- Email: [email protected]
A capacitive technique to assess water content in extra virgin olive oils
Journal of Food Engineering 116 (2013) 246–252
Contents lists available at SciVerse ScienceDirect
Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng
A capacitive technique to assess water content in extra virgin olive oils Luigi Ragni a, Eleonora Iaccheri a, Chiara Cevoli a, Annachiara Berardinelli a,⇑, Alessandra Bendini b, Tullia Gallina Toschi b a b
Department of Agricultural Economics and Engeneering, University of Bologna, P.zza Goidanich 60, 47521 Cesena (FC), Italy Department of Food Science, University of Bologna, P.zza Goidanich 60, 47521 Cesena (FC), Italy
a r t i c l e
i n f o
Article history: Received 25 June 2012 Received in revised form 17 September 2012 Accepted 17 October 2012 Available online 27 October 2012 Keywords: Extra virgin olive oil Water content Fatty acid composition Capacitance
a b s t r a c t The present research investigated the correlations between capacitance and water content of extra virgin olive oils (EVOO). A commercial capacitor probe for radio applications and an LCR meter were used for electric tests in the frequency range from 500 Hz to 512 kHz. Seventeen samples of different EVOO with a moisture content ranging from 178 to 1321 mg/kg oil were selected for study. To assess the influence of moisture only, the oil with the maximum water content was filtered down to 288 mg/kg oil and five samples with intermediate water contents were prepared and submitted to electrical measurements. Subsequently, the capacitance of all 17 EVOO samples was measured at selected frequencies. Water content and capacitance for filtered oil were linearly correlated, showing R2 values up to 0.959 and a root mean square error (RMSE) of 48 mg/kg oil at 2 kHz. When the main oil composition together with the moisture changed, the trend was always linear, but the correlation (R2) decreased to 0.818 with an RMSE value of 123 mg/kg oil (at 8 kHz), so that the system remains suitable for screening of water content. Strong correspondences between capacitance and impedance suggest that these measurements can be carried out with very simple instruments for measuring voltage and current. Some novel linear correlations also emerged between water content, capacitance and fatty acid composition. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction The water content of commercial extra virgin olive oils (EVOO) usually ranges from 0.3 to 2.0 g of water/kg of oil. The type of production process, primarily the extraction and filtration procedures, affects the moisture content in EVOO (Cerretani et al., 2010; Lozano-Sánchez et al., 2010). Water in oil is evenly dispersed as micro drops (Petrakis, 2006). The dispersion, in the form of fine or micro emulsions, can be stabilized by aggregation and dissolution of polar substances such as salts, free acids, diglycerides, phospholipids, alcohols and phenols (Cerretani et al., 2008). Moisture and volatile matter in EVOO should not be higher than 0.2% (kg/kg), which is the limit recommended by the ‘‘International Olive Oil Council’’ (IOOC, 2009) and the ‘‘Codex Alimentarius’’ (CODEX STAN, 1981). The water content, which is indirectly correlated with some taste characteristics, such as pungency and bitterness, can also play a role in stability and preservation of quality during storage (Lercker et al., 1994; Fregapane et al., 2006; Ambrosone et al., 2007). The most commonly accepted methods for measurement of water content in EVOO are Karl Fischer titration (AOAC, 1998) and drying by heating (both moisture and volatile content) (ISO ⇑ Corresponding author. Tel.: +39 0547 338113; fax: +39 0547 382348. E-mail address: [email protected] (A. Berardinelli). 0260-8774/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jfoodeng.2012.10.031
662, 1998). These methods are time-consuming, and usually require large amounts of reagents. Spectroscopic techniques have been studied to assess quality parameters, such as chemical composition, and oil adulteration. The most widely used are near infrared (NIR) (Armenta et al., 2010) and Fourier transform infrared (FT-IR) spectroscopy (Vlachos et al., 2006; Cerretani et al., 2010), often in association with chemiometric statistical analysis such as partial least square regression (PLS). NIR techniques for the determination of the water content have been successfully tested on olive fruits (Jimenez et al., 2000), olive pomace (Muik et al., 2004) and in olive oils in production processes (Bendini et al., 2007), conditions where the water content is much higher than in commercial EVOO. According to Bendini et al. (2007), R2 values of 0.913 (RMSE = 1.29%) was obtained for the prediction of moisture content in olive oil samples characterized by a moisture content ranging from 45% to 73%. Several commercial oil analyzers based on NIR technology have been used to assess the moisture content in olive fruit, paste and pomace (Armenta et al., 2010). Water content in commercial olive oils ranging from 290 to 1402 mg/kg oil was predicted by FT-IR attenuated total reflectance (ATR) spectroscopy with a coefficient of determination of 0.89 (Cerretani et al., 2010). A best performance was obtained by 31 P-NMR spectroscopy in olive oils where the moisture content (ranging from 204 to 769 mg/kg oil) was predicted with an R2 value up to 0.98 (Hatzakis and Dais, 2008). Similar results (R2 = 0.984)
247
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
were achieved by using the time domain reflectometry (TDR) on samples of EVOO with water content ranging from 714 to 2008 mg of water/kg of oil (Ragni et al., 2012). Terahertz timedomain spectroscopy (THz-TDS) has also been used to investigate the water content in synthetic oils (Gorenflo et al., 2006). The goodness of prediction, expressed by the relative root mean square error (RMSE), was up to 24 mg/kg oil in samples where the water content ranged from 430 to 3280 mg/kg oil. NMR, FT-IR, TDR and THz-TDS are expensive techniques. Spectroscopy often requires time and a high level of competence for data manipulation. Capacitive methods based on parallel plates or other types of capacitive sensors have been previously explored to predict some properties in several kinds of foods: water in wheat (Berbert and Stenning, 1996), water and bulk density in safflower (Sacilik et al., 2007), ripening of banana fruits (Soltani, 2011) and freshness parameters of shell eggs (Ragni et al., 2006, 2008). Herein, we explore a simple capacitive technique to asses the water content in EVOO that uses a discontinued commercial variable capacitor as a probe, together with basic statistical analysis. This technique was firstly tested for moisture content determination in the same commercial EVOO, but with different water content obtained by filtration. For this reason, it could be also of interest to assess the water impurities in mineral-based insulating oils, if the oil composition remains constant (Farooq, 1996; Itahashi et al., 1993, 1994; Baird et al., 2006; Liland et al., 2008). Several different EVOO samples, characterized by a moisture content ranging from 178 to 1321 mg/kg oil, were also used to set up a prediction tool for oils with different chemical composition. In fact, based on the previous results of Rudan-Tasik and Klofutar (1999) and Lizhi et al. (2008), the dielectric properties of edible oils can be affected by the composition of fatty acids, and in particular by the unsaturation level (by the water remaining after filtration). Correlation between moisture content, capacitance and fatty acids were assessed. The proposed technique could represent a valuable instrument to monitor the filtration process or to perform an approximate classification of the water content in commercial EVOO, where the fatty acids composition greatly varies from oil to oil.
content. The water content of the prepared five samples was then determined according to the standard ISO 662, Method B (1998). 2.1.2. Measurements on different commercial oils Seventeen samples of different commercial oils were selected to build a dataset of moisture content ranging from 178 to 1321 mg/kg oil. The water content was again determined by the standard the ISO 662, method B. The chemical characterization of the main oil constituents was conducted by determining the fatty acid percentage. For this determination, alkaline esterification producing fatty acids methyl esters and followed by gas chromatography (Bendini et al., 2006) were used. Measurements were carried out in triplicate. The peak retention time was compared to those of the GLC 463 FAME standard mixture (Nu-Chek, Elysian, MN) to identify the peaks. Fatty acid were grouped according to their degree of unsaturation in the three conventional classes: saturated (SFA), monounsaturated (MUFA) and polyunsaturated (PUFA). 2.2. Electrical measurements Capacitance measurements on oils were made by using an instrumental chain consisting of a variable capacitor as a probe together with a LCR meter (LCR-8101G, GW-Instek, Good Will Instrument Co. Ltd, Taiwan) (Fig. 1). The capacitor has a maximum rated capacitance of 470 pF in air; the fixed and mobile armatures consist of 12 and 10 plates, respectively (Fig. 1). The capacitance can be manually changed by the rotation of the axis which involves modification in the total plate surface of the capacitor. The probe was inserted in an oil container made of PET with the following dimensions: 82 mm (length), 77 mm (width) and 55 mm (high). The container housing the capacitor was filled with 150 g of oil in such a way to fully cover the plate armatures of the capacitor. The container with capacitor and oil was placed in a metal box to shield the system from radio electromagnetic interference. In order to avoid the retention of bubbles during the oil filling, measurements were made after the rotation of the plates (vacuum application could be taken in the account as method to remove air bubbles for system improving). Rotation of the mobile plate armature with respect to the fixed one also allowed to easily remove the oil between subsequent tests by washing the probe with
(LCR meter)
Personal computer
2. Materials and methods 2.1. Samples preparation and oil composition
Top view 2.1.1. Measurements on filtered oil mixtures To assess the correlation between the amount of water and the measured capacitance excluding the influence of the variation of fatty acid composition among oils, the sample with the maximum moisture content among the EVOO collected for the measurements in the following sub-section 2.1.2 (1321 mg/kg oil) was gently filtered by insufflation of argon gas (12 l/min) for 3 h by using the patented method number WO 2009/107096 A2 (Cerretani et al., 2009), and subsequently filtrated by gravity with paper sheets (Filter papers 55 mm, Whatman) to obtain a minimum content of water. The amount of water was determined in triplicate by drying 10 g of oil in an oven according to the standard ISO 662, Method B (1998). This mass loss method overestimates the water content, including the volatile compounds, but the error should be considered minor and acceptable since they do not exceed 50 mg/kg oil in EVOO (Angerosa et al., 2004). Five samples with different percentages of non-filtered and filtered oils were prepared to obtain oils with intermediate water
Closed probe
Metallic box PET oil and capacitor probe container
Opened probe
Rotating plates
Fig. 1. Instrumental set up with details of the capacitor probe.
248
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
n-hexane. Measurements of capacitance at 21 ± 1 °C were carried out at frequencies of 500 Hz, 2 kHz, 8 kHz, 32 kHz, 128 kHz and 512 kHz for the five oil samples with different moisture contents obtained by filtration. Each measurement was the average of 50 subsequent measurements; the averaging process required about 20 s. Capacitance measurements were always conducted in triplicate. In all tests, the voltage was set to 100 mV. To consider some limits related to the behaviour of the system and the consequent oil dielectric constant determination, frequency sweep measurements were carried out from 100 Hz to 10 kHz (steps 49.5 Hz) with the capacitor empty and filled with the oil having the maximum water content. We know, in fact, that the capacitance, C, in Farads, of a parallel plate capacitor can be calculated with the following formula:
C ¼ er e0
A d
ð1Þ
where er is the relative dielectric constant of the material, e0 is the absolute dielectric constant (dielectric constant of vacuum, 8.85 1012 F/m), A is the area of the plates, in m2, d is the distance between the plates, in m. If area and distance do not change, er can be roughly calculated by dividing the capacitance with the material by the capacitance with air. This is a reasonable method to determine the static capacitance with direct current, but it becomes less appropriate with alternating current, because the real capacitor and all the instrumental chain have a behaviour that is strictly frequency dependent. Some measurements of impedance with air and oil were also made to assess which part of it can be explained by the capacitive reactance. The impedance Z, which is a frequency dependent parameter, contains three different contributors, namely ohmic resistance, reactive capacitance Xc and reactive inductance Xl. The reactive capacitance can be calculated by:
Xc ¼
1 2pfC
ð2Þ
where f is the frequency, in Hz, and C, is the capacitance, in F. In a real capacitor Xc explains most part of the impedance Z. In an electrical circuit, the impedance is related to the voltage, V, and the current, I, according to the formula:
Z¼
V I
ð3Þ
where V is in V, and I is in A. Thus, the capacitance can be calculated by a simple measurement of voltage and current. The frequency response of an industrial capacitor with rated capacitance of 1 nF was also measured for comparison.
3. Results and discussion 3.1. Water content and EVOO composition The filtration procedure reduced the water content from 1321 mg/kg oil to 288 mg/kg oil. The five samples obtained with different percentage of non-filtered and filtered oils were characterized by moisture contents of 1321 ± 31, 920 ± 120, 753 ± 24, 513 ± 7.5 and 288 ± 54 mg/kg oil. The water content of the 17 EVOO samples is shown in Table 1. It ranged from 178 mg/kg to 1321 mg/kg oil. The minimum and maximum difference between subsequent water content in different EVOO samples were 10 mg/kg oil and 266 mg/kg oil, respectively. The fatty acid composition (Table 2) reveals differences between SFA, MUFA and PUFA of the different samples up to 66.58%, 83.19% and 23.48%, respectively. The fatty acid percentages in Table 2 are in accordance with literature data for EVOO category, being the contents of single fatty acids within the legal limits for EVOO (Anonymous, 2011). 3.2. Correlation of capacitance and impedance with oil composition 3.2.1. Filtered oil mixtures The results of the linear regression analysis conducted between capacitance and water content of the five oil samples characterized by the same fatty acid composition containing moisture in the range from 288 to 1321 mg/kg oil, at the explored frequencies (between 500 Hz and 512 kHz), are summarized in Table 3. A general linear trend between capacitance and water emerged for all frequencies. The R2 values were between 0.781 and 0.962. The root mean square error (RMSEC) ranged from 71 mg/kg oil to 170 mg/ kg oil. The coefficient of variation appeared negligible between replications and was not higher than 0.21%. The frequencies of 2 kHz and 8 kHz were selected to measure the capacitance of all EVOO samples. This choice was made by considering the results obtained for the filtered oil samples (Table 3). High R2, high differences of capacitance (measured in the water
Table 1 Water content of EVOO samples.
2.3. Statistical analysis The correlation between capacitance, water and fatty acids content were investigated by using simple regression analysis. Calibration and cross validation (leave-one-out method) models were built. The coefficient of determination and the p-level values were reported. To assess the prediction power of the models the values of the root mean square error in calibration (RMSEC) was obtained. The RMSE was calculated according to the following equation:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn 2 i¼1 ðxo;1 xp;i Þ RMSE ¼ n
The two frequencies which give the best results in term of coefficient of determination and maximum variation of capacitance between samples of filtered oils were then used for the subsequent measurement on the 17 EVOO to build a simple statistical model for moisture content prediction.
ð4Þ
where xo,1 are the observed values, xp,i are the predicted values, and n is number of values. Cross validation was performed only for the models built with the seventeen EVOO samples to predict the water content by using the capacitance values.
Sample
Water content (mg/kg of oil)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
178 (8) 294 (53) 560 (3) 741 (13) 760 (29) 815 (36) 832 (18) 898 (57) 908 (7) 921 (52) 965 (31) 981 (52) 1003 (70) 1036 (26) 1113 (13) 1249 (14) 1321 (99)
Values in parentheses are standard deviations.
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
range) and low frequencies, where the cable connection have a minor influence, were preferred. The impedance and the capacitive reactance calculated by Eq. (2) for the probe without and with oil (moisture = 1321 mg/kg) are reported for different frequencies in Table 4. From these data, it can be noted that Xc is enough to well describe the impedance with a ratio Z/Xc ranging from 0.99997 to 1.00018.
18.69 68.75 12.57
Values in parentheses are standard deviations. SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids.
15.18 78.12 6.69 15.04 78.83 6.13 SFA MUFA PUFA
15.97 77.83 6.23
14.19 80.09 5.71
14.42 82.63 2.95
14.25 79.35 6.39
16.94 73.07 9.99
16.02 74.63 9.35
12.44 80.19 7.16
15.81 73.10 11.10
15.05 77.30 7.64
16.02 75.13 8.85
15.93 77.11 6.96
15.04 79.59 5.26
15.01 79.94 4.90
15.72 75.30 8.41
(0.29) (0.02) (0.03) (0.02) (0.00) (0.00) (0.50) (0.04) (0.03) (0.03) (0.00) (0.01) (0.01) (0.10) 17
(0.04) 15.22 (0.06) 0.13 (0.02) 1.10 (0.01) 0.13 (0.05) 0.19 (0.07) 2.52 (0.13) 64.25 (0.03) 2.81 (0.06) 11.85 (0.04) 0.40 (0.05) 0.72 (0.02) 0.26 (0.02) 0.10 (0.00) 0.31
16
(0.06) 12.48 (0.01) 0.08 (0.46) 0.80 (0.02) 0.01 (0.04) 0.05 (0.00) 2.67 (0.73) 71.24 (0.17) 2.79 (0.02) 7.67 (0.01) 0.43 (0.03) 0.73 (0.00) 0.34 (0.00) 0.13 (0.00) 0.00 (0.07) 11.13 (0.01) 0.10 (0.03) 0.54 (0.01) 0.02 (0.03) 0.05 (0.01) 3.56 (0.15) 75.72 (0.08) 3.28 (0.05) 4.25 (0.01) 0.29 (0.02) 0.65 (0.05) 0.25 (0.00) 0.00 (0.00) 0.00
15 14
(0.22) 11.17 (0.02) 0.10 (0.04) 0.86 (0.00) 0.04 (0.00) 0.07 (0.01) 3.56 (0.54) 75.20 (0.09) 3.08 (0.03) 4.61 (0.04) 0.27 (0.00) 0.64 (0.01) 0.28 (0.05) 0.00 (0.04) 0.00 (0.81) 12.17 (0.02) 0.12 (0.12) 1.03 (0.00) 0.07 (0.01) 0.12 (0.05) 3.04 (0.79) 72.96 (0.10) 2.62 (0.06) 6.25 (0.08) 0.41 (0.02) 0.71 (0.03) 0.26 (0.05) 0.17 (0.01) 0.09
13 12
(0.02) 12.57 (0.02) 0.14 (0.01) 1.11 (0.01) 0.08 (0.00) 0.16 (0.05) 2.51 (0.07) 70.58 (0.01) 2.84 (0.02) 8.21 (0.01) 0.44 (0.00) 0.64 (0.01) 0.30 (0.05) 0.15 (0.00) 0.28 (1.03) 10.98 (0.02) 0.14 (0.14) 0.73 (0.01) 0.10 (0.00) 0.18 (0.06) 3.08 (1.22) 73.54 (0.19) 2.41 (0.06) 6.90 (0.02) 0.44 (0.04) 0.74 (0.05) 0.31 (0.04) 0.11 (0.00) 0.34
11 10
(0.12) 12.43 (0.02) 0.12 (0.01) 1.18 (0.00) 0.08 (0.01) 0.11 (0.05) 2.69 (0.53) 68.81 (0.27) 2.59 (0.04) 10.39 (0.01) 0.47 (0.01) 0.70 (0.29) 0.29 (0.00) 0.14 (0.00) 0.00 (0.21) 9.73 (0.02) 0.08 (0.04) 0.35 (0.00) 0.03 (0.00) 0.04 (0.02) 2.39 (0.43) 77.20 (0.15) 2.31 (0.00) 6.54 (0.02) 0.29 (0.02) 0.62 (0.03) 0.20 (0.01) 0.00 (0.00) 0.00
9 8
(0.03) 12.61 (0.00) 0.13 (0.00) 1.08 (0.02) 0.08 (0.01) 0.12 (0.02) 2.76 (0.09) 70.45 (0.00) 2.55 (0.04) 8.62 (0.01) 0.44 (0.01) 0.73 (0.00) 0.30 (0.01) 0.14 (0.03) 0.00
7
(1.16) 11.67 (0.01) 0.12 (0.07) 0.89 (0.00) 0.06 (0.01) 0.10 (0.06) 2.86 (1.47) 74.36 (0.33) 2.40 (0.00) 6.02 (0.04) 0.39 (0.01) 0.67 (0.05) 0.26 (0.04) 0.13 (0.08) 0.07
6
(0.04) 11.07 (0.01) 0.09 (0.02) 0.83 (0.01) 0.05 (0.02) 0.10 (0.03) 3.21 (0.04) 75.17 (0.03) 2.37 (0.01) 5.45 (0.01) 0.45 (0.02) 0.68 (0.04) 0.26 (0.01) 0.13 (0.00) 0.13
5
(0.16) 14.10 (0.02) 0.15 (0.01) 1.31 (0.01) 0.09 (0.01) 0.17 (0.05) 2.24 (0.18) 67.96 (0.09) 3.17 (0.01) 9.30 (0.03) 0.38 (0.02) 0.69 (0.01) 0.31 (0.00) 0.13 (0.03) 0.00 (0.40) 10.49 (0.00) 0.14 (0.00) 0.70 (0.00) 0.06 (0.02) 0.13 (0.12) 3.08 (1.77) 75.85 (0.14) 2.25 (2.41) 5.71 (0.03) 0.40 (0.01) 0.68 (0.01) 0.28 (0.02) 0.11 (0.04) 0.11
4 3
(0.20) 10.41 (0.00) 0.12 (0.01) 0.74 (0.02) 0.05 (0.03) 0.08 (0.01) 3.37 (0.15) 79.11 (0.05) 2.32 (0.02) 2.33 (0.02) 0.40 (0.00) 0.62 (0.03) 0.26 (0.00) 0.13 (0.00) 0.06 (0.06) 11.08 (0.00) 0.12 (0.01) 0.69 (0.00) 0.03 (0.01) 0.07 (0.02) 2.63 (0.19) 76.11 (0.16) 2.78 (0.02) 5.06 (0.02) 0.45 (0.21) 0.65 (0.01) 0.32 (0.05) 0.00 (0.00) 0.00
2 1
C16:0 11.79 C16:l n-9 0.12 C16:l n-7 0.94 C17:0 0.05 C17:l 0.09 C18:0 3.60 C18:l n-9 73.39 C18:l n-7 3.05 C18:2 5.68 C20:0 0.29 C18:3 n-3 0.55 C20:l 0.25 C22:0 0.24 C24:0 0.00
EVOO
Fatty acids
Table 2 Fatty acid composition of EVOO samples. Values are in percentage (%) of the total fatty acid composition.
249
3.2.2. Different commercial oils The correlations between the capacitance and the water content of all the oils for the above-mentioned frequencies are shown in Figs. 2 and 3. Predicted and observed values for the cross validation are shown in Figs. 4 and 5. As expected, these results show lower coefficients of determination compared to those obtained for the filtered oils. R2 values of 0.804 and 0.818 were obtained for frequencies of 2 kHz and 8 kHz, respectively. The RMSEC was 127 mg/kg oil and 123 mg/kg oil, instead the RMSECV was 142 and 139 mg/kg for the two frequencies respectively. The different composition of the 17 oils involves an unavoidable ‘noise’ that perturbs the electrical information related to the water content. The relative dielectric constant of some SFA (saturated fatty acids), MUFA (monounsaturated fatty acids) and PUFA (polyunsaturated fatty acids) were measured using the method of Lizhi et al. (2008) in the range from 100 Hz to 1 MHz, and it was roughly invariant up to a frequency of 100 kHz. The values of e r ranged from about 2.3 for palmitic and stearic acid (75 °C) to 2.5 for linolenic and caprilic acids (25 °C), oleic and linoleic acids (25 °C) assuming intermediate values. Although the e r of water is more than 30 times higher than that of the fatty acid, the mass variations in the oil composition between SFA, MUFA and PUFA were higher than that due to the moisture content. Moreover, other components (not measured in this investigation), such as diglycerides, can play a role in the determination of the dielectric constant of the oil. A significant, although weak linear correlation between MUFA and water content was found for the examined oils (R2 = 0.277, p-level = 0.037), one outlier removed (Fig. 6). At present, in the light of these results, we can only suggest that the tendency to retain moisture in commercial EVOO oil may be related to the balance of different types of fatty acids in the lipid matrix. A linear correlation (R2 = 0.627) was also obtained between capacitance and MUFA (Fig. 7). Despite this, the use of SFA, MUFA and PUFA in multiple linear regressions models or other simple functions did not involve an increase in power prediction of the water content. 3.3. Limit for dielectric constant calculation The calculation of the relative dielectric constant through the ratio between capacitance with oil and without oil, at the different frequencies, cannot be conducted with this probe. The frequency response of the probe appeared to be affected by several regions where the capacitance suddenly drops to lower values compared to that which should characterise an ideal capacitor. Fig. 8 shows this behaviour in the restricted range from 100 Hz to 10 kHz. The frequency response of an industrial capacitor with rated capacitance of 1 nF is also shown in the graph for comparison. Capacitance steps of the capacitor probe appear at different frequencies with and without oil, so that the calculation of the dielectric constant needed to describe very small differences between oils with different water contents, at different frequencies, failed. The above-mentioned behaviour can be attributed to the fact that the capacitor probe is a complex component, characterised by a large metallic armature, very far from an ideal capacitor. Antenna effects, reflections and resonances can occur at some frequencies with a dielectric (air) and at other frequencies with another dielectric (oil). These phenomena largely deviates the probe behav-
250
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
Table 3 Analysis of the correlations between capacitance and water content (from 288 to 1321 mg/kg) in the frequency range from 500 Hz to 512 kHz. Frequency (kHz)
Function
R2
DCHI-LO (pF)
DC%
RMSEC (mg/kg of oil)
0.5 2 8 32 128 512
W = 118.38C – 159,057 W = 118.51C – 158,678 W = 142.49C – 190,205 W = 171.63C – 229,882 W = 247.72C – 331,600 W = 147.07C – 197,111
0.895 0.959 0.962 0.959 0.781 0.784
8.350 8.390 7.085 6.030 3.047 6.710
0.605 0.599 0.536 0.450 0.227 0.500
111 48 118 73 170 169
W, water content (mg/kg oil); C, capacitance (pF); DCHI-LO, difference between the highest and the lowest capacitance measured values; DC%, ratio between DCHI-LO and the lowest capacitance value 100; RMSE, root mean square error.
Table 4 Impedance and reactive capacitance of the probe in air and with oil with a water content of 1321 mg/kg oil (n.12, in Tables 1 and 2), at different frequencies. Frequency (Hz)
Air
500 2000 8000 32,000 128,000 512,000
Oil
Z (X)
Xc (X)
Z (X)
Xc (X)
693,130 173,412 43,389 10,854 2716 678
693,151 173,413 43,389 10,854 2716 678
235,237 58,908 14,806 3691 926 230
235,228 58,926 14,806 3691 926 230
Z/Xc air
Z/Xc oil
0.99997 0.99999 0.99999 1.00000 0.99994 1.00018
1.00004 0.99970 1.00002 1.00001 1.00005 0.99997
Z, impedance; Xc, capacitive reactance; for the Z/Xc calculation, six significant digits of precision for Z and Xc representation were used.
1300
2
1500
R = 0.804 RMSEC = 127 mg/kg oil
Predicted water content (mg/kg oil)
Water content (mg/kg oil)
1500
1100 900 700 500 300
2
R = 0.756 RMSECV = 142 mg/kg oil
1300 1100 900 700 500 300 100
100 1333
1338
1343
1348
1353
100
300
Capacitance (pF)
1500
Predicted water content (mg/kg oil)
2
Water content (mg/kg oil)
900
1100
1300
1500
Fig. 4. Predict and observed values for the water content at 2 kHz (cross validation).
1500 R = 0.818 RMSEC = 123 mg/kgoil
1100 900 700 500 300 100 1335
700
Observed water content (mg/kg oil)
Fig. 2. Correlation between capacitance and water content for all oils at 2 kHz.
1300
500
2
R = 0.767 RMSECV = 139 mg/kg oil
1300 1100 900 700 500 300 100
1338
1341
1344
1347
1350
Capacitance (pF) Fig. 3. Correlation between capacitance and water content for all oils at 8 kHz.
100
300
500
700
900
1100
1300
1500
Observed water content (mg/kg oil) Fig. 5. Predict and observed values for the water content at 8 kHz (cross validation).
251
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252 2
R = 0.277
Water content (mg/kg oil)
1500 1300 1100 900 700
4. Conclusions
500 300 100 67
71
75
79
83
MUFA (%) Fig. 6. Correlation between monounsaturated fatty acids (MUFA) and water content for all oils.
1354
Capacitance (pF)
iour from that of a smaller, more compact industrial capacitor (Fig. 8). Thus, we could only calculate an approximate value of the relative dielectric constant of oil, which resulted in the explored frequency range from 2.933 to 2.942. Moreover, at a certain frequency, the dielectric constant variation due to the different moisture content is the same of the capacitance variation and assumed values between 0.227% and 0.605% (for filtered samples).
2
R = 0.627
1349
1344
1339
Capacitive techniques based on multiple parallel plate capacitors can be taken into consideration for the prediction of extremely low water content in filtered EVOO. A water content from 0.03% to 0.13% (kg/kg) of an EVOO can be predicted by using the capacitance value in a simple linear regression with an R2 values ranging from 0.781 to 0.962 for some explored frequencies in the kHz range with root mean square error up to 48 mg/kg oil for the best model. Correlations for EVOO where the chemical composition greatly varies (fatty acid, etc.) was lower than that of the oil where only the moisture changed (R2 = 0.818, at the frequency of 8 kHz). For this reason, this system can be used for water determination in EVOO if a precision expressed as root mean square error not higher than around 142 mg/kg oil is requested. A maximum of uncertainty of 50 mg/kg oil as to be added to take into account possible contribution of volatile compounds in the model constructions. A significant linear correlation (p-level = 0.037) was obtained between monounsaturated fatty acid composition and water content, and a quite good correlation (R2 = 0.627) was seen between the capacitance value and the monounsaturated fatty acid percentage in commercial EVOO oils, behaviours that deserve further investigation. Measurements of impedance demonstrated that it is well described only by the capacitive reactance, so that a simple instrumental system able to measure voltage and current can be set up and tested for the prediction of some chemical characteristics of oils.
1334 67
71
75
79
83
Acknowledgements
MUFA (%) Fig. 7. Correlation between capacitance and monounsaturated fatty acids (MUFA) for all oils at 2 kHz.
The authors thank Dr. Enrico Valli and Mr. Anton Kuti for their collaboration in laboratory activities addressed to the oils characterization.
Olive Oil 1355
Air
I.C.
460
960
459.5
950
459
940
458.5
930
458
920
Olive Oil
1345
Air 1340 1335
Capacitance (pF)
Capacitance (pF)
1350
1330
Industrial Capacitor (I.C.) 1325
0
2000
4000 6000 Frequency (Hz)
8000
10000
Fig. 8. Capacitance for the probe in air and oil with a water content of 1321 mg/kg oil (n. 12, in Tables 1 and 2), together with an industrial capacitor with a rated capacitance of 1 nF, in the range 100 Hz to 10 kHz.
252
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
The research was carried out with the financial contribution of the Italian Ministry of the Instruction, University and Research in the framework of the PRIN project ‘‘Innovative technologic applications to improve the extraction yield of virgin oils from olives and seeds and for rapid controls of their quality’’. References Ambrosone, L., Mosca, M., Ceglie, A., 2007. Impact of edible surfactants on the oxidation of olive oil in water-in-oil emulsions. Food Hydrocolloids 21, 1163– 1171. Angerosa, F., Servili, M., Selvaggini, R., Taticchi, A., Esposto, S., Montedoro, G.F., 2004. Volatile compounds in virgin olive oil: occurrence and their relationship with the quality. Journal of Chromatography A 1054, 17–31. Anonymous, 2011. On the characteristics of olive oil and olive pomace oil and the relevant methods of analysis. EC Commission Regulation (EEC) No. 61/2011 of 24 January 2011 amending Regulation No. 2568/91, Official Journal European Community 23, 1–14. AOAC Official Method 984.20, 1998. Moisture in Oils and Fats. Karl Fischer Method. Official Methods of Analysis of AOAC International, vol. I, 16th ed. AOAC, Gaithersburg, MD, pp. 41–42 (Chapter 41). Armenta, S., Moros, J., Garrigues, S., de la Guardia, M., 2010. Determination of olive oil parameters by near infrared spectrometry. In: Olives and Olive Oil in Health and Disease Prevention. Elsevier (Chapter 58). Baird, P.J., Herman, H., Stevens, G.C., Jarman, P.N., 2006. Spectroscopic measurement and analysis of water and oil in transformer insulating paper. IEEE Transactions on Dielectrics and Electrical Insulation 13, 293–308. Bendini, A., Cerretani, L., Vecchi, S., Carrasco-Pancorbo, A., Lercker, G., 2006. Protective effects of extra virgin olive oil phenolics on oxidative stability in the presence or absence of copper ions. Journal of Agricultural and Food Chemistry 54, 4880–4887. Bendini, A., Cerretani, L., Di Virgilio, F., Belloni, P., Lercker, G., Gallina Toschi, T., 2007. In-process monitoring in industrial olive mill by means of FT-NIR. European Journal of Lipid Science and Technology 109, 498–504. Berbert, P.A., Stenning, B.C., 1996. On-line moisture content measurement of wheat. Journal of Agricultural Engineering Research 65, 287–296. Cerretani, L., Bendini, A., Barbieri, S., Lercker, G., 2008. Osservazioni preliminari riguardo alla variazione di alcune caratteristiche chimiche di oli vergini da olive sottoposti a processi di deodorazione ‘‘soft’’. La Rivista Italiana delle Sostanze Grasse 85, 75–82. Cerretani L., Rocculi P., Bendini A., Romani S., Bacci A., 2009. Oil Clarifying Process and Apparatus for Implementing the Process. International Publication Number WO 2009/107096 A2. Cerretani, L., Giuliani, A., Maggio, R.M., Bendini, A., Gallina Toschi, T., Cichelli, A., 2010a. Rapid FTIR determination of water, phenolics and antioxidant. European Journal of Lipid Science and Technology 112, 1150–1157. Codex Standard for Olive Oils and Olive Pomace Oils, 1981. Codex Stan, p. 33. Farooq, K., 1996. The effect of particulate and water contamination on the dielectric strength of insulating oils. In: Proceeding of the IEEE International Symposium on Electrical Insulation, Millbrae, CA. Fregapane, G., Lavelli, V., León, S., Kapuralin, J., Salvador, M.D., 2006. Effect of filtration on virgin olive oil stability during storage. European Journal of Lipid Science and Technology 108, 134–142. Gorenflo, S., Tauer, U., Hinkov, I., Lambrecht, A., Buchner, R., Helm, H., 2006. Dielectric properties of oil–water complexes using terahertz transmission spectroscopy. Chemical Physics Letters 421, 494–498.
Hatzakis, E., Dais, P., 2008. Determination of water content in olive oil by 13P NMR spectroscopy. Journal of Agricultural and Food Chemistry 56, 1866–1872. International Olive Oil Council, 2009. Trade Standard Applying to Olive Oils and Olive-Pomace Oils. COI/T.15/NC No 3/Rev. (4 November). ISO 662, 1998. Animal and Vegetable Fats and Oils – Determination of Moisture and Volatile Matter Content, second ed. Itahashi, S., Ueta, T., Sakurai, H., Mitsui, H., Sone, M., 1993. Effect of water dimer in insulating oil on conduction phenomena. In Proceedings of the IEEE Conference on Electrical Insulation and Dielectric Phenomena, Pocono Manor, PA. Itahashi, S., Sakurai, H., Iwasawa, T., Mitsui, H., Sato, T., Sone, M., 1994. Effect of water in insulating oil on conduction phenomena under high electrical field. In: Proceedings of the IEEE Conference on Electrical Insulation and Dielectric Phenomena, Arlington, TX. Jimenez, A., Izquierdo, E., Rodriguez, F., Duenas, J.I., Tortosa, C., 2000. Determination of fat and moisture in olives by near-infrared reflectance spectroscopy. Grasas y Aceites 51, 311–315. Lercker, G., Frega, N., Bocci, F., Servidio, G., 1994. Veiled extravirgin olive oils: dispersion response related to oil quality. Journal of the American Oil Chemists’ Society 71, 657–658. Liland, K.B., Eidnes, K., Bjorneklett, K., Hvidsten, S., 2008. Measurement of solubility and water content of insulating oils for HV XLPE cable terminations. In: Proceedings of the IEEE International Symposium on Electrical Insulation. Vancouver, Canada. Lizhi, H., Toyoda, K., Ihara, I., 2008. Dielectric properties of edible oils and fatty acids as a functionof frequency, temperature, moisture and composition. Journal of Food Engineering 88, 151–158. Lozano-Sánchez, J., Cerretani, L., Bendini, A., Segura-Carretero, A., FernándezGutiérrez, A., 2010. Filtration process of extra virgin olive oil: effect on minor components, oxidative stability and sensorial and physicochemical characteristics. Trends in Food Science and Technology 21, 201–211. Muik, B., Lendl, B., Molina-Dìaz, A., Pérez-Villarejo., L., Ayora-Canãda, M.J., 2004. Determination of oil and water content in olive pomace using near infrared and Raman spectrometry. A comparative study. Analytical and Bioanalytical Chemistry 379, 35–41. Petrakis, C., 2006. Olive oil extraction. In: Boskou, D. (Ed.), Olive Oil Chemistry and Technology. AOCS Publishing, Champaign, Illinois, pp. 191–223. Ragni, L., Gradari, P., Berardinelli, A., Giunchi, A., Guarnieri, A., 2006. Predicting Quality Parameters of shell eggs using a simple technique based on the dielectric properties. Biosystems Engineering 94, 255–262. Ragni, L., Berardinelli, A., Guarnieri, A., 2008. A dielectric technique based on a onechip network analyser to predict the quality indices of shell eggs. Biosystems Engineering 100, 470–478. Ragni, L., Berardinelli, A., Cevoli, C., Valli, E., 2012. Assessment of the water content in extra virgin olive oils by Time Domain Reflectometry (TDR) and Partial Least Squares (PLS) regression methods. Journal of Food Engineering 111, 66–72. Rudan-Tasic, D., Klofutar, C., 1999. Characteristics of vegetable oils of some Slovene manufactures. Acta Chimica Slovenica 46, 511–521. Sacilik, K., Tarimci, C., Colak, A., 2007. Moisture content and bulk density dependence of dielectric properties of safflower seed in the radio frequency range. Journal of Food Engineering 78, 1111–1116. Soltani, M., Alimardani, R., Omid, M., 2011. Evaluating banana ripening status from measuring dielectric properties. Journal of Food Engineering 105, 625–631. Vlachos, N., Skopelitis, Y., Psaroudaki, M., Konstantinidou, V., Chatzilazarou, A., Tegou, E., 2006. Applications of Fourier transform-infrared spectroscopy to edible oils. Analytica Chimica Acta 573–574, 459–465.
Contents lists available at SciVerse ScienceDirect
Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng
A capacitive technique to assess water content in extra virgin olive oils Luigi Ragni a, Eleonora Iaccheri a, Chiara Cevoli a, Annachiara Berardinelli a,⇑, Alessandra Bendini b, Tullia Gallina Toschi b a b
Department of Agricultural Economics and Engeneering, University of Bologna, P.zza Goidanich 60, 47521 Cesena (FC), Italy Department of Food Science, University of Bologna, P.zza Goidanich 60, 47521 Cesena (FC), Italy
a r t i c l e
i n f o
Article history: Received 25 June 2012 Received in revised form 17 September 2012 Accepted 17 October 2012 Available online 27 October 2012 Keywords: Extra virgin olive oil Water content Fatty acid composition Capacitance
a b s t r a c t The present research investigated the correlations between capacitance and water content of extra virgin olive oils (EVOO). A commercial capacitor probe for radio applications and an LCR meter were used for electric tests in the frequency range from 500 Hz to 512 kHz. Seventeen samples of different EVOO with a moisture content ranging from 178 to 1321 mg/kg oil were selected for study. To assess the influence of moisture only, the oil with the maximum water content was filtered down to 288 mg/kg oil and five samples with intermediate water contents were prepared and submitted to electrical measurements. Subsequently, the capacitance of all 17 EVOO samples was measured at selected frequencies. Water content and capacitance for filtered oil were linearly correlated, showing R2 values up to 0.959 and a root mean square error (RMSE) of 48 mg/kg oil at 2 kHz. When the main oil composition together with the moisture changed, the trend was always linear, but the correlation (R2) decreased to 0.818 with an RMSE value of 123 mg/kg oil (at 8 kHz), so that the system remains suitable for screening of water content. Strong correspondences between capacitance and impedance suggest that these measurements can be carried out with very simple instruments for measuring voltage and current. Some novel linear correlations also emerged between water content, capacitance and fatty acid composition. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction The water content of commercial extra virgin olive oils (EVOO) usually ranges from 0.3 to 2.0 g of water/kg of oil. The type of production process, primarily the extraction and filtration procedures, affects the moisture content in EVOO (Cerretani et al., 2010; Lozano-Sánchez et al., 2010). Water in oil is evenly dispersed as micro drops (Petrakis, 2006). The dispersion, in the form of fine or micro emulsions, can be stabilized by aggregation and dissolution of polar substances such as salts, free acids, diglycerides, phospholipids, alcohols and phenols (Cerretani et al., 2008). Moisture and volatile matter in EVOO should not be higher than 0.2% (kg/kg), which is the limit recommended by the ‘‘International Olive Oil Council’’ (IOOC, 2009) and the ‘‘Codex Alimentarius’’ (CODEX STAN, 1981). The water content, which is indirectly correlated with some taste characteristics, such as pungency and bitterness, can also play a role in stability and preservation of quality during storage (Lercker et al., 1994; Fregapane et al., 2006; Ambrosone et al., 2007). The most commonly accepted methods for measurement of water content in EVOO are Karl Fischer titration (AOAC, 1998) and drying by heating (both moisture and volatile content) (ISO ⇑ Corresponding author. Tel.: +39 0547 338113; fax: +39 0547 382348. E-mail address: [email protected] (A. Berardinelli). 0260-8774/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jfoodeng.2012.10.031
662, 1998). These methods are time-consuming, and usually require large amounts of reagents. Spectroscopic techniques have been studied to assess quality parameters, such as chemical composition, and oil adulteration. The most widely used are near infrared (NIR) (Armenta et al., 2010) and Fourier transform infrared (FT-IR) spectroscopy (Vlachos et al., 2006; Cerretani et al., 2010), often in association with chemiometric statistical analysis such as partial least square regression (PLS). NIR techniques for the determination of the water content have been successfully tested on olive fruits (Jimenez et al., 2000), olive pomace (Muik et al., 2004) and in olive oils in production processes (Bendini et al., 2007), conditions where the water content is much higher than in commercial EVOO. According to Bendini et al. (2007), R2 values of 0.913 (RMSE = 1.29%) was obtained for the prediction of moisture content in olive oil samples characterized by a moisture content ranging from 45% to 73%. Several commercial oil analyzers based on NIR technology have been used to assess the moisture content in olive fruit, paste and pomace (Armenta et al., 2010). Water content in commercial olive oils ranging from 290 to 1402 mg/kg oil was predicted by FT-IR attenuated total reflectance (ATR) spectroscopy with a coefficient of determination of 0.89 (Cerretani et al., 2010). A best performance was obtained by 31 P-NMR spectroscopy in olive oils where the moisture content (ranging from 204 to 769 mg/kg oil) was predicted with an R2 value up to 0.98 (Hatzakis and Dais, 2008). Similar results (R2 = 0.984)
247
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
were achieved by using the time domain reflectometry (TDR) on samples of EVOO with water content ranging from 714 to 2008 mg of water/kg of oil (Ragni et al., 2012). Terahertz timedomain spectroscopy (THz-TDS) has also been used to investigate the water content in synthetic oils (Gorenflo et al., 2006). The goodness of prediction, expressed by the relative root mean square error (RMSE), was up to 24 mg/kg oil in samples where the water content ranged from 430 to 3280 mg/kg oil. NMR, FT-IR, TDR and THz-TDS are expensive techniques. Spectroscopy often requires time and a high level of competence for data manipulation. Capacitive methods based on parallel plates or other types of capacitive sensors have been previously explored to predict some properties in several kinds of foods: water in wheat (Berbert and Stenning, 1996), water and bulk density in safflower (Sacilik et al., 2007), ripening of banana fruits (Soltani, 2011) and freshness parameters of shell eggs (Ragni et al., 2006, 2008). Herein, we explore a simple capacitive technique to asses the water content in EVOO that uses a discontinued commercial variable capacitor as a probe, together with basic statistical analysis. This technique was firstly tested for moisture content determination in the same commercial EVOO, but with different water content obtained by filtration. For this reason, it could be also of interest to assess the water impurities in mineral-based insulating oils, if the oil composition remains constant (Farooq, 1996; Itahashi et al., 1993, 1994; Baird et al., 2006; Liland et al., 2008). Several different EVOO samples, characterized by a moisture content ranging from 178 to 1321 mg/kg oil, were also used to set up a prediction tool for oils with different chemical composition. In fact, based on the previous results of Rudan-Tasik and Klofutar (1999) and Lizhi et al. (2008), the dielectric properties of edible oils can be affected by the composition of fatty acids, and in particular by the unsaturation level (by the water remaining after filtration). Correlation between moisture content, capacitance and fatty acids were assessed. The proposed technique could represent a valuable instrument to monitor the filtration process or to perform an approximate classification of the water content in commercial EVOO, where the fatty acids composition greatly varies from oil to oil.
content. The water content of the prepared five samples was then determined according to the standard ISO 662, Method B (1998). 2.1.2. Measurements on different commercial oils Seventeen samples of different commercial oils were selected to build a dataset of moisture content ranging from 178 to 1321 mg/kg oil. The water content was again determined by the standard the ISO 662, method B. The chemical characterization of the main oil constituents was conducted by determining the fatty acid percentage. For this determination, alkaline esterification producing fatty acids methyl esters and followed by gas chromatography (Bendini et al., 2006) were used. Measurements were carried out in triplicate. The peak retention time was compared to those of the GLC 463 FAME standard mixture (Nu-Chek, Elysian, MN) to identify the peaks. Fatty acid were grouped according to their degree of unsaturation in the three conventional classes: saturated (SFA), monounsaturated (MUFA) and polyunsaturated (PUFA). 2.2. Electrical measurements Capacitance measurements on oils were made by using an instrumental chain consisting of a variable capacitor as a probe together with a LCR meter (LCR-8101G, GW-Instek, Good Will Instrument Co. Ltd, Taiwan) (Fig. 1). The capacitor has a maximum rated capacitance of 470 pF in air; the fixed and mobile armatures consist of 12 and 10 plates, respectively (Fig. 1). The capacitance can be manually changed by the rotation of the axis which involves modification in the total plate surface of the capacitor. The probe was inserted in an oil container made of PET with the following dimensions: 82 mm (length), 77 mm (width) and 55 mm (high). The container housing the capacitor was filled with 150 g of oil in such a way to fully cover the plate armatures of the capacitor. The container with capacitor and oil was placed in a metal box to shield the system from radio electromagnetic interference. In order to avoid the retention of bubbles during the oil filling, measurements were made after the rotation of the plates (vacuum application could be taken in the account as method to remove air bubbles for system improving). Rotation of the mobile plate armature with respect to the fixed one also allowed to easily remove the oil between subsequent tests by washing the probe with
(LCR meter)
Personal computer
2. Materials and methods 2.1. Samples preparation and oil composition
Top view 2.1.1. Measurements on filtered oil mixtures To assess the correlation between the amount of water and the measured capacitance excluding the influence of the variation of fatty acid composition among oils, the sample with the maximum moisture content among the EVOO collected for the measurements in the following sub-section 2.1.2 (1321 mg/kg oil) was gently filtered by insufflation of argon gas (12 l/min) for 3 h by using the patented method number WO 2009/107096 A2 (Cerretani et al., 2009), and subsequently filtrated by gravity with paper sheets (Filter papers 55 mm, Whatman) to obtain a minimum content of water. The amount of water was determined in triplicate by drying 10 g of oil in an oven according to the standard ISO 662, Method B (1998). This mass loss method overestimates the water content, including the volatile compounds, but the error should be considered minor and acceptable since they do not exceed 50 mg/kg oil in EVOO (Angerosa et al., 2004). Five samples with different percentages of non-filtered and filtered oils were prepared to obtain oils with intermediate water
Closed probe
Metallic box PET oil and capacitor probe container
Opened probe
Rotating plates
Fig. 1. Instrumental set up with details of the capacitor probe.
248
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
n-hexane. Measurements of capacitance at 21 ± 1 °C were carried out at frequencies of 500 Hz, 2 kHz, 8 kHz, 32 kHz, 128 kHz and 512 kHz for the five oil samples with different moisture contents obtained by filtration. Each measurement was the average of 50 subsequent measurements; the averaging process required about 20 s. Capacitance measurements were always conducted in triplicate. In all tests, the voltage was set to 100 mV. To consider some limits related to the behaviour of the system and the consequent oil dielectric constant determination, frequency sweep measurements were carried out from 100 Hz to 10 kHz (steps 49.5 Hz) with the capacitor empty and filled with the oil having the maximum water content. We know, in fact, that the capacitance, C, in Farads, of a parallel plate capacitor can be calculated with the following formula:
C ¼ er e0
A d
ð1Þ
where er is the relative dielectric constant of the material, e0 is the absolute dielectric constant (dielectric constant of vacuum, 8.85 1012 F/m), A is the area of the plates, in m2, d is the distance between the plates, in m. If area and distance do not change, er can be roughly calculated by dividing the capacitance with the material by the capacitance with air. This is a reasonable method to determine the static capacitance with direct current, but it becomes less appropriate with alternating current, because the real capacitor and all the instrumental chain have a behaviour that is strictly frequency dependent. Some measurements of impedance with air and oil were also made to assess which part of it can be explained by the capacitive reactance. The impedance Z, which is a frequency dependent parameter, contains three different contributors, namely ohmic resistance, reactive capacitance Xc and reactive inductance Xl. The reactive capacitance can be calculated by:
Xc ¼
1 2pfC
ð2Þ
where f is the frequency, in Hz, and C, is the capacitance, in F. In a real capacitor Xc explains most part of the impedance Z. In an electrical circuit, the impedance is related to the voltage, V, and the current, I, according to the formula:
Z¼
V I
ð3Þ
where V is in V, and I is in A. Thus, the capacitance can be calculated by a simple measurement of voltage and current. The frequency response of an industrial capacitor with rated capacitance of 1 nF was also measured for comparison.
3. Results and discussion 3.1. Water content and EVOO composition The filtration procedure reduced the water content from 1321 mg/kg oil to 288 mg/kg oil. The five samples obtained with different percentage of non-filtered and filtered oils were characterized by moisture contents of 1321 ± 31, 920 ± 120, 753 ± 24, 513 ± 7.5 and 288 ± 54 mg/kg oil. The water content of the 17 EVOO samples is shown in Table 1. It ranged from 178 mg/kg to 1321 mg/kg oil. The minimum and maximum difference between subsequent water content in different EVOO samples were 10 mg/kg oil and 266 mg/kg oil, respectively. The fatty acid composition (Table 2) reveals differences between SFA, MUFA and PUFA of the different samples up to 66.58%, 83.19% and 23.48%, respectively. The fatty acid percentages in Table 2 are in accordance with literature data for EVOO category, being the contents of single fatty acids within the legal limits for EVOO (Anonymous, 2011). 3.2. Correlation of capacitance and impedance with oil composition 3.2.1. Filtered oil mixtures The results of the linear regression analysis conducted between capacitance and water content of the five oil samples characterized by the same fatty acid composition containing moisture in the range from 288 to 1321 mg/kg oil, at the explored frequencies (between 500 Hz and 512 kHz), are summarized in Table 3. A general linear trend between capacitance and water emerged for all frequencies. The R2 values were between 0.781 and 0.962. The root mean square error (RMSEC) ranged from 71 mg/kg oil to 170 mg/ kg oil. The coefficient of variation appeared negligible between replications and was not higher than 0.21%. The frequencies of 2 kHz and 8 kHz were selected to measure the capacitance of all EVOO samples. This choice was made by considering the results obtained for the filtered oil samples (Table 3). High R2, high differences of capacitance (measured in the water
Table 1 Water content of EVOO samples.
2.3. Statistical analysis The correlation between capacitance, water and fatty acids content were investigated by using simple regression analysis. Calibration and cross validation (leave-one-out method) models were built. The coefficient of determination and the p-level values were reported. To assess the prediction power of the models the values of the root mean square error in calibration (RMSEC) was obtained. The RMSE was calculated according to the following equation:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn 2 i¼1 ðxo;1 xp;i Þ RMSE ¼ n
The two frequencies which give the best results in term of coefficient of determination and maximum variation of capacitance between samples of filtered oils were then used for the subsequent measurement on the 17 EVOO to build a simple statistical model for moisture content prediction.
ð4Þ
where xo,1 are the observed values, xp,i are the predicted values, and n is number of values. Cross validation was performed only for the models built with the seventeen EVOO samples to predict the water content by using the capacitance values.
Sample
Water content (mg/kg of oil)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
178 (8) 294 (53) 560 (3) 741 (13) 760 (29) 815 (36) 832 (18) 898 (57) 908 (7) 921 (52) 965 (31) 981 (52) 1003 (70) 1036 (26) 1113 (13) 1249 (14) 1321 (99)
Values in parentheses are standard deviations.
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
range) and low frequencies, where the cable connection have a minor influence, were preferred. The impedance and the capacitive reactance calculated by Eq. (2) for the probe without and with oil (moisture = 1321 mg/kg) are reported for different frequencies in Table 4. From these data, it can be noted that Xc is enough to well describe the impedance with a ratio Z/Xc ranging from 0.99997 to 1.00018.
18.69 68.75 12.57
Values in parentheses are standard deviations. SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids.
15.18 78.12 6.69 15.04 78.83 6.13 SFA MUFA PUFA
15.97 77.83 6.23
14.19 80.09 5.71
14.42 82.63 2.95
14.25 79.35 6.39
16.94 73.07 9.99
16.02 74.63 9.35
12.44 80.19 7.16
15.81 73.10 11.10
15.05 77.30 7.64
16.02 75.13 8.85
15.93 77.11 6.96
15.04 79.59 5.26
15.01 79.94 4.90
15.72 75.30 8.41
(0.29) (0.02) (0.03) (0.02) (0.00) (0.00) (0.50) (0.04) (0.03) (0.03) (0.00) (0.01) (0.01) (0.10) 17
(0.04) 15.22 (0.06) 0.13 (0.02) 1.10 (0.01) 0.13 (0.05) 0.19 (0.07) 2.52 (0.13) 64.25 (0.03) 2.81 (0.06) 11.85 (0.04) 0.40 (0.05) 0.72 (0.02) 0.26 (0.02) 0.10 (0.00) 0.31
16
(0.06) 12.48 (0.01) 0.08 (0.46) 0.80 (0.02) 0.01 (0.04) 0.05 (0.00) 2.67 (0.73) 71.24 (0.17) 2.79 (0.02) 7.67 (0.01) 0.43 (0.03) 0.73 (0.00) 0.34 (0.00) 0.13 (0.00) 0.00 (0.07) 11.13 (0.01) 0.10 (0.03) 0.54 (0.01) 0.02 (0.03) 0.05 (0.01) 3.56 (0.15) 75.72 (0.08) 3.28 (0.05) 4.25 (0.01) 0.29 (0.02) 0.65 (0.05) 0.25 (0.00) 0.00 (0.00) 0.00
15 14
(0.22) 11.17 (0.02) 0.10 (0.04) 0.86 (0.00) 0.04 (0.00) 0.07 (0.01) 3.56 (0.54) 75.20 (0.09) 3.08 (0.03) 4.61 (0.04) 0.27 (0.00) 0.64 (0.01) 0.28 (0.05) 0.00 (0.04) 0.00 (0.81) 12.17 (0.02) 0.12 (0.12) 1.03 (0.00) 0.07 (0.01) 0.12 (0.05) 3.04 (0.79) 72.96 (0.10) 2.62 (0.06) 6.25 (0.08) 0.41 (0.02) 0.71 (0.03) 0.26 (0.05) 0.17 (0.01) 0.09
13 12
(0.02) 12.57 (0.02) 0.14 (0.01) 1.11 (0.01) 0.08 (0.00) 0.16 (0.05) 2.51 (0.07) 70.58 (0.01) 2.84 (0.02) 8.21 (0.01) 0.44 (0.00) 0.64 (0.01) 0.30 (0.05) 0.15 (0.00) 0.28 (1.03) 10.98 (0.02) 0.14 (0.14) 0.73 (0.01) 0.10 (0.00) 0.18 (0.06) 3.08 (1.22) 73.54 (0.19) 2.41 (0.06) 6.90 (0.02) 0.44 (0.04) 0.74 (0.05) 0.31 (0.04) 0.11 (0.00) 0.34
11 10
(0.12) 12.43 (0.02) 0.12 (0.01) 1.18 (0.00) 0.08 (0.01) 0.11 (0.05) 2.69 (0.53) 68.81 (0.27) 2.59 (0.04) 10.39 (0.01) 0.47 (0.01) 0.70 (0.29) 0.29 (0.00) 0.14 (0.00) 0.00 (0.21) 9.73 (0.02) 0.08 (0.04) 0.35 (0.00) 0.03 (0.00) 0.04 (0.02) 2.39 (0.43) 77.20 (0.15) 2.31 (0.00) 6.54 (0.02) 0.29 (0.02) 0.62 (0.03) 0.20 (0.01) 0.00 (0.00) 0.00
9 8
(0.03) 12.61 (0.00) 0.13 (0.00) 1.08 (0.02) 0.08 (0.01) 0.12 (0.02) 2.76 (0.09) 70.45 (0.00) 2.55 (0.04) 8.62 (0.01) 0.44 (0.01) 0.73 (0.00) 0.30 (0.01) 0.14 (0.03) 0.00
7
(1.16) 11.67 (0.01) 0.12 (0.07) 0.89 (0.00) 0.06 (0.01) 0.10 (0.06) 2.86 (1.47) 74.36 (0.33) 2.40 (0.00) 6.02 (0.04) 0.39 (0.01) 0.67 (0.05) 0.26 (0.04) 0.13 (0.08) 0.07
6
(0.04) 11.07 (0.01) 0.09 (0.02) 0.83 (0.01) 0.05 (0.02) 0.10 (0.03) 3.21 (0.04) 75.17 (0.03) 2.37 (0.01) 5.45 (0.01) 0.45 (0.02) 0.68 (0.04) 0.26 (0.01) 0.13 (0.00) 0.13
5
(0.16) 14.10 (0.02) 0.15 (0.01) 1.31 (0.01) 0.09 (0.01) 0.17 (0.05) 2.24 (0.18) 67.96 (0.09) 3.17 (0.01) 9.30 (0.03) 0.38 (0.02) 0.69 (0.01) 0.31 (0.00) 0.13 (0.03) 0.00 (0.40) 10.49 (0.00) 0.14 (0.00) 0.70 (0.00) 0.06 (0.02) 0.13 (0.12) 3.08 (1.77) 75.85 (0.14) 2.25 (2.41) 5.71 (0.03) 0.40 (0.01) 0.68 (0.01) 0.28 (0.02) 0.11 (0.04) 0.11
4 3
(0.20) 10.41 (0.00) 0.12 (0.01) 0.74 (0.02) 0.05 (0.03) 0.08 (0.01) 3.37 (0.15) 79.11 (0.05) 2.32 (0.02) 2.33 (0.02) 0.40 (0.00) 0.62 (0.03) 0.26 (0.00) 0.13 (0.00) 0.06 (0.06) 11.08 (0.00) 0.12 (0.01) 0.69 (0.00) 0.03 (0.01) 0.07 (0.02) 2.63 (0.19) 76.11 (0.16) 2.78 (0.02) 5.06 (0.02) 0.45 (0.21) 0.65 (0.01) 0.32 (0.05) 0.00 (0.00) 0.00
2 1
C16:0 11.79 C16:l n-9 0.12 C16:l n-7 0.94 C17:0 0.05 C17:l 0.09 C18:0 3.60 C18:l n-9 73.39 C18:l n-7 3.05 C18:2 5.68 C20:0 0.29 C18:3 n-3 0.55 C20:l 0.25 C22:0 0.24 C24:0 0.00
EVOO
Fatty acids
Table 2 Fatty acid composition of EVOO samples. Values are in percentage (%) of the total fatty acid composition.
249
3.2.2. Different commercial oils The correlations between the capacitance and the water content of all the oils for the above-mentioned frequencies are shown in Figs. 2 and 3. Predicted and observed values for the cross validation are shown in Figs. 4 and 5. As expected, these results show lower coefficients of determination compared to those obtained for the filtered oils. R2 values of 0.804 and 0.818 were obtained for frequencies of 2 kHz and 8 kHz, respectively. The RMSEC was 127 mg/kg oil and 123 mg/kg oil, instead the RMSECV was 142 and 139 mg/kg for the two frequencies respectively. The different composition of the 17 oils involves an unavoidable ‘noise’ that perturbs the electrical information related to the water content. The relative dielectric constant of some SFA (saturated fatty acids), MUFA (monounsaturated fatty acids) and PUFA (polyunsaturated fatty acids) were measured using the method of Lizhi et al. (2008) in the range from 100 Hz to 1 MHz, and it was roughly invariant up to a frequency of 100 kHz. The values of e r ranged from about 2.3 for palmitic and stearic acid (75 °C) to 2.5 for linolenic and caprilic acids (25 °C), oleic and linoleic acids (25 °C) assuming intermediate values. Although the e r of water is more than 30 times higher than that of the fatty acid, the mass variations in the oil composition between SFA, MUFA and PUFA were higher than that due to the moisture content. Moreover, other components (not measured in this investigation), such as diglycerides, can play a role in the determination of the dielectric constant of the oil. A significant, although weak linear correlation between MUFA and water content was found for the examined oils (R2 = 0.277, p-level = 0.037), one outlier removed (Fig. 6). At present, in the light of these results, we can only suggest that the tendency to retain moisture in commercial EVOO oil may be related to the balance of different types of fatty acids in the lipid matrix. A linear correlation (R2 = 0.627) was also obtained between capacitance and MUFA (Fig. 7). Despite this, the use of SFA, MUFA and PUFA in multiple linear regressions models or other simple functions did not involve an increase in power prediction of the water content. 3.3. Limit for dielectric constant calculation The calculation of the relative dielectric constant through the ratio between capacitance with oil and without oil, at the different frequencies, cannot be conducted with this probe. The frequency response of the probe appeared to be affected by several regions where the capacitance suddenly drops to lower values compared to that which should characterise an ideal capacitor. Fig. 8 shows this behaviour in the restricted range from 100 Hz to 10 kHz. The frequency response of an industrial capacitor with rated capacitance of 1 nF is also shown in the graph for comparison. Capacitance steps of the capacitor probe appear at different frequencies with and without oil, so that the calculation of the dielectric constant needed to describe very small differences between oils with different water contents, at different frequencies, failed. The above-mentioned behaviour can be attributed to the fact that the capacitor probe is a complex component, characterised by a large metallic armature, very far from an ideal capacitor. Antenna effects, reflections and resonances can occur at some frequencies with a dielectric (air) and at other frequencies with another dielectric (oil). These phenomena largely deviates the probe behav-
250
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
Table 3 Analysis of the correlations between capacitance and water content (from 288 to 1321 mg/kg) in the frequency range from 500 Hz to 512 kHz. Frequency (kHz)
Function
R2
DCHI-LO (pF)
DC%
RMSEC (mg/kg of oil)
0.5 2 8 32 128 512
W = 118.38C – 159,057 W = 118.51C – 158,678 W = 142.49C – 190,205 W = 171.63C – 229,882 W = 247.72C – 331,600 W = 147.07C – 197,111
0.895 0.959 0.962 0.959 0.781 0.784
8.350 8.390 7.085 6.030 3.047 6.710
0.605 0.599 0.536 0.450 0.227 0.500
111 48 118 73 170 169
W, water content (mg/kg oil); C, capacitance (pF); DCHI-LO, difference between the highest and the lowest capacitance measured values; DC%, ratio between DCHI-LO and the lowest capacitance value 100; RMSE, root mean square error.
Table 4 Impedance and reactive capacitance of the probe in air and with oil with a water content of 1321 mg/kg oil (n.12, in Tables 1 and 2), at different frequencies. Frequency (Hz)
Air
500 2000 8000 32,000 128,000 512,000
Oil
Z (X)
Xc (X)
Z (X)
Xc (X)
693,130 173,412 43,389 10,854 2716 678
693,151 173,413 43,389 10,854 2716 678
235,237 58,908 14,806 3691 926 230
235,228 58,926 14,806 3691 926 230
Z/Xc air
Z/Xc oil
0.99997 0.99999 0.99999 1.00000 0.99994 1.00018
1.00004 0.99970 1.00002 1.00001 1.00005 0.99997
Z, impedance; Xc, capacitive reactance; for the Z/Xc calculation, six significant digits of precision for Z and Xc representation were used.
1300
2
1500
R = 0.804 RMSEC = 127 mg/kg oil
Predicted water content (mg/kg oil)
Water content (mg/kg oil)
1500
1100 900 700 500 300
2
R = 0.756 RMSECV = 142 mg/kg oil
1300 1100 900 700 500 300 100
100 1333
1338
1343
1348
1353
100
300
Capacitance (pF)
1500
Predicted water content (mg/kg oil)
2
Water content (mg/kg oil)
900
1100
1300
1500
Fig. 4. Predict and observed values for the water content at 2 kHz (cross validation).
1500 R = 0.818 RMSEC = 123 mg/kgoil
1100 900 700 500 300 100 1335
700
Observed water content (mg/kg oil)
Fig. 2. Correlation between capacitance and water content for all oils at 2 kHz.
1300
500
2
R = 0.767 RMSECV = 139 mg/kg oil
1300 1100 900 700 500 300 100
1338
1341
1344
1347
1350
Capacitance (pF) Fig. 3. Correlation between capacitance and water content for all oils at 8 kHz.
100
300
500
700
900
1100
1300
1500
Observed water content (mg/kg oil) Fig. 5. Predict and observed values for the water content at 8 kHz (cross validation).
251
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252 2
R = 0.277
Water content (mg/kg oil)
1500 1300 1100 900 700
4. Conclusions
500 300 100 67
71
75
79
83
MUFA (%) Fig. 6. Correlation between monounsaturated fatty acids (MUFA) and water content for all oils.
1354
Capacitance (pF)
iour from that of a smaller, more compact industrial capacitor (Fig. 8). Thus, we could only calculate an approximate value of the relative dielectric constant of oil, which resulted in the explored frequency range from 2.933 to 2.942. Moreover, at a certain frequency, the dielectric constant variation due to the different moisture content is the same of the capacitance variation and assumed values between 0.227% and 0.605% (for filtered samples).
2
R = 0.627
1349
1344
1339
Capacitive techniques based on multiple parallel plate capacitors can be taken into consideration for the prediction of extremely low water content in filtered EVOO. A water content from 0.03% to 0.13% (kg/kg) of an EVOO can be predicted by using the capacitance value in a simple linear regression with an R2 values ranging from 0.781 to 0.962 for some explored frequencies in the kHz range with root mean square error up to 48 mg/kg oil for the best model. Correlations for EVOO where the chemical composition greatly varies (fatty acid, etc.) was lower than that of the oil where only the moisture changed (R2 = 0.818, at the frequency of 8 kHz). For this reason, this system can be used for water determination in EVOO if a precision expressed as root mean square error not higher than around 142 mg/kg oil is requested. A maximum of uncertainty of 50 mg/kg oil as to be added to take into account possible contribution of volatile compounds in the model constructions. A significant linear correlation (p-level = 0.037) was obtained between monounsaturated fatty acid composition and water content, and a quite good correlation (R2 = 0.627) was seen between the capacitance value and the monounsaturated fatty acid percentage in commercial EVOO oils, behaviours that deserve further investigation. Measurements of impedance demonstrated that it is well described only by the capacitive reactance, so that a simple instrumental system able to measure voltage and current can be set up and tested for the prediction of some chemical characteristics of oils.
1334 67
71
75
79
83
Acknowledgements
MUFA (%) Fig. 7. Correlation between capacitance and monounsaturated fatty acids (MUFA) for all oils at 2 kHz.
The authors thank Dr. Enrico Valli and Mr. Anton Kuti for their collaboration in laboratory activities addressed to the oils characterization.
Olive Oil 1355
Air
I.C.
460
960
459.5
950
459
940
458.5
930
458
920
Olive Oil
1345
Air 1340 1335
Capacitance (pF)
Capacitance (pF)
1350
1330
Industrial Capacitor (I.C.) 1325
0
2000
4000 6000 Frequency (Hz)
8000
10000
Fig. 8. Capacitance for the probe in air and oil with a water content of 1321 mg/kg oil (n. 12, in Tables 1 and 2), together with an industrial capacitor with a rated capacitance of 1 nF, in the range 100 Hz to 10 kHz.
252
L. Ragni et al. / Journal of Food Engineering 116 (2013) 246–252
The research was carried out with the financial contribution of the Italian Ministry of the Instruction, University and Research in the framework of the PRIN project ‘‘Innovative technologic applications to improve the extraction yield of virgin oils from olives and seeds and for rapid controls of their quality’’. References Ambrosone, L., Mosca, M., Ceglie, A., 2007. Impact of edible surfactants on the oxidation of olive oil in water-in-oil emulsions. Food Hydrocolloids 21, 1163– 1171. Angerosa, F., Servili, M., Selvaggini, R., Taticchi, A., Esposto, S., Montedoro, G.F., 2004. Volatile compounds in virgin olive oil: occurrence and their relationship with the quality. Journal of Chromatography A 1054, 17–31. Anonymous, 2011. On the characteristics of olive oil and olive pomace oil and the relevant methods of analysis. EC Commission Regulation (EEC) No. 61/2011 of 24 January 2011 amending Regulation No. 2568/91, Official Journal European Community 23, 1–14. AOAC Official Method 984.20, 1998. Moisture in Oils and Fats. Karl Fischer Method. Official Methods of Analysis of AOAC International, vol. I, 16th ed. AOAC, Gaithersburg, MD, pp. 41–42 (Chapter 41). Armenta, S., Moros, J., Garrigues, S., de la Guardia, M., 2010. Determination of olive oil parameters by near infrared spectrometry. In: Olives and Olive Oil in Health and Disease Prevention. Elsevier (Chapter 58). Baird, P.J., Herman, H., Stevens, G.C., Jarman, P.N., 2006. Spectroscopic measurement and analysis of water and oil in transformer insulating paper. IEEE Transactions on Dielectrics and Electrical Insulation 13, 293–308. Bendini, A., Cerretani, L., Vecchi, S., Carrasco-Pancorbo, A., Lercker, G., 2006. Protective effects of extra virgin olive oil phenolics on oxidative stability in the presence or absence of copper ions. Journal of Agricultural and Food Chemistry 54, 4880–4887. Bendini, A., Cerretani, L., Di Virgilio, F., Belloni, P., Lercker, G., Gallina Toschi, T., 2007. In-process monitoring in industrial olive mill by means of FT-NIR. European Journal of Lipid Science and Technology 109, 498–504. Berbert, P.A., Stenning, B.C., 1996. On-line moisture content measurement of wheat. Journal of Agricultural Engineering Research 65, 287–296. Cerretani, L., Bendini, A., Barbieri, S., Lercker, G., 2008. Osservazioni preliminari riguardo alla variazione di alcune caratteristiche chimiche di oli vergini da olive sottoposti a processi di deodorazione ‘‘soft’’. La Rivista Italiana delle Sostanze Grasse 85, 75–82. Cerretani L., Rocculi P., Bendini A., Romani S., Bacci A., 2009. Oil Clarifying Process and Apparatus for Implementing the Process. International Publication Number WO 2009/107096 A2. Cerretani, L., Giuliani, A., Maggio, R.M., Bendini, A., Gallina Toschi, T., Cichelli, A., 2010a. Rapid FTIR determination of water, phenolics and antioxidant. European Journal of Lipid Science and Technology 112, 1150–1157. Codex Standard for Olive Oils and Olive Pomace Oils, 1981. Codex Stan, p. 33. Farooq, K., 1996. The effect of particulate and water contamination on the dielectric strength of insulating oils. In: Proceeding of the IEEE International Symposium on Electrical Insulation, Millbrae, CA. Fregapane, G., Lavelli, V., León, S., Kapuralin, J., Salvador, M.D., 2006. Effect of filtration on virgin olive oil stability during storage. European Journal of Lipid Science and Technology 108, 134–142. Gorenflo, S., Tauer, U., Hinkov, I., Lambrecht, A., Buchner, R., Helm, H., 2006. Dielectric properties of oil–water complexes using terahertz transmission spectroscopy. Chemical Physics Letters 421, 494–498.
Hatzakis, E., Dais, P., 2008. Determination of water content in olive oil by 13P NMR spectroscopy. Journal of Agricultural and Food Chemistry 56, 1866–1872. International Olive Oil Council, 2009. Trade Standard Applying to Olive Oils and Olive-Pomace Oils. COI/T.15/NC No 3/Rev. (4 November). ISO 662, 1998. Animal and Vegetable Fats and Oils – Determination of Moisture and Volatile Matter Content, second ed. Itahashi, S., Ueta, T., Sakurai, H., Mitsui, H., Sone, M., 1993. Effect of water dimer in insulating oil on conduction phenomena. In Proceedings of the IEEE Conference on Electrical Insulation and Dielectric Phenomena, Pocono Manor, PA. Itahashi, S., Sakurai, H., Iwasawa, T., Mitsui, H., Sato, T., Sone, M., 1994. Effect of water in insulating oil on conduction phenomena under high electrical field. In: Proceedings of the IEEE Conference on Electrical Insulation and Dielectric Phenomena, Arlington, TX. Jimenez, A., Izquierdo, E., Rodriguez, F., Duenas, J.I., Tortosa, C., 2000. Determination of fat and moisture in olives by near-infrared reflectance spectroscopy. Grasas y Aceites 51, 311–315. Lercker, G., Frega, N., Bocci, F., Servidio, G., 1994. Veiled extravirgin olive oils: dispersion response related to oil quality. Journal of the American Oil Chemists’ Society 71, 657–658. Liland, K.B., Eidnes, K., Bjorneklett, K., Hvidsten, S., 2008. Measurement of solubility and water content of insulating oils for HV XLPE cable terminations. In: Proceedings of the IEEE International Symposium on Electrical Insulation. Vancouver, Canada. Lizhi, H., Toyoda, K., Ihara, I., 2008. Dielectric properties of edible oils and fatty acids as a functionof frequency, temperature, moisture and composition. Journal of Food Engineering 88, 151–158. Lozano-Sánchez, J., Cerretani, L., Bendini, A., Segura-Carretero, A., FernándezGutiérrez, A., 2010. Filtration process of extra virgin olive oil: effect on minor components, oxidative stability and sensorial and physicochemical characteristics. Trends in Food Science and Technology 21, 201–211. Muik, B., Lendl, B., Molina-Dìaz, A., Pérez-Villarejo., L., Ayora-Canãda, M.J., 2004. Determination of oil and water content in olive pomace using near infrared and Raman spectrometry. A comparative study. Analytical and Bioanalytical Chemistry 379, 35–41. Petrakis, C., 2006. Olive oil extraction. In: Boskou, D. (Ed.), Olive Oil Chemistry and Technology. AOCS Publishing, Champaign, Illinois, pp. 191–223. Ragni, L., Gradari, P., Berardinelli, A., Giunchi, A., Guarnieri, A., 2006. Predicting Quality Parameters of shell eggs using a simple technique based on the dielectric properties. Biosystems Engineering 94, 255–262. Ragni, L., Berardinelli, A., Guarnieri, A., 2008. A dielectric technique based on a onechip network analyser to predict the quality indices of shell eggs. Biosystems Engineering 100, 470–478. Ragni, L., Berardinelli, A., Cevoli, C., Valli, E., 2012. Assessment of the water content in extra virgin olive oils by Time Domain Reflectometry (TDR) and Partial Least Squares (PLS) regression methods. Journal of Food Engineering 111, 66–72. Rudan-Tasic, D., Klofutar, C., 1999. Characteristics of vegetable oils of some Slovene manufactures. Acta Chimica Slovenica 46, 511–521. Sacilik, K., Tarimci, C., Colak, A., 2007. Moisture content and bulk density dependence of dielectric properties of safflower seed in the radio frequency range. Journal of Food Engineering 78, 1111–1116. Soltani, M., Alimardani, R., Omid, M., 2011. Evaluating banana ripening status from measuring dielectric properties. Journal of Food Engineering 105, 625–631. Vlachos, N., Skopelitis, Y., Psaroudaki, M., Konstantinidou, V., Chatzilazarou, A., Tegou, E., 2006. Applications of Fourier transform-infrared spectroscopy to edible oils. Analytica Chimica Acta 573–574, 459–465.