Reorganization of Dynamic Self-Assemblies of Cellulose Diacetate in

Biomacromolecules 2002, 3, 1276-1285

1276

Reorganization of Dynamic Self-Assemblies of Cellulose Diacetate in Solution: Dynamical Critical-like Fluctuations in the Lower Critical Solution Temperature System Yoshisuke Tsunashima,* Hiroyuki Kawanishi,† and Fumitaka Horii Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan Received June 10, 2002; Revised Manuscript Received August 20, 2002

Dynamics of cellulose diacetate (CDA, the total degree of substitution (TDS) ) 2.44) in dimethylacetamide (DMAc) in dilute solution was investigated at 2, 10, 20, 30, 40, 49.7, and 61.5 °C through dynamic light scattering in the quiescent state. The following three facts were made clear. First, CDA existed in three types of structures in the polar solvent, DMAc; one is a single CDA chain, and the others are dynamic structures, or self-assemblies, which were formed temporarily and locally by the solvent-mediated hydrogen bonding between the intermolecular C-6 position hydroxyls of the anhydroglucose units in the CDA backbone. Second, CDA showed a nature of low-temperature solubility in DMAc, that is, CDA is expected to dissolve molecularly below -12 °C but to take a phase separation above 65 °C, where two structures such as collapses of a single CDA chain and an aggregate appear. Third, a reorganization in the dynamic structures was detected at the temperature T* ) 33.8 °C. At this temperature, two dynamic structures showed the discontinuity in their correlation lengths, whereas the single CDA realized an uncorrelated chain state in the dynamical sense. In view of the low-temperature solubility of CDA in DMAc, this abnormal behavior around T* was explained by dynamical critical-like fluctuations if T* were treated as a kind of lower critical solution temperature (LCST) in the CDA/DMAc system. Here, the self-assemblies arise as the dynamical fluctuations under the spinodal decomposition situation and the competition between the hydrogen bonding (HB) and the hydrophobic interaction (HPhI) makes the conformation of CDA chains change drastically. In this scheme, the solvent-mediated HB and HPhI play important roles in the structure reorganization of cellulose derivatives in strong electronegative solvents, though HB and HPhI cooperate with the inherent chain helicality. I. Introduction In the previous papers,1-5 we have examined the solution dynamics of cellulose diacetates (CDA) in polar solvent at 30 °C by performing measurements on the viscosity, the sedimentation velocity/equilibrium,1 and the dynamic light scattering (DLS) in quiescence2,3 and in circular Couette flow.4,5 It was found that a CDA fraction (Mw ) 1.70 × 105, Mw/Mn ) 1.23, the total degree of substitution of hydroxyls by O-acetyls (TDS) ) 2.44) showed peculiar behavior in dimethylacetamide (DMAc), summarized as follows. (1) The CDA chain exhibited semiflexiblity, but the chain rigidity was low, and it depended strictly on the solution environment, and (2) CDA formed two large transient clusters of a regulated size (the self-assembly), and the regulation was controlled again by the environmental effects exerted on the CDA chains in solution. Because the present CDA sample was prepared by partial hydrolysis of cellulose triacetate (CTA, TDS ) 3), the CDA chain was composed of various kinds of the anhydroglucose units (AGU) of which three hydroxyl groups at C-2, -3, and -6 positions were substituted partially by O-acetyls. Three hydroxyls in each AGU were substituted by O-acetyls by * To whom correspondence should be addressed. † Present address: Fuji Photo Film Co., Ltd., Minami-Ashigara, Kanagawa 250-0193, Japan.

2.44 on aVerage along the CDA chain contour, that is, the AGU sequence along the chain was irregular in the distribution of O-acetyl substitution. Thus, the above-mentioned peculiarity of CDA in solution was attributed mainly to the hydrogen bonding (HB) that would be formed by specific AGU sequences (i) within a single chain (the intramolecular HB) and (ii) between different chains (the intermolecular HB) with the help of polar solvent DMAc.1-5 The intramolecular HB could be induced mainly by an O-5 position ring oxygen and a C-3′ position hydroxyl incorporated partially with the inherent helical nature and with hydrophobic interaction HPhI of the CDA chain. Here, two AGUs of O-5 and C-3′ would not necessarily be restricted to the adjacent ones but to ones separated by several units. The intramolecular HB made the single CDA chain quasi-flexible or semiflexible, that is, the persistence length Q of CDA was as small as 8.0 nm in the quiescent state but became larger (i.e., stiffer with Q ) 43 nm) in the external force fields such as the ultracentrifuge.1 On the other hand, the intermolecular HB could be formed between C-6 position hydroxyls in the specified AGU sequences of chains, which would result in association. This associated cluster changed its structure with the variation of the solution environment because various ways of the intermolecular HB, which cooperates with HPhI, could be realized in accordance with the change in solvent, temperature, concentration, external

10.1021/bm0200682 CCC: $22.00 © 2002 American Chemical Society Published on Web 09/26/2002

Reorganization of Dynamic Self-Assemblies

forces, and so on. In addition, the structures showed a few steps of shear-rate-dependent transitions under shear field as weak as Couette flow (simple shear flow),4,5 though the structures broke finally into the molecularly dispersed single chain under a shear field such as capillary viscous flow or ultracentrifugal field.1 In this sense, the HB formation by CDA in solution was of a solvent-mediated character. These characteristics of the solvent-mediated HB interaction make us expect that the structures of CDA associates would change or reorganize with temperature, and probably the single CDA chain would too. Ethyl(hydroxyethyl)cellulose gave an example, in which the stable and steady aggregations have been conducted with the variation of temperature.6-11 In the present paper, we intended to examine how the dynamic structures in the CDA/DMAc system can be affected by temperature. To attain this aim, we conducted DLS measurements on dilute CDA/DMAc solutions in the temperature range of 2-61.5 °C. We then detected three structures, that is, two dynamic structures and a single CDA chain, above 2 °C. However, a smaller dynamic structure disappeared and transferred to a larger one at 2 °C, and then the larger one turned into a single chain at -12 °C. This suggests that CDA could disperse molecularly below -12 °C. In addition, above 65 °C, the phase separation could occur, and two structures of a collapsed CDA chain and one kind of aggregate would appear. In addition, a reorganization of dynamical clusters was detected at a temperature T* ) 33.8 °C. These complex features could be explained in terms of the dynamical critical-like fluctuations under unstable state provided that T* is regarded as a kind of lower critical solution temperature (LCST). It is the first instance, to our knowledge, that the existence of T* and the low-temperature solubility of CDA in solution is certified quantitatively. II. Experiment A. Materials and Solution Preparation. The CDA sample used was Fr.1C, which is one of twelve fractions prepared as described already.1,2 The weight-average molecular weight (Mw ) 1.70 × 105) and the polydispersity index (Mz/Mw ) 1.23) were determined by the sedimentation equilibrium method.1,2 The total degree of substitution (TDS) of hydroxyls by O-acetyl groups, determined by 13C NMR measurements after propanoation,12 was 2.44, and the individual degree of substitutions (IDS) in the AGU were 0.85, 0.85, and 0.74 at the C-2, -3, and -6 position hydroxyls, respectively. It should be noticed that the hydroxyls remain dominantly at the C-6 position in the present sample. The CDA sample was mixed with distilled DMAc at 25 °C and was homogenized under continuous stirring at 30 °C. The stock homogeneous solution thus prepared was left for a few days in the thermostat-controlled air box around 30 °C with intermittent stirring. Four different solutions were then prepared by the dilution method. For the DLS experiments, the solution was filtered again into the precision light scattering cell of 12 mm φ through a 0.2 µm pore size filter (Sartorius) and was homogenized at 25 °C. The obtained polymer mass concentrations, c (grams CDA per gram solution), were 1.32 × 10-3 (c1), 2.25 × 10-3 (c2), 3.45 ×

Biomacromolecules, Vol. 3, No. 6, 2002 1277

Figure 1. The refractive index, n, of DMAc (a) vs the wavelength of light λ at four temperatures. The parabolic curve is drawn by the method of least squares at each temperature. In panel b, the refractive index at 488 nm, n488, and the density F0 of DMAc are plotted against temperature T (°C). The data are well represented by the straight line.

10-3 (c3), and 4.81 × 10-3 (c4), all of which were in the dilute solution region. B. Physical Constants of Solvent. Extra-pure reagent DMAc (Tokyo Kasei) was purified by drying over phosphorus pentaoxide and then fractionally distilled in a 1.5m-high column under nitrogen atmosphere at 21 mmHg. The boiling point of the collected fraction was 71 °C. The density, F0, was measured at four different temperatures with the use of a laboratory-made Lipkin-Davison pycnometer of 30ml capacity. It was 0.9501, 0.9407, 0.9315, and 0.9223 g cm-3 at 10.00, 20.00, 30.00, and 40.00 ( 0.03 °C, respectively. The plot of F0 against temperature is shown in Figure 1b by unfilled squares. It gave the relation that F0 (g cm-3) ) -9.260 × 10-4T (°C) + 0.9593. The F0 values at T ) 2 °C and other temperatures were estimated from this relation. The refractive index of DMAc, n, was measured on a Pulfrich refractometer (Shimadzu) at the Na D line (589 nm), Hg e line (546 nm), and Hg g line (436 nm) at the four temperatures described above. The results were plotted against the wavelength of the light, λ, in Figure 1a. A parabola was well fitted to the data for each temperature. The refractive index at 488 nm, n488, was obtained by interpolation at λ ) 488 (the broken line) with the result that n488 ) 1.44851, 1.44481, 1.44047, and 1.43619 at 11.65, 20.40, 30.00, and 40.00 ( 0.05 °C, respectively. Figure 1b shows the temperature dependence of n488 by unfilled circles, which is well represented by the straight-line n488 ) -4.364

1278

Biomacromolecules, Vol. 3, No. 6, 2002

× 10-4T (°C) + 1.45363. This relation was used to evaluate n488 at 2 °C and other temperatures. For the determination of the solvent viscosity, η0, the flow times of the solvent and pure water were measured at 2-60 °C in a capillary viscometer. Then, η0 of DMAc at each temperature was estimated by combining the flow times with the water viscosity and with the densities of water and DMAc. The obtained η0 value was well expressed by the relation that log(η0) ) 0.1229 - 6.16 × 10-3T (°C). It should finally be noted that the freezing point of DMAc is -20 °C. C. Dynamic Light Scattering Measurements. DLS measurements were conducted by the homodyne method on a homemade LS apparatus1 through a multiple-τ digital correlator unit (ALV-5000/E) at seven different temperatures of 2-61.5 °C. The Vv component of the incident light of a single-frequency 488 nm line emitted from an etalonequipped argon-ion laser (3 W, Spectra Physics) was measured for 40-60 min, as has been described elsewhere in detail.4 The normalized intensity time correlation function, g(2)(t), detected was analyzed mainly by the inverse Laplace transformation (ILT)13 program with the complementary use of the CONTIN14,15 and the histogram16,17 programs. Here, the Siegert relation, g(2)(t) ) 1 + a|g(1)(t)|2, was used. If the decay rate distribution, G(Γ), can be reduced to several separated peaks i, as was the present case as shown below, we call afresh them mode i of the averaged decay rate, Γi, and of the averaged relaxation time, τi (τi ) 1/Γi). The histogram program was found to be absolutely effective to analyze g(2)(t) data for weak scattering intensity. In any case, the deduction of G(Γ) from g(2)(t) was made by using these three programs with the consistent/satisfactory results with each other because g(2)(t) was obtained with a high coherency, that is, with a large a value in the Siegert relation. III. Results and Discusion A. Three Modes of Motions for CDA in Solution. An example of the measured g(2)(t) is shown in Figure 2a, where g(2)(t) is plotted against the logarithmic time t for a solution of c ) 3.24 × 10-3 g cm-3 at a scattering angle θ ) 30° and at temperature T ) 20 °C. Here, the g(2)(t) was analyzed by the ILT program, and the result is given in Figure 2b in the form of the decay rate distribution, G(Γ), or the fractional amplitude, plotted against Γ. The figure reveals that three separated peaks appear in G(Γ). Each peak is narrow and well-separated from each other. The uniqueness of three peaks in the shape, the position, and their relative amplitudes was confirmed to within (2-5% by the repeating analyses of the same g(2)(t) data in the way that we changed the range of data points to be analyzed, the range of the first and the final Γ to be searched, and the number of grid points in the range of Γ searched. Three separated peaks similar to Figure 2b were deduced from the other g(2)(t) data at different temperatures of 10-61.5 °C and different concentrations of c1-c4. Only one exception was the case at 2 °C, where only two separated peaks were detected (see below). In any case, we resolved three modes from the distribution G(Γ), which corresponded to three peaks. We assigned these modes to modes I, II, and III in decreasing order of Γ, as is indicated

Tsunashima et al.

Figure 2. The normalized intensity time correlation function, g(2)(t), (a) vs time t for CDA in DMAc of c ) 3.24 × 10-3 g cm-3 at θ ) 30° and T ) 20 °C. Panel b shows the decay rate distribution G(Γ) obtained from the above g(2)(t) data by the inverse Laplace transformation program. Three modes of motions are detected separately.

in Figure 2b and represented them newly by the mean decay rate Γi (i ) I, II, and III). Here, the Γi value was evaluated by averaging Γ only over G(Γ) that was concerned with mode i. At all concentrations and temperatures measured, the mean decay rate Γi for each mode was found to show the squared scattering vector, q, dependence (Γi ∝ q2), where the magnitude of the scattering vector q at the scattering angle θ is defined by q ≡ |q| ) (4πn0/λ) sin(θ/2) with n0 being the refractive index of solvent at the wavelength λ in a vacuum. Three modes in the present system are of diffusive nature, accordingly. Figure 3a-c demonstrates this feature for modes I, II, and III, respectively, for four polymer mass concentrations c1-c4 at 20 °C. Here, Γ(c)/q2 is plotted against q2 and gives a q-independent constant value at each c. The diffusion coefficient at a given c, D(c), is thus determined as a mean value, D(c) ) [Γ(c)/q2]av, by averaging over five q2 region. The D(c) values for other temperatures at 2-61.5 °C were also determined by exactly the same procedure as described here. In the previous works on CDA in DMAc at 30 °C,1-5 we have already detected three modes and discussed in detail their behavior. Namely, the dynamical characteristics for three modes of motions were cross-checked over the viscosity, sedimentation, and DLS data in quiescence and in sheared field. We have then assigned three modes as follows: mode I is the single chain diffusion of a CDA molecule, and modes II and III are the dynamic structures, or clusters. These structures were created locally and temporarily in solution, namely, they were not stable but were in process of the creation-disappearance. In the present study, we treated these three modes at 2-61.5 °C by exactly the same



Figure 4. D(c) vs c plots for mode I of CDA in DMAc at seven different temperatures of 2-61.5 °C.

Figure 3. The decay rates divided by the scattering vector q2, or Γ/q2, vs q2 for four different polymer mass concentrations of CDA in DMAc at 20 °C: (a) mode I; (b) mode II; (c) mode III. The concentrations c (×10-3 g cm-3) are (0) 4.53, (O) 3.24, (4) 2.12, and (3) 1.24.

Figure 5. Temperature dependence of (a) the concentrationdependent coefficients kD and (b) the hydrodynamic radius RH for mode I of CDA in DMAc.

viewpoint as that at 30 °C, that is, they are the single chain diffusion and the dynamic structures, respectively. B. Better Solvent Nature of a Single CDA Chain at Lower Temperature. Figure 4 shows the concentration dependence of DI(c) for mode I at seven temperatures of T ) 2-61.5 °C. The data at each T are well represented by the straight line, which is expressed by the relation DI(c) ) D0,I(1 + kD,Ic). Here D0,I and kD,I are the translational

diffusion coefficient at infinite dilution and the concentrationdependent coefficient, or the dynamical second virial coefficient, respectively. It is found from Figure 4 that, with increasing temperature, D0,I increases but the slope of the line changes from positive to negative around 30-40 °C. This temperature behavior is given in Figure 5a in the form of a kD,I vs T plot. kD,I decreases sharply with increasing T, giving 411 cm3 g-1 at 2 °C and -189 cm3 g-1 at 61.5 °C

1280


Figure 6. D(c) vs c plots for mode II of CDA in DMAc at six different temperatures of 10-61.5 °C. Mode II was not detected at 2 °C.

and taking zero at T* ) 33.8 °C. Because mode I represents the behavior of a single CDA chain, the change of sign in kD,I means that the pairwise hydrodynamic and thermodynamic intermolecular interactions between CDA molecules vary from repulsive to attractive at 33.8 °C, that is, the CDA chain is in good solvent state below 33.8 °C. Corresponding to this trend, the hydrodynamic radius of CDA chains, RH, which was estimated from RH ) kBT/(6πη0D0,I) and is plotted against T in Figure 5b, increases from 1.34 to 72.5 nm with the decrease of temperature from 61.5 to 2 °C. Thus, the hydrodynamic dimension of single CDA chains expands with decreasing temperature; that is, the single CDA molecule is prominently in more good-solvent state at lower temperature, which signifies the low-temperature solubility of CDA in DMAc. On the other hand, the extremely small RH (1.34 nm) at 61.5 °C might indicate a globular collapse of a CDA chain at such a high temperature as 61.5 °C, which will be discussed in the later section. The situation that kD,I ) 0 at T* ) 33.8 °C might be an indication that the CDA chain would be apparently in a noninteractiVe state dynamically due to the cancellation between second- and higher-order intermolecular interactions. C. Dynamic Structures and Their Thermal Reorganization. As described already, the decay rates Γi of modes II and III show the q2 dependence, that is, Γi ) Diq2 (i ) II and III) at different T and c. Modes II and III represent the dynamic structures18 of CDA in DMAc, and the decay rate represents the speed with which a tentative grown-up of concentrations relaxes down to the level of the bulk concentration.1 Thus, the structure neither holds a thermodynamically stable state nor diffuses such as a usual diffusion particle.1 The correlation length to within which the fluctuations associate can be estimated from the diffusion coefficient D(c) as ξ(c) ≡ kBT/(6πη0D(c)). Figure 6 shows a series of DII(c) vs c plots for mode II at six different temperatures of

Tsunashima et al.

Figure 7. D(c) vs c plots for mode III of CDA in DMAc at seven different temperatures of 2-61.5 °C.

10-61.5 °C. Data at 2 °C is absent because mode II was not observed at this temperature. It is very interesting that, judging from extrapolation to c f 0 within our experimental degree of dilution, D0,II at c f 0 does not agree with D0,I for mode I because D0,I is larger by 1 order of magnitude (see Figure 4). At infinite dilution, the CDA solution contains not only the uncorrelated single chains but also the dynamic structures. This would be abnormal for usual solution systems but is characteristic of cellulosic solutions, which could be attributed to the intermolecular HB. Thus, the concentrationdependent coefficient kD,II estimated from DII(c) for dynamic structures is meaningful, and it represents the pairwise interaction between structures. The intercept D0,II at infinite dilution gives the correlation length ξII as defined by ξII ≡ kBT/(6πη0D0,II). From Figure 6, it is found that kD,II is negative over all of the temperatures measured. The diffusion coefficient DIII(c) for mode III is also shown in Figure 7 by the DIII(c) vs c plots for seven temperatures at 2-61.5 °C. Each plot gives a negative slope, and kD,III is negative in the range of 2-61.5 °C, accordingly. kD,II and kD,III at each T are summarized in Table 1. The temperature dependence of kD,II and kD,III are plotted against T in the lower part of Figure 8. To make clearer the discussion that will appear below, kD,I for single chains is replotted in the upper part. The detected peculiar characteristics in the dynamic pairwise interaction can be summarized as follows. First, kD,II for mode II is negative but exhibits a convex upward with a maximum around a specific temperature, T* ) 33.8 °C. The maximum value is nearly zero, and the temperature agrees with T* at kD,I ) 0 for single chains. Thus, the dynamic structures of mode II are apparently in a dynamically noninteractive state to each other at T* ) 33.8 °C. The situation that kD ) 0 could be achieved not by vanishing of two-body interactions but by a complex


Reorganization of Dynamic Self-Assemblies Table 1. Characteristics of the Single Chain and Dynamic Structures Observed for CDA in DMAc at Various Temperatures temp (°C) D0 (10-7 cm2 s-1) RH (10-6 cm) kD (cm3 g-1)

Mode I (Single Chain) 2 10 20 30 0.212 0.457 0.762 1.62 7.25 3.92 2.85 1.64 411 287 70.6 85.0

temp (°C) D0 (10-8 cm2 s-1) ξ (10-6 cm) kD (cm3 g-1)

Mode II (Dynamic Structure) 2 10 20 30 40 a 4.04 3.50 2.53 4.46 a 4.43 6.22 10.4 6.73 a -98.7 -48.4 -1.20 -23.5

temp (°C) D0 (10-8 cm2 s-1) ξ (10-6 cm) kD (cm3 g-1)

Mode III (Dynamic Structure) 2 10 20 30 40 1.10 1.27 1.37 1.37 2.65 14.0 14.1 15.9 19.4 11.3 -161 -144 -124 -105 -163

a

40 2.36 1.27 -71.7

49.7 7.90 0.455 -177

61.5 32.9 0.134 -189

49.7 1.29 2.68 -48.0

61.5 22.6 1.94 -90.6

49.7 3.06 11.7 -150

61.5 4.83 9.10 -113

Mode II was not detected (see the text).

Figure 8. The concentration-dependent coefficients, kD, for the three modes of CDA in DMAc plotted against temperature T.

cancellation between two-, three-, and higher-body interactions. In this sense, various types of interactions and structures could be accessible in this situation. In the present system, the interactions are mainly HB and hydrophobic interaction (HPhI) that are mediated by solvents. It is thus expected that the more the temperature is apart from T* to each side, the more dominant the HB or HPhI interaction becomes and the more distinct the dynamic structures are. Second, kD,III for mode III increases linear-like with T on both sides of T* but the trend is divided into parts at T* ) 33.8 °C, namely, kD,III drops suddenly from -105 to -163 cm3 g-1 at 30-40 °C. This discontinuity indicates also a reorganization of dynamic structure from a lower- to a higher-temperature type at T*. In addition, we can see in Figure 8 that kD,II seems to join kD,III on both edges of the parabola, that is, kD,II and kD,III are equal to each other at 8.6 and at 65 °C. A melding of mode II and III could be achieved at these two temperatures. Because kD,II is always less negative than kD,III and varies more widely with temperature

Figure 9. Temperature dependence of the hydrodynamic radius, RH, for mode I and the correlation lengths, ξ, for modes II and III of CDA in DMAc.

than kD,III, it is conceivable that mode II is more fragile than mode III in the structural stability. In Figure 9, the correlation lengths, ξ, for modes II and III are plotted against temperature, together with the hydrodynamic radius of mode I, RH. In contrast to RH, which increases monotonically with decreasing temperature, ξ of the two dynamic structures exhibits complicated behavior. In other words, ξII and ξIII seem to show a singularity around T* ) 33.8 °C in the way that they rise sharply toward infinity from both sides of T*. The dynamic structure represents the concentration fluctuations, and the correlation length ξ stands for its fluctuation width or the amplitude. Therefore, the extreme increase of ξII and ξIII at T* means that modes II and III would amplify their concentration fluctuations excessively and critically as T approaches to T*. In the case of mode II, we could see, from the combination with Figure 8, that the extreme fluctuations II are driven by the steep decrease of the attractive interstructure interactions with T f T*, followed by an apparent zero interaction at T*, that is, the negative kD,II approaches zero as T f T*. In the case of mode III, the situation is slightly different from the case of mode II. The amplitude of concentration fluctuations III becomes critical and drastic at T*. However, when T* is approached from lower T, the attractive interstructure interactions decrease and the fluctuations would disperse infinitely, but the approach from higher T induces the fluctuations of large but definite amplitude at T* because of an increase of the attractive interstructure interactions (more negative kD,III) with decreasing T. These strange features might come from an unstable nature of the single CDA chain (mode I) at T*, where the intermolecular interactions are made apparently canceled out (i.e., kD,I ) 0) under a delicate balance between the multiorders of interactions acting on the chains. The HB and HPhI compete with each other for the optimum structure and the dominant interaction, either HB or HPhI, controls the chain conformation. The inversion in competing strength of HB and HPhI occurs critically at T*, and it induces the change in the way of chain clustering, which would result in the reorganization of dynamic structures. In this connection, a kind of conforma-

1282


Figure 10. The RH vs T

-1

Tsunashima et al.

plots for mode I of CDA in DMAc.

tional change in the single chain will be demonstrated as described in the later section. Some interesting features are also revealed on ξII and ξIII far below and above T*. In the region below T*, as shown in Figure 9, ξII decreases sharply from 104 to 44 nm with decreasing T from 30 to 10 °C and comes to coincide with RH at 8.6 °C. The coincidence of ξII with RH does not always mean the union of modes I and II but indicates the reorganization of mode II to mode III because kD,II becomes equal to kD,III at 8.6 °C (Figure 8). On the other hand, ξIII decreases from 194 to 141 nm with decreasing T from 30 to 10 °C at T < T* and retains its level below 10 °C. It seems that ξIII unites with RH, or ξIII changes into RH, around -12.0 °C. Thus, the molecular dispersion of a single CDA chain may be realized in solution below -12.0 °C. In the region of T > T*, however, ξII decreases sharply with T, while ξIII holds a constant value (ca. 115 nm) around 40-50 °C and then decreases very steeply with T. ξII and ξIII come to join each other around 65 °C. The complementary evidence is given in Figure 8, where the interactions of dynamic structures for modes II and III are attractive above 40 °C and kD,II becomes equal to kD,III at 65 °C. Thus, modes II and III unite to a single cluster at this temperature. This feature indicates that the phase separation occurs at a high temperature such as 65 °C, that is, CDA separates into two structures, a collapsed single chain and an aggregate that merges modes II and III. This indication could be conducted below. D. Reorganization in the Structure of Single CDA Chains. The temperature dependence of the single-chain size is reexamined in Figure 10 in the form of a log(RH) vs T -1 plot, or so-called the Arrhenius type of plot. The plot shows that RH follows on a straight line (1) below 40 °C (i.e., T -1 > 3.2 × 10-3) and then sifts to another line (2) above 40 °C. The line 1 gives a relation, log(RH (cm)) ) -11.35 + 1.695 × 103(T (K))-1, which predicts that mode I coincides with mode III at T** ) -12.0 °C, as mentioned above. The both sizes could there take an equal value, RH** ) 140 nm, as is demonstrated in Figure 9. The necessity of the line 2 in Figure 10 would mean that the single CDA chain changes the way of the diffusion motion around T* ) 33.8 °C, as is the case of the structural changes for modes II and III at T* (see Figure 9). The change could be considered as follows: Below T* (on line 1), a single CDA chain experiences the Brownian diffusion motion and realizes its complete molecular dispersion at a temperature below -12.0 °C, which

Figure 11. The fractional amplitudes, fI, for mode I (single chains, 0) and fII+III for mode II + III (dynamic structures, O) of CDA in DMAc at the scattering angle θ ) 30° are plotted against temperature T (K) for four polymer mass concentrations, c1-c4. The mass concentrations (10-3 g CDA/g solution) are c1 ) 1.32, c2 ) 2.25, c3 ) 3.45, and c4 ) 4.81.

temperature is noticed as T**. This supports the trend of low-temperature solubility of CDA in polar solvents. On the other hand, above T* (on line 2), the single CDA chain begins to suffer a retardation in the diffusion motion, which might be executed by strong intermolecular attractions in a poor-solvent situation that is indicated by the negative large kD,I above T* (Figure 8). The conformational change of a CDA chain could take place because of the dominant HPhI to HB above T*. A change in the way of intramolecular HB could also occur because of the inversion in the interaction from HB to HPhI, which will be discussed below. Thus, the mobility above T* decreases by about 1.5 times that below T*, which was estimated from the slopes of lines 2 and 1, or the quasi-activation energy. E. Variation of Fractional Amplitudes for Three Modes with Temperature. In Figure 11, as an example, the fractional amplitudes, fi, of mode I (single chains) and modes II + III (sum of two dynamic structures) at the scattering angle θ ) 30° are plotted against the absolute temperature T (K) for four polymer mass concentrations c1-c4 (c1 > c4). For all of the concentrations, fI for mode I takes nearly a constant 0.25 below 313 K but decreases at T > 313 K and goes to zero at 334.5 K (61.5 °C). The fractional sum of modes II and III, fII+III, is vice versa, namely, it takes a constant 0.75 below 313 K and then increases to unity with increasing T. The tendency of fI f 0 around 65 °C indicates that a coiled single chain collapses into an extremely small


grobule, the size RH being 1.34 nm at 61.5 °C, as is discussed above and is shown in Figure 5b. This small RH contributes little to the fractional amplitude of the scattering intensity for mode I. On the other hand, the trend of fII+III f 1 around 65 °C reflects dominant appearance of the cluster modes II and III, both of which would unite together and would appear as an aggregated precipitate above 65 °C. In general, the amplitude f is proportional to the weight fraction, the weight mass, and the static structure factor, S(q), of the structure concerned.19 At present, however, it is difficult to estimate the weight fraction of each structure because of lack in the static LS data. A break on the fI and fII+III lines around T* makes us confirm that T* is a unique temperature for the present CDA/ DMAc system. The competition of HB and HPhI under thermodynamically unstable state at T* could change the way of interactions in both intra- and intermolecular levels, which induces the rearrangement of chain conformations and of clustering. Thus, the size transition in both single chain and dynamic structures could be realized at T*. The structural change of cellulose acetate (CA) in solution has been reported for CA (TDS ) 2.5 and 3.0) by vacuum ultraviolet circular dichroism (CD),20 optical rotatory dispersion (ORD),21,22 and 2D nuclear Overhauser effect spectrometry (NOESY) 13C NMR relaxation times.23 However, they suggested indirectly the micro/macroscopic structural transition below 33 or at 53-55 °C. F. Dynamical Critical-like Fluctuations and QuasiLCST Behavior. The state of kD,I ) kD,II ) 0, which is realized at T*, would make us imagine that the chains become unstable and that a variety of cluster formation could be amplified at T* because here the intermolecular interactions are canceled out under a delicate balance between multiorders of interactions acting on the chains. Taking the low-temperature solubility of CDA into account, the peculiar features mentioned above for CDA in DMAc could compare with the dynamical critical fluctuations in the LCST system provided that T* were regarded as the critical temperature. In Figure 12, all data of ξII, ξIII, and RH are doublelogarithmically plotted against |T - T*|, as mimic to the usual critical fluctuation.24,25 ξII below and above T* can be summed up to a straight master line with a negative slope. ξIII shows nearly the same behavior as ξII. However, the RH data are separated into two straight lines with a negative (for T > T*) and a positive (for T < T*) slopes. The slope, or the critical exponent ν in the expression of ξ ∝ |T - T*|-ν, is estimated to be 0.15, 0.68, 1.5, and -0.63 for ξIII, ξII, and RH at T > T* and RH at T < T*, respectively. The value ν ) 0.68 is close to the theoretical universal value, 0.625.21,22 Thus, the dynamic behavior observed in the present study could be explained as the aspect expectable inside the binodal line on the phase diagram. Here, the system is in an unstable state and the temperature-concentration diagram has a shape of convex-downward (see Figure 14 below). Because the low critical solution temperature (LCST) should sit in the bottom of the convex-downward curve, T* is not the real LCST but a kind of transition temperature where the inversion occurs drastically in the way of concentration fluctuations. In this sense, the aspect observed in the present CDA/DMAc system


Figure 12. The critical fluctuation-like behavior of CDA in DMAc in the temperature range of 2-61.5 °C. ξII, ξIII, and RH for modes II, III, and I, respectively, are double-logarithmically plotted against |T T*| with T* assuming the critical temperature 33.8 °C in the LCST system. The symbols are as follows: (2)ξII; (9)ξIII; (O) RH(T>T*); ()) RH(T T* and RH at T < T*, respectively, as is given for each line.

Figure 13. The formation of the intermolecular hydrogen bond (HB) induced between the locally C-6 OH rich sequences along the chain backbone. Polar solvents DMAc situate between the intermolecular OHs and mediate the HB interaction. The mediation of HB by solvent is affected by the strength between HB and the hydrophobic interaction (HPhI), which strength depends strongly on the solution environments: (+) C-6 position OH; (b) polar solvents.

is regarded as dynamical critical-like fluctuations in the LCST system with T* being a quasi-critical temperature. The stable homogeneous solution could be achieved downward on the binodal line, or in lower temperature region, and the molecular dispersion of single CDA chains could be realized below T** (-12.0 °C). G. Dynamical Phase Diagram and the Mechanism of Structure Reorganization. Figure 13 shows the scheme of the intermolecular HB formation in the present solution. The sequential irregularity of C-6 position OH along the CDA chain contour would give several “locally C-6 OH rich” parts. The HB formation is performed selectively between these parts with the help of polar solvents. The solvents situate between the intermolecular OHs and mediate the strength of the HB interaction. This mediation depends strongly on the solution environments, such as temperature and concentration, and affects the strength of HB against HPhI. As schematically shown in Figure 14, the phase diagram for the present system can be represented by the low-temperature solubility. The spinodal and the binodal lines are a shape of convex downward. At temperatures below the binodal line,

1284


Tsunashima et al.

Figure 14. Dynamical phase diagram for the CDA/DMAc solution system. The spinodal and binodal lines are convex downward, and the solution is stable/homogeneous below these lines. At B above T* ) 34 °C, the formation of dynamic structures are regulated mainly by HPhI, while at A below T*, HB becomes dominant and it regulates the way of structure formations. See detail in Figure 15 and the text.

the solution is homogeneous and stable, and the molecular dispersion could be attained at T e -12 °C, while the phase separation would occur at 65 °C. The solution situated inside the spinodal line is unstable, and the variety types of dynamic structures are realized because of large concentration fluctuations. At B above T* ) 34 °C, the formation of dynamic structures are regulated mainly by HPhI, while at A below T*, HB becomes dominant and it regulates the way of structure formations. The structure and the chain reorganizations take place in accordance with the advance of HB against HPhI or vice versa, which changes with temperature. As illustrated in Figure 15, the inversion in the strength of HB and HPhI occurs drastically at T*. Around A, because of the dominant OH groups that distribute near-randomly on the chain surface, HB is formed between these OH groups and the dynamic structures develop over a long range with stable contact with polar solvents. On the other hand, the HPhI becomes dominant around B. A large numbers of hydrophobic groups situate on the chain surface, but the OH groups are not so many. These OH groups, however, protect the chains from precipitation. The “locally C-6 OH rich” sequences on the chain surface help the chain to form HB and induce dynamical clustering structures. The structures, however, are neither as large nor as stable as compared with those at A because of the hydrophobic atmosphere in solution. In both situations, the polar solvent cooperates with OH groups and mediates the strength of HB and HPhI as temperature changes from A to B. Conclusion Dynamics of CDA/DMAc dilute solutions was investigated at 2, 10, 20, 30, 40, 49.7, and 61.5 °C through DLS in the

Figure 15. Structure and chain reorganizations in accordance with the advance of HB against HPhI or vice versa with temperature. The inversion takes place drastically at T*. Around A, the HB is formed between the near-randomly distributed OH groups. The dynamic structures develop largely and are not so unstable. Around B, the HPhI becomes dominant. Many hydrophobic groups come out on the chain surface but little for the OH groups. These OH groups, however, make the chains soluble in solution, and the locally OH rich parts on the chain surface induce the intermolecular HB and construct dynamic structures. The polar solvent cooperates with OH groups and mediates the strength of HB and HPhI with the change of temperature: (+) OH groups; (-) hydrophobic groups; (b) polar solvents.

quiescent state. Three modes of motions were basically observed in the temperature range of 10-61.5 °C, while only two modes were detected at 2 °C. All three modes were the diffusion motions, the fast mode being assigned to the single chain diffusion and the other two to the dynamic structures of CDA clusters. The single CDA chain had a nature of lowtemperature solubility in polar solvent, and the hydrodynamic radius increases remarkably with decreasing temperature. The dynamic structures are formed in solution temporarily and locally by solvent-mediated HB between the intermolecular C-6 position hydroxyls of specified AGU sequences. At T < 33.8 °C, the correlation lengths of two dynamic structures decreased with decreasing temperature, and the homogeneous solution of single CDA molecules was expected to appear solely at T e -12 °C, whereas above 33.8 °C, all of three structures decreased in their sizes or the correlation lengths, and the phase separation was indicated around 65 °C, where the collapsed CDA single chain and an aggregate would appear. In addition, a curious structural reorganization was observed at a specified temperature T* ) 33.8 °C. At T*, the single CDA took a noninteractive state in the dynamical sense, whereas two dynamic structures became infinity in ξ and gave discontinuity/maximum in kD. These features can be understood as the dynamical critical-like fluctuations in


the LCST system provided that T* is regarded as the quasicritical temperature. These fluctuations would be expected solely for cellulose derivatives in strong electronegative solvents if the derivatives contain hydroxyl groups in the AGU units. The HB and HPhI, which conduct under the LCST type of phase diagram, would make the solubility of CDA in polar solvents extremely complex and irregular. In other words, both the chain architecture and the inherent helical nature of CDA could cooperate with the solution environment in polar solvents. The cooperation could enhance various types of the intra- and intermolecular HB and HPhI, and the competitions between two interactions would originate a variety of specified dynamical clusters in solution. The dynamical structure reorganization and the phase separation in this system will be verified in the forthcoming paper,26 where the single chain dispersion will be examined on our laboratory-made low-temperature DLS apparatus that works in the range of T from 0 to -90 °C. The dynamical behavior at the specified temperatures of T* ) 33.8 °C, T** ) -12 °C, and 65 °C will also be discussed there. References and Notes (1) Kawanishi, H.; Tsunashima, Y.; Okada, S.; Horii, F. J. Chem. Phys. 1998, 108, 6014. (2) Kawanishi, H.; Tsunashima, Y.; Horii, F. J. Chem. Phys. 1998, 109, 11027. (3) Tsunashima, Y.; Kawanishi, H.; Nomura, R.; Horii, F. Macromolecules 1999, 32, 5330. (4) Tsunashima, Y.; Kawanishi, H. J. Chem. Phys. 1999, 111, 3294. (5) Kawanishi, H.; Tsunashima, Y.; Horii, F. Macromolecules 2000, 33, 2092. (6) Goldszal, A.; Costeux, S.; Djabourov, M. Colloids Surf., A 1996, 112, 141.

Biomacromolecules, Vol. 3, No. 6, 2002 1285 (7) Porsch, B.; Nilsson, S.; Sundelo¨f, L.-O. Macromolecules 1997, 30, 4626. (8) Kjøniksen, A.-L.; Nystro¨m, B.; Lindman, B. Langmuir 1998, 14, 5039. (9) Kjøniksen, A.-L.; Nystro¨m, B.; Lindman, B. Colloids Surf. 1999, A149, 347. (10) Nyde´n, M.; So¨derman, O. Macromolecules 1998, 31, 4990. (11) Ostrovskii, D.; Kjøniksen, A.-L.; Nystro¨m, B.; Torell, L. M. Macromolecules 1999, 32, 1534. (12) (a) Hattori, K. Master’s Thesis, Kyoto University, Kyoto, Japan, 1999. (b) Tsunashima, Y.; Hattori, K. J. Colloid Interface Sci. 2000, 228, 279. (c) Tsunashima, Y.; Hattori, K.; Horii, F. Biomacromolecules 2001, 2, 991. (13) Chu, B. Laser Light Scattering, 2nd ed.; Academic: New York, 1991. (14) Proventure, S. W. Comput. Phys. Commun. 1982, 27, 213, 229. (15) Dynamic Light Scattering, The Method and Some Applications; Brown, W., Ed.; Clarendon: Oxford, U.K., 1993. (16) Tsunashima, Y.; Nemoto, N.; Kurata, M. Macromolecules 1983, 16, 584. (17) Gulari, E.; Gulari, E.; Tsunashima, Y.; Chu, B. J. Chem. Phys. 1979, 70, 3965. (18) Doi, M.; Edwards, S. F. The Theory of Polymer Dynamics; Clarendon: Oxford, U.K., 1986. (19) Tsunashima, Y.; Suzuki, S. J. Phys. Chem. 1999, B103, 8675. (20) Stipanovic, A. J.; Stevens, E. S. J. Appl. Polym. Sci., Appl. Polym. Symp. 1983, 37, 277. (21) Mukherjee, S.; Marchessault, R. H.; Sarko, A. Biopolymers 1972, 11, 291. (22) Ritcey, A. M.; Gray, D. K. Biopolymers 1988, 27, 479. (23) Buchanan, C. M.; Hyatt, J. A.; Lowman, D. W. J. Am. Chem. Soc. 1989, 111, 7312. (24) Onuki, A. Polymer Physics and Dynamics of Phase Separation (in Japanese); Iwanami: Tokyo, 1992. (25) Gebhardt, W.; Krey, U. Phase Transition and Critical Phenomena (Translated in Japanese); Yoshioka: Kyoto, Japan, 1992. (26) Tsunashima, Y., to be submitted for publication.

BM0200682

Reorganization of Dynamic Self-Assemblies of Cellulose Diacetate in

Recommend Documents