A schematic diagram of the stress sensor is shown in Fig. 1. In essence it is an optical-deflection AFM without a sample . The housing for the sensor is machined from a single block of aluminum. The electrochemical cell is made from Teflon with a glass window through which the optical beam passes. The gold-coated lever itself is the working electrode and reference and counter electrodes are positioned within the cell for electrochemical control. Electrical contact to the lever is made via a thin insulated wire which is glued to the lever holder with conducting epoxy resin. This connection is then insulated with an inert, insulating epoxy resin. The quality of the voltammograms that we have obtained indicates that there is minimal contamination caused by the epoxy resin.
The gold coating on the as-supplied silicon nitride cantilevers is stripped with aqua regia and a clean (111) oriented gold surface is then deposited by vacuum evaporation on the tip side of the lever. The gold film deposition is carried out at a pressure of less than 10"7 mbar and at a temperature of 270°C and the films are then annealed at the same temperature for 4 h before being allowed to cool down slowly (Balzers UTT400
UHV evaporation system). This procedure is similar to that used for the vacuum evaporation of well-oriented gold (111) films on mica substrates , The (111) orientation of the films was then determined by X-ray diffraction. The stress change can be determined directly from the optical-beam deflection signal. The stress change is given by Stoney's equation:
where a\ and 02 are the surface stresses in the two faces of the cantilever, E the Young's modulus of the lever, t the thickness of the lever, n the Poisson ratio of the lever and r the radius of curvature of the bent cantilever. The values of E (1.5 xlO11 Nm"2) and n (0.3) are those of bulk amorphous silicon nitride. In order to interpret these lever deflections it is necessary to assume that the surface stress of the silicon nitride face is constant. The voltages applied to the working electrode are small (not more than +1V). This has been shown not to alter the electrical characteristics of silicon nitride , so the stress changes can be regarded as surface stress changes of the gold electrode surface. The quoted thickness of the levers is 0.6mm. This was confirmed by measuring the lever resonant frequency. The photocurrents are proportional to the measured output voltages A and B. However, care must be taken to consider the effect of a reflected beam from the glass/solution boundary also impinging upon the detector. To take this into account we measure the A +B signal when the beam is just missing the tip of the lever. This is then subtracted as the background signal (A+B)0fj. By considering the geometry of the measurement system and applying Stoney's formula (above), the stress change Act can be written as:
where (A +B)0„ is the signal from the detector when the laser beam is focused on the lever. From Eq. (2) it is clear that using this approach only changes in the surface stress are experimentally accessible rather than the absolute surface stress value, because there is no reference point of known stress.
The linearity of the photodiode detector was checked using neutral density filters and over the range of operation the output voltages are proportional to the incident beam intensity. Some care is required in the beam alignment to ensure that all of the reflected beam impinges upon the detector compared to the earlier error term, this is negligible.
The photodiode-lever separation can be measured sufficiently accurately to ensure that it is a negligible source of error. The value of this distance varies slightly from scan to scan, depending on the beam alignment, but is typically 12.0 mm. Finally one must consider the geometrical and mechanical properties of the cantilever. Because the stoichiometry of the cantilevers can vary from that of bulk amorphous silicon nitride, the values of the Young's modulus and Poisson ratio are not very accurately known. We have used quoted values for bulk silicon nitride, but the accuracy of these values is hard to assess. As mentioned earlier, the quoted thickness of the levers was checked by measuring the lever resonant frequency, and the width of the levers is accurately known (10 j^m).
The error in the effective cantilever length corresponds to a percentage error in the stress changes of -6%. Furthermore, there could also be some variation in the measured stress changes because of differences in the surface roughness and morphology of individual gold films.
Electrochemical control was provided by an Eco-Chimie PSTAT10 potentiostat, which facilitated the simultaneous measurement of current and the lever deflection signal. For the silver UPD work a clean silver wire was used as the reference electrode, and for the electrocapillary curve a silver/silver chloride reference electrode was used. In all cases a platinum coil served as the counter electrode. All the electrolyte solutions were made with 18.2MW water (Elga). Silver sulfate (Specpure, Johnson Matthey), silver fluoride (99.9+%, Aldrich), potassium chloride (99.999%, Aldrich), perchloric acid (Aristar 70%, BDH) and sulfuric acid (Aristar, BDH) were used for making the electrolyte solutions. All the electrolytes were de-aerated with high-purity nitrogen prior to use.
Was this article helpful?