Available online at www.sciencedirect.

com

Food Control 19 (2008) 873878

www.elsevier.com/locate/foodcont

Rapid gas-chromatographic method for the determination of diacetyl

in milk, fermented milk and butter
V. Macciola a, G. Candela b, A. De Leonardis
a

a,*

Dipartimento di Scienze e Tecnologie Agro-Alimentari, Ambientali e Microbiologiche, Universita` degli Studi del Molise, Via De Sanctis,
86100 Campobasso, Italy
b
Mediterranea Biotecnologie s.r.l. Contrada Rivolta del Re Zona Industriale A, 86039 Termoli (CB), Italy
Received 19 June 2006; received in revised form 18 August 2007; accepted 24 August 2007

Abstract
In this work, a simple and fast method for the determination of diacetyl by gas-chromatographic technique coupled with ame ionisation detector (GLC-FID) was developed. Diacetyl is the typical butter avour, but it is also commonly present in others fermented
dairy products. Recently, diacetyl determination has also attracted interest because it is one of the parameters on which lactic acid
bacteria (L.A.B.) are characterized and valued. Only acetone and 2,3-pentanedione were used as chemicals. After centrifugation of
acetonemilk mixture, supernatant was ltered and directly injected into gas-chromatographic apparatus, without a further purication
procedure step.
This method was accurate and precise; diacetyl recovery on milk was 97% and the detection limit was 1 mg L1. Finally, by using this
method, diacetyl was easily determined in fresh and high-temperature treated milk, commercial butter, yoghurt and also in a series of
L.A.B. performance tests.
2007 Elsevier Ltd. All rights reserved.
Keywords: Diacetyl; Milk avour; Butter avour; Gas-chromatography

1. Introduction
In all the world, the production and consumption of fermented milk products are increasing (Baron, Roy, & Vuillemard, 2000). Simultaneously, the commercial production
and use of lactic acid bacteria (L.A.B.), in the dairy-industry, are also increasing. L.A.B.s quality are principally
evaluated on their ability to metabolise with some milk
components, during specic biochemical processes such
as glycolysis, proteolysis, lipolysis and diacetyl production.
Diacetyl or 2,3 butanedione is the typical butter avour/
aroma, but it is also commonly found in other fermented
dairy-products such as sour cream, yoghurt and others.
Recently, mutant strains have been constructed by genetic

Corresponding author. Tel.: +39 0874 404641; fax: +39 0874 404652.
E-mail address: antomac@unimol.it (A. De Leonardis).

0956-7135/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodcont.2007.08.014

engineering to increase the production of diacetyl in cottage cheese or other soft cheese (Law, 2001).
The precursor of diacetyl is citric acid; cows milk contains approximately 1750 mg citrate per litre (Fox, Law,
McSweeney, & Wallace, 1993). At pH 5.05.2 lactic acid
displaces citric acid from its salts; free citric acid in the
presence of lactose is converted into CO2, diacetyl, acetoin
and 2-acetolactate by citrate-utilising (Cit+) strains of Lactococcus lactis subsp. Lactis (Fox, Lucey, & Cogan, 1990;
McSweeney & Fox, 2004).
The biochemical pathway is: citrate ! oxalacetate !
pyruvate ! acetolactate ! diacetyl/acetoin/2,3-butandiol
(Palles, Beresford, Condon, & Cogan, 1998).
Recently, diacetyl has also attracted interest since it is
one of the parameters on which L.A.B. are characterized
and valued (Beshkova, Simova, Frengova, Simov, & Dimitrov, 2003). Indeed, the quantication of diacetyl in fermented milk products is highly aected by analytical

874

V. Macciola et al. / Food Control 19 (2008) 873878

methods used to determine it (Escamilla, Valdes, Soriano,

& Tomasini, 2000). At present, a specic and internationally proven method to determine diacetyl in milk and milk
products has been not validated. Some time ago, the common analytical method carried out a steam distillation step
to separate diacetyl from the sample, followed by a colorimetric reaction (Pack, Sandine, Elliker, Day, & Lindsay,
1964; Prill & Hammer, 1938; Walsh & Cogan, 1974; Westerfeld, 1945). This procedure was laborious, poorly selective and inaccurate. In fact, colorimetric determination is
not specic and the diacetyl measured was in reality the
sum of diacetyl and acetoin. Furthermore, during the
steam distillation step, 2-acetolactate contained in the samples could be converted by chemical decarboxylation to
diacetyl and acetoin causing an overestimation of both
compounds (Cronin & Rispin, 1996; Veringa, Verburg, &
Stadhouders, 1984).
Direct analysis of diacetyl by gasliquid chromatography (GLC) was performed and validated in bacterial culture supernatant (Lee & Drucker, 1975). The gaschromatographic analysis is more selective, but direct analysis in milk products is very dicult owing to interferences
with other components (Thomhill & Cogan, 1984).
Recently, diacetyl in milk products has often been determined by headspace technique coupled with gasliquid
chromatography, using ame-ionization or mass spectrometer detectors. Diacetyl measured by headspace technique
is conventionally considered as the true value of total diacetyl, but actually, only its volatile part is detected (Monnet, Schmitt, & Divies, 1994). Repartition of diacetyl
between liquid and air phases is aected by several factors,
such as temperature, vapour pressure and sample matrix
composition. In particular, protein and fat inuence the
diacetyls coecient of partition (Haahr, Bredie, Stahnke,
Jensen, & Refsgaard, 2000; Lee, Lo, Richter, & Dill,
1995). Among other techniques, HPLC, spectrophotomeric
and uorometric methods have been used to determine diacetyl in milk products (Guerra Hernandez, Garca Estepa,
& Rodriguez Rivas, 1995; Matsuura, Fujiyama, Minagawa, & Sawa, 1990; Zeppa, Conterno, & Gerbi, 2001).
In this research, a new method to easily and rapidly
determine the diacetyl content of milk products was developed. The gas-chromatographic technique coupled with a
ame ionisation detector has been used. The method was
rst validated and then, used to determine diacetyl on several milk products and for a series of L.A.B. performance
tests.

milk, drawn directly from a local breeding farm; (2) two

samples of UHT commercial milk; (3) four samples of commercial butter; and (4) three samples of commercial
yoghurt. For the L.A.B. performance test, fermented milk
samples were prepared by Mediterranea Biotecnologie s.r.l.
(Termoli, CB, Italy), a specialised company that produces
L.A.B. starters. Dierent mixtures of Lactococcus cultures
from Mediterranea Biotecnologie collections were cultured
in sterilized milk at the following experimental growing
conditions: sample no. 1 was incubated at 30 C for 6 h,
sample no. 2 at 20 C for 16 h and nally, sample no. 3
at 10 C for 17 h. Sterilized milk was analysed as a control.
All samples were stored chilled until required for analysis.
2.2. Sample preparation for GLC analysis
Extraction of diacetyl was performed using acetone. The
butter was melted for two minutes at 40 C. About 2 g of
sample (milk, fermented milk or butter) were accurately
weighted, and shaken vigorously for 30 s with 2 mL of acetone and with 50 lg mL1 of 2,3-pentanedione as internal
standard. After centrifugation at 4000g for 5 min, the supernatant was ltered through a 0.20 lm disposable syringe
membrane lter (Sartorius AG, Gottingen, Germany) and
subsequently injected directly to the gas-chromatographic
apparatus. Final concentration of diacetyl (expressed as
lg of diacetyl per g of sample) was calculated using the
following formula:
Diacetyl lg g1

wIS  AD
AIS  W s

wIS is lg of internal standard (2,3-pentanedione); AD is the

area GLC-FID peak of diacetyl; AIS is the area GLC-FID
peak of internal standard; and Ws is the g of sample used.
2.3. Gas-chromatographic analysis
Gas-chromatographic analysis of diacetyl was performed using a gas-chromatograph Model 8000 Finnigan
(Milan, Italy) equipped with a ame ionisation detector
(FID). One 30 m ZB-Wax (Phenomenex, Torrance, CA,
USA) capillary column, with 0.32 mm i.d., 0.5 lm lm
and 100% polyethylene glycol phase, was used. Gas-chromatographic parameters were: carrier gas helium at
50 kPa ow; injected amount 1 lL; split mode injection
at 1:15 splitting ratio; injector and detector temperatures
were 250 C and 260 C, respectively; oven temperature
was running from 50 C to 240 C at 7 C min1.

2. Materials and methods

2.4. Validation of method
2.1. Materials
Only acetone of reagent grade quality (C.Erba, Rodano,
MI, Italy) was used as a solvent. The standard compounds,
2,3-pentanedione and diacetyl, were from Sigma Chemical
Co. (St. Louis, MO, USA). The milk products analysed
were: (1) several samples of fresh raw cow, sheep or goats

Linearity was determined with solutions of acetone:water 50:50 (v/v) containing diacetyl standard at concentrations between 0.30 and 0.67 mg L1. Each point was
repeated six times. Accuracy was estimated from the slope
of regression line between diacetyl added and detected.
Repeatability or precision of the method was determined

V. Macciola et al. / Food Control 19 (2008) 873878

875

by calculating the coecient of variation (CV%) of the six

independent and consecutive determinations for each
added diacetyl concentration. Detection limit (DL) was
established by running the acetone:water solution six times
with the aim to measure baseline noise at the respective
retention time of diacetyl; DL was thus calculated as the
concentration showing a signal value at three times the
noise.
2.5. Recovery assay
The eciency of the extraction procedure was evaluated
by measuring the recovery percentage of known amounts
of diacetyl standard added to a pasteurised commercial
milk sample. The same milk was used, with and without
diacetyl added, and four subsequent and independent measurements of the diacetyl were carried out such as described
above.
2.6. Statistical analysis
Diacetyl determinations were generally carried out with
four or six independent replicates. Mean, standard deviation (SD), coecient of variation (CV), variance analysis
and linearity were statistically analysed by SPSS software
(SPSS Inc. Headquarters, Chicago) at a signicance level
of 0.05.

3. Results and discussion

3.1. Diacetyl recognition
Firstly, the method development was focused on the
preparation of the sample for GLC analysis. Acetone of
reagent grade quality was used as a solvent to dissolve diacetyl. Acetone showed itself to be an optimal solvent for
diacetyl, as the two compounds have a similar chemical
structure, both being ketones. In addition, acetone also
had a signicant advantage in instantly precipitating milk
proteins and colloids.
After centrifugation of the acetonemilk mixture, the
solids were well separated on the bottom of the tube; the
supernatant appeared clear and, after ltration it was
directly injected into GLC apparatus without a further
purication step.
Secondly, the performance of gas-chromatographic
analysis was ne-tuned. Typical GLC-FID chromatograms
of diacetyl in milk are shown in Fig. 1.
In the chromatogram, the diacetyls peak is shown to be
well separated and its retention time was 4.48 0.03 min,
while the 2,3-pentanedione (internal standard) eluted at
5.52 0.03 min. The diacetyl peak recognition was performed by comparing its retention time with that of the relative standard. The quantitative measurement of diacetyl
was calculated from the diacetyl peak area compared with
that of the 2,3-pentanedione.

Fig. 1. Example of GLC-FID chromatograms of diacetyl in a UHT cows

milk: (a) without diacetyl added; (b) with diacetyl added (200 lg g1). The
peaks are: no. 1. diacetyl; no. 2. 2,3-pentanedione (IS).

3.2. Validation of the GLC method

The precision and accuracy of the method were rst
evaluated by the determination of increasing amounts of
pure diacetyl added to an acetone:water 50:50 v/v solution.
Each diacetyl concentration was measured by GLC-FID
apparatus using six replicates. Fig. 2 plots the amount of
diacetyl added versus the amount detected.
The R2 value, between theoretical and measured concentrations, was 0.9984; this high value indicated good linearity in the range of concentrations tested. The linear slope
could be considered as an expression of the methods accuracy; in fact, a slope value of 1.00 corresponds theoretically

876

V. Macciola et al. / Food Control 19 (2008) 873878

Table 2
Recovery assay of diacetyl added to one UHT milk and determined by the
GLC method

70
y = 0.98 x
R2 = 0.9984

diacetyl detected (mg L-1)

60
50

Diacetyl
added,
lg g1

Diacetyl
detected,
lg g1

Coecient of
variation, %

Values subtracted
of those of the
control lg g1

Percentage
recovery, %

0
13
25
50
100
200

25 2
37 3
49 5
73 6
130 8
215 10

8.0
8.1
10.0
8.2
6.2
8.8

12
24
48
105
190

92.3
96.0
96.9
105.0
95.0

40
30
20
10
0
0

10

20

30

40

50

60

70

diacetyl added (mg L-1)

80

Fig. 2. Relationship between diacetyl detected and added in acetone:water

50:50 (v/v) solution. Each point indicates the mean of six independent
determinations.

Table 1
Repeatability of the GLC method, calculated on the recovery of dierent
concentrations of diacetyl added to a 50:50 (v/v) acetone:water solution
Diacetyl added
(mg L1)

Diacetyl detected
(mg L1)

Standard
deviation, SD

Coecient of
variation, CV%

0.3
1.7
3.3
5.0
6.7
8.3
10.0
11.7
13.3
16.7
33.3
66.7

0.3
1.6
3.6
5.4
6.9
8.3
9.9
10.3
14.2
15.2
33.7
65.2

0.01
0.11
0.17
0.30
0.43
0.61
0.50
0.50
0.70
0.84
1.09
1.50
Mean

3.3
6.7
4.7
5.5
6.2
7.4
5.1
4.8
4.9
5.5
3.2
2.3
5.0

For each concentration, data represent the mean and standard deviation
of six independent determinations.

to the best level of accuracy, equal to 100% of recovery.

The experimental linear slope was estimated 0.98
(y = 0.98x).
Precision of the GLC method was measured by calculation of the coecient of variation relative to the six replicates carried out for each of the concentrations tested
used to construct the calibration curve. All results are
shown in Table 1.
On average, the coecient of variation was equal to
5.0%, showing satisfactory precision for the concentrations
tested. Finally, the detection limit was calculated as
1 mg L1.
3.3. Recovery assay on milk
The eciency of the extraction procedure was evaluated
by determination of recovery percentage of known
amounts of diacetyl standard added to a UHT commercial
milk sample. All results are reported in Table 2.

Mean
Standard deviation
% CV

8.2
1.2
15.1

97.0
4.8
4.9

Data are the mean of four independent determinations.

Diacetyl was unexpectedly found in signicant amounts

(equal to 25 lg g1) also in the milk used as a control. After
subtracting the control value, the concentration of diacetyl
detected in all samples was equivalent to that added.
Recovery percentage was, on the average, 97.0%, while
coecient of variation was equal to 8.2%. Thus, the precision and accuracy of the GLC method was satisfactory in
the analysis of the milk sample.
3.4. Diacetyl contents in milk, fermented milk and butter
samples
In the nal part of this research, the GLC method was
used to determine diacetyl in dierent milk product samples. The samples and their diacetyl contents, detected with
this GLC method, are shown in Table 3.
Milk and yoghurt were easily mixed with acetone and,
after centrifugation and ltration, the nal solutions
appeared clear and ready for GLC analysis. Butter samples
were rst heated for few minutes at 40 C, in order to
achieve better dissolution in acetone. Each sample was
independently analysed four times.
The diacetyl contents found were very variable between
the samples. Mean values found were 91 37 lg g1 for
fresh raw milk, 68 50 lg g1 for commercial UHT milk,
21 8 lg g1 for butter and nally, 43 17 lg g1 for
yoghurt. The repeatability of single measurements were
generally good; coecient of variation was, on average,
5.5% and in all cases less than 8.0%.
Milk samples had the highest values of diacetyl, and also
the most variability.
Diacetyl contents in commercial UHT milk were
32 1.2 lg g1 and 103 5.5 lg g1 in sample no. 1 and
no. 2 (Table 2), respectively.
Nevertheless, signicant amounts of diacetyl were
detected also in fresh raw milk samples. Fresh raw goats
milk samples showed the highest content of diacetyl, with
values higher than 100 lg g1. Diacetyl measured on fresh
raw sheeps milk was lower (58 2.0 lg g1) whilst for
fresh cows raw milk samples, diacetyl contents were

V. Macciola et al. / Food Control 19 (2008) 873878

Table 3
Diacetyl content (lg g1) on several samples of milk, butter and yoghurt
Samples

Diacetyl, lg g1

SD

CV, %

877

Table 4
Production of diacetyl in sterilised milk from dierent mixs of lactic acid
bacteria
Diacetyl detected, lg g1

Commercial UHT milk

1
32
2
103
Mean SD
68 50

1.2
5.5

Fresh raw milk

3 Cow
4 Cow
5 Goat
6 Goat
7 Sheep
Mean SD

96
48
133
118
58
91 37

3.9
2.6
6.0
9.3
2.0

Butter
8
9
10
11
Mean SD

32
16
15
21
21 8

1.8
0.8
1.3
1.4

5.6
5.0
6.2
6.7

Yoghurt
12
13
14
Mean SD

42
27
61
43 17

2.5
1.8
3.6

6.0
6.7
5.9
5.5 1.2

3.8
5.3

5.4
4.5
7.9
3.4
6.1

Mean values of four replicates. SD, standard deviation; CV, coecient of

variation.

96 3.9 lg g1 in one case and 48 2.6 lg g1 in the

other. As only a small number of milk samples were analysed, it was dicult to propose the natural diacetyl content of milk. Diacetyl in unfermented milk could be
directly produced during lactation, or it could be derived
by enzymatic modication of some milk compounds, as
acetoin, a-acetolactate or an unknown chemical precursor.
On the other hand, knowledge about the diacetyl content
of unfermented milk is not available in signicant amount
in the scientic literature. The presence of signicant
amounts of diacetyl in the fresh raw milk sample, if conrmed, is certainly interesting be used as a basis for further
research in dairy science.
With regards to butter and yoghurt (Table 3), diacetyl
contents measured by GLC method was higher than values
generally reported in the scientic literature and these differences were probably due to the analytical techniques
zdemir, 2002; Beshkova et al.,
used (Bakirci, C
elik, & O
2003; Guerra Hernandez et al., 1995). In particular, when
headspace analysis is used, only volatile diacetyl was found
(Monnet et al., 1994). Milk and milk product compositions, as well as time and temperature of incubation, could
aect the partition coecient of diacetyl and thus the quantitative results (Haahr et al., 2000; Lee et al., 1995). On the
contrary, when direct methods have been used, total diacetyl has been detected at the highest concentrations, similar
to those found in this study (Levata-Jovanovic & Sandine,
1996).
Finally, diacetyl was measured in dierent fermented
milk products obtained under laboratory controlled conditions (Table 4). Three dierent L.A.B. starter mixtures were

Sterilized milk (control)

Fermented milk samples
1
2
3

36 2a
35 5a
48 7b
50 6b

Mean values standard deviation of four replicates. The letters indicate a

signicant dierence at P < 0.05.

tested. L.A.B. starter mixtures were composed by mesophilic homofermentative and heterofermentative species of
Lactococcus. L.A.B. starter mixtures were cultured in sterilized milk, under varying incubation conditions depending
on the type of L.A.B. mix. At the end of incubation, diacetyl
was determined by the GLC method on the fermented
milk. Sterilized milk was as also analysed as the control.
In sample no.1 (Table 4) diacetyl detected was 35
2 lg g1 which is similar to the control (36 5 lg g1).
On the contrary, diacetyl contents of the sample no. 2
and no. 3 (Table 4) were higher and signicantly dierent
compared with the control. In particular, the values of diacetyl found were 48 7 lg g1 on the sample no. 2 and
50 6 lg g1 on the sample no. 3. The conditions, especially temperature, used to culture fermented milk samples,
certainly aects the production of diacetyl. In fact, sample
no.1 was incubated at a temperature (30 C for 6 h) higher
than those used for the sample no. 2 (20 C for 16 h) and
the sample no. 3 (10 C for 17 h). The high temperature
of incubation had promoted the growth of homofermentative Lactococcus strains that did not produce diacetyl. On
the contrary, the lower temperature of incubation had
selectively and positively aected the growth of heterofermentative strains responsible of the production of diacetyl.
This last experiment suggested that GLC methods can
be proposed to monitor diacetyl in fermented milk processing and to test L.A.B. starter ability.
4. Conclusion
A new method for the determination of diacetyl in milk
products by gas-chromatographic technique was developed. The method was relatively simple, rapid, sensitive
and highly selective. Preparation of samples was very easy,
fast and low-cost owing to only acetone being required as a
solvent. The present method did not need specialized apparatus additional to that required for GLC-FID. Finally, the
method was shown to be accurate and precise.
On the milk, diacetyl recovery was estimated at 97% and
coecient of variation was generally less than 9%.
Finally, diacetyl was determined on fresh and high-temperature treated milk, commercial butter, yoghurt and also
in a L.A.B. performance test using the new GLC method.
The method could be proposed as a routine technique
for the determination of diacetyl in milk products.

878

V. Macciola et al. / Food Control 19 (2008) 873878

References
Baron, M., Roy, D., & Vuillemard, J. C. (2000). Biochemical characteristics of fermented milk produced by mixed-cultures of lactic starters
and bidobacteria. Lait, 80, 465478.
zdemir, C. (2002). The eects of commercial
Bakirci, I., C
elik, S., & O
starter culture and storage temperature on the oxidative stability and
diacetyl production in butter. International Journal of Dairy Technology, 55, 177181.
Beshkova, D. M., Simova, E. D., Frengova, G. I., Simov, Z. I., &
Dimitrov, Zh. P. (2003). Production of volatile aroma compounds by
ker starter cultures. International Dairy Journal, 13, 529535.
Cronin, D. A., & Rispin, E. (1996). The determination of diacetyl in
cultured buttermilk in the presence of the unstable precursor aacetolactic acid. In A. J. Taylor & D. S. Mottram (Eds.). Flavour
science, recent developments (pp. 118121). Cambridge: The Royal
Society of Chemistry.
Escamilla, M. L., Valdes, S. E., Soriano, J., & Tomasini, A. (2000). Eect
of some nutritional and environmental parameters on the production
of diacetyl and on starch consumption by Pediococcus pentosaceus and
Lactobacillus acidophilus in submerged cultures. Journal of Applied
Microbiology, 88, 142153.
Fox, P. F., Lucey, J. A., & Cogan, T. M. (1990). Glycolysis and related
reactions during cheese manufacture and ripening. Critical Reviews in
Food Science and Nutrition, 29, 237253.
Fox, P. F., Law, J., McSweeney, P. L. H., & Wallace, J. (1993).
Biochemistry of cheese ripening. In Cheese: Chemistry, physics and
microbiology (2nd ed.. In P. F. Fox (Ed.). General aspects (Vol. 1,
pp. 389438). London: Chapman & Hall.
Guerra Hernandez, E. J., Garca Estepa, R., & Rodriguez Rivas, I. (1995).
Analysis of diacetyl in yogurt by two new spectrophotometric and
uorometric methods. Food Chemistry, 53, 315319.
Haahr, A. M., Bredie, W. L. P., Stahnke, L. H., Jensen, B., & Refsgaard,
H. H. F. (2000). Flavour release of aldehydes and diacetyl in oil/water
systems. Food Chemistry, 71, 355362.
Law, B. A. (2001). Controlled and accelerated cheese ripening: The
research base for new technologies. International Dairy Journal, 11,
383398.
Lee, S. M., & Drucker, D. B. (1975). Analysis of acetoin and diacetyl in
bacterial culture supernatants by gasliquid chromatography. Journal
Clinical Microbiology, 2, 162164.

Lee, K. D., Lo, C. G., Richter, R. L., & Dill, C. W. (1995). Eect of milk
composition on the partition coecients of diacetyl, acetaldehyde, and
ethanol in acidied milk products. Journal of Dairy Science, 78,
26662674.
Levata-Jovanovic, M., & Sandine, W. E. (1996). Citrate utilization and
diacetyl production by various strains of Leuconostoc mesenteroides
ssp. Cremoris. Journal of Dairy Science, 79, 19281935.
Matsuura, H., Fujiyama, K., Minagawa, N., & Sawa, J. (1990).
Determination of acetoin, diacetyl and acetaldehyde in food by
HPLC. Bunseki Kagaku, 39, 405409.
McSweeney, P. L. H., & Fox, P. F. (2004). Metabolism of residual lactose
and of lactate and citrate. In Cheese: Chemistry, physics and microbiology (3rd ed.. In P. F. Fox, P. L. H. McSweeney, T. M. Cogan, & T.
P. Guinee (Eds.). General aspects (Vol. 1, pp. 361372). London:
Chapman & Hall.
Monnet, C., Schmitt, P., & Divies, C. (1994). Method for assaying volatile
compounds by headspace gas-chromatography and application to
growing starter cultures. Journal of Dairy Science, 77, 18091815.
Pack, M. Y., Sandine, W. E., Elliker, P. R., Day, E. A., & Lindsay, R. C.
(1964). Owades and Jakovac method for diacetyl determination in
mixed-strain starters. Journal of Dairy Science, 47, 981985.
Palles, T., Beresford, T., Condon, S., & Cogan, T. M. (1998). Citrate
metabolism in Lactobacillus casei and Lactobacillus plantarum. Journal
of Applied Microbiology, 85, 147154.
Prill, E. A., & Hammer, B. W. (1938). A colorimetric method for the
micro-determination of diacetyl. Iowa State College Journal of Science,
12, 385395.
Thomhill, P. J., & Cogan, T. M. (1984). Use of gasliquid chromatography to determine the end products of growth of lactic acid bacteria.
Applied and Environmental Microbiology, 47, 12501255.
Veringa, H. A., Verburg, E. H., & Stadhouders, J. (1984). Determination
of diacetyl in dairy products containing a-acetolactic acid. Netherlands
Milk and Dairy Journal, 38, 251263.
Walsh, B., & Cogan, T. M. (1974). Separation and estimation of acetoin
and diacetyl in milk. Journal of Dairy Research, 41, 2530.
Westerfeld, W. W. (1945). A colorimetric determination of blood acetoin.
Journal of Biological Chemistry, 161, 495502.
Zeppa, G., Conterno, L., & Gerbi, G. (2001). Determination of organic
acids, sugars, diacetyl and acetoin in cheese by high-performance
liquid chromatography. Journal of Agriculture and Food Chemistry, 49,
27222726.