Inside the Xilinx Kintex-7 FPGA: A closer look at the first FPGA to use HKMG technology
Kevin Gibb
4/5/2012 1:59 PM EDT
A closer look at the Kintex-7 FPGA
TMSC's HPL NMOS and PMOS transistors, as seen in the Kintex-7 FPGA, are shown below. The two transistors are made using a gate-last process, where the TiN/HfO2/oxide gate dielectric is first deposited, followed by the deposition, patterning and etching of the sacrificial polysilicon gates. Silicon nitride sidewall spacers are then formed along the sides of the gates and are used to define the source/drain regions.
The sacrificial polysilicon gates are then removed and different gate metals are deposited into the NMOS and PMOS gate regions. The bottom portions of the metal gates include the work function metals, TiAlN for the NMOS and TiN for the PMOS transistors, as can be seen in the TEM images.
And perhaps as a nod to cost savings, TSMC has eschewed strain engineering to boost the transistors’ performance. Instead, rotated wafers are used that place the transistor channels in a <100> orientation to boost the PMOS drive current. This avoids the need for embedded SiGe PMOS source/drain regions used by Intel (and by TMSC’s HP) process. (Note: The <100> refers to a direction in the silicon lattice, in this case the direction of the current flow through the channel of the transistor.)
To read the full article, click here
Related Semiconductor IP
- DSP-Based 112G SerDes
- XTAL oscillator in TSMC-7nm
- GPU
- V-by-One Verification IP
- AI model compression IP
Related Articles
- Generate FPGA designs from M-code
- FPGA design and verification using Simulink
- Designing FPGA Based Reliable Systems Using Virtex-5 System Monitor
- Xilinx responds to Altera's FPGA benchmarks
Latest Articles
- SEAM-V: A Hybrid-Decoupled RISC-V Vector Processor with Backend-Visible EP Context for Sustained Vector Throughput
- New Number Formats for FFT IP Cores in Optical OFDM Transceivers
- Reducing Power Consumption of Embedded Dynamic Memories with ECCs
- NIFA: Nonlinear IMC enhanced FPGA for efficient ML inference
- A 32-channel event-based bio-signal analog front-end with adaptive delta and pulse frequency encoding