LPS001

This content was archived on June 24, 2013.

Laboratory Procedure for the Determination
of benzene in beverages

This content was archived on June 24, 2013.

Scope

A method was employed for analysis of benzene in carbonated and non carbonated soft drinks, cocktail mixes, de-alcoholized wines and sports/energy drinks.

Principle of the method

The method utilizes headspace and gas chromatography-mass spectrometry (GC-MS) in a selected ion monitoring mode (SIM).

A quantity of sample (1 g) is transferred to a headspace vial and a known amount of deuterated benzene (d6) is added as an internal standard. After an equilibration at room temperature, the headspace (50 μL) from the vial is injected into GC-MS. The signals of characteristic ions for benzene m/z (78, 77, 51) and deuterated benzene (84, 82, 52) were compared to those obtained from calibration standards.

Benzene concentration in samples is calculated by utilizing an isotope dilution methodology, using ions with m/z of 78 and 84.

Two variants of the analytical method were used, Method #1 and Method #2.

The only difference between them is the use of sodium sulphate in headspace vials in Method #2. Method #1 does not use this reagent. Since both methods share most of the methodology, only relevant differences in procedure are given for Method #2.

Method #1

Materials and methods

Safety: Benzene is a known human carcinogen and, to prevent exposure, its solutions should be handled in a fume hood using appropriate precautions.

Reagents

Methanol (HPLC grade) was obtained from EM Science (Gibbstown, NJ). Water was obtained from a Millipore MilliQ (Billerica, MA) purification system. All other reagents were of analytical grade. Benzene 99.8 % (CAS #71-43-2) and d 6-benzene, isotopic purity 99.95 atom % D, were from Sigma-Aldrich (St. Louis, MO).

Apparatus

Gas Chromatograph-Mass Spectrometer (GC-MS), model Agilent 6890/5973N with 7673 injector equipped with a 100 µL syringe (Palo Alto, CA); capillary column 30 m x 0.25 mm x 0.25 µm DB-624, (Agilent); 1 mm injection sleeve (Restek, Bellefonte, PA); orbital shaker, OR100 (Cole Parmer, Vernon Hills, IL), 1 and 5 mL variable, Signature series, pipettors (VWR, West Chester, Pa)

GC-MS operating conditions

Carrier gas: helium, constant flow 1.6 mL/min, oven temperature profile: initial, 35°C (1 min), rate (1), 10°C/min. to 100°C , rate (2), 35°C/min. to 250°C, hold 5 min.; injector temp.: 100°C; inj. volume: 50 µL; needle sampling depth: 22 mm; viscosity delay: 1 s; injection mode: splitless, purge 1 min; ionization mode: 70 eV EI+; source temperature 230 ˚C; scan mode: selected ion monitoring (SIM); ions (m/z): 78, 77, 51, (benzene), 84, 82, 52, (d 6-benzene); dwell time 50 ms each ion.

Standards

All stock and spiking solutions of benzene and d 6-benzene were prepared in methanol and stored in the fridge (0 to 4˚C) for no longer than 10 weeks. Calibration solutions were prepared in water and stored in the fridge. Preparation of calibration standards: headspace vials (60 mL capacity) were filled with water at temp. <4°C within a couple mm from the brim, spiked directly with methanolic solution of d 6-benzene and benzene (65 µL of methanol), capped, mixed, and left at in the fridge overnight to give standards with benzene concentration of 0.2, 0.5, 1, 5, 25 and 50 μg/L and d 6-benzene 4.5 μg/L. For analysis, standards (1 mL) were pipetted from the headspace vials into 2 mL vials cooled to 0°C.

Commercial foods

This method ws used for beverages likely to contain benzene as a result of processing. Samples of drinks for the study were purchased from a retail food outlet in the Ottawa area. Samples were selected on a basis of having benzoic acid/benzoate listed as one of the ingredients. A few samples containing cranberry juice (which contains naturally occurring benzoic acid) were also included in a study.

Typical sample preparation

All samples are run at least in duplicate.

Samples were cooled overnight in the fridge (0 to 4˚C) and placed in crushed ice prior to sampling. Sample, (1 mL) was added via pipettor into a 2 mL pre-tarred autosampler vial, cooled to 0°C as well (in an aluminum block), spiked with a 5 µL of methanolic solution of d 6-benzene and capped. This sampling /spiking step was done within 10-15 sec. The concentration of d 6-benzene in the sample vial was approximately 4.5 μg/L. Weight of the each sample was recorded as well its density (aliquots of samples were left overnight, lightly capped, at room temp. to facilitate dissipation of carbon dioxide present in carbonated products).

Samples were equilibrated at room temperature for at least 3 hours by gentle shaking (to avoid septum contamination) on an orbital shaker at 200 rpm.

Headspace sampling was conducted at room temperature (22-24˚C).

Spiking experiments

Samples (~1 L volume), analysed previously, and containing benzene at the level of approx. 1 and 10 μg/L were spiked in their original containers with solution of benzene in methanol (~10 µL) to achieve additional benzene levels of 1 and 10 μg/L respectively. After mixing, samples were left at 4°C overnight to equilibrate before analysis.

Water, containing only native benzene at 5 μg/L and prepared as for standards, was also used to assess recovery.

Results

Under chromatographic conditions described above, retentions times of d 6-benzene and benzene were 6.43 and 6.47 (± 0.01) min., respectively.

Quality Assurance/Quality Control

Sample calculations

Benzene concentration in samples was calculated by employing d 6-benzene as internal standard utilizing an isotope dilution methodology.

Results were reported as an average of at least two determinations from the same container unless otherwise noted. Values for duplicates had to be within ± 20% to be considered valid.

Limit of detection

The injections of 0.5 ppb standards of benzene on three different days gave average signal to noise (S/N) of 12, 4.7, 4.1 in the m/z 78, 77 and 51 channel, respectively. Using criterion signal to noise (S/N) 3:1, (peak-to-peak noise definition), the limit of detection (LOD) with subsequent confirmation was thus calculated to be 0.4 μg/L taking into account the signal of the least abundant confirmation ion. Since the quantitation is achieved using more abundant ion m/z 78, the limit of quantitation (LOQ) is, in this case, equal to LOD.

Calculations of LOD based on analyses of replicates produced a similar value of LOD.

Linearity

The linearity within the range of 0.5, 1, 5, 25 and 50 μg/L was better than 0.999. The calibration curves were not forced through origin and any small positive residuals were included in calculations.

Replicate (within-run) analysis from the same sample container

Analysis of replicates (n=8), (Strawberry Daiquiri Mix) gave average concentration, standard deviation (SD), and relative standard deviation (RSD), of 0.4 μg/L , 0.05 and 13 %, respectively.

Analysis of replicates (n=8), (Margarita Mix) gave average concentration, SD, and RSD, of 13.4 μg/L, 0.83 and 6.2 %, respectively.

Analysis of replicates (n=11), (Blue Raspberry Mix) gave average concentration, SD, and RSD, of 2.6 μg/L, 0.093 and 3.6 %, respectively.

Spiking experiments

At the spike level of 1 μg/L (n=3), (Strawberry Daiquiri Mix) the recovery was 61 % with a RSD of 5.0 %.

At the spike level of 10 μg/L (n=3), (Margarita Mix) the recovery was 90 % with a RSD of 4.3 %.

The average recovery of 5 μg/L water standard (n=3) was 85 % with a RSD of 0.5 %.

Method #2

(only relevant differences from Method #1 are listed)

Reagents

Sodium sulphate (EMD Chemicals, Darmstadt, Germany) was heated to 600°C overnight prior to use and stored at 100°C in the laboratory oven.

Apparatus

Ultrasonic bath, (VWR, West Chester, Pa).

Typical sample preparation

All samples were run at least in a duplicate.

Samples were cooled overnight in the fridge (0 to 4˚C) and placed in crushed ice prior to sampling. Sample, (1 mL) was added via pipettor into a 2 mL pre-tarred autosampler vial containing 0.3 g of sodium sulphate, cooled to 0 °C as well (in an aluminum block), spiked with a 5 µL of methanolic solution of d 6-benzene and capped. This sampling /spiking step was done within 10-15 sec. The concentration of d 6-benzene in the sample vial was approximately 4.5 μg/L. Weight of the each sample was recorded as well its density (aliquots of samples were left overnight, lightly capped, at room temp. to facilitate dissipation of carbon dioxide present in carbonated products).

After vortexing for 5 sec. and sonication (temperature of the vial was kept to less than 30°C) and vortexing again for 5 sec., sample was equilibrated at room temperature for at least 30 min. Headspace sampling was conducted at room temperature (22-24˚C)

Quality Assurance/Quality Control

Limit of detection

The injection of 0.5 ppb standard of benzene gave signal to noise (S/N) of 23, 17, 13 in the m/z 78, 77 and 51 channel, respectively. Using criterion signal to noise (S/N) 3:1, (peak-to-peak noise definition), the limit of detection (LOD) with subsequent confirmation was thus calculated to be approximately 0.1 μg/L, taking into account the signal of the least abundant confirmation ion. Since the quantitation is achieved using more abundant ion m/z 78, the limit of quantitation (LOQ) is, in this case, equal LOD.

Linearity

The linearity within the range 0.2, 0.5, 1, 5, 25 and 50 µg/L was better than 0.9998. The curves were not forced through origin and small positive residuals were included in calculations.

Replicate analysis and spike recoveries

At the spike level of 5 μg/L (cranberry juice at 0.7 ppm), the recovery was 82.5% with a relative standard deviation (RSD) of 4.6%, (n=3).

The recovery of 5 μg/L benzene in water standard was 87% with a RSD of 9.9% (n=3).