Temperature-Reversible Ultrathin Films of N-IsopropylacrylamideTerpolymer Adsorbed at the Solid-Liquid Interface

Lei Wan,† Harender S. Bisht,‡ Ye-Zi You,‡ David Oupicky,‡ and Guangzhao Mao*,†

Departments of Chemical Engineering and Materials Science and Pharmaceutical Sciences,Wayne State UniVersity, Detroit, Michigan 48202

ReceiVed June 19, 2007. In Final Form: September 4, 2007

This article describes the stability and reversibility of ultrathin films ofN-isopropylacrylamide (NIPA)-vinylimidazole(VI)-poly(ethylene glycol) (PEG) graft terpolymer adsorbed at the solid-liquid interface upon temperature cyclingfrom below to above its phase transition temperature. The coil-to-globule and globule-to-coil phase transitions werecaptured by in situ fluid tapping atomic force microscopy (AFM). The film thickness of 1 nm was determined by AFM,X-ray photoelectron spectroscopy, and X-ray reflectivity. The concentration required to reach full coverage was foundto be higher when adsorption occurred below the phase transition temperature. From 23 to 42°C, the adsorbed NIPAterpolymer film was observed to be molecularly smooth, corresponding to the close-packed structure of flexiblepolymer coils. Particles containing between one and a few globules appeared abruptly at the interface at 42-43 °C,the same temperature as the solution phase transition temperature, which was determined by dynamic light scattering.The size of the particles did not change with temperature, whereas the number of particles increased with increasingtemperature up to 60°C. The particles correspond to the collapsed and associated state of the globules. The filmmorphological changes were found to be reversible upon temperature cycling. Subtle differences were observedbetween dip-coated and spin-coated films that are consistent with a higher degree of molecular freedom for spin-coatedfilms. The study contributes to the fundamental understanding and applications of smart ultrathin films and coatings.

Introduction

Smart temperature-responsive polymers (STRPs) that undergolarge and abrupt physical or chemical changes in response tosmall changes in temperature1,2 are attractive materials for drugdelivery,3-5 biosensing and immunoassays,6-8 biocatalysis,9-11

and size-selective separations.12-14The near-physiological lowercritical solution temperatures (LCSTs) ofN-isopropylacrylamide-(NIPA-) based polymers15make them suitable for bioconjugation.For example, precipitation of genetically engineered elastin-likepolypeptides and NIPA copolymers, induced by a local hyper-thermia at 40-42°C, has been used to selectively increase theiraccumulation in solid tumors.16 It is well-known that, when thesolution temperature is raised above the LCST, the NIPA chaincollapses from a fully hydrated random coil to a dehydratedcompact globule and that the globule goes back to the coil below

the LCST, i.e., reversible coil-to-globule and coil-to-globule phasetransitions occur.17,18Subsequently, the solution phase separatesinto a NIPA-rich phase consisting of aggregated globules anda dilute solution phase. The LCST is driven by entropy to freethe bound water molecules and to promote intramolecularhydrogen bonding and hydrophobic interactions. The LCST ofNIPA-based polymers can be precisely varied in synthesis byintroducing hydrophilic or hydrophobic co-monomers.19,20TheLCST is also sensitive to the ionic strength21 and the presenceof surfactants.22,23

This article focuses on the thin film structure and propertiesof a new class of NIPA terpolymers that incorporate 1-vinylimi-dazole (VI) and poly(ethylene glycol) (PEG) grafted chains.24

It describes the reversible changes in film morphology as afunction of solution temperature, as captured by in situ atomicforce microscopy (AFM). The positive charges of the VImonomers below its pKa allow the polymer to adsorb onto thenegatively charged mica. The weak base VI adds pH sensitivityto the polymer near its pKa (which is 6). The phase transitiontemperature increases to 45°C with increasing VI content andwith decreasing pH because the charged VI units perturb thehydrogen-bonding structure of the hydrating water molecules.24

The LCST of the terpolymer increases from temperatures belowto temperatures above body temperature with a slight decreaseof pH from 7 to 6, rendering it a promising material for biomedicalapplications. The PEG chains prevent aggregation of the polymerparticles in solution. The presence of PEG chains in the structure

* Corresponding author. E-mail: gzmao@eng.wayne.edu.† Department of Chemical Engineering and Materials Science.‡ Department of Pharmaceutical Sciences.(1) Schild, H. G.Prog. Polym. Sci.1992, 17, 163.(2) Gil, E. S.; Hudson, S. M.Prog. Polym. Sci.2004, 29, 1173.(3) Dong, L. C.; Hoffman, A. S.J. Controlled Release1990, 13, 21-31.(4) Kwon, I. C.; Bae, Y. H.; Kim, S. W.Nature1991, 354, 291-293.(5) Qiu, Y.; Park, K.AdV. Drug DeliVery ReV. 2001, 53, 321-339.(6) Monji, N.; Hoffman, A. S.Appl. Biochem. Biotechnol.1987, 14, 107-120.(7) Stayton, P. S.; Shimoboji, T.; Long, C.; Chilkoti, A.; Chen, G. H.; Harris,

J. M.; Hoffman, A. S.Nature1995, 378, 472-474.(8) Kuroda, K.; Swager, T. M.Macromolecules2004, 37, 716-724.(9) Takeuchi, S.; Omodaka, I.; Hasegawa, K.; Maeda, Y.; Kitano, H.Makromol.

Chem.-Macromol. Chem. Phys.1993, 194, 1991-1999.(10) Bhattacharya, S.; Moss, R. A.; Ringsdorf, H.; Simon, J.Langmuir1997,

13, 1869-1872.(11) Bergbreiter, D. E.; Case, B. L.; Liu, Y. S.; Caraway, J. W.Macromolecules

1998, 31, 6053-6062.(12) Ito, Y.; Kotera, S.; Inaba, M.; Kono, K.; Imanishi, Y.Polymer1990, 31,

2157-2161.(13) Iwata, H.; Oodate, M.; Uyama, Y.; Amemiya, H.; Ikada, Y.J. Membr.

Sci.1991, 55, 119-130.(14) Park, Y. S.; Ito, Y.; Imanishi, Y.Langmuir1998, 14, 910-914.(15) Heskins, M.; Guillet, J. E.; James, E.J. Macromol. Sci. A: Chem.1968,

2, 1441.(16) Meyer, D. E.; Shin, B. C.; Kong, G. A.; Dewhirst, M. W.; Chilkoti, A.

J. Controlled Release2001, 74, 213-224.

(17) Goldstein, R. E.J. Chem. Phys.1984, 80, 5340-5341.(18) Kubota, K.; Fujishige, S.; Ando, I.J. Phys. Chem.1990, 94, 5154-5158.(19) Chen, G. H.; Hoffman, A. S.Nature1995, 373, 49-52.(20) Neradovic, D.; Hinrichs, W. L. J.; Kettenes-van, den Bosch, J. J.; Hennink,

W. E. Macromol. Rapid Commun.1999, 20, 577-581.(21) Inomata, H.; Goto, S.; Saito, S.Langmuir1992, 8, 1030-1031.(22) Inomata, H.; Goto, S.; Otake, K.; Saito, S.Langmuir1992, 8, 687-690.(23) Meewes, M.; Ricka, J.; Desilva, M.; Nyffenegger, R.; Binkert, T.

Macromolecules1991, 24, 5811-5816.(24) Bisht, H. S.; Wan, L.; Mao, G.; Oupicky, D.Polymer2005, 46, 7945.

12159Langmuir2007,23, 12159-12166

10.1021/la701819q CCC: $37.00 © 2007 American Chemical SocietyPublished on Web 10/27/2007

of the terpolymers provides them with pH-independent colloidalstabilization to prevent phase separation.

In this work, the interfacial structure was characterized byAFM, X-ray photoelectron spectroscopy (XPS), X-ray reflectivity(XRR), and contact angle goniometry. The temperature-responsive interfacial structure was compared to the polymerbehavior in solution as studied by dynamic light scattering (DLS).The NIPA terpolymer was deposited by dip coating and spincoating onto mica. A thermal unit capable of varying thetemperature in an AFM fluid cell from ambient temperature to60 °C enabled the real-time observation of molecular structuralchanges in the adsorbed film at the solution-mica interface.Real-time AFM captured a structural transition from smooth toparticulate topography in 1-nm-thick films when the temperaturewas raised above 42-43°C. This interfacial transition was foundto be reversible and to coincide with the LCST of the polymerin solution.

Materials and Methods

Materials. Deionized water with resistivity equal to or greaterthan 18 MΩ cm (Nanopure System, Barnstead) was used in solutionpreparation. Grade V5 muscovite mica was purchased from TedPella and hand cleaved just before use. Mica was used as the mainsubstrate in all measurements except XRR where silicon wafer wasused. One-sided polished N-type silicon (100) wafers (test gradewith a resistivity of 1-2 Ω cm and a thickness of 500( 25 µm)were purchased from Wafer World. The precut silicon was rinsedwith ethanol and acetone and then oxidized by the RCA method.25,26

The samples were cleaned at 60°C under ultrasonication in 1:2:8HCl/H2O2/deionized water (by volume) for 30 min and then in 1:2:7NH4OH/H2O2/deionized water (by volume) for another 30 min.Phosphate buffer was purchased from Fisher Scientific and used asreceived.N-Isopropylacrylamide (NIPA), 1-vinylimidazole (VI),poly(ethylene glycol) methyl ether methacrylate with averagemolecular weight 2080 (PEGMA, 50% solution in water), andazobisisobutyronitrile (AIBN) were purchased from Aldrich.

NIPA Terpolymer Synthesis. NIPA (0.67 M), VI (0.078 M),PEGMA (15.3 mM), and AIBN (7 mM) were dissolved in acetone,transferred into a glass ampule, and deoxygenated by a stream ofN2. The ampule was sealed, and polymerization was carried out ina water bath at 60°C for 24 h. The reaction mixture was then addeddropwise to cold diethyl ether. The precipitated polymer was isolatedby filtration, reprecipitated from methanol solution, and driedovernight in a vacuum before analysis. The composition of theterpolymer was determined in D2O solutions by1H NMR analysisusing a Varian 400 MHz NMR spectrometer. Areas of-N-CHproton of NIPA (∼4 ppm), protons of the VI cycle (-CH, ∼7.2ppm), and methylene protons of PEG (3.6 ppm) were used to calculatethe composition of the terpolymer. The VI content was also confirmedby titration with 0.01 M NaOH. The number-average (Mn) and weight-average (Mw) molecular weights and the polydispersity (Mw/Mn) ofthe polymer were determined by size-exclusion chromatography(SEC) using a Shimadzu LC-10ADVP liquid chromatographequipped with a Shimadzu CTO-10ASVP column oven and a WatersStyragel HR-4E 7.8×300 mm column. The system was also equippedwith a seven-angle BIMwA static light-scattering detector and aBIDNDC differential refractometer (both from Brookhaven Instru-ments). The BIMwA detector was equipped with a 30 mW verticallypolarized solid-state laser (660 nm) as a light source. Tetrahydrofuranwas used as the eluent at a flow rate of 1 mL/min at 30°C. GPCdata were analyzed using PSS WinGPC Unity software from PolymerStandards Services.

Film Preparation. The NIPA terpolymer was dissolved in 50mM phosphate buffer (pH 5) to obtain 1 g/L stock solution. Thestock solution was diluted with deionized water to obtain different

concentrations with pH values between 4.8 and 5.0. The NIPAterpolymer films were prepared using either dip coating or spincoating at room temperature (23°C). Some films were made at 55°C, which is above the LCST. In dip coating, freshly cleaved micawas immersed in the terpolymer solution for 30 min, rinsed withdeionized water, and dried with N2. In spin coating, 100µL ofterpolymer solution was placed on the substrate, and after a 1-minstanding time, the substrate was spun at 3000 rpm for 1 min(PM101DT-R485 photo resist spinner, Headway Research). Thespin-coated film was rinsed with deionized water to remove excesspolymer and dried with N2. In XRR experiments, an RCA-cleanedsilicon wafer was used instead of mica.

Dynamic Light Scattering (DLS). DLS was used to determinethe hydrodynamic diameter andú potential of the terpolymer insolution as a function of temperature. A ZetaPlus Particle Size andZeta Potential Analyzer (Brookhaven Instruments) equipped witha 35 mW solid-state laser (658 nm) was used. Scattered light wasdetected at 90°. The LCST of the terpolymer was estimated fromthe onset temperature in the scattering intensity versus temperaturecurve using a polymer concentration of 0.5 g/L (pH 5). Meanhydrodynamic diameters were calculated for the size distribution byweight, assuming a log-normal distribution using the suppliedalgorithm, and the results are expressed as mean( standard deviationof three runs.ú potential values were calculated from measuredvelocities using the Smoluchowski equation, and the results areexpressed as mean( standard deviation of 10 runs.

X-ray Photoelectron Spectroscopy (XPS).XPS analysis of thethin film samples on mica was conducted with a PHI 5500spectrometer (Perkin-Elmer) equipped with an aluminum KR X-rayradiation source (1486.6 eV) and an AugerScan system control (RBDEnterprises). The pressure in the chamber was below 2× 10-9 Torrbefore the data were taken, and the voltage and current of the anodewere 15 kV and 13.5 mA, respectively. The takeoff angle was setat 45°. The pass energies for survey and multiplex scans were 117.40and 23.50 eV, respectively. The binding energy scale was referencedby setting the C 1s peak maximum at 285.0 eV.

X-ray Reflectivity (XRR). Polymer films deposited on siliconwafers were studied with a SmartLab high-resolutionθ/θ XRD systemusing Cu KR radiation (λ ) 1.54 Å) (Rigaku). The scan range was0-10° with a step size of 0.01° and a speed of 1°/min. The incidentand receiving slit sizes were 0.05 and 0.25 mm, respectively. Thereflectivity data were fitted with established algorithms to obtain thefilm thickness.27,28

Contact Angle Goniometry. The contact angle was measuredwith an NRL contact angle goniometer (model 100, Rame-Hart) inthe laboratory atmosphere. A 20-µL droplet of the sample was placedon the substrate, and contact angles were read on both sides of thedroplet. Five droplets were placed at various spots near the centerof the substrates, and contact angles were averaged (with a typicalerror of measurement of(3°).

AFM. AFM imaging was conducted using a Nanoscope IIIMultiMode atomic force microscope equipped with a type Epiezoelectric scanner with maximum scan ranges of 10µm (X andY directions) and 2.5µm (Z direction) from VEECO/DigitalInstruments. Ex situ AFM imaging of samples was conducted intapping mode (oscillation frequency≈ 250-300 kHz) in ambientatmosphere using etched silicon probes (TESP, VEECO) with anominal radius of curvature of less than 10 nm. In situ AFM imagingwas conducted in liquid tapping mode (oscillation frequency≈ 8kHz, line scan rate) 2-3 µm/s) using silicon nitride probes (NP,VEECO) with a radius of curvature of 20 nm and a cantilever springconstant of 0.38 N/m as provided by the manufacturer. In situ AFMimaging of the phase transition was conducted using a temperaturecontroller (model 2216e, Eurotherm) capable of controlling the fluidtemperature in the AFM fluid cell from room temperature to 60°C.The AFM scanner was calibrated separately when used with thetemperature controller. The heating unit, placed directly under the

(25) Kern, W.J. Electrochem. Soc.1990, 137, 1887-1892.(26) Wu, B.; Mao, G.; Ng, K. Y. S.Colloids Surf. A: Physicochem. Eng.

Aspects2000, 162, 203-213.

(27) Klappe, J. G. E.; Fewster, P. F.J. Appl. Crystallogr.1994, 27, 103-110.(28) Dane, A. D.; Veldhuis, A.; de Boer, D. K. G.; Leenaers, A. J. G.; Buydens,

L. M. C. Physica B1998, 253, 254-268.

12160 Langmuir, Vol. 23, No. 24, 2007 Wan et al.

fluid cell but isolated from the scanner by a spacer block, allowedfor rapid changes (approximately 4°C/min for heating and 2°C/minfor cooling) of the solution temperature. Only height images areshown. Height images were plane-fit in the fast scan direction withno additional filtering operation. The surface roughness of the filmswas determined using the root-mean-square surface roughnessRq

) x(∑zi2/n), wherezi is the height value andn is the number of

pixels in the image.

Results and Discussion

Temperature-Responsive Structure in Solution.The ter-polymer used here contained 12 mol % VI and 1 mol % PEGgrafts (Figure 1) according to1H NMR analysis. The VI contentwas also confirmed by titration with 0.01 M NaOH. Althoughgel formation due to unwanted cross-linking during the copo-lymerization of NIPA and PEG methacrylates has been reported,29

our SEC analysis showed a monomodal distribution of molecularweights with no evidence of gel formation (Figure 1). The weight-average molecular weight (Mw) of the prepared terpolymer was2.06 × 105, and the polydispersity (Mw/Mn) was 2.51. Therelatively high polydispersity of molecular weights was due tosignificant differences in the reactivity of PEG macromonomerand NIPA and VI monomers.30

The phase transition temperature of the terpolymer wasdetermined from the temperature dependence of the scatteringintensity measured at a 90° scattering angle from 0.5 g/L solutionat pH 5 in a glass cuvette. The temperature was increased stepwisein 1 °C increments in the range of 25-55 °C, and scatteringintensity and mean hydrodynamic radius were measured after 3min of equilibration at each temperature. The phase transitiontemperature at 40-43 °C was determined by the sharp increaseof the scattering intensity in Figure 2a. Whereas the terpolymerexists as a unimer solution below the transition with an averagehydrodynamic diameter of around 25 nm, it forms particles withsizes in the range of 44-55 nm. The size decrease above thephase transition temperature was not accompanied by a corre-sponding decrease in scattering intensity, suggesting furtherassociation of more densely packed particles. Measurement oftheú potential (Table 1) revealed that the particles are positivelycharged, suggesting localization of the charged VI units on thesurface of the coils and associated globules. The positive chargesenable the adsorption of the NIPA polymer on negatively chargedmica.

Film Thickness Measurements.The previously studied NIPApolymer films generally had film thicknesses above 10 nm.31-37

This study focuses on the structural transition exhibited in ultrathinNIPA films with thicknesses approaching 1 nm, which meansthat the polymer chains were confined to a 2-D adsorbed statewith the majority of segments in close proximity to the substrate.The film thickness was significantly below the radius of gyrationof the NIPA terpolymer coils in solution. The film thicknessinfluences the extent of conformational transition of NIPApolymers at the interface. AFM, XPS, and XRR were used todetermine the film thickness below the LCST. The concentrationused for all three methods was 0.05 g/L. We estimated a minimumfilm thickness of 0.8-0.9 nm for the dip-coated film on micaby an AFM scratch test. The scratch test was conducted on anarea of 400× 400 nm2 by scanning with an AFM tip in contactmode at 3 Hz and a constant force between 100 and 150 nN for10 min. The film thickness was determined by AFM sectionalheight analysis of the height difference between the previouslyuntouched and the scratched area. The film thickness was alsoestimated by an XPS analysis method from the literature.38 Thepresence of the NIPA polymer was indicated by the appearanceof the N 1s peak (399.5 eV). Assuming a homogeneous film, theN/Si molar ratio is a function of the film thickness and degreeof surface coverage. The thickness thus computed from thevariation in N/Si ratio with concentration was 0.8-0.9 nm fora dip-coated film on mica.39The thickness of the dip-coated filmon an RCA-treated silicon wafer was determined to be 1.3( 0.3nm by XRR. The film thickness was found to be lower for thespin-coated films, 0.8( 0.2 nm, also determined by XRR. Thissmall but finite difference in film thickness for the first layernext to the substrate between the spin-coated and dip-coatedsamples is opposite to what has been reported in the layer-by-

(29) Virtanen, J.; Baron, C.; Tenhu, H.Macromolecules2000, 33, 336-341.(30) Gramm, S.; Komber, H.; Schmaljohann, D.J. Polym. Sci. A: Polym.

Chem.2005, 43, 142-148.

(31) Suzuki, A.; Yamazaki, M.; Kobiki, Y.; Suzuki, H.Macromolecules1997,30, 2350-2354.

(32) Zhu, P. W.; Napper, D. H.Phys. ReV. E 1998, 57, 3101-3106.(33) Yakushiji, T.; Sakai, K.; Kikuchi, A.; Aoyagi, T.; Sakurai, Y.; Okano,

T. Langmuir1998, 14, 4657-4662.(34) Lee, L. T.; Jean, B.; Menelle, A.Langmuir1999, 15, 3267-3272.(35) Fu, Q.; Rao, G. V. R.; Basame, S. B.; Keller, D. J.; Artyushkova, K.;

Fulghum, J. E.; Lopez, G. P.J. Am. Chem. Soc.2004, 126, 8904-8905.(36) Sun, T.; Wang, G.; Feng, L.; Liu, B.; Ma, Y.; Jiang, L.; Dhu, D.Angew.

Chem., Int. Ed. 2004, 43, 357-360.(37) Cheng, X. H.; Canavan, H. E.; Stein, M. J.; Hull, J. R.; Kweskin, S. J.;

Wagner, M. S.; Somorjai, G. A.; Castner, D. G.; Ratner, B. D.Langmuir2005,21, 7833-7841.

(38) Callewaert, M.; Grandfils, C.; Boulange-Petermann, L.; Rouxhet, P. G.J. Colloid Interface Sci.2004, 276, 299-305.

Figure 1. Chemical structure of the NIPA terpolymer and itsmolecular weight distribution as determined by size-exclusionchromatography.

Figure 2. Hydrodynamic radius dependence on temperature(pH 5).

Table 1. Dependence ofú Potential in 0.5 g/L NIPA TerpolymerSolution (pH 5) on Temperature

temperature(°C)

ú potential(mV)

25 8.9( 0.740 8.2( 1.448 20.4( 4.1

Temperature-ReVersible Films of NIPA Terpolymer Langmuir, Vol. 23, No. 24, 200712161

layer (LbL) multilayer film literature.40,41 In LbL films, spin-coated films generally exhibit a higher film thickness than dip-coated films. It should be pointed out that, in all of these cases,similarly to dip coating, the substrate containing the spin-coatedfilms was thoroughly rinsed with water after deposition of eachlayer to remove nonadsorbing polymers. In conclusion, all of thefilm thickness values point to a very thin layer of 2-D coils forthe adsorbed NIPA terpolymer. Therefore, unlike previous studiesof NIPA phase transition at an interface where the majority ofthe chain segments were free from the substrate, almost all ofthe NIPA terpolymer segments can be assumed to be in directcontact with the substrate in this case via electrostatic interactionsbetween the positively charged VI unit and the negatively chargedmica.

Temperature-Responsive Wettability.The contact angle ofwater was measured on films spin-coated from 0.05 g/L polymersolution on mica and on a silicon wafer. The contact angles weremeasured below (23°C) and above (55°C) the LCST. For thecontact angle measurements above the LCST, the films wereimmersed and allowed to swell in water at 55°C for 20 min andthen left in the oven to dry at 55°C before measurements. Thecontact angle of the film on a silicon wafer changed from 40.2°( 2.0° below the LCST to 54.4° ( 5.0° above the LCST. Anincrease in contact angle is expected as a polymer acquires amore hydrophobic nature above its LCST. However, the contactangle measured for the mica-supported film decreased from40.4° ( 2.2° below the LCST to 33.3° ( 2.8° above the LCST,probably as a result of exposure of the bare mica substrate afterthe coil-to-globule transition.

Temperature-Responsive Structure at the Solid-LiquidInterface. The adsorption behavior of the NIPA terpolymer onmica as a function of solution concentration and temperaturewas studied by ex situ and in situ AFM. The film coverage wasmonitored to determine the concentration range necessary forfull coverage. Figure 3a-d shows adsorbed films on mica withincreasing concentration from 10-4 to 10-2 g/L (pH 5) and anadsorption temperature of 23°C. The films were rinsed, dried,and imaged by AFM in tapping mode in air. The features inFigure 3a correspond to individual polymer coils with an averagediameter of 12.0 nm. The heights of the features, 0.3-0.5 nm,are similar to the thicknesses of individual polymer coils flat-tened by surface adsorption as reported in the literature.42-45

Occasionally, highly stretched coils or chains were observed. At5 × 10-4 g/L (Figure 3b), the coils associated with each otherto form larger domains that coexisted with remaining individualcoils. The height of the domains was identical to that of the coils.Adsorption at 10-3 g/L yielded smooth and full monolayercoverage (Figure 3c). The root-mean-square surface roughness,

Rq, was 0.2 nm in a 1× 1 µm2 area. It is evident that 1× 10-3

g/L is close to the minimum concentration necessary for fullcoverage because only a few dark spots existed in the otherwisecontinuous film. Films adsorbed at concentrations above 1×10-3g/L remained smooth and continuous (Figure 3d). Adsorptionat a lower pH (i.e., 3.0) using 0.01 M phosphate buffer producedstretched chains (Figure 3e) whose contour length was estimatedusing WSxM software (Nanotec Electronica, version 4.0). Themeasured contour length of 88.0( 35.0 nm is significantly lowerthan the calculated value of 448 nm for NIPA terpolymer witha molecular weight of 2.06× 105 g/mol.46 Such a discrepancybetween AFM-determined and theoretical values is often observedand attributed to the combined effect of the necklace chainstructure and the spatial resolution limit of the AFM.43,47 Theside chains were invisible in these AFM images, probably becauseof its low percentage (<5 mol %) and short length (<5.3 nm).However, most chains were in the coil state at pH 5. The coilsize measured on films prepared from 2× 10-4 g/L solution atpH 5 was 0.4( 0.1 nm in height and 23.0( 7.0 nm in widthaccording to AFM sectional height analysis. The radius of

(39) To determine the film thickness, we used a method provided by reference38. The relative sensitivity factors wereiN ) 0.477 andiSi ) 0.328, and thephotoionization cross sections wereσN ) 1.8 andσSi ) 0.817 as provided by theXPS manufacturer. The elemental concentrations in the adsorbed polymer filmand mica substratum were calculated to be CN

ad ) 10.46 M and CSisu ) 16.68

M, respectively. The inelastic electron mean free paths were calculated to beλSisu

) 3.33 nm in mica andλSiad ) 3.99 nm andλN

ad ) 3.11 nm in the polymer film,assuming a mica density of 2.8 g/cm3 and a polymer density of 1.127 g/cm3. TheN-to-Si molar ratio in the film was determined to be 4.9%:12.3%. The filmthickness at full coverage was obtained from the surface coverage versus thicknessplot.

(40) Cho, J.; Char, K.; Hong, J. D.; Lee, K. B.AdV. Mater.2001, 13, 1076-1078.

(41) Chiarelli, P. A.; Johal, M. S.; Casson, J. L.; Roberts, J. B.; Robinson, J.M.; Wang, H. L.AdV. Mater. 2001, 13, 1167-1171.

(42) Minko, S.; Kiriy, A.; Gorodyska, G.; Stamm, M.J. Am. Chem. Soc.2002,124, 3218-3219.

(43) Kiriy, A.; Gorodyska, G.; Minko, S.; Jaeger, W.; Stepanek, P.; Stamm,M. J. Am. Chem. Soc.2002, 124, 13454-13462.

(44) Kumaki, J.; Hashimoto, T.J. Am. Chem. Soc.2003, 125, 4907-4917.(45) Minko, S.; Roiter, Y.Curr. Opin. Colloid Interface Sci.2005, 10, 9-15.

(46) Calculation of the polymer contour length was performed as follows:The molecular weight of the NIPA polymer was 2.06× 105 g/mol. The averagemolecular weight of one unit was 115 g/mol when the polymer contained 87%NIPA, 12% VI, and 1% PEG. Thus, we obtained an average degree ofpolymerization of 1790, which corresponds to an average chain length of 448 nm.

(47) Kirwan, L. J.; Papastavrou, G.; Borkovec, M.; Behrens, S. H.Nano Lett.2004, 4, 149-152.

Figure 3. AFM images (tapping mode in air) of NIPA terpolymerfilms made at room temperature from different concentrations: (a)1 × 10-4, (b) 5 × 10-4, (c) 1 × 10-3, and (d) 1× 10-2 g/L. Thefilms were dip-coated on mica at 23°C. Thez range is 4 nm. (e)AFM height image (tapping mode in deionized water) of a spin-coated film on mica at room temperature from 5× 10-4 g/L andphosphate buffer (pH 3). Thez range is 3 nm. The scan size of theinset image is 115 nm.

12162 Langmuir, Vol. 23, No. 24, 2007 Wan et al.

gyration,Rg, for a NIPA homopolymer coil in dilute solution canbe related to its weight-average molecular weight,Mw, by theexpressionRg ) 0.011Mw

0.59 (e.g., a polymer of 2× 105 g/molshould have anRg of 148 Å).48 The hydrodynamic diametermeasured by DLS was around 25 nm. The coil dimensionsmeasured by AFM contain the tip-convoluted portion and alsoreflect significant coil flattening upon adsorption.

The adsorption of the NIPA polymer on mica was also studiedat 55 °C, above its LCST. Adsorption and rinse cycles werecarried out with solutions at 55°C. The film morphology wasstudied by AFM after film drying. Figure 4 presents the surfacefeatures of the films made in the concentration range between1 × 10-4 and 1× 10-2 g/L above the LCST. No chain-likefeatures were observed above the LCST. Particles with heightsof 0.7( 0.1 nm and widths of 10.1( 3.5 nm were seen to coexistwith larger domains at 1× 10-4 g/L (Figure 4a). The NIPAglobules were reported to contain a significant amount of water(∼66%) with a density of 0.34 g/cm3 in solution.49 The diameterof a single NIPA globule with a molecular weight of 2× 105

g/mol was therefore estimated to be 6.2 nm. Again, the AFM-determined globule size at the interface does not exactly matchits expected size in solution. The mica surface appeared to befully covered by polymer globules at 5× 10-4 g/L (Figure 4b),a lower saturation concentration than required for adsorptionbelow the LCST. The early onset of full surface coverage is dueto the significant decrease in solvent quality above the LCST.The surface roughness,Rq, in Figure 4b was 0.2 nm whenmeasured in a 1× 1 µm2 area. A further increase in solutionconcentration resulted in the adsorption of large aggregatedglobules onto the initial smooth layer (Figure 4c). The smootharea had the same surface roughness as Figure 4b. The particleshad an average diameter of 20.3( 10.0 nm and a height of 2.8( 1.0 nm. This suggests that, at these concentrations, the globulesare no longer stable in solution against coalescence and someof the aggregates precipitate onto the smooth first layer.

To monitor the coil-to-globule and globule-to-coil phasetransitions of the NIPA terpolymer at the solid-liquid interface,we employed a heating stage to gradually and precisely vary thesolution temperature of the AFM fluid cell while capturing theresponse of the adsorbed NIPA polymer film to the temperaturechange. The temperature was changed by 1-5 °C at a time from23 °C and maintained at the final temperature until sufficientAFM data were collected, which generally took between 0.5 and1 h. Data also were collected during the cooling cycle after amaximum temperature of 56°C had been. Surface features didnot change during the measurement period, indicating that thesurface features were stable and that the structural changesoccurred rapidly (<5 min). The abrupt transition with respect totemperature and time was consistent with the transition occurringin solution. Figure 5 is a sequence of images captured during oneheating and cooling cycle in 1× 10-2 g/L NIPA terpolymersolution. In this case, the experiment was carried out completelyin the AFM fluid cell, including the time for polymer adsorptionto make the initial layer and subsequent heating and coolingcycles in the same polymer solution. For comparison purposes,heating and cooling experiments done in pure water on samplesprepared by dip coating prior to their placement below the AFMfluid cell showed essentially the same results as described below,further demonstrating the stability of the films. Above itssaturation concentration, the NIPA terpolymer formed a smoothlayer on mica after 30 min of incubation with a typicalRq valueof 0.2 nm for a 1× 1 µm2 area (Figure 5a). The film remainedfeatureless from 23°C to 40°C (data not shown). At 40°C, afew particles appeared with a surface coverage density about 6particles/µm2(Figure 5b). More particles emerged with increasingtemperature and reached coverage density of 40 particles/µm2

at 50°C (Figure 5c-e). Subsequently, the solution temperaturewas gradually reduced from 50°C to room temperature, and theparticle density was observed to decrease with decreasingtemperature, reaching a value of 4 particles/µm2 at 40°C. Noparticles remained below 40°C. The particle size did not changewith temperature. Therefore, the particle nucleation appears tobe activated by heat, but the nucleated particles show littletendency for further association or growth. The particle size wasdetermined to be 3.6( 1.5 nm in height and 25.0( 15.0 nmin width by sectional height analysis. Assuming a density of 0.34g/cm3, we estimated that each particle contained one or a smallnumber of molecules (<3.6 molecules). Therefore, the mor-phological transition observed here represents a molecularconformational transition at the single-molecule level from 2-DNIPA chains. The same results were again obtained in a secondheating and cooling cycle, indicating the reversibility and stabilityof the adsorbed film. The small particle size can be attributedto the limited degree of freedom from physical adsorption andthe colloidal stabilization by the charged VI units and PEG chainsin the terpolymer. In the coil-to-globule transition, the hydrophilicsegments prefer the surface region of the particle, whereas themore hydrophobic NIPA makes up the core region.

The same in situ heating/cooling experiments were conductedon the spin-coated films. Figure 6 presents the image sequencecaptured during one heating and cooling cycle on the spin-coatedfilm in deionized water. In the temperature range between roomtemperature and 42°C, the film appeared to be largely flat, withparticles appearing suddenly between 42 and 43°C with Rq )0.2 nm for a 1× 1 µm2 area (Figure 6b). The number of particlesincreased with increasing temperature above 42°C. The particlesize, 2.7( 1.0 nm in height and 35.0( 15.0 nm in width, whichtranslates to 5.2 molecules per particle, was unchanged withtemperature. At 42°C, the film returned to the flat morphology

(48) Zhang, J.; Pelton, R.Colloids Surf. A: Physicochem. Eng. Aspects1999,156, 111-122.

(49) Wu, C.; X., W.Phys. ReV. Lett. 1998, 80 4092-4094.

Figure 4. AFM images (tapping mode in air) of the NIPA terpolymeron mica dip-coated at 55°C from aqueous solutions of differentconcentrations: (a) 1× 10-4, (b) 5× 10-4, and (c) 1× 10-2 g/L.Thez range is (a,b) 5 and (c) 20 nm. The arrows in image a pointto individual globules.

Temperature-ReVersible Films of NIPA Terpolymer Langmuir, Vol. 23, No. 24, 200712163

and remained unchanged down to 23°C. Figure 7 presents thenumber of particles measured in AFM images as a function oftemperature for both dip-coated and spin-coated films during thefirst heating and cooling cycle. A higher particle density wasfound on the spin-coated film than the dip-coated film, eventhough the spin-coated film was thinner according to the XRRmeasurement. It is interesting that such a large difference wasfound for films with similar film thicknesses and initial surface

roughness values. The difference between spin coating and dipcoating suggests a higher degree of molecular freedom in spin-coated than dip-coated films. It is possible that more ionic bondsare established during the dip-coating process. This differencecan be used to further tune the responsive properties of the surface-attached molecules. It should be mentioned that the number ofparticles decreased somewhat during the second heating and

Figure 5. AFM images (fluid tapping mode) of mica immersed in1 × 10-2 g/L polymer solution during one heating and coolingcycle: (a) at room temperature after 30-min incubation; after beingheated to (b) 40, (c) 43, (d) 46, and (e) 50°C; and after being cooledto (f) 46, (g) 43, (h) 40, and (i) 30°C. Thez range is 10 nm.

Figure 6. AFM images of a spin-coated NIPA terpolymer film(deposition concentration) 5 × 10-2 g/L) immersed in deionizedwater during one heating and cooling cycle: (a) at room temperature;after being heated to (b) 43, (c) 46, (d) 48, and (e) 49°C; and afterbeing cooled to (f) 46, (g) 44, (h) 43, (i) 42, and (j) 30°C. Thezrange is 10 nm.

12164 Langmuir, Vol. 23, No. 24, 2007 Wan et al.

cooling cycle, indicating some degree of film material loss. Thiscould be due to the fact that the substrate in this case was immersedin water and not the polymer solution.

The coil-to-globule transition experienced by the surface-adsorbed NIPA terpolymer is summarized here. As temperatureis raised above 42-43 °C, the 2-D closely packed coils changeinto compact globules. The number of single globules oraggregates of a few globules increases with increasing temperaturewhile their size remains constant. The process is completelyreversible during one heating and cooling cycle.

Reversible coil-to-globule and globule-to-coil transitions havebeen observed in extremely dilute solutions by light scattering.49,50

However, observation of the coil-to-globule transition in solutionis challenging because it requires high-molecular-weight sampleswith nearly monodisperse molecular weight distributions. Inaddition, the observation of individual completely collapsedglobules could be prevented by the onset of molecular aggregationand phase separation prior to the completion of chain collapse.Therefore, the AFM technique offers a feasible tool for the under-standing of coil-to-globule phase behavior in polymer systems.

Our results show that the incorporation of a positively chargedco-monomer, in this case, the VI unit, enhances the stability ofthe NIPA film without compromising its ability to undergo aconformational transition. On the other hand, when the polymeris adsorbed through hydrogen bonding between the NIPA unitand the surface silanol group of the silica particle, the LCST inthe adsorbed layer was shown to be completely suppressed forvery thin films.51Without any strongly adsorbing units, a previousstudy of NIPA homopolymer films dip-coated onto glass showedthat the lack of shear control upon rinsing could result in poorreproducibility of the coil, globule, and aggregate particledistributions.38 The aggregate size above the LCST was muchgreater in the NIPA homopolymer films. It is likely that, whenthe terpolymer undergoes the coil-to-globule transition, thecharged VI units and hydrophilic PEG grafts migrate toward theparticle surface because of their incompatibility with thehydrophobic core. Theú potentials provide evidence of a VI-rich corona layer above the LCST. The otherwise expected particleaggregation due to an increase in van der Waals forces withparticle size increase is thus prevented by the existence of thecorona layer rich in VI and PEG. A recent study showed aninteresting “memory” property of biotinylated amino-NIPA onmica.52 The position and shape of the aggregated domains weremaintained during subsequent heating and cooling cycles. Thenumber of layers was more than one, and the aggregated domain

size was in the tens to hundreds of nanometers. In our study, wewere not able to identify any positional memory of the nucleatedparticles because of the small size and high density of the particles.In addition, the AFM imaging of spin-coated films in the polymersolution above the LCST showed no evidence of molecularprecipitation from solution. This and the constant and smallparticle size demonstrate the effectiveness of the VI and PEGin enhancing the colloidal stability of NIPA polymers.

The increase in particle number with increasing temperaturecould be due to compositional heterogeneity in the sample. Theterpolymer had a wide molecular weight distribution with apolydispersity of 2.51. This compositional heterogeneity arisesfrom the significant differences in the reactivity of PEGmacromonomers and NIPA and VI monomers.30 In our previousstudy of the solution properties of a terpolymer series,24we foundthe phase transition temperature to be dependent on both thetotal VI content and the VI degree of ionization. The slightlyacidic pKa (∼6) of the terpolymer makes the phase transitiontemperature particularly sensitive to the VI unit. Block copolymersof NIPA with hydrophilic polymers such as PEG often exhibitphase transition temperatures similar to those of NIPA homo-polymers because the cooperative domains in NIPA blocks thatundergo phase transitions are not significantly perturbed by theother component.53,54However, if the sample contains polymerchains with a PEG composition of higher than 2%, this couldresult in an increase of the LCST by 5°C.24 The AFM techniquethus shows promise in revealing a degree of heterogeneity in thecoil-to-globule phase transition that cannot easily be extractedfrom light scattering data.

The specific polymer architecture of the NIPA terpolymerwith 12% VI and PEG grafts provides an ideal balance inmaintaining film stability and temperature-responsive properties.The adsorbed ultrathin films exhibit rapid and reversible coil-to-globule transitions in response to small changes in tempera-ture and stability against in-plane and out-of-plane molecularaggregation. These attributes should make them desirable forapplications in drug and gene delivery and as molecular switchesof protein activity.

Conclusions

Our in situ AFM study shows the transition between coilsbelow the LCST and globules above the LCST of a NIPAterpolymer that parallels its bulk solution behavior. It suggeststhat the NIPA segment retains its freedom to undergo the coil-to-globule transition with the VI acting as an anchoring segment.The AFM results also indicate that the electrostatic interactionbetween the positively charge VI and the negatively chargedmica is strong enough for the film to withstand temperaturecycling. The polymer architecture ensures an ideal force balanceto maintain film integrity and responsive properties. AFM is apowerful tool for the study of polymer and biopolymer structuresand dynamics at surfaces because it offers both high spatialresolution approaching 1 Å and a solution environment withminimal disturbance to the hydrated state. The phase transitionat the LCST is closely related to the state of association of watermolecules, and thus, it is desirable to monitor the NIPA-basedpolymer structure and dynamics in an aqueous environment inwhich abundant water molecules are free to associate or dissociatewith the NIPA segments. In addition, the drying process is knownto cause unevenness in film topography due to dewetting and to

(50) Wang, X.; Qiu, X.; Wu, C.Macromolecules1998, 31, 2972-2976.(51) Petit, L.; Bouteiller, L.; Brulet, A.; Lafuma, F.; Hourdet, D.Langmuir

2007, 23, 147-158.(52) Pelah, A.; Luduena, S. J.; Jares-Erijman, E. A.; Szleifer, I.; Pietrasanta,

L. I.; Jovin, T. M. Langmuir2006, 22, 9682-9686.

(53) Virtanen, J.; Holappa, S.; Lemmetyinen, H.; Tenhu, H.Macromolecules2002, 35, 4763-4769.

(54) Oupicky, D.; Reschel, T.; Konak, C.; Oupicka, L.Macromolecules2003,36, 6863-6872.

Figure 7. Variation in the number of particles per unit area as afunction of temperature for dip-coated and spin-coated films.

Temperature-ReVersible Films of NIPA Terpolymer Langmuir, Vol. 23, No. 24, 200712165

perturb the natural state of biopolymers. Thus, in situ AFMimaging is highly relevant to the eventual biomedical applicationsof temperature-responsive polymers.

Acknowledgment. This work was supported partially by NSFGrants CTS-0221586 and CTS-0553533 (G.M.) and NIH Grant

EB0043588 from the National Institute of Biomedical Imagingand Bioengineering (D.O.). The authors also acknowledge thefinancial support of a grant for nanoresearch from the Office ofthe Vice President for Research at Wayne State University.

LA701819Q

12166 Langmuir, Vol. 23, No. 24, 2007 Wan et al.