c3gc41759j
TRANSCRIPT
Green Chemistry
PAPER
Cite this: Green Chem., 2014, 16,1789
Received 27th August 2013,Accepted 16th October 2013
DOI: 10.1039/c3gc41759j
www.rsc.org/greenchem
Carbohydrate-based PBT copolyesters from acyclic diol derived from naturally occurring tartaricacid: a comparative study regarding meltpolycondensation and solid-state modification†
Cristina Lavilla,a Erik Gubbels,b Abdelilah Alla,a Antxon Martínez de Ilarduya,a
Bart A. J. Noordover,*b Cor E. Koningb,c and Sebastián Muñoz-Guerraa
2,3-O-Methylene-L-threitol (Thx) is a cyclic carbohydrate-based diol prepared by acetalization and sub-
sequent reduction of the naturally occurring tartaric acid. The structure of Thx consists of a 1,3-dioxolane
ring with two attached primary hydroxyl groups. Two series of partially bio-based poly(butylene tere-
phthalate) (PBT) copolyesters were prepared using Thx as a comonomer by melt polycondensation (MP)
and solid-state modification (SSM). Fully random copolyesters were obtained after MP using mixtures of
Thx and 1,4-butanediol in combination with dimethyl terephthalate. Copolyesters with a unique block-
like chemical microstructure were prepared by the incorporation of Thx into the amorphous phase of PBT
by SSM. The partial replacement of the 1,4-butanediol units by Thx resulted in satisfactory thermal stabi-
lities and gave rise to an increase of the Tg values, this effect was comparable for copolyesters prepared
by MP and SSM. The partially bio-based materials prepared by SSM displayed higher melting points and
easier crystallization from the melt, due to the presence of long PBT sequences in the backbone of the
copolyester. The incorporation of Thx in the copolyester backbone enhanced the hydrolytic degradation
of the materials with respect to the degradation of pure PBT.
1. Introduction
Nature produces approx. 140 × 109 tons of carbohydrates fromcarbon dioxide and water annually, making these compounds themost abundantly available organic materials on earth. Humanityonly uses approx. 4% of this feedstock for food and non-foodpurposes.1 Therefore this class of compounds is an extraordi-nary source of chemicals, capable of providing a wide varietyof building blocks for polycondensates.2–4 Thus, the develop-ment of carbohydrate-based polycondensates has the potentialto significantly reduce the amount of petroleum consumed forpolymer production. However, there are only a few examples of
commercially available carbohydrate-derived polymers, whichis mainly due to the relatively high price of these materialscompared to their petroleum-based counterparts. Anotherlimitation of carbohydrates as building blocks in the synthesisof linear polycondensates is their inherent multi-functionality.Although some linear polycondensates have been synthesizedusing carbohydrate-based monomers bearing pendanthydroxyl groups,5,6 the most commonly used strategy is block-ing them with stable protecting groups, and thus retainingonly two reactive functions to carry out linearpolycondensation.7,8
Among carbohydrates, tartaric acid (2,3-dihydroxy-succinicacid) stands out by its easy accessibility and relative simplicityin terms of its molecular structure. This aldaric acid is foundin many plants and fermented grape juice.9 From tartaric acid,various monomers have been prepared and explored as carbo-hydrate-based building blocks in the synthesis of polyconden-sates, usually with the pendant hydroxyl groups protected inseveral forms.10–14 Aromatic polyesters are used in a wide arrayof applications due to their high performance as thermoplasticmaterials with good thermal and mechanical properties.15
Moreover, their high resistance towards hydrolytic or enzy-matic degradation is an added advantage when these polymersare used in durable applications. On the other hand, this
†Electronic supplementary information (ESI) available: Detailed experimentalprocedure. TGA traces; WAXD profiles, Bragg spacings and crystallinity indexes;DSC traces for SSMPB82Thx18T randomization experiments; isothermal crystalliza-tion; SEM micrographs and 1H NMR spectra of the residue of PBT,MPPB82Thx18T,
SSMPB82Thx18T and PThxT upon incubation at pH 2.0 at 80 °C;1H NMR spectra of 2,3-O-methylene-L-threitol after incubation at pH 2.0, 7.4 and10.5 at 80 °C. See DOI: 10.1039/c3gc41759j
aDepartament d’Enginyeria Química, Universitat Politècnica de Catalunya, ETSEIB,
Diagonal 647, 08028 Barcelona, SpainbLaboratory of Polymer Materials, Eindhoven University of Technology, Den Dolech 2,
P.O. Box 513, 5600 MB Eindhoven, The Netherlands. E-mail: [email protected] Coating Resins, Ceintuurbaan 5, Zwolle, The Netherlands
This journal is © The Royal Society of Chemistry 2014 Green Chem., 2014, 16, 1789–1798 | 1789
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article OnlineView Journal | View Issue
durability is considered a negative aspect if such polyestersend up in the natural environment. Therefore, it is desirable insome applications to increase the degradability of thesematerials, especially for products with short life cycles. Theproperties of such aromatic polyesters can be adjusted bycopolymerization, preferably with bio-based comonomers, inorder to make these materials more environmentally friendlyand more sustainable.16 However, the incorporation of bio-based acyclic monomers usually results in a lowered glass-tran-sition temperature and reduced stiffness.11,16,17 For example,2,3-O-methyl-threitol, an acyclic diol with its secondaryhydroxyl groups protected as methoxy ethers, has beenemployed in the preparation of aromatic copolyesters.11 Onthe other hand, carbohydrate-derived monomers with a cyclicstructure distinguish themselves by providing polycondensateswith improved properties, especially those related to polymerchain stiffness.18–22 Two cyclic acetalized carbohydrate-baseddiols derived from tartaric acid, i.e. 2,3-O-isopropylidene-L-threitol and 2,3-O-methylene-L-threitol, with the pendant func-tional groups protected as isopropylidene acetal and methy-lene acetal, respectively, have been recently employed in thepreparation of polyurethanes.12
In this work, the cyclic diol 2,3-O-methylene-L-threitol (Thx)is employed as a building block for the preparation of poly-(butylene terephthalate) (PBT) copolyesters. Whereas isopropy-lidene acetals are known to be unstable and easily removed inacidic media,12,23,24 the protection by methylene acetal groupsallows for the preparation of polycondensates with a satisfac-tory stability during processing and handling, since themethylene acetal rings are much more resistant toopening.25,26 Two different techniques, i.e. melt polycondensa-tion (MP) and solid-state modification (SSM), will be used toprepare the novel bio-based copolyesters from the aforemen-tioned Thx. The widely known MP technique consists of react-ing the monomers Thx, 1,4-butanediol, and dimethylterephthalate in the melt, viz. in two successive stages invol-ving transesterification and polycondensation reactions. TheSSM technique27 is based on the same principles as the com-monly used solid-state polycondensation (SSP), which isusually employed to increase the molecular weight of semi-crystalline polyesters and polyamides by transesterificationand transamidation reactions, respectively.28,29 Both SSM andSSP techniques are carried out at a reaction temperature just(20–30 °C) below the melting temperature of the crystallinephase of the polymer, and they take advantage of the mole-cular mobility in the amorphous phase of the polymer rela-tively far above Tg. During the SSM process, the monomer Thxis inserted exclusively into the mobile amorphous fraction(MAF) of poly(butylene terephthalate); mobility restrictionsprevent the crystalline phase and the rigid amorphous fraction(RAF) from taking part in transesterification reactions.30–32
The synthesis, chemical microstructure and thermal propertiesof the resulting MP- and SSM-prepared bio-based copolyesterswill be studied and discussed in detail. Moreover, a hydrolyticdegradation study will be carried out to evaluate the influenceof the Thx units on the degradation of the copolyester, and to
ascertain the stability of the acetal group forming part of thecarbohydrate-based residues.
2. Experimental section
The cyclic diol 2,3-O-methylene-L-threitol (Thx) was producedfrom commercially available dimethyl-L-tartrate. Syntheticdetails are provided in the ESI† file linked to this paper.
The two series of copolyesters, MPPBxThxyT andSSMPBxThxyT, were prepared by melt polycondensation (MP) orsolid-state modification (SSM), respectively. x and y indicatethe mole percentages (mol%) of 1,4-butanediol and Thx,respectively, in the resulting copolyester.
MPPBxThxyT were obtained from a mixture of 1,4-butanediol(1,4BD), the cyclic diol Thx and dimethyl terephthalate (DMT)with a predetermined composition by a two-stage melt poly-condensation process, using DBTO as a catalyst (0.6% molarwith respect to monomers). The synthetic procedure andexperimental conditions for the preparation of MPPBxThxyTcopolyesters as well as the parent PThxT homopolyester aredetailed in the ESI.†
SSMPBxThxyT copolyesters were prepared by solid-statemodification of PBT with the cyclic diol Thx using DBTO as acatalyst (0.36% molar with respect to PBT). The physical mix-tures of PBT (Mn = 23.3 kg mol−1 and Mw = 47.1 kg mol−1,determined by SEC), Thx and DBTO having predeterminedcompositions were prepared from solution, using a commonsolvent approach as described elsewhere,32 and were intro-duced into the reactor to perform the SSM reactions at 160 °Cunder controlled nitrogen flow. The detailed procedure ofpreparation of SSMPBxThxyT copolyesters can be found in theESI† file linked to this paper.
The chemical constitution, composition and microstructureof polyesters and copolyesters were determined by nuclear mag-netic resonance (NMR) spectroscopy and the molecular weightswere measured by size exclusion chromatography (SEC). Thermalproperties were determined by differential scanning calori-metry (DSC) and thermogravimetric analysis (TGA). Wide-angle X-ray diffraction (WAXD) measurements were performedto determine the crystallinity of the copolyesters. The hydro-lytic degradation of both series of copolyesters was evaluatedby incubating the materials in a citric acid buffer solution at80 °C. The molecular characteristics of these materials werestudied using SEC, NMR spectroscopy and scanning electronmicroscopy (SEM). Details on the aforementioned analyticaltechniques are provided in the ESI.†
3. Results and discussion3.1. Polymer synthesis
The bio-based compound 2,3-O-methylene-L-threitol (Thx) wasprepared following the route depicted in Scheme 1. Thependant hydroxyl groups from naturally occurring L-tartaricacid were protected by internal acetalization leading to the
Paper Green Chemistry
1790 | Green Chem., 2014, 16, 1789–1798 This journal is © The Royal Society of Chemistry 2014
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online
formation of a 1,3-dioxolane ring. The resulting product,dimethyl 2,3-O-methylene-L-tartrate, was subsequently reducedto obtain the desired Thx diol monomer. This cyclic diolpossesses a 2-fold axis of symmetry, and will therefore producestereo-regular polymer chains, and its two primary hydroxylgroups are expected to display the same reactivities towards apolycondensation reaction.
Thx was used as a diol comonomer in the synthesis of thebio-based copolyesters by two different approaches, viz. meltpolycondensation (MP) and solid-state modification (SSM), asdepicted in Scheme 2.
MPPBxThxyT copolyesters were obtained by reactingdimethyl terephthalate (DMT) with mixtures with varying com-positions of Thx and 1,4-butanediol (1,4BD) in the melt. Trans-esterification reactions were initiated at 160 °C in order toprevent volatilization of the diols. As the reaction progressed,the temperature was progressively increased to avoid crystalli-zation of oligomers. The subsequent polycondensation reac-tions were performed under reduced pressure at temperaturesin the range of 220–250 °C. Lower temperatures and longerreaction times were used for copolyesters with higher Thx con-tents, to avoid decomposition of this thermally sensitive com-pound. Dibutyl tin oxide (DBTO) was the catalyst of choice asreplacement of the commonly used titanium(IV) tetrabutoxide(TBT). Results obtained from the synthesis of dimethyl 2,3:4,5-di-O-methylene-galactarate-based aliphatic polyesters by meltpolycondensation demonstrated the higher activity of DBTO asa catalyst compared to TBT.33 MPPBxThxyT copolyesters wereobtained with weight average molecular weights between 39
and 43 kg mol−1 and polydispersity indices in the range of2.1–2.4 (Table 1). Besides the copolyesters prepared by MP,solid-state modification (SSM) was used to prepare copoly-esters with similar overall chemical compositions. Usually,SSM is performed at 20–30 °C below the melting point of asemicrystalline polymer.27,30,31 However, in this present studythe SSM of poly(butylene terephthalate) (PBT, Tm = 225 °C)with the cyclic diol Thx was conducted at 160 °C under a rela-tively low nitrogen flow, to minimize the volatilization of thecarbohydrate-based monomer. DBTO was also used as a cata-lyst for the SSM reactions, because of its high activity in thelow temperature SSM processes.32 To ensure that theSSMPBxThxyT copolyesters had similar molecular weights com-pared to their MP counterparts, all the SSM reactions weremonitored using size exclusion chromatography (SEC). Similarmolecular weights are necessary in order to make a reliablecomparison of their thermal properties. The molecular weightof the copolyesters prepared by SSM showed the expectedbehavior. Initially, the molecular weight of the polymerdecreased, indicating that chain scission of the PBT chains byalcoholysis had occurred. Subsequently, the molecular weightof the PBT/Thx-based copolyesters increased steadily overtime, showing that chain recombination with elimination ofthe 1,4-butanediol condensation product became the dominat-ing process. The SSM reactions were stopped when Mn and Mw
values of 17–18 and 39–46 kg mol−1 were reached, respectively(Table 1). The chemical composition and structure of bothseries of copolyesters were studied using proton NMR spectro-scopy. The 1H and 13C NMR spectra of a representative Thx-based copolyester, in this case prepared by melt polycondensa-tion, are presented in Fig. 1.
The 1H NMR corroborated their expected chemical struc-ture, and confirmed that Thx had successfully been incorpor-ated into the backbone of the copolyesters, either by MP orSSM. The molar composition of the two series was determinedby integration of the signals corresponding to 1,4BD and Thx.The compositions of the copolyesters prepared by MP werefound to be very close to the feed ratio (Table 1). However, theSSMPBxThxyT copolyesters showed a significant deviation fromtheir feed composition. Even though the SSM reaction con-ditions were very mild, a certain amount of comonomer wasevaporated from the reactor during the SSM process.
3.2. Chemical microstructure
The chemical microstructure of the MPPBxThxyT andSSMPBxThxyT copolyesters was elucidated using quantitative
Scheme 1 Preparation of the bio-based 2,3-O-methylene-L-threitol.
Scheme 2 Preparation of the MPPBxThxyT and SSMPBxThxyT copolyesters.
Green Chemistry Paper
This journal is © The Royal Society of Chemistry 2014 Green Chem., 2014, 16, 1789–1798 | 1791
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online
13C NMR spectroscopy. As shown in Fig. 1, the resonances ofall magnetically different carbon atoms present in the back-bone of the copolyesters are well resolved. The resonancearising from the non-protonated aromatic carbon atom provesto be sensitive to the sequence distribution of the units alongthe copolyester backbone at the level of dyads.34,35 This reso-nance has split into four well-resolved peaks which areobserved in the 133–135 ppm chemical shift interval. Thesefour resonances correspond to the different types of dyads (BB,BThx and ThxB, ThxThx) which are present along the copoly-ester chain, and their peak areas display a dependence on thecopolyester composition (Fig. 2).
The molar fractions (N) of the different dyads in the co-polyester chain were determined by integration of their corres-ponding resonances. Based on these results, the number-average sequence lengths n of the butylene-terephthalate (B)and threitylene-terephthalate (Thx) sequences, as well as thedegree of randomness (R), are determined for each copolyesterby using the equations given below.36,37
nB ¼ ðNBB þ 0:5ðNBThx þ NThxBÞÞ=ð0:5ðNBThx þ NThxBÞÞ
nThx ¼ ðNThxThx þ 0:5ðNBThx þ NThxBÞÞ=ð0:5ðNBThx þ NThxBÞÞ
R ¼ ð1=nBÞ þ ð1=nThxÞ
R-values close to unity are indicative of a fully randomchemical microstructure. When the R-value decreases towardszero or increases towards two, the chemical microstructurewould be more blocky or alternating, respectively. Results fromthese calculations showed that the sequence distribution ofT
able
1Molarco
mposition,m
olecu
larweightan
dmicrostructure
ofthetw
oseriesofThx-basedco
polyesters
Cop
olyester
Molar
compo
sition
Molecular
weigh
tMicrostructure
Cop
olyester
aFe
edDyadcontent
Num
beraverag
esequ
ence
lengths
Ran
domnessd
Ran
domnesse
XB
XThx
XB
XThx
Mnb
Mwb
Ðb
NBBc
NBThx/ThxB
cNThxT
hxc
n Bn T
hx
RRc
PBT
100
010
00
2330
047
100
2.0
MPPB
96Thx 4T
95.9
4.1
955
1870
042
500
2.3
83.8
15.3
0.9
11.9
1.1
0.98
MPPB
91Thx 9T
90.8
9.2
9010
1700
041
300
2.4
78.5
19.9
1.7
8.9
1.2
0.97
MPPB
84Thx 1
6T
84.2
15.8
8515
1680
040
500
2.4
70.7
26.2
3.1
6.4
1.2
0.97
MPPB
82Thx 1
8T
81.8
18.2
8020
1750
041
100
2.3
60.5
34.5
5.0
4.5
1.3
1.00
MPPB
77Thx 2
3T
76.7
23.3
7525
1660
040
300
2.4
57.4
36.1
6.5
4.2
1.4
0.98
MPPB
71Thx 2
9T
70.9
29.1
7030
1600
039
100
2.4
47.4
42.6
10.0
3.2
1.5
0.99
SSMPB
96Thx 4T
95.7
4.3
7723
1840
039
400
2.1
93.0
6.1
1.0
31.6
1.3
0.79
0.83
SSMPB
93Thx 7T
92.6
7.4
8317
1780
046
200
2.4
88.2
10.0
1.8
18.6
1.4
0.79
0.86
SSMPB
91Thx 9T
90.6
9.4
919
1740
039
600
2.2
83.9
12.9
3.2
14.0
1.5
0.74
0.84
SSMPB
89Thx 1
1T
88.7
11.3
6733
1720
041
000
2.4
76.3
18.1
5.6
9.4
1.6
0.72
0.84
SSMPB
82Thx 1
8T
81.7
18.3
7129
1710
042
800
2.5
70.5
22.1
7.4
7.4
1.7
0.73
0.90
PThxT
010
00
100
3800
8800
2.3
aMolar
compo
sition
determ
ined
by1H
NMR
spectroscopy.bNum
beran
dweigh
t-averag
emolecular
weigh
tsin
gmol
−1an
dpo
lydisp
ersities
measu
red
bySE
Cin
HFIPag
ainst
PMMA
stan
dards.
cNBB,N
BThx/ThxB
andNThxT
hxindicate
therelative
molar
amou
ntof
thedyad
sas
obtained
from
13CNMR.d
Ran
domnessinde
xof
copo
lyesters
statistically
calculated
ontheba
sisof
the13CNMRan
alysis.e
Ran
domnessinde
xof
theam
orph
ousfraction
ofcopo
lyesters,d
etermined
usingthecorrection
method
repo
rted
byJansenet
al.30
Fig. 1 1H (top) and 13C (bottom) NMR spectra of MPPB71Thx29T copoly-ester. A schematic representation of the copolyester backbone is alsopresented.
Paper Green Chemistry
1792 | Green Chem., 2014, 16, 1789–1798 This journal is © The Royal Society of Chemistry 2014
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online
the MPPBxThxyT copolyesters is essentially random for thewhole range of compositions, with values of R near unity(Table 1). Conversely, the R-values for the SSMPBxThxyT co-polyesters deviated significantly from unity, falling within the0.72–0.79 range. These copolyesters possess therefore a block-like, non-random overall chemical microstructure, as reportedfor other copolyesters obtained by SSM.30–32 Since the crystal-line phase does not participate in the SSM process due tomobility restrictions, relatively long PBT sequences areexpected to be retained in the SSMPBxThxyT copolyesters. ThesePBT sequences are included in the calculation of the overallchemical microstructure, since these values are determinedfrom solution 13C NMR data and both the random copolyesterblocks and the relatively long PBT sequences of the copolyestermacromolecules are dissolved for the NMR analysis. It is desir-able to evaluate the chemical microstructure of the amorphousphase only, disregarding the contribution of the crystallinedomains. To this end, Jansen et al. have developed a methodto correct the R-values for the presence of the crystallinephase, yielding the randomness of the amorphous phaseonly.30 The chemical microstructure of the complete amor-phous phase of the SSMPBxThxyT copolyesters was determinedby using this calculation method, which implements dataobtained by differential scanning calorimetry (DSC), as dis-cussed below. The corrected R-values of the amorphous phaseare shown in Table 1. These R-values are closer to unity, thusindicating that the amorphous phase would be practicallyrandom. Hence, by SSM the Thx is incorporated into the
mobile amorphous fraction in a similar way as through meltpolycondensation, even though in SSM only the most mobilechain segments participate in the transesterification reactions.The main conclusion that can be drawn from this study is thatby using MP and SSM, partially bio-based copolyesters withdifferent overall chemical microstructures can be prepared.MP leads to fully random copolyesters, whereas the SSM-prepared copolyesters show a blocky overall microstructure,formed by long neat PBT sequences, originating from the crys-talline phase present during SSM, and random copolyester seg-ments that are located in the amorphous phase and that haveparticipated in the transesterification reactions.
3.3. Thermal properties
The thermal behavior of both the MPPBxThxyT and theSSMPBxThxyT copolyesters has been comparatively studied byTGA and DSC. The thermal parameters resulting from theseanalyses are shown in Table 2, where the corresponding datafor the parent homopolyesters PThxT and PBT are alsoincluded for comparison.
The thermal stability of the MP- and SSM-prepared copoly-esters, as well as the homopolyester PThxT, was evaluated byTGA under an inert atmosphere and compared with PBT. TheTGA traces are depicted in the ESI† file (Fig. S1 and S2). Thethermal properties derived from the shown thermograms, viz.the onset decomposition temperature (T5%) and the tempera-ture of maximum degradation rate (Td), are listed in Table 2.The weight loss profiles obtained for the copolyesters reveal a
Fig. 2 13C NMR signals used for the microstructure analysis of the MPPBxThxyT (left) and SSMPBxThxyT (right) copolyesters with schematic represen-tation of the dyads to which they are assigned.
Green Chemistry Paper
This journal is © The Royal Society of Chemistry 2014 Green Chem., 2014, 16, 1789–1798 | 1793
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online
thermal degradation mechanism involving one main degra-dation step and leaving a final residue of 2–10% of the initialweight. None of the polyesters show significant weight loss attemperatures below 360 °C. The T5% and Td of PBT are deter-mined to be 371 and 408 °C, respectively. A higher thermalstability is observed for the PThxT homopolyester, whichshows the aforementioned decomposition parameters at 387and 441 °C, respectively. As expected, the T5% and Td for boththe MPPBxThxyT and the SSMPBxThxyT copolyesters are found atintermediate values between those of their two parent homo-polyesters. No noteworthy differences were detected in thethermal stability of the bio-based copolyesters prepared by MPand SSM containing similar amounts of Thx. The valuable con-clusion drawn from this comparative thermogravimetric studyis that the insertion of the carbohydrate-based cyclic Thxyields copolyesters with enhanced thermal stability withrespect to pure PBT.
The DSC analysis revealed that the incorporation of Thxinto the backbone of PBT induced significant changes in theglass-transition temperature (Tg) when compared to neat PBT(Table 2). The dependence of the Tg values on the Thx contentis plotted in Fig. 3. Remarkably, PThxT has a Tg value 50 °Chigher than PBT, proving a more rigid main chain structure.Consequently, the partial replacement of the 1,4-butanediolunits by Thx residues yields an increase in Tg, which showsno dependency on the copolyester preparation method (MP orSSM).
Melting temperatures (Tm) and enthalpies (ΔHm) weremeasured by DSC and are listed in Table 2. PBT is a semicrystal-line polyester with a melting temperature of 225 °C. Copolymeri-zation of PBT with other compounds commonly affects themelting and crystallization behavior significantly. Endothermicpeaks, indicative of the presence of a crystalline phase, wereobserved for all MPPBxThxyT and SSMPBxThxyT copolyesters,
whereas PThxT did not show such a behavior and proves to befully amorphous. The Tm of the copolyesters proved to be depen-dent on the Thx content and also on the preparation method.Since in the first heating run the samples prepared by MP orSSM have a different thermal history, the SSM samples beingannealed during the SSM treatment, a comparative analysis ofthe Tm will be performed on the results obtained from thesecond heating run, after crystallization from the melt (dis-cussed below). Melting and crystallization enthalpies of the MPand SSM copolyesters, however, will be compared for firstheating, cooling and second heating.
From DSC analysis of the obtained copolyesters it is clearthat the materials prepared by SSM exhibit higher first heatingmelting enthalpies compared to those prepared by MP exhibit-ing a similar overall chemical composition (Table 2). Thishigher enthalpy is due to the segmented nature of these
Table 2 Thermal properties of the two series of Thx-based copolyesters
Copolyester
TGA DSC
First heatinge Coolinge Second heatinge
T5%a
(°C)Td
b
(°C)Wc
(%)Tg
d
(°C)Tm(°C)
ΔHm(J g−1)
Tc(°C)
ΔHc(J g−1)
Tc(°C)
ΔHc(J g−1)
Tm(°C)
ΔHm(J g−1)
PBT 371 408 2 31 225 60.5 193 52.0 — — 224 49.0MPPB96Thx4T 375 410 5 39 212 42.8 183 40.5 — — 203/212 f 39.5MPPB91Thx9T 375 409 5 41 206 41.5 179 36.7 — — 199/208 f 35.8MPPB84Thx16T 375 412 5 45 197 38.6 167 29.9 — — 192/199 f 28.1MPPB82Thx18T 376 411 6 49 189 32.5 152 27.3 — — 182/191 f 26.5MPPB77Thx23T 373 412 6 51 185 31.3 144 25.2 — — 178/186 f 24.9MPPB71Thx29T 374 414 10 56 169 24.5 116 16.7 91.6 3.3 172 19.9SSMPB96Thx4T 372 408 5 35 214/222 f 76.2 188 47.9 — — 209/219 f 47.6SSMPB93Thx7T 373 409 8 37 211/218 f 75.2 182 46.7 — — 204/214 f 45.5SSMPB91Thx9T 372 410 6 38 210/218 f 74.3 181 46.1 — — 202/213 f 42.9SSMPB89Thx11T 372 409 6 41 207/216 f 66.5 172 37.1 — — 198/208 f 36.9SSMPB82Thx18T 372 409 8 44 209 63.2 167 35.8 — — 194/203 f 35.4PThxT 387 441 10 83 — — — — — — — —
a Temperature at which 5% weight loss was observed. b Temperature of the maximum degradation rate. c Remaining weight at 600 °C. dGlass-transition temperature taken as the inflection point of the heating DSC traces recorded at 20 °C min−1 during the third heating run. eMelting(Tm) and crystallization (Tc) temperatures, and melting (ΔHm) and crystallization (ΔHc) enthalpies measured by DSC at heating/cooling rates of10 °C min−1. fMultiple melting peaks.
Fig. 3 Glass-transition temperature versus composition plots for theMPPBxThxyT and the SSMPBxThxyT copolyesters. The triangle representsthe Tg of the homopolyester PBT.
Paper Green Chemistry
1794 | Green Chem., 2014, 16, 1789–1798 This journal is © The Royal Society of Chemistry 2014
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online
SSM-prepared copolyesters and annealing of the copolyesterduring the SSM reaction. In order to determine the crystallinityof the Thx-based samples, to elucidate if co-crystallization hadtaken place and to complement the DSC data, wide-angle X-raydiffraction (WAXD) was performed on the two series of as syn-thesized copolyesters as well as on the PThxT and PBT homo-polyesters. The WAXD profiles and the most important Braggspacings present therein are shown in the ESI (Fig. S3 andTable S1†). The pattern of PBT is characterized by five promi-nent reflections at 5.6, 5.1, 4.3, 3.8 and 3.6 Å. Essentially thesame pattern, in terms of spacing and relative intensities, isshared by all MPPBxThxyT and SSMPBxThxyT copolyesters,revealing that the triclinic crystal structure of PBT38 has beenretained and that co-crystallization of ThxT repeat units in thePBT crystals does not occur. In agreement with DSC results,PThxT did not show any discrete reflections characteristic ofcrystalline material. The crystallinity index of the materials wasestimated as the quotient between the crystalline and totalarea of the X-ray diffraction patterns (Table S1†). A noticeabledecrease in crystallinity with increasing Thx content wasobserved for the MP series, which was expected in view of thelack of co-crystallizability of the ThxT repeat units with BTrepeat units. Regarding SSMPBxThxyT copolyesters, a slightdecrease in crystallinity was also observed during first heatingwhen the Thx content was increased, indicating that the crystal-line phase was not completely retained during the SSM process.This phenomenon was observed before in other copolyestersprepared by SSM, and can be attributed to the participation ofthe outer side of the PBT crystallites in trans-reactions, therebygradually dissolving part of the crystallites.32,39
From the data presented in Table 2, it can be noted that allthe MPPBxThxyT and SSMPBxThxyT copolyesters were able tocrystallize from the melt, and that the crystallization tempera-ture (Tc) and the crystallization enthalpy (ΔHc) both decreasewith increasing Thx content. Besides the fact that Thx-basedrepeat units cannot (co)-crystallize (see earlier), the increasingamount of stiffer Thx containing segments makes crystalliza-tion from the melt more difficult. Nevertheless, when theresults for the MP series are compared to those obtained forcopolyesters prepared by SSM at similar compositions, thematerials obtained by SSM showed higher Tc and ΔHc values.This was expected, since longer PBT sequences are present inthe SSM-prepared copolyesters, which facilitate the crystalliza-tion of the materials. A further detailed crystallization study isdescribed below.
The thermal properties obtained from the second heatingrun of the MP- and SSM-prepared samples can be compara-tively studied since they have similar thermal histories andhave all been crystallized from the melt. After this melt-crystal-lization, MPPBxThxyT copolyesters recovered about 70–95% oftheir initial crystallinity and displayed no changes in theirmelting temperatures. However, when the SSM copolyesterswere crystallized from the melt and reheated, there was a defi-nite change in their melting behavior. Both the Tm and ΔHm
decreased significantly when the first and second heating werecompared. A reason for this difference is the (partial) ran-
domization of the chemical microstructure by transesterifica-tion reactions occurring in the melt. This randomization is anentropically driven process and is catalyzed by the presence ofDBTO, which was added to the physical mixtures to facilitatethe Thx incorporation. To study the randomization pheno-menon, samples of SSMPB82Thx18T were kept in the melt at250 °C for predetermined times and after this treatment theirchemical microstructure and thermal properties were studiedusing 13C NMR and DSC, respectively (Fig. S4†). A decrease inthe melting temperature is observed as a function of the timeduring which the sample was kept in the melt, which wasaccompanied by an increase of the R-value from 0.73 to 0.92,showing that almost complete randomization of the chemicalmicrostructure occurred after long residence times at 250 °C.Therefore, the changes in the thermal behavior between thefirst and second heating run are caused by the (at least partial)randomization of the chemical microstructure yielding shortercrystallizable PBT sequences. When the Tm values obtainedfrom the second heating runs of the copolyesters prepared bySSM are compared to those prepared by MP, noticeable differ-ences are observed and an overview is given in Fig. 4. Interest-ingly, both Tm and ΔHm recorded during the second heatingrun of the samples prepared by SSM are higher than thosemeasured for the MPPBxThxyT series, the higher melting pointsof the SSM samples being related to the formation of thickerlamellae from the longer PBT sequences. This implies thateven though randomization of the blocky chemical microstruc-ture of the SSM-prepared copolyesters does occur to someextent during recording the DSC traces, it is not yet so exten-sive as to yield fully random copolyesters for which only rela-tively thin lamellae with low melting points can be formed.
3.4. Isothermal crystallization
As mentioned above, all the MPPBxThxyT and SSMPBxThxyTcopolyesters as well as PBT are able to crystallize from themelt. Since crystallization from the melt is relevant for proces-sing, the isothermal crystallization of MPPB96Thx4T,MPPB91Thx9T,
SSMPB96Thx4T,SSMPB91Thx9T and PBT were com-
paratively studied in the 195–205 °C temperature interval. All
Fig. 4 Overview of the melting temperature during the second heatingrun of the Thx-based copolyesters prepared by MP and SSM. The tri-angle represents the homopolyester PBT.
Green Chemistry Paper
This journal is © The Royal Society of Chemistry 2014 Green Chem., 2014, 16, 1789–1798 | 1795
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online
the polyester materials could be isothermally crystallized at200 °C. The evolution of the relative crystallinity, Xc, versuscrystallization time is shown in Fig. 5.
The analysis of the isothermal crystallization experimentswas performed using the Avrami method, which relates thefraction of non-crystallized polymer (1 − Xc) to the isothermalcrystallization time. The Avrami equation which was used tothis end is presented below.
1� Xc ¼ exp ½�kðt� t0Þn�
The observed onset (t0) and half-crystallization (t1/2) times,as well as the corresponding calculated crystal growth (k) andAvrami parameters (n), which can be determined by plottinglog[−ln(1 − Xc)] vs. log(t − t0), are given for each experiment(Table 3). It can be noted that for all the copolyesters theAvrami exponent n increases with temperature, the valuesobtained being in the 2.0–2.6 range. The double-logarithmicplot of these data indicated that only primary crystallizationtakes place in the selected time interval (Fig. S5†). An increase
in crystallization temperature yielded a delay in the onset ofcrystallization besides the decrease in crystallization rate.Moreover, it was observed that the presence of minor amountsof Thx noticeably depressed the crystallizability of PBT. Fur-thermore, the comparison of crystallization data forMPPB96Thx4T and SSMPB96Thx4T evidenced that crystallizabilityis more repressed for a fully random copolyester prepared byMP than for the more blocky SSM-prepared copolyester. Inaccordance, SSMPB91Thx9T was found to crystallize at 200 °Cwith a higher crystallization rate than the analogousMPPB91Thx9T copolyester.
The conclusion that can be drawn from this comparativestudy is that the presence of Thx units depresses the crystalliz-ability of terephthalate-based polyesters, and that the crystalli-zation rate depends on the chemical microstructure of thepolymeric material. SSM allows for the preparation of morereadily crystallizable copolyesters, since long PBT sequencesare retained during the SSM process.
3.5. Hydrolytic degradation
To investigate the influence of the incorporation of Thx on thehydrolytic degradation of PBT, a comparative study of the twohomopolyesters PThxT and PBT in addition to the copolyesterswith 18% Thx prepared by MP and SSM was carried out at pH2.0 at 80 °C. Under these conditions, it is possible to study thedegradation in a reasonable amount of time because of therelatively high hydrolysis rate. Changes in sample weight andmolecular weight at increasing incubation times are depictedin Fig. 6.
PThxT underwent a weight loss of approx. 16% after sixweeks of incubation time, which was accompanied by adecrease in the molecular weight of about 50%. This is in con-trast to PBT, because the latter did not show any weight loss orsignificant decrease in molecular weights under these con-ditions. The decreases in the sample weight and the molecularweight were also noticeable for MPPB82Thx18T andSSMPB82Thx18T, evidencing the positive effect of the Thx unitson the hydrolytic degradability. Furthermore, SEM analysisshowed more apparent physical degradation of the surfaces ofMPPB82Thx18T,
SSMPB82Thx18T and PThxT compared to PBT(Fig. S6†). The decrease of the sample weight and molecularweight of the two copolyesters were found to be intermediatebetween those of PBT and PThxT homopolyesters, althoughthe copolyester prepared by MP showed more extensive hydro-lytic degradation compared to the sample prepared by SSM.This can be explained by the higher crystallinity ofSSMPB82Thx18T compared to MPPB82Thx18T. The relatively longPBT sequences in the sample prepared by SSM do limit thedegradation. The melting enthalpies of the initial samplesobtained by solvent casting were 45.9 and 28.2 J g−1, respect-ively. It is noteworthy to mention that the melting enthalpiesof the remaining parts of both the copolyesters increased up to59.3 and 46.6 J g−1 after 6 weeks of incubation, respectively.Such an increase in crystallinity is indicative that hydrolysishas taken place in the amorphous phase, as is usuallyobserved in semicrystalline polymers.
Fig. 5 Relative crystallinity versus time plot of PBT, MPPB96Thx4T,MPPB91Thx9T,
SSMPB96Thx4T and SSMPB91Thx9T isothermally crystallizedat 200 °C.
Table 3 Isothermal crystallization dataa
Copolyester Tc (°C) t0 (min) t1/2 (min) n −logk Tm (°C)
PBT 200 0.19 0.82 2.14 −0.25 223.9205 0.51 2.68 2.45 1.02 225.5
MPPB96Thx4T 195 0.41 4.82 2.56 1.77 210.0200 0.55 12.77 2.61 2.99 212.1
MPPB91Thx9T 195 1.58 9.52 2.06 2.04 207.0200 6.64 33.01 2.08 3.14 209.5
SSMPB96Thx4T 195 0.19 1.54 2.31 0.48 215.8200 0.36 4.74 2.62 1.82 217.9
SSMPB91Thx9T 195 0.23 5.95 2.01 1.68 213.0200 1.74 17.29 2.17 2.74 215.2
a Tc: crystallization temperature; t0: onset crystallization time; t1/2: half-crystallization time; n: Avrami index; k: growth-rate parameter; Tm:melting temperature.
Paper Green Chemistry
1796 | Green Chem., 2014, 16, 1789–1798 This journal is © The Royal Society of Chemistry 2014
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online
To gain insight into the degradation mechanism of thepolyester chain at the molecular level, 1H NMR spectra wererecorded for the polymer residues after incubation (Fig. 7 andESI file, Fig. S7–S9†). The spectrum of the residue of PThxTdisplayed, in addition to the signals characteristic for thepolymer, an increase in the signal arising from CH2OH endgroups. This is indicative of the reduced molecular weight ofthe incubated sample; this observation is in full agreementwith data provided by the SEC analysis (Fig. 6). Furthermore, afull stability of the cyclic Thx structure against hydrolysis canbe inferred from the total absence of any signal indicative ofthe hydrolysis of the acetal group in the MPPB82Thx18T,SSMPB82Thx18T and PThxT spectra. This result is rather strikingbecause acetal groups are known to be sensitive to acidic con-ditions,40 and the opening of the dioxolane ring might betherefore expected to happen to some extent. To corroboratethe stability of the cyclic structure against hydrolysis, 2,3-O-methylene-L-threitol was incubated in aqueous buffer at pH2.0, 7.4 and 10.5 at 80 °C for 6 weeks. The spectra recorded atthe end of the incubation period are shown in the ESI file(Fig. S10†); all spectra correspond to the structure of the
original diol without any sign which would indicate hydrolysisof the acetal group taking place.
4. Conclusions
The cyclic diol 2,3-O-methylene-L-threitol (Thx) derived fromnaturally occurring tartaric acid has been used as a carbo-hydrate-based building block in the preparation of poly(buty-lene terephthalate) (PBT) copolyesters. Two series of partiallybio-based copolyesters with comparable molecular weightswere prepared by either melt polycondensation (MP) of Thxwith dimethyl terephthalate and 1,4-butanediol, or by thesolid-state modification (SSM) of PBT with Thx. Copolyesterswith compositions very close to the feed ratio and a randomchemical microstructure were obtained by MP. Conversely,SSM yielded only the partial incorporation of the Thx in thePBT backbone and led to a limited range of compositions.These copolyesters had a block-like overall chemical micro-structure, formed by PBT sequences present in the crystallinephase during the SSM process and a practically random amor-phous phase, which takes part in the transesterification reac-tions. However, the blocky microstructure of the copolyestersprepared by SSM randomized after prolonged times in themelt.
The incorporation of the cyclic bio-based Thx affordedcopolyesters displaying a satisfactory thermal stability andexhibiting increased Tg values. Furthermore, the unique block-like chemical microstructure obtained after SSM yielded co-polyesters with remarkable thermal properties. The SSM-prepared materials showed better crystallizability from themelt because of the long PBT sequences present in the back-bone of these copolyesters. Furthermore, they displayed higherTm values and a higher crystallinity compared to their MPcounterparts, although these properties were not completelyrecovered after crystallizing from the melt as found for theMP-prepared samples.
Fig. 6 Degradation of PBT, MPPB82Thx18T,SSMPB82Thx18T and PThxT at
pH 2.0 at 80 °C. Remaining weight (a) and molecular weight (b) versustime.
Fig. 7 1H NMR spectra in CDCl3–TFA-d of PThxT after incubation at pH2.0 at 80 °C for 6 weeks (top) and initial sample (bottom).
Green Chemistry Paper
This journal is © The Royal Society of Chemistry 2014 Green Chem., 2014, 16, 1789–1798 | 1797
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online
The presence of the Thx significantly increased the hydro-lytic degradability of PBT. This effect was more pronounced forthe materials prepared by MP because of their lower crystalli-nity. The degradation of Thx-containing polyesters happenedthrough hydrolysis of the main chain ester bonds withoutalteration of the acetal structure.
Acknowledgements
The authors thank the Bio-Based Performance Materials (BPM)Program (The Netherlands), MINECO (Spain) (MAT2009-14053-C02-02 and MAT2012-38044-CO3-03 grants) and AGAUR (Cata-lonia) (2009SGR1469 grant) for financial support. The authorsare also indebted to MEC (Spain) for the FPU grant awarded toCristina Lavilla.
References
1 F. W. Lichtenthaler, Carbohydrates as Organic RawMaterials, in Ullmann’s Encyclopedia of Industrial Chemistry,2010, DOI: 10.1002/14356007.n05_n07.
2 R. Wool and S. Sun, Biobased Polymers and Composites,Academic Press, New York, 2005.
3 J. A. Galbis and M. G. García-Martín, Sugars as Monomers,in Monomers, Polymers and Composites from RenewableResources, ed. M. N. Belgacem and A. Gandini, Elsevier,Oxford, 2008, pp. 89–114.
4 C. K. S. Pillai, Des. Monomers Polym., 2010, 13, 87–121.5 D. E. Kiely, L. Chen and T. H. Lin, J. Am. Chem. Soc., 1994,
116, 571–578.6 D. E. Kiely, L. Chen and T. H. Lin, J. Polym. Sci., Part A:
Polym. Chem., 2000, 38, 594–603.7 J. A. Galbis and M. G. García-Martín, Top. Curr. Chem.,
2010, 295, 147–176.8 A. Gandini, Green Chem., 2011, 13, 1061–1083.9 G. T. Blair and J. J. DeFraties, Hydroxy Dicarboxylic Acids,
in Kirk-Othmer Encyclopedia of Chemical Technology, 2000,DOI: 10.1002/0471238961.0825041802120109.a01.
10 C. Japu, A. Martínez de Ilarduya, A. Alla and S. Muñoz-Guerra, Polymer, 2013, 54, 1573–1582.
11 D. P. R. Kint, E. Wigström, A. Martínez de Ilarduya, A. Allaand S. Muñoz-Guerra, J. Polym. Sci., Part A: Polym. Chem.,2001, 39, 3250–3262.
12 R. Marín and S. Muñoz-Guerra, J. Polym. Sci., Part A: Polym.Chem., 2008, 46, 7996–8012.
13 A. Alla, J. Oxelbark, A. Rodríguez-Galán and S. Muñoz-Guerra, Polymer, 2005, 46, 2854–2861.
14 A. Alla, A. Rodríguez-Galán and S. Muñoz-Guerra, Polymer,2000, 41, 6995–7002.
15 R. R. Gallucci and B. R. Patel, Poly(Butylene Terephthalate),in Modern Polyesters, Chemistry and Technology of Polyestersand Copolyesters, ed. J. Scheirs and T. E. Long, John Wiley &Sons, Chichester, 2004, pp. 293–321.
16 W. D. Li, J. B. Zeng, X. L. Lou, J. J. Zhang and Y. Z. Wang,Polym. Chem., 2012, 3, 1344–1353.
17 F. Zamora, K. Hakkou, A. Alla, M. Rivas, I. Roffé,M. Mancera, S. Muñoz-Guerra and J. A. Galbis, J. Polym.Sci., Part A: Polym. Chem., 2005, 43, 4570–4577.
18 A. Gandini, D. Coelho, M. Gomes, B. Reis andA. J. D. Silvestre, J. Mater. Chem., 2009, 19, 8656–8664.
19 E. Gubbels, L. Jasinska-Walc and C. E. Koning, J. Polym.Sci., Part A: Polym. Chem., 2013, 51, 890–898.
20 C. Lavilla, A. Martínez de Ilarduya, A. Alla and S. Muñoz-Guerra, Polym. Chem., 2013, 4, 282–289.
21 F. Fenouillot, A. Rosseau, G. Colomines, R. Saint-Loup andJ. P. Pascault, Prog. Polym. Sci., 2010, 35, 578–622.
22 H. R. Kricheldorf, G. Behnken and M. Sell, J. Macromol.Sci., Pure Appl. Chem., 2007, 44, 679–684.
23 S. Dhamaniya and J. Jacob, Polymer, 2010, 51, 5392–5399.24 R. V. Gómez and O. Varela, Macromolecules, 2009, 42, 8112–
8117.25 C. Lavilla and S. Muñoz-Guerra, Polym. Degrad. Stab., 2012,
97, 1762–1771.26 C. Lavilla, A. Alla, A. Martínez de Ilarduya and S. Muñoz-
Guerra, Biomacromolecules, 2013, 14, 781–793.27 M. A. G. Jansen, J. G. P. Goossens, G. de Wit, C. Bailly and
C. E. Koning, Anal. Chim. Acta, 2006, 557, 19–30.28 S. N. Vouyiouka, E. K. Karakatsani and C. D. Papaspyrides,
Prog. Polym. Sci., 2005, 30, 10–37.29 C. D. Papaspyrides and S. N. Vouyiouka, Fundamentals of
Solid-State Polymerization, in Solid State Polymerization, ed.C. D. Papaspyrides and S. N. Vouyiouka, John Wiley & Sons,Hoboken, 2009, pp. 2–158.
30 M. A. G. Jansen, J. G. P. Goossens, G. de Wit, C. Bailly,C. Schick and C. E. Koning, Macromolecules, 2005, 38,10658–10666.
31 E. Gubbels, L. Jasinska-Walc, D. Hermida Merino,H. Goossens and C. Koning, Macromolecules, 2013, 46,3975–3984.
32 C. Lavilla, E. Gubbels, A. Martínez de Ilarduya,B. A. J. Noordover, C. E. Koning and S. Muñoz-Guerra,Macromolecules, 2013, 46, 4335–4345.
33 C. Lavilla, A. Alla, A. Martínez de Ilarduya, E. Benito,M. G. García-Martín, J. A. Galbis and S. Muñoz-Guerra, Bio-macromolecules, 2011, 12, 2642–2652.
34 R. A. Newmark, J. Polym. Sci., 1980, 18, 559–563.35 H. R. Kricheldorf, Makromol. Chem., 1978, 179, 2133–2143.36 Y. Yamadera and M. Murano, J. Polym. Sci., Part A: Polym.
Chem., 1967, 5, 2259–2268.37 R. C. Randall, in Polymer Sequence Determination, Academic
press, New York, 1977, p. 71.38 I. H. Hall, in The Determination of the Structures of Aromatic
Polyesters from their Wide Angle X-Ray Diffraction Patterns, ed.I. H. Hall, Elsevier Applied Science, London, 1984, pp. 39–78.
39 M. A. G. Jansen, L. H. Wu, J. G. P. Goossens, G. de Wit,C. Bailly, C. E. Koning and G. Portale, J. Polym. Sci., Part A:Polym. Chem., 2008, 46, 1203–1217.
40 M. B. Smith and J. March, March’s Advanced Organic Chem-istry, Wiley-Interscience, Hoboken, 2007, p. 523, 1270.
Paper Green Chemistry
1798 | Green Chem., 2014, 16, 1789–1798 This journal is © The Royal Society of Chemistry 2014
Publ
ishe
d on
16
Oct
ober
201
3. D
ownl
oade
d by
Uni
vers
ity o
f U
tah
on 0
7/05
/201
4 16
:15:
01.
View Article Online