Embedded MTJ in Interconnects via Dummy Metal Level

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The semiconductor industry faces challenges in embedding magnetic tunnel junctions (MTJ) between metal levels in tight pitch interconnects due to vertical dimension constraints and the risk of unwanted interaction with titanium nitride (TiN) during wet cleaning in MRAM device fabrication.

Innovation Solution

A method is developed to insert MTJ between Mx and Mx+2 levels using a MTJ hardmask with a wet clean stop layer, allowing compatibility with small pitch interconnects and TiN metal hardmasks, involving specific steps such as forming a capping layer, filling vias with metals like tungsten or ruthenium, and forming nitride spacers to ensure compatibility and precision.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of moving object

If MTJ is inserted between metal levels with tight pitch (e.g., 90 nm or 80 nm), then memory cell size is reduced, but vertical dimension constraint prevents insertion between adjacent metal levels (Mx and Mx+1)

Engineering Contradiction:
Improvememory cell sizeVSAvoidvertical dimension (via height)
Core Design Contradiction:
Area of moving objectVSLength of stationary object

Solution Approach 1:

The patent inserts a dummy metal level (Mx+1) between the original Mx and Mx+1 levels, effectively adding a dimensional layer to accommodate the MTJ structure. This allows the MTJ to be positioned between Mx and Mx+2 levels with sufficient vertical spacing (120 nm via height), resolving the vertical dimension constraint while maintaining tight pitch in the horizontal plane.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Manufacturing precision

If TiN metal hardmask is used for trench and via patterning in tight pitch interconnects, then patterning precision is improved, but wet cleaning after trench etching increases risk of unwanted interaction with MTJ

Engineering Contradiction:
Improvepatterning precisionVSAvoidunwanted interaction with MTJ during wet cleaning
Core Design Contradiction:
Manufacturing precisionVSObject-affected harmful factors

Solution Approach 1:

The patent introduces a protective capping layer (e.g., silicon nitride or silicon oxide) over the MTJ structure during the wet cleaning process. This intermediary layer prevents direct contact between the wet cleaning chemicals and the MTJ, eliminating unwanted interactions while allowing the TiN hardmask to maintain its patterning precision function.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If MTJ minimum height of 120 nm is required, then MTJ functionality is ensured, but insertion between Mx and Mx+1 levels with 100 nm or 75 nm via height becomes impossible

Engineering Contradiction:
ImproveMTJ functionalityVSAvoidvia height
Core Design Contradiction:
ReliabilityVSLength of stationary object

Solution Approach 1:

The patent segments the interconnect structure by inserting a dummy metal level (Mx+1) between the original Mx and Mx+1 levels. This segmentation creates separate via regions: one from Mx to Mx+1 (120 nm height for MTJ accommodation) and another from Mx+1 to Mx+2 (original via height), allowing the MTJ to achieve its required 120 nm minimum height while maintaining the original tight pitch interconnect architecture.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS10483461B2Embedded MRAM in interconnects and method for producing the same
Publication Date: 2019.11.19 GLOBALFOUNDRIES SINGAPORE PTE LTD
  • US10483461B2 patent drawing
  • US10483461B2 patent drawing
  • US10483461B2 patent drawing

AI summary

Method of forming embedded MRAM in interconnects using a metal hard mask process and the resulting device are provided. Embodiments include forming a first interlayer dielectric (ILD) layer including a first metal (Mx) level; forming a capping layer over the first ILD layer; forming magnetic tunnel junction (MTJ) structures formed in a second ILD over the first capping layer; forming a second metal (Mx+1) level in the second ILD layer; forming a second capping layer over the second ILD layer; and forming a third metal (Mx+2) level in a third ILD layer over the second capping layer.