Development of a Multianalyte Method for the Determination of Phenolic Compounds in Residues from Beverage Production Using a Cyclofructan-Based Column in HILIC-MS/MS

Abstract

This study expands and shows the versatility of applications in hydrophilic interaction chromatography coupled with triple quadrupole mass spectrometry (HILIC-MS/MS) using a FRULIC-N column to determine 31 phenolic compounds in beverage industry residues (coffee, apple juice, beer, and wine). Optimization, via Doehlert design, determined the acetonitrile proportion (50-90%) and pH (3.0-7.0) in the mobile phase (MP). A 2² factorial design assessed the concentration of ammonium acetate (10-110 mmol L^-1) in the MP and column temperature (20-60 °C). Injection volume (5-45 µL) was univariately optimized. Separations at 60 °C reduced chromatographic run time to less than 11 min. Gradient elution (500 µL min^-1) employed an MP comprising acetonitrile and ammonium acetate (110 mmol L^-1, pH 7). Validation demonstrated average coefficient of determination (R²) 0.9707, limit of detection 0.001-0.503 mg L^-1 and quantification 0.004-1.524 mg L^-1, precision < 12.2%, recovery 92.4-110.8%. The method was applied to five residues, indicating its viability for determining phenolics in beverage production residues.

Keywords:
HILIC; phenolic compounds; by-products; beverage industry; method validation

Introduction

In recent years, free radical chemistry has gathered considerable attention due to its association with aging and related diseases. The human body possesses an intricate defense mechanism against free radicals, which are consistently generated during normal cellular metabolism and various pathological events. Free radicals can be stabilized or neutralized by antioxidant compounds, which may be either endogenous or derived from dietary sources. In the latter scenario, certain foods may contain diverse types and concentrations of antioxidants in their composition, each offering distinct health benefits.¹,²,³,⁴

Phenolics are bioactive compounds with antioxidant potential, produced as secondary metabolites in plants.⁵ Residues generated by the beverage industry represent a promising source of phenolic compounds. However, in this process these materials are essentially treated as animal feed, due to their high protein and fiber content, or simply discarded in landfills.⁶ The recovery of phenolic compounds from food residues has drawn considerable attention due to their wide array of properties, including anti-inflammatory, antioxidant, and antimicrobial activities.⁷,⁸ Considering this, extracts rich in bioactive compounds obtained from beverage production residue can be used as ingredients in food and pharmaceutical matrices during the development of new products or in replacement of synthetic antioxidants. Concurrently, their reuse yields environmental, social, and economic advantages.⁹

These compounds represent a chemically diverse group, making the quantification of phenolic phytochemical content challenging. The analytical methods documented in the literature¹⁰,¹¹ encompassing the quantification of total phenolics, as well as individual assessment and/or analyses of specific categories of phenolics predominantly employ reversed phase liquid chromatography (RP-LC) with diode array detector (DAD) or mass spectrometry (MS). In these cases, the use of RP-LC generally involves long analysis time. In addition, the highly aqueous mobile phase (MP) commonly used in RP-LC, can hinder the desolvation process in electrospray ionization-mass spectrometry (ESI-MS) reducing the sensitivity.¹¹,¹²,¹³,¹⁴

Another separation mode with interesting characteristics for the separation of polar compounds that has been extensively studied in recent years is the hydrophilic interaction liquid chromatography (HILIC). This mode provides an alternative approach to effectively separate polar compounds on polar stationary phase (SP) and has evolved a lot with the development of new types of SP. In particular, the FRULIC-N column draws attention and has potential to separate phenolic compounds by HILIC considering its excellent hydrophilicity attributed to the abundance of hydroxyl groups in native cyclofructan-6 (CF6).¹⁵,¹⁶ FRULIC-N was originally developed for the separation of nucleic acid compounds, derived from purines, and carbohydrates, and to date, there is no application of this column for the separation of multiple phenolics in the literature.¹⁶ Other methods commonly use SP based on pure silica, zwitterionic functionalities, amide, and cross-linked diol for analysis of specific categories of phenolic compounds.¹⁷

The HILIC mode combines a hydrophilic SP with a MP composed of a high percentage of organic solvent, generally acetonitrile, and a small amount of pure water or buffer. This system provides a mixed retention mechanism that involves hydrogen bonding, dipolar interactions, electrostatic interactions, adsorption, and also partitioning between the MP and the water-rich layer that is solvating the surface of the SP.¹⁸,¹⁹ The use of more volatile MP increases the sensitivity of the MS, which is a great advantage for the determination of phenolic compounds. Thus, the development of new methods for phenolic compounds exploring HILIC can increase the amount of phenolics determined, increase the sensitivity, and reduce separation time when compared to other methods from the literature.¹¹,¹³,¹⁴,²⁰,²¹,²² To access the optimal separation conditions for these new HILIC methods, important separation parameters such as additives on the MP, temperature, and pH can be efficiently selected using design of experiments. This approach stands out among some works in the literature,¹¹,¹³,¹⁴,²²,²³ as it can allow the observation of interaction effects between the factors that affect the separations, and the simultaneous optimization controls analyte retention, selectivity, and peak shape.

This paper proposes a method for the simultaneous determination of 31 phenolic compounds using a FRULIC-N column in HILIC-MS/MS. Designs of experiments were employed to select the proportion of acetonitrile, pH, concentrations of additives in the MP, and the column temperature. The optimized method was validated according to the Eurachem²⁴ and Association of Official Analytical Chemists (AOAC).²⁵ Finally, after ensuring the reliability of the results, the new HILIC-MS/MS method was applied for the determination of the phenolic compounds in various residues generated by the beverage industry.

Experimental

Reagents and solutions

All reagents utilized were of analytical grade. The water employed was purified using a Milli-Q system (Millipore, Bedford, USA) with a resistivity of 18.2 MΩ cm at 25 °C. Methanol and acetonitrile solvents were acquired from Merck (Darmstadt, Germany), while ethanol and ethyl ether were obtained from Vetec (Rio de Janeiro, Brazil). Ammonium formate, acetic acid, and ammonium hydroxide reagents were purchased from Sigma-Aldrich (St. Louis, USA). Analytical standards, including 4-methylumbelliferone, vanillin, coniferaldehyde, protocatechin acid, umbelliferone, gallic acid, vanillic acid, syringaldehyde, caffeic acid, ferulic acid, scopoletin, sinapaldehyde, resveratrol, sinapic acid, chrysin, pinocembrin, naringenin, kaempferol, catechin, hispidulin, quercetin, taxifolin, chlorogenic acid, rosmarinic acid, isoquercetrin, rutin, p-coumaric acid, vitexin and isorientin were acquired from Sigma-Aldrich (St. Louis, USA); salicylic acid was obtained from Labsynth (Diadema, Brazil), and syringic acid was obtained from Spectrum (Karachi, Pakistan). Figure S1, presented in Supplementary Information (SI section), presents the chemical structure of the phenolic compounds analyzed. Stock solutions of these standards were prepared in methanol, resulting in a final concentration of approximately 1000 mg L^-1, and stored in a freezer at approximately –20 °C until the time of analysis.

Samples

The residues of beer utilized were provided by a craft beer brewery and were collected after the mashing process. In the beer formulation, the following ingredients were used: 37% Pilsen malt; 32% wheat malt; 8% rolled oats; 8% wheat flakes from Agrária and 10% Munich malt, 5% Carafoam/Carapils malt from Weyermann. The wine residue was collected after the mashing process during the artisanal production of two distinct wines made from two grape varieties, Merlot and Nebbiolo, obtained from a winery in Urupema, Santa Catarina, Brazil. The ground coffee used was cultivated in Santa Luzia, Minas Gerais, Brazil, and it was purchased in the local market. The coffee was brewed and the coffee grounds were collected for further analysis. The industrial residue of apple juice was generated from Fuji and Gala apple varieties cultivated in São Joaquim, Santa Catarina, Brazil.

Sample preparation

Sample preparation followed the methodology outlined by Schulz et al.²⁶ The samples were dried in an oven (S100SDS, Biopar, Nova Marilândia, Brazil) at 41.4 ± 1.8 °C for 24 h and subsequently, ground using an analytical mill (A11 Basic, IKA, Deutschland, Germany). The particle size of the samples was standardized using an electromagnetic sieve (I-1016-B, Pavitest, São José da Lapa, Brazil) equipped with an 850 µm (20 TY) sieve.

The lipid fraction of the dry residue sample from beverage production (1 g) was extracted using 10 mL of hexane in an ultrasonic bath at 25 °C for 15 min, followed by centrifugation at 2,000 × g for 15 min. The defatted sample then underwent acid hydrolysis with 5 mL of methanol and 5 mL of hydrochloric acid (6 mol L^-1) in an oven at 85 °C for 30 min. Subsequently, the solution (pH 2) was subjected to partitioning (three times) with 10 mL of ethyl ether, with each partition resulting in extracts that were centrifuged using a tube centrifuge (Q222M, Quimis, Diadema, Brazil) at 2,195 × g for 5 min. The supernatants were pooled and the organic solvent was removed via rotary evaporation (RV 10, IKA, Deutschland, Germany). The dry extract obtained was dissolved in 1 mL of methanol and further diluted up to 30-fold using acetonitrile before being injected into the HILIC-MS/MS system for phenolic determination.

Instrumentation and HILIC-MS/MS method

The method was developed using an Agilent® Technologies 1200 Series HPLC (Waldbronn, Germany), equipped with an online degasser (G1322A), a quaternary pump (G1311A), a column oven (G1316A) and an autosampler (G1367B). The chromatographic system was coupled to a Qtrap 3200 triple quadrupole mass spectrometer with an electrospray source ionization (ESI), Turbo V ion source/TurboIonSpray, (Applied Biosystems/ MDS Sciex, Concord, Canada). The Analyst software²⁷ was employed for data acquisition and analysis. Separations were conducted using a FRULIC-N column (150 mm, 4.6 mm internal diameter, 5 µm particle size) obtained from AZYP (Arlington, USA). Mobile phase A consisted of acetonitrile:ammonium acetate (110 mmol L^-1) and an aqueous ammonium acetate solution (110 mmol L^-1) was used as mobile phase B. The column was conditioned with 95% A for 60 min at the first time before analysis and 10 min between runs. Separation was performed at 60 °C using the following gradient elution (A mobile phase): 95% from 0-2 min, 95-40% from 2-6 min, 40% from 6-11 min, and 40-95% from 11-13 mim. The flow rate was set at 500 µL min^-1, and the injection volume was 25 µL. The mass spectrometer parameters employed were 10 psi curtain gas, source temperature at 600 °C, 50 psi nebulizer gas, 50 psi drying gas, and the capillary needle maintained at –4,500 V. All phenolic compounds were monitored and quantified using the multiple reaction monitoring (MRM) mode.

Optimization of the analyte-dependent MS/MS parameters

The optimization of dependent MS/MS parameters, which included declustering potential (DP); entry potential (EP); collision cell entrance potential (CEP); collision energy (CE), and collision cell exit potential (CXP), was conducted via direct infusion of analytical standards diluted to a concentration of 1.0 mg L^-1. To achieve this, the electrospray capillary needle was maintained at 10 mm in the vertical position and at 5 mm in the horizontal position. The settings for the ESI were as follows: source temperature 600 °C, curtain gas 10 psi, ion gas source 1 (sheath gas) 50 psi, ion gas source 2 (drying gas) 50 psi, and ion spray voltage of –4,500 V. The syringe containing the diluted phenolic standard solution was connected to an MS/MS channel mechanism, with the infusion pump at a flow rate of 10 μL min^-1. The mass spectrometer was configured to operate in negative mode, and the scan range was selected based on the molecular mass of each individual analytical standard.

Optimization of the HILIC-MS/MS method

HILIC column selection

Initially, three HILIC columns composed of different materials were compared to optimize the method: (1) FRULIC-N (150 × 4.6 mm internal diameter, 5 μm; AZYP, USA; (2) LARIHC CF6-P (150 × 4.6 mm internal diameter, 5 µm; AZYP, USA); (3) ZIC-HILIC (150 × 4.6 mm internal diameter, 3.5 µm; Merck, USA). To assess the efficiency of these columns, a mixture of phenolic compounds at a concentration of 3 mg L^-1 was injected and the MP used consisted of 75% acetonitrile and 25% ammonium formate (15 mmol L^-1). The injection volume was maintained at 10 μL, the temperature at 30 °C, and the MP flow rate at 500 μL min^-1 for all columns. The selection of the column was based on the following parameters: the average number of theoretical plates (N) (equation 1) per meter considering the dimensions of each column evaluated, the last analyte retention time (t_R), and the separation profile.

(1) $\text{N}=16{\left(\frac{{\text{t}}_{\text{R}}}{{\text{W}}_{\text{b}}}\right)}^2$

where t_R is the retention time and W_b is the peak width for a given analyte.

Mobile phase composition

Organic solvent ratio and pH

To determine the optimal MP composition, a multivariate optimization strategy was employed based on the Doehlert design (Table S1, SI section). This approach involved varying the proportion of acetonitrile in the range of 50-90% v/v and the pH of the aqueous phase from 3.0 to 7.0. Solutions of acetic acid and ammonium hydroxide were used to adjust the pH of the aqueous phase and the values were measured using a calibrated pH-meter. The choice of ammonium acetate as an additive and the studied pH range aligns with the recommendations provided by the FRULIC-N column manufacturer and its volatility prevents the crystallization of the MP during the desolvation of the drops in the ESI-MS/MS system. The experimental design included the incorporation of central points to calculate the pure error using Statistica® software.²⁸ During these experiments, a mixture of analytes at a concentration of 3 mg L^-1 was injected, while parameters such as column temperature and injection volume were held constant at 30 °C and 10 μL, respectively. All experiments were conducted in duplicate and randomized to detect potential systematic errors. The t_R data were extracted from the resulting chromatograms and the retention factor (k’) was calculated for each analyte using equation 2:

(2) $\text{k'}=\frac{{\text{t}}_{\text{R}}-{\text{t}}_{\text{zero}}}{{\text{t}}_{\text{zero}}}$

where t_R is the retention time and t_zero is the zero retention time.

The zero retention time (t_zero) was experimentally determined using benzyl alcohol as a marker considering its ease of ionization in ESI-MS/MS. Initially, benzyl alcohol was injected into the MS/MS system. Subsequently, benzyl alcohol was injected into the LC-MS/MS system using a MP consisting of 95% aqueous phase (ammonium acetate 15 mmol L^-1) and 5% organic phase (acetonitrile). The flow rate was set at 500 µL min^-1, the injection volume at 10 µL, and the column temperature at 30 °C. The t_R data of benzyl alcohol were obtained from the resulting chromatograms, allowing the determination of the t_zero of the FRULIC-N column.

Furthermore, taking into account the number of analytes under investigation, a compromise condition was established in accordance with equation 3:

(3) $\text{Response\_factor}=\frac{\text{MGk}\text{'}}{{\text{MG}}_{{\text{w}}_{\text{b}}}}$

where MGk’ is the geometric mean of the retention factor; MGw_b is the geometric mean of the base width. To determine the optimal MP composition, the Doehlert design (Table S1) was evaluated using the response in equation 3. The statistical analyses were carried out using Statistica® software.²⁸

Effect of salt concentration and temperature

To optimize the concentration of additives and the column temperature, a 2² factorial design was employed, including triplicate measurements at the central point, as shown in Table S2 (SI section). Three different concentrations of additives in the MP were evaluated, ranging from 10 to 110 mmol L^-1 at pH 7. The column temperature varied between 20 and 60 °C, in accordance with the specified limits of the manufacturer. For this experiment, a mixture of phenolic compounds at a concentration of 3 mg L^-1 was injected while maintaining an isocratic MP composition of 90:10% v/v acetonitrile:ammonium acetate. Equation 3 was applied to analyze the data obtained using Statistica® software.²⁸

Injection volume

The optimization of the injection volume was conducted using univariate analysis. A mixture containing 3 mg L^-1 of phenolic compounds was injected into the FRULIC-N column, which was maintained at 60 °C with an isocratic MP composed of acetonitrile:ammonium acetate 110 mmol L^-1 with 90:10% v/v and flow rate 500 μL min^-1. The injection volumes were evaluated in the range of 5 to 45 µL. To determine the optimal injection volume, the geometric means of peak parameters, including area, height, and signal width, were utilized.

Elution gradient selection

Given the structural diversity of phenolic compounds and the objective of enhancing chromatographic separation, two different MP elution gradients were assessed, as detailed in Table S3 (SI section). First, all gradients were evaluated for the separation profile and t_R of the compounds. Then, aqueous ammonium acetate (110 mmol L^-1) was added to phase A of the selected gradient in order to improve the separation profile of the compounds.

Validation parameters and statistical analysis

The method underwent validation in accordance with the Eurachem²⁴ and AOAC,²⁵ utilizing tests with standard solutions and samples enriched with phenolic standards. The parameters assessed were linearity, matrix effects, precision, accuracy, limit of detection (LOD) and limit of quantification (LOQ).

Linearity was assessed through calibration curves constructed in triplicate, in at least five concentration levels within the working range of the analyte. The parameters of the calibration curves were determined using linear regression, and the statistical significance of the linear model was verified using the F test.²⁹ The normality of the residuals was confirmed through the Shapiro-Wilk test.³⁰ All statistical tests were conducted at a 95% confidence level employing Statistica²⁸ and Excel³¹ softwares.

In this study, the standard addition calibration method was applied to overcome matrix effects, considering that beverage production waste samples have a variable composition that cannot be adequately represented by a limited set of standards and the absence of blanks sample. Standard addition curves were constructed in the methanolic extract generated from the sample preparation. Analytical standards in the same working range as the external calibration curve were added to the extract of each sample. The matrix effects were evaluated by comparing the slopes obtained for the standard solution and matrix calibration curves. The ratio between mean slopes for each calibration curve was calculated. Slope ratios below 0.9 or above 1.1 indicate ion suppression and ion enhancement, respectively, while values within this range suggest negligible matrix effects.³²

Instrumental precision was assessed by performing ten consecutive injections (n = 10) of phenolic analytical standards at an intermediate concentration. Intra-assay precision was evaluated by the injection of three concentration levels of standard solutions, with two consecutive injections (n = 2). Inter-assay precision was determined over two days, involving two consecutive injections on each day (n = 4). Relative standard deviation (RSD) values were calculated for both peak area and t_R.

Accuracy was verified through recovery tests on samples spiked with phenolic standards at three concentration levels, with duplicate injections. LOD and LOQ were determined using signal-to-noise ratios (S/N) of 10:1 for LOQ and 3:1 for LOD.³³

Results and Discussion

Optimization of the analyte-dependent MS/MS parameter

Initially, the phenolic analytical standards were individually injected into the MS, yielding three fragmentation patterns that enabled the equipment to operate in MRM mode. In this mode, the m/z of a precursor ion and the m/z of its fragments are selected and the monitored responses are only the mass variations relevant to the process. This information ensures the selectivity of the method by confirming the identity of the analytical standard using the three fragmentation patterns. Subsequently, the quantification was carried out using the transition with the highest signal intensity. Given that the compounds examined in this study were phenolics, it was understood that the fragments generated from the precursor ions would involve variations associated with the loss of hydrogen [M – H]^–, which is why the ESI source was maintained in the negative mode. The mass spectrometer parameters for each analyte are provided in Table 1.

Table 1
Mass spectrometry parameters for the development of the method for the 31 phenolic compounds using HILIC-MS/MS^a

Analyte	Precursor ion [M – H]– / m/z	Product ion / m/z	DP / V	EP / V	CEP / V	CE / V	CXP / V
4-Methyllumbelliferone	174.9	133.1 / 119.0 / 147.0	–60.0	–4.5	–16.0	–30.0	–2.0
Salicylic acid	136.9	93.0 / 65.0 / 75.1	–30.0	–4.5	–12.0	–22.0	–2.0
Vanillin	150.9	136.0 / 92.0 / 108.0	–25.0	–4.5	–12.0	–16.0	–2.0
Coniferaldehyde	177.0	162.0 / 161.3 / 134.1	–30.0	–4.5	–10.0	–16.0	0.0
Protocatechin acid	152.9	109.0 / 108.4 / 91.0	–30.0	–10.5	–14.0	–20.0	–2.0
Umbelliferone	160.9	133.1 / 177.1 / 105.0	–45.0	–4.5	–12.0	–24.0	–2.0
Gallic acid	168.9	125.0 / 79.0 / 51.0	–35.0	–4.5	–14.0	–20.0	–2.0
Vanillic acid	166.9	108.0 / 123.0 / 152.0	–30.0	–4.5	–14.0	–24.0	–2.0
Syringaldehyde	180.9	151.0 / 166.0 / 122.9	–25.0	–4.5	–18.0	–22.0	0.0
Caffeic acid	178.9	135.0 / 134.4 / 107.0	–30.0	–4.5	–14.0	–22.0	–2.0
Ferulic acid	192.8	134.0 / 177.8 / 149.0	–30.0	–4.5	–16.0	–18.0	–2.0
Scopoletine	190.9	176.0 / 103.9 / 148.0	–25.0	–9.5	–14.0	–18.0	0.0
Synapaldehyde	207.0	177.0 / 192.1 / 149.1	–25.0	–4.5	–10.0	–20.0	0.0
Syringic acid	196.9	121.1 / 153.0 / 182.0	–35.0	–4.5	–16.0	–28.0	–2.0
Resveratrol	226.9	142.9 / 184.8 / 115.1	–45.0	–4.5	–22.0	–34.0	–2.0
Synapic acid	223.0	148.8 / 164.0 / 121.1	–30.0	–4.5	–18.0	–28.0	0.0
Chrysin	252.9	62.9 / 65.0 / 143.1	–65.0	–4.0	–24.0	–52.0	–8.0
Pinocembrin	255.0	151.0 / 65.0 / 62.9	–55.0	–4.0	–14.0	–30.0	0.0
Naringenin	270.9	151.0 / 119.1 / 65.0	–55.0	–4.5	–18.0	–26.0	–2.0
Kaempferol	284.9	93.0 / 117.0 / 65.0	–75.0	–4.5	–18.0	–54.0	–2.0
Catechin	289.0	109.0 / 123.0 / 245.1	–45.0	–4.5	–14.0	–38.0	–2.0
Hispidulin	298.9	284.0 / 65.0 / 137.0	–45.0	–4.5	–18.0	–34.0	–4.0
Quercetin	300.9	151.0 / 121.1 / 65.0	–60.0	–4.5	–16.0	–34.0	0.0
Taxifolin	303.0	125.1 / 285.0 / 57.0	–35.0	–4.5	–16.0	–26.0	–2.0
Chlorogenic acid	353.1	191.0 / 85.0 / 93.0	–25.0	–4.5	–18.0	–24.0	–2.0
Rosmarinic acid	359.8	161.0 / 132.9 / 134.8	–35.0	–4.5	–26.0	–24.0	0.0
Isoquercetin	463.1	300.0 / 271.1 / 301.1	–70.0	–4.5	–34.0	–28.0	–4.0
Rutin	609.2	300.1 / 271.1 / 301.1	–80.0	–4.5	–44.0	–64.0	–2.0
p-Coumaric acid	162.9	119.1 / 93.0 / 117.0	–35.0	–4.5	–14.0	–22.0	–2.0
Vitexin	430.8	310.9 / 282.9 / 116.9	–55.0	–6.5	–20.0	–22.0	–2.0
Isorientin	446.8	326.9 / 356.9 / 133.0	–40.0	–5.5	–32.0	–20.0	–2.0

• ^aExperimental conditions: the capillary needle was maintained at –4500 V in the negative mode; curtain gas 10 psi; source temperature 600 °C; nebulizer gas 50 psi; drying gas 50 psi; scan duration 1 s; dwell time 50 ms. Optimization was carried out by directly infusing an aqueous solution of each analyte (1.0 mg L^-1); HILIC-MS/MS: hydrophilic interaction chromatography coupled with triple quadrupole mass spectrometry; DP: declustering potential; EP: entrance potential; CEP: collision cell entrance potential; CE: collision energy; CXP: collision cell exit potential.

HILIC column selection

First, acetonitrile was chosen as an aprotic organic component in the MP, considering the polar SP evaluated and the structure of the phenolic compounds. Acetonitrile does not compete for polar active sites on the surface of the SP and consequently natural formation of water layers occurs on the surface and analytes with strong hydrogen bonding potentials can interact easily.³⁴ Initially, ammonium formate was used as an ionic additive for the MP considering the variety of applications in the literature,²²,³⁵,³⁶,³⁷ good solubility in organic-rich MP, and its compatible volatility with ESI-MS.³⁸ Some characteristics and performance parameters of the columns (1) FRULIC-N; (2) LARIHC CF6-P and (3) ZIC-HILIC were evaluated and are presented in Table 2.

Table 2
Chromatographic parameters for the selection of HILIC columns regarding the separation of phenolic compounds in HILIC-MS/MS^a

Column	FRULIC-N (1)	LARIHC CF6-P (2)	ZIC-HILIC (3)
Bonded phase structure	native cyclofructan-6 (CF-6)		sulfobetaine
Number of platesb,c / (N m-1)	3774.9	3708.3	4497.4
Last analyte retention time / min	7.27	4.60	9.06

• ^aExperimental conditions: phenolic standards 3 mg L 1; injection volume of 10 μL; column temperature of 30 °C; mobile phase acetonitrile:ammonium formate (15 mmol L¹) 75:25% v/v (isocratic elution); flow rate 500 μL min¹; ^bnumber of plates N=16(trWb)2 per column length (m), where t_R: retention time and W_b: peak width for a given analyte; carithmetic average; HILIC-MS/MS: hydrophilic interaction chromatography coupled with triple quadrupole mass spectrometry.

The columns (1) and (2) are based on SP made of CF6, which belongs to a family of oligosaccharides comprising D-fructofuranose units linked to a crown ether, with each fructofuranose unit containing four stereogenic centers and three hydroxyl groups. Column (1) is made with native CF6 and features neutral high porosity polyhydroxy groups and a solid inner core, resulting in a shorter diffusion path for the analytes and the MP. This design helps reduce band broadening caused by poor mass transfer in systems with slow kinetics. The column (2) employs a SP based on CF6 derivatized with alkyl groups, which decrease hydrogen bonding possibilities and increase non-polar interactions between analyte and SP.³⁹ On the other hand, column (3) has a SP based on zwitterionic compounds, characterized by having both positive and negative charges on its surface.⁴⁰

In HILIC, the eluting strength of the MP increases as the water concentration on the MP increases. In all the columns evaluated, pinocembrin was the first analyte to elute, which was attributed to its greater hydrophobic character (log P 3.14) and gallic acid was the last (log P 0.70), but this trend was not observed for all the compounds (Table S4, SI section). The wide pK_a range of the analytes, a diversity of molar mass, and other structural characteristics demonstrated influence on the compounds retention mechanisms. Considering this, the elution order is an indication that there were a variety of interactions involved in the separation process. The similarities in the elution order of columns (1) and (2) are justified by the SP composed of CF6. However, while column (1) is made by native CF6 with high hydrophilicity, column (2) presents derivatizations with alkyl groups which reduce polar interactions, as hydrogen bonding, and consequently the t_R of the studied compounds, as shown in Figure 1. Lastly, column (3) presented a slightly different elution order due to the zwitterionic SP which combines ionic interactions with polar interactions, but with a low contribution of hydrogen bonding interactions, to retain the analyte.⁴¹

Figure 1
Chromatograms of phenolic standards (3 mg L^-1) obtained by three different columns to optimize the method: (1) FRULIC-N; (2) LARIHC CF6-P; (3) ZIC-HILIC. Experimental conditions: mobile phase composed of acetonitrile and ammonium formate (15 mmol L^-1) with 75:25% v/v (isocratic elution); flow rate 500 μL min^-1; injection volume 10 μL; column temperature 30 °C.

Based on the profiles obtained (Figure 1), column (1) demonstrated superior separation, with a greater distribution of t_R of the analyte throughout the chromatographic run, indicating its potential for later optimizations using gradient elution mode. Conversely, columns (2) and (3) exhibited coelution of chromatographic signals in t_R close to the experimental t_zero (2.80 min), suggesting restricted optimization possibilities. Furthermore, column (3) exhibited higher t_R of the last analyte, prolonging the analysis time (Table 2). The parameters presented in Table 2 support the choice of column (1), which exhibited the second-highest number of plates, indicating the separation efficiency of the method. Additionally, it displayed the second lowest t_R for the last analyte, enabling a higher analytical frequency.

Mobile phase composition

Organic solvent ratio and pH

The pH optimization of the MP was carried out considering the variety of phenolic compounds studied, which have a pK_a range from 2.8 to 9.5 (Table S4), and also keeping safe conditions for the FRULIC-N column, which can be damaged in a strongly alkaline environment.¹⁶ Figure 2 presents the response surface generated using the Doehlert design for determining the MP composition. The response surface is defined by equation 3, where the response is based on the relationship between the geometric mean of the retention factor (representing solute and SP interaction) and the signal widths.

Figure 2
Response surface obtained using the Doehlert design for the optimization of the organic solvent ratio and pH in the separation of 31 phenolic compounds by HILIC-MS/MS. Experimental conditions: acetonitrile of 50 to 90% v/v - phase A, and the pH of the aqueous phase of 3.0 to 7.0 (solutions of acetic acid and ammonium hydroxide) - phase B; phenolic standards 3 mg L^-1; flow rate 500 μL min^-1; FRULIC-N column; injection volume 10 μL; column temperature set at 30 °C.

According to the response surface in Figure 2, the optimal conditions were achieved when the organic phase reached 90% v/v, regardless of the pH levels considered. As indicated in Table S4, the phenolic acids have pK_a values ranging from 2.8 to 4.2, and for other phenolic compounds, values can vary from 6.1 to 9.5. The pH was not statistically significant, but the interaction between pH and % acetonitrile was statistically significant in the studied range (Table S5, SI section). Consequently, the chosen MP composition was acetonitrile:ammonium acetate 90:10% (v/v), maintaining the pH at 7. The statistical model obtained through the Doehlert design exhibited a coefficient of determination (R²) of 0.9944 and a pure error of 8.88E-5. The pure error value obtained indicated a low variability between replicates, that together with R² > 0.99, confirmed the alignment of the experimental data with the predicted model at a 95% confidence level and thus, lack of fit was not observed. The analyses of variance (ANOVA) results of this optimization are provided in Table S5. The statistical model is represented by the following equation:

(1) $\text{Z}=2.401-6.395\text{E}-2\left(\text{x}\right)+4.700\text{E}-4{\left(\text{x}\right)}^2-6.412\text{E}-2\left(\text{y}\right)+6.250\text{E}-5{\left(\text{y}\right)}^2+8.625\text{E}-4\left(\text{x}\right)\left(\text{y}\right)$

Effect of salt concentration and temperature

In HILIC, when analyzing ionizable compounds, the use of additives, such as ammonium acetate, is essential. These additives are used typically at concentrations ranging from 5 to 100 mmol L^-1 while maintaining a high proportion of organic solvent. The incorporation of additives can be advantageous in suppressing undesirable electrostatic interactions, including repulsive and attractive forces, between the SP and analytes.¹⁶ Especially on the FRULIC-N column, a complex positively charged is observed when alkali metals or ammonium ions are used on the MP. It occurs due to the presence of a crown ether core on the CF6 structure. Thus, this effect can be modulated with the addition of ammonium acetate to the MP composition generating an electrostatic interaction for anionic analytes.⁴² Furthermore, the salt concentration can affect the thickness of the stagnant water layer on the surface of the SP, further altering the retention by hydrophilic partition, and also increasing the eluting strength of the MP.¹⁶ These types of SP, based on crown ethers and macrocycles, have been used to separate anions chromatographically since the 90’s.⁴² Considering this, by optimizing the appropriate concentration of the ammonium additive, more symmetrical signals at appropriate t_R can be achieved.

Column temperature is a crucial parameter in HILIC separations, significantly impacting analyte diffusion, the viscosity of the MP, and the enthalpy transfer of analytes between the mobile and SP. The column temperature can influence analyte retention and selectivity based on thermodynamic considerations.³⁴ Consequently, these parameters were simultaneously evaluated by varying the column temperature (20 to 60 °C) and the concentration of ammonium acetate (10 to 110 mmol L^-1) in the MP, while maintaining pH 7. Using equation 3 as the response, it was possible to generate the surface depicted in Figure 3.

Figure 3
Response surface obtained using a 2² factorial design for the optimization of additive concentration and column temperature in the separation of 31 phenolic compounds by HILIC-MS/MS. Experimental conditions: mobile phase with isocratic composition of acetonitrile 90% v/v - phase A, and ammonium acetate solution 10% v/v with a concentration of 10 to 110 mmol L^-1 in the pH 7.0 - phase B; column temperature ranging from 20 to 60 °C; phenolic standards 3 mg L^-1; flow rate 500 μL min^-1; FRULIC-N column; injection volume 10 μL.

According to the response surface in Figure 3, the additive concentration did not exhibit statistical significance within the studied range; however, its interaction with temperature was statistically significant. Furthermore, higher temperatures were observed to increase the response significantly. In HILIC, increasing the temperature can increase the diffusion coefficient, resulting in narrower peaks in a shorter t_R.³⁴ The lack of model fit was statistically significant due to the low variability of the replicates, which resulted in a small value of pure error (1.21E-3). Furthermore, the model presented a good R² of 0.8139. The ANOVA results are provided in Table S6 (SI section). Hence, the model could still be used to continue the method development and optimization. It was observed that additive concentration and temperature have complementary effects and the simultaneous optimization was a convenient approach to control analyte retention, improving selectivity and peak shape. As a result, the column temperature was set at 60 °C, and the concentration of ammonium acetate in the MP was established at 110 mmol L^-1. Despite the high concentration of ammonium acetate for ESI-MS/MS, the final condition was considered optimal to suppress undesirable interactions between the SP and the analytes, reducing broadening without compromising the peak area.

Injection volume

The optimization of the injection volume was conducted univariately, and the results, including the geometric mean of peak parameters such as area, height, and signal width, are depicted in Figure S2 (SI section). The results implied the selection of a 25 μL injection volume, considering the marginal increase in height and area when using volumes below this threshold. Additionally, with a 25 μL volume, the base width of the analytical signal remained close to the lowest value obtained in the study.

Elution gradient selection

According to the Doehlert design, the optimal conditions were reached when the aqueous phase reached only 10% v/v and the organic phase 90% v/v. However, this MP is considered weak in HILIC to elute strongly polar compounds. In HILIC, retention increases with increasing analyte polarity, and the elution of these analytes is boosted by increasing the MP water content, as aforementioned. Considering the structural diversity of phenolic compounds and the amount of analytes proposed in this work, the study of gradient mode was necessary.

Considering that the range of the HILIC mechanism varies between columns, depending on the hydrophilicity of the S P, with or without ionic sites, and the nature of the analyte, the gradients evaluated in Table S3 initially involved a weak MP, at the lowest concentration of aqueous phase (5% v/v). Thereby, it was expected a weak retention and quick elution of the less polar analytes (those with the highest log P). The subsequent steps evaluated the increase in the strength of the MP through the increase of the aqueous phase to 40% v/v (gradient I) or 60% v/v (gradient II), allowing the elution of the most polar phenolic compounds. This evaluation was carried out because it was observed that the increase in the proportion of the aqueous phase resulted in a decrease in the t_R of the most retained compounds, mainly phenolic acids, and an improvement in the separation profile (Table S7, SI section). In both gradient conditions studied, the analytes with log P < 0 (more polar) presented the highest t_R, but there was no trend for this parameter. This observation suggests that there were different retention mechanisms occurring simultaneously during the separation process and the main SP-analyte-MP interactions likely varied for different analytes. In general, it was observed that ionized compounds eluted last, suggesting that ionic interactions or coulombic interactions occurred between these analytes and the SP. To select the appropriate gradient, the separation profile with t_R distribution and best peak shapes within the minimum execution time was considered as a compromise condition. Then, the gradient II was selected as the optimal condition. Finally, the addition of ammonium acetate in phase A (acetonitrile) in gradient II reduced the t_R of the ionized compounds (Table S7) but did not affect the retention of neutral compounds. In summary, the separation profile was improved by the salt addition (Figure 4). The addition of ammonium acetate in the organic phase also increased the MP strength and probably generated electrostatic interactions with anionic analytes. In addition, the CF6 sites of SP can be saturated with ammonium ions, leading to a reduction in the retention of these analytes.⁴² The improvement of the separation profile can also be attributed to the increased stability of the MP viscosity and pH, which, in turn, affects the diffusion of the analyte and t_R.¹⁹,³⁴

Figure 4
Chromatogram of phenolic standards (3 mg L^-1) obtained using the optimized HILIC-MS/MS method. Experimental conditions: acetonitrile:ammonium acetate (110 mmol L^-1) - phase A, and ammonium acetate solution (110 mmol L^-1) - phase B; gradient elution 95% A 0-2 min, 95-40% A 2-6 min, 40% A 6-11 min, and 40-95% A 11-13 min; flow rate 500 μL min^-1; FRULIC-N column; injection volume 25 μL; column temperature set at 60 °C. (1) 4-methyllumbelliferone (3.54 min); (2) salicylic acid (8.00 min); (3) vanillin (3.38 min); (4) coniferaldehyde (3.29 min); (5) protocatechin acid (9.60 min); (6) umbelliferone (3.55 min); (7) gallic acid (10.01 min); (8) vanillic acid (7.47 min); (9) syringaldehyde (3.38 min); (10) caffeic acid (9.35 min); (11) ferulic acid (6.78 min); (12) scopoletine (3.55 min); (13) synapaldehyde (3.27 min); (14) syringic acid (8.51 min); (15) resveratrol (3.69 min); (16) synapic acid (7.05 min); (17) chrysin (3.26 min); (18) pinocembrin (3.19 min); (19) naringenin (3.40 min); (20) kaempferol (3.59 min); (21) catechin (6.01 min); (22) hispidulin (3.61 min); (23) quercetin (4.85 min); (24) taxifolin (5.01 min); (25) chlorogenic acid (9.79 min); (26) rosmarinic acid (9.49 min); (27) isoquercetin (9.20 min); (28) rutin (9.35 min); (29) p-coumaric acid (7.33 min); (30) vitexin (9.18 min); (31) isorientin (9.35 min).

Finally, the use of an elution gradient implies the need to condition the column with the initial MP at the end of each chromatographic run to establish the equilibrium of the initial analysis conditions. According to McCalley,⁴³ equilibrium is achieved in gradient elution by conditioning 12 column volumes for a wide range of SP in HILIC. In this study, the conditioning time of 10 min, corresponding to 2 volumes of the FRULIC-N column was considered adequate to obtain good precision results.

Validation parameters and statistical analysis

The linearity hypotheses were confirmed for both the external calibration curves (Table 3) and the standard addition curves in the matrix (Table S8, SI section). According to the results, each analyte exhibited different calibration ranges. The determination coefficients (R²) were satisfactory, all exceeding 0.90. The profiles obtained for residual analysis were considered random, showing no trend in the data (data not shown). The residuals were normally distributed based on the Shapiro-Wilk test with a significance level of 5% (W_cal > W_tab, indicating that the data did not deviate significantly from a normal distribution, Table 3). The F test indicated that, with a 95% confidence level, the lack of fit was not statistically significant in the linear models (Table 3). The results obtained from the performed tests suggested that the use of the ordinary least squares (OLS) method was appropriate for both external calibration curves and standard addition curves in the matrix. In general, the LOD ranged from 0.001 to 0.503 mg L^-1, while the LOQ ranged from 0.004 to 1.524 mg L^-1. The exception was gallic acid which presented a LOD of 1.701 mg L^-1 and a LOQ of 5.155 mg L^-1 (Table 3). Despite that, the limits were appropriate for gallic acid determinations in the samples.

Table 3
Analytical parameters of the external calibration curves obtained for the 31 phenolic compounds using the developed method

Analyte	Calibration range / (mg L-1)	Slope / (L mg-1)	Standard error-slope	Intercept	Standard error-intercept	Coefficient of determination / R2	LOD / (mg L-1)	LOQ / (mg L-1)	F cal	Wtab	Wcal
4-Methyllumbelliferone	0.04-1.00	2.0208E5	5.473E3	7.995E3	2.470E3	0.9920			1362.9	0.859	0.871
	1.00-5.00	9.1581E4	1.1343E4	1.2935E5	3.5100E4	0.9561	0.009	0.026	65.18	0.762	0.895
Salicylic acid	0.05-0.63	8.2335E6	1.5904E5	–5.226E3	8.0486E4	0.9966			2680.1	0.829	0.940
	0.63-6.25	5.8857E6	3.4186E5	3.4887E6	1.2247E6	0.9801	0.018	0.054	296.4	0.803	0.971
Vanillin	0.10-1.04	1.4800E6	4.7472E4	3.0155E4	2.7221E4	0.9928			972.0	0.818	0.939
	1.04-7.8	5.7354E5	6.8717E4	1.3322E6	2.9771E5	0.9321	0.034	0.103	69.7	0.788	0.911
Coniferaldehyde	0.01-0.25	2.3184E6	5.6773E4	9.843E3	6.114E3	0.9952			1667.6	0.829	0.849
	0.25-3.79	1.1988E6	4.6606E4	5.0355E5	7.7318E4	0.9880	0.002	0.007	661.6	0.829	0.862
Protocatechin acid	0.38-1.02	5.4486E4	1.0038E4	–4.924E3	7.010E3	0.9076			29.4	0.762	0.903
	1.02-12.75	2.4839E5	1.0829E4	–3.616E5	7.9085E4	0.9887	0.108	0.327	526.1	0.803	0.910
Umbelliferone	0.02-0.33	7.3842E5	1.1639E4	6.965E3	1.613E3	0.9980			4024.8	0.829	0.855
	0.33-1.63	5.2827E5	4.6388E4	1.3039E5	4.5408E4	0.9554	0.005	0.016	129.7	0.803	0.972
Gallic acid	7.50-25.00	8.9703E4	8.317E3	–7.290E5	1.3948E5	0.9749	1.701	5.155	116.3	0.762	0.889
Vanillic acid	0.06-1.27	3.1385E5	2.0252E4	1.1604E4	1.2438E4	0.9598			240.2	0.850	0.869
	1.27-7.63	2.7566E5	2.1629E4	1.1570E5	9.9999E4	0.9819	0.020	0.061	162.4	0.762	0.987
Syringaldehyde	0.06-1.00	4.3160E5	8.558E3	1.1231E4	4.228E3	0.9964			2543.2	0.842	0.911
	1.00-12.50	1.5268E5	1.3456E4	4.5510E5	9.0314E4	0.9480	0.020	0.061	128.7	0.818	0.952
Caffeic acid	0.03-0.13	1.4745E6	1.9941E5	1.9407E4	1.5015E4	0.9563			54.6	0.803	0.940
	0.13-1.25	1.9891E6	4.3902E4	–4.518E4	3.0983E4	0.9966	0.008	0.023	2052.8	0.818	0.966
Ferulic acid	0.03-0.13	4.9217E5	1.9828E4	–817.1	1.507E3	0.9903	0.009	0.027	616.1	0.803	0.958
	0.13-1.30	5.5230E5	2.1456E4	5.943E3	1.5293E4	0.9895	0.009	0.027	662.5	0.818	0.973
Scopoletine	0.01-0.29	1.0846E6	2.1770E4	2.854E3	2.716E3	0.9967			2482.2	0.829	0.865
	0.29-2.92	6.6444E5	3.4537E4	2.2018E5	4.5670E4	0.9813	0.001	0.004	370.1	0.818	0.901
Synapaldehyde	0.03-0.13	3.0499E5	2.9321E4	1.580E3	2.207E3	0.9470			108.1	0.803	0.955
	0.13-5.00	1.7717E5	6.378E3	6.0515E4	1.4018E4	0.9884	0.009	0.029	771.5	0.842	0.847
Syringic acid	0.38-1.25	3.3994E5	2.1440E4	7.983E3	1.7300E4	0.9804			251.3	0.788	0.962
	1.25-10.00	2.8363E5	1.6113E4	1.8940E5	9.3756E4	0.9840	0.098	0.296	309.8	0.788	0.971
Resveratrol	0.04-0.14	2.6496E5	9.352E3	3.928E3	8.60E2	0.9938			802.5	0.788	0.965
	0.14-4.27	1.7841E5	7.324E3	6.2574E4	1.3015E4	0.9850	0.012	0.036	593.3	0.842	0.866
Synapic acid	0.08-1.25	2.2934E5	1.5521E4	1.1395E4	9.816E3	0.9602			218.3	0.842	0.909
	1.25-7.50	2.5487E5	2.5282E4	5.6437E4	1.1481E5	0.9618	0.024	0.071	101.6	0.762	0.984
Chrysin	0.02-0.38	4.1969E5	1.5198E4	7.412E3	2.496E3	0.9896			762.5	0.829	0.843
	0.38-3.75	2.4621E5	7.454E3	1.0755E5	1.2971E4	0.9936	0.005	0.015	1090.7	0.818	0.850
Pinocembrin	0.02-0.73	3.5315E5	6.609E3	8.853E3	2.128E3	0.9958			2854.9	0.866	0.872
	0.73-4.35	1.1956E5	1.8029E4	2.2790E5	4.7489E4	0.9148	0.008	0.025	43.9	0.762	0.983
Naringenin	0.04-0.27	1.6910E5	5.382E3	1.409E3	6.88E2	0.9939			987.1	0.803	0.831
	0.27-1.33	1.7031E5	5.517E3	5.052E3	4.411E3	0.9937	0.011	0.035	952.9	0.803	0.974
Kaempferol	0.08-1.24	8.5802E4	4.523E3	–5.284E3	2.832E3	0.9755			359.7	0.842	0.874
	1.24-7.43	7.0884E4	8.318E3	5.1934E4	3.7398E4	0.9471	0.026	0.080	72.6	0.762	0.981
Catechin	0.29-1.48	3.7936E4	1.796E3	4.86E2	1.600E3	0.9866			445.7	0.803	0.954
	1.48-14.78	1.0233E5	4.870E3	–1.053E5	4.1281E4	0.9865	0.066	0.200	441.5	0.803	0.918
Hispidulin	0.01-1.32	7.9548E5	3.0578E4	–7.35E2	2.285E3	0.9897			676.7	0.818	0.966
	1.32-5.30	7.2355E5	4.1009E4	1.5601E5	1.3438E5	0. 9905	0.005	0.015	311.3	0.762	0.923
Quercetin	0.13-0.64	1.4165E5	2.0714E4	–6.160E3	8.810E3	0.9196			46.8	0.762	0.884
	0.64-12.83	6.9657E5	3.7638E4	2.1902E5	2.8574E5	0.9855	0.043	0.129	342.5	0.788	0.961
Taxifolin	0.06-1.25	1.4482E5	1.1062E4	–5.307E3	6.674E3	0.9445			171.4	0.860	0.869
	1.25-12.5	1.9983E5	1.3344E4	1.0667E5	9.5617E4	0.9738	0.022	0.067	224.2	0.803	0.985
Chlorogenic acid	0.25-2.57	2.8679E4	1.947E3	–9.217E3	2.262E3	0.9685			216.8	0.818	0.823
	2.57-12.85	4.0841E4	1.832E3	–5.6673E4	1.4547E4	0.9900	0.050	0.150	496.7	0.788	0.952
Rosmarinic acid	0.03-1.69	5.4136E4	4.614E3	–5.62E2	4.391E3	0. 9582			137.6	0.818	0.919
	1.69-16.89	1.1060E5	4.817E3	–1.5444E5	4.9373E4	0.9905	0.004	0.013	527.1	0.788	0.935
Isoquercetin	1.09-13.62	2.840E3	2.63E2	2.919E3	1.927E3	0.9427	0.360	1.091	116.2	0.818	0.948
Rutin	2.60-13.00	2.954E3	2.40E2	–28.1	1.927E3	0.9678	0.503	1.524	151.5	0.788	0.899
p-Coumaric acid	0.05-1.04	1.1003E6	4.1930E4	–1.6068E4	2.0554E4	0.9856			688.6	0.850	0.856
	1.04-13.00	1.0992E6	3.0545E4	2.7048E5	2.1321E5	0.9946	0.016	0.048	1295.2	0.818	0.927
Vitexin	0.08-0.99	2.9560E4	2.762E3	8.33E2	1.461E3	0.9342			114.5	0.829	0.899
	0.99-9.93	1.9743E4	9.80E2	9.607E3	5.903E3	0.9878	0.022	0.067	405.8	0.788	0.969
Isorientin	0.60-5.98	1.9075E4	1.419E3	1.1798E4	4.862E3	0.9676	0.132	0.400	180.5	0.803	0.881

• LOD: limit of detection; LOQ: limit of quantification; F_cal: calculated value in the F-test; W_tab: critical value (significance level of 5%) in the Shapiro-Wilk test; W_cal: value calculated in the Shapiro-Wilk test.

The results obtained in the evaluation of the matrix effect indicated significant differences in the slopes of the curves in different matrices, except for the analytes scopoletine, kaempferol, hispidulin, and taxifolin (Table S8). Consequently, the significance of the matrix effect in the determination of 27 phenolic compounds using the proposed method became apparent. Considering this, the determination of the analytes in the examined residues from beverage production was conducted using external calibration when no matrix effect was observed (slope ratios between 0.9 to 1.1, Table S8), and standard addition calibrations were employed when the effect was present (slope ratios below 0.9 or above 1.1, Table S8).

Instrumental precision was conducted to assess the capability of the analytical instrument to generate accurate and reproducible data. The RSD (relative standard deviation) results for instrumental precision, evaluating parameters such as migration time and peak area, demonstrated RSD values ranging from 0.1 to 10.7% (Table S9, SI section). Intra-assay and inter-assay precision values, considering t_R and signal area, fell within the range of 0.0 to 12.2% RSD and 0.0 to 11.7% RSD, respectively. However, when considering the mean, the values for intra-assay and inter-assay precision were 2.6 and 3.7% RSD, respectively. The system pressure was monitored by 62 consecutive injections of standard-fortified samples, obtaining an RSD of 8.7%. The AOAC guidelines consider RSD values up to 11% suitable for concentration levels in the order of mg L^-1.²⁵ These results demonstrated that the instrument was suitable for use in the validation procedures and that sample preparation likely contributes to minimizing irreversible adsorption of compounds onto the column.

The average results obtained by adding and recovering phenolic compounds at three different concentration levels in residue samples from beverage production are presented in Table 4.

Table 4
Recovery of 31 phenolic compounds for the developed method, evaluated for five different residue matrices from beverage production at three concentration levels

Recovery / %
Sample	Coffee			Beer			Apple juice			Wine (Merlot)			Wine (Nebiolo)
Added concentration level	L	A	H	L	A	H	L	A	H	L	A	H	L	A	H
4-Methyllumbelliferone	100.4	108.5	99.5	90.5	104.9	107.1	113.3	87.3	103.6	90.1	112.3	102.5	101.2	118.1	95.2
Salicylic acid	99.3	108.6	96.5	99.0	112.2	105.1	100.2	108.0	96.4	96.5	113.1	97.9	99.7	105.5	108.3
Vanillin	104.2	112.5	95.7	94.7	92.6	90.1	87.2	121.0	93.5	97.3	101.6	105.2	102.2	116.3	93.8
Coniferaldehyde	116.1	109.3	92.6	121.0	100.3	96.2	99.3	106.8	96.3	95.6	115.2	102.9	110.4	94.4	114.7
Protocatechin acid	87.0	101.9	113.9	103.3	93.9	107.5	110.7	93.8	106.4	99.2	108.0	97.1	111.8	93.0	96.2
Umbelliferone	106.2	113.3	97.6	93.0	107.1	106.5	109.1	95.8	103.2	88.8	90.2	121.3	99.5	108.1	96.4
Gallic acid	108.0	109.9	95.0	93.3	90.6	100.3	89.6	86.9	96.5	122.1	117.5	106.2	95.1	100.2	95.7
Vanillic acid	97.7	114.4	108.9	91.9	110.2	109.2	89.1	109.0	96.2	117.1	93.4	96.7	110.3	93.8	106.5
Syringaldehyde	77.2	120.3	92.6	105.7	103.6	96.0	105.5	93.3	100.6	97.9	93.5	107.4	97.0	98.1	96.8
Caffeic acid	75.6	89.2	114.5	116.7	99.9	95.9	116.0	100.7	98.5	111.4	102.8	97.6	115.9	90.0	101.6
Ferulic acid	99.1	108.1	101.1	98.9	94.7	102.4	100.0	99.4	100.7	106.5	102.9	104.6	104.3	103.1	98.2
Scopoletine	98.8	110.7	110.0	113.7	118.0	108.8	104.9	111.0	93.0	97.4	103.1	92.3	114.4	95.9	97.6
Synapaldehyde	118.4	119.7	92.8	106.2	112.0	96.5	87.8	107.9	93.0	96.9	100.2	104.4	100.4	93.3	91.1
Syringic acid	114.1	118.5	101.0	96.8	111.2	97.4	94.9	115.4	109.6	107.7	110.5	104.5	99.4	116.9	107.2
Resveratrol	113.3	124.4	106.0	100.8	114.2	115.3	97.6	117.1	91.0	101.9	105.7	99.2	93.8	81.8	105.7
Synapic acid	111.2	99.1	100.8	108.3	111.6	102.9	98.7	103.3	95.4	99.8	110.1	98.8	97.3	95.3	106.3
Chrysin	99.8	94.8	99.5	102.4	112.0	90.6	78.9	107.4	98.0	116.6	103.4	106.9	104.6	95.8	94.4
Pinocembrin	95.4	103.8	93.6	100.1	109.5	96.5	107.7	110.9	95.9	114.2	98.0	102.3	116.1	114.2	95.0
Naringenin	105.5	111.4	101.7	96.0	115.0	94.5	110.4	96.2	92.9	110.3	98.6	98.4	90.9	93.3	101.1
Kaempferol	74.4	81.1	96.2	91.6	107.9	96.7	101.6	91.6	94.0	101.2	114.4	105.5	99.3	96.0	104.4
Catechin	116.0	89.9	108.1	87.9	101.2	105.6	86.2	90.8	110.5	93.5	89.5	108.6	88.0	101.9	106.1
Hispidulin	98.1	106.2	100.2	100.5	104.9	91.6	106.8	106.2	95.7	102.3	111.1	97.8	95.9	93.0	93.9
Quercetin	95.6	81.0	112.5	93.9	108.9	125.3	93.8	84.5	110.4	95.1	82.4	113.3	103.9	104.4	102.5
Taxifolin	92.6	102.1	112.6	119.3	115.6	98.8	84.2	119.8	100.1	105.4	103.0	94.9	109.6	108.8	112.6
Chlorogenic acid	90.4	86.3	109.9	80.2	86.2	103.7	93.6	96.4	111.6	110.4	93.0	99.4	98.5	95.4	113.3
Rosmarinic acid	120.4	74.3	102.6	91.3	98.2	93.9	102.8	115.1	110.8	76.2	87.8	109.6	85.9	114.7	98.7
Isoquercetin	98.2	117.2	100.9	110.5	90.1	103.3	99.4	99.9	101.1	94.4	97.0	101.9	74.0	105.3	103.8
Rutin	103.9	81.0	106.8	116.8	108.8	96.6	90.0	111.3	96.4	103.5	96.8	101.4	91.1	111.7	97.6
p-Coumaric acid	105.3	90.6	97.2	114.3	101.6	98.0	97.5	109.4	96.5	106.0	103.1	95.8	100.4	102.1	110.6
Vitexin	93.2	113.1	106.5	92.5	104.0	95.2	110.2	103.2	100.2	93.0	110.8	91.1	106.7	105.3	103.2
Isorientin	96.0	97.1	101.6	111.2	103.3	98.5	80.9	110.1	118.1	90.7	109.1	97.5	92.1	95.6	105.3

• L: low; A: average; H: high.

The calculated recoveries ranged from 92.4 to 110.8% (Table 4) at the selected concentration levels and were considered suitable for analytes present in samples at mg L^-1 level, according to AOAC guidelines.²⁵

In general, the validation results demonstrate that the separation efficiency of the HILIC method contributes to the precision, LOD, and LOQ. The combination with sample preparation (including 30-fold dilution) establishes optimal conditions for recovery and linearity testing due to the effective extraction of phenolics from samples with a complex matrix.

Method application in samples of residues from beverage production

The proposed method was applied to five different samples of residues from beverage production (chromatograms in Figures S3-S5, SI section), and the results are presented in Table 5.

Table 5
Concentration of the 31 phenolic compounds investigated in samples of residues from beverage production after applying the developed method

Analyte	Average value ± relative standard deviation / (mg per g of dry sample)
	Coffee	Beer	Apple juice	Wine (Merlot)	Wine (Nebbiolo)
4-Methyllumbelliferone	< LOD	< LOD	< LOD	< LOD	< LOD
Salicylic acid	< LOD	< LOD	3.27 ± 0.28	< LOQ	< LOQ
Vanillin	< LOD	< LOD	< LOQ	8.69 ± 0.02	8.48 ± 0.27
Coniferaldehyde	< LOQ	9.36 ± 0.39	4.84 ± 0.85	< LOQ	< LOD
Protocatechin acid	28.2 ± 0.8	< LOD	16.1 ± 0.7	18.6 ± 0.2	10.1 ± 2.5
Umbelliferone	3.94 ± 0.54	< LOQ	< LOQ	< LOQ	< LOQ
Gallic acid	< LOD	14.8 ± 3.2	5.35 ± 1.01	26.8 ± 1.4	39.7 ± 2.2
Vanillic acid	< LOD	< LOD	< LOD	0.96 ± 0.16	1.46 ± 0.63
Syringaldehyde	< LOD	< LOQ	< LOQ	13.0 ± 0.9	12.0 ± 1.3
Caffeic acid	7.98 ± 2.53	0.08 ± 0.01	< LOD	9.32 ± 2.30	< LOQ
Ferulic acid	7.20 ± 1.18	34.2 ± 3.9	< LOD	< LOD	< LOD
Scopoletine	0.03 ± 0.01	< LOD	< LOD	< LOD	< LOD
Synapaldehyde	7.34 ± 0.06	42.1 ± 0.5	9.86 ± 3.34	< LOD	< LOD
Syringic acid	< LOD	< LOD	< LOD	20.6 ± 3.7	9.14 ± 2.44
Resveratrol	< LOD	< LOD	< LOD	< LOD	3.73 ± 0.47
Synapic acid	< LOD	< LOD	< LOD	< LOD	< LOD
Chrysin	< LOD	< LOD	< LOD	< LOD	< LOD
Pinocembrim	< LOD	< LOD	< LOD	< LOD	< LOD
Naringenin	< LOD	< LOD	< LOD	< LOQ	3.30 ± 1.03
Kaempferol	< LOD	< LOD	< LOD	0.52 ± 0.06	1.06 ± 0.23
Catechin	< LOD	< LOD	< LOD	< LOD	69.3 ± 4.2
Hispidulin	< LOD	< LOQ	< LOD	< LOQ	< LOQ
Quercetin	< LOQ	< LOD	< LOD	4.41 ± 0.83	6.87 ± 1.15
Taxifolin	< LOD	< LOD	< LOD	0.69 ± 0.60	8.10 ± 1.66
Chlorogenic acid	65.3 ± 0.6	< LOD	10.3 ± 0.4	< LOD	31.1 ± 1.7
Rosmarinic acid	< LOD	< LOD	< LOD	< LOD	< LOQ
Isoquercetin	< LOD	< LOD	< LOD	< LOD	< LOD
Rutin	< LOD	< LOD	< LOD	< LOD	< LOD
p-Coumaric acid	9.18 ± 0.64	1.06 ± 0.12	< LOQ	5.41 ± 0.34	< LOD
Vitexin	< LOD	< LOD	< LOD	< LOD	< LOD
Isorientin	< LOD	< LOD	< LOD	< LOD	< LOD
Total quantified phenolics	129.2 ± 6.4	101.6 ± 8.2	49.8 ± 6.6	108.9 ± 10.5	204.4 ± 20.0

• LOD: limit of detection (mg L^-1): 4-methyllumbelliferone = 0.009; salicylic acid = 0.018; vanillin = 0.034; coniferaldehyde = 0.002; protocatechin acid = 0.108; umbelliferone = 0.005; gallic acid = 1.701; vanillic acid = 0.020; syringaldehyde = 0.020; caffeic acid = 0.008; ferulic acid = 0.009; scopoletine = 0.001; synapaldehyde = 0.009; syringic acid = 0.098; resveratrol = 0.012; synapic acid = 0.024; chrysin = 0.005; pinocembrim = 0.008; naringenin = 0.011; kaempferol = 0.026; catechin = 0.066; hispidulin = 0.005; quercetin = 0.043; taxifolin = 0.022; chlorogenic acid = 0.050; rosmarinic acid = 0.004; isoquercetin = 0.360; rutin = 0.503; p-coumaric acid = 0.016; vitexin = 0.022; isorientin = 0.135; LOQ: limit of quantification (mg L^-1): 4-methyllumbelliferone = 0.026; salicylic acid = 0.054; vanillin = 0.103; coniferaldehyde = 0.007; protocatechin acid = 0.327; umbelliferone = 0.016; gallic acid = 5.155; vanillic acid = 0.061; syringaldehyde = 0.061; caffeic acid = 0.023; ferulic acid = 0.027; scopoletine = 0.004; synapaldehyde = 0.029; syringic acid = 0.296; resveratrol = 0.036; synapic acid = 0.071; chrysin = 0.015; pinocembrim = 0.025; naringenin = 0.035; kaempferol = 0.080; catechin = 0.200; hispidulin = 0.015; quercetin = 0.129; taxifolin = 0.067; chlorogenic acid = 0.0150; rosmarinic acid = 0.013; isoquercetin = 1.091; rutin = 1.524; p-coumaric acid = 0.048; vitexin = 0.067; isorientin = 0.400.

In this study, 21 out of 31 analytes were quantified in the samples. Phenolic compounds such as 4-methyllumbelliferone, synapic acid, chrysin, pinocembrin, isoquercetin, rutin, vitexin, and isorientin presented concentrations lower than the LOD, while hispidulin and rosmarinic acid were below the LOQ for all tested samples. Wine production residues exhibited the highest diversity of phenolic compounds. According to Kekelidze et al.,⁴⁴ the total phenolic content level is a key indicator of the health benefits of wines, as it is associated with antioxidant capacity due to the presence of compounds such as gallic acid, catechin, quercetin, and caffeic acid, among others. Eighteen analytes were identified in the Nebbiolo grape variety, with catechin, gallic, and chlorogenic acid showing the highest concentrations. Gallic, syringic, and protocatechin acid were the predominant phenolics in the Merlot variety, with 16 of the analytes of the method identified. Chlorogenic and protocatechuic acids had the highest concentrations among the 10 phenolics identified in the residues of coffee production. Residues from beer and apple juice production contained 10 phenolic compounds, with 6 of them present in quantifiable amounts. According to the method, sinapaldehyde, along with ferulic and gallic acids, exhibited the highest levels in the residues of beer production. Furthermore, residues of apple juice had the lowest concentration of phenolic compounds among the evaluated residues.

Comparison of the proposed method with others from the literature

The developed method was compared with recent methodologies described in the literature¹¹,¹³,¹⁴,²⁰,²¹,²²,²³,⁴⁵,⁴⁶ for the determination of phenolic compounds in fruit and vegetable extracts (Table S10, SI section).

The methods mentioned can be categorized into those that identify and quantify specific groups or classes of phenolic compounds. However, the increasing consumption of these compounds due to their health benefits requires the development of methodologies capable of determining a wide range of analytes in foods. In the cited works, analyses were typically conducted using LC coupled with sensitive detectors such as MS. Nevertheless, most of them employed RP-LC,¹¹,¹³,¹⁴,²⁰,²¹,⁴⁵ with only a few utilizing the HILIC mode.²²,²³,⁴⁶ In the proposed method, a new separation of 31 phenolic compounds was presented using a FRULIC-N column in HILIC mode, expanding the range of applications and demonstrating its versatility. In particular, the SP of the FRULIC-N column presents excellent hydrophilicity and offers different possibilities of SP-analyte interactions. This column combined with an acetonitrile-rich MP has been shown to be useful for retaining and separating a wide range of compounds with structural characteristics as diverse as phenolics. Despite these potential benefits, few applications of HILIC for the separation of phenolics have been reported in the literature¹⁷,²²,⁴⁶ and the methods described mostly involve columns with SP composed of pure silica or zwitterionic functional groups. Other methods²³ present the phenolic profile in samples without quantification or validation process due to the absence of analytical standards. According to the literature, the proposed method is the first to employ the FRULIC-N column to determine a wide range of phenolic compounds. The developed method expanded the range of phenolic compounds separated by the HILIC mode and reduced the analysis time by increasing the temperature.¹⁷,²²,²³,⁴⁶ Moreover, the new method was able to determine the phenolic compounds with greater sensitivity than other methods that use RP-LC due to the advantages of HILIC, such as its compatibility with the MS system. The results of LOD and LOQ for the phenolic compounds in the proposed method were better than those obtained by Pico et al.¹¹ (protocatechuic acid, vanillic acid, caffeic acid, ferulic acid, syringic acid, kaempferol, catechin, and p-coumaric acid); Irakli et al.,¹⁴ (vanillic acid, caffeic acid, sinapic acid, chrysin, narigenin, kaempferol, rosmarinic acid and p-coumaric acid); Machado et al.,¹³ (quercetin); Setyaningsih et al.,²¹ (vanillin, protocatechin acid, vanillic acid, caffeic acid, ferulic acid, syringic acid, synapic acid, quercetin, chlorogenic acid, p-coumaric acid) and Qiao et al.,²⁰ (vanillic acid, ferulic acid, syringic acid, p-coumaric acid).

Furthermore, the methods described do not cover analytes such as 4-methylumbelliferone, salicylic acid, coniferalddehyde, umbelliferone, syringaldehyde, scopoletin, sinapaldehyde, pinocembrin, hispidulin, isoquercetin, vitexin, isorientin that were determined in the developed method with LOD < 0.36 and LOQ < 1.09 mg L^-1.

In the developed method, optimization strategies based on multivariate analyses were explored to select the optimal conditions among the factors studied. This approach stands out among some works in the literature¹¹,¹³,¹⁴,²²,²³,⁴⁶ as it allowed the observation of complementary effects between the factors and the simultaneous optimization was convenient for the development of the method.

The validation process of the proposed method was carefully carried out according to Eurachem²⁴ and AOAC²⁵ for several analytical parameters using appropriate statistical tools. The developed method was applied to samples with matrices as complex as those in other works in the literature considering the presence of multiple components and the variable nature of the different beverage industry residues studied.

This study was able to separate 31 phenolic compounds, including phenolic acids, stilbenes, phenolic aldehydes, coumarins, and flavonoids, in less than 11 min of chromatographic separation time, indicating diverse quantification and high-throughput analysis when compared to other methods from the literature.¹¹,¹³,²⁰,²¹,²²,⁴⁶

Conclusions

An analytical method using HILIC-MS/MS was developed and optimized for the simultaneous determination of 31 phenolic compounds in a chromatographic separation time of less than 11 min achieved with the help of increasing the temperature on a FRULIC-N column. Considering its excellent hydrophilicity, a FRULIC-N column was used for the first time to separate a large number of phenolic compounds, expanding the range of HILIC determinations. The designs of experiments were employed as an important tool for optimizing the parameters that affect the separations of the phenolic compounds. The analytical parameters of the developed method were evaluated and considered in compliance with the guidelines employed. The developed method provided greater sensitivity in a shorter analysis time for phenolics than RP-LC methods reported in the literature. The application of the developed method to the selected samples allowed the identification of 21 analytes, indicating that wine production residues have the greatest diversity of phenolics among the samples analyzed. Finally, this method has the potential to evaluate 31 phenolic compounds in by-products of beverage production, including coffee, apple juice, beer, and wine.

Supplementary Information

Supplementary data are available free of charge at http://jbcs.sbq.org.br as PDF file.

Acknowledgments

The authors are thankful to Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) e INCT-Catálise/FAPESC/CNPq/ CAPES (FAPESC-2019TR0847, CNPq-444061/2018). We are also grateful to Farma Service Bioextract for support.

References

• ¹
  Oliveira, D. C. X.; Rosa, F. T.; Simões-Ambrósio, L.; Jordao, A. A.; Deminice, R.; Nutrition 2019, 63-64, 29. [Crossref]
  » Crossref
• ²
  Sarker, U.; Oba, S.; Food Chem. 2018, 252, 72. [Crossref]
  » Crossref
• ³
  Peña-Bautista, C.; Baquero, M.; Vento, M.; Cháfer-Pericás, C.; Clin. Chim. Acta 2019, 491, 85. [Crossref]
  » Crossref
• ⁴
  Sihag, S.; Pal, A.; Ravikant; Saharan, V.; Biocatal. Agric. Biotechnol. 2022, 42, 102368. [Crossref]
  » Crossref
• ⁵
  Núñez-López, G.; Herrera-González, A.; Hernández, L.; Amaya-Delgado, L.; Sandoval, G.; Gschaedler, A.; Arrizon, J.; Remaud-Simeon, M.; Morel, S.; Enzyme Microb. Technol. 2019, 122, 19. [Crossref]
  » Crossref
• ⁶
  Petrón, M. J.; Andrés, A. I.; Esteban, G.; Timón, M. L.; J. Cereal Sci. 2021, 98, 103162. [Crossref]
  » Crossref
• ⁷
  Zekeya, N.; Ibrahim, M.; Mamiro, B.; Ndossi, H.; Kilonzo, M.; Mkangara, M.; Chacha, M.; Chilongola, J.; Kideghesho, J.; Saudi J. Biol. Sci. 2022, 29, 103273. [Crossref]
  » Crossref
• ⁸
  Solomakou, N.; Loukri, A.; Tsafrakidou, P.; Michaelidou, A. M.; Mourtzinos, I.; Goula, A. M.; Sustainaible Chem. Pharm. 2022, 25, 100592. [Crossref]
  » Crossref
• ⁹
  Silva, S. S.; Justi, M.; Chagnoleau, J. B.; Papaiconomou, N.; Fernandez, X.; Santos, S. A. O.; Passos, H.; Ferreira, A. M.; Coutinho, J. A. P.; Sep. Purif. Technol. 2023, 304, 122344. [Crossref]
  » Crossref
• ¹⁰
  Kaanin-Boudraa, G.; Brahmi, F.; Wrona, M.; Nerín, C.; Moudache, M.; Mouhoubi, K.; Madani, K.; Boulekbache-Makhlouf, L.; LWT--Food Sci. Technol. 2021, 151, 112158. [Crossref]
  » Crossref
• ¹¹
  Pico, J.; Yan, Y.; Gerbrandt, E. M.; Castellarin, S. D.; J. Food Compos. Anal. 2022, 108, 104412. [Crossref]
  » Crossref
• ¹²
  Muller, M.; de Villiers, A.; J. Chromatogr. A 2023, 1692, 463843. [Crossref]
  » Crossref
• ¹³
  Machado, P. G.; Londero, D. S.; Farias, C. A. A.; Pudenzi, M. A.; Barcia, M. T.; Ballus, C. A.; Food Chem. 2024, 432, 137296. [Crossref]
  » Crossref
• ¹⁴
  Irakli, M.; Skendi, A.; Bouloumpasi, E.; Chatzopoulou, P.; Biliaderis, C. G.; Antioxidants 2021, 10, 2016. [Crossref]
  » Crossref
• ¹⁵
  Shu, Y.; Lang, J. C.; Breitbach, Z. S.; Qiu, H.; Smuts, J. P.; Kiyono-Shimobe, M.; Yasuda, M.; Armstrong, D. W.; J. Chromatogr. A 2015, 1390, 50. [Crossref]
  » Crossref
• ¹⁶
  Qiao, L.; Shi, X.; Xu, G.; TrAC, Trends Anal. Chem. 2016, 81, 23. [Crossref]
  » Crossref
• ¹⁷
  Sentkowska, A.; Pyrzynska, K. In Polyphenol in Plants; 2^nd ed.; Watson, R. R., ed.; Elsevier: London, UK, 2019, ch. 20. [Crossref]
  » Crossref
• ¹⁸
  Wang, C.; Jiang, C.; Armstrong, D. W.; J. Sep. Sci. 2008, 31, 1980. [Crossref]
  » Crossref
• ¹⁹
  Silva, C. G. A.; Bottoli, C. B. G.; Collins, C. H.; Quim. Nova 2016, 39, 210. [Crossref]
  » Crossref
• ²⁰
  Qiao, B.; Li, C.; Liang, C.; Li, X.; Tian, M.; Li, Q.; Zhao, C.; Fu, Y.; S. Afr. J. Bot. 2022, 148, 387. [Crossref]
  » Crossref
• ²¹
  Setyaningsih, W.; Saputro, I. E.; Carrera, C. A.; Palma, M.; Food Chem. 2019, 288, 221. [Crossref]
  » Crossref
• ²²
  Sentkowska, A.; Pyrzyńska, K.; LWT--Food Sci. Technol. 2018, 93, 641. [Crossref]
  » Crossref
• ²³
  Steevensz, A. J.; MacKinnon, S. L.; Hankinson, R.; Craft, C.; Connan, S.; Stengel, D. B.; Melanson, J. E.; Phytochem. Anal. 2012, 23, 547. [Crossref]
  » Crossref
• ²⁴
  Eurachem; The Fitness for Purpose of Analytical Methods, 2^nd ed.; Magnusson, B.; Örnemark, U., eds.; 2014. [Link] accessed in August 2024
  » Link
• ²⁵
  Association of Official Analytical Chemists (AOAC); Appendix F: Guidelines for Standard Method Performance Requirements, AOAC: Gaithersburg, 2016. [Link] accessed in August 2024
  » Link
• ²⁶
  Schulz, M.; Borges, G. S. C.; Gonzaga, L. V.; Seraglio, S. K. T.; Olivo, I. S.; Azevedo, M. S.; Nehring, P.; de Gois, J. S.; de Almeida, T. S.; Vitali, L.; Spudeit, D. A.; Micke, G. A.; Borges, D. L. G.; Fett, R.; Food Res. Int. 2015, 77, 125. [Crossref]
  » Crossref
• ²⁷
  Analyst®, version 1.6.2; AB Sciex Pte. Ltd., Concord, CA, USA, 2013.
• ²⁸
  Statistica®, version 13.5.0.17; TIBCO Software Inc., Palo Alto, CA, USA, 2018.
• ²⁹
  Snedecor, G. W.; Cochran, W. G.; Statistical Methods, 8^th ed.; Iowa State University Press: Ames, 1989.
• ³⁰
  Shapiro, S. S.; Wilk, M. B.; Biometrika 1965, 52, 591. [Crossref]
  » Crossref
• ³¹
  Microsoft Excel®, version 14.0; Microsoft Corporation, Redmond, WA, USA, 2010.
• ³²
  Hoff, R. B.; Rübensam, G.; Jank, L.; Barreto, F.; Peralba, M. C. R.; Pizzolato, T. M.; Díaz-Cruz, M. S.; Barceló, D.; Talanta 2015, 132, 443. [Crossref]
  » Crossref
• ³³
  Bruce, P.; Minkkinen, P.; Riekkola, M. L.; Microchim. Acta 1998, 128, 93. [Crossref]
  » Crossref
• ³⁴
  Hao, Z.; Xiao, B.; Weng, N.; J. Sep. Sci. 2008, 31, 1449. [Crossref]
  » Crossref
• ³⁵
  Helali, Y.; Bourez, A.; Marchant, A.; Heyden, Y. V.; Antwerpen, P. V.; Delporte, C.; Talanta 2024, 270, 125541. [Crossref]
  » Crossref
• ³⁶
  Xu, Q.; Tadjimukhamedov, F. K.; J. Pharm. Biomed. Anal. 2022, 219, 114936. [Crossref]
  » Crossref
• ³⁷
  Sentkowska, A.; Biesaga, M.; Pyrzynska, K.; Talanta 2013, 115, 284. [Crossref]
  » Crossref
• ³⁸
  Taraji, M.; Haddad, P. R.; Amos, R. I. J.; Talebi, M.; Szucs, R.; Dolan, J. W.; Pohl, C. A.; Anal. Chim. Acta 2018, 1000, 20. [Crossref]
  » Crossref
• ³⁹
  Dolzan, M. D.; Spudeit, D. A.; Breitbach, Z. S.; Barber, W. E.; Micke, G. A.; Armstrong, D. W.; J. Chromatogr. A 2014, 1365, 124. [Crossref]
  » Crossref
• ⁴⁰
  Vallaro, M.; Ermondi, G.; Caron, G.; Eur. J. Pharm. Sci. 2020, 145, 105232. [Crossref]
  » Crossref
• ⁴¹
  Qiu, H.; Loukotková, L.; Sun, P.; Tesařová, E.; Bosáková, Z.; Armstrong, D. W.; J. Chromatogr. A 2011, 1218, 270. [Crossref]
  » Crossref
• ⁴²
  Padivitage, N. L. T.; Dissanayake, M. K.; Armstrong, D. W.; Anal. Bioanal. Chem. 2013, 405, 8837. [Crossref]
  » Crossref
• ⁴³
  McCalley, D. V.; J. Chromatogr. A 2020, 1612, 460655. [Crossref]
  » Crossref
• ⁴⁴
  Kekelidze, I.; Ebelashvili, N.; Japaridze, M.; Chankvetadze, B.; Chankvetadze, L.; Ann. Agrar. Sci. 2018, 16, 34. [Crossref]
  » Crossref
• ⁴⁵
  Nia, N. N.; Hadjmohammadi, M. R.; Microchem. J. 2021, 170, 106721. [Crossref]
  » Crossref
• ⁴⁶
  Sentkowska, A.; Biesaga, M.; Pyrzynska, K.; J. Anal. Methods Chem. 2016, 2016, 1. [Crossref]
  » Crossref

Edited by

Editor handled this article: Andrea R. Chaves (Associate)

Publication Dates

• Publication in this collection
  23 Sept 2024
• Date of issue
  2025

History

• Received
  08 Apr 2024
• Accepted
  09 Sept 2024