LARKIN, LITSKY AND FULLER

nate b)eanls inoculated with S. faccalis wvas successful. Atemperatuire of 88 C for 1 miniute wN-as sufficient to ob-taimi a 100 per cenit kill.

It FERE NC,L SBURTON, N. 0. 1949 Comparisoni of coliform and entero-

coccus organisms as indices of pollution in frozen foods.Food Research, 14, 434-438.

BT-TTERFIELD, C. T. 1932 The selection of dilution water forbacteriological dilutions. J. Bacteriol., 23, 355-368.

LARKIN, E. P., LITSKY, W., AND FULLER, J. E,. 1955 Fecalstreptococci in frozen foods. I. A bacteriological survey ofsome commercially frozen foods. Appl. Microbiol., 3, 98-101.

PEDERSON, C. S. 1947 Significance of bacteria in frozenvegetables. Food Research, 12, 429-438.

RIUDOLFS, W., FALK, I,. I,., ANI) RAGOTZKIE, It. A. 1951aContamination of vegetables grown in l)olluted soil. I.Bacterial conttamination. Sewage ani(l Ind. Wastes, 23,253-268.

BRDOLFS, W., FAI,K, I. IL., ANI) RAG(OTZKIE, R. A. 19511)Contamination of vegetables growni in polluted soil. IV.Bacterial decontamination. Sewage and Ind. Wastes, 23,739-751.

RUDOLFS, W., FALK, L. L., AND RAGOTZKIE, R. A. 1951c Coni-tamination of vegetables grown in polluted soil. VI. Ap-plication of results. Sewage and Ind. Wastes, 23, 992-1000.

WEBER, G. R., AND BLACK, JL. A. 1948 Laboratory procedurefor evaluating practical performance of quaternary ammo-nium and other germicides proposed for sanitizing foodutensils. Am. J. Public Health, 38,1405-1417.

Microbiological Process ReportContinuous Fermentation

A Discussion of Its Principles and Applications

WILLIAM D. MAXON

Research Laboratories, The Upjohn Coompany, Kalarnazoo, Michigan

Received for publication October 28, 1954

For the purposes of this discussion, fermentation willbe defined in its broadest sense. It is a process in whichmicroorganisms catalyze the conversion of a suitablesubstrate into a desired product. The process may oIrmay not require oxygen, and the microorganisms them-selves may be included among the desired products.Thus, the definition in no way limits the process toone of anaerobic dissimilation, as it does in the classicsense. Fermentation processes may be classified accord-ing to the manner in which the substrate is added andthe product withdrawn. Thus, in a batch fermentation,the substrate is charged initially, and, when the fer-mentation is complete, the product is withdrawn.Continuous operation involves addition of the sub-strate in an unbroken stream and withdrawal of fer-mented medium in the same manner. In semicontinu-ous processes the substrate is added and the productremoved at intervals, and in this way certain charac-teristics both of batch and continuous operation arecombined. The present discussion will be confined tothe strictly continuous case, with occasional referenceto those batch and semicontinuous processes whichact as prototypes or serve as illustrations. Continuousfermentations may be of many types with respect tothe equipment used and with respect to the flow offermenting media through that equipment. In analyz-

ing them one type will be considered as basic, that is,the homogeneous flow fermentor. Here, the fermenta-tion occurs in a single vessel, agitated so that its con-tents are homogeneous. The nutrient stream is coIn-tinuously supplied, and the fermentor contents, withthe product they contain, are continuously withdrawnin order to establish a steady state. Other types of con-tinuous fermentation will be discussed as modificationsof this basic system.A continuous fermentation has striking advantages

over the corresponding batch process. The primaryadvantage is a marked reduction in processing time,with the same holding capacity of equipment or, alter-natively, a marked reduction in equipment size forthe same rate of production. Furthermore, as a conse-quence of steady-state operation, the product can beexpected to have greater uniformity. A continuous fer-mentation is more adaptable to instrumental control,and it is better integrated into such other parts of theover-all process as the preparation of medium and therecovery of product, which may be operated moreeconomically and efficiently in a continuous manner.From an experimental point of view, a continuous fer-mentation is ideal. At steady state, the time factor iseliminated, and the effect of nutrients, temperature,pH, agitation, aeration, and other factors upon the

110

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CONTINUOUS FERMENTATION

fermentation may be established easily without thecomplication of continually changing conditions. Therate of growth, as will be seen later, becomes an inde-pendent variable (to a degree) and may be studied forits effect on yield and other dependent factors. Monod(1950) presented an excellent theoretical discussionof the use of continuous fermentation as an experi-mental tool. Novick and Szilard (1950, 1951) used thistool effectively for the study of mutation rates in bac-teria; Maxon and Johnson (1952, 1953) used it for astudy of aeration and agitation in submerged culture.

There are, however, disadvantages in the use of con-tinuous fermentation which have limited its wide-spread adoption. Perhaps the major drawback isinherent in the nature of the operation. If the fermen-tation step in a process is operated continuously, it isdesirable to have all other steps of preparation andrecovery also in continuous operation. In such a proc-ess, a failure in one step will force a complete shut-down. Since fermentation, as in the case of the manufac-ture of antibiotics, may be extremely complex, anoccasional failure must be expected. If these failuresbecome frequent, continuous operation will lose itsadvantages, whereas in batch operation only one batchis lost with each failure. Failure may come from manysources, mechanical or instrumental. There are others,such as contamination and genetic instability, commononlv to certain fermentations. These will be discussedlater.

Although the limitations of continuous fermentationare recognized, they are not in every case insurmount-able. In the simpler fermentations, such as the propaga-tion of food and fodder yeast and the production ofethyl alcohol, the method has proved its worth. Asour experience increases it is probable that more andmore fermentation products will be made in this manner.

This discussion approaches the subject of continu-ous fermentation first from a theoretical viewpoint.Equations that govern the relationships of throughput,microbial propagation, product formation, substrateutilization, and so forth are presented. However, sincea fermentation is a highly complex living, system, a

strictly mathematical treatment of all the factors isneither possible nor, for the present, desirable. Instead,the practical significance of a few key equations is dis-cussed and an attempt made to define the scope of theirapplication. Combination of theory and experienceleads to a description of the effect of operating variableson yield, productivity, and the like in various continu-ous systems.For this discussion, fermentation processes will be

classified according to their purpose:I. Propagation of microorganisms

II. Product formationSeveral variations of the basic homogeneous flow fer-

mentor will be considered:

A. Single vesselB. MultivesselC. RecyclingD. Two phaseFor each process type, the theoretical discussion

will be followed by a description of actual operations,as recorded in the literature. To assist the reader infollowing the development of the equations, all symbolsand their meanings are listed below:

C The concentration of contaminating organisms in thefermentor or effluent stream.

c The rate of growth of contaminating organisms, con-centration per unit time.

F The rate of continuous feed to the fermentor, volumeper unit time.

K, A constant in the kinetic expression for the effect ofsubstrate concentration on growth rate (equation 7).

1\L The concentration of mutant organisms in the fermentoror effluent stream.

m The rate of growth of mutant organisms, concentrationper unit time.

P The concentration of fermentation product in the fer-mentor or effluent stream.

p The rate of product formation, concentration per unittime.

r The throughput rate, volume of feed per unit time perunit operating volume.

S The concentration of substrate limiting to growth inthe fermentor or effluent stream.

s The rate of utilization of substrate, concentration perunit time.

V The operating volume of liquid in the fermentor.X The concentration of the desired organism in the fer-

mentor or effluent stream.x The rate of growth of desired organism, concentration

per unit time.Y The yield of desired organism on the substrate used, x/s.Yp The yield of fermentation product on the substrate

used, p/s.0 Time.01, The holdup or retention time in a fermentor, 1/r.0,,' The total holdup time in a series of fermentors.,I The Naperian growth rate of desired organisms, coIn-

centration per unit time per unit concentration.0 The proportion of a cell population that mutates per

generation.Subscripts

a Refers to conditions in the first fermentor in a series.b Refers to conditions in the second fermentor in a series.c Refers to contaminating organisms.i Refers to initial conditions at time, 0, equal to zero.m Refers to mutant organisms.o Refers to conditions in feed.p Refers to fermentation product.

Propagation of Microorganisms

Theoretical considerations-single vessel. Followingthe derivations of Monod (1950) and Golle (1953) forthis type of continuous fermentation, it will be assumedthat an agitated and aerated fermentor is being con-

tinuously fed fresh nutrient medium at a constantrate, F, and that fermenting medium is being coIn-

ill

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

WILLIAM D. MAXON

tinuously withdrawn from the fermentor at the samerate. Thus, a constant volume of fermenting medium,V, is maintained in the fermentor. It will be furtherassumed that the contents of the fermentor are beingagitated sufficiently to insure homogeneity and thatthe entering fresh medium is thoroughly mixed withthe fermenting medium in a short time. The amount ofagitation necessary for optimum fermentation is usu-ally more than enough for this purpose.A material balance for one component of the system

leads to a generally applicable equation:

1) FXo + xAT = FX + V dX 'd6

That is, the rate at which a component is being fed ti) afermentor plus the rate at which it is being producedis e(lual to the rate at whieh it is wxithdrawin plus therate at which it accumulates in the fermentor. For thepresent, let us assume that there is no product, in thiscase microbial cells, in the entering stream. If equation1 is applied to this component, then X0 = 0 aind:

2) xV = FX + V dX/dOWe shall define several more terms at this point:

r The throughput rate = F/V.61, The holdup or retention time = 1/r.,u The Naperian growth rate (Monod, 1950) = x/X =

(In 2/generation time).

When r is incorporated in equation 2, we find:

:3) dX/dO = x - Xr

And, since at steady state the concentration of cellsis not changing (dX/d6 = 0)

4) x/X = 1 = r

Y and ,u are incorporated into equation 3 to give:

6) dS/dO = r (S0- S) - (X/Y),uUsing equation 6 we can demonstrate the stability

of the steady state, r = ,u. At equilibrium dS/dO iszero, of course, but if the concentration of cells is dis-turbed so that X is less than its equilibrium value,then (X/Y), will be less than r (So - S). Since dS/dOwill then be greater than zero, S, the substrate con-centration, will increase. Now, the growth of cells is,of course, an enzymatic reaction, or rather the summa-tion of many such reactions, and may be expected tofollow the general behavior of such reactions withrespect to the effect of substrate concentration on rate(Michaelis and Menteni, 1913). Thus, as an approxima-tion at least:

7) S1. = jOin ax K + S

Data from Harris et al. (1948b) demonstrate that thistype of relationship exists in the growth of Torula utilison wood hydrolysate and sulfite waste liquor. Equation7 indicates that increasing S causes ,u to increase, and/. becomes greater than r. A rearrangement of equation.3 givres:

8) dXXdG = x/X - r = A- r

Thus, with ,u greater than r, X will increase in the ex-ponential fashion dictated by equation 8. As X in-creases, the growth rate will diminish once againuntil steady state is reached at ,u = r. Here, dS/dOmust be zero, and from equation 6:

9)By closer consideration of the behavior of this system

we can demonstrate what the forces are that bringabout this equilibrium condition in which the Naperiangrowth rate is equal to the throughput rate, and we canalso show what the limitations of such an equilibriumare.

If a fermentation is supplied with adequate aeration,the metabolic rates will ordinarily be limited by theavailability of one of the required nutrients. When thegeneral material balance for a continuous system, equa-tion 1, is applied to the substrate that is limiting togrowvth:

FSo- sV = FS + V dS/dO

and:

5) dS/do = r (S(- S) - s

Now, a yield constant, Y, the proportion of substrateused that is ineorporated into the cells formed, is de-fined:

Y = x/s

X = Y (So - S).

Similar reasoning will demonstrate the return tosteady state which occurs when X is greater than itsequilibrium value. This derivation is valid only whenthe yield, Y, has one definite and unique value at eachgrowth rate in the given system. This is essentiallytrue for purely aerobic metabolism, but it is not truein certain other cases (see Mixed metabolism).

Consider now what happens to a fermentation of thissort operating at steady state when the throughputrate is changed. If r is decreased, equation 6 shows thatdS/dO is less than zero and S decreases. DecreasedS causes decreased 1., according to equation 7. Theeffect continues until once again ,1 = r and steady stateis re-established at the new throughput rate. As before,X = Y (So - S). A similar re-establishment of equi-librium will occur if r is increased.There is a limit, however, imposed theoretically by

the fact that the growth rate will not exceed a certainmaximum. Under optimum conditions, this maximumis specified by equation 7 at the point where S = S,,or essentially, since K, is usually small with respect to

112

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CONTINUOUS FERMENTATION

So, IU = itmax- If r exceeds Amnax then the cells will"wash out," and X will decrease to zero in an exponen-tial manner according to equation 8:

dXXdO = rX-I'.

These material-balance and kinetic derivations formthe theoretical basis for most continuous fermentationprocesses. We are led to the following major conclu-sions for the case of continuous addition and with-drawal from a single vessel, in which purely aerobicmetabolism is occurring to produce microorganismsas the desired product:

1) The Naperian growth rate, ,u, is equal to thethroughput rate, r, for steady state. This is a stableequilibrium which is regained despite disturbances,provided the growth rate is less than its maximum.

2) The population at steady state is uniquely deter-mined at each throughput rate by the yield of cellsfrom substrate, the concentration of substrate enter-ing, and the concentration of substrate in the fermen-tor: X = Y (S0 - S).

3) The Naperian growth rate is a function of thesubstrate concentration in the fermentor, and thusthe substrate concentration depends upon the through-put rate.

In a practical process, the desired results are a maxi-mum rate of product formation for a given fermenta-tion capacity and at the same time a small proportionof unconverted raw material in the effluent. That is,we want to maximize x and minimize S. In order tomaximize x, it is necessary to consider the factorsthat contribute to it. From equation 4:

10) x = rX = r Y (So - S)

As r is increased, x will also increase antil decreasing Yand increasing S cause x to reach a maximum and thendecline. If the substrate is not highly valuable withrelation to the product, then the throughput ratewhich gives this maximum rate of product formationwill be economically optimal. However, if the unusedsubstrate is of relative value, a lower throughput ratewill be necessary to minimize this loss. In many cases

of microbial propagation, as pointed out by Monod(1950), the yield is nearly independent of the growthrate. Furthermore, it is often the case that the con-

stant, KS, of equation 7 is small relative to the sub-strate concentration. This means that r, the through-put rate, may be held at a value only slightly less thanMmax without a significant decrease in product forma-tion rate or an appreciable concentration of substratein the effluent.

Further increases in product formation can often beachieved by increasing the concentration of substratein the feed, So. By this procedure the cell concentrationis increased, while the throughput rate is held constant.

This is possible, however, only if the growth rate islimited by the substrate concentration. It may be thatin an aerobic propagation process the availability ofoxygen will limit the growth rate. This will almostcertainly become the case as the cell concentration isincreased by increasing the substrate concentrationin the feed. When this state is reached, x can only befurther increased by increasing the rate of oxygentransfer to the respiring cells. Otherwise a higher con-centration of cells must be accompanied by a lowerthroughput rate. The experience of the Germans inproducing food yeast by the Waldhof Standard Process,as reported by Saeman, Locke, and Dickerman (1945,1946), was that higher substrate concentratioins neces-sitated lower addition rates. The yeast output remainedthe same.

Increase of production rate by increasing the avail-able oxygen is possible only within limits. A highlyaerated culture contains a large proportion of entrappedair. The actual amount of liquid holding capacity, V,of a given fermentation vessel is thereby proportion-ately decreased. This, of course, diminishes the totaloutput rate, F, for a given throughput rate, r. The maxi-mum output that can be achieved must be determinedexperimentally. In selecting the proper degree of aera-tion, a balance must be struck between the desirableeffect of higher cell concentrations and the undesirableeffect of decreased operating volume. High degrees ofaeration will also necessitate the increased use of de-foamers, mechanical or chemical.

Recycling the organism. The productive capacity of acontinuous propagation may be further increasedover that indicated by equation 10 by the expedientof recycling a portion of the cells produced. If thegeneral material balance equation (equation 1) isused, but Xo, the concentration of cells in the enteringstream, is not zero, then:

* FXo+xV = FX +VdX/dO

and by further manipulations in the same manner asbefore:

11) dX/do = x - (X - X0) r

Again at steady state dX/dO = 0, and r no longer isequal to ,u, but:

12) x MXr = or r = XX- Xo X -Xn

Thus, the throughput rate may be increased over thegrowth rate by the factor X (X - X0). It was shownin the case where there was no recycling that X =

Y (So - S) (equation 9). Similar reasoning will demon-strate that for the present case X = Y (So - S) + Xo.

Let us assume some values for the factors involved inorder to demonstrate the magnitude of the advantageobtained by recycling. Suppose that yeast is being

113

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

WILLIAM D. MAXON

produced continuously on a medium containing 10grams/liter (S0) of glucose, the limiting nutrient. Thegrowth rate is 0.3 hr-1, and under these conditions theconcentration (S) of glucose in the effluent is negligible.The yield is found to be 50 per cent at this growthrate. If there is no recycling, then the productioncapacity, x, from equation 10 is:

x = r Y (S0- S), where r = A

= 0.3 X 0.5 X 10 = 1.5 grams/liter hour.

Suppose that recycling is used, that is, cells are sepa-rated from the product stream and returned to thefermentor with the feed. Assume that the rate of re-cycling is such that Xo is 10 grams/liter. Theii:

x = r Y(S0 -S), where r = , X

= M [Y(So -S) + XO]

= 0.3 (0.5 X 10 + 10) = 4.5 grams/liter hour.

By increasing the concentration of active growingcells, the catalyst for this reaction, by a factor of 3,as a result of recycling % of the cells produced, afactor of 3 increase is realized in productive capacity.This is the equivalent of increasing the limiting sub-strate concentration in the feed, as previously dis-cussed. It has the additional advantage of not necessi-tating a change in So, which might involve an expensiveoncentration process. In an aerobic process, recycling,

as with increasing substrate concentration, will beadvantageous only when excess oxygen is available orcan be made available without excessive loss of hold-ing capacity.More than one vessel. It is sometimes desirable to

operate a continuous fermentation in several vesselsin series. That is, the product of the first fermentor flowscontinuously as feed into a second fermentor, and so oa,the stream passing through as many vessels as desired.

In discussing the theoretical behavior of such a sys-tem let us assume that there are two fermentors inseries. The first of these will conform to the equationsdeveloped for the single vessel; the second behaves as asingle vessel with recycling. The capacity of the com-bination may then be compared to that of a single ves-sel by use of the appropriate relations already derived.For a single vessel the holdup time is Oh = l/I. For twotanks in series the holdup time is the sum of that ineach:

Oh' = 1/# + I/M (Xb; X.) (See equationi 12.)

In a propagation system where the growth rate isnot a marked function of substrate coincentration abovea certain minimum it may be that: u = ua = M,b TIl

this case it may be shown from the above equationsthat:

Oh = (2 - X ) Oh.

Since the concentration of cells is greater in the sec-ond tank (Xb) than in the first (Xa), it is evident thatthe holdup time, Oh', for the two tanks in series is upto two times longer than that for a single tank, Oh.Thus, there is no advantage in using more than onefermentor if the growth rate is the same in each.However, if the fermentation is such that a higher

growth rate can be maintained in the first tank becauseof the presence of a higher concentration of limitingsubstrate, then an advantage can be demonstrated.In this case Ma will be greater than lb while Ab may equalM, the growth rate in a single tank. This situation wouldalso occur if there were two substrates in the medium,one capable of supporting a higher growth rate thanthe second. In the first tank, the more easily assimil-able nutrient would be used, and the harder to as-similate nutrient left as the limiting substrate in thesecond fermentor. In either case:

Oh' = I/Ma + Oh (I - X )

If we assume, for example, that Ma = 2,M and thatXa = 0.9 Xb, then

Oh = 1/2M + Oh (1 - 0.9) = 0.5 Oh + 0.1 Oh = 0.6 Oh.

The holdup time for the two fermentors in series isreduced significantly over that for a single fermentordoing the same job. This means, of course, a reduc-tion in total tankage for the same rate of productformation.Mixed metabolism. In the previous considerations,

the simplifying assumption was made that in aerobicmicrobial propagation the substrate is converted by asingle definite series of metabolic processes into cellularmaterial, the over-all result being, for example:

glucose (s) -* yeast (x) + CO2.This is approximately true in some cases, such as inthe propagation of Torula utilis. It is not true, however,in others. In bakers' yeast, Saccharomyces cerevisiae:

glucose (s) -* yeast (x) + ethanol (p) + CO2.

Here, there are two processes competing for theavailable substrate. It has been shown (Maxon aIndJohnson, 1953) that even under excess aeration bothprocesses may operate in a ratio depending upon therate of substrate utilization (figure 1), but that below acertain rate only the oxidative metabolism is evidentand I10 ethanol is formed.

In a fermentation of this type, the yield will be highlydependent on the substrate concentration, and thus

114

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CONTINUOUS FERMENTATION

on the growth rate and the throughput rate. This isbecause one of the competing metabolic processes givesa lower yield than the other. If r is changed in the rangeof growth rates where both types of metabolism areoccurring, the changing yield will cause the steady-state population to change (equation 9). Finn and Wil-son (1953) demonstrated this effect in the aerobicpropagation of Bacterium linens.

It may also be that in a continuous system wheretwo competing types of metabolism are occurring thesteady-state population is not a unique function of thethroughput rate, as was seen to be the case with purelyaerobic metabolism. Suppose, for example, that a por-tion of a continuous culture is rapidly withdrawn andreplaced by fresh medium. This brings about a simul-taneous decrease in the cell concentration and an in-crease in the substrate concentration away from theirformer values. However, the steady state may not berestored by a return to these values as with unmixedmetabolism. (See Theoretical Considerations.) Instead,since th'e changed conditions cause changed metabolismand reduced yield, a new steady state is established ata lower cell concentration. This type of behavior alsohas been shown to occur with Bacterium linens (Finnand Wilson, 1953).

In a practical process, propagation should, of course,be carried out with the highest yielding metabolicprocess in sole operation. The region of mixed metabo-lism is to be avoided. Thus, with bakers' yeast, for ex-ample, the growth rate should be maintained at lessthan that which brings about' the formation of ethanol.

Applications. The prime practical example of micro-bial propagation by purely aerobic metabolism in onecontinuously operated vessel is found in the manufac-ture of food and feed yeast. A thorough descriptionof this industry has recently become available (Wiley,1954). Many production installations have been oper-ated, but their general characteristics are similar.Briefly, the fermentation is carried out in a single,large fermentor. For example, the fermentor used atLake States Yeast Corp., Rhinelander, Wisconsin(Inskeep, et al., 1951) has an operating capacity of45,000 gallons of air-liquid emulsion or 20,000 gallonsof liquor. Such fermentors are provided with agitationand a supply of air. The Waldhof system has met withgreat success for this purpose; it has been used inseveral installations in Germany (Saeman, Locke andDickerman, 1945, 1946; Holderby, 1946) and in thiscountry (Inskeep et al., 1951). Agitation and air areboth supplied by a single, hollow-bladed impeller.This is located near the bottom of the tank under a

central draft tube. Circulation down through this tubecaused by the impeller brings about effective mixingand aeration and at the same time beats the foam intothe emulsion. Cooling by internal or external heatexchange is required. Fresh medium is continuously

70

40

0U

Rf 30

20

10 [

2 3 4 5 6 7 8

M1.1P.IMOL.E:S GLUCOSE PER HOUR PER GRAM OF YEAST (DRY WEIGHT)

FIG. 1. The effect of glucose utilization rate upon themetabolism of bakers' yeast under excess aeration. Per cent

yield is g dry weight of yeast per 100 g glucose used. Per centglucose glycolyzed is g glucose to ethanol per 100 g glucoseused. Medium B: 1 per cent glucose. Medium C: 10 per centglucose. (From Maxon, W. D., and Johnson, M. J. Ind. Eng.

Chem., 45, Figure 6, page 2559, 1953.)

fed and product withdrawn. The limiting nutrient isusually the fermentable sugar and may be supplied inmolasses, sulfite waste liquor, wood hydrolysate, or

agricultural wastes. Supplementation with salts ofnitrogen, phosphorus, potassium, and occasionallyother elements is usually required. Aqueous ammoniais often added to serve the dual purpose of pH con-

trol and nitrogen source. A process in which the ni-trogen source acts as the limiting nutrient has beenpatented by Seidel (1943).The organism most often used in food and feed yeast

processes is Torula utilis. Other similar organismshave been used with success in Germany, for example,Candida arborea and certain poorly identified and mixedcultures (Saeman, Locke and Dickerman, 1945, 1946).The growth of S. cerevisiae in continuous culture

has been the subject of investigation for some time.DeBeeze and Rosenblatt (1943) discussed a number ofthe early processes. Irvin (1954) also mentioned some

of the work in this field. A patent assigned to the Fleish-mann Company (Buhrig, 1929) describes a two-vesselsystem, a main fermentor and an auxiliary finishingfermentor operated continuously in series. Sak (1928,1932a, 1932b) holds patents involving continuousyeast propagation in multiple vessel systems with re-

cycling. Other patents on processes of this type are

held by Harrison (1930), Olsen (1930), and Daranyi(1936). Unger et al. (1942) worked with a pilot plantfor the continuous propagation of distillers' yeast in a

single vessel and arrived at design figures for a full-scale yeast plant.

Propagation by these methods is in no way limited

to yeast. For example, Gerhardt (1946) found thatBrucella suis could conveniently be maintained in

* MEDIUM B (pH 5.0 - 5.2)

X MEDIUM C (pS 4.6 - 4.7)

EXCESS AERATION

x_/ PERCENT YIELDN

* PERCENTGLUCOSEGLYCOLYZED

_@_Y _ 9_ 1 |-

ll5

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

WILLIAM D. MAXON

TABLE 1. Tlypical produictive capacities fo)r tnicrobial propaujatiorn as reported in the literature: Continaous addition, and wvithdrawalfromit a single vessel

Organism

Torula utilisY'orula utilisTorula utilisT'orula utilisUnknown species

Saccharomyces cerevisiaeSaccharomyces cerevisiae

A erobacter aerogenes

Brucella suis

Substrate

Sulfite waste liquorSulfite waste liquorWood hydrolysateFermentation residuePear waste

Wood hydrolysateGlycose-synthetic me-

diumCasein-yeast extract

Glucose tryptose

2.61.74.60.918

4.510

* Dry weight of cells.

continuous culture on a laboratory scale. Elsworthand Meakin (1954) presented information on thecontinuous propagation of Aerobacter aerogenes in a 20-liter vessel both on glycerol as a limiting substrate andon a casein-yeast extract medium. No successful processof this type for the production of a mycelial organismhas been reported, however, except for a brief mentionof the continuous propagation of Aspergillus oryzae as

a medium supplement in a paper by Ruf et al. (1948).Attempts were made in Germany to manufactureOidium lactis, a mycelium-forming organism, the so-

called "Biosyn" (Saeman, Locke and Dickerman, 1945,1946). Difficulties with contamination in continuousfermentation were encountered and forced abandon-ment of this type of operation.

In table 1, the rate of product formation and otherpertinent data are given for several representative con-

tinuous fermentations reported in the literature. Datawere taken for industrial, pilot, and laboratory opera-tions. Comparison of the productivity with that ofbatch fermentations demonstrates the marked advan-tage in this respect of a continuous process.

Contamination and genetic instability. Contamina-tion in fermentation may be a serious problem. A con-

tinuous fermentation, which is designed to operate forlong periods of time, is particularly liable to the occa-

sional introduction of undesired organisms. The mere

entry of foreign organisms into a continuous culturedoes not, however, assure its failure. Golle (1953) de-veloped equations for the theoretical behavior offermentation under these conditions. These equationsare paraphrased here:Suppose that a single vessel in continuous operation

is being contaminated with a foreign organism at a rateC0r. Then, by application of the general materialbalance equation, as in the development of equation 11

for the case of recycling:

13) dC/dO= c- (C-Co)r

75905598

9499.5

4.52.82.22.81.7

3.73.1

1.3

3.7

Production Capacity

2.1 g/liter hr*2.8 g/liter hr8.9 g/liter hr1.5 g/liter hr27 g/liter hr

5.7 g/liter hr13.2 g/liter hr

2.7 g/liter hr

6.) X 104 cells/literhr

Reference

Saeman et al. 1945Harris et al. 1948aHarris et al. 1948aHarris et al. 1948aRamage and Thomp-

son, 1948Harris et al. 1948bMaxon and Johnson,

1953Elsworth and Meakin,

1954Gerhardt, 1946

The growth rate of the foreign organism, /uL, is definedby: A, = c/C. Use of these quantities in equation 13gives:

14) dC/d6 = C(,c- r) + rC0

Upon integration:

15) ln C((Q r) + rC = -r)r) +

where C is the concentration of contaminating or-

ganisms at any time, 0, and Ci is the initial concentra-tion at 0 = 0.

There are three possibilities with respect to thegrowth rate of the foreign organism in the fermentationthat it is contaminating. It can be greater than, equalto, or less than the throughput rate. If I., is greater thanr (which equals ,u), then equation 15 shows that C willincrease exponentially with time. This will continueuntil the steady-state concentration of limiting sub-strate is reduced to the point where IA, = r accordingto a relationship such as is expressed by equation 7.Under these conditions, the growth rate of the originalorganism will be less than r, and its concentration willdecrease to zero exponentially according to equation 8.After a period of time the original organism will be en-

tirely replaced by the contaminating organism. Aninfection by such an organism, once started, will resultin complete failure of the fermentation.

If the growth rate of the foreign organism is equal tothe throughput rate, then according to equation 14 itsconcentration will increase linearly at a rate rC0. Con-tamination by an organism with such a growth rate willnot be serious if the rate of entry can be kept negligiblysmall.

If the growth rate of the contaminant is less than thethroughput rate, equation 15 shows that its concentra-tion will approach a limit at infinite time:

C-C.r

r -c

Reducing Reducing Holdup TimeSugar in Feed Sugar Used Oi, = 1/r hrs

116

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CONTINUOUS FERMENTATION

A contaminationi by such an organism will become seri-ous only if its rate of entry is extremely high anid itsgrowth rate only slightly less than that of the desiredorganism.

In the manufacture of food and feed yeast cointami-nation has been no problem. In fact, no great precau-tions are taken to prevent it. This is because it is possibleto maintain a high throughput rate under nutrient con-ditions that are unfavorable for the growth of undesiredorganisms. That is, the pH is kept low, and media suchas sulfite waste liquor and wood hydrolysate, whichcontain inhibiting substances, are employed.The Aerobacter aerogenes culture described by Els-

worth and Meakin (1954) was maintained continuouslyfor 2600 hours. Proper precautions prevented entranceof foreign organisms. Only experience with a given con-tinuous fermentation, however, can establish howserious a deterrent to successful operation contamina-tion will be. It is probable that with strict care mostmicroorganisms can be produced free of contamina-tion.A problem related to contamination is mutation.

When mutation of an organism in steady-state cultureoccurs, the fate of the mutant depends upon relation-ships similar to those for a foreign organism. Equationsfor these relationships have been developed by Golle(1953) and by Novick and Szilard (1950, 1951).When mutation occurs in a single vessel in continuous

operation, it is governed by the following materialbalance, where 4 is the proportion of the total popula-tion that mutates in each generation. The number ofgenerations per unit time can be shown to equal,u/ln 2. Then:

dM/dO = m - rM + 4XMA/ln 2

or

16) dM/dO = M(Mm - r) + IXA/ln 2

This equation is of the same form as equation 14 for thecase of contamination. In the same line of reasoningthen, if the growth rate of the mutant, MAm, is greaterthan the throughput rate, the entire culture will bereplaced by the mutant form. If Mm is equal to thethroughput rate, the concentration, M, will increaselinearly at a rate 4XMi/ln 2, and if Mlm is less than thethroughput rate, M will approach a limiting value:

M =,XX (rIn 2 k-MmA

From these relationships it is evident that a con-tinuous propagation system acts as a mutant selector.Novick and Szilard (1950, 1951), indeed, made ingenioususe of this fact for the study of mutation rates ofEscherichia coli. It should be recognized that the or-

ganism in a steady-state system will not necessarily beof the same genetic form as that of the organism orig-inally introduced. It will, rather, be the spontaneous

mutant with the highest affiniity for the limiting nutri-ent.

This is usually an advantage when the production ofmicroorganisms is the goal, since a higher productivitycan be achieved. It may be in a product formationfermentation, however, that a mutation toward highersubstrate affinity will be accompanied by a lower rateof product elaboration. The eventual population, result-ing from mutant selection, will then have a lower pro-ductivity than the original culture. The seriousness ofthis problem would have to be determined experi-mentally for each process. It is not one with a readysolution.By microscopic examination of bacteria in a culture

chamber, Powell (1954) showed that the generationtime of individual bacteria varies widely from theaverage (from 6 to 50 minutes for Streptococcus faecalis).This is an erratic variation and is not consistentlyinherited by later generations. Powell concluded thatselection in a continuous culture will not occur as aresult of this type of variation, but only as a result ofactual mutation.

PRODUCT FORMATION

Although many important applications of continu-ous fermentation lie in the field of microorganism pro-duction, there are still more, whether actually in use ormerely proposed, in the field of product formation.Antibiotics, enzymes, solvents, organic acids, and so onare all fermentation products that might be advantage-ously made in a continuous system.

Theoretical considerations. As with propagation, thesimplest continuous product formation system consistsof a single vessel with continuous addition of feed andcontinuous withdrawal of fermenting medium. Similarequations apply in this case. The growth rate of theorganism is again equal to the throughput rate, and theconcentration of cells is, as before, determined by

9) X = Y(So- S).

It is no longer desired, however, to maximize x, therate of cell formation per unit operating capacity, butrather to maximize the rate of product formation, p.This rate is determined by material balance and kineticconsiderations, as is the growth rate, and similar equa-tions can be used to express the relationships. Thematerial balance expression is based on the yield ofproduct, Yp, from the rate-limiting substrate and isanalogous to equation 10:

17) p = rP = rYp(So - S)

The kinetic factors are summarized by:

18) p = XX

The specific rate of product formation per unit ofcell concentration, X, is, of course, a function of the

117

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

WILLIAM D. MAXON

limitiiig substrate conicentrationi, S, anid the concentra-tion of product, P.Under givein conditionis of throughput rate, einviron-

ment, and nutritioni, the cell concentrationi is fixed at asteady-state value according to the previous considera-tions for microbial propagation. The quantities S, Y,P, and X will also be fixed at equilibrium levels, as de-fined by equations 17 and 18. In this way the produc-tivity, p, at steady state is established.

Product formation associated with growth. In certainfermentations, the growth process and the productformation process are closely linked. The same substrateconcentration may be limiting to the rate of both proc-esses, and these rates may respond to changes in thisconcentration in a similar manner. Such a fermentationis the alcoholic fermentation of glucose by yeast. Here,it is the energy resulting from the anaerobic dissimila-tion of glucose that permits the yeast to proliferate. Incases of this sort, the effect of throughput rate on pro-ductivity in a single vessel may be analyzed accordingto the concepts described for microbial propagation. Asr is increased, the limiting substrate concentration, S,must increase according to equation 7, so that ,u remainsequal to r. This increased S causes the specific rate ofproduct formation, X, to increase also, since X and ,uare similar functions of S. In this manner, productivityincreases with increasing throughput, as specified byequations 17 and 18, until increased residual substrateand decreased population affect the rate of productformation sufficiently to cause a decline. The econom-ically optimum throughput rate will be less than thatfor maximum product formation rate by an amount de-pending upon the relative value of the unused substrate.

It is possible to predetermine the effect of throughputrate upon population level in continuous, anaerobicyeast fermentation by the method of Adams and Hun-gate (1950). Residual glucose, and presumably ethanolconcentration, may also be predicted. The procedure isto run a batch fermentation on the desired nutrientmedium. Periodic analyses reveal the cell concentration,X, the glucose concentration, S, and the ethanol con-centration, P, at various times. The growth rate, ,i, ata given time is found by dividing x, the slope of thegrowth curve, by X. If a continuous fermentation isoperated at this growth rate (determined by thethroughput rate), the equilibrium levels of X and Swill be the same as those in the batch fermentation.That is, the conditions in a continuous fermentationand those in a batch fermentation are the same whenthe growth rates are equal. After a period of continuousoperation on certain media, it may be that a differentsubstrate will become limiting, nitrogen rather thanglucose, for example. In this case, the concentrationsX, P, and S will change from the values predicted byanalysis of the batch fermentation.

It is often advantageous in a continuous product-

formationi fermentatioin to operate with several vesselsin series. When product formation is associated withgrowth, the advaiitage of this type of operation is ob-tainied under the same conditions as for microbialpropagation. That is, the advantage exists only whenhigher growth rates can be maintained in the first tanksin the series, while the lowered growth rates in the finaltanks permit a more complete use of the substrate. Sucha situation is apt to prevail in alcohol fermentation, forexample. In the anaerobic yeast fermentation, it appearsthat growth rates near the maximum require high con-centrations of limiting substrate and low concentrationsof the toxic product, ethanol. Thus, in a single vessel,the high throughput rates necessary for maximum pro-ductivity are not compatible with the low effluent sub-strate and high product concentration also necessary.This reasoning explains the wide acceptance of multi-vessel systems for continuous alcoholic fermentation.As in microbial propagation, the continuous fermenta-

tion for product formation may be increased in produc-tivity by raising the concentration of nutrient in thefeed, So. This increases the concentration of both cellsand product. Equations 17 and 18 demonstrate theeffect on p. This procedure has limitations, since it isknown that X is a function of P. That is, high productconcentrations inhibit the product-forming enzymes.They may also inhibit the enzymes of growth. Thus,the point is reached where increasing So will decreaseYP while S rises. Rate of product formation will rise nofurther.An effect similar to increasing substrate concentra-

tion can be achieved by recycling the organism. Thishas the effect of increasing X without affecting X (equa-tion 18). In this manner, the productivity is enhanced.A higher throughput rate is possible while Yp and S re-main the same (equation 17). The limit is reached whenproduct inhibition of A and X causes X and Yp to fall,and S to rise significantly.

It is possible, of course, to return all the cells to thefermentation, that is, total recycling. If this is done, theconcentration of cells in the fermenting medium willcontinue to rise until growth ceases completely (orrather growth rate and autolysis rate are equal) as aresult of product toxicity, lack of nutrients, or otherundefined factors. In such a case, the throughput ratemay be as high as desired without decreasing X.Productivity will still, however, depend on throughputrate, as indicated by equations 17 and 18.

Product formation not associated with growth. Inthe above paragraphs, the case in which the same sub-strate is limiting both to growth rate and productformation rate has been under consideration, and bothrates reacted similarly to the concentration of this sub-strate. This is often not the situation. For example, ifan aerobic yeast fermentation is fed a medium low innitrogen, the growth rate may be nitrogen-limited while

118

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CONTINUOUS FERMENTATION

glycolysis proceeds at such a rate that it is limited bythe concentration of glucose. More important examplesoccur in the formation of other fermentation products,enzymes, antibiotics, and pigments, for example, whichare less directly connected with the growth process. Forillustration, in the batch penicillin fermentation, thereare two distinct phases. For the first 24 hours growthoccurs on the rapidly assimilable carbohydrate sources.The pH is low. The remainder of the fermentation evi-dences little growth. The pH is high. It is in this periodthat penicillin production rate reaches its maximum.Thus, the conditions required for growth and thoserequired for productivity are widely different.When this is the situation, a single-vessel continuous

process without recycling can only poorly provide theproper environment both for growth and productformation. High throughput rate, which is necessary forhigh productivity (equation 17), necessitates highgrowth rate (since ,u = r). The equilibrium conditionsthat permit high growth rate will, on the other hand,be incompatible with a satisfactorily high rate of prod-uct formation. The two metabolic processes must beseparated for the best results.

There have been few reported attempts to accomplishthis separation of functions in a continuous system.The theory predicts that it can be done by recycling theorganism, by use of a multivessel system, or by use of atwo-phase system.

Two-phase systems. In our previous considerations,the fermenting mixture consisted of a more or lesshomogeneous suspension of microorganisms in a suitablemedium. The product formed was separated from thecells in a separate stage of processing by filtration orcentrifugation. This single-phase type of fermentationis usually the most convenient. In certain cases, how-ever, it is possible to separate the system into twophases. The major example is the vinegar or acetic acidfermentation. Here the fermentor or "generator" con-sists of a mass of beech wood shavings upon which theAcetobacter species which is responsible for the desiredoxidative conversion of ethanol to acetic acid is fixed.In this way the organism is maintained in the solidphase while the substrate and product remain in theliquid stream which trickles through.There are further examples of two-phase continuous

fermentations (see Applications). Usually the phasesare solid and liquid. There are cases where the organismis in a liquid phase while the substrate is in a second,immiscible liquid or a gaseous phase.A two-phase system does not comply with the cri-

terion of homogeneity which was important to thetheoretical relations developed earlier. Fermentativeconversion does not occur in one or more equilibriumstages, but in a continuum. Thus, the concentration ofacetic acid in a vinegar generator rises gradually to itsmaximum in the effluent. Such a situation is not repre-

sented by the equations already derived. Similar princi-ples may be used, however, to evolve the relationshipsfor various processes of this sort.

Applications. Many continuous fermentation proc-esses for product formation have been successfullyoperated. The majority of those reported in the litera-ture have been for the anaerobic production of ethanolby yeast. While the objectives are similar, the mannerin which these fermentations are run is widely varied.The differences occur in the microorganism, themedium, the conditions, and in other respects. Examplesare available of every type of continuous operati6n thathas been described here. Included are single-vesselsystems with and without recycling, multivessel systemswith from 2 to 12 vessels in series, two-phase systemsand others which display major and minor modificationsof these principles.

Bilford et al. (1942) investigated a single-vessel con-tinuous alcoholic fermentation process on a laboratoryscale. Their tests were run in a small fermentor agitatedeither mechanically or with carbon dioxide. The nutri-ent media were glucose (10 to 12 per cent) in yeast wateror molasses (Cuban blackstrap, refined and beet, 12to 13 per cent reducing sugar), and these could be fer-mented to completion with holdup times of 4 to 7 hours.The effluent reducing sugar concentration ranged from0.1 to 1.5 per cent. The Guilaume-Boulanger process(DeBeeze and Rosenblatt, 1943) used in Europe in thelarge beet sugar and molasses alcohol plants is a modi-fied single-vessel type. The main conversion is carriedout in a large continuously operated fermentor. In orderto allow complete usage of the sugar a 212 hour batchfermentation of the effluent is carried on in small auxili-ary tanks.

Molasses provides a readily used source of carbohy-drate and contains no substances inhibitory to thealcoholic fermentation. It is, therefore, rapidly and com-pletely used in continuous operation. Multivessel opera-tion is undesirable, and recycling is not necessary. Thelow pH and high throughput rate minimize the chanceof contamination.On certain other substrates, advantage must be taken

of the multivessel and recycling principles to permithigh, continuous productivity.The process developed for the conversion of acid-

hydrolyzed grain carbohydrates to alcohol by Joseph E.Seagram and Sons, Inc. (Altsheler et al., 1947; Ruf etal., 1948) involves the use of two vessels. In the pilotoperations described by Ruf et al. (1948), the holdingtime is 8 hours in the primary fermentor, 4 to 6 hoursin the secondary. Both tanks are mechanically agitated.Fermentations up to 9 days in pilot plant and 6 monthsin laboratory were obtained without contamination oryeast degeneration. Figure 2 shows some analyses madeduring the course of a pilot-scale run. In the two runscompared, corn mash alone was used in the first, and

119

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

WILLIAM D. MAXON

YEAST 300-6-J 1OCELL cotrs 0a^ID

PRRYNFERMENTER - -o

REOIJCING ,'*MATERIAL 4 a

AS0

GLUCOSE 3G / OOML RY F1ER

2~~~~~~~~~

YfIELD

SECONDARYFERMENTER 0 0 ._

440 - - M.RM INo SEWLEMENT

42_____ a - MM2-SUPPLEMENT AT 20 MOMUS I 6 24 32- 40 40 M 04

TIM E - HOURS

FIG. 2. A two-vessel continuous alcoholic fermentation ofacid-hydrolyzed corn mash. Comparison of data obtained withand without submerged culture mold supplement. (From Ruf,E. W., Stark, W. H., Smith, L. A. and Allen, E. E. AlcoholicFermentation of Acid-hydrolyzed Grain Mashes, Ind. Eng.Chem., 40, 1154-1158, Figure 4, p. 1157.)

in the second submerged culture Aspergillus oryzaewas added as a supplement. The supplementation evi-dently changed the limiting nutrient. This permitted aslightly higher cell population in the primary fermentor,a lower concentration of reducing sugar in both fer-mentors, and a higher alcohol concentration in the sec-ondary fermentor. This example demonstrates the ad-vantages of continuous operation for experimentation.The method of Alzola (1942, 1945) for continuous

fermentation also uses the multivessel principle. Here,five fermentors of equal size are connected in series.Agitation in the fourth and fifth is supplied by carbondioxide pumped from the first and second. The concen-tration of nutrient in each tank, as measured by specificgravity, declines as follows: Second tank, 70 B6; thirdtank, 40 B6; fourth tank, 2.50 Be; fifth tank, 20 B6.The additional advantages of recycling are made use

of in the alcoholic fermentation of wood hydrolysatesand of sulfite waste liquor, substrates less readily usedby the yeast. For a general reference see McCarthy(1954) and Saeman and Andreasen (1954). Multivesselinstallations are used, and the effluent yeast is sepa-rated by centrifugation as a cream to be added with thefeed to the first fermentor in the series. This type ofprocess is quite widely used in Europe (Saeman andAndreasen, 1954; Saeman, Locke and Dickerman, 1945,1946; DeBeeze and Rosenblatt, 1943) and plants areoperated by the Puget Sound Pulp and Timber Com-pany, Bellingham, Washington (Ericsson, 1947; Mc-Carthy, 1954) and by Commercial Alcohols Limited,Gatineau, Quebec (McCarthy, 1954). The fermentorsare large, up to 120,000 gallons capacity at Gatineau,and are agitated either mechanically or with earbondioxide. From 3 to 8 tanks are arranged in series. Thetotal holdup time varies from 6 to 24 hours. Up to 12per cent of the fermenting medium is yeast (pressed).

Pilot studies in a six-fermentor system of this typehave been made by Harris et al. (1948c). Torula utiliswas used to ferment wood hydrolysate containing 5per cent reducing sugar. At steady state with a total of24 hours retention time 80-82 per cent of the fer-mentable sugar was used, and an alcohol yield of 40per cent on total fermentable sugar was obtained. Fromtank to tank the equilibrium sugar concentration was:5.2 per cent (feed), 4.3 per cent (first tank), 3.3 per cent(second tank), 2.3 per cent (third tank), 1.5 per cent(fourth tank), 1.1 per cent (fifth tank), and 0.8 per cent(sixth tank).The patent literature details some special equipment

for fermentations of this sort. Victerero (1948) arrangesthe equilibrium stages vertically in order to facilitateflow of gas and medium from tank to tank. Scholler andco-workers (1937, 1940) describe a special fermentationapparatus that embodies the principles of a multivesselcontinuous system and is provided with a final set-tling chamber where the yeast separates for return withthe feed.Two-phase systems are not commonly used in the

alcoholic fermentation. The "Fesselhefe" method (Sae-man, Locke and Dickerman, 1945), in which the yeastis bounid on twigs, has been used in Germany, however,for a sulfite waste liquor process. This type of operationhas been patented by Romer (1924). An Indian patent(Council of Scientific and Industrial Research, 1952)has appeared on the use of a battery of columns con-taining pumice impregnated with yeast. Molasses wasthe carbohydrate source, and less than 0.5 per centsugar remained in the effluent.

There are relatively few instances of continuousfermentation being used for the manufacture of prod-ucts other than ethyl alcohol. There is, of course, thecase of acetic acid and vinegar, as mentioned before. Areview of this process has been made by Vaughn (1954).In 1931, Whittier and Rogers developed a laboratory-,scale single-vessel continuous fermentation for the con-version of whey to lactic acid.

Interesting applications of continuous fermentationin the petroleum industry have been patented by theStandard Oil Development Company (Taggart, 1946)and the Texaco Development Company (Zobell, 1953).The Standard Oil process is a two-phase fermentationcarried out in a bubble-cap column. Light, gaseous hy-drocarbons are continuously converted to the cor-responding fatty acids and their esters through theoxidative action of such organisms as Bacillus paraf-finicus n. sp., B. methanicus, and B. ethanicus n. sp. Thebacteria in a salt solution pass downward over the trays.The gaseous hydrocarbons mixed with air pass upwardthrough the bubble-caps. The contact that results per-mits oxidation of the substrate and proliferation of thebacteria. The product is recovered from the aqueousstream. The process is one of growth-associated productformation. It involves several equilibrium stages and

120

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CONTINUOUS FERMENTATION

may, therefore, be classified as a multivessel system, butit is unique in that the substrate stream and the bac-terial catalyst stream are in separate phases flowingcountercurrently.The Texaco process has some similarities. Here, how-

ever, the reaction is reductive. The complex sulfur com-pounds that occur in crude petroleum are often unde-sirable in the final product. By this method they areconverted to easily removed mercaptans and hydrogensulfide. An organism such as Desulfovibrio desulfuricansis employed. It can use hydrogen for the reduction ofsulfur compounds by virtue of the hydrogenase enzymethat it produces. For continuous operation a bubble-cap column is used. The crude hydrocarbon flows up-ward and the aqueous medium containing the bacteriaflows downward. Hydrogen is supplied at the bottomof the column and flows out the top with the gaseousreaction products. Alternatively, the hydrogen may besupplied in situ by the action of microorganisms, forexample, Clostridia on a glucose medium. This process,like the Standard Oil process, is a two-phase multistagesystem. Product formation is growth-associated.A few attempts have been made to use continuous

fermentation for the production of antibiotics. Theprocess of the Joseph E. Seagram and Sons., Inc.(Kolachov and Schneider, 1952) is a notable example.A single aerated and agitated fermentor is employed.It is continuously fed a lactose, glucose, corn steepliquor medium, and the penicillin-containing, fermentedmedium is continuously withdrawn. A holdup time of48 hours is maintained. More rapid throughput wouldnecessitate a high growth rate, incompatible with peni-cillin production. The yield of 600 units/ml gives a pro-ductive capacity of 12.5 units/ml hour.

Other antibiotic fermentations of a continuous naturehave been reported. Moor (1945) describes a labora-tory-scale two-phase penicillin surface culture process.A matt of Penicillium notatum is grown on the surfaceof a corn steep medium. Fresh medium, fermentedproduct, and the mycelial matt move continuouslythrough the apparatus. A similar process has been pat-ented (Lilly, 1933). Here, the mycelium grows on theoutside of filter cloth tubes where it is aerated. Themedium to be fermented passes down inside these tubes.Japanese workers (Wakaki et al., 1952) described a

semicontinuous submerged culture process for penicillinfermentation.The acceptance of continuous methods for the manu-

facture of food and feed yeast and fermentation alcoholis easily understood. These industries depend upon a

large volume and a low profit margin. Thus, equipmentof tremendous capacity operating at maximum outputis required. High efficiency and low waste are necessi-ties. The advantages gained in productivity for a givenequipment capacity by continuous fermentation are

great. Furthermore, the time lost in cleanup and prepa-ration necessary to batch fermentation is obviated. The

difficulties that could arise from contamination or yeastdegeneration are not encountered, and no expensiveprecautions are required to prevent them. For thesereasons the continuous process may be expected to haveever wider application in these industries.As the technology of fermentation develops, it seems

inevitable that manufacturers will turn to this type ofoperation for other products. The advantages to begained, whether increased output, automatic control,or product uniformity, all become available when thedifficulties of contamination, genetic instability, and soforth are solved. Meanwhile, continuous fermentationson a laboratory scale will have great value in the col-lection of experimental data. This discussion was in-tended to summarize the information on the subjectpresently available and to provide a background for newdevelopments and a starting point for new ideas.

REFERENCES

ADAMS, S. L., AND HUNGATE, R. E. 1950 Continuous fer-mentation cycle times. Ind. Eng. Chem., 42, 1815-1818.

ALTSHELER, W. B., MOLLETT, H. W., BROWN, E. H. C., STARK,W. H., AND SMITH, W. H. 1947 Design of a two-bushelper day continuous alcohol unit. Chem. Eng. Progr., 43,467-472.

ALZOLA, F. 1942 New process of continuous fermentation.Intern. Sugar J., 274.

ALZOLA, F. 1945 System of continuous fermentation. U. S.Patent 2,371,208.

BILFORD, H. R., SCALF, R. E., STARK, W. H., AND KOLACHOV,P. J. 1942 Ind. Eng. Chem., 34, 1406-1410.

BUHRIG, W. H. F. 1929 Process for the manufacture ofyeast. U. S. Patent 1,730,876.

COUNCIL OF SCIENTIFIC AND INDUSTRIAL RESEARCH. 1952Fermentation process. Indian Patent 43,542.

DARANYI, S. C. 1936 Manufacture of yeast. U. S. Patent2,035,048.

DEBECZE, G., AND ROSENBLATT, M. 1943 Continuous fer-mentation. Am. Brewer, 76, 11-16, 30, 32, 34.

ELSWORTH, R., AND MEAKIN, L. R. P. 1954 Laboratory andpilot plant equipment for the continuous culture of bac-teria with examples of its use. Chemistry & Industry,July 24, 926-927.

ERICSSON, E. 0. 1947 Alcohol from sulfite waste liquor.Chem. Eng. Progr., 43, 165-167.

FINN, R. K., AND WILSON, R. E. 1953 Population dynamicsof a continuous propagator for microorganisms. J. Agr.Food Chem., 1, 66-69.

GERHARDT, P. 1946 Brucella suis in aerated broth culture:Continuous culture studies. J. Bacteriol., 52, 283-292.

GOLLE, H. A. 1953 Theoretical considerations of a con-tinuous culture system. J. Agr. Food Chem., 1, 789-793.

HARRIS, E. E., HANNAN, M. L., MARQUARDT, R. R., AND BUBL,J. L. 1948a Fermentation of wood hydrolysates byTorula utilis. Ind. Eng. Chem., 40, 1216-1220.

HARRIS, E. E., SAEMAN, J. F., MARQUARDT, R. R., HANNAN,H. L., AND ROGERS, S. C. 1948b Fodder yeast fromwood hydrolysates and still residues. Ind. Eng. Chem.,40, 1220-1223.

HARRIS, E. E., HANNAN, M. L., AND MARQUARDT, R. R. 1948cProduction of food yeast from wood hydrolysates. Ind.Eng. Chem., 40, 2068-2072.

HARRISON, A. P. 1930 Process for the manufacture of yeast.U. S. Patent 1.761,789.

121

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

WILLIAM D. MAXON

HOLDERBY, J. M. 1946 Waldhof process for production offood yeast. FIAT Final Report No. 619.

INSKEEP, G. C., WILEY, A. J., HOLDERBY, J. M., AND HUGHES,L. P. 1951 Food yeast from sulfite liquor. Ind. Eng.Chem., 45, 1702-1711.

IRVIN, R. 1954 Commercial yeast manufacture, in IndustrialFermentations, ed. by UNDERKOFLER, L. A., AND HICKEY,R. J., Chemical Publishing Co., Inc., New York.

KOLACHOV, P. J., AND SCHNEIDER, W. C. 1952 Continuousprocess for penicillin production. U. S. Patent 2,609,327.

LILLY, C. H. 1933 Process of mold fermentation and ap-paratus therefor. U. S. Patent 1,936,983.

MAXON, W. D., AND JOHNSON, M. J. 1952 The effect ofaeration on the carbon metabolism of Saccharomycescerevisiae. 121st Meeting, Am. Chem. Soc., Milwaukee, Wis.

MAXON, W. D., AND JOHNSON, M. J. 1953 Aeration studies onpropagation of bakers' yeast. Ind. Eng. Chem., 45, 2554-2560.

AICCARTHY, J. L. 1954 Alcoholic fermentation of sulfitewaste liquor, in Industrial Fermentations, ed. by UNDER-KOFLER, L. A., AND HICKEY, R. J. Chemical PublishingCo., New York.

MIICHAELIS, L., AND MENTEN, Ml. L. 1913 Die Kinetik derInvertinwirkung. Biochem., Z., 49, 333-369.

MONOD, J. 1950 La technique de culture continue, theorieet applications. Ann. Inst. Pasteur, 79, 390-410.

MNOOR, W. A. 1945 The continuous cultivation of micro-organisms. Science, 102, 594-595.

NOVICK, A., AND SZILARD, L. 1950 Experiments with theChemostat on spontaneous mutations of bacteria. Proc.Natl. Acad. Sci., 36, 708-719.

NOVICK, A., AND SZILARD, L. 1951 Genetic mechanisms inbacteria and bacterial viruses. Experiments on spontane-ous and chemically induced mutations of bacteria growingin the Chemostat. Cold Spring Harbor Symposia Quant.Biol., 16, 337-343.

OL,SEN, M. A. J. C., AND INTERNATIONAL YEAST CO., LTD. 1930Perfectionnements relatifs aux procedes de preparation delevure. French Patent 676,835.

POWELL, E. 0. 1954 Populations of bacteria: The averageand the individual. Chemistry & Industry, July 24, 928.

RAMAGE, W. D., AND THOMPSON, J. H. 1948 Producing yeastfrom processing wastes, in Yeasts in Feeding, Proceedingsof a Symposium, ed. by BRENNER, S., Jos. Schlitz BrewingCo. et al. Milwaukee, Wis.

RUF, R. W., STARK, W. H., SMITH, L. A., AND ALLEN, E. E.1948 Alcoholic fermentation of acid-hydrolyzed grainmashes. Ind. Eng. Chem., 40, 1154-1158.

ROMER, A. 1924 Fermentation process. U. S. Patent1,510,195.

SAEMAN, J. F., AND ANDREASEN, A. A. 1954 The productionof alcohol from wood waste, in Industrial Fermentations,ed. by UNDERKOFLER, L. A., AND HICKEY, R. J. ChemicalPublishing Co., Inc., New York.

SAEMAN, J. F., LOCKE, E. G., AND DICKERMAN, G. K. 1945The production of wood sugar in Germany and its con-version to yeast and alcohol. FIAT Final Report No. 499.

SAEMAN, J. F., LOCKE, E. G., AND DICKERMAN, G. K. 1946The production of wood sugar in Germany and its con-version to yeast and alcohol. Paper Trade J., 123, 38-46.

SAK, S. 1928 Improved process of producing yeast. Brit.Patent 294,134.

SAK, S. 1932a Yeast propagation controlling method. U. S.Patent 1,891,841.

SAK, S. 1932b Improvements in the manufacture of yeast.Australian Patent 8021/32.

SCHOLLER, H., AND EICKENMEYER, A. 1937 Method of fer-menting solutions and producing yeast. U. S. Patent2,083,347.

SCHOLLER, H., AND SEIDEL, M. 1940 Method of and ap-paratus for fermenting solutions. U. S. Patent 2,188,192.

SEIDEL, MI. 1943 Proced6 de preparation de levures et autresmicroorganismes. French Patent 878,985.

TAGGART, M. S., JR. 1946 Utilization of hydrocarbons. U. S.Patent 2,396,900.

UNGER, E. D., STARK, W. H., SCALF, R. E., AND KOLACHOV,P. J. 1942 Continuous aerobic process for distiller'sveast. Ind. Eng. Chem., 34, 1402-1405.

VAUGHN, R. H. 1954 Acetic acid-vinegar, in Industrial Fer-mentations, ed. by UNDERKOFLER, L. A., AND HICKEY,R. J. Chemical Publishing Co., Inc., New York.

VICTORERO, F. A. 1948 Apparatus for continuous fermenta-tion. U. S. Patent 2,450,218.

WAKAKI, S., ISHIDA, T., YOSHINOBU, Y., MIZUHARA, K.,MASUDA, K., AND YOSHIDA, K. 1952 Continuous fer-mentation of penicillin. J. Antibiotics (Japan), 5, 622-630.

WHITTIER, E. O., AND ROGERS, L. A. 1931 Continuous fer-mentation in the production of lactic acid. Ind. Eng.Chem., 23, 532-534.

WILEY, A. J. 1954 Food and feed yeast, in Industrial Fer-mentations, ed. by UNDERKOFLER, L. A., AND HICKEY,R. J. Chemical Publishing Co., Inc., New York.

ZOBELL, C. E. 1953 Process for removing sulfur from petro-leum hydrocarbons and apparatus. U. S. Patent2,641],564.

I 22

on July 5, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from