production in one of our cartoning units. We are already seeing half-gallons coming on to the shelves of some supermarkets, and I think it is likely that we shall see an increasing trend towards larger sizes in a few years. If we just look a t what has happened in the United States in the growth of supermarket business and less frequent shopping, we can see how the larger siies have taken over that sector of the business (Fig. 7).

However, there is still a j ob to be done in improving the design of the paper cartons to make them more acceptable to the customer. Some customers still d o not understand how to open the various designs of carton, have difficulty in doing so and distrust them as a result. Their image needs to be improved.

Alongside this trend to larger packages. we have to look for cheaper forms of packaging, and it is again noticeable that in the United States, with the growth of the gallon and half-gallon container, there has been a move towards plastic packaging (Fig. 8). Pastic half-gallons are already on sale in this country, and I think we shall see an increasing use of this packaging material for ordinary pasteurized milk in future years.

1970 1980

Source: USOA

Fig. 8 USA percentage sales by packaging type.

S U M M A R Y

In conclusion, I summarize my paper by listing what I feelsome future trends are likely to be:

I continuation of doorstep sales but with some inevitable

2 further competition on the retail price; 3 an increase in process capacity before a decline; 4 a t best a levelling off of liquid milk volume; 5 increasing percentage of milk production to manufacture; 6 wider range of products available; 7 growth of the market for lower fat milks; 8 larger size containers; 9 new packaging materials and designs.

All in all, this is a rather exciting, if somewhat uncertain and changing prospect for the future.

change;

ONE-DAY SYMPOSIUM

19 October 1982 a t the Scientific Societies Lecture Theatre, Savile Row, London W 1

Subject: Flavour problems in stored milk and cream

Instrumental methods in determining flavour

D. J . MANNING AND J . C. PRICE Physical Sciences Department, National Institute for Research in Dairying, Shinfield, Reading, RG2 9AT

The chemical basis qf'jlavour is explained, and the application qf'chromatographic, and mass spectroscopic techniques for the separation and identification of coniportnds that contributr t o the flavours of dairy products is brit?flj, described.

When we talk about instrumental methods of determining flavour we are in fact referring to chemical analysis, because flavour is chemical. In milk and other dairy foods we are concerned with products of proteolysis, lipolysis, fat oxidation and bacterial metabolism, and in heat-treated foods, with the

products of Maillard reactions. I t is these products that are responsible for flavour, and their analysis assists in defining the flavour of dairy foods.

There are three areas in which instrumental analysis can be applied to flavour: product flavours, which can be either natural

Journal of the Society of Dairy Technology, Vol. 36, No. 2, April 1983 33

or manulactured; off-flavours or taints; and quality control. The first of these isanarea in which instrumentation has had the greatest impact over the last 10 t o 15 years, where a large number of compounds present in trace amounts usually need to be identified. The contribution each individual compound makes to the flavour of the product is usually determined by relating its concentration to its threshold value. The threshold value, the minimum concentration that can be detected by our senses, is important because usually it is the minor components in terms of concentration that have the greatest impact on flavour. Methanethiol, for example, is considered to be an important component of Cheddar cheese flavour and is present in cheese a t concentrations in the region of 24 ppb, but its threshold value in air is approximately 2 ppb. The second type of problem, off-flavours and taints, is usually regarded by flavour chemists as being easier to solve than the study of product flavours. The reason for this is that product flavours are usually composed of a number of components, each contribut- ing notes t o the overall flavour, whereas off-flavours and taints can usually be related to a single compound or class of compound. The compounds responsible for taints or off- flavours are usually easily recognized using techniques to be described later. As an example of the relation between chemical compounds and off-flavours. Manning (1979) has shown that when the concentration of hydrogen sulphide in the headspace exceeds 5 ng/ml, sulphide defects are observed in the cheese.

The application of instrumental methods for the quality assessment of dairy products is relatively new and usually requires the monitoring of compounds known to be either beneficial or detrimental to the flavour of the product. The quality of Cheddar cheese, for example, is adversely affected by the presence of fruity defects, and the development of these defects can be predicted by measuring the concentrations of ethanol in the cheese before maturation (Manning, 1979).

In order t o illustrate the instrumental approach to flavour, it is worth considering the work that has been carried out on the flavour of milk. From the flavour chemist’s point of view, milk presents a particular problem in that it is relatively bland, and its flavour is unlikely to be related to any particular impact compound. More likely its flavour is made up from contri- butions by many compounds. Milk has been studied mainly to determine flavour changes due to processing since heat treatment frequently results in the production of a cooked flavour. In processed milk stored for long periods, stale and oxidized flavours can also occur. Under the mild conditions of pasteurization few volatiles are produced, and the main effect on flavour is t o make the milk even more bland through loss of volatiles already present in the raw milk.

To determine the nature of the compounds responsible for the flavour of the milk, it is necessary first to isolate the volatiles. Probably the most comprehensive study of milk flavour has been made by Badings ( 1980) and he used vacuum distillation to isolate the volatiles from the milk. The resulting distillate, which was a dilute aqueous solution of volatiles, was freeze- concentrated and extracted with an organicsolvent. Theextract contained in excess of 100 compounds which had to be separated before identification. The separation of a large number of volatiles requires a technique which has become essential for flavour studies, namely, gas chromatography. With a suitable column and optimum gas chromatograph conditions the mixture of volatiles in the milk extract can beseparated into its components.

There have been considerable developments in gas chromato- graphy in recent years, particularly in relation to column design. When packed columns were in general use, a milk extract containing about 100 compounds would have yielded a chromatogram with a s few as 30 or so peaks. There were two reasons for this. First, because of the poor separating power of these columns, each peak in the chromatogram would contain two or more compounds, and, second, somecomponents would be at too low a concentration for them to be detected or absorbed onto the stationary phase of the column. The only way

of dealing with multicomponent peaks was to carry out repeat chromatographic runs using a range of stationary phases, but in many cases this only led to other multicomponent peaks of different composition. The problem of poor separation and adsorption of volatiles on the column has been largely overcome by the introduction of glass capillary columns, which have much greater separating power and minimize adsorption onto the stationary phase. An additional advantage of these columns is that, by virtue of their greater separating powers, a large range of stationary phases is not required for separating complex mixtures.

To overcome the failure to detect important flavour compounds. present a t too low a concentration to be detected by the chromatograph, it is necessary to supplement the gas chromatograph detector by the human nose. This technique, which is essential for detecting taints, involves smelling the components as they elute from the end of the gas chromato- graph column. This can be done by insertinga splittingdevice in front of the detector so that part of the effluent gas is diverted to a smelling port. By smelling the effluent gas and recording odour descriptions on the chart paper or on tape, each compound as it elutes can be assessed as a potential contributor to the flavour of the product.

To identify compounds in a mixture after they have been separated by gas chromatography, we depend very much upon the mass spectrometer. The mass spectrometer is the most sensitive instrument available to the flavour chemist for identifying flavour compounds. But contrary to common belief, obtaining a mass spectrum does not automatically provide an identification of a compound, unless it is fairly common and its spectrum is available in a library. It can. however. usually provide a molecular weight and information about the compound’s structure from the way it fragments after ionization. Nevertheless it is often useful to obtain further information about a compound’s structure from its infrared, ultraviolet or nuclear magnetic resonance spectrum.

To obtain mass spectra, compounds separated by gas chromatography must be transferred to the mass spectrometer. I f packed columns are used and the compounds are in abundance, it is possible to trap the compounds a s they elute from the column through the smelling port and to transfer them to the mass spectrometer through one of its inlet systems. However, rarely in flavour work are we fortunate enough to be able to use this technique. The most efficient way to transfer samples to the mass spectrometer is to lead the effluent gas from the gas chromatograph through an interface. The function of the interface is to reduce the pressure from atmospheric at the outlet of the column to about mm Hg in the ion source, and this is achieved by removing the bulk of the carrier gas during thq transfer. A considerable amount of work has been done to develop various interfaces such a s the frit, the membrane and, finally, the jet separator, but the introduction of narrow-bore capillary columns has eliminated the need to use an interface. Fast pumping on modern mass spectrometers is sufficient t o cope with the relatively low flow rates used with capillary columns, enabling the total effluent to be fed into the ion source. I f packed columns or wide-bore capillaries are used it is still necessary to use a molecular separator.

Using gas chromatography-mass spectrometry, Badings (1980) has shown that virtually all of the compounds found in UHT milk are present at lower concentrations in raw and pasteurized milk. He has also shown that, even in U H T milk, the individual components are present below their threshold value and their contribution to milk flavour can occur only a s a result of synergism.

The use of distillation for isolating volatiles is very time- consuming and, more important, due to thedifferent recoveries, the relative proportions of the volatiles in the distillate is unlikely to be the sameas those in the original milk. A technique that overcomes these problems is the headspace technique. This can be used either as a direct method, when samples of vapour are taken from above the product contained in a closed vessel. o r

34 Journal of the Society of Dair.y Technology. Val. 36. No. 2. April 1983

Apparatus for gas entrainment of volatiles in milk. A, milk; B. charcoal scrubber; C , charcoal trap; D, 2-litre measuring cylinder; N , flow rate 60 ml/min.

as an extended headspace when volatiles. swept from the product with nitrogen gas, are concentrated by trapping and subsequently regenerated onto the gas chromatograph column. The apparatus used for the extended headspace analysis of milk is shown in the Figure. Volatiles collected in thecharcoal t rapare eluted with carbon disulphide, which is then injected on to the gas chromatograph. This technique has been used to study the changes in volatile composition of milks that have undergone different heat treatments.

It has been mentioned that mass spectra alone d o not always provide a positive identification. Infrared spectroscopy can be particularly useful as a supplementary tool to mass spectro- metry, since it provides valuable information about the nature of functional groups in the molecule as well as providing a fingerprint for absolute identification. where a reference compound is available. Further information about the nuclear environment of functional groups can be obtained from a nuclear magnetic resonance spectrum. Unfortunately. neither of

these techniques has a sensitivity comparable with mass spectrometry. and attempts to link infrared directly to gas chromatography have not been entirely satisfactory. Under optimum conditions one can rarely obtain infrared or nuclear magnetic resonance spectra with less than a few micrograms of material whereas the mass spectrometer can operate routinely with a few nanograms. Using a special technique with the mass spectrometer tuned to specific ions the sensitivity can be increased to lo-'* g or less.

Until recently. gas chromatography. a s a separating tech- nique, has dominated the instrumental approach to flavour. This is understandable if one accepts that aroma is the most important part of flavour and is usually associated with compounds with sufficient volatility to be amenable to gas chromatography. However, it is becoming increasingly appa- rent that less volatile materials have an important role in flavour, in addition to providing the basic taste sensations. I t has been suggested that involatile components can affect flavour by their interactions with volatile molecules. McGugan. Emmons & Larmond (1979) have recently concluded that, although the quality of the flavour of Cheddar cheese is undoubtedly associated with quite volatile compounds, i t is the involatile fraction. through its interaction with volatile compounds, that largely determines the flavour intensity.

I t is this new interest in involatile components of flavour that has led to an increasing interest in high-performance liquid chromatography (HPLC) . Apart from its obvious use in separating involatile compounds, such as amino acids, sugars and triglycerides. it can also be effective in separating heat-labile compounds, which cannot be satisfactorily dealt with by gas chromatography. I t is unfortunate that, unlike gas chromato- graphy, there is n o universal detector for HPLC. Because of the variety of detectors for use with HPLC, such as ultraviolet absorption, fluorescence and refractive index detection. it is impossible to specify a sensitivity for this technique. The ultimate universal detector for H P L C would be the mass spectrometer, in the same way that it is used as a gas chromatographic detector. However, although a moving belt is available a s an interface between HPLC and the mass spectrometer, it is very costly and appears to require further development before it can be regarded as a routine technique.

K E F E R E N C E S

Radings. H . T. (1980) N v t / i d ~ r i d \ M i / k Duirj. Jourria/. 34. 9. Manning. D. J . ( 1979) Jourrial ((/' Dair j , Re.wardi. 46. 523. McGugan, W. A.. Emmons, D. B. & Larmond, Elizabeth. (1979)

Journal of Dairy Science. 62,398.

Journal of the Society of Dairy Technology, Vol. 36, No. 2. April 1983 35