In microelectronic devices, barrier layers must be placed between the semiconductor silicon and the metal wiring connecting different parts of its surface. The barrier keeps the metal from diffusing into the silicon and ruining its transistor characteristics. Aluminum and tungsten are the metals commonly used for these circuits. In the near future, copper will also be used because of its lower electrical resistance and better durability against electro-migration. In the absence of a barrier layer, aluminum would alloy with the silicon, producing etch pits that can short out the electrical circuits; tungsten would peel off of the silicon dioxide insulating layers; or copper would diffuse into the silicon and provide deleterious recombination centers for the electrons and holes. The barrier layers must cover the sidewalls and bottom of the etched features with a film thickness about the same as on the outer surface. In other words, the barrier layer should have step coverage close to one.

Titanium nitride is the material that is usually used as the barrier layer. The titanium nitride is ordinarily formed by the process of reactive sputtering of a titanium target in a low pressure of nitrogen gas. The sputtered material has been satisfactory for the production of computer chips with feature sizes down to about one-quarter of a micron. As the industry tries to make the circuits operate faster and store more information, the feature sizes are being reduced. For feature sizes less than about one-quarter of a micron, sputtering does not cover adequately the sides and bottoms of the narrow holes and trenches that are etched a micron deep into the substrates. Thus a critical need is perceived for barrier layers deposited by a process that has better step coverage than sputtering can provide. Another problem for the use of titanium nitride in future generations of computer chips is that it may not be an effective diffusion barrier for thicknesses below about 30 nm. For features below 0.25 micron, thinner diffusion barriers will be needed, so that the barrier material does not take up too much of the hole. Titanium nitride films have a microcrystalline structure that allows diffusion of copper through thin titanium nitride barriers along boundaries between the microcrystalline grains.

Amorphous diffusion barriers are expected to perform better than microcrystalline ones, because amorphous materials lack intergranular pathways for diffusion. Amorphous tantalum nitride or niobium nitride form the most conductive known thin barriers to diffusion of copper. Unfortunately, the sputtering processes commonly used to make amorphous tantalum nitride or niobium nitride do not provide adequate step coverage.

Under a previous NSF grant, the PI discovered a process for CVD of amorphous niobium nitride with excellent step coverage at temperatures below 400 oC. In order to gain commercial acceptance for this new CVD process, a pure precursor with completely reproducible properties is needed, not the currently available mixture with unpredictable proportions. Analytical methods need to be established to verify the composition and purity of the precursor. The chemical properties of the precursor must be studied, including its reactivity to materials of construction, air and water, and its stability in storage. Various physical properties, such as vapor pressure, density and viscosity, must also be measured. The composition and quantities of the CVD reaction byproducts must be determined, so that they can be neutralized and disposed of properly. Finally, the kinetics and reaction mechanism must be understood so that proper chemical engineering of the reactor can be accomplished and rational control of the reaction conditions can be practiced. The proposed research will establish this fundamental knowledge base needed for commercialization of CVD niobium nitride barriers in the microelectronics industry. ***

National Science Foundation (NSF)
Division of Electrical, Communications and Cyber Systems (ECCS)
Application #
Program Officer
Usha Varshney
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Harvard University
United States
Zip Code