International Journal of Biological Macromolecules 72 (2015) 855–861

Contents lists available at ScienceDirect

International Journal of Biological Macromolecules

journal homepage: www.elsevier.com/locate/ijbiomac

Research Article

Effective dispersion and crosslinking in PVA/cellulose fiber

biocomposites via solid-state mechanochemistry
Yan Niu, Xiaofang Zhang, Xu He, Jiangqi Zhao, Wei Zhang ∗ , Canhui Lu ∗
State Key Laboratory of Polymer Materials Engineering, Polymer Research Institute at Sichuan University, Chengdu 610065, China

a r t i c l e i n f o a b s t r a c t

Article history: A mechanochemical approach to improve the dispersion and the degree of crosslinking between cel-
Received 17 June 2014 lulose fiber and polymer matrix is presented herein to create high performance poly(vinyl alcohol)
Received in revised form 30 August 2014 (PVA)/cellulose biocomposites in a solvent-free and catalyst-free system. During a pan-milling process,
Accepted 22 September 2014
the hydrogen bonds in both cellulose and PVA were effectively broken up, and the released hydroxyl
Available online 7 October 2014
groups could react with succinic anhydride (SA) to form covalent bonds between the two components.
This stress-induced chemical reaction was verified by fourier transform infrared spectroscopy. The reac-
Keywords:
tion kinetics was discussed according to the conversion rate of SA during the pan-milling process. Soxhlet
Biocomposites
Fiber-reinforced composites
extraction with hot water showed that the crosslinked PVA/cellulose retained more PVA in the composites
Mechanical properties due to the homogeneous and heterogeneous crosslinking. Scanning electron microscope images indicated
Thermal properties the dispersion and interfacial interactions between PVA and cellulose were largely improved. The result-
Interface/interphase ing composites exhibited remarkably enhanced mechanical properties. The tensile strength increased
from 8.8 MPa (without mechanochemical treatment) to 18.2 MPa, and elongation at break increased from
76.8 to 361.7% after the treatment. Their thermal stability was also significantly improved.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction (69 GPa) [4]. Besides, they also offer a number of other advan-
tages over conventional reinforcing materials, such as low cost,
Over the past decade, fiber-reinforced polymer composites worldwide availability, biodegradability, and thermal recyclabil-
are experiencing increased demand in various applications such ity by combustion. Particular attraction is its low density which
as aerospace, vehicles, sport equipment, etc. due to their sig- leads to high specific mechanical properties [4,5]. These unique
nificantly improved strength and toughness over neat polymers. properties have made cellulose an ideal candidate for high per-
Several important factors should be addressed to fully realize the formance biocomposites. Various biodegradable polymers, such
reinforcing potential of the final materials. First, homogeneous fiber as chitosan [6], starch [7], polycaprolactone (PCL) [8,9], poly(3-
dispersion in a polymer matrix is desirable to reduce unreinforced hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) [10,11], polylactic
region and therefore distribute the load evenly in the composite. acid (PLA) [12], poly(furfuryl alcohol) (PFA) [13], and polyvinyl alco-
Second, to achieve efficient stress transfer, the interfacial adhesion hol (PVA) [14–16] have been explored as potential matrices for
between polymer matrix and fiber must be strong enough to pre- this kind of composites. PVA is the most widely produced water-
vent the fiber from being pulled out. The covalent bond formation soluble synthetic polymer in the world. It is also a versatile polymer
between polymer matrix and fiber has been a common method to with broad applications due to its biodegradability, biocompatibil-
improve the composite strength [1–3]. ity, high tensile strength, excellent adhesive properties, chemical
In recent years, the use of natural fibers as reinforcements in resistance and gas barrier properties.
polymers has attracted much attention due to the environmen- Although both cellulose and PVA have plenty of hydroxyl groups
tal concern. Cellulose fibers exhibit high mechanical properties along their molecular chains, direct incorporation of pristine cel-
with the maximum macroscopic Young’s modulus (up to 138 GPa) lulose fibers into PVA matrix cannot produce composites with
even higher than those of aluminum (69 GPa) and glass fiber good mechanical properties [17]. This is due to the fact that most
hydroxyl groups in cellulose and PVA molecules have already
formed either intra- or inter-molecular hydrogen bonds. There-
∗ Corresponding authors. Tel.: +86 28 85460607; fax: +86 28 85402465. fore, it is difficult to form new hydrogen bonds between the
E-mail addresses: [email protected] (W. Zhang), [email protected] (C. Lu). two components by simple mixing, leading to poor interfacial

http://dx.doi.org/10.1016/j.ijbiomac.2014.09.042
0141-8130/© 2014 Elsevier B.V. All rights reserved.
856 Y. Niu et al. / International Journal of Biological Macromolecules 72 (2015) 855–861

adhesion, and deteriorated mechanical performance of the com- 2. Material and Methods
posites. Moreover, the long pristine cellulose fibers can hardly
disperse well within a polymer matrix during conventional com- 2.1. Materials
posite fabrication processes. This situation is even worse for
the melt processing as the high viscosity of polymers will sup- Polyvinyl alcohol 1799, with a degree of polymerization
press free movement of fibers. As a result, the entangled and 1750 ± 50 and a degree of alcoholysis of 99%, was provided by
agglomerated cellulose fibers may generate stress-concentrated SINOPEC Sichuan Vinylon Works, China. Analytical grade suc-
points in the composites and further deteriorate their perfor- cinic anhydride (SA) was purchased from Tianjin Dibo Chemical
mance. Chemical crosslinking in composites is an efficient way Plant, China. Formamide and ethanol were obtained from Chengdu
to achieve desired compatibility [18]. Chabba et al. crosslinked Kelong Chemical Plant, China. Cellulose fibers were hardwood
soy flour/flax yarns composites using glutaraldehyde (GA) as the pulp fibers with 90% cellulose content. Deionized water was used
crosslinking agent to improve the tensile and thermal properties throughout this experiment.
[19]. Li et al. also used GA to prepare crosslinked chitosan/PVA
blend beads with high mechanical strength [20]. Teramoto et al. 2.2. Mechanochemical crosslinking between PVA and cellulose
reported that the miscibility and mechanical properties of pul- fibers
lulan/PVA composites were significantly improved after being
crosslinked with glyoxal [21]. More recently, the crosslinking The crosslinking between PVA and cellulose was achieved using
among cellulose, poly(methyl vinyl ether-co-maleic acid), and SA as a crosslinker through co-milling in the pan-mill. The strong
poly(ethylene glycol) was reported [22]. However, all these reac- compressive and shear forces generated by the pan-mill can induce
tions are solvent based. When organic solvents are used, the mechanochemical esterification reaction among the three ingredi-
emission of toxic solvents may cause serious pollutions to the envi- ents. PVA, cellulose, and SA with a composition of 100:30:13 by
ronment. weight were blended using in a high speed laboratory mixer (FW-
Stress-induced reaction, which is also well-known as 100, Beijing, China) at 26,000 rpm for 5 min. The blend was then fed
mechanochemistry, is an interdisciplinary science based on into a hopper set at the middle of the moving pan. Milled powder
mechanics and chemistry. The pulverization of polymers is a was discharged from the brim of the pans. The discharged powder
typical process of mechanochemistry [23]. Compared to the con- was then collected for the next milling cycle. At predetermined
ventional solvent-based chemical reactions, this approach appears milling cycles (0, 5, 10, 15, 20 cycles, respectively), part of the
to be an energy intensive and environmentally benign technique. milled power was collected for analysis. The milling process was
As a heterogeneous solid-phase reaction, several processes may maintained at ambient temperature with a water cooling system.
occur during the mechanochemical synthesis: dispersion of the The rotating speed of pan-mill was set at 30 rpm under a specific
components, generation of the contacts between them, the mass pressure.
transfer at the contact zones, and chemical reaction. All these pro-
cesses are induced only by the mechanical impact on the mixture of 2.3. Melt processing of crosslinked PVA/cellulose fibers composites
solid reagents. In fact, mechanochemistry has been widely applied
in metal alloying and for production of inorganic salts. However, The obtained pan-milled powder was ethanol extracted to
very few organic reactions have been performed via this process, remove residual SA prior to being blended with plasticizers.
especially in chemical modification of polymers. Furthermore, The plasticizers were prepared by mixing formamide and water
the existing research on polymer pulverization was concentrated in equal weight. All ingredients (composite powders: plasticiz-
either on improving the dispersion of particles in polymer matrix ers = 1:1, by weight) were blended in a laboratory mixer (FW-100,
[24–27], or solely on the synthetic aspects [28–30]. As far as we Beijing, China) at 26,000 rpm for 3 min and sealed for one week
know, taking advantages of the multi-functional mechanochem- before thermal compounding. The blend was then melt mixed
istry to simultaneously realize effective dispersion and sufficient using a Brabender Plasticorder PL2000 mixing machine (Duisburg,
chemical crosslinking for fabricating high performance composites Germany) at 150 ◦ C and 30 rpm for 10 min. The obtained paste was
has rarely been reported before. kept at ambient condition for 24 h. After that, the paste was filled in
The objective of this investigation is to achieve good disper- a metal mold (150 mm in diameter and 1 mm in thickness) under
sion of cellulose fibers in PVA matrix while having their interface a pressure of 10 MPa at 175 ◦ C for 15 min. For comparison, samples
covalently crosslinked via solid-state mechanochemistry. A self- without mechanochemical treatment were prepared in parallel fol-
designed pan-mill was used in the present study. The details of lowing the same processing procedures.
the pan-mill equipment can be found in our previous publica-
tion [31]. Early studies showed that this equipment was much 2.4. Fourier transform infrared spectroscopy (FTIR) analysis
more efficient for mechano-activation of cellulose compared to
conventional ball-mill, and acetylated cellulose was successfully FTIR spectra were studied with a Nicolet 560 spectrometer
synthesized by co-milling of cellulose and acetic anhydride in (Nicolet Co., USA), taking 32 scans for each sample with a resolu-
solvent- and catalyst-free conditions [31–33]. Succinic anhydride tion of 2 cm−1 , ranging from 400 to 4000 cm−1 . Approximately 2 mg
(SA), a non-toxic solid reactant which is quite stable at room tem- of pristine cellulose powder (obtained by milling cellulose fibers
perature, was utilized as a crosslinking agent to form covalent with a mortar), neat PVA and crosslinked PVA/cellulose composite
bonds between PVA and cellulose through ring-opening esterifi- powder (Soxhlet extracted with ethanol for 48 h) were mixed with
cation. After that, the obtained powders were ethanol extracted to 200–300 mg KBr, respectively. Pellets of 1 cm in diameter were then
eliminate the residue SA, followed by melt mixing and compression obtained in a standard mold under a pressure of 4 MPa. Both test-
molding with plasticizers. The crosslinking was verified by fourier ing samples and KBr were dried before pellet preparation and were
transform infrared spectra (FTIR) and weight loss against water subjected to FTIR analysis immediately.
extraction. The dispersion of cellulose fibers in polymer matrix was
detected using scanning electron microscope (SEM). The mechan- 2.5. Evaluation of reaction degree of SA during pan-milling
ical properties and thermal stability of these biocomposites were
investigated and compared with those without mechanochemical The degree of mechanochemical reaction was evaluated by SA
treatment. conversion after pan-milling. Unreacted SA can be removed easily
 Y. Niu et al. / International Journal of Biological Macromolecules 72 (2015) 855–861 857

during Soxhlet extraction with ethanol. Therefore, the SA conver-

sion is expressed in Eq. (1):
 (Wa − Wb )

Conversion rate (%) = 1 − × 100%, (1)
˛Wa
where Wa and Wb represent the dried weight of powder samples
before and after Soxhlet extraction, respectively. ␣ stands for the
SA fraction in the composition.

2.6. Weight loss against water extraction

PVA can be easily dissolved in hot water, whereas crosslinked

PVA only swells under the same condition. Water extraction exper-
iment on samples (after removal of residue SA) was performed to
verify the crosslinking structure in the composite. The weight loss
was calculated by Eq. (2):
 (W − W ) 
c d
Weight loss (%) = × 100%, (2)
Wc
Fig. 1. FTIR analysis of cellulose, PVA and PVA/cellulose/SA co-milled for 20 cycles.
where Wc and Wd represent the dried weight of the powder
samples before and after Soxhlet extraction, respectively. findings indicate mechanochemical esterification reaction among
PVA, cellulose and SA indeed happened during the pan-milling
2.7. Tensile tests process. The absence of peaks at 1850 and 1780 cm−1 of this spec-
trum confirms that the products were free of unreacted SA after
The compression molded sheets were equilibrated in ambi- purification. Similar to those multi-component crosslinking reac-
ent environment for one week at a relative humidity of 75%. The tions in solvent [17,21], the homogeneous chemical crosslinking
dumbbell-shaped specimens were then punched out from the between two hydroxyl groups from either PVA or cellulose may also
sheets by a tensile specimen cutter according to ASTM D 15–54 T. take place during the solid-state mechanochemical process (see
The stress-strain properties were measured by an Instron 5567 Uni- Fig. 2). It is important to note that the C O absorption peak may not
versal Testing Machine (Instron Co., USA) at a crosshead speed of be responsible merely for ester linkage. As illustrated in Fig. 2, the
50 mm/min. At least five measurements for each composite were monoester derivative with the formed succinyl groups contributes
performed. a portion of the peak. This is because that the monoester derivative
is produced at first when SA reacts with hydroxyl-bearing materi-
2.8. Morphology observation als. The resultant carboxyl groups are less reactive as compared to
the corresponding acid anhydride. It is difficult for those carboxyl
Scanning electron microscope (SEM, model JEOL JSM-5600, groups to be completely converted into ester bonds. This claim was
Japan) was used to observe the liquid nitrogen fractured surface also supported by Yoshimura et al. who had studied the cotton
of composites. Prior to SEM evaluation, the samples were sputter- cellulose/SA crosslinking system [37].
coated with gold to avoid charging during tests.
3.2. Weight loss against water extraction
2.9. Thermogravimetric analysis
PVA is a water-soluble polymer, and its solubility depends on
Thermal gravimetric analysis (TGA) was conducted on TA-2000 the degree of polymerization (DP) as well as the degree of alcohol-
analyzer (TA Instruments, USA) from room temperature to 600 ◦ C ysis (DA). The PVA 1799 used in this work is hard to be dissolved in
at a heating rate of 20 ◦ C/min. The measurements were operated water except for under heating over 80 ◦ C, owing to its high DP and
under a nitrogen purge (25 mL/min). All samples were dried in a DA. During the Soxhlet extraction process, the temperature of sol-
vacuum oven at 100 ◦ C for 2 h before the testing. About 5 mg sam- vent water was higher than 90 ◦ C. Therefore, it was easy to dissolve
ples were used in each test. and remove the uncrosslinked PVA from the composite, leaving
the crosslinked portion swelled. Fig. 3 shows the weight loss of the
3. Results and Discussion samples during the Soxhlet extraction with water. The control sam-
ple without mechanochemical treatment exhibits an almost linear
3.1. FTIR analysis weight loss as a function of extraction time. This trend is signifi-
cantly reduced for the PVA/cellulose/SA samples. The weight loss
It has been reported that acetic anhydride itself is difficult to of control sample is 65.2% after 300 min extraction, which means
react with cellulose before activation which weakens the inter- 84.8% of total PVA has been extracted into water, considering PVA
and intra-molecular hydrogen bonds [34]. Traditional ways for the fraction in the composite. On the contrast, the co-milled sample
esterification reactions between cellulose and anhydride always with SA shows 30.1% weight loss, thereby only 43.1% of total PVA
require the use of a large amount of solvents in the presence of cata- was extracted. These results strongly confirmed the crosslinking
lysts at elevated temperature [35,36]. In this study, the crosslinking structure of the material which has prevented extraction of PVA
among PVA, cellulose and SA was realized at ambient temperature into water.
via solvent- and catalyst-free mechanochemical reaction. FTIR was
used to verify this chemical reaction. Fig. 1 compares FTIR spectra of 3.3. Reaction kinetics
cellulose, PVA and co-milled PVA/cellulose/SA samples. The spec-
trum of PVA/cellulose/SA sample exhibits a less intensive hydroxyl PVA and cellulose cannot be dissolved in ethanol which is a
stretching adsorption peak at 3400 cm−1 . In addition, a new peak good solvent for SA. The unreacted SA can be thoroughly removed
appears at 1725 cm−1 assigned to C O stretching vibration. These from the composite by Soxhlet extraction using ethanol as a solvent
858 Y. Niu et al. / International Journal of Biological Macromolecules 72 (2015) 855–861

Fig. 2. Scheme of the mechanism of the mechanochemical esterification between PVA, cellulose and SA.

while retaining the PVA, cellulose and reacted SA in the composi- 3.4. Mechanical properties of melt processed composites.
tion. Therefore, it is possible to evaluate the conversion rate of SA
from a comparison of the weight loss after Soxhlet extraction with A multi-component plasticizer comprised of formamide and
initial SA weight. Fig. 4 illustrates the conversion rate of SA dur- water was adopted for the melt processing of PVA composites [38].
ing the pan-milling process. The conversion rate increased quickly It is well-known that highly crosslinked polymer cannot be melt
at the early milling stage, followed by a relatively slow increase. processed. However, as indicated above, the crosslinking density
The conversion rate of SA reached 61.6% after 20 cycles of co- of PVA was not too high since 43.1% PVA could be removed from
milling. the composites after 300 min of water extraction. Therefore, this
The solid-state esterification among PVA, cellulose, and SA composite was still thermoplastically processable in the presence
is based on the mechanism of mechanochemical activation of of plasticizers.
hydroxyl groups of cellulose and PVA. When cellulose and PVA are All samples, including PVA, PVA/cellulose, and crosslinked
subjected to mechanical forces, their massive hydrogen bond net- PVA/cellulose composites were fabricated by melt mixing followed
work can be destroyed, releasing highly reactive hydroxyl groups by compression molding. Their tensile strength (TS), elongation at
on the fresh particle surface which are capable to react with SA break (EL), and modulus were recorded from tensile testing and
even in solvent- and catalyst-free conditions. Moreover, because all summarized in Table 1. It is evident that the direct incorporation
the reactants are in solid form, the esterification thus mainly hap- of cellulose fibers into PVA cannot yield high strength composites.
pens at the particle surface. In this study, PVA and cellulose after Their TS and EL were significantly lower as compared to those of
pan-milling are several to several tens of microns in size [31]. There- PVA, indicating a weak load transfer from the matrix to fibers. How-
fore, the crosslinking is considered to take place at a macroscopic ever, their moduli were the highest due to the hardening effect
phase. when the high modulus cellulose fibers were added into soft chain

Fig. 3. Weight loss of powder samples of (A) PVA/cellulose (B) Crosslinked

PVA/cellulose during Soxhlet extraction using water as the solvent. Fig. 4. Conversion rate of SA as a function of pan-milling cycles.
 Y. Niu et al. / International Journal of Biological Macromolecules 72 (2015) 855–861 859

Table 1
Mechanical properties of PVA and PVA/cellulose composites.

Sample Strength Elongation at Modulus

(MPa) break (%) (MPa)

PVA 14.0 ± 0.5 339.3 ± 10.0 44.7 ± 3.3

PVA/cellulose 8.8 ± 0.6 76.8 ± 7.2 96.3 ± 9.6
Crosslinked PVA/cellulose 18.2 ± 0.8 361.7 ± 22.1 65.7 ± 6.0

polymers such as PVA. The mechanical properties of the compos-

ites were drastically improved after mechanochemical treatment.
The resultant crosslinked PVA/cellulose biocomposites exhibited
TS, EL, and modulus of 18.2 MPa, 361.7% and 65.7 MPa, respec-
tively, demonstrating a remarkably improved performance. It is
interesting to notice the crosslinked composites showed a slightly
increased EL as compared to PVA. This phenomenon seems to be
contradictive to other composites with strengthened interfacial
adhesion as they generally exhibit decreased EL. A possible rea-
son lies on the fact that the mechaochemical reaction also brings
some highly hydrophilic carboxyl groups into the system, retain-
ing more water in the composite. Consequently, a more plasticizing
effect was exhibited, leading to a longer strain. However, this effect
may also reduce the materials’ moduli. Furthermore, the shortened
and less entangled fibers after pan-milling may be responsible for
the decreased moduli as well.

3.5. Morphological observation on the fractured surface of

composites

Fig. 5 gives SEM images of liquid nitrogen fractured surface of

PVA composites. The PVA component exhibited relatively smooth
texture. This was because part of PVA, which had not been
crosslinked, could be melted during compression molding. How-
ever, those regions containing cellulose fibers (identified using
red circles on the images) were much rougher because cellulose
could not be melted. The fiber dispersion and interfacial adhesion
between PVA and cellulose were largely improved by mecha-
nochemical treatment. As shown in Fig. 5(A) for PVA/cellulose
composite made by simple blending, the coarse fractured sur-
face with twisted fiber agglomerates was observed, suggesting
the inferior fibers dispersion and poor adhesion between two
phases. On the contrast, no fiber agglomeration was seen for the
mechanochemically crosslinked composite owing to the excel-
lent mixing and dispersion from pan-milling. The fine cellulose
particles were evenly dispersed in the matrix from the global
view in Fig. 5(B). The cellulose fibers were firmly embed-
Fig. 5. SEM images of liquid nitrogen fractured surface of (A) PVA/cellulose com-
ded in PVA matrix. As indicated in Fig. 5(C), the broken fiber posite by simple blending (B, C) mechanochemically crosslinked PVA/cellulose
cross-section (white dot) together with quite a few PVA fibrils stuck composite.
to fiber surface in response to external forces strongly demonstrate
the favorable interfacial adhesion between PVA and cellulose due to
the covalent crosslinking. The imposed load can be efficiently trans- 3.6. Thermal stability measurements
ferred from matrix to the reinforcing cellulose fibers, and therefore
improved performance could be achieved. The thermal stability of pristine PVA, PVA/cellulose and
For fiber-reinforced composite materials, it is generally crosslinked PVA/cellulose composite were compared through
accepted that fibers with a relatively high aspect ratio are beneficial thermogravimetric analysis (TGA). Both TGA and derivative ther-
for the composites’ physical properties. However, the dispersion of mogravimetric (DTG) curves are provided in Fig. 6. As shown in
fibers in matrix and the adhesion between them are also crucial fac- the figure, PVA experiences two major weight losses in the tem-
tors for an efficient stress transfer. During the mechanochemical perature range of 200-500 ◦ C. The first one begins at 241 ◦ C with
process, the aspect ratio of cellulose fibers is reduced due to the a major degradation peak at 263 ◦ C according DTG curves, which
strong shearing forces [31]. Besides, the molecular weight of the mainly involves dehydration and formation of some volatile prod-
milled matter can be degraded, which worsens its mechanical prop- ucts. The other one starts at 418 ◦ C, which involves the degradation
erties [39,40]. However, our results did reveal positive effects of of polyene residue to yield carbon and hydrocarbon [41]. Both
mechanochemical treatment. The excellent fiber dispersion and PVA/cellulose and crosslinked PVA/cellulose composites show a
the crosslinking structure could compensate the negative influence relatively slow weight loss before the major degradation, owing to
from milling and yield products with improved performance. the evaporation of plasticizer in the composite. A higher weight loss
860 Y. Niu et al. / International Journal of Biological Macromolecules 72 (2015) 855–861

strong bonding, the mechanochemical crosslinking treatment of

PVA/cellulose/SA under solvent-free and catalyst-free conditions
was reported herein. As expected, the chemical crosslinking with
SA happened when the hydroxyl groups in cellulose and PVA
were mechanochemically activated by the compressive and shear
forces. Meanwhile, the pulverization process could distribute cel-
lulose fibers homogeneously in the polymer matrix. The cutting
of cellulose fibers during milling eliminated entanglement and
agglomeration of long fibers. In addition, it produced a large spe-
cific surface area which was favorable for a stronger interfacial
interaction. The obtained biocomposites exhibited highly enhanced
mechanical properties together with significantly improved ther-
mal stability.

Acknowledgements

The authors would like to thank National Natural Science Foun-

dation of China (51303112, 51473100 and 51433006) and Young
Scholar Foundation of Sichuan University (2012SCU11074) for
financial support of this work.

References

[1] A.K. Bledzki, J. Gassan, Prog. Polym. Sci. 24 (1999) 221–274.

[2] L. Yu, K. Dean, L. Lin, Prog. Polym. Sci. 31 (2006) 576–602.
[3] J. Gao, M.E. Itkis, A. Yu, E. Bekyarova, B. Zhao, R.C. Haddon, J. Am. Chem. Soc.
127 (2005) 3847–3854.
[4] S.J. Eichhorn, A. Dufresne, M. Aranguren, N.E. Marcovich, J.R. Capadona, S.J.
Rowan, C. Weder, W. Thielemans, M. Roman, S. Renneckar, J. Mater. Sci. 45
(2010) 1–33.
[5] I. Siró, D. Plackett, Cellulose 17 (2010) 459–494.
[6] S.C. Angadi, L.S. Manjeshwar, T.M. Aminabhavi, Int. J. Biol. Macromol. 51 (2012)
45–55.
[7] M.E. Vallejosa, A.A.S. Curvelo, E.M. Teixeira, F.M. Mendes, A.J.F. Carvalho, F.E.
Felissia, M.C. Area, Ind. Crop. Prod. 33 (2011) 739–746.
[8] Q. Zhao, S. Wang, X. Cheng, R.C. Yam, D. Kong, R.K. Li, Biomacromolecules 11
(2010) 1364–1369.
[9] H. Lönnberg, K. Larsson, T. Lindström, A. Hult, E. Malmström, ACS. Appl. Mater.
Interfaces 3 (2011) 1426–1433.
[10] R. Bhardwaj, A.K. Mohanty, L.T. Drzal, F. Pourboghrat, M. Misra, Biomacro-
molecules 7 (2006) 2044–2051.
[11] E. Ten, J. Turtle, D. Bahr, L. Jiang, M. Wolcott, Polymer 51 (2010) 2652–2660.
[12] L. Suryanegara, A.N. Nakagaito, H. Yano, Cellulose 17 (2010) 771–778.
Fig. 6. TGA and DTG curves for pristine PVA, PVA/cellulose and crosslinked [13] H. Dekaa, M. Misraa, A. Mohanty, Ind. Crop. Prod. 41 (2013) 94–101.
[14] W. Zhang, X. Yang, C. Li, M. Liang, C. Lu, Y. Deng, Carbohydr. Polym. 83 (2011)
PVA/cellulose.
257–263.
[15] K. Das, D. Ray, N.R. Bandyopadhyay, S. Sahoo, A.K. Mohanty, M. Misra, Compos.
Part B: Eng. 42 (2011) 376–381.
[16] A. Abdulkhania, E.H. Marvasta, A. Ashorib, Y. Hamzeha, A.N. Karimia, Int. J. Biol.
is recorded for the crosslinked sample in this stage. As discussed Macromol. 62 (2013) 379–386.
above, the presence of carboxyl groups can retain more water in [17] S.K. Ozaki, M.B.B. Monterio, H. Yano, Y. Imamura, M.F. Souza, Polym. Degrad.
composition, which corresponds to the greater weight loss. The Stabil. 87 (2005) 293–299.
[18] Z. Wang, E. Wang, S. Zhang, Z. Wang, Y. Ren, Ind. Crop. Prod. 29 (2009) 133–138.
onset temperature for major degradation of PVA/cellulose is 291 ◦ C [19] S. Chabba, G.F. Matthews, A.N. Netravali, Green Chem. 7 (2005) 576–581.
with the major degradation peak at 329 ◦ C, whereas these values [20] M. Li, S. Cheng, H. Yan, Green Chem. 9 (2007) 894–898.
are 307 and 340 ◦ C for the crosslinked PVA/cellulose composite, [21] N. Teramoto, M. Saitoh, J. Kuroiwa, M. Shibata, R. Yosomiya, J. Appl. Polym. Sci.
82 (2001) 2273–2280.
respectively. The results indicate that the incorporation of cellu- [22] L. Goetz, A. Mathew, K. Oksman, P. Gatenholm, A.J. Ragauskas, Carbohydr.
lose into PVA could remarkably enhance the thermal stability of Polym. 75 (2009) 85–89.
final composites. This is in accordance with previous report on cel- [23] X. Xu, Q. Wang, X. Kong, X. Zhang, J. Huang, Plast. Rubber Compos. Process Appl.
25 (1996) 152–158.
lulose/PVA composite films [41]. In addition, the results also show [24] A.P. Smith, H. Ade, S.D. Smith, C.C. Koch, R.J. Spontak, Macromolecules 34 (2001)
that chemical crosslinking in PVA/cellulose composites can further 1536–1538.
improve their thermal stability. Such phenomena have been discov- [25] K. Wakabayashi, C. Pierre, D.A. Dikin, R.S. Ruoff, T. Ramanathan, L.C. Brinson,
J.M. Torkelson, Macromolecules 41 (2008) 1905–1908.
ered in crosslinked PVA/TiO2 composites too [42]. It is worth noting
[26] J. González-Benito, G. González-Gaitano, Macromolecules 41 (2008)
that the crosslinked composite yielded significantly higher amount 4777–4785.
of carbon residues after testing. Previously, other crosslinked PVA [27] J. Masuda, J.M. Torkelson, Macromolecules 41 (2008) 5974–5977.
[28] M. Kuzuya, S. Kondo, A. Noguchi, Macromolecules 24 (1991) 4047–4053.
composites were also reported to have more carbon residues than
[29] J. Huang, J.A. Moore, J.H. Acquaye, R.B. Kaner, Macromolecules 38 (2005)
the uncrosslinked ones [42,43]. 317–321.
[30] A. Voronov, A. Kohut, A. Synytska, W. Peukert, J. Appl. Polym. Sci. 104 (2007)
3708–3714.
4. Conclusions [31] W. Zhang, M. Liang, C. Lu, Cellulose 14 (2007) 447–456.
[32] W. Zhang, X. Zhang, M. Liang, C. Lu, Compos. Sci. Technol. 68 (2008) 2479–2484.
[33] W. Zhang, C. Li, M. Liang, Y. Geng, C. Lu, J. Hazard. Mater. 181 (2010)
The potential property improvement of any composites depends 468–473.
on the degree of dispersion and the bonding ability between [34] A. Biswas, R.L. Shogren, J.L. Willett, Biomacromolecules 6 (2005) 1843–1845.
matrix and reinforcing phase. To realize good dispersion and [35] F. Gellerstedt, P. Gatenholm, Cellulose 6 (1999) 103–121.
 Y. Niu et al. / International Journal of Biological Macromolecules 72 (2015) 855–861 861

[36] E. Zini, M. Scandola, Biomacromolecules 4 (2003) 821–827. [41] J. Lu, T. Wang, L.T. Drzal, Compos. Part A: Appl. Sci. Manufacturing 39 (2008)
[37] T. Yoshimura, K. Matsuo, R. Fujioka, J. Appl. Polym. Sci. 99 (2006) 3251–3256. 738–746.
[38] N. Chen, L. Li, Q. Wang, Plast. Rubber Compos. 36 (2007) 283–290. [42] C.C. Yang, J.Membrane Sci. 288 (2007) 51–60.
[39] R. Chen, C. Yi, H. Wu, S. Guo, Carbohydr. Polym. 81 (2010) 188–195. [43] K. Das, D. Ray, N.R. Bandyopadhyay, A. Gupta, S. Sengupta, S. Sahoo, A. Mohanty,
[40] P.J. Flory, J. Am. Chem. Soc. 67 (1945) 2048–2050. M. Misra, Ind. Eng. Chem. Res. 49 (2010) 2176–2185.