1962
Vol. 46, No. 9
INDUSTRIAL AND E N G I N E E R I N G CHEMISTRY
phase is also an important factor, as the separation achieved is dependent on both equilibrium and rat,e factors. The interrelation of these factors can be clearly demonstrated by a MeCabe-Thiele type analysis of a continuous system. A more t'horough investigation of these fact,ors is being made and will be used to correlate the data given in this paper. Figure 14 shows that the maximum concentration of the ethl-1ene glycol effluent is an increasing function of the initial salt concentration. However, the final ethylene glycol concentration is dependent upon bot,h the initial salt and the initial glycol concentrations. Ion exclusion is a relatively economical niet,hod of separation, because, a t constant flow rate, the whole unit can be set up on a t,ime cycle, thus eliminating the necessit'y of constant attention. T h e major operating costs are for pumping and for rinse water. Aside from the lo^ operating costs involved, ion exclusion may be advantageously employed in the separation of pH-sensitive, heat-sensitive, or corrosiITe materials. CONCLUSION
Ion exclusion is no longer just a laborat'ory curiosity. It, has been developed to a point .iT-here it is a potential industrial method for the purification of a number of chemicals. K i t h the technique of operation described here, it is possible to improve the product concentration greatly \vitli no loss in resin-separating capacity. At first thought, t,he volume of liquor t,hat must be
recycled may appear to be a problem, as the forward and backward mixing must be kept a t a minimum. On an industrial scale, this excess volume of recycle liquor may be handled in at least two ways: (1)by using the same technique as in the laboratory-namely, a series of small inventory tanks; ( 2 ) by making a portion of the resin bed act as a storage for the portion doing the separating by installing a series of resin beds properly valved, or by using a single unit with a series of inlets and outlets. There are also other versions or modes of operation that will give similar results. h OM EX C L ATUR E
concentration of solut,ein effluent concentration of solute in feed solution distribution coefficient - concentration of solute in resin phase concentration of solute in liquid phase Ti, = effluent volume V r = bulk volume of resin bed
C,
= C; = Kd =
LITERATURE CITED
(I) KIayer, S. W., and Tompkins, E. R., J .
B7n. Chem. SOC.,69, 2566
(1947).
( 2 ) Simpson, D. IT., and Wheaton, R. X , , Chem. Eng. Prou.. 50,45
(1954).
(3) Wheat,on, R. hl., and Bauman, VI. C., Ann. AV.Y . Acad. Sci., 57, 159 (Nov. 11, 1953). (4) Wheaton, R. AI., and Bauman, W.C., IKD.ESG. CHEII.,45, 228
(1953).
RECEIVED for review January 19, 1954.
ACCEPTEDM a y 5 , 1954. Presented before the Division of Industrial and Engineering Chemistry at t h e 125th Meeting of the AXEEICASCHEXICAL SOCIETY, Kansas City, 1\10.
F. P. REDING AND ALEXANDER BROWN Research and Development Department, Carbide and Carbon Chemicals C o . , South Charleston, W . F'a.
URISG the cryst'allization process of many crystalline polymers the individual tallites form in clusters or aggregates called spherulites ( e , 3: 7 ) . Actually, the Tyord spher. ulite is somevhat of a misnomer, for t,he aggregates are not only spherically shaped but also frequentlv occur as rod- or sheafshaped bundles. These different, shapes of aggregates are i!lustrated in Figure 1, a n electron micrograph of a thin cajt film of polyethylene. Morse and Donna?. (11) have shorn that the first two types of aggregates are simply stages of growth of the truly spherical spherulite. At first the long axes of the crystallites or fibril bundles of crystallites are parallel to one another, but as groivt'h proceeds t,he crystallite fibrils fan out from the ends of the original bundle and form the sheaflike structure and eventually radiate out in all directions and form the tru1:- epherical structure. Although the long axes of the crystallit'e fibrils are parallel to the rod and sheaf axes and to the radii of the spherulite, this does not necessarily mean that the long chain axes of the polymer molecules are similarly oriented. Polychlorotrifluorethylene (fluorothene) is a crystalline polgBier whjch has been shown by Price ( I $ ) to exhibit spherulitic structures when observed betu;een crossed polaroids in a light microscope. Figure 2 s h o w how truly spherical spherulites appear in the polarizing microscope; there are four light qmdrants separated by a black c r o w In the follo~vingdiscussion of the aggregates in fluorothene all three types of aggregatcs are called spherulites.
The fluorothene uqed in the study (CF3-15) had a melt viscosity of 15 megapoises a t 230' C. The figures are photomicrographe of films, 5 mils thick, of CF3-15 or thermally degraded CF3-IS a t 25' C. SPHERULITE STRUCTURE
There has been considerable discussion in the lit'erature concerning the st,ructure of spherulites. From measurements of the refract'ive indices parallel and perpendicular to the radii of spherulites and parallel and perpendicular t o the draw direction of a cold-drawn section of film, Bryant (1 ) has concluded for polyethylene that t,he long chain axes of the polymer molecules must be arranged nearly a t right angles to t,he radii of the spherulites. Similarly, from a comparison of the interference colors observed in the different quadrants of the spherulites in the polarizing microscope using the first-order red interference plate, and a comparison of the colors observed when a cold-drawn section of polymer is similarly inspect,ed, Jenckel and coworkers ( 5 , 6 ) have concluded that the polymer molecules of polyurethanes and polyainides are perpendicular t'o the radii of the spherulitcs. Thus spherulites have been compared to a ball of string. The evidence which leads to this conclusion not only for the above polymers but also for fluorothene is summarized in Figure 3. When a section of cold-drawn fluorothene is placed between crossed polaroids in a microscope and a quartz interferencc wedge is inserted with the fast asis of the wedge parallel t o the draw
September 1954
INDUSTRIAL AND ENGINEERING CHEMISTRY
1963
the polymer chains were rolled up like a ball of string in the spherulites. Figure 5 is a picture of a broken section of thermally degraded fluorothene taken between crossed polaroids. There are several spherulites broken through their centers, lines of cleavage being along the spherulite radii. In addition, there are several cracks in the spherulites, all of which are along the radii. This same phenomenon has been observed in the spherulites of polyethylene. Figure 6 is an electron micrograph of the broken edge of a section of polyethylene. It is clearly seen that the crack has propagated through oriented molecular layers in the spherulite. Thus one is forced to conclude that n-hatever the true structure of the spherulite, the long chain axes must be in such an arrangement that cleavage will take place along the spherulite radii. FACTORS AFFECTING SPHERULITE SIZE
Figure 1. Electron Micrograph of Thin Cast Film of Polyethylene Showing different stnges of spherulite growth, X 12,500
direction of the film, the observed colors go down in spectral order. Therefore, the fast axis of the polymer film is perpendicular to the draw direction and consequently perpendicular to the long axes of the polymer molecules. When a spherulite is observed between crossed polaroids using the first-order red interference plate as shown in Figure 3, quadrants 1 and 3 are yellow and quadrants 2 and 4 are blue. This indicates that the fast optical axis in the spherulite is parallel or nearly parallel to the radii in all quadrants and therefore the polymer chains must be nearly perpendicular to the spherulite radii. The structure of the spherulites in both fluorothene and polyethylene and probably the other polymers as well, how ever, cannot be so simple as is indicated by this procedure. The polymer chains cannot be simply perpendicular to the radii of the spherulites. One indication of this is given by inserting a quartz interference wedge instead of the first-order red interference plate. If the spherulite were like a ball of string, the interference colors in quadrants 2 and 4 would go up in spectral order and the colors in quadrants 1 and 3 would go down in spectral order. However, for polyethylene and fluorothene this is not the c a s e t h e colors go up in spectral order in all quadrants. This indicates the complexity of the orientation of the long chain axes of the polymer molecules in the spherulites. Recently, in order to account for other interference effects that are observed in spherulites, Keller (8, 9) has suggested that the long chain axes of bundles of polymer molecules form spirals or helices and that a number of these helices form a spherulite. If the helical axis is parallel to the spherulite radii, this structure is a t least consistent with all the data. I n the first place, such a structure could give a preferential orientation of the long chain axes of the polymer molecules in the spherulites perpendicular to the radii as illustrated in Figure 4. Three crystallites or crystallite fibrils are shown with their long direction parallel to the radii of the spherulite in conformance with the electron microscope observations. However, the individual polymer chains are nearly perpendicular to the long direction of the crystallites and to the radii of the spherulite. Such a structure would also account for another phenomenon that has been observed in spherulites, that the spherulites break or cleave along the radii as illustrated by the wide black line. This cleavage would be very hard to explain if
When molten fluorothene is cooled below the melting point, crystallization and spheruljtr formation do not, occur immediately. Price ( I d ) has shown that when the resin is cooled from the melt and held a t a temperature b e h e e n 175' and 190' C., first unaggregated or nearly unaggregated crystallites are formed and then these crystalline regions transform more slowly into spherulites. I n this temperature range the spherulites are heterogeneously nucleated and only relatively few of them are formed, but if they are allowed to grow for a long enough time the entire sample will be filled with spherulites which range in diameter from 0.1 to 0.5 mm. Similarly, when the sample is slowly cooled from the melt, only relatively few but very large spherulites are formed and fill the entire sample. Below 175' C. Price has shown that the spherulites are also homogeneously nucleated and that a great many more of them are formed than between 175" and 190' C. Thus as fluorothene is cooled more and more rapidly from the melt, many more, and much smaller, spherulites are formed.
Figure 2.
Truly Spherical Spherulites in Fluorothene Polarizing microscope, X 150
Figures 7 and 8 illustrate the decrease in spherulite size n hen the rate of cooling from the melt is increased. Figure 7 is a photomicrograph of a sample of fluorothene that was cooled from 225O to 100" C. in 0.5 hour. Although the spherulites are somewhat obscured by overlapping, they are in general rather large. Figure 8 shows a sample which was cooled through the above range in 5 minutes; the spherulites in this case are very much smaller. If fluorothene is cooled very rapidly from the melt, water quenched, a large number of small crystallites are formed but the spherulites either are so small as to be unobservable or are nonexistent. The polymer chains become 80 immobile a t room temperature that the crystallization is essentially stopped. Therefore a quenched sample is nonspherulitic and considerably less
1964
INDUSTRIAL A N D E N G I N E E R I N G C H E M I S T R Y
Yol. 46,No. 9
crystalline than an annealed sample. IIofTman ( 4 ) has shown that a rapidly cooled sample is about 35% crystalline, whereas the annealed resin is about SOYo cq-stalliiie.
other. Once a flaw is formed in a spherulite, it, is eaaily propngated t,hrough the long-range regions of onentation and the spherulite breaks in two as shown in Figure 5 . This figure illu~trates the broken edge of a degraded and brittle film of fluorothene. The crack not only travels along the radii of the PCL ..iiOIOS spherulites, but also on occasion travels between two spherulit'es. Regardless of whether the cracks propaga,te BRIEN'rED POLYMER F I L M SPHC 4f u L I T E through or between tlie spherulite*, horvever, the spheru-I-lites constitute easily fractured regions of weakness in the polymer. JYhen the spherulite5 are large, the fracture of a single spherdite means that an appreciable portion of thcl cross section of the sample is broken. \Vork in the Balceiitc~ Laboratories ( 1 0 ) has shoxn that t8he voltage breakdowi potential is relatively low when the resin is slowly cooled from the melt and the I>realido\vn potential decreases w h w quenched resin is heat-aged. These effects almost certainly arise from the fact t,hat the easily fractured regions of orientation provide avenues for the passage of the spark. If the Figure 3. Optical E + i d e n w o f ibei-age Orientation of spherulites Lvere not, present. a single flaw would he propaLong Chain Axes of I ' o l v n i ~ ~ .Wolwrrles in Sphei d i t m gated only the length of the rrystallite involved, a mattc.1 of 100 or 200 A. instead of the 0.01- to 0.5-mm. diaincit(~r of the sphrrulite. As the niolecular weight of the polymer is reduced, the chainThe embrittlement of quenched fluorothene upon heating become more mobile and the crystallization process and spherutween 130" and 210" C. (the melting point) is caused not only lite formation become more rapid, T h e n the resin is highly by the formation of spherulites, but also by the subatantial indegraded, even quenched samples are spherulitic. I n other stallinity. The embrittlenient caused by the inwords, the time required for cry~tallizatioriand spherulite iormacreax in crystallinity, however, is probably not eo great as that tion is smaI1 in comparison to the time of quenching. This is caused by the spherulite formation. Thcre are good indication!< illustrated in Figure 9, which shows an air-quenched sample of that the increase in brittleness in the first €em-minutes of heating highly degraded fluorothene (thermally degraded to a melt, visarises prim;t,rily from the rapid increase in the nuinher or' cry+ cosity of 0.02 megapoise at 230" '2.). The spherulites are large t:t!iites and thnt the extreme embrittlenicnt. upon further hc:lliiiy Undegraded fluorothene reein shows no evidence of spherulitic structures in the polarizing microscope follon-ing the same trentment. A somewhat different situation exists {Then a quenched resin is reheated to a temperature slightly below the melting point. The crystallinity increases very rapidly in the first few minuteof heat,ing and further heating causes orily a moderate additional increase in crystallinity. However, in this further heating, spherdit= are formed, as is demonstrated by the fact Ghat the resin bec,omes opalescent,. The lower molwxlar TTeight or lightly degraded polymers form much larger spherulites (and form then; more rapidly) than the high mo!ecular weight resins. Because of the large number of crystallization centers present, t,he spherulites which grow on heating quenched resins ai'e considerably smaller and probably less perfectly formed than thow grown by sloivly cooling the melt,. This is seen in Figure 10, zi photomicrograph of a quenched sample which v a s subsequently heated for 12 hours a t 190' C.; the individual spherulites appear as small light spots. The magnification of X l 5 0 has been useti i o the spherulites can be compared to those in Figure 7 . At much higher magnification, the individual quadrants of the sphcrulites in Figure 9 can be dist,inguished. Figure '4. &lost Probable Brrangenient of Polymer Chains in Spherulites RELATION BETWEEN SPHERULTTE9 AUD PHYSICAL
i
h i -
PROPERTIES
Large black line inc1icatc.i direction of c l e a r a m
In order t o scatter a nnticeahle amount of light, and cause opal-. sscence, the optical inhornogcneities in a substance must bc, larger than 500 -1.in diameter. In fluorothene the individual crystallites are probably only 100 to 300 A . long and are not largcl enough to cloud the resin. Consequently, a quenched resin which is crystalline but not, spherulit,ic is clear. On the other hand, spherulites very much larger than 500 3. in diameter scatter a considerable amount of light, and an annealed sample mhich is spherulit,ic is hazy. Two other properties of BuoroThene which are re1at)ed to the spherulites are the embrittlement and the decrease in voltage breakdown potential on prolonged heating. Both properties appear to arise to a large extent from the fact that adjacent cryBtallites in the spherulites are oriented nearly parallel to one an-
arises from the formation of spherulites. This is shown in Figure 11; where the increase in crystallinity as measured by the increase in density and the increase in brit,tlenees are shown as funclions of the heat treatment, t,ime. [Although Price ( I d ) ha:^ stated that the amorphous and crystalline portions of Huorothenc probably have nearly the samr densities a t room tcmperature and that density can therelore not be used as a measure of crystallinity, the aut'hors have found that the density at 25" ically, the number and size of the spherulites certainly can be reduced and the undesirable properties minimized. Quenching is one method of reducing the size of the spherulit,es and impairing perfection of the oriented layera within the spherulites. Of course, when the resin is reheated but not melted, the spherulites gron-, but they do not become so large as they would if the resin had not been quenched. Cold drawing t,he resin and orienting t'he crystallites is another means of destroying the spherulites. This process is particularly effective against the heat. emhrittlement of fluorothene tubing, used for \Tire insulation. Upon heating, the cold-dravn, oriented resin becomes just. as crystalline as the unorient,ed polymer, but t'he crystallites remain parallel to each other and the spherulites do not form. Consequently, the cold-drawn fluorothene tubing does not heat embrittle and has a higher use t,emperature than the unoriented or partially oriented tubing. \$'hen an oriented tube is bent, most of the stress is parallel to the long chain axes of the molecules and the resin is very resistant to breaking. Highly oriented tubing does become fibrous under extreme bentling, but with normal usage this would not be serious.
h
Clarity, heat emhrittlement, and voltage breakdown of fluorothene are all closely correlated with the size and perfection of the spherulites in the resin. The spherulite formation is a iesult of the mode of crystallization, %hereas the crystallinity is an inherent property of the resin. If the polymer is modified so that a substantial reduction in crystallinity occurs, the spherulites can, oi courfie, be eliminated, but many of the desirable properties of the resin will also be sacrificed Although spherulites are normally formed, there is no basic reason why thev cannot be eliminated n-ithout reducing crystallinity. Even if the ideal
6 ' E r5
217,
el6
211
SUGGESTIOVS FOR MINIMIZIhG SPHERULITE SIZE
Vol. 46, No. 9
--t
DENSITY V S TIME
--eBRITTLENESS
I
VS
TIME
..' 0 e
4
e
8
10
L
2
T i m OF HEATING. H O U R S AT ISO'C.
Figure 11. Increases in Crystallinity (Density) and Brittleness of Fluorothene as Functions of Heating Time The spherulite formation is a phenomenon of the mode of crystalliiation rather than an inherent property of the polymer. I t would be most desirable to find some physical treatment which ITould induce a large number of spherulites toformsimultaneously, so that the individual spherulites could not hecome large hefort: their growth is blocked by other crystalline regions. Quenching the resin accomplishes this l o a certain extent, but it is only partially effective against both heat embrittlement and opalescence. Other treatments, such as the use of finrly divided inert fillers or other miscihle resins, might prove very effective nithout sacrificing the desirable propeities of the resins. The aim here would be tmofold: to find an extraneous material that vould induce more nucleation and/or act as a physical block to the formation of large spherulites. COSCLUSlON s
Figure 10. Spherulites in Air-Quenched Film of Fluorothene Subsequently Heated for 12 Hours at 190" C. Polarizing microscope, X 150
The structure of spherulite8 is such that cracks are propagated along the spherulite radii. The spherulites are responsible to a large extent for the lack of clarity, embrittlement, and low voltage breakdown potential of fluorothene. These undesirable properties could be eliminated wthout sacrificing any of the desirable properties of the resin if the spherulites could be eliminated without substantially reducing the degree of crystallinity. Several methods of reducing the size and perfection of the spEc~-ulites are discussed.
September 1954
INDUSTRIAL AND ENGINEERING CHEMISTRY ACKNOWLEDGMENT
The authors wish to thank E. R. Kalter for taking the photomicrographs used in this paper. LITERkTURE CITED
(1) Bryant,
W.M . D., J . Polunao Sci., 2, 547 (1947).
(2) Bunn, C. W., and Alcock, T. C., Trans. Faraday SOC.,41, 317 (1946). (3) Gabler, R., .~atzLrwissenscha/ten, 35, 284 (1948). (4) Hoffman, J. D., J . Am. Chem. Soc., 74, 1696 (1952). (5) Jenckel, V. E., and Klein, E., Kolloid Z., 118, 86 (1050).
1961
Jenckel, V. E., Tiege, E., and Hinrichs, W., Ibid., 129, 19 (1952). Jenckel, V. E., and Wilsing, H., 2.Elektrochem., 53,4 (1949). Keller, A , Nature, 169,913 (1952). Ibid., 171, 170 (1953). (IO) Maddock, B. H., private communication. (11) Morse, H. W., and Donnay, J. D. H., Am. Mineralogist, 21, 391 (1936). (12) Price, F. P.,’J. Am. Chem. SOC.,74, 311 (1952). RECEIVEDfor review May 19, 1953. ACCEPTED May 3 , 1054. Presented before the Division of Industrial and Engineering Chemistry, Symposium on Fluorine Chemistry, at the 124th Meeting of the AMERICAN CHCXICAL SOCIETY, Chicago, Ill.
Calculation of Liquid-Liquid Extraction Processes Correction I n the article, “Calculation of Liquid-Liquid Extraction ProcESG. CHEM.,46, 16 (1954)], esses,” by Edward G. Scheibel [IND. some of the curves for batchwise extraction with fresh solvent in Figure 4 were incorrectly plotted, and the revised figure is shown helow. These corrections affect the calculations described in the first column of page 19, so that the continuous operation is slightly better than originally indicated. Beginning with line 9, the paragraph should read as follows: According to the lover family of curves, the value of
LD
must
have been 0.58 and the average value of D was therefore 2.9. If this recovery was satisfactory, it could be accomplished in five continuous countercurrent stages a t a value of E = 1.20, whence
L
= 0.41.
Thus, the same job can be done with 41% of the
total solvent if employed in a countercurrent manner. Also, if the original amount of solvent was employed, E would be equal to 2.9 and the recovery would then be better than 99.5%. I n addition, all of the Rz terms in Equations 11 to 14 on page 23 should be R;.
IO
9 8 7 6
5 4
3 2
E 1.0
.e .6
.S .4
.3 .2
Figure 4.
Graphical Representation of Equations 1 and 6
