Regioselectively Substituted 6-O- and 2,3-Di-O-acetyl-6-O

Two kinds of regioselectively substituted cellulose derivatives, i.e., 6-O-triphenylmethylcellulose (6TC) and 2,3-di-O-acetyl-6-O-triphenylmethylcellu...
0 downloads 0 Views 141KB Size
Download PDF
Biomacromolecules 2001, 2, 991-1000

991

Regioselectively Substituted 6-O- and 2,3-Di-O-acetyl-6-O-triphenylmethylcellulose: Its Chain Dynamics and Hydrophobic Association in Polar Solvents† Yoshisuke Tsunashima,* Kimihiko Hattori,‡ Hiroyuki Kawanishi,§ and Fumitaka Horii Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan Received March 27, 2001

Two kinds of regioselectively substituted cellulose derivatives, i.e., 6-O-triphenylmethylcellulose (6TC) and 2,3-di-O-acetyl-6-O-triphenylmethylcellulose (2,3Ac6TC), were prepared via cellulose. In these samples, C-6 position hydroxyls in the anhydroglucose units (AGU) along the cellulose chain were selectively substituted by the hydrophobic triphenylmethyl groups, but C-2 and -3 position hydroxyls remained in 6TC or were substituted completely by O-acetyls in 2,3Ac6TC. Their chain dynamics in polar solvents, dimethyl sulfoxide (DMSO) and N,N-dimethylacetamide (DMAc), in dilute solution were investigated by dynamic light scattering in the viewpoint of cluster formation. The results were compared with those of cellulose diacetates (CDA) in DMAc where three hydroxyls in the AGU were statistically substituted up to 2.44 by O-acetyls but hydroxyls at C-6 positions remained predominantly. It was found that 6TC and 2,3Ac6TC formed a dynamic structure about 10 times larger than single chains and that the structure would be a temporary and local association due to concentration fluctuations (dynamic structures) which were originated from the hydrophobic interactions between intermolecular triphenylmethyl groups. The dynamics and structures were in clear contrast to those of CDA where a solvent-mediated hydrogen bonding between intermolecular C-6 position hydroxyls was essential to cluster formation. The present structures were so weak as to dissipate easily under low shear field. Introduction Cellulose is a structural polysaccharide and is insoluble in organic solvents because of its β-1,4-glucosidic bonds between the anhydroglucose units (AGU). Cellulose acetates (CA) are cellulose derivatives of which hydroxyl groups at C-2, -3, and -6 positions in each AGU are substituted by O-acetyl groups to within three in the degree of substitution (DS). The solubility of CA in various solvents depends strongly on the distribution of hydroxyl groups in the AGU, i.e., the individual degree of substitution (IDS) of three hydroxyls, as well as the total DS (TDS). Moreover, very recently, we have found that1 CA is generally in a quasiflexible chain state in solution and that the chain conformation depends on the solution conditions and on the architecture of CA chain. Here the chain architecture denotes the sequential distribution of the AGUs with specified IDS along a chain contour. The quasi-flexibilty of the chain is also denoted as follows; the chain stiffness is between flexible and usual semiflexible chains, and in addition, the stiffness in solution is not constant; i.e., it is not intrinsic to the chemical structure of the chain but is regulated by the environmental factors such as solvent quality,1 concentration,2 * To whom correspondence should be addressed. † Presented at the 38th Macromolecular IUPAC Symposium, Warsaw, Poland, July 9-14, 2000. ‡ Present address: Mitsubishi Rayon, Co., Ootake, Hiroshima 739-0693, Japan. § Present address: Fuji Photo Film, Co., Minami-ashigara, Kanagawa 250-01, Japan.

temperature,3 external fields,1,4 and so on. In other words, the quasiflexibility of a single CA chain in solution is specified by double effects of the chain stereoregularity, which induces the inherent helicalicy in CA chains, and the intramolecular hydrogen bonds that are correlated with the environmental conditions. On the other hand, chain clustering occurs in solution due to the intermolecular hydrogen bonds, which are also mediated by solvent polarity and by the surroundings. Thus, the modest solubility and/or cluster formation of CA in solution, which are usual for CA, is deeply related to the intra- and intermolecular hydrogen bond formations that would be controlled by both the chain architecture and the surroundings. Although these strange tendencies of CA in solution have sometimes been reported in several papers,5-12 but no direct discussion has been done in detail on the relationship between the surroundings and the chain architecture. Our previous studies1-4,13,14 may be the first, in our knowledge, to discuss quantitatively this strangeness in terms both of the nonequal distribution of hydroxyls in the AGU and of the sequential distribution of various types of AGUs along the chain contour. The results were following. For randomly substituted cellulose diacetate (CDA) of TDS ) 2.44, the hydroxyls at three positions in AGU were not substituted equally by O-acetyls, i.e., the unsubstituted hydroxyls at C-2, -3, and -6 positions were 25%, 25%, and 75%, respectively,13,14 and the predominance of hydroxyls at C-6 position resulted in formation of two huge associated structures in a polar solvent, N,N-dimethylacetamide (DMAc).

10.1021/bm010069e CCC: $20.00 © 2001 American Chemical Society Published on Web 06/19/2001

992

Biomacromolecules, Vol. 2, No. 3, 2001

Tsunashima et al.

Figure 1. (a) Scheme for regioselective preparation of cellulose acetates from cellulose and 13C NMR spectra for (b) randomly substituted commercial cellulose acetate of DS ) 1.75 and (c) regioselectively substituted 2,3-di-O-acetylcellulose (2,3AcC) in DMSO-d6 at 100 MHz at 40 °C.

These structures coexisted always with molecularly dispersed single chains. The associated structure was not a steadystable but a temporary cluster of concentration fluctuations ()dynamic structures) in a creation-dissipation mechanism, the clustering being motivated by hydrogen bonds between intermolecular C-6 position hydroxyls directly or via polar solvents.1,15 Expressing the length over which the fluctuations correlate in the solution by a size ξ, we have obtained that ξ was larger by 2-4 order of magnitude than single chains, though the length does not always represent the real size of structures. The amount and the sequential distribution of C-6 position hydroxyls on CDA affect strongly the way of clustering in solution. It is thus essential to control regularly the sequence of C-6 position hydroxyls, as well as the IDS at C-6 positions. The control can be achieved by the

regioselective substitution of hydroxyl groups in cellulose by O-acetyl groups.16 Actually various types of polysaccharide derivatives have been prepared by this method. Changing our viewpoint to analyze the substituent patterns in cellulose derivatives, we could emphasize recent developments on mathematical models for the pattern calculation in polysaccharide derivatives,17-21 where various cellulose and amylose ethers have been studied in excellent agreement with the experimental data. We prepared three cellulose derivatives by the regioselective substitution; 6-O-triphenylmethylcellulose (6TC), 2,3-di-O-acetyl-6-O-triphenylmethylcellulose (2,3Ac-6TC) and 2,3-di-O-acetylcellulose (2,3AcC ) regioselectively substituted CDA).14 Figure 1a exhibits the scheme for regioselective preparation of these cellulose derivatives from

6-O-Triphenylmethylcelluloses

cellulose. The 6TC chain was characterized by the architecture that every C-6 position hydroxyl in the chain was completely taken place by the hydrophobic trityl ()triphenylmethyl ) Tr) group and that all the C-2,3 positions hydroxyls remained; i.e., IDS ) 0 at C-2 and -3 positions and [IDS by Tr] ) 1 at C-6 position. On the other hand, 2,3Ac6TC was characterized by IDS ) 1 at C-2 and -3 positions and [IDS by Tr] ) 1 at C-6 position. Thus, 6TC and 2,3Ac6TC are homopolymers composed of a single kind of repeating AGU, there being an evenness of the substituents along the chain. They are expected not to form intermolecular hydrogen bonds, or not to induce any association, in polar solvents because of the lack of C-6 position hydroxyls. As indicated already for randomly substituted CA,1,15 the intramolecular hydrogen bonds are easily formed between the C-3′ position hydroxyl and the neighboring O-5 ring-position oxygen and/ or between the C-2 and the neighboring C-6′ position hydroxyls. These intramolecular interactions give strong influence on the chain conformation and the stiffness in solution, while the C-6 position hydroxyls are essential to formation of intermolecular hydrogen bonds, or associations. In the present article, we thus took 6TC and 2,3Ac6TC as typical samples of the even substituents along the chain and of lacking in the C-6 position hydroxyls. We then tried to investigate their dynamic behavior in polar solvents and to certify the essential role1,15 of C-6 position hydroxyls on intermolecular associations of cellulose derivatives. The dynamic light scattering (DLS) technique was applied to dilute solutions of 6TC in quiescent equilibrium state and in the nonequilibrium state under Couette (a simple shear) flow field. For 2,3Ac6TC, on the other hand, DLS measurements were carried out only in the equilibrium state. Experimental Section Materials and Preparation of Solutions. 6TC was prepared from cellulose via two steps (see Figure 1a for the scheme).14 First, cellulose (CF11, Whatman) was added to a mixed solution of 7.5% lithium chloride/DMAc and heated to 150 °C over 1-1.5 h. The solution was kept for 20 min at 140-150 °C and cooled slowly down to room temperature over 24 h to obtain a clear viscous solution, followed by agitation for 2-3 days to achieve complete dissolution of cellulose. Second, a mixed solution of trityl chloride (2.5 mol/1.0 mol of cellulose)/pyridine was added to the above cellulose solution and heated to 90 °C. Tritylation was carried out for 4 h at 90 °C, and then the solution was cooled slowly to room temperature. These reactions at high temperature were made under Ar atmosphere. The reaction mixture was poured into methanol to obtain 6TC as white precipitant. The precipitant was purified by repeating washings with methanol and by filtration and finally dried under vacuum. The identification of 6TC was made by elementary analysis of C, H, and N, and by 13C NMR spectra16,22 conducted in dimethyl sulfoxide (DMSO-d6) at 100 MHz (see below). The molecular weight of 6TC was determined by side exclusion chromatography (SEC) with TSK-GEL R-M column (Tosoh) in DMAc at 40 °C. It would be no use in determining the molecular weight of 6TC by static light scattering because

Biomacromolecules, Vol. 2, No. 3, 2001 993

6TC coexisted always with its dynamic associates in solution, as shown below. The obtained weight-averaged molecular weight Mw was 1.87 × 105 with Mw/Mn ) 1.2 (based on the polystyrene standards). 2,3Ac6TC was prepared by homogeneous acetylation of 6TC obtained above.14 The 6TC was first dissolved in pyridine and heated at 90 °C under argon atmosphere, followed by selective substitution of C-2 and -3 hydroxyls by O-acetyls with acetic anhydride for 48 h. The reaction mixture was cooled and then poured into methanol. 2,3Ac6TC was obtained as a white precipitate, which was purified by the same procedure as described for 6TC. It was identified by elementary analysis of C, H, and N, and by 13C NMR spectra16,22 measured in dimethyl sulfoxide (DMSO-d6) at 100 MHz. Figure 1c shows, as an example, 13C NMR spectra of 2,3-di-O-acetylcellulose (2,3AcC) prepared from 2,3Ac6TC by regioselective substitution of 6-trityls by hydroxyls, while Figure 1b demonstrates spectra of randomly substituted commercial cellulose acetate (RC) of TDS ) 1.75. The C1, C4 hydroxyl signals for unsubstituted C-2,3 positions and the C6′ hydroxyl signal for substituted C-6 position in RC are not observed in Figure 1c, which certifies to identification of 2,3AcC; i.e., C-2 and -3 positions are substituted but C-6 position is unsubstituted by O-acetyls. The polar solvent DMSO was of specially prepared spectro-grade (Nacalai Tesque) and was kept in a dry-sealed bottle until use. Its purity was checked by measuring the refractive index n0 for the Na D-line at 30 °C (Pulfrich refractometer, Shimadzu) with good agreement of the measured value 1.47491 with the reference value 1.47486. The n0 value at 488 nm at 30 °C was obtained to be 1.48243 by interpolation of the three measured n0 values for Na D-line (589.3 nm) and Hg g- and e-lines (435.9, 546.1 nm) at 30 °C. The density F0 and the viscosity η0 were 1.091 g cm-3 and 1.814 cP at 30 °C, respectively. Another polar solvent, DMAc, was prepared, as described elsewhere,1 by purifying an extrapure reagent one (Tokyo Kasei) through column distillation. Its purity was ascertained1 with the characteristics n0(488 nm) ) 1.4405, F0 ) 0.932 g cm-3, and η0 ) 0.838 cP at 30 °C. An original solution of a given mass concentration c0 was prepared for each system of 6TC/DMSO and 2,3Ac6TC/ DMAc solutions. That is, a polymer sample and the corresponding polar solvent were mixed by weight at room temperature and held in a drybox at 35-40 °C for about 1 week with intermittent stirring. The c0 values were 6.95 × 10-3 and 6.92 × 10-3 g cm-3 at 30 °C for 6TC/DMSO and 2,3Ac6TC/DMAc, respectively. It is worthwhile to note here that 6TC can dissolve in DMSO but not in DMAc, and vice versa for 2,3Ac6TC. Solutions of lower mass concentration were prepared by mixing the original solution and solvent in LS cells through 0.45 and 0.2 µm pore-size filters (Sartorius), respectively. These solutions were kept for few days at 35 °C with intermittent stirring and stored at about 30 °C just prior to use. Dynamic Light Scattering in the Equilibrium State and in Simple Shear Flow. DLS measurements in quiescent state were made through a laboratory-made DLS system23 at the scattering angle θ ) 10-150°, to which a multiple-τ digital

994

Biomacromolecules, Vol. 2, No. 3, 2001

Tsunashima et al.

correlator (ALV-5000/E) was connected. A single-frequency 488 nm line emitted from an etalon-equipped Argon-ion laser (Spectra Physics, 3W) was put into a test solution, and the Vv component of the scattered light from the solution was measured for 30 min by the homodyne method at 30 °C. DLS in Couette flow field was also carried out through another laboratory-made apparatus of a concentric rotating cylinder viscometer24-26 at a fixed angle θ ) 75.6° at 30 °C. The solution to be tested was filtered again just prior to measurement through a 0.2 µm pore-size filter (Sartorius) into the gap (2.0 mm) between concentric rotating and stationary cylinders. The shear rate γ of 0.132-7.93 s-1 was applied to the solution by rotating the inner rotor with the eddy current, which was induced by the electromagnetic interaction between a copper ring in the rotor and the outer rotating permanent magnet. The shear rate and the solution viscosity under DLS measurements were evaluated by counting simultaneously the number of revolution of the rotor through an outer observation system. The normalized time correlation function of the scattered-light intensity g(2)(t) was measure by the homodyne method for 20 min just 5 min after the onset of shear. Analyses of the measured g(2)(t) were made by the CONTIN and/or the inverse Laplace transformation (ILT) methods, and the decay rate distribution G(Γ) was obtained with good consistency to each analysis. In the present analyses, the decay rate function used was of singleexponential decay form, not of stretched exponential one. Therefore, G(Γ) can be defined by the single-exponential decay function g(1)(t) as g(2)(t) ) 1 + a|g(1)(t)|2, g(1)(t) )

∫ G(Γ) exp(-Γt) dΓ

(1)

Γ is the decay rate of motion and a is a measure representing the degree of coherency on the scattering volume (0 < a e 1). In the case where G(Γ) can be given by i pieces of discrete peaks, as was true in the present case shown below, we denote each peak as the ith mode and represent its motion by the mean decay rate Γi. The evaluation of Γi can be made by averaging Γ only over the ith discrete distribution Gi(Γ). Assignments of Dynamic Processes Observed under Couette Flow Field. The dynamic processes measured by DLS under simple shear flow has already been discussed by one of us with typical materials, polystyrene-latex particles in aqueous solution24 and flexible polymers in good solvent (polystyrene25 and poly(R-methylstyrene)26 in benzene). It has been made clear that the particle diffusion, the solution viscosity ηsoln, the direction of the scattering vector to the shear velocity q‚W, and the magnitude of applied shear rate γ were the keys to restrain observable chain dynamics.24,26 The theoretical study on g(2)(t) for dispersed particles under Couette flow has shown that the diffusion motion of particles competes with the convection motion and g(2)(t) takes the form24,26 g(2)(t) ) 1 + exp[-2(t /τD)H(t) - (1/2)(t/τS)2]

(2)

H(t) ) 1 + (t/τf) sinθq cosθq cosφq + (1/3)(t/τf)2 sin2θq cos2φq (3) τD ) 1/Dq2, τS ) 1/qvγw, τf ) 1/γ

(4)

Figure 2. (a) g(2)(t) vs logarithmic time t plot for a solution of 6TC in DMSO (c ) 6.95 × 10-3 g cm-3) in quiescent state at 30 °C. The scattering angle θ is 10, 30, and 150°. (b) g(2)(t) vs linear t plot for the above data at θ ) 30°.

Here τD is the time scale for the Brownian diffusion of the particle, and the time scales τS and τf are associated with the convection motions of particles. The convection induces generally a sigmoidal oscillating feature in the g(2)(t) profile. w is the laser width in the scattering volume, and θq and φq are the angles specifying the LS optical geometry.24 The effect of convection of particles on g(2)(t) could be reduced to a negligible level when no sigmoidal feature appears in the linear plot of g(2)(t) vs t under the criterion that τS/τD > 1/ , which was discussed in refs 24 and 26. In this situation, 2 the stretching and deformation of flexible polymers would be little detected in the observable amounts, and the translational diffusion and the segmental motions, if qRG >1, could be deduced from g(2)(t) without ambiguity. Here RG is the root-mean-square radius of gyration of the polymer. The present study under Couette flow was made in such the condition as specified above. Thus, the analyses of g(2)(t) under Couette flow were performed with the same procedures as done in the quiescent state. Results and Discussion A. Dynamics of 6TC in DMSO. Two Types of Diffusion Motions in the Quiescent State. Figure 2a shows, as an example, g(2)(t) data obtained in quiescent state for a 6TC solution of c0 ) 6.95×10-3 g cm-3 at three scattering angles, θ ) 10, 30, and 150°. Here g(2)(t) is plotted against the logarithmic time t. The same data at θ ) 30° are also plotted against the linear time t in Figure 2b to demonstrate that the data points show no sigmoidal oscillating decay feature. g(2)(t) gives a high coherency with the large amplitude a (see eq 1) and exhibits high accuracy in the analyses of G(Γ). The decay rate distribution G(Γ) deduced by the CONTIN method at θ ) 150° is plotted against the decay rate Γ in Figure 3. There are two discrete peaks, which are called slow and fast

6-O-Triphenylmethylcelluloses

Figure 3. Decay rate distribution G(Γ) plotted against the decay rate Γ for a solution of 6TC in DMSO (c ) 6.95 × 10-3 g cm-3) at θ ) 150°.

Figure 4. Variation of G(Γ) as a function of the scattering angle θ for a solution of 6TC in DMSO (c ) 6.95 × 10-3 g cm-3) at 30 °C. Two peaks are denoted as slow and fast modes in increasing order of Γ.

modes in increasing order of Γ, and the corresponding mean decay rate is denoted as Γs and Γf, respectively, as mentioned above. In Figure 4 the distributions G(Γ) at other angles θ ) 10-150° are also demonstrated. It is found that two modes appear at all θ measured. Fast mode appears clearer at higher scattering angles, while slow mode is detectable at each θ with nearly the same amplitude. Clustering by Hydrophobic Interactions. The Γs and Γf values estimated are plotted against the squared scattering vector q2 for solutions of different concentration c in Figure 5. For each c, the data points are represented well by a straight line passing through the origin. Both fast and slow modes are diffusion motions accordingly, and the slope gives the diffusion coefficient at finite c, i.e., D(c). To determine D(c) more precisely, we replotted Γ(c)/q2 as a function of q2 for each c and evaluated D(c) as a constant Γ(c)/q2 value independent of q2. D(c) thus evaluated is plotted against c for fast and slow modes in Figure 6. The data points for each mode give a straight line of negative slope, which is

Biomacromolecules, Vol. 2, No. 3, 2001 995

Figure 5. Mean decay rates for fast and slow modes, Γf and Γs, plotted against the scattering vector q2 for 6TC in DMSO at 30 °C. The polymer mass concentrations c/10-3 g cm-3 are as follows: (O) c0 ) 6.95; (4) c1 ) 2.78; (0) c2 ) 1.53.

Figure 6. D(c) vs c plots for fast and slow modes of 6TC in DMSO at 30 °C.

expressed by kD < 0 in the equation D(c) ) D0 (1 + kDc)

(5)

Here D0 is the diffusion coefficient at infinite dilution and kD is the dynamic second virial coefficient, or a measure of hydrodynamic intermolecular interactions, of 6TC in DMSO. The negative kD for fast and slow modes indicates that the attractive interaction works between 6TC in the present solution; the solution being not in a good but in a poor solvent state and having a tendency toward cluster formation. The D0 and kD values thus estimated for fast and slow modes are summarized in Table 1, together with the fluctuating length ξ(c) which is defined tentatively by ξi(c) ) kBT/6πη0Di(c) for mode i (i ) s and f). Here kB is the Boltzmann constant and T is the absolute temperature. The magnitude of ξf(c), 9.3-10.1 nm, indicates that fast mode is the diffusion of molecularly dispersed 6TC chains and ξf(c ) 0) results in the Stokes radius RH of single 6TC chains. The reason is that this size is comparable to RH(c) ) 10-12

996

Biomacromolecules, Vol. 2, No. 3, 2001

Tsunashima et al.

Table 1. D(c), ξ(c), and the Infinite Dilution Values D0 and kD for 6TC in DMSO in Quiescent State at 30 °C mode fast

slow

concn, 10-3 g cm-3

D(c), 10-8 cm2 s-1

RH(c), ξ(c), 10-7 cm

6.95 2.78 1.53 0.00 6.95 2.78 1.53 0.00

12.17 12.34 13.30 13.22 0.808 0.946 1.004 1.052

10.1 9.99 9.27 9.32 153 130 123 117

kD, cm3 g-1

-12.6

-33.7

nm obtained for molecularly dispersed CDA in DMAc (Mw ) 1.70 × 105) in dilute solution at 30 °C1,13 and that the DP 460 estimated from the present 6TC with Mw ) 1.87 × 105 and with its AGU mass 404 is comparable roughly to DP 270 for the CDA. The dissolution of CDA in DMAc to the molecular level has already been certified in the ultracentrifugal field of sedimentation equilibrium/velocity and in the capillary viscometric field of γ = 103 s-1, where the external forces broke dynamical clusters into molecular pieces.1 On the other hand, ξs(c ) 0) for slow mode is 117 nm, which is about 12 times larger than the size of single chains and is in the similar order of magnitude of a smaller dynamic structure detected in CDA/DMAc.13 The latter fact might make us imagine that there were a cluster formation mechanism similar to that in the previous CDA/DMAc system13 where the intermolecular hydrogen bond played an essential role in clustering and the induced two dynamical clusters had the sizes larger by about 2 and 4 order of magnitudes than the single chains, respectively. In the present 6TC/DMSO system, however, there is only one dynamical cluster and the size is at most 10 times larger than single chains. In addition, no hydroxyl exists at the C-6 positions in the 6TC chains, trityl groups sitting there instead. The present situation is thus in clear contrast to the CDA/DMAc system. In 6TC/DMSO, the trityls of hydrophobic nature would be squeezed out of a polar-solvent medium, DMSO, and would huddle together. Therefore, the cluster in 6TC/ DMSO could be recognized as a representative of a dynamic association that was formed by the intermolecular hydrophobic interactions between C-6 position trityls, because bulky trityl groups should be protected from precipitation in the hydrophilic DMSO atmosphere. It would result in formation of a temporary hydrophobic core covered with a shell of many hydrophilic hydroxyls that exist at C-2 and -3 positions in 6TC. This hydrophobic interaction is of nondirectional nature, and so the association would be very weak compared to the structure formed by CDA in DMAc.1-4,13 These expectations will be certified by the results described below. Here it should be noted that the dynamic association does not mean a thermodynamically stable structure but a temporary buildup of 6TC concentration due to concentration fluctuations in solution, and the buildup decreases down to the level of average (bulk) concentration with the decay rate Γs. Thus, the diffusion gives the so-called cooperative diffusion coefficient Dcoop, and the correlation length ξ

Figure 7. (a) g(2)(t) vs logarithmic time t plot for a solution of 6TC in DMSO (c ) 1.53 × 10-3 g cm-3) under shear of γ ) 0.980 s-1 at 30 °C. (b) Same data plotted against the linear time t. No significant sigmoidal trend is observed in g(2)(t).

defined by Dcoop ) kBT/6πη0Dcoop represents the length within which the fluctuation correlates. Behavior in Shear Flow. Figure 7 shows, as an example, g(2)(t) data obtained under shear of γ ) 0.980 s-1 for a 6TC solution of c ) 1.53 × 10-3 g cm-3 and θ ) 75.6°. The time t of the abscissa is expressed in the logarithmic and the linear scales in Figure 7, parts a and b, respectively. Figure 7b demonstrates that the data points show no significant sigmoidal oscillating decay feature and that there is no convection effect with the satisfaction of τS/τD > 1/2. The deduced decay rate distribution G(Γ) gives two separate peaks as shown in the middle part (γ ) 0.980 s-1) of Figure 8. The same analytical procedure was applied to the g(2)(t) data at other shear rates to check the convection effect. The results confirmed that τS/τD was in the range of 1/2 ∼ 8 at γ < 7.93 s-1 and was slightly smaller than 1/2 at γ ) 7.93 s-1. Thus, almost all g(2)(t)s were free from convection in the shear range of 0-7.93 s-1 used. Convection-free g(2)(t)s at every γ were analyzed by the same procedure as described above and the results are summarized in Figure 8. Here the decay rate distribution G(Γ), evaluated from the ILT method, is plotted against Γq-2 as a function of γ for the solution of 1.53 × 10-3 g cm-3 at 30 °C. There are two peaks in G(Γ) in the range of γ ) 0.285-0.980 s-1 but only one at γ g 1.37 s-1. In other words, the slow mode of the smaller Γq-2, which was also detected in equilibrium state γ ) 0 (see Figures 3 and 4), shifts its position rightward and finally disappears with the increase of γ. On the other hand, the fast mode of the larger Γq-2 retains its position around 1 × 10-7 cm2 s-1. The increase in Γq-2 of the slow mode, i.e., an apparent larger mobility of the slow mode, with increasing γ could be considered as follows. Slow mode would be the dynamic structure due to local concentration fluctuations, and so the structure would be temporary and a creation-dissipation process of clustering could happen repeatedly in solution. It

6-O-Triphenylmethylcelluloses

Biomacromolecules, Vol. 2, No. 3, 2001 997

Figure 9. Relative amplitude of fast and slow modes plotted against shear rate γ for a solution of 6TC in DMSO (c2 ) 1.53 × 10-3 g cm-3) at 30 °C.

Figure 8. Variation of G(Γ) with shear rate γ plotted against Γq-2 for a solution of 6TC in DMSO (c2 ) 1.53 × 10-3 g cm-3) at 30 °C. Two peaks are denoted as slow and fast modes in increasing order of Γ/q2.

is thus conceivable that the structure would be more fragile at larger γ and could brake into a molecularly dispersed single 6TC chain () fast mode) around γ > 1 s-1. This drastic change in dynamic structures can be demonstrated more clearly in Figure 9, where the relative amplitudes of fast and slow modes are plotted against γ. A threshold γthr, above which the destruction of structure occurs, is found to be 1.1 s-1. This value is a contrast to γthr = 103 s-1 for CDA in DMAc, where CDA comes to disperse molecularly.13 Moreover, the value γthr ) 1.1 s-1 is surprisingly low when compared with a critical shear rate for conspicuous chain deformation reported in nonequilibrium molecular dynamics (NEMD) simulations27,28 or in Brownian dynamics (BD) simulations29,30 under rheological fields. The rotation of the chain around its center of mass would be caused in the gradient shear field of Couette flow. If it exists, the effect could appear in g(1)(t) for the Vv component of the scattered light as31 g(1)VV(t) ) g(1)ISO(t) + (4/3)g(1)VH(t)

(6)

g(1)ISO(t) ) 〈N〉R2 exp[-Dq2t]

(7)

g(1)VH(t) ) (〈N〉β2/15) exp[-(Dq2 + 6Θ)t]

(8)

Here the analogy of an anisotropic particle is used under the assumption that particle rotation and translation are independent in dilute solutions. g(1)ISO(t) and g(1)VH(t) are the isotropic and the anisotropic parts of the time correlation function due to the isotropic R ) (1/3)(R| + 2R⊥) and the anisotropic β ) (R| - R⊥) polarizabilities, respectively, and 〈N〉 and Θ are the average number of particles in the scattering volume and the rotational diffusion coefficient, respectively. Equations 6-8 fail when the coupling between

translational and rotational diffusion occurs. The coupling could be discussed by an ellipsoid of revolution. However, if the anisotropy difference between the parallel and the perpendicular diffusion coefficients, (D| - D⊥) and (Θ| Θ⊥), would be small, the coupling could be ignored.31 This situation would hold for the present slow mode unless the mode is a dynamic structure of a highly anisotropic shape, or β = 0. It is convincing that (D| - D⊥) and (Θ| - Θ⊥) would be negligibly small and, with a negligible β in eq 8, there is scarcely any effect of rotational diffusion on translation diffusion, that is, g(1)VV(t) ) g(1)ISO(t) in eq 6. Stability of Dynamic Structure under Shear. The g(2)(t) profile specific to Couette flow was found to return very quickly to its equilibrium g(2)(t) form when switching off the shear. Thus, the relaxation time for the return was within a few second, which was a limiting minimum interval for the repeating measurement of g(2)(t) on the present instrument. In other words, the dynamic structure formed under shear could be realized only when an appropriate shear was applied to the solution; the shear energy might be consumed to produce the dynamic structure. Conversely, an excess shear force would destroy the structure. This would be a kind of shear enhanced clustering/dissipating phenomenon. The change of solution viscosity with shear was detected by measuring the speed of revolution of the rotor, as described above, and no shear thinning of the viscosity was detected in all γ values measured. This phenomenon could be explainable by a temporary structure, not a thermodynamically stable one, in dilute solution. Thus, it is highly convincing that the structure in 6TC/DMSO under shear is a weak dynamical structure formed transiently due to concentration fluctuations. The structure formation could be originated by an attractive interaction, or the intermolecular hydrophobic interaction operating between the C-6 position trityls in a polar solvent, DMSO. The C-2 and -3 position hydroxyls on 6TC, on the other hand, would be ineffective to formation of dynamic structure but would be helpful to form the hydrophilic corona and to stabilize the structure in solution. The size of the dynamic structure, which is estimated from the diffusion coefficient Ds(c), represents the correlation length of fluctuations and is defined by ξs(c) ) kBT/6πη0Ds(c). Figure 10 shows the variation of the correlation length for slow mode ξs and the Stokes radius for fast mode RH as a function of γ. Table 2 summarizes the results for 6TC/

998

Biomacromolecules, Vol. 2, No. 3, 2001

Tsunashima et al.

Figure 10. Shear rate dependence of the correlation length for slow mode ξs and of the Stokes radius for fast mode RH in a solution of 6TC in DMSO (c2 ) 1.53 × 10-3 g cm-3) at 30 °C. The filled symbols at γ ) 0 s-1 (b and 2) denote the fluctuation length for slow mode and RH for fast mode ()single chains), respectively, in the equilibrium state.

Figure 11. Dependence of ξs on γ for two dynamic structures detected in the present 6TC/DMSO and in the previous CDA/DMAc solutions at 30 °C:13 (O and 4) a dynamic clustering by hydrophobic interaction and the Stokes radius RH for single chains, respectively, in 6TC/DMSO; (0 and 9) a dynamic clustering made by a solventmediated hydrogen bonding at finite and zero shear rates, respectively, in CDA/DMAc.13

Table 2. Shear Rate Dependence of D(c) and ξ(c) for 6TC in DMSO in Couette Flow at 30 °C and at c ) 1.53 × 10-3 g cm-3 mode fast

slow

shear rate γ, s-1

D(c), 10-8 cm2 s-1

RH(c), ξ(c), 10-7 cm

0.00 0.132 0.285 0.693 0.980 1.37 2.71 5.45 7.93 0.00 0.132 0.285 0.693 0.980 1.37 2.71 5.45 7.93

13.2 9.43 10.8 12.0 11.9 11.9 11.9 12.6 11.8 1.004 0.852 0.616 0.911 5.10 _ _ _ _

9.27 13.0 11.3 10.2 10.3 10.2 10.2 9.64 10.5 123 145 200 135 23.9 _ _ _ _

DMSO at c ) 1.53 × 10-3 g cm-3 at 30 °C. As shown in Figure 10, ξs is 123 nm at γ ) 0 and becomes twice as large ()200 nm) when γ increases to 0.285 s-1 but decreases steeply to the level of single CDA molecules (ca. 10.5 nm) around γ ) 1.1 s-1. This trend contrasts clearly with that of CDA structures in DMAc where the hydrogen bonds between the intermolacular C-6 position hydroxyls plays a main role in the clustering and resists to shear flow. The aspect of resistance to shear is demonstrated in Figure 11 for 6TC and CDA, where the structure changes are shown by the ξs vs γ plots in the range of γ smaller than 6 s-1. When a slight shear rate of γ < 0.1 s-1 is applied to the solution, CDA forms an extremely large dynamic structure that expands to the correlation length of 105 nm. This size is larger by about 3 orders of magnitude than that of 6TC and the structure is very stable against shear within the range of 10 s-1. The breakdown of the structure into single chains occurs only above γ = 103 s-1, as described previously.13 On the other

Figure 12. Variation of G(Γ) as a function of the scattering angle θ for a solution of 2,3Ac6TC in DMAc (c ) 6.92 × 10-3 g cm-3) at 30 °C. Two peaks are denoted as slow and fast modes in increasing order of Γ.

hand, the correlation length of present 6TC in DMSO is small, 200 nm at most, and the structure is not resistible to external forces such as γ = 1.1 s-1. It might be the reason that the structure is constructed by nondirectional hydrophobic interactions which operate between trityl groups in polar solvent. B. Dynamics of 2,3Ac6TC in DMAc. Dynamic Structures. Figure 12 shows the results of the decay rate distribution obtained by the CONTIN method for a solution (c ) 6.92 × 10-3 g cm-3) of 2,3Ac6TC in DMAc at 30 °C. Two separated peaks are detected at all scattering angles θ measured. The peaks were denoted as slow and fast modes in increasing order of Γ. The slow mode appears always at all θ, while the fast mode is less observable at lower θ. The mean decay rates of each mode, Γs and Γf, are plotted against q2 for different polymer mass concentrations c in Figure 13.

6-O-Triphenylmethylcelluloses

Biomacromolecules, Vol. 2, No. 3, 2001 999

Figure 13. Mean decay rates for fast and slow modes, Γf and Γs, plotted against q2 for 2,3Ac6TC in DMAc at 30 °C. The polymer mass concentrations c/10-3 g cm-3 are 6.92 (O), 3.51 (4), and 1.79 (0).

Figure 14. D(c) vs c plots for fast and slow modes of 2,3Ac6TC in DMAc at 30 °C. Table 3. D(c), ξ(c), and the Infinite Dilution Values D0 and kD for 2,3Ac6TC in DMAc in Quiescent State at 30 °C mode fast

slow

concn, 10-3 g cm-3

D(c), 10-8 cm2 s-1

RH(c), ξ(c), 10-7 cm

6.92 3.51 1.79 0.00 6.92 3.51 1.79 0.00

25.1 24.7 23.4 23.2 2.14 2.07 2.53 2.49

10.6 10.7 11.4 11.4 124 128 105 106

kD, cm3 g-1

12.4

-24.4

The data points for each c are represented well by a straight line passing through the origin. Thus, fast and slow modes are of diffusive nature and the diffusion coefficient D(c) can be estimated from the slope of each line. Figure 14 shows D(c) as a function of c, which represents the relation D(c) ) D0 (1 + kDc) with D0 the infinite dilution value. The dynamic second virial coefficient kD is positive (repulsive interactions) for structures of fast mode, but negative (attractive interactions) for those of slow mode. The D0 values give a length ξf ) 11.4 nm and ξs ) 106 nm for fast and slow modes at infinite dilution, respectively, as are summarized in Table 3. This ξf value is comparable to that of single 6TC chains in DMSO, suggesting that the fast mode represents single 2,3Ac6TC chains. Thus, ξf(c ) 0) means

the Stokes radius of single chains; RH(c ) 0) ) 11.4 nm. Whereas slow mode would be a dynamical association of single chains, which could realize tentatively, at a given local space and time, due to concentration fluctuations in solution. Its correlation length is larger by about 10 times than the size of single chains. This situation is quite similar to the case of 6TC clusters in DMSO mentioned above. It is thus convincing that an association of 2,3Ac6TC is formed by the hydrophobic interaction between the intermolecular C-6 position trityls. States of Molecular Dispersion. Comparing kD values of single chains (fast modes) for 2,3Ac6TC and 6TC (Figures 14 and 6, and Tables 3 and 1), we can find that the former is positive (12.4 cm3 g-1) and the latter is negative (-12.6 cm3 g-1), with the same absolute magnitude. This would reveal the difference in solubility of single 2,3Ac6TC and 6TC chains; i.e., 2,3Ac6TC dissolves molecularly and stable with the repulsive intermolecular interactions (kD > 0), while 6TC is in unstable dissolution in single chains (kD < 0) and tends to associate each other. The difference would come from the affinity of C-2,3 position O-acetyls (in 2,3Ac6TC) and hydroxyls (in 6TC) to association. In this sense, C-2,3 position hydroxyls might have some effect on the intermolecular clustering of CA chains in an indirect way such that the intramolecular hydrogen bond could control the molecular dispersion of CA in solution. C. Conclusion. Regioselectively prepared cellulose derivatives 6TC and 2,3Ac6TC, whose C-6 position hydroxyls were regularly substituted by hydrophobic trityl groups, showed a formation of dynamical clusters in polar solvents in quiescent state. The cluster coexisted always with its molecularly dispersed single chain in solution. However, the cluster would be a temporarily and locally created dynamic structure ()the concentration fluctuations) due to the intermolecular hydrophobic interactions between C-6 position trityls in the AGUs. Although the size ()correlation length) was about 10 times larger than the Stokes radius of single chains, the structure was so fragile that it broke easily into single chains at a shear rate as low as 1.1 s-1. This weak resistance to shear was clearly different from the case for dynamical clusters of CDA, where the intermolecular hydrogen bonds between C-6 position hydroxyls hold the structure and resist shearing as high as 103 s-1. This fact indicate that the C-2 and -3 position hydroxyls, which exist in 6TC chains but not in 2,3Ac6TC chains, did directly take little part in intermolecular cluster formation in polar solvents. An essential key for dissolution of CA is the way of working of the inter- and intramolecular hydrogen bondings and the hydrophobic interactions in a given solution environment. References and Notes (1) Kawanishi, H.; Tsunashima, Y.; Okada, S.; Horii, F. J. Chem. Phys. 1998, 108, 6014. (2) Tsunashima, Y.; Kawanishi, H..; Nomura, R.; Horii, F. Macromolecules 1999, 32, 5330. (3) Kawanishi, H.; Tsunashima, Y.; Horii, F. Polym. Prepr. Jpn. 1998, 47, 3679. (4) (a) Kawanishi, H.; Tsunashima, Y.; Horii, F. Macromolecules 2000, 33, 2092. (b) Kawanishi, H.; Doctoral Thesis, Kyoto University, March 2000.

1000

Biomacromolecules, Vol. 2, No. 3, 2001

(5) Jeffrey, G. A.; Saenger, W. Hydrogen Bonding in Biological Systems, Springer-Verlag: Berlin, 1999; pp 33, 216. (6) Johnson, D. C. In Cellulose Chemistry and its Applications; Nevell P., Zeronian S. H., Eds.; Ellis Horwood: Chichester, England, 1985, p 181. (7) Klohr, E.; Zugenmaier, P. Macromol. Symp. 1997, 120, 219. (8) Burger, J.; Kettenbach, G.; Klu¨fers, P. Macromol. Symp. 1995, 99, 113. (9) Kamide, K. Macromol. Chem., Macromol. Symp. 1994, 83, 233. (10) Kubic, S.; Ho¨ller, O.; Steinert, A.; Tolksdorf, M.; Wulff, G. Macromol. Symp. 1995, 99, 93. (11) Balser, K. In Polysaccharides; Burchard, W., Ed.; Springer, Berlin, 1985; p 84. (12) Schulz, L.; Burchard, W. Macromol. Symp. 1995, 99, 57. (13) Kawanishi, H.; Tsunashima, T.; Horii, F. J. Chem. Phys. 1998, 109, 11027. (14) Hattori, K. Master’s thesis, Kyoto University, March 1999. (15) (a) Kondo, T. J. Polym. Sci., Poly. Phys. Ed. 1997, 35, 717. (b) Kondo, T.; Miyamoto, T. Polymer 1998, 39, 1123. (c) Kondo, T. Doctoral Thesis, Kyoto University, 2000. (16) Iwata, T.; Azuma, J.; Okamura, K.; Muramoto, M.; Chun, B. Carbohydr. Res. 1992, 224, 277. (17) Mischnick, P.; Hennig, C. Biomacromolecules 2001, 2, 180. (18) Arisz, P. W. F.; Kauw, H. J. J.; Boon, J. J. Carbohydr. Res. 1995, 271, 1.

Tsunashima et al. (19) Steeneken, P. A. M.; Woortmann, A. J. J. Carbohydr. Res. 1994, 258, 207. (20) Van der Burgt, Y. E. M.; Bergsma, J.; Bleeker, I. P.; Mijland, P. J. H. C.; Van der Kerk-van Hoof, A.; Kamerling, J. P.; Vliegenthart, J. F. G. Carbohydr. Res. 1998, 312, 201. (21) Richardson, S.; Nilsson, G. S.; Bergquist, K.-E.; Gorton, L.; Mischnick, P. Carbohydr. Res. 2000, 328, 365. (22) Tezuka, Y. Carbohydr. Res. 1993, 241, 285. (23) Nemoto, N.; Tsunashima, Y.; Kurata, M. Polym. J. 1981, 13, 827. (24) (a) Tsunashima, Y. J. Phys. Soc. Jpn. 1992, 61, 2763. (b)Tsunashima, Y.; Odani, H. Bull. Inst. Chem. Res., Kyoto UniV. 1991, 69, 322. (25) Tsunashima, Y. J. Chem. Phys. 1995, 102, 4673. (26) Tsunashima, Y. J. Chem. Phys. 1999, 110, 12211. (27) Baranyai, A.; Cummings, P. T. J. Chem. Phys. 1995, 103, 10217. (28) Travis, K. P.; Daivis, P. J.; Evans, D. J. J. Chem. Phys. 1995, 103, 1109. (29) Lopez Cascales, J. J.; Navarro, S.; Garcia de la Torre, J. Macromolecules 1992, 25, 3574. (30) Wang, S. Q. J. Chem. Phys. 1990, 92, 7618. (31) Dynamic Light Scattering; Berne, B. J., Pecora, R., Eds.; John Wiley & Sons, Inc.: New York, 1976; Chapter 7.

BM010069E