EPA-R2-73-051a June 1973 Environmental Protection Technology Series !• iiiiiiiiiB ^lliiiiii I ------- EPA-R2-73-051a Development of Aqueous Processes for Removing NOx from Flue Gases - Addendum by Gilford A. Chappell Esso Research and Engineering Co. Government Research Laboratory Linden, New Jersey 07036 Contract No. 68-02-0220 Program Element No. 1A2014 EPA Project Officer: D.A.Kemnitz Control Systems Laboratory National Environmental Research Center Research Triangle Park, North Carolina 27711 Prepared for OFFICE OF RESEARCH AND MONITORING U.S. ENVIRONMENTAL PROTECTION AGENCY WASHINGTON, D.C. 20460 June 1973 ------- This report has been reviewed by the Environmental Protection Agency and approved for publication. Approval does not signify that the contents necessarily reflect the views and policies of the Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use. 11 ------- ABSTRACT This report summarizes the findings of the laboratory program for "Development of Aqueous Processes for Removing NOX and SC>2 from Combustion Flue Gases." This project is the second phase of the flue gas scrubbing work sponsored by EPA under Contract No. 68-02-0220. The results of the Phase I program are contained in report EPA-R2-72-051, entitled, "Development of the Aqueous Processes for Removing NOX from Flue Gases." The present report contains discussions of analytical techniques and scrubber design in addition to experimental results obtained with a vertical spray tower scrubber. The blended flue gases passed up the unpacked glass column countercurrent to the absorbing solution which was sprayed down from the top. The scrubbing experiments showed: • N02 and S02 are effectively absorbed by 1.0 molar Na2SO~ solutions. 9 NC>2 absorption by 1.0 molar NaOH solution is enhanced by the presence of S02 in the flue gas. « Neither NO nor NOo is effectively absorbed by 1.0 molar NaOH solution in the absence of S02> and NO absorption is not improved by the presence of SO-. » Increasing the L/G ratio improves N02 and S02 absorption by 1.0 molar Na^SO.,. © Under similar scrubbing conditions Mg(OH)2 slurry is not as effective as Na?SO solution for NO absorption. The data show that sulfite solutions would effectively absorb NOX and S02 from flue gases provided the NO (mostly NO) has been oxidized to NO,, upstream from the scrubber. ACKNOWLEDGEMENTS This work was conducted by the Government Research Laboratory of the Esso Research and Engineering Company for the Environmental Protection Agency under EPA Contract 68-02-0220. Dr. Gilford A. Chappell, the Principal Investigator for the work reported herein, was a member of the Air Conservation Section managed by Mr. Alvin Skopp. The invaluable assistance of Mr. William Moss in the laboratory during the entire project is sincerely appreciated. Mr. Douglas Kemnitz, was the EPA Technical Project Officer during the program. - iii - ------- - IV - ------- TABLE OF CONTENTS ABSTRACT iii 1. INTRODUCTION 1 2 . LABORATORY STUDIES 4 2 .1 Ana lyt ica 1 Techniques 4 2.1.1 Analysis of Solutions for Nitrite and Nitrate Levels 4 2.1.2 Measurement of NO Levels in Flue Gases 10 2 .2 Scrubbing System • 13 2.2.1 Flue Gas Blending 13 2.2.2 Flue Gas Scrubber 14 2.3 Results of Flue Gas Scrubbing Experiments 16 3 . CONCLUSIONS AND RECOMMENDATIONS 20 3 .1 Cone lus ions 20 3.2 Recommendations for Future Work 21 APPENDIX - Measurement of Molar Absorptivities for Nitrite and Nitrate 22 - v - ------- 1. INTRODUCTION The oxides of nitrogen (NO, N02) are essential components in the formation of photochemical smog in addition to being pollutants in their own right. Sulfur dioxide (S02) is a major air pollutant. The major sources of these oxides is the large fossil fuel fired boilers such as those found in electric power generating plants. Nitric oxide (NO) is formed in the high temperature zone of the furnace by the reaction between atmospheric nitrogen and oxygen; as the combustion gases cool, a small percentage (10%) of the NO is oxidized to N02« Collectively, these two oxides are referred to as 'NOx'- If the fuels contain organically bound nitrogen, as do coal and oil, part of this nitrogen is converted to NOX during combustion. Similarly, sulfur containing species present in the fuel provide a source for S02- The composition of different flue gases is shown in Table 1. TABLE 1 TYPICAL COMPOSITIONS OF'FLUE GASES Volume % Combustion Of Component N2 co2 H20 °2 so2 .NO x Participates grams/ft3 Coal(a) ( 76.2 14.2 6.0 3.3 0.2 0.5(e) Dil(b) 77.0 12.0 8.0 3.0 0.15 0.07(d' 0.01 Gas(c) 72.3 9.1 16.8 1.8 — ) __ (a) Calculated for burning with 20% excess air a typical high volatile bituminous coal of the following com- position = carbon -70.1%, oxygen -6.6%, hydrogen -4.9%, nitrogen -1.4%, sulfur -3.0%, ash -12.7%, and H20 -1.3%. (b) Calculated a typical residual fuel oil of the following composition = 86.5% carbon, 10.3% hydrogen, 2.5% sulfur, 0.7% nitrogen with 20% excess air. (c) Calculated for burning natural gas with 10% excess air. (d) This is an average value. Actual values range from 0.01% to 0.15%. (e) Assumes 90% particulates removal. ------- - 2 - In order to remove NOX from flue gases, Esso Research and Engineering Company carried out a flue gas scrubbing program which was sponsored by EPA. The program consisted of screening various aqueous absorbents for NOX absorption potential. The results of the screening program are contained in the report EPA-R2-72-051 . ' The main conclusions of the batch screening studies were: • The addition of N(>2 to flue gas to improve NOX (mostly NO) absorption does not appear promising. While the presence of N0£ does improve the absorption of NO, the magnitude of the increase is insufficient to support a viable process. • Sulfite solutions and slurries are efficient N02-S02 absorbents. Soluble sulfites (Na2S03) are better N02 absorbers than insoluble slurries (CaSOn) because of the higher level of sulfite ion in solution. • Calcium, magnesium, and zinc hydroxide slurries are effective N02~S02 absorbers . The sulfite formed when S02 is absorbed is necessary for efficient N02 scrubbing. • Limestone (CaC03) is also a good N02-S02 absorbent for the same rea- sons as for Ca(OH)2. • N02 scrubbing is enhanced by removing oxygen from the flue gas or by adding an anti-oxidant such as hydroquinone to the scrubbing solution. « Sulf ide solutions are excellent N02 and S02 absorbers but do generate a small amount of NO. • Part of the absorbed S02 is oxidized to sulfate. These results led to the second phase of the program which was to design, construct, and test a continuous, gas scrubbing system for simultaneous removal of N02 and S02 from blended flue gases. The underlying assumption is that flue gas NOX (mostly NO) from a real plant could be oxidized to N02 upstream from the gas scrubber. This may be technically feasible using ozone or a catalyst. The basic approach is to use the sulfite formed during S02 absorption to remove the N02 from the same gas stream. S02 + 20H~ - > S03~ + 20H~ + S03~ .+ 2N02 - > S04 + 2N02" + HO The resulting nitrite ion (N02~) may subsequently be oxidized to nitrate ion (N03~) by the molecular oxygen (3%) present in the flue gas. In addition to testing sulfite ion scrubbing, certain analytical techniques needed improvement. These included the spectrophotometric procedure for analyzing solutions for nitrite and nitrate levels, and the procedure for measuring nitric oxide (NO) levels in damp flue gas. ------- - 3 - The objectives of the program were to: (1) Design and construct a continuous flue gas scrubber for SC>2 absorption. (2) Obtain scrubbing data using several absorbents to verify the results of the screening study. (3) Develop and improve the analytical procedures so that accurate mater ia 1 ba lances may be obtained. These will be discussed in detail in the following section. ------- - 4 - 2. LABORATORY STUDIES 2.1 Analytical Techniques This section discusses the procedures for solution analysis and for gas analysis. ,..•.. 2.1.1 Analysis of Solutions for Nitrite and Nitrate Levels When absorbed by aqueous solutions NOX is converted primarily to nitrite and nitrate ions whose levels must be accurately determined in order to insure a satisfactory NOx material balance. J.'H. Wetters and K. L. Uglum [Analytical Chemistry, 42, 335 (1970] have described a spectrophotometric technique capable of the direct, simultaneous deter- mination of nitrate and nitrite levels in aqueous solutions. The pro- cedure involves taking two ultraviolet absorbance readings (302 nm and 355 nm) on the solution followed by calculation of the levels using molar absorptivities determined with standard solutions of nitrite and nitrate ions. The authors claim a lower detection limit, using 1.0 cm cells, of 0.02 mg/ml for nitrite and 0.09 mg/ml for nitrate. Three molar absorptivities are required because nitrite ion absorbs at 302 nm and 355 nm whereas nitrate ion absorbs only at 302 nm. Thus, the nitrite level in an unknown solution may be determined by a single absorbance reading at 355 nm whereas the nitrate level must be calculated from- readings taken at both wavelengths. For example, the nitrite level in a solution containing both nitrite and nitrate may be calculated from the absorbance reading at 355 nm using Beer's Law. A = ebC A = measured absorbance e = molar absorptivity b = cell path length (centimeters) ! C = solute concentration (molarity) The nitrite concentration is calculated from the following quantities. .355 355 ,_ The nitrate determination is slightly more involved because the measured absorption at 302 nm contains contributions from both species 302 302 L ^ 302 ^0 A - £- bC- + £- bC- ------- - 5 - However, the concentration of nitrite is already known from the reading taken at 355 nm. = 355 Substituting into the expression for the total absorption at 302 nm gives, after cancelling the equal path lengths, . .302 / N02-\ .355 . .302 A =1 I A :NO, .355 302 N03 Therefore, the nitrate level may be determined from three molar absorptivities, two absorbance measurements, and the cell path length. In order to measure the molar absorptivities, stock solutions of NaN02 and KN03 were prepared from reagent grade chemicals and dis.tilled water.. Samples of solution were placed in 4.0 cm qua.rtz cells which were inserted into an Optica. Spectrophotometer for absorbance measure-' ments. The results, tabulated in the Appendix, gave the following molar absorptivities and average deviations: 302 = 9.4 + 0.2 355 = 24.2 + 0.7 ,302 "NO,/ 7.5 '+ 0.2 These values were used to test the procedure by analyzing prepared solutions containing both NaN02 and KN03. These 'results are shown in Table 2.. ------- - 6 - Table 2 Spectrophotometric Analysis of Nitrite-Nitrate Mixtures Actual Solution Molarity Measured Solution Molarity Exp # I 2 3 4 5 6 6a* N00~ ,__ £ . 0.00488 0.00488 0.00975 0.00520 0.00520 0.0104 0.0104 N00" - J — 0.00908 0.00908 0.00908 0.00908 0.00908 0.00908 0.00908 NO," £. 0.00500 . 0.00485 0.00986 .0.00525 0.00526 0.0102 0.0104 NO,,' j 0.00961 0.00908 0.00958 0.00930 0.00922 0.00958 0.00951 NO ' 2.5 0.6 1.1 1.0 1.2 1.5 0 N0_ 5.8 0 5.5 2.4 1.5 5.5 •4.7 Used a 1.0 cm cell instead of the 4.0 cm cell used in all the other experiments . . . '' Although the nitrate measurements tended to be high, the overall results were satisfactory. Since the bulk of the flue gas scrubbing experiments were to be carried out in the presence of sulfite ion, it was important to ascertain the effect of sulfite ion on the analysis for nitrate and nitrite levels. The Spectrophotometric absorbance of sodium sulfite (Na2SQ3) solutions exhibited a strong pH dependence as shown in Table 3. Table 3 Effect of pH on the Molar Absorptivity of Sulfite Ion Molar Absorptivity Measured at EiL 9.7 7.7 302 nm 0.0038 0.131 355 nm 0.0018 0.0035 ------- - 7 - The natural pH of 1.0 molar Na2SC>3 is 9.7. Addition of hydrochloric acid, which does not absorb high at either wavelength, was necessary to lower the pH to 7.7. The dramatic increase in molar absorptivity at 302 nm was likely due to the formation of bisulfite ion. S°3~ + H3° - * HS°3~ + H2° This would not seriously affect the determination of nitrite levels at the lower pH but would completely negate any attempt to measure nitrate levels. Thus, to minimize sulfite interference requires that the solution be made sufficiently alkaline (pH>-9) prior to analysis. In order to verify this observation, solutions containing nitrite and sulfite were analyzed at both wavelengths. In the absence of nitrate each wavelength gives an independent measurement of the nitrite con- centration and any interference should show up in the reading taken at 302 nm. Duplicate experiments on a solution containing 0.00975 molar nitrite and 0.90 molar Na2S03 (pH = 9.8) produced consistent results; at 302 nm the measured nitrite level was 0.00994 molar (27, error) and at 355 nm the measured level was 0.0104 molar (57» error). Another sample of the same solution was treated with a small quantity of concentrated hydrochloric acid to lower the pH to 7.8. Spectro- photometric measurements then showed the nitrite values to be 0.0131 molar at 302 nm and 0.00894 molar at 355 nm. These values differ by +35% and -87,, respectively, from the original nitrite level. The +357, error is consistent with sulfite interference whereas the -87, error is less obvious. The dropwise addition, with stirring, of concentrated HCl produced locally high acidities in the region of drop impingement. This very low pH condition lasted momentarily until the mixing effect of stirring caused the pll to shift toward more alkaline values. However, part of the nitrite may have been lost via breakdown of the nitrous acid formed at low p-H . ' 2HN02 - : - > N2 N20 (g) — - — > N0(g) + NO (g) (escape from solution) The problem was eliminated by improved stirring and by introducing the HCl below the surface of the solution. Alkaline scrubbing solutions will absorb C02 to form carbonates which may or may not be soluble depending on the cations present. Because soluble carbonate interferes with spectrophotometric determination of nitrite and nitrate, the species must be removed prior to analysis. This is easily accomplished by acidifying the solution which expels the car- bonate as C02 gas. Subsequently, the pH must be raised to eliminate sulfite interference. In order to check the effect of'pH cycling, several experiments were made with no carbonate present. Table 4 shows the results from two runs using solutions containing nitrite and sulfite ions . ------- - 8 - Table 4 Effect of pH Cycling on Nitrite Readings; After pH Cycling* Original Solution [N02~] % Error Exp # IM>2lLi* [S03 ] pH 302 nm 355 nm _p_H 302 nm 355 nm 1C 0.00956 0.88 9.8 0.00924 0.00966 10.1 -3. 1. 2C 0.00953 0.88 9.8 0.00974 0.00957 11.9 2. •£ 1. * The pH was lowered to 7.7 with concentrated HC1. This was followed by addition of concentrated NaOH to raise the pH. The slight dilution effect was taken into account in the calculations. ** All concentrations. ar,e in units of molarity. Since the pH cycling showed no effect on nitrite measurements in the presence of sulflte, several more experiments were made with solutions containing nitrite, nitrate, and sulfites as shown in Table 5. Table 5 Effect of pH Cycling on Nitrite and Nitrate Readings Original Solution After pH Cycling* % Error Exp # IN0^* [N0"] pH IN021_ [NO] pH [NO^] [NO 3C -0.00520 0.00908 9.5 0.00557 0.00892 11.6 7.1 -1.8 4C 0.00520 0.00908 9.6 0.00524 0.00936 11.7 1. 3.1 5C 0.00520 0.00908 9.4 0.00540 0.00918 11.5 4.6 1.7 * Same as in Table 4. ** Same as in Table A. All solutions contained The effect of pH cycling appears to be more pronounced in Table 5 than in Table 4 although the only difference is the presence of nitrate ion. Because the goal of pH cycling was to eliminate carbonate and sulfite spectral interferences, it was decided to move on to work with solutions containing all four ions - N03", N02", S03= and C03~. Stock solutions containing the four ions gave totally unsatisfactory analyses after pH cycling as shown in Table 6. ------- - 9 - Table 6 Effect of pH Cycling on Solutions Containing Nitrite, Nitrate, Sulfite, and Carbonate Ions Original Solution After pH Cycling* 7. Error Exp # [NO^'l** fNO-~l pH 6C 0.00520 0.00908 11.10.00143 0.0103 .9.5 -73. 13, 7C 0.00520 0.00908 11.l" 0.00131 0.0109 9.8 -75. 20, 8C 0.00520 0.00908 11.10.00384 0.0100 9.8 -26. 10, * Same as in Table 4. ** Same as in Table 4. All solutions contained 0.85 M SO ~ and 0.85M CO . These results indicate that nitrite is disappearing during pH cycling and is not being totally converted to nitrate. It is likely that nitrite is being stripped.from the solution by the effervescence produced during HCl addition. As discussed previously the nitrite ion may decompose into the two gases, NO and N02, which may then diffuse into a nearby bubble of C02 gas instead of recombining to reform nitrite ion.' co2 •> escape from solution <• — - - ^ eoudiJc 1. 1. win oi_< icii_ A.^ii bubbles mixing eliminates, nitrite ion * -local acidity Of the two gases, NO and N02 , which may escape, N02 may react with water to form more nitrate. Nitric oxide is very unreactive and once entrapped in a gas bubble, will be stripped from the solution. The overall sequence shows the possibility of one nitrate being formed for every three nitrites which disappear. ^cid NO 2HN03. + NO 3HNO (nitrite) acld» HO + 2ND + HN03 (nitrate) This sequence is approximately consistent with the results of experiments 6C and 7C and accounts for the high levels of nitrate found after pH cycling. In experiment 8C, however, we can account for only 40% of the total excess nitrate error. Since the cycling in this experiment lowered the pH only to 8.3 as compared to 6.5 and 7.7 for 6C and 7C, ------- vi' - 10 - I y , more ca rbona Li- sliould remain in solul ion t.o raise the 302 nm reading (carbonate, has Less influence, at 355 nin) . The fact thai the pll was only lowered to 8.3 also explains the 1/3 l.e.ss error in nitrite level for experiment 8C . Since less acid was added, leas nitrite decomposed. If the solutions listed in Table 6 do not undergo pH cycling, but are analyzed directly (pH = 11.1) the resulting nitrite and nitrate values are in excess by approximately 1270. » All further attempts at removing soluble carbonate by acid addition failed to yield satisfactory results. At this juncture we turned to precipitation techniques with the hope of resolving the problem. A concentrated solution of CaClZ was added to the test solution containing all four of the important ions. Unfortunately, calcium carbonate and calcium hydroxide precipitates gelled the entire sample. After breaking the gel and centrifuging the mixture the clean solution was analyzed spectrophotometrically for nitrite and nitrate. Unfortunately, the results were highly erratic and totally unsat is factory . At this point we, decided not to pursue this question any further but concluded that a total nitrogen level would have to suffice for those few samples containing high levels of soluble carbonate. A complete NOX material balance would still be possible for these samples; only the ratio of nitrite to nitrate would be unavailable. A reasonable approximation to the ratio could be obtained by extrapolation from data taken from a similar absorbent in which carbonate is insoluble. The results of this effort can be briefly summarized: • Raising the pH to >9 , effectively minimizes sulfite interference in the spectrophotometric determination of nitrite and nitrate. « High levels of soluble carbonate interfere with the nitrite and nitrate measurements; no simple procedure was found which would eliminate this problem. Fortunately, a total NOX materia.l balance is still possible for those few samples exhibiting this problem. 2.1.2 Measurement of NO Levels in Flue Gases Flue gases contain three gaseous pollutants: 2. N02 3. NO Two Dupont 400 spectrometers were used for on-line analysis of S02 and N02 • These instruments were also employed in the screening study referred to previously. The hot (130°F), damp, flue gas flowed through the heated gas analyzer cells where the appropriate measurements were made. No pretreating of flue gas was necessary. The analyzer readings were recorded continuously on strip charts. ------- - 11 - The; detenu i ii;il i on of NO levels in the flue c,,is w.-is more comp I i edited. During I he earlier screening study a lUu-kman NDTR (non-l)i spers i ve Inlr.i-Ke.d) w;is used lor NO mr;i sure.iucnl: . Hn I or I un,-i I e I v , I he i ns t runienl was sensitive1 l"o water v;ipor which required thai the w.-il.e.r he removed prior to ana l.-ys Is . This w;is not a simple, task because ol the complex reaction chemistry associated with NO, N(>2, S0;> and liquid water. A cumbersome yet reliable technique evolved in which the. [hie gas passed through a sequence of calibrated traps designed to remove N02, S02 and water vapor. The complexity of the system created problems and introduced a delay time in the ,NDIR output. ; In order to improve the analytical procedure for NO, we . discussed chemiluminescence techniques with Thermoelectron Corporation which manufactures an instrument for NO analysis. Their conventional instrument was unable to handle damp gases directly but it seemed that specially heated inlet lines could resolve that problem. As long as the water vapor did not condense in the inlet system, no difficulties should arise. Consequently, Thermpelectron Corporation modified a standard instrument to our specifications and loaned it to us for test experiments. In the first series of tests we compared the response of the NDIR and the chemiluminescence unit (CML) to a gas blend consist- ing of nitrogen and varying amounts of NO. Table 7 contains the results . Table 7 Comparison of the CML with the NDIR NDIR Response Series # CML Response 287* 198 475 120 360 287* 202 485 115 365 Comments CML inlet system is cold; heater not on 2 395* 200 87 353 480 460* 110 175 440 277 265 250 302 395* 208 87 360 485 460* 106 185 450 290 280 265 318 CML inlet system heater is on 'Gas blend contains 12% C02 in Series # 3 runs except for ' last run steam (^57=) added steam (<57o)+370 00 No C0n * At the beginning of each series, both instruments were adjusted to give the same correct reading. ------- - 12 - The two instruments were connected in parallel to the main line carrying the blended gas.' The agreement is quite good between the NDIR and the CML which indicated satisfactory reliability for the CML. The CML has two readout modes; one for NO only and the other for total NOx- The second series of tests was designed to check the agree- ment between these two modes and to determine the influence of all the flue gas constituents on the NO reading under normal operating conditions. The NO level was fixed at 600 ppm initially and left ' unchanged during the entire test. The results are shown in Table 8. Table 8 Effect of Flue Gas Components on CML Reading ' • NO Reading NO* Reading Exp . # (ppm) (ppm) \ Comments Id 588 590 gas contains N2 and NO 2d 598 598 3% 02 added 3d 600 600 12% C02 added to the above 4d 600 600 107» steam added to the above 5d 600 600 one hour after steam added 6d 590 1625 £ 1000 ppm N02 added to above 7d . 580 1500 ^ 3000 ppm S02 added to above 8d 580 1500 . S02 source turned off 9d 580 1500 S02 source turned back on The data in Table 8 show the CML to be reliable, stable, and not significantly affected by the various constituents of flue gases. The modified instrument circumvented all the previous problems associated with the NDIR and, in addition, provided a check on our Dupont 400 N02 analyzer. The total NOX readout on the CML gave the sum of NO plus N02 from which the N02 level could be calculated by subtracting the NO reading. The final test was made with the NDIR again connected in parallel with the CML and both analyzing a hot, blended flue gas containing N2. 37= 02, 127, C02, '10% H20, and NO. The NDIR was equipped with the proper traps for pretreating the flue gas. The CML read 116 ppm NO and the NDIR read 111 ppm NO, which is good agreement. Based on these results, we consider the problem of measuring NO in a damp flue gas containing N02 and S02 to be resolved. ------- - 13 - 2 .2 Scrubbing System 2.2.1 FJLue Gas Blending The gas blending system was capable of producing synthetic flue gas consisting of 107, steam, 127,, C02, 37, 02, 3000 ppm S02, 1000 ppm N02, 1000 ppm NO and N2• The concentration of any component could be varied over a wide range. In addition,the system was capable of total flow rates of 80 to 800 SCFH. Figure 1 depicts the total scrubbing system and include a schematic of the gas blending system. All of the gases were derived from pure gas sources except for oxygen, whose source was compressed air. Nitrogen, air, and steam were available in the laboratory whereas C02> N02> NO, and S02 were supplied from cylinders. In order to minimize gas phase reactions between S02 and N02» each, was thoroughly diluted before mixing. The pure SOo was blended with air, C02, and NO, while the pure N02 was dilutee with N2 • After mixing the two diluted streams the blend passed into a heated steam box to receive the appropriate steam flow. The steam supplied to the lab was wet so the steam box provided the heat to dry it out. The dewpoint of the blended flue gas exiting from the steam box was approximately 115°F so that all downstream lines had to be heated . The biggest problem encountered involved N02• This substance is a liquid (N20^) under normal conditions which boils at 70°F under a pressure of one atmosphere. The. vapor pressure rises to 17 psig at 100°F. A small heated shed was constructed just outside the laboratory to hold one cylinder, each of pure N02 and pure S02• It was probably unnecessary to heat the.S02 because of its high vapor pressure but since both cylinders had to be outdoors in winter, we decided to take the extra precaution against the possibility of unusually low temperatures. The shed temperature was maintained at 100-115°F by a thermostatically controlled electric heater.' Because of its low vapor pressure, conventional corrosive gas regulators would not regulate N02 flow effectively. We tried to control the flow with a metering valve but temperature fluctuations in the shed generated large uncontrolled changes in gas flows. Eventually we obtained a specially modified corrosive gas pressure regulator which could effectively regulate the N02 pressure. However, even this controller failed once because of the highly corrosive nature of pure N02• The stainless steel lines carrying the pure NOo were heated up to the point of dilution with nitrogen. The N02 flowmeter was ineffective because droplets of liquid.formed inside the glass barrel; we were unable to conveniently provide sufficient heat to the flowmeter. To simplify the situation, the flowmeter was replaced with a straight piece of stainless steel tubing with a metering valve. .The N02 level in the flue gas was obtained by opening the metering valve until the proper response occurred in the Dupont 400 N02 analyzer. When fixing the various pollutant con- centrations, the scrubber column shown in Figure 1 was by-passed and the flue gas went directly to the gas analyzers. ------- - 14 - All lines were SS 316 stainless steel. All lines delivering gases from their sources were 1/4" i.d. except for the nitrogen and steam lines which were 1/2" i.d. Beyond the first level of blending all lines were 1/2" i.d. Each flowmeter had a pressure gauge on the downstream side and each flowmeter was calibrated with a wet test meter. The blending manifold functioned very well after the problems discussed above were resolved. 2.2.2 , Flue Gas Scrubber The continuous tower scrubber is shown in Figure 1. The vertical glass column is constructed of sections of glass pipe each of which is 8 inches long and 2 inches inside diameter. The sections are coupled with threaded aluminum collars. The collars and the glass pipe were obtained from the Fisher-Porter'Company. The design makes it easy to change the height of the column. Each section contains a thermocouple and a port for removing samples of scrubbing fluids. The top of the column held a demister head packed with glass wool for removing entrained droplets from the gas stream. The entire column sat on a 50 liter heated flask which served as the scrubbing fluid resevoir. The flask was provided with a stirrer. Pressure gauges were located at the top and bottom of the column and a water manometer measured the pressure drop across the entire column. A circulating pump withdrew fluid from the resevoir and pumped it up to the top to be sprayed down the column, countercurrent to the gas flow. The rough pumping rate was controlled at the pump with final adjustment being made at the flowmeter downstream from the pump. A pH electrode was situated in the pump intake for resevoir pH determinations. The temperature of the scrubbing fluid was usually kept at 125°F. The gas and liquid flow rates were sufficiently high to maintain the glass column temperature at 125°F without additional heating or insulation. Also,it was unnecessary to heat or insulate the pump lines . All thermocouples on the scrubbing tower and on the heated lines were connected to temperature recorders . Temperature controllers were used wherever heat was required. When making a scrubbing experiment the following sequence was followed: • Fix the flue gas composition and flow rate while by-passing the scrubber. ------- Figure 1 Flow Schematic of Scrubbing Unit heated steam box L -=3- Steam t r?4i t 0 n Air c\ O o A + A -> Vent Electrical Heater NO-NOX Chemil-- urainescenjce Dupont 400 •F —> Vent Recorder and Control Console ------- - 16 - • Turn on the circulating pump and set rate to predetermined value. • Wait 5 to 10 minutes for column temperature to stabilize. • Switch flue gas through the scrubber. • Monitor pH and temperatures. • Periodically withdraw liquid samples. e Periodically by-pass the scrubber to check the composition of the flue gas. The system performed well; the next section discusses the results of our scrubbing experiments. 2.3 Results of Flue Gas Scrubbing Experiments The construction, testing and trouble-shooting of the flue gas scrubbing system took much more time than anticipated. Con- sequently, only a small fraction of the scheduled scrubbing experiments were completed. Despite these shortcomings, however, an important series.of scrubbing experiments was completed which verified the capability of sulfite ion to effectively absorb N02 from flue gas. The results are shown in Table 9 and were obtained under identical conditions of temperature and flow rates. These results allow several conclusions. The first is that sulfite ion is a much better absorber of N02 than hydroxide ion. Unfortunately neither is effective for NO absorption which also verifies the results obtained in our previous screening study. Also the data show that the presence of S02 in the flue gas enhances N02 absorption by alkaline solutions initially containing no sulfite (see runs Cl and Dl). When absorbed, S02 produces sulfite ion which subsequently reacts with N02, thereby removing it from the gas stream. A similar effect is not observed for NO; the presence of S02 is not significantly beneficial for NO absorption (see runs Al and Bl). Since S02 absorption is primarily dependent on pH, both solutions'were effective for removing essentially all of the S02• These experiments were made with an open column which does not provide good contacting between gas and liquid, although it does give minimum pressure drop across the column. A small amount of column packing or even some liquid distribution plates would enhance contacting and, no doubt, improve the N02 absorption efficiency. However, the Na2S03 solution does quite well considering the poor con- tacting between the two phases. ------- - 17 - Table 9 Flue Gas Scrubbing with NaOH and Na?SCL Solutions Exp # Al A2 Bl B2 Cl C2 Dl D2 Input NO 660 640 680 680 -- -- Levels N02 -- 680 690 690 690 (ppm) SO? 2450 2500 — 2700 2700 % NO 12. 0 19. 6. — --, Absorption N02 -- 12. 83. 48. 83. S02 98. 98. — 99. 99. . Scrubbing Solution L.OM NaOH l.OM Na2SO l.OM NaOH l.OM Na2S03 l.OM NaOH l.OM Na2S03 l.OM NaOH l.OM Na0S00 ' NOTES: . (1) The bulk gas composition was 10% H20, 3% 02, 12% C02, N2. (2) Scrubbing solution temperature = 128°F. (3) 'Gas flow rate = 5. SCFM (4.5 ft/sec in column) (4) L/G = 20. (5) Liquid flow rate = 4. liters per minute. (6) Pressure inside column - 3. psig. (7) Five.glass pipe sections in column; pressure drop across entire column = 3" H20. (8) Each run lasted 15 minutes; took about five minutes for system to stabilize after going on line. (9) Resevoir initially contained 15.. liters of solution. (10) Solution pH was constant during brief runs: l.OM NaOH, pH = 13.; l.OM Na2S03, pH = 9. ; (11) No solution analyses were made. ------- - 18 - Increasing the weight ratio of liquid flow rate to' gas flow rate improves the scrubbing efficiency for N02 and SC>2 as shown in Figure 2. Figure 2 Effect of L/G on NO and SO. Scrubbing with Na SO Percent Absorption 100 — 10 20 30 40 50 NOTES: (1) Initial N02 level = 620 ppm (2) Initial S02 level = 1800 ppm (3) Resevoir contained 13 liters of l.OM Na2S03 (4) Total gas Flow rate = 5. SCFM (5) Scrubbing temperature = 125°F Under these conditions the N02 absorption appears to be limited to 9070 whereas the S02 removal, as usual, is better. In another scrubbing experiment, using the same column con- figuration as before, a magnesium hydroxide slurry was employed as the absorbent for N02 and S02• Table 10 contains the results. ------- - 1.9 - Table LO Flue Gas Scrubbing Using Magnesia Slurry Run time (min) - 0 10 % N02 Absorbed 0 26. % S02 Absorbed 0 95. L/G 10. 10. pH 9.4 9.0 20 35. 95. 10. 8.8 30 35. 95. 10. 8.7 . 40 34. 99. 20. 7.5 50 38. 99. 20. 6.7 60 30. 90. 20. 5.6 NOTES: (1) Initial NOo and S02 levels = 920 ppm and 2600 ppm, respectively (2) Resevoir contained 13. liters of slurry; 10. g Mg(OH)2 per liter. (3) Total pressure drop across column =3.5" H20 (4) Scrubbing temperature = 125°F. (5) Total gas flow rate - 5. SCFM. (6) At L/G = 10, liquid flow rate was 2. liters/min. (7) At 40 minutes on line, reset input levels to 710 ppm N02 and 1920 ppm S02- (8) At 50 minutes, added 2. grams of hydroquinone to scrubbing fluid. These results show that about 1/3 of the N02 is absorbed by the slurry. Doubling the L/G did not have any effect. In the screening program a magnesia slurry (7.4 g/1) absorbed 5870 of the N02 from a flue gas containing 830 ppm N02 and 2460 ppm S02• The poor removal in the present experi- ment implies poor contacting. The screening studies also showed a marked absorption improvement with hydroquinone addition to the slurry. This was not observed in the present case, although the amount of the antioxidant added may have been too small to be effective or else the run did not last long enough for the influence to be demonstrated. In any case more data is required before any conclusions may be drawn concerning the effect of hydroquinone. These scrubbing results verify that sulfite is a good NO- absorbent but a poor NO absorbent. - Also, the effect of L/G is quantified for Na2S03 solutions. However, these results strongly.suggest that more information is required on these and other absorbing systems before engineering and economic evaluations 'can be initiated. The bulk of this information was to be obtained in the third and final phase of the project. ------- - 20 - 3. CONCLUSIONS AND RECOMMENDATIONS 3 .1 Conclusions The second phase of the flue gas scrubbing program was designed to realistically apply the results of the Phase I screening study to the removal of N02 and S02 from flue gases. The primary goals of the project were to design, construct and test a continuous flue gas scrubbing system which would employ sulfite species as reactive absorbents. The program required that suitable analytical procdures be developed so that accurate NOX material balances may be obtained. The conclusions are: (1) Analytical (a) Sulfite ion interferes with the spectrophotometric determina- tion of nitrite and nitrate levels in scrubbing solutions. This interference may be minimized by raising the solution pH above 9 prior to analysis. (b) High levels of soluble carbonate ion will also interfere with the nitrite and nitrate analysis. ' This problem was not resolved. Fortunately, few samples will'exhibit this condition. (c) Difficulties were associated with using a Non Dispersive Infra-Red instrument for analyzing NO levels in a gas stream containing NO, N02» S02 and H20. The problems disappeared when we changed to a chemiluminescence instru- ment modified with a heated inlet system. (2) Results with continuous scrubber (a) The removal of N02 from flue gas is enhanced by the presence of sulfite in the absorbent or S02 in the flue gas. This verifies the results of the Phase I screening study. (b) NO absorption is not affected by sulfite or S02• (c) S02 absorption is excellent in alkaline media. (d) Increasing the L/G ratio improves the N02 absorption in l.OM Na2S03- (e) Only 1/3 of the N02 was scrubbed out by a magnesia slurry. The results are essentially in agreement with the earlier screening study. However, much more data is required before we can effectively evaluate the overall process. ------- - 21 - 3.2 Recommendations for Future Work Our studies have shown that flue gas scrubbing in 3 continuous, open column unit has promise. The following recommendations list the work which remains to be done. • Obtain more scrubbing data using different absorbents such as lime slurry, limestone slurry, and ammonium hydroxide. • Determine the effect of varying process parameters such as column height, superficial gas velocity, and column packing. e Obtain "complete material balances for NOX and S02 • • Investigate solution regeneration techniques. e The oxidation of NOX to NQ2 upstream from the scrubber is vital. Studies should be instituted to optimize this reaction under flue gas conditions. The final goal is to successfully combine NOX oxidation, gas scrubbing, and solution regeneration into an integrated process which cleans the flue gas and minimizes the impact of scrubbing products upon the environment. ------- - 22 - APPENDIX TABLE A1 Measurement of Molar Absorptivities for Nitrite and Nitrate Solution Molarity 0.00975 0.00975 0.01040 0.01040 0.00975 0.00975 •0.00975 0.01040 0.00908 0.00520 0.00520 0.00488 0.00488 0.00908 0.00908 0.01816 0.02270 0.01816 NOTES : (1) Used standard and KN03 . (2) Used an Optica N02~ €302 9.87 9.74 9.52 9.38 9.02 9.25 9.28 9.38 • 9.14 9.00 9.64 9.48 -- -- -- -- "* ™ solutions of e355 25.4 25.3 23.7 23.4 24.1 24.7 24.7 25.0 -- 23.2 23.9 23.6 23.7 -- -- -- -- — mt reagent N03" e302 _ _ -- . -- -- -- -- -- 7.35 -- -- • -- -- 7.43 7.43 7.74 7.65 7.71 grade NaN02 Spectrophotometer . (3) All measurements made with 4.0 cm (4) Used Beer's Law to calculate molar quartz cell. absorptivity A = ebc A = measured absorbance e = molar absorptivity b = cell path length in centimeters c = solute concentration in molarity ------- Unclassified Security Classification DOCUMENT CONTROL DATA - R 8, D (Security clalslflcatlon ol title, body ol abstract and Indexing annotation mutt be entered when the overall report Is clamlllfd) l.-ORIGINATING ACTIVITY (Corporate author) Esso Research and Engineering Company P.O. Box 8 : Linden, New Jersey 07036 . la. REPORT SECURITY CLASSIFICATION Unclassified 26. CROUP Air Conservation 3. REPORT TITLE Development of Aqueous Processes for Removing NOx from Flue Gases . . . Addendum 4. DESCRIPTIVE NOTES (Type ol report and Inclusive dates) Final Report, July 1; 1972 - June 1, 1973 5. AUTHOR(S) (Flrit name, middle Initial, lamt nama) Gilford A.. Chappell «. REPORT DATE June 1973 ta. CONTRACT OR GRANT NO. EPA 68-02-0220 b. PROJECT NO. C. , ^' •' d. Ta. TOTAL NO. OP PACES 7b. NO. OF KEFS 25 2 »B. ORIGINATOR'S REPORT NUM"BER<3) GRU.2PJAA.73 ab. OTHER REPORT NO(S) (Any other number* that may be maalgnad this report) 10. DISTRIBUTION STATEMENT Unclassified - Distribution Unlimited II. SUPPLEMENTARY NOTES 12. SPONSORING MILITARY ACTIVITY Office of Research and Monitoring U.S. Environmental Protection Agency 13. ABSTRACT This report summarizes the findings of the laboratory program for "Development of Aqueous Processes for Removing NOX and S02 from Combustion Flue Gases." This project is the second phase of the flue gas scrubbing work sponsored by EPA under Contract No. 68-02-0220. The results of the Phase I program are contained in report EPA-R2-72-051, entitled, "Development of the Aqueous Processes for Removing NO^ from Flue Gases." The present report contains'discussions of analytical techniques and scrubber design in addition to experimental results obtained with a vertical spray tower scrubber. The blended flue gases passed up the unpacked glass column countercurrent ,-tb the absorbing solution which was sprayed down from the top. The scrubbing experiments showed: o N02 and S02 are effectively absorbed by 1.0 molar Na2S03 solutions. o N02 absorption by 1.0 molar NaOH solution is enhanced by the presence of S02 in the flue gas. • Neither NO nor N02 is effectively absorbed by 1.0 molar NaOH solution in the absence of S02, and NO absorption is not improved by the presence of S02 • • Increasing the L/G ratio improves N02 and S02 absorption by 1.0 molar 0 Under, similar scrubbing conditions Mg(OH)2 slurry is not as effective as solution for N02 absorption. The data show that sulfite solutions would effectively absorb NOx and S02 from flue gases provided the NOx (mostly NO) has been oxidized to N02 upstream from the scrubber. DD /.T..1473 MKPLACKB DO FORM 147*. I.JAN O4. WHICH ID OOBOLKTB PON AMMY U»B. Unclassified Security Cloooiflcotion ------- Unclassified Security Classification 14. K«V WORD* Absorption Air Conservation Air Pollution Chemical Analysis Flue Gases Flue Gas Scrubbing Gas Scrubbing Nitrogen Oxides NO . x S00 2 Sulfites LINK A MOLB WT LINK • KOLB: •FT LINK C ' MOLC »T , i Security Claaelficotlon ------- BIBLIOGRAPHIC DATA SHEET 1. Report Nb. EPA-R2-73-051a 3. Recipient's Accession No 4. Title and Subtitle Development of Aqueous Processes for Removing NOx from Flue Gases -- Addendum 5- Report Date June 1973 6. 7. Author(s) Gilford A. Chappell 8- Performing Organization Kept. No. 9. Performing Organization Name and Address 10. Project/Task/Work Unit Mr EPA, Office of Research and Monitoring NERC/RTP, Control Systems Laboratory Research Triangle Park, North Carolina 27711 11. Contract/Grant No. 68-02-0220 12. Sponsoring Organization Name and Address Esso Research and Engineering Co. Government Research Laboratory Linden, New Jersey 07036 13. Type of Report & Period Covered 14. 15. Supplementary Notes 16. Abstracts rp^g repOrt summarizes the findings of a laboratory program for developing aqueous processes for removing NOx and SO2 from combustion flue gases. It discusses analytical techniques and scrubber design, as well as results obtained experimentally with a vertical spray tower scrubber: blended flue gases passed up an unpacked glass column, countercurrent to the absorbing solution which was sprayed down from the top. The experiments showed that: NO2 and SO2 are effect- ively absorbed by 1.0 molar Na2SO3 solutions; NO2 absorption by 1. 0 molar NaOH solution is enhanced by SO2 in the flue gas; neither NO nor NO2 is effectively absorbed by 1. 0 molar NaOH solution in the absence of SO2 (NO absorption is not improved by SO2); increasing the L/G ratio improves NO2 and SO2 absorption by 1.0 molar Na2SO3; and under similar conditions, Mg(OH)2 slurry is not as effective as Na2SQ3 solution for NO2 absorption. 17. Key Words and Document Analy Air pollution Nitrogen oxides Sulfur dioxide Combustion Flue, gases Chemical analysis Design Washing Spraying 17b. Identifiers/Open-Ended Terms Air pollution control Stationary sources Aqueous processes Scrubbers Liquid/gas ratio sis. 17o. Descriptors Sodium sulfites Absorbers (materials) Sodium hydroxide Nitrogen oxide Nitrogen dioxide Magnesium hydroxides Sulfites Slurries 17c. COSATI Field/Group 13B, 7A 18. Availability Statement Unlimited 19. Security Class (This R c port) UNCLASSIFIED 20. Security Class (This Page UNCLASSIFIED 21- No. of Pages 25 22. Price FORM NTIS-35 (REV. 3-72) USCOMM-DC ------- |