A Recovery Scheme fo r . . .

Poisoned Ion Exchange Resins

MAYER B. GOREN’

Kerr-McGee Oil Industries, Inc., Oklahoma City, Okla.

10s exchange unit operations have played an increasingly important role in hydrometallurgical recovery processes (6, 77) : and have been extended to the re- covery of uranium.

This recovery scheme is susceptible to accumulation of relatively nonelutable substances in the resin beads so that op- eration becomes less efficient. Mechan- ical poisoning is due to accumulation of insoluble fouling materials within the resin matrix; chemical poisoning, to rela- tively irreversible adsorption of ionic material on exchange sites.

Both types are characterized by pre- mature or persistent leakage of mate- rial ordinarily sorbed (early “break- through“), while saturation capacity is affected less by mechanical than by chemical fouling. Mechanical fouling also interferes with normal elution by physical blocking of resin pores. Because of the value of ion exchange resins, it is not practical to discard them after short use.

Among the agents contributing to the poisoning problem in uranium recovery, silica, polythionate, cobalticyanide, and molybdenum have been most prominent. Other metals (zirconium, titanium) and anions (phosphate), as well as water- dispersible organic materials, have played a lesser role (7-6, 8, 9 ) .

Figure 1 compares loading character- istics of fresh IRA-400 (Rohm & Haas Co.) and after some 100 cycles when feed liquors contained minor proportions of poisoning elements. Feed used for ex- amining loading characteristics was a synthetic liquor containing 1.0 gram of uranium oxide ( U 3 0 ~ ) per liter a t p H 1.5. A background of 30 grams of sulfate per liter was added as magnesium sulfate; retention time was 3 minutes. Thus fresh resin exhibits 1% leakage a t a uranium oxide loading of about 55 mg. per cc. of resin; poisoned resin, a t only 10 mg. per cc.

Figure 2 shows elution characteristics of the same resins after loading to satura- tion. (Eluent, O.BAlr sodium chloride- 0.1-V hydrochloric acid.) Even though poisoned resin has a lower saturation capacity, more eluent is required for

Present address, 5950 hlcIntyre Rd., Rt. 1, Box 299A, Golden, Colo.

Uranium recovery by ion ex- change processes may be- come less of a headache with the development of this process for restoring the resins. Treatment with strong caustic, then strong acid at &month intervals can effec- tively prolong resin life

stripping it to the same degree of “bar- renness” in the effluent than does fresh Of elutions (Figure 2) became intoler- resin. ably prolonged. Spectrographic analy-

sis of ashed resins indicated that other

Resin Poisoning at Shiprock contaminants were hafnium, silicon, iron, and phosphate; in addition organic . -

The solubilization scheme employed a t the Kerr-McGee ShiDrock. N. M..

complex humates were found.

mill during the first year of operation was a n “acid cure” process wherein the ore (largely carnotite-roscoelite deposits from the salt wash of the Morrison forma- tion) is pugged with approximately 600 pounds of 60% sulfuric acid per ton of ore and allowed to cure in a pile for some 16 hours. Heat of reaction raises the temperature above 100” C., and mineral values are effectively solubilized.

Water leaching then dissolves metal values, and leach liquor is separated from the solids by conventional methods. Acid cure is a very vigorous, nonselective process and solubilizes a considerable portion of undesirable minerals. The leach liquor, which subsequently pro- ceeds to ion exchange, might typically have the followirig composition :

Grams/Liter us08 1-1.5 vzo5 4-6 AlzOj 8-12 Fe 3-5 so4-- 50-100 Pod--- 0.1-0.3 Sios, Mo, Zr, Ti Minor quantities

Pilot plant studies indicated that molybdenum might be the major poison- ing agent because of an essentially irreversible adsorption on quaternary sites. Such poisoning is ordinarily read- ily alleviated by regular elution with 5 to 10% sodium hydroxide or smaller amounts of causticcombinedwith cheaper reagents such as salt (9 ) . I n full scale plant operation poisoning was more rapid than anticipated, and zirconium and ti- tanium, in addition to molybdenum, were the chief contaminants.

Within 40 cycles ash content of the resin was 7yo and increased to 14 to 20% for 130-cycle resin. Break-through capacity dropped to about 307, of that of unfouled resin. Effective saturation capacity decreased markedly, as columns could not be kept on stream because of excessive leakage. and “tailing-out’’

Development of Regeneration Process

Laboratory tests indicated that satura- tion capacity (during the not too aggra- vated stages of poisoning) a t long reten- tion times was not drastically lowered, but break-through capacity a t normal flow rate had been affected considerably. This was interpreted to mean that fouling was largely mechanical (intrdresin) rather than chemical. Treatment with caustic removed onlv about half the inorganic material along with organic color bodies. On the other hand, 1 to 2 s acid (hydrochloric, nitric, or sulfuric) dissolved only a trace (largely iron) of the precipitated materials.

Resins can be restored to essentially ash-free condition and original operating characteristics by prolonged treatment with (ethylenedinitri1o)tetraacetic acid- caustic (7) ; however, the procedure was too costly to be practical.

Strong Acid-Strong Base Regenera- tion. The predominantly mechanical nature of the fouling prompted consider- ation of the poisoned resin therefore somewhat as an ore. Because constit- uents deposited within the resin matrix had been solubilized in the acid cure, similar, though necessarily less drastic, treatment might redissolve the de- posited material.

Exploratory experiments indicated that slow percolation of 1 to 1 sulfuric acid (about 65%) through the resin a t ambient temperature very effectively removed considerable quantities of in- organic material, among which molyb- denum, titanium, zirconium, and iron were identified. However, the resin was irreversibly darkened by this treatment, and more dilute acid was tried; 48% (1 to 2.0) and 42 5% (1 to 2.5) sulfuric acid were very effective in slow percola- tion (about 15 hours or longer) and had no destructive influence on the resin.

VOL. 51, NO. 4 APRIL 1959 539

7 c

I I O Z “ 30 4 U s o

BED VOLUMES FEED

Figure 1 . Uranium oxide loading characteristics of fresh and poisoned resins Poisoned resin shows 1 % leakage at a loading of only 10 mg. per cc.

0 Fresh resin X Fouled resin

Examination of 40-cycle resin ( 7 7 , ash) after strong sulfuric acid treatment indicated that almost all titanium and zirconium had been removed; however, some molybdenum and silica still re- mained. These Lvere then largely re- moved by prolonged treatment with l0Yc sodium hydroxide which also displaced large (apparently) quantities of organic color bodies. Resin so treated \vas re- stored to an ash level of 0.55Y0 and to operating characteristics essentially in- distinguishable from those of fresh resin.

The capacity of “brand ne\v” resin is actually some 120% that of “restored” resin; however, even under ideal condi- tions this excess 2070 capacity is lost in early use possibly as a result of conver- sion of quaternary to less basic sites. Thus after about 15 to 20 cycles. ne\v resin capacity in a nonpoisoning environ- ment levels off a t approximately 80 to 85% of that of ”brand new” resins.

This strong acid-moderately strong alkali treatment does not appear to be harmful to the resin, and in one respect (increased porosity) is actually beneficial even to new resin (7 ) . Heavily poisoned resin was restored by an abbreviated strong acid-base process \vherein about 9Orc of the ash was removed with an increase in resin volume of 1.3% and in moisture content from 41.1 to 45.17,.

Efficiency of ash dissolution at moder- ate contact times (less than 6 hours) begins to decrease as degree of fouling increases. Thus with 100-cycle resin (ash content, 1470) some 25 column volumes of 1 to 2 sulfuric acid at 4-hour contact time and 13 column volumes of lOy0 caustic reduced ash content only to 1.51 to 2.67yG. Equal or better re- sults are obtainable \vith smaller reagent quantities a t longer contact times.

Inverse Process. During strong acid treatment of the resin, the effluent was examined periodically to check the progress of metal dissolution. After iron had been essentially completely displaced, metallic constituents in the effluent were precipitated by excess base but apparently not by excess car-

2 :

Figure 2. Elution characteristics o f fresh and poisoned 5 resins 3

W*(

-I More eluent i s required for stripping poisoned w . resin to a given degree . of effluent barrenness -

2 0 Fresh resin X Fouled resin m

0‘1

r? 3

0

I \

5

2 1 4 5 b 7 8 9 80

B E D V O L U M E S EFFLUENT

bonate-or at most they gave only a slight turbidity in excess carbonate. These metals were identified as zirco- nium and. to a lesser extent. titanium.

This behavior is somewhat puzzling in that neither zirconium nor titanium is knoivn to form stable carbonate complexes (70). Nevertheless zirconyl solutions can be caused to precipitate and the precipitate “redissolved” by excess carbonate. Redissolution may \vel1 be a colloidal phenomenon. as strong heating (or addition of sodium hydroxide) produces irreversible pre- cipitation of zirconium.

Percolation of dilute (lOyc) sodium carbonate through the poisoned resin eluted considerable quantities of zir- conium, molybdenum. and phosphate. along with lesser quantities of titanium. After carbonate treatment (or even after treatment with strong sodium hydroxide), residual metal oxides in the resin Lvere considerably more susceptible to leaching by acidic reagents than before alkaline treatment. so that the amount of acid required to remove metallic constituents completely from even very heavily fouled resins \vas markedly lowered.

Because silicate in the resin might undergo a cerrain amount of dehydra- tion under the influence of strong acid with resultant intractability toivard sub- sequent basic dissolution. it is regarded as fortuitous that alkaline treatment as a first step obviates this dehydration and promotes subsequent dissolution of tita- nium. zirconium, and other metals by the acid.

Reagent requirements for the funda- mental and inverse processes are com- pared in ‘Table 1: which also summarizes over-all effectiveness in terms of residual ash and loading and elution charac- teristics.

Conversion of relatively intractable metal contaminants into acid-soluble species by alkaline treatment is inter- preted as arising out of a metathetical reaction involving precipitated impurities and caustic and:or carbonate. In the

Shiprock resins, zirconium. titanium, and possibly other metals appear to be precipitated within the resin as phos- phates, which accounts for their inert- ness to all but very strong acids. Metath- esis Lvith carbonate or h)-droxide might convert acid-insoluble zircon)-l phos- phate into acid-soluble zirconyl hydrox- ide: possibly according to Equaiion 1, facile dissolution of zircon)-l hydroxide proceeding according to Equation 2. 221-0? P 2 0 , 5 H 2 0 + 12NaOH -+

4Na,$PO4 + ZrlOa(OHjz + 15Hs0 ( 1 ) Zr?Ol(OH)- + 4H2SOq -

2Zr(S04) : + jH,O ( 2 )

Additional Requirements. The “molybdenum blue” formed to a small extent by the in situ reduction of ad- sorbed molybdenum is more easil!, eluted by caustic or causlic-salt mixture if it is first reoxidized. Organic materials (humates) are also eiuted from the resin by alkaline reagents and are more effectively removed after mild oxidation. In practice, contacting the fouled resin with ammonium nitrate-sulfuric acid (about 0.75.Y-0.25.Y) effects these osi- dations satisfactorily. To minimize os- motic shock it is expedient to convert the resin from a hydroxide (or carbonate) form to the sulfate or chloride form (by contact with the appropriate salt) be- fore the resin is in contact with strong acid. This salt treatment elutes addi- tional traces of molybdenum, organics, and generally frees some surface dirt. \Vith these refinements? the most effec- tive sequence of operations is as follo\vs:

It et ci 1-

tion Column Time,

Operation Concn., % i-01. Hr. NHaNO.4-

H12S04 0 . 7 5 S - 0 . 2 5 S 5 2-3 Rinse NazCOa 5-10 5 8-10 N a O H - N a C I 5-10 of each 5-10 12 Rinse N a C l 5-10 5 3-5 &SO4 20-45 2.5-5 16-24

When this sequence was carried out on heavily fouled resin, ash content was

540 INDUSTRIAL AND ENGINEERING CHEMISTRY

reduced from 14.6 to 0.1 1 yc, and operat- ing characteristics were restored to normal. Under less ideal conditions in plant practice, ash content was reduced to 0.54% (967, dissolution). In practice, the first three steps can be omitted with relatively little loss of efficiency and some saving in reagents. This has been done at Shiprock several times.

laboratory Studies

All materials were ordinary C.P. reagents. Ion exchange resins were samples of -4mberlite IR.4-400 (Rohm & Haas Co.) , taken a t various times from cells of the Navajo Uranium Division mill, Shiprock, K. M.

Capacity tests were carried out in small laboratory columns, 17-mni. inside diam- eter and 110 cm. long. Columns had the usual feed, backwash, and effluent lines, the resin being supported on a 1- inch bed of fine gravel resting on a Witt plate. Influent liquors \rere grav- ity fed by a constant-head device to obviate changes in flow rate.

Capacity determinations were carried out xvith synthetic pregnant liquors prepared by dissolving relatively pure ammonium diuranate in about 407, sulfuric acid. diluting, and adding sufficient magnesium sulfate to provide 30 grams of sulfate per liter a t ultimate dilution. Final adjustments in volume afforded 1.0 gram of uranium oxide ( U 3 0 ~ ) per liter. and p H was adjusted to 1.5 with either sulfuric acid or concen- trated ammonium hydroxide as needed.

Flow rate in loading synthetic preg- nant liquor was maintained a t 10 ml. per minute through a 75-ml. bed (about 3.3-minute retention time). Samples of

effluent \vere collected periodically and assayed by standard fluorometric pro- cedures to determine extent of uranium leakage. T o compare resin samples, break-through capacity was arbitrarily defined as the loading at which lYO leakage occurred (10 mg. of uranium oxide per liter in the effluent). After break-through capacity had been deter- mined, an equal volume of synthetic pregnant liquor was passed through the column to load the resin to saturation. Before elution, the resin was rinsed and back-washed as in standard practice.

Elution characteristics of resin samples were determined on resins loaded to saturation. Eluent was a solution 0 . 9 S in sodium chloride and 0.1-j- in hydro- chloric acid and was fed at 3 ml. per minute (about 10-minute retention time). Effluent was sampled periodically to determine the degree of eluate barrenness a t a given level. Table I summarizes results of the various procedures.

Regeneration Processes

EDTA-Sodium Hydroxide Process ( 7 ) . A sample of freshly eluted resin (68 cycles, 9.17, ash) was transferred to a column and treated as follows :

(Ethvlenedinitri1o)tetraacetic acid (EDTA). 100 grams, and 58 grams of sodium chloride were suspended in about 900 ml. of water; 40 grams of sodium hydroxide as a SOYc solution was added to dissolve EDTA. The pH of the solu- tion after dilution to 1 liter was 6 0. T\senty-seven column volumes of this solution fed at 30-minute retention time Lvere required to elute all material precipitable by base. The resin \vas then well \sashed.

R E S I N R E C O V E R Y

Four column volumes of 1070 sodium chloride solution were fed at 10-minute retention t i n e to elute EDTA. Next, 10 column volumes of 1070 sodium hydroxide removed considerable organic material (dark red-broLvn) during an %hour contact. Molybdenum and silica \\'ere likewise dissolved. Finally, three column volumes of 67, sodium chloricle solution [rere followed by rinse.

The washed resin (0 .60 j , ash) ex- hibited break-through and saturation capacities for uranium oxide of 51 and 61 mg. per cc. of \vet settled resin, re- spectively. Eluate a t 13 bed volumes assayed 7.5 mg. of uranium oxide per liter. By comparison, used unpoisoned resin has break-through and saturation capacities for uranium oxide of 57 to 60 and 65 to 70 mg. per cc. of resin.

\Vhen 40-cycle resin (77; ash) was treated in the same manner, ash content was reduced to 0.337;: break-through and saturation capacities were 55 and 69 mg. per cc. of resin, respectively.

Strong Sulfuric Acid-Sodium Hy- droxide Process. Resin (40 cycles; 7Yo ash) was treated IS follo~vs:

Fifteen bed volumes of 1 to 2 sulfuric acid were percolated through the resin for 12 hours. Effluent \vas turbid \vith dissolved and reprecipitated solids. ?rletals identified by qualitative tests included iron, molybdmum. zirconium, titanium. and uranium. To\vard the end of this treatment qualitative tests for titanium and zirconium \\-ere faint.

The resin \vas well rinsed and treated with 10 column volumes of 107; sodium hydroxide for 8 hours. This treatment removed essentially the remainder of the molybdenum? silica, and organic color bodies. The resin \\-as then converted

Table I. Resin Regeneration Processes In plant practice regeneration with the condensed inverse process was as effective as the complete version and required much less time

Process EDTA

Effluent

ILesiii _____ M g . G O I /

Contact Cagarity. Mg. L O ? 'c'c. *issay,

lteageiit Sequence \ ~ o l u m e ~ Hr. Initial Filial Break-through Saturatiori Liter" Bed Time, Ash, yo

10% EDTA, 1.V NaCl, 1.1- NaOH 27 33 7 0 .33 55 69 ...

10% NaCl 4 2 9 . 1 0 . 6 51 61 < 10

6% NaCl 3 ... 10Tc NaOH 10 8

48% H i s 0 1 10% NaOH lOc/, NaCl

15 12 7 0 .55 55 67 10 8 4-5 3

<10

Complete inverse process 0.8-1- NH~NOB-O.Z~.V 9 7 . 5 14.6 0 .11 54-57 63-66 < 10

10% Na?C03 7 8

10% NaCl 6 4 427, H2S01 4 . 5 20

10% NaOH 8 8

Condensed inverse process 10% NaOH-10% NaCl 6 15 18.7 0.71 57 lOTc NaCl 6 6 207, H2S01 7 24

... ...

Fresh resin capacity On 13th bed volume.

... ... .. .. 60 72 < 10

VOL. 51, NO. 4 APRIL 1959 541

to the chloride form with salt and rinsed. Residual ash was reduced to 0.5570;

break-through and saturation capacities wrre determined as 55 and 67 mg. per cc. of resin, respectivelv; eluate con- centration at 13 bed volumes was 5 to 10 mg. of uranium oxide per liter.

When contact times during acid and alkali treatment were reduced to 4 and 2 hours. respective1)-, reagent require- ments \sere essentially doubled to achieve the same degree of regeneration.

Complete Inverse Process. Resin (100 cycles. 14.67, ash) was eluted a t 20-minute retention time with 9 bed volumes of 0.8aV ammonium nitrate- 0.25&V sulfuric acid. Effluent contained principally iron and a small amount of uranium. The resin was then washed with 6 to 7 bed volumes of water.

Seven bed volumes of 10% sodium carbonate solution were fed a t 30- to 40- minute retention time. Alkaline effluent was orange from elution of organic material. During carbonate elution, the effluent Isas tested for the presence of base-precipitable material, and after 7 bed volumes these substances were no longer being eluted from the resin. Spectrographic analysis of the carbonate effluent showed :

Major constituents : Mo, Zr, Ti, V, Hf Minor to trace quantities : Fe, AI, Ca, Mg, Cu

An effluent sample was made strongly basic with sodium hydroxide, and the precipitate which formed was collected, washed. and dried. Spectrographic analysis indicated the principal constitu- ent to be zirconium; 1 to 10% titanium was estimated to be present. Silicon, haf- nium, copper, and iron were also found.

Eight column volumes of 10% sodium hydroxide were percolated a t 30-minute retention time, with molybdenum issuing in trace quantities a t the end of this treatment. Caustic effluent was red- brown from organic material. A com- posite of the liquor, analyzed spectro- graphically, contained:

Major constituents: Mo, V Minor to trace quantities : Si, Ca, Fe, Al, Mg

The resin was washed and backwashed to remove macroscopic solids, and 6 bed volumes of 10% sodium chloride were fed a t 15- to 20-minute retention time. The salt displaced additional color bodies and a small amount of molybdenum. A washed sample of the resin was then dried, carbonized, and ignited to leave ash equivalent to 3.457, based on resin. Thus ash level was re- duced by 75y0. Qualitative tests were confirmed by spectrographic analysis :

Major constituents : Zr, Ti Significant quantities : Fe, Hf, Si, Ca Trace amounts: Al, Sr, Mg, V, Cr, Pb, Sn

The resin was finally treated with 427, sulfuric acid (4.5 column volumes)

for 18 to 20 hours. Initial acidic effluent was so concentrated in zirconium that when a few drops of phosphoric acid were added to indicate its presence the solution instantly gelled.

Other qualitative tests indicated that the effluent contained largely iron and titanium. After 4.5 bed volumes. quali- tative tests for these metals were nega- tive. The effluent was analyzed spec- trographically with the following results:

Major constituents : Zr, Ti Minor constituents: Fe, Ca, Hf, Sn, Cu Trace amounts : Si, Al, Mg, Cr

T h e resin was thoroughly washed and backwashed to free a small amount of deflocculated macro solids. A dried sample was charred and ignited, leaving a residue amounting to 0.1 1% of the resin. Thus this process removes better than 99% of the fouling constituents in very heavily fouled resin.

Break-through and saturation capaci- ties for uranium oxide were 54 to 57 and 63 to 66 mg. per cc. of resin, respec- tively. Elution of the saturated column afforded a n effluent (at 12 bed volumes) assaying below 10 mg. of uranium oxide per liter. By comparison, untreated 100-cycle resin (Figure 1) showed almost continuous leakage during loading, and over 19 bed volumes were required to achieve a similar degree of barrenness in the effluent solution.

Condensed Inverse Process. Eluted and well washed 146-cycle resin (18.770 ash, 4 . ~ 3 8 7 ~ molybdenum) was sus- pended in very dilute caustic and gently agitated. Agitation freed a consider- able quantity of surface slimes. Molyb- denum and phosphate appeared in the liquid almost immediately.

The resin was transferred to a column and over a 15-hour period 6 bed volumes of IOyG sodium hydroxide-10y0 sodium chloride were fed. Molybdenum and phosphate were almost undetectable in the column effluent a t the end of this treatment. Washing was followed by percolation of 6 bed volumes of 10% sodium chloride (6 hours); then the column was washed, and dissolution of residual metals was continued with 2070 sulfuric acid (7 bed volumes) for 24 hours. Final effluent contained traces of zirconium, but titanium was not de- tectable. Washed and dried resin con- tained 0.7101, residual ash, and regen- erated resin had a break-through capacity for uranium oxide of 57.5 mg. per cc. of resin.

Plant Practice

For convenience in preparation and storage of solutions involved in resin regeneration two 12 x 12 foot mild steel tanks and a n 8 X 10 foot lead-lined steel tank were constructed. Caustic-salt solution is prepared a t ground level in a small stainless tank (about 75-gallon capacity) fitted with a Lightnin’ mixer.

Lt‘ater, salt, and flake caustic are fed simultaneously to this small tank, and concentrated solution is pumped as pre- pared into one of the large tanks where it is diluted to ultimate strength. The solution is air lanced, covered with about a n inch of water, and allowed to stand until sufEciently cooled for use.

Sulfuric acid is made up in the lead- lined tank which is fitted with a Tefion air lance for mixing and cooling and with a Hastelloy-C valve and nipple. One low-capacity pump services all rea- gent tanks and delivers reagents through a Carlon hose-Uscolite nipple and flange which attaches to the upper distributor valve of the cell being re- generated. Flow rates throughout the procedure are in general slow.

The elaborate inverse regeneration process described required some 72 hours per cell in plant practice under considerably less than ideal conditions. Resin thus treated was restored to full operating capacity and to a n ash level of 0.54% (96Yo reduction). The con- densed inverse process has been carried out with equally good results in about 44 hours. Although the method appears drastic, resins so treated at 4-month in- tervals for 2 years have shown no dete- rioration. This process should be of value for restoring ion exchange resins poisoned in hydrometallurgical as well as nonmetallurgical processes in general.

Acknowledgment

T h e contribution of Eugene IVood- ward in translating these laboratory results into practical plant operation is acknowledged with thanks.

literature Cited

(1) Dickert, C. T., Preuss, .4. F., U. S. Atomic Energy Comm. Rept. RMO-2532 (March 1955).

(2) Fischer, S. X., Ibid., RMO-2510 (.August 1952).

(3) Higgins, I. R., Zbid., ORNL 54-2-179 (February 1954).

(4) Kennedy, R. H., Zbid., ACCO-31 (February 19531.

(5) Matthew, Dale, Anaconda Copper Co., Bluewater, N. M., private commu- nication, 1955.

(6) Mindler, A. B., Paulson, C. F., Am. Inst. Mining Met. Engrs., Los Angeles, Calif., February 1953.

(7) Preuss, A. F., Rohm & Haas Co., Philadelphia, Pa., private communica- tion, 1955.

(8) Preuss, A. F., U. S. Atomic Energy Comm. Rept. RMO-2523 (June 1953).

(9) Quinlan, K. P., Barry, R. J., U. S. Atomic Energy Comm. Rept. WIN-27 (.August 19561.

(10) Sidgwick, N. V., “The Chemical Elements and Their Compounds,” p. 643, Oxford Univ. Press, London, 1950.

(11) Susman, S., Nachod, F. C., Wood, W., IND. ENG. CHEM. 37, 618 (1945).

RECEIVED for review January 6, 1958 ACCEPTED November 10, 1958

Division of Industrial and Engineering Chemistry, 132nd Meeting, ACS, New York, N. Y., September 1957.

542 INDUSTRIAL A N D ENGINEERING CHEMISTRY