SURFACE CHARACTERIZATION OF MEMS ANOMALY

John D. Olivas and Stephen Bolin
Jet Propulsion Laboratory
(818) 354-0413

INTRODUCTION

MEMS is an acronym for microelectromechanical systems and are classed as either sensors or actuators. In general, they perform the same functions as conventional macroscopic sensors and actuators, but on a microscopic scale where overall component size is on the order of millimeters. For the present study, the MEMS device investigated was a hydrophone sensor currently being developed by JPL for the US Navy. Specifically, the study focused on issues of surface quality surrounding a portion of the component which had undergone a visible change during the manufacturing process. The indications had been observed on several components from several lots upon final inspection and was suspected to account for adverse effects on the performance of the hydrophone.

In general, MEMS may be made by a variety of microfabrication techniques. The fabrication methods can be divided between additive and subtractive techniques. The hydrophones investigated presently were fabricated by additive techniques which included plating, various forms of vapor and liquid phase deposition and diffusion bonding methods. Some form of controlled etching was also used to remove material and oxidize the exposed silicon. The device is a two piece silicon structure which has a tunneling tip and gold lead on one surface and the sensing surface and framing on the other. It was not clear where or how in the process the observed indications were formed.

The purpose of the present study was to characterize the differences between the region in question (A) and the reference region (B) of the silicon with tunneling tip shown in Figure 1. The surface of the region in question, macroscopically, was characterized by a whitish hazy appearance.

Figure 1: Macrophoto of silicon hydrophone with tunneling tip. Region in question identified by A and reference region identified by B.

The objective of the study was to better understand the process by nondestructively evaluating the observed indications.

EXPERIMENTAL TECHNIQUE

X-ray Photoelectron Spectroscopy (XPS). XPS is a surface sensitive spectroscopic technique which extracts quantitative (>0.1% at.) and qualitative chemical information to depths of 10-100 Å below the surface. The theory of operation extends from the fact that all elements, except hydrogen have core electrons not used in bonding with other elements. Recognizing the quantized nature of electrons, the binding energies of core electrons are element specific. In XPS, a sample surface is irradiated by a source of low energy x-rays (Mg Ka of 1253.6 eV). The interaction of the x-rays with the core electrons result in photoionization and the ejection of photoelectrons and associated Auger electrons. Quantitative detection of these electrons which escape from the surface region are the two features primarily associated with XPS analysis. In addition, information regarding particular species is discerned by evaluating shifts in the peak position due to variation in binding energies.

RESULTS AND DISCUSSION

Results. High magnification images of the surface topography were obtained using scanning electron microscopy. The images of both the region in question and a reference area are depicted in Figure 2. Initial inspection noted the presence of depressions in the region in question and other surface irregularities. These were not present in the reference region. While the severity of the surface irregularities were consistent with the observed macroscopic pattern, observations did not lead to conclusions as to the source.

Figure 2: SEM surface images of regions identified by A (left) and B (right) in Figure 1.

Figure 3: XPS survey spectra for as received and sputtered surface of regions identified by A (left) and B (right) in Figure 1.

Results - XPS. The XPS survey spectra from region A and B are given in Figure 3. The surfaces were analyzed in both the as received and argon sputtered conditions Sputtering, approximately 180 sec, was performed to remove surface contamination resulting form normal adsorption of ambient gases (O2, H2O, CO, CO2, etc.).

In general, the presence of silicon, oxygen and carbon was as expected considering the nature of the substrate and background contamination. The most notable difference between regions A and B was the presence of a small amount (~0.5% at.) of gold in area B. It should be noted that this spectrum was taken in an area which was expected to be free of gold. The presence of gold in these insulating areas may also be cause for component failure due to electrical shorting. Sputtering demonstrated the percentage of gold contained in the surface region was >1% at. and not merely surface contamination. Rather, this indicated that the gold was present below residual surface contaminants. The reduction of the carbon 1s peak and subsequent increase in the silicon 2p and oxygen 1s peaks were indicative of ambient carbonaceous species contaminating the surface.

A detailed peak analysis of the carbon 1s peak in the both regions before and after argon sputtering gave an indication as to differences in the chemical nature of the surface. This is shown in Figure 4.

Both areas demonstrated bonding characteristic of C-C type bonding in the before and after argon sputtering conditions. However, both areas show the presence of carbonyl groups. All conditions noted the presence of C-O, however, the presence of C=O in the region of question before sputtering was indicative of chemical differences between the two regions, probably a result of processing. Due to peak overlap between the C-C and C-H primary C 1s peaks, it is difficult to asses whether the C=O is a result of residual hydrocarbons or may have a hydrocarbon basis.

Figure 5: Silicon 2p detail peak for as received and sputtered surface of region identified by region A (left) and region B (right). Primary SiO2 peak located at 103.4 eV.

Figure 4: C 1s detail peak for as received and sputtered surface of region identified by region A (left) and region B (right). Primary C-C peak located at 284.6 eV.

A detailed peak analysis of the silicon 2p peak in the both regions before and after argon sputtering further indicated differences in the chemical makeup of the two regions. This is given in Figure 5.

Figure 5: Silicon 2p detail peak for as received and sputtered surface of region identified by region A (left) and region B (right). Primary SiO2 peak located at 103.4 eV.

The principal peak was associated with Si-O type bonding typical of SiO2. The principal difference noted was that after sputtering of the region in question, there appeared to be a broadening effect of the silicon 2p peak which may have been associated with a procedure that partially reduced the SiO2 to a suboxide of SiOX type. This indicated that the process was most probably an aggressive attack on the surface due to the inherent stability of SiO2 in ambient condition.

CONCLUSIONS

The present study nondestructively evaluated surface quality of a MEMS hydrophone sensor currently being developed by JPL, using XPS. While the root cause of the defects was not determined, several characteristics indicative of potential problems associated with the process were established. The presence of C=O and peak broadening of the Silicon 2p peak in the region of question before sputtering was indicative of differences which resulted from the processing which caused the indications observed. This was significant when recognizing that both the control and the region in question were in the same proximal location and processed under similar conditions yet spectroscopically were significantly different. Additionally, the presence of gold in the control region indicated that process variables are not completely understood. The next step in this ongoing investigation is to work closely with the process and identify the variables which will eventually lead to the indications observed and the process controls needed to prevent this from happening in the future.