University of Gothenburg and
Chalmers University of Technology
CO and C3H8 measurements
with
FTIR spectroscopy
By
Malin Olsén
Jorge Castro
Mikael Johansson
10 November, 1998
Abstract
We have used Fourier Transform InfraRed spectroscopy, FTIR, to determine the concentrations of the exhaust gases: carbon monoxide (CO) and propane (C3H8). The samples where exhaust from a Ford Escort -86 with automatic choke. We applied Beer-Lambert’s law to calculate the concentrations of CO and C3H8. We compared the samples when starting the car with a cold and a warm engine.
The concentration of CO for cold engine and warm engine was 1.34 - 2.24 % and 0.76 - 1.27 % respectively. The concentrations of C3H8 for the cold engine was 0.07 % and under detection limit when the engine was warm. We found that a warm engine causes less pollution due to a more effective combustion.
Contents
1 Introduction *
2 FTIR *
2.1 The Michelson interferometer
*3 Beer-Lambert’s Law *
4 Experiment methods *
4.1 Measurements
*4.2 Analysis and Calculations
*5 Results *
6 Discussion *
7 References *
Carbon monoxide, CO, is interesting because it is one of the most common polluting gases in the atmosphere with a concentration of 0.1 ppm; 100 ppm have been detected in heavily trafficked roads and urban areas. Most of the CO has antropogenic origin. It is formed at incomplete combustion in e. g. gasoline engines. In Sweden, the vehicles produce about 3 million tons, 2/3 of the yearly discharge. The presence of propane, C3H8, is an evidence of low combustion rate, which happens when the oxygen concentration is low, namely the larger hydrocarbons breaks down to shorter chains.
Our task was to examine exhaust gases, especially CO and C3H8 from an engine, using infrared spectroscopy. We found it interesting to study exhaust from a car that was started when the engine was cold and when it was started with a warm engine, in this report we will call the first case CE and the second WE.
2 FTIR
In a molecule, atoms vibrate and rotate relative to their centre of mass. If the vibration energy is small the motion can be approximated to Simple Harmonic Motion, SHM. These vibrational and rotatioal phenomena are used to determine the compounds of matter and its structure. In order to do this, the interaction between radiation and matter may be used. There are different techniques to analyse this interaction. Fourier Transform InfraRed Spectroscopy, FTIR, has given good results in analyses of gas compounds. FTIR-technology offers a unique flexibility considering gases. The absorbance can be determined in all or parts of the infrared (IR) region, not only for one specific wavelength. The radiation absorbed by the molecule corresponds to the difference in vibrational and rotational energy levels. The least concentration that can be detected with FTIR is about 0.2 ppm.
2.1 The Michelson interferometer
There are many different interferometers but a commonly used one is the Michelson interferometer. This is a device that splits an electromagnetic beam in two directions to recombine them later so that the intensity variation, registered by a detector, can be determined as a function of the path difference between them. The interferometer contains two orthogonal mirrors: one movable and one fixed, as shown in Figure 1. If the fixed and the movable mirrors are located at the same distance from the beam splitter, the two rays are in phase relative to each other. At this point, the rays interfere constructively so the beam that reaches the detector has the intensity of the sum of the two rays and has the same intensity as the source. Interference happens when two beams recombine.
Figure 1. The principle of the Michelson interferometer.
If the movable mirror moves a distance of 
, where 
is the wavelength, the retardation will be 
, resulting in destructive interference at the detector. With further removal of 
, the rays will be in phase again and constructive interference occurs. The mirror moves with constant velocity and a sine signal will be detected with a maximum, when the retardation passes 
, where 
is a natural number. The signal is a function of the retardation, 
, which is represented as 
. The intensity of the signal at each point, 
, is the same as the source intensity,
. For other values of the size of the signal is
.
If 
the expression becomes
.
The interferogram from a monochromatic source can be expressed by the equation
.
But the amplitude of the detector signal does not only depend on the intensity of the source but also of the system. 
must be corrected with a factor 
so that it yields
,
which can be written as
.
The parameter 
gives the intensity of the source modified to the system. 
is a cosine Fourier transform of . The Fourier transform of gives the spectra and that is what the detector register to make an interferogram. Now the source can be polychromatic, the interferogram at each point is the sum of the interference from all incoming wavelengths. If the source has continually wavelength the interferogram can be described as
.
When Fourier transforming 
, the intensity of the source,, can be calculated as
.
Now the intensity can be described at each and every wavelength area. describes the source’s blackbody distribution, which is energy reduced in the specific wavelength area where the sample absorbs energy. The FTIR is able to measure the absorption spectra of the sample, the so-called fingerprints of the gas compounds. Beer-Lambert’s law is applied to compute the concentration of the gas compounds.
Assume we have a cell with an absorbing gas (see Figure 2), the gas is exposed with light of intensity, I0, through the sample the intensity I decreases and the intensity, It, is transmitted. The intensity decreases depending on the length of the cell, b, the number of the absorbing molecules in the cell (the concentration), c, and a constant, k, that depends on the molecule’s cross-section. If we study a volume element we get the relation:
,
the intensity, I, before the current interval must considered. The minus sign appears because the intensity decreases.
Figure 2: Illustration of how the intensity decreases trough the sample. I0 is the source intensity, I the reduced intensity, It the transmitted intensity and b is the length of cell.
Integration of
,
gives
.
The absorbance, 
is defined as
and the transmittance, 
, as
.
With small values of 
in Equation (1), we can approximate the logarithmic dependence of the absorbance to a linear one. The linearity gives the following expression for the unknown concentration, cdet,
(2),
where ccal is the concentration of the calibration gas (which is known) and Acal and Adet are the absorbance of each sample (the result from the FTIR measurements).
The FTIR, model 1725X, and it had a detector of Triglycine Sulphate. The source was MIR and the windows of the cell where made of Zink-Celenide.
To collect the gas we used balloons and a pet-bottle. We put the balloons on the bottle and it was then placed in the exhaust pipe, this is illustrated in Figure 3. The bottom was cut off from the bottle and to fit the pipe, the bottom was made smaller.
Figure 3. The samples were taken with a pet-bottle and a balloon,
which was placed in the exhaust-pipe.
Before we put the pet-bottle in place we waited a second to let the air in the system be blown out. When the balloon was full we pulled the bottle out of the exhaust pipe. We repeated this process six times. The number of measurements is shown in Table 1.
Table 1. The number of samples from starting the engine with a cold engine, CE, and with a warm engine, WE.
The measurements gave us 19 samples, which we scanned once in the FTIR-spectrometer and got a spectrum for each sample. We took mean values of the CE spectra and of the WE spectra as well. Determination of the concentration can be done according to Equation (2).
The spectra of the mean values for CE and WE are plotted in Figure 4 with calibrations of CO, C3H8, H20 and CO2. The exhaust constitutes mostly of CO2 and H20. These two gases do not interfere with CO and C3H8, that is they absorbe IR for different wavenumbers, see Figure 4.
Calibration spectra and our average measurements
Figure 4. The peak for wavenumber 
is carbondioxide. For 
and 
we can identify water. But the most interesting is 
which is CO and 
that is C3H8. The conclusion is that the interference from water and carbondioxide is minimal for the relevant wavenumbers of CO and C3H8.
To determine the concentration of CO we have used two calibration gases with concentrations of 0.6-1.0 % and 100 ppm. The spectra of these calibration gases and the spectra of CE and WE are shown in Figure 5.
Carbon monoxide spectra
Figure 5. The symmetry of the spectra of CO is due to the molecule structure. The figure shows that the amount of CO is higher for CE than WE. The red line is the calibration gas with a concentration of 0.6-1.0 % and the cyan line has a concentration of 100 ppm.
With Figure 5 as a basis, we computed the concentration of CO. We compared the absorbance of the calibration gases with the absorbance of the samples, the result should be the same for every wavenumber. We determined the concentration for six different wavenumbers, see Table 2, and took the mean value.
Table 2. The concentration, c, of CO for CE and WE at different wavenumbers.
The absorbance of C3H8 is more complex due to its structure. To calculate the concentration we looked for wavenumbers where the curves are symmetrical, see Figure 6. For concentration of C3H8 with cold start, see Table 3.
Propane spectra
Figure 6. Absorbance spectra of the average values for CE in the C3H8 region. It also shows that the absorbance for WE is under detection limit.
Table 3. The concentration of C3H8 for CE at different wavenumbers.
The concentration of CO for CE was 1.34-2.24 % and for WE 0.76-1.27 %. The concentration of C3H8 is for CE 0.07 % and for WE the rate was too low to determine.
We found that the combustion is better if the engine is warm. This can be explained with the fact that the temperature in the pistons is higher and therefore easier to exploit the energy in the molecule.
We assumed that the logarithmic dependence in Equation (1) can be approximated to a linear dependence because of the low concentration rates of CO and C3H8. This means that the relation between the concentrations should be constant. We tested this assumption by dividing the concentrations of our samples with the concentration of our calibration. The result showed that our assumption is acceptable for the low concentrations in our samples. This procedure was repeated for both of the gases.
We found that it is easy to calculate the concentration of diatomic molecules, because of the symmetry in the spectra. But for more complicated molecules, e.g. C3H8, the pattern will be more difficult to analyse. We had two different calibration spectra to compare our measurements with, so we could see which wavenumbers that had the same development in three spectra.
The concentration rates of absorbing gases might be influenced by the improved capacity because of higher temperature of the gas. In the CO molecule the rotational energy levels change when the gas temperature increases, resulting in higher values of absorbance. We believe that this has not happened in our case, because our samples had reached room temperature and therefore had the same temperature as our calibration gas.
The approved concentration limit for CO in vehicle exhaust is in Sweden 3.5 % for cars manufactured 1986, no such limit are available for C3H8. Our results show that our samples are under this limit, both when starting the car with a cold engine and a warm engine. An engine heater would improve the combustion rate in order to decrease the amount of exhaust gases, at least for CO and C3H8.