"' _— —— 7' "' . . , . . . , . ._. _. . ‘. .,,.‘. . A - ‘ .. "HUM... K ....- cw.--..quwmao-uvrrvn-v-«r-wrv —-.--—-—-'—-‘ ELECTRON-BEAM PROCESSING or PGWBERED MATERIAL ‘ Thesis for the Degree of M. S. MICHISAN STAY'E UNEVERSWY WAYRE H. CLIFFORD 1963 JflESli W3“ My 95.2.1394; a. . ~;’ rwggxo we? 1 v' W) ‘ ., . ‘ ‘ _ , . . ' 1 .' '7 n ‘ ‘ .. .I‘ \, - __ Fl) “4.... 5.) LC. 9" I Y .- - .. -.'L--- maids .1; ‘5? 1.)! #. *w‘ 3",” f'F- ‘ 1r ABSTRACT ELECTRON—BEAM PROCESSING OF POWDERED MATERIAL by Wayne H. Clifford As the use of ionizing radiation to induce chemical changes increases, it is natural that more efficient ways be sought to utilize such radiation. The characteristic of an electron beam is very high power delivery to a very small volume of material compared to other kinds of radiation. For irradiation of powdered materials on a conveyor belt—-the method now commonly used——this means that, for uniform irradiation, a thin sample must be used, resulting in much of the beam being absorbed in the belt. If a thick enough layer is irradiated to utilize the whole beam, there will be an extreme dose variation. The purpose of this research was to demonstrate how these problems might be overcome by the use of a fluidized bed. In this case a pulsed fluidized bed (since pulsing enables the fluidization of a greater variety of particle sizes and densities) was used. The bed was two feet high by one foot in diameter and was filled with approximately twenty—five pounds of methylcellulose powder. The fluidiz— ing gas was nitrogen from a forty-five gallon surge tank which was filled to ten or fifteen pounds per square inch Wayne H. Clifford gage pressure. A solenoid valve in the line from the surge tank to the fluidized bed was generally open for one second and then closed for one second. Fourteen different batch runs were made, of which half gave good product uniformity as determined by the viscosity of a two per cent aqueous solution at twenty degrees Celcius. The other runs involved poor fluidiza— tion and the presence of a "dead spot" in the bed. A continuous feed and drawoff mixing run (without radiation) was made where the bed was filled with previously irradiated methylcellulose and fresh material was fed. The results of this run corresponded to a perfectly backmixed model of the system. A continuous feed and drawoff irradiation run was made starting at approximately steady-state conditions. In spite of difficulties in maintaining the feed and draw— off mechanism a relatively uniform product was obtained. In an effort to estimate the efficiency of utiliza— tion of the electron beam, thin-layer samples of methyl- cellulose were irradiated to various doses. A comparison of the dose and resulting viscosity for the batch fluidized bed material and the thin layer material showed that less dose was required in the fluidized bed (half as much in some cases), to achieve the same viscosity as in the thin layer samples. ITxa‘W-nr - . _ ELECTRON-BEAM PROCESSING OF POWDERED MATERIAL By Wayne H. Clifford A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemical Engineering 1968 ACKNOWLEDGMENTS This work was performed under a research contract from the Dow Chemical Company. I would like to express my appreciation to Dr. William Bickert and Mr. Richard Patterson of the Michigan State University Agricultural Engineering Department for their assistance in the Operation of the Resonant Transformer both during the dosinetry runs and irradiation experiments. I would also like to thank Mr. George Atchison of the Dow Chemical Company for the use of the Van de Graaff Accelerator and aid in setting up for irradiation runs of it, and to Miss Irene Takahashi, also of the Dow Chemical Company, for the molecular weight distribution determinations. Finally, I would especially like to thank my Major Professor, Dr. Bruce W. Wilkinson, for his aid in all phases of this research. ii TABLE OF CONTENTS ACKNOWLEDGMENTS LIST OF TABLES LIST OF ILLUSTRATIONS LIST OF APPENDICES 1. INTRODUCTION. 2. DESCRIPTION OF THE PULSED FLUIDIZED BED REACTOR. . . . . . . . . 3. BATCH IRRADIATION EXPERIMENTS . A. CONTINUOUS FEED AND DRAWOFF EXPERIMENTS. 5. DISCUSSION AND CONCLUSIONS 5.1 Radiation Effects. . . . 5.2 Fluid Bed Uniformity Studie 6. RECOMMENDATIONS FOR FUTURE WORK APPENDICES. BIBLIOGRAPHY iii Page ii iv vii 1? 2A 28 28 33 40 A2 76 L417” Table 10. LIST OF TABLES Data for Batch Irradiation Runs Made on the MSU Resonant Transformer . Data for Batch Irradiation Runs Made with the Dow Van de Graaff Generator Molecular Weight Data for Ten Centipoise Starting Material . . . . Amounts of Methocel Product Obtained for the First Twenty Minutes of the Continuous Mixing Run. . . Data From the Continuous Feed and Drawoff Mixing Study Continuous Operation of the Pulsed Fluidized Bed Relative Dose Rates for the Three Methods of Irradiation of Methocel and Their Effectivenesses. . . . . . Dosage Calculations for the Thin-Layer Samples Based on One Square Centimeter of Dish Area. . . . . . . Dose—Viscosity Data for 4000 Centipoise Methocel Irradiated in Thin Layers Viscosity Resulting from Mixtures of Methocel iV Page 18 19 23 25 26 27 3O 6O 7O 73 Figure 10. ll. 12. 13. LIST OF ILLUSTRATIONS Depth-Dose Profile for 500 Kilovolt Electrons. . . . . . . Depth—Dose Profile for i 10 Percent Dose Variation Using 500 Kilovolt Electrons Photograph of Pulsed Fluidized Bed Reactor Set Up for Batch Processing. . Schematic Diagram of the Batch Pulsed Fluidized Bed Reactor. . . . Photograph of the Pulsed Fluidized Bed Reactor with Continuous Feed and Drawoff Schematic Diagram of the Pulsed Fluidized Bed Reactor with Continuous Feed and Drawoff Plot of the Continuous Feed and Drawoff Mixing Run Using the Perfectly Backmixed Model. . . . . Molecular Weight Distributions for Unirradiated Methocel HC 10 and Three Irradiated Samples . Photograph of the Ionization Chamber. Cutaway Diagram of the Ionization Chamber Circuit Schematic for the Ionization Chamber Surface Dose Rate in the Plane 35 Centimeters below the Beam Window at 1000 Kilovolts Peak and 1.0 Milliampere Beam Current for the Michigan State University Resonant Transformer Depth-Dose Curve at 35 Centimeters below the Beam Window of the Resonant Transformer on the Beam Axis at 1000 Kilovolts Peak Page 13 1A 36 38 AA 45 45 A8 50 Figure Page 1A. Plot of the Average Dose to a Given Depth of Material at 35 Centimeters below the Resonant Transformer Beam Window on the Beam Axis at 1000 Kilovolts Peak. . 51 15. The Variation of Dose Rate with Distance from the Vertical Beam Axis for the Dow Chemical Company Van de Graaff Generator in a plane 12 Inches below the Accelerator Window. . . . . . . . 57 16. Particle Size Distribution for Methocel. . 65 17. Photomicrograph of A000 Centipoise Methocel 6O HG, Premium Grade . . . . 66 18. Viscosity Resulting from Various Radiation Doses Applied to Thin Layer Samples of Methocel . . . . . . . . 71 19. Theoretical and Experimental Viscosities Resulting from Mixtures of Methocel . . 75 vi Appendix A. B. LIST OF APPENDICES Radiation Dose Calculations Physical Properties of Methocel. Methocel Viscosity Determination Thin—Layer Irradiation of Methocel. Viscosities of Mixtures of Methocel vii Page ’43 62 67 69 72 1. INTRODUCTION The application of high energy radiation to the processing of plastics, chemicals, medical products and food has been investigated extensively in the last two decades. This field of research was initiated by the availability of significant quantities of radioactive isotopes resulting from the preparation of nuclear weapons materials. Once the idea of desirable radiation— induced reactions had been raised, the field was broadened to include the effects of high energy particulate radia— tion (principally electrons) as produced by accelerators. The use of electron beams has been demonstrated with regard to film irradiation (preparation of cross-linked polyethy- lene film which has higher temperature stability); electri— cal insulation (cross-linked wire coating materials); irradiation of food materials for sprout—inhibition, pasteurization or sterilization; and several other fields. Similarly, radiation from radioisotopes has been demon— strated to be applicable for the catalysis of various chemical reactions as well as several of the processes listed above with regard to electron beam processing. In the design of irradiation experiments as well as the design of irradiation production facilities, one of the problems encountered in the technique to be used is l to attain uniform radiation exposure to the entire mass being processed. The problem applies to both electron beam and radioisotope radiation (generally gamma rays) but is particularly severe with electron beams. In the electron beam radiation processing of powdered materials, the common technique involves spread- ing the powder in a thin film on a suitable conveyor belt. The belt then carries the powder beneath the electron beam at a rate suitable to achieve the proper radiation dose. The permissible depth of material on the belt is controlled by the absorption characteristics of the material and the electron beam available. _For a monoenergetic beam of electrons, the typical variation of radiation dose with sample depth may be seen in Figure 1. If, as is the usual case, it is desired to expose the entire sample being processed to a relatively uniform radiation dose, a problem is encountered. A variation of radiation exposure or dose from 60 to 100% of the maximum is observed in the top layers of material (near the radiation source). Like- wise, an even wider variation in absorbed dose is encountered as the sample depth is increased beyond the peak dose. Thus, in order to achieve relatively uniform doses, it is necessary to limit the depth of material being irradiated, as shown in Figure 2. Obviously, this reduces the effi— ciency of radiation utilization since the radiation falling outside the chosen depth is wasted. Furthermore, if the limx relative dose, "/0 100 l. 75 50 25 n I 0 20 40 60 80 100 depth of absorber, mils at unit gravity Figure l. Depth—Dose Profile for 500 Kilovolt Electrons relative dose, % 100 F 4 U1 U1 C N U“ O \" / // .10.. beam // . 80 depth of absorber, mils at unit gravity Figure 2. Depth—Dose Profile for i 10 Percent Dose Variation Using 5 00 Kilovolt Electrons 100 depth of sample is not carefully controlled, product will be produced which has somewhat lower radiation dose than desired. Conversely, too thin a layer of sample reduces the efficiency of radiation processing by wasting a larger portion of the beam. Another solution applicable to particles which are small in relation to the depth of penetration of the electron beam is to use a very deep bed of the material and to keep the particles in constant motion in and out of the electron beam——a fluidized bed. T. 1 2 it has been found ’ ’3 that, for fine particles, there is a tendency for caking in a continuously fluidized bed. This tendency is overcome with pulsing: i.e., a period of high flow of the fluidizing gas followed by a period of no or low flow. A deep fluidized bed ensures that all of the beam Striking it will be absorbed in the material. Presumably, as the fluidized material moves into and out of the beam region, it experiences a radiation eXposure equal to the average echsure of the fluidized bed. Of course, it 18 AA possible that uniform irradiation is not achieved, but, R. P. Ievey, Jr., "Gas—Solids Contacting Method," United States Patent #3,l6A,AAO, January 5, 1965. 2L. Massimilla, G. Volpicelli, and G. Raso, ”A Study of Pulsing Gas Fluidization of Beds of Particles," Chemical Engineering Progress Symposium Series 62, 62, 63 (1966). 2) ’F. A. Zenz, and D. F. Othmer, Fluidization and Fluid—Particle Systems (New York: Reinhold Publishing Co., 1—...- rather, non—uniform irradiation followed by intimate mixing of the fluidized material is experienced. The purpose of the present work was to show the degree of uniformity of the final material and to relate the chemical effect to the average dose to the bed. A fluidized bed capable of pulsed flow of the fluid— izing gas was constructed. Several samples of Methocel, a particular methyl cellulose product of The Dow Chemical Company, were irradiated in it on a batch basis to demon- (/1 trate the degree of uniformity of the final product; The bed was then modified for continuous feed and drawoff. An unsteady-state run was made without the electron beam where unirradiated Methocel was continuously fed to a bed of previously irradiated Methocel. This test was used to demonstrate the applicability of modeling the fluidized bed as a perfectly stirred backmix reactor. Finally, a continuous feed and drawoff irradiation run was made start— U) ing near teady—state conditions. The uniformity of exposure of the Methocel was deter— mined by measuring the resulting 2% solution viscosity at 20.03C. The radiation absorbed causes chain scission in the methyl cellulose molecule and thus quantitatively lowers the molecular weight (and solution viscosity) of the material. In addition, the approximate molecular weight distribution of irradiated material was determined in an effort to ascertain the radiation exposure homogeneity. 2. DESCRIPTION OF THE PULSED FLUIDIZED BED REACTOR The fluidized bed reactor used in these eXperiments was built, tested for proper fluidization by visual examination and dynamic pressure measurements, and modi- fied until apparently satisfactory results were obtained. Then, after the batch experiments were completed, it was modified for continuous feed and drawoff. Figures 3 and A are a photograph and a schematic diagram of the batch fluidized bed reactor. The bed of powdered material was contained in a vertical thick—walled glass cylinder 12 inches in diameter and 2A inches high. The bed was supported at the bottom by a fine mesh stainless steel screen. The nitrogen dis— tributor under the screen was a conical section 6 inches high, 12 inches in diameter at the top, connected to a 1—inch by 3/A inch reducing tee at the bottom. A second ([1 tainless steel screen was placed inside the nitrogen dis— tributor about A inches below the bed support screen just before batch run number 10. One of the horizontal 3/A inch ends of the copper tee led to a 10 psi safety relief valve while the other led to the nitrogen supply. Eight inches from the tee on the supply side was a 3/A inch copper tee with its center connected to a pressure 7 0.0 O I .0. .0. 0: 0.9.0.... 7,. 3.2: : ,..r...:.m3.vfit.,.tzbai .eyv-.uL.-.. « 00 0 Q I» v. n O :e...~.. 0 ‘0 Que... 000 do... . o: 3.. . . . v '- ecoeov.o....sovs . a O . o e . .r».wu:. ooovt . mmmwufuiqbvfl u... DIIIIIIDDLD... 1111,1v _‘.O o . Photograph of Pulsed Fluidized Bed Reactor Set Up for Batch Processing Figure 3 nitrogen source safety valve L surge chamber vent ‘\ solenoid E (—— vats vent F"! F" ‘I . reactor | ' filte'r‘r) I beam , | I | window I ' L - J |_ -3 _ L -J fluid bed reactor 4 I presiure safety valve tap Figure A. Schematic Diagram of the Batch Pulsed Fluidized Bed Reactor lO transducer (which allowed the measurement of the pressure pulse—time characteristics). The line continued through the second tee for another 2 inches to a 90 degree elbow and 18 inches of vertical 3/A—inch rubber hose. A 1—inch 90 degree elbow led to one of the large solenoid valves and through it to the surge tank. The solenoid valve used was of type JJ from the Atkomatic Valve Company, and “1 had 1-inch fittings and a 1-inch port. The surge tank was a galvanized steel tank A feet “,~. high and 20 inches in diameter and had a capacity of about A5 gallons. A 30 psi safety valve was connected to the surge tank. A small solenoid valve served to control the nitrogen feed to the surge tank. This valve was type 8115 made by the Alco Valve Company. The feed to the solenoid was made through 3/A inch copper pipe, which led to 3/8 inch flexible copper tubing and then to the nitrogen regulators. The recorded pressures were read from the downstream gauge on the regulators during dynamic flow conditions. The bottom part of the filter housing, located just above the reactor, was a conical steel section 12 inches in diameter at the bottom, 2A inches in diameter at the top, and 6 inches high. The top part of the filter housing was a right circular cylinder 2A inches in diam- eter by 10 inches high, with a 13 inch wide cutout through the center. Two filters were supported inside the two ll remaining 10 inch high chambers. They were Cuno No. 520A5-l-Al—CA filters with a 10 micron maximum pass dimension. The reactor electron beam port was located in the center of the cutout part of the filter housing. It was 6 inches in diameter, covered by aluminum foil and a wire support screen. The latter was a square grid with two .0A2 inch diameter wires per inch. There was a pressure gauge on the outside of one filter chamber. The control system for the filters included a large solenoid valve exiting to the air, and a small solenoid valve connected to 3/A inch copper lines from the nitrogen supply for each filter. The large solenoid was open and the small one closed for 17A seconds for normal filtering followed by a flow reversal for 6 seconds to achieve filter blowback in the 180 second period. The blowback times for the two filters were 87 seconds apart. The bed gas pulse control system was on a variable period (normally 2 seconds was used) with the large solenoid valve between the surge tank and reactor Open and the snail feed valve closed for half the period. A single rubber gasket served to seal the filter housing and the glass cylinder,and two more gaskets, one on either side of the bed support screen,sealed the bottom of the reactor. A vibrator was used on the sloping part of the filter housing for some of the runs. There was never more than a half pound of Methocel on the sloping 12 part of the filter housing. An external air blast past the reactor window was provided to assure that overheating of the window did not take place. Care was taken that all parts of the equipment were grounded during operation. The reactor, surge tank, and timers were supported on a steel support frame which was on wheels for mobility. The bed sampling device was a 3 foot long plastic tube fitted with an internal piston. The piston was held in plaCEthile the tube wasthrust into the bed of Methocel through the opened reactor beam port. In this manner, it was possible to ”core" the bed and thus sample it verti- cally. It was observed that a "dead spot" existed at times in the bed at the bottom of the side nearest the surge tank. This was thought to be a result from the nitrogen flow which was horizontal and away from the surge tank for 10 inches before it reached the nitrogen distribution cone. Had there been room under the resonant transformer, the line leading to the distribution come would have been modified to give 6 or 8 inches of vertical flow before reaching the distribution cone. In this manner, it is believed that the bed would have been more uniformly fluidized. There were several changes in the reactor for con- tinuous processing (see Figures 5 and 6). The small solenoid valve feeding the surge tank was removed to l3 . ’40; -~ so a. a ~Y"; -q} u svnu. . ll diet t 5‘ . 94..¢00MH cw .& \0. RI . a. ‘sa. be! "r". .41 "o, 4 x. 1 . _.. '.:,. .‘A ‘0 .s. .- O , - ‘ i I Photograph of Pulsed Fluidized Bed Reactor with Continuous Feed and Drawoff Figure 5. 1A nitrogen .. source K vents\ A r———\ feed tank L—-r4l ( W i r:7l I I l I I. -J surge tank '- vibrator «\baffle plate fluid bed sample thinnble /’é= prOduCt product tank line Figure 6. Schematic Diagram of Pulsed Fluidized Bed Reactor with Continuous Feed and Drawoff l5 reduce the resistance in the line. Methocel feed to the reactor was made from a tank 35 inches high by 12 inches in diameter emptying into a 2—inch aluminum pipe with a 90 degree elbow. There was a thin aluminum dam covering all but the bottom half inch of this pipe. The pipe was sealed to the bottom of the feed tank and the side of the filter housing with large thin-walled rubber tubing which allowed the pipe to vibrate freely. A vibrator was used on the feed pipe to control the flow rate of powder. It was a Syntron Company VibraFlow feeder model FTO with an electric voltage controller model FCTO by the same company, with a 0 to 100 scale of settings. In the center of the reactor, a thin vertical galva- nized steel plate was installed in three hinged sections so it could be removed through the beam port. This plate came to within 3/A inch of the beam port and the bottom was 7—1/2 inches above the bed support screen. The Methocel thus had to travel down the sloping part of the filter housing, down one side of the reactor, under the dividing plate, and up the other side to the exit port. The powder exit port was a 1/2 inch hard copper tube which protruded inside the reactor to the bottom of the filter housing. Thus, unless it was overloaded and blocked off, the exit port maintained the bed height at the top of the glass cylinder. The line leading to the product tank was clear plastic tubing 1/2 inch inside diameter, 30 feet long (so the product l6 tank could be located outside the radiation room). The product sampling device was located just before the product tank. It was a hard c0pper 3/8 inch tee, with normal horizontal flow straight through to the product tank. The center of the tee led up 5 inches on a A5 degree angle to a movable steel cut-off plate. The plate had a 3/8 inch hole leading to a 90 degree elbow and back down at a A5 degree angle to a paper thimble. For sampling, the plate was moved so the holes matched for a few seconds and then moved back. An automotive air filter on top of the product tank was provided so there would be less resistance to flow out through the product tank than through the reactor filters. The nitrogen feed manifold was installed so that three tanks could be used at a time and independently removed and replaced. The nitrogen tanks were mounted on the stand for the product tank, with 30 feet of 3/A inch hose leading from the nitrogen manifold to the surge tank and filter blowback lines. 3. BATCH IRRADIATION EXPERIMENTS Fourteen batch irradiation runs were made, eight of these using the 1 Mev (peak) Michigan State University Resonant Transformer located in room 105 of the Agri— cultural Engineering Building, and six using the 2 Mev Van de Graaff Accelerator in the Radiochemistry Laboratory of the Dow Chemical Company in Midland. In order to eliminate variations between batches of Methocel, a large quantity of a single batch was acquired. Since the effect of radiation is most striking for high viscosity materials, it was decided to use A000 centipoise material. Thus, the Methocel used in all but two runs was 65 HG Standard Methocel of A000 centipoise nominal vis— cosity (lot number 060262—T). This was found to have an actual viscosity (as determined by the procedure described in Appendix C) of about A500 centipoise. One of the other two runs used granular Methocel of 5800 centipoise vis— cosity, and the other used 10 centipoise nominal viscosity MC Standard (lot number 02107A). Tables 1 and 2 summarize the data for the batch runs. The Methocel used in Runs 1 and 2 had been in the reactor for some time before the irradiation run. It was used to determine good fluidization conditions from the standpoint of visual observation. The fluidization was still good at 17 I." ‘I It." .I‘lli'llll . l“ I 18 m.m c.m o.m m.m Ammo m.HHc ca mm mm as cm 02 oz me mm.o ma m ma 3H2 DH mm ass osmfi cc» mew mcflxae tapas c . coasmNaeasdc . m a ca o m woos: :00 w o.m m: mm me oz cop mo posoos an s osma cmem cmca czma coo: m .oom m2 mm me o: 0: mo sm.c m m m 2H m.mm sac cmo cmc mmc mmo coo: mm x: mm m: 02 cm mm sm.c m.s m mm .cwc 3H H.mm m com :zm :mm mmm mmm mcflxfis popes m Hm.c m :ofipwmflcflsam mm .cmo mm m: me oz 09 and mo pozoocd mooscfipcoo : owe: cmca mmm no: em: Ho: coo: mm .cmc mm w: x9 om 0: QB 02 mo mo.c ca m m 2H c.cm .cw sac mmo mac mmc mmc ozo H .oc mm.o m m z m o mm m: we cm oz 09 can no pounced cwma cmHH cmHH coma csHH csHH ccc: : .cmo mm m2 we cm oz 05 o: mc mm.o m m H DH mm cop who 2H :oHpHmOQ was some Hwfiuopms mcflusmpm mcmawon .CHE “mafia .oom .oEHp mums zufimoomfi> soapsaom msomsom Rm .cmo on cofipwficwppH macho omasm new omoo owmpm>< .0: cam poEpogmcmpB uneconom cm: ecu so mom: mcsm coHpmflchpH copmm new moon H mqm COHusHom mzomsom H 6Home on :oHpmHomLLH oHozo mmHsm cam owoo ommsm>< .oc 23m Loompococ gammgo on cm> zoo esp cpHs coma mczm COHumHUmLLH Lopwm pom mama m m4m<9 20 the end of Run 1. There was no buildup of Methocel on the reactor walls, the slanting filter housing walls, or the underside of the beam window. At the end of Run 2, the fluidization was observed for 1 minute before sampling, and found to be less than the best. Runs 3 and A were segmented into two 5 minute periods. In between the two periods the fluidization was observed to be poor. In an effort to remedy this, a large pulse of nitrogen was sent through the bed by allowing the surge tank to fill for about 15 seconds with the timer off. The extra blast of nitrogen thus produced helped to break up the bed and improve the fluidization. The nitrogen pressure was 30 psig for the runs as compared to 20 psig for runs 1 and 2. Run A was made on the same Methocel after fluidizing it for 10 minutes. The nitrogen pressure was 15 psig for this case. Run 5 was segmented into 3 periods of 2, 3, and 2—1/2 minutes with examination of the reactor in between. The nitrogen pressure was 30 psig. After 2 minutes the fluidization was good, but after 5 minutes some stagnation was evident, so a large pulse was used to break it up. Run 6, with the slower pulse time, showed less bed expansion than the previous runs. There was evidence of dead areas in the bed and some buildup of Methocel on the slanting walls of the filter housing. At the end of the run (after sampling) the bed was mixed with pulsing for several minutes, but the fluidization was poor. This was 21 perhaps due to lower tank nitrogen pressure since the flow of nitrogen to the system was controlled in part by critical flow through the pressure regulator on the cylinder. Run 7 was then made on the mixed bed. The nitrogen pressure used was 30 psig for this continuous run. The fluidization was still satisfactory at the end. Run 8 was somewhat unfortunate in that the air a blower used to cool the beam window was not used. The window thus overheated and failed. Run 9 was essentially a repeat of 8 with the blower on. The nitrogen pressure ; was 30 psig. The irradiation was stopped in the middle of the run so the reactor could be examined. The fluidiza— tion was found to be poor, so a large pulse was sent through the bed. A significant discoloration of the glass reactor wall was observed. Thus, a part of the radiation was missing the bed (the bed was lower than expected). It was theorized that the poor fluidization might be due to low tank pressure since both runs used the same tanks. Runs 10 and 11 used fresh nitrogen tanks with the nitrogen pressure again set at 30 psig. The fluidization appeared good throughout both runs, although some Methocel collected on the slanted side of the filter housing. Run 12 was a duplicate of the above runs, except that no samples were taken in the middle. Run 13 was made using granular Methocel. With each pulse the whole bed moved up and down with relatively little 22 mixing. The reactor and the method of operating it were not optimized for the granular material. Run 1A was made on 10 centipoise feed to provide material for molecular weight distribution testing at the Dow Chemical Company. Similar material was also irradiated in a thin layer so that there was only a small dose varia— tion similar to methods used in Appendix D. The molecular weight distributions and average data are given in Table 3. 23 .mCMQEoo HMOHEmco 30o map no Hammcmxme .9 .H mo coxmu who; mums mmmee* 000.0H oom.HH oom.mH ooo.cH oom.mH ooom.OH *ze mmmtm>m hopes: :.c: m.cH :.mH m.HH T-M ccccmH uia :.sm c.mm :.mm c.mm mm“ Doom: Mm m.mH o.:m m.:m c.mm Tad ooo:H m.- c.: m.cH THH c.MH cam cccm w o 3.: m.m m.m .H H.: m.m m.HH o.m :.w m.m omHongcoo ampHmoomfiH awn ”Mane... 2. .22. 2.....- H .303 H .30Q «con cmmHsm aHome couwcHsHm «nmmmH pewsmmohp LommH xOHcB pommH chB com Hm :Hce oH Emm HmHsopmz wchnwpm moo cH now open pcwHoz ansooHoz m mqm¢9 A. CONTINUOUS FEED AND DRAWOFF EXPERIMENTS There were two continuous feed and drawoff experi— ments. The first did not involve irradiation, but simply mixing previously irradiated Methocel with an unirra- diated batch. The object in the second was the steady state continuous processing of Methocel. Thus, the bed was filled with irradiated material and fed unirradiated Methocel while under irradiation. For the mixing run, the bed was charged with 21 pounds of previously irradiated Methocel and the feeder was filled with the unirradiated 65 HG A000. The nitrogen pressure was 10 psig, and the feeder was set at 60. Table A shows the weight of product with time and the average drawoff rate for the first 20 minutes operation. At 16-1/2 minutes into the run the product line became plugged, so the run was temporarily interrupted. Next, the feeder was set at 80, and the operation continued for 15 minutes, at the end of which a cumulative total of 12.06 pounds of product had been collected (for 35 minutes operation). Then the nitrogen pressure was gradually increased to 15 psig with the feeder off to partially empty the bed. In this way an additional 2.69 pounds of Methocel were emptied from the bed to the product 2A 25 TABLE A Amounts of Methocel Product Obtained for the First 20 Minutes of the Continuous Mixing Run Cumulative Run Total Average Time, pounds of Exit Flow Minutes Methocel Rate, lb/min 5 1.5 0.30 ’ 10 2.9A 0.29 15 A.8l 0.32 20 6.81 0.3A tank. Next, the feeder was set at 100 and started, and the mixing continued for 30 minutes more, during which 13.06 pounds of additional product were collected. In the whole run, about 28 pounds of product were collected. Thus, the bed was changed 1.A times. The viscosity data, along with the values derived from the viscosity which were used in the discussion, may be seen in Table 5. The continuous feed and drawoff irradiation run started with 20 pounds of 770 centipoise Methocel which had been irradiated previously. The nitrogen pressure was 15 psig and the feeder was set at 100. The Resonant Trans— former beam out current was set at 0.1 milliamperes instead of 1.0 as had been used before, because the feed and drawoff 26 TABLE 5 Data From the Continuous Feed and Drawoff Mixing Study Time, min. up, cp ué/B log (pg/8 - ui/B) O 323 2.059 0.0035 2 316 2.053 0.0060 A 395 2.111 - 0.0195 6 A20 2.128 — 0.0273 8 A80 2.16A - 0.0AA3 10 518 2.185 — 0.05A0 12 58A 2.217, - 0.0706 1A 659 2.251 - 0.0883 16* { 678 2.259 - 0.0926 6A3 2.2AA — 0.08A6 18* { 77A 2.297 - 0.12u9 729 2.280 - 0.10A0 20* { 836 2.319 - 0.1261 801 2.306 - 0.1192 25 938 2.352 - 0.1A57 30 1157 2.A2l - 0.1898 35 127A 2.AAA — 0.1986 A0 1A99 2.A9A — 0.2A18 A5 1636 2.522 - 0.2636 50 1866 2.56A — 0.298A 55 2287 2.630 - 0.3595 60 216A 2.612 - 0.3A20 *Two viscosimeters were used on these samples. 27 rates were so low. The feed was 65 HG A000, as usual. Table 6 shows the product rates and exit Viscosities as a function of time during the run. During this run, the bed level in the reactor dropped somewhat. This caused the drawoff rate to be higher than the feed rate. It is thought that the feed rate to the bed was probably continuous at about 0.12 pounds per minute. This was much smaller than planned. TABLE 6 Continuous Operation of the Pulsed Fluidized Bed Rate of Run time, Exit viscosity, Total drawoff, drawoff, min. centipoise lbs. 1b/min 770 583 10 550 15 62A A.5 0.30 20 667 25 639 30 623 7.06 0.17 35 678 A0 6A5 A5 662 8.69 0.11 50 838 55 802 60 691 61.5 10.81- 0.13 5. DISCUSSION AND CONCLUSIONS 5.1. Radiation Effects The mechanism of breakdown of methyl cellulose and cellulose when irradiated in the dry state has been shown to involve a free radical chain reaction where the rate of degradation is proportional to the concentration of free radicals.1’2 The free radicals are destroyed by two mechanisms. Bimolecular termination involves the combina— tion of two free radicals to make a stable molecule. Alternatively, free radical scavengers such as molecular oxygen (meaning 02) may result in the destruction of radicals. The faCt that free radical loss is nearly second order to cellulose2 shows that the bimolecular termination is the most important destruction mechanism. It was also found that free radicals could exist for several days in dry irradiated cellulose in an inert gas, and that their destruction was about 10 times as fast in air.2 In thin layer electron beam processing where a single particle sees the beam continuously while it is being 1F. A. Blouin, et al., "The Effect of Gamma Radia- tion on the Chemical Properties of Methyl Cellulose," Textile Research Journal 33, 153-158, February, 196A. 2R. E. Florin, and L. A. Wall, "Electron spin Resonance of Gamma—Irradiated Cellulose," Journal of Polymer Science Part A, I, 1163—1173 (1963). 28 29 processed, the concentration of free radicals builds quickly to a steady state concentration where the rate of production due to the radiation is equal to the rate of termination. For the same total dose, then, a lower dose rate generally results in a steady state free radical concentration which is not as much lower as the dose rate. Thus, the resulting radiation effect is greater. In the pulsed fluidized bed, a particle "sees" the beam for a short time before moving away. Thus, it gets irradiated in short pulses. If it spent the same amount of time in the beam as the thin layer particle, it would experience a much more effective eXposure to free radicals because it would see the concentration while they were dying out many times instead of just once. Table 7 is a summary of the four best batch irradia— tion runs showing the dose, resulting viscosity, and thin layer dose for the same viscosity taken from Figure 18. It shows, at least in a qualitative way, the effect of using the Van de Graaff generator versus the Resonant Transformer. As may be seen, the Van de Graaff beam Spot is smaller and the resultant average dose rate higher. This means that an individual particle will "see" about three times as much dose rate while in the beam. If, in fact, the bimolecular termination process does predominate, this higher dose rate should more nearly approximate the thin layer work and the amount of radia- tion required to achieve a given degree of degradation 30 c.H m.H w.m c.m m.m con: omoc ow onoo noeeH sees no onoom I: c mo mo mo monocH .xomadm ohmsom Aomoo EdemeE go How unmoH pm wcH Iso>HHoc some mo moh< no.0 mm.o so.H Ho.H sm.o moonemoz ampHmoomH> mean one now omoo mohmchHQB ooOH oomm omo meo owHH onnoonoeoo .mpHmoomH> pampHSmom no.0 oH.o am.o om.o mm.o noonemoz fiomoo Hopoo owmno>< :m.m omH.o oso.o a:o.o o:o.o oozene non moonemoz “moms onoo ommho>< m.mm oom o.oo o.oo o.oo oon\noene Aooonc open onoc oommgsm nosoH ease oH m m H .oz com .9 .m cheapo mo cm> - soEgo mews chOmo oopso COH ma m cm: zoo m B p m cm: m .p .o m .HO mommoco>Hpoommm sHone com Hooonpoz coHpmHUMLmH mo moocpoz oopce one how mopom omom m mqm¢e o>eooHom 31 should be higher than with the resonant transformer results. A comparison of the last row on Table 7 would indicate just such an effect. The ratio of thin layer radiation dose required to achieve an equivalent degrada- tion of the solution viscosity is considerably higher with the resonant transformer fluid bed experiments than with the Dow fluid bed experiments which are, in turn roughly equivalent to the thin layer experiments. This effect may be due to any of several causes: First is the instantaneous radiation intensity effect. The fact that the dose rate with the fluid bed tests was higher than in the thin film work would imply that the total dose requirements would be higher with these cases as compared to the thin film work. In fact, however, the converse was true with the fluid bed work always using less radiation. Further, since the dose rate with the Van de Graaff is the highest of the three cases, it should be the poorest with regard to radiation efficiency. Again, this was not observed. The second is the pulse or "shutter" effect. In the case of the fluid bed work, the particle of Methocel is eXposed to "pulses" of radiation as the result of its movement into and out of the beam. In such a case, the average free radical concentration would be lower and thus the radiation would be used more efficiently. This is consistent with the data reported. The effect is apparently 32 much less pronounced with the Van de Graaff and this is perhaps due to the effect of higher radiation intensity with this machine. The third is oxygen radical scavenging. The pulsed bed work was conducted in an atmosphere of nitrogen whereas the thin layer work was done in air. It is possible that sufficient oxygen was present in the thin r1 layer work to destroy a significant amount‘of the free radicals. If such were the case, it is difficult to ~ rationalize why the effect was not observed with the Van de Graaff runs. Finally, one might suspect the dosimetry calcula- tions used for Table 7. Below 10 centimeters from the Resonant Transformer beam window the surface dose rate on the beam axis follows the inverse square decrease law.1 The axial surface dose rate for the bed, at 35 centimeters below the beam window, was 96.0 kilorads per second, while the like value for the thin layer runs, at A7 centimeters, was 55.5 kilorads per second, in very good agreement with the inverse square law. On the other hand, inherent in the calculations involving the Van de Graaff was the assumption that all of the radiation emitted from the accelerator was absorbed in the bed. The error in this assumption is demonstrated (at least qualitatively) by the 1R. 0. Nicholas, "The Application of High-Energy Electrons to Some Grain-Infesting Pests" (unpublished Ph.D. thesis, Department of Agricultural Engineering, Michigan State University, 1958). 33 discoloration observed in the reactor glass walls. Thus it is likely that, if anything, the radiation dose reported for these runs is too high and this would tend to bring the results more nearly into agreement with the resonant transformer results. In conclusion, then, one might state that the radia- tion required to achieve a given degree of product degrada— tion is apparently less with the nitrogen pulsed fluid bed operation than with thin film irradiation in air. The reason for the lower dose requirements may be due either to lower effective radiation intensity (and free radical concentration) with the fluid bed or to the oxygen scaveng— ing in the thin film work. 5.2. Fluid Bed Uniformity Studies Results have been obtained to indicate that it is possible to achieve good product uniformity within a fluidized bed. Within the experimental apparatus used, a localized "dead spot" of material was often observed. The tendency of the bed toward having a dead spot at the bottom on the side of the reactor next to the surge tank was so strong that only two samples were really necessary to determine the uniformity of the product of a run-—one from that position and the other from the center of the bed. This dead spot is probably not critical in constant feed and drawoff work since it will only decrease the effec— tive size of the reactor. The significance of this would 3A be that a normal free particle will spend a larger fraction of the time in the beam, on the average, since it cannot use the dead volume. The simplest model of the continuous feed and drawoff pulsed fluidized bed is a perfect backmix reactor, where the reaction is the mixing of Methocel samples of dif- ferent Viscosities. The composition of the misture leav- ing the reactor, for pure component A in the reactor at first and pure component B entering may be found from the model by making a differential balance on component B in the reactor: (Input) - (Output) Rate of Accumulation F — F ' XB d(W - XB) /dt where F is the constant feed and drawoff rate, W is the constant amount of material in the bed, and KB is the weight fraction of component B in the bed. Rearranging and integrating gives: fXB ‘1 dXB = ft dt o F (1 - XB) o XB = l - e-(E/W)t From the Appendix one finds 35 1/8 1/8 _ 1/8 H X BuB p — AHA +X and since _ _ -(F/W)t XA — 1 — XB - e u1/8 = e-(F/W)t 1/8 1/8 1/8 -(F/W)t p HA + uB e “6/8 _ uJig/8 = (“é/8 _ “1/8) e-(F/W)t 1/8 1/8 _ 1/8 1/8 .A3A F 1/8 1/8 so a plot of log10 (u ) versus t should be a B — up straight line of Slope -.A3A F/W. Table 5 gives the viscosity data from the continuous run along with the calculations for the theoretical straight line plot, which is shown in Figure 7. The slope from the plot was —0.006l/minute, and, with the input and output at about .3 lb/min, and 21 pounds in the bed, -.A3A F/W is calculated to be -.0062. The continuous feed and drawoff mixing run shows that the perfect backmix reactor model is close to the actual physical situation, as evidenced by the straightness of the plot in Figure 7. The continuous radiation run was not absolutely necessary, since the method had been shown 12 18 log10(“13/ '“P/) .40 36 Figure 7. Plot of the Continuous Feed and Drawoff Mixing Run Using the Perfectly Backmixed Model 37 in the mixing run. Of course, it is possible that the feeder difficulties were due somehow to the radiation, but the problem was not in the fluidized bed. The best fluidization conditions, which were found by visual observation and demonstrated in batch irradia- tion runs and the continuous mixing run, were a 2 second goulsing period and 15 psig nitrogen pressure, with 20 to 225 pounds of Methocel in the reactor. This provides a very eefficient utilization of the radiation because each particle rruoves in and out of the beam to extend the utilization of ifxree radicals while they are dying out. The question of whether the material is actually ruseceiving uniform irradiation or just being well mixed is r1<:>t as easily resolved, however. A review of Table 3 would Zirfldicate that the thin layer irradiation material has ecxzperienced a marked decrease in the number average molecu— lEagr weight whereas the fluid bed irradiation material Sflcows little or no change. This might be explained on the béissis that the thin layer material has experienced a rather CCDrusistent destruction of the molecules due to the irradia- ti—C3n. An analysis of the molecular weight distributions 355 shown in Figure 8 indicates very little difference in dj-Estribution between the fluid bed and thin film samples. If? anything, it would appear that the fluid bed samples exlperienced a somewhat less uniform exposure than the thin film materials since the former shows a slightly larger :2.-'1‘ molecular weight 38 150, 000 T r l l I r l I I l l 100,000 7 7 80,000 - ‘ 60,000 ' J 40, 000 '- H 30,000 E a 20,000 _' - 15,000 ' - blank 10, 000 - J 8,000 7 ‘ 6, 000 P \ - C:B.A 4’ 000 l l 1 l l 1 l l 4 J 1 95 90 80 'N) 60 50 40 30 20 10 53 2 l \venflu percentsrnalknrthan Figure 8. Molecular Weight Distributions for Unirradiated Methocel HC 10 and Three Irradiated Samples. E and C were irradiated in the fluidized ted at RC and Fr, while sample A was irradiated in a thin layer. 39 fraction of low molecular weight material. Further, the sample taken from the relatively stagnant (BF) region of the bed appears to be slightly poorer than the other fluid bed samples. The difference between the samples is small, however, compared to the deviation from the unirradiated sample. An alternative technique of analyzing the uniformity of irradiation revolved about the total dose required to give a certain degree of molecular weight degradation. If the bed had resulted from a physical mixture of high and low viscosity materials, the total dose requirements would have been higher than if a uniformly average viscosity had been used. Unfortunately, the diScrepancy between fluid bed and batch samples with regard to radiation require- ments (see above) has masked this effect. Therefore, although it has not been possible to prove that the bed had, in fact, achieved uniform irradia— tion, it would appear that the product produced (as characterized by the molecular weight distributions) by the two irradiation schemes is very similar. Thus, the irradiation by fluid bed does not give a grossly nonuniform irradiation. 6. RECOMMENDATIONS FOR FUTURE WORK Future work with batch irradiation in the pulsed fluidized bed should be directed to the following areas: 1. The "dead" Spot should be eliminated. This can perhaps be done by redirecting the nitrogen stream as it enters the distribution cone. 2. The apparently lower radiation dose requirement for fluid bed work needs to be confirmed. Batch thin film irradiations in a nitrogen atmosphere would be helpful as would the irradiation of batch samples beneath a rotating shutter to give a pulsed effect. 3. A more rigorous technique of comparing the radiation homogeneity is required. It is possible that better molecular weight determinations would be helpful and, further, the results of number 2 above would permit the application of the mixing technique discussed in the results. A. Fluidization conditions need further optimiza— tion. The optimum pulse size and frequency might improve the fluidization conditions while minimizing the gas requirements. Future work on continuous pulsed fluidized bed irradiation must start with finding a more reliable feed and drawoff system. Once this is done the dose uniformity A0 A1 should be investigated, either with molecular weight dis— tribution studies on a material such as Methocel or per— haps with a microbiological technique where bacteria or spores can be mixed in with the material to be processed. The average dose means nothing to a single bacterium if it has not been hit with radiation, and, if uniformity of microbial kill can be shown this method of processing could be applied to food materials. It is also suggested that a smaller fluid bed be made (say, 6 inches in diameter by 18 inches high). With this, all the handling difficulties would be lessened as would be the amount of nitrogen used. This would also open the possibility of Operating closer to the generator beam window and putting in a length of vertical pipe just below the nitrogen distribution cone. It would also decrease the amount of material needed for a run. This might be important for continuous runs. APPENDICES A2 APPENDIX A RADIATION DOSE CALCULATIONS Three methods of determining high radiation doses (the Fricke Ferrous Sulfate dosimeter, a Faraday cup, and a small ionization chamber) were tested with the conclusion that the ionization chamber is of most value. A photo— graph of the ionization chamber used is shown in-Figure 9, and a cut away diagram is shown in Figure 10. The start- ing point of the design was that of Lawton,l with the addition of a thermocouple well. The ionization chamber consists of two 5 inch by 1/8 inch thick aluminum plates, with a 1 inch hole in their centers. As can be seen in the diagram, the hole is not cut off straight, but the last sixteenth of an inch slants out fOr a quarter inch, so the hole is l—l/2 inches in diameter at the top or bottom. This is to prevent dis- tortion of the beam due to scattering off the plate or the attraction the grounded plate would have for electrons. The two quartz disks were .OAA inches thick, with a one inch hole and big enough to fit snugly in position. A notch was cut in the top of the lower disk to permit the 1E. J. Lawton, and J. S. Balwit, "Ionization Chambers for Measuring Cathode—Ray Dose Rate," General Electric Report No. RL—6l8, November, 1951. A3 ,1:_':;'»¢ a {391.4 1 ‘wt: :fi 0* I. a a v I. {In Photograph of Ionization Chamber Figure 9 thermocoup well I. :2. T :— , [ll 7"“ ‘1 "€an 1 y elec‘tj'odes 1 electrode quartz thermocouple lead trough insulators lead trough Figure 10. Cutaway Diagram of the Ionization Chamber electron beam $111 —450 volts r'-""-'""l I - -L l - - I L _______ .1 \ionization chamber Keithly 210 electrometer Figure 11. Circuit Sche will with shunt matic for the Ionization Chamber A6 electrical contact with the center electrode. Contact‘ with the outer electrodes was made through the aluminum plates, since they were in contact and grounded. Figure 11 shows the circuit schematic used for radiation measurements. In the calculation of the cell constant, the conver- sion factor which relates the current measured to the dose in the chamber, use is made of the fact that it takes 3A.O electron volts of energy absorbed in the chamber. The chamber constant, c, is given as follows: (3A.0 ev )( electron_ )(l.602x10-12 erg) GleCtrO“ 1.602-410"19 coulomb eV (10.6 coulomb) _ 3A0 erg sec. p—amp _ sec. u—amp To get c in terms of rads, the mass of air in the chamber must be calculated. Under these conditions air is essentially a perfect gas, so the perfect gas law will be used. 28.8g/g mole ) q 2113 in’—mmHg/oR/g mole 9.A2x10_u grams, 3 ll +—ll"o and since A7 1.0 rad = 100 ergs/gram then 3A0 erg» , ( 10“ T)(gram rad) C = sec. uamp 9.A2 . P 100 ergs ‘ T rad C I P (3610) sec. uamp where T is in degrees Rankine and P is in millimeters of mercury. Under typical conditions of 90°F (5A9.7°R) and 7A0 millimeters pressure, the constant is 2680 rads per second per microampere. Irradiation of Methocel was carried out in three different ways. The first was in thin layers with the Michigan State University Resonant Transformer; the second was in the fluidized bed with the same source; and the last was at the Dow Chemical Company in Midland, Michigan, with the Van de Graaff generator. For the fluidization runs at Michigan State University the ionization chamber was supported at various points in a plane 35 centimeters below the accelerator window. This corresponded approxi- mately to the tOp of the bed during fluidization. Figure 12 shows the results of that work, along with the fitted parabola used to approximate the curve for integration. The parabola is D = 96.0 — .98Ar2 kilorads per second when r is in inches. To get the average dose, one simply integrates the dose over the area of the bed and divides by the area. A8 100 surface dose, kilorads/second 6O east-west n 2 north- south so i 2 ‘ : i 6 4 2 d 2 4 6 north or east south or west inches from vertical beam axis Figure 12. Surface Dose Rate in the Plane 35 Centimeters below the Beam Window at 1000 Kilovolts Peak and 1.0 Milliampere Beam Current for the Michigan State University Resonant Transformer _ 1‘ IF- A, ,. — 1 6 1 6 2 D = A fOD 2nr dr = §E— f0 2nr (96.0—.98Ar ) dr 2 L16 2 _ 1 96.0r _ .98Ar = 96.0(36) _ .98u(36) ’I8 2 A O 36 72 0| n 78.3 Krads/second at the surface at l Mev, 100 uamp beam. To go from the surface dose to the average total dose, use is made of a depth—dose curve, such as the one shown in Figure 10. This curve was made by placing aluminum absorbers of various thicknesses over the ioniza- tion chamber and measuring the resulting dose. From this one can determine the average dose to any depth of graphically integrating the cruve. This is shown in Figure 1A. From Figure 13 one observes that the whole beam will be absorbed in the first 0.3 grams per square centimeter of material, and from Figure 1A one sees that the average dose to that depth is 66.A% of the surface dose. Since the area considered here is a circle one foot in diameter (or 730 square centimeters) the top 219 grams of material will receive an average of 66.A% of the sur— face dose. Then the total beam is given by: Krads sec U3 || (dose)(mass) = 78.3 (.66A)(2l9 grams) 6 ggam-rads second 11.38 x 10 relative dose rate 50 120 80- 60- 40' 20. l i l 1 I 1 l 0 50 100 150 200 250 300 depfllinlmfilfigrarns per square centhneter Figure 13. Depth-Dose Curve at 35 Centimeters below the Beam Window of the Resonant Transformer on the Beam Axis at 1000 Kilovolts Peak 51 120 100 80 60 40 relative ave rage dose 20 ul- L l i ' + 0 100 200 300 400 500 depth in milligrams per square centimeter Figure 1A. Plot of the Average Dose to a Given Depth of Material.at35 Centimeters below the Resonant Transformer Beam Window on the Beam Axes at 1000 Kilovolts Peak 52 The beam window on tOp of the fluidized bed was found to have a thickness of A milligrams per square centimeter, so one may calculate the energy absorbed in it in a like manner. The weight would be 3 grams, and from Figure 1A the average dose in the first A milligrams is 101.5% of the surface dose, so the beam absorbed in the fluidized bed window is: Krads sec U1 ll (dose)(mass) = 78.3 (1.015)(3) .2A x 106 gram—rads/sec. As will be shown later, the wire support grid lets 83.9 per cent of the beam striking it through. Then the Methocel will receive 11.1A times 0.839, or 9.3A x (10)6 gram rads per second. As_an example, the average dose for a 5 minute irradiation of a 25 pound sample of Methocel under these conditions would give: 6 gerads (300 sec)( 1 )( l6 ) sec 5 min 25 lb A53.6 g C II I9.3Ax10 2A7,000 rads in 5 minutes The total power generated by the Resonant Transformer, converted to dose rate units, may be calculated from the root mean square voltage and beam out current: 107 ergs)(gram—rad) 3 -3 (7O7X10 v)(10 amp)(watt.sec 100 ergs 7 gram rads sec = 7.07x10 This compares with 9.31 x 106 gram rads per second net power delivered to the fluidized bed, as calculated above, which is 13.2% of the total power. To determine if this is reasonable, the losses from the initial beam will be calculated. Just before the beam leaves the accelerator it is a rectified sine wave.1 The losses include absorption in the .0075 inch titanium accelerator beam window, 35 centi— meters of air, the iron wire support screen for the reactor window, and the aluminum foil reactor window. In addition, the beam tends to spread out after it leaves the accelera- tor, so the fraction lost in the reactor walls must be estimated. Since titanium has a density of A.5 grams per cubic centimeter, the .0075 inch accelerator window has a face density of 85.8 milligrams per square centimeter. The average dE/dx for electrons in this energy range is 1.6 Mev square centimeter per gram. The loss in the titanium would then be (85.8 mg/cm2) : (1.6 Mev cm2/g) or .137 Mev. 1J. A. Knowlton, G. R. Mahn, and J. W. Ranftl, "The Resonant Transformer: A Source of High-Energy Electrons," Nucleonics ll—ll 6A—65, 1953. 5A The beam out current is that which actually leaves the accelerator window, but the root mean square voltage will now be 0.570 Mev instead of .707 Mev. At 70°F and 7A0 millimeters of mercury pressure the density of air is 1.16 milligrams per cubic centimeter, so the 35 centimeters of air represents A0.7 milligrams per square centimeter thickness. Since air is a slightly better absorber of electrons than aluminum (by about 19%) and the depth-dose curve was made for aluminum, this is equivalent to A8.3 milligrams per square centimeter of aluminum. The reactor window is aluminum foil A milligrams per square centimeter. The average dose to this depth is found to be 1.08 Do’ so the air and window together absorb 56.5 Do' The average dose to 300 milligrams per square centimeter is 0.66A Do’ so the total available is 199.2 Do' Then the air and window take out 28.A% of the beam, so the fraction transmitted is 0.716. The iron wire in the reactor window support screen is .0A2 inches thick, with a distance of 0.A58 inches between adjacent strands. It was assumed that the wire, although round, absorbed all the beam which struck it, because the amount of beam transmitted through the thin parts is offset by the amount absorbed by the solder at the corners and along some parts of the wire. The ratio of the area taken up by the wire to the total area is .161, so the screen will transmit 0.839 of the beam striking it. 55 It has been reported1 that in a horizontal plane the variation of dose with radial distance from the center of the beam is described by the error function: D = D ~ exp(—r2/2a2) where r is the radial distance from the vertical axis of the beam, D0 is the dose at the vertical axis, and a is the equivalent of the statistical standard deviation. A weighted average of the data from Figure 12 was used to compute a, which was 6.38 inches. To find the total beam, the product of the differential area, 2hr dr, and the dose at r is integrated from r = 0 to infinity D = f 2hr . DC ° exp(—r2/2a2) dr = 20a2 DC For the amount of the beam falling inside the reactor, the integration is carried out from r =-0 to 6 inches. 6 2 2 2 D1 = f 20? ° DC exp(—r /2a ) dr = 2wa DO 0 2 2 (l _ .‘5 /2(6.38) ) 1R. C. Nicholas, "The Application of High—Energy Electrons to Some Grain-Infesting Pests" (Ph.D. Thesis, Department of Agricultural Engineering, Michigan State University, 1958). 56 U II 2 _ 2wa Dc(l - .6A2) — .358 Dt Then the fraction of the beam which is absorbed in the Methocel is 0.358 of that which enters the reactor. Then the net beam striking the Methocel will be the root mean square voltage of the electrons leaving the accelerator multiplied by the beam out current and the fractions transmitted by air and the aluminum reactor window, the wire support screen, and the fraction which falls inside the reactor. (270x103 v)(10’3 amp)(.7l6)(.839)(.358) = 122 watts Converted to dose rate units, this is 12.2 x 106 gram rads per second as compared with 9.31 x 106 actually measured by the ion chamber. Figure 15 (which was provided by the Dow Chemical Company) was used to calculate the dose to the bed for the irradiation work performed on the Dow Van de Graaff generator. First, the curve was integrated numerically in the same way as was done previously to determine the total power delivered to the plane of the fluidized bed. 00 P = f 2 rvp(r) dr = Z 2nr p(r) Ar 0 the relative dose rate C, P/P 57 l l. O «I. . O O 0.8-r . $ 5 5' o. 6"- l O O 0.4—? 0.24 I O D 0 ~ ~ ' 2 i .L . l l . 10 8 6 4 2 O 2 4 6 8 10 inches from center of beam Figure 15. The Variation of Dose Rate with Distance from the Vertical Beam Axis for the Dow Chemical Company Van de Graaff Generator in a plane 12 Inches below the Accelerator Window 58 where p(r) is pC multiplied by the value found in Figure 5 15. p was 3.13 x 10_ watts/cm2, according to information c from the Dow Chemical Company, at 250 microamperes of beam current (the current used for these irradiations). The resulting total power was A25 watts, corresponding to 1.7 Mev or a loss of 0.3 Mev in the accelerator window and intervening air. At 1.7 Mev the dE/dx in aluminum is 1.A7 Mev cm2/gram, so the A milligram per cm2 reactor window would decrease the beam energy by .006 Mev. Then the reactor window would transmit 0.997 of the energy striking it. As discussed above the wire support screen transmits 0.839 of the beam striking it. It was also found that the beam used, which had a three—inch scan, was 97.3% inside the reactor. The technique for determining this was to take the integrated power curve calculated from Figure 15, and find the fraction of the total power falling outside the reactor as the center of the curve was moved from the center of the reactor to 1.5 inches from the center of the reactor (a 3 inch beam scan). At the center 98.0% of the beam was inside the reactor, while at 1.5 inches from the center 96.A% of the beam fell inside. The total dose rate for the Midland irradiations was: 59 107 erg)(g rad ) “25 watts ('839)(°97u)('997)(watt SEC 100 erg = 3.A6 x 107 g—Egg SEC For a 1 minute irradiation of 25 pounds of material this gives a dose of 7 g_rad 60 sec lb 1 _ M-rad 3.A6x10 sec ( min )(A53.6g)(25 lb) 7 '183 min Dose calculations for the thin layer samples involve a set of calculations similar to those for the fluidized bed above but with the additional determination of the amount of backscatter from the glass holding the sample. The average atomic number for pyrex is 9.A, so the back- scatter will be 19.5% of the radiation incident to the pyrex.l Table 8 shows the calculations used for the thin layer samples. The average surface dose over the area was found from ionization chamber measurements 55.6 kilorads per second at 1.0 milliamp beam out current. The calculations were made on a per square centimeter basis, so the total beam is: B = 55.5 Kggid (.66u) 0.3 Eiégi = 9050 gram rads cm sec cm 1R. 0. Nicholas, "The Application of High-Energy Electrons to Some Grain-Infesting Pests" (Ph.D. Thesis, Department of Agricultural Engineering, Michigan State University, 1958). 60 TABLE 8 Dosage Calculations for the Thin—Layer Samples Based on One Square Centimeter of Dish Area Sample weight, grams Dish area, cm2 Simple thickness, mg/cm2 Foil thickness, mg/cm2 Do’ surface dose rate, Krads/sec Sample plus foil thickness, mg/cm2 Average dose to that depth from Figure Total forward energy absorbed, gram—rads/sec Total beam (from text) gram-rads/sec Energy striking glass, g-rads/sec Backscatter fraction for pyrex Backscatter energy to sample g—rads/sec Energy absorbed in foil per DO Energy absorbed in foil in g—rads/sec Energy absorbed in sample, g-rads/sec Dose rate to sample,* Krads/sec 6.36 63.6 100 none 55.55 100 1.072 DO 5950 9050 5100 0.195 1000 0 6950 69.5 9.54 63.6 150 5 55.5 155 1.007 DO 8660 9050 2390 0.195 A60 1.019 0.005 220 9120 59-0 *At 1000 Kv peak energy and 11.0 ua beam current. In the same way absorbed in the sample The difference between and foil and the total the pyrex dish, and 19. sample. 61 the amount of forward beam energy and foil covering were calculated. the energy absorbed by the sample beam is the amount of beam striking 5% of this is backscattered to the APPENDIX B PHYSICAL PROPERTIES OF METHOCEL The bulk density of Methocel varies from about 0.3 P1 to 0.5 grams per cubic centimeter,1 and is affected by the amount of packing. The particle density was determined with a pycnometer (using heptane as a wetting agent), and found to be 1.28 grams per cubic centimeter. An analysis 1 of the particle—size distribution was made down to HA microns least dimension with a set of sieves and down to 7 microns equivalent spherical diameter with a Roller particle—size analyzer.2 Sieve and roller analysis data are comparable only for spherical particles. As may be seen in Figure 17 (a photomicrograph of Methocel particles) Methocel can hardly be considered spherical. In order to compare the two sets of data, it was assumed that the particles were right circular cylinders where the length is 4 times the diameter, Dc' The equivalent spherical diameter, Ds’ which is the diameter of a sphere of equal volume, can be easily calculated by equating the volumes: lMethocel, The Dow Chemical Company, 1962. 2Roller, United States Bureau of Mines, Technical Publication 490, 1931. 62 63 ND2 C u — 1 3 — 33 _E_ ( DC) — 6 NDS or D — D The sphericity, w, of an object is the surface area of a sphere of equal volume divided by the surface area of the object, so: n D n D S 2 _ _ C _ W - 2 - 2 - —ET5 - .73“ c nDc n D 2(—F—) + NDC(u DC) From Perry,l pages 5—59, it may be seen that: = 18 uu I— _ DS / ETE;:E / K: where K1 - .8A3 log Tfi%§ for conditions in the Stokes—law region. The roller analysis yields D3, 0 - ———————l 111.1 .1 O : = DS — / g(pp:37 so that Ds Ds / K1 DC /6 From the above formula, K1 is .887 and: 1R. H. Perry, 0. H. Chilton, and s. D. Kirkpatrick, eds., Chemical Engineers' Handbook, Nth ed., (New York: McGraw—Hill, Inc., 1963). 6A c s / .887 j 6 = DO (1.062)/(1.8l7) .58“ D: S Thus, this correction was applied to the roller data. The results are plotted in Figure 16. 65 95 O = 4000 Cp, Sieve 90 - O A = 400 Cp, Sieve o D = 400 Cp, Roller (as per text) 80 ’- .. 70 P ; A to L 4 ° 0 O 50 -‘ A ’ +- ' A 40 A .3 30 ' 4.: H 20 - 0 g” A g l in 10 i' 4...) 53 U 5 I- H a) o. 2 L a 1 ' u .5 " D .2 ' 1 r- .05 h .01 J l J l 1 l J 1 A 200 10080 60 40 20 10 8 6 4 particle size in microns Figure 16. Particle Size Distribution for Methocel 66 scale ->| l‘- 5014 Figure 17. Photomicrograph of 4000 Centipoise Methocel 60 HG, Premium Grade APPENDIX C METHOCEL VISCOSITY DETERMINATION The 2 per cent solution Viscosities of Methocel were determined according to the ASTM designation 1 D 13A7—6A.‘ A moisture determination was made on each sample and sufficient Methocel was weighed out to give 2.00 grams of dry material. This was placed in a tall 8 ounce jar with 98 grams of 90°C water. The jar was covered and the mixture was stirred with a mechanical stirrer for 35 minutes. During the final 25 minutes the jar was in an ice bath. The solution was then spun in a clinical centrifuge for 15 minutes to remove all bubbles and finally placed in a viscometer in a 20.00 Celsius water bath for 30 minutes. At least A determinations of the flow time were made. The viscometer was chosen so that the flow time was between 30 and 150 seconds to minimize non-Newtonian behavior. The densities of all solutions were assumed to be 1.00 grams per cubic centimeter. Except for the low viscosity samples, all viscom— eters were calibrated with standard viscosity oils as l"Standard Methods of Testing Methylcellulose," A.S.T.M. Designation D—13A7—6U, 196A. 67 68 called for in the ASTM procedure for testing methylcel— lulose. The Methocel 10 HG samples were related to the assumed viscosity of the unirradiated material which had been determined by the Dow Chemical Company. In many cases the viscosity tended to drift down- ward while the readings were being taken. This was especially true for irradiated samples. A five minute time lag between readings might produce a 0.2 or 0.3 second difference in flow time. Most of the final average values, therefore, represented about 10 minutes standing in the water bath after the flow time determinations had begun. This error, which was less than 1.5% in all cases, was not significant for the use to which the data were put because the change for a sample was always small compared to the viscosity differences between samples. APPENDIX D THIN—LAYER IRRADIATION OF METHOCEL In order to determine the relationship of absorbed radiation dose and resulting 2 per cent solution Viscosities, # 9 samples of the 4000 centipoise material were irradiated in thin layers. Six 9.6M gram samples were placed in 9- centimeter glass dishes to make an even layer 150 milli— T: grams per square centimeter deep. They were covered with aluminum foil 5 milligrams per square centimeter thick. Three other samples of 6.36 grams each were placed in similar dishes with no foil covering, making a layer 100 milligrams per square centimeter deep. The samples were irradiated A7 centimeters below the accelerator window for various times and beam currents and at 1000 kilovolts peak energy. The calculations of the dose rate at 1.0 milliampere beam current are shown in Table 8. Table 9 shows the beam current, irradiation time, average dose, and resulting 2 per cent solution viscosi— ties for the nine samples and two controls. The dose— viscosity results are plotted in Figure 18. Several attempts to find a simple two or three factor analytical function to fit the curve in Figure 18 were made. However, no such function was found, so the points were taken off the curve where dose—viscosity data are needed. 69 70 a“. m2H 02am 0.mm 2.0 00 0H2 000a m.mm 2.0 m2 0HOH 0mm ©.mm 2.0 2m 0©m2 0 1| II 0 m.©fi 00mm m.m2 2.0 00m m.wm 0mm2 m.>2 2.0 02m 22H 0mmm m.m2 2.0 0mH mmm 0NMH m.w2 2.0 om OE owe TE :5 cm 0m0H 0mm m.w2 m.0 2m 0222 0 I: II 0 omfiomflpcoo mumsx oow\00Lx me mucooom “moflpHmOOmH> nmmow Hmpoe pcohuso Smog we «pcngSO Emom noEHu whomoaxm soapSHOm 2m 0.H pm mpmh mmom mammmq CHQB CH UopwfivmhhH Hmoozpoz omfiomfipcoo 0002 how sumo mpflmoomfi>lomoo 0 mqm