STM, QCM, and the Windshield Wiper Effect - American Chemical

STM, QCM, and the Windshield Wiper Effect: A Joint. Theoretical-Experimental Study of Adsorbate Mobility and. Lubrication at High Sliding Rates...
0 downloads 0 Views 176KB Size
9606

Langmuir 2006, 22, 9606-9609

STM, QCM, and the Windshield Wiper Effect: A Joint Theoretical-Experimental Study of Adsorbate Mobility and Lubrication at High Sliding Rates M. Abdelmaksoud,†,§ S. M. Lee,† C. W. Padgett,‡ D. L. Irving,‡ D. W. Brenner,‡ and J. Krim*,† Department of Physics and Department of Materials Science and Engineering, North Carolina State UniVersity, Raleigh, North Carolina 26795 ReceiVed June 22, 2006. In Final Form: August 22, 2006 We have observed that when mobile adsorbed films of benzene, tricresyl phosphate, and tertiary-butyl phenyl phosphate are present on the surface electrode of a quartz crystal microbalance (QCM), oscillation of the QCM produces clearer scanning tunneling microscope (STM) images of the electrode surface. This is in contrast to an immobile overlayer of iodobenzene, where oscillation of the QCM does not affect image quality. This observation is attributed to a “windshield wiper effect”, where at MHz frequencies the tip motion maintains a region of the surface where the absorbate concentration is reduced, which leads to a clearer image. A straightforward model is presented that supports this conclusion and that provides guidelines for effective lubrication of contacts operating at MHz frequencies.

I. Introduction Many of the issues central to the chemical and mechanical stability of moving nanostructures are closely related to the field of nanotribology.1 These include thermal versus nonequilibrium chemistry, heat flow, chemical mechanisms and rates, and the role of defects, adsorbed films, and microstructure. A fundamental understanding of these phenomena is lacking, however, because they occur at buried interfaces that are extremely difficult to characterize experimentally during the tribological process. A major challenge is therefore to perform experiments, supported by theory and modeling, that can isolate individual causal relations. When a quartz crystal microbalance (QCM) is employed in combination with a scanning tunneling microscope (STM), imaging of the buried sliding interface becomes attainable, and the relative sliding speeds and contact areas explored are much closer to likely operational conditions for nanoscale devices than other nanotribological techniques.2 We focus here on our observation that when mobile films are present on the surface of a QCM, the oscillation of the QCM can lead to a clearer STM image of the electrode than when the QCM is stationary. The first experimental observation of this effect was recorded on a blend of tertiary-butyl phenyl phosphate (TBPP) molecules on platinum. Unfortunately, this system was prepared under conditions too poorly defined to propose a sound interpretation of the result and to make meaningful comparisons with theory.3 Reported in this paper is a set of STM-QCM measurements for two additional mobile overlayers, benzene and tricresyl * Author to whom correspondence should be addressed. † Department of Physics. ‡ Department of Materials Science and Engineering. § Present address: Physics Department, University of Cairo, Giza, Egypt. (1) Krim, J. Surf. Sci. 2002, 500, 741-758. Krim, J. Phys. World 2005, 18, 31-33. (2) Borovsky, B.; Mason, B. L.; Krim, J. J. Appl. Phys. 2000, 88, 4017-4021. (3) Borovsky, B.; Abdelmaksoud, M.; Krim, J. In Nanotribology: Critical Assessment and Research Needs; Hsu, S., Ying, Z. C., Eds.; Kluwer: Boston, MA, 2002; pp 361-375. Krim, J.; Abdelmaksoud, M.; Borovsky, B.; Winder, S. M. In Dynamics and Friction in Submicrometer Confining Systems; Braiman, Y., Drake, J. M., Family, F., Eds.; Oxford University Press: New York, 2004, ACS Symposium Series 2004, 882, 1.

phosphate (TCP) on copper, in which this same effect is observed. Also reported are the results for an immobile overlayer, iodobenzene on copper, under the same conditions, where it was found that operation of the QCM does not improve the quality of the surface image. Our interpretation of these experimental results is that the presence of an electrically insulating overlayer between the tip and surface (whether strongly or weakly adsorbed) results in images in which the resolution of surface features is difficult. When this layer is mobile, however, we propose that the oscillation of the QCM acts as a “windshield wiper” that is able to maintain a region of the substrate below the tip in which the concentration of the overlayer is sufficiently reduced that clearer images of the substrate are obtained. To help validate this interpretation, we also report results from preliminary modeling in which the steady-state overlayer concentration below an oscillating tip has been calculated as a function of the product of the overlayer diffusion coefficient and one-half of the tip oscillation period. The modeling demonstrates that the windshield wiper effect can be apparent at MHz oscillations but not at kHz (and slower) oscillations for a physically meaningful range of mobile overlayer diffusion coefficients.

II. STM Imaging of a Clean QCM Electrode The basic component of a QCM is a single crystal of quartz that has very little internal dissipation (or friction). As a result, it oscillates at an extremely sharp resonance frequency (usually 5-10 MHz) that is determined by its elastic constants and mass.4 The oscillations are driven by applying a voltage to thin metal electrodes that are deposited on the surface of the quartz in a manner that produces a crystalline texture, generally (111) in nature.5 Atomically thin films of a different material are then adsorbed onto the electrodes. The extra mass of the adsorbed layer lowers the resonance frequency of the microbalance, and the resonance is broadened by any frictional dissipation because of the relative motion of the adsorbed layer and the microbalance. (4) Stockbridge, C. D. Vacuum Microbalance Techniques; Plenum: New York, 1966; Vol. 5. (5) Chopra, K. L. Thin Film Phenomena; McGraw-Hill: New York, 1969. Krim, J. Thin Solid Films 1986, 137, 297-303.

10.1021/la061797w CCC: $33.50 © 2006 American Chemical Society Published on Web 10/06/2006

STM, QCM, and the Windshield Wiper Effect

By simultaneously measuring the shift in frequency and the broadening of the resonance (as evidenced by a decrease in the amplitude of vibration of the microbalance), the sliding friction of the layer with respect to the metal substrate can be deduced.6 The STM produces a single asperity contact that can image the QCM in both stationary and sliding conditions. The microbalance, whose surface is oscillating in transverse shear motion at speeds that can reach 10 m/s, can be employed to measure the uptake rate and frictional properties of adsorbed species. The STMQCM combination allows access to a range of contact pressures and sliding speeds that are comparable to those that will be encountered in nanoscale devices. Traditional instruments such as the atomic force microscope (AFM) and the surface forces apparatus (SFA) fail to access either the required sliding speeds or contact pressures, respectively. Furthermore, the use of an STM avoids the problem of jump-to-contact associated with an AFM, allowing the applied normal force to be varied continuously in a controlled fashion. A remarkable consequence of the STM-QCM combination is that the amplitude of vibration at the surface of the QCM may be directly measured from the STM images.2 This ability to image a vibrating surface with an STM may be attributed to the three widely separated time scales involved and the exponential dependence of the tunneling current on tip-surface separation. The characteristic frequencies of the scanner (Hz), the feedback loop (kHz), and the QCM (MHz) are each separated by 3 orders of magnitude. Conventional STM operation relies on the feedback loop being much faster than the scanner (in constant-current mode). For the STM-QCM, the fact that the vibrations of the QCM are in turn much faster than the feedback loop causes the tip to be held at a separation from the surface that on average (over many surface oscillation cycles) gives the desired tunneling current (typically 1 nA). Qualitatively, the closest approach of the surface to the tip during each cycle is weighted most heavily in this average, because of the exponential dependence of the tunneling current on separation, so the tip is held at an altitude sufficient to avoid direct contact of the tip and surface. This allows the STM tip to image the vibrating surface without crashing, as is evident from a lack of damage to the surface. In addition, there is no observable normal load on the crystal when the tip is in tunneling contact, provided that both the tip and electrode surfaces are clean metals in high vacuum (e∼10-8 Torr). The presence of even a thin electrically insulating adsorbed layer on the tip or electrode surfaces results in a QCM frequency shift during tunneling contact, attributable to the normal load required to squeeze the two metal surfaces sufficiently close together for tunneling.7 For all samples reported in the present work, overtone polished A-type transverse cut 8 MHz quartz crystals8 were cleaned with acetone using an ultrasonic cleaner and were rinsed with methanol and deionized water before being introduced to the vacuum chamber. Copper films, whose thickness is 1000 Å, were deposited onto the QCM by a thermal evaporation source that contained 99.999% pure copper pellets. The base pressure of the deposition system was 3 × 10-10 Torr. After deposition of the electrode, the QCM was transferred in situ to the STM sample holder for the STM-QCM studies. For studies of lubricated contact, vapor was admitted to the chamber to allow films to condense on the copper surface electrodes, as described in the following section. (6) Krim, J.; Widom, A. Phys. ReV. B 1988, 38, 12184-12189. (7) Borovsky, B.; Krim, J.; Asif, S. A. S.; Wahl, K. J. J. Appl. Phys. 2001, 90, 6391-6396. Flanigan, C. M.; Desai, M.; Shull, K. R. Langmuir 2000, 16, 9825-9829. (8) Colorado Crystal Corporation, 2303 West 8th Street, Loveland, CO 80537, Part number: CCAT1BK-1004-000.

Langmuir, Vol. 22, No. 23, 2006 9607

Figure 1. 200 nm × 200 nm STM images of a clean copper QCM electrode while stationary (left) and oscillating (right) in a diagonal direction from upper left to lower right (indicated by the arrow).

Figure 2. STM images of TBPP on platinum (from ref 3) (left) and TCP on copper (111) (right) that have been recorded while the QCM is stationary (noisy regions) and oscillating.

Shown in Figure 1 are 200 nm × 200 nm STM images of a QCM copper film electrode imaged while stationary (left) and oscillating (right). Because the scan rate is much less than the oscillation frequency, features in the image to the right appear slightly elongated along the direction of the oscillation (diagonal from the upper left to lower right). Other than this slight elongation, the image quality of the bare surface appears largely independent of the oscillation.

III. Lubricated Contact Three systems composed of mobile lubricants with slightly different properties (TBPP on platinum, TCP and benzene on copper), and an immobile overlayer (iodobenzene on copper) were considered. TBPP is a blend of molecules that has demonstrated high-quality performance as a macroscopic lubricant at elevated temperatures; it exhibits oxidation inhibiting characteristics as well as several other desirable tribological properties, such as the reduction of wear. The images of TBPP used in Figure 2 were reported earlier.3 TCP is commonly used in synthetic oil blends as an antiwear additive. The commercial product used for the experiments is a mixture of three isomeric forms, ortho, meta, and para. Both TBPP and TCP are adsorbed with little to no surface chemical reactivity at room temperature on platinum and copper, respectively, with characteristic slip times on the order of nanoseconds.3,9 Benzene is highly mobile on copper, even at low temperatures.10 Temperature-programmed desorption studies of benzene on copper (111) show a distinct peak at 228 K that is due to desorption from terraces and a tail in the spectrum that extends to temperatures higher than 300 K that is due to desorption from steps and other defects.11 Iodobenzene reacts with copper at 175 K to create (9) Abdelmaksoud, M.; Bender, J.; Krim, J. Phys. ReV. Lett. 2004, 92, art # 176101. Abdelmaksoud, M.; Bender, J.; Krim, J. Tribol. Lett. 2002, 13, 179. (10) Stranick, S. J.; Kamna, M. M.; Weiss, P. S. Surf. Sci. 1995, 338, 41-59. (11) Lukas, S.; Vollmer, S.; Witte, G.; Woll, C. J. Chem. Phys. 2001, 114, 10123.

9608 Langmuir, Vol. 22, No. 23, 2006

Abdelmaksoud et al.

IV. Theoretical Analysis

Figure 3. Benzene (left) and iodobenzene (right) STM images recorded with the QCM stationary (upper half of each image) and oscillating (lower half of each image). The windshield wiper effect is present for benzene but not for iodobenzene.

iodine and phenyl groups, both of which are strongly chemically bonded to the surface.12,13 At temperatures above 300 K, the phenyl groups react to form biphenyl that is subsequently desorbed. The iodine remains on the surface until temperatures in excess of 900 K. Preparation of the TBPP on platinum was described earlier.3 For the other overlayers, the copper-coated QCM electrode was prepared as described above, and vapor was admitted to the chamber to allow films to condense on the copper surface electrodes. 99% pure benzene was further purified by freezethaw cycles before dosing on the copper film. To have a moderate amount of benzene on the copper surface, the chamber was backfilled to 1 × 10-3 Torr at 300 K with benzene vapor. For the iodobenzene, a pressure of 6 × 10-4 Torr was exposed to the crystal to ensure that the copper was densely coated with a chemisorbed overlayer. STM images for the TBPP on platinum and TCP on copper are shown in Figure 2. The blurry regions were recorded with the QCM held stationary, while the regions in which the substrate is more clearly visible were recorded with the QCM vibrating. The images taken during vibration are reminiscent of those of the clean copper electrode with and without QCM oscillation (Figure 1), while an unstable image with indiscernible surface features occurs for both cases with the QCM held stationary. STM images of benzene and iodobenzene on copper are shown in the left and right panels, respectively, of Figure 3. Benzene shows the same behavior as the other mobile overlayers, an unstable image with blurry surface features with the QCM held stationary (top of the image) and a surface with distinguishable surface features with the QCM oscillating (bottom of the image). In contrast, for the chemisorbed iodobenzene overlayer, there is no difference between images taken with the QCM held stationary (top of the image) and the QCM oscillating (bottom of the image). Under both conditions, the surface features are largely indistinguishable, in sharp contrast to the results with the clean copper electrode coating. From these results, it is clear that the image of the surface improves dramatically for the mobile overlayers with the QCM oscillating but does not improve for the immobile overlayer. As discussed above, we interpret this ability to produce a clearer image of the surface during high-speed oscillation as a windshield wiper effect where the rubbing action of the vibrating QCM electrode against the STM tip creates a region below the tip where the concentration of the overlayer is reduced. (12) Xi, M.; Bent, B. E. Surf. Sci. 1992, 278, 19-32. (13) Jiang, D.; Sumpter, B. G.; Dai, S. J. Am. Chem. Soc. 2006, 128, 60306031.

A straightforward model is used to establish the combined range of tip oscillation speeds and overlayer diffusion rates under which the proposed windshield wiper effect is feasible. When diffusion relative to tip oscillation is too rapid, for example, sustaining an evacuated region with tip motion alone is impossible. Similarly, too slow a diffusion rate for a mobile overlayer relative to the tip frequency would result in the windshield wiper effect being apparent both with and without the QCM operating. Finally, a completely immobile overlayer would not show the windshield wiper effect at any sliding rate because the overlayer concentration would not be changed by the tip motion. We assume that the concentration of adsorbate in a circular region with diameter defined by the tip oscillation amplitude is instantaneously and uniformly reduced by some fraction at each cycle time tc, which we define as one-half of the period of the tip oscillation. Assuming that the diffusion coefficient is independent of coverage, the adsorbate concentration profile resulting from diffusion back into the evacuated region is obtained by solving the continuum equation14

C(r,tc) ) 2



∑ 2 n)1

r0

J0(βnr)

βnJ1(βnr0)

2

2

c

e(-βn Dt )

r r′J0(βnr′)F(r′)dr′ + C0 ∫r′)0 0

(1)

In this expression, r and r′ are the distance from the center of the evacuated region, rt is the radius of the partially evacuated region, r0 is the distance from the center of the partially evacuated region at which the concentration C0 remains constant, D is the adsorbate diffusion coefficient, and βn’s are the nth roots to the first- and second-order Bessel functions JO and J1 (e.g., J0(βnro) ) 0). The function F(r′) is the initial concentration profile for the system under the tip, defined as the deviation from C0, so that F(r′) ) (f - 1) C0 where f equals the fraction removed by the tip. To obtain the steady-state concentration due to multiple passes of the tip, the calculated concentration within the area affected by the tip at each cycle time is uniformly reduced by some given fraction, and the concentration for the entire surface is recalculated from eq 1. This process is continued until the concentration below the tip at the end of each cycle has converged to a steadystate value. The resulting converged concentration fractions at the origin for a tip oscillation amplitude of 0.1 µm and fractions removed of 100%, 50%, and 10% at the end of each cycle are plotted as the symbols in Figure 4 as a function of the product of the diffusion coefficient and the cycle time. The solid lines are fits to the data of the logistic function

P(γ) )

1 1 + e-(aγ+b)

(2)

where γ is the logarithm of the product of the overlayer diffusion coefficient and the cycle time, and a and b are fitting parameters. This function provides an accurate fit to the converged concentrations, with a standard deviation of 0.6%. Values for the fitting constants for several tip radii and fractions removed are given in the table. From these data, the range of diffusion coefficients over which the windshield wiper effect will be apparent for MHz oscillations but not for kHz (or slower) oscillations can be determined assuming some fraction of concentration removed by the tip at the each cycle. For a tip (14) Ozisik, M. N. Heat Conduction; John Wiley & Sons: New York, 1993.

STM, QCM, and the Windshield Wiper Effect

Langmuir, Vol. 22, No. 23, 2006 9609

range 0.087-0.265 eV. These values are comparable to available experimental data for hydrocarbons on metal surfaces.15 For benzene on copper, temperature-programmed desorption data yields an energy of 0.65 eV for binding to steps on a (111) surface.11 Corrugation ratios are typically in the range 0.2-0.4,16 which puts this estimate for the migration energy within our predicted values. Experimental measurements that provide a more direct validation of the model are left to future work.

V. Conclusions

Figure 4. Calculated steady-state concentrations at the origin as a function of the product of the overlayer diffusion coefficient and the cycle time (γ in eq 2, which is the log of the product D × tc), assuming a tip oscillation amplitude of 0.1 µm. The calculations assume that 10% of the overlayer is removed from under the tip at each cycle (diamonds), 50% is removed (squares), and 10% is removed (triangles). Table 1. Fitting Constants for Eq 2 for Several Sets of Conditions % evacuated

rt (µm)

a

b

10% 50% 50% 50% 100%

0.1 0.01 0.1 0.2 0.1

2.613 2.662 2.834 3.117 3.052

7.411 9.994 6.191 5.349 5.919

oscillation amplitude of 0.01 µm, which corresponds to the experiment described above, for the windshield wiper effect to occur at a 10 MHz oscillation but not for oscillations of 10 kHz or slower our modeling predicts that the diffusion coefficient must be between about 3.5 × 10-5 and 3.6 × 10-8 cm2/s assuming that the concentration is reduced by 50% at each cycle and that the difference in image occurs if the converged concentration at the origin is 50% or less than the initial concentration. Surface diffusion coefficients for our overlayer-substrate systems have not been measured, and therefore a direct comparison between the theory and experiment is not possible at this time. We can, however, make some general conclusions regarding the physical significance of our model. Assuming Arrhenius behavior for diffusion and the “universal” prefactor of 1 × 10-3 cm2/s (which is derived using an attempt frequency of 1013 s-1 and a nearest-neighbor distance of a few angstroms),15 our modeling predicts that the activation energy for diffusion should be in the

Our experimental studies have shown that for a QCM-STM combination with mobile overlayers, oscillation of the QCM produces STM images in which surface features are more distinguishable than in images produced with the QCM off. In contrast, for an immobile overlayer due to adsorption of iodobenzene on copper, oscillation of the QCM does not affect the image quality. This result is interpreted as being due to a windshield wiper effect in which the tip oscillates so fast (MHz) with respect to the electrode with the QCM operating that it creates a patch on the surface with a depleted concentration of overlayer that allows clearer images of the substrate. This proposed mechanism is supported by a theoretical analysis that shows that diffusion of typical weakly bound overlayers on metal surfaces falls into the regime in which the overlayer concentration may not be fully recovered on the time and length scale of the STM-QCM operation. With a highly mobile film, holes or clear areas produced in this way may not be permanent, however, and may readily fill with additional lubricant after the QCM is turned off or the tip is moved to a different position. The ability of a lubricating film to replenish its depleted areas is a requirement for effective lubrication for macroscopic scale machines. For nanoscale machines, however, where sliding rates may be comparable to the diffusion rate for the lubricant, the same lubricant may not be optimal, and other lubricating schemes, perhaps involving a combination of bound and mobile lubricant, may be appropriate. Acknowledgment. This work has been supported by the Extreme Friction MURI program, AFOSR grant FA9550-041-0381, and in part by DOE grant DE-FG02-01ER45936, ONR grant N00014-04-1-0263, and NSF grant DMR0320743. LA061797W (15) Barth, J. V. Surf. Sci. Rep. 2000, 40, 75-149. (16) See, for example: Arena, M. V.; Westre, E. D.; George, S. M. J. Chem. Phys. 1992, 96, 808-816. Arena, M. V.; Westre, E. D.; George, S. M. J. Chem. Phys. 1991, 94, 4001-4008.