Table 111. Separation of C S - B O ' ~and ~ Zr-Nbg5 in 0.5 Formal Ammonium Acid Fluoride Using the Seven-Cell Unit
I s n I 0pc
Cs-Br
I37
2 r-N
b9j
Final acidity Cs-Bal37 Zr-Nb95
Final acidity Flow rate, nil. miin.
Initial Is0 topP ..icticity, D.P.S.a X 108
Per Cent of Initial Isotop? Acticity in Solutions after l1,'2-Hr, Electrodialysis Anion Ifhed Cati 0 12 Experiment 1 2.21 0.0 18.0 81.6 1.68 63.5 34.0 0.0 -v 1.29 0.24 0.49
Experiment 2 2.07 0.0 1.74 69.7 .v 1,09 150
9.9
89.6
29.2 0.21 109
0.31 200
0.0
Disintqratiun, Po- second.
Table IV. Summary of Results from the Seven-Cell Unit i n a Mixture of 0.5N Nitric Acid and 0.5 F Ammonium Acid Fluoride
Initial Per Cent of Initial Isotope Isotope Acticity in Solutions after Actiuity, D.P.S.O 2-Hr. Electrodialysis Is0 tofie X 106 lnion Feed Cation Experiment 1 Cs-BrI3' 1.92 0.0 9 5 90 3 Zr-l\b93 1.61 34.9 62 1 0 0 Ce-Prli4 2.11 0.0 98.0 0.0 Sr-\'90 4.43 0.0 70.0 30.0 Pm14: 3.95 0.0 98 0 0.0 Flow rate! ml. min. 150 109 200 Experiment 2 8.2 91.8 2.37 0.0 63.7 0.0 1.64 30.7 99 . 0 0.0 3.24 0.0 Sr-YgO 4.75 0.0 60.0 39.': Pm147 4.02 0.0 99.0 0 0 Flow rate: mi. min. 150 109 200 Disinfegraiio?zsper second. I
Acknowledgment
T h e authors thank C. \V. Brauer for providing some of the d a t a for various phases of this work and D. E. 'l'routner for his assistancr i n suggcsting a mixed isotope counting technique. literature Cited (1) Xmphlrtt. C . B.. Fixation of Highly Active \Vasres in Solid Form. L X ' C 5,'19. Atomic Energy Resfarch Establishment, Harwell. England. 1956. (2) Brnedict. M.. Pigford. T. H.. "Nuclear Chemical Engineeri q . " p. 204. McGraw-Hill, New York, 1957. (3) Bruce. F. K.. Fletcher, J. M.. Hyman. H . H., Katz. J. J.,eds., "Progress in Nuclrar Energy 111-Procrss Chemistry," p. 345351. McGra\\.-Hill. Nris York. 1956. (4) Bub. G. ,I.. \\.ebb. \ V . H., Rei,. Sci. Inslr. 32, No. 7, 857 (1961). (5) Durham. K. \V.. Gouldrn. P. D.: Electrodialysis of Fission Product Solutions. Chalk Rivrr, Ontario. AECL-437, CRDC614, 1957. (6) Mason. 1'. .I.,,Juda. \Valtcr: "Applications of Ion Exchange
Membranes in Electrodialysis," Chem. Eng. Progr. Symn,b. S u . 5 5 , KO.24. 155 (1959). (7) Mason. E. X.. Parsi. E. J., ".Applications of Ion Transfer Membranes in Nuclear Chemical Processing," Second United Nations International Conference on the Peaceful Uses of Atomic Energy, Geneva? Switzerland. .A/Conf.. 15/P/502, 1958. 18) Parsi. E. J . . "The Electrochemical Utilization of Ion Exchange Membranes in AEC ODerations." ORNL-1812. Oak Ridre National Laboratory, O a i Ridge. Tenn., 1955. (9) Schubert. J.. Conn. E. E.. .Y~icleonics 4, 6 (1949). (10) Sieqel. J. M.. Rw..\fad. Phys. 18, 513-44: J . Am. Ci7rm. Suc. 6 8 , 2411 (1946). RECEIVED for review .Iugust 21. 1961 .ACCEPTED March 12. 1962 \i-ork financially supported by V. S.Xtomic Enrrgy Commission under Contract No. .AT (1 1-1) 770.
SUPERSATURATION OF SULFATES IN ELECT ROD IA LYSIS CA R L BE RG
E R A N D R0 BERT M
.
L U R I E, Ionics, Inc., 752 Sixth St., Cambiidge. .liass.
Small quantities ( 2 5 p.p.m.) of sodium hexametaphosphate and carboxymethylcellulose stabilize calcium and magnesium sulfate solutions in electrodialysis equipment at levels up to 225% of saturation. When calcium sulfate precipitates, the salt (CaS04.2H?O) i s first found within the unit. Precipitation first occurs at stagnant zones next to the membranes, the points of highest concentraiion. Four polyphosphates and CMC appreciably extend the time before the supersaturated solutions precipitate. The synergistic effect of SHMP and CMC has been noted. Higher temperatures and low pH's hasten hydrolysis of the polyphosphate and lead to lower stability. Silica sols appear to b e effective destabilizers for supersaturated Cas04 solutions, since a slowly settling precipitate i s produced within the unit which does not clog the flow paths.
L E C T R O D I . ~ I . Y S I S has
become a kvidely used method of deIt is a n especially valuable and economic technique for deniinrralizing brackish ground waters found in desert regions around the world as \re11 as in many other locations. h-ot uncoinmonly these waters have high levels of calcium and magnesium sulfates and may even be saturated. Since the electrodialysis process electrically transfers the salt from the product stwain to a waste stream: the \vaste stream
E salting Ivater.
must necessai il! become supersaturated if the raw \\-ater is saturated. .4 further complication is the disposal of the waste stream which. if poured out on the ground. might find its way back to the original ivell. In addition. it is desirable to desalt all of the !rater rather than just a portion. The primary objective of this work has been the determination of rhe conditions ivhich control the allowable level of superVOL.
1
NO. 3 J U L Y 1 9 6 2
229
saturation of calcium sulfate in the waste stream of electrodialysis demineralizers. The minimum relative volume of the waste stream is determined by the maximum permissible salt concentration in this stream. The waste stream could be further reduced by precipitating the salts from the supersaturated effluent and recycling the saturated solution for further concentration. Thus, damp salt would be the waste from this procedure. T h e work was divided into investigating addirives in stirred flasks and studying the actual effect of the operating conditions of a n electrodialysis demineralizer on the allowable level of supersaturation. A laboratory electrodialysis unit was employed to investigate the amount and location of precipitate as a function of the level of supersaturation, the additives, current density, flow rates, time, pH, and the presence of other salts in addition to CaS04. A commercial unit (Mark 11-4) was successfully operated. The results from four other commercial units that satisfactorily performed under supersaturated conditions are also reported.
Supersaturation Deflnitions I n defining supersaturation, two points of view may be taken. First, one may take a solution of salts which is supersaturated in calcium sulfate and allow it to equilibrate by precipitating enough calcium sulfate so that the solution is saturated. The degree of supersaturation may then be expressed as the percentage of saturation.
yosaturation
=
d ( C a + 2 ) (Soq-') initial final d ( C a + 2 ) (so,-')
x
100
The parentheses indicate the ionic concentration. However, one may also take a supersaturated solution and add water until the solution is just saturated. I n this case the final solubility product, (Ca+2) (s04-') final, is different, since the saturation concentration is affected by the concentration of the other salts which are present.
EXAMPLE.Suppose a solution contains 120 meq. per liter of C a s 0 4 and 240 meq. per liter of NaCl and the excess C a s 0 4 is allowed to precipitate. From Figure 1 (4,72) by trial and error, the saturation concentration of C a s 0 4 in 240 meq. of NaCl per liter is found to be 58 meq. per liter, and therefore
However, suppose water is added until the solution is saturated-that is, the CaS04/NaC1 ratio remains 0.5. Again, by trial and error from Figure 1, the CaSOl will be saturated when it has a concentration of 45 meq. per liter; NaCl will be 90 meq. per liter, since the ratio remains 0.5. Noiv:
These values are obviously very different numerically. If only the one salt being considered is present, the two methods are identical. Only when other salts which affect the solubility are introduced does this distinction become important. In this work the former technique was used-i.e., the supersaturated component was allowed to precipitate.
Stirred Flask Experiments i\dditives were tested by stirring 225 ml. of solution in 250-ml. Erlenmeyer flasks with magnetic stirrers. Saturated solutions (200%) of calcium sulfate and sodium chloride (1 to 2 equivalent ratio) have been used to screen additives. These were made by mixing a calcium chloride solution with a sodium chloride and sodium sulfare solution. The additives were used at 25-p.p.m. concentration and were added to the sodium chloride-sodium sulfate solution before the calcium chloride was blended. The stirring speed is set by measuring the height of the vortex in the flask and adjusting so that the heights are equal. Two observarions were made. cloud times and precipitation times. The former is the time at the first sign of haziness. -4flashlight is held at a right angle to the line of vision, so that scattering is observed. Precipitation time is the time when solids settle to the bottom of the flask when the stirring is stopped. At this time, the solution may be hazy. The latter is much easier to pinpoint and was used exclusively in the latter part of the experimentation. The additives found effective in preventing clouding of a 2007, saturated solution for at least an hour are sodium hexametaphosphate (SHMP), sodium pyrophosphate, sodium tetraphosphate. sodium tripolyphosphate, and carboxymethylcellulose (CMC, Hercules 7 M). .4dditives that did not prevent precipitation are several polyacrylic acids [Polyco 296 \$', Polyco 296 N (Borden Chemical Co.)], Acrysol A-3 (Rohm and Haas), citric acid, gelatin,
Table 1. Effect of Sodium Hexametaphosphate on Cloud Time for Supersaturated Calcium Sulfate Solutions at Room Temperature
70Saturation 200
0 A
HISO, K>SO,NolSO4
-
MpSO6+
-c--
X
-__~ ' 0
100
Figure 1.
200
300
a
-
400 500 600 700 BO€ 900 TOTAL SOLUTION CONCENTRATlON.ME0ILITER
MpCiiKCI NoCI
___ -
-~ 1000
Solubility of CaS04.2H20 in aqueous solutions a t
25" C. 230
I&EC PROCESS DESIGN AND DEVELOPMENT
250 300 400
450
SHA1fP:P.P..26. '0 0.05 0.5 1 0 2.5 10.0 15.0 20.0 25.0 250.0 25.0 25.0 25.0 250.0 250.0
Cloud Time 2 minutes 10 minutes 1 hour 2 hours 8-16 hours 1 day 3-4 days 3-4 days .4bout 10 days About 10 days 40-60 minutes 50 minutes 15 minutes 30-90 minutes 90 minutes
modified cellulose (CMHEC 37 51>Satrosol, Ceron A? C, T; all Hercules Powder Co.), calcium lignosulfonate, sodium ethylenediamine tetraacetate, dialdehyde starch (Miles Chemical Co.), alginic acid, sodium phosphate, sodium alginate, and a silica sol (Du Pont). The effect of sodium hexametaphosphate concentration and supersaturation level is tabulated in Table I , T h e data in Table I indicate that 200% saturation (for Ca/Na = 0.5, SO,/Cl = 0.5) is about the limiting concentration that can be stabilized by S H M P . Solutions of calcium, magnesium, and sodium sulfates (0.43/0.32/0.25 equivalent ratios) have also been stabilized Ivith the various phosphates. T h e effect of S H M P concentration is shown in Table 11. S H M P with C M C shows a marked improvement and indicates a substantial synergistic effect. Table I11 shows the further effect of supersaturation level on precipitation time. LYhen both C M C and S H M P are used, 2257, saturation appears to be a maximum safe concentration. T h e effect of temperature may be seen in Table I\’. The higher temperature makes the solution less stable. Table \shows that slightly acid solutions are less stable than a neutral solution. Higher temperatures and more acid solutions both accelerate the hydrolysis of polyphosphates, yielding phosphates ( 7 , 14). Phosphates d o not stabilize supersaturated solutions, and, in fact, form very insoluble precipitates with calcium. Destabilization. T h e destabilization of the SHMP-stabilized solution is of interest for the precipitation of salts outside of the electrodialysis unit. Precipitation of the salts outside of the demineralizer and filtration would permit the recirculation of a saturated solution to the concentrating compartments. Damp salts would be the waste product rather than a large volume of highly saline water. The polyphosphate molecule is presumably adsorbed on nuclei of calcium sulfate. By thus disrupting the crystal growth, precipitation is prevented. T h e multivalent hexametaphosphate also aids in stabilization of colloidal particles. Polyvalent ions such as aluminum are known to flocculate such particles? since the three positive charges depress the stabilizing charge around the nuclei and allow coagulation (5). Discharge a t electrodes in solution also causes flocculation of colloids (7, 8 ) . The addition of polyvalent ions and the use of such electrodes have both been successfully tried as methods for precipitation. Table V I gives the results of flocculation experiments with aluminum ions. Two electrodes a t 6-volt potential caused coagulation of a 2007, supersaturated solution ivithin a half hour. Boiling reduces the stability of colloidal particles (6) as well as greatly accelerating the hydrolysis of the polyphosphate, and in this case immediate precipitation is noted. Ceron C, a modified polyvalent cellulose (Hercules), was added (about 1%) ; flocculation resulted immediately. Seeding a n SHMPstabilized solution with freshly precipitated calcium sulfate did not destabilize the solution. One other series of experiments was carried out (Table VII). It was thought that the excess calcium sulfate might be precipitated out as nonsettling or very slowly settling fine particles, if many nuclei could be introduced into the system as a n alternative to stabilizing the particles. This offers the possibility of precipitation within the electrodialysis equipment in such fine particles that no clogging occurs. However, means of coagulating and filtering would have to be devised. This sol did not cause precipitation of a n SHMPstabilized solution. I n practice, the use of sodium hexametaphosphate is much more economical than silica sols.
Table II.
Effect of Sodium Hexametaphosphate on Cloud Time
(200% saturation Ca!Mg,”aSOk,
30” C.)
Pptn. Additioe SHMP SHMP SHMP SHMP,
Quantity,
P.P.M.
Cloud T i m e , Days
25
10
Time, Days 12
4
7.1
in
Table 111. Effect of Supersaturation Level on Precipitation Time in Presence of Carboxymelhylcellulose and Sodium Hexametaphosphate
(CaSOJNaCl
70Saturation
Additives SHMP. CMC SHhlP, CMC SHMP, CMC SHMP. CMC SHMP: CMC SHMP SHMP a
=
Pptn. T i m e ,
Cloud Time
Days >45 15 2 2 --stateeffects to be studied. T h e second method requires large reservoirs of solution for long runs! and the compositions that could be examined were limited to those that could be synthesized. Feed-Bleed System. I n this feed and bleed method the concentrating stream is continuously recycled through the stack (Figure 3). Fresh lvater is added to this stream and the overflow is disposed of. T h e rate of water addition (or stream disposal) and the current in the unit (or rate of salt addition to the stream) determine the concentration in the stream. T h e additive is placed in the \yarer \vhich enters the concentrating stream in the feed and bleed system. T h e stack is disassembled after each run and the location and amount of precipitate are noted. Samples, taken at the 232
ANOM WASTE
WATER
l&EC PROCESS D E S I G N A N D DEVELOPMENT
inlet and outlet of the unit during the run, are subsequently analyzed. Precipitation occurred within the unit when no stabilizing additives were used, even when the effluent stream \vas stable (did not cloud) for more than a day. Thus recirculating the stream and allowing old nuclei to develop in the stream d o not affect the development of crystals in the stack. T h e growth of crystals in the stack from a stream which does not cloud in a day is d u e to the concentration gradients in the flow path. Thus, at the membrane surfaces in the concentrating stream the salt is of higher concentration than the mixed center portion of the flowing solution. iyithout additives, precipitation occurred in all cases after 1 to 4 hours' running, except where the current was very low (4.5 ma. per sq. cm.). This is added evidence that concentration gradients cause localized high concentrations near the membrane which are unstable and precipitate. Sodium hexametaphosphate was then added to the feed water, so that the concentrating stream was 25 p.p.ni. (Table
LIII). I n these experiments, precipitate generally fornicd bet\veen
CiLUTERECCLE
__ I
I
i
t
i I
+
I
TOUNIT
ADDITIVES
FPR CONC.
FEED" 20 G L T D N K
5TIRRED
q c OMRROW BLED"
Figure 3. Feed-bleed flowsheet for electrodialysis supersaturation study
Figure 4.
Calcium sulfate dihydrate crystals as formed on spacers
the anion membrane and the spacer; i n the above experiments (with no additives) precipitate formed directly in the flow path. The crystals are large blade-shaped. This is typical of slow formation with no stirring, which is the condition between the spacer and membrane. The crystals have been analyzed as calcium sulfate dihydrate. It is not clear why the growth occurs a t the anion and not the cation membrane. This may well be related to the fact that polarization first occurs at the anion. Photographs have been taken of these crystals (Figures 4 and 5). The electrodialysis experiments reported in Table VI11 were conducted a t room temperature. Samples from the concentrating stream were withdrawn and put in agitated flasks where cloud times were observed. Two runs in this series resulted in precipitation in the flaw path: R u n 32 was at a high level of supenaturation (278 to 29070) and run 33 was a t a p H of 3 to 4, which possibly caused hydrolysis of the SHMP. A similar series of runs was made with a solution of calcium, magnesium, and sodium sulfates in the equivalent ratio of 0.43/0.32/0.25. Sulfate solutions have lower polarization limits ( 7 7 ) than
Figure 5.
chloride, presumably due to the lower degree of ionization of the former (2). Thus, as the current density rises, the sulfate ions do not move to the membrane fast enough and hydroxyl ions are transferred. This leads to a basic concentration stream and a n acid dilute stream. Hence carbonate scale forms in the concentrated stream unless acid is added. Table VI11 lists some of the runs made with Ca/Mg/Na sulfate. I n this last series of experiments, 13 ma. per s q . cm. was found to be the highest current density that could be maintained without precipitation in the flow path. T h e current densities of practical interest in commercial units are about 5 ma. per sq. cm. Flask runs indicated that 25 p . p m of SHMP with 25 p.p.rn. of C M C (Hercules CMC-7M) gave more stable solutions, so runs 42 and 43 tested this combination. A slight reduction in the amount of precipitate was observed: Five gallons of 2007, saturated solution were collected from run 43. This was continuously recycled through a D11 centrifugal pump with no signs of destabilization. I n earlier experiments without stabilizers in the solution, immediate precipitation occurred when the solution was passed through a pump. Another series of runs, not reported hei.e, showed that vary-
Gypsum crystals
Upper left.
Crystals taken from spacei Center and r!ght. Crystals on spacer
VOL
1
NO. 3
JULY 1 9 6 2
233
Table VIII.
Electrodialysis Test Data
Current
% Saturation R~~ Ma./ Concn. Concn. Time, Run Sq. Cm. in out Hours Density,
17 21 26
18 18 4.5
142 119 125
150 134 130
43/4 21/4 3
27 28 29 30
18 18 18 18
170 155 222
176 i 26 162 223
4 5 7 5
31 32 33
18 18 18
290 278 188
295 290 192
36 37 40 42
18 15 13 13
182 188 190 188
190 214 200 208
43
13
195
212
iio
Time to Cloud
25 P.P.M. Sodium Hexametaphosphate, CaS04/NaCl = 0.5 1 month Slight ppt. between anion and spacer Hazy immediately Slight ppt. between anion and spacer Hazy immediately Slight ppt. between anion and spacer Hazy immediately, Slight ppt. between anion and spacer no ppt. after 1 month 3 Hazy ppt. overnight Slight ppt. between anion and spacer, very slight in flow path 73/4 Hazy immediately Ppt. in flow path, on cation and anion membranes 6 Hazy immediately pH 3.1-4.4, heavy ppt. in flow path
25 P.P.M. 3 5L/2 6 6 6
Sodium Hexametaphosphate, Ca/Mg/NaSOc = 0.43/0.32/0.23 1 day Carbonate on anion, ppt. in flow path, between spacer and anion About 1 month Ppt. between anion and spacer, very slight ppt. in flow path 1-3 days Ppt. between anion and spacer 2 days Very slight ppt. on spacers; 25 p.p.m. SHMP and 23 p.p.m. CMC; velocity 39 cm./sec. 3 days Very slight ppt. on spacers; 25 p.p.m. SHMP and 25 p.p.m. CMC; velocity 50.5 cm./sec.
ing the flow velocity from 20 to 50 cm. per second and the holdup from 1 to 4 liters had n o measurable effect. Finally, a commercial unit (Ionics, Inc., Mark 11-4) has been operated continuously for two days using five cell pairs, Ca/Mg/NaS04, and 25 p.p.m. of S H M P and 25 p.p.m. of CIVIC. The linear velocity was 51 cm. per second in the concentrating stream. The current was 19.8 amperes for 1550 sq. cm. of 12.8 ma. per sq. cm. The rest of the operation conditions duplicated run 43, the concentration being about 210% saturation. The p H was adjusted to 4.5 to give a Langelier Index of about -1.5 ( 3 ) . When the stack was opened, very slight precipitate between the spacer and anion membrane was observed after 6 hours in run 43. Thus, further growth of crystals does not take place and continuous operation under these conditions should be trouble-free. Four commercial units have used satisfactorily supersaturated concentrating streams with sodium hexametaphosphate. NEW YORKSTATETHRUWAY. The concentration of salts in the water at the Junius Pond restaurant on the New York State Thruway is about SOY0 calcium sulfate and 20% magnesium sulfate. This unit has been running at 1457, saturation with 13 p.p.m. of S H M P with no precipitation of calcium sulfate, The unit operates on a cycle which recirculates the dilute Btream from a small tank until the proper concentration is obtained and then the water is discharged to a large storage tank. The concentrating stream varies in concentration, since the current changes during the cycle. However, the concentrating stream is always supersaturated and thus the use of such materials as S H M P is definitely effective in an electrodialysis unit for prevention of calcium sulfate precipitations. ROSWELL, L'EWMEXICO. At Roswell, New Mexico, Ionics, Inc., has installed six units drawing water from six separate sources. Two of these (sites 2 and 5) are operating with calcium sulfate supersaturation of 130 and 12070, respectively. Five of these units utilize 25 p,p.m, of SHMP. At site 2 the \cater also contains 22 p,p.m. of strontium and 1.5 p , p , m , of 234
,Votes
No Additive, CaS04/NaCl = 0.5 40 minutes Heavy precipitation 1 day Very slight precipitation '/z hour No precipitation
l & E C P R O C E S S DESIGN A N D DEVELOPMENT
barium, which were 6 and 66 times saturation concentrations. precipitation has been found in operation, so that these salts are also adequately handled with SHMP. HAVRE, MONTANA.Precipitation occurred within this unit a t 0 and 6 p.p.m. of SHMP, but not at 25 p.p.m. The precipitate was mainly barium sulfate. The unit is operating below saturation in calcium and strontium, but barium sulfate is 12 times saturation. The unit has been satisfactorily operated to date for 3 months lvith no signs of precipitation.
KO
Discussion The above data from flask and electrodialysis experiments may be summarized as follows : 1. The electrodialysis unit has been operated successfully a t 210 to 225% saturation with CaS04/NaC1 (0.5 ratio) and with Ca/Mg/NaSOd (0.43/0.32/0.25) utilizing 25 p.p.m. of S H M P and 25 p,p.m. of S H M P with 25 p.p.m. of CiMC as stabilizing additives. Large blade crystals of calcium sulfate dihydrate form between the anion membrane and the spacer. An 8-hour and a 48-hour run showed the same crystals. Commercial experience has also shown that the crystals will not grow out into the flow path, where they would cause high pressure drops and a reduction in current. Large blade crystals are characteristic of growth in unstirred solutions. If the membrane-spacer junction is not a very tight fit, there will be a conductive path through the membrane, along the membrane-spacer junction, and into the flow path. Since the liquid in the junction is stagnant and the salt concentration is highest a t the membrane (Figure 6 ) , the level of supersaturation can rise very high. Precipitation was not noted a t the cation membrane, even though this is a n analogous situation. 2. Precipitation occurs first within the unit. Even with extensive precipitation, the effluent may not cloud for a day or more (run 21). The precipitate is calcium sulfate dihydrate. Figure 6 shows that the current is carried in the flow paths by movement of cations and anions, but in the perm-selective
membranes the same current is carried only (about 95%) by cations or anions. In the solution 457c of the current is carried by calcium and 557, by sulfate in calcium sulfate solutions. Thus in the diluting path the sulfate ion is removed through the membrane faster than it is brought u p to the membrane by the electric current. T h e remaining sulfate diffuses up to the membrane. T h e sulfate concentration at the anion membrane becomes reduced until a t high current densities it is depleted at the surface. Hydroxyl ion is then transferred. IYhen 2 X 10-5 meq. of OH- is transferred per milliequivalent of current, the limiting current density is reached (by definition). I n practice, carbonate may then be precipitated in the concentrating stream Ivhere this base is coming through. At the cation membrane. hydrogen ion is transferred under similar circumstances. The limiting current density is usually correlated as the current density-normality in the dilute streamthat is, as the ionic concentration in the dilute stream decreases, the hydroxyl transfer becomes significant a t lower current densities. I n the flow path, the higher the velocity, the greater is the mixing that occurs and thus the ions will not be depleted at the membrane as readily. In the stagnant zone in the membrane-spacer interface, the following effects take place :
4. Four polyphosphates and carboxymethylcellulose have been found effective in appreciably extending the time before the supersaturated solutions precipitate. The synergistic effect of S H M P and CMC has been noted. Higher temperatures and low pH's hasten hydrolysis of the polvphosphate and thus lead to lower stability ( 7 , 73, 74). SHk4P-stabilized solutions may be destabilized by heating, adding aluminum or other multivalent (more than divalent) salts, impressing a potential with electrodes. and adding other flocculating agents. These facts are consistent with the hypothesis that the polyvalent additives are adsorbed on a nucleus of gypsum and disrupt the structure and thereby prevent further growth. The charge of the adsorbed molecule also prevents or hinders ions of like charge from approaching the nuclei. At 2007, saturation in a Ca/S04/NaC1 = 0.5 solution, the excess calcium sulfate concentration is 28.7 mmoles per liter. If this is stabilized with 25 p.p.m. of sodium hexametaphosphate (molecular weight 1490; hence 25 p.p.m. is 0.0168 mmole per liter), there is one molecule of S H M P for every 1700 molecules of calcium sulfate. A sphere containing 1700 molecules of calcium sulfate dihydrate would weigh 1700
6x1023
SpacerMembrane .4nion
Cell Dilute
Dilute
Cation
Concentrate
.Anion
Concentrate
Cation
Effect in Stagnant Zone Salt concentration drops very low, slightly acid Salt concentration drops very low, slightly basic Salt concentration very high, basic Salt concentration very high, acidic
gram
X 172.2 grams = 4.9 X
If such a particle has the same density as large crystals of gypsum. it would have a volume of
The diameter of a sphere would be V.6 n 3 = T
Thus the anion membrane-concentrate cell location is most favorable for precipitation, especially if it is assumed that carbonate will nucleate sulfate. Also calcium sulfate solubility is very markedly increased in hydrochloric acid, so that the cation-concentrate location would retard precipitation of calcium sulfate if chloride is also present. It can be concluded that the nucleation of precipitate in the concentrating stream outside of the unit is negligible? especially in the light of the very long shelf life of 200% stabilized saturated solution. Hence the holdup volume and recirculation rate are unimportant. The flow velocity (within the region studied 30 to 50 cm. per second) (runs 40, 42, and 43) also is not important, since the precipitation occurs in a sheltered, protected volume. 3. ,4s current density rises, the tendency toward internal precipitation increases (see runs 36, 37, 40, 21, and 26). This is consistent with the above discussion. As current density rises, more hydroxyl is transferred, the salt concentration a t the spacer-membrane interface is higher, and thus precipitation occurs to a greater extent.
DILUTING STREAM
CONCENTRAM STREAM
DILUTING STREAM
ANOOE
D
47
o'212 X 6 X 10-18 = 0.74
x
10-6cm.
= 74 X 10-8 cm. =
74 A.
=
7.4 m y
This is a colloidal sized particle and reasonable for such a stable, clear suspension. A molecule of the polyphosphate (approximately Na16P14043) would be approximately 3.5 mp long ( P - 0 bond length is 1.52 A , ) , and assuming the molecule is linear (73) with bond angles of 109", the area (length X Lvidth) would be 31 sq. mp, This could lie on the spherical particle and be about '/6 of the circumference in length. If fully ionized, it would have 16 negative charges. Of course, there could well be more than one polyphosphate molecule to a particle and hence larger size particles. Thus:
s o . of
Polyphosphates per Particle 1 2 3 4 5 10 100 171
Particle Diameter,
l r e a of
Sphere,
My
sq. MI1
7.4 9.3 10.7 11.75 12.7 15.9 34.4 40.9
172 282 356 432 505 800 3720 5250
7'
.hea o,f Sphere CoLered 18 22 5 26 28 8 30 8 39 0 83 100
Charges per Particle ( 16 per Polyphosphate) 16 32 48 64 80 160 1600 2730
SPACER
STAGNANT ZONE LOCATION OFCRYSTALS
Figure
6.
Electrodialysis cell
Even a t 100% coverage the particle is definitely colloidal (40.9 mp). Such particles may be destabilized or coagulated in several ways. Multivalent ions such as aluminum will cause the VOL. 1
NO. 3
JULY 1 9 6 2
235
counter-ion layer (made u p of the sodium ion from the S H M P and other ions in the sclution) to move closer to the particles. Collision of t\vo particles (and subsequent coagulation) is then more likely. Contact a t a n electrode is also a method for coagulation. Heating the solution gives greater thermal energy to the particles. \vhich aids coagulation. These methods have been tried and all work with these solutions. 5. .\ silica sol \vas used to precipitate unstabilized supersaturated solutions. It gave a precipitate which settled very slo\vly (I, 2 to 1 hour). -4 very slo\vly settling precipitate might allow precipitation Lvirhin the unit without clogging the flow paths. Acknowledgment
.A literature survey by E. Gutoff and laboratory assistance of -A. J. Giuffrida and 1Iarcia E. Berg are gratefully acknoivledged. Data from field experience ivere obtained from I V . B. Iaconelli.
Brackish \Vaters,” Dept. of Interior, OS\$‘ Progress Rept. 11 (December 1956). (4) , , Kellev. K. K.. Southard. J. C.. .Anderson. C. T.. U. S. Bur. Mines Tech. Paper 625 (1941). (5) Kruyt, H. R., “Colloid Science.” Vol. 1, pp. 81, 129, 302, Elsevier, New York, 1950. (6) Lewis, IV. K., Squires. L., Broughton, G., “Industrial Chemistry of Colloidal and Amorphous Materials,” pp. 122, 183, Macmillan, New York, 1948. (7) Ibid., p. 438. (8) McBain, J. \V., “Colloid Science,“ p. 191, D. C. Heath, A-eiv York, 1950. (9) Mason, E. A , , Juda, IV., Chem. Eng. Progr. Symposium Set. 5 5 , No. 24, 155 (1959). (10) Mason, E. A , , Kirkham, T. A . Zbid.. 5 5 , No. 24, 173 (1959), (11) Rosenberg, N. \V., Tirrell, C. E., IND.ENG. CHEM.49, 780 (1957). (12) Seidell, .4.,Linke, I V . F., “Solubilities of Inorganic and Metal Organic Compounds,” 4th ed., Vol. I, Van Nostrand, Princeton, N. J., 1958. (13) Van IVazer, J. R.: Callis, C. F., Chern. Rev. 58, 1011-46 (1958). (14) V i n IVazer J. R.. Griffith. E. J., McCullough, J. F., J . A m . Chern. Soc. 77, 287 (1955). RECEIVED for review October 26, 1961 .ACCEPTED April 12, 1962
literature Cited
(1) Bell, R. N., IND.ENG.CHEM.39, 136 (1947). (2) Davies. C. W., “Structure of Electrolytic Solutions,“ I V . J. Hames. rd.. Chap. 3. IViley. New York, 1959. (3) Ionics. Inc.. “Design. Construction, Field Testing and Cost Analysis of an Experimental Electrodialysis Demineralizer for
Division of IVater and IVaste Chemistry, Symposium on Saline Il’ater Conversion. 139th Meeting, ACS. St. Louis, Mo., March 1961. LVork supported by the Office of Saline IVater, Department of Interior, under Contract 14-01-001-180.
CORRESPONDENCE ENERGY-NEW SURFACE RELATIONSHIP IN PARTICLE CRUSHING SIR: The research described by R. A. Zeleny and E. L. Piret [”Dissipation of Energy in Singlc Particle Crushing,” ISD. E N G . CHEM. PROCESS DESIGNDEVELOP.1, 37 (1962)] is of considerable interest. It shows a linear relationship betlveen heat generation and nelv surface produced in the crushed material, the slope of the line for quartz being equivalent to 7.7 X lo4 ergs per sq. cin. This result is similar to Method of those \vhich I obtained in the early 1950’s [“-4 Predicting the Performance of Commercial Mills in the Fine Grinding of Brittle hIaterials?” Trans. Znst. M i n i n g M d . 63,
that the performance ratios of the various mi!ls related to the unit impact crusher were the same for the three brittle materials examined. The results obtained are quoted in Table I. The energy losses in tne double pendulum crusher used by Zeleny and Piret are almost certainly less than in the unit impact crusher used by me: in lvnich the particles are crushed by a close1:- fitting piston sliding in a hollow anvil. Nevertheless. the rnergy usage for quartz quoted by them is gratifyingly closr to those quoted in the table.
211 (1953-4)].
G. L o d e Fairs
I n this \%rork2I compared the performance of a number of types of commercial grinding equipment with that of a unit impact crusher. I obtained a linear relationship bet\veen the net energy input and the new surface produced. I also showed
.21l11
A. Unit impact crusher B. Batch ball mill C. Swing hammer mill with coarse clearances D. Swing hammer mill with fine clearances E. Attrition mill with static classifier
Table 1. Comparison of Mill Efficiencies Aaerape EnergylUnit .Vew Surface, Ergs/Sq. Cm. Limestone Barytes .inliydrite 9.96 X l o 4 6.81 X 10‘ 6.33 X lo1 1 . 0 2 x 105 6 . 4 3 X lo4 6.42 X 10‘
9 . 2 5 X IO‘ 1.95 2.77
x x
Limestone 1 0.98
Performance Ratio Barytes iinhydrite 1 1 1.06 0.99
6.84 X lo4
6.06 X 10‘
1.08
1 .oo
1.05
x
1 . 0 2 x 105
0.51
0.67
0,62
x
0.36
0.36
0.37
105
1.02
1oj
105
1.89 X 10%
~~
236
General Chemicals Division, Diuision Engineering Department. Imperial Chemical Industries, Ltd., Runcorn, Cheshire. England
ILEC PROCESS DESIGN A N D DEVELOPMENT
1.71
105
