Mechanical Probes Move Beyond Optics
E-beam and FIB technologies help mechanical probes meet measurement challenges.
by Alexander E. Braun Assotiate Editor -- Semiconductor International, 4/1/1998
Waveform measurement tools -- whether logic, memory or mixed signal - are experiencing a rapid evolution. Three years ago, when linewidths were large, as in 0.5 µm CMOS processes, using optical microscope-based probing stations for mixed-signal applications with mechanical probes was not an issue. As CDs have shrunk, users have hit obstacles.
The limiting factor is not the probe, but the tool that manipulates it. The perception of how good a probe is may be unrelated to its electrical characteristics. Instead, it lies on ease of use and manipulation to touch on a line and measure a signal. Regarded solely by electrical performance, mechanical probes give satisfactory results for mixed-signal. Manipulated properly, probe performance is strong.
Since an e-beam's capacitance is 0 pf, it is a logical choice for probing logic and memory parts. The very best mechanical probe on the market today has a 100 femtofarad capacitance, and for logic and memory design debug and failure analysis cannot compete with an e-beam probe that does not alter signal amplitude.
| Screen capture showing a mechanical probe in the same field of view as the die. A hole has been made to access the buried conductor. The active probe has a 0.2 µm radius, and the shaft is 10 µm. Depth focus of the FIB microscope is more than 200 µm (Source: Schlumberger Automated Test Equipment) |
Some vendors, such as Schlumberger (San Jose, Calif.) have developed mechanical probes in a vacuum chamber for use with e-beam and FIB. The idea is to use e-beam's superior timing accuracy to measure small existing nodes inside the chamber. Voltage accuracy is about 50 mV, so if the user goes below 50 mV, e.g., for mixed-signal devices, the mechanical probe comes into its own. The combination of an e-beam contactless probe and a mechanical probe in a vacuum chamber provides not only the necessary resolution to manipulate the probe, but also the capability to measure an entire range of signals. Partnering a FIB with a mechanical probe provides the capability to probe mixed-signal and doing FIB probe point to access deep lines (see Figure).
Three microscope specs determine the easy and accurate manipulation of mechanical probes: image resolution, depth of focus (DOF) and working distance.
A top-of-the-line optical microscope's resolution is an area about 100 nm in size. Thus, when working with a 0.25 µm line, one sees a spot size about half its size -- not the best visibility. A 1 keV SEM can go below 100 nm to a <2 nm spot size. A FIB carries this to <5 nm. This is an important consideration if one reflects that when 0.18 µm (180 nm) processes come on-line, linewidths will be 0.25 µm.
DOF is the ability to see two things simultaneously. An optical microscope must get close to the object to see it. Focus is very sensitive to z-axis distance. An optical microscope's DOF is a few microns. So, when manipulating a probe, the choice must be made to see well either the wafer surface or the probe head -- both cannot be seen well simultaneously. This situation can result in a crash into the device surface during probe manipulation.
A SEM's DOF is a few tens of microns, a FIB's a few hundreds of microns. Thus, even if the probe is far from the device surface, it is possible to see well and in focus both the probe head and the device surface. There is little risk of "flying low" and damaging the device surface.
The third spec, working distance, means an optical microscope's objective lens must be very close to the target - 1 mm or less. The probe must slide between the lens and the wafer. A SEM or FIB offers much greater working distance -- 10 mm with an e-beam, up to 40 mm with a FIB -- without jeopardizing image resolution. Thus, the trend for the mechanical probing station of the future no longer lies with an optical microscope, but with a SEM or FIB.
Choosing between a SEM- or FIB-equipped station is simple. If the object tested is a multiple-layer mixed-signal device, little can be accessed from the top. If it is necessary to access the deep metal oxide, a probe point is required - a minute, beam-created hole. Here, the choice is FIB, because it has that capability. FIB microscopes will probably find wider acceptance because of this and, allied with the mechanical probe, will continue serving critical measurement needs.