Analysis of Ethanol in Gasoline by Gas Chromatography and Infrared Spectroscopy

Abstract
The United States government claims that gasoline is 10% ethanol by volume. In this
experiment this claim is tested by using two different methods. Gas Chromatography and Fourier
Transform Infrared Spectroscopy are the two different methods which are used by creating a
standard addition curve. After running both methods FTIR shows three different peaks for
ethanol, giving concentrations to be 9+3%, 11+3%, and 12+2%. All the FTIR concentrations
fell within the expected U.S. standard range. GC on the other hand did not have such a narrow
or close results. The percent volume of ethanol found using the GC is 31+144%. Almost all of
the results statistically were significantly different, which means that they do not agree with each
other within the experimental error.

Introduction
Oil products have become very important for daily living on the modern 21
st
century.
One of the most important and maybe most used, not just here in the United States but
worldwide would be gasoline. Gasoline is a mixture of primarily hydrocarbons and small
amounts of additives (Streva, 2011). For years it has generated energy to keep most of the world
moving by powering most industrial and personal vehicles. With such high demand of gasoline,
added to the decreasing supply of oil, it is no surprise that prices have drastically increased
through the year.
Modern gasoline has many additives that can help reduce harmful emissions as well as
increase miles driven with few amounts of gas needed. Ethanol is one of these such additives.
Ethanol is a two carbon chemical with an OH attachment that helps oxygenate the gasoline,
meaning that it will help the combustion of the gas, creating a cleaner more efficient burn (U.S.
Energy Information Administration, 2013). In 2012 about 134 billion gallons of gasoline were
consumes in the United States alone and 10% of that is ethanol additive (U.S. Energy
Information Administration, 2013). Ethanol then is a cheap additive that extends gasoline
supplies which in turn lowers prices. Some countries such as Brazil use gasoline that contain 20-
25% ethanol by volume in order to increase clean performance and lower gasoline consumption
(Streva, 2011).
To determine the ethanol percent by volume, two different methods of detection and
analysis will be used. The first method will be making a standard addition of ethanol. In a
standard addition an amount of unknown sample, here ethanol in gasoline, is added to several
different vials, each with the same amount of sample. Then in each vial, staring with 0, you
increase the amount of pure 100% ethanol into each vial, increasing its concentration. Then
samples will be ran through a Gas Chromatography (GC) that has a flame Ionization detector
(FID).
GC is an instrument in which gaseous sample or analyte is moved though a long metal
column that has a stationary phase that separates individual chemicals in the sample (Harris
2007). The GC can have a varied detectors that are specialized to detect certain compounds. In
this GC a flame ionization detector will be used. Flame ionization detectors detect hydrocarbons,
like ethanol, by burning the eluent with a mixture of H
2
and air then detecting carbon ions that
are produced (Harris, 2007). Those ions released in the process are then converted to voltage,
amplified and finally turned into a digital signal (Harris, 2007). GC will then separate each
eluent and then the detector will give a signal that shows the time taken to travel and it intensity
of each eluent. Once all the sample have been run, ethanol signals will be put into a linear graph
which will, when x = 0, give the concentration of ethanol.
The second method will use a standard addition sample of ethanol again, but will be
made for a fourier transform infrared spectroscopy (FTIR). Since this instrument is much more
sensitive than the GC, smaller amounts of ethanol are needed when adding to the gasoline in the
vials. FTIR is a charged coupled device that uses light to measure an entire spectrum of
concretions are once. The advantage of using a FTIR is that it has low noise being emitted.
Measurement of Ethanol in GC and FTIR
The Chemistry 316 lab manual was followed to analyze the volume percent of ethanol in
gasoline (Haan, 2014). To analyze the sample in both the GC and the FTIR a standard addition
curve was first made. To make this standard addition curve for the GC, a mixture containing 5
mL of gasoline and 1mL of propanol was made (Haan, 2014). Five different small glass vials,
containing 1 mL of this mixture were then added different amount of pure ethanol (Haan, 2014).
The amounts added where 0 L, 50 L, 100 L, 150 L, and 200 L of ethanol (Haan, 2014).
Since the FTIR is much more sensitive than the GC the standard addition was made different.
The FTIRs standard addition curve was 1 mL of gasoline with five different volumes of ethanol,
these volumes were 0 L, 25 L, 50 L, 75L, and 100L (Haan, 2014).
The percent volumes for each sample where then calculated as shown below along with
calculated percentages of each sample by increasing volume.
FTIR: 25L eth. * 100 GC: 50L eth. * 100
25L eth. + 1000 L gasoline 50L eth + 833 L gasoline

2.44%, 4.76%, 6.98%, 9.09% 5.66%, 10.7%, 15.3%, 19.4%
FTIR Results
A Thermo Scientific Nicolet 380 FTIR machine was used to analyze the FTIR samples.
The samples rendered significant peaks for ethanol at wavelengths 3334.484 m, 1091.566 m, and
1051.066 m (Fig 1). Knowing that the three wavelength peaks were form ethanol, all three peak
heights (absorbance) were graphed against the four ethanol spike % volumes (figure 2-4).

Figure 1. FTIR using a Nicolet 380 of sample containing 50 L spike of ethanol. The three
ethanol peaks detected wavelengths are 3.33e3m, 1.09e3m, and 1.05e3m.

Figure 2. This graph shows the absorbance of the four different % ethanol spikes at peak
wavelength 3.33e3 m.



0.00E+00
5.00E-01
1.00E+00
1.50E+00
2.00E+00
2.50E+00
3.00E+00
3.50E+00
4.00E+00
3.00E+02 8.00E+02 1.30E+03 1.80E+03 2.30E+03 2.80E+03 3.30E+03 3.80E+03
A
b
s
o
b
a
n
c
e

(
a
.
u
.
)

waveleght (m)
FTIR 50 uL Ethanol
y = 6.875E-02x + 6.383E-01
R = 9.970E-01
0.600
0.700
0.800
0.900
1.000
1.100
1.200
1.300
1.400
0.0 2.0 4.0 6.0 8.0 10.0
A
b
o
s
b
a
n
c
e

(
a
.
u
.
)

% volume ethanol
Absobance vs % vol Eth. at wavelenght 3.33e3 m

Figure 3. This graph shows the absorbance of the four different % ethanol spikes at peak
wavelength 1.09e3 m.

Figure 4. This graph shows the absorbance of the four different % ethanol spikes at peak
wavelength 1.05e3 m.
Using the slope from figures 2-4 we can then calculate for the % volume of ethanol on
each peak by using equation (1) (Haan, 2014). Where x is the % volume of ethanol, b the y-
intercept, and m the slope of the line. The % volumes for the three peeks were 9% at 3.33e3m,
11% at 1.09e3m, and 12% at 1.05e3m.

x = b/m (1)
y = 5.237E-02x + 5.523E-01
R = 9.968E-01
0.500
0.600
0.700
0.800
0.900
1.000
1.100
0.0 2.0 4.0 6.0 8.0 10.0
A
b
s
o
b
a
n
c
e

(
a
.
u
.
)

% volume ethanol
Absobance vs % vol Eth. at wavelenght 1.09e3 m
y = 8.957E-02x + 1.044E+00
R = 9.986E-01
1.200
1.300
1.400
1.500
1.600
1.700
1.800
1.900
0.0 2.0 4.0 6.0 8.0 10.0
A
b
s
o
r
b
a
n
c
e

(
a
.
u
.
)

% volume ethanol
Absobance vs % vol Eth. at wavelenght 1.05e3 m
Table 1 shows the concentrations of all three peaks at a 95% confidence interval. To
calculate the 95% confidence interval equation (2) was used. In this equation x is the
concentration, t is the Students t for n-2 degrees of freedom, sx is the sample standard deviation
or standard error (Harris, 2007).
x + tsx (2)

Table 1 also shows the Limit of Detection (LOD) and Limit of Quantification (LOQ) for all three
wavelength peeks. The LOD is the lowest analyte signal that the machine can detect and the
LOQ is the smallest signal that the machine can make an accurate measurement (Harris, 2007).
The equations for LOD and LOQ are shown below (3) and (4).
LOD = 3s/m (3)

LOQ = 10s/m (4)
In both the LOD and LOQ, s is the standard deviation of the three measured blanks and m is the
slope of the line.

Wavelength
Peak
% vol.
Ethanol
LOD
(% vol)
LOQ
(% vol)
3.33E+03 9 + 3 1.03 3.42
1.09E+03 11 + 3 1.21 4.02
1.05E+03 12 + 2 1.33 4.47
Table 1. Table showing the results for the % volume ethanol in each peak wavelength known to
represent ethanol. The table also shows the LOD and LOQ of each peak using the standard
deviation of three different blanks taken for each peak.

GC Results
A GC Hewlett Packard 5890 Series II was used by injecting 1 L of sample in each run.
When analyzing the samples with the machine a retention time of .927 minute was found for the
ethanol. This was found by the increase of area counts at the .927 minute when known higher
volumes of ethanol were injected.
Although four different samples were applied, the spiked 200 L sample seemed to be an
outlier. To be able to get rid of this data point a grubs test at the 95 % confidence interval was
ran using equation (5).
(5)
Xsuspect is the point in question, x is the average of the four points, and s is the standard
deviation. If the G is smaller than the G table where your n-1 being the degrees of freedom. G for
the spiked 200 L sample was 1.10, and when looking at a G test table at n = 3 for limit of
detection G is 1.15. Therefor the point should be kept according to the Grubbs test since G
calculated is smaller than G table. Unfortunately if the 200 L spiked, or 19.4% ethanol is added
to the graph a negative slope is given, which is logically impossible to use in a standard addition
curve is, so although the point passed the G test it was still removed from the data.
After removing the outlier point, the other data was graphed as shown in figure 5. Using
equation 1 and along with the information in figure 5, the % volume of ethanol in the gasoline
sample was calculated at 31 % ethanol. Table 2 shows the confidence interval at 95%, calculated
using equation (2), along with the LOD and LOQ, calculated from equation (3) and (4).


Figure 5. Absorbance vs % volume ethanol standard addition curve. Data gathered from GC
Hewlett Packard 5890 Series II. Only shows points with at 5.66%, 10.7%, and 15.3% ethanol.
The point at 19.4% ethanol was removed due to a negative slope, rendering impossible to use for
standard addition curve.


Table 2. Table showing the results of % volume ethanol of gasoline using a GC. The LOD and
LOQ of the GC when using the standard deviation from three different zero spiked samples.

Comparing GC to FTIR
y = 4.389E+05x + 1.344E+07
R = 9.302E-01
14000000.000
15000000.000
16000000.000
17000000.000
18000000.000
19000000.000
20000000.000
21000000.000
2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
A
b
s
o
r
b
a
n
c
e

(
a
.
u
.
)

% volume ethanol
Abosrbance vs % volume Eth. using GC
% vol Ethanol LOD (% vol) LOQ (% vol)
31 + 144 5.91 19.7
Since there are no t-test that compare three different replicate measurements, there was
no way to run a t-test for all three FTIR peeks simultaneously. Therefore two of the FTIR spiked
samples were compared at a time. The first FTIR peaks compared where peaks 3.33e3m and
1.09e3m as shown in table 2 named FTIR 1. The second FTIR groups analyzed were peaks
3.33e3m and 1.05e3m shown as FTIR 2 in table 2. The third group analyzed was peak 1.05e3m
and 1.09e3m shown as FTIR 3 in table 2. All three groups were statistically analyzed using case
2 or equation (6 ) (Harris, 2007). The FTIR t-test show that there is a significant difference in
two sets of the FTIR samples.
(6)
Again since we had three different comparable answers, each FTIR peak was also
compared with those of the GC using case 2 method (Harris, 2007). All the t-calculated were
greater than the t-table, but a different equation was used to solve for the statistics since the F-
calculated was greater than F-table. Instead of using equation 6, equation 7 was used, with
degree of freedom of 2. Table 3 shows the different t-calculation results for GC and each FTIR
peak.
(7)

Table 3. FTIR peaks 1 analyzes peak at 3.33e3m and 1.09e3. FTIR peaks 2 analyzes peaks at
3.33e3 and 1.05e3. FTIR peaks 3 analyzes peaks 1.05e3 and 1.09e3. T-table numbers came form
315 book (book).
Discussion
The % volume of ethanol that the U.S. claims is in gasoline is 10% by volume. As shown
in table 1, all the FTIR samples agree with this claim by having 10% in their range. At peak
1.05e3 m the range goes form 10% to 14% ethanol in gasoline, almost missing the U.S.
standards. The other two peaks comfortably range in the U.S. standards of ethanol in gasoline,
and all three were tested to be accurate to the 95% confidence interval. The GC samples met the
U.S. standards, but only because it had a high uncertainty as shown in table 2. This sample
t-calculated t-table 95% t-calc comparison
FTIR peaks 1 1.47 2.78 t-cal < t-table
FTIR peaks 2 3.26 2.78 t-cal > t-table
FTIR peaks 3 4.78 2.78 t-cal > t-table
GC/FTIR 3.33e3 14.3 4.3 t-cal > t-table
GC/FTIR 1.09e3 14.3 4.3 t-cal > t-table
GC/FTIR 1.05e3 14.3 4.3 t-cal > t-table
resulted having a 31% volume of ethanol + 144%, therefore the sample could have 0% to 175%
volume ethanol.
The most accurate method would have to be the FTIR. The FTIR results where much
closer to the known amount of 10% volume of ethanol. The GC method gave a 31% volume of
ethanol and although the uncertainty does include the 10%, it is much too large to really know
what the concentration really is. The FTIR is also more precise having all the % volumes one or
two away from the U.S. standards and when looking at each of the coefficient of determination
or R^2, they are all close to 1 by .01. On the other hand the GC method had a huge uncertainty
and its coefficient of determination is off by .07.
Statistically looking at the results will show that only peaks 3.33e3 and 1.09e3 are not
statistically different. On the other hand the other two combinations of peaks observed are
statistically different and do have some correlation. When looking at the FTIR peaks and
comparing each to the GC results you can see in table 3 that they all have a significant difference
and are correlated to each other. Their t-calculated are larger than their t-table, which means they
are significantly different.
All the standard addition curve from both the FTIR and the GC are larger than both their
LOD and LOQ. Although all the results were lower, the FTIR is a more sensitive machine. The
FTIRs highest LOD and LOQ, 1.33 and 4.47, are smaller than that of the GC. The GC LOD and
LOQ were at 5.91 and 19.7. This sensitivity is one of the reasons why FTIR is selected to
analyze ethanol. It is also chosen because ethanol has an O-H bond, which has specific
wavelength peaks that are easily detected. GC, although are not as sensitive, are also selective for
ethanol because you can use detectors like FID that are specifically made to detect hydrocarbons.
Whenever an experiment is conducted some kind of error is present and can have
indeterminate and determinant error present. When analyzing the % volume of ethanol in this
experiment indeterminate error can be present as instrumental noise effecting the results being
put out by each machine. Another noticeable indeterminate error is the students experimental
inexperience in running the experiment. Besides indeterminate error this experiment could also
have determinant error, or errors that should be corrected when noticed. A couple of determinate
errors that could have occurred would be the fact that the instruments used were miscalibrated or
uncalibrated. The determinte error that seems more reasonable for any errors in the experiment
would be parallax error when pipetting solution when using the volumetric pipettes.
Conclusion
The % volume of ethanol was analyzed using two methods selective to ethanol, GC and
FTIR. Both methods rendered different results. FTIR gave three different peaks that were
specifically known for ethanol, these peaks were 3.33e3m, 1.09e3m, and 1.05e3m. These three
peaks rendered a % volume of ethanol to be 9+3%, 11+3%, and 12+2%. These result range are
within the U.S.A. measured standards, confirming their measurements. The GC did not have
such favorable results with a % volume ethanol of 31+144. Although the range fits within the
10% ethanol the U.S. claims, the uncertainty is so high that it is hard to conclude how much
ethanol is really in gasoline. Only one pair of FTIR peaks 3.33e3 and 1.09e3 were found not to
be significantly different with each other, but the rest of the peaks when compared were
significantly different. The GC and all the FTIR peaks were significantly different form each
other. This means that although the FTIR % volume ethanol are similar they do not agree with
each other within the experimental error. This is also true for the GC and FTIR peaks.

























REFERENCES
Haan, J., Experiments for Quantitative Analysis Laboratory. Spring 2014. pg. 16-17
Harris, D. C. (2007). Quantitative Chemical Analysis (7
th
Ed.). New York, NY: W. H. Freeman
and Company.
IHS Chemical. (2014). http://www.ihs.com/products/chemical/planning/ceh/gasoline-octane-
improvers.aspx viewed on February 4, 2014.
Streva, E. , Pasa, V. , & Sodr, J. (2011). Aging effects on gasolineethanol blend properties and
composition. Fuel, 90(1), 215-219.
U.S. Energy Information Administration. (April 12, 2013).
http://www.eia.gov/tools/faqs/faq.cfm?id=27&t=10 viewed on February 4, 2014.