Anal. Chem. l S S 5 , 57,1586-1590
(4) Kokhoff, I. M.;Stenger V. A. "Volumetric Analysis"; Wlley: New York, 1957; Vol. 3, p 384. (5) Fritz, J. S.; DuVal, D. L.; Barron, R. E. Anal. Cbem. 1984, 56, 1177.
RECEIVED for review January 10,1985. Accepted March 22,
1985. Ames Laboratory is operated for the U.S. Department Energy under Contract No. W-7405-ENG-84. This work was supported by the Director of Energy Research, Office of Basic Energy Sciences. Of
Determination of Sulfur Dioxide, Nitrogen Oxides, and Carbon Dioxide in Emissions from Electric Utility Plants by Alkaline Permanganate Sampling and Ion Chromatography John H. Margeson,* Joseph E.Knoll, a n d M. Rodney Midgett Environmental Monitoring Systems Laboratory, US.Environmental Protection Agency, Research Triangle Park, North Carolina 27711 Guy B. Oldaker, 111, and Wayne E. Reynolds Entropy Environmentalists, Inc., Post Office Box 12291, Research Triangle Park, North Carolina 27709
A manual 24-h integrated method for determlnlng SO,, NO,, and CO, In emlsslons from electric utlllty plants was developed and fleld tested downstream from an SO, control system. Samples were collected In alkallne potasslum permanganate solution Contained In restrlcted-orlflce Implngers. SO, Is oxldlzed to SO-: and S202-, and NO, (NO NO2) is oxldlzed to NO,-; CO, Is converted to C03,-. Samples were analyzed by Ion chromatography. The method showed 100 % collectlon efficiency for all three pollutants at a sample flow rate of 35 cma/mln and was found to be unbiased relative to Independent monltorlng systems. At a flow rate of 75 cm3/ mln, the collectlon efficiency was still 100% for SO, and NO,, but dropped to 80 % for COP The relative standard devlatlon for the three determlnatlons was as follows: SO,, 0.2-8%; NO,, 3 4 % ; and CO,, 2-8%. Further work Is needed to determine the error In the SO, determlnatlon caused by H,S04 absorptlon.
Environmental Protection Agency (EPA) regulations (1) require that electric utility plants monitor SOz and NO, emissions on a continuous (24 h) basis. COz continuous emission data are also needed so that emissions can be reported in the units of the EPA standard (pounds of pollutant/million Btu of heat input). These regulations also require that continuous monitoring be utilized to determine the efficiency of flue gas desulfurization (FGD) systems that are used to reduce SO2 emissions. This, of course, means monitoring upstream and downstream from the control system. We report here on a manual 24-h integrated sampling method in which all three pollutants are collected in one sample, by conversion to their ions, and then analyzed by ion chromatography. Thus, the method is cost effective in that all three pollutants are quantified with one monitoring system. In addition to being used as a separate routine monitoring method, the method could be used in place of SO2 and NO, instrumental methods, when they are inoperative. Work on development of this method for use upstream and downstream from FGD systems, with field testing downstream from a FGD system, is the subject of this paper. EXPERIMENTAL SECTION Laboratory Samplings. Simulated emission atmospheres were generated by flow dilution of SO2,NO, and COPatmospheres contained in cylinders the contents of which had been analyzed;
water vapor was added by bubbling the dilution g:.s (02in N,) through water. Samples were collected in 0.25 M KMn04-1.25 M NaOH solution contained in three restricted-orifice impingers using the sampling system shown in Figure 1. Because of the high concentration of COz in source emissions, it was desirable to maximize the OH- concentration in the absorbing solution. Therefore, NaOH, KOH, and LiOH were tested for solubility in 0.25 M KMn04 solution at 10 "C. NaOH had the highest solubility, but increasing the concentration from 1.25 to 1.50 M precipitated KMnO,. Therefore, 1.25 M NaOH was used in all samplings. Field Testing. Field testing was conducted at a coal-fired electric utility plant with a generating capacity of 917 MW; the coal being burned contained 4.2% (w/w) S. Sampling was conducted downstream from particulate and SO2 emission control systems. SOz emissions were controlled by a wet injection process based on MgO and CaO. The stack gas temperature was 126 OF; the gas was saturated with water and contained water droplets. Reference Measurements. Data obtained from continuous emission monitors (CEMS) were intended for use as the standard for determining the SOz,NO,., and COz collection efficiency (CE) of the alkaline permanganate (A-P) sampling systems. However, the SOz CEM exhibited a negative bias presumably due to the high moisture content of the emissions. This bias was observed in spite of the sample conditioning system (see below). Therefore, EPA Method 6 (ref 1,appendix A, pp 455-460) was used as the reference measurement for the SOz determinations. During each 24-h run, 20-min samples were taken at discrete hourly intervals. The SO2concentration for each 24-h run was the arithmetic mean of the 24 EPA Method 6 results. NO,, COz, and Oz concentrations were determined with a transportable continuous emission monitoring system which was designed, operated, and calibrated according to a published protocol (2). Samples were extracted from the stack using a 180 cm long, 10 mm i.d. heated borosilicate-glass probe. A polyethylene bottle was fitted at the probe inlet in order to remove water droplets and thereby minimize demands on the sample conditioning system. (The same system was used with the probe for taking Method 6 samples.) A plug of cleaned (2-propanol/ water) borosilicate-glass wool was placed within the probe and near its inlet to remove particulate matter. The temperature of the gas stream at the plug was monitored with a thermocouple and maintained at approximately 302 OF. The probe outlet was connected to a conditioningsystem which consisted of a condenser, designed to lower the water dew point of the sample stream to 0 "C, and a 120 cm long Perma Pure dryer which operated to further lower the dew point. The outlet of the conditioningsystem was connected to a manifold which served three analyzers. NO,, COz, and Oz concentrations were determined with a Beckman Model 951 (chemiluminescence),a Beckman Model 864 (NDIR),
0003-2700/85/0357-1586$01.50/00 1985 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 57, NO. 8, JULY 1985 *
w CRITICAL ORIFICE, VACUUMGAUGE
nr PROBE. P O R T ALKALINE. PERMANGANATE SAMPLING
IMPINGER TIP: 1.5 m m 1.0.
PUMP, NEEDLE VALVE, SURGE TANK, ROTAMETER, SILICA GEL, DRY GASMETER
Flgure 1. Alkaline permanganate sampling system.
and a Teledyne Model 320P4 analyzer, respectively. The analyzers were calibrated with NO or Ozin Nz and COzin air atmospheres contained in cylinders; concentrations were traceable to NBS Standard Reference Materials. System calibration checks entailed introduction of gases via a tee located at the probe outlet. These checks were performed before and after each 24-h run, and 3 to 4 times during the run, for each analyzer. Stack NO,, COz, and Ozconcentrations were recorded on strip charts. Concentration data were reduced to 24-h averages by taking the arithmetic mean of concentrations recorded at 15-min intervals (-90 data points). The exception to this procedure was the computation of the COz concentration for run 1. During the final 3 h of the run, the COz analyzer malfunctioned; EPA Method 3 (ref 1, appendix A, pp 422-427) was used to provide 20-min integrated samples. Alkaline Permanganate (A-P) Sampling Trains. This sampling system was designed to collect H#04 in addition to SOz, NO,, and COz. The heated probe was of the same design as that used for the reference measurements. The probe outlet projected into an electrically heated box and therein was connected to a borosilicate-glass manifold having four outlets. Each outlet was connected to separate HzS04adsorption tubes. These tubes were designed according to the plug procedure described by Cheney et al. (3). Two plugs of cleaned borosilicate-glasswool, approximately 1g each and separated by a 25-mm air space, were contained in a 140 X 10 mm glass tube. The air temperature within the box was monitored with a thermocouple and maintained at 263 and 272 O F for runs 1and 2, respectively. These temperatures were above the water dew point but below the estimated HzSO4 dew point of 280 OF. (No condensation was observed within the adsorption tubes during either run). This system was checked in the laboratory for SO2 loss on the glass wool. No loss was detected. Each of the four outlets from the HzS04adsorption tubes was connected to an A-P sampling train as described in Figure 1. Impinger design is discussed in the text. Each impinger contained 200 mL of absorbing solution (250 mL was inadvertently used in a few impingers). Sample flow rates were controlled with critical orifices. Sample volumes were measured with Singer dry test meters having a capacity of 1L/revolution. The positioning in the stack of the probes for the different monitoring systems is illustrated in Figure 2. A-P Sample Preparation and Analysis. Individual impinger contents were diluted to 500 mL, a 50-mL aliquot was taken, and the Mn0,- was reduced with 1.5 M HzOzby the reaction HzOz + 2Mn04- = 2MnO2 + 2O2 + 20HCompletion of the reaction is indicated by the presence of a clear supernatant liquid on settling of the MnO,. Excess HzOz is decomposed catalytically by Mn02 Samples were then filtered through GF/C filter paper (Whatman) and diluted to volume.
Flgure 2. Position of monitoring systems in the stack.
Samples were analyzed for NO< and SO>- by ion chromatography (IC) using a Dionex Model 2110i ion chromatograph, equipped with an ion exchange analytical column (Dionex 35311) and a fiber suppressor. The eluent was 1.5 mM Na2C03-1.9 mM NaHC03 at a flow rate of 2.0 mL/min. The sample loop volume was 12 pL. Samples and standards (prepared from reagent grade KNOB or Na2SO4 in deionized water) were analyzed twice. Calibration curves (least-sqwes fit) were generated by using peak areas, which were determined with a Hewlett-Packard Model 3390A integrator. CO2- was determined by using the same instrument and procedure as above, except that an ion exclusion analytical column (Dionex 30890) was used without suppression and HzO was used as the eluent at a flow rate of 1.7 mL/min. Calibration standards were prepared from reagent grade NaZCO3. The calibration curves for NO, and SO>- determinations were linear. With COBz,the curve was parabolic. In this analysis, Cotis deposited onto the head of a cation exchange column in the H+ form, Thus, H2C03is formed, eluted, and sent to the detector cell. Since HzCO3 is a weak acid, the result of its ionization is that the relationship between HzC03 concentration and conductivity is parabolic. A-P field samples were spiked with NO,, SO2-, and CO2-; the recoveries were 97.2 0.8%, 92.8 2.2%, and 106.4 A 4.4%, respectively. EPA NO, and SO2 quality assurance samples, and 1 N NazC03 which had been analyzed titrimetrically (Fisher Scientific),were also analyzed to check the accuracy of the analysis procedure. The recoveries were 99.5 f 2.2%, 101.3 f 2.2%, and 102.3 f 3.4% for NO3-, SOq2-, and CO2-, respectively. SzOs2-analysis was carried out with a Dionex Model 10 ion chromatograph equipped with a 50-mm analytical column (Dionex 35310), a packed bed suppressor, and a strip chart recorder. The eluent was 0.07 M NazC03at a flow rate of 2.0 mL/min. The sample loop volume was 0.5 mL. Because of the high HzC03 concentration generated during suppression, the initial noise level was excessive. It was reduced to an acceptable (normal) level by inserting two guard columns (in series) after the detector. The resultant increase in pressure shifted the H2C03/C02equilibrium in the detector cell so that more COz was in solution. It was established by external calibration with Sz02-standards (prepared from NazSz06)that the S2OsZ-concentration-peak height relationship was linear (r = 0.999). The time available for analysis, before suppressor regeneration was necessary, was only 95 min. Therefore, one standard was analyzed which had a response close to that of the samples and the SZOs2-concentration determined by proportion. This procedure allowed the analysis of three samples, a standard, and a blank before regeneration. The relative standard deviation for this analysis was approximately 6%. HzS04 Analysis. Has04 was determined by extracting the glass wool with 100% isopropyl alcohol (>91% recovery) and analyzing the extract for SO:- by IC. Isopropyl alcohol rather than water was used to extract the HzS04so that any particulate sulfate that may have penetrated the filter (in the probe) would not be extracted (4). The accuracy and precision of this analysis were approximately *12 and *IO%, respectively.
RESULTS AND DISCUSSION Alkaline permanganate solution has been used to oxidize NO, (NO + NO2) emissions from electric utility plants to
ANALYTICAL CHEMISTRY, VOL. 57, NO. 8, JULY 1955
NO3-, and it is the basis for a 1-h sampling procedure with IC analysis (ref 1,September 27,1984, pp 38232-38238). Since SO2 is also oxidized in this system, it was decided to investigate the possibility of determining both NO, and SO2 simultaneously on a 24-h basis. The reactions for the oxidation of NO, which accounts for 95% of the NO, emissions, have been published (5); the rate-controlling step is the solubility of NO. SO2 oxidation apparently involves the reactions 2Mn04- + 3So3'-
+ H20 = 3s04'- + 2Mn02 + 20H(1)
+ 6SOS2-+ 4H20 = 3s202- + 2Mn02 + 80H(2)
The rate-controlling step for reaction 1is the solubility of SOz. The rate of dithionate (S20& formation in reaction 2 becomes significant at high SO2 concentrations. This rate may be increased by the formation of Mn02 (6). C02 is collected by the reaction
COz + 20H- = C032-
The rate-controlling step here is the well-known slow reaction rate of C02 with OH- (7). The SO2 concentration before a FGD control system typically ranges from 500 to 3000 ppm, and it is reduced to about 200 to 50 ppm depending on the efficiency of the control system. The NO, concentration is usually in the 200-500 ppm range and passes through the SO2 control system essentially unaffected. The C 0 2 concentration is typically 12%. A method for SO2, NO,, and C02must be able to sample these widely different concentrations at close to 100% efficiency. The method must also sample at a constant rate for 24 h to avoid biasing the data. Laboratory Results. The first laboratory experiments involved using three-restricted-orifice impingers in series, to sample an atmosphere containing 1300 ppm SO2and 160 ppm NO. The orifices had an inside diameter of 1.5 mm. This orifice size was chosen to produce a high surface to volume ratio so as to maximize NO absorption. After sampling for 10 h at a low flow rate, flow through the impingers ceased, because the orifice in the first impinger was plugged with MnOz. Ultimately, it was necessary to use an open tube, 11 mm i.d., in place of the restricted orifice in the first impinger to avoid plugging and to obtain a constant flow rate for 24 h. This system was used to sample an atmosphere containing 2700 ppm SOz and 300 ppm NO. Constant flow rate was indeed maintained for 25 h. However, it was necessary for the inlet to the open tube to be elevated (2.5 cm) from the bottom of the impinger so as to prevent the inlet from being submerged in Mn02, which collected at the bottom of the impinger. Once the problem of obtaining a constant flow rate for 24 h had been solved, attention was directed toward sampling simulated atmospheres to determine S02, NO,, and COZ CE and to determine the contribution of S20S2formation to SO2 CE. Low sampling flow rates (25-80 cm3/min) were used to maximize absorption of the pollutants. The results showed that: 1. The SO2 CE ranged from 87 to 96% a t flow rates of 40-80 cm3/min. All of the SO2 (as sod2-and S20SE)was found in the first impinger. At 200 and 2700 ppm SO2 sampled, S20S2-formation contributed