RICHARD F. GAERTNERI and J. A. KENT West Virginia University, Morgantown, W. Va.
Conversion in a Continuous Photochemical Reactor A significant advance can be made if practical design methods are developed, using fundamental data on flow characteristics, extinction coefficients, a n d reaction velocity constants EmCmoMAGmTIC radiation for activation of chemical reactions has been used to a sign'ificant extent in commercial processes for less than 30 years. Most industrial applications have been batchwise, but the trend is toward continuous reactor systems. Annular concentric reaction vessels in photocatalytic chlorinators having temperature control chambers about a linear, high wattage, high pressure mercury-arc lamp improve both liquid- and vapor-phase reactions (7). A similar apparatus yielded faster reaction rates in the irradiation of vitamin D (7). Elliptical, spray, bubble, and film types of reactors are mentioned (2, 3 ) . Practically nothing has been published on the operating characteristics or design techniques of continuous photochemical reactors or the effect of flow pattern on conversion or of turbulence in securing higher reaction rates (3). The development of a two-stage annular type of photochemical reactor for chlorination of benzene (5) represents the only reference on reactor design. The investigation was undertaken to study variables affecting conversion in a continuous photochemical reactor of the elliptical type having a concentric annular reaction chamber. It included an experimental study of the effect of reactant flow rate and of annular reaction chamber dimensions on the photolysis of an aqueous uranyl oxalate solution. A theoretical expression was derived relating the size of the reaction chamber, intensity of light source, extinction coefficient of reactant, and laminar flow rate of reactant with the conversion of a reactant solution having only a primary light reaction. Experimental Equipment
'
separated by pipe spacers to facilitate good heat transfer from the reflector to the surrounding atmosphere and joined with four steel rods, 5 feet long. Twentygage bright rolled aluminum sheet was used because it is easily shaped and light in weight, and has high reflectance and resistance to corrosion. I t has a reflectivity of about 83% with light of 2500-A. wave length, and 92% with light of 4000-A. wave length (6). The reflectivity of silver a t these wave lengths is 30 and 87%; of chromium, 55 and 70%; and of nickel, 42 and 52%. The reflecting surface of the aluminum was polished with alundum powder 2 months prior to the first experimental run to permit a layer of aluminum oxide to form, thus ensuring constant reflectivity while the reactor was in operation. The light source was a 3000-watt General Electric UA-9 Uvariac mercury-vapor arc lamp, with a lighted length of 50 inches and a diameter of 15/18 inches, mounted at the upper focus of the reflector. In this position the danger of material coming into contact with the hot lamp was reduced and an accessible position was provided for the annular reaction chamber at the lower focus. The outer tube of that portion of the annular reaction chamber which passed through the reactor was made of Vycor brand 7910 glass. This glass is highly transparent to both visible and ultraviolet light, and transmits 7070 of the radiation of 2540-A. wave length through a thickness of 2 mm. The tube had a nominal diameter of 1 inch and a length of 5 feet. To eliminate turbulence in the reactant solution arising from entrance effects, the Vycor tube was joined to a Pyrex glass tube of the same diameter, extending 3 feet in front of the reactor. The Pyrex tube,
and all other transparent tubing containing reactant solution prior to passing through the reactor, were coated with black paint to prevent premature conversion of the reactant solution by extraneous light. High pressure Pyrex gage glass in 10-foot lengths and four nominal sizes//s, a/4, 5/8, and l / ~inchwas used as the inner tube of the chamber. Table I gives average values of a series of measurements determined with a micrometer along each tube.
Table 1.
Annular Reaction Chamber Data
1
Outer tube, Vycor 1-inch Inner tube, Pyrex ?/a-inch 3/a-inch 6/g-inch l/S-inch
Diameter, Inch
CrossSectional Area, Sq.Inch
0.984 I.D.
...
0.855 O.D. 0.733 O.D. 0.609 0.486 O.D.
0.186 0.338 0.469 0.575
The ,maximum allowable surface temperature of the glass envelope of the light source was about 1200' F. T o prevent overheating of the light tube, and excessive temperature of the reactor, a fan provided forced-air cooling. Controlled with a Variac, it produced an adjustable air velocity in the reactor u p to 10 feet per second. A metering pump with variable speed transmission fed the reactant solution through the reactor. Its capacity ranged from 0 to 6.4 cubic feet per hour. To minimize temperature rise in the reactant solution resulting from radiation from the mercury arc lamp and the heat
An elliptical photochemical reactor was chosen because of its flexibility. The functions of light transmittance and of transferring the heat of reaction are separate. Therefore a study may be made a t constant light intensity without interference from the light source. Rectangular wooden ribs having an elliptical hole cut in the center were used as the framework. The ribs were Present address, University of Illinois, Urbana, Ill. VOL. 50, NO. 9
SEPTEMBER 1958
1223
13.0
12.0
Olameter of Outer Tube: 0.984" 1.D.
11.0
Diameter ot Inner Tube:
ion 9.0 Iy
t
6.0
ct
6.0
i
7.0
c
5
f
5.0 4.0
9.0 2.0
r .O 0 0
400
1200
800
Volumelrlo
I600
Flow R o t e ,
2000
2400
2600
5200
CC Per Wlnute
Figure 2. Effect of volumetric flow rate on conversion shows good reproducibility
of chemical reaction, water at 32' F. was continuously circnlated through the inner tube of the chamber cocurrent with reactant solution passing through the annulus.
Procedure The photolysis of oxalic acid sensitized by uranyl ion in the form of uranyl nitrate was chosen because it has a zeroorder light reaction, a temperature coefficient of approximately unity, no dark reaction, great width of absorption band, small effect of added electrolytes, and simplicity of analysis (8). The reactant solution, approximately 0.05 M in oxalic acid and 0.01M in uranyl nitrate, was prepared in a 12-gallon glass carboy shielded from light. After the solution had been thoroughly mixed and cooled to approximately 50' F., oxalic acid concentration was determined volumetrically in several portions with 0.1N potassium permanganate
(4). The analysis was reproducible to within 1 to 2 parts per 1000. A typical run was begun after the light source had attained constant operation-several minutes after power was supplied, indicated by constant current and voltage to the lamp, and by degree of fluctuation of the air temperature a t the center of the reactor. Reactant solution was then pumped through the annular space of the reaction chamber a t a constant rate until the temperature of the solution leaving the reactor indicated a steady-state condition. The temperature rise of the solution passing through the reactor varied with flow rate, but ranged between 7' and 40" F. The reaction was as follows, the star representing the activated state: UOz++
HzCz04
+
H&zOa*
+ hp
-.t
(UOz+')*
(UOz++)* +
HzC204*
+
uoz++ --, HzO + COz + CO
Table II. Typical Experimental Run Data Indicated Vol. Light O.D. of Pump low Linear Contact Temp. of Liquid, Run Inner Tube, Speed, Rate, Velocity, Time, OF. % No. In. Unitless Cc./Min. In./Min. See. Inlet Outlet Conversion 3
0.855
5 15 20
420 840 1270 1680
137.7 275.4 416.5 551.0
24.4 12.2 8.1 6.1
49.9 48.2 47.7 47.5
82.0 69.0 62.5 59.5
4.41 2.18 1.37 1.07
10
8
0.733
3 4 6 10 20
250 340 505 840 1680
45.1 61.4 91.2 151.5 303.2
74.5 54.7 36.8 22.2 11.1
54.5 53.5 52.5 51.7 50.4
85.5 83.5 79.5 70.2 64.5
12.08 8.65 5.97 3.56 1.40
5
0.609
5 10 15 20 25 30 35
420 840 1680 2100 2520 2950
54.6 109.4 165.2 218.8 273.2 328.0 384.0
61.5 30.7 20.3 15.4 12.3 10.2 8.8
48.5 46.5 46.0 45.9 45.8 45.8 45.8
82.0 68.8 61.0 57.1 55.5 54.0 53.0
10.65 5.30 3.59 2.78 2.31 2.22 1.69
6 7 8 9 12 35
505 590 670 760 1010 2950
53.6 62.6 71.1 80.7 107.2 313.2
62.7 53.7 47.3 41.6 31.3 10.7
49.5 49.5 49.0 49.5 49.0 48.6
91.0 85.0 83.5 80.0 73.0 57.0
9.70 8.30 7.40 6.67 5.08 1.78
10
1 224
0.486
1270
INDUSTRIAL AND ENGINEERING CHEMISTRY
A 250-ml. portion of the solution leaving the reactor was collected in a glass flask, and placed in a dark cabinet for several hours to cool and to permit any undissolved gases to escape. After mixing to ensure a representative sample of the run, oxalic acid was determined in several 50-ml. aliquots. Runs were made over the entire range of the metering pump capacity, using the four sizes of annular reaction chambers. Although the literature states that temperature had little effect on the quantum yield for the photolysis of uranyl oxalate ( 8 ) , runs were made to determine if the temperature rise in the reactant solution affected conversion. At a given flow rate, conversion was determined a t two values of temperature rise, with and without circulation of cooling water through the reactor chamber. No difference in conversion was observed. Several repetitions of earlier runs showed that the intensity of the light source remained constant throughout the experiment.
Theoretical Analysis A reactant, as a single homogeneous liquid phase, is passed through the annular portion of the chamber (Figure l ) in the presence of light; it is assumed that no secondary reaction takes place other than that which results directly from the absorption of electromagnetic radiation. If the reactant is passed through the annulus a t a low rate, in laminar flow, a velocity gradient will be established. At any point on the annular cross section, the reactant liquid will flow with a different linear velocity, and will have a different retention time in the annulus than a t some other position. According to the Roscoe-Bunsen law of photochemistry (9), the amount of material transformed in a photochemical reaction is proportional to the product of the light intensity and the time of illumination. The retention time in the annulus, or the time of illumination, is a function of the velocity gradient and varies with the position of reactant liquid in the annular cross section. The light intensity in the annulus is an exponential function of the light path distance, and varies with the position of the reactant liquid. As illumination time and light intensity both depend upon POsition, a different conversion is obtained in each incremental "laminar layer" of liquid passing through the annular tube. The average conversion in all the laminar layers is the conversion measured experimentally. This value, in total moles converted per total volume of reactant liquid passing through the reaction chamber, may be correlated mathematically with the experimentally determined variables of the system.
CONTINUOUS PHOTOCHEMICAL REACTOR 13.0 12.0 I io
t0.0
9.0
where 1 and 2 refer to the laminar layers a t the surface of the inner and outer tubes, respectively, which constitute the annular reaction chamber. In terms of a laminar layer of solution, the Roscoe-Bunsen photochemical law is written :
M = KI&
:8.0 L
5 u e
7.0
60 60
e 4.0
a.
(2)
Diameter of lnnrr Tube:
30 2.0
or
I .o
dN = KZzt dq
0
(3)
0
20
10
30
40
z@-'
(4)
t =
L/v
(5)
(14)
dq = u dA
(6)
Purday (70) showed that for an annular chahnel:
dN = K l o ~ - u " L dA
(7)
and,
however, dA = 2Hrdr x =.ra
(8)
-I
alsb, A = n(r22
(9)
After combining Equations 7, 8, and 9, and integrating with respect to r, the following equation is derived:
- r?)
B = L/t
(16) (17)
and, C' =
c/c
(18)
Combining Equations 14, 15, 16, 17, and 18 gives: (rlu
- 1) e% -
7d]
(10)
This relationship gives an expression for the numerator of Equation 1. T o obtain a relationship for the denominator Purday (70) has shown that the velocity gradient for a liquid in laminar flow in an annular channel is given by: u =
C' = K'Io
but, dq = u dA = v(2rr dr)
from Equations 1, 10, and 13,
BO
Under constant operating conditions Equation 19 shows direct linear relationship between conversion and the average time the reactant is in contact with light in the reactor. I t also indicates how reaction chamber dimensions, initial light intensity, and light extinction coefficient of the reactant influence conversion. Constant K'Z,, best determined experimentally, may be approximated by calculations based on Einstein's law of photochemical equivalence and of knowledge of the quantum yield for the photochemical reaction. Evaluation of Theoretical Correlation
Application of Equation 19 to the system studied experimentally required determination of the effective extinction coefficient of the light emitted by the UA-9 mercury arc lamp through the uranyl nitrate reactant solution. The per cent light transmittance through the reactant solution was determined with
Average Light Transmittance of UA-9 Photochemical Lamp through 1 Cm. of Reactant Solution Wave Length Range Power Output Light Transmittance, % Weighted A. Watts %
' Near-UV
(12)
By combining Equations 11 and 12, and integrating, the following result is obtained:
70
Table 111.
Middle UV
(11)
BO
Effect of light contact time on conversion is linear
Figure 3.
I, =
Therefore,
60
Contact Time, Seconds
Lambert's law describing the intensity of light transmitted to a laminar layer of reactant solution a t a distance x from the inside surface of the outer glass tube is:
Visible
2800-2900 2900-3000 3000-3100 3100-3200 3200-3300 3300-3400 3400-3500 3500-3600 3600-3700 3700-3800 3800-4000 4000-4100 4100-4300 4300-4 400 4400-5400 5400-5500 5500-5700 5700-5800 5800-7600
0.01 0.06 0.24 1.73 0.13 1.32 0.39 0.53 53.10 0.79 1.93 40.80 2.42 70.80 6.53 84.30 1.60 78.50 9.72
...
0.02 0.07 0.49
0.04 0.37 0.11 0.15 14.96 0.22 0.54 11.50 0.68 19.95 1.84 23.72 0.45 22.11 2.74
... ...
... ... ... 0.01
2.0 7.0 26.0 47.5 64.7 73.5 78.0 74.0 67.0 75.0 90.0 98.8 98.8 98.8 98.8
0.01 0.02 0.03 0.07 9.68 0.16 0.42 8.51 0.46 14.97 1.65 23.42 0.45 21.82 2.71 84.39
... 2.0
Av.
C = N/q = VOL. 50, NO. 9
SEPTEMBER 1958
1225
.
. Cooling Water Reservoir
t
a Beckman quartz spectrophotometer between 3200 and 7600 A. By using these data and the spectral power output data supplied by the manufacturer of the lamp, average per cent light transmittance was calculated (Table 111). Taking this value of average per cent light transmittance as the light transmittance of a pseudomonochromatic light, the extinction coefficient was calculated by Lambert’s law, giving a value of 0.4320 per inch.
The constant term relating conversion and flow rate in Equation 19 is made up of two factors, K’lo, and a slope constant, s, defined as: s =
rzu - I
-
(rlu
-
1)e~(~1-~2)
c A uz
(20)
K’Io is constant for a particular light source-reactant system, but the slope constant depends on the dimensions of the annular reaction chamber used in the photochemical reactor. The slope constants representative of the four reaction chambers used were calculated (Table IV). They were almost numerically equal, average deviation from the mean being less than 1%. Discussion of Results
The relationship between volumetric flow rate of reactant solution and per cent conversion of oxalic acid is shown in Figure 2. Each parameter represents the diameter of the inner tube of one of the annular reaction chambers. The data used in plotting each parametric curve were taken from two or more runs, performed at different times.
Table IV.
Nom. Diameter of Inner Tube, In.
Slope Constant,
6/8
a 3.19 3.21 3.18
1/2
3.08
’/E 3/4
1226
Slope Constants
As no experimental data were eliminated from the plots and the points generally fall on a smooth curve, reproducibility of data appears good. Figure 3 shows the effect of light contact time on conversion. As predicted by Equation 19, this is a linear relationship. A single line may be passed through the points representing all four annular reaction chambers without serious discrepancy. This agrees with the fact that the slope constants calculated from Equation 19 were almost numerically equal. As several terms, including an exponential term, comprise the slope constant, the relative variation of the terms with respect to each other was such that for the particular extinction coefficient and range of reaction chamber dimensions used, the over-all cancellation effect between terms caused the slope constant to remain essentially unchanged. Had another light source or reactant been used. there is no reason to believe that the slope constants would have behaved in the same manner. If the UA-9 lamp was replaced by a low pressure mercury germicidal lamp of lower power output emitting 93% of its total light at 2.537 A , , the initial light intensity would be lower and the extinction coefficient different from that obtained with the UA-9 lamp. Conversion would vary directly with contact time in the reactor, but the rate of increase of conversion with contact time would depend upon the extinction coefficient, which in turn depends upon the radiant energy distribution. Therefore, at a certain value of contact time it would be possible to have a higher conversion with the less powerful light source. This depends on light absorption characteristics of the reactant. Future studies should be made with some of the more common reactant systems, especially those exhibiting secondary chemical reactions which would involve additional kinetic interpretation of conversion. Nomenclature A = cross-sectional sq. inches
INDUSTRIAL AND ENSINEERING CHEMISTRY
original concentration of reactant, moles per liter C = average conversion, total moles converted per liter of reactant passing through annulus C’ = fraction of reactant converted, C/’c S = total moles converted per second I, = intensity of light at distance x from inside surface of outer glass tube IO = initial light intensity at inside surface of outer glass tube K,K‘ = constants L = illuminated length of annular reaction chamber, inches M = rate of conversion in a laminar layer having an incremental volume (dq), moles,/liter, second n = ratio of pressure to viscosity, P / z P = pressure drop through annular reaction chamber r1 = outside radius of inner tube, inches rp = inside radius of outer tube, inches r = distance of laminar layer from center of annular reaction chamber, inches s = slope constant in Equation 19 = rgu - 1 - ( r l u - l ) & - r ? ’ cAu2 liter per mole S = ratio of tube radii, rz/’rl t = length of time reactant in laminar layer is in contact with light, seconds t = average contact time of reactant in all laminar layers with light, seconds u = light extinction coefficient, llinch u = linear velocitv of reactant in laminar layer, inches per second 3 = average linear velocity of reactant in all laminar layers, inches per second q = volumetric flow rate, cc. per second x = distance of laminar layer from inside surface of outer tube. inches z = viscosity, lb./inch second “ c
=
literature Cited
(1) Anderson, W. T., I ~ DENC. . CHEM. 39, 844 (1947). (2) Beggs, E. W.. Sci. American 178, 109 (1948). (3) Doede, C. M., Walker, C. A., Chem. Eng. 62, 159 (1955). (4) Fowler, R. M., Bright, H. A,, J . Research Natl. Standards 15, 493 (1935). (5) Governale, L. J., Clarke, J. T., Chem. Eng. Progr. 5 2 , 281 (1956). ( 6 ) Koller, L. R., “Ultraviolet Radiation,” Wiley, New York, 1952. (7) Krautz, Erich, Abhandl. braunschweig. wiss. Ges. 4, 5 (1952). (8) Leighton, W. G., Forbes, G. S.. J . Am. Chem. SOC.52, 3139 (1930). (9) Noyes, W. A., Leighton, P. A., “Photochemistry of Gases,’’ Reinhold, New York, 1941. (IO) Purday, H. F. P., “Introduction to the Mechanics of Viscous Flow,” Dover Publications, New York, 1949. RECEIVED for review March 7, 1957 ACCEPTEDJanuary 17, 1958
area of annulus,
Work supported by Engineering Experiment Station, West Virginia University.