AFM and X-ray Studies of Crystal and Ionic Domain Morphology in Poly(ethylene-co-methacrylic acid)Ionomers
Bryan B.Sauer*and R.Scott McLean
DuPont Central Rearch and Development,Experimental Station,Wilmington,Delaware19880-0356 Received June27,2000
ABSTRACT:New AFM methods were applied to resolve the morphology of the ionic domains in poly-(ethylene-co-methacrylic acid)copolymers.Using standard tapping AFM techniques with a precision in the lateral dimensions on the order of a nanometer,the crystalline lamellar structure and overall morphology consisting of stacks of ethylene-rich lamellae parated by sometimes broad noncrystalline regions were characterized.By operating the AFM under special low oscillation amplitude conditions where tip-ionic cluster interactions could be induced to dominate the pha signal,it was shown that this method could be ud to uniquely resolve ionic-rich regions and individual ionic domains.The individual domains were found to be on the order of2nm in diameter.Small-angle X-ray scattering (SAXS)characterization was ud to confirm some aspects of the morphology and to contrast the different levels of resolution of the two techniques for both lamell
hilltop
ar crystals and ionic domains.By quential images taken under different tapping AFM conditions,the“softer”amorphous regions were found to be the richest in ionic domains.Lamellar morphology and perfection were found to be controlled by mobility in the melt which depends on acid level,neutralization level,and counterion type in the different ionomers studied.The crystalline domains in the metal neutralized ionomers were compared to the unneutralized “acid”form of the ionomer.Data for the acid form verified that no ionic domains exist in this material.
Introduction
Ionomers such as poly(ethylene-ran-methacrylic acid) neutralized with Zn and other metal ions are examples of the Surlyn family of copolymers.Ionic clustering in many polymers including the ethylene-bad iono-mers,1-6perfluorinated sulfonic acid,3,6-11and other ionomers2-6,12bad on block copolymers has been reviewed.Theoretical models for ionic clustering de-scribe the energetics and chain conformation effects in limiting cluster sizes.3-9,12Some theories have been criticized becau they do not consider the constraints impod by the crystallites.3Complete theories are apparently unavailable becau of the many complexi-ties including energetics of ion interactions,steric chain effects,interfacial tension and interface interactions, and possibly chain mobility constraints due to crystal-linity.The importance of the latter must be qualified becau even i
n the melt there are ionic domains comparable in size which contribute to small-angle X-ray scattering,2,13strongly suggesting that steric chain constraints12are dominant in limiting ionic domain size. The nature of steric effects involving the chain structure includes the covalent attachment of hydro-carbon chains to the ionic groups and forms the basis of the model of Yarusso and Cooper2,14where the distance between ionic domains depends on both the diameter of the ionic domain and the thickness of the attached hydrocarbon chains at the“surfaces”of the domains.The morphology studies bad mostly on small-angle X-ray scattering(SAXS)show that this model of interionic domain scattering provides an excel-lent description of many ionomer systems,and the results show that other models may be less applicable.2 Effects of dispersity of domain spacings are probably important but have not been considered.
Although the influence of polymer structures on the domain spacings is relatively well established,studies of domain size are very controversial.The results on different polymers show interesting structure/property relationships of ionic domain size,2but the absolute domain sizes are qualitative becau of possible model-dependent contributions to the analysis.The detailed relationship of domain sizes to the miblocky structure of the polymer chain with ionic species parated by hydrocarbon blocks of different length in different Surlyn compositions is not available.I
n one early model,12chain structure and the energetics of chain stretching resulting from steric effects due to the copolymer chain structure were considered in addition to the ionic interactions.Related to this is the steric constraints which evolve due to the accumulation of hydrocarbon chains around the ionic domains(or other “hard”domains)at a given volume fraction of ionic species,and the could provide kinetic limitations to domain formation.
SAXS,1,2,6,7,10,11,13neutron scattering,15transmission electron spectroscopy(TEM),6and dynamic mechanical data5,6all can be qualitatively correlated with various model predictions.5From the start,2,16-18interpretation of the SAXS peak associated with the ionic domains has been controversial in poly(ethylene-ran-methacrylic acid) ionomers.The“higher angle”SAXS peak maximum was attributed to either interparticle or intraparticle scat-tering,and veral diver ionic domain morphologies were postulated including sphere,2lamellae,17and core/ shell.18Recent experiments using extended X-ray ab-sorption fine structure spectroscopy also provide de-tailed information about the coordination number and chemistry of the ionic species19,20and provide some evidence that is consistent with nonspherical ionic domains.20Recent scanning(STEM)data have provided new direct evidence of a spherical shape,21,22the STEM21,22studies characterized domain diameters of approximately2-3nm for Zn-neutralized Surlyn,21,22 and the domains are found to have a relatively low level of dispersity in sizes with sizes only slightly dependent
7939
Macromolecules2000,33,7939-7949
10.1021/ma001110t CCC:$19.00©2000American Chemical Society
Published on Web09/16/2000
on neutralization level.The STEM results22provide valuable model-independent characterization of domain size and shape which suggests that many available theoretical models referenced above are not accurate and should be refined.They strongly suggest that the earlier estimates of domain diameters on the order of1 nm for Surlyn2,12are far too small.Additional SAXS studies including“resonant”SAXS studies,23which u wavelength to change the contrast due to the prence of metal ions,have provided additional details of ionic domain morphology.
Becau the electron microscopy ction thickness is not exactly controlled and is large relative to the ionic domain size,the position in space of ionic domains
relative to the ethylene crystal ,stacks of lamellae)is difficult to discern.In addition,the overall pha structure and organization including the poly-ethylene lamellae are not easily resolvablringworm
e by electron microscopy becau of the high degree of imperfection of the ethylene crystals.The low degree of order also leads to difficulties in modeling of SAXS data and to difficulties in characterization by wide-angle X-ray diffraction in ionomers and other ethylene copolymers where WAXD gives a total percent crystallinity much lower than that derived from DSC and other techniques becau of the poor crystal perfection.24,25
Recent work has verified that tapping AFM has resolution similar to that of TEM for some polymer morphological features such as crystal lamellae near the surface,26-31with the additional advantage of AFM that thin ctions are not needed.It has been demonstrated that AFM can resolve certain polymer microstructural features at the nanometer scale,with the limitation that the particular orientation and depth of domains under the surface can modify aspects of the actual resolution of such features.27,28,32However,the application of AFM in the ionomer field has mainly addresd surface textural features,33-35and only recently have new AFM methods been developed and applied to the examination of ionic clusters.36We will discuss data which show that a combination of AFM techniques can provide additional morphological information not readily available from any other real-space technique becau of the ability of AFM to characterize the ethylene lamellae in addition to using novel imaging methods36to determine the relative morphological position of the ioni
c species and domains.The real space organization and network of ionic and nonpolar aggregates is relevant to trans-port,3,7-9surface,adhesion,and mechanical properties.4 We will also contrast the results with new SAXS data on some of the systems in an attempt to obtain a more detailed picture of ionic domain morphology. Experimental Section
Materials and Preparation.The Surlyn samples are from DuPont and are bad on poly(ethylene-co-methacrylic acid) polymers.The are metal neutralized with the degrees of neutralization and acid comonomer levels listed in Table1and a sample designation implemented by indicating the metal neutralization species followed by the degree of neutralization, e.g.,Zn58for58%neutralization.The ethylene blocks form crystallites which are characterized by broad melting DSC regions with peak melting points around90-110°C depending on the composition,23which can be compared with a melting point for linear polyethylene of about133°C.
Surfaces with low levels of long-range roughness can improve resolution of nanocrystal structures in tapping AFM studies.28Obtaining smooth surfaces from the melt for poly-mers that crystallize can be difficult becau crystallization in some polymers roughens the surface.Becau of this,we developed methods to fabricate very smooth surfaces for the ethylene-bad ionomers,although problems with roughness are less vere with copolymers due to lower degrees of crystallinity.Ther
mal melt or“melt-presd”surfaces were obtained by pressing a silane grafted silicon disk onto the Surlyn melt at180°C and cooling to room temperature.In some cas the samples were annealed at elevated tempera-tures.The levels of crystallinity prent near the surface are probably difficult to control,and there was no attempt at determining quantitative reproducibility of levels of crystal-linity.
To coat the silicon wafer,a lf-adsorbed monolayer of octadecyltrichlorosilane was deposited from standard hexa-decane mixed solvents.37The silicon disk was removed from the Surlyn surface after cooling from the melt,leaving a surface almost as smooth as the silicon wafer.This type of silicon surface treatment prevents reaction of materials such as molten nylon and Surlyn with the Si-OH groups on the inorganic oxide surface of the bare silicon,thus minimizing adhesion and acting as a permanent mold relea.For the most part the silane remains on the silicon wafer.Evidence for this includes the fact that a single treated disk can be ud to prepare veral dozen samples from the melt.In one example discusd at the end of this paper,we do note the interesting conquences of degradation of this silane layer.This sample preparation method can be contrasted with a common method typically ud in the literature which is to rub polymer on a substrate,sometimes at elevated temperatures.This leaves thin layers with interesting morphologies,27but the relation-ship to morphologies in macroscopically thick samples is sometimes tenuous.
Methods.SAXS data were collected using a compact Kratky camera using Cu K R(λ)1.54Å)radiation.The data were procesd after background subtraction and desmearing.
A TA Instruments(Newcastle,DE)2920DSC was ud for the thermal characterization.
Tapping mode AFM was ud to obtain height and pha imaging data simultaneously on a Nanoscope IIIa from Digital Instruments,Santa Barbara,CA.Microfabricated cantilevers or silicon probes(Nanoprobes,Digital Instruments)with125µm long cantilevers were ud at their fundamental resonance frequencies which typically varied from270to350kHz depending on the cantilever.Very small tip radii(5-10nm) and stiff cantilevers are necessary for the∼1nm lateral resolution needed for the studies.The images prented here are not filtered.
In tapping mode where the tip makes intermittent contact with the surface.The tapping forces are roughly adjusted by the ratio(R)A eng/A free)of the engaged(A eng)or t point amplitude to the free air amplitude(A free),27keeping the frequency relative to the resonance peak controlled on the low-frequency side32,38to avoid artifacts.This ratio of amplitudes which are ud in feedback control was adjusted to R)0.4-
0.7for“moderate force”imaging(or normal tapping),27,32with
a free air amplitude of∼50(10nm.It is known that the pha lag data are nsitive to local stiffness differences of species or domains in the top veral nanometers from the outermost surface.27,28,32,38
Table1.Summary of AFM Crystal Dimensions and Other
英语俗语
天津效果图制作Lamellar Properties a
L SAXSother than
(nm)
l c,AFM
(nm)b
L AFM
(nm)c
length
(nm)d Na54(Na,10%acid,54%neut)8.856-830
Zn18(Zn,8.7%acid,18%neut)105-71150-100 Zn58(Zn,15%acid,58%neut)7-81240-50 Na59(Na,15%acid,59%neut)85-6815 Nucrel(15%acid,0%neut)8-1015-20100-200
a Neutralizing metal species,weight percent acid in the ba polymer,and percent acid groups neutralized are given.
b l c,AFM)crystal thickness perpendicular to the plane of the lamellae.
c L)spacing or long perio
d between crystals.d Averag
e length o
f crystals in the lon
g direction.
7940Sauer and McLean Macromolecules,Vol.33,No.21,2000
webtrends
By operating at lower amplitudes of oscillation of about 8(3nm,36pha lag data that are innsitive to stiffness with certain systems can be obtained with appropriate experi-mental control.The ratio of tip oscillation amplitudes(R)for low oscillation amplitude imaging conditions were adjusted to the same ratios as for our normal tapping conditions.Recent work has shown that this lower energy applied to the canti-lever allows one to directly image the ionic clusters at or near the surface.36By lowering the oscillation amplitude of the cantilever,it is possible that attractive tip interactions with the polar ionic domains can dominate the AFM pha signal, giving ri to relatively high contrast in the absolute magni-tude of the pha shift.As discusd below,the unneutralized acid form of Surlyn and other controls were ud to prove that polar domains cannot be detected unless they are prent as aggregates containing ionized species in this low oscillation amplitude mode.
A third method utilizes a very light tapping force39and was ud to look for polar domains or contrast bad on hydropho-bicity differences at the outer few angstroms of the surface, i.e.,a clor to true surface-specific imaging technique.This mode is implemented by keeping the free air oscillation amplitude the same as in normal ,50nm),but R is incread substantially to about0.9(0.05reducing the tip
to surface interaction forces substantially.39Again,this method is uful for determining surface com
position of the outermost fraction of a nanometer.Ionic species are not required for contrast in this mode,and the contrast is dominated by hydrophobicity differences and not stiffness differences.39No contrast was obtained in images for any of the Surlyn surfaces imaged using this AFM technique.Water contact angles on fresh Surlyn surfaces demonstrate a surface composition similar to pure polyethylene,so presumably the very light tapping method is also detecting a hydrocarbon layer coating the entire outermost fraction of a nanometer of the surface.
青云翻译Details of the AFM pha imaging methods were prented recently,36and a brief summary is as follows:
1.Normal tapping under moderate forces us stiffness contrast to resolve domains from the top surface down to about 5-10nm.
木已成舟
say you say me2.Low oscillation amplitude tapping is a new method which resolves ionic domains down to∼2-5nm below the surface. Ionic species or domains are necessary for contrast in this method of imaging.
3.Very light tapping39is nsitive to hydrophobicity differ-ences in the outermost fraction of a nanometer of the surface. Ionic species are not necessary for contrast.Certain rough surfaces cannot be studied in this mode.39
Results
Several small-angle X-ray scattering studies have characterized details of morphology including the ionic clusters and the“lamellar”crystals of the ethylene-rich gments in Surlyn.Data for Surlyn Zn18,Na54,and Na59are shown in Figure1(e Table1for composi-tions).The maxima at low angles are the Bragg peaks from the stacking of the ethylene lamellae.Lorentz corrections40,41are needed for the low angle peak becau the lamellae are anisotropic in two dimensions. The correction is simply performed by multiplying the y-axis intensity by the momentum transfer squared,q2 (q)2π/(λ)sin2θ),with2θthe scattering angle.Long periods[L)1.54Å/(2θmax)]obtained are only slightly different for corrected and uncorrected data.Values from corrected data are10,8.8,and8.0nm for Zn18, Na54,and Na59,respectively.The compare well with tho determined directly by AFM for lamellae oriented approximately perpendicular to the surface(Table1). The broad peaks at2θmax∼4°are due to the interionic domain spacings,and for such isotropic(spherical21,22) objects no Lorentz correction should be applied.40Figure 1shows the small shifts in the peaks for such correction procedures,except in the ca of Zn18where no higher angle peak was detected due to the low level of ionic species.Even though no peak is detected in SAXS,the prence of ionic domains is verified by STEM22and also AFM described below,helping to confirm that the“ionic peak”i
n SAXS is from interparticle scattering and not intraparticle.Since the uncorrected data should be ud, the convert to interparticle distances of about L IP) 2.1nm[center-to-center distance)L IP)1.54Å/(2θmax)], and the domain diameters from SAXS have been estimated to be about d)1.7nm for related Surlyn systems42or d)∼2nm from TEM.21,22The relatively large diameter which is comparable to L suggests that ionic domains are quite clo together on , the diameters are large enough that there is reduced space between ionic domains for the two systems with relatively high acid and neutralization levels.The cau of this is the exclusion from the crystalline ethylene-rich lamellae and possibly a partial exclusion of ionic species and domains from the interior of lamellar“stack”regions,causing a concentration of ionic domains outside of the lamellar region as confirmed by AFM data below. Simultaneously obtained AFM height and pha data for Surlyn Zn18were obtained under normal tapping conditions(Figure2A,B)and low oscillation amplitude tapping(Figure2C,D).The total scan boxes for(A)-(D)are500nm×500nm.In the plots white is high in topography and high in pha,and high pha indicates high stiffness regions in normal tapping(Figure2B). Generally,in normal tapping with carefully controlled frequencies relative to the cantilever resonance peak, the hard domains are high regions becau of higher deformability of the soft pha by the tip,28and the stiff regions are also high in pha.27,28The ionic domains are not nd or imaged in the height or pha data obtained under normal tapping conditions(Figure2A,
B). They should reside outside of the hard crystal“lamellae”in softer(darker)amorphous pha regions.The ethyl-ene lamellae show organization into parallel orienta-tions to differing degrees and are relatively well per-fected becau of the low degree of neutralization of this Surlyn.As with most polymers of this type,28,38the outermost layer of the surface is covered with about1 nm of amorphous material.Amorphous polyethylene gments are the low-energy component and probably dominate the outermost surface.We must imagechela clinton
through Figure1.Small-angle X-ray data for samples prepared by cooling from190to25°C.Sample compositions bad on numbers given in the legend can be found in Table1.
Macromolecules,Vol.33,No.21,2000Poly(ethylene-co-methacrylic acid)Ionomers7941
this layer,and one must consider that this may lower the contrast and sharpness of boundaries of lamellae imaged below the surface.
Pha images quentially taken under normal (Fig-ure 2B)followed by low oscillation amplitude (D)condi-tions are distinctly different,while the height data from normal (A)and low oscillation energy (C)tapping are similar although contrast is somewhat better in (A).The identical 500×500nm spot was studied in this comparison.Normal tapping pha data are nsitive to stiffness differences and exhibit mainly the details of lamellar organization in Figure 2B,somewhat similar to the normal tapping height data in Figure 2A.In Figure 2D the contrast is derived from ionic concentra-tion differences near the surface.If ionic domains have the same “stiffness”as amorphous polyethylene,then we should not expect any contrast from ionic domains with AFM in normal “pha”imaging as is found in Figure 2B.To help guide the eye,we have provided arrows that point to the soft regions where lamellae are abnt (darker regions in Figure 2A -C),with the identical points
marked in Figure 2D.The comparison shows that the correspond to bright regions rich in ionic domains.Many ionic domains are clustered into the white regions and cannot be individually resolved in Figure 2D becau they are overlapped with each other.Many parate ones are resolvable as small white dots in Figure 2D and have diameters on the order of 2nm consistent with recent SAXS data and analysis which gave a diameter of 1.7nm for a slightly different Surlyn material 42and with high-resolution TEM which gave diameters of about 2-3nm depending on thermal history.21,22Figure 2E gives a magnified view of the indicated region from Figure 2D,and ionic domain sizes on the order of 2-3nm in diameter are relatively well resolved in veral regions.The contribution of inter-pha species and possibly reduced resolution due to the ionic domains residing at some finite and even variable depth beneath the hydrocarbon layer make the AFM somewhat qualitative.We have taken higher magnifica-tion scans,but the resolution of single ionic domains is not improved significantly,possibly becau of larger degrees of sample deformation becau of the higher residence time of the tip on the surface during
the
Figure 2.Images of a Surlyn Zn18surface in AFM normal (A,B)and low oscillation amplitude (C -E)tapping.The surface was prepared by cooling from 190to 25°C,annealing at 90°C for 2min,before cooling back to room temperature.The height data from normal (A)and low oscillation energy (C)tapping show some similarities,while the pha data in (B)and (D)on the same spots show dramatic differences.Normal tapping pha data (B)are nsitive to stiffness contrast,and low oscillation amplitude tapping (D)derives its contrast from ionic-rich regions just below the surface (e text).Arrows indicate reprentative soft (dark)regions in the normal tapping pha data (B),and the exact corresponding spots are indicated in (D),indicating high populations of ionic species (light)in low oscillation amplitude tapping.The identical 500×500nm spot was studied in (A)-(D),and (E)is a magnified (150nm ×150nm)region from the boxed area in (D).The scales are 0-10nm for height data and 0-20°for pha data.
7942Sauer and McLean Macromolecules,Vol.33,No.21,2000
lateral scanning at a constant scan frequency.This lack of improved resolution in very high magnification AFM scans was also discusd in previous studies of nanop-arated copolymer systems.28
Parts B and D of Figure 2were not taken simulta-neously but were carefully obtained rescans over the same imaged region.The importance of this is that any contributions of thermal drift and other factors which could contribute to an offt would make much of the interpretation invalid.Parts A and B of Figure 2were first obtained simultaneously,then height and pha data were obtained by scanning from the bottom of the imaged region (not shown),and finally parts C and D of Figure 2were obtained by scanning from the top again.The offt could be gauged from the various nanofeatures and was about 3nm for the first two experiments and less than a nanometer between Figure 2A,B compared to Figure 2C,D.Thus,we can prove that thermal drift can be minimal even for very high magnification scans.
Figure 3shows tapping data for Surlyn Na54which has a somewhat different lamellar morphology becau of the higher degree of acid comonomer and especially higher level of neutralization (Table 1)compared to Surlyn Zn18in Figure 2.Simultaneously obtained AFM height and pha data for Surlyn Na54were obtained under normal tapping conditions (Figure 3A,B)and show the dramatically smaller lamellae prent becau of the higher level of acid and neutralization.Each scan box in Figure 3A -D is 500nm ×500nm.The ionic domains are not nd or imaged in Figure 3A,B becau the data were obtained under normal tapping conditions.The ethylene lamellae show organi
zation into parallel orientations to differing degrees and barely show any anisotropy in dimensions.Pha images quentially taken under normal (Figure 3B)followed by low oscillation amplitude (D)conditions show dis-tinctly different contrast and different length-scale morphologies.The height data from normal (A)and low oscillation energy (C)tapping are similar,as was the ca in the ries of experiments in Figure 2.Normal tapping pha data are nsitive to stiffness differences and exhibit mainly the details of lamellar organization in Figure 3B,somewhat similar to the normal tapping height data in Figure 3A.In Figure 3D the contrast is derived from ionic domains near the surface.To help guide the eye,we have provided arrows that point to the soft regions where lamellae are abnt
(darker
Figure 3.Images of a Surlyn Na54surface in AFM normal (A,B)and low oscillation amplitude (C -E)tapping.The surface was prepared by cooling from 190to 25°C,annealing at 90°C for 2min,before cooling back to room temperature.The height data from normal (A)and low oscillation energy (C)tapping show some similarities,while the pha data in (B)and (D)on the same spots show dramatic differences.Arrows indicate reprentative soft (dark)regions in the normal tapping pha data (B),and the exact corresponding spots are indicated in (D),indicating high populations of ionic species (light)in low oscillation amplitude tapping.The identical 500×500nm spot was studied in (A)-(D),and (E)is a magnified (150nm ×150nm)region from the boxed area in (D).The scales are 0-10nm for height data and 0-20°for pha data.
Macromolecules,Vol.33,No.21,2000Poly(ethylene-co -methacrylic acid)Ionomers 7943
