TUNABLE DIODE LASER SPECTROSCOPY FOR HIGH TEMPERATURE COMBUSTION CONTROL
Ketil Gorm Paulsen Dung Do Dang
Neo Monitors AS Norsk Elektro Optikk AS
Solheimveien 62 A Solheimveien 62 A
1471 Lørenskog, Norway 1471 Lørenskog, Norway
TUNABLE DIODE LASER, SPECTROSCOPY, CO, O2, HIGH TEMPERATURE, OPTICAL FIBER, COMBUSTION CONTROL
Gas analyzers based on tunable diode laser absorption spectroscopy (TDLAS) have over the recent years seen increased use in process control due to the possibility of non-contact in-situ measurements, high dynamic range, and negligible cross-interference from background gases. However, high temperature applications have been a challenge due to high process temperatures as well as very high ambient temperatures for the laser electronics. This paper presents an approach that makes measurements at high ambient temperatures feasible. By utilizing wavelengths around 760 nm and 2,3 m combined with fiber optics, remote alignment as well as state of the art signal processing it is possible to enhance the benefits of TDLAS. Oxygen (O2) and ppm levels of carbon monoxide (CO) have successfully been measured across stack in industrial processes with gas temperatures up to 1500 °C.
TDLAS based gas analyzers introduced the possibility for non-contact in-situ measurements in combination with high dynamic range and negligible cross-interference from background gases. These features have probably been main drivers for an increase in installed base as well as the main reasons for using TDLAS in a multitude of applications. With better understanding of environmental issues came increased focus on combustion processes: Better and more accurate measurements of carbon monoxide (CO) as close as possible to the combustion zone can be used for optimization of the combustion which will result in a significant reduction of polluting gases with nitrogen oxide (NOx) as the most important. A secondary but most welcome result is a significant reduction of power consumption.
In the near-infrared region carbon dioxide (CO) is normally measured at the second overtone around 1600 nm but the sensitivity requirements combined with potential interference from other gases from the combustion like water vapor (H2O) and carbon dioxide (CO2), makes it necessary to move to the first overtone at 2300 nm. O2 absorbs at 760 nm where cross-interference from other gases is very limited. This physical phenomenon will in many cases be a limiting factor for the use of TDLAS technique. By searching spectroscopic databases such as HITRAN  or for higher temperature HITEMP , it is possible to identify suitable absorption lines. Interference-free measurements are obtained when the measured line(s) are sufficiently separated from lines belonging to other gases present in the process. Figure 1 shows the dependence of the transmission and line shape on the gas temperature for oxygen (c=4%).
FIGURE 1. CHANGES IN O2 ABSORPTION LINES WITH INCREASED TEMPERATURE
The telecom industry has been a major driver in laser as well as fiber development and several commercially available TDLAS analyzers utilize optical fibers to improve flexibility and access to processes. The downside is however the introduction of more optical noise that reduces the signal to noise ratio. Additionally, as optical fibers and other components have been developed for the telecom wavelengths between 1300 nm and 1700 nm, fiber solutions for oxygen (O2) as well as low concentration measurements of carbon monoxide (CO) have not been successful. These measurements have thus been made commercially available with TDLAS analyzers using traditional free space optics.
Commercially available TDLS instruments were modified to measure the concentration of oxygen (O2) and carbon monoxide (CO) at high temperatures.
FIGURE 2. BASIC SETUP FOR FIBER-COUPLED TDLS IN HIGH TEMPERATURE MEASUREMENTS FOR COMBUSTION CONTROL
Laser sources 1 and 2 in figure 2 are guided by optical fibers to one transmitter unit. A single lens is used to collimate both laser beams across the stack. On the receiver side a beam splitter is used to separate the two wavelengths to two detectors, one for each laser wavelength. Second harmonic detection is performed on each detector signal and the electronics provides concentration signals for each gas. This setup allows moving a significant amount of electronics away from the process entry points where the ambient temperature is close to 100 °C. Remote laser - or rather fiber - alignment is maintained by a servo controlled system.
USE OF FIBERS IN TDLAS
Optical fibers consist of a doped silica (SiO2) core and a surrounding layer of lower refractive index glass with a typical diameter of 125 m. Light is guided in the core due to total internal reflections as indicated in figure 3 . The core and the cladding are protected by an acrylate coating layer. Single-mode fibers differ from multi-mode fibers mainly by the much smaller core diameter: Less than 10 m for single-mode fibers compared to more than 50 m for multi-mode fibers.
FIGURE 3. PROPAGATION OF LASER LIGHT IN AN OPTICAL FIBER 
Both single- and multi-mode fibers have transmission losses which are given by:
I = I0exp(-αL) (1)
I is the intensity of the light received,
I0 is the intensity of the light transmitted,
α is the loss factor of a given fiber,
L is the length of the fiber.
There are several physical reasons for losses, mainly Rayleigh scattering, silica impurities, water absorption, and at higher wavelengths silica lattice interaction. All these losses are summarized in the loss factor α.
At wavelengths above 2 m the loss due to lattice interaction grows very quickly. Above 2 m only a few meters of optical fiber will cause significant losses as illustrated in figure 4. This limits the practical separation of the laser and transmitter unit in fiber coupled TDLAS applications.
FIGURE 4. LOSSES IN OPTICAL FIBER AS FUNCTION OF WAVELENGTH 
Additional losses are introduced by bending the fiber. Large bending losses occur when the radius is close to the critical bending radius Rc
is the wavelength,
n1 is the refractive index of the core,
n2 is the refractive index of the cladding.
Tunable Diode Laser Absorption Spectrometers are used to measure gas concentrations based on Beer-Lambert’s law. It defines the relation between the transmission of the laser beam and the concentration of a gas sample as follows:
T = exp(-Sg(f)NL) (3)
T is the transmission,
S is the absorption line strength,
g(f) is the line shape function,
N is the concentration of absorbing gas molecules,
and L is the optical path length.
By knowing S, g(f), and L, and measuring T the concentration N can be calculated. The conventional way of obtaining such quantitative information is to compare the measured response to a prior performed calibration.
A standard measurement scheme is utilized: The laser is temperature stabilized so that the emitting wavelength coincides with the center wavelength of the target absorption line. The laser frequency is scanned over the absorption line with a repetition rate of 50 – 100 Hz. A high frequency modulation of the laser frequency is applied on top of the wavelength scan. A detector located at the receiver side measures the light absorption caused by the target gas molecules. The second harmonic (2f) component of the absorption signal is selected for further analysis by using a lock-in amplifier for the measurement. A more detailed description of the measurement principle is given by Linnerud et al .
HIGH TEMPERATURE MEASUREMENTS
Using fibers without introducing too high losses due to coupling is one of the major challenges of the setup. To reduce the amount of high temperature tests, initial work has been done with a sample of carbon monoxide (CO) at room temperature to estimate the expected absorption at high temperatures. The entire measurement setup was later moved to the initial field tests with high temperature process gases.
UTILZATION OF FIBER FOR LASER IN THE 2.3 m RANGE
Initial measurements have been performed with a cell with an optical path length of L = 2mm and a carbon monoxide (CO) concentration of c = 5%. A standard coupled laser at approximately 2,3 m and single-mode fiber were used. A measured spectrum is given in figure 5 a). The signal to noise ratio (SNR) is used to quantify the quality of a measured signal and is given by:
Psignal is the power of the signal,
Pnoise is the power of the noise,
Asignal is the amplitude of the signal,
Anoise is the amplitude of the noise.
The signal to noise ratio for the spectrum in figure 5 a) was approximately 4. Use of single-mode fibers caused unexpected high bend losses, and a bend radius of a few cm led to significant losses in optical power.
These results encouraged tests with the single-mode fiber replaced by a multi-mode fiber in an otherwise identical setup. The measured spectrum is illustrated in figure 5 b). This improved the signal to noise ratio to approximately 110, making it possible to measure ppm levels of carbon monoxide (CO) at room temperature. Use of multi-mode fiber improved the bending losses as well. This setup was judged to be sufficiently robust to enter the field tests, and calibration to 500 ppm carbon monoxide (CO) was made prior to shipment.
FIGURE 5. SIGNAL TO NOISE RATIOS CARBON MONOXIDE MEASURE-MENTS WITH SINGLE-MODE AND MULTI-MODE FIBERS, RESPECTIVELY.
EXPERIENCE FROM FIELD TESTS
Field tests were carried out at a glass factory in the Netherlands. The basic process in their facility is to heat glass pellets and melt into fluid glass which is used for bottle production. Oxygen is added to obtain optimum process conditions, and the exhaust consists mainly of CO and water vapor. As a part of an energy conservation scheme, the gas flow direction is reversed at regular intervals, i.e. input becomes output and vice versa. In this setup the optical path length was approximately 3 m and the ambient temperature near the stack was around 100 °C. The transmitter unit was air cooled to 65 °C and the receiver unit was water cooled to 75 °C. The test setup was installed in parallel to existing water cooled TDLAS analyzers operative on site. The field test process is schematically illustrated in figure 6.
FIGURE 6. SCHEMATIC ILLUSTRATION OF FIELD TEST PROCESS
Long term measurements were performed over a period of four months at input and output and measured concentrations as well as optical transmission were constantly logged.
Examples of long term measurements are given in figure 7. The carbon monoxide concentration in a typical cycle is shown in figure 7a. During the first part of the cycle (the first 20 minutes) no carbon monoxide was measured at the measurement point. Then the flow was reversed, and in the remaining time, the carbon monoxide concentration varies significantly. This is consistent with results from earlier performed open path TDLS measurements. Figure 7b shows the first part of the cycle in a different scale and a signal corresponding to 10 ppm of CO is observed.
FIGURE 7. CONCENTRATIONS AT INPUT/OUTPUT AND TRANSMISSION
Second harmonic signals of the carbon monoxide measurements are illustrated in figure 8. The carbon monoxide signals for 4022ppm and 261ppm are measured when the port is operated as output. The water vapour (H2O) concentration is close to 25%. The two H2O lines can be seen on left hand side of the CO line. While the CO concentration is changing the water lines are remaining constant. Figure 8 shows the CO concentration when this port is used as input of O2. In this case both CO and H2O concentrations are reduced.
FIGURE 8. SECOND HARMONIC SIGNALS FROM A CO MEASUREMENT
Single-mode and multi-mode fibers that originally were made to work in the telecom spectral range from 1300 to 1700 nm have successfully been tested for both oxygen (760 nm) and carbon monoxide (2300 nm). Laboratory tests showed that utilization of optical fibers even with lasers around 2,3 m is possible with acceptable optical losses.
Several very critical issues related to mounting and treatment of the fibers have been solved and a concept making use of a standard multi-mode fiber for wavelengths at 760 nm and 2300 nm have been qualified. However, due to the very high attenuation for light sources above 2 m practical fiber lengths are limited to a few meters. For oxygen at 760nm the fiber length can be increased because of lower attenuation for this wavelength. The noise introduced by the fiber was reduced to an acceptable level and field tests have shown that carbon monoxide can be detected at ppm levels at temperatures up to 1500 °C.
Still the fiber coupled laser solution increases the optical noise by a factor of 5 to 7 compared to free air lasers. This leads to a detection limit corresponding to relative absorption for laser sources at 760nm and 2,3 m. Similar detection limits can be expected for fiber coupled laser sources in the telecom wavelength range.
The authors would like to acknowledge Dr. Peter Geiser at Norsk Elektro Optikk AS for his simulations as well his splendid work as discussion partner and Mr. Peter Kaspersen at Norsk Elektro Optikk AS for his inputs and in particular for his ability to communicate an overview. The authors would also like to thank Mr. Marco van Kersbergen at TNO for his cooperation at the field test facilities in Dongen, Netherlands.
1.L.S. Rothman, I.E. Gordon, A. Barbe, D.C. Benner, P.F. Bernath, M. Birk, et al, “The HITRAN 2008 molecular spectroscopic database,” JQSRT 110, 533-572 (2009).
2.L.S. Rothman, I.E. Gordon, R.J. Barber, H. Dothe, R.R. Gamache, A. Goldman, V. Perevalov, S.A. Tashkun, J. Tennyson, “HITEMP, the high-temperature molecular spectroscopic database”, JQSRT 111, 2139-2150 (2010)
3.J. M. Senior, “Optical Fiber Communications – Principles and practice”, Prentice-Hall International Series in Optoelectronics, 1985