Note 93: Detection of Benzene in Carbonated Beverages with Purge & Trap Thermal Desorption GC/MS


By Robert Frey, Ronald E. Shomo II and John J. Manura, Scientific Instrument Services, 1027 Old York Road, Ringoes, NJ 08551 (presented at Pittcon2007)


The recent discovery of benzene in soft drink beverages has created a firestorm of controversy within the public health, FDA, and public domain. The current FDA permissible limit for benzene in drinking water is 5 ppb.1 However; the FDA currently has no regulatory limits for benzene in beverages other than bottled water. Benzene is found in air from emissions from burning coal and oil, vapors from fossil fuels, and motor vehicle exhaust. Benzene is a known carcinogen and one of the most regulated chemicals in the world.

The recent controversy began after an independent laboratory submitted a report to the FDA in late 2005 indicating a number of commercially available soft drink beverages contained benzene in concentration levels as high as 75 ppb.1 Where was this contamination originating when the manufacturing process typically used highly filtered water to produce these beverages? One of the theories puts the blame of the elevated benzene levels on the combination of benzoate salts and ascorbic acid found in some of these drinks. The preservative sodium benzoate combined with ascorbic acid under the proper conditions (high heat and/or light exposure) has been shown to lead to the formation of benzene via the decarboxylation of benzoic acid.2

These conditions have led to another question, was the purge & trap conditions that the FDA employed contributing to the formation of benzene in these beverages? The FDA method involved the use of purge & trap GC/MS in which the samples were heated to 100°C for 30 minutes.3 Our focus in this work was to examine the effects of heat on the formation of benzene utilizing purge & trap in conjunction with a commercial thermal desorption unit coupled to a GC/MS.

Materials and Methods

Three brands of orange flavored beverages were purchased at local grocery stores for analysis. Brand X was a diet carbonated drink, Brand Y was a non-diet carbonated drink, and Brand Z was a noncarbonated non-diet drink.

Desorption tubes (4 mm stainless steel id x 4 inch) were packed with 200 mg of Tenax® TA sorbent and heat conditioned in a 24 port conditioning oven for 4 hours at 320°C with an ultra-pure Nitrogen gas purge at 25 ml/min. Tubes were removed from the heat conditioning oven then allowed to cool for 10 minutes and capped with stainless steel caps with a PTFE seal to avoid adsorption of contaminants. 300 milliliters of liquid sample was placed in a glass three neck round bottom flask connected to an Scientific Instrument Services, Inc.™ (SIS) purge and trap sampling system. Ultra-pure nitrogen gas was used to sparge the sample and a conditioned desorption tube was threaded on to the purge and trap sampling system. (Figure 1)

Figure 1 - Schematic of
SIS Purge & Trap sampling system.

The flask containing the sample was lowered into a heated water bath for a period of 20 minutes while the nitrogen purge gas sparge was adjusted to 25 ml/minute. An equivalent dry purge gas of 25 ml/minute was blown through the gas sampling system to prevent moisture from the beverage from condensing on the sorbent bed in the desorption tube.

Samples were collected at 20, 50, and 90 °C to examine the effects of temperature on the formation of benzene. Duplicate samples of each beverage were sampled at each temperature.

The desorption tube was removed from the purge and trap sampling system and connected to a SIS ADS2000 thermal desorption system. A 35 mm desorption needle was connected to one end of the desorption tube. The needle acts as the transfer line when desorption takes place insuring the highest transfer of sample to the GC and eliminates any carryover effects observed in many other desorption systems.

The desorption tube was placed in a 12 sample holding carousel. The Thermal Desorption unit (Figure 2) picked up the desorption tube from the carousel, performed a 1 minute He purge using the GC's carrier gas, and then injected the needle through the septum of the GC injection port. After the injection port pressure stabilized, all of GC carrier gas flow was diverted through the desorption tube.

Figure 2 - Schematic of
SIS AutoDesorb system.

Desorption blocks heated to 150°C then closed around the desorption tube and heated the tube for 5 minutes. During the desorption process, a cryofocusing unit (SIS CryoTrap) inside the GC oven cooled a 2" section of a guard column (DB-5MS 0.53 mm ID with a 1.5 μm film thickness) to -180°C with liquid nitrogen. This trapped the desorbed compounds during the desorption process improving the chromatography by cryofocussing the components.

Thirty seconds after the desorption process was completed, the cryogenic cooling was stopped and the cryofocusing unit heated up to 200°C for 3 minutes to elute the trapped components at the head of the capillary column.

The Thermal Desorption controller then sent a remote start signal to the GC to begin the temperature ramp from 40°C to 280°C at a rate of 10°C per minute for a total GC run time of 26 minutes. The MS transfer line was held at 280°C. The GC (Agilent 6890) oven contained a capillary column HP5MS (250 μm film thickness x 0.25 mm ID x 60m length). The GC injection port was held at 280°C and the GC was run in "split" mode set at a 5:1 ratio. Total helium flow through the desorption tube was 8.8 ml/minute for the length of the 5 minute desorption run. Helium flow through the GC capillary column was 1.0ml/minute. A Mass Spectrometer (Agilent 5973MSD) was operated in Electron Impact Mode (EI) and scanned over a mass range between 35 and 500 daltons.

All analytes were identified using Agilent ChemStation software with the NIST AMDIS software using the NIST mass spec library as well as the Wiley NBS mass spec library running in the SIM mode (Selection Ion Monitoring). Quantification was performed utilizing a d6-benzene standard using the identical purge & trap method performed on the orange beverages.

Results and Discussion

All three beverages showed the presence of benzene by this technique. Benzene levels were found to be between 0.14 pg/μl and 10.1 pg/μl. One would expect if temperature was not a factor in the production of benzene in the samples that the levels of benzene found would be the same, within experimental error, regardless of the temperature at which the samples were collected. However, the data shows a definite trend towards higher benzene concentrations at the more elevated temperatures. This can be seen most clearly for Brand Z shown in Figure 3. Brand Z had the highest levels of benzene and made this comparison easier. Also included in this figure is the blank conditioned Tenax packed desorption tube run showing no evidence of any benzene present in the adsorbent. A tabulated form of the data can be found in Table 1

Table 1
SamplePurge & Trap Temperature(°C)Benzene (pg/μl)

Brand X also showed an increase in benzene levels (Figure 4) with increasing temperatures but the chromatograms are less dramatic as with Brand Z. Brand Y showed an increase up to the 90 °C level (Figure 5), but at this temperature we had a difficult time keeping the lid on the flask as the pressure from the carbonation kept popping off the lid to the purge & trap sampling device, resulting in lost volatiles. This brand seemed to have more carbonation than the other brands.

Figure 3 - GC chromatograms of Brand Z.
Figure 4 - GC chromatograms of Brand X.
Figure 5 - GC chromatograms of Brand Y.

The sensitivity of the ADS2000 allows us to detect benzene at these low levels without needing to use SIM mode. This allows us to detect and identify the entire volatile and semi-volatile components present in these beverages. A couple of examples are illustrated in Figure 3. The major volatile component in this chromatogram was identified as Limonene. Decanal was also identified and is shown in the same chromatogram.


We have shown that there is a definite trend towards higher benzene levels found in these beverages as the temperature is increased. This is most likely a result of the previously reported decarboxylation of the benzoates in the presence of ascorbic acid to form benzene.2 This shows that one needs to be very aware in the method development stage of an analytical test, that the analysis is truly reporting the analyte of interest and not contributing to the levels of that analyte as a result of the test. There needs to be a very thorough method validation step done to insure the quality of the analytical method.

It should also be noted that since the initial independent lab report and the subsequent media blitz, many of the previously reported contaminated beverages have since changed the formulations of their products to eliminate either benzoates or ascorbic acid.

Tenax® is a registered trademark of Buchem BV.