Fuel 82 (2003) 23–33

www.fuelfirst.com

Determination of benzene and total aromatics in commercial gasolines

using packed column GC and NMR techniquesq
A.P. Singh, S. Mukherji, A.K. Tewari, W.R. Kalsi, A.S. Sarpal*
Indian oil corporation limited, R&D Center, Faridabad 121007, India
Received 3 December 2001; accepted 31 May 2002; available online 19 July 2002

Abstract
A simple, accurate and rapid method has been developed for the estimation of benzene and total aromatics (0.5 – 50% w/w) including
heavier aromatics (C8, C9, and C10) in commercial gasoline using packed column GC and NMR techniques. The benzene content can be
estimated as low as 0.1% w/w. The response of a flame ionisation detector (FID) to each major aromatic group in gasoline was calculated
using internal standard. The results have been compared with the NMR and standard ASTM D5580 methods. The limitations of NMR and
GC techniques for the estimation of total aromatics particularly in reformulated gasoline containing oxygenates have been discussed and
attempts have been made to overcome problems associated with the analysis. The results obtained by both the techniques for a number of
commercial gasoline samples containing olefins or free of olefins received from different refineries processing variety of crudes using
different refining technologies have shown excellent correlation. The 1H NMR method has a wider scope, convenient and fast, and also
applicable to heavier naphtha. The method can routinely be adopted for the quality control of commercial gasoline at refinery as well as
marketing terminals for monitoring benzene and total aromatic content. The time consumption for single run using gas chromatographic
technique is approximately 35 min. q 2002 Published by Elsevier Science Ltd.
Keywords: Benzene; Total aromatics; Gasoline; Packed column GC; 1H NMR

1. Introduction properties. The determination of benzene and the aromatic

The quality of the gasoline products is generally
controlled by blending different refinery streams which in
turn is reflected by engine performance. However, in recent
years, compositional data are also being considered
significantly in terms of their benzene and total aromatic
content. Different regions of the world have their own
specifications namely EN-228 in Europe, ASTM D4814 in
USA, JIS K 2202 in Japan and IS 2796 in India. The details
of proposed changes in gasoline quality in Europe from the
year 2000 (Euro II to V) onwards include additional controls
on vapour pressure, gasoline composition in terms of
benzene, aromatics and olefins as well as sulphur levels.
The reduction in gasoline vapour pressure is required to
control volatile organic compounds (VOC) emission,
benzene, aromatics and olefins contents to reduce toxic
materials. In USA, the gasoline quality control is done via
empirical equations relating vehicle emissions to fuel
* Corresponding author. Fax: þ 91-129-286-221.
E-mail address: sarpalas@iocrnd.stpn.net (A.S. Sarpal).
q
Published first on the web via Fuelfirst.com—http://www.fuelfirst.com
0016-2361/03/$ - see front matter q 2002 Published by Elsevier Science Ltd.
PII: S 0 0 1 6 - 2 3 6 1 ( 0 2 ) 0 0 1 5 7 - 6
content of gasoline are necessary to assess product quality
and to meet new fuel regulations.
The regulation limiting the concentration of benzene and
the total aromatic content of finished gasoline has been a
concern to the regulating authorities. In order to reduce the
ozone reactivity and toxicity of automotive evaporative and
exhaust emissions, benzene and total aromatic content
measurement in gasoline has been the focus of considerable
attention, and many standards and non-standard methods
have been developed. The method ASTM D1319 [1]
fluorescent indicator absorption (FIA) is the oldest method
for the estimation of total aromatics which provides total
aromatic content. The method does not provide reasonable
precision, if total aromatic content is at a level of
20 – 25 vol%. The EPA originally specified a multicolumn
GC method, ASTM D3606-92 [2] for benzene and toluene
analysis in gasoline and aviation gasoline samples. It has
been mentioned that the method is not applicable to
gasolines containing ethanol and methanol. The method
ASTM D4420 [3] describes the determination of benzene,
toluene, C8, C9 and heavier aromatics and total aromatics in
finished motor gasoline. The method involves the analysis
24 A.P. Singh et al. / Fuel 82 (2003) 23–33
using capillary column. It has not been mentioned whether
ASTM D4420 method is applicable to gasoline containing
oxygenates. ASTM D5134 [4] standard test method for
detailed analysis of petroleum naphthas through n-nonane
which is a single column capillary GC method for
estimating PIONA at each carbon number and limits the
analysis up to n-C9 range. The method provides detailed
component analysis along with benzene and total aromatic
content. The analysis time for single run for reformate
sample is more than 90 min which is another limitation if
only benzene and total aromatics are required.
The major problems associated with the multicolumn gas
chromatograph are high landed cost, time consumption and
operational complexities and also these sophisticated
equipments require extensive maintenance. The ASTM
D5443 [5] which is multicolumn and multiswitching valve
method provides the determination of paraffins, naphthenes
and aromatic by carbon number for sample having boiling
points lower than 200 8C. The method does not provide
individual aromatics except benzene and toluene and the
total analysis time is 70 min. The ASTM D5580 [6] is the
most important and versatile method which can be used for
gasoline containing oxygenates (alcohols and ethers)
employing multidimensional GC with packed/capillary
polar and non-polar columns. The potential oxygenates
found in gasoline do not interfere with the analysis of
benzene and other aromatics by this test method. The
method requires stepwise analyses. In the first step benzene
and toluene are quantified. In the second step analysis of
other aromatics, viz. ethyl benzene, o-xylene, m-p-xylene
and C9 and heavier aromatics content is completed. The
total analysis time is approximately 45 min for a single
sample. The IP425 [7] is concerned with the determination
of benzene content of petrol by gas chromatography which
also utilizes add on column switching technique and does
not provide total aromatics content.
The ASTM D5769 [8] is a targeted component analysis
method. The total aromatics content is calculated by
summing the concentration of individual aromatic com-
pounds. The method does not give accurate analysis if the
sample contains aromatics, which are not indicated in the
target list. In this method numerous calibration curves are
needed as it utilizes extracted ion abundance method for
quantitation which is less reliable than FID. Poole et al. [9]
have described the use of liquid organic salt column for
separating benzene and toluene in gasoline.
Selection of stationary liquid phase plays vital role for
achieving a particular separation. The polarity (solvent
strength) and selectivity are the important characteristics of
a stationary phase. The polarity and selectivity character-
istics are temperature-dependent [10]. The column selectiv-
ity is influenced by the nature of stationary phase and
temperature of column along with the choice of support
material whether treated or untreated. Separation depends
on boiling point differences, structure of components and
molecular weights. The polarity has been used to tabulate
various liquid stationary phases [11,12]. The increased
polarity of stationary phases helps to improve separation by
retaining solvents with polar functional groups longer than
non-polar solvents by specific polar interactions. The liquid
phases should have low viscosity and high wetting ability to
form uniform film on the support material/column wall [13].
The standard procedures [3 – 6] have recommended station-
ary phases such as TCEP [1,2,3-Tris (2-cyano ethyoxy)
propane], OV-275 (cyanosilicones) for analysis of benzene,
toluene, xylene (BTX) and total aromatics in gasoline range
petroleum products.
Over the years, a number of other chromatographic and
spectroscopic methods for gasoline analysis have been
developed. SFC and multidimensional methods like heart
cut GC, HPLC [14] LC/GC, SFC/GC [15,16] and GC-FTIR
[17] and NMR [18 – 22] are being utilized by different
workers for gasoline analysis. Recently a novel method for
quantitative determination of BTX and total aromatic
compounds in gasoline by comprehensive two-dimensional
gas chromatography (GC £ GC) has been developed by
Frysinger et al. [23]. The total aromatics, paraffins,
naphthenes and olefins (PONA) content of straight run
and cracked gasoline range products can conveniently be
estimated by NMR method based on average group
molecular weight concept. A direct method based on 1H
and 13C NMR spectroscopic technique for the estimation of
oxygenates (ethers and alcohols) and total aromatics has
been described [21,22].
There has been consistent and regular efforts put by many
workers to improve the ease, precision and accuracy of
gasoline analysis. Research workers are actively exploring
new tests for the determination of gasoline composition. All
the above described methods are either expensive or
cumbersome requiring dedicated instruments. In view of
above, a simple, accurate and fast method for the estimation
of benzene and total aromatic content has been developed
based on GC packed column with TCEPE as stationary phase,
FID detector. NMR technique has also been utilized for
comparison of the analytical results obtained by GC method.
2. Experimental
2.1. Gas chromatograph
All analyses were performed on Chemito 8610 GC
equipped with FID (flame ionisation detector). The data
were processed using a computer and printer using
Winchrom software.
2.2. Columns and conditions
A column packed with strongly polar phase tetra cyano
ethoxylated penta-erythritol (TCEPE) 15% coated over
gaschrom Q, (80-100) mesh (12 ft £ 1/8 in. dia.) stainless
steel, has been utilized for the analyses. The following
 A.P. Singh et al. / Fuel 82 (2003) 23–33
analytical conditions have been used:
injector temperature: 250 8C
detector temperature: 250 8C
oven program: 90– 120 8C
initial hold up time: 5 min
ramp rate: 6 8C/min
final hold up time: 25 min
carrier gas flow rate: 40 ml/min (helium)
sample size: 0.2 ml.
Another packed stainless steel (12 ft £ 1/8 in.) column
with highly polar TCEP stationary phase coated over
gaschrom Q (80 – 100) mesh was prepared for comparison
of test results with two stationery phases. The analytical
conditions were kept identical as described above and only
flow of carrier gas was higher (50 ml) in order to achieve
almost same resolution of benzene as obtained using
TCEPE column. The comparison of properties of TCEPE
versus TCEP is given in Table 1.
2.3. Standards
Six calibration blends containing benzene, toluene,
ethylbenzene, m-p-xylenes, o-xylene, 1,2,4-trimethylben-
zene and 1,2,3,5-tetramethylbenzene were prepared in
heptane/isooctane with concentrations spanning the
expected range for each in gasoline. n-Tridecane (n-C13)
was added at 4% w/w in each solution as internal standard
(IS). The internal standard was selected because it was not
present in commercial gasoline samples taken for study and
clearly separates from all gasoline components.
2.4. Samples
Total 90 samples were collected from different refineries
and terminals involved in processing the crudes by different
processes as conventional, hydrotreating, hydrocracking,
etc. These 90 samples were divided into three categories.
The first category was composed of different gasoline
blending stocks, i.e. straight run naphtha (SRN), reformate,
hydrocracker naphtha, cocker naphtha, fluid catalytic
cracking naphtha (FCC) and heavy naphtha. A second
category of samples consisted of commercial gasoline being
sold in the market. Third category of samples consisted of
Table 1
Comparison of properties of TCEPE versus TCEP
No. Property
1 Polarity
2 Solubility
3 Temperature (8C), min./max. limits
4 McReynolds constant (probe compound benzene)
5 Colour
25
selected commercial gasoline containing oxygenates such as
alcohols and ethers.
2.5. Calibration
The calibration blends were analysed as per conditions
described earlier. The calibration graphs were plotted
between the concentration of aromatics and the area ratio
of aromatic and IS. These reference samples were analysed
five times. Individual benzene, toluene, m-p-xylenes,
o-xylene, ethylbenzene, 1,2,4-trimethylbenzene and
1,2,3,5-tetramethylbenzene peaks were integrated and
quantified in terms of w/w% against internal standard in
accordance with ASTM D5580 [3] calculation procedure.
GC-FID calibration data has been presented in Table 2
which shows slope, intercept and correlation coefficient
(R 2). A correlation coefficient 0.99 shows the detector
linearity in the above working range for most of these
compounds. All but one of correlation coefficient values
exceeded the 0.990 specified in the ASTM D5580 method.
2.6. 1H NMR parameters
The 1H NMR spectra of 5% solution of the samples in
CDCl 3 (w/w) were recorded on a 300 MHz NMR
spectrometer. The spectral parameters are: number of
scans (NS) ¼ 64, relaxation delay (RD) ¼ 10 s, pulse
spectral width (PW) ¼ 19 ms.
In order to get reproducible and highly resolved
spectrum, the instrument was shimmed properly before
carrying out the sample recordings. This is necessary for
proper base line correction and obtaining narrow width of
the full signals and for proper integration. The full spectrum
was integrated thrice and average values of integration of
respective regions were taken for use in the equations for the
estimation of benzene and total aromatics. Since the integral
intensities of the aromatics region, especially benzene
signal, are smaller (1 –10%) compared to the total integral
intensity of the full region, the aromatic region was
amplified both horizontally (X-direction) and vertically
(Y-direction) to observe clear demarcation between benzene
and other aromatic regions. The integral values were
measured by both auto mode for higher values of benzene
(. 0.5%) and manually (benzene , 0.5% and
aromatics , 2.0% w/w) depending upon the level of the
benzene or total aromatics.
TCEPE TCEP
Highly polar Highly polar
Chloroform, toluene Chloroform, methanol
30/175 0/175
526 594
Water colour Water colour
26 A.P. Singh et al. / Fuel 82 (2003) 23–33

Table 2 Table 4
GC Calibration data Identification of aromatic hydrocarbons with known standards (Fig. 1)

Compound Range (% w/w) Slope Y intercept R 2 Peak Identity Retention time RRT value

Benzene 0.5– 5.0 1.072 0.0035 0.999 1 n-C11 (undecane) 3.88 0.45
Toluene 1–20 1.051 0.0033 0.998 2 Benzene 4.72 0.55
Ethylbenzene 1–5.0 0.895 0.0070 0.997 3 n-C12 (dodecane) 5.91 0.69
m-p-Xylene 2–20 0.901 0.0085 0.996 4 Toluene 7.24 0.84
o-Xylene 1–10.0 0.912 0.0080 0.992 5 n-C13 (tridecane) 8.58 1.00
1,2,4-Trimethylbenzene 1–5.0 0.875 0.170 0.990 6 Ethylbenzene 9.59 1.12
1,2,3,5-Tetramethylbenzene 1–5.0 1.521 0.194 0.982 7 m-p-Xylenes 10.14 1.18
8 Isopropylbenzene 10.63 1.24
9 o-Xylene þ n-propylbenzene 11.97 1.40
2.7. Method validation 10 t-Butylbenzene 12.61 1.47
11 1,3,5-Trimethylbenzene 13.33 1.55
Known blends of aromatics with benzene, toluene, 12 1-Methyl 2-ethylbenzene 14.58 1.70
xylenes, 1,2,4-trimethylbenzene, ethylbenzene and 1,2,3,5- 13 1,2,4-trimethylbenzene 15.71 1.83
14 o-Diethylbenzene 18.16 2.12
tetramethylbenzene were prepared in a blend (1:1 ratio) of
15 1,2,3-Trimethylbenzene 19.41 2.26
heptane and isooctane in the range (benzene 0.1 –5.0%) and 16 n-Pentylbenzene 20.61 2.40
other aromatics (10 – 50% w/w). These ranges are generally 17 1,2,4,5-Tetramethylbenzene 25.36 2.96
found in commercial gasoline. The prepared blends were 18 1,2,3,5-tetramethylbenzene 26.03 3.03
analysed with the incorporation of n-C13 as IS in order to 19 1,2,3,4-Tetramethylbenzene 32.88 3.83
validate the results. The results are given in Table 3.
in small amounts. The proposed method limits to the
analysis of gasoline samples containing carbon range not
3. Results and discussion more than n-C11. The column selectivity is such that it
retains only aromatics and provides discrete separations
among the various aromatics. The m-p-xylenes co-elute and
3.1. GC method ethylbenzene elutes before m-p-xylenes. Table 4 shows the
detailed list of aromatic hydrocarbons that have been
The optimum separation conditions and polar phase identified from a commercial gasoline sample. The table
loading were selected as described in Section 2 that has also shows the relative retention time (RRT) with respect to
provided a discrete separation of benzene and other n-C13 which has been taken as internal standard. The
aromatics from the other components of commercial retention time of n-C13 has been taken as 1.00; with respect
gasoline. A chromatogram of commercial gasoline on to this the retention times of other aromatic hydrocarbons
TCEPE column is shown in Fig. 1. The total analysis time are given. The elution pattern of aromatics from m-p-xylene
is 35 min while helium is used as carrier gas. In the
onwards follows the order: isoproylbenzene, o-xylene þ
proposed method benzene elutes after n-C11, n-C12 between
n-propylbenzene, t-butylbenzene, 1,3,5-trimethyl benzene,
benzene and toluene and n-C13 elutes after toluene. The
1-methyl 2 ethylbenzene, 1,2,4-trimethylbenzene,
n-C14 elutes after m-p-xylene but before o-xylene. In the
o-diethylbenzene, 1,2,3-trimethylbenzene, 1,2,3,5-tetra-
selected column, paraffins and naphthenes elute as a single
methylbenzene and 1,2,3,4-tetramethylbenzene. Among
peak and benzene after n-C11. The n-paraffins were added in
the C9 aromatics, 1,2,4-trimethylbenzene is clearly separ-
the samples in order to verify whether these n-paraffins
ated from other aromatics. The C9 aromatics have been
separates clearly from other aromatics or not. Therefore, the
quantified using 1,2,4-trimethylbenzene calibration curve,
quantitation of aromatics is not affected if these are present
while C10 and higher aromatics have been quantified using
Table 3 1,2,3,5-tetramethylbenzene calibration curve. In the pro-
GC analysis results obtained for known calibration blends posed method there is no co-elution of MTBE/TAME/
Sample Benzene Total aromatics ETBE which are commercial oxygenates being added in
gasoline as octane boosters. They also elute before n-C11
a
Estimated Actual Estimateda Actual and similarly all the olefines up to C10 elute before n-C11
paraffin which has been confirmed experimentally. There is
Blend no. 1 1.10 ^ 0.06b 1.05 30.21 ^ 0.07b 30.34
Blend no. 2 2.24 ^ 0.03 2.18 35.28 ^ 0.08 35.50
no co-elution of olefinic or naphthenic components along
Blend no. 3 2.20 ^ 0.03 2.14 12.48 ^ 0.06 12.60 with the aromatics.
Blend no. 4 3.15 ^ 0.04 3.12 45.52 ^ 0.10 45.62 The packed column prepared from TCEP stationary
Blend no. 5 5.18 ^ 0.05 5.20 55.62 ^ 0.12 55.75 phase was evaluated under the conditions described in
a
Average of five determinations. Section 2. Fig. 3 shows the GC chromatogram of a
b
Average ^ standard deviations. commercial gasoline on TCEP column. Benzene elutes
 A.P. Singh et al. / Fuel 82 (2003) 23–33
Fig. 1. GC chromatogram of a commercial gasoline sample on TCEPE column.
after n-C11 however, in the selected conditions n-C12 co-
elutes with benzene and n-C13 with toluene. On comparison
of chromatograms in Figs. 2 and 3, it is quite evident that
TCEPE column has provided better separations of ethyl-
benzene from m-p-xylenes and t-butylbenzene from
o-xylene. However, on TCEP column t-butylbenzene has
co-eluted with o-xylene.
3.2. Estimation of benzene
A known amount of n-C13 (4% w/w) was added in all the
samples. The samples were chromatographed and the peaks
were identified from their retention times (tR) by running the
standards under the similar conditions. The area counts of
benzene and n-C13 peaks were noted and ratios of their area
counts were calculated. The amount of benzene present in
the samples was calculated from the calibration graph
already plotted between their area ratio and percentage of
benzene as described in Section 2. It is further rec-
ommended that for better result, a fresh blend of the
27
expected range of benzene should be prepared daily and GC
should be calibrated before the analysis. In case of naphtha
range samples containing only up to C8 aromatics the
analysis takes only 15 min. The commercial gasoline
samples containing MTBE, TAME or other ethers do not
interfere in benzene estimation in this method. However the
samples containing ethanol or 2-propanol interfere with
benzene analysis and both ethanol as well as 2-propanol co-
elute with benzene peak giving higher value of benzene.
Therefore care must be taken while analysing the reformu-
lated gasoline samples containing alcohols. However,
methanol does not interfere and elutes just before the
benzene peak (Fig. 2). The sample containing alcohols
should be tested by other reported methods such as GC [19,
20] and NMR [17,18]. In most of the samples containing
only up to C9 aromatics, the analysis is completed within
25 min. In few cases the samples containing C10 and higher
aromatics, the analysis time has been found to be 33 min.
The analyses of a commercial gasoline sample were also
performed on TCEP column at different oven temperatures,
28 A.P. Singh et al. / Fuel 82 (2003) 23–33
Fig. 2. GC chromatogram for commercial gasoline containing oxygenates (ethanol/isopropanol) on TCEPE column. (1) Benzene þ ethanol þ isopropanol, (2)
toluene.
i.e. 80 and 100 8C under similar condition as those of
TCEPE column. It was found that with change in
temperature and carrier gas flow, it did not improve
resolution on TCEP column, i.e. n-C12 could not be
separated from benzene peak. It has also been observed
that on TCEP column methanol could be separated from
benzene or n-C11. However, on both the columns ethanol
co-elutes with benzene. If we compare the analysis data on
two columns it was found that for benzene estimation both
the columns (TCEPE and TCEP) gave the same results.
3.3. Estimation of total aromatics
A weighed amount of IS (4% w/w) was added in all the
samples to be analysed and the solutions were prepared in
volumetric flasks. The selected samples were stored in
refrigerator at about 5 8C. It is important to mention here
that gasoline samples cannot be stored in volumetric flasks
in a refrigerator at 5 8C for more than a day without loss of
volatiles therefore all the prepared (blended) samples should
be analysed on the same day. A known amount (0.2 ml) of
the sample containing IS was injected into a gas
chromatograph using 10 ml hamilton syringe. The com-
ponents eluting after n-C11, i.e. benzene and other higher
aromatics were identified by the comparison of retention
time of standard mixture analysed under similar conditions.
The benzene, toluene, and C8 aromatics, viz. m-p-xylenes,
o-xylene and ethylbenzene were individually identified. The
C9 aromatics, viz. 1,2,3/1,3,5/1,2,4-trimethylbenzenes, were
also identified. Table 4 shows retention times and RRT
values of identified hydrocarbons analysed in the developed
GC conditions. Other components eluting after 1,2,3-
trimethylbenzene to 1,2,3,4-tetramethylbenzene were
taken as C10 aromatics. Most of the peaks eluting in the
range of C9 and C10 aromatics which are found in 95%
commercial gasolines have been identified and only few
aromatics which may be at trace levels could not be
identified due to non-availability of pure standard com-
pounds. The area counts of individual peak were noted and
the area ratios of these components were calculated and the
amount of these components were computed from the
calibration graph. The quantitation of C9 aromatics were
carried out from the calibration graph prepared using 1,2,4-
trimethylbenzene. The C10 aromatics were quantified using
calibration graph of 1,2,3,5-tetramethylbenzene. After this
quantitation, the values of individual group aromatics C6,
 A.P. Singh et al. / Fuel 82 (2003) 23–33
Fig. 3. GC chromatogram for commercial gasoline on TCEP column. (1) Benzene, (2) ethyl benzene, (3) o-Xylene and t-butyl benzene.
C7, C8, C9 and C10 were summed up to get the total aromatic
content in these samples.
It has been observed from the data of large number of
samples used in the present study from different sources
(straight run, hydrotreated, hydrocracked, etc.) that the
contribution of C10 aromatics towards the total aromatics
found to be in the range 0.5 –2.0% w/w. Therefore, the error
in calculation utilizing 1,2,3,5-tetramethylbenzene does not
contribute significant error in the estimation of total
aromatics as the response factors of the FID does not vary
more than 10% for the substituted C10 aromatic compounds
which has been confirmed experimentally. However, it has
been observed that among alcohols, if ethanol/2-propanol is
added in the gasoline, the real quantitation of benzene in
presence of the ethanol/2-propanol is not possible because
of the co-elution problem. However if methanol is added to
the gasoline, the problem of co-elution with benzene is not
observed since methanol gets separated prior to the benzene.
It is proposed to analyse the samples for benzene content
using our previous developed procedure [24] where the
exact quantitation of benzene, toluene and xylenes is also
possible even in the presence of these common oxygenates.
The NMR method developed earlier, which can estimate all
29
types of oxygenates (alcohols and ethers) in different types
of gasoline, can also be used directly [21,22].
3.4. Estimation of benzene and total aromatics by NMR
technique
3.4.1. Aromatics
The 1H NMR spectrum of the gasoline sample is shown in
Fig. 4. The respective characteristic region of benzene (sharp
signal at 7.35 ppm), aromatics (6.5–8.0 ppm), a-substituents
of aromatics (2.05–3.0 ppm), olefins (4.5–6.0 ppm) and
saturates (0.6–2.0 ppm) are also marked in the spectrum.
The spectral regions corresponding to oxygenates such as
methanol, ethanol, etc. are marked between 3.0 and 3.9 ppm.
Thus there are clear spectral separation among the different
types of hydrocarbon groups as observed in the 1H NMR
spectrum. The total aromatics content has been estimated by
our earlier published method as described in Refs. [18–20]. In
these methods, the relative average group molecular weight of
aromatics, olefins and saturates are estimated by the use of
certain derivations which take into account the integral
intensities of the various regions of 1H NMR spectrum.
The aromatic component analysis of commercial
30 A.P. Singh et al. / Fuel 82 (2003) 23–33
Fig. 4. 1H NMR spectra of gasoline sample containing oxygenates. (1) Ethanol, (2) methanol, (3) t-amyl ether.
gasoline samples from different refineries with different
crude sources and those by Frysinger and co-workers [23]
and Methiesen and Lubeck [25] by GC identify and list out
the types of aromatics which are generally found in the
gasoline range samples. The alkyl substituents such as
methyl, ethyl and isopropyl give signals in the region of 2–
3 ppm. Thus the contribution of C8 (xylenes, ethyl, etc.), C9
(trimethyl as mesitylene, etc.) and C10 (tetramethyl and
diethyl, etc.) aromatics are systematically considered and
included in the equations for the estimation of total
aromatics. The other type of heavier aromatics such as
biphenyls, indenes, indanes, naphthalenes, etc. which are
likely to be present in hydrocracked naphtha (HCN) or
FCCU-HCN (heavier naphtha) have also been included in
the total aromatic content. The proton signals due to these
components will appear in the regions of aromatics (6.4 – 8.0
and 2.05 – 3.5 ppm). However, the signals due to half of the
substituents (other than alpha and beta methylene) of the
components like o-butyltoluene, t-butylbenzene or butyl-
benzene (i.e. C4 substituents), etc. will overlap with the
signals due to saturates in the region of 0.5 –2.0 ppm. This
will marginally underestimate the total aromatics as the
average length of alkyl substituents has been assumed to be
C2. The error will be insignificant, as the total concen-
trations of these components are not more than 2.0%
compared to other types of aromatics. The aromatics are
then estimated by the following equation:
Aromatics ð%w=wÞ ¼ ðAw =Tw Þ £ 100
where Aw and Tw are the relative group molecular weight of
total aromatics and total, respectively. The above equation
is applicable to all type of naphtha and gasoline samples
(conventional, hydrocracked and cracked, i.e. containing
olefins). Different spectral treatment is required for various
types of gasoline samples formulated with different refinery
streams. The details of methods for samples containing
olefins are given in Ref. [19] and samples of hydrocracked
and straight run, i.e. not containing olefins are given in Refs.
[18,20].
The samples containing oxygenates such as alcohols and
ethers require different methodology to be adopted. The
details of the method are given in Ref. [22]. The absorption
regions of the oxygenates (alcohols and ethers) fall in the
chemical shift regions of 3.0 – 4.0 ppm, which is free from
the overlapping from benzene (7.35 ppm) and aromatics
(6.5 –8.0 ppm) [21]. Therefore, reformulated samples con-
taining ethanol and 2-propanol can also be analysed without
overlapping with the benzene signals, which is a limitation
of the proposed GC method. In this method the relative
molecular weight of the oxygenates present are estimated
separately by the use of certain equations and then added to
the group molecular weight of aromatics, olefins and
saturates to get the total group molecular weight of the
sample.
It has been realised to develop generalised equations so
that without prior knowledge about the types of the streams
ð1Þ or gasoline samples, the hydrocarbon type analyses could be
 A.P. Singh et al. / Fuel 82 (2003) 23–33
carried out. The NMR technique provides good discrimi-
nation between the various types of fuel samples as signals
due to different type of hydrocarbon and oxygenates appear
distinctly and separately [18 –22]. The generalised devel-
oped equations incorporate all regions of the spectra, viz.
aromatics, benzene, olefins, alpha-substituents, oxygenates
( – OCH2, – OCH3) and saturates. The basis of their
derivations can be found in Refs. [18,19,21,22]. The
following equations can be used for the estimation of
parameters Aw and Tw for applying in Eq. (1)
Tw ¼ 13A þ 20:5B þ 9C þ 7D þ 5E þ 14:36O þ ½Ox=n
£ Mw
Aw ¼ 13A þ 20:5B þ 9C
where A, B, C, etc. are the integral intensities of the regions
as marked in the spectrum and also given in Table 5 and Mw
is the molecular weight of the oxygenates as identified in the
sample.
It has been proved that total aromatics content estimated
by the above Eqs. (1) –(3) and equations mentioned in Refs.
[18 – 22] give accurate and consistent results. The last two
terms, viz. 14.36 O and [Ox/n ] £ Mw are due to the presence
of olefins and oxygenates. Therefore, depending upon the
presence of olefins and oxygenates, these terms can be
applied. If these two quantities or any one of these are found
to be absent, the integral values corresponding to these
groups should be used as zero. The aromatic content of
samples from different refineries involved in the processing
of different crude under different refinery processes have
been determined and compared with those of GC results.
The correlation co-efficient has been estimated to be R ¼
0:99 (Graph 1).
The 1H NMR method described above has a wider scope,
convenient and tolerate larger variation in the boiling ranges
of the products like HCN and reformulated gasoline
compared to single or multicolumn standard methods. The
method is rapid and complete analysis of aromatics and
benzene content of a sample can be carried out in 10 min.
The targeted and untargeted aromatics (mostly heavier
aromatics and unsaturated alkyl aromatics) as described in
Table 5
Chemical shift regions of hydrogen type in 1H NMR spectra
Hydrogen type Chemical shift (ppm)
Aromatic hydrogen (A) 8.0–6.4
Olefins (O) 6.0–4.5
Oxygenates (Ox) 3.9–3.0
Aromatic substituents
a CH (B) 3.0–2.8
a CH2 (B) 2.8–2.4
a CH2 (C) 2.4–2.05
Protons on a carbon adjacent to double bond 2.05– 1.5
CH signal of paraffinic part (D) 1.6–1.4
CH2 signals of paraffinic part (D) 1.4–1.05
CH2 signals of paraffinic part (E) 1.05– 0.6
31
ð2Þ
ð3Þ Graph 1. Correlation between NMR and GC methods for estimation of total
aromatics.
the method ASTM D5580 and in the subsequent modifi-
cation of this method suggested by Methiesen and Lubeck
[25] are all accounted for in the total aromatic content.
3.4.2. Benzene content
As discussed in Section 3.4.2, the benzene exhibits a
sharp and distinct signal at 7.35 ppm which is free from the
overlapping of signals of other types of mono aromatics
(toluene, ethylbenzene, mesitylene, etc.) present in naphtha/
gasoline range samples. However there is a rise in base line
at the demarcation of benzene signal and other aromatic
signals. In order to remove this, the aromatic region between
6.5 and 7.5 ppm was expanded along the X and Y directions
as shown in the inset of the spectrum of Fig. 3. The X – Y
expansion facilitates the determination of integral areas due
to benzene signal accurately and reproducibly.
The following equations have been used for the
determination of benzene in gasoline range products
ðIB £ 78 £ 100Þ
Benzene ð% w=wÞ ¼ ¼ 1300 £ IB=Tw ð4Þ
ð6 £ Tw Þ
where IB is the integral intensity of the sharp signal due to
benzene at 7.35 ppm.
The samples containing ethanol were simultaneously
analysed by NMR and GC. In such samples, estimation of
benzene content by NMR has advantage over GC method.
GC showed higher benzene content because of co-elution of
ethanol with benzene. The NMR spectra of blended gasoline
showed separate signals for benzene (7.3 ppm) and ethanol
(3.62 ppm), and no error in the estimation was observed.
The data by NMR on 90 samples including 10 samples
containing oxygenates have been compared with the data of
GC method and were treated statistically (Graph 2). The
correlation coefficient R was found to be 0.98. It has been
observed that there is a large difference to the extent of 25%
in the values of GC and NMR when the benzene content was
less than 0.5% w/w. This has been verified by analysing the
known blends of benzene containing 0.1– 0.5 wt%. The
NMR results have been found to vary between 20 and 25%.
32 A.P. Singh et al. / Fuel 82 (2003) 23–33

Table 6
Correlation data

Sample Benzene (% w/w) Aromatics (% w/w)

ASTM D5580 GC NMR ASTM D5580 GC NMR

Gasoline-1 1.04 1.01 1.02 32.8 32.5 33.9

Gasoline-2 1.01 1.02 0.98 35.9 35.6 37.1
Gasoline-3 1.25 1.27 1.19 33.8 34.4 34.8
Gasoline-4 1.26 1.30 1.30 34.1 34.2 33.1
Gasoline-5 0.95 1.01 0.95 19.7 19.5 20.8
Gasoline-6 2.48 2.45 2.38 47.2 48.8 48.6
Gasoline-7 0.88 0.90 0.82 33.5 33.2 32.8
Gasoline-8 0.95 0.92 0.86 33.0 33.0 33.2
Gasoline-9 1.27 1.23 1.28 41.1 41.5 40.4
Gasoline-10 1.34 1.36 1.26 30.9 31.5 32.1

Graph 2. Correlation between NMR and GC methods for estimation of

benzene content greater than 0.5%.
This has been attributed to the error in the measurement of
integral areas of benzene signals due to very poor signal-to-
noise ratio. The correlation coefficient in this case was
determined to be 0.94 (Graph 3).
3.5. Correlation with ASTM D5580-97 method
The results of 10 samples as obtained by the developed
GC as well as NMR methods have been correlated with the
results obtained by ASTM D5580 (Table 6). It is evident
from the results obtained by the three methods that the
correlation is excellent for both benzene as well as total
aromatic content. However, the small differences observed
in data of few samples have been attributed to the adoption
of different methodology and calibration standards. It has
also been observed that the samples in which only C9
aromatics are present in addition to other aromatics, the
difference in the total aromatics is negligible compared to
sample containing C9 and C10 aromatics. The result for total
aromatic content in three gasolines, viz. gasoline 6, 9 and 10
were found to be higher. The reason for this is that the
amount of C10 aromatics are higher in these samples. In the
proposed GC method, we have quantified all the C10
aromatics from a calibration graph of 1,2,3,5-tetramethyl-
benzene while in ASTM D5580 the quantitation is from a
calibration of 1,2,4-trimethylbenzene. It has been found that
the response of FID for these two types of compounds is
different and the detector (FID) has lower response for
1,2,3,5-tetramethylbenzene as compared to 1,2,4-trimethyl-
benzene in the range 5– 6%. The response factor of 1,2,3,5-
tetramethylbenzene has been taken into consideration for
reporting C10 aromatics. Therefore the difference is obvious
which is evident from Table 6. In other samples where
mainly up to C9 aromatics are present the difference in
values is insignificant. However all the results for benzene
as well as total aromatics are within the reproducibility
limits of the method ASTM D5580. The results obtained by
NMR method for benzene and total aromatics are also
within the reproducibility limits of the method ASTM
D5580.
3.6. Repeatability

The repeatability for benzene and total aromatics by both

the methods are estimated from the data on seven samples
(total aromatics in the range 1 – 45% and benzene in the
range 1 –10 wt%) which were analysed in replicates five
times. The ANOVA analysis on the data has been carried
out and variation between the sample results was considered
for the estimation of repeatability (r ). The ANOVA analysis
gives a value for standard deviation (sr). The repeatability
(r ) has been calculated from the following equation [26]:
p
r ¼ t £ 2 £ sr

where t is the student t, two tailed value, for n ¼ N 2 1 for a

given confidence, usually 95%. The estimated values for the
NMR and GC methods are given in Table 7. The
Graph 3. Correlation between NMR and GC methods for estimation of repeatability data on TCEPE column was found to be better
benzene content lesser than 0.5%. compared to TCEP column as given in Table 8.
 A.P. Singh et al. / Fuel 82 (2003) 23–33 33

Table 7 been solved using NMR method. Thus complete analysis of

Repeatability estimates by GC and NMR in gasoline/blends benzene and aromatic content in the gasoline samples
Component Range in mass (wt%) GC NMR containing oxygenates, i.e. ethers and alcohols is possible
by the combination of NMR and GC techniques.
Total aromatics 10–45 0.62 1.58
Benzene 1–10 0.10 0.20
References
Table 8
Precision data of TCEPE versus TCEP column [1] American Society for Testing and Materials. ASTM D1319-98,
Annual book of ASTM standards, vol. 05.01; 1999.
Type Averagea (% w/w) SD (% w/w) Repeatability [2] American Society for Testing and Materials. ASTM D3606-96,
Annual book of ASTM standards, vol. 05.02; 1997.
TCEPE TCEP TCEPE TCEP [3] American Society for Testing and Materials. ASTM D4420-94,
Annual book of ASTM standards, vol. 05.02; 1997.
Benzene 1.49 0.004 0.005 0.03 0.04 [4] American Society for Testing and Materials. ASTM D5134-98,
6.01 0.06 0.08 0.05 0.08 Annual book of ASTM standards, vol. 05.03; 1999.
Toluene 11.30 0.08 0.08 0.10 0.12 [5] American Society for Testing and Materials. ASTM D5443-93,
19.74 0.13 0.13 0.14 0.15 Annual book of ASTM standards, vol. 05.03; 1997.
Xylenes 11.80 0.04 0.08 0.11 0.10 [6] American Society for Testing and Materials. ASTM D5580-95,
8.83 0.08 0.09 0.08 0.09 Annual book of ASTM standards, vol. 05.03; 1997.
t-Butyl benzene 2.49 0.07 0.07 0.07 0.08 [7] The institute of Petroleum. Standard methods, IP 425, vol. 02; 1999.
6.50 0.06 0.07 0.12 0.11 [8] American Society for Testing and Materials. ASTM D5769-95,
Total aromatics 26.09 0.15 0.17 0.25 0.32 Annual book of ASTM standard, vol. 05.03; 1997.
41.15 0.28 0.38 0.42 0.53 [9] Poole SK, Furton KG, poole CF. J Chromatogr Sci 1988;26:67.
a
[10] Poole CF, Poole SK. Chem Rev 1989;89:377 –95.
Average of 10 measurements. [11] Walter B, Hawes SJ. J Chromatogr Sci 1977;15:185 –90.
[12] Mc Closkey DH, Hawes SJ. J Chromatogr Sci 1975;13:185–90.
4. Conclusions [13] Grob RL, editor. Modern practice of gas chromatography. New York:
Wiley; 1995. p. 123– 224.
The proposed GC method is very simple, convenient, fast [14] Pauls RE. J Chromatogr Sci 1985;23:437.
and analysis time is less than 35 min as compared to [15] Pauls; RE. Adv Chromagr 1995;259:35.
expensive and cumbersome multidimensional GC method [16] Diehl JW, Finkbeiner JW, Disanzo FP. Anal Chem 2015;(1996):67.
[17] American Society for Testing and Materials. ASTM D5986-96,
(ASTM D5580) for the gasolines samples containing carbon
Annual book of ASTM standards, vol. 05.03; 1992.
range up to n-C11 only. Since the proposed method requires [18] Kapur GS, Singh AP, Sarpal AS. Fuel 2000;79(9):1023 –9.
a single packed column, it can be used for monitoring the [19] Sarpal AS, Kapur GS, Mukherjee S, Tiwari AK. Fuel 2001;80(4):
quality of the product in the refinery equipped with simple 521–8.
GC instruments. The 1H NMR method described has a wider [20] Sarpal AS, Apparao NVR. Research Ind 1986;31:64.
[21] Kalsi WR, Sarpal AS, Jain SK, Srivastava SP, Bhatnagar AK. Energy
scope, rapid and convenient, and tolerates larger variations
Fuel 1995;15:2976.
in the boiling range of the products. The results obtained by [22] Sarpal AS, Kapur GS, Mukherjee S, Jain SK. Energy Fuel 1997;11:
both GC and NMR methods are comparable and repeatable. 666–7.
The combined GC and NMR methods are quite versatile and [23] Frysinger GS, Gaines RB, Ledford Jr. EB. J High Resol Chromatogr
applicable to both cracked, hydrofinished and hydrocracked 1999;22(4):195–200.
[24] Vatsala S, Singh AP, Kalsi WR, Basu B, Jain SK, Shrivastva SP,
gasoline including oxygenated products. The problem of co-
Bhatnager AK. Chromatographia 1995;40(9/10):607–10.
elution as faced in GC for estimation of benzene content and [25] Methiesen MD, Lubeck AJ. J Chromatogr Sci 1998;36:449–56.
total aromatics including heavier ones and in the presence of [26] Farrant TJ. Practical statistics for the analytical scientist: a bench
oxygenates such as (2-propanol and ethanol) and ethers has guide. Royal Society Chemistry; 1997.