Manufacturers can now visualize voids and phases in copper-pillar interconnects with unprecedented resolution, thanks to Synchrotron X-ray nanotomography.
For the past 5 decades, the microelectronics industry has witnessed an exponential increase in computing power, owing to the ability to pack greater numbers of transistors onto silicon chips. Thus, this has placed the industry at the center of economic and social development. However, this incredible trend has been jeopardized due to issues such as heat management and current leakage, despite the fact that transistors are consistently becoming smaller, faster, and more efficient.
In an effort to build processors to embrace ever increasing data sets and novel IT paradigms (such as the Internet of Things), manufacturers are considering advanced “More than Moore” chip architectures – including chips based on enhanced 3D integration of chips.
Thus, 3D fabrication processes are fast approaching a high degree of complexity, with copper pillars that can pass vertically between silicon planes, eventually providing a promising thoughsilicon-via (TSV) technology.
The key to ensuring high speed communication between different layers of the 3D stacked structure is reliable interconnects. Copper pillars present a major manufacturing challenge, since they measure between 1—20 microns in diameter and are formed by drilling small holes in silicon. Thus, any defects or voids within the material can cause spikes in the current density, which may end up damaging components and thereby affecting the device’s performance.
To visualize the voids in copper pillars in a prototype 3D integrated chip, IRT Nanoelec partner ST Microelectronics used X-ray nano-tomography and fluorescence at ESRF nanoimaging beamline ID16A. By studying the location of voids and elemental phases, the aim was to characterize pillar’s mechanical stability and thus the quality of the interconnect.
The ESRF’s new synchrotron X-ray techniques are able to provide unique capabilities for non-destructive imaging of 3D architectures. Further, the existence of voids in the sample (with a resolution of just 23 nm) was revealed by the extremely small spot size and high penetration depth of the X-ray beam.
Meanwhile, X-ray fluorescence enabled the team to establish the presence of different elements – including the complex phases resulting from alloys formed during soldering (see image).
Nano-computed tomography of a copper pillar reveals the distribution of voids (left), while fluorescence tomography slices at different heights (right) show the elemental phases in the soldered region on top of the pillar: silver (red), nickel (blue), tin (green) and copper (grey). [P Bleuet et al. SPIE Developments in x-ray tomography IX, 2014, San Diego]
The ESRF beamline ID16A provides a unique nanoprobe for quantitative 3D characterization of the morphology as well as the elemental composition of samples in their native state. This technique, when applied to copper pillars, reveals that voids can be resolved at length scales more than 10x shorter than what is possible using commercial laboratory equipment.
This experiment’s success led to an investment by ST Microelectronics in a long-term proposal to characterize even smaller voids in copper pillars and to explore other promising TSV architectures. The ultimate aim of this is to apply the technique to ensembles of TSVs and perform in-operando analysis on whole working devices.
- ESRF beamline ID16A is a unique experimental station – using advanced X-ray optics and a long baseline, thus producing a high-brilliance beam focused down to nanometer size
- Quantitative 3D characterization of the morphology and the elemental composition of specimens in their native state is made possible by Nano computed tomography (nanoCT) and X-ray fluorescence (XRF)
- NanoCT requires the collection of a large number of magnified phase contrast radiographs while the sample is turned over a 180-degree angle, while XRF involves scanning the sample through the tight focus while collecting the fluorescence signal characteristic of the different elements present in the device
- 3D Scanning time of three hours, producing a data set of several tens of GB per sample. Conversely, 2D radiographs can be obtained in a fraction of a second
- Improved resolution made possible by upgrades to the ESRF source, allowing smaller device structures to be studied during in operando experiments
This information has been sourced, reviewed and adapted from materials provided by The Platform for Advanced Characterisation Grenoble (PAC-G).
For more information on this source, please visit The Platform for Advanced Characterisation Grenoble (PAC-G).