Development and Validation of a Novel Free Fatty Acid Butyl Ester Gas Chromatography Method for the Determination of Free Fatty Acids in Dairy Products

Article abstract of DOI:10.1021/acs.jafc.8b05462

Accurate quantification of free fatty acids in dairy products is important for both product quality control and legislative purposes. In this study, a novel fatty acid butyl ester method was developed, where extracted free fatty acids are converted to butyl esters prior to gas chromatography with flame ionization detection. The method was comprehensively validated to establish linearity (20-700 mg/L; R² > 0.9964), limits of detection (5-8 mg/L), limits of quantification (15-20 mg/L), accuracy (1.6-5.4% relative error), interday precision (4.4-5.3% relative standard deviation), and intraday precision (0.9-5.6% relative standard deviation) for each individual free fatty acid. A total of 17 dairy samples were analyzed, covering diverse sample matrices, fat content, and degrees of lipolysis. The method was compared to direct on-column injection and fatty acid methyl ester methods and overcomes limitations associated with these methods, such as either column-phase absorption or deterioration, accurate quantification of short-chain free fatty acids, and underestimation of polyunsaturated free fatty acid.

Full text of DOI:10.1021/acs.jafc.8b05462

Subscriber access provided by University of South Dakota
New Analytical Methods
Development and Validation of a Novel Free Fatty
Acid Butyl Ester Gas Chromatography Method for the
Determination of Free Fatty Acids in Dairy Products.
David T Mannion, Ambrose Furey, and Kieran Noel Kilcawley
J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.8b05462 • Publication Date (Web): 07 Dec 2018
Downloaded from http://pubs.acs.org on December 13, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 21
Journal of Agricultural and Food Chemistry
1
2
3
Development and Validation of a Novel Free Fatty Acid Butyl Ester
Gas Chromatography Method for the Determination of Free Fatty
Acids in Dairy Products.
4
David T. Mannion†‡, Ambrose Furey‡, Kieran N Kilcawley†*.
5
6
†Food Quality and Sensory Department, Teagasc Food Research Centre, Moorepark, Fermoy, Co. Cork, Ireland, P61 C996
‡ Physical Sciences Department, Cork Institute of Technology, Bishopstown, Cork, Ireland. T12 P928
7
8
9
0
1
2
3
4
5
6
7
ABSTRACT: Accurate quantification of free fatty acids in dairy products is important for both product
quality control and for legislative purposes. In this study a novel fatty acid butyl ester method was
developed where extracted free fatty acids are converted to butyl esters prior to gas chromatography
flame ionization detection. The method was comprehensively validated to establish linearity (20-700
1
1
1
1
1
1
1
1
2
mg/L, R > 0.9964), limits of detection (5-8 mg/L), limits of quantification (15-20 mg/L), accuracy (1.6-
5.4 % relative error), inter-day precision (4.4-5.3 % relative standard deviation) and intra-day precision
(0.9-5.6 % relative standard deviation) for each individual free fatty acid. Seventeen dairy samples were
analyzed covering diverse sample matrices, fat content and degrees of lipolysis. The method was
compared to direct on-column injection and fatty acid methyl ester methods and overcomes limitations
associated with these methods such as either column phase absorption or deterioration, accurate
quantification of short chain free fatty acids and the underestimation of polyunsaturated FFA.
1
8
INTRODUCTION:
19
20
21
22
Free fatty acids (FFAs) are important components in dairy fat as they have a range of potential
1
,2
attributes; flavor development, anti-microbial activity, nutrition and functional characteristics . The
accurate quantification of FFAs in dairy products is also important for quality control, legislative
purposes, authentication, food research and product development. Therefore it is important to have an
1
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 2 of 21
23
24
25
26
27
28
29
30
31
32
33
34
35
accurate, reliable and robust method that can be used to quantify FFAs in all dairy products. Some
difficulties arise in relation to the quantification of FFAs in dairy products, which relate to the wide
variation in fat content (~0.5 to >80 %), degree of lipolysis (0.5 to 82 %), and because of the differing
1
solubility and volatility of FFAs based on chain length . Therefore fat extraction techniques need to be
able to take into account these differences and efficiently extract both water soluble short chain FFAs
and organic soluble medium and long chain FFAs. A method should also avoid the use of an
evaporation step to prevent losses of volatile short chain FFAs and remove or negate any water that may
be present in the sample as it can interfere in the extraction of water soluble FFAs. The most common
approach to quantify FFAs is the use of gas chromatography flame ionization detection (GC-FID),
which is likely to remain the case as the equipment is relatively cheap, simple, robust, reproducible,
widely available and arguably more accurate than mass selective detectors for quantification purposes.
1
While there have been some recent advances in methodology, which have been reviewed , development
in FFA determination in dairy products has been largely lacking for the past several decades.
3
3
3
3
4
4
4
4
4
4
4
4
6
7
8
9
0
1
2
3
4
5
6
7
The two most commonly utilized GC-FID methods to date, fatty acid methyl esterification (FAME)
where the FFAs are esterified in the GC injector to methyl esters using tetramethylammonium
3
hydroxide as a catalyst and direct on-column injection where the isolated FFA extract is injected
4
directly into the GC , were assessed for linearity, accuracy, limits of detection (LOD) and limits of
5
quantification (LOQ) by Mannion et al. . Prior to this study, these two methods have been in existence
for the past 20+ years without a comprehensive assessment of their quantitative suitability for use with a
diverse range of dairy samples matrices that are typically tested for in regulatory and food testing
laboratories. The direct on-column approach had lower levels of LOD and LOQ, mainly because the
FFAs are directly injected onto the GC as opposed to a split injection used in the FAME method.
However, both methods had significant disadvantages. The direct injection on-column method results in
5
accumulative deterioration of the column phase and irreversible FFA absorption , which adversely
impacts method robustness and the quantification of some longer chain FFAs. In the FAME method,
2
ACS Paragon Plus Environment

Page 3 of 21
Journal of Agricultural and Food Chemistry
4
4
5
5
5
8
9
0
1
2
butyric acid co-eluted with the injection solvent impacting on its quantification, especially at low
concentrations. Artifact peaks, which are believed to be a result of by-products from the
tetramethylammonium hydroxide (TMAH) reaction, were found to interfere with the quantification of
other short-chain FFAs. Also, losses of polyunsaturated FFAs were observed because of the strong
alkaline nature of the TMAH reagent5.
53
54
55
56
57
58
59
60
61
62
This study was undertaken to create a new GC-FID method that overcomes existing limitations of the
FAME and direct on-column methods, but that is also robust, accurate and with a low LOD and LOQ.
We believe that combining aspects of both the FAME and direct on-column methods offers the best
potential. The use of solid phase extraction to isolate the FFAs from glycerides (mono-, di and tri-
4
glycerides) as used in the direct on-column method provides a relatively pure FFA extract. This can be
converted to butyl esters instead of methyl esters to potentially overcome issues with losses of short
chain volatiles during evaporation steps, and with issues related to water solubility and co-elution with
the injection peak, as these butyl esters are of higher molecular weight, more non-polar, less volatile and
less water soluble. Butyl esters have been previously used to quantify short chain FFAs in dairy
products and for the determination of FFAs in milk fat and cheese8-11.
6
,7
8
6
6
6
6
6
6
6
3
4
5
6
7
8
9
Very little detailed information on the analysis of FFAs in dairy products exists, especially in relation to
method validation, accuracy, precision and robustness. The term “dairy” encompasses a very wide and
diverse range of products and most methods described are limited in their application to specific dairy
products. Therefore having a single quantifiable robust method would be a significant analytical
advancement and of great benefit for both regulatory and food testing laboratories. This study covers the
validation of a fatty acid butyl ester (FABE) method for the quantification of FFAs in a range of dairy
products and compares it to the existing widely used FAME and direct on-column methods.
7
7
0
1
MATERIALS AND METHODS
3
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 4 of 21
5
72
73
74
75
76
Solvents, standards and samples were sourced as per Mannion et al. . Infant formula, milk powder, and
enzyme modified cheese (EMC) powder samples were stored under nitrogen in sealed containers at
room temperature. Milk, yoghurt, butter, EMC paste and ice cream samples were transferred into sterile
containers which were frozen at -18 °C until required. All cheese samples were vacuum packed and
frozen at -18 °C. All samples were stored in dark conditions.
7
7
7
8
8
8
8
8
8
7
8
9
0
1
2
3
4
5
FABE analysis was carried out on a Varian CP3800 gas chromatograph (Aquilant, Dublin 22, Ireland)
equipped with a CP8400 auto-sampler and flame ionization detector. The column was a CP FFAP CB
capillary column (25 m x 250µm I.D., 0.32 µm phase thickness - Agilent Technologies, Ireland). A
Varian 1079 programmed temperature vaporizer (PTV) injector was used and held at 250 °C for the
entire run. A split of 1:50 was used. The inlet liner was a split/splitless wool packed liner with a volume
of 250µl. An injection volume of 0.7µl was used, which was calculated to yield a vapor cloud of 208µl.
The column oven was held at 40 °C for 2 min and raised to 240 °C at 7.5 °C/min, which was held for 5
min. The total runtime was 33.67 min. The FID was operated at 300 °C. The carrier gas was helium and
was held at a constant flow of 1.2 mL/min.
8
8
8
8
9
6
7
8
9
0
A 7696A Sample Prep Workbench (Agilent Technologies, Little Island, Cork, Ireland) was employed in
the preparation of all standards for calibration and validation studies, and for the derivitization reactions
(butyl ester derivitization using BF in butanol as a catalyst). This instrument has been applied by
3
5
Mannion et al in previous work for FAME analysis. It was configured with a 100 µl gas tight syringe in
the back tower, and a 500 µl gas tight syringe in the front tower.
91
92
93
94
Lipid extraction and solid phase extraction were carried out as per the procedure outlined by Mannion et
5
al. . To briefly describe, lipids were extracted from the sample using heptane/diethyl ether (1:1 v/v).
Solid phase extraction was carried out on this extract to isolate the FFAs from the rest of the lipid
mixture, using 5 mL of 2% (v/v) formic acid in diethyl ether.
9
9
5
6
The FFA extract (60 µl) was transferred into a capped 2 mL amber GC vial containing a 400 µl glass
insert. BF in butanol reagent (60 µl) was added to the sealed vial containing the FFA extract. This was
3
4
ACS Paragon Plus Environment

Page 5 of 21
Journal of Agricultural and Food Chemistry
9
9
9
7
8
9
vortexed for 10 seconds and then was heated at 80 °C for 1 h. After heating, 30 µl of deionized water
and 30 µl of heptane were added and further vortexed for 1 min. Phase separation occurs between the
aqueous and organic layer (butyl esters reside in organic layer). An aliquot of 80 µl of the organic layer
was transferred into a capped 2 mL amber vial with a 400 µl glass insert containing 200 µl of saturated
NaCl solution and vortexed for 1 min. After mixing, 60 µl of the upper layer containing the butyl esters
was transferred into a capped 2 mL amber vial containing a 250 µl glass insert. A 0.7 µl aliquot of this
solution was sampled for GC analysis.
1
1
1
1
00
01
02
03
1
1
1
1
1
1
1
1
1
04
05
06
07
08
09
10
11
12
The extraction procedures described (lipid extraction followed by solid phase extraction) were applied
to six replicates (n=6) of each sample. Three of these replicates were spiked with 0.5 mg of each FFA
(C4:0-C18:3), this was done by adding 0.5 mL of a solution containing each FFA at 1000 ppm
concentration in heptane, with the exception of the EMC samples where 5 mL of this solution was
added. These spiked samples were used for recovery determinations by comparing the difference in
measurements against the un-spiked samples. The collected 5 mL extracts of 2 % formic acid/diethyl
ether (FA/DE) from the solid phase extraction step were treated according to the derivitization protocol
as described, and analyzed by GC-FID. For samples where the concentration was greater than the
calibration range, the extract was diluted and re-analyzed until within range.
113
114
115
116
117
118
119
120
All standard mixtures, which were used for instrument calibration and method validation studies; were
5
prepared as described by Mannion et al. . These were prepared using the Nu-Chek certified FFA
calibration mix, which contained FFAs C4:0-C18:3. The standards were collected in 2 % FA/DE
solution, converted to butyl esters using the protocol described, using the sample prep workbench
(Agilent Technologies Ltd, Ireland). The internal standard (ISTD) (C5:0, C11:0 and C17:0) was added
during the dilution step. Five point calibration curves were established with a concentration range of 20-
700 ppm for all FABEs with the ISTDs at a concentration of 200 ppm. Calibration curves were
established based on the correction factors outlined in Mannion et al.5.
5
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 6 of 21
1
1
1
1
1
21
22
23
24
25
For analysis of each dairy product, a batch consisted of sixteen samples (n=16). This comprised of three
un-spiked samples and three spiked samples for recovery determinations; totaling six samples (n=6),
with each sample derivitized in duplicate (n=12). Four standards were included with each sample batch
(n=16) to monitor inter-day precision, as part of the method validation study. Intra-day precision was
also monitored, where ten standards (n=10) were analyzed sequentially in one day.
1
1
1
26
27
28
RESULTS & DISCUSSION
Method Validation
129
130
131
132
133
134
135
136
The linearity was investigated where the FFAs were collected in 2 % FA/DE as per standards and
derivitized to butyl esters. The linear range was established at 20-700 ppm. The calibration data
obtained where line equation, correlation coefficient and linear range are illustrated in Table 1. Linearity
was obtained over these concentration ranges where correlation coefficients of ≥0.996 were achieved for
all butyl esters. LOD and LOQ were evaluated based on the signal to noise ratio. LOD was established
at 3 times and LOQ at 10 times the signal to noise level (Table 1), which was 5 ppm (LOD) and 15 ppm
(LOQ) for most FFAs. The exceptions were C4:0-C8:0, which were at 8 ppm and 20 ppm for LOD and
LOQ respectively.
1
1
1
1
37
38
39
40
Precision was monitored over the course of the analysis by analyzing FFA standards (C4:0-C18:3) at
100 ppm concentration in 2% FA/DE, which were converted to butyl esters using the sample prep
workbench (Table 1). Percentage (%) relative standard deviation (RSD) is used to express precision,
which is calculated by the following formula:
푆퐷
1
41
%
x 100
푅푆퐷 = 푥
1
1
1
42
43
44
Where SD is the standard deviation and 푥 is the mean. Intra-day repeatability was established by
analyzing ten samples (n=10) in one day with 0.9-5.6 % RSD achieved across all butyl esters. The inter-
day study was structured so that precision was monitored during the analysis of the dairy products. Four
6
ACS Paragon Plus Environment

Page 7 of 21
Journal of Agricultural and Food Chemistry
1
1
1
45
46
47
FFA samples were analyzed in the same batch with each dairy product (n=72), which were analyzed
over the course of approximately 1 month for the inter-day study. Excellent repeatability, 4.4-5.4 %
RSD, was achieved across all esters.
1
1
1
1
1
1
1
1
48
49
50
51
52
53
54
55
The accuracy of the method was expressed as the relative error (RE) of each FFA across the
concentration range 20-700 ppm prepared in 2 % FA/DE, with the acids subsequently converted to butyl
esters prior to analysis. Table 1 shows the average error across this concentration range, a more detailed
breakdown is shown in Table S-1 (Supporting Information). The RE is based on the true value of the
butyl ester in solution compared with the value obtained from the analysis. The average values were
obtained from the analysis of 9 calibration curves, along with the replicate RSD values. Across all the
esters the highest error was 5.6-19.5 % at 20 ppm, the RE % is significantly reduced at the higher
concentration range (100-700 ppm) yielding 0.2-5.0 % across all esters.
1
1
56
57
Application to Dairy Samples
158
159
160
161
162
163
164
165
166
A typical chromatogram of FABE analysis is shown in Figure 1 (Special reserve Cheddar cheese
sample). The complete separation of C4:0-C18:3 butyl esters are evident. The results obtained from the
5
FABE method for all samples are displayed in Table 2. The study by Mannion et al. employed a direct
on-column approach and FAME approach to analyze FFAs in these same samples. A summary of
results obtained from these methods compared to the FABE method are displayed in Table 3. Detailed
results are shown in Tables S-2 to S-4. Values reported are the average values obtained from six
replicates (n=6).
Recovery was determined by measuring the difference between samples spiked
with 0.5 mg of each FFA (C4:0-C18:3), except for the EMCs where 5mg was added, and un-spiked
samples.
1
1
1
67
68
69
For the cheese samples (Brie, processed cheese, light Cheddar, mild Cheddar, special reserve Cheddar
and Blue Stilton) total FFA recoveries were >95 % with the exception of processed cheese where a total
FFA recovery of 85 % was obtained. When reviewing the individual recoveries for processed cheese the
7
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 8 of 21
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
majority of the FFAs achieved >92 % recovery with the exception of C14:0, C16:0, C18:0, and C18:1,
where a much lower recovery (>46 %) was achieved. The processed cheese sample showed some
variation in the recoveries obtained between the direct on column (88 %), FAME (102 %) and FABE
(88 %) methods. Also there was a higher variation in the measured ppm concentrations between the 3
methods; 1246 ppm for FABE, 1104 ppm for the direct on-column and 952 ppm for the FAME (Table
3) for processed cheese (RSD 13.4 %), when compared to the other cheeses analyzed (5.2-9.7 % RSD).
Processed cheese is not a natural cheese and contains emulsifiers which may have adversely impacted
fat separation and thus isolation of the FFAs from the lipid extract, either during the solvent extraction
and/or solid phase extraction steps. For the solvent extraction, diethyl ether/heptane (1:1 v/v) was used.
Increasing the ratio of the non-polar heptane to the more polar diethyl ether, might improve the
recoveries of the longer chain FFAs, however this could potentially reduce the recoveries of the short
chain acids. Despite the lower recoveries obtained for the longer chain acids, consistent results were
achieved with an RSD 1-7 % across the range of FFAs. While complete recovery of FFAs greater than
C14:0 was not achieved from the processed cheese sample, the method displayed analytical precision.
FFA concentrations varied considerably across the different cheese types.
185
186
187
188
189
190
191
192
193
194
Infant formula, milk powder and yoghurt samples achieved total recoveries of >94 %, although the short
chain FFAs were below the established limit of quantification (LOQ 20 ppm). Total FFA concentrations
were relatively low in the powder and yoghurt samples, at 296 ppm and 217 ppm for milk powder and
infant formula, respectively, and at 223 ppm for yoghurt. Concentrations of 661 ppm and 90 ppm were
determined for ice cream and milk, respectively. However, the analysis of the milk and ice cream
samples had a high RSD for some individual acids (45 % for C14:0 in ice cream and 23 % for C16:0 in
milk) and poor recovery for others (133 % for C16:0 and 165 % for C18:0 in ice cream). As highlighted
5
by Mannion et al. , the extraction method employed doesn’t seem to be effective in obtaining complete
recovery of all FFAs from these samples (ice cream and milk) nor does it seem capable of achieving
consistent results. This is demonstrated in Table 3 where 51.4 % and 64.3 % RSD was obtained for milk
8
ACS Paragon Plus Environment

Page 9 of 21
Journal of Agricultural and Food Chemistry
1
1
1
1
1
2
2
95
96
97
98
99
00
01
and ice cream respectively between the three methods. Despite what seems to be excellent total
recoveries for the FABE method (109 % and 98 % for ice cream and milk respectively), the
inconsistency of the analysis demonstrates that the extraction procedure is not suitable for the FFA
analysis of these samples. This is most likely a result of the fact that the milk fat globule membrane is
largely intact potentially inhibiting the solvent during fat extraction. The addition of a suitable solvent to
disrupt the lipoprotein complex is necessary, such as the use of ethanol as described in the analysis of
milk by De Jong and Badings4.
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
For the butter samples high recoveries (90-106 %) and excellent RSD (1-6 %) values were obtained
across all butyl esters. The FABE method seems accurate and precise with high recoveries, however
when the results are compared (Table 3) to those achieved by the other methods, there was a significant
difference. The FFA concentrations for butter by the direct on-column, FAME and FABE methods were
1492, 1295 and 1960 ppm, respectively. These differences could not be solely attributed to instrument
variance or accumulative absorption of FFAs on the column phase. Further investigation highlighted the
presence of glycerides in the FFA extract for this sample type. The solid phase extraction step describes
the use of 10 mL of 20 % DE in hexane to remove glycerides from the column prior to collecting the
FFAs in 5 mL of 2 % FA/DE. As butter has a very high fat content (~ 80 %) it seems that the washing
step was insufficient in removing all the glycerides from the solid phase extraction cartridge. As an
acidic catalyst BF is capable of esterifying FFAs as well as trans-esterifying glycerides, thus if FFAs
3
and glycerides are not completely separated, glycerides will inadvertently be converted to butyl esters
and quantified as FFAs. To confirm the presence of glycerides in the butter extract after solid phase
extraction, the use of TMAH was employed. TMAH can be used to form methyl esters of both FFAs
3
and triglycerides in the same solution , by forming ammonium salts of FFAs in an aqueous layer and
forming methyl esters of glycerides in a separate organic layer. The ammonium salts of FFAs are
subsequently converted to methyl esters within the GC injector without any interference from
glycerides. This derivitization procedure was carried out on the butter FFA extracts with the resulting
9
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 10 of 21
2
2
2
2
2
2
2
2
2
2
20
21
22
23
24
25
26
27
28
29
organic layer analyzed by GC-FID. Figure S-1 is a chromatogram obtained from the analysis of the
organic layer after TMAH derivitization of a butter FFA extract. While this is only a qualitative
determination, fatty acid peaks were evident and had to result from the presence of glycerides. A
comparison of these peak areas against methylated FFA extracts (Figure S-2) confirms considerable
responses, indicating the presence of glycerides. Thus these glycerides are converted to butyl esters
resulting in the over estimation of FFAs in butter when using the FABE method. This was an unforeseen
outcome of the analysis and has not previously been reported. Thus for FABE analyses of butter
samples the solid phase extraction step needs to be modified to ensure complete removal of the
glycerides prior to derivatization. This could be easily achieved by increasing the amount of solvent
during the washing steps prior to collecting the FFAs in the 2 % FA/DE solution.
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
Out of all the samples analyzed in this study the EMCs had the highest FFA concentrations ranging
from 44,712–131,622 ppm. Recovery determinations for the powder samples were greater (88 % and 93
% for butter powder and cream powder respectively) than for the paste samples (47-82 %). The
recovery of longer chain FFAs (>C14:0) was lower in the EMC samples and was even more pronounced
for the paste samples. The lowest total recovery was achieved for the natural blue cheese paste samples
at 47 % with the cream paste and butter paste at 70 % and 82 %, respectively. It was difficult to form a
5
reliable conclusion on the recovery determinations reported in Mannion et al , where some of the
recovery determinations calculated were not dependable, e.g. -27 % (C18:1 from nat. blue cheese paste)
and 177 % (C16:0 from cream paste). This was believed to be partly a result of the accumulative
absorption of FFAs onto the column phase because of the extremely high concentrations of FFAs in
these samples. Given that this issue was not experienced with butyl esters, it must be considered that the
recovery losses were because of the practical aspects of the extraction method and not a result of issues
with the instrument analysis. After a review of the area responses it seems overloading of the
aminopropyl columns occurred with the spiked EMC samples because of the very high concentration of
FFAs present. For recovery determinations where a known amount of FFAs (C4:0-C18:3) was added to
1
0
ACS Paragon Plus Environment

Page 11 of 21
Journal of Agricultural and Food Chemistry
2
2
2
2
2
2
2
2
45
46
47
48
49
50
51
52
the sample, it was necessary to add a large amount of this FFA mix (5 mg of each FFA) to the EMCs,
where only 0.5 mg of each FFA was sufficient for the other sample types in this study. This was
necessary to ensure that sufficient measurable differences in area responses could be achieved between
the spiked and un-spiked samples, especially when the measured concentration was outside the linear
range of the analysis and a dilution of the sample extract was employed. A consequence of this however
was that the capacity of the aminopropyl cartridges was exceeded resulting in a loss of FFAs. There are
obvious solutions to this problem; a further reduction of the EMC sample volume (0.5 g was used in this
study) and/or greater capacity aminopropyl columns.
2
2
2
2
2
2
2
2
2
2
2
2
53
54
55
56
57
58
59
60
61
62
63
64
The application of the FABE method yielded excellent RSD values for most samples, with the cheese
samples (1-7 % RSD), EMCs (1-8 % RSD), infant milk formula (2 % RSD), milk powder (5 % RSD),
butter (2 % RSD) and yoghurt (8 % RSD) all displaying excellent consistency between the replicates.
As discussed, the analysis of ice cream and milk yielded poor results between the replicates. Although
ice cream obtained 12 % RSD for the total FFA measurement, when the individual measurements of
FFAs were studied higher RSD values were observed between the replicates. RSDs of 21 %, 26 % and
45 % were obtained for C10:0, C12:0 and for C14:0, respectively. Milk yielded 33 % RSD for the total
FFA measurements, the individual FFA RSD measurements ranged from 6-37 % with 101 % RSD
being obtained for C18:3 butyl ester replicates. For the remaining samples (Cheese, EMCs, powders,
butter and yoghurt) the FABE method displayed consistency in its application. The lipid extraction,
solid phase extraction, derivitizing to butyl esters using automation, and the analysis of the butyl esters
were all performed with excellent precision (1-8 % RSD).
2
2
65
66
FABE verses Direct On-Column and FAME
267
268
269
The effect of the BF solution on column integrity was a potential concern because of its acidic nature,
3
5
because of experiences with the direct on-column method with formic acid . Figure 2 displays
chromatograms of two typical calibration standards at different stages of column lifetime (~120
1
1
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 12 of 21
270
271
272
273
274
275
276
277
278
279
280
281
282
analyses apart) using the FABE method. No retention time shift was evident. The only retention time
shift experienced was after routine maintenance of the instrument (septa and liner replacement, column
cutting) performed every 200 injections. The lack of any significant adverse impact of the BF solution
3
on the column phase was likely a result of the split ratio (1:50) and low injection volume (0.7 µl). The
butyl ester method is significantly more robust than the direct on-column method. The issue of sample
carryover was a major drawback of the direct on-column method, necessitating the employment of
formic acid blanks between every injection, which exacerbated column degradation, retention time
shifts and increased overall analysis times. It was evident from preliminary analysis butyl esters did not
yield any sample carryover, even when handling samples with very high concentrations of FFAs.
Figures S-3 and S-4 show the chromatogram of a blank injection post a 700 ppm FFA calibration mix
after the direct on-column method and FABE method, respectively. FFA accumulative column phase
absorption is clearly evident in the direct on-column method with no peaks present in the FABE
method.
2
2
2
2
2
2
2
2
2
2
2
2
83
84
85
86
87
88
89
90
91
92
93
94
A main advantage of the direct on-column method over the FAME analysis was the resolution of all
5
fatty acid peaks including C4:0 . Also no additional reagent is required for the FFA extract prior to
analysis and therefore there are no resulting artefact peaks. For the FAME method methyl butyrate
eluted very close to the solvent peak and there was an artefact peak (5 min RT) believed to be
5
trimethylamine that interfered with analysis at low concentrations (<20 ppm) (Figure 3). In the case of
the FABE method (Figure 4) butyl butyrate elutes (8.8 min RT) after butanol (~7 min RT) despite the
fact that butanol is in excess for the butyl derivitization step; thus, no interference is evident. BF also
3
created artefact peaks (8.6 min and 8.9 min RT), that eluted very close to butyl butyrate but did not
interfere with quantification. For validation the LOD was reported at 3 times signal to noise and LOQ at
10 times signal to noise. At these thresholds LOD was established at 8 ppm and LOQ at 20 ppm for
C4:0 for the FABE method. This was also compared against serial dilutions of a butyl ester sample and
it was found that quantifiable results were achieved as low as 5 ppm (Figure S-5), despite these
1
2
ACS Paragon Plus Environment

Page 13 of 21
Journal of Agricultural and Food Chemistry
2
2
2
95
96
97
responses being below the LOQ threshold (Table S-5). Thus, quantifiable results for the butyl esters can
be obtained at low concentrations, in contrast to the FAME method where the C4:0 methyl ester could
not be reliably quantified below 20 ppm.
2
2
98
99
FABE as an Alternative for Determination of FFAs in Dairy
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
00
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
The FABE method overcame the significant limitations of the direct on-column method relating to FFA
accumulation onto the column phase and column phase degradation allowing for a more robust analysis
with stable retention times and longer column life. The analysis of FFAs as esters seems to be a logical
solution to issues of sample carry over and column degradation. A further advantage is the ability to
incorporate automation into the method to reduce labor, human error and solvent usage. In relation to
the direct comparison of FABE and FAME, butyl esters are much more suitable than methyl esters for
the analysis of short chain fatty acids in dairy samples because of their decreased volatility, particularly
evident in the determination of C4:0. Even when using higher split ratios and lower injection volumes
the LOD and LOQ thresholds of the FABE method were similar to the FAME method. The application
of the FABE method proved suitable for the majority of samples in this study. However, there remains
scope to improve the extraction protocol, or add additional steps, to allow for analysis of samples such
as milk and ice cream because of the integrity of the milk fat globule membrane. Also consideration
needs to be given to the level of fat content, degree of lipolysis and sample matrix, as highlighted by the
EMC, butter and processed cheese samples. Overall the FABE approach is robust, sensitive, accurate
and precise and is a more suitable alternative to existing methods for determining the range of FFAs
found in dairy products.
3
16
3
3
17
18
ASSOCIATED CONTENT
Supporting Information
1
3
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 14 of 21
3
3
3
3
3
19
20
21
22
23
Chromatographic plots of on-column, FAME and FABE analysis. Tables displaying some of the
validation data for the FABE method, along with results of the various dairy samples comparing the
direct on-column, FAME and FABE methods.
The Supporting Information is available free of charge on the ACS Publications website.
3
24
AUTHOR INFORMATION
3
3
25
26
Corresponding Author
*Email: kieran.kilcawly@teagasc.ie
3
27
Author Contributions
3
3
3
3
3
3
28
29
30
31
32
33
The manuscript was written through the contribution of authors †‡ and †* and reviewed by ‡. All
authors have given approval to the final version of the manuscript. Authors †*/ and ‡ designed the
experiment and author †‡ undertook the study.
Notes
The authors would like to thank Agilent Technologies Inc. (A Delaware Corporation, 2850 Centerville
Road, Wilmington, Delaware 19808 USA) for the support of this study.
3
34
REFERENCES
335
336
337
338
339
340
341
(1) Mannion, D. T.; Furey, A.; Kilcawley, K. N. Free fatty acid quantitfaction in dairy products. Int.
J. Dairy Technol. 2016, 69, 1-12.
(2) Fox, P. F.; Wallace, J. M. Formation of flavor compounds in cheese. In Advances in Applied
Microbiology, Saul, L. N.; Allen, I. L., Eds.; Academic Press, 1997, pp 17-85.
(3) Martínez-Castro, I.; Alonso, L.; Juárez, M. Gas chromatographic analysis of free fatty acids and
glycerides of milk fat using tetramethylammonium hydroxide as catalyste. Chromatographia 1986, 21,
37-40.
1
4
ACS Paragon Plus Environment

Page 15 of 21
Journal of Agricultural and Food Chemistry
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
(4) De Jong, C.; Badings, H. T. Determination of free fatty acids in milk and cheese procedures for
extraction, clean up, and capillary gas chromatographic analysis. J. High Resolut. Chromatogr. 1990,
13, 94-98.
(5) Mannion, D. T.; Furey, A.; Kilcawley, K. N. Comparison and validation of 2 analytical methods
for the determination of free fatty acids in dairy products by gas chromatography with flame ionization
detection. J. Dairy Sci. 2016, 69, 5047-5063.
(6) Iverson, J.; Sheppard, A. Butyl ester preparation for gas-liquid chromatographic determination of
fatty acids in butter. J.- Assoc. Off. Anal. Chem. 1977, 60, 284.
(7) Gander, G. W.; Jensen, R. G.; Sampugna, J. Analysis of milk fatty acids by gas-liquid
chromatography. J. Dairy Sci. 1962, 45, 323-328.
(8) Kim Ha, J.; Lindsay, R. C. Method for the quantitative analysis of volatile free and total branched-
chain fatty acids in cheese and milk fat. J. Dairy Sci. 1990, 73, 1988-1999.
(9) Attaie, R.; Richter, R. L. Formation of volatile free fatty acids during ripening of Cheddar-like
hard goat cheese. J. Dairy Sci. 1996, 79, 717-724.
(10) Pavia, M.; Trujillo, A. J.; Sendra, E.; Guamis, B.; Ferragut, V. Free fatty acid content of
Manchego-type cheese salted by brine vacuum impregnation. Int. Dairy J. 2000, 10, 563-568.
(11) Buffa, M. N.; Guamis, B.; Pavia, M.; Trujillo, A. J. Lipolysis in cheese made from raw,
pasteurized or high-pressure-treated goats’ milk. Int. Dairy J. 2001, 11, 175-179.
1
5
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 16 of 21
TABLES
Table 1. Validation data for the fatty acid butyl ester method. Displayed are the linear calibration, limits of detection
LOD), limits of quantification (LOQ), intra-day and inter-day relative standard deviation (RSD) and relative error results.
(
Linear
Range
Intra-
Day
RSD (%) RSD (%)
Inter-
Day
Relative
Error
(%)
Butyl
ester
Correlation Coefficient
LODa
(ppm)
LOQa
(ppm)
b
c
d
line Equation
2
(R )
(ppm)
C4:0
C6:0
y = 0.9022x + 0.0139
y = 1.1366x - 0.0203
y = 1.3231x - 0.0643
y = 0.9868x - 0.0016
y = 1.143x - 0.0109
y = 1.1947x + 0.006
y = 1.0876x - 0.0205
y = 1.1067x - 0.0021
y = 1.068x - 0.0064
y = 1.0478x - 0.007
y = 1.0616x - 0.0064
20-700
20-700
20-700
20-700
20-700
20-700
20-700
20-700
20-700
20-700
20-700
0.9996
0.9995
0.9964
0.9999
0.9998
0.9978
0.9996
0.9999
0.9999
0.9999
0.9999
8
8
8
5
5
5
5
5
5
5
5
20
20
20
15
15
15
15
15
15
15
15
2.3
1.9
4.7
1.8
1.9
5.6
2.3
2.0
0.9
2.1
1.3
4.9
5.1
5.1
4.4
4.8
5.1
4.9
5.2
4.7
5.4
5.3
2.4
2.2
5.4
1.6
3
C8:0
C10:0
C12:0
C14:0
C16:0
C18:0
C18:1
C18:2
C18:3
2.4
3.4
3.9
3.3
4.2
4.3
aLOD stablished at 3 times signal to noise. LOQ established at 10 times signal to noise.
bAverage values obtained from ten samples (n=10) analyzed in a single day.
cAverage values obtained from seventy two samples (n=72) analyzed over 1 month.
dAverage values obtained from across 20-700 ppm concentration range.
1
6
ACS Paragon Plus Environment

Page 17 of 21
Journal of Agricultural and Food Chemistry
1
7
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 18 of 21
Table 2. Fatty acid butyl ester (FABE) analysis of dairy samples. Displayed are the average values of six samples (n=6). Included are the measured ppm, % recovery and % RSD
values of the ppm replicates.
Blue
Cheese
Paste
Processed
Cheese
Light
Mild
Special
Blue
Milk
Infant
Ice
Cream
Butter Nat.Cream Cream
Butter
Paste
Brie
Butter Yoghurt
Milk
Cheddar Cheddar Reserve Stilton Powder Formula
Powder
Powder
Paste
C4:0
C6:0
ppm
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
38
96
4
39
97
1
35
97
4
64
95
1
86
97
1
158
99
12
92
5
5
94
4
16
106
5
11
103
6
10
103
3
54
102
2
88
103
1
193
94
2
607
90
1
250
101
1
608
106
1
76
100
2
43
101
2
1960
101
1
12
96
10
7
99
12
6
97
13
5
97
10
6
11
100
7
4
97
6
3386
97
1942
98
6497
112
11
5552
110
13
5899
51
2
8
10
5
16
100
5
16
98
5
14
101
2
17
103
2
28
100
3
61
104
2
7
-
98
-
5
100
9
-
99
-
1524
98
8
967
105
12
2935
87
10
3514
88
2
3746
52
4
99
13
6
93
12
10
91
9
C8:0
15
101
4
17
97
4
12
96
8
18
96
4
22
98
3
38
97
2
-
7
101
8
-
1533
94
14
1017
112
10
3337
71
14
2188
78
6
2205
64
6
100
-
102
-
C10:0
C12:0
C14:0
C16:0
C18:0
C18:1
C18:2
C18:3
Total
37
99
2
37
96
2
34
98
3
44
99
1
60
99
3
81
101
3
-
95
-
8
-
99
-
3369
98
6
1904
95
9
7151
93
10
4245
103
5
4681
56
5
100
21
18
98
26
30
98
45
254
133
14
270
165
9
55
99
2
53
92
3
46
100
4
62
100
1
81
100
1
110
104
5
12
96
8
9
97
2
4
98
7
4569
91
6
2125
98
14
7108
89
6
4962
95
3
5256
46
3
97
14
9
96
9
140
97
2
139
76
11
386
46
9
98
97
4
170
103
1
213
94
2
305
115
5
26
97
10
90
87
6
10
97
2
8
96
4
9931
75
15
5457
102
11
17960
70
13
17389
76
8
14995
56
5
389
94
2
274
100
4
474
100
2
577
88
1
811
135
5
69
90
2
89
93
7
38
96
23
27
95
18
10
94
37
9
23364
101
5
15471
86
8
46728
29
6
33657
95
5
42694
54
4
161
96
3
129
85
8
114
103
3
181
97
2
205
93
1
288
106
5
49
94
4
33
98
3
61
95
10
12
96
14
12
95
6
4741
79
5
4296
82
10
9689
55
6
11411
65
1
11201
36
3
274
93
3
342
61
7
138
97
3
343
100
1
577
97
1
1265
170
4
61
88
9
49
90
2
47
103
21
11
97
8
12925
68
4
10551
72
10
26835
44
4
21449
55
1
19626
28
2
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
Recoverya
RSD (%)
ppm
32
95
4
62
95
3
27
98
9
33
121
5
50
94
3
176
105
4
12
96
8
29
92
3
1308
82
5
698
86
13
2457
58
4
1610
67
2
2526
38
2
95
-
22
96
3
23
96
7
19
98
6
27
96
3
31
95
2
59
95
3
10
101
13
296
94
5
9
103
2
-
-
10
93
101
90
98
33
526
87
5
252
89
12
895
63
10
743
71
8
518
40
2
100
-
101
-
1183
97
2
1246
85
7
814
99
3
1436
101
1
1936
96
1
3355
112
4
217
98
2
223
96
8
661
109
12
67206
88
3
44712
93
8
131622 106748 113377
70
Recoverya
RSD (%)
82
1
47
3
4
aCalculated from the addition of 0.5mg of each FFA (C4:0-C18:3) to sample (spiked sample) and measuring difference against un-spiked sample. For EMC samples 5mg of each FFA was added.
1
8
ACS Paragon Plus Environment

Page 19 of 21
Journal of Agricultural and Food Chemistry
Table 3. Summary of total FFA measurements obtained from on-column, fatty acid methyl ester (FAME) and fatty acid
butyl ester (FABE) analysis of dairy samples. Displayed are the average ppm values (n=6) and the % relative standard
deviation (RSD) in the measurements between the three methods.
Direct On-
Column
Sample
FAMEa
FABE
RSD (%)
a
Brie
1135
1104
1067
952
1183
1246
5.2
13.4
Processed Cheese
Light Cheddar
Mild Cheddar
969
1525
2289
3109
337
827
1627
2316
2856
333
814
1436
1936
3355
296
9.8
6.2
Special Reserve Cheddar
Blue Stilton
9.7
8.0
Milk Powder
7.1
Infant Formula
Butter
173
227
217
13.8
21.6
14.9
64.3
51.4
15.9
5.2
1492
182
1295
246
1960
223
Yoghurt
Ice Cream
323
176
661
Milk
290
286
90
Natural Butter Powder
Natural Cream Powder
Natural Cream paste
Natural Butter paste
Natural Blue Cheese paste
73652
48680
90949
49274
114969
98129
104376
67206
44712
131622
106748
113377
126615
106079
116085
6.9
4.6
5.5
aResults obtained from Mannion et al.5.
FIGURES
a
Figure 1. FABE chromatogram of special reserve Cheddar. Peaks are: 1 butanol; 2 butyl butyrate; x unknown; 3 butyl valerate ; 4 butyl
a
caproate; 5 butyl caprylate; 6 butyl caprate; 7 butyl undecanoate ; 8 butyl laurate; 9 butyl myristate; 10 butyl palmitate; 11 butyl
a
heptadecanoate ; 12 butyl stearate; 13 butyl oleate; 14 butyl linoleate; 15 butyl linolenate.
aInternal Standard
1
9
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 20 of 21
Figure 2. Two chromatograms of a typical butyl ester standard calibration mix analyzed at different stages in the column lifetime, approx.
20 analyses apart. Peak retention times remain stable.
1
Figure 3. Chromatogram of a FAME calibration standard
xUnknown artefact (likely TMA)
Figure 4. Chromatogram of a FABE calibration standard.
xUnknown
2
0
ACS Paragon Plus Environment

Page 21 of 21
Journal of Agricultural and Food Chemistry
Table of Contents artwork
OH
+OH
O
H+
'
R OH
R
OH
H
R
C
C
R
C
OH
OH
+O
Fatty Acid
Protonation of acid
Addition of alcohol
R'
Proton transfer
OH
+OH
H+
O
-
H
R
C
O
R
C
R
C
H
OR'
OR'
OR'
+
Ester
Deprotonation
Elimination of water
H O
2
2
1
ACS Paragon Plus Environment