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

Summary

The time-dependent dielectric breakdown (TDDB) in on-chip interconnect stacks is one of the most critical failure mechanisms for microelectronic devices. This paper demonstrates the procedure of an in situ TDDB experiment in the transmission electron microscope, which opens a possibility to study the failure mechanism in microelectronic products.

Abstract

The time-dependent dielectric breakdown (TDDB) in on-chip interconnect stacks is one of the most critical failure mechanisms for microelectronic devices. The aggressive scaling of feature sizes, both on devices and interconnects, leads to serious challenges to ensure the required product reliability. Standard reliability tests and post-mortem failure analysis provide only limited information about the physics of failure mechanisms and degradation kinetics. Therefore it is necessary to develop new experimental approaches and procedures to study the TDDB failure mechanisms and degradation kinetics in particular. In this paper, an in situ experimental methodology in the transmission electron microscope (TEM) is demonstrated to investigate the TDDB degradation and failure mechanisms in Cu/ULK interconnect stacks. High quality imaging and chemical analysis are used to study the kinetic process. The in situ electrical test is integrated into the TEM to provide an elevated electrical field to the dielectrics. Electron tomography is utilized to characterize the directed Cu diffusion in the insulating dielectrics. This experimental procedure opens a possibility to study the failure mechanism in interconnect stacks of microelectronic products, and it could also be extended to other structures in active devices.

Introduction

Since Cu interconnects were firstly introduced into the ultra-large-scale integration (ULSI) technology in 1997 ¹, low-k and ultra-low-k (ULK) dielectrics have been adopted into the back-end-of-line (BEoL) as the insulating materials between on-chip interconnects. The combination of new materials, e.g., Cu for reduced resistance and low-k/ULK dielectrics for lower capacitance, overcomes the effects of increased resistance-capacitance (RC) delay caused by interconnect dimensional shrinkage ^{2, 3}. However, this benefit was encroached by the continuing aggressive scaling of microelectronic devices in recent years. The use of low-k/ULK materials results in various challenges in the manufacturing process and for the product reliability, particularly if the interconnect pitch reaches about 100 nm or less ^4-6.

TDDB refers to the physical failure mechanism of a dielectric material as a function of time under an electric field. The TDDB reliability test is usually carried out under accelerated conditions (elevated electrical field and/or elevated temperature).

The TDDB in on-chip interconnect stacks is one of the most critical failure mechanisms for the microelectronic devices, which has already raised intense concerns in the reliability community. It will continue to be in the spotlight of reliability engineers since ULK dielectrics with even weaker electrical and mechanical properties are being integrated into the devices in advanced technology nodes.

Dedicated experiments have been performed to investigate the TDDB failure mechanism ^7-9, and a significant amount of effort has been invested to develop models which describe the relationship between electric field and lifetime of the devices ^10-13. The existing studies benefit the community of reliability engineers in microelectronics; however, many challenges still exist and many questions still need to be answered in detail. For example, proven models to describe the physical failure mechanism and degradation kinetics in the TDDB process and the respective experimental verification are still lacking. As a particular need, a more appropriate model is needed to substitute the conservative √E-model ¹⁴.

As a very important part of the TDDB investigation, typical failure analysis is facing an unprecedented challenge, i.e., providing comprehensive and hard evidence to explain the physics of failure mechanisms and degradation kinetics. Apparently, inspecting millions of vias and meters of nanoscale Cu lines one by one and ex situ imaging the failure site is not the appropriate choice to hurdle this challenge, because it is very time consuming, and only limited information about the kinetics of the damage mechanism can be provided. Therefore, an urgent task has emerged to develop and to optimize experiments and to get a better procedure to study the TDDB failure mechanisms and degradation kinetics.

In this paper, we will demonstrate an in situ experimental methodology to investigate the TDDB failure mechanism in Cu/ULK interconnect stacks. A TEM with the ability of high quality imaging and chemical analysis is used to study the kinetic process at dedicated test structures. The in situ electrical test is integrated into the TEM experiment to provide an elevated electrical field to the dielectrics. A customized “tip-to-tip” structure, consisting of fully encapsulated Cu interconnects and insulated by a ULK material, is designed in the 32 nm CMOS technology node. The experimental procedure described here can also be extended to other structures in active devices.

Protocol

1. Preparing the Sample for the Focused Ion Beam (FIB) Thinning (Figure 1)

1. Cleave the full wafer into small chips (~ 10 mm by 10 mm) with a diamond scribe.
2. Mark the positions of the “tip-to-tip” structure on the chips.
3. Saw the chip with a dicing machine to obtain bars of 60 µm by 2 mm size. The bar includes the “tip-to-tip” structure in the center.
4. Glue the target bar on a Cu half ring using the super glue. Next, glue the bar on a Cu sample stage also using the super glue. Then, use silver paste to set the conduction between the half ring and the copper sample stage.
   Note: When handling the sample, make sure to always wear an antistatic wrist strap to prevent electrostatic discharges, which may damage the sensitive structure in the sample.

2. FIB Thinning in the Scanning Electron Microscope (Figure 2)

1. Put the sample obtained in step 1 on an SEM sample stage and place the stage carefully into the SEM.
2. Chose the deposition mode, and set up the dimensions (area and thickness) of the needed Pt protection layer. Always use a 30 kV ion beam to maintain the highest precision. Tune the current to get the satisfied efficiency, dependent on the dimensions of the needed Pt layer.
   1. Deposit a Pt line to contact one pad to the Cu stage (ground potential). Subsequently, deposit a thick Pt layer on top of the “tip-to-tip” structure, which is very important to minimize the ion damage during the FIB thinning process and to reinforce the thin lamella. This is a standard procedure used in FIB preparation.
   2. Take caution not to introduce any conductive paths between the two pads on top of the “tip-to-tip” structure through the Pt layer when performing the Pt deposition. Any conductive path will short the electrical circuit (Figure 2A and B).
3. FIB milling
   1. Use a voltage of 30 kV and current of 10 pA for the final cut. Thin the target bar into an H-bar TEM lamella with a thickness between 150 and 180 nm.
   2. Cut a notch close to the pad (V+ pad) which will be touched by a transducer tip in the TEM. Use the notch as a marker to identify the correct pad in the TEM.

3. Sample Transfer from the SEM to the TEM

1. Put on the antistatic wrist strap before touching the sample.
2. Dismount the prepared H-bar sample from the SEM stage. Keep the sample on the Cu stage when removing it from the SEM.
3. Fix the Cu stage onto the TEM holder. Move the transducer tip of the TEM holder close to the test structure (a few hundreds of micrometers away from the test structure) under the optical microscope.
   1. Insert the TEM holder into the TEM carefully. Do not utilize any cleaning treatment (e.g., plasma cleaning) during the transfer process, otherwise the lamella may be influenced.
4. Keep the time for the sample transfer within 15 min or shorter to avoid too much exposure to ambient moisture and oxygen.

4. Establishing the Electrical Connection (Figure 3)

1. Connect the TEM holder to its control system and the SourceMeter. Then switch on the control system and the SourceMeter.
2. Monitor the transducer tip in the TEM when doing the coarse approach of the transducer tip to the test structure by tuning the knobs on the TEM holder.
   1. Move the transducer tip of the TEM holder close to the V+ pad (≤ 500 nm). Bring the transducer tip to the same level (Z: height) as the pad. Tune the position of the tip and make the tip face the center of the V+ pad.
3. Contact the transducer tip to the V+ pad. Set a very low voltage on the tip (0.5 V to about 1 V) while approaching the pad. Monitor the current simultaneously to make sure the contact is established.

5. In Situ TDDB Experiment

1. Use an accelerated voltage of 200 kV in the TEM. Move the electron beam to the area of interest; choose a proper magnification and focus the image.
2. Use low illumination steps (≤ 8) to reduce the beam damage on the test structure. Use a condenser aperture to localize the illumination area only within the thin part of the H-bar sample.
3. Apply a constant voltage (≤ 40 V) on the “tip-to-tip” structure using the SourceMeter while recording the TEM images in situ (2-3 frames/sec). Record the images automatically by using a self-scripted code, e.g., using the DigitalMicrograph software.
4. Pause the experiment when seeing an apparent diffusion of metal into the ULK dielectrics and do the Electron Spectroscopic Imaging (ESI) chemical analysis.
   1. Insert the filter slit aperture into the Omega energy filter in the TEM.
   2. Tune the width of the filter silt aperture to get a proper energy width (10–20 eV) in the electron energy loss spectrum (EELS).
   3. Shift the energy to the copper M-edge adsorption peak in the EELS.
   4. Go back to the imaging mode to acquire an energy filtered TEM image at the Cu M-edge absorption peak.
   5. Shift the energy to the pre-edge of the copper M-edge and get another energy filtered TEM image.
   6. Correct the drift of the sample between the two images.
   7. Divide the first image by the second one to get the jump ratio image of Cu.
5. Continue the TDDB experiment: reapply a constant voltage (≤ 40 V) on the “tip-to-tip” structure using the SourceMeter and record the TEM images.

6. Computed Tomography

1. Perform TEM computed tomography when the TDDB experiment is finished, to get 3D distribution information about the diffused particles.
2. Tilt the sample and record a tilt series of 138°. Use a tilt step of 1°, and record the image during every step in the bright field (BF) STEM mode.
3. Reconstruct the series (includes aligning images, determining tilt axis, reconstructing volume and segmentation to form the 3D tomographic volume).

Results

Figure 4 shows bright field (BF) TEM images from an in situ test. There are partially breached TaN/Ta barriers and pre-existing Cu atoms in the ULK dielectrics before the electrical test (Figure 4A) due to extended storage in ambient. After only 376 sec at 40 V, 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 ^15-16. The diffused Cu particle...

Discussion

The prerequisite of success in the TDDB experiment is good sample preparation, especially in the FIB milling process in the SEM. Firstly, a thick Pt layer on top of the “tip-to-tip” structure has to be deposited. The thickness and the size of the Pt layer can be adjusted by the SEM operator, but have to follow three principles: (1) The thickness and the size are enough to protect the target area from possible ion beam damage during the whole milling process; (2) There is still a relatively thick Pt layer (...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Automatic Dicing Saw	DISCO Kiru-Kezuru-Migaku Technologies
Scanning Electron Microscope	Zeiss	Zeiss Nvision 40
Picoindentor	Hysitron	Hysitron Pi95
Keithley SourceMeter	Keithley	Keithley 2602/237
Transmission Electron Microscope	FEI	FEI Tecnai F20
Transmission Electron Microscope	Zeiss	Zeiss Libra 200