Machine direction orientation of high density polyethylene (HDPE):Barrier and optical properties
Tirtha Chatterjee a, *, Rajen Patel b, John Garnett IV b, 1, Rajesh Paradkar c, Shouren Ge d,Lizhi Liu d, Kenneth T. Forziati Jr. e, Nik Shah e
a Analytical Sciences, The Dow Chemical Company, Midland, MI 48667, USAb Performance Packaging, The Dow Chemical Company, Freeport, TX 77541, USAc Analytical Technology Center, The Dow Chemical Company, Freeport, TX 77541, USAd Analytical Sciences, The Dow Chemical Company, Freeport, TX 77541, USAe Parkinson Technologies Inc., Woonsocket, RI 02895, USA
a r t i c l e i n f o
Article history:Received 9 April 2014Received in revised form30 May 2014Accepted 8 June 2014Available online 16 June 2014
Keywords:HDPEBarrier propertiesOptical properties
* Corresponding author.E-mail address: [email protected] (T. Chatterje
1 Present address: RQ&I Packaging Development68102, USA.
In this study, blown films were produced from three different high density polyethylene (HDPE) resins.Further, these blown films were oriented along the machine direction using either traditional machinedirection orientation (MDO) or compression roll drawing (CRD MDO) technology. The stretched/orientedfilms were compared for their water vapor transport and optical properties as a function of the extent oforientation (i.e., draw ratio, DR). In addition to that, these properties were directly compared betweenthe oriented and unoriented (blown) films of the same thickness. The present study revealed that thewater vapor transmission rate (WVTR) as a function of draw ratio (DR) went through a maxima. Detailedmorphological characterization showed that initially, at low DR, stretching resulted in lamellar organi-zation (along the MD). Consequently, the tortuosity in the diffusion path decreased and the overall WVTRincreased. However, beyond a critical DR, lamellar crystals transformed to microfibrillar crystals. Duringthis process the amorphous content of the HDPE chains decreased and geometric redistribution of theamorphous phase occurred along with some orientation. As a combination of all these effects, the WVTRwas found to decrease beyond the critical DR. Unlike transport properties, optical properties mono-tonically improved as a function of DR. Further, for both, the barrier and optical properties, the absolutevalues were found to be a strong function of the resin physical properties. Finally, compression rolldrawing technology (CRD MDO) utilizes a patented process that simultaneously applies a stretching andcompressive force to orient polymer chains homogeneously without any tiger striping (uneven defor-mation). Consequently, CRD MDO stretched samples showed a lower surface roughness compared totheir traditional MDO stretched counterpart. Therefore, for the same resin and DR, CRD MDO stretchedsamples displayed markedly better barrier properties and optics, especially lower total haze and highergloss.
Polymers are widely used in modern food and beverage pack-aging industries. One of the most essential requirements for pack-aging applications is high barrier properties. Additionally, highbarrier polymers are relevant to other industrial applications
e)., ConAgra Food, Omaha, NE
including gas separation membrane, encapsulation of flexibleelectronic devices, and packaging of healthcare and pharmaceuticalgoods among others [1e5]. It is well known that the orientation ofpolymers is responsible for enhancing many properties, specificallybarrier, mechanical and optical. At the same time, poor puncturestrength and tear along the direction of orientation somewhatrestrict their commercial uses in the packaging industry, especiallyfor uniaxially oriented polymers. Nevertheless, improved barrierproperties due to orientation offer economically and environmen-tally viable solutions such that a reduction in packaging materialthickness can take place along with the extension of shelf-lifewithout compromising food quality and safety [5,6]. The deciding
T. Chatterjee et al. / Polymer 55 (2014) 4102e4115 4103
factors that determine high barrier properties in polymers are theirchemical structure and morphology. The nature of the permeantmolecule plays an important role in polymer selection. Forexample, inclusion of increasingly polar substituent in the ethylenicbackbone unit increases the polymer cohesive energy density andreduces the oxygen transport rate significantly. Similarly, polymerswith high cohesive energy density are expected to be a good barrierto non-polar aromatic molecules. In contrast, moisture/water vapormolecules, being strongly polar, can pass through polymers withhigh cohesive energy density relatively easily [5,6].
The other significant contributor to barrier properties is thehierarchical polymer molecular architecture. For semi-crystallinepolymers the polymer crystals act as impermeable barriers andthe small molecules diffuse through the amorphous regions. Thereare several mechanisms reported to explain the improvement inbarrier properties for oriented polymers. One proposed mechanismis that, orientation of polymer chains result in a decrease in freevolume of the amorphous region which causes a reduction istransport rate [7]. An alternative argument is orientation rendershigher tortuosity (through fractionation and crystal lamellaealignment) to the diffusion path (perpendicular to the alignment)and hence, the overall barrier property improves [8,9]. Anothercontention is that orientation induces crystallization and theoverall crystallinity increases. Moreover, barrier property improvesdue to microfibrillar crystal formation and denser packing in theamorphous regions brought about by the orientation [10,11].Consequently, understanding the processingestructure relation fororiented polymer films is fundamental to understand their barrierand other properties.
High water vapor transmission barrier (or low water vaportransmission rate, WVTR) is one of the most unique properties ofhigh density polyethylene (HDPE). This is one of the major reasonsbehind using HDPE films in food packaging applications. Stretchingof the HDPE chains transforms randomly distributed spheruliticcrystals into anisotropic fibrillar morphology which is the reasonbehind the improvement in barrier properties [12e14]. Most ofthese stretching induced morphology transition and correspondingbarrier property studies have been performed under biaxial ori-entations, where a predominant deformation is applied in themachine direction (MD) with simultaneous stretching in thetransverse direction (TD) [15e20]. Peterlin and coworkers studiedthe morphology and barrier properties of uniaxially stretchedHDPE films [11,21]. However, these films were uniaxially orientedusing a tensile testing instrument which is not practiced in industryfor large scale production. Additionally, this group mostly focusedon transportation of small aromatic molecules through orientedfilms.
Commercially, uniaxial drawing of the polymer films is per-formed using machine direction orientation (MDO) technology. Inthe MDO process, drawing takes place in a semi-solid state (pro-cessing temperature is above the glass transition temperature butbelow the polymer melting temperature) through tensile forcegenerated by two rollers which are rotating in opposite directionsat increasing surface speeds. It is the differential in linear surfacespeeds, not the direction of rotation that stretches the film. TheMDO unit can operate in-line or off-line with extrusion and iscompatible with both the blown and cast film productions. Addi-tionally, orientation makes the film flat and stiff and as a resultMDO films run many times faster in downstream converting op-erations [22]. Even after being used in the industry for over 40 years[22,23], there are only a few studies available on the morphology[24e27] of the HDPE films oriented through the MDO process andthe consequences on the overall barrier properties [8,10]. Paulosand Thomas [10] reported that both the water vapor and oxygentransmission rates decrease monotonically with increasing film
draw ratio. This is attributed to the increased orientation of thecrystalline and amorphous phases simultaneously and the micro-fibrillar crystal structure formation brought about by the drawingprocess. In contrast, Breese and Beaucage [8] have found that thewater vapor and oxygen transmission rates (WVTR and OTR,respectively), as a function of HDPE draw ratio (DR), go through amaximum around a critical draw ratio ~6e7X. It implies that at lowDR (<6e7X), the transport rate through oriented HDPE films in-creases compared to their un-oriented counterparts. The reasonbehind this trend is a transformation from the crystal lamellaeorientation with lamellae normal parallel to the draw directionbelow the critical draw ratio to chain orientation (where crystallamellae organize themselves in a zigzag format with the drawdirection) above the critical draw ratio. The zigzag orientation ofthe lamellae results in an increase in the tortuosity of the diffusionpath which subsequently reduces the WVTR. But, with someexception [8], none of these studies investigate the role of startingresin physical properties on the final (barrier and optical) proper-ties of the MDO processed HDPE films.
In this context, we have systematically studied the barrier(water vapor transport) and optical (clarity, gloss and total haze)properties of MDO processed HDPE films prepared from threedifferent resins which varied in their physical properties (density,molecular weight, molecular weight distribution, etc.). Besides thetraditional MDO, a patented process called compression rolldrawing technology (CRD MDO) has also been utilized whichenabled us to orient the HDPE films uniformly below their naturaldraw ratio. The focus of our work was to understand the influencesof (a) uniaxial (MD) orientation and (b) resin physical properties onthe final barrier and optical properties of the oriented films. Thestarting blown films made from three different HDPE resins werestretched to a wide range of draw ratio (DR, ~3Xe10X) using eitherthe traditional MDO or CRD MDO technologies. To understand theinfluence of stretching, the stretched film properties were directlycompared with corresponding blown films of the same thickness(control samples). Additionally, in order to understand the influ-ence of resin physical properties, the stretched film properties werestudied as a function of extent of orientation (i.e., DR). This studyshows that for any DR and resin type, the optical propertiesimprove upon stretching. In contrast, at low DR barrier propertybecomes worse (i.e., WVTR increases with increasing DR) and itimproves only above a critical DR. We also found that the absolutevalues of the barrier and optical properties are more influenced bythe resin selection than the stretching operation. Finally, all theseresults are rationalized from a processingestructureepropertypoint of view.
1.1. Traditional machine direction orientation (MDO) andcompression roll drawing (CRD MDO)
The MDO process consists of four major steps: (a) pre-heating,(b) drawing, (c) annealing, and (d) cooling [28]. The processingflow sheet has been presented in Fig. 1. In the beginning, the pre-heating roller heats the polymer film to a temperature that is~5e7 �C below their differential scanning calorimetry (DSC) peakmelting temperature. This pre-heated film is fed into a slowdrawing roll with nip roller, which has the same rolling speed as thepre-heating rollers. The stretching takes place in a semi-solid statethrough tensile force generated by two rollers in the drawing zonewhich are rotating in the opposite directions at increasing surfacespeeds and effectively orienting the film on a continuous basis inthe MD. The ratio of the speeds of these two rollers controls thedraw ratio (DR). The draw ratio can be defined as DR ¼ Thickness ofthe un-stretched film/thickness of the stretched film z Speed of thefast drawing roll/speed of the slow drawing roll. The distance
Table 1Properties of three Dow resins used in this study.
a Target values. Melt index data weremeasured at 2.16 kg. Measured I21 value forResin-3 is 9.5.
b Gel permeation chromatography data. Provided in the Supporting information.c DSC data. Reported data are taken from the 2nd heating cycle.
Preheat 1
Preheat 2(Slow Draw)
(Fast Draw)
Annealing
Cooling
Primary Draw Gap
SecondaryDraw Gap
Infeed
Outfeed
F.D.
S.D.
Fig. 1. MDO processing steps. Note that the only difference between the traditionalMDO and CRD MDO processes is the gap between the slow-draw and fast-draw rollersin the drawing zone (primary draw gap). In traditional MDO process the gap is muchlarger in contrast to negligible gap maintained in the CRD MDO process.
T. Chatterjee et al. / Polymer 55 (2014) 4102e41154104
between these two drawing rollers is kept as low as possible tominimize neck-in in the transverse direction with a typical roller-to-roller gap between being 50e80 mm. The oriented film isannealed at an elevated temperature for a period of time to allowstress relaxation and hence, improves dimensional stability. Thetemperature of the annealing thermal roller is ~5e10 �C below theDSC onset melting temperature with typical annealing time ~1e5 s.Annealing at this temperature reduces the amount of filmshrinkage during heat sealing. Finally, the films are cooled bycontacting the rollers that are at ambient temperature. Generally,chilled water passes through rollers that cool the film to ambienttemperature.
One of the major challenges of the traditional MDO technologyarises from inherent polymer properties. A conventional cold-drawing of polymer forms ‘necking’ (in thickness) due to non-uniform distribution of strains along the length of the material.Only above a critical draw ratio, referred to as the natural draw ratio,can uniform drawing/stretching along the specimen length berealized. In the non-uniform drawing zone, film thickness differ-ence leads to optical difference between the strain hardened parts(thinner film) and the non-strain hardened parts (thicker film). Thisis commonly referred to as Tiger Striping due to the visual resem-blance to stripes that this non-uniform film stretching produces.This shortcoming prevents the use of traditional MDO technologyin stretching any polymer film below their natural draw ratio. Thischallenge can be overcome using the compression roll drawing(CRD MDO), a patented technology [29e31]. In contrast to tradi-tional MDO process in which the gap between the slow-draw andfast-draw rollers (where stretching is accomplished) is typically setgreater than the thickness of the materials being stretched, the CRDMDO technology maintains a minimal gap (rollers separated onlyby the film thickness). Because of this, traditional MDO allows thematerial to contract more freely in the lateral (cross/transverse)direction than the CRD MDO technology. Due to limited lateralcontraction, in CRD MDO, the stretching rollers apply a simulta-neous compressive force in addition to the MD drawing force.Under these circumstances, the perfect cylindrical symmetryaround the stretching direction is not obtained which typicallyrenders some orientation in the transverse direction [32]. However,
in the CRD MDO process, the applied tensile force [ compressiveforce and therefore, the films are largely stretched uniaxially. Weclearly mention that unlike traditional biaxial stretching (such as ina Tenter frame) in CRD MDO process there is no substantial strainapplied in the transverse direction. Further, the CRDMDO stretchedfilm mechanical properties (e.g. tensile, elongation) were found tobe largely improved in MD and not in TD. For one of the repre-sentative resins, for which the traditional MDO and CRD MDOprocessing could be performed at the same DR (4X), these me-chanical properties (MD and TD) are presented in Supportinginformation and compared with a non-stretched blown filmproperties. These data suggest that the CRD MDO process essen-tially orients the film uniaxially (MD) and for the same DR theextent of orientation between the traditional MDO and CRD MDOprocessed films are comparable. Therefore, in this study, in terms ofuniaxial orientation, no distinction was considered between thetraditional MDO and CRD MDO processed films. Further, since thedrawing is performed at an elevated temperature and the gap be-tween the rollers is essentially the film thickness, the CRD MDOprocess is effective for low DR operations (or when the drawn filmthickness is at least 1mil or 0.00100). The present study utilized boththe traditional MDO and CRD MDO processes to orient the HDPEfilms. With one exception, the films were stretched using the CRDMDO process below their natural draw ratio (low DR regime) andtraditional MDO process above their natural draw ratio (high DRregime). Finally, hereafter, for simplicity, the traditional MDO and CRDMDO processes will be referred to as the MDO and CRD processes,respectively.
2. Materials and methods
2.1. HDPE resins and their physical properties
Three different resins, all manufactured by the Dow ChemicalCompany, were selected for this study. Physical and thermalproperties of these resins have been listed in Table 1. Resin-2 is anucleated HDPE. Resin-3 contains hexene comonomer. The desir-able resin properties for the MDO technology are (a) high overallmolecular weight (Mw) and (b) broad bimodal or multimodal mo-lecular weight distribution.
2.2. Uniaxial drawing/stretching of the films
For all three resins monolayer blown films of 8 mil(1 mil ¼ 0.00100) thickness were prepared at a Dow Freeport, TXfacility. These films will be referred to as ‘starting films’ from hereon throughout the manuscript since these were the starting ma-terials for MDO/CRD operations. Gauge uniformity and initial/un-stretched film quality are very important factors. Presence ofwrinkles, gels, holes, and gauge non-uniformity, etc., are detri-mental to a successful machine direction stretching operation. Thetypical blow up ratio (BUR) maintained for starting film productions
Table 3Operating temperatures of different rollers in MDO/CRD processing.
a DSC data. Reported data are taken from the 2nd heating cycle.
T. Chatterjee et al. / Polymer 55 (2014) 4102e4115 4105
was 2.65 for which the resulting layflat was 33.400. The frost lineheight was 2500. Additionally, 1.0, 1.5 and 2.0-mil thickness blownfilms were prepared from each resin. These films were notstretched (using MDO/CRD process) and were used for comparisonof properties with stretched film of the same thickness. These filmswill be referred to as ‘control films’ throughout the remainingmanuscript. It is worth noting here that for thick blown films thesurface morphology inside and outside of the bubble is not oftenthe same due to inherent poor heat transfer properties of polymer.However, such a morphology gradient was not expected to besignificant for 8 mil thick films and no further studies wereconducted.
All the MDO and CRD processing, that has been reported here,were performed at the Marshall and Williams Plastics (MW) BiaxLab, Parkinson Technologies Inc., 100 Goldstein Drive, Woonsocket,Rhode Island 02895, USA. To avoid non-uniform stretching, at lowDR, the compression roll drawing (CRD MDO) was utilized asnecessary. A detailed list of samples prepared using the traditionalMDO and/or CRD has been reported in Table 2.
The draw ratios reported in Table 2 are the target values basedon the roller speed ratios. The exact draw ratios were calculatedusing the film thicknesses (starting film and drawn films) measuredduring the property measurements (as per ASTM protocol). ForResin-1, the films can be drawn using the CRD process for DR �5Xand the lowest DR available for uniform stretching (without tigerstriping) using the MDO process was 7.5X. Similarly, for Resin-2,films with DR �7X were drawn using the CRD technology andthe lowest DR available for MDO stretching without tiger stripingwas 10X. Therefore, in both cases, there was a DR gap which wasnot accessible through either technology. For Resin-3, the MDOoperation produced uniformly stretched film (without any tigerstriping) at DR as low as 3X. Presence of high MW fraction in Resin-3 (Fig. S2, Supporting information) leads to high concentration oftie chains which presumably enable uniform stretching withouttiger striping even at low draw ratios. The maximum DR obtainedfor Resin-3 was 8X. Additionally, for DR ¼ 3X and 4X, orientedResin-3 films were prepared using the CRD process as well. Thesesamples provided direct comparison of drawn film propertiesprepared through the traditional MDO and CRD MDO processes,respectively. All the processing temperatures for the MDO/CRDdrawing employed in this study are presented in Table 3
2.3. Measurement of barrier and optical properties
The water vapor transmission rate (WVTR) was measured usinga Mocon Permatran-W Model 3/33 instrument. Each sample wastested in duplicate, subjected to 90% relative humidity (RH), at38 �C and the average value was reported. Permatran-W 3/33 in-strument follows ASTM F1249.
Optical properties namely total haze, clarity and 45� gloss weremeasured at our in-house characterization laboratory at The DowChemical Company, Freeport, TX. All the reported optical properties
Table 2Different uniaxially drawn samples prepared using the traditional MDO and/or CRDMDO process.
Resin MDO drawratio
CRD MDOdraw ratio
Comments
Resin-1 7.5X, 8X, 8.5X 3Xe5X DR range >5X and <7.5X was inaccessiblethrough either technology
Resin-2 10X, 10.5X 3Xe7X DR range >7X and <10X was inaccessiblethrough either technology
Resin-3 3Xe8X 3X, 4X Can be drawn using conventional MDO fordraw ratio as low as 3X
data were based on measurements performed with five differentspecimens per sample. Total haze and clarity were measured usingthe Gardner Haze Gard Plus™ instrument. It should be noted thatthis instrument is designed to measure haze following the ASTM D1003 protocol [33] (measure of light intensity that is scatteredmorethan 2.5� from the axis of incident light). A high haze value rep-resents poor optical properties.
Clarity measurement performed using the Gardner Haze GardPlus™ in clarity mode does not follow the ASTM D 1746 method-ology [34]. Technically, clarity data should correlate with the loss ofresolution of an object viewed through the film. As per ASTM D1746 protocol, the experiment geometry for clarity/transparencymeasurement should use an aperture appropriate to exclude lightscattered by more than 0.1� [34]. Clarity values in the Haze GardPlus™ instrument are calculated using the intensities received bythe ring and center photosensors. The ring photosensor is designedfor the haze measurement (sensitive to light scattered > 2.5�).However, the center photosensor is toowide (angular width 1.4�) toexclude light that is scattered more than 0.1� from the axis ofincident light. Therefore, clarity measurements performed by theHaze Gard Plus™ depend on both the attenuation of the main beam(detected by the center sensor) and 3.2e4.0� angular range scat-tered light reaching the ring sensor [35]. In this case a value of 100%indicates that no light beam is detected by the ring photosensor(i.e., no light scattered within the angular range 3.2e4.0�). A zeroclarity value indicates an equal amount of light intensity detectedby both the ring and center photosensors. Due to these measure-ment limitations, significant differences are generally observedbetween the clarity datameasured by the Gardner Haze Gard Plus™and Zebedee CL-100 clarity meter (this instrument is in compliancewith ASTM D 1746 protocol) [35]. Nevertheless, considering theease of measurement the Gardner Haze Gard Plus™ clarity data arewidely used in the industrial laboratories and also reported inpublications [24,35e39]. We caution the readers that the clarityvalues reported here should be treated as a transparency metricand the relative comparison of values are more meaningful torepresent the film optical properties than their absolutemagnitude.
45� gloss was measured using a BYK-Gardner micro-gloss 45�
instrument following the ASTM D 2475 protocol [40]. A set ofparallel beams strikes the surface at an incident angle q (here 45�)and the specular reflection from the film is recorded in thephotodetector. The sample intensity is normalized with respect to astandard measured under the identical condition and expressed as%. Since gloss of a film is a strong surface property, carewas taken toobtain a flat surface while mounting the sample and the samestandard was used for the normalization procedure.
2.4. Film morphology and characterization of the oriented state
2.4.1. Differential scanning calorimetry (DSC)DSC measurements were conducted using a DSC Q1000 V9.9
Build 303 calorimeter instrument. To obtain the peak meltingtemperature of the resins, a heatecooleheat cycle was used.Initially the sample was quickly heated to 180 �C and held at thattemperature for 5 min to remove all previous thermal history. The
T. Chatterjee et al. / Polymer 55 (2014) 4102e41154106
peak melting temperature of the HDPE is ~130e135 �C, and thesample was expected to be completely molten at 180 �C. Further,after the isothermal hold at the melt state, the sample was cooledto �40 �C with a cooling rate of 10 �C/min. The sample was held atthe low temperature (�40 �C) for 5 min and was subsequentlyheated to 150 �C using a heating rate of 10 �C/min. All the HDPEresin melting temperatures and enthalpy of fusion data have beenreported using the second melting curve (Tables 1 and 3).
For uniaxially oriented samples and starting films, specimenswere heated to 150 �C using a heating rate of 10 �C/min. All thereported values were collected from the first heating data.
2.4.2. Wide-angle X-ray diffraction (WAXD)Samples were analyzed using Bruker-AXS GADDS systemwith a
HiStar area detector. A 0.3 mm beam size was used for the X-rayexperiment. Sample to detector distance was 6 cm. Cu1a radiationwith wavelength of 1.54 Å was used for the measurements. Theinstrument was calibrated with Corundum standard (Al2O3). Datawere collected in the transmission mode with the detector posi-tioned at 0� and covering the angular range (2q) of data collectionfrom 3 to 33�. The data were analyzed using the MDI JADE® XRDPattern Processing Software.
2.4.3. Light scattering (LS)The configuration for the small-angle laser light scattering
(SALLS) consisted of a Compass 315 laser with a wavelengthl ¼ 538 nm, two polarized lens adjusted for the HV (horizontal)configurations, velum paper for image capture, and a Pixels cameraequipped with a Pentax 50 mm Macro F2.8 camera lens. The Pixelscamera was positioned and focused to capture the SALLS image onthe velum. After adjusting the sample position, the sample wasremoved and replaced with a 3.3333 mm diffraction grating film forsystem calibration. The scattering images were analyzed using theSCATTER image analysis software.
2.4.4. Atomic force microscopy (AFM)The samples were mounted on the stage using double-sided
carbon tape and then blow-cleaned with Duster for AFM analysis.AFM images were captured at ambient temperature by using aVeeco (now Bruker) Icon AFM system with a Mikromasch probe.The probe had a spring constant of 40 N/m and a resonant fre-quency in the vicinity of 170 kHz. An imaging frequency of 0.5e2 Hzwas used with a set point ratio of ~0.8.
(a)
Fig. 2. (a) Water vapor transmission rate data obtained for the drawn films made of three divertical line divides the DR range into two zones: CRD MDO drawn (low DR or left side of thvertical line) zones. The above zones are applicable for Resin-1 and Resin-2 films only. All thesamples. All the WVTR data were corrected for the film thickness. Dotted lines (through da
2.4.5. Birefringence measurementRefractive index of free standing films was measured using a
Metricon 2010 Prism Coupler. The Prism Coupler utilized a632.8 nm HeeNe laser source. Refractive index of the prism usedfor the measurement was 1.9648. The refractive index rangemeasurable using this prism was 1.0422e1.6879 and was wellsuited for polyethylene samples. A detailed discussion of the theoryof measuring birefringence of free standing polymer films using thePrism Coupler can be found elsewhere [41e43]. The Prism Couplerallowed measurement of refractive index in the machine direction(MD), transverse direction (TD) and normal to the plane of the filmsurface.
In a polymer film, characterization of the three principalrefractive indices yields considerable information about theanisotropy of the film. Measuring three refractive indices allowsmeasurement of birefringence, i.e., a measure of the averageorientation, in all three directions of the film, in addition tomeasuring the average refractive index of the film. The averagerefractive index of the film is simply the average of the refractiveindices in the three principle directions of the film[Navg ¼ (Nx þ Ny þ Nz)/3]. Further, by definition: Dzy ¼ Nz � Ny,Dzx ¼ Nz � Nx and Dxy ¼ Nx � Ny, where, Nz is defined as therefractive index in the machine direction (MD) (sometimes alsodesignated as Nk), Ny is the refractive index perpendicular to themachine direction (transverse direction, TD) but in the same plane(i.e., N⊥) and finally, Nx is defined as the refractive index normal tothe sample surface (ND). The in-plane birefringence is thereforecalculated as (Nz � Ny or Nk � N⊥).
3. Results and discussion
3.1. Barrier properties
Uniaxial orientation of the HDPE films has significant effect ontheir barrier properties. The water vapor transmission rate (WVTR)data, measured at 38 �C and 90% RH, have been presented as afunction of draw ratio (DR) in Fig. 2a. In this plot, uniaxially orientedResin-1 and Resin-2 films at low DR (left side of the dotted verticalline) were stretched using the CRD process whereas, the high DRdata (right side of the dotted vertical line) were collected on filmsstretched using the MDO process. For Resin-3, all the WVTR datareported in this plot were collected on the MDO drawn films. As afunction of DR, the WVTR initially increased to reach a peak
(b)
fferent Dow resins. Values for the starting films are also shown (within the box). Dottede dotted vertical line) and traditional MDO drawn (high DR or right side of the dottedWVTR data reported for Resin-3 films were measured using the traditional MDO drawnta points) are only a guide to the eye. (b) Normalized WVTR as a function of DR.
T. Chatterjee et al. / Polymer 55 (2014) 4102e4115 4107
(maximum) followed by a decrease. This trend is similar to theWVTR data reported for uniaxially oriented HDPE films by Breeseand Beaucage [8]. For Resin-2 and -3 stretched films, the WVTRpeak appeared at DR ~7X and 4X, respectively. However, thismaximumwas not obvious for Resin-1. This was presumably due toa DR gap between 5X and 7.5X which was inaccessible througheither MDO or CRD processes. For this resin, the CRD drawn sam-ples (i.e., at low DR) exhibited higher WVTR compared to the MDOdrawn samples (i.e., at higher DR). Following this trend, for Resin-1,the WVTR maximum seemed to exist between the DR 5X and 7.5X.
Along with different stretched films, the WVTR data for thestarting films are also presented in Fig. 2a. In terms of barrierproperties, the starting films exhibit the following order Resin-3 < Resin-1 < Resin-2. Being a nucleated resin, Resin-2 films areexpected to demonstrate superior barrier properties due to higherfractional crystallinity (as evident from the density data and frac-tional crystallinity data presented later) and smaller crystal size. Onthe other hand, low crystal content (as evident from the densitydata and fractional crystallinity data presented later) and a verybroad MWD (Table 1) are presumably responsible for poor barrierproperties displayed by the Resin-3 samples. The focus of this workis to understand the role of orientation (MDO/CRD) on the barrierproperties of the stretched films. Therefore, the drawn films WVTRvalues were normalized using the corresponding starting filmWVTR values and are presented in Fig. 2b. The normalized WVTRvalues are independent of the starting film morphology andrepresent the influence of film orientation on barrier properties.The major observations from normalized WVTR data are: (a) thepeak/maximumWVTR ratio is ~1.5 for Resin-1 and Resin-3 whereas~2.3 for Resin-2 and (b) the minimum WVTR ratio (typicallyobserved at the highest DR) is ~0.7 for Resin-1 and -3 opposed to avalue of 0.9 for Resin-2. Therefore, while formost of the DRwindowthe Resin-2 films demonstrate the lowestWVTR values, the effect oforientation is relatively limited for these films. In contrast, at anyDR, Resin-3 films demonstrate the worst barrier property (i.e., thehighest WVTR values), however, orientation significantly improvesthe barrier properties compared to the corresponding starting film.
Further, for a given resin, the effect of orientation on overallbarrier properties has been compared with their control samples(blown films of the same thickness but without any post-processing/orientation) and presented in Fig. 3. For Resin-1 withfilm thickness of 2.0 and 1.5 mil, the WVTR values were found to besimilar for the control (non-drawn) and drawn (4X and 5X DR,respectively) samples. However, for 1-mil sample thickness, theMD-oriented sample showed better barrier properties or lowerWVTR. Therefore, for Resin-1, any substantial difference in theWVTR between the control and drawn sample was only realized athigh draw ratio (~1 mil film thickness or at DR ~8X). A similar
Fig. 3. Comparison of the WVTR between the MD drawn samples and control samples (blodifferent film thicknesses. For the drawn samples (MDO/CRD), the lower the thickness is trepresents an ascending order in DR. All samples were drawn from an initial 8-mil thick m
observation has been made for Resin-2 as well. In this case, for anintermediate DR (1.5-mil sample thickness), the WVTR of thedrawn sample was even higher than the control sample. But at ahigher DR (1-mil sample thickness), the drawn sample WVTR waslower than their un-oriented counterpart. In contrast, for Resin-3,MD orientation of the film resulted in WVTR reduction comparedto control films for all three thicknesses studied here. In conclusion,these data suggest that: (a) as a function of DR, the WVTR initiallyincreases and later decreases after going through a maximum, (b)the DR at which the WVTR maximum appears is a function of theresin type, and (c) due to stretching or orientation in the MD,improvement in overall barrier properties is realized for Resin-3 forall DRs while for Resin-1 and Resin-2 the benefits are realized onlyat high draw ratios (>7X).
3.2. Optical properties
Orientation of crystal lamellae in a preferred direction improvesthe polymer film optical properties. Previously Ajji and coworkershave reported a substantial decrease in haze and increase in clarityafter the MD orientation of HDPE films [24]. In Fig. 4, opticalproperties namely clarity, total haze and 45� gloss have been pre-sented as a function of draw ratio. In general, as the draw ratioincreased, the clarity and gloss of the film increased, whereas thetotal haze decreased. Due to experimental limitation associatedwith clarity measurement, it was hard to comment on visual ap-pearances of the Resin-1 and Resin-2 drawn films as a function ofDR. For Resin-3, it was safe to say that the MD stretching of the filmreduced some large scale heterogeneity (likely associated withcrystal lamellae orientation) as reflected in monotonic increase inclarity values as a function of DR. The clarity improvementobserved in Resin-3 drawn films was much greater than what wasexpected due to reduction in film thickness [35]. In addition to that,stretching of the Resin-3 films rendered significant improvement infilm resolution which was visually apparent. Unlike clarity, forResin-1 and Resin-2 stretched films, the 45� gloss initially increasedwith the DR and later decreased a little at very high DR. Thisdecrease at high DR is a surface effect. As mentioned in theExperimental section, light scattered in non-specular directionsreduces the measured gloss value. With increasing surface rough-ness (overall root mean square roughness), the light intensitytransmitted in the specular direction decreases exponentiallyresulting in a lower gloss value [44,45]. For Resin-1 and Resin-2, atvery high DR, the samples were stretched using the MDO tech-nology whereas at low DR samples were stretched through the CRDtechnology (Table 2). Due to the presence of a small amount ofcompressive force during the CRD operation, the mean surfaceroughness (of the oriented films) is expected to be lower for this
wn films of the same thickness but without any post-processing/orientation) for threehe higher is the draw ratio. Therefore, a descending order in thickness axis (abscissa)onolayer blown film. Note that the WVTR axis scales are different for different resins.
Fig. 4. Stretched film (a) clarity, (b) total haze and (c) 45� gloss as a function of draw ratio. Higher clarity, higher gloss and lower total haze are the desirable optical properties forpackaging applications. Note that the ordinate scales are different for different optical properties.
T. Chatterjee et al. / Polymer 55 (2014) 4102e41154108
technology compared to the MDO processed samples, as will bediscussed shortly. Likely due to similar reasons, the total hazeinitially monotonically decreasedwith increasing DR in Resin-1 andResin-2 stretched films and later increased for the MDO drawnfilms at high DR. In contrast, gloss and total haze data presentedhere for the Resin-3 films were measured using the MDO drawnsamples only. In this case, the gloss increased and the total hazedecreased monotonically as the DR increased, as expected. There-fore, it is concluded that uniaxial orientation of HDPE filmsimproved their optical properties. In cases where the surfaceroughness affects the property, the CRD processed samples aresuperior to the MDO processed samples. Finally, the loss of reso-lution increases (i.e., clarity decreases) when the distance betweenthe object and film is increased [44]. From optics points of view, thehaze of a film qualifies the loss of contrast (a high value indicatespoor optical properties) and the clarity measurement qualifies theloss of resolution (a low value indicates poor optical properties). Inmost of the packaging applications, the viewed objects are locatedclosed to the film and hence, from packaging point of view low haze(related to contrast) is a more desirable criteria than high clarity(related to resolution) for film optical performances.
Fig. 5. Comparison of optical properties total haze (top) and 45� gloss (bottom) between thlower DR yields thicker film. Therefore, a descending order in thickness axis (abscissa) repreCRD stretching. Ordinate-axis scales are different for different plots.
Direct comparison of optical properties (total haze and 45�
gloss) between uniaxially oriented films and their control sampleshave been presented in Fig. 5. The clarity comparison is not shownhere due to their measurement limitation. However, we mentionthat for Resin-1 and Resin-2, the Haze Gard Plus™measured clarityvalues of the drawn and control films were similar all three filmthicknesses (2.0, 1.5 and 1.0 mil) studied here. In contrast, for allthree thicknesses, a significant increase in clarity was observed forResin-3 drawn film compared to corresponding control film whichwas also in accordance with visual inspection. In general, Resin-3drawn film clarity was poorer than Resin-1 and Resin-2 controlsamples of the comparable thickness. This suggests that the choiceof resin is a more important criterion for the final clarity of the filmthan the stretching operation.
MD orientation significantly reduced the total haze of thestretched films compared to control samples for all three resins.Note that the control and drawn samples were of the same thick-ness and therefore these haze values were directly comparable.This suggests that the stretching operation possibly reduced surfaceroughness along with a decrease in orientation and/or densityfluctuations which led to lowering of the film haze. However, the
e drawn films and control samples of the same thickness. For the stretched samples, asents an ascending order in DR. Control samples were blown films without any MDO or
T. Chatterjee et al. / Polymer 55 (2014) 4102e4115 4109
total haze values were higher for Resin-3 stretched films comparedto the stretched films made from the other two resins (for the samefilm thickness) as illustrated in Figs. 4b and 5. Finally, 45� gloss isalso a function (albeit weak) of resin selection. Resin-3 control filmswere found to have substantially lower gloss value compared toResin-1 and Resin-2 control films of similar thickness. A substantialincrease in the % gloss value was observed for oriented samplescompared to the corresponding control blown film for all threeresins studied here. Again, for the oriented samples, the absolutegloss values were resin specific. The % gloss values were ~80 and~90 for Resin-1 and Resin-2 stretched films, respectively and almostinvariant to draw ratio. Contrary to that, for Resin-3 films, the %gloss values were quite small and even at the highest DR film (8X,film thickness ~1 mil) it reached a value of ~40 only. All these re-sults indicate that uniaxial orientation improves the optical prop-erties of the film compared to their non-oriented counterpart.However, the absolute values of the optical properties, such asclarity, gloss and total haze are more influenced by the resin se-lection than the stretching operation.
3.3. Crystallinity, orientation and morphology
Polymer structures are governed by processing conditions.Further, polymer properties are controlled by their hierarchicalstructure. Therefore, it is essential to understand themorphologicalchanges in the HDPE during the stretching operation to explainoverall barrier and optical properties of the films discussed so far.Strain-induced chain/lamellar orientation and crystallization arewell-documented in polymer literature [46e50]. DSC meltingtraces for Resin-1 starting film and a representative oriented sample(DR ¼ 5X) are presented in Fig. 6a. Similar data for Resin-2 and -3are reported in the Supporting information. For the MDO processedpolypropylene films, two distinct melting peaks are typicallyobserved [39,51,52]. This double melting phenomenon is rational-ized as the melting of the shish-kebab crystal structure where theshish melting temperature is higher than the kebab melting tem-perature [49]. In contrast, oriented HDPE crystals do not exhibitdual melting behavior which is consistent with our observation[15,24]. It is argued that under uniaxial strain, HDPE crystals arerelatively easily deformed to form fibrillar structures. Consistentwith this argument we observed an average 2.2 �C higher peakmelting temperature for the oriented films compared to the cor-responding un-oriented starting films (first heating cycle data). Thisimplies that stretching strongly affects the crystal morphology andthe averagemelting temperature of lamellar crystals are lower than
(a)
Fig. 6. (a) Representative DSC melting traces for Resin-1 starting film and a stretched sampfunction of DR for uniaxially oriented HDPE films made of three different resins. Values for
the fibrillar structure. Further, during the stretching process, straininduced crystallization occurs which results in an increase inpolymer crystal content. In Fig. 6b, fractional polymer crystallinityhas been presented against the draw ratio. The fractional crystal-linity (xc) for semi-crystalline HDPE was calculated from:xc ¼ DH=DH0
m, where DH is the enthalpy of fusion of the sample(the area under the melting peak in non-isothermal DSC thermo-gram) and DH0
m is the enthalpy of fusion for 100% crystalline HDPE.The enthalpy of fusion of 100% crystalline HDPE is taken to be 293 J/g [53]. For both, the stretched films and starting films, the DH valueswere obtained from the first heating cycle data.
For starting films, Resin-2 blown film exhibited the highestfractional crystallinity, ~0.68 followed by Resin-1 and Resin-3 films,respectively. This is consistent as Resin-2 is a nucleated, Resin-1 is anon-nucleated and Resin-3 is a lower density resin. The presence ofshort chain branching (incorporated through hexene comonomer)in Resin-3 makes it less crystalline (xc ~ 0.60). Short-chainbranching disturbs the regularity of the PE chain, consequentlythe crystallization process is hindered and the crystalline contentdecreases. This trend is also evident from the overall density ofthese resins reported in Table 1. For all three resins, the fractionalcrystallinity of the drawn films increased monotonically withincreasing DR. The increase in crystallinity upon stretching, albeitsmall, is assigned to strain induced crystallizationwhich gives birthto new crystals. A power-law model fitted the xc vs. DR datareasonably well. Nevertheless, the WVTR trend as a function of DRcannot be attributed to the increase in overall crystal content uponorientation alone since the crystallinity increased with DRmonotonically.
Diffusion of gas or small molecules through polymers dependson free volume content of the polymer and the tortuosity in thediffusion path. Typically crystal lamellae, being the hard imper-meable object, act as diffusion barriers in the crystalline polymers.Therefore, orientation/organization of the crystal lamellae impactsthe diffusion process and overall transport rate. It is noteworthy tomention that the MDO or CRD process orients lamellae or polymerchains in the machine direction. However, the MDO or CRD pro-cessing was performed on a monolayer 8-mil thick blown film(starting film). In the film blowing process, the polymer melt isextruded through a narrow gap while being drawn in the machineand transverse direction, inducing a small degree of machine di-rection orientation. In the high stalk film blowing process, the stalkis allowed to cool for a longer time prior to the film being blown upin the transverse direction. This extended cooling allows for morerelaxation and the formation of larger lamellae stacks that aremore
(b)
le (DR ¼ 5X). Heating rates are 10 �C/min. (b) The fractional crystallinity (xc) data as athe starting films are also shown (within the box).
Fig. 7. Orientation order parameter measured from birefringence data as a function ofdraw ratio. Orientation order monotonically increases with DR but the WVTR shows amaximum as a function of DR (Fig. 2). Therefore, orientation order parameter alonecannot describe the barrier properties exhibited by the MD-stretched films.
T. Chatterjee et al. / Polymer 55 (2014) 4102e41154110
randomly orientated. In contrast, when a film is blown in a moreconventional way (without high stalk), the polymer is immediatelydrawn/blown up above the air ring. This technique results in a rapidcrystallization, generating smaller lamellae that are more orientedin the machine direction. In our case the blown films were manu-factured without a high stalk condition. Therefore, it is essential toidentify the initial orientation of the crystals in the starting filmsand subsequent changes due to the MDO or CRD processing.
The degree of orientation is quantitatively measured throughthe orientation order parameter or Herman's orientation parameter(fH) defined as: fH ¼ ð3⟨cos2 q⟩� 1Þ=2, where, q is the angle be-tween the lamellar normal and the stretching direction. Since theaverage square cosine may vary between 0 and 1, the orientationorder parameter varies between �1/2 and 1.0. For random orien-tation, fH ¼ 0, for complete orientation q ¼ 0 and fH ¼ 1. For perfectorientation in the trans-direction (while stretching is along theMD), q ¼ 90� and fH ¼ �1/2. Therefore, as fH / 1.0, the degree oforientation along the stretching direction (here MD) becomes moreperfect. Here, it is necessary to discuss the orientation orderparameter calculations based on different analytical measurementtechniques. Small-angle X-ray scattering probes the crystal packingstructure (lamellae or extended chain crystals) and hence the orderparameter calculated based on azimuthal intensity distribution (atthe scattering vector where the primary reflection is observed)characterizes the crystal orientation order parameter [54,55].Wide-angle X-ray scattering investigates crystals structure formand orientation. From intensity distributions of different crystalplane reflections their orientation order parameter can be calcu-lated [54]. Similarly small-angle light scattering (SALS) data can beutilized to calculate mesoscale orientation order [56,57]. However,all these scattering data essentially deliver orientation orderparameter of the crystalline phase at different hierarchical lengthscales. In a semi-crystalline polymer, crystal lamellae are volumefilling and therefore, the amorphous segments of the chains mustbe included within the alternating crystal/amorphous lamellae.This implies that the orientation of crystal lamellae throughstretchingmay orient the amorphous segment of the chains as well.The change in birefringence as a function of stretching follows thestress-optical law and the orientation order parameter calculatedfrom the birefringence data correspond to both the crystalline andamorphous ordering (overall molecular orientation) [18]. A detaileddiscussion and the relation between the orientation order param-eter calculated based on different measurement techniques (X-ray,Birefringence and Infrared Dichroism) can be found elsewhere[58e60].
Previously Breese and Beaucage [8] reported that for MD ori-ented HDPE films the orientation order parameter calculated basedon the birefringence data correlates with the corresponding WVTRdata better than the same calculated using SAXS, WAXS or Fouriertransform-cross polarized optical microscopy measurements. Theorientation order parameter from birefringence data can bemeasured using the following expression [58,60]:Nk � N⊥ ¼ Dn ¼ fH*Dn0. In this expression Nk and N⊥ are therefractive indices measured in the MD and TD, respectively. Dn0 isthe maximum or intrinsic birefringence which is 0.063 for HDPE athigh extrusion ratio [61]. The calculated fH values have been pre-sented in Fig. 7. The overall orientation order parameter was foundto be ~0 or slightly negative for the blown films (starting films).Therefore, the overall polymer chains were randomly oriented instarting films. The order parameter could not be measured for theResin-3 blown film (starting film) and its' 3X stretched sample sincethese filmswere too hazy tomeasure refractive index confidently. Itis apparent from Fig. 7 that as the draw ratio increased, the fH valueincreased and it tended to reach an asymptotic value of ~0.8 at highDR. As mentioned before, birefringence measurements correspond
to overall chain orientation. Therefore, stretching yields orientationof both the crystalline and amorphous segments of the HDPE chain.For any given DR, the fH values do not depend much on the resintype. For the small set of resins studied here, different extent ofdrawing (i.e., different DR) brings about the same state of polymerchain orientation. Consequently, the absolute values of both theoptical and overall barrier properties are a strong function of theresin physical properties (density, molecular weight and their dis-tribution) rather than the stretching process.
It was hypothesized that the changes in crystal morphologyalong with the crystal orientation were responsible for improvedbarrier properties. Therefore, crystal morphologies were studiedindependently by a combination of different scattering techniquesand atomic force microscopy. 2-D wide-angle X-ray and small-angle laser light scattering profiles of the Resin-2 films, stretchedto different draw ratios, have been shown in Fig. 8. These arerepresentative data sets and similar observations were made forResin-1 and Resin-3 stretched films as well (Supportinginformation). Wide-angle X-ray scattering image for the “startingfilm” (non-stretched blown film, 8-mil thick) exhibited weakorientation with (110) plane preferably oriented in MD direction,which is typical for PE blown film. It should be noted that orien-tation seenwith X-ray diffraction represents crystal orientation in amuch smaller scale than what light scattering look at. Thus,different orientation degree can be yielded for the same samplewith different technique as the scale range studied is different. Inpolyethylene, the unit cell is orthorhombic (a ¼ 7.42 Å, b ¼ 4.95 Å,c ¼ 2.55 Å) and the polymer chains run along the c-axis. Forrandomly oriented samples, possessing a spherulitic morphology,the c-axis of the unit cell (chain direction) is oriented at an angle34.4� to the lamellar normal direction [62e64]. For any periodicstructure, the sharpness of the reflection in the azimuthal planereveals the extent of orientation of the structural normal. For Resin-2 starting film, the Herman orientation factor was found to be�0.05and 0.16 when calculated from the azimuthal intensity distributionof (110) and (200) plane reflections, respectively (MD was used asthe reference direction for orientation evaluation).
The c-axis or chain orientation can be understood by followingthe texture of the (110) plane reflection. The (110) plane reflectionbecame concentrated at equatorial plane (along TD) indicatinggradual ordering of the chains parallel to the stretch direction (MD).
Fig. 8. Wide-angle X-ray scattering (WAXS) and HV light scattering (LS) data for Resin-2 blown film stretched at different draw ratios. Top panel represents the WAXS data and thebottom panel shows the corresponding LS data.
T. Chatterjee et al. / Polymer 55 (2014) 4102e4115 4111
Simultaneous diminished intensity of the (200) plane reflectionsuggested an increasing a-axis orientation normal to the c-axis andparallel to the thin dimension of the film [65]. Such observation isconsistent with the WAXS profile of the MD-drawn HDPE that hasbeen previously reported [10]. The orientation order parameterscalculated were found to be 0.44, 0.54, 0.64, 0.68, 0.69 and 0.71 for3X CRD, 4X CRD, 5X CRD, 6X CRD, 7X CRD, and 10X MDO samples,respectively. These orientation order parameters were calculatedbased on azimuthal dependence of (110) plane reflections with TDas reference direction. Initially, the order parameter increasedstrongly up to DR ¼ 6X where the c-axis of the unit cell becameprogressively aligned along the draw direction. Mirabella and co-workers claimed that at this condition, chain segments in both theamorphous and crystalline regions were pulled into near-perfectorientation along the draw direction [25]. Beyond this DR, allchains were aligned parallel to the MD and no further alignmentwas possible. Therefore, very weak change in the orientation orderparameter was found beyond that point. The HV patterns in SALSdepend on optical anisotropy and orientation fluctuation. In small-angle light scattering (SALS), the starting film exhibited a weakfourfold symmetrical HV scattering pattern consistent with spher-ulite structure with little orientation. At low DR regime (3Xe5X),below the critical DR (associated with the maximum observed forinWVTR data), fibrils parallel to the MD directionwere observed asevident from the equatorial intensity streaks (albeit weak) in SALSpattern. This suggests that drawing at a temperature close to themelting temperature caused localized and partial melting ofweaker and thinner crystals and subsequent microfibrillation[10,15,66]. At high DR regime (6Xe10X), close to and above thecritical DR, besides strong equatorial intensity streaks (i.e., scat-tering from rod-like structures or fibrils), a weak four-lobe patternwas also observed. For the lobes, intensity was maximum at thecenter and gradually decreased with increasing scattering vector.Further, the lobes were flattened along the stretching direction.This four-lobe pattern suggested that some residual and immaturespherulites still existed in the stretched films. For extruded HDPEfilms with different draw ratios, similar HV LS pattern has beenattributed to flattened spherulites [67] with the major axisperpendicular to the stretching direction [68,69].
The transition from spherulitic to rod-like microfibrillarmorphology was further investigated in real space through atomicforce microscope phase imaging (Fig. 9). In this set of images allfilms were made of Resin-2. Any noticeable orientation was notobserved in the starting film as is evident from isotropic lamellaestructure in the MDeTD plane. At 3X CRD drawn samples, most ofthe lamellae rearranged perpendicular to the MD; i.e., lamellar
thickness or lamellar normals were parallel to the draw direction.At an intermediate DR, 5X CRD drawn sample, some molecularbundles were observed. At this DR, also a transition fromperpendicular lamellae to fiber-like structure appeared. At ahigher DR (7X CRD or 10X MDO) more bundles and fiber-likestructures were observed on the surface. Consistent with scat-tering studies, the transition in crystal morphology occurredaround DR ~ 6X. Similar scattering and AFM data confirmingspherulite to microfibrillar crystal transformation upon stretchingfor Resin-1 and Resin-3 films are reported in the Supportinginformation. Finally, AFM images did not show any spherulitesas observed in SALS pattern (Fig. 8). The drawn films wereannealed and cooled at 82 and 33 �C, respectively, where thesespherulites were likely formed during secondary crystallization.Scattering probes the bulk film and covers a statistically large area,while AFM is a surface technique which (in our case) was con-ducted within a very limited area (3 mm � 3 mm). In future largerarea AFM scans are indicated to investigate the presence ofspherulites on film surface.
Before any drawing operation, in starting films, all lamellae wererandomly oriented in 3D macroscopic volume. Random orientationoffered substantial tortuosity in the diffusion path and therefore,the un-oriented films exhibited a low WVTR. At low DR regime,perpendicular crystal lamellae morphology was observed whichindicated that lamellae normals were parallel to the draw direction.Therefore, at low DR samples, polymer chains subtend an angle of34.4� with the draw direction. With increased stretching, polymerchains start to get aligned along the stretching direction. In thisregime, the WVTR increased due to a decrease in the diffusion pathtortuosity arising from the lamellae orientation. Further, the dila-tion of materials (non-cohesive damage) under uniaxial stretchingdue to Poisson ratio mismatch between the amorphous (Poissonratio ~ 0.4e0.5) and crystalline phases (Poisson ratio ~ 0.3) inducesome free volume [70,71]. Such free volume induction is known tobe reduced or eventually eliminated at high temperature drawingdue to increased chain mobility and partial melting of crystals[46,70,72]. In this study, even when the films were stretched atclose to the crystal melting temperature (Table 3), we presumesome free volume induction due to incomplete or partial melting ofcrystal lamellae and strain induced crystallization. This additionalfree volume also contributed, albeit small, to the initial increase inWVTR as a function of DR.
At a critical DR, all the lamellae were presumably optimallyoriented with their normals subtending an angle of ±34.4� with themachine direction (i.e., c-axis was k to the draw direction) and atthis point the maximum in WVTR was realized. It is worthwhile to
Fig. 9. AFM phase images from the non-stretched ‘starting film’ and stretched films at different draw ratios. All films presented here were made using Resin-2.
T. Chatterjee et al. / Polymer 55 (2014) 4102e41154112
recognize that around the critical DR the orientation orderparameter calculated based on (110) plane WAXS reflectionapproached an asymptotic value. For isotactic-polypropylene (i-PP)films, drawn at high temperature, Samuels previously reported thatthe long spacing (obtained from SAXS measurements) increaseswith stretching below the critical DR and remains constant beyondit [73,74]. This suggests an optimum c-axis as well as lamellar ori-entations is required prior to lamellar to fibrillar morphologytransition [72e76]. Further, stretching at close to the crystalmelting temperature eases the inter-lamellar deformation throughcrystal cleavage and subsequent fibrillar structure formation.Beyond the critical DR, unraveling or transformation of lamellaeprobably follows the mechanism presented by Matsumoto andcoworkers [77]. In this condition, localized melting causes thinnerand weaker lamellae where some of these possibly begin to unzipand form long, rigid fibrils oriented in the machine direction, asindicated in Fig. 9 (AFM phase images for DR 5X).
Under strain, bigger crystals break and smaller crystals aregenerated as was evident from an increase in fractional crystallinity(Fig. 6b) with increasing DR. During this process lamellae weremostly fractured in the lateral direction [25]. Light scattering dataalso showed some immature spherulite formation in this condition.Further drawing beyond the critical DR yielded an increased for-mation of large aspect ratio fibrillar crystals decorated with smallerlamellae all around the azimuthal plane. At this point it is notevident whether the lamellar growth in the transverse direction isepitaxial in nature i.e, a shish-kebab structure is formed or a row-nucleated lamellae stacked around the central fibrillar crystal isformed. As mentioned before, the crystal lamellae act as barriers tothe adsorption and diffusion process. Most of the transport takesplace through the amorphous part of the polymer chains. In addi-tion to that evolution of microfibrillar crystal structure rendered achange in the geometric distribution of the amorphous regions.Further, the amorphous phase also got oriented (to some extent)due to stretching as was evident from the birefringence data. Pre-sumably, a combination of all three, (a) a lower amorphous chain
content, (b) their orientation along the MD (i.e., amorphoussegment of chain becomes more crystal-like) [74], and (c) changesin the geometric distribution of amorphous phase due to microfi-brillar crystal formation resulted in the observed trends in watervapor transport. Consequently, beyond the critical DR, the watervapor transport rate gradually decreased with stretching.
Previously, Keller [78] showed that for polyethylene high mo-lecular weight tail is important for controlling the formation ofextended chain fibrils in row nucleated structures. It was later [79]shown that the minimum strain rate needed to stretch polymerchains from the random coil state to the fully stretched state(extended conformation) decreases as molecular weight increases.However, for a given PE sample, microfibrillar structure evolutionunder shear is independent of the shear rate and occurs at a criticalstrain [80]. This suggests that the deformation of high molecularweight polymer chains controls this transition. Therefore, resinwith high polydispersity or high Z-averagemolecular weight (Mz) isexpected to exhibit an earlier transition since those high molecularweight chains cannot relax on the time scale of experiment [80].Consistent with that an earlier lamellar to microfibrillar crystalstransition (or equivalently observation of the maximum WVTR at alower critical DR) was observed for Resin-3 which contained anextremely high molecular weight tail (Table 1).
3.4. Comparison of barrier and optical properties between the MDOand CRD processed films
One major aspect which remained unanswered is the relativeperformance between the MDO and CRD processed films when theresin selection and film thickness (equivalently DR) are held con-stant. In general it is hard to compare these two technologies sincethey are typically effective at different DR zones. Compression rolldrawing is effective for low draw ratios where the traditional MDOoperation would result in tiger striping of the film (DR < naturaldraw ratio). On the other hand, the traditional MDO operation canuniformly stretch HDPE films above their natural draw ratio which
T. Chatterjee et al. / Polymer 55 (2014) 4102e4115 4113
is typically ~6e7X (at cold draw condition). It is well-known thatthe natural draw ratio of HDPE depends on different processingconditions (such as choice of catalyst, single vs. dual reactor pro-cess, etc.) [81]. Nevertheless, stretching of an HDPE film using theCRD technology above the natural draw ratio requires a highstarting film thickness (10 mil or above) to avoid direct roller toroller contact. For Resin-3, films could be stretched at 3X and 4Xdraw ratio using the traditional MDO operation without any tigerstriping. It may be argued that the uneven stretched regions werenot visible for Resin-3 which demonstrated inherently poor opticalproperties (due to presence of large crystals and broad crystal sizedistribution). To verify this, the film thickness was independentlymeasured at 30 random spots and based on that it was concludedthat even at low DR (3X and 4X), the MDO stretched films were ofuniform thickness.
The comparison between the MDO and CRD stretched films interms of their water vapor transport rate and optical propertieshave been presented in Table 4. In both cases (DR ¼ 3X and 4X), theCRD stretched samples exhibited lower WVTR. It is also apparentthat the films stretched using the CRD technology displayed su-perior optical properties. The clarity of the CRD drawn films was~80% compared to ~30e50% for the MDO stretched film. In parallel,only at high DR (8X), the MDO drawn Resin-3 film showed 80%clarity (Fig. 4). The optical property difference between the MDOand CRD drawn films was more dramatic for % gloss and total hazemeasurements. In fact, 3X and 4X CRD stretched Resin-3 film glosswas comparable with Resin-1 and Resin-3 CRD stretched films (forthe same DR). Hence, by using the CRD technology the opticalproperties of stretched Resin-3 films improved significantly andwere comparable with similarly stretched Resin-1 and Resin-2films.
Total haze of a film is contributed by both the bulk and surfacehaze. It is well recognized that for polyethylene films, surface hazeis the dominating contributor to the total haze behavior [82,83].Similarly, with increasing surface roughness the scattering in non-specular directions increases leading to a lower gloss value [35,45].Therefore, it was conjectured that the superior optical properties(higher gloss and lower total haze) of CRD drawn films stem fromthe lower surface roughness rendered by the application ofcompressive force during the CRD stretching. This was verifiedthrough the AFM topography images of Resin-3 films 3X stretchedusing the CRD and MDO technologies, respectively (Fig. 10). Theroot mean square roughness (Sq) is defined as the geometric meandeviation of the all the points on the surface from themean value ofthe data (center surface plane). The mean Sq values were~214 ± 28 nm for the MDO stretched sample and ~100 ± 24 nm forthe CRD stretched samples, respectively. This topography imageanalysis undoubtedly proves that the CRD process yields lowersurface roughness. Hence it is concluded that uniaxial stretchingimproves the optical properties of the film by orienting the crystals.Additionally, transformation of crystal morphology helps inimproving optical properties. For example, due to the largerspherulites size compared to fibrils, the films with fibrillar
Table 4Comparison of barrier and optical properties for the films with the same draw ratiobut processed through different (MDO/CRD) technologies. All films were made ofResin-3.
morphology are expected to show a much lower haze than thosewith spherulitic structure. In addition to that, CRD technology re-sults in a smoother surface compared to the MDO technology. Thelower surface roughness is the reason behind better optical prop-erties displayed by films oriented using the CRD technology.
4. Summary
In this study, three proprietary Dow HDPE resins were selectedwhich varied in their physical properties (density, molecularweight, molecular weight distribution, etc.). Blown films of thesethree resins were uniaxially stretched in the machine directionusing either traditional machine direction orientation (MDO) orcompression roll drawing (CRD MDO) technologies. The compres-sion roll drawing was effective in stretching films uniformly belowthe polymer natural draw ratio. In contrast, the traditional MDOtechnology stretched films uniformly above the natural draw ratio.For all three resins, initially the oriented films demonstrated anincrease in the water vapor transport rate (WVTR) with increasingdraw ratio (DR) and beyond a critical DR it gradually decreased.Overall, for all three resins, improvement in barrier properties (i.e.,reduction of WVTR) was realized at high draw ratio relative to thenon-stretched control samples (of the same thickness).
Films made from all three resins exhibited improvement inoptical properties (higher clarity and gloss and lower total haze)after stretching. Unlike traditional MDO, the CRD MDO processapplied a little amount of compressive force during stretching(along with tensile force) which resulted in lower surface rough-ness (compared to the traditional MDO stretched films). As a resultthe CRD MDO stretched samples exhibited markedly better opticalproperties compared to traditional MDO stretched samples (for thesame resin and the same DR).
The above results were rationalized in terms of changes in thecrystal content, morphology and orientation induced by uniaxialstretching. The fractional crystallinity of the polymer increasedmonotonically with increasing stretching. However, the absolutechange was quite small and less than 5% even at the highest drawratio studied for each resin. Birefringence data revealed that theuniaxial stretching resulted in molecular orientation (both thecrystalline and amorphous chain segments) along the draw direc-tion. The orientation order parameter calculated based on thebirefringence data increased rapidly at low DR and finally reachedan asymptotic value of ~0.8 at high DR. Further, this parameter wasindependent of the resin selection for any DR. This suggested thatirrespective of the physical properties of the resin/starting film, theMDO/CRD process oriented the polymer chains to the same extent(for the same DR). Therefore, for the same extent of stretching/orientation, the stretched film properties were found to be a strongfunction of the inherent resin physical properties.
Stretching induced transformations in both the crystalline andamorphous phases were presumably responsible for the WVTRtrend observed as a function of the DR. Before the stretchingoperation, in starting films the spherulitic crystal lamellae wererandomly organized which provided tortuosity to the diffusionpath. As the films were stretched, the unit cell c-axis or the chainaxis got oriented parallel to the draw direction. In this DR zone, thecrystal lamellae organization resulted in a decrease in diffusionpath tortuosity and consequently the WVTR increased. Beyond acritical DR (at which the maximum in WVTR as well as optimum c-axis orientation were observed), polymer chains started to unfoldfrom the lamellae and form fiber-like crystal structure. Simulta-neously, smaller crystals were formed decorating around the fiber-structure (transverse to the draw direction). This transformation incrystal morphology resulted in geometric redistribution of amor-phous regions in the 3-D macroscopic volume. In addition to this,
Fig. 10. Atomic force microscopy topography images and the corresponding 3D view.
T. Chatterjee et al. / Polymer 55 (2014) 4102e41154114
the chain amorphous segments were oriented due to stretchingand the overall amorphous content of the chain decreased. Sincethe sorption and diffusion of gas molecules takes place through theamorphous region, their geometric redistribution, orientation and aslight decrease in overall volume fraction hinders the transportprocess. Therefore, above the critical DR, a decrease in the WVTRwith increasing DR was observed.
Acknowledgment
Authors thank Drs. Todd Pangburn, Ayush Bafna and BrianLandes for their insightful comments. We thank Dr. Tianzi Huangfor the GPC data. TC thanks Blake Barthlome and Jon (Wes) Hobsonfor help in blown film production. This research was supported bythe Dow Chemical Company.
Appendix A. Supporting information
Supporting information related to this article can be found on-line at http://dx.doi.org/10.1016/j.polymer.2014.06.029.
References
[1] Lewis J. Mater Today 2006;9(4):38e45.[2] Granstrom J, Gallstedt M, Villet M, Moon JS, Chatterjee T. Thin Solid Films
2010;518(18):5282e7.[3] Granstrom J, Villet M, Chatterjee T, Gerbec JA, Jerkunica E, Roy A. Appl Phys
Lett 2009;95(9).
[4] Ahmad J, Bazaka K, Anderson LJ, White RD, Jacob MV. Renew Sustain EnergyRev 2013;27:104e17.
[5] Lagaron JM, Catala R, Gavara R. Mater Sci Technol 2004;20(1):1e7.[6] Miller KS, Krochta JM. Trends Food Sci Technol 1997;8(7):228e37.[7] Sha HG, Harrison IR. J Polym Sci B e Polym Phys 1992;30(8):915e22.[8] Breese DR, Beaucage G. Effects of machine direction orientation (MDO) on the
moisture and oxygen properties of HMW-PE films. Las Vegas, NV: TAPPI;2005.
[9] Salame M. The use of barrier polymers in food and beverage packaging. In:Finlayson KM, editor. Plastic film technology. Technomic, vol. 1; 1989.pp. 132e45.
[10] Paulos JP, Thomas EL. J Appl Polym Sci 1980;25(1):15e23.[11] Williams JL, Peterlin A. J Polym Sci A-2 e Polym Phys 1971;9(8):1483e94.[12] Meinel G, Morosoff N, Peterlin A. J Polym Sci A-2 e Polym Phys 1970;8(10):
1723e40.[13] Meinel G, Peterlin A. J Polym Sci A-2 e Polym Phys 1971;9(1):67e83.[14] Meinel G, Peterlin A. Eur Polym J 1971;7(6):657e70.[15] Ratta V, Wilkes GL, Su TK. Polymer 2001;42(21):9059e71.[16] Zhang XM, Elkoun S, Ajji A, Huneault MA. Polymer 2004;45(1):217e29.[17] Prasad A, Shroff R, Rane S, Beaucage G. Polymer 2001;42(7):3103e13.[18] Ajji A, Zhang X, Elkoun S. Polymer 2005;46(11):3838e46.[19] Yu TH, Wilkes GL. Polymer 1996;37(21):4675e87.[20] Chau CC, Raspor OC. J Polym Sci B e Polym Phys 1990;28(5):631e45.[21] Peterlin A, Williams JL, Stannett V. J Polym Sci A-2 e Polym Phys 1967;5:
957e72.[22] Schut JH. Plast Technol 2005;(February).[23] Breese DR. Convert Q 2011;(Quarter 2):4.[24] Tabatabaei SH, Parent L, Cigana P, Ajji A, Carreau PJ. J Plast Film Sheeting
2009;25(3e4):235e49.[25] Bafna A, McFaddin D, Beaucage G, Merrick-Mack J, Mirabella FM. J Polym Sci B
e Polym Phys 2007;45(14):1834e44.[26] Breese DR, Beaucage G. Curr Opin Solid State Mater Sci 2004;8(6):439e48.[27] Breese DR, Beaucage G. J Polym Sci B e Polym Phys 2008;46(6):607e18.[28] Breese DR. Modeling the effects of solid state orientation on blown high
molecular weight high density polyethylene films: a composite theory
T. Chatterjee et al. / Polymer 55 (2014) 4102e4115 4115
approach. vol. Master of Science. Department of Chemical Materials Engi-neering, University of Cincinnati; 2005. p. 159.
[29] Van Erden DL, Enriquez MC, Noren DW. US Patent 5387388, Method forproducing oriented plastic strap. Illinois Tool Works Inc.; 1995
[30] Van Erden DL, Enriquez MC, Noren DW. Method and apparatus for producingsimultaneously milled and stretched plastic strap. US Patent 5405699,Signode Corporation; 1995.
[31] Van Erden DL, Enriquez MC. Method and apparatus for producing orientedplastic strap, and strap produced thereby. US Patent 5837349, Illinois ToolWorks Inc.; 1998.
[32] Uejo H, Hoshino S. J Appl Polym Sci 1970;14(2):317e28.[33] Materials ASfTa. Standard test method for haze and luminous transmittance of
transparent plastics, standard test method D 1003-00; 2000.[34] Materials ASfTa. Standard test method for transparency of plastic sheeting,
standard test method D 1746-97; 1997.[35] Fratini CM. Study of the morphology and optical properties of propylene/
ethylene copolymer films. vol. Ph.D. Department of Chemistry, VirginiaPolytechnic Institute and State University; 2006. p. 196
[36] Kristiansen M, Werner M, Tervoort T, Smith P, Blomenhofer M, Schmidt HW.Macromolecules 2003;36(14):5150e6.
[37] DeMuse MT. J Plast Film Sheeting 2002;18:17e23.[38] Wood SS, Pocius AV. J Plast Film Sheeting 2001;17:62e87.[39] Sadeghi F, Carreau PJ. Can J Chem Eng 2008;86(6):1103e10.[40] Materials ASfTa. Standard test method for specular gloss of plastic films and
solid plastics, standard test method D 2457-97; 1997.[41] www.metricon.com.[42] Kersten RT. Opt Commun 1975;13(3):327e9.[43] Tien PK, Ulrich R, Martin RJ. Appl Phys Lett 1969;14(9):291.[44] Willmouth FM. Transparency, translucency, and gloss. Amsterdam: Elsevier;
1986.[45] Assender H, Bliznyuk V, Porfyrakis K. Science 2002;297(5583):973e6.[46] Lin L, Argon AS. J Mater Sci 1994;29(2):294e323.[47] Lavine MS, Waheed N, Rutledge GC. Polymer 2003;44(5):1771e9.[48] Kornfield JA, Kumaraswamy G, Issaian AM. Ind Eng Chem Res 2002;41(25):
6383e92.[49] Somani RH, Yang L, Zhu L, Hsiao BS. Polymer 2005;46(20):8587e623.[50] Swartjes FHM, Peters GWM, Rastogi S, Meijer HEH. Int Polym Process
2003;18(1):53e66.[51] Savolainen A. Polym Eng Sci 1990;30(19):1258e64.[52] Tabatabaei SH, Carreau PJ, Ajji A. Polymer 2009;50(17):4228e40.[53] Mark J. Physical properties of polymer handbook. 4th ed. Springer; 2006.
[54] Chatterjee T, Mitchell CA, Hadjiev VG, Krishnamoorti R. Adv Mater2007;19(22):3850e3.
[55] Chatterjee T, Mitchell CA, Hadjiev VG, Krishnamoorti R. Macromolecules2012;45(23):9357e63.
[56] Hashimoto T, Prudhomm Re, Keedy DA, Stein RS. J Polym Sci B e Polym Phys1973;11(4):693e708.
[57] Hashimoto T, Prudhomm Re, Stein RS. J Polym Sci B e Polym Phys 1973;11(4):709e36.
[58] Stein RS. J Polym Sci 1958;31(123):327e34.[59] Stein RS. J Polym Sci 1958;31(123):335e43.[60] White JL, Spruiell JE. Polym Eng Sci 1983;23(5):247e56.[61] Pietralla M, Grossmann HP, Kruger JK. J Polym Sci B e Polym Phys 1982;20(7):
1193e205.[62] Bassett DC, Hodge AM. Proc R Soc Lond Ser A e Math Phys Eng Sci
1981;377(1768):25.[63] Bassett DC, Olley RH, Alraheil IAM. Polymer 1988;29(9):1539e43.[64] Gautam S, Balijepalli S, Rutledge GC. Macromolecules 2000;33(24):9136e45.[65] Young RJ, Bowden PB. J Mater Sci 1973;8(8):1177e84.[66] Rhodes MB, Stein RS. J Polym Sci 1960;45(146):521e4.[67] Baranov VG, Frenkel SY, Volkov TI, Zurabian RS, Gromov VI. J Polym Sci C e
Polym Symp 1972;(38):61.[68] Hashimoto T, Todo A, Itoi H, Kawai H. Macromolecules 1977;10(2):377e84.[69] Hashimoto T, Ishido S, Kawai H. Macromolecules 1978;11(6):1210e5.[70] Nitta KH, Yamana M. Poisson's ratio and mechanical nonlinearity under ten-
sile deformation in crystalline polymers. In: Vicente JD, editor. Rheology.InTech; 2012. pp. 113e32.
[71] Addiego F, Dahoun A, G'Sell C, Hiver JM. Polymer 2006;47(12):4387e99.[72] Butler MF, Donald AM. Macromolecules 1998;31(18):6234e49.[73] Samuels RJ. J Polym Sci A-2 e Polym Phys 1968;6(A-2):1101e39.[74] Samuels RJ. Structured polymer properties. John Wiley & Sons, Inc.; 1974.[75] GaucherMiri V, Francois P, Seguela R. J Polym Sci B e Polym Phys 1996;34(6):
1113e25.[76] Vickers ME, Fischer H. Polymer 1995;36(13):2667e70.[77] Matsumot T, Kawai T, Maeda H. Makromol Chem 1967;107(Aug):250e2.[78] Bashir Z, Odell JA, Keller A. J Mater Sci 1986;21(11):3993e4002.[79] Keller A, Kolnaar JWH. Prog Colloids Polym Sci 1993;92:81e102.[80] Balzano L, Rastogi S, Peters G. Macromolecules 2011;44(8):2926e33.[81] Kuriyagawa M, Nitta KH. Polymer 2011;52(15):3469e77.[82] Kojima M, Magill JH, Lin JS, Magonov SN. Chem Mater 1997;9(5):1145e53.[83] Johnson MB, Wilkes GL, Sukhadia AM, Rohlfing DC. J Appl Polym Sci
