preparing the inoculum lasted 24 h at a temperature of 26 °C.

The xanthan biosynthesis is performed in a 25,662 L (6.7 m height and 2.2 m diameter) vessel (P-09). The bioprocess duration is 96 h, the tempera-ture is 30 °C and the rate of aeration is 2 vvm. The obtained cultivation broth is pasterised (P-10) with an operating throughput of 20,530 L/h, at a temperature of 100 °C, for 60 min in order to inactivate the pro-duction microorganism after which it is centrifuged in the 8 units of disc-stack centrifuge (P-11) to remove the majority of the biomass (throughput 2571 L/h).


131

After centrifugation, the supernatant is mixed (P-12; throughput 363126 kg/h) with ethanol and a saturated solution of KCl in order to precipitate and recover biosynthesized xanthan. The obtained mix-ture is then centrifuged with 7 units of disc-stack cen-trifuge (P-13; throughput 2541 L/h). The solid phase obtained after centrifugation is then spray-dried (P-14) and the obtained product is xanthan.

The liquid fraction obtained after centrifugation is distilled with a column (P-16; height of 3.3 m and diameter of 0.8 m), which recovers 93.36% of ethanol which is then used for xanthan precipitation. The stil-lage obtained after distillation is mixed with the bio-mass stream obtained after the first centrifugation (P-11) and concentrated in an evaporator with a heat transfer area of 7.03 m2 and saturated steam as the transfer agent. After evaporation (P-18), the mixture is dried in a rotary dryer (P-19) with a drying capacity of 345.2 kg/h, drum area 42.1 m2 using air as the drying gas. The obtained product can be further processed and used as animal feed, manure, etc.

The economic analysis for the suggested pro-cess model was performed in SuperPro Designer® software and is shown as costs of working capital, operating costs (raw material and utility costs, costs

that are facility-dependent, labour costs) and unit pro-duction costs.

Working capital is a measure of both a com-pany’s efficiency and its short-term financial health. Working capital is calculated by subtracting current liabilities from current assets. It indicates whether a company has enough short-term assets to cover its short-term debt. Current assets include cash, market-able securities, inventory, accounts receivable and other short-term assets to be used within the year. Current liabilities include accounts payable and the current portion of long-term debt. These are debts that are due within the year. Operating costs are exp-enses associated with the maintenance and adminis-tration of a business on a day-to-day basis. The oper-ating cost is a component of operating income and is usually reflected on a company’s income statement. While operating costs generally do not include capital outlays, they can include many components of oper-ating a business, including: accounting and legal fees, bank charges, sales and marketing costs, travel exp-enses, entertainment costs, non-capitalized research and development expenses, office supply costs, rent, repair and maintenance costs, utility expenses ans salary and wage expenses. Production rate repre-

Figure 1. Process flow diagram of xanthan production using vegetable oil industry wastewater or semi-synthetic medium.


132

sents the number or quantity of goods that can be produced during a given period of time. It is calcul-ated by using the annual available amount of vegat-able oil industry wastewater, batch size, xanthan pro-duced per batch and number of batches per year. A unit cost is the total expenditure incurred by a com-pany to produce, store and sell one unit of a particular product or service. Unit costs include all fixed costs, or overhead costs, and all variable costs, or direct material costs and direct labor costs, involved in pro-duction. Total revenue is the amount of cash obtained from selling all of the goods produced on a yearly basis.


Kinetics of xanthan biosynthesis

In this paper, the kinetics of the batch xanthan production process using the production microorg-anism Xanthomonas campestris ATCC 13951 in a 3 L bioreactor was monitored. Additionally, these experi-ments examined the larger scale applicability of the developed mathematical models obtained in the pre-vious research phase [22] which focused on optimiz-ing carbon, nitrogen and phosphorus source content in a medium based on vegetable oil industry waste-water in order to produce the maximum amount of desired product with minimum leftover nutrients.

For the purpose of kinetic modelling and defin-ing the kinetics of biomass and product formation as well as carbon source consumption, it is significant to monitor the contents of biomass, xanthan and the car-bon source in defined time intervals of the bioprocess. In order to obtain these results without changing the

working volume of the bioreaction, the values of other significant indicators of xanthan biosynthesis, such as apparent viscosity of the cultivation medium and con-tents of the residual nitrogen and phosphorus, were measured only in the 96th hour of the process. The obtained results show that the content of xanthan, apparent viscosity of the cultivation medium, residual contents of the carbon source, nitrogen and phos-phorus in the 3 L bioreactor were 12.52 g/L, 41.90 mPa·s, 2.55, 0.015 and 0.008 g/L, respectively. If these values are compared with the ones predicted by the developed model [22], which are 11.79 g/L for the content of xanthan, 38.61 mPa·s for apparent viscosity of the cultivation medium, 3.953 g/L for resi-dual content of carbon source, 0.016 g/L for residual content of nitrogen and 0.008 g/L for residual content of phosphorus, it can be seen that the two sets of values match to a large degree. Higher values of xan-than content and apparent viscosity of the cultivation medium, as well as lower or matching values of resi-dual nutrients can be a consequence of more inten-sive conditions in the bioreactor specifically agitation and aeration, which improves oxygen mass transfer and thereby increases the amount of dissolved oxy-gen which positively affects the bioprocess [26,27].

Figure 2 shows the kinetics of biomass multi-plication (X), product formation (P) and kinetics of carbon source consumption (S) using vegetable oil indutry wastewater as a basis for the cultivation medium for xanthan biosynthesis (a) and semi-syn-thetic medium (b). This figure shows the obtained experimental data and curves obtained from fitting this data in the suggested kinetic models using the

Figure 2. Kinetics of biomass multiplication (X), product formation (P) and kinetics of carbon source consumption (S) using vegetable oil

indutry wastewater (a) and semi-synthetic medium (b) as a basis for the cultivation medium.


133

SigmaPlot®11 software package. Experimental data in Figure 2a show that biomass content slowly but constantly rises immediately after medium inocul-ation. An intensive increase in biomass content can be seen from the 8th to the 36th hour of biosynthesis and then increases less intensively until the 48th hour, when it becomes stationary until the end of the bio-process with a biomass value of 2.38 g/L. This value is slightly lower than the biomass content (2.59 g/L) obtained during xanthan production using a medium based on diluted kitchen wastewater [28]. Intensive xanthan production lasts approximately from the 8th to the 48th hour of biosynthesis after which xanthan con-tent less intensively but constantly increases until the end of the bioprocess, when it reaches 12.52 g/L. Carbon source content during the process constantly decreases until it reaches the value of 2.55 g/L.

By performing the process under optimal con-ditions, it is possible to define its kinetics, and deve-loping kinetic models is very significant for the dev-elopment of biotechnological processes considering they affect process control, cost reduction and inc-rease the quality of the final product [25]. Table 1 shows the experimental and predicted values of kin-etic parameters of the xanthan biosynthesis on vegetable oil industry wastewater based medium and semi-synthetic medium.

Experimental data for biomass content was fitted into a logistic equation and the obtained model curve has a rising trajectory and a trend similar to the curve of general microbiological growth [17]. Addi-tionally, as a result of fitting experimental data to the aforementioned equation, the values for initial bio-mass content (X0), maximum biomass content (Xm) and maximum specific growth rate (μm) were obtained as predicted by the model (Table 1) with a R2 value of 0.9990. Initial cell concentration value predicted by the model is 0.124 g/L and is approximately the same as the experimental value. Additionally, the values of

maximum biomass content and maximum specific growth rate predicted by the model are 2.365 g/L and 0.118 h-1, respectively, are very similar to the experi-mental values (2.38 g/L and 0.127 h-1, respectively). According to research by Lo et al. [29], the value of specific growth rate, using the same strain of product-ion microorganism, was 0.06-0.12 h-1 and depends on cultivation conditions. All of the obtained results indi-cate that the applied logistic equation represents a suitable kinetic model for the growth of X. campestris in the applied experimental conditions and using a cultivation medium based on vegetable oil industry wastewater.

In order to obtain kinetic parameters and des-cribe xanthan production the Luedeking-Piret equat-ion was used and the results of fitting obtained expe-rimental values of xanthan content to this model are shown in Figure 2a. The applied model shows that xanthan production rate depends on biomass content and biomass formation rate. Therefore, the values of initial biomass content, maximum biomass content and maximum specific growth rate predicted by fitting experimental data for biomass to the logistic equation were used as constants in the Luedeking-Piret model used to fit experimental data of xanthan content during biosynthesis in the used conditions. This large degree of matching between experimental and values predicted by the model, as well as the fact that the obtained value of R2 is 0.9997, indicates that the Luedeking-Piret model can be applied to describe xanthan production using a cultivation medium based on vegetable oil industry wastewater.

In order to describe the kinetics of carbon source consumption, a modified Luedeking-Piret equation was used and the results of fitting obtained experi-mental data to this model are shown in Figure 2a. Since changes in carbon source content are affected by biomass content and biomass formation rate, according to the applied model, the values of initial

Table 1. Kinetic parameters of the xanthan biosynthesis on vegetable oil industry wastewater-based medium and semi-synthetic medium

Parameter Vegetable oil industry wastewater as a basis for cultivation medium Semi-synthetic medium

Values predicted by the model Experimentally obtained values Values predicted by the model

Experimentally obtained values

X0 / g L-1 0.124 ± 0.007 0.120 ± 0.010 0.145 ± 0.012 0.130 ± 0.003

Xm / g L-1 2.365 ± 0.012 2.380 ± 0.060 2.517 ± 0.019 2.560 ± 0.090

μm / h-1 0.118 ± 0.025 0.127 0.119 ± 0.004 0.146

α / gP gX-1 4.551 ± 0.023 4.486 4.864 ± 0.054 4.668

β / gP gX-1·h-1 0.013 ± 0.001 0.013 0.013 ± 0.001 0.014

S0 / g L-1 15.870 ± 0.133 15.650 ± 0.350 15.535 ± 0.122 15.890 ± 0.380

γ / gS gX-1 5.021 ± 0.127 4.214 4.467 ± 0.107 3.680

δ / gS gX-1·h-1 0.015 ± 0.002 0.020 0.019 ± 0.002 0.023


134

biomass content, maximum biomass content and maximum specific growth rate, predicted by fitting experimental data for biomass to the logistic equation, were used as constants in the modified Luedeking- -Piret model used to fit experimental data of carbon source content during xanthan biosynthesis. The obtained results show that experimentally determined values of initial carbon source content (15.65 g/L) and values predicted by the model (15.87 g/L) match to a large degree. Additionally, values of the parameter that is growth asociated (γ) and non-growth asso-ciated (δ) calculated based on experimental results and values predicted by the model were similar, with the statistical indicator (R2) of 0.9970. The obtained results show that the modified Luedeking-Piret equat-ion represents a suitable model for the description of carbon source consumption during xanthan biosyn-thesis in the used experimental conditions, using a cultivation medium based on vegetable oil industry wastewater.

In order to generate a model of a xanthan pro-duction facility using a cultivation medium based on vegetable oil industry wastewater and to access the efficiency of this bioprocess, xanthan biosynthesis was also performed on a semi-synthetic glucose medium whose significant kinetic parameters were determined and kinetic models for biomass and pro-duct formations and carbon source consumption were defined (Table 1). The process on the semi-synthethic medium was used to compare all indicators of bio-technological process efficiency with the wastewater-based cultivation medium. The semi-synthetic medium with glucose was selected because this carbon source is commonly used for xanthan biotechno-logical production [5,15].

Figure 2b shows the kinetics of biomass multi-plication (X), product formation (P) and kinetics of carbon source consumption (S) using semi-synthetic medium for xanthan biosynthesis. Experimentaly obtained values of biomass, shown in Figure 2b, int-ensively increase after the 4th hour until the last 36 h of biosinthesis where it becomes stationary and rem-ains almost unchanged at a value of 2.56 g/L until the end of the bioprocess. The obtained results show that glucose content constantly decreases during biosyn-thesis from 15.89 to 1.55 g/L. Xanthan content slightly increases during the first 4 h, after which it intensively increases to 12.71 g/L in the 60th hour and reaches a final value of 13.62 g/L at the end of the process.

The aforementioned models were used to des-cribe the kinetics of biomass and product formation and carbon source consumption during xanthan bio-sythesis on semi-synthetic medium. The logistic

equation was applied to fit experimental data of bio-mass content and the obtained R2 value was 0.9979 which indicates that the used logistic equation is excellent in describing the behavior (growth) of the used production microorganism on a semi-synthetic medium with glucose. Additionally, a high degree of matching of experimental and values predicted by the model, as well as the fact that the value of R2 was 0.9986 shows that the Luedeking-Piret model can be used to describe xanthan production on a semi-syn-thetic medium with glucose. The obtained results and the R2 value of 0.9976 shows that the applied Luede-king-Piret equation represents a suitable model for describing the consumption of the carbon source during xanthan biosynthesis on a semi-synthetic medium with glucose.

Simulation of xanthan biosynthesis process

As a part of this research, SuperPro Designer® was used to generate a model of the xanthan pro-duction bioprocess using vegetable oil industry waste-water and compare it to the bioprocess on the semi-synthetic medium with glucose. The results of this comparative analysis can point out key segments of the observed production process which can be used in further research to improve these segments and thereby the productivity of the entire bioprocess.

Results obtained in the previous part of the paper which contains kinetic parameters obtained by defining kinetic models for biomass formation, xan-than biosynthesis and carbon source consumption during xanthan biosynthesis in the medium based on vegetable oil industry wastewater as well as the semi-synthetic medium with glucose, were applied in the developed bioprocess model shown in Figure 1. The reactions and conversions in the bioprocess model are defined by the kinetics described in the previous part of this paper (Table 1). All data for input streams and operational conditions was obtained based on experimental results, while data regarding equipment and the bioprocess itself were obtained directly from the used software package.

Table 2 shows the data obtained from the simul-ation model of the xanthan production bioprocess as a result of using defined kinetic models (Table 1). The obtained content of the media based on vegetable oil industry wastewater as well as glucose, before and after bioreaction, shows that the simulation can suffi-ciently describe experimental results.

As a result of simulating the obtained results in a developed model, Table 3 shows the economic ana-lysis of the model of the xanthan production biopro-cess in media based on vegetable oil industry waste-


135

water as well as glucose. The economic analysis of the developed model did not consider investment costs, which assumes that a facility for the production of xanthan using the aforementioned media already exists, which means that the working capital has the same value of 16,000 $, while the operating costs dif-fer slightly and are 179,000 and 183,000 $/year, res-pectively.

The production rate of the xanthan production process using vegetable oil industry wastewater is 21,294.29 kg/year, while using a semi-synthetic med-ium gives a value of 23,107.97 kg/year. Additionally, unit production cost is 8.57 and 7.72 $/kg for waste-water and glucose-based media, respectively, which indicates some significant aspects of the bioprocess need to be improved in order to increase its profit-ability and thereby make the use of vegetable oil ind-ustry wastewater as a raw material cost effective. However, it is important to emphasise that since recycling/reusing wastewaters is one of the goals of sustainable development [30], the obtained results are very significant from an environmental perspect-ive and the suggested model represents a useful tool in scaling up this process from the laboratory to ind-ustrial level.

Results of the analysis of the simulation model unambiguously show that in order for the process to be economically viable, the amount of the obtained product must be higher. With the goal of reducing pro-duction costs and increasing the competitiveness of xanthan on the market, it is necessary to improve the productivity of the process by developing a more effi-cient production process, optimizing cultivation media content or isolating new strains of X. campestris

which have better production capabilities in the given experimental conditions [31].

Table 3. Economic analysis of the proposed process model for xanthan production on vegetable oil industry wastewater and semi-synthetic glucose-based medium; MP – main product

Parameter Vegetable oil industry

wastewater based medium

Semi-synthetic medium

Working capital, $ 16,000 16,000

Operating cost, $/year 183,000 178,000

Production rate, kg MP/year

21,294.29 23,107.97

Unit production cost, $/kg MP

8.57 7.72

Total revenues, $/year 108,000 117,000

Since the improvement of the amount of the desired product can be realized by using a suitable cultivation media for its production, it is significant to choose and optimize cultivation media content reg-arding the most important micro and macronutrients for its production [13,32-35]. Additionally, it is import-ant to choose and optimize suitable process paramet-ers such as agitation, aeration rate, cultivation time and bioreactor geometry [36-38]. Increasing the pro-fitability of the xanthan production process can also be realized by improving the quality of the obtained biopolymer, increasing its molecular mass and the content of acetates and pyruvates in the molecule [4]. This would increase the value of xanthan on the mar-ket for its use in food, cosmetic and pharmaceutical products.

Table 2. Simulated medium content before and after bioreaction, obtained using vegetable oil industry wastewater and semi-synthetic glucose based medium; miscellaneous - water, product of metabolism and leftover nutrients

Stream content Vegetable oil industry wastewater as a basis for

cultivation medium Semi-synthetic medium

Before bioreaction X / g L-1 0.12 0.15

X / % 0.01 0.02

S / g L-1 15.69 15.30

S / % 1.68 1.64

Miscellaneous, % 98.31 98.35

After bioreaction X / g L-1 2.36 2.51

X / % 0.25 0.27

S / g L-1 2.66 0.65

S / % 0.28 0.07

P / g L-1 12.51 14.20

P / % 1.34 1.52

Miscellaneous, % 98.13 98.14


136

CONCLUSION

The results obtained by examining xanthan pro-duction on a vegetable oil industry wastewater-based medium show that there is great potential for its application as a raw material for the biotechnological production of xanthan. The developed process model for xanthan production presents the basis for a 21,294.29 and 23,107.97 kg/year xanthan production plant that uses vegetable oil industry wastewate-based medium and semi-synthetic medium. There-fore, it is necessary to further improve the production process in order to increase its efficiency. The results of this research, obtained from kinetic modelling as well as bioprocess simulation based on experimental data using the selected software, are a technologic-ally reliable source of data and as such represent a basis for defining a general design for the suggested biotechnological process.

REFERENCES

[1] E. Heinzle, A. Biwer, C. Cooney, Development of Sus-tainable Bioprocesses, John Wiley and Sons Ltd., New York, 2006

[2] J.A. Sokolowski, C.M. Banks, Principles of Modeling and Simulation A Multidisciplinary Approach, John Wiley and Sons Ltd., New York, 2009

[3] T. Shanklin, K. Roper, P.K. Yegneswaran, M.R. Marten, Biotechnol. Bioeng. 72 (2011) 483-489.

[4] F.Garcia-Ochoa, V.E. Santos, J.A. Casas, E. Gomez, Biotechnol. Adv. 18 (2000) 549-579

[5] S. Rosalam, R. England, Enzyme Microb. Technol. 39 (2006) 197–207

[6] S. Velu, V. Velayutham, Int. J. ChemTech Res. 6 (2014) 2475-83

[7] S. Moshaf, Z. Hamidi-Esfahani, M.H. Azizi, World Acad. Sci. Eng. Technol. 5 (2011) 471-474

[8] V. Gunasekar, K.R. Reshma, G. Treesa, D. Gowdhaman, V. Ponnusami, Carbohydr. Polym. 102 (2014) 669– 673

[9] S.L. Gilani, G.D. Najafpour, H.D. Heydarzadeh, H. Zare, Chem. Ind. Chem. Eng. Q. 17 (2011) 179−187

[10] M Marbrouk, A. El Ahwany, M. Beliah, S. Sabry, Life Sci. 10 (2013) 1660-1667

[11] C.R.R. Carignatto, K.S.M. Oliveira, V.M.G de Lima, P. de Oliva Neto, Indian J. Microbiol. 51 (2011) 283–288

[12] M.F. Silva, R.C.G. Fornari, M.A. Mazutti, D. Oliveira, F.F. Padilha, A.J. Cichoski, R.L. Cansian, M.D. Luccio, H. Treichel, J. Food Eng. 90 (2009) 119-112

[13] R.B. Salah, K. Chaari, S. Besbes, N. Ktari, C. Blecker, C. Deroanne, H. Attia, Food Chem. 121 (2010) 627–633

[14] T. Gumus, A.S.D. M. Mirik, M. Arici, Y. Aysan, Food Sci. Biotechnol. 19 (2010) 201-206

[15] Palaniraj, V. Jayaraman, J. Food Eng. 106 (2011) 1–12

[16] S. Faria, P.A. Vieira, M.M. Resende, E.J. El Ribeiro, V.L. Cardoso, Food Sci. Technol. 43 (2010) 498–506

[17] F. Garcıa-Ochoa, V.E. Santos, A. Alcon, Enzyme Microb. Technol. 17 (1995) 206–217

[18] R.A. Moraine, P. Rogovin, Biotechnol. Bioeng. 13 (1971) 381–391

[19] R.M. Weiss, D.F. Ollis, Biotechnol. Bioeng. 22 (1980) 859–873

[20] A. Pinches, L.J. Pallent, Biotechnol. Bioeng. 28 (1986) 1484–1496

[21] F. Letisse, N.D. Lindley, G. Roux, Biotechnol. Prog. 19 (2003) 822–827

[22] B. Bajić, D. Vučurović, S. Dodić, Z. Rončević, J. Gra-hovac, J. Dodić, Chem. Ind. Chem. Eng. Q. 23 (2017) 329−339

[23] K. Herlich, Official Methods of Analysis of the Association of Official Analytical Chemists, fifth ed., Association of Official Analytical Chemists, Arlington, 1990, p. 758

[24] M.E.J. Gales, E.C. Julian, R.C. Kroner, J. Am. Water Works Assoc. 58 (1966) 1363-1368

[25] J. Dodić, D. Vučurović, S. Dodić, J. Grahovac, S. Popov, N. Nedeljković, Appl. Energy 99 (2012) 192-197

[26] R. Cheng, L. Lin, Y. Zhang, J. Ind. Microbiol. Biotechnol. 39 (2012) 799-803

[27] C.D. Borges, A.S. Moreira, C.T. Vendruscolo, M.A.Z. Ayub, Rev. Argent. Microbiol. 40 (2008) 81-85

[28] P. Li, T. Li, Y. Zeng, X. Li, X. Jiang, Y. Wang, T. Xie, Y. Zhang, Carbohydr. Polym. 151 (2016) 684–691

[29] Y.M. Lo, S.T. Yang, D.B. Min, Appl. Microbiol. Biotechnol. 47 (1997) 689-694

[30] S.C. Jhansi, S.K. Mishra, J. Sustainable Dev. 10 (2013) 1–15

[31] F. Kamal, H. Mehrgan, M.M. Assadi, S.A. Mortazavi, Iran. Biomed. J. 7 (2003) 91-98

[32] S. Jeeva, T.S. Mohan, A. Palavesam, N.C.J. Packia Lekshmi, J.R. Brindha, J. Microbiol. Biotechnol. Res. 1 (2011) 175-182

[33] D.K. Khosravi, S.F. Reyhani, B. Nasernejad, Afr. J. Bio-technol. 10 (2011) 13520-13527

[34] S.V. Niknezhad, M.A. Asadollahi, A. Zamani, D. Biria, Food Sci. Biotechnol. 24 (2015) 453-460

[35] T. Ghashghaei, M.R. Soudi, S. Hoseinkhani, Appl. Food Biotechnol. 3 (2016) 15-23

[36] A. Amanullah, S. Satti, A.W. Nienow, Biotechnol. Prog. 14 (1998) 265–269

[37] S.K. Psomas, M. Liakopoulou-Kyriakides, D.A. Kyriakidis, Biochem. Eng. J. 35 (2007) 273–280

[38] M. Mirik, A.S. Demirci, T. Gumus, M. Arici, Food Sci. Biotechnol. 20 (2011) 1243-1247.


137

BOJANA Ž. BAJIĆ

DAMJAN G. VUČUROVIĆ SINIŠA N. DODIĆ

ZORANA Z. RONČEVIĆ JOVANA A. GRAHOVAC

JELENA M. DODIĆ

Univerzitet u Novom Sadu, Tehnološki fakultet Novi Sad, Katedra za biotehnologiju i farmaceutsko

inženjerstvo, Bulevar cara Lazara 1, 21000 Novi Sad, Srbija

NAUČNI RAD

BIOTEHNOLOŠKA PROIZVODNJA KSANTANA NA OTPADNOJ VODI IZ PROIZVODNJE JESTIVOG ULJA. DEO II: KINETIČKO MODELOVANJE I SIMULACIJA PROCESA

Ksantan je mikrobiološki polimer koji ima široku primenu u različitim granama industrije. Očekuje se da će u narednoj deceniji potražnja za ksantanom u velikoj meri porasti usled čega je neophodno konstantno razvijati sve aspekata ovog biotehnološkog pro-cesa. Cilj ovog istraživanja je ispitivanje kinetike diskontinualne kultivacije Xanthomonas campestris ATCC 13951 primenom otpadne vode iz proizvodnje jestivog ulja kao osnove kultivacionog medijuma za proizvodnju biopolimera ksantana. Kinetičko modelo-vanje je veoma značajno za kontrolu procesa, smanjenje njegovih troškova i povećanje kvaliteta gotovog proizvoda. Izvođenjem biosinteze ksantana na medijumu sa optimi-zovanim sastavom dobijene su eksperimetalne vrednosti sadržaja biomase, izvora ugljenika i željenog proizvoda koje su primenjene za određivanje kinetike biosinteze. Logistička, Luedeking-Piret i modifikovana Luedeking-Piret jednačina primenjene su za opisivanje kinetike nаstајаnjа biоmаsе, kinеtikе generisanja prоizvоdа i kinеtikе pоtrо-šnjе izvоrа uglјеnikа tоkоm biоsintеzе ksаntаnа, redosledom. Pored toga, primenom simulacionog softvera (SuperPro Designer®) razvijen je model bioprocesa proizvodnje ksantana i izvedena njegova ekonomska analiza. Prоduktivnоst prоcеsа prоizvоdnjе ksаntаnа primеnоm оtpаdnе vоdе iz prоizvоdnjе јеstivоg ulја kао оsnоvе kultivаciоnоg mеdiјumа iznоsi 21294,29 kg/god, dоk primеnоm pоlusintеtičkе pоdlоgе sа glukоzоm imа vrеdnоst оd 23107,97 kg/god. Simulacioni model, razvijen na osnovu definisanih kinetičkih modela, predstavlja odličnu osnovu za dalje poboljšanje i povećanje efikas-nosti bioprocesa proizvodnje ksantana.

Ključne reči: diskontinualna kultivacija, kinetičko modelovanje, ksantan, otpadne vode iz proizvodnje jestivog ulja.




139

TUTUK DJOKO KUSWORO

HADIYANTO HADIYANTO DEARISKA DEARISKA

LUTFI NUGRAHA

Chemical Engineering Department, Faculty of Engineering, University of Diponegoro, Jl. Prof. Sudharto, Tembalang, Semarang, Indonesia

SCIENTIFIC PAPER

UDC 628.3.034.2:66

ENHANCEMENT OF SEPARATION PERFORMANCE OF ASYMMETRIC CELLULOSE ACETATE MEMBRANE FOR PRODUCED WATER TREATMENT USING RESPONSE SURFACE METHODOLOGY

Article Highlights • Effects of cellulose acetate concentration on membrane performance for produced

water treatment • Effects of polyethylene glycol addition on membrane performance for produced water

treatment • Effects of nonsolvent addition on membrane performance for produced water treat-

ment • Optimization of the cellulose acetate membrane fabrication using response surface

methodology Abstract

Produced water is the wastewater generated from the process of exploration in oil and gas production, which needs special treatment. A membrane with cellulose acetate is widely used for produced water treatment, but further developments and improvements are still required. Therefore, it is important to determine the factors of separation efficiency of an ultrathin cellulose acetate membrane by assessing the influence of the composition of the dope solution. The response surface methodology was employed to determine the optimal conditions for this application. The investigations were conducted by varying the cellulose acetate polymer concentration at 18-20 wt.%, polyethylene glycol 4000 at 2-3 wt.% and nonsolvent addition at 3-5 wt.%. The evaluation of membrane performance for the produced water treatment was performed in a dead-end filtration cell with permeate water flux and rejection parameters for turbidity, total dissolved solids, Ca2+, Mg2+ and sulfides of produced water upstream and downstream of the membrane. The optimal composition of the dope solution was: 19 wt.% of cellulose acetate, 3 wt.% of polyethylene glycol, and 5.67 wt.% of non-solvent.

Keywords: optimization, asymmetric membrane, cellulose acetate, pro-duced water, response surface methodology.

Produced water from the petroleum and natural gas industry can contain organic and inorganic mat-erials. The content of the produced water depends on the geographical location of oil wells, the type of rock

Correspondence: T.D. Kusworo, Chemical Engineering Department, Faculty of Engineering, University of Diponegoro, Jl. Prof. Sudharto, Tembalang, Semarang, 50239, Indonesia. E-mail: [email protected] Paper received: 12 January, 2017 Paper revised: 18 July, 2017 Paper accepted: 25 July, 2017

https://doi.org/10.2298/CICEQ170112026K

structures, the type of hydrocarbon, as well as the various additive compounds used during exploration and production [1,2]. Moreover, it also contains high total dissolved solids (TDS), oils and fats and other organic and inorganic contaminants [3,4]. The util-ization of a membrane for produced water treatment is expected to be more effective in removing pollut-ants. However, the separation performance of exist-ing cellulose acetate membranes must be improved because of their low permeability and selectivity per-formance. Various methods for increasing the effici-

T.D. KUSWORO et al.: ENHANCEMENT OF SEPARATION PERFORMANCE… Chem. Ind. Chem. Eng. Q. 24 (2) 139−147 (2018)

140

ency of the membrane separation process have been proposed by previous researchers [5-7]. Neverthe-less, the optimal conditions were rarely studied.

There are many variables that can be inves-tigated in order to improve the efficiency of membrane separation processes such as membrane composit-ion, membrane properties and membrane fabrication process [8]. The transport properties of the mem-brane depend on its structural properties such as por-osity, hydrophilicity, and pore size, whereby those properties were influenced by three main factors: polymer composition, additives and nonsolvents. Kus-woro and his coworkers stated that polymer concen-tration was the dominant factor during the fabrication of a high-performance membrane [9]. Lai and his coworkers also indicated that the addition of non-sol-vent in the solution and porosity of the membrane significantly increase the morphology of membrane [10]. In addition, Xu and Qusay also reported that different amounts of added additives greatly affect the performance and membrane morphology [11]. A higher concentration of nonsolvent additives will inc-rease the molecular weight cut-off (MWCO) and membrane permeation (flux). Furthermore, the addi-tives have been introduced to increase hydrophilicity and diffusion of solute transport properties through a polysulfone hollow fiber membrane. These investig-ations were mostly applied for reverse osmosis, ultra-filtration and gas separation membranes. The effects of non-solvent addition on the performance of a nano-filtration membrane have not yet been systematically investigated. Therefore, the investigation on the effect of the composition of the dialysis membrane to the membrane performance is very important [12].

The mechanical design of experiments and mathematical modeling techniques were considered to simplify the use of the pertinent variables. The use of statistical design of experiments, such as factorial designs and RSM, has been applied in many inves-tigations. Ismail and Lai studied the influence and interaction factors in the manufacture of membranes using RSM [13]. In their study, they used a factorial design in order to obtain the most influential factor in the separation performance. Ismail and Lai also rep-orted that the shear rate, concentration of polymer and solvent amount are the most prominent factors of membrane separation performances [13]. Idris and co-workers performed a response surface methodol-ogy technique in order to develop mathematical models and to optimize the aqueous generation pro-cess in the thin film membrane fabrication process [12]. The final conclusion showed that the technique

was very useful in optimizing processes and mathe-matical modeling of thin film composite fabrication.

As discussed in the preceeding paragraph, the performance of the membrane can be affected by the concentration of the polymer, the ratio of non-solvent and additives. Therefore, obtaining the optimal con-ditions of cellulose acetate membrane fabrication for produced water treatment is very important. By emp-loying response surface methodology, the objective of this research was to generate the appropriate mathe-matical model and to demonstrate that the response surface model could serve as a tool to perform and optimize control variables in the CA nanofiltration membrane for produced water treatment.

.MATERİALS AND METHODS

The materials used in this research included CA (MKR Chemicals Semarang, Indonesia), which was used as a membrane forming polymer, PEG 4000 (Sigma Aldrich Chemie GmbH, Steinheim, Germany), as an additive, acetone 99.75% (Mallinckrodt Chem-icals, Dublin, Ireland), and distilled water (UPT Integ-rated Laboratory of Diponegoro University). The pro-duced water samples were obtained from PT. Pert-amina E & P, Ltd. (Cirebon, Indonesia).

This study was divided into three stages. The first stage is the production of cellulose acetate mem-branes, followed by optimization using RSM for mem-brane applications in the processing of produced water, and phase characterization. At this stage, the cellulose acetate membrane was fabricated by pre-paring solutions with compositions of 18, 19, and 20 wt.% cellulose acetate polymers, 2-4 wt.% PEG 4000 and 3-5 wt.% acetone as the solvent. For membrane casting the phase inversion method was used and was carried out on a glass plate using a casting knife. The membrane was immersed into the coagulation bath with distilled water as the nonsolvent for 1 h, fol-lowed by immersion in a different coagulation bath at ambient temperature (30±2 °C) for 24 h. The mem-brane was dried using an oven at a temperature of 60 °C for 24 h. After this process, the membrane was ready to be applied for the treatment of produced water. Rejection and permeate water flux measure-ments were conducted using a dead-end separation system. The membrane effective area in the module was determined to be 12.57 cm2. Before performing the permeability test, the membrane compaction pro-cess was conducted by using distilled water for 30 min. After the compaction process, the distilled water was replaced with produced water and kept at a cons-tant temperature of 30±2 °C. Produced water flux


141

determination was done by measuring the volume of produced water every 15 min. Membrane rejection was performed by determining the concentration of Mg2+, Ca2+, TDS and S2- before and after passing through the membrane barrier. Determination of TDS was carried out using a TDS meter while for the ana-lysis of Mg2+, Ca2+ and S2- , the titrimetric method was used. Figure 1 illustrates the simple process diagram of a dead-end cell apparatus for a produced water separation process using a prepared CA membrane. The permeate water flux was calculated by the fol-lowing equation [14]:

J = V/(PAt) (1)

where J = flux (L m-2 bar-1 h-1), V = permeate volume (L), P = pressure (bar), t = time (h) A = membrane effective area (m2). Determination of the coefficient of rejection was done by analyzing the concentration of pollutants in the upstream and downstream from the membrane.

R = 100(1-Cp/Cf) (2)

where R is percent of rejection, Cp is permeate con-centration and Cf is feed concentration. The manufac-ture parameter of membranes was optimized using a technique called response surface methodology (RSM) [15]. The central composite design (CCD) was used to design the number of experiments. Three indpendent variable trials were used for controlled variables such as concentration CA (X1), polyethylene glycol (X2), and nonsolvent (X3). Lower, upper, center and star point of the design were encoded as -1, 1, 0, and α, where +1 indicates a high level, low level –1, α = 2(n/4) (n = number of variables or factors). The star point was added to the design to produce an estimate arch in the mathematic model and it took 17 experi-mental runs [16]. Based on this design, the total num-ber of experimental runs are 2k + 2k + no, where k is the number of independent variables and no is the number of repeated experiments at the center point. For statistical calculation, the variable Xi was coded as xi according to equation:

o( ) /i iX x x xδ= − (3)

where xi is a dimensionless number of variables i, xi is the original value of the variable i, xo is the value xi at the center point, and δx is a step change, respectively [16]. Response surface methodology resulted in a second-degree polynomial equation. The equations were used as a prediction of the effect of experi-mental variables and their interactions with the res-ponse variables. Each response can be presented by

Y and x as independent variables which are expres-sed by a quadratic mathematical model:

3 32

1 1i o j j ij i j jj j

j i j jY x x x xβ β β β

= < == + + + (4)

where Yi is the predicted response, βo is the offset term, βj the linear effect, βij interaction effect, and βjj is the squared effect. In this study, the flux and rejection of Ca2+, Mg2+, sulfide, and TDS were investigated as the responses of the experimental result. The res-ponse contour and surface plots, analysis of variance, and standard deviation were developed using Statis-tica software and ANOVA was performed for statis-tical analysis of the model. This analysis included the Fisher’s F-test (overall model significance), its asso-ciated probability p(F), correlation coefficient R, and determination coefficient R2 which measures the goodness of fit of the regression model. It also inc-ludes the student’s t-value for the estimated coef-ficients and the associated probabilities p(t). For each variable, the quadratic models were represented as contour plots and surface plots.


Characterization of Produced Water

The produced water was characterized to obtain the concentration of contaminants in the produced water (Table 1). The initial characteristics of the pro-duced water were used for rejection calculation and the analysis of results. The characterization results showed that the produced water still had a high content of S2- (1536 mg/L), Ca2+ (2834 mg/L) and Mg2+ (267 mg/L) as well as TDS (6500 mg/L).

Table 1. Characterization of produced water

No. Parameter Unit Numerical value

1. 2. 3. 4. 5. 6. 7.

TDS Turbidity Sulfide

Ca2+

Mg2+ COD

Oil content

mg/L NTU mg/L mg/L mg/L mg/L mg/L

6500 80.6 1536 2834 267

150.82 0.15

Optimization of membrane CA with PEG and nonsolvent addition using RSM

The most significant and influential factors were investigated in order to optimize the process of form-ing an ultrathin cellulose acetate membrane. For interactions and effects that consist of CA concen-tration (X1), PEG concentration (X2) and the non-sol-


142

vent (X3), the permeate flux and solute rejection were observed. The comparison between experimental and prediction responses of 17 runs is presented in Table 2.

Optimization of membrane CA for permeate water flux

The flux was the highest (44.25%) in run 16 in which 19 wt.% CA concentration, 3 wt.%, PEG and 5.67 wt.% of non-solvent were used, as shown in Table 2. This result is not significantly different from the predicted results (43.19%), with the percentage error of 1.05%. This is supported by ANOVA analysis, which evaluates the accuracy and significance of the experimental results.

As shown in Table 2, the influence of the three control variables on CA membrane separation per-formance is represented in term of permeate water flux. The interaction between the factors were deter-

mined by developing the mathematical model and the significant parameters of the model are shown in Table 3.

From Table 3, the value of F-value for the regression is defined as MSreg/MSres, where MSreg is a mean square of regression, which is obtained by dividing the sum of squares of regression with the degree of freedom. MSres is the mean square of the residuals data. F-value also shows the influence of variables on the model with the hypothesis H0 (there is no influence of variables on the model); and H1

(there is influence of variables on the model). In this experiment, the F value of the calculation (Fmodel) is 16.03 and higher than Ftable value (F0.05;9.10 = 3.68) [17], meaning that H0 must be rejected and that the independent variables xi contributed the effects to the proposed model [16]. The accuracy of this model can

Table 2. Factorial central composite experimental design for fabrication of ultrathin cellulose acetate (CA) membrane and the observed response in terms of flux; PFU: permeate flux unit (L h-1 m-2 bar-1)

Run Independent variables Dependent variables Error%

Dependent variables Error% Conc. of Ca

wt.% Conc. of PEG

wt.% Conc. of non--solvent, wt.%

Yo Flux, PFU Yp Flux, PFU Yo TDS rejection, %

Yp TDS rejection, %

1 18 2 3 35.21 35.65 0.44 95.61 95.52 0.09

2 18 4 5 31.15 30.90 0.25 87.72 86.58 1.14

3 20 2 5 25.46 27.82 2.36 83.77 85.63 1.86

4 20 4 3 31.19 32.43 1.24 88.16 88.81 0.65

5 19 3 4 41.66 39.93 1.72 98.25 97.67 0.58

6 18 2 5 32.93 32.80 0.13 92.11 90.01 2.09

7 18 4 3 34.87 33.63 1.24 95.18 91.87 3.31

8 20 2 3 22.95 24.32 1.37 79.82 79.52 0.30

9 20 4 5 35.36 36.04 0.68 96.49 95.14 1.35

10 19 3 4 38.89 39.93 1.05 96.93 97.67 0.74

11 17.33 3 4 29.15 30.39 1.24 85.53 88.80 3.27

12 20.67 3 4 28.06 25.22 2.84 83.77 82.57 1.20

13 19 1.33 4 25.35 23.47 1.88 81.58 81.26 0.32

14 19 4.67 4 28.39 28.66 0.28 83.77 86.16 2.39

15 19 3 2.33 43.10 42.55 0.54 98.25 99.38 1.13

16 19 3 5.67 44.25 43.19 1.05 99.12 100.06 0.94

17 19 3 4 38.94 39.93 0.99 98.25 97.67 0.58

Table 3. ANOVA of permeate water flux

Source Sum of squares Degree of freedom Mean square F Value F0.05 Value R2

ANOVA of Flux

SS regression 637.95 9.00 70.88 16.03 3.68 0.95

S.S. error 30.95 7.00 4.42 - - -

S.S. total 669.90 - - - - -

ANOVA of TDS Rejection

SS regression 444.21 9.00 49.36 17.02 3.68 0.96

S.S. error 20.30 7.00 2.90 - - - S.S. total 464.51 - - - - -


143

also be seen from the value of the regression coefficient, R2 = 0.95. This indicates that 95% of the total variation has been covered by the model [9]. The mathematical model was developed as:

pred 1 2 3

2 2 21 2 3 1 2

2 3 1 3

39.93 1.55 1.55 0.19

4.33 4.95 1.05 2.532.53 0.03

Y X X X

X X X X XX X X X

= − + + −

− − + + ++ +

(5)

From Eq. (5), it can be seen that three factors dominantly influence the value of Y (membrane flux). The effects were described as the linear effect, quad-ratic effect, and interaction effect. Based on regres-sion coefficient values, it can be concluded that the dominant factors are concentration of CA (X1) and PEG (X2). The introduction of nonsolvent in this case does not have significant effect due to the fact that the ranges of nonsolvent concentrations are beyond the optimal range. The X1 (CA concentration) has a nega-tive coefficient that indicates the increase in concen-tration of CA will reduce the permeate flux. on the other hand, the X2 (PEG) has a positive coefficient; it indicates that with the increase in concentration of PEG, the value of flux will also increase. The quad-ratic effect exhibits negative coefficients for X1 and X2. This is because the process variable has exceeded the optimal conditions and therefore the permeate flux decreases. The interactions between CA concentra-tion and PEG concentration in the dope solution are strong, as well as the interaction between PEG con-centration and nonsolvent concentration, while the interaction between CA concentration and non-sol-vent concentration is very weak. This phenomenon shows that CA and PEG concentrations have influ-ences in the structural formation of the membrane. Both CA and PEG serve important roles in pore form-ation. Moreover, PEG also serves as a hydrophilicity enhancing agent. The interaction between PEG con-centration and the nonsolvent has significant influ-ence on the separation performance. The non-solvent (water) has a high affinity to PEG molecules. When the nonsolvent is added to the dope solution, it attracts the PEG and it leaves the void fractions in the membrane material.

The optimal composition in fabricating a CA membrane can be depicted from 3D surface contours. The contour graphs in this study represent a rising ridge surface. The surface plot and contour plot showed the optimal conditions corresponding to the maximum of flux, because the linear P-value is lower than the square P-value. The surface plot and contour plot are presented in Figure 1.

In Figure 1, the nonsolvent concentration has less significant influence on the yield, where the inc-

rease in non-solvent concentration slightly increases the yield. On the other hand, the concentration of CA has a negative effect on the yield, meaning that the increase in concentrations of CA causes a decrease in the yield. The concentration of CA has a greater impact on the yield than non-solvent concentration. It can be shown through the coefficient values in the regression equations. The higher coeficient value indicates greater influence on the response variable. It is clear that the region of the highest water flux of 99.12% corresponds to the concentration CA of 19 wt.%, PEG concentration of 3 wt.%, and non-solvent of 5.67 wt.% (Table 2).

Figure 1. Surface and contour plot, the effect of the

concentration of cellulose acetate and PEG in the fix of non-solvent to the membranes flux.

Optimization of membrane TDS towards rejection

The evaluation of Ca2+, Mg2+, S-2 and TDS reject-

ion were performed and the responses are presented in Table 2. The TDS rejection was studied in order to analyze the influence of soluble mineral pollutants.

Based on the F-value (Table 3), the Fmodel of TDS rejection (17.02) is higher than the value of Ftable (F0.05 = 3.68). It indicates that the three variables have effect on TDS rejection [9]. According to the value of Fmodel, which is higher than Ftable, the decision is to reject H0, which means that the independent variables Xi influence response variables [18]. The accuracy of this model can be observed in the value of the deter-mination coefficient (R2) of TDS rejection, as shown in Table 4. The regression equation has an R2 value of 0.96 and the p-value of < 0.05, which means the model fits according to experiment results. [9]. The regression equation for TDS rejection is:


144

TDS 1 2 3

2 2 21 2 3 1 2

2 3 1 3

80.90 0.94 0.84 0.01

3.90 3.97 0.50 2.282.39 0.53

Y X X X

X X X X XX X X X

= − + + −

− − + + ++ +

(6)

From Eq. (6), it can be seen that the most influential factors that affect membrane separation performance in terms of TDS rejection are CA con-centration (X1) and PEG concentration (X2). CA con-centration contributed to the formation of a dense layer in the membrane, while PEG contributed to pore formation during membrane fabrication. The dense layer and membrane porosity are structural properties of the membrane that controls the separation per-formance. The non-solvent in this case does not have a significant effect, because the range of the non-sol-vent concentration used in this experiment is out of the optimal range. The X1 coefficient (concentration CA) has a negative coefficient (-0.94), it indicates that the decrease in the concentration of CA will increase the rejection rate. On the other hand, the increase in concentration of PEG, the rejection performance will decrease. The quadratic effect shows a negative sign for X1 and X2 that indicates the optimal concentration for each variable. If the variables exceed the optimal concentration, it would result in low rejection value.

The maximum TDS rejection is depicted in Fig-ure 2. On the Ca2+ and sulfide rejections, the concen-tration of CA exhibits the dominant influence on the rejection performance. Another case in Mg2+ and TDS rejections, the concentration of PEG has the greatest influence on rejection performance. Furthermore, the optimal process variables in membrane fabrication are 19 wt.% CA, 3 wt.% PEG, and 5.67 wt.% non-sol-vent.

Regression coefficient significance

Table 4 exhibits the regression coefficient signi-ficance of the flux and rejection models. The student’s test was used to determine the significance of the reg-ression coefficient [9]. The coefficients with single fac-

Figure 2. Surface and contour plot, the effect of the

concentration of cellulose acetate and PEG in the fix of non-solvent towards TDS rejection.

tors represent the effects of a certain factor, while the coefficients with two factors and second order equat-ion terms represent the interaction between two fac-tors and the quadratic effect, respectively. Table 4 shows that the coefficient of concentration of poly-ethylene glycol as additive contributing the most signi-ficant effect to the determination of optimal fluxes with p-value of 0.03. Moreover, the polymer concentration and cellulose acetate had a significant negative effect, with p-value of 0.03. Figure 3 shows that with the increase in CA concentration in the dope solution, the membrane flux declines. The results are consist-

Table 4. Multiple regression results and regression coefficient significance for permeate water flux and TDS rejection

Parameter Term Water flux TDS Rejection

Coefficient t-value p-value Coefficient t-value p-value

βo 39.93 32.98237 0.000000 80.98 79.85227 0.000000

β1 X1 –1.55 -2.71246 0.030092 -0.94 0.36433 0.726368

β2 X2 1.55 2.72097 0.029724 0.84 -0.77849 0.461780

β3 X3 0.19 0.33365 0.748416 0.01 0.21368 0.836889

β12 X1 X2 2.53 3.40747 0.011327 2.28 -0.44754 0.668008

β13 X1 X3 2.53 2.13302 0.070356 2.38 0.06302 0.951510

β23 X2 X3 0.03 0.03748 0.971146 0.53 -0.11659 0.910463

β11 X12 –4.33 -6.87218 0.000237 -3.90 0.00000 1.000000

β22 X22 –4.95 -7.85559 0.000102 -3.97 0.00000 1.000000

β33 X32 1.05 1.66574 0.139707 0.50 0.00000 1.000000

R2 0.95 0.96

R 0.89 0.90

(6)


145

ent with previous studies, where the polymer concen-tration contributes to controling the permeate water flux of the membrane. The higher concentration of polymer in the dope solution produces high density membrane material [9]. The experiment also showed that increasing the concentration of PEG in the dope solution would increase the permeate flux. This is in accordance with previous research conducted by Arthanareeswaran et al. [19], who showed that inc-reasing the concentration of PEG will enhance the flux of water through improvement in pore formation. The addition of PEG to the dope solution has influ-ence on pore formation, wherein the pores formed should become bigger with the addition of PEG. PEG will stimulate the formation of macrovoids [12]. The finger-like macrovoids will increase the membrane flux. Beside the formation of pores, the PEG addition improved the hydrophilic properties of the membrane. Therefore, the permeation of water increased.

Time (hour)

0.2 0.4 0.6 0.8 1.0

Flux

(L.h

-1m

-2ba

r-1)

0

20

40

60

80 CA 18 wt%CA 19 wt%CA 20 wt%CA 17.33 wt%CA 20.67 wt%

Figure 3. Effect of cellulose acetate concentration on the flux.

The quadratic effects of CA and PEG concentra-tion provide significant effect to the separation per-formance with level of significance (p-values) of 0.0002 and 0.0001, respectively. A significant factor represents the interaction between X1 and X2 (p-value = = 0.011 < 0.05). The polymer concentration and addi-tive combination could be used in controlling the vis-cosity of the dope solution during fabrication of asym-metric cellulose acetate process. Viscous dope sol-ution causes improvement of the pore size of mem-brane which will produce asymmertric membrane with high performance in terms of fluxes.

The effect of polymer concentration on TDS rejection was also significant, with p-value of 0.008. One of the performance parameters of membrane separation is TDS rejection, which represents the mineral pollutants which are soluble in the produced

water. Figures 4 and 5 show the effect of polymer concentration on the rejection of Ca2+ and sulfide. The highest rejection was observed in the membrane with 20 wt.% concentration of CA, and the lowest rejection was observed in the membrane with 18 wt.% of CA. It can be seen that the rejection increases with an inc-rease in concentration of cellulose acetate. This phe-nomenon might occur due to the increase in the con-centration of polymer which will increase the viscosity of the casting film. Moreover, the diffusivity between the components in the system will be lower during the process of solidification of the casting solution and then inhibit the precipitation process and lead the sur-face membranes to having smaller pores [19].

Time (hour)

0.2 0.4 0.6 0.8 1.0

Ca2+

Rej

ectio

n (%

)

40

45

50

55

CA 18 wt%CA 19 wt%CA 20 wt%CA 17.33 wt%CA 20.67 wt%

Figure 4. cellulose acetate concentration effect on Ca2+ rejection.

Time (hour)

0.2 0.4 0.6 0.8 1.0

S2- R

ejec

tion

(%)

92

94

96

98

CA 18 wt%CA 19 wt%CA 20 wt%CA 17.33 wt%CA 20.67 wt%

Figure 5. cellulose acetate concentration effect on S2- rejection.

The ANOVA analysis to obtain the best pre-dicted parameters is depicted in Table 5. The maxi-mum flux was achieved when CA concentration, PEG additive and nonsolvent were 18.86, 3.12 and 4.01


146

wt.%, respectively. As depicted in Table 5, experi-mental values are 89.29 and 98.18 for Mg2+ and S-2 rejection, respectively. The error percentages between the experimental and predicted results are 6.30 and 1.29% for flux and TDS rejection, respect-ively. From these results, it can be confirmed that the statistical model is feasible for predicting experi-mental conditions by obtaining the optimal flux and rejection for produced water treatment using a CA nanofiltration membrane.

CONCLUSIONS

The optimal separation performance in terms of flux and rejection were determined. The optimal pro-cess variables were 18.86 wt.% CA concentration with 3.12 wt.% PEG together with 4.01 wt.% of the non-solvent, exhibiting the best separation perform-ance in terms of flux and rejection for produced water treatment using a cellulose acetate membrane. The main effects and interactions were successfully deter-mined using response surface methodology. Gener-ally, the significant variables in membrane fabrication were CA concentration in the total solid followed by the amount of the PEG additive. The errors between the experimental and predicted, using a model deve-loped from response surface methodology, were 6.30% for flux, 3.20% for TDS rejection, that are within the acceptable limit (5%).

Acknowledgements

The authors would like to thank the Waste Treatment Laboratory of University of Diponegoro for the support.

List of abbreviations

MWCO: Molecular weight cut off RSM: Response surface methodology CCD: Central composite design FFD: Fractional factorial design CA: Cellulose acetate PEG: Polyethylene glycol J: Permeate water flux V: Volume

P: Pressure A: Effective area of membrane t: Time Rj: Rejection CP : Concentration of pollutant in downstream Cf : Concentration of pollutant in upstream X1: Independent variable represents as polymer con-centration X2: Independent variable represents as additive con-centration X3: Independent variable represents as non-solvent concentration α: Axial point Xi : Independent variable of i xi : Dimensionless number of i x0: Value of xi in center point δx: Step change Yi : Response of i βo: Offset term βj : Linear effect βij : Interaction effect βjj : Quadratic effect p(F): Associated probability R: Correlation coefficient H0 : Null hypothesis H1: Alternative hypothesis Fmodel: Ratio of two variances from model Ftable: Ratio of two variances from statistic table Y0: Experimental response YP: Predicted response

REFERENCES

[1] F.R. Ahmadun, A. Pendashteh, L.C. Abdullah, D.R.A. Biak, S.S. Madaeni, Z.Z. Abidin, J. Hazard. Mater. 170 (2009). 530–551

[2] S. Alzahrani, A.W. Mohammad, N. Hilal, P. Abdullah, dan O. Jaafar, Desalination. 135 (2013). 18–26

[3] S. Mondal, S.R. Wickramasinghe. J. Membr. Sci. 322 (2008). 162-170

[4] T.D. Kusworo, N, Aryanti, R.A. Anggita, T.A.D. Setyorini, D.P. Utomo. J. Environ. Sci. Technol. 10 (2017). 35-43

[5] X.H. Zhen, S.L. Yu, B.F. Wang, H.F. Zheng. J. Environ. Sci. 18 (2006). 1077-1081

Table 5. Responses comparison between experimental results and predictions

Response Independent variables Dependent variables Error, %

CA, wt.% PEG, wt.% Non-solvent, wt.% Yo / % Yp / %

Flux, L h-1 m-2 bar-1 18.86 3.12 4.01 42.97 45.86 6.30

Ca rejection, % 18.86 3.12 4.01 60.11 62.92 4.46

Mg rejection, % 18.86 3.12 4.01 89.29 88.33 1.08

S Rejection, % 18.86 3.12 4.01 98.18 99.75 1.57

TDS rejection, % 18.86 3.12 4.01 82.92 85.68 3.2


147

[6] Y. Ma, F. Shi, J. Ma, M. Wu, J. Zhang, C. Gao. Desalin-ation 272 (2011) 51-58

[7] A. Motta, C. Borges, K. Esquerre, A. Kiperstok. J. Membr. Sci. 469 (2014). 371-378

[8] T.D. Kusworo, Budiyono, A.I. Wibowo, G.D. Harjanto, A.D. Yudisthira and F.B. Iswanto, Res. J. Appl. Sci. Eng. Technol. 7 (2014). 3852-3859

[9] T.D. Kusworo, A.F. Ismail, A. Mustafa, J. Eng. Sci. Technol. 10 (2015). 1116 – 1130

[10] J.Y. Lai, F. C. Lin, T. T. Wu, D. W. Wang. J. Membr. Sci. 155 (1999). 31-43

[11] Z.L. Xu, F.A. Qusay. J. Membr. Sci. 233 (2004). 101-111

[12] A. Idris, L.K. Yet, M.Y. Noordin, C.M. Kee. J. Sci. Technol. 49 (2008) 38–39

[13] A.F. Ismail, P.Y. Lay. Sep. Purif. Tech. 40 (2004) 191–207

[14] M.S.F Da Silva, Polyamide and Polyetherimide Organic Solvent Nanofiltration Membranes. PhD Thesis, University of Nova De Lisboa, 2007

[15] J.A. Cornell, Experiments with Mixtures: Designs, Models, and the analysis of Mixture Data, John Willey & Sons, Inc., New York, 2002

[16] T.D. Kusworo. Fabrication and characterization of flat sheet and hollow fiber mixed matrix membrane for gas separation, PhD Thesis, Universiti Teknologi Malaysia, 2008

[17] C. Dougherty, Introduction to Economics, 2nd ed., University Press, Oxford, 2012

[18] E. Saljoughi, M. Sadrzadeh, T. Mohammadi. J. Membr. Sci. 326 (2009). 627–634

[19] G. Arthanareeswaran, D. Mohan, M. Raajenthiren. J. Membr. Sci. 350 (2010).130-138.

TUTUK DJOKO KUSWORO HADIYANTO HADIYANTO

DEARISKA DEARISKA LUTFI NUGRAHA

Chemical Engineering Department, Faculty of Engineering, University of

Diponegoro, Jl. Prof. Sudharto, Tembalang, Semarang, Indonesia

NAUČNI RAD

POBOLJŠANJE SEPARACIONIH PERFORMANSI ASIMETRIČNE CELULOZNO ACETATNE MEMBRANE ZA OBRADU OTPADNE VODE PRIMENOM METODE ODZIVNE POVRŠINE

Otpadna voda nastala tokom eksploatacije nafte i gasa zahteva poseban tretman. Celulozno acetatna membrana se odavno široko primenjuje u tretmanu otpadnih voda, ali je potreban njen dalji razvoj i poboljšanje. Zbog toga je važno utvrditi faktore efikas-nosti razdvajanja ultratanke celulozno acetatne membrane procenom uticaja polimera na sastav rastvora. Optimalni uslovi za ovu primenu određeni su primenom metode odzivne površine. Ispitivanja su sprovedena variranjem koncentracije celulozno acetat-nog polimera od 18 do 20 mas.%, polietilen glikola 4000 od 2 do 3% i dodavanjem nerastvarača u opsegu od 3 do 5 mas.%. Evaluacija membranskih performansi za obradu otpadne vode urađena je u “dead-end” filtracionoj ćeliji u odnosu na ostvareni vodeni fluks, mutnoću, ukupno rastvorene supstance, sadržaj Ca2+, Mg2+ i sulfida otpad-ne vode pre i posle filtriranja. Optimalni sastav polimera u rastvoru bio je 19% celuloznog acetata, 3% polietilen glikola i 5,67% nerastvarača.

Ključne reči: optimizacija, asimetrična membrana, celulozni acetat, proizvedena voda, metoda odzivne površine.




149

DRAGAN PSODOROV1

VERA LAZIĆ1 MARIJANA AČANSKI1

ĐORĐE PSODOROV2 SENKA POPOVIĆ1

DRAGANA PLAVŠIĆ 2

KRISTIAN PASTOR1

DANIJELA ŠUPUT1 ZVONKO NJEŽIĆ2

1Faculty of Technology, University of Novi Sad, Novi Sad, Serbia

2Institute of Food Technology in Novi Sad, University of Novi Sad,

Novi Sad, Serbia

SCIENTIFIC PAPER

UDC 664.6:54:664.66.03:66

FATTY ACID PROFILE CHANGES IN RICOTTA-FILLED PASTRY DURING STORAGE INVESTIGATED BY A GC/MS-ANOVA

Article Highlights • Ricotta-filled pastry was packed in a special multilayered packaging material • Fatty acids profile of Ricotta cheese filled pastry was examined by GC/MS • Significant changes of most fatty acids were not observed after product storing • Significant change was observed only for cis,cis-9,12-octadecadienoic (linoleic) acid • It is appropriate to pack and storage Ricotta-filled pastry for the period of 4 weeks Abstract

Fatty acid composition of Ricotta cheese filled bakery product was examined using a GC-MS method immediately after production and packaging in the case of a control sample, and after production, packaging under air atmosphere in a seven-layer packaging material consisting of PE/Ad/PA/Ad/PE/Ad/PET, and storing during a four weeks period at room temperature, in the case of the experimental samples. The statistical significance of the fatty acid profile change was examined using ANOVA method. The results of this research showed that there are no significant changes of fatty acids composition and content after defined storing period, with the exception of diunsaturated cis,cis-9,12-octadeca-dienoic (linoleic) acid, whose average content was reduced by 83.705%. How-ever, a small amount of linoleic acid was converted to cis,trans-9,11-octadeca-dienoic (conjugated linoleic) acid. Therefore, it could be considered as appro-priate to pack and storage Ricotta-filled pastry for the period of four weeks, considering the insignificant changes of fatty acid composition and content.

Keywords: fatty acid, GC-MS, packaging material, Ricotta-filled pastry, storage.

Bakery products are an important part of people’s diet, being a good source of energy, nece-ssary for daily organism functioning [1]. These facts represent an important reason for the mass pro-duction of this kind of food [2] which, if not consumed in the recommended period, will be thrown away [3]. Food scientists' aims are pointed towards extending the product's expiration date, in order to preserve starting freshness as long as possible and to reduce food waste. However, freshness-preserving problems

Correspondence: K. Pastor, Faculty of Technology, University of Novi Sad, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia. E-mail: [email protected] Paper received: 12 April, 2017 Paper revised: 24 June, 2017 Paper accepted: 25 July, 2017

https://doi.org/10.2298/CICEQ170412027P

are mostly based on microbiological contamination of products, which is not an exclusive cause of food qua-lity and expiration date reduction. The changes that occur during the storing period could be caused by different factors that affect the physicochemical char-acteristics of the product.

Ricotta cheese filled laminated pastry belongs to a group of complex bakery products, whereby the pro-duct contains cheese filling, besides the base created by laminated dough. Ricotta is a type of cheese, which is produced by combining heat and acid treat-ment, in order to cause whey protein coagulation. Due to nutritional characteristics, high aw value and low concentration of salt, ricotta has no more than few days of shelf life, which mostly depends on the quality of the packaging material and the atmosphere inside the packaging [4]. However, considering the fact that

D. PSODOROV et al.: FATTY ACID PROFILE CHANGES IN RICOTTA-FILLED… Chem. Ind. Chem. Eng. Q. 24 (2) 149−155 (2018)

150

bakery products with cheese filling are subjected to heat treatment, the result could be a reduction of a total number of microorganisms and the change in nutritional and sensory characteristics of the product.

The packaging material and the packaging pro-cess might play an important role in preserving the initial product characteristics. Packaging materials with good barrier properties allow the stabile storing conditions and preserve food products from external conditions, to a certain extent [5]. Some synthetic-based multilayer packaging materials possess good barrier properties towards the oxygen, moisture and aromatic compounds, and therefore could affect the preservation of freshness [6,7]. Polyamide(PA)-based packaging materials possess good barrier properties towards gasses and aromatic compounds. Thus, using polyamide with polyethylene (PE) films, enhanced moisture barrier properties might be obtained [8]. However, it is necessary to determine whether the packaging in the air atmosphere affects potential change of fatty acids in ricotta-filled lamin-ated pastry. Alam & Goyal pointed out that oxygen remained inside the packaging could affect the cheese quality [9]. Considering that, it is assumed that pastry containing dairy products could undergo a certain change of fatty acids during the storing pro-cess. A significant fatty acid change might cause a certain modification of the nutritional and sensorial properties of a food product.

Besides microbiological defects and starch ret-rogradation, as the main causes of the pastry pro-ducts quality deterioration, it is also necessary to fol-low the changes of fatty acids contents, especially in food with high fat concentrations. During the storing period, the pastry products with increased fat content are subjected to oxidation, which lowers nutritional and sensorial values of the products [10,11]. Oxid-ation and lipolyses are certainly some of undesirable appearances that might occur during cheese product storage, whereby the negative effect is the change of nutritional and sensorial properties of the product [12].

Autooxidation - the reaction of double bonds in unsaturated fatty acids with the atmospheric oxygen - could have a detrimental effect on oganoleptic and toxicological changes of the product [11], and could cause the food spoilage, but may also potentially threaten the health of consumers [13].

Lately, the impact of various fatty acids on human health has been increasingly studied. The cause and effect relationships, between daily intake of fatty acids and cardiovascular diseases, degen-erative and inflammatory arthritis, cancer and osteo-poroses, were determined. Moreover, some fatty

acids are recognized as a trigger of the serum low density cholesterol (LDL) increase. Changes in fatty acids composition and content, during the storing period, are particularly interesting, due to the potential transformation of double bonds from cis to trans form, whose daily intake, according to the references, should be limited to the minimal amount [14]. Milk is a good source of fat in human nutrition, and especially saturated fatty acids, because they are present in the amount of 70% [15]. However, the concentration of trans fatty acids (TFA) ranges between 3 and 6%. An excessive intake of TFA could have a negative impact on human health, but the moderate intake of these fats, especially milk originated, is considered safe [16].

Several different analytical methods were recog-nized for determination of lipid oxidation. However, the standard method for all types of food products has not been established. Gas chromatography coupled with mass spectrometry proved to be a good choice to determine the lipid and carbohydrate content of ama-ranth flour [17]. Lipid oxidation can also be assessed by quantitatively measuring the loss of initial sub-strates. In foods containing fats or oils, unsaturated fatty acids are the main reactants whose composition changes significantly during oxidation. Changes in fatty acid composition provide an indirect measure of the extent of lipid oxidation.

The aim of this work was to determine whether a significant change of fatty acid composition and con-tent would occur in ricotta-filled pastry, packed under air atmosphere in specially selected packaging mat-erial, after four weeks of storage at 4 °C (±2 °C).

EXPERIMENTAL

Ricotta-filled pastry production

Two identical dough batches were prepared in Ekomil pite d.o.o. factory, in Bačka Palanka, Serbia. Every batch consisted of wheat flour (255 g), water (90 g) and salt (5 g), which is a measure for one pastry product. The mixtures were kneaded in the mixer (Kemper SP 30, Germany) for 8 min in the first gear and 4 min in the second gear. The laminating of the dough was conducted applying a special machine (Mateks Makina, Turkey) in order to reach 0.4 mm dough thickness, followed by the addition of Ricotta cheese (100 g) and the final shaping into the pipe form. The pastry was then baked for 40 min at 220 °C and cooled down for 20 min at 20 °C. Cold pastry was packed into the seven-layer packaging material, con-sisted of PE/Ad/PA/Ad/PE/Ad/PET in the air atmo-sphere by DZQ vacuum packager (DZQ-600/2SB, China). After the packaging process, a control sample


151

was taken from the paste created from one batch, while the experimental sample was taken from the paste created from the other batch. The control sample was analyzed immediately after the pack-aging process (A). The pastry samples were taken both from the edges (A1), and from the middle part of the pastry, in which cheese dominates (A2). Experi-mental sample was stored at 4 °C (±2 °C). and the analysis was performed after four weeks (B). Like-wise, the edges of the pastry were labeled as B1, and the middle part as B2.

Preparation of samples

Five grams of the edge samples (A1, B1) and five grams of the middle parts (A2, B2) were weighed and homogenized in a blender. A sample portion (1.61 g each) was collected from every sample and transferred to centrifugation test tubes. The 3 ml of methylene-chloride was added for extraction of lipo-solubile substances, and the prepared mixture was firstly homogenized for 1 min in the Vortex mixer. Fur-thermore, the samples were centrifuged for 10 min at 2000 rpm. One ml of the lipid supernatant from every sample was transferred into vials, and 50 µl of the derivatization reagent (0.2 M TMSH, trimethylsulfo-nium-hydroxide in methanol, Macherey-Nagel) was added, by which trans-esterification reaction of fatty acids, from triacylglycerol into corresponding evapor-ative fatty acid methyl esters, was conducted [18-22].

GC-MS Analysis

Analysis of lipid (fatty acid) profile from the edge and the middle parts of the investigated samples was conducted by the application of gas chromatography (GC, Agilent Technologies 7890) coupled with the mass-spectrometric detection device (MS, Agilent Technologies MSD 5975). The standard conditions were applied for the analysis of fatty acid methyl esters by GC-MS system. An electron ionization method with the energy of 70 eV was applied. A DB-5 MS column (60 m×0.25 mm×25 μm) was used and the following temperature program was applied: 50– –130 °C, 30 °C·min-1 and 130–280 °C, 15 °C·min-1, whereas at the end of every analysis the temperature was held for 8 min at 280 °C. The temperature of the injector was 250 °C, and the helium flow, as a gas carrier, was 1.1563 ml·min-1. One μl of a solution from every analyzed sample was injected with the split- -ratio of 1:50 [23,24].

Data processing

The obtained chromatograms were processed using a MSD ChemStation Data Analysis (Agilent Technologies) program, and the fatty acid peaks, in

the form of the corresponding methyl esters, were identified by the examination and comparison of their characteristic fragmentations with the Wiley 275 mass spectra library, using the AMDIS, with a probability-based matching (a match quality of 95% minimum was used as a criterion). Surface areas of detected fatty acid methyl esters were integrated from total ion current (TIC) chromatograms, both automatically and manually in control and experimental pastry samples. Mean values were determined and afterwards com-pared using ANOVA statistical method.


Figure 1 shows total ion current chromatograms (TIC) of the edge samples of ricotta-filled pastry (A1 and B1, Figure 1A) and the middle parts of the same pastry (A2 and B2, Figure 1B), whereas the pastry samples A1 and A2 were analyzed immediately after the packaging process and samples B1 and B2 after packaging and storing for the defined period of four weeks. By looking at the presented chromatograms a high similarity of eluting peaks could be observed.

Table 1 lists the detected fatty acids, their ret-ention times, relative content with regard to the most abundant hexadecanoic (palmitic) acid (100%), and standard deviations.

It might be concluded that the presence of fatty acids, which originate from Ricotta cheese applied for the pastry production, has been typical for the milk fat, whose composition and content could be influ-enced by various factors [25]. Numerical values of the peak surface areas were obtained by a more rapid and convenient automatic integration of the detected fatty acid methyl esters. The relative contents of every fatty acid detected compared to the most abundant palmitic acid (abundance 100%) were subjected to statistical calculations. By the application of the ana-lysis of variance (ANOVA), the relations between the variance of intergroup variability were tested for every detected fatty acid in analyzed edge and middle part pastry samples.

The results did not show statistically significant differences in the contents of any detected fatty acid between the samples of the pastry product analyzed after the packaging (A1 and A2) and the pastry pro-duct subjected to storing (B1 and B2). Obtained F value did not exceed an F-critical value in any case. Likewise, p value was higher than 0.05 in every case.

An exception was the diunsaturated cis,cis-9,12- -octadecadienoic acid (linoleic acid), whose F value (42.8850) exceeded F–critical (9.5520). Moreover, p value (0.0062) was lower than 0.05. That implies the


152

Figure 1. Total ion current fatty acids chromatogram of the edges (A1 and B1) and the middle part (A2 and B2) of Ricotta-filled pastry

after packaging (A) and after packaging and storing (B).

Table 1. Retention times and relative content of fatty acids detected in Ricotta-filled pastry

No. Rt / min Fatty acid Index A1, % A2, % SD / % B1, % B2, % SD / %

1 6.88 Octanoic (caprylic acid) C8:0 2.57 2.87 0.12 2.17 2.34 0.21

2 8.39 cis-4-decenoic acid C10:1ω6 0.60 0.64 0.15 0.22 0.43 0.03

3 8.45 Decanoic acid C10:0 7.06 7.90 0.17 6.19 6.43 0.59

4 10.05 Dodecanoic (lauric acid) C12:0 9.42 10.37 0.11 8.31 8.46 0.67

5 10.84 Tridecanoic acid C13:0 0.26 0.30 0.09 0.13 0.26 0.03

6 11.31 12-Methyltridecanoic acid C14:0 0.40 0.43 0.03 0.33 0.38 0.02

7 11.51 cis-9-Tetradecenoic acid (myristoleic acid) C14:1ω5 2.86 3.09 0.42 3.10 2.50 0.16

8 11.57 Tetradecanoic (myristic) C14:0 36.13 39.14 0.65 34.36 33.44 2.13

9 12.09 12-Methyltetradecanoic acid C15:0 1.58 1.81 0.04 1.60 1.66 0.16

10 12.29 Pentadecanoic acid C15:0 3.58 3.96 0.03 3.81 3.76 0.27

11 12.86 cis-9-Hexadecenoic acid (palmitoleic acid) C16:1ω7 6.28 5.66 0.41 5.83 5.25 0.44

12 12.98 Hexadecanoic (palmitic acid) C16:0 100.00 100.00 0 100.00 100.00 0

13 13.49 4-Methylhexadecanoic acid C17:0 1.61 1.41 0.08 1.70 1.58 0.14

14 13.54 cis-10-Heptadecenoic acid C17:1ω7 1.07 1.07 0.07 0.91 1.01 0

15 13.68 Heptadecanoic acid C17:0 2.26 2.49 0.22 2.31 2.00 0.16

16 14.22 cis,cis-9,12-Octadecadienoic acid (linoleic acid) C18:2ω6 46.63 44.61 1.05 9.68 8.19 1.43

17 14.26 cis-9-Octadecenoic acid (oleic acid) C18:1ω9 93.18 93.36 3.85 87.75 82.31 0.13

18 14.30 trans-9-Octadecenoic acid (elaidic acid) C18:1ω9 11.89 13.38 1.89 14.38 11.71 1.05

19 14.36 trans-15-Octadecenoic acid C18:1ω9 3.10 3.02 0.90 3.45 4.73 0.06

20 14.41 Octadecanoic acid (stearic acid) C18:0 47.74 48.82 0.12 48.08 47.91 0.76


153

Table 1. Continued

No. Rt / min Fatty acid Index A1, % A2, % SD / % B1, % B2, % SD / %

21 14.58 cis,trans-9,11-Octadecadienoic acid (conjugated linoleic acid) C18:2ω7 2.17 2.35 0.15 2.67 2.46 0.13

22 15.03 cis-10-Nonadecenoic acid C19:1ω9 0.29 0.33 0.01 0.29 0.28 0.03

23 15.20 Nonadecanoic acid C19:0 0.30 0.28 0.01 0.24 0.25 0.01

24 15.87 cis-11-Eicosenoic acid (gondoic acid) C20:1ω9 0.96 0.94 0.15 0.83 0.62 0.01

25 16.09 Eicosanoic acid (arachidic acid) C20:0 0.81 0.87 0.05 0.60 0.53 0.04

significant statistical change in this fatty acid content after the defined storage period.

Due to the aforementioned, it was decided to select each fatty acid containing 18 carbon atoms in the molecule, and eluting time between 14.16 and 14.50 min, as the area of interest (Figure 2). The aim was to perform another method of chromatogram integration - a manual integration, to verify the results obtained using an automatic integration, and make a statistical comparison between contents of the fatty acids that eluate between mentioned retention times, obtained from the samples A and B. Indicated fatty acids includes: hexadecanoic (palmitic acid), cis,cis-

-9,12-octadecadienoic (linoleic acid), cis-9-octadece-noic (oleic acid), trans-9-octadecenoic (elaidic acid), trans-15-octadecenoic, octadecanoic (stearic acid) and cis,trans-9,11-octadecadienoic acid (conjugated linoleic acid).

Figure 2 presents a zoomed part of a TIC chromatogram with the focus on the retention time interval of the area of interest, for the samples of the edge (A1 and B1, Fig. 2.A) and the middle (A2 and B2, Fig. 2.B) of Ricotta-filled pastry samples analyzed immediately after production and packaging (A1 and A2) and the pastry samples packed and stored for the defined time period (B1 and B2).

Fig. 2. Enlarged total ion current chromatogram of fatty acids with focus on the area of interest from the edges (A1 and B1) and the

middle part (A2 and B2) of Ricotta-filled pastry after packaging (A) and after packaging and storing (B).


154

The difference in linoleic acid (cis,cis-9,12-octadecadienoic) content between the samples of edge (A1 and B1) and the middle parts (A2 and B2, Figure 2) of the bakery product analyzed immediately after the packaging (A1 and A2) and the same pro-duct analyzed after packaging and storing in a def-ined time period (B1 and B2), could be obviously obs-erved in Fig. 2.

By repeated statistical data analysis of the area of interest, using the ANOVA test, it was concluded that only diunsaturated cis,cis-9,12-octadecadienoic (linoleic) acid suffered statistically significant losses during the storing period, owing to oxidation and deg-radation ability, due to the presence of two double bonds in the molecule. The content of linoleic acid was decreased by 85.55% in the samples of the edges and by 81.85% in the samples of the middle parts of the pastry (mean value 83.7%). Considering the moderate increase of conjugated linoleic (cis,trans-9,11-octadecadienoic) acid content in the area of interest (by 24.72%), we assume that cis,cis- -9,12-octadecadienoic (linoleic) acid was most pro-bably partially transformed in a more stabile trans iso-mer, cis,trans-9,11-octadecadienoic (conjugated lino-leic acid). Furthermore, it was also probaly catabo-lized partially to the products of lower molecular weight, such as aldehydes, ketones and alcohols, which could not be detected by the applied analytical method [26]. On the other hand, a slight increase in the content of the cis-9-octadecenoic (oleic) acid was observed based on the standard deviation values, although ANOVA test evaluated this change as insig-nificant.

Even though the concentration of conjugated linoleic acid, specifically, did not increase signific-antly, the increase of the fatty acid content in the trans form was detected. However, conjugated linoleic acid was frequently labeled as beneficial for human health, because it reportedly shows anticarcinogenic, anti-atherogenic and antidiabetic effects [16,27]. On the other hand, EFSA reported (2010) that there are no convincing evidences about the beneficial effect of conjugated linoleic acid on human health. In accord-ance to above mentioned report, there are no recom-mended daily intakes of conjugated linoleic acid.

CONCLUSION

The storing of Ricotta-filled pastry product packed in a specially selected multilayered packaging material during the defined time period could be sig-nified as appropriate, considering the insignificant changes of fatty acids composition and content. GC-

-MS analysis proved that the choice of packaging material, storing period and packaging conditions were adequate for the fatty acid preservation in Ric-otta-filled pastry. However, it is important to empha-size that fatty acid analysis, performed without micro-biological and sensory analysis, shouldn’t have an effect on decision of storing period and the best before date of the food product. The composition of initially detected fatty acids remained unchanged after specified time period of storage, except in the case of diunsaturated cis,cis-9,12-octadecadienoic (linoleic) acid. A slight increase of cis,trans-9,11-octadeca-dienoic acid (conjugated linoleic acid) occurred, due to the oxidation process, but it didn’t essentially affect the nutritional value of the bakery product, due to its low concentration. Additionaly, a certain amount of linoleic acid was probably transformed into the pro-ducts of lower molecular mass. The general good manufacturer practice requires the maintenance of microbiological and sensory properties of the starting food product, without any reference to the potential nutritional changes, such as fatty acid profile changes. Therefore, the necessity of fatty acid ana-lysis due to a food product sustainability measure-ment, should be considered. GC-MS fatty acid ana-lysis perform along with the microbiological and sen-sory analysis, should answer the questions whether Ricotta-filled pastry product could be stored for a selected period of time.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the financial support from the Ministry of Education, Science and Technological Development of the Republic of Serbia.

REFERENCES

[1] E.M. Ali Karrar, Int. J. Nutr. Food Sci. 3 (2014) 592-596

[2] H.P.S. Nagi, J. Kaur, J.B.N. Dar, S. Sharma, Am. J. Food Technol. 7 (2012) 301-310

[3] P. Amani, L. Gadde, in: International Europian Forum, Innsbruck-Igls, Austria, 2015, pp. 17-21

[4] M.A. Del Nobile, A. Conte, A.L. Incoronato, O. Panza, Afr. J. Microbiol. Res. 3 (2009) 137-142

[5] V. Lazić, J. Budinski-Simendić, J. Gvozdenović, B. Simendić, Acta Phys. Pol., A 117 (2010) 855-858

[6] S.K. Meshram, S.K. Mendiratta, B.P. Mishra, B.D. Sharma, S. Chand, P.P. Prabhakaran, Int. J. Adv. Res. Technol. 3 (2014) 90-101

[7] N. Hromiš, V. Lazić, S. Popović, S. Markov, Ž. Vaštag, D. Šuput, S. Bulut, V. Tomović, J. Food Nutr. Res. 55 (2016) 78-88

[8] M. Kirwan, S. Plant, J. Strawbridge, in: Food and beverage packaging technology, 2nd ed., R. Coles, M.


155

Kirwan, Eds., John Wiley & Sons, Hoboken, NJ, 2011, pp. 157-212

[9] T. Alam, G.K. Goyal. J. Food Sci. Technol. 48 (2011) 120-123

[10] T.K. Dalsgaard, J. Sørensen, M. Bakman, L. Vognusen, C. Nebel, R. Albrechtsen, J. Nielsen, Dairy Sci. Technol. 90 (2010) 565-577

[11] F. Pop, D. Boltea, J. Agroalim. Proces. Technol. 20 (2014) 39-45

[12] S. Khoshgozaran, M.A. Azizi, N. Bagheripoor-Fallah, Dairy Sci. Technol. 92 (2011) 1-24

[13] S. Chudy, J. Pikul, M. Rudzińska, J. Food Nutr. Res. 54 (2015) 31-40

[14] European Food Safety Authority. Scientific opinion on dietary reference values for fats, including saturated fatty acids, monounsaturated fatty acids, trans fatty acids, and cholesterol, EFSA J. 8 (2010) 1-107

[15] F. Farbod, A. Kalbasi, S. Moini, Z. Emam-Djomeh, H. Razavi, A. Mortazavi, H. Beleshti, J. Nutr. Food Sci. 3 (2013) 1-8

[16] T. Ritvanen, T. Putkonen, K. Peltonen, Food Nutr. Sci. 3 (2012) 1189-1196

[17] E. Cequier-Sánchez, C. Rodríguez, Á.G. Ravelo, R. Zárate, J. Agric. Food Chem. 56 (2008) 4297–4303

[18] R. Anthony, B.Đ. Psodorov, M. Ačanski, Đ. Vujić, J. Brkljača, D. Psodorov, Chem. Ind. Chem. Eng. Q. 21 (2014) 71-76

[19] R.A. Moreau, M.J. Powell, V. Singh, J. Am. Oil Chem. Soc. 80 (2003) 1063-1067

[20] J. Dron, R. Linke, E. Rosenberg, M. Schreiner, J. Chromatogr., A 1047(1) (2004) 111-116

[21] A. Renil, B. Stuart, Frontiers Energy Res. 2 (2015) 1-5

[22] http://www.mn-net.com/tabid/10234/default.aspx

[23] M. Ačanski, Đ. Vujić, Đ. Psodorov, Food Chem. 172 (2015) 314–317

[24] K. Pastor, M. Ačanski, Đ. Vujić, Đ. Jovanović, S. Wien-koop, Chromatographia 79 (2016) 1387–1393

[25] P.J. Moate, W. Chalupa, R.C. Boston, I.J. Lean, J. Dairy Sci. 90 (2007) 4730-4739

[26] I.H. Han, A.S. Csallany, JAOCS 86 (2009) 253-260

[27] P.W. Parodi, J. Dairy Sci. 82 (1999) 1339-1349.

DRAGAN PSODOROV1 VERA LAZIĆ1

MARIJANA AČANSKI1

ĐORĐE PSODOROV2 SENKA POPOVIĆ1

DRAGANA PLAVŠIĆ 2

KRISTIAN PASTOR1

DANIJELA ŠUPUT1 ZVONKO NJEŽIĆ2

1Tehnološki fakultet, Univerzitet u Novom Sadu, Bulevar cara Lazara 1,

21000 Novi Sad, Srbija 2Institut za prehrambene tehnologije u

Novom Sadu, Univerzitet u Novom Sadu, Bulevar cara Lazara 1, 21000

Novi Sad, Srbija

NAUČNI RAD

PROMENE U PROFILU MASNIH KISELINA U PECIVU PUNJENOM RICOTTA SIROM U TOKU SKLADIŠTENJA ISPITANE PRIMENOM GC/MS-ANOVA

Sastav masnih kiselina pekarskog proizvoda sa Ricotta sirom ispitan je metodom GC-MS, odmah nakon proizvodnje i pakovanja u slučaju kontrolnog uzorka, a u slučaju ekspe-rimentalnih uzoraka nakon proizvodnje, pakovanja u atmosferi vazduha u sedmo-strukom ambalažnom materijalu koji se sastoji od PE / Ad / PA / Ad / PE / Ad / PET i skladištenja u periodu od četiri nedelje na sobnoj temperaturi. Statističi značajna pro-mena u profilu masnih kiselina ispitana je pomoću ANOVA. Rezultati ovog istraživanja pokazali su da nema značajnih promena u sastavu i sadržaju masnih kiselina i nakon definisanog perioda čuvanja, sa izuzetkom dinezasićene cis,cis-9,12-oktadekadienske (linolne) kiseline, čiji je prosečni sadržaj smanjen za 83,705%. Međutim, mala količina linolne kiseline pretvorena je u cis,trans-9,11-oktadekadiensku (konjugovanu linolnu) kiselinu. Prema tome, pakovanje i skladištenje peciva napunjenog Ricotta sirom u tra-janju od četiri nedelje može se smatrati odgovarajućim, s obzirom na beznačajne pro-mene sastava i sadržaja masnih kiselina.

Ključne reči: masne kiseline, GC-MS, ambalažni materijal, punjeno pecivo Ricotta sirom, skladištenje.




157

FAEZEH SHARIFI

MANSOUR JAHANGIRI

Faculty of Chemical, Petroleum and Gas Eng., Semnan University,

Semnan, I.R. Iran

SCIENTIFIC PAPER

UDC 577.161.2:66.02:615.451.2

INVESTIGATION OF THE STABILITY OF VITAMIN D IN EMULSION-BASED DELIVERY SYSTEMS

Article Highlights • Vitamin D was successfully encapsulated by emulsion procedure • The soy protein isolate can protect 85% of vitamin D from degradation • The size distribution of nanocapsules and polydispersity index were 104 nm and 0.4,

respectively • The SEM analysis of dried nanocapsules of vitamin D shows the round shape and

homogeneity dispersion Abstract

Vitamin D is a nutraceutical agent, which is necessary for good health. However, the sufficient amount of this vitamin needed for daily intake is not found in most foods which leads to many producers choosing to develop vitamin-enriched products. Vitamin D is sensitive to the exposure to oxygen and high temperature. To protect it against degradation during food processing, emulsion-based delivery is preferred. The more stable emulsion leads to higher protection of vitamin D. The present study investigated the effects of different factors, such as the choice of biopolymer, pH, ionic strength, and temperature, on emulsion stability. As emulsions with smaller particles are known to be more stable, the minimum concentrations of the biopolymers under study allowing the minimum size of particles were determined. The results obtained were the following: gum arabic 7 %, 468 nm; maltodextrin 2 %, 266 nm; Whey protein concentrate (WPC) 0.5 %, 190 nm; Soy protein isolate (SI) 4 %, 132 nm. Among the different biopolymers and the emulsion conditions studied, the soy protein isolate emulsion provided the highest protection of vitamin D (85 %) at 4 wt% concentration, pH 7 and 25 °C. SEM analysis of the dried nanocapsules of the soy protein isolate emulsion revealed homogeneous and uniform dispersion of particles.

Keywords: vitamin D, emulsion, stability, soy protein isolate, nanocap-sules.

Using emulsion systems is a way to protect sensitive nutraceuticals from chemical degradation and is preferred in the food industry and drug delivery systems [1]. An emulsion is a suitable way of dis-persing lipophilic bioactives in aqueous environments to be further used in foods and pharmaceuticals. In

Correspondence: M. Jahangiri, Faculty of Chemical, Petroleum and Gas Eng., Semnan University, Semnan, Zip code: 35196-45399, I.R. Iran. E-mail: [email protected] Paper received: 8 April, 2016 Paper revised: 23 April, 2017 Paper accepted: 8 August, 2017

https://doi.org/10.2298/CICEQ160408028S

the human gastrointestinal tract, the lipid breaks down and makes mixed micelles that can solubilize and transfer other lipids. Vitamin D is a fat-soluble nutra-ceutical, which is very important for calcium and phosphorus absorption and necessary for cardiovas-cular health, helping to prevent cancer and improve the immune system [2]. Vitamin D exists in the form of vitamin D2 and D3, the latter being more bioavailable than the former. Many people around the world suffer from insufficient vitamin D, mainly caused by avoid-ance of sun exposure for preventing melanoma. Lack of this vitamin can be complemented by food fortific-ation. Vitamin D is therefore added to foods in the

F. SHARIFI, M. JAHANGIRI: INVESTIGATION OF THE STABILITY OF VITAMIN D… Chem. Ind. Chem. Eng. Q. 24 (2) 157−167 (2018)

158

forms of vitamin D2 or D3 in accordance with applic-able legislation [3].

Vitamin D is an important nutraceutical that is not water-soluble and is vulnerable to heat and oxy-gen exposure. Due to its importance, the subject of vitamin D encapsulation has been thoroughly studied by researchers [2-9]. In addition, different methods exist for encapsulation, such as salting-out, solvent evaporation and emulsification, which are the most common methods in food industries. Depending on the physical and chemical properties of nutrition and its application in food industry, specific methods of the encapsulation were designed [4]. Two different methods were designed for encapsulation of dairy products with vitamin D, precrystalline vitamin D and emulsified vitamin D. The results showed that emul-sified vitamin D is more stable in the products [5]. The influence of a number of parameters, such as differ-ent surfactants and stirring conditions, of spontane-ous emulsification of vitamin D was studied, and sup-ported the idea that it has great potential for food and pharmaceutical applications [6]. Complex nanopar-ticles from the polymer of carboxymethyl chitosan (CC) and soy protein isolate (SI) encapsulated vit-amin D. The effects of Ca2+ concentration, pH and the polymer ratio (CC/SI) on the formation of the nano-particles were investigated. The results indicated that the obtained nanoparticles reduced the release of vitamin D in simulated gastric fluid and increased its release under intestinal conditions [2]. In another study, corn protein hydrolysate encapsulated vitamin D where a novel carrier was used to enhance phys-icochemical stability and bioavailability of vitamin D. The results showed that the carrier could improve the bioavailability of vitamin D when exposed to the light [7]. Vitamin D was also encapsulated by protein-fatty acid and the findings show that by using different pro-teins and fatty acids, the stability of vitamin D inc-reased when exposed to temperature of 37 °C or UV light [8].

Emulsification such as microemulsions and nanoemulsions, are the most convenient methods to encapsulate lipophilic nutritional compounds such as vitamin D within an aqueous phase [9]. The emulsi-fication has been proven as an efficient method to increase vitamin D water solubility and stabilize it against surrounding conditions. In this way, a stable vitamin D emulsion can be easily used directly in its aqueous phase in foods such as dairy products, juices and soft drinks.

The emulsion stability over time and under envi-ronmental conditions is extremely important. It can be measured by different mechanisms, such as cream-

ing index, flocculation, sedimentation, coalescence between droplets and phase inversion [10]. An imp-ortant point in emulsion stability is to achieve a sta-bilized droplet by the use of the emulsifier [11]. The emulsifier is adsorbed to the surface of the oil droplet and improves oil stability [12]. Also, the emulsifier reduces the interfacial tension and prevents droplet aggregation [13].

Different kinds of emulsifiers are being used in the food and pharmaceutical industries, including proteins, polysaccharides and surfactants [14], but natural emulsifiers are generally preferred over syn-thetic ones. Polysaccharides are most often used in emulsions, and proteins less [11]. Each bioemulsifier has its advantages and disadvantages during emulsi-fication. Most globular proteins after homogenization produce smaller particles, but these particles are not stable over time [15]. Many studies demonstrate that polysaccharides in emulsions are more stable than proteins during storage [16]. However, they need a higher concentration in emulsion, compared to pro-teins [17]. Gum arabic and modified starch are two other common types of emulsifiers, which have good water solubility [18]. Soy protein isolate also has high water solubility potential and balanced amino acid, which can permit good interaction with nutraceuticals [2].

In the present study, the effect of four different emulsifiers and salts was evaluated for the first time. Besides, this research investigates the effects of dif-ferent emulsifiers, such as gum arabic, maltodextrin, soy protein isolate, 35% whey protein concentrate, and the effects of factors such as the choice of bio-polymer, pH, ionic strength and temperature on the stability of an emulsion, which can provide protection of vitamin D. The aim of this work is the studying of using new bioemulsifiers like soy protein isolate and maltodextrin in order to improve the stability of the emulsion of vitamin D. Also, Ca2(PO3)4 was used for measuring the ionic strength of the emulsion stability. The stable produced emulsion is converted to the capsules of vitamin D by spray dryer. The size and dispersion of the capsules were also investigated.


Vitamin D was purchased from Sigma Aldrich. Medium-chain triglycerides(MCT) (Miglyol® 812) oil was purchased from Warner Graham Company. Gum arabic (GA) (viscosity 25%, Brookfield LVL, 60 rpm at 20 °C, 150 mPa s) was supplied by Nexira. Soy pro-tein isolate (SI) was obtained from Foodchem (the soy flour contained 54% protein and less than 1% oil (dry basis)). Maltodextrin (MD, dextrose equivalent


159

18-20) was purchased from the National Starch Com-pany. Whey protein concentrate 35% (WPC 35%) was provided by Behtampowder Company. Sodium chloride, calcium phosphate, sodium citrate and sodium azide were provided by Sigma Aldrich. Double distilled water was used for all emulsions.

Preparation of oil in water emulsions

The four emulsifiers such as WPC, GA, SI, MD with concentrations between 0.5-7 wt.% were dis-persed in buffer solutions containing 10.0 mM sodium citrate, 0.01% sodium azide, at pH 3. The oil phase contained 2.5 wt.% vitamin D, which was dissolved in medium-chain triglycerides (MCT). These two phases were mixed by using an M133/1281-0 high-speed mixer (Biospec Products, Inc., Bartlesville, OK, USA) for 5 min at 10,000 rpm. Then, 4 wt.% of the oil phase was added to 96 wt.% aqueous phase at 25 °C and the emulsion was homogenized [19,20]. Then the emulsion was passed through an M-110L high-pres-sure microfluidizer for five passes at 9,000 psi (Mic-rofluidics, Newton, MA, USA). The procedure was repeated using each emulsifier individually.

Investigation of different environmental conditions on emulsion stability

First, four emulsifiers (35% whey protein con-centrate, gum arabic, soy protein isolate and malto-dextrin) with different concentrations are considered. The best concentration for a stable emulsion is sel-ected. Different parameters such as pH, which ranges between 2 and 8, and two salts, NaCl and Ca2(PO3)4 with different concentrations and temperatures between 30 and 90 °C, were analyzed. The stability of the emulsion is measured by a particle size analyzer and creaming index. The best conditions of emulsifier type, pH, salts concentrations and temperature are chosen. Then, four stable emulsifier emulsions are prepared at these conditions and analyzed during one month in order to protect vitamin D. Finally, the stable emulsion which can protect vitamin D better than the others is our desirable emulsion. The desirable emul-sion is converted to capsules. These capsules were analyzed by SEM and particle size analyzer. The morphology of these capsules is very important. The surface of these capsules should be smooth without cracks in which vitamin D couldn’t penetrate to the outside of the capsules.

First, different concentrations of the emulsifiers were tested, and the emulsions that exhibited the highest stability (the most suitable concentration of emulsifier) were selected for further assessing their stability under different conditions. In order to charac-terize the pH, salt and temperature stability, particle

size and zeta potential measurements of the emul-sions under different conditions were performed.

pH stability

Emulsions were prepared with pH values in the range 2 < pH < 8 by aqueous buffer solutions. By using NaOH and HCl, the pH values were adjusted to desired values and then 25 ml of emulsions were transferred to glass tubes at ambient (25 °C) tem-perature for 24 h before the analysis [19].

Salt stability

The emulsions with pH 7.0 were selected to test different amounts of NaCl (0-600) mM and Ca2(PO3)4 0-100 mM, as well as of buffer solutions. The emul-sions were mixed for 15 min at 8000 rpm and stored at 25 °C overnight before the analysis [19].

Stability to heating

Emulsions with pH 7.0 and the concentration of salt that causes the minimum size distribution were placed in a water bath at a fixed temperature (from 30 to 90 °C). Then, the emulsion samples were stored at ambient temperature for 24 h prior to the analysis [21].

Measurement of particle size

The mean particle size and size distribution of the samples were measured by dynamic light scat-tering (Nano-ZS, Malvern instrument, UK). Samples were diluted 1:100 by sodium citrate buffer at the same pH and ionic composition before the measure-ment of particle size. Mean droplet size was reported as Z-average diameter. The ζ-potential of droplets was determined by using a particle electrophoresis instrument (ZEN3600, Nano-series, Zetasizer, Mal-vern instrument, UK).

Creaming index

An amount of 25 ml of samples was placed in a tube and stored at 25 °C for 7 days after the prepar-ation. The boundary between the two phases could be easily seen: phase one was oil-rich, while phase two was turbid and in a larger amount at the bottom of the tube. The height of the oil phase (HOP) may be used to calculate the creaming index (CI) as the per-cent ratio of HOP to the height of total emulsion.

Creaming index = 100×(the height of oil phase)/ /(height of total emulsion) (1)

Vitamin D analysis

At first, solutions of 30, 60, 90 and 120 µg of vitamin D in ethanol were prepared as standard. The standard curve was obtained by regression analysis (y = 0.0121x + 0.0432) with R2 = 0.997. Then, 10 ml


160

of the vitamin D emulsion samples and 0.5 ml of aqueous potassium hydroxide (KOH, 60%) aliquots were poured into test tubes. The tubes were sealed and vortexed by Phonix, Araraquara (Sao Paulo, Bra-zil, Ap560). Then, the tubes were placed in a water bath for 30 min and shaken for 5 min. 30 min after shaking, the tubes with the samples were cooled in an ice-water bath and allowed to rest for 10 min. After that, 4 ml ethanol was added to each sample and vor-texed. The tubes with the samples were centrifuged at 2000g (Sigma, USA) and the ethanol was collected and transferred to a UV-160 Shimadzu spectrophoto-meter (Japan), for determining vitamin D concentra-tion at 265 nm. The tests were repeated three times [3].

Spray drying the emulsion

The most common form of vitamins, which is easily used in the food and pharmaceutical industries, is powder. From the different emulsions of vitamin D tested, the stable ones were selected for spray dry-ing. The stable emulsions were spray dried using a mini-spray drier (B-290, Buchi, Switzerland). The inlet temperature was 180 °C and outlet temperature was 89 °C. The emulsion feed rate was 4 ml/min.

SEM analysis

For scanning electron microscopy (SEM) ana-lysis, the samples were sprinkled on double adhesive tape fixed on aluminum stubs and then put into vacuum desiccators to dry out. The dried emulsion (capsules) was coated with gold-palladium using a sputter coater machine model 7620 C for 180 s. The SEM analysis of the capsules was carried out with a Stereoscan S360 model (Leica, Cambridge, UK). The capsules were observed at an accelerating voltage of 20 kV.

All samples were prepared in triplicate and the results have been reported as a means with its stan-dard deviation. The stability of the emulsions was det-

ermined by a particle size analyzer, and by assessing ζ-potential and creaming index.


Emulsifier concentration

In this section, different biopolymers with differ-ent concentrations were investigated to determine the minimum droplet diameter. These samples were pro-duced by 4% oil phase and 96% aqueous phase. Emulsions with the minimum droplet diameter are very stable during storage. When the size of the drop-lets increases gravitational separation occurs and the two oil and aqueous phases are formed [22]. Figure 1 shows that an increasing emulsifier concentration causes the size of the emulsion droplets to decrease, and the droplets are completely covered and satur-ated by the emulsifiers [23]. By increasing the emul-sifier concentration, the droplet size was reduced because there was a sufficient amount of emulsifier for the oil droplets [24]. The minimum concentration and minimum size of particles obtained were as fol-lows: gum arabic 7%, 468 nm, maltodextrin 2%, 266 nm, WPC 0.5% 190 nm, SI 4%, 132 nm. The dec-rease in droplet diameter when increasing the emul-sifier concentration was observed only for gum arabic. There was no gradual increase or decrease when using other emulsifiers. Emulsion samples with mini-mum particle size and minimum concentration were used for further investigations.

Influence of pH on emulsion stability

It is important to assess the effect of pH on the stability of different emulsions when considering applications of the studied emulsions in the food ind-ustry, since products such as soft drinks and dairy drinks have different pH values, acidic and neutral, respectively. We investigated the particle size in emulsion samples with varying pH values: 2 < pH < 8.

Figure 1. The effect of emulsifier concentration on the mean droplet particle size.


161

The average particle size of WPC at low pH (2 < pH < 4) and at high pH (7 < pH < 8) was lower than that reached at pH 5.0, which is the isoelectric point of WPC. The particle size of soy protein isolate was higher around pH 4 compared to the particle size at pH 2, 3 and that at pH 6, 7 and 8. Gum arabic and maltodextrin with particle sizes of 416-466 nm and 250-272 nm, respectively, were constant in the whole pH range 2 < pH < 8. Figure 2a shows that gum ara-bic and maltodextrin are stable at all pH values, because these polysaccharides can fully cover the oil droplets, and as a result the van der Waals forces between droplets decrease [12] and steric repulsion is stronger than electrostatic repulsion. Protein droplets, like soy proteins and WPC, are not stable at their iso-electric point because the electrostatic repulsion between droplets is reduced. At the isoelectric point negative and positive charges are equal while net protein is without charge, so the electrostatic repul-sion is dominated by steric repulsion. Thus, under the effect of attractive Vander Waals forces, droplets aggregate [16].

The influence of pH on droplets’ ζ-potential is shown in Figure 2b. The ζ-potential of WPC droplets was highly positive at low pH and reduced to negative values by increasing the pH, while at pH 4 to pH 5, it became zero. The ζ-potential of SI droplets showed the same behavior as those of WPC: at low pH the droplets were positive, while at high pH they were negative. The dependence of droplet charge on the pH is related to the isoelectric point of proteins [27]. The isoelectric point of WPC and SI is between 4-5 and 5, respectively. At a pH below the isoelectric point, the concentration of H+ is high, and the amino groups of the proteins are positively charged (NH3

+) and carboxyl groups are neutral (COOH), so the net protein charge is positive. At high pH, the concen-tration of H+ is low, the carboxyl groups are negative (COO-) and the amino groups are neutral, so the net charge of protein is negative. At the pI of protein, the positive and negative charges are balanced and the total protein net charge is zero [28]. The interaction between the droplets of the emulsifier is electrostatic repulsion instead of steric repulsion. Therefore, at a

Figure 2. a) Effect of pH on the mean droplet particle size of gum arabic, maltodextrin, whey protein concentrated, soy protein isolate

emulsions (0 mM salt). b) Influence of pH on the droplet charge (ζ-potential) of gum arabic, maltodextrin, whey protein concentrated, soy protein isolate emulsions.


162

pH around the pI, the droplets have a high tendency to aggregate because of the van der Waals attraction between droplets and as a consequence the emulsion is not stable [29,30]. The charge of GA and MD drop-lets is negative at all pH values, which demonstrates the presence of negative groups (COO-) on these polysaccharides. The negative charge of GA is higher than that of MD, and it may be the linear charge den-sity of GA, which is higher than that of MD [14,18]. The negative charge of droplets can cause transfer of metals, which increase the oxidation of lipids. The negative droplets of emulsion can absorb the positive metals in food emulsion and promote the lipid oxid-ation. The reduction of pH below 5 causes the nega-tive charge of the droplets to decrease because it is below the pKa values of carboxyl groups and they lose their charges [31].

Influence of ionic strength on emulsion stability

Many solutes in food emulsions are ionic and capable of being ionized, which depends on the nat-

ure of the food and emulsions, and the oil which is emulsified [32]. Salt is one of the important factors of coalescence, affecting emulsion stability. We exam-ined the effect of ionic strength using two salts: NaCl (0-600 mM), used to achieve better absorption of vit-amin D, and Ca3(PO4)2 (0-70 mM), used to enrich the emulsion with calcium. No significant changes in par-ticle size were observed for maltodextrin and gum arabic with increasing NaCl and Ca3(PO4)2 concentra-tion. This consistency may be related to the steric repulsion between the droplets of GA and MD [33], and to the absence of electrostatic repulsion.

Both whey protein and soy protein isolate became unstable with the addition of the salts. Figure 3a and b reveal that the particle size of proteins increased remarkably (WPC at NaCl > 200 mM and SI at NaCl > 100 mM and WPC at Ca3(PO4)2 > 20 mM and SI at Ca3(PO4)2 > 10 mM). The droplets are not stable at higher salt concentrations because of the reduction of electrostatic repulsion among the drop-lets. The ions of the salts can screen the electrostatic

Figure 3. a) The effect of NaCl concentration on the mean droplet particle size of gum arabic, maltodextrin, whey protein concentrated,

soy protein isolate emulsions (pH 7.0). b) Effect of Ca3(PO4)2 concentration on the mean droplet particle size of gum arabic, maltodextrin, whey protein concentrated, soy protein isolate emulsions (pH 7.0).


163

interaction and they can make bridges among the charged droplets [22]. The electrostatic repulsion is not strong enough to overcome the attractive van der Waals interaction [30]. These protein emulsifiers are not stable over time, thus gravitational separation occurs and the creaming layer is observed at the top of the emulsions. The Ca3(PO4)2 has multivalent ions. Its screening and binding effect is greater than that of NaCl, and it can destabilize the emulsion in a lower concentration than that necessary for NaCl [32]. Fig-ures 4a and 4b indicate that the negative charge of WPC and SI droplets decreased when increasing the salt concentration [34].

These results may be attributed to electrostatic screening and ion binding of positive charge of Na+ and Ca2+ on negatively charged groups (COO-) of pro-teins by electrostatic attraction screening of the net charge of proteins. The surface charge of multivalent ions of Ca2+ is higher than that of Na+ and can screen the proteins. Net charge and ion binding would dec-rease more the ζ-potential. The ζ- potential of GA and MD underwent little changes when increasing the salt concentration, which would be the effect of charge compensation in the interfacial structure or composi-tion with increasing ionic strength [35].

Effect of thermal treatment on emulsion stability

The application of an emulsion at different ranges of temperature, as well as thermal processing, such as sterilization, pasteurization and cooking, causes emulsion degradation [36]. Therefore, we investigated the influence of temperature (between 30 and 90 °C) on the particle size and creaming stability of GA, WPC, SI and MD. The other experimental con-ditions were as follows: time – 15 min, salt concen-tration – between 0 and 100 mM of NaCl, pH 7.0. Figure 5a illustrates a significant effect of temperature on the particle size of the emulsion in the absence of salts. As shown in Figure 5b, when increasing the temperature and salt concentration to 100 mM, the GA and MD emulsions were stable and there were no remarkable changes in emulsion particle size. How-ever, the WPC emulsion was unstable at 70 °C and its particle size increased considerably. Similarly, the SI emulsion was not stable at 80 °C and its particles increased in size. It has been reported that emulsions stabilized by protein are unstable when exposed to thermal processing, because these proteins unfold the oil droplets when temperatures exceed critical values, and their reactive groups like non-polar or

Figure 4. a) The effect of NaCl concentration on the (ζ-potential) of gum arabic, maltodextrin, whey protein concentrated, soy protein

isolate emulsions (pH 7.0). b) Effect of Ca3(PO4)2 concentration on the (ζ-potential) of gum arabic, maltodextrin, whey protein concentrated, soy protein isolate emulsions (pH 7.0).


164

sulfhydryl groups are located in their intersurface [36], which increase the attractive interaction among pro-teins adsorbing each other or different droplets [19]. By adding the salts, the electrostatic repulsion is not strong enough to overcome the hydrophobic inter-action and van der Waals attraction, which leads to an increase in droplet flocculation [25,37,38]. Heating the soy protein isolate emulsion in the presence of salts shows an unstable emulsion and leads to dis-ruption of the quaternary structure of proteins and dis-sociation of the subunits. This dissociation causes proteins to unfold, and increase the hydrophobicity of surfaces and interactions between polypeptide chains [39,40]. GA and MD are polysaccharides, which do not unfold at high temperatures.

Stability of emulsions under environmental conditions

The emulsions with the least particle size were prepared and stored in test tubes at 25 °C (environ-

mental condition) for 30 days, and every 7 days the samples were assessed to determine the amount of vitamin D available. The results (see Figure 6) show that the soy protein emulsion offered the best pro-tection, preventing vitamin D oxidation and degrad-ation. The final amount of vitamin D found in each emulsion was as follows: 85, 74, 69.5 and 70% for SI, WPC, MD and GA, respectively. The mechanism by which soy protein offers protection to vitamin D con-sists in the following: the vitamin cannot be easily mobile due to the binding proteins and the oxidizing agent has restricted accessibility to vitamin D due to the protein barrier [40]. Another important factor is the decrease of lipid oxidation by the reduction of emul-sion particle size, which should limit the issues of oxidation. Thus, the minimum amount of vitamin D would be oxidized in smaller droplets [41].

Figure 5. a) Effect of temperature on the mean droplet particle size of gum arabic, maltodextrin, whey protein concentrated, soy protein

isolate emulsions (0 mM salt, pH 7.0). b) Effect of temperature on the mean droplet particle size of gum arabic, maltodextrin, whey protein concentrated, soy protein isolate emulsions (100 mM salt, pH 7.0).


165

SEM analysis of spray dried emulsions

Most of the time emulsions are used in powder form in the food industry. We carried out the SEM analysis of the SI emulsion, which proved the best ability to protect vitamin D from the series of emul-sifiers investigated, by first spray drying it. Figure 7 shows the particle size distribution of the nanocap-

sules, which were of 104 nm and had a polydispersity index of 0.4. SEM images of spray dried nanocap-sules in Figure 8 indicate that the droplets of the nanocapsules had a round shape and were dispersed homogeneously. In addition, no cracks could be obs-erved on the surface of the nanocapsules.

Figure 6. Comparison of different emulsifiers on preventing of Vitamin D.

Figure 7. Particle size distribution of spray dried nanocapsules of vitamin D emulsified by soy protein isolate.

Figure 8. SEM analysis of spray dried soy protein emulsion.


166

CONCLUSION

This study has investigated the effect of different biopolymers, such as two polysaccharides (gum ara-bic and maltodextrin) and two proteins (35% whey protein concentrate and soy protein isolate) on the formation of emulsions with the objective to provide protection to vitamin D against degradation. The mini-mum concentration of emulsifier necessary to achieve the minimum droplet size was found to be the follow-ing: gum arabic 7%, 468 nm; maltodextrin 2%, 266 nm; WPC 0.5% 190 nm; and SI 4%, 132 nm. Among these emulsifiers, the SI emulsion demonstrated the best ability to produce small droplets at low emulsifier concentration, and the emulsions were more stable under environmental conditions, as well as under vari-able emulsion conditions. The SI and WPC emulsions exhibited extensive aggregation and destabilization around their isoelectric point (WPC 4 < pH < 5, SI pH 5) at high salt concentration (WPC at NaCl >200 mM and SI at NaCl >100 mM and WPC at Ca3(PO4)2 >20 mM and SI at Ca3(PO4)2 >10 mM). At high tempera-tures (WPC >70 °C, pH 7, 100 mM, SI >80 °C, pH 7, 100 mM NaCl), the results demonstrated changes in electrostatic and hydrophobic interactions and predo-minant van der Waals attraction among droplets. The results achieved in the present study led to the con-clusion that the gum arabic, maltodextrin and WPC emulsions presented inadequate particle nanosize, while the SI emulsion could protect vitamin D against degradation better than the others during one month.

Acknowledgements

The authors acknowledged the Research Vice Presidency of Semnan University and Babol Ab food industry for providing necessary facilities to perform this research.

REFERENCES

[1] B. Hollis, J. Kamerud, S. Selvaag, J. Lorenz, J. Napoli, Clin. Chem. 39 (1993) 529-533

[2] Z. Teng, Y. Luo, Q. Wang, Food Chem. 141 (2013) 524- –532

[3] S. Tamnak, H. Mirhosseini, C.P. Tan, B.T. Amid, M. Kazemi, S. Hedayatnia, Food Hydrocolloids 61 (2016) 599-608

[4] N. Khalid, I. Kobayashi, M.A. Neves, K. Uemura, M. Nakajima, H. Nabetani, Eur. J. Lipid Sci. Technol. (2016) 899-905

[5] S.A. Kazmi, R. Vieth, D. Rousseau, Int. Dairy J. 17 (2007) 753-759

[6] M. Guttoff, A.H. Saberi, D.J. McClements, Food Chem. 171 (2015) 117-122

[7] Y. Lin, Y.-H. Wang, X.-Q. Yang, J. Guo, J.-M. Wang, LWT-Food Sci. Technol. 72 (2016) 510-517

[8] J.N. Pedersen, H.S. Frislev, J.S. Pedersen, D.E. Otzen, J. Dairy Sci. 99 (2016) 7755-7767

[9] K. Ziani, Y. Fang, D.J. McClements, Food Chem. 134 (2012) 1106-1112

[10] M. Haham, S. Ish-Shalom, M. Nodelman, I. Duek, E. Segal, M. Kustanovich, Y.D. Livney, Food Funct. 3 (2012) 737-744

[11] C. Qian, E.A. Decker, H. Xiao, D.J. McClements, J. Am. Oil Chem. Soc. 88 (2010) 47-55

[12] G. Davidov-Pardo, C.E. Gumus, D.J. McClements, Food Chem. 196 (2016) 821-827

[13] H. Singh,A. Sarkar, Adv. Colloid Interface Sci. 165 (2011) 47-57

[14] D. McClements, E. Decker, J. Food Sci. 65 (2000) 1270- –1282

[15] D.J. McClements, Soft Matter. 7 (2011) 2297-2316

[16] F.E. Mitidieri,J.R. Wagner, Food Res. Int. 35 (2002) 547- –557

[17] I. Kralova,J. Sjöblom, J. Dispersion Sci. Technol. 30 (2009) 1363-1383

[18] M. Hu, D.J. McClements, E.A. Decker, J. Agric. Food. Chem. 51 (2003) 1435-1439

[19] A. Sarkar, H. Kamaruddin, A. Bentley, S. Wang, Food Hydrocolloids 57 (2016) 160-168

[20] A. Sarkar, K.K. Goh, H. Singh, Food Hydrocolloids 24 (2010) 534-541

[21] J.-M. Wang, N. Xia, X.-Q. Yang, S.-W. Yin, J.-R. Qi, X.-T. He, D.-B. Yuan, L.-J. Wang, J. Agric. Food. Chem. 60 (2012) 3302-3310

[22] J. Surh, Y.S. Gu, E.A. Decker, D.J. McClements, J. Agric. Food. Chem. 53 (2005) 4236-4244

[23] L.B. Petrovic, V.J. Sovilj, J.M. Katona, J.L. Milanovic, J. Colloid Interface Sci. 342 (2010) 333-339

[24] M. Klein, A. Aserin, I. Svitov, N. Garti, Colloids Surfaces, B: Biointerfaces 77 (2010) 75-81

[25] Z.M. Nazareth, Int J Biol Macromol. 62 (2009) 642-650

[26] J. Surh, E.A. Decker, D.J. McClements, Food Hydrocol-loids 20 (2006) 607-618

[27] R. Chanamai, D. McClements, J. Food Sci. 67 (2002) 120-125

[28] A. Kulmyrzaev, R. Chanamai, D.J. McClements, Food Res. Int. 33 (2000) 15-20

[29] K. Demetriades, J. Coupland, D. McClements, J. Food Sci. 62 (1997) 342-347

[30] J. Ventureira, E. Martínez, M. Añón, Food Hydrocolloids 24 (2010) 551-559

[31] J. Acedo-Carrillo, A. Rosas-Durazo, R. Herrera-Urbina, M. Rinaudo, F. Goycoolea, M. Valdez, Carbohydr. Polym. 65 (2006) 327-336

[32] Y.J. Yuan, O.D. Velev, K. Chen, B.E. Campbell, E.W. Kaler, A.M. Lenhoff, J. Agric. Food. Chem. 50 (2002) 4953-4958

[33] C. Qian, E.A. Decker, H. Xiao, D.J. McClements, Food Chem. 132 (2012) 1221-1229


167

[34] N. Tangsuphoom, J. Coupland, J. Food Sci. 73 (2008) E274-E280

[35] J.N. Israelachvili, Intermolecular and surface forces, third ed., Academic press, Oxford, 2011, pp. 71-80

[36] H. Singh, Food Hydrocolloids 25 (2011) 1938-1944

[37] M. Keerati-u-rai,M. Corredig, Food Hydrocolloids 23 (2009) 2141-2148

[38] B. Yin, W. Deng, K. Xu, L. Huang, P. Yao, J. Colloid Inter-face Sci. 380 (2012) 51-59

[39] M.G. Darewicz, J. Dziuba, M. Dziuba, Pol. J. Food Nutr. Sci. 15 (2006) 79

[40] N.-N. Wu, X. Huang, X.-Q. Yang, J. Guo, E.-L. Zheng, S.-W. Yin, J.-H. Zhu, J.-R. Qi, X.-T. He, J.-B. Zhang, Food Hydrocolloids 28 (2012) 110-120

[41] T. Tokle, D.J. McClements, Food Hydrocolloids 25 (2011) 976-982.

FAEZEH SHARIFI MANSOUR JAHANGIRI

Faculty of Chemical, Petroleum and Gas Eng., Semnan University,

Semnan, I.R. Iran

NAUČNI RAD

ISPITIVANJE STABILNOSTI VITAMINA D U SISTEMIMA ZA ISPORUČIVANJE NA BAZI EMULZIJE

Vitamin D je nutraceutski agens koji je neophodan za zdravlje ljudi. Međutim, količina ovog vitamina u većini namirnica je nedovoljna, pa su mnogi proizvođači razvili proizvode obogaćene vitaminima. Vitamin D je osetljiv na izloženost kiseoniku i visokoj temperaturi. Za zaštitu od degradacije tokom prerade hrane, poželjna je isporuka na bazi emulzije. Stabilnija emulzija dovodi do veće zaštite vitamina D. U ovom radu su istraživani je efekti različitih faktora, kao što su: izbor biopolimera, pH, jonska jačina i temperatura, na sta-bilnost emulzije. Pošto su emulzije sa manjim česticama stabilnije, utvrđene su minimalne koncentracije biopolimera koje omogućavaju minimalnu veličinu čestica. Dobijeni rezultati su sledeći: guma arabika 7% (468 nm), maltodekstrin 2% (266 nm), izolat proteina surutke 0,5% (190 nm) i izolat sojinih proteina 4% (132 nm). Među različitim biopolimerima i ispitanim uslovima emulzije, emulzija izolata proteina soje je obezbedila najveću zaštitu vitamina D (85%) pri koncentraciji od 4%, pH 7 i 25 °C. SEM analiza osušenih nano-kapsula emulzije izolata proteina soje je otkrila homogenu i jedinstvenu disperziju čestica.

Ključne reči: vitamin D, emulzija, stabilnost, izolat sojinih proteina, nanokapsule.




169

SHILIN HUANG

JUAN LI CHANG-FENG YAN

ZHIDA WANG CHANGQING GUO

YAN SHI

Key Laboratory of Renewable Energy, Chinese Academy of

Sciences; Guangdong Key Laboratory of New and Renewable

Energy Research and Development, Guangzhou Institute

of Energy Conversion, Chinese Academy of Sciences, No.2,

Wushan, Tianhe District, Guangzhou, China

SCIENTIFIC PAPER

UDC 546.11:544.4:66

SYNTHESIS AND CHARACTERIZATION OF Cu-X/γ-Al2O3 CATALYST BY INTERMITTENT MICROWAVE IRRADIATION FOR HYDROGEN GENERATION FROM DIMETHYL ETHER STEAM REFORMING

Article Highlights • 2Cu-Fe/72γ-Al2O3 performs excellent for DME steam reforming • The intermittent microwave irradiation provides a rapid method for Cu-Fe/γ-Al2O3

synthesis • Adjusting micro-irradiation on and off time can effectively control the crystallite size of

CuO • Ferric oxide particles are active in water-gas shift reaction, which lower CO

concentration Abstract

A series of Cu-X/γ-Al2O3 (X = Fe, Co, Ni) catalysts were synthesized by a rapid intermittent microwave irradiation method for hydrogen generation from dimethyl ether steam reforming. Different parameters, such as the promoters of X (X = Fe, Co, Ni), microwave irradiation procedure and the ratio of metal to γ-Al2O3, were investigated. The results show that 2Cu-Fe/72γ-Al2O3 has the best performance, for which the agglomeration is prevented, CuO is well dispersed and the catalytic activity is improved. Promoter iron oxide in 2Cu-Fe/9γ-Al2O3 facilitates the water-gas shift reaction, which lead to an increase in the conversion of CO to CO2 and hydrogen yield. Particularly, the 2Cu-Fe/72γ-Al2O3 catalyst, with the best molar ratio of metal to γ-Al2O3, shows a dimethyl ether conversion of >99% and a hydrogen yield of >98% and produces the lowest CO content of 1.4%, indicating that the synergism between dimethyl ether hydrolysis and methanol reforming requires an appropriate balance between the metallic Cu-Fe and the acidγ-Al2O3. The intermittent microwave irradiation technique provides a simple but effective method of the Cu-Fe/γ-Al2O3 synthesis with a good catalyst perform-ance for the dimethyl ether steam reforming.

Keywords: hydrogen generation; dimethyl ether; intermittent microwave; catalytic steam reforming.

Polymer electrolyte membrane fuel cell (PEMFC) vehicles fueled by hydrogen are one of the most pro-mising candidates for the internal combustion engine automobiles due to their high efficiency and environ-mentally friendly properties [1]. Dimethyl ether (DME) Correspondence: C.-F. Yan, Guangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangdong, 510640, China. E-mail: [email protected] Paper received: 2 March, 2016 Paper revised: 9 August, 2016 Paper accepted: 28 August, 2017

https://doi.org/10.2298/CICEQ160203029H

is an ideal liquid fuel as hydrogen carrier for its high H2 content (13.0 wt.% compared to 12.5 wt.% of methanol), easy storage and simple transportation [2]. Besides, DME is inert, non-carcinogenic, non-muta-genic, non-corrosive and virtually non-toxic. It can also be conveniently handled because its physical properties are very similar to liquid petroleum gas (LPG). However, DME now faces overcapacity in China with an annual production of 13 million metric tons per year. Over 90% of DME is used as a blend stock for LPG which is primarily used for combustion in cooking and irradiation [3]. Steam reforming of

S. HUANG et al.: SYNTHESIS AND CHARACTERIZATION OF Cu-X/γ-Al2O3… Chem. Ind. Chem. Eng. Q. 24 (2) 169−178 (2018)

170

DME is an excellent process to produce hydrogen and an attractive route to provide hydrogen as fuel cells on a small or medium scale. An integrated sys-tem of DME steam reforming (Eq. (1)) for hydrogen generation usually involves two consecutive pro-cesses: the DME hydrolysis (Eq. (2)) and the MeOH steam reforming (Eq. (3)). Besides, the reverse water- -gas shift reaction (Eq. (4)) takes place. Generally, a solid acid catalyst such as γ-Al2O3, HZSM-5, zeolites, Ga2O3 or ZrO2 is needed to catalyze the hydrolysis of DME, whereas Cu-, Pt-, Ru-, Pd-, and Ni-based catalysts are used for MeOH steam reforming [4-9]. Consequently, suitable metallic and acid components are required to synthesize the bi-function catalyst for high DME conversion, high H2 selectivity and high stability.

DME steam reforming:

3 3(g) 2 (g) 2 2

O -1298

CH OCH 3H O 2CO 6H ,

122 kJ molKH

+ = +

Δ = (1)

DME hydrolysis:

3 3(g) 2 (g) 3 (g)

O -1298 K

CH OCH H O 2CH OH ,

24 kJ molH

+ =

Δ = (2)

MeOH steam reforming:

O -13 (g) 2 (g) 2 2 298 KCH OH H O CO 3H , 49 kJ molH+ = + = (3)

Reverse water-gas shift reaction:

O -12 2 2 (g) 298 KCO H H O CO, 41 kJ molH+ = + Δ = (4)

DME steam reforming bi-functional catalysts can be prepared by several methods including mechanical mixing [10,11], deposition-precipitation [12] and sol-gel [13]. The mechanical mixing method has been applied in CuZnAlX/HZSM-5 (X = Cr, Zr, Co or Ce) for hydrogen production from steam reforming of DME in our previous work [10,11]. The microwave technology has been used to prepare catalyst material, which reduced the preparation time and optimized the crys-tallization [14,15]. Li et al. used microwave irradiation method in the aged process that give great benefits to both activity and stability of Cu/ZnO/Al2O3 catalyst [16]. In intermittent microwave irradiation (IMI) tech-nique, the irradiation on and off, are alternatively per-formed at controllable temperatures. Compared with the conventional and continuous microwave heating technique, IMI is much easier to control the heating temperature for crystallization and provide relaxation time to protect the particle growth. Here we propose a rapid synthesis method based on the IMI technique for preparing the metal based catalyst Cu-X (X = Fe,

Co, Ni) supported on the most commonly used acid function γ-Al2O3 as the acid catalyst for DME steam reforming of hydrogen generation. To improve the catalytic performance, effects of irradiation procedure, ratio of metal to γ-Al2O3 and experimental condition were investigated in details.

EXPERIMENTAL

Catalyst preparation

2Cu-X/9γ-Al2O3 (X = Fe, Co, Ni) with a Cu: X: γ-Al2O3, mole ratio of 2:1:9, was prepared by the imp-regnation method: an aqueous precursor containing Cu(NO3)2, X(NO)n (X = Fe, Co, Ni) and γ-Al2O3 pow-der was made at room temperature and then was pro-cessed in the microwave (FT-2KW, Guangzhou Futao microwave equipment co. ltd at 2 kW). IMI was in the pulse form of 5 s-on and 5 s-off for 20 times (2Cu-Fe/9γ-Al2O3, 5 s). The catalyst was dried at 110 °C for 12 h and calcined at 400 °C for 5 h.

Effects of the microwave irradiation procedure were investigated by the different irradiation times with a total irradiation time of 100 s, irradiation and relaxation time remain the same ratio of 1:1. 2Cu-Fe/ /9γ-Al2O3 (0 s) meant that no microwave irradiation was applied. 2Cu-Fe/9γ-Al2O3 (20 s) was heated by intermittent microwave in the pulse form of 20 s-on and 20 s-off for 5 times. 2Cu-Fe/9γ-Al2O3 (50 s) was heated in the pulse form of 50 s-on and 50 s-off for 2 times.

Catalyst characterization

X-ray diffraction (XRD) analysis was performed in the PANalytical X’Pert diffractometer (X’Pert PRO MPD, PW3040/60) within the 2-θ ranged from 20 to 80° by a speed of 2° per min with Cu-Kα (λ = = 0.154060 nm) radiation (40 kV, 40 mA). The crys-tallite size of CuO and Fe2O3 was calculated by the Sherrer equation on the basis of the data of the line broadening at half the maximum intensity (full width at half-maximum, FWHM) and the Bragg angle (θ). The Brunauer-Emmett-Teller (BET) specific surface was determined from the adsorption and desorption iso-therms of nitrogen at -196 °C after outgassing pro-cedure under vacuum at 250 °C for 20 h, using a Quantachrome SI-MP-10/PoreMaster 33. The catalyst was characterized by the Scanning electron micro-scope (SEM) with a S-4800 instrument at 2.0 kW. Hydrogen temperature-programmed reduction (H2-TPR) was carried out to estimate the reduction per-formance of the catalyst by Quantachrome ASIQACIV200-2. H2-TPR was conducted in the fol-lowing method: 100 mg of a powder sample was


171

heated at a rate of 10 °C/min up to 120 °C in helium atmosphere and kept for 1 h. The sample was cooled down to 40 °C in the He atmosphere, and then followed by a pure hydrogen flow for 1 h. Sub-sequently, H2-TPR was initiated by a heating rate of 10 °C/min up to 750 °C. The consumption of hyd-rogen was determined by a Thermal conductivity det-ector (TCD) and recorded by an online computer.

Catalytic reaction and analyses of the products

DME steam reforming was executed in a bench-scale fixed-bed reactor (a quartz tube with inner dia-meter of 20 mm) under ambient pressure. The fixed-bed reactor consists of a gas supplier, a catalyst reaction part, an analyzer, and a controller, as shown in Figure 1. The catalyst was placed in the quartz tube and was heated by an electric furnace equipped with a temperature controller. The catalyst (m = 2 g) was reduced in 5 vol.% H2 in N2, heating rate of 10 °C min-1 from room temperature up to 400 °C for 4 h, then the system was balance with N2 for 1 h at the same tem-perature. Standard reaction conditions for DME steam reforming were steam to DME molar ratio of 5 to 1 at a space velocity of 3600 ml/(gcat h). The system was run for 1 h then the product gas was cooled and dried before analyzing by the chromatograph (WUFENG GC522) equipped with a TCD detector and a Flame ionization detector (FID) detector. The steam reform-ing reaction performance over Cu-X/γ-Al2O3 catalysts was evaluated by conversion of the CO2 and selec-tivity CO, CH4. The representative values are given as follows:

DME,in DME,outDME

DME,in

100F F

XF

−= (5)

,out

DME,in DME,out

1002( )

XX

FS

F F=

− (6)

22

HH

DME,in

100

6

FY

F= (7)

where XDME is DME conversion, SX (X = CH4, CO) are CH4 selective and CO selectivity, FDME,,in and FDME,out are the inlet and outlet molar flow rates of DME, respectively. FH2 are the molar flow rate of H2 in the gas out reactant, FX,out (X = CH4, CO) are the molar flow rate of CH4 or CO in the gas out reactant.

RESULTS

Effect of promoter on steam reforming of DME

Figure 2a shows the XRD patterns of 2Cu-X/9γ- -Al2O3 (X = Fe, Co, Ni) and γ-Al2O3. The peaks of γ-Al2O3 and CuO can be seen for all the 2Cu-X/9γ- -Al2O3 (X = Fe, Co, Ni) catalysts. The strong peaks assigned to CuO appears at 2θ 35.6 and 38.5°. The X-ray phase analysis results for the 2Cu-Fe/9γ-Al2O3 (Figure 2a, e) shows the composition of the sample includes the phase CuO, Fe2O3 and γ-Al2O3. A broad-ened low-intensity peak in the range of 2θ 33° indicate the presence of a hematite phase Fe2O3 (2θ = 33.18°). The low-intensity of reflection from crystallized ferric oxide phase allows us to assume that in 2Cu-Fe/9γ- -Al2O3, ferric compounds are found in an X-ray amor-phous or highly dispersed state [17]. The peaks of Co3O4 crystals can be found in 2Cu-Co/9γ-Al2O3 (Fiure 2a, d)). There is no peak of NiO, which may be existed in amorphous or microcrystal state.

Figure 2b shows the results of the H2-TPR of the catalysts. Cu species are the main center of the acti-vity for methanol steam reforming. It was reported that usually there are two CuO species in copper based catalysts, the highly dispersed CuO reacts at lower reduction temperature (peak α) and the bulk CuO species reacts at higher temperature (peak β) [18]. This 2Cu-X/9γ-Al2O3 (X = Fe, Co, Ni) series show that introducing a second metal into the copper

Figure 1. Schematic diagram of DME catalytic reaction apparatus. (MFC: mass flow controller, GC: gas chromatograph).


172

Figure 2. a) XRD patterns and b) H2-TPR profiles of the Cu- based catalysts (a: γ-Al2O3, b: 2Cu/9γ-Al2O3, c: 2Cu-Ni/9γ-Al2O3,

d: 2Cu-Co/9γ-Al2O3, e: 2Cu-Fe/9γ-Al2O3); c) XRD patterns and d) H2-TPR profiles of 2Cu-Fe/9γ-Al2O3 catalysts by different microwave irradiation time. a : 2Cu-Fe/9γ-Al2O3 (0 s), b : 2Cu-Fe/9γ-Al2O3 (5 s), c : 2Cu-Fe/9γ-Al2O3 (20 s), d : 2Cu-Fe/9γ-Al2O3 (50 s); e) XRD

patterns and f) H2-TPR profiles of the catalysts derived from the Cu-Fe/γ-Al2O3 by different loading of 2Cu-Fe andγ-Al2O3 (a: γ-Al2O3, b: 2Cu-Fe/72γ-Al2O3, c: 2Cu-Fe/36γ-Al2O3, d: 2Cu-Fe/18γ-Al2O3, e: 2Cu-Fe/9γ-Al2O3. g) SEM images of 2Cu-Fe/9γ-Al2O3 (5 s) and

h) 2Cu-Fe/9γ-Al2O3 (20 s).


173

based catalyst change the copper reduction property. The change of the reduction temperature for bicom-ponent catalysts can be explained by the several pos-sible factors: the dispersity of the copper oxide in the presence of the second metal, formation of mixed oxides (leading to changing the metal-oxygen bonds). In the presence of the promoter Fe, Co and Ni, the reduction of dispersed CuO species all shifts to higher temperature (Table 1).

2Cu-X/9γ-Al2O3 (X = Fe, Co, Ni) was tested in steam reforming of DME at the condition of H2O to DME of 5:1, temperature of 400 °C and space velocity of 3600 ml g-1 h-1 in Figure 3. All the catalysts show the DME conversions of >99%, while the hydrogen yield differ from each other. 2Cu/9γ-Al2O3 achieved hydrogen yield of 93%, 2Cu-Ni/9γ-Al2O3 and 2Cu- -Co/9γ-Al2O3 obtained 69 and 86%. The results obtained are in agreement with the H2-TPR profiles, according to which for 2Cu-Ni/9γ-Al2O3, 2Cu-Co/9γ- -Al2O3 and 2Cu/9γ-Al2O3, we observe an increase in the CuO reduction temperature. Usually the higher reduction temperature presents lower reducibility and catalytic activity [11], whereas the 2Cu-Fe/9γ-Al2O3 has the highest activity of H2 yield 97% whose reduction temperature is slightly higher than 2Cu/9γ- -Al2O3. It may be due to the presence of ferric oxides within the composition of the catalyst, which are active catalysts for the water-gas shift reaction. As a result, it leads to an increase in the conversion of CO to CO2 and an increase in the hydrogen yield. The effect of ferric additives on the water-gas shift react-ion has also been noted for Fe-Cu/ZrO2 [17], Ni- -Fe/La2O2CO3 [19] and Fe/Ca-Al2O3 [20]. The addition of ferric promotes leading to the better catalytic per-formance and the lowest CO selectivity of 1.4% in the case of 2Cu-Fe/9γ-Al2O3. Therefore, the following study will focus on ferric as the promoter.

Figure 3. Performance of DME steam reforming over

2Cu-X/9γ-Al2O3 (X = Fe, Co, Ni) catalysts: mcat = 2 g, Tempera-ture = 400 °C, H2O/DME = 5/1, space velocity = 3600 ml g-1 h-1.

Effect of microwave irradiation procedure on steam reforming of DME

Effects of the IMI procedure on 2Cu-Fe/9γ-Al2O3 were studied using different irradiation procedure with a total irradiation time of 100 s as described in the catalysts preparation part. A typical XRD pattern of CuO and Fe2O3 crystal samples is shown in Figure 2c, the major peaks located at 2θ values of 30-70° correspond to the characteristic diffraction of the monoclinic phase of CuO and Fe2O3. No other peak is observed for the impurities such as Cu2O, indicating the obtained products have a high purity. The micro-wave irradiation time affects little on the BET and pore size distribution, but the average crystallite sizes of Cu and Fe2O3 increase from 41.8 and 20.9 nm for 2Cu-Fe/9γ-Al2O3 (5 s) to 52.4 and 31.2 nm for 2Cu- -Fe/9γ-Al2O3 (50 s) (Table 1).

For the 2Cu-Fe/9γ-Al2O3 synthesized using 5 s intermittent irradiation, the average crystallite size of CuO and Fe2O3 were 41.8 and 20.9 nm, respectively, significantly smaller than those synthesized in longer

Table 1. Specific surface area, pore structure and reduction peak temperature and ratio of dispersed CuO to bulk CuO of different catalysts prepared in this work. Chemical composition referred as mole ratio

Catalyst BET (m2 g-1) V / cm3 g-1 d / nm Temperature, °C Dispersed

CuO/Bulk CuO Crystallite size, nm

Dispersed CuO Bulk CuO CuO Fe2O3

γ-Al2O3 256.7 0.50 7.86 - - - - -

2Cu-Fe/9γ-Al2O3 195.7 0.41 8.02 282 319 1.79 41.8 20.9

2Cu-Co/9γ-Al2O3 182.9 0.38 8.40 302 326 1.04 36.4 -

2Cu-Ni/9γ-Al2O3 181.9 0.39 8.47 286 322 0.79 51.1 -

2Cu/9γ-Al2O3 197.7 0.42 8.49 273 323 0.98 62.7 -

2Cu-Fe/9γ-Al2O3 (0 s) 195.6 0.40 8.21 257 319 0.73 42.8 31.5

2Cu-Fe/9γ-Al2O3 (20 s) 193.7 0.38 8.00 259 306 1.04 49 27.8

2Cu-Fe/9γ-Al2O3 (50 s) 192.1 0.39 8.20 258 305 1.04 52.4 31.2

2Cu-Fe/18γ-Al2O3 206.0 0.42 8.00 283 319 2.30 - -

2Cu-Fe/36γ-Al2O3 233.2 0.46 7.96 287 320 2.61 - -

2Cu-Fe/72γ-Al2O3 237.6 0.48 7.95 317 343 4.03 - -


174

intermittent irradiation. The microwave irradiation of short on and off time results in lower reaction tempe-rature of the synthesis reaction system, which limits the nucleation and the grain growth reaction of the CuO and Fe2O3 particles. Shorter irradiation time is required for the dispersion of CuO and Fe2O3 on the γ-Al2O3 supports [21-23]. The results indicate that regulating micro irradiation on and off-time can effect-ively control the crystallite size. From Table 1, 2Cu- -Fe/9γ-Al2O3 (5 s) has the highest dispersed CuO/ /bulk CuO ratio of 1.79. Further increasing the irradi-ation time, the dispersed CuO concentration dec-reases, indicating that the shorter the irradiation time, the higher the ratio of CuO/bulk CuO.

2Cu-Fe/9γ-Al2O3 (5 s) and 2Cu-Fe/9γ-Al2O3 (0 s) obtain over 99% DME conversion in Figure 4. The advantages of optimized IMI technique compared with the conventionally heated one can be seen in the following: 2Cu-Fe/9γ-Al2O3 (5 s) achieves a hydrogen yield of 97%, but the conventionally heated 2Cu- -Fe/9γ-Al2O3 (0 s) achieves 86%. It is noted that the CO concentration of 2Cu-Fe/9γ-Al2O3 (5 s) is 2.3% while of 2Cu-Fe/9γ-Al2O3 (0 s) is 7.2%. Optimized microwave irradiation technique can decrease the CO selectivity of the 2Cu-Fe/9γ-Al2O3 catalyst. Further increasing the irradiation time leads to a lower DME conversion and hydrogen yield, and decreases the CO selectivity. Usually the micro structure such as

BET, pore size distribution and high disperse CuO reduction property affect the catalyst activity. Since the microwave irradiation procedure has little influ-ence on the BET and pore size distribution of the cat-alysts (Table 1), it is the reducibility of CuO that affects the activity of 2Cu-Fe/9γ-Al2O3. 2Cu-Fe/9γ- -Al2O3 (5 s) achieve the >99% DME conversion, 97.5% hydrogen yield and 2.5% CO selectivity. The elevated catalytic property might be due to the high-est ratio of dispersed CuO/bulk CuO of 1.79.

Influence of reaction condition over 2Cu-Fe/9γ-Al2O3

The effect of temperature, steam to DME ratio and space velocity were investigated for the 2Cu- -Fe/9γ-Al2O3 (Figure 5). The runs were conducted in a fixed bed reactor in the 300-400 °C range, as shown in Fig. 5a, where the DME conversion increases with the increase of the reaction temperature. γ-Al2O3 typically shows an excellent DME hydrolysis activity at the temperature above 300 °C due to its moderate acid amount of weak acid strength [24]. The steam reforming reaction is a strong endothermic reaction. Increasing the reaction temperature will facilitate the reaction.

Figure 5b shows the effect of steam to DME ratio on SRD over 2Cu-Fe/9γ-Al2O3. The DME con-version and hydrogen yield increases sharply from 51 and 45% to 95 and 92% as the stoichiometric ratio of steam to DME was 1 to 3. DME conversion and hyd-

Figure 4. Performance of DME steam reforming over 2Cu-Fe/9γ-Al2O3 by different microwave irradiation time: mcat

= 2 g, H2O/DME = 5/1, space velocity = 3600 ml g-1 h-1.


175

rogen yield flatten out as continued increasing the steam to DME ratio. The CO concentration clearly decreases when the steam to DME ratio was raised to 5. The hydrogen yield never reaches 100% for it is consumed by the reverse water-gas shift reaction [25].

Figure 5c shows the influence of space velocity on the catalytic activity. The DME conversion, hyd-rogen yield and CO decrease with the increase of space velocity. An over 99% DME conversion is seen in 3600 ml g-1 h-1 but only 46% in 18000 ml g-1 h-1. The increase of the space velocity leads to an insufficient residence time for the catalytic reaction, which dec-reases the DME conversion and hydrogen yield.

Effect of weight ratio of γ-Al2O3 to 2Cu-Fe on Steam reforming of DME

γ-Al2O3 and copper is used as the active com-ponent of the catalyst for DME hydrolysis and the steam reforming of methanol, respectively. The sel-ection of appropriate ratio of γ-Al2O3 to 2Cu-Fe will significantly improve the performance of the catalyst in the DME reforming reaction.

From the XRD patterns in Figure 2e, 2Cu-Fe/9γ- -Al2O3 (e) clearly shows the characteristic diffraction of the monoclinic phase CuO and Fe2O3. As the ratio of γ-Al2O3 increases, the diffraction from CuO and Fe2O3 becomes so weak, only γ-Al2O3 phase could be detected in 2Cu-Fe/36γ-Al2O3 (c) and 2Cu-Fe/72γ- -Al2O3 (d), indicating that they exist as small crystals

or in amorphous phase [12,26]. The results are also confirmed by the H2-TPR. With the increase of γ-Al2O3 loading, the CuO reduction peak is weakening and shifted to higher temperature. The CuO reduction peak of 2Cu-Fe/72γ-Al2O3 is broader than that of 2Cu- -Fe/36γ-Al2O3, indicating a higher dispersion of CuO in 2Cu-Fe/72γ-Al2O3. Meanwhile, as the Cu-Fe con-centration decreases, CuO reduction peak shifts to higher temperature, which reveals that it enhances the interaction of CuO and γ-Al2O3. As discussed pre-viously in part 3.2, the ratio of dispersed CuO and bulk CuO strongly affects the catalytic performance. 2Cu-Fe/72γ-Al2O3 has the highest ratio of 4.03 pre-senting the best activity and selectivity (Table 1).

The catalytic performance of different weight ratio of γ-Al2O3 to Cu-Fe starts with 2Cu-Fe/9γ-Al2O3 (Figure 6). High γ-Al2O3 content leads to high DME conversion and hydrogen yield at a temperature below 400 °C. When the reaction temperature reaches 400 °C, all the 2Cu-Fe/Yγ-Al2O3 (Y = 9, 18, 36 or 72) catalysts achieve >99% DME conversion and >96% hydrogen yield. Moreover, the 2Cu-Fe/72γ--Al2O3 catalyst presented the least CO selectivity of 1.4%.

The CH4 concentration in the outlet doesn’t change in the temperature range of 300-400 °C (Fig-ure 6d), where usually methanation occurs in alu-mina-rich composite catalysts at above 350 °C [27].

Figure 5. a) Effect of temperature (H2O/DME = 5/1, space velocity = 3600 ml g-1 h-1), b) steam to DME ratio (temperature = 400 °C,

space velocity = 3600 ml g-1 h-1) and c) space velocity (temperature = 400 °C, H2O/DME = 5/1) in DME steam reforming over 2Cu-Fe/9γ-Al2O3.


176

The best performance of 2Cu-Fe/72γ-Al2O3

comes from: 1) its higher γ-Al2O3 content, which is the effective component for DME hydrolysis; 2) the highest ratio of dispersed CuO to Bulk CuO. γ-Al2O3

rich samples possess a large amount of acid sites, which result in high DME conversion; 3) the well dis-persed CuO and Fe2O3 on the γ-Al2O3 phase. Hence, the weight balance between γ-Al2O3 and Cu-Fe should be regulated very carefully.

Figure 6e shows dimethyl ether conversion as a function of reaction time over the 2Cu-Fe/72γ-Al2O3 catalyst. 2Cu-Fe/72γ-Al2O3 had good stability over 50 h and dimethyl ether conversion stayed around 99%. The 2Cu-Fe/72γ-Al2O3 catalyst had been stable at 400 °C, while usually ordinary Cu-supported catalysts were gradually deactivated [27].

CONCLUSIONS

A series of 2Cu-X/γ-Al2O3 (X = Fe, Co, Ni) cat-alysts were synthesized by a rapid intermittent micro-wave irradiation (IMI) technique for hydrogen gener-ation in dimethyl ether (DME) steam reforming. The promoter of ferric oxides within the composition of 2Cu-Fe/γ-Al2O3 are active catalysts for the water-gas shift reaction which leads to an increase in the con-version of CO to CO2 and an increase in the hydrogen yield.

The preparation of 2Cu-Fe/γ-Al2O3 catalysts by IMI technique changes the average crystallite size of Cu and Fe. The optimized IMI procedure is in the pulse form of 5 s-on and 5 s-off for 20 times. The 2Cu-Fe/72γ-Al2O3 (5 s) catalyst provides >99% DME

Figure 6. Performance of DME steam reforming over 2Cu-Fe/Yγ-Al2O3 (Y = 9, 18, 36, 72). mcat

= 2 g, Temperature = 400 °C, H2O/DME = 5/1, Space Velocity = 3600 ml•g-1•h-1, (e) Time course of 2Cu-Fe/72γ-Al2O3 catalytic activity (H2O/DME = 5/1, Space Velocity = 3600

ml•g-1•h-1)


177

conversion, 97% hydrogen yield and 1.4% CO per-cent by volume. It may be due to the higher ratio of dispersed CuO to bulk CuO leads to better activity and selectivity. The optimal condition IMI technique gives great benefits to hydrogen yield and CO and CH4 selectivity.

Acknowledgments

The authors are grateful for the financial support of CAS Renewable Energy Key Lab., Natural Science Foundation of China (51576201), Natural Science Foundation of Guangdong province (2015A030312007, 2015A030313716), Guangdong Science and Technology Project (2013B091300001, 2014A020216030) Guangzhou Science and Techno-logy Project (2013J4500027).

REFERENCES

[1] U. Eberle, M. Felderhoff, F. Schüth, Angew. Chem. Int. Ed. 48 (2009) 6608-6630

[2] T.A. Semelsberger, R.L. Borup, H.L. Greene, J. Power Sources 156 (2006) 497-511

[3] T.H. Fleisch, A. Basu, R.A. Sills, J. Nat. Gas Sci. Eng. 9 (2012) 94-107

[4] V.V. Galvita, G.L. Semin, V.D. Belyaev, T.M. Yurieva, V.A. Sobyanin, Appl. Catal., A 216 (2001) 85-90

[5] N. Shimoda, K. Faungnawakij, R. Kikuchi, K. Eguchi, Int. J. Hydrogen Energy 36 (2011) 1433-1441

[6] T. Kawabata, H. Matsuoka, T. Shishido, D. Li, Y. Tian, T. Sano, K. Takehira, Appl. Catal., A 308 (2006) 82-90

[7] T. Nishiguchi, K. Oka, T. Matsumoto, H. Kanai, K. Utani, S. Imamura, Appl. Catal., A 301 (2006) 66-74

[8] T. Mathew, Y. Yamada, A. Ueda, H. Shioyama, T. Kobayashi, Appl. Catal., A 286 (2005) 11-22

[9] J. Vicente, J. Ereña, L. Oar-Arteta, M. Olazar, J. Bilbao, A.G. Gayubo, Ind. Eng. Chem. Res. 53 (2014) 3462-3471

[10] C. Yan, H. Hai, C. Guo, W. Li, S. Huang, H. Chen, Int. J. Hydrogen Energy 39 (2014) 10409-10416

[11] C. Yan, H. Hai, R. Hu, C. Guo, S. Huang, W. Li, Y. Wen, Int. J. Hydrogen Energy 39 (2014) 18625-18631

[12] X. Wang, X. Pan, R. Lin, S. Kou, W. Zou, J.-X. Ma, Int. J. Hydrogen Energy 35 (2010) 4060-4068

[13] N. Shimoda, K. Faungnawakij, R. Kikuchi, K. Eguchi, Appl. Catal., A 378 (2010) 234-242

[14] P. Reubroycharoen, T. Vitidsant, Y. Liu, G. Yang, N. Tsubaki, Catal. Commun. 8 (2007) 375-378

[15] S. Qi, B. Yang, Catal. Today 98 (2004) 639-645

[16] H.Z. Hui FAN, Zhong LI, Front. Chem. Sci. Eng. 4 (2010) 445-451

[17] I.V. Deinega, L.Y. Dolgykh, I.L. Stolyarchuk, L.A. Staraya, P.E. Strizhak, E.M. Moroz, V.P. Pakharukova, D.A. Zyuzin, Theor. Exp. Chem. 48 (2013) 386-393

[18] C. Ledesma, J. Llorca, J. Phy. Chem., C 115 (2011) 11624-11632

[19] Q. Shi, Z. Peng, W. Chen, N. Zhang, J. Rare Earths 29 (2011) 861-865

[20] L. Chen, C.K.S. Choong, Z. Zhong, L. Huang, T.P. Ang, L. Hong, J. Lin, J. Catal. 276 (2010) 197-200

[21] Z.Q. Tian, S.P. Jiang, Y.M. Liang, P.K. Shen, J. Phy. Chem., B 110 (2006) 5343-5350

[22] N. Lingaiah, P.S. Sai Prasad, P.K. Rao, L.E. Smart, F.J. Berry, Appl. Catal., A 213 (2001) 189-196

[23] J. Ren, S. Liu, Z. Li, X. Lu, K. Xie, Appl. Catal., A 366 (2009) 93-101

[24] K. Faungnawakij, Y. Tanaka, N. Shimoda, T. Fukunaga, S. Kawashima, R. Kikuchi, K. Eguchi, Appl. Catal., A 304 (2006) 40-48

[25] T.A. Semelsberger, K.C. Ott, R.L. Borup, H.L. Greene, Appl. Catal., B 65 (2006) 291-300

[26] J. Okal, M. Zawadzki, Appl. Catal., B 89 (2009) 22-32

[27] Y. Tanaka, R. Kikuchi, T. Takeguchi, K. Eguchi, Appl. Catal., B 57 (2005) 211-222.


178

SHILIN HUANG

JUAN LI CHANG-FENG YAN

ZHIDA WANG CHANGQING GUO

YAN SHI

Key Laboratory of Renewable Energy, Chinese Academy of Sciences;

Guangdong Key Laboratory of New and Renewable Energy Research and Development, Guangzhou Institute of

Energy Conversion, Chinese Academy of Sciences, No.2, Wushan, Tianhe

District, Guangzhou, China

NAUČNI RAD

SINTEZA I KARAKTERIZACIJA KATALIZATORA Cu-X/γ-Al2O3 INTERMITENTNIM MIKROTALASNIM ZRAČENJEM ZA PROIZVODNJU VODONIKA REFORMOVANJEM DIMETIL ETRA VODENOM PAROM

Serija katalizatora Cu-X/γ-Al2O3 (X = Fe, Co, Ni) je sintetizovana metodom brzog inter-mitentnog mikrotalasnog zračenja za generisanje vodonika reformovanjem dimetil-etra vodenom parom. Ispitani su različiti parametri, kao što su: promoteri X (X = Fe, Co, Ni), postupak mikrotalasnog zračenja i odnos metal/Cu-X/γ-Al2O3. Rezultati pokazuju da 2Cu--Fe/72γ-Al2O3 ima najbolje performanse, jer sprečava aglomeraciju, CuO je dobro disper-govan, a katalitička aktivnost je poboljšana. Promotor oksida gvožđa u 2Cu-Fe/9γ-Al2O3 olakšava reakciju vodene pare, što dovodi do povećanja konverzije CO u CO2 i prinos vodonika. Naročito, katalizator 2Cu-Fe/72γ-Al2O3, sa najboljim molskim odnosom metala i γ-Al2O3, omogućava konverziju dimetil etra veću od 99%, prinos vodonika veći od 98% i najniži sadržaj CO od 1,4%, što ukazuje da sinergizam između hidrolize dimetil-etra i refor-movanja metanozahteva odgovarajuću ravnotežu između metalnog dela Cu-Fe i kiseleγ-Al2O3. Tehnika intermitentnog mikrotalasnog zračenja obezbeđuje jednostavnu, ali efi-kasnu metodu sinteze Cu-X/γ-Al2O3 sa dobrim performansama katalizatora za reformiranje dimetil etra vodenom parom.

Ključne reči: generisanje vodonika; dimetil etar; intermitentno mikrotalasno zra-čenje; katalitička reformiranje vodenom paroma.




179

NATNIRIN

PHANTHUMCHINDA1

TANAPAWARIN RAMPAI1

BUDSABATHIP PRASIRTSAK1 SITANAN THITIPRASERT2

SOMBOON TANASUPAWAT3

SUTTICHAI ASSABUMRUNGRAT4

NUTTHA THONGCHUL2 1Program in Biotechnology, Faculty

of Science, Chulalongkorn Uni-versity, Wangmai, Pathumwan,

Bangkok, Thailand 2Research Unit in Bioconversion/

/Bioseparation for Value-Added Chemical Production, Institute of

Biotechnology and Genetic Engineering, Chulalongkorn

University, Wangmai, Pathumwan, Bangkok,Thailand

3Research Unit in Bioconversion/ /Bioseparation for Value-Added

Chemical Production, Department of Biochemistry and Microbiology,

Faculty of Pharmaceutical Sciences, Wangmai, Pathumwan,

Bangkok, Thailand 4Department of Chemical Eng-

ineering, Faculty of Engineering, Chulalongkorn University,

Wangmai, Pathumwan, Bangkok, Thailand

SCIENTIFIC PAPER

UDC 66.067.1:544.076.34:54

ALTERNATIVE REVERSE OSMOSIS TO PURIFY LACTIC ACID FROM A FERMENTATION BROTH

Article Highlights • Reverse osmosis was used to recover and purify lactic acid from fermentation broth • Cation rejection occurred at the first BWRO unit, where lactic acid was separated • Concentration of lactic acid occurred at the second SWRO unit • Sufficiently high lactate purity and recovery was obtained from two-stage RO units • Donnan potential, pH and operating pressure-controlled ion rejection Abstract

Brackish water reverse osmosis (BWRO) and seawater reverse osmosis (SWRO) membranes were used in a two-stage reverse osmosis (RO) unit to recover, pre-purify, and pre-concentrate lactic acid. Calcium lactate, sodium lactate, and ammonium lactate were used as model feed solutions. The oper-ating pressure showed a pronounced effect on lactate passage through the first BWRO unit, and the Donnan exclusion effect and hydrogen bonding were res-ponsible for cation rejection. Calcium ions were rejected at the BWRO unit because of low diffusion rate and charge interaction at the surface. However, monovalent ions formed hydrogen bonds with the carbonyl group of the mem-brane that allowed passage across the membrane. The second SWRO unit was for pre-concentrating lactic acid. A high lactate purity of 99.2% with a total recovery of 50.5% was acquired from calcium lactate feed solution. Lower purity with higher lactate recovery was obtained when the feed solution was sodium lactate and ammonium lactate. When the actual fermentation broth was used in the two-stage RO unit, a slightly lower recovery and purity of lactic acid were obtained. It was claimed that the total ions present in the fermentation broth were responsible for the low efficiency of the two-stage RO unit.

Keywords: lactic acid; fermentation broth; reverse osmosis; Donnan exclusion effect; ionic strength.

Owing to its chiral structure, consisting of both carboxyl and hydroxyl groups, lactic acid has been extensively used in the food, cosmetic, and pharma- Correspondence: N. Thongchul, Research Unit in Bioconversion/ /Bioseparation for Value-Added Chemical Production, Institute of Biotechnology and Genetic Engineering, Chulalongkorn Univer-sity, Phayathai Road, Wangmai, Pathumwan, Bangkok 10330, Thailand. E-mail: [email protected] Paper received: 14 March, 2017 Paper revised: 15 August, 2017 Paper accepted: 29 August, 2017

https://doi.org/10.2298/CICEQ170314030P

ceutical industries. To date, the market demand for lactic acid has increased considerably because of the global trend toward environmental protection that accelerated the supply of biodegradable plastics. Currently, bacterial fermentation is widely used in industrial lactic acid production. During production, the pH is typically decreased as the acid is produced, and the proper operating pH must be maintained at 5.5–6.5 [1]. pH control is carried out by the addition of bases such as calcium bases (CaCO3 and Ca(OH)2), NaOH and NH4OH [2,3]. This generates 2 lactate species (free lactic acid and lactate salt) in the fer-

N. PHANTHUMCHINDA et al.: ALTERNATIVE REVERSE OSMOSIS TO PURIFY… Chem. Ind. Chem. Eng. Q. 24 (2) 179−190 (2018)

180

mentation broth, depending on the operating pH con-trol. At the end of the fermentation, lactate salts and free lactic acid must be recovered in the form of free acid in the downstream operation units.

In the recovery and purification of lactic acid from the fermentation broth, many major process steps are involved: e.g., primary recovery, product purification and finishing processes [4]. The first step in primary recovery units involves cell separation by centrifugation or microfiltration [4-7]. Several down-stream techniques have been used for recovering and separating lactic acid, including reactive distillation, reactive extraction, adsorption (ion exchange), and membrane separation (electrodialysis) [8]. Both distil-lation and electrodialysis have high energy con-sumption in a large-scale operation because of the low volatility of lactic acid and high electricity loading, respectively [9]. When charged compounds such as amino acids and other organic acids are present, the separation efficiency of electrodialysis decreases as a result of membrane fouling. Electrodialysis fails to reject divalent ions such as calcium and magnesium; therefore, it is not suitable for recovering lactic acid from the conventional calcium base fermentation pro-cess where lactate species are present in the form of calcium lactate and lactic acid [10,11]. In reactive extraction, toxicity of the solvent and final product contamination are usually of concern [12]. Using ion exchangers to recover lactic acid requires large amounts of chemicals, enzymes and process water during the resin regeneration and washing steps. This eventually generates a large effluent loading in waste-water treatment. In addition, precautions should be taken in the pretreatment steps before the feed enters the ion exchangers so as to avoid fouling and resin deterioration [13]. Purified lactic acid then enters the evaporator where water is removed, resulting in con-centrated lactic acid as the finished product [14-16]. As previously mentioned, process integration is necessary in order to achieve good recovery and purification performance as well as cost effectiveness.

Membrane separation provides advantages of low energy consumption and low toxicity. Never-theless, a single membrane unit cannot fulfill lactate recovery, purification, and concentration of lactic acid [7]. Typically, nanofiltration is used prior to the pro-duct finishing step (evaporation in case of lactic acid recovery and purification). Nanofiltration is applied for the removal of trace ions and small, neutral molecules from free lactic acid solution [14]. On the other hand, reverse osmosis (RO) is generally applied for the removal of water in previously published literature [7]. In this study, a two-stage RO membrane-based pro-

cess was developed for recovering, purifying, and concentrating lactic acid from the model solution and the cell-free fermentation broth by applying the appropriate pressure higher than the osmotic pressure of the species of interest. From the principle of the RO process, solute transport occurs by dif-fusion through the membrane depending on mole-cular size and charge [17]. With smaller pore sizes than the nanofiltration membrane, the RO membrane provides high efficiency in the rejection of monovalent ions [18]. It was also reported that some trace organic compounds, such as neutral molecules smaller than the molecular weight cut-off of the membrane, leaked and passed through the membrane [19]. By applying an operating pressure sufficiently higher than the osmotic pressure of the molecules, such molecules can pass through the RO membrane. At the proper pH, free lactic acid passed across the first RO membrane to the permeate side whilst the cation salts remained in the retentate. The second RO unit was for pre-concentrating lactic acid solution. Compared with the other techniques mentioned previously, RO filtration did not require chemicals, enzymes and pro-cess water during the operation. The low volumetric rate of the exit stream from this two-stage RO unit resulted in size reduction of the evaporator and lower capital and operating expenditures.


Chemicals

Calcium lactate (CaLAC), sodium lactate (NaLAC), and ammonium lactate (NH4LAC), pur-chased from Sigma-Aldrich, were used in this study as model solutions. These chemicals were dissolved directly in deionized water to the specified concen-tration (5 g/L lactic acid equivalent). This equivalent mass concentration of lactic acid resulted in different pH values of the solution, e.g., 4, 4, and 9 for CaLAC, NaLAC, and NH4LAC solutions, respectively. To obtain the specific tested pH at 4 and 6, the pH of the model solution was adjusted by 5 M NaOH or 1 M H2SO4. The concentrations of the lactate species at equilibrium (both free lactic acid and its salts) were dependent on the pH. The following stoichiometry describes the presence of lactate species mimicking those that appear in the fermentation processes.

When CaCO3 was used for pH control during the fermentation, both lactic acid and calcium lactate were present in the solution:

3 3

3 2 2 2

2CH CHOHCOOH CaCO(CH CHOHCOO) Ca H O CO

++ +


181

When NaOH was used for pH control during the fermentation, both lactic acid and sodium lactate were present in the solution:

3

3 2

H CHOHCOOH NaOH(CH CHOHCOO)Na H O

++

When NH4OH was used for pH control during the fermentation, both lactic acid and ammonium lactate were present in the solution:

3 4

3 4 2

CH CHOHCOOH NH OH(CH CHOHCOO)NH H O

++

Fermentation broth preparation

Lactate fermentation broth was prepared from the cultivation of Bacillus coagulans BC-013 in a 5 L stirred fermenter. An active 24 h glucose–yeast ext-ract–peptone slant was used to prepare the bacterial suspension. The bacterial suspension (1% inoculum size) was inoculated in a preculture flask containing the preculture medium. The preculture medium con-tained (per liter) 10 g glucose, 15 g yeast extract, 4 g NH4Cl, 0.5 g KH2PO4, 0.5 g K2HPO4, 5 g CaCO3 and 20 mL salt solution. The compositions of the salt sol-ution consisted of (per 10 mL) 400 mg MgSO4⋅7H2O, 20 mg MnSO4⋅5H2O, 20 mg FeSO4⋅7H2O and 20 mg NaCl. The preculture flask was incubated at 50 °C, 200 rpm for 3 h. After that, the preculture flask was transferred into the 5 L stirred fermenter containing 2.5 L sterile preculture medium at 10% inoculum size. The fermenter was operated at 50 °C and agitated at 300 rpm with 1 vvm air. After 3 h, 0.5 L of the fer-mentation medium containing (per L) 720 g glucose was added into the fermenter. Aeration was then stopped. Three different bases, i.e., CaCO3, NaOH, and NH4OH, were used for pH control at 6.5. As a result, 3 different lactate salts, i.e., CaLAC, NaLAC, and NH4LAC, were obtained in the fermentation. Fer-mentation was continued for 48 h until glucose dep-letion. Next, the fermentation broth was harvested.

Cell biomass and soluble, neutral macromolecules such as proteins, sugars, etc. were removed from the fermentation broth by microfiltration and ultrafiltration. The cell-free broth obtained was later used in the two-stage RO unit.

Designing and setting up the RO unit

An in-house RO unit was constructed by a local Thai company (Icrotech Co., Ltd.) for use in this study. The apparatus set-up is illustrated in Figure 1.

Two RO membrane filtration units were sub-sequently connected with auxiliary instruments, inc-luding boost pumps, pressure gauges, flowmeters for the feed, concentrate and permeate, valves and stor-age vessels. Negatively charged brackish water RO (BWRO) elements (DOW FILMTECTM BW60-1812- -75) were installed at the first RO unit for recovering lactate from the model solution while allowing some salts to pass through the membrane at a rejection percentage of 97–99% (Table 1). In the second RO unit, positively charged seawater RO (SWRO) ele-ments (DOW FILMTECTM SW30-2521) were installed for pre-concentrating recovered lactate obtained from the first RO unit where water was expelled. The rejection percentage was higher than 99.4% (Table 1). As a result, lactate was passed through the BWRO unit to the permeate side and later entered the SWRO unit. Lactic acid was then concentrated in the follow-ing SWRO unit and remained in the retentate.

Batch operation was used in the first RO unit where the feed solution entered the unit and the permeate discharged from the first unit to enter the second RO unit later on. In the second RO unit, the retentate was recycled so that most of the remaining water in the retentate was discharged into the per-meate. The maximum operating pressure for the BWRO and SWRO units was set at 6 and 15 bar, res-pectively, owing to the pressure limit of apparatus housing, pumps, and piping systems.

Figure 1. Schematic diagram of the in-house RO unit for recovery and purification of lactic acid.


182

Table 1. Characteristics of DOW FILMTECTM RO membrane elements used in this study

Product name DOW FILMTECTM

BW60 DOW FILMTECTM

SW30

Model BW60-1812-75 SW30-2521

Membrane type Polyamide thin-film composite

Polyamide thin-film composite

Membrane surface area, m2

0.77 1.20

Membrane surface charge

Negative Positive

Dimension, mm×mm 44.5×305 63.5×533

Permeate flow rate, L/h 12 45

Average NaCl rejection, %

97–99 99.4

Maximum applied pressure, bar

10 55

Maximum applied temperature, °C

45 45

pH range 2–11 2–11

Determining the operating conditions

To utilize the principle of RO in the recovery of lactic acid, osmotic pressure was introduced as a key factor determining the process conditions. The oper-ating pressure was adjusted over the osmotic pres-sure of lactic acid so that lactic acid could pass through the BWRO membrane to the permeate, with other fermentation impurities remaining in the retentate. Eq. (1) expresses osmotic pressure as a function of molarity and temperature:

iMRTπ = (1)

where π is the osmotic pressure (bar), i is the dim-ensionless van’t Hoff factor, M is the molar concen-tration of the solution/species of interest (in this case, lactic acid and its salts), R is the gas constant (0.082 L⋅bar/(K⋅mol)) and T is the temperature in K.

From Eq. (1), the osmotic pressures of the 3 model solutions, e.g., CaLAC, NaLAC, and NH4LAC with 5 g/L lactate equivalent to be studied are 2.12 bar, 2.82 and 2.81 bar, respectively. The osmotic pressure of lactic acid solution at 5 g/L is 1.41 bar.

The effects of pH and operating pressure on lactate separation efficiency at the BWRO unit were investigated. The lactate model solution (2 L) at 5 g/L LAC equivalent was adjusted to the tested pH values of 4 and 6 using either NaOH or H2SO4. The oper-ating pressure was varied at 4 and 6 bar. Further increasing the concentration of the lactate model solution to higher than 5 g/L LAC equivalent resulted in higher osmotic pressure that exceeded the maxi-mum pressure threshold in the BWRO apparatus (7

bar) and consequently led to reduced mass flux and separation efficiency. Therefore, the tested concen-tration of the model solution was limited at 5 g/L LAC equivalent.

The model solution (2 L) in the feed tank was fed into the first BWRO unit where free lactic acid was supposed to be separated from other impurity species, including Ca2+, Na+, NH4

+ and SO42− (in case

the model solution was adjusted to the desired pH by H2SO4). The operating temperature was set at 30 °C. The apparatus was run until the collected volume of the permeate of 1.6 L was obtained. Samples (20 mL) were periodically collected from both permeate and retentate for analyses of free lactic acid concentration and ion species.

The permeate that left the BWRO unit and col-lected in the SWRO feed tank (1.6 L) was passed through the SWRO unit where water separation occurred, which resulted in lactate concentration in the retentate. The operating temperature was also set at 30 °C. The effects of pH and pressure on sep-aration efficiency were determined. The tested pres-sure was set at 13 and 15 bar. At the first 5 min of operation, the retentate was recycled into the SWRO feed tank. After the recycling was stopped, the oper-ation was continued until the permeate flux became zero. Samples (20 mL) were periodically collected for analyses of lactic acid and all the major remaining impurities.

Sample analyses

During the runs, samples from the permeate and retentate obtained in each RO unit were collected periodically for determining discharge volume, mea-suring pH and analyzing substances that remained. For ion analyses, the collected samples were ana-lyzed for concentrations of lactic acid and major imp-urity ions, including Ca2+, Na+, N (representing NH4

+), SO4

2−, Cl−, P and Mg2+. L-lactate ion in the sample was analyzed with a

glucose–lactate analyzer (YSI2700, Yellow Spring Ins-truments Inc.) within the detection range of 0–2.67 g/L. The sample size of 25 μL was automatically injected into the reaction chamber where the enzymatic react-ion occurred. The reading of L-lactate concentration was explained by the action of L-lactate oxidase immobilized at the membrane sensor.

An atomic absorption spectrophotometer was used to determine the metal concentration, including Ca2+ and Na+, in the sample. The sample was pre-pared by dilution with 5 vol.% HNO3 solution. An air flame of 13.60 L/min along with an acetylene flame of 2 L/min was used for metal atomization of the sample


183

before reading the atomic absorption with the atomic absorption spectrophotometer (AA280FS, Varian, Inc.). Aqueous standard solutions were prepared by dilution to appropriate concentrations (2, 4, 6, and 8 mg/L for Ca2+, and 5, 10, 20 and 30 mg/L for Na+). The concentration of Ca2+ and Na+ in the sample was calculated by comparing the spectra with the standard calibration curves.

Nitrogen content was determined by the total Kjeldahl nitrogen technique. Nitrogen in the sample was first converted to NH3 by metal-catalyzed acid digestion. The resulting NH3 was separated from the sample by distillation. The released NH3 was cap-tured in a diluted H2SO4 solution. The result repre-sented organic nitrogen after digestion and distillation in the sample. The digestion reagent (catalyst) was prepared by mixing 134 g K2SO4 and 7.3 g CuSO4 in 134 mL concentrated H2SO4. After that, the volume was made up to 1 L. Digestion reagent (50 mL) was added into the sample, and digestion proceeded for 30 min (Buchi, K499). Later, 50 mL boric acid was added into the reaction mixture as the absorbent sol-ution during NH3 distillation (Buchi, K350). Finally, NH3 was determined by titration with a standard sol-ution (Buchi, K350) [20].

Chloride was analyzed by the potentiometric method. The solubilized chloride ion in the sample was measured by a chloride ion-selective electrode during titration (Orion 720A, Labx Inc.). The sample was mixed with concentrated HNO3 before dilution to the proper concentration. Titration was performed with a standard AgNO3 solution as the reference.

Phosphorus was determined by the total phos-phorus method using persulfate digestion. The sample (50 mL) was mixed with 11 N H2SO4 (1 mL). Next, dissolved and particulate phosphorus in the sample was digested with (NH4)2S2O8 (0.4 g) to convert phos-phorus into orthophosphate (mixed and boiled to obtain a final volume of 40 mL). The orthophosphate concentration was measured by a spectrophotometer (Nova Spec 2, Pharmacia Biotech, Inc.) using a standard calibration curve. The calibration curve was prepared from a standard phosphorus solution (0.3– –1.2 mg P/L).

Sulfate ion in the sample was determined by the turbidimetric method. Sulfate ion present in the sample was converted into a BaSO4 suspension under controlled conditions. The sample (80 mL) was mixed with 20 mL buffer solution containing (per liter) 30 g MgCl2⋅6H2O, 5 g CH3COONa⋅3H2O, 1 g KNO3 and 20 mL acetic acid (99%). Then, BaCl2 was added into the reaction mixture to obtain BaSO4 precipitate. The tur-bidity was measured by a spectrophotometer (2100P,

HACH). The concentration was determined using the calibration curve of the standard sulfate solution [20].

Investigating the performance of the two-stage RO unit

The performance of the two-stage RO unit was evaluated using 6 criteria: mass flux of lactic acid, lactic acid separation, ion separation, lactic acid recovery, overall recovery, and purity.

The mass flux of lactic acid (JLA) at the BWRO unit was calculated by the following equation:

LA,BWPLA

mJ

At= (2)

where mLA,BWP is the lactic acid mass (g) passing through the membrane, A is effective membrane surface area (m2), and t is time (h).

The efficiency of the BWRO unit to separate lactic acid from other ions can be explained by lactic acid separation (SLA) in percentage defined by Eq. (3):

LA,BWPLA

LA

100mS

F= (3)

where mLA,BWP is the lactic acid mass (g) passing through the BWRO unit and FLA is the initial mass of lactic acid present in the feed solution (g).

The ion (i) leakage at the BWRO unit can be described by the separation percentage (Si), as seen in Eq. (4):

i,BWPi

i

100mS

F= (4)

where mi,BWP is the mass of ion i (g) moving through the BWRO unit and Fi is the initial mass of the ion (g) present in the feed solution.

The efficiency of the SWRO unit to pre-concen-trate lactic acid can be represented by lactic acid recovery in percentage (RLA) described by Eq. (5):

LA,SWF LA,SWPLA

LA,SWF

100C C

RC

−= (5)

where CLA,SWF and CLA,SWP are lactic acid concentra-tions (g/L) present in the feed solution entering the SWRO unit and the permeate leaving the SWRO unit.

The overall recovery (Roverall) of lactic acid pro-duct obtained from the two-stage RO unit can be described by Eq. (6):

overall LA LA100R S R= (6)

where SLA and RLA were defined from Eqs. (3) and (5), respectively.


184

The purity of lactic acid product (PLA) obtained from the two-stage RO unit can be described by the mass ratio of lactic acid and the total ions that rem-ained in the retentate of the SWRO unit (Eq. (7)).

LA,SWRLA

T,SWR

100mP

m= (7)

where mLA,SWR and mT,SWR represent the mass of lactic acid product and the total mass of ions that remained in the retentate leaving the SWRO unit.


Lactate separation at the BWRO unit

A feed solution of different lactate salts, inc-luding CaLAC, NaLAC, and NH4LAC, was prepared at an equivalent lactate concentration of 5 g/L. To obtain the specific tested pH at 4 and 6, NaOH and H2SO4 were added into the solution to adjust to the desired tested pH, which eventually resulted in changes in the molar concentration of chemical ions present in the model solutions (Table 2).

The obtained model solutions were then trans-ferred into the feed tank to be pumped into the BWRO unit at different operating pressures (4 and 6 bar)

where lactic acid presumably passed through, whereas the other cations remained in the retentate. Figure 2 shows the lactate mass flux passing through the BWRO membrane. The lactate mass flux inc-reased with increasing operating pressure. It appears that the pH strongly influenced lactate transport through the BWRO membrane for CaLAC and NaLAC. On the other hand, pH showed less effect on lactate transport when compared with the operating pressure for NH4LAC.

The separation efficiency of the BWRO unit is expressed by Eqs. (3) and (4) and displayed in Figure 3. It was observed that the operating pressure exhi-bited a strong effect on lactate separation in the BWRO unit. High lactate separation efficiency (% lactate pas-sage) was obtained from all 3 model solutions at 6 bar regardless of changes in pH compared with the runs using an operating pressure of 4 bar. It was sug-gested that increasing the operating pressure from 4 to 6 bar improved lactate separation owing to the larger difference in operating pressure and the osmo-tic pressure of lactic acid generating a larger driving force across the membrane, which eventually resulted in a higher diffusion rate (high lactate flux as seen in Figure 2). Operating pressures higher than the osmo-

Table 2. Molar concentration of chemical species present in different lactate model solutions containing a lactic acid equivalent of 5 g/L at pH 4 and pH 6

Species content, mol/L pH 4 pH 6

CaLAC NaLAC NH4LAC CaLAC NaLAC NH4LAC

Total LAC− Free LA (cal) LAC−

(cal)

0.056 0.024 0.032

0.056 0.025 0.031

0.056 0.018 0.038

0.056 0.024 0.032

0.056 0.025 0.031

0.056 0.018 0.038

Ca2+ 0.016 - - 0.016 - -

Na+ - 0.031 - 0.050 0.086 -

NH4+ - - 0.038 - - 0.038

SO42− - - 0.015 - - 0.007

H3O+ 10−4 10−4 10−4 10−6 10−6 10−6

OH− 10−10 10−10 10−10 10−8 10−8 10−8

Total ions 0.0721 0.0821 0.1091 0.1220 0.1420 0.1010

Figure 2. Lactate flux at the permeate of the BWRO unit operated at 30 °C. CaLAC, NaLAC and NH4LAC containing the initial lactate

equivalent of 5 g/L at different pH values were passed through the BWRO unit at different operating pressures.


185

tic pressure of the solution resulted in an increasing mass flux throughout the membrane and thus the separation efficiency [21]. It should be noted that the osmotic pressure of the solution was increased with increasing concentration of the feed solution [22]. Therefore, operation at a certain pressure with varied feed concentrations resulted in a different permeate flux, and eventually lactate separation efficiency, in the 3 model solutions studied.

From Figure 3, pH was found to be responsible for lactate separation in the BWRO unit. This was related to the amount of total ions present in the feed solution, which played a role in mass transport across the membrane (Table 2). Liew et al. claimed that as the ionic strength of the feed solution increased, the permeate flux decreased as a result of increases in osmotic pressure and viscosity [23]. It was observed that the higher total ion concentration lowered the lac-tate flux (Figures 2 and 3). At the same operating pressure, higher permeate flux resulting in signific-antly higher lactate separation at the BWRO unit was achieved at a lower pH with a rapid diffusion rate (pH 4 compared with pH 6) in the case of CaLAC and NaLAC feed solutions, when the total ion concentra-tion of the feed solution was lower (Table 2). On the other hand, in the case of NH4LAC, slightly increasing permeate flux and lactate separation were obtained at

pH 6. This was presumably due to the slight change in total ion concentration, resulting in a similar ionic strength between the 2 pH values studied. The find-ings in this work confirmed that the pH and the total ion concentration of the feed solution played a role in controlling permeate flux, and thus separation effi-ciency, at the BWRO unit.

Ion rejection by the BWRO membrane

As previously mentioned, the separation of lac-tate from other ions was expected at the BWRO unit. Nonetheless, not only lactate species but also cal-cium, sodium and ammonium ions could pass through the BWRO membrane (Figure 3). Several interaction mechanisms of salt passage through the membrane have been investigated, including convection, dif-fusion, and charge repulsion. It was claimed that both membrane charge and feed ionic strength played a significant role in salt rejection [24]. When a typical feed solution interacted at the surface of the nega-tively charged membrane, the ion shift was generated at the boundary between the membrane and the sol-ution, resulting in an electrical potential known as the Donnan exclusion effect [22]. In the case of uncharged solutes such as undissociated lactic acid, solution transport mainly occurred through diffusion and con-vection. The larger the difference between the oper-

Figure 3. Lactic acid separation at the BWRO unit operated at 30 °C. CaLAC, NaLAC and NH4LAC with the initial lactate equivalent of

5 g/L were passed through the BWRO unit at different operating pressures of 4 and 6 bar.


186

ating pressure and the osmotic pressure, the larger the percentage of undissociated lactic acid that passed through the BWRO membrane. When lactic acid species were present in the dissociated form at an operating pH higher than the pKa value (3.86), the Donnan exclusion effect governed the transport of ion species through the BWRO membrane [25]. Thus, higher lactate rejection was observed in all 3 model solutions at pH 6 owing to a larger electric repulsive force by the negatively charged surface (Figure 3).

Considering the passage of Ca2+, Na+, NH4+,

and H+ through the BWRO membrane, these cations typically bind at the membrane surface. The higher the pH, the more the dissociated lactate and the more the negatively charged membrane brings larger cat-ions to the membrane surface [21]. It was suggested that the larger ions had lower diffusion rates and thus were expected to have lower concentrations in the permeate. Size controlled ion diffusion, and the ability of ions to form hydrogen bonds with the carbonyl group of the polyamide membrane facilitated the pas-sage of such ions [22,23]. In addition, Tu et al. and Zaidi et al. confirmed that salt rejection by the BW30 membrane was dominated mostly by size exclusion [26,27]. Thus, in our work, most of Ca2+ were retained whereas Na+ and NH4

+ apparently passed through the BWRO membrane.

Water permeability and solute rejection at the SWRO unit

Lactate concentration was determined in the SWRO unit under different operating pressures. After passing through the BWRO unit, the model solution was passed through and recirculated in the SWRO unit for 5 min. The sample was collected for analyses of lactate in both the retentate and permeate. The performance of the SWRO unit in terms of lactate recovery and water permeation is shown in Table 3.

Slightly increasing the pressure from 13 to 15 bar did not result in significant changes in SWRO performance. A slight increase in lactate concentra-tion at the retentate with a higher water permeable flux was obtained at 15 bar. From Table 3, it is apparent that a relatively high rejection percentage of 95-99% was obtained at the SWRO unit operated at 13 bar. Therefore, running SWRO at this pressure not only gave an increasing concentration of lactic acid product at the retentate, but the operating cost could also be reduced [28]. Although the operating pressure used in this study did not show a significant effect on lactate recovery owing to the limited applied pressure to the apparatus up to 15 bar, it is believed that with higher operating pressure, higher lactate rejection

rate and water flux should have been obtained, even-tually resulting in increasing concentration of lactic acid product at the retentate [28].

Table 3. Effect of operating pressure on lactate recovery at the SWRO unit. Permeates from the BWRO unit passed through the SWRO unit at 30 °C where water was expelled yielding concentrated lactic acid solution

Starting feed CaLAC NaLAC NH4LAC

At 13 bar

Lactate at permeate, g/L 0.05 0.20 0.09

Lactate at retentate, g/L 6.45 9.68 7.97

Lactate rejection, % 98 95 99

Cation rejection, % 99 99 99

Water flux, L/(m2 h) 11.4 10.8 8.4

At 15 bar

Lactate at permeate, g/L 0.07 0.22 0.11

Lactate at retentate, g/L 7.47 11.50 9.54

Lactate rejection, % 98 95 100

Cation rejection, % 99 99 90

Water flux, L/(m2 h) 13.2 12.0 10.8

A typical RO operation involved the removal of inorganic and organic salts from the aqueous solution [29]. The SWRO membrane used in this work was the positively charged membrane containing free amine groups; therefore, high cation rejection was expected, especially at the lower pH (pH 4) when the feed sol-ution was more protonated and lactic acid was pre-sent more in the undissociated form. Similar to obs-ervations in the BWRO unit, evidence of some NH4

+ leaking out from the SWRO membrane could be exp-lained by hydrogen bonding to the carbonyl group of the polyamide membrane facilitating the passage of NH4

+ through the permeate although Ca2+ and Na+ ions were strongly rejected because of the repulsive force of the positively charged surface [22,23]. Evi-dence of high rejection percentages of both cations (Ca2+, Na+, and NH4

+) and lactate ions confirmed that the SWRO unit was successfully utilized to concen-trate lactate at the retentate by expelling water through the membrane (Table 3).

Total mass balance and efficiency of lactate recovery at the two-stage RO units

Figure 4 presents the total mass balance over the two-stage RO units. The first BWRO unit was considered the key operating unit where lactate ions were separated from the cations, and the second SWRO unit was for concentrating the product remain-ing in the retentate.

From the three model solutions studied, it was found that more than 50% of lactic acid from the feed


187

stream was recovered from the two-stage RO units (Table 4). Compared with the other 2 feed solutions, when the feed stream was CaLAC, a lactic acid purity of 99.2% was obtained. Nonetheless, the total rec-overy seemed to be slightly low (50.5%). It should also be noted that the highest lactic acid purity was obtained with the lowest recovery percentage.

In addition, the efficiency of the two-stage RO unit was tested with the actual lactic acid fermentation broth. The fermentation broths, including CaLAC

broth, NaLAC broth, and NH4LAC broth, primarily passed through the microfiltration and ultrafiltration units where cells and proteins were separated. The solutions were then diluted to obtain the equivalent concentration of lactic acid of 5 g/L before entering the RO units operated at optimized conditions det-ermined before allowing lactic acid recovery and puri-fication. Table 5 presents the efficiency of the two-stage RO units constructed in this study on lactic acid separation and purification from the actual ferment-

Figure 4. Total mass balance over the two-stage RO units. the model solution (a: CaLAC; b: NaLAC; c: NH4LAC) was fed into the

apparatus operated under optimized pH and pressure.

Table 4. Overall recovery and purity of lactic acid from the 3 different model solutions after passing through the two-stage RO units operated at optimized conditions

Feed solution CaLAC NaLAC NH4LAC

Operating conditions pH 4, 6 bar at BWRO pH 4.6, 15 bar at SWRO

pH 4, 6 bar at BWRO pH 4.5, 15 bar at SWRO


Total lactic acid recovery 50.5% 66.4% 70.3%

Purity 99.2% 89.9% 89.7%


188

ation broths. Compared with the model solutions, the overall lactate recovery was similar whereas the pur-ity was lower.

Various ions present in the actual fermentation broth were claimed to be responsible for the lower purity (Figure 5). The amount of total ions present in the feed fermentation broth was higher than that of the model solution; therefore, the Donnan exclusion effect was lowered, resulting in increasing ion pas-sage across the BWRO membrane [24].

From the findings in this study, it can be pre-sumably concluded that the membrane-based pro-cess to recover and purify lactic acid from the fer-mentation broth has 2 major advantages. The first one is that no pretreatment is required for the cell-free fermentation broth before entering the two-stage RO unit to recover, purify, and concentrate lactic acid. In general, pretreatment of the cell-free fermentation broth by acidification using H2SO4 is necessary for lactate recovery by the typical ion exchange resin-based process. Furthermore, using the typical ion exchanger to separate lactic acid from the ferment-ation broth requires 3 main steps including feed stream loading (adsorption), washing (to remove unbound solution from the resins), and lactic acid elu-tion by proper eluant (desorption) [8]. This resulted in the increasing consumption of chemicals, wastewater treatment, and eventually dilution of the fermentation

broth after acidification. Secondly, without pretreat-ment of the cell-free fermentation broth and applying the two-stage RO unit for lactic acid recovery, the volume of cell-free fermentation broth remained unchanged. Therefore, the downstream equipment sizing can be smaller compared with the typical down-stream process using ion exchange resins. Although the broth had to be diluted to 5 g/L before entering the two-stage RO unit, the performance of this unit to recover, pre-purify, and pre-concentrate lactic acid was evident. This strongly indicated the beneficial outcome of this process, especially when we could operate without the pressure limit as experienced in this work with our in-house apparatus.

It is suggested that lactic acid loss in the ret-entate at the BWRO unit can be reduced by stream recycling at the feed tank. In addition, to increase the purity of lactic acid product, the retentate from the SWRO unit is recommended to pass through another membrane unit, i.e., a nanofiltration unit, to eliminate the remaining trace ions. By selecting the proper membrane, it is believed that the purity of lactic acid product can be increased [30]. From the findings of this study, it can be claimed that the two-stage RO membrane process has for the first time been rep-orted as a simple operation but effective in recovering and purifying lactic acid from the fermentation broth.

Table 5. Performance of the two-stage RO unit to recover and purify lactic acid from the fermentation broth

Feed solution CaLAC broth NaLAC broth NH4LAC broth

Operating conditions pH 6, 6 bar at BWRO pH 5.39, 15 bar at SWRO



Lactate passage at BWRO 54.2% 66.9% 72.0%

Lactate rejection at SWRO 100% 97.4% 99.6%

Total lactic acid recovery 54.2% 65.2% 71.7%

Purity 86.1% 73.3% 74.7%

Figure 5. Percentage of ion leakage into the permeate. The different cell-free lactate broth solutions, including CaLAC broth, NaLAC

broth and NH4LAC broth, entered the BWRO unit operated under the optimized conditions previously determined.


189

CONCLUSIONS

The in-house two-stage RO unit constructed in this study was used to recover and purify lactic acid. The acceptable amount of free lactic acid recovered with sufficiently high purity was obtained under opti-mized operating conditions. This unit was applicable to different fermentation broth solutions, including calcium lactate, sodium lactate and ammonium lac-tate. Although the operating pressure was set at a higher value than the lactate osmotic pressure, lac-tate rejection was still observed at the BWRO unit where most of the lactate was expected to pass through the membrane while other ions remained in the retentate. It was found that the total ion concen-tration of the feed solution and the operating pH both played a crucial role in controlling ion leakage across the membrane, thereby controlling both lactate rec-overy and purity of this RO membrane system. From the results obtained in this study, it is suggested that by coupling this two-stage RO unit with the upstream fermentation and primary cell and protein separation (microfiltration and ultrafiltration) units, the simple design of continuous fermentation and lactic acid rec-overy to achieve high productivity in long-term oper-ations is feasible.

Acknowledgments

This work was conducted under the University–Industry Linkage Program with the collaboration of researchers from Chulalongkorn University and PTT Global Chemical Public Company Limited. Partial support by Grant for International Research Inte-gration: Research Pyramid, Ratchadapiseksomphot Endowment Fund (GCURP_58_01_33_01) and Thail-and Research Fund via the Distinguished Research Professor Grant (DPG5880003) was highly appre-ciated. Research facility supported by Chulalongkorn Academic Advancement into its 2nd Century Project (CUAASC) is highly acknowledged. SA and NT are the recipients of the NSTDA Chair Professor Grant (No. 5) funded by the Crown Property Bureau of Thailand and National Science and Technology Dev-elopment Agency.

REFERENCES

[1] A.N. Vaidya, R.A. Pandey, S. Mudliar, M.S. Kumar, T. Chakrabarti, S. Devotta, Crit. Rev. Environ. Sci. Technol. 35 (2005) 429-467

[2] K. Hetenyi, A. Nemeth, B. Sevella, Chem. Eng. Process. 50 (2011) 293-299

[3] S. Nakano, C.U. Ugwu, Y. Tokiwa, Bioresour. Technol. 104 (2012) 791-794

[4] P.F. Stanbury, A. Whitaker, S. J. Hall, In Fermentation Technology Handbook. Pergamon Press, Oxford, 2017, p. 619

[5] Y. Li, A. Shahbazi, C.T. Kadzere, Food. Eng. (2006) 574- –580

[6] M.A.A. Rahman, Y. Tashiro, K. Sonomoto, Biotechnol. Adv. (2013) 877–902

[7] P. Pal, J. Sikder, S. Roy, L. Giorno, Chem. Eng. Process. 48 (2009) 1549–1559

[8] C. Rodrigues, L.P.S. Vandenberghe, A.L. Woiciechowski, J. de Oliveira, L.A.J. Letti, C.R. Soccol, Biotechnol. Bioeng. (2017) 543-556

[9] P. Khunnonkwao, P. Boontawan, D. Haltrich, T. Mais-chberger, A. Boontawan, Process. Biochem. 47 (2012) 1948-1956

[10] J.H. Kim, J.-G. Na, H.J. Shim, Y.K. Chang, J. Membr. Sci. 396 (2012) 110-118

[11] E.G. Lee, S.H. Kang, H.H. Kim, Y.K. Chang, Biotechnol. Bioproc. Eng. 11 (2006) 313–318

[12] K.L. Wasewar, A.A. Yawalkar, J.A. Moulijn, V.G. Pan-garkar, Ind. Eng. Chem. Res. 43 (2004) 5969-5982

[13] H.G. Joglekar, I. Rahman, S. Babu, B.D. Kulkari, A. Joshi, Sep. Purif. Technol. 52 (2006) 1-17

[14] T. Ghaffar, M. Irshad, Z. Anwar, T. Aqil, Z. Zulifqar, A. Tariq, M. Kamran, N. Ehsan, S. Mehmood, J. Radiat. Res. Appl. Sci. (2014) 222-229

[15] C.S.L. Garzon, A.J.J. Straathof, Biotechnol. Adv. (2014) 873-904

[16] B. Wojtyniak, J. Kołodziejczyk, D. Szaniawska, Chem. Eng. (2016) 28-36

[17] C. Bellona, J.E. Drewes, P. Xu, G. Amy, Water. Res. 38 (2004) 2795-2809

[18] S.S. Sablani, R.M. Shafiur, A.K. Datta, A.S. In Food and Bioprocesses Modeling Techniques Handbook, CRC press, Boca Raton, FL, 2007

[19] K. Kosutic, B. Kunst, Desalination 142 (2002) 47-56

[20] D.E Andrew, S.C. Lenore, W.R Eugene, E.G. Arnold, In Water and Wastewater Handbook, A.H.F Mary Ed., American Public Health Association and Water Envi-ronment Federation, Washington DC, 2005, p. 131

[21] V. Freger, T.C. Arnot, J.A. Howell, J. Membr. Sci. 178 (2000) 185-193

[22] M.I. Gonzalez, S. Alvarez, F.A. Riera, R. Alvarez, Desalination 228 (2008) 84-96

[23] M.K.H. Liew, S. Tanaka, M. Morita, Desalination. 101 (1995) 269-277

[24] C. Bartels, R. Franks, S. Rybar, M. Schierach, M. Wilf, Desalination 184 (2005) 185-195

[25] P. Dey, L. Linnanen, P. Pal, Desalination 288 (2012) 47- –57

[26] K.L. Tu, L.D. Nghiem, A.R. Chivas, Chem. Eng. J. 168 (2011) 700-706

[27] S.M.J. Zaidi, F. Fadhillahb, Z. Khanc, A.F. Ismail, Desal-ination 368 (2015) 202-213

[28] M.H. Oo, L. Song, Desalination. 246 (2009) 605-612


190

[29] R.A. Diltz, T.V. Marolla, M.V. Henley, L. Li, Bioresour. Technol. 98 (2007) 686-695

[30] M.-S. Kim, J.-G. Na, M.-K. Lee, H. Ryu, Y.-K. Chang, J.M. Triolo, Y.-M. Yun, D.-H. Kim. Water. Res. 96 (2016) 208-216.

NATNIRIN PHANTHUMCHINDA1

TANAPAWARIN RAMPAI1

BUDSABATHIP PRASIRTSAK1 SITANAN THITIPRASERT2

SOMBOON TANASUPAWAT3

SUTTICHAI ASSABUMRUNGRAT4

NUTTHA THONGCHUL2 1Program in Biotechnology, Faculty of

Science, Chulalongkorn University, Wangmai, Pathumwan, Bangkok,

Thailand 2Research Unit in Bioconversion/

/Bioseparation for Value-Added Chemical Production, Institute of

Biotechnology and Genetic Engineering, Chulalongkorn University,

Wangmai, Pathumwan, Bangkok,Thailand

3Research Unit in Bioconversion/ /Bioseparation for Value-Added

Chemical Production, Department of Biochemistry and Microbiology, Faculty of Pharmaceutical Sciences, Wangmai,

Pathumwan, Bangkok, Thailand 4Department of Chemical Engineering, Faculty of Engineering, Chulalongkorn

University, Wangmai, Pathumwan, Bangkok, Thailand

NAUČNI RAD

ALTERNATIVNA REVERZNA OSMOZA ZA PREČIŠĆAVANJE MLEČNE KISELINE IZ FERMENTACIONE KOMINE

U dvostepenoj jedinici za reverznu osmozu (RO) su korišćene membrane za tretman slane (BVRO) i morske (SWRO) vode za izdvajanje, prethodno prečišćavanje i pre-kon-centrisanje mlečne kiseline. Kalcijum laktat, natrijum laktat i amonijum laktat su koriš-ćeni kao model napojne smeše. Radni pritisak je značajno uticao na prolazak laktata kroz prvu BVRO jedinicu, dok su Donnanov ekskluzioni efekat i vodončne veze odgo-vorni za izdvanje katjona. Kalcijumovi ioni su izdvojeni na jedinici BVRO zbog male brzine difuzije i interakcije naelektrisanja na površini. Međutim, monovalentni joni su for-mirali vodonične veze sa karbonilnim grupama membrane što je omogućilo prolaz preko membrane. Druga SVRO jedinica bila je za prethodno koncentrisanje mlečne kiseline. Visoka laktatna čistoća 99,2% sa ukupnim stepenom separacije 50,5% je postignita iz napojne smeše sa kalcijum laktatom. Niža čistoća sa višim stepenom izdvajanja laktata je postignuta iz napojne smeše sa natrijum i amonijum laktatom. Sa realnom fermenta-cionom kominom ostvareni su manji stepen separacije i čistoća u dvostepenoj RO jedi-nici. Utvrđeno je da su ukupni ioni prisutni u fermentacionoj kominbili odgovorni za nisku efikasnost dvostepene RO jedinice.

Ključne reči: mlečna kiselina; fermentaciona komina; reversna osmoza; Donnan ekskluzioni efekat; jonska jačina.




191

SVETOLIK MAKSIMOVIC1

VANJA TADIC2 JASNA IVANOVIC1

TANJA RADMANOVIC1

STOJA MILOVANOVIC1

MILICA STANKOVIC3 IRENA ZIZOVIC1

1University of Belgrade, Faculty of Technology and Metallurgy,

Belgrade, Serbia 2Institute for Medical Plant

Research “Dr Josif Pančić”, Belgrade, Serbia

3University of Niš, Faculty of Medicine, Niš, Serbia

SCIENTIFIC PAPER

UDC 66.061.3:582.998.16:615.322

UTILIZATION OF THE INTEGRATED PROCESS OF SUPERCRITICAL EXTRACTION AND IMPREGNATION FOR INCORPORATION OF Helichrysum italicum EXTRACT INTO CORN STARCH XEROGEL

Article Highlights • Starch xerogels were impregnated with H. italicum extract by scCO2 extraction-impre-

gnation process • Influence of different process parameters on impregnation yield was studied • Influence of ethanol on the extract’s chemical profile and impregnation yield was det-

ermined • Selected process conditions enabled sufficient quantity of impregnated extract Abstract

Supercritical CO2 extraction of Helichrysum italicum and impregnation of starch xerogels with the extract by using an integrated scCO2 extraction and impreg-nation process were performed at 350 bar and 40 °C in order to produce bio-materials for possible oral intake of the extract. Xerogels produced by air-drying of acetogels and alcogels were used as carriers in the supercritical impregnation process. The effect of ethanol as a co-solvent, contact time, plant material/carrier mass ratio and xerogel preparation on the impregnation loading was studied. The highest impregnation loading (1.26±0.22%) was achieved after 5 h impreg-nation of the xerogel obtained from alcogel using pure scCO2 and plant material/ /carrier mass ratio of 10. Chemical analysis of the extracts showed that the addition of ethanol as co-solvent had a positive effect on scCO2 selectivity to ter-pene fraction and total flavonoids, while it lowered the total phenolic content. Despite the difference in chemical composition, both extracts expressed similar antioxidant activity according to the DPPH and FRAP methods. The integrated process was shown to be a feasible method for isolation and incorporation of bioactive components of H. italicum into starch xerogels.

Keywords: antioxidant activity, Helichrysum italicum, supercritical ext-raction, supercritical impregnation, starch xerogels.

The genus Helichrysum consists of an estimated 600 species, in the sunflower family (Asteraceae). Helichrysum italicum is commonly known as curry plant or everlasting. It is a small aromatic shrub with yellow flowers. The stems are woody at the base and

Correspondence: S. Maksimovic, University of Belgrade, Faculty of Technology and Metallurgy, Karnegijeva 4, 11120 Belgrade, Serbia. E-mail: [email protected] Paper received: 23 February, 2017 Paper revised: 9 August, 2017 Paper accepted: 6 September, 2017

https://doi.org/10.2298/CICEQ170223031M

can reach 30-70 cm in height. It grows on dry, rocky or sandy ground around the Mediterranean [1].

Traditional use of this plant includes the applic-ation for treatment of allergies, colds, cough, skin, liver and gallbladder disorders, inflammation, infect-ions and sleeplessness [1]. Essential oils and extracts of H. italicum are reported for antimicrobial, anti-inf-lammatory, anti-viral, antioxidant and anti-larvicidal activities which is mainly due to the presence of terpenes and phenolic compounds [2].

Although H. italicum is traditionally used for treatment of digestive system disorders, only a few articles in the open literature reported use of H. itali-

S. MAKSIMOVIC et al.: UTILIZATION OF THE INTEGRATED PROCESS… Chem. Ind. Chem. Eng. Q. 24 (0) 000−000 (2018)

192

cum isolates in this area. Recently Rosa et al. [3], per-forming extraction from H. italicum by acetone, rep-orted the protective effect of the extract’s component arzanol on oxidative modification of lipid components induced by Cu2+ in human low-density lipoprotein (LDL) and the reduction of polyunsaturated fatty acids and cholesterol levels, inhibiting the increase of oxid-ative products. Previously, Facino et al. [4] tested free-radical scavenger properties of the single glyco-syl-flavonoids, isolated from the ethanolic extract of H. italicum, and in toto glycosidic fraction with in vitro systems where different reactive oxygen species were generated and on lipid peroxidation in rat liver microsomes. It was shown that in toto fraction inhi-bited superoxide ions and hydroxyl radicals to a les-ser extent than the lipid peroxidation in microsomes. Finally, Rigano et al. [5], performing extraction of H. italicum by ethanol, showed that this extract elicited antispasmodic actions in the isolated mouse ileum and inhibited transit preferentially in the inflamed gut. Biological assays on human colonic epithelial cells indicated the extract’s component 12-acetoxytreme-tone antioxidant effects, specifically by reducing reactive oxygen species [6].

Supercritical fluid technology, including the application of carbon dioxide as the main solvent, has gained wide acceptance during the past decades as an alternative to conventional processes [7]. Carbon dioxide is the most used solvent in supercritical processes because it is safe, readily available and low-cost [8]. Also, it has low critical pressure and tem-perature values (73.86 bar and 31.06 °C, respect-ively). The main advantage of supercritical CO2 (scCO2) extraction compared to conventional extract-

ion processes is the production of solvent-free and highly valued plant extracts, after removing the sol-vent by decompression. On the other hand, scCO2 can induce swelling of polymers and reduce viscosity of the polymer melt by up to an order of magnitude [9]. This and the near-zero surface tension of scCO2 facilitates incorporation of compounds soluble in scCO2 into the polymer matrix [10]. The main advent-ages of supercritical impregnation over the conventio-nal impregnation processes are the production of high-purity and solvent-free impregnated materials, better distribution of solute inside the carrier matrix during the short time compared to conventional pro-cesses and avoidance of toxic reagents and addi-tional treatment of the impregnated products.

Data on extraction from H. italicum using scCO2

were reported in few articles found in the literature [11-18]. In these studies, extraction pressure and temperature were varied in the range of 79.3-350 bar and 35.86-64.14 °C respectively, with the extraction times ranging from 1.5 h to 4 h. Obtained yields were in the range of 0.35-6.31%. It can be noted that there is a lack of data regarding the extract composition and co-solvent effects at higher pressures [19].

Starch is a natural biodegradable polymer suit-able to be a carrier for oral intake of bioactive compo-nents. Impregnation of starch-based materials with bioactive components soluble in scCO2 can be per-formed using supercritical solvent impregnation (SSI). A literature survey on SSI of starch materials for dif-ferent applications is given in Table 1.

The articles listed in Table 1 included starch of various origins: potato and Eurylon7 amylomaize [20], maize [21], corn [22-24], sorghum and rice [25], pea

Table 1. Literature survey of supercritical CO2 impregnation of starch with different active components

Form of starch Active component Pressure

bar Temperature

°C Impregnation time

h Loading,

% Application Reference

Aerogel Ibuprofen, paracetamol

180 40 70 10-22,10-25

Controlled drug delivery

[20]

Microparticles, modified with n-octenil succinate (OSA)

Lavandin essential oil

100-120 40-50 2 2.5-14.7 Biocide formulation [21]

Pregelatinized starch prepared by spray drying

Oleic acid, flax oil 150-300 40-80 8 4.89-11.86, 0.51-6.60

Food packaging [22]

Aerogel Ketoprofen 180 40 1-8 2.1-11.53 Controlled drug delivery

[23]

Aerogel Ketoprofen, Benzoic acid

180 40-55 24 12.84,21.54 Controlled drug delivery

[24]

Spherules and particles Oregano essential oil

80-150 40-50 3-24 2.5-15 Food preservation [25]

Biocomposite films Cinnamaldehyde 150-250 35 3-15 0.1-0.25 Food packaging [26]

Xero- and aerogel Thymol 155 35 24 0.58-4.01 Pharmaceuticals and nutraceuticals

[27]


193

[23], cassava [26] and tapioca [27]. Starch used in the mentioned articles showed considerable potential as a carrier for the production of controlled drug delivery systems. Furthermore, the highest impregnation load-ings were obtained when starch aerogels were imp-regnated due to their outstanding surface area and porosity achieved by supercritical drying [28]. On the other side, production of polymer aerogels requires a substantial flow of scCO2 during the drying process [29]. Starch in the form of xero- and aerogel was used as carrier for SSI with thymol [27]. The study showed that xerogels expressed high thymol loading capa-cities at given experimental conditions (Table 1), whereby the xerogels were easy to produce in a low-cost process. Finally, it can be noted that in all the presented articles SSI of starch with different active components was performed in a batch mode – the carrier and active component were exposed to scCO2 in the same vessel as a batch and after a certain time, decompression was performed.

This research was aimed to investigate the simultaneous scCO2 extraction from H. italicum and impregnation of corn starch xerogels with the extract in order to produce a suitable system for oral intake of H. italicum bioactive components. As previously men-tioned, H. italicum extract was shown to be effective in treatment of disorders in the digestive system. The main objective of this research was feasibility testing of the application of the recently introduced [30] int-egrated supercritical fluid extraction (SFE) and SSI (SFE-SSI) process in isolation of H. italicum extract and its incorporation into corn starch xerogels using scCO2. Influence of the xerogel preparation method as well as the SFE-SSI process parameters (impreg-nation time, mass of plant material to carrier mass ratio and addition of ethanol as co-solvent) on the impregnation loading was analyzed. Furthermore, the influence of ethanol as a co-solvent on the compo-sition and antioxidant activity of H. italicum extracts was studied.

EXPERIMENTAL

Materials

H. italicum flowers were collected in August 2013 in the region Konavle (the southern part of Croatia) and kept away from direct sunlight at room temperature during the drying process. Corn starch (amylose content 28%) was obtained from Starch Industry Jabuka (Serbia) in the form of white to pale yellow powder. The density of starch given by the manufacturer was 1.5 g/cm3. Commercial CO2 (purity 99%) was purchased from Messer-Tehnogas, (Ser-

bia). Commercial acetone (purity 99.7%) and 96% ethanol (purity 99.8%) were purchased from Zorka Pharma-Hemija (Serbia). 2,2-diphenyl-1-picrylhydra-zyl (DPPH, purity ≥99%), 2,4,6-tri(2-pyridyl)-s-triazine (TPTZ, purity ≥99%) and L-ascorbic acid (purity ≥99%) were purchased from Sigma-Aldrich GmbH (Germany). Folin-Ciocalteu reagent was purchased from Merck (Germany). Gallic acid (purity >98%) was obtained from TCI Europe (Belgium).

Starch xerogels preparation

Starch (10 g) was mixed with distillated water in the mass ratio of 1:10. The suspension was heated in a silicon oil bath and stirred with a magnetic stirrer (500 rpm) for 20 min at 100 °C to obtain hydrogels [27]. Hydrogels were immersed into the acetone for 5 days in the fridge at 8 °C to replace the water in the hydrogel pores. The volume of the acetone was the same as the volume of water used for the preparation of hydrogels (100 ml). The obtained acetogel films (thickness of 4.0±0.5 mm) were cut into small discs (diameter of 1 cm). The acetone was removed from the acetogels by air-drying at ambient conditions for 10 days to obtain xerogels (XG-Ac).

Xerogels were also produced from alcogels by gradual replacement of water in starch hydrogels. Hydrogels were poured into Petri dishes and warm 10% ethanol solution in water (temperature of 40 °C) was added. After cooling at ambient conditions, the hydrogel was left in the fridge at 8 °C during the night. Next day, the solvent for water replacement was removed and replaced with 20% ethanol solution and the hydrogel was returned into the fridge where it remained overnight. This procedure was repeated during the following days by using solvents with suc-cessively higher content of ethanol (30-96%) to prevent gel shrinkage (pore collapse). The obtained alcogel films (thickness of 4.0±0.5 mm) were cut into small discs (diameter of 1 cm). The ethanol was rem-oved from the alcogels by air-drying at ambient con-ditions for 10 days to obtain xerogels (XG-Alc).

ScCO2 extraction and impregnation

In order to investigate the influence of ethanol as a co-solvent on chemical composition and anti-oxidant activity of the obtained extract, scCO2 extract-ion of H. italicum was performed first, apart from the integrated SFE-SSI process. The SFE was performed at the pressure of 350 bar and temperature of 40 °C. Moderately low temperature (40 °C) is suggested to prevent thermal degradation of terpenes [31]. Higher CO2 pressure (density) is suggested for it favors ext-raction of phenolic compounds [32]. Due to the high polarity of phenolic compounds, especially flavonoids,


194

an addition of small amounts of organic co-solvents like ethanol and methanol in order to change the solvent polarity and increase the solvating power towards the desired compounds is required [32,33]. Ethanol in quantity of 10 mass% of CO2 consumed in experiment without co-solvent was added to the milled plant material. Extraction was carried out in the extraction part of the experimental unit for the com-bined SFE-SSI process described in details below. The average extraction time was 5.5 h, while the average CO2 flow rate was 0.19 kg/h. The extraction yield (y) was calculated using the following equation:

( ) e

s

100%

mym

= (1)

where me is the mass of obtained extract, while ms is the mass of plant material at beginning of the pro-cess.

The integrated SFE-SSI process was carried out in the previously described [34] high pressure extraction adsorption (HPEA) 500 unit (Eurotechnica GmbH, Germany), Figure 1.

Starch xerogel was placed in the 100 mL ads-orption column, made of stainless steel and designed to be operated at maximum pressure of 690 bar and temperature of 250 °C, while milled H. italicum flo-wers (26.6±0.3 g) were put in the 280 mL stainless steel extractor, designed to be operated at maximum pressure of 534 bar and temperature of 120 °C. In experiments with co-solvent, ethanol in a quantity of 10 mass% of CO2 in a system at operating conditions

in the experiment without co-solvent was placed in the extractor as well. Plant material to carrier mass ratio of 10 and 20 was used. Liquid CO2 was supplied from a CO2 cylinder with a siphon tube, then cooled in a cryostat to prevent vaporization and finally pumped into the system by a liquid metering pump (Milton Roy, France). After the filling of vessels with CO2 and reaching the operational conditions (350 bar and 40 °C) circulation of the solution (CO2 + extract) through both vessels (extractor and adsorption column) during the 5 or 8 h followed. A high pressure gear pump designed to be operated at maximum pressure of 500 bar and temperature of 100 °C was used for circul-ation of the supercritical solution. At the end of each experiment the system was depressurized at the rate of 35 bar/min. The impregnation loading (I) was cal-culated using the following equation:

( ) i

i

% 100m mI

m−= (2)

where mi is the mass of impregnated carrier, while m is the mass of carrier before the impregnation. All experiments were performed in duplicates.

Chromatographic analysis

Samples of H. italicum extract obtained with and without use of ethanol as a co-solvent were used to analyze the chemical composition and antioxidant activity thereof. Prior to the analysis, ethanol was removed from the H. italicum extract by rotary vacuum evaporation. Furthermore, a sample of H.

Figure 1. Schematic view of the HPEA 500 unit.


195

italicum extract incorporated in XG-Alc without co-solvent was also analyzed after its dissolving in a chloroform/methanol mixture (7:3).

Gas chromatography analysis of the extracts was carried out on a HP-5890 Series II GC apparatus (Hewlett-Packard, Waldbronn, Germany), equipped with split-splitless injector and automatic liquid sampler, attached to HP-5 column (25 m×0.32 mm, 0.52 μm fil thickness) and fitted to flame ionization detector (FID). Carrier gas flow rate (H2) was 1 mL/min, split ratio 1:30, injector temperature was 250 °C, detector temperature 300 °C, while column tem-perature was linearly programmed from 40 °C to 260 °C (at rate of 4 °C/min), and then kept isothermally at 260 °C for 10 min. Solutions of samples dissolved in chloroform/methanol mixture (7:3) were consecutively injected in amount of 1 μL. Area percent reports, obtained as result of standard processing of chroma-tograms, were used as the base for the quantification analysis.

The same analytical conditions as those men-tioned for GC/FID were employed for GC/MS ana-lysis, along with column HP-5MS (30 m×0.25 mm, 0.25μm film thickness), using HP G 1800C Series II GCD system (Hewlett-Packard, Palo Alto, CA, USA). Helium was used as a carrier gas. Transfer line was heated at 260 °C. Mass spectra were acquired in EI mode (70 eV), in m/z range 40-450. The amount of 0.2 μL of the sample solution in chloroform/methanol mixture (7:3) was injected. The components of the oil were identified by comparison of their spectra to those from Wiley 275 and NIST/NBS libraries, using different search engines. The experimental values for retention indices were determined by the use of calib-rated Automated Mass Spectral Deconvolution and Identification System Software (Amdis, ver. 2.1), com-pared to those from the available literature (Adams) [35] and used as additional tool to approve MS find-ings.

Determination of total phenolic content

The content of total phenolics in the extracts was determined by a modified Folin–Ciocalteu method [36]. The extracts were diluted in methanol to a final concentration of 1 mg/mL. 0.1 mL of extracts was shaken for 1 min with 0.5 mL of Folin-Ciocalteu reagent and 6 mL of distilled water. After the mixture was shaken, 1.5 mL of 20% Na2CO3 was added and the mixture was shaken once again for 0.5 min. Fin-ally, the solution was brought up to 10 mL by adding distilled water. After incubation for 2 h at room tempe-rature, the absorbance was measured at 750 nm using glass cuvettes against a blank (100 μL of meth-

anol instead of test samples). The total phenolic con-tent was calculated using the standard calibration curve of gallic acid (from 1 to 1.500 μg/mL). Spec-trophotometric measurements were performed by using UV-Vis spectrophotometer HP 8453 (Agilent Technologies, USA) in order to determine total phen-olic and total flavonoid contents. The analysis was performed in triplicate and the results were expressed as milligrams of gallic-acid equivalents (GAE) per g of dried extract.

Determination of total flavonoid content

Total flavonoid content was measured by means of the aluminum chloride colorimetric assay [37]. An aliquot (1 mL) of 0.02, 0.04, 0.06, 0.08, 0.10 mg/mL methanolic catechin solutions or methanolic plant ext-racts (1 mg/mL) was added into a 10 mL volumetric flask containing 4 mL of water. Then 0.3 mL of 5% NaNO2 was added and after 5 min, 0.3 mL of 10% AlCl3, was added. After 6 min, 2 mL of 1 M NaOH was added and the total volume was made up to 10 mL with water. The solution was well mixed and the abs-orbance was measured against the prepared blank at 510 nm. Total flavonoids were expressed as g of cat-echin equivalents (CE) per g of the dry extract.

Determination of antioxidant activity of the extracts

Antioxidant activity of the extracts was mea-sured on the basis of scavenging activities of the stable 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical [38]. 80 μL of extract previously diluted in methanol to obtain at least four different concentrations (0.2 to 1 mg plant material/mL), 2.92 mL of methanol and 1.0 mL of freshly prepared DPPH methanol solution (90 μmol/L) were shaken vigorously and left in the dark for 30 min at room temperature. The absorbance was measured against a blank (methanol) at 517 nm. Inhibition of the DPPH radical was calculated as a percentage (PE, %) using the following equation:

( ) control sample

control

% 100A A

PEA

−= (3)

where Acontrol is the absorbance of the control reaction (containing all reagents except the test compound), and Asample is the absorbance of the test compound. Synthetic antioxidant L-ascorbic acid was used as a positive control and all tests were carried out in tripli-cates.

The FRAP (the ferric reducing ability of plasma) assay is based on the reduction, at low pH, of a yel-low ferric complex (Fe3+-2,4,6-tri(2-pyridyl)-s-triazine, TPTZ) to a blue-colored ferrous complex (Fe2+-TPTZ) by the action of electron-donating antioxidants. The


196

reduction is monitored by measuring the change of absorbance at 593 nm. The measurement was car-ried out according to the method reported by Benzie and Strain [39]. The working FRAP reagent was pre-pared daily by mixing 10 volumes of 300 mM acetate buffer pH 3.6 (containing 6.4 mL 2 M sodium acetate solution and 93.6 mL 2 M acetic acid solution diluted in a volumetric flask (1 L)) with 1 volume of 10 Mm TPTZ (in 40 mM HCl) and with 1 volume of 20 mM ferric chloride. A standard calibration curve was con-structed using aqueous solutions of FeSO4⋅7H2O at different concentrations (0.1–1 mM). For antioxidant activity determination, 100 μL of methanol plant ext-ract (in concentration of 1mg of dry extract/mL) were mixed with 3 mL of the FRAP reagent. The absor-bance readings were started after 10 min and they were performed at 593 nm against a blank (100 μL of methanol instead of test samples). The FRAP value was calculated and expressed as mmol Fe (II) equi-valents per gram of extract (mmol Fe (II)/g) based on the standard calibration curve.

Characterization of xerogels

Morphology of the starch xerogels was inves-tigated by the field emission scanning electron micro-scopy (FE-SEM, Mira 3 XMU TESCAN a.s., Czech Republic) operated at the accelerating voltage of 10 kV. The samples of the starch xerogels were coated with a thin layer of Au/Pd (85/15), using a sputter coater (Polaron SC502, Fisons Instruments, UK) prior to the analysis.

Density and porosity of the xerogel samples were determined by the pycnometer method. The density of the samples (ρx) was calculated using Eq. (4) according to the previously described procedure [40]:

2H O 1

1 2 3x

mm m m

ρρ =

+ − (4)

where m1 is the mass of sample, m2 is the mass of pycnometer and water, m3 is the mass of glass with water and the sample. The density of water used for determination of the same for XG-Ac at 28.9±0.2 °C was 996.0±0.1 kg/m3, while the density of water used for determination of the same for XG-Alc at 27.8±0.2 °C was 996.3±0.1 kg/m3. The measurements were performed in triplicates. Porosity of the xerogels (ε) was calculated using the following equation [41]:

( )starch

% 100 1 xρερ

= −

(5)

whereby ρstarch is the density of raw starch given by the manufacturer.


ScCO2 extraction and characterization of extracts

The extraction of H. italicum with pure scCO2 resulted in an extraction yield of 2.7±0.5%. The addi-tion of ethanol in quantity of 10 mass% of the CO2 consumed in experiment without the co-solvent caused the increase of extraction yield to 5.9±0.7%. The extractions were performed in duplicate. Obtained results indicated strong ethanol’s influence on changing scCO2 solubility power.

Compounds identified in H. italicum extracts obtained with pure scCO2 (97.1%) and using co-sol-vent (93.9%), according to the GC/MS analysis can be classified in 5 groups of compounds (Table 2). Based on presented data, the addition of ethanol led to a 1.5 times increase of the overall content of terpe-nes (total terpenic content) in H. italicum extract and to the reduction of the content of fatty acid esters, aldehydes and alcohols (4 times). The presence of the co-solvent resulted in a significant increase of content of coumarin and amorphene derivatives and hydrocarbons, including waxes. Main components of the H. italicum extract obtained with pure scCO2 were terpenes – neryl acetate (1.5%), ar-curcumene (4.1%), β-selinene (4.6%) and xanthorrhyzol (2.1%), ester methyl caprylate (9.2%), hydrocarbons nonacosane (3.0%) and untriacontane (2.7%) and 3,4-dihydro- -4,4,5,7,8-penthamethyl coumarin-6-ol (5.9%). Main components of the H. italicum extract obtained using co-solvent were terpenes neryl acetate (2.0%), ar-cur-cumene (6.4%), β-selinene (4.9%) and α-selin-11-en- -4-ol (1.8%), 2-methyl-3-oxo-valeric acid methyl ester (5.2%), hydrocarbons nonacosane (5.9 %) and untria-contane (7.4%) and 2α-acetoxy-11-metoxy amorpha- -4,7-diene (7.0%) and 3,4-dihydro-4,4,5,7,8-pentha-methyl coumarin-6-ol (6.4%).

The Folin-Ciocalteu method and aluminum chlo-ride colorimetric assay were used to quantify fractions of higher molecular weight compounds in H. italicum supercritical extracts. Obtained results (Table 3) indi-cated significantly higher total phenolic content in the extract obtained with pure scCO2, compared to values reported for H. italicum essential oil (74±1.642 mg GAE/g sample [42]), ethanolic extract (31.97 1.42 mg GAE/g dry extract [43]) and aqueous extract (15.69±0.17 mg GAE/g dry extract [44]). On the other hand, according to the obtained results (Table 3), addition of ethanol caused an increase of total flavo-noid content by 20%. Kladar et al. [43] determined the


197

total flavonoid content in H. italicum ethanolic extract, being 20.68±0.66 mg QE/g dry extract, expressed as quercetin equivalents, while in this study we obtained 130.27±1.14 and 57.77±1.01 mg CE/g dry extract, expressed as catechin equivalents, for the extract obtained with pure scCO2 and extract obtained using co-solvent, respectively. Results presented in Tables 2 and 3 indicated the significant role of ethanol as a co-solvent in modification of scCO2 selectivity.

Antioxidant activity of H. italicum scCO2 extracts obtained at 350 bar and 40 °C was tested by DPPH free-radical and FRAP assays. Results given in Table 3 indicated moderate antioxidant potential of H. itali-cum extracts, which is in accordance with previously reported data [15,16]. Costa et al. [15] showed that the DPPH radical scavenging ability of the scCO2 extracts of H. italicum isolated at 120 bar and 40 °C increased in a dose-dependent manner (approximat-ely 15–60% at concentration of 0.625–5.0 mg/mL), while the extracts isolated at 90 bar and 40 °C were not able to reduce the DPPH radical at the studied concentrations. On the other side, Poli et al. [16] rep-orted scavenging effect of H. italicum extracts obtained at 260 bar and 50 °C of 10.79±0.3– –95.2±1.5% at concentration of 5–200 μg/mL, tested by DPPH method, 0.5±0.026–0.857±0.04% after 28– –56 h, assessed by the β-carotene bleaching test and none to full superoxide radical scavenging activity at concentration of 25–200 μg/mL. Addition of ethanol as a co-solvent resulted in a slight increase of antioxi-dant activity in both tests. Since the extract obtained with the addition of ethanol contained 40.8% terpenes

(1.5 times higher content than in the extract obtained with pure CO2), whereas the extract obtained with pure scCO2 contained 50.3% fatty acids, esters, alde-hydes and alcohols and 2 times higher total phenolic content than the extract obtained with addition of ethanol, it could be concluded that both types of com-pounds equally contribute to the antioxidant activity of H. italicum extracts.

Characterization of xerogels

The SEM images of dried XG-Ac and XG-Alc were given in Figure 2. Both types of xerogel pos-sessed compact and porous structures which could be due to the completed gelatinization in the applied temperature range.

The determined density of the XG-Ac and XG- -Alc samples was 1285.3±3.2 kg/m3 and 1213.9±6.8 kg/m3, respectively. Corresponding porosity of the XG-Ac and XG-Alc samples was 14.3±0.2% and 19.1±0.5%, respectively.

Integrated SFE-SSI process

Obtained impregnation loadings of H. italicum extract in the xerogels at selected conditions (350 bar and 40 °C) for different process parameters (plant to carrier mass ratio, contact time, use of co-solvent) are listed in Table 4.

In the case of XG-Ac impregnation, the influence of all mentioned parameters was tested. Addition of the co-solvent in the case of XG-Ac resulted in almost two times lower impregnation loading (decrease from 0.99±0.06 to 0.54±0.13%) despite the fact that the

Table 2. Results of GC/MS analysis of H. italicum extracts obtained at 350 bar and 40 °C

Compound group Content, %

Extract obtained with pure scCO2

Extract obtained with addition of co-solvent

Extract obtained after reextraction from impregnated XG-Alc

Terpenes 28.3 40.8 14.3

Fatty acids, esters, aldehydes and alcohols 50.3 13.3 56.6

Hydrocarbons 8.9 18.9 18.1

Coumarin and amorphene derivatives 8.1 20.9 4.5

Others 1.5 0.0 0.2

Total 97.1 93.9 93.7

Table 3. Total phenolics, flavonoids and antioxidant activities of H. italicum extracts obtained at 350 bar and 40 °C; the experimental results were expressed as mean ± standard deviation (SD) of three replicates

Sample TPa, mg GAE/g dry

extract* TFb, mg CE/g

dry extract DPPHc PEd, %

(1 mg/mL) FRAPe, mmol Fe2+/g

dry extract

Extract obtained with pure supercritical CO2 130.27±1.14 40.67±0.53 11.50±0.10 0.32±0.04

Extract obtained with addition of co-solvent 57.77±1.01 48.02±0.31 12.84±0.76 0.34±0.02 aTotal phenolics (TP) was measured by Folin-Ciocalteu method and expressed as gallic-acid equivalents (GAE); btotal flavonoids (TF) was measured by the aluminium chloride colorimetric assay expressed as catechin equivalents (CE); cDPPH radical scavenging activity; dpercentage inhibition (PE); eferric reducing ability (FRAP)


198

solubility of H. italicum extract was higher in the scCO2+EtOH system, than in pure scCO2 (according to the obtained extraction yields). This could be due to the higher affinity of H. italicum extract to the scCO2+EtOH mixture than to the starch xerogel. Inc-rease of the process time from 5 to 8 h resulted in the decrease of the impregnation loading from 0.99±0.06 to 0.83±0.05%. It could be assumed that impreg-nation of the xerogel with H. italicum extract in a particular moment reaches its maximum, after which desorption of the extract takes place [34]. Increase of the plant material/carrier mass ratio from 10 to 20, in the case of XG-Ac, resulted in the decrease of the impregnation loading from 0.99±0.06 to 0.88±0.01%. Therefore, the process time of 5 h, plant material/car-rier mass ratio of 10 and impregnation without co-sol-vent were found to be the optimal conditions with res-pect to the obtained loading in the case of XG-Ac. Based on these results, impregnation of XG-Alc was performed without the co-solvent during 5 h, whereby the plant material/carrier mass ratio was varied. Inc-reasing of the plant material/carrier mass ratio had a negative effect on the impregnation loading. As pre-sented in Table 4, the highest impregnation loading of 1.26±0.22% was achieved for the SFE-SSI of xerogel

obtained from alcogel (XG-Alc) with pure scCO2 during 5 h and with the plant material/carrier mass ratio of 10. Significantly higher impregnation loading for the SFE-SSI of XG-Alc under the same process conditions compared to XG-Ac (Table 4) indicated a positive influence of gradual water replacement during the preparation of XG-Alc. SEM analysis indi-cated that gradual water replacement positively affected porosity of the xerogels (Figure 2) which was confirmed by the determined porosity values (higher porosity was determined for the XG-Alc sample). Therefore, this could be the reason for 27% higher impregnation loading of the extract in XG-Alc (1.26± ±0.22%) in comparison to XG-Ac (0.99±0.06%) at the same SFE-SSI conditions (Table 4).

Finally, xerogels obtained in the SFE-SSI pro-cess with the co-solvent had a more intensive yellow color compared to the one obtained in the SFE-SSI process with pure scCO2. This was assumed to be due to the change of scCO2 selectivity in the pre-sence of ethanol (Table 2).

GC/MS analysis of the H. italicum extract dis-solved from the impregnated XG-Alc resulted in iden-tification of 93.7% compounds. Content of the com-pound groups is listed in Table 2. As can be seen, the

Figure 2. FE-SEM images of starch xerogels obtained by drying of: a – acetogels; b – alcogels.

Table 4. Experimental data for integrated SFE-SSI process; mcos – mass of co-solvent, mc – mass of carrier, ms/mc – plant material to carrier mass ratio, I – impregnation loading

Xerogel type Time, h mCO2, g mcos, g mc, g ms/mc I, %

XG-Ac 5 450 - 2.66 10 0.99±0.06

5 400 45 2.69 10 0.54±0.13

8 450 - 2.66 10 0.83±0.05

5 450 - 1.34 20 0.88±0.01

XG-Alc 5 450 - 2.69 10 1.26±0.22

5 450 - 1.34 20 0.71±0.12


199

carrier showed affinity towards fatty acids, esters, aldehydes and alcohols and hydrocarbons at the exp-ense of terpenes, coumarin and amorphene derivat-ives. Main components of the extract were terpenes ar-curcumene (5.6%), β-selinene (1.8%) and δ-cadi-nene (2.2%), esters methyl isomyristate (6.9%) and methyl myristate (4.0%), oleic acid (6.1%), hydrocar-bons pentacosane (4.3%), nonacosane (3.5%) and untriacontane (4.0%) and 3,4-dihydro-4,4,5,7,8-pen-thamethyl coumarin-6-ol (4.1%).

CONCLUSION

ScCO2 extraction from H. italicum and SSI of starch xerogels with the extract were performed by the integrated SFE-SSI process.

It was shown that the addition of ethanol as a co-solvent changed selectivity of scCO2 increasing total terpenic content as well as the content of cou-marin and amorphene derivatives and flavonoids. Both scCO2 extracts showed moderate antioxidant activity.

In the case of XG-Ac ethanol as a co-solvent significantly lowered the impregnation loading, which led to the conclusion that the extract possesses higher affinity to scCO2+ethanol mixture than to the xerogel. The increase of impregnation time from 5 to 8 h, as well as the plant material to carrier mass ratio caused decrease of the impregnation loading. The highest impregnation loadings were obtained in the processes with pure scCO2, during 5 h and with plant material to carrier mass ratio of 10. Significantly higher impregnation loading for XG-Alc could be attri-buted to the different porosity and morphology of the xerogels which was visible on the SEM images. Obtained results indicated feasibility of the combined SFE-SSI process application in extraction and post-erior impregnation of H. italicum extract into the starch xerogels. In further research a more thorough analysis of the influence of contact time and impreg-nation mode on the impregnation loading and chem-ical composition of the loaded extract is needed.

Acknowledgements

The financial support of the Ministry of Edu-cation, Science and Technological Development of the Republic of Serbia (Grants No. 45001 and 45017) is gratefully acknowledged.

REFERENCES

[1] D. A. Viegas, A. Palmeira-de-Oliveira, L. Salgueiro L, J. Martinez-de-Oliveira, R. Palmeira-de-Oliveira, J. Ethnopharmacol. 151 (2014) 54-65

[2] E. Guinoiseau, V. Lorenzi, A. Luciani, A. Muselli, J. Costa, J. Casanova, L. Berti, Formatex. 2 (2013) 1073- –1080

[3] A. Rosa, F. Pollastro, A. Atzeri, G. Appendino, M. P. Melis, M. Deiana, A. Incani, D. Loru, M. A. Dessi, Chem. Phys. Lipids. 164 (2011) 24-32

[4] R.M. Facino, M. Carini, L. Franzoi, O. Pirola, E. Bosisio, Pharmacol. Res. 22 (1990) 709-721

[5] D. Rigano, C. Formisano, F. Senatore, S. Piacente, E. Pagano, R. Capasso, F. Borelli, A. A. Izzo, J. Ethno-pharmacol. 150 (2013) 901-906

[6] D. Rigano, C. Formisano, E. Pagano, F. Senatore, S. Piacente, M. Masullo, R. Capasso, A. A. Izzo, F. Borelli, Fitoterapia. 99 (2014) 198-203

[7] N. Vyas, M. Y. Khan, S. Panchal, A. Butani, V. Kumar, Int. J. Pharm. Sci. Nanotech. 1 (2009) 303-308

[8] E. Reverchon, I. De Marco, J. Supercrit. Fluids 38 (2006) 146-166

[9] Y. Wang, C. Yang, D. Tomasko, Ind. Eng. Chem. Res. 41 (2002) 1780-1786

[10] I. Kikic, F. Vecchione, Curr. Opin. Solid State Mater. Sci. 7 (2003) 399-405

[11] J. Ivanovic, M. Ristic, D. Skala, J. Supercrit. Fluids. 57 (2011) 129-136

[12] S. Maksimovic, Z. Kesic, I. Lukic, S. Milovanovic, M. Ristic, D. Skala, J. Supercrit. Fluids 84 (2013) 1-12

[13] V. Micic, M. Jotanovic, Z. Lepojevic, V. Aleksic, B. Pejo-vic, JET-M. 1 (2009) 26-31

[14] B. Marongiu, A. Piras, E. Desogus, S. Porcedda, J. Essent. Oil Res. 15 (2003) 120-126

[15] P. Costa, J. M. Loureiro, M. A. Teixeira, A. E. Rodrigues, Ind. Crop. Prod. 77 (2015) 680-683

[16] F. Poli, M. Muzzoli, G. Sacchetti, G. Tassinato, R. Laz-zarin, A. Bruni, Pharm. Biol. 41 (2003) 379-383

[17] S. Jokic, M. Rajic, B. Bilic, M. Molnar, Phytochem. Anal. 27 (2016) 290-295

[18] I. Jerkovic, M. Rajic, Z. Marijanovic, M. Bilic, S. Jokic, Separ. Sci. Technol. 51 (2016) 2925-2931

[19] S. Maksimovic, V. Tadic, D. Skala, I. Zizovic, Phyto-chemistry 138 (2017) 9-28

[20] T. Mehling, I. Smirnova, U. Guenther, R. H. H. Neubert, J. Non-Cryst. Solids 355 (2009) 2472-2479

[21] S. Varona, S. Rodriguez-Rojo, A. Martin, M. J. Cocero, C. M. M. Duarte, J. Supercrit. Fluids 58 (2011) 313-319

[22] L.M. Comin, F. Temelli, M.D.A. Saldana, J. Supercrit. Fluids 61 (2012) 221-228

[23] G.A. Garcia-Gonzalez, I. Smirnova, J. Supercrit. Fluids 79 (2013) 152-158

[24] G.A. Garcia-Gonzales, M. Jin, J. Gerth, C. Alvarez-Lor-enzo, I. Smirnova, Carbohydr. Polym. 117 (2015) 797- –806

[25] A.P. Almeida, S. Rodriguez-Rojo, A.T. Serra, H. Vila- -Real, A.L. Simplicio, I. Delgadilho, S.B. da Costa, L.B. da Costa, I.D. Nogueira, C.M.M. Duarte, Innov. Food Sci. Emerg. Technol. 20 (2013) 140-145


200

[26] A.C. De Souza, A.M.A. Dias, H.C. Sousa, C.C. Tadini, Carbohydr. Polym. 102 (2014) 830-837

[27] S. Milovanovic, I. Jankovic-Castvan, J. Ivanovic, I. Zizovic, Starch. 67 (2015) 174-182.

[28] [28] G. A. Garcia-Gonzales, M. Alnaief, I. Smirnova, Carbohydr. Polym. 86 (2011) 1425-1438

[29] J. Ivanovic, S. Milovanovic, I. Zizovic, Starch 68 (2016) 821-833

[30] M.A. Fanovich, J. Ivanovic, D. Misic, M.V. Alvarez, P. Jaeger, I. Zizovic, R. Eggers, J. Supercrit. Fluids 78 (2013) 42-53

[31] T. Fornari, G. Vicente, E. Vazquez, M.R. Garcia-Risco, G. Reglero, J. Chromatogr., A 1250 (2012) 34-48

[32] M.R. Marostica Junior, A.V. Leite, N.R.V. Dragano, Open Chem. Eng. J. 4 (2010) 51-60

[33] C.G. Pereira, M.A.A. Meireles, Food Bioprocess Technol. 3 (2010) 340-372

[34] J. Ivanovic, S. Milovanovic, M. Stamenic, M.A. Fanovich, P. Jaeger, I. Zizovic, in: Handbook on supercritical fluids: fundamentals, properties and applications, E. Jane Osborne, Nova Science Publishers, New York, 2014, pp. 257-281

[35] R.P. Adams, Identification of Essential Oil Components by Gas Chromatography/Mass Spectrometry, 4th ed., Allured Publishing Corporation, Carol Stream, IL, 2007

[36] V.L. Singleton, J.A. Rossi, Am. J. Enol. Vitic. 16 (1956) 144-158

[37] J. Zhishen, T. Mengcheng, W. Jianming, Food Chem. 64 (1999) 555–559

[38] J.C. Espin, C. Soler-Rivas, H.J. Wichers, J. Agric. Food Chem. 48 (2000) 648-656

[39] I.F.F. Benzie, J.J. Strain, Anal. Biochem. 239 (1996) 70- –76

[40] Q. Xu, X. Ren, Y. Chang, J. Wang, L. Yu, K. Dean, J. Appl. Polym. Sci. 94 (2004) 593-597

[41] A. Salerno, E. Di Maio, S. Iannace, P.A. Netti, J. Super-crit. Fluids 58 (2011) 158-167

[42] H.F. Wang, Y.K. Wang, K.H. Yih, J. Cosmet. Sci. 59 (2008) 509-522

[43] N.V. Kladar, G.T. Anackov, M.M. Rat, B.U. Srdjenovic, N.N. Grujic, E.I. Sefer, B.N. Bozin, Chem. Biodiversity 12 (2015) 419-431

[44] M. Kazazic, M. Djapo, E. Ademovic, Int. J. Pure App. Biosci. 4 (2016) 85-90.

SVETOLIK MAKSIMOVIĆ1

VANJA TADIĆ2 JASNA IVANOVIC1

TANJA RADMANOVIĆ1

STOJA MILOVANOVIC1

MILICA STANKOVIC3 IRENA ZIZOVIC1

1Univerzitet u Beogradu, Tehnološko- –metalurški fakultet, Karnegijeva 4,

11120 Beograd, Srbija 2Institut za proučavanje lekovitog bilja

“Dr Josif Pančić”, Tadeuša Košćuška 1, 11000 Beograd, Srbija

3Univerzitet u Nišu, Medicinski fakultet, Bulevar Zorana Đinđića 81, 18 000 Niš,

Srbija

NAUČNI RAD

PRIMENA INTEGRISANOG PROCESA NATKRITIČNE EKSTRAKCIJE I IMPREGNACIJE ZA INKORPORACIJU EKSTRAKTA Helichrysum italicum U KSEROGEL KUKURUZNOG SKROBA

Ekstrakcija iz smilja (Helichrysum italicum) pomoću natkritičnog CO2 i impregnacija kserogelova skroba ekstraktom primenom integrisanog procesa ekstrakcije i impreg-nacije pomoću natkritičnog CO2 izvođeni su na 350 bar i 40 °C u cilju proizvodnje bio-materijala za potencijalni oralni unos ekstrakta. Kserogelovi dobijeni sušenjem na vaz-duhu acetogelova i alkogelova upotrebljeni su kao nosač u procesu natkritične impreg-nacije. Ispitani su uticaji etanola kao kosolventa, vremena kontakta, odnosa masa biljnog materijala i nosača i načina pripreme kserogela na prinos impregnacije. Najveći prinos impregnacije (1,26±0,22%) postignut je nakon 5 h impregnacije kserogelova dobijenih od alkogelova, upotrebom čistog natkritičnog CO2 i pri odnosu masa biljnog materijala i nosača od 10. Hemijska analiza ekstrakata pokazala je da je dodatak etanola kao kosolventa pozitivno uticao na selektivnost natkritičnog CO2 prema terpe-nima i ukupnom sadržaju flavonoida, dok je ukupan sadržaj fenola smanjen. Bez obzira na razliku u hemijskom sastavu, oba tipa ekstrakta su pokazala slično antioksidantno dejstvo, na osnovu primene DPPH i FRAP testova. Integrisani proces se pokazao kao pogodan metod za izolaciju i inkorporaciju bioaktivnih komponenata smilja (H. Italicum) u kserogelove skroba.

Ključne reči: antioksidantno dejstvo, Helichrysum italicum, natkritična ekstrak-cija, natkritična impregnacija, kserogelovi skroba.

7434(Online) APRIL-JUNE 2018 Vol.24, Number 2, 93-200

Documents

Transcript of 7434(Online) APRIL-JUNE 2018 Vol.24, Number 2, 93-200