Langmuir 2001, 17, 349-360

349

Interpretation of Contrast in Tapping Mode AFM and Shear Force Microscopy. A Study of Nafion
P. J. James, M. Antognozzi,* J. Tamayo, T. J. McMaster, J. M. Newton, and M. J. Miles
University of Bristol, H.H. Wills Physics Laboratory, Tyndall Avenue, Bristol, BS8 1TL, U.K. Received March 6, 2000. In Final Form: November 1, 2000
The origin of phase contrast in tapping-mode atomic force microscopy has been investigated using two complementary scanning probe microscopy techniques, atomic force microscopy and shear force microscopy, which can be classified as a transverse dynamic force microscopy. The sample chosen for this study was Nafion, and specifically the membrane in different hydration states by virtue of its cation form. Differences in probe-sample adhesion throughout a sample, caused by an inhomogeneous distribution of surface water, were an important phase-contrast mechanism. A new variant in three-dimensional force imaging, phase-volume imaging has been a useful tool in the interpretation of phase contrast. With the use of transverse dynamic force microscopy, approach curves were obtained while the frequency spectrum around resonance was measured. This enabled the damping of the probe oscillation amplitude and the shift in its resonant frequency to be decoupled. Knowing the true oscillation amplitude of the probe, it was also possible to determine quantitatively the elastic and dissipative parts of the probe-sample interaction. Distinct regimes were found at different probe-sample separations.

Introduction Tapping-mode atomic force microscopy (AFM) is more suitable than contact mode for imaging delicate samples because of the lower lateral forces. It has been applied to many polymer systems.1-4 It also has the added advantage of being able to obtain phase images and topographical data. Tapping-mode phase imaging is a relatively new AFM technique. It can differentiate between areas with different properties regardless of their topographical nature.5-7 The phase angle is defined as the phase lag of the cantilever oscillation relative to the signal sent to the piezo driving the cantilever. Transverse dynamic force microscopy (TDFM) is a dynamic probe microscopy in which the detected force is perpendicular to the probe, hence transverse. The first use of this technique has been the shear force microscope (ShFM) as a distance control mechanism in scanning nearfield optical microscopy (SNOM).8,9 In the past few years,
Present address: National Power Innogy, Harwell International Business Centre, Harwell, Didcot, OX11 0QA, U.K.

(1) McMaster, T. J.; Hobbs, J. K.; Barham, P. J.; Miles, M. J. AFM Study of in situ Real Time Polymer Crystallization and Spherulite Structure. Probe Microscopy 1997, 1(1), 43-56. (2) Hobbs, J. K.; McMaster, T.J.; Miles, M. J.; Barham, P. J. Direct Observations of the Growth of Spherulites of Poly(hydroxybutyrateco-valerate) Using Atomic Force Microscopy. Polymer 1998; 39(12), 2437-2446. (3) Ratner, B.; Tsukruk, V. V. Scanning Probe Microscopy in Polymers. ACS Symposium Series; American Chemical Society: Washington, DC, 1998. (4) James, P. J.; McMaster, T. J.; Newton, J. M.; Miles, M. J. In Situ Rehydration of Perfluorosulphonate Ion-exchange Membrane Studied by AFM. Polymer 2000, 41(11), 4223-4231. (5) Leclere, Ph.; Lazzaroni, R.; Bredas, J. L.; Yu, J. M.; Dubois, Ph.; Jerome, R. Microdomain Morphology Analysis of Block Copolymers bt Atomic Force Microscopy with Phase Detection Imaging. Langmuir 1996, 12, 4317-4320. (6) Cleveland, J. P.; Anczykowski, B.; Schmid, A. E.; Elings, V. B. Energy Dissipation in Tapping-Mode Atomic Force Microscopy. Appl. Phys. Lett. 1998, 72(20), 2613-2615. (7) Tamayo, J.; Garcia, R. Relationship between Phase Shift and Energy Dissipation in Tapping-Mode Scanning Force Microscopy. Appl. Phys. Lett. 1998, 73(20), 2926-2928. (8) Betzig, E.; Finn, P. L.; Weiner, J. S. Combined Shear Force and Near-field Scanning Optical Microscopy. Appl. Phys. Lett. 1992, 60(20), 2485-2486.

this particular experimental setup has been applied in the study of different samples in which the term shear force was not appropriate. (It recalls the idea of shear between surfaces that is not true in all cases.) For this reason, a more general description was required. In TDFM, the cantilever is oriented perpendicularly to the sample and oscillates parallel to its surface. The interaction between the tip and the sample can be measured at different separations by observing the change in amplitude and the relative phase of the cantilever oscillation. The shear force is often used in TDFM to obtain topographic images of the surface. The oscillation amplitude of the probe decreases monotonically when approaching the surface; by using the amplitude signal in a feedback loop it is therefore possible to scan the surface at constant height. If the system is monitoring amplitude and phase at the same time, it is also possible to record phase information while keeping the amplitude constant. All the TDFM images in the present work were obtained using this technique. TDFM has also been used for force spectroscopy. In this case the probe is held over one point of the sample surface, and its amplitude and phase are recorded in a series of approach and retract cycles. The experimental quantity that characterizes the cantilever and its interaction with the specimen is the frequency spectrum across the resonance peak. An original technique that records this information at different tip-sample distances (real-time frequency spectra) will be described in this article. To finally evaluate forces from these measurements the dynamics of the vibrating probe has to be modeled and assumptions made on the actual interaction force. In our analysis the force is considered as a combination of
(9) Yang, P. C.; Chen, Y.; Vaeziravani, M. Attractive-mode Atomic Force Microscopy with Optical-Detection in an Orthogonal Cantilever Sample Configuration. J. Appl. Phys. 1992, 71(6), 2499-2502. (10) Davy, S.; Spajer, M.; Courjon, D. Influence of the Water Layer on the Shear Force Damping in Near-field Microscopy. Appl. Phys. Lett. 1998, 73(18), 2594-2596. (11) Brunner, R.; Marti, O.; Hollricher, O. Influence of Environmental Conditions on Shear-Force Distance Control in Near-field Optical Microscopy. J. Appl. Phys. 1999, 86(12), 7100-7106.

10.1021/la000332h CCC: $20.00 2001 American Chemical Society Published on Web 12/21/2000

350

Langmuir, Vol. 17, No. 2, 2001

James et al.

dissipative and elastic restoring components. This assumption is based on the experimental evidence of a viscoelastic interaction between the probe and the specimen in normal humidity conditions.10,11 Nafion is a commercially available perfluorosulfonate cation-exchange membrane (CEM) manufactured by E I du Pont de Nemours & Co. Inc. It is generally used as a perm-selective separator in chlor-alkali electrolyzers12,13 and as the electrolyte in solid polymer fuel cells (SPFC). Perfluorosulfonate cation-exchange membranes are used in these applications because of their high ionic conductivity and their high mechanical, thermal, and chemical stability. The industrial applications of Nafion have prompted considerable research summarized by Eisenberg and Yeager in 198214 and more recently by Tant et al. in 1997.15 Structurally, Nafion consists of a hydrophobic tetrafluoroethylene (TFE) backbone with pendant side chains of perfluorinated vinyl ethers terminated by hydrophilic ion-exchange groups. The difference in probespecimen adhesion between the hydrophobic backbone and hydrophilic side-group regions of the polymer allows the spatial distribution of these two regions to be observed using tapping-mode phase imaging.4 The two complementary scanning probe microscopy (SPM) techniques of AFM and TDFM have been used to investigate the difference in phase contrast exhibited by two Nafion samples differing only in cation form (H+ and Cs+). For each ion form of Nafion, the same probes were used under identical imaging conditions for both AFM and TDFM imaging. Initially standard AFM techniques were applied to Nafion and a test sample before investigating the membrane further with many more novel techniques, such as phase-volume imaging and the collection of a real-time frequency spectrum. Modeling of Probe-Sample Interaction To understand fully the reasoning behind the experiments performed, it is essential that the nature of the probe-sample interaction for these two complementary techniques be understood. Atomic Force Microscopy. In standard tapping-mode scanning force microscopy (TMSFM), the tip intermittently contacts the surface, resulting in a minimization of the destructive lateral forces. This allows the study of soft surfaces and/or weakly adsorbed molecules on a substrate. Alternatively, the measurement of the phase lag of the cantilever oscillation with respect to the excitation force contains information about the interaction between the tip and the sample, allowing compositional contrast on heterogeneous surfaces.16 The origin and nature of the phase contrast has been a subject of debate and discussion for the past few years.17-34 During each oscillation cycle beginning with the tip furthest from the specimen surface, the tip feels a negligible force, then a long-range attractive interaction,
(12) Yeager, H. L.; Steck, A. Cation and Water Diffusion in Nafion Ion Exchange Membranes: Influence of Polymer Structure. J. Electrochem. Soc. 1981, 128, 1880-1884. (13) Yeager, H. L.; ODell, B.; Twardowski, Z. Transport Properties of Nafion Membranes in Concentrated Solution Environments. J. Electrochem. Soc. 1982, 129, 85-89. (14) Eisenberg, A.; Yeager, H. L. Perfluorinated Ionomer Membranes. ACS Books: Washington, DC; 1982. (15) Tant, M. R.; Mauritz, K. A.; Wilkes, G. L. Ionomers: Synthesis, Structure, Properties and Applications. Chapman & Hall: London, 1997. (16) Tamayo, J.; Garcia, R. Deformation, Contact Time and Phase Contrast in Tapping Mode Scanning Force Microscopy. Langmuir 1996, 12, 4430-4435. (17) Chen, G. Y.; Warmack, R. J.; Huang, A.; Thundat, T. Harmonic Response of Near-Contact Scanning Force Microscopy. J. Appl. Phys. 1995, 78(3), 1465-1469.

and finally a repulsive force as it approaches and contacts the sample. Despite the complexity of the interaction and its effect on the cantilever dynamics, theoretical simulations and experiments of the cantilever dynamics in air have shown that phase contrast arises from differences in the energy dissipation between the tip and the sample.7,22 This relationship is due to the surprisingly simple harmonic cantilever response. In fact, calculations and experiments show a sinusoidal movement of the cantilever for the usual cantilever parameters in air, that is, spring constant and quality factor on the order of 10 N/m and 100, respectively. This allows the phase shift to be related analytically to the energy dissipated in the tipsample interaction.6,7,35

sin )

( )

QED A + 0 A0 kAA0

(1)

where is the phase angle, /0 is the working frequency/ resonance frequency, A/A0 is the setpoint amplitude/free amplitude, Q is the quality factor, ED is the energy dissipation, and k is the cantilever spring constant.
(18) Anczykowski, B.; Kruger, D.; Babcock, K. L.; Fuchs, H. Basic Properties of Dynamic Force Spectroscopy with the Scanning Force Microscope in Experiment and Simulation. Ultramicroscopy 1996, 66(34), 251-259. (19) Anczykowski, B.; Kruger, D.; Fuchs, H. Cantilever Dynamics in Quasinoncontact Force Microscopy: Spectroscopic Aspects. Phys. Rev. B Condens. Matter 1996, 53(23), 15485-15488. (20) Burnham, N. A.; Behrend, O. P.; Oulevey, F.; Gremaud, G.; Gallo, P. J.; Gourdon, D.; Dupas, E.; Kulik, A. J.; Pollock, H. M.; Briggs, G. A. D. How Does a Tip Tap? Nanotechnology 1997, 8(2), 67-75. (21) Kuhle, A.; Sorensen, A. H.; Bohr, J. Role of Attractive Forces in Tapping Tip Force Microscopy. J. Appl. Phys. 1997, 81(10), 6562-6569. (22) Tamayo, J.; Garcia, R. Effects of Elastic and Inelastic Interactions on Phase Contrast Images in Tapping-Mode Scanning Force Microscopy. Appl. Phys. Lett. 1997, 71(16), 2394-2396. (23) Whangbo, M. H.; Magonov, S. N.; Bengel, H. Tip-Sample Force Interactions and Surface Stiffness in Scanning Probe Microscopy. Probe Microscopy 1997, 1(1), 23-42. (24) Bar, G.; Brandsch, R.; Whangbo, M. H. Description of the Frequency Dependence of the Amplitude and Phase Angle of a Silicon Cantilever Tapping on a Silicon Substrate by the Harmonic Approximation. Surf. Sci. 1998, 411(1-2), L802-L809. (25) Behrend, O. P.; Oulevey, F.; Gourdon, D.; Dupas, E.; Kulik, A. J.; Gremaud, G.; Burnham, N. A. Intermittent Contact: Tapping or Hammering? Appl. Phys. A Materials Sci. Proc. 1998, 66(Pt1SS), S219S221. (26) Garcia, R.; Tamayo, J.; Calleja, M.; Garcia, F. Phase Contrast in Tapping-Mode Scanning Force Microscopy. Appl. Phys. A Solids Surf. 1998, 66(Pt1SS), S309-S312. (27) Hunt, J. P.; Sarid, D. Kinetics of Lossy Grazing Impact Oscillators. Appl. Phys. Lett. 1998, 72(23), 2969-2971. (28) Whangbo, M. H.; Bar, G.; Brandsch, R. Description of Phase Imaging in Tapping Mode Atomic Force Microscopy by Harmonic Approximation. Surf. Sci. 1998, 411(1-2), L794-L801. (29) Bar, G.; Brandsch, R.; Whangbo, M. H. Correlation between Frequency-Sweep Hysteresis and Phase Imaging Instability in Tapping Mode Atomic Force Microscopy. Surf. Sci. 1999, 436(1-3), L715-L723. (30) Bar, G.; Brandsch, R.; Whangbo, M. H. Effect of Tip Sharpness on the Relative Contributions of Attractive and Repulsive Forces in the Phase Imaging of Tapping Mode Atomic Force Microscopy. Surf. Sci. 1999, 422(1-3), L192-L199. (31) Haugstad, G.; Jones, R. R. Mechanisms of Dynamic Force Microscopy on Poly(vinyl alcohol): Region-specific Noncontact and Intermittent Contact Regimes. Ultramicroscopy 1999, 76(1-2), 7786. (32) Nony, L.; Boisgard, R.; Aime, J. P. Nonlinear Dynamical Properties of an Oscillating Tip-Cantilever System in the Tapping Mode. J. Chem. Phys. 1999, 111(4), 1615-1627. (33) Bar, G.; Brandsch, R.; Bruch, M.; Delineau, L.; Whangbo, M. H. Examination of the Relationship between Phase Shift and Energy Dissipation in Tapping Mode Atomic Force Microscopy by Frequency Sweep and Force-Probe Measurements. Surf. Sci. 2000, 444(13), L11 L16. (34) Delineau, L.; Brandsch, R.; Bar, G.; Whangbo, M. H. Harmonic Responses of a Cantilever Interacting with Elastomers in Tapping Mode Atomic Force Microscopy. Surf. Sci. 2000, 448(1), L179-L187. (35) Tamayo, J. Energy Dissipation in Tapping-Mode Scanning Force Microscopy with Low Quality Factors. Appl. Phys. Lett. 1999, 75(22), 3569-3571.

Phase Contrast

Langmuir, Vol. 17, No. 2, 2001 351

Figure 1. A graphical representation of the two solutions to eq 1, demonstrating the two possible imaging regimes and the phase lags associated with them. A more energy-dissipative feature appears light in the noncontact regime and dark in the intermittent contact regime. (DI convention has been used.)

This expression allows experimental phase curves to be interpreted. When the cantilever is far enough from the sample, the tip oscillates freely (A ) A0, ED ) 0) and the phase shift is 90 (0 in the Digital Instruments (DI) software). As the cantilever oscillates in the proximity of the sample, the oscillation is damped (A < A0) as a consequence of the interaction between the tip and the sample, and a linear decrease of the damped amplitude is produced as the probe approaches the specimen. If it is assumed that this interaction is conservative and no energy is dissipated, eq 1 has two solutions, producing the two branches shown in Figure 1. As the cantilever approaches the sample, one branch increases until the phase shift is 180 (-90 in the DI software), whereas the other branch goes toward 0 (90 in the DI software). The first solution corresponds to the noncontact regime in which an attractive interaction is responsible for the damping of the cantilever oscillation. This interaction shifts the cantilever resonance to lower frequencies producing a phase shift lower than 90. The second solution is associated with the intermittent-contact regime, in which the repulsive force produced during the tip-sample contact displaces the resonance to higher frequencies. The DI phase convention will be adopted for the remainder of the article. The effect of a tip-sample interaction, which involves energy dissipation, is the displacement of the noncontact solution to higher phase shifts and the intermittentcontact solution to lower phase shift values. The more dissipative features will appear lighter in the noncontact regime, whereas they will appear darker in the intermittent-contact regime. An experimental curve is a combination of both solutions. As the cantilever approaches, the attractive force is responsible for the damping of the oscillation, and the cantilever oscillates in the noncontact regime. As the cantilever approaches further, the tip strikes the surface intermittently and a sudden change in the phase shift is observed as a consequence of the transition of the cantilever oscillation from noncontact to the intermittent-contact regime. Transverse Dynamic Force Microscopy. The dynamics of an oscillating cylindrical probe can be modeled using the simple harmonic oscillation theory36 but, in this case, the continuum mechanics model for a cylindrical bar has been used.37 The equation describing the trans(36) Sarid, D. Scanning Force Microscopy; Oxford University Press; Oxford, 1991. (37) Drummond-Roby, M. A.; Wetsel, G. C. Measurement of Elastic Force on a Scanned Probe Near a Solid Surface. Appl. Phys. Lett. 1996, 69(24), 3689-3691.

Figure 2. (a) The two possible mechanisms for a decrease in the oscillation amplitude of a probe as it is brought into contact with the sample; a damping or a change in resonance frequency. These two mechanisms are indistinguishable using conventional feedback methods. (b) The experimental setup required to perform a real-time frequency spectrum, enabling the two components of the probe-sample interaction to be separated.

verse oscillation of a cylindrical bar is:

2 u 2u 2 EI 2 u + + A 2 ) 0 t z2 z t

(2)

where u(z,t) is the displacement of the bar, E is the Youngs modulus, I is the second moment of inertia, is the internal friction coefficient, and is the density of the bar. The boundary conditions for the clamped end (z ) 0) are: u(0) ) d0; u/z ) 0 (d0 is the applied vibration to the probe), and for the free end (z ) L) EI2u/z2 ) 0; EI2u/z3 ) 0. The second and third derivatives are proportional to the external torque and the external force, respectively. The oscillation amplitude and resonance frequency of the probe change when it is interacting with the surface. To model the experimental data, the interacting force is described by an elastic and a dissipative component. The first term is responsible for the decrease in the oscillation amplitude, whereas the second term takes into account the shift in resonance frequency. The boundary conditions can be therefore rewritten as:

EI(3u/z3)z)L ) (u/t)z)L + ku(L)

(3)

where is a dissipative coefficient and k is an elastic constant. The importance of a liquid film between the sample and the probe for the TDFM contrast mechanism has already been reported.10,11 When the probe is close to the surface (10 nm) the liquid film becomes confined and displays solidlike behavior.38,39 This gives rise to a viscoelastic shear force,40 which is dependent on the sample
(38) Israelachvili, J. Intermolecular and Surface Forces; Academic Press: London, 1985. (39) Granick, S. Motions and Relaxations of Confined Liquids. Science 1991, 253, 1374-1379.

352

Langmuir, Vol. 17, No. 2, 2001

James et al.

Figure 3. Images of different cation forms of Nafion obtained using the same cantilever under identical imaging conditions. (a) A 1-m tapping-mode AFM topography image of Nafion 115 H+ imaged under ambient conditions. (b) A phase image corresponding to part a (Z-scale, 25 nm and 10, respectively). (c) A 1-m tapping-mode AFM topography image of Nafion 115 Cs+ imaged under ambient conditions. (d) A phase image corresponding to part c (Z-scale, 25 nm and 60, respectively).

surface via the chemical bonding between the first liquid layer and the surface, and on the amount of surface water, which is humidity-dependent. A force with an elastic and a dissipative component can simulate a viscoelastic interaction and for this reason it has been implemented in the described model. Experimental Method
Sample Preparation. The membrane will rehydrate readily if it is exposed to an environment of high relative humidity (RH).41 Careless handling can result in the membrane ion-exchanging from the acid (H+) to a salt form (e.g., Na+ or K+). All Nafion samples therefore were routinely prepared by refluxing with concentrated nitric acid and deionized water (50/50 v:v), then deionized water alone, to ensure that the membrane was in the H+ form and free from any chemical impurities. Strips of Nafion H+ were then converted to the Cs+ form by immersion in a 0.1 M CsNO3 solution for a week. Atomic Force Microscopy. H+ and Cs+ samples were mounted on magnetic stainless steel sample stubs and placed inside a Digital Instruments Extended Multi Mode AFM using version 4.22 of the Nanoscope software (DI-Veeco, S. Barbara, CA). The samples were imaged using tapping-mode phase imaging and a standard silicon cantilever (40 N/m) to provide topographic and corresponding phase images. The samples were imaged using the same cantilever under identical imaging conditions: relative humidity, free amplitude of oscillation, and ratio of setpoint to free amplitude. The relative humidity was controlled by placing the AFM inside a purpose-built environmental chamber allowing the humidity to be kept constant by
(40) Hu, H.-W.; Carson, G. A.; Granick, S. Relaxation Time of Confined Liquids under Shear. Phys. Rev. Lett. 1991, 66(21), 2758-2761. (41) Dreyfus, B.; Gebel, G.; Aldebert, P.; Pineri, M.; Escoubes, M.; Thomas, M. Distribution of the Micelles in Hydrated Perfluorinated Ionomer Membranes from Sans Experiments. J. Phys. 1990, 51(12), 1341-1354.

Figure 4. Variation of IR absorption with cation form. ATR data obtained for the two ion forms of Nafion. The spectra were labeled with aid of reference spectra for Nafion and Teflon50 and normalized to the CF2 peaks. Comparison of the two curves clearly illustrates the greater water content of the H+ form.
passing nitrogen gas through molecular sieve material.4 The properties of the two ion forms of Nafion were investigated further by obtaining attenuated total reflection Fourier transform infrared spectra using a Nicolet 510P spectrometer. The effect of surface water on tip-sample adhesion was investigated by using a specimen surface prepared to have hydrophobic and hydrophilic domains. The test sample was prepared by cleaning glass with detergent, rinsing with water, dipping in a mixture of chromic and sulfuric acid, rinsing with water, dipping in 5 M NaOH, rinsing with water, and then drying vertically in an oven. When dry, the slide was placed onto a

Phase Contrast

Langmuir, Vol. 17, No. 2, 2001 353

Figure 5. A mixed hydrophilic/hydrophobic test sample consisting of silane evaporated onto cleaned glass. (a) An optical image of the test sample. Water droplets are clearly visible on the right-hand side of the interface running from middle top to bottom left. (b) A force distance curve obtained to the left of the interface in part a, the energy required to break free 1.2f J. (c) A force distance curve obtained to the left of the interface in part a, the energy required to break free was 7.6f J.
small jar of aminopropyltriethoxysilane, which was allowed to evaporate onto the slide for 5 min. The adhesive properties of the test sample were then studied by obtaining multiple-force distance curves over the two regions. Force volume imaging allows an image to be built up of the tip-sample interaction at each pixel from an array of force curves.42 It has been applied primarily to the study of elastic and adhesive properties of nonhomogeneous substrates.43-46 Although the importance of phase-distance curves has been recognized,47 the next logical progression, phase-volume imaging, had yet to be taken. At each pixel in the image, the phase contrast can be obtained for any tip-sample separation, hence the word volume. This is a novel AFM technique on which no reports have been published to date. In practice, it works in exactly the same way as the force-volume imaging except that amplitude is used for the feedback loop and it is the phase signal rather than the deflection signal that is monitored. Phase-volume (PV) images of the different cation forms of Nafion, consisting of 64 64 pixels with 64 data points per approach and retract curve, were obtained. At each pixel the cantilever approached the surface until such time as the oscillation amplitude decreased to the preset trigger before
(42) Hoh, J. H.; Heinz, W. F.; Hassan, E. A. Force Volume Support Note No. 240. Digital Instruments: 1997. (43) Cappella, B.; Dietler, G. Force-distance Curves by Atomic Force Microscopy. Surf. Sci. Rep. 1999, 34, 1-104. (44) Reynaud, C.; Sommer, F.; Quet, C.; El Bounia, N.; Duc, T. M. Quantitative Determination of Youngs Modulus on a Biphase Polymer System Using Atomic Force Microscopy. Surf. Interface Anal. 2000, 30(1), 185-189. (45) Raiteri, R.; Butt, H. J.; Beyer, D.; Jonas, S. Heterogeneous Polymer-Containing Films: A Comparison of Macroscopic Properties with Microscopic Properties Determined by Atomic Force Microscopy. Phys. Chem. Chem. Phys. 1999, 1(20), 4881-4887. (46) Beake, B. D.; Leggett, G. J.; Shipway, P. H. Frictional, Adhesive and Mechanical Properties of Polyester Films Probed by Scanning Force Microscopy. Surf. Interface Anal. 1999, 27(12), 1084-1091. (47) Chen, X.; Davies, M. C.; Roberts, C. J.; Tendler, S. J. B.; Williams, P. M.; Davies, J.; Dawkes, A. C.; Edwards, J. C. Interpretation of Tapping Mode Atomic Force Microscopy Data Using Amplitude-Phase-Distance Measurements. Ultramicroscopy 1998, 75(3), 171-181. (48) Antognozzi, M.; Haschke, H.; Miles, M. J. A New Method to Measure the Oscillation of a Cylindrical Cantilever: The Laser Reflection Detection System. Rev. Sci. Instrum. 2000; in press.

withdrawing and moving onto the next pixel. All the images were obtained with the cantilever initially at resonance to simplify interpretation. A Nafion 115 H+ sample was imaged under ambient conditions using phase-volume imaging at several different free amplitudes and ratios of set point to free amplitude. A scan size of 500 500 nm2 coupled with an array of 64 64 pixels was used to ensure that features comparable with the cluster size could be detected. The images could then be analyzed by taking slices through the images at specific tip-sample distances. Phase images of Nafion 115 Cs+ were then obtained using the same cantilever under the same conditions with a variety of different free amplitudes and ratios of set points to free amplitude to determine the optimum conditions. These were then used to obtain the PV image. Transverse Dynamic Force Microscopy. All the experiments detailed below were performed using an in-house built transverse dynamic force microscope. The probe was mounted on a piezoelectric actuator which drove it at one of its resonant modes. The oscillation amplitude was detected using the laser reflection detection system (LRDS)48 that provides the true measurement of the vibration amplitude necessary to quantify the tip-sample interaction. Typical values for the oscillation amplitude are about 10 nm. An uncoated optical fiber probe was used rather than a metallic probe49 owing to the similarity of its surface chemistry to that of a silicon tapping-mode cantilever. The probe was prepared using the same method as that for SNOM probes.50 To test the importance of the relative humidity in the TDFM contrast mechanism the microscope was placed inside an inhouse built environmental chamber. A Nafion H+ sample was mounted onto a 1.5-cm-diameter magnetic stainless steel sample stub and placed in the transverse dynamic force microscopy. The relative humidity could be reduced using nitrogen gas passed through molecular sieve material or increased by bubbling nitrogen gas through water and into the chamber. Once the desired humidity had been reached, the sample was allowed to
(49) Nam, A. J.; Teren, A.; Lusby, T. A.; Melmed, A. J. Benign Making of Sharp Tips for STM and FIM: Pt, Ir, Au, Pd, and Rh. J. Vac. Sci. Technol. 1995, B13(4), 1556-1559. (50) Williamson, R. L.; Miles, M. J. Melt-drawn Scanning Near-field Optical Probe Profiles. J. Appl. Phys. 1996, 80(9), 4804-4812.

354

Langmuir, Vol. 17, No. 2, 2001

James et al.

Figure 6. (a) A 500-nm tapping mode AFM topography image of Nafion H+ imaged under ambient conditions obtained in bistable regime. Depressions can be seen in the topographic image; these artifacts correspond to the transitions from the noncontact to the intermittent-contact regime. (b) A phase image corresponding to part a. A much larger phase range can be observed in a bistable regime than can be obtained working in any one regime. The Z-scales are 20 nm and 180. (c) A line profile of the topographic image (a) over four transitions between regimes, at each of which a depression of up to 5 nm can be observed. (d) A line profile of the phase image (b) over four transitions between regimes, at each of which phase shifts in excess of 90 can be observed.
equilibrate. Topographic and corresponding phase images were obtained at high (50%) and low (10%) humidities. In addition to the images, amplitude-distance curves were obtained at each of the humidities. This process was then repeated for the Cs+ ion form of the membrane. In tapping-mode AFM and TDFM, the feedback mechanism uses the probe oscillation amplitude signal to control the probesample separation. Unfortunately, using conventional feedback methods, it is impossible to differentiate between a damping of the amplitude and a shift in resonance frequency (Figure 2a). As described previously the liquid confined between the probe and the surface may be responsible for a decrease in the oscillation amplitude and a resonance frequency shift. These effects can be detected by measuring the frequency spectrum of the probe while it is approaching the surface. In this way the two components of the interaction are separated and measured as a function of tip-sample distance. Using the transverse dynamic force microscopy it is possible to perform a real-time frequency spectrum by simultaneously exciting two modes of the probe. The first frequency is kept constant at the first (or second) resonant peak and is used to monitor the oscillation amplitude during the approach and retract cycle. The second frequency is swept from just below to just above the second (or first) resonance frequency of the free probe, by using the sweep mode of the Philips 5192 signal generator. This mode of operation will sweep the driving frequency up and down continuously, and the sweep time can be as low as 50 ms. The amplitude and the phase spectra are monitored continuously as the probe approaches the sample surface (Figure 2b) using a Labview recording system.

Results and Discussion Atomic Force Microscopy. Tapping-mode topography and corresponding phase images of Nafion H+ obtained under ambient conditions are shown in Figure 3 a and b, respectively. The images of Nafion Cs+ obtained using the same cantilever under identical imaging conditions

are shown in Figures 3c and 3d, respectively. There is a marked difference in the phase contrast between the two ion forms. The phase range is significantly larger in the Cs+ ion, 60 as opposed to 10 for the H+ form. There is no significant difference in the topography images, therefore topographic coupling is not responsible for the change in phase contrast. Attenuated total reflection (ATR) data obtained for the two ion forms of Nafion are shown in Figure 4. The spectra were labeled with the aid of reference spectra for Nafion and Teflon51 and normalized to the CF2 peaks. Comparison of the two curves clearly illustrates the greater water content of the H+ form. This combined with the greater charge density associated with the Cs+ ion, which would be screened less effectively, points toward an electrostatic force being responsible for the difference in phase contrast. An optical image of the mixed hydrophobic/hydrophilic test sample is shown in Figure 5a. Water droplets are clearly visible on the right-hand side of the interface between the silanized glass and cleaned glass, running diagonally from top right to bottom left. The interface is not particularly sharp because of the way in which the silane was evaporated onto the glass. Force curves obtained to the left and right of the interface are shown in Figures 5b and 5c, respectively. The energy required for the cantilever to break free from the surface was considerably higher to the right of the interface where the water droplets are clearly visible. The energy can be calculated from the area under the curve, if it is assumed
(51) Kuptsov, A. H.; Zhizhin, G. N. Handbook of Fourier Transform Raman and Infrared Spectra of Polymers; Elsvier Science: Amsterdam, 1998.

Phase Contrast

Langmuir, Vol. 17, No. 2, 2001 355

Figure 7. Phase images corresponding to slices through the phase-volume image of Nafion Cs+ for cantilever positions from 0 to 48 nm at intervals of 3.7 nm showing the wide range of possible contrasts (Z-scale, 60). Note that the cantilever position is arbitrary and it does not indicate the cantilever-sample distance. The zero position is determined as the cantilever position in which the amplitude is damped by 90%.

that Hookes law is obeyed and that the cantilever spring constant is 0.12 N/m. The pull-off energy was 1.2f J and 7.6f J for areas left and right of the interface, respectively. It is clear from these results that any surface water dramatically increases the adhesive force between the tip and sample. Although these features are over an order of magnitude larger than the proposed cluster size in Nafion, this contrast mechanism would still apply when imaging at higher resolutions. The elastic response of Nafion increases with frequency,52 at the frequencies at which the cantilever is being driven 250 kHz, and so the viscoelastic energy loss would be negligible. Therefore any energy dissipation

in the tapping interaction is primarily the result of tipsample adhesion rather than a viscoelastic energy loss. Differences in surface adhesion over a sample caused by an inhomogeneous distribution of surface water are probably a very important phase-contrast mechanism. Normal tapping-mode topography and phase images of Nafion, obtained in a bistable regime are shown in Figures 6a and 6b, respectively. In a bistable regime, the cantilever
(52) Eisenberg, A.; King, M. Ion-Containing Polymers; Physical Properties and Structure, Vol. 2; Academic Press: London, 1977; p 164. (53) Kuhle, A.; Sorensen, A. H.; Zandbergen, J. B.; Bohr, J. Contrast Artifacts in Tapping Tip Atomic Force Microscopy. Appl. Phys. A Solids Surf. 1998, 66(Pt1 SS), S329-S332.

356

Langmuir, Vol. 17, No. 2, 2001

James et al.

can oscillate in both noncontact and intermittent-contact regimes.53-56 From the mathematical point of view, the state of the oscillation depends on the initial conditions. In the practice, the cantilever oscillates randomly in noncontact and intermittent contact during the scanning, producing a flipping of the phase shift between negative and positive values (DI software). The sudden transitions from noncontact to intermittent-contact regime are accompanied by artifacts in the form of depressions in the topography image. Line profiles across both of the images are shown in Figures 6c and 6d. The line profiles cover four transitions between the noncontact and intermittent-contact regimes. At each of these transitions, corresponding to the intermittent-contact regime, a depression of up to 5 nm (25% of the Z-scale) and a phase shift of move than 90 is observed in the topographic and phase image, respectively. This result illustrates the importance of avoiding a bistable regime in phase imaging of a surface. Fourteen slices have been taken through a phase-volume image of Nafion Cs+ at intervals of 3.7 nm. The oscillation amplitude of the cantilever when free was in excess of 50 nm. Each of the slices shows the phase contrast that would have been obtained in traditional tapping mode phase imaging at different set points (Figure 7). At cantileversample distances slightly larger than the free amplitude (cantilever position 37 nm), the interaction is too weak to produce any meaningful phase contrast. A significant higher phase shift ( > 60) is observed in certain regions for a cantilever position from 22 to 33 nm. In these regions, the cantilever oscillates in the intermittent-contact regime. Once below a tip-sample separation of about 15 nm, the images are predominantly in the intermittent-contact regime and the phase contrast is reduced to about 20. However, a few points, which show up as white squares in the dark regions, are still in the noncontact regime, even for a damped amplitude lower than 10%. If slices taken in the noncontact (cantilever position, 44.3 nm) and intermittent-contact regimes are compared, the same features are clearly visible in both images, there has been a contrast reversal, however. This is consistent with eq 1 and implies energy dissipation in the noncontact regime. To interpret the evolution of the phase contrast with the damped amplitude, a topography and corresponding phase image and the three types of amplitude-distance curves, obtained during phase volume imaging of a Nafion Cs+ sample using a free amplitude in excess of 5V and a set point of 2.5V, are shown in Figure 8. The location from which the phase-distance curves were taken has been clearly marked in the phase image (Figure 8b). In one case (Figure 8c), the phase distance curve starts off in the noncontact regime before swiftly moving into the intermittent contact regime where the phase angle tends to 90, indicating that there is little energy dissipation. The second type of curve (Figure 8d) takes noticeably more force to move from the noncontact to intermittent-contact regimes because of damping caused by an attractive force, which is probably electrostatic. Once in the intermittent contact regime the phase angle is considerably lower, 60, indicating that more energy is being dissipated. Inside regions made up of the second type of curve; a third type
(54) Behrend, O. P.; Odoni, L.; Loubet, J. L.; Burnham, N. A. Phase Imaging: Deep or Superficial? Appl. Phys. Lett. 1999, 75(17), 25512553. (55) SanPaulo, A.; Garcia, R. High-Resolution Imaging of Antibodies by Tapping-Mode Atomic Force Microscopy: Attractive and Repulsive Tip-Sample Interaction Regimes. Biophys. J. 2000, 78(3), 1599-1605. (56) Garcia, R.; SanPaulo, A. Amplitude Curves and Operating Regimes in Dynamic Atomic Force Microscopy. Ultramicroscopy 2000, 82(1-4), 79-83.

Figure 8. Phase-volume images (500 nm) of Nafion Cs+ imaged under ambient conditions. (a) A topography image (Z-scale, 20 nm); (b) a phase image corresponding to part a (Z-scale, 60). The phase-distance curves can be divided into three types. (c) A curve that moves quickly from the noncontact to intermittentcontact regime. (d) A curve that requires more force to move into the intermittent-contact regime. (e) A third type that remains in the noncontact regime can be found within regions made up of the second type of curve. Note that the data have been plotted with an offset x-axis to not obscure the data.

of curve (Figure 8e) can be found where, despite the very high free amplitude and low set point, the cantilever never actually made contact with the surface because of particularly strong damping. The corresponding points in the topography image (Figure 8a) appear high, because these points are in the noncontact regime, whereas the remainder of the image is in the intermittent-contact regime. The behavior of the first type of curve (Figure 8c) can be attributed to the hydrophobic backbone of the membrane, whereas that of the second (Figure 8d) and third (Figure 8e) type of curve can be attributed to the ion-rich regions of the membrane. The inability to image the Cs+ ion form completely in the intermittent-contact regime, using the same cantilever as that used for the H+ form, despite using double the free amplitude, can again be explained by the greater charge density associated with the Cs+ ion, which is screened less effectively because of the lower water content. This produces a strong longrange attractive interaction that shifts the resonance frequency of the cantilever, therefore reducing its oscillation amplitude until the set point is reached. The second type of curve (Figure 8d) corresponds to the hydrophilic region that surrounds the Cs+ ions. These regions show a lower phase shift with respect to the hydrophobic regions in the intermittent-contact regime, associated to a higher energy dissipated between the tip and the sample. The energy dissipation would be due to preferential water adsorption to the hydrophilic regions. The unbalance between the lower adhesion when the water neck forms and the needed force to break the meniscus should be the responsible factor of energy dissipation. The contrast reversal observed in the phase volume in the noncontact regime supports this energy dissipation model. Only the

Phase Contrast

Langmuir, Vol. 17, No. 2, 2001 357

Figure 9. A comparison of AFM and TDFM images. (a) A 1-m tapping-mode AFM topography image of a Nafion H+ sample imaged under ambient conditions (Z-scale, 15 nm). (b) A phase image corresponding to part a (Z-scale, 30). (c) A 1-m TDFM topography image of Nafion H+ imaged under ambient conditions (Z-scale, 400 nm). The resolution of the TDFM is lower than that of the AFM owing to the size of the probe and thermal drift, which is more of a problem for the TDFM because of the longer scan times, 30 min compared with 4 min for the AFM. (d) Corresponding TDFM phase image (Z-scale, 30).

rupture of a water meniscus between the tip and the sample could produce energy dissipation in the noncontact regime. Although terminology such as soft and hard tapping23 is useful, what is really important is the tapping regime in which the experiment is performed, that is whether it is noncontact or intermittent contact. It is particularly important not to work in a bistable regime because of the height artifacts that can be produced. If the phase contrast observed is greater than 90, it is certain that this is the case. Several articles cite phase ranges of 90 or higher23; it is likely that the images were obtained in a bistable regime. Consequently, any topography data across the bistable regime is at least partly artifactual. Many of the effects observed in other articles57-59 including contrast reversal in the topography and phase images may be attributed to working in a bistable regime or moving from one regime to another. To use phase imaging successfully, it is essential to establish the tip-sample interaction regime by obtaining a phase-distance curve and adjusting the free amplitude
(57) McLean, S. R.; Sauer, B. B. Nano-deformation of Crystalline Domains during Tensile Stretching Studied by Atomic Force Microscopy. J. Polym. Sci. Part B Polym. Phys. 1999, 37(8), 859-866. (58) McLean, S. R.; Sauer, B. B. Tapping-Mode AFM Studies Using Phase Detection for the Resolution of the Nanophases in Segmented Polyurethanes and Other Block Copolymers. Macromolecules 1997, 30(26), 8314-8317. (59) Sauer, B. B.; McLean, R. S.; Thomas, R. R. Tapping Mode AFM Studies of Nano-phases on Fluorine-containing Polyester Coatings and Octadecyltrichlorosilane Monolayers. Langmuir 1998, 14(11), 30453051. (60) Hsu, W. Y.; Gierke, T. D. Ion-Transport and Clustering in Nafion Perfluorinated Membranes. J. Membr. Sci. 1983, 13(3), 307-326.

and set point accordingly. During the study it was observed that different cantilevers required different amounts of force to move from the noncontact to intermittent-contact regimes. This is consistent with a recent study into the effects of tip sharpness on the contrast in phase imaging.30 The transition occurred more easily with a sharper tip. A higher attractive force appears with blunter tips, as a consequence of the larger effective contact area for interaction. Transverse Dynamic Force Microscopy. A 1-m tapping-mode AFM topography image, its corresponding phase image and TDFM topography, and phase images of Nafion are shown in Figure 9. It is apparent from the topography images that the resolution of the TDFM on this sample is somewhat lower than that of the AFM. This is probably the result of two main effects: first, the size of the end of the particular TDFM probe, and second, thermal drift, because scan times are about 30 min, rather than the 4 min for the AFM. The nature of the sample may also have a bearing on the ultimate resolution, because the cluster-network model of Nafion60 postulates a large-scale organization of clusters with transient connective tubes which are in constant flux. Four amplitude-distance curves obtained using Nafion H+ and Cs+ samples across a range of humidities from 10 to 52 are shown in Figure 10. The effect of humidity is evident for both the samples, but it is easier to explain in Nafion H+ probably because of its higher hydrophilicity. The difference between the points at which the amplitude starts to be damped and drops to zero indicates the thickness of the water layer over the surface (6 nm for the H+ from at 46%RH). The hysteresis in the retract curve is probably caused by the presence of a capillary

358

Langmuir, Vol. 17, No. 2, 2001

James et al.

Figure 10. The effect of relative humidity on surface water thickness. Four TDFM amplitude-distance curves obtained using Nafion H+ and Cs+ samples at range of humidities. (a) Nafion H+ at 46%RH, (b) Nafion H+ at 10%RH, (c) Nafion Cs+ at 52%RH, (d) Nafion Cs+ at 10%RH. As the humidity decreases, the slope of the approach curves becomes steeper, the evidence of a liquid-neck formation is no longer detectable and, in the Nafion H+, the hysteresis between the approach and retract curves decreases. The depth of the surface water layer was lower for the Cs+ ion form than for the H+ form.

neck. At lower humidities the approach curves become steeper, the neck formation is not clearly detectable, and the hysteresis in the curves decreases, which indicates a decrease in the thickness of the water layer. Comparison of the approach and retract curves for the Cs+ and H+ samples at 10% humidity shows a steeper approach curve, which indicates that the surface is drier. This would be consistent with the infrared data in Figure 4, which clearly demonstrated the differences in water content of the ion forms of the membrane. In Nafion Cs+, the noncontinuous nature of the surface liquid film causes differences in the force curves. Topography and corresponding phase images of Nafion H+ and Cs+ were obtained during the investigation. The phase range obtained using TDFM was significantly lower than that observed with tapping mode AFM because of the different contrast mechanism involved. In tappingmode AFM, the phase contrast was caused by the tip going in and out of the water layer, whereas the TDFM probe is always inside the water layer moving from side to side. Differences in the hydrophilicity of the sample, resulting in preferential water adsorption to some areas, would still be detectable because of the increased drag on the probe in these areas. A change in phase contrast with humidity was observed; the phase contrast is lower at the lower humidity, which is consistent with a recent AFM study.4 The phase contrast for the Cs+ form was lower than that of the H+ again indicating that there is less surface water for this ion form. A real-time frequency spectrum surface is shown in Figure 11. The surface is obtained from the frequency spectra taken at different tip-sample distances. The vertical gray plane represents the reference frequency

used to measure the approach amplitude curve (white line). It is clear that, when the probe approaches the surface, the resonance frequency (dotted line) does not change in a monotonic way on a nanometric scale, as usually expected.37 Looking at the approach curve it is possible to distinguish four regions, which are separated by black lines. Fitting this surface with the model described previously (eqs 2 and 3), the elastic and dissipative parts of the interaction can be calculated. The results of this analysis are shown in Figure 12 in which the four regions marked in Figure 11 have been emphasized using black vertical lines. In region A, the probe is free, until it reaches a distance of 16 nm from the surface when the oscillation amplitude of the probe drops as a consequence of the formation of a capillary condensation neck. In region B, the dissipative force remains almost constant while the elastic component increases. This is responsible for the decrease in the approach-curve amplitude. (A small shift of the frequency peak produces a significant change in the amplitude at resonance.) In region C, the dissipative component dominates the elastic component and is responsible for the large decrease in the oscillation amplitude. In region D, the slope of the elastic force is steeper than the dissipative force, and the elastic component will eventually become predominant for z smaller than 4 nm. These results clearly show that the information provided by the approach curve alone is insufficient to determine the kind of probesample interaction. The real-time frequency spectrum is therefore a unique tool to decouple and quantify the two components giving a better insight into the nature of the interaction.

Phase Contrast

Langmuir, Vol. 17, No. 2, 2001 359

Figure 11. A real-time frequency spectrum obtained using the TDFM. The black curves on the surface are frequency spectra taken at different tip-sample distances. The vertical gray plane represents the reference frequency used to measure the approach amplitude curve (white line), and a dotted line represents the resonant frequency. Using the mathematical model highlighted in the text, it is possible to evaluate the elastic and dissipative components of the tip probe from these data.

Figure 12. Analysis of real-time frequency spectrum obtained using the TDFM. The approach curve (thick line), the dissipative component (dotted line), and the elastic component (continuous line) of the probe-sample interaction. The slope of the force curves in the different regions (a, b, c, d) is added to emphasize the predominant component.

Conclusions The origin of phase contrast in Nafion has been investigated using the two complementary SPM techniques of AFM and TDFM. A variety of standard and new techniques, namely phase-volume imaging and a realtime frequency spectrum were used. Force curves obtained over a mixed hydrophobic/ hydrophilic test sample showed a much larger adhesive force over the water-rich regions. An increase in relative humidity resulted in an increase in the thickness of the

surface water layer and the phase contrast observed with both SPM techniques. Therefore differences in probesample adhesion, caused by an inhomogeneous distribution of surface water, are an important phase-contrast mechanism. Phase-volume imaging has been a useful tool in the interpretation of phase contrast. It has clearly demonstrated the wide range of phase contrasts that can be observed on the same sample. Moving from the noncontact to intermittent-contact regime resulted in a contrast

360

Langmuir, Vol. 17, No. 2, 2001

James et al.

inversion. The most dissipative features were light in the noncontact and dark in intermittent-contact regimes. When working in a bistable regime, height artifacts were produced in the topographic images at the points where the transition occurred. The anomalous results for the different cation forms of Nafion and those in many published studies can be attributed to working in a bistable regime or moving from one regime to another. The importance of phase-distance curves has been highlighted as a prerequisite to imaging to ensure that imaging takes place in any one regime, rather than relying on a standard set of operating conditions. The sharpness of the tip influenced the phase contrast observed, altering the force required to move from one regime to another. It is therefore necessary to obtain a phase-distance curve if a cantilever is damaged or changed to ensure that imaging continues in the same regime. The phase-volume images of Nafion consisted of two main types of phase-distance curves. The first curve moved quickly from the noncontact to the intermittent-contact regime, and once there the phase angle of 90 indicated little energy dissipation. These regions were attributed to the hydrophobic backbone. The second type of curve required considerably more force to enter the intermittentcontact regime, and once there the phase angle of 60 indicated more energy dissipation. These regions were attributed to the ion-rich regions that would damp the cantilever oscillation with an attractive electrostatic force at longer distances, then, once in contact, dissipate more energy owing to their greater affinity for water. A greater force was required to image the Cs+ ion form in the intermittent-contact regime compared with the H+ form because of the lower water content and therefore reduced screening of the Cs+ ions charge. When compared with AFM (dynamic mode), it is clear that TDFM differs in two main aspects: the shape and the orientation of the probe with respect to the specimen

surface. In TDFM the cantilevers have a cylindrical tapered shape and are mounted perpendicular to the specimen surface, which allows accurate control of the tip-sample distance, ecause the probe is not extensible in the vertical direction. This characteristic makes the TDFM a more suitable tool for force spectroscopy: the probe does not jump to contact during the approach and a constant load rate can be applied by just keeping the approach or retract speed constant. (Both these aspects are problematic in AFM.) Thus far efforts have been concentrated mainly on determining the energy loss mechanism in the tip-sample interaction and have neglected the effect of resonance frequency shifts. A real-time frequency spectrum was obtained to decouple the two effects of change in resonance frequency and damping of the oscillation. It was also possible to determine quantitatively the elastic and dissipative parts of the interaction by accurately modeling the dynamic of the TDFM probes. Distinct regimes were found at different probe-sample separations. Although tapping-mode phase imaging remains a very useful tool for identifying and mapping regions of different properties regardless of their topographical nature, the interpretation is not always trivial. There have been several pit falls for the unwary, namely phase inversion and height artifacts. Acknowledgment. The authors would like to thank Anna Halter for her assistance with the ATR analysis and Andy Humphris for his help with the real time frequency spectrum technique. This work was supported financially by the EPSRC and National Power PLC as part of their ongoing research into regenerative fuel cell technology.
LA000332H