Improved Quantification by Nuclear Magnetic Resonance Spectroscopy of the Fatty Acid Ester Composition of Extra Virgin Olive Oils

Analysis of foods, which are typically highly complex mixtures, by 1H NMR can be difficult because the prevalence of signal overlap complicates characterization and quantification. The various components of a food sample may have a wide range of concentrations, leading to a high dynamic range NMR spectrum and complicating the analysis of less concentrated species. One source of this complication is the presence of 13C satellites, peaks that appear either side of a parent peak with ∼0.56% of its intensity. Satellites of concentrated species can easily be comparable in intensity to the signals of minor components, and can partly or wholly obscure them. This is commonly seen in olive oil samples, leading to inaccurate calculation of the fatty acid ester composition of the oil, used for determining the quality of edible oils and for detecting adulteration. Here, we show that the recently introduced Destruction of Interfering Satellites by Perfect Echo Low-pass filtration (DISPEL) experiment is able to suppress 13C satellites and can substantially improve the accuracy of integration of minor signals. The DISPEL experiment does not require any complicated optimization, working “out of the box” with standard parameters, and incurs no significant loss of sensitivity. It has the potential to become the default experiment, replacing conventional 1D 1H NMR, for quantitative analysis of olive oil.


Table of Contents
Experimental details Table S1. Phase cycle for each pulse and the receiver in the DISPEL experiment.

Pulse
Phase Cycle . DISPEL pulse sequence used here. The narrow and wide black rectangles represent 90⁰ and 180⁰ pulses, respectively. The τ delay was set to 4.05 ms, and τ1, τ2, τ3 and τ4 to 3.2, 1.1, 3.95 and 1.56 ms, respectively, giving excellent satellite suppression for all 1 H-13 C coupling constants from 120 to 360 Hz. 1 EVOO 1 spectra Figure S2. Full 1 H NMR spectrum of the EVOO 1 sample.
The two spectral regions displayed in Figure S3 are regions of the EVOO 1 spectrum where the 13 C satellites of H and E/F, respectively, overlap with signals of minor components marked with asterisks in Figures S3c and d. These signals are only revealed after satellite suppression by DISPEL, demonstrating how characterisation, or even observation, of these signals would be very difficult with only a 1 H pulse-acquire experiment. The methylene protons sn-1,3 of signal H are strongly coupled to each other, as can be seen by the "roofing" effect in the 1 H spectrum in Figure S4a, distorting the multiplets from the 1:1:1:1 relative peak inetnsities expected in the case of weak coupling. When DISPEL is used, the roofing effect is reversed, as shown in Figure S4b. Figure S4. 500 MHz (a) conventional 1 H NMR and (b) DISPEL spectra showing the effects of DISPEL on the diastereotopic CH2s of H of the EVOO 1 sample. The roofing in (a) caused by strong coupling between the two protons is reversed by the DISPEL sequence, as shown in (b). Integration of (b) would give a slightly different value, but as H is not used for the FEC determination here this is not a problem.

AMX spectra
A sample of 2-bromothiophene (AMX spin system) is used to demostrate the effectiveness of the DISPEL experiment at surpressing 13 C satellites and for quantitation purposes. Figure S5c shows the DISPEL spectrum of the AMX sample and Figure S5d the spectrum with GARP-4 13 C composite pulse decoupling. The superiority of the DISPEL experiment is clear; there is no loss in resolution, whereas the decoupling experiment suffers because of the limited acquisition time usable. Because of the complexity of EVOO spectra it is difficult to show that both I and J lose an integral value equal to that of the satellites they overlap with when using DISPEL, but the equivalent change can be demonstrated straightforwardly in the simpler spectrrum of 2-bromothiophene, as seen in Table S2. The relative integral of of the S4'+UI region decreases in the DISPEL experiment by almost exactly the integral of S4.

EVOO data fitting
FECs of the five EVOO samples were determined by minimising the sum of the squares of the differences between the experimental values of the integral ratios in the right hand sides of equations (1) to (5) and the integral ratios back-calculated from trial compositions. Tables S3 and S4 show the experimental integral ratios, back-calculated ratios and residuals for data from the conventional 1 H spectra and the DISPEL spectra. Only three of the parameters are over-determined (because of the choice in reference 25 to integrate methyl protons spearately from methylene) so the residuals are zero for integral ratio (1). eq (1) eq (2) eq (3) eq (4) eq (5) EVOO sample Experimental ratios