Motorola and Infineon Improve APC Strategy
Laura Peters, Senior Editor -- Semiconductor International, 2/1/2001
The group at Motorola, headed by Perry Tapp, presented a methodology for lithography equipment control that reduces lot-to-lot variability, drives CD to target and minimizes overlay errors. Using the Catalyst process control platform and an Oracle database, Motorola engineers built ALEC, the Advanced Lithography Equipment Controller, a program that can control CDs within 2.5% of the target value and reduce overlay mean errors to <60 nm.
ALEC works by collecting and analyzing CD and overlay data from previous lots, selecting an appropriate model, retrieving the model configuration, information and associated lot data, and executing algorithms to predict optimum dose and overlay parameters. It then downloads those parameters to the stepper. The program uses overlay and dose formulas to calculate model-based recommended values. Using this approach, the engineers emphasize the ultimate reliance on high-quality metrology data. They recommend an appropriate sampling strategy, overlay data filtering with flier removal capability and accurate CD-SEM measurements for lot averaging.
The ALEC architecture was designed specifically for integration with other applications such as MatLab. Importantly, ALEC's model-specific offsets enable overlay and dose corrections beyond current tool matching capabilities.
|
|
Like the Motorola approach, Infineon uses a feedforward/ feedback loop between the stepper's control of exposure dose and CD and overlay measurements to meet specifications and improve yield. The APC system uses reticle- and stepper-dependent exposure dose and CD target, as well as stepper-dependent overlay offsets. Between the photolithography database, ACE CD control, ACE overlay control, CIM database and metrology tools, information is fed through to modify exposure dose and overlay parameters.
The equipment and process monitoring (EPM) portion of Infineon's APC program treats raw stepper data using filters and algorithms and then assigns key numbers including 90 raw data points and 50 calculated parameters that reflect reticle CD, wafer alignment, focus, process times, etc. Key number tracking of individual process parameters allows rapid identification of tool problems such as misalignment, stage rotation problems and double exposure.
For additional information on yield management, go to www.semiconductor.net/yield