Learn more about R&D chemical mechanical polishing by requesting our FREE informational CD.

Download our CMP White Paper


FREE subscription
Subscribe for free to receive each issue of Semiconductor Today magazine and weekly news brief.


21 December 2007


Micromem makes first fully functioning MRAM cells

Fabless GaAs firm Micromem Technologies Inc of Toronto, Canada says that it has manufactured the first foundry-grade fully functioning magnetic random access memory (MRAM) cells. This culminates an intensive three-year R&D proof-of-concept phase followed by a foundry phase focused on manufacturability and scalability of its MRAM product.

In late summer Micromem contracted engineering company Strategic Solutions of Grass Valley, CA, USA as a partner to design and implement its MRAM technology. To orchestrate a knowledge transfer of work by Dr Harry Ruda and his team at the University of Toronto, Strategic Solutions created a manufacturing design-of-experiment and a reticle design (consisting of thousands of MRAM bit cells in various configurations) as well as a test plan document, in order to communicate manufacturing requirements to a GaAs foundry.

In September, Micromem contracted Global Communication Services Inc (GCS) of Torrance, CA, USA as its partner foundry. The MRAM design entered the foundry on September 17. Phase One of the contract aims to establish the manufacturability of the MRAM design, via tests to determine the optimum configuration for memory performance and the foundry manufacturing costs.

Micromem says that the development team at Strategic Solutions and GCS have worked closely to meet the aggressive schedule initiated this autumn. The initial foundry phase has now generated much statistically validated MRAM performance data. During January Micromem will analyze this large data set. The firm’s initial data review indicates that the test plan was successful and that the data set is rich with scalable data that indicates a clear path to an optimized MRAM cell design.

Also, in January, Micromem aims to deliver functioning MRAM cells in specific package formats for third-party radiation-hard testing, focused on medical device sterilization radiation requirements, galactic radiation test protocols for space-based applications, and weapons radiation protocols. Phase One should be completed in late January with the manufacture of working multi-bit MRAM cells. Micromem plans to then communicate the results of the Phase One test plan and to release MRAM performance data in early February.

With Phase One foundry work ahead of schedule, Micromem has therefore now contracted Strategic Solutions to begin designing a reticle design and test plan for a 64-bit MRAM cell. These arrays will be transistorized and packaged to allow third parties to test and validate standard memory test protocols, including power, speed and robustness. In that same foundry run, Micromem says that, driven by the Phase One test plan data set, it will take the necessary steps to further optimize the fundamental MRAM bit cell architecture. It is anticipated that 64-bit MRAM arrays will be available for client testing in 3-4 months.

With the progress of Phase One, the availability of foundry-grade MRAM cells packaged for third-party testing, and the release of the 64-bit MRAM array into the next foundry phase, Micromem says that it will be equipped to engage clients with targeted product development and integration requirements. These steps were mandatory for holding discussions in earnest with clients that have expressed interest in partnering. Micromem claims that, compared with competing MRAM cell architectures with multiple thin-film layers, Phase One has proven that the memory is simple to manufacture and is robust in its design.

A further outcome of the initial foundry phase is the high-sensitivity performance of the thin-film Hall cross-sensor component of the MRAM cell structure, says program director Steven Van Fleet. “This has afforded the company an opportunity to increase our product portfolio with a high-sensitivity, low-cost Hall cross-sensor.” The February 2006 Frost & Sullivan Report on ‘Extremely Sensitive Magnetic Sensors’ indicated a 2010 market size projection of $2bn, with Hall sensors potentially representing about 80% of the market, he adds. In 2005 the total world magnetic sensor markets was about $1bn.

Micromem says that it has completed preliminary market discussions with targeted clients, and that there appears to be a good revenue opportunity to supply standalone Hall cross-sensors for incorporation into the various applications. In January the firm will deliver Hall cross-sensor evaluation packages to several clients that have agreed to test and provide performance and cost/benefit feedback.

CEO Joe Fuda adds that Micromem is now focusing on its client capture strategy, which will proceed with three parallel market streams, aiming to secure:

  • a developmental and market development partner for GaAs-based MRAM, focusing primarily on military, health-care and automotive applications;
  • end-user clients for Hall cross-sensor applications requiring high sensitivity in thin-film solutions;
  • a developmental and market development partner for conversion of the GaAs memory technology to silicon germanium, allowing the firm to participate in the larger silicon-based memory space (focusing on lower-cost, higher-density memory requirements).

See related item:

Micromem's Hall cross sensor meets expectations; on schedule for functioning MRAM by year end

Micromem’s GaAs MRAM device processed at GCS

Micromem engages GCS as foundry for MEMS

Search: GaAs