NMR characterization of cellulose acetate: Mole fraction of monomers in cellulose acetate determined from carbonyl carbon resonances

Carbohydrate Polymers 170 (2017) 23–32

Contents lists available at ScienceDirect

Carbohydrate Polymers journal homepage: www.elsevier.com/locate/carbpol

NMR characterization of cellulose acetate: Mole fraction of monomers in cellulose acetate determined from carbonyl carbon resonances Hiroyuki Kono ∗ , Chinatsu Oka, Ryota Kishimoto, Sayaka Fujita Division of Applied Chemistry and Biochemistry, National Institute of Technology, Tomakomai College, Nishikioka 443, Tomakomai, Hokkaido 059 1275, Japan

a r t i c l e

i n f o

Article history: Received 3 February 2017 Received in revised form 3 April 2017 Accepted 21 April 2017 Keywords: Cellulose acetate Substituent distribution Monomer composition Nuclear magnetic resonance

a b s t r a c t Cellulose acetate (CA) samples with varying degrees of substitution were prepared via homogeneous acetylation in 1-butyl-3-methylimidazolium chloride ([BMIM]Cl) and by the acid-hydrolysis of cellulose triacetate in acetic acid. Quantitative analysis of the 13 C NMR spectra facilitated the assignment of the carbonyl carbon shifts of the 2-mono-, 3-mono-, 6-mono-, 2,3-di-, 2,6-di-, 3,6-di-, and 2,3,6-tri-substituted anhydroglucose units (AGUs), and the determination of the mole fraction of 7 AGUs and unsubstituted AGU in the CA chains. This shed some light on the mechanism of CA production in homogeneous reaction systems. In addition, comparison of the mole fractions of the 8 AGUs suggested that the acetone solubility of CA strongly related to the AGU composition. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction Cellulose acetate (CA) is a cellulose ester that is widely used in various materials (Bifari, Bahadar Khan, Alamry, Asiri, & Akhatar, 2016; Fisher et al., 2008). CA is generally produced by the esterification of the hydroxyl groups at the 2-, 3-, and 6-positions of the anhydroglucose units (AGUs) of cellulose with acetic anhydrides. The degree of substitution (DS), i.e., the average number of acetyl groups per AGU, is related to the physicochemical properties of CA. Commercial CA is produced through two-step reactions involving the acetylation of cellulose and partial hydrolysis of CA (Wu et al., 2004). First, the hydroxyl groups of cellulose are acetylated by acetic anhydride in the presence of acetic acid and concentrated sulfuric acid to form cellulose triacetate (CTA, DS = 3). Next, the acetyl groups of CTA are partially hydrolyzed upon addition of water to give products with the desired DS. The DS of commercial CAs is generally 1.9–2.5, because such samples are soluble in acetone (Fisher et al., 2008). On the other hand, CAs produced by one-step acetylation processes are not soluble in acetone, even if the DS is 1.9–2.5 (Heinze, Schwikal, & Barthel, 2005). Therefore, the

Abbreviations: AGUs, anhydroglucose units; [BMIM]Cl, 1-butyl-3methylimidazolium chloride; CA, cellulose acetate; CMC, carboxymethylcellulose; CTA, cellulose triacetate; DS, degrees of substitution; EDA, electron donor–electron accepter; FWHM, full width at half maximum; HSQC, heteronuclear single quantum coherence; INEPT, Insensitive nuclei enhanced by polarization transfer; ␹, mole fraction; MC, methylcellulose; NOE, nuclear Overhauser effect; ROE, rotaing frame Overhauser effect. ∗ Corresponding author. E-mail address: [email protected] (H. Kono). http://dx.doi.org/10.1016/j.carbpol.2017.04.061 0144-8617/© 2017 Elsevier Ltd. All rights reserved.

substituent distribution at the 2-, 3-, and 6-positions of CA prepared by two-step reactions differs from that prepared by the one-step procedure, and these differences affect the solubility of CA. Important properties of CA which could influence their structure-property relationships include DS, substituent distribution, and monomer composition (Fisher et al., 2008). Quantitative 13 C NMR is commonly used to estimate the DS and substitution distribution of CA (Kamide & Okajima, 1981; Buchanan, Edgar, Hyatt, & Wilson, 1991; Tezuka, 1993; Tezuka & Tsuchiya, 1995; Kono, Hashimoto, & Shimizu, 2015). DS can be determined from the integral values of the methyl and/or carbonyl carbon resonances of the substituent acetyl groups (Kamide & Okajima, 1981). To determine the substituent distribution of CA, perpropionated CA derivatives can be analyzed following the complete propionation of unsubstituted CA hydroxyl groups (Tezuka & Tsuchiya, 1995). The 13 C NMR spectrum of perpropionated CA exhibits two sets of three carbonyl carbon resonances, which correspond to acetyl groups at the 2-, 3-, and 6-positions and propionyl groups at the 2-, 3-, and 6-positions. Quantitative evaluation of the acetyl and propionyl triplet resonances permits the determination of the DS at the 2-, 3-, and 6-positions (DS2 , DS3 , and DS6 , respectively) of CA. The composition of eight monomers comprising CA, i.e. un-, 2-mono-, 3-mono-, 6-mono-, 2,3-di-, 2,6-di-, 3,6-di-, and 2,3,6tri-substituted AGUs, has been determined by acid or enzymatic hydrolysis of the polymeric CA chains followed by chromatographic separation and quantification (Glasser, McCartney, & Samaranayake, 1994; Saake, Horner, & Puls, 1998: Chap. 15). This approach can be used to estimate the mole fraction (␹) of the monomers, but the chemical heterogeneity of CA in the intact poly-

24

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

meric chains is generally lost. In addition, the acid-hydrolysis of glucoside bonds is generally accompanied by the hydrolysis of the acetyl groups, and the complete depolymerization of CA to the monomer units is generally difficult. Thus, methods for the quantitative estimation of AGUs in CA are required to elucidate structure-property relationships in CAs and for quality control in industrial CA production. Notably, we previously reported the two-dimensional (2D) NMR of a series of sodium carboxymethyl cellulose (CMC) samples with DS = 0.68–2.86 (Kono, Anai, Hashimoto, & Shimizu, 2015; Kono, Oshima, Hashimoto, Shimizu, & Tajima, 2016a; Kono, Oshima, Hashimoto, Shimizu, & Tajima, 2016b) as well as those of methylcellulose (MC) samples with DS = 0.66–2.44 (Kono, Fujita, & Tajima, 2017). The 1 H and 13 C chemical shifts of eight AGUs could be assigned using correlation spectra. The chemical shift of each AGU was unaffected by the neighboring substituents or the DS. In addition, quantitative 13 C NMR of the samples provided the ␹ of the eight AGUs comprising these cellulose ethers. The ␹ of the eight AGUs revealed the reaction mechanism of CMC and MC production (Kono et al., 2016a Kono et al., 2016b; Kono et al., 2017). For CA, a similar 2D NMR approach was applied to samples with DS values ranging from 0.92 to 2.92 prepared from CTA (DS = 2.92) by acid-hydrolysis; the 1 H and 13 C chemical shifts of the eight AGUs were assigned (Kono, Hashimoto et al., 2015), and were unaffected by the neighboring substituents and the DS. However, the ␹ of the eight AGUs of CA could not be determined by 13 C NMR because the changes in chemical shift of C2, C3, and C6 of CA upon substitution with acetyl groups were smaller than CMC and MC (Kono, Hashimoto et al., 2015). For example, substitution of the methyl groups at the 2-, 3-, and 6-positions caused a downfield shift in the signals of the directly attached C2, C3, and C6 of cellulose (␤-effect, i.e. through two bonds from the substituent) to 9.6–10.1 ppm. The large ␤-effect caused a separation in the 13 C chemical shifts of the ring carbons of CMC and MC, which permitted the estimation of ␹ of the eight AGUs in these cellulose ethers. On the other hand, ␤-effects caused by the acetyl substituent groups at the 2-, 3-, and 6-positions of cellulose were only 0, 0.2, and 2.8 ppm for the directly attached carbons, and ␥-effects, i.e., through three bonds from the substituent, ranged from −4.3 to +0.2 ppm. Moreover, the chemical shifts of the ring carbons of eight AGUs of CA overlap, which complicates their quantitative estimation (Kono et al., 2016a; Kono et al., 2017). In the present study, in order to quantify the eight AGUs in CA chains, CA samples (DS = 1.26–2.70) were prepared from cellulose in 1-butyl-3-methylimidazolium chloride ([BMIM]Cl) via esterification with acetic anhydride (Schlufter, Schmauder, Dorn, & Heinze, 2006). Quantitative 13 C NMR spectra of the CA samples were measured, and lineshape analysis of the carbonyl carbon regions was performed to assign the chemical shifts of the carbonyl carbons and to determine the change in ␹ for each AGU against the DS. In addition, a similar analysis was applied for CA samples (DS = 2.31–1.28) prepared from CTA in acetic acid by hydrolysis with sulfuric acid (Kono, Hashimoto et al., 2015). Based on these results, the mechanisms for the acetylation of cellulose and the hydrolysis of acetyl groups in CA in homogeneous reaction systems are discussed.

2. Experimental 2.1. Materials Purified pulp (97% ␣-cellulose) was provided by Daicel Co., Japan. [BMIM]Cl and CTA (DS = 2.92) were purchased from SigmaAldrich Inc. (USA). Other chemicals employed in this work were of

a chemically pure grade and were used as received without further purification.

2.2. Preparation of CA samples CA samples (CA 1–3) were prepared from purified pulp dissolved in [BMIM]Cl, according to a previous report (Schlufter et al., 2006). The cellulose pulp (0.45 g, 2.8 mmol for AGU) was dissolved in 7.5 g of [BMIM]Cl with stirring at 353 K for 15 min. After complete dissolution of cellulose, acetic anhydride (0.85 g, 8.4 mmol) was added to the solution. After heating at 353 K for 1 h, the solution was poured into 100 mL of water to precipitate CA. The precipitate was filtered on a sintered crucible, suspended in water (∼50 mL), dialyzed against a stream of distilled water for 3 d, and lyophilized to obtain CA 1. Similarly, CA 2 and CA 3 were prepared by setting the feed amount of acetic anhydride to 1.4 g (13.9 mmol) and 2.0 g (19.5 mmol), respectively. HCA 1–3 were prepared from commercial CTA by acidhydrolysis with sulfuric acid, according to a previously reported procedure (Kono, Hashimoto et al., 2015). Briefly, 1.0 g of CTA was dissolved in 15 mL of acetic acid. Sulfuric acid (0.40 g) was added, followed by water (1.5 mL). The mixture was allowed to react at 358 K for 10, 40, and 70 min to obtain HCA 1–3, respectively. At the end of the reaction, 20% aqueous solution of magnesium acetate was added to neutralize the sulfuric acid. The precipitate was washed in ethanol. The precipitates were suspended in water, dialyzed against water for 3 d, and lyophilized. All samples were stored in a desiccator under vacuum until use.

2.3. NMR experiments Each CA sample (∼35 mg) was dissolved in 700 ␮L of dimethyl sulfoxide-d6 (DMSO-d6 ) containing 0.03% tetramethylsilane (TMS, isotropic purity 99.9%, Sigma-Aldrich Inc.) and transferred to an NMR tube (diameter; 5 mm, Wilmad-LabGlass Co., USA). NMR data were acquired on a two-channel 500 MHz Bruker AVIII spectrometer equipped with a Bruker z-gradient dual-resonance BBFO probe (Bruker BioSpin, GmbH, Germany) at 363 K. 1 H NMR spectra were obtained by setting the 1 H flip angle and number of scans to 30◦ and 32, respectively. Quantitative 13 C NMR spectra were obtained using the 1 H inverse-gated decoupling method (Kono, Hashimoto et al., 2015) with a 13 C flip angle, repetition time, and scan number of 30◦ , 30 s, and 6144, respectively. 1 H–13 C heteronuclear single quantum coherence (HSQC) data were acquired on a 1024 × 256 point matrix for the full spectrum, with 64 scans per increment. The interpulse delay and repetition time in the HSQC experiment were set to 3.44 ms (corresponds to 1/4 JCH ) and 2 s, respectively, and a sine-squared window function was applied along both dimensions prior to the Fourier transform. All 1 H and 13 C chemical shifts were referenced to the methyl resonance of TMS at 0 ppm. Lineshape analysis of the quantitative 13 C NMR spectra was performed using a simulation software package (“Solaguide” in TopSpin ver. 3.1, Bruker BioSpin) and a nonlinear least-squares method to fit a Lorentzian function to the spectra (Noll and Pires, 1980). The baselines of the 13 C spectra were precisely subtracted prior to each fit. The fit was performed using an initial Lorentzian line with a full width at half maximum (FWHM) of 5–15 Hz, and the chemical shift and amplitude of the line were manually adjusted to fit the experimental spectra to achieve a coefficient of determination (r2 ) above 0.85. In addition, the FWHM, chemical shift, and amplitude of the Lorentzian lines were fit using the simplex algorithm to minimize the least-squares difference between the experimental and simulated spectra (Kono et al., 2017). In the lineshape analysis, the quality of the fits was set to r2 > 0.96.

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

3. Results and discussion 3.1. DS and individual DS of CA samples Fig. 1a shows the quantitative 13 C NMR spectra of CA 1–3; the assignments and integral values for the resonances were obtained when the sum of the integral values for the C1 resonances at 103–98 ppm was set to 1. The DS of the CA samples was determined by quantifying the carbonyl carbon resonances at 170–168 ppm with respect to the C1 resonances; it agreed with the integral values of the methyl carbon resonances at 21–18 ppm (abbreviated as DSMe in Table 1). DS2 , DS3 , and DS6 could be determined by quantifying the 13 C resonance splitting in the C1, C4, and C6 regions based on the following interpretation (Kono, Hashimoto et al., 2015): the chemical shifts of C1 when the neighboring C2 was substituted with an acetyl group shifted from 101 ppm to 99 ppm, those of C4 when the neighboring C3 was substituted shifted from 79 ppm to 76–74 ppm, and the substituted C6 was deshielded and shifted from 60 ppm to 61–63 ppm. Based on the integral values for the C1, C4, and C6 regions, DS2 , DS3 , and DS6 of CA 1–3 are summarized in Table 1. The sum of DS2 , DS3 , and DS6 for each sample (DSring ) was in agreement with the DS value. Acetyl substituent distribution of CA 1–3 decreased as C6  C3 > C2, which was in complete agreement with a previous report on the substituent distribution of CA samples prepared by the homogeneous acetylation of cellulose dissolved in [BMIM]Cl (Schlufter et al., 2006). 3.2. Assignment of carbonyl carbon resonances Fig. 2a shows the carbonyl region (170–168 ppm) of the quantitative 13 C NMR spectra of CA 1–3. The carbonyl carbons of the acetyl groups at the 2-, 3-, and 6-positions of CAs appeared at 168.6–168.1 ppm, 169.2–168.6 ppm, and 169.7–169.2 ppm, respectively (Kamide & Okajima, 1981). Previously (Kono, 2013; Kono, Hashimoto et al., 2015), 2D NMR spectroscopic analysis of CA samples with DS = 2.92–0.92 revealed that the 1 H and 13 C chemical shifts of the eight AGU rings were not affected by the neighboring substituents or the DS, suggesting that each carbonyl carbon in the AGUs has a constant 13 C chemical shift. Therefore, lineshape analysis was performed for the carbonyl carbon region of CA samples assuming that there were total 12 resonance lines in the carbonyl carbon region: 4 carbonyl carbons at the 2-position due to 2-mono-, 2,3-di-, 2,6-di-, and 2,3,6-tri-substituted AGUs in the 168.6–168.1 ppm, those at the 3-position due to 3-mono-, 2,3-di-, 3,6-di-, and 2,3,6-tri-AGUs in 169.2–168.6 ppm, and those at the 6-positions due to 6-mono-, 2,6-di-, 3,6-di-, and 2,3,6-tri-AGUs. A nonlinear least-square method was used in the lineshape analysis, assuming that the 12 lines were Lorentzian lineshapes for all 13 C resonances (Kono et al., 2017). The result of the lineshape analysis is shown in Fig. 2a. The spectral lines for the carbonyl carbons at the 2-, 3-, and 6-positions were well reproduced by 4 Lorentzian lines, and the errors for the calculated (dotted lines) and experimental spectra (solid lines) were within ±4% (r2 > 0.96). Chemical shifts, relative height, full-width at half-maximum (FWHM), and integral value of lines 1–12 (abbreviated as I1 –I12 ) for CA 1–3 when the sum of I1 –I12 was set to 1 are summarized in Table 2. The chemical shifts of lines 2 (169.55 ppm), 7 (168.79 ppm), and 10 (168.36 ppm) were in good agreement with those of the carbonyl carbons at the 6-, 3-, and 2positions of CTA (Kono, 2013), indicating that these lines could be assigned to the carbonyl carbons in the 2,3,6-tri-substituted AGU. This assignment was confirmed by the finding that the I2 , I7 , and I10 were nearly the same for all CA samples. In addition, I4 and I6 were almost equivalent, indicating that these lines could be assigned to carbonyl carbons in the 3- and 6-positions of 3,6-di-substituted AGU, respectively. Similarly, I1 was in good agreement with I11 , and I8 and I9 were similar for all samples. These findings indicated

25

that lines 1 and 11 could be assigned to the carbonyl carbons at the 6- and 2-positions of the 2,6-di-AGU, and that lines 8 and 9 corresponded to those at the 3- and 2-positions of the 2,3-di-AGU, respectively. Finally, lines 3, 5, and 12 could be assigned to the carbonyl carbons of the 6-mono-, 3-mono-, and 2-mono-substituted AGUs, respectively (Fig. 2a and Table 2). To confirm the assignment, the 1 H–13 C HSQC spectrum of the CA 2 was measured (Fig. 3). The HSQC spectra provide correlations between directly coupled 1 H–13 C spins (Fig. 3). As depicted by arrows, acetyl substitutions at the 2-, 3-, and 6-positions caused a change in the 1 H and 13 C chemical shifts of the correlation signals owing to acetyl substituent effects (Kono, Hashimoto et al., 2015). Correlations due to the 2,3,6-tri-, 2,3-di-, 2,6-di-, 3,6-di-, and 6-diAGUs could be observed in addition to very weak correlations due to the 3-mono-AGU, while correlations due to the unsubstituted and 2-mono-substituted AGUs were not observed. Due to the carbonyl carbons at the 3-position of 3-mono-AGU and the 2-position of 2-mono-AGU, I5 and I12 for CA 2 equaled 0.01 and 0, respectively, while the other I values ranged from 0.41–0.11 (Table 2). The correlation signals in the HSQC spectrum reflect the ␹ of the AGUs in CA 2, which supported the assignments of the carbonyl carbons. 3.3. Change in the monomer composition of CA against DS Next, the ␹ values of the eight AGUs were determined from the lineshape analysis of CA 1–3 (Table 2). Because the sum of I1 –I12 was set to 1, ␹ was determined from I1 –I12 and DS (Eqs. (1)–(8)): ␹2 = I12 × DS

(1)

␹3 = I5 × DS

(2)

␹6 = I3 × DS

(3)

␹23 = (I8 + I9 ) × DS/2

(4)

␹26 = (I1 + I11 ) × DS/2

(5)

␹36 = (I4 + I6 ) × DS/2

(6)

␹236 = (I2 + I7 + I10 ) × DS/3

(7)

and ␹n = 1–(2 +3 +6 +23 +26 +36 +236 ),

(8)

where ␹n , ␹2 , ␹3 , ␹23 , ␹26 , ␹36 , and ␹236 are the mole fractions of unsubstituted, 2-mono-, 3-mono-, 6-mono-, 2,3-di-, 2,6-di-, 3,6-di, and 2,3,6-tri-substituted AGUs, respectively. Thus, ␹ of the mono, di-, and tri-substituted AGUs (␹mono , ␹di , and ␹tri , respectively) could be calculated using Eqs. (9)–(11): ␹mono = ␹2 + ␹3 + ␹6

(9)

␹di = ␹23 + ␹26 + ␹36

(10)

␹tri = ␹236 .

(11)

In addition, DS2 , DS3 , and DS6 could be also determined using Eqs. (12)–(14): DS2 = ␹2 + ␹23 + ␹26 + ␹236

(12)

DS3 = ␹3 + ␹23 + ␹36 + ␹236

(13)

DS6 = ␹6 + ␹26 + ␹36 + ␹236

(14)

The ␹, DS2 , DS3 , and DS6 values for CA 1–3 (Eqs. 1–14) are plotted against DS in Fig. 4a. Buchanan et al. (1991) prepared a series of labeled [1-13 C]acetyl cellulose samples from the acid hydrolysis of CA, and 15 carbonyl carbon resonances were detected by 1D selective pulse INEPT NMR (INAPT). They suggested that the carbonyl resonance was shifted downfield by 0.01–0.06 ppm due to hydrogen bonding between

26

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

Fig. 1. Quantitative 13 C NMR spectra of (a) CA 1–3 and (b) HCA 1–3 dissolved in DMSO-d6 recorded at 363 K. C1 , C4 , and C6 are the resonances of C1 when the neighboring 2-position is unsubstituted, C4 when the neighboring 3-position is unsubstituted, and unsubstituted C6, respectively.

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

27

Table 1 DS, individual DS at the 2-, 3-, and 6-positions (DS2 , DS3 , and DS6 , respectively), and acetone solubility of CA 1–3 and HCA 1–3. Samples

DSa

DSMe b

DS2

DS3

DS6

DSring c

Acetone solubility

CA 1 CA 2 CA 3 HCA 1 HCA 2 HCA 3

1.26 2.05 2.70 2.31 1.81 1.28

1.28 2.06 2.72 2.30 1.83 1.31

0.21 0.45 0.80 0.74 0.52 0.28

0.26 0.66 0.92 0.79 0.54 0.32

0.79 0.94 0.97 0.78 0.75 0.68

1.26 2.05 2.69 2.31 1.81 1.28

Insoluble Insoluble Insoluble Soluble Insoluble Insoluble

a b c

Total integral values for carbonyl carbon resonances in the quantitative 13 C NMR spectra of the samples. Total integral values for methyl carbon resonances in the quantitative 13 C NMR spectra of the samples. Sum of DS2 , DS3 , and DS6 .

Fig. 2. Lineshape analysis in the carbonyl regions of the quantitative 13 C NMR spectra of (a) CA 1–3 and (b) HCA 1–3. The analyses were performed using 12 Lorentzian lines. The peaks are labeled as ij, where i is the substituted position the carbon of AGU (i = 2, 3, or 6), and j is the substituted position(s) of the hydroxyl group(s); e.g., 223 indicates the carbonyl carbon resonance of an acetyl group substituted at the 2-position of 2,3-di-substituted AGU.

AGU’s or chains. In the INAPT experiments, the 1 H resonance(s) of each AGU were selectively irradiated by the soft pulse. However, as previously reported (Kono, Hashimoto et al., 2015), 1 H resonances of CA samples except for DS = 3 overlap, and thus the selective excitation of each 1 H resonance is difficult. In this study, 12 carbonyl carbon resonances were indirectly assigned by lineshape analysis, and DS2 , DS3 , and DS6 derived by applying the ␹ values to Eqs. 12–14 (Table 3) were in good agreement with those derived from the quantification of C1, C4, and C6 (Table 1).

3.4. Mechanism of homogeneous acetylation of cellulose in [BMIM]Cl The plots of the ␹ values of eight AGUs in CA 1–3 against DS (Fig. 4a) shed light on the mechanism of the acetyl substitution of cellulose in [BMIM]Cl. Initially, the substitution at the 6-position of cellulose occurs to produce 6-mono-substituted AGU, while that at the 3-position simultaneously occurs to form the 3-monosubstituted AGU. On the other hand, 2-mono-substituted AGU is hardly produced, as confirmed from the ␹6 (0.63), ␹3 (0.11), and ␹2

28

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

Table 2 Assignment, chemical shift, height, FWHM, and Integral values (Ii ) for lines 1–12 in the carbonyl carbon resonances of CA 1–3 and HCA 1–3. Samples

CA 1

CA 2

CA 3

HCA 1

HCA 2

HCA 3

Line number Assignmenta ␦/ppm

1 626 169.63

2 6236 169.55

3 66 169.50

4 636 169.43

5 33 169.06

6 336 168.85

7 3236 168.79

8 323 168.74

9 223 168.48

10 2236 168.36

11 226 168.27

12 22 168.22

Heightb FWHM/Hz Ii c Heightb FWHM/Hz Ii c Heightb FWHM/Hz Ii c Heightb FWHM/Hz Ii c Heightb FWHM/Hz Ii c Heightb FWHM/Hz Ii c

0.27 8.1 0.10 0.83 6.2 0.20 0.15 2.7 0.04 0.17 9.9 0.12 0.59 6.2 0.07 0.29 9.9 0.06

0.21 6.5 0.06 0.99 4.1 0.15 1.00 6.4 0.74 1.00 7.5 0.51 1.00 12.4 0.22 0.24 7.5 0.06

1.00 14.1 0.63 0.93 5.6 0.19 0.07 4.1 0.03 0.14 8.1 0.07 0.92 24.3 0.40 1.00 8.1 0.45

0.08 0.6 0.02 1.00 10.8 0.41 0.32 4.2 0.15 0.19 7.2 0.09 0.30 10.7 0.06 0.26 7.2 0.09

0.13 18.2 0.11 0.02 7.5 0.01 0.01 6.4 0.01 0.34 4.8 0.10 0.21 20.6 0.08 0.24 4.8 0.07

0.10 5.2 0.02 0.92 10.8 0.38 0.24 5.3 0.15 0.27 6.2 0.09 0.31 13.5 0.08 0.30 6.2 0.09

0.23 7.7 0.08 0.37 9.4 0.13 0.93 6.9 0.73 0.95 7.7 0.49 0.58 21.2 0.22 0.23 7.7 0.06

0.13 8.3 0.05 0.50 7.8 0.15 0.08 4.6 0.04 0.26 6.2 0.10 0.78 12.4 0.17 0.42 6.2 0.12

0.14 8.6 0.05 0.69 6.1 0.15 0.11 4.6 0.06 0.20 8.4 0.10 0.81 12.9 0.19 0.49 8.4 0.12

0.24 7.52 0.08 0.29 10.5 0.11 0.75 8.1 0.69 0.82 8.8 0.49 0.81 15.2 0.22 0.23 8.8 0.06

0.11 11.6 0.06 0.50 10.1 0.17 0.10 4.1 0.04 0.20 8.9 0.12 0.24 12.2 0.05 0.26 8.9 0.05

0.06 5.84 0.02 0.00 3.7 0.00 0.02 5.2 0.01 0.05 7.2 0.02 0.26 10.3 0.05 0.18 7.2 0.04

a Assignment of the carbonyl carbons are labeled as ij , where i is the position of substituted acetyl group (i = 2, 3, or 6), and subscript j is the position of the AGU substituted with acetyl group(s); e.g., 626 indicates the carbonyl carbon resonance at the 6-position of the 2,6-di-substituted AGU. b The highest height among lines 1–12 was set to 1. c Area of each line is labeled as Ii , where subscript i is the line number. The sum of I1 –I12 was set to 1.

Fig. 3. Expanded ring carbon region of the 2D 1 H–13 C HSQC spectrum of CA 2. The 1 H and 13 C correlation signals are labeled as ij , where i is the carbon number of the AGU ring, and j is the substituted position(s) of the hydroxyl group(s); e.g., 1236 indicates the directly coupled 1 H–13 C spin pair of ring carbon and proton(s) at the 1-position of the 2,3,6-tri-substituted AGU. The assignments were carried out according to a previous report (Kono, Hashimoto et al., 2015). The arrows in the spectrum indicate the change in chemical shift caused by the acetyl substitution at the 2-, 3-, and 6-positions of AGU. Correlation signals due to the 6-mono-, 2,3-di-, 2,6-di, 3,6-di-, and 2,3,6-tri-substituted AGUs were major, and those due to the 3-mono-AGU were minor.

(0.02) of CA 1 (at DS = 1.28). Subsequent substitution preferentially occurs at the unsubstituted hydroxyl groups at the 3-position of the 6-mono-subststituted AGU to produce 3,6-di-substituted AGU, as confirmed from the changes in ␹6 and ␹26 when the DS ranged

from 1.28 to 2.05; ␹6 drastically decreased from 0.63 to 0.19, while ␹36 rapidly increased from 0.02 to 0.39. In addition, ␹2 was essentially constant at ∼0.01 during the DS change, while ␹26 steadily increased from 0.08 to 0.18, indicating that the 2,6-di-substituted

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

29

Fig. 4. DS2 , DS3 , DS6 , and mole fractions (␹) of the AGUs in CA 1–3 (a) and HCA 1–3 (b) plotted as a function DS determined by lineshape analysis (Fig. 2 and Table 2).

AGU was produced via subsequent substitution at the 2-position of the 6-mono-substituted AGU. Similarly, the 2,3-di-substituted AGU was mainly produced via substitution at the 2-position of the

3-mono-substituted AGU, because the ␹3 values decreased from 0.11 to 0.01 when DS increased. Finally, the hydroxyl groups of

30

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

Table 3 Individual DS at the 2-, 3-, and 6-positions (DS2 , DS3 , and DS6 , respectively) and DSs of CA 1–3 and HCA 1–3 determined from lineshape analysis (Fig. 2). Samples

DS2 a

DS3 a

DS6 a

DSb

CA 1 CA 2 CA 3 HCA 1 HCA 2 HCA 3

0.21 0.44 0.81 0.74 0.51 0.28

0.26 0.66 0.93 0.78 0.55 0.34

0.81 0.95 0.96 0.79 0.75 0.66

1.28 2.05 2.70 2.31 1.81 1.28

a DS2 , DS3 , and DS6 were determined by applying the ␹ values of eight AGUs to Eqs 12–14, respectively. b DS is the sum of DS2 , DS3 , and DS6 .

the di-substituted AGUs were substituted to produce the 2,3,6-trisubstituted AGU. The reactivity of the cellulose hydroxyl groups during acetylation using acetic anhydride was previously reported; Miyamoto et al. (Miyamoto, Sato, Shibata, Takahashi, & Inagaki, 1985) investigated the substituent distribution of CA samples produced from cellulose by an industrial solution-acetylation process employing sulfuric acid and reported the reactivity of cellulose hydroxyl groups as OH(6) > OH(2) > OH (3), as the primary OH(6) is more reactive than the secondary hydroxyl groups, and the acidity of OH(2) is higher than that of OH(3). In the solution acetylation of cellulose in [BMIM]Cl, the order of reactivity was OH(6)  OH(3) > OH(2) (Schlufter et al., 2006), which was confirmed here. In addition, the preferable formation of the 3,6-di-substituted AGU via the 6-mono-AGU was revealed in the homogeneous reaction system. This was related to the strong interaction between [BMIM]Cl and the hydroxyl groups of cellulose (Payal & Balasubramanian, 2014; Hassan, Mutelet, & Bouroukba, 2015). With respect to the dissolution of cellulose in [BMIM]Cl, Remsing et al. (Remsing, Swatloski, Rogers, & Moyna, 2006) reported that Cl− anions and BMIM+ cations were involved in cellulose dissolution and a stoichiometric interaction between the Cl− and hydroxyl groups of cellulose occurred based on 13 C, and 35/37 Cl NMR relaxation measurements. In addition, Kosan et al. (Kosan, Michels, & Meister, 2008) proposed the structure of cellulose in [BMIM]Cl. In the model structure, the oxygen and hydrogen atoms of cellulose hydroxyls form electron donor–electron acceptor (EDA) complexes with Cl- and [BMIM] of [BMIM]Cl, respectively, and the formation of the EDA complex occurs primarily between the 6- and 3-position of hydroxyl groups of neighboring cellulose chains, cleaving the strong intermolecular hydrogen-bonds between cellulose chains. The formation of hydrogen bonds between hydroxyl groups in the 3- and 6-positions of cellulose and [BMIM]Cl should enhance the nucleophilicity of the oxygen of the hydroxyl groups in the these positions. The ␹ values as a function of DS, as well as substituent distribution determined in this study, support the model structure of the EDA complex in [BMIM]Cl (Fig. 5). 3.5. Change in monomer composition of CA during acid-hydrolysis Next, the ␹ values of HCA 1–3 were determined in order to estimate the reaction mechanism. As shown in Fig. 1b, the integration of carbonyl carbon resonances in the quantitative 13 C NMR spectra of HCA 1–3 provided their DS (2.73, 1.80, and 1.26, respectively), and DS2 , DS3 , and DS6 could be estimated by the integral values of the split resonances for C1, C4, and C6. As summarized in Table 1, the DS of HCA 1–3 indicated that the hydrolysis of the acetyl groups occurred in the following order: C2 >C3  C6, which was in complete agreement with previous reports on the substitution distribution of CA samples prepared via hydrolysis (Miyamoto et al., 1985).

Fig. 5. Electron donor–electron accepter complex formation between cellulose hydroxyl groups and [BMIM]Cl.

Lineshape analysis was applied to the carbonyl carbon regions in the 13 C spectra of HCA 1–3; the results are shown in Fig. 2b and Table 2. The calculated spectra (dotted lines) using lines 1–12 almost fit the experimental spectra (solid lines) with r2 > 0.96 (Fig. 2b). The DS2 , DS3 , DS6 , and ␹ values of the eight AGUs were determined (Table 2) and plotted against DS by applying the I1 –I12 values for each HCA to Eqs. 1–14 in section 3.3 (Fig. 4b). The plots of ␹ as a function of DS (Fig. 4b) suggested a mechanism for the acid-catalyzed hydrolysis of CTA. Initial hydrolysis randomly occurs at the acetyl groups at the 2-, 3-, and 6-postions of CTA because ␹23 (= 0.10), ␹26 (= 0.12), and ␹36 (=0.09) are almost equivalent at DS = 2.31. The subsequent hydrolysis reaction preferably occurs at the 2-position of the 2,6-di-AGU and at the 3-position of the 3,6-di-AGU to form the 6-mono-AGU with a decrease in DS from 2.61 to 1.28, because ␹6 increased from 0.07 to 0.45. On the other hand, the hydrolysis of the 6-position of the 2,6-di- and 3,6di-AGUs hardly occurred, because ␹2 and ␹3 were approximately constant (0.03–0.05 and 0.10–0.08) when the DS change, despite the drastic increase of ␹6 . Thus, the formation of the 2-mono- and 3-mono-AGUs mainly occurred via the hydrolysis of the 3- and 2-positions of 2,3-di-AGU, respectively. The formed mono-AGUs, mainly 6-mono-AGU, were finally converted into unsubstituted AGUs to form cellulose. Notably, the accumulation of the ␹26 and ␹36 of CA chains was not observed before the increase of ␹6 , indicating that the rate of the hydrolysis of the acetyl groups at the 2- and 3-positions was greater than that at the 6-position (Fig. 4b). This could be explained by the mechanism of the acid-catalyzed ester hydrolysis and the order of electron charge density of the cellulose hydroxyl groups at the 2-, 3-, and 6-positions (Kono et al., 2002). The hydrolysis of the acetyl groups begins with the abstraction of a proton from a hydroxonium ion in the solvent by the ester to make the ester carbonyl more electrophilic. The oxygen of water functions as a nucleophile and attacks the electrophilic carbonyl carbon, which results in the formation of a cationic tetrahedral intermediate. Next, a proton (a hydrogen ion) from water binds to the carbonyl carbon and is transferred to the oxygen atom of the ester derived from the cellulose hydroxyl by the unpaired electrons of the oxygen atom, which facilitates the dissociation of the hydroxyl group from the ester. Thus, the electronegativity of the oxygen atom in the hydroxyl groups of cellulose promotes the acid-catalyzed ester hydrolysis. Because the electron density of the hydroxyl groups of cellulose ranks as OH(2) > OH(3) > OH(6) (Kono et al., 2002), the electronegativity of the hydroxyl groups in cellulose reflects the rate of hydrolysis at the 2-, 3-, and 6-positions. Therefore, the very fast hydrolysis at the

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

2- and 3-positions resulted in a prompt increase of ␹6 in CA chains, without increasing ␹26 and ␹36 . 3.6. Relationship between monomer composition and acetone-solubility As mentioned above, because there is an efficient one-pot acetylation process to obtain CA with complete acetone-solubility, commercial acetone-soluble CA (DS = 1.9–2.5) is produced by the two-step reaction process. The first step is the complete acetylation of cellulose via acetylation (Tang, Hon, & Zhu, 1997), and the second step is the partial hydrolysis of CTA by sulfonic. Therefore, there is a close relationship between the substituent distribution and acetone-solubility of CA (Fisher et al., 2008). In this study, CA 1–3 were produced from cellulose by a onestep reaction, and HCA 1–3 were produced from the acid-hydrolysis of CTA; the samples correspond to the CA sample produced by the industrial two-step process. Among these samples, only HCA 1 (DS = 2.31) exhibited solubility in acetone, while the other samples did not (Table 1). Comparison of the DS2 , DS3 , DS6 , and ␹ profiles obtained by the one-step acetylation process of cellulose (Fig. 4a) with those from the acid-hydrolysis of CTA (Fig. 4b) in the DS range of 1.9-2.5 sheds light on the relationship between substituent distribution and acetone solubility. In the case of substituent distribution, the one-step reaction provided different DS2 , DS3 , and DS6 values in the range of 1.9–2.5, while these values were relatively uniform in the hydrolysis reaction. In addition, in contrast to the finding that the ␹mono , ␹di , and ␹tri for the hydrolyzed CA were the same around a DS value of 2.1, the non-uniformity of these values was particularly remarkable in CA prepared from the one-step reaction process. ␹di is higher keeps at higher value in the DS range of 1.9–2.5, while ␹mono and ␹tri were considerably lower than ␹di . Notably, ␹2 and ␹3 of hydrolyzed CA were ca. 0.05 and 0.08, respectively, while those produced in the one-step reaction were almost negligible. Moreover, ␹23 , ␹26 , and ␹36 of the hydrolyzed CA samples were between 0.08–0.14, while ␹36 was far higher than ␹23 and ␹26 in the one-step reaction system. Based on these findings, the hydrolyzed CA in the DS region required for acetone solubility had a highly homogeneous substituent distribution, and ␹ values compared with the CA samples prepared by the one-step acetylation of cellulose dissolved in [BMIM]Cl. 3.7. Monomer sequences along CA chains Lineshape analysis of carbonyl peak patterns allowed mole fractions of eight AGUs to be determined, indicating that 13 C chemical shifts of carbonyl carbons reflect the magnetic inequivalence of their locations within each AGU. Therefore, analysis of carbonyl resonances was not suited to determine the sequence of the eight AGUs in the CA chain. Polymer properties, such as solubility in different solvents, are generally affected by the sequence and composition of constituent monomers. Regarding the through-space interaction of CA acetyl substituents, Tezuka (Tezuka, 1994) provided an excellent way of determining the sequence of AGUs in the polymer chain based on the 1 H–1 H nuclear Overhauser effect (NOE) spectra of CTA. Though-space interaction between acetyl 1 H atoms at the 3- and 6-positions via neighboring AGUs could be detected in CTA dissolved in DMSO-d6 , while NOE correlations were observed between 3- and 6-positions as well as between 2- and 6-positions in CTA dissolved in CDCl3 , indicating the different CTA conformations in these solvents. In other words, low-polar solvents (such as CHCl3 ) were preferred to highly polar ones (such as DMSO) for the spectroscopic characterization of the vicinal AGUs of CA. However, the solubility of CA samples strongly depends on solvent polarity and DS (Fisher et al., 2008), with NMR characterization of CA samples in low-polar solvents being restricted by the DS of CA samples. Thus,

31

NOE- and/or rotating frame Overhauser effect (ROE)-based NMR measurements in low-polar solvents provide additional important information on the distribution of AGU units in the CA chain. 4. Conclusions This study revealed that CA samples produced by different methods, i.e., homogeneous acetylation of cellulose and hydrolysis of cellulose triacetate, exhibit different substitution patterns of the eight AGUs. In addition, we observed that the ␹ values of AGUs strongly affected the solubility of CA in acetone. Therefore, the method used in this study for estimating the ␹ values of the eight AGUs comprising CA chains represents a valuable tool for exploring structure-property relationships in CA, and could be applied for the structural characterization of other cellulose esters, as well as for quality control in industrial CA production. The accumulation of NMR data for other CA samples prepared under various reaction conditions, such as temperature, reaction system (homogenous and heterogeneous), esterification reagents, solvents, will further reveal the mechanism of CA production and hydrolysis of acetyl groups in CA. Acknowledgements Funding: This work was supported by the Japan Society for Promotion of Science (JSPS) [grant number JP16K05802]. References Bifari, E. N., Bahadar Khan, S., Alamry, K. A., Asiri, A. M., & Akhatar, K. (2016). Cellulose acetate based nanocomposites for biomedical applications: A review. Current Pharmaceutical Design, 22, 3009–3017. Buchanan, C. M., Edgar, K. J., Hyatt, J. A., & Wilson, A. K. (1991). Preparation of cellulose [1-carbon-13]acetates and determination of monomer composition by NMR spectroscopy. Macromolecules, 24, 3050–3059. Fisher, S., Thümmler, S., Volkert, B., Hettrich, K., Schmidt, I., & Fisher, K. (2008). Properties and applications of cellulose acetate. Macromolecular Symposia, 261, 89–96. Glasser, W. G., McCartney, B. K., & Samaranayake, G. (1994). Cellulose derivatives with a low degree of substitution: 3. The biodegradability of cellulose esters using a simple enzyme assay. Biotechnology Progress, 10, 214–219. Hassan, E.-S. R. E., Mutelet, F., & Bouroukba, M. (2015). Experimental and theoretical study of carbohydrate-ionic liquid interactions. Carbohydrate Polymers, 127, 316–324. Heinze, T., Schwikal, K., & Barthel, S. (2005). Ionic liquids as reaction medium in cellulose functionalization. Macromolecular Bioscience, 5, 520–525. Kamide, K., & Okajima, K. (1981). Determination of distribution of O-acetyl group in trihydric alcohol units of cellulose acetate by carbon-13 nuclear magnetic resonance analysis. Polymer Journal, 13, 127–133. Kono, H., Yunoki, S., Shikano, T., Fujiwara, M., Erata, T., & Takai, M. (2002). CP/MAS 13 C NMR study of cellulose and cellulose derivatives. 1: Complete assignment of the CP/MAS 13 C NMR spectrum of the native cellulose. Journal of the American Chemical Society, 124, 7506–7511. Kono, H., Oshima, K., Hashimoto, H., Shimizu, Y., & Tajima, K. (2016a). NMR characterization of sodium carboxymethyl cellulose: Substituent distribution and mole fraction of monomers in the polymer chains. Carbohydrate Polymers, 146, 1–9. Kono, H., Oshima, K., Hashimoto, H., Shimizu, Y., & Tajima, K. (2016b). NMR characterization of sodium carboxymethyl cellulose 2: Chemical shift assignment and conformation analysis of substituent groups. Carbohydrate Polymers, 150, 241–249. Kono, H., Fujita, S., & Tajima, K. (2017). NMR characterization of methylcellulose: Chemical shift assignment and mole fraction of monomers in the polymer chains. Carbohydrate Polymers, 157, 728–738. Kono, H. (2013). Chemical shift assignment of the complicated monomers comprising cellulose acetate by two-dimensional NMR spectroscopy. Carbohydrate Research, 375, 136–144. Kono, H., Anai, H., Hashimoto, H., & Shimizu, Y. (2015). 13 C-detection two-dimensional NMR approaches for cellulose derivatives. Cellulose, 22, 2972–2942. Kono, H., Hashimoto, H., & Shimizu, Y. (2015). NMR characterization of cellulose acetate: Chemical shift assignments, substituent effects: And chemical shift additivity. Carbohydrate Polymers, 118, 91–110. Kosan, B., Michels, C., & Meister, F. (2008). Dissolution and forming of cellulose with ionic liquids. Cellulose, 15, 59–66.

32

H. Kono et al. / Carbohydrate Polymers 170 (2017) 23–32

Miyamoto, T., Sato, Y., Shibata, T., Takahashi, M., & Inagaki, H. (1985). 13 C NMR spectral studies on the distribution of substituents in water-soluble cellulose acetate. Journal of Polymer Science Part A: Polymer Chemistry, 23, 1373–1381. Noll, R. J., & Pires, A. (1980). A new nonlinear least square algorithm for Voigt spectral lines. Applied Spectroscopy, 34, 351–360. Payal, R. S., & Balasubramanian, S. (2014). Dissolution of cellulose in ionic liquids: An ab initio molecular dynamics simulation study. Physical Chemistry Chemical Physics, 16, 17458–17465. Remsing, R. C., Swatloski, R. P., Rogers, R. D., & Monya, G. (2006). Mechanism of cellulose dissolution in the ionic liquid 1-n-butyl-3-methylimidazolium chloride: A 13 C and 35/37 Cl NMR relaxation study on model systems. Chemical Communications, 1271–1273. Saake, B., Horner St, & Puls, J. (1998). Progress in the enzymatic hydrolysis of cellulose derivatives. ACS Symposium Series, 688, 201–216. Schlufter, K., Schmauder, H.-P., Dorn, S., & Heinze, T. (2006). Efficient homogeneous chemical modification of bacterial cellulose in the ionic liquid 1-N-butyl-3-methylimidazolium chloride. Macromolecular Rapid Communications, 27, 1670–1676.

Tang, L.-G., Hon, D. N.-S., & Zhu, Y.-Q. (1997). An investigation in solution acetylation of cellulose by microscopic techniques. Journal of Applied Polymer Science, 64, 1953–1960. Tezuka, Y., & Tsuchiya, Y. (1995). Determination of substituent distribution in cellulose acetate by means of a 13 C NMR study on its propanoated derivative. Carbohydrate Research, 273, 83–91. Tezuka, Y. (1993). 13 C NMR determination of the distribution of two ester substituents in cellulose acetate butyrate. Carbohydrate Research, 241, 285–290. Tezuka, Y. (1994). Nuclear Overhauser effect spectroscopy (NOESY) detection of the specific interaction between substituents in cellulose and amylose triacetates. Biopolymers, 34, 1477–1482. Wu, J., Zhang, J., Zhang, H., He, J., Ren, Q., & Guo, M. (2004). Homogeneous acetylation of cellulose in a new ionic liquid. Biomacromolecules, 5, 266–268.