3D NAND requires new approaches to automated metrology and process control

Jack Hager, Sr. Product Marketing Manager, Materials & Structural Analysis, Thermo Fisher Scientific

NAND memory manufacturers will continue their rapid transition from planar to three-dimensional (3D) stacked architectures, with 64- and 96-layer devices expected to come into full production in 2018. This transition is driven by the significant advantages offered by 3D NAND in both size and cost. Stacking multiplies the number of bits that can be stored in the same footprint, thus reducing cost per bit in a process where costs are roughly proportional to the area of the device on the wafer. It also relaxes the resolution require- ments on photolithographic processes, permitting the use of less expensive tools and technologies. The market for high capacity 3D NAND currently comprises solid state drive and mobile applications, but this space is expected to expand.

As high layer-count 3D NAND moves into production, manufacturers are looking for metrology and process control solutions that can measure the critical dimensions (CD) of the very high aspect-ratio tube-within- tube vertical structures used to trap charge and connect individual memory cells across many layers. Critical dimension scanning electron microscopy (CD-SEM) is limited by its top-down point-of-view and challenge to see below the surface. Optical techniques can look below the surface, but are limited, especially in development and early ramp phases, by the need to develop complex models based on empirical data from the measured structures.

Existing subsurface techniques are continuing to evolve to provide solution pathways. Focused ion beams (FIB) can cut conventional or oblique cross sections to reveal structural information at varying depths to measure with a scanning electron micro- scope (SEM). This technique takes advantage of the relatively large dimensions of particular 3D NAND features to collect sub surface data quickly. Another approach creates thin-section samples in the horizontal plane (planar) at various depths for imaging in a transmission electron microscope (TEM). The TEM approach provides much higher resolution and the ability to enhance contrast among the multiple concentric layers inside each “container” by adding information from other analytical signals, such as Energy Dispersive Spectroscopy (EDS). Both approaches are being enhanced to provide automated, robust and repeatable process control insights.