Direct injection gas chromatographic/mass spectrometric analysis for denatonium benzoate in specific denatured alcohol formulations

Article abstract of DOI:10.1021/ac980093r

Direct injection GC/MS was investigated for the analysis of benzyldiethyl(2,6-xylylcarbamoylmethyl)ammonium benzoate (Bitrex), a quaternary ammonium salt, in various Canadian denatured alcohol formulations. Bitrex yielded predominantly a peak due to the neutral diethylamine derivative (I). The structure of I, elucidated by MS and NMR, is strongly related to that of the cation of Bitrex. Compound I was formed from Bitrex in the heated injector port of the GC via a decomposition reaction similar to Stevens rearrangement. The response of I was found to be dependent on the injector port temperature, and the optimal temperature was determined to be in the range 250-350°C. The GC/MS response of I in SIM mode was used to quantify Bitrex. The effects of the codenaturants sucrose octaacetate (SOA), diethyl phthalate (DEP), and camphor, which are present at much higher concentration than Bitrex in several formulations, were also investigated. The presence of SOA enhanced the response of the analyte considerably, while DEP and camphor had no significant effect. All standard curves of Bitrex (1-16 ppm) in different alcohol matrixes were fitted by second-order polynomial functions, with coefficients of determination (R²) routinely in the range 0.998-0.999. The analysis time was 18 min, and the within-run precision was <4%. The results of this study point to the potential of the GC/MS technique as a quantitative tool for Bitrex in various alcohol formulations.

Full text of DOI:10.1021/ac980093r

Anal. Chem. 1998, 70, 4389-4393
Direct Injection Gas Chromatographic/Mass
Spectrometric Analysis for Denatonium Benzoate
in Specific Denatured Alcohol Formulations
Lay-Keow Ng,*,† Michel Hupe´,† Jean Harnois,‡ and Andre´ H. Lawrence†
Research and Development Division and Excise Laboratory Division, Laboratory & Scientific Services Directorate,
Revenue Canada, Ottawa, Ontario, Canada K1A 0L5
matrixes, such as alcohol toilet preparations,1,2 rapeseed oils3 and
denatured alcohols.4 All these methods involve preconcentration
of Bitrex prior to chromatographic analysis. They are time-
consuming, require a large sample volume, and may suffer errors
inherent to the extraction process. Therefore, a direct approach
which does not involve any sample preparation for the determi-
nation of Bitrex in alcohol formulations is highly desirable.
Quaternary ammonium salts are thermally unstable. They
decompose at elevated temperatures to neutral molecules which
could be vaporized and subsequently analyzed by various separa-
tion techniques. Gas chromatography is an ideal analytical
technique for the determination of quaternary ammonium salts
because the heated injector port of the instrument allows in situ
decomposition of the salts. As such, transformation to neutral
molecules, vaporization, and analysis of the characteristic pyroly-
sates can be performed in a single step. Direct injector pyrolysis
has been previously applied to the analysis of quaternary am-
monium salts.5-7 For example, benzalkonium chlorides were
thermally degraded to alkyldimethylamines upon injection into a
gas chromatograph, and the pyrolysis products were used to
determine the homologous composition of the salt. Similarly,
Bitrex could be assayed by analyzing the nitrogen-containing
pyrolysates. In several Canadian formulations containing Bitrex,
other denaturants such as diethyl phthalate (DEP), camphor, and
sucrose octaacetate (SOA) can also be present. The concentration
levels of these components are usually 30-400 times higher than
that of Bitrex. Trace analysis of Bitrex in the presence of large
amounts of other denaturants requires a highly selective and
sensitive detection technique.
Direct injection GC/ MS was investigated for the analysis
of benzyldiethyl(2,6-xylylcarbamoylmethyl)ammonium ben-
zoate (Bitrex), a quaternary ammonium salt, in various
Canadian denatured alcohol formulations. Bitrex yielded
predominantly a peak due to the neutral diethylamine
derivative (I). The structure of I, elucidated by MS and
NMR, is strongly related to that of the cation of Bitrex.
Compound I was formed from Bitrex in the heated injector
port of the GC via a decomposition reaction similar to
Stevens rearrangement. The response of I was found to
be dependent on the injector port temperature, and the
optimal temperature was determined to be in the range
2 5 0 -3 5 0 °C. The GC/ MS response of I in SIM mode
was used to quantify Bitrex. The effects of the codena-
turants sucrose octaacetate (SOA), diethyl phthalate (DEP),
and camphor, which are present at much higher concen-
tration than Bitrex in several formulations, were also
investigated. The presence of SOA enhanced the re-
sponse of the analyte considerably, while DEP and
camphor had no significant effect. All standard curves
of Bitrex (1 -1 6 ppm) in different alcohol matrixes were
fitted by second-order polynomial functions, with coef-
ficients of determination (R2 ) routinely in the range
0 .9 9 8 -0 .9 9 9 . The analysis time was 1 8 min, and the
within-run precision was <4 %. The results of this study
point to the potential of the GC/ MS technique as a
quantitative tool for Bitrex in various alcohol formulations.
Bitrex, chemically known as benzyldiethyl(2,6-xylylcarbam-
oylmethyl)ammonium benzoate, is a highly bitter compound. It
is among the common substances used for denaturing alcohol.
The Canadian Government Regulations specify the level of Bitrex
in several formulations of specially denatured alcohol to be 7 ppm.
The rate of duty on the denatured alcohols depends on whether
the specifications are met. Various methods based on thin-layer
chromatography and high-performance liquid chromatography
have been applied to the determination of Bitrex in different
The objective of this study was to show that direct injection
GC/ MS operated in a selected ion monitoring mode (SIM) is a
feasible approach for the determination of Bitrex in various alcohol
formulations. Quantification of Bitrex was based on the deter-
mination of a rearranged pyrolysis product (I). The determination
of the structure of I by MS and NMR spectroscopy, and the
analytical characteristics of the technique, are described.
* To whom correspondence should be addressed. E-mail: rlkng@revcan.ca.
Fax: 613-952-7825.
(3) Damon, C. E.; Pettitt, B. C., Jr. J. Chromatogr. 1 9 8 0 , 195, 243-249.
(4) Kovar, J.; Loyer M. J. Liq. Chromatogr. 1 9 8 4 , 7, 2163-2177.
(5) Ng, L.-K.; Hupe´, M.; Harris, A. G. J. Chromatogr. 1 9 8 6 , 351, 554-559.
(6) Barry, B. W.; Saunders, G. M. J. Pharm. Sci. 1 9 7 1 , 60, 645-647.
(7) Lopez, A. F.; Peralta de Ariza, M. T.; Orio, O. A. J. High-Resolut. Chromatogr.
1 9 8 9 , 12, 503-504.
† Research and Development Division.
‡ Excise Laboratory Division.
(1) Glover, M. J.; Blake, A. J. Analyst 1 9 7 2 , 97, 891-896.
(2) Sogden, C.; Mayne, T. G.; Loscombe, C. R. Analyst 1 9 7 8 , 103, 653-656.
S0003-2700(98)00093-6 CCC: $15.00 © 1998 American Chemical Society
Published on Web 09/04/1998
Analytical Chemistry, Vol. 70, No. 20, October 15, 1998 4389

Figure 1. Major pyrolysis pathways of Bitrex in the GC injector port.
EXPERIMENTAL SECTION
in 32 768 data points over a spectral width of 16.92 kHz, with a
90° pulse width of 10 µs and a relaxation delay of 5 s. A double-
quantum-filtered COSY 2D NMR spectrum was collected in a 1024
× 512 matrix with a spectral width of 2.026 kHz.
Chemicals. Bitrex (98%), ethyl stearate (99%), diethyl phtha-
late (99%), sucrose octaacetate (99%), and camphor (96%) were
purchased from Aldrich Chemical Co. (Milwaukee, WI) and used
as received, and 95% (v/ v) ethyl alcohol was supplied by Les
Alcools de Commerce Co. (Quebec, Canada).
P reparation of Compound I. About 35 mg of Bitrex was
placed in a Pyrex tube, 4 mm i.d. and 75 mm long. Glass wool
was introduced from both ends of the tube to confine Bitrex to a
5-mm section of the tube. A small heating block at 250 °C was
mounted over the tube where Bitrex was contained while N2 was
being passed through the tube at 30 mL/ min. The volatile
products of the pyrolysis, such as benzoic acid, benzyl benzoate,
and lidocaine, were flushed away using a N2 stream. The less
volatile components were dissolved in ∼2 mL of CH2Cl2, and the
solution was concentrated to ∼100 µL under N2 before being
added to a SPE C-18 cartridge (Waters, MA). Compound I was
eluted using ∼2 mL of CH2Cl2. The eluent was evaporated to
dryness under N2 and redissolved in deuterated CHCl3 for GC/
MS and NMR analysis.
Gas Chromatography/ Mass Spectrometry Analysis. Quan-
tification was carried out using a Hewlett-Packard GC/ MSD model
6890 system. The system included a 6890 GC with electronic
pressure control, the MSD, an autosampler, and a Chemstation.
The analytical column was a DB5ms (J&W Scientific, Folsom, CA),
30 m long, 0.25 mm i.d., and 0.5 µm film thickness. Helium was
the carrier gas at a constant flow of 1.1 mL/ min. A 1-µL aliquot
of the alcoholic solution was introduced into the GC/ MS system
in the splitless mode, and the injector was purged after 1 min.
The injector glass sleeve was a 4-mm-i.d. borosilicate single-taper
type liner with a small glass wool plug at the tapered end. The
temperature program was as follows: initial 85 °C held for 1 min;
30 °C/ min to 300 °C, held for 9.83 min. The total run time was
18 min. The injector and interface temperatures are 275 and 290
°C, respectively. The mass-selective detector was operated in the
SIM mode, with a solvent delay of 8 min. Two groups of ions,
m/ z 176 and 233, and m/ z 88 and 101, were used to monitor Bitrex
and ethyl stearate, the internal standard (ISTD), respectively. The
dwell time for each ion was 150 ms. The ions at m/ z 176 and 88
were used for quantifying Bitrex and ISTD, respectively.
Qualitative analysis was also performed on a Varian Saturn III
GC/ MS system which was equipped with a Star 3400CX GC, an
8200 CX autosampler, an Alltech EPC 1000 electronic pressure
control system, the Saturn III Ion Trap, and a Chemstation. GC
column and conditions were the same as described above. The
trap was kept at 280 °C and was scanned from 40 to 500 amu at
a rate of 1 scan/ s.
RESULTS AND DISCUSSIONS
Structure of the Major P yrolysis P roduct. Quaternary
ammonium compounds are known to decompose thermally by
cleavage of a C-N bond at the quaternary nitrogen with the
formation of a tertiary amine and another neutral compound.5,7
In general, debenzylation is favored over dealkylation (Figure 1,
route b). In this work, in addition to the expected pyrolysis
products, lidocaine and benzyl benzoate, the thermal decomposi-
tion of Bitrex in the GC injection port gave a major product with
a retention time of 10.25 min (Figure 2) corresponding to the
diethylamine derivative I, and benzoic acid (Figure 1, route a).
Bitrex contains a strong electron-withdrawing 2,6-xylycarbam-
oyl group attached to one of the methylene carbons around the
quaternary nitrogen atom. Compounds of similar structure have
been reported to undergo Stevens rearrangement in the presence
of a base to give rearranged tertiary amines.8 The reaction
involves loss of a labile proton from the methylene group adjacent
Nuclear Magnetic Resonance Analysis. 1H and 13C NMR
spectra of I and Bitrex were recorded at 25 °C on a JEOL 270
spectrometer operating at 270 MHz and using CDCl3 as solvent.
1
The H NMR spectra were collected in 16 384 data points over a
4.052-kHz spectral width, with a 45° pulse width of 5 µs and a
relaxation delay of 1 s. The 13C NMR spectrum of I was acquired
(8) March, J. Advanced Organic Chemistry, 2nd ed.; McGraw-Hill Book Co.: New
York, 1977; Chapter 18.
4390 Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

Figure 2. Total ion chromatogram resulting from injection of an
alcohol solution of Bitrex into GC/MS in full-scan mode: (1) benzoic
acid, (2) benzyl benzoate, (3) lidocaine, and (4) compound I.
to the 2,6-xylylcarbamoyl moiety with subsequent migration of a
substituent on the quaternary nitrogen. The formation of I and
benzoic acid from Bitrex in the heated GC injector port can be
explained via a similar rearrangement. In this case, the benzoate
ion serves as the base, and the benzyl substituent is the migrating
group.
The diethylamine derivative I was characterized by MS and
NMR. The EI mass spectrum of I obtained from a quadrupole
mass filter shows a base peak at m/ z 176, with small peaks at
m/ z 148 and 233, while the molecular ion (m/ z 324) was not
observed (Figure 3a). The spectrum obtained from ion trap
spectrometry is very similar to that in Figure 3a, except that, in
addition, a distinct peak at m/ z 325 was detected (Figure 3b).
This is the [M + 1]+ ion which is often formed in ion storage
mass spectrometry.9 This experiment provides evidence that I
has a molecular weight of 324. The fragments at m/ z 176 and
233 are ions resulting from R-cleavage with loss of 2,6-xylycar-
bamoyl and benzyl groups, respectively, when the tertiary nitrogen
is the charge-initiation site. R-Cleavage at the carbonyl group
yields the peak at m/ z 148 when I is ionized at the carbonyl
oxygen.
Figure 3. Mass spectra of compound I obtained from (a) quadrupole
mass spectrometer and (b) ion-trap mass spectrometer.
Compound I had the same retention time and mass spectrum
as the major peak in the total ion chromatogram obtained from
direct injection of Bitrex, but its 1H NMR spectrum (Figure 4)
was entirely different from that of Bitrex. The assignments of
peaks are shown in Table 1. The methine proton (a) and the
benzylic protons (b and c) constitute a spin system. The two
benzylic protons are not chemical shift equivalent since they are
attached to an asymmetric carbon. Each proton is split by the
other and by the neighboring methine proton, resulting in a four-
peak pattern. The methine proton is coupled almost equally to
both benzylic protons (Jab ) 6.6 Hz, Jac ) 5.6 Hz) and, therefore,
appears as a triplet (doublet of doublet) at δ 3.84. The methylene
protons (d and e) on the ethyl groups appear as strongly coupled
spin systems at δ 2.54 and δ 2.72. This is in agreement with
previously reported nonequivalence of the methylene protons,
Figure 4. COSY NMR of compound I. Peaks marked with an
asterisk are due to impurities.
even if they are more than one atom away from the chiral center.10
The shift difference between the two methylene protons is small
(48.6 Hz) compared with a geminal coupling constant of 13.4 Hz.
Consequently, the AB pattern from the geminal coupling is quite
distorted; the inner peaks are strong, while the outer peaks are
weak. Each methylene proton is also split by the methyl protons
at δ 1.1 with slightly different coupling constants, 7.0 and 6.8 Hz.
Some of the peaks coincide, giving rise to six peaks. The singlet
at δ 7.05 is assigned to the aromatic protons of the 2,6-
xylycarbamoyl group, since the three aromatic protons of lidocaine,
(9) Pannell, L. K.; Pu Q.-L.; Fales, H. M.; Mason, R. T.; Stephenson, J. L. Anal.
Chem. 1 9 8 9 , 61, 2500-2503.
(10) Silverstein, R. M.; Bassler, G. C.; Morrill, T. C. Spectrometric Identification
of Organic Compounds, 4th ed.; John Wiley & Sons: New York, 1981; Chapter
4, p 208.
Analytical Chemistry, Vol. 70, No. 20, October 15, 1998 4391

Table 1. NMR Data for Compound I
proton
(type, no of protons)
chemical shift δ,
ppm (multiplicity)
coupling
constant J, Hz
a (methine, 1)
3.84 (d of d)
3.48 (d of d)
2.91 (d of d)
Jac ) 6.6, Jab ) 5.6
Jab ) 5.6, Jbc ) 13.7
Jac ) 6.6, Jbc ) 13.7
b (methylene, 1)
c (methylene, 1)
d,e (methylene, 4)
2.54 (m), 2.72 (m) Jde ) 13.4, Jgd ) 7.0,
Jge ) 6.8
f (methyl, 6)
2.2 (s)
Figure 5. Effects of matrix on the GC/MS response of Bitrex.
g (methyl, 6)
aromatic
1.1 (t)
7.05 (s)
Jgd ) 7.0, Jge ) 6.8
(2,6-xylylcarbamoyl, 3)
aromatic (benzylic, 5)
7.2-7.4 (m)
carbon-13 (type)
chemical shift δ, ppm
1 (methyl)
14.3
45
31.5
66
172
2 (methylene)
3 (methylene)
4 (methine)
5 (carbonyl)
6 (methyl)
18.6
aromatic (2,6-xylylcarbamoyl)
aromatic (benzylic)
127-133.5
128-130
which contains the same aromatic moiety, also appear as a singlet
at δ 7.07. The COSY spectrum (Figure 4) confirms the connectiv-
ity of the coupled nuclei described above. Peak areas of the
absorptions of the proton nuclei are in agreement with their
relative abundances. The chemical shifts of carbon-13 nuclei,
shown in Table 1, are also consistent with the structure.
Analytical Characteristics. Compound I was used for
quantification of Bitrex because it had a structure highly related
to Bitrex, and it gave the strongest GC/ MS response of all the
products of pyrolysis. Since I was formed by heat-induced
rearrangement/ decomposition of Bitrex in the injector, the
response may vary considerably with the injector temperature.
The effect of injector port temperature over the range 200-360
°C was investigated, using a 7 ppm solution of Bitrex in alcohol.
The analysis was carried out in SIM mode, and the ion at m/ z
176 was used for quantification. At each temperature, four
replicates were analyzed. The average response of I rose sharply
from 200 to 250 °C and leveled off thereafter. In all the studies
described below, the injector temperature was set at 275 °C. The
replicates of the same sample were run nonconsecutively. Ethyl
stearate, the internal standard (ISTD), was added to each test
solution at a concentration of 6 ppm. A stable isotope-labeled
Bitrex would be an ideal ISTD; it was, however, not used in this
study because of the unavailability of the labeled compound.
The precision was determined by performing five replicate
analyses of each of the three alcohol solutions containing 7 ppm
of Bitrex in different matrixes. The relative standard deviations
(RSDs) of the measurements of the response ratio (I/ ISTD) were
typically <4%. In all cases, no significant carryover of the analyte
and the ISTD was observed in the blanks that were run after the
Bitrex solutions.
Figure 6. Standard curves of Bitrex in different matrixes. M1,
absence of codenaturants; M2, 1000 ppm camphor; M3, 210 ppm
SOA, 400 ppm camphor, and 2800 ppm DEP; and M4, 980 ppm SOA.
Several Canadian alcohol formulations also contain other
substances as denaturants in addition to Bitrex. These codena-
turants, camphor, DEP, and SOA, which eluted at 4.6, 6.6, and
14.0 min, respectively, were well separated from I at 10.25 min
and the ISTD at 8.8 min. Although they did not interfere with
peak integration, they might have an effect on the quantification
of Bitrex. The possible matrix effects of camphor, DEP, and SOA
on the response of I were determined using a series of standard
alcohol solutions containing 16 ppm of Bitrex and one of the other
denaturants in varying concentrations covering the typical range
found in Canadian denatured alcohol formulations. As shown
clearly in Figure 5, the response ratio of I to ISTD was not
significantly affected by the presence of camphor (500-2000 ppm)
and DEP (500-3500 ppm), as confirmed by ANOVA tests (P >
0.1). A Bitrex solution containing 2000 ppm of camphor and 3500
ppm of DEP also gave similar results. However, the response
ratio of I to the ISTD increased considerably with the concentra-
tion of SOA (500-2000 ppm). SOA was analyzed separately, and
the results confirmed the absence of any impurity that coeluted
with I and would interfere with the integration.
The effects of SOA were further illustrated by the standard
curves of Bitrex (Figure 6). Seven Bitrex standard solutions,
covering the range 1-16.3 ppm, in typical denatured alcohol
matrixes of Canadian formulationssM1 (absence of any other
denaturants), M2 (1000 ppm camphor), M3 (210 ppm SOA, 400
ppm camphor, and 2800 ppm DEP), and M4 (980 ppm SOA)s
4392 Analytical Chemistry, Vol. 70, No. 20, October 15, 1998

and matrix blanks were analyzed in triplicates. The averaged
response ratios of I/ ISTD were plotted against Bitrex concentra-
tion. The standard curves, when fitted with linear regression,
yielded y-residuals which showed a marked trend. However,
second-order polynomial functions provided satisfactory fits. The
coefficients of determination, R2, were invariably in the range
0.998-0.999, and the intercepts were <(4% of the response of
Bitrex at a concentration of 7 ppm. The 95% confidence bands of
the standard curves of matrixes M3 and M4 did not enclose the
standard curve of M1. These results are expected because SOA
was present in both M3 and M4, while M1 contained only Bitrex.
On the other hand, the fitted curves of M1 and M2 were not
significantly different since SOA was not present in M2. In all
cases, the area counts of ISTD did not vary significantly (P >
0.05) across the calibration solutions; therefore, any change in
response ratio was related to the response of I. In light of these
findings, for accurate prediction of Bitrex in a sample containing
SOA, it is necessary to match the matrix of the calibration
solutions with that of the sample. The nonlinearity of the standard
curves indicates that the response factor of I increased with the
concentration of Bitrex. It is possible that the decomposition of
Bitrex in the injector port is an intermolecular process. The
mechanism of pyrolysis is presently under investigation.
The nonlinear data can, however, approximate a linear fit if
the range is suitably small. For the most accurate measure-
ments,11 the point corresponding to the specified level of 7 ppm
is bracketed by closely spaced calibration standards. Measure-
ments in the range of 5.6-8.4 ppm of Bitrex were practically linear
(R2 ) 0.999), although the line did not pass through the origin (P
) 0.001). The 95% confidence interval of Bitrex at 7 ppm was
(1.5% of the concentration.
CONCLUSIONS
Direct injection GC/ MS (SIM) has been demonstrated to be
a feasible technique for analyzing Bitrex in various alcohol
formulations. The predominant pyrolysate, I, has a structure
highly related to Bitrex and can serve as an analyte representative
of Bitrex. The valuable feature of this approach remains the ability
of GC/ MS to quantify Bitrex in the presence of large amounts of
other denaturants, such as camphor, DEP, and SOA, in alcohol
solutions without any sample preparation.
ACKNOWLEDGMENT
We acknowledge P. Neudorfl, D. Perreault, J. Hardy, and K.
Smith for technical support. We particularly thank T. C. Leung
for his interest and support during this study.
Received for review January 29, 1998. Accepted July 10,
1998.
(11) Mitchell, D. G.; Mills, W. N.; Garden J. S. Anal. Chem. 1 9 7 7 , 49, 1655-
1660.
AC980093R
Analytical Chemistry, Vol. 70, No. 20, October 15, 1998 4393