Raman Microimaging of Polymer Blends

Raman Microimaging of Polymer Blends R. APPEL, T. W . ZERDA,* and W. H. WADDELL Department of Physics, Texas Christian University, P.O. Box 289940 Fort Worth, Texas 76129 (R.A., T.W.Z.); and ExxonMobil Chemical Company, 5200 Bayway Dr., Baytown, Texas 77520 (W.H.W.) Raman m icroimaging was used to estimate the effect of the silica Žller on phase separation in binary polymer blends com posed of brominated poly(isobutylene-co-para-methylstyrene) (BIM S) and cis-1-4-polybutadiene (BR). The domain sizes, relative concentration of polymer components within domains, and distribution of particulate silica Žller and zinc stearate curative were characterized for blends of different compositions and history of aging treatments. The presence of increased concentrations of precipitated silica results in better polymer morphology since domain sizes are reduced. Increased temperature treatment also decreases domain sizes up to about 150 8C, but treatment at 200 8C appears to induce separation of the elastomer com ponents. Silica is usually found near the cen ters of the BIMS domains. There is little difference in silica distribution before and after curing. Index Headings: Raman spectroscopy; Raman imaging; Polymer; Filler distribution. INT RODUCTIO N Raman spectroscopy is an analytical method that has undergone an impressive growth. In the past, this technique required highly skilled operators and expensive, fragile instrumentation. The continuing developments of diode lasers, holographic Žlters, and gratings and the efŽciency of CCD (charge-coupled device) detectors transformed Raman spectroscopy into an efŽcient analytical technique. Finally, the addition of optical m icroscopes coupled with the use of Žber optics has shown m any advantages of Raman methods over other types of analytical techniques. In particular, use of a confocal microscope enables the imaging of speciŽc chemical species at a spatial resolution of less than one micrometer and a depth of Želd of as little as one m icrom eter.1 Recently, we applied Raman m icroimaging to study the polymer morphology in binary mixtures of two elastomers: brominated poly(isobutylene-co-para-methylstyrene) (BIMS) and cis-1-4-polybutadiene (BR). 2 This initial study demonstrated that Raman microimaging can be used to characterize the internal structure of elastomer blends. Now, we report the distribution of BIMS and BR binary blends in silica-Žlled systems. Formulations comprising BIMS, BR, silica, accelerator, zinc stearate, antioxidants, and other compounding ingredients were examined. The goals of this study were to (1) characterize the sizes of the polymer domains in the presence of the silica Žller; (2) evaluate the distribution of the particulate silica Žller, zinc stearate, and other detectable additives within each phase of the binary blend; (3) identify any effect of aging on the distribution of BIMS within blends of different compositions; and (4) compare the distribuReceived 3 April 2000; accepted 2 August 2000. * Author to whom correspondence should be sent. Volume 54, Number 11, 2000 tion of com ponents in the uncured and cured rubber samples. EXPERIMENTAL Raman spectra were obtained by illuminating the sample with an argon laser operating at the wavelength of 514 nm and the power of 10 m W. The laser beam was focused by a 1003 objective, numerical aperture (NA) 5 0.95, of a confocal microscope (Olympus BH2), and the diffraction-limited width of the laser beam focused on the sample was less than 0.4 mm in diameter. 2 The lateral resolution is also affected by the size of the pinhole in the back focal plane; in our case it was a circular pinhole with a diameter of 0.5 mm. To determine the spatial resolution, we used Raman microimaging to m easure the size of latex spheres of precisely determined diameters, 2.02 mm, supplied by E. Fuller. For the 0.5 mm pinhole the lateral resolution was less than 1 mm . The scattered light was collected by the same objective, and the spectral analysis was accomplished with an axial transmissive sp ectro graph (K aiser O ptical S y stem s, H olo S pec) equipped with a Princeton Instrument CCD camera. Use of high-dispersion gratings allowed recording of Raman spectra within the frequency range from 150 to 2400 cm 2 1 . The spacing between two pixels corresponded to 1.5 cm 2 1 , and the spectral resolution was determined by the width of the slit, 50 mm, which resulted in broadening of atomic lines of neon and argon to about 5–7 cm 2 1. For studying the composition of the rubber sample, the position of the specimen with respect to the incident beam must be very carefully controlled. The precision of the x-y-z positioning stage also affects the spatial resolution. The computer-controlled x-y-z table (Oriel, Encoder Mike) was capable of advancing the specimen in either direction in steps as sm all as 0.1 mm . However, the measured effects were m uch larger, and only negligibly small changes in the polymer distribution were observed when the sample was displaced in those very small 0.1 mm steps; thus larger steps were used. Each sample was scanned in two different areas: 20 mm 3 40 mm with the use of steps of 2 mm , and 10 mm 3 10 mm in 0.5 mm intervals. Larger area scans did not provide new information and will not be discussed. For selected samples, linear scans were obtained, all greater than 200 mm long and in several cases approaching 500 mm . At each point Raman spectra were recorded for 30 s. Note that it is difŽcult to record spectra for very long run times without suffering interruptions. For example, when the alignment of the m icroscope is accidentally changed or a sample is moved, the imaging experiment m ust be restarted. Another problem encountered with rubber samples is that of laser heating. If care was not taken, the laser beam would burn a small hole in the rubber specimen when exposed 0003-7028 / 00 / 5411-1559$2.00 / 0 q 2000 Society for Applied Spectroscop y APPLIED SPECTROSCOPY 1559 for an extended period of time. These data were discarded. Use of sm all pinholes is necessary to achieve good resolution along the optical axis.3–5 Unfortunately, use of sm all pinholes reduces the amount of light gathered by the detector, which necessitates longer exposure times. To shorten the duration of individual experiments while still maintaining high spatial resolution, we prepared selected samples in the form of thin sections, 1 mm thick, which allowed the use of a pinhole 0.5 mm in diameter and a relatively short exposure time of 30 s. Contour maps obtained with large pinholes for thin sections were similar to those obtained for the bulk systems with small pinholes. Blends of two commercial polymers were studied: BIMS, a brominated poly(isobutylene-co-para-methylstyrene) elastom er from ExxonMobil (Exxpro 8745), and BR, a polybutadiene predominantly in the cis conformation (.98%) from Goodyear (Budene 1207). Different blends were prepared, and their compositions are listed in Table I. The numbered samples indicate the binary blends consisting only of polymers and silica; and lettered samples represent silica-Žlled rubber compounds containing other additives to protect or cure the compound. Sam ples A, B, and C were m ixed in three stages with the use of a Banbury (size BR) internal mixer at 60 rpm. The Žrst stage consists of adding elastomers and m ixing for 30 s, adding silica, and mixing an additional 90 s, and adding processing aids such as oil and m ixing for a total of 300 s. The second mixing stage mixes the Žrststep masterbatch for 60 s, adds the zinc stearate, and m ixes for 180 s total. The third stage mixes the second-step masterbatch and curatives for 150 s. All other samples simply used a Žrst-stage mix. Sam ples #3 and #4, which do not contain additives, were subsequently heat treated at 100, 150, and 200 8C by pressing the blends inside the platens, under 50 psi pressure for about 20 m in. Sam ples A, B, and C were studied as uncured and cured samples. For these samples, we determined the morphology of the two polymers, the silica Žller distribution and zinc stearate location. Analysis of the distribution of other compounding ingredients has not yet been achieved due to their relatively low concentrations or due to the presence of overlapping Raman bands. RESULTS AND DISCUSSIO N Raman Spectra. Absolute intensities of Raman bands are extremely difŽcult to calibrate. Raman intensity depends on the population of scattering m olecules, intensity of the incident laser, sensitivity of the detector, incident angle, scattering geometry, light collection angle, sample surface roughness, absorption of light by optical elements (such as lenses, diffraction gratings and Žlters), and exposure time. M ost of those param eters are very difŽcult to calculate, and in most cases it is impossible even to estimate them. For this reason, Raman spectroscopy is not used to determine the absolute concentration of one component in a mixture, which is a typical application of IR spectroscopy. However, the relative concentration of species can be easily determined. For example, in the 1560 Volume 54, Number 11, 2000 case of a binary mixture it is sufŽcient to identify two Raman bands well separated from neighboring bands, if each band is characteristic of only one component. Their relative intensity ratio is independent of laser uctuations, scattering geometry, etc., and is proportional to the concentration ratio of the two components in the binary m ixture. 2,3 To analyze the relative concentration of the elastomers, we selected the following Raman bands: BIM S, the 714 cm 2 1 band assigned to the rocking m ode of the m ethylene g roup s; cis-1-4 -p oly butadiene, the C 5C stretching vibration centered at 1648 cm 2 1; precipitated silica, the broad band centered at about 488 cm 2 1 ; and zinc stearate, the band centered at about 1118 cm 2 1 and assigned to the bending mode of the –C–O–C– group. These peaks are identiŽed in Fig. 1, which represents typical spectra for the blends under study, before and after curing. These peaks are well isolated from their neighbors, and their intensities can be readily determined. To improve the accuracy of the results, we calculated the intensity ratios using integrated values over the frequency range from (nmax 2 15) cm 2 1 to (nmax 1 15) cm 2 1, where nm ax is the frequency of the band maximum intensity. The integrated intensities were calculated after background subtraction. All intensity ratios were calculated with respect to the integrated intensity of the band due to the cis-polybutadiene. Thus, the data represent the relative concentration of various components to the concentration of the cis-polybutadiene. The m easurement errors vary. For the BIMS/BR ratio, the experimental error is estimated to be less than 15%, but for the silica/BR and the zinc stearate/BR intensity ratio, errors exceed 20 and 25%, respectively. Blends with Silica and Curatives. After the sample is cured, the intensity of the C5C stretching mode of cis1-4-polybutadiene decreases. The reduction in the 1648 cm 2 1 band intensity and the form ation of a shoulder at 1604 cm 2 1 are results of the curing process. These changes can be explained in terms of crosslinking, which reduces the population of the C5C bonds and concomitant formation of m olecular species having vibrations at slightly different frequencies. The intensity ratios discussed here were obtained in reference to the intensity of the 1648 cm 2 1 band. Similar results were obtained when a different butadiene band centered at about 1280 cm 2 1 , which does not change signiŽcantly during the curing process, was used in the normalization procedure. However, since we wanted to monitor changes occurring during the curing process, we focused on the C5C band. Intensity Calibration. Even in binary m ixtures, BIM S and BR do not mix well, and for the same specimen Raman intensities vary from one scattering location to another, because of local uctuations in concentrations of the components. However, for a long scan, in which hundreds or thousands of readings are recorded, the average intensity ratio depends only on the initial concentrations of the polym ers. Indeed, as seen in Fig. 2, the average values of the I BIM S /I B R ratio for the 30:70, 40:60, and 50: 50 BIMS/BR blends yield straight lines of different slopes for the cured and uncured samples. We used linear equations to Žt the data in Fig. 2, which were later used to calculate the BIM S concentration in units of phr (parts per hundred weight of rubber) from the intensity ratios obtained during the measurem ents. The results of those F IG . 1. Typical Raman spectra for the BIMS/BR blends with silica, zinc stearate, thiate U, and other additives. The spectrum was corrected to give a at background across the spectral range. Small letters indicate peaks used to calculate relative concentration of various components. (a) The peak at about 490 cm 2 1 is assigned to silica; (b) the band at 714 cm 2 1 is due to the rocking CH 2 mode of the BIM S backbone; (c) the band centered at 1118 cm 2 1 is assigned to hydrocarbon chain vibrations of zinc stearate; (d ) the peak at 1648 cm 2 1 is assigned to the C5C stretching vibrations of the cis-polybutadiene backbone. calculations are shown in Fig. 3, which illustrates distributions of BIM S in uncured and cured samples. This procedure was limited only to the BIM S concentration calculations. We could not use this method to cal- culate the concentration of silica since only two different silica contents (30 and 45 phr) were used, which was not enough to provide sufŽcient conŽdence in the calibration function. Therefore, in Fig. 3 the silica distribution is characterized only by providing relative intensities of Raman bands due to silica and BR. In the experiments reported in this study, the concentration of zinc stearate was constant and equal to 5 phr; consequently no attempt was made to calibrate the intensity ratios of Raman bands due to zinc stearate and BR. Polymer Morphology. In Fig. 3 we show contour m aps for the 50:50 BIM S/BR blend before and after curing. Similar results were obtained for other blends. These maps depict only the distribution of the BIMS and BR elastomers. The polymer domain sizes do not change as TABLE I. Compositions of polymer blends studied by Raman spectroscopy. Sample composition (phr) F IG . 2. Average Raman intensity ratio of the peaks due to BIMS and BR plotted against the phr ratio of the two components. Circles represent data for the cured blends, and squares for the uncured blends. The solid and broken lines represen t best linear Žts to the experimental data. In both cases the root mean squared value was better than 0.1%. The equations represen ting these linear Žts were used to calculate the BIMS concentration in the units of phr. a Sample BIMS BR Zinc stearate Silica Oil, Flexon 766 1 2 3 4 5 6 7 Aa Ba Ca 0 25 50 75 50 50 50 40 50 30 100 75 50 25 50 50 50 60 50 70 – – – – – – – 5 5 5 30 30 30 30 10 50 70 45 45 45 10 10 10 10 10 10 10 10 10 10 A, B, and C contain a different oil and other curatives. APPLIED SPECTROSCOPY 1561 1562 Volume 54, Number 11, 2000 a result of the curing process. Regardless of the composition of the samples, whether or not cured, the measured domain sizes are less than one micrometer in diameter. An accompanying atomic force m icroscopy (AFM ) study identiŽed sm aller domains with diameters of about 0.3 mm.6 This discrepancy may be attributed to a better resolution of the AFM technique and aggregate formation. It is important to understand that some of the domains form aggregates, which can be large, and in some cases exceed 5 mm. With increasing concentration of the cispolybutadiene in the sample, the BR domains appear to be slightly larger than the BIMS domains, but the effect is small and within experimental error. M apping experiments discussed in this paper essentially characterize distributions of components within a surface layer 1 mm thick. The blends prepared with silica and other additives contained 45 phr of silica by weight, which corresponds to about 15% by volum e. For the BIM S component, the calculated volum e contributions for samples A, B, and C were 38, 30, and 23%, respectively. Assuming that the volume distribution is identical to the surface distribution, in Fig. 4 we overlapped the two surface maps shown in Fig. 3, one depicting silica distribution and the other BIMS and BR for the cured sample A. Silica is shown in black, BIMS in red, and BR in white. The cut-off concentrations for the silica, BIMS, and BR were found experimentally by matching the areas covered by different components with the values predicted from the volum e concentrations. Although this Žgure does not provide information on the domain sizes, it illustrates well the locations of silica domains, and allows identiŽcation of regions rich in the BIMS and BR elastomers. Comparing Fig. 4 with Fig. 3, it is evident that the single polymer phase domains are rare, and most of the domains are a m ixture of both components. As seen in Fig. 3, their relative contributions vary. The fact that BR is present throughout the sample can be interpreted as a result of good penetration of this component into the BIM S domains. However, an alternative explanation is also possible. The polymer blends were studied in the form of thin sections, 1 mm thick, and it is possible that two different domains were present as distinct layers in the section. In this scenario, laser light could penetrate through the top domain and enter the domain below. As a result the detector would register signals originating from two different single-phase domains. At this stage of our research it is impossible to verify which interpretation is correct. Additional studies are planned to verify whether single-phase domains exist in the BIM S/BR blends. Silica Distribution. The contour maps in Fig. 5 depict silica distribution. Different colors indicate different absolute values of the intensity ratios of the peaks assigned to silica and BR. For clarity reasons, we did not label F IG . 6. Distribution of BIMS in the binary blends with BR, samples #2 (bottom), #3 (center), and #4 (top). The concentration of silica was 30 phr. each color for individual graphs and provided only the maximum ranges in the Žgure captions. Precipitated silica appears to be preferentially adsorbed in, or in close proximity to, the domains enriched in BIMS. The regions with the highest concentrations of silica are usually 1 mm removed from the center of a BIMS domain. Actually, for many of the BIMS domains, a decreased concentration of silica is observed at the center of the BIM S dom ain; the maximum silica concentration never coincides with the center of the domain. Silica is also found to reside within the BR domains. This effect is especially pronounced for sample C, which has the highest concentration of cis-polybutadiene. It is important to note that there are areas within the samples where the concentration of silica is so small that it is undetectable and it appears to be essentially absent. Because of experimental error resulting from low signal-to-noise ratios, we cannot unequivocally conclude whether in those domains the silica concentration is just very small or the silica is completely absent. In the cured samples, silica is always within or near the boundaries of the BIM S dom ains. Prior to curing of the compound, silica is also present in the BR domains. However, after compound curing, the amount of silica found inside the BR domains is drastically reduced, indicating desorption of the BR from the silica surface and adsorption of BIM S onto the silica surface. Zinc Stearate. Zinc stearate is an ionic species and can dissociate in rubber mixes. In this experiment we monitored only Raman bands due to the C–O–C vibrations of the fatty acids. Thus, Fig. 5 illustrates the distribution of the stearate chains. Consequently, no inform ation is available on the distribution of the zinc ions and/or of any zinc oxide. It is seen that stearic acid concentrates in small areas, which appear to be less than 1 mm in di- ¬ F IG . 3. Contour maps for the 50:50 BIMS/BR blend with silica and other curatives before (left column) and after curing (right column). The lower maps depict distribution of BIM S in the units of phr, the upper distribution of silica in relative numbers. The scale for the x- and y-axes is in micrometers. The spectra were obtained in 0.5 mm intervals with an arbitrarily chosen origin. F IG . 4. Contour map for the cured blend A, with the 50:50 BIMS/BR ratio. The regions in red correspond to about 41% surface area and represent the BIMS domains. Silica domains are shown in black and cover about 15%. The remaining surface is covered by the BR component. FIG . 5. Raman intensity ratio of zinc stearate vs. BR for the sample C, with the 30:70 BIMS/BR ratio. Colors represent areas of similar intensity ratios. APPLIED SPECTROSCOPY 1563 TABLE III. The average domain sizes, d, and the magnitude of uctuations in the relative intensity ratio, DI, for the BIMS/BR blends. ameter. In the cured samples, stearate concentrations can be found near the silica agglom erates, or near the boundaries of the BIM S domains. It is interesting to observe that the band assigned to the stearate chain vibrations and centered at 1118 cm 2 1 for the bulk zinc stearate is shifted in the blends toward lower frequencies. The shift is relatively sm all, about 4 cm 2 1, and is independent of the state of cure of the samples. The shift is toward lower frequencies and is probably caused by the chemical dissociating. Because zinc stearate can be found within the BIMS domains, one could speculate that the zinc ions are attracted to the electronegative brom ine atom of the BIMS polym er, which also results in higher stearate concentrations. However, in uncured samples where large concentrations of stearate chains are detected in BR, we also obser ve that the shapes of the Raman peaks due to the cis-polybutadiene are distorted and appear similar to those m easured for the cured samples. The band shapes change only when special interactions are present. Although the changes in the C5C band appear to be similar to those observed after the curing process, it is impossible to explain the nature of these interactions. Polymer Blends Containing Silica. To estimate domain sizes for polymer blends prepared only with silica and no curatives, we used long linear scans. Such measurem ents are much faster but provide similar information on polymer structure. Figure 6 illustrates the polymer morphology for rubber systems comprising BIMS, BR, and silica. Figure 7 shows the polymer morphology in the 50:50 blends prepared with various amounts of silica. Table II lists the average domain sizes. In the presence of silica, polymer domains are sm aller. As the concentra- 30 phr silica 45 phr silica d (mm) DI d (mm) DI d (mm) DI 25:75 30:70 40:60 50:50 72:25 4.3 – – 6.0 4.8 3.82 – – 2.50 1.80 ,2.0 – – 2.5 3.1 0.03 – 1.45 4.25 – 2.6a 2.2a 2.5a – – 0.25 0.45 0.60 – a F IG . 7. Distribution of BIM S in the 50:50 blend with BR and various concentrations of silica; from bottom to top we show samp les #5, #3, #6, and #7, respectively. No silica 2 BIM S/ BR ratio Average for both the cured and uncured blends. tion of silica is increased, the morphology of the domains becomes increasingly more uniform. This conclusion is based on the observed uctuations, DI, of the ratio of the Raman peak intensities. Lower DI values correspond to sm aller differences in the concentration of BIM S and, hence, m ore uniform distribution. For the 50:50 and 75:25 BIMS/BR blends, the aggregates of domains can also be clearly identiŽed (see Fig. 6). The average sizes of BIMS domain aggregates are about 6.0 and 4.8 mm for the respective 50:50 and 75:25 blends prepared without silica (compare Table III). The second elastomer is present, although its concentration usually is much smaller. For example, in sample #4, the 75:25 blend, cis-p oly butad iene is w ell distribu ted throughout the sample and can be found within domains formed predominantly by BIM S. The effect of silica on the morphology of the polymers in these blends is different from that observed in the 25:75 BIMS/BR blend. The concentration uctuations and the domain sizes are sm aller, but these reductions are less dramatic than those for the sample with the low concentration of BIM S. When the concentration of BIMS is 75 phr, the domains appear to be continuous, form ing large areas of sizes exceeding 20 mm (compare Fig. 6). Fluctuations in intensity ratio, DI, increase compared with those observed using silica (compare Table II). Fluctuations are proportional to the uctuations in polymer concentration. For the 75:25 BIMS/BR blend, the addition of silica does not improve, and m ay even worsen, the uniformity of the distribution of the two elastomers. To illustrate this effect better, Fig. 6 depicts the distribution curves for three samples, #2– TABLE II. The average domain sizes, d, and the magnitude of uctuations in the relative intensity ratio, DI, for the 50:50 BIMS/ BR blend as a function of silica concentration. Concentration of silica (phr) d (mm) DI 0 10 30 50 70 6.0 3.5 4.1 2.1 3.0 2.50 2.27 1.54 1.27 0.80 1564 Volume 54, Number 11, 2000 F IG . 8. Tem perature dependence of the distribution of BIM S in the 50: 50 blend with BR, sample #3. Concentration of silica was 30 phr. F IG . 9. Temperature dependence of the distribution of BIMS in the 75: 25 blend with BR, sam ple #4. Concentration of silica was 30 phr. #4, which have equal concentrations of silica at 30 phr. In this Žgure, the scan lines are offset for clarity reasons; the vertical axis is the same. Aging Effects. The temperature dependence of the distribution of BIMS in the 50:50 and 75:25 blends with 30 phr of silica was investigated via the next series of experiments. Linear scans obtained for both samples at various temperatures are shown in Figs. 8 and 9, respectively. For the 50:50 blend, the heat treatments at 100 and 150 8C resulted in progressively sm aller domain sizes and better distribution of BIMS, as the polymer m orphology becomes co-continuous. The linear scans show smaller domains and reduced uctuations of the BIMS concentration (compare Fig. 8 and also data in Table IV). However, upon heat treating at the higher temperature of 200 8C there is an increased separation of the components. This result has been reproduced several times for different sections of sample #3. For sample #4 heat treatment also results in better dispersion, but the improvements are less pronounced (compare Fig. 9 and Table IV). Degradation of cis-1-4-polybutadiene was monitored by observing the changes in the shape of the C5C stretching vibration at 1648 cm 2 1 . For freshly exposed surfaces, this band is sharp and narrow. With time elapsed, the band broadens and a shoulder appears at about 1604 cm 2 1, which steadily grows in intensity at the expense of the intensity of the initial 1648 cm 2 1 band (compare Fig. 10). To the Žrst approximation, the changes appear to be similar to those caused by the curing process. For samples stored at room-temperature conditions for four weeks, both components of the doublet TABLE IV. Domain sizes, d, and the m agnitude of uctuations in the relative intensity ratio, DI, for the 50:50 and 75:25 blends as a function of temperature. History of heat treatment Untreated 100 8C 150 8C 200 8C Sample #3 50:50 BIMS/BR, 30 phr SiO 2 Sample #4 75:25 BIMS/BR, 30 phr SiO 2 d (mm ) DI d (mm ) DI 4.1 2.1 1.8 2.3 1.45 0.95 0.57 0.95 3.1 2.4 2.5 2.2 4.25 3.75 3.70 3.25 F IG . 10. Changes in the Ram an spectra of the C5C vibrations of cispolybutadiene caused by degradation of the cured blend B. A is the spectrum obtained shortly after a new surface was made, spectrum B was taken after storing the specim en at room conditions for four weeks, and C is the spectrum obtained after 72 h of irradiation by a 200 W mercu ry lamp. were of similar magnitudes, and it was impossible to resolve them. The appearance of the second peak and its subsequent growth is associated with the degradation of the BR elastomer. After three days of illuminating the cured blend B with a 200 W mercury lamp, the intensity of the C5C band was dramatically reduced, and in addition to the 1604 cm 2 1 band, a broad peak at 1750 cm 2 1 also appeared (see Fig. 10C). The low-frequency component is most likely due to polybutadiene segments coupled to modiŽed hydrocarbon chains, and the high-freAPPLIED SPECTROSCOPY 1565 quency 1750 cm 2 1 band probably signiŽes the presence of carbonyl groups. Similar changes in this spectral region were recorded by IR absorption.7 Finally, it is found that BR degradation as monitored by the disappearance of the 1648 cm 2 1 peak proceeds faster in the sectional samples (several micrometers thick) than in the bulk material. This is likely a direct result of the increased surface area of the sectioned sample permitting oxygen to readily penetrate. By monitoring the intensities of the 1750, 1648, and 1604 cm 2 1 bands it is possible to study the kinetics of cis-polybutadiene degradation. CO NCLUSIO N We demonstrated that Raman m icroimiging can be used to map the distribution of polymers, silica Žller, and additives of relatively high concentration. Raman microimaging experiments conŽrm ed that the BIMS elastomer forms domains in the binary blends with polybutadiene. Contour m apping allows estimation of the size of these domains—their sizes, shapes, and the average distance between them. The relative concentration of polymer components within domains and the distribution of the silica Žller and zinc stearate additive were calculated. Since the blends studied were prepared without the use of an antioxidant, cis-polybutadene degrades at a rate dependent upon temperature, wavelength of the incident light and its uence, and sample thickness. Thus, to compare blends of different compositions and treatment protocols, it is important to perform the imaging experiment on freshly exposed surfaces, preferably immediately after or up to one day of generating the new surface, in order to maintain the same environmental conditions. In the absence of precipitated silica, BIMS and BR do not mix well, with the phase separation of domains being very pronounced. Many areas consist of a single polymer component, which can be clearly identiŽed. The degree to which the domains are enriched in one component depends on the composition of the use of binary blends. Upon inclusion of silica the polymer m orphology is completely changed. Increased concentrations of silica result 1566 Volume 54, Number 11, 2000 in increased viscosity of the compound and generation of greater shear deformations during the internal m ixing process, which results in a more uniform blend morphology. Each polymer is dispersed throughout the m icro regions, and single-component domains are seldom obser ved. Individual polymer domain sizes are reduced, and the distribution of the two polymers is more uniform, since m onitored uctuations in polymer concentration are reduced. The blend m orphology of the uncured samples is dependent upon temperature. Increased temperature decreases domain sizes. The extent of this effect depends on the blend composition. However, heating to 200 8C may reverse this process. For the 50:50 blend with 30 phr of silica, the separation of the components increased, while for the 75:25 BIMS/BR blend it remained the same. This is a surprising effect as we expected a continuous reduction in domain sizes and better distribution of components with increased temperature. It is observed that silica and stearate chains are found in the close proximity to the centers of the BIMS domains. On numerous occasions Raman imaging showed silica just in between two or more sharp BIMS domains, but never at the very center of the BIM S domain. For the uncured samples, silica and stearate chains can sometimes be observed in the BR component; this is more obvious for samples with high BR concentrations. However, after curing, stearate is observed only in the BIM S domains. 1. S. R. Goldstein, L. H. Kidder, T. M. Herne, I. W. Levin, and E. N. Lewis, J. Microscopy 184, 35 (1995). 2. R. Appel, T. W. Zerda, and W. Waddell, Rubber World 219, 32 (1999). 3. L. M arkwort and B. Kip, J. Appl. Polymer Sci. 61, 231 (1996). 4. S. Hajatdoost, M. Olsthoorn, and J. Yarw ood, Appl. Spectrosc. 51, 1784 (1997). 5. R. Appel, W. Xu, T. W. Zerda, and Z. Hu, M acromolecules, 31, 5071 (1998). 6. A. H. Tsou, unpublished data presented at the ACS Meeting, San Francisco, California (2000). 7. S.-M. Wang, J.-R. Chang, and R. C.-C. Tsiang, Polym. Degrad. Stabi. 52, 51 (1995).