Table II, X-Ray Diffraction Analysis of the Curing Process in Phenolics Cure MaxiPerioHr., mum, dicity, -2 e A. Sample 176' C . 18.79 18.48 18.32 18.24 18.21 18.18 18.18 18.16
0 26.25 113.75 189.0 250.5 294.25 421.75 480.5
4.719 4.797 4.838 4.859 4.867 4.875 4.875 4.880
c
tion and r is the interatomic distance in the resin which gives rise to scattering. The parameter s is related to the Bragg angle by the equation s =
4n
- sin 0
x
The function i(s) is derived from the experimental intensity after correction for polarization, absorption, and incoherent scattering. The method of calculating this curve is described by Klug and Alexander (78).
model of the phenolic resin where aromatic rings joined by methylene groups are assumed coplanar. Lengths of the aromatic and aliphatic C-C bonds are assumed to be 1.38 and 1.54 A., respectively, and the methylene group is assumed to exhibit the tetrahedral bond angle of 109" 28'. Steric hindrance from hydrogens ortho to the methylene group actually prevents coplanarity, and some of the longer distances represented by bars, therefore, are slightly underestimated. Nevertheless, good correspondence generally exists between locations of the bars and maxima of the scattering curve. T h e peak in the radial distribution function at 4.8 A., however. is higher than expected from arrangement of the bars drawn from the model. Therefore, this peak probably represents chain separation in the polymer. The most prominent maximum in polymer radial distribution curves is usually related to interchain distances (74, 78). Moreover, models of phenolic polymers indicate important distances of amroximatelv 5 A. between atoms in neighboring nonbonded phenolic groups. That these intermolecular distances increase with increasing cure time (Table I) is surprising because
shrinkage generally occurs during cure of phenolic resins ( 9 ) . This may be explained by cross linking during the cure. In the early stages of polymerization, the phenolic resin remains liquid and gelation occurs rather abruptly. This suggests that formation of chains rather than three-dimensional networks is favored a t first. The same conclusion is reached from statistical and steric considerations. I n the resin studied, the phenol-formaldehyde ratio of approximately unity is suitable for synthesis of linear structures. Therefore, at the beginning of cure, the resin exhibits the characteristic intermolecular distance observed in models when linear chains are placed side by .side. The curing process proceeds with further cross linking and ramification of the polymer, so that packing becomes less efficient. Gross shrinkage of the resin is attributed to loss of vclatiles and eliminatian of comparatively large lacunae.
Acknowledgment
I I
Discussion The vertical bars of Figure 3 represent interatomic distances for a simplified
The authors are indebted to Donald S. Stang for preparing the resin and to Murray Bloom for preparing the cured samples.
ROBERT A. SPURR, EDWARD H. ERATH, and HOWARD MYERS Research Laboratories, Hughes Aircraft Co., Culver City, Calif. DANIEL C. PEASE Department of Anatomy, University of California, Los Angeles, Calif.
Curing Process in Phenolic Resin
Electron-Microscopic Analysis Existence of micelles is verified by observation, but rather than being units having definite boundaries, they result from differences in cohesiveness
A
FUNDAMENTAL difficulty in the theory of high polymer structure is lack of agreement between theory and experiment in estimating tensile strength. De Boer, for example, has calculated ( 9 ) the theoretical tensile strength for phenol-formaldehyde polymers as 4300 kg. per sq. mm. on the assumption that primary valence bonds are broken in rupture; the value is greater than 39 kg. per sq. mm. if it is assumed that only van der Waals forces are involved. H e quotes an experimental value of 7.8 kg. per sq. mm. Therefore, the resin is less than one fifth as strong as would be expected from a calculation based on weak surface forces. To account for this discrepancy, Houwink (75, 76) adapted to resins the theory of Smekal (23), according to which discrepancies between calculated and observed strengths of crystalline materials
can be explained by the presence of Lockerstellen (structural defects). Near these defects, concentration of stresses leads to the low strengths observed. These defects are thought to arise from the chemical process of condensation, which begins simultaneously at a number of reactive centers. As long as the molecular weight increases by adding small molecules-e.g., dimethylolphenols-a molecule will probably be added at each active site of the growing network. As material of low molecular weight in the vicinity of the network is depleted, the reaction is retarded, because the probability that a large molecule will be properly oriented for addition is slight. The resulting structure is pictured as a collection of dense regions held together by methylene links to form a spongelike skeleton (isogel) having
cavities filled with a viscous liquid condensate of lower molecular weight. Megson ( 7 7 , 79) and others have provided support for this picture by measuring molecular models. An alternative theory, proposed by Stager (25) and others, was developed by studying thin films swollen by acetone. According to Stager, resin molecules build u p into three-dimensional, intermeshed spherocolloidal particles which are embedded in a matrix of lower molecular weight. H e believes there is no need to assume that molecules must be linked by primary (methylene) bonds to give rise to a rigid skeleton; gelation is assumed to result from association and interpenetration of molecules. Stager's model does not, however, adequately account for the abrupt change in physical properties and in solubility experienced VOL. 49, NO. 1 1
NOVEMBER 1957
1839
Figure 4. Shadowed replica o f fracture surface; 480-houi. cure. fracture edges A.
Micellar units, on left, form linear aggregates to center and right
X90,OOO. B.
Bands show where plastic has flaked along
No evidence of micelles here; flne stipple is chromium grain
e
Figure A.
5. Micelle arrangement in shadowed replica o f fracture surface
Radiating fracture lines, 250-hour cure, suggest linear arrangement, X36,OOO
Figure 6. A.
I
Experimental Procedure The previously described phenol-formaldehyde plastic was used. T h e most useful micrographs were obtained from chromium-shadowed replicas of frac-
1 840
Linear aggregates unrelated to fracture lines in this uncured resin, X 36,000
Sections with no evidence of micelles X 90,000 B. Except possibly a t the edge
Except for the light area the left and upper edge, this unshadowed section o f the uncured resin appears homogeneous
when a phenolic resin passes into the C stage, nor for the retention of strain ( 7 3 ) when a resin is cooled under load No matter which of these two theories of phenolic resin structure is preferred, existence of molecular clusters, possibly visible in the electron microscope, must be assumed. These clusters are called micelles, leaving open the question as to whether they are joined to one another by chemical bonds. With the exception of the single photograph by Barkhuff and Reynolds, reproduced by Carswell (%), no direct evidence of micelles from electron microscopy has been presented in the literature.
B.
in the upper right corner, this shadowed section, 250-hour cure, shows no micelles
ture surfaces. and the features discussed are larger than the grain size of the chromium, which is readily apparent at high magnification. It was necessary to adapt customary techniques so that almost microscopic chips of the plastic could be examined. Small flakes of the resin were broken away from the wafers under a low-power binocular microscope. Every effort was made to keep these freshly fractured surfaces clean during subsequent handling. The small flakes were floated on water in a watch glass and then covered with a film of thick Parlodion solution in amyl acetate. The watch glass was heated to hasten hardening of the film, and additional drops of Parlodion solution were added on top of the specimens as needed to build up a sufficiently thick film. The film with the specimens attached then was removed from the water surface and dried. The resin flakes were peeled away from the Parlodion film, which
INDUSTRIAL AND ENGINEERING CHEMISTRY
had formed a replica of their surfaces. The Parlodion replicas were mounted in the vacuum evaporator and shadowed with chromium metal applied at an angle of approximatel>-20 degrees. Following this procedure, a film of carbon was evaporated at right angles to the surface. The Parlodion was dissolved with amyl acetate or acetone, leaving behind the preshadowed carbon replication of its surface. These replicas were examined with an RCA electron microscope. Ultra-thin sections of the plastic samples were also prepared. A PorterBlum ultramicrotome was used with glass knives in the manner pioneered by biologists (27). T o be suitable for electron microscopy, samples of organic materials must ordinarily be substantially less than 0.1 micron thick; it was possible to make such sections of all samples. I t was necessary to keep the block size extremely small (a few tenths of a milli-
E N G I N E E R I N G A S P E C T S OF P O L Y M E R PROCESSES
A A.
Figure 7.
Figure 8. A.
Sections showing basic micellar patterns
Centers of individual micelles seem less dense than interstitial material in this unshadowed section o f the un-
cured resin; X90,OOO E . Shadowed section after 2 5 0 hour cure; X90,OOO
Ribbons flaked off in fracturing;
In this positive replica o f a fracture surface, numerous fracture lines run horizontally and the plastic ribbons remain dangling; X 10,700
6.
edge is a t left;
b
480-hour cure Micellar structure in this ribbon, exposed across a tear, is evident; X90,OOO
b
Figure 9 (left). This shadowed replica shows that when viscous, uncured resin i s streaked on glass, fairly uniform flattened droplets as well as linear aggregates occur
meter), and to change knives frequently. Some sections were examined directly and others were shadowed with chromium to reveal surface structure. Although smooth sections of all of the cured samples were obtained, poorly cut sections having internal disruption and tearing also gave useful views. A resin sample, having no cure or treatment in the press and being extremely viscous, was smeared on a glass slide as thinly as possible. This slide was shadowed with chromium at an angle of approximately 20 degrees and a carbon film was built up on its surface by rightangle evaporation. The resin was removed by acetone and the shadowed carbon replica was transferred to specimen screens. These various methods supplemented each other-an observation made after one method of preparation could usually be confirmed with a second method.
Results As expected, many fracture surfaces were too rough and irregular to provide useful information on basic organization of the plastic. Sometimes, however (Figure 4,A), the surface was textured with globular particles of reasonably uniform size, and at others (Figure 4,B), the surface was nearly smooth. The fine stippling in Figure 4,B corresponds to the grain size of the shadowed chromium and is without significance in the interpretation of the picture. The dilemma occurs where some micrographs indicate a micellar organization while others seemingly deny it. This is explained by the fact that although the plastic is made u p of micellar units, the micelles are not much stronger than the intermicellar zones; thus, the micelles themselves are readily cleavable. Un-
der some conditions they are visible, and under others, not. Measurement for the average size of plastic micelles as seen on fracture surfaces was attempted. Independent measurements by .four investigators have agreed that the average diameter is about 400 A. Assigning limits to the size range is difficult-some larger profiles may not represent individual micelles, and smaller ones may represent micelles so deeply embedded in the plastic that only a portion of their surrace is exposed. Micelle diameters were calculated from measurements made of photographs by two observers (Table 11) and for each datum, about 40 measurements of typical micelles were made. After 480 hours of cure a t 176' C., the average diameter of the samples had decreased. This may have been caused by degradation of VOL. 49, NO. 11
NOVEMBER 1957
1841
the micelles from thermal agitation and oxidation. Whether or not linear aggregates of micelles exist was particularly interesting and suggestions of these often occur near cracks (Figure 4,A). Occasionally surfaces show radiating patterns lvith a high degree of linearity (Figure 5,A). Not much reliance should be placed on such figures, however, because such patterns depend on the forces involved a t the time of fracture. O n the other hand, Figure 5,B is a unique micrograph where presence of linear aggregates unrelated to fracture lines cannot be doubted. But this pattern is rarely seen-perhaps such a high degree of organization is not generally present in the plastic; or linear organization may not impart particular strength to the plastic, so that cleavage occurs at random in relation to the axes of aggregation. I n sectioned plastic, micelles can again be demonstrated only under some conditions. The unshadowed section in Figure 6,A and the shadowed section in Figure 6,B were cut smoothly and no trace of a micellar organization appears, except possibly a t the edges of the sample. O n the other hand: Figures 7,A and 7,B. demonstrate a basic micellar pattern in both unshadowed and shadowed sections. For some samples, micellar patterns appeared when cutting was relatively poor, the knife was dull, or the edge chattered. T h e plastic apparently shattered internally! separating and isolating the micelles. O n the other hand, good sectioning with a sharp knife cleaved individual micelles with a minimum of disturbance and left essentially smooth surfaces which revealed no internal structure. Figure 7,A, particularly revealing, shows a commonly seen t).pe of structure where individual micelles seem to have centers more dense than the interstitial material. Disruption has occurred in the latter zone-possibly by stretching or fracturing the material. I n fractures, a curious phenomenon sometimes occurred where long, thin strips of plastic were separated by cleavage from the main mass. These strips were sufficiently thin to be more or less transparent to electrons and, except for being made without compression distortion that a knife’s edge might produce, they resemble sections (Figure 8,$). They were on the plastic block, and when the Parlodion replica of its surface
Table II.
was made, they were embedded in the Parlodion and carried along with the replica. Being embedded in the Parlodion, they were only partly shadowed by the chromium. After the carbon film was deposited and the Parlodion dissolved, they were left lying freely on the surface. In Figure 8,B, one of these strips is nakedly exposed across a tear in the replica and a micellar pattern identical with that in Figure 7,A appears. Even thinner strips of the original plastic sometimes adhered to the replica in fracture lines (Figures 4,.4 and B ) . Here, internal detail is obscured by the overlying chromium shadowing, but the minimum diameter of these strips corresponds to the diameter of a micelle. Sometimes, individual micelles are completely isolated. The highly viscous resin without press treatment or cure, studied by streaking on glass and replication, suggested micellar aggregation as well. Sizes of individual droplets deposited on glass were fairly uniform (Figure 9) with a n average diameter of approximately 1200 A . as compared with the micellar size in cured plastic of about 400 A . However, these droplets are flattened against the glass to an undetermined extent; therefore, their volume may correspond with rhat of globular micelles of solid, treated resins. These droplets are believed definable structural units-clearly linear aggregates are sometimes visible (Figure 9) and they lie in random directions quite apart from the direction of smear. These aggregates may represent the genesis of the linear patterns shown in Figures 4,A and 5,B. Obviously, from the patterns shown in Figure 9, lateral branching and also lateral aggregation is possible.
Conclusions Phenolformaldehvde resins hake an inhomogeneous structure but micelles are not units with a definite boundary. N o great physical differences occur between the micellar cores and their peripheries because individual micelles can be cleaved or sectioned without difficultv. Thus, micellar structure represents differences in cohesiveness in a continuum of plastic. Data developed by Houwink using the model of a liquid-filled sponge conform generallv with those obtained from electron microscopy and described here. Because differences between Houmink’s and Stager’s theories relate only to char-
Micelle Size
v5.
Cure Time
A ~-
Cure at 176’ C., Hr.
Micelle &am.
0 250.5 480.5
413 427 134
1 842
Observer I Standard deviation 148 53 55
INDUSTRIAL AND ENGINEERING CHEMISTRY
Micelle diam. 420 349 176
Observer I1 Standard deviation 168
62 72
acter of bonding, electron microscopy cannot decide between them. Megson demonstrated by molecular models that, because of steric interactions, the dihydroxyphenylmethanes cannot achieve planar structures. As methylolphenol molecules are added, resin molecules become highly kinked, and active sites on benzene nuclei become inaccessible. Therefore, linkage of large resin molecules into an ordered lattice by methylene bridges is improbable. Megson suggests that fully cured phenolic plastic has few cross links, and an open structure containing numerous voids. Again, this concept is compatible with data from the electron microscope.
Literature Cited CHEM.SOC., Div. Paint, Plastics, and Printing Inks, Symposium on Cure in Thermosetting Resins, 127th Meeting, Cincinnati, Ohio, April 1955. ( 2 ) Am. SOC. Testing Materials, Philadelphia, Pa., ’4STM Standards on Plastics, D 570-42, 1942. (3) Zbid.,D 648-45T, 1945. (4) Ihid.: D 494-46, 1946. ( 5 ) Ibid., D 785-51, 1951. (6) Barkhuff: R. A . , Carswell, T. S.,IKD. EKG.CHEM.36, 461 (1946). (7) Boer. J. H.de, TTans. Faraday Soc. 32, 10 (1936). (8) Carswell, T. S., “Phenoplasts,” Chap. 5. Intrrscience. New York. 1947. ( 9 ) Ibid,,-Chap. 8. (10) Dannenherg, H., Harp. W. K., Jr., Anal. Chem. 28, 8 6 (1956). (11) Finn, S. R., Megson. N. J. L., Whittaker, E. J. W., Chem. @ Ind. 69, S849 (1950). (12) Gardner, H. A , , Sward, G. G., “Physical and Chemical Examination of Paints. Varnishes, Lacquers, and Colors,” p. 395, H. A. Gardner Lahoratory, Bethesda, Md., 1950. (13) Hetenyi, M., J. Appl. Phys. 10, 295 (1939). (14) Houwink, R., “Elastomers and Plastomers, General Theory.” Ch. 6, Elsevier, Amsterdam, 1950. (15) Houwink, R., J . Soc. Chem. Ind. (London) 5 5 , 247T (1936). (16) Wouwink, R., Trans. Faraday Sac. 32, 122 (1936). (17) Hunter, R . F., Vand? V., J . Appl. Chem. 1, 298 (1951). (18) Klug, H. P., Alexander, L. E., “Xray Diffraction Procedures,” Ch. 11; Wiley, New York. 1954. (19) Xleqson, h-,J. L., J.SQC.Chem. Ind. ( k o n d o n ) 6 7 , 155 (1948). (20) Megson, N. J. L.: JVood, W. A., Xature 140, 642 (1937). (21) Porter, K. R., Bloom, J.>Anat. Record 117, 685 (1953). (22) Siegfried, W., Sanger, R . , Stager, H., Kunststofftagung 1940, p. 5. (23) Smekal, A . G., Handbuch der Physik, 2nd ed., vol. 24, pt. 2. p. 795, Springer Verlag, Berlin, 1953. (24) Sofer, G. A , , Dietz, A. G. H., Huscr, E. A , , IND.ENG.CHEM.45, 2743 (1953). (25) Staiger, H., Sanger, R., Siegfried, W.: Helv. Phys. Acta 12, 561 (1939). RECEIVED for review June 24, 1957 ACCEPTED August 30: 1957 Division of Industrial and Engineering Chemistry, Engineering Aspects of Polymer Processes, 131st Meeting, ACS, Miami, Fla., April 1957. (1 )
h i .