Simultaneous Determination of SO, and NO, Concentrations by Ultraviolet Spectro- TECHNICAL PAPER hotometry Phil Harris PRESENTED AT. ISA Analysis Division Spring Symposium, New Orleans, LA April, 1997 AMETEK’ WESTERN RESEARCHSimultaneous Determination of SO, and NO, Concentrations by Ultraviolet Spectrophotometry Phil Harris Senior Scientist BOVAR Western Research 8 Manning Close NE Calgary, Alberta, Canada KEYWORDS Sulfur Dioxide, Nitric Oxides, Ultraviolet Absorption, Emissions Monitoring ABSTRACT A multicomponent stack gas analyzer, capable of simultaneous measurement of SO;, NO and NO; has been developed and proven. This analyzer employs a unique, high resolution ultraviolet (UV) photometer, with many advantages over other technologies in this application. UV photometry is the industry accepted standard for SO, measurements. In addition, itis argued that UV photometry has ‘many benefits over chemiluminescence for NO, measurement. Ultraviolet absorption photometry ‘enables direct, simultaneous measurement of NO, NO;, and SO; with minimal span drift. This method does not suffer the interference or quenching effects caused by water and CO, as observed on chemiluminescence analyzers. UV photometry allows direct measurement of NO; without the use of a thermal converter- an important factor in measuring the higher relative levels of NO; found in emissions from low-NO, gas turbines. The sample system associated with a UV analyzer is simpler as it does ont require a vacuum system, dilution probes, or capillaries. Finally, there are lower capital and operating ‘costs associated with a multi-component UV analyzer.INTRODUCTION ‘The measurement of NO, emissions levels is of primary importance due to the adverse effects which result from elevated concentrations of these species in ambient air. These effects include acidic deposition on natural resources, increased levels of photochemical smog with the associated impacts on public health, and reduced yields of crops. Currently, electrical power generation, particularly from coal fired boilers, accounts for approximately 40% of the total NO, emissions in North America. The majority of boilers use coal or oil as a fuel source, and thus produce emissions of both SO; and NO,. ‘These facilities are required to perform continuous on-line monitoring of the emissions levels for SO, and NO, and have specific performance requirements. The accuracy of the monitoring system is periodically proven during a Relative Accuracy Test Audit (RATA). In such testing, the source tester is required to use methods which are approved by the Environmental Protection Agency (EPA). Most source testers prefer to use instrumental rather than wet chemical methods. Instrumental methods are simpler and faster to implement, provide immediate results and feedback, and can be performed at lower cost to the client. The instrumental methods for NO, and SO, which have achieved high levels of acceptance are the chemiluminescent method for NO, (Method 7E) and the ultraviolet absorption photometric method for SO, (Method 6C). Given that the majority of facilities which contribute to NO, emissions also emit SO,, the value of a single analytical technique which could accurately and simultaneously determine the SO, and NO, concentrations is apparent. The technique should maintain the simplicity of the standard method for SO,, provide equivalent accuracy for NO, as the reference method, and be able to determine total NO, directly without the need for chemical reagents or converters. Ultraviolet absorption photometry has the capability to unify the measurement of NO, and SO;, Such analysis can be performed accurately and reliably in the field. The analytical technique has been proven in a number of continuous stack emissions monitoring (CSEM) applications [1]-The method can be used to determine the concentration of NO, NO; and SO; directly, with no reagents, no chemical converters, and a simple, direct extractive sample system with no vacuum system requirements. In fact, the method can be used without the need for a sample drier, provided that the sample cell of the analyzer and the sample lines be maintained above the dewpoint of the analyte gas. Ithas been proposed to the EPA that a new instrumental method for NO, (Method 7F) be promulgated. In this method, the concentration of NO, in a sample gas is determined by direct ultraviolet absorption spectroscopy. The method has sufficient analytical accuracy to meet the requirements of most emissions monitoring applications for power and secondary boilers. In such applications, the same analyzer may be used to determine the SO, concentration. It is also applicable to gas turbines and gas compressors, where the ability to directly measure the concentration of NO, without the need for a converter and converter switching is a distinct asset. Since the analytical technique involves only direct extraction of a sample of analyte gas, drying the sample if required, and allowing the analyte gas to flow through the sample cell of the analyzer, the sampling methodology is robust and reliable.ANALYTICAL ADVANTAGES TO UV METHOD The proposed method presents a number of advantages over the existing instrumental method in terms of the analytical technique, sample system reliability, and costs to both the source tester and his client. The analytical advantages of the UV method are best described by performing a direct comparison of this technique to the existing instrumental method. NO INTERFERENCE OR QUENCHING FROM COMMON STACK GAS CONSTITUENTS The primary species present in combustion source stack gas are typically nitrogen, water vapor, carbon dioxide and oxygen. These species have no significant absorption in the near UV region of the spectrum where NO, NO, and SO, are measured. In addition, there are no matrix effects in gas phase UV spectroscopy, provided that the light source intensity is sufficiently low to prevent photolysis of the sample. This condition can easily be met by ensuring that the wavelength selection is performed prior to allowing the gas to pass through the sample cell. Chemiluminescent NO, analyzers do not exhibit any direct interference from the common stack gas constituents, However, there are significant matrix effects. Both CO, and water vapor quench the chemiluminescent reaction [2]. NO, can be measured through the chemiluminescent reaction of ozone with NO. When ozone and NO react, NO; is produced in an electronically excited state. The excited NO, can return to the ground state by either giving off a photon of light, or by transferring the energy to another chemical species. Chemiluminescent NO, analyzers are operated at low pressures (typically -25 to -30 inches H,0) to minimize the number of collisions between gas molecules, and thus prevent the transfer of energy from the excited NO; to other gas molecules. However, when the composition of the gas is changed, and especially when polyatomic molecules like CO, and HO are present in the stream, the energy transfer efficiency changes. In a typical case with N, and NO in the analyte stream, approximately 10 % of the excited NO; molecules may give off a photon. The span factor for the instrument is set to relate the number of photons produced per molecule of NO in the feed stream, given this degree of quenching. When the analyte stream ‘composition is changed, to include CO,, a new span factor is required to account for the increased ability of the carrier gas to absorb the energy from NO, before the chemiluminescence reaction occurs. In general, the degree of quenching is significantly greater for water vapor than it is for CO;. However, water vapor is for the most part removed by the drier in the sample system. CO, quenching has been reported numerous times in the literature, and documents pertaining to CO; quenching are available from all specialty gas ‘manufacturers to address client concems when EPA protocol | calibration gases which contain NO blended with CO; in nitrogen carrier gas are used to calibrate chemiluminescent NO, analyzers. Such quenching effects are not observed on ultraviolet absorption photometric analyzers [3]. DIRECT MEASUREMENT OF NO, WITHOUT THE USE OF A THERMAL CONVERTER NO, can be measured directly on an ultraviolet absorption analyzer, with no need for chemical reagents or thermal conversion to NO. The NO; measurement is interference free from SO, and NO, and can beperformed with sub-ppm levels of accuracy. Given that ultraviolet absorption analyzers do not exhibit span drift, the NO, accuracy need only be checked periodically in a laboratory environment, To measure NO; on a chemiluminescent analyzer, it must first be converted to NO. The background NO is first determined with the NO, converter bypassed. The total NO is then determined with the sample gas passing through the thermal converter. The NO; concentration is then calculated by subtracting these two results. This process does not provide an accurate measurement of NO,, since the NO, concentration is typically small relative to the NO concentration. Assuming that the NO; concentration is 5 % of the total ‘NO, and that the accuracy of each of the individual signals is +/- 1 %, the NO, concentration thus calculated will have a typical accuracy of approximately +/- 30%. Furthermore, each time a RATA test is performed a converter efficiency test must also be performed. The efficiency of the thermal converter has been observed to degrade substantially over time, and is strongly dependent on the water content of the sample stream. The NO, converter efficiency requirements of Method 7E contribute significantly to the time required to perform a RATA test, and can lead to delays while a converter is regenerated o replaced. SPAN DRIFT Ultraviolet absorption photometry, when performed at sufficiently high resolution, may be viewed as a first principle measurement of the concentrations of the analyte gas. The molar extinction coefficients of a given species are a fundamental property of the material, and have no significant temperature or pressure dependance over the range of environmental conditions occurring ina RATA test. There is a gas density effect, but this is known to follow the ideal gas law. Since these devices operate at ambient temperatures and pressures in the source testers' trailer, typical variations in pressure and temperature are only about one ‘or two percent (assuming 10 mmHg pressure change in 760 mmHg, and temperature variations on the order of 3 K) Furthermore, including pressure and temperature monitoring in the analytical bench can completely negate this effect. Provided that proper temperature and pressure compensation is performed, there should benno span drift on a well designed UV absorption analyzer. In practical applications, these analyzers have displayed span drifts of less than 1% of full scale over months of continuous operation. In addition, the span factor is extremely reproducible from analyzer to analyzer, with typical variations being less than 2 %. Chemiluminescent analyzers either use dilution probes or vacuum systems to minimize the effects of quenching on the analytical system. Each of these methods rely on the performance of a critical orifice or capillary to produce a constant flow rate into the analytical system. Changes in process pressure, pumping efficiencies, or contamination of the critical orifice can result in significant flow or pressure changes at the analytical system. The response of the analyzer changes in direct proportion to these variations. Furthermore, the response factor of the analyzer is a strong function of the detector electronics (high voltage supply to the PMT). Thus, span variations are common on such devices, and one does not expect a reproducible span setting from test to test. DIRECT AND SIMULTANEOUS MEASUREMENT OF NO, NO; AND SO; ‘The ultraviolet absorption method provides direct and simultaneous measurement of all of the components of interest. The concentration of NO, NO;, SO, and total NO, are determined on a real-time basis on a singleanalyzer. The use of a single analyzer for this measurement simplifies the plumbing requirements of the analytical system being used in the RATA test, Furthermore, any sample system bias caused by air leaks into the system will affect all species equally. Thus, these effects can be readily observed as they will introduce a systematic bias relative to the plant CSEM. In contrast, an instrument rack configured for the measurement of SO, and NO, using the existing methods requires that the sample flow be split after the drier to direct a portion of the flow to both analyzers. Since the analyzers are effectively looking at separate sample streams, and using different analytical techniques, it can be problematic to determine the presence of a leak in the separate portions of the sample system. POTENTIAL FOR HOT/WET ANALYSIS Ultraviolet absorption photometry suffers no interference from water vapor, and thus may be employed on a direct extractive, hot/wet basis eliminating the need for a sample drier in the sample system. This removes some of the potential for sample bias, since both NO; and SO, are water soluble. Furthermore, the resultant concentrations are direct measurements of the actual concentration of the analyte species in the stack, and do not require water vapor content corrections. Assuming that a stack velocity measurement is also performed, the mass emissions of SO, and NO, may be calculated directly. Due to water vapor quenching and converter efficiency effects, chemiluminescence analysis can not be performed on a hot wet basis without also employing dilution. PRINCIPLE OF OPERATION To perform the simultaneous analysis of $02, NO and NO2 the analytical bench must possess the following attributes: 1) sufficient baseline stability and signal to noise ratio to measure low concentrations of the analyte species, 2) exceptional reproducibility of response such that the effects of the overlapping spectral bands may be minimized, 3) high spectral resolution to ensure linearity of response. Baseline stability is determined primarily by the choice of optical components and by the design of the photometer [4]. Dual beam spectrometers provide the most accurate and reliable measurements available in absorption spectroscopy [5], and minimize the effects of light souree drift. Reproducibility of response in absorption analyzers is determined primarily by the wavelength stability of the photometer. Wavelength drift may result in variations in both span and linearity. Hollow cathode lamps act as wavelength reference lamps for many applications, but may also be used as the light source in a UV photometer. The use of such lamps provides a reliable and simple method to ensure that there is no wavelength drift in the spectrometer. In addition, the extremely narrow bandwidth of the emission lines provides unparalleled resolution, and thus ensures the linearity of response of the analyzer [6,7]. The Western Research 900 series of photometers has been specifically designed to meet these requirements. The design of this analyzer, in comparison to other photometers, has been discussed in other papers and presentations [4,6]. Of particular interest to this specific application is the absorption profiles of the species of interest. While all chemical species exhibit absorbance in some portion of the ultraviolet spectrum (from 10‘nanometers to 400 nanometers), only those with relatively low energy excited states can be observed in the near UV region from 200 to 400 nanometers. Below 200 nanometers, the common constituents of air (such as oxygen, nitrogen, carbon dioxide and water) begin to absorb, and thus this region is typically not used for process or emissions monitoring applications. However, no absorption from oxygen, carbon dioxide, nitrogen or water vapor is observed in the near ultraviolet region of the spectrum. This allows near ultraviolet absorption spectroscopy to be used reliably for the determination of those trace compounds which do display an absorbance in this wavelength region. Nitric oxide absorbs in the 200 to 230 nanometer of the near UV spectrum (see Figure 1). The absorption in this region is the result of electronic transitions from the ground state to the lowest first excited state. The three principal bands observed are the result of differing vibrational transitions (as would be observed in the infrared) occurring during the electronic transitions. FIGURE 1. THE NEAR ULTRAVIOLET SPECTRUM OF NO The UV absorption spectrum of sulfur dioxide exhibits bands due to two distinctly separated electronic transitions. The bands in the 200 to 230 nm region (Fig 2) of the spectrum are due to an electronic transition from the ground state to the second excited state, while the bands in the 250 to 350 nm (Fig 3) region are due to a transition from the ground state to the first excited state. Each of these electronic bands exhibits fine structure due to rotational and vibrational transitions. In this application, it is particularly important to note that the sulfur dioxide spectrum exhibits a local ‘minimum in the vicinity of 215 nanometers, which coincides with one of the local maximum of the nitric oxide spectrum (Figure 1). This is important in that it allows for the measurement of NO using the large absorption band at 214.8 nanometers without concem that the signal will be "swamped" by the SO; absorption Nitrogen Dioxide absorbs throughout the ultraviolet and visible regions of the spectrum. The absorbtion in the visible (primarily from the blue to green regions of the spectrum) gives rise to the characteristic ‘orange-brown color of NO2 gas. The absorption spectrum consists of a large number of mostly ill-defined ‘bands each of which consists of a large number of irregularly spaced lines. The absorption maximum occurs at about 400 nanometers (see figures 4 and 5).FIGURE 2. THE NEAR UV SPECTRUM OF SO, (210 TO 230 nm) FIGURE3. THE NEAR UV SPECTRUM OF SO, (230 TO 350 nm) Based on the data presented in the previous figures, the following conclusions may be drawn, Nitrogen dioxide can be measured with maximum sensitivity in the 400 nanometer region of the spectrum. In this wavelength region, there is no overlap of the sulfur dioxide or nitric oxide absorbance, and thus the measurement does not require any correction for the presence of these species. The concentration of nitrogen dioxide present in the sample gas may be determined directly from equation 3, using an absorbance measurement in this wavelength region and a known molar absorptivity. Sulfur dioxide can be measured with excellent sensitivity in the 280 to 300 nanometer region of the spectrum. Nitrogen dioxide has a non-zero molar absorptivity in this wavelength region. Thus, accurate SO; analysis will require that the absorbance measured in this region be corrected for the presence of NO;.3. Nitric oxide can in principle be measured using any one of the three bands centred at 204.7 nm, 215 nm, or 226.5 nm. Since both SO, and NO, absorb in this wavelength region, determination of the NO concentration will require that the analyzer be capable of measuring SO, and NO,, and applying appropriate corrections to the absorbance signal at the measuring wavelength chosen. The maximum sensitivity can be achieved by focusing on the band at 214.8 nanometers. This absorption maximum conveniently lies in the region of an absorption minimum for sulfur dioxide. ‘The analysis of these three species by ultraviolet absorption spectroscopy requires, at a minimum, that three analytical wavelengths be chosen to measure the nitric oxide, sulfur dioxide and nitrogen dioxide concentrations. In practice, a fourth analytical wavelength is typically included to provide improved baseline stability. Once accurate determination of the nitric oxide and nitrogen dioxide concentration has. been achieved, the NO, concentration is calculated as the sum of these species. FIGURE 4. THE NEAR UV SPECTRUM OF NO, (200 TO 280 nm) i FIGURES. ‘THE UV-VISIBLE SPECTRUM OF NO, (275 TO 550 nm)FIELD DATA ‘The UV photometric technique was tested via a direct head to head comparison with the results obtained from a chemiluminescent analyzer at a coal fire generating station. The data acquisition began with the initial direct calibration of the instruments, and an initial system bias check. The system bias check is useful in determining the effect of the sample line and sample handling system on the species concentration, Subsequently, a series of nine stack gas runs were performed, each of 30 minutes duration, interspersed with system calibration checks. The process was completed by performing a final direct calibration. Both the UV and chemiluminescent analyzers were shown to be accurate to within 1 ppm during the initial direct calibration. The data acquired during the system calibration(Table I) for the UV analyzer demonstrates that the analyzer is extremely stable in both it's response to SO, and NO, During the SO, system calibrations with a 449 ppm cylinder, the instrument response was 447 + 0.7 ppm for the first UV analyzer and 446 + 1.1 ppm for the second Similarly, the response for these instruments to a 449 ppm cylinder of NO, was 449.3 0.3 ppm and 447.8 + 1.2 for units #1 and #2 respectively. The chemiluminescent analyzers showed comparably stable results in their response to NO,, although the measured concentrations were somewhat less than the tag value. The response of the first chemi to 449 ppm of NO, was 441.7 + 1.6 ppm, while that of the second was 438.1 + 0.8 ppm. The consistently low response of the chemiluminescent analyzers to the calibration gas during system calibrations may be caused by the presence of a species in the sample lines which interferes with the performance of the converter in the chemiluminescent analyzer. Note also that zero drift was negligible for both the ultraviolet and chemiluminescent analyzers. During on-line measurements of the flue gas, both analyzers performed reliably, and gave stable ‘measurements of the flue gas concentration. As mentioned previously, only the UV analyzers was capable of measuring both SO; and NO, simultaneously. During the on-line measurements, it was again noted that the chemiluminescent analyzer reported consistently lower concentrations of NO, than the UV analyzers. Asa final check, a direct calibration was again performed after completing all of the system calibrations and on-line measurements. The results for the 248 ppm tag value NO, cylinder are virtually identical to those obtained in the initial direct calibration, for both analyzers The final direct calibration for the higher concentration of NO, was performed using a different cylinder than that used for the initial direct calibration, and thus the concentrations reported can not be compared directly. The two analytical methods agreed to within one percent during both the initial and final direct calibration. It is obvious that both analyzers remained accurate during the entire duration of the runs, and that there was no significant change in the calibration factors as determined by direct calibration. The data upon which these results are based is summarized in the attached tables. CONCLUSIONS The direct and simultaneous measurement of NO, and SO, can be achieved using ultraviolet absorptionspectrophotometry. In order to do so, the spectrometer must possess high resolution, in order to provide a linear response to the nitric oxide and sulfur dioxide in the short wavelength regions of the spectrum. ‘The analytical technique has been well established, with over 50 installations, and has proven to be reliable and robust. For coal fired or oil fired burners, the technique provides a reliable and cost effective method to satisfy emission monitoring needs for both species. The SO2 measurement provided is consistant with EPA Method 6C, and thus the analyzer can currently be used in source testing for SO;. The ultraviolet method has also been submitted to the Environmental Protection Agency for review as an alternative instrumental method for NO,. REFERENCES [1] Acklin, M. W., McCullough, M., Tolk, J. D. “Utility Engineerings experience in the design, equipment selection, installation and operation of CEMS for utilities", AWMA Acid Rain and Electric Utilities; Permits, Allowances, Monitoring and Meteorology Specialty Conference, Tempe, AZ, Jan 23-25, 1995. [2] Folsom, A. B., and Courtney, C. W., “Accuracy of Chemiluminescent Analyzers Measuring Nitric Oxide in Stack Gases”, Journal of the Air Pollution Association, Vol. 29, No. 11, 1979. [3] Imbrogno, F., and Harris, P. “Ultraviolet Absorption Analyzers for CEM Applications”, Continuous Emissions Monitoring for Process Control and Regulatory Compliance, Edmonton, AB June 1996 [4] Harris P. and Adam H. “The design and practical application of UV process photometers Part II. Geometry and Component Selection, ISA 96, Chicago, IL October 1996. [5] Henderson, Brian. Optical Spectrometer, Handbook of Optics II. M. Bass et el Ed. McGraw Hill, NY, 1995. [6] Adam H. And Harris P. “The design and practical application of UV process photometers Part I. Resolution Effects” L’Analyse Industriel, Paris, France, 1996. [7] Saltzman, R. S., Ultraviolet/Visible (UV/Vis) Absorption Analyzers, in Analytical Instrumentation R. E Sherman (Ed) 1996.TABLE. _ FIELD CALIBRATION FOR UV AND CHEMILUMINESCENT ANALYZERS Tne [orenaTion [caLcas wi [we [een [oma 50,40 ppm wos | ane | ae ue Manors pm | 209 | an3 | 20a | ams Higind.e7m | toa | aoa | ana | aaa $0,409 ppm was | an va va Manor ppm | 208 | az | aioe | ann oe ce |S. mem wasn | as ve ve wee | _syem | inom | aee | fant | ate | aise Hynétm | a1 | ata | aos | 82 coi Syem | 50,469 pm wea | aes o* us ors | callin | ignore ppm | a2 | ae2 | aos | ars vas. | _syxem | 0, 449ppm wea | auto va va ioe | caternon [Hiinoceorem | ase | aero | air | aes 136. | Sam | $0,400 ppm usa | ais ™ os ist | cattrton [ignore ppm | wos | ama | ano | are waz | sytem | $0,449 9m usa | asa Py ma 23s | calor [Hipnoaopem | 2 | ase | atte | oz cra | sym | 50,409 ppm wa | ar om ve 338 | caloatin [Higanoreo pom | aoa | 7a | aor | ane oes | Suen | $0,400 ppm wie | eo os va 501 | calbaton |Hignnirei ppm | 19 | ans | aos | ane sae. | samem | 50,449 ppm waa | ano ua oa 60: | cation [rignoreorem | 9s | aes | ates | ait ocor- | die | miano.aieppm | 24a | 21 | 254 | 206 vers | calvin | Hid.esepmme | asa | asa | ous | Asn Notes: 1. Zero Drift was negligible during all calibrations and online runs 2. —‘Thechemi analyzers not capable of ealyzing fr sulfur dioxide 3. Thechemianayers were obeved to repent give low readings ding sytem caltratonpefomed though he ample aster system ‘The final direct calibration was performed using a different cylinder than that used for the intial direct calibration. The tag concentration for the new cylinder was 454 ppm NO,. Given that both the UV and the ‘Chemi analyzers reported this tank at approximately 460 ppm, itis likely that there was @ one percent error in the tag value.TABLE I PROCESS DATA FOR UV AND CHEMILUMINESCENT ANALYZERS TIME srecies— [uvi v2 cHemi1 | CHEMI2 80, 4864 484.1 wa wa NO 398.5 396.7 ma wa 22:37 - 23:06 No, 136 151 a wa No, 409.1 4106 396.1 388.4 $0, 489.2 4872, wa wa NO 3998 4013 wa wa 25:31 24:00 NO, 147 159 wa a No, 44s 4159 4019 393.0 80, 4862 486.6 wa wa NO 3984 400.3 wa wa 0019 = 00:88 NO, 149 156 wa va NO. 433 4145 400.7 391s 80, 480.2 4790 wa wa NO 396.4 3982 wa va 01:07 01:36 No, 147 157 wa va NO, aud 4126 398.1 388.6 $0, 4795 4786 wa wa NO 3962 398.1 wa a O15 02:24 NO, 151 156 wa a NO, aia 4124 399.8 3890 $0, 4824 482.0 we wa NO 396.6 3984 ve wa 0n63-05:12 NO, 152 158 fa va NO, airs 4128 3998 389.1 50, 4770 4767) wa wa NO 3995 4014 ve a 03:31 - 04100 NO, 156 165 we wa No, 415.1 4165 402.1 390.7 $0, 4814 4812 wa wa NO 4022 4036 we a 0419-0848 No, 157 170 a va No, 4179 4191 4043 3920 $0, 4766 4759 wa wa NO 4018. 403.5 ne wa 05:07 05:36 NO, 161 175 aa va NO. 4180 4196 4022 389.2