edfas.org 7 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 25 NO. 2 fabrication facilities, as material properties can change depending on how those thin films were deposited. An overview of a delayering tool is shown in Fig. 4. Figure 4a shows the tool itself, including a single wafer load lock and a process chamber. By minimizing pump down volume, the load lock reduces the analysis cycle time compared to venting the entire system for each chip. As the actual etch step takes a few to tens of minutes, it would be dominated by vacuum pumping and venting. In Fig. 4b the stage is in the center, complete with tilt, rotation, water cooling, and shutters. The ion source, to the left in 4b, is shown better in 4c. The source is 12 cm diameter. Etch selectivity can be controlled through several methods. The physical sputtering process, by which the chips are etched, is not trivial to model and is captured in the single parameter called the “sputter yield.” Sputter yield is the average number of substrate atoms kicked out per incident atom. This is a mainly physical process, like “bowling with atoms.” Sputter yields vary from 0.5 to 2. The sputter yield is dependent on several factors including the materials, both gas and etch material, the ion energy, and the angle of incidence of the ion to the surface. None of these trends are easily predictable for a given material, and do not follow simple equations. Adding reactive gases such as O2 to the source can provide further control. The result is that, since different materials follow different trends, by only changing one or two parameters, such as the tilt and beam voltage, two materials sputter yields (etch rates) can be matched. An example of this is shown in Fig. 5. In Fig. 5, an experiment was run by keeping the beam voltages, currents, RF power, and gases all constant while only changing the incidence angle. The etch rate was measured by looking at the mass spectrometer signal strength after calibrating with a test etch sample. At incident angles below 70 degrees, the etch rate of the soft Al film is higher. At angles above 70 degrees, the etch rate of the SiO2 is higher. Thus, there is a sweet spot around 70 degrees where the etch rates are matched and the selectivity drops to zero. This experiment was performed with separate thin films, but it has been observed that the architecture of the chip can affect the result (the ratio of Al to SiO2 as well as linewidths). SECONDARY ION MASS SPECTROSCOPY When an energetic ion impinges on a solid surface, atoms from the surface are ejected or sputtered. A fraction of these ejected atoms are also ionized. These socalled “secondary ions” contain information about the elemental, isotopic, and molecular composition of the surface. Secondary ion mass spectroscopy uses a quadrupole mass spectrometer to measure the masses of ions entering the instrument (Fig. 6). Depending on the mass and charge of the ions entering the electric and magnetic fields, their trajectory will be deviated greater or less. Some will be collected by the detector and most rejected. By scanning the voltages and currents in the device, a full spectrum of signal versus mass can be collected. A typical device on the market can scan 0 to 300 AMU in less than a second and has a working range from 50 counts per second (limited by noise) up to 107 counts per second. Fig. 5 Etch rate of SiO2 and Al thin films as incidence angle (tilt) is changed. At around 70 degrees incidence, the etch rates are the same and selectivity drops to zero. Gas, beam currents, and voltages are constant in the experiment. Fig. 6 An illustration of an ion source etching a substrate and ions being collected by the SIMS.
RkJQdWJsaXNoZXIy MTMyMzg5NA==