Fuel 89 (2010) 99–104

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Quality control of gasoline by 1H NMR: Aromatics, olefinics, paraffinics,

and oxygenated and benzene contents
Carlos R. Kaiser a,*, Joana L. Borges a, Anderson R. dos Santos a, Débora A. Azevedo a, Luiz A. D0 Avila b
a
Instituto de Química, Universidade Federal do Rio de Janeiro, CT, Bloco A, Ilha do Fundão, 21949-909 Rio de Janeiro – RJ, Brazil
b
Escola de Química, Universidade Federal do Rio de Janeiro, CT, Bloco E, Ilha do Fundão, 21949-909 Rio de Janeiro – RJ, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: A simple and fast 1H NMR method, without any pretreatment, was developed for quality control of gas-
Received 18 November 2008 oline. It is based on the average group molecular weight approach and relative-content concept involving
Received in revised form 10 June 2009 aromatics, olefinics and paraffinics, including also ethanol and benzene contents. The ethanol content
Accepted 15 June 2009
was evaluated for Brazilian samples, but the method can be easily adapted to any oxygenated compound
Available online 3 July 2009
(ex. MTBE), and to gasoline from other countries. Twenty two laboratory prepared gasoline samples (gas-
oline from Brazilian refineries plus hydrocarbon solvents) and thirty four real (i.e., Brazilian gas stations)
Keywords:
gasoline samples were tested. The routine quality control carried out through the usual physicochemical
Gasoline
Quality control
analyses reached a level of confidence of 75% and 73% in detecting non conformity in laboratory and real
1
H NMR gasoline samples, respectively. The NMR method was very superior reaching 100% and 97% of confidence,
Hydrocarbon relative-contents respectively. It was better suited for laboratories with high sample throughput since measurement time
Ethanol is short and only one NMR experiment is needed per sample.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction because of its dependence in the ethanol production in the country,

and it is regulated by the Brazilian National Agency of Petroleum,
The Quality control of gasoline is performed everywhere in the Natural Gas and Biofuels (ANP) [1–7].
world by refineries, distribution companies and government During the last six years, our research group has been working on
inspection departments. It is a crucial procedure, as gasoline must the improvement of gasoline quality control in Brazil, using physico-
follow the quality standards set for local markets, meeting the chemical, chromatographic, mass spectrometric and statistical ap-
requirements of car engines and with minimum possible damage proaches. The need for such studies relies on the high level of non
to the environment. Obtained by mixing several refinery streams conformity cases found in the commercial gasoline distributed by
at predetermined concentrations, gasoline is a complex mixture gas stations around the country. The quality control of the Brazilian
of volatile flammable liquid hydrocarbons containing hundreds of gasoline is routinely verified by physicochemical analysis according
different natural components (divided in paraffinics, olefinics, to ANP regulations and, as demonstrated in previous works, it is not
and aromatics), a few sulfur containing compounds, and some adequate when low levels of hydrocarbon solvents are intentionally
additives like methyl tert-butyl ether (MTBE) or anhydrous ethyl added to the gasoline. These solvents are of low cost, and such inten-
alcohol (Alc or ethanol) [1–5]. tional addition characterizes an adulteration. Its detection is usually
Some physicochemical parameters of gasoline as fractionating difficult due to the similar hydrocarbon constitution and physico-
distillation temperatures, motor octane number, antiknock index, chemical properties between the solvents and the gasoline. Other
etc, must be analysed before this product can be commercialized. research groups have also published works on this subject, demon-
All of them are extremely dependent on the hydrocarbon distribu- strating that only physicochemical parameters are not enough when
tion between aromatic, olefinic and paraffinic compounds, and the sophisticated hydrocarbon solvent adulterations are involved. Try-
content of the oxygenated compound (MTBE or Alc), added as oc- ing to overcome such difficulty, diverse spectroscopic (IR), spectro-
tane enhancer. In Brazil, ethanol (Alc) is the selected additive for metric (Mass), chromatographic (GC) and chemometric methods
gasoline, mainly due to its high production (obtained from sugar have been applied, and although there has been evident improve-
cane), and because pollutant emissions, such as carbon monoxide, ment in quality, some lacks are still a reality [3–18].
are reduced when a high content of ethanol is used in gasoline (20– Development of new analytical methods or methodologies for
25%). The exact quantity added to gasoline changes continuously quality control of commercial gasoline, including solvent adultera-
tion, remains a challenge for academic and forensic research. Thus,
* Corresponding author. Tel.: +55 21 25627736; fax: +55 21 25627944. to insert new insights into this research field, we present here an
E-mail address: [email protected] (C.R. Kaiser). application based on Nuclear Magnetic Resonance Spectroscopy

0016-2361/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2009.06.023
100 C.R. Kaiser et al. / Fuel 89 (2010) 99–104

(NMR). Some interesting studies using NMR in gasoline composi- pressure for 10% v/v between 45 and 60 °C (T10), 50% v/v between
tion analyses showed promising results when compared with 90 and 110 °C (T50), 90% v/v between 160 and 190 °C (T90), 98% v/
other techniques and also with physicochemical parameters. How- v between 195 and 215 °C (T98), distillation residue of 0–2% v/v
ever, effective quality control and adulteration detection involving (RE), motor octane number (MON), antiknock index (AKI). The eth-
NMR analyses have not being in fact undertaken before [19–26]. As anol (Alc, maximum 25% v/v) and benzene contents (B, maximum
related above, the hydrocarbon composition and oxygenated addi- of 1% v/v) were also determined (ABNT/NBR13992 and ABNT/
tive content are determinant to the physicochemical characteris- D6277) [7].
1
tics and quality of the gasoline. Thus, here we describe, for H NMR spectra were obtained on a 300 MHz BRUKER spec-
quality control purposes, a 1H NMR method, based on the average trometer from 0.3 mL gasoline samples diluted to 0.6 mL with
group molecular weight approach and relative-content concept, CDCl3/TMS in a 5 mm (o.d.) tube, calibrated to the internal tetra-
involving aromatics, olefinics and paraffinics, and also including methylsilane (TMS) signal at 0,0 ppm. The measure conditions
the ethanol and benzene contents in commercial gasoline. It is car- were: 298 K temperature; 3300 Hz spectral width; 32 K spectral
ried out in few minutes and does not require any pretreatment. size; 8 ls pulse (p/2); 5 s acquisition time; 5 s relaxation delay;
The ethanol content was evaluated, but the method can also be 32 scans. Integration has been done after baseline correction. Mea-
easily adapted to any other oxygenated compound (ex. MTBE), surements were performed in triplicate.
and to gasoline from other countries.
2.3. Determination of the gasoline composition
2. Experimental
The composition in aromatic, olefinic, and paraffinic hydrocar-
2.1. Reagents and samples bons, and ethanol and benzene compounds was carried out inte-
grating the 1H NMR signals in the regions of interest (Fig. 1), and
Refinery gasoline samples were supplied by two different local inserting the obtained values in the equations below. These equa-
companies and after the addition of the anhydrous ethanol (25% v/ tions were adapted from the average group molecular weight ap-
v), it was considered in this work as the reference gasolines (LG1-X proach [19–26], in such a way as to allow the application for
and LG1-Y). The refinery gasolines must be of high quality, what is quality control purposes. In all equations, VX is related to the rela-
controlled by Federal laws. This certified quality is a refinery gaso- tive partial volume of the X group substances; NX is proportional to
line pre-requisite before reaching the marked (gas station real gas- the number of X molecules, MWX is related to the mean molecular
olines), but always with the previous addition of ethanol weight of the X group substances and qX is related to the mean
performed by the distribution companies. density of the X group substances.
Anhydrous ethanol and deuterium chloroform (99.8% D, with
0.01% TMS) were supplied by Tedia-Brazil (Rio de Janeiro, Brazil). 2.3.1. Aromatics
Three commercial solvents were donated by chemistry industries:
NAr ¼ ðAr þ A1 þ 1=2A2 þ 1=3A3 Þ  ð2=3TÞ=6
light aliphatic hydrocarbons (LA, distillation range 52–128 °C), aro-
matic hydrocarbons (AR, distillation range 110–142 °C) and heavy aH ¼ ðAr1 þ Ar2 þ BÞ=NAr aM ¼ A3 =3NAr aE ¼ A2 =2NAr aP ¼ A1 =NAr
aliphatic hydrocarbons (HA, distillation range 162–183 °C). MWAr ¼ 72 þ aH þ ð15aM Þ þ ð29aE Þ þ ð43aP Þ
Laboratory gasoline samples were prepared by mixing 25.0 mL qAr ¼ 868:0 Kg=m3 VAr ¼ ðNAr  MWAr Þ=qAr
of ethanol and 37.5 mL of each refinery gasoline (X and Y) with var-
%VolðArÞ ¼ 100 VAr =ðVAr þ VO þ V P þ VAlc Þ
iable quantities of solvents to a final volume of 100 mL (Table 1),
leading to 20 samples (LG2-X to LG11-X and LG2-Y to LG11-Y).
Ar, A1, A2, A3 are the integral values from the regions indicated
Two control or reference laboratory gasoline samples (LG1-X and
in Fig. 1. The aH, aM, aE, e aP terms are related to the mean number
LG1-Y) with 25.0 mL of ethanol and 75.0 mL of each refinery gaso-
of aromatic hydrogens and its methyl, ethyl, isopropyl substituents
line samples were also included. Commercial gasoline samples (gas
(side chain hydrogens from aromatics). The benzene (B) integral is
station gasolines) were obtained from local gas stations in a total of
measured separately to calculate its content (see below), but the
34 samples (CG1 to CG34).
value is included in the aromatic content. In the case of the residual
CHCl3 signal (CDCl3 solvent is 99.8% D pure) it is discounted from
2.2. Physicochemical and 1H NMR parameters
the aromatics, the CDCl3 solvent spectrum is acquired and the
CHCl3 and the TMS signals are integrated. These integral data are
Standard physicochemical specification tests (ABNT/NBR9619,
transported to the gasoline analyses (NAr) in which only the TMS
ASTM/D86, ABNT/MB457, ASTM/D2700, ASTM/D2699) were deter-
(T) can be individually integrated (between 0.25 and 0.25 ppm
mined for all samples as follow: distillation range at atmospheric
to include the 1JCH satellite).

Table 1
2.3.2. Olefinics
Composition (% v/v) of laboratory gasoline samples.*
NO ¼ 1=2ðO1 þ 1=2O4 Þ þ 1=2O2 þ O3 þ 1=2O5
Sample Refinery gasoline Ethanol LA HA AR
q ¼ ð1=2ðO1 þ 1=2O4 Þ þ O2 þ 3O3 þ O5 Þ=NO
LG1 75.0 25.0 – – –
LG2 37.5 25.0 – – 37.5 qH ¼ 4  qM  qE; qM ¼ q  qE; qE ¼ 1=2ðOe =NO Þ
LG3 37.5 25.0 – 37.5 –
LG4 37.5 25.0 37.5 – –
MWO ¼ 24 þ qH þ 15qM þ 29qE
LG5 37.5 25.0 – 18.8 18.8
qO ¼ 669:0 Kg=m3 VO ¼ ðNO  MWO Þ=qO
LG6 37.5 25.0 18.8 – 18.8
LG7 37.5 25.0 18.8 18.8 – %VolðOÞ ¼ 100VO =ðVAr þ VO þ VP þ VAlc Þ
LG8 37.5 25.0 12.5 12.5 12.5
LG9 37.5 25.0 26.1 5.7 5.7 O1, O2, O3, O4, O5, Oe are the integral values from the regions
LG10 37.5 25.0 5.7 26.1 5.7 indicated in Fig. 1. The qH, qM, and qE terms are related to the mean
LG11 37.5 25.0 5.7 5.7 26.1
number of olefinic hydrogens and its methyl and ethyl substituents
*
Solvents added: LA, light aliphatics; HA, heavy aliphatics; AR, aromatics. (side chain hydrogens from olefins).
 C.R. Kaiser et al. / Fuel 89 (2010) 99–104 101

ethanol

benzene
TMS

Fig. 1. Typical 1H NMR spectrum of a Brazilian commercial gasoline, showing the regions to be integrated and benzene, ethanol and TMS signals.

Table 2
2.3.3. Paraffinics 1
H NMR determination of laboratory gasoline and solvents compositions (% v/v).
IP ¼ Al  ð6A1 þ 3=2A2 Þ  3=2ðOe Þ  ð1 þ ðqM =qE ÞÞ  ðQ Þ Samplea Compositionb
3
HP ¼ 15;MWP ¼ 93:0; qP ¼ 668:0 kg=m ; VP ¼ ðIP  MWP Þ=ðHP  qP Þ Ar O P Alc B
%VolðPÞ ¼ 100VP =ðVAr þ VO þ VP þ VAlc Þ LG1-X 18.2 15.2 40.6 26.0 0.3
LG2-X 45.3 6.7 22.7 25.3 0.1
Al is the integral value of the region indicated in Fig. 1. HP is re- LG3-X 11.7 6.6 56.8 24.8 0.1
LG4-X 10.1 7.6 56.6 25.7 0.4
lated to the hydrogen number of atoms per molecule (2n + 2). The LG5-X 28.2 6.8 39.5 25.4 0.1
contributions of the ethanol (Q), aromatic (A1 and A2) and olefinic LG6-X 28.0 7.3 39.1 25.6 0.3
(OE) substituents are adequately subtracted. LG7-X 11.4 7.3 55.3 26.0 0.2
LG8-X 22.3 7.3 44.9 25.5 0.2
LG9-X 16.1 7.6 50.7 25.6 0.3
2.3.4. Benzene (B)
LG10-X 16.9 7.1 50.7 25.4 0.2
LG11-X 35.1 7.3 32.5 25.2 0.2
NB ¼ B=6; MWB ¼ 78:1; qB ¼ 879:0 Kg=m3 ; LG1-Y 20.9 13.0 40.1 26.1 0.6
VB ¼ ðNB  MWB Þ=qB%VolðBÞ ¼ 100VB =ðVAr þ VO þ VP þ VAlc Þ LG2-Y 45.5 7.6 22.2 24.8 0.3
LG3-Y 12.7 6.3 56.3 24.6 0.3
B is the integral value of the singlet indicated in Fig. 1 (between LG4-Y 11.4 7.0 55.2 26.4 0.5
7.4 and 7.3 ppm). LG5-Y 29.4 6.7 39.3 24.7 0.3
LG6-Y 28.2 6.8 39.6 25.4 0.1
LG7-Y 11.7 6.5 56.3 25.5 0.4
2.3.5. Ethanol (Alc) LG8-Y 22.6 6.8 45.8 24.8 0.4
LG9-Y 16.3 6.8 51.0 25.9 0.4
NAlc ¼ Q =3 MWAlc ¼ 46; qAlc ¼ 791:5 Kg=m3 ; VAlc ¼ ðNAlc  MWAlc Þ=qAlc LG10-Y 18.0 7.5 49.3 25.2 0.3
%VolðAlcÞ ¼ 100VAlc =ðVAr þ VO þ VP þ VAlc Þ LG11-Y 35.0 6.6 33.3 25.1 0.3
LA 3.9 1.7 94.4 0.0 0.6
HA 9.5 0.6 90.0 0.0 0.0
Q is the integral value of the region indicated in Fig. 1. It is con-
AR 91.5 0.2 8.3 0.0 0.3
sidered the sum of the CH2 and OH signals, and is equally sub-
a
tracted from the paraffinic content (due to CH3 group from LG, laboratory gasoline; X, refinery X; Y, refinery Y; LA, light aliphatic solvent;
HA, heavy aliphatic solvent; AR, aromatic solvent.
ethanol, see Fig. 1). This integral takes also into account the 1JCH sa- b
Ar, aromatics; O, olefinics; P, paraffinics; Alc, ethanol; B, benzene. See text for
tellite from the CH2 signal and eventual fluctuations of the hydro- more details.
xyl signal due to the presence of small amount of water.

3. Results and discussion paring five fresh samples, measuring the spectra and applying the
proposed equations to obtain the constituent percentages (% v/v).
3.1. Establishing the 1H NMR quality criteria The observed variances from the mean value were very low, show-
ing good repeatability of the method: ±0.8% v/v for aromatics,
To establish the 1H NMR method, twenty laboratory gasoline ±0.9% v/v for olefinics, ±0.5% v/v for aliphatics, ±0.9% v/v for ethanol
samples were prepared (Table 1) by mixing ethanol and one of and ±0.3% v/v for benzene.
independent refinery gasolines (X or Y) with three distinct solvents The direct inspection of the data in Table 2 shows that some
(LA, HA, AR). The calculated contents obtained by the 1H NMR mea- hydrocarbon distribution variations are easily detected (e.g.,
surements with application of the equations for laboratory refer- LG2-X or LG2-Y) while others are not so evident (e.g., LG8-X and
ence gasolines (LG1), laboratory gasolines (LG2 to LG11) and LG8-Y). However, it should be noted that great quantities of
solvents are shown in Table 2. The LG8-X sample (Table 1) was paraffinic (LA or HA) or aromatic (AR) solvents in the mixtures
used for the repeatability test of the proposed NMR method by pre- (LG2 to LG4) causes a substantial unbalance in the P/Ar relation
102 C.R. Kaiser et al. / Fuel 89 (2010) 99–104

if compared with the original reference gasolines. Another, and the Y and LG9-Y). This result means that 25% of these samples were
most important observation in Table 2, is that the olefinic content classified as conform gasoline when only routine physicochemical
decrease considerably in all mixtures (LG2 to LG11) because of the quality control was applied. The present NMR method could not
dilution effect with commercial solvents, which have a very low distinguish between the presence of light or heavy aliphatic hydro-
concentration of olefins as characteristic (see individual LA, HA carbon solvents, but did distinguish easily the presence of aliphatic
and AR compositions). or aromatic solvents (see LG-2, LG-3, LG-4, for X and Y refinery gas-
Thus, for more precise, simpler and quicker evaluation, we oline samples). For composite solvent mixtures (LG-5 to LG-11, for
introduced the inspection of the NMR data in the concept of ‘‘rela- X and Y refinery gasoline samples) the NMR method shows at least
tive-contents” involving the NMR measurements of the hydrocar- three out of range hydrocarbon parameters (relative-contents),
bon contents, % v/v (aromatics, Ar; olefinics, O; and paraffinics, either with high (26.1 and 18.8% v/v) or low (12.5 and 5.7% v/v) sol-
P), and acceptable ranges, as follow: vent contents.
P=Ar ¼ 1:7—2:5; P=ðAr þ OÞ ¼ 1:0—1:4; P=O
3.2. 1H NMR method as quality control
¼ 2:2—3:5; Ar=O ¼ 1:0—1:8
Thirty four gas station gasoline samples were analysed applying
The acceptable ranges from the above relations were obtained
the proposed 1H NMR quality criteria in a real situation. The data
by inspecting the fluctuation in the measured NMR hydrocarbon
collection is listed in Table 4. It includes the NMR final result (R)
content from three distinct production streams of each refinery
indicating a commercial gas station gasoline as conform (C) or
gasoline (X and Y). The variations in a refinery production line
non-conform (NC) by the present 1H NMR method. Physicochemi-
are small but taking them into account, make the NMR analysis
cal control was also included when out of range parameters indi-
for future analyses of real gasolines (gas stations) independent
cated the sample as non-conform.
from the reference gasolines (refineries) and ethanol content. Thus,
Benzene contents of the commercial gasoline samples were be-
as quality criteria of the present 1H NMR method we established
low 1% for all samples and for both methods (NMR and physico-
that the accepted (or conform) gasoline must present at least three
chemical) suggesting, together with the laboratory gasoline data
relative-contents with acceptable ranges. The ethanol and benzene
(Table 3), that this hazardous compound is not a critical problem
compositions were also verified as ANP regulations [7] establish a
in Brazilian gasolines.
maximum benzene content of 1% v/v and an ethanol content of 25%
NMR control of the commercial gasoline samples indicated
v/v. Table 3 illustrates the final result of the proposed 1H NMR
eight samples not conform for ethanol content whereas the phys-
method for the laboratory gasolines, including the physicochemi-
icochemical control indicated only two. That was relevant for NMR
cal data for comparison.
control, as these samples with ethanol non conformity (CG5, 10,
In Table 3, both NMR and physicochemical analyses were in
13, 14, 30, 32, 34, 35) also showed significant out of range hydro-
agreement with the standard parameter for the ethanol and ben-
carbon contents (at least, in three relative-contents).
zene content for all the laboratory samples. The 1H NMR method
NMR analyses of all commercial gasoline samples detected
showed non-conformity results (two or more out values from the
eighteen that were not conform, while physicochemical control
relative-contents ranges) for all the twenty hydrocarbon solvent
indicated only ten of them. It should be noted that these ten sam-
mixtures (LG-2 to LG-11), whereas the usual physicochemical anal-
ples showed the greatest out of range relative-contents in the NMR
yses classified five of them as conform (LG4-X, LG6-X, LG4-Y, LG6-

Table 3
1
H NMR and physicochemical quality control of laboratory gasoline samples.a

Sampleb 1
H NMR controlc Physicochemical controld
(out of range parameters)
P/Ar P/(Ar+O) P/O Ar/O Alc B
[1.7–2.5] [1.0–1.4] [2.2–3.5] [1.0–1.8] [25 ± 1%] [61%]
LG1-X 2.2 1.2 2.7 1.2 26 61 –
LG2-X 0.5 0.4 3.4 6.7 25 61 T10, T50
LG3-X 4.9 3.1 8.6 1.8 25 61 T10, T50, T90, T98, RE, MON, AKI
LG4-X 5.6 3.2 7.4 1.3 26 61 –
LG5-X 1.4 1.1 5.8 4.1 25 61 T10, T50, T98
LG6-X 1.4 1.1 5.3 3.8 26 61 –
LG7-X 4.8 3.0 7.6 1.6 26 61 T10, T98
LG8-X 2.0 1.5 6.1 3.0 26 61 T10, T50
LG9-X 3.2 2.1 6.7 2.1 26 61 T10
LG10-X 3.0 2.1 7.2 2.4 25 61 T10, T50, RE, MON, AKI
LG11-X 0.9 0.8 4.5 4.8 25 61 T10, T50
LG1-Y 1.9 1.2 3.1 1.6 26 61 –
LG2-Y 0.5 0.4 2.9 6.0 25 61 T10, T50
LG3-Y 4.4 3.0 8.9 2.0 25 61 T10, T50, T90, T98, MON, AKI
LG4-Y 4.8 3.0 7.9 1.6 26 61 –
LG5-Y 1.3 1.1 5.9 4.4 25 61 T10, T50, T98, RE
LG6-Y 1.4 1.1 5.8 4.1 25 61 –
LG7-Y 4.8 3.1 8.7 1.8 26 61 T98, MON, AKI
LG8-Y 2.0 1.6 6.8 3.3 25 61 T50
LG9-Y 3.1 2.2 7.5 2.4 26 61 –
LG10-Y 2.7 1.9 6.6 2.4 25 61 T10, T50,T90, T98, AKI
LG11-Y 1.0 0.8 5.1 5.3 25 61 T10, T50
a
In range (bold) and out of range (italic) NMR (% v/v) and physicochemical parameters.
b
LG, laboratory gasoline (see Table 1); X, refinery X; Y, refinery Y.
c
Ar, aromatics; O, olefinics; P, paraffinics; Alc, ethanol; B, benzene.
d
See experimental.
 C.R. Kaiser et al. / Fuel 89 (2010) 99–104 103

Table 4
1
H NMR and physicochemical quality control of commercial gasoline samples.a

Sampleb 1
H NMR controlc Physicochemical controld
(out of range parameters)
P/Ar P/(Ar+O) P/O Ar/O Alc B (R)
[1.7–2.5] [1.0–1.4] [2.2–3.5] [1.0–1.8] [25±1%] [61%]
CG1 9.3 4.0 6.9 0.7 25 61 NC MON, AKI
CG2 2.0 1.0 2.1 1.1 26 61 C –
CG3 2.4 1.4 3.2 1.3 26 61 C –
CG4 2.5 1.5 3.5 1.4 25 61 C –
CG5 1.6 0.9 1.8 1.1 27 61 NC –
CG6 2.6 1.5 3.8 1.5 26 61 NC –
CG7 2.4 1.4 3.3 1.4 26 61 C –
CG8 2.2 1.2 2.7 1.2 26 61 C –
CG9 2.5 1.4 3.3 1.3 26 61 C –
CG10 3.5 1.9 4.2 1.2 28 61 NC –
CG11 2.1 1.1 2.4 1.2 25 61 C –
CG12 4.3 2.2 4.5 1.0 26 61 NC –
CG13 12.1 4.3 6.7 0.6 29 61 NC MON, Alc
CG14 12.7 5.0 8.3 0.7 23 61 NC MON, AKI
CG15 2.1 1.3 3.6 1.6 26 61 C –
CG16 3.6 2.2 5.9 1.6 26 61 NC T10, T90, T98
CG17 8.6 4.8 10.8 1.2 25 61 NC T1, T50,T90, T98, MON, AKI
CG18 2.5 1.4 2.9 1.2 26 61 C –
CG19 9.4 8.2 63.8 6.7 25 61 NC T10, T90, T98, AKI
CG20 1,2 0.8 2.0 1.7 26 61 NC –
CG21 1.8 1.0 2.2 1.2 26 61 C –
CG22 6.8 4.1 10.7 1.6 26 61 NC T10, T90, T98
CG23 2.0 1.1 2.2 1.1 25 61 C –
CG24 1.8 1.0 2.0 1.1 26 61 C –
CG25 2.3 1.1 2.2 1.0 26 61 C –
CG26 2.4 1.2 2.5 1.0 26 61 C –
CG27 2.0 1.1 2.5 1.2 26 61 C –
CG28 13.7 10.9 53.7 3.9 26 61 NC T10, T90, T98, MON, AKI
CG29 3,7 2,0 4.3 1.2 28 61 NC T90, T98, Alc
CG30 2.3 1.1 2.3 1.0 26 61 C –
CG31 2.1 1.6 6.0 2.8 23 61 NC –
CG32 26.1 20.6 96.9 3.7 26 61 NC T50,T90, T98, MON, AKI
CG33 2.6 1.6 3.9 1.5 22 61 NC –
CG34 1.6 1.0 2.9 1.9 23 61 NC –
a
In range (bold) and out of range (italic) NMR (% v/v) and physicochemical parameters.
b
CG, commercial gasoline.
c
Ar, aromatics; O, olefinics; P, paraffinics; Alc, ethanol; B, benzene; (R), NMR result; C, NMR conform gasoline; NC, NMR not conform gasoline.
d
See experimental.

hydrocarbon determination. Again, it seems that physicochemical ied samples. Sample CG15 was a distinct sample as its solvent mar-
control was not sensitive enough to detect low variations in the ker content was really very low (bellow 10 ppb). We believe that
gasoline constituents, mainly when sophisticated additions of this gasoline was not intentionally altered, but had been in contact
hydrocarbon solvents are involved (solvent adulterations). with solvents or adulterated gasoline in stock, transport or gas sta-
Also with the aim of verifying possible solvent adulterations, tion reservoirs. It must be clear that, in Brazil, the detection of the
the commercial gasoline samples were submitted to hydrocarbon marker characterizes the gasoline adulterated with solvents, but its
solvent marker analysis. The addition of this marker is, by ANP absence does not indicate that the gasoline is conform (without
regulation (ANP 274), mandatory for solvent importers or distrib- adulteration). For example, Sample CG14 was shown to be not con-
utors. However, only a few of the ANP’s quality control laborato- forming by both the NMR and physicochemical methods but no
ries around the country are certified to carry out this marker marker was detected. This means that not all hydrocarbonic sol-
analysis in gasolines, and because it is a time consuming and vents distributed in the country are being marked and thus, they
expensive analysis, only very suspicious gasoline samples are are also not conform to the Brazilian respective laws.
tested for the marker in routine analyses. The present study found The acquired knowledge of our group throughout the years
five commercial gasolines in which the marker was present: CG10, studying the quality control of gasolines was determinant to the
CG15, CG32, CG33 and CG34. The non conformity of these gaso- significant advances described in the present work. For example,
lines was detected by both NMR and physicochemical control the 1H NMR technique was applied to overcome some limitations
for sample CG32 and only by NMR control for samples CG10, in detecting low concentration adulterations by aromatics or light
CG33, CG34. That is, only the NMR method detected all the adul- aliphatics using GC–FID, GC–MS and statistical analysis found in
terated or non-conform commercial gasoline samples. Sample previous studies [3–5]. Other researcher groups, using other tech-
CG32 showed a high level of the marker (above 100 ppb) while niques, also pointed out some bordering concentrations in detect-
in the other samples the level was considered low (bellow ing hydrocarbon solvent adulterations [6,13,14]. In the present
40 ppb). These results show once again that physicochemical con- work, it was simulated up to 6% v/v of minimum intentional hydro-
trol is not suitable for detecting low level adulterations of hydro- carbon solvent additions to construct the quality control model,
carbon solvents. and all of them were detected. Thus, in this level of detection,
The 1H NMR method failed only for one sample (CG15) among the simple and low time analysis proposed NMR method showed
thirty four, or in other words it got 97% of confidence for the stud- to be the best choice for gasoline quality investigation.
104 C.R. Kaiser et al. / Fuel 89 (2010) 99–104

4. Conclusion [9] Flumignan DL, Ferreira FO, Tininis AG, Lopes MN, Oliveira JE. Development,
optimization and validation of gas chromatographic fingerprinting of
Brazilian commercial gasoline for quality control. J Chromatogr A
We established for quality control purposes a 1H NMR method 2008;1202:181–8.
based on the relative-contents (aromatics, olefinics and paraffinics) [10] Flumignan DL, Ferreira FO, Tininis AG, Oliveira JE. Multivariate calibrations in
gas chromatographic profiles for prediction of several physicochemical
concept and also on the ethanol and benzene contents in gasolines.
parameters of Brazilian commercial gasoline. Chemom Intell Lab Syst
It is worthy of note that no pretreatment of the sample is necessary 2008;92:53–60.
and that the result can be obtained in only one measurement with [11] Kardamakis AA, Mouchtaris A, Pasadakis N. Linear predictive spectral coding
and independent component analysis in identifying gasoline constituents
up to 6 min. The present method evaluates ethanol content, but
using infrared spectroscopy. Chemom Intell Lab Syst 2007;89:51–8.
can be easily adapted to any oxygenated compound (ex. MTBE), [12] Oliveira FS, Teixeira LSG, Araujo MCU, Korn M. Screening analysis to detect
and to gasolines from other countries. The method attained 100% adulterations in Brazilian gasoline samples using distillation curves. Fuel
and 97% of confidence in non conformity detection of simulated 2004;83:917–23.
[13] Brudzewski K, Kesik A, Kołodziejczyk K, Zborowska U, Ulaczyk J. Gasoline
and of real gasolines, respectively. On the other hand, physico- quality prediction using gas chromatography and FTIR spectroscopy: an
chemical analyses lead to only 75% and 73% confidence, artificial intelligence approach. Fuel 2006;85:553–8.
respectively. [14] Skrobot VL, Castro EVR, Pereira RCC, Pasa VMD, Fortes ICP. Identification of
adulteration of gasoline applying multivariate data analysis techniques HCA
This method can be applied as pre-analysis in routine quality and KNN in chromatographic data. Energy Fuels 2005;19:2350–6.
control of gasolines, in view of possible automation, given that it [15] Barbeira PJS, Pereira RCC, Corgozinho CNC. Identification of gasoline origin by
allows analyses of a great number of samples, and employs only physical and chemical properties and multivariate analysis. Energy Fuels
2007;21:2212–5.
one NMR dedicated instrument. [16] Skrobot VL, Castro EVR, Pereira RCC, Pasa VMD, Fortes ICP. Use of principal
component analysis (PCA) and linear discriminant analysis (LDA) in gas
Acknowledgements chromatographic (GC) data in the investigation of gasoline adulteration.
Energy Fuels 2007;21:3394–400.
[17] Pereira RCC, Skrobot VL, Castro EVR, Fortes ICP, Pasa VMD. Determination of
The authors acknowledge CNPq, the Brazilian Research Council, gasoline adulteration by principal components analysis-linear discriminant
for financial support and fellowships. analysis applied to FTIR spectra. Energy Fuels 2006;20:1097–102.
[18] Ré-Poppi N, Almeida FFP, Cardoso CAL, Raposo Jr JL, Viana LH, Silva TQ, et al.
Screening analysis of type C Brazilian gasoline by gas chromatography – flame
References ionization detector. Fuel 2009;88:418–23.
[19] Kalsi WR, Sarpal AS, Jain SK, Srivastava SP, Bhatnagar AK. Determination of
[1] Pasadakis N, Kardamakis AA. Identifying constituents in commercial gasoline oxygenates in gasolines by 1H nuclear magnetic resonance spectroscopy.
using Fourier transform-infrared spectroscopy and independent component Energy Fuels 1995;9:574–9.
analysis. Anal Chim Acta 2006;578:250–5. [20] Sarpal AS, Kapur GS, Mukherjee S, Jain SK. Estimation of oxygenates in gasoline
[2] Pavén JLP, Sánchez MN, Laespada MEF, Pinto CG, Cordero BM. Determination of by 13C NMR spectroscopy. Energy Fuel 1997;11:662–7.
benzene in gasoline using direct injection-mass spectrometry. Anal Chim Acta [21] Sarpal AS, Kapur GS, Mukherjee S, Tiwari AK. PONA analyses of cracked
2006;576:156–62. gasoline by 1H NMR spectroscopy. Part II. Fuel 2001;80:521–8.
[3] Wiedemann LSM, D́Avila LA, Azevedo DA. Brazilian gasoline quality: study of [22] Kapur GS, Singh AP, Sarpal AS. Determination of aromatics and naphthenes
adulteration by statistical analysis and gas chromatography. J Braz Chem Soc in straight run gasoline by 1H NMR spectroscopy. Part I. Fuel 2000;79:
2005;16:139–46. 1023–9.
[4] Wiedemann LSM, D́Avila LA, Azevedo DA. Adulteration detection of Brazilian [23] Singh AP, Mukherjee S, Tewari AK, Kalsi WR, Sarpal AS. Determination of
gasoline samples by statistical analysis. Fuel 2005;84:467–73. benzene and total aromatics in commercial gasolines using packed column GC
[5] Moreira LS, D́Avila LA, Azevedo DA. Automotive gasoline quality analysis by and NMR techniques. Fuel 2003;82:23–33.
gas chromatography: study of adulteration. Chromatographia 2003;58:501–5. [24] Bansal V, Krishna GJ, Singh AP, Gupta AK, Sarpal AS. Determination of
[6] Takeshita EV, Rezende RVP, Souza SMAGU, Souza AAU. Influence of solvent hydrocarbons types and oxygenates in motor gasoline: a comparative study by
addition on the physicochemical properties of Brazilian gasoline. Fuel different analytical techniques. Energy Fuels 2008;22:410–5.
2008;87:2168–77. [25] Burri J, Crockett R, Hany R, Rentsch D. Gasoline composition determined by 1H
[7] ANP, Agência Nacional de Petróleo, Gás Natural e Biocombustíveis (Brazilian NMR spectroscopy. Fuel 2004;83:187–93.
National Agency of Petroleum, Natural Gas and Biofuels). <www.anp.com.br>. [26] Meusinger R, Moros R. Determination of octane numbers of gasoline
[8] Flumignan DL, Tininis AG, Ferreira FO, Oliveira JE. Screening Brazilian C compounds from their chemical structure by 13C NMR spectroscopy and
gasoline quality: application of the SIMCA chemometric method to gas neural networks. Fuel 2001;80:613–21.
chromatographic data. Anal Chim Acta 2007;595:128–35.