I
CARL
1. MALM, LEO J.
TANGHE, and R. E. GLEGG
Cellulose Technology Division, Eastman Kodak Co., Rochester 4, N. Y.
Effect of Solvenf and Catalyst on
...
Viscosity of Cellulose Esters Changing catalyst and/or solvent gives different viscosity reductions a t equivalent esterification rates
THIS
work was designed to determine the effects of various mixtures of organic solvents, acetic anhydride, a n d acid catalysts on cleavage of the cellulose chain and esterification of the hydroxyl groups. To avoid the variable diffusion and distribution of acetic anhydride and catalyst encountered in heterogeneous systems containing fibrous cellulose, reactions of a partially esterified cellulose were studied in solution. I n this model system esterification could be completed in a relatively short time, after which chain cleavage could be studied separately.
Experimental Cellulose Ester. A cellulose acetate butyrate was used, because of its relatively high hydroxyl content and solubility in a wide variety of solvents. Analyses of Cellulose Acetate Butyrate Primary hydroxyl, wt. % Secondary hydroxyl, wt. % Total hydroxyl, wt. % Acetyl, wt. % Butyryl, wt. % OH groups/g.u. Ac groups/g.u. Bu groups/g.u. Inherent viscosity in 90y0 methylene chloride-10% methanol Disodium hydrogen phosphate (stabilizing salt), wt.
Solvents.
1.33 1.79 3.12 19.3 27.1 0.54 1.33 1.13 1.47 0.04
Reactions were studied in
99.7y0 acetic anhydride, and in mixtures containing 10 weight yoacetic anhydride and 90% each of acetic acid, dioxane, ethylene chloride, ethyl acetate, and isopropyl acetate. All liquids were redistilled. T h e solvent-acetic anhydride mixtures were refluxed to remove small amounts of water which could interfere with early stages of acetylation. Ester Solutions. T h e ester was dried a t 110' C. for 2 hours. Solutions were made just prior to use, by vigorously stirring 5 grams of dry ester into 100 grams of solvent mixture. Solution was usually complete within 15 minutes. This concentration provides a large
excess of acetic anhydride, because only 0.94 gram is consumed in completely esterifying 5 grams of the ester. A corresponding amount of acetic acid is formed, so small changes in composition occur. Catalysts. Only catalysts that d o not combine with cellulose during esterification were used. Three sulfonic acids were compared with perchloric acid, which is a powerful catalyst for cellulose acetylation. Perchloric acid was reagent grade ( 7 2 weight yo); sulfoacetic acid was prepared from sulfuric acid and acetic anhydride, a n d isolated as the crystalline monohydrate; methanesulfonic acid was a redistilled Eastman Kodak product (boiling point, 153" C. a t 5 mm.); methanedisulfonic acid was obtained from Tennessee Eastman as a n aqueous solution containing 52 weight 70acid. T h e catalysts (except methanesulfonic acid) were added from stock solutions in acetic acid of such concentrations that 1 ml. per 100 grams of the ester solution gave the required molal catalyst concentration. Enough acetic anhydride was added to the sulfoacetic acid and methanedisulfonic acid stock solutions to react with free water or water of crystallization. Methanesulfonic acid was added undiluted. No measurements were made with this catalyst in dioxaneacetic anhydride, because a solid addition compound separated out when methanesulfonic acid was added to the solvent. Temperature. All reactions were studied a t 25' C. Because addition of catalyst caused a temperature rise, ester solutions were precooled. Precipitation. Water is a suitable precipitant when the ester is dissolved in acetic acid-acetic anhydride or dioxaneacetic anhydride. Samples were precipitated from ethylene chloride-acetic anhydride into 65y0 3A denatured alcohol, followed by a wash in 65% alcohol t o prevent samples' becoming horny during subsequent washings; from ethyl acetate-acetic anhydride into 30% 3A alcohol; and from isopropyl acetateacetic anhydride into 45% 3A alcohol T h e choice of precipitant was based on experiments in which 100 grams of a 5y0
solution of the ester (no catalyst) was precipitated in 1.5 liters of alcohol-water of various compositions. These proportions are typical of those used throughout this study. T h e precipitates were examined for their physical condition, and viscosities of the dried samples were determined in 90% methylene chloride10% methanol by weight. T h e concentration which gave a fluffy precipitate, with the same viscosity as the original ester, and involved use of the smallest amount of alcohol (Table I), was chosen; 20% acetic acid was chosen for ester solutions in acetic anhydride. ~
~
~~~~~~
Table 1. Effect of Precipitating Ester from Solvent Mixtures into Alcohol
dlcohol Concn.,
7%
Inherent Viscosity in 90% MeC12-10% MeOH EtClzEtOAc- IdoPrOAc .le20 .le20 -lC?O
-
90 80
70 60 50 40 30 20
10
...
1.80 1.70 1.61 1.57 1.53 1.48 1.48 1.46* 1.45*
1.56 1.46 1.47
... Q
a
... ...
...
1.64 1.56 1.54 1.49 1.47 b b b
Ester remained dissolved in droplets of organic liquid phase, so there was no precipitate, or partial precipitation. * Precipitates were too gelatinous, sticky, or horny to isolate, handle, or wash conveniently. a
Table II. Catalyst Concentrations Required to Produce Comparable Esterification Rates in Various Solvents W e r e Determined by Trial and Error
Solvent in Combination with Acetic Anhydride Acetic acid Dioxane Ethylene chloride Ethyl acetate Isopropyl acetate Acetic anhydride
VOL. 51, NO. 12
Catalyst Concentritt'iori (Molal) PerMethane Sulfochloric sulfonic acetic acid acid acid 0.002 0.00045
0.20
...
0.0058 0.070
0.00035 0.00042
0.154 0.61
0.0058
0.00042
0.462
0.0058
0.00035
0.154
0.0005
DECEMBER 1959
...
1483
llli
Figure 1. Measurements of change in flow time and inherent viscosities correlate well
0.28
=
2.303 log relative viscosity 0.25
The values differ very little from the usual intrinsic viscosities obtained by extrapolation to infinite dilution. Results and Discussion
l 100
i
i 80
l
l 60
l
l
l
40
i
l
20
1 0
% of original flow time of reaction mixture
Table 111.
Inherent Viscosity of Precipitated Samples Remained Constant from 20 to 120 Minutes after Catalyst Addition
(Soh-eiit. 90% methylene chloride-lO% methanol by weight) Minutes after
Catalyst Addition 10 15 20 30 50
75 120
I
-I
1.40 1.32 1.33 1.32 1.33 1.32 1.38
1.30 1.31 1.32 1.30 1.28 1.31 1.30
I.
I
1.40 1.36 1.32 1.32 1.29 1.26 1.36
I
I
3.0
9 I 2.0
2
e ,-e 1.0 I-
I
c3
w
3 0 0
50
20 40 60 REACTION TIME (MIN)
0
IO
20
80
30
REACTION TIME (DAYS) Figure 2. At equivalent esterification rates loss of viscosity in acetic acidacetic anhydride contrasts sharply with that in other solvents Acetic anhydride (0.00035ML H Ethylene chloride-acetic anhydride 10.00035M). Ethyl acetate-acetic anhydride (0.00042M). A Isopropyl acetate-acetic anhydride (0.00042M). A Dioxane-acetic anhydride (0.00045M). 0 Acetic acid-acetic anhydride (0.002M)
1484
1.42 1.40 1.34 1.33 1.34 1.32 1.32
1.44 1.41 1.35 1.36 1.34 1.34 1.32
1.36 1.34 1.32 1.33 1.32 1.32 1.33
1.35 1.33 1.33 1.34 1.31 1.31 1.32
Washing and Drying. All precipitated samples were washed (with vigorous mechanical stirring) twice with distilled water: once in slightly alkaline water (by adding small amounts of sodium carbonate to obtain a permanent pink color with phenolphthalein indicator), and finally in another change of distilled water. Between washings, the sample was recovered in a muslin bag, where it received more washings. The samples were squeezed as dry as possible, dried overnight at 50' C., and eventually a t 60' C. for 3 hours in the vacuum of a water aspirator. Esterification. Changes in total hydroxyl content of the ester, determined by carbanilation ( 3 ) , were used as a measure of esterification. The reaction mixtures from which samples were isolated for this purpose were held a t 25' C. for study of chain cleavage. Chain Cleavage. Changes in viscosity were used as an indication of rate of chain cleavage. Measurements were made on aliquots of the reaction mixtures and on precipitated samples. Flow times of 10 ml. of the reaction mixture were measured in OstwaldCannon-Fenske viscometers at 25 C . Changes in viscosity were used as a guide in determining time intervals a t which to isolate samples by precipitation. Inherent viscosities ( 7 ) of precipitated samples, determined in 90% methylene chloride-I 0% methanol by weight, were calculated from the results a t a single concentration (0.25 gram of ester per 100 ml. of solution), using the expression
INDUSTRIAL AND ENGINEERING CHEMISTRY
Viscosity measurements were made on products from the reaction mixtures used for following esterification. The changes in viscosity were studied essentially on the resulting triester. This is important, because separate experiments showed that faster loss of viscosity occurs at lower degrees of substitution. In evaluating the effects of solvents and catalysts on reduction in viscosity, a t least two bases of comparison may be used. Adjusting catalyst concentrations so that esterification rates are the same in different solvents was strongly favored, because it seemed that under these conditions the solutions must have at least one common property. Catalyst concentrations to give comparable esterificarion rates icere determined by trial and error (Table 11). The second possibility was to compare the reduction in viscosity caused by equimolar amounts of catalysts in different solvents, without regard to esterification rates. This provided data concerning the influence of solvent on the activities of catalysts in promoting esterification (acetylation). However, equimolar amounts of catalyst gave equivalrnt esterification rates in some solvents, and direct comparisons could be made on the first basis. The esterification rate was evaluated from the chanqe in per cent total hydroxyl in the ester with time after addition of catalyst. The cellulose acetate butyrate contained 3.12% total hydroxyl originally. Viscosity Changes of Reaction Mixtures and Precipitated Samples
Immediately after addition of catalyst to some ester solutions, changes in flow time occurred which could not be attributed to esterification or cleavage of the glycosidic bonds. This effect was especially large when the solvent was ethylene chloride-acetic anhydridewhen 0.00035M perchloric acid was added, the flow time a t once fell from 809 to 110 seconds and remained a t this value for 30 days. I t was therefore decided to calculate changes in flow time on the basis of values 60 minutes after addition of catalyst. AS esterification was almost complete by this time, the viscosity could not be influenced by variations in hydroxyl content of the ester. In the routine determination of inherent viscosity, 0.25 gram of the ester is dissolved in 100 ml. of 9070 methylene chloride-lO% methanol, regardless of type or amount of acyl substituent.
*
C E L L U L O S E ESTER V I S C O S I T Y Thus inherent viscosity would decrease as acetylation progressed, even when no chain cleavage had occurred. In these studies the ester was essentially trisubstituted soon after addition of catalyst; therefore. i t was necessary to decide on a viscosity value other than that of the original ester for calculating percentage changes. Viscosities of representative samples, precipitated during esrerification, remained constant between 20 and 120 minutes after addition of catalyst (Table 111). T h e mean viscosity value for 35 samples is 1.32 'l'otal hydroxyl varied from 0.13 to 0.70% at 20 minutes, decreasing later. Per cent changes in viscosity of precipitated samples were therefore calculated on the basis of 1.32 for the undegraded triester, rather than the original value of 1.47. In many cases, there was no change from 1.32 after 30 days of reaction. Figure 1 shows the relationship between percentage changes in flow time of reaction mixtures and inherent viscosities in methylene chloridemethanol of precipitated samples. The latter always show smaller changes.
REACTION TIME (MIN)
Consideration that the 66 points represent all combinations of solvent-catalyst systems, the two measurements correlate well. For this reason, the results are shown only as changes in viscosity of precipitated samples in methylene chloride-methanol (Figures 2 to 6).
Solvent
0
10
20
30
REACTION TIME (DAYS) Figure 3. Esterification is slower with methanesulfonic acid than with perchloric acid, but viscosity reduction is faster except in acetic acid-acetic anhydride
Catalyst Per- Methanechloric sulfonic Sulfoacetic acid acid acid
Acetic acid-AcnO Dioxane-AcpO Ethylene chloride-AcpO E t h y l acetateAcnO Isopropyl acetate-AcpO Acetic anhydride
14
5
la
...
la
0.08
6
1.3
...
5
0.3
1
5
0.4
1
6
1.3
12
-4rbitrary value assigned for comparison of catalyst in different solvents. Equimolar cal alyst concentrations do not have same catalytic activity in acetic acrid-acetic anhydride. 'a
Acetic anhydride (0.154M). A Isopropyl acetate-acetic anhydride (0.462M). Ethyl(0.1 54M). ene chloride-acetic anhydride 0 Ethyl acetote-acetic anhydride (0.61M ) . 0 Acetic acid-acetic anhydride ( 0 . 2 0 M ) .
-
-
A
0
Perchloric Acid. At equivalent ester-
Table IV. Catalytic Activities for Acetylation Are Strongly Influenced by Solvent
-
-
Viscosity Reduction at Equivalent Esterification Rates
ification rates (Figure 2,A), there is a sharp contrast between the loss of viscosity (Figure 2,B) in acetic acid-acetic anhydride and absence of measurable change in acetic anhydride, dioxaneacetic anhydride, ethylene chlorideacetic anhydride: ethyl acetate-acetic anhydride, or isopropyl acetate-acetic anhydride. Methanesulfonic Acid. .4s a group, ebsterificationrates with methanesulfonic acid (Figure 3,A) are slower than with perchloric acid (Figure 2J), but reduction of viscosity is faster under the influence of methanesulfonic acid in all solvents except acetic acid-acetic anhydride (Figure 3,B). Again, the least change in viscosity occurred in acetic anhydride, and most in acetic acidacetic anhydride. T h e latter would have been even greater if more catalyst had been used to make the esterification rate more nearly equivalent to that in the other three solvents (Figure 3 4 ) . The most conspicuous difference is in the case of ethylene chloride-acetic anhydride, ethyl acetate-acetic anhydride, and isopropyl acetate-acetic anhydride; much more loss of viscosity occurred than with perchloric acid. Sulfoacetic Acid. This catalyst produces effects similar to methanesulfonic acid (Figure 4,A and B ) .
1
I
I
1
1
1
REACTION
I
A A* I
IIME (HOURS) I
I
ga ap
50 0
8
16
24
32
REACTION TIME (DAYS)
Figure 4. Sulfoacetic acid effects are similar to those of methanesulfonic acid Acetic anhydride ( 0 . 0 0 0 5 M ) . Ethyl acetate-acetic anhydride ( 0 . 0 0 5 8 M ) . A Isopropyl acetate-acetic anhydride (0.0058M). A Dioxane-acetic anhydride (0.070M). 0 Acetic acid-acetic anhydride ( 0 . 0 0 5 8 M ) .
These results show that chain cleavage is dependent on the solvent and the catalyst. There is consistently more reduction of viscositk- in acetic acidacetic anhydride than in the other solvent-acetic anhydride mixtures or in acetic anhydride. Therefore acetic acid-acetic anhydride is a convenient solvent for which to summarize the effect of catalysts a t equivalent esterification rates (Figure 5 , A and B ) . Reduction in viscosity is least with perchloric acid, and increases in the order methanesulfonic acid, methanedisulfonic acid, sulfoacetic acid. Esterification Rates and Catalytic Activities
Relative catalytic activities can be deduced by comparing the amounts of catalyst which produce the same esterification rate (Table 11)--if O.lhi and 0.5M catalyst produce the same rate in solvents A and B, respectively, the catalyst is five times as active in A as in B. If the activity of each catalyst in acetic acid-acetic anhydride is assigned a n arbitrary value of 1, the relative values in the other solvents are as listed in Table IV. T h e amounts of perchloric acid added were very small (Table 11). T h e effective concentrations were even VOL. 51, NO. 12
DECEMBER 1959
1485
smaller, because impurities used u p some of the catalyst, the amount varying with the solvent system. Some loss of catalyst was due to neutralization by the small amount of disodium hydrogen phosphate present in the ester as a stabilizing salt (0.04 part per 100 parts of ester). I n the case of ethylene chloride-acetic anhydride, ethyl acetate-acetic anhydride, and dioxane-acetic anhydride, more than 0.0002M, 0.0003M, and 0.0004M perchloric acid, respectively, had to be added before esterification was initiated. Corrections of this order of magnitude would make the effective concentrations of perchloric acid much different from the values shown in Table 11, except for acetic acid-acetic anhydride, where the catalyst concentration was higher (0.002M). O n this basis, the relative catalytic activities for perchloric acid are probably much larger than those listed in Table I V . T h e concentrations of methanesulfonic acid are so large in relation to the amount of catalyst neutralized by impurities, that corrections would be insignificant. This is also true for sulfoacetic acid, except with acetic anhydride alone (Table 11). These catalytic activities should be interpreted with caution, because approximations are involved in some cases, and they refer to one set of conditions. They may be influenced by many factors, including ester and catalyst concentration. Table I V cannot be used for comparison of the three catalysts, because esterification rates were not adjusted to the same level. I n comparing the activities in 90% solvent-lOyo acetic anhydride with those in acetic
3
20
0
401 0
40 60 80 100 REACTION TIME (MINI
I
I
4
8
I
I
I
12 16 20 REACTION TIME (DAYS)
I
24
120
I
28
'
Figure 5. Reduction in viscosity is least with perchloric acid in acetic acid-acetic anhydride
0
Perchloric acid (0.00105M). W Methanemonosulfonic acid (0.20M). A Methanedisulfonic acid (0.003M). 0 Sulfoacetic acid
(0.067M).
1486
anhydride alone, the change in concentration of the esterifying agent should be kept in mind. Kevertheless, these results show that catalytic activities are strongly influenced by the solvent. Methanesulfonic and sulfoacetic acids are equally or more active for acetylation in acetic acidacetic anhydride than in the other combinations of solvent-acetic anhydride. However, perchloric acid is much less active in acetic acid-acetic anhydride than in the other systems (Table IV). This anomaly is compounded by the fact that perchloric acid is much more active toward chain cleavage in acetic acid-acetic anhydride than in the other solvents, a t the same esterification rates (Figure 2,B). Another surprising result is seen in a comparison of acetic anhydride alone with the solvent-acetic anhydride systems (Table IV). Sulfoacetic acid shows a marked increase in activity in acetic anhydride; this is not the case for perchloric or methanesulfonic acid. T h e catalyst concentrations required to produce the same esterification rate in different solvents do not cause the same changes in viscosity. This is illustrated by sulfoacetic acid, where e q u i m o l a r a m o u n t s of c a t a l y s t (0.0058 M ) produced equivalent esterification rates (Figure 4,A) in acetic acid-acetic anhydride and isopropyl acetate-acetic anhydride, but much less viscosity reduction in the latter (Figure 4 , B ) . This was confirmed a t a higher concentration of sulfoacetic acid (0.067M) (Figure 6,A and B ) . O n e possible explanation is that esterification and chain cleavage are controlled by different factors. O n the other hand, if the reaction mixtures had been adjusted to equal acidities during esterification, these may have shifted to different values as esterification became complete. T h e acidity is influenced by water ( 2 ) and by other hydroxylic molecules including cellulose (4). Changes in acidity as hydroxyl is consumed may vary with the catalyst and solvent, so that the reduction in viscosity may have been studied under conditions differing from those a t the start of the reaction. There is little or no change in viscosity when the reactions are run in acetic anhydride with any catalysts (Figures 2 to 4 3 ) . Therefore, the observed losses in viscosity must be due, a t least indirectly, to the other solvent component or the catalyst. In the absence of catalyst, the hydroxyl content of the ester changed slowly and the viscosity remained essentially constant after 29 days in the solvent-anhydride mixtures (Table V). These results cannot necessarily be extrapolated to apply to heterogeneous reactions containing cellulose as the starting material, but the more definite
INDUSTRIAL AND ENGINEERING CHEMISTRY
REACTION TIME ( M I N )
0
2
4 6 8 1 REACTION TIME [DAYS)
0
Figure 6. When equimolar catalyst concentrations produce the same esterification rates in different solvents, the viscosity reduction i s not always the same 0.067M sulfoacetic acid. A Isopropyl acetateacetic anhydride. 0 Acetic acid-acetic anhydride.
Table V. Hydroxyl Content Changed Slowly and Viscosity of Ester Was Essentially Constant after 29 Days in Solvent-Anhydridewithout Catalyst
Inherent
Wt. 70 Viscositya Total in MeC12-
Acetic acid-AcZO Dioxane-AcnO Ethylene chlorideAczO Ethyl acetate-AczO Isopropyl acetate-Ac20 (Original ester)
Hydroxyl
MeOH
2.49 2.56
1.46 1.45
2.33 2.43 2.55 (3.12)
1.46 1.44
1.45 (1.47)
Small changes from original viscosity are due to increases in weight of glucose unit resulting from esterification. Inherent viscosity measurements are made on standard weight of sampIe, regardless of degree of substitution.
trends provide a basis for examining such systems. literature Cited
(1) Cragg, L. H., J . Colloid Sci. 1, 261 (,1 946) -.-,. ~ (2) Ludwig, F. J., Adams, K. H., J. Am. Chem. Sac. 76, 3853 (1954). ( 3 ) Malm, C. J . , Tanghe, L. J., Laird, B. C., Smith, G. D., Anal. Chem. 26, I88 (1954). (4) Tanghe, L. J., unpublished data. RECEIVED for review May 4, 1959 ACCEPTED August 6 , 1959 Division of Cellulose Chemistry, 135th Meeting, ACS, Boston, Mass., April 1959.
