New tests show in real-time that cracks can run on top of and through metal layers.
BY K. VANSTREELS, I. DE WOLF, H. ZAHEDMANESH, H. BENDER, M. GONZALEZ, J. LEFEBVRE, AND S. BHOWMICK, imec, Leuven, Belgium
Imec, with the help of equipment supplier Hysitron, developed a new test method to study crack formation in Cu/low-k back-end-of-line (BEOL) stacks. By combining a PicoIndenter, a scanning electron microscope (SEM) and unique sample preparation using a focused ion beam (FIB), it becomes possible to visualize in real-time crack initiation and propagation (See video). Insight into this reliability issue allows to optimize BEOL design, material choices and process steps to strengthen the BEOL.
Porosity of low-k materials affects the BEOL’s mechanical strength
Porous low-k materials are introduced in the BEOL of chips to improve its performance. More specifically, low-k materials prevent leakage between metal interconnections of the circuitry and minimize the time delay. In research, low-k materials with k-values lower than 2.0 are tested. These are very porous and reduce the mechanical strength of the BEOL stack.
SEM picture of crack formation in the back-end-of-line.
Due to the reduced mechanical strength of Cu/ low-k BEOL stacks, cracks can be formed when local mechanical stresses become too high. This can occur during chip processing, during packaging, and during use of the end products. The stresses can be caused by temperature fluctuations, due to thermal mismatch of materials; or by shrinkage of materials during curing; by local forces during bonding; or even due to external mechanical impact caused by drop or shock. Since this crack formation is an important reliability issue for future technology nodes, tests are being developed to gain insight into this problem. For example, there is the four-point bending test and the BABSI test which measure the force at which cracking starts. However, with these tests it is not possible to follow in-situ and in real-time how the crack initiates and propagates through the BEOL stack. With the new test method, this hidden world reveals itself.
The new test
FIGURE 1. Schematic overview of the sample preparation. Using a Focus Ion Beam, a double clamped BEOL beam sample is made.
For the new test, a PicoIndenter is integrated into a SEM microscope. The sample has to be prepared in such a way that a beam is formed out of the back-end- of-line stack (see FIGURE 1). Imec was able to make such BEOL beams with the Focused Ion Beam (FIB) technique. This involved several steps as depicted in FIGURE 1. The sample is then placed in the SEM. With the PicoIndenter, a gradual force is applied on top of the beam while the SEM continuously images the cross-section of the BEOL (side view of the beam). In this way, a movie can be made revealing the crack initiation and propagation, while at the same time measuring the force that is applied (FIGURE 2).
FIGURE 2. Force-displacement curve and corresponding SEM pictures, as measured with the new test method. The pictures reveal beam bending (a to c), crack initiation (d) and crack growth (e to h).
Conclusions and future work
This new test has proven to be very relevant for further development of Cu/low-k BEOL stacks. In this phase, only a few BEOL beams – with different dimensions – were tested and measured. From these initial results it can be concluded that cracks run on top of metal layers in this device, and even through metal layers. This can point out that the interface at the top of metal layers should be strengthened for the studied technology, either by BEOL design, material choices or optimization of process steps (such as cleaning).
By setting up more experiments, a model can be made (FEM) to predict crack formation in specific BEOL stacks. The experiments will allow to validate the model. In this way, this new test method is an important tool in the development of reliable chips made in future technology nodes with copper and low-k materials in the back-end-of-line.
This article is based on the paper “In-situ scanning electron microscopy study of fracture events during back-end-of-line microbeam bending tests” which was published in Applied Physics Letters 105, 213102 (2014). AUTHORS: K. VANSTREELS, I. DE WOLF, H. ZAHEDMANESH, H. BENDER, M. GONZALEZ, J. LEFEBVRE, AND S. BHOWMICK