F BACTERIA: ANALYZER-DESIGN '_ DIALYSISCULTUREO V . V, ROWTH, AND DIALYSlSI AERAmNJ NUTRIENT TRANSFER. G Thesis for the Degree of Ph. I), WCHIGAN STATE UNNERS‘TY HAROLD E. B. HUMPHREY 1970 ‘ gag--3.» This is to certify that the thesis entitled DIALYSIS CULTURE OF BACTERIA: DIALYZER DESIGN, NUTRIENT TRANSFER, GRWTH, AND DIALYSIS AERATION presented by Harold E. B. Humphrey has been accepted towards fulfillment of the requirements for Ph.D. Microbiology degree in {Aéfl/hfi/éz, L1 L421QC ( v , V Malor professor Date June 30, 1970 0-169 _ :1 'V U-' 33.113 nzszcx, 3;“ A new and we: :25 been constrw ‘ x L'.‘ ;:ess with stainlc mugs. Clam-cf _~-'>.v .8". fi‘ ' ‘ 1 Les, St . £52.} ; . .1..d 0‘ mfg-5‘6 Mung and Effie-C‘- ~inner. The dial' ABSTRACT DIALYSIS CULTURE OF BACTERIA: DIALYZER DESIGN, NUTRIENT TRANSFER, GROWTH AND DIALYSIS AERATION BY Harold E. B. Humphrey A new and versatile dialyzer designed for biological application has been constructed. This instrument resembles a rectangular filter press with stainless steel end plates, one containing entry and exit fittings. Clamped between these is an alternating series of sheet membranes, stainless steel frames, and molded silicone rubber separa- tors. Each separator integrally provides gasketing, entry ports, and a field of pyramidal elements which support the membrane with minimal masking and effectively induce turbulent flow within each dialysis chamber. The dialyzer measures 43 cm.x 12.7 cm x 15 cm exteriorly, has an effective area of 288 cm2 per membrane, has a resident volume of 65 cc per dialysis chamber, and may be expanded by addition of cham- bers and membranes. The dialyzer was incorporated into a dialyzer-dialysis culture sys- tem which utilized a fermentor and separate reservoir of 3 to 10 liters, respectively. The efficiency of this system for nutrient transfer was evaluated by determination of half equilibrium times (ETSO) and over- all permeability coefficients (Pm). Mean values of 144 i 15 min (ETSO) and 4.7 1;.47 x 10.3 cm/min (Pm) were obtained with a 1% glucose solu- tion and a single dialyzer membrane. Continuous Operation, autoclaving, dialyzer position, and fluid direction had no effect on dialysis per- formance. Optimum.rates were obtained with an area of .2880 m2 and bulk F‘— _ his ‘-A-‘ fix rates abo.'e .,. Erma conventions. mile nutrient mi. :z-e eppssite side. afrariatian bent: :iiialysis cultur 213:, and extensit: a? comparable n;- ‘f‘ -:: . . 4. c..ect on diam p HAROLD E. B. HUMPHREY flow rates above .5 L/min. The variations in performance associated with large overall liquid volumes and volume ratios and the signifi- cance Of internal fluid dynamics were discussed. The effectiveness of this dialysis culture system was demonstrated by the growth Of Serratia marcescens. Liquid cultures were circulated from a conventional 5 liter fermentor through one side Of the dialyzer while nutrient media from a separate reservoir was circulated through the Opposite side. A series Of growth trials established the extent of variation between repeated results and demonstrated the superiority Of dialysis cultures, in terms Of culture viability, culture concentra- tion, and extension Of the eXponential and stationary growth phases, over comparable nondialysis cultures. Fluid circulation velocities had no effect on dialysis growth. A 1:10 fermentor tO reservoir volume ra- tio, obtained by increasing the reservoir volume, produced Optimal bio- logical performance. The addition Of dialyzer membranes resulted in improved culture concentrations with 15 membranes (.4320 m2) producing a maximum.viable cell concentration Of 283 billion cells/ml. Diffusion- al access to the nutrient reservoir was shown to be instrumental in maintaining a culture environment which permitted the high dialysis growth yields, an extension Of the stationary phase for over five days, and the concentration Of cells on a relatively weak growth medium. In several instances the dialysis growth response to Operational conditions was similar to that shown previously for nutrient transfer. Culture aeration by dialysis gas transfer was demonstrated by in- corporating silicone rubber membranes into the dialyzer. Liquid cul- tures Of Serratia marcescens were circulated from a conventional 5 liter unSparged fermentor through the dialyzer past one face of the irranesh) V1116 '5.-'.n ' ' l -‘ it.-...a:ec bUDblC . £3; Ia5.;fi.-‘ re, salv- .. a... .. _. Ugh“ seam to cram tranes at the em; “v3. . Max:511 5%,: 'EL 0: "“1395 When dia: [L'Zure SYStEn HAROLD E. B. HUMPHREY membranes(s) while humidified gas (air) was circulated past the Op- posite face. This aeration system enabled the use of non-sterile air, eliminated bubble-liquid interfacial effects, and reduced the power and antifoam requirements. The membrane represented the greatest re- sistance to oxygen transfer. This was reduced by the addition Of mem- branes at the expense Of Operational efficiency. A single membrane (.0288 m2) produced a KLa Of 6.2 x 10'2 min"1 and an OTR Of .0018 m moles 02/L/min. Six membranes (.1728 m2) increased these values to 28 x 10-2 min“1 and .0250 m moles OZ/L/min respectively. Increases in liquid velocity especially, gas velocity, oxygen partial pressure, and overall gas pressure improved the dialysis transfer of oxygen. Although the rate Of oxygen transfer with six membranes was only l/20th of that for a well Sparged fermentor, dialysis-aerated cultures attained equal or greater pOpulation densities. In addition, comparable results were Obtained when dialysis-aeration was combined with a nutrient-dialysis culture system. c;! \’I' ”Y": anal 335’" in D" .GI't' DIALYSIS CULTURE OF BACTERIA: -DIALYZER DESIGN, NUTRIENT TRANSFER, GROWTH, AND DIALYSIS AERATION BY Harold E. B. Humphrey A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree Of DOCTOR OF PHILOSOPHY Department of Microbiology and Public Health 1970 } k» @- — 45:5 .. 9 /".;:Z.5)' - x3 ‘5. I wish to dedicate this thesis in memory of my father, Harold E. B. Humphrey, Sr., whose example Of patience and persistence has been indispensable. ii :0! who 1 “Hugeaent our 1:; (1:4. .1 .:. ...... an. DaLJCHCE; ; v ~‘.‘> «LAY. ACKNOWLEDGMENTS I wish to express my gratitude tO my wife, Jane, for her help and encouragement during the course Of this work; to our families for their faith and patience; and to Dr. Philipp Gerhardt for providing the Oppor— tunity. iii l .‘aTfi‘V ' av,— u 'b' “ .II'1- ‘shug...\'L\L'_ 1.1 List 05 i p 27‘": t 7 n ‘ ‘J an J-.\~.g\. L I ‘ Rab. ‘ ‘ et - ._ . ‘9 j '7! H I n. i' ' “n Jud-:ZLR D;bT,’t\; "v- TABLE OF CONTENTS 1. PRELIMINARIES 1.1 LiSt Of Tables. 0 o o o o o o 1.2 List Of Figures . . . . . . 2. GENERAL INTRODUCTION . . . . . . . . 3. DIALYZER DESIGN. . . . . . . . . . . 3.1 Introduction and Historical . 3.2 Description . . . . . . . . . 3.3 Summary . . . . . . . . . . . 4. NUTRIENT TRANSFER CHARACTERISTICS. . 4.1 Introduction. . . . . . . . . 4.2 Materials and Methods . 4.3 Results . . . . . . . . . . . 4.4 Discussion... . . . . . . . . 4.5 Summary . . . . . . . . . . . 5 - GROWTH OF A TYPE BACTERIUM . . 5.1 Introduction and Historical . 5.1.1 Introduction . . . . 5.1.2 Historical . . . . 5.2 Materials and Methods . . . . iv Page vii ix 15 l6 16 18 24 44 58 59 59 59 63 75 C ‘I mus: i3 Results .S AE“I~ ta 5.3 5.4 5.5 TABLE OF CONTENTS (CONTINUED) Results . . Discussion. . . . . ... . . . Summary . . . . . . . . . . . . . 6. DIALYSIS AERATION . . . . . . . . . . . . . . . 6.1 6.2 6.3 6.4 6.5 6.6 Introduction and Objectives . . Historical and Theory . . . . 6.2.1 Oxygen Demands of Microbial Cultures . 6.2.2 Oxygen States . 6.2.3 Nondialysis Aeration Methods . 6.2.4 Mechanism Of Oxygen Transfer . 6.2.5 Principles and Problems Of Bubble Aeration . . . . . . . . . . . . . 6.2.6 Mechanism Of Membrane Oxygen-Transfer. 6.2.7 Principles Of Dialysis Aeration-Silicone Membranes. . . . . . . . . 6.2.8 Applications Of Dialysis Aeration. Materials and Methods . . . . . Results . . . . . . . . . . . . . 6.4.1 Physical Aeration Characteristics. 6.4.2 Culture Demand . . . . . 6.4.3 Biological Aeration Characteristics. Discussion . . . . . . . . Summary . Page 82 105 116 117 117 121 121 124 129 133 141 146 150 157 163 175 175 196 200 209 224 h. In" 22313.1 SL'H‘LXR': BIBLIEMFEY. . YAIERL-‘ILS REFS? 1"" 1? Vfllj ....\ .. , _ 13.1 Dialj.':-. TABLE OF CONTENTS (CONCLUDED) Page 7. GENERAL SUMMARY . . . . . . . . . . . . . . . . . . . . . . 226 8. BIBLIOGRAPHY. . . . . . . . . . . . . . . . . . . . . . . . 229 9. MATERIALS REFERENCES. . . . . . . . . . . . . . . . . . . . 241 10. APPENDIX. . . . . . . . . . . . . . . . . . . . . . . . . . 244 10.1 Dialyzer Cost Analysis. . . . . . . . . . . . . . . 244 vi I r ..L. (T! (.4) l l.’ 1! LS Inttuence Influence direction Iniiuence Influence dialysis InfIUence flUld hC1;< glucose 3;, Table 10 11 12 13 14- 1.1 LIST OF TABLES Influence of steam sterilization on glucose dialysis Influence Of duration Of dialyzer Operation on glucose dialysis Influence Of dialyzer Operating position and fluid flow direction on glucose dialysis Influence Of membrane hydration time on glucose dialysis Influence of manufactured lots Of membranes on glucose dialysis Influence Of number Of membranes on dialyzer volume and fluid hold-up time Influence of the total volume of the dialysis system on glucose mass transfer and dialysis rates Page 28 28 28 29 29 31 40 Influence Of fermentor-to-reservoir volume ratios on glu-. cose mass transfer and dialysis rates Influence Of temperature on glucose dialysis and diffu- sivity Influence Of the initial reservoir concentration on glucose dialysis rates, diffusivity, and mass transfer Comparison of total and viable cell concentrations in dialysis and nondialysis growth trials with Serratia marcescens Influence Of selected growth medium on cell production in dialysis and nondialysis cultures Influence Of fermentor-to-reservoir volume ratios on viable cell concentration during dialysis culture of Serratia marcescens Gas permeation coefficient Of membrane films vii 43 53 55 86 99 101 154 s u 0‘ r9 N) r...) L. I [\J ‘~- IN) (J1 . 1 An exasp.e graphical Influence loss Influence COEIILCLCC Influence internal .t'! - Ina ‘uLnC transfer ~ Influencg gen trans IREIUEHC£ f£3V rate Calculat. at three Dental on dialysis ‘ LIST OF TABLES (CONCLUDED) Table Page 15 An example Of the computations necessary for the graphical datermination Of the overall KLa value 171 16 Influence Of the aeration method on culture liquid loss 176 17 Influence Of the gas selected on the oxygen transfer coefficients for the dialysis-aeration system 179 18 Influence of gas velocity through the dialyzer on the internal gas pressure 179 19 Influence Of air velocity and membrane area on oxygen transfer rates for the dialysis aerator 183 20 Influence Of liquid velocity and membrane area on oxy- gen transfer rates for the dialysis aerator 186 21 Influence Of membrane area, liquid flow rate, and air flow rate on dialysis-aeration oxygen-transfer rates 187 22 Calculated liquid film resistance to oxygen transfer at three liquid flow rates through the dialyzer-aerator 189 23 Influence Of membrane area on the calculated and experi- mental oxygen transfer rate and coefficient values for dialysis aeration 192 24 Theoretical and experimental oxygen transfer coeffi- cients for dialysis aeration, and estimation of the membrane area necessary tO meet active culture require- ments 195 25 Oxygen demand Of a Serratia marcescens culture 198 26 Volumetric oxygen transfer coefficients during propa- gation of Serratia marcescens in a conventional fermen- tor 198 viii (7‘ (I? H Dimension. silicone ; Schematic deter2inii Least Sqt fractive cartelat: and reset ' . F trial g1: Exazple ; deEErzina Illf luean; Sis. Fer (Curve A) rate tnrr Correlat; zer and : to the :1 Figure 10 11 12 1.2 LIST OF FIGURES Exterior view of the dialyzer assembled with ten membrane units A stainless steel separator support frame (left) and molded silicone rubber separator (right) Dimensional view Of a corner portion of the molded silicone rubber separator, magnified about 10 x Schematic diagram Of the dialysis system used for determining glucose transfer Least squares fit Of the gorrelation curve for re- fractive index values (n 25) and glucose concentrations Correlation Of glucose concentration in the fermentor and reservoir with time during an equilibrium dialysis trial with a 300 ml fermentor and a 1000 ml reservoir Example of a glucose dialysis rate plot used for ET50 determinations Influence of the number Of membranes on glucose dialy- sis. Fermentor to reservoir volume ratios of 1:3 (curve A) and 1:1 (curves B and C) were used Correlation Of glucose dialysis with the fluid flow rate through the dialyzer Correlation between fluid flow rate through the dialy- zer and pressure drOp in the dialyzer chambers adjacent to the membrane Influence Of dialyzer membrane area on the pressure drOp across a given dialyzer chamber at fluid flow rates of l and 2 L/min Influence Of flow rate through the dialyzer on the pres- sure drOp across a given dialyzer chamber with 5 and 10 membranes installed ix Page 11 12 19 21 25 26 30 32 34 35 36 ... nrvre . .~“ d (\J *4 [\J LIV (‘9 \I Influence Influence cose dial‘ Influence initial c Influence glucose half equi Rapresent of tVO 2e €3Ch surf channel Figure 22, 13 14 15 16 17 18 19 20 21 23, 24 25 26 27 28 LIST OF FIGURES (CONTINUED) Influence Of Operating temperature on glucose dialysis Influence of initial reservoir concentration on glu- cose dialysis Influence Of concentration on glucose diffusivity and initial dialysis rate Influence of fermentor-to-reservoir volume ratios on glucose dialysis and mass transfer necessary to reach half equilibrium concentration Representation Of fluid flow past the Opposing faces Of two membranes, with a laminar film immediately along each surface and turbulent bulk flow in the central channel Schematic diagrams of the culture systems used. TOp; a dialyzer-dialysis culture system. Bottom; a non- dialysis "control" culture system Assembly Of the eXperimental dialysis growth system Growth curve Of Serratia marcescens; mean Of four duplicated batch nondialysis cultures Growth curve Of Serratia marcescens; mean Of dupli- cated dialysis cultures with ten membranes (.288 m.) in the dialyzer Growth curves of Serratia marcescens, in dialysis and comparable nondialysis culture Influence of dialyzer membrane area on growth of Serratia marcescens Influence Of dialyzer membrane area on cell mass of Serratia marcescens after 48 hours growth under the same conditions given in the preceding figure Correlation between dialysis growth and solute (glucose) concentration in the fermentor and reser- voir Correlation between nondialysis growth and solute (glucose) concentration Page 37 38 39 42 51 76 78 83 89 90 92 95 96 743w? A‘ gal. 39 Influence the viabl over an e H Cozputed I fernent c: cultures 31 Influenc._ ETJ-‘th c: 33 Diagram c Pflase to 33 Dlagra: N/ ‘ )5 Diagram ; Culture ‘: 3) Relatxm. 36 LIST OF FIGURES (CONTINUED) Figure Page 29 Influence Of dialysis culture and membrane area on the viable concentration Of Serratia marcescens cells over an extended growth period 98 30 Computed and experimental effect of the reservoir-to- fermentor volume ratio on the cell density of dialysis cultures 102 31 Influence of liquid flow rate through the dialyzer on growth of Serratia marcescens 104 32 Diagram of the transfer of gas molecules from the gas ' phase to a cell suspended in the liquid phase 137 33 Diagram of the dialysis-aeration culture system 164 34 Diagram Of the dialysis-aeration-—dia1ysis-nutrient culture system 165 35 Relationship between percent saturation and PPM con- centration Of dissolved oxygen at 30 C 169 36 Oxygen uptake by 3 liters of medium in a 5 liter fer- mentor aerated by dialysis 170 37 Graphical determination Of the overall volumetric oxy- gen transfer coefficient (KLa) for the dialysis aera- tion trial shown in Figure 36 172 38 Influence of air velocity and membrane area on dialysis- aeration oxygen transfer rates 182 39 Influence Of liquid velocity and membrane area on dialy- sis aeration oxygen transfer rates 185 40 Influence Of membrane area on experimental and calculated oxygen transfer rates and oxygen transfer coefficients for the dialysis-aeration system 191 41 Correlation Of the oxygen demand and growth in an air sparged, nutrient-dialysis culture Of Serratia marces- cens 199 42 Dissolved oxygen concentrations in the culture medium during batch growth of Serratia marcescens under Spar er and dialysis-aeration with a membrane area Of .1728 m (6 membranes) 201 xi 75:: i3 Growth (2 broth u: I branes, ‘__ LJI h (J ’1 ' i m 9.4 ..J ‘1) r v Figure 43 44 45 LIST OF FIGURES (CONCLUDED) Growth of aggratia marcescens on Trypticase soy broth under dialysis-aeration (6 silicone rubber mem- branes, .1728 m area) and Sparger aeration Growth Of Serratia marcescens on water diffusate Of Trypticase soy broth under; dialysis-nutrient and dialysis-aeration, dialysis-nutrient and sparger- aeration, and nondialysis nutrient and sparger-aera- tion culture conditions Correlation between growth Of Serratia marcescens and the culture aeration method and agitation xii Page 202 204 206 Dialysis ha.- ‘tiraug’n a seziper ingrapertion to first erployec' 5" separation of cc": 3 :erzng ez-rperiner. 2 ‘f'n ' fine n tiL'E ’as 2. GENERAL INTRODUCTION Dialysis has been defined as the selective tranSport Of solutes through a semipermeable membrane (interface) in the direction Of and in prOportion to a concentration gradient (102). The principle was first employed by Thomas Graham over 108 years ago for the effective separation Of colloids from crystalloids in urine (70). These pio- neering experiments showed that the quantity of solute diffused in a given time was prOportional to the concentration Of the original solution, a principle from which Fick's law Of diffusion has evolved. The dialysis process continues tO be used for the separation, purifi- cation, and identification of solutes in research laboratories. The commercial economies afforded by dialysis Of spent process liquors for recovery Of processing chemicals constituted the primary focus and application of the dialysis process during the first half of this century (10). More recently the basic principle Of dialysis has been extended to a number Of unique applications such as artificial organs, ultrafiltration, fuel cells, packaging films, reverse osmosis, solute concentration, food processing, and administration Of anesthesia. Except for some isolated applications at the beginning Of this century, dialysis was largely ignored in microbiology. However, since the early 1950's a surprisingly large number Of microorganisms have been studied with igiyiggg dialysis culture; see Schultz and Gerhardt (141). The motives for employing the dialysis principle were as varied 1 évc‘onal I‘EECVa * “an and inde PCP-C cells Uith‘n a r4 231:: end product feeiaack inhib“ t: 5:lic end product iniependent O;€:I‘£ :etc'n or cont inu At first ti. as the genera selected for use: extension Of eXponential growth; maintenance Of high viable cell concentrations in prolonged stationary growth; diffusional access to a nutrient reservoir; dilution and dif- fusional removal Of toxic or inhibitory culture metabolites; separa- tion and independent manipulation Of process regions; propagation Of cells within a particulate free medium; recovery Of nondiffusible meta- bolic end products within the particulate-free medium; prevention Of feedback inhibition by the dilution and removal Of diffusible meta- bolic end products to the reservoir region; symbiotic prOpagation; and independent Operation Of the fermentor and/or reservoir regions on a batch or continuous basis. At first the dialysis culture equipment designed for use in most Of the above investigations involved implantation Of membrane sacs, suspension Of dialysis sacs in vessels containing nutrients, fabrica- tion Of flasks with membrane bottoms, or other schemes. Frequently these were either poorly designed, adapted for a specific application, could not be duplicated, or were not adequately described. The lack Of uniform or sound equipment design, as well as little or no reported data on the performance characteristics Of the system employed, has made evaluation Of the experimental results difficult and subject to question on the basis of the unknown influence Of the equipment or system itself. Recently, a dialysis culture system based on the separation and independent Operation Of the fermentor, reservoir, and exchanger re- gions has been deveIOped by Gallup and Gerhardt (57). Such an arrange- ment affords a high degree of Operational flexibility and control, uti- lizes conventional fermentation equipment, can be scaled to a larger size, and reduce: earlier experimer fer subsequent d..- sis culture conc- In this the. -. ‘7' --.‘e dens. strati: an; to extend, cc culture aeration The first c H 18 ‘lali’zer used w- £13m of a size, and reduces the technical limitations and uncertainty found with earlier experimental schemes. This system established a reliable basis for subsequent development, demonstration, and extension Of the dialy- sis culture concept. In this thesis I seek tO refine and extend the basic concept de- velOped by Gallup and Gerhardt. The goal is to present a comprehen- sive demonstration and analysis Of the Operational parameters and the biological performance Of dialysis culture within the identifiable con- fines Of a system utilizing a new and Specifically designed dialyzer and to extend, convincingly, the dialysis concept to a new application, culture aeration. The first Objective (Section 3) is to describe thoroughly the new dialyzer used in this work. This instrument represents the culmina- tion Of a design Specific for biological application, which has evol- ved through several preliminary models over several years. With the develOpment Of a workable and satisfactory dialyzer, it became possible to analyze the nutrient transfer characteristics Of the entire dialysis culture system and the dialyzer itself under var- ious Operational modes similar to those which might be employed during propagation of microorganisms. The glucose-transfer data gathered (Section 4) will provide the background information necessary for the identification and eXplanation Of Operational Optimums and the Opera- tional limitations of this system, permitting a more meaningful evalu- ation of the dialysis growth trials. The third Objective (Section 5) is to demonstrate by a series Of dialysis growth trials the capacity of the system for prOpagation Of a bacterial culture and its superiority over conventional methods. An azzezgt will be 7 to the preceding as to basic dial: Ine final c‘: tie cancent note: rnzranes with 'r. r=s=ful ' ~~ (15.61th- apprsacn to 51:22: 1° liqud could ‘~ :- ‘V is ea and IESlStav attempt will be made to relate the Observed dialysis growth results to the preceding solute transfer characteristics for the system as well as to basic dialysis principles. The final Objective (Section 6) is to extend, successfully, the dialysis concept to the problem Of culture aeration. Features which have made dialysis attractive for culture solute transfer also make the concept potentially applicable to gas transfer. Recently develOped membranes with high gas permeability have made this practicable. Suc- cessful development Of the dialysis aeration process will Open a new approach to supplying a gas environment to liquid cultures, especially heretofore sensitive Species or those with complex or changing gas de- mands. It also represents a system by which oxygen transfer from gas to liquid could be investigated with a fixed and defined interfacial area and resistance. LA) L Ca 1 lup an- ( aerating a dial: 3. DIALYZER DESIGN 3.1 Introduction and Historical Gallup and Gerhardt (57) have outlined several schemes for Operating a dialysis culture Of'WhiCh the dialyzer-dialysis system appeared most practical and compatible with conventional fermentation equipment. The system may be Operated in four modes: fully batch, fully continuous, batch fermentor and continuous reservoir, or con- tinuous fermentor and batch reservoir basis; see Figure 8 in Schultz and Gerhardt (141). In all Of these the key to successful Operation is a suitably designed diffusion exchanger, or dialyzer. A plastic and stainless steel laboratory dialyzer designed for chemical application (Graver Water Conditioning Company) was tried first (57). Although this unit successfully enabled the demonstration of a prototype dialysis system, its usefulness was limited by construc- tion materials, membrane capacity, and fluid dynamics. Other commer- cial dialysis devices are available but most suffer from similar nega- tive characteristics. This situation with chemical dialyzers apparent- ly results from the historical application Of dialysis to chemical separations such as caustic soda from rayon steep liquor, sugar ex- traction, drug purification, and acid recovery from metal plating liquors (10). The most nearly satisfactory commercial exchangers for microbiology are those developed as artificial lungs and kidneys. Such dialyzers have 5 ,- :rane type, in 'v~' with support urn. bundle of fine f aai-r'rane csnfir. a:.e geocetr.‘ ( sary fer succes: flih’ capacity, After unsu: fete suitable 'n»; bl b (b “Iii-Or Unit :. AL). .. LI. ?. t ‘0 been designed in three basic forms: the coil type, in which membrane tubing is spirally wrapped around a mesh cylinder (93); the plate-and- frame type, in which a series Of flat membranes are sandwiched together with support units (144, 56, 46); and the capillary type, in which a bundle of fine membrane tubes is used (153). The rectangular plate- and-frame configuration was found to be mathematically the most prefer- able geometry (168) and seems most applicable to a microbiological cul- ture system because of its simplicity, capacity, and durability. How- ever, even the best Of these hemodialyzers lack certain features neces- sary for successful culture application: namely sterilizability, bulk flow capacity, and full turbulence on both sides Of the membrane. After unsuccessful efforts to modify a chemical dialyzer, and be- fore suitable hemodialyzers became available, Gerhardt, Pederson and the author undertook to design a dialyzer eXplicitly for dialysis cul- ture applications. The design criteria for such a dialyzer included the following requirements: variable membrane area; adaptability for using different types Of membranes; a geometric form.which could be scaled up tO industrial dimensions; minimal priming volume (and there- fore minimum fluid holdup); positive gasketing; sterilizability; con- struction from nontoxic materials; induction Of turbulence at the mem- brane face; and capacity for fluids of various viscosities. Construc- tion Of a series of models and prototypes culminated in the design and assembly Of a dialyzer which satisfied these design criteria. This instrument is described below. a. .‘ 1 11522858. :10 - izg bolts ” are EC sian unirormly 5 is blank except A :V-‘S .5 e a ‘15 in each cc which are sealeq plate (not Show: lively , althougi gaskets nd met k 1““ EnCes - Citati s des 3.2 Description The dialyzer (Figure I) basically resembles a rectangular filter press and consists Of two end plates which compress an alternating series Of molded separators, sheet membranes, gaskets, and frames. Each end plate is constructed from type 316 stainless steel (4-M)* cut 1/2 inch thick, 17 inches long, and 5 inches wide. The faces are machined to flatness. Holes that accommodate 1/4 inch diameter stainless steel clamp- ing bolts are equally Spaced around the perimeter to distribute compres- sion uniformly and to align components during assembly. The back plate is blank except for the clamping holes. The front plate has threaded holes in each corner to accommodate four Swagelok entry fittings (l4eM), which are sealed with O-ring gaskets and teflon thread tape. A back plate (not shown) Of 3/4 inch pyrex glass (l3-M) may be used alterna- tively, although this back plate requires the addition of 5 teflon-cush- ioned steel bars to accommodate the clamping bolts. It permits visual inspection of the fluids as they pass through the dialyzer. This fea- ture was useful during physical evaluation Of the dialyzer and in mon- itoring cultures. Each of the repeating functional units consists Of a flat membrane, gaskets and metal frames on both sides Of the faces of two separators. *References designated with an "M" are found in the material reference citations. this unit nay Du the separatets and ends, ant in. hers. The no)...“ u’ \ . seizly still r 1223, 5 inches 15 pcunds. :‘my type cf In continuity NI regular Vislzing With, .0016 in Neither of thee 3f gaskets. 1'12. Ill-EDS! (15.31) '— .. ’ u This unit may be visualized as a rectangular box 1/4 inch thick with the separators as tap and bottom, gaskets and frames for the Sides and ends, and the membrane bisecting the box into two dialysis cham- bers. The volume of each chamber is 65 ml and the eXposed area Of the membrane is .0288 m2. The total membrane area may be increased simply by adding additional units between the end plates. Fifteen units, with a total membrane area Of .5320 m2, have been successfully employed. Since a single unit is only 1/4 inch thick, the 15-unit as- sembly still remains compact and easy to handle, measuring 17 inches long, 5 inches wide and 4 3/4 inches thick, and weighing approximately 15 pounds. Any type Of sheet membrane material may be used in this instrument. In continuity with previous work (57), membrane sheets were cut from regular Visking regenerated cellulose dialysis tubing Of 3 in. flat width, .0016 in. dry thickness, and 5 munominal pore diameter (28-M). Neither of these membrane materials are self sealing, necessitating use Of gaskets. These are 1/32 inch thick, 30 durometer S-2000 silicone rubber (lS-M), Cut by means Of a die in the shape Of the stainless steel separator support frames. The separator support frames (Figure 2), made Of type 316 stainless steel sheets, are 16 inches long, 4 inches wide, and .05 inches thick. They are cut to fit over, and match, the perimeter sealing edges Of the molded silicone rubber separators, as illustrated. These frames give rigidity to the perimeter sealing edges of the separators, bridge the triangular entry ports, and provide an even surface for seating the mem- brane (and membrane gaskets) over these critical areas. Figure 1. Exterior view of the dialyzer assembled with ten membrane units. The separa’. has been a difli zers have metal fluid conduits 3 . Izese separator turbulence and p typical of the i Henodialyzt and require the areror units co: i‘ztn cone field (it curar types, | tile membrane tu" “bins (138). Pe‘ In' 10 The separator, a combination membrane support and fluid chamber, has been a difficult but essential design feature. Industrial dialy- zers have metal or plastic separator plates with appropriately placed fluid conduits and wire screens (136), rubber coated extruded metal mesh (152), machined grooves (6, 85, 94, 139), or blade-like grids (76, 162) fastened or molded in the central area for membrane support. These separator styles generally are successful in promoting fluid turbulence and providing support for the thick and rugged membranes typical of the industrial chemical dialysis processes. Hemodialyzers, which require support for thin delicate membranes and require the passage of fragile blood cells, are designed with sep- arator units composed of screens, longitudinally grooved mats, or mats with cone fields, e.g. Leonard and Bluemle (102). In representative circular types, plastic or neOprene screens are wrapped adjacent to the membrane tubing around a cylinder (99), or are placed inside the tubing (138). In representative rectangular types, sheet membranes are compressed between plates with longitudinal grooves or ridges (89, 30, 46) or between mats with a field Of staggered cone or column support elements (65, 84, 17, 122). In an evaluation of these styles, Peirce (120) concluded that the staggered cone design provided superior efficiency, fluid flow, and minimal membrane masking. Our dialyzer separators are shown in Figures 2 and 3. They are custom molded from silicone rubber (15eM), comparable to Dow Corning Corporation medical grade S-2000 Of 50-durometer hardness. The separa- tor in one piece provides gasketing, entry holes and ports, and fields of membrane-support elements. It is completely reversible. The four entry holes in a series Of separators and frames correspond with the 11 Figure 2. A stainless steel separator support frame (left) and molded silicone rubber separator (right). Gaskets also are used, which are shaped the same as the steel frame. Figure 3. Dimensional view Of a corner portion of the molded silicone rubber separator, magnified about 10 x. The gasketing beads, entry hole, triangular entry port with hemispherical supports, and pyrimidal membrane support elements are visible. entry taps in t' triangular entrj cal supports ft: tree. This are perimeter of ti; eral alignment provide uniforr arisen turbule‘ the nee-brane Ni eiges creating a h‘ v f9 .r the Ididtn 5'. y. I a II]? at t L. .le :‘C- d; atity *‘n *-9 molde. c." : :i‘fe Ye a ‘ r5. 1. A‘l‘ l3 . entry taps in the front end plate. Separator entry holes lead into triangular entry ports (Figure 3) which contain a number of hemispheri- cal supports for the bridging portion of the steel frame positioned over them. This area is critical with reSpect to potential leakage. The perimeter of the separator has two molded .003 inch high beads for seal- ing this surface. The membrane support elements are unique. They consist of snubbed equilateral pyramids, each 1/8 inch in height, set in a field with lat- eral alignment and longitudinal staggering (Figure 3). These elements provide uniform membrane support with minimum surface masking and develOp maximum turbulence as fluid strikes their faces and is diverted against the membrane with a scrubbing action. Fluid also passes around their edges creating eddies, as on the downstream side of a weir. In addition, the arrangement of the entire field provides even distribution of fluid over the width, but a tortuous path over the lenth, of the separator. This unusual design minimizes stagnant fluid movement and reduces laminar flow at the membrane surface, while maximizing bulk fluid mixing and flow capacity. The molded silicone rubber separator just described represents the culmination of a series of prototype designs developed over the past five years. It supersedes a preliminary prototype which was used in some of the initial solute transfer evaluations reported in the follow- ing section. The preliminary separator has been fully described (see H. E. B. Humphrey, M.S. Thesis, University of Michigan, 1965). It was composed of two metal frames, similar to that shown in Figure 2, a 1/16 inch thick silicone rubber spacer gasket with fluid entry channels cut in opposing corners, and a molded neoprene mat for membrane support and ‘mbulence incur - - a. rarity formed ._ A cost an presented in ti The dial_.':. “-‘it'n prop gatiz‘. raintains ster; accemoc‘ates i:— bclism of nutr; internal Press; into a culture 14 turbulence induction. The mat was trimmed to fit into the central cavity formed by the above frames and Spacer gasket. The current prototype separator combines the form and function of the former Spacer gasket and mesh mat into a single molded unit. A cost analysis for construction of an Operational dialyzer is presented in the appendix. The dialyzer is designed to accommodate the problems associated with prOpagating microbial cultures. It may be sterilized by steam, maintains sterility by positive seals, is nontoxic to growing cells, accommodates increasing culture viscosities associated with cell meta- bolism of nutrient media, allows high bulk fluid flow without undue internal pressures or holdup, and is easily assembled and integrated into a culture system. -:;>e; .. A u ‘I. nLv ac ~l.. v s .C .C ..C fl... . .5 at c. c. ‘_¢. \ sq xx; 3.3 Summary A new and versatile dialyzer designed for biological use has been described. It resembles a rectangular filter press with stainless steel end plates, one of which contains entry and exit fittings. Clamped between these are a series of sheet membranes, thin stainless steel frames, and molded silicone rubber separators. Each side of a separator forms, in conjunction with a frame and membrane, a dialysis chamber. The separator in one piece provides gasketing, entry ports, and membrane support. The field of pyrimidal support elements on each face induces turbulence as well as supports the membrane, .0288 m2 effective area, with minimal masking. Dialyzer area is expanded by installation of additional separators and membranes. This instrument represents a reliable and efficient exchanger which may be incorporated into a functional dialyzer-dialysis culture scheme.. 15 Q. ... :— Q. 5‘ L» A: 4. NUTRIENT TRANSFER CHARACTERISTICS 4.1 Introduction The key to a successful dialysis culture system is separation Of the three process regions (fermentor, reservoir, and exchanger) to en- able independent control Of each (57). Although a pilot dialyzer-dialy- sis system was successfully demonstrated utilizing conventional fermen- tor vessels and a commercial chemical dialyzer’unsuccessful efforts to modify the exchanger to the necessary Specifications have prevented a complete analysis of the system. The essential function of an exchanger, is to exchange diffusible materials from one process region through the semi-permeable membrane to another region and to retain nondiffusible materials in the desired region under culture conditions. The dialyzer described in the preceding section satisfied such Specifications. Possession Of a satisfactory dialyzer permitted a critical and complete analysis Of physical, physiological, and culture influences on the solute dialysis characteristics Of this dialyzer-dialysis system. The effect Of these Operational procedures, which are indicative Of dialyzer performance, must be known in order to establish the parameters, limitations, and Optimums for efficient dialysis culture Operation. U1- timately such information will aid in the selection Of satisfactory Op- erational schemes and Specifications for further dialysis applications. Information of this type, with the exception of membrane testing (61), has been lacking. Consequently, an analysis of solute transfer 16 ‘ l7 characteristics for our dialyzer and dialysis culture system under po- tential Operating conditions is presented below. All Operational pro- cedures were examined, even those in which the results appear Obvious, in order to thoroughly document and confirm the influence of these on performance. In addition, an explanation was sought for the selection Of and the results Observed with certain Operational conditions and procedures. Although many Of the conclusions were anticipated from the laws Of diffusion, this work contributes a meaningful and necessary evaluation Of nutrient dialysis within the system prOposed for dialysis culture of microorganisms. .\ 5n: 5 a 1.3: d in :ixer (3; reaperart (Q! 4.2 Materials and Methods The dialyzer was evaluated under conditions which simulated those for a growth trial with the dialyzer-dialysis culture system diagram- med in Figure 4. A 2-1iter filter flask or a 5-1iter fermentor and a lh-liter fermentor (24-M) served as the culture and reservoir vessels respectively. These were situated in a water bath equipped with a mixer (22-M), thermo-regulator (3-M), and immersion heater (Z-M) for temperature control. The dialyzer was equipped with presoaked Visking regenerated cellulose dialysis membranes (28-M) and connected to the vessels by 1/4 inch I.D. rubber tubing (ZS-M). A positive displace- ment variable Speed pump (27-M) inserted in the reservoir and fermentor circuits circulated the fluids from their reSpective vessels through the dialyzer and back. Glass "T's" capped with vaccine bottle stoppers were also placed in the fluid circuits to allow sample removal by syr- inge and needle. Unless noted Otherwise, the reservoir was charged with one liter Of a 1% glucose solution (which approximates the concentra- tion in a growth medium), the fermentor contained 300 m1 Of distilled water, the liquid flow rates were 2 liters per minute, the dialyzer held one membrane (.0288 m2), and the temperature was maintained at 30. C. The dialyzer employed in most of these tests as well as all sub- sequent experiments and trials in this thesis was the model described in the preceding section. A preliminary prototype dialyzer, referred 18 l9 uo. ooooooooo o . _ 0.: _ .10....0'. : ...-R z... ' :0. f... - -,-r.' 3?:- '.'- .:.:. o.0.'.:.'n.o 0-.‘.I.'..:'.': oz': : z I ...:zo-c“; .:::::::’:.I.I GLUCOSE RESERVOIR FERMENTOR DIALYZER Figure 4. Schematic diagram Of the dialysis system used for deter- mining glucose transfer. l o—- .‘ :— t, ... ...c ..v‘ ...- V :b.’.». \ -~ 1 “'7." .. C ..4-- A»... ‘v; c 7. W ‘ I .-.~-' ‘ . ‘ '“o‘t. .t: " 032?.516 c .323 ~—: "so -.: cu - ‘56.“!‘v V 1‘ 4.1 '1' 'J . «\ A -.a A L0“ ._: I r‘... \‘ 1 ‘ w: L ‘1 k: Q =13" . 20 to in the same section, was used for tests measuring the influence Of autoclaving (Table 1), duration of operation (Table 2), dialyzer posi- tion and flow direction (Table 3), membrane hydration (Table 4), mem- brane lots (Table 5), and temperature (Table 9 and Figure 13). These trials were reported in a previous thesis (H.E.B. Humphrey, M.S. thesis, University of Michigan, 1965) and are included here in order to provide complete documentation of dialyzer performance. This work was not re- peated with the current model dialyzer, which utilizes molded sili- cone rubber separators, because separator selection would not appreci- ably alter the conclusions derived from the above trials. Each experiment involved filling the fermentor and reservoir with their respective fluids (in a 1:3 ratio unless specified otherwise), assembling the dialyzer, connecting the necessary conduits, selecting pump and mixing speeds, and removing an initial sample Of the fluids. The system was then set into Operation and fluid samples removed peri- odically and analyzed for glucose concentration. The quantity of glu- cose transferred by dialysis from the reservoir to the fermentor was determined indirectly by measuring the refractive index (5-M) or direct- ly by the colorimetric anthrone analysis. Glucose concentration is pro- portional (i .052) to refractive index, when no other solutes are present, from .1% to 1.2% (Figure 5). The anthrone method was employed for solu- tions containing less than .1% glucose or for Situations where interfer- ing solutes were present. Hydraulic pressure measurements were made by inserting glass tubes with water or mercury columns into the conduits near the dialyzer entry and exit taps. The columns also could be connected via syringe needles inserted through the separator wall into an individual dialysis chamber. l.2*' 10“ h 8 20.5. (EZWOZOO : _ 6. “mOnVDJmV ~ 4” L. zuomml 21 I I I I I2? LO" “ E as a :2); .6— .. 8 .J (9 I- 9* " 5 i .2}. .4 0L 1 1 l l W 33320 53360 534(1) INDEX OF REFRACTION Figure 5. Least squares fit of the correlation curve for refractive index values (n85) and glucose concentrations. [4 ){easurez area bei The dialyzer (35.3, in C‘s/Z: for the fined a 5111;: 22 Measurement Of column heights indicated pressure gradients across the area being tested. The parameters used for evaluating glucose dialysis rates in this dialyzer-dialysis system were the half-equilibrium concentration time (ETSO’ in minutes) and the calculated permeability coefficient (Fm, in cm/min). These were determined from glucose concentration data for the reservoir and fermentor regions during Operation, and are de- fined as follows: ET50 = Number Of minutes required for the solute concentration in the fermentor to reach one half of the equilibrium concentration for the whole system. ’6 II The coefficient describing the permeation (dialysis) rate for a particular solute and membrane material within a system of defined volume and interfacial area. It is related to the ET50 by: ln 2 '1! II (1/v1 + 1/v2) Am ET50 Where V1 and V2 = the fermentor and reser- voir volumes reSpectively and A,m represents the dialyzer membrane area. The ET50 notation is a convenient means for evaluation of the dialysis process because it accounts for membrane factors, solute gradients, bulk fluid flow, and fluid turbulence and because it is easily detemine tr refer :atiar first. t‘. the di 23 determined. The ET50 and Fm notations apply only to the diffusional transfer of solutes as determined in this section. These could be develOped for application in a fermentation system if factors con- sidering the rate and extent Of cell growth as well as growth inhibi- tory product formation were included in the basic equations. For each dialysis experiment an equilibrium and half-equilibrium concen- tration are calculated for the system.and the solute receiving vessel from the concentration of the initial sample. Samples removed during the dialysis run are analyzed for glucose concentration. A time plot is made (Figure 6 is an example of such a plot) for each experiment from which the rate Of solute transfer can be determined and reported as the ET or the calculated Fm value. The influence of an imposed 50 experimental variable on the solute transfer rate between the reser- voir and fermentor is reflected in ETSO and thvalues. Diminishing ET values indicated more rapid transfer and increasing Fm values 50 indicated superior efficiency. « r.‘_“ C “v“. ‘y A C. ‘- F 1‘ 4-3 339.122 The process employed in all Of the experimental work was equilibrium dialysis. This involved continuous circulation of the respective fluids through the dialyzer. The ongoing transfer of solute produced a contin- uously diminishing concentration gradient. Therefore, both the mass and concentration Of solute gradually increased in the fermentor and decreased in the reservoir until the system approached a concentration equilibrium and point (Figure 6). The illustration, a typical dialysis trial, Shows an asymptotic approach to the quilibrium condition and a greater total mass of solute present in the reservoir than in the fermentor, through- out the process and at the end point. This reserve quantity Of solute, a consequence Of the larger (3.3 times) reservoir volume, represents one of the advantages dialysis holds for biological applications as will be shown in a later section. Each dialysis trial was analyzed for glucose transfer to the fer- mentor, and the data were plotted as illustrated in Figure 7. As Shown, the ET50 (half equilibrium time) value for the trial was determined by extrapolation from the calculated half-equilibrium concentration for the system in question. The dialyzer was equipped with one membrane and the dialysis system Operated at: 2 liters/min liquid flow rates; 1 to 3.3 fermentor to reservoir volume ratio; 1% initial glucose concentration in the reservoir; and 300 C temperature. The mean ET for twelve 50 duplicate runs was 144 min i 15 min over a range Of 110 min to 170 min. 24 24.5 Figure 6. Correlation of glucose concentration in the fermentor and reservoir with time during an equilibrium dialysis trial with a 300 ml fermentor and a 1000 ml reservoir. (G) [33 A .‘3 ’3 ‘ fi ( .,LICL’SF 9*" AL I \l h.) ("J 9.4 9.0 82 § 7.8 U) (I) § 7.4 8% 2.0 O O D -..I 0 LG L2 Figure 25 .- EQUILIBRIUM RESERVOIR 50 FERMENTOR 6. I I00 1 1 200 300 408 MINUTES E QUILIBRIUM GLUCOSE 7: h .0 I II IIIIL S... 1.13.1.4 2. .3.» A; 31w «30.8. ...: 4. 26 80 50 7. GLUCOSE EQUILIBRIUM 30 ETso f c, l 1 If 1 l _l -h 0 20 60 I00 I40 I80 220 260 3C1) MINUTES Figure 7. Example Of a glucose dialysis rate plot used for ETSO determinations . 27 The mean Fm'was 4.69 x 10.3 cm/min i .47 x 10.3 cm/min over a range Of 4.25 x 10.3 to 5.68 x 10.3 cm/min. The Operational conditions given above and the mean values Obtained for them represent the standards against which all subsequent data in this section were compared. The influence of steam sterilization (Table 1), length Of Opera- tion (Table 2), and fluid flow direction within the dialyzer and its position during Operation (Table 3) were evaluated. These manipula- tions did not change the solute dialysis rates within the standard de- viations Shown above. Membranes should be hydrated for at least ten minutes before use (Table 4). Differences in dialysis rates were noted with different manufacturer's lots Of membranes (Table 5). Enlarging the dialyzer membrane area resulted in superior ET50 values but poorer efficiency (Pm) until an Optimum area was reached (Figure 8). For the 1% glucose system, 7-10 membranes (.202 to .288 m2) produced Optimum ET of 40 minutes (for a 1:3 fermentor to reservoir 50's volume ratio) and 60 minutes (for a 1:1 ratio). Further addition Of area did not significantly change these values. Increasing the dialy- zer area also increased the internal volume and fluid holdup time (Table 6). Predicted and measured holdup times did not corresPond at the largest areas. A leveling trend in dialysis rate was Observed as the liquid flow rates through the dialyzer were increased (Figure 9). With a single membrane in the dialyzer, a significant reduction in ET50 (from 268 minutes to 226 minutes) occurred as the flow rate was increased from 1/2 L/min to 1 L/min. Above 1 L/min, the ET did not change Signifi- 50 cantly. Additional membrane area lowered the overall values but did not change this trend. Fluid turbulence, measured by visual Table 1. Influence of steam sterilization on glucose dialysis. EXTENT OF AUTOCLAVING ETSO Pm ( Hours ) (‘Min ) ( Cm/Min ) 0 150 .00453 2 166 .00426 4 156 .00435 6 158 .00430 Table 2. Influence Of duration Of dialyzer Operation on glucose dialysis. LENGTH OF OPERATION ETSO Pm ( Hours ) ( Min ) ( Cm/Min ) 0 140 .00472 24 122 .00557 70 145 .00468 109 154 .00441 135 150 .00453 Table 3. Influence Of dialyzer Operating position and fluid flow direction on glucose dialysis. COUNTERCURRENT FLOW CONCURRENT FLOW DIALYZER POSITION ET 5 0 (Mi n) Pm (Cm/ Min) ETSO (Min) Pm (Cm/Min) Horizontal 148 Vertical 152 .00459 .00447 152 .00447 148 .00459 Table 4. 29 Table 4. Influence of membrane hydration time on glucose dialysis LENGTH OF HYDRATION ET50 Pm BEFORE USE ( Min ) ( Cm/Min ) 3 Minutes 194 .00350 4 Minutes 182 .00363 10 Minutes 139 .00488 2 Days 142 .00480 * Table 5. Influence Ofmanufactured lots Of membranes on glucose dialysis MEMBRANE LOT ET ( Min P Cm/Min L0 ) 43 ) June 1964 135 to 160 .00503 to .00425 September 1964 160 to 180 .00425 to .00377 * Both lots were Visking regenerated cellulose membranes ( Union Carbide Corporation ); regular type, .003 ,I‘ pore diameter. . a. A LO '- «2.2» . .. ...F m 30 TOTAL MEMBRANE AREA (M2) Q .144 . .432 360 T 1 as I ._ 280 ‘- - .004 2 2 \ - .003 2 8 2 O. - .002 O 5 IO I5 NUMBER OF MEMBRANES Figure 8. Influence Of the number of membranes on glucose dialysis. Fermentor to reservoir volume ratios Of 1:3 (curve A) and 1:1 (curves B and C) were used. 31 0H Hm osofl owns. mm 0H Hm mas owns. OH 0H NH cam ossH. m w s end memo. H a N no o o Aoomv Aommv Aazv Amzv geese: anam menace angm mmzammzmz maezmszmmxm amaagaogau mngo> mumsnaun umuhfimwc co mocmunama wo Hones: mo ooaoaamnH .c manna .22).: 00L: Wu 26:3 32 *' I F I 7 I“ — 280—- I; g: O 240-- If) i_. LLI A 200- A “0005 O ‘.004 O - .003 LL 1 J _ I I _._a 0 I 2 3 4 5 ‘ FLOW RATE (L/MINI Figure 9. Correlation Of glucose dialysis with the fluid flow rate through the dialyzer. (CM/MINI PM 33 Observation of gas bubble movement within the dialyzer, first occurred at 1/2 L/min and continued to the maximum flow of 5 L/min. The pres- sure drOp (AF) across each Of the dialysis chambers adjacent to a mem- brane rose, but not identically, as flow rates were increased (Figure 10). This increase in pressure gradient across a given chamber was small (0-80 cm.H20), was reduced as additional membranes were added to the dialyzer, and was Similar at flow rates of l and 2 L/min when five or more membranes were present in the dialyzer (Figure 11). However, the pressure drOp for 5 and 10 membranes differed when flow rates greater than 2 L/min were used (Figure 12). The pressure drOp across individual chambers was less than across the whole dialyzer, but the response to applied conditions was similar for both. The temperature under which the dialysis system was Operated had a nearly linear effect on glucose dialysis rates, which rose with tem- perature from.4'C to 30.0 (Figure 13). The best transfer and efficiency occurred at 60°C. When the initial reservoir glucose concentration was increased above 10 g/L (1%) Slower dialysis rates and poorer efficiency were Ob- served (Figure 14). The diffusivity Of glucose declined as concentra- tion rose (Figure 15). In contrast to ET values, the initial dialy- 50 sis rates (over the first 100 minutes) rose when the same increases in concentration were used (Figure 15). The dialysis systemmwas easily enlarged by use Of larger vessels and/or liquid quantities. The resulting total volume increase, from .6 liters to 6 liters for example, resulted in higher ET and Pm 50 values (Table 7). Although the initial concentration (10 g/L) was kept constant, an increase in volume caused an overall increase in the total “I {'73. ’CJ '1 A" w. (MM H60) AP LOG Figure'lO. 34 I I I I I 5. I t I .J O TCP SIDE A BOTTOM SIDE I .A l___ - h ,L -1- 1-...-- LAM“ ---- L I_' I 2. .3 4- E> ES FLOW RATE (LIMIN) Correlation between fluid flow rate through the dialyzer and pressure drop in the dialyzer chambers adjacent to the membrane. “G ..I h... C V C 6‘ .1 a: at; Q. P... v A C a 35 I \ O —- I 1-x:~..t;t~..=. erow ._ h (D N we 4- w is" 20*- '7 t) Q. q * ‘ ‘ “WC? ‘ “" ““V I0— "‘ J L L 5 I ’f‘ I K} 01 . :«o’sr‘rfi .r‘v'" ,c-‘e‘gjf‘x I. If I‘-IL-'I‘-.'z::.r_t'-. r2.” fatLtea vii-«IN Q MI. 5.“ Figure 11. Influence of dialyzer membrane area on the pressure drOp across a given dialyzer chamber at fluid flow rates of 1 and 2 L/min . 36 I I I I 5 "- '1 0 0 4 —— __ $2 5 I'-.IEIe'iE§?;lI\-EES A. .9 CL <1 2 -—- o , ._ g If} Ic'IEI'rd\3I\r u If?) 0 a/ I - ... Q _I __ I _ - _ _I I O I 2 3 L 4 FLOW RATE (LN/SEN} Figure 12. Influence Of flow rate through the dialyzer On the pressure drOp across a given dialyzer chamber with 5 and 10 membranes installed. 263‘. a va.:».. n0 T. eh. A.“ n K V\' Figure 37 340 .- I I r I I I 1 fl , ° - .006 260 - _ 0 £3; 3 IE 0 r .004 L0 {—1 LL: I80 _ I—. '1 .- q loo — a ‘c .002 L I ' 1 L, l 0 IO 20 30 40 50 60 TEMPERATURE Figure 13. Influence Of Operating temperature on glucose dialysis. PM I «CM/Mam r. .. I x. 0 ..q no. .... «a.» PVC «2.3: 0...: I; 38 I __ I I 340‘ 300 " 2 E O ‘0 260- I" - 004 LLI 63 ° 220“ .. . x0 .uf J!— I I I 0 50 I00 I50 'LUCGSE CCI‘JCE? 7:77;}. 30?: (133‘...) Figure 14. Influence Of initial reservoir concentration on glucose dialysis. ‘ ' PM (CM/MINI ....C. x 0.1.4.. $3.0. >F.)‘V.Dli.k‘6 initi 39 .68 1 I :q , 4(3 0‘ .60 '- J; L) ll) ‘1’ “k 9 o _ -l a: C) 3 .52 - t7) 0 :3 It O 5 --fi2 fM4a- . L I I. O 50 Im I50 GLUCOSE CONCENTRATION (GIL) Figure 15. Influence Of concentration on glucose diffusivity and initial dialysis rate. (WW I LOG INITIAL UALYSIS RATE ommoo. owe 00.5 m.cm oooo ooom ooom 40 waqoo. «Na mm.m H.¢H ooom oomH oomH mflqoo. omu m¢.N n.m ooom oooH OOOH ommoo. «mm ow.H N.n coma com com ammoo. oNH ow. ~.m com com com sawunwfiwavm mHm: sommm — Aawzxaov Aawzv OH OO>HOOOM omoonfiw va hfifimwuHaH Omoomaw Amy Hmuos ufio>ummom uOucOEuom am omam moezaémm mHoEmmmm , 35 $542, .mmumu mfimhamfiv mam nommcmuu mmms Omoonfiw no Eoumhm mfimhflmwm msu mo m55~o> Hmuou can no mononamaH .n manna 41 mass Of glucose initially present in the reservoir. This resulted in an increase in the mass transfer necessary to reach half equilibrium. Manipulation Of vessel size or liquid volumes also made changes in the fermentor to reservoir volume ratios possible. A low ratio (1:1) and small overall volume (600 m1) produced the fastest dialysis rate (126 min). ‘When the ratio was increased to 1:10 by enlarging the reservoir, the ET values increased markedly. Further ratio incre- 50 ments produced no Significant change in the values (Figure 16). The mass of glucose present in the reservoir both initially and at half equilibrium corre8pondingly rose as the ratio was increased. However the mass transfer necessary to produce a half equilibrium concentration increased up to a 1:10 ratio but not above (Figure 16 and Table 8). Thus the dialysis rate and necessary mass transfer to the fermentor corresponded over the range of volume ratios tested. ("I r: (L) 42 0913 Hovas OJ. I9)I:B.-.ISNV81 ssvw «2 uomom _uoucmaumm on OH wm>wmoom Omoosfiu AUV omoonau on em moezaame «853mm SEE :5 ESQ? .mmumu mwmhflmwm new ummmcmuu mmma omooafim no mowumu oEaHO> uHo>uommunounuouaoauom mo mucosfimmH .w oHan II .1 ' L‘a:l\ , . .G S I. S r n . Q. A. . Au ad a... a: G E ”a c. a. E l U E U .. . E .2 D. S .2 .L 7.. a u ... a L. a r 1 10a .. t ...... aL a Co NC 3 «C C C. . e :1 C d a u 4 0 AL 3: ..1 .. . v . 0 a: a: .... a. u 4.. . ?. 2. sud ..& LE 11 .2 v . ha. I. b ...J o . D G. .. a a he. .4 P hi ...t a 4.4 Discussion The dialyzer and the dialyzer-dialysis system were designed for cultivation Of microorganisms. Because Of this Objective, solute dialy- sis was evaluated under conditions resembling, as closely as possible, those necessary for growth Of bacteria or allied with microbiological processes. For example, a temperature Of 30’C, circulation rates of 1 to 2 liters per minute, and Visking regenerated cellulose membranes have been shown to be satisfactory for growth of Serratia marcescens and other bacterial species (57, 61). Consequently they were adOpted as standard testing conditions. Similarly, glucose was employed as the test solute because a fast assay method was available and because this sugar is a re- quired growth-limiting nutrient for most bacterial species. Glucose is included in both a chemically defined medium (at 2% concentration) (145) and in Trypticase soy broth (at .25% concentration), both Of which have proven satisfactory for culturing the standard test organism, Serratia marcescens . Glucose dialysis was reported as ET 0 (half equilibrium time) and 5 Pm (permeability coefficient) values. The ET50 notation was used pre- viously (61, 34) as a means Of evaluating dialysis membranes. The con- cept originated as a factor for indicating electrolyte penetration of algal vacuoles (31) and was used as a coefficient for evaluation of transport in artificial kidney devices (69). The ET50 designation repre- sents a meaningful criterion for evaluation of dialyzer performance but 44 corn :hro volu P (g) In 45 is limited by dependance on the system volume and membrane area (141). The overall permeability coefficient (Pig'which includes the above and all other performance parameters was developed and adapted in order to provide a second and all encompassing factor for evaluation Of the dialyzer. The overall coefficient, Pm, was derived (J. Schultz, personal communication) from Manegold's formula (109) for diffusion of solutes through a membrane film.and relates solute concentration, time and volume to permeability: . c1 (1/v1 + 1/v2) - (cl/v2 + c2/v2) 1n _ _ . — (1/v1 + 1/v2) Pm Am t, (4.1) c° - 0° 1 2 V1 J Where C1 is the solute concentration in chamber 1 (fermentor) at any time t, C? and C3 are the initial solute concentrations in chambers 1 and 2 (fermentor and reservoir respectively), V1 and V2 are the volumes Of chambers 1 and 2, and Am the membrane area. Manegold's formula was develOped to measure the permeability of membrane materials and was based on the transfer Of an identifiable solute from one Side to the other in accordance with the basic laws of diffusion. It was assumed that A represented the effective membrane surface available for solute transfer, the chamber volumes were constant, the fluids were well mixed, the initial concentrations in chambers 1 and 2 were known, and the con- centration in one Of the chambers could be measured as a function Of time. If the concentration in one of the chambers (fermentor) is initially zero (C? = 0), the left-hand side Of the above equation 4.1 is simplified so that 1n (1 - cllce) = - (1/V1 + 1/v2) Pm Am t ; (4.2) 3: :3‘ (1) pt) (I I m P“ ‘i 3 H! m n (7" flat; £13 £51: absE alt, OVEI 46 Where Ce is the final equilibrium concentration in both chambers, cv c = 22 (4.3) It“??? When the system reaches half equilibrium, t = ET50 and C1 = Ce/Z, and the left hand side Of equation 4.2 becomes 1 - g/z C e 1n = 1n 1/2. (4.4) Equation 4.1 then is written: 1n 1/2 = - (1/v1 + 1/v2) Pm Am ET (4.5) 50' Rearranging: (4 6) 1n 2 1n 2 ET50 = or Fm = (1/v1 + 1/v2) Pm Am (1/v1 + 1/v2) AmET 50 Thus, two criteria exist for reporting solute transfer from exper- imental data: an ET value encompassing membrane thickness and material 50 factors, fluid flow rates, turbulence, and solute concentration; and a Fm value accounting for the volume and membrane area. These provide a basis for evaluation, modification, and scale up of the dialysis system. Design of the dialyzer for operation with cultures implied that sev- eral important criteria be met: compatibility with metabolizing cells, sterilization and maintanance of sterility, consistent performance over long periods of Operation, freedom to place the instrument anywhere with- in the pilot plant, choice Of fluid flow direction within it, and infor- mation concerning membrane treatment and consistency. The results veri- fied that these conditions were satisfied. Moreover, the durability and reliably consistent performance Of the dialyzer was demonstrated by the absence of equipment deterioration by steam sterilization and by the un- altered solute transfer characteristics during continuous Operation, for over six days. These were important because steam represents the only dependable method for biological sterilization and the continuity Of s, H "I .O.A 5 ..-..x ...»...v .1 all. .P‘ a a I .»5 .1‘ bl» u on C l .D . «J 3. In 1” S D. a a. .... C. a u.” . L ~ .. . E E Q. 0 C .5 A” .: .. . .. . a a. n .I. .u I.“ a .... .1 :1 at; P a a: FIT use .4 .l a y C. ..IA firm 6L t c» I 47 Operation is essential for microbial propagation in either batch or continuous culture. Independent control and placement Of equipment components, eSpecial- ly the dialyzer, is a valuable and convenient asset of this culture sys- tem. Historically, dialysis exchangers were constructed with vertical membranes and chambers to take full advantage of the counter current cur- culation produced at the membrane surfaces by gravity Stratification Of the solutions as they moved through the exchanger (36). This scheme main- tained the maximum concentration gradients necessary for effective dialy- sis but limited bulk fluid flow and reduced the freedom Of exchanger placement within the system. Counter current flow, which maintains a maximal concentration gradient linearly along the full surface Of the membrane, facilitates efficient mass transfer. This has been achieved in hemodialyzers positioned either horizonally or vertically by fluid pumping (101, 102). The physical characteristics of a microbial culture, such as gas bubbles, viscosity, particulate nutrient suspensions and sed- imentation Of cells, may supersede the importance Of maintaining maximal mass transfer conditions and will influence dialyzer placement and Opera- tion. For example, aerated broths containing air bubbles require a ver- tical fluid flow to prevent entrapment Of the bubbles within the dialy- sis chamber. The flexibility Of the dialysis system facilitates satis- faction Of such demands without jeOpardizing solute transfer performance. Visking regenerated cellulose membranes provide good glucose dif- fusion rates, 8-10 grams per hour (167), are sufficiently durable, and were shown superior to other membrane materials for dialysis applications (61). Preparation and selection Of these membranes was necessary for con- Sistent performance within the established range. A minimum hydration 48 time was necessary and differences were Observed between manufactured membrane lots. Although these variations were not Of a great enough magnitude to affect the dialysis transfer Of solutes in principle, they were sufficient to influence the precision Of individual experiments or groups Of comparable experiments. Therefore the membrane factor must be recognized and accounted for during dialysis investigations. Independent control of the dialyzer affords a means Of increasing dialysis capacity. This was accomplished by addition Of dialysis cham- bers and membranes to the instrument. EXpansion of membrane area per- mits adaptation Of the dialyzer to a fermentation system of any desired size. If an excessively large area is required, two or more dialyzers may be added in parallel. Addition Of membrane area (.0288 m2 for each membrane) improved dialysis of 1% glucose solutions until an Optimum was reached above which no substantial improvement was observed. This effect has been Observed before (D. M. Gallup, Ph.D. Thesis, University of Michigan, 1962) and represents the point at which membrane area ceases to limit dialysis. Seven membranes were Optimal for a 1:3.3 fermentor to reservoir volume ratio (1.3 liters total volume) and ten membranes for a 1:1 ratio (6 liters total volume). Because the second system had a larger volume it required a greater mass transfer to reach half equilibrium (7.6 g vs 1.8 g, see Table 7) and, therefore, required a larger area to achieve Optimum dialysis. These results are in accord with Fick's dif- fusion law which states that solute transfer rates are prOportional to the exposed membrane area. ‘Moreover they substantiate the principle that membrane surface should be as large as possible in relation tO the volume Of liquid to be dialyzed (36). "IV... a: .d .I . AL A... b - n31? a F. a Q ' L‘Qn -‘ n 49 The surprising decrease in dialyzer efficiency with additional membrane area was believed tO be associated with a change in internal fluid dynamics as dialysis chambers were added. Enlargement of the dialyzer increased the internal volume and the fluid holdup time as measured by the time required for an injected swarm of bubbles or dye to clear the dialyzer (Table 6). However the holdup times did not cor- respond directly with the area used. Bubble swarms began tO clear the dialyzer within 4-5 seconds regardless of the area. It was concluded that a given volume Of fluid was not distributed evenly among all of the dialysis chambers and therefore was not only unexposed to some Of the available membrane area but also left the dialyzer sooner than ex- pected. This effect multiplied as chambers were added, resulting in re- duced efficiency. Moreover, the increased internal volume associated with additional dialysis chambers also reduced efficiency by reducing the flow rate across a given membrane surface. This premise was sub- stantiated with a 15 membrane dialyzer where an increase in Pm from 2 x 10.3 to 2.7 x 10.3 cm/min was observed when the pump rate was raised from 2 to 6 L/min respectively. A similar reduction in efficiency with addition Of membrane area has been attributed to unequal flow through the separate chambers in hemodialyzers of similar design (52). The influence Of Operational conditions on solute dialysis was analyzed. In particular fluid velocity, temperature, solute concentra- tion, and fermentor and reservoir volumes and volume ratios were examined. Fluid velocity past a membrane surface generally determines the thickness Of the laminar film immediately adjacent to it. In addition, fluid velocity through a dialysis chamber governs the degree of turbulence and bulk mixing within the vicinity of the membrane. The fluid dynamics and . . . II i. w A «E. E .5 a 3 E r r C, a .c A-.. .d e n r 1C h 2.. I. . .D a L t t EL 5 I a n a L O fin r1 .: 2H t. 3. n4 .t ha I In“ .C «C .w a ...-.. n... .1 a .. ..L ....I... 0 .I fit. ...- _..._ I . Au t .wL I. t ‘4 sL VA 0 \St: He P313 “31‘ Id Co; at L! t 50 within the dialyzer are depicted in Figure 17. Both separator design and fluid velocity contribute to Optimization of the above conditions. Adequate mixing maintains the solute concentration gradient and thin laminar films reduce membrane diffusion resistance. Both of these aid dialysis transfer rates. The success of separator design was Shown by Observance of fluid turbulence at fluid flow rates Of 1/2 L/min or more. Achievement Of Optimized fluid dynamics within the dialyzer was Shown by stabilization Of ET and Pm values at fluid velocities of 2 L/min 50 or above. Increments in pump rates increased the pressure drOp across the whole dialyzer and across the dialysis chambers on either Side Of a membrane. The Observed pressure gradients were relatively small, simi- lar for both chambers, and similar in chambers positioned adjacent to or distant from the front plate Of the dialyzer. These results suggested that hydrodynamic conditions were similar for all dialysis chambers with- in the dialyzer and that the dialyzer presents an efficiently small re- sistance to fluid flow. The hydraulic gradients, although analogous to those Observed in smooth pipes, could not be used as criteria for deter- mining turbulence within a dialysis chamber. The fluid dynamics within the rectangular dialysis chamber were not the same as those within a round pipe. Therefore visual Observation through the glass back plate of the dialyzer (as indicated above) was used for this determination. The hydraulic pressure gradients diminished when membranes and dialysis chambers were added. This was attributed to the expanded internal volume and constant pump rate. Resoration Of the gradient values to those charac- teristic Of the dialyzer with a few membranes was accomplished by raising the pump rate. The resultant improvement in dialysis efficiency (see ————e> I I W m I I III ' I! ’I ... .. --- _. - _-. 2...... .. ..---- --.- ...—-...-. -- ...- .... .. ---... --.-.- ....— ._... -_. --.“... --.---.. _ _ ... .. ...“... _._ _.,..- .... ....-. _ .___._- _._-.. -_.... ... -- .. _- .... ..-- -....- .... _- ... ._....-.._. .. _ . -- - ...- --- .. L. - ...—.....- ...- .. ...—...- ._ -..- -... \--...... .__. ._---. 5.. -. - .. .. ....-- _ ...-.. ...—.... _ - .. - .. .. ———.-—_._.— - .— ._--'—-_-. _. --.. —————_— ....___—. .———..._.__ — ...—a u-.....---..—... _.~ ..-...- v— m- .._ u. .4. '—.__...—.___..._ - _—- -—.__.’-_-_r._.. _.__. .——.—.—.~-——. _. .— --..-” —- ... c ...—...... __ ____ .- .—.‘_~--—. -...__ -- ———..-__ ..- -.---- _ .———- -..—.- .--. -, -~- -——-.__—_.._. ..-—-.— -W— ...—— m .—---—__ ~-___.__-- .» -——- .—.—-—.___.—.-- —. _. “...--.-.-fi- — .— ..——~-—...--.~.._.-. ..-— .._._.._.-— -- .- .--—— .——..—.__—_-_. ..—.—. -—.—- "—4.“..— h--. .— .———u-—_. -— -_ .7-. .- ———.—-—- ,_---....— ”..—__. .—_.-~ _ ..fi......-..-.-... —.— -__..-_--._ -... ..——..__-—.-_ —- .- ...— .-——__. -_..-._—. .—-——- --. ......— - ——.—. - ...... —_— —.-..—— -..—L - -.. _ _ —v ~... --—~_._.-_ . .— -.—-- _ -....— _. ——_—_ ..— 0.- -..-— a-.—..___——— ....-— _-_-—~..—_.— -___, .--—4-.--.- __ .- —— - -.--.._._ hm _._ _- —— ... - _ .______ ”h... . - ..- -—-—--.~— M I<————>I MEMBRANE IS->I FILIA Figure 17. Representation Of fluid flow past the Opposing faces Of two membranes, with a laminar film immediately along each surface and turbulent bulk flow in the central channel. and Gerhardt (L41). I, |\ TURBULENCE 51 MEMBRANE Reproduced from Schultz ...—.-.--m _-_._ --_—»-—--_.o ”a -. __.~._ _.—__.. n... -.-r—._.—.. — — ...—.--- "_“-——— —— —- .---.—.——- ._ -. ...-.H” . .-.._.. .~ _. ...“..- .~-— .— -. _ --— .. .... .. -..-— — - .- .- .. ..- .—.—.—. ..——_ -..— .. -_ ..~ _ .. m*—. —-———-. ...—.- “ -.._ ---_ .. -.- -.“ .-- --.—.———-— \- - ---—o— ... -_ .. . __..__, ... -4--. _.—. _~- -.——~ ...—... --.. .._.——.- . ...- «.-.-”-..... - ... .... “-..—..- —--_~——_—‘_ —— —-_ ——-.— -..—..— -- .-— _._-___~‘_ —. —’ —.—.———- .— -_ ..-. “ugh“ -- ..- ... __-~_... ____--__._. ......— -_. -- -._ -__.__.--__.- .._- -—_...——._.—___ o — .. -..... --. __ .—->____. ..- -.. -m ._ ...-“ _._— —_ .__. ---i...._._.. —--.———.-————~ ------___——.......__ ——-. ~- _— -——. -. -. -- ...- — _---.-—-__ ...~~— -- _ ‘-- ..-—— .— - - -———.-——. q ——-— ...-... ...—~— —_ *wm—_ __...._.— ...--fl —— _ -_—_ ___ —_ _.-._. ... -- --—-<—-——— ———-—_—-— _.. --.— -_----._q. .—__._.____——_ - _ — -———— ._.___,,___-.______.___ .b— M¢4—~~.¢ -——__q. fl-.H———-— —___._ ...._ _.—-— ..~ , ~.— —- _. ..._. - _ m.- ..— ILhfl WHILE '.I H..TI a 52 the example cited previously) indicated that efficient internal fluid conditions could be maintained by manipulatiOn Of the fluid flow rates within the dialyzer and that the expanded internal volume associated with additional membrane area could be compensated for. Dialysis efficiency was closely related tO the temperature Of Opera- tion. An increase in dialysis rates with temperature was expected since diffusion is directly related to temperature (21). Warmer conditions produce greater molecular activity and reduced viscosity. Both of these increase glucose diffusivity (see Table 9), which explains the correla- tion Of dialysis performance with temperature. Thus the dialysis culture system would achieve its best efficiency when used to propagate thermo- philic organisms. 0n the other hand reduced dialysis efficiency must be considered if the same system were tO be used with psychrOphiles. Most Of the experimental data was Obtained at 30'C (meSOphilic Optimum) be- cause this group Of microorganisms are the most numerous and commercially the most useful. A biological growth medium may contain glucose in concentrations ranging from 1.5 g/L (Trypticase soy broth) to 141 g/L (synthetic medium) depending on the requirements Of the organism used or the metabolic pro- cess being explored. Glucose concentrations in this range were placed in the reservoir and dialyzed against a fermentor containing an equal volume Of water to demonstrate the capacity Of the dialysis system for these conditions. Initial dialysis rates (over the first 100 minutes) increased as the reservoir mass increased from 3.8 g to 141 g. This was expected since the rate of dialysis, like diffusion, increases as the concentration gradient across the membrane becomes greater (Fick's law). However, the half equilibrium time became longer and dialysis efficiency ble 7. .s LP 1...! she .. IN 53 Table 9. Influence of temperature on glucose dialysis and diffusivity. TEMPERATURE ET50 Pm DIFFUSIVITY (0 c ) (Min) (Cm/Min) (10‘S sz/Sec) 4 331 .00205 .365 16 233 .00291 .540 30 144 .00472 .793 54 114 .00596 1.326 A\~ .. I ... u 1. ~ . S 111 t . 4 . . .3 . . .C ..C .. . .3 n... a U at D. e 5 .C 11 .5 C U S w. I fit : . .. L 0 .... ~ . S 2. e a . .C L H J. an n hr” .6 a .C S 8.. .. a .1 ..z.. 3 O a . .04 a T. I .3 4L 4 a .... 4 r .v. :4. a» i. ..l .6L .1... v 4 t 5.1 P. 54 poorer as the concentration gradient (reservoir concentration) rose over the same range. This had not been anticipated and suggested that increased reservoir concentration produced apparently Opposing effects on dialysis. The larger glucose concentrations increased fluid viscosity, decreased solute diffusivity, and increased mass Of solute which must diffuse across the membrane to achieve the half equilibrium concentration as well as the increased concentration gradient. These effects are tabulated in Table 10. Thus it appears that, at the onset of dialysis, mass transfer was rapid in response to the solute concentration gradient. As dialysis proceeded to- ward equilibrium the gradient was reduced and its influence on mass trans- fer was lessened, allowing the influence Of diffusivity, viscosity, and especially required mass transfer to prevail. The reduction in dialyzer efficiency resulting from the greater required mass transfer associated with increased reservoir solute concentration could be improved by addi- tional membrane area. Independent control of the fermentor and reservoir regions permits expansion Of the dialysis system to accommodate any desired Operating volume. 'When these were increased in 1:1 ratios, the total system volume rose and correspondingly, the reservoir solute mass as well as the mass transfer required to reach the half equilibrium concentration, rose. The latter increase eXplains the Observed reduction in glucose dialysis rates as the total volume rose from .6 L to 6 L. The corresponding in- crease in dialysis coefficients suggested that the dialyzer remained relatively efficient in transferring the additional volume and solute mass involved. Again the addition Of membrane area, which would main- tain the necessary area to dialysate volume relationship, would improve performance in large volume applications. 55 owe. oaH. ommoo. omm oo.mm o.HsH ohm. mmo. mmwoo. emu om.mH m.~o umo. owo. mHmoo. sow oo.~H s.om one. mac. somoo. New om.o s.s~ mam. Hmo. mmmoo. mam os.m o.mH Nos. 600. mmmoo. NHN ma. m.m Aomm\~su m-OHV Acaz\uv “cazxaoV Aamzv Auv Amy . s cm zDHMmHquom Esau a< mmaz ~9H>HmsmmHn maam mmamzams Awmsmmwb .mmumu mwmhfimwp mmoozaw co :OHumuucmOcOO uwo>uommu Hmfiuwcw osu mo mocmsflmaH .oH wanna 56 The ratio between fermentor and reservoir volumes was increased by increasing the latter while the former was held constant. Because the reservoir region would be the simplest and least expensive region tO ex- pand in a cell prOpagation system, the ratios were increased in this man- ner allowing retention Of the same fermentor in each case. This permitted analysis Of the ratio effect on a key feature Of dialysis culturing: the concentration of culture by confinement within a limited volume which is possible only by exchange Of dialyzable nutrients from a reservoir. Dialysis performance became decidedly poorer when the ratio increased from 1:1 to 1:10 but remained constant at larger ratios. The explanation again resided in the mass Of solute which had to be transferred through the dialyzer tO attain half equilibrium. For 1:1, 1:3.2, and 1:10 ratios the required mass transfer rose from .8 g to 1.4 g (Table 8). However, ratios above 1:10 showed no significant increase in required mass trans- fer which was reflected by the constant ET50 values Observed at these ra- tios. This effect was due to the fermentor volume, being constant, becom- ing an increasing smaller portion Of the total system volume as the ratio rose by increasing the reservoir volume. Thus, half-equilibrium could be established in a 1:30 system as quickly as in a 1:10 system because only 2% to 4% of the reservoir mass needed to be dialyzed. A 1:1 ratio in constrast, required 25% Of the reservoir mass to be transferred. Since the reservoirs Of large ratio systems retain 96% - 98% Of their solute mass at half equilibrium, a large reservoir Of additional solute is pro- vided for transfer into the fermentor if equilibrium conditions are upset, as occurs during growth. Comparing Tables 7 and 8 reveals, as just shown, that the total volume effect predominated over the volume ratio effect. For example, the 1:1 system with a 6 L total volume required 680 minutes In ’LJ 57 to transfer 7.6 g Of glucose while a 1:18.5 system with nearly the same volume required only 270 minutes tO reach half equilibrium because a transfer Of only 1.4 g of glucose was required. Thus the system with a larger volume and the larger required mass transfer will either have slower performance, or will require additional dialyzer area to attain equal performance. These results demonstrate the importance of identify- ing the conditions under which the dialysis system will Operate, SO neces- sary adjustments may be made to attain the most efficient or effective performance. 4 . 5 Summary The previously described dialyzer was incorporated into a dialyzer- dialysis culture system which utilized a fermentor and separate reservoir Of 3 tO 10 liters and 10 to 30 liters, respectively. Nutrient transfer (glucose) for this system was evaluated by determination Of half-equili- brium times (ETSO) and overall permeability coefficients (Pm). Mean values Of 144 i- 15 min (ETSO) and 4.7 i .47 x 10"3 cm/min (Pm) were Ob- tained with a 1% glucose solution and a Single dialysis membrane. Con- tinuous Operation, autoclaving, dialyzer position, and fluid direction had no effect on dialysis performance. Optimum rates were Obtained with 2 an area of .2016 m to .2880 m2, depending on the volume ratio used, and bulk flow rates above .5 L/min. The decrease in performance associated with large overall liquid volumes or fermentor-to-reservoir volume ratios and the Significance Of internal fluid dynamics were discussed. 58 Ch ' U {H he Dyer. 5. GRWTH OF A TYPE BACTERIUM 5.1 Introduction and Historical 5.1.1 Introduction Microorganisms are traditionally grown in conventional fermentors when large yields are desired. Although these are relatively Simple to assemble and Operate the efficiency Of the process is low and the Opera- tional flexibility limited, precluding adoption Of new or novel culture schemes. The continuous culture technique improves the culture environ- ment within a conventional system but the costly control and monitoring equipment make it impractical for routine use. The dialysis culture technique can not only improve the culture environment but also increases the Operational flexibility Of a conventional culture system. The application Of dialysis to the culture Of microorganisms began with the use Of simple equipment. Seventy years ago it was discovered tthat: a pneumococcal culture confined within a dialysis membrane sac not Only grew well but exhibited enhanced virulence and toxin production (24) . Since those initial experiments, the dialysis culture technique has been widely exploited tO achieve results superior to those Of con- ventional fermentations. A historical survey of the literature pertinent to the evolution Of the dialysis culture technique and the major contribu- tions concerning its design, apparatus, theory, and application was re- Cent 1y compiled by members of this laboratory, including contributions by this author, and was published as a review (141). This survey, 59 .n\ It .I,‘ 60 including results from this laboratory, revealed that major consequences Of growing cells in dialysis culture were: the prolongation Of active multiplication to reach higher maximum populations which had high via- bility and Often extraordinarily high densities; the stabilization Of the maximum stationary and terminal phases; the production Of cells free from macromolecular constituents Of the medium; the removal and dilution of diffusible growth inhibiting solutes; the increased production and accumulation Of culture products such as extracellular enzymes, toxins, or spores; and the Opportunity to prepare and study unique interbiotic relationships such as symbiosis and antagonism. A surprisingly large variety of organisms have been studied in dialysis culture: 23 genera Of bacteria, 6 genera Of fungi, 6 genera Of protozoa and 6 lines Of tis- sue cells. ‘Most Of the investigations to date were made with the same basic dialysis culture design used in the initial investigation seventy years ago, a dialysis sac containing the culture suspended within a nutrient reservoir, or with small custom built dialysis chambers (141). These systems, while sufficient for Specific investigations, are inherently limited in capacity, capability for control, scale-up, and Operational modifications. The closest approach to a large scale system using the dialysis sac scheme was devised by Sterne and Wentzel (151) who used a large intussuscepted dialysis sac (3.5 liters Of culture) surrounded by 35 liters Of medium in a carboy to produce high yields of pure and potent botulinum toxin. Hedéh (75) produced high molecular weight extracellular products by using a 15 meter long dialysis bag, containing a culture which was circulated and aerated by an external pulse-aeration pump, suspended in a fermentor of nutrient. Both of these devices although _ a U " Culture Iuantitati‘ diaII‘Sis s The a fer-3.52:8“ in this th ture SE’SIE Pyrex PIPE nediun res or. a shaki measly 50 system V85 Embraces, the genera vith a var Sext, fermentatit by conduits coiled dial was 61 quantitatively large suffer from the above limitations of this type Of dialysis Scheme. The aim Of the investigations in this laboratory have been to im- prove the basic design Of dialysis culture equipment, to test the ap- plicability of membranes, to analyze the performance (chemical and bio- logical) Of the dialysis system, and to develop a functional dialysis fermentation system. Several important studies preceded those reported in this thesis. The first was the development of a flask dialysis cul- ture system (61). This twin-chambered flask is constructed of stock Pyrex pipe fittings, with a supported membrane clamped between the lower medium reservoir portion and the upper culture portion and can be mounted on a shaking machine. Several Of these units could be Operated simulta- neously for experimental duplication and control. The dialysis flask system was used to evaluate the suitability and diffusion properties Of membranes, to assess the variables of dialysis culture, and demonstrate the general usefulness Of the concentrating effect of dialysis culture with a variety Of microorganisms. Next, the dialysis culture concept was extended to a pilot scale fermentation system where the culture was remote from, but connected by conduits and pumps with, the nutrient reservoir (57). Initially coiled dialysis tubing and conventional fermentor vessels were used tO construct a fermentor-dialysis system, where the nutrients from a remote reservoir were circulated through tubing placed in the fermentor. In a reservoir-dialysis system, the culture fermentor was separate and the tubing was placed within the reservoir vessel. The third and best scheme was a dialyzer-dialysis culture system where the principle regions -- culture, reservoir, and dialysis -- were separated and could be an ad] maul 193353 wgiul vlwviInIiIunup 62 controlled independently. This system necessitated the use Of a dialy- zer. A plate-and-frame industrial chemical dialyzer, which used sheet membranes, was employed. The dialyzer-dialysis system demonstrated the potential and usefulness Of dialysis for large scale fermentations when growth trials revealed high densities Of Serratia marcescens cultures and Showed that Operational manipulations such as supplemental feeding and osmotic dehydration Of the culture for further concentration could be successfully employed. These studies also showed that the key to the successful Operation Of a dialysis culture system is a suitably designed dialyzer. The design, construction, and analysis Of such a dialyzer rep- resents the initial portion Of the current study, is presented in the pre- ceding sections (Sections 3 and 4), and was reported earlier (Abstract, 152nd Meeting, American Chemical Society, New York, 1966). The prototype dialyzer (Section 3) was incorporated into the dialy- zer-dialysis culture scheme in the experiments which follow. The Objec- tives of these investigations were: (a) to examine the suitability Of the dialyzer for dialysis culture; (b) to examine the influence Of Opera- tional variables on dialysis growth and to compare their influence on solute transfer; (c) to demonstrate and compare the characteristics Of dialysis and nondialysis growth; and (d) to illustrate and analyze some Of the attributes Of dialysis culture. The test organism Serratia marcescens was chosen in order to retain continuity with previous in- vestigations and to permit comparison. IL 5.1.2 means 5 ucts re 5.1.2 Historical The selective transport Of solutes through a membrane barrier makes dialysis a potentially useful technique for biological processes as a means by which nutrients could be made available tO and metabolic prod- ucts removed from a confined microbial culture. This was first recog- nized nearly seventy-five years ago when Metchnikoff, Roux and Salimbeni demonstrated the solubility of cholera toxin by implanting collodion sacs containing cholera vibrios in the peritoneal cavity Of animals (113). Shortly thereafter Carnot and Fournier (24) extended the use Of dialysis to an in vitro situation by suspending a collodion sac containing pneu- mococci cultures in a laboratory flask containing ordinary growth medium, also in an effort to demonstrate the presence of a diffusible toxin. Subsequently the dialysis principle has been employed in numerous bio- logical investigations including: a variety Of in vivo and in vitro schemes for prOpagating a large variety of microorganisms; the produc- tion, purification, and/or concentration Of both diffusible and nondif- fusible culture products; membrane testing; surface growth and filtra- tion; the study of interbiotic relationships and effects such as synergism and antagonism; gas exchange; and peritoneal and hemodialysis (141). In addition, the availability Of new types Of synthetic membranes with a greater range Of permeabilities has extended application Of the dialysis process into the administration of anesthesia (51), the prolonged admin- istration Of drugs (9, 50) and Of deficient enzymes (25) by capsule 63 implanI 0530515 culture tubes, ing dif of the culture free dj 1y re e (1‘! A thirg brane : (IO-pol1 leCOI: the se; rous me meanan Ge can be iuring 1 and is 1 SUre the bial CuI vitro die 64 implants in animals, and osmotic pressure filtration such as "reverse osmosis" for water purification (60). The membrane interface represents the key element of a dialysis culture system. The most commonly used schemes employ membrane sheets, tubes, or sacs which sequester the microbial pOpulation while maintain- ing diffusional access to a larger nutrient reservoir. The integrity of the membrane is both assumed and essential, especially when a pure culture is desired, interbiotic relationships are to be studied, or cell free diffusible products are to be recovered from the reservoir. Typical- ly regenerated cellulose membranes "ordinary dialysis membrane" with a mean pore diameter Of 5 V are used although other materials such as plastic or polyvinyl chloride membrane filters with a range Of pore diam- eters as small as 25 matare becoming pOpular for Specific applications. A third and relatively new type Of membrane, a nonporous synthetic mem- brane film, has been applied to gas diffusion. However one Of these, a co-polymer material based on co-polyether-ester compounds (poly-oxyethylene glycolmethylene bis 4 phenyl isocyanate), also has been proven useful for the separation Of related solutes (106, 107). AS shown for other non-po- rous membranes, mass transfer through these films is related tO solute- membrane interactions. Generally a membrane is assumed suitable for dialysis culture if it can be sterilized, has an adequate solute transfer rate, is not destroyed during the culture process, has a porosity smaller than the organism used, and is free Of rips, holes or flaws. Several tests are available to in- sure the latter (141). Penetration Of the dialysis membrane by the micro- bial culture apparently has not been a problem for either in vivo or in vitro dialysis. Our own experience with dialysis culture has not revealed sis cult ination Reserve to a de or a cc 1: purest of Shizelj. N 65 revealed such a problem (57, 61, current results) and a dialyzer-dialy- sis culture has been Operated for as long as seven days without contam- ination Of the reservoir (see the results presented in this Section). Reservoir contamination, when it did occur, was readily attributable to a demonstratable hole or tear in the membrane, an unsealed gasket, or a contaminating organism from an external source. It is generally recognized that membrane filters and other fine porosity materials when used in pressure filtration will eventually pass bacteria. The length Of retention is related to the concentra- tion Of organisms, duration of the filtration process, and the filtra- tion pressures and flow rates (143, 155). Singer recently showed that Pseudomonas cells did not penetrate membrane filters for a least 7 days (143). During the demonstration Of a differential dialysis flask system it was discovered that Staphylococcus aureus cells penetrated the membrane filter but not the dialysis membrane (79). This was con- sidered unusual because no pressure differential existed across either membrane barrier. Even more remarkable were the studies of Can which revealed that over twenty species Of bacteria and yeast were capable Of "dialyzing through" normal dialysis membranes (58). A review and discussion Of bacterial penetratability was included in the recent re- view by Schultz and Gerhardt (141). Because this represents a critical aspect Of the dialysis culture process additional related evidence will be presented here. Gan attributed membrane penetration to submicroscopic viable units Of the parent bacteria. He recently supported this supposition by demon- strating, at growth-discouraging temperatures, that such dialyzable units Of Shigella sonnei passed through collodion filters, could not be the visi bat: aezh the P8531v9 ‘ that Such 66 sedimented, and increased in size and changed in morphology until they became recognizable bacteria (59). He claimed this supported the theory that the bacterial life cycle included a filterable "in- visible" form such as an extracorporal gene or reproductive unit. Two alternate explanations for penetration were prOposed by Schultz and Gerhardt (141). The first suggested the formation of plastic forms (e.g. SpherOplasts or L-forms) by plasmOptySis Of the bacteria as a result Of the osmotic and ionic gradients across the membrane. The second involved the growth or diapedetic transfer Of the organisms through the membrane Openings. The demonstrated pene- tration Of membrane filters by bacterial L-forms and illustrated growth Of mouse peritoneal cell processes through membrane filters were cited as supporting examples. The appearance of bone tumor cells on the exterior surface of millipore filter growth chambers was recently Shown to be the result of the growth Of cell processes through the filter (54). Photomicrographs Of embedded and sectioned specimens illustrated that cytoplasmic extensions and collagin fibrils had grown through the tortuous pore channels apparently without de- forming them. Significantly, this study showed that diapedesis oc- curred with membranes which have a pore diameter (10 mph) which is only twice as large as that Of a regenerated cellulose membrane (5 Eye. The growth or diapedesis Of organisms through the pores repre- sents a more plausible explanation of membrane penetration than the passive diffusion of submicroscopic viable units. Cliver showed that such submicroscOpic particles, 30-85 yrviruses for instance, do not necessarily penetrate a membrane but instead absorb to the surface even though the mean pore diameter is 285 times the virus ' :13 .lm diate W351 no: 1 Ier : sure 67 diameter (28). This phenomenon, Observed under pressure filtration, was related tO the chemical composition Of the membrane material, ex- cept for porosities close to the virus diameter where penetration was influenced by the diluent used. These results pose the possibility that Gan's bacterial units might behave in a similar manner and would not penetrate a dialysis membrane, which is thicker and has much smal- ler pores, by passing through it especially in the absence Of a pres- sure gradient. The significance Of the failure of Gan's control par- ticles to pass through the dialysis membrane which had been penetrated by the bacterial submicroscopic units could be criticized on the basis of the above phenomenon. This would permit Speculation that dialyzabil- ity might still be Simply due to the presence Of some large pores in the dialysis tubing. Examination Of various membranes by transmission and stereoscan- ning electron micrOSCOpy has revealed that the cross sectional channels and cavities are quite tortuous, have random widths, and have numerous blind pockets (43, 95, 54, 77). These structural features have been cited as an explanation for the entrapment Of particles smaller than the rated pore size (54). Further, the examination Of membrane fil- ters after use showed that the organisms are concentrated on and with- in the upper half Of the membrane thickness, with few if any being found beyond the mid point (77). Thus, intact bacteria dO not readily penetrate membranes and it would be reasonable to speculate again that (gan's presumed submicrOSCOpic units would have difficulty successfully diaalyzing through the narrow tortuous channels Of the relatively thick dixalysis membrane under the passive pressure conditions described. However, these same structural conditions have not retarded the growth 13“ ‘l'.’ proce and :' biati 68 Of tissue cells within and through membranes and the illustrated ability Of this process to occur with porosities approaching that Of dialysis membranes lends support to the growth and diapedesis explana- tion of bacterial dialyzability. It is possible that the environmen- tal conditions Of Gan's tests stimulated such processes within the parent culture. Certainly electron micrOSCOpic examination of Gan's dialysis membranes would be a logical extension Of his investigations and could be valuable in establishing the validity Of the growth and diapedesis mechanism Of membrane penetration. Dialysis culture has been employed in a wide variety Of culture processes which include cell production, formation Of nondiffusible and diffusible products, mammalian cell and virus production, inter- biotic culture systems, gas exchange, periotoneal and hemodialysis, culture chamber implants, and rumen symbiodialysis (141). Since this publication, additional investigations concerning the application Of the dialysis culture technique have appeared. These are presented be- low and should serve as an updating for the above literature review and the Supplemental List Of References on Dialysis Culture Of Micro- organisms (available from P. Gerhardt by request). Schultz and Gerhardt (141) compiled a table (Table 5) listing the investigations in which nondiffusible products were Obtained with dialy- Sis culture in vitro. Two additional studies have been published and Should be included in the above list. The first concerns the produc- tion of thermostable hemolysin from Staphylococcus epideomidis cultures grown on a layer of cellophane covering the surface Of a brain heart infusion agar plate (90). Higher yields Of hemolysin were recovered from the surface dialysis cultures than from cultures grown directly on the pendei lular I vs F1, (I) r- in a s in ecu produc' exploi' Sc ture ha C°$P0U1 Product markedly culture I fIOm naph aIIC’Iled dj VIIlch rESu quEntly a . l 69 on the semi-solid medium. The second investigation utilized the sus- pended dialysis-bag technique to concentrate and recover an extracel- lular enzyme, proteinase, from Streptococcus faecalis cultures (114). The single dialysis apparatus, a cluster of dialysis tubes suspended in a six liter flask of growth medium, did not represent an advancement in equipment design and was similar to that used by previously for the production of highly potent tetanus toxin (91). Both of these schemes exploited the semipermeability of the dialysis membrane as a means for confining and concentrating a viable culture and its enzyme product while allowing diffussional access to the necessary nutrients in the flask. Schultz and Gerhardt found only a few reports where dialysis cul- ture had been used as a means to enhance the production of diffusible compounds. The technique has merit because it would allow metabolic products which are lethal or inhibiting to be continuously removed and diluted in the reservoir, resulting in higher growth and productiv- ity as well as simplified product recovery from the reservoir. This principle was recently used to remove a "growth inhibitory factor," tentatively identified as dialyzable lauric acid, from yeast cultures (4). Dilution of the inhibiting fatty acid by dialysis permitted markedly improved growth in hexadecane or decane medium. A similar culture response has been found for the production of salicylic acid from naphthalene (l). Dialysis cultures of Pseudomonas fluorescens allowed dialytic removal of the cellular product, salicylic acid, which resulted in a greatly enhanced production of cells and conse- quently a 20-fold improvement in salicylic acid production over non- dialysis cultures. In addition, the product could be easily recovered tinuall sure, 1 P31Vsac Culiure. Sible H} C “‘10 r9133 CUICUrin a cum”.E . miHera Circulati} against fI 0f 6, 26, as a! 70 from the cell free reservoir during the fermentation. Finally, it has been discovered that membrane phase separation and product removal by diffusion are useful for the examination of enzyme -- substrate -- product formation reactions (T. A. Butterworth, D. I. C. Wang, and A. J. Sinskey. Abstract, 158th National Meeting, American Chemical Society, New York, 1970). Judicious selection of ultrafiltration mem- brane porosity allowed an active enzyme preparation,O(-amylase from Bacillus subtilis, to be retained and continually re-used within a diffusion chamber which served as a reaction vessel. Substrate, con- tinually pumped into the chamber under constant ultrafiltration pres- sure, reacted with the enzyme and the reaction products, di-oligo-and polysaccharides, passed through the membrane for collection while the enzyme and unreacted substrate were retained within the chamber. This technique prevented the loss of enzyme permitting its continual re-use for product formation. An increase in culture density is a proven attribute of dialysis culture. Diffusional access to a nutrient reservoir is largely respon- sible for the enhanced growth. This was aptly demonstrated by Marino who reported that dialyzing Norcardia salmonicolor cultures after batch culturing vastly improved the cell yield (110). This organism achieved a culture density of 4-5 g/L when grown conventionally on a hydrocarbon -- mineral salts medium. When the same culture was then dialyzed, by circulating it through dialysis tubing coiled within a 40 L reservoir, against fresh medium the culture resumed growth and achieved densities of 6, 26, and 47 g/L after 1, 7, and 16 days respectively. The improve- ment was attributed to the removal and/or dilution by dialysis of growth inhibiting products which had accumulated. trated ture p system I; we (I) Stant elizina tissues b0 plantations 71 Sortland and Wilke achieved high culture densities of Streptococ- cus faecalis by utilizing a quasi~dialysis-continuous culture process (146). They designed a rotating microfiltration (millipore membranes) fermentor which could be operated on a continuous nutrient feed basis, batchwise, and with or without filtration. The desire to improve the time efficiency of continuous culture, to study the effect of concen- trated cell density on growth characteristics, and to analyze the cul- ture products free of cells prompted the development of this culture system. Essentially, this scheme allowed retention of the cells normal- ly washed out during continuous culture while still permitting a con- stant feed of nutrients through the fermentor. Theoretically this eliminated interferring fluctuations in nutrients and growth rates while maintaining a high culture density. The system apparently met expecta- tions. High cell densities, up to 40% packed cell volume, were obtained which were 45 times greater than achieved in simple batch culture. A practically cell free (filter efficiency was high but not absolute) cul- ture effluent allowed precise chemical analysis which showed that glu- cose was converted almost stoichiometrically to lactic acid with no con- sumption for cell maintenance. In addition high cell densities had no apparent influence on growth characteristics and changes in cell yield per mole of glucose consumed unexplainably changed abruptly during growth. As found with most dialysis processes, this system could prove useful on an industrial scale if costs were lower. Schultz and Gerhardt cited examples where the dialysis culture technique has been applied toward propagation of mammalian cells and tissues both in vitro and in vivo, the latter by dialysis chamber im- plantations. The reported investigations were directed toward pr:- v.” '1 72 propagation of the cells and tissues themselves, virus production, and examination of cell differentation, antibody production, homgraph sur- vival, and other interbotic relationships. Several additional studies concerning in vitro dialysis culture of mammalian cells have appeared since the review. One involves the modification of a tissue culture chamber by inserting a membrane partition which allows diffusional ac— cess to the cell clones (11). The construction of a dialysis chamber is not new but the use of Nuclepore membranes (the .5? to .8/0pores are formed byauparticle bombardment) is. The membrane was autoclavable, nontoxic, unpenetratable by the cells, and diffused radioiodinated serum albumen molecules with equilibrium between the chambers being reached in 7-9 hours. The authors felt that this culture system would enable cer- tain essential metabolites from a feeder culture in one chamber to dif- fuse through the membrane to supply and enhance cell cloning in the other chamber. They also prOposed the use of this device for small scale continuous perfusion. A far more complex dialysis tissue culture system described as a multichamber circumfusion system has been under development by Rose and co-workers for over ten years (132). First they noted that differentia- tion in growing tissue explants was superior when the multipurpose cul- ture chambers were overlayed with dialysis membranes which improved cell anchorage to the glass cover plate (135). Then they created the circumfu- sion system by modifying twelve of these chambers for the continuous per- fusion of nutrient fluids and connected them with the necessary pumps, reservoirs, and conduits to form a compact self-contained system enclosed in a plexiglass incubator (132). Nutrient fluid, circulated by pumps, wflas separated from the tissue explants by the membranes which permitted 1. 5.. r# BED: OUSE PUIQOSE far bEtz bet IOta YdroSta 73 diffusional exchange but not direct contact with the culture environ- ment. Oxygen was supplied to the fluids by exchange through the teflon conduits and carbon dioxide from an exchanger, composed of teflon tubes, submerged in the nutrient reservoir. Detailed descriptions of the equip- ment and its Operation have been illustrated (132). This dialysis system was shown to successfully prOpagate various lines of tissue cells, en- hance cellular differentiation, and, in a separate study, allow continu- ouse prOpagation of mouse melanoma cells in a serum free environment for 418 days (133). Continuous and indirect contact with the culture environ- ment by in vitro dialysis more closely resembled a natural in vivo envi- ronment than conventional or individual culture chambers. It was felt that this enabledsuperior control of a constant physiological pH and the provision and retention of basic and vital nutrients and cellular secre- tions respectively creating a growth enhancing microenvironment. In ad- dition this system provided the interconnection of multiple culture units, the use of'a reliable respiratory system, and the containment of all the components in a temperature controlled air circulated incubator. Al- though attributes of dialysis exchange have been applied to bacterial nutrition and aeration (57, 61, and Section G of this thesis) this repre- sented an advance in technique and scale size for the application of di- alysis to mammalian tissue culture. Recently Rose further improved the circumfusion system for multi- purpose culture chambers by redesigning the individual culture chambers for better membrane seals and by adding the physical dimensions of cham- ber rotation, additional culture chambers, alternating introchamber hydrostatic pressures, and fluid switching mechanisms(134). This work represented a refinement of the previous system and was reportedly capill for th. ture u: vivo cc illustr perzitt for the tissue 1 with i1] 74 superior for the maintenance of cellular differentiation. The major design changes eliminated membrane wrinkling in the culture chambers, reduced sedimentation deposits, simulated in vivo venous and arterial capillary pressures, enlarged the capacity of the system, and provided for the selective direction of fluid nutrients through a choice of cul- ture units. The elimination of sediments and closer simulation of in vivo conditions reportedly enhanced the results for the tissue lines illustrated. The additional chambers and fluid switching capability permitted continuous or intermittent membrane mediated fluid contact for the study of such interbiotic relationships as symbiosis or host- tissue reactions. The work represents further refinement of hardware with illustrated examples of its operation. I“. fi' (941), 5 charged volume d (ii-M). (Figure 39" x 20 t0! (3‘M Stant Ce: 5.2 Materials and Methods Dialysis growth trials were conducted in a dialyzer-dialysis cul- ture system (Figure 18 and 19) similar to the one developed by Gallup (61) and the same as the one described in the preceding section (Sec- tion 4.2). The culture vessel was a conventional 5 liter fermentor (24-M) which was charged with 3 liters of Trypticase soy broth (30 g/L) (9-M), synthetic medium (145), or distilled water. The reservoir was charged with 10 to 26 liters of one of these media. Depending on the volume desired it consisted of either one or two 14 liter fermentors (24-M). The latter were connected in series by conduits and a pump (Figure 19). The fermentor and reservoir vessels were situated in a 30" x 20" x 15" water bath equipped with a mixer (22-M), thermoregula- tor (3-M), and 1000 watt immersion heater (Z-M) for maintaining a con- stant temperature of 30.0. The prototype dialyzer described previously (Section 3.2) contained 1 to 15 (.0288 m2 to .4320 m2 area) presoaked Visking regenerated cellulose dialysis membranes (28-M) and was connected to the vessels by 1/4 inch I.D. rubber tubing (ZS-M). A positive diSplace- ment variable speed pump (27‘M) inserted in the reservoir and fermentor circuits circulated the fluids from their respective vessels through the dialyzer and back. The stainless steel heads of these pumps were remov- able which permitted steam sterilization of the entire dialysis culture system as a unit. Glass "T's" capped with vaccine bottle stoppers were also placed in the fluid circuits to allow asceptic inoculation and 75 F -——— n ”I“: 76 NUTRENT FERMENTOR DIALYZER RESERVOIR ”(E—77 FERMENTOR Figure 18. Schematic diagrams of the culture systems used. Top; A dialyzer-dialysis culture system. Bottom; A nondialysis ”control" culture system. 77 Figure 19. Assembly of the experimental dialysis growth system. Two 14 liter fermentors are joined to form a 30 liter medium reservoir (left) which is connected to the dialyzer (center) by rubber tubing. The culture fermentor (right) is connected to the dialyzer in the same way. Pumps (foregound) circulate the respective fluids through the dialyzer. [h systa" n resen’i‘i [ whit? in the 53 augh [he Figure 19 sample and res the ves 79 sample removal by syringe and needle. The contents of the fermentor and reservoir were mixed by means of the pumped fluid circulation and the vessel impellors. The latter were connected to 1/8 H.P. motors (lO-M) by flexible drive shafts (21-M) with speed being controlled by speed regulator power units (l7-M). Culture aeration was accomplished by vigorous agitation and sparging. Prefiltered (lZ-M) air from the building service was regulated by a flowmeter (18-M), sterilized by a 12 inch fiberglass packed filter, and humidified before entering the single orifice sparger of the fermentor. Gases exiting the fermentor passed through a foam trap and another fiberglass filter. The reser- voir was not aerated but was vented by a filter and slowly mixed. Nondialysis batch "control" growth trials were identical to the dialysis growth trials, except that the reservoir and dialyzer were not used. In this case, the culture fermentor was charged with growth media which was circulated, sampled, and aerated in the same manner as above. The dialysis and nondialysis schemes are compared in Figure 18. The test organism was Serratia marcescens strain 8 UK (29-M). The choice of this organism maintained continuity with previous dialysis culture work, permitted identification of equipment leaks or contamina- tion by its pigmentation, and allowed meaningful evaluation of growth results because its nutrition and growth has been documented (145, 100). A uniformly-pigmented colony was picked from a streak plate and trans- ferred to a 500 m1 flask containing 100 ml of Trypticase soy broth. This was incubated on a rotary shaker (23-M) for 10 hours at 30’C. The resulting culture served as an inoculum for the growth trials. This was approximately a one percent by volume or 70 to 200 million cells per m1 concentration. IESE‘ 80 The initiation and follow-though of all growth trials was similar. The equipment was assembled as described and illustrated. Preparation of inoculum and media was the same in companion experiments. Several types of trials were used: (a) "control" where the fermentor contained 3 liters of growth medium; (b) dialysis where both the fermentor and reservoir initiallylxxeived growth media; and (c) dialysis where the fermentor was initially charged with water, the reservoir charged with growth medium, and a 10 hour dialysis exchange period run before inocu- lation of the fermentor. Upon selection of the type of trial to be run, the necessary equip- ment was assembled, filled, and steam sterilized as an integral unit for 40-50 minutes at 121.C. Routinely 3 or 2.6 liters were used in the fer- mentor and 9 or 26 liters in the reservoir, equivalent to fermentor to reservoir ratios of 1:3 and 1:10, respectively. All fluids were steri- lized within their respective vessels, except for glucose which was added asceptically prior to inoculation. Upon cooling, the system was assem- bled as illustrated in Figure 19. After temperature equilibration to 30%., the fermentor was inoculated with 2.6 or 3 m1 of the shake cul- ture to provide an initial cell concentration of approximately 70 million cells per ml in the fermentor. Unless specified otherwise, the culture system was operated at pump velocities of 2 L/min., air velocity of 9 9 L/min, agitation of 365 r.p.m. (fermentor) and 50 r.p.m. (reservoir), a pH of 7, and a temperature of 30.0. During the 48 hour growth trials, culture samples were periodically removed and assayed for total and viable cell numbers. The amount of cell matter was assayed by dry weight and Optical density. Total counts were obtained by direct microsc0pic observation using a conven- tional Petroff-Hauser counting chamber and appropriate dilutions. 81 Viable counts were obtained by the conventional surface plating tech- nique on Trypicase soy agar. Dry weights were determined by centrifug- ing, washing, resuspending and oven drying (ll-M) culture samples at 105°C to a constant weight. Optical density was measured at 620 mglby a calorimeter (6-M) on 1:10 sample dilutions. All samples were corrected for liquid loss. Initial volumes were maintained by addition of sterile water from an attached reservoir or by syringe as needed. Foam was con- trolled by addition of approximately 50 ppm of polyprOpylene glycol (16-M) prior to sterilization. Finally, all growth trials were monitored for pH (7-M), contamination, and loss of volume due to evaporation, sampling, foaming, or leaks. .a, : ‘ I l . duCte tor E dial} abOvé a Si: Sis E 5.3 Results The experimental growth trials reported here were intended to illus- trate the attributes of a dialysis culture system, demonstrate the suita- bility of this prototype dialyzer, and show the influence Of some operating variables on the growth of Serratia marcescens cultures. Analysis of the growth characteristics of the test organism in nondialysis batch culture provided a standard reference against which subsequent dialysis growth trials could be compared. A series of duplicated nondialysis cultures were grown in 3 liters of Trypticase soy broth under the routine condi- tions listed previously. A composite growth curve (Figure 20) revealed that these achieved a mean viable cell concentration of 55 billion cells/ml after 36 hours, which decreased to a mean of 46 billion cells/ml after 48 hours. The peak cell concentrations ranged from 43 to 88 billion cells/ml with a standard deviation Of‘i 22 billion cells/ml between repeated trials. During the exponential growth phase the organism had a mean generation time of 30.1 minutes i;4 minutes (s.d.). A similar series of duplicate growth trials with dialysis were con- ducted using 3 liters and 10 liters of Trypticase soy broth in the fermen- tor and reservoir reSpectively (a 1:3.2 fermentor to reservoir ratio), ten dialyzer membranes (.288 m2 area), and the same Operating conditions as above. Significantly superior viable cell concentrations were reached and a similar repeatability Of results were Observed (Figure 21). These dialy- Sis growth trials achieved a mean peak cell concentration of 120 billion 82 1H1! h...” meJUMU NJE<~> AHJ tabla h hllllliida \ batCh 83 ES LOG VIABLE CELLS PER ML ' (D 8 .4 L l 1 J :2 24 36 48 HOURS Figure 20. Growth curve of Serratia marcescens; mean Of four duplicated batch nondialysis cultures. —»\d avbtwurk NU - ..H.!\ in.» .er..(x.\’ “\rr\ ~ 84 I I r I DIALYSIS II - // wwwwwwwwwwww /' / I '1'! / >- IO—- / — a: I m °~ I m j I “J I U I “e? [I > 9 F- —— C) I O I ..J I I I I I I 8 - l,’ ‘ a I I l L L ..r. .1... C) (I .0 (r; ‘3 Figure 21. Growth curve_of Serratia marcescens; mean Of duplicated dialysis cultures with ten membranes ( .288mz ) in the dialyzer. 85 cells/ml after 48 hours, had a range Of peak values from 100 to 140 billion cells/ml. The exponential dialysis growth rate, the mean gen- eration time, 35.4 minutes i .6 minutes (s.d.), and the standard devia- tion, i_20 billion viable cells/ml between repeated trials were similar to those for the nondialysis cultures. Although there was a difference in inoculum concentration between the two sets of growth trials (Figure 21) this was shown, in subsequent experiments, to have no significant effect on final culture concentrations. Thus the repeated superiority of dialysis cultures, an extended exponential growth phase and a greater final culture density, could be accepted with confidence. Another attribute Of dialysis culture was the maintenance Of a greater prOportion of viable cells during a 48 hour growth trial (Table 11). The examples illustrated show that: the correlation between total and viable cells was superior for dialysis culture in all cases; the proportion of viable cells in a culture decreased as growth progressed; dialysis culture with a small membrane area, series A, produced cell concentrations similar to those of comparable nondialysis but with a superior prOportion of viable cells per total (89% vs 75%) after 48 hours; and dialysis culture, utilizing an Optimal membrane area and a water dif- fusate medium, series B, produced both superior cell concentrations (283 billion vs 8 billion) and a markedly greater proportion of viable cells per total (85% vs 42%) after 48 hours of growth. The preceding data demonstrates the successful Operation of this dialysis cultue system with the prototype dialyzer and confirms two po- tential advantages of dialysis culture, increased culture density and improved culture viability. Microbial prOpagation in a clear particu- late-free medium is desirable in certain fermentations, especially when .Eswpoa muons how Ommowumhua maflcfimucou HHO>uOmOu on» Eoum wo~>~mwp mummsmmww umuma ocu %Hco pmcwmucoo >-mauwsfi “OucoauOm can OposB mamas» :u3Onw mwmhamwp .m .Eswvoa suoun hOm OmmuwummuH wmofimucoo %Hfimfluwcw HMO>uOmOu pom uOucOEqu nuon muons mfimwuu susouw mwmhfiwfiv .4 x 86 Ne w as He as we eeesseeeeoz tHeeeeeEOO m A aomme.v . mm new omm as mes wee memeeeaez N eeesbeea mm NO on Ca ms mm mamhflmwwuoz mHnmummsou < A aesmo.v . mm 3 me 3 mm 8 89382 N entrees eaeee> eHeee> esteemm «Heme> fleece eeeeeem efieee> Heeoe zmamwm expanse emmHmMm memos ms mmee< memos em mmam< a: see magma mo mongsHm .wcoomwouma mauwuumm nuw3 mfimwuu susouw mwmhfimwwcoa can wwwzamwp cw mcowumuucmocoo HHOO OHAOH> was Hmuou mo :Owwummaoo .HH OHan 87 recovery of a metabolic product or a "clean" suspension of cells is sought. Such a fermentor environment can be Obtained and a continual adequate nutrient supply maintained in dialysis culture. Such dialysis cultures, when Operated optimally and with a large membrane area (15 membranes, .4320 m2 area), achieved high population densities and cell yields (Figures 22, 23, 24). As illustrated, the significant superiority of dialysis culture was demonstrated in terms of higher viable cell den- sity, total cell density, and cell mass. The comparable nondialysis cul- ture in this instance was initially identical, the fermentor containing 2.6 liters of Trypticase soy broth diffusate, but did not receive the benefit of continuous nutrient exchange by dialysis after inoculation. The advantage Of dialyzing a large nutrient reservoir was clearly shown by the longer exponential phase and by the continuous increase in viable cell concentrations and cell mass in the later stages Of the trial (48 hours). In contrast these parameters in nondialysis declined or remain- ed stationary. Probably the most important variable Of the dialysis culture system was expansion Of dialyzer membrane area. When a 2.6 liter fermentor con- taining the water diffusate of the Trypticase soy broth in the 26 liter reservoir (1:10 ratio) was inoculated with Serratia marcescens, dialyzer membrane areas Of .0288 m2 (l membrane), .0864 m2 (3 membranes) and .1728 m2 (6 membranes) produced similar rates of growth and only slightly increasing viable cell concentrations after 48 hours (Figure 25). As shown an increase in area to .2880~m2 (10 membranes) significantly improved cell concentra- tions from 80 billion cells/ml for six membranes to 210 billion cells/ml. Further expansion to .4320 m2 (15 membranes) produced only a small addi- tional increment in final viable cell concentration to 283 billion cells/m1, 88 Figures 22, 23 and 24. Growth curves of Serratia marcescens, in dialysis and comparable nondialysis culture. Operational conditions were: 2.6 liters of Trypticase soy diffusate in the fermentor and 26 liters of Trypticase soy broth in the reservoir (a 1:10 ratio), 15 membranes in the dialyzer, 30 C, 9 L/min Sparging, 365 r.p.m. agitation, and 2 L/min liquid flow rates. 89 ' 9 p «>2 b ‘ g m 1‘ (D a a )- 1.. 4 § .. (5 N O O - l A 1 S 2 a 9 0° 1! H3d lH‘JnM ILHO on ‘M 438 :57 ‘3”. WI”; x SNIJITIH C Q ' I. #5 Ir. I3 2 m & 0 , I f I T I G <5, P <1 *3? L_ I O OIAU "JS l l o l2 E m a: N 1H ‘53:: S'IH‘J ‘WlOl 5.11 ‘M H3d STUD TWA, IO QNOFHrfl ’ H2 _ g, e I: ‘ O |( Q m *1“ N ”m d3d $1133 318W SUI HOURS HOURS HOURS Figures 22, 23, 24 . J2 \ WJJWU WJNwQCI 0 P... J 90 ] I l T IOnoIS MEMBRANES O O l I *- 9" O C O 8 O I,3,m6 MEMBRANES ..J z ’ ... — —~ ~ § _ \ // ‘ ‘u “ ‘ —i 93 'OF / ‘ ~"’ ~ ~ ...I “ I41 0 NONUALYSIS LU ...) CD ‘5 > C) O ...] 9 r- ‘ 8 I 1 L 0 l2 24 36 48 HOURS Figure 25. Influence Of dialyzer membrane area on growth Of 2 S was marcescens. ,Each membrane has an effective area of .0288m , e fermentor contained water diffusate Of Trypticase soy broth, and a 1 '- 10 fermentor to reservoir volume ratio was used. 91 the highest value recorded for the system. These latter results com- pared favorably with the earlier Observation that nutrient transfer rates Optimized at a dialyzer area Of .288 m2 (10 membranes) (Figure 8). An increase in membrane area appeared to extend the duration of the ex- ponential growth phase enabling the attainment Of higher culture concen- trations. Area enlargement did not however increase the prOportion of viable cells within the culture (Table 11). The advantage of additional area was further illustrated by the final cell mass (dry weight) for the growth trials (Figure 26). Here the addition Of area above ten membranes produced a more decided increase in yield than shown by viable cell counts. This difference was because mass measurements included both viable and nonviable cells. Dialysis cultures in all cases significantly exceeded the results Obtained in comparable nondialysis cultures. In ad- dition, dialysis under these conditions enable a prolonged linear-station- .ary growth phase in contrast to the decline in culture viability Observed in nondialysis. The preceding growth trials revealed that dialysis cultures consist- enntly achieved superior culture concentrations and viability, extended eaacponential growth phases, and prolonged linear and/or stationary growth Iptiases. In addition, these results were enhanced by an increase in mem- t>1rane area. These trials indicated that adequate diffusional access to £1 nutrient reservoir was the key to the success Of the dialysis culture 5337£item. It was shown earlier that this process, by virtue Of the con- tiIluous circulation Of the culture and reservoir fluids through the di- a1372er, represented equilibrium dialysis (Figure 6). As shown, solutes ‘1 i‘iailyzed from the reservoir into the fermentor until the system reached ecl'~1:i.librium. The same process occurs in a culture situation (Figure 27). r H a ~..\ T.<.. q.- \ f ~ _ e l e . a t L ‘5‘ LL DENSITY (GM/L) E C 4:. C) l 92 ()4 C) I ICC} I A - O l I l I I O 3 a o 22 IS .'!I+:"*v r‘ l“‘3rnflfifit'7‘ rI I..II+'I:.,;CL¥‘\ ”F r‘J'ItIr'ltltk\;fii \3‘ 8 Figure 26. Influence of dialyzer membrane area on cell mass of Serratia marcescens after 48 hours growth under the same conditions given in the preceding figure. The data point on the absissa represents the value for a nondialysis growth trial. ' 93 Here the nutrients, a glucose-citrate salts synthetic medium containing 2.5 g/L glucose which was the solute measured, transferred fixmlthe 30 L reservoir to the 3 L fermentor containing water;during a 12 hour equili- bration period. In this example the dialyzer had a single membrane and the system had not reached equilibrium (which was 2.2 g/L or 86% of the initial reservoir concentration) at the time of culture inoculation. However, the fermentor had received 1.8 g/L of glucose (80% of equili- brium or 69% of the initial concentration) which was adequate for rapid growth. Additional membrane area would enhance the rate Of glucose transfer prior to inoculation. Figure 27 shows that culture growth, especially during the exponential phase, caused a pronounced decrease in fermentor nutrients. Diffusional access to the reservoir prevented the decrease in the fermentor from extending below 30% of the initial glucose concentration and allowed the nutrient concentration to gradually recover during the growth trial. Reservoir nutrients were not expended by the end Of the trial. They had only decreased to 70% Of their initial value, and a new fermentor-reservoir equilibrium Of 1.5 g/L or 57% of the initial concentration, was in the process Of being established. In contrast, a comparable nondialysis growth trial with the same growth medium experienced a 90% decline in nutrient concentration within 12 hours (Figure 28). This limited the extent of growth (26 vs 139 billion cells per ml) and caused the appearance Of a decline growth phase. Dif- fusional access to the nutrient reservoir appeared to be an important attribute Of the dialysis culture system. This permitted sufficient transfer to prevent the exhaustion of fermentor nutrients and allowed attainment and maintenance Of superior culture concentration. In ad- dition these growth trials 94 Figure 27. Correlation between dialysis growth and solute (glucose) concentration in the fermentor and reservoir; the time period ~12 to 0 hours represents the equilibration of the 3 liter fermentor and 30 liter nutrient reservoir prior to culture inoculation. A single membrane (.0288 m2) was used in the dialyzer. 95 “LE SELLS / ML ‘ A V If‘A L.) .l \IIA’: ,5 112 ’7. CF MA X {M U (3 OJ ‘9' ‘4 OO. WEDOI mOHZwEmmm 50>mmmwc I. \ MI I‘ \Jt...' ... ’ IN i T 3 A I,» III) Fv \L‘ Figure 27. 96 '1“! $1139 B'BVIA HOWIXVW :00 X s 2 8 9 so HOURS I1 I I I I ‘ T fie? -.‘8 _¢ N me! 0 l I, l .J_ It I l c) 1' 8 8 .9 8 ° NOWHLNBOWO BUTTOS ‘MJJNI :D ’5 Figure 28. Correlation between nondialysis growth and solute (glucose) concentration. 97 demonstrated that the presence Of metabolizing cells had no effect on the dialysis mass transfer process other than to shift the equilibrium end point of the system. The benefit derived from diffusional access to a nutrient reser- voir was vividly illustrated when growth trials were extended to five days or more (Figure 29). Dialysis cultures of Serratia marcescens achieved substantially higher culture densities which were maintained in a prolonged stationary growth phase for up to five days. An enlarged dialyzer membrane area had no influence Of the stationary phase other than to raise the level Of cell concentration during this period. In contrast, nondialysis cultures had a substantial reduction in viable cell concentration over the same time Span and showed little or no sta- tionary growth. Sufficient nutrients remained in the reservoir after six days to further extend dialysis growth an additional six days. These results demonstrate the value of the dialysis culture reservoir for the supply Of nutrients and the dilution of growth limiting solutes within the culture. As a result highly concentrated viable cultures were consistently maintained for long periods. This characteristic could prove advantageous for the production of intermediary metabolic products which generally are produced most rapidly during the stationary growth phase. Growth media, on the basis Of glucose concentration, appeared to influence nondialysis cell yields (Table 12). The greatest culture den- sities occurred in the medium with the greatest glucose concentration, with progressively poorer yields occurring in weaker media. This de- pendance on nutrient strength was not Observed with dialysis cultures, which achieved higher cell yields in all the growth media. The -J «3L \ .... J J m... I . I\ N} ...Mk I- Vs...“ /.\~\./ "VJ Pad J LOG VIAELE CELLS/ML. ‘2 T I I I I I I0 Iv’IENIERANES A _n O 3 k‘fEfvfiEIRIC‘II‘ILS __ Il - Q 8 \ . \\ \ \ E \ \\\\ 7“ __ ’ \ "‘ K] NONDIAL‘I'SIS I» 9 .— 4J9 l . I I I I co I 2 s - 4 e s 98 DAYS Figure 29. Influence Of dialysis culture and membrane area on the viable concentration Of Serratia marcescens cells over an extended growth period. 99 .uoumfimwv mnu cw moamuasoa mo Hones: one Oumcwwmmw muawuumnam * n.H on ONH o.oH SOHOOZ afiumsuahm cowcnOhInuHEm m H.N om ONH m.~ nuoum mom OmmOfiumhuH N S a m S 03 8 a; foam sow 383939 no 333mg .833 H Sn 0 Sn en GEO 2528:va zofléazmozou 3:5 555, $83.5 eon—.2823 manage 5 223m: .mOHOOHDO Auv mfimzamwvaoa mow «any mammamwp cw cowuosvoum HHOO co Esfimoa cusouw mouoofimm mo ouconamaH .NH magma 100 superiority of dialysis over nondialysis was eSpecially apparent with the weakest growth medium, water diffusate Of Trypticas soy broth, in which dialysis produced 9 times more cells. This yield was further im- proved, to the highest shown in the table, by an increase in membrane area. The results again exemplify the advantage gained by diffusional access to a nutrient reservoir. In this case dialysis Of a large reser- voir Of dilute growth media prevented the growth limiting depletion of the fermentor nutrients during the fermentation. As expected when a small concentration gradient is involved, the process responded favorably to increased membrane area by providing additional nutrients sufficient for greater cell growth. The separate fermentor, reservoir, and dialyzer regions of the dialysis culture system allowed independant control Of each. A number Of fermentor-to-reservoir volume ratios were Obtained by changing the volumes used in these vessels. Solute transfer experiments indicated that a 1:10 fermentor to reservoir volume ratio was Optimal (Figure 16). Previously published growth trials with Serratia marcescens in a reser- voir-dialysis culture system showed that the same ratio produced the best culture yields (57). Growth trials with the same organism demon- strated that a 1:10 ratio achieved higher culture yields than smaller (1:4) ratios in the present dialyzer-dialysis culture system (Table 13). These results correlate with the calculated differences in culture yields anticipated for different ratios in this system (Figure 30). The fer- mentor-to-reservoir ratio could be changed either by reducing the fer- mentor volume (series A, Table 13) or by increasing the reservoir volume (series B, Table 13). The results illustrate that the greatest cell den- sities and culture yields were Obtained with the latter adjustment. 101 .OESHO> ug0>uomou Ow Ommouoafi cm Ou Ono owcmsu owumu .m mowumm .OEOHO> uoucmsuom mo aowuozvmu Ou use owowso Owumu e4 mowuom * omm OON ndfi OHHH om ©.N 00m ONH om OHH NH m m HON mmH OMH oHuH ma m.H 8m 02 om .3 S m .4 Amoucoaaom \szHHHHuHV mason wq uoum< musom «N umum< ufio>ummom uoa:o8uom QHWHW RAND AZ mmm mAAmU mgm¢H> ho mZOHAAHm OHH¢M ARV MZDHO> *mmHmmm .mcoomooume mwumuuom mo manuaao mwmhamwv wawuuv cowumuuoOOcOO Hflwo OHnmw> :O mowumu OeDHO> HHO>uOmOuIOquOuaoauOm mo OuaonamcH .mH OHan 102 I I I I I ‘ [Yang 50" // VF . .... // / / / / / 3 40" // "‘ \ / 2 / 9 / / / vi. >- ... 4:! _. C 30 / a F O (é) / us / 0 / :‘I 20—- / ,, v - UJ /’ ._::'r5 Q) ' / ,_ ———————————— \I L V9,,“ k) V? \M: I I I I 20 3'? 40 ‘“O HOURS Figure 30. Computed and eXperimental effect Of the reservoir-to-fermentor volume ratio on the cell density Of dialysis cultures. 103 Although reduction Of the fermentor volume achieved a 1:10 ratio, the beneficial effect of the proportionately larger reservoir was lost due to the inefficiency Of the smaller 1:3 liter volume in the 5 liter fer- mentor. Consequently this scheme although able to produce a higher cell concentration did not achieve as great a culture yield as anticipated (Series A). The same ratio, 1:10, when Obtained by enlargement of the reservoir while retaining the original fermentor volume, 3 liters, pro- duced both a higher cell density and culture yield (series B). This data further demonstrated the biological advantage gained from the nu- trient supply and toxic dilution capacity of a large reservoir in the dialysis culture system. Reservoir enlargement to ten times the fermen- tor volume appears tO be the most convenient, most practical, and most productive Operational scheme for this dialysis Culture syStem. The velocity of the fluids circulating through the dialyzer could be varied by altering the pump speeds. Previously it was shown that a minimal velocity Of .5 L/min was necessary to produce turbulent flow within the dialyzer and that flow rates greater than 2 L/min had no effect on solute mass transfer rates (Figure 9). Neither the growth rate nor the cell concentration of Serratia marcescens cultures were significantly changed when the dialysis system was circulated at veloc- ities above 2 L/min (Figure 31). 104 ' 7 ' ' I I I O O H —- —~ 0 6 L/MIN .21 O 2 L/MIN \ IO - '4 9 ..J Lu (J u: .I 00 ‘3 > (9 C) .J 9 - __ I D 8 I I I 0 IE 24 36 48 HOURS Figure 31; Influence of of Serratia marcescens. liquid flow rate through the dialyzer on growth 5.4 Discussion The experiments presented in this section were not intended to introduce the dialysis culture concept or dialyzer-dialysis culture system. This has been accomplished (57). Instead these investigations were conducted to demonstrate the biological compatibility and perform- ance of the dialyzer described in Section 3 when used in a dialyzer-dialy- sis culture system. In addition tO showing the successful Operation and confirming the previously claimed superiority of dialysis culture, the growth trials were also intended to define the influence Of various Op- erational procedures on biological performance. The data were examined to determine if a correlation existed between solute transfer and culture growth, and an attempt was made to delineate the key factors which con- tribute to the biological superiority of the dialysis culture process. In order to critically analyze dialysis cultures, the routine pro- cedures used in fermentation growth trials were individually examined and found to have little or no influence on interpretable results. This permitted conclusions derived from dialysis and nondialysis results to be attributed confidently to the culture process and not to the influ- ence of Operational conditions, such as pH, buffering, agitation speed, sparging flow rates, antifoam agent, or inoculum size. However culture circulation was an exception. Uncirculated nondialysis cultures showed a decline phase between the 24th and 48th hour of the growth trial. This was less pronounced in circulated nondialysis trials and absent 105 106 in dialysis trials. Although various velocities had no effect, the dynamics created by the positive displacement gear pumps apparently re- duced cell clumping or improved culture aeration or culture mixing re- sulting in improved viable cell concentrations during the later stages of the growth trial. Thus nondialysis trials were always circulated in order to accurately compare the growth results with those of dialysis trials which, by necessity, were circulated. The absence of a clearly defined decline phase in 48 hour nondialy- sis cultures was Observed only when Trypticase soy broth was used. Simi- lar nondialysis cultures using synthetic glucose citrate salts or water diffusate of Trypticase soy broth media exhibited decline growth phases. Dialysis cultures showed a prolonged stationary phase and no decline phase regardless Of the nutrient medium used. Nutrient selection did, however, influence the degree by which dialysis culture densities sur- passed the comparable controls. This emphasized the care which must be exercised in selecting duplicate culture procedures for both dialysis and nondialysis when comparison between growth trials are intended. Failure to do this as detected in previous reports (57) reduces the credibility of the conclusions drawn from the experimental work. The many variables Of a microbial fermentation which must be con- trolled Or accounted for Often make experimental duplication difficult. The variation observed between supposedly identical fermentations has prompted many to regard the fermentation process as an art and has made the term "growth trial" a more apprOpriate description of these investi- gations. A series Of repeated dialysis and nondialysis growth trials were made. These permitted an estimate of the standard deviation, range, and mean viable cell counts which could be expected for both types Of 107 culture techniques. This statistical information helped evaluate data and determine the significance of comparative assays. The duplicated nondialysis "controls" produced growth curves which were similar to those reported previously for batch growth Of Serratia marcescens on Trypticase soy broth (57). However, the mean peak cell concentrations achieved in these trials were unexplainably lower than before. Duplicated dialysis cultures using the same nutrient medium produced growth curves comparatively similar to nondialysis except that the mean peak cell concentrations were significantly higher (2 fold) and the culture densities continued to increase in a linear manner through- out the later stages of the trials. Additional membrane area and a larger fermentor to reservoir volume ratio further improved the concen- tration Of cells in dialysis culture to better than a 5 fold advantage over nondialysis. These results all agreed, in principle, with those reported previously for dialysis cultures (57). The similarity in ex- ponential growth rates and in viable cell assay standard deviation values between the two culture systems showed that dialysis had no adverse ef- fect on bacterial multiplication. The series Of duplicated growth trials demonstrated that dialysis results were repeatable within certain limits. The trials also showed that this dialysis culture system equalled or exceeded the expected superiority over comparable nondialysis. The primary biological advantages gained from employing dialysis culture appeared to include an extention of the eXponential growth phase, a continual linear increase in culture concentration in the latter stages Of the growth trial, an improved culture viability, and as a result of these a greater concentration Of cells within a defined fermentor volume. Although the results did not surpass those reported for the original 108 prototype system, the conclusions Obtained were in agreement with those predicted for the dialysis culture technique (141). Several growth trials were made to correlate the biological per- formance of this dialyzer and dialysis culture system with its solute transfer characteristics. Solute transfer data (Section 4) had shown that a fluid velocity of 1 L/min and a volume ratio Of 1:10 produced Optimal glucose transfer rates. The same was found to be true for prOpagation of Serratia marcescens. Increased fluid velocity to 6 L/min produced no improvement in growth over a 2 L/min rate and a 1:10 ratio produced significantly better growth, eSpecially toward the end of the trial, than a 1:4 ratio. The demonstration that a 1:10 ratio was Op- timal agreed with earlier data for other dialysis culture schemes which showed that both smaller and larger ratios produced poorer growth den- sities (57, 61). It was noted that, with this dialyzer-dialysis culture system, the best culture densities were Obtained if the fermentor volume was kept constant (3 literS)while the reservoir volume was increased. In contrast, superior culture densities had been achieved in the dialysis flask and reservoir-dialysis culture systems when the Optimal ratio was Obtained by a reduction in fermentor volume. In both of these the mem- brane interface was an integral part Of the reservoir region. With the dialyzer-dialysis culture system it was separate and a reduction in the fermentor volume reduced the liquid level within this 5 liter vessel to a point where its Operational efficiency declined. It is expected that this problem could be alleviated by the use of a smaller fermentor ves- sel. However on an industrial scale it would be simpler and more economi- cal to increase the size Of the reservoir while maintaining a constant standard fermentor size. The results indicated a direct relationship 109 between Optimal solute transfer conditions and microbial growth. Further, it was evident that this dialyzer-dialysis culture system reSponded to the advantage of a prOportionately larger nutrient reservoir in a manner similar to that Observed for other dialysis schemes and in accordance with theoretical predictions for the system (141). Enlarging the dialyzer membrane area improved culture yields in the same manner as it had improved soluted transfer efficiency (Section 4). A comparison Of growth curves when 6 (.1728 m2), 10 (.2880 m2), and 15 (.4320 m2) membranes were used suggested that the Optimum area for solute transfer (10 membranes) also held true for growth since the dif- ference in culture density between 10 and 15 membranes was only 5%. How- ever the highest viable cell concentrations recorded for this dialysis system (283 billion cells/ml) were Obtained when the larger area was used. It is presumed that, on the basis of growth, the Optimum membrane area would vary with the organism and growth medium used. Further in- creases in area were not tested so the Optimum for these culture conditions was not precisely identified. It is possible that subsequent additions be- yond 10 membranes would produce increases in cell concentrations, but of dimishing magnitude as indicated with the comparison between 10 and 15 membranes. Such results indicate that either the diffusional demands of the culture were approached or the efficiency of the dialyzer declined. The first is difficult to determine since the dialysis system might well eliminate the limitations imposed by toxic accumulations or nutrient defi- ciencies permitting growth to increase to a point where another factor Isuch as aeration became limiting. A reduction in dialyzer efficiency is also possible since other data (Section 4 and Section 6) has shown that the addition of dialyzer membranes resulted in an increase in 110 internal volume, an increased fluid hold-up time, and a decrease in oxy- gen transfer efficiency. The performance Of this dialysis culture system equaled or exceeded the levels eXpected from previous work. This was especially apparent in terms of culture concentration superiority over nondialysis, a Specific attribute Of the dialysis system. The growth trials showed that, on Trypticase soy broth, a 5 fold increase in culture concentration was achieved. On water diffusate of Trypticase soy broth, this rose to a surprising 35 fold advantage. These results were, in the first case, equal to and, in the second case, greater than those reported previously (57). Although reasonable high culture concentrations were achieved these did not surpass the densities reported for the original prototype system, which had a smaller membrane area (57). This difference was not necessarily attributed to a poorer efficiency for the present dialyzer because it was noted that the current nondialysis "control" cultures also achieved lower cell densities than in earlier work. It is likely that the above differences in culture yields may be associated with the organisms themselves or with subtle differences in the growth media or the culture procedures. The dialyzer-dialysis culture technique physically separated the fermentor, reservoir, and diffusional exchange regions of the system in- to independently controlled units. This allowed the adjustments in volumes, volume ratios, and membrane area necessary to provide the 0p- timal solute transfer conditions for the best use of various growth media or for the culture demands of various microorganisms. Such an Operational flexibility resulted in two biological attributes of dialysis culture which have not been emphasized; the ability to produce high culture lll yields on relatively weak culture media and the extension Of a stationary growth phase for surprisingly long periods. The selected nutrient medium was found to directly influence the culture densities produced in nondialysis cultures. Here the concen- tration Of viable cells corresponded with the quantity Of carbon source (glucose) contained in the medium with a low carbon water diffusate broth producing the poorest yield. Examination of the concentration of glucose remaining in the fermentor during a growth trial revealed that its reduction to a low level, possibly a critical concentration, coin- cided with the termination of exponential growth. Dialysis cultures with the same types Of media produced greater culture yields with the glucose concentration of the medium having little influence on the ex- tent of growth. Moreover, when the membrane area was sufficiently en- larged dialysis cultures achieved surprisingly high cell concentrations on the low glucose water diffusate broth. In these trials the glucose concentration within the fermentor fell to a level which was over three times higher than that in the above nondialysis and which remained con- stant or rose during the remainder of the trial. Maintenance Of a higher nutrient concentration in the fermentor allowed an extension Of the ex- ponential growth phase, which resulted in the production of higher cul- ture densities and a prolonged linear-stationary growth phase after the 24th hour of the trial. These results indicate a direct relationship between diffusional access to a nutrient reservoir and the creation and maintenance of an adequate culture environment which permits production Of higher culture concentrations regardless Of the strength Of the selected growth medium. 112 The maintenance Of a high concentration Of viable cells in a pro- longed stationary growth phase further illustrated the advantage of dialyzing a nutrient reservoir. This growth phase was continued for over six days without a loss in culture density. After the fifth day the reservoir still contained 60% of its initial glucose concentration. These trials indicated that a dialysis culture could be continuously operated for up to two weeks as a closed system without the risk Of nutrient limitation or culture contamination. Conceivably a single dialysis culture Operated for several days could produce as great a viable cell yield as several batch nondialysis cultures. In addition to the economies gained by such an Operation,the prolonged stationary phase produced by dialysis cultures provided an extended time during which the synthesis Of culture products, many of which are formed only during the stationary growth period, could continue. Long term dialysis cultures might well represent a commercially economical method for ob- taining high yields Of such products. In addition these dialysis cul- tures with or without the adaption of the reservoir to a continuous Operation could represent a useful alternative to the more complex con- tinuous culture system for some fermentation studies. The dialyzer and the dialysis culture system as Operated in these trials was not only compatible with the test organism but demonstrated biological performance which equaled or exceeded that anticipated for the system. Apparently the only influence exerted by a growing culture on the dialysis process were changes in the solute concentration gradient as nutrients were consumed andrlhe dialysis equilibrium concentration. NO appreciable changes in the mass transfer characteristics of the sys- tém were noted. Diffusional access to the nutrient reservoir repeatedly 113 appeared to represent a key factor in the success Of dialysis growth trials. The successful concentration Of viable cells on a weak nutrient, increased duration of the exponential growth phase, the extension Of the stationary growth phase, increased culture viability, and growth reSponse to increased membrane area which were demonstrated characteristics Of dialysis culture were largely the result of access to the nutrient reser- voir. The additional nutrient supply contained in the reservoir effec- tively prevented or greatly delayed the onset Of growth limiting con- ditions within the culture fermentor. The reserve solutes were continual- ly diffused into the fermentor during the growth period. This reduced the magnitude of nutrient depletion within the fermentor during exponential growth and provided a continuing replenishment Of consumed nutrients es- pecially during the later stages Of growth. As a result, higher nutrient levels were maintained in the fermentor throughout the growth trial. These levels represented an improved culture environment which contributed to the attainment Of a higher culture con- centration ceiling for a given set of culture conditions within the con- fines Of a fixed culture volume. Correspondingly the prOportion of vi- able cells within the culture increased especially during the later stages. The dialysis reservoir also represented a sink for removal and dilu- tion of diffusible metabolic byproducts, some Of which become toxic upon accumulation. The growth of a microbial culture is generally limited by the depletion of an essential growth factor or the accumulation of a growth inhibiting factor. It is likely that the increase in dialysis culture lev- els and viability were also due to the capacity of the dialysis system to remove toxic solutes from the fermentor and dilute them within the reser- voir. This function may be the reason larger volume ratios (achieved by 114 increasing the reservoir volume) enhanced culture yields. The dilution of toxic solutes effectively improved the culture environment by pre- venting or slowing the rate of their accumulation within the fermentor. Thus the presence of the dialysis reservoir had the effect Of enlarging, on a molecular basis, the fermentor region while maintaining a constant volume within the relatively small physical confines of a given fermen- tor capacity. The net effect and a valuable attribute of the dialysis culture technique was the production Of higher viable cell concentrations within a small culture volume than would be possible by nondialysis. The separation of the fermentor and reservoir regions within the dialyzer-dialysis culture system provided a convenient method for large scale symbiotic culture schemes. The former was discovered to be feasible vvhen the accidental contamination of the reservoir by Bacillus megaterium vmas found to have no effect on the growth, yield, or purity Of the fermen- ‘tor culture. This indicated that a sufficiently large (13 to 26 liters) rxaservoir contained enough nutrients to Support both an inoculated and a ccnntaminating culture within its confines for at least 48 hours. TMoreover, thus dialysis membranes allowed diffusional exchange between the two cul- tnxres while maintaining separation Of the cells insuring the purity of both culinares. Such an application of dialysis culture to the study of symbiotic The growth of Bordetella pertussis Cultnmres has been reported elsewhere. cultmlres for vaccine production have been enhanced by dialysis against Mbacterium gseudodiphtheriticum in a dialysis flask system (13). The influence Of mixed cultures on a bacterial Species was studied by the use of a dialysis device with membrane filters to simulate a mixed culture en- Virotltnent (140). A study Of the "satellite phenomenon" Of HemOphilus -££l§3£ffanzae was made by dialyzing these cultures against a Staphlococcus 115 culture contained in a cellophane sac (130). Dialysis cells have been used tO determine the associative interrelationships between various microorganisms (117). Each of the above investigations employed small scale or crude dialysis devices. The dialyzer-dialysis culture system, on the basis Of the above results, could be employed for symbiotic studies and would afford the benefit of Operational flexibility and a large scale to such investigations. Finally, the lack of negative consequences caused by reservoir con- tamination indicated that dialysis cultures could be successfully prOpa- gated on unsterilized growth media. This could be accomplished by dialyz- ing sterilized water from the fermentor against the reservoir containing the unsterilized nutrients until a sufficient quantity of solute had transferred to enable inoculation and growth. Such a scheme would pre- clude the need to sterilize the reservoir nutrients and would allow the tise of heat liable growth media without the danger of culture contamina- tion. 5.5 Summary A model plate-and-frame dialyzer (Section 3) containing regenerated cellulose dialysis membranes was incorporated into a dialyzer-dialysis culture system. Liquid cultures Of a type bacterium, Serratia marcescens, were circulated from a conventional 5 liter fermentor through one side of the dialyzer while nutrient media from a separate reservoir was circulated through the Opposite side. A series of growth trials established the ex- tent Of variation between repeated results and demonstrated the superiority of dialysis cultures, in terms of culture viability, culture concentration, and extension of the exponential and stationary growth phases, over compar- able nondialysis cultures. Fluid circulation velocities had no effect on dialysis growth. A 1:10 fermentor to reservoir volume ratio, obtained by increasing the reservoir volume, produced Optimal biological performance. The addition of dialyzer membranes resulted in improved culture concentra- tions with 15 membranes (.4320 m2), producing a maximum viable cell concen- tration Of 283 billion cells/ml. Diffusional access to the nutrient reser- voir was shown to be instrumental in maintaining a culture environment which permitted the high dialysis growth yields, an extension of the sta- tionary phase for over five days, and the concentration Of cells on a relatively weak growth medium. In several instances the dialysis growth response to Operational conditions was similar to that shown previously for nutrient transfer. 116 6. DIALYSIS AERATION 6.1 Introduction and Objectives The use of membranes for the diffusional exchange of nutrients and cellular products during microbial growth can be readily extended to in- clude the exchange of gaseous nutrients and products, namely oxygen and carbon dioxide. Dialysis aeration was first demonstrated 22 years ago by Gladstone (64) for the production of protective antigens from Bacillus anthracis cultures. Air was passed over the surface of cellOphane sacs containing the culture. More recently an attempt was made to employ this technique for the growth of Serratia marcescens cultures in the bottom Of a dialysis flask (61) and in a small dialysis fermentation system (62). Gas or gas-saturated liquid contacted the membranes and diffused through the aqueous channels Of the membrane material. Although this system transferred sufficient oxygen when a relatively porous mem- brane filter was used, the resulting cell density and growth rate were less than in a conventionally aerated control. Independently, Brewer (l9) develOped and patented a relatively crude apparatus which employed an outer envelOpe made from polyethylene sheets for gas transfer to an aerobic culture contained within the device. The membrane barrier is classically considered to be a micro-hetero- porous seive with pores or Open spaces which fill with the liquid solvent. frhus solute (gas) is transferred by diffusion through the bulk solvent filling the membrane not through the membrane material itself. Selective 117 118 permeability under these conditions is a function of solute dimensions (volume) and pore diameters as well as electrostatic forces, entrOpy factors, absorption, and membrane swelling. The rate Of transport is inversely proportional to the square root Of the Inolecular weight, dif- fusivity in the fluid, and bulk fluid movement (7, 159). In the past decade a new type Of membrane has been develOped com- pletely devoid Of pores, channels or crystallinity (16). These membranes are typically fabricated from methyl or dimethyl silicone rubber or from fluorocarbon polymers and are called permselective films or solute trans- port membranes. TranSport through this type of membrane is a function of solubility into and diffusion through the membrane polymer. Gases dis- solve directly into one face Of the membrane, diffuse through its thick- ness, and escape from the other face. Silicone rubber, in addition to being strong, is selectively and highly permeable to oxygen and carbon dioxide but not to water. Their usefulness for biological applications has been demonstrated in prosthetic artificial lungs (45, 38), artificial placentas (138, 172), anesthesia administration (51), and underwater life support (128). The develOpment and availability of permselective membranes and their successful use in blood oxygenation suggested applicability to the prOpagation of aerobic microorganisms. We first tested this idea in a dialysis culture system in 1965 and reported preliminary results in 1966 (H. E. B. Humphrey and P. Gerhardt, 1966, American Chemical Society 152nd Meeting, New York, Abstracts Of Papers p. Q-2). Three basic experimental schemes were used and analyzed in this study: a batch scheme utilizing conventional sparging and agitation as a control system; a dialysis aeration scheme utilizing a dialyzer and 119 membrane exchange as the sole source of aeration; and a full dialysis scheme utilizing one dialyzer for aeration and a second dialyzer for nutrient supply. The third scheme represents the ultimate in dialysis culturing -- a microbial population solely supported by exchange through membranes and totally isolated from the exterior environment. The sys- tems were evaluated for oxygen transfer characteristics and then evaluated for microbial growth characteristics. The potential advantages expected for dialysis aeration are as follows: A) Economy of Operation and Safety; (a) The need for air ster- ilization is eliminated, (b) the required quantity Of anti- foam materials is reduced, (c) the existence of positive pressure within the culture vessel is eliminated, thereby reducing the hazard of aerosol contamination, and (d) the membrane provides a known and defined gas-liquid interface, the gas and liquid films of which may be manipulated by velocity adjustments. B) Improved Culture Environment; (a) C02 stripping is reduced, (b) the denaturing effect of violent agitation and direct gas-cell interfacial contact is eliminated, (c) the evap- oration Of culture liquid is reduced, (d) the uniform and even diffusion of gas molecules into the liquid eliminates the fluctuation in dissolved oxygen associated with sparged gas bubbles, and (e) the character Of the dissolved gas en- vironment is better controlled by partial pressure adjust- ment in the gas phase prior tO membrane diffusion into the liquid. IMH‘II. - 120 The Objectives Of this section of the thesis were: (a) to assemble a historical and theoretical background on aeration of microbial cul- tures; (b) to examine and compare quantitatively the oxygen transfer rates of dialysis and sparged aeration systems; and (c) to examine the feasibility of using permselective membranes for gas exchange in mass culture of a typical aerobic bacterium. 6.2 Historical and Theory 6.2.1 Oxygen demand of microbial cultures. Oxygen transfer is commonly regarded as a critical factor in the development Of an aerobic fermentation process. An aeration system must be designed to provide an adequate dissolved oxygen environment for ex- ponential multiplication and other physiological phases. The success of an aerated fermentation will depend on Operational efficiency and on the influence which the resultant physical and chemical conditions exert upon the growth and metabolism Of the organism. A large portion Of industrial research has been concerned with the measurement and manipulation of these conditions in an attempt to Optimize the desired microbiological process. An equally concentrated effort has been directed toward examination of the organisms themselves with reSpect to oxygen requirements, since the quantity of dissolved oxygen required by a multiplying culture largely dictates the design Of the aeration system. Oxygen demand is a function of both the physiological state and the concentration of the culture. Clifton (27) demonstrated that Aerobacter aerogenes or Escherichia coli cultures had the highest uptake on a per cell basis when young but that the overall culture demand was greatest later, during the decline physio- logical stage when the high concentration Of cells dominated uptake rates. This was confirmed later by manometric measurements of Streptomyces griseus cultures(68). Zobell and Stadler (175) demonstrated that the oxygen demand Of lake bacteria rose, both on the per cell and total 121 122 overall basis, when the nutrient concentration was increased. Thus Optimal or super-Optimal nutrient conditions can render a standard aera- tion supply system inadequate and create a condition in which oxygen is growth-limiting. This condition has been experienced in dialysis cul- ture. Supplemental addition of nutrients permitted a culture concentra- tion Of l x 1012 organisms/ml, and the cessation of growth was attributed to insufficient oxygen transfer, which had been adequate at lower cell concentrations (57). Provision of oxygen is further complicated by changes in metabolic function. Sporulation, for instance, creates a peak demand period as adaptive enzymes begin synthesizing complex organic bonds and molecules (71). The physiologic state, the metabolic function and the population density of the organism appear, therefore, to repre- sent important considerations in the design, operation, and apparent ef- ficiency Of a culture aeration system. Overall oxygen demand constitutes a limiting condition which must be satisfied by fermentation equipment in order to achieve Optimal aero- bic growth. Finn (48) noted that relatively little was known about the peak demand of cultures actively growing in rich nutrients. He found that most of the reported values had been Obtained under artificial con- ditions different from those Of the actual fermentation. Such data did not reflect conditions within the culture nor the influence of environ- mental factors as growth proceeded. Factors considered important to oxygen demand were nutrient concentration, accumulation of toxic and products or loss of volatile intermediates, nitrogen source and growth factor quantities, oxygen supply, and cell accumulation. Some of these affected oxygen solubility and Others, cellular respiration. 123 The lack of accurate demand values was primarily due to the lack Of a method for measuring dissolved oxygen, and therefore oxygen demand, within the culture itself. The same problem also prevented accurate evaluation of aeration equipment. The methods which have been used until recently include: manometric analysis of removed samples, titri- metric analysis Of cell-less sodium sulfite solutions, and polarographic analysis. In each case the data Obtained failed to represent an accurate picture Of the actual situation because a substitute solution was used, samples had to be removed, cells were not present, or the probe used was subject to deterioration by the culture. The develOpment of a membrane-covered, bimetallic electrode by Clark (26) represented a break-through in measuring dissolved oxygen. Subse- quent improvement in the electrode and membrane materials (66, 82) and the perfection of a reliable, long-lived, steam-sterilizable oxygen electrode (86, 15) have permitted precise and continuous monitoring Of the culture during a fermentation process. A "dynamic" (i.e. continuous) measurement technique for evaluation Of "in situ" oxygen transfer, oxygen concentration, and culture demand has been develOped (156) as a result Of the availability of such oxygen probes. The method reportedly produced results as precise as those from a calibrated manometer, but without the problems and errors associated with other detection methods. The value of this "dynamic" technique is exemplified in Table 15 (below). The data, from an active Serratia marcescens culture, show both the occurrence and magnitude Of oxygen demand for the culture as the fermentation progressed. This information not only identified the peak oxygen transfer but also in- dicated periods when such transfer could be reduced. The latter repre- sents an Operational economy and permits elimination of potentially toxic excess Oxygen conditions. 124 Maintenance of Optimal oxygen conditions as well as satisfaction of culture demands have been design goals in industrial fermentations. A tailored aeration system, which produces an Optimal condition through- out the physiological stages of the culture, can result in advantageous cell or product yields. In an effort to avoid either excessive or de- ficient oxygenation, complex and expensive monitoring and control de- vices have been develOped. These detect, record, analyze, and adjust aeration in order to maintain culture aeration within predetermined Op- timal levels. Generally the components include oxygen probes, small computors, and elaborate automated valving, flow meters, and gas selec- tors for delivery of the correct oxygen concentration to the sparger (72, 108). Realization Of the importance Of adequate supply and maintenance of dissolved oxygen for the productive success of aerobic fermentations has justified the development Of more efficient fermentors and aeration monitors. 6.2.2 Oxygen states Because oxygen functions as the primary metabolic electron acceptor, aerobic microorganisms are sensitive to excesses or deficiencies. A cul- ture exposed to subOptimal conditions may reSpond with an altered physi- ology or submaximal growth and viability. With reSpect to oxygen a fer- mentation may eXperience a deficient, optimal or toxic state depending on aeration efficiency and control. A deficient oxygen state may result in death, growth inhibition, or metabolic disruption in bacteria. Oxygen defficiency was shown to occur dramatically in Acetobacter cells within 15 seconds (48) after air flow 125 disruption. Oxygen supply and concentration also influences the dura- tion Of the exponential growth phase, which in turn affects the concen- tration of cells produced. Rahn (25) demonstrated that oxygen exhaustion restricted the growth of Pseudomonas fluorescens in an ordinary test tube to a population level of 2 x 106 cells per ml. ‘More recently, oxygen also was shown to limit the concentration of Serratia marcescens in dialy- sis culture at a level of 2 x 1012 cells per ml (57). Other manifestations attributed tO insufficient oxygen conditions include a shift in Operating enzyme systems, a change in physiological function (inhibition Of sporulation), a decrease in product formation, or a change in growth rate (44, 71, 18, 105, 97). Such results apparently occur because the oxygen concentration falls below a minimal tolerance. This minimum has been defined as the critical oxygen tension, a term in- troduced by Gerard and Falk in 1931 (73) to describe the point at which growth rates cease to be independent Of oxygen concentration. The value for this concentration varies with temperature, the organism, and measure- ment precision. There has been little agreement on the level of the critical tension, largely because of the lack Of precise oxygen determination at low Con- centrations. Reported values for bacteria range from .003 to .005 m moles Oz/L (48, 100). Not only has the value of the critical oxygen tension been unsure but until recently little was known about the relationship between oxygen concentration and cellular respiration rates at oxygen levels approaching and within the critical range. This relationship was recently evaluated and categorized with respect to the critical oxygen concentration value (73). Oxygen concentration above the critical value had no effect on respiration rates and represented an excess condition. 126 Concentrations below the critical value directly influenced respiration rates and represented a limited condition. Between these two a trans- ition condition was detected where respiration rates and oxygen con- centrations fluctuated in an unique manner. The unusual oscillations found in the transition stage led the authors to propose a new defini- tion of the critical oxygen law; "the critical oxygen tension is that concentration above which the respiration rate Of an organism is inde- pendent Of changes in dissolved oxygen concentration and beIOW‘WhiCh dissolved oxygen and oxygen uptake may vary" (73). If oxygen levels are not maintained above the critical value for the duration of a fer- mentation exponential growth rates, viability and culture density will decline (5, 165). The Optimal oxygen state represents the range of oxygen concentra- tions which are ideal for maximal cell growth and function. The dissolved oxygen concentrations characteristic of this state are specific for each species and occur between the critical low values and toxic high values. The principal goal Of fermentation design, scale-up, and Operation has been the attainment and maintenance Of an Optimal oxygen state, the im- portance Of WhiCh has justified development Of the elaborate oxygen sens- ing and control devices mentioned previously. Despite these efforts an optimal oxygen state seldom is maintained throughout aerobic liquid fer- mentations. The inherent inefficiency of liquid aeration and the varia- tion in culture oxygen demand as the cells age and multiply are the causa- tive factors. In addition, the Optimum oxygen state for product synthesis or a physiological function such as Sporulation is Often different frOm that for growth (67, 71). Fermentor aeration techniques are generally in- sensitive and unresponsive to changing demands and therefore are unable 127 continually to produce the necessary Optimal conditions, with a result- ing loss in efficiency and product yield. Thg_£2§ig_oxygen §£§£g_represents conditions in which excess oxygen retards growth or inhibits a metabolic function. Although this state is less recognized than the previous two, it was implicated in restricting the length of the lag growth phase which occurred after inoculation of an aerobic fermentation, eSpecially if the cells were young, the inocula small or the nutrients dilute (175). In addition, excessive oxygen con- centrations have been shown to inhibit antibiotic synthesis in certain fermentations (18). Oxygen toxicity has been attributed to shock or respiratory enzyme poisoning from sudden exposure to high concentrations, slow adaption of key enzyme systems to high oxygen levels, unfavorable conditions due to carbon dioxide or nitrogen stripping, or oxidation Of metabolic inter- mediates (126, 160, 165, 35). Although use of pure oxygen instead Of air generally magnifies the inhibitory effect, some organism such as Aspergillus niger are stimulated by excess oxygen concentration (173). Thus, oxygen toxicity probably represents a combination Of effects which do not affect all organisms in a universal manner. However, it is pos- sible that a toxic state may occur at some point, especially early in the fermentation process. More importantly, sparger aerated cultures may expose a portion Of the culture directly to gas bubble interfaces at any given moment. This represents a source of oxygen contact at the cellular level which may be inhibitory to sensitive cells, even through the overall dissolved oxygen level is within acceptable limits. The frequency of such cell-bubble contact is regarded sufficient enough so that it has been prOposed as an alternate pathway for oxygen transfer (12). 128 The three oxygen states and their influence on culture viability or yield take on added significance when fermentor aeration is con- sidered at the cellular level. Under existing Sparged and agitated liquid aeration techniques, an individual cell may alternately be directly exposed to air bubbles, exposed to varying concentrations of dissolved oxygen, and exposed to no dissolved oxygen. The idea that cells become oxygen starved during deep liquid fermentation was prOposed as an explanation for the difference in duration Of exponential growth in a shake flask versus that for deep liquid culture Of Pseudomonas fluorescens (125). The authors concluded that a disproportion existed between oxygen demand and supply in deep liquid cultures, which resulted in death or slower growth. More recently the dissolved oxygen concen- tration was found to vary throughout an aerated fermentor broth, and the variation was attributed to inadequate bulk mixing (123). In addition, realization that an area Of decreasingly concentrated oxygen trails behind sparged air bubbles as they rise led to the conclusion that bacterial cells are exposed to varying degrees of oxygen concentration as they circulate through these bubble trails. The variation in dissolved oxygen throughout the liquid, was reported actually to be stagnant under some mixing condi- tions (116) so that the actual critical oxygen concentration may be higher than commonly believed. Although overall aeration may appear adequate on the basis of gross dissolved oxygen analysis, it thus appears that the cells themselves alternate between regions Of Optimal and subOptimal oxy- gen cOnditions as they circulate through the Sparged fermentor. It is difficult to compare this effect precisely against the success of a fer- mentation. However, the physiological influence of these suboptimal oxy- gen states strongly suggests that improved aeration at the cellular level would result in superior fermentation yields overall. 129 6.2.3 Nondialysis aeration methods Adequate delivery of oxygen from an air stream into a fermentation broth has been one of the most critical as well as most difficult prob- lems facing fermentation technologists. A wealth Of literature has ac- cumulated concerning the design, improvement, and Operation of fermentor aeration equipment. The oxygen transfer process has been examined and theoretical equations, based on mass transfer principles, have been de- velOped to explain Observations, facilitate predictions and aid perform- ance evaluation. NO attempt will be made to review fully all aSpects of fermentor aeration as several pertinent and comprehensive reviews have been published (39, 48, 149, 127, 104, 161) Early research in fermen- tor aeration emphasized develOpment of mechanical devices for transfer Of gaseous oxygen from the air stream into the liquid. The low solubility of oxygen in aqueous liquids makes this transfer difficult and has focused attention on development of efficient Operational techniques. Liquid aera- tors have been classified according to the way they distribute air. Those which have practical application to industrial fermentors include fixed metallic distributors with bored holes (e.g. single or multiple-orifice, hollow-tube spargers), finely porous aerators (e.g. sintered glass or metal foam bubblers), fixed distributors with various Openings other than bored holes, and mechanical air distributors (e.g. rotating hollow porous agita- tors) (39). Various modifications have been made on the basic aerator designs and on the associated equipment, agitators and baffles that are found in modern fermentor aeration systems. The German-designed Waldhof fermentor represents one of the successful variations. This design popularized the 130 use of a central draft tube suspended over a hollow shafted agitator which dispersed air bubbles near the bottom of the vessel. Additional air and any foam created during aeration was drawn down the draft tube and rediSpersed in the liquid. Lower air requirements and elimination of antifoam agents make this design desirable although power require- ments are greater than more conventional fermentros (142). It is be- lieved that the liquid vortex depth which is enhanced by the draft tube is related to oxygen transfer and a key to the effectiveness of the Waldhof design (158). Another aeration system eliminated the need for compressed air by utilizing the suction created by rotating turbine blades within a hollow cylinder to draw air into the cylinder from an adjacent intake pipe (83). A similar device utilizing a rotating tur- bine and fixed stator blades through which air bubbles are dispersed horizontally into the liquid was recently reported (42). Other modified systems include the use of acoustic air velocities through a single ori- fice sparger to create very small bubbles and violent agitation (2) and the use Of a sparger which may be moved vertically to better locate gas dispersion and improve gas-liquid contact time (49). The search for superior aeration techniques has not been limited only to the modification of conventionally designed equipment. Ingenious methods have been prOposed which utilize chemical reactions, electrical currents, or pumps and valves. One Of the most successful techniques, one which set a precedent for circulation of a culture to an aerator on the exterior of a fermentor was devised by Strich (154). The culture broth was pumped from the upper level of the fermentor through an ex- ternal helical gas exchanger and returned to the bottom level of the fermentor. Aeration was accomplished by passage of very fine gas bubbles 131 from a continuous air stream through a microporous, (20-60/‘diameter) ceramic or metallic diaphragm into the circulating culture liquid. This system produced extraordinarily fine bubble aeration and afforded control Of oxygenation rates through manipulation of the gas and liquid velocities. Another unique aeration technique employed a pulsating aerator which uti- lized a single air filter and a siphon device to produce positive and nega- tive air pulses. The positive pulse forced air bubbles into the culture, the negative pulse removed foam and exhaust air from the culture, and the succeeding positive pulse forced fresh air and the foam into the culture. The culture is maintained under negative pressure, a desirable feature for growth Of pathogenic organisms (74). Generation Of adequate oxygen for submerged culture Of Pseudomonas fluorescens by electrolysis of the cul- ture medium has been demonstrated on a small scale (137) and an electro- chemical cell employing a KOH solution has been shown to concentrate oxy- gen from the air into 99.5% pure oxygen at a rate of .21b/hr (17). Finally, periodic additions of H2 02 to a culture containing catalase have been shown as a satisfactory method for generation of oxygen "in situ" (169). Although such prOposals are ingenious, their application to large scale or commercial fermentations is limited by the costs involved. It has been found that, ideally, an aeration system should provide for an even distribution Of gas throughout the liquid and for a maximal gas- liquid surface contact for a maximal length Of time. Consideration must also be made for the design of fermentor tank geometry, baffle placement, volume and height ratios, agitator design and operation, and air velocity (112, 81, 104, 48, 53, 161). These considerations increase in signifi- cance when the design of large scale deep liquid aerators is attempted. 132 As shown, the principle techniques for improving aeration efficiency have involved manipulations which are designed to raise the value of the oxygen transfer coefficient (K a). This approach continues to be held in L high esteem by fermentation technologists (161) and with Finn's conclusion that the principle means for aeration improvement lay in attempting to in- crease KLa values or raise the driving force (C* - CL) (48). The latter would require hyperbaric oxygen conditions or the use Of pure oxygen, both Of which are economically impractical on a larger scale. The Objective in fermentor aeration has been to assure that dissolved oxygen is maintained in excess Of the critical concentration. Although the importance Of this has been recognized, there has been little agree- ment on the value Of the critical concentration (55). It has generally been assumed that oxygen uptake rates and respiration rates would be in- dependent Of fluctuations in dissolved oxygen concentration when the fer- mentor was Operated so that the detectable liquid oxygen concentration never fell as low as the critical level (123, 127). Wise condluded that this type of Operation was basically inefficient and wasteful of oxygen (166). The variations in dissolved oxygen observed both at a given point and throughout the fermentor have been attributed to the inefficiencies Of mixing and have not been regarded as important to the fermentation as long as the overall level was above the critical tension (123). Although these assumptions are partially correct, they fail to recognize a funda- mental consideration in microbial aeration, namely that the overall "macroscale" oxygen conditions are magnified, with respect to the re- spiring cells, in the "microscale" cellular environment. Thus, inef- ficiencies in mixing or sparging which produce varying dissolved oxygen levels in a bubble aerated fermentor actually produce pockets within 133 the culture which are suboptimal or even critical for aerobic growth. Cells present in these locations for even a short time may have their essential metabolic functions disrupted to some degree. In addition the repeated encounter with optimal, subOptimal, and excess oxygen conditions may produce uneven metabolic rates and potentially submaxi- mal culture yields. It is entirely possible that a consistent and fully adequate oxygen environment at the cellular level is unattain- able by bubble sparging. 6.2.4 Mechanism of oxygen transfer The conclusion that fluctuating oxygen conditions exist within con- ventionally aerated fermentors is plausible in view Of past and ongoing attempts to improve oxygen transfer by manipulation of agitation, sparger design, and gas velocity. If the importance of the micro-level oxygen environment is significant and if the inadequacy Of bubble aeration is an acceptable premise, then there exists a need for an alternative method of delivering oxygen to the cell suspension. Dialysis aeration is pro- posed as a feasible alternative. In order to understand dialysis (mem- brane) aeration and to appreciate its relationship to conventional aera- tion, the oxygen transfer mechanism and theory should be examined. The solution Of oxygen into water or a culture broth is considered to be a gas absorption process in which gas molecules are transferred from gas bubbles (gas phase) through a gas-liquid interface and into the liquid phase. This represents a mass transfer in which the interface and the thin gas and liquid films adjacent to it are the principle rate- determining factors. The process is commonly explained by the Whitman 134 two-film transfer theory, in which the stagnant films adjacent to the gas-liquid interface are assumed to control the rate of mass transfer (3). The Whitman theory provides the basis for derivation Of the familiar gas transfer laws and volumetric mass transfer coefficients from the basic mass transfer equation, N=Ko (AC)A , (6.1) where; N = the mass transfer rate Ko = the overall mass transfer coefficient C = the oxygen concentration driving force, and A = the interfacial area between the gas and liquid phases. Oxygen transfer from.the gas phase to cells suspended in liquid in- volves a series of transfer steps from the bulk gas to the gas film, the gas film to the interface, the interface to the liquid film, the liquid film to the bulk liquid, the bulk liquid to the liquid film surrounding a cell, and the cell liquid film to the interior and/or enzyme receptors Of the cell (see Figure 32). This concept was first applied to oxygen transfer by Bartholomew (8) twenty years ago. A mass transfer equation, similar to equation 6.1, may be written for each Of these steps. The mass transfer coefficient, k, for each step is characteristic of the trans- fer in that region. The rate of transfer in each step Of this process is determined by the concentration gradient and the resistance to transfer characteristic for each region. The latter is 135 designated as the reciprocal of the mass transfer coefficient in each case (21 ) and the overall resistance to mass transfer is also k the sum of the individual resistances (l = + l + l ) (127). 1 KO k1 k2 k7 As shown in the diagram (Figure 32) oxygenation Of liquid involves five oxygen transfer steps with an additional two steps involved in trans- fer into the interior Of the cell itself. As previously stated, the inter- face and the adjacent gas and liquid films (k2, k and k4) are regarded 3 as the key steps in this mass transfer process. The other steps (k1, k5, k6, and k.7 ) are considered to be far less important in comparison. Experimental evidence has shown that the bulk gas (k1) and bulk liquid (k5) transfer steps are diffusional processes, Obey Fick's law Of dif- fusion (J = -D 513 ) and do not limit the rate of oxygen transfer dx (127, 15). The transfer steps associated with cells (k6 and k7) have no influence on oxygen absorption into the liquid and therefore are significant only to the biological aspect Of aeration. Borkowski and Johnson (15) Showed that the liquid film surrounding the cells had no importance to cellular aeration as long as the cells were suspended with- in the liquid. The transfer step (k7) into the cellular material may be disregarded in the case Of bacterial cells where the enzymatic oxy- gen receptors lie on the plasma membrane. With cells such as yeast where these sites lie deeper within the cell matrix, diffusion through the protOplasm may become limiting when very low oxygen concentrations are encountered (87). The foregoing examination of the mass transfer steps supports the emphasis placed on the gas-liquid interface and its films by the Whitman 136 Figure 32. Diagram Of the transfer of gas molecules from the gas phase to a cell suspended in the liquid phase. A mass transfer equation may be written for each step of the transfer process with k1 - k7 repre- senting the mass transfer coefficient in each case. 137 2......“ 050.1. 9:0... 38 . 2...“. _ _ 9:0... wUdEmsz— N x 53.“. m<0 r‘wcio.~‘ -' a . W‘ ‘MICHWI‘HR‘. :3 f .’:.‘.-“..'~‘£'." I mqe xnnm Figure 32. 138 theory. Numerous experimental determinations have directly confirmed that these most probably do represent the key rate limiting steps in oxygen transfer (22, 8, 48, 171, 115, 80, 127, 3). Although unaminous consensus does not exist, most investigators feel that the liquid film represents the greatest resistance to mass transfer and therefore con- trols the aeration process. For example, investigations concerning measured thermodynamic effects on KLa values and on diffusion coefficients for aqueous and gas phases as well as the low viscosity Of gas in compari- son to liquid, showed that the gas film would have to be far thicker than found at the bubble-liquid interface in order to be a significant rate limiting Obstacle to gas transfer (48, 8). Thus the assumptions of the Whitman theory Of gas transfer appear valid. This theory with its ems phasis on the gas and liquid films is, along with interfacial considera- tions, essential for understanding dialysis aeration. A summary of the key features Of the two-film.theory and the development of the familiar gas transfer equations follows, and is based on explanations which have appeared in two texts (149, 3). The important steps for oxygen absorption in an aqueous solution involve the gas film, the interface, and the liquid film. Mass transfer equations may be written to represent oxygen transport through each of these regions the sum Of'WhiCh represents the overall mass transfer pro- cess; N = KO (ACOZ)A = kg (P - P1)A = ki (Pi - Ci)A = kl (c:i - c )A , (6.2) L ll‘ Thus the overall concentration gradient driving force , IAC (P - CL) , O2 02 is a composite Of the gradients across the gas film, interface, and liquid 139 film regions. Each region is characterized by its mass transfer coef- ficient which is largely determined by the resistance to mass transfer found in the region. As discussed previously the overall resistance to transfer is a composite sum of the individual steps involved l/K = l/k + l/k, + l/k . (6.3) 0 g 1 L In magnitude of importance the resistance to tranSport offered by the interface itself is considered to be negligible. Thus equation 6.2 can be simplified by elimination of the interfacial term; (6.4) N = KO A (P-CL) = kg A (P - Pi) = kL A (Ci - CL) . Oxygen solubility in aqueous liquids is low (7.6 mg/ml at 30°C for air saturated water, slightly less for biological broths), is dependent on the gas phase oxygen partial pressure, and obeys Henry's Law C02= I/H P02 or P02 = H C02 . (6.5) Because saturation of the liquid with oxygen is the Object of aeration the process aims toward achieving an equilibrium.between the gas and liquid phases, where Pi = C1: or P* = 0* , ‘where the * represents the equilibrium values of concentration and partial pressure. Equation 6.4 may therefore be written; N = KO A (p - CL) = kg A (P-N) = kL A (c*-cL) . (6.6) Applying Henry's law under equilibrium conditions , P* e H c H 0* e P . (6.7) Substituting this into equation 6.6 gives a mass transfer rate equation written in terms Of oxygen concentration; = . = * .- -2 * .- N K() A (P CL) kg A H (C CL) kl. A (C CL) . (6.8) Steel (149) showed that H l for air saturated broth permitting this factor to be removed in the above equation 6.8 which may be further simplified; = ‘k .- = ... N Ko A (C CL) kg + KL A (6* CL) . (6.9) It was previously shown that the liquid film is believed to control the rate Of oxygen transfer. This implies that, _l_,> _l_ , and although k k L 1 -1 +1 +1 K k (6.3) the other resistances are insignificant with respect to the liquid film allowing the following assumption to be made, 1 m l and Koch KL. Ko kL Since the individual film coefficients can only be estimated by graphical interpolation the overall term KL is comonly used. Therefore, the overall oxygen transfer equation 6.9 may be made in terms of the liquid film alone which permits evaluation of oxygen transfer rates (OTR) and oxygen transfer coefficients on the basis of bulk liquid dissolved oxygen cencentration measurement, N = KL A (0* - CL) . (6.10) 141 The gas-liquid interfacial area varies with fermentor volume and Opera- tional procedures. Since the rate of oxygen dissolution is proportional to the interfacial area and is inversely proportional to the liquid volume used, the oxygen transfer equation is commonly written in terms of the volumetric mass transfer coefficient which combines the area and volume factors, ‘A = a . Thus KLA/V = KLa ‘where KLa is the volumetric V mass transfer coefficient and the units used are; KLa = l/min, KL = cm/min , A = cm? , V = cm. , and N = d§_= OTR , (6.11) dt where OTR is m Moles 02/L/Min . With these definitions the oxygen trans- fer equation 6.10 takes the form which has been commonly used by fermenta- tion technologists ; OTR = KLa (0* - CL) . (6.12) Olson and Johnson (118) were the first to use the KLa term as a criterion for gas transfer evaluation. It has since become the standard index for reporting the overall oxygen absorption capacity of fermentor aeration equipment and a valuable means for comparison Of different fer- mentor designs or sizes. 6.2.5 Principles and Problems of Bubble Aeration Aeration of liquid cultures is, as previously described, conven- tionally accomplished by gas bubble production and mixing within a cylindrical fermentor containing a sparger and driven agitator. The establishment of an equilibrium between dissolved oxygen in the liquid and the gas (bubbles) phase represents the goal of the process. Ideally this would be accomplished by the total dissolution of each Sparger 142 produced gas bubble as it rose and was gently mixed to evenly distrib- ute the dissolved oxygen throughout the liquid volume. Unfortunately such ideal oxygen transfer does not occur,mainly because bubbles re- leased into the bottom of a liquid tank are dynamic and do not behave uniformly. Because of this, oxygen is not uniformly dispersed in the liduid, much of the gas remains in the gas phase and as a result, aera- tion efficiency is relatively low, about 1% to 2% (48). The key elements in sparger aeration are the maximization Of the bubble-liquid interfacial area and the maximization of the duration of contact between bubbles and liquid. This agrees with the importance placed on the interface by the Whitman two-film oxygen transfer theory, as explained previously. In an attempt to improve oxygen transfer rates fermentation techno- logists have attempted to increase the interfacial area by manipulating operational procedures and/or equipment design, in order to increase the air volume input, produce smaller bubbles, or delay the escape of bubbles and improve mixing (104). Other techniques designed to improve oxygen solubility such as raising the temperatures, gas pressure, or oxygen partial pressure have not proven economical on a large scale basis. Flushing high volumes of air through a fermentor provides a large quantity of gas, usually with some back pressure within the vessel (which raises the partial pressure of oxygen), but also is very inefficient. High flow rates cause large bubble formation, bubble coalescence, exces- sive foaming, creation Of gas channels within the liquid, and air leak- age around the agitator shaft and other seals. Thus any advantage gain- ed by the additional oxygen is offset by the disadvantages Of this pro- cedure. 143 A high rate of liquid agitation will also improve oxygen transfer rates because it will improve bubble dispersion throughout the liquid, reduce bubble coalescence, delay the escape of bubbles from the liquid, and increase the effective interfacial area of the bubbles by moving them about, thereby reducing their liquid film thickness and producing surface regeneration (a net increase in interfacial area) (33, 41, 23). Vigorous agitation reduces stagnant zones within the liquid and in- creases the overall interfacial area for oxygen transfer per given vol- ume Of gas sparged. However, there appears to be an optimal speed above which undesirable effects outweigh any improvements in efficiency. Ex- cessive agitation may produce bubble coalescence and "flooding" around the impeller blades, cause excessive liquid evaporation and foaming, produce over-aeration conditions which may retard cell growth, or pro- duce a shear effect which may damage the cells (14, 104, 23). The total surface area exposed to the liquid and the internal par- tial pressure Of oxygen are inverse functionsof gas bubble diameter (32, 2). Pattle concluded from his studies of gas solution from rising bubbles that small gas bubbles held in suspension for as long as pos- sible produced the most effective oxygen transfer (119). In addition, small gas bubbles tend to rise slowly and dissolve more completely (33). In order to exploit the advantages of small bubble aeration investigators have attempted to design and operate aeration systems which could produce consistently fine gas bubbles, with the ideal being about .2 and .4 cm in diameter (41). For example, a single orifice sparger was Operated at acoustic air velocities, and a "P-jet" Sparger was used torelease small bubbles (2, 119). Although these systems are effective, they require either excessively high Operating pressures or impractically small 144 orifice Openings. Therefore the commonly used Spargers produce a variety Of bubble sizes and represent a compromise between efficiency and econom- ics. In an attempt to understand and improve bubble aeration some excel- lent studies have been made of bubble dynamics. Photostrobe measurements and high speed motion pictures have been used to study the fate of bubbles after release from the sparger (174). Upon release a bubble rises, dis- solving as it moves and leaving a trail Of oxygen rich liquid (123). Bub- bles of different sizes were found to behave differently and rise at dif- ferent rates. Small bubbles act as spheres and larger bubbles change shape while ascending (96). The initial size determines the height trav- eled before disappearance, although mixing will influence this because it changes the bubble path from vertical to a more dispersed path (119). Hy- drostatic pressure, initial bubble volume, and liquid viscosity influence the rate of rise. This rate would be eXpected to increase as a bubble ascends because the hydrostatic pressure is reduced allowing an increase in size and lift. However Calderbank has shown that bubbles of a given size actually rise at a constant velocity (22). Bubbles suspended in liquid tend to coalesce, which reduces the net surface area and increases the rise velocity. In addition to diSperSing bubbles, mixing also re- tards coalescence. The addition of hydrOphilic organic substances such as lactic acid, glycol, and glycerol have also been successful in retard- ing coalescence and allowing prolonged bubble life and surface area (22, 23, 119). Examination of the literature reveals that bubble behavior is both dynamic and unpredictable. The foregoing description of sparged bubble behavior represented idealized observations under defined experimental 145 conditions. The conditions within a fermentation are far more com- plicated and profoundly influence bubble dynamics and the efficiency of oxygen transfer. For example Calderbank observed that the rise velocity of gas bubbles was independent of the liquid viscosity (22). However, fermentation broths. ’ ex- perience viscosity changes with the age and growth Of the culture. This has been shown to cause an 85% reduction in oxygen transfer rates as com- pared tO only an 8% reduction in cell-free medium (40). It is probable, therefore, that oxygen transfer efficiency, which is influenced by bub- ble dynamics and oxygen solubility, is directly related to the surface tension, the viscosity, and the presence of solutes, antifoam agents, and suspended particles which are characteristic of fermentation broths. In addition these liquid characteristics along with Sparger design, fer- mentor geometry, and agitation influence bubble shape, rate of formation, coalescence, velocity Of rise, distribution, interfacial area, and dura- tion of existence (37). It may be concluded that in conventional sparged and agitated fer- mentors adequate liquid aeration is best accomplished by the production Of relatively small bubbles which are held in suspension as long as pos- sible and are well diSpersed throughout the liquid. Although careful selection of equipment design and Operation may improve oxygenation ef- ficiency up to 30%, stagnant regions continue to exist within the fer- mentor and sharp dissolved oxygen gradations continue to exist within the liquid. As a result, cell yields do not always correspondingly im- prove with improved aeration efficiency (150). Therefore it appears that the bubble aeration technique, although able to deliver a net trans- fer Of oxygen to liquid, is inherently inefficient and may, as a result, 146 limit the optimal oxygenation of fermentation broths, especially at the microscOpic-cellular level. 6.2.6 Mechanism of Membrane Oxygen-Transfer The transfer of oxygen from the gas phase through a membrane into the liquid phase is similar in principle to oxygen transfer from bubbles, except that the phases are separated by a defined membrane interface which has transfer characteristics unique to the membrane material used. Transfer of solutes through solid or membrane barriers has been adequately examined and discussed by Barrer (7), Brubaker and Kammermeyer (20), and more recently by Tuwiner (159). Except for the influence of the thicker interface, oxygen transfer through a membrane is fundamentally no differ- ent than mass transfer from bubbles and may also be explained by the Whitman two-film mass transfer theory as outlined previously in Section 6.2.4. This theory accounts for the gas film, the liquid film, and the interface (membrane) factors which constitute the principle elements which control oxygen transfer in dialysis aeration. The key equations explaining membrane oxygen transfer as described in several references are reviewed below (3, 159). Dialysis aeration involves the same transfer steps as visualized in Figure 32. The key difference between dialysis and bubble aeration lies in the importance Of the membrane resistance factor ( l_ ) and k3 the gas and liquid dynamics adjacent to the membrane interface. The interface, regarded as a small and immeasurable factor in bubble 147 aeration, is tangible and has a major influence on membrane aeration. In addition, the design of the dialyzer allows a greater degree Of con- trol of the bulk gas and liquid as they flow past the membrane(s). These velocities influence the thickness of the stagnant (laminar) gas and liquid films and accordingly their resistances to transfer .1 and ‘1 . k k 2 4 The influence of fluid velocity and dialyzer Operation on the fluid dynamics adjacent to the membrane were discussed in Section 4.4 of this thesis, were diagrammed in Figure 17, and will influence the dialysis aeration system in a similar manner. As shown previously, the overall oxygen transfer process is repre- sented by the basic mass transfer equation, (6.1), which is the summation of the mass transfer steps 1 through 5 (Figure 32); bulk gas membrane liquid bulk gas film interface film liquid N = KO(AC)A = kbg(6C)A = kg(4C)A = km(AC)A = kL(4C)A = kbL(AC)A. (6.13) Transfer through the bulk gas and bulk liquid is a diffusion process, obeys and may be represented by Fick's Law of Diffusion, and does not represent a significant limitation to oxygen transfer. Thus, these two factors can be assumed small and neglected, as explained previously. This assumption may be made with confidence in the dialysis aeration system because the dynamics within the dialyzer were shown to produce excellent bulk turbu- lence, which promotes adequate mixing and aids bulk diffusion (Section 4 of this thesis). Equation 6.13 may now be reduced to; N = kg(AC)A r km(lC)A - kL(AC)A . (6.14) Tuwiner (159) described membrane transport as a diffusional process. This makes the membrane transfer coefficient in equation 6.13 equivalent . .. .. ,. .. ., . . _. e . . ..s 4 _ e . . . .. , .. i. 5 . _ . .— . . . . I . p .. v . a .. 4 . . . \ .. s . _ .. 4. v . e . . . .. , 4, I .. . 4. . . . . . . e . . . . u _ .. . u . . .. . . . . . .o . . . . . . . . e . 1.1!..6 3|. v. 148 to a diffusion coefficient; N = km (AC) a N N = D/X (AC) A and km ‘3 D/X, (6.15) where D = membrane diffusion coefficient and X = the membrane thickness. The Object of dialysis aeration, as with bubble aeration, is the establishment of an equilibrium state between the gas and liquid phases which results in the saturation Of the liquid with dissolved oxygen. Thus the reasoning used previously to develop the oxygen transfer expression in terms of steady state concentration expressions can be applied here. Thus equation 6.14 becomes; 2 ll kg (P - Pi) A ==km (Pi - Ci) A.=-kL (Ci - CL) A (6.16) and 2 ll * * * g * kg (P - P ) A ==km (P - C ) A kL (C - CL) A (6.17) where the * denotes the equilibrium values. Henry's Law is again used to write the oxygen transfer relationship in terms of liquid oxygen concen- tration; _ + * = * A N-kg kL (C-CL) A k-(C-C)A. (6.18) Under these steady state conditions the membrane has a constant con- centration gradient across its thickness and because of its fixed transfer characteristics the membrane factor in the above expression will be constant 149 permitting equation 6.18 to be written in terms of the liquid film equili- brium concentrations; m ( I) 0 ( 0 ) As shown in equation 6.3 the overall resistance to mass transfer equals the sum of the individual resistances. Although the membrane coefficient and resistance are major elements in dialysis oxygen transfer, in contrast to the significance placed on the interface in bubble transfer, they are fixed characteristics of the material and may be calculated separately from diffusion data, IIS km D/X (6.20) allowing ommission of (km) from the overall oxygen transfer equation. This permits the expression of membrane oxygen transfer to be made in terms of the gas and liquid film coefficients and the liquid film oxygen concentra- tion. The importance Of the liquid film in contrast to the gas film was shown previously. This relationship, also applies to the membrane transfer process permitting the overall ex- pression of dialysis aeration to be made solely in terms of the liquid film; N=KOA(AC)\"N=KLA(C*-CL). (6.21) 150 Although the interfacial area is measurable in dialysis aeration, the use of the volumetric oxygen transfer coefficient which has been the accepted criteria for evaluation of gas transfer for twenty years, is preferred in order to facilitate comparison to previous data. Thus the proportion KL ,3 = KLa applies and equation 6.21 is written in familiar terms; * OTR = KLa (C - C ) . (6.22) L With the use of equation 6.22, dialysis aeration may be evaluated in a manner similar to bubble aeration. OTR and CL values are determined by dissolved oxygen analysis, C* values are characteristic of oxygen satura- tion in the liquid used, and equations 6.21 and 6.22 are used to determine the overall transfer coefficient and the volumetric transfer coefficient, respectively. 6.2.7 Principles of Dialysis Aeration - Silicone Membranes Oxygen is transferred to a liquid in a uniform, continuous, con- trollable, defined, and predictable manner by dialysis aeration. This contrasts with the unpredictable and variable transfer associated with dynamic bubble aeration, as discussed in Section 6.2.5. Unlike sparging, dialysis aeration has a gas-liquid interfacial area which is stable and of known dimensions, a degree of control over the thickness of the 151 gas and liquid films, a stable oxygen partial pressure in the gas phase and therefore a reasonably constant concentration gradient across the interface, an even diffusion of oxygen into the liquid phase which re- duces foaming and sharp oxygen concentration variations within the liquid, and an exposure of the total liquid volume to the oxygen ex- change interface by virtue Of the fluid circulation and the turbulent relatively thin fluid film passing through the dialyzer. Although the aeration of a microbial culture by membrane exchange was demonstrated on a small scale over twenty-two years ago (64), the technique was largely ignored until the oxygenation of blood was suc- cessfully demonstrated by oxygen transfer through membranes (ethyl cellu- lose) contained in an external independent exchanger (29). The ability to handle large volumes and the elimination of toxic, direct, blood-gas contact represent the features which spurred interest in this technique. During the ensuing years considerable effort has been expended in select- ing suitable membrane materials, analyzing fluid dynamics and oxygen trans- fer properties, and designing efficient exchangers. The develOpment of artificial lungs has been thoroughly recorded over the past sixteen years in the Transactions of the American Society for Artificial Internal Organs. These reports of successful applications of membrane aeration using ex- changers similar tO our model dialyzer served as the impetus for applying this large scale technique to microbial aeration. The membrane interface represents the key element in dialysis aera- tion. Early artificial lungs utilized a variety of membrane materials including cellulose. These did not provide sufficient gas transmission rates to oxygenate blood adequately. The subsequent use of Teflon mem- branes and, more recently, silicone rubber membranes (which are 4 to 100 152 times more permeable to oxygen and carbon dioxide) has permitted the maintenance of up to 92% oxygen saturation with a reasonable membrane area (122, 38, 98). The oxygen transfer rates of these membranes, es- pecially the silicone rubber, exceed the rate of transfer through blood. This demonstrated that these membranes do not limit oxygen transfer and confirmed that the fluid film represents the major resistance to oxygen transfer (122). This validates the use of oxygen transfer equations 6.21 and 6.22, which are based on the assumption that the liquid film consti- tutes the major resistance to transfer. Investigations with a variety of membrane oxygenators has shown that membrane gas transfer involves two phenomena, intramembrane transport and phase boundary transport (88). This emphasizes the importance of the membrane and its surface films. The first involves molecular diffusion, which is expressed by Fick's Law of Diffusion J = -D A_C_ X The composition of the membrane material determines oxygen diffusivity, while membrane thickness and area as well as the temperature of operation influence transport rates. Membranes as thin as .0005" have been success- fully employed although the increased strength associated with thicker membranes v".003" is preferred (98). The superior gas permeability of the newer membranes has focused interest on the second phenomena. The liquid film is considered to be the major resistance governing gas trans- fer and reduction of its thickness is important to improving liquid oxy- genation (111, 102). Greater velocity, faster agitation, and counter current gas-liquid flow have been shown to improve oxygen transfer ef- ficiency by as much as 50% (122, 98). These techniques improve oxygen transfer to the bulk liquid by reducing the laminar layer of liquid 153 adjacent to the membrane and rapidly exchanging this film, which quickly becomes saturated with oxygen)with fresh unsaturated fluid. In addition the continuous flow of fresh gas over the Opposite face of the membrane maintains a constant partial pressure of gas-phase oxygen,which aids ex- change efficiency and contrasts with the reduction in gas volume and par- tial pressure when gas bubbles rise and dissolve. Other factors which influence overall gas transport to a lesser extent include gas concentra- tion (air vs. pure oxygen), membrane area, and membrane support and dialy- sis chamber design. These are discussed elsewhere in this thesis (Section 6.4). The key to the efficient membrane oxygenation of blood and to the successful dialysis aeration of a microbial culture has been the avail- ability of Silicone rubber membranes. Its high permeability to oxygen and carbon dioxide and its biological compatibility make it superior to other materials. The permeability prOperties Of several types of commonly used membranesare compared in Table 14. The pheonomenon of gas movement through membranes has been examined and reviewed by Barrer (7), Brubaker and Kammermeyer (20), Tuwiner (159), Rickles (128), and Kaufmann and Leonard (88). Generally, transport through a material is a function of crystalinity, thickness, electrostatic attrac- tion, energy gradients, tortuosity factors, temperature, and concentration gradients and is a process which obeys diffusion equations (Fick's laws). It was recently noted that the nature of the newer membrane polymers and the variety Of interacting factors Operating in membrane permeation make it impossible to write meaningful equations which can predict the process accurately (129). The mechanism for gas transfer through silicone rubber is reported to be analagous to transfer through solid surfaces (Dow data 154 Table 14. Gas permeation coefficients of membrane films *. MEMBRANE TYPE Pm , m , CO2 Natural Rubber and 1 _ Polyvinyl Chloride Polyethylene 2 11 Regenerated Cellulose 3-9 - Ethylcellulose 17 70 Teflon 30 78 Silicone Rubber 1210 6310 * Values were obtained from Gas Transmission Rates 2f Plastic Films, Dow Corning Corporation, Midland, Michigan , and from references ( 29, 122, 131 ). ** cc gas min mZ atm/mil thickness 155 sheet, Gag Tommiesion in Plastic Films, 1959). This permeation has been described as a three-step process as follows: condensation on and solution into the membrane surface, which is a function of the inter- action between the gas molecule and the membrane polymer; molecular dif- fusion through the thickness Of the membrane; and dissolution and evapora- tion of the gas molecule at the Opposite membrane surface (147, 103). Braley (16) has described silicone rubber as a homogenous organo- silicone compound consisting Of a highly viscous non-crystalline polymer with linear sequences of alternating silicone and oxygen atoms and with organic methyl or ethyl molecules attached to each silicone atom: CH3 CH3 -Si - O - Si - 0 CH3 CH3 n The number of these sequences per polymer molecule, n, determines the viscosity of the material and is about 5000 for pliable rubber films. Here these polymers are vulcanized, forming a three dimensional film Of polymer chains lightly cross-linked between the carbon atoms. The vul- canizing agent (dichlorobenzoylperOxide) haloginates the methyl groups permitting cross bonding but does not become a part of the film after vulcanization. This produces a pliable rubber film which consists solely of the cross linked organopolymer and the silica filler ($102 for strength). As a result Of this purity silicone rubber is nontoxic and biologically compatible with blood, tissues, and bacteria. Because silicone rubber films have no pores or open Spaces and show no appreciable bulk flow Of aqueous solvents they are categorized as plas- tic membranes. Solute movement through these films is analogous to the three-step process explained previously, with the exception that the move- ment through the membrane material itself does not follow the classical 156 diffusional transport concept given above. Rapid gas transport in sili- cone rubber is attributed to the high solubility of the gas molecules in the organo-silicone polymer, the high diffusivity characteristics of the polymer, the absence of crystallinity, and the flexibility of the bonds in the silicone-oxygen-silicone chain (16, 131). Because the high perme- ability is related tO the solubility Of gases in silicone rubber, the trans- port process is written in terms of both solubility and diffusion; Pm = S-D (6.23) Thus, the equation for mass transfer across silicone membranes (in equa- tions 6.17 and 6.18 given in Section 6.2.6) is written; N = PmA (AC) (6.24) X where the membrane transfer coefficient, km, and its relationship to diffusion and membrane thickness, k.m g' D/X, is replaced by the perme- ability coefficient, Pm’ and the new relationship which emphasizes solubility (45, 131), NS D;§ (6.25) X The selectivity of gas solubility and the exact process of gas move- ment through the silicone polymer is not known. It is possible that gas molecules squeeze past the pliable polymer bonds or the cross-linkage bonds, that a chemical exchange occurs between gaseous oxygen molecules and the polymer oxygen molecules, or that the methyl groups of the organo- silicone polymer rotate around its long axis opening "void spaces" which enhance the movement of gas molecules. Solubility and solute-membrane 157 interaction play a key role in mass transfer with this material. For instance carbon dioxide permeates the film 10 times faster than much smaller and lighter helium molecules, a condition which is contrary to a strictly diffusional model for tranSport (107, 131). The higher sol- ubility shown by gases with a higher boiling point suggests that molecu- lar energy may also play a role in molecular transmission through the silicone polymer. Despite the inability to clearly identify the mass transfer process, the characteristics of silicone rubber make it an excellent vehicle for the uniform transfer of gas molecules from the gas to the liquid phase without direct gas-liquid contact. 6.2.8 Applications of Dialysis Aeration The develOpment of suitable membrane films has been the major technological advance necessary for serious interest in the dialysis aeration process. Careful material selection, equipment design, and equipment construction have eliminated most of the problems associated with the assembly and operation of membrane aeration equipment. In addition membrane manufacturing has become more sophisticated, per- mitting the production of reliable thin films, dacron mesh-reinforced films, capillary tube membranes, and co-polymer films incorporating molded membrane support elements (16, 131, 47, 121). Recently these developments have primarily involved silicone rubber and are in response to the successful application of this material to the refinement of artificial lung devices. The potential of membrane exchange has also been reported for the following applications: prolonged drug therapy, 158 administration Of general anesthesia, separation, purification, concen- tration, and/or measurement of gas mixtures, oxygen extraction from water, and artificial placenta devices (50, 51, 164, 131, 172). On- going deve10pments concerning the application of silicone materials are periodically reviewed and summarized in the Bulletin of the Dow Corning Center for Aid to Medical Research, Midland, Michigan. The concept of culture aeration by membraneous gas exchange was introduced before the discovery of gas permeable films. Gladstone (64) used a crudely constructed combination of cellOphane sacs contained within a glass nutrient broth chamber. Air passed through a cylindrical inner sac which was surrounded by a more coarse outer sac with the culture being contained in the Space between the two. The Bacillus anthracis culture was aerated by gas diffusion through the membranes. This small scale dialysis-aeration scheme prevented autolysis and the liberation of proteolytic enzymes and delayed Sporulation, resulting in a 25-fold increase in antigen production. The improved yield was attributed to the elimination of the need for toxic antifoam agents, a necessity with bubble aeration, and to the elimination Of the de- naturing effect of direct gas-liquid and gas-cell contact. These exemplify two of the advantages which dialysis aeration can provide for culture prOpagation. Another small scale dialysis-aeration scheme was reported more recently by Brewer (19). This consisted of a small dialysis fermenta- tion system described as a closed outer envelOpe containing the culture and an inner semipermeable envelope which comprised the nutrient chamber. The device was provided with inlet and outlet ports for culture sampling and intermittent or continuous nutrient circulation. When the outer 159 envelope was constructed Of thin polyethylene or polypropylene, oxygen from the atmosphere was transmitted to the culture and permitted aerobic growth which was demonstrated with Corynebacterium diptheriae and Bacillus stearothermophilus. With this scheme the culture received nutrients by dialysis from the inner envelOpe and oxygen by gas per- meation of the outer envelOpe. Both of the above dialysis aeration systems were limited in size, constructed Of fragile materials, and in the second case relied upon passive oxygen transfer. Their oxygen transfer capabilities were limited by the selected membrane materials and available membrane area. In addition no provisions were made for fluid circulation in order to reduce the liquid film resistance to oxygen transfer. Al- though these devices demonstrated the potential value Of dialysis aeration, their practicality for larger scale production of micro- organisms was limited. Furthermore because no data were shown regard- ing the performance parameters Of these aeration processes, no calcu- lations could be made to predict their efficiency or adaptability. Dialysis aeration was demonstrated on a somewhat larger scale when Serratia marcescens cultures were propagated in the bottom of a dialysis shake flask, a culture system specifically designed for biological applications (61). The culture was separated from the atmOSphere by a membrane and a thin water layer. Aeration was pro- vided by oxygen diffusion through the water layer and membrane interface. When a membrane filter was used, viable cell growth was Observed to be markedly higher than the anaerobic control but less dense than growth achieved by conventional culture shaking. This concept was explored further with a 12-liter dialysis fermentation 160 system using, for the first time, an external membrane exchanger (62). In this demonstration the nutrient supply, which was aerated by con- ventional Sparging, was circulated through the dialyzer which contained regenerated cellulose membranes. Oxygen transfer occurred from the oxygenated nutrient fluid across the membrane surface to the culture, which was circulated through the dialyzer from the fermentor. Viable cell densities of 1.4 x 1010 cells/ml were reported after 48 hours growth. Although these results were inferior to the cell densities achieved by conventional fermentor aeration, they did illustrate the principle of microbial aeration by indirect oxygenation. The last two examples represent an advance in equipment design, reliability, and biological compatability. Although the membrane area available for oxygen transfer was greater, the potential Of these systems was again limited by the membrane material used. Regenerated cellulose has a low permeability to oxygen (see Table 14) and membrane filters will not dependably confine bacteria (79). Unfortunately no examination of the oxygen transfer characteristics was reported for either system. In addition,the Operation of both of these schemes was dependent upon the saturation capacity of the liquid, resulting in a much lower oxygen concentration than with air and therefore a lower oxygen concentration gradient across the membrane. The dialysis flask system relied upon passive oxygen availability to the liquid film covering the membrane, had no means of manipulating the gas or liquid phases, and had a limited culture capacity. As a result this experimental system is limited in adaptability to larger scale operations and has its primary value as a test bed for membrane evaluation. The dialyzer-dialysis culture system on the other hand was designed for 161 large scale use, permitted, by virtue Of its independent exchanger, adjustments in liquid flow rates, membrane area, and culture volume, and could be used as a model system for further investigation of the characteristics of dialysis-aeration. Microbial cultures are conventionally aerated by internal sparging. and agitation. External aeration of the fermentation has seldom been attempted, the only notable example being an external aeration system designed by Stich which produced fine bubbles by forcing air through porous materials into the circulating fermentation broth (154). The dialyzer-dialysis culture system designed by Gallup and Gerhardt (57) provided a means of examining the principle of membrane aeration to a full scale microbial fermentation. The availability of highly per- meable, leak-free silicone rubber membranes and the successful applica- tion of these to external membrane aeration Of blood demonstrated the potential of this scheme for fulfilling the demands of a fragile biological system. Theoretically the dialysis-aeration technique has several features (listed in Section 6.1) which could be advantageous tO a microbial fermentation, especially when oxygen sensitive or pathogenic cultures are used. The examination Of these desirable features, the analysis of aeration characteristics, and the examination of the effect of uniform bubble free oxygenation on microbial growth provide reasonable justification for investigation of dialysis-aeration. The serious consideration of dialysis-aeration as a substitution for conventional aeration requires that two principle questions be answered: 1) will the experimental system perform reliably under the rigors of fermentation conditions and provide sufficient oxygen transfer rates; 162 and 2) will this system which has a gas-liquid interfacial area which is relatively small support cell growth approaching or exceeding culture densities Obtained by conventional aeration? Preliminary investigations utilizing a prototype of the dialyzer described in Section 3 showed that, when equipped with silicone rubber membranes and using pure oxygen, the dialysis-aeration system provided sufficient oxygen transfer to theoretically support a 1.5 liter culture of Saccharomyces cervisiae (G. F. Bennett, personal communication). The preliminary analysis of the oxygen transfer characteristics for the system (presented at the 152nd American Chemical Society meeting, 1966) have been extended and an analysis of both the physical and biological characteristics of dialysis-aeration have been completed and are presented in the following parts of this section. 6.3 Materials and Methods Dialysis aeration experiments were conducted with the same fermen- tation equipment described for dialysis growth trials. Adaptation for aeration required only minor Operational changes and inclusion of gas FF permeable membranes in the dialyzer. The Operational configurations used were: non-dialysis control, with conventional air sparging and agitation (Figure 18); dialysis aeration plus nondialysis nutrients (Figure 33); and dialysis aeration-dialysis nutrients (total dialysis) with one dialyzer for gas exchange and another for nutrient exchange (Figure 34). Dialysis aeration also required pumps, conduits, a dialyzer, air flow meters (18 M), air pressure gauges (19 M), a dissolved oxygen probe and analyzer (8 M), a fabricated nylon "T", threaded to hold the dissolved oxygen probe in the fermentor liquid conduit line, a chart recorder (26 M), and most importantly, silicone rubber membranes (20-M) cut and fitted to the dialyzer. In the case where both aeration and nutrition were accomplished by dialysis, two dialyzers were assembled. One contained silicone rubber membranes separating air flow and fer- mentor liquid flow, and the other contained Visking regenerated cellulose membranes separating nutrient reservoir liquid flow and fermentor liquid flow. The silicone rubber membranes chosen for dialysis aeration were .007 inches thick, dacron-weave reinforced, "Silastic" B-2000 type (produced by the Dow Corning Corporation). Assembled in the dialyzer, 163 164 .EOumzm ououfiso sowumuomumflmmamwo on» mo Emnmmwn $2 mwN>4<5 .; m SHEILC '1 mm ...-"m mOszémmu I—-— l.- --.vu 1mm! Emwmxrm mmakmao zofidmmaxlmaejflo .mm Ouswwm 165 .aouwhm ombufiao uaoauubaamamhamaw In aowumuomumfimmflmwo man no Seams“: .dm ouswwm m_o>mmmmm «me/.35 mmNZSQ moezmsmm... AszSz E4 0000000.“. Emem>m_wmz.fiao m_m>I_<_o IZO_._. factors which represented the approach to liquid - gas equilibrium C* , which is the oxygen 169 T7 I I I I F I00— I - '— 30.. ._ 60— ...I 23 Q t.— <1 a u- E 4;;)-- ~— <1 (.0 0° ZO~ - C 1 I. I L ' I - I J. U ti (I ii) 5 I3 F’Ifl Figure 35. Relationship between percent saturation and PPM concentration Of dissolved oxygen at 30°C. ._i;:::rr 170 SATURATKNI EN CMYG ‘7. j.- p— p— I‘\) \D N 1‘)! '— OJ “x O 5 IO I5 NHNUTES Figure 36. Oxygen uptake by 3 liters of medium in a 5 liter fermentor aerated by dialysis. ' 171 Table 15. An example Of the computations necessary for the graphical a value, as illustrated in determination of the overall K Figure 37 *. L TIME DISSOLVED OXYGEN COMPUTATIONS * (Min) ( % Saturation) CL (av) I Ce - CL(av) fl. 0 16 I 33 67 Ifi l 50 2 56 61 39 4 66 6 75 78 22 8 81 10 85 86 14 12 88 14 91 92 8 16 94 18 95 95.5 4.5 20 96 22 97 97.5 2.5 24 98 26 99 99.5 .5 28 100 * Dissolved oxygen values are from the dialysis aeration trial shown in Figure 36. C* represents the equilibrium concentration end point for this gas- liquid system (100% saturation or 7. 3 PPM) and C (aV) represents the average dissolved oxygen concentration over thé desig- 1. nated time intervals Of the aeration tria 172 g '3 L) . L L) S? (9 -(D .J O -- .. ‘ I I O ' IO 20 30 MINUTES Figure 37. Graphical determination of the overall volumetric oxygen transfer coefficient (KLa) for the dialysis aeration trial shown in Figure 36. 173 saturation end point Of the liquid aeration Operation (C* = C av = L saturation = 7.6 ppm»). These factors C* - CL av were the integrated values (1n (0* - CL av)= KLa t + constant) of the oxygen transfer equation Ell-E- = KLa (C* - CLav)) over the aeration period. They provided one of the coordinates 1n (0* - C av), t being the other, for the L graphical solution of the overall KLa value for the experiment (Figure 37). The KLa value was obtained from the slope of the plot as shown or could be determined by solving the oxygen transfer equation for each time interval Of the trial and averaging the values. Demonstration of cell synthesis and viability in dialysis aerated .cultures of Serratia marcescens represented the final and most important evaluation of this aeration technique. These growth trials were con- ducted with the same equipment and in the same manner as described above for oxygen transfer analysis and previously for growth analysis (preceding section Of this thesis) with the exception that the sole source Of oxygen was by dialysis aeration (conventional air sparging in comparative controls) and the fermentor was inoculated with viable cells. As described previously, the culture was sampled via syringe and sampling ports and was analyzed for total and viable cell concentrations and dry 'weight. In addition, oxygen uptake as well as culture oxygen demand were measured during growth by either the sampling method or the dynamic degassing-gassing method (156). The first method required removal of a culture sample which was placed in a stirred, oxygen-saturated beaker of broth. The utilization of dissolved oxygen by the reSpiring cells was detected by the dissolved oxygen electrode. The second method was more convenient and eliminated sampling errors and atmospheric oxygen con- tamination because it allowed oxygen uptake analysis at any time without 174 sampling. Instead, the culture aeration was stOpped and the dissolved oxygen utilization of the culture was detected as it circulated past the electrode in the external conduit. Deoxygenation of the culture fluid was indicative Of uptake and demand, and oxygenation of the fluid upon resumption of aeration was indicative of the oxygen transfer capacity of the aeration system under actual culture conditions. This technique also allowed determination of KLa values under these con- ditions. However one problem was observed which reduced the useful- ness of this method. The electrode frequently lost accuracy toward the end of long growth periods. Presumably this was due to poisoning of the electrode by the salts in the growth medium. 6.4 Results 6.4.1 Physical Aeration Characteristics The dialysis-aeration technique produced no observable positive gas pressure within the fermentor vessel, a characteristic which significantly reduced the possibility of air leaks at the fermentor gaskets and agitator shaft bearings. In contrast, a conventionally sparged fermentor develOped an internal gas pressure of 2 lb/sq. inch to 5 lb/sq. inch depending on the gas velocity used. Thus dialysis- aeration lowered the risk of aerosol contamination, which represents an important consideration especially for the propagation Of pathogenic microorganisms. A second attribute of dialysis aeration was the elimination of the need for gas sterilization. The silicone rubber membranes com- pletely separated the culture from direct contact with the gas phase and were impervious to bacteria. Therefore these membranes when prOperly sealed in the dialyzer, protected the culture from contamina- tion by the incoming air and protected the outgoing air from contamina- tion by the culture. This eliminated the necessity for gas sterilization at either site. In addition, dialysis aeration did not, as shown above, produce a positive pressure or gas flow within the fermentor, permitting the use of a simple and inexpensive vent on the culture vessel. Diffi- culty in Obtaining a perfect membrane seal in the dialyzer was occasion- ally experienced during the investigations. This should not detract from the potential economy which dialysis-aeration affords because it represented a technological problem which has been reported for 175 176 artificial lung devices and reliably solved by the use of silicone gasket cements. Another attribute of dialysis-aeration was the reduction in liquid loss due to evaporation and foam formation, which are associated with high volume sparging. Analysis of the liquid lost during conven- tional and dialysis-aeration did not however show as large a reduction in liquid loss as eXpected (Table 16). Table 16 Influence of the aeration method on culture liquid loss. CULTURE METHOD VOLUME LOSS (ML) Non-aerated batch control* 175 Sparged air -- Nondialysis nutrients 460 Sparged air -- Dialysis nutrients l membrane 750 Sparged air -- Dialysis nutrients 6 membranes 1300 Sparged air -- Dialysis nutrients 10 membranes 2300 Dialysis air -- Nondialysis nutrient+ 350 * Culture vessel was not aerated or agitated +Average of 5 trials The large losses associated with the Sparged dialysis-nutrient system were attributed to the osmotic transfer of water through the dialysis membranes to the nutrient solution. Because membrane oxygenation does not produce a high throughput of gas bubbles within the fermentor, it was believed that evaporation would be nearly eliminated. The data 177 suggested that a 350 ml volume loss represents a base line liquid loss over the course Of the fermentation and that Sparger aeration at the rate used, 9 L/min air velocity, did not cause as great a liquid loss as anticipated. Higher air velocities would be expected to produce greater evaporation in conventional aeration than in dialysis aeration. The elimination of gas bubbles should reduce foaming and permit the quantity of necessary antifoam to be reduced or eliminated altogether. This was Observed to be the case when a culture was dialysis-aerated and not agitated. In contrast, sparging under the same conditions produced unmanageable foam within the fermentor. However as will be shown below in Figure 45 agitation was necessary in order to achieve the greatest cell density in a dialysis-aerated culture. Although the quantity required was comparatively lower, the initial addition Of antifoam was required under such culture conditions. The rate of oxygen transfer is directly proportional to the magni- tude of the oxygen concentration gradient from the gas phase to the liquid phase. An increase in the partial pressure of oxygen in the gas phase will enlarge this gradient and therefore the driving force for mass transfer. The selection of gas used for dialysis-aeration was found to influence the oxygen transfer coefficients for the aera- tion system (Table 17). As shown, the gas with the greatest partial pressure, pure oxygen, produced the highest volumetric oxygen transfer coefficient. However its use was regarded as impractical for long term or large scale fermentations because of the expense involved. Humidified air showed a slightly lower oxygen transfer coefficient than dry air. This was attributed to the additional transfer resistance produced by the liquid film which formed on the gas side of the membrane by the 178 moist air. Humid air represented the best gas selection for the dialysis aeration method because it eliminated liquid evaporation via the mem- brane interface. Reduction Of the gas-liquid vapor gradient, by use of moist air, has been shown to prevent the transfer of water vapor across silicone membranes (131). The partial pressure of oxygen can also be increased by an increase in overall air pressure. Under the selected standard mode Of Operation, (30°C, 2 L/min liquid velocity, and 25 L/min gas velocity) no pressure gradient existed across the dialyzer membrane (e.g. the difference in pressure between the liquid and the gas chambers of the dialyzer was zero). Insertion of an air filter in the outgoing gas conduit restricted gas flow and raised the internal pressure to .13 atm. The addition of membranes did not affect this value. With the filter attached the in- ternal gas pressure increased when the gas velocity passing through the dialyzer was raised (Table 18). These increases in the pressure gradient were relatively small, would produce only small increments in the oxygen partial pressure, and would not appreciably improve the oxygen transfer driving force. Thus the manipulation Of gas flow as a means of increas- ing the partial pressure of oxygen did not appear to be justified. The pressure drOp across the length of a dialyzer air chamber was very small (.021 atmI§ the above standard Operational conditions). Slight changes, corresponding in nature to those Observed for total pressure in Table 18, were Observed when the gas or liquid velocities were increased. These data demonstrated that the dialyzer as designed presented little resiStance to air passage. The addition of membranes produced a small reduction in pressure drOp across a given chamber which indicated that the increase in internal volume associated with additional 179 Table 17. Influence of the gas selected on the oxygen transfer coefficients for the dialysis aeration system. GAS USED KLa (Min-1) * -2 Oxygen 4.2 x 10 Air 3.4 x 10'2 . . -2 Hum1d Air 3.1 x 10 Nitrogen 0 * With one membrane (.0288 m 2) in the dialyzer. Table 18. Influence of the gas velocity through the dialyzer on the internal gas pressure. GAS VELOCITY GAS PRESSURE ( L/Min ) ( Atm ) * 7 o 15 .07 21 .10 25 .13 31 .17 * Positive pressure above the ambient atmospheric pressure. 180 chambers further reduced the resistance to gas flow. Variations in pres- sure drop were noted for different chambers within the dialyzer suggest- ing that the membranes and/or membrane separators flexed or bulged when gas and/or liquid velocities were changed. This was further sub- stantiated by the observation that these variations in pressure drOp were more pronounced with six membranes than with one. It is possible that the distortion of the internal dialyzer components may limit the membrane area and gas velocity which are Operationally practical. The standard Operational conditions adapted for these and subsequent dialysis aeration investigations did not appear to be excessive. Al- though no membrane failure occurred at the maximums of Operation the risk of membrane rupture and gasket failure (gas leakage was occasion- ally Observed at 31 L/min) should be considered carefully, especially if the system was to be Operated under conditions approaching the maximal limits. The separate and independently controlled membrane exchanger repre- sented the most important characteristic Of the dialysis-aeration system. This scheme permitted a combination of independent adjustments to be made on the operation of the gas and liquid phases. In addition the gas-liquid interfacial area was a measurable entity and could be changed by adding or removing membranes from the dialyzer. The influence Of these operational adjustments on the oxygen transfer capability of the dialysis-aeration system were evaluated in terms of oxygen transfer rates (OTR, in millimoles Of oxygen/liter/min) and/or volumetric oxygen trans- fer coefficients (KLa, in min-1) in the following experiments. The influence Of air velocity on oxygen transfer was analyzed while the membrane area and liquid velocity were held constant. An increase in air velocity from 9 L/min to 25 L/min or 31 L/min produced only a 181 small increment in OTR values unless a large membrane area was used (Figure 38 and Table 19). As shown, an increase in membrane area improved the OTR at all air velocities with the greatest improvement occurring with six membranes and 31 L/min. The data values given in the table Show that an increase in air velocity from 9 L/min to 25 L/min produced a 50% increase in OTR.when 1, 2, 3, or 4 membranes were used, a 360% increase in OTR.when 6 membranes were used, and a 5% increase in OTR for the sparger aerated control. However the OTR's for Sparger aeration were over twenty times greater than that for dialysis-aeration Operated with 6 membranes and at the maximum gas velocity. CorreSpond- ing improvement in dialysis-aeration KLa values have been Observed (Bennett, personal communication). The improved oxygen transfer was primarily attributable to a reduction in gas-film resistance at the membrane surface and to the increase in gas pressure and correSponding oxygen partial pressure associated with greater air velocity. The more pronounced improvement in OTR when six membranes were used re- flected the value of additional interfacial transfer area. The effectiveness of this area was enhanced by greater air velocity which increased dialyzer ventilation and reversed the reduction of internal pressure drop associated with added dialyzer chambers. It appeared that dialysis-aeration transferred oxygen more efficiently when Operated at highest practical air velocity and the maximum possible membrane area. The influence of liquid velocity on dialysis-aeration oxygen transfer rates was analyzed in a manner corresponding to that given above. In this case, gas velocity was held constant while the liquid velocity was varied. An increase in liquid velocity from 1 L/min to 182 I I I O - ’ SPARGER"CONTROL" .... 0mmmmomw meM'W'mM '3‘ E \l \ -| - .. N O MEMfiBéIANE (I) . [LI 6’ \l C) d”””,/”13 3 O E o E -2 ~— ._ 2?. 2...... g ’0 Jaw—— \I -07 0.... ..3 .... ...... l I- I I 0 I0 20 30 AIR VELOCITY (L/MIN) Figure 38. Influence of air velocity and membrane area on dialysis- .aeration oxygen transfer rates. 183 .uouhfimeo ecu swnousu :H2\A N no huHOOHO> weavea ucmumooo m um ooumnoeo e ommo. ooao. so no. moNo. oooo. ono. mmoo. oHoo. mm me. ovo. oooo. omoo. omoo. osoo. AH oo. smoo. «moo. mmoo. eeoo. NHoo. o aomezoo oNAA. Aoo ones. Aeo eooo..Amo homo. ANV A oomo. AHo Aeazxao MH< Qmwfimm A NE v mmu< oawufifimz flaw mmcwunamz MO Hun—:52 EHUQHM> mH< A eAz\a\eerz.s o mac e .uoumuom memefimwo ecu you moumu nommcwsu cowkxo so mono ocmuofioe one hquOHo> new mo mucosflmcH .mH OHomH 184 3 L/min produced similar 70% to 80% increases in OTR values for all the membrane areas used (Figure 39 and Table 20). An increase in membrane area resulted in a larger OTR value at each liquid velocity. Again, the OTR value for the Sparged control was far greater than that recorded for dialysis-aeration at the maximal liquid flow rate. Liquid velocity increments have been shown to produce corresponding increases in dialysis- aeration KLa values (Bennett, personal communication). The improvements in oxygen transfer rates were attributed to a reduction in the liquid film thickness and resistance at the membrane surface and an improve- ment in bulk liquid mixing both of which are essential to efficient membrane transfer and occur with increased flow rates. These results correspond to those Observed earlier for solute dialysis (Figure 9 of Section 4 of this thesis) and demonstrate that dialysis transfer Of oxygen was maximized when the system was Operated at the greatest practical liquid velocity and maximum possible membrane area. The preceding results revealed that increases in gas and liquid velocities raised OTR values and that, generally, liquid velocity increments produced the largest improvements in oxygen transfer rates. This was further illustrated by increases from intermediate to maximal or near maximal Operational flow rates (Table 21). Upward adjustments 111 liquid velocity consistently produced the greatest percent change 111 OTR values regardless of the membrane area selected. Liquid velOOity Iwas again confirmed as the most important factor when the transfer vvithin a single dialysis unit was analyzed. Increments in gas and lirpiid velocities from the minimum to the maximum Operational values, 9 L/min to 31 L/min and 1 L/min to 3 L/min respectively, produced respective 50% and 78% increases in OTR values. In addition it was '185 I I I 0__ SPARGER"CONTROL" _ ...]- .... MEMBRANE NO. -6‘ /(3" LOG OTR (mMOLES 02/1. /MI/v) I 2 3 LIQUID VELOCITY (L/MINI Figure 39.' Influence of liquid velocity and membrane area on dialysis- aeration oxygen transfer rates. 186 mo mufioofio> new ucmumcoo m an woumumoo x ooNo. ovo. emoo. maoo. m oooo. omoo. omoo. oaoo. N moso. mmoo. oNoo. eaoo. A oNAH. Aoo sooo. Amo somo. ANV ooNo. AHV Aeaz\ao A a v mon<.ocmunEoz pom N mocwuoEoz mo Hooabz A eaz\a\eeaoz_s V moo NHHUQHM> QHDUHA % .uoumuom mwmkfimwo osu pom moumu nommcmuu cowhxo so mono ocmuoaoe new AOHOOHO> oHOUwH mo moconamaH .om oHomH 187 . oazhafloz s E :98 ... on oeNo. ooHo. om moNo. ooao. o one ovo. omoo. oN msoo. omoo. m eon mNoo. oHoo. ANA voo. voo. A $55 Aces 8 Aces No oozéo A323 moo SEE E mmzémzee ezmommm moo zo mozooaozH zone oHooHa ezmommm meo zo oozmoaoze zoom mH< mo ammzpz e .moumu nommcmuuucowhxo cowumummnmwthwwo so many 36am new new .ouou 30am oHOUHH .moum moonoaoa mo mononamaH .HN OHAMH 188 found that the OTR for the highest air velocity (.0350 m moles OZ/L/min) could be further improved (by 30%) to .0410 m moles OZ/L/min (the high- est recorded value for the dialysis aeration system, using six silicone rubber membranes) by an increase in liquid velocity to 3 L/min. These results show that the dialysis-aeration system delivered the greatest oxygen transfer at the maximum Operational conditions and that liquid velocity through the dialyzer was more important for improving oxygen transfer rates than air velocity. The comparatively greater gain in oxygen transfer Obtained by raising the liquid velocity, which accordingly reduced the liquid film thickness at the membrane surface, prompted an attempt to calcu- late the liquid film resistance and its response to fluid velocities. This was possible because, for the first time in a fermentor aeration system, the exact area and oxygen transfer properties Of the gas-liquid interface were known and constant during operation. For this analysis a single silicone rubber membrane (.0288 1112 area), a 3 liter liquid volume, a constant 17 L/min air velocity, and varying liquid velocities were used. The oxygen transfer rate (OTR), volumetric oxygen transfer coefficient (KLa), overall mass transfer coefficient (K0), and overall, membrane, and combination gas and liquid film resistance __ , 1 K art-- .1 R O m values (Table 22) were determined by the previously described methods, the equations and relationships described in Section 6.2.6, and the + + following relationships: KLa = K , ‘1 .1 k k .1. 0 km L g as .1. V K at and 1 ; where _]_. _ + _l_ k k k l 1 KO 111 kL/ L L g * (See footnote on page 190.) 189 .momoo Ham a“ GHB\A NH on cam; Ones moumu sofiw new ooawm uaoumcoo one HHmSm moanmmm mg oocmuwwmou SHAH new .oafio> mocmuwwmou EHHM mam can ownvwfi vocfioeoo m muaomoumom as .oom: mos :HE\EO mmoo. mo Aaxv ucowowmmooo Hommsmnu moms ooumHoOHmO w sues ocmuone OHumeHm xOficu HHS m one x _.I u o mom sow 2:. SS. 1 38. m of. No? . 3- . I 2. now oom BB. oooo. 2+ I 38. N 2. . _.I me? I I 2 now won ioo. omoo. I Soo. H O O Aeweeeo .e the: as: e: A3558 e ATeee was Aeweeeo .e Aestflmo .262 so A355 oozfimeomm mmomzéo mommzfie 8% 34m manage zmoer use 38o * .uOumuomunoumawHo onu nmnounu mouse BOHm ofisvwa mounu um powwowuu cowhxo Ou mocmumwmou Eafim ofinvfia woumfinofimo .NN oHomH 190 As shown an increase in liquid velocity through the dialyzer from 1 L/min to 2 L/min produced a 14% increase in OTR and a corresponding 12% decrease in the calculated liquid film resistance to oxygen transfer. A further velocity increase from 2 L/min to 3 L/min resulted in an additional 56% increase and 40% decrease, respectively. The correlation between OTR improvement and liquid film resistance reduction supported the imporv tance placed on liquid velocity in dialysis-aeration and substantiated the validity of the assumption made in equation EBand reported by others (48). The liquid film clearly represents the largest and therefore most important factor in gas-liquid oxygen transfer. Enlargement of the dialyzer membrane area represented a direct and measurable change in the gas-liquid interfacial area which is essential to oxygen transfer. When the dialysis aeration system was Operated at an air velocity of 25 L/min and a liquid velocity of 2 L/min a single silicone rubber membrane (.0288 m2 area) produced an OTR of .0018 m moles -2 - OZ/L/min and a K a of 6.2 x 10 min 1 (Figure 40 and Table 23). An L increase in area to six membranes (.1728 m2) produced a 12 fold increase in OTR (to .0205 m moles Oz/L/min) and a 4 fold increase in KLa (to 25.6 x 10.2 min-1). Previous experiments had shown similar increases in OTR * (From preceding page) The [H term represented a combination gas and I L liquid film resistance value. The available instrumentation permitted precise dissolved oxygen analysis. However gas analysis equipment was not available preventing precise evaluation of the gas film transfer co- efficient and resistance values. Because the liquid film mass transfer resistance- has been reported to be far greater than the gas fihn re- sistance W 7) and because the gas velocity was great enough to mini- mize gas £11h resgstance and was held constant it was felt that the data from this eXperiment would be primarily indicative of changes in the liquid film and its resistance to oxygen transfer. 71 191 O -\ 3 3 \ \l -|-- \ 6" (0 Lu \l 9 § E \. E O Q) '2'“ O \l -3 0 Figure 40. oxygen transfer aeration system. O ___l I l 2 :3 4 5 ' NUMBER OF MEMBRANES Influence of membrane area on experimental and calculated rates-and oxygen transfer coefficients for the dialysis- 192 GHE\A mm mo mowufiooHo> vwsvwa pom “Hm um .hHm>Huoonwou :HE\A N can pooHEHouop ouoB moonp Housmaflummxm x mowo. mnmo. N-oH x mmoo. ammo. ~-oH x mace. quo. N-oH x mmoo. mmwo. N-oH x mHoo. mmfio. N-oH x m.mm N- o.wH N- m.mH N- q.HH N- N.@ N- x om.nm x do.m~ x ww.wH x mm.NH x mN.o # Afimucmawuoaxmv _ ApmumHDonov * afimucoEwuomme _ Awmumfiaofimov AEEENO 3?: 5 ES Afincwzv mam Amcmunamfimz wwmoé mmdemzmz mo ammZDz .aowumumm mammamfip now mmaam> ucmwowmmmoo paw mum“ mommamuu cowhxo HmucmEHquXo pom poumfisofimo ecu no mono mamunsoa mo ooconfimaH .MN mHan 193 when membrane area was increased (Figures 38 and 39 and Tables 19 and 20). These results clearly demonstrated the increase in oxygen transfer gained by increasing the interfacial transfer area as predicted by the oxygen transfer equations. The dialyzer has been successfully Operated with up to 10 membranes installed and this system could incorporate two or more dialyzers in order to increase the total available membrane area. Unfor- tunately additional .007 inch thick reinforced silicone rubber membranes were unavailable from the manufacturer precluding investigations with greater than 6 membranes. The defined area and known permeability of the aeration membranes permitted the analysis of dialysis-aeration gas transfer efficiency and the prediction of the membrane area required for microbial aeration (Figure 40 and Tables 23 and 24). A comparison of the calculated and experimental KLa and OTR values for the dialysis-aeration system Operated at 25 L/min air velocity, 2 L/min liquid velocity, 3 liter liquid volume, and varying membrane areas (.0288 mz/membrane) showed a close agreement between the K a values when one or two membranes were used but an increasingly wider L disagreement in K a values as additional membrane area was added (Table L 23 and Figure 40). In addition, the data showed a large difference between calculated and experimental OTR values, even.with a single mem: brane. The calculated values were determined from published permeability Specifications for the .007 inch thick silicone rubber membranes. The calculated KLa and OTR values of 6.25 x 10-2 L/min respectively for one membrane were obtained after considerations min.1 and .0153 m moles 02/ had been made for liquid volume, area, prOportion Of oxygen in air, and the assumption of a maximal concentration gradient driving force. Because the calculated and experimental K a values were approximately the same, L 194 it was assumed that the membrane Specifications and the calculations were reasonably accurate. The large discrepancies between calculated and ex- perimental OTR values were therefore attributed to errors in dissolved oxygen detection, timing, and data evaluation. As shown in the materials and methods section, values selected to represent the OTR for a given ex- periment were those which occurred during the most rapid increase in liquid oxygenation. Such a determination required interpretation of the portion Of the dissolved oxygen trace (represented by Figure 36) with the fewest data points and was therefore subject to the greatest amount of error. In contrast the graphical determination of a KLa value utilized numerous data points (as shown in Table 15) accordingly improving accuracy, as confirmed in Table 23 for a single membrane. It may be concluded that the reported OTR values for all eXperiments were subject to this error and tend to be lower than the actual values. However the consistent interpre- tation of dissolved oxygen traces and the demonstrated duplication of resultant values offset the detracting effect of this error on the ex- perimental trends Observed or the conclusions drawn. The greater accuracy associated with KLa values enhanced their use- fulness as criteria for dialysis-aeration efficiency determinations and membrane area predictions. The addition of six membranes (.1728 m2 area) was expected, assuming 100% efficiency, to result in a six fold increase in the volumetric oxygen transfer coefficient. The experimental data however revealed an increase Of only slightly greater than 4 fold (Table 23) resulting in a lower efficiency, 75% (Table 24). Repeated eXperiments with La values which could be eXpected between duplicated trials, :12 x 10.2 min-1, (Figure 40). The best KLa six membranes established the range in K Value recorded for six membranes under the Operational conditions was 195 . Eum\ B\cHE\~o we mmm mo huHHHnmoauom wmwmwooam m sufia mocmunEOE OHummHHm HHE w cwaounu Emouum ume cm Bonn cowhxo mo uwmmcmuu any so pummm * N-oH x cm emu N-oH x oH.m~ N-OH x om.nm 0 ON - ma ma 09 N-oH x OH ems ~-oH x o~.o ~-oH x m~.o H Asmav aucmauammm AxooHv A assuage manaa> v A Hangmaaumaxm V A emumfiaofimu V mum: nmnmmz a wozmHOHmmm a mmzHuom uooa Ou hummmmooc mono mcmunaoa ocu mo cowumswumo pom .coHuwuom m«m%awwp now mucmfiowmmmoo nomwcmuu cmmhxo Hmucoawuomxw pom HmOHumnoonH .um mHan 196 28.1 x 10.2 min-1, a dialysis-aeration efficiency of 75%. The reduction in efficiency associated with larger membrane areas was attributed to the previously reported changes in internal dialyzer pressures, variations in fluid and gas dynamics, reduction in bulk flow and turbulence, uneven distribution of gas and fluid through the dialysis chambers, and resultant increases in film resistances all of WhiCh contribute to the reduction in efficient mass transfer through the membranes. The results show that greater oxygen transfer occurs when larger membrane areas are used and that Operational adjustments in liquid and/or air velocities would permit more effective utilization of this area and probably increase aeration efficiency. The volumetric oxygen transfer coefficient for an active Serratia -2 - marcescens culture was determined to vary between 10 x 10 min 1 and 80 x 10.2 min.1 (Table 26). The membrane area necessary to satisfy the maximum oxygen transfer requirements of this organism by dialysis-aeration was predicted from the experimental KLa values and eXpected efficiency for this system (Table 24). As shown, highly (100%) efficient dialysis aeration would require approximately 13 membranes (.3744 m2 area). On the basis of the lower (75%) efficiency Observed for the system a larger ‘membrane area would be needed. These membrane requirements would vary with the oxygen demand of the organism selected. 6.4.2 Culture Demand Evaluation Of the biological aSpects of dialysis-aeration required the use of an aerobic test organism and knowledge of its oxygen demand. Serratia marcescens 8-UK.was used previously in the dialysis culture 197 system, met this specification, was found to have a maximum oxygen demand of .66 m moles Oil/min (Table 25), and was reported to reSpond to in- creased aeration efficiency by increased cell synthesis (145). During the course Of a batch culture the oxygen demand of this organism rose from a low initial value (.11 m moles Oz/L/min) to the above peak value, which occurred at the end Of exponential growth and returned tO a low value (.13 - .14 m.moles Oz/L/min) during stationary growth (Figure 41 and Table 25). These oxygen demands were slightly greater than those previously reported for the same strain using a similar analytical technique (.075 to .377 m moles 02/L/min) (100) and were in general agreement with the values reported for several other organism (48). The volumetric oxygen transfer coefficients for the growing Serratia marcescens culture were also determined. The dynamic determination method (156) utilizing a dissolved oxygen probe permitted an estimation of the La values which prevailed during the propagation of a conven- tional control culture (Table 26). The range of these experimental range of K values, from a high of 80.4 x 10"2 min.1 during exponential growth to a low of 10 x 10-2 min"1 during stationary growth, corresponded to those lreported for other cultures (48), 156). As noted previously, the maximal ItLa Obtained with the six membrane dialysis-aeration system was 28.1 x 10.2 min.1 which is less than the peak value for this viable culture. The initial conclusion from these results suggest that the cell density and degen demand achieved with conventional aeration could not be equalled ‘With dialysis-aeration unless a greater membrane area were used. 198 Table 25. Oxygen demand of a Serratia marcescens culture. HOUR DRY WEIGHT OXYGEN DEMAND (mg/ml) ( m Molele/Min ) ( m.MOles/mg wt/Min ) 2 .12 .11 .00091 6 2.17 .66 .00030 12 5.25 .33 .00006 24 7.08 .14 .00002 48 14.10 .13 .00001 Table 26. Volumetric oxygen transfer coefficients during propagation of Serratia marcescens in a conventional fermentor. GROWTH PERIOD VIABLE CELLS KLa (Hours) (Billions/Ml) (Min-1) -2 2 .56 80.4 x 10 5 2.50 36.0 x 10'2 8 12.00 10.0 x 10'2 199 - Ntwm/‘b 831% w ONVWBG NEOAXO % Q “I T I 480 I I I I I I I I I II- I Q 0) a) C .2 ‘IW 53d T133- 318VIA 901 _‘ Figure 41. Correlation of the oxygen demand and growth in an air- sPar'ged, nutrient-dialysis culture of Serratia marcescens . HOURS 200 6.4.3 Biological Aeration Characteristics The previous data (Figure 41) showed that the maximal oxygen demand of a Serratia marcescens culture occurred during and toward the end Of the exponential growth phase, i.e. the 6th to the 12th hour. The oxygen demand associated with concentrated cell growth had a diminishing effect on the dissolved oxygen concentration of the fermentation broth (Figure 42). This was reduced from approximately 100% saturation at the initia- tion Of growth to a minimum of 40% saturation for Sparger aeration and 10% saturation for dialysis aeration between the 6th and 12th hours of growth, the period correSponding to that for the maximal demand. As illustrated, the dissolved oxygen concentration rose after this period and remained relatively constant at 80% saturation and 12%-l4% saturation respectively for the duration of the 48 hour growth trial. Dialysis- aeration resulted in a greater reduction in and poorer recovery of culture Oxygen saturation levels during the course Of the fermentation. This corresponded with the smaller oxygen transfer rates previously observed for this aeration process. However, dialysis-aeration provided culture Oxygen levels superior to anaerobic conditions (no active aeration) and did maintain, at all times, a dissolved oxygen concentration significantly greater than the reported range of concentrations ( .003 m moles OZ/L (-7% saturation) to .050 m moles 02/L (5% saturation)) critical for microbial growth (48, 100) (Figure 42). Dialysis-aeration (6 membranes,.l725 m2 area) did not limit the BrOWth or concentration of Serratia marcescens batch cultures (Figure 43). Duplicated growth trials showed that dialysis-aerated cultures attained Si‘Inilar but longer exponential growth phases and slightly greater mean 201 [X3 1‘ I I SPARGED AIR 80 . f __ GI s 60- - 2 (I B ‘1 (D 40» .. é )— X C) a5 20"- —1 DIALYSIS AIR ‘1 —L _-_- -___ 03mm. CCNQENTRATION C) l j J - O :2 24 36 43 HOURS Figure 42. Dissolved oxygen concentrations in the culture medium during batch growth of Serratia marcescens under sparger and dialysis- -----'----1r- aeration with a membrane area of .1728 m (6 membranes). 202 II ... N- T DIALYSIS AIR 13.-...‘T ‘34 ”EPARGER AIR 1I \ (f) \I q LL] 9 Lu \I m S x Q) Q q 9 -- ._ II 8/ J I I 1 O 12 24 36 48 HOURS Figure 43. Growth of Serratia marcescens on Trypticasesoy broth under 'dialysis-aeration (6 silicone rubber membranes, .1728 m2 area) and Sparger aeration. ‘ 203 concentrations Of viable cells than comparable conventionally aerated cultures. As illustrated, dialysis-aeration permitted a mean peak cell concentration Of 107 billion viable cells/m1 within a range Of 86 to 120 billion cells/ml to be achieved at the 12th hour Of repeated 48 hour growth trials in Trypticase soy broth. Sparger aerated cultures by com- parison achieved a peak concentration of only 55 billion cells/ml within a range Of 43 to 88 billion cells/m1. In both cases the viable cell concentrations declined to 86 and 46 billion cells/ml reSpectively during the stationary phase of growth. The results were repeatable and varia- tions in growth data throughout the trials, the standard deviation was i_20 - 24 billion cells/ml, were similar for both aeration methods. The unexpected ability of dialysis-aeration, even with a suboptimal membrane area, to support high densities of Serratia marcescens cultures was further demonstrated in growth trials utilizing: 1) dialysis-aeration and dialysis-nutrients, 2) sparger-aeration and dialysis-nutrients, and 3) Sparger-aeration and nondialysis-nutrients (Figure 44). The first represented a "total dialysis" culture system where the entire culture environment was provided by membrane exChange. The second system was similar to the conventional dialysis culture trials reported previously (Section 5) utilizing conventional Sparger aeration. The third system represented a conventional nondialysis "control" fermentation which typified the batch culture process commonly used in the fermentation industry and served as a basis for comparison. As illustrated, the dialysis-aeration and dialysis-nutrient culture system supported a mean concentration of viable Serratia marcescens of 73 billion cells/ml after 48 hours Of growth. Although this was only half as great as the cell density achieved by the sparger-aeration and dialysis-nutrient culture 204 \I E \ (I) \l a: 9 IO - lu ~I .. m 3 _ x 9 __ o DIALYSIS NUTRIENT ond ._ 8 SPARGER AIR 4 A A DIALYSIS NUTRIENT and / g DIALYSIS AIR _ a NONDIALYSIS CONTROL 8 A I W I l I _ 0 I2 24 36 48 HOURS - Figure 44. Growth of Serratia marcescens on water diffusate of Trypticase soy broth under; dialysis-nutrient and dialysis-aeration, dialysis-nutrient and sparger-aeration, and nondialysis-nutrient and spargeriaeration culture‘ conditions. 205 system (mean of 147 billion cells/ml) after the same period of growth, the range of values observed between individual duplicated trials for each system often overlapped, indicating that the culture densities supported by the two were in closer correSpondence than the mean values alone suggested. The "total" dialysis culture system produced a viable cell density over eight times greater than the comparable nondialysis "control" culture system (mean of 8.5 billion cells/ml after 48 hours), which again demonstrated the superiority of dialysis-aeration over con- ventional sparger-aeration. The greatly enlarged difference between the dialysis—aeration and Sparger-aeration growth results shown here was due, in part, to the nutrient supply. In all Of the growth trials shown in Figure 44 the nutrients were provided by dialyzing the 3 liter culture fermentor, initially containing water, for 12 hours against a 12 liter reservoir containing Trypticase soy broth. Thus at inoculatiOn all of the systems contained a similar concentration of water diffusate medium. During growth the nondialysis Sparger-aerated "control" system did not receive the additional nutrients provided in the other two nutrient- dialysis systems. This contributed therefore to lower cell concentra- tions which were shown for the "control" in Figure 44. The results represented by Figures 43 and 44 clearly demonstrated that dialysis- aeration met the oxygen demand Of a concentrated viable Serratia marcescens culture and was equal or superior to sparger-aeration as a means of de- livering sufficient oxygen to the growing cells. Agitation Of the fermentation broth was unnecessary for adequate oxygen transfer by dialysis-aeration. .A greater viable cell density was achieved in a nonagitated dialysis-aerated fermentation than in a non-agitated sparger-aerated fermentation (Figure 45). Conventional 206 .. A T AGITATED DIALYSIS AIR 1 a 4 I AGITATED SPARGER 'AIR 1 3 \ a) H :I‘ In .0_ NONAGITATED DIALYSIS AIR 4 U ~0- lU . ~J 2 - NONAGITATED SPARGER AIR 3 . o o ~J S- . - 8% l l l I 0 I2 , 24 36 48 HOURS Figure 45. Correlation between growth of Serratia marcescens and the culture aeration method and agitation. 207 sparger-aeration produces bubbles which must be mixed and diapersed in order to facilitate oxygen transfer tO the liquid. Agitation has been shown to directly affect cell growth and yield in conventional fermenta- tions (163). Dialysis-aeration by nature of its permeation and diffusion of gaseous oxygen through the membrane interface into a relatively thin and turbulent liquid film does not require agitation for adequate oxygen transfer. Therefore elimination of agitation will not have as great an effect on oxygen availability to the cells of a dialysis-aerated culture as shown in the figure. However it was found that superior cell yields could be attained when the dialysis-aerated culture was agitated at 300 rpm (Figure 45). As shown, a Sparger-aerated and agitated (365 rpm) culture also attained improved growth, as expected, but did not achieve as great a cell concentration as the comparable dialysis-aerated and agitated culture. In both instances, with and without agitation, dialysis-aeration again demonstrated superior cell yields. The results indicate that agita- tion was necessary in order to achieve the biological potential of the aeration system used. With sparger-aeration, agitation primarily facili- tated oxygen transfer itself. In the case of dialysis-aeration, agitation represented the mixing necessary to reduce culture stagnation and to re- duce the liquid film around the cells themselves. It has been shown that an agitation rate of approximately 250 rpm is necessary to reduce the liquid film resistance around the cells to a negligible amount (15). That this was probably the case with dialysis aeration was illustrated by the nonagitated growth curves in Figure 45, where sparger aeration demonstra- ted a greater exponential growth rate. This was attributed to the advantage gained by the mixing action as the Sparged bubbles rose through the culture, an effect which was absent in the dialysis-aerated growth 208 trial. The necessity for agitation in order to achieve maximal biologi- cal growth with dialysis-aeration eliminated two potential advantages Of this aeration method, the complete elimination of antifoam agents and elimination of agitation power requirements. The results Of the growth trials with Serratia marcescens cultures revealed that dialysis-aeration not only supplied sufficient oxygen to the culture for growth but was able to support concentrated cell densi- ties equal to or greater than those achieved in comparable conventionally aerated trials. These results were repeatable and strongly support the feasibility of membrane aeration as a reasonable substitute for Sparger- aeration, at least on the scale demonstrated in these experiments. 6.5 Discussion This experimental work was intended to demonstrate the operational characteristics of the dialysis-aeratiOn system and its practical poten- tial for gassing microbial liquid cultures. The results were surprising. Dialysis-aeration not only was demonstrated to have potential but actually matched or exceeded conventional aeration in supporting high concentra- tions of an aerobic bacterium with high oxygen demand. These results were unexpected in view of the lower oxygen transfer rates and coefficients achieved by dialysis-aeration in comparison to conventional aerationg. and the utilization of membrane areas below that predicted to be Optimal. Apparently the manner by which dialysis-aeration introduced oxygen into the liquid medium (by molecular permeation and diffusion through a meme brane, in contrast to diffusion from rising bubbles) was more efficient in providing an adequate supply of dissolved oxygen at the bacterial cell- liquid interface. In other words, satisfaction of the oxygen requirement of the organisms themselves could be accomplished by a means which was up to 40 times poorer in mass transfer of oxygen into liquid alone. Analysis of the physical Operation of the dialysis-aeration system revealed several features potentially useful in the fermentation process. These included the elimination of positive pressure within the fermentor, the elimination of gas sterilization, the elimination of gas bubbles within the culture liquid and subsequent reduction of liquid evaporation by high volume sparging, the reduction of necessary antifoam agents, and the reduction of high agitation rates and correSponding power requirements. 209 210 The growing culture was effectively separated and isolated from the fermentor air supply by the dialyzer membranes. The Dacron-reinforced silicone rubber membranes have been proven reliable and rupture free in artificial lung devices and performed correspondingly in these investi- gations. The seal between the air and culture permitted the elimination of both entering and exiting gas sterilization, thereby effecting an operational economy unique to the dialysis aeration system. This same phase separation within the dialyzer and the elimination of gas Sparging reduced the pressure within the fermentor vessel to zero. The elimina- tion of positive pressure within the fermentor represented a feature of this aeration system particularly desirable for the culture of hazardous organisms because it greatly reduced the danger of aerosol contamination Of the laboratory area. Investigation of the Operational characteristics of the dialysis- aeration system revealed that the equipment capably handled gas velocities as high as 31 L/min, liquid velocities as high as 3 L/min, and up to six membranes in the dialyzer. With the exception of membrane area these represented maximums. Although no excessive pressures were detected at these values it was noted that pressure distribution was not equal in all dialyzer chambers and some flexing of internal components was Observed. The risk of membrane rupture and gasket leakage represented potential drawbacks for the system, although neither proved to be a prob- lem during these investigations. In order to insure the integrity of the system.va1ues of 2 L/min liquid velocity, 25 L/min air velocity and 300 rpm agitation were selected as operational standards. These repre- sented a reasonable compromise between maximal oxygen transfer, economi- cal Operation with minimal internal dialyzer stress, and a practical 211 efficiency which permitted confidence in the long term Operational re- liability Of the dialysis-aeration system. The bubblessdiffusion Of oxygen through the dialyzer membranes eliminated, to a large extent, the conventional function of agitation in liquid aeration. Indeed, little or no improvement in OTR values was observed when a dialysis-aerated nutrient broth was agitated. Be- cause Of this it was hypothesized that culture agitation might be eliminated altogether, therby precluding the need for antifoam agents and fermentor power requirements. However, growth trials revealed that this was not possible. Elimination Of culture agitation resulted in poorer growth yields for both sparger-aeration and dialysis-aerated cultures. The first was expected because agitation has been shown to be essential to the gas bubble dispersion and interfacial area maximiza- tion necessary for adequate sparger oxygenation (8, 23). Nonagitated dialysis-aerated fermentations achieved greater culture densities, demonstrating that agitation played a less important role in culture aeration by this method. Although this indicated that the above hypothesis was partially correct, the superior growth Observed when dialysis-aerated cultures were agitated showed that agitation could not be totally eliminated. Evidently the fluid circulation to and from the fermentor and the fluid turbulence within the dialyzer (which were sufficient for Optimal oxygen transfer to the liquid broth alone) were not sufficient to provide the bulk culture mixing necessary for Optimal oxygen transfer to the cells themselves. It has been reported that culture agitation of 250 rpm (.3 watts/L) was necessary in order to keep cells suSpended and minimize the formation of a stagnant liquid film, an oxygen transfer barrier, on the cell surfaces (15). Thus, at least minimal agitation 212 was required for maximal growth in dialysis-aerated cultures and as a result antifoam chemicals and fermentor power requirements could be reduced but not eliminated with this aeration system. The rate of dialysis-aeration Of nutrient broth was improved by increasing the air velocity, the liquid velocity, the membrane area, and the oxygen partial pressure. Increases in membrane area and liquid velocity produced the greatest improvements, results which correspond to those reported for artificial lungs (122, 56). Membrane area, since it represented the sole gas-liquid interface, was the critical variable in dialysis-aeration and the superior oxygen transfer rates observed with larger areas were expected. Unfortunately an insufficient supply of silicone rubber membranes of .007 inch thickness prevented a complete analysis of dialyzer membrane area on dialysis-aeration performance. An increase in area up to .1728 m2 (6 membranes), the maximum dialyzer area used, produced a 12 fold increase in OTR to .0205 m moles 02/L/min. This value, however, was only l/40th as great as the OTR produced by Sparger-aeration at the same air velocity. Additional membrane area would be eXpected to produce further increases in OTR values. However, the exact magnitude of improvement was difficult to assess because it was Observed that dialyzer mass transfer efficiency declined with the addition of membranes. Thus the actual oxygen transfer coefficient with six membranes was only 75% of the value calculated for that area. This decline in efficiency complicated prediction of the area necessary to produce a coefficient equal to that of a growing culture. Rough estimates indicated that as few as 13 membranes (assuming 100% efficiency) or as many as 20 membranes (assuming a poorer efficiency) might be required to meet the demands Of a growing culture. The Operational flexibility of the dialysis-aeration system did however allow adjustments in gas 213 and liquid velocities which effectively improved oxygen transfer rates and efficiency. An increase in these to the Operational maximums pro- duced a 2 fold increase in OTR to .0410 m moles OZ/L/min, the best value achieved for the system. The decline in oxygen transfer efficiency associated with the addi- tion of membranes has been Observed with artificial lungs (52) and was attributed to dynamic conditions within the dialyzer. The addition of membranes enlarged the internal volume Of the dialyzer which resulted in an uneven distribution of gas and fluid through the dialysis chambers, a reduction in bulk fluid velocity and turbulence in any given chamber, an increase in the laminar film thickness at the membrane surfaces, a decrease in internal gas pressure, and a greater flexing of internal dialyzer components all of which contributed in some degree to the re- duction of optimal conditions for mass transfer and therefore dialyzer efficiency. It appeared that the diminution in efficiency would limit the number of membranes which could be effectively used in a single. dialyzer. However the addition of two or more dialyzers to the system represented an alternate method for increasing the overall dialysis- aeration membrane area. Liquid velocity represented the second greatest influence on dialysis-aeration performance. Increases in liquid velocity resulted in substantially greater OTR improvement than corresponding increases in gas velocity, suggesting that the greatest resistance to oxygen trans- fer lay in the liquid phase. In both cases a velocity increase exhibited a larger improvement in performance when six membranes were used than when a single membrane was used. This effect was attributed to the change in internal dynamics which accompanied the addition of membranes. 214 As explained above, liquid and/or gas velocity increases helped to restore the optimal mass transfer conditions which prevailed when the dialyzer was equipped with one or two membranes and was most efficient. Faster flow through the dialyzer was believed to improve bulk movement, maintain a maximal gas-liquid concentration gradient, maximize turbu- lence and mixing at the membrane surfaces, and most importantly reduce film resistances by minimizing the thickness of the gas and liquid films at the membrane interfaces. Other investigations have shown that a direct relationship exists between flow rates and liquid turbulence and the thickness of laminar liquid films within the chambers of a dialyzer (63); These dynamic manifestations of velocity all influence the rate of oxygen transfer to some degree. The improvement in oxygen transfer gained by increasing gas and especially fluid velocities as observed with this dialysis-aeration system correSponds to the influence exhibited by velocity on solute transfer with the same dialyzer as reported in Section 4 and corresponds to the results reported elsewhere for other dialyzers of similar design (56, 102). The results of the velocity experiments demonstrated that liquid velocity had the greatest influence on OTR values. This data suggested that the liquid film was more important to oxygen transfer than the gas film. The status of liquid film resistance as a major governing factor in oxygen transfer has been emphasized in the literature (22, 23, 48) and was assumed in the theoretical discussion presented earlier in this section. Thus, on the basis of experimental data, the use of equation Na(0TR)= K a (C* - CL) L which was based on the importance of the liquid phase was valid for the 215 analysis of dialysis-aeration performance. Because the laminar films (gas and liquid) at the surface of the membranes represent mass transfer barriers which can be Operationally reduced and are regarded as a critical factor in membrane permeability it was deemed desirable to calculate these film resistances and the effect of velocity on their magnitude. In addition the dialysis-aeration system, by nature of its defined inter- facial membrane area and membrane resistance provided an Opportunity to evaluate film resistance more accurately than is possible with con- ventional bubble aerators. Ideally both the gas and liquid phase oxygen concentrations would be analyzed so that the transfer of oxygen from the gas into the liquid would be precisely known. However gas analysis equipment was not available necessitating the use of only dissolved oxygen data, a constant gas velocity, and a modified liquid phase re- sistance factor ‘1 in order to estimate the liquid film resistance 1‘11 and the influence of liquid velocity on it. The results showed a good correlation between liquid velocity increases, OTR improvements, and liquid film resistance decreases. These calculations supported the principle that increased liquid velocity improved dialysis-aeration because it reduced the laminar liquid film on the membrane surface thereby reducing a resistance to oxygen transfer. The determinations were made with the same constant gas velocity, which maintained a constant gas film resistance in all cases. These considerations plus agreement of the results with other published data (22, 48) and the previous theoretical considerations lent validity to the calculations and to the conclusion derived from them. Although the resistance values Obtained in these determinations could be criticized because the gas film contribution was not determined, the conclusions drawn from the principle they demon- strate should still be valid. The dialysis-aeration system may well 216 represent an excellent test bed for further studies of the aeration process. Additional investigations With gas analysis and dissolved oxygen analysis equipment would be a logical extension of the work begun here and might prove valuable in elucidating the magnitude and importance of the gas and liquid films in gas transfer through a mem- brane interface. The separation Of components and independent Operation of the gas and liquid phases in the dialysis-aeration system permitted a greater degree Of Operational flexibility than commonly afforded with conven- tional fermentation equipment. Manipulations, eSpecially with the gas phase, were possible without the develOpment of positive pressures, foaming, or mixing problems. The low solubility of oxygen in aqueous liquids made any adjustments which would increase the partial pressure of oxygen in the gas phase desirable in that they would increase the concentration gradient driving force for oxygen transfer. The use of counter-current liquid and gas flow and pure oxygen represented possible advantageous manipulations. The first, which was employed, improved oxygen transfer on the basis of the superior longitudinal gas and liquid oxygen concentration gradients which were created in the dialysis chambers on each side of the membrane. The use of pure oxygen however was economically impractical in terms of the long duration of most fermen- tations. Humidified air was selected as the most desirable gas for use in this system because it is relatively inexpensive and it reduced the possibility of liquid loss from the fermentor by vapor transfer through the membranes. Oxygen transfer could also be improved by increasing the overall pressure on the gas side of the dialyzer membranes. Generally the gas 217 pressure within the dialyzer was low. Restricting the effluent gas flow or increasing the overall gas velocity did increase this pressure to a small degree but the advantage gained in terms of improved oxygen partial pressure was not considered great enough to justify the risk of gasket or membrane failure or reduced gas throughput. Therefore the Operational gas and liquid velocities selected for general use were chosen as a reasonable compromise between operational reliability and oxygen transfer efficiency. Upward adjustments in gas velocity were found to be valuable in restoring the pressure of the gas side of the dialyzer when six mem- branes were used to that when a single membrane was used and the dialyzer was most efficient. In this case faster air velocity replaced the pressure which had been disipated by the increased internal vOlume associated with added membranes. The results of the physical aeration experiments demonstrated that the dialysis-aeration system does deliver oxygen to the liquid circulated through the dialyzer. The system had several characteristics advantageous to fermentations. As expected from theoretical predictions the membrane interface proved to be the principle factor in oxygen transfer with additional area providing an overall reduction in transfer resistance and surprisingly a prOportionate reduction in dialyzer efficiency. Liquid velocity increases reduced the liquid film thickness at the dialyzer membrane surfaces and represented the most effective Operational manipulation for improving dialysis-aeration rates. Although other operational adjustments which improved oxygen transfer were possible, the dialysis-aeration system did not quantitatively approach the oxygen transfer capacity Of a conventional Sparger-aeration system which pro- duced OTR values at least 20 times greater. 218 Dialysis-aerated Serratia marcescens cultures produced viable cell densities at least equal to and in some cases as much as two times greater than those Of comparable sparger-aerated cultures and eight times greater than previously reported results on this scale (62). The results of these dialysis-aerated growth trials were eSpecially surprising in view of the preceding discussion and the fact that the Operational conditions used (25 L/min gas velocity, 6 membranes, and 2 L/min liquid velocity) achieved an oxygen transfer rate (.0205 m.moles 02/L/min) which was approximately 40 times poorer than Sparger-aeration. In addition the calculated KLa (28 x 10.2 min-1) for this system was less than the maxi- mum of the range recorded for a Serratia marcescens culture. Equally remarkable were the results for nutrient-dialysis, dialysis-aeration growth trials. These achieved culture densities as great as 73 billion cells per ml which was over half of the density achieved in a Sparger- aerated, nutrient-dialysis system. The latter system has been noted for its potential for producing high concentrations of viable cells. The cell densities achieved in the nutrient-dialysis, dialysis-aeration system were again significantly greater than those achieved in a comparable Sparger-aerated control culture. The growth data could be accepted with confidence because the results Observed here for the control cultures corresponded closely with those recorded previously (Section 5) and the standard deviations for duplicated trials in all cases were similar to those Obtained in previous growth eXperiments. Examination Of the dissolved oxygen data for the growth trials showed that the dialysis-aerated cultures experienced a perceptible drOp in dissolved oxygen. Although this decline was greater than with Sparger- aeration, the dissolved oxygen concentration did not fall below the 219 critical level. It appears, therefore, that a Serratia marcescens culture of reasonably high density can be supported under conditions of low dissolved oxygen concentration as long as the dissolved oxygen con- centration is maintained above the critical level. This is supported by experiments which have shown that cellular oxygen uptake was inde- pendent of oxygen concentration provided it was in excess of the critical level (149). Further, the results suggest that the higher dissolved oxygen levels maintained by Sparger-aeration have no beneficial influence on cell viability or growth in batch cultures. This suggests that highly Sparged Operations waste a large portion of the oxygen introduced into the fermentor. This contention is supported by the low efficiency, 1% - 2%, reported for conventional aeration systems (48). In addition, Sparging increases evaporation losses, tends to strip CO2 from the cul- ture, and may produce oxygen concentrations which could be toxic to cell synthesis in some instances. These have been reported to be detri- mental to cell metabolism or synthesis and are effectively avoided with dialysis-aeration (35, 67, 78, 124). It may be concluded that dialysis- aeration, although inferior in oxygen transfer capacity with the membrane area used, was able to deliver dissolved oxygen in such a manner that the critical oxygen level was exceeded at all times, thereby enabling production and maintenance of a concentration of viable cells equal to or slightly better than that produced in sparged fermentations. The improvement in culture conditions provided by the removal of nutrient limitations and dilution of toxic accumulations by a conventional, Sparged dialysis culture system resulted in improved growth and higher culture densities. Under these conditions the excess dissolved oxygen provided by Sparger-aeration apparently contributed the additional oxygen 220 necessary for the greater viable cell density which resulted. Although the same system under dialysis-aeration achieved improved cell densities, these were less than shown above. Apparently the available membrane area was insufficient to provide the additional oxygen necessary to support such high culture populations. Nonetheless the results for dialysis-aerationjwhich did approach those for the conventional dialysis culture)were impressive when the small membrane area and inferior physi- cal transfer rates are considered. It would be reasonable to expect that equal or superior growth results could be achieved in the dialysis- aerated, nutrient-dialysis system if a greater oxygen exchange interface were available by addition of more silicone rubber membranes to the dialysis aerator. Incorporation of both nutrient-dialysis and dialysis- aeration into a single "total dialysis" system represented a culture system which would be potentially capable of producing high cell yields for prolonged periods while allowing independent control of the system components and providing isolation and maintenance of the culture by membrane interfaces. The results shown here establish the "total dialysis" system.as an Operationally feasible and perhaps a practical method for prOpagation of microorganisms. The transfer of oxygen by permeation of a gas through a highly receptive membrane interface represents the key to the success of the dialysis-aeration system and may provide an explanation to the seemingly contradictory physical transfer and biological growth results. The principle difference between dialysis-aeration and conventional Sparger- aeration lies in the transfer of gaseous oxygen through a silicone rubber membrane interface instead of a gas-liquid bubble interface. Although the latter has a greater area and a smaller resistance for mass transfer, certain dynamic conditions attributed to bubbles suspended in liquid tend 221 to reduce the efficiency and effectiveness of this form of aeration and produce a widely varying oxygen environment for the biological system. The dialysis-aeration system eliminates introduction of gas bubbles in the fermentation and shifts the transfer of oxygen from the interface of the bubbles rising through the liquid to the dialyzer membrane inter- £E££§.Which separate the thin turbulent films of gas and liquid. This method provides a continuous evenly distributed'transfer Of oxygen into the turbulent liquid film circulating from the fermentor by direct molecular diffusion from the membrane surface. This eliminates the sharp differences in oxygen concentration between bubbles, bubble trails, and the liquid and the rise, coalescence, and disappearance of bubbles all of which are characteristic of conventional sparger aeration. It is entirely possible that such a diffusional transfer better provides adequate dissolved oxygen throughout the entire culture volume than does sparging. It would follow that such a condition would improve and Optimize the microscale oxygen environment at the cell surfaces and would result, as Observed, in greater cell synthesis than eXpected from the macroscale oxygen transfer rates for the aeration system. In addition the elimination Of Sparged bubbles and reduction of foaming probably enhanced growth by reducing the metabolic shock or denaturing effects associated with exposure to large variations in dissolved oxygen concentration or violent agitation. The liquid flow rates selected for the dialysis-aeration growth trials (2 L/min) insured that the entire culture volume was cycled through the dialyzer every 1 1/2 minutes, assuming adequate mixing within the fermentor. The culture was continually exposed to an oxygen-rich inter- face, the net effect Of‘WhiCh was a more thorough distribution of dissolved oxygen and the elimination of stagnant poorly aerated areas 222 throughout the culture. Again this provided a more adequate transfer and retention Of oxygen within the vicinity of the reSpiring cells than does Sparging and may well eXplain why a dialysis-aeration system with a smaller interfacial area and poorer overall oxygen transfer capacity was able to support equal or superior culture densities. The successful application of dialysis-aeration was dependent on the availability of a suitable membrane material. A large resistance to mass transfer or too large a porosity have made most membrane materials inadequate or inapprOpriate for the oxygen transfer necessary to support a viable biosystem (61, 122). The development of thin reinforced silicone rubber membranes has made dialysis-aeration feasible. The composition of this membrane's chemical skeleton is such that molecular void Spaces are created by the easily rotated methyl groups within the polydimethyl Siloxane polymers. Although no liquid passage or seiving occurs this characteristic aids oxygen permeability because it reduces the entrOpy associated with membrane diffusion. This factor is, in most membranes, two to three times greater than that of a gas-liquid interface (7). In addition to the favorable tranSport characteristics of silicone rubber membranes, the continuous counter current circulation of oxygen depleted liquid from the fermentor and fresh oxygen (air) through the dialyzer permitted maintenance of a maximal oxygen concentration gradient at all times. This allowed the culture to be exposed to a relatively constant and maximal oxygen partial pressure throughout the fermentation, a char- acteristic which contributed to the effective support of microbial growth. Separation of the culture and gas phases also facilitated more precise control of these. The composition or partial pressure of the gas to which the culture was exposed was easily adjusted by manipulation of the gas flow through the dialyzer. Changes in the gaseous environment 223 to suit specific culture needs can be implemented quickly in this manner. Thus aerobic, anaerobic or intermediate conditions could be produced, maintained, or adjusted in order to satisfy culture requirements for either growth or metabolic synthesis. This Operational flexibility enhances the applicability Of dialysis-aeration to the wide variety of known fermentation processes. Dialysis-aeration appears to represent a feasible alternative to conventional aeration methods. The growth trial results indicate that this system deserves serious consideration both on the basis of its ability to support concentrated growth and on the Operational flexibility which permits more precise control of the culture environment. Although the dialysis transfer of oxygen was relatively low, Operational adjust- ments and the addition of membranes sufficiently increased its magnitude so that surprisingly good growth yields were obtained. The explanation for the excellent culture yields evidently lay in the elimination of gas bubbles from the fermentation medium and in the highly efficient process of oxygen diffusion and transfer through the Silicone rubber membrane interfaces in the dialyzer. The successful demonstration Of the Opera- tion and the applicability of dialysis culture, dialysis-aeration, or the combination of the two, "total dialysis", to known and experimental fermentation processes warrants continued investigation of the features and performance of these membrane mediated systems. 6.6 Summary A dialysis-aeration system was designed which incorporated the previously described dialyzer, assembled with gas-transfer membrane sheets, into a modified dialysis culture system. Liquid nutrient medium or a Serratia marcescens culture was circulated from a conventional 5 liter fermentor through the dialyzer past one face of the membrane(s). Humidified gas (air) circulated past the Opposite face. Unlike con- ventional air Sparging, this aeration system enabled the use of nonsterile air, eliminated bubble-liquid interfacial effects, and re- duced power and antifoam requirements. The greatest resistance to oxygen transfer, the membrane interface, could be reduced by increasing the area. A single membrane in the dialyzer (.0288 m2) produced a volumetric oxygen transfer coefficient (KLa) of 6.2 x 10.2 min.1 and an oxygen trans- fer rate (OTR) of .0018 m.moles 02/L/min. Six membranes (.1728 m2) in- creased these values to 28 x 10"2 min.1 and .0205 m moles OZ/L/min, respectively, but lowered the overall transfer efficiency of the dialyzer. Increases in liquid velocity, gas velocity, oxygen partial pressure, and overall gas pressure improved dialysis transfer of oxygen, with the first Showing the greatest effect. Even though the OTR.with six membranes was l/20th of that for a well sparged fermentor, dialysis-aerated cultures attained equal or greater pOpulation densities. In addition, comparable results were obtained when dialysis aeration was used in conjunction with nutrient-dialysis. Apparently the elimination of gas bubbles and the mechanism of membrane-mediated molecular oxygen transport effectively 224 III I D .30 225 satisfied the biological demands of a culture even though the physical transfer rates for the dialysis-aeration system indicated that it would be insufficient. 7. GENERAL SUMMARY Contributions were made to the design and construction of a new and versatile dialyzer for use in dialysis culture of microorganisms. It is assembled as a press with two rectangular stainless steel end plates, one of which contains entry and exit fittings. The plates compress an alter- nating series of Sheet membranes, thin stainless steel frames, and molded silicone rubber separators. Each separator forms a chamber adjacent to the membrane and in one piece provides gasketing, entry holes and chamber ports. A field of pyramidal elements, on each face of the separator, point-support the membrane and induce turbulent fluid flow within the chamber. The dialyzer measures 43 x 12.7 cm, has an effective area of 288 cm2 per membrane, and retains a volume of 65 ml per chamber. The dialyzer was incorporated in a dialyzer-dialysis culture system which utilized a reservoir and a culture fermentor with 10 to 30 liters and 3 to 10 liters, respectively. Dialysis efficiency for the system was evaluated by determining the half-equilibrium time (ETSO) and the overall permeability coefficient (Pm) for transfer of glucose solutions. Con- tinuous membrane use, autoclaving, dialyzer position, and fluid flow direction had no effect on transfer performance. However, ambient tem- perature, bulk fluid flow rates, and glucose concentration significantly affected transfer. A mean ET50 Of 144 1:15 min and a Pm Of 4.7 x 10-3 cm/min occurred with a 1% glucose solution and a single membrane. Opti- mum solute transfer rates were obtained with an area of 2,050 cm2 (7 mem- branes) and bulk flow rates above .5 liters per minute. Increasing the 226 227 volume of the total system or the ratio of fermentor:reservoir volume slowed these rates. In all cases, solute transfer from reservoir to fermentor approached an equilibrium concentration for the system. Dialysis efficiency was also evaluated by culturing Serratia.g§£- cescens, in comparison with nondialysis "control" growth trials. The bulk flow rate of the culture had no effect on the viability or cell yield. Increasing the initial nutrient concentration increased the growth rates but not the final cell yield, although the effect was more pronounced with a defined medium than with a natural broth. Continuously providing additional nutrients by dialysis maintained greater cell con- centrations in the fermentor than those observed in nondialysis fermen- tors, where nutrients were depleted during growth. Cell densities as high as 2.85 x 1011 cells/ml were maintained by dialysis for as long as six days. Increasing the fermentor:reservoir volume ratio had little effect on final cell yield. Increasing the dialyzer membrane area only Slightly affected growth rates, but Significantly increased final cell yields from 12.5 mg/ml dry wt. with 1 membrane to 40 mg/ml dry wt. with 15 membranes., The dialysis system with a fermentor:reservoir ratio Of 1:10 and with 15 membranes produced a maximum of 3.0 x 1011 cells/ml, with a 5% difference between viable and total cell counts. In contrast, .08 x 1011 cells/ml and a 58% difference were Obtained with a comparable nondialysis control. A dialyzer was equipped with silicone rubber membranes which separated culture flow (2 L/min) on one Side from humidified air flow (25 L/min) on the opposite side. This method for providing air by dialysis was incor- porated both into a dialysis culture system and a nondialysis culture system. Volumetric oxygen transfer coefficients (KLa, min-1) and oxygen 228 transfer rates (OTR, memoles 02/L/min) were near predicted values; they increased with gas and liquid velocity through the dialyzer, with p 02, and with culture mixing; and they increased from 6.2 x 10.2 and .0018, respectively, with one membrane (.0288 m2) to 28 x 10.2 and .025 with six membranes (.1728 m?) in the dialyzer. Although the OTR with six membranes was . 1720 Iof that in a well Sparged fermentor, dialysis aerated cultures consistently attained pOpulations of Serratia marcescens at least as high (.5 - 1.5 x 1011 cells per ml) as those in the control situation. Comparable results were Obtained when dialysis aeration was used in conjunction with nutrient dialysis. Thus, despite a lower measured rate of oxygen transfer, dialysis aeration appeared at least as efficient in meeting the oxygen demand of an aerobic culture (.003 mwmoles 02/mg cell wt/min) as conventional Sparging. Furthermore, dialysis aeration enabled use of nonsterile air, eliminated air-liquid interfacial effects, and reduced antifoam and power requirements, culture evaporation, and aerosol danger as compared to conventional air Sparging. BIBLIOGRAPHY 10. ll. 12. 8. BIBLIOGRAPHY ABBOTT, B. J. and P. GERHARDT. 1970. Dialysis fermentation. 1. Enhanced production of salicylic acid from naphthalene by Pseudomonas fluorescens. Biotechnol. Bioeng., Ig.press. ACHORN, G. B. and J. L. SCHWAB. 1948. A method for the aeration of liquid cultures of microorganisms. Science 107:377-378. AIBA, S. A., E. HUMPHREY, and N. FT‘WIIIJS. 1965. Biochemical Engineering. Univ. of Tokyo Press, Tokyo, Japan. AIDA, T., and K. YAMAGUCHI. 1969. Studies on the utilization of hydrocarbons by yeasts. .Abr. Biol. Chem. 3§;1244-1250. ALEXANDER, M. and P. W. WILSON. 1954. Large-scale production Of the azotobacter for enzymes. Appl. Microbiol. 2;135-l40. ARENDT, B. F. and R. H. MOORE. 1907. Filter Press. U. S. Patent 863,894. BARRER, R. M. 1941 Diffusion in and through solids. Cambridge Uhiv. Press, London. BARTHOLOMW, W. H., E. 0. KARCW, M. R. SFAT, and R. H. WILHEIM. 1950. Oxygen transfer and agitation in submerged fermenta- tions: Parts I and II. Ind. Eng. Chem. figgl801-1815. BASS, P., R..A. PURDON and J. N. WILEY. 1965. Prolonged administra- tion of atropine or histamine in a silicone rubber implant. Nature 208:591-592. BASSETT, H. P. 1938. Super-filtration by dialysis. Chem. Met. Eng. 4§;254-255. BATZDORF, U., R. S. KNOX. S. M. POKRESS, and J. C. KENNADY. 1969. Membrane partitioning of the rose-type chamber for the study of metabolic interactions between different cultures. Stain Technol. 44571-74. BENNETT, G. F: and L. L. KEMP. 1964. Oxygen transfer mechanisms in the gluconic acid fermentation by Pseudomonas ovalis. Biotechnol. Bioeng. 95347-360. 229 l3. 14. 15. l6. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 230 BLACK, S. H. 1966. Enhanced growth of Bordetella pertussis in dialysis culture. Nature 209:105-106. BLAKEBROUGH, N. and K. SAMBAMDRTHY. 1964. Performance of turbine impellers in Sparger-aerated fermentation vessels. J. Appl. BORKOWSKI, J. D; and M. J. JOHNSON. 1967. Long lived steam-steri- lizable membrane probes for dissolved oxygen measurements. Biotechnol. Bioengin. 2:635-639. BRALEY, S. A. 1964. The medical silicones. Trans. Am. Soc. Art. Int. Org. 195240-243. BRAMSON,‘M., J. HILL, J. OSBORN, F. GERBODE. 1969. Partial veno- arterial perfusion with membrane oxygenation and diastolic augmentation. Trans. Am. Soc. Art. Int. Org. 123412. BRANDL, E., A. SCHMID, and H. STEINER. 1966. Aeration in submerged fermentation. Biotechnol. Bioeng. 83297-213. BREWER, J. H. 1965. Apparatus for culturing microorganisms. British Patent 1,037,759. BRUBAKER, D. W. and K. KAMMERMEYER. 1953. Flow of gases through plastic membranes. Ind. Eng. Chem. 42:1148-1152. BULL, H. B. 1964. An introduction to physical biochemistry. F. A. Davis Company, Philadelphia, Pa. CALDERBANK, P. H. 1958. Part I: the interfacial area in gas-liquid contacting with mechanical agitation. Trans. Inst. Chem. Eng. 36:443-463. CALDERBANK, P. H. 1959. Part II: mass transfer coefficients in gas-liquid contacting with and without mechanical agitation. Trans. Inst. Chem. Eng. 31:173-185. CARNOT, P. and L. FOURNIER. 1900. Recherches sur le pneumocoque et ses toxines. Arch. Med. Exptl. 12:357-378. CHANG, T. M. S. and M. J. POZNANSKY. 1968. Semipermeable micro- capsules containing catalase for enzyme replacement in acata- lasaemic mice. Nature 218:243. CLARK, L. C. 1956. Membrane covered oxygen electrode. Trans. Amer. Soc. Art. Int. Org. 2341-46. CLIFTON, C. E. 1937. A comparison of the metabolic activities of Aerobacter aerogenes, Eberthella typhi, and Escherichia coli. J. Bacteriol. 325145-162. ‘osuIIu-l CII774II‘IIII'HI I-. III I-IIu I..~.I:::.IIHI ."IJL'I .II .o' I'D/..I-‘II a. ‘ \ ."III -"I-'\"" :AlllIrI' alullu. SIC/‘(IK'IL Jenna-«LI III. 111 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 231 OLIVER, D. O. 1968. Virus interactions with membrane filters. Biotechnol. Bioeng. 19:877-889. CLOWES, G. H. A., A. L. HOPKINS, and W. E. NEVILLE. 1956. An artificial lung dependent upon diffusion Of oxygen and carbon dioxide through plastic membranes. J. Thoracic. Surg. 32; 630-637. COLE, J. J., T. L. POLLARD, and J. S. MURRAY. 1963. Studies on the midified polyprOpylene Kiil dialyzer. Trans. Am. Soc. Art. Int. Org. 2:67-70. COLLANDER, R. and H. BARLUND. 1933. Permeabilitatsstudien an chara ceratOphylla. II. Die permeabilitat fur nichtelektrolyte. Acta. Botan. Fennica. 1151-114. COOPER, C. M., G. A. FERNSTROM, and S. A. MILLER. 1944. Performance of agitated gas-liquid contractors. Ind. Eng. Chem. §§:504-509. COPPOCK, P. D. and G. T. MEIKLEJOHN. 1951. The behavior Of gas bubbles in relation to mass transfer. Trans. Inst. Chem. Eng. ‘22:75-86. CRAIG, L. C. 1964. Differential dialysis. Science 144:1093-1099. DAGLEY, S., E. A. DAWES, and G. A. MORRISON. 1951. The effect of aeration on the growth of Aerobacter aerogenes and Escherichia coli with reference to the Pasteur mechanism. J. Bacteriol. 1:433-441. DANIEL, F. K. 1957. Dialysis. 13’ Kirk and Othmer (ed.), EncycloPedia of chemical Technology, Vol. 5. DATTA, R. L., D. H. NAPIER, and D. M. NEWITT. 1950. The prOperties and behavior of gas bubbles formed at a circular orifice. Trans. Inst. Chem. Eng. 28514-26. DAY, 3. w., D. c. CRYSTAL, c. L. WAGNER, w. R. KORESKI and J. M. KRANZ. 1964. Combination membrane oxygenator-dialyzer. Trans. Am. Soc. Art. Int. Org. 19:69-73. DeBECZE, G. and A. J. LIEFMANN. 1944. Aeration in the production of compressed yeast. Ind. Eng. Chem. 363882-890. DEINDORFER, F. H. and E. L. GADEN. 1955. Effects Of liquid physical properties on oxygen transfer in penicillin fermentation. Appl. Microbiol. 3;253-257. DEINDORFER, F. H. and A. E. HUMPHREY. 1961. Mass transfer from individual gas bubbles. Ind. Eng. Chem. 53:755-759. (I) :La:.;‘3 ._ l1I-. I _.I) I_n ‘ F ' II ‘ .5..- .0... . . -IJ . GEL/.021.) ___. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 232 EBNER, H., K. POHL and A. ENENKEL. 1967. Self-priming aerator and mechanical defoamer for microbiological processes. Biotechnol. Bioeng.‘2:357-364. ENGLAND, M. A. 1969. ‘Millipore filters studied in isolation and in vitro by transmission electron micrOSCOpy and sterioscanning electron micrOSCOpy. Exper. Cell Res. 24:222-230. ENGLESBERG, E., J. B. LEVY and A. GIBOR. 1954. Some enzymatic changes accompanying the shift from anaerobiosis to aerobiosis in Pasteurella pestis. J. Bacteriol. 68:178-185. ESMOND, W. G. and N. R. DIBELIUS. 1965. Perselective ultra-thin disposable silicone rubber membrane blood oxygenator: preliminary report. Trans. Am. Soc. Art. Int. Org. 11:325-329. ESMOND, W. G., M. STRAUCH, A. ZAPATA, F. HERNANDEZ, E. COX, A. LEWITINN and S. MOORE. 1967. Chronic periodic hemodialysis in Maryland. Bull. Sch. Med., University of Maryland. ‘21:2-16. FILIPPI, R. P., F. C. TOMPKINS, J. H. PORTER, R. S. TIMMINS and M. J. BUCKLEY. 1968. The capillary membrane blood oxygenator: in vitro and in vivo gas exchange measurements. Trans. Am. Soc. Art. Int. Org. 14:236-241. FINN, R. K. 1954. Agitation-aeration in the laboratory and in industry. Bacteriol. Rev. 18:254-274. FLYNN, D. S. and M. D. LILLY. 1967. A method for the control of dissolved oxygen tension in microbial cultures. Biotechnol. Bioeng.‘2:515-53l. FOLKMAN, J. and D. M. LONG. 1964. The use of silicone rubber as a carrier for prolonged drug therapy. J. Surg. Res. 43139-142. FOLKMAN, J., D. LONG and R. ROSENBAUM. 1966. Silicone rubber: a new diffusion prOperty useful for general anesthesia. Science 154:148-149. FREEMAN, R. B'.‘, J“. G. SETTER, J. F: MAHER, and G. E. SCHREINER. 1964. Characteristics and comparative efficiencies of coil and parallel flow hemodialyzers. Trans. Am. Soc. Art. Int. Org.'1Q:l74-182. FRIEDMAN, A. M. and E. N. LIGHTFOOT. 1957. Oxygen absorption in agitated tanks. Ind. Eng. Chem. 4251227-1230. FRIEDMAN, B., P. BLAIS, and P. SHAFFER. 1968. Fine structure of millipore filters. J. Cell Biol. 223208-211. GADEN, E. L., (ed.). 1964. 1962 fermentation review, agitation and mass transfer. Biotechnol. Bioeng. 6:14-17. 56. 57. 58. 59. 60. 61. 62. 63. 65. 66. 67. 68. 69. 70. 71. 233 GALLETTI, P. M., M. A. HOPF, and C. E. PEIRCE. 1962. A membrane lung-kidney. Trans. Am. Soc. Artif. Int. Org. 8:47-52. GALLUP, D. M. and P. GERHARDT. 1963. Dialysis fermentor system for concentrated culture of microorganisms. Appl. Microbiol. 11:506-512. GAN, H. K. 1963. Dialysis studies. Experiments dealing with the dialyzability of bacteria. A preliminary report. J. Hyg. Epidemiol. Microbiol. Immunol. 15422-435. GAN, H. K. 1969. Further studies on the dialyzability of bacteria. J. Hyg. Epidemiol. Microbiol. Immunol. 11:124-131. GENTRY, R. E. 1967. Reverse Osmosis, a pleasant inversion. Environ. Sci. Technol. 1:124-131. GERHARDT, P. and D. M. GALLUP. 1963. Dialysis flask for concen- trated culture of microorganisms. J. Bacteriol. 86:919-929. GERHARDT, P. and D. M. GALLUP. 1965. Process and apparatus for dialysis fermentation. U. S. Patent 3,186,917. GINZBURG, B. Z. and A. KATCHALSKY. 1963. The frictional coefficients of the flows of non-electrolytes through artificial membranes. J. Gen. Physiol. 41:403-418. GLADSTONE, G. P. 1948. Immunity to anthrax production of cell-free protective antigen in cellOphane sacs. Brit. J. Exp. Pathol. ‘ggz379-389. GOBEL, C. and L. BLUEMLE. 1959. Dialyzer. U. S. Patent 3,077,268. GORE, J. H. and T. PHILLIPS. 1964. Dissolved oxygen electrode. Biotechnol. Bioeng. 63491-496. GOTTLIEB, S. F. 1966. Bacterial nutritional approach to mechanism of oxygen toxicity. J. Bacteriol. 21:1021-1025. GOTTLIEB, D. and H. W. ANDERSON. 1948. The reSpiration of Strepto- coccus griseus. Science 107:172-173. GRAHAM, R. H. and T. P. RADERMACHER. 1952. Effects of dialyzer construction on transfer rates. J. Lab. Clin. Med. 49;27l-283. GRAHAM, T. 1861. Liquid diffusion applied to analysis. Proc. Roy. Soc. (London) 151:183-187. HALVORSON, H. O. 1957. Rapid and simultaneous Sporulation. J. Appl. Bacteriol. 195305-314. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 234 HARRIS, G., D. T. MOUNT, W. F. McLIMANS, K. TUNNAH, S. SCHEELE and G. E. MOORE. 1966. Gas monitor and control unit for cell culture systems. Biotechnol. Bioeng. 8:489-509. HARRISON, D. E. F. and S. J. PIRT. 1967. The influence of dissolved oxygen concentration on the respiration and glucose metabolism of Klebsiella aerogenes during growth. J. Gen. Microbiol. 46: 193-211. HEDEN, C. G. 1957. Pulsating aeration of mixed cultures. Nature 179:324-325. HEDEN, C. G. 1958. Large scale dialysis bag culture Of bacteria. Abst. 7th Intern. Cong. Microbiol. p. 410. HEIBIG, E. 1932. Dialyzer. U. S. Patent 1,849,622. HENNEBERG, G., and B. CRODEL. 1953-54. Electron microscOpic illus- tration of bacteria in sections of membrane filters. Zentralbl. Bakt. Abt. l Originale 160:605-613. HERBERT, D. 1965. Microbial respiration and oxygen tension. J. Gen. Microbiol. 415-8. HEROLD, J. D., J. S. SCHULTZ, and P. GERHARDT. 1967. Differential dialysis culture for separation and concentration of a macro- molecular product. Appl. Microbiol. 15:1192-1197. HIXON, A. and E. L. GADEN. 1950. Oxygen transfer in submerged fermentation. Ind. Eng. Chem. 4131792-1801. HOLLAND, F. A. and F. S. CHAPMAN. 1966. Liquid mixing and processing in stirred tanks. Reinhold Pub. Corp., New York. HOSPODKA, J. and A. CASLAVSKY. 1965. Design and application of electrodes for the determination of dissolved oxygen. Fol. Microbiol.,19:186-l99. HUMFELD, H. 1947. An improved laboratory scale fermentor for sub- merged culture investigations. J. Bacteriol. 54:689-696. HUNT, J. R., J. H. SADLER, J. H. SHINABERGER and P. M. GALLETTI. 1968. Laboratory and clinical evaluation of a small counter- current dialyzer, the Miniklung. Trans. Am. Soc. Art. Int. Org.'14:109-113. ISREELI, J. 1958. Dialysis apparatus. U. S. Parent 2,864,507 and U. S. Patent 3,072,259. JOHNSON, M. J., J. BORKOWSKI and C. ENGBLOM. 1964. Steam sterilizable probes for dissolved oxygen measurement. Biotechnol. Bioeng. 6:457-468. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 235 JOHNSON, M. J. 1967. Aerobic microbial growth at low oxygen con- centrations. J. Bacteriol. 24:101-108. KAUFMANN, T. G. and E. F. LEANARD. 1968. Studies of intramembrane transport: a phenomenological approach. A. I. Ch. E. Journal 14:110-117. KIIL, F. 1960. Development of a parallel flow artificial kidney in plastics. Acta. Chir. Scand. Suppl. 253:142-150. KLECK, J. L. and J. A. DONAHUE. 1968. Production of thermostable hemolysin by culture of Staphylococcus epideomidis. J. Infect. Dis. 118:317-323. KOCH, W. and D. KAPLAN. 1953. A simple method for Obtaining highly potent tetanus toxin. J. Immunol. 19:1-5. KOLFF, W. J. 1954. Dialysis in treatment of uremia. Arch. Int. Med. 24:142-160. KOLFF, W. J. 1957. The artificial kidney--past, present and'future. Circulation 12:285-294. KOLLSAMAN, P. 1959. Dialyzer. U. S. Patent 2,891,900. KOROSY, F. de, and E. ZEIGERSON. 1966. Electronmicroscopy of permselective membranes. Israel J. Chem. 4:85-95. KRIEGER, J. M., G. W. MULHOLLAND and C. S. DICKEY. 1967. Diffusion coefficients for gases in liquids from.rates of solution of small gas bubbles. J. Phys. Chem. 11:1123-1129. LAMANNA, C. and M. F. MALLETTE. 1965. Basic Bacteriology, p. 521-526. Williams and Wilkins 00., Baltimore, Maryland. LANDE, A. J., s. J. DOS, R. G. CARLSON, R. A. PERSCHAU, R. P. LANGE, L. s. SOUSTEGARD, and c. w. LILLEHEI. 1967. A new membrane oxygenator-dialyzer. Surg. Clin. N. Amer. 41: 1461-1470. LAVENDER, A. R., F. W. MARKLEY and M. FORLAND. 1968. .A new pump- less, parallel-flow hemodialyzer. Trans. Am. Soc. Art. Int. Org.,14:92-98. LEMP, J. F. 1960. Polarographic measurement of oxygen supply and demand during aerobic propagation of a bacterial culture. J. Biochem. and Microbiol. Technol. Eng.‘1:215-225. LEONARD, E. F. and In W. BLUEMLE. 1958. Factors influencing permeability in extracorporeal hemodialysis. Trans. Am. Soc. Art. Int. Org. 4:4-13. ~4 LI .91; 1'43 (f I I" 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 236 LEONARD, E. F. and L. W. BLUEMLE. 1959. Engineering in medicine: design of an artificial kidney. Trans. New York Acad. Sci. .21z585-598. LI, N. N., R. B. LONG and E. J. HENLEY. 1965. Membrane separation processes. Ind. Eng. Chem. 21:18-29. LOCKHART, W. R. and R. W. SQUIRES. 1963. Aeration in the laboratory. Adv. Appl. Microbiol. 2:157-187. LONGMUIR, I. S. 1954. Respiration rate of bacteria as a function of oxygen concentration. Biochem. J. 22:81-87. LYMAN, D. J. 1964. New synthetic membranes for the dialysis of blood. Trans. Am. Soc. Art. Int. Org. 12:17-20. LYMAN, D. J. and B. H. L00. 1967. New synthetic membranes for dialysis. IV. A co-polymer-urethane membrane system. J. Biomed. Mater. Res. 1:17-26. MACLENNAN, D. G. and S. J. PIRT. 1966. Automatic control of dissoived oxygen concentration in stirred microbial cultures. J. Gen. Microbiol. 42:289-302. Manegold, E. 1929. Die dialyse durch kollodiummembranen und der qusammenhang zwischen dialyse, diffusion and membranstruktur. Kolloid-Z. 42:372-395. MARINO, S. P. 1968. Growing increased yields of microorganisms. Chem. Abstracts 22:916. MARX, T. I., B. R. BALDWIN, and D. S. MILLER. 1962. Factors in- fluencing oxygen uptake by blood in membrane oxygenators. Ann. Surg. 156:204-212. MAXON, W. D. and M. J. JOHNSON. 1953. Aeration studies on prOpa- gation of baker's yiest. Ind. Eng. Chem. 42:255482560. METCHINKOFF, E., E. ROUX and T. SALIMBENI. 1896. Toxine et anti- toxine cholerique. Ann. Inst. Past. 12:257-282. MILLNER, S. N. 1969. Apparatus for preparation Of bacterial extracellular enzymes. Appl. Microbiol. 11:639-640. MUIRA, Y. and S. HIROTA. 1966. Transfer Of oxygen supplied by aeration in submerged fermentation. J. Ferm. Technol. 44: 890-901. NEWITT, D. M., G. C. SHIPP and C. R. BLACK. 1951. Recent devel- Opments in the theory and practice of agitating and mixing. Trans. Inst. Chem. Eng. 22:278-289. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 237 NURMIKKO, V. 1955. The dialysis technique in the study of the vitamins and amino acids affecting associations of micro- organisms. Acta. Chem. Scand. 2:1317-1322. OLSON, B. H. and M. J. JOHNSON. 1946. Factors producing high yeast yields in synthetic medium. J. Bacteriol. 21:235-246. PATTLE, R. E. 1950. The aeration of liquids: Parts I and 11. Trans. Inst. Chem. Eng. 22:27-37. PEIRCE, E. C. 1962. The membrane lung, a new multiple support for teflon film. Surgery St. Louis. 22:777-783. PEIRCE, E. C. and N. R. DIBELIUS. 1968. The membrane lung: studies with a new high permeability co-polymer membrane. Trans. Am. Soc. Art. Int. Org. 14:220-226. PEIRCE, E. C. and G. PEIRCE. 1963. The membrane lung: the influence of membrane characteristics and lung design on gas exchange. J. Surg. Res. 2:67-75. PHILLIPS, D. H. and M. J. JOHNSON. 1961. Aeration in fermenta- tions. J. Biochem. Microbiol. Technol. Eng. 2:277-309. PIRT, J. S. 1957. The oxygen demand of growing cultures of an aerobacter species determined by means of continuous culture technique. J. Gen.‘Microbiol.‘12:59-75. RAHN, O. and G. L. RICHARDSON. 1941. Oxygen demand and oxygen supply. J. Bacteriol. 41:225-249. RAHN, O. and G. L. RICHARDSON. 1942. Oxygen demand and oxygen supply. J. Bacteriol. 44:321-332. RICHARDS, J. W. 1961. Studies in aeration and agitation. Prog. Ind. Microbiol. 2:143-172. RICKLES, R. N. 1967. Membranes: technology and economics. Chemical Process Monograph Series, Noyes DevelOpment Corp., Park Ridge, New Jersey. RICKLES, R. N. and H. Z. FRIEDLANDER. 1966. The basics of membrane permeation. Chem. Eng. 12:163-168. RITTER, H. 1949. A method for cultivating HemOphiluS influenzae. J. Bacteriol. 21:474-475. ROBB, W. L. 1965. Thin silicone membranes: their permeation prOperties and some applications. Gen. Electric Res. and Devel. Ctr. Report #65-C-03l. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 238 ROSE, G. G. 1967. The circumfusion system for multipurpose culture chambers. I. Introduction to the mechanics, techniques, and basic results of a 12 chamber (in vitro) closed circulatory system. J. Cell Biol. 22:89-112. ROSE, G. G., M. KUMEGAWA, H. NIKAI, M. CATTONI, and F. HU. 1969. The HFH-l8 mouse melanoma in roller tube, chamber, and circum- fusion system cultures. Cancer Res. 22:2010-2033. ROSE, G. G., M. KUMEGNWA, H. NIKAI, M. BRACHO, and M. CATTONI. 1970. The dual-rotary circumfusion system for Mark 11 culture chambers. I. Design, control, and monitoring of the systems and the cultures. Microvas. Res. 2:24-60. ROSE, G. G., C. M. POMERAT, T. O. SHINDLER, and J. B. TRUNNEL. 1958. A cellOphane-strip technique for culturing tissue in multipurpose culture chambers. J. Biophys. Biochem. Cytol. 4:761-764. ROSENAK, S. S. 1953. Continuous extracorporeal dialyzer. U. S. Patent 2,650,709. SADOFF, H. L. 1955. The effect of electrolysis on the growth of an aerobic bacterium. Ph.D. thesis, Univ. of Illinois. SARIN, C. L., A. SEN GUPA, H. P. TAYLOR and W. J. KOLFF. 1966. Further develOpment Of an artificial placenta with the use of a membrane oxygenator and venovenous perfusion. Surgery 22:754-760. SAVINO.,F.;M. 1963. Artificial kidney. U. S. Parent 3,074,559. SCHAUMBURG, F. D. and E. J. KIRSCH. 1966. Anaerobic simulated mixed culture system. Appl. Microbibl. 14:761-766. SCHULTZ, J. S. and P. GERHARDT. 1969. Dialysis culture of micro- organisms: design, theory, and results. Bacterial. Rev. 33:1-47. SILCOX, H. E. and S. B. LEE. 1948. Fermentation. Ind. Eng. Chem. 42:1602-1607. SINGER, S. 1970. Pseudomonas penetration Of membrane filters. Tech. Rept. 70-1, Pall Corporation, New York. p. l-l3. SKEGGS, L. T. and J. R. LEONARDS. 1948. Studies on an artificial kidney: 1. Preliminary results with a new type of continuous dialyzer. Science 108:212-213. SMITH, C. G. and M. J.‘JOHNSON. 1954. Aeration requirements for the growth of aerobic microorganisms. J. Bacteriol. 22:346-350. I). 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 239 SORTLAND, L. D. and C. R. WILKE. 1969. Growth of Streptococcus faecalus in dense culture. Biotechnol. Bioeng. 11:805-841. STANNETT, V. and M. SZWARC. 1955. The permeability of polymer films to gasses--a sinple relationship. J. Polymer Science 16:89-91. STARKS, O. B. and J. KOFFLER. 1949. Aerating liquids by agita- tion on a mechanical Shaker. Science 109:495-496. STEEL, R. 1958. Biochemical Engineering. Macmillan Press. New York. STEELE, R. and W. D. MAXON. 1966. Dissolved oxygen measurements in pilot and production scale novobiocin fermentations. Biotechnol. Bioeng. 2:97-108. STERNE, M. and L. M. WENTZEL. 1950. .A new method for the large- scale production Of high-titre botulinum formoltoxoid types C and D. J. Immunol. 22:175-183. STEVENS, HENRY P. 1938. Apparatus for purifying rubber latex. U. S. Patent 2,127,791. STEWART, R. D., J. c. CERNY and H. I. MAHON. 1964. The capillary kidney: preliminary report. Uhiv. Mich. Med. Ctr. J. 22: 116-118. STRICH, E. G. 1962. A method of producing yeast. U. S. Patent 3,068,155. SYKES, G. 1965. Disinfection by sterilization, pp. 188-201. 12. Sterilization by Filtration, 2nd Ed., J. B. Lippincott, Philadelphia. ' TAGUCHI, H. and A. E. HUMPHREY.- 1966. Dynamic measurement of the volumetric oxygen transfer coefficient in fermentation systems. J. Ferm. Technol. 44:881-889. TSAO, G. T. N. and L. L. KEMPE. 1960. Oxygen transfer in fermen- tation systems. I. Use of gluconic acid fermentation for determination of instantaneous oxygen transfer rates. J. Biochem- Microbiol. Technol. Eng. 2:129-L42. TSAO, G. T. N. 1968. Vortex behavior in the Waldhof fermentor. Biotechnol. Bioeng. 12:177-188. TUWINER, S. B. 1962. Diffusion and membrane technology. A.C.S. monograph #156. Reinhold Publishing Corp., New York. VALLEY, G. and L. F. RETTGER. 1927. The influence of carbon dioxide on bacteria. J. Bacteriol. 24:101-135. . .. 4.1.1.“: 1' {"0" Lri 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 240 WANG, D. I-C. and A. E. HUMPHREY. 1968. Developments in agita- tion and aeration of fermentation systems. Prog. Ind. Microbiol..2:1-34. WEBER, G. H. 1940. Dialyzer grid support. U. S. Patent 2,225,024. WEST, J. M. and E. L. GADEN. 1959. Agitation effects in yeast prOpagation. J. Biochem. Microbiol. Technol. Eng. 1:163-172. WINSLEY, H. S. and J. FOLKMAN. 1967. Silicone rubber: oxygen, carbon dioxide and nitrous oxide measurement in gas mixtures. Science 157:103. WINSLOW, C. E. A., H. H. WALKER, and M. SUTERMEISTER. 1932. The influence of aeration and of sodium chloride upon the growth of bacteria in various media. J. Bacteriol. 24:185-208. WISE, W. S. 1951. The measurement of aeration of culture media. J. Gen. Microbiol. 2:167-177. WOLF, A. V., D. G. REMP, J. E. KILEY and G. D. CURRIE. 1951. Artificial kidney function: kinetics of hemodialysis. J. Clin. Invest. 22:1062-1070. WOLF, L. and S. ZALTZMAN. 1968. Optimum geometry for artificial kidney dialyzer. Chem. Eng. Prog. Sump. Ser. 24:109-111. WOLNAK, B. and L. F. BARRINGTON. 1962. Chemical oxygen genera- tion "in situ". U. S. Patent 3,041,250. WYNVEEN, R. A; and K. M. MONTGOMERY. 1967. An experimental oxygen concentrating system. AerOSpace Medicine. ‘22z712-718. YOSHIDA, F., A. IKEDA, S. IMAKAWA, and Y. MUIRA. 1960. Oxygen absorption rates in stirred gas-liquid contactors. Ind. Eng. Chem. 22:435-438. ZAPOLu‘W. M., J. E. PIERCE, G. G. VUREK, R. L. BOWMAN, and T. KOLOBOW. 1969. Artificial placenta: two days Of total extrauterine support of the isolated premature lamb fetus. Science 166:617-618. ZETELAKI, K. and K. VAS. 1968. The role of aeration and agita- tion in the production Of glucose oxidase in submerbed culture. Biotechnol. Bioeng. 19:45-59. ZIEMINSKI, S. A. and D. R. RAYMOND: 1968. Experimental study of the behavior of single bubbles. Chem. Eng. Sci.‘22:l7-28. ZOBELL, C. E. and J. STADLER. 1940. The effect of oxygen tension Of the oxygen uptake of lake bacteria. J. Bacteriol. 22: 307-322. MATERIAL REFERENCES 10 ll 12 9. MATERIAL REFERENCES American Instrument Company, Inc. Silver Spring, Maryland American Instrument Company, Inc. Silver Spring, Maryland American Instrument Company, Inc. Silver Spring, Maryland Bamel Corporation 8051 W. Chicago Blvd. Detroit, Michigan Bausch and Lomb, Inc. 60967 Bausch Street Rochester, New York 14602 Bausch and Lamb, Inc. 60967 Bausch Street Rochester, New YorkI 14602 Beckman Instruments, Inc. 24755 Five Mile Road Detroit,‘Michigan Beckman Instruments, Inc. Scientific and Process Instr. Div. 24755 Five Mile Road Detroit, Michigan BiOQuest, Inc. 1640 Gorsuch Avenue Baltimore, Maryland Bodine Electric Company 2500 West Bradley Place Chicago, Illinois Central Scientific Company 1700 Irving Park Road Chicago, Illinois 60613 Chemical Rubber Company 2310 Superior Avenue Cleveland, Ohio 241 Electronic Relay Lolog-Flexible-type Immersion Heater, 1000 Watt "Quickset" Bimetal Thermo- regulator Stainless Steel Plates and Bolts Precision Refractometer Spectronic 20 Colorimeter Model 76 EXpandomatic pH Meter and Frit Junction Combination Electrode 39030 Oxygen Analyzer Model 77 Trypticase-Soy Broth QAgar 1.5%) Bodine Type N Shunt wound D.C. Motor 1/8 H.P. Cenco Drying Oven #95470. Air Line CRC "Pollutex" Filters 13 M 14 M 15 M 16 M 17 M 18 M 19 M 20 M 21 M 22 M 23 M 24 M 25 M 26 M 242 Corning Glass WOrks Corning, New York Crawford Fitting Company 884 East 140th Street Cleveland, Ohio Detroit Silicone Rubber Co. 10439 Northlawn Detroit, Michigan 48204 Dow Chemical Company 1000 Main Street Midland, Michigan Edon Industrial Products 4412 Fernlee Royal Oak, Michigan Gelman Instrument Company P. O. Box 1448 Ann Arbor, Michigan Marshalltown Manufacturing, Inc. Marshalltown, Iowa Medical Products Division Down Corning Corporation Midland, Michigan Michigan Industrial Sales 2540 Park Avenue Detroit, Michigan ‘Mixing Equipment Company, Inc. 135 Mt. Read Boulevard Rochester, New York New Brunswick Scientific Company 1130 Somerset Street New Brunswick, New Jersey New Brunswick Scientific Company Somerset Street P. 0. Box 606 New Brunswick, New Jersey Sargent Company 4647 West Foster Avenue Chicago, Illinois #7740 Pyrex Glass Back Plates Swagelok Fittings Silicone Rubber Separators and Gaskets Polyglycol P-2000 Edon Power Unit Model 624-3 Flowmeters Pressure Gauges Silastic Oxygenator Membrane Elliott Flexible Drive Shaft SB-27l Lightnin Model L Mixer Gyrotory Tier Shaker Model G52 Fermentor Models F-05 and F-l4 Rubber Tubing Used - 1/4" I. Diam. x 3/16" Wall Thickness Recorder Model S-R 27 M 28 M 29 M 243 Tuthill Pump Company 1716 W. Hubbard Street Chicago, Illinois Union Carbide Corporation Food Prod. Div. 6733 W. 65th Street Chicago, Illinois U.S. Army Chemical Corps Fort Detrick, Maryland Maisch Biological Metering Pump. Model HQDCC, 11029104 Dialysis Tubing ("Visking") Regenerated Cellulose Serratia W strain 8-UK t‘ 'nlmxfii‘c I :1 APPENDIX I.“ "7" 10. APPENDIX 10.1 Dialyzer Cost Ana1ysis An attempt was made to construct the dialyzer from commercially F- available materials, such as Swagelok fittings and standard size bolts IV '.'o-.I_W a and nuts, to reduce costs. However the final design included some custom built parts, such as the molded Silicone separators which required design and construction of custom master dies ($3,000) and the professional mold- ing and trimming. The values given below include the labor costs for machining, drilling, trimming, etc., for each piece but do not reflect the time and effort Spent in develOpment or purchase of manufacturing machinery (i.e. mold dies or trimmers). Twelve dialyzers have been manu- factured to specifications, with the Detroit Silicone Rubber Company responsible for separator and gasket production and the University of Michigan machine shop responsible for the preparation of the metal parts. 244 245 A. Exterior Dialyzer Shell Stainless steel front plate $ 88.00 Stainless steel back plate 88.00 Alternate pyrex glass back plate ($ 30.00) 12 bolts and nuts, 4 swageloks, 5 clamping bars 20.00 TOTAL $196.00 ($138.00) B. Dialysis Membrane Unit 2 stainless steel separator frames $ 12.00 2 1/32 inch silicone rubber gaskets 4.00 2 silicone rubber separators 24.00 1 dialysis membrane 0.00 TOTAL PER UNIT $ 40.00 For each membrane used in the dialyzer the above components are necessary. Thus, as additional membranes are added a similar number of units will be required and the cost will rise prOportionally. “III II'III’II I 6 o 3 o 3 9 2 1 "I H N“ " H "I H H H