Adsorption of nitrogen oxides from waste gas - Environmental Science

Publication Date: February 1967. ACS Legacy Archive. Cite this:Environ. Sci. Technol. 1967, 1, 2, 151-156. Note: In lieu of an abstract, this is the a...
0 downloads 0 Views 505KB Size
Adsorption of Nitrogen Oxides from Waste Gas B. B. Sundaresan,' C. I. Harding, F. P. May, and E. R. Hendrickson Air Pollution Research Laboratory, University of Florida, Gainesville, Fla. 32601

rn Experimental studies showed that a commercial zeolite

(molecular sieve) was more efficient than silica ge! in removing low concentrations of nitrogen oxides from gas streams. Further studies proved that the zeolite could effectively adsorb NO, from waste gas streams in a nitric acid plant. Adsorbed NO, was recovered as enriched NO, and H N 0 3 by regenerating the bed at elevated temperatures with hot air and/or steam. Tests indicate that such a system could be incorporated into a n existing nitric acid plant, preventing release of significant quantities of NO, into the atmosphere. A patent application has been filed.

itrogen oxides result from a variety of high temperature combustion processes, such as power boilers and automobile engines, from sulfuric acid production by the chamber process, and from the manufacture of nitric acid. These highly reactive oxides are one of the major air contaminants in many cities. Their major role in community air pollution is the photochemical oxidation of organic material to produce Los Angeles-type smog. Limits for emissions of oxides of nitrogen from automobile exhausts have been established to attack the single largest source of these oxides. This work was undertaken to investigate the feasibility of recovering the oxides from waste gas streams in nitric acid manufacture. Catalytic oxidation of ammonia t o form nitric oxide is commonly used in nitric acid production. This process consists of passing a mixture of preheated air and ammonia over a suitable catalyst, cooling the gas by heat exchange with the incoming air, diluting with secondary air, and absorbing nitrogen dioxide in recirculating nitric acid. The entire process consists of three essential steps. + 5 o2( P t - Pd catalyst 1500-1700° F.)+ 4 NO + 6 ~~0 4

2 NO

+

+

0 2

-+

2 NO2

(1) (2)

+

3 NO3 H20 2 "03 NO (3) The reaction of Equation 3 leads to the release of oxides of nitrogen in the tail gas, which also contains about 3.00 to 4.00 volume % oxygen. These oxides (NO,) in the tail gas are essentially in the form of NO as they enter the stack, but conditions are favorable for their conversion to NOz when discharged into the atmosphere. NO, NOz, and N2O4equilibria represented in Equation 4 indicate that this conversion takes place under normal atmospheric conditions (Kolthoff and Elving, 1961). 32' t o 283'F.

-+

302' t o 114S'F.

'2NOP. ' 2 N O + Oz (4) Colorless Brown Colorless F o r nitrogen oxides concentrations of about 0.40 volume o r less it is uneconomical to effect further conversion to nitric N20,

1

Present address, University of Madras, India.

acid in the absorption tower. Concentrations above 0.02 volume may constitute a potential air pollution problem. Nitric acid manufacturers have been searching for a suitable method to recover and utilize the released N O and thus improve plant efficiency. A review of the literature on methods for removing nitrogen oxides from waste gas streams indicates that these methods may be grouped broadly in four categories: absorption in aqueous solutions, vapor phase reactions. catalytic combustion, and adsorption. Peters and Holman (1955) report that both gas- and liquidphase reactions involving nitrogen dioxides occur when gases containing dilute concentrations of nitrogen dioxide come in contact with various aqueous solutions. In general, the efficiency of nitrogen oxide removal from low concentrations by absorption was found to be poor. Ammonia has been made to react with dilute nitrogen oxides to form aerosols of ",NOU and ",NOs which were collected by acoustic coagulation. Waste gases have also been treated by mixing with reducing gases such as Hp o r natural gas and subsequent combustion over catalysts such as platinum alloys, Ni-Cr, or platinum on Al2O3. Adsorption of NO? o n silica gel at low temperatures and subsequent desorption at 100" C. were reported by Almquist et a/. (1925). Foster and Daniels (1951) used silica gel to recover nitric oxide resulting from the atmospheric oxidation of nitrogen. The 1.00 to 1.50 volume nitric oxide in flue gas was adsorbed and oxidized to nitrogen dioxide in a silica gel bed by chemisorption. NO2 was recovered by heating the gel o r recirculating a hot mixture of NOp and air. Baker et al. (1952) found that the rate of catalytic oxidation of nitric oxide on silica gel could be expressed by the rate of the surface reaction between oxygen in the gas phase and the nitric oxide complex adsorbed on the gel. However, adsorption of nitrogen oxides (NO,) at low concentrations of 0.10 to 0.40 volume using silica gel o r other commercial adsorbents has rarely been attempted. The literature survey o n these four methods of removing low concentrations of nitrogen oxides disclosed serious technical and/or economic drawbacks to all the methods except adsorption. The little information that was available o n adsorption was encouraging, so this process was picked for the first experimental work. In the end, results with adsorption were so encouraging that no other method was tested experimentally.

z

z

Molecular Sieces as Adsorbents I n addition to silica gel and alumina, commercial zeolites such as molecular sieves have been widely used as adsorbents in air-drying operations. This common knowledge, plus the information that there were available commercial grades of molecular sieves which had a high degree of acid resistance, generated an interest in these zeolites as potential adsorbents for nitrogen oxides. Silicates, alumina, and one o r more cations such as potassium, sodium, o r calcium in varying degree form the basis of these commercial zeolites. A tetrahedron o r pyramid of Volume 1, Number 2, February 1967 151

four oxygen atoms around silicon o r aluminum ion is the basic building block. Na+, K+, o r Ca2+ electrically neutralizes the aluminacentered tetrahedra (Chemical Engineering, 1964). The corner oxygen atoms are shared by each of the tetrahedrons and so the structure extends in all directions. Adsorbent properties of these crystalline structures depend upon the honeycomb of cavities of about 11-A. diameter linked by 4- to 9-A. pores. The cavities and pores are uniform in size and hence “pore size distribution” normally referred to in the case of other adsorbents does not apply to molecular sieves. This uniformity in cavities and pores introduces an entirely new factor which overcomes many limitations of other adsorbents. The uniform distribution of cavities and pores permits sieving o r screening of molecules of different effective sizes. Molecules smaller than the effective pore opening in the crystal would normally be adsorbed. Molecular sieves adsorb water molecules in preference to other molecules because the sieve pores have been created during the manufacturing process by driving off the water of hydration without causing

collapse o r rearrangement of the crystalline structure. Unsaturated hydrocarbons are adsorbed in preference to saturated hydrocarbons. Polar molecules are adsorbed in preference to nonpolar ones. The strong electrostatic field within the crystal lattice of aluminosilicate enables it to function as a catalyst also. Experimentul An experimental program was initiated to determine the feasibility of removing nitrogen oxides by adsorption from gas streams containing low concentrations of the oxides. Silica gel and a commercial zeolite were selected as the first adsorbents to be studied. Two stainless steel columns, each 3 inches in i.d. and 28 inches high, with thernlocouples, flowmeters, and pressure gages were assembled as indicated in Figure 1 . Two gas cylinders with metering devices were used to prepare simulated nitric acid plant tail gas for the system. The first cylinder contained a mixture of oxygen and nitrogen with 3.50 volume oxygen, while the second one contained nitric oxide (NO). Flow rates from each cylinder were adjusted to give the required gaseous mixture containing 0.15 to 0.30 volume NO. A dry mixture was used in the first series of tests to select a suitable adsorption medium. Subsequently, to simulate better conditions in nitric acid plants, the mixed gas was passed through a bubbler to introduce moisture into the gas stream. Samples of the gas entering and leaving the column were Figure 1. Experimental adsorption apparatus

Silica gel drvers Air heater . C. Rotameter D. Bubbler, stainless steel E . Pressure gage, stainless steel F. Multipoint switch G . Wire mesh, No. 100, stainless steel H . Sampling flask I. Cylinder, NO gas J . Cylinder, 3Sz 02,96.5% NI K. Sampling port Adsorbent Coluinn.1. Silica gel, 6-16 mesh Column 11. Molecular sieve A.

B.

Air

152 Environmental Science and Technolog)

collected periodically, in 2-liter spherical flasks. Several turnovers of gas were passed through the flask before it was removed for analysis. Thirty milliliters of 3.00 hydrogen peroxide was introduced into the flask, which was shaken periodically during the next 2 hours. The nitric acid formed was titrated with 0.01N NaOH and the volume per cent NO, (as NO,) computed after a suitable correction for temperature and pressure differences. Each cycle of operation consisted of adsorbing NO, from a n input gas a t 75" to 95" F. and 1.0 to 8.0 p.s.i.g., stripping NO, a t an elevated temperature in the form NOz and H N 0 3 with steam and/or hot air, and cooling the bed. The two columns were fed separately with different gaseous mixtures. Inlet and exit gas samples were collected periodically for analysis. When the NO, concentration in the exit gas reached the same level as in the input gas, the run was discontinued. The columns were regenerated a t elevated temperatures with hot air and/or steam. The amounts of NO, and H N 0 3 released during stripping were measured to permit computation of a material balance for each cycle.

Performance of the silica gel and the zeolite is compared in Figure 2. NO, removal was appreciably higher in the zeolite than in the silica gel. Hence, the next series of studies using wet gas was restricted to zeolite beds. The first series of runs was conducted in the laboratory using composite wet and dry gases from cylinders. The results obtained were so encouraging that the experimental unit was transferred to a nitric acid plant. Tail gas from the plant was used for further studies reported here. Results and Discussion Eleven cycles of experiments were conducted using nitric acid plant tail gas independently o n each of the two columns, having the same zeolite. Operating conditions, NO, recovery, and moisture removal in all the cycles are summarized in Table I. A typical cycle is shown in Figure 3 to indicate removal pattern and temperature variation in the columns. F o r cycle 1-11 (Figure 3), with feed gas at 2.35 SCFM, exit gas NO, concentration was less than 10 p.p.m. for about 3 hours. From this point on, a gradual buildup of NO, in the

~-

Table I. NO, Removal in Commercial Zeolite Columns (Plant tail gas for feed.) Lb. NO, Removed

Input Gas Temu..

Space Veloc., Vol./Hr. Vol.

Mean Column Temp.,

100-Lb. Bed At 200 p.p.m. End exit of cycle NO,

Lb. H 9 0 Removed 100-Lb. BedAt 200 End p.p.m. of exit cycle NO,

F.

P.s.i.g.

Linear Veloc., Ft ./Min.

1 2 3 4

82 78 80 84

2.00 2.00 2.00 7.50

20.0 20.0 20.0 51 . O

525 525 525 1310

92 85 89 102

2400 1600 3200 800

5.75 5.00 3.80 3.45

6.69 5.86 4.44 4.68

8.26 6.03 5.66 4.26

9.95 6.98 7.60 6.11

5 6 7 8

79 86 77 77

7.50 8.00 7.00 2.10

51 . O 53.0 53.0 27.0

1310 1370 1370 675

94 102 95 92

1050 1000 700 300

4.48 3.16 4.96 4.24

5.76 5.11 5.58 4.96

4.45 3.72 5.23 4.39

5.87 6.55 6.16 5.21

9 10 11

82 94 106

2.80 5.50 6.20

27.0 48.0 48.0

675 1250 1250

95 102 109

200 800 1500

4.42 3.25 2.16

4.42 4.26 3.62

4.43 4.27 2.72

4.43 5.96 5.44

Cycle

.

I

Exit Concn at End F. of Cycle COLUMN I O

COLUMN I1

a

1 2 3 4

87 77 78 80

2.00 1.20 1.60 4.00

20.0 20.0 20.0 40.0

525 525 525 1050

93 85 84 93

1600 1500 2400 600

5.13 4.10 4.91 3.88

5.76 5.03 5.63 5.26

8.26 6.03 5.66 4.26

9.95 6.98 7.60 6.11

5 6 7 8

84 79 79 77

6.70 5.50 6.70 2.00

51.0 51 . O 53.0 27.0

1310 1310 1370 685

97 94 95 91

700 800 1000 400

3.51 4.36 3.68 3.67

4.42 5.38 5.16 4.67

3.50 4.27 3.97 3.71

5.18 5.80 5.74 4.93

9 10 11

82 94 106

2.20 5.80 5.50

27.0 48.0 48.0

685 1250 1250

96 103 110

200 1000 1200

3.96 2.67 2.24

4.13 3.82 3.63

3.98 4.42 2.98

4.18 7.09 4.51

Inlet concentration varied from 1800 to 2500 p.p.m.

Volume 1, Number 2, February 1967

153

exit gas occurred, reaching 1500 p.p.m. at the end of 7 hours. Although some tests were continued until the adsorbent was completely exhausted, most were stopped short of this point and special attention was given to data at an exit concentration of 200 p . p m An arbitrary level of 200 p . p m was selected as the acceptable upper limit for NO, in the exit gas. At the

0

1

3.00

Exit NOx Concentration (1000ppm as N 0 2 )

7 r

.-Vc

8

Slhca gel

1.00

6-16 mesh

x

_I

/

L

- - -on-3

e-------100

0

200

300

400

500

600

E x i t NOx Concentration

700

( p p r n as

BOO

TCoIurnn-Middle

XII-2

-1-3

3

f 3f

ilj,, 0

u 0

0.4

0.8

,

,

,

1.2

1.6

2.0

2.4

Exit NOx Concentration (1000pprn as N O 2 1

e c

f

L

NOp)

Figure 2. NO, removal in silica gel and zeolite

g B

900 1000

Figure 4. NO, removal in repeated cycles

I ;m

90 ;. i 80

-

o*

240 1

2000

800 Exit Gas

400

Time

Figure 3.

- Hours

Typical NO, removal with zeolite

Column I, cycle 11 Column diameter. 3 inches i d . Height 2 ft. 4 inches Molecular sieve. 4.90 lb. Bottom. 2 inches from base Temperature recordings Middle. 14 inches Top. 26 inches Gas flow rate. 2.35 SCFM Linear velocity. 48.0 ft./min. Space velocity. 1250.0 cu. ft. gas/hr.-cu. ft. bed 154 Environmental Science and Technology

200-p.p.m. limit for run 1-11, 2.16 pounds of NO, and 2.72 pounds of HzO per 100 pounds of adsorbent were retained in the bed (Table I). At the end of the cycle, the corresponding values were 3.62 and 5.44, respectively. Volume percentage moisture in the feed gas varied between 0.70 and 0.80 and remained less than 0.20 in the exit gas until the end of the cycle. It appears that moisture was still being removed in the bed, even though the NO, saturation limit had been reached. The simultaneous removal of NO, and HzOmakes the system well suited for treating tail gas from nitric acid plants. Temperatures were recorded at three points in the bed with thermocouples 2, 14, and 26 inches above the base with upward flow of feed gas. Adsorption being exothermic, the bottom part of the bed first exposed to feed gas showed an immediate rise in temperature, but stabilized at input gas temperature level. The middle portion of the column showed a sharp increase in temperature, as high as 35" F. above the input gas temperature. This apparently was due to a combination of convection and the upward movement of the adsorption zone. As equilibrium conditions were reached, the top of the bed also became saturated with NO,. The top temperature reached a maximum and then tended to decrease along with the middle temperature. In process design, the maximum temperature rise in the column should be taken into account in optimizing linear and space velocities. Cycles 1-1 to 3 and 11-1 to 3 were run with the input gas at

77" to 87" F., 2.0 to 3.0 p.s.i.g., with a linear velocity of 20.0 feet per minute, and space velocity of 525 cu. ft. of gas per hour per cu. ft. of bed. NO, removal at different exit concentrations for these six cycles is shown in the lower graph in Figure 4. Cycles 1-4 to 7 and 11-6 and 7 were made with the linear velocity increased to 51.0 to 53.0 feet per minute, and space velocity between 1310 and 1370 cu. ft. of gas per hour per cu. ft. of bed. Cycles 1-1 and 11-1 were made with fresh beds of zeolite, which accounts for the higher NO, removal. After the first few cycles, there was a slight

reduction in NO, loading, but it remained more or less uniform beyond a certain limit, as when the zeolites are used in air-drying operations. In that operation there is a reduction of 20 to 30% in capacity during the first 200 or so cycles, beyond which there is no appreciable reduction (Harwell, 1965). At the 200-p.p.m. level and a linear velocity of 20.0 feet per minute in cycle 1-3, NO, loading was 3.90 pounds per 100 pounds of bed. At the same exit level, in cycle 1-4 (upper graph in Figure 4), NO, loading was 3.45 pounds per 100 pounds of bed with the linear velocity at 51.0 feet per minute.

Airr-

LL'

300 W

4l

3

3

e

e

E

F

0

J----,5O 100 ,I

4.00

4.001

5

L

Lo 21

ln 0

f f

3.00

C ._ > c),

3

"03

b

.-p 2.00

b

p

2.00

L

W c

5

L

c 01

L

f

3.00

c ._0 >

1.00

E

01 VI

0.00 0

l"LL I

0.00

"03

0 2

3

6

8

Time

IO

- Hours

12

14

Steam -,r-Air

1

Adsorption

]Heat + A i r --Regeneration

-

Figure 5. NO, recovery using hot air in column I

Column I. Cycle 5 Column diameter. 3 inches i.d. Height. 2 ft. 4 inches Molecular sieve. 4.90 lb. Desorption Heat. 300-375"F. Air flow. 0.50 SCFM Input NO,[as NOa].2.78 g. moles Recovery As NO, AS HNOa Total

G . Moles

z

1.22 1.26 1.48

43.90 45.40 89.30

16

2

3

6

1

Total Input, G.

Cycle

Moles

1 2 3 4 5 6 7 8 9 10 11

3.36 2.83 2.14 2.26 2.78 2.46 2.69 2.42 2.40 2.06 1.75

Table 11. NO, Recovery from Commercial Zeolite Columns Recovery of NO, in Feed AS ”01 AS NOT Acid strength, G. G. Moles Methoda % Moles % COLUMN I

z

I I I I I I I I I1 I1 I1

25.20 27.20 36.30 19.30 48.20 56.50 56.40 57.60 33,40 19.40 26,80

...

...

...

0.20 0.69 0.37 1.26 1.16 1.30 1.26 2.67 1.53 0.66

7.10 32.40 16.40 45.40 47.20 48.30 52.20 ... 74.40

2.03

...

... 71,80

...

...

1.20 1.22 0.93 0.79 0.95 0.13 0.13 0.57

53.10 43.90 37.80 29.40 39.30

2.56 1 27

79.40 43.80

...

6.20 .,.

Total G.

Moles ...

2.23 ... 1.58 2.48 2.09 2.09 2.21 2.80 1.66 1.23

z ... 78.90 ... 69.29 89.30 85.00 77.70 91.50 ...

90.60 ...

COLUMN I1 1 2 3 4 5 6 7 8 9 10 11 a

3.23 2.90 2.87 2.68 2.26 2.75 2.64 2.38 2.11 1.94 1.65

I I I I I I I I

I1 I1 I1

30.80 28.60 49.70 35.20 39.30 55.00 58.50 58.90 24.60 22.50 17.60

0.27 0.45 0.40 0.34 0.83 1.18 1.20 1.40 2.10 1.33 1.08

5.30 15.50 14.00 12.70 36.70 42,90 45.50 ... . . .

68.60 65,50

. .

...

0.71 0.75 1.02 0.99 1.06 0.08 0.25 0.32

26.60 33.70 36.80 37.50 . . . . . .

12.90 19.40

2.83 1 ,72 ... 1.05 1.58 2.20 2.19 2.46 2.18 1.58 1.40

84.70 59.30

... 39.30 70.00 89.70 83.00 ...

... 81.50 84.90

I = Heat plus air. I1 = Heat plus steam plus air. ~

This reduction apparently is due to the increase in linear velocity. NO, and moisture retained in the bed were recovered during regeneration by hot air and heat in the form of enriched NO, and “Os. The recovery data for all the cycles (1-1 to 11 and 11-1 to 11) are summarized in Table 11. A typical recovery cycle (1-5) is shown in histogram format in Figure 5 . Out of 2.78 gram moles of NO, (as NOz) retained in the bed, 1.22 gram moles (43.90%) as enriched NO, and 1.26 gram moles (45.40x) as 48.2% H N 0 3 were recovered. About 80 to 85% of the total input was stripped out in about 40 minutes, compared to the input time of 7 hours 20 minutes. Recovery could also be effected by using a combination of heat, steam, and hot air. Figure 6 shows a typical recovery pattern, in which 74.40x of total input was recovered as H N 0 3 of 19.40% acid strength. The balance was recovered as enriched NO,. About 80 to 85% of total input was recovered in less than 30 minutes. By varying recovery conditions the recovery could be effected either as enriched NO, and higher strength H N 0 3 o r mostly as lower strength H N 0 3 . In either case, there are several places in the nitric acid manufacturing process where either enriched NO, or 20 to 25% H N 0 3 could be reintroduced, thereby increasing plant production. Thompson (1966) has estimated that in a 300-ton acid plant about 4 to 5 tons per day or 60% nitric acid now being wasted could be added 156 Environmental Science and Technology

~

to production by feeding back the recovered NO, into the process stream. Conclusions A commercial zeolite can remove nitrogen oxides along with an appreciable amount of moisture from nitric acid plant tail gas. The oxides can be recovered as enriched NO, and H N 0 3 for return to the processing system. A patent application on the process has been filed with the U. S. Patent Office. References Almquist, J. A., Gaddy, V. L., Braham, J. M., Ind. Eng.’Chem. 17, 599-603 (1925). Baker, R. W., Wong, H. N., Hougen, 0. A., Chem. Eng. Progr. Syn7p. Ser. 48 (4), 103-9 (1952). Chem. Eng. 71,52-3 (Dec. 21, 1964). Foster, E. G., Daniels, F., Znd. Eng. Chem. 43, 986-92-(1951). Harwell, T. W., Orlando, Fla., personal communication, 1965. Kolthoff, I. M., Elving, P. J., “Treatise on Analytical Chemistry,” Part 11, Vol. 5, pp. 217-303, Interscience, New York, 1961. Peters, M. S., Holman, J. C., Znd. Eng. Chem. 47, 2536 (1955). Thompson, L. H., Consulting Chemical Engineer, Birmingham, Ala., personal communication, 1966. Receiced for reciew Nocember 25, 1966. Accepted February 6 , 1967. Work supporred by a grant from Nitram Chemicals, Inc., Tampa, Fla., and Wilson & Toomer Fertilizer, Co., Jacksoncille, Fla., to the Air Pollution Research Laboratory, Unicersity of Florida.