Synthetic-Resin Ion Exchangers in Water Purification Operating Characteristics
.
ROBERT J. MYERS AND JOHN W. EASTES The Resinous Products and Chemical Company, Ino., Philadelphia, Penna. The operating characteristics of two typical ion-exchange resins have been studied in an apparatus designed to evaluate the probable performance of the commercial products in full-size exchanger beds. The study has proved that the ion-exchange resins are stable in acid, neutral, or alkaline solutions, have high capacity, and can be regenerated with a high degree of chemical efficiency. The capacity of the cation exchanger which may be used in the hydrogen as well
OLLOWING the discovery of Adams and Holmes (I) in 1935 that certain synthetic resins possessed ion-exchange properties, a number of publications and patents (7) have appeared dealing with ion-exchange resins. Whereas the older exchange adsorbents were of a siliceous or humus character, and the more recently developed carbonaceous zeolites (derived from the sulfonation of coal, lignite, peat, etc.) were based upon natural organic substances of poorly defined chemical constitution, the synthetic-resin ion exchangers offered for the first time a nonsiliceous character combined with “tailor-made” physical and chemical properties to any desired extent or degree, as a consequence of the well characterized chemical nature of such products. In the field of water purification, the newer cation exchangers (such as those derived from modified phenol- or tannin-formaldehyde resins) are of special interest in view of the higher operating efficiencies, better thermal and mechanical stabilities, and freedom from silica pickup when employed in industrial and domestic water-sof tening. In 1939 the Resinous Products and Chemical Company became the sole American licensees tinder the basic patents of Adams and Holmes (1). Nearly two years have been spent in extensive research on synthetic resinous exchangers to bring the materials to successful commercial application. Two resins, suitable for use in the water purification field, have been developed and are known as Amberlite IR-1 and Amberlite IR-4. Amberlite IR-1 is a cation-exchange resin which operates either in the sodium cycle (to replace hardnessDroducine: ions bv sodium) or in the hydrogen cvcle (to remove ill cation;). G b e r l i t e IR-4 is an inion-exchinge ‘adsorbent which is used primarily with Amberlite IR-1 in the hydrogen cvcle to remove all dissolved salts from water. “ To evaluate the probable behavior of these two ion-exchange resins when employed in the purification of water, a study was made of the performance of typical exchanger beds
a s the sodium cycle was found to be independent of particle size and hardness of water, and constant over long periods of repeated use. Preferential adsorption of certain ions was noted. The anion-exchanger resin exhibited a high degree of chemical efficiency in anion removal, combined with capacities of an extraordinary magnitude. Rates of regeneration and operation are of significance in the operation of the anion-exchanger units.
under a variety of conditions simulating those found in commercial practice. Whereas Seyb (II), Richter (IO), and Griessbach (6) have reported on the performance of ionexchange resin installations in boiler feed-water purification plants in Europe, no study of a semicommercial installation under a variety of operating conditions such as those encountered in this country has yet been made. This paper covers the first phase of the investigation.
F
Apparatus and Materials An apparatus of larger dimensions than the usual laboratory equipment was constructed for the purpose of investigating the behavior of the resins under varying rates of flow, conditions of regeneration, and other variables pertinent to the operation of large size exchanger units. This apparatus (Figure 1) consisted of glass cylinders which contained the resin bed, with fittings of S/d-inch hard rubber pipe and rubber hose, the supply tanks bein rubber-lined or having a resistant coating. Valves consisted woodworker’s clamps on the rubber hoses. This construction was variable enough so that any connection could be made or changed in a few minutes. Cylinders up t o 8-inch diameter and any hei ht can be used for the resin bed. #he first study in this a paratus was that of Amberlite I R 1 in the calcium-sodium cycg, using a resin bed of 0.0909 cubic foot volume which filled the 4inch column to a height of 12.5 inches. The column of resin was supported by a perforated plate of phenol-formaldehyde resin covered with a, 50-mesh Monel metal wire screen. The resin used was substantially of -20 40 mesh (U. S. Staadard) particles. This size was selected as representative of commercial exchangers which do not 5 v e too much resistance to flow. The character of Amberlite IR-1 is such that i t a capacity for ion exchange is independent of particle
07
+
size (8).
Sodium-Calcium Exchange In this studs the resin was saturated with calcium bv flowing over it a ;ohtion of calcium chloride of known strength, taking samples of the effluent at appropriate intervals, and analyzing them for calcium. In this way the break-through
1203
1204
INDUSTRIAL AND ENGINEERING CHEMISTRY
point and concentration changes after the break-through point were determined. Thereafter the resin was regenerated by flowing over it a solution of sodium chloride of known strength] taking samples of the regenerate effluent a t intervals] and analyzing for calcium. In this way a complete curve for the regeneration was mapped out. UPSLOWREGENERATION. The effectiveness of different concentrations of salt in upflow regeneration was tried, using rates a t least sufficient to maintain the bed a t full teeter. Table I shows data pertinent to these regenerations.
TABLE I. Regeneretion No.
% NaCl Used
‘CTPFLOW REGENERATIONS
Sample Size, Lb.
Rate of Flow Gal./Sq. Ft./M;n.
% Bed Expansion
As Figure 2 shows, there was considerable difference in the effectiveness of salt solutions of different concentrations. The 4 per cent solution was found to be most efficient. Figure 3 shows the relation of salt concentration to concentration of calcium in the regenerate effluent and indicates that a mass-action relation exists. Figure 4 shows the efficiency of different salt concentrations in terms of capacities attainable at different salt values. The discontinuities at the beginning of some of the curves is due to the headroom and consequent dilution of solution in and above the resin by the water already present. The extent of this dilution effect is more noticeable a t the higher salt concentrations. INTERMITTENT REQENERATION. To avoid the dilution effect and to study the effect of the time of exposure to the regenerant, a series of experiments were run in which the regenerations were carried out in a manner termed “intermittent flow”. This consisted of draining the resin by gravity and running into it by upflow an amount of salt solution which, when the bed settled, would a little more than cover the resin. After standing for a time, usually 30 minutes, this solution was drained and the process repeated with a fresh portion of salt solution. In this way the points shown in Figure 5 were obtained; times of exposure, other than 30 minutes, are indicated. That they all lie on the same curye shows that 30 minutes was ample time of exposure for the salt solution. The mass effect of concentration is well illustrated by Figure 5, especially by the 4 per cent curve in which the discontinuity is caused by a switch to 18 per cent salt solution after partial regeneration with the 4 per cent solution. Figure 6 shows the relation between salt used and calcium removed. The 13 per cent solution was most efficient. In Figure 7 the data are plotted in terms of capacity attained a t different salt values and reveal that sometimes the 11 per cent solution was most economical in salt consumption. Table I1 shows other data concerning this series of regenerations. DOWNFLOW REGENERATION. In operating an exchanger bed by downward flow there is no motion of the particles as in upflow operation. The entire bed remains stationary with
TABLE11.
Vol. 33, No. 9
INTERMITTENT-FLOW
Regeneration No.
yo KaCl Used
11 12 13 14 15
10.50 7.27 3.66 18.24 12.60
REGENERATION
Sample Size, Lb.
4 4.5 4 4 4
Hours of.Contact 1/2:2:15 1/2 1/2:1 1/2:1 1/2
the liquid flowing down through the spaces between the resin particles without turbulence and thus diminishing the dilution of fresh incoming salt solution by water or spent salt solution. This has the effect of applying fresh salt solution continuously to the most highly regenerated portions of the bed in order, which is an advantage over intermittent flow in addition to a saving of time. The data for a series of experiments using downflow regeneration are shown in Table I11 and Figures 8, 9, and 10. These results show by comparison with previous data that downward flow is the most efficient and practical method of regeneration, and the salt concentration should be 4-7 per cent. TABLE 111. DOWNFLOW REQENERATION Regeneration No.
% NaCl
Sample Size, Lb.
Used
Rate of Flow Gal./&. Ft./M\n.
0.83 1.17 1.32 1.38 2.66
Using the results obtained with 4 per cent salt solution, the following behavior of Amberlite IR-1 may be expected: Capa?ity, Grains CaCOs/Cu. F t . Resin
8,400 10,000 12,200
FIGURE
Lb. NaCI/ Kilograin CaCOa
0.37 0.40 0.48
Lb. NaCI/ Cu. Ft. Resin
3.5 4.0 5.5
Capacity, Greins CaCOa/Cu. Ft. Resin
Lb. NaCI/ Kilogram CaCOa
14,400 15,500 29,800
1. FOUR-INCH EXCHANQER UNIT
0.59 0.65 2.72
Lb. NaCI/ Cu. Ft. Resin
8.2 10.3 81
I N D U S T R I A L AND E N G I N E E R I N G C H E M I S T R Y
September, 1941
L
I
I
I
I
I
1205
of water softened, wash water requirements, labor, and time out of service. This partial regeneration is sometimes termed “bed starvation”. One of the disadvantages of partial regeneration or bed starvation (4) has been that for some types of exchanges the capacity of the unit diminishes with each successive regeneration in spite of the fact that the same quantity of salt was used in each partial regeneration. For some exchanges this diminution of capacity stops after several cycles and levels off to some steady value which is lower than that given by the first partial regeneration. In other exchanges, however, this diminution of capacity does not stop but continues until the capacity is so low that it is no longer profitable to operate. At this point the unit is treated with a large quantity of salt which restores its original capacity SO that partial regeneration may be resumed. In order to examine Amberlite IR-1 under partial regeneration, two pojnts were selected on the 4 per cent salt curve of Figure 10. Fourteen cycles were carried out on the first point, four cycles on the second point. Tables IV, V, and V I show the values calculated from the 4 per cent curve and those actually obtained in the fourteen cycles. Inasmuch as the Amberlite IR-1 showed no diminution in capacity upon repeated partial regeneration and since one field unit has gone two hundred cycles without diminution of capacity, it is possible to conclude that the resin shows a constant capacity under any partial regeneration and that the capacity is dependent upon the extent of regeneration. Other points brought out by these tables are that the resin takes up substantially as much calcium to the break-through point as was removed by the preceding regeneration, and that the break-through capacity is independent of the water hardness. Many exchangers do not possess these properties.
I
9000 8000 7000
0“ eQ00 U
d U
TABLE Iv.
15,800-GRAIN PARTIAL REGENERATION 4 % NrtCl 7% NaCl (Actual) ka%a (Actual) 16,800 16,600 15,800 Capacitz, grains CaCOs/cu. ft: resin 10.8 10.4 11.0 Lb. Na 1 required/cu. ft. resin 0.70 0.663 0.66 Lb. NaCl used/kilomain CaCOa removed 7 effioienoy of s<-u6ed 24 26.2 24 L$. NaCl repuired for bed 0.9s 0.948 1.00 Lb. CaCOs required .by bed for 16,800 grains/cu. ft. oaprtcity 0.205 0,2020 0.206 Values,taken as the average from Table VIA, obtained from a 4.9% NaCl solution.
1 5000 a: 0:
4000 3000 2000
1000
1.0
FIGURE
7.0 3.0 4.0 5.0 POUNDS N a C l USED
100 50
6.0
30
PFRC€NT - . IO
20
FFflClfNCY .
5
2 (Above). REMOVAL OF CALCIUM BY UPFLOW REGENERATION
3 (Below). CONCENTIUTION OF CALCIUM EFFLUENT DURING UPFLOW REGENERATION
FIGURE
IN
PARTIAL REQENERATION. In the actual operation of exchanger units it is not customary or economical practice to regenerate the bed to its highest capacity after each run to exhaustion because] in common with other exchange materials as shown by the curves for Amberlite IR-1, the amount of salt required to regenerate to maximum capacity would make the operation impractical. In contrast, a partial regeneration, although producing a lower capacity, is practical because of the greater efficiency of the salt used. The extent to which a unit is regenerated is dependent upon several competing economic factors, among which are cost of salt, amount
I
I
0.6
LO
1.5
I
I
’
2.0
2.5
POUND5 NaCZ PER KILOGRAIN C a C 0 5
F I G U4.~ EFFICIENCY OB UPFLOW REOENE~RATION
1
BE
Vol. 33, No. 9
INDUSTRIAL A N D E N G I N E E R I N G I C H E M I S T R Y
1206
TABLE V.
6600-GRAIN
PARTIAL R.EGENERATION
Capacity rains CaCOa/ou. f t . resin Lb. NaCiAu. f t . resin Lb. NaCl/kilo rain CaCOs removed % e5cienoy of;salt used Lb. NaCl required for bed Lb. CaCOs in bed
Calcd. from 495 NaCl Curve 6600 2.2 0.34 49.0 0.20 0.086
24000
Found for 5.2% NaCl (Table VIB) 7060 2.3 0.32 52.1 0,209 0,092
?low 18000
,
m
TABLE
Regeneration No.
Lb. NaCl Used
VI.
PARTIAL Lb. NaCI/ Kilograin CaC03
Lb. CaC08 A.
21
0.962
22
0.870
23
0.910
24
0.945
25
0.870
26
0.950
27
0.870
28
0.864
29
1.009
30
0.943
31
1.020
32
0.950
33
0,998
34
0.995
Av. for all
0,948
35
0.163
36
0.207
37
0.217
38
0,229
39
0,267
40
0,211
Av. for all
0.219 0.209
Capacity, Grains
%,Ef- CaCOa/Cu.
fioiency
0.1840 0.1823 0.1868 0.1898 0.1921 0.1903 0,2030 0.2050 0.2106 0.2140 0.2047 0.2025 0.2068 0,2140 0.2100 0.1982 0.1962 0.2140 0.2036 0.2135 0.2013 0.2163 0.2080 0.1860 0.2100 0.2015 0.2020 0,2020
01607 0.591 0.684 0.664 0.614 0.600 0.577 0.588 0.729 0.735 0.630 0.662 0.683 0.723 0.628 0,637 0.767 0.679 0.706 0.704 0.663
0,0815 0,0792 0,0941 0,0950 0,1035 0,1037 0,1005 0,1047 0,1208 0,1210 0,0892 0.0882 0.0984 0,0917
...
27.5 28.3 24.4 25.1 27.2 27.8 28.9 28.4 22 .9 22.7 26.5 25.2 24.4 23.1 26.6 26.2 21.8 24.8 23.6 23.7 25.2
Run No.
Ft. Resin
15,800-Grain Regeneration 22.2 0.752 14,160 22.2 14,030 0,753 25.1 0.666 14,380 25.5 0.655 14,620 24.7 0.677 14,790 24.4 0.683 14,650 25.1 0.666 15,830
B.
2
REGENERATSON DATA
6600-Grain Regeneration 0.260 64.2 6260 0,295 56.5 6098 0,314 53.2 7245 0.311 53.7 7315 0.299 55.7 7970 0,299 55.7 7980 0.325 51.3 7740 0.313 53.2 8050 0,316 52.7 9300 0.316 52.7 9320 0.338 49.3 6870 0.842 38.8 6770 0.319 52.3 7570 0.821 52.0 7060
This emphasizes the distinction between “break-through” capacity, at which hardness first appears in the effluent t o a measurable extent, and “total” capacity, a t which the effluent has the same hardness as the water being softened. Thus Amberlite IR-1would be well suited for use when completely softened water was demanded. M ~when partially ~ soft ~ water ~ only was required, IR-1 would have the advantage of requiring only a simple hardness test to determine when the unit was exhausted, instead of elaborate testing as required by exchange materials which deliver only partially softened water and gradually approach total capacity by delivering water of increasing hardness. Figure 11 shows that 95 per cent or more of the capacity of Amberlite IR-1is used up in delivering water of zero hardness and thatless than 5 per cent of the total capacity remains after the break through, under widely different conditions of regeneration and water hardness.
P.P.M. CaCOa Grains Used CaCOs i n R u n in Bed
22
ld,78O 16,220 16 480 15:760 15,580 15,920 16 480 16:170 15,260 15,370 16,480 15.670 16,450 15,500 16,660 16,020 14,330 16,170 15,520 15,550 15,550
l5000
d
486
23
507
24
440
25
495
26
507
27
477
28
523
29
498
30
502
31
565
32
127
33
49
34
1445
35
3445
.,
.. .
36
592
37
592
38
592
39
500
40
500
41
517
1288 1276 1307 1328 1344 1333 1421
...
1433 1474 1498 1432 1417 1448 1498 1470 1387 1373 1498 1425 1495 1411 1514 1491 1302 1470 1410 1413 14130 628 554 659 665 725 726 704 732 845 847 624 617 688 651
2.0 3.0 4.0 POUNDS NaCl USED
0.5 1.0
5.0
FIGURE5 (Above). CONCENTRATION OF CALCIUM IN EFFLUENT DURING INTERMITTENT-FLOW REGENERATION FIGURE 6 (Below). REMOVAL OF CALCIUM BY INTERMITTENT-FLOW REGENERATION
i
TABLE VII.
REGESERATIOS
19, DOWNFLOW WITH 4 P E R CEXT SODIUM
CHLORIDE
P. P. 31. CaCOSin Effluents 16,890 10,550 ~ 6,070 4,290 3,080 2,450 2,030 1.810 1,310 1,070 980 677 875 654 537 537 495 326 214 131
Lb. NaCi 0.20 0 .~ 40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 5.20 3.80 4.40 4.80 5.40
Total Lb. CaCOa 0.0844 0.1372~ 0.1675 0.1890 0.2044 0.2166 0.2269 0.2340 0,2414 0.2468 0.2517 0,2361 0,2595 0.2628 0.2633 0.2682 0.2756 0,2815 0.2840 0.2869
Lb. NaCI/Lb. CaCOs 2.35 2 . 9 1, 3.58 4.23 4.88 5.53 6.17 6.82 7.45 8.09 8.74 9.38 10.02 10.63 11.30 11.92 13.78 15.62 16.90 18.79
Lb. NaC1/ Kilograin CaC08 0.3386 0.417 0.51? 0.60~
0.698 0.790 0.879 0.972 1.OG2 1.156 1.249 1.360 1.429 1.520 1.618
1.703 1.969 2.230 2.415 2.685
yqEffiaency 49.3 40.0 32.6 27.6 23.9 21.1 19.0 17.2 15.7
14.4 13.3 12.2 11.r 11.0 10.3 9.8 8.5 7.5 6.9 6.2
cg$,y’ CaCOz,’ CU.
Ft.
6,500 10,380 12,900 14,550 15,730 16,700 17,490 18,100 18,580 19,020 19,380 19,700 19,980 20,250 24,050 20,680 21,200
21,700
21,880 22,100
1 Cu. Ft. Reein Lb. Lb. x a c 1 CaCOs 2.2 0.929 4.4 1.508 6.6 1.843 8.8 2.079 11.0 2.249 13.2 2.383 15.4 2.495 17.6 2.583 19.8 2.657 22.0 2.719 24.2 2.765 26.4 2.818 28.6 2.854 30.8 2.891 33.3 2.921 35.2 2.950 41.8 3.032 48.4 3.097 52.8 3.124 59.4 3.156
£ samples of effluent were analyzed. 4
INDUSTRIAL AND ENGINEERING CHEMISTRY
September, 1941
The foregoing data demonstrate that the capacity of Amberlite IR-1 is dependent only upon the extent of regenerationi. e., by the amount of calcium removed in the regenerationand that we may use the data obtained for the regeneration curves to calculate the capacity of the resin at any salt value. It is convenient to derive the mathematical equations for these relations. PPRCeNT
30
10
EFFICIENCY
9
IO
I
0.5 1.0 1.5 2.0 POUNDS NaCl PER KILOGRAIN CaCOj
I
2.5
I
1207
HEADLOSSES. Figure 16 indicates that resistance to downflow is a linear -function of the rate of flow (actual values are determined by the past history and condition of the bed). RINSEWATERREQUIREMENTS. After an exchanger bed is regenerated, it contains a mixture of sodium and calcium chlorides in solution, which must be washed from the bed before the unit is put into service again. Obviously the amount of rinse water, time, and effort required for this operation is a factor in the economy of operation of the unit. The removal of salt solution from Amberlite IR-1 requires no soaking or leaching, but demands only enough water to meet the mechanical requirements of dilution, diffusion, etc., t o remove the salt. Rinsing the unit by upflow operation requires more water than downflow, owing to the greater volume of salt solution present (freeboard); 26.4 gallons of water per cubic foot of resin were required to wash out all the sodium chloride from
I
FIGURE7. EFFICIENCYOF INTERMITTENTFLOW REGENERATION Plotting the data for the 4 per cent curve (Table VII) of Figure 9 on semilogarithmic paper gives a straight line (Figure 12) over the useful range of salt used. This portion of the curve is described by the general equation Y = nlogX
+b
Transferring these data to a one-cubic-foot basis gives the curve of Figure 13, where: Y = Ib. CaC08 removed or adsorbed/cu. f t resin = lb. NaCl used/cu. ft. resin C = capacity, kilograins CaCOa/cu. ft. resin
X
Comparison of the capacity of Amberlite IR-1 for the adsorption of calcium with that of other water softeners shows (Figure 14 and Table VIII) that IR-1 has higher capacities on any basis of comparison. If the comparison was on a weight basis, IR-1 would appear even better when compared to inorganic exchangers, for only 28-30 pounds of dry resin are needed per cubic foot of wet exchanger bed. The true density of wet Amberlite IR-1 is 1.7 grams resin per cc. (109 pounds per cubic foot). BACKWASH RATESAND BEDEXPANSION.The low density of synthetic resins has the advantage of permitting a low backwash rate to give full bed teeter to classify and remove dirt from the bed. This is of special importance where only low water pressure is available. Since the relation between rate of flow, bed expansion, and pressure for an exchanger unit is dependent not only upon the material and its grade in the softener bed, but upon temperature of water, construction of distributing systems, etc., the data given here must be interpreted with these qualifications. Figure 15 shows the relation between rate of flow and per cent bed expansion. The difference in pressure required for a rate of 0 to 5.5 gallons per square foot per minute was of the order of 0.25 inch of mercury, which indicates that the resin bed was giving practically no resistance to flow.
I
I
I
I
I
I
I
I
OF CALCIUM IN FIGURE 8 (Above). CONCENTRATION EFFLUENT DURING DOWNFLOW REGENERATION OF CALCIUM BY DOWNFremi 9 (Below). REMOVAL FLOW REGENERATION
INDUSTRIAL A N D ENGINEERING CHEMISTRY
1208
TABLEVIII.
-
COXPARISOX OF AMBERLITEIR-1
WITH
Vol. 33, No. 9
OTHER PRODUCTS
7 A m b e r l i t e IR-1Commercial Product Grains Lb. Lb. Grains Lb. CaCOa/ NaCl/ NaCl! CaCOa/ NaC1/ ou.it. cu. f t . kilogram cu.ft. ou.ft. resin resin CaCOa Name resin resin 10,000 4 0.40a Greensand (3) 3,800 4 6,000 4 4 0.40Synthetic (pptd.) zeolite ( 5 ) 10,000 4,000 4 0.4O5 Same 10,000 4 8,000 4 0.40' Synthetic (fused) zeolite (3) 10,000 4 Clay group (S) 5,000 4 10,000 4 0,405 8,000 4 4 0.40" Same 10,000 6 0.49" Gel zeolite ( 8 ) 7,500 6 12,500 10,000 6 0.490 Same 12,500 6 2,600 0.96" 0.37 Greensand (9) 3.14 8,400 2,800 1.260 Same 0.46 11,400 5 . 1 a 4,000 1.405 Synthetic zeolite (9) 0.35 7 200 2.5a 5,500 2.750 Same 0.50 12:700 6 . P Carbonaceous zeolite (9) 7,000 2.45 0.35 7,200 2 . 5 a 2,800 0.40 Zeolite (6) 1.12" 10,000 4= 5,900 4 . 0 Carbonaceous ieolite 4 0.405 10,000 a Values calculated from given data.
the unit in downflow rinsing. The calcium disappeared from the rinse effluent almost a t once. The major portion of the rinse was soft and was required merely t o remove the sodium chloride or chloride ion. This is in accord with the studies described below.
Lb. NaCl/ kilo rain cab02 1.05a 0.675 1.005 0.50'
0.805 0.50a 0.80" 0.605
0.37
0.45 0.35 0.50 0.35
0.40
0.68"
FIGURE 11. BREAK-THROUGH AND TOTAL CAPACITY 0 indicates strength of
solution used.
Adsorption of Calcium in the Presence of Sodium Since a resin saturated with calcium can be regenerated to the sodium form by use of a sufficient concentration of sodium ions, it is obvious that the presence of a sufficient amount of sodium ions in a hard water would prevent the calcium from being adsorbed by the resin. Since all natural waters contain some sodium salts, it is important t o have some knowledge of the influence of different concentrations of sodium ions on the adsorption of calcium by cation exchangers that are t o be used in the sodium form for water softening. PERCENT 100 50
30
20
obtained previously in the absence of sodium ions. Moreover, the high concentrations that are tolerated before the resin fails to soften a considerable portion of the water to zero hardness are noteworthy; in addition, a t the very high concentrations the resin is still removing some of the calcium. This behavior clearly shows the possibility of using the resin to purify or recover traces of metals from solutions which contain a high concentration of a second ion.
EFFICIENCY IO
9
FIGURE10. EFFICIENCY OF DOWNFLOW REGENERATION
This was determined on the cation exchange resin, Amberlite IR1, through the use of synthetic waters containing known amounts of calcium and sodium chloride or sodium hydroxide. The resin bed was regenerated in a standard manner. Figure 17 shows the results obtained on several mixtures of calcium and sodium chlorides. As would be predicted, the presence of sufficient sodium ions hinders the adsorption of calcium ions; but the significant point revealed in this work is the high concentration of sodium ions tolerated before the capacity of the resin for calcium adsorption is diminished a t all, below the value of 12,700 grains per cubic foot,
FIGURE12 (Above). SEMILOG GRAPHOF REGENERATION OF 0.0909 CUBICFOOTOF AMBERLITE IR-1 WITH 4 PIUR CENTSODIUM CHLORIDE Equation of straight line: Y 0.166 log X 4- 0.204 FIGURE13 (Below). SEMILOG GRAPHOF REGENERATION OF 1 CUBICFOOT OB AMBPRLITE IR-1 WITH 4 PERCENTSODIUM CHLORIDE
.
-
Equation of straight line: Y C = 13.02 log X
1.86 log
4-2.170
X
+ 0.310:
September, 1941
.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Figure 18 shows the stoichiometric relations that hold during adsorption of calcium in the calcium-sodium cycle. These data indicate that the chloride concentration of raw water and effluent are the same, that the total cation 810 concentration of raw water and ef16 fluent is the same, 8 and that for a t?I? given amount of s" calcium adsorbed, J8 an equivalent quantity of so$4 dium ion is put in its place. Thus the term "base exmwvs N ~ C IPER KI~OGRAINc.coI change" is a true FIQUW14. COMPARISON OF EFFIdescription of the CIBNCIES wat e r-so f t e n i n g action of the resin.
8
Effect of pH on Calcium Adsorption Certain cation exchange resins, such as Amberlite IR-1, can be regenerated with strong acid to the hydrogen form or with sodium carbonate to the sodium form without any harm to the resin; therefore we have no reason to fear the action of water of any pH 5 00 on the resin. It is zdW of interest, howP ever, to ascertain c 3w the influence of DH k on the actual ada 1.00 3 / -COMPLETE I)W MOTION sorption. L * 10% x PARTIAL BED MOTION As shown below, A Irn PI I t the effluent from the resin in the IO IO 30 PO IO 60 70 hydrogen PERCENT BED EXPANSION - - cycle - has a pH as low as 1.8, FIQURBI 15. RATBOF UPWARD FLOW and therefore uersu8 BEDEXPANSION cium was taken out of water a t this pH. To test the effect of high pH, a water containing 500 D. I). m. calcium carbonate was made up to pH 9.03 by the addition of sodium hydroxide (20 p.-p. m. NaOH as calcium carbonate required). The capacity of the resin for calcium a t this pH was unchanged from its previous value.
..
0
s
1209
removed from acid solutions, etc. The stoichiometry of ion adsorption in the hydrogen cycle by $10 Amberlite IR-1 is illus: trated by Figures 19 and 20. Figure 19 indicates that calcium is completely adsorbed to a high capacity, the adsorbed calcium being replaced by an exactly equivalent quantity of hydrogen ion which PRrSSVRE, INCHE5 MERCURY gives rise to a constant FIGURE 16. RATB OF DOWNpH of solution up to WARD FLOWversus PRESSDRH~ (7' C. OR 44.0' F.) the break-through point. After the break-through point the hydrogen-ion concentration and pH of solution vary in accordance with the amount of calcium being adsorbed. Figure 20 shows the same relation to hold for the adsorption of sodium ions. In this work the calcium in the effluent was determined by soap titration, pH with the glass electrode, and liberated acid by titration with standard sodium hydroxide solution The quantitative relation which exists between the cations adsorbed and the liberated acid affords an accurate and convenient method of determining the amount of total cations in a mixture, or the amount of single cation in a pure solution. Through the use of appropriate laboratory-size adsorption columns several different cations have been quantitatively determined. The change in pH and acid concentration with the amount of cation being adsorbed affords an accurate and convenient method of checking the operation of the exchanger in the hydrogen cycle. I n the hydrogen cycle the resin bed is regenerated by flowing over it a solution of an acid, usually sulfuric or hydrochloric. In general, the same relations of concentration effects and efficiency prevail in regeneration with acids as do regenerations with sodium chloride in the sodium cycle. Figure 21 illustrates the results obtained by using different concentrations of hydrochloric acid on the resin bed saturated with calcium, and indicates a 4 per cent solution of hydrochloric acid to be satisfactory. The inefficiency at the start of the 4 per cent curve was due to a slow start, and consequent dilu-
Hydrogen Cycle The ability of synthetic-resin cation exchangers to be regenerated by acids to give a stable hydrogen exchanger, which can exchange hydrogen for all metallic ions and thereby reduce the solids content of the water, is one of the outstanding advantages possessed by this type of exchanger over mineral exchangers. The ability to operate in the hydrogen cycle opens up many possibilities in the water treatment and recovery fields hitherto not possible. For instance, carbonates and bicarbonates can be completely eliminated from water, the p H of water can be adjusted by appropriate blending of treated and untreated water, metals from dilute solutions can be recovered uncontaminated with alkali metals, trace metals oan be
0.02
0.04
0.06 0.08 0.10 POUNDS C ~ C OEXPOXD ~
0.12
ro RESIN
0.14
0.6
0.18
F~omt~l 17. ADSORPTION OB CALCIUM IN TEE PRESENCE OB SODIUM Run Number Water Used C a + * a s p . p. m. CaCOs (CaCln) m. CaCOs (NaC1)
%ttl,""&,G
42
43 622
44
45 47 48 49 . -_ 513 634 613 529 531 284 507 960 1516 1989 6061 12120 5976 1 a 2 4 11 22 21
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
1210
10.8
'1 3k
80
whereas solutions of salts such as ammonium sulfate are "split" to a considerable extent. For this reason, t o remove all ions from solution, the order of treatment is first with Amberlite IR-1in the hydrogen form and then with Amberlite IR-4.
PER CUBIC FOOT RESIN
KII0GRAINS CaCOs
12.3
Vol. 33, No. 9
13.8
PERCENT
Lu
IO0 50
z
60
8
EFFICIENCY
30
16
u3 12
p: p:
(L
%!
. . . I .JI
I.. . I . .,I 0.08
0.10 0.12 POUND5 CaCO,
0.14 0.16 0.18 EXPOSED TO R€5lN
FIGURE 18. ADSORPTIOKO F CALCIUM OF SODIUM (RUN42)
$ 8 I
020
IK THE PRESENCE
Water used: 513 p. p. m. CaCOa (CaCla), 507 p. p . m. CaCOa (RTaCl), 1020 p. p. m. CaCOs (total cations), 1020 p . p. m. CaCO3 ( C Y ) . X = total C1- and/or total cations ( N a + Ca++) 0 = sodium ions @ = calcium ions
+
tion in the freeboard above the resin bed and diffusion in the resin bed. One precaution t o be observed in regeneration in the hydrogen cycle is that the acid used for regeneration must form soluble salts with the cations adsorbed on the resin. The effluent from the hydrogen cycle of a hydrogen-exchange resin, such as Amberlite IR-1, is lower in solids content than the original water through the complete removal of metallic cations, fromwhich the carbon dioxide can be removed by aeration. Treatment z p H OF EFFLUENT of this acid effluent Im by an anion exchanger, such as A m b e r l i t e IR-4, removes all of the anion and gives an effluent which is comparable to high-quality distilled water.
9 2 4
a
3
2.5
0.5 1.0 1.5 2.0 POUND5 NaCl PER KILOGRAIN CeCO,
FIGURE 21. EFFICIENCY O F REGEKERATIOK OF AMBERLITEIR-1 WITH HYDROCHLORIC ACID FROM THE CALCIUM FORM R a t e of flow, 1.4 gallons per square foot per minute.
The capacity of Amberlite IR-4 for anion adsorption is somewhat dependent (8) upon the concentration of the anion in solution. Therefore, for a point of reference, 500 p. p. m. acid was used in the work reported here. The particle size of the resin used was -20 +40, a practical size for low resistance to flow. The capacity of Amberlite IR-4 is essentially independent of particle size over the range of commercial sizes (8). I n the activated wet form the resin bed used contained 908 grams (2 pounds) of dry resin and occupied a volume of 0.100 cubic foot. The apparent wet density was 0.32 gram per cc. (20 pounds per cubic foot), and the true wet density was 1.26 grams resin per cc. (79 pounds per cubic foot).
L
I \ -IZd0 wW(0s CeCOs E~fUIfD TO RESIN
FIGURE 19 (Above). ADSORPTION OF CALCIUM BY AMBERLITE IR-1IN THE HYDROGEN CYCLE Break-through capacity, 15.3 kilograins CSCOI per cubic f o o t resin.
FIGURE 20 (Below). ADSORPTION OF SODIUM BY AMBERLITEIR-1 IN THE HYDROGEN CYCLE Break-through aa acity 13.7 kilograins C ~ C O per~ c u t i c foAt resin.
$
I:
Adsorption of Anions Amberlite IR-4 removes from solutions those anions, except carbonate, which are in their acid forms, but does not extensively remove anions from solutions of neutral salts. Thus solutions of sodium chloride are but s l i g h t 1y "split"
d
-LOO
2of
L
J
I
j
~
I,
::
4
PPM HCI
"
J "
30t0.04
Ob
Oh
Oh
0.30 Pi0 " P e U H W Hg504 FXfV5ED TO RESIN
060
OF HYDROFIGURE 22 (Above). ADSORPTIOI~ CHLORIC ACID (500 P. P. M.) BY AMBERLITE IR-4
Break-through cagacity, 30 2 kilograins CaCOa per cubic f o o t resin.
FIGURE 23 (Below). ADSORPTIONOF SULFURIC ACID (500 P. P. M.) BY AMBERLITEIR-4 Break-through capacity, 43.5 kilograins CaCOi per cubic foot resin.
INDUSTRIAL AND ENGINEERING CHEMISTRY
September, 1941
PERCENT €FFlClENCY 100
50
30
100
50
30
POUND5 N d 1 PER KILOGRAIN &COJ
25. FIGURE 24. EFFECT OF RATE FIGURE
EFFECT
OF CONCENTRATION OF OF FLOW ON EFFICIENCY OF REGENERATION OF THE CHLORIDE SODIUM CARBONATE FORMOF AMBERLITE IR-1 SOLUTIONON EFFIWITH A 4 PERCENTSOLUTIONCIENCY OF REOENERATION OF THE CHLOOF SODIUM CARBONATE RIDE FORM OF AMBERThe numbers indioate the rate of flow a8 gallons per square foot per LITE IR-4 -
minute.
Rate of flow,0.18 gallon per square foot per minute.
When Amberlite IR-4 is converted from the free base to the acid salt by adsorption of acid, followed by regeneration, the bed volume changes (“breathes”). Thus the resin, when in the chloride form, occupies a volume 28 per cent greater than when in the activated state. Amberlite IR-4 exhibits a preferential adsorption of one kind of ion over another; sulfate, for instance, has the preference over chloride. This is analogous to the preferential adsorption of cations by Amberlite IR-1. The high quality of effluent produced from acid solutions by removal of the acids with Amberlite IR4 is illustrated in Figures 22 and 23. These figures illustrate the complete removal of chloride and sulfate to extremely high capacities of the resin, a sharp change in anion concentration and pH OCcurring at and after the break-through point. This sharp change in pH affords a convenient means of detecting the breakthrough point by use of suitable indicators or the glass electrode. The figures also show the extremely high capacity of the resin for anion adsorption and the preferential adsorption of sulfate over chloride.
Regeneration o f Amberlite IR-4 with Sodium Carbonate Amberlite IRA can be regenerated with alkalies such as sodium hydroxide, sodium carbonate, and ammonium hydroxide. Data presented here pertain only to regeneration with sodium carbonate solutions of various strengths. Only downflow regenerations were considered, which were carried out in a manner similar to those for Amberlite IR-1. In the regenera-
1211
tion of Amberlite PPRCENT EFFICIENCY 30 IR-4 with sodium 100 50 c a r b o n a t e solution, the rate of flow (Figure 24) zzB greatly influences the efficiency of re8 24 generation, slow rates of flow being the most efficient. In this and other figures the rate, 0.18 g a l l o n p e r square foot per minute, is the equivalent of 30minute exposure of 2 a given portion of solution to the bed in static state. 0.10 0.50 1.00 After 30-minute POUNDS NOCZ PER KlLOGRklN CrCO3 exposure this FIGURE 26. EFFECT OF CONCENTRAtionwasfor~edout TION OF SODIUM CARBONATE SOLUTION AND R.Al’E O F FLOW (IN PARENwithafresh wortion which t h e n reTHESES ~ 1 -GALLONS 3 PER SQUARE FOOT PER MINUTE) ON THE EFFImained for 30 minCIENCY OB REOENERATION OF THE All o t h e r CHLORIDE FORM OF AMBERLITEIR-4 ratesrefertosteady flow. As Figure 25 shows, the concentration of regenerant solution also affects the efficiency of regeneration. Part of this effect can be attributed to the fact that a higher concentration at a given rate of stream flow inherently produces a greater rate of application of active material. However, Figure 26, while illustrating the balance of effects produced by changes in concentration and rates of flow, reveals that at a given rate of application of active material, concentration produces some of the effect on efficiency indicated by Figure 25. From these data on the regeneration of Amberlite IR-4 from the chloride form with sodium carbonate solutions, it is evident that in operating practice a choice must be made. A practical rate of flow and a concentration of solution must be chosen which will give a high efficiency at this rate of flow yet require only a reasonable time for the application of the desired amount of active material to the resin bed. Thus, in
FIGURE 27. COMMERCIAL INSTALLATION USINGAMBERLITEIR-1
INDUSTRIAL AND ENGINEERING CHEMISTRY
1212
general, a 2 per cent solution of sodium carbonate would be most practical. One of the most striking characteristics of this resin, in contrast with exchange material in general, is its extremely high capacity for adsorption and the high chemical efficiency of regeneration. As might be expected, the rinse water requirements of Amberlite IR-4 are dependent upon the rate of flow used, the high rates of flow being rather inefficient. In general, at a flow rate of 1 gallon per square foot per minute, the equivalent of 60 gallons of water per cubic foot of resin was required. Philadelphia tap water was used for rinsing downflow. Complete removal of sodium salts from the resin bed after regeneration by sodium carbonate solution is necessary; otherwise these neutral salts will appear in the effluent during the beginning of the next run. Figure 27 is a photograph of a commercial installation for water treatment in the powerhouse of a large chemical manufacturing plant. The two units are 7 X 4 foot rubber-line tanks with stainless steel fittings. At present both units are filled with 30-inch beds of cation exchangers, one with Amberlite and the other with a carbonaceous cation exchanger for comparison. Both are operating in the sodium cycle to give completely softened water from a raw water whose average hardness is 3.0 grains (51 p. p, m.) of calcium carbonate. The Amberlite IR-1 unit has operated successfully at rates of 1to 9 gallons per square foot per minute, the head loss (downflow operation) of the unit being substantively 1 pound per
Vol. 33, No. 9
square inch a t these rates. The raw water pressure is 60 to 70 pounds per square inch. The performance of the two units is still under study and will be reported in detail in a later paper. To date, the resin exchanger has given excellent performance.
Literature Cited Adams, B. A., and Holmes, E. L., J . SOC.Chem. I d . , 54, 1-6T (1935); Brit. Patents 450,308-9 (June 13, 1936), 474,361 (Nov. 25, 1937) ; French Patents 796,796-7 (April 15, 1936) ; U. S. Patents 2,104,501 (Jan. 4, 1938). 2,151,883 (March 28, 1938). 2,191,853 (Feb. 27, 1940).
Am. Water Softener Co., unpublished data. Babbit and Doland, “Water Supply Engineering”, p. 593 (1939). Collins, L. F., J . Am. Water Works Aasoc., 29, 1472 (1937). Davis, D. E., Ibid., .29, 1517 (1937). Griessbaoh. R., “fiber die Herstellung und Anwendung neuer Austauschadsorbienten, inbesondere auf Harzbasis”, Berlin, Verlag Chemie, 1939; Angew. Chem., 52,215 (1939); Melliand Temtilber., 20, 577 (1939).
Myers, R. J., Eastes, J. mi., and Myers, F. J., IND.ENQ.CHEIM., 33, 697 (1941).
Myers, R. J., Eastes, J. W., and Urquhart, D. J., Ibid., to be aublished. 018on, H. M., Ohio Conf. Water PurZfication, 18th Ann. Rept., 1938, 98-101.
Richter, A., Angew. Chem., 52, 679 (1939) : Melliand Textitber., 29, 579 (1939).
Seyb, E., Chem. Fabrik,1940, 30. PRUSENTUD before the Division of Water, Sewage, and Sanitation Chemistry at the l O l s t hleeting of the American Chemical Sooiety, St. Louis, Mo.
CORRESPONDENCE Effect of Sodium Sulfate on Alkali Resistance of Portland Cement TDA mixture in cement was shown by Kenned a t o promote better dispersion with consequent improved griniability, increased strength, and better setting qualities. Siliceous materials have been shown to increase the resistance of concrete to alkali and the cement can be changed to the double sulfoaluminate. In sea water attack, although such mixtures usually set slowly. this way the expansion, which accompanies the double salt forOn the other hand numerous admixtures advocated to inmation, will occur before the cement has hardened, and specicrease the strength of cement or to impart special properties, mens made from such a mix might be sound under certain condias alkali resistance, waterproofness, or workability, have such tions in contact with sulfate solutions. been tested and found to be worthless for the purpose claimed Unfortunately, however, the sulfoaluminate forms crystals or to decrease strength to such a degree as to nullify any advantwhich continue to grow in certain orientations with changes in age gained in other respects. temperature and moisture content. Such growth is very apt to Although sodium sulfate is among the materials cause disintegration of conc.rete.1 The sodium hydroxide set regarded as objectionable in concrete mixes, the resu?ttg% free by interaction with the lime from the hydrated cement may show that addition of sodium sulfate markedly improved alkali react in many unfortunate ways, including painful attack upon resistance and that tensile strength was also increased especially the epidermis of the applicators. at early ages. All cement contains some alkali and ’is more or It is hoped that these facts will suggest the desirability of makless caustic, and only trial will show whether the mixture is “hot”. ing adequate tests before investing much money in portland Even then the workmen can be protected. cement-sodium sulfate mixes. The references cited by Anderegg do not show the effect of usF. 0. ANDEREGG ing sodium sulfate, as he apparently used calcium sulfate. In Newark, Ohio our experiments only sodium sulfate improved either strength or alkali resistance. The mechanism of stabilization advanced * * . e by Anderegg does not explain why other soluble sulfates, which would likewise react with the aluminate, do not stabilize the ceSIR: Anderegg emphasizes the general attitude toward the ment. It seems doubtful, therefore, that a simple reaction of sulpresence of sodium sulfate in cement and concrete. Effloresfate with the calcium aluminate suffices t o explain the results. cence, “hotness” or causticity, and weakened concrete are some of The authors entirely agree that further investigation should be the evil effects often described as caused by sodium sulfate. Usumade before investment of much money. ally alkali resistance is secured by controlling the cement comAn error in the caption of our Figure 1 (page 692) has been position and limiting the alumina content. noted; 10 per cent was the strength of the sodium sulfate soluMany attempts have been.made t o improve the quality of tion instead of 4 per cent as printed. concrete by admixture of various compounds to the cement or PHILIP H. DELANO concrete mix, and some of these have proved beneficial for specific purposes. For example, the use of small amounts of 12061/$ Riverside Drive SIR: By the addition of sodium sulfate to portland cement,
as proposed by Delano and Weber [IND. ENG.CHEM.,33, 692 (1941)] it is true that most of the tricalcium aluminate found in
1 Anderegg, F. O., Proc. Am. Concrete Fab. 28. 1931.
I n a t . , 23, 332 (1929); Rock Producta,
Tusoaloosa, Ala. 1
IND. ENQ.CHEX., 28, 963 (1936).