forming crystalline polymer-nano interphase structures for high-modulus and high- tensile/strength...

Full Paper

Forming Crystalline Polymer-Nano InterphaseStructures for High-Modulus and High-Tensile/Strength Composite Fibers

Jiangsha Meng, Yiying Zhang, Kenan Song, Marilyn L. Minus*

PVA/single-walled nanotube (SWNT) composite fibers are fabricated using a steady shear-flowgel-spinning method. The resultant fibers show excellent tensile strength, modulus, andtoughness of 4.9 GPa, 128 GPa, and 202 J � g�1, respectively. Templated interfacial
crystallization of PVA in the vicinity of SWNT iscontrolled by tailoring the degree of undercoolingof PVA during the composite solution preparation.WAXD shows that the templated crystallizationbehavior of the PVA at the SWNT interfacialregion is new. PVA/SWNT fibers that exhibitinterfacial structure show a predominant crystalli-zation plane of (001) as compared to the (101)plane seen in PVA/SWNT fibers without a distinctinterfacial structure. This demonstrates that thePVA interfacial region around SWNT has densercrystalline chain-packing.
J. Meng, Y. Zhang, K. Song, Prof. M. L. MinusDepartment of Mechanical and Industrial Engineering,Northeastern University, Boston MA 02115, USAE-mail: [email protected]

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com

Early View Publication; these are NOT the fin

lution-state spinning. For solid-state spinning
1. Introduction
The fabrication of polymer-based carbon nanotubes (CNT)

reinforced fiberswith high concentration of CNT (>10wt%)

has been a study of interest in recent years, motivated by

the potential to produce composites with outstanding

mechanical, electrical, and thermal properties.[1–12] Two

main strategies are typically used to produce high-

concentration CNT-based fiber: (i) solid-state spinning

and (ii) so

processes, CNT fibers are spun by pulling CNT strands

directly from CNT forests,[1,2] or from the CNT aerogels that

are formed during growth in the chemical vapor deposition

(CVD) reactor.[4,10] The mechanical performance of the

resultant fibers strongly depends on the quality of the CNT

source, as well as the fiber spinning process, and the post-

spinning treatments.[13] It requires the nanotubes not only

to be as long and as free of defects as possible, but also

aligned along the fiber axis with great precision for the

tubes to contribute to better axial properties in the fiber.

However, the CNT length and defect control depends on the

CNT synthesis processes. Additionally, CNT alignment in

the fiber is affected by several mechanisms: (i) twisting

processes duringfiber spinning (i.e.,which canmisalign the

Macromol. Mater. Eng. 2013, DOI: 10.1002/mame.201300025 1

al page numbers, use DOI for citation !! R

www.mme-journal.de

J. Meng et al.

2

REa

CNT by introducing a twisting angle with respect to the

fiber axis); and (ii) post-spinning treatments such as

densification, used to increase the contact area between

CNT (i.e., improves the load transfer). Such post-processing

treatments may also misalign the CNT due to unevenly

distributed densification as well as the introduction of

different-sized vacancies between CNT. It has been shown

that the theoretical Young’s modulus of the fiber can

decrease to approximately 80% of its magnitudewith even

a minor change in the degree of orientation (global CNT

alignment of �58) of the CNT with respect to the fiber

axis.[14] Therefore, the quality of the CNT source as well

as the spinning and post-spinning processes need to be

well-controlled in order to improve the fiber mechanical

properties.

The second strategy for processing high-concentration

polymer/CNT composites uses the solution-state spinning

method. In this case, CNT are first dispersed into solvents

or polymer solutions and the composite dopes are

subsequently spun into fibers.[3,5,12,15] The mechanical

performance of the resultant fibers will depend on several

factors including the dispersion of CNT, alignment of

CNT, the CNT length, and the polymer/CNT interfacial

strength.[16] If the dispersion quality of CNT is not well-

controlled, CNT tend to entangle and result in misalign-

ment and/or poor interaction with polymer matrix in the

composite fiber. In addition, preservation of CNT length

in the fiber is also important to reduce defects which

will occur at the CNT ends and result in premature fiber

failure.

In situations where CNT dispersions are prepared

utilizing polymer solutions, the CNT bundles are exfoliated

and the bundle reformation can be prevented by the

presence of the polymer chains which adsorb to the CNT.

The adsorption of the polymer to the CNT may lead to the

formation of an interphase structure between polymer and

CNT. The interphase is considered to be the major factor

of maximizing the mechanical contribution from CNT

due to its ability to provide load transfer from polymer

matrix to CNT.[17]

Crystallization processes provide a mechanism for

the formation of the polymer interfacial structure on

CNT. It has been shown that CNT can act as a nucleating

agent for polymer crystallization,[18–20] and the crystalline

polymer/CNT interface has been observed for various

polymers.[14,21–23] This interfacial polymer crystallization

on CNT improves the polymer/CNT binding energy,

which could lead to better load transfer from polymer

matrices to CNT upon stretching.[24]

Poly(vinyl) alcohol (PVA) is a commonly used polymer

to produce CNT reinforced composite fibers.[3,5,12,25–29]

The first continuous fibers with high concentration of

CNT were produced in PVA solution, where the single-

walled nanotubes (SWNT) suspension was injected into

Macromol. Mater. Eng. 2013, DOI

� 2013 WILEY-VCH Verlag Gmb

rly View Publication; these are NOT the final pag

a rotating PVA flow to produce SWNT fibers (>60 wt% of

SWNT) with Young’s modulus of 15 GPa and tensile

strength of 0.15 GPa.[12] A few years later, the spinning

process was refined and much better SWNT fibers was

achieved with average Young’s modulus of 244 GPa

and average tensile strength of 2.9 GPa.[3,5] It should be

noted that these high tensile modulus and strength

properties (244 and 2.9 GPa, respectively) correspond to

fibers with very small diameter (�1.4mm). As the diameter

increased the properties of PVA/SWNT fibers were found

to decrease. Fibers with diameters 10mm or higher

exhibited average modulus and tensile strength below

30 and 500 MPa, respectively.[3] The composite fibers

produced and discussed in this work have diameters of

�10mm. In addition to the composite property trends for

PVA/SWNT fibers, it has also been reported that the PVA

has the ability to form an extended-chain crystalline

interface on SWNT.[22] Resultant PVA/SWNT composite

fibers (with �1 wt% SWNT loading) which exhibit this

extended-chain morphology, show great improvement in

mechanical performance.[14]

In this work, the PVA/SWNT composite fibers were

produced using a steady shear-flow apparatus. Crystalli-

zation temperature was controlled during the composite

solution preparation to facilitate the polymer interfacial

crystal growth in the vicinity of the SWNT. The shear-

flow during spinning and fiber hot-drawing were

used to promote high orientation of the PVA and

SWNT. The composite fibers exhibited oriented inter-

facial structure which resulted in high mechanical

performance with maximum Young’s modulus, tensile

strength, and toughness of 128 GPa, 4.9 GPa, and

202 J � g�1, respectively.

2. Experimental Section

2.1. Materials

PVA (molecular weight 325 000 g �mol�1, Mowiol 235) was

obtained from Kuraray America Incorporation. SWNT (purity

>90 wt%, ash <1.5 wt%) was obtained from Cheaptubes

Incorporation. Dimethyl sulfoxide (DMSO) was purchased from

Sigma-Aldrich and used as received.

2.2. Solution Preparation

SWNTwas dispersed in DMSO solvent (at a concentration of 21mg

ofSWNTmL�1ofDMSO)usingabathsonicator (FS30manufactured

by Fisher Scientific) for a period of 48 h. After sonication, the PVA

powder and additional DMSO solvent were added into the

dispersion to achieve a final PVA concentration of 5 wt% in DMSO

and SWNT concentration of 10wt%with respect to solid content of

PVA in the solution. The mixture was subsequently heated and

homogenized (T10 ULTRA-TURRAXmanufactured by IKA) for 20 h

: 10.1002/mame.201300025

H & Co. KGaA, Weinheim www.MaterialsViews.com

e numbers, use DOI for citation !!

Forming Crystalline Polymer-Nano Interphase Structures . . .

www.mme-journal.de

in order to further disperse SWNT and completely dissolve all PVA.

Two different solution batches were prepared by using separate

heat treatments during homogenization: (i) the first batch of

solution (S-A) was homogenized under a constant hot plate

temperatureof120 8C;and(ii) thesecondbatchofsolution(S-B)was

homogenized using a hot plate temperature 160 8C for the first 2 h

and subsequently cooled down to 120 8C. The change in tempera-

ture used for the S-B solution induced PVA crystallization during

homogenization.

2.3. SWNT Dispersions

The nanotube dispersions were studied by electron microscopy in

order to determine the length distribution of the SWNT after

sonication.Diluted droplets of thedispersionsweredried on silicon

and imaged by means of scanning electron microscopy (SEM).

Imageswere analyzed using the software package Image J (version

1.44o). It was found that after 48 h of bath sonication the SWNT

length varied considerably. The average SWNT length distribution

based on more than 100 measurements of distinct tubes was

1.5�1.1mm.[32] This experimental length value is used in

conjunction with the theoretical analysis of the mechanical

properties for the PVA/SWNT composite fibers produced in this

work.

2.4. Fiber Preparation

Rotation of a cylindrical stir-bar in a flat bottom flask (gap size

between the cylinder and flask was 10 mm) was used to create

shear-flow. The rotation speed of the stir-bar ranged from50 to 100

RPM.Amethanol spinningbathwasusedandmaintainedat5–8 8C

Table 1. Processing parameters for PVA control and PVA/SWNT fiber

Fabrication Parameters/drawing conditions

sonication time [h]

temperature [8C]

homogenization time [h]

hot plate temperature [8C]

shear flow spinning needle gauge/diameter [mm]

gelation temperature [8C]

syringe push speed [mL � s–1]take-up speed [rpm]

hot drawing process as-spun draw ratio

coagulation time [h]

drawing temperatures [8C] 1st sta

2nd st

draw ratio 1st sta

2nd st

Macromol. Mater. Eng. 2013, DO

� 2013 WILEY-VCH Verlag Gmwww.MaterialsViews.com

Early View Publication; these are NO

for the purpose of gelation. By using a glass syringe (Thermo

ScientificHPLC 100mL), the PVA/SWNT solutionswere injected at a

speedof10mL � s�1 througha22-gaugeblunt tipneedle into thecold

methanol subjected to shear flow. The as-spun fibers were further

immersed in cold methanol for 1 h before being drawn on the hot

plate. Twohot-drawing stageswereusedat temperatures of 90and

160 8C, respectively. The drawn fibers were subsequently cut into

specimens for tensile tests. The spun fibers from the S-A and S-B

solutions are referred to as F-A and F-B fibers, respectively. The

specific gel-spinning and hot-drawing parameters used for the F-A

and F/B composite fibers are listed in Table 1.

2.5. Sample Characterization

Scanning electron microscopy was performed on Supra-25

(operating voltage 5 kV, manufactured by Zeiss). SEM samples

were placed on carbon tape and sputter coated with platinum for

imaging.Tensile testswereperformedusingadynamicmechanical

analyzer G2-RSA model (manufactured by TA Instruments) using

gauge length of 10mmand strain rate of 0.05mm � s�1.Wide-angle

X-ray diffraction (WAXD) patterns were obtained on multi-

filament bundles using Rigaku MicroMax-002 Microfocus X-ray

Generator (operating voltage at 45 kV, current at 0.66 mA, Cu Ka,

l¼0.1541 nm) manufactured by Rigaku Americas Corporation.

3. Results and Discussion

SEM micrographs of the fractured ends of the F-A and F-B

composite fibers are shown in Figure 1. For F-A fibers

(Figure 1a and b), no interfacial coating of PVA is observed

s.

F-A F-B Control fibers

48 48 –

50 50

20 20 PVA was dissolved by

mechanically stirring

on 120 8C hot plate

To: 120 Tf: 120

22/0.394 22/0.394 22/0.394

5–8 5–8

10 10 10

50 50

1.3 1.3 1.3

1 1

ge 90 90 90

age 160 160

ge 5.0 4.6 9.0

age 2.2 2.1

I: 10.1002/mame.201300025

bH & Co. KGaA, Weinheim 3

T the final page numbers, use DOI for citation !! R

Figure 1. SEM images of fiber fracture surfaces for (a) the F-A fiber and (b) high-magnification of F-A fiber showing no evidence of interphase PVA structure aroundSWNT (see arrow), (c) the F-B fiber and (d) high-magnification of the F-b fiber showingdistinct PVA interphase structure around the SWNT (see arrow).

www.mme-journal.de

J. Meng et al.

4

REa

around the SWNT bundles on the fractured surface. In

contrast, very distinct interfacial crystallization of PVAwas

observed for F-B fibers (Figure 1d). The thickness of the

interfacial coating (b) can be calculated from

Tab

Par

eff

ten

elo

ten

tou

eff

eff

int

eff

a)Effec

calcul

manu

respec

rly V

b ¼ Dcoating � DSWNT

2ð1Þ

le 2. Mechanical properties of PVA control and PVA/SWNT composite fibers.

ameter Control fibers

ective fiber diametera)[mm] 22

sile modulus [GPa] 21.8� 3.0

ngation-to-break [%] 11.4� 1.7

sile Strength [GPa] 1.2� 0.3

ghness [J � g�1] 55.8� 12.3

ective modulus of SWNT [GPa] – 20

ective modulus of SWNT with polymer

erphase contribution

–

ective tensile strength of SWNT [GPa] –

tive fiber diameter Dwas calculated as D ¼ 2ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiwc=ðrclcpÞ

pusing the weight method,

ated as rc ¼ rPVAð1� wÞ þ rSWNTw, where rPVA is 1.345 g � cm�3 for fully crystallize

facturer andw is theweightconcentrationof theSWNTin thecompositefibers (wc and

tively; b)Calculated value for aligned SWNTs (orientation factor, ho¼1); c)Calculated

Macromol. Mater. Eng. 2013, DOI: 10.1002/mame.20130

� 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinhe

iew Publication; these are NOT the final page numbers, us

where Dcoating is the average total thick-

ness of the coated SWNT bundle

(�26.7� 5.8 nm), and DSWNT is the

average diameter of the SWNT bundle

(�10.3� 4.4 nm). Based on this calcula-

tion, the average thickness (b) of the

interfacial coating is calculated to be

�8.2 nm.

The mechanical properties of control,

F-A and F-B fibers are listed in Table 2.

Compared to the control fibers, the

Young’s modulus of F-A and F-B fibers

shows an increase of 67 and 446%,

respectively. Comparatively, the tensile

strength of the F-A and F-B fibers also

increases by 125 and 308%, respectively.

The toughness of the F-A and F-B fibers

increases by 82 and 208%, respectively.

The highest tensile strength for the F-B

fiber (4.9 GPa) is larger than some

commercially available polymer high-

performance fibers, such as the PVA

fibers produced for concrete reinforce-

ment (0.88 GPa) and Kevlar (3.6 GPa).

The highest toughness from F-B fibers

(202 J � g�1) is bigger than that of Kevlar

(33 J � g�1) and spider dragline silk (165 J � g�1).[30] Figure 2a

shows the typical strain/stress curves for the control and

composite fibers (F-A and F-B). The stress/strain curves

show that at the same value of elongation at break (�11%),

the F-A fibers show intermediate improvement ofmechan-

ical performance as compared to the control fiber. In

addition, the F-Bfiberswith the interfacial crystallizationof

F-A fibers F-B fibers

9.2 9.9

36.3� 1.3 119.1� 8.6

10.7� 0.7 9.7� 1.1

2.5� 0.1 4.4� 0.5

101.4� 11.4 171.6� 30.4

1.5b)–1042.1c) 1539.8b)–8574.4c)

– 577.1b)–2967.6c)

16.9b)–88.1c) 47.1b)–253.0c)

and The density of the composite fiber rc was

d PVA, rSWNT is 2.1 g � cm�3 according to the

lc areweightand lengthof the compositefibers,

value for randomly oriented SWNTs (ho¼1/5).

0025

im www.MaterialsViews.com

e DOI for citation !!

Figure 2. (a) Stress/strain curve for PVA/SWNT control and composite fibers. Comparedto the control fiber, F-A fiber (no interphase) shows intermediate improvement on both


www.mme-journal.de

PVA have an even greater improvement

in Young’s modulus and tensile strength

as compared to the F-A fibers. Figure 2b

shows a comparison of the Young’s

modulus and tensile strength distribu-

tion of the control, F-A and F-B fibers

versus some of the commercially avail-

able fibers as well as various other PVA/

SWNT fibers.[3,12,27] It can be clearly seen

that the F-B fibers produce in this work

combine both high modulus and high

tensile strength, and this is most likely

due to the presence of interphase PVA

structure which contributes to better

stress-transfer between the SWNT and

the matrix.

The SWNT enhancement efficiency

was studied by calculating the effective

Young’s modulus and tensile strength

of the SWNT based on the mechanical

properties of control, F-A, and F-B fibers.

The composite Young’s modulus and

tensile strength are given by Equation 2

and 3, respectively:[16]

modulus and tensile strength. F-B (interphase present) shows the most mechanicalimprovement with tensile strength of 4.9 GPa. (b) Comparison of Young’s modulus andtensile strength distributions of the control, F-A and F-B fibers in this work versus thesame properties of other commercially available high-performance fibers and variousPVA/SWNT composite fibers [(i) Vigolo et al.,>60 wt% of SWNT; (ii) Xu et al., 0.3 wt% ofSWNT (iii) Young et al., 2–31% of SWNT].[3,12,27] (c) Effective Young’s modulus and (d)tensile strength of SWNT for both F-A and F-B fibers as a function of the orientationfactor (ho). Both effective Young’s modulus and tensile strength of SWNT graduallyapproach real values of SWNT as (ho) increases from 1/5 (randomly distributed) to 1

www.M

Yc ¼ ðh0h1YSWNT�eff

� YPVAÞVSWNT þ YPVA ð2Þ

sc ¼ ðh0h1sSWNT�eff

� sPVAÞVSWNT þ sPVA ð3Þ

(perfectly aligned). The SWNT length (L) variation has been taken into account whengenerating the theoretical curves.

where VSWNT is the volume fraction of

SWNT in the composite fiber. YSWNT-eff,

YPVA and Yc are the effective Young’s

modulus of the SWNT, control PVA and composite fiber,

respectively. sSWNT-eff, sPVA and sc are the effective tensile

strength of SWNT, control PVA and composite fibers,

respectively. ho is the orientation factor which has values

of ho¼ 1 for perfectly aligned fibers and ho¼ 1/5 for

randomly oriented fibers. h1 is the length efficiency factor

described by Equation 4 [31]:

h1 ¼ 1� tan hðalSWNT=DSWNTÞalSWNT=DSWNT

ð4Þ

where a is given by Equation 5 and 6, respectively:

for Young0s modulus calculation:

a ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi�3YPVA

2YSWNT�effln VSWNT

r ð5Þ

for tensile strength calculation:

a ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi�3sPVA

2sSWNT�effln VSWNT

r ð6Þ


� 2013 WILEY-VCH Verlag GmaterialsViews.com


DSWNT is taken as the bundle diameter of the SWNT

measured from SEM analysis of the fractured composite

(10.3� 4.4 nm) and lSWNT is the average length of

the SWNT after 48 h of sonication and determined to be

1.5� 1.1mm.[32] The average (�1.5mm) as well as the

upper (�2.6mm) and lower (�0.4mm) nanotube length

distributions are taken into account when determining

the effective modulus and strength of the CNT in the

composites. The calculated curves for effective YSWNT-eff and

sSWNT-eff as a function of ho are shown in Figure 2c and d.

The theoretical curves are given by calculating Yc and

sc values using ho¼ 1, YSWNT (Experimental)¼ 1470 GPa, and

sSWNT (Experimental)¼ 52 GPa[33] (experimental values of

SWNT bundles) conditions first, and subsequently using Ycand sc values as constants in Equation 2 and (3) to compute

the relationship between ho and YSWNT-eff or sSWNT-eff.

According to the calculation, the effective YSWNT-eff for

the F-B fibers ranges from 1539.8 GPa (ho¼ 1) to 8574.4 GPa

(ho¼ 1/5) and the effective sSWNT-eff for the F-B fibers range

I: 10.1002/mame.201300025



www.mme-journal.de

J. Meng et al.

6

REa

from 47.1 GPa (ho¼ 1) to 253.0 GPa (ho¼ 1/5). These results

suggest that as the orientation factor changes from1/5 to 1,

the calculated values approach the real properties of SWNT

(as mentioned), which have been experimentally and

theoretically determined.[33–37]

Since the fibers are spun under shear-flow and subse-

quently drawn, the SWNT in the fiber systems will tend

to be more aligned than randomly distributed. Therefore,

recognizing that a certain degree of disorientation of SWNT

will remain in the composite fiber, the minimum effective

Young’s modulus and tensile strength of the SWNT for F-B

fibers are 1539.8 and 47.1 GPa, respectively. By comparing

these values with the same analysis for the F-A fibers, the

effectiveYSWNT-eff and sSWNT-eff values of F-Bfibers are 664%

higher and 179% higher, respectively. This result confirms

that by inducing interfacial crystallization of PVA during

fiber processing, a stronger PVA/SWNT interphasehas been

created, and subsequently better load transfer from the

PVA matrix to SWNT has been achieved.

Based on this theoretical analysis using Equation 2, 4, 5,

and6, thepredictedeffectivemodulusvalueatho¼ 1 for the

SWNT in theF-B is ratherhigh. Typically, bothexperimental

andtheoretical SWNTmodulusvaluesof1TPaorgreaterare

associated with individual single-wall nanotubes as

opposed to bundles/ropes. In this work the PVA/SWNT

fibers contain SWNT bundles of �10 nm in diameter, and

the predicted and measured modulus values reported for

SWNT bundles are usually only a few hundred giga-

pascals.[33,42,43] For the F-A fibers where there is no

Figure 3. (a) WAXD data for the F-A fibers shows a major crystallizapattern for the proposed PVA unit cell arrangement withmajor (101) recrystallization reflection for the (001) plane, followed by the simulatmajor (001) reflection. Crystal structure comparison of the F-A and F-Bin a-c projection, and (b2) in a-b projection. The closest O—Odistancewconstructed using the simulatedWAXD pattern results in a-c projectioto be 0.2785 nm.




interphase present, therefore the bulk-PVA matrix is

reinforced by the SWNT bundles. The resultant predicted

effective modulus for fully aligned SWNT is �201 GPa,

and this value is within the ranges experimentally and

theoretically obtained for nanotube bundles.[42,43] Howev-

er, for the F-B fibers themicroscopy studies (Figure 1) show

that the micro-structure consists of interphase and bulk

PVA. For this reason, it is important to understand the

influence of this interphase structure on the mechanical

performance, and how this may influence the reinforce-

ment efficiency of the SWNT in the composite fiber.

WAXD data for both the F-A and F-B composite fibers is

shown in Figure 3a. Interestingly, compared to the F-A

fibers WAXD pattern, the main crystallization plane of

the F-B fibers shifted from themost common (101) plane at

2u¼ 19.38 (i.e., main crystallization plane for the F-A fibers

and typically forPVA) to (001)planeat2u¼ 16.78 (Figure3a).This change of the predominant crystallization plane has

never been reported before. Crystal sizes (L) for the F-A

and F-B fibers on their predominant crystallization planes

were calculated based on the diffraction peak full-width at

maximum B at 2u angle and the wavelength (l) of X-ray

source using Scherrer’s equation,[38]

tion reflflectioned WAfibers:asmen and (

: 10.100

H & Co

e nu

Bð2uÞ ¼ KlL cosðuÞ ð7Þ

whereK is the Scherrer constant and taken to be 0.9, l is the

X-ray wavelength of 0.1541 nm and u is the Bragg angle in

ection for the (101) plane, followed by the simulated WAXD. The experimental WAXD pattern of F-B fibers shows amajorXD pattern for the proposed PVA unit cell arrangement with(b1) unit cell arrangement for the F-A fibers (i.e., Bunn unit cell)asured to be 0.3252 nm. (c1) Unit cell structure for the F-B fibersc2) in a-b projection. The closest O—O distance was measured

2/mame.201300025

. KGaA, Weinheim www.MaterialsViews.com

mbers, use DOI for citation !!

Table 3. Lattice parameters, crystal sizes and minimum hydrogen bond distances for F-A and F-B fibers with their predominant crystalplanes.

Sample

Lattice parameters

b

Crystal

size [nm]

Hydrogen bond

distance [nm]a [nm] b [nm] c [nm]

F-A fiber [predominant (101) crystal plane] 0.781 0.252 0.581 918420 4.48 0.3087a)

F-B Fiber [predominant (001) crystal plane] 0.781 0.252 0.530 918420 8.18 0.2785a)

a)The closest hydrogenbondingdistancesweremeasured fromthe crystal structures of theF-A andF-Bfibers. Thedistancevalue calibration

was done by the simulation software.

Figure 4. Schematic of O—H � � �O hydrogen bond formationbetween two polymer molecules. The total hydrogen bonddistance is composed of length of the hydroxyl bond and theH � � �O interaction distance determined by the non-boundedpotential.


www.mme-journal.de

radians. The calculated crystal sizes of the predominant

crystallization planes for the F-A and F-B fibers are 4.48 nm

(101)and8.18nm(001), respectively.Noticeably, thecrystal

size on (001) plane for F-B fibers is very similar to the

average thickness of the interfacial coating (�8.2 nm)

measured from the SEM image (Figure 1d). This indicates

that the F-B fibers consists of mostly interphase polymer

where the (001) crystallizationplane ispredominant,where

the change in crystallization for the interphase PVA is

templated by the SWNT.

To better understand the crystallization and strengthen-

ing mechanism of the F-B fibers with (001) crystallization

plane, the PVA crystal unit cell models for both the F-A

and F-B fibers were built (Figure 3b and c). The lattice

parameters (a, b, c) for both the F-A and F-B fibers were

calculated from the known indexed peaks present in

the WAXD data (Table 3). The F-A fibers with main

crystallization plane (101) was built based on the typical

PVA unit cell proposed by Bunn,[39] which has amonoclinic

lattice system with two PVA chains P21/m symmetrically

located in diagonal positions. However, since Bunn’s

structure only describes the unit cell with (101) crystalliza-

tion plane, a new unit cell structure is proposed to better

describe the F-B fibers with major crystallization plane

of (001). In this hypothetical unit cell, each PVA chain was

first rotated with respect to the CH-group center where the

carbon backbone lies parallel to (001) plane. Then the two

chains were separated along c-axis to reach the final

positions, where one chain located at (001) plane and the

other chain located at the a-b plane. Figure 3b and c show

the comparison of the PVA unit cells of both composite

fibers in a-c projection and a-b projection. WAXD pattern

simulation was performed using the unit cell structures

(Figure 3a). The simulated patterns for the proposed unit

cell structure for the F-B fiber shows a distinct (001) peak at

2u¼ 16.78 and a shallow (101) peak at 2u¼ 18.98, which is

in agreement with the real WAXD data. (101) and (201)

peaks are slightly shifted in the simulated pattern due to

the perfection of the PVA crystal in the model.

The strengthening mechanism for both the F-A and F-B

fiber crystal structures with respect to the predominant




planes was studied by examining the O—H � � �O hydrogen

bonding between the two PVA chains. As illustrated in

Figure 4, the O—O distance is composed of the length of

the hydroxyl bond (OH group) and the length of the H � � �Onon-bounded potential. It has been shown that the van

der Waals radius for O is around 0.14 nm, which means

any oxygen atoms that are close to around 0.28 nm

can experience repulsion.[40] More precisely, the strongest

O—H � � �O bonding can be found at O—O distance of

0.273 nm by bond valence calculation. Therefore, a O—O

distance reduction from 0.3252 nm in F-A fiber crystal

structures on the (101) plane, to 0.2785 nm in F-B fiber

crystal structuresonthe (001)plane thatmeasured fromthe

models indicates that as the crystallization plane changed

from (101) to (001) the polymer chains become more

adjacent to one another and as a result experience stronger

inter-chain hydrogen bonding. This strong interaction

I: 10.1002/mame.201300025



Figure 5. Schematic depicting the interfacial structure in a (a)PVA/SWNT fiber displaying no interphase structure (F-A fibers),where voids may be present between the SWNT surface andsemi-crystalline polymer matrix; and (b) PVA/SWNT fiberdisplaying interphase structure (F-B fibers), where voidpresence is reduced between the SWNT surface and interphasepolymer due to PVA crystalline interface.

www.mme-journal.de

J. Meng et al.

8

REa

between polymer chains introduces more chain resistance

upon stretching and may also result in the greater

mechanical improvement as observed for the F-B fibers.

A schematic illustrating the PVA chain morphology in

the vicinity of SWNT for both the F-A fiber (no interphase)

and F-B fiber (interphase present) is shown in Figure 5.

Without the interfacial crystallization presence (Figure 5a),

voids may exists between the matrix interface at the

surface of the SWNT. This may be due to the polymer semi-

crystalline morphology which consists of both unevenly

distributed folded-chain crystals and amorphous chains.

The presence of voids will decrease the contact area




between PVA chains and SWNT, lower the stress transfer

efficiency from polymer matrix to the SWNT dramatically,

and result failure due to delamination at the bulk polymer/

SWNT interface (Figure 1b).

On the other hand, when interfacial crystallization is

present (Figure 5b), the interphase regions consisting of the

compact PVA crystal structure (based on the WAXD data)

provides closer and more complete contacts with SWNT

that may allow for more efficient stress transfer. The

presence of voids is most likely between the highly

crystalline interphase and the semi/crystalline bulk

polymer interface. However, the transition from PVA

crystalline interphase regions to semi-crystalline PVA

regions may be more natural (i.e., PVA chain in both

regions can interlock with each other) as compared to the

transition from SWNT surface to the polymer chains due to

the different energy states of PVAand SWNT. Therefore, the

voids for the latter case may be smaller than the former

case, and result in fiber failure between the crystalline

interphase and bulk/polymer interface (Figure 1d).

Based on the X-ray and SEM evidence, it is clear that in

addition to presence of bulk-PVA the interphase-PVA is also

predominant throughout the F-B fiber. To determine the

effect of this templated polymer structure on themechani-

cal properties in the composite, additional theoretical

analysis for the modulus of the F-B PVA/SWNT fiber is

performed. This is done by modifying Equation 2. The

theoretical contribution of the interphase-PVA is deter-

mined using

: 10.100

H & Co

e nu

Yc ¼ VPVAYPVA þ V interphaseY interphase ð8Þ

where Vinterphase is the volume fraction of the PVA

interphase in the vicinity of the SWNT. YPVA and Yinterphaseare the Young’s modulus of the PVA control fibers

(i.e., bulk PVA) and fully crystalline (i.e., interphase PVA),

respectively. The volume fraction of the interphase-PVA is

calculated using the crystal size obtained fromWAXD and

SEM data assuming the each SWNT bundle is uniformly

coated by interphase PVA, and was determined to be

�38.32%. Yinterphase was taken to be 255 GPa, and this

value is based on experimental X-ray analysis for fully

crystalline PVA along the axial direction.[41] X-ray analysis

in this work confirms that the interphase PVA is fully

crystalline and oriented in the fiber. To determine the

contribution of the interphase PVA, the SWNT bundle

is considered tobeavoid (Figure6a). Therefore, themodulus

contribution of the interphase PVA is calculated to be

�97.7 GPa. Using this calculation, the overall composite

modulus is estimated to be �109.7 GPa. This is below

the real composite average modulus of �119 GPa. For this

reason, it is determined that in addition to role of the SWNT

as a template for PVA interphase crystallization, the CNT

also play a role in reinforcing the PVA in the composite.

2/mame.201300025

. KGaA, Weinheim www.MaterialsViews.com

mbers, use DOI for citation !!

Figure 6. (a) Schematic showing the SWNT bundle as a void, wherethis structure is used to determine the effective moduluscontribution for the interphase-PVA in the F-B fibers. (b) Curvesshowing the effective modulus of SWNT as a function of orientationfactor (ho) taking into account the interphase-PVA in the F-B fiber.A twofold increase of stress transfer from the PVA matrix to theSWNT is observed in the F-B fibers. (c) A bar chart showing theeffective modulus contributions from each component present inthe control PVA, and F-A and F-B PVA/SWNT fibers.


www.mme-journal.de




Equation 2 is further modified to now take into account

the potential effect of the interphase PVA (see Equation 9).

In this way, a more accurate prediction of the effective

modulus reinforcement of the SWNT bundle in the F-B

composite is found. Equation 9 is given by:

I: 10.10

bH & Co

T the

Yc ¼ ðh0h1YSWNT�eff2 � YPVA � Y interphaseÞVSWNT

þ VPVAYPVA þ V interphaseY interphase ð9Þ

where YSWNT-eff2 is the new effective Young’s modulus of

the SWNTafter taking the interphase PVAcontribution into

account. Based on this analysis, the effective modulus for

the SWNT ranges from 2293.1 to 434.1 GPa for randomly

(ho¼ 1/5) to fully aligned (ho¼ 1) tubes, respectively

(Figure 6b). It should be noted that by including the

interphase contribution the approximation for the modu-

lus of the SWNT ismore in linewith othermeasured values

for SWNT bundles.[42,43] It should also be noted that the

effective modulus of the SWNT in the F-B fiber is still

significantly higher than what was observed for the F-A

fibers (no interphase; i.e., 434.1 GPa as compared to 201.5

GPa at ho¼ 1).

Figure 6c shows the effective modulus contributions

from each component present in the control, F-A, and F-B

fibers. It is clear that the presence of the SWNT templated

interphase-PVA significantly affects the fiber properties,

and results in an increase of stress transfer from thematrix

to the filler. Although experimental analysis for the

mechanical contribution from the polymer interphase is

currently being pursued to fully examine the PVA/SWNT

interfacial strengthening mechanism, the mechanical

testing results, crystal structure analysis, theoretical

analysis, and morphology visualization all suggest that

the formation of a polymer interphase in the vicinity of

SWNT plays a very important role in the improvement of

the composite fiber mechanical properties.

4. Conclusion

PVA/SWNT composite fibers with 10 wt% loading were

produced with a unique combination of high-performance

properties (tensile strength up to 4.9 GPa, modulus of 128

GPa, and toughness of 202 J � g�1) by controlling polymer

crystallization during gel-spinning under shear flow. As a

result, a new mechanism for PVA crystallization (001)

crystallization plane is observed of the interphase polymer.

A PVA unit cell is proposed for the predominant (001)

crystallization plane based on both real and simulated

WAXD data. The hypothetical unit cell suggests a closer

distance and stronger hydrogen bonding between two

adjacent PVA chains, which may also contribute to the

improvement of the fiber performance. The calculations of

the effective Young’s modulus and tensile strength are

02/mame.201300025

. KGaA, Weinheim 9

final page numbers, use DOI for citation !! R

www.mme-journal.de

J. Meng et al.

10

REa

much larger for the PVA/SWNT fibers which exhibit

interphase structure. Creation of the polymer interphase

region is a key component for improving stress-transfer,

where this clearly observed in the F-B fibers as compared to

the F-A (no interphase) fibers.

Acknowledgements: The authors thank the Air Force Office ofScientific Research (AFOSR FA 9950-11-1-0153) for financialsupport of this research work.

Received: January 22, 2013; Revised: March 29, 2013; Publishedonline: DOI: 10.1002/mame.201300025

Keywords: carbon nanotubes; fibers; interfaces; nanocomposites;poly(vinyl alcohol)

[1] M. Zhang, K. R. Atkinson, R. H. Baughman, Science 2004, 306,1358.

[2] X. B. Zhang, K. L. Jiang, C. Teng, P. Liu, L. Zhang, J. Kong, T. H.Zhang, Q. Q. Li, S. S. Fan, Adv. Mater. 2006, 18, 1505.

[3] K. Young, F. M. Blighe, J. J. Vilatela, A. H. Windle, I. A. Kinloch,L. B. Deng, R. J. Young, J. N. Coleman, ACS Nano 2010, 4, 6989.

[4] L. Y. Li, I. A. Kinloch, A. H. Windle, Science 2004, 304, 276.[5] J. M. Razal, J. N. Coleman, E. Munoz, B. Lund, Y. Gogotsi, H. Ye,

S. Collins, A. B. Dalton, R. H. Baughman, Adv. Funct. Mater.2007, 17, 2918.

[6] K. Liu, Y. H. Sun, R. F. Zhou, H. Y. Zhu, J. P. Wang, L. Liu, S. S. Fan,K. L. Jiang, Nanotechnology 2010, 21, 045708.

[7] A. B. Dalton, S. Collins, J. Razal, E. Munoz, V. H. Ebron, B. G.Kim, J. N. Coleman, J. P. Ferraris, R. H. Baughman, J. Mater.Chem. 2004, 14, 1.

[8] X. H. Zhong, Y. L. Li, Y. K. Liu, X. H. Qiao, Y. Feng, J. Liang, J. Jin, L.Zhu, F. Hou, J. Y. Li, Adv. Mater. 2010, 22, 692.

[9] S. Boncel, R. M. Sundaram, A. H. Windle, K. Koziol, ACS Nano2011, 5, 9339.

[10] K. Koziol, J. Vilatela, A. Moisala, M. Motta, P. Cunniff,M. Sennett, A. Windle, Science 2007, 318, 1892.

[11] C. D. Tran, S. Lucas, D. G. Phillips, L. K. Randeniya, R. H.Baughman, T. Tran-Cong, Nanotechnology 2011, 22, 145302.

[12] B. Vigolo, A. Penicaud, C. Coulon, C. Sauder, R. Pailler, C. Journet,P. Bernier, P. Poulin, Science 2000, 290, 1331.

[13] T. W. Chou, L. M. Gao, E. T. Thostenson, Z. G. Zhang, J. H. Byun,Compos. Sci. Technol. 2010, 70, 1.

[14] M. L. Minus, H. G. Chae, S. Kumar, Macromol. Chem. Phys.2009, 210, 1799.




[15] A. B. Dalton, S. Collins, E. Munoz, J. M. Razal, V. H. Ebron, J. P.Ferraris, J. N. Coleman, B. G. Kim, R. H. Baughman, Nature2003, 423, 703.

[16] J. N. Coleman, U. Khan, W. J. Blau, Y. K. Gun’ko, Carbon 2006,44, 1624.

[17] J. N. Coleman, M. Cadek, R. Blake, V. Nicolosi, K. P. Ryan,C. Belton, A. Fonseca, J. B. Nagy, Y. K. Gun’ko, W. J. Blau, Adv.Funct. Mater. 2004, 14, 791.

[18] K. P. Ryan, M. Cadek, A. Drury, M. Ruether, W. J. Blau, J. N.Coleman, Fuller. Nanotub. Carbon Nanostruct. 2005, 13, 431.

[19] C. Y. Li, L. Y. Li, W. W. Cai, S. L. Kodjie, Adv. Mater. 2005, 17,1198.

[20] K. P. Ryan, S. M. Lipson, A. Drury, M. Cadek, M. Ruether, S. M.O’Flaherty, V. Barron, B. McCarthy, H. J. Byrne, W. J. Blau,Chem. Phys. Lett. 2004, 391, 329.

[21] H. G. Chae, M. L. Minus, S. Kumar, Polymer 2006, 47, 3494.[22] M. L. Minus, H. G. Chae, S. Kumar, Polymer 2006, 47, 3705.[23] S. Kanagaraj, F. R. Varanda, T. V. Zhil’tsova, M. S. A. Oliveira,

J. A. O. Sim̌es, Compos. Sci. Technol. 2007, 67, 3071.[24] J. N. Coleman, M. S. Ferreira, Appl. Phys. Lett. 2004, 84, 798.[25] W. Liu, X. H. Zhang, G. Xu, P. D. Bradford, X. Wang, H. B. Zhao,

Y. Y. Zhang, Q. X. Jia, F. G. Yuan, Q.W. Li, Carbon 2011, 49, 4786.[26] M. Naebe, T. Lin, W. Tian, L. M. Dai, X. G. Wang,

Nanotechnology 2007, 18, 225605.[27] X. Xu, A. J. Uddin, K. Aoki, Y. Gotoh, T. Saito, M. Yumura,

Carbon 2010, 48, 1977.[28] L. B. Deng, S. J. Eichhorn, C. C. Kao, R. J. Young,ACS Appl. Mater.

Interfaces 2011, 3, 433.[29] K. K. H. Wong, M. Zinke-Allmang, J. L. Hutter, S. Hrapovic, J. H.

T. Luong, W. Wan, Carbon 2009, 47, 2571.[30] F. Vollrath, B. Madsen, Z. Z. Shao, Proc. R. Soc. Lond. Ser. B Biol.

Sci. 2001, 268, 2339.[31] G. P. Carman, K. L. Reifsnider, Compos. Sci. Technol. 1992, 43,

137.[32] Y. Zhang, M. L. Minus, Conference Presentation at ACS Spring

Meeting 2012, PMSE (429).[33] M. F. Yu, B. S. Files, S. Arepalli, R. S. Ruoff, Phys. Rev. Lett. 2000,

84, 5552.[34] M. M. J. Treacy, T. W. Ebbesen, J. M. Gibson, Nature 1996, 381,

678.[35] T. Natsuki, K. Tantrakarn, M. Endo, Carbon 2004, 42, 39.[36] R. S. Ruoff, D. Qian, W. K. Liu, C. R. Phys. 2003, 4, 993.[37] T. Belytschko, S. P. Xiao, G. C. Schatz, R. S. Ruoff, Phys. Rev. B

2002, 65, 235430.[38] P. Scherrer, Nachr. Ges. Wiss. G€ottingen 1918, 26, 98.[39] C. W. Bunn, Nature 1948, 161, 929.[40] I. D. Brown, Acta Crystallogr. A 1976, 32, 24.[41] I. Sakurada, T. Ito, K. Nakamae, J. Polym. Sci. C 1966, 15, 75.[42] C. H. Ke, M. Zheng, I. T. Bae, J. Appl. Phys. 2010, 107, 104305.[43] J. P. Lu, Phys. Rev. Lett. 1997, 79, 1297.

I: 10.1002/mame.201300025

H & Co. KGaA, Weinheim www.MaterialsViews.com

e numbers, use DOI for citation !!

forming crystalline polymer-nano interphase structures for high-modulus and high- tensile/strength...

Documents