Note 14: Identification of Volatiles and Semi-Volatiles In Carbonated Colas

by Santford Overton and John J. Manura


A large number of volatile and semi-volatile organics are present in commercial beverages which are responsible for each of their unique flavors. With increased shelf life or exposure to environmental conditions the relative amounts of these organics can vary due to evaporation or chemical changes. Off-odors and unusual taste development occur in some carbonated colas with increased shelf life, or stored at temperatures higher than normal. Problems associated with off-odor/off-taste development are thought to be related to the manufacturer's formulation, interaction of the volatile components with the different types of container lining, or foreign material introduction. Previous studies showed the greatest odor problems were in bottles with ethylene vinyl acetate (EVA) closure liners containing titanium dioxide (1). These off-odors were probably due to the presence of terpene degradation products. Light intensity and temperature are also thought to influence off-odor development. Increased shelf life in varying light and temperature conditions may stimulate oxidative reactions. Wiley et al. (1) suggest that a turpentine-like off-odor development IS due to excess p-cymene in the cola resulting from increased shelf life at higher temperatures. The primary sources of p-cymene appear to be from the conversion of terpinene and limonene, natural constituents of lemon oils used in cola formulations. The purpose of this study is to analyze several brands of carbonated colas in aluminum cans to compare the flavor profiles of different manufactures' brands and to quantify cymene and limonene within these products. Quantification of cymene and limonene is conducted to determine their relationship of off-odor/off-taste development with increased shelf life.

Figure 1

Figure 1 - Purge & Trap System


A Purge & Trap System (Scientific Instrument Services, Ringoes, NJ) was used for the purging of the volatile and semi-volatile organics from the carbonated colas and subsequent trapping of the purged organic compounds on preconditioned glass-lined stainless steel desorption tubes (GLT) containing the porous polymer Tenax® TA. The Purge and Trap System (Figure 1) consists of a sparge gas inlet connected to a stainless steel purging needle that is inserted through an adaptor fitting into a 10 ml test tube. This purging needle provides for the helium gas flow through the cola sample to purge the organics from the liquid sample and carry them onto the adsorbent resin. The dry purge gas inlet is located at a right angle to the sparge gas inlet at the top of the apparatus. The purpose of the dry purge is to reduce the water vapor condensation on the adsorbent trap. This problem can be especially troublesome when isolating volatiles from aqueous solutions at high temperatures. Although the adsorbent traps packed with Tenax have a low affinity for water, it is inevitable that some condensation can occur in the trap due to the high relative humidity of the sparge gas as it exits the apparatus. When moisture condenses on the adsorbent, it can block the pores of the resin matrix and thereby drastically reduce the diffusion of volatile organics into the trapping resins. This will result in reduced trapping efficiency. The use of the dry purge will eliminate this problem of water condensation on the Tenax resin. Opposite the dry purge inlet is the connector for the GLT desorption tube containing the adsorbent resin. The sample collection vessel for this analysis was a 10.0 ml purge and trap glass test tube. A heater blanket was placed around this glass tube to heat the tube and contents to 50 degrees C using a regulated temperature controller. This heating of the sample served to reduce the viscosity of the cola to permit better purging of the liquid sample as well as higher molecular weight compounds from the liquid samples for subsequent trapping on the adsorbent trap.

All experiments were conducted using a Scientific Instrument Services Model TD-2 Short Path Thermal Desorption System accessory (2) connected to the injection port of an HP 5890 Series II GC interfaced to an HP 5971 Mass Selective Detector. The mass spectrometer was operated in the electron impact mode (EI) at 70 eV and scanned from 35 to 350 daltons during the GC run for the total ion chromatogram. The mass spectrometer scanned from 90 to 140 for the quantitative analysis of cymene and limonene.

A short 0.5 meter by 0.53 mm diameter fused silica precolumn was attached to the injection port end of a 30 meter x .25 mm i.d. DB-5MS capillary column containing a 0.50 µm film thickness. The GC injection port was set to 250 degrees C, and a 10:1 split was used during the injection step. The GC oven was maintained at -40 degrees C during the desorption process and then temperature programmed to 300 degrees C at a rate of 10 degrees per minute for the total ion chromatogram. For the quantitative analysis of cymene and limonene, the GC oven was maintained at 30 degrees C during the desorption and extraction process and then temperature programmed to 70 degrees C at 10 degrees per minute, to 100 degrees C at 1 degrees per minute, and to 300 degrees C at 30 degrees per minute.


Several brands of carbonated colas were analyzed to compare the flavor profiles of different manufacturers' brands and to quantify cymene and limonene to determine their relationship to off-odor/off-taste development. Cymene and limonene were also examined in two brands of colas, a 2 year old cherry flavored cola and a 25 year old cola, to determine any relationship to off-odor/off-taste development. Another brand of cola was sampled and analyzed at 4 days, 1 week and 3 weeks from an opened aluminum can to determine any changes in cymene and limonene concentrations and cymene/limonene ratios.

Sample sizes of 1 to 5 ml of several carbonated colas with varying shelf lives were pipetted into a 10 ml test tube and heated to 50 degrees C. Samples were sparged with high purity helium at 15 to 20 ml/min and a dry purge of 20 ml/min for 10 minutes using a Scientific Instrument Services Liquid Purge System (Figure 1). Volatile analytes were gas extracted and carried to a preconditioned 4.0 mm i.d. glass-lined stainless steel desorption tube packed with 100 mg of Tenax TA. Once the samples were collected, they were spiked with 400 ng/µl of d-14 cymene internal standard by injecting 1 µl of a 400 ng/µl of a d-14 cymene stock solution in methanol by syringe injection into the Tenax matrix.

The desorption tubes with sample and internal standard were than attached to the Short Path Thermal Desorption System and a syringe needle attached. The sample was injected into the GC injection port and thermally desorbed in the GC injection port at desorption block temperatures of 200 degrees C for 10 minutes at a purge flow of 10 ml/min, and a GC injection split ratio of 10:1.


The quantification of cymene and limonene in carbonated colas was accomplished using d-14 cymene as a matrix-spiked surrogate internal standard with GC-MS detection. To prepare the calibration curve data, analytical standards of d-14 cymene (MSD Isotopes) were prepared from the standard stock solutions of the d-14 cymene by diluting this solution with methanol using volumetric glassware to prepare 250 ml of a stock solution of d-14 cymene at a concentration of 400 ng. A 23.3 µl sample of cymene (Aldrich Chemical Company, Inc.) and a 23.8 µl sample of limonene (Aldrich Chemical Company, Inc.) were injected into a 10 ml volumetric flask. The volume of the flask was adjusted to exactly 10 ml using the 400 ng/µl d-14 cymene stock solution. Starting from this solution mixture (2000ng/µl cymene and limonene and 400ng/µl of d-14 cymene), a series of log and half log dilutions were prepared down to a final concentration of 3.9 ng/µl of cymene and limonene, using the d-14 cymene stock solution as the diluent.

 Therefore, the following stock solutions were obtained in ng/ul:

Stock Solution       Limonene/Cymene      D-14 Cymene                              

                      Concentrate         Concentrate

-----------------    ---------------     ------------

Spiking Solution           0                 400

       A                2000                 400

       B                1000                 400

       C                 500                 400

       D                 250                 400

       E                 125                 400

       F                62.5                 400

       G                31.2                 400

       H                15.6                 400

       I                 7.8                 400

       J                3.91                 400

One µl each of these solutions were then injected onto the top of a Tenax TA desorption trap, purged with high purity helium for 2 minutes to purge the MeOH from the desorption tube and then injected into the GC injection port at 10.0 ml/min with a 10:1 split, thermally desorbed into the GC and analyzed as previously described. Mass chromatograms for the molecular ion species of d-14 cymene (m/z 130), cymene (m/z 119) and limonene (m/z 93) were then generated and the resulting data integrated via the computer.


The data generated from the following concentrations of cymene and limonene in 400 ng/µl d-14 cymene were used in creating the calibration curves: 2000 ng/µl, 1000ng/µl, 500 ng/µl, 250 ng/µl, 125ng/µl and 31.25 ng/µl. The peak area integrations were used to generate the calibration curves for cymene and limonene relative to d-14 cymene internal standard. Curves were constructed by plotting the ratio of cymene or limonene peak area to the area of d-14 cymene area (y-axis) versus the concentration of cymene or limonene (x-axis) in nanograms. A linear calibration curve was obtained for cymene with a correlation coefficient of 0.999 (Figure 2). The linear calibration curve for limonene exhibited a correlation coefficient of 0.998 (Figure 2). To quantify the cymene and limonene levels in carbonated colas, all samples were spiked with 400 ng/µl of d-14 cymene, as previously described. The ratio of cymene or limonene to d-14 cymene internal standard peak areas were then determined, and the cymene and limonene concentrations were then calculated from the calibration curves.

Figure 2a

Figure 2b

Figure 2 - Calibration Curves for the Quantification of Cymene and Limonene.


Monterpenes such as myrcene, cymene, limonene, terpinene and terpinolene were identified in each of the colas examined (Figure 3) with the exception of a 25 year old cola which appeared to undergo oxidative degradation of the terpenes over time, resulting in the production of acetic acid and 2-ethyl hexanol ( Figure 4). Additional compounds which were found in the colas included the alcohols endo-fenchol and endo-borneol and 1,8-cineole, a chief constituent of oil of eucalyptus ( Figures 3-5). The flavors furfural and benzaldehyde were also detected in the cherry flavored colas (Figure 5). Even though each cola had its own distinct chromatograph, they were found to contain many common compounds.

Figure 3a

Figure 3b

Figure 3b

Figure 3 - Colas , 1 ml. Collected for 10 min at 15 ml/min With 15 ml/min Dry Purge and Thermally Desorbed At 200 Degrees C For 10 min

Figure 4

Figure 4 - Twenty-Five-Year-Old Cola, 1 ml. Collected For 10 min at 15 ml/min With 15 ml/min Dry Purge and Thermally Desorbed At 200 Degrees C For 10 min

Figure 5

Figure 5 - Cherry Cola, 1 ml. Collected For 10 min at 15 ml/min With 15 ml/min Dry Purge and Thermally Desorbed At 200 Degrees C For 10 min

Increased shelf life had a significant effect on both the 2-year-old cherry flavored cola and the 25-year-old cola that were tested. Limonene concentrations significantly decreased in both colas from those of the control, while cymene concentrations increased with accompanying increases in cymene/limonene ratios (Table I). The more drastic effect occurred in the 25 year old cola where cymene was not detected and only a trace of limonene was present. This was probably due to increased oxidative reactions over time. This trend of decreased limonene concentrations and increased cymene/limonene ratios over time also occurred in the colas which were sampled at 4 days, 1 week and 3 weeks from an opened aluminum can.

TABLE I - Comparison of Cymene and Limonene in Colas With different Shelf Lives

                                      Cymene(ug/ml) Limonene(ug/ml) Cym/Lim

-------------------------------------- ------------ -------------   -------

Cherry Cola (1992 Control) l                 101.9       1957.2      0.05

Cherry Cola (1990) 2 yr. old                 195.5        104.9      1.86

Cola (1992 Control)                          105.6       8930.9      0.01

Cola (1967)                              Not Detected      37.8        -

Cherry Cola - freshly opened (Control)       101.9       1957.2       .05

Cherry Cola - opened (4 days)                 66.1        585.7       .11

Cherry Cola - opened (1 week)                 22.2         58.8       .38

Cherry Cola - opened (3 weeks)                14.2         15.2       .93


The Short Path Thermal Desorption System used in conjunction with the Liquid Purge System permits the identification and accurate quantification of trace levels of cymene and limonene and other volatiles and semi-volatiles in carbonated colas. Results indicate that limonene levels in the various colas decreased with respect to increased shelf life. It appears that the reduction in the limonene levels and not so much any changes in the cymene levels play a major role in the increased cymene/limonene ratios and in off-odor and unusual taste development over time. In the colas with a longer shelf life, as the 25-year-old cola that was examined, oxidative reactions occurred resulting in the degradation of the terpenoid compounds and in the formation of by-products like acetic acid and substituted alcohols. This technique has also been applied to other applications such as: quantification of benzene and toluene in food products (3), and flavors and fragrances in food products (4&5), commercial products (6) and plant material (7).


Hartman, T.G., Karmas, K., Chen, J., Shevade, A., Deagro, M., and H. Hwang. 1992. Determination of

    Vanillin, Other Phenolic Compounds, and Flavors in Vanilla Beans. ACS Symposium Series 506. Phenolic

    Compounds in Food and Their Effects on Health I. Chi-Tang Ho, Chang Y. Lee, and Mon-Tuan Huang,

    Editors. pp.60-76.

Manura, J.J., Overton, S.V., Baker, C.W. and Manos, J.N. 1990. Short Path Thermal Desorption-Design and

    Theory. The Mass spec Source. Vol. XIII (4): 22-28.

Manura, J.J. 1993. Quantitation of BHT in Food and Food Packaging by Short Path Thermal Desorption.

    LCGC Vol. 11 (2): 140-146.

Methodologies for the Quantification of Purge and Trap Thermal Desorption and Direct Thermal Desorption

    Analyses. S.I.S. Application Note No. 9, September 1991.

Patt, J.M., Hartman, T.G., Creekmore, R.W., Elliott, J.J., Schal, C., Lech, J., and R.T. Rosen. 1992. The

    Floral Odour of Peltandra Virginica contains Novel Trimethyl-2,5-Dioxabicyclo [3.2.1.] Nonanes.

    Phytochemistry. Vol. 31 (2): 487-491.

Quantification of Naphthalene in a Contaminated Pharmaceutical Product by Short Path Thermal Desorption.

    S.I.S. Application Note No. 10, May 1992.

Wiley, R.C., M.K. Louie and M.J. Sheu. 1984. Studies on Turpentine-Like Off-Odor in Cola. Journal of

    Food Science. Vol. 49: 485-488, 497.

Tenax® is a registered trademark of Buchem BV.