**1. Introduction**

Olive oil is considered to be one of the best edible oils and an essential component in the Mediterranean diet due to extraordinary organoleptic qualities and a large number of health benefits. According to the *Codex alimentarius* of the Food and Agriculture Organization of the United Nations (FAO) [1], olive oils are classified in three categories: virgin olive, refined olive, and refined olive-pomace oils. These, in turn, are divided into different grades depending on their organoleptic qualities, median defects, and color, among other attributes. From the hierarchy list of grades among these categories, extra virgin olive oil (EVOO) is considered to have the highest nutritional value with various health benefits. Among its nutritional properties, EVOO possesses high antioxidant activity [2,3], exhibits anti-inflammatory effects [3,4], improves the metabolism of carbohydrates in patients with type-2 diabetes [5–8], reduces blood pressure and the risk of hypertension [7,9], and improves vasodilation [10,11], to name a few. These many health benefits have boosted the popularity of olive oil in recent decades [12], although this popularity has also brought about other problems associated with the adulteration and/or deliberate mislabeling of EVOO [13,14]. One of the principal motivations for olive-oil fraud is the large price gap

**Citation:** Chavez-Angel, E.; Puertas, B.; Kreuzer, M.; Soliva Fortuny, R.; Ng, R.C.; Castro-Alvarez, A.; Sotomayor Torres, C.M. Spectroscopic and Thermal Characterization of Extra Virgin Olive Oil Adulterated with Edible Oils. *Foods* **2022**, *11*, 1304. https:// doi.org/10.3390/foods11091304

Academic Editor: Daniel Cozzolino

Received: 14 April 2022 Accepted: 28 April 2022 Published: 29 April 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

between EVOO and other non-olive oils or even between EVOO and other types of olive oils. Due to its relative scarcity and high production/selling price, unscrupulous processors have been fined for adulterating EVOO with large amounts of cheaper oils. EVOO itself is a vegetal fat with high levels of monounsaturated fatty acids (e.g., 78%) and low levels of saturated acids (e.g., 14%), in contrast to cheap seed oils (e.g., sunflower, corn, and soybean), which have high levels of polyunsaturated fats [1]. Consequently, adulteration with other oils results in the loss of many of the healthy properties of EVOO.

There is a long list of properties that can be tested to ensure the quality of EVOO [15–17]. The standard and official methods to characterize EVOO are gas chromatography (GC) and high-performance liquid-chromatography (HPLC). GC is mainly used to determine the composition of the saponifiable fraction, which contains fatty acids and their derivatives, as well as the unsaponifiable fraction, which contains waxes, aliphatic alcohols, tocopherols, and phenolic compounds, among others. On the other hand, HPLC is mainly used to determine the structure of triglycerides, the quantity of pigments such as chlorophylls and carotenes, and other quality parameters (other than purity). Apart from these official methods, there are a number of alternative and complementary methods that have been suggested over the past decade. Among them, infrared and Raman spectroscopy are gaining attention [18–22].

This work aims at evaluating the ability to detect traces of adulteration in EVOO with three spectroscopic techniques: Raman, photoluminescence (PL), and Fourier-transform infrared (FTIR) spectroscopies. In addition, we explore the use of thermal conductivity as a potential new parameter to be used as a detection tool. Despite its relative measurement simplicity, thermal conductivity has, to date, been overlooked as a figure of merit to determine food purity. The combination of all of these techniques provides an easy method, free of sample pre-processing, to ascertain the quality and authenticity of food.

#### **2. Materials and Methods**

Twenty samples of EVOO were intentionally adulterated using five different types of edible oils: sunflower (La Masia, "masiasol", Sevilla, Spain), high oleic (HO) sunflower (Carrefour, "Aceite refinado de girasol", Madrid, Spain), 95–5% soybean–nut blend (La española, "Soy plus", Jaen, Spain), corn (Coosol, "Maiz", Jaen, Spain), and olive-pomace (Carrefour, "Aceite refinado de orujo de oliva", Madrid, Spain), in volume concentrations of 5%, 10%, 20%, and 50%. A single type of EVOO (Salvatge "Les Garrigues", Lleida, Spain) was adulterated, and the oil was provided directly from the factory to guarantee its purity. All samples were stored in a dry place protected from light to preserve their quality (see Figure S1a, Supplementary Materials).

The Raman and photoluminescence spectra were recorded using the same equipment (a T64000 Raman spectrometer using a liquid-nitrogen-cooled Symphony CCD manufactured by HORIBA Jobin Yvon, Chilly-Mazarin, France) optimized in the visible regime (400–800 nm). It was used in single-grating mode with 2400 and 300 lines per mm and a spectral resolution of at least 0.4 cm<sup>−</sup><sup>1</sup> and 0.2 nm for Raman and photoluminescence, respectively. The use of 2400 lines for Raman measurements provides a very high frequency resolution at the cost of a small frequency window. On the other hand, 300 lines allow for a larger spectroscopic window which is ideal for the broad PL signal. The measurements were performed by focusing a diode laser (532 nm) onto a transparent quartz cuvette with a 10× long working distance microscope objective (see Figure S1b,c, Supplementary Materials). The power of the laser was kept as low as possible (<0.5 mW) to avoid any possible damage from self heating of the samples. For the photoluminescence measurements (also known as fluorescence spectroscopy), all samples were measured using 3 accumulations with the same integration time of 0.3 s with a fixed focal plane, to allow for direct comparison between each sample. For each sample, 5 to 10 spectra were recorded at positions on the sample.

FTIR spectra were recorded (64 co-added scans) by a Hyperion 3000 infrared (IR) microscope coupled to a Vertex 70 spectrometer manufactured by Bruker (Billerica, MA,

USA) at the infrared beamline MIRAS of the ALBA synchrotron [23]. Data was recorded with a liquid-nitrogen-cooled MCT detector. A 2–5 μL drop of oil was placed on the center of a piece of ZnSe glass and pressured with a second slide to create a homogenous oil film. The setup was used in the transmission configuration with a spectral resolution of 4 cm<sup>−</sup><sup>1</sup> with a Globar as the infrared light source. The IR light was focused onto the ZnSe slide using a 30× Schwarzschild condenser and collected with a matching objective.

Principal component analysis (PCA) was used to treat the FTIR spectra using the software Orange Data Mining [24]. For each sample, 50 spectra at different sample positions were recorded and concatenated in a large matrix, as displayed in Table S1 in the supplementary information. Prior to the calculation, baseline corrections, spectral normalization, and Savitzky–Golay filters (for smoothing) and derivatives (to reduce scatter effects), were applied to process the spectra (see Figure S2, Supplementary Materials).

The thermal conductivity (*k*) was determined by the bidirectional three-omega (3ω) method [25,26] over the temperature range T = 298–400 K. The bidirectional 3ω method is based on the measurement of a rise in temperature that is produced by an alternating current (AC) passing through a metallic strip via the Joule heating effect. The metal line is composed of four rectangular pads connected by pins to a narrow wire that is used simultaneously as both a heater and temperature sensor (see Figure S1d, Supplementary Materials). The two outer pads are used to apply the AC current while the inner pads are used to measure the third component voltage (3ω-voltage), which contains the information regarding the temperature rise Δ*T*. Metal heaters (Cr:Au, 5:95 nm) were deposited by physical vapor deposition onto quartz substrates (5 × 5 × 0.5 mm3). For the measurement, a drop of oil (~10 μL) was placed on top of the 3ω heater. First, an empty 3ω cell was measured (reference). Then, a second measurement took place using the same cell after the sample to be studied was placed on top of the heater. Assuming that heat transfer occurs only across sample-heater-substrate interfaces, the total measured temperature change of the heater (Δ*TTotal*) is given by [25]:

$$\frac{1}{\Delta T\_{Total}} = \frac{1}{\Delta T\_{Sample}} + \frac{1}{\Delta T\_{Substrate}}\tag{1}$$

where Δ*TSample* and Δ*TSustrate* correspond to the temperature fluctuations induced by the sample (oil) and substrate (reference) located at the top and the bottom of the heater, respectively. Lubner et al. [26] showed that the error associated with this interface assumption (Equation (1)) can be as small as 1% if three experimental conditions are fulfilled: (i) the ratio of the thermal diffusivities of the sample (*α*oil) and the substrate (αSub), <sup>α</sup>oil/<sup>α</sup>Sub > <sup>10</sup>−1; (ii) the ratio of the thermal conductivities is in the range 10−<sup>2</sup> < *k*oil/*k*Substrate < 1; and (iii) the excitation frequencies are <100 Hz (low frequency limit). In our case, the room-temperature thermal diffusivity of the oils fluctuated within the range of (0.5–0.8) × 10−<sup>7</sup> m<sup>2</sup> s<sup>−</sup><sup>1</sup> [27,28], while that of the quartz fluctuated within the range of (0.8–1) × 10−<sup>7</sup> m<sup>2</sup> s<sup>−</sup><sup>1</sup> [29], i.e., 1 > <sup>α</sup>oil/<sup>α</sup>Sub > 0.5. The *k* of quartz is ~1.2–1.4 W m<sup>−</sup><sup>1</sup> K−<sup>1</sup> [29,30], and the *k* of oils was ~0.15–0.17 W m<sup>−</sup><sup>1</sup> K−<sup>1</sup> [31,32], i.e., *koil*/*kSubstrat*e < 1. The frequency range used here was (5–100) Hz, which falls within the low frequency limit.

The relationship between the temperature rise and the heat generation rate can be expressed as [33,34]:

$$
\Delta T = \frac{P}{l\pi k} \int\_0^\infty \frac{\sin^2(xb)}{\left(xb\right)^2 \sqrt{\mathbf{x}^2 + iq^2}} dx \tag{2}
$$

$$q = \sqrt{4\pi f/a} = \sqrt{4\pi f C\_V/k} \tag{3}$$

$$
\Delta T = \Delta T\_X + i\Delta T\_Y \tag{4}
$$

where Δ*T* is the complex temperature rise oscillation; *b* and *l* are the heater's half width (5 μm) and length (1 mm), respectively; *q* is the inverse of thermal penetration depth; *CV* is the volumetric heat capacity; *i* = √−1 is the imaginary number; *f* is the excitation

frequency; and *P* is the AC power. The real and the imaginary parts are proportional to the in-phase ('X') and quadrature ('Y') components of three-omega voltage.

Finally, the thermal conductivity of the oils was found by least square fitting of the inphase signal using *k* and *CV* as fitting variables. A detailed description of the bidirectional technique and the full development of the equations can be found in the supporting information of our previous works [35,36].
