Ethanol Volatility in Fermentation Systems. Salt ... - ACS Publications

remove dead cells and regulate the extent of buildup of components to .... 0. 0.2. 0.4. 0.6. 0.8. 1.0. Mole Frocfion Ethanol. (Solt-free basis). Figur...
1 downloads 0 Views 702KB Size
332

Ind. Eng. Chem. Fundam.

1984,23, 332-337

Ethanol Volatility in Fermentation Systems. Salt and Metabolite Effects Brlan L. Malorella, Charles R. Wllke, and Harvey W. Blanch" Lawrence Berkeley Laboratoty and Department of Chemical Engineering, Universiv of California, Berkeley, California 94720

The effect of dissolved species on the relative volatility of ethanol in fermentation systems is evaluated. New data are presented showing that the enhancement in volatility due to dissolved salts varies with ethanol concentration and is largest for the dilute ethanol concentrationstypical in fermentation broths. Dissolved sugars and metabolites also affect relative volatility. The most commonly applied model of volatility enhancement does not incorporate the effect of ethanol concentration, and published enhancement factors measured at high salt and ethanol concentration are not applicable to the conditions of a fermentation broth.

Introduction Volatility Enhancement in Fermentation Systems. During wine production, ethanol is stripped from the wine by C02 evolved in the fermentation. In industrial ethanol production, ethanol is initially concentrated in a stripping still. In new fermentation processes such as vacuum fermentation (Maiorella et al., 1981; Cysewski and Wilke, 1977) or extractive fermentation (Wang et al., 1981; Finn, 1966), ethanol is removed from the fermenting broth (by flashing under vacuum or by extraction, respectively) continuously as it is produced. In all of these processes, the ethanol is removed from a solution containing dissolved species and the effect of these species on the ethanollwater equilibrium is not well understood. Proposed new fermentation processes will concentrate the dissolved species and, for these processes, the effect of the dissolved components may become important. In vacuum fermentation, the feed solution is continuously added to a fermentor under vacuum. Fermentation takes place in the broth, and ethanol is boiled away as it is produced (to eliminate end product inhibition). Nonvolatile feed components which are not metabolized build up in the fermentor. A small liquid bleed is taken to remove dead cells and regulate the extent of buildup of components to levels below where they would inhibit the fermentation reaction. It is desirable, however, to maintain the bleed as small as possible to take most of the product in the concentrated purified vapor form. Likewise, in conventional fermentation, to reduce the liquid waste disposal load, stillage is now recycled (with added substrate) to the fermentor, again concentrating nonvolatile dissolved species (Wall et al., 1981; Ronkainen, 1978). Similar concentration effects occur in extractive and membrane extractive fermentation processes. In one proposed extraction process, a salt (KC1) is added to the fermentation broth to enhance the ethanolfwater extractive separation (Murphy et al., 1982). The potentially large effect of dissolved species on ethanol volatility is well illustrated by the use of added salts in extractive distillation to produce pure ethanol. The azeotrope may be broken by the addition of 6 mol % potassium acetate (Cook and Furter, 1968). The HIAG process, using a eutectic mixture of potassium and sodium acetate at saturation, was utilized in over one hundred plants between 1930 and 1950, producing anhydrous ethanol from 94 w t % feed (Gorham, 1933; Webb, 1964). The effects of salts a t saturation and the application of salts in extractive ethanol distillation have been comprehen-

Table I. Inhibitory Levels of Salts on Ethanolic Fermentation concn, M at 20% a t 80% inhibition" inhibition" NaCl 0.24 0.45 KCl 0.60 1.74 KHzPO4 0.66 0.73 CaCIZ 0.08 0.23 MgClz 0.58 0.92 MgS04 0.63 0.97 NH4Cl 0.12 0.46 (NH,)zSO4 0.11 0.24 "Indicated 20% and 80% reductions of cell mass productivity in continuous fermentation.

sively reviewed elsewhere (Furter, 1977; Maiorella et al., 1984a). This paper considers the effects of salts at the lower concentrations which may be found in fermentation broth. Table I summarizes levels at which various salts (present in most feeds) becomes inhibitory to yeasts (Maiorella et al., 198413). These inhibitory levels set the limits for operation of the fermentation and are the maximum concentrations that need to be considered in assessing effects of salts on volatility enhancement in ethanol fermentation systems. Volatility Enhancement Model A relation derived by incorporating electrostatic and induced dipole effects caused by added charge species is most commonly used to model the salt effect in solution. The activity coefficient of the nonaqueous solvent in a salt solution is represented by a power series in the concentrations of the salt and nonaqueous solvent (Long et al., 1952; Kirkwood, 1943) In yi = C kl,,XimX,' l,m=O

(1)

where Xi and X,are mole fractions of the nonaqueous solvent and of the salt, respectively. When the interaction constants (kl,J are calculated from pure component properties using electrostatic principles (Kirkwood, 1943), the model fails to predict salting phenomena quantitatively. If, however, the constants are fitted to experimental equilibrium data, then a valuable interpolative model results. Xi is then generally taken as the nonaqueous solvent concentration on a salt free basis

0196-4313/84/1023-0332$01.50/00 1984 American Chemlcal Society

Ind. Eng. Chem. Fundam., Vol. 23, No. 3, 1984

Table 11. Enhancement Factors for Salts at Saturation” salt enhancement factor 8.3 ammonium chloride 11.1 sodium chloride 9.7 potassium chloride 6.9 sodium nitrate lead nitrate 8.1 mercuric chloride -1.4 mercuric bromide -2.1 mercuric iodide 0.2 barium nitrate 4.6 potassium sulfate 2.3 ammonium sulfate 8.1 cuprous chloride 5.5

361

I

I

I

I

I

333

Results of Johnston and Furter (1960).

and the fitted constants reflect (in a complex way) all the various force effects of the salt on the components of the solution. At high dilution, eq 1 reduces to In Ti = ~ O , O+ ko,lXi + kl,&s

(2)

Assuming a similar expression for the effect of the salt on water, a very simple relation for the enhancement of nonaqueous solvent volatility by a salt can be derived (Johnson and Furter, 1960)

K, . In - = (klo,l - kw0,,)X,

KO

I

0

0.2

I

I

I

I

0.4

0.6

0.8

1.0

Mole Frocfion Ethanol (Solt-free b a s i s )

Figure 1. Effect of sodium acetate on ethanol/water relative volatility.

I

I

S l o p e = k : 5.3

(3)

where kio,l and kwo,lare constanta as in eq 2 but taken from the calculation of yi and y, (activity coefficients for the nonaqueous solvent and water), respectively. K , and KO are solvent relative volatilities with and without salt

0.2

0

(4) with Xi and X, given on a salt free basis; kio,l is a measure of the effect of the salt on the nonaqueous solvent; kWO,l is a measure of the effect of the salt on the water component of the solution. The difference between these factors is referred to simply as the enhancement factor, k, and is a measure of the effect of the salt on the relative volatilities of the solvents.

K,

In - = kX,

KO

This simple relation is a limiting law (for low salt and fixed nonaqueous solvent concentration only). Further, it should be valid only when the change in boiling point due to the salt is small, so that the activity coefficients computed with and without salt are at the same temperature-again this is true for low salt concentration. Maiorella et al. (1984a) summarized 44 reported experimental evaluations of the salt effect for the system ethanol/water. The data are actually quite limited in that almost all are for salts at saturation, and data are rarely available for ethanol concentrations below 10 mol % (salt free basis). Johnson and Furter (1960) applied eq 5 to the ethanol/water system with 12 different saltsat saturation. The values of k (Table 11)for the systems tested were all remarkably constant and a single value of k for each system was able to predict the equilibrium curve duplicating the data to within approximately 1 mol % throughout. There is no apparent underlying basis for the success of eq 5 (which was derived only as a limiting law for the case of

002

004

006

008

010

012

Mole Fraction Ammonium Chloride

Figure 2. Effect of ammonium chloride on ethanol/water relative volatility. Results of Jaques and Furter (1974).

dilute salts) in modeling the behavior of the wide range of systemsof Table I1 under the more complex conditions of salts at saturation. The derivation of eq 5 from the general form of eq 1suggests that a more complex relation should be required. While eq 5 has proved successful in correlating the effects of a wide range of simple salts in the ethanol/water system at saturation, it does fail for many important cases. For the system ethanol/water/potassium acetate at saturation, Meranda and Furter (1966) found large enhancements (with the azeotrope completely eliminated), but the the value of k varied by a factor of 3 as ethanol concentration was decreased from X, = 0.9 to X, = 0.1. For the system ethanol/water/sodium acetate at saturation, a large enhancement was also found, but the values of k were approximately constant with varying ethanol concentration (Meranda and Furter, 1971). Bedrossian and Cheh (1974) have applied eq 5 to the sodium acetate system under conditions of varying levels of salt at low concentrations (0.75 to 3.0 g/L). Under these conditions (unlike the behavior at salt saturation) the value of k was found to vary significantly with ethanol concentration as shown in Figure 1. As the ethanol concentration is decreased, k initially slowly decreases, but at low ethanol concentrations (similar to those found in a fermenting broth) the value of k rises by a factor of 5. Recent work by Jaques and Furter (1974) raises further questions as to the broad applicability of eq 5. In this work, eq 5 was tested under the conditions for which it was derived-constant nonaqueous solvent concentration and

334

:e

Ind. Eng. Chem. Fundam., Vol. 23, No. 3, 1984

Table 111. Enhancement Factor for Dilute Salt Solutions and Solutions at Salt Saturation enhancement factor, k at saturation dilute salt salt 11.1 8.3 sodium chloride potassium chloride 9.7 7.8 ammonium chloride 8.3 5.3 sodium bromide 6.7

varying low salt concentration. For both ethanol and methanol systems, the equation worked very well giving nearly constant k values (Figure 2). However, the values of k found under these conditions of low salt concentration (approximately 0.02 to 0.12 mole fraction) were consistently lower than those found for the salts at saturation (Table 111). Very little information is available on the effect of mixed salts on ethanol/water relative volatility. For many no-2polar systems at low salt and nonaqueous solvent concentrations, the effect of multiple salts is approximately additive (Long and McDevit, 1952; Morrison, 1955) so that

K,

In - = E k j X ,

KO

OTHMER EOUlLlBRlUM STILL

Refrigeroted Total Condenser

Product Receiver

U

Therm-syph on Heater

Figure 3. Othmer equilibrium still used in vapor/liquid equilibrium studies.

j=1

where X j and kj are the mole fraction and enhancement factor for the j t h salt component added. This additivity of effects has also been reported for a limited number of polar nonaqueous solvent systems (Larsson, 1931; Gross, 1933). For the ethanol/water system the only data are for mixed halide salts at saturation (Meranda and Furter, 1972). At low ethanol concentration (