edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 27 NO. 1 20 current-carrying wire 50 µm below the surface where the sensor is placed, would be ~ 5 µm. OPERATION OF A QDM FROM THE FA ENGINEER’S PERSPECTIVE Operation of a QDM involves a two-step process: Step 1: The DUT must be prepared within the biasing conditions. Secondly, the sensor positioning must take place. The interaction of the DUT with the quantum sensor can be achieved in two main ways. Either the quantum sensor is a stand-alone diamond chip and is placed directly on the sample, or it is integrated into the microscope objective and is brought close to the sample. In the stand-alone case (as illustrated in Fig. 1b), the sensor can directly be placed on top of the region of interest on the DUT. The diamond is in contact with one side of the sample, therefore achieving a minimal stand-off distance to the area of interest (limited by the flatness of the sample surface). Note that despite being placed on the surface, the sensor collects information through the layers of the DUT. As an example, a QDM that uses the stand-alone sensor configuration and has a sensitivity of ~ 5 µT(Hz)-0.5 will be able to image currents of ~ 1 mA buried at a distance of up to ~ 300 µm with a reasonable signal to noise ratio of 3 within ~ 20 seconds. In the case of diamond sensors integrated into the optical setup,[13] automation is essential for aligning the sensor with the region of interest, while tilt correction should ensure flatness. The depth reach and resolution of the integrated sensor will diminish as the distance between the objective head and the sam- ple increases. Generally, the diamond sensors are usually cuboid chips with sides of length 1-4 mm, which allows to image a FoV up to 4 mm x 4 mm. The stand-alone case is appealing when samples have rough and hard-to-reach surfaces (typical for wide band-gap devices such as GaN), because the objective can be moved farther from the sample stage, while the integrated case is appealing when samples are flat and automation is desired, e.g. when stitching data over a wafer. Step 2: After sensor positioning is done, the measurement is automated and requires no interference from the operator. Magnetic field maps are created, and the source current density images are extracted. The operator can use infrared (IR) backside images and their knowledge of the layout (if available) to precisely pinpoint current paths in either a single or multiple layers. QDM IN THE FA WORKFLOW As an MCI method, the QDM can identify all failures addressable by traditional MCI techniques while also overcoming certain prior limitations.[7] Generally, the addressable faults include shorts, leakages, and highresistance faults. However, the QDM can also identify true non-resistive opens using AC-currents on the DUT, which remained a challenge for many imaging methods to this date.[1] Table 1 gives a summary of addressable failure modes. QDM is a versatile tool that can be incorporated into the FA workflow in two different points. First, it can be used as a nondestructive inspection tool before sample preparation. This could either be over-the-package or after decapsulation without thinning and polishing. It is generic to use LIT for this step in FA labs that have access to such a device. Here, QDM can be seen as a supplement to the hotspot information, but also has the potential to surpass the depth detection and resolution capabilities of LIT. These prospects are actively being investigated. Second, the QDM can also be used Table 1 List of addressable failures for packages and dies for QDM. Package-level Package high-R Die-level Die high-R Shorts, leakages, opens Bumping voids C4 and µ-bumps, vias Power shorts, IDD leakage, I/O leakage Upper metal defects Fig. 2 Flowchart of a generic FA workflow. Steps 2 and 4 are value-proposition points for QDM as a versatile tool that can be used both before and after sample preparation, with different benefits: nondestructive overview and very high lateral resolution, respectively.
RkJQdWJsaXNoZXIy MTYyMzk3NQ==