All kits
40 nm · Planar bulk, SiON

LM40PB-L

40 nm planar bulk, low power

Status

First development release (0.1.0-dev).

A generated NAND2 cell from LM40PB-L, layer by layer at true heights
A generated NAND2 cell from LM40PB-L, layer by layer at true heights

LM40PB-L is a complete 40 nm-class low-power kit: planar bulk transistors with a silicon-oxynitride gate oxide, three threshold-voltage flavours, two cell heights and a nine-layer copper interconnect. It is the reference kit of the family and the first one taken all the way to characterized and checked libraries.

Why 40 nm

The 40 nm generation remains one of the most used technologies in production: microcontrollers, connectivity, power management, automotive and IoT chips. It is also the last generation of the classic planar transistor, which makes it the natural baseline for every comparison in our family: what changes when the gate wraps around the channel, when wires get narrower, when the supply drops.

Technology at a glance

Supply1.05 V nominal; 0.95 and 1.15 V corners
Contacted gate pitch146 nm
Gate length32 nm (LVT), 35 nm (SVT), 38 nm (HVT)
Gate oxideSiON, 1.2 nm physical, 1.83 nm electrical
Metal stack9 copper layers, 113 nm minimum pitch, low-k dielectric
Cell heights9 tracks (1017 nm) and 12 tracks (1356 nm)
Threshold voltagesHVT, SVT, LVT
Device modelBSIM4

Transistors

Each flavour is fitted to cited targets for off-state leakage and on-state drive, and to cited gate capacitance, so that both speed and switching energy are realistic.

DeviceLeakage (nA/µm)Drive (µA/µm)
nMOS HVT / SVT / LVT0.05 / 0.5 / 5536 / 638 / 700
pMOS HVT / SVT / LVT0.05 / 0.5 / 5335 / 389 / 438
Id–Vg and Id–Vd of the six core devices
Id–Vg and Id–Vd of the six core devices

Interconnect

Nine copper layers, from 113 nm pitch for local wiring to 729 nm for power. Resistance includes the size effect of narrow copper lines (surface and grain-boundary scattering), and vias are modelled with an interface term calibrated on a published 45 nm reference.

The full stack to scale, from gates to the top metal
The full stack to scale, from gates to the top metal
Resistance and capacitance per metal layer
Resistance and capacitance per metal layer

Standard cells

  • 130 cells per library: inverters and buffers, NAND/NOR/AND/OR with 2–4 inputs, AOI/OAI from 21 to 222, MUX, XOR/XNOR, latch, flip-flops with and without reset, clock gate, tie, fill, well tap and decoupling cells; drive strengths X1–X8.
  • 9-track libraries in all three threshold flavours: a NAND2 gate density of about 2.2 million gates per mm².
  • 12-track library with about 40 % stronger transistors per cell for performance-critical blocks (timing characterization in progress).
A D flip-flop from the 9-track library
A D flip-flop from the 9-track library

Timing and power

The 9-track libraries are characterized at 13 process, voltage and temperature corners: all five process corners for SVT, and typical, slow hot and fast cold for HVT and LVT. A fanout-of-four inverter delay at the typical corner is about 23 ps (SVT) and 16 ps (LVT).

Inverter delay and leakage across corners
Inverter delay and leakage across corners

Verified

  • Every cell from its drawn layout: extracted independently and tested for its logic function, 130/130 per library.
  • Against 45 nm-class references it was not fitted to (gate capacitance, delay per stage, drive, input capacitance, via resistance, cell area): all within band.
  • Against transistor-level simulation of real logic paths (an 8×8 multiplier and a counter, driven by real input vectors): the timing libraries are on the safe side throughout, with a mean deviation of about 2 %.
Library timing vs transistor-level simulation on real paths
Library timing vs transistor-level simulation on real paths

These checks also found and fixed real problems before release, among them XOR and MUX timing tables measured for one switching direction only, and AOI/OAI tables measured in their fastest input state. All are documented.

How much slower is a chip after ten years at 125 °C? We model the two main transistor wear-out mechanisms from published 40 nm-class measurements:

  • Bias-temperature instability in pMOS, which raises the threshold voltage whenever the transistor is on;
  • Hot-carrier injection in nMOS, which damages the transistor while it switches.

The degraded transistors are characterized into aged timing libraries that any timing tool can use directly. Ten years at nominal voltage shifts the pMOS threshold by about 40 mV in typical use and up to 60 mV under constant stress.

Threshold shift vs time (pMOS) and current loss (nMOS)
Threshold shift vs time (pMOS) and current loss (nMOS)

Built from

International roadmap tables (ITRS 2009), published 45 nm process papers, the FreePDK45 research kit, and peer-reviewed reliability measurements.

Interactive 3D cells

Generated cells, layer by layer at true heights. Pick a cell to load it.

LM40PB-L 3D

Drag to rotate, scroll or pinch to zoom. Heights are drawn at 2× for legibility.

Privacy settings

Choose what this site may use. You can change your mind at any time from the Privacy settings link at the bottom of every page.

What the site needs to work, and the record of the choice you make here, kept in your browser. No cookies are set.

Always on

Would let us count visits and see which pages are useful, without advertising or cross-site tracking. We do not use any analytics today, so nothing is loaded whichever you choose.

Cookie policy · Privacy policy