Professional Documents
Culture Documents
PIIS0022030292779685
PIIS0022030292779685
PIIS0022030292779685
TABLE 2. Three-minute, double injection sampling se- duce the scorched off-flavor. The milks were
quence stored in the memory of the automatic headspace
sampler.
placed in foil-covered, 946-m1 glass bottles
and subjected to selected heat treatments. The
Action Time sulfurous and rich milks were heated on a
(min:s) steam bath at 75’C for 2 min at 97’C for 15
Insert sampling needle 0:Ol min, respectively. The caramelized milks were
Pressure on 0:02 autoclaved in a pressure cooker at 95 kPa (14
Pressure off 022 psi) for 15 min. At the end of heat treatment,
Vent on 0:38 samples were cooled by placing them in an ice
Vent off 0:40
Begin injection 0:42
water bath.
End injection 1 :20 The milks were analyzed by the GC method
Pressure on 1 :22 described with the results reported as peak
Pressure off 1 :42 areas.
Vent on 158
Vent off 2:oo
Begin injection, second vial 2:02 Sensory and Instrumental Analyses
End injection, second vial 255
Withdraw sampling needle 257 To study the relationship between the ana-
lytical results and sensory scores, skim and
whole milks were brought to 75’C and held at
door was closed, and the run proceeded auto- that temperature for different times. A sensory
matically from that point. panel rated the milks according to the amount
In order to have reproducible injections, the of cooked flavor that had developed. The head-
timing of GC and HSS events was controlled space method was used to follow changes in
electronically; printed queues came from the the levels of the sulfur volatiles. The sample
integrator to tell the operator when to remove procurement, heat treatment, sensory study,
the cryotrap and close the oven door. The GC and analytical measurements were completed
events were controlled with a keystroke pro- in 4 d.
gram stored in the memory of the GC (Table On the 1st d of the study, 18.9 L (5 gal) of
1). This program started the GC run, advanced skim and 18.9 L of whole pasteurized milk
the sample carousel of the HSS, and initiated were procured, and 15.1 of the 18.9 L were
the HSS injection sequence. The HSS injection loaded into 18.9-L “shotgun pails” (round
sequence was another program stored within stainless steel pails having inside radii of about
the memory of the HSS, which controlled the 24 cm) and heated with constant stirring in a
timing of the HSS (Table 2). 97°C water bath until the temperature of the
milk had reached 75’C. At this point, the pails
Fat Interference
were transferred to another bath, which was
held at 75 to 80’C. After intervals of 1, 5, 10,
To study the effect of the presence of fat in 20, and 30 min, milk was transferred from the
this particular system, DMS and dibutylsulfide stainless steel pails into foil-covered .98-L (1-
were dissolved in distilled water and in whole qt) jars and placed on ice. Once cool, they
milk at a concentration of 100 ng/g. The solu- were stored in the dark at 3 to 5’C until
tions were analyzed under the conditions just sensory testing.
described. Sensory testing took place 16 h after heat
treatment. The sensory panel consisted of 21
ADSA Heat Treatments volunteers, all of whom were students or fac-
ulty at the University of Minnesota Depart-
Pasteurized skim and whole milks, which ment of Food Science and Nutrition. The
were purchased locally, were subjected to the judges’ experience varied considerably, so they
first three of the four heat treatments were first presented with a training set consist-
prescribed by the ADSA Committee on ing of four samples: severely cooked (30 min)
Nomenclature and Standards of Milk Off- skim milk, severely cooked whole milk, and
Flavors (16) to produce sulfurous, rich, and freshly pasteurized s k i m and whole milks.
caramelized milks. We did not attempt to pro- They were asked to taste the reference samples
Journal of Dairy Science Vol. 75, No. 8, 1992
HEADSPACE ON HEATED MILK 2101
TABLE 3. Two-factor ANOVA of fat (skim or whole
31 milk) and cook (sulfurous, rich, or caramelized) on H2S
peak area.
Source df SS MS F Test P Value
Fat 1 18,449 18,449 2.508 .1644
Cook 2 134,620 67,310 9.15 .0151
Fat x
cook 2 67,336 33,668 4.577 .0621
Error 6 44,139 7351
i3
Y
I 1120
m
a
Q
a,
e
=3
m
c
Q
0
e
R
r
0 10 20 30
Sulfurous Heated Carmelized
Heating time at 75°C (rnin)
Cook Level
Figure 6. Comparison of the cooked flavor score (A)
Figure 5. Effect of different heat treatments on the to the hydrogen sulfide peak area (m) for skim milk. Error
relative amount of H2S detected in milk.rep. = Replicate. bars correspond to 1 SE.
Whatever the cause, the implications for were moderate, and the correlations for whole
instrumental measures of cooked flavor are milk were weak.
clear: the level of sulfur volatiles is a useful The Spearman procedure indicates that a
predictor of cooked flavor only for mild heat relationship exists between analytical and sen-
treatments. It is possible that a more useful sory results for skim milk, but it does not
relationship might have been found if a non- reveal what the relationship is. Figures 6 and 7
selective detector had been used because the reveal that the flavor scores increase on heat
caramelized flavor in milk is thought to come treatment and then remain fairly constant. The
from the products of browning reactions, same pattern occurs for the H2S areas, suggest-
which may contain little sulfur (16). ing that H2S could be used for predicting
sensory scores. In the case of DMS, there is no
such relationship between sensory scores and
Statistical Analyses
peak areas. There appears to be more DMS in
Spearman's rank correlation coefficients the untreated pasteurized milk than in any
were generated to see whether there was a
statistically significant correlation between
sensory scores and adjusted peak areas (Table
5 ) . The Spearman procedure was chosen be-
cause it is robust against nonlinear relation-
ships and outliers (7). The prescribed ranges
for weak, moderate, and strong correlations are
0 to f.5,f . 5 to .8, f.8 to 1, respectively. By
these criteria, the correlations for skim milk
other sample. Except for an increase at approx- 5 Buttery, R. G., D. G. Guadagni, and L. C. Ling. 1973.
imately 5 min, which may or may not be Flavor compounds: volatilities in vegetable oil and oil-
significant, the overall trend is of decreasing water mixtures. Estimation of odor thresholds. J.
Agric. Food Chem. 21:198.
amounts of DMS on further heating. 6Buttery, R. G.,and R. Teranishi. 1963. Food vapor
analysis. Measurement of fat autooxidation and
CONCLUSIONS browning aldehydes in food vapors by direct vapor
injection gas liquid chromatography. J. Agric. Food
In conclusion, the headspace method used Chem 11:504.
in this research, which requires no sample 7 Devore, J., and R. Peck. 1986. Statistics; the Explora-
tion and Analysis of Data. West Publ. Co., St. Paul,
preparation, had adequate sensitivity for this h4N.
study, evidenced by the ability to detect H 2 S 8 Ferretti, A. 1973. Inhibition of cooked flavor in heated
and DMS in pasteurized milks that did not milk by the use of additives. J. Agric. Food Chem. 21:
have a noticeable cooked flavor. Also, there 939.
gHachenburg, H., and A. P. Schmidt. 1977. Gas
was a moderately strong correlation between Chromatographic Headspace Analysis. John Wiley &
the level of H 2 S and the intensity of the Sons, New York, NY.
cooked flavor in skim milk. Sulfur volatiles lOHaytham, A., J.A.P. Jaddou, and D. I. Manning. 1978.
were maximal after moderate heat treatments Chemical analysis of flavor volatiles in heat-treated
and then decreased or remained the same on milks. J. Dairy Res. 45:391.
11 Jennings, W. G., S. Viljhalmsson, and W. L. Dunkley.
more severe heat treatment. This might limit 1962. Direct gas chromatography of milk vapors. J.
their usefulness in a quality control setting Food Sci. 27:306.
because a mild heated flavor is not a severe 12 Kolb, B., D. Boege, and L. S. Ettre. 1988. Advances
defect. in headspace gas chromatography: instrumentation
and applications. Am. Lab. 20:33.
There was no correlation between GC and 13Mabbitt, L. A., and G. McKinnon. 1963. The detec-
sensory data for milk that contains fat. The tion of volatile components of milk by gas-liquid
milk fat could have been acting as a competing chromatography and its possible application in assess-
solvent for the flavor volatiles, lowering their ing keeping quality and flavour. J. Dairy Res. 30:359.
vapor pressure below the detectable limits of 14Patton. S., D. A. Forss, and E. A. Day. 1956. Methyl
sulfide and the flavor of milk. J. Dairy Sci. 39:1469.
the analytical system. 15Reddy. M. C., R. Bassette, G. Ward, and J. R. Dun-
ham. 1967. Relationship of methyl sulfide and flavor
REFERENCES score of milk. J. Dairy Sci. 50:147.
16Shipe. W. F., R. Bassette, D. D. Deane, W. L. Dunk-
1 Aston, J. W., and K. Douglas. 1981. Detection of ley, E. G. Hammond, W. J. Harper, D. H. Kleyn, M.
carbonyl sulfide in Cheddar cheese headspace. J. E. Morgan, J. H. Nelson, and R. A. Scanlan. 1978.
Dairy Res. 48:473. Off-flavors of milk: nomenclature, standards, and bib-
2Badings. H. T., and C. de Jong. 1984. Headspace liography. J. Dairy Sci. 61:855.
analysis for the study of aroma compounds in milk 17 Sundararajan, N. R., J. Tobias, and R. Whitney. 1967.
and dairy products. Page 401 in Analysis of Volatiles. Quantitative procedure for gas chromatographic
P. Schreier, ed. Walter de Gruyter & Co., New York, analysis of headspace vapor over sterile concentrated
NY. milk. J. Dairy Sci. 51:1169.
3 Badings, H. T., R. Neeter, and J.J.G. van der Pol. 18 Wylie. P. L. 1986. Headspace analysis with cryogenic
1978. Reduction of heated flavour in heated milk and focusing: a procedure for increasing the sensitivity of
milk products by L-cystine. Lebensm. Wiss. Technol. automated capillary headspace analysis. Chro-
11:237. matographia 21:251.
4Bodyfelt. F. W., J. Tobias, and G. M. Trout. 1988. 19 Wylie. P. L. 1988. Comparison of headspace with
The Sensory Analysis of Dairy Products. Van Nos- purge and trap techniques for the analysis of volatile
trand Reinhold, New York, NY. priority pollutants. J. Am. Waterworks Assoc. 80:65.