Polymer 53 (2012) 50e59

Contents lists available at SciVerse ScienceDirect

Polymer
journal homepage: www.elsevier.com/locate/polymer

Novel poly(vinyl alcohol)-g-poly(hydroxy acid) copolymers: Synthesis and characterization

Ainhoa Lejardi a, Agustin Etxeberria b, Emilio Meaurio a, Jose-Ramon Sarasua a, *
a b

School of Engineering, University of the Basque Country (EHU-UPV), Alameda de Urquijo s/n, 48013 Bilbao, Spain Department of Polymer Science and Technology, Institute of Polymer Materials, University of the Basque Country (EHU-UPV), M. de Lardizabal 3, 20018 Donostia, Spain

a r t i c l e i n f o
Article history: Received 6 October 2011 Received in revised form 8 November 2011 Accepted 17 November 2011 Available online 25 November 2011 Keywords: Poly(vinyl alcohol) (PVA) Grafting Hydroxy acids (HA)

a b s t r a c t
Poly(vinyl alcohol) (PVA) was reacted with three hydroxy acids (HA), namely D,L-lactic acid (LA), glycolic acid (GA) and D,L-3-hydroxybutyric acid (HB). The graft copolymers obtained were thoroughly characterized by 1H and 13C NMR, FTIR, and DSC. Copolymer compositions were in the range 14e45 mol% HA with average lateral chain lengths in the range 1.1e1.3. The C]O stretching band, arising from the lateral polyester chains, presents signicant differences from that of pure polyesters. In case of the VALA and VAGA copolymers, carbonyl groups are almost completely interassociated with hydroxyl groups and as a result the carbonyl band presents a single contribution; however, splitting appears in VAHB. The band at about 1735 cm1 already observed for Poly(3-hydroxybutyrate) P3HB was reexamined in the light of molecular models for VAHB and the splitting observed was attributed to CeHO]C and to OeHO] C hydrogen bonding. The thermal analysis of copolymers demonstrates that esterication suppresses crystallinity and increases free volume, both accounting for a Tg reduction with regard to PVA. The stronger intermolecular hydrogen bonding interactions found in PVA with the chemically modied PVAs opens an interesting way towards miscibility with polyesters and other polymer systems containing carbonyl groups. 2011 Elsevier Ltd. All rights reserved.

1. Introduction In recent years the number of research works devoted to the synthesis of macromolecular designs (such as block, comb, brush or star polymers) based on biocompatible and biodegradable polymer materials has shown a noticeable increase [1]. The control of the macromolecular structure can lead to versatile polymers with tuned chemical, physical and mechanical properties. The synthesis of Bio-related polymers can be accomplished using different methodologies such as gene technology, enzymatic polymerization and nally pure chemical synthesis reactions. Polymers containing reactive groups are one interesting starting point for the architectures based on chemical synthesis. Poly(vinyl alcohol) (PVA) is a well known water-soluble polymer with widespread commercial applications [2]. The repeating unit of PVA contains reactive hydroxyl groups that can be converted to ester groups through esterication with acids or ring opening reactions. Due to its biodegradable nature PVA is a particular synthetic vinyl polymer, although biodegradability is somewhat

* Corresponding author. E-mail address: jr.sarasua@ehu.es (J.-R. Sarasua). 0032-3861/$ e see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.polymer.2011.11.029

low. Furthermore, PVA has been cleared by the US Food and Drug Administration (FDA) as a safe component for coating and adhesives in contact with fatty foods [3]. It is also used as a component of cosmetics, bacteriostatic agents, externally applied medicines and biomedical materials [2e5]. As biomedical material, PVA has found applications for the production of soft contact lenses or articial muscles [2]. In these cases, the base polymer is crosslinked to avoid dissolving in the aqueous medium. There are, however, alternative routes to increase the water resistance of PVA. One of them is blending PVA with biodegradable materials of hydrophobic nature to obtain polymer blends with tuned hydrophilicities. Biodegradable polyesters (such as polylactides) can be appropriate blending counterparts since they carry hydrogen bond acceptor groups, needed to promote the miscibility of the system through the establishment of specic interactions. However, these blends show phase separation and poor compatibility according to different investigations [6e8]. Chemical modication of PVA, particularly grafting with biodegradable hydroxy acids, might improve its miscibility with hydrogen bond acceptor counterparts for several reasons. First the introduction of bulkier lateral groups should decrease the high autoassociation density of PVA, which generally impairs its interassociation. Second, the modication reaction moves away the

A. Lejardi et al. / Polymer 53 (2012) 50e59

51

hydroxyl groups bonded to the chain skeleton of PVA to the end of the lateral chains, where the formation of intermolecular hydrogen bonds should be favoured due to reduced steric hindrances [9]. Grafting of PVA with Lactic acid has been already reported [10e12], but some properties of the resulting products are apparently inconsistent. Carlotti et al. grafted PVA with LA after water removal (i.e., in the melt) without catalysts, obtaining watersoluble polymers in all cases [11]. They reported LA compositions ranging from 14 to 33 mol % and estimated grafted chain lengths below two repeat units (grafted chain lengths could not be determined exactly because of overlapping of the proton peaks in the 1H NMR spectra). On the other hand, Ding et al grafted PVA with Lactic acid by melt polycondensation using stannous chloride as catalyst, obtaining non-water-soluble polymers in all cases [12]. These authors reported LA compositions from 21 to 53 mol % and the degrees of polymerization were in the 1.7e2.2 range. Surprisingly, even though some of the copolymers obtained by both researchers show very similar compositions, water resistances are completely different. The present paper deals with the synthesis and detailed characterization of PVA modied with lactic acid (LA), glycolic acid (GA) and hydroxybutyric acid (HB). Grafting PVA with HB was investigated for the rst time in our lab as described here and in ref. 14. Grafting with GA has only been reported by Carlotti et al, but the copolymers were not characterized in detail (particularly the polymerization degree of the lateral chains was only roughly estimated) [11]. Uncatalyzed grafting conditions will be used for two reasons: on the one hand, we would like to understand the structural features explaining the improved water solubility of these materials. On the other hand, their blends should cover wider hydrophilicity ranges. Moreover, we are more interested in substituting the largest possible amount of vinyl alcohol groups rather than in lengthening the lateral chains. Actually, our best bet would be the synthesis of the corresponding poly(vinyl hydroxyester)s; i.e., the polymers resulting from the double bond addition polymerization of vinyl glycolate, vinyl lactate and vinyl hydroxybutyrate. However, neither the monomers nor the polymers are commercially available, and the synthesis of the poly(vinyl hydroxyester)s starting from the raw chemical compounds can be a formidable research work. Therefore, a simpler route is chosen in this paper: the esterication of PVA with the corresponding hydroxy acids. The miscibility of these copolymers with selected counterparts has been already investigated in our research group. Blends with poly(-caprolactone) (PCL) show phase separation, but compatibility is strongly improved [13]. It is necessary to realize that, unfortunately, the vinyl alcohol repeat units still dominate the composition of the obtained graft copolymers, and their total substitution seems not actually possible. Blends with poly(vinyl pyrrolidone) PVP show stronger intermolecular interactions than those reported for other hydroxylated polymers; and particularly the formation of double hydrogen bonds is reported for the rst time in polymer blends in the solid state [14].

close to atactic. Also, 1% of residual acetylation grade was determined [16]. 2.1. Esterication of PVA PVA was esteried with varying feed ratios of LA, GA and HB (see scheme 1 and Table 1) to obtain modied polymers of different composition. In a specic example, PVA (10 g) was dissolved in hot distilled water (50 mL). LA (20 g) was then slowly added to the reactor. The esterication process was carried out in two stages, rst the solution was maintained at 100  C for 2 h, and then water was removed till dryness by evaporation at about 120  C. The product was redissolved in water at 25  C and reprecipitated twice in acetone (10 times in volume) to remove unreacted acid and the poly(lactid acid) chains formed as byproduct. The precipitate was additionally puried in a Soxhlet for 24 h with chloroform to assure complete removal of residual monomer and byproducts. Finally, the modied PVA was dried under vacuum at 30  C for 24 h. 2.2. Differential scanning calorimetry (DSC) Calorimetric studies were carried out on a differential scanning calorimeter (Mettler DSC 2920) in sealed aluminium pans. The weight of sample used was in the 5e10 mg range. The scans were carried out from 20 to 220  C at a heating rate of 20  C min1. The instrument was calibrated using an Indium Standard (melting point 156.6  C). 2.3. Infrared spectroscopy (FTIR) Infrared spectra of modied PVA were recorded on a Nicolet AVATAR 370 Fourier transform infrared spectrophotometer (FTIR). Samples were prepared by casting 0.2 wt % hexauoroisopropanol (HFIP) solutions on KBr pellets. 2.4. Nuclear Magnetic Resonance (NMR)
1 H and 13C NMR spectra were recorded in a Bruker Avance DPX 300, which corresponds to 300.16 and 75.5 MHz frequencies for 1H and 13C respectively, in 5 mm o.d. sample tubes using 0.7 mL of deuterated dimethyl sulfoxide (DMSOd6), at room temperature

Scheme 1. Esterication of PVA with D,L-lactic acid.

Table 1 Reaction feed ratios, naming conventions and glass transitions for the polymers synthesized in this study. Hydroxy acid Molar feed ratio [Acid]/[PVA] e 1/1 4/1 1/4 4/1 4/1 Copolymer Namea PVA VALA28 VALA45 VAGA24 VAGA44 VAHB14 Tg ( C) 74 57 68 48 38 60

2. Materials and methods The biodegradable hydroxy acids selected to modify the PVA chain, D,L-lactic acid (LA), glycolic acid (GA) and hydroxybutyric acid (HB), were obtained from SigmaeAldrich. Poly(vinyl alcohol), PVA (Mw 20,000), was also obtained from SigmaeAldrich and characterized by 13C NMR. The analysis of the methine Carbon at triad sequence level [15] gave the following relative intensities: (mm) 0.207, (rm) 0.515 and (rr) 0.278. The isotacticity coefcient s, where s (m) (mm) (mr)/2, took a value of 0.465,
e LA LA GA GA HB

a The naming convention for the copolymers uses the letters VA to indicate vinyl alcohol, the following two letters refer to the hydroxy acid used in the modication reaction, and the last two numbers indicate the hydroxy acid molar % according to 1 H NMR.

52

A. Lejardi et al. / Polymer 53 (2012) 50e59

(z30  C). Experimental conditions were as follows: a) for 1H, 10 mg of sample; 3 s acquisition time; 1 s delay time and 32 scans; b) for 13 C, 60 mg, inverse gated decoupled sequence; 1.74 s acquisition time; 5 s delay time and more than 10,000 scans; c) for 2D-NMR HSQC (heteronuclear single quantum correlation), the number of collected complex points 2048 for 1H dimension with a recycle delay of 1.5 s, the number of transients 32 and 512 time increments where recorded in the 13C dimension. 2.5. Solubility tests The solubilities of PVA and of the grafted copolymers have been tested in Hexauoroisopropanol (HFIP), N,N-dimethyl formamide (DMF), Dimethyl sulfoxide (DMSO), water, methanol, ethanol and tetrahydrofurane (THF). About 0.05 g of polymer and 1 ml of solvent were gently shaked for about 1 h to observe the solubility of the polymers. PVA solutions were heated to disrupt the crystalline structure, but solubility was also observed at room temperature. 3. Results and discussion 3.1. Characterization by NMR Fig. 1 shows the 1H NMR spectra for PVA and selected graft copolymers and Scheme 2 lists the distinguishable H and C atoms present in PVA and in the VAegeHA copolymers. As can be seen, the 1H NMR indicates that the grafting reaction has only partially esteried the OH groups of PVA. All the signals have been assigned to the H labelled in Scheme 2 and their chemical shifts are shown in Table 2. In the cases where solvent and sample signals are overlapped, chemical shifts have been obtained from the 2D HSQC spectra (discussed later). All the values are in good agreement with those reported for the corresponding homopolymers and similar copolymers [12,17e19].

NMR is a well-known technique to obtain the composition of copolymers among other valuable information. For the VAHB copolymers, the following integration regions can be dened (see Fig. 1b and Table 2): - A: Intensity of the methylene H in the side hydroxybutyryl moieties (d and h, Scheme 2). - B. Intensity of the methylene H in the main chain (a and a0 , Scheme 2) plus intensity of the methyl H in the lateral hydroxybutyryl units (f and j, Scheme 2). From these integrals, the molar composition of the VAHB copolymers can be easily calculated using the following equation:

HB mol fraction

A B A=2

(1)

Similarly, the integration regions for the VAGA copolymers are (Fig. 1d, Table 2): - C. Intensity of the methylene H in the grafted glycolyl units (r and t, Scheme 2) plus intensity of the methyne H in the main chain (b and b000 , Scheme 2) plus intensity of the hydroxyl H. - D. Intensity of the methylene H in the main chain (a and a000 , Scheme 2). From these integrals, the molar fraction of VA units in the VAGA copolymers can be simply calculated as D/C, hence:

GA mol fraction

CD C

(2)

However, VALA copolymers do not show any independent signal corresponding to the grafted hydroxy acid units (recall integration region A in HB) or to the VA units (recall integration region C in GA). The methine groups located in the main chain are overlapped with

Fig. 1. 300 MHz 1H spectra: a) pure PVA; b) VAHB14; c) VALA45; d) VAGA44.

A. Lejardi et al. / Polymer 53 (2012) 50e59

53

a'

b' d e
z 3

- CH2 - CH OH

- CH2 - CH c

g h i C - CH2 - CH - OH CH3 j b'''

with solvent or VA methine signals. Anyway, the intensity areas of the observed internal and terminal peaks allow us to calculate the degree of polymerization of the grafted ester chain, DP, using the following expression [12,19].

VA

g-HB

f a'''

DP

I 1 T

(4)

- CH2 - CH n o k l O - C - CH - O -z C - CH - OH O CH3 m LA O CH3 p

- CH2 - CH s t q r O - C - CH2 - O -z C - CH2 - OH O GA O

Scheme 2. Distinguishable H and C atoms present in PVA and in the VAegeHA copolymers.

the methine groups in the grafted lactyl chains (b00 and b overlap with l and o respectively) and the methylene H in the VA units (a) also overlap with the methyl H located in the internal lactyl units (m). As a consequence, the degree of substitution, DS, and the degree of polymerization of the grafted chain, DP (DP z 1, Scheme 2), must be obtained in order to accurately calculate the molar composition of the VALA copolymers. The DP of the grafted chain can be obtained from the ratio between internal (I) and terminal (T) H of the hydroxy acids. However, two independent signals have only been observed in case of HB, see the two lines in the A region of Fig. 1b). Consequently, 13C NMR spectra have been recorded (Fig. 2) using a sequence that assures that intensity can be used quantitatively in order to calculate the molar compositions and the degree of polymerization of the grafted chain. The chemical shifts of the signals have been assigned to the C atoms labelled in Scheme 2 and their chemical shifts are shown in Table 2. For the VALA copolymers, the composition can be calculated from:

The determined DP values are shown in Table 3. DP is close to 1.1 in all the samples, excluding VAGA44 where its value is 1.3. Thus, the grafted ester chain does not practically grow since only one hydroxy acid unit is added per esteried OH group. Both 1H and 13C spectra show additional splitting in the VA methylene signal, arising from the copolymer sequences. Taking into account that only three signals can be observed, main chain sequence analysis must be based on diads which will be denoted as VAeVA, gVAeVA and gVAegVA. Their analysis informs about the microstructure and also provides an alternative way to calculate the degree of substitution, DS. Unfortunately DMSO overlaps with the gVAegVA signal in the 13C spectra. Regarding the 1H spectra, even though the methyl groups of the internal repeat units overlap with the metylene groups in the HB and LA cases, the diad fractions can be still accurately calculated using the previously determined internal/terminal intensity ratio. Furthermore, as can be seen in Fig. 3bed), well-distinguishable signals can be obtained using the 2D HSQC technique for the three sequences of VA methylene. The sequence mole fractions have been obtained from both 1H and 2D HSQC spectra and average values are given in Table 3. As can be seen, the gVAeVA and gVAegVA mole fractions increase with the modication degree. In fact, the gVAegVA signal is hardly observed in samples with lower modication degrees, see Fig. 3b). The diad mole fractions allows calculating the degree of substitution of the PVA chain (DS) [20]:

DS gVA gVA

gVA VA 2

(5)

LA mol fraction

E E0 F F

(3)

Similar expressions can be deduced for both VAHB and VAGA. Table 3 compiles the molar compositions obtained from 13C NMR, in good agreement with the results measured from 1H NMR. The signals of the internal (I) and terminal (T) Carbons are easily identied in the 13C spectra, see Fig. 2bed). Actually, both methyl and carbonyl give two independent lines in VAHB and VALA copolymers while in the case of VAGA splitting is only observed for the latter. Other Carbons are also sensitive but their signals overlap
Table 2 H and C chemical shifts, in ppm from TMS, corresponding to PVAegeHA copolymers. Polymer PVA Label a b OH a0 b0 c d e f g h i j

The determined DS values are given in Table 3. As can be seen, the hydroxy acid molar fraction, (HA), is in general only slightly superior to DS; providing qualitative support to the small degrees of polymerization of the grafted chain, DP, calculated before according to Eq. (4). The parameter h provides a measure of the blocky character of the polymer at the diads level [20]:

VA gVA VA gVA 2VAgVA 21 HAgVA

(6)

Eq. (6) can only actually be applied if the grafted ester chain length, DP, is equal to 1, condition that is approximately fullled by

dH
1.45a 3.89a 4.32 1.9e1.6a 5.13 e 2.57 5.12 1.87b e 2.30 4.00 1.11b

dC
45e44a 68e64a e 43e38a 69.2 174.1 40.6 67.4 18.1b 171.6b 44.7 63.8 23.7b

Polymer VAegeLA

Label a00 b00 k l m n o p a000 b000 q r s t

dH
1.9e1.6a 5.14 e 4.98 1.42 e 4.1 1.27 1.9e1.6a 5.12 e 4.65 e 4.0b

dC
43e38a 69.9 169.8 68.9 17.0 174.1 66.3 20.6 43e38a 69.9 174.7 69.2 172.6b 60.1b

VAegeHB

VAegeGA

a b

These signals are splitted due to both tacticity and copolymer microstructures. See text and Figs. 1e4. In these cases, given position corresponds to the most intense line.

54

A. Lejardi et al. / Polymer 53 (2012) 50e59

Fig. 2. 75.5 MHz

13

C spectra: a) pure PVA; b) VAHB14; c) VALA28; d) VAGA24.

the copolymers investigated in this paper. Using the data shown in Table 3, we have obtained values of h in the 0.90e1.03 interval, indicating random distribution of diads (h 1). A similar behaviour was reported for the re-acetylation reaction of poly(vinyl alcohol) [21]. The 2D HSQC spectra of the three grafted PVA copolymers also show additional signals in the region of the VA methine (low eld region, Fig. 4) that must be ascribed again to microstructural effects, and can therefore be used to corroborate spectral results calculated above. First, distinguishable signals for internal and terminal groups of the side chain are observed in Fig. 4b)ed). For instance, in Fig. 4d) two signals corresponding to internal (r) and terminal (s) glycolyl methylene pairs are clearly observed. A similar event is observed in Fig. 3b)ec) in the HB and LA methyl region. These intensities provide an alternative route to calculate DP values according to Eq. (6), and have been used in Table 3 to calculate the average results provided. Second, the PVA methine signal gives two well separated signals: one for the esteried methines, at approximately 5.1e70 ppm, and another complex signal in the region around
Table 3 Hydroxy acid molar fraction, sequence distribution, degree of substitution of PVA (DS), random character (h) and degree of polymerization (DP z 1) in the grafted poly(vinyl alcohol)s. HA mol fraction
1

3.7e66 ppm (see Fig. 4aed), assigned to the non-esteried VA methine. Splitting of the latter signal must be assigned to copolymer triads. We propose that the signals at approximately 3.73e64.6 and 3.69e66.4 ppm are due to gVAeVAeVA triads and the signal at 3.51e64.3 ppm to gVAeVAegVA sequences, while the signals located at the same positions of the pure PVA belong to VAeVAeVA triads and all the triads containing gVA as central segment are located at 5.1e70 ppm. This assignment, similar to that reported for poly(vinyl acetate-co-vinyl alcohol) [16], can be conrmed applying the expression fullled by the relative integrals of diads and triads:

VA VA VA VA VA VA VA gVA=2 VA gVA VA VA gVA 2gVA VA gVA gVA gVA 1 VA VA VA gVA

(7)

Main chain sequence mole fractiona C (VAeVA) 0.77 0.56 0.34 0.65 0.44 (VAgVA) 0.22 0.38 0.50 0.30 0.46 (gVAegVA) z0.01 0.06 0.16 0.05 0.10

DS

DPb

13

VAHB14 VALA28 VALA45 VAGA24 VAGA44

a b

0.14 0.28 0.45 0.24 0.44

0.12 0.28 0.42 0.22 0.40

0.12 0.25 0.41 0.20 0.33

0.95 0.94 1.03 0.90 0.94

1.17 1.08 1.15 1.15 1.27

13

Average values of those obtained from 1H and 2D HSQC spectra. Average values obtained using internal-terminal signal relations from 1H, and 2D HSQC spectra.

Eq. (7) provides an alternative route to calculate diad sequences and consequently DS values using Eq. (5). For instance, in the case of VALA28 the relative integrals of (VAeVAeVA), (VAeVAegVA), (gVAeVAegVA) sequences are 0.42, 0.31, and 0.03 respectively, and they provide the relative integrals of 0.575, 0.37 and 0.055 for the VAeVA, VAegVA, geVAegVA sequences, respectively, which are in excellent agreement with the values reported in Table 3. Moreover, the values DP and DS values obtained from the region shown in Fig. 4 are in good agreement with the ones calculated from the other spectra, thus, Table 3 compiles the average values. In addition, we have to point out that the relative integrals of the three signals of the VAeVAeVA triad of PVA, splitted due to tacticity, remain basically constant in the copolymers. Thus, it can be concluded that tacticity does not inuence the grafting reaction. Finally, the integration of the 2D HSQC signals for quantitative purposes deserves an ending remark. In that type of experiments, cross-signal intensity depends on the particular coupling constants, J, among other parameters [22]. Hence, only integrals of the same CeH pairs have been related. In this case, it is possible to assume that each CeH pair will give the same intensity independently of

A. Lejardi et al. / Polymer 53 (2012) 50e59

55

Fig. 3. High-eld region in the 2D HSQC spectra: a) pure PVA; b) VAHB14; c) VALA28; d) VAGA24. Letters refer to the labels given in Scheme 2 and sequences cited in the text.

Fig. 4. Low-eld region in the 2D HSQC spectra: a) pure PVA; b) VAHB14; c) VALA28; d) VAGA24. Letters refer to the labels given in Scheme 2.

56

A. Lejardi et al. / Polymer 53 (2012) 50e59

the sequence as well as the substituent nature [23]. Taking into account the fact that determined relative values for DP, copolymer sequences and DS are in good agreement with the calculated ones from both 1H and 13C, the validity of the assumption is conrmed.

3.2. Characterization by FTIR and analysis of the interactions Fig. 5 shows the infrared spectra of pure PVA and selected graft copolymers. Both the intense band located at about 1750 cm1, characteristic of the carbonyl stretching mode of esters, and the band located at about 1200 cm1, attributed to the CeO stretching mode of the ester moiety, suggest successful esterication. Since these polymers contain one hydroxyl group per vinyl repeating unit, the area of the hydroxyl stretching region between 3000 and 3800 cm1 has been used to normalize the spectra in Fig. 5. Hence, the intensity of the C]O stretching band in the normalized spectra should be approximately proportional to the HA/VA molar ratio in the copolymers. Notice the approximate nature of this assumption, since the molar extinction coefcients for both the OH and the C]O groups also depend on their association patterns [24]. Fig. 6 plots the ratio of intensities IC]O/IOH versus the molar composition of the modied polymers (in terms of molar %, as dened in Table 3). The experimental points t a linear curve passing close to the origin. Therefore, the composition of the copolymers can be estimated multiplying the ratio of intensities IC]O/IOH by the factor 200 (the slope of the plot in Fig. 6). The slight deviation from the origin observed in the actual plot is attributed to the absorption corresponding to residual acetylation (about 1% mol according to 1H NMR spectra) in PVA. The analysis of C]O stretching band of the graft copolymers (Fig. 7) is of special interest since it provides information about fundamental aspects such as the association behaviour. Before dealing with the analysis of the carbonyl band for the PVA sample grafted with Lactic acid, namely VALA28, it is necessary to recall the specic features of the C]O stretching band of polylactides. The typical spectrum of amorphous PLAs shows a complex prole [25,26]; but second derivative spectroscopy reveals three components located at about 1776, 1759 cm1 and 1749 cm1. The splitting of this band has been attributed to conformational sensitivity arising from the through bond coupling of the C]O stretching mode; and the observed components were attributed to gg, gt and tt conformers respectively [25,26]. However, in the C]O stretching band for the VALA28 copolymer (Fig. 7) different components can not be observed. As has been shown above, the length of the LA

Fig. 6. Molar composition versus IC]O/IOH. The insert sketches the integration ranges.

residues in VALA28 is close to unity in most cases; therefore the through bond coupling mechanism does not exist in the lateral residues because of the lack of neighbours to couple with. Therefore, conformational splitting cannot be expected, and the LA residues should be observed at the same location regardless of their conformation, as actually occurs according to Fig. 7. The location corresponding to the unperturberd C]O stretching mode of PLAs is 1763 cm1 [25,27], however the C]O stretching band for VALA28 in Fig. 7 is located at 1733 cm1, red shifted by 30 cm1. The observed shifting can be attributed to hydrogen bonding, therefore, all C]O groups in VALA28 are apparently hydrogen bonded, since no component can be observed at higher wavenumbers attributable to free C]O groups. The achievement of complete interassociation for the C]O groups is in fact another interesting feature of these systems; compared, for example, with the typical behaviour observed for polymer blends between polyesters and hydroxyl containing polymers, where the fraction of associated C]O groups always attains a limiting value [24,28]. This difference between the graft copolymers and the blends can be attributed to the occurrence of intramolecular screening in the latter [29], impairing the interassociation even at high dilution levels.

VALA28 VAGA44 VAHB14

Absorbance

Y" VAHB14

1800

1780

1760

1740

1720
-1

1700

1680

Wavenumber (cm )
Fig. 5. FTIR spectra of pure PVA and selected graft copolymers normalized according to the area of the hydroxyl stretching band (3000e3800 cm1). Fig. 7. Autoscaled C]O stretching band for selected graft copolymers, and second derivative spectrum of VAHB14.

A. Lejardi et al. / Polymer 53 (2012) 50e59

57

In case of the copolymers grafted with glycolic acid, (see the spectrum for VAGA44 in Fig. 3), a single gaussian band is again observed, and similar arguments can be provided. The chain backbone for the glycolic acid residue is identical to that of the lactic acid residue; therefore through bond coupling can also be expected in PGA chains. However, the C]O stretching band of PGA has not been studied in detail, the unperturbed frequency is unknown, and the red shift relative to this value cannot be accurately obtained. Taking for reference the location reported for the C]O stretching band of amorphous PGA [30], 1760 cm1, the band observed for VAGA44 (located at 1737 cm1, see Fig. 7) shows a red shift of about 23 cm1. The magnitude of this result suggests that carbonyl groups are again hydrogen bonded. However, since it has not been calculated using the unperturbed wavenumber for reference, it should not be used to compare the relative strength of the interactions. Distinctively, the spectrum of the C]O stretching band of PVA modied with hydroxybutyric acid, VAHB14, shows two components located at about 1734 and 1716 cm1 according to second derivative techniques. Once again, the spectral features of the carbonyl stretching band of PHB will be recalled to aid in the analysis of this spectral region. The presence of at least three major contributions at about 1748, 1735 and 1723 cm1 is generally accepted in the C]O stretching region of semicrystalline PHB [31]. The band at 1748 cm1 is assigned to free C]O groups in the amorphous phase. The component at 1723 cm1 is assigned to carbonyl groups in the crystalline phase, and its shift to lower wavenumbers is attributed to intermolecular CeHO hydrogen bonding between lateral methyl groups and C]O groups [32]. But the origin of the band at about 1735 cm1, occurring in both PHB and VAHB14, is still unclear. The band located at 1716 cm1 for VAHB14 is red shifted 32 cm1 relative to the free component (expected at about 1748 cm1 as in PHB), therefore it can be attributed to amorphous C]O groups hydrogen bonded to hydroxyl groups. Notice also that conformational sensitivity arising from through bond coupling should not be invoked for VAHB14 or PHB, since neighbouring C]O groups are spaced by at least two CeC bonds. Using 2D IR spectroscopy, Padermshoke et al. assigned the band at 1735 cm1 in PHB to a minor crystalline component with a less ordered structure [33]. On the other hand, aided with QM calculations Sato et al. [34] attributed it to carbonyl groups in the amorphous phase establishing similar intermolecular interactions to those existing in the crystalline phase (a hypothesis that seems quite restrictive for an amorphous phase since it implies proper relative orientations for both interacting groups and chain segments, see ref. 33). However, those hypotheses do not hold in VAHB14. According to DSC results (see following sections), the modied polymer is completely amorphous, therefore the band at 1735 cm1 cannot be attributed to a crystalline component. Regarding the hypothesis of Sato et al, the probability of intermolecular CH3O]C hydrogen bonding contacts in VAHB14 is very small due to the low HB composition (14 mol %). Therefore, an alternative explanation is necessary for this band.

In a parallel study in preparation, we have investigated the formation of intramolecular CeHO hydrogen bonds to explore a possible connection with the band at about 1735 cm1. The relevance of these interactions in the conformation of organic compounds has been already reported in recent papers using both experimental techniques and high level ab initio MO calculations [35]. We have performed high level QM ab initio calculations on the model compound methyl (3R)-3-hydroxybutanoate (MHB), chosen to represent the lateral chain of the VAHB14 copolymers. Particularly, the frequency calculations for two of the geometries, the tggt and the all trans geometry, show C]O stretching bands shifted by about 14 cm1 relative to the free band. In the former conformers, the formation of strong intramolecular CH3O]C hydrogen bonds is detected, and in the latter double 1,5 (gauche effect) and 1,6 CeHO interactions have been detected. These interactions have been depicted in Fig. 8a) and b) for MHB, and similar structures can be proposed to explain the absorption al about 1735 cm1 in PHB. The occurrence of CeHO interactions has been reported in other biodegradable polyesters; for example it is responsible of the stereocomplexation between the L and D isomers of polylactides [27]. 3.3. Thermal behaviour Fig. 9 and Table 1 show respectively the second scan DSC traces for the polymers synthesized in this study and their glass transition temperatures. As can be seen, the Tg of the PVA sample used in this work is 74  C and the melting peak temperature is 222  C, both values within the typical range for this polymer [2]. Chemical modication of PVA introduces important differences on its thermal behaviour. Crystallinity is suppressed for the modied polymers, a result that can be attributed to the presence of bulky side groups impairing crystalline ordering. In addition, the glass transition temperature of the modied polymers decreases compared to that of PVA. To explain this variation, the main factors to consider are the inuence of crystallinity on chain mobility, the free volume effects and the association behaviour. Regarding the inuence of crystallinity on chain mobility, crystalline lamellae is known to constrain the mobility of the surrounding amorphous phase compared to the amorphous phase located at larger distances [36]. The fraction of the amorphous material close to the lamellae is usually denoted RAF (rigid amorphous fraction) and the fraction of the remaining amorphous material with higher mobility is denoted MAF (mobile amorphous fraction). In PVA, being a highly crystalline material, most of the amorphous phase is expected to exist in the form of RAF; however, in the modied PVAs, the RAF decreases to zero due to the amorphous nature of these materials. Since the Tg of the RAF is usually about 10e20  C higher than the Tg of the MAF [36], the change in Tg arising from the suppression of crystallinity can account for most of the reduction observed in the modied polymers. Regarding the free volume effects, the introduction of side groups in a linear polymer is expected to increase

Fig. 8. Geometries for the a) tggt and b) tttt conformers of MHB optimized at the MP2/6e311G(d,p) level of theory.

58

A. Lejardi et al. / Polymer 53 (2012) 50e59

PVA

VAHB14 VAGA44

VAGA24 VALA45 VALA28

50

100

150

200

250

Temperature (C)
Fig. 9. DSC scans for PVA and the modied polymers synthesized in this study.

Table 4 Solubility of PVA and the graft copolymers investigated in this paper. Water PVA VALA28 VALA45 VAGA24 VAGA44 VAHB14 MeOH e e e e EtOH e e e e THF e e e e e e HFIP DMSO DMF e e e e

NMR reveals copolymer compositions in the range 14e45 mol% HA with lateral chain lengths in the range 1.1e1.3. Also, the grafting reaction produced a random copolymer from the backbone point of view. Analysis by FTIR of the C]O stretching band arising from the lateral polyester chains, reveals a completely different prole from that of the long chain homopolymers. In case of the copolymers with GA and LA, this is a result of a single contribution, behaviour that can be attributed to the absence of through bond coupling. In case of the copolymers with HB, two bands can be observed, located at 1735 and 1716 cm1. The former band also occurs in P3HB, and its analysis excludes the hypotheses reported in the literature to explain its origin. From our results it can be attributed to intramolecular CeHO hydrogen bonding, while the band at 1716 cm1 can be attributed to OeHO hydrogen bonds. In case of the VALA and VAGA copolymers carbonyl groups are completely interassociated with hydroxyl groups, but in case of the VAHB copolymers, OeHO hydrogen bonds coexist with CeHO hydrogen bonds. Regarding supramolecular arrangements in PVA, grafting of hydroxy acid repeat units along the PVA backbone suppressed crystallinity and increased free volume, both accounting for the Tg reduction observed. In addition, stronger intermolecular hydrogen bonding interactions with PVP were found in chemically modied PVA systems with regard to unmodied PVA, opening an interesting way in the search of miscibility with polyesters through blending.

EXO>

Highly swollen gel-like structures can be observed.

Acknowledgements The authors are thankful for funds from Basque Government, Department of Education, Universities and Research (GIC10/152-IT334-10 and ITT444-10) and from MICINN (BIO2010-21542-C02-01). A.L thanks the University of the Basque Country for a predoctoral grant.

the free volume and should therefore reduce the glass transition temperature. The last factor to be considered is the interassociation behaviour, and is the most difcult to anticipate. The lateral groups introduced in the linear chain can participate in the formation of hydrogen bonds, and may increase the glass transition temperature of the polymer. In that case, and considering the experimental results, the last two factors discussed here (free volume effects and interassociation behaviour) seem to counteract each other. 3.4. Effect of grafting on solubility Table 4 shows the solubility of the graft-copolymers in different solvents. In general, grafting with LA increases noticeably the solubility of the materials, particularly compared to grafting with GA. In case of grafting with HB, the small grafting degree prevents comparing the effect of this pendant group. As can be seen, all the graft copolymers are completely soluble in HFIP and DMSO. The copolymers also form homogeneous solutions in water, excluding VAGA44, which forms highly swollen gellike structures in aqueous medium. Recalling that PVAegeLA copolymers with lateral chain polymerization degrees above 1.7 have been obtained by Ding et al., reporting non-water-soluble materials in all cases, it seems that the reduced water solubility of the VAGA44 copolymers can be most probably attributed to the larger length of its lateral chains. It seems from these results that the accurate determination of the length of the lateral chains seems mandatory to understand the solubility of the copolymers in aqueous medium. 4. Conclusions PVA was successfully modied with selected hydroxy acids, D,Llactic acid (LA), glycolic acid (GA) and D,L-3-hydroxybutyric acid (HB) yielding graft copolymers. Characterization by 1H and 13C

References
[1] Okano T, editor. Biorelated polymers and gels: controlled release and applications in biomedical engineering. London: Academic Press; 1998. [2] Chiellini E, Corti A, DAntone S, Solaro R. Prog Polym Sci 2003;28:963e1014. [3] Aoi K, Aoi H, Okada M. Macomol Chem Phys 2002;203:1018e28. [4] Li G, Cai Q, Bei J, Wang S. Polym Adv Technol 2003;14:239e44. [5] Cordewener FW, Dijkgraaf LC, Ong JL, Agrawal CM, Zardeneta G, Milam SB, Schmitz JP. J Biomed Mater Res 2000;50:59e66. [6] De Kesel C, Lefvre C, Nagy JB, David C. Polymer 1999;40:1969e78. [7] Shuai X, He Y, Asakawa N, Inoue Y. J Appl Pol Sci 2001;81:762e72. [8] Zhou ZX, Wang XL, Wang YZ, Yang KK, Chen SC, Wu G, Li J. Polym Int 2006;55: 383e90. [9] Sawatari C, Kondo T. Macromolecules 1999;32:1949e55. [10] Onyari JM, Huang SJ. Polym Prepr (Am Chem Soc. Div Polym Chem.) 1996;37: 145e6. [11] Carlotti SJ, Giani-Beaune O, Schue F. J Appl Polym Sci 2001;80:142e7. [12] Ding J, Chen S-C, Wang X-L, Wang Y-Z. Ind Eng Chem Res 2009;48:788e93. [13] Lejardi A, Meaurio E, Monasterio N, Zuza E, Etxeberria A, Sarasua JR. ANTEC Proc; 2009:2957e61. [14] Lejardi A, Meaurio E, Fernandez J, Sarasua JR. Macromolecules 2011;44: 7351e63. [15] Ovenall DW. Macromolecules 1984;17:1458e64. [16] Fernndez MD, Fernndez MJ. J Appl Polym Sci 2008;107:2509e19. [17] Brandolini AJ, Hills DD. NMR spectra of polymers and polymer additives. New York: Marcel Dekker; 2000. [18] Pham Q-T, Petiaud R, Waton H, Llauro MF. Proton and carbon NMR spectra of polymers. 5th ed. Chichester (UK): Wiley; 2001. 2003.142-147. [19] Guerrouani N, Mas A, Schu F. J Appl Polym Sci 2009;113:1188e97. [20] Herbert IR. Statistical analysis of copolymer sequence distribution. In: Ibbett RN, editor. NMR spectroscopy of polymers. London: Blackie Academic & Proffesional; 1993. p. 50e79. Chapter 2 pp. 50e79. [21] Moritori T, Fujiwara Y. Macromolecules 1977;10:532e5. [22] del Ro JC, Rencoret J, Marques G, Gutirrez A, Ibarra D, Santos JI, JimnezBarbero J, Zhang L, Martnez AT. J Agric Food Chem 2008;56:9525e34. [23] Ruiz de Luzuriaga A, Pomposo JA, Grande H. Etxeberria A Macromol Rapid Commun 2009;30:932e5.

A. Lejardi et al. / Polymer 53 (2012) 50e59 [24] Coleman MM, Graf JF, Painter PC. Specic interactions and the miscibility of polymer blends. Lancaster, PA: Technomic Publishing Inc.; 1991. [25] Meaurio E, Zuza E, Lpez-Rodrguez N, Sarasua JR. J Phys Chem B 2006;110: 5790e800. [26] Meaurio E, Martinez de Arenaza I, Lizundia E, Sarasua JR. Macromolecules 2009;42:5717e27. [27] Sarasua JR, Lpez-Rodrguez N, Lpez-Arraiza A, Meaurio E. Macromolecules 2005;38:8362e71. [28] Martinez de Arenaza I, Meaurio E, Coto B, Sarasua JR. Polymer 2010;51: 4431e8. [29] Painter PC, Veytsman B, Kumar S, Shenoy S, Graf JF, Xu Y, Coleman MM. Macromolecules 1997;30:932e42.

59

[30] Kister G, Cassanas G, Vert M. Spectrochim Acta A 1997;53:1399e403. [31] Huang H, Hu Y, Zhang J, Sato H, Zhang H, Noda I, Ozaki Y. J Phys Chem B 2005; 109:19175e83. [32] Sato H, Murakami R, Padermshoke A, Hirose F, Senda K, Noda I, Ozaki Y. Macromolecules 2004;37:7203e13. [33] Padersmshoke A, Sato H, Katsumoto Y, Ekgasit S, Noda I, Ozaki Y. Vib Spectrosc 2004;36:241e9. [34] Sato H, Dybal J, Murakami R, Noda I, Ozaki Y. J Mol Struct 2005;744-747: 35e46. [35] Takahashi O, Kohno Y, Nishio M. Chem Rev 2010;110:6049e76. [36] Zuza E, Ugartemendia JM, Lopez A, Meaurio E, Lejardi A, Sarasua JR. Polymer 2008;49:4427e32.