In Situ Time-dependent Dielectric Breakdown in the Transmission Electron Microscope: A Possibility to Understand the Failure Mechanism in Microelectronic Devices

The overall goal of the following experiment is to improve the procedure. Used to study the time dependent dielectric breakdown failure mechanisms and degradation kinetics in copper, ultra low K interconnect stacks. This is achieved by designing and fabricating a dedicated tip to tip structure consisting of fully encapsulated copper interconnects and insulated by an ultra low K material inside a full wafer.

As a second step, prepare an intact H bar sample using focused ion beam thinning in the scanning electron microscope. Next, establish the electrical connection and perform in situ electrical tests in a transmission electron microscope. To study the degradation kinetics and failure mechanisms of the copper ultra low K interconnect stacks.

The results show the degradation kinetics and the time dependent dielectric breakdown failure mechanisms in copper, ultra low K interconnect stacks. Failure analysis is typically done after the electrical test, which is done at the wafer level, always packaged parts. This only gives a very limited amount of information on the true failure mechanism.

The main advantage of our technique is the fact that we're able to follow the failure mechanism in situ in the TEM, the procedure will be shown by Yvo Duncan, our technician, Mula, our TM expert, and John a PhD student in our lab. To begin, put on an antistatic wrist strap to prevent damage to the sample. Next, mark the positions of tip to tip structure on the wafer.

Then use a diamond scribe to cleave the full wafer into small chips that are approximately 10 millimeters by 10 millimeters square. Fix the chips on a six inches silicon wafer with hot wax using a dicing machine. Saw each chip to obtain bars that are 60 microns by two millimeters and are centered on the tip to tip structure.

Then glue the target bar on a copper half ring using super glue. Next, glue the bar on a copper sample stage. Also using the super glue.

Then use silver paste to set the conduction between the half ring and the copper sample stage. Place the sample onto the SEM stage and place the stage carefully into the microscope. Next, choose the deposition mode to add a platinum protection layer using a 30 kilovolt ion beam.

Tune the current to around 150 pico amps in order to get a satisfied efficiency for the desired platinum layer dimensions. Deposit a platinum line to contact one pad to the copper stage as the ground potential in a similar manner. Deposit a thick platinum layer on top of the tip to tip structure.

This is very important as it minimizes the ion damage during the focused ion beam thinning process and reinforces the thin lamella. Take care not to introduce any conductive connections between the two pads of the tip to tip structure on top. Otherwise, you will destroy your structure.

Use a voltage of 30 kilovolts and a current of 10 pico amps to thin the target bar into an H bar. TEM lamella with a thickness between 150 and 180 nanometers. Cut a notch close to the positive voltage pad using focused ion beam thinning.

The notch will be used as a marker to identify the correct pad in the TEM and will be touched by a transducer tip in the transmission electron microscope. Put on the antistatic wrist strap before touching the sample. Then remove the prepared hbar sample from the SEM stage while still keeping it on the copper stage.

Affix the copper stage onto the TEM holder and move the transducer tip of the TEM holder close to the test structure under an optical microscope. Next, insert the sample into the TEM Carefully keep the time for the sample transfer within 15 minutes or shorter. To avoid too much exposure to ambient moisture and oxygen, connect the TEM holder to its control system and the source meter.

Then switch on both the control system and the source meter. Begin the course approach of the transducer tip to the test structure by tuning the knobs on the TEM holder. During the approach, monitor the transducer tip in the window.

Next, move the transducer tip of the TEM holder to within 500 nanometers of the positive voltage pad. Bring the transducer tip to the same Z height as the pad, and then tune the position of the tip to make the tip face the center of the positive voltage pad. Set a potential of 0.5 volts to about one volt on the tip while approaching the positive voltage pad.

Monitor the current simultaneously to know when contact is established. Move the electron beam to the area of interest. Choose a proper magnification and focus on the image.

Use low illumination steps to reduce the beam damage on the test structure. Also use a condenser aperture to localize the illumination area only within the thin part of the hbar sample. Apply a constant voltage of less than or equal to 40 volts on the tip to tip structure using the source meter while recording two to three frames of the TEM images in C two.

Pause the experiment when seeing an apparent diffusion of metal into the ULK dielectrics and perform electron spectroscopic imaging chemical analysis. To do this. First, insert the filter slit aperture into the omega energy filter in the TEM.

Tune the width of the filter slit aperture to get a proper energy width of 10 to 20 electron volts in the electron energy loss spectrum. Next, shift the energy to the copper M Edge absorption peak in the electron energy loss spectrum. Go back to the imaging mode to acquire an energy filtered TEM image at the copper M Edge absorption peak.

Then shift the energy to the pre edge of the copper M Edge and get another energy filtered TEM image, correct the drift of the sample between the two images, and then divide the first image by the second one to get the jump ratio image of copper. To get a three dimensional distribution of information about the diffused particles, tilt the sample and record a tilt series starting at 138 degrees. Use a tilt step of one degree and record images during every step in the bright field scanning mode.

Finally, reconstruct the series of images using standard techniques to form a three dimensional tomographic volume. Here is a brightfield TEM image from an in CQ test. There are partially breached tantalum nitride, tantalum barriers, and preexisting copper atoms in the ultra locate dielectrics Before the electrical test due to extended storage in ambient conditions after only 376 seconds at 40 volts, the dielectric breakdown started and was accompanied with two major migration pathways of copper from the M1 metal having a positive potential with reference to the ground side.

The diffused copper particles in the ultra locate dielectrics are also visible in a flawless sample. The tip to tip structure stays intact without any damage in the tanin nitrite tanin barrier for more than 50 minutes until the breakdown occurred. It appears that metal atoms migrated into the silicon oxide from the bottom corner of the M1 metal having a positive potential indicated by a red arrow.

The electron spectroscopic chemical analysis shown here proves that there is a migration path of copper at the fracture interface between the layers. This could not be detected from the contrast of the bright field TEM images. The main advantage of our technique is that we're able to follow the TDB mechanism in situ and the TEM.

What we saw is that in the case of a breached barrier, that massive copper diffusion can occur, and even in the case that there's an intact barrier, alum nitrite, tantalum based, we saw that at the very thinnest positions, the solution can occur and then diffusion follows as well. And this gives a very, very strong emphasis on the fact that the barrier process has to be controlled, especially if we think forward to downscaling of non-electronic products.