Kelvin Probes and Kelvin Probe Systems are currently in a redesign.  Please check back in 4-6 Months.


We are able to synthesize a wide variety of ionic interfaces by depositing ions in or on molecular-scale films, using a newly developed, very low energy, molecular ion source. This accomplishment is illustrated in two examples, one of hydronium ions interacting with amorphous water films, the other, of cesium ions with hexane multilayers. In these cases we report on stress-induced dielectric responses and ion diffusion, as probed by Kelvin probe measurements of the potentials created by the ions in the films.

Ions play an important role in most chemical systems in everyday life. For example, the chemistry of soils or cells is often dominated by dissolved ions interacting with molecules and the surrounding interfaces at thermal energies. Ion beams, on the other hand, made in vacuum chambers, typically smash into their targets at tens to hundreds of eV’s of energy, far in excess of the bond energies of molecules, leading to high energy, fragmentation-dominated chemistry. We describe an experimental apparatus which combines the ease of manipulations of ions afforded by ion beams with the low ion energies encountered in solutions. We show how a chemical system can be assembled stepwise and then ions can be monitored as the system is allowed to evolve. This technique should enable new ways to measure complex processes involving ion migration and charge transfer mechanisms, with applications in fields as diverse as photosynthesis, photography, and electrochemistry.

A recent article discussed soft-landing of ions [1], and a paper from Graham Cooks’ group reported [2] exciting results for soft-landing of ions on self-assembled monolayers. However, Cooks’s lab used about 10 eV ion energies, and had to rely on the cushioning effects of bulky ions and special films on the target to successfully soft-land ions. Our system operates at a much lower ion beam energy, permitting a much broader application of ion soft landing. It was developed in collaboration with Ellison and Biesecker at the University of Colorado [3], and delivers as low as 0.5 eV beams at several to tens of nanoamps of mass-selected molecular ions. We have used it for NH4+, D3O+, H2O+, and Cs+, for example.

We prepared targets by condensing at low temperature a thin film of a desired composition, to a thickness ranging from a few atomic layers to as many as 30,000, on top of a single crystal Pt(111) surface. Such a structure is schematically shown in Fig. 1a. The ions initially have 300 eV of energy, and are decelerated down to the final impact energy on the last 1 cm of their path. The ions hit the target with about 1 to 0.5 eV, slowly enough that they do not normally break apart either the target molecules or themselves.

The first experiment shows how we can measure ion migration through thin insulating solids or liquids. We prepared thin films of hexane ices, either crystalline or amorphous. The amorphous ice is, when above the glass temperature, a true liquid, even though the temperature is below the normal freezing point. [5].

We monitored the voltage these ions produced across the film, using a Kelvin probe. A Kelvin probe [6] is a vibrating gold plated plate held near but not touching the surface, as shown in Fig 1a. It measures the difference in work function between the probe and the target. This includes the voltage caused by the ions sitting on or in the surface film we grow. If we subtract the signal from the clean surface, we get the “film voltage” , the potential difference between the top of the assembled film and the substrate. This voltage is proportional to the amount of ions placed, to their average height above the substrate, and inversely proportional to the hexane dielectric constant, which should be approximately 1.9.

We wanted Kelvin probe measurements fairly often, during temperature ramps at about .16 to 1 K/s. For this we used the “fitted sine” data analysis mode of the McAllister software. For a drive frequency of ω, this software takes the digitized Kelvin probe signal, and calculates the best fit sin(ωt) amplitude and phase.We operate the Kelvin probe head at about 105 Hz (resonant frequency is about 65 Hz). 105 Hz is 7/4 of 60 Hz. To optimize data collection, and rejection of 60 Hz noise, we digitize in one continuous span precisely 7 complete cycles of 105 Hz. When the contact potential difference is within range of the to ±5 volts of the backing potential, with no further averaging, with 2 steps of backing potential, ( -4, 0, and +4 V) we get contact potential differences about 1 per second with about 10 mV of noise. We require about 200 ms of dead time delay between measurements at different backing potentials, a major time cost. We have a 0.25 inch diameter probe about .2 mm from the sample. We typically stop our closed cycle refrigerator used for sample cooling during the measurements, as this causes too much vibration. During heating the sample bows slightly. We have used the auto-tracking mode of the electronics to compensate for this, but we seem to be at the high-speed limit of its capability, so more typically we moved the probe slightly further from the sample than otherwise optimum, and ran with the auto-tracking off. We use a modified version of the software with an additional output channel, which feeds out an analog version of the calculated contact potential difference(at 0.01, .1, 1, 10, 100 gain, ±5V maximum output). This is inputted by our main data collection program.

Figure 1

Figure 1.
a. Ion deposition and diffusion. Soft landing of ions to create ionic interface shown for case of hexane. Physical phenomena occurring during heating are shown, corresponding to indicated features in b).
b. Voltage across the hexane film due to the presence of Cs+ ions, during temperature ramp. The first step ending at 95 K marks full crystallization, the second the multilayer desorption of hexane. Desorption rate shown in upper inset [12]. The voltage across 3-methyl pentane (dotted lower trace) shows no crystallization.

Often in these measurements our contact potential differences exceeded the +/- 5 range of the backing potential: we have measured as high as 70 V changes. This forces the software to extrapolate to the zero crossing of sinusoidal signal versus backing potential. This degraded the signal-to-noise of the contact potential difference, which seems a shame, given the high magnitude of the signal under those conditions. Probably a new algorithm could improve this situation (under our unusual conditions), and we will discuss this with the vendor.

A 37 monolayer thick, amorphous hexane film was deposited at 30 K . Next, 0.3 µC of Cs+ ions over 1 cm2 (about 0.001 monolayers) was placed on top. This gives rise to 3.7 V film voltage across the 200 Å film. To keep the ion energy below 1 eV despite the charging, the target voltage was adjusted several times during the ion deposition. From these numbers a capacitance of 70 nF is calculated and by taking into account the exposed area a dielectric constant of approximately 2 is determined, in good agreement with the expected dielectric constant.

If we monitor the film voltage as we slowly ramp the temperature, as shown in Fig. 1b, we expect at some temperature the ions will become mobile and will begin to migrate through the hexane film under the influence of the electric field. This begins to occur at about 70 K (see fig. 1b) and is evident by an increasingly rapid decline of the film voltage. At any given temperature, the voltage relative to the starting value provides a direct measurement of the average location of the ions (see the right axis in fig. 1b). Since the temperature ramp rate used was 0.33 K/s, an average ion velocity in the 85-95 K range is about 1 Å/s. The corresponding “ion mobility” (drift velocity over electric field intensity) can then be determined from the electric field of 2×108 V/m to be 6×10-19 m2V-1s-1.

This average value for the ion mobility is not the whole story in the case of hexane. As seen in Fig. 1b, a sudden transition takes place at 95 K. From separate infrared spectroscopy studies, we found that amorphous hexane begins to crystallize between 80 and 90 K, and is completely crystalline by 95 K. Cs+ ions are known from our studies with pre-crystallized hexane (not shown here) to move only slightly or not at all through crystalline hexane. So, once hexane begins to crystallize, ion motion can only occur in the amorphous spaces between the crystallites, as suggested in Fig. 1a. This creates a complicated ion migration kinetics, which ends with the ions being locked in place as the hexane fully crystallizes.

Fig. 1b shows also the chamber pressure rise due to desorbing hexane which occurs near 140 K. The ions move readily through crystalline hexane near the desorption temperature. The last small change in the film voltage takes place in the 180-240 K range and is associated with the increase in work function of Pt during the desorption of the last monolayer of hexane.

Figure 1b shows the simpler behavior using 3-methyl pentane instead of n-hexane. 3-methyl pentane is known to be an excellent glass former, and resistant to crystallization [5]. For a similar ion dose and film thickness, the ions are seen to move cleanly through the film. From those data we can abstract the activation energy of diffusion, which is 24 kJ/mole (0.25 eV).

The second experiment tackles ion motion in water. We would like to understand what the nature of ion diffusion of hydronium ions (D3O+) is, whether quantum tunneling is important, and how the electric field strength can alter the barriers to diffusion.

An important difference from hexane or other non-polar solvents, is that the dielectric constant of polar water can be very high (80 to 300), and strongly temperature dependent. However, it is known that the ability of condensed water to rotate in response to an applied field is thermally activated, and should be quenched below 150 K for crystalline water, and below 130 K for amorphous water.[8]

Figure 2 shows the results for a 60 monolayer-thick film of amorphous deuterated water deposited at 30 K with 0.6µC of deuterated hydronium ions placed on top. The magnitude of the voltage established across the water layer indicates that the capacitor has a dielectric constant of approximately 2.1. Thus the water rotations are frozen at 30K.

Figure 2

Figure 2. D3O+ ions on amorphous deuterated water. Voltage across film during temperature ramp is shown. The steep drop is caused by drastic change in the effective dielectric constant of amorphous ice; the second drop is caused by the desorption of water from the Pt crystal, shown in inset.

As the sample temperature is raised (see fig. 2) a dramatic, irreversible drop in the film voltage is observed below 70 K. An equivalent drop is seen when an insulating layer of hexane is added between the water layer and the ions, which indicates that the voltage drop seen near 50 K is due to a large change in the water’s dielectric constant. Studies have shown that amorphous water deposited at very low temperature is highly stressed, has low average density (0.6 g/cm2), and even contains empty cavities (microvoids).[9] These structural defects heal progressively during heating from 20 to 120 K. During this healing the water molecules can orient in the electric field, causing the dielectric constant to “turn on” at these very low temperatures.

Between 80 and 160 K, the film voltage remains nearly constant. Around 160 K, a small voltage change coincides with a shoulder peak in the water desorption signal (also shown in Fig. 2) which signifies the crystallization of the amorphous ice [10]. Also at about this temperature, the water starts desorbing and the ions are free to move closer to the metal. The last step in the film voltage signal reflects this motion, although part of this change is related to the complete removal of the last monolayer of water from the Pt surface.

An important conclusion beyond these particular experiments is that the soft-landing of ions using this ultra-low energy ion source makes it possible to routinely do what has been considered as impossible thought experiments [11], for a very wide range of ions and substrates. Inspired by many previous pioneering ion-surface approaches [1][4], we took this technique to its logical, technologically feasible, next step turning it into an exciting new tool for interfacial studies.

References and Notes

  1. A. W. Kleyn, Science 275, 1440 (1997)
  2. S. A. Miller, H. Luo, S. J. Pachuta, and R. G. Cooks, Science 275, 1447 (1997)
  3. J. P. Biesecker, G. B. Ellison, H. Wang, M. J. Iedema, A. A. Tsekouras, and J. P. Cowin (submitted to Rev. Sci. Instrum.)
  4. D. R. Strongin, J. K. Mowlem, K. G. Lynn, and Y. Kong,, Rev. Sci. Instrum. 63, 175 (1992); H. Kang, S. R. Kasi, J. W. Rabalais, J. Chem. Phys. 88, 5882 (1988)
  5. C. A. Angell, Science 267, 1924 (1995); G. Fischer, E. Fischer, Molec. Photochem. 8, 279 (1977)
  6. Kelvin probe is a McAllister Technical Services KP5000 system. See also J. W. He and P. R. Norton, J. Chem. Phys., 89 (1988) 1170, and references therein.
  7. M. Tuckerman, K. Laasonen, M. Sprik, and M. Parrinello, J. Chem. Phys. 103, 150 (1995); N. Agmon, Chem. Phys. Lett. 244, 456 (1995)
  8. G. P. Johari, J. Chem. Phys. 105, 7079 (1996); P.V. Hobbs, “Ice Physics” Clarendon Press, Oxford (1974), p. 123
  9. D. E. Brown, S. M. George, C. Huang, E. K. L. Wong, K. B. Rider, R. S. Smith, and B. D. Kay, J. Phys. Chem. 100, 4988 (1996)
  10. R. S. Smith, C. Huang, E. K. L. Wong, and B. D. Kay, Surf. Sci. 367, L13 (1996)
  11. F. D. Wagner, “Simulation of the Electrochemical Double layer in Ultrahigh Vacuum”, in Structure of Electrified Interfaces, J. Lipkowski, P. N. Ross eds., Wiley-VCR Pub, 1993, p 309-400
  12. Desorption peak appears split because of the partial obstruction by the Kelvin probe.