Technique for rapid detection of phthalates in water and beverages

Journal of Food Engineering 116 (2013) 515–523

Contents lists available at SciVerse ScienceDirect

Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng

Technique for rapid detection of phthalates in water and beverages Asif I. Zia a,d, Mohd Syaifudin Abdul Rahman a, Subhas Chandra Mukhopadhyay a,⇑, Pak-Lam Yu a, I.H. Al-Bahadly a, Chinthaka P. Gooneratne b, Jü´rgen Kosel b, Tai-Shan Liao c a

School of Engineering and Advanced Technology, Massey University, Palmerston North, New Zealand Sensing, Magnetism and Microsystems Group, King Abdullah University of Science and Technology, Saudi Arabia c Instrument Technology Research Centre, National Applied Research Laboratories, Hsinchu, Taiwan d Department of Physics, COMSATS Institute of Information Technology, Islamabad, Pakistan b

a r t i c l e

i n f o

Article history: Received 3 October 2012 Received in revised form 18 November 2012 Accepted 18 December 2012 Available online 2 January 2013 Keywords: Interdigital sensors Phthalates DEHP Impedance spectroscopy Constant phase element Electrochemical

a b s t r a c t The teratogenic and carcinogenic effects of phthalate esters on living beings are proven in toxicology studies. These ubiquitous food and environmental pollutants pose a great danger to the human race due to their extraordinary use as a plasticizer in the consumer product industry. Contemporary detection techniques used for phthalates require a high level of skills, expensive equipment and longer analysis time than the presented technique. Presented research work introduces a real time non-invasive detection technique using a new type of silicon substrate based planar interdigital (ID) sensor fabricated on basis of thin film micro-electromechanical system (MEMS) semiconductor device fabrication technology. Electrochemical impedance spectroscopy (EIS) was used in conjunction with the fabricated sensor to detect phthalates in deionized water. Various concentrations of di(2-ethylhexyl) phthalate (DEHP) as low as 2 ppb to a higher level of 2 ppm in deionized water were detected distinctively using new planar ID sensor based EIS sensing system. Dip testing method was used to obtain the conductance and dielectric properties of the bulk samples. Parylene C polymer coating was used as a passivation layer on the surface of the fabricated sensor to reduce the influence of Faradaic currents. In addition, inherent dielectric properties of the coating enhanced the sensitivity of the capacitive type sensor. Electrochemical spectrum analysis algorithm was used to model experimentally observed impedance spectrum to deduce constant phase element (CPE) equivalent circuit to analyse the kinetic processes taking place inside the electrochemical cell. Curve fitting technique was used to extract the values of the circuit components and explain experimental results on theoretical grounds. The sensor performance was tested by adding DEHP to an energy drink at concentrations above and below the minimal risk level (MRL) limit set by the ATSDR (Agency for Toxic Substances & Disease Registry), USA. Results showed that the new sensor was able to detect different concentrations of phthalates in energy drinks. The experimental outcomes provided sufficient indication to favour the development of a low cost detection system for rapid quantification of phthalates in beverages for industrial use. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction Phthalates, chemically known as 1-2-Benzenedicarboxylic acid esters, are internationally recognized as key industrial chemicals with a large number of commercial uses, such as additives, solvents and plasticizers due to their low cost (Schettler, 2005). Phthalates with higher molecular weight i.e. DINP (Diisononyl phthalate) and DEHP constitute about 80% of the phthalate production for the reason of their use as plasticizer in almost every plastic product including food and beverage packaging and medical applications, while low molecular weight phthalates, such as diethyl ⇑ Corresponding author. E-mail addresses: [email protected] (A.I. Zia), s.c.mukhopadhyay@massey. ac.nz (S.C. Mukhopadhyay). 0260-8774/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jfoodeng.2012.12.024

phthalate (DEP) and dimethyl phthalate (DMP), are used in cosmetics, insecticides, paints and pharmaceutical applications (Heudorf et al., 2007). Phthalates added as plasticizers, do not covalently bond to the molecular structure of the resulting product, therefore, their potential for non-occupational exposure to the environment is high. They leach and migrate into packed food and beverages, gas out in the atmosphere or enter directly into human body fluids through medical products (Hauser et al., 2004). Molecular structures of DEHP and DINP are illustrated in Fig. 1. Human exposure to phthalates is a major concern for adverse human health risk. Phthalates have been characterized as environmental endocrine-disrupting compounds (EDCs) by many health monitoring agencies in the world due to their observed reproductive and developmental defects in rodents (USCPSC, 2010). It was conclude that DEHP poses highest toxicity threat to

516

A.I. Zia et al. / Journal of Food Engineering 116 (2013) 515–523

2. Existing detection techniques

Fig. 1. Molecular structure of (a) DEHP and (b) DINP.

human race, especially to children under 12 months of age and pregnant and nursing mothers (Schettler, 2005). A number of recent researches have suggested declining trend in the reproductive hormones levels in male adults (Meeker et al., 2009) and elevated risk of breast cancer in females (Lopez-Carrillo et al., 2010). Human beings become in contact with phthalates in their environment by three major routes; dermal, inhalation and oral (dietary) intake. Oral route is the most important among all as it contributes the highest exposure rates to the human beings. For example DEHP intake occurs via phthalate contaminated food, water and other liquids and through mouthing of toys and teethers in children (Swan, 2008). Due to its leach-ability several researchers reported migration of phthalates from food packaging (Lee et al., 2011), PET (PETE, polyethylene terephthalate) bottled beverages and mineral water (Montuori et al., 2008) and from corks of glass bottles (Sendón et al., 2012). World Health Organization (WHO) and European Union have limited DEHP in their water policy by setting the guideline value at 8.0 lg/L in fresh and drinking waters in a published list of priority compounds posing endocrine disrupting hazard to human (WHO, 2011). Table 1 shows risk assessment of phthalates by agencies in EU, US and Canada. In May 2011, MoH Taiwan reported illegitimate addition of phthalates by a beverage manufacturing company. DEHP was added as a clouding agent in the drink. Later, on investigation, phthalates were found in certain medicines, foods and beverages manufactured in Taiwan. DEHP has legitimately been used as plasticizers in PVC food contact packaging material, but its use as food additive has never been allowed due to its health risk. It should be noted that the tolerable daily intake (TDI) limit for DEHP in Taiwan and Hong Kong is 1.5 ppm. Long-time ingestion of DEHP with food at levels above the TDI can create hormonal imbalance in the human body. This incident created a huge wave of suspicion for all packed beverages throughout the world. Importers demanded testing certifications from the manufacturing companies which caused huge workload on the test laboratories. Due to expensive and time consuming testing manufacturers have to face production loss and pay additional cost.

Phthalates detection and measurement is purely a laboratory based procedure. The ubiquitous presence of this compound as a contaminant seriously limits its minimal detection level. Even in most controlled laboratory setup it cannot generally be accurately quantified below about 2-ppb (Cao, 2008). Gas chromatography (GC) is the most commonly used technique for detection and quantification of phthalates metabolites (Wagner and Oehlmann, 2009). High performance liquid chromatography (HPLC) is used to measure phthalate concentrations in blood plasma and urine sample at low detection limits (WHO, 2011). Chromatography technique is used to separate complex mixtures of organic compounds with each compound quantified by its specific detector. DEHP is measured using electron capture detector (ECD) (Wagner and Oehlmann, 2009) and flame ionization detector (FID) (LopezCarrillo et al., 2010). Liquid chromatography (LC) coupled with mass spectrometry (MS) or ultra violet (UV) detection are also a few commonly used techniques for detection of phthalates (Giam et al., 1975). Almost all contemporary analytical techniques used to detect phthalates in food products and beverages require laboratory environment with stringent conditions over sampling procedures. The equipment required is expensive and complicated, produces laboratory waste, needs the expertise of trained professionals and involve high testing cost in addition to a long testing time. There is a paramount requirement for a low cost real time testing system which could be used for instant screening of food and beverage products to detect the presence of phthalates that could readily be installed in an industrial setup. The discussed design and methodology of the developed sensing technique possesses properties of speed, in situ testing with an additional benefit of low cost.

3. Development of planar interdigital sensors Planar interdigital (ID) sensors, are digit-like or finger-like, parallel, in-plane electrodes patterned in a periodic fashion on a solid substrate used to observe the capacitive reactance produced as an outcome of applied alternating electric field perturbations. The capacitive reactance behaves as a function of system properties. The applied field fringes through the material to perform dielectric profiling and observe conduction properties of the semi insulating test materials. Conventional ID sensors are based on repeated patterns of one sensing and one excitation (working electrode) on a solid substrate to achieve single side access and non-destructive testing of the material under study. (Mamishev et al., 2001) have reported the evaluation profiling by using a three wavelength interdigital sensor. The penetration depth of the fringing electric field depends on the spatial wavelength (distance between sensing

Table 1 Risk assessment by ASTDR, US-EPA, EU-CSTEE, Health Canada (Hauser et al., 2004). Phthalate type

DEHP DEHP DEHP DEHP DINP DEP DBP DBP a b c

Minimal risk level. Tolerable daily intake. Reference dose levels.

Risk assessment of phthalates Country, region

Committee/year

mg/kg bodyweight/day

MRLa/TDIb/RfDc

USA USA Canada EU EU USA USA USA

US-EPA, 1993b ATSDR, 2002 Health Canada, 1994 CSTEE, 1998a,b CSTEE, 1998a,b ATSDR, 1995 ATSDR, 2001 US-EPA, 1990

0.02 0.1 0.044 0.050 0.25 7 0.5 0.1

RfD MRL TDI TDI TDI MRL MRL RfD

A.I. Zia et al. / Journal of Food Engineering 116 (2013) 515–523

517

and working electrode) of the interdigital structures (Mamishev et al., 2004). In biosensing applications where only a few lL of the sample is required to be pipetted on the sensing area make conventional ID sensors a good candidate as biosensors; therefore, ID sensors have most conveniently been used for the detection of food borne pathogens. (Radke and Alocilja, 2005; Rahman et al., 2012; Wang et al., 2012). The dielectric behaviour of nematic liquid crystal film was reported using ID capacitive sensors (Abu-Abed and Lindquist, 2008). Fig. 2a displays the layout of conventional ID sensor in comparison to the layout of the sensor used in the presented research (Fig. 2b). 3.1. Sensor design and simulation New types of planar interdigital (ID) structures were designed, modelled, simulated and fabricated with a greater number of sensing electrodes compared to the number of working electrodes. More sensing electrodes between two working electrodes enhance the magnitude of penetration depth parameter for the fringing electric field in the sensor. This parameter provides an edge to the new sensors over the conventional ID sensors in dielectric and impedance profiling of the bulk sample. New sensors proved to differ from the conventional interdigital sensors with respect to size, design, performance and sensitivity. The new types of sensors were designed using finite element modelling software COMSOL MultiphysicsÒ employing AC/DC module in 3D workspace quasi-static mode. Two sensors were designed by keeping the substrate material and thickness, electrode material and thickness and effective sensing area at constant values. Design geometry was drawn in two configurations; (a) five sensing electrodes between two working electrodes and pitch length (distance between two adjacent electrodes) set at 50 lm, coded as (1-5-50) and (b) eleven sensing electrodes between two working electrodes with pitch length set at 50 lm, coded as (1-11-50). The pitch length for both the sensors was kept at a constant value of 50 lm. Sensing area of each sensor was fixed to 2.5 mm  2.5 mm, electrode area 25 lm  2.425 mm and electrode thickness to 520 nm was taken as standard for both designs. Fig. 2b shows the layout measurements for the new type of sensor. A single crystal silicon substrate with a thickness 525 lm and a relative permittivity 4.2 was modelled for the sensor design as a base material in COMSOL so as to realize the optimized design for a 4 inch single polished side P type silicon wafer commercially available in semiconductor fabrication facility. In order to simulate the design for the experimental conditions, the sensing surface was sequentially exposed to a bulk medium having relative permittivity of 80.1 and 5.1. These values

Fig. 2a. Conventional interdigital sensor.

Fig. 2b. New design Si base ID sensor (1-11-50).

correspond to the relative permittivity of water and DEHP. The simulation results provided valuable estimations of capacitance, stored electrical energy, field intensity and penetration depth of the fringing electric field lines. Fig. 3a shows the image of electric field simulation for 1-5-50. The average penetration depth of electric field for deionized water MilliQ is calculated to be 212.5 lm. Fig. 3b shows the electric field simulation image for 1-11-50 with an average penetration depth of 437.6 lm. Table 2 shows the simulated value for capacitance and electrostatic energy density (J) using relative permittivity 80.1 (MilliQ) and 5.1 (DEHP). The value for electrostatic energy density We is calculated by the software using the equation:

We ¼

Q2 2C

ð1Þ

where Q is the charge stored and C is the capacitance. The software calculates electrostatic energy density by integrating scalar product of electric displacement (D) and electric field intensity (E) over the domain as given below:

Fig. 3a. Electric field penetration depth (1-5-50).

518

A.I. Zia et al. / Journal of Food Engineering 116 (2013) 515–523

in presence of redox probe in addition to its inherent dielectric properties making it a better choice for capacitive sensing. After development of coated photoresist on top of polymer the bonding pads were opened using plasma etching process. Fig. 4 shows (a) sensors in wafer form and (b) individually against scale. 4. Materials and methods 4.1. Electrochemical impedance spectroscopy

Fig. 3b. Electric field penetration depth (1-11-50).

Table 2 Calculated capacitance and energy density values using COMSOL MultiphysicsÒ modelling simulation. Sensor type

Vo (V)

1-11-50 1-5-50

We ¼

Z

10 10

Freq. (kHz)

10 10

Electrostatic energy density (J)

Capacitance (pF)

MilliQ er ¼ 80:1

DEHP er ¼ 5:1

MilliQ er ¼ 80:1

DEHP er ¼ 5:1

6.9e12 4.2e12

7.6e13 4.9e13

13.94 8.46

1.52 0.92

ðD  EÞdX

ð2Þ

X

C¼

Q2 C 2 DV 2 2W e ¼ )C¼ 2W e 2W e DV 2

ð3Þ

V is the applied voltage to the working electrode. 3.2. Sensor fabrication The simulation results by finite element modelling depicted better performance capability of 1-11-50 in comparison to 1-550 intimating the fabrication of 1-11-50. The sensor was fabricated by photolithography and etching techniques on a single crystal Silicon/Silicon Dioxide (Si/SiO2) 4 inch (diameter) wafer, 525 lm thick. 36 workable sensors were patterned on one wafer with each sensor having dimension of 10 mm by 10 mm and sensing area of 6.25 mm2. The patterns were written for periodic interdigital structure with 11 sensing between two excitation electrodes for pitch length of 50 lm with the width of electrodes and the sensing area set to values of 25 lm and 2.5 mm  2.5 mm respectively. The sensors were fabricated using MEMS technology involving the steps of photo resist coating, UV exposed ID pattern transfer, plasma etching, metal deposition by DC magnetron sputtering and lift off. Gold was used as electrode material due to inertness and flexibility in the methods available for its deposition as thin film electrodes. 500 nm of Gold (Au) were sputtered on top of 20 nm Chromium (Cr) to provide proper adhesion of the ID structures on the substrate. After lift-off, the wafer was coated with a 1 lm layer of parylene C by a conformal parylene coater. Parylene C is a polymeric coating and provided a uniform and pinhole free layer of polymer on the sensing area which can withstand continuous exposure to air at high temperatures for long periods of time. It avoids corrosion of electrodes by chemical reaction during testing

Electrochemical impedance spectroscopy (EIS) is a versatile and popular technique used to describe the response of an electrochemical process taking place at the electrode solution interface consequent to a low amplitude sinusoidal perturbation as a function of frequency applied through a capacitive probe. This technique has been reported for meat inspection (Mukhopadhyay and Gooneratne, 2007), to estimate the dielectric properties of milk (Mukhopadhyay et al., 2006), detection of contaminated seafood with marine bio-toxin (Mohd Syaifudin et al., 2009), evaluate electrical properties of food (Li et al., 2011) and impedance characteristics of dried pulp (Wu et al., 2008). Other than real time testing one of the most promising qualities of impedance spectroscopy is labelfree detection for biological and chemical analysis. The impedance Z of a system is determined by applying a small signal perturbation as a function of frequency and analysing the resultant current through the system in terms of amplitude and phase shift compared to the voltage-time function. The complex value of impedance can thus be translated in terms of its real and imaginary values at different frequencies by obtaining phase shift between applied and received signals. The results of an impedance measurement can be graphically demonstrated using bode and Nyquist (Cole–Cole) plot for all applied frequencies with Real part of impedance Z plotted along X-axis and Imaginary part plotted along Y-axis in the later. The consequent impedance spectrum thus obtained, allows the characterization of surface, layers and concentrations in addition to the exchange and diffusion processes describing the system kinetics. This is achieved by analysing the impedance spectrum on basis of an equivalent circuit commonly consisting of series and/or parallel combination of resistances and capacitances representing the different electrochemical and physiochemical properties of the system under analysis (Lisdat and Schäfer, 2008). The most frequently cited equivalent circuit used to interpret EIS experimental findings in electrical model form is Randles equivalent circuit as shown in Fig. 5a. It shows uncompensated solution resistance Rs in series to a parallel combination of double layer capacitance Cdl to the charge transfer resistance Rct in series with Warburg impedance Zw (Lisdat and Schäfer, 2008). The Nyquist plot (Cole–Cole plot) for the equivalent circuit comprises a semi-circular region followed by a 45° straight line as shown in Fig. 5b. The straight line represents a faster mass-transfer limited process at lower frequencies whereas the semi-circular portion describes a relatively slower charge transfer limited process at higher frequencies. Cdl can be calculated from the frequency at the maximum of semi-circular region in the Nyquist plot using x = 2pf = 1/ RctCdl whereas, Rct is calculated by extrapolating the semicircle to Zreal axis as shown in Fig. 2b. The double layer capacitance Cdl and charge transfer resistance Rct are the key electrical parameters in determination of impedance change for analytical studies of the system kinetics in detection of the phthalates. In-depth study of electrochemical impedance spectroscopy technique could be found in (Barsoukov and Macdonald, 2005; Wang, 2006). 4.2. Sample preparation The objective of this research is to test the detection response of the fabricated sensor for the two phthalate esters; di(2-ethylhexyl)

A.I. Zia et al. / Journal of Food Engineering 116 (2013) 515–523

519

Fig. 4. (a) Sensors in wafer form. (b) individual sensor with testing jig shown against scale.

Fig. 5. (a) Randles equivalent circuit. (b) Nyquist plot for the equivalent circuit.

phthalate (DEHP) and diisononyl phthalate (DINP). Gravimetrically prepared 99.5% pure solutions of DEHP and DINP at a concentration of 100 lg/mL (100 ppm) in ethanol were procured from ChemServiceÒ USA. Two set of working solutions of these compounds were prepared by serial dilution method. The first set was prepared with deionized water MilliQÒ (MILLIPORE USA) at concentration of 0.002 through 2 ppm concentration levels of DEHP. The second set of samples was prepared at a concentration level of 0.1 through 20 ppm of DINP in ethanol. 20 ppm concentration from this set was used as stock solution. 99.7% pure research grade ethanol was used to serial dilute the stock to achieve lower concentrations. pH of control and working solutions were tested using IQ Scientific Instrument Inc. USA after calibration with buffer solutions. Hioki 3522-50 LCR Hi precision tester (Japan) was used for the EIS experiments. The equipment was interfaced with a desktop computer through RS-232C hardware interface device. Automatic data acquisition software programmed in Lab viewÒ was used to generate Microsoft excelÒ data file in real time. The LCR tester was calibrated with built-in open circuit and short circuit tests in order to offset any stray capacitance appearing due to the testing leads. All experiments were performed using slow mode of testing to achieve error rate of <0.05%. For reliable results the device was set to write an average of 8 readings of impedance values at each single frequency applied to the sensor. A constant voltage 1Vrms sinusoidal signal was applied to the sensor with a frequency sweep of 1 Hz to 100 kHz with 20 data points per decade on log scale. An average of three experiments was used to further ensure reproducibility and reliability of the results. The sensor was cleaned with acetone, rinsed with deionized water and dried before next test. It was observed experimentally that soldering the sensor’s connection pads for bonding connecting wires instantly degrades the sensors detection performance. A special jig with gold coated pins was

developed to avoid soldering the electrodes and make good resistance free contact with the connection pads of the sensor. Fig. 6 shows (a) the test bench setup and (b) sensor testing the sample using dip testing method.

5. Results and discussions 5.1. DEHP detection in MilliQ Deionized Milli-Q water (18 MX cm) with 2 ppm ethanol (99.7%) was used as control solution. The pH of control was 7.3 at 23 °C in the laboratory environment. 2 ppm concentration of DEHP was used as stock solution to obtain the lower concentrations of 0.2 through 0.002 ppm by serial dilution. The pH of the stock solution was measured to be 6.95 at 23 °C. Experiments were conducted immediately after sample preparation at 23 °C with a humidity level of 47% in laboratory environment. Fig. 7 shows the plot for imaginary and real part of impedance vs. frequency for all the four concentrations of DEHP. The capacitive reactance (Zimag) shows a good variation with changing concentrations of DEHP especially at lower frequency range (10–400 Hz). At higher frequencies (500 Hz to 5 kHz) the change in impedance occurs at very low rate. This change is described in terms of sensitivity in the analysis discussed in later part of this paper. The corresponding change in Zreal is not dominant and its rate of change is much less in comparison to its other counterpart; capacitive reactance (Zimag). This shows the increase in capacitance of the sensor with increase in DEHP concentration which is due to the inherent dielectric properties of the target DEHP molecule. The real part represents the static resistance of the solution appearing due to the faradaic currents passing through the solution. The coated polymer stops

520

A.I. Zia et al. / Journal of Food Engineering 116 (2013) 515–523

Fig. 6. (a) Test bench setup. (b) sample under test by dip testing method.

Fig. 8. Nyquist plot.

Fig. 7. Imaginary and real part of impedance vs. frequency.

the ion exchange through the electrode surface. The Nyquist plot shown in Fig. 8 demonstrates: (1) the mass transfer process in form of a straight line at 45° which corresponds to Warburg resistance and (2) charge transfer process by semi-circular region of different diameters corresponding to the DEHP concentrations in the test solutions. The diameter of the semicircle dictates the value of charge transfer resistance Rct. Correspondingly, Rct changes the double layer capacitance Cdl at the sensor-solution interface which shows the ability of the sensor to monitor the variation of DEHP as low as 0.002 ppm in deionized water. 5.1.1. Data analyses using non-linear least square curve fitting Using Randles cell model, the real and imaginary impedance at the electrode solution interface could be derived as shown by the following equation:

ZðxÞ ¼ Rs þ

Rct 2 R2 C 2 ct dl

1þx



jxR2ct C dl 1 þ x2 R2ct C 2dl

ð4Þ

The sensitivity of the sensor is evaluated on the basis of capacitive reactance which is dominant over change in the real part of the measured impedance spectra. Eq. (5) gives the relation leading to calculate the percentage sensitivity of the sensor with reference to the control solution.

Sensitivity ð%Þ ¼

Zimag ðcontrolÞ  Zimag ðsampleÞ  100 Zimag ðcontrolÞ

ð5Þ

The calculated values of sensitivity (%) plotted in Fig. 9 depict that the sensor remains sensitive to DEHP concentration in MilliQ during a fair range of frequency i.e. 10–1400 Hz, and its sensitivity declines to low values above 5 kHz frequency range. Fig. 9 accounts for the calculated percentage sensitivity values using Eq. (5) plotted for different concentrations of DEHP in the test solutions at 200, 790, 1400 and 5000 Hz frequencies. Fig. 10 shows bode plot demonstrating the behaviour of phase angle and absolute impedance against frequency. The equivalent circuit was estimated using theoretical calculations by electrochemical spectrum analyser algorithm. The circuit parameters were estimated by non-linear least square fitting technique which fits the measured impedance data on theoretically

A.I. Zia et al. / Journal of Food Engineering 116 (2013) 515–523

521

estimated values. The algorithm performs statistical analysis to calculate the residual mean square r 2amplitude for experimentally observed values in measured spectra against the calculated values based on the theoretical response of suggested equivalent circuit using the following relation.

r 2amplitude ¼

N X ðZ 0

iobs

i¼1

 Z 0icalc Þ2 þ ðZ 00iobs  Z 00icalc Þ2 002 Z 02 iobs þ Z iobs

ð6Þ

Fig. 11a. Equivalent circuit proposed by least square curve fitting.

where Z 0iobs gives observed real impedance, Z 0icalc gives calculated real impedance, Z 00iobs gives observed imaginary impedance, Z 00icalc gives calculated imaginary impedance. r 2amplitude determines the deviation of the experimentally observed data from the optimal solution. The optimization of the calculated data is achieved by the number of iterations it takes to solve the mathematical model for the proposed equivalent circuit. The value of r2amplitude in the range of 104 shows optimal fitting with an error rate of less than 5% in calculating the values of equivalent circuit components. Fig. 11a shows the equivalent circuit interpreting the kinetics of the electrochemical model obtained in consequence of non-linear least square fitting. Fig. 11b depicts

Fig. 11b. Least square curve fitting plot for absolute value of impedance. Solid line shows calculated curve, markers show experimental data points.

Fig. 9. % Sensitivity vs. DEHP concentration. Fig. 12a. Least square curve fitting plot for imaginary value of impedance. Solid line shows calculated curve, markers show experimental data points.

Fig. 10. Bode plot for DEHP concentrations in MilliQ.

Fig. 12b. Least square curve fitting plot for phase shift. Solid line shows calculated curve, markers show experimental data points.

522

A.I. Zia et al. / Journal of Food Engineering 116 (2013) 515–523

Table 3 Results for the equivalent circuit components’ parameters, error %ages, residual mean square values obtained from non-linear least square fitting for DEHP concentrations from 0.002 ppm to 2 ppm in 1–100 kHz frequency range. Equivalent circuit parameters

Units

DEHP concentrations in deionized water 0.002 ppm DEHP

Cdl Cad Rct Zw Rs r2amplitude Iterations

F F X X X

0.02 ppm DEHP

0.2 ppm DEHP

2 ppm DEHP

Value

Error (%)

Value

Error (%)

Value

Error (%)

Value

Error (%)

1.5842E9 5.8891E9 91,152 2.1872E+6 2889.9 0.00075375

0.32204 2.7213 0.5703 1.6126 2.8226

1.7E9 1.326E8 64,578 1.7024E+6 2845.7 0.0009797

0.93117 4.057 1.2952 3.4118 4.4782

2.3134E9 1.2799E8 42,339 5.1313E+5 2681.6 0.0003502

0.42549 2.4528 0.41634 2.6064 1.9857

1.003E8 1.148E8 48,805 2.915E+5 2021.1 0.0002925

0.52117 4.0362 0.68871 4.4795 3.1013

300

300

Fig. 13a. Real and imaginary part of the impedance spectra for different concentrations of DEHP in energy drink plotted against frequency.

300

300

residual mean square values obtained from non-linear least square fitting for DEHP concentrations from 0.002 ppm to 2 ppm in 1–100 kHz frequency range. The analyses show that the double layer capacitance Cdl increases from 1.5 pF to 10 pF with increasing concentration of DEHP which verifies the inherent dielectric properties of DEHP molecule present in the sample solutions with rising concentrations. The charge transfer resistance Rct decreases from 91.1 kX to 48.8 kX. Same kind of decrease is observed in Warburg impedance Zw which reduces from 2.1 MX to 0.29 MX with increasing molecular density of DEHP in the sample solutions. The solution resistance Rs, which depends on the ionic concentration in the sample solution, also falls from 2.8 kX to 2 kX value indicating increase in conductivity of the sample solutions with increasing concentration of DEHP. Another parameter Cad was observed to build up with increasing concentration of DEHP in samples from 5.8 nF to 11.4 nF which is attributed to the formation of layer of DEHP on gold electrode surface. It is caused by the diffusion of ions from the electrode surface into the solution as a result of transfer of charges (electrons) into the electrode from the bulk solution (Kerner and Pajkossy, 2002). 5.2. DEHP detection in energy drink

Fig. 13b. Bode plot for the measured impedance spectra for different concentrations of DEHP in energy drinks.

the fitting plot for absolute value of Z. Fig. 12a shows curve fitting for Nyquist and Fig. 12b displays the curve fitting for phase shift. It should be noted that the solid line in all curve fitting plots sketches the theoretically calculated curves and the markers display the plot of experimentally observed corresponding values. Table 3 presents the equivalent circuit components’ parameters, error %ages, and

Due to strict control and stringent testing procedures, it was assumed that the beverages available in the local New Zealand markets are phthalate free. The sensor’s response was tested with added DEHP in one of the locally available energy drink, branded ‘‘Lift Plus Extra’’. This drink is supplied in glass bottle. PET bottled drink was not selected for testing due to the possibility of leached phthalate in the drink. Three working solutions with 2, 6 and 10 ppm concentrations of DEHP were prepared at 23 °C temperature and were tested immediately after sample preparation. These concentration levels were selected to note the detection response of the sensor below and above the MRL limit set by ATSDR, USA. Uncontaminated original drink was used as control for reference. Fig. 13a shows the plot for real and imaginary part of the impedance spectra for the contaminated energy drink as compared to the uncontaminated drink. The sensor displays change of impedance in the capacitive reactance dominating the corresponding real impedance (Zreal) change through a frequency range of 10 Hz to 1.3 kHz. Fig. 13b displays the bode plot for the same frequency range. It has been observed that the change in absolute impedance Zabs is visible only at lower frequencies whereas the phase shift is distinctive for pure and contaminated samples especially for DEHP concentration as high as 10 ppm. 6. Conclusion Non-invasive, real time detection technique for ubiquitous environmental endocrine-disrupting compounds in water and

A.I. Zia et al. / Journal of Food Engineering 116 (2013) 515–523

energy drink has been reported. Optimization of design for MEMS based interdigital sensors has been achieved on basis of mathematical modelling using finite element modelling software COMSOL MultiphysicsÒ. Optimized design of sensor was fabricated using thin film semiconductor device fabrication techniques. Sensor was able to detect phthalate ester, DEHP, in deionized water. Analyses of the results have been carried using non-linear least square curve fitting statistical technique to discuss the kinetic processes in the electrochemical cell on basis of Randles cell model. Constant phase element (CPE) equivalent circuit was suggested by analyses of the experimental data. The experimental data matched the theoretical model at an error rate of less than 5%. Sensitivity of the sensor at different frequencies was discussed with respect to the change in capacitive reactance observed experimentally and modelled theoretically. Sensor’s performance has been evaluated for commercially available drink which proved its phthalates detection ability in such products. The promising outcome of this research paves way to the future work in order to improve sensor’s sensitivity and induce selectivity to the present design. Further experiments are being conducted to study the effects of other parameters on sensors detection characteristics. Quantification of the detected phthalate pollutant with selectivity induction is the major challenge in the road map for commercialization of this rapid assay technique.

Acknowledgements The authors would like to thank Massey University, New Zealand, for providing the best possible research facilities. The authors are obliged to COMSATS Institute of Information Technology and Higher Education Commission Pakistan, for providing support and funds to work on this project. Special thanks to all researches referenced throughout the paper whose valuable research has guided the way through to this research work, and to all whom that had fruitful discussions and collaborations with the authors.

References Abu-Abed, A.S., Lindquist, R.G., 2008. Capacitive interdigital sensor with inhomogeneous nematic liquid crystal film. Progress in Electromagnetics Research B 7, 75–87. Barsoukov, E., Macdonald, J.R., 2005. Impedance Spectroscopy: Theory, Experiment, and Applications. Wiley-Interscience. Cao, X.L., 2008. Determination of phthalates and adipate in bottled water by headspace solid-phase microextraction and gas chromatography/mass spectrometry. Journal of Chromatography A 1178 (1–2), 231–238. Giam, C., Chan, H., Neff, G., 1975. Rapid and inexpensive method for detection of polychlorinated biphenyls and phthalates in air. Analytical Chemistry 47 (13), 2319–2320. Hauser, R., Duty, S., Godfrey-Bailey, L., Calafat, A.M., 2004. Medications as a source of human exposure to phthalates. Environmental Health Perspectives 112 (6), 751.

523

Heudorf, U., Mersch-Sundermann, V., Angerer, J., 2007. Phthalates: toxicology and exposure. International Journal of Hygiene and Environmental Health 210 (5), 623–634. Kerner, Z., Pajkossy, T., 2002. Measurement of adsorption rates of anions on Au (1 1 1) electrodes by impedance spectroscopy. Electrochimica Acta 47 (13), 2055–2063. Lee, M.R., Lai, F.Y., Dou, J., Lin, K.L., Chung, L.W., 2011. Determination of trace leaching phthalate esters in water and urine from plastic containers by solidphase microextraction and gas chromatography–mass spectrometry. Analytical Letters 44 (4), 676–686. Li, X., Toyoda, K., Ihara, I., 2011. Coagulation process of soymilk characterized by electrical impedance spectroscopy. Journal of Food Engineering 105 (3), 563– 568. Lisdat, F., Schäfer, D., 2008. The use of electrochemical impedance spectroscopy for biosensing. Analytical and Bioanalytical Chemistry 391 (5), 1555–1567. Lopez-Carrillo, L., Hernandez-Ramirez, R.U., Calafat, A.M., Torres-Sanchez, L., Galvan-Portillo, M., Needham, L.L., Ruiz-Ramos, R., Cebrian, M.E., 2010. Exposure to phthalates and breast cancer risk in Northern Mexico. Environmental Health Perspectives 118 (4), 539–544. Mamishev, A.V., Du, Y., Bau, J.H., Lesieutre, B.C., Zahn, M., 2001. Evaluation of diffusion-driven material property profiles using three-wavelength interdigital sensor. IEEE Transactions on Dielectrics and Electrical Insulation 8 (5), 785–798. Mamishev, A.V., Sundara-Rajan, K., Fumin, Y., Yanqing, D., Zahn, M., 2004. Interdigital sensors and transducers. Proceedings of the IEEE 92 (5), 808–845. Meeker, J.D., Calafat, A.M., Hauser, R., 2009. Urinary metabolites of di(2-ethylhexyl) phthalate are associated with decreased steroid hormone levels in adult men. Journal of Andrology 30 (3), 287. Mohd Syaifudin, A., Jayasundera, K., Mukhopadhyay, S., 2009. A low cost novel sensing system for detection of dangerous marine biotoxins in seafood. Sensors and Actuators B: Chemical 137 (1), 67–75. Montuori, P., Jover, E., Morgantini, M., Bayona, J.M., Triassi, M., 2008. Assessing human exposure to phthalic acid and phthalate esters from mineral water stored in polyethylene terephthalate and glass bottles. Food Additives and Contaminants 25 (4), 511–518. Mukhopadhyay, S.C., Gooneratne, C.P., 2007. A novel planar-type biosensor for noninvasive meat inspection. Sensors Journal, IEEE 7 (9), 1340–1346. Mukhopadhyay, S.C., Gooneratne, C.P., Gupta, G.S., Demidenko, S.N., 2006. A lowcost sensing system for quality monitoring of dairy products. IEEE Transactions on Instrumentation and Measurement 55 (4), 1331–1338. Radke, S.M., Alocilja, E.C., 2005. A microfabricated biosensor for detecting foodborne bioterrorism agents. Sensors Journal, IEEE 5 (4), 744–750. Rahman, M.S.A., Mukhopadhyay, S.C., Yu, P.L., Goicoechea, J., Matias, I.R., Gooneratne, C.P., Kosel, J., 2012. Detection of bacterial endotoxin in food: new planar interdigital sensors based approach. Journal of Food Engineering. Schettler, T., 2005. Human exposure to phthalates via consumer products. International Journal of Andrology 29 (1), 134–139. Sendón, R., Sanches-Silva, A., Bustos, J., Martín, P., Martínez, N., Cirugeda, M.E., 2012. Detection of migration of phthalates from agglomerated cork stoppers using HPLC–MS/MS. Journal of Separation Science 35 (10–11), 1319–1326. Swan, S.H., 2008. Environmental phthalate exposure in relation to reproductive outcomes and other health endpoints in humans. Environmental Research 108 (2), 177–184. USCPSC, 2010. Toxicity Review of DEHP. Bethesda, MD, USA. Wagner, M., Oehlmann, J., 2009. Endocrine disruptors in bottled mineral water: total estrogenic burden and migration from plastic bottles. Environmental Science and Pollution Research 16 (3), 278–286. Wang, J., 2006. Analytical Electrochemistry. Wiley-VCH. Wang, Y.X., Ye, Z.Z., Ying, Y.B., 2012. New trends in impedimetric biosensors for the detection of foodborne pathogenic bacteria. Sensors 12 (3), 3449–3471. WHO, 2011. WHO | Guidelines for Drinking-Water Quality, fourth ed. World Health Organization. Wu, L., Ogawa, Y., Tagawa, A., 2008. Electrical impedance spectroscopy analysis of eggplant pulp and effects of drying and freezing-thawing treatments on its impedance characteristics. Journal of Food Engineering 87 (2), 274–280.