Rubber-Modified Thermoset Resin - American Chemical Society

6 Characterization of a Carboxyl-Terminated Polybutadiene

Downloaded by MONASH UNIV on March 20, 2013 | http://pubs.acs.org Publication Date: December 5, 1984 | doi: 10.1021/ba-1984-0208.ch006

Molecular Weight Distribution, Functionality Distribution, and Microstructure Η. Ε. D I E M , D. J. H A R M O N , R. A. KOMOROSKI, J. B. PAUSCH, and R. J. BERTSCH The BFGoodrich Research and Development Center, Βrecksville, O H 44141 A commercial carboxyl-terminated polybutadiene was separated by preparative gel permeation chromatography (GPC) into 12 fractions of relatively narrow molecular weight. Information on the molecular weight, molecular weight distribution, functionality, and microstructure of each fraction was determined. Acid-equivalent weights were obtained from three end-group counting (number of carboxyls) measurements, proton NMR and IR spectroscopies, and titration. M values were obtained from vapor pressure osmometry and GPC. Results from the end-group counting methods are statistically indistinguishable. By comparing the pooled data from the end-group counting methods with the osmometry data, we calculate the functionality of the polymer to be 1.9 carboxyl groups per molecule. M values from GPC (polystyrene standards) are about 70% high. The highest molecular weight fraction (3.5% of the polymer) is unique. It consists of branched molecules with an average of about one branch per molecule. We hypothesize that the branching resulted from the reaction of a terminal nitrite on one chain end with a terminal carboxyl on another chain end. The microstructure is uniform across the molecular weight distribution. No "cage product" from coupling of the azo catalyst fragments is found, although a small loss in carboxyl is indicated by the end-group accounting. This discrepancy is assigned to the necessity to estimate carboxyl equivalents for one fraction, plus accumulated errors. n

n

JR.EACTIVE LIQUID POLYMERS (RLP),

such as H Y C A R C T B , are used i n applications where they are actually monomers for condensation po lymerizations. Monofunctional or polyfunctional constituents i n d i 0065-2393/84/0208-0065/$06.00/0 © 1984 American Chemical Society

In Rubber-Modified Thermoset Resins; Riew, C., et al.; Advances in Chemistry; American Chemical Society: Washington, DC, 1984.


66

R U B B E R - M O D I F I E D T H E R M O S E T RESINS

functional monomers have a profound effect on the condensation pro cess and on the physical properties of the resulting condensation polymer. L i t t l e information exists on the detailed molecular structure of such polymers, in particular the extent to which these materials contain monofunctional or polyfunctional chains. This chapter was designed to provide that information. C T B 2000X165 is a polybutadiene ( P B D ) synthesized free radi cally by using as initiator 4,4'-azobis[4-eyanopentanoie acid] ( A D V A ) Registry N o . [2638-94-0] (I, 2). If termination is via coupling, a car boxyl group should be on every chain end. However, termination by o t h e r than c o u p l i n g of radicals or transfer reactions ( e . g . , to monomer, polymer, solvent, or impurities) would lead to monofunc tional chains and/or to chains with more than two functional groups. Little agreement exists in the literature on the critically significant question of the functionality of such telechelic polymers. Karatavykh et al. (I) found little branching in P B D made with A D V A in acetone at 70 °C, and concluded that the role of transfer reactions in the polymerization was insignificant. Similarly, Ray (3) found that i n the polymerization of styrene at 60 °C, combination was the major ter mination mode of the radicals. O n the other hand, Valuev (4, 5), examining copolymers of butadiene with acrylonitrile (AN) and with methacrylic acid, found an average functionality of two, yet found a large amount of trifunctional polymer. Finally, Reed (6, 7) has ex amined carboxyl-terminated polymers from A D V A and found widely varying functionality, usually greater than two. A careful fractionation of C T B 2000X165 was carried out, and each f r a c t i o n was c h a r a c t e r i z e d to p r o v i d e s t r u c t u r e , m o l e c u l a r weight, molecular weight distribution, and functionality information. N u m b e r average molecular weights were obtained by both colligative and end-group counting methods [IR and N M R spectroscopies, and equivalent parts per hundred parts resin ( E P H R ) ] . B y comparing the data from these two methods, we assayed the distribution of carboxyl end groups and branches across the fractions. Although gel perme ation chromatography ( G P C ) fractionation procedures (8) have been used by earlier workers, several of the end-group counting methods we report have not previously been used on telechelic polymers. Experimental Preparative Fractionation. GPC (Waters Associates AnaPrep) was used to fractionate CTB 2000X165 by molecular size. The conditions for the fraction ation were as follows: Five 4ft x 2.5 in. i.d. columns packed with Styragel of porosities 10 (1), 10 (1), 500 (2), and 100 (1) A. The solvent was distilled toluene; the flow rate was 50 mL/min.; and the fraction size was 390 mL. Eleven injections were made for a total throughput of 110 g of polymer. At the end of the run, each fraction was contained in 4.3 L of toluene. The toluene was removed by use of a rotary evaporator after addition of 0.5% Antioxidant 2246 (A02246) (American Cyanamid) (based on polymer weight) to the solution. Antioxidant was required because the antioxidant originally present in the 4

3


6.

DIEM ET AL.

Carboxyl-Terminated Polybutadiene

67

polymer would all be in the lowest molecular weight fraction. The fractions were taken down to about 150-mL volume, and final drying took place in a vacuum oven at 40 °C for ~ 96 h (to constant weight). The weight percent values were obtained by dividing the dry weight of each fraction by the weight of polymer charged. Analytical G P C . Each of the fractions was analyzed for molecular size distribution (MWD) with a Waters Model 200 GPC modified for high-speed operation with microparticulate (10 jxm) columns. The conditions for the analysis were as follows: The columns were 5-|xm Styragel 30 cm x 7.8 mm i.d.; 10 , 10 , 10 , 10 , and 10 A pore size (Waters Associates designation). The solvent was tetrahydrofuran (THF) (Fisher); the flow rate was —2.0 mL/min; the sample size was 200 |xL of 0.2% concentrate; and the detector was refractive index (An) at 4 x attenuation. The columns were calibrated by using narrow size distribution polystyrenes (PS). Because no universal calibration constants are available for these polymers, all GPC data were reported in terms of PS equivalent molecular weight. Vapor Pressure Osmometry. The number average molecular weight (M ) of each fraction was determined with vapor pressure osmometry (VPO) (Corona/ Wescan Model 232A molecular weight apparatus). The solvent used was ethyl acetate at 37 °C. The instrument was calibrated with benzil (MW 210.2) and sucrose octaacetate (MW 678.6). An instrument constant of 42495 was obtained (K = MW • AH/C where C is in moles per liter). At least four concentrations were measured for each sample, and AR (change in resistance) was obtained. AR/C (where C is in grams per liter) was then plotted against C and (by using a least squares fit) extrapolated to zero concentration. The data were corrected for the amount of antioxidant in each fraction. M was then calculated from 2


3

4

5

6

n

n

Functionality by Acid Group Titration. A sample (~1 g) was weighed into a 4-oz jar, and 50 mL of a toluene/2-propanol (60/40 by volume) mixture was added. The sample was mixed on a magnetic stirrer for —20 min, or until dissolved. The mixture was then titrated with -0.1 N K O H in ethanol (stan dardized) with a phenolphthalein indicator. The following equation was used to calculate acid content. C O O H EPHR = ^ L t i t r a n t ) ( N o f t i r a n ) (grams sample) x 10 t

t

The acid-equivalent weight (AEW) of each fraction was then calculated from A E W = 100/COOH EPHR. Proton N M R Analysis of M . Proton Fourier transform N M R (*H FTNMR) spectra were obtained at ambient temperature at 200 M H z on a Bruker WH-200 spectrometer. Samples were dissolved in CDC1 with tetramethylsilane (TMS) as internal standard and were run in 5-mm o.d. N M R tubes. Typical conditions were 90° rf pulses of 4.6-|xs duration; pulse repetition rate, 9.1 s; and line broadening, 0.4 Hz. The *H FTNMR spectra of the preparative GPC frac tions were used to obtain A E W values. The triplet at about 2.5 ppm (Figure 1) is assigned to the methylene protons next to a C O O H group on the basis of model compounds and expected chemical shifts. A E W values were calculated for comparison to other data. The calculation is based on the ratio of the peak area due to alkene protons (adjusted for 1,2-units) to the area due to the end groups. Hence, number of butadiene units = (E + D/2)/2, number of C O O H = C/2, number of butadiene units/COOH = (E + D/2)/C, and A E W = (number of units/COOH) (54.0) + (126), where 54 and 126 are the molecular n

3


68


C H0 CCH2


2

'fZ-NMR spectrum of fraction 2000X165.

Figure 1. The 200-MHz

5 of CTB

weight of a monomer unit and an end group, respectively. An adjustment was made for a small amount of antioxidant detected in Fractions 9-12. Carbon-13 N M R Analysis of Microstructure. Carbon-13 FTNMR spectra were obtained at ambient temperature at 50.3 M H z on the Bruker WH-200 spectrometer. Samples dissolved in CDC1 with TMS were run in 20-mm o.d. tubes. Typical conditions were 90° rf pulses of 35-|xs duration; 12,195-Hz spectral width accumulated in 32,000 points with quadrature detection; pulse repetition rate, 10.7 s; and line broadening, 0.5-2 Hz. Increasing the pulse repetition rate to 20 s did not noticeably change the spectrum, hence peak intensities are not attenuated by incomplete relaxation to equilibrium. For reasons of sensitivity, all of the C spectra were acquired in standard fashion with nuclear Overhauser enhancement (NOE). To ensure that variable NOEs were not affecting our microstructural results, a C spectrum of a fraction was acquired with N O E suppression (9). No significant change was observed relative to the results ob tained with N O E . Carbon-13 N M R spectra of PBDs containing cis-1,4-, trans-1,4-, and 1,2units have been reported, and peak assignments made {10, 11). Using the pub lished assignments, we obtained cis-, trans-, and 1,2- compositions for the pre parative GPC fractions with C - N M R spectroscopy. A typical C spectrum is shown in Figure 2. The percent vinyl was obtained from the ratio of the vinyl carbon areas + X ) to the area due to all olefinic backbone carbons (X + X + Z). For fractions 9 and 10, the area Z is corrected for the presence of A02246 by using the peak at about 31 ppm and the spectrum of A02246. The remaining non-1,2-unit content is divided between cis-1,4- and trans- 1,4-units on the basis of the ratio of the respective peaks due to backbone C H (see Fig ure 2). Determination of A E W by IR Spectroscopy. A E W values were calculated for each fraction from the relationship A E W = 126.14/weight fraction of chain ends. The weight fraction of chain ends was obtained for each fraction by an absorbance subtraction method developed for this problem; the details will be published elsewhere (12). The method treats carboxyl-terminated polybutadiene (CTPB) as a mixture of PBD and of ends, and applies a binary mixture analysis (13). Traditional methods based on titration standards were used for eonfirma3

1 3

1 3

13

1 3

2

Y

2

2



1

2

A

I

•

1 i

Figure 2. The 50.3-MHz

4 _ J L 80

_l ppm

I 70

I

I

I

I

I

L_

3 of CTB

x J L j J J U ULA_JL. C-NMR spectrum of fraction 2000X165.

13

L_


CD

as

5

5

©

O

70

RUBBER-MODIFIED THERMOSET RESINS

1Q0

i3

63


\

r

J r r i I \

s V

A

A J M i

\

i

i

y

40

i

0

) 2000

1800

1600

1400 1

WAVENUMBER (CM )

Figure 3. IR spectra of fraction 2 (upper curve) and fraction 7 (lower curve) of CTB 2000X165. tion. Each fraction was dissolved in carbon tetrachloride (40 g/L), and a signalaveraged, solvent-compensated spectrum (0.2-mm cell) was recorded between 1900 and 1300 c m . Measurements were made on a Perkin-Elmer 180 spec trometer at a spectral slit width of 1.8 c m , a scan rate of 96 cm~Vs, and a data interval of 0.5 c m ; two scans were averaged. The index bands for the analysis, which can be seen in the spectra of Figure 3, are the carboxyl dimer absorption at 1714 c m " , for the chain ends, and the vinyl double bond stretch at 1639 c m , for the PBD. The analysis was made, two fractions at a time, by subtracting the spectra of the two fractions. First, the scaling factor (F ) required to null the P B D in one subtraction was determined. Then the factor (F ) re quired to null the carboxyl band in the inverse subtraction was determined. The weight fraction of chain ends in the two fractions, x and y, were calculated from these scaling factors. - 1

- 1

- 1

1

- 1

x

2

1 - F, 1 - FiF

(1 ~ Fi)F - ^TTTT FjFg)

p

2

2

V

=

7i (1

=

2

F

X

By taking the lowest molecular weight fraction (fraction 12) with each of the other fractions in turn, the weight fraction of chain ends was calculated for all three fractions except 1, 10, and 11.


6.

DIEM ET AL.


71


3

DECREASING MOLECULAR SIZE

Figure 4. Anaprep preparative GPC size distribution profile. The microstructure of the P B D portion of the whole polymer was deter mined by an in-house procedure that had been calibrated by ^ - N M R mea surements. The microstructures of the fractions were not determined directly. However, the fractions were judged to have the same microstructure because the trans and vinyl bands could be nulled out of any pair of spectra with the same scaling factor. Results F i g u r e 4 shows a chromatogram taken during the preparative G P C fractionation. O n e of the objectives of this study was to prepare narrow fractions of C T P B suitable for calibrating the G P C analysis. The results of the fractionation, plotted i n Figure 5, show that the efficiency of the fractionation is good, and the fractions are of suitably narrow distribution. The analytical G P C data (Table I) are plotted i n Figure 6, i n w h i c h the solid line is the least squares line calculated omitting fraction 1. The M data from the colligative methods and the A E W data from the end-group counting methods are collected i n Table II. In general, the molecular weight progresses from high to low values through the fractions. However, the molecular weight of fraction 1 is lower than that of fraction 2 with all methods except N M R spec troscopy and G P C . The M values obtained with the analytical G P C (calibrated w i t h narrow size distribution PS fractions) are some 70% above the values obtained from V P O and are about equally above the values obtained from the various end-group counting methods as w e l l . T h u s , the n e e d for recalibration is obvious. W i t h the data plotted as the solid line i n F i g u r e 7 (omitting fraction 1) the relation s h i p M , , = 0.5697 M „ is f o u n d . T h e standard e r r o r of the n

n

n

L

"actual

DC

n

PS

relationship is 525. The microstructural results are given i n Table III. Essentially no change is observed over the entire molecular weight range. Values determined by N M R and IR spectroscopy for the whole polymer are In Rubber-Modified Thermoset Resins; Riew, C., et al.; Advances in Chemistry; American Chemical Society: Washington, DC, 1984.

72


CUMULATIVE DISTRIBUTION CURVES

100-


Original

0

6 LOG PSMW

Figure 5. Fractionation efficiency for several GPC fractions from Figure 4. in agreement. A full characterization of the sequence distribution of the cis-, trans-, and 1,2-units was not performed. Comparison of the C spectra of the various fractions indicates no major change of sequence distribution with molecular weight. Comparison to spectra in the literature suggests a random distribution of the various units (2). We did not detect any additional resonances in either the H or C - N M R spectrum of fraction 1 that could be attributed to car1 3

1

1 3

Table I. Analytical GPC Results on CTB 2000X165 Fraction

M

Original 1 2 3 4 5 6 7 8 9 10 11 12

14,900 38,500 23,100 15,100 9,620 8,110 6,560 5,720 5,600 4,440 3,870 3,000 2,670

w

M

n

9,000 28,300 19,100 12,600 8,300 5,960 4,920 4,290 4,070 3,470 3,070 2,760 2,470

Peaks 12,700 30,300 18,900 12,700 8,080 7,230 5,790 4,910 4,650 3,730 3,170 2,840 2,690

N O T E : Excluding fraction 1, M = 1.1923 M + 284; and r = 0.9947. m

n

2


6.

73

Carboxyl-1"erminated Polybutadiene

DIEM ET AL.

36

32

28

o IS


20

16 12

8

4

0

4

8

12

16

M x I0"

20

24

28

3

n

Figure 6. Molecular weight values for fractions of CTB 2000X165 obtained from analytical GPC.

Table H . M Values for CTB 2000X165 Fractions r

Fraction 1 2 3 4 5 6 7 8 9 10 11 12 Whole polymer a b

Weight Percent

VPO

EPHR

IR

NMR

GPC

3.47 14.51 25.35 18.76 11.97 7.96 5.42 3.90 3.04 2.61 1.64 1.21

7,178 10,365 8,365 4,941 3,821 3,256 2,588 2,266 1,534 833 1,560 1,389

3,437 5,406 4,762 2,591 2,020 1,642 1,376 1,134 1,035 — 903 769

4,885 5,592 3,983 2,880 2,305 1,805 1,511 1,468 1,312 — — 983

7,100 6,250 4,410 3,335 2,485 1,975 1,625 1,530 1,285 1,415 1,235 1,050

28,300 19,100 12,600 8,300 5,960 4,920 4,290 4,070 3,470 3,070 2,760 2,470

3,601

1,961

—

3,150

9,000

a

a

a

b

Data given as acid equivalent weights. Based on calibration with PS.


74



12

0

8 6 5

4

4

8

3

2

12

16

I

20

24

—FRACTION

28

32

3

M x l 0 " , GPC n

Figure 7. M values from end-group counting methods vs. analytical GPC. n

M from a

boxyl-containing branches or any defect structures. However, the presence of such structures is indicated by the low A E W obtained for this fraction. Moreover, the carboxyl region of the IR spectrum could not be n u l l e d out by the carboxyl region of any other fraction. A n absorption always remained that overlapped the 1 7 1 4 - c m acid dimer band. (An I R estimate of the molecular weight of fraction 1 was obtained from the linear plot of the IR absorbance ratio A / A versus A E W for the other fractions. This estimate is obviously low as a result of the overlap at 1714 c m . ) These observations suggest a unique structure i n fraction 1. -1

1 6 3 9

1 7 1 4

- 1

Table III. Microstructure of HYCAR 2000X165 Fractions cis-J,4-Units trans-1,4-Units

Fraction Whole polymer 1 2 b 4 5 6 7 8 9 10 3

C c

0

23(22) 22 22 21(21) 22 23 22 22 23 21 25

57(58) 58 58 58(59) 59 58 59 59 58 58 60

1,2-Units 20(20) 20 20 20(20) 19 19 19 19 19 21 15

Data given as percentages. Values in parentheses were obtained by IR spectroscopy. Values in parentheses were obtained with N O E suppression. Corrected for the presence of antioxidant.

NOTE: a b

c


6.

DIEM ET AL.

75



Discussion The agreement among the three end-group counting methods and V P O is quite gratifying. The data from each pair of methods were submitted to linear regression analysis. Data for fractions 10 and 11, as w e l l as for fraction 1, were omitted from this analysis, because the IR spectra of fractions 10 and 11 showed large amounts of antioxidant, especially i n 10. If all the 1.5 parts of A 0 2 2 4 6 added initially had eluted with these two fractions, the antioxidant should constitute nearly 35% of the sum of the weights of fractions 10 and 11. Absorbance subtraction showed that fractions 9 and 12 also contained more A 0 2 2 4 6 than the other fractions. Because this additional A 0 2 2 4 6 was only tenths of a percent, data from fractions 9 and 12 were used in c a l c u l a t i o n s . R e g r e s s i o n l i n e s w e r e c a l c u l a t e d for each p a i r of methods with first one method as the independent variable, then the other. Standard error of estimate values were obtained with each line, and the algebraic mean of the regression lines was taken as the least squares equation. Correlation coefficients were uniformly high, and precision was good in most instances. In comparing the M values from the various methods (M values were calculated for the IR, N M R , and E P H R methods by assuming two carboxyls per chain), V P O values were consistently the lowest, N M R were consistently the highest, and E P H R and IR results were intermediate. The differences are not really large, however. The as sumption of two carboxyls per chain can be tested with the slope of the plot of V P O (y) against any end-group method (x). The slopes of those plots are 1.87, 2.08, and 1.82 for the E P H R , IR, and N M R methods, respectively. However, the standard errors of the slopes are large enough so that these values are not statistically different, and the data can be pooled. The slope of the plot of V P O against the pooled end-group methods is 1.90. If our assumption (termination by coupling only) is correct, the slope should be 2. Thus, we have 1.9 functional groups per chain with a precision of about 0.1. The standard error of estimate of the pooled data is 502, which is i n line with the data for the methods taken by pairs. W e attempted to account for the fate of the carboxyl end groups in the fractionation, as detailed in Table IV. About 15% of the carboxyl groups i n the starting polymer were not accounted for by the mea sured fractions. Part of this loss may result from the necessity of estimating the E P H R for fraction 10, w h i c h contained much antiox idant. W e calculated that the loss is equivalent to about 1% of the total weight as catalyst "cage" product. However, tests showed that, although the preliminary filtration step may have removed traces of (insoluble) cage product, 1% could not have been present. Thus the loss in carboxyl is probably a result of accumulated weighing and titration inaccuracies. Fraction 1 is clearly unique. The anomaly is striking in Figure n

n


76


Table IV. End-Group Accounting


Fraction

Weight Percent

1 2 3 4 5 6 7 8 9 10 11 12

3.47 14.51 25.35 18.76 11.97 7.96 5.42 3.90 3.04 2.61 1.64 1.21 99.84

Lost

0.16

EPHR

Equivalents

0.0291 0.0185 0.0210 0.0386 0.0495 0.0609 0.0727 0.0882 0.0966 (0.1)* 0.1108 0.1300

0.00101 0.00268 0.00532 0.00724 0.00593 0.00485 0.00394 0.00344 0.00294 0.00261 0.00182 0.00157 0.04335 (0.051 whole sample) 0.00765

a

(4.78)

c

Corrected for 0.5% A02241. Estimated. 0.97% of total as cage product; cage product E P H R = 0.7928. a

b c

7. I n this figure, the M values from N M R , IR, E P H R , and V P O measurements are plotted against the G P C M values (excluding frac tions 10 and 11). T h e M values for the end-group counting methods are obtained b y m u l t i p l y i n g the A E W b y 1.9. T h e solid line is the least squares line calculated, omitting fraction 1, b y pooling the e n d g r o u p c o u n t i n g a n d V P O data. T h e fact that a l l t h e e n d - g r o u p counting methods find far too many carboxyls suggests that fraction 1 is branched. However, the V P O result does not seem i n accord with this; agreement among the methods is poor. O n e sort of " b r a n c h " can be proposed to harmonize the data. Twenty-five years ago, Grassie and coworkers were studying the de velopment of yellow color i n poly aery lonitrile. They found (14, 15) that a potent catalyst for the color reaction was the imino anhydride, - C ( 0 ) O C ( N H ) - , that formed when a small amount of acrylic acid (2-propanoic acid) was copolymerized w i t h the acrylonitrile and the acid hydroxyl was added across the C = N bond. In some of our pre vious characterization work on R L P s , a cyclic imino anhydride was postulated as the starting point i n heat aging changes (development of color, increased viscosity, and loss of carboxyl and nitrile groups) in butadiene-acrylonitrile copolymers (16). Some support for this structure was obtained b y absorbance subtraction I R spectroscopy. In the present case, w e suggest the possibility that fraction 1 consists of chains i n w h i c h the carboxyl at the e n d of one chain has reacted with the nitrile at the e n d of another to give ( C N ) C ( C H ) C H C H C ( 0 ) O C ( N H ) C ( C H ) ( C H C H C O O H ) . Such a reaction is really chain n

n

n

3

3

2

2


2

2


6.

DIEM ET A L .


77

extension, rather than branching, because the branch is only three carbons long and the molecular weight is doubled. W i t h one such branch per molecule, N M R and IR spectroscopy should observe three carboxyls rather than two (except that absorp tion from the imino anhydride w i l l complicate the IR spectra). But because an anhydride would be expected to hydrolyze easily, E P H R titration should observe four carboxyls. Similarly, V P O should mea sure two molecules per mole because interchange with the ethyl acetate solvent w o u l d split the anhydride. Using this logic and Figure 7, we find that the number of branches per molecule in fraction 1 is 0.6 from N M R , 1.3 from V P O , and 1.6 from E P H R data. As a result of the carboxyl overlap, only a maximum value, 1.5, can be calculated from IR data. These values are in reasonable agreement, especially considering that in the other fractions, V P O and E P H R consistently found more carboxyls than N M R and that all the methods have their smallest accuracy for fraction 1. W e take an average value of 1.2 as the number of branches per chain in fraction 1. Note the dashed line in Figure 7 that is the least squares line calculated omitting both fractions 1 and 2. Although all the points for fraction 2 fall below the line, the precision is such that no firm con clusions can be drawn about this fraction. But the data suggest that some branching may also exist in fraction 2. (The use of the dashed line only increases the calculated number of branches per chain in fraction 1 to —1.6.) Is there any experimental evidence for an imino anhydride struc ture in fraction 1? W e have already noted that nothing was observed with C - N M R spectroscopy that might relate to branches. This result may occur simply because the concentration of such structures would be not more than one-third that of the carboxyl groups, which is already quite low. However, carbonyl groups are among the strongest absorbers in IR spectroscopy and can be observed at very low con centrations. W e noted previously that the carboxyl region in fraction 1 could not be nulled in absorbance subtraction by the carboxyl re gion of any other fraction. This result is shown in Figure 8, where the spectrum of fraction 2 is subtracted from fraction 1. Although the noise level is quite high because of the low concentration of functional groups, w h e n the carboxyl dimer band is brought to zero at 1714 c m , appreciable absorption remains. F r a c t i o n 1 contains some other carbonyl function with a maximum near 1740 c m " and possibly a second weaker band at —1705 c m . W e have found no spectra of imino anhydride model compounds in the literature, but, by analogy with open chain anhydrides and imides (17), one might expect a pair of bands in this region. By itself, this evidence is weak, but it does support the proposed method by which "branching" could arise in fraction 1. Also, the overlap of this band makes the carboxyl dimer band too strong i n the IR spectrum of fraction 1, and as a result, IR measures appreciably more carboxyl than N M R . 1 3

- 1

1

- 1


78


1Q0

A

J

E)

ft

I

/

\ Downloaded by MONASH UNIV on March 20, 2013 | http://pubs.acs.org Publication Date: December 5, 1984 | doi: 10.1021/ba-1984-0208.ch006

63

\l P

1/\ V

Si

A r I

I

0

I— 2000

1800

1600

1400 1

WAVENUMBER (CM )

Figure 8. Difference in spectra obtained from IR absorbance subtraction: (fraction 1) — factor (fraction 2). Both spectra are A absorbance of 0.1 AUFS: upper curve, factor just sufficient to null the vinyl band at 1640 cm' ; lower curve, factor large enough to bring the (COOH) band at 1714 cm to (slightly below) zero. 1

2

-1

Finally, indirect evidence of the proposed structure is obtained by considering that fraction 1 d i d , i n fact, elute first o n the G P C (i.e., it was the highest molecular weight fraction). Because the mech anism we propose is really chain extension, w h i c h doubles the m o lecular weight w i t h only a three carbon atom branch, such a result w o u l d be expected. T h e amount of a branched fraction such as frac tion 1 that is found i n a given lot of polymer w i l l probably correlate with the heat history and time elapsed since manufacture of the given polymer. Thus, the observation of Valuev (4, 5) of a polymer with an average functionality of about two but with a large fraction of t r i functional material is not necessarily inconsistent with this work. In summary, C T B 2000X165 contains a small fraction (3.5%) w i t h about three carboxyls p e r chain. T h e remaining polymer consists of about 90% difunctional chains and about 10% monofunctional chains.


6.


DIEM ET AL.

The microstructure was uniform across the molecular weight distri bution. The relationship of the actual M to that calculated from G P C (with PS calibration) is constant across the total M W D and allows a simple conversion. n

Acknowledgments


The authors thank the following people for their assistance: W. R. Barr, for preparative G P C fractionation; J. E . Shockcor, for N M R spectroscopic measurements; R. J. Stevens, for E P H R measure ments; and A . M . Fairlie, for statistical analysis support. Literature Cited 1. Karatavykh, V. P.; Barantsevich, Ye. N.; Ivanchev, S. S., Polym. Sci. USSR Engl. Transl. 1976. 18, 991. 2. French, D . M., Rubber Chem. Technol. 1969, 42, 71. 3. Ray, G. Palit, S. R. Indian J. Chem. 1971, 9, 1124. 4. Valuev, V. I.; Tsvetskovskü, I. B.; Trizna, N . N.; Schlyakhter, R. A. Metody Anal. Kontrolya Kach. Prod. Khim. Prom-sti 1978, 3, 24. 5. Valuev, V. I.; Shlyakhter, R. A . ; Pavlova, I. L . ; Rozinova, O. A. Polymer Sci. USSR 1973, 15, 3097. 6. Reed, S. F., Jr. J. Polym. Sci. Part A-1 1971, 9, 2147. 7. Ibid 1972, 10, 649. 8. Law, R. D. J. Polym. Sci. Part A-1 1971, 9, 589. 9. Freeman, R. Hill, H . D . W.; Kaptein, R. J. Magn. Reson. 1972, 7, 327. 10. Elgert, K.-F.; Quack, G . ; Stutzel, B. Polymer 1975, 16, 154. 11. Clague, A. D. H.; Van Broekhoven, J. A. M.; Blaauw, L. P. Macromolecules 1974, 7, 348. 12. Diem, H . E . unpublished results. 13. Diem, H . E . ; Krimm, S. Appl. Spectrosc. 1981, 35, 421. 14. Grassie, N.; McNeill, I. C. J. Polym. Sci. 1959, 34, 211. 15. Grassie, N.; Hay, J. N . J. Polym. Sci. 1962, 56, 189. 16. Diem, H . E . private communication. 17. Bellamy, L. J. "Advances in IR Group Frequencies"; Methuen: London, 1968. ;

;

RECEIVED

for review November 18, 1983.

ACCEPTED

April 16, 1984.


Rubber-Modified Thermoset Resin - American Chemical Society

Recommend Documents