Proximate composition, functional properties and quantitative analysis of benzoyl peroxide and benzoic acid in wheat flour samples: effect on wheat flour quality

Sample collection

Test samples included commercial SF (n = 4, SF) purchased from the local supplier from different flour mills (with additives) and a control sample (n = 1, WF) without additives was prepared by mechanical grinding the Triticum aestivum L. seeds (Inqulab 19), harvested from the crop grown on the experimental field of University of Agriculture, Faisalabad, under optimized field conditions (16/8 D/N; 23 ± 1 °C; 14 Inch water) without any fertilizer and insecticide. The variety of wheat seeds and growth conditions were kept same as used by flour mills to generate SF, to avoid ambiguity in results. The WF flour sample was passed through sieve (75 µm size) before packing them into air-tight plastic containers. The size was in agreement with the particle size of SF purchased from flour mills.

Determination of functional properties

Bulk density of flour was determined as described previously (Lam et al., 2008) by following ASTM D1895B prescribed procedure. The sample was allowed to flow freely in a circular container (0.615 L) with a suspended funnel of opening diameter (1.5 cm). The height of funnel was kept about 20 cm and the powder was stirred continuously to avoid clogging inside the opening. Container with the sample was dropped few times from the height of 150 mm to allow settling and release of air. Weight of the container with the sample was determined and weight/volume (loose bulk density) was obtained. Density was determined through the formula, $d={\displaystyle \frac{m}{v}}$, where mass (g) is the weight of the sample and volume (mL) is the volume of the material. Water absorption capacity and oil absorption capacity (WAC and OAC respectively) were determined through the method described by Beuchat (1977). In this method, 1 g of sample was allowed to mix with 10 mL of distilled water for about 30 s. Sample was then allowed to stand at room temperature (25 ± 2 °C) for the next 30 min and centrifuged at 3,000 rpm (30 min). Volume of the supernatant was determined and WAC (mg/mL) was calculated by formula $\mathrm{W}\mathrm{A}\mathrm{C}=V_{\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{a}\mathrm{l}}-V_{\mathrm{f}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{l}}$, where V is the volume of water (mL). Similar procedure was repeated for OAC determination using soybean oil (Sp. Gravity 0.9092) as the absorbing agent. Emulsifying stability and emulsifying activity (ES and EA respectively) were determined by the method described by Neto et al. (2001) with modifications. About five mL of flour dispersion (10 mg/mL of water) was homogenized with five mL of soybean oil for 1 min through vigorous shaking. The emulsion was then centrifuged (2–6; Sigma, Neustadt, Germany) at 1,100 rpm for 10 min. Height (cm) of emulsified layer (ELH) was deducted from the total height of the tube contents (TC) to estimate the EA by the equation $\mathrm{E}\mathrm{A}=\left({\displaystyle \frac{\mathrm{E}\mathrm{L}\mathrm{H}}{\mathrm{T}\mathrm{C}}}\right)\times 100$. ES was calculated by heating the emulsion at 80 °C for 30 min before centrifuging at 1,300 rpm for 10 min. ES was then calculated by $\mathrm{E}\mathrm{S}=\left({\displaystyle \frac{\mathrm{E}\mathrm{L}\mathrm{H}\mathrm{A}}{\mathrm{T}\mathrm{C}\mathrm{A}}}\right)\times 100$, where ELHA is the height of emulsified layer after heating and TCA is total content of the tube before heating. The effect of EA and ES was determined by varying the capacity of the flour samples. Foaming stability and capacity (FS and FC) of flours were determined by the method reported by Coffmann & Garciaj (1977), where about 100 mL of distilled water was mixed with 10 g of flour. The suspension was mixed vigorously for 5 min on the magnetic stirrer (MSC Digital, IRMCO, Berlin, Germany). The initial solution volume $V_1$ and the final solution volume $V_2$ were recorded (using graduated cylinder). Foaming Capacity (FC) was also calculated from the formula $\mathrm{F}\mathrm{C}=\left({\displaystyle \frac{V_2-V_1}{V_1}}\right)\times 100$. Foaming Stability (FS) was also determined by the foam volume that was left for 8 h and expressed as percentage of initial foam volume. To determine the gelation properties; Least gelatinization concentration (LGC) and gelatinization temperature (GnT), distilled water sample suspensions (2–10% w/v) were prepared. About 10 mL of these suspensions were transferred into the test tubes. In the boiling water bath these test tubes were heated for 1 h and then cooled for 2 h at 4 °C in a refrigerator. Least gelation concentrations were taken when the samples did not fall from the inverted test tubes.

Proximate composition

Proximate composition was determined through FT-NIR spectroscopy (TENSOR 37 FTIR Spectrometer; Burker, Bremen, Germany) as well as conventional methods to get reliable results. These properties included moisture content (MC), ash contents (AC), crude protein (CP), gluten and starch contents. Other properties such as crude fiber and fat (CFF) were determined through solvent extractions. For crude fiber about 2.5–3 g of sample was transferred to Soxhlet and extracted with petroleum ether. The air dried extracted sample was transferred to one L conical flask. Whole boiling dilute sulfuric acid (1.25% w/v) was transferred to the flask containing sample and immediately connected with a water cooled reflux condenser and heat it so that the contents start boiling in 1 min. Few drops of octanol were added to avoid bumping and risk of foaming. The flask was continuously shaken and boiled for 30 min. Flask is then removed and filtered through fine linen (18 threads to a cm) placed in a funnel and washed with boiling water until no longer acid to litmus is achieved. The residue on linen was then washed with the 200 mL of boiling NaOH (1.25% w/v). Immediately the flask was connected to the reflux condenser and boiled for 30 min. The residue was then filtered through filtering cloth and residues were thoroughly washed with boiling water. The residues were transferred to Gooch crucible having thin compact layer of ignited asbestos. Finally wash the residues with 15 mL ethyl alcohol. The Gooch crucible was dried in air oven at 105 ± 2 °C in an air oven until constant weight is achieved. The crucible was weighed and placed in a muffle furnace to burnt all carbonaceous matter. Crucible was then cooled and weighed again. Crude fiber was then calculated by the following formula; $\mathrm{c}\mathrm{r}\mathrm{u}\mathrm{d}\mathrm{e}\mathrm{f}\mathrm{i}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{\%}\mathrm{b}\mathrm{y}\mathrm{w}\mathrm{t}={\displaystyle \frac{\left(W_1-W_2\right)\times 100}{W}}$, where W₁ is weight of Gooch crucible before ashing, W₂ is weight of Gooch Crucible containing asbestos and ash and W is weight of the dried material taken for the test. The fat content of wheat was determined using the method of Association of Official Analytical chemists (AOAC International, 2000) with petroleum ether as solvent. In this method the fat was extracted by petroleum ether. Sample of about 3.0 g was weighed and added in a labeled thimble. About 150 mL of petroleum ether was then added to it and boiled up to 40–60 °C in 250 mL boiling flask. The thimble was tightly plugged with cotton wool and Soxhlet apparatus was assembled to allow it to reflux for 24 h the thimble was carefully removed and petroleum ether from the top was drained into another container for further use. The petroleum ether was then evaporated completely in a hot air oven and desiccator was weighed following cooling. The weight was determined by formula; $\mathrm{f}\mathrm{a}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{\%}\mathrm{b}\mathrm{y}\mathrm{w}\mathrm{t}={\displaystyle \frac{W_1\times 100}{W}}$, where W₁ is the weight of the residue in desiccator and W is initial weight of the sample.

Since the bulk density (BD; ρ) varies with MC, therefore, it was determined following procedure described in ASABE standard S358.2 (Theerarattananoon et al., 2011). During this procedure, 100 g of sample was dried in a forced air convection oven (IM-115, Berlin, Germany) at 103 °C for 24 h. The sample was then weighed on digital balance (0.01 g precision; TE-313S-DS, Göettingen, Germany) and MC was calculated by the equation; $\mathrm{M}\mathrm{C}=W_{\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{a}\mathrm{l}}-W_{\mathrm{f}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{l}}$. MC was also determined through FT-NIR spectroscopy. Flour samples of different moisture level were utilized to develop the model for calibration and multivariate analysis was conducted after gathering their spectra. Unknown samples were then analyzed against the calibration curve to get the moisture contents. Crude protein was determined through semi micro-Kjeldahl method (AACC adopted method 46-13; AACC, 1995) and auto protein analyzer (Kjeltec 2400 auto-analyzer; FOSS Analytical, Hillerød, Denmark). Here 1 g of flour sample was used along with keeping nitrogen to protein conversion factor of 5.7. AACC method 38-12 was utilized to determine the gluten contents of the selected flour samples (25 g each).

Analytical method

NIR Omega G Analyser (Bruins Instruments, Salem, NH, USA) was employed to analyze different parameters (protein, starch, fat, moisture and gluten) of the flour and grain samples. The spectral transmissions range was 700–1,100 nm with 5 nm scan increment, measured at controlled room conditions of 24 ± 1 °C, RH 34 ± 2%.

MIR spectroscopy

The FTIR transmission spectra were recorded at Burker-TENSOR 37 FTIR spectrometer with Michelson interferometer. Working range of the spectrometer was 4,000–12,000 cm⁻¹ and spectra generated were interpreted on the basis overtones of different functional groups in the product. Resolution of spectrometer was 4 cm⁻¹ (max scan interval value was 2 cm⁻¹) with maximum scan time kept at 5 s. MIR spectra were recorded at Opus 6.0 Burker Software using Attenuated Total Reflectance (ATR) unit. The reference spectrum (empty sample bottle) was utilized as background measurement before loading in sample’s spectra. About 8–10 g of sample was added in the sample bottle to generate the spectra in diffused reflectance mode. Three spectra per sample were recorded by rotating the sample bottle at 120°. The measurements were carried out under the controlled conditions of temperature 24 ± 1 °C, RH 34 ± 2%. Each spectra were the average of three scans per object.

BP and BA quantification

Calibration graph for BPO

Standard stock solution (SS) was generated by dissolving pure BPO (60 mg/L) in diethyl ether (100% purity). Working standards were obtained by diluting SS with appropriate volume of diethyl ether. For BA stock solution, pure compound (100 mg) was dissolved in 100 mL of methanol and working standards were prepared by diluting the stock solution. Calibration curve was then generated by plotting the absorbance value against concentration.

Extraction procedure for BPO and BA

The standard procedure was carried out (at room temperature) in a flask with grinding stopper. About 100 mL of diethyl ether was added to 50 g of flour (both control and the one which is already mixed and sifted with BPO). This mixture was shaken vigorously on a magnetic stirrer for 10 min and left to settle for 15 min. Upper layer of this solution (containing the products of reaction) was withdrawn through the pipette and transferred into Falcon polypropylene tube (10 mL) and held into ice until HPLC analysis.

HPLC method

The supernatant was analyzed by Waters 600 HPLC system at Inertsil ODS-80A column (5 μm, 4.6 × 250 mm; GL Science, Tokyo, Japan) equipped Inertsil ODS-3 guard column (10 × 4 mm i.d.) and Waters 2,996 Photodiode array detector. The detection wavelength was kept at 235 nm and column oven at 40 °C. For isocratic separation the conditions were as follows: Water (Solvent A), acetonitrile (CAN; solvent B) and Benzoic acid (Solvent C); 55% B:45% A as mobile phase for 1 mL/min. The gradient conditions for analysis were as follows: Water-glacial acetic acid (1,000:1) (Solvent A), ACN-glacial acetic acid (1,000:1) (solvent B); 18% B (10 min) was increased to 60% B (11–15 min hold) at flow rate 1.2 mL/min and column temperature 35 °C.

Statistical analysis

All samples were analyzed in triplicate (both biological and experimental replicates), therefore standard error of mean (SEM) was applied using the Statistixl 1.9 Add-in package within Excel 2007. Two-way analysis of variance (ANOVA) was conducted, followed by post-hoc Tukey tests that separated the treatment into groups (at P < 0.01 and P < 0.05). The aim was to give the significant difference in the data sets from different flour samples which was not achievable through univariate or one-way ANOVA.

Functional properties

The oil absorption capacity (OAC; Table 1) showed that the SF sample has highest lipophilic tendency of about 2.87 mL/g. Highest OAC (188 mL/100 g compared to 146 mL/100 g for WF) and WAC (408 g/100 mL compared to 140 g/100 mL for WF) were obtained for SF4. All considered flour mill samples had almost similar results for OAC. Water Absorption Capacity (WAC) was, however, higher (140 g/100 mL) for WF compared to SF (<123 g/100 mL).

The results revealed that emulsifying Activity (EA) was higher for WF (43.7%), whereas, stability (ES) is higher for all SF samples (<42%). EA and ES of the WF and SF may also vary due to the milling process. The emulsifying properties varied inversely, which means that WF had highest EA and lowest ES. Foaming Capacity (FC) and Foaming stability (FS) collectively form the foaming properties of any flour. Both of these properties are directly proportional to one another, which were observed to be higher for WF (12.9% FC and 1.94% FS). FC and FS of WF is more (<12% and <1% respectively) compared to all SF samples (>9% and >1% respectively). A highly significant difference (P < 0.01) was observed when values were compared statistically with those of WF.

Gelation capacity

Gelation capacity (including gelatinization temperature GnT and least Gelatinization concentration LGC) are attributed and controlled by the balance between hydrophilic interactions and repulsive electrostatic interactions between the water/oil and proteins (Casanova et al., 2008). Results (Table 1) indicated that WF had higher gelation capacity (GnT = 59.21 °C and LGC = 8%) compared to all SF samples (P < 0.001 when datasets were compared with the WF dataset). It can also be observed that both considered parameters for gelation capacity are directly related to each other such that increase in one also showed increase in the other.

Proximate composition

The proximate composition (moisture, crude fiber, fat, ash, starch and damaged starch) of all the flour samples is as summarized in the Table 2. Moisture content of SF is less (3.84–4.25%), protein contents of WF were higher (8.9% compared to 4.6% for SF) and total starch was also high for WF (76.92% compared to 50.21% for SF). The results indicated that intense milling processes have detrimental effect on several properties of the flour. Most of the components such as crude protein, gluten, damaged starch datasets for SF showed highly significant difference (P < 0.01) when compared with the WF dataset. This showed that quality of flour was deteriorated while processing and refining.

Benzoyl peroxide concentration

Benzoyl peroxide and BA were determined simultaneously using gradient analysis (Table 3). The retention times were observed to be 17.5 min for BP and 7.8 min for BA (Figs. 1 and 2). The maximum absorption of BPO was obtained at 195 nm and 235 nm, however, a wavelength of 235 nm was kept as standard for measuring the BP in the samples, considering the possible interference with the food ingredients appearing at 195 nm. The calibration curve had excellent linearity for the range of 0.05–16 µg/g for BPO and 0.2–15 µg/g for BA. The acquisition following isocratic gradient with 55% ACN, showed an excellent linearity for BP, however BA was not detected. Therefore, measurements at gradient conditions were kept as standard for the analysis.

Benzoyl peroxide is a free radical initiator and it causes the oxidation of carotenoids by free radical mechanism. The process (Fig. 3) leads to the formation of benzoic acid (BA) as a by-product (Saiz, Manrique & Fritz, 2001; Shan et al., 2007; Sumnu & Sahin, 2008). No BA or BP were observed for the WF sample (Fig. 2). In the controlled samples, 30 µg/g of BP was added to the wheat flour samples (grown under standard field conditions) and 99.5% recovery (29.5 µg/g) was observed soon after adding. After 3 h of bleaching, the amount of BP reduced significantly to 4 µg/g (Fig. 4) that reached to zero after 8 h of exposure. The contents of BA were observed to be 2.84 µg/g as recovered soon after the addition. This quantity increased to 8.9 µg/g after 3 h and increased further to 13.5 µg/g after 8 h of exposure. The contents of BA were determined again after 12 h though very small increase in quantity was observed (13.75 µg/g), which shows that the process of conversion got stabilized with slight variation. To confirm this, the flour was analyzed again after 16 h and quantity of BA didn’t change much (13.77 µg/g). Local standards for maximum acceptable quantity have not yet been specified however, international standards (60 µg/g for BP; 0–5 mg/kg for BA as per body weight by JECFA acceptable daily intakes) were considered in the current study. Analysis of WF and SF available commercially for BP and BA showed higher rates of BA in the SF samples (Table 4). This indicated higher amount of BA intake when SF based products are consumed.

Estimation of daily intake

Estimation of dietary exposure of BA through just wheat flour consumption was estimated (Table 5). Since flour is important parameter of people diet, therefore, it was deemed necessary to give the level of exposure through flour only. Rest of the food groups such as noodles and drinks, though unavoidable, also contain certain amount of BA but they were not considered in the current study. As evaluated through questionnaire study most of the considered subjects (72.5) were consuming SF purchased from the flour mills.

High Performance Liquid Chromatography was used for quantification of BP in both white and WF samples. BA was also determined because BP was decomposed into benzoic acid within limited days (Fig. 4). A research on bleaching agents including BPO and BA by HPLC during bleaching process of wheat flour. The retention time of BP was 17.5 min and that of BA was 7.6 min. After 30 h of bleaching BPO concentration was 11 ppm. After 3 months its concentration was reduced to 6 ppm. These results demonstrated that when benzoyl peroxide added to flour their greater amount was decomposed into benzoic acid within limited days of treatment. The analytical results of present study showed that that retention time of BPO was 17.5 min and BA was 7.5 min at 235 nm. In WF samples no content of BPO and BA were found. In white flour samples BP content ranges from 6.6 to 21 mg/kg and BA content ranges from 13 to 28 mg/kg.