t
thickness of orifice plate absolute temperature upstream of orifice ratio of principal specific heats mass density of gas a t orifice upstream conditions = substituted gas
= = Y = ?(T,P) =
T
Literature Cited
American Society of Mechanical Engineers, New York, “Fluid Meters,” 5th ed., 1959. Andersen, J. W., Friedman, R., Rev. Sei. Znstr. 20, 61 (1949). British Standards Institution, British Standard 1042, Part 1, 1964. Brokaw, R. S., Natl. Aeronautics and Space Administration, N.A.S.A. T.N.R-81 (1961).
Chakraborti, P. K., Gray, P., Trans. Faraday SOC.61, 2422 (1965). Engel, F. V. A., French, J. W. E., Engineering, London 142, 410 (1936) Eubank, P. T., Znd. Eng. Chem. Process Design Develop. 5 , 203 (1966). Ghazi, H. S. A.S.M.E., Paper 65-WA/FM-3 (1965). Hogh, M. S., Hulf, H. J., “Proceedings of 7th World Petroleum Congress,” Vol. VI, Elsevier, London, 1967. Kastner, L. J., McVeigh, J. C., Proc. Inst. Mech. Eng. 180, 331 (1966). Levy, A., J . Sei. Znstr. 41, 449 (1964). Linford, A , , “Flow Measurement and Meters,” E. & F. N. Spon, London, 1961. RECEIVED for review January 4, 1968 ACCEPTEDMay 27, 1968
BATCH, RECYCLE REACTOR FOR SLOW PHOTOCHEMICAL REACTIONS A.
E. CASSANO, T A K E S H I
M A T S U U R A , A N D J.
M. S M I T H
University of California, Davis, C a y . 95616
A batch-type recycle reactor was developed for studying the kinetics of slow photochemical processes. The equipment performed satisfactorily when tested with acetone photolysis. Solvable operating problems were encountered due to pressure fluctuations caused by the recycle pump. In particular, air leakage in minute amounts induced photooxidation. Enough experimental data were obtained to establish the main products of the photooxidation and photolysis reactions. At the operating conditions (870 mm. of Hg, 97” C., and 4 to 17% acetone in helium) the rate of ethane formation was less than first order in acetone. reactors have come into widespread use (Biskis and Smith, 1963; Butt et al., 1962; Hougen, 1951; Korbach and Stewart, 1964; Maymo and Smith, 1966; Perkins and Rase, 1958) as a means to retain differential reactor operation and yet not be restricted to systems suitable for accurate measurement of small composition changes. Control of velocities through the reactor by recycle also permits reduction in diffusion resistances and temperature variations. Disadvantages may arise due to effects of reaction by-products, the long time sometimes necessary to achieve steady state in the circulation section of the apparatus (for flow-type recycle reactors), and the difficulty in operating a t predetermined compositions. T h e batch-operated recycle reactor appears to be particularly well suited for studying the engineering kinetics of slow, photochemical reactions. Slow photoreactions are not unusual and may occur (1) when the quantum yield is low-for example, less than unity as in nonchain kinetics-and (2) for gaseous reactions. Slow reactions are more likely for gases because the rate of the primary step in a photoreaction is proportional to the amount of light absorbed. This depends upon the absorptivity and concentration of the absorbing species. For gaseous systems both cy and C are commonly orders of magnitude less than for liquids. Increasing the light intensity improves rate of slow photoreactions. However, the increase is limited because of the rather severe restrictions in intensity in available lamps. The improvement is further diminished if radiation of discrete wavelength regions is needed, for this must be achieved by filters, which diminish the intensity. Previously slow photoreactions have been studied by analyzing batch reactors of reasonable size, operated long enough to detect significant changes in concentration. This procedure RECYCLE
is questionable because of mixing problems. A method of achieving complete mixing without the mixing device influencing the radiation path has not been developed. Then there is the further uncertainty of possible heterogeneous termination steps on the surface of the stirring device and their effect on the kinetics. If no mixing is voluntarily introduced, the analysis of the results is uncertain in view of temperature and concentration gradients, Finally, batch results must be adapted to a commercial flow reactor, which probably is the ultimate goal. This report describes a batch photoreactor and shows its applicability for studying the kinetics of the photolysis of acetone vapor (in helium) at about atmospheric pressure. Acetone Photolysis
Gas-phase photolysis has been studied extensively in batch systems (Calvert and Pitts, 1966; Davis, 1947; Noyes et al., 1956; Steacie, 1954). I t apparently has nonchain kinetics and at atmospheric pressure wall reactions are insignificant. I t has been mentioned as a n actinometer (Calvert and Pitts, 1966 ; Melville, 1964) at low concentrations (acetone pressure less than 50 mm. of Hg) and temperatures greater than 130” C. I t has a low absorptivity, the maximum value being aboui 33 liters/(mole) (cm.), defined in terms of loge(Z/Io), near 2800 A. Absorption at lower absorptivities extends from 3300 to 2200 A. with an additional absorption band below 1960 A. (Calvert and Pitts, 1966; Davis, 1947). These features combine to give a slow photolysis. In studying the kinetics of fast, gaseous photoreactionsfor example, chlorinations (Cassano and Smith, 1966, 1967)it is necessary to use small reactors to reduce conversion and VOL. 7
NO. 4
NOVEMBER 1 9 6 8
655
temperature changes. The light intensity a t the reactor wall is evaluated in situ using an actinometer. This has been done using liquid actinometers, but their absorptivities are several orders of magnitude greater than those for gaseous photoreactions. This difference can introduce errors in the intensity evaluation due to changes in reflection in the high and low absorptivity systems. It would be desirable to use, as an actinometer, a reaction mixture which has about the same absorptivity in the same regions of the spectrum as that for the main reaction. When the latter is a chlorination, the acetone actinometer meets these requirements well. If acetone were used in this fashion in a single pass through the small chlorination reactor, the conversion would be too low to measure accurately. This difficulty might be overcome, however, by operating the actinometer system as a batch recycle reactor. This use for recycle photoreactors would be advantageous whenever a fast main reaction and slow actinometer system are combined. T h e photolysis of acetone is believed to include the following main reactions: CH3COCHa CH3COCH3*
+ hv
-P
CHaCOCH3*
+ M (inert gas)
4
CH3COCH3
(1) (2)
Other reactions which may occur are: 2CH3CO. + (CHaC0)z
+ CHaCOCH, CH4 + CH3COCHz. CHiCOCHz. + CH3. + CH3COCHz-CHa
CHI*
+
(7) (8)
(9)
Thus the main products are C O and C2Hs with biacetyl, methane, and methyl ethyl ketone as potential by-products. At temperatures above 130' C., C O and C Z Hare ~ formed with a quantum yield of C O of unity. It is in this range of conditions that acetone might confidently be used as a gaseous actinometer. I n batch recycle operation leaks can have a pronounced effect. This is particularly important in photoreactions where very small concentrations of inhibitors can drastically reduce the main reaction. I n this study of acetone photolysis it was found that oxygen, in concentrations as low as 300 to 400 p.p.m., leaking into the system from the air, completely inhibited the main photolysis (photooxidation is not detectable at oxygen levels less than 60 p.p.m.). Photooxidation occurred in our system when oxygen was present in high enough concentrations. The possible reactions in the 02-acetone system are numerous and complex (Hoare and Pearson, 1964; Pearson, 1963). The main products are CO, Con, H C H O , and methanol, but many others like HzO, CH4, and formic acid have been reported. The available data indicate that at most conditions more COZ and C O are produced than C H I O H and H C H O . Reactions which have been postulated to explain the formation of COz, C H 3 0 H , H C H O , and water are: 0 2
+ CH3CO. + C H ~ C O C H P CH3C03H + CH3COCHz. CHaCO. + CHIO* coz CHoCOaH COz + C H 3 0 H (1 0 ) 2CH30. CHaOH + H C H O 4
0 2 -
4
-+
656
l&EC FUNDAMENTALS
+ + M (inert) CH302. + M CH3* + H C H O + OH. CH302' + OH. CHIOH +
CHI.
0 2
-+
0 2 -
-+
0 2
(11) (12) (13)
Formaldehyde can be oxidized in the presence of oxygen (Whitmore, 1937) to give water and carbon dioxide. Photooxidation increases with temperature and above 200' C. is believed to occur by a chain mechanism. Experimental
Apparatus. The schematic diagram of the apparatus (Figure 1) shows the recycle system (heavy lines), reflectorlamp-reactor system, filter solution section, feed system (left side of diagram), and analytical section (broken lines). Recycle System. The reaction mixture was circulated through the reactor, 3, or bypass 2 and reservoir 1. T h e reservoir (about 5500 ml.) was of 316 stainless steel with Teflon gaskets. Samples were withdrawn after the reactor or bypass through line 5 to the analytical section. The Kemlon duplex compressor, 4, was a diaphragm type with a maximum discharge pressure of 20 p.s.i.g. and capacity of 2 cu. feet per minute. The diaphragms, valves, and heads were stainless steel with gaskets of Teflon. T h e flow rate could be changed by varying the pump speed or adjusting the stroke of the pistons. The sampling system in the chromatograph caused a relatively large pressure drop. This eliminated the possibility of using a glass centrifugal pump of the type developed by Russian investigators. The whole system could be evacuated (V.P. = vacuum pump) and purged (21). Reflector-Reactor-Lamp System. The lamp, 7, and reactor, 3, were placed a t the foci of a n elliptical reflector system, 6, as used in previous studies (Cassano and Smith, 1966, 1967). The reflector was made from a sheet (0.032-inch thickness) of specularly finished aluminum, highly polished. The dimensions were: major axis 600 mm., minor axis 447 mm. (eccentricity 0.666), and height 445 mm. Filter solutions for controlling the wave length distribution of the radiation were circulated through three jackets, 13, surrounding the reactor. Some of the details of mounting the lamp and reactor in the reflector, and the dimensions of the lamp jackets, are shown in Figure 2. Air was circulated through the space between lamp and filter solutions for cooling. The reactor tube (Figure 2) was of optically clear Amersil quartz, 20-mm. i.d., 23-mm. o.d., with a lighted length of 200 mm. The single jacket for heating the reactor was made from a 32-mm. i.d., 35-mm. 0.d. quartz tube. Air was used for heating, since no transparent liquid, stable to radiation in the 2000- to 4000-A. region at temperatures above 100' C., was found. The longitudinal temperature gradient in the reactor, when operated at 97' i 1' C., was no more than 2' C. A Hanovia LL, 189A10, 1200-watt, high pressure quartz lamp (mercury vapor) was used (Hanovia, 1959). The lamp was connected with a voltage stabilizer and transformer, and its output was checked with an Eppley, 16-junction thermopile, mounted at the mid-height of the reflector on the lamp side. Fluctuations in the thermopile reading were within &5%. Filter Solutions. The filter solutions were preirradiated to ensure stability and then circulated a t a controlled temperature through the lamp jackets, as shown in Figures 1 and 2. A 1.5 molal NiS04.6HzO aqueous solution at 27.0' C. flowed through the outer jacket, 0.25 molal CoS04.7HzO solution a t 27.5' C. through the middle jacket, and distilled water at 29' C. through the inner jacket. Teflon-lined pumps (2-gallon per minute capacity) were used for circulation, and no metal parts were in contact with the solutions. Feed System. T h e known feed composition was obtained by bubbling helium (99.99% purity) through a saturator, 17, maintained a t 1' to 2' C. below room temperature. The required concentration was achieved by mixing this stream a t a measured flow rate with a stream of pure helium, also metered as noted in Figure 1. T h e feed stream could be sent to either the reservoir of the recirculation system or the analysis section for calibration of the chromatograph. Manometers, M , were installed at four locations to measure the pressure a t different stages of evacuation, feeding, and
P SllICO geldrier8 M V. At. v.R C
@
@
Manometer8 Vacuum manometer Vacuum pump Flow c o n t r o l l e r r Lamp v o l t a g e stabilizer Potentlometer for thermopile (9) Figure 1 .
Heating elements One of three (14,15,/6) filtering solution circulating systems.
@) @
0
Recycle reactor system
operating periods of the process. This turned out to be an important measurement because of the pulsating characteristics of the recirculating pump. Analytical Section. A Varian Aerograph, Model 1520B chromatograph, with two columns was used to determine the significant products and by-products of the reactions. Seven-port, tube-type, gas sampling valves allowed continuous flow, whether the valve was in the sampling or injection position. T h e reaction mixture was diverted from the recycle system by means of valve 5 to the chromatograph for a controlled period. A set of four solenoid valves, s, was installed, so that samples taken a t the same time could be analyzed a t different times in the two columns. The total volume of the batch system was 6620 ml., of which the photoreactor constituted 62.8 ml. The total volume does not include that of the feed system, which was isolated during operation by valves 18, 19, and 20. The apparatus worked satisfactorily, although two problems, introduced by the pressure fluctuations in the circulating pump, were always present: Traces of air could not be completely evacuated from the check valves, even after long pumping periods. If the oxygen concentration so introduced was high enough, photooxidation occurred. Pressure fluctuations a t the pump inlet could not be completely eliminated. Since the sampling valves were rather close to the pump suction, these fluctuations made it difficult to know the exact pressure inside the sample loops. The problem was solved because the exact initial concentration of acetone was known.
Operating Conditions. Feeding the recycle system took about 2 hours. Samples were taken for analysis a t hourly intervals and required about 30 minutes in each column. Total operating time for a run was from 3 to 7 hours, depending upon the concentration of acetone (range investigated, 4 to 17 mole yo) in the feed and the conversion desired. For acetone photolysis, and using a thermal conductivity detector, conversions of acetone as low as 0.1% could be observed. T h e operating temperature was constant at 97' =k 2' C., and the
Teflon
,
n11n nf - , . d
;/
Nylon Tube LD. O.D. (mm.) I 42 46 2 57 61 3 70 74 4 85 89
Figure 2.
Detail of reactor and filter-lamp systems
pressure level (average value in the reservoir) ranged from 850 to 890 mm. of Hg. Analyses made for different recycle rates indicated that the composition in the reservoir was uniform. The bypass line was used to feed the system initially and to bypass the reactor until the lamp output and temperatures were stabilized. Tubing in all the system was either glass or 316 stainless steel, '/d-inch in the recycle section and '/B-inch in the analytical section. Fittings and valves were either glass or 316 stainless steel. Flows were controlled with Moore differential pressure controllers and measured with ball flowmeters, except that in the feed system soap film meters were used to improve accuracy. Radiation Characteristics of System
T o make runs at constant light intensity, the transmission of the filter solutions must not change during radiation. This was checked by using the stabilization period of the new lamp VOL. 7
NO. 4 NOVEMBER 1 9 6 8
657
Table I .
Lamp Output, Filter Transmission, and Acetone Absorption ffAC,"
Liters/ FAX,
AA, A .
Einsteins/Sec.
>3500 3500-3400 3400-3300 3300-3200 3200-3100 3100-3000 3000-2900 2900-2800 2800-2700 2700-2600 2600-2500 2500-2400 2400-2300 2300-2200
