Food Chemistry 150 (2014) 429–437

Contents lists available at ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Analytical Methods

Deep-frying food in extra virgin olive oil: A study by 1H nuclear magneticresonance of the influence of food nature on the evolving composition ofthe frying medium

0308-8146/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.foodchem.2013.11.015

⇑ Corresponding author. Tel.: +34 945 013081; fax: +34 945 013014.E-mail address: mariadolores.guillen@ehu.es (M.D. Guillén).

Andrea Martínez-Yusta, María Dolores Guillén ⇑Food Technology, Faculty of Pharmacy, Lascaray Research Center, University of the Basque Country (UPV-EHU), Paseo de la Universidad No. 7, 01006 Vitoria, Spain

a r t i c l e i n f o

Article history:Received 14 April 2013Received in revised form 13 July 2013Accepted 3 November 2013Available online 13 November 2013

Keywords:Extra virgin olive oilDeep-fryingFood lipids migrationFrying media compositionDegradationAcyl groupsAldehydesEpoxidesAlcoholsHydrolysisIodine ValuePercentage in weight of polar compounds

a b s t r a c t

Three series of fourteen deep-frying experiments on three foods of very different compositions werecarried out using extra virgin olive oil as the original frying medium. The aim of the study was to establishhow the nature of the food being fried influences the composition of the frying medium. The changes inthe composition of the frying media referred to the evolution of molar percentage of the different kinds ofacyl groups, as well as to the evolution of the concentration of newly formed compounds such asaldehydes, epoxystearyl groups, primary and secondary alcohols were monitored in a simultaneousway by 1H NMR spectroscopy. Changes due to hydrolytic processes were also considered. In addition,the evolution of the Iodine Value and percentage in weight of polar compounds was also analysed. Theinfluence of food lipids migration to the frying medium on the composition of this latter was evidenced.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Frying is a culinary process in which both oil and food are mod-ified. The chemical and physical changes provoked in the food re-sult in the acquisition of desirable attributes. The fatty mediumalso undergoes very complex chemical and physical changes, andthe formation of toxic compounds with important repercussionson health is also possible. This culinary process is in worldwideused and due to its complexity, many of its aspects remain un-known and require further research.

The study of the modifications of oils, either heated at fryingtemperature in absence of food, or used in frying experiments,has been accomplished by methodologies which determine eitherphysical or chemical parameters, such as Iodine Value, conjugateddiene values, polar compounds percentage, peroxide and anisidinevalues or concentration of dimmers, polymers of triglycerides aswell as of oxidised triglycerides, (Andrikopoulos, Kalogeropoulos,Falirea, & Barbagianni, 2002; Dobarganes & Márquez-Ruiz, 2007;

Kalogianni, Karapantsios, & Miller, 2011; Sánchez-Gimeno, Negu-eruela, Benito, Vercet, & Oria, 2008). It should be pointed out thatfrom some of the above mentioned parameters it is not easy toextract information about the chemical changes occurring in thefriyng medium. Furthermore, the influence of the composition ofthe food being fried on the chemical changes in the frying mediumhas been the subject of only a few studies.

This paper deals with the latter subject. To this aim, a study ofthe evolution of the composition of the frying medium throughoutthree series of frying experiments on three foods is carried out.This is compared with the evolution of the composition of the ori-ginal oil submitted to frying temperature in the same fryer but inthe absence of food.

In both cases, 1H NMR spectroscopy is applied as in previousstudies (Guillén & Uriarte, 2011, 2012a,b,c). The foods selectedfor the study differ not only in lipid content but also in lipid nature.

This work may provide new information about the effect of thefrying process on changes in the composition of the frying mediumin comparison with its simple heating; furthermore, this study mayalso help to clarify the influence of food components on the evolu-tion of the composition of the frying medium by the possible

430 A. Martínez-Yusta, M.D. Guillén / Food Chemistry 150 (2014) 429–437

occurrence of the reactions between components of both systems(medium and food), or by potential migration of the lipidic foodcomponents to the frying medium.

2. Materials and methods

2.1. Oil and food samples

The oil used in this study was extra virgin olive oil, acquired in alocal supermarket; its composition given in molar percentage ofthe several kinds of acyl groups (linolenic, Ln, 0.8%; linoleic, L,6.5%; oleic, O, 80.8%; and saturated, S, 11.9%) was determined fromits 1H NMR spectrum as in previous studies (Guillén & Uriarte,2012a).

One of the foods submitted to frying is frozen commercialdough of Spanish doughnut whose average estimated compositionis 29.6% water, 8.0% proteins, 60.8% carbohydrates, and 0.8% lipids,these lipids being made up of 4.0% Ln, 55.0% L, 15.5% O, and 21.5%S. The other food involved in the study is pork adipose tissue, withan approximate composition of 18.0% water, 4.7% proteins, and76.7% lipids, these latter constituted by 1.7% Ln, 10.2% L, 46.8% O,and 41.3% S. The third food involved is farmed salmon fillets (Salmosalar) whose approximate composition is 55.5% water, 17.6%proteins, and 26.2% lipids, these latter formed of 16.0%, omega-3(x-3) groups, 10.9% L, 55.9% O or monounsaturated (MU), and16.8% S. All of them were acquired in local supermarkets. The com-position of the lipids of the pork adipose tissue as well as of thefarmed salmon fillets, involved in this study, were determined by1H NMR as in previous studies (Guillén & Uriarte, 2012a; Vidal,Manzanos, Goicoechea, & Guillén, 2012), after extraction usingultrasounds and carbon disulphide as solvent. It should also be no-ticed that the omega-3 groups of these salmon lipids contain4.0 ± 0.0% of docosahexaenoic (DHA) acyl groups and 4.7 ± 0.3% ofeicosapentaenoic (EPA) acyl groups.

2.2. Experiment of heating at frying temperature of the extra virginolive oil

For comparative purposes, before carrying out the deep-fryingexperiments, the extra virgin olive oil was submitted to the sameheating conditions but without food. To this aim, 4000 cm3 of extravirgin olive oil was submitted at 190 �C in an industrial fryer(Franke ECO4, 230 V, 10 A, 2.3 kW) for periods of 8 h/day, for4 days (the total heating time was 32 h and 30 min); the oil wasstored at room temperature between the heating episodes. Thedimensions of the stainless steel tank of the fryer were 15 cmwide � 30 cm long � 17 cm high. During heating no oil was replen-ished. Oil samples were periodically taken and, when necessary,refrigerated until their study in order to avoid or hinder the contin-uation of the degradation process. This heating experiment wascarried out in duplicate before the deep-frying experiments witheach food. This extra virgin olive oil submitted to heating at fryingtemperature was named EVO.

2.3. Deep-frying experiments

Three series of deep-frying experiments, one with each kind offood (Spanish doughnut, pork adipose tissue, and farmed salmonfillets), were carried out. The same initial amount of extra virginolive oil as described above was used for each series of deep-fryingexperiments, as was the same industrial fryer and heatingconditions. In each series of deep-frying experiments, the oil washeated for 32 h and 30 min (8 h/day for 4 days); in this time 14frying experiments were carried out. The first frying experimentin each series was done just when the oil reached 190 �C, and

the subsequent frying experiments after regular intervals of2.5 h. The deep-frying time was in all cases 1 min. In the case ofthe Spanish doughnut each frying experiment was carried out witheight pieces of defrosted dough shaped into cylinders 10 cm inlength and 1 cm in diameter, weighing 40 g in total. In the casesof pork adipose tissue and salmon fillets, six pieces (7 cm by3 cm by 1.75 cm of pork adipose tissue) and four pieces (7 cm by5 cm by 3 cm of salmon fillet), weighing in total 250 g, were friedsimultaneously in each frying experiment.

During each series of deep-frying experiments no oil wasreplenished. Samples of oil of the three series of frying experimentswere taken before and after each frying experiment. Each series offrying experiments was carried out in duplicated. The frying mediainvolved in the deep-frying of dough of Spanish doughnut, porkadipose tissue and salmon fillets were named DEVO, PEVO andSEVO respectively.

2.4. Monitoring of the evolution of the frying media composition by 1Hnuclear magnetic resonance spectroscopy

The evolution of the composition of the frying media wasmonitored by 1H nuclear magnetic resonance. To this aim, the 1Hnuclear magnetic resonance spectra of EVO, DEVO, PEVO and SEVOthroughout the several experiments (heating and frying experi-ments), were registered in a Bruker Avance 400 spectrometer oper-ating at 400 MHz. All operating conditions were as in previousstudies (Guillén & Uriarte, 2012a).

The 1H NMR spectra of the original oil and of the oil (EVO) andfrying media (DEVO, PEVO and SEVO) after 32.5 h of heating orfourteen frying experiments respectively are shown in Fig. 1.Table 1 shows the assignment of the main signals of these spectra;the assignment of some of the signals was made as in previousstudies (Guillén & Ruiz, 2003a,b, 2005a,b,c; Guillén & Uriarte,2012a; Sopelana, Arizabaleta, Ibargoitia, & Guillén, 2013; Vidalet al., 2012) and that of other signals has been made in this paperfor the first time. In this study triolein, triestearin, trilinolein andtrilinolenin as well as oleic and linoleic acids and some primaryand secondary alcohols as well as aldehydes of different typesand certain 1,2-diglycerides, and 1,3-diglycerides, were used asstandard compounds for both the assignment of certain signalsof the spectra and for quantitative purposes.

2.5. Determination of the Iodine Value

The determination of the Iodine Value (IV) was made from 1HNMR data by using a previously developed approach which provedthat Iodine Value (Guillén & Ruiz, 2003b) is related to the percent-age of olefinic protons (OP%) by the equation

IV ¼ 10:54þ 13:39 OP%: ð1Þ

The percentage of olefinic protons (OP%) in the oil can be directlydetermined from the area of signal R in Fig. 1.

2.6. Determination of the percentage in weight of polar compounds

This parameter was determined throughout the heating or fry-ing experiments by using a Testo 265 instrument. The measure-ments are based on the dielectric constant of the oil and aredirectly transformed by the instrument into percentage in weightof polar compounds (PC%).

2.7. Statistic and kinetic studies

The statistical package IMB SPSS Statistics (Version 19) wasused to find equations that fit molar percentages of the severalkinds of acyl groups of the oil or frying media and heating time.

Fig. 1. 1H NMR spectra of the original extra virgin olive oil, and of EVO, DEVO, PEVO and SEVO media after 32.5 h of processing together with the enlargement of certainsignals. Signal J0 of SEVO medium before frying (f0) and after 4 (f4), 7 (f7), 10 (f10) and 14 (f14) frying experiments.

Table 1Assignment to different kinds of protons of some of the main signals, showed in Fig. 1,appeared in the 1H NMR spectra of EVO, DEVO, PEVO and SEVO fatty mediathroughout the heating or frying experiments. The signal letter agrees with those inFig. 1.

Signal Chemical shift (ppm) Functional group

A 0.68 (s) –CH3 (position 18; b-sitosterol)B 0.70 (s) –CH3 (position 18; D5-stigmasterol)C 0.83–0.93 (t-o-br) –CH3 (saturated, oleic and linoleic acyl and

other groups)D 0.93–1.03 (t) –CH3 (omega-3 acyl and other groups)E 1.22–1.42 (o-br) –(CH2)n– (acyl and other groups)F 1.52–1.70 (o-br) –OCO–CH2–CH2– (acyl groups, except DHA

and similars)G 1.94–2.14 (o-br) –CH2–CH@CH– (acyl and other groups,

except DHA and similars partially)H 2.23–2.36 (t-o-br) –OCO–CH2– (acyl group, except DHA and

similars)I 2.63 (m)

-CH-CH-O ((E)-9,10-epoxystearyl groups)

J 2.70–2.86 (o-br) @HC–CH2–CH@ (acyl and other groups)K 2.88 (m)

-CH-CH-O ((Z)-9,10-epoxystearyl groups)

L 3.54–3.59 (m) –CHOH– (secondary alcohols)M 3.62 (t) –CH2OH– (primary alcohols)N 3.71 (d) –CH2OH (1,2-diglycerides)O 4.04–4.10 (m) –CHOH (1,3-diglycerides)P 4.10–4.32 (dd,dd) –CH2OCOR (glyceryl groups)Q 5.20–5.26 (tt) >CHOCOR (glyceryl groups)R 5.26–5.40 (br) –CH@CH– (acyl and other groups)S 9.49 (d) –CHO ((E)-2-alkenals)T 9.52 (d) –CHO ((E,E)-2,4-alkadienals)U 9.75 (t) –CHO (n-alkanals)V 9.78 (t) –CHO (4-oxo-alkanals)

s: singlet; t: triplet; o: overlapping; br: broad signal; DHA: docosahexaenoic acylgroups; m: multiplet; d: doublet; dd,dd: double doublet; tt: triple triplet.

A. Martínez-Yusta, M.D. Guillén / Food Chemistry 150 (2014) 429–437 431

Likewise, equations that fit the concentration of aldehydes, epox-ides, alcohols, or diglycerides (expressed as mmol per mol of tri-glycerides) and heating time were also found using the same

statistical package. In the same way, equations that relate heatingtime and either percentage in weight of polar compounds, or Io-dine Value were obtained.

3. Results and discussion

3.1. Changes in the composition pattern of acyl groups of the oil (EVO)and of the frying media (DEVO, PEVO and SEVO) throughout theseveral processes

Triglycerides as main oil components underwent changes as aconsequence of different kinds of reactions. Some of these onlyaffect the chains of their acyl groups with the triglyceride structureremaining intact. Amongst these can be cited: reactions thatmodify the acyl group chains by incorporation of new additionaloxygenated functional groups (hydroperoxy, hydroxy, epoxy, etc.)or of new unsaturations patterns (conjugated dienic or trienic sys-tems) leading to intermediate or final compounds; reactions thatbring about the breaking of acyl group chains leading to the forma-tion of both small molecules exhibiting different functional groupsand triglycerides having truncated acyl groups; and polymerisationreactions that, modify the acyl group chains by disappearance ofdouble bonds, forming dimers, oligomers or polymers of intact orof modified triglycerides. The occurrence of all these reactionshas as its consequence the modification of the composition patternof the acyl groups.

3.1.1. Heating of extra virgin olive oil at frying temperature for aprolonged period of time

It is known that the prolonged heating at frying temperature ofextra virgin olive oil modifies the composition pattern of its acylgroups by diminution of the concentration of the three unsaturatedacyl groups without the total disappearance of any of them(Guillén & Uriarte, 2012a); this finding has been corroborated bythis study. The determination of the molar percentage of each kindof acyl group of EVO throughout 32.5 h of heating was carried out

432 A. Martínez-Yusta, M.D. Guillén / Food Chemistry 150 (2014) 429–437

as in previous studies from 1H NMR spectral data (Guillén &Uriarte, 2012a) from the area (A) of signals of different kinds ofprotons shown in Fig. 1 and Table 1. The molar percentage of lino-lenic (Ln) and linoleic groups (L), was determined from the area ofthe signals of bis-allilic protons of linolenic (AJLn) and linoleic (AJL)groups respectively and of the protons of glyceryl groups (AP) byusing the equations

Ln% ¼ 100 ðAJLn=3APÞ; ð2Þ

and

L% ¼ 100 ð2AJL=3APÞ; ð3Þ

the molar percentage of oleic (or monounsaturated) groups (Oor MU) was determined by the equation

O ðorMUÞ% ¼ 100 ½ðAG � 2AJL � AJLnÞ=3AP� ð4Þ

in which the area of mono-allilic protons (AG) is involved; andthat of saturated plus modified acyl groups (S + M), by the equation

ðSþMÞ% ¼ 100 ½1� ðAG=3APÞ�: ð5Þ

These determinations were carried out considering that thecontribution either of diglycerides or of monoglycerides is verysmall; in fact only 1,2-diglycerides were found initially and theirinclusion in the equations does not modify the results due to thefact that their abundance is hundred times smaller than that oftriglycerides.

The equations that relate, with high correlation coefficients, themolar percentage of different kinds of acyl groups and heatingtime, together with the experimental data are represented inFig. 2a–d. These results, as expected, indicate that the degradationaffects the three unsaturated acyl groups. The rate of decrease ofthe molar percentage of oleic (0.1372) is much higher than thatof linoleic (0.0578) and this is higher than that of linolenic(0.0105) acyl groups; and the molar percentage of saturated plusmodified acyl groups increases at a rate of 0.2054. These valuesare similar to those found for other virgin olive oil in the previousstudy above mentioned.

3.1.2. Deep-frying experiments of dough of Spanish doughnutsAs above, the determination of the molar percentage of the sev-

eral kinds of acyl groups of the frying medium (DEVO) throughoutthe fourteen successive deep-frying experiments of dough of Span-ish doughnuts was carried out. In the same way as before, the mo-lar percentage of the several kinds of acyl groups and the heatingtime fits very well to linear equations, which are represented inFig. 2a–d.

It can be observed that the evolution of the molar percentage ofthe different kinds of acyl groups with the heating time in EVO andDEVO samples coincides greatly. This indicates that the degrada-tion of acyl groups in the simple heating of extra virgin olive oiland when it is used to fry Spanish doughnut dough is very similar,though slightly smaller in the latter case. This small differencecould be attributed to the sudden decrease of temperature in thesystem when the food is repeatedly introduced into the oil, mean-ing that DEVO consequently remains less time at frying tempera-ture than EVO, although the frying temperature is rapidlyrecovered. It should be commented on that when frying Spanishdoughnut dough the effect of possible migration of lipids fromthe food to the frying medium is not appreciable due to their lowcontent in this food.

3.1.3. Deep-frying experiments of pork adipose tissueThe same determinations as above were made from 1H NMR

spectra signals of PEVO. The equations that relate the molarpercentage of the different kinds of acyl groups in PEVO and theheating time are also represented in Fig. 2a–d. As in EVO and DEVO,

the molar percentages of the three unsaturated acyl groups de-crease throughout the heating time, but in this case there are cleardifferences. The rate of diminution of the molar percentage of lin-olenic and linoleic acyl groups is somewhat smaller in PEVO(0.0076, 0.0419) than in EVO (0.0105 and 0.0578) and DEVO(0.0117 and 0.0542); however, that of oleic or monounsaturatedacyl groups is much larger (0.2194 in PEVO versus 0.1372 in EVOand 0.1358 in DEVO). This difference has its consequence that afterfourteen frying experiments the molar percentage of oleic groupsin PEVO is much smaller (72.9%) than in EVO (76.1%) and DEVO(76.3%) and the opposite is true for saturated plus modified acylgroups.

As proved above, the successive sudden decrease and increase ofthe frying medium temperature in the frying experiments only af-fect the rate of variation of the molar percentage of the different acylgroups to a very small extent. Nor can this greater decrease in themolar percentage of oleic groups be attributed to a higher degrada-tion of this acyl group in the frying medium when pork adipose tis-sue is fried (PEVO) than in the above mentioned cases (EVO andDEVO). This fact only can be explained by migration of triglyceridesfrom pork adipose tissue to the frying medium (PEVO) during eachfrying experiment, thus modifying the composition of the latter.

This hypothesis has been proved. As above commented, beforeand after each frying experiment the molar percentage of the dif-ferent kinds of acyl groups was determined, and greater changeswere observed in the molar percentage of oleic (or monounsatu-rated) acyl groups, during each frying experiment (which takesone minute) than between frying experiments (taking 2.5 h).Fig. 2e represents the molar percentage of oleic (or monounsatu-rated) acyl groups before and after each frying experiment, show-ing that the data before frying are aligned over the data after frying,both lines being parallel (having almost the same slope). Data inthis figure indicate that the molar percentage of oleic (or monoun-saturated) acyl groups in PEVO decreases suddenly after each fry-ing experiment (effect of migration of triglycerides) and muchmore slowly between frying experiments (effect of thermal degra-dation). Which is to say that in each frying experiment PEVO is en-riched with triglycerides coming from pork adipose tissue, in sucha way that the molar proportions of the acyl groups in PEVO aremodified, diminishing the molar percentage of oleic (or monoun-saturated) and increasing that of linolenic, linoleic and saturatedacyl groups in agreement with the proportions of these groups inthe lipids of pork adipose tissue; the effect of thermal degradationshould be added to this.

This same conclusion is reached from data referring to the lossof weight of the pork adipose tissue during frying, a fact also ob-served in previous studies on pork loin chops submitted to frying(Ramírez & Cava, 2005). The average loss in each frying experi-ment is around 32.5 g, presumably due to water and lipids. Theeffect of the incorporation of the pork adipose tissue lipids intothe frying medium was estimated taking into account the compo-sition of the extra virgin olive oil, that of the original pork adiposetissue, as well as the loss of weight of this latter; in this estima-tion it was assumed that the loss of water and lipids in the porktissue was proportional to their concentration in the original porktissue. In this way, the molar percentage of the different kinds ofacyl groups in this hypothetical frying medium, after fourteen fry-ing experiments, assuming no degradation, is Ln, 0.9%, L, 6.8%, Oor MU, 77.6%, and S + M, 14.7%. This hypothetical frying mediumwould have a higher proportion of linolenic, linoleic and espe-cially of saturated acyl groups than the original EVO, and a muchsmaller proportion of oleic ones. Assuming a rate of thermal deg-radation, of the several acyl groups in this hypothetical fryingmedium, similar to that found in DEVO, the molar percentages,after 32.5 h of heating, of the acyl groups of this hypothetical fry-ing medium are (Ln, 0.5%, L, 5.0%, O or MU, 73.2%, and S + M,

Fig. 2. Evolution, in EVO (�), DEVO (h), PEVO (N) and SEVO (X) media, of the molar percentage of the groups: (a) linolenic (Ln); (b) linoleic (L); (c) oleic or monounsaturated(O or MU); (d) saturated plus modified (S + M). (e) Evolution of the molar percentage of monounsaturated (O or MU) acyl groups before (d) and after (s) each fryingexperiment in PEVO medium. (f) Evolution of the molar percentage of omega-3 except Ln before (d) and after (s) each frying experiment in SEVO medium. Averageexperimental points and lines corresponding to the regression equations.

A. Martínez-Yusta, M.D. Guillén / Food Chemistry 150 (2014) 429–437 433

21.2%) very similar to those found experimentally in PEVO (seeFig. 2a–d).

These results prove that main changes in the molar proportionsof acyl groups in PEVO are due to migration of the triglyceridesfrom pork tissue to the frying medium and that the rate of degra-dation suffered by unsaturated acyl groups in PEVO is similar tothat undergone in DEVO and EVO media. As consequence afterfourteen frying experiments of pork adipose tissue the molar per-centage of O (or MU) in PEVO is much lower than that in DEVO andEVO, whereas that of S + M is much higher (see Fig. 2c and d).

3.1.4. Deep-frying experiments of salmon filletsAs above, in the series of frying experiments of salmon fillets

the frying medium SEVO composition is affected by both migrationof salmon lipids and thermal degradation of the unsaturated acylgroups. In this case the migration is detected by the simple obser-vation of the 1H NMR spectra, as the enlargement of signal J shows;in this signal (the typical overlapped triplets of bis-allilic protons oflinoleic and linolenic groups) after the second frying experiment(2.5 h of heating), appears also overlapped the signal J0, due toother bis-allilic protons of omega-3 groups, amongst which are

docosahexaenoic and eicosapentaenoic groups. The observablepart of signal, J0 (see Fig. 1), increases in the spectrum of SEVO withthe number of frying experiments. The presence of this signal indi-cates that SEVO frying medium contains omega-3 acyl groups (dif-ferent from linolenic) coming from the migration of fish lipids. Themolar percentage of these omega-3 no linolenic groups in SEVO,before and after each frying experiment, can be calculated fromdata derived from 1H NMR spectra of SEVO and of salmon filletlipids. These have been represented in Fig. 2f; it can be observedthat data after frying are aligned over data before frying. This indi-cates that each frying experiment involves a clear increase of thiskind of acyl groups, and that, in spite of the thermal degradationits concentration increases as the number of frying experimentsincreases.

Like pork adipose tissue, salmon fillets also undergo loss ofweight during frying presumably due to water and lipids. Takinginto account the same assumptions above mentioned for pork adi-pose tissue an average loss of 9.6 g of salmon fillet lipids in eachfrying experiment has been estimated. From the composition of ex-tra virgin olive oil and of salmon fillets it is possible to determinethe molar percentage of the different kinds of acyl groups in a

434 A. Martínez-Yusta, M.D. Guillén / Food Chemistry 150 (2014) 429–437

hypothetical frying medium made up of the fish lipids added,during the fourteen frying experiments, to the original extra virginolive oil; the composition of this hypothetical medium is 0.6% ofomega-3 (no Ln), 0.8% of Ln, 6.7% of L, 79.9% of O or MU, and12.1% of S. Assuming a rate of degradation of the unsaturated acylgroups of this hypothetical medium similar to that of DEVO, after32.5 h of heating the acyl group molar percentage pattern will be0.3% of omega-3 (no Ln), 0.4% of Ln, 4.9% of L, 75.5% of O or MU,and 18.7% of S; these predicted values are very close to the foundin SEVO after fourteen frying experiments (see Fig. 2a–d). This con-stitutes an additional proof of the influence of the fish salmon lipidmigration on SEVO composition. The migration of fish lipids to thefrying media has also been observed by other authors (Sioen et al.,2006).

3.2. Compounds present in the frying media derived from degradationof acyl groups

It is evident from the above results that the heating of edibleoils at frying temperature brings about modifications of the longchains of acyl groups. In the classical scheme of oil oxidation, itis considered that the first kinds of functional groups formed inthe acyl group chains are hydroperoxides and also associatedconjugated dienic systems (Guillén & Ruiz, 2005b). However, inagreement with previous studies of oil degradation at frying tem-perature, none of the above mentioned functional groups weredetected (Guillén & Uriarte, 2009, 2012a,b,c) by this technique; ifthese functional groups are formed, at this temperature in the con-ditions of this study, they degrade very quickly in such a way thatthey do not accumulate in the oil. This fact is not only observed inEVO but also in DEVO, PEVO and SEVO. In spite of this, differentkinds of compounds coming from acyl groups degradation areformed; some of these, such as aldehydes, alcohols, and epoxides,are observable by 1H NMR.

3.2.1. Occurrence of aldehydes3.2.1.1. Aldehydes in EVO. Previous studies have shown that whenedible oils are submitted to frying temperature, aldehydes arethe earliest newly formed compounds detectable by 1H NMR(Guillén & Uriarte, 2009, 2012a,b,c). If the oil is extra virgin oliveoil the aldehydes detected by this technique in the oil liquid matrixare n-alkanals, (E)-2-alkenals, (E,E)-2,4-alkadienals and other onestentatively identified as 4-oxo-alkanals (Guillén & Uriarte, 2012a).This is also confirmed in the study of the EVO evolution throughoutheating, the first three types of aldehydes being detected after 2.5 hof heating and the fourth after 12.5 h. The concentration of the sev-eral kinds of aldehydes (Al) in the frying media, at any heatingtime, expressed as mmol per mol of triglyceride (TG), can be deter-mined by the equation

ð½Alðmmol=mol TGÞ� ¼ 4000 ðAAl=APÞÞ ð6Þ

where AAl is the area of the signal of the –CHO protons of the severalkind of aldehydes (indicated in Table 1, and shown in Fig. 1 with theletters S, T, U, and V), and AP is the area of the signal P, due to the –CH2OCOR protons of triglycerides (TG).

The concentration of the several kinds of aldehydes varies withthe heating time and both variables fit, with high correlation coef-ficients, to second or first degree equations which are representedtogether with the experimental data in Fig. 3a.

3.2.1.2. Aldehydes in DEVO, PEVO and SEVO. It might be thought thatsome aldehydes, other than those above mentioned, could beformed during the deep-frying experiments; however, the typesof aldehydes present in DEVO, PEVO or SEVO media are the sameas in EVO, although some differences have been found in their con-centrations; they are observable by 1H NMR spectroscopy after

2.5 h of heating, except 4-oxo-alkanals which signal appears inthe spectrum of DEVO, PEVO and SEVO media after 17.5, 15 and20 h of heating respectively, a heating time which is somewhat lar-ger than in EVO. In Fig. 3b–d are represented experimental concen-trations of the different kinds of aldehydes versus heating time andthe corresponding regression equations.

Data represented in Fig. 3 indicate that (E)-2-alkenals are in thegreatest concentration in all media followed by n-alkanals;although in the case of SEVO these two kinds of aldehydes are pres-ent in almost the same concentration. In summary, as data after32.5 h of processing indicate, PEVO medium is the poorest inaldehydes due to its low concentration of n-alkanals and of (E,E)-2,4-alkadienals, followed by SEVO medium due to its low concen-tration in (E)-2-alkenals, EVO being the richest in aldehydes of thefour media.

The fact that the concentration of aldehydes in DEVO, and espe-cially in PEVO and SEVO media is lower than in EVO could be dueto several causes. One of them could be the slightly shorter heatingtimes during which DEVO, PEVO and SEVO were submitted to thefrying temperature; this is a consequence of temperature goingdown in the system as a result of the introduction of the food ineach frying experiment. Another cause could be the greater facilityfor aldehydes to escape towards the atmosphere during frying thanduring heating, because of greater movement of the fatty mediumand water escape. And finally, another cause could be the estab-lishment of possible reactions, such as Maillard reactions, betweenaldehydes and some food components such as proteins.

3.2.2. Occurrence of epoxidesThe formation of epoxides (EPO) supported on acyl chains

during heating of extra virgin olive oil at frying temperature hasbeen observed by 1H NMR in previous studies (Guillén & Uriarte,2012a). These are detected by the appearance in the spectra ofmultiplets centered at 2.63 ppm ((E)-9,10-epoxystearyl groups)and at 2.88 ppm ((Z)-9,10-epoxystearyl groups) (see Table 1 andFig. 1); these functional groups are detected, after 7.5 h of heatingin the spectra of all frying media, later than aldehydes. The concen-tration of these functional groups was determined by using theequations

ð½E� EPO ðmmol=mol TGÞ� ¼ 2000 ðAI=APÞÞ ð7Þ

and

ð½Z � EPO ðmmol=mol TGÞ� ¼ 2000 ðAK=APÞÞ ð8Þ

where AI and AK are the areas of the corresponding signals indicatedin Fig. 1 and Table 1. The concentrations thus obtained and the cor-responding heating times fit, with high correlation coefficients, toequations which are represented in Fig. 4a and b together withexperimental points. It has been observed that there are no greatdifferences in concentration of these functional groups in the sev-eral media. However, in all cases the concentration of the epoxytrans-isomer is higher than that of the epoxy cis-isomer.

3.2.3. Occurrence of alcoholsIn the 1H NMR spectra of all media, after 7.5 h of heating at fry-

ing temperature, signals attributable to protons of secondary (SA)(multiplets centered near 3.56 ppm) and primary alcohols (PA)(triplet centered at 3.62 ppm) also appear (see Table 1 andFig. 1). In a previous study on extra virgin olive oil submitted tohigh temperature these signals, although observed, were neitherassigned nor considered (Guillén & Uriarte, 2012a). However, theformation of hydroxyl groups in edible oils submitted to fryingtemperature has been previously reported (Schwartz, Rady, &Castaneda, 1994). The concentration of these alcohol groups wasdetermined by using the equations

Fig. 3. Evolution of the concentration of total aldehydes (�), n-alkanals (X), (E)-2-alkenals (h), (E,E)-2,4-alkadienals (N), and 4-oxo-alkanals ( ) throughout the processingtime in: (a) EVO medium; (b) DEVO medium; (c) PEVO medium; (d) and SEVO medium. Average experimental points and lines corresponding to the regression equations.

A. Martínez-Yusta, M.D. Guillén / Food Chemistry 150 (2014) 429–437 435

ð½SA ðmmol=mol TGÞ� ¼ 4000 ðAL=APÞÞ ð9Þ

and

ð½PA ðmmol=mol TGÞ� ¼ 2000 ðAM=APÞÞ ð10Þ

where AL and AM are the area of the signals L and M indicated inFig. 1 and Table 1. This concentration increases with the heatingtime in all frying media and both parameters fit, with high correla-tion coefficients, to equations represented in Fig. 4c and d togetherwith experimental points. It can be observed that the concentra-tions of alcohols are similar in the four media and that of secondaryalcohols is much higher than that of primary alcohols, and of thesame order as that of (E)-9,10-epoxystearyl groups.

3.3. Evaluation of the occurrence of hydrolytic reactions in extra virginolive oil submitted to deep-frying processes

Hydrolysis has been cited as being amongst the multiplereactions that edible oils can undergo at frying temperature, (Choe& Min, 2007) giving fatty acids and di- or mono-glycerides. How-ever, in the original oil the main components are triglycerides,whereas di- and mono-glycerides, if they are present, are in verylow concentrations.

Observation of the 1H NMR spectra of the extra virgin olive oilthroughout the heating (EVO medium), and of the same oil submit-ted to the several series of frying experiments (DEVO, PEVO andSEVO media) could lead to the conclusion that hydrolytic reactionsoccur to a small extent in these processes.

3.3.1. Occurrence of 1,2-diglyceridesThe original extra virgin olive oil used in this study contains

1,2-diglycerides, as the doublet at 3.71 ppm of its 1H NMR

spectrum shows (see signal N in Fig. 1 and Table 1), the signalsof 1,3-diglycerides and also 1- and 2- mono-glycerides being ab-sent (Sopelana et al., 2013). The initial concentration of 1,2-digly-cerides in this oil in relation to that of triglycerides (TG) is of13 mmol/mol TG; this has been determined by the equation

ð½1;2� DG ðmmol=mol TGÞ� ¼ 2000 ðAN=APÞÞ ð11Þ

where AN is the area of the signal N indicated in Fig. 1 and Table 1.When this oil is heated at 190 �C in absence of food (EVO med-

ium) firstly the concentration of 1,2-DG, in the first 2.5 h of heat-ing, decreases reaching a concentration around 7 mmol/mol TG.From this heating time onwards, the concentration of these com-pounds increases very slowly, to reach, in the following thirtyhours, concentrations which are slightly lower than the original(see Fig. 5a). It can be observed that after 32.5 h of heating SEVOmedium is slightly richer in 1,2-diglycerides than the other media,followed by DEVO, this latter being slightly different from EVO andPEVO. Taking into account that water is required for this reaction,these small differences could be associated to the different concen-trations of water in the food submitted to frying, salmon filletsbeing the richest, followed by dough of Spanish doughnuts.

3.3.2. Occurrence of 1,3-diglyceridesThe formation of 1,3-diglycerides involves triglycerides under-

going hydrolytic reactions. Despite signals of 1,3-diglyceridesbeing very small in the spectrum of the original oil, they increaseslightly with the heating time in all frying media. These signals,indicated with the letter O in Fig. 1 are also given in Table 1. AsFig. 1 shows, these signals are very small and overlap partially withsignal P, making their quantification impossible. In spite of this, thearea of the not overlapped part has been determined in order toknow the rate of increase with the heating time in the different

Fig. 4. Evolution, with the processing time, in EVO (�), DEVO (h), PEVO (N) and SEVO (�) media of the concentration of: (a and b) (E)- and (Z)-9,10-epoxystearyl groups (E- orZ-9,10-EPO); (c and d) primary (PA) and secondary alcohols (SA). Average experimental points and lines corresponding to the regression equations.

436 A. Martínez-Yusta, M.D. Guillén / Food Chemistry 150 (2014) 429–437

media. The obtained results indicate that at the same processingtime, once again SEVO followed by DEVO are the media in whichthe concentrations of 1,3-diglycerides are higher, the smaller beingin EVO and PEVO media; this could be due to the same reasonsabove mentioned for 1,2-diglycerides.

3.3.3. Occurrence of fatty acidsThese can be formed in hydrolytic processes. The concentration

of fatty acids is very small in the original oil. These compounds areobservable in the spectrum through a part of its triplet signal oftheir –CH2 protons in alpha position in relation to the carboxylicgroup (indicated by H0 in Fig. 1); this signal overlaps with the trip-lets of the same kind of protons of acyl chains of triglycerides; forthis reason, and due to the small abundance of fatty acids, this sig-nal cannot be used for a correct quantification of its concentration.However from this partial signal H0 information of the changes in

Fig. 5. Evolution, with the processing time, in EVO (�), DEVO (h), PEVO (N) and SEVO (�(c) and percentage in weight of polar compounds (PC%). Average experimental points a

the concentration of fatty acids in the several media can be ex-tracted. To this aim the area of H0 signal was determined in thespectra of the several media at different processing time and thevalues obtained fit well with the heating time, allowing one topredict in which media the concentration would be higher after32.5 h of processing. The results obtained indicate again that SEVOand DEVO are the richest media in fatty acids, whereas PEVO is thepoorest. This agrees with the above observed in relation to 1,2- and1,3-diglycerides.

3.4. Evolution of Iodine Value of the several media throughout theprocessing time

It is known that Iodine Value decreases in oils submitted tofrying temperature as a consequence of the disappearance of theacyl group double bonds due to polymerisation and oxidation

) media of the concentration of: (a) 1,2-diglycerides (1,2-DG); (b) Iodine Value (IV);nd lines corresponding to the regression equations.

A. Martínez-Yusta, M.D. Guillén / Food Chemistry 150 (2014) 429–437 437

reactions. Iodine Value is easy to determine from the integration ofthe olefinic protons signal in the 1H NMR spectra, as it has beencommented before. In this way the Iodine Value of the severalmedia was determined throughout the processing time. Bothparameters (Iodine Value and heating time) fit, with high correla-tion coefficients, to equations represented in Fig. 5b, together withthe average experimental points. It can be observed that PEVOshows the highest rate of decrease of IV, whereas SEVO showsthe lowest rate, EVO and DEVO having an intermediate rate. Thisis because the concentration of omega-3 polyunsaturated acylgroups in SEVO medium increases progressively with the process-ing time as a consequence of fish lipid migration to this medium,thus reducing the loss of unsaturation; however the concentrationof saturated acyl groups in PEVO medium increases progressivelyas a consequence of the migration of pork adipose tissue lipids tothis medium, thus increasing its saturation. It is evident thatalthough the degradation reached by thermal effect can be similarin the four media, they show different iodine values; this showsthe weakness of this parameter to indicate the thermal degrada-tion reached by frying media.

3.5. Evolution of the percentage in weight of polar compounds in theseveral media throughout the processing time

The percentage in weight of the polar compounds of the severalmedia increases progressively with the processing time due to thegrowing concentration of polar compounds as oil degradation ad-vances; however in PEVO and SEVO media this evolution is alsoinfluenced by the food lipid migration. The experimental PC% dataand the regression equations that relate these data and heatingtime are represented in Fig. 5c. As can be observed, PEVO showsthe lowest rate of increase of PC%, whereas DEVO and EVO showthe highest values and SEVO a value very close to these latteralthough slightly smaller. The influence of lipid migration fromthe food to the frying medium on this parameter seems to beimportant.

4. Conclusions

In the deep-frying process, when extra virgin olive oil is used,the changes undergone by this frying medium are due firstly tothe degradation of the oil components by thermal effects; in addi-tion, the composition of the food involved in the frying process alsoinfluences the changes in the frying medium composition. If thefood contains lipids, these migrate to the frying medium, thusmodifying its composition; in turn, the lipids coming from the foodundergo thermal degradation, also contributing to the modificationof the frying medium. Furthermore, reactions between frying med-ium and food components cannot be excluded from also contribut-ing to changes in the composition of the former. Some of thechanges in the frying medium composition are not only observedin the concentration of main components, such as the compositionpattern of acyl groups, but also in the concentration of newlyformed oxygenated compounds, mainly in that of aldehydes; like-wise, although in a very subtle way, the hydrolysis degree of thefrying medium is affected by food composition. All these changes,in their turn, influence in different ways on parameters such as Io-dine Value and percentage in weight of polar compounds. This isthe first time that the study of the compositional changes under-gone by the frying medium during deep-frying has been carriedout by 1H NMR spectroscopy. It pays attention, in a global way,to not only the thermal effect but also to the influence of foodcomposition. It covers modifications due not only to the acyl grouppattern composition, but also to the newly formed oxygenated

compounds and hydrolytic reactions, and even to their unsatura-tion degree.

Acknowledgements

This work has been supported by the Spanish Ministry of Econ-omy and Competitiveness (MINECO, AGL2012-36466), the BasqueGovernment (GIC10/IT-463-10), and the University of the BasqueCountry (UPV/EHU, UFI-11/21). A. Martínez-Yusta thanks the Bas-que Government for a predoctoral fellowship.

References

Andrikopoulos, N. K., Kalogeropoulos, N., Falirea, A., & Barbagianni, M. N. (2002).Performance of virgin olive oil and vegetable shortening during domestic deep-frying and pan-frying of potatoes. International Journal of Food Science andTechnology, 37(2), 177–190.

Choe, E., & Min, D. B. (2007). Chemistry of deep-fat frying oils. Journal of FoodScience, 72(5), R77–R86.

Dobarganes, C. M., & Márquez-Ruiz, G. (2007). Formation and analysis of oxidizedmonomeric, dimeric and higher oligomeric triglycerides. In M. D. Erickson (Ed.),Deep frying: Chemistry, nutrition and practical applications (2nd ed., pp. 87–110).Champaign: AOCS Press.

Guillén, M. D., & Ruiz, A. (2003a). Edible oils: Discrimination by 1H nuclear magneticresonance. Journal of the Science of Food and Agriculture, 83(4), 338–346.

Guillén, M. D., & Ruiz, A. (2003b). Rapid simultaneous determination by protonNMR of unsaturation and composition of acyl groups in vegetable oils. EuropeanJournal of Lipid Science and Technology, 105(11), 688–696.

Guillén, M. D., & Ruiz, A. (2005a). Study by proton nuclear magnetic resonance ofthe thermal oxidation of oils rich in oleic acyl groups. Journal of American OilChemists’ Society, 82(5), 349–355.

Guillén, M. D., & Ruiz, A. (2005b). Oxidation process of oils with high content oflinoleic acyl groups and formation of toxic hydroperoxy and hydroxyalkenals. Astudy by 1H nuclear magnetic resonance. Journal of the Science of Food andAgriculture, 85(14), 2413–2420.

Guillén, M. D., & Ruiz, A. (2005c). Monitoring the oxidation of unsaturated oils andformation of oxygenated aldehydes by proton NMR. European Journal of LipidScience and Technology, 107(1), 36–47.

Guillén, M. D., & Uriarte, P. S. (2009). Contribution to further understanding of theevolution of sunflower oil submitted to frying temperature in a domestic fryer:Study by 1H nuclear magnetic resonance. Journal of Agricultural and FoodChemistry, 57(17), 7790–7799.

Guillén, M. D., & Uriarte, P. S. (2011). A very simple, fast, and non-destructiveapproach to predict the time at which edible oils submitted to high temperaturereach the established limits of safety. Food Chemistry, 127(2), 802–806.

Guillén, M. D., & Uriarte, P. S. (2012a). Study by 1H NMR spectroscopy of theevolution of extra virgin olive oil composition submitted to frying temperaturein an industrial fryer for a prolonged period of time. Food Chemistry, 134(1),162–172.

Guillén, M. D., & Uriarte, P. S. (2012b). Monitoring by 1H nuclear magneticresonance of the changes in the composition of virgin linseed oil heated atfrying temperature. Comparison with the evolution of other edible oils. FoodControl, 28(1), 59–68.

Guillén, M. D., & Uriarte, P. S. (2012c). Simultaneous control of the evolution of thepercentage in weight of polar compounds, iodine value, acyl groups proportionsand aldehydes concentrations in sunflower oil submitted to frying temperaturein an industrial fryer. Food Control, 24(2), 50–56.

Kalogianni, E. P., Karapantsios, T. D., & Miller, R. (2011). Effect of repeated frying onthe viscosity, density and dynamic interfacial tension of palm and olive oil.Journal of Food Engineering, 105(1), 169–179.

Ramírez, M. R., & Cava, R. (2005). Changes in colour, lipid oxidation and fatty acidcomposition of pork loin chops as affected by the type of culinary frying fat.LWT – Food Science and Technology, 38(7), 726–734.

Sánchez-Gimeno, A. C., Negueruela, A. I., Benito, M., Vercet, A., & Oria, R. (2008).Some physical changes in Bajo Aragon extra virgin olive oil during the fryingprocess. Food Chemistry, 110(3), 654–658.

Schwartz, D. P., Rady, A. H., & Castaneda, S. (1994). The formation of oxo- andhydroxy-fatty acids in heated fats and oils. Journal of the American Oil Chemists’Society, 71(4), 441–444.

Sioen, I., Haak, L., Raes, K., Hermans, C., De Henauw, S., De Smet, S., et al. (2006).Effects of pan-frying in margarine and olive oil on the fatty acid composition ofcod and salmon. Food Chemistry, 98(4), 609–617.

Sopelana, P., Arizabaleta, I., Ibargoitia, M. L., & Guillén, M. D. (2013).Characterization of the lipidic components of margarines by 1H nuclearmagnetic resonance. Food Chemistry, 141(4), 3357–3364.

Vidal, N. P., Manzanos, M. J., Goicoechea, E., & Guillén, M. D. (2012). Quality offarmed and wild sea bass lipids studied by 1H NMR: Usefulness of thistechnique for differentiation on a qualitative and a quantitative basis. FoodChemistry, 135(3), 1583–1591.