Citrus Station Mimeo Series 61-6
September 21, 1960
Flavor Research Investigations
I. The Recovery and Analytical Separation of Volatile Flavor Components
of Valencia Orange Gas Chromatography
Richard W. Wolford and John A. Attaway
The following papers are presented as a continuation of the flavor research
investigations reported at the 1959 Citrus Processorst Meeting. During the year
work on this project has been intensified. We have been fortunate in having
additional personnel assigned to the project, thus permitting an extension of our
basic research efforts. Also, the expansion of our gas chromatographic instru-
mentation and its application to the problem has contributed significantly to the
results herein presented.
The study of the natural flavor of citrus juices, citrus oils, and processed
citrus products requires the use of the same general concepts as in any study of
natural flavors and aromas. Also, it might be said that the same formidable
difficulties are encountered.
Certain characteristics of the chemical composition of natural flavors may
serve as a guide to outlining the course of such an investigation. Generally
speaking the natural flavors seem to share the following characteristics: (1)
they normally consist of a great many components, some of which are present in
very large proportion, while the majority of the components are present in very
small concentrations; (2) the flavor components are present as minor constituents
compared to the total of all classes of compounds in the fruits or vegetables,
however, in their low concentration they produce tremendous flavor effects; (3)
the flavor characteristics are highly specific to the molecular configuration of
the molecules involved; and (4) the flavor components tend to be labile, thermally
unstable compounds, which require the most sensitive and specific of chemical and
physical techniques for their recovery and separation.
In the natural citrus flavors we have an extremely large number of flavor
components, covering the entire chemical spectrum from mixtures that are largely
hydrocarbons to mixtures of highly polar and reactive substances.
The basic problems in these investigations involve the concentration of the
small quantities of material recovered from large quantities of juice, the
fractionation of the recovered concentrated material, and the identification of
the separated components.
To date the recovery of the volatile components from gross quantities of
citrus juice has been carried out in a commercial essence recovery systemI. We
have been fortunate in being provided with relatively large quantities of this
material for our studies. The availability of this material has permitted our
placing primary emphasis on the development and application of methods for the
satisfactory separation of the volatile components and subsequently their identi-
1 Samples of orange essence supplied through cooperation of E. J. Kelly and
Associates, Inc., and Libby, McNeill and Libby.
Florida Citrus Experiment Station and Florida Citrus Commission,
Lake Alfred, Florida. 992 9/21/60 RWW
The analytical results and identifications of individual components in
these investigations have been carried out principally on Valencia orange
essence since we had some assurance of having a single variety of oranges during
that portion of the concentrate processing season.
The isolation of the volatile components from the essence has been studied
using a number of procedures, but not all of the methods produced a material
satisfactory for gas chromatography. The main requirement of gas chromatography
is that the material be volatilizable at the temperature of operation; and it is
preferred that the moisture content be at a minimum. The following method has
met the requirements of this sensitive procedure.
A quantity of aqueous essence (2 liters) is first saturated with anhydrous
sodium sulfate. The salt saturated solution is then extracted with 500 ml of
diethyl ether with vigorous agitation in a beaker. The two phases are allowed
to separate and the solvent phase is removed. The water phase is extracted
again with a like amount of ether and the solvent phase is removed. Both of
the ether extracts are combined and evaporated on a steam bath in an Erlenmeyer
flask. The evaporation is carefully watched to avoid a temperature rise above
400C. At that point isopentane (approximately 50 ml) is added and the water
which is dissolved in the ether breaks out. Separation of the water from the
solvent extract is accomplished in a separatory funnel. The evaporation is con-
tinued until most of the solvents are removed at which time a light yellow-
colored oil is obtained. If a temperature rise is incurred at any time, addi-
tional amounts of isopentane are added until the water-free oil is obtained.
From a 2 liter quantity of the aqueous essence, we normally obtain about 5 to
8 ml of the oil constituents.
The gas chromatographic analyses have been carried out on two instruments.
The Perkin-Elmer 154-C Vapor Fractometer used in our beginning work and the F & M
Scientific Linear Temperature Programmed Gas Chromatograph, which has been in use
for approximately five months. The liquid stationary phases employed in the
Vapor Fractometer, an isothermal instrument, were Ucon Polar #9 and Carbowax 1540.
Some studies were also made with a ten foot Craig polyester succinate column.
Normally, a two meter column was used. The liquid phases were fixed on either
0-22 Firebrick or Chromosorb regular. The best conditions of operation with the
Vapor Fractometer were obtained using a two meter Carbowax 1540 column at a
temperature of 1500C and a carrier gas (helium) flow rate of 40 to 50 ml/min.
Much of the information to be presented would not have been possible to
obtain in the relatively short period of time if it had not been for the newer
approach, that of temperature programming the gas chromatographic column. Com-
parisons between runs made under both isothermal conditions and temperature
programming show tremendous improvements in resolution and separating power in
the latter system. For example, an an-.lysis cx the *extrar..ed essence oil under
isothermal conditions normally shows the resolution of soxe 18 to 23 component
Florida Citrus Experiment Station
and Florida Citrus Commission,
Lake Alfred, Florida.
992 a 9/21/60 RWW
peaks. When the same material is separated in the temperature programmed
system, the number of component peaks including shoulders will number from
35 to 40.
One of the principal limitations in the use of temperature programming is
in the availability of column stationary liquids which are applicable to the
temperature range encountered. The normal run in these investigations starts
with a temperature of 5000 on the column. The temperature is then increased
at a fixed predetermined rate and stopped at a pre-set limit, usually about
2450C. Thus, one must use a liquid stationary phase with a minimum vapor
pressure under the conditions of operations and permit a sufficiently high
temperature without decomposition. It had been determined previously in these
studies that a polar stationary phase is more desirable in performing the
separation of the volatile components. Our best separations have been obtained
on a six foot Carbowax 20M column. Other parameters employed in the programmed
gas chromatographic separations were as follows: the detector block at a
temperature of 25500C under closely controlled conditions, the injection port
maintained at about 1850C, and the pressure to the system maintained at 25
p.s.i. with the helium (carrier gas) flow rate adjusted to 50 ml/min. The de-
tection system used is of the thermistor type for thermal conductivity measure-
ment, which is provided with a split stream arrangement in that the reference
thermistor is in an atmosphere of helium but receives a flow only one fifth of
the amount passing over the sensing thermistor. There is a short path from the
column exit to the sensing detector which is geometrically located to provide
for the full flow of carrier gas plus the eluted sample. Normally, a bridge
current of 10 milliamperes is used. The sensitivity of the detectors may be
increased somewhat by increasing the bridge power but the life of the thermistors
would be reduced at the higher current settings. From 8 to 10 chromatographic
separations may be made in 8 hours using this system. This compares with half
that amount on the isothermal instrument. The F & M instrument has versatility
in that it may also be used as an isothermal instrument while the temperature
may be raised at any point in the scan or may be run as a combined isothermal
and temperature programmed system if desired.
The new approaches to isolation, separation, and identification of natural
flavor components can add much to the development of a better understanding of
the scientific aspects of flavor. The application of these methods have some
advantages but also certain limitations. Regardless of the availability of
improved instrumentation, the problem deals with basic chemistry and physics
and the real success obtained comes from a broad but cautious approach to the
Florida Citrus Experiment Station
and Florida Citrus Commission,
Lake Alfred, Florida.
992 b 9/21/60 RWW