Recovery of Potassium Magnesium Sulfate Double Salt from

Seawater Bittern

Jose A. Fernandez Lozano

Universidad de Oriente, Nucleo de Anzoategui, Puerto La Cruz, Venezuela

Seawater bittern from the production of halite contains about 2.6 % potassium sulfate and 4.3% magnesium SUI-
fate, the recovery of which is made difficult by the presence of other salts. A novel method for the recovery of
potassium sulfate and magnesium sulfate as potassium magnesium sulfate double salt (PMS) was developed.
This method is based on the solubility characteristics of these salts in aqueous methanol solutions. It was found
that selection of the appropriate mole fraction of the organic solvent, concentration of bittern, salting time, and
temperature permits the recovery of salt containing up to 23% K 2 0 and 13% MgO, making it a very desirable
direct use fertilizer. Laboratory and I-ton-per-day pilot plant studies have demonstrated the chemical and me-
chanical feasibility of the process. Under optimum conditions of operation the recovery efficiency for potassium
sulfate is about 98% with only 0.21 % methanol losses.

Introduction magnesium sulfate slurry; (d) filtering the thickened slurry

Seawater bittern is an almost unlimited reserve of several and washing the crystals; (e) recovering the methanol from
potential chemicals which, at the present time, are attracting the filtrate by distillation; and (f) drying the solid product and
much attention. From the standpoint of an ore reserve, sea- recovering the methanol vapors. These two objectives have
water contains a low grade ore which necessarily has to be been met by the investigation summarized in this paper. The
concentrated by evaporation of the water present in it. Sea- process outlined in the objectives was found to work effi-
water bittern is obtained as a by-product of two evaporation ciently.
processes: the solar salt industry and saline water conversion Experimental Crystallizer
plants, but with few exceptions these rich ores are wasted.
Previous works related to the recovery of chemicals from The experimental crystallizer is schematically shown in
seawater bittern have been predominantly in the development Figure 1 and was designed to accomplish rapid mixing of
of conventional processes of evaporation-crystallization, ion measured amounts of two liquid phases and rapid separation
exchange, and solvent extraction, and these have been widely of the resulting crystalline precipitate formed. Provision was
reviewed in the literature. This paper is devoted to new made for setting and maintaining a constant temperature, for
technology for the recovery of potassium, magnesium, and measuring the time of reaction, and agitation.
sulfate as potassium magnesium sulfate double salt (PMS), The experimental procedure for the study of the equilib-
which promises a practical solution for the exploitation of rium solubility and crystallization kinetics involved mea-
seawater bittern. surement of the mass of crystals formed as a function of
The separation of water-soluble salts has frequently been methanol concentration and reaction time under controlled
achieved by introducing to the system an organic solvent in conditions of temperature, agitation, and reactants concen-
which one of the salts is insoluble and thus effects selective tration. Each experiment involved a batch crystallization
separation of that component. Based on this principle, some which gave one point on the curve. Repeated experiments with
interesting separation methods for the recovery of salts from different concentrations of methanol and reaction time es-
wastes and other effluents have been proposed (Roller, 1946; tablished the integral functions for a given set of conditions.
Gee et al., 1947; Gaska, 1966,1967; Hoppe, 1968, and Gaska The crystallization was initiated by rapid mixing of meth-
and Goodenough, 1969). No prior studies of the use of meth- anol from the injector with the bittern in the stirred batch
'anol or ethanol for the recovery of potassium magnesium crystallizer. Air pressure was used for rapid injection of the
sulfate double salt from seawater bittern have been reported. methanol into the crystallizer. The crystals formed were
The potassium magnesium sulfate salt is a very useful fer- rapidly separated from the mother liquor by filtration,
tilizer; its importance in agriculture has been well established washed, dried, weighed, and analyzed. The apparatus and
(Walsh, 1972; Borskey, 1973).The emphasis of this paper has experimental considerations are important in this research,
been on the experimental techniques and approach necessary but for the sake of brevity only a limited description is offered
for the recovery of potassium magnesium sulfate double salt here.
of fertilizer quality.
Laboratory Work
Scope of This Investigation General Consideration. The composition of the seawater
The main objectives of the present investigation were: (1) bittern used in this work assuming a hypothetical combination
to devise and to demonstrate in the laboratory and pilot plant of the main ions is shown in Table I. The solubility in aqueous
a simple crystallization process for the separation and recovery methanol of KC1, Na2S04, and other salts, presented in Table
of potassium magnesium sulfate double salt (PMS) from I, was determined and the results are shown in Figure 2. The
seawater bittern; (2) to determine the optimum variables of solubility of NaC1, KC1, MgC12, and Na2S04 is significant only
the developed process shown in Figure 4, consisting of: (a) as contrasted with the solubility of KpSO4 and MgSO4.
solar concentration of bittern from a halite plant to remove It is apparent from Figure 2 that both KzSO4 and MgS04
halite and to raise the potassium concentration; (b) mixing are considerably less soluble in aqueous methanol solutions
the methanol with the concentrated bittern for the selective than the other salts. On this basis, if methanol is added to
crystallization of PMS salt; (c) thickening of the potassium seawater bittern both salts will be selectively precipitated,

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 3, 1976 445
Table I. Composition of Bittern Used in This Study as a Function of Bittern Density

Main components of Composition,96 wt

seawater bittern,
solar concentrated Sp. gr. 1.320 Sp. gr. 1.310 Sp. gr. 1.282 Sp. gr. 1.262 Sp. gr. 1.208 s p . gr. 1.107

NazCl 2.81 3.94 7.88 12.74 25.95 11.85

MgCL 23.74 ' 18.55 12.46 11.05 3.57 1.79
4.95 4.63 3.15 2.62 0.81 0.38
MgS04 3.51 4.95 7.25 4.34 1.51 0.71
+
H2O O.S." 64.99 67.93 69.26 69.25 68.26 85.27
O.S. = other salts. Sp. gr. = specific gravity.

12
I 'P (I -0
21

-ti-4

Figure 1. Schematic diagram of the crystallization system used in

this study: 1, crystallizer vent valve and charge port; 2, injector
vent valve and charge port; 3, filter; 4, crystallizer receiver; 5, crys-
tallizer; 6, injector; 7, injector charge valve; 8, crystallizer charge
valve; 9, water bath temperature controller; 10, crystallizer metal-
lic thermometer; 11, air pressure gauge; 12, air pressure regulator;
13, air pressure valve; 14, electrical control valve switch; 15, injec-
tor outlet valve; 16, crystallizer discharge valve; 17, crystallizer
METHANOL ,IN SOLUTION, WT %
pressurization valve; 18, crystallizer stirrer; 19, air bottle; 20, crys-
tallizer jacket; 21, injector jacket; 22, water circulating pump. Figure 2. Solubility of various salts in aqueous methanol solutions
a t 30 O C .

with potassium and magnesium ions competing for the sulfate

ion. Of both salts, K&04 is the most insoluble and this char- bittern a t 30 "C, for various methanol concentrations, con-
acteristic is important in the competition for the sulfate ion, sisted mainly of KzS04, MgS04, and water of hydration. The
as will be shown in this work. results are summarized in Table 11. It is clear from these data
The concentration of seawater was conducted in large rec- that, at low methanol concentration, most of the sodium
tangular solar evaporation ponds. The evaporation proceeded chloride and sodium sulfate remained in the liquid phase while
under natural conditions with the temperature of the bittern most of the potassium sulfate and magnesium sulfate were
varying between 26 and 45 "C. The bittern from the ponds was found in the solid phase, as hydrated potassium magnesium
conducted through plastic pipes to plastic lined storage tanks sulfate.
and stored for 2 h at 35 "C before feeding it to the crystallizer. The percentage of sodium chloride, sodium sulfate, and
The ionic composition of the liquid and solid phases was water of hydration in the solid phase rises with increasing
chemically evaluated by the procedure proposed by (Rafols, methanol concentration in solution while potassium sulfate
1969) and by spectrographic analysis. Tables 11, 111, and IV decreases. A point worth noting here is that at 40% methanol
summarize the composition of the different crystal crops ob- concentration the solid phase is almost pure monohydrate
tained. Table I1 also shows the mineral species in the solid potassium magnesium sulfate.
phases which were identified by their optical properties under At about 70% methanol concentration in solution, the sol-
a polarized microscope, and the results are confirmed by x-ray ubility of potassium sulfate and magnesium sulfate reverses
analysis. itself. At this point, the concentration of methanol is high
Seawater Bittern Test. The test was begun by charging enough to solvate the pPtassium and magnesium ions, the
11. of seawater bittern to the constant temperature crystallizer result being the increased solubility of the said salts. This
(Figure 1). A minimum of 30 min was allowed for thermal effect is more pronounced for potassium sulfate.
equilibrium to be established. The results of the solid phase identification presented in
The crystallization was started by adding the methanol to Table I1 show that hydrated potassium magnesium sulfate
the bittern in the crystallizer. The rate of addition was ad- and hydrated sodium magnesium sulfate double salts are the
justed to 500 ml/s, and the stirrer control and timer were main components of the different crystal crops recovered. The
actuated automatically as the methanol injector valve was degree of salt hydration increases as the methanol concen-
opened. The crystal crops obtained from 1.31 specific gravity tration in solution increases to 70% to decrease thereafter.

446 Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 3, 1976
Table 11. Composition and Amount of the Salts Recovered per Liter of Bittern as a Function of Methanol in Solution"

Composition, % wt
Methanol in Salts recovered,
solution, % wt g/l. NaCl Na2S04 K2S04 MgS04 H2O Solid phase in addition to NaCl

40 78.15 0.10 2.25 54.68 37.38 5.61 K2S0pMgS04.H20 + NazS04

45 90.54 0.54 3.52 46.68 38.69 KzSO4.MgSOC2HZ0+ Na2S04-MgS04.2H~O+
11.52
MgS042H20
50 105.16 0.65 3.86 42.23 37.13 +
16.45 K2S04.MgSOC3H20 Na2S04.MgS04.3HzO +
MgS0~3Hz0
55 118.15 0.81 4.15 41.64 36.85 +
16.62 K2S04-MgSOc3H~0 Na2SO~MgS04.3Hz0+
MgS04.3H20
60 139 42 1.51 4.26 41.23 36.59 +
16.53 K2S04.MgS04.3HzO Na2SO.g.MgS0~3Hz0+
MFSOA-~H~O
65 151.16 1.85 4.71 39.75 36.97 16.65 KzS04-MgS04.3H26 + NazS04.MgS0~3HzO+
MgS04.3HzO
70 155.65 3.45 8.56 27.75 38.17 +
22.54 K2SO~MgS04.4H20 NazS04.MgS04-4HzO +
Md304-4HzO
80 138.54 2.69 21.01 23.44 36.38 16.52 K~S04.MgS0~3Hz6 + N a z S 0 ~ M g S 0 ~ 3 H+z 0
MgS04.3HzO
90 121.16 1.25 25.46 21.08 36.15 +
16.10 K z S 0 ~ M g S 0 ~ . 3 H 2 0Na2S04.MgS04.3HzO +
MgS04.3HzO
Traces of NazB407 and Cas04 were found
a Specific gravity = 1.31; salting time = 3 min; rpm = 300; temperature = 30 "C.

Table 111. Composition and Amount of the Salts Recovered per Liter of Bittern with Methanol as a Function of Bittern
Density"

Bittern Salts Composition, % wt

density, recovered,
g/cm3 gll. Cas04 NaCl Na2S04 K2S04 MgS04 H20

1.107 12.52 28.90 0.62 57.09 0.75 2.32 10.45

1.208 81.25 0.93 82.85 14.40 0.95 1.20 -
1.262 99.46 - 3.16 81.73 3.63 6.88 4.52
1.282 106.16 - 3.15 34.89 25.75 33.52 2.64
1.310 139.42 - 1.61 4.26 41.23 36.61 16.53
1.318 145.63 - 0.40 5.42 42.75 34.95 16.38
a Methanol in solution = 60 % wt; temperature = 30 "C: salting time = 3 min; rpm = 300.

Table IV. Composition and Amount of the Salts Recovered per Liter of Bittern as a Function of Ethanol in Solutionn

Ethanol Salts Composition, % wt

in solution, recovered,
% wt gll. NaCl Na2S04 MgS04 HzO
40 149.25 20.65 0.96 23.25 46.28 8.95
50 155.16 21.46 1.05 23.78 44.55 9.16
60 160.18 23.17 1.15 24.36 41.32 10.08
70 120.54 19.38 7.20 26.96 39.67 6.70
80 88.60 17.86 10.66 28.96 38.69 3.93
Traces of calcium, boron, and strontium
a Bittern density = 1.31; salting time = 3 min; rpm = 300; temperature = 30 "C.

Sodium chloride and traces of calcium sulfate and sodium cerned, were obtained with bittern of 1.31 specific gravity or
borate were also identified. However, the samples examined above. A series of experiments led to the conclusion that al-
did not show the presence of glaserite. cohol decomposition, if it occurs, is below the range of labo-
The polythermal composition of the solid phase vs. salting ratory detection.
time and polythermal data are not reproduced here for the The effect of temperature on the kinetics of PMS crystal-
sake of brevity, but two points are worth noting: the compo- lization from seawater bittern with methanol was determined
sition of the solid phase is independent of salting time u p to at 30 and 15 "C. Typical results, as summarized in Figure 3,
about 20 min. Beyond this point, the concentration of sodium indicate that the initial rate of crystallization is very rapid,
chloride and sodium sulfate shows a significant increase. Also, reaching maximum yield in about 30 to 40 s, and because of
the temperature range of 30-40 "C was the best compromise this fast rate of reaction the crystals formed are very fine, with
between yield and purity. a size average of about 30 g. The same figure shows that, a t
The test was repeated with bitterns of different density at increasing reaction time, the rate of crystallization becomes
30 *C and 60% methanol concentration. T h e results summa- independent of temperature. At a low reaction time a slight
rized in Table I11 reveal that the crystal crops recovered a t decrease in the rate of crystallization as temperature increases
specific gravity below 1.31 were too contaminated with sodium is observed. It is apparent from the results that a salting time
chloride and sodium sulfate to be of marketable value. T h e of about 3 min is sufficient.
most promising results, as far as purity and yield are con- A test was conducted using ethanol as the selective pre-

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 3, 1976 447
 LL FWRES m ms SALTMLD BITTERN
1+1( YEAR EITTERN 1,5649610
SPU I26
QS : OTWR SUTS
E W W R I T E D WATER
Hz0 503,410 4

qea 7

WS04 42, I 3 2 0

io- ~ ~- ~
,
I I
I
I

1 I I

I
I MAWESUM
I o*xy
PLWT
cipitant with bittern of 1.31specific gravity a t 30 "C. The re-
sults are presented in Table IV and indicate that ethanol is I6,lOZ B

a very effective precipitant for sodium chloride, potassium Figure 4. Material balance flow diagram for the recovery of potas-
sulfate, magnesium sulfate, and to a lesser degree for sodium sium magnesium sulfate double salt from seawater bittern by the
sulfate. This table shows also that the solubility reverses itself methanolation process.
a t about 60% ethanol concentration.
A combined analysis of Table I1 and Table IV indicates that crystallization was also a critical operation as far as yield and
ethanol is a more effective salt dehydrating agent than purity of product were concerned. Retention time, methanol
methanol, but methanol is preferred as a precipitating agent concentration, and temperature govern to a great degree the
because of its higher selectivity for potassium sulfate. The yield and purity of the product.
results presented in Table I1 and Figure 3 formed the basis for The 1.31 specific gravity bittern containing 4.63% K2S04
for the development of the novel process described in this 4.95% MgS04, and 22-49'??of other salts a t 30 "C, and 97% pure
work. methanol a t 28,"C, were fed continuously to a closed tank
The crystals were filtered and washed in the crystallizer. provided with agitation. A two-phase system, consisting of salt
An exhaustive series of laboratory tests demonstrated that in an aqueous methanol solution, formed immediately a t ap-
about 2.2 g of methanol per gram of unwashed crystals was proximately 38 OC.
sufficient for total displacement of the filtrate. The washed Considerable thought and study were devoted to the design
crystals containing about 52% by weight of methanol were of the crystallizer in order that the requisites of mixing,
dried a t 100 "C in an electrical muffle for 2 h, weighed, and growth, and removal of the crystals would be met. A vigorous
analyzed. agitation zone was necessary to ensure uniformity throughout
the magma. Final design was a circular deep tank 0.40 m high
Small Scale Pilot Plant and 0.35 m in diameter with four equally spaced baffles and
Economic considerations of any process for the recovery of a somewhat conical bottom with a discharge hole in the center.
chemicals, using an organic solvent, disclosed that the solvent A double turbine type stirrer, centrally mounted, and 2.5 h p
losses in the operation must be kept very low. For this reason per 1000 gal was found to ensure excellent blending. This
it seemed advisable to pursue a pilot plant development to system worked satisfactorily with a salting time of 3 min.
explore the fundamental characteristics of the process. The Thickening. The third step is the thickening of the
foregoing laboratory and pilot plant studies demonstrated the PMS-aqueous methanol bittern slurry. This step was added
feasibility of the process. The flow diagram for the l-ton- in order to concentrate the dilute feed to the vacuum filter
per-day pilot plant was the same as that outlined in Figure 4. and, thus, make the operation more efficient.
Concentration of the Bittern. Laboratory experience Filtration. One of the most important unit operations is
indicated that this operation was critical as far as purity and undoubtedly filtering and washing the crystals. This step
yield of product were concerned. In brief, the operation controls in great part the solvent loss and ultimate product
comprised the concentration of seawater saltfield bittern from quality, and must be conducted in a vapor-tight continuous
1.26 to 1.31 specific gravity in solar evaporation ponds. This unit to minimize solvent losses. A hooded 3 X 1 rotary vacuum
step raises the potassium concentration, and most important, filter fulfilled the requirements. The minimum wash rate that
reduces the sodium concentration, which is a strong compet- gives efficiency better than 98% is 2.2 g/g of unwashed product.
itor for the sulfate ion during crystallization. In the process Product purity is comparable to optimum laboratory washing.
of this investigation it was found that the crystal crops ob- The filter capacity far exceded that of other plant units, and
tained from bitterns with sodium concentration above 2.2% at full capacity the rate of filtration was about 815 kg m-2 h-l,
were too contaminated with Na2S04 to be of fertilizer value. but the filter effective area was decreased to one-tenth of its
Crystallization. Laboratory experiences indicated that original value through the use of an impermeable canvas.

448 Ind. Eng. Chern., Process Des. Dev., Vol. 15, No. 3, 1976
 Drying. This operation is a critical step as far as loss of operation, it is assumed to represent a true evaluation of the
methanol is concerned, but the design was not a major obstacle overall methanol recovery efficiency.
since a commercially available indirect screw-conveyor drier Loss of 0.10%of the methanol processed in the distillation
was adequate. The condensed methanol vapors from the column was good for a unit of this capacity. Filtration repre-
completely enclosed drier were collected in a vertical tube sents a large loss, and losses of 0.007 kg of methanol/kg of salt
condenser. The wet cake from the filter was discharged by produced were obtained in the drying operation.
gravity into the drier. The solid shows good flowability
properties and the operation had no outstanding mechanical Material of Construction
problems.
Distillation. I t was necessary to recover continuously the Corrosion was a major problem encountered in the pilot
methanol discharged in the filtrate and washings. The re- plant, although materials of construction were selected in
quired unit consisted of a 12 f t high, &in. diameter stainless accordance with small scale tests. The greatest difficulty lay
in the distillation column, crystallizer, and filter which were
steel column equipped with 10 plates; each plate contained
6 bubble caps. Heat was supplied to the still by a fuel gas di- constructed of type 302 stainless steel. After some time of
rect-fired reboiler. Still feed was pumped from a storage operation some corrosion was observed. Additional tests are
through a preheater inserted in the top of the column, where necessary for the selection of the best material for process
the ascending vapors provided the required amount of pre- requirements.
heat. The methanol product was collected in an intermediate
receiver after liquefaction in a horizontal tube condenser, and Conclusions
the still bottoms almost free from sulfate ions were conducted A simple crystallization method using methanol as the se-
to the experimental magnesium oxide recovering plant. lective precipitant may be employed for the efficient recovery
of potassium magnesium sulfate double salt of fertilizer
Evaluation of the Process
quality. The method could also be used with particular ad-
The primary objective of the pilot plant development was vantage in preparing a pure magnesium chloride bittern for
to evaluate this process with respect to potential commercial use in the manufacture of high purity magnesium oxide.
application. The preceding discussion of pilot plant results From the results presented in this work, it is evident that
is devoted mainly to mechanical operation, recovery of the the developed process offers enough incentive to utilize it for
solvent, and quality of the product. While these studies were the recovery of salts of fertilizer quality from the virtually
being conducted, the optimum operating conditions were inexhaustible and inexpensive seawater bittern.
being established, and, from this result, the material balance
flow diagram (Figure 4) was prepared.
The 1561 961 tons of seawater bittern fed to the solar L i t e r a t u r e Cited
evaporator (Figure 4) is the by-product of a one millon tons Borskey, J. W., Bulletins Pa. 25 4-60 and F. No 00303, international Minerals
and Chemical Co., Agricultural Division, Carlsbad, N.M., 1973.
per year NaCl plant, which at the present time is being wasted. Gaska, R. A., U.S. Patent 3 231 340 to Dow Chemical Co. (Jan 25, 1966).
It is important to note here that NaCl salt and MgClz rich Gaska, R. A., U.S. Patent 3 359 076 (Dec 19, 1967).
Gaska, R. A.. Goodenough, R. D., The Northern Ohio Geological Society, Inc.,
residue are produced as by-products in quantities sufficient 111 International Symposium on Salt, Cleveland, Ohio, Apr 21-74, 1969.
to almost defray the cost of recovering the fertilizer product. Gee, E. A., et al., lnd. Eng. Chern., 39 (9), 1178-1188 (1947).
The rich MgClz residue, free of most of the sulfate ion, is a very Hoppe, H., Chem. Eng. Prog., 12, 61-63 (1968).
Rafols. J. M., The Northern Ohio Geological Society, Inc., Cleveland, Ohio, pp
desirable feed for the magnesium oxide production plant. 186-194, 1970.
Roller, P. S..U.S. Patent 2 302 668 (June 25, 1946).
Recovery of Methanol Walsh, M. L. University of Wisconsin, Madison, Wis., private communication
to the author. 1972.
The lowest loss of methanol in the pilot plant was 0.027
kg/kg of salt produced with an equivalent average figure of Receiued for reuiew February 11, 1975
0.031. As the former figure was maintained over several days’ Accepted March 6,1976

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 3, 1976 449